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>
38 #include <netinet/in.h> /* struct sockaddr_in */
39 #include <sys/counter.h>
42 typedef int rnh_preadd_entry_f_t(u_int fibnum, const struct sockaddr *addr,
43 const struct sockaddr *mask, struct nhop_object *nh);
46 struct radix_head head;
47 rn_matchaddr_f_t *rnh_matchaddr; /* longest match for sockaddr */
48 rn_addaddr_f_t *rnh_addaddr; /* add based on sockaddr*/
49 rn_deladdr_f_t *rnh_deladdr; /* remove based on sockaddr */
50 rn_lookup_f_t *rnh_lookup; /* exact match for sockaddr */
51 rn_walktree_t *rnh_walktree; /* traverse tree */
52 rn_walktree_from_t *rnh_walktree_from; /* traverse tree below a */
53 rnh_preadd_entry_f_t *rnh_preadd; /* hook to alter record prior to insertion */
54 rt_gen_t rnh_gen; /* generation counter */
55 int rnh_multipath; /* multipath capable ? */
56 struct radix_node rnh_nodes[3]; /* empty tree for common case */
57 struct rmlock rib_lock; /* config/data path lock */
58 struct radix_mask_head rmhead; /* masks radix head */
59 struct vnet *rib_vnet; /* vnet pointer */
60 int rib_family; /* AF of the rtable */
61 u_int rib_fibnum; /* fib number */
62 struct callout expire_callout; /* Callout for expiring dynamic routes */
63 time_t next_expire; /* Next expire run ts */
64 struct nh_control *nh_control; /* nexthop subsystem data */
67 #define RIB_RLOCK_TRACKER struct rm_priotracker _rib_tracker
68 #define RIB_LOCK_INIT(rh) rm_init(&(rh)->rib_lock, "rib head lock")
69 #define RIB_LOCK_DESTROY(rh) rm_destroy(&(rh)->rib_lock)
70 #define RIB_RLOCK(rh) rm_rlock(&(rh)->rib_lock, &_rib_tracker)
71 #define RIB_RUNLOCK(rh) rm_runlock(&(rh)->rib_lock, &_rib_tracker)
72 #define RIB_WLOCK(rh) rm_wlock(&(rh)->rib_lock)
73 #define RIB_WUNLOCK(rh) rm_wunlock(&(rh)->rib_lock)
74 #define RIB_LOCK_ASSERT(rh) rm_assert(&(rh)->rib_lock, RA_LOCKED)
75 #define RIB_WLOCK_ASSERT(rh) rm_assert(&(rh)->rib_lock, RA_WLOCKED)
78 #define RIB_MAX_RETRIES 3
80 /* Macro for verifying fields in af-specific 'struct route' structures */
81 #define CHK_STRUCT_FIELD_GENERIC(_s1, _f1, _s2, _f2) \
82 _Static_assert(sizeof(((_s1 *)0)->_f1) == sizeof(((_s2 *)0)->_f2), \
83 "Fields " #_f1 " and " #_f2 " size differs"); \
84 _Static_assert(__offsetof(_s1, _f1) == __offsetof(_s2, _f2), \
85 "Fields " #_f1 " and " #_f2 " offset differs");
87 #define _CHK_ROUTE_FIELD(_route_new, _field) \
88 CHK_STRUCT_FIELD_GENERIC(struct route, _field, _route_new, _field)
90 #define CHK_STRUCT_ROUTE_FIELDS(_route_new) \
91 _CHK_ROUTE_FIELD(_route_new, ro_nh) \
92 _CHK_ROUTE_FIELD(_route_new, ro_lle) \
93 _CHK_ROUTE_FIELD(_route_new, ro_prepend)\
94 _CHK_ROUTE_FIELD(_route_new, ro_plen) \
95 _CHK_ROUTE_FIELD(_route_new, ro_flags) \
96 _CHK_ROUTE_FIELD(_route_new, ro_mtu) \
97 _CHK_ROUTE_FIELD(_route_new, spare)
99 #define CHK_STRUCT_ROUTE_COMPAT(_ro_new, _dst_new) \
100 CHK_STRUCT_ROUTE_FIELDS(_ro_new); \
101 _Static_assert(__offsetof(struct route, ro_dst) == __offsetof(_ro_new, _dst_new),\
102 "ro_dst and " #_dst_new " are at different offset")
104 struct rib_head *rt_tables_get_rnh(int fib, int family);
105 void rt_mpath_init_rnh(struct rib_head *rnh);
107 VNET_PCPUSTAT_DECLARE(struct rtstat, rtstat);
108 #define RTSTAT_ADD(name, val) \
109 VNET_PCPUSTAT_ADD(struct rtstat, rtstat, name, (val))
110 #define RTSTAT_INC(name) RTSTAT_ADD(name, 1)
113 struct radix_node rt_nodes[2]; /* tree glue, and other values */
115 * XXX struct rtentry must begin with a struct radix_node (or two!)
116 * because the code does some casts of a 'struct radix_node *'
117 * to a 'struct rtentry *'
119 #define rt_key(r) (*((struct sockaddr **)(&(r)->rt_nodes->rn_key)))
120 #define rt_mask(r) (*((struct sockaddr **)(&(r)->rt_nodes->rn_mask)))
121 #define rt_key_const(r) (*((const struct sockaddr * const *)(&(r)->rt_nodes->rn_key)))
122 #define rt_mask_const(r) (*((const struct sockaddr * const *)(&(r)->rt_nodes->rn_mask)))
125 * 2 radix_node structurs above consists of 2x6 pointers, leaving
126 * 4 pointers (32 bytes) of the second cache line on amd64.
129 struct nhop_object *rt_nhop; /* nexthop data */
132 * Destination address storage.
133 * sizeof(struct sockaddr_in6) == 28, however
134 * the dataplane-relevant part (e.g. address) lies
135 * at offset 8..24, making the address not crossing
136 * cacheline boundary.
138 struct sockaddr_in rt_dst4;
139 struct sockaddr_in6 rt_dst6;
140 struct sockaddr rt_dst;
144 int rt_flags; /* up/down?, host/net */
145 int rt_refcnt; /* # held references */
146 u_long rt_weight; /* absolute weight */
147 u_long rt_expire; /* lifetime for route, e.g. redirect */
148 #define rt_endzero rt_mtx
149 struct mtx rt_mtx; /* mutex for routing entry */
150 struct rtentry *rt_chain; /* pointer to next rtentry to delete */
153 #define RT_LOCK_INIT(_rt) \
154 mtx_init(&(_rt)->rt_mtx, "rtentry", NULL, MTX_DEF | MTX_DUPOK | MTX_NEW)
155 #define RT_LOCK(_rt) mtx_lock(&(_rt)->rt_mtx)
156 #define RT_UNLOCK(_rt) mtx_unlock(&(_rt)->rt_mtx)
157 #define RT_LOCK_DESTROY(_rt) mtx_destroy(&(_rt)->rt_mtx)
158 #define RT_LOCK_ASSERT(_rt) mtx_assert(&(_rt)->rt_mtx, MA_OWNED)
159 #define RT_UNLOCK_COND(_rt) do { \
160 if (mtx_owned(&(_rt)->rt_mtx)) \
161 mtx_unlock(&(_rt)->rt_mtx); \
164 #define RT_ADDREF(_rt) do { \
165 RT_LOCK_ASSERT(_rt); \
166 KASSERT((_rt)->rt_refcnt >= 0, \
167 ("negative refcnt %d", (_rt)->rt_refcnt)); \
168 (_rt)->rt_refcnt++; \
171 #define RT_REMREF(_rt) do { \
172 RT_LOCK_ASSERT(_rt); \
173 KASSERT((_rt)->rt_refcnt > 0, \
174 ("bogus refcnt %d", (_rt)->rt_refcnt)); \
175 (_rt)->rt_refcnt--; \
178 #define RTFREE_LOCKED(_rt) do { \
179 if ((_rt)->rt_refcnt <= 1) \
185 /* guard against invalid refs */ \
189 #define RTFREE(_rt) do { \
191 RTFREE_LOCKED(_rt); \
195 * With the split between the routing entry and the nexthop,
196 * rt_flags has to be split between these 2 entries. As rtentry
197 * mostly contains prefix data and is thought to be generic enough
198 * so one can transparently change the nexthop pointer w/o requiring
199 * any other rtentry changes, most of rt_flags shifts to the particular nexthop.
202 * RTF_UP: rtentry, as an indication that it is linked.
203 * RTF_HOST: rtentry, nhop. The latter indication is needed for the datapath
204 * RTF_DYNAMIC: nhop, to make rtentry generic.
205 * RTF_MODIFIED: nhop, to make rtentry generic. (legacy)
206 * -- "native" path (nhop) properties:
207 * RTF_GATEWAY, RTF_STATIC, RTF_PROTO1, RTF_PROTO2, RTF_PROTO3, RTF_FIXEDMTU,
208 * RTF_PINNED, RTF_REJECT, RTF_BLACKHOLE, RTF_BROADCAST
211 /* Nexthop rt flags mask */
212 #define NHOP_RT_FLAG_MASK (RTF_GATEWAY | RTF_HOST | RTF_REJECT | RTF_DYNAMIC | \
213 RTF_MODIFIED | RTF_STATIC | RTF_BLACKHOLE | RTF_PROTO1 | RTF_PROTO2 | \
214 RTF_PROTO3 | RTF_FIXEDMTU | RTF_PINNED | RTF_BROADCAST)
216 /* rtentry rt flag mask */
217 #define RTE_RT_FLAG_MASK (RTF_UP | RTF_HOST)
219 /* Nexthop selection */
220 #define _NH2MP(_nh) ((struct nhgrp_object *)(_nh))
221 #define _SELECT_NHOP(_nh, _flowid) \
222 (_NH2MP(_nh))->nhops[(_flowid) % (_NH2MP(_nh))->mp_size]
223 #define _RT_SELECT_NHOP(_nh, _flowid) \
224 ((!NH_IS_MULTIPATH(_nh)) ? (_nh) : _SELECT_NHOP(_nh, _flowid))
225 #define RT_SELECT_NHOP(_rt, _flowid) _RT_SELECT_NHOP((_rt)->rt_nhop, _flowid)
227 /* rte<>nhop translation */
228 static inline uint16_t
229 fib_rte_to_nh_flags(int rt_flags)
233 res = (rt_flags & RTF_REJECT) ? NHF_REJECT : 0;
234 res |= (rt_flags & RTF_HOST) ? NHF_HOST : 0;
235 res |= (rt_flags & RTF_BLACKHOLE) ? NHF_BLACKHOLE : 0;
236 res |= (rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) ? NHF_REDIRECT : 0;
237 res |= (rt_flags & RTF_BROADCAST) ? NHF_BROADCAST : 0;
238 res |= (rt_flags & RTF_GATEWAY) ? NHF_GATEWAY : 0;
243 void tmproutes_update(struct rib_head *rnh, struct rtentry *rt);
244 void tmproutes_init(struct rib_head *rh);
245 void tmproutes_destroy(struct rib_head *rh);