From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail-qk1-f201.google.com (mail-qk1-f201.google.com [209.85.222.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 58A323EDE59 for ; Tue, 12 May 2026 17:52:56 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.222.201 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1778608377; cv=none; b=jigWh69QYId6cLmBWUx1l/7I3sQGan9m417saLzEqo64Wb94mSHp8hXyYwjeBiDa+yS0dvmkQved2AsbOKX5jtNynMXqN01HzuXm3ybAsSFjXxFOZCoBBJeb7/tzb/G817ZPUTiEg/Zl9iLQUGDBZQnRQBHqSuD/kNY+r04mw1k= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1778608377; c=relaxed/simple; bh=evUa0TAAI3eXZ0FsA0tYRXKhf6FkGowWBBWRmcVr8NU=; h=Date:Mime-Version:Message-ID:Subject:From:To:Cc:Content-Type; b=c8pLzXtwVzRCvTT42gaTif1MppOR2VdqX9acLIBXpa6WQEjKxANetrJw7KbsVy3r/O2t0zXbQpoGkKxXsqxEAU1/pttbZb54t6uxED3Cao4mmvfWG1t2dcIbcZpHJNNesv+KumyPNHs2U0nh2OFLMvJl05ACskHOLZjGf2DEEnY= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--edumazet.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=c5ZK9/RH; arc=none smtp.client-ip=209.85.222.201 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--edumazet.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="c5ZK9/RH" Received: by mail-qk1-f201.google.com with SMTP id af79cd13be357-908438e5f47so859490585a.0 for ; Tue, 12 May 2026 10:52:56 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20251104; t=1778608375; x=1779213175; darn=vger.kernel.org; h=cc:to:from:subject:message-id:mime-version:date:from:to:cc:subject :date:message-id:reply-to; bh=ML1j1QWf0h4V2ae6GBFcVrt8m1Q9rBwZAydsZktoI+w=; b=c5ZK9/RHHgrM9zockfU7X0y0ZJ09+aMsKk6lJI0LO2vmKPnhn0J5LHzWFCn+XqTNH6 SVi5MUQYlLwsOJWwYmDCMP+zQ2/ZAlhy8KDq740SZty3kw3hx9eNCEVtpvyIR/rGrEWM +wlspWZt4V2MNmEiI3J9LOmBH/ch9qi4V2GBDor2hXGtDLpSZ0daSuSpn3xJyL2VcNPq vyoItaS3WIf4lZkG53nL+vFlvpKp8p+ZQXGCscOwZasaYe+hJlRHTIWdzbHKYh5yPNCr x04Nf3lvqHPh8Kp22hPscmCIzslR++WomhAAlUWw9+WD9s8JwmPZ/9AHEqC80nO3t/3R T43w== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20251104; t=1778608375; x=1779213175; h=cc:to:from:subject:message-id:mime-version:date:x-gm-message-state :from:to:cc:subject:date:message-id:reply-to; bh=ML1j1QWf0h4V2ae6GBFcVrt8m1Q9rBwZAydsZktoI+w=; b=PSJrisCm9zLJiq1VONZfBi3y/EqfYXWwvgPAMXWxE6hwPH/MoFoaUBMAPTVNaFumbR EDT5ItcEfaQYdutpeNMxJWtIWQtL15as8PVVLnUTagDYI3vq0mt81kGyBD5/ri0L1ckt K+zrLIdVI2yF47vRRwhUl67KVk662HDbY63pQdp5Ycu0F7zvVy6rugXY3ie0pxrdkGCv T/15tPgyBcEdVTwieaCc1k5wDP1K8+eplC4mCO2O0+1+dvUGgzVM9KcvV93K5ZDkmLMj n1eR41MCWTN0rhE2b8MFtMQYdQ562WCmYVIDREM9lTNU8sMENYSg8bV9895Q/rY9Fn0m hVYQ== X-Forwarded-Encrypted: i=1; AFNElJ+6xOW7DbLQy2S0x1t3XtbZzFceAIMkPxmYL2IheIjvDTpE21pp6dEN9b7cL4LNOpeoslnq18Q=@vger.kernel.org X-Gm-Message-State: AOJu0YwghC4n4ONYByUFux6itlqSCmZgwHmKnxb362s6rxRd62Tvrtdo icXCmNgCvM9FCbOEXOE/Ah4ckImAENwFN1EL7YKbHgt1IYENrcaRPxJqrxCUTeHQRzn8id+WNrE cDdCu9yhhtAcK5g== X-Received: from qknvk12.prod.google.com ([2002:a05:620a:70cc:b0:8ee:edbf:bd43]) (user=edumazet job=prod-delivery.src-stubby-dispatcher) by 2002:a05:620a:4110:b0:8ef:f1c0:ab8c with SMTP id af79cd13be357-90f7f932f5dmr17774285a.6.1778608375044; Tue, 12 May 2026 10:52:55 -0700 (PDT) Date: Tue, 12 May 2026 17:52:50 +0000 Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 X-Mailer: git-send-email 2.54.0.563.g4f69b47b94-goog Message-ID: <20260512175250.346486-1-edumazet@google.com> Subject: [PATCH v2 net] tcp: fix stale per-CPU tcp_tw_isn leak enabling ISN prediction From: Eric Dumazet To: "David S . Miller" , Jakub Kicinski , Paolo Abeni Cc: Simon Horman , Neal Cardwell , Kuniyuki Iwashima , netdev@vger.kernel.org, eric.dumazet@gmail.com, Eric Dumazet Content-Type: text/plain; charset="UTF-8" Blamed commit moved the TIME_WAIT-derived ISN from the skb control block to a per-CPU variable, assuming the value would always be consumed by tcp_conn_request() for the same packet that wrote it. That assumption is violated by multiple drop paths between the producer (__this_cpu_write(tcp_tw_isn, isn) in tcp_v{4,6}_rcv()) and the consumer (tcp_conn_request()): - min_ttl / min_hopcount check - xfrm policy check - tcp_inbound_hash() MD5/AO mismatch - tcp_filter() eBPF/SO_ATTACH_FILTER drop - th->syn && th->fin discard in tcp_rcv_state_process() TCP_LISTEN - psp_sk_rx_policy_check() in tcp_v{4,6}_do_rcv() - tcp_checksum_complete() in tcp_v{4,6}_do_rcv() - tcp_v{4,6}_cookie_check() returning NULL When a packet is dropped on any of these paths, tcp_tw_isn is left set. The next SYN processed on the same CPU then consumes the non zero value in tcp_conn_request(), receiving a predictable ISN. We could fix this by clearing tcp_tw_isn at tcp_v{4,6}_do_rcv() start, at the expense of slower fast path. This patch moves back tcp_tw_isn to skb, but not in skb->cb[] which was the original problem. Instead, union it with skb->mark / skb->reserved_tailroom which are unused in TCP receive path. Fixes: 41eecbd712b7 ("tcp: replace TCP_SKB_CB(skb)->tcp_tw_isn with a per-cpu field") Reported-by: Jakub Kicinski Signed-off-by: Eric Dumazet --- v2: get rid of the per-cpu storage, this was too much hassle. include/linux/skbuff.h | 1 + include/net/tcp.h | 3 --- net/ipv4/tcp.c | 3 --- net/ipv4/tcp_input.c | 15 ++++++--------- net/ipv4/tcp_ipv4.c | 5 ++--- net/ipv4/tcp_minisocks.c | 5 +++-- net/ipv6/tcp_ipv6.c | 5 ++--- 7 files changed, 14 insertions(+), 23 deletions(-) diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h index 2bcf78a4de7b9edb0d1342319d4340c0a9997eeb..93c6072c97b65ebabb5dd0245edff0728ff082d1 100644 --- a/include/linux/skbuff.h +++ b/include/linux/skbuff.h @@ -1066,6 +1066,7 @@ struct sk_buff { union { __u32 mark; __u32 reserved_tailroom; + u32 tcp_tw_isn; }; union { diff --git a/include/net/tcp.h b/include/net/tcp.h index ecbadcb3a7446cb18c245e670ba49ff574dfaff7..74835c51e0d55ee9b35d422d442f951e37d95cbc 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -65,8 +65,6 @@ static inline void tcp_orphan_count_dec(void) this_cpu_dec(tcp_orphan_count); } -DECLARE_PER_CPU(u32, tcp_tw_isn); - void tcp_time_wait(struct sock *sk, int state, int timeo); #define MAX_TCP_HEADER L1_CACHE_ALIGN(128 + MAX_HEADER) @@ -479,7 +477,6 @@ enum tcp_tw_status { enum tcp_tw_status tcp_timewait_state_process(struct inet_timewait_sock *tw, struct sk_buff *skb, const struct tcphdr *th, - u32 *tw_isn, enum skb_drop_reason *drop_reason); struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb, struct request_sock *req, bool fastopen, diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 432fa28e47d4c8ef5d50339bfdf7da0ea8772b94..389a7cc17110daa5b3b490b3c339e53e212969f8 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -299,9 +299,6 @@ enum { DEFINE_PER_CPU(unsigned int, tcp_orphan_count); EXPORT_PER_CPU_SYMBOL_GPL(tcp_orphan_count); -DEFINE_PER_CPU(u32, tcp_tw_isn); -EXPORT_PER_CPU_SYMBOL_GPL(tcp_tw_isn); - long sysctl_tcp_mem[3] __read_mostly; DEFINE_PER_CPU(int, tcp_memory_per_cpu_fw_alloc); diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index d5c9e65d97606d8eb57aba8ebc2373adf1bed62b..203660dc94efddfb6a79db8b5fce4d54dbe0b4ab 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -7594,25 +7594,22 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, struct net *net = sock_net(sk); struct sock *fastopen_sk = NULL; union tcp_seq_and_ts_off st; + u32 isn = skb->tcp_tw_isn; struct request_sock *req; bool want_cookie = false; struct dst_entry *dst; struct flowi fl; u8 syncookies; - u32 isn; #ifdef CONFIG_TCP_AO const struct tcp_ao_hdr *aoh; #endif - isn = __this_cpu_read(tcp_tw_isn); - if (isn) { - /* TW buckets are converted to open requests without - * limitations, they conserve resources and peer is - * evidently real one. - */ - __this_cpu_write(tcp_tw_isn, 0); - } else { + /* TW buckets are converted to open requests without + * limitations, they conserve resources and peer is + * evidently real one. + */ + if (!isn) { syncookies = READ_ONCE(net->ipv4.sysctl_tcp_syncookies); if (syncookies == 2 || inet_csk_reqsk_queue_is_full(sk)) { diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index c0526cc0398049fb34b5de20a1175d54942e80cd..719cfab3cd0296652791b9720fc25c9e437ca403 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2077,7 +2077,6 @@ int tcp_v4_rcv(struct sk_buff *skb) struct sock *sk = NULL; bool refcounted; int ret; - u32 isn; drop_reason = SKB_DROP_REASON_NOT_SPECIFIED; if (skb->pkt_type != PACKET_HOST) @@ -2198,6 +2197,7 @@ int tcp_v4_rcv(struct sk_buff *skb) } } + skb->tcp_tw_isn = 0; process: if (static_branch_unlikely(&ip4_min_ttl)) { /* min_ttl can be changed concurrently from do_ip_setsockopt() */ @@ -2299,7 +2299,7 @@ int tcp_v4_rcv(struct sk_buff *skb) goto csum_error; } - tw_status = tcp_timewait_state_process(inet_twsk(sk), skb, th, &isn, + tw_status = tcp_timewait_state_process(inet_twsk(sk), skb, th, &drop_reason); switch (tw_status) { case TCP_TW_SYN: { @@ -2313,7 +2313,6 @@ int tcp_v4_rcv(struct sk_buff *skb) sk = sk2; tcp_v4_restore_cb(skb); refcounted = false; - __this_cpu_write(tcp_tw_isn, isn); goto process; } diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index e6092c3ac840bdc1f62d4435c414e7f79edc10c2..8396f396f326336a31bde408dfffcdd7de95187b 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -99,7 +99,7 @@ static void twsk_rcv_nxt_update(struct tcp_timewait_sock *tcptw, u32 seq, */ enum tcp_tw_status tcp_timewait_state_process(struct inet_timewait_sock *tw, struct sk_buff *skb, - const struct tcphdr *th, u32 *tw_isn, + const struct tcphdr *th, enum skb_drop_reason *drop_reason) { struct tcp_timewait_sock *tcptw = tcp_twsk((struct sock *)tw); @@ -255,9 +255,10 @@ tcp_timewait_state_process(struct inet_timewait_sock *tw, struct sk_buff *skb, (tmp_opt.saw_tstamp && (s32)(READ_ONCE(tcptw->tw_ts_recent) - tmp_opt.rcv_tsval) < 0))) { u32 isn = tcptw->tw_snd_nxt + 65535 + 2; + if (isn == 0) isn++; - *tw_isn = isn; + skb->tcp_tw_isn = isn; return TCP_TW_SYN; } diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c index d13d49bfef19457cc5902cb556605a80f4c0ab2c..208ec75703df17d5e53e133b888b6b93a5ec04e6 100644 --- a/net/ipv6/tcp_ipv6.c +++ b/net/ipv6/tcp_ipv6.c @@ -1722,7 +1722,6 @@ INDIRECT_CALLABLE_SCOPE int tcp_v6_rcv(struct sk_buff *skb) struct sock *sk = NULL; bool refcounted; int ret; - u32 isn; drop_reason = SKB_DROP_REASON_NOT_SPECIFIED; if (skb->pkt_type != PACKET_HOST) @@ -1839,6 +1838,7 @@ INDIRECT_CALLABLE_SCOPE int tcp_v6_rcv(struct sk_buff *skb) } } + skb->tcp_tw_isn = 0; process: if (static_branch_unlikely(&ip6_min_hopcount)) { /* min_hopcount can be changed concurrently from do_ipv6_setsockopt() */ @@ -1937,7 +1937,7 @@ INDIRECT_CALLABLE_SCOPE int tcp_v6_rcv(struct sk_buff *skb) goto csum_error; } - tw_status = tcp_timewait_state_process(inet_twsk(sk), skb, th, &isn, + tw_status = tcp_timewait_state_process(inet_twsk(sk), skb, th, &drop_reason); switch (tw_status) { case TCP_TW_SYN: @@ -1956,7 +1956,6 @@ INDIRECT_CALLABLE_SCOPE int tcp_v6_rcv(struct sk_buff *skb) sk = sk2; tcp_v6_restore_cb(skb); refcounted = false; - __this_cpu_write(tcp_tw_isn, isn); goto process; } -- 2.54.0.563.g4f69b47b94-goog