The SYN-RECEIVED request-socket path in tcp_check_req() accepts an in-window RST without requiring SEG.SEQ to exactly match RCV.NXT. A non-exact RST therefore removes the request instead of eliciting a challenge ACK.
RFC 9293 section 3.10.7.4 applies the RFC 5961 reset check in SYN-RECEIVED: an exact RST resets the connection, while a non-exact in-window RST must trigger a challenge ACK and be dropped. Apply that check before the ACK-field validation, following the RFC sequence-number, RST, then ACK processing order. Factor the per-netns challenge ACK quota out of tcp_send_challenge_ack() so request sockets can share it. Use the request socket's send_ack() callback and its own out-of-window ACK timestamp to send and rate-limit the response. Reported-by: Yuxiang Yang <[email protected]> Reported-by: Yizhou Zhao <[email protected]> Reported-by: Ao Wang <[email protected]> Reported-by: Xuewei Feng <[email protected]> Reported-by: Qi Li <[email protected]> Reported-by: Ke Xu <[email protected]> Fixes: 282f23c6ee34 ("tcp: implement RFC 5961 3.2") Cc: [email protected] Signed-off-by: Yuxiang Yang <[email protected]> --- include/net/tcp.h | 2 ++ net/ipv4/tcp_input.c | 56 ++++++++++++++++++++++++++++++---------- net/ipv4/tcp_minisocks.c | 12 ++++++++- 3 files changed, 56 insertions(+), 14 deletions(-) diff --git a/include/net/tcp.h b/include/net/tcp.h index 6d376ea4d..2c5b88953 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -1974,6 +1974,8 @@ static inline void tcp_fast_path_check(struct sock *sk) bool tcp_oow_rate_limited(struct net *net, const struct sk_buff *skb, int mib_idx, u32 *last_oow_ack_time); +void tcp_reqsk_send_challenge_ack(struct sock *sk, struct sk_buff *skb, + struct request_sock *req); static inline void tcp_mib_init(struct net *net) { diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 61045a888..daff93d51 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -4038,24 +4038,17 @@ static void tcp_send_ack_reflect_ect(struct sock *sk, bool accecn_reflector) __tcp_send_ack(sk, tp->rcv_nxt, flags); } -/* RFC 5961 7 [ACK Throttling] */ -static void tcp_send_challenge_ack(struct sock *sk, bool accecn_reflector) +/* Consume one slot from the per-netns RFC 5961 challenge ACK quota. + * Returns true if a challenge ACK may be sent. + */ +static bool tcp_challenge_ack_allowed(struct net *net) { - struct tcp_sock *tp = tcp_sk(sk); - struct net *net = sock_net(sk); u32 count, now, ack_limit; - /* First check our per-socket dupack rate limit. */ - if (__tcp_oow_rate_limited(net, - LINUX_MIB_TCPACKSKIPPEDCHALLENGE, - &tp->last_oow_ack_time)) - return; - ack_limit = READ_ONCE(net->ipv4.sysctl_tcp_challenge_ack_limit); if (ack_limit == INT_MAX) - goto send_ack; + return true; - /* Then check host-wide RFC 5961 rate limit. */ now = jiffies / HZ; if (now != READ_ONCE(net->ipv4.tcp_challenge_timestamp)) { u32 half = (ack_limit + 1) >> 1; @@ -4067,12 +4060,49 @@ static void tcp_send_challenge_ack(struct sock *sk, bool accecn_reflector) count = READ_ONCE(net->ipv4.tcp_challenge_count); if (count > 0) { WRITE_ONCE(net->ipv4.tcp_challenge_count, count - 1); -send_ack: + return true; + } + return false; +} + +/* RFC 5961 7 [ACK Throttling] */ +static void tcp_send_challenge_ack(struct sock *sk, bool accecn_reflector) +{ + struct tcp_sock *tp = tcp_sk(sk); + struct net *net = sock_net(sk); + + /* First check our per-socket dupack rate limit. */ + if (__tcp_oow_rate_limited(net, + LINUX_MIB_TCPACKSKIPPEDCHALLENGE, + &tp->last_oow_ack_time)) + return; + + /* Then check the per-netns RFC 5961 rate limit. */ + if (tcp_challenge_ack_allowed(net)) { NET_INC_STATS(net, LINUX_MIB_TCPCHALLENGEACK); tcp_send_ack_reflect_ect(sk, accecn_reflector); } } +/* Send a challenge ACK from a SYN-RECEIVED request socket. Uses + * __tcp_oow_rate_limited() directly so that an RST carrying payload + * cannot bypass the per-request rate limit. + */ +void tcp_reqsk_send_challenge_ack(struct sock *sk, struct sk_buff *skb, + struct request_sock *req) +{ + struct net *net = sock_net(sk); + + if (__tcp_oow_rate_limited(net, LINUX_MIB_TCPACKSKIPPEDCHALLENGE, + &tcp_rsk(req)->last_oow_ack_time)) + return; + + if (tcp_challenge_ack_allowed(net)) { + NET_INC_STATS(net, LINUX_MIB_TCPCHALLENGEACK); + req->rsk_ops->send_ack(sk, skb, req); + } +} + static void tcp_store_ts_recent(struct tcp_sock *tp) { tp->rx_opt.ts_recent = tp->rx_opt.rcv_tsval; diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index ddc4b17a8..6ab3e3a0b 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -833,7 +833,7 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb, * elsewhere and is checked directly against the child socket rather * than req because user data may have been sent out. */ - if ((flg & TCP_FLAG_ACK) && !fastopen && + if ((flg & TCP_FLAG_ACK) && !(flg & TCP_FLAG_RST) && !fastopen && (TCP_SKB_CB(skb)->ack_seq != tcp_rsk(req)->snt_isn + 1)) return sk; @@ -872,6 +872,16 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb, flg &= ~TCP_FLAG_SYN; } + /* RFC 5961 section 3.2, as clarified by RFC 9293 section + * 3.10.7.4, requires a challenge ACK for a non-exact + * in-window RST in SYN-RECEIVED. + */ + if ((flg & TCP_FLAG_RST) && + TCP_SKB_CB(skb)->seq != tcp_rsk(req)->rcv_nxt) { + tcp_reqsk_send_challenge_ack(sk, skb, req); + return NULL; + } + /* RFC793: "second check the RST bit" and * "fourth, check the SYN bit" */ -- 2.34.1
