4 * Copyright 2017 6WIND S.A.
5 * Copyright 2017 Mellanox.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
11 * * Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * * Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in
15 * the documentation and/or other materials provided with the
17 * * Neither the name of 6WIND S.A. nor the names of its
18 * contributors may be used to endorse or promote products derived
19 * from this software without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
24 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
25 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
26 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
27 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
28 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
29 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
30 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
31 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
36 #include <linux/netlink.h>
38 #include <sys/socket.h>
41 #include <rte_malloc.h>
42 #include <tap_netlink.h>
43 #include <rte_random.h>
45 /* Must be quite large to support dumping a huge list of QDISC or filters. */
46 #define BUF_SIZE (32 * 1024) /* Size of the buffer to receive kernel messages */
47 #define SNDBUF_SIZE 32768 /* Send buffer size for the netlink socket */
48 #define RCVBUF_SIZE 32768 /* Receive buffer size for the netlink socket */
52 struct nested_tail *prev;
56 * Initialize a netlink socket for communicating with the kernel.
59 * Set it to a netlink group value (e.g. RTMGRP_LINK) to receive messages for
60 * specific netlink multicast groups. Otherwise, no subscription will be made.
63 * netlink socket file descriptor on success, -1 otherwise.
66 nl_init(uint32_t nl_groups)
68 int fd, sndbuf_size = SNDBUF_SIZE, rcvbuf_size = RCVBUF_SIZE;
69 struct sockaddr_nl local = {
70 .nl_family = AF_NETLINK,
71 .nl_groups = nl_groups,
74 fd = socket(AF_NETLINK, SOCK_RAW | SOCK_CLOEXEC, NETLINK_ROUTE);
76 RTE_LOG(ERR, PMD, "Unable to create a netlink socket\n");
79 if (setsockopt(fd, SOL_SOCKET, SO_SNDBUF, &sndbuf_size, sizeof(int))) {
80 RTE_LOG(ERR, PMD, "Unable to set socket buffer send size\n");
83 if (setsockopt(fd, SOL_SOCKET, SO_RCVBUF, &rcvbuf_size, sizeof(int))) {
84 RTE_LOG(ERR, PMD, "Unable to set socket buffer receive size\n");
87 if (bind(fd, (struct sockaddr *)&local, sizeof(local)) < 0) {
88 RTE_LOG(ERR, PMD, "Unable to bind to the netlink socket\n");
95 * Clean up a netlink socket once all communicating with the kernel is finished.
98 * The netlink socket file descriptor used for communication.
101 * 0 on success, -1 otherwise.
104 nl_final(int nlsk_fd)
106 if (close(nlsk_fd)) {
107 RTE_LOG(ERR, PMD, "Failed to close netlink socket: %s (%d)\n",
108 strerror(errno), errno);
115 * Send a message to the kernel on the netlink socket.
118 * The netlink socket file descriptor used for communication.
120 * The netlink message send to the kernel.
123 * the number of sent bytes on success, -1 otherwise.
126 nl_send(int nlsk_fd, struct nlmsghdr *nh)
128 /* man 7 netlink EXAMPLE */
129 struct sockaddr_nl sa = {
130 .nl_family = AF_NETLINK,
134 .iov_len = nh->nlmsg_len,
136 struct msghdr msg = {
138 .msg_namelen = sizeof(sa),
144 nh->nlmsg_pid = 0; /* communication with the kernel uses pid 0 */
145 nh->nlmsg_seq = (uint32_t)rte_rand();
146 send_bytes = sendmsg(nlsk_fd, &msg, 0);
147 if (send_bytes < 0) {
148 RTE_LOG(ERR, PMD, "Failed to send netlink message: %s (%d)\n",
149 strerror(errno), errno);
156 * Check that the kernel sends an appropriate ACK in response to an nl_send().
159 * The netlink socket file descriptor used for communication.
162 * 0 on success, -1 otherwise with errno set.
165 nl_recv_ack(int nlsk_fd)
167 return nl_recv(nlsk_fd, NULL, NULL);
171 * Receive a message from the kernel on the netlink socket, following an
175 * The netlink socket file descriptor used for communication.
177 * The callback function to call for each netlink message received.
178 * @param[in, out] arg
179 * Custom arguments for the callback.
182 * 0 on success, -1 otherwise with errno set.
185 nl_recv(int nlsk_fd, int (*cb)(struct nlmsghdr *, void *arg), void *arg)
187 /* man 7 netlink EXAMPLE */
188 struct sockaddr_nl sa;
192 .iov_len = sizeof(buf),
194 struct msghdr msg = {
196 .msg_namelen = sizeof(sa),
198 /* One message at a time */
208 recv_bytes = recvmsg(nlsk_fd, &msg, 0);
211 for (nh = (struct nlmsghdr *)buf;
212 NLMSG_OK(nh, (unsigned int)recv_bytes);
213 nh = NLMSG_NEXT(nh, recv_bytes)) {
214 if (nh->nlmsg_type == NLMSG_ERROR) {
215 struct nlmsgerr *err_data = NLMSG_DATA(nh);
217 if (err_data->error < 0) {
218 errno = -err_data->error;
224 /* Multi-part msgs and their trailing DONE message. */
225 if (nh->nlmsg_flags & NLM_F_MULTI) {
226 if (nh->nlmsg_type == NLMSG_DONE)
238 * Append a netlink attribute to a message.
241 * The netlink message to parse, received from the kernel.
243 * The type of attribute to append.
244 * @param[in] data_len
245 * The length of the data to append.
247 * The data to append.
250 nlattr_add(struct nlmsghdr *nh, unsigned short type,
251 unsigned int data_len, const void *data)
253 /* see man 3 rtnetlink */
256 rta = (struct rtattr *)NLMSG_TAIL(nh);
257 rta->rta_len = RTA_LENGTH(data_len);
258 rta->rta_type = type;
259 memcpy(RTA_DATA(rta), data, data_len);
260 nh->nlmsg_len = NLMSG_ALIGN(nh->nlmsg_len) + RTA_ALIGN(rta->rta_len);
264 * Append a uint8_t netlink attribute to a message.
267 * The netlink message to parse, received from the kernel.
269 * The type of attribute to append.
271 * The data to append.
274 nlattr_add8(struct nlmsghdr *nh, unsigned short type, uint8_t data)
276 nlattr_add(nh, type, sizeof(uint8_t), &data);
280 * Append a uint16_t netlink attribute to a message.
283 * The netlink message to parse, received from the kernel.
285 * The type of attribute to append.
287 * The data to append.
290 nlattr_add16(struct nlmsghdr *nh, unsigned short type, uint16_t data)
292 nlattr_add(nh, type, sizeof(uint16_t), &data);
296 * Append a uint16_t netlink attribute to a message.
299 * The netlink message to parse, received from the kernel.
301 * The type of attribute to append.
303 * The data to append.
306 nlattr_add32(struct nlmsghdr *nh, unsigned short type, uint32_t data)
308 nlattr_add(nh, type, sizeof(uint32_t), &data);
312 * Start a nested netlink attribute.
313 * It must be followed later by a call to nlattr_nested_finish().
315 * @param[in, out] msg
316 * The netlink message where to edit the nested_tails metadata.
318 * The nested attribute type to append.
321 * -1 if adding a nested netlink attribute failed, 0 otherwise.
324 nlattr_nested_start(struct nlmsg *msg, uint16_t type)
326 struct nested_tail *tail;
328 tail = rte_zmalloc(NULL, sizeof(struct nested_tail), 0);
331 "Couldn't allocate memory for nested netlink"
336 tail->tail = (struct rtattr *)NLMSG_TAIL(&msg->nh);
338 nlattr_add(&msg->nh, type, 0, NULL);
340 tail->prev = msg->nested_tails;
342 msg->nested_tails = tail;
348 * End a nested netlink attribute.
349 * It follows a call to nlattr_nested_start().
350 * In effect, it will modify the nested attribute length to include every bytes
351 * from the nested attribute start, up to here.
353 * @param[in, out] msg
354 * The netlink message where to edit the nested_tails metadata.
357 nlattr_nested_finish(struct nlmsg *msg)
359 struct nested_tail *tail = msg->nested_tails;
361 tail->tail->rta_len = (char *)NLMSG_TAIL(&msg->nh) - (char *)tail->tail;
364 msg->nested_tails = tail->prev;