blob: 77c7908b7d73f28683d7ce993e1011822b948eb5 [file] [log] [blame]
Andrew Scullb4b6d4a2019-01-02 15:54:55 +00001/*
2 * Copyright (c) 2005 Voltaire Inc. All rights reserved.
3 * Copyright (c) 2005 Intel Corporation. All rights reserved.
4 *
5 * This software is available to you under a choice of one of two
6 * licenses. You may choose to be licensed under the terms of the GNU
7 * General Public License (GPL) Version 2, available from the file
8 * COPYING in the main directory of this source tree, or the
9 * OpenIB.org BSD license below:
10 *
11 * Redistribution and use in source and binary forms, with or
12 * without modification, are permitted provided that the following
13 * conditions are met:
14 *
15 * - Redistributions of source code must retain the above
16 * copyright notice, this list of conditions and the following
17 * disclaimer.
18 *
19 * - Redistributions in binary form must reproduce the above
20 * copyright notice, this list of conditions and the following
21 * disclaimer in the documentation and/or other materials
22 * provided with the distribution.
23 *
24 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
25 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
26 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
27 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
28 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
29 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
30 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
31 * SOFTWARE.
32 */
33
34#if !defined(IB_ADDR_H)
35#define IB_ADDR_H
36
37#include <linux/in.h>
38#include <linux/in6.h>
39#include <linux/if_arp.h>
40#include <linux/netdevice.h>
41#include <linux/inetdevice.h>
42#include <linux/socket.h>
43#include <linux/if_vlan.h>
44#include <net/ipv6.h>
45#include <net/if_inet6.h>
46#include <net/ip.h>
47#include <rdma/ib_verbs.h>
48#include <rdma/ib_pack.h>
49#include <net/ipv6.h>
50#include <net/net_namespace.h>
51
52/**
53 * struct rdma_dev_addr - Contains resolved RDMA hardware addresses
54 * @src_dev_addr: Source MAC address.
55 * @dst_dev_addr: Destination MAC address.
56 * @broadcast: Broadcast address of the device.
57 * @dev_type: The interface hardware type of the device.
58 * @bound_dev_if: An optional device interface index.
59 * @transport: The transport type used.
60 * @net: Network namespace containing the bound_dev_if net_dev.
61 * @sgid_attr: GID attribute to use for identified SGID
62 */
63struct rdma_dev_addr {
64 unsigned char src_dev_addr[MAX_ADDR_LEN];
65 unsigned char dst_dev_addr[MAX_ADDR_LEN];
66 unsigned char broadcast[MAX_ADDR_LEN];
67 unsigned short dev_type;
68 int bound_dev_if;
69 enum rdma_transport_type transport;
70 struct net *net;
71 const struct ib_gid_attr *sgid_attr;
72 enum rdma_network_type network;
73 int hoplimit;
74};
75
76/**
77 * rdma_translate_ip - Translate a local IP address to an RDMA hardware
78 * address.
79 *
80 * The dev_addr->net field must be initialized.
81 */
82int rdma_translate_ip(const struct sockaddr *addr,
83 struct rdma_dev_addr *dev_addr);
84
85/**
86 * rdma_resolve_ip - Resolve source and destination IP addresses to
87 * RDMA hardware addresses.
88 * @src_addr: An optional source address to use in the resolution. If a
89 * source address is not provided, a usable address will be returned via
90 * the callback.
91 * @dst_addr: The destination address to resolve.
92 * @addr: A reference to a data location that will receive the resolved
93 * addresses. The data location must remain valid until the callback has
94 * been invoked. The net field of the addr struct must be valid.
95 * @timeout_ms: Amount of time to wait for the address resolution to complete.
96 * @callback: Call invoked once address resolution has completed, timed out,
97 * or been canceled. A status of 0 indicates success.
98 * @context: User-specified context associated with the call.
99 */
100int rdma_resolve_ip(struct sockaddr *src_addr, const struct sockaddr *dst_addr,
101 struct rdma_dev_addr *addr, int timeout_ms,
102 void (*callback)(int status, struct sockaddr *src_addr,
103 struct rdma_dev_addr *addr, void *context),
104 void *context);
105
106void rdma_addr_cancel(struct rdma_dev_addr *addr);
107
108void rdma_copy_addr(struct rdma_dev_addr *dev_addr,
109 const struct net_device *dev,
110 const unsigned char *dst_dev_addr);
111
112int rdma_addr_size(const struct sockaddr *addr);
113int rdma_addr_size_in6(struct sockaddr_in6 *addr);
114int rdma_addr_size_kss(struct __kernel_sockaddr_storage *addr);
115
116static inline u16 ib_addr_get_pkey(struct rdma_dev_addr *dev_addr)
117{
118 return ((u16)dev_addr->broadcast[8] << 8) | (u16)dev_addr->broadcast[9];
119}
120
121static inline void ib_addr_set_pkey(struct rdma_dev_addr *dev_addr, u16 pkey)
122{
123 dev_addr->broadcast[8] = pkey >> 8;
124 dev_addr->broadcast[9] = (unsigned char) pkey;
125}
126
127static inline void ib_addr_get_mgid(struct rdma_dev_addr *dev_addr,
128 union ib_gid *gid)
129{
130 memcpy(gid, dev_addr->broadcast + 4, sizeof *gid);
131}
132
133static inline int rdma_addr_gid_offset(struct rdma_dev_addr *dev_addr)
134{
135 return dev_addr->dev_type == ARPHRD_INFINIBAND ? 4 : 0;
136}
137
138static inline u16 rdma_vlan_dev_vlan_id(const struct net_device *dev)
139{
140 return is_vlan_dev(dev) ? vlan_dev_vlan_id(dev) : 0xffff;
141}
142
143static inline int rdma_ip2gid(struct sockaddr *addr, union ib_gid *gid)
144{
145 switch (addr->sa_family) {
146 case AF_INET:
147 ipv6_addr_set_v4mapped(((struct sockaddr_in *)
148 addr)->sin_addr.s_addr,
149 (struct in6_addr *)gid);
150 break;
151 case AF_INET6:
152 *(struct in6_addr *)&gid->raw =
153 ((struct sockaddr_in6 *)addr)->sin6_addr;
154 break;
155 default:
156 return -EINVAL;
157 }
158 return 0;
159}
160
161/* Important - sockaddr should be a union of sockaddr_in and sockaddr_in6 */
162static inline void rdma_gid2ip(struct sockaddr *out, const union ib_gid *gid)
163{
164 if (ipv6_addr_v4mapped((struct in6_addr *)gid)) {
165 struct sockaddr_in *out_in = (struct sockaddr_in *)out;
166 memset(out_in, 0, sizeof(*out_in));
167 out_in->sin_family = AF_INET;
168 memcpy(&out_in->sin_addr.s_addr, gid->raw + 12, 4);
169 } else {
170 struct sockaddr_in6 *out_in = (struct sockaddr_in6 *)out;
171 memset(out_in, 0, sizeof(*out_in));
172 out_in->sin6_family = AF_INET6;
173 memcpy(&out_in->sin6_addr.s6_addr, gid->raw, 16);
174 }
175}
176
177/*
178 * rdma_get/set_sgid/dgid() APIs are applicable to IB, and iWarp.
179 * They are not applicable to RoCE.
180 * RoCE GIDs are derived from the IP addresses.
181 */
182static inline void rdma_addr_get_sgid(struct rdma_dev_addr *dev_addr, union ib_gid *gid)
183{
184 memcpy(gid, dev_addr->src_dev_addr + rdma_addr_gid_offset(dev_addr),
185 sizeof(*gid));
186}
187
188static inline void rdma_addr_set_sgid(struct rdma_dev_addr *dev_addr, union ib_gid *gid)
189{
190 memcpy(dev_addr->src_dev_addr + rdma_addr_gid_offset(dev_addr), gid, sizeof *gid);
191}
192
193static inline void rdma_addr_get_dgid(struct rdma_dev_addr *dev_addr, union ib_gid *gid)
194{
195 memcpy(gid, dev_addr->dst_dev_addr + rdma_addr_gid_offset(dev_addr), sizeof *gid);
196}
197
198static inline void rdma_addr_set_dgid(struct rdma_dev_addr *dev_addr, union ib_gid *gid)
199{
200 memcpy(dev_addr->dst_dev_addr + rdma_addr_gid_offset(dev_addr), gid, sizeof *gid);
201}
202
203static inline enum ib_mtu iboe_get_mtu(int mtu)
204{
205 /*
206 * Reduce IB headers from effective IBoE MTU.
207 */
208 mtu = mtu - (IB_GRH_BYTES + IB_UDP_BYTES + IB_BTH_BYTES +
209 IB_EXT_XRC_BYTES + IB_EXT_ATOMICETH_BYTES +
210 IB_ICRC_BYTES);
211
212 if (mtu >= ib_mtu_enum_to_int(IB_MTU_4096))
213 return IB_MTU_4096;
214 else if (mtu >= ib_mtu_enum_to_int(IB_MTU_2048))
215 return IB_MTU_2048;
216 else if (mtu >= ib_mtu_enum_to_int(IB_MTU_1024))
217 return IB_MTU_1024;
218 else if (mtu >= ib_mtu_enum_to_int(IB_MTU_512))
219 return IB_MTU_512;
220 else if (mtu >= ib_mtu_enum_to_int(IB_MTU_256))
221 return IB_MTU_256;
222 else
223 return 0;
224}
225
226static inline int iboe_get_rate(struct net_device *dev)
227{
228 struct ethtool_link_ksettings cmd;
229 int err;
230
231 rtnl_lock();
232 err = __ethtool_get_link_ksettings(dev, &cmd);
233 rtnl_unlock();
234 if (err)
235 return IB_RATE_PORT_CURRENT;
236
237 if (cmd.base.speed >= 40000)
238 return IB_RATE_40_GBPS;
239 else if (cmd.base.speed >= 30000)
240 return IB_RATE_30_GBPS;
241 else if (cmd.base.speed >= 20000)
242 return IB_RATE_20_GBPS;
243 else if (cmd.base.speed >= 10000)
244 return IB_RATE_10_GBPS;
245 else
246 return IB_RATE_PORT_CURRENT;
247}
248
249static inline int rdma_link_local_addr(struct in6_addr *addr)
250{
251 if (addr->s6_addr32[0] == htonl(0xfe800000) &&
252 addr->s6_addr32[1] == 0)
253 return 1;
254
255 return 0;
256}
257
258static inline void rdma_get_ll_mac(struct in6_addr *addr, u8 *mac)
259{
260 memcpy(mac, &addr->s6_addr[8], 3);
261 memcpy(mac + 3, &addr->s6_addr[13], 3);
262 mac[0] ^= 2;
263}
264
265static inline int rdma_is_multicast_addr(struct in6_addr *addr)
266{
267 __be32 ipv4_addr;
268
269 if (addr->s6_addr[0] == 0xff)
270 return 1;
271
272 ipv4_addr = addr->s6_addr32[3];
273 return (ipv6_addr_v4mapped(addr) && ipv4_is_multicast(ipv4_addr));
274}
275
276static inline void rdma_get_mcast_mac(struct in6_addr *addr, u8 *mac)
277{
278 int i;
279
280 mac[0] = 0x33;
281 mac[1] = 0x33;
282 for (i = 2; i < 6; ++i)
283 mac[i] = addr->s6_addr[i + 10];
284}
285
286static inline u16 rdma_get_vlan_id(union ib_gid *dgid)
287{
288 u16 vid;
289
290 vid = dgid->raw[11] << 8 | dgid->raw[12];
291 return vid < 0x1000 ? vid : 0xffff;
292}
293
294static inline struct net_device *rdma_vlan_dev_real_dev(const struct net_device *dev)
295{
296 return is_vlan_dev(dev) ? vlan_dev_real_dev(dev) : NULL;
297}
298
299#endif /* IB_ADDR_H */