mptcp: fix rcv buffer auto-tuning

The MPTCP code uses the assumption that the tcp_win_from_space() helper
does not use any TCP-specific field, and thus works correctly operating
on an MPTCP socket.

The commit dfa2f0483360 ("tcp: get rid of sysctl_tcp_adv_win_scale")
broke such assumption, and as a consequence most MPTCP connections stall
on zero-window event due to auto-tuning changing the rcv buffer size
quite randomly.

Address the issue syncing again the MPTCP auto-tuning code with the TCP
one. To achieve that, factor out the windows size logic in socket
independent helpers, and reuse them in mptcp_rcv_space_adjust(). The
MPTCP level scaling_ratio is selected as the minimum one from the all
the subflows, as a worst-case estimate.

Fixes: dfa2f0483360 ("tcp: get rid of sysctl_tcp_adv_win_scale")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Co-developed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Link: https://lore.kernel.org/r/20230720-upstream-net-next-20230720-mptcp-fix-rcv-buffer-auto-tuning-v1-1-175ef12b8380@tessares.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

3 files changed, 15 insertions, 10 deletions

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 22323b1fa184..68da26539970 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -90,6 +90,7 @@ static int __mptcp_socket_create(struct mptcp_sock *msk)
 	if (err)
 		return err;
 
+	msk->scaling_ratio = tcp_sk(ssock->sk)->scaling_ratio;
 	WRITE_ONCE(msk->first, ssock->sk);
 	WRITE_ONCE(msk->subflow, ssock);
 	subflow = mptcp_subflow_ctx(ssock->sk);
@@ -1881,6 +1882,7 @@ static void mptcp_rcv_space_adjust(struct mptcp_sock *msk, int copied)
 {
 	struct mptcp_subflow_context *subflow;
 	struct sock *sk = (struct sock *)msk;
+	u8 scaling_ratio = U8_MAX;
 	u32 time, advmss = 1;
 	u64 rtt_us, mstamp;
 
@@ -1911,9 +1913,11 @@ static void mptcp_rcv_space_adjust(struct mptcp_sock *msk, int copied)
 
 		rtt_us = max(sf_rtt_us, rtt_us);
 		advmss = max(sf_advmss, advmss);
+		scaling_ratio = min(tp->scaling_ratio, scaling_ratio);
 	}
 
 	msk->rcvq_space.rtt_us = rtt_us;
+	msk->scaling_ratio = scaling_ratio;
 	if (time < (rtt_us >> 3) || rtt_us == 0)
 		return;
 
@@ -1922,8 +1926,8 @@ static void mptcp_rcv_space_adjust(struct mptcp_sock *msk, int copied)
 
 	if (READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_moderate_rcvbuf) &&
 	    !(sk->sk_userlocks & SOCK_RCVBUF_LOCK)) {
-		int rcvmem, rcvbuf;
 		u64 rcvwin, grow;
+		int rcvbuf;
 
 		rcvwin = ((u64)msk->rcvq_space.copied << 1) + 16 * advmss;
 
@@ -1932,18 +1936,13 @@ static void mptcp_rcv_space_adjust(struct mptcp_sock *msk, int copied)
 		do_div(grow, msk->rcvq_space.space);
 		rcvwin += (grow << 1);
 
-		rcvmem = SKB_TRUESIZE(advmss + MAX_TCP_HEADER);
-		while (tcp_win_from_space(sk, rcvmem) < advmss)
-			rcvmem += 128;
-
-		do_div(rcvwin, advmss);
-		rcvbuf = min_t(u64, rcvwin * rcvmem,
+		rcvbuf = min_t(u64, __tcp_space_from_win(scaling_ratio, rcvwin),
 			       READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_rmem[2]));
 
 		if (rcvbuf > sk->sk_rcvbuf) {
 			u32 window_clamp;
 
-			window_clamp = tcp_win_from_space(sk, rcvbuf);
+			window_clamp = __tcp_win_from_space(scaling_ratio, rcvbuf);
 			WRITE_ONCE(sk->sk_rcvbuf, rcvbuf);
 
 			/* Make subflows follow along.  If we do not do this, we
diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
index 37fbe22e2433..795f422e8597 100644
--- a/net/mptcp/protocol.h
+++ b/net/mptcp/protocol.h
@@ -321,6 +321,7 @@ struct mptcp_sock {
 		u64	time;	/* start time of measurement window */
 		u64	rtt_us; /* last maximum rtt of subflows */
 	} rcvq_space;
+	u8		scaling_ratio;
 
 	u32		subflow_id;
 	u32		setsockopt_seq;
@@ -351,9 +352,14 @@ static inline int __mptcp_rmem(const struct sock *sk)
 	return atomic_read(&sk->sk_rmem_alloc) - READ_ONCE(mptcp_sk(sk)->rmem_released);
 }
 
+static inline int mptcp_win_from_space(const struct sock *sk, int space)
+{
+	return __tcp_win_from_space(mptcp_sk(sk)->scaling_ratio, space);
+}
+
 static inline int __mptcp_space(const struct sock *sk)
 {
-	return tcp_win_from_space(sk, READ_ONCE(sk->sk_rcvbuf) - __mptcp_rmem(sk));
+	return mptcp_win_from_space(sk, READ_ONCE(sk->sk_rcvbuf) - __mptcp_rmem(sk));
 }
 
 static inline struct mptcp_data_frag *mptcp_send_head(const struct sock *sk)
diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index 9ee3b7abbaf6..ad7080fabb2f 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -1359,7 +1359,7 @@ void mptcp_space(const struct sock *ssk, int *space, int *full_space)
 	const struct sock *sk = subflow->conn;
 
 	*space = __mptcp_space(sk);
-	*full_space = tcp_full_space(sk);
+	*full_space = mptcp_win_from_space(sk, READ_ONCE(sk->sk_rcvbuf));
 }
 
 void __mptcp_error_report(struct sock *sk)