tjh.dev
Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
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#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 */
62struct rdma_dev_addr {
63 unsigned char src_dev_addr[MAX_ADDR_LEN];
64 unsigned char dst_dev_addr[MAX_ADDR_LEN];
65 unsigned char broadcast[MAX_ADDR_LEN];
66 unsigned short dev_type;
67 int bound_dev_if;
68 enum rdma_transport_type transport;
69 struct net *net;
70 enum rdma_network_type network;
71 int hoplimit;
72};
73
74/**
75 * rdma_translate_ip - Translate a local IP address to an RDMA hardware
76 * address.
77 *
78 * The dev_addr->net field must be initialized.
79 */
80int rdma_translate_ip(const struct sockaddr *addr,
81 struct rdma_dev_addr *dev_addr);
82
83/**
84 * rdma_resolve_ip - Resolve source and destination IP addresses to
85 * RDMA hardware addresses.
86 * @src_addr: An optional source address to use in the resolution. If a
87 * source address is not provided, a usable address will be returned via
88 * the callback.
89 * @dst_addr: The destination address to resolve.
90 * @addr: A reference to a data location that will receive the resolved
91 * addresses. The data location must remain valid until the callback has
92 * been invoked. The net field of the addr struct must be valid.
93 * @timeout_ms: Amount of time to wait for the address resolution to complete.
94 * @callback: Call invoked once address resolution has completed, timed out,
95 * or been canceled. A status of 0 indicates success.
96 * @context: User-specified context associated with the call.
97 */
98int rdma_resolve_ip(struct sockaddr *src_addr, struct sockaddr *dst_addr,
99 struct rdma_dev_addr *addr, int timeout_ms,
100 void (*callback)(int status, struct sockaddr *src_addr,
101 struct rdma_dev_addr *addr, void *context),
102 void *context);
103
104void rdma_addr_cancel(struct rdma_dev_addr *addr);
105
106void rdma_copy_addr(struct rdma_dev_addr *dev_addr,
107 const struct net_device *dev,
108 const unsigned char *dst_dev_addr);
109
110int rdma_addr_size(struct sockaddr *addr);
111int rdma_addr_size_in6(struct sockaddr_in6 *addr);
112int rdma_addr_size_kss(struct __kernel_sockaddr_storage *addr);
113
114static inline u16 ib_addr_get_pkey(struct rdma_dev_addr *dev_addr)
115{
116 return ((u16)dev_addr->broadcast[8] << 8) | (u16)dev_addr->broadcast[9];
117}
118
119static inline void ib_addr_set_pkey(struct rdma_dev_addr *dev_addr, u16 pkey)
120{
121 dev_addr->broadcast[8] = pkey >> 8;
122 dev_addr->broadcast[9] = (unsigned char) pkey;
123}
124
125static inline void ib_addr_get_mgid(struct rdma_dev_addr *dev_addr,
126 union ib_gid *gid)
127{
128 memcpy(gid, dev_addr->broadcast + 4, sizeof *gid);
129}
130
131static inline int rdma_addr_gid_offset(struct rdma_dev_addr *dev_addr)
132{
133 return dev_addr->dev_type == ARPHRD_INFINIBAND ? 4 : 0;
134}
135
136static inline u16 rdma_vlan_dev_vlan_id(const struct net_device *dev)
137{
138 return is_vlan_dev(dev) ? vlan_dev_vlan_id(dev) : 0xffff;
139}
140
141static inline int rdma_ip2gid(struct sockaddr *addr, union ib_gid *gid)
142{
143 switch (addr->sa_family) {
144 case AF_INET:
145 ipv6_addr_set_v4mapped(((struct sockaddr_in *)
146 addr)->sin_addr.s_addr,
147 (struct in6_addr *)gid);
148 break;
149 case AF_INET6:
150 *(struct in6_addr *)&gid->raw =
151 ((struct sockaddr_in6 *)addr)->sin6_addr;
152 break;
153 default:
154 return -EINVAL;
155 }
156 return 0;
157}
158
159/* Important - sockaddr should be a union of sockaddr_in and sockaddr_in6 */
160static inline void rdma_gid2ip(struct sockaddr *out, const union ib_gid *gid)
161{
162 if (ipv6_addr_v4mapped((struct in6_addr *)gid)) {
163 struct sockaddr_in *out_in = (struct sockaddr_in *)out;
164 memset(out_in, 0, sizeof(*out_in));
165 out_in->sin_family = AF_INET;
166 memcpy(&out_in->sin_addr.s_addr, gid->raw + 12, 4);
167 } else {
168 struct sockaddr_in6 *out_in = (struct sockaddr_in6 *)out;
169 memset(out_in, 0, sizeof(*out_in));
170 out_in->sin6_family = AF_INET6;
171 memcpy(&out_in->sin6_addr.s6_addr, gid->raw, 16);
172 }
173}
174
175/*
176 * rdma_get/set_sgid/dgid() APIs are applicable to IB, and iWarp.
177 * They are not applicable to RoCE.
178 * RoCE GIDs are derived from the IP addresses.
179 */
180static inline void rdma_addr_get_sgid(struct rdma_dev_addr *dev_addr, union ib_gid *gid)
181{
182 memcpy(gid, dev_addr->src_dev_addr + rdma_addr_gid_offset(dev_addr),
183 sizeof(*gid));
184}
185
186static inline void rdma_addr_set_sgid(struct rdma_dev_addr *dev_addr, union ib_gid *gid)
187{
188 memcpy(dev_addr->src_dev_addr + rdma_addr_gid_offset(dev_addr), gid, sizeof *gid);
189}
190
191static inline void rdma_addr_get_dgid(struct rdma_dev_addr *dev_addr, union ib_gid *gid)
192{
193 memcpy(gid, dev_addr->dst_dev_addr + rdma_addr_gid_offset(dev_addr), sizeof *gid);
194}
195
196static inline void rdma_addr_set_dgid(struct rdma_dev_addr *dev_addr, union ib_gid *gid)
197{
198 memcpy(dev_addr->dst_dev_addr + rdma_addr_gid_offset(dev_addr), gid, sizeof *gid);
199}
200
201static inline enum ib_mtu iboe_get_mtu(int mtu)
202{
203 /*
204 * Reduce IB headers from effective IBoE MTU.
205 */
206 mtu = mtu - (IB_GRH_BYTES + IB_UDP_BYTES + IB_BTH_BYTES +
207 IB_EXT_XRC_BYTES + IB_EXT_ATOMICETH_BYTES +
208 IB_ICRC_BYTES);
209
210 if (mtu >= ib_mtu_enum_to_int(IB_MTU_4096))
211 return IB_MTU_4096;
212 else if (mtu >= ib_mtu_enum_to_int(IB_MTU_2048))
213 return IB_MTU_2048;
214 else if (mtu >= ib_mtu_enum_to_int(IB_MTU_1024))
215 return IB_MTU_1024;
216 else if (mtu >= ib_mtu_enum_to_int(IB_MTU_512))
217 return IB_MTU_512;
218 else if (mtu >= ib_mtu_enum_to_int(IB_MTU_256))
219 return IB_MTU_256;
220 else
221 return 0;
222}
223
224static inline int iboe_get_rate(struct net_device *dev)
225{
226 struct ethtool_link_ksettings cmd;
227 int err;
228
229 rtnl_lock();
230 err = __ethtool_get_link_ksettings(dev, &cmd);
231 rtnl_unlock();
232 if (err)
233 return IB_RATE_PORT_CURRENT;
234
235 if (cmd.base.speed >= 40000)
236 return IB_RATE_40_GBPS;
237 else if (cmd.base.speed >= 30000)
238 return IB_RATE_30_GBPS;
239 else if (cmd.base.speed >= 20000)
240 return IB_RATE_20_GBPS;
241 else if (cmd.base.speed >= 10000)
242 return IB_RATE_10_GBPS;
243 else
244 return IB_RATE_PORT_CURRENT;
245}
246
247static inline int rdma_link_local_addr(struct in6_addr *addr)
248{
249 if (addr->s6_addr32[0] == htonl(0xfe800000) &&
250 addr->s6_addr32[1] == 0)
251 return 1;
252
253 return 0;
254}
255
256static inline void rdma_get_ll_mac(struct in6_addr *addr, u8 *mac)
257{
258 memcpy(mac, &addr->s6_addr[8], 3);
259 memcpy(mac + 3, &addr->s6_addr[13], 3);
260 mac[0] ^= 2;
261}
262
263static inline int rdma_is_multicast_addr(struct in6_addr *addr)
264{
265 __be32 ipv4_addr;
266
267 if (addr->s6_addr[0] == 0xff)
268 return 1;
269
270 ipv4_addr = addr->s6_addr32[3];
271 return (ipv6_addr_v4mapped(addr) && ipv4_is_multicast(ipv4_addr));
272}
273
274static inline void rdma_get_mcast_mac(struct in6_addr *addr, u8 *mac)
275{
276 int i;
277
278 mac[0] = 0x33;
279 mac[1] = 0x33;
280 for (i = 2; i < 6; ++i)
281 mac[i] = addr->s6_addr[i + 10];
282}
283
284static inline u16 rdma_get_vlan_id(union ib_gid *dgid)
285{
286 u16 vid;
287
288 vid = dgid->raw[11] << 8 | dgid->raw[12];
289 return vid < 0x1000 ? vid : 0xffff;
290}
291
292static inline struct net_device *rdma_vlan_dev_real_dev(const struct net_device *dev)
293{
294 return is_vlan_dev(dev) ? vlan_dev_real_dev(dev) : NULL;
295}
296
297#endif /* IB_ADDR_H */