ib_addr.h revision 331769
1/*
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
51struct rdma_addr_client {
52	atomic_t refcount;
53	struct completion comp;
54};
55
56/**
57 * rdma_addr_register_client - Register an address client.
58 */
59void rdma_addr_register_client(struct rdma_addr_client *client);
60
61/**
62 * rdma_addr_unregister_client - Deregister an address client.
63 * @client: Client object to deregister.
64 */
65void rdma_addr_unregister_client(struct rdma_addr_client *client);
66
67/**
68 * struct rdma_dev_addr - Contains resolved RDMA hardware addresses
69 * @src_dev_addr:	Source MAC address.
70 * @dst_dev_addr:	Destination MAC address.
71 * @broadcast:		Broadcast address of the device.
72 * @dev_type:		The interface hardware type of the device.
73 * @bound_dev_if:	An optional device interface index.
74 * @transport:		The transport type used.
75 * @net:		Network namespace containing the bound_dev_if net_dev.
76 */
77struct vnet;
78struct rdma_dev_addr {
79	unsigned char src_dev_addr[MAX_ADDR_LEN];
80	unsigned char dst_dev_addr[MAX_ADDR_LEN];
81	unsigned char broadcast[MAX_ADDR_LEN];
82	unsigned short dev_type;
83	int bound_dev_if;
84	enum rdma_transport_type transport;
85	struct vnet *net;
86	enum rdma_network_type network;
87	int hoplimit;
88};
89
90/**
91 * rdma_translate_ip - Translate a local IP address to an RDMA hardware
92 *   address.
93 *
94 * The dev_addr->net field must be initialized.
95 */
96int rdma_translate_ip(const struct sockaddr *addr,
97		      struct rdma_dev_addr *dev_addr, u16 *vlan_id);
98
99/**
100 * rdma_resolve_ip - Resolve source and destination IP addresses to
101 *   RDMA hardware addresses.
102 * @client: Address client associated with request.
103 * @src_addr: An optional source address to use in the resolution.  If a
104 *   source address is not provided, a usable address will be returned via
105 *   the callback.
106 * @dst_addr: The destination address to resolve.
107 * @addr: A reference to a data location that will receive the resolved
108 *   addresses.  The data location must remain valid until the callback has
109 *   been invoked. The net field of the addr struct must be valid.
110 * @timeout_ms: Amount of time to wait for the address resolution to complete.
111 * @callback: Call invoked once address resolution has completed, timed out,
112 *   or been canceled.  A status of 0 indicates success.
113 * @context: User-specified context associated with the call.
114 */
115int rdma_resolve_ip(struct rdma_addr_client *client,
116		    struct sockaddr *src_addr, struct sockaddr *dst_addr,
117		    struct rdma_dev_addr *addr, int timeout_ms,
118		    void (*callback)(int status, struct sockaddr *src_addr,
119				     struct rdma_dev_addr *addr, void *context),
120		    void *context);
121
122int rdma_resolve_ip_route(struct sockaddr *src_addr,
123			  const struct sockaddr *dst_addr,
124			  struct rdma_dev_addr *addr);
125
126void rdma_addr_cancel(struct rdma_dev_addr *addr);
127
128int rdma_copy_addr(struct rdma_dev_addr *dev_addr, struct net_device *dev,
129	      const unsigned char *dst_dev_addr);
130
131int rdma_addr_size(struct sockaddr *addr);
132
133int rdma_addr_find_smac_by_sgid(union ib_gid *sgid, u8 *smac, u16 *vlan_id);
134int rdma_addr_find_l2_eth_by_grh(const union ib_gid *sgid,
135				 const union ib_gid *dgid,
136				 u8 *smac, u16 *vlan_id, int *if_index,
137				 int *hoplimit);
138
139static inline u16 ib_addr_get_pkey(struct rdma_dev_addr *dev_addr)
140{
141	return ((u16)dev_addr->broadcast[8] << 8) | (u16)dev_addr->broadcast[9];
142}
143
144static inline void ib_addr_set_pkey(struct rdma_dev_addr *dev_addr, u16 pkey)
145{
146	dev_addr->broadcast[8] = pkey >> 8;
147	dev_addr->broadcast[9] = (unsigned char) pkey;
148}
149
150static inline void ib_addr_get_mgid(struct rdma_dev_addr *dev_addr,
151				    union ib_gid *gid)
152{
153	memcpy(gid, dev_addr->broadcast + 4, sizeof *gid);
154}
155
156static inline int rdma_addr_gid_offset(struct rdma_dev_addr *dev_addr)
157{
158	return dev_addr->dev_type == ARPHRD_INFINIBAND ? 4 : 0;
159}
160
161static inline u16 rdma_vlan_dev_vlan_id(const struct net_device *dev)
162{
163	uint16_t tag;
164
165	if (VLAN_TAG(__DECONST(struct ifnet *, dev), &tag) != 0)
166		return 0xffff;
167	return tag;
168}
169
170static inline int rdma_ip2gid(const struct sockaddr *addr, union ib_gid *gid)
171{
172	switch (addr->sa_family) {
173	case AF_INET:
174		ipv6_addr_set_v4mapped(((const struct sockaddr_in *)
175					addr)->sin_addr.s_addr,
176				       (struct in6_addr *)gid);
177		break;
178	case AF_INET6:
179		memcpy(gid->raw, &((const struct sockaddr_in6 *)addr)->sin6_addr, 16);
180		/* make sure scope ID gets zeroed inside GID */
181		if (IN6_IS_SCOPE_LINKLOCAL((struct in6_addr *)gid->raw) ||
182		    IN6_IS_ADDR_MC_INTFACELOCAL((struct in6_addr *)gid->raw)) {
183			gid->raw[2] = 0;
184			gid->raw[3] = 0;
185		}
186		break;
187	default:
188		return -EINVAL;
189	}
190	return 0;
191}
192
193/* Important - sockaddr should be a union of sockaddr_in and sockaddr_in6 */
194static inline void rdma_gid2ip(struct sockaddr *out, const union ib_gid *gid)
195{
196	if (ipv6_addr_v4mapped((const struct in6_addr *)gid)) {
197		struct sockaddr_in *out_in = (struct sockaddr_in *)out;
198		memset(out_in, 0, sizeof(*out_in));
199		out_in->sin_len = sizeof(*out_in);
200		out_in->sin_family = AF_INET;
201		memcpy(&out_in->sin_addr.s_addr, gid->raw + 12, 4);
202	} else {
203		struct sockaddr_in6 *out_in = (struct sockaddr_in6 *)out;
204		memset(out_in, 0, sizeof(*out_in));
205		out_in->sin6_len = sizeof(*out_in);
206		out_in->sin6_family = AF_INET6;
207		memcpy(&out_in->sin6_addr.s6_addr, gid->raw, 16);
208	}
209}
210
211static inline void iboe_addr_get_sgid(struct rdma_dev_addr *dev_addr,
212				      union ib_gid *gid)
213{
214	struct net_device *dev;
215	struct ifaddr *ifa;
216
217	dev = dev_get_by_index(&init_net, dev_addr->bound_dev_if);
218	if (dev) {
219		TAILQ_FOREACH(ifa, &dev->if_addrhead, ifa_link) {
220			if (ifa->ifa_addr == NULL ||
221			    ifa->ifa_addr->sa_family != AF_INET)
222				continue;
223			ipv6_addr_set_v4mapped(((struct sockaddr_in *)
224					       ifa->ifa_addr)->sin_addr.s_addr,
225					       (struct in6_addr *)gid);
226			break;
227		}
228		dev_put(dev);
229	}
230}
231
232static inline void rdma_addr_get_sgid(struct rdma_dev_addr *dev_addr, union ib_gid *gid)
233{
234	if (dev_addr->transport == RDMA_TRANSPORT_IB &&
235	    dev_addr->dev_type != ARPHRD_INFINIBAND)
236		iboe_addr_get_sgid(dev_addr, gid);
237	else
238		memcpy(gid, dev_addr->src_dev_addr +
239		       rdma_addr_gid_offset(dev_addr), sizeof *gid);
240}
241
242static inline void rdma_addr_set_sgid(struct rdma_dev_addr *dev_addr, union ib_gid *gid)
243{
244	memcpy(dev_addr->src_dev_addr + rdma_addr_gid_offset(dev_addr), gid, sizeof *gid);
245}
246
247static inline void rdma_addr_get_dgid(struct rdma_dev_addr *dev_addr, union ib_gid *gid)
248{
249	memcpy(gid, dev_addr->dst_dev_addr + rdma_addr_gid_offset(dev_addr), sizeof *gid);
250}
251
252static inline void rdma_addr_set_dgid(struct rdma_dev_addr *dev_addr, union ib_gid *gid)
253{
254	memcpy(dev_addr->dst_dev_addr + rdma_addr_gid_offset(dev_addr), gid, sizeof *gid);
255}
256
257static inline enum ib_mtu iboe_get_mtu(int mtu)
258{
259	/*
260	 * reduce IB headers from effective IBoE MTU. 28 stands for
261	 * atomic header which is the biggest possible header after BTH
262	 */
263	mtu = mtu - IB_GRH_BYTES - IB_BTH_BYTES - 28;
264
265	if (mtu >= ib_mtu_enum_to_int(IB_MTU_4096))
266		return IB_MTU_4096;
267	else if (mtu >= ib_mtu_enum_to_int(IB_MTU_2048))
268		return IB_MTU_2048;
269	else if (mtu >= ib_mtu_enum_to_int(IB_MTU_1024))
270		return IB_MTU_1024;
271	else if (mtu >= ib_mtu_enum_to_int(IB_MTU_512))
272		return IB_MTU_512;
273	else if (mtu >= ib_mtu_enum_to_int(IB_MTU_256))
274		return IB_MTU_256;
275	else
276		return 0;
277}
278
279static inline int iboe_get_rate(struct net_device *dev)
280{
281	uint64_t baudrate = dev->if_baudrate;
282#ifdef if_baudrate_pf
283	int exp;
284	for (exp = dev->if_baudrate_pf; exp > 0; exp--)
285		baudrate *= 10;
286#endif
287	if (baudrate >= IF_Gbps(40))
288		return IB_RATE_40_GBPS;
289	else if (baudrate >= IF_Gbps(30))
290		return IB_RATE_30_GBPS;
291	else if (baudrate >= IF_Gbps(20))
292		return IB_RATE_20_GBPS;
293	else if (baudrate >= IF_Gbps(10))
294		return IB_RATE_10_GBPS;
295	else
296		return IB_RATE_PORT_CURRENT;
297}
298
299static inline int rdma_link_local_addr(struct in6_addr *addr)
300{
301	if (addr->s6_addr32[0] == htonl(0xfe800000) &&
302	    addr->s6_addr32[1] == 0)
303		return 1;
304
305	return 0;
306}
307
308static inline void rdma_get_ll_mac(struct in6_addr *addr, u8 *mac)
309{
310	memcpy(mac, &addr->s6_addr[8], 3);
311	memcpy(mac + 3, &addr->s6_addr[13], 3);
312	mac[0] ^= 2;
313}
314
315static inline int rdma_is_multicast_addr(struct in6_addr *addr)
316{
317	return addr->s6_addr[0] == 0xff;
318}
319
320static inline void rdma_get_mcast_mac(struct in6_addr *addr, u8 *mac)
321{
322	int i;
323
324	mac[0] = 0x33;
325	mac[1] = 0x33;
326	for (i = 2; i < 6; ++i)
327		mac[i] = addr->s6_addr[i + 10];
328}
329
330static inline u16 rdma_get_vlan_id(union ib_gid *dgid)
331{
332	u16 vid;
333
334	vid = dgid->raw[11] << 8 | dgid->raw[12];
335	return vid < 0x1000 ? vid : 0xffff;
336}
337
338static inline struct net_device *rdma_vlan_dev_real_dev(const struct net_device *dev)
339{
340	return VLAN_TRUNKDEV(__DECONST(struct ifnet *, dev));
341}
342
343#endif /* IB_ADDR_H */
344