xref: /titanic_52/usr/src/uts/common/inet/ip_stack.h (revision efd4c9b63ad77503c101fc6c2ed8ba96c9d52964)
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 #ifndef	_INET_IP_STACK_H
28 #define	_INET_IP_STACK_H
29 
30 #ifdef	__cplusplus
31 extern "C" {
32 #endif
33 
34 #include <sys/netstack.h>
35 #include <netinet/igmp_var.h>
36 #include <sys/modhash.h>
37 
38 #ifdef _KERNEL
39 #include <sys/list.h>
40 
41 
42 /*
43  * IP statistics.
44  */
45 #define	IP_STAT(ipst, x)	((ipst)->ips_ip_statistics.x.value.ui64++)
46 #define	IP_STAT_UPDATE(ipst, x, n) \
47 		((ipst)->ips_ip_statistics.x.value.ui64 += (n))
48 
49 typedef struct ip_stat {
50 	kstat_named_t	ip_udp_fannorm;
51 	kstat_named_t	ip_udp_fanmb;
52 	kstat_named_t	ip_recv_pullup;
53 	kstat_named_t	ip_db_ref;
54 	kstat_named_t	ip_notaligned;
55 	kstat_named_t	ip_multimblk;
56 	kstat_named_t	ip_opt;
57 	kstat_named_t	ipsec_proto_ahesp;
58 	kstat_named_t	ip_conn_flputbq;
59 	kstat_named_t	ip_conn_walk_drain;
60 	kstat_named_t   ip_out_sw_cksum;
61 	kstat_named_t	ip_out_sw_cksum_bytes;
62 	kstat_named_t   ip_in_sw_cksum;
63 	kstat_named_t   ip_ire_reclaim_calls;
64 	kstat_named_t   ip_ire_reclaim_deleted;
65 	kstat_named_t   ip_nce_reclaim_calls;
66 	kstat_named_t   ip_nce_reclaim_deleted;
67 	kstat_named_t   ip_dce_reclaim_calls;
68 	kstat_named_t   ip_dce_reclaim_deleted;
69 	kstat_named_t	ip_tcp_in_full_hw_cksum_err;
70 	kstat_named_t	ip_tcp_in_part_hw_cksum_err;
71 	kstat_named_t	ip_tcp_in_sw_cksum_err;
72 	kstat_named_t	ip_udp_in_full_hw_cksum_err;
73 	kstat_named_t	ip_udp_in_part_hw_cksum_err;
74 	kstat_named_t	ip_udp_in_sw_cksum_err;
75 	kstat_named_t	conn_in_recvdstaddr;
76 	kstat_named_t	conn_in_recvopts;
77 	kstat_named_t	conn_in_recvif;
78 	kstat_named_t	conn_in_recvslla;
79 	kstat_named_t	conn_in_recvucred;
80 	kstat_named_t	conn_in_recvttl;
81 	kstat_named_t	conn_in_recvhopopts;
82 	kstat_named_t	conn_in_recvhoplimit;
83 	kstat_named_t	conn_in_recvdstopts;
84 	kstat_named_t	conn_in_recvrthdrdstopts;
85 	kstat_named_t	conn_in_recvrthdr;
86 	kstat_named_t	conn_in_recvpktinfo;
87 	kstat_named_t	conn_in_recvtclass;
88 	kstat_named_t	conn_in_timestamp;
89 } ip_stat_t;
90 
91 
92 /*
93  * IP6 statistics.
94  */
95 #define	IP6_STAT(ipst, x)	((ipst)->ips_ip6_statistics.x.value.ui64++)
96 #define	IP6_STAT_UPDATE(ipst, x, n)	\
97 	((ipst)->ips_ip6_statistics.x.value.ui64 += (n))
98 
99 typedef struct ip6_stat {
100 	kstat_named_t	ip6_udp_fannorm;
101 	kstat_named_t	ip6_udp_fanmb;
102 	kstat_named_t	ip6_recv_pullup;
103 	kstat_named_t	ip6_db_ref;
104 	kstat_named_t	ip6_notaligned;
105 	kstat_named_t	ip6_multimblk;
106 	kstat_named_t	ipsec_proto_ahesp;
107 	kstat_named_t   ip6_out_sw_cksum;
108 	kstat_named_t	ip6_out_sw_cksum_bytes;
109 	kstat_named_t   ip6_in_sw_cksum;
110 	kstat_named_t	ip6_tcp_in_full_hw_cksum_err;
111 	kstat_named_t	ip6_tcp_in_part_hw_cksum_err;
112 	kstat_named_t	ip6_tcp_in_sw_cksum_err;
113 	kstat_named_t	ip6_udp_in_full_hw_cksum_err;
114 	kstat_named_t	ip6_udp_in_part_hw_cksum_err;
115 	kstat_named_t	ip6_udp_in_sw_cksum_err;
116 	kstat_named_t	ip6_frag_mdt_pkt_out;
117 	kstat_named_t	ip6_frag_mdt_discarded;
118 	kstat_named_t	ip6_frag_mdt_allocfail;
119 	kstat_named_t	ip6_frag_mdt_addpdescfail;
120 	kstat_named_t	ip6_frag_mdt_allocd;
121 } ip6_stat_t;
122 
123 typedef struct ire_stats {
124 	uint64_t ire_stats_alloced;	/* # of ires alloced */
125 	uint64_t ire_stats_freed;	/* # of ires freed */
126 	uint64_t ire_stats_inserted;	/* # of ires inserted in the bucket */
127 	uint64_t ire_stats_deleted;	/* # of ires deleted from the bucket */
128 } ire_stats_t;
129 
130 #define	TX_FANOUT_SIZE	128
131 #define	IDLHASHINDEX(X)	\
132 	((((uintptr_t)(X) >> 2) + ((uintptr_t)(X) >> 9)) & (TX_FANOUT_SIZE - 1))
133 
134 /* Data structure to represent addresses */
135 typedef struct srcid_map {
136 	struct srcid_map	*sm_next;
137 	in6_addr_t		sm_addr;	/* Local address */
138 	uint_t			sm_srcid;	/* source id */
139 	uint_t			sm_refcnt;	/* > 1 ipif with same addr? */
140 	zoneid_t		sm_zoneid;	/* zone id */
141 } srcid_map_t;
142 
143 /*
144  * IP stack instances
145  */
146 struct ip_stack {
147 	netstack_t	*ips_netstack;	/* Common netstack */
148 
149 	uint_t			ips_src_generation;	/* Both IPv4 and IPv6 */
150 
151 	struct mod_prop_info_s	*ips_propinfo_tbl; 	/* ip tunables table */
152 
153 	mib2_ipIfStatsEntry_t	ips_ip_mib;	/* SNMP fixed size info */
154 	mib2_icmp_t	ips_icmp_mib;
155 	/*
156 	 * IPv6 mibs when the interface (ill) is not known.
157 	 * When the ill is known the per-interface mib in the ill is used.
158 	 */
159 	mib2_ipIfStatsEntry_t	ips_ip6_mib;
160 	mib2_ipv6IfIcmpEntry_t	ips_icmp6_mib;
161 
162 	struct igmpstat		ips_igmpstat;
163 
164 	kstat_t		*ips_ip_mibkp;	/* kstat exporting ip_mib data */
165 	kstat_t		*ips_icmp_mibkp; /* kstat exporting icmp_mib data */
166 	kstat_t		*ips_ip_kstat;
167 	ip_stat_t	ips_ip_statistics;
168 	kstat_t		*ips_ip6_kstat;
169 	ip6_stat_t	ips_ip6_statistics;
170 
171 /* ip.c */
172 	kmutex_t	ips_igmp_timer_lock;
173 	kmutex_t	ips_mld_timer_lock;
174 	kmutex_t	ips_ip_mi_lock;
175 	kmutex_t	ips_ip_addr_avail_lock;
176 	krwlock_t	ips_ill_g_lock;
177 
178 	krwlock_t	ips_ill_g_usesrc_lock;
179 
180 	/* Taskq dispatcher for capability operations */
181 	kmutex_t	ips_capab_taskq_lock;
182 	kcondvar_t	ips_capab_taskq_cv;
183 	mblk_t		*ips_capab_taskq_head;
184 	mblk_t		*ips_capab_taskq_tail;
185 	kthread_t	*ips_capab_taskq_thread;
186 	boolean_t	ips_capab_taskq_quit;
187 
188 /* ipclassifier.c - keep in ip_stack_t */
189 	/* ipclassifier hash tables */
190 	struct connf_s	*ips_rts_clients;
191 	struct connf_s	*ips_ipcl_conn_fanout;
192 	struct connf_s	*ips_ipcl_bind_fanout;
193 	struct connf_s	*ips_ipcl_proto_fanout_v4;
194 	struct connf_s	*ips_ipcl_proto_fanout_v6;
195 	struct connf_s	*ips_ipcl_udp_fanout;
196 	struct connf_s	*ips_ipcl_raw_fanout;		/* RAW SCTP sockets */
197 	struct connf_s	*ips_ipcl_iptun_fanout;
198 	uint_t		ips_ipcl_conn_fanout_size;
199 	uint_t		ips_ipcl_bind_fanout_size;
200 	uint_t		ips_ipcl_udp_fanout_size;
201 	uint_t		ips_ipcl_raw_fanout_size;
202 	uint_t		ips_ipcl_iptun_fanout_size;
203 	struct connf_s	*ips_ipcl_globalhash_fanout;
204 	int		ips_conn_g_index;
205 
206 /* ip.c */
207 	/* Following protected by igmp_timer_lock */
208 	int 		ips_igmp_time_to_next;	/* Time since last timeout */
209 	int 		ips_igmp_timer_scheduled_last;
210 	int		ips_igmp_deferred_next;
211 	timeout_id_t	ips_igmp_timeout_id;
212 	boolean_t	ips_igmp_timer_setter_active;
213 
214 	/* Following protected by mld_timer_lock */
215 	int 		ips_mld_time_to_next;	/* Time since last timeout */
216 	int 		ips_mld_timer_scheduled_last;
217 	int		ips_mld_deferred_next;
218 	timeout_id_t	ips_mld_timeout_id;
219 	boolean_t	ips_mld_timer_setter_active;
220 
221 	/* Protected by igmp_slowtimeout_lock */
222 	timeout_id_t	ips_igmp_slowtimeout_id;
223 	kmutex_t	ips_igmp_slowtimeout_lock;
224 
225 	/* Protected by mld_slowtimeout_lock */
226 	timeout_id_t	ips_mld_slowtimeout_id;
227 	kmutex_t	ips_mld_slowtimeout_lock;
228 
229 	/* IPv4 forwarding table */
230 	struct radix_node_head *ips_ip_ftable;
231 
232 #define	IPV6_ABITS		128
233 #define	IP6_MASK_TABLE_SIZE	(IPV6_ABITS + 1)	/* 129 ptrs */
234 	struct irb	*ips_ip_forwarding_table_v6[IP6_MASK_TABLE_SIZE];
235 
236 	/*
237 	 * ire_ft_init_lock is used while initializing ip_forwarding_table
238 	 * dynamically in ire_add.
239 	 */
240 	kmutex_t	ips_ire_ft_init_lock;
241 
242 	/*
243 	 * This is the IPv6 counterpart of RADIX_NODE_HEAD_LOCK. It is used
244 	 * to prevent adds and deletes while we are doing a ftable_lookup
245 	 * and extracting the ire_generation.
246 	 */
247 	krwlock_t	ips_ip6_ire_head_lock;
248 
249 	uint32_t	ips_ip6_ftable_hash_size;
250 
251 	ire_stats_t 	ips_ire_stats_v4;	/* IPv4 ire statistics */
252 	ire_stats_t 	ips_ire_stats_v6;	/* IPv6 ire statistics */
253 
254 	/* Count how many condemned objects for kmem_cache callbacks */
255 	uint32_t	ips_num_ire_condemned;
256 	uint32_t	ips_num_nce_condemned;
257 	uint32_t	ips_num_dce_condemned;
258 
259 	struct ire_s	*ips_ire_reject_v4;	/* For unreachable dests */
260 	struct ire_s	*ips_ire_reject_v6;	/* For unreachable dests */
261 	struct ire_s	*ips_ire_blackhole_v4;	/* For temporary failures */
262 	struct ire_s	*ips_ire_blackhole_v6;	/* For temporary failures */
263 
264 	/* ips_ire_dep_lock protects ire_dep_* relationship between IREs */
265 	krwlock_t	ips_ire_dep_lock;
266 
267 	/* Destination Cache Entries */
268 	struct dce_s	*ips_dce_default;
269 	uint_t		ips_dce_hashsize;
270 	struct dcb_s	*ips_dce_hash_v4;
271 	struct dcb_s	*ips_dce_hash_v6;
272 
273 	/* pending binds */
274 	mblk_t		*ips_ip6_asp_pending_ops;
275 	mblk_t		*ips_ip6_asp_pending_ops_tail;
276 
277 	/* Synchronize updates with table usage */
278 	mblk_t		*ips_ip6_asp_pending_update; /* pending table updates */
279 
280 	boolean_t	ips_ip6_asp_uip;	/* table update in progress */
281 	kmutex_t	ips_ip6_asp_lock;	/* protect all the above */
282 	uint32_t	ips_ip6_asp_refcnt;	/* outstanding references */
283 
284 	struct ip6_asp	*ips_ip6_asp_table;
285 	/* The number of policy entries in the table */
286 	uint_t		ips_ip6_asp_table_count;
287 
288 	struct conn_s	*ips_ip_g_mrouter;
289 
290 	/* Time since last icmp_pkt_err */
291 	clock_t		ips_icmp_pkt_err_last;
292 	/* Number of packets sent in burst */
293 	uint_t		ips_icmp_pkt_err_sent;
294 
295 	/* Protected by ip_mi_lock */
296 	void		*ips_ip_g_head;	/* IP Instance Data List Head */
297 	void		*ips_arp_g_head; /* ARP Instance Data List Head */
298 
299 	/* Multirouting stuff */
300 	/* Interval (in ms) between consecutive 'bad MTU' warnings */
301 	hrtime_t	ips_ip_multirt_log_interval;
302 	/* Time since last warning issued. */
303 	hrtime_t	ips_multirt_bad_mtu_last_time;
304 
305 	/*
306 	 * CGTP hooks. Enabling and disabling of hooks is controlled by an
307 	 * IP tunable 'ips_ip_cgtp_filter'.
308 	 */
309 	struct cgtp_filter_ops *ips_ip_cgtp_filter_ops;
310 
311 	struct ipsq_s	*ips_ipsq_g_head;
312 	uint_t		ips_ill_index;	/* Used to assign interface indicies */
313 	/* When set search for unused index */
314 	boolean_t	ips_ill_index_wrap;
315 
316 	uint_t		ips_loopback_packets;
317 
318 	/* NDP/NCE structures for IPv4 and IPv6 */
319 	struct ndp_g_s	*ips_ndp4;
320 	struct ndp_g_s	*ips_ndp6;
321 
322 	/* ip_mroute stuff */
323 	kmutex_t	ips_ip_g_mrouter_mutex;
324 
325 	struct mrtstat	*ips_mrtstat;	/* Stats for netstat */
326 	int		ips_saved_ip_forwarding;
327 
328 	/* numvifs is only a hint about the max interface being used. */
329 	ushort_t	ips_numvifs;
330 	kmutex_t	ips_numvifs_mutex;
331 
332 	struct vif	*ips_vifs;
333 	struct mfcb	*ips_mfcs;	/* kernel routing table	*/
334 	struct tbf	*ips_tbfs;
335 	/*
336 	 * One-back cache used to locate a tunnel's vif,
337 	 * given a datagram's src ip address.
338 	 */
339 	ipaddr_t	ips_last_encap_src;
340 	struct vif	*ips_last_encap_vif;
341 	kmutex_t	ips_last_encap_lock;	/* Protects the above */
342 
343 	/*
344 	 * reg_vif_num is protected by numvifs_mutex
345 	 */
346 	/* Whether or not special PIM assert processing is enabled. */
347 	ushort_t	ips_reg_vif_num; 	/* Index to Register vif */
348 	int		ips_pim_assert;
349 
350 	union ill_g_head_u *ips_ill_g_heads;   /* ILL List Head */
351 
352 	kstat_t		*ips_loopback_ksp;
353 
354 	/* Array of conn drain lists */
355 	struct idl_tx_list_s	*ips_idl_tx_list;
356 	uint_t		ips_conn_drain_list_cnt; /* Count of conn_drain_list */
357 
358 	/*
359 	 * ID used to assign next free one.
360 	 * Increases by one. Once it wraps we search for an unused ID.
361 	 */
362 	uint_t		ips_ip_src_id;
363 	boolean_t	ips_srcid_wrapped;
364 
365 	struct srcid_map *ips_srcid_head;
366 	krwlock_t	ips_srcid_lock;
367 
368 	uint64_t	ips_ipif_g_seqid;	/* Used only for sctp_addr.c */
369 	union phyint_list_u *ips_phyint_g_list;	/* start of phyint list */
370 
371 /* ip_netinfo.c */
372 	hook_family_t	ips_ipv4root;
373 	hook_family_t	ips_ipv6root;
374 	hook_family_t	ips_arproot;
375 
376 	net_handle_t		ips_ipv4_net_data;
377 	net_handle_t		ips_ipv6_net_data;
378 	net_handle_t		ips_arp_net_data;
379 
380 	/*
381 	 * Hooks for firewalling
382 	 */
383 	hook_event_t		ips_ip4_physical_in_event;
384 	hook_event_t		ips_ip4_physical_out_event;
385 	hook_event_t		ips_ip4_forwarding_event;
386 	hook_event_t		ips_ip4_loopback_in_event;
387 	hook_event_t		ips_ip4_loopback_out_event;
388 
389 	hook_event_t		ips_ip6_physical_in_event;
390 	hook_event_t		ips_ip6_physical_out_event;
391 	hook_event_t		ips_ip6_forwarding_event;
392 	hook_event_t		ips_ip6_loopback_in_event;
393 	hook_event_t		ips_ip6_loopback_out_event;
394 
395 	hook_event_t		ips_arp_physical_in_event;
396 	hook_event_t		ips_arp_physical_out_event;
397 	hook_event_t		ips_arp_nic_events;
398 
399 	hook_event_token_t	ips_ipv4firewall_physical_in;
400 	hook_event_token_t	ips_ipv4firewall_physical_out;
401 	hook_event_token_t	ips_ipv4firewall_forwarding;
402 	hook_event_token_t	ips_ipv4firewall_loopback_in;
403 	hook_event_token_t	ips_ipv4firewall_loopback_out;
404 
405 	hook_event_token_t	ips_ipv6firewall_physical_in;
406 	hook_event_token_t	ips_ipv6firewall_physical_out;
407 	hook_event_token_t	ips_ipv6firewall_forwarding;
408 	hook_event_token_t	ips_ipv6firewall_loopback_in;
409 	hook_event_token_t	ips_ipv6firewall_loopback_out;
410 
411 	hook_event_t		ips_ip4_nic_events;
412 	hook_event_t		ips_ip6_nic_events;
413 	hook_event_token_t	ips_ipv4nicevents;
414 	hook_event_token_t	ips_ipv6nicevents;
415 
416 	hook_event_token_t	ips_arp_physical_in;
417 	hook_event_token_t	ips_arp_physical_out;
418 	hook_event_token_t	ips_arpnicevents;
419 
420 	net_handle_t		ips_ip4_observe_pr;
421 	net_handle_t		ips_ip6_observe_pr;
422 	hook_event_t		ips_ip4_observe;
423 	hook_event_t		ips_ip6_observe;
424 	hook_event_token_t	ips_ipv4observing;
425 	hook_event_token_t	ips_ipv6observing;
426 
427 	struct __ldi_ident	*ips_ldi_ident;
428 
429 /* ipmp.c */
430 	krwlock_t		ips_ipmp_lock;
431 	mod_hash_t		*ips_ipmp_grp_hash;
432 
433 };
434 typedef struct ip_stack ip_stack_t;
435 
436 /* Finding an ip_stack_t */
437 #define	CONNQ_TO_IPST(_q)	(Q_TO_CONN(_q)->conn_netstack->netstack_ip)
438 #define	ILLQ_TO_IPST(_q)	(((ill_t *)(_q)->q_ptr)->ill_ipst)
439 #define	PHYINT_TO_IPST(phyi)	((phyi)->phyint_ipsq->ipsq_ipst)
440 
441 #else /* _KERNEL */
442 typedef int ip_stack_t;
443 #endif /* _KERNEL */
444 
445 #ifdef	__cplusplus
446 }
447 #endif
448 
449 #endif	/* _INET_IP_STACK_H */
450