2 * Copyright (c) 2015-2016
3 * Alexander V. Chernikov <melifaro@FreeBSD.org>
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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.
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
32 #ifndef _NET_ROUTE_VAR_H_
33 #define _NET_ROUTE_VAR_H_
36 #include <net/radix.h>
39 #include <sys/epoch.h>
40 #include <netinet/in.h> /* struct sockaddr_in */
41 #include <sys/counter.h>
42 #include <net/route/nhop.h>
45 #define DPRINTF(_fmt, ...) printf("%s: " _fmt "\n", __func__ , ## __VA_ARGS__)
47 #define DPRINTF(_fmt, ...)
51 typedef int rnh_preadd_entry_f_t(u_int fibnum, const struct sockaddr *addr,
52 const struct sockaddr *mask, struct nhop_object *nh);
55 struct radix_head head;
56 rn_matchaddr_f_t *rnh_matchaddr; /* longest match for sockaddr */
57 rn_addaddr_f_t *rnh_addaddr; /* add based on sockaddr*/
58 rn_deladdr_f_t *rnh_deladdr; /* remove based on sockaddr */
59 rn_lookup_f_t *rnh_lookup; /* exact match for sockaddr */
60 rn_walktree_t *rnh_walktree; /* traverse tree */
61 rn_walktree_from_t *rnh_walktree_from; /* traverse tree below a */
62 rnh_preadd_entry_f_t *rnh_preadd; /* hook to alter record prior to insertion */
63 rt_gen_t rnh_gen; /* generation counter */
64 int rnh_multipath; /* multipath capable ? */
65 struct radix_node rnh_nodes[3]; /* empty tree for common case */
66 struct rmlock rib_lock; /* config/data path lock */
67 struct radix_mask_head rmhead; /* masks radix head */
68 struct vnet *rib_vnet; /* vnet pointer */
69 int rib_family; /* AF of the rtable */
70 u_int rib_fibnum; /* fib number */
71 struct callout expire_callout; /* Callout for expiring dynamic routes */
72 time_t next_expire; /* Next expire run ts */
73 struct nh_control *nh_control; /* nexthop subsystem data */
74 CK_STAILQ_HEAD(, rib_subscription) rnh_subscribers;/* notification subscribers */
77 #define RIB_RLOCK_TRACKER struct rm_priotracker _rib_tracker
78 #define RIB_LOCK_INIT(rh) rm_init(&(rh)->rib_lock, "rib head lock")
79 #define RIB_LOCK_DESTROY(rh) rm_destroy(&(rh)->rib_lock)
80 #define RIB_RLOCK(rh) rm_rlock(&(rh)->rib_lock, &_rib_tracker)
81 #define RIB_RUNLOCK(rh) rm_runlock(&(rh)->rib_lock, &_rib_tracker)
82 #define RIB_WLOCK(rh) rm_wlock(&(rh)->rib_lock)
83 #define RIB_WUNLOCK(rh) rm_wunlock(&(rh)->rib_lock)
84 #define RIB_LOCK_ASSERT(rh) rm_assert(&(rh)->rib_lock, RA_LOCKED)
85 #define RIB_WLOCK_ASSERT(rh) rm_assert(&(rh)->rib_lock, RA_WLOCKED)
88 #define RIB_MAX_RETRIES 3
89 #define RT_MAXFIBS UINT16_MAX
90 #define RIB_MAX_MPATH_WIDTH 64
92 /* Macro for verifying fields in af-specific 'struct route' structures */
93 #define CHK_STRUCT_FIELD_GENERIC(_s1, _f1, _s2, _f2) \
94 _Static_assert(sizeof(((_s1 *)0)->_f1) == sizeof(((_s2 *)0)->_f2), \
95 "Fields " #_f1 " and " #_f2 " size differs"); \
96 _Static_assert(__offsetof(_s1, _f1) == __offsetof(_s2, _f2), \
97 "Fields " #_f1 " and " #_f2 " offset differs");
99 #define _CHK_ROUTE_FIELD(_route_new, _field) \
100 CHK_STRUCT_FIELD_GENERIC(struct route, _field, _route_new, _field)
102 #define CHK_STRUCT_ROUTE_FIELDS(_route_new) \
103 _CHK_ROUTE_FIELD(_route_new, ro_nh) \
104 _CHK_ROUTE_FIELD(_route_new, ro_lle) \
105 _CHK_ROUTE_FIELD(_route_new, ro_prepend)\
106 _CHK_ROUTE_FIELD(_route_new, ro_plen) \
107 _CHK_ROUTE_FIELD(_route_new, ro_flags) \
108 _CHK_ROUTE_FIELD(_route_new, ro_mtu) \
109 _CHK_ROUTE_FIELD(_route_new, spare)
111 #define CHK_STRUCT_ROUTE_COMPAT(_ro_new, _dst_new) \
112 CHK_STRUCT_ROUTE_FIELDS(_ro_new); \
113 _Static_assert(__offsetof(struct route, ro_dst) == __offsetof(_ro_new, _dst_new),\
114 "ro_dst and " #_dst_new " are at different offset")
116 struct rib_head *rt_tables_get_rnh(uint32_t table, sa_family_t family);
117 int rt_getifa_fib(struct rt_addrinfo *info, u_int fibnum);
120 VNET_PCPUSTAT_DECLARE(struct rtstat, rtstat);
121 #define RTSTAT_ADD(name, val) \
122 VNET_PCPUSTAT_ADD(struct rtstat, rtstat, name, (val))
123 #define RTSTAT_INC(name) RTSTAT_ADD(name, 1)
126 * Convert a 'struct radix_node *' to a 'struct rtentry *'.
127 * The operation can be done safely (in this code) because a
128 * 'struct rtentry' starts with two 'struct radix_node''s, the first
129 * one representing leaf nodes in the routing tree, which is
130 * what the code in radix.c passes us as a 'struct radix_node'.
132 * But because there are a lot of assumptions in this conversion,
133 * do not cast explicitly, but always use the macro below.
135 #define RNTORT(p) ((struct rtentry *)(p))
138 struct radix_node rt_nodes[2]; /* tree glue, and other values */
140 * XXX struct rtentry must begin with a struct radix_node (or two!)
141 * because the code does some casts of a 'struct radix_node *'
142 * to a 'struct rtentry *'
144 #define rt_key(r) (*((struct sockaddr **)(&(r)->rt_nodes->rn_key)))
145 #define rt_mask(r) (*((struct sockaddr **)(&(r)->rt_nodes->rn_mask)))
146 #define rt_key_const(r) (*((const struct sockaddr * const *)(&(r)->rt_nodes->rn_key)))
147 #define rt_mask_const(r) (*((const struct sockaddr * const *)(&(r)->rt_nodes->rn_mask)))
150 * 2 radix_node structurs above consists of 2x6 pointers, leaving
151 * 4 pointers (32 bytes) of the second cache line on amd64.
154 struct nhop_object *rt_nhop; /* nexthop data */
157 * Destination address storage.
158 * sizeof(struct sockaddr_in6) == 28, however
159 * the dataplane-relevant part (e.g. address) lies
160 * at offset 8..24, making the address not crossing
161 * cacheline boundary.
163 struct sockaddr_in rt_dst4;
164 struct sockaddr_in6 rt_dst6;
165 struct sockaddr rt_dst;
169 int rte_flags; /* up/down?, host/net */
170 u_long rt_weight; /* absolute weight */
171 u_long rt_expire; /* lifetime for route, e.g. redirect */
172 struct rtentry *rt_chain; /* pointer to next rtentry to delete */
173 struct epoch_context rt_epoch_ctx; /* net epoch tracker */
177 * With the split between the routing entry and the nexthop,
178 * rt_flags has to be split between these 2 entries. As rtentry
179 * mostly contains prefix data and is thought to be generic enough
180 * so one can transparently change the nexthop pointer w/o requiring
181 * any other rtentry changes, most of rt_flags shifts to the particular nexthop.
184 * RTF_UP: rtentry, as an indication that it is linked.
185 * RTF_HOST: rtentry, nhop. The latter indication is needed for the datapath
186 * RTF_DYNAMIC: nhop, to make rtentry generic.
187 * RTF_MODIFIED: nhop, to make rtentry generic. (legacy)
188 * -- "native" path (nhop) properties:
189 * RTF_GATEWAY, RTF_STATIC, RTF_PROTO1, RTF_PROTO2, RTF_PROTO3, RTF_FIXEDMTU,
190 * RTF_PINNED, RTF_REJECT, RTF_BLACKHOLE, RTF_BROADCAST
193 /* Nexthop rt flags mask */
194 #define NHOP_RT_FLAG_MASK (RTF_GATEWAY | RTF_HOST | RTF_REJECT | RTF_DYNAMIC | \
195 RTF_MODIFIED | RTF_STATIC | RTF_BLACKHOLE | RTF_PROTO1 | RTF_PROTO2 | \
196 RTF_PROTO3 | RTF_FIXEDMTU | RTF_PINNED | RTF_BROADCAST)
198 /* rtentry rt flag mask */
199 #define RTE_RT_FLAG_MASK (RTF_UP | RTF_HOST)
201 /* route_temporal.c */
202 void tmproutes_update(struct rib_head *rnh, struct rtentry *rt);
203 void tmproutes_init(struct rib_head *rh);
204 void tmproutes_destroy(struct rib_head *rh);
207 struct route_nhop_data {
209 struct nhop_object *rnd_nhop;
210 struct nhgrp_object *rnd_nhgrp;
215 int change_route_nhop(struct rib_head *rnh, struct rtentry *rt,
216 struct rt_addrinfo *info, struct route_nhop_data *rnd,
217 struct rib_cmd_info *rc);
218 int change_route_conditional(struct rib_head *rnh, struct rtentry *rt,
219 struct rt_addrinfo *info, struct route_nhop_data *nhd_orig,
220 struct route_nhop_data *nhd_new, struct rib_cmd_info *rc);
221 struct rtentry *lookup_prefix(struct rib_head *rnh,
222 const struct rt_addrinfo *info, struct route_nhop_data *rnd);
224 bool nhop_can_multipath(const struct nhop_object *nh);
225 bool match_nhop_gw(const struct nhop_object *nh, const struct sockaddr *gw);
226 int check_info_match_nhop(const struct rt_addrinfo *info,
227 const struct rtentry *rt, const struct nhop_object *nh);
228 int can_override_nhop(const struct rt_addrinfo *info,
229 const struct nhop_object *nh);
231 void vnet_rtzone_init(void);
232 void vnet_rtzone_destroy(void);
235 void rib_init_subscriptions(struct rib_head *rnh);
236 void rib_destroy_subscriptions(struct rib_head *rnh);
239 void nhops_init(void);
240 int nhops_init_rib(struct rib_head *rh);
241 void nhops_destroy_rib(struct rib_head *rh);
242 void nhop_ref_object(struct nhop_object *nh);
243 int nhop_try_ref_object(struct nhop_object *nh);
244 void nhop_free_any(struct nhop_object *nh);
246 void nhop_set_type(struct nhop_object *nh, enum nhop_type nh_type);
247 void nhop_set_rtflags(struct nhop_object *nh, int rt_flags);
249 int nhop_create_from_info(struct rib_head *rnh, struct rt_addrinfo *info,
250 struct nhop_object **nh_ret);
251 int nhop_create_from_nhop(struct rib_head *rnh, const struct nhop_object *nh_orig,
252 struct rt_addrinfo *info, struct nhop_object **pnh_priv);
254 void nhops_update_ifmtu(struct rib_head *rh, struct ifnet *ifp, uint32_t mtu);
255 int nhops_dump_sysctl(struct rib_head *rh, struct sysctl_req *w);
258 #define MPF_MULTIPATH 0x08 /* need to be consistent with NHF_MULTIPATH */
260 struct nhgrp_object {
261 uint16_t nhg_flags; /* nexthop group flags */
262 uint8_t nhg_size; /* dataplain group size */
264 struct nhop_object *nhops[0]; /* nhops */
267 static inline struct nhop_object *
268 nhop_select(struct nhop_object *nh, uint32_t flowid)
272 if (NH_IS_NHGRP(nh)) {
273 struct nhgrp_object *nhg = (struct nhgrp_object *)nh;
274 nh = nhg->nhops[flowid % nhg->nhg_size];
281 struct weightened_nhop;
284 int add_route_mpath(struct rib_head *rnh, struct rt_addrinfo *info,
285 struct rtentry *rt, struct route_nhop_data *rnd_add,
286 struct route_nhop_data *rnd_orig, struct rib_cmd_info *rc);
287 int del_route_mpath(struct rib_head *rh, struct rt_addrinfo *info,
288 struct rtentry *rt, struct nhgrp_object *nhg, struct rib_cmd_info *rc);
291 int nhgrp_ctl_init(struct nh_control *ctl);
292 void nhgrp_ctl_free(struct nh_control *ctl);
293 void nhgrp_ctl_unlink_all(struct nh_control *ctl);
297 int nhgrp_dump_sysctl(struct rib_head *rh, struct sysctl_req *w);
299 int nhgrp_get_group(struct rib_head *rh, struct weightened_nhop *wn,
300 int num_nhops, struct route_nhop_data *rnd);
301 typedef bool nhgrp_filter_cb_t(const struct nhop_object *nh, void *data);
302 int nhgrp_get_filtered_group(struct rib_head *rh, const struct nhgrp_object *src,
303 nhgrp_filter_cb_t flt_func, void *flt_data, struct route_nhop_data *rnd);
304 int nhgrp_get_addition_group(struct rib_head *rnh,
305 struct route_nhop_data *rnd_orig, struct route_nhop_data *rnd_add,
306 struct route_nhop_data *rnd_new);
308 void nhgrp_free(struct nhgrp_object *nhg);