master
1/*-
2 * Copyright (c) 2015-2016
3 * Alexander V. Chernikov <melifaro@FreeBSD.org>
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of the University nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 */
29
30#ifndef _NET_ROUTE_VAR_H_
31#define _NET_ROUTE_VAR_H_
32
33#ifndef RNF_NORMAL
34#include <net/radix.h>
35#endif
36#include <sys/ck.h>
37#include <sys/epoch.h>
38#include <netinet/in.h> /* struct sockaddr_in */
39#include <sys/counter.h>
40#include <net/route/nhop.h>
41
42struct nh_control;
43/* Sets prefix-specific nexthop flags (NHF_DEFAULT, RTF/NHF_HOST, RTF_BROADCAST,..) */
44typedef int rnh_set_nh_pfxflags_f_t(u_int fibnum, const struct sockaddr *addr,
45 const struct sockaddr *mask, struct nhop_object *nh);
46/* Fills in family-specific details that are not yet set up (mtu, nhop type, ..) */
47typedef int rnh_augment_nh_f_t(u_int fibnum, struct nhop_object *nh);
48
49struct rib_head {
50 struct radix_head head;
51 rn_matchaddr_f_t *rnh_matchaddr; /* longest match for sockaddr */
52 rn_addaddr_f_t *rnh_addaddr; /* add based on sockaddr*/
53 rn_deladdr_f_t *rnh_deladdr; /* remove based on sockaddr */
54 rn_lookup_f_t *rnh_lookup; /* exact match for sockaddr */
55 rn_walktree_t *rnh_walktree; /* traverse tree */
56 rn_walktree_from_t *rnh_walktree_from; /* traverse tree below a */
57 rnh_set_nh_pfxflags_f_t *rnh_set_nh_pfxflags; /* hook to alter record prior to insertion */
58 rt_gen_t rnh_gen; /* datapath generation counter */
59 int rnh_multipath; /* multipath capable ? */
60 struct radix_node rnh_nodes[3]; /* empty tree for common case */
61 struct rmlock rib_lock; /* config/data path lock */
62 struct radix_mask_head rmhead; /* masks radix head */
63 struct vnet *rib_vnet; /* vnet pointer */
64 int rib_family; /* AF of the rtable */
65 u_int rib_fibnum; /* fib number */
66 struct callout expire_callout; /* Callout for expiring dynamic routes */
67 time_t next_expire; /* Next expire run ts */
68 uint32_t rnh_prefixes; /* Number of prefixes */
69 rt_gen_t rnh_gen_rib; /* fib algo: rib generation counter */
70 uint32_t rib_dying:1; /* rib is detaching */
71 uint32_t rib_algo_fixed:1;/* fixed algorithm */
72 uint32_t rib_algo_init:1;/* algo init done */
73 struct nh_control *nh_control; /* nexthop subsystem data */
74 rnh_augment_nh_f_t *rnh_augment_nh;/* hook to alter nexthop prior to insertion */
75 CK_STAILQ_HEAD(, rib_subscription) rnh_subscribers;/* notification subscribers */
76};
77
78#define RIB_RLOCK_TRACKER struct rm_priotracker _rib_tracker
79#define RIB_LOCK_INIT(rh) rm_init_flags(&(rh)->rib_lock, "rib head lock", RM_DUPOK)
80#define RIB_LOCK_DESTROY(rh) rm_destroy(&(rh)->rib_lock)
81#define RIB_RLOCK(rh) rm_rlock(&(rh)->rib_lock, &_rib_tracker)
82#define RIB_RUNLOCK(rh) rm_runlock(&(rh)->rib_lock, &_rib_tracker)
83#define RIB_WLOCK(rh) rm_wlock(&(rh)->rib_lock)
84#define RIB_WUNLOCK(rh) rm_wunlock(&(rh)->rib_lock)
85#define RIB_LOCK_ASSERT(rh) rm_assert(&(rh)->rib_lock, RA_LOCKED)
86#define RIB_WLOCK_ASSERT(rh) rm_assert(&(rh)->rib_lock, RA_WLOCKED)
87
88/* Constants */
89#define RIB_MAX_RETRIES 3
90#define RT_MAXFIBS UINT16_MAX
91#define RIB_MAX_MPATH_WIDTH 64
92
93/* Macro for verifying fields in af-specific 'struct route' structures */
94#define CHK_STRUCT_FIELD_GENERIC(_s1, _f1, _s2, _f2) \
95_Static_assert(sizeof(((_s1 *)0)->_f1) == sizeof(((_s2 *)0)->_f2), \
96 "Fields " #_f1 " and " #_f2 " size differs"); \
97_Static_assert(__offsetof(_s1, _f1) == __offsetof(_s2, _f2), \
98 "Fields " #_f1 " and " #_f2 " offset differs");
99
100#define _CHK_ROUTE_FIELD(_route_new, _field) \
101 CHK_STRUCT_FIELD_GENERIC(struct route, _field, _route_new, _field)
102
103#define CHK_STRUCT_ROUTE_FIELDS(_route_new) \
104 _CHK_ROUTE_FIELD(_route_new, ro_nh) \
105 _CHK_ROUTE_FIELD(_route_new, ro_lle) \
106 _CHK_ROUTE_FIELD(_route_new, ro_prepend)\
107 _CHK_ROUTE_FIELD(_route_new, ro_plen) \
108 _CHK_ROUTE_FIELD(_route_new, ro_flags) \
109 _CHK_ROUTE_FIELD(_route_new, ro_mtu) \
110 _CHK_ROUTE_FIELD(_route_new, spare)
111
112#define CHK_STRUCT_ROUTE_COMPAT(_ro_new, _dst_new) \
113CHK_STRUCT_ROUTE_FIELDS(_ro_new); \
114_Static_assert(__offsetof(struct route, ro_dst) == __offsetof(_ro_new, _dst_new),\
115 "ro_dst and " #_dst_new " are at different offset")
116
117static inline void
118rib_bump_gen(struct rib_head *rnh)
119{
120#ifdef FIB_ALGO
121 rnh->rnh_gen_rib++;
122#else
123 rnh->rnh_gen++;
124#endif
125}
126
127struct rib_head *rt_tables_get_rnh(uint32_t table, sa_family_t family);
128int rt_getifa_fib(struct rt_addrinfo *info, u_int fibnum);
129struct rib_cmd_info;
130
131VNET_PCPUSTAT_DECLARE(struct rtstat, rtstat);
132#define RTSTAT_ADD(name, val) \
133 VNET_PCPUSTAT_ADD(struct rtstat, rtstat, name, (val))
134#define RTSTAT_INC(name) RTSTAT_ADD(name, 1)
135
136/*
137 * Convert a 'struct radix_node *' to a 'struct rtentry *'.
138 * The operation can be done safely (in this code) because a
139 * 'struct rtentry' starts with two 'struct radix_node''s, the first
140 * one representing leaf nodes in the routing tree, which is
141 * what the code in radix.c passes us as a 'struct radix_node'.
142 *
143 * But because there are a lot of assumptions in this conversion,
144 * do not cast explicitly, but always use the macro below.
145 */
146#define RNTORT(p) ((struct rtentry *)(p))
147
148struct rtentry {
149 struct radix_node rt_nodes[2]; /* tree glue, and other values */
150 /*
151 * XXX struct rtentry must begin with a struct radix_node (or two!)
152 * because the code does some casts of a 'struct radix_node *'
153 * to a 'struct rtentry *'
154 */
155#define rt_key(r) (*((struct sockaddr **)(&(r)->rt_nodes->rn_key)))
156#define rt_mask(r) (*((struct sockaddr **)(&(r)->rt_nodes->rn_mask)))
157#define rt_key_const(r) (*((const struct sockaddr * const *)(&(r)->rt_nodes->rn_key)))
158#define rt_mask_const(r) (*((const struct sockaddr * const *)(&(r)->rt_nodes->rn_mask)))
159
160 /*
161 * 2 radix_node structurs above consists of 2x6 pointers, leaving
162 * 4 pointers (32 bytes) of the second cache line on amd64.
163 *
164 */
165 struct nhop_object *rt_nhop; /* nexthop data */
166 union {
167 /*
168 * Destination address storage.
169 * sizeof(struct sockaddr_in6) == 28, however
170 * the dataplane-relevant part (e.g. address) lies
171 * at offset 8..24, making the address not crossing
172 * cacheline boundary.
173 */
174 struct sockaddr_in rt_dst4;
175 struct sockaddr_in6 rt_dst6;
176 struct sockaddr rt_dst;
177 char rt_dstb[28];
178 };
179
180 int rte_flags; /* up/down?, host/net */
181 u_long rt_weight; /* absolute weight */
182 struct rtentry *rt_chain; /* pointer to next rtentry to delete */
183 struct epoch_context rt_epoch_ctx; /* net epoch tracker */
184};
185
186/*
187 * With the split between the routing entry and the nexthop,
188 * rt_flags has to be split between these 2 entries. As rtentry
189 * mostly contains prefix data and is thought to be generic enough
190 * so one can transparently change the nexthop pointer w/o requiring
191 * any other rtentry changes, most of rt_flags shifts to the particular nexthop.
192 * /
193 *
194 * RTF_UP: rtentry, as an indication that it is linked.
195 * RTF_HOST: rtentry, nhop. The latter indication is needed for the datapath
196 * RTF_DYNAMIC: nhop, to make rtentry generic.
197 * RTF_MODIFIED: nhop, to make rtentry generic. (legacy)
198 * -- "native" path (nhop) properties:
199 * RTF_GATEWAY, RTF_STATIC, RTF_PROTO1, RTF_PROTO2, RTF_PROTO3, RTF_FIXEDMTU,
200 * RTF_PINNED, RTF_REJECT, RTF_BLACKHOLE, RTF_BROADCAST
201 */
202
203/* rtentry rt flag mask */
204#define RTE_RT_FLAG_MASK (RTF_UP | RTF_HOST)
205
206/* route_temporal.c */
207void tmproutes_update(struct rib_head *rnh, struct rtentry *rt, struct nhop_object *nh);
208void tmproutes_init(struct rib_head *rh);
209void tmproutes_destroy(struct rib_head *rh);
210
211/* route_ctl.c */
212struct route_nhop_data;
213int change_route(struct rib_head *rnh, struct rtentry *rt,
214 struct route_nhop_data *rnd, struct rib_cmd_info *rc);
215int change_route_conditional(struct rib_head *rnh, struct rtentry *rt,
216 struct route_nhop_data *nhd_orig, struct route_nhop_data *nhd_new,
217 struct rib_cmd_info *rc);
218struct rtentry *lookup_prefix(struct rib_head *rnh,
219 const struct rt_addrinfo *info, struct route_nhop_data *rnd);
220struct rtentry *lookup_prefix_rt(struct rib_head *rnh, const struct rtentry *rt,
221 struct route_nhop_data *rnd);
222int rib_copy_route(struct rtentry *rt, const struct route_nhop_data *rnd_src,
223 struct rib_head *rh_dst, struct rib_cmd_info *rc);
224
225bool nhop_can_multipath(const struct nhop_object *nh);
226bool match_nhop_gw(const struct nhop_object *nh, const struct sockaddr *gw);
227int check_info_match_nhop(const struct rt_addrinfo *info,
228 const struct rtentry *rt, const struct nhop_object *nh);
229bool rib_can_4o6_nhop(void);
230
231/* route_rtentry.c */
232void vnet_rtzone_init(void);
233void vnet_rtzone_destroy(void);
234void rt_free(struct rtentry *rt);
235void rt_free_immediate(struct rtentry *rt);
236struct rtentry *rt_alloc(struct rib_head *rnh, const struct sockaddr *dst,
237 struct sockaddr *netmask);
238
239/* subscriptions */
240void rib_init_subscriptions(struct rib_head *rnh);
241void rib_destroy_subscriptions(struct rib_head *rnh);
242
243/* route_ifaddrs.c */
244void rib_copy_kernel_routes(struct rib_head *rh_src, struct rib_head *rh_dst);
245
246/* Nexhops */
247void nhops_init(void);
248int nhops_init_rib(struct rib_head *rh);
249void nhops_destroy_rib(struct rib_head *rh);
250void nhop_ref_object(struct nhop_object *nh);
251int nhop_try_ref_object(struct nhop_object *nh);
252void nhop_ref_any(struct nhop_object *nh);
253void nhop_free_any(struct nhop_object *nh);
254struct nhop_object *nhop_get_nhop_internal(struct rib_head *rnh,
255 struct nhop_object *nh, int *perror);
256
257bool nhop_check_gateway(int upper_family, int neigh_family);
258
259int nhop_create_from_info(struct rib_head *rnh, struct rt_addrinfo *info,
260 struct nhop_object **nh_ret);
261int nhop_create_from_nhop(struct rib_head *rnh, const struct nhop_object *nh_orig,
262 struct rt_addrinfo *info, struct nhop_object **pnh_priv);
263
264void nhops_update_ifmtu(struct rib_head *rh, struct ifnet *ifp, uint32_t mtu);
265int nhops_dump_sysctl(struct rib_head *rh, struct sysctl_req *w);
266
267/* MULTIPATH */
268#define MPF_MULTIPATH 0x08 /* need to be consistent with NHF_MULTIPATH */
269
270struct nhgrp_object {
271 uint16_t nhg_flags; /* nexthop group flags */
272 uint8_t nhg_size; /* dataplain group size */
273 uint8_t spare;
274 struct nhop_object *nhops[0]; /* nhops */
275};
276
277static inline struct nhop_object *
278nhop_select(struct nhop_object *nh, uint32_t flowid)
279{
280
281#ifdef ROUTE_MPATH
282 if (NH_IS_NHGRP(nh)) {
283 struct nhgrp_object *nhg = (struct nhgrp_object *)nh;
284 nh = nhg->nhops[flowid % nhg->nhg_size];
285 }
286#endif
287 return (nh);
288}
289
290
291struct weightened_nhop;
292
293/* mpath_ctl.c */
294int add_route_mpath(struct rib_head *rnh, struct rt_addrinfo *info,
295 struct rtentry *rt, struct route_nhop_data *rnd_add,
296 struct route_nhop_data *rnd_orig, struct rib_cmd_info *rc);
297
298/* nhgrp.c */
299int nhgrp_ctl_init(struct nh_control *ctl);
300void nhgrp_ctl_free(struct nh_control *ctl);
301void nhgrp_ctl_unlink_all(struct nh_control *ctl);
302
303
304/* nhgrp_ctl.c */
305int nhgrp_dump_sysctl(struct rib_head *rh, struct sysctl_req *w);
306
307int nhgrp_get_filtered_group(struct rib_head *rh, const struct rtentry *rt,
308 const struct nhgrp_object *src, rib_filter_f_t flt_func, void *flt_data,
309 struct route_nhop_data *rnd);
310int nhgrp_get_addition_group(struct rib_head *rnh,
311 struct route_nhop_data *rnd_orig, struct route_nhop_data *rnd_add,
312 struct route_nhop_data *rnd_new);
313
314void nhgrp_ref_object(struct nhgrp_object *nhg);
315uint32_t nhgrp_get_idx(const struct nhgrp_object *nhg);
316void nhgrp_free(struct nhgrp_object *nhg);
317
318/* rtsock */
319int rtsock_routemsg(int cmd, struct rtentry *rt, struct nhop_object *nh,
320 int fibnum);
321int rtsock_routemsg_info(int cmd, struct rt_addrinfo *info, int fibnum);
322int rtsock_addrmsg(int cmd, struct ifaddr *ifa, int fibnum);
323
324
325/* lookup_framework.c */
326void fib_grow_rtables(uint32_t new_num_tables);
327void fib_setup_family(int family, uint32_t num_tables);
328void fib_destroy_rib(struct rib_head *rh);
329void vnet_fib_init(void);
330void vnet_fib_destroy(void);
331
332/* Entropy data used for outbound hashing */
333#define MPATH_ENTROPY_KEY_LEN 40
334extern uint8_t mpath_entropy_key[MPATH_ENTROPY_KEY_LEN];
335
336#endif