[PATCH net] udp: annotate data-races around up->pending

[Eric Dumazet <edumazet@google.com>]

up->pending can be read without holding the socket lock,
as pointed out by syzbot [1]

Add READ_ONCE() in lockless contexts, and WRITE_ONCE()
on write side.

[1]
BUG: KCSAN: data-race in udpv6_sendmsg / udpv6_sendmsg

write to 0xffff88814e5eadf0 of 4 bytes by task 15547 on cpu 1:
 udpv6_sendmsg+0x1405/0x1530 net/ipv6/udp.c:1596
 inet6_sendmsg+0x63/0x80 net/ipv6/af_inet6.c:657
 sock_sendmsg_nosec net/socket.c:730 [inline]
 __sock_sendmsg net/socket.c:745 [inline]
 __sys_sendto+0x257/0x310 net/socket.c:2192
 __do_sys_sendto net/socket.c:2204 [inline]
 __se_sys_sendto net/socket.c:2200 [inline]
 __x64_sys_sendto+0x78/0x90 net/socket.c:2200
 do_syscall_x64 arch/x86/entry/common.c:52 [inline]
 do_syscall_64+0x44/0x110 arch/x86/entry/common.c:83
 entry_SYSCALL_64_after_hwframe+0x63/0x6b

read to 0xffff88814e5eadf0 of 4 bytes by task 15551 on cpu 0:
 udpv6_sendmsg+0x22c/0x1530 net/ipv6/udp.c:1373
 inet6_sendmsg+0x63/0x80 net/ipv6/af_inet6.c:657
 sock_sendmsg_nosec net/socket.c:730 [inline]
 __sock_sendmsg net/socket.c:745 [inline]
 ____sys_sendmsg+0x37c/0x4d0 net/socket.c:2586
 ___sys_sendmsg net/socket.c:2640 [inline]
 __sys_sendmmsg+0x269/0x500 net/socket.c:2726
 __do_sys_sendmmsg net/socket.c:2755 [inline]
 __se_sys_sendmmsg net/socket.c:2752 [inline]
 __x64_sys_sendmmsg+0x57/0x60 net/socket.c:2752
 do_syscall_x64 arch/x86/entry/common.c:52 [inline]
 do_syscall_64+0x44/0x110 arch/x86/entry/common.c:83
 entry_SYSCALL_64_after_hwframe+0x63/0x6b

value changed: 0x00000000 -> 0x0000000a

Reported by Kernel Concurrency Sanitizer on:
CPU: 0 PID: 15551 Comm: syz-executor.1 Tainted: G        W          6.7.0-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 11/17/2023

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Reported-by: syzbot+8d482d0e407f665d9d10@syzkaller.appspotmail.com
Link: https://lore.kernel.org/netdev/0000000000009e46c3060ebcdffd@google.com/
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 net/ipv4/udp.c | 12 ++++++------
 net/ipv6/udp.c | 16 ++++++++--------
 2 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index 89e5a806b82e9c83b583d454e1b58b7838068f04..5f742d0b9e0794f994a8f2637d237d15d76549f4 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -805,7 +805,7 @@ void udp_flush_pending_frames(struct sock *sk)
 
 	if (up->pending) {
 		up->len = 0;
-		up->pending = 0;
+		WRITE_ONCE(up->pending, 0);
 		ip_flush_pending_frames(sk);
 	}
 }
@@ -993,7 +993,7 @@ int udp_push_pending_frames(struct sock *sk)
 
 out:
 	up->len = 0;
-	up->pending = 0;
+	WRITE_ONCE(up->pending, 0);
 	return err;
 }
 EXPORT_SYMBOL(udp_push_pending_frames);
@@ -1070,7 +1070,7 @@ int udp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 	getfrag = is_udplite ? udplite_getfrag : ip_generic_getfrag;
 
 	fl4 = &inet->cork.fl.u.ip4;
-	if (up->pending) {
+	if (READ_ONCE(up->pending)) {
 		/*
 		 * There are pending frames.
 		 * The socket lock must be held while it's corked.
@@ -1269,7 +1269,7 @@ int udp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 	fl4->saddr = saddr;
 	fl4->fl4_dport = dport;
 	fl4->fl4_sport = inet->inet_sport;
-	up->pending = AF_INET;
+	WRITE_ONCE(up->pending, AF_INET);
 
 do_append_data:
 	up->len += ulen;
@@ -1281,7 +1281,7 @@ int udp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 	else if (!corkreq)
 		err = udp_push_pending_frames(sk);
 	else if (unlikely(skb_queue_empty(&sk->sk_write_queue)))
-		up->pending = 0;
+		WRITE_ONCE(up->pending, 0);
 	release_sock(sk);
 
 out:
@@ -1319,7 +1319,7 @@ void udp_splice_eof(struct socket *sock)
 	struct sock *sk = sock->sk;
 	struct udp_sock *up = udp_sk(sk);
 
-	if (!up->pending || udp_test_bit(CORK, sk))
+	if (!READ_ONCE(up->pending) || udp_test_bit(CORK, sk))
 		return;
 
 	lock_sock(sk);
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index 594e3f23c12909fe6f245bf31057278169cd85c5..3f2249b4cd5f6a594dd9768e29f20f0d9a57faed 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -1135,7 +1135,7 @@ static void udp_v6_flush_pending_frames(struct sock *sk)
 		udp_flush_pending_frames(sk);
 	else if (up->pending) {
 		up->len = 0;
-		up->pending = 0;
+		WRITE_ONCE(up->pending, 0);
 		ip6_flush_pending_frames(sk);
 	}
 }
@@ -1313,7 +1313,7 @@ static int udp_v6_push_pending_frames(struct sock *sk)
 			      &inet_sk(sk)->cork.base);
 out:
 	up->len = 0;
-	up->pending = 0;
+	WRITE_ONCE(up->pending, 0);
 	return err;
 }
 
@@ -1370,7 +1370,7 @@ int udpv6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 		default:
 			return -EINVAL;
 		}
-	} else if (!up->pending) {
+	} else if (!READ_ONCE(up->pending)) {
 		if (sk->sk_state != TCP_ESTABLISHED)
 			return -EDESTADDRREQ;
 		daddr = &sk->sk_v6_daddr;
@@ -1401,8 +1401,8 @@ int udpv6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 		return -EMSGSIZE;
 
 	getfrag  =  is_udplite ?  udplite_getfrag : ip_generic_getfrag;
-	if (up->pending) {
-		if (up->pending == AF_INET)
+	if (READ_ONCE(up->pending)) {
+		if (READ_ONCE(up->pending) == AF_INET)
 			return udp_sendmsg(sk, msg, len);
 		/*
 		 * There are pending frames.
@@ -1593,7 +1593,7 @@ int udpv6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 		goto out;
 	}
 
-	up->pending = AF_INET6;
+	WRITE_ONCE(up->pending, AF_INET6);
 
 do_append_data:
 	if (ipc6.dontfrag < 0)
@@ -1607,7 +1607,7 @@ int udpv6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 	else if (!corkreq)
 		err = udp_v6_push_pending_frames(sk);
 	else if (unlikely(skb_queue_empty(&sk->sk_write_queue)))
-		up->pending = 0;
+		WRITE_ONCE(up->pending, 0);
 
 	if (err > 0)
 		err = inet6_test_bit(RECVERR6, sk) ? net_xmit_errno(err) : 0;
@@ -1648,7 +1648,7 @@ static void udpv6_splice_eof(struct socket *sock)
 	struct sock *sk = sock->sk;
 	struct udp_sock *up = udp_sk(sk);
 
-	if (!up->pending || udp_test_bit(CORK, sk))
+	if (!READ_ONCE(up->pending) || udp_test_bit(CORK, sk))
 		return;
 
 	lock_sock(sk);
-- 
2.43.0.275.g3460e3d667-goog