1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 */
31
32 #include "opt_inet.h"
33 #include "opt_inet6.h"
34 #include "opt_ipsec.h"
35
36 #include <sys/param.h>
37 #include <sys/systm.h>
38 #include <sys/callout.h>
39 #include <sys/kernel.h>
40 #include <sys/sysctl.h>
41 #include <sys/malloc.h>
42 #include <sys/mbuf.h>
43 #include <sys/priv.h>
44 #include <sys/proc.h>
45 #include <sys/socket.h>
46 #include <sys/socketvar.h>
47 #include <sys/syslog.h>
48 #include <sys/protosw.h>
49 #include <sys/random.h>
50
51 #include <vm/uma.h>
52
53 #include <net/route.h>
54 #include <net/if.h>
55 #include <net/if_var.h>
56 #include <net/vnet.h>
57
58 #include <netinet/in.h>
59 #include <netinet/in_kdtrace.h>
60 #include <netinet/in_pcb.h>
61 #include <netinet/in_systm.h>
62 #include <netinet/in_var.h>
63 #include <netinet/ip.h>
64 #include <netinet/ip_icmp.h>
65 #include <netinet/ip_var.h>
66 #ifdef INET6
67 #include <netinet/ip6.h>
68 #include <netinet6/in6_pcb.h>
69 #include <netinet6/ip6_var.h>
70 #include <netinet6/scope6_var.h>
71 #include <netinet6/nd6.h>
72 #endif
73 #include <netinet/tcp.h>
74 #include <netinet/tcp_fsm.h>
75 #include <netinet/tcp_seq.h>
76 #include <netinet/tcp_timer.h>
77 #include <netinet/tcp_var.h>
78 #include <netinet/tcpip.h>
79
80 #include <netinet/udp.h>
81 #include <netinet/udp_var.h>
82
83 #include <netipsec/ipsec_support.h>
84
85 #include <machine/in_cksum.h>
86
87 #include <security/mac/mac_framework.h>
88
89 static u_int
tcp_eff_msl(struct tcpcb * tp)90 tcp_eff_msl(struct tcpcb *tp)
91 {
92 struct inpcb *inp = tptoinpcb(tp);
93 #ifdef INET6
94 bool isipv6 = inp->inp_inc.inc_flags & INC_ISIPV6;
95 #endif
96
97 if (
98 #ifdef INET6
99 isipv6 ? in6_localip(&inp->in6p_faddr) :
100 #endif
101 #ifdef INET
102 in_localip(inp->inp_faddr))
103 #else
104 false)
105 #endif
106 return (V_tcp_msl_local);
107 else
108 return (V_tcp_msl);
109 }
110
111 /*
112 * Move a TCP connection into TIME_WAIT state.
113 * inp is locked, and is unlocked before returning.
114 *
115 * This function used to free tcpcb and allocate a compressed TCP time-wait
116 * structure tcptw. This served well for 20 years but is no longer relevant
117 * on modern machines in the modern internet. However, the function remains
118 * so that TCP stacks require less modification and we don't burn the bridge
119 * to go back to using compressed time-wait.
120 */
121 void
tcp_twstart(struct tcpcb * tp)122 tcp_twstart(struct tcpcb *tp)
123 {
124 struct inpcb *inp = tptoinpcb(tp);
125
126 NET_EPOCH_ASSERT();
127 INP_WLOCK_ASSERT(inp);
128
129 /* A dropped inp should never transition to TIME_WAIT state. */
130 KASSERT((inp->inp_flags & INP_DROPPED) == 0, ("tcp_twstart: "
131 "(inp->inp_flags & INP_DROPPED) != 0"));
132
133 tcp_state_change(tp, TCPS_TIME_WAIT);
134 tcp_free_sackholes(tp);
135 soisdisconnected(inp->inp_socket);
136
137 if (tp->t_flags & TF_ACKNOW)
138 (void) tcp_output(tp);
139
140 tcp_timer_activate(tp, TT_2MSL, 2 * tcp_eff_msl(tp));
141 INP_WUNLOCK(inp);
142 }
143
144 /*
145 * Returns true if the TIME_WAIT state was killed and we should start over,
146 * looking for a pcb in the listen state. Otherwise returns false and frees
147 * the mbuf.
148 *
149 * For pure SYN-segments the PCB shall be read-locked and the tcpopt pointer
150 * may be NULL. For the rest write-lock and valid tcpopt.
151 */
152 bool
tcp_twcheck(struct inpcb * inp,struct tcpopt * to,struct tcphdr * th,struct mbuf * m,int tlen)153 tcp_twcheck(struct inpcb *inp, struct tcpopt *to, struct tcphdr *th,
154 struct mbuf *m, int tlen)
155 {
156 struct tcpcb *tp = intotcpcb(inp);
157 char *s;
158 int thflags;
159 tcp_seq seq;
160
161 NET_EPOCH_ASSERT();
162 INP_LOCK_ASSERT(inp);
163
164 thflags = tcp_get_flags(th);
165 #ifdef INVARIANTS
166 if ((thflags & (TH_SYN | TH_ACK)) == TH_SYN)
167 INP_RLOCK_ASSERT(inp);
168 else {
169 INP_WLOCK_ASSERT(inp);
170 KASSERT(to != NULL,
171 ("%s: called without options on a non-SYN segment",
172 __func__));
173 }
174 #endif
175
176 /*
177 * NOTE: for FIN_WAIT_2 (to be added later),
178 * must validate sequence number before accepting RST
179 */
180
181 /*
182 * If the segment contains RST:
183 * Drop the segment - see Stevens, vol. 2, p. 964 and
184 * RFC 1337.
185 */
186 if (thflags & TH_RST)
187 goto drop;
188
189 #if 0
190 /* PAWS not needed at the moment */
191 /*
192 * RFC 1323 PAWS: If we have a timestamp reply on this segment
193 * and it's less than ts_recent, drop it.
194 */
195 if ((to.to_flags & TOF_TS) != 0 && tp->ts_recent &&
196 TSTMP_LT(to.to_tsval, tp->ts_recent)) {
197 if ((thflags & TH_ACK) == 0)
198 goto drop;
199 goto ack;
200 }
201 /*
202 * ts_recent is never updated because we never accept new segments.
203 */
204 #endif
205
206 /* Honor the drop_synfin sysctl variable. */
207 if ((thflags & TH_SYN) && (thflags & TH_FIN) && V_drop_synfin) {
208 if ((s = tcp_log_addrs(&inp->inp_inc, th, NULL, NULL))) {
209 log(LOG_DEBUG, "%s; %s: "
210 "SYN|FIN segment ignored (based on "
211 "sysctl setting)\n", s, __func__);
212 free(s, M_TCPLOG);
213 }
214 goto drop;
215 }
216
217 /*
218 * If a new connection request is received
219 * while in TIME_WAIT, drop the old connection
220 * and start over if the sequence numbers
221 * are above the previous ones.
222 * Allow UDP port number changes in this case.
223 */
224 if (((thflags & (TH_SYN | TH_ACK)) == TH_SYN) &&
225 SEQ_GT(th->th_seq, tp->rcv_nxt)) {
226 /*
227 * In case we can't upgrade our lock just pretend we have
228 * lost this packet.
229 */
230 if (INP_TRY_UPGRADE(inp) == 0)
231 goto drop;
232 if ((tp = tcp_close(tp)) != NULL)
233 INP_WUNLOCK(inp);
234 TCPSTAT_INC(tcps_tw_recycles);
235 return (true);
236 }
237
238 /*
239 * Send RST if UDP port numbers don't match
240 */
241 if (tp->t_port != m->m_pkthdr.tcp_tun_port) {
242 if (tcp_get_flags(th) & TH_ACK) {
243 tcp_respond(tp, mtod(m, void *), th, m,
244 (tcp_seq)0, th->th_ack, TH_RST);
245 } else {
246 if (tcp_get_flags(th) & TH_SYN)
247 tlen++;
248 if (tcp_get_flags(th) & TH_FIN)
249 tlen++;
250 tcp_respond(tp, mtod(m, void *), th, m,
251 th->th_seq+tlen, (tcp_seq)0, TH_RST|TH_ACK);
252 }
253 INP_UNLOCK(inp);
254 TCPSTAT_INC(tcps_tw_resets);
255 return (false);
256 }
257
258 /*
259 * Drop the segment if it does not contain an ACK.
260 */
261 if ((thflags & TH_ACK) == 0)
262 goto drop;
263
264 INP_WLOCK_ASSERT(inp);
265
266 /*
267 * If timestamps were negotiated during SYN/ACK and a
268 * segment without a timestamp is received, silently drop
269 * the segment, unless the missing timestamps are tolerated.
270 * See section 3.2 of RFC 7323.
271 */
272 if (((to->to_flags & TOF_TS) == 0) && (tp->ts_recent != 0) &&
273 (V_tcp_tolerate_missing_ts == 0)) {
274 goto drop;
275 }
276
277 /*
278 * Reset the 2MSL timer if this is a duplicate FIN.
279 */
280 if (thflags & TH_FIN) {
281 seq = th->th_seq + tlen + (thflags & TH_SYN ? 1 : 0);
282 if (seq + 1 == tp->rcv_nxt)
283 tcp_timer_activate(tp, TT_2MSL, 2 * tcp_eff_msl(tp));
284 }
285
286 /*
287 * Acknowledge the segment if it has data or is not a duplicate ACK.
288 */
289 if (thflags != TH_ACK || tlen != 0 ||
290 th->th_seq != tp->rcv_nxt || th->th_ack != tp->snd_nxt) {
291 TCP_PROBE5(receive, NULL, NULL, m, NULL, th);
292 tcp_respond(tp, mtod(m, void *), th, m, tp->rcv_nxt,
293 tp->snd_nxt, TH_ACK);
294 INP_UNLOCK(inp);
295 TCPSTAT_INC(tcps_tw_responds);
296 return (false);
297 }
298 drop:
299 TCP_PROBE5(receive, NULL, NULL, m, NULL, th);
300 INP_UNLOCK(inp);
301 m_freem(m);
302 return (false);
303 }
304