From 8281b7ec5c56b71cb2cc5a1728b41607be66959c Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Wed, 13 Jul 2022 13:51:51 -0700
Subject: ip: Fix data-races around sysctl_ip_default_ttl.

While reading sysctl_ip_default_ttl, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its readers.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/route.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'include')

diff --git a/include/net/route.h b/include/net/route.h
index 991a3985712d..bbcf2aba149f 100644
--- a/include/net/route.h
+++ b/include/net/route.h
@@ -373,7 +373,7 @@ static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
 	struct net *net = dev_net(dst->dev);
 
 	if (hoplimit == 0)
-		hoplimit = net->ipv4.sysctl_ip_default_ttl;
+		hoplimit = READ_ONCE(net->ipv4.sysctl_ip_default_ttl);
 	return hoplimit;
 }
 
-- 
cgit v1.3-18-gd494


From 60c158dc7b1f0558f6cadd5b50d0386da0000d50 Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Wed, 13 Jul 2022 13:51:53 -0700
Subject: ip: Fix data-races around sysctl_ip_fwd_use_pmtu.

While reading sysctl_ip_fwd_use_pmtu, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its readers.

Fixes: f87c10a8aa1e ("ipv4: introduce ip_dst_mtu_maybe_forward and protect forwarding path against pmtu spoofing")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/ip.h | 2 +-
 net/ipv4/route.c | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

(limited to 'include')

diff --git a/include/net/ip.h b/include/net/ip.h
index 26fffda78cca..05fe313f72fa 100644
--- a/include/net/ip.h
+++ b/include/net/ip.h
@@ -446,7 +446,7 @@ static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
 	struct net *net = dev_net(dst->dev);
 	unsigned int mtu;
 
-	if (net->ipv4.sysctl_ip_fwd_use_pmtu ||
+	if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) ||
 	    ip_mtu_locked(dst) ||
 	    !forwarding) {
 		mtu = rt->rt_pmtu;
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 356f535f3443..91c4f60de75a 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -1398,7 +1398,7 @@ u32 ip_mtu_from_fib_result(struct fib_result *res, __be32 daddr)
 	struct fib_info *fi = res->fi;
 	u32 mtu = 0;
 
-	if (dev_net(dev)->ipv4.sysctl_ip_fwd_use_pmtu ||
+	if (READ_ONCE(dev_net(dev)->ipv4.sysctl_ip_fwd_use_pmtu) ||
 	    fi->fib_metrics->metrics[RTAX_LOCK - 1] & (1 << RTAX_MTU))
 		mtu = fi->fib_mtu;
 
-- 
cgit v1.3-18-gd494


From 289d3b21fb0bfc94c4e98f10635bba1824e5f83c Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Wed, 13 Jul 2022 13:51:55 -0700
Subject: ip: Fix data-races around sysctl_ip_nonlocal_bind.

While reading sysctl_ip_nonlocal_bind, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its readers.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/inet_sock.h | 2 +-
 net/sctp/protocol.c     | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

(limited to 'include')

diff --git a/include/net/inet_sock.h b/include/net/inet_sock.h
index daead5fb389a..68d337775564 100644
--- a/include/net/inet_sock.h
+++ b/include/net/inet_sock.h
@@ -374,7 +374,7 @@ static inline bool inet_get_convert_csum(struct sock *sk)
 static inline bool inet_can_nonlocal_bind(struct net *net,
 					  struct inet_sock *inet)
 {
-	return net->ipv4.sysctl_ip_nonlocal_bind ||
+	return READ_ONCE(net->ipv4.sysctl_ip_nonlocal_bind) ||
 		inet->freebind || inet->transparent;
 }
 
diff --git a/net/sctp/protocol.c b/net/sctp/protocol.c
index 35928fefae33..1a094b087d88 100644
--- a/net/sctp/protocol.c
+++ b/net/sctp/protocol.c
@@ -358,7 +358,7 @@ static int sctp_v4_available(union sctp_addr *addr, struct sctp_sock *sp)
 	if (addr->v4.sin_addr.s_addr != htonl(INADDR_ANY) &&
 	   ret != RTN_LOCAL &&
 	   !sp->inet.freebind &&
-	   !net->ipv4.sysctl_ip_nonlocal_bind)
+	    !READ_ONCE(net->ipv4.sysctl_ip_nonlocal_bind))
 		return 0;
 
 	if (ipv6_only_sock(sctp_opt2sk(sp)))
-- 
cgit v1.3-18-gd494


From 85d0b4dbd74b95cc492b1f4e34497d3f894f5d9a Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Wed, 13 Jul 2022 13:51:57 -0700
Subject: ip: Fix a data-race around sysctl_fwmark_reflect.

While reading sysctl_fwmark_reflect, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its reader.

Fixes: e110861f8609 ("net: add a sysctl to reflect the fwmark on replies")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/ip.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'include')

diff --git a/include/net/ip.h b/include/net/ip.h
index 05fe313f72fa..4a15b6bcb4b8 100644
--- a/include/net/ip.h
+++ b/include/net/ip.h
@@ -384,7 +384,7 @@ void ipfrag_init(void);
 void ip_static_sysctl_init(void);
 
 #define IP4_REPLY_MARK(net, mark) \
-	((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
+	(READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0)
 
 static inline bool ip_is_fragment(const struct iphdr *iph)
 {
-- 
cgit v1.3-18-gd494


From 1a0008f9df59451d0a17806c1ee1a19857032fa8 Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Wed, 13 Jul 2022 13:51:58 -0700
Subject: tcp/dccp: Fix a data-race around sysctl_tcp_fwmark_accept.

While reading sysctl_tcp_fwmark_accept, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its reader.

Fixes: 84f39b08d786 ("net: support marking accepting TCP sockets")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/inet_sock.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'include')

diff --git a/include/net/inet_sock.h b/include/net/inet_sock.h
index 68d337775564..b29108f0973a 100644
--- a/include/net/inet_sock.h
+++ b/include/net/inet_sock.h
@@ -107,7 +107,8 @@ static inline struct inet_request_sock *inet_rsk(const struct request_sock *sk)
 
 static inline u32 inet_request_mark(const struct sock *sk, struct sk_buff *skb)
 {
-	if (!sk->sk_mark && sock_net(sk)->ipv4.sysctl_tcp_fwmark_accept)
+	if (!sk->sk_mark &&
+	    READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_fwmark_accept))
 		return skb->mark;
 
 	return sk->sk_mark;
-- 
cgit v1.3-18-gd494


From 08a75f10679470552a3a443f9aefd1399604d31d Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Wed, 13 Jul 2022 13:51:59 -0700
Subject: tcp: Fix data-races around sysctl_tcp_l3mdev_accept.

While reading sysctl_tcp_l3mdev_accept, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its readers.

Fixes: 6dd9a14e92e5 ("net: Allow accepted sockets to be bound to l3mdev domain")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/inet_hashtables.h | 2 +-
 include/net/inet_sock.h       | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

(limited to 'include')

diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h
index ebfa3df6f8dc..fd6b510d114b 100644
--- a/include/net/inet_hashtables.h
+++ b/include/net/inet_hashtables.h
@@ -179,7 +179,7 @@ static inline bool inet_sk_bound_dev_eq(struct net *net, int bound_dev_if,
 					int dif, int sdif)
 {
 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
-	return inet_bound_dev_eq(!!net->ipv4.sysctl_tcp_l3mdev_accept,
+	return inet_bound_dev_eq(!!READ_ONCE(net->ipv4.sysctl_tcp_l3mdev_accept),
 				 bound_dev_if, dif, sdif);
 #else
 	return inet_bound_dev_eq(true, bound_dev_if, dif, sdif);
diff --git a/include/net/inet_sock.h b/include/net/inet_sock.h
index b29108f0973a..6395f6b9a5d2 100644
--- a/include/net/inet_sock.h
+++ b/include/net/inet_sock.h
@@ -121,7 +121,7 @@ static inline int inet_request_bound_dev_if(const struct sock *sk,
 #ifdef CONFIG_NET_L3_MASTER_DEV
 	struct net *net = sock_net(sk);
 
-	if (!bound_dev_if && net->ipv4.sysctl_tcp_l3mdev_accept)
+	if (!bound_dev_if && READ_ONCE(net->ipv4.sysctl_tcp_l3mdev_accept))
 		return l3mdev_master_ifindex_by_index(net, skb->skb_iif);
 #endif
 
@@ -133,7 +133,7 @@ static inline int inet_sk_bound_l3mdev(const struct sock *sk)
 #ifdef CONFIG_NET_L3_MASTER_DEV
 	struct net *net = sock_net(sk);
 
-	if (!net->ipv4.sysctl_tcp_l3mdev_accept)
+	if (!READ_ONCE(net->ipv4.sysctl_tcp_l3mdev_accept))
 		return l3mdev_master_ifindex_by_index(net,
 						      sk->sk_bound_dev_if);
 #endif
-- 
cgit v1.3-18-gd494


From 11052589cf5c0bab3b4884d423d5f60c38fcf25d Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Wed, 13 Jul 2022 10:52:07 -0700
Subject: tcp/udp: Make early_demux back namespacified.

Commit e21145a9871a ("ipv4: namespacify ip_early_demux sysctl knob") made
it possible to enable/disable early_demux on a per-netns basis.  Then, we
introduced two knobs, tcp_early_demux and udp_early_demux, to switch it for
TCP/UDP in commit dddb64bcb346 ("net: Add sysctl to toggle early demux for
tcp and udp").  However, the .proc_handler() was wrong and actually
disabled us from changing the behaviour in each netns.

We can execute early_demux if net.ipv4.ip_early_demux is on and each proto
.early_demux() handler is not NULL.  When we toggle (tcp|udp)_early_demux,
the change itself is saved in each netns variable, but the .early_demux()
handler is a global variable, so the handler is switched based on the
init_net's sysctl variable.  Thus, netns (tcp|udp)_early_demux knobs have
nothing to do with the logic.  Whether we CAN execute proto .early_demux()
is always decided by init_net's sysctl knob, and whether we DO it or not is
by each netns ip_early_demux knob.

This patch namespacifies (tcp|udp)_early_demux again.  For now, the users
of the .early_demux() handler are TCP and UDP only, and they are called
directly to avoid retpoline.  So, we can remove the .early_demux() handler
from inet6?_protos and need not dereference them in ip6?_rcv_finish_core().
If another proto needs .early_demux(), we can restore it at that time.

Fixes: dddb64bcb346 ("net: Add sysctl to toggle early demux for tcp and udp")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://lore.kernel.org/r/20220713175207.7727-1-kuniyu@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 include/net/protocol.h     |  4 ----
 include/net/tcp.h          |  2 +-
 include/net/udp.h          |  2 +-
 net/ipv4/af_inet.c         | 14 ++---------
 net/ipv4/ip_input.c        | 37 +++++++++++++++++------------
 net/ipv4/sysctl_net_ipv4.c | 59 ++--------------------------------------------
 net/ipv6/ip6_input.c       | 23 ++++++++++--------
 net/ipv6/tcp_ipv6.c        |  9 ++-----
 net/ipv6/udp.c             |  9 ++-----
 9 files changed, 45 insertions(+), 114 deletions(-)

(limited to 'include')

diff --git a/include/net/protocol.h b/include/net/protocol.h
index f51c06ae365f..6aef8cb11cc8 100644
--- a/include/net/protocol.h
+++ b/include/net/protocol.h
@@ -35,8 +35,6 @@
 
 /* This is used to register protocols. */
 struct net_protocol {
-	int			(*early_demux)(struct sk_buff *skb);
-	int			(*early_demux_handler)(struct sk_buff *skb);
 	int			(*handler)(struct sk_buff *skb);
 
 	/* This returns an error if we weren't able to handle the error. */
@@ -52,8 +50,6 @@ struct net_protocol {
 
 #if IS_ENABLED(CONFIG_IPV6)
 struct inet6_protocol {
-	void	(*early_demux)(struct sk_buff *skb);
-	void    (*early_demux_handler)(struct sk_buff *skb);
 	int	(*handler)(struct sk_buff *skb);
 
 	/* This returns an error if we weren't able to handle the error. */
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 1e99f5c61f84..1636c55e798b 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -932,7 +932,7 @@ extern const struct inet_connection_sock_af_ops ipv6_specific;
 
 INDIRECT_CALLABLE_DECLARE(void tcp_v6_send_check(struct sock *sk, struct sk_buff *skb));
 INDIRECT_CALLABLE_DECLARE(int tcp_v6_rcv(struct sk_buff *skb));
-INDIRECT_CALLABLE_DECLARE(void tcp_v6_early_demux(struct sk_buff *skb));
+void tcp_v6_early_demux(struct sk_buff *skb);
 
 #endif
 
diff --git a/include/net/udp.h b/include/net/udp.h
index b83a00330566..bb4c227299cc 100644
--- a/include/net/udp.h
+++ b/include/net/udp.h
@@ -167,7 +167,7 @@ static inline void udp_csum_pull_header(struct sk_buff *skb)
 typedef struct sock *(*udp_lookup_t)(const struct sk_buff *skb, __be16 sport,
 				     __be16 dport);
 
-INDIRECT_CALLABLE_DECLARE(void udp_v6_early_demux(struct sk_buff *));
+void udp_v6_early_demux(struct sk_buff *skb);
 INDIRECT_CALLABLE_DECLARE(int udpv6_rcv(struct sk_buff *));
 
 struct sk_buff *__udp_gso_segment(struct sk_buff *gso_skb,
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index 4bc24f9e38b3..8baef2f3fc4b 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -1710,24 +1710,14 @@ static const struct net_protocol igmp_protocol = {
 };
 #endif
 
-/* thinking of making this const? Don't.
- * early_demux can change based on sysctl.
- */
-static struct net_protocol tcp_protocol = {
-	.early_demux	=	tcp_v4_early_demux,
-	.early_demux_handler =  tcp_v4_early_demux,
+static const struct net_protocol tcp_protocol = {
 	.handler	=	tcp_v4_rcv,
 	.err_handler	=	tcp_v4_err,
 	.no_policy	=	1,
 	.icmp_strict_tag_validation = 1,
 };
 
-/* thinking of making this const? Don't.
- * early_demux can change based on sysctl.
- */
-static struct net_protocol udp_protocol = {
-	.early_demux =	udp_v4_early_demux,
-	.early_demux_handler =	udp_v4_early_demux,
+static const struct net_protocol udp_protocol = {
 	.handler =	udp_rcv,
 	.err_handler =	udp_err,
 	.no_policy =	1,
diff --git a/net/ipv4/ip_input.c b/net/ipv4/ip_input.c
index b1165f717cd1..1b512390b3cf 100644
--- a/net/ipv4/ip_input.c
+++ b/net/ipv4/ip_input.c
@@ -312,14 +312,13 @@ static bool ip_can_use_hint(const struct sk_buff *skb, const struct iphdr *iph,
 	       ip_hdr(hint)->tos == iph->tos;
 }
 
-INDIRECT_CALLABLE_DECLARE(int udp_v4_early_demux(struct sk_buff *));
-INDIRECT_CALLABLE_DECLARE(int tcp_v4_early_demux(struct sk_buff *));
+int tcp_v4_early_demux(struct sk_buff *skb);
+int udp_v4_early_demux(struct sk_buff *skb);
 static int ip_rcv_finish_core(struct net *net, struct sock *sk,
 			      struct sk_buff *skb, struct net_device *dev,
 			      const struct sk_buff *hint)
 {
 	const struct iphdr *iph = ip_hdr(skb);
-	int (*edemux)(struct sk_buff *skb);
 	int err, drop_reason;
 	struct rtable *rt;
 
@@ -332,21 +331,29 @@ static int ip_rcv_finish_core(struct net *net, struct sock *sk,
 			goto drop_error;
 	}
 
-	if (net->ipv4.sysctl_ip_early_demux &&
+	if (READ_ONCE(net->ipv4.sysctl_ip_early_demux) &&
 	    !skb_dst(skb) &&
 	    !skb->sk &&
 	    !ip_is_fragment(iph)) {
-		const struct net_protocol *ipprot;
-		int protocol = iph->protocol;
-
-		ipprot = rcu_dereference(inet_protos[protocol]);
-		if (ipprot && (edemux = READ_ONCE(ipprot->early_demux))) {
-			err = INDIRECT_CALL_2(edemux, tcp_v4_early_demux,
-					      udp_v4_early_demux, skb);
-			if (unlikely(err))
-				goto drop_error;
-			/* must reload iph, skb->head might have changed */
-			iph = ip_hdr(skb);
+		switch (iph->protocol) {
+		case IPPROTO_TCP:
+			if (READ_ONCE(net->ipv4.sysctl_tcp_early_demux)) {
+				tcp_v4_early_demux(skb);
+
+				/* must reload iph, skb->head might have changed */
+				iph = ip_hdr(skb);
+			}
+			break;
+		case IPPROTO_UDP:
+			if (READ_ONCE(net->ipv4.sysctl_udp_early_demux)) {
+				err = udp_v4_early_demux(skb);
+				if (unlikely(err))
+					goto drop_error;
+
+				/* must reload iph, skb->head might have changed */
+				iph = ip_hdr(skb);
+			}
+			break;
 		}
 	}
 
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 108fd86f2718..130e9c130311 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -350,61 +350,6 @@ bad_key:
 	return ret;
 }
 
-static void proc_configure_early_demux(int enabled, int protocol)
-{
-	struct net_protocol *ipprot;
-#if IS_ENABLED(CONFIG_IPV6)
-	struct inet6_protocol *ip6prot;
-#endif
-
-	rcu_read_lock();
-
-	ipprot = rcu_dereference(inet_protos[protocol]);
-	if (ipprot)
-		ipprot->early_demux = enabled ? ipprot->early_demux_handler :
-						NULL;
-
-#if IS_ENABLED(CONFIG_IPV6)
-	ip6prot = rcu_dereference(inet6_protos[protocol]);
-	if (ip6prot)
-		ip6prot->early_demux = enabled ? ip6prot->early_demux_handler :
-						 NULL;
-#endif
-	rcu_read_unlock();
-}
-
-static int proc_tcp_early_demux(struct ctl_table *table, int write,
-				void *buffer, size_t *lenp, loff_t *ppos)
-{
-	int ret = 0;
-
-	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
-
-	if (write && !ret) {
-		int enabled = init_net.ipv4.sysctl_tcp_early_demux;
-
-		proc_configure_early_demux(enabled, IPPROTO_TCP);
-	}
-
-	return ret;
-}
-
-static int proc_udp_early_demux(struct ctl_table *table, int write,
-				void *buffer, size_t *lenp, loff_t *ppos)
-{
-	int ret = 0;
-
-	ret = proc_dou8vec_minmax(table, write, buffer, lenp, ppos);
-
-	if (write && !ret) {
-		int enabled = init_net.ipv4.sysctl_udp_early_demux;
-
-		proc_configure_early_demux(enabled, IPPROTO_UDP);
-	}
-
-	return ret;
-}
-
 static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
 					     int write, void *buffer,
 					     size_t *lenp, loff_t *ppos)
@@ -707,14 +652,14 @@ static struct ctl_table ipv4_net_table[] = {
 		.data           = &init_net.ipv4.sysctl_udp_early_demux,
 		.maxlen         = sizeof(u8),
 		.mode           = 0644,
-		.proc_handler   = proc_udp_early_demux
+		.proc_handler   = proc_dou8vec_minmax,
 	},
 	{
 		.procname       = "tcp_early_demux",
 		.data           = &init_net.ipv4.sysctl_tcp_early_demux,
 		.maxlen         = sizeof(u8),
 		.mode           = 0644,
-		.proc_handler   = proc_tcp_early_demux
+		.proc_handler   = proc_dou8vec_minmax,
 	},
 	{
 		.procname       = "nexthop_compat_mode",
diff --git a/net/ipv6/ip6_input.c b/net/ipv6/ip6_input.c
index 0322cc86b84e..e1ebf5e42ebe 100644
--- a/net/ipv6/ip6_input.c
+++ b/net/ipv6/ip6_input.c
@@ -45,20 +45,23 @@
 #include <net/inet_ecn.h>
 #include <net/dst_metadata.h>
 
-INDIRECT_CALLABLE_DECLARE(void tcp_v6_early_demux(struct sk_buff *));
 static void ip6_rcv_finish_core(struct net *net, struct sock *sk,
 				struct sk_buff *skb)
 {
-	void (*edemux)(struct sk_buff *skb);
-
-	if (net->ipv4.sysctl_ip_early_demux && !skb_dst(skb) && skb->sk == NULL) {
-		const struct inet6_protocol *ipprot;
-
-		ipprot = rcu_dereference(inet6_protos[ipv6_hdr(skb)->nexthdr]);
-		if (ipprot && (edemux = READ_ONCE(ipprot->early_demux)))
-			INDIRECT_CALL_2(edemux, tcp_v6_early_demux,
-					udp_v6_early_demux, skb);
+	if (READ_ONCE(net->ipv4.sysctl_ip_early_demux) &&
+	    !skb_dst(skb) && !skb->sk) {
+		switch (ipv6_hdr(skb)->nexthdr) {
+		case IPPROTO_TCP:
+			if (READ_ONCE(net->ipv4.sysctl_tcp_early_demux))
+				tcp_v6_early_demux(skb);
+			break;
+		case IPPROTO_UDP:
+			if (READ_ONCE(net->ipv4.sysctl_udp_early_demux))
+				udp_v6_early_demux(skb);
+			break;
+		}
 	}
+
 	if (!skb_valid_dst(skb))
 		ip6_route_input(skb);
 }
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index f37dd4aa91c6..9d3ede293258 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -1822,7 +1822,7 @@ do_time_wait:
 	goto discard_it;
 }
 
-INDIRECT_CALLABLE_SCOPE void tcp_v6_early_demux(struct sk_buff *skb)
+void tcp_v6_early_demux(struct sk_buff *skb)
 {
 	const struct ipv6hdr *hdr;
 	const struct tcphdr *th;
@@ -2176,12 +2176,7 @@ struct proto tcpv6_prot = {
 };
 EXPORT_SYMBOL_GPL(tcpv6_prot);
 
-/* thinking of making this const? Don't.
- * early_demux can change based on sysctl.
- */
-static struct inet6_protocol tcpv6_protocol = {
-	.early_demux	=	tcp_v6_early_demux,
-	.early_demux_handler =  tcp_v6_early_demux,
+static const struct inet6_protocol tcpv6_protocol = {
 	.handler	=	tcp_v6_rcv,
 	.err_handler	=	tcp_v6_err,
 	.flags		=	INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL,
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index 55afd7f39c04..e2f2e087a753 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -1052,7 +1052,7 @@ static struct sock *__udp6_lib_demux_lookup(struct net *net,
 	return NULL;
 }
 
-INDIRECT_CALLABLE_SCOPE void udp_v6_early_demux(struct sk_buff *skb)
+void udp_v6_early_demux(struct sk_buff *skb)
 {
 	struct net *net = dev_net(skb->dev);
 	const struct udphdr *uh;
@@ -1660,12 +1660,7 @@ int udpv6_getsockopt(struct sock *sk, int level, int optname,
 	return ipv6_getsockopt(sk, level, optname, optval, optlen);
 }
 
-/* thinking of making this const? Don't.
- * early_demux can change based on sysctl.
- */
-static struct inet6_protocol udpv6_protocol = {
-	.early_demux	=	udp_v6_early_demux,
-	.early_demux_handler =  udp_v6_early_demux,
+static const struct inet6_protocol udpv6_protocol = {
 	.handler	=	udpv6_rcv,
 	.err_handler	=	udpv6_err,
 	.flags		=	INET6_PROTO_NOPOLICY|INET6_PROTO_FINAL,
-- 
cgit v1.3-18-gd494


From 76c16d3e19446deea98b7883f261758b96b8781a Mon Sep 17 00:00:00 2001
From: Wong Vee Khee <vee.khee.wong@linux.intel.com>
Date: Thu, 14 Jul 2022 15:54:27 +0800
Subject: net: stmmac: switch to use interrupt for hw crosstimestamping

Using current implementation of polling mode, there is high chances we
will hit into timeout error when running phc2sys. Hence, update the
implementation of hardware crosstimestamping to use the MAC interrupt
service routine instead of polling for TSIS bit in the MAC Timestamp
Interrupt Status register to be set.

Cc: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Wong Vee Khee <vee.khee.wong@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 drivers/net/ethernet/stmicro/stmmac/dwmac-intel.c  | 25 ++++++++++++++--------
 drivers/net/ethernet/stmicro/stmmac/dwmac4.h       |  3 ++-
 drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c  |  4 ++++
 drivers/net/ethernet/stmicro/stmmac/stmmac.h       |  1 +
 .../net/ethernet/stmicro/stmmac/stmmac_hwtstamp.c  |  5 +++++
 drivers/net/ethernet/stmicro/stmmac/stmmac_ptp.c   | 12 +----------
 include/linux/stmmac.h                             |  1 +
 7 files changed, 30 insertions(+), 21 deletions(-)

(limited to 'include')

diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-intel.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-intel.c
index 38fe77d1035e..3fe720c5dc9f 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac-intel.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-intel.c
@@ -298,6 +298,11 @@ static void get_arttime(struct mii_bus *mii, int intel_adhoc_addr,
 	*art_time = ns;
 }
 
+static int stmmac_cross_ts_isr(struct stmmac_priv *priv)
+{
+	return (readl(priv->ioaddr + GMAC_INT_STATUS) & GMAC_INT_TSIE);
+}
+
 static int intel_crosststamp(ktime_t *device,
 			     struct system_counterval_t *system,
 			     void *ctx)
@@ -313,8 +318,6 @@ static int intel_crosststamp(ktime_t *device,
 	u32 num_snapshot;
 	u32 gpio_value;
 	u32 acr_value;
-	int ret;
-	u32 v;
 	int i;
 
 	if (!boot_cpu_has(X86_FEATURE_ART))
@@ -328,6 +331,8 @@ static int intel_crosststamp(ktime_t *device,
 	if (priv->plat->ext_snapshot_en)
 		return -EBUSY;
 
+	priv->plat->int_snapshot_en = 1;
+
 	mutex_lock(&priv->aux_ts_lock);
 	/* Enable Internal snapshot trigger */
 	acr_value = readl(ptpaddr + PTP_ACR);
@@ -347,6 +352,7 @@ static int intel_crosststamp(ktime_t *device,
 		break;
 	default:
 		mutex_unlock(&priv->aux_ts_lock);
+		priv->plat->int_snapshot_en = 0;
 		return -EINVAL;
 	}
 	writel(acr_value, ptpaddr + PTP_ACR);
@@ -368,13 +374,12 @@ static int intel_crosststamp(ktime_t *device,
 	gpio_value |= GMAC_GPO1;
 	writel(gpio_value, ioaddr + GMAC_GPIO_STATUS);
 
-	/* Poll for time sync operation done */
-	ret = readl_poll_timeout(priv->ioaddr + GMAC_INT_STATUS, v,
-				 (v & GMAC_INT_TSIE), 100, 10000);
-
-	if (ret == -ETIMEDOUT) {
-		pr_err("%s: Wait for time sync operation timeout\n", __func__);
-		return ret;
+	/* Time sync done Indication - Interrupt method */
+	if (!wait_event_interruptible_timeout(priv->tstamp_busy_wait,
+					      stmmac_cross_ts_isr(priv),
+					      HZ / 100)) {
+		priv->plat->int_snapshot_en = 0;
+		return -ETIMEDOUT;
 	}
 
 	num_snapshot = (readl(ioaddr + GMAC_TIMESTAMP_STATUS) &
@@ -392,6 +397,7 @@ static int intel_crosststamp(ktime_t *device,
 	}
 
 	system->cycles *= intel_priv->crossts_adj;
+	priv->plat->int_snapshot_en = 0;
 
 	return 0;
 }
@@ -576,6 +582,7 @@ static int intel_mgbe_common_data(struct pci_dev *pdev,
 
 	plat->has_crossts = true;
 	plat->crosststamp = intel_crosststamp;
+	plat->int_snapshot_en = 0;
 
 	/* Setup MSI vector offset specific to Intel mGbE controller */
 	plat->msi_mac_vec = 29;
diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac4.h b/drivers/net/ethernet/stmicro/stmmac/dwmac4.h
index 462ca7ed095a..71dad409f78b 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac4.h
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac4.h
@@ -150,7 +150,8 @@
 #define	GMAC_PCS_IRQ_DEFAULT	(GMAC_INT_RGSMIIS | GMAC_INT_PCS_LINK |	\
 				 GMAC_INT_PCS_ANE)
 
-#define	GMAC_INT_DEFAULT_ENABLE	(GMAC_INT_PMT_EN | GMAC_INT_LPI_EN)
+#define	GMAC_INT_DEFAULT_ENABLE	(GMAC_INT_PMT_EN | GMAC_INT_LPI_EN | \
+				 GMAC_INT_TSIE)
 
 enum dwmac4_irq_status {
 	time_stamp_irq = 0x00001000,
diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c b/drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c
index fd41db65fe1d..d5299dd13e85 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c
@@ -23,6 +23,7 @@
 static void dwmac4_core_init(struct mac_device_info *hw,
 			     struct net_device *dev)
 {
+	struct stmmac_priv *priv = netdev_priv(dev);
 	void __iomem *ioaddr = hw->pcsr;
 	u32 value = readl(ioaddr + GMAC_CONFIG);
 
@@ -58,6 +59,9 @@ static void dwmac4_core_init(struct mac_device_info *hw,
 		value |= GMAC_INT_FPE_EN;
 
 	writel(value, ioaddr + GMAC_INT_EN);
+
+	if (GMAC_INT_DEFAULT_ENABLE & GMAC_INT_TSIE)
+		init_waitqueue_head(&priv->tstamp_busy_wait);
 }
 
 static void dwmac4_rx_queue_enable(struct mac_device_info *hw,
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
index 57970ae2178d..f9e83964aa7e 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
@@ -266,6 +266,7 @@ struct stmmac_priv {
 	rwlock_t ptp_lock;
 	/* Protects auxiliary snapshot registers from concurrent access. */
 	struct mutex aux_ts_lock;
+	wait_queue_head_t tstamp_busy_wait;
 
 	void __iomem *mmcaddr;
 	void __iomem *ptpaddr;
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_hwtstamp.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_hwtstamp.c
index 92d32940aff0..764832f4dae1 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_hwtstamp.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_hwtstamp.c
@@ -179,6 +179,11 @@ static void timestamp_interrupt(struct stmmac_priv *priv)
 	u64 ptp_time;
 	int i;
 
+	if (priv->plat->int_snapshot_en) {
+		wake_up(&priv->tstamp_busy_wait);
+		return;
+	}
+
 	tsync_int = readl(priv->ioaddr + GMAC_INT_STATUS) & GMAC_INT_TSIE;
 
 	if (!tsync_int)
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_ptp.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_ptp.c
index e45fb191d8e6..4d11980dcd64 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_ptp.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_ptp.c
@@ -175,11 +175,10 @@ static int stmmac_enable(struct ptp_clock_info *ptp,
 	struct stmmac_priv *priv =
 	    container_of(ptp, struct stmmac_priv, ptp_clock_ops);
 	void __iomem *ptpaddr = priv->ptpaddr;
-	void __iomem *ioaddr = priv->hw->pcsr;
 	struct stmmac_pps_cfg *cfg;
-	u32 intr_value, acr_value;
 	int ret = -EOPNOTSUPP;
 	unsigned long flags;
+	u32 acr_value;
 
 	switch (rq->type) {
 	case PTP_CLK_REQ_PEROUT:
@@ -213,19 +212,10 @@ static int stmmac_enable(struct ptp_clock_info *ptp,
 			netdev_dbg(priv->dev, "Auxiliary Snapshot %d enabled.\n",
 				   priv->plat->ext_snapshot_num >>
 				   PTP_ACR_ATSEN_SHIFT);
-			/* Enable Timestamp Interrupt */
-			intr_value = readl(ioaddr + GMAC_INT_EN);
-			intr_value |= GMAC_INT_TSIE;
-			writel(intr_value, ioaddr + GMAC_INT_EN);
-
 		} else {
 			netdev_dbg(priv->dev, "Auxiliary Snapshot %d disabled.\n",
 				   priv->plat->ext_snapshot_num >>
 				   PTP_ACR_ATSEN_SHIFT);
-			/* Disable Timestamp Interrupt */
-			intr_value = readl(ioaddr + GMAC_INT_EN);
-			intr_value &= ~GMAC_INT_TSIE;
-			writel(intr_value, ioaddr + GMAC_INT_EN);
 		}
 		writel(acr_value, ptpaddr + PTP_ACR);
 		mutex_unlock(&priv->aux_ts_lock);
diff --git a/include/linux/stmmac.h b/include/linux/stmmac.h
index 29917850f079..8df475db88c0 100644
--- a/include/linux/stmmac.h
+++ b/include/linux/stmmac.h
@@ -260,6 +260,7 @@ struct plat_stmmacenet_data {
 	bool has_crossts;
 	int int_snapshot_num;
 	int ext_snapshot_num;
+	bool int_snapshot_en;
 	bool ext_snapshot_en;
 	bool multi_msi_en;
 	int msi_mac_vec;
-- 
cgit v1.3-18-gd494


From f2f316e287e6c2e3a1c5bab8d9b77ee03daa0463 Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Fri, 15 Jul 2022 10:17:45 -0700
Subject: tcp: Fix data-races around keepalive sysctl knobs.

While reading sysctl_tcp_keepalive_(time|probes|intvl), they can be changed
concurrently.  Thus, we need to add READ_ONCE() to their readers.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/tcp.h | 9 ++++++---
 net/smc/smc_llc.c | 2 +-
 2 files changed, 7 insertions(+), 4 deletions(-)

(limited to 'include')

diff --git a/include/net/tcp.h b/include/net/tcp.h
index 1636c55e798b..204478d5d388 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -1493,21 +1493,24 @@ static inline int keepalive_intvl_when(const struct tcp_sock *tp)
 {
 	struct net *net = sock_net((struct sock *)tp);
 
-	return tp->keepalive_intvl ? : net->ipv4.sysctl_tcp_keepalive_intvl;
+	return tp->keepalive_intvl ? :
+		READ_ONCE(net->ipv4.sysctl_tcp_keepalive_intvl);
 }
 
 static inline int keepalive_time_when(const struct tcp_sock *tp)
 {
 	struct net *net = sock_net((struct sock *)tp);
 
-	return tp->keepalive_time ? : net->ipv4.sysctl_tcp_keepalive_time;
+	return tp->keepalive_time ? :
+		READ_ONCE(net->ipv4.sysctl_tcp_keepalive_time);
 }
 
 static inline int keepalive_probes(const struct tcp_sock *tp)
 {
 	struct net *net = sock_net((struct sock *)tp);
 
-	return tp->keepalive_probes ? : net->ipv4.sysctl_tcp_keepalive_probes;
+	return tp->keepalive_probes ? :
+		READ_ONCE(net->ipv4.sysctl_tcp_keepalive_probes);
 }
 
 static inline u32 keepalive_time_elapsed(const struct tcp_sock *tp)
diff --git a/net/smc/smc_llc.c b/net/smc/smc_llc.c
index c4d057b2941d..0bde36b56472 100644
--- a/net/smc/smc_llc.c
+++ b/net/smc/smc_llc.c
@@ -2122,7 +2122,7 @@ void smc_llc_lgr_init(struct smc_link_group *lgr, struct smc_sock *smc)
 	init_waitqueue_head(&lgr->llc_flow_waiter);
 	init_waitqueue_head(&lgr->llc_msg_waiter);
 	mutex_init(&lgr->llc_conf_mutex);
-	lgr->llc_testlink_time = net->ipv4.sysctl_tcp_keepalive_time;
+	lgr->llc_testlink_time = READ_ONCE(net->ipv4.sysctl_tcp_keepalive_time);
 }
 
 /* called after lgr was removed from lgr_list */
-- 
cgit v1.3-18-gd494


From 39e24435a776e9de5c6dd188836cf2523547804b Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Fri, 15 Jul 2022 10:17:50 -0700
Subject: tcp: Fix data-races around some timeout sysctl knobs.

While reading these sysctl knobs, they can be changed concurrently.
Thus, we need to add READ_ONCE() to their readers.

  - tcp_retries1
  - tcp_retries2
  - tcp_orphan_retries
  - tcp_fin_timeout

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/tcp.h     |  3 ++-
 net/ipv4/tcp.c        |  2 +-
 net/ipv4/tcp_output.c |  2 +-
 net/ipv4/tcp_timer.c  | 10 +++++-----
 4 files changed, 9 insertions(+), 8 deletions(-)

(limited to 'include')

diff --git a/include/net/tcp.h b/include/net/tcp.h
index 204478d5d388..23ccaa386746 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -1523,7 +1523,8 @@ static inline u32 keepalive_time_elapsed(const struct tcp_sock *tp)
 
 static inline int tcp_fin_time(const struct sock *sk)
 {
-	int fin_timeout = tcp_sk(sk)->linger2 ? : sock_net(sk)->ipv4.sysctl_tcp_fin_timeout;
+	int fin_timeout = tcp_sk(sk)->linger2 ? :
+		READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_fin_timeout);
 	const int rto = inet_csk(sk)->icsk_rto;
 
 	if (fin_timeout < (rto << 2) - (rto >> 1))
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index b3632fa5df07..b1b1bcbc4f60 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -3973,7 +3973,7 @@ static int do_tcp_getsockopt(struct sock *sk, int level,
 	case TCP_LINGER2:
 		val = tp->linger2;
 		if (val >= 0)
-			val = (val ? : net->ipv4.sysctl_tcp_fin_timeout) / HZ;
+			val = (val ? : READ_ONCE(net->ipv4.sysctl_tcp_fin_timeout)) / HZ;
 		break;
 	case TCP_DEFER_ACCEPT:
 		val = retrans_to_secs(icsk->icsk_accept_queue.rskq_defer_accept,
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 02ab3a9c6657..3b3552d292a5 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -4090,7 +4090,7 @@ void tcp_send_probe0(struct sock *sk)
 
 	icsk->icsk_probes_out++;
 	if (err <= 0) {
-		if (icsk->icsk_backoff < net->ipv4.sysctl_tcp_retries2)
+		if (icsk->icsk_backoff < READ_ONCE(net->ipv4.sysctl_tcp_retries2))
 			icsk->icsk_backoff++;
 		timeout = tcp_probe0_when(sk, TCP_RTO_MAX);
 	} else {
diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c
index a234704e8163..ec5277becc6a 100644
--- a/net/ipv4/tcp_timer.c
+++ b/net/ipv4/tcp_timer.c
@@ -143,7 +143,7 @@ static int tcp_out_of_resources(struct sock *sk, bool do_reset)
  */
 static int tcp_orphan_retries(struct sock *sk, bool alive)
 {
-	int retries = sock_net(sk)->ipv4.sysctl_tcp_orphan_retries; /* May be zero. */
+	int retries = READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_orphan_retries); /* May be zero. */
 
 	/* We know from an ICMP that something is wrong. */
 	if (sk->sk_err_soft && !alive)
@@ -243,14 +243,14 @@ static int tcp_write_timeout(struct sock *sk)
 			READ_ONCE(net->ipv4.sysctl_tcp_syn_retries);
 		expired = icsk->icsk_retransmits >= retry_until;
 	} else {
-		if (retransmits_timed_out(sk, net->ipv4.sysctl_tcp_retries1, 0)) {
+		if (retransmits_timed_out(sk, READ_ONCE(net->ipv4.sysctl_tcp_retries1), 0)) {
 			/* Black hole detection */
 			tcp_mtu_probing(icsk, sk);
 
 			__dst_negative_advice(sk);
 		}
 
-		retry_until = net->ipv4.sysctl_tcp_retries2;
+		retry_until = READ_ONCE(net->ipv4.sysctl_tcp_retries2);
 		if (sock_flag(sk, SOCK_DEAD)) {
 			const bool alive = icsk->icsk_rto < TCP_RTO_MAX;
 
@@ -381,7 +381,7 @@ static void tcp_probe_timer(struct sock *sk)
 		 msecs_to_jiffies(icsk->icsk_user_timeout))
 		goto abort;
 
-	max_probes = sock_net(sk)->ipv4.sysctl_tcp_retries2;
+	max_probes = READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_retries2);
 	if (sock_flag(sk, SOCK_DEAD)) {
 		const bool alive = inet_csk_rto_backoff(icsk, TCP_RTO_MAX) < TCP_RTO_MAX;
 
@@ -589,7 +589,7 @@ out_reset_timer:
 	}
 	inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
 				  tcp_clamp_rto_to_user_timeout(sk), TCP_RTO_MAX);
-	if (retransmits_timed_out(sk, net->ipv4.sysctl_tcp_retries1 + 1, 0))
+	if (retransmits_timed_out(sk, READ_ONCE(net->ipv4.sysctl_tcp_retries1) + 1, 0))
 		__sk_dst_reset(sk);
 
 out:;
-- 
cgit v1.3-18-gd494


From 55be873695ed8912eb77ff46d1d1cadf028bd0f3 Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Fri, 15 Jul 2022 10:17:51 -0700
Subject: tcp: Fix a data-race around sysctl_tcp_notsent_lowat.

While reading sysctl_tcp_notsent_lowat, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its reader.

Fixes: c9bee3b7fdec ("tcp: TCP_NOTSENT_LOWAT socket option")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/tcp.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'include')

diff --git a/include/net/tcp.h b/include/net/tcp.h
index 23ccaa386746..6ee1fb4fb292 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -2027,7 +2027,7 @@ void __tcp_v4_send_check(struct sk_buff *skb, __be32 saddr, __be32 daddr);
 static inline u32 tcp_notsent_lowat(const struct tcp_sock *tp)
 {
 	struct net *net = sock_net((struct sock *)tp);
-	return tp->notsent_lowat ?: net->ipv4.sysctl_tcp_notsent_lowat;
+	return tp->notsent_lowat ?: READ_ONCE(net->ipv4.sysctl_tcp_notsent_lowat);
 }
 
 bool tcp_stream_memory_free(const struct sock *sk, int wake);
-- 
cgit v1.3-18-gd494


From 30e22a6ebca039572ce9bc10f1934f4eabfb5b7f Mon Sep 17 00:00:00 2001
From: Taehee Yoo <ap420073@gmail.com>
Date: Sun, 17 Jul 2022 16:09:03 +0000
Subject: amt: use workqueue for gateway side message handling

There are some synchronization issues(amt->status, amt->req_cnt, etc)
if the interface is in gateway mode because gateway message handlers
are processed concurrently.
This applies a work queue for processing these messages instead of
expanding the locking context.

So, the purposes of this patch are to fix exist race conditions and to make
gateway to be able to validate a gateway status more correctly.

When the AMT gateway interface is created, it tries to establish to relay.
The establishment step looks stateless, but it should be managed well.
In order to handle messages in the gateway, it saves the current
status(i.e. AMT_STATUS_XXX).
This patch makes gateway code to be worked with a single thread.

Now, all messages except the multicast are triggered(received or
delay expired), and these messages will be stored in the event
queue(amt->events).
Then, the single worker processes stored messages asynchronously one
by one.
The multicast data message type will be still processed immediately.

Now, amt->lock is only needed to access the event queue(amt->events)
if an interface is the gateway mode.

Fixes: cbc21dc1cfe9 ("amt: add data plane of amt interface")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
---
 drivers/net/amt.c | 159 ++++++++++++++++++++++++++++++++++++++++++++++++------
 include/net/amt.h |  20 +++++++
 2 files changed, 164 insertions(+), 15 deletions(-)

(limited to 'include')

diff --git a/drivers/net/amt.c b/drivers/net/amt.c
index 89563d1b2a3b..9e2d1992b349 100644
--- a/drivers/net/amt.c
+++ b/drivers/net/amt.c
@@ -900,6 +900,28 @@ static void amt_send_mld_gq(struct amt_dev *amt, struct amt_tunnel_list *tunnel)
 }
 #endif
 
+static bool amt_queue_event(struct amt_dev *amt, enum amt_event event,
+			    struct sk_buff *skb)
+{
+	int index;
+
+	spin_lock_bh(&amt->lock);
+	if (amt->nr_events >= AMT_MAX_EVENTS) {
+		spin_unlock_bh(&amt->lock);
+		return 1;
+	}
+
+	index = (amt->event_idx + amt->nr_events) % AMT_MAX_EVENTS;
+	amt->events[index].event = event;
+	amt->events[index].skb = skb;
+	amt->nr_events++;
+	amt->event_idx %= AMT_MAX_EVENTS;
+	queue_work(amt_wq, &amt->event_wq);
+	spin_unlock_bh(&amt->lock);
+
+	return 0;
+}
+
 static void amt_secret_work(struct work_struct *work)
 {
 	struct amt_dev *amt = container_of(to_delayed_work(work),
@@ -913,12 +935,8 @@ static void amt_secret_work(struct work_struct *work)
 			 msecs_to_jiffies(AMT_SECRET_TIMEOUT));
 }
 
-static void amt_discovery_work(struct work_struct *work)
+static void amt_event_send_discovery(struct amt_dev *amt)
 {
-	struct amt_dev *amt = container_of(to_delayed_work(work),
-					   struct amt_dev,
-					   discovery_wq);
-
 	spin_lock_bh(&amt->lock);
 	if (amt->status > AMT_STATUS_SENT_DISCOVERY)
 		goto out;
@@ -933,11 +951,19 @@ out:
 	spin_unlock_bh(&amt->lock);
 }
 
-static void amt_req_work(struct work_struct *work)
+static void amt_discovery_work(struct work_struct *work)
 {
 	struct amt_dev *amt = container_of(to_delayed_work(work),
 					   struct amt_dev,
-					   req_wq);
+					   discovery_wq);
+
+	if (amt_queue_event(amt, AMT_EVENT_SEND_DISCOVERY, NULL))
+		mod_delayed_work(amt_wq, &amt->discovery_wq,
+				 msecs_to_jiffies(AMT_DISCOVERY_TIMEOUT));
+}
+
+static void amt_event_send_request(struct amt_dev *amt)
+{
 	u32 exp;
 
 	spin_lock_bh(&amt->lock);
@@ -967,6 +993,17 @@ out:
 	spin_unlock_bh(&amt->lock);
 }
 
+static void amt_req_work(struct work_struct *work)
+{
+	struct amt_dev *amt = container_of(to_delayed_work(work),
+					   struct amt_dev,
+					   req_wq);
+
+	if (amt_queue_event(amt, AMT_EVENT_SEND_REQUEST, NULL))
+		mod_delayed_work(amt_wq, &amt->req_wq,
+				 msecs_to_jiffies(100));
+}
+
 static bool amt_send_membership_update(struct amt_dev *amt,
 				       struct sk_buff *skb,
 				       bool v6)
@@ -2392,12 +2429,14 @@ static bool amt_membership_query_handler(struct amt_dev *amt,
 	skb->pkt_type = PACKET_MULTICAST;
 	skb->ip_summed = CHECKSUM_NONE;
 	len = skb->len;
+	local_bh_disable();
 	if (__netif_rx(skb) == NET_RX_SUCCESS) {
 		amt_update_gw_status(amt, AMT_STATUS_RECEIVED_QUERY, true);
 		dev_sw_netstats_rx_add(amt->dev, len);
 	} else {
 		amt->dev->stats.rx_dropped++;
 	}
+	local_bh_enable();
 
 	return false;
 }
@@ -2688,6 +2727,38 @@ send:
 	return false;
 }
 
+static void amt_gw_rcv(struct amt_dev *amt, struct sk_buff *skb)
+{
+	int type = amt_parse_type(skb);
+	int err = 1;
+
+	if (type == -1)
+		goto drop;
+
+	if (amt->mode == AMT_MODE_GATEWAY) {
+		switch (type) {
+		case AMT_MSG_ADVERTISEMENT:
+			err = amt_advertisement_handler(amt, skb);
+			break;
+		case AMT_MSG_MEMBERSHIP_QUERY:
+			err = amt_membership_query_handler(amt, skb);
+			if (!err)
+				return;
+			break;
+		default:
+			netdev_dbg(amt->dev, "Invalid type of Gateway\n");
+			break;
+		}
+	}
+drop:
+	if (err) {
+		amt->dev->stats.rx_dropped++;
+		kfree_skb(skb);
+	} else {
+		consume_skb(skb);
+	}
+}
+
 static int amt_rcv(struct sock *sk, struct sk_buff *skb)
 {
 	struct amt_dev *amt;
@@ -2719,8 +2790,12 @@ static int amt_rcv(struct sock *sk, struct sk_buff *skb)
 				err = true;
 				goto drop;
 			}
-			err = amt_advertisement_handler(amt, skb);
-			break;
+			if (amt_queue_event(amt, AMT_EVENT_RECEIVE, skb)) {
+				netdev_dbg(amt->dev, "AMT Event queue full\n");
+				err = true;
+				goto drop;
+			}
+			goto out;
 		case AMT_MSG_MULTICAST_DATA:
 			if (iph->saddr != amt->remote_ip) {
 				netdev_dbg(amt->dev, "Invalid Relay IP\n");
@@ -2738,11 +2813,12 @@ static int amt_rcv(struct sock *sk, struct sk_buff *skb)
 				err = true;
 				goto drop;
 			}
-			err = amt_membership_query_handler(amt, skb);
-			if (err)
+			if (amt_queue_event(amt, AMT_EVENT_RECEIVE, skb)) {
+				netdev_dbg(amt->dev, "AMT Event queue full\n");
+				err = true;
 				goto drop;
-			else
-				goto out;
+			}
+			goto out;
 		default:
 			err = true;
 			netdev_dbg(amt->dev, "Invalid type of Gateway\n");
@@ -2780,6 +2856,46 @@ out:
 	return 0;
 }
 
+static void amt_event_work(struct work_struct *work)
+{
+	struct amt_dev *amt = container_of(work, struct amt_dev, event_wq);
+	struct sk_buff *skb;
+	u8 event;
+	int i;
+
+	for (i = 0; i < AMT_MAX_EVENTS; i++) {
+		spin_lock_bh(&amt->lock);
+		if (amt->nr_events == 0) {
+			spin_unlock_bh(&amt->lock);
+			return;
+		}
+		event = amt->events[amt->event_idx].event;
+		skb = amt->events[amt->event_idx].skb;
+		amt->events[amt->event_idx].event = AMT_EVENT_NONE;
+		amt->events[amt->event_idx].skb = NULL;
+		amt->nr_events--;
+		amt->event_idx++;
+		amt->event_idx %= AMT_MAX_EVENTS;
+		spin_unlock_bh(&amt->lock);
+
+		switch (event) {
+		case AMT_EVENT_RECEIVE:
+			amt_gw_rcv(amt, skb);
+			break;
+		case AMT_EVENT_SEND_DISCOVERY:
+			amt_event_send_discovery(amt);
+			break;
+		case AMT_EVENT_SEND_REQUEST:
+			amt_event_send_request(amt);
+			break;
+		default:
+			if (skb)
+				kfree_skb(skb);
+			break;
+		}
+	}
+}
+
 static int amt_err_lookup(struct sock *sk, struct sk_buff *skb)
 {
 	struct amt_dev *amt;
@@ -2867,6 +2983,8 @@ static int amt_dev_open(struct net_device *dev)
 
 	amt->ready4 = false;
 	amt->ready6 = false;
+	amt->event_idx = 0;
+	amt->nr_events = 0;
 
 	err = amt_socket_create(amt);
 	if (err)
@@ -2892,6 +3010,8 @@ static int amt_dev_stop(struct net_device *dev)
 	struct amt_dev *amt = netdev_priv(dev);
 	struct amt_tunnel_list *tunnel, *tmp;
 	struct socket *sock;
+	struct sk_buff *skb;
+	int i;
 
 	cancel_delayed_work_sync(&amt->req_wq);
 	cancel_delayed_work_sync(&amt->discovery_wq);
@@ -2904,6 +3024,15 @@ static int amt_dev_stop(struct net_device *dev)
 	if (sock)
 		udp_tunnel_sock_release(sock);
 
+	cancel_work_sync(&amt->event_wq);
+	for (i = 0; i < AMT_MAX_EVENTS; i++) {
+		skb = amt->events[i].skb;
+		if (skb)
+			kfree_skb(skb);
+		amt->events[i].event = AMT_EVENT_NONE;
+		amt->events[i].skb = NULL;
+	}
+
 	amt->ready4 = false;
 	amt->ready6 = false;
 	amt->req_cnt = 0;
@@ -3146,8 +3275,8 @@ static int amt_newlink(struct net *net, struct net_device *dev,
 	INIT_DELAYED_WORK(&amt->discovery_wq, amt_discovery_work);
 	INIT_DELAYED_WORK(&amt->req_wq, amt_req_work);
 	INIT_DELAYED_WORK(&amt->secret_wq, amt_secret_work);
+	INIT_WORK(&amt->event_wq, amt_event_work);
 	INIT_LIST_HEAD(&amt->tunnel_list);
-
 	return 0;
 err:
 	dev_put(amt->stream_dev);
@@ -3280,7 +3409,7 @@ static int __init amt_init(void)
 	if (err < 0)
 		goto unregister_notifier;
 
-	amt_wq = alloc_workqueue("amt", WQ_UNBOUND, 1);
+	amt_wq = alloc_workqueue("amt", WQ_UNBOUND, 0);
 	if (!amt_wq) {
 		err = -ENOMEM;
 		goto rtnl_unregister;
diff --git a/include/net/amt.h b/include/net/amt.h
index 0e40c3d64fcf..08fc30cf2f34 100644
--- a/include/net/amt.h
+++ b/include/net/amt.h
@@ -78,6 +78,15 @@ enum amt_status {
 
 #define AMT_STATUS_MAX (__AMT_STATUS_MAX - 1)
 
+/* Gateway events only */
+enum amt_event {
+	AMT_EVENT_NONE,
+	AMT_EVENT_RECEIVE,
+	AMT_EVENT_SEND_DISCOVERY,
+	AMT_EVENT_SEND_REQUEST,
+	__AMT_EVENT_MAX,
+};
+
 struct amt_header {
 #if defined(__LITTLE_ENDIAN_BITFIELD)
 	u8 type:4,
@@ -292,6 +301,12 @@ struct amt_group_node {
 	struct hlist_head	sources[];
 };
 
+#define AMT_MAX_EVENTS	16
+struct amt_events {
+	enum amt_event event;
+	struct sk_buff *skb;
+};
+
 struct amt_dev {
 	struct net_device       *dev;
 	struct net_device       *stream_dev;
@@ -308,6 +323,7 @@ struct amt_dev {
 	struct delayed_work     req_wq;
 	/* Protected by RTNL */
 	struct delayed_work     secret_wq;
+	struct work_struct	event_wq;
 	/* AMT status */
 	enum amt_status		status;
 	/* Generated key */
@@ -345,6 +361,10 @@ struct amt_dev {
 	/* Used only in gateway mode */
 	u64			mac:48,
 				reserved:16;
+	/* AMT gateway side message handler queue */
+	struct amt_events	events[AMT_MAX_EVENTS];
+	u8			event_idx;
+	u8			nr_events;
 };
 
 #define AMT_TOS			0xc0
-- 
cgit v1.3-18-gd494


From 9b55c20f83369dd54541d9ddbe3a018a8377f451 Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Mon, 18 Jul 2022 10:26:42 -0700
Subject: ip: Fix data-races around sysctl_ip_prot_sock.

sysctl_ip_prot_sock is accessed concurrently, and there is always a chance
of data-race.  So, all readers and writers need some basic protection to
avoid load/store-tearing.

Fixes: 4548b683b781 ("Introduce a sysctl that modifies the value of PROT_SOCK.")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/ip.h           | 2 +-
 net/ipv4/sysctl_net_ipv4.c | 6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)

(limited to 'include')

diff --git a/include/net/ip.h b/include/net/ip.h
index 4a15b6bcb4b8..1c979fd1904c 100644
--- a/include/net/ip.h
+++ b/include/net/ip.h
@@ -357,7 +357,7 @@ static inline bool sysctl_dev_name_is_allowed(const char *name)
 
 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
 {
-	return port < net->ipv4.sysctl_ip_prot_sock;
+	return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
 }
 
 #else
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 130e9c130311..5490c285668b 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -84,7 +84,7 @@ static int ipv4_local_port_range(struct ctl_table *table, int write,
 		 * port limit.
 		 */
 		if ((range[1] < range[0]) ||
-		    (range[0] < net->ipv4.sysctl_ip_prot_sock))
+		    (range[0] < READ_ONCE(net->ipv4.sysctl_ip_prot_sock)))
 			ret = -EINVAL;
 		else
 			set_local_port_range(net, range);
@@ -110,7 +110,7 @@ static int ipv4_privileged_ports(struct ctl_table *table, int write,
 		.extra2 = &ip_privileged_port_max,
 	};
 
-	pports = net->ipv4.sysctl_ip_prot_sock;
+	pports = READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
 
 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
 
@@ -122,7 +122,7 @@ static int ipv4_privileged_ports(struct ctl_table *table, int write,
 		if (range[0] < pports)
 			ret = -EINVAL;
 		else
-			net->ipv4.sysctl_ip_prot_sock = pports;
+			WRITE_ONCE(net->ipv4.sysctl_ip_prot_sock, pports);
 	}
 
 	return ret;
-- 
cgit v1.3-18-gd494


From 3d72bb4188c708bb16758c60822fc4dda7a95174 Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Mon, 18 Jul 2022 10:26:43 -0700
Subject: udp: Fix a data-race around sysctl_udp_l3mdev_accept.

While reading sysctl_udp_l3mdev_accept, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its reader.

Fixes: 63a6fff353d0 ("net: Avoid receiving packets with an l3mdev on unbound UDP sockets")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/udp.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'include')

diff --git a/include/net/udp.h b/include/net/udp.h
index bb4c227299cc..8dd4aa1485a6 100644
--- a/include/net/udp.h
+++ b/include/net/udp.h
@@ -238,7 +238,7 @@ static inline bool udp_sk_bound_dev_eq(struct net *net, int bound_dev_if,
 				       int dif, int sdif)
 {
 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
-	return inet_bound_dev_eq(!!net->ipv4.sysctl_udp_l3mdev_accept,
+	return inet_bound_dev_eq(!!READ_ONCE(net->ipv4.sysctl_udp_l3mdev_accept),
 				 bound_dev_if, dif, sdif);
 #else
 	return inet_bound_dev_eq(true, bound_dev_if, dif, sdif);
-- 
cgit v1.3-18-gd494


From 4845b5713ab18a1bb6e31d1fbb4d600240b8b691 Mon Sep 17 00:00:00 2001
From: Kuniyuki Iwashima <kuniyu@amazon.com>
Date: Mon, 18 Jul 2022 10:26:48 -0700
Subject: tcp: Fix data-races around sysctl_tcp_slow_start_after_idle.

While reading sysctl_tcp_slow_start_after_idle, it can be changed
concurrently.  Thus, we need to add READ_ONCE() to its readers.

Fixes: 35089bb203f4 ("[TCP]: Add tcp_slow_start_after_idle sysctl.")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/tcp.h     | 4 ++--
 net/ipv4/tcp_output.c | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

(limited to 'include')

diff --git a/include/net/tcp.h b/include/net/tcp.h
index 6ee1fb4fb292..071735e10872 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -1403,8 +1403,8 @@ static inline void tcp_slow_start_after_idle_check(struct sock *sk)
 	struct tcp_sock *tp = tcp_sk(sk);
 	s32 delta;
 
-	if (!sock_net(sk)->ipv4.sysctl_tcp_slow_start_after_idle || tp->packets_out ||
-	    ca_ops->cong_control)
+	if (!READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_slow_start_after_idle) ||
+	    tp->packets_out || ca_ops->cong_control)
 		return;
 	delta = tcp_jiffies32 - tp->lsndtime;
 	if (delta > inet_csk(sk)->icsk_rto)
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 898fcdcb7989..51120407c570 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -1898,7 +1898,7 @@ static void tcp_cwnd_validate(struct sock *sk, bool is_cwnd_limited)
 		if (tp->packets_out > tp->snd_cwnd_used)
 			tp->snd_cwnd_used = tp->packets_out;
 
-		if (sock_net(sk)->ipv4.sysctl_tcp_slow_start_after_idle &&
+		if (READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_slow_start_after_idle) &&
 		    (s32)(tcp_jiffies32 - tp->snd_cwnd_stamp) >= inet_csk(sk)->icsk_rto &&
 		    !ca_ops->cong_control)
 			tcp_cwnd_application_limited(sk);
-- 
cgit v1.3-18-gd494