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/tcp_log_buf.h>
79 #include <netinet/tcpip.h>
80
81 #include <netinet/udp.h>
82 #include <netinet/udp_var.h>
83
84 #include <netipsec/ipsec_support.h>
85
86 #include <machine/in_cksum.h>
87
88 #include <security/mac/mac_framework.h>
89
90 static u_int
tcp_eff_msl(struct tcpcb * tp)91 tcp_eff_msl(struct tcpcb *tp)
92 {
93 struct inpcb *inp = tptoinpcb(tp);
94 #ifdef INET6
95 bool isipv6 = inp->inp_inc.inc_flags & INC_ISIPV6;
96 #endif
97
98 if (
99 #ifdef INET6
100 isipv6 ? in6_localip(&inp->in6p_faddr) :
101 #endif
102 #ifdef INET
103 in_localip(inp->inp_faddr))
104 #else
105 false)
106 #endif
107 return (V_tcp_msl_local);
108 else
109 return (V_tcp_msl);
110 }
111
112 /*
113 * Move a TCP connection into TIME_WAIT state.
114 * inp is locked, and is unlocked before returning.
115 *
116 * This function used to free tcpcb and allocate a compressed TCP time-wait
117 * structure tcptw. This served well for 20 years but is no longer relevant
118 * on modern machines in the modern internet. However, the function remains
119 * so that TCP stacks require less modification and we don't burn the bridge
120 * to go back to using compressed time-wait.
121 */
122 void
tcp_twstart(struct tcpcb * tp)123 tcp_twstart(struct tcpcb *tp)
124 {
125 struct inpcb *inp = tptoinpcb(tp);
126
127 NET_EPOCH_ASSERT();
128 INP_WLOCK_ASSERT(inp);
129
130 /* A dropped inp should never transition to TIME_WAIT state. */
131 KASSERT((inp->inp_flags & INP_DROPPED) == 0, ("tcp_twstart: "
132 "(inp->inp_flags & INP_DROPPED) != 0"));
133
134 tcp_state_change(tp, TCPS_TIME_WAIT);
135 tcp_free_sackholes(tp);
136 soisdisconnected(inp->inp_socket);
137
138 if (tp->t_flags & TF_ACKNOW)
139 (void) tcp_output(tp);
140
141 tcp_timer_activate(tp, TT_2MSL, 2 * tcp_eff_msl(tp));
142 INP_WUNLOCK(inp);
143 }
144
145 /*
146 * Returns true if the TIME_WAIT state was killed and we should start over,
147 * looking for a pcb in the listen state. Otherwise returns false and frees
148 * the mbuf.
149 *
150 * For pure SYN-segments the PCB shall be read-locked and the tcpopt pointer
151 * may be NULL. For the rest write-lock and valid tcpopt.
152 */
153 bool
tcp_twcheck(struct inpcb * inp,struct tcpopt * to,struct tcphdr * th,struct mbuf * m,int tlen)154 tcp_twcheck(struct inpcb *inp, struct tcpopt *to, struct tcphdr *th,
155 struct mbuf *m, int tlen)
156 {
157 struct tcpcb *tp = intotcpcb(inp);
158 char *s;
159 int thflags;
160 tcp_seq seq;
161
162 NET_EPOCH_ASSERT();
163 INP_LOCK_ASSERT(inp);
164
165 thflags = tcp_get_flags(th);
166 #ifdef INVARIANTS
167 if ((thflags & (TH_SYN | TH_ACK)) == TH_SYN)
168 INP_RLOCK_ASSERT(inp);
169 else {
170 INP_WLOCK_ASSERT(inp);
171 KASSERT(to != NULL,
172 ("%s: called without options on a non-SYN segment",
173 __func__));
174 }
175 #endif
176
177 /*
178 * NOTE: for FIN_WAIT_2 (to be added later),
179 * must validate sequence number before accepting RST
180 */
181
182 /*
183 * If the segment contains RST:
184 * Drop the segment - see Stevens, vol. 2, p. 964 and
185 * RFC 1337.
186 */
187 if (thflags & TH_RST)
188 goto drop;
189
190 #if 0
191 /* PAWS not needed at the moment */
192 /*
193 * RFC 1323 PAWS: If we have a timestamp reply on this segment
194 * and it's less than ts_recent, drop it.
195 */
196 if ((to.to_flags & TOF_TS) != 0 && tp->ts_recent &&
197 TSTMP_LT(to.to_tsval, tp->ts_recent)) {
198 if ((thflags & TH_ACK) == 0)
199 goto drop;
200 goto ack;
201 }
202 /*
203 * ts_recent is never updated because we never accept new segments.
204 */
205 #endif
206
207 /* Honor the drop_synfin sysctl variable. */
208 if ((thflags & TH_SYN) && (thflags & TH_FIN) && V_drop_synfin) {
209 if ((s = tcp_log_addrs(&inp->inp_inc, th, NULL, NULL))) {
210 log(LOG_DEBUG, "%s; %s: "
211 "SYN|FIN segment ignored (based on "
212 "sysctl setting)\n", s, __func__);
213 free(s, M_TCPLOG);
214 }
215 goto drop;
216 }
217
218 /*
219 * If a new connection request is received
220 * while in TIME_WAIT, drop the old connection
221 * and start over if the sequence numbers
222 * are above the previous ones.
223 * Allow UDP port number changes in this case.
224 */
225 if (((thflags & (TH_SYN | TH_ACK)) == TH_SYN) &&
226 SEQ_GT(th->th_seq, tp->rcv_nxt)) {
227 /*
228 * In case we can't upgrade our lock just pretend we have
229 * lost this packet.
230 */
231 if (INP_TRY_UPGRADE(inp) == 0)
232 goto drop;
233 if ((tp = tcp_close(tp)) != NULL)
234 INP_WUNLOCK(inp);
235 TCPSTAT_INC(tcps_tw_recycles);
236 return (true);
237 }
238
239 /*
240 * Send RST if UDP port numbers don't match
241 */
242 if (tp->t_port != m->m_pkthdr.tcp_tun_port) {
243 if (tcp_get_flags(th) & TH_ACK) {
244 tcp_respond(tp, mtod(m, void *), th, m,
245 (tcp_seq)0, th->th_ack, TH_RST);
246 } else {
247 if (tcp_get_flags(th) & TH_SYN)
248 tlen++;
249 if (tcp_get_flags(th) & TH_FIN)
250 tlen++;
251 tcp_respond(tp, mtod(m, void *), th, m,
252 th->th_seq+tlen, (tcp_seq)0, TH_RST|TH_ACK);
253 }
254 INP_UNLOCK(inp);
255 TCPSTAT_INC(tcps_tw_resets);
256 return (false);
257 }
258
259 /*
260 * Drop the segment if it does not contain an ACK.
261 */
262 if ((thflags & TH_ACK) == 0)
263 goto drop;
264
265 INP_WLOCK_ASSERT(inp);
266
267 /*
268 * If timestamps were negotiated during SYN/ACK and a
269 * segment without a timestamp is received, silently drop
270 * the segment, unless the missing timestamps are tolerated.
271 * See section 3.2 of RFC 7323.
272 */
273 if (((to->to_flags & TOF_TS) == 0) && (tp->ts_recent != 0) &&
274 (V_tcp_tolerate_missing_ts == 0)) {
275 goto drop;
276 }
277
278 /*
279 * Reset the 2MSL timer if this is a duplicate FIN.
280 */
281 if (thflags & TH_FIN) {
282 seq = th->th_seq + tlen + (thflags & TH_SYN ? 1 : 0);
283 if (seq + 1 == tp->rcv_nxt)
284 tcp_timer_activate(tp, TT_2MSL, 2 * tcp_eff_msl(tp));
285 }
286
287 /*
288 * Acknowledge the segment if it has data or is not a duplicate ACK.
289 */
290 if ((thflags & (TH_SYN | TH_FIN)) != 0 || tlen != 0 ||
291 th->th_seq != tp->rcv_nxt || th->th_ack != tp->snd_nxt) {
292 TCP_LOG_EVENT(tp, th, NULL, NULL, TCP_LOG_IN, 0, tlen, NULL,
293 true);
294 TCP_PROBE5(receive, NULL, NULL, m, NULL, th);
295 tcp_respond(tp, mtod(m, void *), th, m, tp->rcv_nxt,
296 tp->snd_nxt, TH_ACK);
297 INP_UNLOCK(inp);
298 TCPSTAT_INC(tcps_tw_responds);
299 return (false);
300 }
301 drop:
302 TCP_PROBE5(receive, NULL, NULL, m, NULL, th);
303 INP_UNLOCK(inp);
304 m_freem(m);
305 return (false);
306 }
307