xref: /illumos-gate/usr/src/uts/common/inet/ip_stack.h (revision 221e47fb)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License (the "License").
6  * You may not use this file except in compliance with the License.
7  *
8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9  * or http://www.opensolaris.org/os/licensing.
10  * See the License for the specific language governing permissions
11  * and limitations under the License.
12  *
13  * When distributing Covered Code, include this CDDL HEADER in each
14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15  * If applicable, add the following below this CDDL HEADER, with the
16  * fields enclosed by brackets "[]" replaced with your own identifying
17  * information: Portions Copyright [yyyy] [name of copyright owner]
18  *
19  * CDDL HEADER END
20  */
21 
22 /*
23  * Copyright 2010 Sun Microsystems, Inc.  All rights reserved.
24  * Use is subject to license terms.
25  */
26 
27 /*
28  * Copyright 2019 Joyent, Inc.
29  * Copyright 2020 OmniOS Community Edition (OmniOSce) Association.
30  */
31 
32 #ifndef	_INET_IP_STACK_H
33 #define	_INET_IP_STACK_H
34 
35 #ifdef	__cplusplus
36 extern "C" {
37 #endif
38 
39 #include <sys/netstack.h>
40 #include <netinet/igmp_var.h>
41 #include <sys/modhash.h>
42 
43 #ifdef _KERNEL
44 #include <sys/list.h>
45 
46 
47 /*
48  * IP statistics.
49  */
50 #define	IP_STAT(ipst, x)	((ipst)->ips_ip_statistics.x.value.ui64++)
51 #define	IP_STAT_UPDATE(ipst, x, n) \
52 		((ipst)->ips_ip_statistics.x.value.ui64 += (n))
53 
54 typedef struct ip_stat {
55 	kstat_named_t	ip_udp_fannorm;
56 	kstat_named_t	ip_udp_fanmb;
57 	kstat_named_t	ip_recv_pullup;
58 	kstat_named_t	ip_db_ref;
59 	kstat_named_t	ip_notaligned;
60 	kstat_named_t	ip_multimblk;
61 	kstat_named_t	ip_opt;
62 	kstat_named_t	ipsec_proto_ahesp;
63 	kstat_named_t	ip_conn_flputbq;
64 	kstat_named_t	ip_conn_walk_drain;
65 	kstat_named_t   ip_out_sw_cksum;
66 	kstat_named_t	ip_out_sw_cksum_bytes;
67 	kstat_named_t   ip_in_sw_cksum;
68 	kstat_named_t   ip_ire_reclaim_calls;
69 	kstat_named_t   ip_ire_reclaim_deleted;
70 	kstat_named_t   ip_nce_reclaim_calls;
71 	kstat_named_t   ip_nce_reclaim_deleted;
72 	kstat_named_t   ip_nce_mcast_reclaim_calls;
73 	kstat_named_t   ip_nce_mcast_reclaim_deleted;
74 	kstat_named_t   ip_nce_mcast_reclaim_tqfail;
75 	kstat_named_t   ip_dce_reclaim_calls;
76 	kstat_named_t   ip_dce_reclaim_deleted;
77 	kstat_named_t	ip_tcp_in_full_hw_cksum_err;
78 	kstat_named_t	ip_tcp_in_part_hw_cksum_err;
79 	kstat_named_t	ip_tcp_in_sw_cksum_err;
80 	kstat_named_t	ip_udp_in_full_hw_cksum_err;
81 	kstat_named_t	ip_udp_in_part_hw_cksum_err;
82 	kstat_named_t	ip_udp_in_sw_cksum_err;
83 	kstat_named_t	conn_in_recvdstaddr;
84 	kstat_named_t	conn_in_recvopts;
85 	kstat_named_t	conn_in_recvif;
86 	kstat_named_t	conn_in_recvslla;
87 	kstat_named_t	conn_in_recvucred;
88 	kstat_named_t	conn_in_recvttl;
89 	kstat_named_t	conn_in_recvtos;
90 	kstat_named_t	conn_in_recvhopopts;
91 	kstat_named_t	conn_in_recvhoplimit;
92 	kstat_named_t	conn_in_recvdstopts;
93 	kstat_named_t	conn_in_recvrthdrdstopts;
94 	kstat_named_t	conn_in_recvrthdr;
95 	kstat_named_t	conn_in_recvpktinfo;
96 	kstat_named_t	conn_in_recvtclass;
97 	kstat_named_t	conn_in_timestamp;
98 } ip_stat_t;
99 
100 
101 /*
102  * IP6 statistics.
103  */
104 #define	IP6_STAT(ipst, x)	((ipst)->ips_ip6_statistics.x.value.ui64++)
105 #define	IP6_STAT_UPDATE(ipst, x, n)	\
106 	((ipst)->ips_ip6_statistics.x.value.ui64 += (n))
107 
108 typedef struct ip6_stat {
109 	kstat_named_t	ip6_udp_fannorm;
110 	kstat_named_t	ip6_udp_fanmb;
111 	kstat_named_t	ip6_recv_pullup;
112 	kstat_named_t	ip6_db_ref;
113 	kstat_named_t	ip6_notaligned;
114 	kstat_named_t	ip6_multimblk;
115 	kstat_named_t	ipsec_proto_ahesp;
116 	kstat_named_t   ip6_out_sw_cksum;
117 	kstat_named_t	ip6_out_sw_cksum_bytes;
118 	kstat_named_t   ip6_in_sw_cksum;
119 	kstat_named_t	ip6_tcp_in_full_hw_cksum_err;
120 	kstat_named_t	ip6_tcp_in_part_hw_cksum_err;
121 	kstat_named_t	ip6_tcp_in_sw_cksum_err;
122 	kstat_named_t	ip6_udp_in_full_hw_cksum_err;
123 	kstat_named_t	ip6_udp_in_part_hw_cksum_err;
124 	kstat_named_t	ip6_udp_in_sw_cksum_err;
125 } ip6_stat_t;
126 
127 typedef struct ire_stats {
128 	uint64_t ire_stats_alloced;	/* # of ires alloced */
129 	uint64_t ire_stats_freed;	/* # of ires freed */
130 	uint64_t ire_stats_inserted;	/* # of ires inserted in the bucket */
131 	uint64_t ire_stats_deleted;	/* # of ires deleted from the bucket */
132 } ire_stats_t;
133 
134 #define	TX_FANOUT_SIZE	128
135 #define	IDLHASHINDEX(X)	\
136 	((((uintptr_t)(X) >> 2) + ((uintptr_t)(X) >> 9)) & (TX_FANOUT_SIZE - 1))
137 
138 /* Data structure to represent addresses */
139 typedef struct srcid_map {
140 	struct srcid_map	*sm_next;
141 	in6_addr_t		sm_addr;	/* Local address */
142 	uint_t			sm_srcid;	/* source id */
143 	uint_t			sm_refcnt;	/* > 1 ipif with same addr? */
144 	zoneid_t		sm_zoneid;	/* zone id */
145 } srcid_map_t;
146 
147 /*
148  * IP stack instances
149  */
150 struct ip_stack {
151 	netstack_t	*ips_netstack;	/* Common netstack */
152 
153 	uint_t			ips_src_generation;	/* Both IPv4 and IPv6 */
154 
155 	struct mod_prop_info_s	*ips_propinfo_tbl;	/* ip tunables table */
156 
157 	mib2_ipIfStatsEntry_t	ips_ip_mib;	/* SNMP fixed size info */
158 	mib2_icmp_t	ips_icmp_mib;
159 	/*
160 	 * IPv6 mibs when the interface (ill) is not known.
161 	 * When the ill is known the per-interface mib in the ill is used.
162 	 */
163 	mib2_ipIfStatsEntry_t	ips_ip6_mib;
164 	mib2_ipv6IfIcmpEntry_t	ips_icmp6_mib;
165 
166 	struct igmpstat		ips_igmpstat;
167 
168 	kstat_t		*ips_ip_mibkp;	/* kstat exporting ip_mib data */
169 	kstat_t		*ips_icmp_mibkp; /* kstat exporting icmp_mib data */
170 	kstat_t		*ips_ip_kstat;
171 	ip_stat_t	ips_ip_statistics;
172 	kstat_t		*ips_ip6_kstat;
173 	ip6_stat_t	ips_ip6_statistics;
174 
175 /* ip.c */
176 	kmutex_t	ips_igmp_timer_lock;
177 	kmutex_t	ips_mld_timer_lock;
178 	kmutex_t	ips_ip_mi_lock;
179 	kmutex_t	ips_ip_addr_avail_lock;
180 	krwlock_t	ips_ill_g_lock;
181 
182 	krwlock_t	ips_ill_g_usesrc_lock;
183 
184 	/* Taskq dispatcher for capability operations */
185 	kmutex_t	ips_capab_taskq_lock;
186 	kcondvar_t	ips_capab_taskq_cv;
187 	mblk_t		*ips_capab_taskq_head;
188 	mblk_t		*ips_capab_taskq_tail;
189 	kthread_t	*ips_capab_taskq_thread;
190 	boolean_t	ips_capab_taskq_quit;
191 
192 /* ipclassifier.c - keep in ip_stack_t */
193 	/* ipclassifier hash tables */
194 	struct connf_s	*ips_rts_clients;
195 	struct connf_s	*ips_ipcl_conn_fanout;
196 	struct connf_s	*ips_ipcl_bind_fanout;
197 	struct connf_s	*ips_ipcl_proto_fanout_v4;
198 	struct connf_s	*ips_ipcl_proto_fanout_v6;
199 	struct connf_s	*ips_ipcl_udp_fanout;
200 	struct connf_s	*ips_ipcl_raw_fanout;		/* RAW SCTP sockets */
201 	struct connf_s	*ips_ipcl_iptun_fanout;
202 	uint_t		ips_ipcl_conn_fanout_size;
203 	uint_t		ips_ipcl_bind_fanout_size;
204 	uint_t		ips_ipcl_udp_fanout_size;
205 	uint_t		ips_ipcl_raw_fanout_size;
206 	uint_t		ips_ipcl_iptun_fanout_size;
207 	struct connf_s	*ips_ipcl_globalhash_fanout;
208 	int		ips_conn_g_index;
209 
210 /* ip.c */
211 	/* Following protected by igmp_timer_lock */
212 	int		ips_igmp_time_to_next;	/* Time since last timeout */
213 	int		ips_igmp_timer_scheduled_last;
214 	int		ips_igmp_deferred_next;
215 	timeout_id_t	ips_igmp_timeout_id;
216 	boolean_t	ips_igmp_timer_setter_active;
217 	boolean_t	ips_igmp_timer_quiesce;
218 
219 	/* Following protected by mld_timer_lock */
220 	int		ips_mld_time_to_next;	/* Time since last timeout */
221 	int		ips_mld_timer_scheduled_last;
222 	int		ips_mld_deferred_next;
223 	timeout_id_t	ips_mld_timeout_id;
224 	boolean_t	ips_mld_timer_setter_active;
225 	boolean_t	ips_mld_timer_quiesce;
226 
227 	/* Protected by igmp_slowtimeout_lock */
228 	timeout_id_t	ips_igmp_slowtimeout_id;
229 	kmutex_t	ips_igmp_slowtimeout_lock;
230 	boolean_t	ips_igmp_slowtimeout_quiesce;
231 
232 	/* Protected by mld_slowtimeout_lock */
233 	timeout_id_t	ips_mld_slowtimeout_id;
234 	kmutex_t	ips_mld_slowtimeout_lock;
235 	boolean_t	ips_mld_slowtimeout_quiesce;
236 
237 	/* IPv4 forwarding table */
238 	struct radix_node_head *ips_ip_ftable;
239 
240 #define	IPV6_ABITS		128
241 #define	IP6_MASK_TABLE_SIZE	(IPV6_ABITS + 1)	/* 129 ptrs */
242 	struct irb	*ips_ip_forwarding_table_v6[IP6_MASK_TABLE_SIZE];
243 
244 	/*
245 	 * ire_ft_init_lock is used while initializing ip_forwarding_table
246 	 * dynamically in ire_add.
247 	 */
248 	kmutex_t	ips_ire_ft_init_lock;
249 
250 	/*
251 	 * This is the IPv6 counterpart of RADIX_NODE_HEAD_LOCK. It is used
252 	 * to prevent adds and deletes while we are doing a ftable_lookup
253 	 * and extracting the ire_generation.
254 	 */
255 	krwlock_t	ips_ip6_ire_head_lock;
256 
257 	uint32_t	ips_ip6_ftable_hash_size;
258 
259 	ire_stats_t	ips_ire_stats_v4;	/* IPv4 ire statistics */
260 	ire_stats_t	ips_ire_stats_v6;	/* IPv6 ire statistics */
261 
262 	/* Count how many condemned objects for kmem_cache callbacks */
263 	uint32_t	ips_num_ire_condemned;
264 	uint32_t	ips_num_nce_condemned;
265 	uint32_t	ips_num_dce_condemned;
266 
267 	struct ire_s	*ips_ire_reject_v4;	/* For unreachable dests */
268 	struct ire_s	*ips_ire_reject_v6;	/* For unreachable dests */
269 	struct ire_s	*ips_ire_blackhole_v4;	/* For temporary failures */
270 	struct ire_s	*ips_ire_blackhole_v6;	/* For temporary failures */
271 
272 	/* ips_ire_dep_lock protects ire_dep_* relationship between IREs */
273 	krwlock_t	ips_ire_dep_lock;
274 
275 	/* Destination Cache Entries */
276 	struct dce_s	*ips_dce_default;
277 	uint_t		ips_dce_hashsize;
278 	struct dcb_s	*ips_dce_hash_v4;
279 	struct dcb_s	*ips_dce_hash_v6;
280 	uint_t		ips_dce_reclaim_needed;
281 
282 	/* pending binds */
283 	mblk_t		*ips_ip6_asp_pending_ops;
284 	mblk_t		*ips_ip6_asp_pending_ops_tail;
285 
286 	/* Synchronize updates with table usage */
287 	mblk_t		*ips_ip6_asp_pending_update; /* pending table updates */
288 
289 	boolean_t	ips_ip6_asp_uip;	/* table update in progress */
290 	kmutex_t	ips_ip6_asp_lock;	/* protect all the above */
291 	uint32_t	ips_ip6_asp_refcnt;	/* outstanding references */
292 
293 	struct ip6_asp	*ips_ip6_asp_table;
294 	/* The number of policy entries in the table */
295 	uint_t		ips_ip6_asp_table_count;
296 
297 	struct conn_s	*ips_ip_g_mrouter;
298 
299 	/* Time since last icmp_pkt_err */
300 	clock_t		ips_icmp_pkt_err_last;
301 	/* Number of packets sent in burst */
302 	uint_t		ips_icmp_pkt_err_sent;
303 
304 	/* Protected by ip_mi_lock */
305 	void		*ips_ip_g_head;	/* IP Instance Data List Head */
306 	void		*ips_arp_g_head; /* ARP Instance Data List Head */
307 
308 	/* Multirouting stuff */
309 	/* Interval (in ms) between consecutive 'bad MTU' warnings */
310 	hrtime_t	ips_ip_multirt_log_interval;
311 	/* Time since last warning issued. */
312 	hrtime_t	ips_multirt_bad_mtu_last_time;
313 
314 	/*
315 	 * CGTP hooks. Enabling and disabling of hooks is controlled by an
316 	 * IP tunable 'ips_ip_cgtp_filter'.
317 	 */
318 	struct cgtp_filter_ops *ips_ip_cgtp_filter_ops;
319 
320 	struct ipsq_s	*ips_ipsq_g_head;
321 	uint_t		ips_ill_index;	/* Used to assign interface indicies */
322 	/* When set search for unused index */
323 	boolean_t	ips_ill_index_wrap;
324 
325 	uint_t		ips_loopback_packets;
326 
327 	/* NDP/NCE structures for IPv4 and IPv6 */
328 	struct ndp_g_s	*ips_ndp4;
329 	struct ndp_g_s	*ips_ndp6;
330 
331 	/* ip_mroute stuff */
332 	kmutex_t	ips_ip_g_mrouter_mutex;
333 
334 	struct mrtstat	*ips_mrtstat;	/* Stats for netstat */
335 	int		ips_saved_ip_forwarding;
336 
337 	/* numvifs is only a hint about the max interface being used. */
338 	ushort_t	ips_numvifs;
339 	kmutex_t	ips_numvifs_mutex;
340 
341 	struct vif	*ips_vifs;
342 	struct mfcb	*ips_mfcs;	/* kernel routing table	*/
343 	struct tbf	*ips_tbfs;
344 	/*
345 	 * One-back cache used to locate a tunnel's vif,
346 	 * given a datagram's src ip address.
347 	 */
348 	ipaddr_t	ips_last_encap_src;
349 	struct vif	*ips_last_encap_vif;
350 	kmutex_t	ips_last_encap_lock;	/* Protects the above */
351 
352 	/*
353 	 * reg_vif_num is protected by numvifs_mutex
354 	 */
355 	/* Whether or not special PIM assert processing is enabled. */
356 	ushort_t	ips_reg_vif_num;	/* Index to Register vif */
357 	int		ips_pim_assert;
358 
359 	union ill_g_head_u *ips_ill_g_heads;   /* ILL List Head */
360 
361 	kstat_t		*ips_loopback_ksp;
362 
363 	/* Array of conn drain lists */
364 	struct idl_tx_list_s	*ips_idl_tx_list;
365 	uint_t		ips_conn_drain_list_cnt; /* Count of conn_drain_list */
366 
367 	/*
368 	 * ID used to assign next free one.
369 	 * Increases by one. Once it wraps we search for an unused ID.
370 	 */
371 	uint_t		ips_ip_src_id;
372 	boolean_t	ips_srcid_wrapped;
373 
374 	struct srcid_map *ips_srcid_head;
375 	krwlock_t	ips_srcid_lock;
376 
377 	uint64_t	ips_ipif_g_seqid;	/* Used only for sctp_addr.c */
378 	union phyint_list_u *ips_phyint_g_list;	/* start of phyint list */
379 
380 /* ip_netinfo.c */
381 	hook_family_t	ips_ipv4root;
382 	hook_family_t	ips_ipv6root;
383 	hook_family_t	ips_arproot;
384 
385 	net_handle_t		ips_ipv4_net_data;
386 	net_handle_t		ips_ipv6_net_data;
387 	net_handle_t		ips_arp_net_data;
388 
389 	/*
390 	 * Hooks for firewalling
391 	 */
392 	hook_event_t		ips_ip4_physical_in_event;
393 	hook_event_t		ips_ip4_physical_out_event;
394 	hook_event_t		ips_ip4_forwarding_event;
395 	hook_event_t		ips_ip4_loopback_in_event;
396 	hook_event_t		ips_ip4_loopback_out_event;
397 
398 	hook_event_t		ips_ip6_physical_in_event;
399 	hook_event_t		ips_ip6_physical_out_event;
400 	hook_event_t		ips_ip6_forwarding_event;
401 	hook_event_t		ips_ip6_loopback_in_event;
402 	hook_event_t		ips_ip6_loopback_out_event;
403 
404 	hook_event_t		ips_arp_physical_in_event;
405 	hook_event_t		ips_arp_physical_out_event;
406 	hook_event_t		ips_arp_nic_events;
407 
408 	hook_event_token_t	ips_ipv4firewall_physical_in;
409 	hook_event_token_t	ips_ipv4firewall_physical_out;
410 	hook_event_token_t	ips_ipv4firewall_forwarding;
411 	hook_event_token_t	ips_ipv4firewall_loopback_in;
412 	hook_event_token_t	ips_ipv4firewall_loopback_out;
413 
414 	hook_event_token_t	ips_ipv6firewall_physical_in;
415 	hook_event_token_t	ips_ipv6firewall_physical_out;
416 	hook_event_token_t	ips_ipv6firewall_forwarding;
417 	hook_event_token_t	ips_ipv6firewall_loopback_in;
418 	hook_event_token_t	ips_ipv6firewall_loopback_out;
419 
420 	hook_event_t		ips_ip4_nic_events;
421 	hook_event_t		ips_ip6_nic_events;
422 	hook_event_token_t	ips_ipv4nicevents;
423 	hook_event_token_t	ips_ipv6nicevents;
424 
425 	hook_event_token_t	ips_arp_physical_in;
426 	hook_event_token_t	ips_arp_physical_out;
427 	hook_event_token_t	ips_arpnicevents;
428 
429 	net_handle_t		ips_ip4_observe_pr;
430 	net_handle_t		ips_ip6_observe_pr;
431 	hook_event_t		ips_ip4_observe;
432 	hook_event_t		ips_ip6_observe;
433 	hook_event_token_t	ips_ipv4observing;
434 	hook_event_token_t	ips_ipv6observing;
435 
436 	struct __ldi_ident	*ips_ldi_ident;
437 
438 /* ipmp.c */
439 	krwlock_t		ips_ipmp_lock;
440 	mod_hash_t		*ips_ipmp_grp_hash;
441 
442 };
443 typedef struct ip_stack ip_stack_t;
444 
445 /* Finding an ip_stack_t */
446 #define	CONNQ_TO_IPST(_q)	(Q_TO_CONN(_q)->conn_netstack->netstack_ip)
447 #define	ILLQ_TO_IPST(_q)	(((ill_t *)(_q)->q_ptr)->ill_ipst)
448 #define	PHYINT_TO_IPST(phyi)	((phyi)->phyint_ipsq->ipsq_ipst)
449 
450 #else /* _KERNEL */
451 typedef int ip_stack_t;
452 #endif /* _KERNEL */
453 
454 #ifdef	__cplusplus
455 }
456 #endif
457 
458 #endif	/* _INET_IP_STACK_H */
459