xref: /titanic_41/usr/src/uts/common/inet/ip_stack.h (revision fc724630b14603e4c1147df68b7bf45f7de7431f)
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 2009 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  * IP statistics.
43  */
44 #define	IP_STAT(ipst, x)	((ipst)->ips_ip_statistics.x.value.ui64++)
45 #define	IP_STAT_UPDATE(ipst, x, n) \
46 		((ipst)->ips_ip_statistics.x.value.ui64 += (n))
47 
48 typedef struct ip_stat {
49 	kstat_named_t	ipsec_fanout_proto;
50 	kstat_named_t	ip_udp_fannorm;
51 	kstat_named_t	ip_udp_fanmb;
52 	kstat_named_t	ip_udp_fanothers;
53 	kstat_named_t	ip_udp_fast_path;
54 	kstat_named_t	ip_udp_slow_path;
55 	kstat_named_t	ip_udp_input_err;
56 	kstat_named_t	ip_tcppullup;
57 	kstat_named_t	ip_tcpoptions;
58 	kstat_named_t	ip_multipkttcp;
59 	kstat_named_t	ip_tcp_fast_path;
60 	kstat_named_t	ip_tcp_slow_path;
61 	kstat_named_t	ip_tcp_input_error;
62 	kstat_named_t	ip_db_ref;
63 	kstat_named_t	ip_notaligned1;
64 	kstat_named_t	ip_notaligned2;
65 	kstat_named_t	ip_multimblk3;
66 	kstat_named_t	ip_multimblk4;
67 	kstat_named_t	ip_ipoptions;
68 	kstat_named_t	ip_classify_fail;
69 	kstat_named_t	ip_opt;
70 	kstat_named_t	ip_udp_rput_local;
71 	kstat_named_t	ipsec_proto_ahesp;
72 	kstat_named_t	ip_conn_flputbq;
73 	kstat_named_t	ip_conn_walk_drain;
74 	kstat_named_t   ip_out_sw_cksum;
75 	kstat_named_t   ip_in_sw_cksum;
76 	kstat_named_t   ip_trash_ire_reclaim_calls;
77 	kstat_named_t   ip_trash_ire_reclaim_success;
78 	kstat_named_t   ip_ire_arp_timer_expired;
79 	kstat_named_t   ip_ire_redirect_timer_expired;
80 	kstat_named_t	ip_ire_pmtu_timer_expired;
81 	kstat_named_t	ip_input_multi_squeue;
82 	kstat_named_t	ip_tcp_in_full_hw_cksum_err;
83 	kstat_named_t	ip_tcp_in_part_hw_cksum_err;
84 	kstat_named_t	ip_tcp_in_sw_cksum_err;
85 	kstat_named_t	ip_tcp_out_sw_cksum_bytes;
86 	kstat_named_t	ip_udp_in_full_hw_cksum_err;
87 	kstat_named_t	ip_udp_in_part_hw_cksum_err;
88 	kstat_named_t	ip_udp_in_sw_cksum_err;
89 	kstat_named_t	ip_udp_out_sw_cksum_bytes;
90 	kstat_named_t	ip_frag_mdt_pkt_out;
91 	kstat_named_t	ip_frag_mdt_discarded;
92 	kstat_named_t	ip_frag_mdt_allocfail;
93 	kstat_named_t	ip_frag_mdt_addpdescfail;
94 	kstat_named_t	ip_frag_mdt_allocd;
95 } ip_stat_t;
96 
97 
98 /*
99  * IP6 statistics.
100  */
101 #define	IP6_STAT(ipst, x)	((ipst)->ips_ip6_statistics.x.value.ui64++)
102 #define	IP6_STAT_UPDATE(ipst, x, n)	\
103 	((ipst)->ips_ip6_statistics.x.value.ui64 += (n))
104 
105 typedef struct ip6_stat {
106 	kstat_named_t	ip6_udp_fast_path;
107 	kstat_named_t	ip6_udp_slow_path;
108 	kstat_named_t	ip6_udp_fannorm;
109 	kstat_named_t	ip6_udp_fanmb;
110 	kstat_named_t   ip6_out_sw_cksum;
111 	kstat_named_t   ip6_in_sw_cksum;
112 	kstat_named_t	ip6_tcp_in_full_hw_cksum_err;
113 	kstat_named_t	ip6_tcp_in_part_hw_cksum_err;
114 	kstat_named_t	ip6_tcp_in_sw_cksum_err;
115 	kstat_named_t	ip6_tcp_out_sw_cksum_bytes;
116 	kstat_named_t	ip6_udp_in_full_hw_cksum_err;
117 	kstat_named_t	ip6_udp_in_part_hw_cksum_err;
118 	kstat_named_t	ip6_udp_in_sw_cksum_err;
119 	kstat_named_t	ip6_udp_out_sw_cksum_bytes;
120 	kstat_named_t	ip6_frag_mdt_pkt_out;
121 	kstat_named_t	ip6_frag_mdt_discarded;
122 	kstat_named_t	ip6_frag_mdt_allocfail;
123 	kstat_named_t	ip6_frag_mdt_addpdescfail;
124 	kstat_named_t	ip6_frag_mdt_allocd;
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 /*
139  * IP stack instances
140  */
141 struct ip_stack {
142 	netstack_t	*ips_netstack;	/* Common netstack */
143 
144 	struct ipparam_s	*ips_param_arr; 	/* ndd variable table */
145 	struct ipndp_s		*ips_ndp_arr;
146 
147 	mib2_ipIfStatsEntry_t	ips_ip_mib;	/* SNMP fixed size info */
148 	mib2_icmp_t	ips_icmp_mib;
149 	/*
150 	 * IPv6 mibs when the interface (ill) is not known.
151 	 * When the ill is known the per-interface mib in the ill is used.
152 	 */
153 	mib2_ipIfStatsEntry_t	ips_ip6_mib;
154 	mib2_ipv6IfIcmpEntry_t	ips_icmp6_mib;
155 
156 	struct igmpstat		ips_igmpstat;
157 
158 	kstat_t		*ips_ip_mibkp;	/* kstat exporting ip_mib data */
159 	kstat_t		*ips_icmp_mibkp; /* kstat exporting icmp_mib data */
160 	kstat_t		*ips_ip_kstat;
161 	ip_stat_t	ips_ip_statistics;
162 	kstat_t		*ips_ip6_kstat;
163 	ip6_stat_t	ips_ip6_statistics;
164 
165 /* ip.c */
166 	krwlock_t	ips_ip_g_nd_lock;
167 	kmutex_t	ips_igmp_timer_lock;
168 	kmutex_t	ips_mld_timer_lock;
169 	kmutex_t	ips_ip_mi_lock;
170 	kmutex_t	ips_ip_addr_avail_lock;
171 	krwlock_t	ips_ill_g_lock;
172 	krwlock_t	ips_ipsec_capab_ills_lock;
173 				/* protects the list of IPsec capable ills */
174 	struct ipsec_capab_ill_s *ips_ipsec_capab_ills_ah;
175 	struct ipsec_capab_ill_s *ips_ipsec_capab_ills_esp;
176 
177 	krwlock_t	ips_ill_g_usesrc_lock;
178 
179 	/* Taskq dispatcher for capability operations */
180 	kmutex_t	ips_capab_taskq_lock;
181 	kcondvar_t	ips_capab_taskq_cv;
182 	list_t		ips_capab_taskq_list;
183 	kthread_t	*ips_capab_taskq_thread;
184 	boolean_t	ips_capab_taskq_quit;
185 
186 /* ipclassifier.c - keep in ip_stack_t */
187 	/* ipclassifier hash tables */
188 	struct connf_s	*ips_rts_clients;
189 	struct connf_s	*ips_ipcl_conn_fanout;
190 	struct connf_s	*ips_ipcl_bind_fanout;
191 	struct connf_s	*ips_ipcl_proto_fanout;
192 	struct connf_s	*ips_ipcl_proto_fanout_v6;
193 	struct connf_s	*ips_ipcl_udp_fanout;
194 	struct connf_s	*ips_ipcl_raw_fanout;
195 	uint_t		ips_ipcl_conn_fanout_size;
196 	uint_t		ips_ipcl_bind_fanout_size;
197 	uint_t		ips_ipcl_udp_fanout_size;
198 	uint_t		ips_ipcl_raw_fanout_size;
199 	struct connf_s	*ips_ipcl_globalhash_fanout;
200 	int		ips_conn_g_index;
201 
202 /* ip.c */
203 	/* Following protected by igmp_timer_lock */
204 	int 		ips_igmp_time_to_next;	/* Time since last timeout */
205 	int 		ips_igmp_timer_scheduled_last;
206 	int		ips_igmp_deferred_next;
207 	timeout_id_t	ips_igmp_timeout_id;
208 	boolean_t	ips_igmp_timer_setter_active;
209 
210 	/* Following protected by mld_timer_lock */
211 	int 		ips_mld_time_to_next;	/* Time since last timeout */
212 	int 		ips_mld_timer_scheduled_last;
213 	int		ips_mld_deferred_next;
214 	timeout_id_t	ips_mld_timeout_id;
215 	boolean_t	ips_mld_timer_setter_active;
216 
217 	/* Protected by igmp_slowtimeout_lock */
218 	timeout_id_t	ips_igmp_slowtimeout_id;
219 	kmutex_t	ips_igmp_slowtimeout_lock;
220 
221 	/* Protected by mld_slowtimeout_lock */
222 	timeout_id_t	ips_mld_slowtimeout_id;
223 	kmutex_t	ips_mld_slowtimeout_lock;
224 
225 	/* IPv4 forwarding table */
226 	struct radix_node_head *ips_ip_ftable;
227 
228 	/* This is dynamically allocated in ip_ire_init */
229 	struct irb	 *ips_ip_cache_table;
230 
231 #define	IPV6_ABITS		128
232 #define	IP6_MASK_TABLE_SIZE	(IPV6_ABITS + 1)	/* 129 ptrs */
233 
234 	struct irb	*ips_ip_forwarding_table_v6[IP6_MASK_TABLE_SIZE];
235 	/* This is dynamically allocated in ip_ire_init */
236 	struct irb	*ips_ip_cache_table_v6;
237 
238 	uint32_t	ips_ire_handle;
239 	/*
240 	 * ire_ft_init_lock is used while initializing ip_forwarding_table
241 	 * dynamically in ire_add.
242 	 */
243 	kmutex_t	ips_ire_ft_init_lock;
244 	kmutex_t	ips_ire_handle_lock;	/* Protects ire_handle */
245 
246 	uint32_t	ips_ip_cache_table_size;
247 	uint32_t	ips_ip6_cache_table_size;
248 	uint32_t	ips_ip6_ftable_hash_size;
249 
250 	ire_stats_t 	ips_ire_stats_v4;	/* IPv4 ire statistics */
251 	ire_stats_t 	ips_ire_stats_v6;	/* IPv6 ire statistics */
252 
253 	/* pending binds */
254 	mblk_t		*ips_ip6_asp_pending_ops;
255 	mblk_t		*ips_ip6_asp_pending_ops_tail;
256 
257 	/* Synchronize updates with table usage */
258 	mblk_t		*ips_ip6_asp_pending_update; /* pending table updates */
259 
260 	boolean_t	ips_ip6_asp_uip;	/* table update in progress */
261 	kmutex_t	ips_ip6_asp_lock;	/* protect all the above */
262 	uint32_t	ips_ip6_asp_refcnt;	/* outstanding references */
263 
264 	struct ip6_asp	*ips_ip6_asp_table;
265 	/* The number of policy entries in the table */
266 	uint_t		ips_ip6_asp_table_count;
267 
268 	int		ips_ip_g_forward;
269 	int		ips_ipv6_forward;
270 
271 	time_t		ips_ip_g_frag_timeout;
272 	clock_t		ips_ip_g_frag_timo_ms;
273 
274 	struct conn_s	*ips_ip_g_mrouter;
275 
276 	/* Time since last icmp_pkt_err */
277 	clock_t		ips_icmp_pkt_err_last;
278 	/* Number of packets sent in burst */
279 	uint_t		ips_icmp_pkt_err_sent;
280 
281 	/* Protected by ip_mi_lock */
282 	void		*ips_ip_g_head;		/* Instance Data List Head */
283 
284 	caddr_t		ips_ip_g_nd;		/* Named Dispatch List Head */
285 
286 	/* Multirouting stuff */
287 	/* Interval (in ms) between consecutive 'bad MTU' warnings */
288 	hrtime_t	ips_ip_multirt_log_interval;
289 	/* Time since last warning issued. */
290 	hrtime_t	ips_multirt_bad_mtu_last_time;
291 
292 	struct cgtp_filter_ops *ips_ip_cgtp_filter_ops;	/* CGTP hooks */
293 	boolean_t	ips_ip_cgtp_filter;	/* Enable/disable CGTP hooks */
294 
295 	kmutex_t	ips_ip_trash_timer_lock;
296 	timeout_id_t	ips_ip_ire_expire_id;	/* IRE expiration timer. */
297 	struct ipsq_s	*ips_ipsq_g_head;
298 	uint_t		ips_ill_index;	/* Used to assign interface indicies */
299 	/* When set search for unused index */
300 	boolean_t	ips_ill_index_wrap;
301 
302 	clock_t		ips_ip_ire_arp_time_elapsed;
303 			/* Time since IRE cache last flushed */
304 	clock_t		ips_ip_ire_rd_time_elapsed;
305 			/* ... redirect IREs last flushed */
306 	clock_t		ips_ip_ire_pmtu_time_elapsed;
307 			/* Time since path mtu increase */
308 
309 	uint_t		ips_ip_redirect_cnt;
310 			/* Num of redirect routes in ftable */
311 	uint_t		ips_ipv6_ire_default_count;
312 			/* Number of IPv6 IRE_DEFAULT entries */
313 	uint_t		ips_ipv6_ire_default_index;
314 			/* Walking IPv6 index used to mod in */
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_g_forward;
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;
369 	union phyint_list_u *ips_phyint_g_list;	/* start of phyint list */
370 
371 /* ip_neti.c */
372 	hook_family_t	ips_ipv4root;
373 	hook_family_t	ips_ipv6root;
374 
375 	/*
376 	 * Hooks for firewalling
377 	 */
378 	hook_event_t		ips_ip4_physical_in_event;
379 	hook_event_t		ips_ip4_physical_out_event;
380 	hook_event_t		ips_ip4_forwarding_event;
381 	hook_event_t		ips_ip4_loopback_in_event;
382 	hook_event_t		ips_ip4_loopback_out_event;
383 	hook_event_t		ips_ip4_nic_events;
384 	hook_event_t		ips_ip6_physical_in_event;
385 	hook_event_t		ips_ip6_physical_out_event;
386 	hook_event_t		ips_ip6_forwarding_event;
387 	hook_event_t		ips_ip6_loopback_in_event;
388 	hook_event_t		ips_ip6_loopback_out_event;
389 	hook_event_t		ips_ip6_nic_events;
390 
391 	hook_event_token_t	ips_ipv4firewall_physical_in;
392 	hook_event_token_t	ips_ipv4firewall_physical_out;
393 	hook_event_token_t	ips_ipv4firewall_forwarding;
394 	hook_event_token_t	ips_ipv4firewall_loopback_in;
395 	hook_event_token_t	ips_ipv4firewall_loopback_out;
396 	hook_event_token_t	ips_ipv4nicevents;
397 	hook_event_token_t	ips_ipv6firewall_physical_in;
398 	hook_event_token_t	ips_ipv6firewall_physical_out;
399 	hook_event_token_t	ips_ipv6firewall_forwarding;
400 	hook_event_token_t	ips_ipv6firewall_loopback_in;
401 	hook_event_token_t	ips_ipv6firewall_loopback_out;
402 	hook_event_token_t	ips_ipv6nicevents;
403 
404 	net_handle_t		ips_ipv4_net_data;
405 	net_handle_t		ips_ipv6_net_data;
406 
407 	boolean_t		ips_ipobs_enabled;
408 	list_t			ips_ipobs_cb_list;
409 	kmutex_t		ips_ipobs_cb_lock;
410 	uint_t			ips_ipobs_cb_nwalkers;
411 	kcondvar_t		ips_ipobs_cb_cv;
412 
413 	struct __ldi_ident	*ips_ldi_ident;
414 
415 /* ipmp.c */
416 	krwlock_t		ips_ipmp_lock;
417 	mod_hash_t		*ips_ipmp_grp_hash;
418 
419 /* igmp.c */
420 	/* multicast restart timers thread logic */
421 	kmutex_t		ips_mrt_lock;
422 	uint_t			ips_mrt_flags;
423 	kcondvar_t		ips_mrt_cv;
424 	kcondvar_t		ips_mrt_done_cv;
425 	kthread_t		*ips_mrt_thread;
426 };
427 typedef struct ip_stack ip_stack_t;
428 
429 /* Finding an ip_stack_t */
430 #define	CONNQ_TO_IPST(_q)	(Q_TO_CONN(_q)->conn_netstack->netstack_ip)
431 #define	ILLQ_TO_IPST(_q)	(((ill_t *)(_q)->q_ptr)->ill_ipst)
432 #define	PHYINT_TO_IPST(phyi)	((phyi)->phyint_ipsq->ipsq_ipst)
433 
434 #else /* _KERNEL */
435 typedef int ip_stack_t;
436 #endif /* _KERNEL */
437 
438 #ifdef	__cplusplus
439 }
440 #endif
441 
442 #endif	/* _INET_IP_STACK_H */
443