Merge branch 'net-sysctl-races-part-6'

Kuniyuki Iwashima says:

====================
sysctl: Fix data-races around ipv4_net_table (Round 6, Final).

This series fixes data-races around 11 knobs after tcp_pacing_ss_ratio
ipv4_net_table, and this is the final round for ipv4_net_table.

While at it, other data-races around these related knobs are fixed.

  - decnet_mem
  - decnet_rmem
  - tipc_rmem

There are still 58 tables possibly missing some fixes under net/.

  $ grep -rnE "struct ctl_table.*?\[\] =" net/ | wc -l
  60
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

10 files changed, 36 insertions, 31 deletions

diff --git a/include/net/sock.h b/include/net/sock.h
index 9fa54762e07752..7a48991cdb198a 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -2843,18 +2843,18 @@ static inline int sk_get_wmem0(const struct sock *sk, const struct proto *proto)
 {
 	/* Does this proto have per netns sysctl_wmem ? */
 	if (proto->sysctl_wmem_offset)
-		return *(int *)((void *)sock_net(sk) + proto->sysctl_wmem_offset);
+		return READ_ONCE(*(int *)((void *)sock_net(sk) + proto->sysctl_wmem_offset));
 
-	return *proto->sysctl_wmem;
+	return READ_ONCE(*proto->sysctl_wmem);
 }
 
 static inline int sk_get_rmem0(const struct sock *sk, const struct proto *proto)
 {
 	/* Does this proto have per netns sysctl_rmem ? */
 	if (proto->sysctl_rmem_offset)
-		return *(int *)((void *)sock_net(sk) + proto->sysctl_rmem_offset);
+		return READ_ONCE(*(int *)((void *)sock_net(sk) + proto->sysctl_rmem_offset));
 
-	return *proto->sysctl_rmem;
+	return READ_ONCE(*proto->sysctl_rmem);
 }
 
 /* Default TCP Small queue budget is ~1 ms of data (1sec >> 10)
diff --git a/net/decnet/af_decnet.c b/net/decnet/af_decnet.c
index dc92a67baea394..7d542eb4617295 100644
--- a/net/decnet/af_decnet.c
+++ b/net/decnet/af_decnet.c
@@ -480,8 +480,8 @@ static struct sock *dn_alloc_sock(struct net *net, struct socket *sock, gfp_t gf
 	sk->sk_family      = PF_DECnet;
 	sk->sk_protocol    = 0;
 	sk->sk_allocation  = gfp;
-	sk->sk_sndbuf	   = sysctl_decnet_wmem[1];
-	sk->sk_rcvbuf	   = sysctl_decnet_rmem[1];
+	sk->sk_sndbuf	   = READ_ONCE(sysctl_decnet_wmem[1]);
+	sk->sk_rcvbuf	   = READ_ONCE(sysctl_decnet_rmem[1]);
 
 	/* Initialization of DECnet Session Control Port		*/
 	scp = DN_SK(sk);
diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c
index 46e8a5125853ae..452ff177e4da95 100644
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -1042,6 +1042,7 @@ fib_find_matching_alias(struct net *net, const struct fib_rt_info *fri)
 
 void fib_alias_hw_flags_set(struct net *net, const struct fib_rt_info *fri)
 {
+	u8 fib_notify_on_flag_change;
 	struct fib_alias *fa_match;
 	struct sk_buff *skb;
 	int err;
@@ -1063,14 +1064,16 @@ void fib_alias_hw_flags_set(struct net *net, const struct fib_rt_info *fri)
 	WRITE_ONCE(fa_match->offload, fri->offload);
 	WRITE_ONCE(fa_match->trap, fri->trap);
 
+	fib_notify_on_flag_change = READ_ONCE(net->ipv4.sysctl_fib_notify_on_flag_change);
+
 	/* 2 means send notifications only if offload_failed was changed. */
-	if (net->ipv4.sysctl_fib_notify_on_flag_change == 2 &&
+	if (fib_notify_on_flag_change == 2 &&
 	    READ_ONCE(fa_match->offload_failed) == fri->offload_failed)
 		goto out;
 
 	WRITE_ONCE(fa_match->offload_failed, fri->offload_failed);
 
-	if (!net->ipv4.sysctl_fib_notify_on_flag_change)
+	if (!fib_notify_on_flag_change)
 		goto out;
 
 	skb = nlmsg_new(fib_nlmsg_size(fa_match->fa_info), GFP_ATOMIC);
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index a11e5de3a4c366..002a4a04efbe07 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -452,8 +452,8 @@ void tcp_init_sock(struct sock *sk)
 
 	icsk->icsk_sync_mss = tcp_sync_mss;
 
-	WRITE_ONCE(sk->sk_sndbuf, sock_net(sk)->ipv4.sysctl_tcp_wmem[1]);
-	WRITE_ONCE(sk->sk_rcvbuf, sock_net(sk)->ipv4.sysctl_tcp_rmem[1]);
+	WRITE_ONCE(sk->sk_sndbuf, READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_wmem[1]));
+	WRITE_ONCE(sk->sk_rcvbuf, READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_rmem[1]));
 
 	sk_sockets_allocated_inc(sk);
 }
@@ -1724,7 +1724,7 @@ int tcp_set_rcvlowat(struct sock *sk, int val)
 	if (sk->sk_userlocks & SOCK_RCVBUF_LOCK)
 		cap = sk->sk_rcvbuf >> 1;
 	else
-		cap = sock_net(sk)->ipv4.sysctl_tcp_rmem[2] >> 1;
+		cap = READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_rmem[2]) >> 1;
 	val = min(val, cap);
 	WRITE_ONCE(sk->sk_rcvlowat, val ? : 1);
 
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index c799f39cb774d9..b1637990d57084 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -426,7 +426,7 @@ static void tcp_sndbuf_expand(struct sock *sk)
 
 	if (sk->sk_sndbuf < sndmem)
 		WRITE_ONCE(sk->sk_sndbuf,
-			   min(sndmem, sock_net(sk)->ipv4.sysctl_tcp_wmem[2]));
+			   min(sndmem, READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_wmem[2])));
 }
 
 /* 2. Tuning advertised window (window_clamp, rcv_ssthresh)
@@ -461,7 +461,7 @@ static int __tcp_grow_window(const struct sock *sk, const struct sk_buff *skb,
 	struct tcp_sock *tp = tcp_sk(sk);
 	/* Optimize this! */
 	int truesize = tcp_win_from_space(sk, skbtruesize) >> 1;
-	int window = tcp_win_from_space(sk, sock_net(sk)->ipv4.sysctl_tcp_rmem[2]) >> 1;
+	int window = tcp_win_from_space(sk, READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_rmem[2])) >> 1;
 
 	while (tp->rcv_ssthresh <= window) {
 		if (truesize <= skb->len)
@@ -574,16 +574,17 @@ static void tcp_clamp_window(struct sock *sk)
 	struct tcp_sock *tp = tcp_sk(sk);
 	struct inet_connection_sock *icsk = inet_csk(sk);
 	struct net *net = sock_net(sk);
+	int rmem2;
 
 	icsk->icsk_ack.quick = 0;
+	rmem2 = READ_ONCE(net->ipv4.sysctl_tcp_rmem[2]);
 
-	if (sk->sk_rcvbuf < net->ipv4.sysctl_tcp_rmem[2] &&
+	if (sk->sk_rcvbuf < rmem2 &&
 	    !(sk->sk_userlocks & SOCK_RCVBUF_LOCK) &&
 	    !tcp_under_memory_pressure(sk) &&
 	    sk_memory_allocated(sk) < sk_prot_mem_limits(sk, 0)) {
 		WRITE_ONCE(sk->sk_rcvbuf,
-			   min(atomic_read(&sk->sk_rmem_alloc),
-			       net->ipv4.sysctl_tcp_rmem[2]));
+			   min(atomic_read(&sk->sk_rmem_alloc), rmem2));
 	}
 	if (atomic_read(&sk->sk_rmem_alloc) > sk->sk_rcvbuf)
 		tp->rcv_ssthresh = min(tp->window_clamp, 2U * tp->advmss);
@@ -745,7 +746,7 @@ void tcp_rcv_space_adjust(struct sock *sk)
 
 		do_div(rcvwin, tp->advmss);
 		rcvbuf = min_t(u64, rcvwin * rcvmem,
-			       sock_net(sk)->ipv4.sysctl_tcp_rmem[2]);
+			       READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_rmem[2]));
 		if (rcvbuf > sk->sk_rcvbuf) {
 			WRITE_ONCE(sk->sk_rcvbuf, rcvbuf);
 
@@ -910,9 +911,9 @@ static void tcp_update_pacing_rate(struct sock *sk)
 	 *	 end of slow start and should slow down.
 	 */
 	if (tcp_snd_cwnd(tp) < tp->snd_ssthresh / 2)
-		rate *= sock_net(sk)->ipv4.sysctl_tcp_pacing_ss_ratio;
+		rate *= READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_pacing_ss_ratio);
 	else
-		rate *= sock_net(sk)->ipv4.sysctl_tcp_pacing_ca_ratio;
+		rate *= READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_pacing_ca_ratio);
 
 	rate *= max(tcp_snd_cwnd(tp), tp->packets_out);
 
@@ -5520,7 +5521,7 @@ send_now:
 	}
 
 	if (!tcp_is_sack(tp) ||
-	    tp->compressed_ack >= sock_net(sk)->ipv4.sysctl_tcp_comp_sack_nr)
+	    tp->compressed_ack >= READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_comp_sack_nr))
 		goto send_now;
 
 	if (tp->compressed_ack_rcv_nxt != tp->rcv_nxt) {
@@ -5541,11 +5542,12 @@ send_now:
 	if (tp->srtt_us && tp->srtt_us < rtt)
 		rtt = tp->srtt_us;
 
-	delay = min_t(unsigned long, sock_net(sk)->ipv4.sysctl_tcp_comp_sack_delay_ns,
+	delay = min_t(unsigned long,
+		      READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_comp_sack_delay_ns),
 		      rtt * (NSEC_PER_USEC >> 3)/20);
 	sock_hold(sk);
 	hrtimer_start_range_ns(&tp->compressed_ack_timer, ns_to_ktime(delay),
-			       sock_net(sk)->ipv4.sysctl_tcp_comp_sack_slack_ns,
+			       READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_comp_sack_slack_ns),
 			       HRTIMER_MODE_REL_PINNED_SOFT);
 }
 
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index d16e6e40f47ba4..586c102ce152d0 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -1006,7 +1006,7 @@ static int tcp_v4_send_synack(const struct sock *sk, struct dst_entry *dst,
 	if (skb) {
 		__tcp_v4_send_check(skb, ireq->ir_loc_addr, ireq->ir_rmt_addr);
 
-		tos = sock_net(sk)->ipv4.sysctl_tcp_reflect_tos ?
+		tos = READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_reflect_tos) ?
 				(tcp_rsk(req)->syn_tos & ~INET_ECN_MASK) |
 				(inet_sk(sk)->tos & INET_ECN_MASK) :
 				inet_sk(sk)->tos;
@@ -1526,7 +1526,7 @@ struct sock *tcp_v4_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
 	/* Set ToS of the new socket based upon the value of incoming SYN.
 	 * ECT bits are set later in tcp_init_transfer().
 	 */
-	if (sock_net(sk)->ipv4.sysctl_tcp_reflect_tos)
+	if (READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_reflect_tos))
 		newinet->tos = tcp_rsk(req)->syn_tos & ~INET_ECN_MASK;
 
 	if (!dst) {
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 2efe41c84ee8a3..4c376b6d87649c 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -238,7 +238,7 @@ void tcp_select_initial_window(const struct sock *sk, int __space, __u32 mss,
 	*rcv_wscale = 0;
 	if (wscale_ok) {
 		/* Set window scaling on max possible window */
-		space = max_t(u32, space, sock_net(sk)->ipv4.sysctl_tcp_rmem[2]);
+		space = max_t(u32, space, READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_rmem[2]));
 		space = max_t(u32, space, sysctl_rmem_max);
 		space = min_t(u32, space, *window_clamp);
 		*rcv_wscale = clamp_t(int, ilog2(space) - 15,
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 9d3ede2932582e..be09941fe6d9a8 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -546,7 +546,7 @@ static int tcp_v6_send_synack(const struct sock *sk, struct dst_entry *dst,
 		if (np->repflow && ireq->pktopts)
 			fl6->flowlabel = ip6_flowlabel(ipv6_hdr(ireq->pktopts));
 
-		tclass = sock_net(sk)->ipv4.sysctl_tcp_reflect_tos ?
+		tclass = READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_reflect_tos) ?
 				(tcp_rsk(req)->syn_tos & ~INET_ECN_MASK) |
 				(np->tclass & INET_ECN_MASK) :
 				np->tclass;
@@ -1314,7 +1314,7 @@ static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk, struct sk_buff *
 	/* Set ToS of the new socket based upon the value of incoming SYN.
 	 * ECT bits are set later in tcp_init_transfer().
 	 */
-	if (sock_net(sk)->ipv4.sysctl_tcp_reflect_tos)
+	if (READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_reflect_tos))
 		newnp->tclass = tcp_rsk(req)->syn_tos & ~INET_ECN_MASK;
 
 	/* Clone native IPv6 options from listening socket (if any)
diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 9bbd8cbe0acb04..7e1518bb6115d5 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -1926,7 +1926,7 @@ static void mptcp_rcv_space_adjust(struct mptcp_sock *msk, int copied)
 
 		do_div(rcvwin, advmss);
 		rcvbuf = min_t(u64, rcvwin * rcvmem,
-			       sock_net(sk)->ipv4.sysctl_tcp_rmem[2]);
+			       READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_rmem[2]));
 
 		if (rcvbuf > sk->sk_rcvbuf) {
 			u32 window_clamp;
@@ -2669,8 +2669,8 @@ static int mptcp_init_sock(struct sock *sk)
 	mptcp_ca_reset(sk);
 
 	sk_sockets_allocated_inc(sk);
-	sk->sk_rcvbuf = sock_net(sk)->ipv4.sysctl_tcp_rmem[1];
-	sk->sk_sndbuf = sock_net(sk)->ipv4.sysctl_tcp_wmem[1];
+	sk->sk_rcvbuf = READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_rmem[1]);
+	sk->sk_sndbuf = READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_wmem[1]);
 
 	return 0;
 }
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index 43509c7e90fc28..f1c3b8eb4b3d33 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -517,7 +517,7 @@ static int tipc_sk_create(struct net *net, struct socket *sock,
 	timer_setup(&sk->sk_timer, tipc_sk_timeout, 0);
 	sk->sk_shutdown = 0;
 	sk->sk_backlog_rcv = tipc_sk_backlog_rcv;
-	sk->sk_rcvbuf = sysctl_tipc_rmem[1];
+	sk->sk_rcvbuf = READ_ONCE(sysctl_tipc_rmem[1]);
 	sk->sk_data_ready = tipc_data_ready;
 	sk->sk_write_space = tipc_write_space;
 	sk->sk_destruct = tipc_sock_destruct;