From 4faeee0cf8a5d88d63cdbc3bab124fb0e6aed08c Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 26 May 2023 16:34:58 +0000 Subject: tcp: deny tcp_disconnect() when threads are waiting Historically connect(AF_UNSPEC) has been abused by syzkaller and other fuzzers to trigger various bugs. A recent one triggers a divide-by-zero [1], and Paolo Abeni was able to diagnose the issue. tcp_recvmsg_locked() has tests about sk_state being not TCP_LISTEN and TCP REPAIR mode being not used. Then later if socket lock is released in sk_wait_data(), another thread can call connect(AF_UNSPEC), then make this socket a TCP listener. When recvmsg() is resumed, it can eventually call tcp_cleanup_rbuf() and attempt a divide by 0 in tcp_rcv_space_adjust() [1] This patch adds a new socket field, counting number of threads blocked in sk_wait_event() and inet_wait_for_connect(). If this counter is not zero, tcp_disconnect() returns an error. This patch adds code in blocking socket system calls, thus should not hurt performance of non blocking ones. Note that we probably could revert commit 499350a5a6e7 ("tcp: initialize rcv_mss to TCP_MIN_MSS instead of 0") to restore original tcpi_rcv_mss meaning (was 0 if no payload was ever received on a socket) [1] divide error: 0000 [#1] PREEMPT SMP KASAN CPU: 0 PID: 13832 Comm: syz-executor.5 Not tainted 6.3.0-rc4-syzkaller-00224-g00c7b5f4ddc5 #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 03/02/2023 RIP: 0010:tcp_rcv_space_adjust+0x36e/0x9d0 net/ipv4/tcp_input.c:740 Code: 00 00 00 00 fc ff df 4c 89 64 24 48 8b 44 24 04 44 89 f9 41 81 c7 80 03 00 00 c1 e1 04 44 29 f0 48 63 c9 48 01 e9 48 0f af c1 <49> f7 f6 48 8d 04 41 48 89 44 24 40 48 8b 44 24 30 48 c1 e8 03 48 RSP: 0018:ffffc900033af660 EFLAGS: 00010206 RAX: 4a66b76cbade2c48 RBX: ffff888076640cc0 RCX: 00000000c334e4ac RDX: 0000000000000000 RSI: dffffc0000000000 RDI: 0000000000000001 RBP: 00000000c324e86c R08: 0000000000000001 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000000 R12: ffff8880766417f8 R13: ffff888028fbb980 R14: 0000000000000000 R15: 0000000000010344 FS: 00007f5bffbfe700(0000) GS:ffff8880b9800000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000001b32f25000 CR3: 000000007ced0000 CR4: 00000000003506f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: tcp_recvmsg_locked+0x100e/0x22e0 net/ipv4/tcp.c:2616 tcp_recvmsg+0x117/0x620 net/ipv4/tcp.c:2681 inet6_recvmsg+0x114/0x640 net/ipv6/af_inet6.c:670 sock_recvmsg_nosec net/socket.c:1017 [inline] sock_recvmsg+0xe2/0x160 net/socket.c:1038 ____sys_recvmsg+0x210/0x5a0 net/socket.c:2720 ___sys_recvmsg+0xf2/0x180 net/socket.c:2762 do_recvmmsg+0x25e/0x6e0 net/socket.c:2856 __sys_recvmmsg net/socket.c:2935 [inline] __do_sys_recvmmsg net/socket.c:2958 [inline] __se_sys_recvmmsg net/socket.c:2951 [inline] __x64_sys_recvmmsg+0x20f/0x260 net/socket.c:2951 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x39/0xb0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd RIP: 0033:0x7f5c0108c0f9 Code: 28 00 00 00 75 05 48 83 c4 28 c3 e8 f1 19 00 00 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007f5bffbfe168 EFLAGS: 00000246 ORIG_RAX: 000000000000012b RAX: ffffffffffffffda RBX: 00007f5c011ac050 RCX: 00007f5c0108c0f9 RDX: 0000000000000001 RSI: 0000000020000bc0 RDI: 0000000000000003 RBP: 00007f5c010e7b39 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000122 R11: 0000000000000246 R12: 0000000000000000 R13: 00007f5c012cfb1f R14: 00007f5bffbfe300 R15: 0000000000022000 Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Reported-by: syzbot Reported-by: Paolo Abeni Diagnosed-by: Paolo Abeni Signed-off-by: Eric Dumazet Tested-by: Paolo Abeni Link: https://lore.kernel.org/r/20230526163458.2880232-1-edumazet@google.com Signed-off-by: Jakub Kicinski --- net/ipv4/af_inet.c | 2 ++ net/ipv4/inet_connection_sock.c | 1 + net/ipv4/tcp.c | 6 ++++++ 3 files changed, 9 insertions(+) (limited to 'net/ipv4') diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c index c4aab3aacbd8..4a76ebf793b8 100644 --- a/net/ipv4/af_inet.c +++ b/net/ipv4/af_inet.c @@ -586,6 +586,7 @@ static long inet_wait_for_connect(struct sock *sk, long timeo, int writebias) add_wait_queue(sk_sleep(sk), &wait); sk->sk_write_pending += writebias; + sk->sk_wait_pending++; /* Basic assumption: if someone sets sk->sk_err, he _must_ * change state of the socket from TCP_SYN_*. @@ -601,6 +602,7 @@ static long inet_wait_for_connect(struct sock *sk, long timeo, int writebias) } remove_wait_queue(sk_sleep(sk), &wait); sk->sk_write_pending -= writebias; + sk->sk_wait_pending--; return timeo; } diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c index 65ad4251f6fd..1386787eaf1a 100644 --- a/net/ipv4/inet_connection_sock.c +++ b/net/ipv4/inet_connection_sock.c @@ -1142,6 +1142,7 @@ struct sock *inet_csk_clone_lock(const struct sock *sk, if (newsk) { struct inet_connection_sock *newicsk = inet_csk(newsk); + newsk->sk_wait_pending = 0; inet_sk_set_state(newsk, TCP_SYN_RECV); newicsk->icsk_bind_hash = NULL; newicsk->icsk_bind2_hash = NULL; diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index a60f6f4e7cd9..635607ac37e4 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -3081,6 +3081,12 @@ int tcp_disconnect(struct sock *sk, int flags) int old_state = sk->sk_state; u32 seq; + /* Deny disconnect if other threads are blocked in sk_wait_event() + * or inet_wait_for_connect(). + */ + if (sk->sk_wait_pending) + return -EBUSY; + if (old_state != TCP_CLOSE) tcp_set_state(sk, TCP_CLOSE); -- cgit From 34dfde4ad87b84d21278a7e19d92b5b2c68e6c4d Mon Sep 17 00:00:00 2001 From: Cambda Zhu Date: Sat, 27 May 2023 12:03:17 +0800 Subject: tcp: Return user_mss for TCP_MAXSEG in CLOSE/LISTEN state if user_mss set This patch replaces the tp->mss_cache check in getting TCP_MAXSEG with tp->rx_opt.user_mss check for CLOSE/LISTEN sock. Since tp->mss_cache is initialized with TCP_MSS_DEFAULT, checking if it's zero is probably a bug. With this change, getting TCP_MAXSEG before connecting will return default MSS normally, and return user_mss if user_mss is set. Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Reported-by: Jack Yang Suggested-by: Eric Dumazet Link: https://lore.kernel.org/netdev/CANn89i+3kL9pYtkxkwxwNMzvC_w3LNUum_2=3u+UyLBmGmifHA@mail.gmail.com/#t Signed-off-by: Cambda Zhu Link: https://lore.kernel.org/netdev/14D45862-36EA-4076-974C-EA67513C92F6@linux.alibaba.com/ Reviewed-by: Jason Xing Reviewed-by: Eric Dumazet Link: https://lore.kernel.org/r/20230527040317.68247-1-cambda@linux.alibaba.com Signed-off-by: Jakub Kicinski --- net/ipv4/tcp.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 635607ac37e4..8d20d9221238 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -4078,7 +4078,8 @@ int do_tcp_getsockopt(struct sock *sk, int level, switch (optname) { case TCP_MAXSEG: val = tp->mss_cache; - if (!val && ((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_LISTEN))) + if (tp->rx_opt.user_mss && + ((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_LISTEN))) val = tp->rx_opt.user_mss; if (tp->repair) val = tp->rx_opt.mss_clamp; -- cgit From 30c6f0bf9579debce27e45fac34fdc97e46acacc Mon Sep 17 00:00:00 2001 From: fuyuanli Date: Wed, 31 May 2023 16:01:50 +0800 Subject: tcp: fix mishandling when the sack compression is deferred. In this patch, we mainly try to handle sending a compressed ack correctly if it's deferred. Here are more details in the old logic: When sack compression is triggered in the tcp_compressed_ack_kick(), if the sock is owned by user, it will set TCP_DELACK_TIMER_DEFERRED and then defer to the release cb phrase. Later once user releases the sock, tcp_delack_timer_handler() should send a ack as expected, which, however, cannot happen due to lack of ICSK_ACK_TIMER flag. Therefore, the receiver would not sent an ack until the sender's retransmission timeout. It definitely increases unnecessary latency. Fixes: 5d9f4262b7ea ("tcp: add SACK compression") Suggested-by: Eric Dumazet Signed-off-by: fuyuanli Signed-off-by: Jason Xing Link: https://lore.kernel.org/netdev/20230529113804.GA20300@didi-ThinkCentre-M920t-N000/ Reviewed-by: Eric Dumazet Link: https://lore.kernel.org/r/20230531080150.GA20424@didi-ThinkCentre-M920t-N000 Signed-off-by: Paolo Abeni --- net/ipv4/tcp_input.c | 2 +- net/ipv4/tcp_timer.c | 16 +++++++++++++--- 2 files changed, 14 insertions(+), 4 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 61b6710f337a..bf8b22218dd4 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -4530,7 +4530,7 @@ static void tcp_sack_maybe_coalesce(struct tcp_sock *tp) } } -static void tcp_sack_compress_send_ack(struct sock *sk) +void tcp_sack_compress_send_ack(struct sock *sk) { struct tcp_sock *tp = tcp_sk(sk); diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c index b839c2f91292..39eb947fe392 100644 --- a/net/ipv4/tcp_timer.c +++ b/net/ipv4/tcp_timer.c @@ -290,9 +290,19 @@ static int tcp_write_timeout(struct sock *sk) void tcp_delack_timer_handler(struct sock *sk) { struct inet_connection_sock *icsk = inet_csk(sk); + struct tcp_sock *tp = tcp_sk(sk); - if (((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_LISTEN)) || - !(icsk->icsk_ack.pending & ICSK_ACK_TIMER)) + if ((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_LISTEN)) + return; + + /* Handling the sack compression case */ + if (tp->compressed_ack) { + tcp_mstamp_refresh(tp); + tcp_sack_compress_send_ack(sk); + return; + } + + if (!(icsk->icsk_ack.pending & ICSK_ACK_TIMER)) return; if (time_after(icsk->icsk_ack.timeout, jiffies)) { @@ -312,7 +322,7 @@ void tcp_delack_timer_handler(struct sock *sk) inet_csk_exit_pingpong_mode(sk); icsk->icsk_ack.ato = TCP_ATO_MIN; } - tcp_mstamp_refresh(tcp_sk(sk)); + tcp_mstamp_refresh(tp); tcp_send_ack(sk); __NET_INC_STATS(sock_net(sk), LINUX_MIB_DELAYEDACKS); } -- cgit