1 /*- 2 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994, 1995 3 * The Regents of the University of California. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 4. Neither the name of the University nor the names of its contributors 14 * may be used to endorse or promote products derived from this software 15 * without specific prior written permission. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 * 29 * @(#)tcp_input.c 8.12 (Berkeley) 5/24/95 30 */ 31 32 #include <sys/cdefs.h> 33 __FBSDID("$FreeBSD$"); 34 35 #include "opt_inet.h" 36 #include "opt_inet6.h" 37 #include "opt_tcpdebug.h" 38 39 #include <sys/param.h> 40 #include <sys/kernel.h> 41 #include <sys/malloc.h> 42 #include <sys/mbuf.h> 43 #include <sys/socket.h> 44 #include <sys/socketvar.h> 45 #include <sys/sysctl.h> 46 #include <sys/syslog.h> 47 #include <sys/systm.h> 48 #include <sys/vimage.h> 49 50 #include <vm/uma.h> 51 52 #include <net/if.h> 53 #include <net/route.h> 54 55 #include <netinet/in.h> 56 #include <netinet/in_pcb.h> 57 #include <netinet/in_systm.h> 58 #include <netinet/in_var.h> 59 #include <netinet/ip.h> 60 #include <netinet/ip_var.h> 61 #include <netinet/ip_options.h> 62 #include <netinet/ip6.h> 63 #include <netinet6/in6_pcb.h> 64 #include <netinet6/ip6_var.h> 65 #include <netinet6/nd6.h> 66 #include <netinet/tcp.h> 67 #include <netinet/tcp_fsm.h> 68 #include <netinet/tcp_seq.h> 69 #include <netinet/tcp_timer.h> 70 #include <netinet/tcp_var.h> 71 #include <netinet6/tcp6_var.h> 72 #include <netinet/tcpip.h> 73 #ifdef TCPDEBUG 74 #include <netinet/tcp_debug.h> 75 #endif /* TCPDEBUG */ 76 #include <netinet/vinet.h> 77 78 #ifdef VIMAGE_GLOBALS 79 static int tcp_reass_maxseg; 80 int tcp_reass_qsize; 81 static int tcp_reass_maxqlen; 82 static int tcp_reass_overflows; 83 #endif 84 85 SYSCTL_NODE(_net_inet_tcp, OID_AUTO, reass, CTLFLAG_RW, 0, 86 "TCP Segment Reassembly Queue"); 87 88 SYSCTL_V_INT(V_NET, vnet_inet, _net_inet_tcp_reass, OID_AUTO, maxsegments, 89 CTLFLAG_RDTUN, tcp_reass_maxseg, 0, 90 "Global maximum number of TCP Segments in Reassembly Queue"); 91 92 SYSCTL_V_INT(V_NET, vnet_inet, _net_inet_tcp_reass, OID_AUTO, cursegments, 93 CTLFLAG_RD, tcp_reass_qsize, 0, 94 "Global number of TCP Segments currently in Reassembly Queue"); 95 96 SYSCTL_V_INT(V_NET, vnet_inet, _net_inet_tcp_reass, OID_AUTO, maxqlen, 97 CTLFLAG_RW, tcp_reass_maxqlen, 0, 98 "Maximum number of TCP Segments per individual Reassembly Queue"); 99 100 SYSCTL_V_INT(V_NET, vnet_inet, _net_inet_tcp_reass, OID_AUTO, overflows, 101 CTLFLAG_RD, tcp_reass_overflows, 0, 102 "Global number of TCP Segment Reassembly Queue Overflows"); 103 104 /* Initialize TCP reassembly queue */ 105 static void 106 tcp_reass_zone_change(void *tag) 107 { 108 INIT_VNET_INET(curvnet); 109 110 V_tcp_reass_maxseg = nmbclusters / 16; 111 uma_zone_set_max(V_tcp_reass_zone, V_tcp_reass_maxseg); 112 } 113 114 #ifdef VIMAGE_GLOBALS 115 uma_zone_t tcp_reass_zone; 116 #endif 117 118 void 119 tcp_reass_init(void) 120 { 121 INIT_VNET_INET(curvnet); 122 123 V_tcp_reass_maxseg = 0; 124 V_tcp_reass_qsize = 0; 125 V_tcp_reass_maxqlen = 48; 126 V_tcp_reass_overflows = 0; 127 128 V_tcp_reass_maxseg = nmbclusters / 16; 129 TUNABLE_INT_FETCH("net.inet.tcp.reass.maxsegments", 130 &V_tcp_reass_maxseg); 131 V_tcp_reass_zone = uma_zcreate("tcpreass", sizeof (struct tseg_qent), 132 NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); 133 uma_zone_set_max(V_tcp_reass_zone, V_tcp_reass_maxseg); 134 EVENTHANDLER_REGISTER(nmbclusters_change, 135 tcp_reass_zone_change, NULL, EVENTHANDLER_PRI_ANY); 136 } 137 138 int 139 tcp_reass(struct tcpcb *tp, struct tcphdr *th, int *tlenp, struct mbuf *m) 140 { 141 INIT_VNET_INET(curvnet); 142 struct tseg_qent *q; 143 struct tseg_qent *p = NULL; 144 struct tseg_qent *nq; 145 struct tseg_qent *te = NULL; 146 struct socket *so = tp->t_inpcb->inp_socket; 147 int flags; 148 149 INP_WLOCK_ASSERT(tp->t_inpcb); 150 151 /* 152 * XXX: tcp_reass() is rather inefficient with its data structures 153 * and should be rewritten (see NetBSD for optimizations). 154 */ 155 156 /* 157 * Call with th==NULL after become established to 158 * force pre-ESTABLISHED data up to user socket. 159 */ 160 if (th == NULL) 161 goto present; 162 163 /* 164 * Limit the number of segments in the reassembly queue to prevent 165 * holding on to too many segments (and thus running out of mbufs). 166 * Make sure to let the missing segment through which caused this 167 * queue. Always keep one global queue entry spare to be able to 168 * process the missing segment. 169 */ 170 if (th->th_seq != tp->rcv_nxt && 171 (V_tcp_reass_qsize + 1 >= V_tcp_reass_maxseg || 172 tp->t_segqlen >= V_tcp_reass_maxqlen)) { 173 V_tcp_reass_overflows++; 174 TCPSTAT_INC(tcps_rcvmemdrop); 175 m_freem(m); 176 *tlenp = 0; 177 return (0); 178 } 179 180 /* 181 * Allocate a new queue entry. If we can't, or hit the zone limit 182 * just drop the pkt. 183 */ 184 te = uma_zalloc(V_tcp_reass_zone, M_NOWAIT); 185 if (te == NULL) { 186 TCPSTAT_INC(tcps_rcvmemdrop); 187 m_freem(m); 188 *tlenp = 0; 189 return (0); 190 } 191 tp->t_segqlen++; 192 V_tcp_reass_qsize++; 193 194 /* 195 * Find a segment which begins after this one does. 196 */ 197 LIST_FOREACH(q, &tp->t_segq, tqe_q) { 198 if (SEQ_GT(q->tqe_th->th_seq, th->th_seq)) 199 break; 200 p = q; 201 } 202 203 /* 204 * If there is a preceding segment, it may provide some of 205 * our data already. If so, drop the data from the incoming 206 * segment. If it provides all of our data, drop us. 207 */ 208 if (p != NULL) { 209 int i; 210 /* conversion to int (in i) handles seq wraparound */ 211 i = p->tqe_th->th_seq + p->tqe_len - th->th_seq; 212 if (i > 0) { 213 if (i >= *tlenp) { 214 TCPSTAT_INC(tcps_rcvduppack); 215 TCPSTAT_ADD(tcps_rcvdupbyte, *tlenp); 216 m_freem(m); 217 uma_zfree(V_tcp_reass_zone, te); 218 tp->t_segqlen--; 219 V_tcp_reass_qsize--; 220 /* 221 * Try to present any queued data 222 * at the left window edge to the user. 223 * This is needed after the 3-WHS 224 * completes. 225 */ 226 goto present; /* ??? */ 227 } 228 m_adj(m, i); 229 *tlenp -= i; 230 th->th_seq += i; 231 } 232 } 233 TCPSTAT_INC(tcps_rcvoopack); 234 TCPSTAT_ADD(tcps_rcvoobyte, *tlenp); 235 236 /* 237 * While we overlap succeeding segments trim them or, 238 * if they are completely covered, dequeue them. 239 */ 240 while (q) { 241 int i = (th->th_seq + *tlenp) - q->tqe_th->th_seq; 242 if (i <= 0) 243 break; 244 if (i < q->tqe_len) { 245 q->tqe_th->th_seq += i; 246 q->tqe_len -= i; 247 m_adj(q->tqe_m, i); 248 break; 249 } 250 251 nq = LIST_NEXT(q, tqe_q); 252 LIST_REMOVE(q, tqe_q); 253 m_freem(q->tqe_m); 254 uma_zfree(V_tcp_reass_zone, q); 255 tp->t_segqlen--; 256 V_tcp_reass_qsize--; 257 q = nq; 258 } 259 260 /* Insert the new segment queue entry into place. */ 261 te->tqe_m = m; 262 te->tqe_th = th; 263 te->tqe_len = *tlenp; 264 265 if (p == NULL) { 266 LIST_INSERT_HEAD(&tp->t_segq, te, tqe_q); 267 } else { 268 LIST_INSERT_AFTER(p, te, tqe_q); 269 } 270 271 present: 272 /* 273 * Present data to user, advancing rcv_nxt through 274 * completed sequence space. 275 */ 276 if (!TCPS_HAVEESTABLISHED(tp->t_state)) 277 return (0); 278 q = LIST_FIRST(&tp->t_segq); 279 if (!q || q->tqe_th->th_seq != tp->rcv_nxt) 280 return (0); 281 SOCKBUF_LOCK(&so->so_rcv); 282 do { 283 tp->rcv_nxt += q->tqe_len; 284 flags = q->tqe_th->th_flags & TH_FIN; 285 nq = LIST_NEXT(q, tqe_q); 286 LIST_REMOVE(q, tqe_q); 287 if (so->so_rcv.sb_state & SBS_CANTRCVMORE) 288 m_freem(q->tqe_m); 289 else 290 sbappendstream_locked(&so->so_rcv, q->tqe_m); 291 uma_zfree(V_tcp_reass_zone, q); 292 tp->t_segqlen--; 293 V_tcp_reass_qsize--; 294 q = nq; 295 } while (q && q->tqe_th->th_seq == tp->rcv_nxt); 296 ND6_HINT(tp); 297 sorwakeup_locked(so); 298 return (flags); 299 } 300