[PATCH net-next 1/2] tcp: randomize tcp timestamp offsets for each connection

[PATCH net-next 1/2] tcp: randomize tcp timestamp offsets for each connection

[Florian Westphal]

jiffies based timestamps allow for easy inference of number of devices
behind NAT translators and also makes tracking of hosts simpler.

commit ceaa1fef65a7c2e ("tcp: adding a per-socket timestamp offset")
added the main infrastructure that is needed for per-connection ts
randomization, in particular writing/reading the on-wire tcp header
format takes the offset into account so rest of stack can use normal
tcp_time_stamp (jiffies).

So only two items are left:
 - add a tsoffset for request sockets
 - extend the tcp isn generator to also return another 32bit number
   in addition to the ISN.

Re-use of ISN generator also means timestamps are still monotonically
increasing for same connection quadruple, i.e. PAWS will still work.

Includes fixes from Eric Dumazet.

Cc: Mirja Kühlewind <mirja.kuehlewind@tik.ee.ethz.ch>
Signed-off-by: Florian Westphal <fw@strlen.de>
---
 include/linux/tcp.h      |  1 +
 include/net/secure_seq.h |  8 ++++----
 include/net/tcp.h        |  2 +-
 net/core/secure_seq.c    | 10 ++++++----
 net/ipv4/syncookies.c    |  1 +
 net/ipv4/tcp_input.c     |  7 ++++++-
 net/ipv4/tcp_ipv4.c      |  9 +++++----
 net/ipv4/tcp_minisocks.c |  4 +++-
 net/ipv4/tcp_output.c    |  2 +-
 net/ipv6/syncookies.c    |  1 +
 net/ipv6/tcp_ipv6.c      | 10 ++++++----
 11 files changed, 35 insertions(+), 20 deletions(-)

diff --git a/include/linux/tcp.h b/include/linux/tcp.h
index 32a7c7e35b71..2408bcc579f1 100644
--- a/include/linux/tcp.h
+++ b/include/linux/tcp.h
@@ -123,6 +123,7 @@ struct tcp_request_sock {
 	u32				txhash;
 	u32				rcv_isn;
 	u32				snt_isn;
+	u32				ts_off;
 	u32				last_oow_ack_time; /* last SYNACK */
 	u32				rcv_nxt; /* the ack # by SYNACK. For
 						  * FastOpen it's the seq#
diff --git a/include/net/secure_seq.h b/include/net/secure_seq.h
index 3f36d45b714a..0caee631a836 100644
--- a/include/net/secure_seq.h
+++ b/include/net/secure_seq.h
@@ -6,10 +6,10 @@
 u32 secure_ipv4_port_ephemeral(__be32 saddr, __be32 daddr, __be16 dport);
 u32 secure_ipv6_port_ephemeral(const __be32 *saddr, const __be32 *daddr,
 			       __be16 dport);
-__u32 secure_tcp_sequence_number(__be32 saddr, __be32 daddr,
-				 __be16 sport, __be16 dport);
-__u32 secure_tcpv6_sequence_number(const __be32 *saddr, const __be32 *daddr,
-				   __be16 sport, __be16 dport);
+u32 secure_tcp_sequence_number(__be32 saddr, __be32 daddr,
+			       __be16 sport, __be16 dport, u32 *tsoff);
+u32 secure_tcpv6_sequence_number(const __be32 *saddr, const __be32 *daddr,
+				 __be16 sport, __be16 dport, u32 *tsoff);
 u64 secure_dccp_sequence_number(__be32 saddr, __be32 daddr,
 				__be16 sport, __be16 dport);
 u64 secure_dccpv6_sequence_number(__be32 *saddr, __be32 *daddr,
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 7de80739adab..1c09d909bd43 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -1809,7 +1809,7 @@ struct tcp_request_sock_ops {
 	struct dst_entry *(*route_req)(const struct sock *sk, struct flowi *fl,
 				       const struct request_sock *req,
 				       bool *strict);
-	__u32 (*init_seq)(const struct sk_buff *skb);
+	__u32 (*init_seq)(const struct sk_buff *skb, u32 *tsoff);
 	int (*send_synack)(const struct sock *sk, struct dst_entry *dst,
 			   struct flowi *fl, struct request_sock *req,
 			   struct tcp_fastopen_cookie *foc,
diff --git a/net/core/secure_seq.c b/net/core/secure_seq.c
index fd3ce461fbe6..a8d6062cbb4a 100644
--- a/net/core/secure_seq.c
+++ b/net/core/secure_seq.c
@@ -40,8 +40,8 @@ static u32 seq_scale(u32 seq)
 #endif
 
 #if IS_ENABLED(CONFIG_IPV6)
-__u32 secure_tcpv6_sequence_number(const __be32 *saddr, const __be32 *daddr,
-				   __be16 sport, __be16 dport)
+u32 secure_tcpv6_sequence_number(const __be32 *saddr, const __be32 *daddr,
+				 __be16 sport, __be16 dport, u32 *tsoff)
 {
 	u32 secret[MD5_MESSAGE_BYTES / 4];
 	u32 hash[MD5_DIGEST_WORDS];
@@ -58,6 +58,7 @@ __u32 secure_tcpv6_sequence_number(const __be32 *saddr, const __be32 *daddr,
 
 	md5_transform(hash, secret);
 
+	*tsoff = hash[1];
 	return seq_scale(hash[0]);
 }
 EXPORT_SYMBOL(secure_tcpv6_sequence_number);
@@ -86,8 +87,8 @@ EXPORT_SYMBOL(secure_ipv6_port_ephemeral);
 
 #ifdef CONFIG_INET
 
-__u32 secure_tcp_sequence_number(__be32 saddr, __be32 daddr,
-				 __be16 sport, __be16 dport)
+u32 secure_tcp_sequence_number(__be32 saddr, __be32 daddr,
+			       __be16 sport, __be16 dport, u32 *tsoff)
 {
 	u32 hash[MD5_DIGEST_WORDS];
 
@@ -99,6 +100,7 @@ __u32 secure_tcp_sequence_number(__be32 saddr, __be32 daddr,
 
 	md5_transform(hash, net_secret);
 
+	*tsoff = hash[1];
 	return seq_scale(hash[0]);
 }
 
diff --git a/net/ipv4/syncookies.c b/net/ipv4/syncookies.c
index 0dc6286272aa..3e88467d70ee 100644
--- a/net/ipv4/syncookies.c
+++ b/net/ipv4/syncookies.c
@@ -334,6 +334,7 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb)
 	treq = tcp_rsk(req);
 	treq->rcv_isn		= ntohl(th->seq) - 1;
 	treq->snt_isn		= cookie;
+	treq->ts_off		= 0;
 	req->mss		= mss;
 	ireq->ir_num		= ntohs(th->dest);
 	ireq->ir_rmt_port	= th->source;
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 22e6a2097ff6..1b1921c71f7c 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -6301,6 +6301,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,
 		goto drop;
 
 	tcp_rsk(req)->af_specific = af_ops;
+	tcp_rsk(req)->ts_off = 0;
 
 	tcp_clear_options(&tmp_opt);
 	tmp_opt.mss_clamp = af_ops->mss_clamp;
@@ -6322,6 +6323,9 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,
 	if (security_inet_conn_request(sk, skb, req))
 		goto drop_and_free;
 
+	if (isn && tmp_opt.tstamp_ok)
+		af_ops->init_seq(skb, &tcp_rsk(req)->ts_off);
+
 	if (!want_cookie && !isn) {
 		/* VJ's idea. We save last timestamp seen
 		 * from the destination in peer table, when entering
@@ -6362,7 +6366,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,
 			goto drop_and_release;
 		}
 
-		isn = af_ops->init_seq(skb);
+		isn = af_ops->init_seq(skb, &tcp_rsk(req)->ts_off);
 	}
 	if (!dst) {
 		dst = af_ops->route_req(sk, &fl, req, NULL);
@@ -6374,6 +6378,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,
 
 	if (want_cookie) {
 		isn = cookie_init_sequence(af_ops, sk, skb, &req->mss);
+		tcp_rsk(req)->ts_off = 0;
 		req->cookie_ts = tmp_opt.tstamp_ok;
 		if (!tmp_opt.tstamp_ok)
 			inet_rsk(req)->ecn_ok = 0;
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 5555eb86e549..b50f05905ced 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -95,12 +95,12 @@ static int tcp_v4_md5_hash_hdr(char *md5_hash, const struct tcp_md5sig_key *key,
 struct inet_hashinfo tcp_hashinfo;
 EXPORT_SYMBOL(tcp_hashinfo);
 
-static  __u32 tcp_v4_init_sequence(const struct sk_buff *skb)
+static u32 tcp_v4_init_sequence(const struct sk_buff *skb, u32 *tsoff)
 {
 	return secure_tcp_sequence_number(ip_hdr(skb)->daddr,
 					  ip_hdr(skb)->saddr,
 					  tcp_hdr(skb)->dest,
-					  tcp_hdr(skb)->source);
+					  tcp_hdr(skb)->source, tsoff);
 }
 
 int tcp_twsk_unique(struct sock *sk, struct sock *sktw, void *twp)
@@ -237,7 +237,8 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 		tp->write_seq = secure_tcp_sequence_number(inet->inet_saddr,
 							   inet->inet_daddr,
 							   inet->inet_sport,
-							   usin->sin_port);
+							   usin->sin_port,
+							   &tp->tsoffset);
 
 	inet->inet_id = tp->write_seq ^ jiffies;
 
@@ -824,7 +825,7 @@ static void tcp_v4_reqsk_send_ack(const struct sock *sk, struct sk_buff *skb,
 	tcp_v4_send_ack(sk, skb, seq,
 			tcp_rsk(req)->rcv_nxt,
 			req->rsk_rcv_wnd >> inet_rsk(req)->rcv_wscale,
-			tcp_time_stamp,
+			tcp_time_stamp + tcp_rsk(req)->ts_off,
 			req->ts_recent,
 			0,
 			tcp_md5_do_lookup(sk, (union tcp_md5_addr *)&ip_hdr(skb)->daddr,
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index 6234ebaa7db1..28ce5ee831f5 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -532,7 +532,7 @@ struct sock *tcp_create_openreq_child(const struct sock *sk,
 			newtp->rx_opt.ts_recent_stamp = 0;
 			newtp->tcp_header_len = sizeof(struct tcphdr);
 		}
-		newtp->tsoffset = 0;
+		newtp->tsoffset = treq->ts_off;
 #ifdef CONFIG_TCP_MD5SIG
 		newtp->md5sig_info = NULL;	/*XXX*/
 		if (newtp->af_specific->md5_lookup(sk, newsk))
@@ -581,6 +581,8 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb,
 
 		if (tmp_opt.saw_tstamp) {
 			tmp_opt.ts_recent = req->ts_recent;
+			if (tmp_opt.rcv_tsecr)
+				tmp_opt.rcv_tsecr -= tcp_rsk(req)->ts_off;
 			/* We do not store true stamp, but it is not required,
 			 * it can be estimated (approximately)
 			 * from another data.
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 19105b46a304..1b6d5f34bf45 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -640,7 +640,7 @@ static unsigned int tcp_synack_options(struct request_sock *req,
 	}
 	if (likely(ireq->tstamp_ok)) {
 		opts->options |= OPTION_TS;
-		opts->tsval = tcp_skb_timestamp(skb);
+		opts->tsval = tcp_skb_timestamp(skb) + tcp_rsk(req)->ts_off;
 		opts->tsecr = req->ts_recent;
 		remaining -= TCPOLEN_TSTAMP_ALIGNED;
 	}
diff --git a/net/ipv6/syncookies.c b/net/ipv6/syncookies.c
index 97830a6a9cbb..a4d49760bf43 100644
--- a/net/ipv6/syncookies.c
+++ b/net/ipv6/syncookies.c
@@ -209,6 +209,7 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb)
 	treq->snt_synack.v64	= 0;
 	treq->rcv_isn = ntohl(th->seq) - 1;
 	treq->snt_isn = cookie;
+	treq->ts_off = 0;
 
 	/*
 	 * We need to lookup the dst_entry to get the correct window size.
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 28ec0a2e7b72..a2185a214abc 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -101,12 +101,12 @@ static void inet6_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb)
 	}
 }
 
-static __u32 tcp_v6_init_sequence(const struct sk_buff *skb)
+static u32 tcp_v6_init_sequence(const struct sk_buff *skb, u32 *tsoff)
 {
 	return secure_tcpv6_sequence_number(ipv6_hdr(skb)->daddr.s6_addr32,
 					    ipv6_hdr(skb)->saddr.s6_addr32,
 					    tcp_hdr(skb)->dest,
-					    tcp_hdr(skb)->source);
+					    tcp_hdr(skb)->source, tsoff);
 }
 
 static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
@@ -283,7 +283,8 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
 		tp->write_seq = secure_tcpv6_sequence_number(np->saddr.s6_addr32,
 							     sk->sk_v6_daddr.s6_addr32,
 							     inet->inet_sport,
-							     inet->inet_dport);
+							     inet->inet_dport,
+							     &tp->tsoffset);
 
 	err = tcp_connect(sk);
 	if (err)
@@ -956,7 +957,8 @@ static void tcp_v6_reqsk_send_ack(const struct sock *sk, struct sk_buff *skb,
 			tcp_rsk(req)->snt_isn + 1 : tcp_sk(sk)->snd_nxt,
 			tcp_rsk(req)->rcv_nxt,
 			req->rsk_rcv_wnd >> inet_rsk(req)->rcv_wscale,
-			tcp_time_stamp, req->ts_recent, sk->sk_bound_dev_if,
+			tcp_time_stamp + tcp_rsk(req)->ts_off,
+			req->ts_recent, sk->sk_bound_dev_if,
 			tcp_v6_md5_do_lookup(sk, &ipv6_hdr(skb)->daddr),
 			0, 0);
 }
-- 
2.7.3

[PATCH net-next 2/2] tcp: allow to turn tcp timestamp randomization off

[Florian Westphal]

Eric says: "By looking at tcpdump, and TS val of xmit packets of multiple
flows, we can deduct the relative qdisc delays (think of fq pacing).
This should work even if we have one flow per remote peer."

Having random per flow (or host) offsets doesn't allow that anymore so add
a way to turn this off.

Suggested-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
---
 Documentation/networking/ip-sysctl.txt |  9 +++++++--
 net/ipv4/tcp_input.c                   | 10 +++++++---
 2 files changed, 14 insertions(+), 5 deletions(-)

diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt
index 5af48dd7c5fc..de2448313799 100644
--- a/Documentation/networking/ip-sysctl.txt
+++ b/Documentation/networking/ip-sysctl.txt
@@ -610,8 +610,13 @@ tcp_syn_retries - INTEGER
 	with the current initial RTO of 1second. With this the final timeout
 	for an active TCP connection attempt will happen after 127seconds.
 
-tcp_timestamps - BOOLEAN
-	Enable timestamps as defined in RFC1323.
+tcp_timestamps - INTEGER
+Enable timestamps as defined in RFC1323.
+	0: Disabled.
+	1: Enable timestamps as defined in RFC1323.
+	2: Like 1, but also use a random offset for each connection
+	rather than only using the current time.
+	Default: 2
 
 tcp_min_tso_segs - INTEGER
 	Minimal number of segments per TSO frame.
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 1b1921c71f7c..ebed73703198 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -76,7 +76,7 @@
 #include <asm/unaligned.h>
 #include <linux/errqueue.h>
 
-int sysctl_tcp_timestamps __read_mostly = 1;
+int sysctl_tcp_timestamps __read_mostly = 2;
 int sysctl_tcp_window_scaling __read_mostly = 1;
 int sysctl_tcp_sack __read_mostly = 1;
 int sysctl_tcp_fack __read_mostly = 1;
@@ -6323,10 +6323,12 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,
 	if (security_inet_conn_request(sk, skb, req))
 		goto drop_and_free;
 
-	if (isn && tmp_opt.tstamp_ok)
+	if (isn && tmp_opt.tstamp_ok && sysctl_tcp_timestamps == 2)
 		af_ops->init_seq(skb, &tcp_rsk(req)->ts_off);
 
 	if (!want_cookie && !isn) {
+		u32 ts_off;
+
 		/* VJ's idea. We save last timestamp seen
 		 * from the destination in peer table, when entering
 		 * state TIME-WAIT, and check against it before
@@ -6366,7 +6368,9 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,
 			goto drop_and_release;
 		}
 
-		isn = af_ops->init_seq(skb, &tcp_rsk(req)->ts_off);
+		isn = af_ops->init_seq(skb, &ts_off);
+		if (sysctl_tcp_timestamps == 2)
+			tcp_rsk(req)->ts_off = ts_off;
 	}
 	if (!dst) {
 		dst = af_ops->route_req(sk, &fl, req, NULL);
-- 
2.7.3

Re: [PATCH net-next 1/2] tcp: randomize tcp timestamp offsets for each connection

[Eric Dumazet]

On Tue, 2016-11-29 at 16:45 +0100, Florian Westphal wrote:
> jiffies based timestamps allow for easy inference of number of devices
> behind NAT translators and also makes tracking of hosts simpler.
> 
> commit ceaa1fef65a7c2e ("tcp: adding a per-socket timestamp offset")
> added the main infrastructure that is needed for per-connection ts
> randomization, in particular writing/reading the on-wire tcp header
> format takes the offset into account so rest of stack can use normal
> tcp_time_stamp (jiffies).
> 
> So only two items are left:
>  - add a tsoffset for request sockets
>  - extend the tcp isn generator to also return another 32bit number
>    in addition to the ISN.
> 
> Re-use of ISN generator also means timestamps are still monotonically
> increasing for same connection quadruple, i.e. PAWS will still work.
> 
> Includes fixes from Eric Dumazet.
> 
> Cc: Mirja Kühlewind <mirja.kuehlewind@tik.ee.ethz.ch>
> Signed-off-by: Florian Westphal <fw@strlen.de>
> ---

Very nice work, thanks !

Acked-by: Eric Dumazet <edumazet@google.com>

Re: [PATCH net-next 2/2] tcp: allow to turn tcp timestamp randomization off

[Eric Dumazet]

On Tue, 2016-11-29 at 16:45 +0100, Florian Westphal wrote:
> Eric says: "By looking at tcpdump, and TS val of xmit packets of multiple
> flows, we can deduct the relative qdisc delays (think of fq pacing).
> This should work even if we have one flow per remote peer."
> 
> Having random per flow (or host) offsets doesn't allow that anymore so add
> a way to turn this off.
> 
> Suggested-by: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Florian Westphal <fw@strlen.de>
> ---

Excellent, thanks !

Acked-by: Eric Dumazet <edumazet@google.com>

Re: [PATCH net-next 2/2] tcp: allow to turn tcp timestamp randomization off

[Florian Westphal]

Eric Dumazet <eric.dumazet@gmail.com> wrote:
> On Tue, 2016-11-29 at 16:45 +0100, Florian Westphal wrote:
> > Eric says: "By looking at tcpdump, and TS val of xmit packets of multiple
> > flows, we can deduct the relative qdisc delays (think of fq pacing).
> > This should work even if we have one flow per remote peer."
> > 
> > Having random per flow (or host) offsets doesn't allow that anymore so add
> > a way to turn this off.
> > 
> > Suggested-by: Eric Dumazet <edumazet@google.com>
> > Signed-off-by: Florian Westphal <fw@strlen.de>
> > ---
> 
> Excellent, thanks !
> 
> Acked-by: Eric Dumazet <edumazet@google.com>

Thanks for the ack, I missed connect() side though so this doesn't work
for outgoing connections, sorry :-/

I will send a v2.