Print this page
dccp: starting module template
| Split |
Close |
| Expand all |
| Collapse all |
--- old/usr/src/uts/common/inet/ip.h
+++ new/usr/src/uts/common/inet/ip.h
1 1 /*
2 2 * CDDL HEADER START
3 3 *
4 4 * The contents of this file are subject to the terms of the
5 5 * Common Development and Distribution License (the "License").
6 6 * You may not use this file except in compliance with the License.
7 7 *
8 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 9 * or http://www.opensolaris.org/os/licensing.
10 10 * See the License for the specific language governing permissions
11 11 * and limitations under the License.
12 12 *
13 13 * When distributing Covered Code, include this CDDL HEADER in each
14 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 15 * If applicable, add the following below this CDDL HEADER, with the
16 16 * fields enclosed by brackets "[]" replaced with your own identifying
17 17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 18 *
19 19 * CDDL HEADER END
20 20 */
21 21
22 22 /*
23 23 * Copyright (c) 1991, 2010, Oracle and/or its affiliates. All rights reserved.
24 24 * Copyright (c) 1990 Mentat Inc.
25 25 */
26 26
27 27 #ifndef _INET_IP_H
28 28 #define _INET_IP_H
29 29
30 30 #ifdef __cplusplus
31 31 extern "C" {
32 32 #endif
33 33
34 34 #include <sys/isa_defs.h>
35 35 #include <sys/types.h>
36 36 #include <inet/mib2.h>
37 37 #include <inet/nd.h>
38 38 #include <sys/atomic.h>
39 39 #include <net/if_dl.h>
40 40 #include <net/if.h>
41 41 #include <netinet/ip.h>
42 42 #include <netinet/igmp.h>
43 43 #include <sys/neti.h>
44 44 #include <sys/hook.h>
45 45 #include <sys/hook_event.h>
46 46 #include <sys/hook_impl.h>
47 47 #include <inet/ip_stack.h>
48 48
49 49 #ifdef _KERNEL
50 50 #include <netinet/ip6.h>
51 51 #include <sys/avl.h>
52 52 #include <sys/list.h>
53 53 #include <sys/vmem.h>
54 54 #include <sys/squeue.h>
55 55 #include <net/route.h>
56 56 #include <sys/systm.h>
57 57 #include <net/radix.h>
58 58 #include <sys/modhash.h>
59 59
60 60 #ifdef DEBUG
61 61 #define CONN_DEBUG
62 62 #endif
63 63
64 64 #define IP_DEBUG
65 65 /*
66 66 * The mt-streams(9F) flags for the IP module; put here so that other
67 67 * "drivers" that are actually IP (e.g., ICMP, UDP) can use the same set
68 68 * of flags.
69 69 */
70 70 #define IP_DEVMTFLAGS D_MP
71 71 #endif /* _KERNEL */
72 72
73 73 #define IP_MOD_NAME "ip"
74 74 #define IP_DEV_NAME "/dev/ip"
75 75 #define IP6_DEV_NAME "/dev/ip6"
76 76
|
↓ open down ↓ |
76 lines elided |
↑ open up ↑ |
77 77 #define UDP_MOD_NAME "udp"
78 78 #define UDP_DEV_NAME "/dev/udp"
79 79 #define UDP6_DEV_NAME "/dev/udp6"
80 80
81 81 #define TCP_MOD_NAME "tcp"
82 82 #define TCP_DEV_NAME "/dev/tcp"
83 83 #define TCP6_DEV_NAME "/dev/tcp6"
84 84
85 85 #define SCTP_MOD_NAME "sctp"
86 86
87 +#define DCCP_MOD_NAME "dccp"
88 +#define DCCP_DEV_NAME "/dev/dccp"
89 +#define DCCP6_DEV_NAME "/dev/dccp6"
90 +
87 91 #ifndef _IPADDR_T
88 92 #define _IPADDR_T
89 93 typedef uint32_t ipaddr_t;
90 94 #endif
91 95
92 96 /* Number of bits in an address */
93 97 #define IP_ABITS 32
94 98 #define IPV4_ABITS IP_ABITS
95 99 #define IPV6_ABITS 128
96 100 #define IP_MAX_HW_LEN 40
97 101
98 102 #define IP_HOST_MASK (ipaddr_t)0xffffffffU
99 103
100 104 #define IP_CSUM(mp, off, sum) (~ip_cksum(mp, off, sum) & 0xFFFF)
101 105 #define IP_CSUM_PARTIAL(mp, off, sum) ip_cksum(mp, off, sum)
102 106 #define IP_BCSUM_PARTIAL(bp, len, sum) bcksum(bp, len, sum)
103 107
104 108 #define ILL_FRAG_HASH_TBL_COUNT ((unsigned int)64)
105 109 #define ILL_FRAG_HASH_TBL_SIZE (ILL_FRAG_HASH_TBL_COUNT * sizeof (ipfb_t))
106 110
107 111 #define IPV4_ADDR_LEN 4
108 112 #define IP_ADDR_LEN IPV4_ADDR_LEN
109 113 #define IP_ARP_PROTO_TYPE 0x0800
110 114
111 115 #define IPV4_VERSION 4
112 116 #define IP_VERSION IPV4_VERSION
113 117 #define IP_SIMPLE_HDR_LENGTH_IN_WORDS 5
114 118 #define IP_SIMPLE_HDR_LENGTH 20
115 119 #define IP_MAX_HDR_LENGTH 60
116 120
117 121 #define IP_MAX_OPT_LENGTH (IP_MAX_HDR_LENGTH-IP_SIMPLE_HDR_LENGTH)
118 122
119 123 #define IP_MIN_MTU (IP_MAX_HDR_LENGTH + 8) /* 68 bytes */
120 124
121 125 /*
122 126 * XXX IP_MAXPACKET is defined in <netinet/ip.h> as well. At some point the
123 127 * 2 files should be cleaned up to remove all redundant definitions.
124 128 */
125 129 #define IP_MAXPACKET 65535
126 130 #define IP_SIMPLE_HDR_VERSION \
127 131 ((IP_VERSION << 4) | IP_SIMPLE_HDR_LENGTH_IN_WORDS)
128 132
129 133 #define UDPH_SIZE 8
130 134
131 135 /*
132 136 * Constants and type definitions to support IP IOCTL commands
133 137 */
134 138 #define IP_IOCTL (('i'<<8)|'p')
135 139 #define IP_IOC_IRE_DELETE 4
136 140 #define IP_IOC_IRE_DELETE_NO_REPLY 5
137 141 #define IP_IOC_RTS_REQUEST 7
138 142
139 143 /* Common definitions used by IP IOCTL data structures */
140 144 typedef struct ipllcmd_s {
141 145 uint_t ipllc_cmd;
142 146 uint_t ipllc_name_offset;
143 147 uint_t ipllc_name_length;
144 148 } ipllc_t;
145 149
146 150 /* IP IRE Delete Command Structure. */
147 151 typedef struct ipid_s {
148 152 ipllc_t ipid_ipllc;
149 153 uint_t ipid_ire_type;
150 154 uint_t ipid_addr_offset;
151 155 uint_t ipid_addr_length;
152 156 uint_t ipid_mask_offset;
153 157 uint_t ipid_mask_length;
154 158 } ipid_t;
155 159
156 160 #define ipid_cmd ipid_ipllc.ipllc_cmd
157 161
158 162 #ifdef _KERNEL
159 163 /*
160 164 * Temporary state for ip options parser.
161 165 */
162 166 typedef struct ipoptp_s
163 167 {
164 168 uint8_t *ipoptp_next; /* next option to look at */
165 169 uint8_t *ipoptp_end; /* end of options */
166 170 uint8_t *ipoptp_cur; /* start of current option */
167 171 uint8_t ipoptp_len; /* length of current option */
168 172 uint32_t ipoptp_flags;
169 173 } ipoptp_t;
170 174
171 175 /*
172 176 * Flag(s) for ipoptp_flags
173 177 */
174 178 #define IPOPTP_ERROR 0x00000001
175 179 #endif /* _KERNEL */
176 180
177 181 /* Controls forwarding of IP packets, set via ipadm(1M)/ndd(1M) */
178 182 #define IP_FORWARD_NEVER 0
179 183 #define IP_FORWARD_ALWAYS 1
180 184
181 185 #define WE_ARE_FORWARDING(ipst) ((ipst)->ips_ip_forwarding == IP_FORWARD_ALWAYS)
182 186
183 187 #define IPH_HDR_LENGTH(ipha) \
184 188 ((int)(((ipha_t *)ipha)->ipha_version_and_hdr_length & 0xF) << 2)
185 189
186 190 #define IPH_HDR_VERSION(ipha) \
187 191 ((int)(((ipha_t *)ipha)->ipha_version_and_hdr_length) >> 4)
188 192
189 193 #ifdef _KERNEL
190 194 /*
191 195 * IP reassembly macros. We hide starting and ending offsets in b_next and
192 196 * b_prev of messages on the reassembly queue. The messages are chained using
193 197 * b_cont. These macros are used in ip_reassemble() so we don't have to see
194 198 * the ugly casts and assignments.
195 199 * Note that the offsets are <= 64k i.e. a uint_t is sufficient to represent
196 200 * them.
197 201 */
198 202 #define IP_REASS_START(mp) ((uint_t)(uintptr_t)((mp)->b_next))
199 203 #define IP_REASS_SET_START(mp, u) \
200 204 ((mp)->b_next = (mblk_t *)(uintptr_t)(u))
201 205 #define IP_REASS_END(mp) ((uint_t)(uintptr_t)((mp)->b_prev))
202 206 #define IP_REASS_SET_END(mp, u) \
203 207 ((mp)->b_prev = (mblk_t *)(uintptr_t)(u))
204 208
205 209 #define IP_REASS_COMPLETE 0x1
206 210 #define IP_REASS_PARTIAL 0x2
207 211 #define IP_REASS_FAILED 0x4
208 212
209 213 /*
|
↓ open down ↓ |
113 lines elided |
↑ open up ↑ |
210 214 * Test to determine whether this is a module instance of IP or a
211 215 * driver instance of IP.
212 216 */
213 217 #define CONN_Q(q) (WR(q)->q_next == NULL)
214 218
215 219 #define Q_TO_CONN(q) ((conn_t *)(q)->q_ptr)
216 220 #define Q_TO_TCP(q) (Q_TO_CONN((q))->conn_tcp)
217 221 #define Q_TO_UDP(q) (Q_TO_CONN((q))->conn_udp)
218 222 #define Q_TO_ICMP(q) (Q_TO_CONN((q))->conn_icmp)
219 223 #define Q_TO_RTS(q) (Q_TO_CONN((q))->conn_rts)
224 +#define Q_TO_DCCP(q) (Q_TO_CONN((q))->conn_dccp)
220 225
221 226 #define CONNP_TO_WQ(connp) ((connp)->conn_wq)
222 227 #define CONNP_TO_RQ(connp) ((connp)->conn_rq)
223 228
224 229 #define GRAB_CONN_LOCK(q) { \
225 230 if (q != NULL && CONN_Q(q)) \
226 231 mutex_enter(&(Q_TO_CONN(q))->conn_lock); \
227 232 }
228 233
229 234 #define RELEASE_CONN_LOCK(q) { \
230 235 if (q != NULL && CONN_Q(q)) \
231 236 mutex_exit(&(Q_TO_CONN(q))->conn_lock); \
232 237 }
233 238
234 239 /*
235 240 * Ref counter macros for ioctls. This provides a guard for TCP to stop
236 241 * tcp_close from removing the rq/wq whilst an ioctl is still in flight on the
237 242 * stream. The ioctl could have been queued on e.g. an ipsq. tcp_close will wait
238 243 * until the ioctlref count is zero before proceeding.
239 244 * Ideally conn_oper_pending_ill would be used for this purpose. However, in the
240 245 * case where an ioctl is aborted or interrupted, it can be cleared prematurely.
241 246 * There are also some race possibilities between ip and the stream head which
242 247 * can also end up with conn_oper_pending_ill being cleared prematurely. So, to
243 248 * avoid these situations, we use a dedicated ref counter for ioctls which is
244 249 * used in addition to and in parallel with the normal conn_ref count.
245 250 */
246 251 #define CONN_INC_IOCTLREF_LOCKED(connp) { \
247 252 ASSERT(MUTEX_HELD(&(connp)->conn_lock)); \
248 253 DTRACE_PROBE1(conn__inc__ioctlref, conn_t *, (connp)); \
249 254 (connp)->conn_ioctlref++; \
250 255 mutex_exit(&(connp)->conn_lock); \
251 256 }
252 257
253 258 #define CONN_INC_IOCTLREF(connp) { \
254 259 mutex_enter(&(connp)->conn_lock); \
255 260 CONN_INC_IOCTLREF_LOCKED(connp); \
256 261 }
257 262
258 263 #define CONN_DEC_IOCTLREF(connp) { \
259 264 mutex_enter(&(connp)->conn_lock); \
260 265 DTRACE_PROBE1(conn__dec__ioctlref, conn_t *, (connp)); \
261 266 /* Make sure conn_ioctlref will not underflow. */ \
262 267 ASSERT((connp)->conn_ioctlref != 0); \
263 268 if ((--(connp)->conn_ioctlref == 0) && \
264 269 ((connp)->conn_state_flags & CONN_CLOSING)) { \
265 270 cv_broadcast(&(connp)->conn_cv); \
266 271 } \
267 272 mutex_exit(&(connp)->conn_lock); \
268 273 }
269 274
270 275
271 276 /*
272 277 * Complete the pending operation. Usually an ioctl. Can also
273 278 * be a bind or option management request that got enqueued
274 279 * in an ipsq_t. Called on completion of the operation.
275 280 */
276 281 #define CONN_OPER_PENDING_DONE(connp) { \
277 282 mutex_enter(&(connp)->conn_lock); \
278 283 (connp)->conn_oper_pending_ill = NULL; \
279 284 cv_broadcast(&(connp)->conn_refcv); \
280 285 mutex_exit(&(connp)->conn_lock); \
281 286 CONN_DEC_REF(connp); \
282 287 }
283 288
284 289 /*
285 290 * Values for squeue switch:
286 291 */
287 292 #define IP_SQUEUE_ENTER_NODRAIN 1
288 293 #define IP_SQUEUE_ENTER 2
289 294 #define IP_SQUEUE_FILL 3
290 295
291 296 extern int ip_squeue_flag;
292 297
293 298 /* IP Fragmentation Reassembly Header */
294 299 typedef struct ipf_s {
295 300 struct ipf_s *ipf_hash_next;
296 301 struct ipf_s **ipf_ptphn; /* Pointer to previous hash next. */
297 302 uint32_t ipf_ident; /* Ident to match. */
298 303 uint8_t ipf_protocol; /* Protocol to match. */
299 304 uchar_t ipf_last_frag_seen : 1; /* Last fragment seen ? */
300 305 time_t ipf_timestamp; /* Reassembly start time. */
301 306 mblk_t *ipf_mp; /* mblk we live in. */
302 307 mblk_t *ipf_tail_mp; /* Frag queue tail pointer. */
303 308 int ipf_hole_cnt; /* Number of holes (hard-case). */
304 309 int ipf_end; /* Tail end offset (0 -> hard-case). */
305 310 uint_t ipf_gen; /* Frag queue generation */
306 311 size_t ipf_count; /* Count of bytes used by frag */
307 312 uint_t ipf_nf_hdr_len; /* Length of nonfragmented header */
308 313 in6_addr_t ipf_v6src; /* IPv6 source address */
309 314 in6_addr_t ipf_v6dst; /* IPv6 dest address */
310 315 uint_t ipf_prev_nexthdr_offset; /* Offset for nexthdr value */
311 316 uint8_t ipf_ecn; /* ECN info for the fragments */
312 317 uint8_t ipf_num_dups; /* Number of times dup frags recvd */
313 318 uint16_t ipf_checksum_flags; /* Hardware checksum flags */
314 319 uint32_t ipf_checksum; /* Partial checksum of fragment data */
315 320 } ipf_t;
316 321
317 322 /*
318 323 * IPv4 Fragments
319 324 */
320 325 #define IS_V4_FRAGMENT(ipha_fragment_offset_and_flags) \
321 326 (((ntohs(ipha_fragment_offset_and_flags) & IPH_OFFSET) != 0) || \
322 327 ((ntohs(ipha_fragment_offset_and_flags) & IPH_MF) != 0))
323 328
324 329 #define ipf_src V4_PART_OF_V6(ipf_v6src)
325 330 #define ipf_dst V4_PART_OF_V6(ipf_v6dst)
326 331
327 332 #endif /* _KERNEL */
328 333
329 334 /* ICMP types */
330 335 #define ICMP_ECHO_REPLY 0
331 336 #define ICMP_DEST_UNREACHABLE 3
332 337 #define ICMP_SOURCE_QUENCH 4
333 338 #define ICMP_REDIRECT 5
334 339 #define ICMP_ECHO_REQUEST 8
335 340 #define ICMP_ROUTER_ADVERTISEMENT 9
336 341 #define ICMP_ROUTER_SOLICITATION 10
337 342 #define ICMP_TIME_EXCEEDED 11
338 343 #define ICMP_PARAM_PROBLEM 12
339 344 #define ICMP_TIME_STAMP_REQUEST 13
340 345 #define ICMP_TIME_STAMP_REPLY 14
341 346 #define ICMP_INFO_REQUEST 15
342 347 #define ICMP_INFO_REPLY 16
343 348 #define ICMP_ADDRESS_MASK_REQUEST 17
344 349 #define ICMP_ADDRESS_MASK_REPLY 18
345 350
346 351 /* Evaluates to true if the ICMP type is an ICMP error */
347 352 #define ICMP_IS_ERROR(type) ( \
348 353 (type) == ICMP_DEST_UNREACHABLE || \
349 354 (type) == ICMP_SOURCE_QUENCH || \
350 355 (type) == ICMP_TIME_EXCEEDED || \
351 356 (type) == ICMP_PARAM_PROBLEM)
352 357
353 358 /* ICMP_TIME_EXCEEDED codes */
354 359 #define ICMP_TTL_EXCEEDED 0
355 360 #define ICMP_REASSEMBLY_TIME_EXCEEDED 1
356 361
357 362 /* ICMP_DEST_UNREACHABLE codes */
358 363 #define ICMP_NET_UNREACHABLE 0
359 364 #define ICMP_HOST_UNREACHABLE 1
360 365 #define ICMP_PROTOCOL_UNREACHABLE 2
361 366 #define ICMP_PORT_UNREACHABLE 3
362 367 #define ICMP_FRAGMENTATION_NEEDED 4
363 368 #define ICMP_SOURCE_ROUTE_FAILED 5
364 369 #define ICMP_DEST_NET_UNKNOWN 6
365 370 #define ICMP_DEST_HOST_UNKNOWN 7
366 371 #define ICMP_SRC_HOST_ISOLATED 8
367 372 #define ICMP_DEST_NET_UNREACH_ADMIN 9
368 373 #define ICMP_DEST_HOST_UNREACH_ADMIN 10
369 374 #define ICMP_DEST_NET_UNREACH_TOS 11
370 375 #define ICMP_DEST_HOST_UNREACH_TOS 12
371 376
372 377 /* ICMP Header Structure */
373 378 typedef struct icmph_s {
374 379 uint8_t icmph_type;
375 380 uint8_t icmph_code;
376 381 uint16_t icmph_checksum;
377 382 union {
378 383 struct { /* ECHO request/response structure */
379 384 uint16_t u_echo_ident;
380 385 uint16_t u_echo_seqnum;
381 386 } u_echo;
382 387 struct { /* Destination unreachable structure */
383 388 uint16_t u_du_zero;
384 389 uint16_t u_du_mtu;
385 390 } u_du;
386 391 struct { /* Parameter problem structure */
387 392 uint8_t u_pp_ptr;
388 393 uint8_t u_pp_rsvd[3];
389 394 } u_pp;
390 395 struct { /* Redirect structure */
391 396 ipaddr_t u_rd_gateway;
392 397 } u_rd;
393 398 } icmph_u;
394 399 } icmph_t;
395 400
396 401 #define icmph_echo_ident icmph_u.u_echo.u_echo_ident
397 402 #define icmph_echo_seqnum icmph_u.u_echo.u_echo_seqnum
398 403 #define icmph_du_zero icmph_u.u_du.u_du_zero
399 404 #define icmph_du_mtu icmph_u.u_du.u_du_mtu
400 405 #define icmph_pp_ptr icmph_u.u_pp.u_pp_ptr
401 406 #define icmph_rd_gateway icmph_u.u_rd.u_rd_gateway
402 407
403 408 #define ICMPH_SIZE 8
404 409
405 410 /*
406 411 * Minimum length of transport layer header included in an ICMP error
407 412 * message for it to be considered valid.
408 413 */
409 414 #define ICMP_MIN_TP_HDR_LEN 8
410 415
411 416 /* Aligned IP header */
412 417 typedef struct ipha_s {
413 418 uint8_t ipha_version_and_hdr_length;
414 419 uint8_t ipha_type_of_service;
415 420 uint16_t ipha_length;
416 421 uint16_t ipha_ident;
417 422 uint16_t ipha_fragment_offset_and_flags;
418 423 uint8_t ipha_ttl;
419 424 uint8_t ipha_protocol;
420 425 uint16_t ipha_hdr_checksum;
421 426 ipaddr_t ipha_src;
422 427 ipaddr_t ipha_dst;
423 428 } ipha_t;
424 429
425 430 /*
426 431 * IP Flags
427 432 *
428 433 * Some of these constant names are copied for the DTrace IP provider in
429 434 * usr/src/lib/libdtrace/common/{ip.d.in, ip.sed.in}, which should be kept
430 435 * in sync.
431 436 */
432 437 #define IPH_DF 0x4000 /* Don't fragment */
433 438 #define IPH_MF 0x2000 /* More fragments to come */
434 439 #define IPH_OFFSET 0x1FFF /* Where the offset lives */
435 440
436 441 /* Byte-order specific values */
437 442 #ifdef _BIG_ENDIAN
438 443 #define IPH_DF_HTONS 0x4000 /* Don't fragment */
439 444 #define IPH_MF_HTONS 0x2000 /* More fragments to come */
440 445 #define IPH_OFFSET_HTONS 0x1FFF /* Where the offset lives */
441 446 #else
442 447 #define IPH_DF_HTONS 0x0040 /* Don't fragment */
443 448 #define IPH_MF_HTONS 0x0020 /* More fragments to come */
444 449 #define IPH_OFFSET_HTONS 0xFF1F /* Where the offset lives */
445 450 #endif
446 451
447 452 /* ECN code points for IPv4 TOS byte and IPv6 traffic class octet. */
448 453 #define IPH_ECN_NECT 0x0 /* Not ECN-Capable Transport */
449 454 #define IPH_ECN_ECT1 0x1 /* ECN-Capable Transport, ECT(1) */
450 455 #define IPH_ECN_ECT0 0x2 /* ECN-Capable Transport, ECT(0) */
451 456 #define IPH_ECN_CE 0x3 /* ECN-Congestion Experienced (CE) */
452 457
453 458 struct ill_s;
454 459
455 460 typedef void ip_v6intfid_func_t(struct ill_s *, in6_addr_t *);
456 461 typedef void ip_v6mapinfo_func_t(struct ill_s *, uchar_t *, uchar_t *);
457 462 typedef void ip_v4mapinfo_func_t(struct ill_s *, uchar_t *, uchar_t *);
458 463
459 464 /* IP Mac info structure */
460 465 typedef struct ip_m_s {
461 466 t_uscalar_t ip_m_mac_type; /* From <sys/dlpi.h> */
462 467 int ip_m_type; /* From <net/if_types.h> */
463 468 t_uscalar_t ip_m_ipv4sap;
464 469 t_uscalar_t ip_m_ipv6sap;
465 470 ip_v4mapinfo_func_t *ip_m_v4mapping;
466 471 ip_v6mapinfo_func_t *ip_m_v6mapping;
467 472 ip_v6intfid_func_t *ip_m_v6intfid;
468 473 ip_v6intfid_func_t *ip_m_v6destintfid;
469 474 } ip_m_t;
470 475
471 476 /*
472 477 * The following functions attempt to reduce the link layer dependency
473 478 * of the IP stack. The current set of link specific operations are:
474 479 * a. map from IPv4 class D (224.0/4) multicast address range or the
475 480 * IPv6 multicast address range (ff00::/8) to the link layer multicast
476 481 * address.
477 482 * b. derive the default IPv6 interface identifier from the interface.
478 483 * c. derive the default IPv6 destination interface identifier from
479 484 * the interface (point-to-point only).
480 485 */
481 486 extern void ip_mcast_mapping(struct ill_s *, uchar_t *, uchar_t *);
482 487 /* ip_m_v6*intfid return void and are never NULL */
483 488 #define MEDIA_V6INTFID(ip_m, ill, v6ptr) (ip_m)->ip_m_v6intfid(ill, v6ptr)
484 489 #define MEDIA_V6DESTINTFID(ip_m, ill, v6ptr) \
485 490 (ip_m)->ip_m_v6destintfid(ill, v6ptr)
486 491
487 492 /* Router entry types */
488 493 #define IRE_BROADCAST 0x0001 /* Route entry for broadcast address */
489 494 #define IRE_DEFAULT 0x0002 /* Route entry for default gateway */
490 495 #define IRE_LOCAL 0x0004 /* Route entry for local address */
491 496 #define IRE_LOOPBACK 0x0008 /* Route entry for loopback address */
492 497 #define IRE_PREFIX 0x0010 /* Route entry for prefix routes */
493 498 #ifndef _KERNEL
494 499 /* Keep so user-level still compiles */
495 500 #define IRE_CACHE 0x0020 /* Cached Route entry */
496 501 #endif
497 502 #define IRE_IF_NORESOLVER 0x0040 /* Route entry for local interface */
498 503 /* net without any address mapping. */
499 504 #define IRE_IF_RESOLVER 0x0080 /* Route entry for local interface */
500 505 /* net with resolver. */
501 506 #define IRE_HOST 0x0100 /* Host route entry */
502 507 /* Keep so user-level still compiles */
503 508 #define IRE_HOST_REDIRECT 0x0200 /* only used for T_SVR4_OPTMGMT_REQ */
504 509 #define IRE_IF_CLONE 0x0400 /* Per host clone of IRE_IF */
505 510 #define IRE_MULTICAST 0x0800 /* Special - not in table */
506 511 #define IRE_NOROUTE 0x1000 /* Special - not in table */
507 512
508 513 #define IRE_INTERFACE (IRE_IF_NORESOLVER | IRE_IF_RESOLVER)
509 514
510 515 #define IRE_IF_ALL (IRE_IF_NORESOLVER | IRE_IF_RESOLVER | \
511 516 IRE_IF_CLONE)
512 517 #define IRE_OFFSUBNET (IRE_DEFAULT | IRE_PREFIX | IRE_HOST)
513 518 #define IRE_OFFLINK IRE_OFFSUBNET
514 519 /*
515 520 * Note that we view IRE_NOROUTE as ONLINK since we can "send" to them without
516 521 * going through a router; the result of sending will be an error/icmp error.
517 522 */
518 523 #define IRE_ONLINK (IRE_IF_ALL|IRE_LOCAL|IRE_LOOPBACK| \
519 524 IRE_BROADCAST|IRE_MULTICAST|IRE_NOROUTE)
520 525
521 526 /* Arguments to ire_flush_cache() */
522 527 #define IRE_FLUSH_DELETE 0
523 528 #define IRE_FLUSH_ADD 1
524 529 #define IRE_FLUSH_GWCHANGE 2
525 530
526 531 /*
527 532 * Flags to ire_route_recursive
528 533 */
529 534 #define IRR_NONE 0
530 535 #define IRR_ALLOCATE 1 /* OK to allocate IRE_IF_CLONE */
531 536 #define IRR_INCOMPLETE 2 /* OK to return incomplete chain */
532 537
533 538 /*
534 539 * Open/close synchronization flags.
535 540 * These are kept in a separate field in the conn and the synchronization
536 541 * depends on the atomic 32 bit access to that field.
537 542 */
538 543 #define CONN_CLOSING 0x01 /* ip_close waiting for ip_wsrv */
539 544 #define CONN_CONDEMNED 0x02 /* conn is closing, no more refs */
540 545 #define CONN_INCIPIENT 0x04 /* conn not yet visible, no refs */
541 546 #define CONN_QUIESCED 0x08 /* conn is now quiescent */
542 547 #define CONN_UPDATE_ILL 0x10 /* conn_update_ill in progress */
543 548
544 549 /*
545 550 * Flags for dce_flags field. Specifies which information has been set.
546 551 * dce_ident is always present, but the other ones are identified by the flags.
547 552 */
548 553 #define DCEF_DEFAULT 0x0001 /* Default DCE - no pmtu or uinfo */
549 554 #define DCEF_PMTU 0x0002 /* Different than interface MTU */
550 555 #define DCEF_UINFO 0x0004 /* dce_uinfo set */
551 556 #define DCEF_TOO_SMALL_PMTU 0x0008 /* Smaller than IPv4/IPv6 MIN */
552 557
553 558 #ifdef _KERNEL
554 559 /*
555 560 * Extra structures need for per-src-addr filtering (IGMPv3/MLDv2)
556 561 */
557 562 #define MAX_FILTER_SIZE 64
558 563
559 564 typedef struct slist_s {
560 565 int sl_numsrc;
561 566 in6_addr_t sl_addr[MAX_FILTER_SIZE];
562 567 } slist_t;
563 568
564 569 /*
565 570 * Following struct is used to maintain retransmission state for
566 571 * a multicast group. One rtx_state_t struct is an in-line field
567 572 * of the ilm_t struct; the slist_ts in the rtx_state_t struct are
568 573 * alloc'd as needed.
569 574 */
570 575 typedef struct rtx_state_s {
571 576 uint_t rtx_timer; /* retrans timer */
572 577 int rtx_cnt; /* retrans count */
573 578 int rtx_fmode_cnt; /* retrans count for fmode change */
574 579 slist_t *rtx_allow;
575 580 slist_t *rtx_block;
576 581 } rtx_state_t;
577 582
578 583 /*
579 584 * Used to construct list of multicast address records that will be
580 585 * sent in a single listener report.
581 586 */
582 587 typedef struct mrec_s {
583 588 struct mrec_s *mrec_next;
584 589 uint8_t mrec_type;
585 590 uint8_t mrec_auxlen; /* currently unused */
586 591 in6_addr_t mrec_group;
587 592 slist_t mrec_srcs;
588 593 } mrec_t;
589 594
590 595 /* Group membership list per upper conn */
591 596
592 597 /*
593 598 * We record the multicast information from the socket option in
594 599 * ilg_ifaddr/ilg_ifindex. This allows rejoining the group in the case when
595 600 * the ifaddr (or ifindex) disappears and later reappears, potentially on
596 601 * a different ill. The IPv6 multicast socket options and ioctls all specify
597 602 * the interface using an ifindex. For IPv4 some socket options/ioctls use
598 603 * the interface address and others use the index. We record here the method
599 604 * that was actually used (and leave the other of ilg_ifaddr or ilg_ifindex)
600 605 * at zero so that we can rejoin the way the application intended.
601 606 *
602 607 * We track the ill on which we will or already have joined an ilm using
603 608 * ilg_ill. When we have succeeded joining the ilm and have a refhold on it
604 609 * then we set ilg_ilm. Thus intentionally there is a window where ilg_ill is
605 610 * set and ilg_ilm is not set. This allows clearing ilg_ill as a signal that
606 611 * the ill is being unplumbed and the ilm should be discarded.
607 612 *
608 613 * ilg records the state of multicast memberships of a socket end point.
609 614 * ilm records the state of multicast memberships with the driver and is
610 615 * maintained per interface.
611 616 *
612 617 * The ilg state is protected by conn_ilg_lock.
613 618 * The ilg will not be freed until ilg_refcnt drops to zero.
614 619 */
615 620 typedef struct ilg_s {
616 621 struct ilg_s *ilg_next;
617 622 struct ilg_s **ilg_ptpn;
618 623 struct conn_s *ilg_connp; /* Back pointer to get lock */
619 624 in6_addr_t ilg_v6group;
620 625 ipaddr_t ilg_ifaddr; /* For some IPv4 cases */
621 626 uint_t ilg_ifindex; /* IPv6 and some other IPv4 cases */
622 627 struct ill_s *ilg_ill; /* Where ilm is joined. No refhold */
623 628 struct ilm_s *ilg_ilm; /* With ilm_refhold */
624 629 uint_t ilg_refcnt;
625 630 mcast_record_t ilg_fmode; /* MODE_IS_INCLUDE/MODE_IS_EXCLUDE */
626 631 slist_t *ilg_filter;
627 632 boolean_t ilg_condemned; /* Conceptually deleted */
628 633 } ilg_t;
629 634
630 635 /*
631 636 * Multicast address list entry for ill.
632 637 * ilm_ill is used by IPv4 and IPv6
633 638 *
634 639 * The ilm state (and other multicast state on the ill) is protected by
635 640 * ill_mcast_lock. Operations that change state on both an ilg and ilm
636 641 * in addition use ill_mcast_serializer to ensure that we can't have
637 642 * interleaving between e.g., add and delete operations for the same conn_t,
638 643 * group, and ill. The ill_mcast_serializer is also used to ensure that
639 644 * multicast group joins do not occur on an interface that is in the process
640 645 * of joining an IPMP group.
641 646 *
642 647 * The comment below (and for other netstack_t references) refers
643 648 * to the fact that we only do netstack_hold in particular cases,
644 649 * such as the references from open endpoints (ill_t and conn_t's
645 650 * pointers). Internally within IP we rely on IP's ability to cleanup e.g.
646 651 * ire_t's when an ill goes away.
647 652 */
648 653 typedef struct ilm_s {
649 654 in6_addr_t ilm_v6addr;
650 655 int ilm_refcnt;
651 656 uint_t ilm_timer; /* IGMP/MLD query resp timer, in msec */
652 657 struct ilm_s *ilm_next; /* Linked list for each ill */
653 658 uint_t ilm_state; /* state of the membership */
654 659 struct ill_s *ilm_ill; /* Back pointer to ill - ill_ilm_cnt */
655 660 zoneid_t ilm_zoneid;
656 661 int ilm_no_ilg_cnt; /* number of joins w/ no ilg */
657 662 mcast_record_t ilm_fmode; /* MODE_IS_INCLUDE/MODE_IS_EXCLUDE */
658 663 slist_t *ilm_filter; /* source filter list */
659 664 slist_t *ilm_pendsrcs; /* relevant src addrs for pending req */
660 665 rtx_state_t ilm_rtx; /* SCR retransmission state */
661 666 ipaddr_t ilm_ifaddr; /* For IPv4 netstat */
662 667 ip_stack_t *ilm_ipst; /* Does not have a netstack_hold */
663 668 } ilm_t;
664 669
665 670 #define ilm_addr V4_PART_OF_V6(ilm_v6addr)
666 671
667 672 /*
668 673 * Soft reference to an IPsec SA.
669 674 *
670 675 * On relative terms, conn's can be persistent (living as long as the
671 676 * processes which create them), while SA's are ephemeral (dying when
672 677 * they hit their time-based or byte-based lifetimes).
673 678 *
674 679 * We could hold a hard reference to an SA from an ipsec_latch_t,
675 680 * but this would cause expired SA's to linger for a potentially
676 681 * unbounded time.
677 682 *
678 683 * Instead, we remember the hash bucket number and bucket generation
679 684 * in addition to the pointer. The bucket generation is incremented on
680 685 * each deletion.
681 686 */
682 687 typedef struct ipsa_ref_s
683 688 {
684 689 struct ipsa_s *ipsr_sa;
685 690 struct isaf_s *ipsr_bucket;
686 691 uint64_t ipsr_gen;
687 692 } ipsa_ref_t;
688 693
689 694 /*
690 695 * IPsec "latching" state.
691 696 *
692 697 * In the presence of IPsec policy, fully-bound conn's bind a connection
693 698 * to more than just the 5-tuple, but also a specific IPsec action and
694 699 * identity-pair.
695 700 * The identity pair is accessed from both the receive and transmit side
696 701 * hence it is maintained in the ipsec_latch_t structure. conn_latch and
697 702 * ixa_ipsec_latch points to it.
698 703 * The policy and actions are stored in conn_latch_in_policy and
699 704 * conn_latch_in_action for the inbound side, and in ixa_ipsec_policy and
700 705 * ixa_ipsec_action for the transmit side.
701 706 *
702 707 * As an optimization, we also cache soft references to IPsec SA's in
703 708 * ip_xmit_attr_t so that we can fast-path around most of the work needed for
704 709 * outbound IPsec SA selection.
705 710 */
706 711 typedef struct ipsec_latch_s
707 712 {
708 713 kmutex_t ipl_lock;
709 714 uint32_t ipl_refcnt;
710 715
711 716 struct ipsid_s *ipl_local_cid;
712 717 struct ipsid_s *ipl_remote_cid;
713 718 unsigned int
714 719 ipl_ids_latched : 1,
715 720
716 721 ipl_pad_to_bit_31 : 31;
717 722 } ipsec_latch_t;
718 723
719 724 #define IPLATCH_REFHOLD(ipl) { \
720 725 atomic_add_32(&(ipl)->ipl_refcnt, 1); \
721 726 ASSERT((ipl)->ipl_refcnt != 0); \
722 727 }
723 728
724 729 #define IPLATCH_REFRELE(ipl) { \
725 730 ASSERT((ipl)->ipl_refcnt != 0); \
726 731 membar_exit(); \
727 732 if (atomic_add_32_nv(&(ipl)->ipl_refcnt, -1) == 0) \
728 733 iplatch_free(ipl); \
729 734 }
730 735
731 736 /*
732 737 * peer identity structure.
733 738 */
734 739 typedef struct conn_s conn_t;
735 740
736 741 /*
737 742 * This is used to match an inbound/outbound datagram with policy.
738 743 */
739 744 typedef struct ipsec_selector {
740 745 in6_addr_t ips_local_addr_v6;
741 746 in6_addr_t ips_remote_addr_v6;
742 747 uint16_t ips_local_port;
743 748 uint16_t ips_remote_port;
744 749 uint8_t ips_icmp_type;
745 750 uint8_t ips_icmp_code;
746 751 uint8_t ips_protocol;
747 752 uint8_t ips_isv4 : 1,
748 753 ips_is_icmp_inv_acq: 1;
749 754 } ipsec_selector_t;
750 755
751 756 /*
752 757 * Note that we put v4 addresses in the *first* 32-bit word of the
753 758 * selector rather than the last to simplify the prefix match/mask code
754 759 * in spd.c
755 760 */
756 761 #define ips_local_addr_v4 ips_local_addr_v6.s6_addr32[0]
757 762 #define ips_remote_addr_v4 ips_remote_addr_v6.s6_addr32[0]
758 763
759 764 /* Values used in IP by IPSEC Code */
760 765 #define IPSEC_OUTBOUND B_TRUE
761 766 #define IPSEC_INBOUND B_FALSE
762 767
763 768 /*
764 769 * There are two variants in policy failures. The packet may come in
765 770 * secure when not needed (IPSEC_POLICY_???_NOT_NEEDED) or it may not
766 771 * have the desired level of protection (IPSEC_POLICY_MISMATCH).
767 772 */
768 773 #define IPSEC_POLICY_NOT_NEEDED 0
769 774 #define IPSEC_POLICY_MISMATCH 1
770 775 #define IPSEC_POLICY_AUTH_NOT_NEEDED 2
771 776 #define IPSEC_POLICY_ENCR_NOT_NEEDED 3
772 777 #define IPSEC_POLICY_SE_NOT_NEEDED 4
773 778 #define IPSEC_POLICY_MAX 5 /* Always max + 1. */
774 779
775 780 /*
776 781 * Check with IPSEC inbound policy if
777 782 *
778 783 * 1) per-socket policy is present - indicated by conn_in_enforce_policy.
779 784 * 2) Or if we have not cached policy on the conn and the global policy is
780 785 * non-empty.
781 786 */
782 787 #define CONN_INBOUND_POLICY_PRESENT(connp, ipss) \
783 788 ((connp)->conn_in_enforce_policy || \
784 789 (!((connp)->conn_policy_cached) && \
785 790 (ipss)->ipsec_inbound_v4_policy_present))
786 791
787 792 #define CONN_INBOUND_POLICY_PRESENT_V6(connp, ipss) \
788 793 ((connp)->conn_in_enforce_policy || \
789 794 (!(connp)->conn_policy_cached && \
790 795 (ipss)->ipsec_inbound_v6_policy_present))
791 796
792 797 #define CONN_OUTBOUND_POLICY_PRESENT(connp, ipss) \
793 798 ((connp)->conn_out_enforce_policy || \
794 799 (!((connp)->conn_policy_cached) && \
795 800 (ipss)->ipsec_outbound_v4_policy_present))
796 801
797 802 #define CONN_OUTBOUND_POLICY_PRESENT_V6(connp, ipss) \
798 803 ((connp)->conn_out_enforce_policy || \
799 804 (!(connp)->conn_policy_cached && \
800 805 (ipss)->ipsec_outbound_v6_policy_present))
801 806
802 807 /*
803 808 * Information cached in IRE for upper layer protocol (ULP).
804 809 */
805 810 typedef struct iulp_s {
806 811 boolean_t iulp_set; /* Is any metric set? */
807 812 uint32_t iulp_ssthresh; /* Slow start threshold (TCP). */
808 813 clock_t iulp_rtt; /* Guestimate in millisecs. */
809 814 clock_t iulp_rtt_sd; /* Cached value of RTT variance. */
810 815 uint32_t iulp_spipe; /* Send pipe size. */
811 816 uint32_t iulp_rpipe; /* Receive pipe size. */
812 817 uint32_t iulp_rtomax; /* Max round trip timeout. */
813 818 uint32_t iulp_sack; /* Use SACK option (TCP)? */
814 819 uint32_t iulp_mtu; /* Setable with routing sockets */
815 820
816 821 uint32_t
817 822 iulp_tstamp_ok : 1, /* Use timestamp option (TCP)? */
818 823 iulp_wscale_ok : 1, /* Use window scale option (TCP)? */
819 824 iulp_ecn_ok : 1, /* Enable ECN (for TCP)? */
820 825 iulp_pmtud_ok : 1, /* Enable PMTUd? */
821 826
822 827 /* These three are passed out by ip_set_destination */
823 828 iulp_localnet: 1, /* IRE_ONLINK */
824 829 iulp_loopback: 1, /* IRE_LOOPBACK */
825 830 iulp_local: 1, /* IRE_LOCAL */
826 831
827 832 iulp_not_used : 25;
828 833 } iulp_t;
829 834
830 835 /*
831 836 * The conn drain list structure (idl_t), protected by idl_lock. Each conn_t
832 837 * inserted in the list points back at this idl_t using conn_idl, and is
833 838 * chained by conn_drain_next and conn_drain_prev, which are also protected by
834 839 * idl_lock. When flow control is relieved, either ip_wsrv() (STREAMS) or
835 840 * ill_flow_enable() (non-STREAMS) will call conn_drain().
836 841 *
837 842 * The conn drain list, idl_t, itself is part of tx cookie list structure.
838 843 * A tx cookie list points to a blocked Tx ring and contains the list of
839 844 * all conn's that are blocked due to the flow-controlled Tx ring (via
840 845 * the idl drain list). Note that a link can have multiple Tx rings. The
841 846 * drain list will store the conn's blocked due to Tx ring being flow
842 847 * controlled.
843 848 */
844 849
845 850 typedef uintptr_t ip_mac_tx_cookie_t;
846 851 typedef struct idl_s idl_t;
847 852 typedef struct idl_tx_list_s idl_tx_list_t;
848 853
849 854 struct idl_tx_list_s {
850 855 ip_mac_tx_cookie_t txl_cookie;
851 856 kmutex_t txl_lock; /* Lock for this list */
852 857 idl_t *txl_drain_list;
853 858 int txl_drain_index;
854 859 };
855 860
856 861 struct idl_s {
857 862 conn_t *idl_conn; /* Head of drain list */
858 863 kmutex_t idl_lock; /* Lock for this list */
859 864 idl_tx_list_t *idl_itl;
860 865 };
861 866
862 867 /*
863 868 * Interface route structure which holds the necessary information to recreate
864 869 * routes that are tied to an interface i.e. have ire_ill set.
865 870 *
866 871 * These routes which were initially created via a routing socket or via the
867 872 * SIOCADDRT ioctl may be gateway routes (RTF_GATEWAY being set) or may be
868 873 * traditional interface routes. When an ill comes back up after being
869 874 * down, this information will be used to recreate the routes. These
870 875 * are part of an mblk_t chain that hangs off of the ILL (ill_saved_ire_mp).
871 876 */
872 877 typedef struct ifrt_s {
873 878 ushort_t ifrt_type; /* Type of IRE */
874 879 in6_addr_t ifrt_v6addr; /* Address IRE represents. */
875 880 in6_addr_t ifrt_v6gateway_addr; /* Gateway if IRE_OFFLINK */
876 881 in6_addr_t ifrt_v6setsrc_addr; /* Src addr if RTF_SETSRC */
877 882 in6_addr_t ifrt_v6mask; /* Mask for matching IRE. */
878 883 uint32_t ifrt_flags; /* flags related to route */
879 884 iulp_t ifrt_metrics; /* Routing socket metrics */
880 885 zoneid_t ifrt_zoneid; /* zoneid for route */
881 886 } ifrt_t;
882 887
883 888 #define ifrt_addr V4_PART_OF_V6(ifrt_v6addr)
884 889 #define ifrt_gateway_addr V4_PART_OF_V6(ifrt_v6gateway_addr)
885 890 #define ifrt_mask V4_PART_OF_V6(ifrt_v6mask)
886 891 #define ifrt_setsrc_addr V4_PART_OF_V6(ifrt_v6setsrc_addr)
887 892
888 893 /* Number of IP addresses that can be hosted on a physical interface */
889 894 #define MAX_ADDRS_PER_IF 8192
890 895 /*
891 896 * Number of Source addresses to be considered for source address
892 897 * selection. Used by ipif_select_source_v4/v6.
893 898 */
894 899 #define MAX_IPIF_SELECT_SOURCE 50
895 900
896 901 #ifdef IP_DEBUG
897 902 /*
898 903 * Trace refholds and refreles for debugging.
899 904 */
900 905 #define TR_STACK_DEPTH 14
901 906 typedef struct tr_buf_s {
902 907 int tr_depth;
903 908 clock_t tr_time;
904 909 pc_t tr_stack[TR_STACK_DEPTH];
905 910 } tr_buf_t;
906 911
907 912 typedef struct th_trace_s {
908 913 int th_refcnt;
909 914 uint_t th_trace_lastref;
910 915 kthread_t *th_id;
911 916 #define TR_BUF_MAX 38
912 917 tr_buf_t th_trbuf[TR_BUF_MAX];
913 918 } th_trace_t;
914 919
915 920 typedef struct th_hash_s {
916 921 list_node_t thh_link;
917 922 mod_hash_t *thh_hash;
918 923 ip_stack_t *thh_ipst;
919 924 } th_hash_t;
920 925 #endif
921 926
922 927 /* The following are ipif_state_flags */
923 928 #define IPIF_CONDEMNED 0x1 /* The ipif is being removed */
924 929 #define IPIF_CHANGING 0x2 /* A critcal ipif field is changing */
925 930 #define IPIF_SET_LINKLOCAL 0x10 /* transient flag during bringup */
926 931
927 932 /* IP interface structure, one per local address */
928 933 typedef struct ipif_s {
929 934 struct ipif_s *ipif_next;
930 935 struct ill_s *ipif_ill; /* Back pointer to our ill */
931 936 int ipif_id; /* Logical unit number */
932 937 in6_addr_t ipif_v6lcl_addr; /* Local IP address for this if. */
933 938 in6_addr_t ipif_v6subnet; /* Subnet prefix for this if. */
934 939 in6_addr_t ipif_v6net_mask; /* Net mask for this interface. */
935 940 in6_addr_t ipif_v6brd_addr; /* Broadcast addr for this interface. */
936 941 in6_addr_t ipif_v6pp_dst_addr; /* Point-to-point dest address. */
937 942 uint64_t ipif_flags; /* Interface flags. */
938 943 uint_t ipif_ire_type; /* IRE_LOCAL or IRE_LOOPBACK */
939 944
940 945 /*
941 946 * The packet count in the ipif contain the sum of the
942 947 * packet counts in dead IRE_LOCAL/LOOPBACK for this ipif.
943 948 */
944 949 uint_t ipif_ib_pkt_count; /* Inbound packets for our dead IREs */
945 950
946 951 /* Exclusive bit fields, protected by ipsq_t */
947 952 unsigned int
948 953 ipif_was_up : 1, /* ipif was up before */
949 954 ipif_addr_ready : 1, /* DAD is done */
950 955 ipif_was_dup : 1, /* DAD had failed */
951 956 ipif_added_nce : 1, /* nce added for local address */
952 957
953 958 ipif_pad_to_31 : 28;
954 959
955 960 ilm_t *ipif_allhosts_ilm; /* For all-nodes join */
956 961 ilm_t *ipif_solmulti_ilm; /* For IPv6 solicited multicast join */
957 962
958 963 uint_t ipif_seqid; /* unique index across all ills */
959 964 uint_t ipif_state_flags; /* See IPIF_* flag defs above */
960 965 uint_t ipif_refcnt; /* active consistent reader cnt */
961 966
962 967 zoneid_t ipif_zoneid; /* zone ID number */
963 968 timeout_id_t ipif_recovery_id; /* Timer for DAD recovery */
964 969 boolean_t ipif_trace_disable; /* True when alloc fails */
965 970 /*
966 971 * For an IPMP interface, ipif_bound_ill tracks the ill whose hardware
967 972 * information this ipif is associated with via ARP/NDP. We can use
968 973 * an ill pointer (rather than an index) because only ills that are
969 974 * part of a group will be pointed to, and an ill cannot disappear
970 975 * while it's in a group.
971 976 */
972 977 struct ill_s *ipif_bound_ill;
973 978 struct ipif_s *ipif_bound_next; /* bound ipif chain */
974 979 boolean_t ipif_bound; /* B_TRUE if we successfully bound */
975 980
976 981 struct ire_s *ipif_ire_local; /* Our IRE_LOCAL or LOOPBACK */
977 982 struct ire_s *ipif_ire_if; /* Our IRE_INTERFACE */
978 983 } ipif_t;
979 984
980 985 /*
981 986 * The following table lists the protection levels of the various members
982 987 * of the ipif_t. The following notation is used.
983 988 *
984 989 * Write once - Written to only once at the time of bringing up
985 990 * the interface and can be safely read after the bringup without any lock.
986 991 *
987 992 * ipsq - Need to execute in the ipsq to perform the indicated access.
988 993 *
989 994 * ill_lock - Need to hold this mutex to perform the indicated access.
990 995 *
991 996 * ill_g_lock - Need to hold this rw lock as reader/writer for read access or
992 997 * write access respectively.
993 998 *
994 999 * down ill - Written to only when the ill is down (i.e all ipifs are down)
995 1000 * up ill - Read only when the ill is up (i.e. at least 1 ipif is up)
996 1001 *
997 1002 * Table of ipif_t members and their protection
998 1003 *
999 1004 * ipif_next ipsq + ill_lock + ipsq OR ill_lock OR
1000 1005 * ill_g_lock ill_g_lock
1001 1006 * ipif_ill ipsq + down ipif write once
1002 1007 * ipif_id ipsq + down ipif write once
1003 1008 * ipif_v6lcl_addr ipsq + down ipif up ipif
1004 1009 * ipif_v6subnet ipsq + down ipif up ipif
1005 1010 * ipif_v6net_mask ipsq + down ipif up ipif
1006 1011 *
1007 1012 * ipif_v6brd_addr
1008 1013 * ipif_v6pp_dst_addr
1009 1014 * ipif_flags ill_lock ill_lock
1010 1015 * ipif_ire_type ipsq + down ill up ill
1011 1016 *
1012 1017 * ipif_ib_pkt_count Approx
1013 1018 *
1014 1019 * bit fields ill_lock ill_lock
1015 1020 *
1016 1021 * ipif_allhosts_ilm ipsq ipsq
1017 1022 * ipif_solmulti_ilm ipsq ipsq
1018 1023 *
1019 1024 * ipif_seqid ipsq Write once
1020 1025 *
1021 1026 * ipif_state_flags ill_lock ill_lock
1022 1027 * ipif_refcnt ill_lock ill_lock
1023 1028 * ipif_bound_ill ipsq + ipmp_lock ipsq OR ipmp_lock
1024 1029 * ipif_bound_next ipsq ipsq
1025 1030 * ipif_bound ipsq ipsq
1026 1031 *
1027 1032 * ipif_ire_local ipsq + ips_ill_g_lock ipsq OR ips_ill_g_lock
1028 1033 * ipif_ire_if ipsq + ips_ill_g_lock ipsq OR ips_ill_g_lock
1029 1034 */
1030 1035
1031 1036 /*
1032 1037 * Return values from ip_laddr_verify_{v4,v6}
1033 1038 */
1034 1039 typedef enum { IPVL_UNICAST_UP, IPVL_UNICAST_DOWN, IPVL_MCAST, IPVL_BCAST,
1035 1040 IPVL_BAD} ip_laddr_t;
1036 1041
1037 1042
1038 1043 #define IP_TR_HASH(tid) ((((uintptr_t)tid) >> 6) & (IP_TR_HASH_MAX - 1))
1039 1044
1040 1045 #ifdef DEBUG
1041 1046 #define IPIF_TRACE_REF(ipif) ipif_trace_ref(ipif)
1042 1047 #define ILL_TRACE_REF(ill) ill_trace_ref(ill)
1043 1048 #define IPIF_UNTRACE_REF(ipif) ipif_untrace_ref(ipif)
1044 1049 #define ILL_UNTRACE_REF(ill) ill_untrace_ref(ill)
1045 1050 #else
1046 1051 #define IPIF_TRACE_REF(ipif)
1047 1052 #define ILL_TRACE_REF(ill)
1048 1053 #define IPIF_UNTRACE_REF(ipif)
1049 1054 #define ILL_UNTRACE_REF(ill)
1050 1055 #endif
1051 1056
1052 1057 /* IPv4 compatibility macros */
1053 1058 #define ipif_lcl_addr V4_PART_OF_V6(ipif_v6lcl_addr)
1054 1059 #define ipif_subnet V4_PART_OF_V6(ipif_v6subnet)
1055 1060 #define ipif_net_mask V4_PART_OF_V6(ipif_v6net_mask)
1056 1061 #define ipif_brd_addr V4_PART_OF_V6(ipif_v6brd_addr)
1057 1062 #define ipif_pp_dst_addr V4_PART_OF_V6(ipif_v6pp_dst_addr)
1058 1063
1059 1064 /* Macros for easy backreferences to the ill. */
1060 1065 #define ipif_isv6 ipif_ill->ill_isv6
1061 1066
1062 1067 #define SIOCLIFADDR_NDX 112 /* ndx of SIOCLIFADDR in the ndx ioctl table */
1063 1068
1064 1069 /*
1065 1070 * mode value for ip_ioctl_finish for finishing an ioctl
1066 1071 */
1067 1072 #define CONN_CLOSE 1 /* No mi_copy */
1068 1073 #define COPYOUT 2 /* do an mi_copyout if needed */
1069 1074 #define NO_COPYOUT 3 /* do an mi_copy_done */
1070 1075 #define IPI2MODE(ipi) ((ipi)->ipi_flags & IPI_GET_CMD ? COPYOUT : NO_COPYOUT)
1071 1076
1072 1077 /*
1073 1078 * The IP-MT design revolves around the serialization objects ipsq_t (IPSQ)
1074 1079 * and ipxop_t (exclusive operation or "xop"). Becoming "writer" on an IPSQ
1075 1080 * ensures that no other threads can become "writer" on any IPSQs sharing that
1076 1081 * IPSQ's xop until the writer thread is done.
1077 1082 *
1078 1083 * Each phyint points to one IPSQ that remains fixed over the phyint's life.
1079 1084 * Each IPSQ points to one xop that can change over the IPSQ's life. If a
1080 1085 * phyint is *not* in an IPMP group, then its IPSQ will refer to the IPSQ's
1081 1086 * "own" xop (ipsq_ownxop). If a phyint *is* part of an IPMP group, then its
1082 1087 * IPSQ will refer to the "group" xop, which is shorthand for the xop of the
1083 1088 * IPSQ of the IPMP meta-interface's phyint. Thus, all phyints that are part
1084 1089 * of the same IPMP group will have their IPSQ's point to the group xop, and
1085 1090 * thus becoming "writer" on any phyint in the group will prevent any other
1086 1091 * writer on any other phyint in the group. All IPSQs sharing the same xop
1087 1092 * are chained together through ipsq_next (in the degenerate common case,
1088 1093 * ipsq_next simply refers to itself). Note that the group xop is guaranteed
1089 1094 * to exist at least as long as there are members in the group, since the IPMP
1090 1095 * meta-interface can only be destroyed if the group is empty.
1091 1096 *
1092 1097 * Incoming exclusive operation requests are enqueued on the IPSQ they arrived
1093 1098 * on rather than the xop. This makes switching xop's (as would happen when a
1094 1099 * phyint leaves an IPMP group) simple, because after the phyint leaves the
1095 1100 * group, any operations enqueued on its IPSQ can be safely processed with
1096 1101 * respect to its new xop, and any operations enqueued on the IPSQs of its
1097 1102 * former group can be processed with respect to their existing group xop.
1098 1103 * Even so, switching xops is a subtle dance; see ipsq_dq() for details.
1099 1104 *
1100 1105 * An IPSQ's "own" xop is embedded within the IPSQ itself since they have have
1101 1106 * identical lifetimes, and because doing so simplifies pointer management.
1102 1107 * While each phyint and IPSQ point to each other, it is not possible to free
1103 1108 * the IPSQ when the phyint is freed, since we may still *inside* the IPSQ
1104 1109 * when the phyint is being freed. Thus, ipsq_phyint is set to NULL when the
1105 1110 * phyint is freed, and the IPSQ free is later done in ipsq_exit().
1106 1111 *
1107 1112 * ipsq_t synchronization: read write
1108 1113 *
1109 1114 * ipsq_xopq_mphead ipx_lock ipx_lock
1110 1115 * ipsq_xopq_mptail ipx_lock ipx_lock
1111 1116 * ipsq_xop_switch_mp ipsq_lock ipsq_lock
1112 1117 * ipsq_phyint write once write once
1113 1118 * ipsq_next RW_READER ill_g_lock RW_WRITER ill_g_lock
1114 1119 * ipsq_xop ipsq_lock or ipsq ipsq_lock + ipsq
1115 1120 * ipsq_swxop ipsq ipsq
1116 1121 * ipsq_ownxop see ipxop_t see ipxop_t
1117 1122 * ipsq_ipst write once write once
1118 1123 *
1119 1124 * ipxop_t synchronization: read write
1120 1125 *
1121 1126 * ipx_writer ipx_lock ipx_lock
1122 1127 * ipx_xop_queued ipx_lock ipx_lock
1123 1128 * ipx_mphead ipx_lock ipx_lock
1124 1129 * ipx_mptail ipx_lock ipx_lock
1125 1130 * ipx_ipsq write once write once
1126 1131 * ips_ipsq_queued ipx_lock ipx_lock
1127 1132 * ipx_waitfor ipsq or ipx_lock ipsq + ipx_lock
1128 1133 * ipx_reentry_cnt ipsq or ipx_lock ipsq + ipx_lock
1129 1134 * ipx_current_done ipsq ipsq
1130 1135 * ipx_current_ioctl ipsq ipsq
1131 1136 * ipx_current_ipif ipsq or ipx_lock ipsq + ipx_lock
1132 1137 * ipx_pending_ipif ipsq or ipx_lock ipsq + ipx_lock
1133 1138 * ipx_pending_mp ipsq or ipx_lock ipsq + ipx_lock
1134 1139 * ipx_forced ipsq ipsq
1135 1140 * ipx_depth ipsq ipsq
1136 1141 * ipx_stack ipsq ipsq
1137 1142 */
1138 1143 typedef struct ipxop_s {
1139 1144 kmutex_t ipx_lock; /* see above */
1140 1145 kthread_t *ipx_writer; /* current owner */
1141 1146 mblk_t *ipx_mphead; /* messages tied to this op */
1142 1147 mblk_t *ipx_mptail;
1143 1148 struct ipsq_s *ipx_ipsq; /* associated ipsq */
1144 1149 boolean_t ipx_ipsq_queued; /* ipsq using xop has queued op */
1145 1150 int ipx_waitfor; /* waiting; values encoded below */
1146 1151 int ipx_reentry_cnt;
1147 1152 boolean_t ipx_current_done; /* is the current operation done? */
1148 1153 int ipx_current_ioctl; /* current ioctl, or 0 if no ioctl */
1149 1154 ipif_t *ipx_current_ipif; /* ipif for current op */
1150 1155 ipif_t *ipx_pending_ipif; /* ipif for ipx_pending_mp */
1151 1156 mblk_t *ipx_pending_mp; /* current ioctl mp while waiting */
1152 1157 boolean_t ipx_forced; /* debugging aid */
1153 1158 #ifdef DEBUG
1154 1159 int ipx_depth; /* debugging aid */
1155 1160 #define IPX_STACK_DEPTH 15
1156 1161 pc_t ipx_stack[IPX_STACK_DEPTH]; /* debugging aid */
1157 1162 #endif
1158 1163 } ipxop_t;
1159 1164
1160 1165 typedef struct ipsq_s {
1161 1166 kmutex_t ipsq_lock; /* see above */
1162 1167 mblk_t *ipsq_switch_mp; /* op to handle right after switch */
1163 1168 mblk_t *ipsq_xopq_mphead; /* list of excl ops (mostly ioctls) */
1164 1169 mblk_t *ipsq_xopq_mptail;
1165 1170 struct phyint *ipsq_phyint; /* associated phyint */
1166 1171 struct ipsq_s *ipsq_next; /* next ipsq sharing ipsq_xop */
1167 1172 struct ipxop_s *ipsq_xop; /* current xop synchronization info */
1168 1173 struct ipxop_s *ipsq_swxop; /* switch xop to on ipsq_exit() */
1169 1174 struct ipxop_s ipsq_ownxop; /* our own xop (may not be in-use) */
1170 1175 ip_stack_t *ipsq_ipst; /* does not have a netstack_hold */
1171 1176 } ipsq_t;
1172 1177
1173 1178 /*
1174 1179 * ipx_waitfor values:
1175 1180 */
1176 1181 enum {
1177 1182 IPIF_DOWN = 1, /* ipif_down() waiting for refcnts to drop */
1178 1183 ILL_DOWN, /* ill_down() waiting for refcnts to drop */
1179 1184 IPIF_FREE, /* ipif_free() waiting for refcnts to drop */
1180 1185 ILL_FREE /* ill unplumb waiting for refcnts to drop */
1181 1186 };
1182 1187
1183 1188 /* Operation types for ipsq_try_enter() */
1184 1189 #define CUR_OP 0 /* request writer within current operation */
1185 1190 #define NEW_OP 1 /* request writer for a new operation */
1186 1191 #define SWITCH_OP 2 /* request writer once IPSQ XOP switches */
1187 1192
1188 1193 /*
1189 1194 * Kstats tracked on each IPMP meta-interface. Order here must match
1190 1195 * ipmp_kstats[] in ip/ipmp.c.
1191 1196 */
1192 1197 enum {
1193 1198 IPMP_KSTAT_OBYTES, IPMP_KSTAT_OBYTES64, IPMP_KSTAT_RBYTES,
1194 1199 IPMP_KSTAT_RBYTES64, IPMP_KSTAT_OPACKETS, IPMP_KSTAT_OPACKETS64,
1195 1200 IPMP_KSTAT_OERRORS, IPMP_KSTAT_IPACKETS, IPMP_KSTAT_IPACKETS64,
1196 1201 IPMP_KSTAT_IERRORS, IPMP_KSTAT_MULTIRCV, IPMP_KSTAT_MULTIXMT,
1197 1202 IPMP_KSTAT_BRDCSTRCV, IPMP_KSTAT_BRDCSTXMT, IPMP_KSTAT_LINK_UP,
1198 1203 IPMP_KSTAT_MAX /* keep last */
1199 1204 };
1200 1205
1201 1206 /*
1202 1207 * phyint represents state that is common to both IPv4 and IPv6 interfaces.
1203 1208 * There is a separate ill_t representing IPv4 and IPv6 which has a
1204 1209 * backpointer to the phyint structure for accessing common state.
1205 1210 */
1206 1211 typedef struct phyint {
1207 1212 struct ill_s *phyint_illv4;
1208 1213 struct ill_s *phyint_illv6;
1209 1214 uint_t phyint_ifindex; /* SIOCSLIFINDEX */
1210 1215 uint64_t phyint_flags;
1211 1216 avl_node_t phyint_avl_by_index; /* avl tree by index */
1212 1217 avl_node_t phyint_avl_by_name; /* avl tree by name */
1213 1218 kmutex_t phyint_lock;
1214 1219 struct ipsq_s *phyint_ipsq; /* back pointer to ipsq */
1215 1220 struct ipmp_grp_s *phyint_grp; /* associated IPMP group */
1216 1221 char phyint_name[LIFNAMSIZ]; /* physical interface name */
1217 1222 uint64_t phyint_kstats0[IPMP_KSTAT_MAX]; /* baseline kstats */
1218 1223 } phyint_t;
1219 1224
1220 1225 #define CACHE_ALIGN_SIZE 64
1221 1226 #define CACHE_ALIGN(align_struct) P2ROUNDUP(sizeof (struct align_struct),\
1222 1227 CACHE_ALIGN_SIZE)
1223 1228 struct _phyint_list_s_ {
1224 1229 avl_tree_t phyint_list_avl_by_index; /* avl tree by index */
1225 1230 avl_tree_t phyint_list_avl_by_name; /* avl tree by name */
1226 1231 };
1227 1232
1228 1233 typedef union phyint_list_u {
1229 1234 struct _phyint_list_s_ phyint_list_s;
1230 1235 char phyint_list_filler[CACHE_ALIGN(_phyint_list_s_)];
1231 1236 } phyint_list_t;
1232 1237
1233 1238 #define phyint_list_avl_by_index phyint_list_s.phyint_list_avl_by_index
1234 1239 #define phyint_list_avl_by_name phyint_list_s.phyint_list_avl_by_name
1235 1240
1236 1241 /*
1237 1242 * Fragmentation hash bucket
1238 1243 */
1239 1244 typedef struct ipfb_s {
1240 1245 struct ipf_s *ipfb_ipf; /* List of ... */
1241 1246 size_t ipfb_count; /* Count of bytes used by frag(s) */
1242 1247 kmutex_t ipfb_lock; /* Protect all ipf in list */
1243 1248 uint_t ipfb_frag_pkts; /* num of distinct fragmented pkts */
1244 1249 } ipfb_t;
1245 1250
1246 1251 /*
1247 1252 * IRE bucket structure. Usually there is an array of such structures,
1248 1253 * each pointing to a linked list of ires. irb_refcnt counts the number
1249 1254 * of walkers of a given hash bucket. Usually the reference count is
1250 1255 * bumped up if the walker wants no IRES to be DELETED while walking the
1251 1256 * list. Bumping up does not PREVENT ADDITION. This allows walking a given
1252 1257 * hash bucket without stumbling up on a free pointer.
1253 1258 *
1254 1259 * irb_t structures in ip_ftable are dynamically allocated and freed.
1255 1260 * In order to identify the irb_t structures that can be safely kmem_free'd
1256 1261 * we need to ensure that
1257 1262 * - the irb_refcnt is quiescent, indicating no other walkers,
1258 1263 * - no other threads or ire's are holding references to the irb,
1259 1264 * i.e., irb_nire == 0,
1260 1265 * - there are no active ire's in the bucket, i.e., irb_ire_cnt == 0
1261 1266 */
1262 1267 typedef struct irb {
1263 1268 struct ire_s *irb_ire; /* First ire in this bucket */
1264 1269 /* Should be first in this struct */
1265 1270 krwlock_t irb_lock; /* Protect this bucket */
1266 1271 uint_t irb_refcnt; /* Protected by irb_lock */
1267 1272 uchar_t irb_marks; /* CONDEMNED ires in this bucket ? */
1268 1273 #define IRB_MARK_CONDEMNED 0x0001 /* Contains some IRE_IS_CONDEMNED */
1269 1274 #define IRB_MARK_DYNAMIC 0x0002 /* Dynamically allocated */
1270 1275 /* Once IPv6 uses radix then IRB_MARK_DYNAMIC will be always be set */
1271 1276 uint_t irb_ire_cnt; /* Num of active IRE in this bucket */
1272 1277 int irb_nire; /* Num of ftable ire's that ref irb */
1273 1278 ip_stack_t *irb_ipst; /* Does not have a netstack_hold */
1274 1279 } irb_t;
1275 1280
1276 1281 /*
1277 1282 * This is the structure used to store the multicast physical addresses
1278 1283 * that an interface has joined.
1279 1284 * The refcnt keeps track of the number of multicast IP addresses mapping
1280 1285 * to a physical multicast address.
1281 1286 */
1282 1287 typedef struct multiphysaddr_s {
1283 1288 struct multiphysaddr_s *mpa_next;
1284 1289 char mpa_addr[IP_MAX_HW_LEN];
1285 1290 int mpa_refcnt;
1286 1291 } multiphysaddr_t;
1287 1292
1288 1293 #define IRB2RT(irb) (rt_t *)((caddr_t)(irb) - offsetof(rt_t, rt_irb))
1289 1294
1290 1295 /* Forward declarations */
1291 1296 struct dce_s;
1292 1297 typedef struct dce_s dce_t;
1293 1298 struct ire_s;
1294 1299 typedef struct ire_s ire_t;
1295 1300 struct ncec_s;
1296 1301 typedef struct ncec_s ncec_t;
1297 1302 struct nce_s;
1298 1303 typedef struct nce_s nce_t;
1299 1304 struct ip_recv_attr_s;
1300 1305 typedef struct ip_recv_attr_s ip_recv_attr_t;
1301 1306 struct ip_xmit_attr_s;
1302 1307 typedef struct ip_xmit_attr_s ip_xmit_attr_t;
1303 1308
1304 1309 struct tsol_ire_gw_secattr_s;
1305 1310 typedef struct tsol_ire_gw_secattr_s tsol_ire_gw_secattr_t;
1306 1311
1307 1312 /*
1308 1313 * This is a structure for a one-element route cache that is passed
1309 1314 * by reference between ip_input and ill_inputfn.
1310 1315 */
1311 1316 typedef struct {
1312 1317 ire_t *rtc_ire;
1313 1318 ipaddr_t rtc_ipaddr;
1314 1319 in6_addr_t rtc_ip6addr;
1315 1320 } rtc_t;
1316 1321
1317 1322 /*
1318 1323 * Note: Temporarily use 64 bits, and will probably go back to 32 bits after
1319 1324 * more cleanup work is done.
1320 1325 */
1321 1326 typedef uint64_t iaflags_t;
1322 1327
1323 1328 /* The ill input function pointer type */
1324 1329 typedef void (*pfillinput_t)(mblk_t *, void *, void *, ip_recv_attr_t *,
1325 1330 rtc_t *);
1326 1331
1327 1332 /* The ire receive function pointer type */
1328 1333 typedef void (*pfirerecv_t)(ire_t *, mblk_t *, void *, ip_recv_attr_t *);
1329 1334
1330 1335 /* The ire send and postfrag function pointer types */
1331 1336 typedef int (*pfiresend_t)(ire_t *, mblk_t *, void *,
1332 1337 ip_xmit_attr_t *, uint32_t *);
1333 1338 typedef int (*pfirepostfrag_t)(mblk_t *, nce_t *, iaflags_t, uint_t, uint32_t,
1334 1339 zoneid_t, zoneid_t, uintptr_t *);
1335 1340
1336 1341
1337 1342 #define IP_V4_G_HEAD 0
1338 1343 #define IP_V6_G_HEAD 1
1339 1344
1340 1345 #define MAX_G_HEADS 2
1341 1346
1342 1347 /*
1343 1348 * unpadded ill_if structure
1344 1349 */
1345 1350 struct _ill_if_s_ {
1346 1351 union ill_if_u *illif_next;
1347 1352 union ill_if_u *illif_prev;
1348 1353 avl_tree_t illif_avl_by_ppa; /* AVL tree sorted on ppa */
1349 1354 vmem_t *illif_ppa_arena; /* ppa index space */
1350 1355 uint16_t illif_mcast_v1; /* hints for */
1351 1356 uint16_t illif_mcast_v2; /* [igmp|mld]_slowtimo */
1352 1357 int illif_name_len; /* name length */
1353 1358 char illif_name[LIFNAMSIZ]; /* name of interface type */
1354 1359 };
1355 1360
1356 1361 /* cache aligned ill_if structure */
1357 1362 typedef union ill_if_u {
1358 1363 struct _ill_if_s_ ill_if_s;
1359 1364 char illif_filler[CACHE_ALIGN(_ill_if_s_)];
1360 1365 } ill_if_t;
1361 1366
1362 1367 #define illif_next ill_if_s.illif_next
1363 1368 #define illif_prev ill_if_s.illif_prev
1364 1369 #define illif_avl_by_ppa ill_if_s.illif_avl_by_ppa
1365 1370 #define illif_ppa_arena ill_if_s.illif_ppa_arena
1366 1371 #define illif_mcast_v1 ill_if_s.illif_mcast_v1
1367 1372 #define illif_mcast_v2 ill_if_s.illif_mcast_v2
1368 1373 #define illif_name ill_if_s.illif_name
1369 1374 #define illif_name_len ill_if_s.illif_name_len
1370 1375
1371 1376 typedef struct ill_walk_context_s {
1372 1377 int ctx_current_list; /* current list being searched */
1373 1378 int ctx_last_list; /* last list to search */
1374 1379 } ill_walk_context_t;
1375 1380
1376 1381 /*
1377 1382 * ill_g_heads structure, one for IPV4 and one for IPV6
1378 1383 */
1379 1384 struct _ill_g_head_s_ {
1380 1385 ill_if_t *ill_g_list_head;
1381 1386 ill_if_t *ill_g_list_tail;
1382 1387 };
1383 1388
1384 1389 typedef union ill_g_head_u {
1385 1390 struct _ill_g_head_s_ ill_g_head_s;
1386 1391 char ill_g_head_filler[CACHE_ALIGN(_ill_g_head_s_)];
1387 1392 } ill_g_head_t;
1388 1393
1389 1394 #define ill_g_list_head ill_g_head_s.ill_g_list_head
1390 1395 #define ill_g_list_tail ill_g_head_s.ill_g_list_tail
1391 1396
1392 1397 #define IP_V4_ILL_G_LIST(ipst) \
1393 1398 (ipst)->ips_ill_g_heads[IP_V4_G_HEAD].ill_g_list_head
1394 1399 #define IP_V6_ILL_G_LIST(ipst) \
1395 1400 (ipst)->ips_ill_g_heads[IP_V6_G_HEAD].ill_g_list_head
1396 1401 #define IP_VX_ILL_G_LIST(i, ipst) \
1397 1402 (ipst)->ips_ill_g_heads[i].ill_g_list_head
1398 1403
1399 1404 #define ILL_START_WALK_V4(ctx_ptr, ipst) \
1400 1405 ill_first(IP_V4_G_HEAD, IP_V4_G_HEAD, ctx_ptr, ipst)
1401 1406 #define ILL_START_WALK_V6(ctx_ptr, ipst) \
1402 1407 ill_first(IP_V6_G_HEAD, IP_V6_G_HEAD, ctx_ptr, ipst)
1403 1408 #define ILL_START_WALK_ALL(ctx_ptr, ipst) \
1404 1409 ill_first(MAX_G_HEADS, MAX_G_HEADS, ctx_ptr, ipst)
1405 1410
1406 1411 /*
1407 1412 * Capabilities, possible flags for ill_capabilities.
1408 1413 */
1409 1414 #define ILL_CAPAB_LSO 0x04 /* Large Send Offload */
1410 1415 #define ILL_CAPAB_HCKSUM 0x08 /* Hardware checksumming */
1411 1416 #define ILL_CAPAB_ZEROCOPY 0x10 /* Zero-copy */
1412 1417 #define ILL_CAPAB_DLD 0x20 /* DLD capabilities */
1413 1418 #define ILL_CAPAB_DLD_POLL 0x40 /* Polling */
1414 1419 #define ILL_CAPAB_DLD_DIRECT 0x80 /* Direct function call */
1415 1420
1416 1421 /*
1417 1422 * Per-ill Hardware Checksumming capbilities.
1418 1423 */
1419 1424 typedef struct ill_hcksum_capab_s ill_hcksum_capab_t;
1420 1425
1421 1426 /*
1422 1427 * Per-ill Zero-copy capabilities.
1423 1428 */
1424 1429 typedef struct ill_zerocopy_capab_s ill_zerocopy_capab_t;
1425 1430
1426 1431 /*
1427 1432 * DLD capbilities.
1428 1433 */
1429 1434 typedef struct ill_dld_capab_s ill_dld_capab_t;
1430 1435
1431 1436 /*
1432 1437 * Per-ill polling resource map.
1433 1438 */
1434 1439 typedef struct ill_rx_ring ill_rx_ring_t;
1435 1440
1436 1441 /*
1437 1442 * Per-ill Large Send Offload capabilities.
1438 1443 */
1439 1444 typedef struct ill_lso_capab_s ill_lso_capab_t;
1440 1445
1441 1446 /* The following are ill_state_flags */
1442 1447 #define ILL_LL_SUBNET_PENDING 0x01 /* Waiting for DL_INFO_ACK from drv */
1443 1448 #define ILL_CONDEMNED 0x02 /* No more new ref's to the ILL */
1444 1449 #define ILL_DL_UNBIND_IN_PROGRESS 0x04 /* UNBIND_REQ is sent */
1445 1450 /*
1446 1451 * ILL_DOWN_IN_PROGRESS is set to ensure the following:
1447 1452 * - no packets are sent to the driver after the DL_UNBIND_REQ is sent,
1448 1453 * - no longstanding references will be acquired on objects that are being
1449 1454 * brought down.
1450 1455 */
1451 1456 #define ILL_DOWN_IN_PROGRESS 0x08
1452 1457
1453 1458 /* Is this an ILL whose source address is used by other ILL's ? */
1454 1459 #define IS_USESRC_ILL(ill) \
1455 1460 (((ill)->ill_usesrc_ifindex == 0) && \
1456 1461 ((ill)->ill_usesrc_grp_next != NULL))
1457 1462
1458 1463 /* Is this a client/consumer of the usesrc ILL ? */
1459 1464 #define IS_USESRC_CLI_ILL(ill) \
1460 1465 (((ill)->ill_usesrc_ifindex != 0) && \
1461 1466 ((ill)->ill_usesrc_grp_next != NULL))
1462 1467
1463 1468 /* Is this an virtual network interface (vni) ILL ? */
1464 1469 #define IS_VNI(ill) \
1465 1470 (((ill)->ill_phyint->phyint_flags & (PHYI_LOOPBACK|PHYI_VIRTUAL)) == \
1466 1471 PHYI_VIRTUAL)
1467 1472
1468 1473 /* Is this a loopback ILL? */
1469 1474 #define IS_LOOPBACK(ill) \
1470 1475 ((ill)->ill_phyint->phyint_flags & PHYI_LOOPBACK)
1471 1476
1472 1477 /* Is this an IPMP meta-interface ILL? */
1473 1478 #define IS_IPMP(ill) \
1474 1479 ((ill)->ill_phyint->phyint_flags & PHYI_IPMP)
1475 1480
1476 1481 /* Is this ILL under an IPMP meta-interface? (aka "in a group?") */
1477 1482 #define IS_UNDER_IPMP(ill) \
1478 1483 ((ill)->ill_grp != NULL && !IS_IPMP(ill))
1479 1484
1480 1485 /* Is ill1 in the same illgrp as ill2? */
1481 1486 #define IS_IN_SAME_ILLGRP(ill1, ill2) \
1482 1487 ((ill1)->ill_grp != NULL && ((ill1)->ill_grp == (ill2)->ill_grp))
1483 1488
1484 1489 /* Is ill1 on the same LAN as ill2? */
1485 1490 #define IS_ON_SAME_LAN(ill1, ill2) \
1486 1491 ((ill1) == (ill2) || IS_IN_SAME_ILLGRP(ill1, ill2))
1487 1492
1488 1493 #define ILL_OTHER(ill) \
1489 1494 ((ill)->ill_isv6 ? (ill)->ill_phyint->phyint_illv4 : \
1490 1495 (ill)->ill_phyint->phyint_illv6)
1491 1496
1492 1497 /*
1493 1498 * IPMP group ILL state structure -- up to two per IPMP group (V4 and V6).
1494 1499 * Created when the V4 and/or V6 IPMP meta-interface is I_PLINK'd. It is
1495 1500 * guaranteed to persist while there are interfaces of that type in the group.
1496 1501 * In general, most fields are accessed outside of the IPSQ (e.g., in the
1497 1502 * datapath), and thus use locks in addition to the IPSQ for protection.
1498 1503 *
1499 1504 * synchronization: read write
1500 1505 *
1501 1506 * ig_if ipsq or ill_g_lock ipsq and ill_g_lock
1502 1507 * ig_actif ipsq or ipmp_lock ipsq and ipmp_lock
1503 1508 * ig_nactif ipsq or ipmp_lock ipsq and ipmp_lock
1504 1509 * ig_next_ill ipsq or ipmp_lock ipsq and ipmp_lock
1505 1510 * ig_ipmp_ill write once write once
1506 1511 * ig_cast_ill ipsq or ipmp_lock ipsq and ipmp_lock
1507 1512 * ig_arpent ipsq ipsq
1508 1513 * ig_mtu ipsq ipsq
1509 1514 * ig_mc_mtu ipsq ipsq
1510 1515 */
1511 1516 typedef struct ipmp_illgrp_s {
1512 1517 list_t ig_if; /* list of all interfaces */
1513 1518 list_t ig_actif; /* list of active interfaces */
1514 1519 uint_t ig_nactif; /* number of active interfaces */
1515 1520 struct ill_s *ig_next_ill; /* next active interface to use */
1516 1521 struct ill_s *ig_ipmp_ill; /* backpointer to IPMP meta-interface */
1517 1522 struct ill_s *ig_cast_ill; /* nominated ill for multi/broadcast */
1518 1523 list_t ig_arpent; /* list of ARP entries */
1519 1524 uint_t ig_mtu; /* ig_ipmp_ill->ill_mtu */
1520 1525 uint_t ig_mc_mtu; /* ig_ipmp_ill->ill_mc_mtu */
1521 1526 } ipmp_illgrp_t;
1522 1527
1523 1528 /*
1524 1529 * IPMP group state structure -- one per IPMP group. Created when the
1525 1530 * IPMP meta-interface is plumbed; it is guaranteed to persist while there
1526 1531 * are interfaces in it.
1527 1532 *
1528 1533 * ipmp_grp_t synchronization: read write
1529 1534 *
1530 1535 * gr_name ipmp_lock ipmp_lock
1531 1536 * gr_ifname write once write once
1532 1537 * gr_mactype ipmp_lock ipmp_lock
1533 1538 * gr_phyint write once write once
1534 1539 * gr_nif ipmp_lock ipmp_lock
1535 1540 * gr_nactif ipsq ipsq
1536 1541 * gr_v4 ipmp_lock ipmp_lock
1537 1542 * gr_v6 ipmp_lock ipmp_lock
1538 1543 * gr_nv4 ipmp_lock ipmp_lock
1539 1544 * gr_nv6 ipmp_lock ipmp_lock
1540 1545 * gr_pendv4 ipmp_lock ipmp_lock
1541 1546 * gr_pendv6 ipmp_lock ipmp_lock
1542 1547 * gr_linkdownmp ipsq ipsq
1543 1548 * gr_ksp ipmp_lock ipmp_lock
1544 1549 * gr_kstats0 atomic atomic
1545 1550 */
1546 1551 typedef struct ipmp_grp_s {
1547 1552 char gr_name[LIFGRNAMSIZ]; /* group name */
1548 1553 char gr_ifname[LIFNAMSIZ]; /* interface name */
1549 1554 t_uscalar_t gr_mactype; /* DLPI mactype of group */
1550 1555 phyint_t *gr_phyint; /* IPMP group phyint */
1551 1556 uint_t gr_nif; /* number of interfaces in group */
1552 1557 uint_t gr_nactif; /* number of active interfaces */
1553 1558 ipmp_illgrp_t *gr_v4; /* V4 group information */
1554 1559 ipmp_illgrp_t *gr_v6; /* V6 group information */
1555 1560 uint_t gr_nv4; /* number of ills in V4 group */
1556 1561 uint_t gr_nv6; /* number of ills in V6 group */
1557 1562 uint_t gr_pendv4; /* number of pending ills in V4 group */
1558 1563 uint_t gr_pendv6; /* number of pending ills in V6 group */
1559 1564 mblk_t *gr_linkdownmp; /* message used to bring link down */
1560 1565 kstat_t *gr_ksp; /* group kstat pointer */
1561 1566 uint64_t gr_kstats0[IPMP_KSTAT_MAX]; /* baseline group kstats */
1562 1567 } ipmp_grp_t;
1563 1568
1564 1569 /*
1565 1570 * IPMP ARP entry -- one per SIOCS*ARP entry tied to the group. Used to keep
1566 1571 * ARP up-to-date as the active set of interfaces in the group changes.
1567 1572 */
1568 1573 typedef struct ipmp_arpent_s {
1569 1574 ipaddr_t ia_ipaddr; /* IP address for this entry */
1570 1575 boolean_t ia_proxyarp; /* proxy ARP entry? */
1571 1576 boolean_t ia_notified; /* ARP notified about this entry? */
1572 1577 list_node_t ia_node; /* next ARP entry in list */
1573 1578 uint16_t ia_flags; /* nce_flags for the address */
1574 1579 size_t ia_lladdr_len;
1575 1580 uchar_t *ia_lladdr;
1576 1581 } ipmp_arpent_t;
1577 1582
1578 1583 struct arl_s;
1579 1584
1580 1585 /*
1581 1586 * Per-ill capabilities.
1582 1587 */
1583 1588 struct ill_hcksum_capab_s {
1584 1589 uint_t ill_hcksum_version; /* interface version */
1585 1590 uint_t ill_hcksum_txflags; /* capabilities on transmit */
1586 1591 };
1587 1592
1588 1593 struct ill_zerocopy_capab_s {
1589 1594 uint_t ill_zerocopy_version; /* interface version */
1590 1595 uint_t ill_zerocopy_flags; /* capabilities */
1591 1596 };
1592 1597
1593 1598 struct ill_lso_capab_s {
1594 1599 uint_t ill_lso_flags; /* capabilities */
1595 1600 uint_t ill_lso_max; /* maximum size of payload */
1596 1601 };
1597 1602
1598 1603 /*
1599 1604 * IP Lower level Structure.
1600 1605 * Instance data structure in ip_open when there is a device below us.
1601 1606 */
1602 1607 typedef struct ill_s {
1603 1608 pfillinput_t ill_inputfn; /* Fast input function selector */
1604 1609 ill_if_t *ill_ifptr; /* pointer to interface type */
1605 1610 queue_t *ill_rq; /* Read queue. */
1606 1611 queue_t *ill_wq; /* Write queue. */
1607 1612
1608 1613 int ill_error; /* Error value sent up by device. */
1609 1614
1610 1615 ipif_t *ill_ipif; /* Interface chain for this ILL. */
1611 1616
1612 1617 uint_t ill_ipif_up_count; /* Number of IPIFs currently up. */
1613 1618 uint_t ill_max_frag; /* Max IDU from DLPI. */
1614 1619 uint_t ill_current_frag; /* Current IDU from DLPI. */
1615 1620 uint_t ill_mtu; /* User-specified MTU; SIOCSLIFMTU */
1616 1621 uint_t ill_mc_mtu; /* MTU for multi/broadcast */
1617 1622 uint_t ill_metric; /* BSD if metric, for compatibility. */
1618 1623 char *ill_name; /* Our name. */
1619 1624 uint_t ill_ipif_dup_count; /* Number of duplicate addresses. */
1620 1625 uint_t ill_name_length; /* Name length, incl. terminator. */
1621 1626 uint_t ill_net_type; /* IRE_IF_RESOLVER/IRE_IF_NORESOLVER. */
1622 1627 /*
1623 1628 * Physical Point of Attachment num. If DLPI style 1 provider
1624 1629 * then this is derived from the devname.
1625 1630 */
1626 1631 uint_t ill_ppa;
1627 1632 t_uscalar_t ill_sap;
1628 1633 t_scalar_t ill_sap_length; /* Including sign (for position) */
1629 1634 uint_t ill_phys_addr_length; /* Excluding the sap. */
1630 1635 uint_t ill_bcast_addr_length; /* Only set when the DL provider */
1631 1636 /* supports broadcast. */
1632 1637 t_uscalar_t ill_mactype;
1633 1638 uint8_t *ill_frag_ptr; /* Reassembly state. */
1634 1639 timeout_id_t ill_frag_timer_id; /* timeout id for the frag timer */
1635 1640 ipfb_t *ill_frag_hash_tbl; /* Fragment hash list head. */
1636 1641
1637 1642 krwlock_t ill_mcast_lock; /* Protects multicast state */
1638 1643 kmutex_t ill_mcast_serializer; /* Serialize across ilg and ilm state */
1639 1644 ilm_t *ill_ilm; /* Multicast membership for ill */
1640 1645 uint_t ill_global_timer; /* for IGMPv3/MLDv2 general queries */
1641 1646 int ill_mcast_type; /* type of router which is querier */
1642 1647 /* on this interface */
1643 1648 uint16_t ill_mcast_v1_time; /* # slow timeouts since last v1 qry */
1644 1649 uint16_t ill_mcast_v2_time; /* # slow timeouts since last v2 qry */
1645 1650 uint8_t ill_mcast_v1_tset; /* 1 => timer is set; 0 => not set */
1646 1651 uint8_t ill_mcast_v2_tset; /* 1 => timer is set; 0 => not set */
1647 1652
1648 1653 uint8_t ill_mcast_rv; /* IGMPv3/MLDv2 robustness variable */
1649 1654 int ill_mcast_qi; /* IGMPv3/MLDv2 query interval var */
1650 1655
1651 1656 /*
1652 1657 * All non-NULL cells between 'ill_first_mp_to_free' and
1653 1658 * 'ill_last_mp_to_free' are freed in ill_delete.
1654 1659 */
1655 1660 #define ill_first_mp_to_free ill_bcast_mp
1656 1661 mblk_t *ill_bcast_mp; /* DLPI header for broadcasts. */
1657 1662 mblk_t *ill_unbind_mp; /* unbind mp from ill_dl_up() */
1658 1663 mblk_t *ill_promiscoff_mp; /* for ill_leave_allmulti() */
1659 1664 mblk_t *ill_dlpi_deferred; /* b_next chain of control messages */
1660 1665 mblk_t *ill_dest_addr_mp; /* mblk which holds ill_dest_addr */
1661 1666 mblk_t *ill_replumb_mp; /* replumb mp from ill_replumb() */
1662 1667 mblk_t *ill_phys_addr_mp; /* mblk which holds ill_phys_addr */
1663 1668 mblk_t *ill_mcast_deferred; /* b_next chain of IGMP/MLD packets */
1664 1669 #define ill_last_mp_to_free ill_mcast_deferred
1665 1670
1666 1671 cred_t *ill_credp; /* opener's credentials */
1667 1672 uint8_t *ill_phys_addr; /* ill_phys_addr_mp->b_rptr + off */
1668 1673 uint8_t *ill_dest_addr; /* ill_dest_addr_mp->b_rptr + off */
1669 1674
1670 1675 uint_t ill_state_flags; /* see ILL_* flags above */
1671 1676
1672 1677 /* Following bit fields protected by ipsq_t */
1673 1678 uint_t
1674 1679 ill_needs_attach : 1,
1675 1680 ill_reserved : 1,
1676 1681 ill_isv6 : 1,
1677 1682 ill_dlpi_style_set : 1,
1678 1683
1679 1684 ill_ifname_pending : 1,
1680 1685 ill_logical_down : 1,
1681 1686 ill_dl_up : 1,
1682 1687 ill_up_ipifs : 1,
1683 1688
1684 1689 ill_note_link : 1, /* supports link-up notification */
1685 1690 ill_capab_reneg : 1, /* capability renegotiation to be done */
1686 1691 ill_dld_capab_inprog : 1, /* direct dld capab call in prog */
1687 1692 ill_need_recover_multicast : 1,
1688 1693
1689 1694 ill_replumbing : 1,
1690 1695 ill_arl_dlpi_pending : 1,
1691 1696 ill_grp_pending : 1,
1692 1697
1693 1698 ill_pad_to_bit_31 : 17;
1694 1699
1695 1700 /* Following bit fields protected by ill_lock */
1696 1701 uint_t
1697 1702 ill_fragtimer_executing : 1,
1698 1703 ill_fragtimer_needrestart : 1,
1699 1704 ill_manual_token : 1, /* system won't override ill_token */
1700 1705 /*
1701 1706 * ill_manual_linklocal : system will not change the
1702 1707 * linklocal whenever ill_token changes.
1703 1708 */
1704 1709 ill_manual_linklocal : 1,
1705 1710
1706 1711 ill_manual_dst_linklocal : 1, /* same for pt-pt dst linklocal */
1707 1712
1708 1713 ill_pad_bit_31 : 27;
1709 1714
1710 1715 /*
1711 1716 * Used in SIOCSIFMUXID and SIOCGIFMUXID for 'ifconfig unplumb'.
1712 1717 */
1713 1718 int ill_muxid; /* muxid returned from plink */
1714 1719
1715 1720 /* Used for IP frag reassembly throttling on a per ILL basis. */
1716 1721 uint_t ill_ipf_gen; /* Generation of next fragment queue */
1717 1722 uint_t ill_frag_count; /* Count of all reassembly mblk bytes */
1718 1723 uint_t ill_frag_free_num_pkts; /* num of fragmented packets to free */
1719 1724 clock_t ill_last_frag_clean_time; /* time when frag's were pruned */
1720 1725 int ill_type; /* From <net/if_types.h> */
1721 1726 uint_t ill_dlpi_multicast_state; /* See below IDS_* */
1722 1727 uint_t ill_dlpi_fastpath_state; /* See below IDS_* */
1723 1728
1724 1729 /*
1725 1730 * Capabilities related fields.
1726 1731 */
1727 1732 uint_t ill_dlpi_capab_state; /* State of capability query, IDCS_* */
1728 1733 uint_t ill_capab_pending_cnt;
1729 1734 uint64_t ill_capabilities; /* Enabled capabilities, ILL_CAPAB_* */
1730 1735 ill_hcksum_capab_t *ill_hcksum_capab; /* H/W cksumming capabilities */
1731 1736 ill_zerocopy_capab_t *ill_zerocopy_capab; /* Zero-copy capabilities */
1732 1737 ill_dld_capab_t *ill_dld_capab; /* DLD capabilities */
1733 1738 ill_lso_capab_t *ill_lso_capab; /* Large Segment Offload capabilities */
1734 1739 mblk_t *ill_capab_reset_mp; /* Preallocated mblk for capab reset */
1735 1740
1736 1741 uint8_t ill_max_hops; /* Maximum hops for any logical interface */
1737 1742 uint_t ill_user_mtu; /* User-specified MTU via SIOCSLIFLNKINFO */
1738 1743 uint32_t ill_reachable_time; /* Value for ND algorithm in msec */
1739 1744 uint32_t ill_reachable_retrans_time; /* Value for ND algorithm msec */
1740 1745 uint_t ill_max_buf; /* Max # of req to buffer for ND */
1741 1746 in6_addr_t ill_token; /* IPv6 interface id */
1742 1747 in6_addr_t ill_dest_token; /* Destination IPv6 interface id */
1743 1748 uint_t ill_token_length;
1744 1749 uint32_t ill_xmit_count; /* ndp max multicast xmits */
1745 1750 mib2_ipIfStatsEntry_t *ill_ip_mib; /* ver indep. interface mib */
1746 1751 mib2_ipv6IfIcmpEntry_t *ill_icmp6_mib; /* Per interface mib */
1747 1752
1748 1753 phyint_t *ill_phyint;
1749 1754 uint64_t ill_flags;
1750 1755
1751 1756 kmutex_t ill_lock; /* Please see table below */
1752 1757 /*
1753 1758 * The ill_nd_lla* fields handle the link layer address option
1754 1759 * from neighbor discovery. This is used for external IPv6
1755 1760 * address resolution.
1756 1761 */
1757 1762 mblk_t *ill_nd_lla_mp; /* mblk which holds ill_nd_lla */
1758 1763 uint8_t *ill_nd_lla; /* Link Layer Address */
1759 1764 uint_t ill_nd_lla_len; /* Link Layer Address length */
1760 1765 /*
1761 1766 * We have 4 phys_addr_req's sent down. This field keeps track
1762 1767 * of which one is pending.
1763 1768 */
1764 1769 t_uscalar_t ill_phys_addr_pend; /* which dl_phys_addr_req pending */
1765 1770 /*
1766 1771 * Used to save errors that occur during plumbing
1767 1772 */
1768 1773 uint_t ill_ifname_pending_err;
1769 1774 avl_node_t ill_avl_byppa; /* avl node based on ppa */
1770 1775 list_t ill_nce; /* pointer to nce_s list */
1771 1776 uint_t ill_refcnt; /* active refcnt by threads */
1772 1777 uint_t ill_ire_cnt; /* ires associated with this ill */
1773 1778 kcondvar_t ill_cv;
1774 1779 uint_t ill_ncec_cnt; /* ncecs associated with this ill */
1775 1780 uint_t ill_nce_cnt; /* nces associated with this ill */
1776 1781 uint_t ill_waiters; /* threads waiting in ipsq_enter */
1777 1782 /*
1778 1783 * Contains the upper read queue pointer of the module immediately
1779 1784 * beneath IP. This field allows IP to validate sub-capability
1780 1785 * acknowledgments coming up from downstream.
1781 1786 */
1782 1787 queue_t *ill_lmod_rq; /* read queue pointer of module below */
1783 1788 uint_t ill_lmod_cnt; /* number of modules beneath IP */
1784 1789 ip_m_t *ill_media; /* media specific params/functions */
1785 1790 t_uscalar_t ill_dlpi_pending; /* Last DLPI primitive issued */
1786 1791 uint_t ill_usesrc_ifindex; /* use src addr from this ILL */
1787 1792 struct ill_s *ill_usesrc_grp_next; /* Next ILL in the usesrc group */
1788 1793 boolean_t ill_trace_disable; /* True when alloc fails */
1789 1794 zoneid_t ill_zoneid;
1790 1795 ip_stack_t *ill_ipst; /* Corresponds to a netstack_hold */
1791 1796 uint32_t ill_dhcpinit; /* IP_DHCPINIT_IFs for ill */
1792 1797 void *ill_flownotify_mh; /* Tx flow ctl, mac cb handle */
1793 1798 uint_t ill_ilm_cnt; /* ilms referencing this ill */
1794 1799 uint_t ill_ipallmulti_cnt; /* ip_join_allmulti() calls */
1795 1800 ilm_t *ill_ipallmulti_ilm;
1796 1801
1797 1802 mblk_t *ill_saved_ire_mp; /* Allocated for each extra IRE */
1798 1803 /* with ire_ill set so they can */
1799 1804 /* survive the ill going down and up. */
1800 1805 kmutex_t ill_saved_ire_lock; /* Protects ill_saved_ire_mp, cnt */
1801 1806 uint_t ill_saved_ire_cnt; /* # entries */
1802 1807 struct arl_ill_common_s *ill_common;
1803 1808 ire_t *ill_ire_multicast; /* IRE_MULTICAST for ill */
1804 1809 clock_t ill_defend_start; /* start of 1 hour period */
1805 1810 uint_t ill_defend_count; /* # of announce/defends per ill */
1806 1811 /*
1807 1812 * IPMP fields.
1808 1813 */
1809 1814 ipmp_illgrp_t *ill_grp; /* IPMP group information */
1810 1815 list_node_t ill_actnode; /* next active ill in group */
1811 1816 list_node_t ill_grpnode; /* next ill in group */
1812 1817 ipif_t *ill_src_ipif; /* source address selection rotor */
1813 1818 ipif_t *ill_move_ipif; /* ipif awaiting move to new ill */
1814 1819 boolean_t ill_nom_cast; /* nominated for mcast/bcast */
1815 1820 uint_t ill_bound_cnt; /* # of data addresses bound to ill */
1816 1821 ipif_t *ill_bound_ipif; /* ipif chain bound to ill */
1817 1822 timeout_id_t ill_refresh_tid; /* ill refresh retry timeout id */
1818 1823
1819 1824 uint32_t ill_mrouter_cnt; /* mrouter allmulti joins */
1820 1825 uint32_t ill_allowed_ips_cnt;
1821 1826 in6_addr_t *ill_allowed_ips;
1822 1827
1823 1828 /* list of multicast physical addresses joined on this ill */
1824 1829 multiphysaddr_t *ill_mphysaddr_list;
1825 1830 } ill_t;
1826 1831
1827 1832 /*
1828 1833 * ILL_FREE_OK() means that there are no incoming pointer references
1829 1834 * to the ill.
1830 1835 */
1831 1836 #define ILL_FREE_OK(ill) \
1832 1837 ((ill)->ill_ire_cnt == 0 && (ill)->ill_ilm_cnt == 0 && \
1833 1838 (ill)->ill_ncec_cnt == 0 && (ill)->ill_nce_cnt == 0)
1834 1839
1835 1840 /*
1836 1841 * An ipif/ill can be marked down only when the ire and ncec references
1837 1842 * to that ipif/ill goes to zero. ILL_DOWN_OK() is a necessary condition
1838 1843 * quiescence checks. See comments above IPIF_DOWN_OK for details
1839 1844 * on why ires and nces are selectively considered for this macro.
1840 1845 */
1841 1846 #define ILL_DOWN_OK(ill) \
1842 1847 (ill->ill_ire_cnt == 0 && ill->ill_ncec_cnt == 0 && \
1843 1848 ill->ill_nce_cnt == 0)
1844 1849
1845 1850 /*
1846 1851 * The following table lists the protection levels of the various members
1847 1852 * of the ill_t. Same notation as that used for ipif_t above is used.
1848 1853 *
1849 1854 * Write Read
1850 1855 *
1851 1856 * ill_ifptr ill_g_lock + s Write once
1852 1857 * ill_rq ipsq Write once
1853 1858 * ill_wq ipsq Write once
1854 1859 *
1855 1860 * ill_error ipsq None
1856 1861 * ill_ipif ill_g_lock + ipsq ill_g_lock OR ipsq
1857 1862 * ill_ipif_up_count ill_lock + ipsq ill_lock OR ipsq
1858 1863 * ill_max_frag ill_lock ill_lock
1859 1864 * ill_current_frag ill_lock ill_lock
1860 1865 *
1861 1866 * ill_name ill_g_lock + ipsq Write once
1862 1867 * ill_name_length ill_g_lock + ipsq Write once
1863 1868 * ill_ndd_name ipsq Write once
1864 1869 * ill_net_type ipsq Write once
1865 1870 * ill_ppa ill_g_lock + ipsq Write once
1866 1871 * ill_sap ipsq + down ill Write once
1867 1872 * ill_sap_length ipsq + down ill Write once
1868 1873 * ill_phys_addr_length ipsq + down ill Write once
1869 1874 *
1870 1875 * ill_bcast_addr_length ipsq ipsq
1871 1876 * ill_mactype ipsq ipsq
1872 1877 * ill_frag_ptr ipsq ipsq
1873 1878 *
1874 1879 * ill_frag_timer_id ill_lock ill_lock
1875 1880 * ill_frag_hash_tbl ipsq up ill
1876 1881 * ill_ilm ill_mcast_lock(WRITER) ill_mcast_lock(READER)
1877 1882 * ill_global_timer ill_mcast_lock(WRITER) ill_mcast_lock(READER)
1878 1883 * ill_mcast_type ill_mcast_lock(WRITER) ill_mcast_lock(READER)
1879 1884 * ill_mcast_v1_time ill_mcast_lock(WRITER) ill_mcast_lock(READER)
1880 1885 * ill_mcast_v2_time ill_mcast_lock(WRITER) ill_mcast_lock(READER)
1881 1886 * ill_mcast_v1_tset ill_mcast_lock(WRITER) ill_mcast_lock(READER)
1882 1887 * ill_mcast_v2_tset ill_mcast_lock(WRITER) ill_mcast_lock(READER)
1883 1888 * ill_mcast_rv ill_mcast_lock(WRITER) ill_mcast_lock(READER)
1884 1889 * ill_mcast_qi ill_mcast_lock(WRITER) ill_mcast_lock(READER)
1885 1890 *
1886 1891 * ill_down_mp ipsq ipsq
1887 1892 * ill_dlpi_deferred ill_lock ill_lock
1888 1893 * ill_dlpi_pending ipsq + ill_lock ipsq or ill_lock or
1889 1894 * absence of ipsq writer.
1890 1895 * ill_phys_addr_mp ipsq + down ill only when ill is up
1891 1896 * ill_mcast_deferred ill_lock ill_lock
1892 1897 * ill_phys_addr ipsq + down ill only when ill is up
1893 1898 * ill_dest_addr_mp ipsq + down ill only when ill is up
1894 1899 * ill_dest_addr ipsq + down ill only when ill is up
1895 1900 *
1896 1901 * ill_state_flags ill_lock ill_lock
1897 1902 * exclusive bit flags ipsq_t ipsq_t
1898 1903 * shared bit flags ill_lock ill_lock
1899 1904 *
1900 1905 * ill_muxid ipsq Not atomic
1901 1906 *
1902 1907 * ill_ipf_gen Not atomic
1903 1908 * ill_frag_count atomics atomics
1904 1909 * ill_type ipsq + down ill only when ill is up
1905 1910 * ill_dlpi_multicast_state ill_lock ill_lock
1906 1911 * ill_dlpi_fastpath_state ill_lock ill_lock
1907 1912 * ill_dlpi_capab_state ipsq ipsq
1908 1913 * ill_max_hops ipsq Not atomic
1909 1914 *
1910 1915 * ill_mtu ill_lock None
1911 1916 * ill_mc_mtu ill_lock None
1912 1917 *
1913 1918 * ill_user_mtu ipsq + ill_lock ill_lock
1914 1919 * ill_reachable_time ipsq + ill_lock ill_lock
1915 1920 * ill_reachable_retrans_time ipsq + ill_lock ill_lock
1916 1921 * ill_max_buf ipsq + ill_lock ill_lock
1917 1922 *
1918 1923 * Next 2 fields need ill_lock because of the get ioctls. They should not
1919 1924 * report partially updated results without executing in the ipsq.
1920 1925 * ill_token ipsq + ill_lock ill_lock
1921 1926 * ill_token_length ipsq + ill_lock ill_lock
1922 1927 * ill_dest_token ipsq + down ill only when ill is up
1923 1928 * ill_xmit_count ipsq + down ill write once
1924 1929 * ill_ip6_mib ipsq + down ill only when ill is up
1925 1930 * ill_icmp6_mib ipsq + down ill only when ill is up
1926 1931 *
1927 1932 * ill_phyint ipsq, ill_g_lock, ill_lock Any of them
1928 1933 * ill_flags ill_lock ill_lock
1929 1934 * ill_nd_lla_mp ipsq + down ill only when ill is up
1930 1935 * ill_nd_lla ipsq + down ill only when ill is up
1931 1936 * ill_nd_lla_len ipsq + down ill only when ill is up
1932 1937 * ill_phys_addr_pend ipsq + down ill only when ill is up
1933 1938 * ill_ifname_pending_err ipsq ipsq
1934 1939 * ill_avl_byppa ipsq, ill_g_lock write once
1935 1940 *
1936 1941 * ill_fastpath_list ill_lock ill_lock
1937 1942 * ill_refcnt ill_lock ill_lock
1938 1943 * ill_ire_cnt ill_lock ill_lock
1939 1944 * ill_cv ill_lock ill_lock
1940 1945 * ill_ncec_cnt ill_lock ill_lock
1941 1946 * ill_nce_cnt ill_lock ill_lock
1942 1947 * ill_ilm_cnt ill_lock ill_lock
1943 1948 * ill_src_ipif ill_g_lock ill_g_lock
1944 1949 * ill_trace ill_lock ill_lock
1945 1950 * ill_usesrc_grp_next ill_g_usesrc_lock ill_g_usesrc_lock
1946 1951 * ill_dhcpinit atomics atomics
1947 1952 * ill_flownotify_mh write once write once
1948 1953 * ill_capab_pending_cnt ipsq ipsq
1949 1954 * ill_ipallmulti_cnt ill_lock ill_lock
1950 1955 * ill_ipallmulti_ilm ill_lock ill_lock
1951 1956 * ill_saved_ire_mp ill_saved_ire_lock ill_saved_ire_lock
1952 1957 * ill_saved_ire_cnt ill_saved_ire_lock ill_saved_ire_lock
1953 1958 * ill_arl ??? ???
1954 1959 * ill_ire_multicast ipsq + quiescent none
1955 1960 * ill_bound_ipif ipsq ipsq
1956 1961 * ill_actnode ipsq + ipmp_lock ipsq OR ipmp_lock
1957 1962 * ill_grpnode ipsq + ill_g_lock ipsq OR ill_g_lock
1958 1963 * ill_src_ipif ill_g_lock ill_g_lock
1959 1964 * ill_move_ipif ipsq ipsq
1960 1965 * ill_nom_cast ipsq ipsq OR advisory
1961 1966 * ill_refresh_tid ill_lock ill_lock
1962 1967 * ill_grp (for IPMP ill) write once write once
1963 1968 * ill_grp (for underlying ill) ipsq + ill_g_lock ipsq OR ill_g_lock
1964 1969 * ill_grp_pending ill_mcast_serializer ill_mcast_serializer
1965 1970 * ill_mrouter_cnt atomics atomics
1966 1971 * ill_mphysaddr_list ill_lock ill_lock
1967 1972 *
1968 1973 * NOTE: It's OK to make heuristic decisions on an underlying interface
1969 1974 * by using IS_UNDER_IPMP() or comparing ill_grp's raw pointer value.
1970 1975 */
1971 1976
1972 1977 /*
1973 1978 * For ioctl restart mechanism see ip_reprocess_ioctl()
1974 1979 */
1975 1980 struct ip_ioctl_cmd_s;
1976 1981
1977 1982 typedef int (*ifunc_t)(ipif_t *, struct sockaddr_in *, queue_t *, mblk_t *,
1978 1983 struct ip_ioctl_cmd_s *, void *);
1979 1984
1980 1985 typedef struct ip_ioctl_cmd_s {
1981 1986 int ipi_cmd;
1982 1987 size_t ipi_copyin_size;
1983 1988 uint_t ipi_flags;
1984 1989 uint_t ipi_cmd_type;
1985 1990 ifunc_t ipi_func;
1986 1991 ifunc_t ipi_func_restart;
1987 1992 } ip_ioctl_cmd_t;
1988 1993
1989 1994 /*
1990 1995 * ipi_cmd_type:
1991 1996 *
1992 1997 * IF_CMD 1 old style ifreq cmd
1993 1998 * LIF_CMD 2 new style lifreq cmd
1994 1999 * ARP_CMD 3 arpreq cmd
1995 2000 * XARP_CMD 4 xarpreq cmd
1996 2001 * MSFILT_CMD 5 multicast source filter cmd
1997 2002 * MISC_CMD 6 misc cmd (not a more specific one above)
1998 2003 */
1999 2004
2000 2005 enum { IF_CMD = 1, LIF_CMD, ARP_CMD, XARP_CMD, MSFILT_CMD, MISC_CMD };
2001 2006
2002 2007 #define IPI_DONTCARE 0 /* For ioctl encoded values that don't matter */
2003 2008
2004 2009 /* Flag values in ipi_flags */
2005 2010 #define IPI_PRIV 0x1 /* Root only command */
2006 2011 #define IPI_MODOK 0x2 /* Permitted on mod instance of IP */
2007 2012 #define IPI_WR 0x4 /* Need to grab writer access */
2008 2013 #define IPI_GET_CMD 0x8 /* branch to mi_copyout on success */
2009 2014 /* unused 0x10 */
2010 2015 #define IPI_NULL_BCONT 0x20 /* ioctl has not data and hence no b_cont */
2011 2016
2012 2017 extern ip_ioctl_cmd_t ip_ndx_ioctl_table[];
2013 2018 extern ip_ioctl_cmd_t ip_misc_ioctl_table[];
2014 2019 extern int ip_ndx_ioctl_count;
2015 2020 extern int ip_misc_ioctl_count;
2016 2021
2017 2022 /* Passed down by ARP to IP during I_PLINK/I_PUNLINK */
2018 2023 typedef struct ipmx_s {
2019 2024 char ipmx_name[LIFNAMSIZ]; /* if name */
2020 2025 uint_t
2021 2026 ipmx_arpdev_stream : 1, /* This is the arp stream */
2022 2027 ipmx_notused : 31;
2023 2028 } ipmx_t;
2024 2029
2025 2030 /*
2026 2031 * State for detecting if a driver supports certain features.
2027 2032 * Support for DL_ENABMULTI_REQ uses ill_dlpi_multicast_state.
2028 2033 * Support for DLPI M_DATA fastpath uses ill_dlpi_fastpath_state.
2029 2034 */
2030 2035 #define IDS_UNKNOWN 0 /* No DLPI request sent */
2031 2036 #define IDS_INPROGRESS 1 /* DLPI request sent */
2032 2037 #define IDS_OK 2 /* DLPI request completed successfully */
2033 2038 #define IDS_FAILED 3 /* DLPI request failed */
2034 2039
2035 2040 /* Support for DL_CAPABILITY_REQ uses ill_dlpi_capab_state. */
2036 2041 enum {
2037 2042 IDCS_UNKNOWN,
2038 2043 IDCS_PROBE_SENT,
2039 2044 IDCS_OK,
2040 2045 IDCS_RESET_SENT,
2041 2046 IDCS_RENEG,
2042 2047 IDCS_FAILED
2043 2048 };
2044 2049
2045 2050 /* Extended NDP Management Structure */
2046 2051 typedef struct ipndp_s {
2047 2052 ndgetf_t ip_ndp_getf;
2048 2053 ndsetf_t ip_ndp_setf;
2049 2054 caddr_t ip_ndp_data;
2050 2055 char *ip_ndp_name;
2051 2056 } ipndp_t;
2052 2057
2053 2058 /* IXA Notification types */
2054 2059 typedef enum {
2055 2060 IXAN_LSO, /* LSO capability change */
2056 2061 IXAN_PMTU, /* PMTU change */
2057 2062 IXAN_ZCOPY /* ZEROCOPY capability change */
2058 2063 } ixa_notify_type_t;
2059 2064
2060 2065 typedef uint_t ixa_notify_arg_t;
2061 2066
2062 2067 typedef void (*ixa_notify_t)(void *, ip_xmit_attr_t *ixa, ixa_notify_type_t,
2063 2068 ixa_notify_arg_t);
2064 2069
2065 2070 /*
2066 2071 * Attribute flags that are common to the transmit and receive attributes
2067 2072 */
2068 2073 #define IAF_IS_IPV4 0x80000000 /* ipsec_*_v4 */
2069 2074 #define IAF_TRUSTED_ICMP 0x40000000 /* ipsec_*_icmp_loopback */
2070 2075 #define IAF_NO_LOOP_ZONEID_SET 0x20000000 /* Zone that shouldn't have */
2071 2076 /* a copy */
2072 2077 #define IAF_LOOPBACK_COPY 0x10000000 /* For multi and broadcast */
2073 2078
2074 2079 #define IAF_MASK 0xf0000000 /* Flags that are common */
2075 2080
2076 2081 /*
2077 2082 * Transmit side attributes used between the transport protocols and IP as
2078 2083 * well as inside IP. It is also used to cache information in the conn_t i.e.
2079 2084 * replaces conn_ire and the IPsec caching in the conn_t.
2080 2085 */
2081 2086 struct ip_xmit_attr_s {
2082 2087 iaflags_t ixa_flags; /* IXAF_*. See below */
2083 2088
2084 2089 uint32_t ixa_free_flags; /* IXA_FREE_*. See below */
2085 2090 uint32_t ixa_refcnt; /* Using atomics */
2086 2091
2087 2092 /*
2088 2093 * Always initialized independently of ixa_flags settings.
2089 2094 * Used by ip_xmit so we keep them up front for cache locality.
2090 2095 */
2091 2096 uint32_t ixa_xmit_hint; /* For ECMP and GLD TX ring fanout */
2092 2097 uint_t ixa_pktlen; /* Always set. For frag and stats */
2093 2098 zoneid_t ixa_zoneid; /* Assumed always set */
2094 2099
2095 2100 /* Always set for conn_ip_output(); might be stale */
2096 2101 /*
2097 2102 * Since TCP keeps the conn_t around past the process going away
2098 2103 * we need to use the "notr" (e.g, ire_refhold_notr) for ixa_ire,
2099 2104 * ixa_nce, and ixa_dce.
2100 2105 */
2101 2106 ire_t *ixa_ire; /* Forwarding table entry */
2102 2107 uint_t ixa_ire_generation;
2103 2108 nce_t *ixa_nce; /* Neighbor cache entry */
2104 2109 dce_t *ixa_dce; /* Destination cache entry */
2105 2110 uint_t ixa_dce_generation;
2106 2111 uint_t ixa_src_generation; /* If IXAF_VERIFY_SOURCE */
2107 2112
2108 2113 uint32_t ixa_src_preferences; /* prefs for src addr select */
2109 2114 uint32_t ixa_pmtu; /* IXAF_VERIFY_PMTU */
2110 2115
2111 2116 /* Set by ULP if IXAF_VERIFY_PMTU; otherwise set by IP */
2112 2117 uint32_t ixa_fragsize;
2113 2118
2114 2119 int8_t ixa_use_min_mtu; /* IXAF_USE_MIN_MTU values */
2115 2120
2116 2121 pfirepostfrag_t ixa_postfragfn; /* Set internally in IP */
2117 2122
2118 2123 in6_addr_t ixa_nexthop_v6; /* IXAF_NEXTHOP_SET */
2119 2124 #define ixa_nexthop_v4 V4_PART_OF_V6(ixa_nexthop_v6)
2120 2125
2121 2126 zoneid_t ixa_no_loop_zoneid; /* IXAF_NO_LOOP_ZONEID_SET */
2122 2127
2123 2128 uint_t ixa_scopeid; /* For IPv6 link-locals */
2124 2129
2125 2130 uint_t ixa_broadcast_ttl; /* IXAF_BROACAST_TTL_SET */
2126 2131
2127 2132 uint_t ixa_multicast_ttl; /* Assumed set for multicast */
2128 2133 uint_t ixa_multicast_ifindex; /* Assumed set for multicast */
2129 2134 ipaddr_t ixa_multicast_ifaddr; /* Assumed set for multicast */
2130 2135
2131 2136 int ixa_raw_cksum_offset; /* If IXAF_SET_RAW_CKSUM */
2132 2137
2133 2138 uint32_t ixa_ident; /* For IPv6 fragment header */
2134 2139
2135 2140 uint64_t ixa_conn_id; /* Used by DTrace */
2136 2141 /*
2137 2142 * Cached LSO information.
2138 2143 */
2139 2144 ill_lso_capab_t ixa_lso_capab; /* Valid when IXAF_LSO_CAPAB */
2140 2145
2141 2146 uint64_t ixa_ipsec_policy_gen; /* Generation from iph_gen */
2142 2147 /*
2143 2148 * The following IPsec fields are only initialized when
2144 2149 * IXAF_IPSEC_SECURE is set. Otherwise they contain garbage.
2145 2150 */
2146 2151 ipsec_latch_t *ixa_ipsec_latch; /* Just the ids */
2147 2152 struct ipsa_s *ixa_ipsec_ah_sa; /* Hard reference SA for AH */
2148 2153 struct ipsa_s *ixa_ipsec_esp_sa; /* Hard reference SA for ESP */
2149 2154 struct ipsec_policy_s *ixa_ipsec_policy; /* why are we here? */
2150 2155 struct ipsec_action_s *ixa_ipsec_action; /* For reflected packets */
2151 2156 ipsa_ref_t ixa_ipsec_ref[2]; /* Soft reference to SA */
2152 2157 /* 0: ESP, 1: AH */
2153 2158
2154 2159 /*
2155 2160 * The selectors here are potentially different than the SPD rule's
2156 2161 * selectors, and we need to have both available for IKEv2.
2157 2162 *
2158 2163 * NOTE: "Source" and "Dest" are w.r.t. outbound datagrams. Ports can
2159 2164 * be zero, and the protocol number is needed to make the ports
2160 2165 * significant.
2161 2166 */
2162 2167 uint16_t ixa_ipsec_src_port; /* Source port number of d-gram. */
2163 2168 uint16_t ixa_ipsec_dst_port; /* Destination port number of d-gram. */
2164 2169 uint8_t ixa_ipsec_icmp_type; /* ICMP type of d-gram */
2165 2170 uint8_t ixa_ipsec_icmp_code; /* ICMP code of d-gram */
2166 2171
2167 2172 sa_family_t ixa_ipsec_inaf; /* Inner address family */
2168 2173 #define IXA_MAX_ADDRLEN 4 /* Max addr len. (in 32-bit words) */
2169 2174 uint32_t ixa_ipsec_insrc[IXA_MAX_ADDRLEN]; /* Inner src address */
2170 2175 uint32_t ixa_ipsec_indst[IXA_MAX_ADDRLEN]; /* Inner dest address */
2171 2176 uint8_t ixa_ipsec_insrcpfx; /* Inner source prefix */
2172 2177 uint8_t ixa_ipsec_indstpfx; /* Inner destination prefix */
2173 2178
2174 2179 uint8_t ixa_ipsec_proto; /* IP protocol number for d-gram. */
2175 2180
2176 2181 /* Always initialized independently of ixa_flags settings */
2177 2182 uint_t ixa_ifindex; /* Assumed always set */
2178 2183 uint16_t ixa_ip_hdr_length; /* Points to ULP header */
2179 2184 uint8_t ixa_protocol; /* Protocol number for ULP cksum */
2180 2185 ts_label_t *ixa_tsl; /* Always set. NULL if not TX */
2181 2186 ip_stack_t *ixa_ipst; /* Always set */
2182 2187 uint32_t ixa_extra_ident; /* Set if LSO */
2183 2188 cred_t *ixa_cred; /* For getpeerucred */
2184 2189 pid_t ixa_cpid; /* For getpeerucred */
2185 2190
2186 2191 #ifdef DEBUG
2187 2192 kthread_t *ixa_curthread; /* For serialization assert */
2188 2193 #endif
2189 2194 squeue_t *ixa_sqp; /* Set from conn_sqp as a hint */
2190 2195 uintptr_t ixa_cookie; /* cookie to use for tx flow control */
2191 2196
2192 2197 /*
2193 2198 * Must be set by ULP if any of IXAF_VERIFY_LSO, IXAF_VERIFY_PMTU,
2194 2199 * or IXAF_VERIFY_ZCOPY is set.
2195 2200 */
2196 2201 ixa_notify_t ixa_notify; /* Registered upcall notify function */
2197 2202 void *ixa_notify_cookie; /* ULP cookie for ixa_notify */
2198 2203 };
2199 2204
2200 2205 /*
2201 2206 * Flags to indicate which transmit attributes are set.
2202 2207 * Split into "xxx_SET" ones which indicate that the "xxx" field it set, and
2203 2208 * single flags.
2204 2209 */
2205 2210 #define IXAF_REACH_CONF 0x00000001 /* Reachability confirmation */
2206 2211 #define IXAF_BROADCAST_TTL_SET 0x00000002 /* ixa_broadcast_ttl valid */
2207 2212 #define IXAF_SET_SOURCE 0x00000004 /* Replace if broadcast */
2208 2213 #define IXAF_USE_MIN_MTU 0x00000008 /* IPV6_USE_MIN_MTU */
2209 2214
2210 2215 #define IXAF_DONTFRAG 0x00000010 /* IP*_DONTFRAG */
2211 2216 #define IXAF_VERIFY_PMTU 0x00000020 /* ixa_pmtu/ixa_fragsize set */
2212 2217 #define IXAF_PMTU_DISCOVERY 0x00000040 /* Create/use PMTU state */
2213 2218 #define IXAF_MULTICAST_LOOP 0x00000080 /* IP_MULTICAST_LOOP */
2214 2219
2215 2220 #define IXAF_IPSEC_SECURE 0x00000100 /* Need IPsec processing */
2216 2221 #define IXAF_UCRED_TSL 0x00000200 /* ixa_tsl from SCM_UCRED */
2217 2222 #define IXAF_DONTROUTE 0x00000400 /* SO_DONTROUTE */
2218 2223 #define IXAF_NO_IPSEC 0x00000800 /* Ignore policy */
2219 2224
2220 2225 #define IXAF_PMTU_TOO_SMALL 0x00001000 /* PMTU too small */
2221 2226 #define IXAF_SET_ULP_CKSUM 0x00002000 /* Calculate ULP checksum */
2222 2227 #define IXAF_VERIFY_SOURCE 0x00004000 /* Check that source is ok */
2223 2228 #define IXAF_NEXTHOP_SET 0x00008000 /* ixa_nexthop set */
2224 2229
2225 2230 #define IXAF_PMTU_IPV4_DF 0x00010000 /* Set IPv4 DF */
2226 2231 #define IXAF_NO_DEV_FLOW_CTL 0x00020000 /* Protocol needs no flow ctl */
2227 2232 #define IXAF_NO_TTL_CHANGE 0x00040000 /* Internal to IP */
2228 2233 #define IXAF_IPV6_ADD_FRAGHDR 0x00080000 /* Add fragment header */
2229 2234
2230 2235 #define IXAF_IPSEC_TUNNEL 0x00100000 /* Tunnel mode */
2231 2236 #define IXAF_NO_PFHOOK 0x00200000 /* Skip xmit pfhook */
2232 2237 #define IXAF_NO_TRACE 0x00400000 /* When back from ARP/ND */
2233 2238 #define IXAF_SCOPEID_SET 0x00800000 /* ixa_scopeid set */
2234 2239
2235 2240 #define IXAF_MULTIRT_MULTICAST 0x01000000 /* MULTIRT for multicast */
2236 2241 #define IXAF_NO_HW_CKSUM 0x02000000 /* Force software cksum */
2237 2242 #define IXAF_SET_RAW_CKSUM 0x04000000 /* Use ixa_raw_cksum_offset */
2238 2243 #define IXAF_IPSEC_GLOBAL_POLICY 0x08000000 /* Policy came from global */
2239 2244
2240 2245 /* Note the following uses bits 0x10000000 through 0x80000000 */
2241 2246 #define IXAF_IS_IPV4 IAF_IS_IPV4
2242 2247 #define IXAF_TRUSTED_ICMP IAF_TRUSTED_ICMP
2243 2248 #define IXAF_NO_LOOP_ZONEID_SET IAF_NO_LOOP_ZONEID_SET
2244 2249 #define IXAF_LOOPBACK_COPY IAF_LOOPBACK_COPY
2245 2250
2246 2251 /* Note: use the upper 32 bits */
2247 2252 #define IXAF_VERIFY_LSO 0x100000000 /* Check LSO capability */
2248 2253 #define IXAF_LSO_CAPAB 0x200000000 /* Capable of LSO */
2249 2254 #define IXAF_VERIFY_ZCOPY 0x400000000 /* Check Zero Copy capability */
2250 2255 #define IXAF_ZCOPY_CAPAB 0x800000000 /* Capable of ZEROCOPY */
2251 2256
2252 2257 /*
2253 2258 * The normal flags for sending packets e.g., icmp errors
2254 2259 */
2255 2260 #define IXAF_BASIC_SIMPLE_V4 \
2256 2261 (IXAF_SET_ULP_CKSUM | IXAF_IS_IPV4 | IXAF_VERIFY_SOURCE)
2257 2262 #define IXAF_BASIC_SIMPLE_V6 (IXAF_SET_ULP_CKSUM | IXAF_VERIFY_SOURCE)
2258 2263
2259 2264 /*
2260 2265 * Normally these fields do not have a hold. But in some cases they do, for
2261 2266 * instance when we've gone through ip_*_attr_to/from_mblk.
2262 2267 * We use ixa_free_flags to indicate that they have a hold and need to be
2263 2268 * released on cleanup.
2264 2269 */
2265 2270 #define IXA_FREE_CRED 0x00000001 /* ixa_cred needs to be rele */
2266 2271 #define IXA_FREE_TSL 0x00000002 /* ixa_tsl needs to be rele */
2267 2272
2268 2273 /*
2269 2274 * Simplistic way to set the ixa_xmit_hint for locally generated traffic
2270 2275 * and forwarded traffic. The shift amount are based on the size of the
2271 2276 * structs to discard the low order bits which don't have much if any variation
2272 2277 * (coloring in kmem_cache_alloc might provide some variation).
2273 2278 *
2274 2279 * Basing the locally generated hint on the address of the conn_t means that
2275 2280 * the packets from the same socket/connection do not get reordered.
2276 2281 * Basing the hint for forwarded traffic on the ill_ring_t means that
2277 2282 * packets from the same NIC+ring are likely to use the same outbound ring
2278 2283 * hence we get low contention on the ring in the transmitting driver.
2279 2284 */
2280 2285 #define CONN_TO_XMIT_HINT(connp) ((uint32_t)(((uintptr_t)connp) >> 11))
2281 2286 #define ILL_RING_TO_XMIT_HINT(ring) ((uint32_t)(((uintptr_t)ring) >> 7))
2282 2287
2283 2288 /*
2284 2289 * IP set Destination Flags used by function ip_set_destination,
2285 2290 * ip_attr_connect, and conn_connect.
2286 2291 */
2287 2292 #define IPDF_ALLOW_MCBC 0x1 /* Allow multi/broadcast */
2288 2293 #define IPDF_VERIFY_DST 0x2 /* Verify destination addr */
2289 2294 #define IPDF_SELECT_SRC 0x4 /* Select source address */
2290 2295 #define IPDF_LSO 0x8 /* Try LSO */
2291 2296 #define IPDF_IPSEC 0x10 /* Set IPsec policy */
2292 2297 #define IPDF_ZONE_IS_GLOBAL 0x20 /* From conn_zone_is_global */
2293 2298 #define IPDF_ZCOPY 0x40 /* Try ZEROCOPY */
2294 2299 #define IPDF_UNIQUE_DCE 0x80 /* Get a per-destination DCE */
2295 2300
2296 2301 /*
2297 2302 * Receive side attributes used between the transport protocols and IP as
2298 2303 * well as inside IP.
2299 2304 */
2300 2305 struct ip_recv_attr_s {
2301 2306 iaflags_t ira_flags; /* See below */
2302 2307
2303 2308 uint32_t ira_free_flags; /* IRA_FREE_*. See below */
2304 2309
2305 2310 /*
2306 2311 * This is a hint for TCP SYN packets.
2307 2312 * Always initialized independently of ira_flags settings
2308 2313 */
2309 2314 squeue_t *ira_sqp;
2310 2315 ill_rx_ring_t *ira_ring; /* Internal to IP */
2311 2316
2312 2317 /* For ip_accept_tcp when IRAF_TARGET_SQP is set */
2313 2318 squeue_t *ira_target_sqp;
2314 2319 mblk_t *ira_target_sqp_mp;
2315 2320
2316 2321 /* Always initialized independently of ira_flags settings */
2317 2322 uint32_t ira_xmit_hint; /* For ECMP and GLD TX ring fanout */
2318 2323 zoneid_t ira_zoneid; /* ALL_ZONES unless local delivery */
2319 2324 uint_t ira_pktlen; /* Always set. For frag and stats */
2320 2325 uint16_t ira_ip_hdr_length; /* Points to ULP header */
2321 2326 uint8_t ira_protocol; /* Protocol number for ULP cksum */
2322 2327 uint_t ira_rifindex; /* Received ifindex */
2323 2328 uint_t ira_ruifindex; /* Received upper ifindex */
2324 2329 ts_label_t *ira_tsl; /* Always set. NULL if not TX */
2325 2330 /*
2326 2331 * ira_rill and ira_ill is set inside IP, but not when conn_recv is
2327 2332 * called; ULPs should use ira_ruifindex instead.
2328 2333 */
2329 2334 ill_t *ira_rill; /* ill where packet came */
2330 2335 ill_t *ira_ill; /* ill where IP address hosted */
2331 2336 cred_t *ira_cred; /* For getpeerucred */
2332 2337 pid_t ira_cpid; /* For getpeerucred */
2333 2338
2334 2339 /* Used when IRAF_VERIFIED_SRC is set; this source was ok */
2335 2340 ipaddr_t ira_verified_src;
2336 2341
2337 2342 /*
2338 2343 * The following IPsec fields are only initialized when
2339 2344 * IRAF_IPSEC_SECURE is set. Otherwise they contain garbage.
2340 2345 */
2341 2346 struct ipsec_action_s *ira_ipsec_action; /* how we made it in.. */
2342 2347 struct ipsa_s *ira_ipsec_ah_sa; /* SA for AH */
2343 2348 struct ipsa_s *ira_ipsec_esp_sa; /* SA for ESP */
2344 2349
2345 2350 ipaddr_t ira_mroute_tunnel; /* IRAF_MROUTE_TUNNEL_SET */
2346 2351
2347 2352 zoneid_t ira_no_loop_zoneid; /* IRAF_NO_LOOP_ZONEID_SET */
2348 2353
2349 2354 uint32_t ira_esp_udp_ports; /* IRAF_ESP_UDP_PORTS */
2350 2355
2351 2356 /*
2352 2357 * For IP_RECVSLLA and ip_ndp_conflict/find_solicitation.
2353 2358 * Same size as max for sockaddr_dl
2354 2359 */
2355 2360 #define IRA_L2SRC_SIZE 244
2356 2361 uint8_t ira_l2src[IRA_L2SRC_SIZE]; /* If IRAF_L2SRC_SET */
2357 2362
2358 2363 /*
2359 2364 * Local handle that we use to do lazy setting of ira_l2src.
2360 2365 * We defer setting l2src until needed but we do before any
2361 2366 * ip_input pullupmsg or copymsg.
2362 2367 */
2363 2368 struct mac_header_info_s *ira_mhip; /* Could be NULL */
2364 2369 };
2365 2370
2366 2371 /*
2367 2372 * Flags to indicate which receive attributes are set.
2368 2373 */
2369 2374 #define IRAF_SYSTEM_LABELED 0x00000001 /* is_system_labeled() */
2370 2375 #define IRAF_IPV4_OPTIONS 0x00000002 /* Performance */
2371 2376 #define IRAF_MULTICAST 0x00000004 /* Was multicast at L3 */
2372 2377 #define IRAF_BROADCAST 0x00000008 /* Was broadcast at L3 */
2373 2378 #define IRAF_MULTIBROADCAST (IRAF_MULTICAST|IRAF_BROADCAST)
2374 2379
2375 2380 #define IRAF_LOOPBACK 0x00000010 /* Looped back by IP */
2376 2381 #define IRAF_VERIFY_IP_CKSUM 0x00000020 /* Need to verify IP */
2377 2382 #define IRAF_VERIFY_ULP_CKSUM 0x00000040 /* Need to verify TCP,UDP,etc */
2378 2383 #define IRAF_SCTP_CSUM_ERR 0x00000080 /* sctp pkt has failed chksum */
2379 2384
2380 2385 #define IRAF_IPSEC_SECURE 0x00000100 /* Passed AH and/or ESP */
2381 2386 #define IRAF_DHCP_UNICAST 0x00000200
2382 2387 #define IRAF_IPSEC_DECAPS 0x00000400 /* Was packet decapsulated */
2383 2388 /* from a matching inner packet? */
2384 2389 #define IRAF_TARGET_SQP 0x00000800 /* ira_target_sqp is set */
2385 2390 #define IRAF_VERIFIED_SRC 0x00001000 /* ira_verified_src set */
2386 2391 #define IRAF_RSVP 0x00002000 /* RSVP packet for rsvpd */
2387 2392 #define IRAF_MROUTE_TUNNEL_SET 0x00004000 /* From ip_mroute_decap */
2388 2393 #define IRAF_PIM_REGISTER 0x00008000 /* From register_mforward */
2389 2394
2390 2395 #define IRAF_TX_MAC_EXEMPTABLE 0x00010000 /* Allow MAC_EXEMPT readdown */
2391 2396 #define IRAF_TX_SHARED_ADDR 0x00020000 /* Arrived on ALL_ZONES addr */
2392 2397 #define IRAF_ESP_UDP_PORTS 0x00040000 /* NAT-traversal packet */
2393 2398 #define IRAF_NO_HW_CKSUM 0x00080000 /* Force software cksum */
2394 2399
2395 2400 #define IRAF_ICMP_ERROR 0x00100000 /* Send to conn_recvicmp */
2396 2401 #define IRAF_ROUTER_ALERT 0x00200000 /* IPv6 router alert */
2397 2402 #define IRAF_L2SRC_SET 0x00400000 /* ira_l2src has been set */
2398 2403 #define IRAF_L2SRC_LOOPBACK 0x00800000 /* Came from us */
2399 2404
2400 2405 #define IRAF_L2DST_MULTICAST 0x01000000 /* Multicast at L2 */
2401 2406 #define IRAF_L2DST_BROADCAST 0x02000000 /* Broadcast at L2 */
2402 2407 /* Unused 0x04000000 */
2403 2408 /* Unused 0x08000000 */
2404 2409
2405 2410 /* Below starts with 0x10000000 */
2406 2411 #define IRAF_IS_IPV4 IAF_IS_IPV4
2407 2412 #define IRAF_TRUSTED_ICMP IAF_TRUSTED_ICMP
2408 2413 #define IRAF_NO_LOOP_ZONEID_SET IAF_NO_LOOP_ZONEID_SET
2409 2414 #define IRAF_LOOPBACK_COPY IAF_LOOPBACK_COPY
2410 2415
2411 2416 /*
2412 2417 * Normally these fields do not have a hold. But in some cases they do, for
2413 2418 * instance when we've gone through ip_*_attr_to/from_mblk.
2414 2419 * We use ira_free_flags to indicate that they have a hold and need to be
2415 2420 * released on cleanup.
2416 2421 */
2417 2422 #define IRA_FREE_CRED 0x00000001 /* ira_cred needs to be rele */
2418 2423 #define IRA_FREE_TSL 0x00000002 /* ira_tsl needs to be rele */
2419 2424
2420 2425 /*
2421 2426 * Optional destination cache entry for path MTU information,
2422 2427 * and ULP metrics.
2423 2428 */
2424 2429 struct dce_s {
2425 2430 uint_t dce_generation; /* Changed since cached? */
2426 2431 uint_t dce_flags; /* See below */
2427 2432 uint_t dce_ipversion; /* IPv4/IPv6 version */
2428 2433 uint32_t dce_pmtu; /* Path MTU if DCEF_PMTU */
2429 2434 uint32_t dce_ident; /* Per destination IP ident. */
2430 2435 iulp_t dce_uinfo; /* Metrics if DCEF_UINFO */
2431 2436
2432 2437 struct dce_s *dce_next;
2433 2438 struct dce_s **dce_ptpn;
2434 2439 struct dcb_s *dce_bucket;
2435 2440
2436 2441 union {
2437 2442 in6_addr_t dceu_v6addr;
2438 2443 ipaddr_t dceu_v4addr;
2439 2444 } dce_u;
2440 2445 #define dce_v4addr dce_u.dceu_v4addr
2441 2446 #define dce_v6addr dce_u.dceu_v6addr
2442 2447 /* Note that for IPv6+IPMP we use the ifindex for the upper interface */
2443 2448 uint_t dce_ifindex; /* For IPv6 link-locals */
2444 2449
2445 2450 kmutex_t dce_lock;
2446 2451 uint_t dce_refcnt;
2447 2452 uint64_t dce_last_change_time; /* Path MTU. In seconds */
2448 2453
2449 2454 ip_stack_t *dce_ipst; /* Does not have a netstack_hold */
2450 2455 };
2451 2456
2452 2457 /*
2453 2458 * Values for dce_generation.
2454 2459 *
2455 2460 * If a DCE has DCE_GENERATION_CONDEMNED, the last dce_refrele should delete
2456 2461 * it.
2457 2462 *
2458 2463 * DCE_GENERATION_VERIFY is never stored in dce_generation but it is
2459 2464 * stored in places that cache DCE (such as ixa_dce_generation).
2460 2465 * It is used as a signal that the cache is stale and needs to be reverified.
2461 2466 */
2462 2467 #define DCE_GENERATION_CONDEMNED 0
2463 2468 #define DCE_GENERATION_VERIFY 1
2464 2469 #define DCE_GENERATION_INITIAL 2
2465 2470 #define DCE_IS_CONDEMNED(dce) \
2466 2471 ((dce)->dce_generation == DCE_GENERATION_CONDEMNED)
2467 2472
2468 2473
2469 2474 /*
2470 2475 * Values for ips_src_generation.
2471 2476 *
2472 2477 * SRC_GENERATION_VERIFY is never stored in ips_src_generation but it is
2473 2478 * stored in places that cache IREs (ixa_src_generation). It is used as a
2474 2479 * signal that the cache is stale and needs to be reverified.
2475 2480 */
2476 2481 #define SRC_GENERATION_VERIFY 0
2477 2482 #define SRC_GENERATION_INITIAL 1
2478 2483
2479 2484 /*
2480 2485 * The kernel stores security attributes of all gateways in a database made
2481 2486 * up of one or more tsol_gcdb_t elements. Each tsol_gcdb_t contains the
2482 2487 * security-related credentials of the gateway. More than one gateways may
2483 2488 * share entries in the database.
2484 2489 *
2485 2490 * The tsol_gc_t structure represents the gateway to credential association,
2486 2491 * and refers to an entry in the database. One or more tsol_gc_t entities are
2487 2492 * grouped together to form one or more tsol_gcgrp_t, each representing the
2488 2493 * list of security attributes specific to the gateway. A gateway may be
2489 2494 * associated with at most one credentials group.
2490 2495 */
2491 2496 struct tsol_gcgrp_s;
2492 2497
2493 2498 extern uchar_t ip6opt_ls; /* TX IPv6 enabler */
2494 2499
2495 2500 /*
2496 2501 * Gateway security credential record.
2497 2502 */
2498 2503 typedef struct tsol_gcdb_s {
2499 2504 uint_t gcdb_refcnt; /* reference count */
2500 2505 struct rtsa_s gcdb_attr; /* security attributes */
2501 2506 #define gcdb_mask gcdb_attr.rtsa_mask
2502 2507 #define gcdb_doi gcdb_attr.rtsa_doi
2503 2508 #define gcdb_slrange gcdb_attr.rtsa_slrange
2504 2509 } tsol_gcdb_t;
2505 2510
2506 2511 /*
2507 2512 * Gateway to credential association.
2508 2513 */
2509 2514 typedef struct tsol_gc_s {
2510 2515 uint_t gc_refcnt; /* reference count */
2511 2516 struct tsol_gcgrp_s *gc_grp; /* pointer to group */
2512 2517 struct tsol_gc_s *gc_prev; /* previous in list */
2513 2518 struct tsol_gc_s *gc_next; /* next in list */
2514 2519 tsol_gcdb_t *gc_db; /* pointer to actual credentials */
2515 2520 } tsol_gc_t;
2516 2521
2517 2522 /*
2518 2523 * Gateway credentials group address.
2519 2524 */
2520 2525 typedef struct tsol_gcgrp_addr_s {
2521 2526 int ga_af; /* address family */
2522 2527 in6_addr_t ga_addr; /* IPv4 mapped or IPv6 address */
2523 2528 } tsol_gcgrp_addr_t;
2524 2529
2525 2530 /*
2526 2531 * Gateway credentials group.
2527 2532 */
2528 2533 typedef struct tsol_gcgrp_s {
2529 2534 uint_t gcgrp_refcnt; /* reference count */
2530 2535 krwlock_t gcgrp_rwlock; /* lock to protect following */
2531 2536 uint_t gcgrp_count; /* number of credentials */
2532 2537 tsol_gc_t *gcgrp_head; /* first credential in list */
2533 2538 tsol_gc_t *gcgrp_tail; /* last credential in list */
2534 2539 tsol_gcgrp_addr_t gcgrp_addr; /* next-hop gateway address */
2535 2540 } tsol_gcgrp_t;
2536 2541
2537 2542 extern kmutex_t gcgrp_lock;
2538 2543
2539 2544 #define GC_REFRELE(p) { \
2540 2545 ASSERT((p)->gc_grp != NULL); \
2541 2546 rw_enter(&(p)->gc_grp->gcgrp_rwlock, RW_WRITER); \
2542 2547 ASSERT((p)->gc_refcnt > 0); \
2543 2548 if (--((p)->gc_refcnt) == 0) \
2544 2549 gc_inactive(p); \
2545 2550 else \
2546 2551 rw_exit(&(p)->gc_grp->gcgrp_rwlock); \
2547 2552 }
2548 2553
2549 2554 #define GCGRP_REFHOLD(p) { \
2550 2555 mutex_enter(&gcgrp_lock); \
2551 2556 ++((p)->gcgrp_refcnt); \
2552 2557 ASSERT((p)->gcgrp_refcnt != 0); \
2553 2558 mutex_exit(&gcgrp_lock); \
2554 2559 }
2555 2560
2556 2561 #define GCGRP_REFRELE(p) { \
2557 2562 mutex_enter(&gcgrp_lock); \
2558 2563 ASSERT((p)->gcgrp_refcnt > 0); \
2559 2564 if (--((p)->gcgrp_refcnt) == 0) \
2560 2565 gcgrp_inactive(p); \
2561 2566 ASSERT(MUTEX_HELD(&gcgrp_lock)); \
2562 2567 mutex_exit(&gcgrp_lock); \
2563 2568 }
2564 2569
2565 2570 /*
2566 2571 * IRE gateway security attributes structure, pointed to by tsol_ire_gw_secattr
2567 2572 */
2568 2573 struct tsol_tnrhc;
2569 2574
2570 2575 struct tsol_ire_gw_secattr_s {
2571 2576 kmutex_t igsa_lock; /* lock to protect following */
2572 2577 struct tsol_tnrhc *igsa_rhc; /* host entry for gateway */
2573 2578 tsol_gc_t *igsa_gc; /* for prefix IREs */
2574 2579 };
2575 2580
2576 2581 void irb_refrele_ftable(irb_t *);
2577 2582
2578 2583 extern struct kmem_cache *rt_entry_cache;
2579 2584
2580 2585 typedef struct ire4 {
2581 2586 ipaddr_t ire4_mask; /* Mask for matching this IRE. */
2582 2587 ipaddr_t ire4_addr; /* Address this IRE represents. */
2583 2588 ipaddr_t ire4_gateway_addr; /* Gateway including for IRE_ONLINK */
2584 2589 ipaddr_t ire4_setsrc_addr; /* RTF_SETSRC */
2585 2590 } ire4_t;
2586 2591
2587 2592 typedef struct ire6 {
2588 2593 in6_addr_t ire6_mask; /* Mask for matching this IRE. */
2589 2594 in6_addr_t ire6_addr; /* Address this IRE represents. */
2590 2595 in6_addr_t ire6_gateway_addr; /* Gateway including for IRE_ONLINK */
2591 2596 in6_addr_t ire6_setsrc_addr; /* RTF_SETSRC */
2592 2597 } ire6_t;
2593 2598
2594 2599 typedef union ire_addr {
2595 2600 ire6_t ire6_u;
2596 2601 ire4_t ire4_u;
2597 2602 } ire_addr_u_t;
2598 2603
2599 2604 /*
2600 2605 * Internet Routing Entry
2601 2606 * When we have multiple identical IREs we logically add them by manipulating
2602 2607 * ire_identical_ref and ire_delete first decrements
2603 2608 * that and when it reaches 1 we know it is the last IRE.
2604 2609 * "identical" is defined as being the same for:
2605 2610 * ire_addr, ire_netmask, ire_gateway, ire_ill, ire_zoneid, and ire_type
2606 2611 * For instance, multiple IRE_BROADCASTs for the same subnet number are
2607 2612 * viewed as identical, and so are the IRE_INTERFACEs when there are
2608 2613 * multiple logical interfaces (on the same ill) with the same subnet prefix.
2609 2614 */
2610 2615 struct ire_s {
2611 2616 struct ire_s *ire_next; /* The hash chain must be first. */
2612 2617 struct ire_s **ire_ptpn; /* Pointer to previous next. */
2613 2618 uint32_t ire_refcnt; /* Number of references */
2614 2619 ill_t *ire_ill;
2615 2620 uint32_t ire_identical_ref; /* IRE_INTERFACE, IRE_BROADCAST */
2616 2621 uchar_t ire_ipversion; /* IPv4/IPv6 version */
2617 2622 ushort_t ire_type; /* Type of IRE */
2618 2623 uint_t ire_generation; /* Generation including CONDEMNED */
2619 2624 uint_t ire_ib_pkt_count; /* Inbound packets for ire_addr */
2620 2625 uint_t ire_ob_pkt_count; /* Outbound packets to ire_addr */
2621 2626 time_t ire_create_time; /* Time (in secs) IRE was created. */
2622 2627 uint32_t ire_flags; /* flags related to route (RTF_*) */
2623 2628 /*
2624 2629 * ire_testhidden is TRUE for INTERFACE IREs of IS_UNDER_IPMP(ill)
2625 2630 * interfaces
2626 2631 */
2627 2632 boolean_t ire_testhidden;
2628 2633 pfirerecv_t ire_recvfn; /* Receive side handling */
2629 2634 pfiresend_t ire_sendfn; /* Send side handling */
2630 2635 pfirepostfrag_t ire_postfragfn; /* Bottom end of send handling */
2631 2636
2632 2637 uint_t ire_masklen; /* # bits in ire_mask{,_v6} */
2633 2638 ire_addr_u_t ire_u; /* IPv4/IPv6 address info. */
2634 2639
2635 2640 irb_t *ire_bucket; /* Hash bucket when ire_ptphn is set */
2636 2641 kmutex_t ire_lock;
2637 2642 clock_t ire_last_used_time; /* For IRE_LOCAL reception */
2638 2643 tsol_ire_gw_secattr_t *ire_gw_secattr; /* gateway security attributes */
2639 2644 zoneid_t ire_zoneid;
2640 2645
2641 2646 /*
2642 2647 * Cached information of where to send packets that match this route.
2643 2648 * The ire_dep_* information is used to determine when ire_nce_cache
2644 2649 * needs to be updated.
2645 2650 * ire_nce_cache is the fastpath for the Neighbor Cache Entry
2646 2651 * for IPv6; arp info for IPv4
2647 2652 * Since this is a cache setup and torn down independently of
2648 2653 * applications we need to use nce_ref{rele,hold}_notr for it.
2649 2654 */
2650 2655 nce_t *ire_nce_cache;
2651 2656
2652 2657 /*
2653 2658 * Quick check whether the ire_type and ire_masklen indicates
2654 2659 * that the IRE can have ire_nce_cache set i.e., whether it is
2655 2660 * IRE_ONLINK and for a single destination.
2656 2661 */
2657 2662 boolean_t ire_nce_capable;
2658 2663
2659 2664 /*
2660 2665 * Dependency tracking so we can safely cache IRE and NCE pointers
2661 2666 * in offlink and onlink IREs.
2662 2667 * These are locked under the ips_ire_dep_lock rwlock. Write held
2663 2668 * when modifying the linkage.
2664 2669 * ire_dep_parent (Also chain towards IRE for nexthop)
2665 2670 * ire_dep_parent_generation: ire_generation of ire_dep_parent
2666 2671 * ire_dep_children (From parent to first child)
2667 2672 * ire_dep_sib_next (linked list of siblings)
2668 2673 * ire_dep_sib_ptpn (linked list of siblings)
2669 2674 *
2670 2675 * The parent has a ire_refhold on each child, and each child has
2671 2676 * an ire_refhold on its parent.
2672 2677 * Since ire_dep_parent is a cache setup and torn down independently of
2673 2678 * applications we need to use ire_ref{rele,hold}_notr for it.
2674 2679 */
2675 2680 ire_t *ire_dep_parent;
2676 2681 ire_t *ire_dep_children;
2677 2682 ire_t *ire_dep_sib_next;
2678 2683 ire_t **ire_dep_sib_ptpn; /* Pointer to previous next */
2679 2684 uint_t ire_dep_parent_generation;
2680 2685
2681 2686 uint_t ire_badcnt; /* Number of times ND_UNREACHABLE */
2682 2687 uint64_t ire_last_badcnt; /* In seconds */
2683 2688
2684 2689 /* ire_defense* and ire_last_used_time are only used on IRE_LOCALs */
2685 2690 uint_t ire_defense_count; /* number of ARP conflicts */
2686 2691 uint_t ire_defense_time; /* last time defended (secs) */
2687 2692
2688 2693 boolean_t ire_trace_disable; /* True when alloc fails */
2689 2694 ip_stack_t *ire_ipst; /* Does not have a netstack_hold */
2690 2695 iulp_t ire_metrics;
2691 2696 /*
2692 2697 * default and prefix routes that are added without explicitly
2693 2698 * specifying the interface are termed "unbound" routes, and will
2694 2699 * have ire_unbound set to true.
2695 2700 */
2696 2701 boolean_t ire_unbound;
2697 2702 };
2698 2703
2699 2704 /* IPv4 compatibility macros */
2700 2705 #define ire_mask ire_u.ire4_u.ire4_mask
2701 2706 #define ire_addr ire_u.ire4_u.ire4_addr
2702 2707 #define ire_gateway_addr ire_u.ire4_u.ire4_gateway_addr
2703 2708 #define ire_setsrc_addr ire_u.ire4_u.ire4_setsrc_addr
2704 2709
2705 2710 #define ire_mask_v6 ire_u.ire6_u.ire6_mask
2706 2711 #define ire_addr_v6 ire_u.ire6_u.ire6_addr
2707 2712 #define ire_gateway_addr_v6 ire_u.ire6_u.ire6_gateway_addr
2708 2713 #define ire_setsrc_addr_v6 ire_u.ire6_u.ire6_setsrc_addr
2709 2714
2710 2715 /*
2711 2716 * Values for ire_generation.
2712 2717 *
2713 2718 * If an IRE is marked with IRE_IS_CONDEMNED, the last walker of
2714 2719 * the bucket should delete this IRE from this bucket.
2715 2720 *
2716 2721 * IRE_GENERATION_VERIFY is never stored in ire_generation but it is
2717 2722 * stored in places that cache IREs (such as ixa_ire_generation and
2718 2723 * ire_dep_parent_generation). It is used as a signal that the cache is
2719 2724 * stale and needs to be reverified.
2720 2725 */
2721 2726 #define IRE_GENERATION_CONDEMNED 0
2722 2727 #define IRE_GENERATION_VERIFY 1
2723 2728 #define IRE_GENERATION_INITIAL 2
2724 2729 #define IRE_IS_CONDEMNED(ire) \
2725 2730 ((ire)->ire_generation == IRE_GENERATION_CONDEMNED)
2726 2731
2727 2732 /* Convenient typedefs for sockaddrs */
2728 2733 typedef struct sockaddr_in sin_t;
2729 2734 typedef struct sockaddr_in6 sin6_t;
2730 2735
2731 2736 /* Name/Value Descriptor. */
2732 2737 typedef struct nv_s {
2733 2738 uint64_t nv_value;
2734 2739 char *nv_name;
2735 2740 } nv_t;
2736 2741
2737 2742 #define ILL_FRAG_HASH(s, i) \
2738 2743 ((ntohl(s) ^ ((i) ^ ((i) >> 8))) % ILL_FRAG_HASH_TBL_COUNT)
2739 2744
2740 2745 /*
2741 2746 * The MAX number of allowed fragmented packets per hash bucket
2742 2747 * calculation is based on the most common mtu size of 1500. This limit
2743 2748 * will work well for other mtu sizes as well.
2744 2749 */
2745 2750 #define COMMON_IP_MTU 1500
2746 2751 #define MAX_FRAG_MIN 10
2747 2752 #define MAX_FRAG_PKTS(ipst) \
2748 2753 MAX(MAX_FRAG_MIN, (2 * (ipst->ips_ip_reass_queue_bytes / \
2749 2754 (COMMON_IP_MTU * ILL_FRAG_HASH_TBL_COUNT))))
2750 2755
2751 2756 /*
2752 2757 * Maximum dups allowed per packet.
2753 2758 */
2754 2759 extern uint_t ip_max_frag_dups;
2755 2760
2756 2761 /*
2757 2762 * Per-packet information for received packets and transmitted.
2758 2763 * Used by the transport protocols when converting between the packet
2759 2764 * and ancillary data and socket options.
2760 2765 *
2761 2766 * Note: This private data structure and related IPPF_* constant
2762 2767 * definitions are exposed to enable compilation of some debugging tools
2763 2768 * like lsof which use struct tcp_t in <inet/tcp.h>. This is intended to be
2764 2769 * a temporary hack and long term alternate interfaces should be defined
2765 2770 * to support the needs of such tools and private definitions moved to
2766 2771 * private headers.
2767 2772 */
2768 2773 struct ip_pkt_s {
2769 2774 uint_t ipp_fields; /* Which fields are valid */
2770 2775 in6_addr_t ipp_addr; /* pktinfo src/dst addr */
2771 2776 #define ipp_addr_v4 V4_PART_OF_V6(ipp_addr)
2772 2777 uint_t ipp_unicast_hops; /* IPV6_UNICAST_HOPS, IP_TTL */
2773 2778 uint_t ipp_hoplimit; /* IPV6_HOPLIMIT */
2774 2779 uint_t ipp_hopoptslen;
2775 2780 uint_t ipp_rthdrdstoptslen;
2776 2781 uint_t ipp_rthdrlen;
2777 2782 uint_t ipp_dstoptslen;
2778 2783 uint_t ipp_fraghdrlen;
2779 2784 ip6_hbh_t *ipp_hopopts;
2780 2785 ip6_dest_t *ipp_rthdrdstopts;
2781 2786 ip6_rthdr_t *ipp_rthdr;
2782 2787 ip6_dest_t *ipp_dstopts;
2783 2788 ip6_frag_t *ipp_fraghdr;
2784 2789 uint8_t ipp_tclass; /* IPV6_TCLASS */
2785 2790 uint8_t ipp_type_of_service; /* IP_TOS */
2786 2791 uint_t ipp_ipv4_options_len; /* Len of IPv4 options */
2787 2792 uint8_t *ipp_ipv4_options; /* Ptr to IPv4 options */
2788 2793 uint_t ipp_label_len_v4; /* Len of TX label for IPv4 */
2789 2794 uint8_t *ipp_label_v4; /* TX label for IPv4 */
2790 2795 uint_t ipp_label_len_v6; /* Len of TX label for IPv6 */
2791 2796 uint8_t *ipp_label_v6; /* TX label for IPv6 */
2792 2797 };
2793 2798 typedef struct ip_pkt_s ip_pkt_t;
2794 2799
2795 2800 extern void ip_pkt_free(ip_pkt_t *); /* free storage inside ip_pkt_t */
2796 2801 extern ipaddr_t ip_pkt_source_route_v4(const ip_pkt_t *);
2797 2802 extern in6_addr_t *ip_pkt_source_route_v6(const ip_pkt_t *);
2798 2803 extern int ip_pkt_copy(ip_pkt_t *, ip_pkt_t *, int);
2799 2804 extern void ip_pkt_source_route_reverse_v4(ip_pkt_t *);
2800 2805
2801 2806 /* ipp_fields values */
2802 2807 #define IPPF_ADDR 0x0001 /* Part of in6_pktinfo: src/dst addr */
2803 2808 #define IPPF_HOPLIMIT 0x0002 /* Overrides unicast and multicast */
2804 2809 #define IPPF_TCLASS 0x0004 /* Overrides class in sin6_flowinfo */
2805 2810
2806 2811 #define IPPF_HOPOPTS 0x0010 /* ipp_hopopts set */
2807 2812 #define IPPF_RTHDR 0x0020 /* ipp_rthdr set */
2808 2813 #define IPPF_RTHDRDSTOPTS 0x0040 /* ipp_rthdrdstopts set */
2809 2814 #define IPPF_DSTOPTS 0x0080 /* ipp_dstopts set */
2810 2815
2811 2816 #define IPPF_IPV4_OPTIONS 0x0100 /* ipp_ipv4_options set */
2812 2817 #define IPPF_LABEL_V4 0x0200 /* ipp_label_v4 set */
2813 2818 #define IPPF_LABEL_V6 0x0400 /* ipp_label_v6 set */
2814 2819
2815 2820 #define IPPF_FRAGHDR 0x0800 /* Used for IPsec receive side */
2816 2821
2817 2822 /*
2818 2823 * Data structure which is passed to conn_opt_get/set.
2819 2824 * The conn_t is included even though it can be inferred from queue_t.
2820 2825 * setsockopt and getsockopt use conn_ixa and conn_xmit_ipp. However,
2821 2826 * when handling ancillary data we use separate ixa and ipps.
2822 2827 */
2823 2828 typedef struct conn_opt_arg_s {
2824 2829 conn_t *coa_connp;
2825 2830 ip_xmit_attr_t *coa_ixa;
2826 2831 ip_pkt_t *coa_ipp;
2827 2832 boolean_t coa_ancillary; /* Ancillary data and not setsockopt */
2828 2833 uint_t coa_changed; /* See below */
2829 2834 } conn_opt_arg_t;
2830 2835
2831 2836 /*
2832 2837 * Flags for what changed.
2833 2838 * If we want to be more efficient in the future we can have more fine
2834 2839 * grained flags e.g., a flag for just IP_TOS changing.
2835 2840 * For now we either call ip_set_destination (for "route changed")
2836 2841 * and/or conn_build_hdr_template/conn_prepend_hdr (for "header changed").
2837 2842 */
2838 2843 #define COA_HEADER_CHANGED 0x0001
2839 2844 #define COA_ROUTE_CHANGED 0x0002
2840 2845 #define COA_RCVBUF_CHANGED 0x0004 /* SO_RCVBUF */
2841 2846 #define COA_SNDBUF_CHANGED 0x0008 /* SO_SNDBUF */
2842 2847 #define COA_WROFF_CHANGED 0x0010 /* Header size changed */
2843 2848 #define COA_ICMP_BIND_NEEDED 0x0020
2844 2849 #define COA_OOBINLINE_CHANGED 0x0040
2845 2850
2846 2851 #define TCP_PORTS_OFFSET 0
2847 2852 #define UDP_PORTS_OFFSET 0
2848 2853
2849 2854 /*
2850 2855 * lookups return the ill/ipif only if the flags are clear OR Iam writer.
2851 2856 * ill / ipif lookup functions increment the refcnt on the ill / ipif only
2852 2857 * after calling these macros. This ensures that the refcnt on the ipif or
2853 2858 * ill will eventually drop down to zero.
2854 2859 */
2855 2860 #define ILL_LOOKUP_FAILED 1 /* Used as error code */
2856 2861 #define IPIF_LOOKUP_FAILED 2 /* Used as error code */
2857 2862
2858 2863 #define ILL_CAN_LOOKUP(ill) \
2859 2864 (!((ill)->ill_state_flags & ILL_CONDEMNED) || \
2860 2865 IAM_WRITER_ILL(ill))
2861 2866
2862 2867 #define ILL_IS_CONDEMNED(ill) \
2863 2868 ((ill)->ill_state_flags & ILL_CONDEMNED)
2864 2869
2865 2870 #define IPIF_CAN_LOOKUP(ipif) \
2866 2871 (!((ipif)->ipif_state_flags & IPIF_CONDEMNED) || \
2867 2872 IAM_WRITER_IPIF(ipif))
2868 2873
2869 2874 #define IPIF_IS_CONDEMNED(ipif) \
2870 2875 ((ipif)->ipif_state_flags & IPIF_CONDEMNED)
2871 2876
2872 2877 #define IPIF_IS_CHANGING(ipif) \
2873 2878 ((ipif)->ipif_state_flags & IPIF_CHANGING)
2874 2879
2875 2880 /* Macros used to assert that this thread is a writer */
2876 2881 #define IAM_WRITER_IPSQ(ipsq) ((ipsq)->ipsq_xop->ipx_writer == curthread)
2877 2882 #define IAM_WRITER_ILL(ill) IAM_WRITER_IPSQ((ill)->ill_phyint->phyint_ipsq)
2878 2883 #define IAM_WRITER_IPIF(ipif) IAM_WRITER_ILL((ipif)->ipif_ill)
2879 2884
2880 2885 /*
2881 2886 * Grab ill locks in the proper order. The order is highest addressed
2882 2887 * ill is locked first.
2883 2888 */
2884 2889 #define GRAB_ILL_LOCKS(ill_1, ill_2) \
2885 2890 { \
2886 2891 if ((ill_1) > (ill_2)) { \
2887 2892 if (ill_1 != NULL) \
2888 2893 mutex_enter(&(ill_1)->ill_lock); \
2889 2894 if (ill_2 != NULL) \
2890 2895 mutex_enter(&(ill_2)->ill_lock); \
2891 2896 } else { \
2892 2897 if (ill_2 != NULL) \
2893 2898 mutex_enter(&(ill_2)->ill_lock); \
2894 2899 if (ill_1 != NULL && ill_1 != ill_2) \
2895 2900 mutex_enter(&(ill_1)->ill_lock); \
2896 2901 } \
2897 2902 }
2898 2903
2899 2904 #define RELEASE_ILL_LOCKS(ill_1, ill_2) \
2900 2905 { \
2901 2906 if (ill_1 != NULL) \
2902 2907 mutex_exit(&(ill_1)->ill_lock); \
2903 2908 if (ill_2 != NULL && ill_2 != ill_1) \
2904 2909 mutex_exit(&(ill_2)->ill_lock); \
2905 2910 }
2906 2911
2907 2912 /* Get the other protocol instance ill */
2908 2913 #define ILL_OTHER(ill) \
2909 2914 ((ill)->ill_isv6 ? (ill)->ill_phyint->phyint_illv4 : \
2910 2915 (ill)->ill_phyint->phyint_illv6)
2911 2916
2912 2917 /* ioctl command info: Ioctl properties extracted and stored in here */
2913 2918 typedef struct cmd_info_s
2914 2919 {
2915 2920 ipif_t *ci_ipif; /* ipif associated with [l]ifreq ioctl's */
2916 2921 sin_t *ci_sin; /* the sin struct passed down */
2917 2922 sin6_t *ci_sin6; /* the sin6_t struct passed down */
2918 2923 struct lifreq *ci_lifr; /* the lifreq struct passed down */
2919 2924 } cmd_info_t;
2920 2925
2921 2926 extern struct kmem_cache *ire_cache;
2922 2927
2923 2928 extern ipaddr_t ip_g_all_ones;
2924 2929
2925 2930 extern uint_t ip_loopback_mtu; /* /etc/system */
2926 2931 extern uint_t ip_loopback_mtuplus;
2927 2932 extern uint_t ip_loopback_mtu_v6plus;
2928 2933
2929 2934 extern vmem_t *ip_minor_arena_sa;
2930 2935 extern vmem_t *ip_minor_arena_la;
2931 2936
2932 2937 /*
2933 2938 * ip_g_forward controls IP forwarding. It takes two values:
2934 2939 * 0: IP_FORWARD_NEVER Don't forward packets ever.
2935 2940 * 1: IP_FORWARD_ALWAYS Forward packets for elsewhere.
2936 2941 *
2937 2942 * RFC1122 says there must be a configuration switch to control forwarding,
2938 2943 * but that the default MUST be to not forward packets ever. Implicit
2939 2944 * control based on configuration of multiple interfaces MUST NOT be
2940 2945 * implemented (Section 3.1). SunOS 4.1 did provide the "automatic" capability
2941 2946 * and, in fact, it was the default. That capability is now provided in the
2942 2947 * /etc/rc2.d/S69inet script.
2943 2948 */
2944 2949
2945 2950 #define ips_ip_respond_to_address_mask_broadcast \
2946 2951 ips_propinfo_tbl[0].prop_cur_bval
2947 2952 #define ips_ip_g_resp_to_echo_bcast ips_propinfo_tbl[1].prop_cur_bval
2948 2953 #define ips_ip_g_resp_to_echo_mcast ips_propinfo_tbl[2].prop_cur_bval
2949 2954 #define ips_ip_g_resp_to_timestamp ips_propinfo_tbl[3].prop_cur_bval
2950 2955 #define ips_ip_g_resp_to_timestamp_bcast ips_propinfo_tbl[4].prop_cur_bval
2951 2956 #define ips_ip_g_send_redirects ips_propinfo_tbl[5].prop_cur_bval
2952 2957 #define ips_ip_g_forward_directed_bcast ips_propinfo_tbl[6].prop_cur_bval
2953 2958 #define ips_ip_mrtdebug ips_propinfo_tbl[7].prop_cur_uval
2954 2959 #define ips_ip_ire_reclaim_fraction ips_propinfo_tbl[8].prop_cur_uval
2955 2960 #define ips_ip_nce_reclaim_fraction ips_propinfo_tbl[9].prop_cur_uval
2956 2961 #define ips_ip_dce_reclaim_fraction ips_propinfo_tbl[10].prop_cur_uval
2957 2962 #define ips_ip_def_ttl ips_propinfo_tbl[11].prop_cur_uval
2958 2963 #define ips_ip_forward_src_routed ips_propinfo_tbl[12].prop_cur_bval
2959 2964 #define ips_ip_wroff_extra ips_propinfo_tbl[13].prop_cur_uval
2960 2965 #define ips_ip_pathmtu_interval ips_propinfo_tbl[14].prop_cur_uval
2961 2966 #define ips_ip_icmp_return ips_propinfo_tbl[15].prop_cur_uval
2962 2967 #define ips_ip_path_mtu_discovery ips_propinfo_tbl[16].prop_cur_bval
2963 2968 #define ips_ip_pmtu_min ips_propinfo_tbl[17].prop_cur_uval
2964 2969 #define ips_ip_ignore_redirect ips_propinfo_tbl[18].prop_cur_bval
2965 2970 #define ips_ip_arp_icmp_error ips_propinfo_tbl[19].prop_cur_bval
2966 2971 #define ips_ip_broadcast_ttl ips_propinfo_tbl[20].prop_cur_uval
2967 2972 #define ips_ip_icmp_err_interval ips_propinfo_tbl[21].prop_cur_uval
2968 2973 #define ips_ip_icmp_err_burst ips_propinfo_tbl[22].prop_cur_uval
2969 2974 #define ips_ip_reass_queue_bytes ips_propinfo_tbl[23].prop_cur_uval
2970 2975 #define ips_ip_strict_dst_multihoming ips_propinfo_tbl[24].prop_cur_uval
2971 2976 #define ips_ip_addrs_per_if ips_propinfo_tbl[25].prop_cur_uval
2972 2977 #define ips_ipsec_override_persocket_policy ips_propinfo_tbl[26].prop_cur_bval
2973 2978 #define ips_icmp_accept_clear_messages ips_propinfo_tbl[27].prop_cur_bval
2974 2979 #define ips_igmp_accept_clear_messages ips_propinfo_tbl[28].prop_cur_bval
2975 2980
2976 2981 /* IPv6 configuration knobs */
2977 2982 #define ips_delay_first_probe_time ips_propinfo_tbl[29].prop_cur_uval
2978 2983 #define ips_max_unicast_solicit ips_propinfo_tbl[30].prop_cur_uval
2979 2984 #define ips_ipv6_def_hops ips_propinfo_tbl[31].prop_cur_uval
2980 2985 #define ips_ipv6_icmp_return ips_propinfo_tbl[32].prop_cur_uval
2981 2986 #define ips_ipv6_forward_src_routed ips_propinfo_tbl[33].prop_cur_bval
2982 2987 #define ips_ipv6_resp_echo_mcast ips_propinfo_tbl[34].prop_cur_bval
2983 2988 #define ips_ipv6_send_redirects ips_propinfo_tbl[35].prop_cur_bval
2984 2989 #define ips_ipv6_ignore_redirect ips_propinfo_tbl[36].prop_cur_bval
2985 2990 #define ips_ipv6_strict_dst_multihoming ips_propinfo_tbl[37].prop_cur_uval
2986 2991 #define ips_src_check ips_propinfo_tbl[38].prop_cur_uval
2987 2992 #define ips_ipsec_policy_log_interval ips_propinfo_tbl[39].prop_cur_uval
2988 2993 #define ips_pim_accept_clear_messages ips_propinfo_tbl[40].prop_cur_bval
2989 2994 #define ips_ip_ndp_unsolicit_interval ips_propinfo_tbl[41].prop_cur_uval
2990 2995 #define ips_ip_ndp_unsolicit_count ips_propinfo_tbl[42].prop_cur_uval
2991 2996 #define ips_ipv6_ignore_home_address_opt ips_propinfo_tbl[43].prop_cur_bval
2992 2997
2993 2998 /* Misc IP configuration knobs */
2994 2999 #define ips_ip_policy_mask ips_propinfo_tbl[44].prop_cur_uval
2995 3000 #define ips_ip_ecmp_behavior ips_propinfo_tbl[45].prop_cur_uval
2996 3001 #define ips_ip_multirt_ttl ips_propinfo_tbl[46].prop_cur_uval
2997 3002 #define ips_ip_ire_badcnt_lifetime ips_propinfo_tbl[47].prop_cur_uval
2998 3003 #define ips_ip_max_temp_idle ips_propinfo_tbl[48].prop_cur_uval
2999 3004 #define ips_ip_max_temp_defend ips_propinfo_tbl[49].prop_cur_uval
3000 3005 #define ips_ip_max_defend ips_propinfo_tbl[50].prop_cur_uval
3001 3006 #define ips_ip_defend_interval ips_propinfo_tbl[51].prop_cur_uval
3002 3007 #define ips_ip_dup_recovery ips_propinfo_tbl[52].prop_cur_uval
3003 3008 #define ips_ip_restrict_interzone_loopback ips_propinfo_tbl[53].prop_cur_bval
3004 3009 #define ips_ip_lso_outbound ips_propinfo_tbl[54].prop_cur_bval
3005 3010 #define ips_igmp_max_version ips_propinfo_tbl[55].prop_cur_uval
3006 3011 #define ips_mld_max_version ips_propinfo_tbl[56].prop_cur_uval
3007 3012 #define ips_ip_forwarding ips_propinfo_tbl[57].prop_cur_bval
3008 3013 #define ips_ipv6_forwarding ips_propinfo_tbl[58].prop_cur_bval
3009 3014 #define ips_ip_reassembly_timeout ips_propinfo_tbl[59].prop_cur_uval
3010 3015 #define ips_ipv6_reassembly_timeout ips_propinfo_tbl[60].prop_cur_uval
3011 3016 #define ips_ip_cgtp_filter ips_propinfo_tbl[61].prop_cur_bval
3012 3017 #define ips_arp_probe_delay ips_propinfo_tbl[62].prop_cur_uval
3013 3018 #define ips_arp_fastprobe_delay ips_propinfo_tbl[63].prop_cur_uval
3014 3019 #define ips_arp_probe_interval ips_propinfo_tbl[64].prop_cur_uval
3015 3020 #define ips_arp_fastprobe_interval ips_propinfo_tbl[65].prop_cur_uval
3016 3021 #define ips_arp_probe_count ips_propinfo_tbl[66].prop_cur_uval
3017 3022 #define ips_arp_fastprobe_count ips_propinfo_tbl[67].prop_cur_uval
3018 3023 #define ips_ipv4_dad_announce_interval ips_propinfo_tbl[68].prop_cur_uval
3019 3024 #define ips_ipv6_dad_announce_interval ips_propinfo_tbl[69].prop_cur_uval
3020 3025 #define ips_arp_defend_interval ips_propinfo_tbl[70].prop_cur_uval
3021 3026 #define ips_arp_defend_rate ips_propinfo_tbl[71].prop_cur_uval
3022 3027 #define ips_ndp_defend_interval ips_propinfo_tbl[72].prop_cur_uval
3023 3028 #define ips_ndp_defend_rate ips_propinfo_tbl[73].prop_cur_uval
3024 3029 #define ips_arp_defend_period ips_propinfo_tbl[74].prop_cur_uval
3025 3030 #define ips_ndp_defend_period ips_propinfo_tbl[75].prop_cur_uval
3026 3031 #define ips_ipv4_icmp_return_pmtu ips_propinfo_tbl[76].prop_cur_bval
3027 3032 #define ips_ipv6_icmp_return_pmtu ips_propinfo_tbl[77].prop_cur_bval
3028 3033 #define ips_ip_arp_publish_count ips_propinfo_tbl[78].prop_cur_uval
3029 3034 #define ips_ip_arp_publish_interval ips_propinfo_tbl[79].prop_cur_uval
3030 3035 #define ips_ip_strict_src_multihoming ips_propinfo_tbl[80].prop_cur_uval
3031 3036 #define ips_ipv6_strict_src_multihoming ips_propinfo_tbl[81].prop_cur_uval
3032 3037 #define ips_ipv6_drop_inbound_icmpv6 ips_propinfo_tbl[82].prop_cur_bval
3033 3038
3034 3039 extern int dohwcksum; /* use h/w cksum if supported by the h/w */
3035 3040 #ifdef ZC_TEST
3036 3041 extern int noswcksum;
3037 3042 #endif
3038 3043
3039 3044 extern char ipif_loopback_name[];
3040 3045
3041 3046 extern nv_t *ire_nv_tbl;
3042 3047
3043 3048 extern struct module_info ip_mod_info;
3044 3049
3045 3050 #define HOOKS4_INTERESTED_PHYSICAL_IN(ipst) \
3046 3051 ((ipst)->ips_ip4_physical_in_event.he_interested)
3047 3052 #define HOOKS6_INTERESTED_PHYSICAL_IN(ipst) \
3048 3053 ((ipst)->ips_ip6_physical_in_event.he_interested)
3049 3054 #define HOOKS4_INTERESTED_PHYSICAL_OUT(ipst) \
3050 3055 ((ipst)->ips_ip4_physical_out_event.he_interested)
3051 3056 #define HOOKS6_INTERESTED_PHYSICAL_OUT(ipst) \
3052 3057 ((ipst)->ips_ip6_physical_out_event.he_interested)
3053 3058 #define HOOKS4_INTERESTED_FORWARDING(ipst) \
3054 3059 ((ipst)->ips_ip4_forwarding_event.he_interested)
3055 3060 #define HOOKS6_INTERESTED_FORWARDING(ipst) \
3056 3061 ((ipst)->ips_ip6_forwarding_event.he_interested)
3057 3062 #define HOOKS4_INTERESTED_LOOPBACK_IN(ipst) \
3058 3063 ((ipst)->ips_ip4_loopback_in_event.he_interested)
3059 3064 #define HOOKS6_INTERESTED_LOOPBACK_IN(ipst) \
3060 3065 ((ipst)->ips_ip6_loopback_in_event.he_interested)
3061 3066 #define HOOKS4_INTERESTED_LOOPBACK_OUT(ipst) \
3062 3067 ((ipst)->ips_ip4_loopback_out_event.he_interested)
3063 3068 #define HOOKS6_INTERESTED_LOOPBACK_OUT(ipst) \
3064 3069 ((ipst)->ips_ip6_loopback_out_event.he_interested)
3065 3070 /*
3066 3071 * Hooks marcos used inside of ip
3067 3072 * The callers use the above INTERESTED macros first, hence
3068 3073 * the he_interested check is superflous.
3069 3074 */
3070 3075 #define FW_HOOKS(_hook, _event, _ilp, _olp, _iph, _fm, _m, _llm, ipst, _err) \
3071 3076 if ((_hook).he_interested) { \
3072 3077 hook_pkt_event_t info; \
3073 3078 \
3074 3079 _NOTE(CONSTCOND) \
3075 3080 ASSERT((_ilp != NULL) || (_olp != NULL)); \
3076 3081 \
3077 3082 FW_SET_ILL_INDEX(info.hpe_ifp, (ill_t *)_ilp); \
3078 3083 FW_SET_ILL_INDEX(info.hpe_ofp, (ill_t *)_olp); \
3079 3084 info.hpe_protocol = ipst->ips_ipv4_net_data; \
3080 3085 info.hpe_hdr = _iph; \
3081 3086 info.hpe_mp = &(_fm); \
3082 3087 info.hpe_mb = _m; \
3083 3088 info.hpe_flags = _llm; \
3084 3089 _err = hook_run(ipst->ips_ipv4_net_data->netd_hooks, \
3085 3090 _event, (hook_data_t)&info); \
3086 3091 if (_err != 0) { \
3087 3092 ip2dbg(("%s hook dropped mblk chain %p hdr %p\n",\
3088 3093 (_hook).he_name, (void *)_fm, (void *)_m)); \
3089 3094 if (_fm != NULL) { \
3090 3095 freemsg(_fm); \
3091 3096 _fm = NULL; \
3092 3097 } \
3093 3098 _iph = NULL; \
3094 3099 _m = NULL; \
3095 3100 } else { \
3096 3101 _iph = info.hpe_hdr; \
3097 3102 _m = info.hpe_mb; \
3098 3103 } \
3099 3104 }
3100 3105
3101 3106 #define FW_HOOKS6(_hook, _event, _ilp, _olp, _iph, _fm, _m, _llm, ipst, _err) \
3102 3107 if ((_hook).he_interested) { \
3103 3108 hook_pkt_event_t info; \
3104 3109 \
3105 3110 _NOTE(CONSTCOND) \
3106 3111 ASSERT((_ilp != NULL) || (_olp != NULL)); \
3107 3112 \
3108 3113 FW_SET_ILL_INDEX(info.hpe_ifp, (ill_t *)_ilp); \
3109 3114 FW_SET_ILL_INDEX(info.hpe_ofp, (ill_t *)_olp); \
3110 3115 info.hpe_protocol = ipst->ips_ipv6_net_data; \
3111 3116 info.hpe_hdr = _iph; \
3112 3117 info.hpe_mp = &(_fm); \
3113 3118 info.hpe_mb = _m; \
3114 3119 info.hpe_flags = _llm; \
3115 3120 _err = hook_run(ipst->ips_ipv6_net_data->netd_hooks, \
3116 3121 _event, (hook_data_t)&info); \
3117 3122 if (_err != 0) { \
3118 3123 ip2dbg(("%s hook dropped mblk chain %p hdr %p\n",\
3119 3124 (_hook).he_name, (void *)_fm, (void *)_m)); \
3120 3125 if (_fm != NULL) { \
3121 3126 freemsg(_fm); \
3122 3127 _fm = NULL; \
3123 3128 } \
3124 3129 _iph = NULL; \
3125 3130 _m = NULL; \
3126 3131 } else { \
3127 3132 _iph = info.hpe_hdr; \
3128 3133 _m = info.hpe_mb; \
3129 3134 } \
3130 3135 }
3131 3136
3132 3137 #define FW_SET_ILL_INDEX(fp, ill) \
3133 3138 _NOTE(CONSTCOND) \
3134 3139 if ((ill) == NULL || (ill)->ill_phyint == NULL) { \
3135 3140 (fp) = 0; \
3136 3141 _NOTE(CONSTCOND) \
3137 3142 } else if (IS_UNDER_IPMP(ill)) { \
3138 3143 (fp) = ipmp_ill_get_ipmp_ifindex(ill); \
3139 3144 } else { \
3140 3145 (fp) = (ill)->ill_phyint->phyint_ifindex; \
3141 3146 }
3142 3147
3143 3148 /*
3144 3149 * Network byte order macros
3145 3150 */
3146 3151 #ifdef _BIG_ENDIAN
3147 3152 #define N_IN_CLASSA_NET IN_CLASSA_NET
3148 3153 #define N_IN_CLASSD_NET IN_CLASSD_NET
3149 3154 #define N_INADDR_UNSPEC_GROUP INADDR_UNSPEC_GROUP
3150 3155 #define N_IN_LOOPBACK_NET (ipaddr_t)0x7f000000U
3151 3156 #else /* _BIG_ENDIAN */
3152 3157 #define N_IN_CLASSA_NET (ipaddr_t)0x000000ffU
3153 3158 #define N_IN_CLASSD_NET (ipaddr_t)0x000000f0U
3154 3159 #define N_INADDR_UNSPEC_GROUP (ipaddr_t)0x000000e0U
3155 3160 #define N_IN_LOOPBACK_NET (ipaddr_t)0x0000007fU
3156 3161 #endif /* _BIG_ENDIAN */
3157 3162 #define CLASSD(addr) (((addr) & N_IN_CLASSD_NET) == N_INADDR_UNSPEC_GROUP)
3158 3163 #define CLASSE(addr) (((addr) & N_IN_CLASSD_NET) == N_IN_CLASSD_NET)
3159 3164 #define IP_LOOPBACK_ADDR(addr) \
3160 3165 (((addr) & N_IN_CLASSA_NET == N_IN_LOOPBACK_NET))
3161 3166
3162 3167 extern int ip_debug;
3163 3168 extern uint_t ip_thread_data;
3164 3169 extern krwlock_t ip_thread_rwlock;
3165 3170 extern list_t ip_thread_list;
3166 3171
3167 3172 #ifdef IP_DEBUG
3168 3173 #include <sys/debug.h>
3169 3174 #include <sys/promif.h>
3170 3175
3171 3176 #define ip0dbg(a) printf a
3172 3177 #define ip1dbg(a) if (ip_debug > 2) printf a
3173 3178 #define ip2dbg(a) if (ip_debug > 3) printf a
3174 3179 #define ip3dbg(a) if (ip_debug > 4) printf a
3175 3180 #else
3176 3181 #define ip0dbg(a) /* */
3177 3182 #define ip1dbg(a) /* */
3178 3183 #define ip2dbg(a) /* */
3179 3184 #define ip3dbg(a) /* */
3180 3185 #endif /* IP_DEBUG */
3181 3186
3182 3187 /* Default MAC-layer address string length for mac_colon_addr */
3183 3188 #define MAC_STR_LEN 128
3184 3189
3185 3190 struct mac_header_info_s;
3186 3191
3187 3192 extern void ill_frag_timer(void *);
3188 3193 extern ill_t *ill_first(int, int, ill_walk_context_t *, ip_stack_t *);
3189 3194 extern ill_t *ill_next(ill_walk_context_t *, ill_t *);
3190 3195 extern void ill_frag_timer_start(ill_t *);
3191 3196 extern void ill_nic_event_dispatch(ill_t *, lif_if_t, nic_event_t,
3192 3197 nic_event_data_t, size_t);
3193 3198 extern mblk_t *ip_carve_mp(mblk_t **, ssize_t);
3194 3199 extern mblk_t *ip_dlpi_alloc(size_t, t_uscalar_t);
3195 3200 extern mblk_t *ip_dlnotify_alloc(uint_t, uint_t);
3196 3201 extern mblk_t *ip_dlnotify_alloc2(uint_t, uint_t, uint_t);
3197 3202 extern char *ip_dot_addr(ipaddr_t, char *);
3198 3203 extern const char *mac_colon_addr(const uint8_t *, size_t, char *, size_t);
3199 3204 extern void ip_lwput(queue_t *, mblk_t *);
3200 3205 extern boolean_t icmp_err_rate_limit(ip_stack_t *);
3201 3206 extern void icmp_frag_needed(mblk_t *, int, ip_recv_attr_t *);
3202 3207 extern mblk_t *icmp_inbound_v4(mblk_t *, ip_recv_attr_t *);
3203 3208 extern void icmp_time_exceeded(mblk_t *, uint8_t, ip_recv_attr_t *);
3204 3209 extern void icmp_unreachable(mblk_t *, uint8_t, ip_recv_attr_t *);
3205 3210 extern boolean_t ip_ipsec_policy_inherit(conn_t *, conn_t *, ip_recv_attr_t *);
3206 3211 extern void *ip_pullup(mblk_t *, ssize_t, ip_recv_attr_t *);
3207 3212 extern void ip_setl2src(mblk_t *, ip_recv_attr_t *, ill_t *);
3208 3213 extern mblk_t *ip_check_and_align_header(mblk_t *, uint_t, ip_recv_attr_t *);
3209 3214 extern mblk_t *ip_check_length(mblk_t *, uchar_t *, ssize_t, uint_t, uint_t,
3210 3215 ip_recv_attr_t *);
3211 3216 extern mblk_t *ip_check_optlen(mblk_t *, ipha_t *, uint_t, uint_t,
3212 3217 ip_recv_attr_t *);
3213 3218 extern mblk_t *ip_fix_dbref(mblk_t *, ip_recv_attr_t *);
3214 3219 extern uint_t ip_cksum(mblk_t *, int, uint32_t);
3215 3220 extern int ip_close(queue_t *, int);
3216 3221 extern uint16_t ip_csum_hdr(ipha_t *);
3217 3222 extern void ip_forward_xmit_v4(nce_t *, ill_t *, mblk_t *, ipha_t *,
3218 3223 ip_recv_attr_t *, uint32_t, uint32_t);
3219 3224 extern boolean_t ip_forward_options(mblk_t *, ipha_t *, ill_t *,
3220 3225 ip_recv_attr_t *);
3221 3226 extern int ip_fragment_v4(mblk_t *, nce_t *, iaflags_t, uint_t, uint32_t,
3222 3227 uint32_t, zoneid_t, zoneid_t, pfirepostfrag_t postfragfn,
3223 3228 uintptr_t *cookie);
3224 3229 extern void ip_proto_not_sup(mblk_t *, ip_recv_attr_t *);
3225 3230 extern void ip_ire_g_fini(void);
3226 3231 extern void ip_ire_g_init(void);
3227 3232 extern void ip_ire_fini(ip_stack_t *);
3228 3233 extern void ip_ire_init(ip_stack_t *);
3229 3234 extern void ip_mdata_to_mhi(ill_t *, mblk_t *, struct mac_header_info_s *);
3230 3235 extern int ip_openv4(queue_t *q, dev_t *devp, int flag, int sflag,
3231 3236 cred_t *credp);
3232 3237 extern int ip_openv6(queue_t *q, dev_t *devp, int flag, int sflag,
3233 3238 cred_t *credp);
3234 3239 extern int ip_reassemble(mblk_t *, ipf_t *, uint_t, boolean_t, ill_t *,
3235 3240 size_t);
3236 3241 extern void ip_rput(queue_t *, mblk_t *);
3237 3242 extern void ip_input(ill_t *, ill_rx_ring_t *, mblk_t *,
3238 3243 struct mac_header_info_s *);
3239 3244 extern void ip_input_v6(ill_t *, ill_rx_ring_t *, mblk_t *,
3240 3245 struct mac_header_info_s *);
3241 3246 extern mblk_t *ip_input_common_v4(ill_t *, ill_rx_ring_t *, mblk_t *,
3242 3247 struct mac_header_info_s *, squeue_t *, mblk_t **, uint_t *);
3243 3248 extern mblk_t *ip_input_common_v6(ill_t *, ill_rx_ring_t *, mblk_t *,
3244 3249 struct mac_header_info_s *, squeue_t *, mblk_t **, uint_t *);
3245 3250 extern void ill_input_full_v4(mblk_t *, void *, void *,
3246 3251 ip_recv_attr_t *, rtc_t *);
3247 3252 extern void ill_input_short_v4(mblk_t *, void *, void *,
3248 3253 ip_recv_attr_t *, rtc_t *);
3249 3254 extern void ill_input_full_v6(mblk_t *, void *, void *,
3250 3255 ip_recv_attr_t *, rtc_t *);
3251 3256 extern void ill_input_short_v6(mblk_t *, void *, void *,
3252 3257 ip_recv_attr_t *, rtc_t *);
3253 3258 extern ipaddr_t ip_input_options(ipha_t *, ipaddr_t, mblk_t *,
3254 3259 ip_recv_attr_t *, int *);
3255 3260 extern boolean_t ip_input_local_options(mblk_t *, ipha_t *, ip_recv_attr_t *);
3256 3261 extern mblk_t *ip_input_fragment(mblk_t *, ipha_t *, ip_recv_attr_t *);
3257 3262 extern mblk_t *ip_input_fragment_v6(mblk_t *, ip6_t *, ip6_frag_t *, uint_t,
3258 3263 ip_recv_attr_t *);
3259 3264 extern void ip_input_post_ipsec(mblk_t *, ip_recv_attr_t *);
3260 3265 extern void ip_fanout_v4(mblk_t *, ipha_t *, ip_recv_attr_t *);
3261 3266 extern void ip_fanout_v6(mblk_t *, ip6_t *, ip_recv_attr_t *);
3262 3267 extern void ip_fanout_proto_conn(conn_t *, mblk_t *, ipha_t *, ip6_t *,
3263 3268 ip_recv_attr_t *);
3264 3269 extern void ip_fanout_proto_v4(mblk_t *, ipha_t *, ip_recv_attr_t *);
3265 3270 extern void ip_fanout_send_icmp_v4(mblk_t *, uint_t, uint_t,
3266 3271 ip_recv_attr_t *);
3267 3272 extern void ip_fanout_udp_conn(conn_t *, mblk_t *, ipha_t *, ip6_t *,
3268 3273 ip_recv_attr_t *);
3269 3274 extern void ip_fanout_udp_multi_v4(mblk_t *, ipha_t *, uint16_t, uint16_t,
3270 3275 ip_recv_attr_t *);
3271 3276 extern mblk_t *zero_spi_check(mblk_t *, ip_recv_attr_t *);
3272 3277 extern void ip_build_hdrs_v4(uchar_t *, uint_t, const ip_pkt_t *, uint8_t);
3273 3278 extern int ip_find_hdr_v4(ipha_t *, ip_pkt_t *, boolean_t);
3274 3279 extern int ip_total_hdrs_len_v4(const ip_pkt_t *);
3275 3280
3276 3281 extern mblk_t *ip_accept_tcp(ill_t *, ill_rx_ring_t *, squeue_t *,
3277 3282 mblk_t *, mblk_t **, uint_t *cnt);
3278 3283 extern void ip_rput_dlpi(ill_t *, mblk_t *);
3279 3284 extern void ip_rput_notdata(ill_t *, mblk_t *);
3280 3285
3281 3286 extern void ip_mib2_add_ip_stats(mib2_ipIfStatsEntry_t *,
3282 3287 mib2_ipIfStatsEntry_t *);
3283 3288 extern void ip_mib2_add_icmp6_stats(mib2_ipv6IfIcmpEntry_t *,
3284 3289 mib2_ipv6IfIcmpEntry_t *);
3285 3290 extern void ip_rput_other(ipsq_t *, queue_t *, mblk_t *, void *);
3286 3291 extern ire_t *ip_check_multihome(void *, ire_t *, ill_t *);
3287 3292 extern void ip_send_potential_redirect_v4(mblk_t *, ipha_t *, ire_t *,
3288 3293 ip_recv_attr_t *);
3289 3294 extern int ip_set_destination_v4(ipaddr_t *, ipaddr_t, ipaddr_t,
3290 3295 ip_xmit_attr_t *, iulp_t *, uint32_t, uint_t);
3291 3296 extern int ip_set_destination_v6(in6_addr_t *, const in6_addr_t *,
3292 3297 const in6_addr_t *, ip_xmit_attr_t *, iulp_t *, uint32_t, uint_t);
3293 3298
3294 3299 extern int ip_output_simple(mblk_t *, ip_xmit_attr_t *);
3295 3300 extern int ip_output_simple_v4(mblk_t *, ip_xmit_attr_t *);
3296 3301 extern int ip_output_simple_v6(mblk_t *, ip_xmit_attr_t *);
3297 3302 extern int ip_output_options(mblk_t *, ipha_t *, ip_xmit_attr_t *,
3298 3303 ill_t *);
3299 3304 extern void ip_output_local_options(ipha_t *, ip_stack_t *);
3300 3305
3301 3306 extern ip_xmit_attr_t *conn_get_ixa(conn_t *, boolean_t);
3302 3307 extern ip_xmit_attr_t *conn_get_ixa_tryhard(conn_t *, boolean_t);
3303 3308 extern ip_xmit_attr_t *conn_replace_ixa(conn_t *, ip_xmit_attr_t *);
3304 3309 extern ip_xmit_attr_t *conn_get_ixa_exclusive(conn_t *);
3305 3310 extern ip_xmit_attr_t *ip_xmit_attr_duplicate(ip_xmit_attr_t *);
3306 3311 extern void ip_xmit_attr_replace_tsl(ip_xmit_attr_t *, ts_label_t *);
3307 3312 extern void ip_xmit_attr_restore_tsl(ip_xmit_attr_t *, cred_t *);
3308 3313 boolean_t ip_recv_attr_replace_label(ip_recv_attr_t *, ts_label_t *);
3309 3314 extern void ixa_inactive(ip_xmit_attr_t *);
3310 3315 extern void ixa_refrele(ip_xmit_attr_t *);
3311 3316 extern boolean_t ixa_check_drain_insert(conn_t *, ip_xmit_attr_t *);
3312 3317 extern void ixa_cleanup(ip_xmit_attr_t *);
3313 3318 extern void ira_cleanup(ip_recv_attr_t *, boolean_t);
3314 3319 extern void ixa_safe_copy(ip_xmit_attr_t *, ip_xmit_attr_t *);
3315 3320
3316 3321 extern int conn_ip_output(mblk_t *, ip_xmit_attr_t *);
3317 3322 extern boolean_t ip_output_verify_local(ip_xmit_attr_t *);
3318 3323 extern mblk_t *ip_output_process_local(mblk_t *, ip_xmit_attr_t *, boolean_t,
3319 3324 boolean_t, conn_t *);
3320 3325
3321 3326 extern int conn_opt_get(conn_opt_arg_t *, t_scalar_t, t_scalar_t,
3322 3327 uchar_t *);
3323 3328 extern int conn_opt_set(conn_opt_arg_t *, t_scalar_t, t_scalar_t, uint_t,
3324 3329 uchar_t *, boolean_t, cred_t *);
3325 3330 extern boolean_t conn_same_as_last_v4(conn_t *, sin_t *);
3326 3331 extern boolean_t conn_same_as_last_v6(conn_t *, sin6_t *);
3327 3332 extern int conn_update_label(const conn_t *, const ip_xmit_attr_t *,
3328 3333 const in6_addr_t *, ip_pkt_t *);
3329 3334
3330 3335 extern int ip_opt_set_multicast_group(conn_t *, t_scalar_t,
3331 3336 uchar_t *, boolean_t, boolean_t);
3332 3337 extern int ip_opt_set_multicast_sources(conn_t *, t_scalar_t,
3333 3338 uchar_t *, boolean_t, boolean_t);
3334 3339 extern int conn_getsockname(conn_t *, struct sockaddr *, uint_t *);
3335 3340 extern int conn_getpeername(conn_t *, struct sockaddr *, uint_t *);
3336 3341
3337 3342 extern int conn_build_hdr_template(conn_t *, uint_t, uint_t,
3338 3343 const in6_addr_t *, const in6_addr_t *, uint32_t);
3339 3344 extern mblk_t *conn_prepend_hdr(ip_xmit_attr_t *, const ip_pkt_t *,
3340 3345 const in6_addr_t *, const in6_addr_t *, uint8_t, uint32_t, uint_t,
3341 3346 mblk_t *, uint_t, uint_t, uint32_t *, int *);
3342 3347 extern void ip_attr_newdst(ip_xmit_attr_t *);
3343 3348 extern void ip_attr_nexthop(const ip_pkt_t *, const ip_xmit_attr_t *,
3344 3349 const in6_addr_t *, in6_addr_t *);
3345 3350 extern int conn_connect(conn_t *, iulp_t *, uint32_t);
3346 3351 extern int ip_attr_connect(const conn_t *, ip_xmit_attr_t *,
3347 3352 const in6_addr_t *, const in6_addr_t *, const in6_addr_t *, in_port_t,
3348 3353 in6_addr_t *, iulp_t *, uint32_t);
3349 3354 extern int conn_inherit_parent(conn_t *, conn_t *);
3350 3355
3351 3356 extern void conn_ixa_cleanup(conn_t *connp, void *arg);
3352 3357
3353 3358 extern boolean_t conn_wantpacket(conn_t *, ip_recv_attr_t *, ipha_t *);
3354 3359 extern uint_t ip_type_v4(ipaddr_t, ip_stack_t *);
3355 3360 extern uint_t ip_type_v6(const in6_addr_t *, ip_stack_t *);
3356 3361
3357 3362 extern void ip_wput_nondata(queue_t *, mblk_t *);
3358 3363 extern void ip_wsrv(queue_t *);
3359 3364 extern char *ip_nv_lookup(nv_t *, int);
3360 3365 extern boolean_t ip_local_addr_ok_v6(const in6_addr_t *, const in6_addr_t *);
3361 3366 extern boolean_t ip_remote_addr_ok_v6(const in6_addr_t *, const in6_addr_t *);
3362 3367 extern ipaddr_t ip_massage_options(ipha_t *, netstack_t *);
3363 3368 extern ipaddr_t ip_net_mask(ipaddr_t);
3364 3369 extern void arp_bringup_done(ill_t *, int);
3365 3370 extern void arp_replumb_done(ill_t *, int);
3366 3371
3367 3372 extern struct qinit iprinitv6;
3368 3373
3369 3374 extern void ipmp_init(ip_stack_t *);
3370 3375 extern void ipmp_destroy(ip_stack_t *);
3371 3376 extern ipmp_grp_t *ipmp_grp_create(const char *, phyint_t *);
3372 3377 extern void ipmp_grp_destroy(ipmp_grp_t *);
3373 3378 extern void ipmp_grp_info(const ipmp_grp_t *, lifgroupinfo_t *);
3374 3379 extern int ipmp_grp_rename(ipmp_grp_t *, const char *);
3375 3380 extern ipmp_grp_t *ipmp_grp_lookup(const char *, ip_stack_t *);
3376 3381 extern int ipmp_grp_vet_phyint(ipmp_grp_t *, phyint_t *);
3377 3382 extern ipmp_illgrp_t *ipmp_illgrp_create(ill_t *);
3378 3383 extern void ipmp_illgrp_destroy(ipmp_illgrp_t *);
3379 3384 extern ill_t *ipmp_illgrp_add_ipif(ipmp_illgrp_t *, ipif_t *);
3380 3385 extern void ipmp_illgrp_del_ipif(ipmp_illgrp_t *, ipif_t *);
3381 3386 extern ill_t *ipmp_illgrp_next_ill(ipmp_illgrp_t *);
3382 3387 extern ill_t *ipmp_illgrp_hold_next_ill(ipmp_illgrp_t *);
3383 3388 extern ill_t *ipmp_illgrp_hold_cast_ill(ipmp_illgrp_t *);
3384 3389 extern ill_t *ipmp_illgrp_ipmp_ill(ipmp_illgrp_t *);
3385 3390 extern void ipmp_illgrp_refresh_mtu(ipmp_illgrp_t *);
3386 3391 extern ipmp_arpent_t *ipmp_illgrp_create_arpent(ipmp_illgrp_t *,
3387 3392 boolean_t, ipaddr_t, uchar_t *, size_t, uint16_t);
3388 3393 extern void ipmp_illgrp_destroy_arpent(ipmp_illgrp_t *, ipmp_arpent_t *);
3389 3394 extern ipmp_arpent_t *ipmp_illgrp_lookup_arpent(ipmp_illgrp_t *, ipaddr_t *);
3390 3395 extern void ipmp_illgrp_refresh_arpent(ipmp_illgrp_t *);
3391 3396 extern void ipmp_illgrp_mark_arpent(ipmp_illgrp_t *, ipmp_arpent_t *);
3392 3397 extern ill_t *ipmp_illgrp_find_ill(ipmp_illgrp_t *, uchar_t *, uint_t);
3393 3398 extern void ipmp_illgrp_link_grp(ipmp_illgrp_t *, ipmp_grp_t *);
3394 3399 extern int ipmp_illgrp_unlink_grp(ipmp_illgrp_t *);
3395 3400 extern uint_t ipmp_ill_get_ipmp_ifindex(const ill_t *);
3396 3401 extern void ipmp_ill_join_illgrp(ill_t *, ipmp_illgrp_t *);
3397 3402 extern void ipmp_ill_leave_illgrp(ill_t *);
3398 3403 extern ill_t *ipmp_ill_hold_ipmp_ill(ill_t *);
3399 3404 extern ill_t *ipmp_ill_hold_xmit_ill(ill_t *, boolean_t);
3400 3405 extern boolean_t ipmp_ill_is_active(ill_t *);
3401 3406 extern void ipmp_ill_refresh_active(ill_t *);
3402 3407 extern void ipmp_phyint_join_grp(phyint_t *, ipmp_grp_t *);
3403 3408 extern void ipmp_phyint_leave_grp(phyint_t *);
3404 3409 extern void ipmp_phyint_refresh_active(phyint_t *);
3405 3410 extern ill_t *ipmp_ipif_bound_ill(const ipif_t *);
3406 3411 extern ill_t *ipmp_ipif_hold_bound_ill(const ipif_t *);
3407 3412 extern boolean_t ipmp_ipif_is_dataaddr(const ipif_t *);
3408 3413 extern boolean_t ipmp_ipif_is_stubaddr(const ipif_t *);
3409 3414 extern boolean_t ipmp_packet_is_probe(mblk_t *, ill_t *);
3410 3415 extern void ipmp_ncec_delete_nce(ncec_t *);
3411 3416 extern void ipmp_ncec_refresh_nce(ncec_t *);
3412 3417
3413 3418 extern void conn_drain_insert(conn_t *, idl_tx_list_t *);
3414 3419 extern void conn_setqfull(conn_t *, boolean_t *);
3415 3420 extern void conn_clrqfull(conn_t *, boolean_t *);
3416 3421 extern int conn_ipsec_length(conn_t *);
3417 3422 extern ipaddr_t ip_get_dst(ipha_t *);
3418 3423 extern uint_t ip_get_pmtu(ip_xmit_attr_t *);
3419 3424 extern uint_t ip_get_base_mtu(ill_t *, ire_t *);
3420 3425 extern mblk_t *ip_output_attach_policy(mblk_t *, ipha_t *, ip6_t *,
3421 3426 const conn_t *, ip_xmit_attr_t *);
3422 3427 extern int ipsec_out_extra_length(ip_xmit_attr_t *);
3423 3428 extern int ipsec_out_process(mblk_t *, ip_xmit_attr_t *);
3424 3429 extern int ip_output_post_ipsec(mblk_t *, ip_xmit_attr_t *);
3425 3430 extern void ipsec_out_to_in(ip_xmit_attr_t *, ill_t *ill,
3426 3431 ip_recv_attr_t *);
3427 3432
3428 3433 extern void ire_cleanup(ire_t *);
3429 3434 extern void ire_inactive(ire_t *);
3430 3435 extern boolean_t irb_inactive(irb_t *);
3431 3436 extern ire_t *ire_unlink(irb_t *);
3432 3437
3433 3438 #ifdef DEBUG
3434 3439 extern boolean_t th_trace_ref(const void *, ip_stack_t *);
3435 3440 extern void th_trace_unref(const void *);
3436 3441 extern void th_trace_cleanup(const void *, boolean_t);
3437 3442 extern void ire_trace_ref(ire_t *);
3438 3443 extern void ire_untrace_ref(ire_t *);
3439 3444 #endif
3440 3445
3441 3446 extern int ip_srcid_insert(const in6_addr_t *, zoneid_t, ip_stack_t *);
3442 3447 extern int ip_srcid_remove(const in6_addr_t *, zoneid_t, ip_stack_t *);
3443 3448 extern void ip_srcid_find_id(uint_t, in6_addr_t *, zoneid_t, netstack_t *);
3444 3449 extern uint_t ip_srcid_find_addr(const in6_addr_t *, zoneid_t, netstack_t *);
3445 3450
3446 3451 extern uint8_t ipoptp_next(ipoptp_t *);
3447 3452 extern uint8_t ipoptp_first(ipoptp_t *, ipha_t *);
3448 3453 extern int ip_opt_get_user(conn_t *, uchar_t *);
3449 3454 extern int ipsec_req_from_conn(conn_t *, ipsec_req_t *, int);
3450 3455 extern int ip_snmp_get(queue_t *q, mblk_t *mctl, int level, boolean_t);
3451 3456 extern int ip_snmp_set(queue_t *q, int, int, uchar_t *, int);
3452 3457 extern void ip_process_ioctl(ipsq_t *, queue_t *, mblk_t *, void *);
3453 3458 extern void ip_quiesce_conn(conn_t *);
3454 3459 extern void ip_reprocess_ioctl(ipsq_t *, queue_t *, mblk_t *, void *);
3455 3460 extern void ip_ioctl_finish(queue_t *, mblk_t *, int, int, ipsq_t *);
3456 3461
3457 3462 extern boolean_t ip_cmpbuf(const void *, uint_t, boolean_t, const void *,
3458 3463 uint_t);
3459 3464 extern boolean_t ip_allocbuf(void **, uint_t *, boolean_t, const void *,
3460 3465 uint_t);
3461 3466 extern void ip_savebuf(void **, uint_t *, boolean_t, const void *, uint_t);
3462 3467
3463 3468 extern boolean_t ipsq_pending_mp_cleanup(ill_t *, conn_t *);
3464 3469 extern void conn_ioctl_cleanup(conn_t *);
3465 3470
3466 3471 extern void ip_unbind(conn_t *);
3467 3472
3468 3473 extern void tnet_init(void);
3469 3474 extern void tnet_fini(void);
3470 3475
3471 3476 /*
3472 3477 * Hook functions to enable cluster networking
3473 3478 * On non-clustered systems these vectors must always be NULL.
3474 3479 */
3475 3480 extern int (*cl_inet_isclusterwide)(netstackid_t stack_id, uint8_t protocol,
3476 3481 sa_family_t addr_family, uint8_t *laddrp, void *args);
3477 3482 extern uint32_t (*cl_inet_ipident)(netstackid_t stack_id, uint8_t protocol,
3478 3483 sa_family_t addr_family, uint8_t *laddrp, uint8_t *faddrp,
3479 3484 void *args);
3480 3485 extern int (*cl_inet_connect2)(netstackid_t stack_id, uint8_t protocol,
3481 3486 boolean_t is_outgoing, sa_family_t addr_family, uint8_t *laddrp,
3482 3487 in_port_t lport, uint8_t *faddrp, in_port_t fport, void *args);
3483 3488 extern void (*cl_inet_getspi)(netstackid_t, uint8_t, uint8_t *, size_t,
3484 3489 void *);
3485 3490 extern void (*cl_inet_getspi)(netstackid_t stack_id, uint8_t protocol,
3486 3491 uint8_t *ptr, size_t len, void *args);
3487 3492 extern int (*cl_inet_checkspi)(netstackid_t stack_id, uint8_t protocol,
3488 3493 uint32_t spi, void *args);
3489 3494 extern void (*cl_inet_deletespi)(netstackid_t stack_id, uint8_t protocol,
3490 3495 uint32_t spi, void *args);
3491 3496 extern void (*cl_inet_idlesa)(netstackid_t, uint8_t, uint32_t,
3492 3497 sa_family_t, in6_addr_t, in6_addr_t, void *);
3493 3498
3494 3499
3495 3500 /* Hooks for CGTP (multirt routes) filtering module */
3496 3501 #define CGTP_FILTER_REV_1 1
3497 3502 #define CGTP_FILTER_REV_2 2
3498 3503 #define CGTP_FILTER_REV_3 3
3499 3504 #define CGTP_FILTER_REV CGTP_FILTER_REV_3
3500 3505
3501 3506 /* cfo_filter and cfo_filter_v6 hooks return values */
3502 3507 #define CGTP_IP_PKT_NOT_CGTP 0
3503 3508 #define CGTP_IP_PKT_PREMIUM 1
3504 3509 #define CGTP_IP_PKT_DUPLICATE 2
3505 3510
3506 3511 /* Version 3 of the filter interface */
3507 3512 typedef struct cgtp_filter_ops {
3508 3513 int cfo_filter_rev; /* CGTP_FILTER_REV_3 */
3509 3514 int (*cfo_change_state)(netstackid_t, int);
3510 3515 int (*cfo_add_dest_v4)(netstackid_t, ipaddr_t, ipaddr_t,
3511 3516 ipaddr_t, ipaddr_t);
3512 3517 int (*cfo_del_dest_v4)(netstackid_t, ipaddr_t, ipaddr_t);
3513 3518 int (*cfo_add_dest_v6)(netstackid_t, in6_addr_t *, in6_addr_t *,
3514 3519 in6_addr_t *, in6_addr_t *);
3515 3520 int (*cfo_del_dest_v6)(netstackid_t, in6_addr_t *, in6_addr_t *);
3516 3521 int (*cfo_filter)(netstackid_t, uint_t, mblk_t *);
3517 3522 int (*cfo_filter_v6)(netstackid_t, uint_t, ip6_t *,
3518 3523 ip6_frag_t *);
3519 3524 } cgtp_filter_ops_t;
3520 3525
3521 3526 #define CGTP_MCAST_SUCCESS 1
3522 3527
3523 3528 /*
3524 3529 * The separate CGTP module needs this global symbol so that it
3525 3530 * can check the version and determine whether to use the old or the new
3526 3531 * version of the filtering interface.
3527 3532 */
3528 3533 extern int ip_cgtp_filter_rev;
3529 3534
3530 3535 extern int ip_cgtp_filter_supported(void);
3531 3536 extern int ip_cgtp_filter_register(netstackid_t, cgtp_filter_ops_t *);
3532 3537 extern int ip_cgtp_filter_unregister(netstackid_t);
3533 3538 extern int ip_cgtp_filter_is_registered(netstackid_t);
3534 3539
3535 3540 /*
3536 3541 * rr_ring_state cycles in the order shown below from RR_FREE through
3537 3542 * RR_FREE_IN_PROG and back to RR_FREE.
3538 3543 */
3539 3544 typedef enum {
3540 3545 RR_FREE, /* Free slot */
3541 3546 RR_SQUEUE_UNBOUND, /* Ring's squeue is unbound */
3542 3547 RR_SQUEUE_BIND_INPROG, /* Ring's squeue bind in progress */
3543 3548 RR_SQUEUE_BOUND, /* Ring's squeue bound to cpu */
3544 3549 RR_FREE_INPROG /* Ring is being freed */
3545 3550 } ip_ring_state_t;
3546 3551
3547 3552 #define ILL_MAX_RINGS 256 /* Max num of rx rings we can manage */
3548 3553 #define ILL_POLLING 0x01 /* Polling in use */
3549 3554
3550 3555 /*
3551 3556 * These functions pointer types are exported by the mac/dls layer.
3552 3557 * we need to duplicate the definitions here because we cannot
3553 3558 * include mac/dls header files here.
3554 3559 */
3555 3560 typedef boolean_t (*ip_mac_intr_disable_t)(void *);
3556 3561 typedef void (*ip_mac_intr_enable_t)(void *);
3557 3562 typedef ip_mac_tx_cookie_t (*ip_dld_tx_t)(void *, mblk_t *,
3558 3563 uint64_t, uint16_t);
3559 3564 typedef void (*ip_flow_enable_t)(void *, ip_mac_tx_cookie_t);
3560 3565 typedef void *(*ip_dld_callb_t)(void *,
3561 3566 ip_flow_enable_t, void *);
3562 3567 typedef boolean_t (*ip_dld_fctl_t)(void *, ip_mac_tx_cookie_t);
3563 3568 typedef int (*ip_capab_func_t)(void *, uint_t,
3564 3569 void *, uint_t);
3565 3570
3566 3571 /*
3567 3572 * POLLING README
3568 3573 * sq_get_pkts() is called to pick packets from softring in poll mode. It
3569 3574 * calls rr_rx to get the chain and process it with rr_ip_accept.
3570 3575 * rr_rx = mac_soft_ring_poll() to pick packets
3571 3576 * rr_ip_accept = ip_accept_tcp() to process packets
3572 3577 */
3573 3578
3574 3579 /*
3575 3580 * XXX: With protocol, service specific squeues, they will have
3576 3581 * specific acceptor functions.
3577 3582 */
3578 3583 typedef mblk_t *(*ip_mac_rx_t)(void *, size_t);
3579 3584 typedef mblk_t *(*ip_accept_t)(ill_t *, ill_rx_ring_t *,
3580 3585 squeue_t *, mblk_t *, mblk_t **, uint_t *);
3581 3586
3582 3587 /*
3583 3588 * rr_intr_enable, rr_intr_disable, rr_rx_handle, rr_rx:
3584 3589 * May be accessed while in the squeue AND after checking that SQS_POLL_CAPAB
3585 3590 * is set.
3586 3591 *
3587 3592 * rr_ring_state: Protected by ill_lock.
3588 3593 */
3589 3594 struct ill_rx_ring {
3590 3595 ip_mac_intr_disable_t rr_intr_disable; /* Interrupt disabling func */
3591 3596 ip_mac_intr_enable_t rr_intr_enable; /* Interrupt enabling func */
3592 3597 void *rr_intr_handle; /* Handle interrupt funcs */
3593 3598 ip_mac_rx_t rr_rx; /* Driver receive function */
3594 3599 ip_accept_t rr_ip_accept; /* IP accept function */
3595 3600 void *rr_rx_handle; /* Handle for Rx ring */
3596 3601 squeue_t *rr_sqp; /* Squeue the ring is bound to */
3597 3602 ill_t *rr_ill; /* back pointer to ill */
3598 3603 ip_ring_state_t rr_ring_state; /* State of this ring */
3599 3604 };
3600 3605
3601 3606 /*
3602 3607 * IP - DLD direct function call capability
3603 3608 * Suffixes, df - dld function, dh - dld handle,
3604 3609 * cf - client (IP) function, ch - client handle
3605 3610 */
3606 3611 typedef struct ill_dld_direct_s { /* DLD provided driver Tx */
3607 3612 ip_dld_tx_t idd_tx_df; /* str_mdata_fastpath_put */
3608 3613 void *idd_tx_dh; /* dld_str_t *dsp */
3609 3614 ip_dld_callb_t idd_tx_cb_df; /* mac_tx_srs_notify */
3610 3615 void *idd_tx_cb_dh; /* mac_client_handle_t *mch */
3611 3616 ip_dld_fctl_t idd_tx_fctl_df; /* mac_tx_is_flow_blocked */
3612 3617 void *idd_tx_fctl_dh; /* mac_client_handle */
3613 3618 } ill_dld_direct_t;
3614 3619
3615 3620 /* IP - DLD polling capability */
3616 3621 typedef struct ill_dld_poll_s {
3617 3622 ill_rx_ring_t idp_ring_tbl[ILL_MAX_RINGS];
3618 3623 } ill_dld_poll_t;
3619 3624
3620 3625 /* Describes ill->ill_dld_capab */
3621 3626 struct ill_dld_capab_s {
3622 3627 ip_capab_func_t idc_capab_df; /* dld_capab_func */
3623 3628 void *idc_capab_dh; /* dld_str_t *dsp */
3624 3629 ill_dld_direct_t idc_direct;
3625 3630 ill_dld_poll_t idc_poll;
3626 3631 };
3627 3632
3628 3633 /*
3629 3634 * IP squeues exports
3630 3635 */
3631 3636 extern boolean_t ip_squeue_fanout;
3632 3637
3633 3638 #define IP_SQUEUE_GET(hint) ip_squeue_random(hint)
3634 3639
3635 3640 extern void ip_squeue_init(void (*)(squeue_t *));
3636 3641 extern squeue_t *ip_squeue_random(uint_t);
3637 3642 extern squeue_t *ip_squeue_get(ill_rx_ring_t *);
3638 3643 extern squeue_t *ip_squeue_getfree(pri_t);
3639 3644 extern int ip_squeue_cpu_move(squeue_t *, processorid_t);
3640 3645 extern void *ip_squeue_add_ring(ill_t *, void *);
3641 3646 extern void ip_squeue_bind_ring(ill_t *, ill_rx_ring_t *, processorid_t);
3642 3647 extern void ip_squeue_clean_ring(ill_t *, ill_rx_ring_t *);
3643 3648 extern void ip_squeue_quiesce_ring(ill_t *, ill_rx_ring_t *);
3644 3649 extern void ip_squeue_restart_ring(ill_t *, ill_rx_ring_t *);
3645 3650 extern void ip_squeue_clean_all(ill_t *);
3646 3651 extern boolean_t ip_source_routed(ipha_t *, ip_stack_t *);
3647 3652
3648 3653 extern void tcp_wput(queue_t *, mblk_t *);
3649 3654
3650 3655 extern int ip_fill_mtuinfo(conn_t *, ip_xmit_attr_t *,
3651 3656 struct ip6_mtuinfo *);
3652 3657 extern hook_t *ipobs_register_hook(netstack_t *, pfv_t);
3653 3658 extern void ipobs_unregister_hook(netstack_t *, hook_t *);
3654 3659 extern void ipobs_hook(mblk_t *, int, zoneid_t, zoneid_t, const ill_t *,
3655 3660 ip_stack_t *);
3656 3661 typedef void (*ipsq_func_t)(ipsq_t *, queue_t *, mblk_t *, void *);
3657 3662
3658 3663 extern void dce_g_init(void);
3659 3664 extern void dce_g_destroy(void);
3660 3665 extern void dce_stack_init(ip_stack_t *);
3661 3666 extern void dce_stack_destroy(ip_stack_t *);
3662 3667 extern void dce_cleanup(uint_t, ip_stack_t *);
3663 3668 extern dce_t *dce_get_default(ip_stack_t *);
3664 3669 extern dce_t *dce_lookup_pkt(mblk_t *, ip_xmit_attr_t *, uint_t *);
3665 3670 extern dce_t *dce_lookup_v4(ipaddr_t, ip_stack_t *, uint_t *);
3666 3671 extern dce_t *dce_lookup_v6(const in6_addr_t *, uint_t, ip_stack_t *,
3667 3672 uint_t *);
3668 3673 extern dce_t *dce_lookup_and_add_v4(ipaddr_t, ip_stack_t *);
3669 3674 extern dce_t *dce_lookup_and_add_v6(const in6_addr_t *, uint_t,
3670 3675 ip_stack_t *);
3671 3676 extern int dce_update_uinfo_v4(ipaddr_t, iulp_t *, ip_stack_t *);
3672 3677 extern int dce_update_uinfo_v6(const in6_addr_t *, uint_t, iulp_t *,
3673 3678 ip_stack_t *);
3674 3679 extern int dce_update_uinfo(const in6_addr_t *, uint_t, iulp_t *,
3675 3680 ip_stack_t *);
3676 3681 extern void dce_increment_generation(dce_t *);
3677 3682 extern void dce_increment_all_generations(boolean_t, ip_stack_t *);
3678 3683 extern void dce_refrele(dce_t *);
3679 3684 extern void dce_refhold(dce_t *);
3680 3685 extern void dce_refrele_notr(dce_t *);
3681 3686 extern void dce_refhold_notr(dce_t *);
3682 3687 mblk_t *ip_snmp_get_mib2_ip_dce(queue_t *, mblk_t *, ip_stack_t *ipst);
3683 3688
3684 3689 extern ip_laddr_t ip_laddr_verify_v4(ipaddr_t, zoneid_t,
3685 3690 ip_stack_t *, boolean_t);
3686 3691 extern ip_laddr_t ip_laddr_verify_v6(const in6_addr_t *, zoneid_t,
3687 3692 ip_stack_t *, boolean_t, uint_t);
3688 3693 extern int ip_laddr_fanout_insert(conn_t *);
3689 3694
3690 3695 extern boolean_t ip_verify_src(mblk_t *, ip_xmit_attr_t *, uint_t *);
3691 3696 extern int ip_verify_ire(mblk_t *, ip_xmit_attr_t *);
3692 3697
3693 3698 extern mblk_t *ip_xmit_attr_to_mblk(ip_xmit_attr_t *);
3694 3699 extern boolean_t ip_xmit_attr_from_mblk(mblk_t *, ip_xmit_attr_t *);
3695 3700 extern mblk_t *ip_xmit_attr_free_mblk(mblk_t *);
3696 3701 extern mblk_t *ip_recv_attr_to_mblk(ip_recv_attr_t *);
3697 3702 extern boolean_t ip_recv_attr_from_mblk(mblk_t *, ip_recv_attr_t *);
3698 3703 extern mblk_t *ip_recv_attr_free_mblk(mblk_t *);
3699 3704 extern boolean_t ip_recv_attr_is_mblk(mblk_t *);
3700 3705
3701 3706 /*
3702 3707 * Squeue tags. Tags only need to be unique when the callback function is the
3703 3708 * same to distinguish between different calls, but we use unique tags for
3704 3709 * convenience anyway.
3705 3710 */
3706 3711 #define SQTAG_IP_INPUT 1
3707 3712 #define SQTAG_TCP_INPUT_ICMP_ERR 2
3708 3713 #define SQTAG_TCP6_INPUT_ICMP_ERR 3
3709 3714 #define SQTAG_IP_TCP_INPUT 4
3710 3715 #define SQTAG_IP6_TCP_INPUT 5
3711 3716 #define SQTAG_IP_TCP_CLOSE 6
3712 3717 #define SQTAG_TCP_OUTPUT 7
3713 3718 #define SQTAG_TCP_TIMER 8
3714 3719 #define SQTAG_TCP_TIMEWAIT 9
3715 3720 #define SQTAG_TCP_ACCEPT_FINISH 10
3716 3721 #define SQTAG_TCP_ACCEPT_FINISH_Q0 11
3717 3722 #define SQTAG_TCP_ACCEPT_PENDING 12
3718 3723 #define SQTAG_TCP_LISTEN_DISCON 13
3719 3724 #define SQTAG_TCP_CONN_REQ_1 14
3720 3725 #define SQTAG_TCP_EAGER_BLOWOFF 15
3721 3726 #define SQTAG_TCP_EAGER_CLEANUP 16
3722 3727 #define SQTAG_TCP_EAGER_CLEANUP_Q0 17
3723 3728 #define SQTAG_TCP_CONN_IND 18
3724 3729 #define SQTAG_TCP_RSRV 19
3725 3730 #define SQTAG_TCP_ABORT_BUCKET 20
3726 3731 #define SQTAG_TCP_REINPUT 21
3727 3732 #define SQTAG_TCP_REINPUT_EAGER 22
3728 3733 #define SQTAG_TCP_INPUT_MCTL 23
3729 3734 #define SQTAG_TCP_RPUTOTHER 24
3730 3735 #define SQTAG_IP_PROTO_AGAIN 25
3731 3736 #define SQTAG_IP_FANOUT_TCP 26
3732 3737 #define SQTAG_IPSQ_CLEAN_RING 27
3733 3738 #define SQTAG_TCP_WPUT_OTHER 28
3734 3739 #define SQTAG_TCP_CONN_REQ_UNBOUND 29
3735 3740 #define SQTAG_TCP_SEND_PENDING 30
3736 3741 #define SQTAG_BIND_RETRY 31
3737 3742 #define SQTAG_UDP_FANOUT 32
3738 3743 #define SQTAG_UDP_INPUT 33
3739 3744 #define SQTAG_UDP_WPUT 34
3740 3745 #define SQTAG_UDP_OUTPUT 35
|
↓ open down ↓ |
3511 lines elided |
↑ open up ↑ |
3741 3746 #define SQTAG_TCP_KSSL_INPUT 36
3742 3747 #define SQTAG_TCP_DROP_Q0 37
3743 3748 #define SQTAG_TCP_CONN_REQ_2 38
3744 3749 #define SQTAG_IP_INPUT_RX_RING 39
3745 3750 #define SQTAG_SQUEUE_CHANGE 40
3746 3751 #define SQTAG_CONNECT_FINISH 41
3747 3752 #define SQTAG_SYNCHRONOUS_OP 42
3748 3753 #define SQTAG_TCP_SHUTDOWN_OUTPUT 43
3749 3754 #define SQTAG_TCP_IXA_CLEANUP 44
3750 3755 #define SQTAG_TCP_SEND_SYNACK 45
3756 +#define SQTAG_IP_DCCP_INPUT 46
3757 +#define SQTAG_DCCP_OUTPUT 47
3758 +#define SQTAG_DCCP_CONN_REQ_UNBOUND 48
3751 3759
3752 3760 extern sin_t sin_null; /* Zero address for quick clears */
3753 3761 extern sin6_t sin6_null; /* Zero address for quick clears */
3754 3762
3755 3763 #endif /* _KERNEL */
3756 3764
3757 3765 #ifdef __cplusplus
3758 3766 }
3759 3767 #endif
3760 3768
3761 3769 #endif /* _INET_IP_H */
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX