1 /*
2 * CDDL HEADER START
3 *
4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
7 *
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
12 *
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 *
19 * CDDL HEADER END
20 */
21 /*
22 * Copyright 2010 Sun Microsystems, Inc. All rights reserved.
23 * Use is subject to license terms.
24 */
25
26 /*
27 * Copyright 2019 OmniOS Community Edition (OmniOSce) Association.
28 */
29
30 #ifndef _INET_IPCLASSIFIER_H
31 #define _INET_IPCLASSIFIER_H
32
33 #ifdef __cplusplus
34 extern "C" {
35 #endif
36
37 #include <inet/common.h>
38 #include <inet/ip.h>
39 #include <inet/mi.h>
40 #include <inet/tcp.h>
41 #include <inet/ip6.h>
42 #include <netinet/in.h> /* for IPPROTO_* constants */
43 #include <sys/sdt.h>
44 #include <sys/socket_proto.h>
45 #include <sys/sunddi.h>
46 #include <sys/sunldi.h>
47
48 typedef void (*edesc_rpf)(void *, mblk_t *, void *, ip_recv_attr_t *);
49 struct icmph_s;
50 struct icmp6_hdr;
51 typedef boolean_t (*edesc_vpf)(conn_t *, void *, struct icmph_s *,
52 struct icmp6_hdr *, ip_recv_attr_t *);
53
54 /*
55 * ==============================
56 * = The CONNECTION =
57 * ==============================
58 */
59
60 /*
61 * The connection structure contains the common information/flags/ref needed.
62 * Implementation will keep the connection struct, the layers (with their
63 * respective data for event i.e. tcp_t if event was tcp_input_data) all in one
64 * contiguous memory location.
65 */
66
67 /* Conn Flags */
68 /* Unused 0x00020000 */
69 /* Unused 0x00040000 */
70 #define IPCL_FULLY_BOUND 0x00080000 /* Bound to correct squeue */
71 /* Unused 0x00100000 */
72 /* Unused 0x00200000 */
73 /* Unused 0x00400000 */
74 #define IPCL_CL_LISTENER 0x00800000 /* Cluster listener */
75 /* Unused 0x01000000 */
76 /* Unused 0x02000000 */
77 /* Unused 0x04000000 */
78 /* Unused 0x08000000 */
79 /* Unused 0x10000000 */
80 /* Unused 0x20000000 */
81 #define IPCL_CONNECTED 0x40000000 /* Conn in connected table */
82 #define IPCL_BOUND 0x80000000 /* Conn in bind table */
83
84 /* Flags identifying the type of conn */
85 #define IPCL_TCPCONN 0x00000001 /* From tcp_conn_cache */
86 #define IPCL_SCTPCONN 0x00000002 /* From sctp_conn_cache */
87 #define IPCL_IPCCONN 0x00000004 /* From ip_conn_cache */
88 #define IPCL_UDPCONN 0x00000008 /* From udp_conn_cache */
89 #define IPCL_RAWIPCONN 0x00000010 /* From rawip_conn_cache */
90 #define IPCL_RTSCONN 0x00000020 /* From rts_conn_cache */
91 /* Unused 0x00000040 */
92 #define IPCL_IPTUN 0x00000080 /* iptun module above us */
93
94 #define IPCL_NONSTR 0x00001000 /* A non-STREAMS socket */
95 /* Unused 0x10000000 */
96
97 #define IPCL_REMOVED 0x00000100
98 #define IPCL_REUSED 0x00000200
99
100 #define IPCL_IS_CONNECTED(connp) \
101 ((connp)->conn_flags & IPCL_CONNECTED)
102
103 #define IPCL_IS_BOUND(connp) \
104 ((connp)->conn_flags & IPCL_BOUND)
105
106 /*
107 * Can't use conn_proto since we need to tell difference
108 * between a real TCP socket and a SOCK_RAW, IPPROTO_TCP.
109 */
110 #define IPCL_IS_TCP(connp) \
111 ((connp)->conn_flags & IPCL_TCPCONN)
112
113 #define IPCL_IS_SCTP(connp) \
114 ((connp)->conn_flags & IPCL_SCTPCONN)
115
116 #define IPCL_IS_UDP(connp) \
117 ((connp)->conn_flags & IPCL_UDPCONN)
118
119 #define IPCL_IS_RAWIP(connp) \
120 ((connp)->conn_flags & IPCL_RAWIPCONN)
121
122 #define IPCL_IS_RTS(connp) \
123 ((connp)->conn_flags & IPCL_RTSCONN)
124
125 #define IPCL_IS_IPTUN(connp) \
126 ((connp)->conn_flags & IPCL_IPTUN)
127
128 #define IPCL_IS_NONSTR(connp) ((connp)->conn_flags & IPCL_NONSTR)
129
130 typedef struct connf_s connf_t;
131
132 typedef struct
133 {
134 int ctb_depth;
135 #define CONN_STACK_DEPTH 15
136 pc_t ctb_stack[CONN_STACK_DEPTH];
137 } conn_trace_t;
138
139 typedef struct ip_helper_minor_info_s {
140 dev_t ip_minfo_dev; /* Device */
141 vmem_t *ip_minfo_arena; /* Arena */
142 } ip_helper_minfo_t;
143
144 /*
145 * ip helper stream info
146 */
147 typedef struct ip_helper_stream_info_s {
148 ldi_handle_t iphs_handle;
149 queue_t *iphs_rq;
150 queue_t *iphs_wq;
151 ip_helper_minfo_t *iphs_minfo;
152 } ip_helper_stream_info_t;
153
154 /*
155 * Mandatory Access Control mode, in conn_t's conn_mac_mode field.
156 * CONN_MAC_DEFAULT: strict enforcement of MAC.
157 * CONN_MAC_AWARE: allows communications between unlabeled systems
158 * and privileged daemons
159 * CONN_MAC_IMPLICIT: allows communications without explicit labels
160 * on the wire with privileged daemons.
161 *
162 * CONN_MAC_IMPLICIT is intended specifically for labeled IPsec key management
163 * in networks which don't pass CIPSO-labeled packets.
164 */
165 #define CONN_MAC_DEFAULT 0
166 #define CONN_MAC_AWARE 1
167 #define CONN_MAC_IMPLICIT 2
168
169 /*
170 * conn receive ancillary definition.
171 *
172 * These are the set of socket options that make the receive side
173 * potentially pass up ancillary data items.
174 * We have a union with an integer so that we can quickly check whether
175 * any ancillary data items need to be added.
176 */
177 typedef struct crb_s {
178 union {
179 uint32_t crbu_all;
180 struct {
181 uint32_t
182 crbb_recvdstaddr : 1, /* IP_RECVDSTADDR option */
183 crbb_recvopts : 1, /* IP_RECVOPTS option */
184 crbb_recvif : 1, /* IP_RECVIF option */
185 crbb_recvslla : 1, /* IP_RECVSLLA option */
186
187 crbb_recvttl : 1, /* IP_RECVTTL option */
188 crbb_ip_recvpktinfo : 1, /* IP*_RECVPKTINFO option */
189 crbb_ipv6_recvhoplimit : 1, /* IPV6_RECVHOPLIMIT option */
190 crbb_ipv6_recvhopopts : 1, /* IPV6_RECVHOPOPTS option */
191
192 crbb_ipv6_recvdstopts : 1, /* IPV6_RECVDSTOPTS option */
193 crbb_ipv6_recvrthdr : 1, /* IPV6_RECVRTHDR option */
194 crbb_old_ipv6_recvdstopts : 1, /* old form of IPV6_DSTOPTS */
195 crbb_ipv6_recvrthdrdstopts : 1, /* IPV6_RECVRTHDRDSTOPTS */
196
197 crbb_ipv6_recvtclass : 1, /* IPV6_RECVTCLASS */
198 crbb_recvucred : 1, /* IP_RECVUCRED option */
199 crbb_timestamp : 1; /* SO_TIMESTAMP "socket" option */
200
201 } crbb;
202 } crbu;
203 } crb_t;
204
205 #define crb_all crbu.crbu_all
206 #define crb_recvdstaddr crbu.crbb.crbb_recvdstaddr
207 #define crb_recvopts crbu.crbb.crbb_recvopts
208 #define crb_recvif crbu.crbb.crbb_recvif
209 #define crb_recvslla crbu.crbb.crbb_recvslla
210 #define crb_recvttl crbu.crbb.crbb_recvttl
211 #define crb_ip_recvpktinfo crbu.crbb.crbb_ip_recvpktinfo
212 #define crb_ipv6_recvhoplimit crbu.crbb.crbb_ipv6_recvhoplimit
213 #define crb_ipv6_recvhopopts crbu.crbb.crbb_ipv6_recvhopopts
214 #define crb_ipv6_recvdstopts crbu.crbb.crbb_ipv6_recvdstopts
215 #define crb_ipv6_recvrthdr crbu.crbb.crbb_ipv6_recvrthdr
216 #define crb_old_ipv6_recvdstopts crbu.crbb.crbb_old_ipv6_recvdstopts
217 #define crb_ipv6_recvrthdrdstopts crbu.crbb.crbb_ipv6_recvrthdrdstopts
218 #define crb_ipv6_recvtclass crbu.crbb.crbb_ipv6_recvtclass
219 #define crb_recvucred crbu.crbb.crbb_recvucred
220 #define crb_timestamp crbu.crbb.crbb_timestamp
221
222 /*
223 * The initial fields in the conn_t are setup by the kmem_cache constructor,
224 * and are preserved when it is freed. Fields after that are bzero'ed when
225 * the conn_t is freed.
226 *
227 * Much of the conn_t is protected by conn_lock.
228 *
229 * conn_lock is also used by some ULPs (like UDP and RAWIP) to protect
230 * their state.
231 */
232 struct conn_s {
233 kmutex_t conn_lock;
234 uint32_t conn_ref; /* Reference counter */
235 uint32_t conn_flags; /* Conn Flags */
236
237 union {
238 tcp_t *cp_tcp; /* Pointer to the tcp struct */
239 struct udp_s *cp_udp; /* Pointer to the udp struct */
240 struct icmp_s *cp_icmp; /* Pointer to rawip struct */
241 struct rts_s *cp_rts; /* Pointer to rts struct */
242 struct iptun_s *cp_iptun; /* Pointer to iptun_t */
243 struct sctp_s *cp_sctp; /* For IPCL_SCTPCONN */
244 void *cp_priv;
245 } conn_proto_priv;
246 #define conn_tcp conn_proto_priv.cp_tcp
247 #define conn_udp conn_proto_priv.cp_udp
248 #define conn_icmp conn_proto_priv.cp_icmp
249 #define conn_rts conn_proto_priv.cp_rts
250 #define conn_iptun conn_proto_priv.cp_iptun
251 #define conn_sctp conn_proto_priv.cp_sctp
252 #define conn_priv conn_proto_priv.cp_priv
253
254 kcondvar_t conn_cv;
255 uint8_t conn_proto; /* protocol type */
256
257 edesc_rpf conn_recv; /* Pointer to recv routine */
258 edesc_rpf conn_recvicmp; /* For ICMP error */
259 edesc_vpf conn_verifyicmp; /* Verify ICMP error */
260
261 ip_xmit_attr_t *conn_ixa; /* Options if no ancil data */
262
263 /* Fields after this are bzero'ed when the conn_t is freed. */
264 #define conn_start_clr conn_recv_ancillary
265
266 /* Options for receive-side ancillary data */
267 crb_t conn_recv_ancillary;
268
269 squeue_t *conn_sqp; /* Squeue for processing */
270 uint_t conn_state_flags; /* IP state flags */
271
272 int conn_lingertime; /* linger time (in seconds) */
273
274 unsigned int
275 conn_on_sqp : 1, /* Conn is being processed */
276 conn_linger : 1, /* SO_LINGER state */
277 conn_useloopback : 1, /* SO_USELOOPBACK state */
278 conn_broadcast : 1, /* SO_BROADCAST state */
279
280 conn_reuseaddr : 1, /* SO_REUSEADDR state */
281 conn_keepalive : 1, /* SO_KEEPALIVE state */
282 conn_multi_router : 1, /* Wants all multicast pkts */
283 conn_unspec_src : 1, /* IP_UNSPEC_SRC */
284
285 conn_policy_cached : 1, /* Is policy cached/latched ? */
286 conn_in_enforce_policy : 1, /* Enforce Policy on inbound */
287 conn_out_enforce_policy : 1, /* Enforce Policy on outbound */
288 conn_debug : 1, /* SO_DEBUG */
289
290 conn_ipv6_v6only : 1, /* IPV6_V6ONLY */
291 conn_oobinline : 1, /* SO_OOBINLINE state */
292 conn_dgram_errind : 1, /* SO_DGRAM_ERRIND state */
293 conn_exclbind : 1, /* SO_EXCLBIND state */
294
295 conn_mdt_ok : 1, /* MDT is permitted */
296 conn_allzones : 1, /* SO_ALLZONES */
297 conn_ipv6_recvpathmtu : 1, /* IPV6_RECVPATHMTU */
298 conn_mcbc_bind : 1, /* Bound to multi/broadcast */
299
300 conn_pad_to_bit_31 : 12;
301
302 boolean_t conn_blocked; /* conn is flow-controlled */
303
304 squeue_t *conn_initial_sqp; /* Squeue at open time */
305 squeue_t *conn_final_sqp; /* Squeue after connect */
306 ill_t *conn_dhcpinit_ill; /* IP_DHCPINIT_IF */
307 ipsec_latch_t *conn_latch; /* latched IDS */
308 struct ipsec_policy_s *conn_latch_in_policy; /* latched policy (in) */
309 struct ipsec_action_s *conn_latch_in_action; /* latched action (in) */
310 uint_t conn_bound_if; /* IP*_BOUND_IF */
311 queue_t *conn_rq; /* Read queue */
312 queue_t *conn_wq; /* Write queue */
313 dev_t conn_dev; /* Minor number */
314 vmem_t *conn_minor_arena; /* Minor arena */
315 ip_helper_stream_info_t *conn_helper_info;
316
317 cred_t *conn_cred; /* Credentials */
318 pid_t conn_cpid; /* pid from open/connect */
319 uint64_t conn_open_time; /* time when this was opened */
320
321 connf_t *conn_g_fanout; /* Global Hash bucket head */
322 struct conn_s *conn_g_next; /* Global Hash chain next */
323 struct conn_s *conn_g_prev; /* Global Hash chain prev */
324 struct ipsec_policy_head_s *conn_policy; /* Configured policy */
325 in6_addr_t conn_bound_addr_v6; /* Address in bind() */
326 #define conn_bound_addr_v4 V4_PART_OF_V6(conn_bound_addr_v6)
327 connf_t *conn_fanout; /* Hash bucket we're part of */
328 struct conn_s *conn_next; /* Hash chain next */
329 struct conn_s *conn_prev; /* Hash chain prev */
330
331 struct {
332 in6_addr_t connua_laddr; /* Local address - match */
333 in6_addr_t connua_faddr; /* Remote address */
334 } connua_v6addr;
335 #define conn_laddr_v4 V4_PART_OF_V6(connua_v6addr.connua_laddr)
336 #define conn_faddr_v4 V4_PART_OF_V6(connua_v6addr.connua_faddr)
337 #define conn_laddr_v6 connua_v6addr.connua_laddr
338 #define conn_faddr_v6 connua_v6addr.connua_faddr
339 in6_addr_t conn_saddr_v6; /* Local address - source */
340 #define conn_saddr_v4 V4_PART_OF_V6(conn_saddr_v6)
341
342 union {
343 /* Used for classifier match performance */
344 uint32_t connu_ports2;
345 struct {
346 in_port_t connu_fport; /* Remote port */
347 in_port_t connu_lport; /* Local port */
348 } connu_ports;
349 } u_port;
350 #define conn_fport u_port.connu_ports.connu_fport
351 #define conn_lport u_port.connu_ports.connu_lport
352 #define conn_ports u_port.connu_ports2
353
354 uint_t conn_incoming_ifindex; /* IP{,V6}_BOUND_IF, scopeid */
355 ill_t *conn_oper_pending_ill; /* pending shared ioctl */
356
357 krwlock_t conn_ilg_lock; /* Protects conn_ilg_* */
358 ilg_t *conn_ilg; /* Group memberships */
359
360 kcondvar_t conn_refcv; /* For conn_oper_pending_ill */
361
362 struct conn_s *conn_drain_next; /* Next conn in drain list */
363 struct conn_s *conn_drain_prev; /* Prev conn in drain list */
364 idl_t *conn_idl; /* Ptr to the drain list head */
365 mblk_t *conn_ipsec_opt_mp; /* ipsec option mblk */
366 zoneid_t conn_zoneid; /* zone connection is in */
367 int conn_rtaware; /* RT_AWARE sockopt value */
368 kcondvar_t conn_sq_cv; /* For non-STREAMS socket IO */
369 sock_upcalls_t *conn_upcalls; /* Upcalls to sockfs */
370 sock_upper_handle_t conn_upper_handle; /* Upper handle: sonode * */
371
372 unsigned int
373 conn_mlp_type : 2, /* mlp_type_t; tsol/tndb.h */
374 conn_anon_mlp : 1, /* user wants anon MLP */
375 conn_anon_port : 1, /* user bound anonymously */
376
377 conn_mac_mode : 2, /* normal/loose/implicit MAC */
378 conn_anon_priv_bind : 1, /* *_ANON_PRIV_BIND state */
379 conn_zone_is_global : 1, /* GLOBAL_ZONEID */
380 conn_isvrrp : 1, /* VRRP control socket */
381 conn_spare : 23;
382
383 boolean_t conn_flow_cntrld;
384 netstack_t *conn_netstack; /* Corresponds to a netstack_hold */
385
386 /*
387 * IP format that packets received for this struct should use.
388 * Value can be IP4_VERSION or IPV6_VERSION.
389 * The sending version is encoded using IXAF_IS_IPV4.
390 */
391 ushort_t conn_ipversion;
392
393 /* Written to only once at the time of opening the endpoint */
394 sa_family_t conn_family; /* Family from socket() call */
395 uint_t conn_so_type; /* Type from socket() call */
396
397 uint_t conn_sndbuf; /* SO_SNDBUF state */
398 uint_t conn_rcvbuf; /* SO_RCVBUF state */
399 uint_t conn_wroff; /* Current write offset */
400
401 uint_t conn_sndlowat; /* Send buffer low water mark */
402 uint_t conn_rcvlowat; /* Recv buffer low water mark */
403
404 uint8_t conn_default_ttl; /* Default TTL/hoplimit */
405
406 uint32_t conn_flowinfo; /* Connected flow id and tclass */
407
408 /*
409 * The most recent address for sendto. Initially set to zero
410 * which is always different than then the destination address
411 * since the send interprets zero as the loopback address.
412 */
413 in6_addr_t conn_v6lastdst;
414 #define conn_v4lastdst V4_PART_OF_V6(conn_v6lastdst)
415 ushort_t conn_lastipversion;
416 in_port_t conn_lastdstport;
417 uint32_t conn_lastflowinfo; /* IPv6-only */
418 uint_t conn_lastscopeid; /* IPv6-only */
419 uint_t conn_lastsrcid; /* Only for AF_INET6 */
420 /*
421 * When we are not connected conn_saddr might be unspecified.
422 * We track the source that was used with conn_v6lastdst here.
423 */
424 in6_addr_t conn_v6lastsrc;
425 #define conn_v4lastsrc V4_PART_OF_V6(conn_v6lastsrc)
426
427 /* Templates for transmitting packets */
428 ip_pkt_t conn_xmit_ipp; /* Options if no ancil data */
429
430 /*
431 * Header template - conn_ht_ulp is a pointer into conn_ht_iphc.
432 * Note that ixa_ip_hdr_length indicates the offset of ht_ulp in
433 * ht_iphc
434 *
435 * The header template is maintained for connected endpoints (and
436 * updated when sticky options are changed) and also for the lastdst.
437 * There is no conflict between those usages since SOCK_DGRAM and
438 * SOCK_RAW can not be used to specify a destination address (with
439 * sendto/sendmsg) if the socket has been connected.
440 */
441 uint8_t *conn_ht_iphc; /* Start of IP header */
442 uint_t conn_ht_iphc_allocated; /* Allocated buffer size */
443 uint_t conn_ht_iphc_len; /* IP+ULP size */
444 uint8_t *conn_ht_ulp; /* Upper-layer header */
445 uint_t conn_ht_ulp_len; /* ULP header len */
446
447 /* Checksum to compensate for source routed packets. Host byte order */
448 uint32_t conn_sum;
449
450 uint32_t conn_ioctlref; /* ioctl ref count */
451 #ifdef CONN_DEBUG
452 #define CONN_TRACE_MAX 10
453 int conn_trace_last; /* ndx of last used tracebuf */
454 conn_trace_t conn_trace_buf[CONN_TRACE_MAX];
455 #endif
456 };
457
458 /*
459 * connf_t - connection fanout data.
460 *
461 * The hash tables and their linkage (conn_t.{hashnextp, hashprevp} are
462 * protected by the per-bucket lock. Each conn_t inserted in the list
463 * points back at the connf_t that heads the bucket.
464 */
465 struct connf_s {
466 struct conn_s *connf_head;
467 kmutex_t connf_lock;
468 };
469
470 #define CONN_INC_REF(connp) { \
471 mutex_enter(&(connp)->conn_lock); \
472 DTRACE_PROBE1(conn__inc__ref, conn_t *, connp); \
473 ASSERT(conn_trace_ref(connp)); \
474 (connp)->conn_ref++; \
475 ASSERT((connp)->conn_ref != 0); \
476 mutex_exit(&(connp)->conn_lock); \
477 }
478
479 #define CONN_INC_REF_LOCKED(connp) { \
480 DTRACE_PROBE1(conn__inc__ref, conn_t *, connp); \
481 ASSERT(MUTEX_HELD(&(connp)->conn_lock)); \
482 ASSERT(conn_trace_ref(connp)); \
483 (connp)->conn_ref++; \
484 ASSERT((connp)->conn_ref != 0); \
485 }
486
487 #define CONN_DEC_REF(connp) { \
488 mutex_enter(&(connp)->conn_lock); \
489 DTRACE_PROBE1(conn__dec__ref, conn_t *, connp); \
490 /* \
491 * The squeue framework always does a CONN_DEC_REF after return \
492 * from TCP. Hence the refcnt must be at least 2 if conn_on_sqp \
493 * is B_TRUE and conn_ref is being decremented. This is to \
494 * account for the mblk being currently processed. \
495 */ \
496 if ((connp)->conn_ref == 0 || \
497 ((connp)->conn_ref == 1 && (connp)->conn_on_sqp)) \
498 cmn_err(CE_PANIC, "CONN_DEC_REF: connp(%p) has ref " \
499 "= %d\n", (void *)(connp), (connp)->conn_ref); \
500 ASSERT(conn_untrace_ref(connp)); \
501 (connp)->conn_ref--; \
502 if ((connp)->conn_ref == 0) { \
503 /* Refcnt can't increase again, safe to drop lock */ \
504 mutex_exit(&(connp)->conn_lock); \
505 ipcl_conn_destroy(connp); \
506 } else { \
507 cv_broadcast(&(connp)->conn_cv); \
508 mutex_exit(&(connp)->conn_lock); \
509 } \
510 }
511
512 /*
513 * For use with subsystems within ip which use ALL_ZONES as a wildcard
514 */
515 #define IPCL_ZONEID(connp) \
516 ((connp)->conn_allzones ? ALL_ZONES : (connp)->conn_zoneid)
517
518 /*
519 * For matching between a conn_t and a zoneid.
520 */
521 #define IPCL_ZONE_MATCH(connp, zoneid) \
522 (((connp)->conn_allzones) || \
523 ((zoneid) == ALL_ZONES) || \
524 (connp)->conn_zoneid == (zoneid))
525
526 /*
527 * On a labeled system, we must treat bindings to ports
528 * on shared IP addresses by sockets with MAC exemption
529 * privilege as being in all zones, as there's
530 * otherwise no way to identify the right receiver.
531 */
532
533 #define IPCL_CONNS_MAC(conn1, conn2) \
534 (((conn1)->conn_mac_mode != CONN_MAC_DEFAULT) || \
535 ((conn2)->conn_mac_mode != CONN_MAC_DEFAULT))
536
537 #define IPCL_BIND_ZONE_MATCH(conn1, conn2) \
538 (IPCL_CONNS_MAC(conn1, conn2) || \
539 IPCL_ZONE_MATCH(conn1, conn2->conn_zoneid) || \
540 IPCL_ZONE_MATCH(conn2, conn1->conn_zoneid))
541
542
543 #define _IPCL_V4_MATCH(v6addr, v4addr) \
544 (V4_PART_OF_V6((v6addr)) == (v4addr) && IN6_IS_ADDR_V4MAPPED(&(v6addr)))
545
546 #define _IPCL_V4_MATCH_ANY(addr) \
547 (IN6_IS_ADDR_V4MAPPED_ANY(&(addr)) || IN6_IS_ADDR_UNSPECIFIED(&(addr)))
548
549
550 /*
551 * IPCL_PROTO_MATCH() and IPCL_PROTO_MATCH_V6() only matches conns with
552 * the specified ira_zoneid or conn_allzones by calling conn_wantpacket.
553 */
554 #define IPCL_PROTO_MATCH(connp, ira, ipha) \
555 ((((connp)->conn_laddr_v4 == INADDR_ANY) || \
556 (((connp)->conn_laddr_v4 == ((ipha)->ipha_dst)) && \
557 (((connp)->conn_faddr_v4 == INADDR_ANY) || \
558 ((connp)->conn_faddr_v4 == ((ipha)->ipha_src))))) && \
559 conn_wantpacket((connp), (ira), (ipha)))
560
561 #define IPCL_PROTO_MATCH_V6(connp, ira, ip6h) \
562 ((IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_laddr_v6) || \
563 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &((ip6h)->ip6_dst)) && \
564 (IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_faddr_v6) || \
565 IN6_ARE_ADDR_EQUAL(&(connp)->conn_faddr_v6, &((ip6h)->ip6_src))))) && \
566 (conn_wantpacket_v6((connp), (ira), (ip6h))))
567
568 #define IPCL_CONN_HASH(src, ports, ipst) \
569 ((unsigned)(ntohl((src)) ^ ((ports) >> 24) ^ ((ports) >> 16) ^ \
570 ((ports) >> 8) ^ (ports)) % (ipst)->ips_ipcl_conn_fanout_size)
571
572 #define IPCL_CONN_HASH_V6(src, ports, ipst) \
573 IPCL_CONN_HASH(V4_PART_OF_V6((src)), (ports), (ipst))
574
575 #define IPCL_CONN_MATCH(connp, proto, src, dst, ports) \
576 ((connp)->conn_proto == (proto) && \
577 (connp)->conn_ports == (ports) && \
578 _IPCL_V4_MATCH((connp)->conn_faddr_v6, (src)) && \
579 _IPCL_V4_MATCH((connp)->conn_laddr_v6, (dst)) && \
580 !(connp)->conn_ipv6_v6only)
581
582 #define IPCL_CONN_MATCH_V6(connp, proto, src, dst, ports) \
583 ((connp)->conn_proto == (proto) && \
584 (connp)->conn_ports == (ports) && \
585 IN6_ARE_ADDR_EQUAL(&(connp)->conn_faddr_v6, &(src)) && \
586 IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &(dst)))
587
588 #define IPCL_PORT_HASH(port, size) \
589 ((((port) >> 8) ^ (port)) & ((size) - 1))
590
591 #define IPCL_BIND_HASH(lport, ipst) \
592 ((unsigned)(((lport) >> 8) ^ (lport)) % \
593 (ipst)->ips_ipcl_bind_fanout_size)
594
595 #define IPCL_BIND_MATCH(connp, proto, laddr, lport) \
596 ((connp)->conn_proto == (proto) && \
597 (connp)->conn_lport == (lport) && \
598 (_IPCL_V4_MATCH_ANY((connp)->conn_laddr_v6) || \
599 _IPCL_V4_MATCH((connp)->conn_laddr_v6, (laddr))) && \
600 !(connp)->conn_ipv6_v6only)
601
602 #define IPCL_BIND_MATCH_V6(connp, proto, laddr, lport) \
603 ((connp)->conn_proto == (proto) && \
604 (connp)->conn_lport == (lport) && \
605 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &(laddr)) || \
606 IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_laddr_v6)))
607
608 /*
609 * We compare conn_laddr since it captures both connected and a bind to
610 * a multicast or broadcast address.
611 * The caller needs to match the zoneid and also call conn_wantpacket
612 * for multicast, broadcast, or when conn_incoming_ifindex is set.
613 */
614 #define IPCL_UDP_MATCH(connp, lport, laddr, fport, faddr) \
615 (((connp)->conn_lport == (lport)) && \
616 ((_IPCL_V4_MATCH_ANY((connp)->conn_laddr_v6) || \
617 (_IPCL_V4_MATCH((connp)->conn_laddr_v6, (laddr)) && \
618 (_IPCL_V4_MATCH_ANY((connp)->conn_faddr_v6) || \
619 (_IPCL_V4_MATCH((connp)->conn_faddr_v6, (faddr)) && \
620 (connp)->conn_fport == (fport)))))) && \
621 !(connp)->conn_ipv6_v6only)
622
623 /*
624 * We compare conn_laddr since it captures both connected and a bind to
625 * a multicast or broadcast address.
626 * The caller needs to match the zoneid and also call conn_wantpacket_v6
627 * for multicast or when conn_incoming_ifindex is set.
628 */
629 #define IPCL_UDP_MATCH_V6(connp, lport, laddr, fport, faddr) \
630 (((connp)->conn_lport == (lport)) && \
631 (IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_laddr_v6) || \
632 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &(laddr)) && \
633 (IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_faddr_v6) || \
634 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_faddr_v6, &(faddr)) && \
635 (connp)->conn_fport == (fport))))))
636
637 #define IPCL_IPTUN_HASH(laddr, faddr) \
638 ((ntohl(laddr) ^ ((ntohl(faddr) << 24) | (ntohl(faddr) >> 8))) % \
639 ipcl_iptun_fanout_size)
640
641 #define IPCL_IPTUN_HASH_V6(laddr, faddr) \
642 IPCL_IPTUN_HASH((laddr)->s6_addr32[0] ^ (laddr)->s6_addr32[1] ^ \
643 (faddr)->s6_addr32[2] ^ (faddr)->s6_addr32[3], \
644 (faddr)->s6_addr32[0] ^ (faddr)->s6_addr32[1] ^ \
645 (laddr)->s6_addr32[2] ^ (laddr)->s6_addr32[3])
646
647 #define IPCL_IPTUN_MATCH(connp, laddr, faddr) \
648 (_IPCL_V4_MATCH((connp)->conn_laddr_v6, (laddr)) && \
649 _IPCL_V4_MATCH((connp)->conn_faddr_v6, (faddr)))
650
651 #define IPCL_IPTUN_MATCH_V6(connp, laddr, faddr) \
652 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, (laddr)) && \
653 IN6_ARE_ADDR_EQUAL(&(connp)->conn_faddr_v6, (faddr)))
654
655 #define IPCL_UDP_HASH(lport, ipst) \
656 IPCL_PORT_HASH(lport, (ipst)->ips_ipcl_udp_fanout_size)
657
658 #define CONN_G_HASH_SIZE 1024
659
660 /* Raw socket hash function. */
661 #define IPCL_RAW_HASH(lport, ipst) \
662 IPCL_PORT_HASH(lport, (ipst)->ips_ipcl_raw_fanout_size)
663
664 /*
665 * This is similar to IPCL_BIND_MATCH except that the local port check
666 * is changed to a wildcard port check.
667 * We compare conn_laddr since it captures both connected and a bind to
668 * a multicast or broadcast address.
669 */
670 #define IPCL_RAW_MATCH(connp, proto, laddr) \
671 ((connp)->conn_proto == (proto) && \
672 (connp)->conn_lport == 0 && \
673 (_IPCL_V4_MATCH_ANY((connp)->conn_laddr_v6) || \
674 _IPCL_V4_MATCH((connp)->conn_laddr_v6, (laddr))))
675
676 #define IPCL_RAW_MATCH_V6(connp, proto, laddr) \
677 ((connp)->conn_proto == (proto) && \
678 (connp)->conn_lport == 0 && \
679 (IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_laddr_v6) || \
680 IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &(laddr))))
681
682 /* Function prototypes */
683 extern void ipcl_g_init(void);
684 extern void ipcl_init(ip_stack_t *);
685 extern void ipcl_g_destroy(void);
686 extern void ipcl_destroy(ip_stack_t *);
687 extern conn_t *ipcl_conn_create(uint32_t, int, netstack_t *);
688 extern void ipcl_conn_destroy(conn_t *);
689
690 void ipcl_hash_insert_wildcard(connf_t *, conn_t *);
691 void ipcl_hash_remove(conn_t *);
692 void ipcl_hash_remove_locked(conn_t *connp, connf_t *connfp);
693
694 extern int ipcl_bind_insert(conn_t *);
695 extern int ipcl_bind_insert_v4(conn_t *);
696 extern int ipcl_bind_insert_v6(conn_t *);
697 extern int ipcl_conn_insert(conn_t *);
698 extern int ipcl_conn_insert_v4(conn_t *);
699 extern int ipcl_conn_insert_v6(conn_t *);
700 extern conn_t *ipcl_get_next_conn(connf_t *, conn_t *, uint32_t);
701
702 conn_t *ipcl_classify_v4(mblk_t *, uint8_t, uint_t, ip_recv_attr_t *,
703 ip_stack_t *);
704 conn_t *ipcl_classify_v6(mblk_t *, uint8_t, uint_t, ip_recv_attr_t *,
705 ip_stack_t *);
706 conn_t *ipcl_classify(mblk_t *, ip_recv_attr_t *, ip_stack_t *);
707 conn_t *ipcl_classify_raw(mblk_t *, uint8_t, uint32_t, ipha_t *,
708 ip6_t *, ip_recv_attr_t *, ip_stack_t *);
709 conn_t *ipcl_iptun_classify_v4(ipaddr_t *, ipaddr_t *, ip_stack_t *);
710 conn_t *ipcl_iptun_classify_v6(in6_addr_t *, in6_addr_t *, ip_stack_t *);
711 void ipcl_globalhash_insert(conn_t *);
712 void ipcl_globalhash_remove(conn_t *);
713 void ipcl_walk(pfv_t, void *, ip_stack_t *);
714 conn_t *ipcl_tcp_lookup_reversed_ipv4(ipha_t *, tcpha_t *, int, ip_stack_t *);
715 conn_t *ipcl_tcp_lookup_reversed_ipv6(ip6_t *, tcpha_t *, int, uint_t,
716 ip_stack_t *);
717 conn_t *ipcl_lookup_listener_v4(uint16_t, ipaddr_t, zoneid_t, ip_stack_t *);
718 conn_t *ipcl_lookup_listener_v6(uint16_t, in6_addr_t *, uint_t, zoneid_t,
719 ip_stack_t *);
720 int conn_trace_ref(conn_t *);
721 int conn_untrace_ref(conn_t *);
722 void ipcl_conn_cleanup(conn_t *);
723 extern uint_t conn_recvancillary_size(conn_t *, crb_t, ip_recv_attr_t *,
724 mblk_t *, ip_pkt_t *);
725 extern void conn_recvancillary_add(conn_t *, crb_t, ip_recv_attr_t *,
726 ip_pkt_t *, uchar_t *, uint_t);
727 conn_t *ipcl_conn_tcp_lookup_reversed_ipv4(conn_t *, ipha_t *, tcpha_t *,
728 ip_stack_t *);
729 conn_t *ipcl_conn_tcp_lookup_reversed_ipv6(conn_t *, ip6_t *, tcpha_t *,
730 ip_stack_t *);
731
732 extern int ip_create_helper_stream(conn_t *, ldi_ident_t);
733 extern void ip_free_helper_stream(conn_t *);
734 extern int ip_helper_stream_setup(queue_t *, dev_t *, int, int,
735 cred_t *, boolean_t);
736 extern mib2_socketInfoEntry_t *conn_get_socket_info(conn_t *,
737 mib2_socketInfoEntry_t *);
738
739 #ifdef __cplusplus
740 }
741 #endif
742
743 #endif /* _INET_IPCLASSIFIER_H */