0001-RTAX_INITRWND-should-be-able-to-bring-the-rcv_ssthresh.patch

0001-RTAX_INITRWND-should-be-able-to-bring-the-rcv_ssthresh.patch

commit 400854c8da31a5798abc871c78dcf6d41089abe1
Author: Marek Majkowski <marek@cloudflare.com>
Date:   Mon Jul 18 13:08:03 2022 +0200

    RTAX_INITRWND should be able to set the rcv_ssthresh above 64KiB
    
    There are three places where we initialize sockets:
     - tcp_output:tcp_connect_init
     - tcp_minisocks:tcp_openreq_init_rwin
     - syncookies
    
    In the first two we already have a call to `tcp_rwnd_init_bpf` and
    `dst_metric(RTAX_INITRWND)` which retrieve the bpf/path initrwnd
    attribute. We use this value to bring `rcv_ssthresh` up, potentially
    above the traditional 64KiB.
    
    With higher initial `rcv_ssthresh` the receiver will open the receive
    window more aggresively, which can improve large BDP flows - large
    throughput and latency.
    
    This patch does not cover the syncookies case.
    
    Signed-off-by: Marek Majkowski <marek@cloudflare.com>

diff --git a/include/net/tcp.h b/include/net/tcp.h
index 8e48dc56837b..16aaf1c6b253 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -449,7 +449,8 @@ void tcp_ld_RTO_revert(struct sock *sk, u32 seq);
 int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb);
 struct sock *tcp_create_openreq_child(const struct sock *sk,
 				      struct request_sock *req,
-				      struct sk_buff *skb);
+				      struct sk_buff *skb,
+				      const struct dst_entry *dst);
 void tcp_ca_openreq_child(struct sock *sk, const struct dst_entry *dst);
 struct sock *tcp_v4_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
 				  struct request_sock *req,
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 228d36692d08..1f84c9fea026 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -1501,7 +1501,7 @@ struct sock *tcp_v4_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
 	if (sk_acceptq_is_full(sk))
 		goto exit_overflow;
 
-	newsk = tcp_create_openreq_child(sk, req, skb);
+	newsk = tcp_create_openreq_child(sk, req, skb, dst);
 	if (!newsk)
 		goto exit_nonewsk;
 
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index 6854bb1fb32b..38360ad93baf 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -359,7 +359,7 @@ void tcp_openreq_init_rwin(struct request_sock *req,
 	int full_space = tcp_full_space(sk_listener);
 	u32 window_clamp;
 	__u8 rcv_wscale;
-	u32 rcv_wnd;
+	int adj_mss;
 	int mss;
 
 	mss = tcp_mss_clamp(tp, dst_metric_advmss(dst));
@@ -372,20 +372,16 @@ void tcp_openreq_init_rwin(struct request_sock *req,
 	    (req->rsk_window_clamp > full_space || req->rsk_window_clamp == 0))
 		req->rsk_window_clamp = full_space;
 
-	rcv_wnd = tcp_rwnd_init_bpf((struct sock *)req);
-	if (rcv_wnd == 0)
-		rcv_wnd = dst_metric(dst, RTAX_INITRWND);
-	else if (full_space < rcv_wnd * mss)
-		full_space = rcv_wnd * mss;
+	adj_mss = mss - (ireq->tstamp_ok ? TCPOLEN_TSTAMP_ALIGNED : 0);
 
 	/* tcp_full_space because it is guaranteed to be the first packet */
 	tcp_select_initial_window(sk_listener, full_space,
-		mss - (ireq->tstamp_ok ? TCPOLEN_TSTAMP_ALIGNED : 0),
+		adj_mss,
 		&req->rsk_rcv_wnd,
 		&req->rsk_window_clamp,
 		ireq->wscale_ok,
 		&rcv_wscale,
-		rcv_wnd);
+		0);
 	ireq->rcv_wscale = rcv_wscale;
 }
 EXPORT_SYMBOL(tcp_openreq_init_rwin);
@@ -448,7 +444,8 @@ static void smc_check_reset_syn_req(struct tcp_sock *oldtp,
  */
 struct sock *tcp_create_openreq_child(const struct sock *sk,
 				      struct request_sock *req,
-				      struct sk_buff *skb)
+				      struct sk_buff *skb,
+				      const struct dst_entry *dst)
 {
 	struct sock *newsk = inet_csk_clone_lock(sk, req, GFP_ATOMIC);
 	const struct inet_request_sock *ireq = inet_rsk(req);
@@ -499,10 +496,19 @@ struct sock *tcp_create_openreq_child(const struct sock *sk,
 		inet_csk_reset_keepalive_timer(newsk,
 					       keepalive_time_when(newtp));
 
+	u32 adj_mss = req->mss - (ireq->tstamp_ok ? TCPOLEN_TSTAMP_ALIGNED : 0);
+
+	u32 rcv_wnd = tcp_rwnd_init_bpf((struct sock *)req);
+	if (rcv_wnd == 0)
+		rcv_wnd = dst_metric(dst, RTAX_INITRWND);
+
 	newtp->rx_opt.tstamp_ok = ireq->tstamp_ok;
 	newtp->rx_opt.sack_ok = ireq->sack_ok;
 	newtp->window_clamp = req->rsk_window_clamp;
-	newtp->rcv_ssthresh = req->rsk_rcv_wnd;
+	if (rcv_wnd) {
+		req->rsk_rcv_wnd = min(req->rsk_rcv_wnd, rcv_wnd * adj_mss);
+	}
+	newtp->rcv_ssthresh = max(req->rsk_rcv_wnd, rcv_wnd * adj_mss);
 	newtp->rcv_wnd = req->rsk_rcv_wnd;
 	newtp->rx_opt.wscale_ok = ireq->wscale_ok;
 	if (newtp->rx_opt.wscale_ok) {
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 18c913a2347a..0f2d4174ea59 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -3642,6 +3642,7 @@ static void tcp_connect_init(struct sock *sk)
 	struct tcp_sock *tp = tcp_sk(sk);
 	__u8 rcv_wscale;
 	u32 rcv_wnd;
+	u32 mss;
 
 	/* We'll fix this up when we get a response from the other end.
 	 * See tcp_input.c:tcp_rcv_state_process case TCP_SYN_SENT.
@@ -3679,8 +3680,10 @@ static void tcp_connect_init(struct sock *sk)
 	if (rcv_wnd == 0)
 		rcv_wnd = dst_metric(dst, RTAX_INITRWND);
 
+	mss = tp->advmss - (tp->rx_opt.ts_recent_stamp ?
+			    tp->tcp_header_len - sizeof(struct tcphdr) : 0);
 	tcp_select_initial_window(sk, tcp_full_space(sk),
-				  tp->advmss - (tp->rx_opt.ts_recent_stamp ? tp->tcp_header_len - sizeof(struct tcphdr) : 0),
+				  mss,
 				  &tp->rcv_wnd,
 				  &tp->window_clamp,
 				  sock_net(sk)->ipv4.sysctl_tcp_window_scaling,
@@ -3688,7 +3691,10 @@ static void tcp_connect_init(struct sock *sk)
 				  rcv_wnd);
 
 	tp->rx_opt.rcv_wscale = rcv_wscale;
-	tp->rcv_ssthresh = tp->rcv_wnd;
+	if (rcv_wnd)
+		tp->rcv_ssthresh = max(tp->rcv_wnd, rcv_wnd * mss);
+	else
+		tp->rcv_ssthresh = tp->rcv_wnd;
 
 	sk->sk_err = 0;
 	sock_reset_flag(sk, SOCK_DONE);
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 70d4890d8d2f..b66dc0efcfd1 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -1267,7 +1267,7 @@ static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk, struct sk_buff *
 			goto out;
 	}
 
-	newsk = tcp_create_openreq_child(sk, req, skb);
+	newsk = tcp_create_openreq_child(sk, req, skb, dst);
 	if (!newsk)
 		goto out_nonewsk;