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 #ifndef _INET_IPCLASSIFIER_H
27 #define _INET_IPCLASSIFIER_H
28
29 #ifdef __cplusplus
30 extern "C" {
31 #endif
32
33 #include <inet/common.h>
34 #include <inet/ip.h>
35 #include <inet/mi.h>
36 #include <inet/tcp.h>
37 #include <inet/ip6.h>
38 #include <netinet/in.h> /* for IPPROTO_* constants */
39 #include <sys/sdt.h>
40 #include <sys/socket_proto.h>
41 #include <sys/sunddi.h>
42 #include <sys/sunldi.h>
43
44 typedef void (*edesc_rpf)(void *, mblk_t *, void *, ip_recv_attr_t *);
45 struct icmph_s;
46 struct icmp6_hdr;
47 typedef boolean_t (*edesc_vpf)(conn_t *, void *, struct icmph_s *,
48 struct icmp6_hdr *, ip_recv_attr_t *);
49
50 /*
51 * ==============================
52 * = The CONNECTION =
53 * ==============================
54 */
55
56 /*
57 * The connection structure contains the common information/flags/ref needed.
58 * Implementation will keep the connection struct, the layers (with their
59 * respective data for event i.e. tcp_t if event was tcp_input_data) all in one
60 * contiguous memory location.
61 */
62
63 /* Conn Flags */
64 /* Unused 0x00020000 */
65 /* Unused 0x00040000 */
66 #define IPCL_FULLY_BOUND 0x00080000 /* Bound to correct squeue */
67 /* Unused 0x00100000 */
68 /* Unused 0x00200000 */
69 /* Unused 0x00400000 */
70 #define IPCL_CL_LISTENER 0x00800000 /* Cluster listener */
71 /* Unused 0x01000000 */
72 /* Unused 0x02000000 */
73 /* Unused 0x04000000 */
74 /* Unused 0x08000000 */
75 /* Unused 0x10000000 */
76 /* Unused 0x20000000 */
77 #define IPCL_CONNECTED 0x40000000 /* Conn in connected table */
78 #define IPCL_BOUND 0x80000000 /* Conn in bind table */
79
80 /* Flags identifying the type of conn */
81 #define IPCL_TCPCONN 0x00000001 /* From tcp_conn_cache */
82 #define IPCL_SCTPCONN 0x00000002 /* From sctp_conn_cache */
83 #define IPCL_IPCCONN 0x00000004 /* From ip_conn_cache */
84 #define IPCL_UDPCONN 0x00000008 /* From udp_conn_cache */
85 #define IPCL_RAWIPCONN 0x00000010 /* From rawip_conn_cache */
86 #define IPCL_RTSCONN 0x00000020 /* From rts_conn_cache */
87 #define IPCL_DCCPCONN 0x00000040 /* From dccp_conn_cache */
88 #define IPCL_IPTUN 0x00000080 /* iptun module above us */
89
90 #define IPCL_NONSTR 0x00001000 /* A non-STREAMS socket */
91 /* Unused 0x10000000 */
92
93 #define IPCL_REMOVED 0x00000100
94 #define IPCL_REUSED 0x00000200
95
96 #define IPCL_IS_CONNECTED(connp) \
97 ((connp)->conn_flags & IPCL_CONNECTED)
98
99 #define IPCL_IS_BOUND(connp) \
100 ((connp)->conn_flags & IPCL_BOUND)
101
102 /*
103 * Can't use conn_proto since we need to tell difference
104 * between a real TCP socket and a SOCK_RAW, IPPROTO_TCP.
105 */
106 #define IPCL_IS_TCP(connp) \
107 ((connp)->conn_flags & IPCL_TCPCONN)
108
109 #define IPCL_IS_SCTP(connp) \
110 ((connp)->conn_flags & IPCL_SCTPCONN)
111
112 #define IPCL_IS_UDP(connp) \
113 ((connp)->conn_flags & IPCL_UDPCONN)
114
115 #define IPCL_IS_RAWIP(connp) \
116 ((connp)->conn_flags & IPCL_RAWIPCONN)
117
118 #define IPCL_IS_RTS(connp) \
119 ((connp)->conn_flags & IPCL_RTSCONN)
120
121 #define IPCL_IS_IPTUN(connp) \
122 ((connp)->conn_flags & IPCL_IPTUN)
123
124 #define IPCL_IS_DCCP(connp) \
125 ((connp)->conn_flags & IPCL_DCCPCONN)
126
127 #define IPCL_IS_NONSTR(connp) ((connp)->conn_flags & IPCL_NONSTR)
128
129 typedef struct connf_s connf_t;
130
131 typedef struct
132 {
133 int ctb_depth;
134 #define CONN_STACK_DEPTH 15
135 pc_t ctb_stack[CONN_STACK_DEPTH];
136 } conn_trace_t;
137
138 typedef struct ip_helper_minor_info_s {
139 dev_t ip_minfo_dev; /* Device */
140 vmem_t *ip_minfo_arena; /* Arena */
141 } ip_helper_minfo_t;
142
143 /*
144 * ip helper stream info
145 */
146 typedef struct ip_helper_stream_info_s {
147 ldi_handle_t iphs_handle;
148 queue_t *iphs_rq;
149 queue_t *iphs_wq;
150 ip_helper_minfo_t *iphs_minfo;
151 } ip_helper_stream_info_t;
152
153 /*
154 * Mandatory Access Control mode, in conn_t's conn_mac_mode field.
155 * CONN_MAC_DEFAULT: strict enforcement of MAC.
156 * CONN_MAC_AWARE: allows communications between unlabeled systems
157 * and privileged daemons
158 * CONN_MAC_IMPLICIT: allows communications without explicit labels
159 * on the wire with privileged daemons.
160 *
161 * CONN_MAC_IMPLICIT is intended specifically for labeled IPsec key management
162 * in networks which don't pass CIPSO-labeled packets.
163 */
164 #define CONN_MAC_DEFAULT 0
165 #define CONN_MAC_AWARE 1
166 #define CONN_MAC_IMPLICIT 2
167
168 /*
169 * conn receive ancillary definition.
170 *
171 * These are the set of socket options that make the receive side
172 * potentially pass up ancillary data items.
173 * We have a union with an integer so that we can quickly check whether
174 * any ancillary data items need to be added.
175 */
176 typedef struct crb_s {
177 union {
178 uint32_t crbu_all;
179 struct {
180 uint32_t
181 crbb_recvdstaddr : 1, /* IP_RECVDSTADDR option */
182 crbb_recvopts : 1, /* IP_RECVOPTS option */
183 crbb_recvif : 1, /* IP_RECVIF option */
184 crbb_recvslla : 1, /* IP_RECVSLLA option */
185
186 crbb_recvttl : 1, /* IP_RECVTTL option */
187 crbb_ip_recvpktinfo : 1, /* IP*_RECVPKTINFO option */
188 crbb_ipv6_recvhoplimit : 1, /* IPV6_RECVHOPLIMIT option */
189 crbb_ipv6_recvhopopts : 1, /* IPV6_RECVHOPOPTS option */
190
191 crbb_ipv6_recvdstopts : 1, /* IPV6_RECVDSTOPTS option */
192 crbb_ipv6_recvrthdr : 1, /* IPV6_RECVRTHDR option */
193 crbb_old_ipv6_recvdstopts : 1, /* old form of IPV6_DSTOPTS */
194 crbb_ipv6_recvrthdrdstopts : 1, /* IPV6_RECVRTHDRDSTOPTS */
195
196 crbb_ipv6_recvtclass : 1, /* IPV6_RECVTCLASS */
197 crbb_recvucred : 1, /* IP_RECVUCRED option */
198 crbb_timestamp : 1; /* SO_TIMESTAMP "socket" option */
199
200 } crbb;
201 } crbu;
202 } crb_t;
203
204 #define crb_all crbu.crbu_all
205 #define crb_recvdstaddr crbu.crbb.crbb_recvdstaddr
206 #define crb_recvopts crbu.crbb.crbb_recvopts
207 #define crb_recvif crbu.crbb.crbb_recvif
208 #define crb_recvslla crbu.crbb.crbb_recvslla
209 #define crb_recvttl crbu.crbb.crbb_recvttl
210 #define crb_ip_recvpktinfo crbu.crbb.crbb_ip_recvpktinfo
211 #define crb_ipv6_recvhoplimit crbu.crbb.crbb_ipv6_recvhoplimit
212 #define crb_ipv6_recvhopopts crbu.crbb.crbb_ipv6_recvhopopts
213 #define crb_ipv6_recvdstopts crbu.crbb.crbb_ipv6_recvdstopts
214 #define crb_ipv6_recvrthdr crbu.crbb.crbb_ipv6_recvrthdr
215 #define crb_old_ipv6_recvdstopts crbu.crbb.crbb_old_ipv6_recvdstopts
216 #define crb_ipv6_recvrthdrdstopts crbu.crbb.crbb_ipv6_recvrthdrdstopts
217 #define crb_ipv6_recvtclass crbu.crbb.crbb_ipv6_recvtclass
218 #define crb_recvucred crbu.crbb.crbb_recvucred
219 #define crb_timestamp crbu.crbb.crbb_timestamp
220
221 /*
222 * The initial fields in the conn_t are setup by the kmem_cache constructor,
223 * and are preserved when it is freed. Fields after that are bzero'ed when
224 * the conn_t is freed.
225 *
226 * Much of the conn_t is protected by conn_lock.
227 *
228 * conn_lock is also used by some ULPs (like UDP and RAWIP) to protect
229 * their state.
230 */
231 struct conn_s {
232 kmutex_t conn_lock;
233 uint32_t conn_ref; /* Reference counter */
234 uint32_t conn_flags; /* Conn Flags */
235
236 union {
237 tcp_t *cp_tcp; /* Pointer to the tcp struct */
238 struct udp_s *cp_udp; /* Pointer to the udp struct */
239 struct icmp_s *cp_icmp; /* Pointer to rawip struct */
240 struct rts_s *cp_rts; /* Pointer to rts struct */
241 struct iptun_s *cp_iptun; /* Pointer to iptun_t */
242 struct sctp_s *cp_sctp; /* For IPCL_SCTPCONN */
243 struct dccp_s *cp_dccp; /* Pointer to dccp struct */
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_dccp conn_proto_priv.cp_dccp
253 #define conn_priv conn_proto_priv.cp_priv
254
255 kcondvar_t conn_cv;
256 uint8_t conn_proto; /* protocol type */
257
258 edesc_rpf conn_recv; /* Pointer to recv routine */
259 edesc_rpf conn_recvicmp; /* For ICMP error */
260 edesc_vpf conn_verifyicmp; /* Verify ICMP error */
261
262 ip_xmit_attr_t *conn_ixa; /* Options if no ancil data */
263
264 /* Fields after this are bzero'ed when the conn_t is freed. */
265 #define conn_start_clr conn_recv_ancillary
266
267 /* Options for receive-side ancillary data */
268 crb_t conn_recv_ancillary;
269
270 squeue_t *conn_sqp; /* Squeue for processing */
271 uint_t conn_state_flags; /* IP state flags */
272
273 int conn_lingertime; /* linger time (in seconds) */
274
275 unsigned int
276 conn_on_sqp : 1, /* Conn is being processed */
277 conn_linger : 1, /* SO_LINGER state */
278 conn_useloopback : 1, /* SO_USELOOPBACK state */
279 conn_broadcast : 1, /* SO_BROADCAST state */
280
281 conn_reuseaddr : 1, /* SO_REUSEADDR state */
282 conn_keepalive : 1, /* SO_KEEPALIVE state */
283 conn_multi_router : 1, /* Wants all multicast pkts */
284 conn_unspec_src : 1, /* IP_UNSPEC_SRC */
285
286 conn_policy_cached : 1, /* Is policy cached/latched ? */
287 conn_in_enforce_policy : 1, /* Enforce Policy on inbound */
288 conn_out_enforce_policy : 1, /* Enforce Policy on outbound */
289 conn_debug : 1, /* SO_DEBUG */
290
291 conn_ipv6_v6only : 1, /* IPV6_V6ONLY */
292 conn_oobinline : 1, /* SO_OOBINLINE state */
293 conn_dgram_errind : 1, /* SO_DGRAM_ERRIND state */
294 conn_exclbind : 1, /* SO_EXCLBIND state */
295
296 conn_mdt_ok : 1, /* MDT is permitted */
297 conn_allzones : 1, /* SO_ALLZONES */
298 conn_ipv6_recvpathmtu : 1, /* IPV6_RECVPATHMTU */
299 conn_mcbc_bind : 1, /* Bound to multi/broadcast */
300
301 conn_pad_to_bit_31 : 12;
302
303 boolean_t conn_blocked; /* conn is flow-controlled */
304
305 squeue_t *conn_initial_sqp; /* Squeue at open time */
306 squeue_t *conn_final_sqp; /* Squeue after connect */
307 ill_t *conn_dhcpinit_ill; /* IP_DHCPINIT_IF */
308 ipsec_latch_t *conn_latch; /* latched IDS */
309 struct ipsec_policy_s *conn_latch_in_policy; /* latched policy (in) */
310 struct ipsec_action_s *conn_latch_in_action; /* latched action (in) */
311 uint_t conn_bound_if; /* IP*_BOUND_IF */
312 queue_t *conn_rq; /* Read queue */
313 queue_t *conn_wq; /* Write queue */
314 dev_t conn_dev; /* Minor number */
315 vmem_t *conn_minor_arena; /* Minor arena */
316 ip_helper_stream_info_t *conn_helper_info;
317
318 cred_t *conn_cred; /* Credentials */
319 pid_t conn_cpid; /* pid from open/connect */
320 uint64_t conn_open_time; /* time when this was opened */
321
322 connf_t *conn_g_fanout; /* Global Hash bucket head */
323 struct conn_s *conn_g_next; /* Global Hash chain next */
324 struct conn_s *conn_g_prev; /* Global Hash chain prev */
325 struct ipsec_policy_head_s *conn_policy; /* Configured policy */
326 in6_addr_t conn_bound_addr_v6; /* Address in bind() */
327 #define conn_bound_addr_v4 V4_PART_OF_V6(conn_bound_addr_v6)
328 connf_t *conn_fanout; /* Hash bucket we're part of */
329 struct conn_s *conn_next; /* Hash chain next */
330 struct conn_s *conn_prev; /* Hash chain prev */
331
332 struct {
333 in6_addr_t connua_laddr; /* Local address - match */
334 in6_addr_t connua_faddr; /* Remote address */
335 } connua_v6addr;
336 #define conn_laddr_v4 V4_PART_OF_V6(connua_v6addr.connua_laddr)
337 #define conn_faddr_v4 V4_PART_OF_V6(connua_v6addr.connua_faddr)
338 #define conn_laddr_v6 connua_v6addr.connua_laddr
339 #define conn_faddr_v6 connua_v6addr.connua_faddr
340 in6_addr_t conn_saddr_v6; /* Local address - source */
341 #define conn_saddr_v4 V4_PART_OF_V6(conn_saddr_v6)
342
343 union {
344 /* Used for classifier match performance */
345 uint32_t connu_ports2;
346 struct {
347 in_port_t connu_fport; /* Remote port */
348 in_port_t connu_lport; /* Local port */
349 } connu_ports;
350 } u_port;
351 #define conn_fport u_port.connu_ports.connu_fport
352 #define conn_lport u_port.connu_ports.connu_lport
353 #define conn_ports u_port.connu_ports2
354
355 uint_t conn_incoming_ifindex; /* IP{,V6}_BOUND_IF, scopeid */
356 ill_t *conn_oper_pending_ill; /* pending shared ioctl */
357
358 krwlock_t conn_ilg_lock; /* Protects conn_ilg_* */
359 ilg_t *conn_ilg; /* Group memberships */
360
361 kcondvar_t conn_refcv; /* For conn_oper_pending_ill */
362
363 struct conn_s *conn_drain_next; /* Next conn in drain list */
364 struct conn_s *conn_drain_prev; /* Prev conn in drain list */
365 idl_t *conn_idl; /* Ptr to the drain list head */
366 mblk_t *conn_ipsec_opt_mp; /* ipsec option mblk */
367 zoneid_t conn_zoneid; /* zone connection is in */
368 int conn_rtaware; /* RT_AWARE sockopt value */
369 kcondvar_t conn_sq_cv; /* For non-STREAMS socket IO */
370 sock_upcalls_t *conn_upcalls; /* Upcalls to sockfs */
371 sock_upper_handle_t conn_upper_handle; /* Upper handle: sonode * */
372
373 unsigned int
374 conn_mlp_type : 2, /* mlp_type_t; tsol/tndb.h */
375 conn_anon_mlp : 1, /* user wants anon MLP */
376 conn_anon_port : 1, /* user bound anonymously */
377
378 conn_mac_mode : 2, /* normal/loose/implicit MAC */
379 conn_anon_priv_bind : 1, /* *_ANON_PRIV_BIND state */
380 conn_zone_is_global : 1, /* GLOBAL_ZONEID */
381 conn_isvrrp : 1, /* VRRP control socket */
382 conn_spare : 23;
383
384 boolean_t conn_flow_cntrld;
385 netstack_t *conn_netstack; /* Corresponds to a netstack_hold */
386
387 /*
388 * IP format that packets received for this struct should use.
389 * Value can be IP4_VERSION or IPV6_VERSION.
390 * The sending version is encoded using IXAF_IS_IPV4.
391 */
392 ushort_t conn_ipversion;
393
394 /* Written to only once at the time of opening the endpoint */
395 sa_family_t conn_family; /* Family from socket() call */
396 uint_t conn_so_type; /* Type from socket() call */
397
398 uint_t conn_sndbuf; /* SO_SNDBUF state */
399 uint_t conn_rcvbuf; /* SO_RCVBUF state */
400 uint_t conn_wroff; /* Current write offset */
401
402 uint_t conn_sndlowat; /* Send buffer low water mark */
403 uint_t conn_rcvlowat; /* Recv buffer low water mark */
404
405 uint8_t conn_default_ttl; /* Default TTL/hoplimit */
406
407 uint32_t conn_flowinfo; /* Connected flow id and tclass */
408
409 /*
410 * The most recent address for sendto. Initially set to zero
411 * which is always different than then the destination address
412 * since the send interprets zero as the loopback address.
413 */
414 in6_addr_t conn_v6lastdst;
415 #define conn_v4lastdst V4_PART_OF_V6(conn_v6lastdst)
416 ushort_t conn_lastipversion;
417 in_port_t conn_lastdstport;
418 uint32_t conn_lastflowinfo; /* IPv6-only */
419 uint_t conn_lastscopeid; /* IPv6-only */
420 uint_t conn_lastsrcid; /* Only for AF_INET6 */
421 /*
422 * When we are not connected conn_saddr might be unspecified.
423 * We track the source that was used with conn_v6lastdst here.
424 */
425 in6_addr_t conn_v6lastsrc;
426 #define conn_v4lastsrc V4_PART_OF_V6(conn_v6lastsrc)
427
428 /* Templates for transmitting packets */
429 ip_pkt_t conn_xmit_ipp; /* Options if no ancil data */
430
431 /*
432 * Header template - conn_ht_ulp is a pointer into conn_ht_iphc.
433 * Note that ixa_ip_hdr_length indicates the offset of ht_ulp in
434 * ht_iphc
435 *
436 * The header template is maintained for connected endpoints (and
437 * updated when sticky options are changed) and also for the lastdst.
438 * There is no conflict between those usages since SOCK_DGRAM and
439 * SOCK_RAW can not be used to specify a destination address (with
440 * sendto/sendmsg) if the socket has been connected.
441 */
442 uint8_t *conn_ht_iphc; /* Start of IP header */
443 uint_t conn_ht_iphc_allocated; /* Allocated buffer size */
444 uint_t conn_ht_iphc_len; /* IP+ULP size */
445 uint8_t *conn_ht_ulp; /* Upper-layer header */
446 uint_t conn_ht_ulp_len; /* ULP header len */
447
448 /* Checksum to compensate for source routed packets. Host byte order */
449 uint32_t conn_sum;
450
451 uint32_t conn_ioctlref; /* ioctl ref count */
452 #ifdef CONN_DEBUG
453 #define CONN_TRACE_MAX 10
454 int conn_trace_last; /* ndx of last used tracebuf */
455 conn_trace_t conn_trace_buf[CONN_TRACE_MAX];
456 #endif
457 };
458
459 /*
460 * connf_t - connection fanout data.
461 *
462 * The hash tables and their linkage (conn_t.{hashnextp, hashprevp} are
463 * protected by the per-bucket lock. Each conn_t inserted in the list
464 * points back at the connf_t that heads the bucket.
465 */
466 struct connf_s {
467 struct conn_s *connf_head;
468 kmutex_t connf_lock;
469 };
470
471 #define CONN_INC_REF(connp) { \
472 mutex_enter(&(connp)->conn_lock); \
473 DTRACE_PROBE1(conn__inc__ref, conn_t *, connp); \
474 ASSERT(conn_trace_ref(connp)); \
475 (connp)->conn_ref++; \
476 ASSERT((connp)->conn_ref != 0); \
477 mutex_exit(&(connp)->conn_lock); \
478 }
479
480 #define CONN_INC_REF_LOCKED(connp) { \
481 DTRACE_PROBE1(conn__inc__ref, conn_t *, connp); \
482 ASSERT(MUTEX_HELD(&(connp)->conn_lock)); \
483 ASSERT(conn_trace_ref(connp)); \
484 (connp)->conn_ref++; \
485 ASSERT((connp)->conn_ref != 0); \
486 }
487
488 #define CONN_DEC_REF(connp) { \
489 mutex_enter(&(connp)->conn_lock); \
490 DTRACE_PROBE1(conn__dec__ref, conn_t *, connp); \
491 /* \
492 * The squeue framework always does a CONN_DEC_REF after return \
493 * from TCP. Hence the refcnt must be at least 2 if conn_on_sqp \
494 * is B_TRUE and conn_ref is being decremented. This is to \
495 * account for the mblk being currently processed. \
496 */ \
497 if ((connp)->conn_ref == 0 || \
498 ((connp)->conn_ref == 1 && (connp)->conn_on_sqp)) \
499 cmn_err(CE_PANIC, "CONN_DEC_REF: connp(%p) has ref " \
500 "= %d\n", (void *)(connp), (connp)->conn_ref); \
501 ASSERT(conn_untrace_ref(connp)); \
502 (connp)->conn_ref--; \
503 if ((connp)->conn_ref == 0) { \
504 /* Refcnt can't increase again, safe to drop lock */ \
505 mutex_exit(&(connp)->conn_lock); \
506 ipcl_conn_destroy(connp); \
507 } else { \
508 cv_broadcast(&(connp)->conn_cv); \
509 mutex_exit(&(connp)->conn_lock); \
510 } \
511 }
512
513 /*
514 * For use with subsystems within ip which use ALL_ZONES as a wildcard
515 */
516 #define IPCL_ZONEID(connp) \
517 ((connp)->conn_allzones ? ALL_ZONES : (connp)->conn_zoneid)
518
519 /*
520 * For matching between a conn_t and a zoneid.
521 */
522 #define IPCL_ZONE_MATCH(connp, zoneid) \
523 (((connp)->conn_allzones) || \
524 ((zoneid) == ALL_ZONES) || \
525 (connp)->conn_zoneid == (zoneid))
526
527 /*
528 * On a labeled system, we must treat bindings to ports
529 * on shared IP addresses by sockets with MAC exemption
530 * privilege as being in all zones, as there's
531 * otherwise no way to identify the right receiver.
532 */
533
534 #define IPCL_CONNS_MAC(conn1, conn2) \
535 (((conn1)->conn_mac_mode != CONN_MAC_DEFAULT) || \
536 ((conn2)->conn_mac_mode != CONN_MAC_DEFAULT))
537
538 #define IPCL_BIND_ZONE_MATCH(conn1, conn2) \
539 (IPCL_CONNS_MAC(conn1, conn2) || \
540 IPCL_ZONE_MATCH(conn1, conn2->conn_zoneid) || \
541 IPCL_ZONE_MATCH(conn2, conn1->conn_zoneid))
542
543
544 #define _IPCL_V4_MATCH(v6addr, v4addr) \
545 (V4_PART_OF_V6((v6addr)) == (v4addr) && IN6_IS_ADDR_V4MAPPED(&(v6addr)))
546
547 #define _IPCL_V4_MATCH_ANY(addr) \
548 (IN6_IS_ADDR_V4MAPPED_ANY(&(addr)) || IN6_IS_ADDR_UNSPECIFIED(&(addr)))
549
550
551 /*
552 * IPCL_PROTO_MATCH() and IPCL_PROTO_MATCH_V6() only matches conns with
553 * the specified ira_zoneid or conn_allzones by calling conn_wantpacket.
554 */
555 #define IPCL_PROTO_MATCH(connp, ira, ipha) \
556 ((((connp)->conn_laddr_v4 == INADDR_ANY) || \
557 (((connp)->conn_laddr_v4 == ((ipha)->ipha_dst)) && \
558 (((connp)->conn_faddr_v4 == INADDR_ANY) || \
559 ((connp)->conn_faddr_v4 == ((ipha)->ipha_src))))) && \
560 conn_wantpacket((connp), (ira), (ipha)))
561
562 #define IPCL_PROTO_MATCH_V6(connp, ira, ip6h) \
563 ((IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_laddr_v6) || \
564 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &((ip6h)->ip6_dst)) && \
565 (IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_faddr_v6) || \
566 IN6_ARE_ADDR_EQUAL(&(connp)->conn_faddr_v6, &((ip6h)->ip6_src))))) && \
567 (conn_wantpacket_v6((connp), (ira), (ip6h))))
568
569 #define IPCL_CONN_HASH(src, ports, ipst) \
570 ((unsigned)(ntohl((src)) ^ ((ports) >> 24) ^ ((ports) >> 16) ^ \
571 ((ports) >> 8) ^ (ports)) % (ipst)->ips_ipcl_conn_fanout_size)
572
573 #define IPCL_CONN_HASH_V6(src, ports, ipst) \
574 IPCL_CONN_HASH(V4_PART_OF_V6((src)), (ports), (ipst))
575
576 #define IPCL_CONN_MATCH(connp, proto, src, dst, ports) \
577 ((connp)->conn_proto == (proto) && \
578 (connp)->conn_ports == (ports) && \
579 _IPCL_V4_MATCH((connp)->conn_faddr_v6, (src)) && \
580 _IPCL_V4_MATCH((connp)->conn_laddr_v6, (dst)) && \
581 !(connp)->conn_ipv6_v6only)
582
583 #define IPCL_CONN_MATCH_V6(connp, proto, src, dst, ports) \
584 ((connp)->conn_proto == (proto) && \
585 (connp)->conn_ports == (ports) && \
586 IN6_ARE_ADDR_EQUAL(&(connp)->conn_faddr_v6, &(src)) && \
587 IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &(dst)))
588
589 #define IPCL_PORT_HASH(port, size) \
590 ((((port) >> 8) ^ (port)) & ((size) - 1))
591
592 #define IPCL_BIND_HASH(lport, ipst) \
593 ((unsigned)(((lport) >> 8) ^ (lport)) % \
594 (ipst)->ips_ipcl_bind_fanout_size)
595
596 #define IPCL_BIND_MATCH(connp, proto, laddr, lport) \
597 ((connp)->conn_proto == (proto) && \
598 (connp)->conn_lport == (lport) && \
599 (_IPCL_V4_MATCH_ANY((connp)->conn_laddr_v6) || \
600 _IPCL_V4_MATCH((connp)->conn_laddr_v6, (laddr))) && \
601 !(connp)->conn_ipv6_v6only)
602
603 #define IPCL_BIND_MATCH_V6(connp, proto, laddr, lport) \
604 ((connp)->conn_proto == (proto) && \
605 (connp)->conn_lport == (lport) && \
606 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &(laddr)) || \
607 IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_laddr_v6)))
608
609 /*
610 * We compare conn_laddr since it captures both connected and a bind to
611 * a multicast or broadcast address.
612 * The caller needs to match the zoneid and also call conn_wantpacket
613 * for multicast, broadcast, or when conn_incoming_ifindex is set.
614 */
615 #define IPCL_UDP_MATCH(connp, lport, laddr, fport, faddr) \
616 (((connp)->conn_lport == (lport)) && \
617 ((_IPCL_V4_MATCH_ANY((connp)->conn_laddr_v6) || \
618 (_IPCL_V4_MATCH((connp)->conn_laddr_v6, (laddr)) && \
619 (_IPCL_V4_MATCH_ANY((connp)->conn_faddr_v6) || \
620 (_IPCL_V4_MATCH((connp)->conn_faddr_v6, (faddr)) && \
621 (connp)->conn_fport == (fport)))))) && \
622 !(connp)->conn_ipv6_v6only)
623
624 /*
625 * We compare conn_laddr since it captures both connected and a bind to
626 * a multicast or broadcast address.
627 * The caller needs to match the zoneid and also call conn_wantpacket_v6
628 * for multicast or when conn_incoming_ifindex is set.
629 */
630 #define IPCL_UDP_MATCH_V6(connp, lport, laddr, fport, faddr) \
631 (((connp)->conn_lport == (lport)) && \
632 (IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_laddr_v6) || \
633 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &(laddr)) && \
634 (IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_faddr_v6) || \
635 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_faddr_v6, &(faddr)) && \
636 (connp)->conn_fport == (fport))))))
637
638 #define IPCL_IPTUN_HASH(laddr, faddr) \
639 ((ntohl(laddr) ^ ((ntohl(faddr) << 24) | (ntohl(faddr) >> 8))) % \
640 ipcl_iptun_fanout_size)
641
642 #define IPCL_IPTUN_HASH_V6(laddr, faddr) \
643 IPCL_IPTUN_HASH((laddr)->s6_addr32[0] ^ (laddr)->s6_addr32[1] ^ \
644 (faddr)->s6_addr32[2] ^ (faddr)->s6_addr32[3], \
645 (faddr)->s6_addr32[0] ^ (faddr)->s6_addr32[1] ^ \
646 (laddr)->s6_addr32[2] ^ (laddr)->s6_addr32[3])
647
648 #define IPCL_IPTUN_MATCH(connp, laddr, faddr) \
649 (_IPCL_V4_MATCH((connp)->conn_laddr_v6, (laddr)) && \
650 _IPCL_V4_MATCH((connp)->conn_faddr_v6, (faddr)))
651
652 #define IPCL_IPTUN_MATCH_V6(connp, laddr, faddr) \
653 (IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, (laddr)) && \
654 IN6_ARE_ADDR_EQUAL(&(connp)->conn_faddr_v6, (faddr)))
655
656 #define IPCL_UDP_HASH(lport, ipst) \
657 IPCL_PORT_HASH(lport, (ipst)->ips_ipcl_udp_fanout_size)
658
659 #define IPCL_DCCP_CONN_HASH(src, ports, ipst) \
660 ((unsigned)(ntohl((src)) ^ ((ports) >> 24) ^ ((ports) >> 16) ^ \
661 ((ports) >> 8) ^ (ports)) % (ipst)->ips_ipcl_dccp_conn_fanout_size)
662
663 #define IPCL_DCCP_CONN_HASH_V6(src, ports, ipst) \
664 IPCL_DCCP_CONN_HASH(V4_PART_OF_V6((src)), (ports), (ipst))
665
666 #define IPCL_DCCP_BIND_HASH(lport, ipst) \
667 ((unsigned)(((lport) >> 8) ^ (lport)) % \
668 (ipst)->ips_ipcl_dccp_bind_fanout_size)
669
670
671 #define CONN_G_HASH_SIZE 1024
672
673 /* Raw socket hash function. */
674 #define IPCL_RAW_HASH(lport, ipst) \
675 IPCL_PORT_HASH(lport, (ipst)->ips_ipcl_raw_fanout_size)
676
677 /*
678 * This is similar to IPCL_BIND_MATCH except that the local port check
679 * is changed to a wildcard port check.
680 * We compare conn_laddr since it captures both connected and a bind to
681 * a multicast or broadcast address.
682 */
683 #define IPCL_RAW_MATCH(connp, proto, laddr) \
684 ((connp)->conn_proto == (proto) && \
685 (connp)->conn_lport == 0 && \
686 (_IPCL_V4_MATCH_ANY((connp)->conn_laddr_v6) || \
687 _IPCL_V4_MATCH((connp)->conn_laddr_v6, (laddr))))
688
689 #define IPCL_RAW_MATCH_V6(connp, proto, laddr) \
690 ((connp)->conn_proto == (proto) && \
691 (connp)->conn_lport == 0 && \
692 (IN6_IS_ADDR_UNSPECIFIED(&(connp)->conn_laddr_v6) || \
693 IN6_ARE_ADDR_EQUAL(&(connp)->conn_laddr_v6, &(laddr))))
694
695 /* Function prototypes */
696 extern void ipcl_g_init(void);
697 extern void ipcl_init(ip_stack_t *);
698 extern void ipcl_g_destroy(void);
699 extern void ipcl_destroy(ip_stack_t *);
700 extern conn_t *ipcl_conn_create(uint32_t, int, netstack_t *);
701 extern void ipcl_conn_destroy(conn_t *);
702
703 void ipcl_hash_insert_wildcard(connf_t *, conn_t *);
704 void ipcl_hash_remove(conn_t *);
705 void ipcl_hash_remove_locked(conn_t *connp, connf_t *connfp);
706
707 extern int ipcl_bind_insert(conn_t *);
708 extern int ipcl_bind_insert_v4(conn_t *);
709 extern int ipcl_bind_insert_v6(conn_t *);
710 extern int ipcl_conn_insert(conn_t *);
711 extern int ipcl_conn_insert_v4(conn_t *);
712 extern int ipcl_conn_insert_v6(conn_t *);
713 extern conn_t *ipcl_get_next_conn(connf_t *, conn_t *, uint32_t);
714
715 conn_t *ipcl_classify_v4(mblk_t *, uint8_t, uint_t, ip_recv_attr_t *,
716 ip_stack_t *);
717 conn_t *ipcl_classify_v6(mblk_t *, uint8_t, uint_t, ip_recv_attr_t *,
718 ip_stack_t *);
719 conn_t *ipcl_classify(mblk_t *, ip_recv_attr_t *, ip_stack_t *);
720 conn_t *ipcl_classify_raw(mblk_t *, uint8_t, uint32_t, ipha_t *,
721 ip6_t *, ip_recv_attr_t *, ip_stack_t *);
722 conn_t *ipcl_iptun_classify_v4(ipaddr_t *, ipaddr_t *, ip_stack_t *);
723 conn_t *ipcl_iptun_classify_v6(in6_addr_t *, in6_addr_t *, ip_stack_t *);
724 void ipcl_globalhash_insert(conn_t *);
725 void ipcl_globalhash_remove(conn_t *);
726 void ipcl_walk(pfv_t, void *, ip_stack_t *);
727 conn_t *ipcl_tcp_lookup_reversed_ipv4(ipha_t *, tcpha_t *, int, ip_stack_t *);
728 conn_t *ipcl_tcp_lookup_reversed_ipv6(ip6_t *, tcpha_t *, int, uint_t,
729 ip_stack_t *);
730 conn_t *ipcl_lookup_listener_v4(uint16_t, ipaddr_t, zoneid_t, ip_stack_t *);
731 conn_t *ipcl_lookup_listener_v6(uint16_t, in6_addr_t *, uint_t, zoneid_t,
732 ip_stack_t *);
733 int conn_trace_ref(conn_t *);
734 int conn_untrace_ref(conn_t *);
735 void ipcl_conn_cleanup(conn_t *);
736 extern uint_t conn_recvancillary_size(conn_t *, crb_t, ip_recv_attr_t *,
737 mblk_t *, ip_pkt_t *);
738 extern void conn_recvancillary_add(conn_t *, crb_t, ip_recv_attr_t *,
739 ip_pkt_t *, uchar_t *, uint_t);
740 conn_t *ipcl_conn_tcp_lookup_reversed_ipv4(conn_t *, ipha_t *, tcpha_t *,
741 ip_stack_t *);
742 conn_t *ipcl_conn_tcp_lookup_reversed_ipv6(conn_t *, ip6_t *, tcpha_t *,
743 ip_stack_t *);
744
745 extern int ip_create_helper_stream(conn_t *, ldi_ident_t);
746 extern void ip_free_helper_stream(conn_t *);
747 extern int ip_helper_stream_setup(queue_t *, dev_t *, int, int,
748 cred_t *, boolean_t);
749
750 #ifdef __cplusplus
751 }
752 #endif
753
754 #endif /* _INET_IPCLASSIFIER_H */