master
1/*-
2 * SPDX-License-Identifier: BSD-2-Clause
3 *
4 * Copyright (c) 2020 Alexander V. Chernikov
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 *
15 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
16 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
19 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
20 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
21 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
22 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
25 * SUCH DAMAGE.
26 */
27
28/*
29 * This header file contains public definitions for the nexthop routing subsystem.
30 */
31
32#ifndef _NET_ROUTE_NHOP_H_
33#define _NET_ROUTE_NHOP_H_
34
35#include <netinet/in.h> /* sockaddr_in && sockaddr_in6 */
36
37#include <sys/counter.h>
38
39enum nhop_type {
40 NH_TYPE_IPV4_ETHER_RSLV = 1, /* IPv4 ethernet without GW */
41 NH_TYPE_IPV4_ETHER_NHOP = 2, /* IPv4 with pre-calculated ethernet encap */
42 NH_TYPE_IPV6_ETHER_RSLV = 3, /* IPv6 ethernet, without GW */
43 NH_TYPE_IPV6_ETHER_NHOP = 4 /* IPv6 with pre-calculated ethernet encap*/
44};
45
46#ifdef _KERNEL
47
48/*
49 * Define shorter version of AF_LINK sockaddr.
50 *
51 * Currently the only use case of AF_LINK gateway is storing
52 * interface index of the interface of the source IPv6 address.
53 * This is used by the IPv6 code for the connections over loopback
54 * interface.
55 *
56 * The structure below copies 'struct sockaddr_dl', reducing the
57 * size of sdl_data buffer, as it is not used. This change
58 * allows to store the AF_LINK gateways in the nhop gateway itself,
59 * simplifying control plane handling.
60 */
61struct sockaddr_dl_short {
62 u_char sdl_len; /* Total length of sockaddr */
63 u_char sdl_family; /* AF_LINK */
64 u_short sdl_index; /* if != 0, system given index for interface */
65 u_char sdl_type; /* interface type */
66 u_char sdl_nlen; /* interface name length, no trailing 0 reqd. */
67 u_char sdl_alen; /* link level address length */
68 u_char sdl_slen; /* link layer selector length */
69 char sdl_data[8]; /* unused */
70};
71
72#define NHOP_RELATED_FLAGS \
73 (RTF_GATEWAY | RTF_HOST | RTF_REJECT | RTF_BLACKHOLE | \
74 RTF_FIXEDMTU | RTF_LOCAL | RTF_BROADCAST | RTF_MULTICAST)
75
76struct nh_control;
77struct nhop_priv;
78
79/*
80 * Struct 'nhop_object' field description:
81 *
82 * nh_flags: NHF_ flags used in the dataplane code. NHF_GATEWAY or NHF_BLACKHOLE
83 * can be examples of such flags.
84 * nh_mtu: ready-to-use nexthop mtu. Already accounts for the link-level header,
85 * interface MTU and protocol-specific limitations.
86 * nh_prepend_len: link-level prepend length. Currently unused.
87 * nh_ifp: logical transmit interface. The one from which if_transmit() will be
88 * called. Guaranteed to be non-NULL.
89 * nh_aifp: ifnet of the source address. Same as nh_ifp except IPv6 loopback
90 * routes. See the example below.
91 * nh_ifa: interface address to use. Guaranteed to be non-NULL.
92 * nh_pksent: counter(9) reflecting the number of packets transmitted.
93 *
94 * gw_: storage suitable to hold AF_INET, AF_INET6 or AF_LINK gateway. More
95 * details ara available in the examples below.
96 *
97 * Examples:
98 *
99 * Direct routes (routes w/o gateway):
100 * NHF_GATEWAY is NOT set.
101 * nh_ifp denotes the logical transmit interface ().
102 * nh_aifp is the same as nh_ifp
103 * gw_sa contains AF_LINK sa with nh_aifp ifindex (compat)
104 * Loopback routes:
105 * NHF_GATEWAY is NOT set.
106 * nh_ifp points to the loopback interface (lo0).
107 * nh_aifp points to the interface where the destination address belongs to.
108 * This is useful in IPv6 link-local-over-loopback communications.
109 * gw_sa contains AF_LINK sa with nh_aifp ifindex (compat)
110 * GW routes:
111 * NHF_GATEWAY is set.
112 * nh_ifp denotes the logical transmit interface.
113 * nh_aifp is the same as nh_ifp
114 * gw_sa contains L3 address (either AF_INET or AF_INET6).
115 *
116 *
117 * Note: struct nhop_object fields are ordered in a way that
118 * supports memcmp-based comparisons.
119 *
120 */
121#define NHOP_END_CMP (__offsetof(struct nhop_object, nh_pksent))
122
123struct nhop_object {
124 uint16_t nh_flags; /* nhop flags */
125 uint16_t nh_mtu; /* nexthop mtu */
126 union {
127 struct sockaddr_in gw4_sa; /* GW accessor as IPv4 */
128 struct sockaddr_in6 gw6_sa; /* GW accessor as IPv6 */
129 struct sockaddr gw_sa;
130 struct sockaddr_dl_short gwl_sa; /* AF_LINK gw (compat) */
131 char gw_buf[28];
132 };
133 struct ifnet *nh_ifp; /* Logical egress interface. Always != NULL */
134 struct ifaddr *nh_ifa; /* interface address to use. Always != NULL */
135 struct ifnet *nh_aifp; /* ifnet of the source address. Always != NULL */
136 counter_u64_t nh_pksent; /* packets sent using this nhop */
137 /* 32 bytes + 4xPTR == 64(amd64) / 48(i386) */
138 uint8_t nh_prepend_len; /* length of prepend data */
139 uint8_t spare[3];
140 uint32_t spare1; /* alignment */
141 char nh_prepend[48]; /* L2 prepend */
142 struct nhop_priv *nh_priv; /* control plane data */
143 /* -- 128 bytes -- */
144};
145
146/*
147 * Nhop validness.
148 *
149 * Currently we verify whether link is up or not on every packet, which can be
150 * quite costy.
151 * TODO: subscribe for the interface notifications and update the nexthops
152 * with NHF_INVALID flag.
153 */
154
155#define NH_IS_VALID(_nh) RT_LINK_IS_UP((_nh)->nh_ifp)
156#define NH_IS_NHGRP(_nh) ((_nh)->nh_flags & NHF_MULTIPATH)
157
158#define NH_FREE(_nh) do { \
159 nhop_free(_nh); \
160 /* guard against invalid refs */ \
161 _nh = NULL; \
162} while (0)
163
164struct weightened_nhop {
165 struct nhop_object *nh;
166 uint32_t weight;
167 uint32_t storage;
168};
169
170void nhop_free(struct nhop_object *nh);
171
172struct sysctl_req;
173struct sockaddr_dl;
174struct rib_head;
175
176/* flags that can be set using nhop_set_rtflags() */
177#define RT_SET_RTFLAGS_MASK (RTF_PROTO1 | RTF_PROTO2 | RTF_PROTO3 | RTF_STATIC)
178#define RT_CHANGE_RTFLAGS_MASK RT_SET_RTFLAGS_MASK
179
180struct nhop_object *nhop_alloc(uint32_t fibnum, int family);
181void nhop_copy(struct nhop_object *nh, const struct nhop_object *nh_orig);
182struct nhop_object *nhop_get_nhop(struct nhop_object *nh, int *perror);
183int nhop_get_unlinked(struct nhop_object *nh);
184
185void nhop_set_direct_gw(struct nhop_object *nh, struct ifnet *ifp);
186bool nhop_set_gw(struct nhop_object *nh, const struct sockaddr *sa, bool is_gw);
187
188
189void nhop_set_mtu(struct nhop_object *nh, uint32_t mtu, bool from_user);
190void nhop_set_rtflags(struct nhop_object *nh, int rt_flags);
191void nhop_set_pxtype_flag(struct nhop_object *nh, int nh_flag);
192void nhop_set_broadcast(struct nhop_object *nh, bool is_broadcast);
193void nhop_set_blackhole(struct nhop_object *nh, int blackhole_rt_flag);
194void nhop_set_pinned(struct nhop_object *nh, bool is_pinned);
195void nhop_set_redirect(struct nhop_object *nh, bool is_redirect);
196void nhop_set_type(struct nhop_object *nh, enum nhop_type nh_type);
197void nhop_set_src(struct nhop_object *nh, struct ifaddr *ifa);
198void nhop_set_transmit_ifp(struct nhop_object *nh, struct ifnet *ifp);
199
200#define NH_ORIGIN_UNSPEC 0 /* not set */
201#define NH_ORIGIN_REDIRECT 1 /* kernel-originated redirect */
202#define NH_ORIGIN_KERNEL 2 /* kernel (interface) routes */
203#define NH_ORIGIN_BOOT 3 /* kernel-originated routes at boot */
204#define NH_ORIGIN_STATIC 4 /* route(8) routes */
205void nhop_set_origin(struct nhop_object *nh, uint8_t origin);
206uint8_t nhop_get_origin(const struct nhop_object *nh);
207
208uint32_t nhop_get_idx(const struct nhop_object *nh);
209uint32_t nhop_get_uidx(const struct nhop_object *nh);
210void nhop_set_uidx(struct nhop_object *nh, uint32_t uidx);
211enum nhop_type nhop_get_type(const struct nhop_object *nh);
212int nhop_get_rtflags(const struct nhop_object *nh);
213struct vnet *nhop_get_vnet(const struct nhop_object *nh);
214struct nhop_object *nhop_select_func(struct nhop_object *nh, uint32_t flowid);
215int nhop_get_upper_family(const struct nhop_object *nh);
216bool nhop_set_upper_family(struct nhop_object *nh, int family);
217int nhop_get_neigh_family(const struct nhop_object *nh);
218uint32_t nhop_get_fibnum(const struct nhop_object *nh);
219void nhop_set_fibnum(struct nhop_object *nh, uint32_t fibnum);
220uint32_t nhop_get_expire(const struct nhop_object *nh);
221void nhop_set_expire(struct nhop_object *nh, uint32_t expire);
222struct rib_head *nhop_get_rh(const struct nhop_object *nh);
223
224struct nhgrp_object;
225struct nhgrp_object *nhgrp_alloc(uint32_t fibnum, int family,
226 struct weightened_nhop *wn, int num_nhops, int *perror);
227struct nhgrp_object *nhgrp_get_nhgrp(struct nhgrp_object *nhg, int *perror);
228void nhgrp_set_uidx(struct nhgrp_object *nhg, uint32_t uidx);
229uint32_t nhgrp_get_uidx(const struct nhgrp_object *nhg);
230uint8_t nhgrp_get_origin(const struct nhgrp_object *nhg);
231void nhgrp_set_origin(struct nhgrp_object *nhg, uint8_t origin);
232#endif /* _KERNEL */
233
234/* Kernel <> userland structures */
235
236/* Structure usage and layout are described in dump_nhop_entry() */
237struct nhop_external {
238 uint32_t nh_len; /* length of the datastructure */
239 uint32_t nh_idx; /* Nexthop index */
240 uint32_t nh_fib; /* Fib nexhop is attached to */
241 uint32_t ifindex; /* transmit interface ifindex */
242 uint32_t aifindex; /* address ifindex */
243 uint8_t prepend_len; /* length of the prepend */
244 uint8_t nh_family; /* address family */
245 uint16_t nh_type; /* nexthop type */
246 uint16_t nh_mtu; /* nexthop mtu */
247
248 uint16_t nh_flags; /* nhop flags */
249 struct in_addr nh_addr; /* GW/DST IPv4 address */
250 struct in_addr nh_src; /* default source IPv4 address */
251 uint64_t nh_pksent;
252 /* control plane */
253 /* lookup key: address, family, type */
254 char nh_prepend[64]; /* L2 prepend */
255 uint64_t nh_refcount; /* number of references */
256};
257
258struct nhop_addrs {
259 uint32_t na_len; /* length of the datastructure */
260 uint16_t gw_sa_off; /* offset of gateway SA */
261 uint16_t src_sa_off; /* offset of src address SA */
262};
263
264#define NHG_C_TYPE_CNHOPS 0x1 /* Control plane nhops list */
265#define NHG_C_TYPE_DNHOPS 0x2 /* Dataplane nhops list */
266struct nhgrp_container {
267 uint32_t nhgc_len; /* container length */
268 uint16_t nhgc_count; /* number of items */
269 uint8_t nhgc_type; /* container type */
270 uint8_t nhgc_subtype; /* container subtype */
271};
272
273struct nhgrp_nhop_external {
274 uint32_t nh_idx;
275 uint32_t nh_weight;
276};
277
278/*
279 * Layout:
280 * - nhgrp_external
281 * - nhgrp_container (control plane nhops list)
282 * - nhgrp_nhop_external
283 * - nhgrp_nhop_external
284 * ..
285 * - nhgrp_container (dataplane nhops list)
286 * - nhgrp_nhop_external
287 * - nhgrp_nhop_external
288 */
289struct nhgrp_external {
290 uint32_t nhg_idx; /* Nexthop group index */
291 uint32_t nhg_refcount; /* number of references */
292};
293
294#endif