Commit | Line | Data |
---|---|---|
3d14c5d2 | 1 | #include <linux/ceph/ceph_debug.h> |
31b8006e SW |
2 | |
3 | #include <linux/crc32c.h> | |
4 | #include <linux/ctype.h> | |
5 | #include <linux/highmem.h> | |
6 | #include <linux/inet.h> | |
7 | #include <linux/kthread.h> | |
8 | #include <linux/net.h> | |
5a0e3ad6 | 9 | #include <linux/slab.h> |
31b8006e SW |
10 | #include <linux/socket.h> |
11 | #include <linux/string.h> | |
68b4476b YS |
12 | #include <linux/bio.h> |
13 | #include <linux/blkdev.h> | |
ee3b56f2 | 14 | #include <linux/dns_resolver.h> |
31b8006e SW |
15 | #include <net/tcp.h> |
16 | ||
3d14c5d2 YS |
17 | #include <linux/ceph/libceph.h> |
18 | #include <linux/ceph/messenger.h> | |
19 | #include <linux/ceph/decode.h> | |
20 | #include <linux/ceph/pagelist.h> | |
bc3b2d7f | 21 | #include <linux/export.h> |
31b8006e SW |
22 | |
23 | /* | |
24 | * Ceph uses the messenger to exchange ceph_msg messages with other | |
25 | * hosts in the system. The messenger provides ordered and reliable | |
26 | * delivery. We tolerate TCP disconnects by reconnecting (with | |
27 | * exponential backoff) in the case of a fault (disconnection, bad | |
28 | * crc, protocol error). Acks allow sent messages to be discarded by | |
29 | * the sender. | |
30 | */ | |
31 | ||
32 | /* static tag bytes (protocol control messages) */ | |
33 | static char tag_msg = CEPH_MSGR_TAG_MSG; | |
34 | static char tag_ack = CEPH_MSGR_TAG_ACK; | |
35 | static char tag_keepalive = CEPH_MSGR_TAG_KEEPALIVE; | |
36 | ||
a6a5349d SW |
37 | #ifdef CONFIG_LOCKDEP |
38 | static struct lock_class_key socket_class; | |
39 | #endif | |
40 | ||
31b8006e SW |
41 | |
42 | static void queue_con(struct ceph_connection *con); | |
43 | static void con_work(struct work_struct *); | |
44 | static void ceph_fault(struct ceph_connection *con); | |
45 | ||
31b8006e SW |
46 | /* |
47 | * nicely render a sockaddr as a string. | |
48 | */ | |
49 | #define MAX_ADDR_STR 20 | |
d06dbaf6 SW |
50 | #define MAX_ADDR_STR_LEN 60 |
51 | static char addr_str[MAX_ADDR_STR][MAX_ADDR_STR_LEN]; | |
31b8006e SW |
52 | static DEFINE_SPINLOCK(addr_str_lock); |
53 | static int last_addr_str; | |
54 | ||
3d14c5d2 | 55 | const char *ceph_pr_addr(const struct sockaddr_storage *ss) |
31b8006e SW |
56 | { |
57 | int i; | |
58 | char *s; | |
59 | struct sockaddr_in *in4 = (void *)ss; | |
31b8006e SW |
60 | struct sockaddr_in6 *in6 = (void *)ss; |
61 | ||
62 | spin_lock(&addr_str_lock); | |
63 | i = last_addr_str++; | |
64 | if (last_addr_str == MAX_ADDR_STR) | |
65 | last_addr_str = 0; | |
66 | spin_unlock(&addr_str_lock); | |
67 | s = addr_str[i]; | |
68 | ||
69 | switch (ss->ss_family) { | |
70 | case AF_INET: | |
d06dbaf6 SW |
71 | snprintf(s, MAX_ADDR_STR_LEN, "%pI4:%u", &in4->sin_addr, |
72 | (unsigned int)ntohs(in4->sin_port)); | |
31b8006e SW |
73 | break; |
74 | ||
75 | case AF_INET6: | |
d06dbaf6 SW |
76 | snprintf(s, MAX_ADDR_STR_LEN, "[%pI6c]:%u", &in6->sin6_addr, |
77 | (unsigned int)ntohs(in6->sin6_port)); | |
31b8006e SW |
78 | break; |
79 | ||
80 | default: | |
12a2f643 SW |
81 | snprintf(s, MAX_ADDR_STR_LEN, "(unknown sockaddr family %d)", |
82 | (int)ss->ss_family); | |
31b8006e SW |
83 | } |
84 | ||
85 | return s; | |
86 | } | |
3d14c5d2 | 87 | EXPORT_SYMBOL(ceph_pr_addr); |
31b8006e | 88 | |
63f2d211 SW |
89 | static void encode_my_addr(struct ceph_messenger *msgr) |
90 | { | |
91 | memcpy(&msgr->my_enc_addr, &msgr->inst.addr, sizeof(msgr->my_enc_addr)); | |
92 | ceph_encode_addr(&msgr->my_enc_addr); | |
93 | } | |
94 | ||
31b8006e SW |
95 | /* |
96 | * work queue for all reading and writing to/from the socket. | |
97 | */ | |
98 | struct workqueue_struct *ceph_msgr_wq; | |
99 | ||
3d14c5d2 | 100 | int ceph_msgr_init(void) |
31b8006e | 101 | { |
f363e45f | 102 | ceph_msgr_wq = alloc_workqueue("ceph-msgr", WQ_NON_REENTRANT, 0); |
d96c9043 SW |
103 | if (!ceph_msgr_wq) { |
104 | pr_err("msgr_init failed to create workqueue\n"); | |
105 | return -ENOMEM; | |
31b8006e SW |
106 | } |
107 | return 0; | |
108 | } | |
3d14c5d2 | 109 | EXPORT_SYMBOL(ceph_msgr_init); |
31b8006e SW |
110 | |
111 | void ceph_msgr_exit(void) | |
112 | { | |
113 | destroy_workqueue(ceph_msgr_wq); | |
114 | } | |
3d14c5d2 | 115 | EXPORT_SYMBOL(ceph_msgr_exit); |
31b8006e | 116 | |
cd84db6e | 117 | void ceph_msgr_flush(void) |
a922d38f SW |
118 | { |
119 | flush_workqueue(ceph_msgr_wq); | |
120 | } | |
3d14c5d2 | 121 | EXPORT_SYMBOL(ceph_msgr_flush); |
a922d38f SW |
122 | |
123 | ||
31b8006e SW |
124 | /* |
125 | * socket callback functions | |
126 | */ | |
127 | ||
128 | /* data available on socket, or listen socket received a connect */ | |
129 | static void ceph_data_ready(struct sock *sk, int count_unused) | |
130 | { | |
131 | struct ceph_connection *con = | |
132 | (struct ceph_connection *)sk->sk_user_data; | |
133 | if (sk->sk_state != TCP_CLOSE_WAIT) { | |
134 | dout("ceph_data_ready on %p state = %lu, queueing work\n", | |
135 | con, con->state); | |
136 | queue_con(con); | |
137 | } | |
138 | } | |
139 | ||
140 | /* socket has buffer space for writing */ | |
141 | static void ceph_write_space(struct sock *sk) | |
142 | { | |
143 | struct ceph_connection *con = | |
144 | (struct ceph_connection *)sk->sk_user_data; | |
145 | ||
182fac26 JS |
146 | /* only queue to workqueue if there is data we want to write, |
147 | * and there is sufficient space in the socket buffer to accept | |
148 | * more data. clear SOCK_NOSPACE so that ceph_write_space() | |
149 | * doesn't get called again until try_write() fills the socket | |
150 | * buffer. See net/ipv4/tcp_input.c:tcp_check_space() | |
151 | * and net/core/stream.c:sk_stream_write_space(). | |
152 | */ | |
31b8006e | 153 | if (test_bit(WRITE_PENDING, &con->state)) { |
182fac26 JS |
154 | if (sk_stream_wspace(sk) >= sk_stream_min_wspace(sk)) { |
155 | dout("ceph_write_space %p queueing write work\n", con); | |
156 | clear_bit(SOCK_NOSPACE, &sk->sk_socket->flags); | |
157 | queue_con(con); | |
158 | } | |
31b8006e SW |
159 | } else { |
160 | dout("ceph_write_space %p nothing to write\n", con); | |
161 | } | |
31b8006e SW |
162 | } |
163 | ||
164 | /* socket's state has changed */ | |
165 | static void ceph_state_change(struct sock *sk) | |
166 | { | |
167 | struct ceph_connection *con = | |
168 | (struct ceph_connection *)sk->sk_user_data; | |
169 | ||
170 | dout("ceph_state_change %p state = %lu sk_state = %u\n", | |
171 | con, con->state, sk->sk_state); | |
172 | ||
173 | if (test_bit(CLOSED, &con->state)) | |
174 | return; | |
175 | ||
176 | switch (sk->sk_state) { | |
177 | case TCP_CLOSE: | |
178 | dout("ceph_state_change TCP_CLOSE\n"); | |
179 | case TCP_CLOSE_WAIT: | |
180 | dout("ceph_state_change TCP_CLOSE_WAIT\n"); | |
181 | if (test_and_set_bit(SOCK_CLOSED, &con->state) == 0) { | |
182 | if (test_bit(CONNECTING, &con->state)) | |
183 | con->error_msg = "connection failed"; | |
184 | else | |
185 | con->error_msg = "socket closed"; | |
186 | queue_con(con); | |
187 | } | |
188 | break; | |
189 | case TCP_ESTABLISHED: | |
190 | dout("ceph_state_change TCP_ESTABLISHED\n"); | |
191 | queue_con(con); | |
192 | break; | |
193 | } | |
194 | } | |
195 | ||
196 | /* | |
197 | * set up socket callbacks | |
198 | */ | |
199 | static void set_sock_callbacks(struct socket *sock, | |
200 | struct ceph_connection *con) | |
201 | { | |
202 | struct sock *sk = sock->sk; | |
203 | sk->sk_user_data = (void *)con; | |
204 | sk->sk_data_ready = ceph_data_ready; | |
205 | sk->sk_write_space = ceph_write_space; | |
206 | sk->sk_state_change = ceph_state_change; | |
207 | } | |
208 | ||
209 | ||
210 | /* | |
211 | * socket helpers | |
212 | */ | |
213 | ||
214 | /* | |
215 | * initiate connection to a remote socket. | |
216 | */ | |
217 | static struct socket *ceph_tcp_connect(struct ceph_connection *con) | |
218 | { | |
f91d3471 | 219 | struct sockaddr_storage *paddr = &con->peer_addr.in_addr; |
31b8006e SW |
220 | struct socket *sock; |
221 | int ret; | |
222 | ||
223 | BUG_ON(con->sock); | |
f91d3471 SW |
224 | ret = sock_create_kern(con->peer_addr.in_addr.ss_family, SOCK_STREAM, |
225 | IPPROTO_TCP, &sock); | |
31b8006e SW |
226 | if (ret) |
227 | return ERR_PTR(ret); | |
228 | con->sock = sock; | |
229 | sock->sk->sk_allocation = GFP_NOFS; | |
230 | ||
a6a5349d SW |
231 | #ifdef CONFIG_LOCKDEP |
232 | lockdep_set_class(&sock->sk->sk_lock, &socket_class); | |
233 | #endif | |
234 | ||
31b8006e SW |
235 | set_sock_callbacks(sock, con); |
236 | ||
3d14c5d2 | 237 | dout("connect %s\n", ceph_pr_addr(&con->peer_addr.in_addr)); |
31b8006e | 238 | |
f91d3471 SW |
239 | ret = sock->ops->connect(sock, (struct sockaddr *)paddr, sizeof(*paddr), |
240 | O_NONBLOCK); | |
31b8006e SW |
241 | if (ret == -EINPROGRESS) { |
242 | dout("connect %s EINPROGRESS sk_state = %u\n", | |
3d14c5d2 | 243 | ceph_pr_addr(&con->peer_addr.in_addr), |
31b8006e SW |
244 | sock->sk->sk_state); |
245 | ret = 0; | |
246 | } | |
247 | if (ret < 0) { | |
248 | pr_err("connect %s error %d\n", | |
3d14c5d2 | 249 | ceph_pr_addr(&con->peer_addr.in_addr), ret); |
31b8006e SW |
250 | sock_release(sock); |
251 | con->sock = NULL; | |
252 | con->error_msg = "connect error"; | |
253 | } | |
254 | ||
255 | if (ret < 0) | |
256 | return ERR_PTR(ret); | |
257 | return sock; | |
258 | } | |
259 | ||
260 | static int ceph_tcp_recvmsg(struct socket *sock, void *buf, size_t len) | |
261 | { | |
262 | struct kvec iov = {buf, len}; | |
263 | struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL }; | |
98bdb0aa | 264 | int r; |
31b8006e | 265 | |
98bdb0aa SW |
266 | r = kernel_recvmsg(sock, &msg, &iov, 1, len, msg.msg_flags); |
267 | if (r == -EAGAIN) | |
268 | r = 0; | |
269 | return r; | |
31b8006e SW |
270 | } |
271 | ||
272 | /* | |
273 | * write something. @more is true if caller will be sending more data | |
274 | * shortly. | |
275 | */ | |
276 | static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov, | |
277 | size_t kvlen, size_t len, int more) | |
278 | { | |
279 | struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL }; | |
42961d23 | 280 | int r; |
31b8006e SW |
281 | |
282 | if (more) | |
283 | msg.msg_flags |= MSG_MORE; | |
284 | else | |
285 | msg.msg_flags |= MSG_EOR; /* superfluous, but what the hell */ | |
286 | ||
42961d23 SW |
287 | r = kernel_sendmsg(sock, &msg, iov, kvlen, len); |
288 | if (r == -EAGAIN) | |
289 | r = 0; | |
290 | return r; | |
31b8006e SW |
291 | } |
292 | ||
293 | ||
294 | /* | |
295 | * Shutdown/close the socket for the given connection. | |
296 | */ | |
297 | static int con_close_socket(struct ceph_connection *con) | |
298 | { | |
299 | int rc; | |
300 | ||
301 | dout("con_close_socket on %p sock %p\n", con, con->sock); | |
302 | if (!con->sock) | |
303 | return 0; | |
304 | set_bit(SOCK_CLOSED, &con->state); | |
305 | rc = con->sock->ops->shutdown(con->sock, SHUT_RDWR); | |
306 | sock_release(con->sock); | |
307 | con->sock = NULL; | |
308 | clear_bit(SOCK_CLOSED, &con->state); | |
309 | return rc; | |
310 | } | |
311 | ||
312 | /* | |
313 | * Reset a connection. Discard all incoming and outgoing messages | |
314 | * and clear *_seq state. | |
315 | */ | |
316 | static void ceph_msg_remove(struct ceph_msg *msg) | |
317 | { | |
318 | list_del_init(&msg->list_head); | |
319 | ceph_msg_put(msg); | |
320 | } | |
321 | static void ceph_msg_remove_list(struct list_head *head) | |
322 | { | |
323 | while (!list_empty(head)) { | |
324 | struct ceph_msg *msg = list_first_entry(head, struct ceph_msg, | |
325 | list_head); | |
326 | ceph_msg_remove(msg); | |
327 | } | |
328 | } | |
329 | ||
330 | static void reset_connection(struct ceph_connection *con) | |
331 | { | |
332 | /* reset connection, out_queue, msg_ and connect_seq */ | |
333 | /* discard existing out_queue and msg_seq */ | |
31b8006e SW |
334 | ceph_msg_remove_list(&con->out_queue); |
335 | ceph_msg_remove_list(&con->out_sent); | |
336 | ||
cf3e5c40 SW |
337 | if (con->in_msg) { |
338 | ceph_msg_put(con->in_msg); | |
339 | con->in_msg = NULL; | |
340 | } | |
341 | ||
31b8006e SW |
342 | con->connect_seq = 0; |
343 | con->out_seq = 0; | |
c86a2930 SW |
344 | if (con->out_msg) { |
345 | ceph_msg_put(con->out_msg); | |
346 | con->out_msg = NULL; | |
347 | } | |
31b8006e | 348 | con->in_seq = 0; |
0e0d5e0c | 349 | con->in_seq_acked = 0; |
31b8006e SW |
350 | } |
351 | ||
352 | /* | |
353 | * mark a peer down. drop any open connections. | |
354 | */ | |
355 | void ceph_con_close(struct ceph_connection *con) | |
356 | { | |
3d14c5d2 YS |
357 | dout("con_close %p peer %s\n", con, |
358 | ceph_pr_addr(&con->peer_addr.in_addr)); | |
31b8006e SW |
359 | set_bit(CLOSED, &con->state); /* in case there's queued work */ |
360 | clear_bit(STANDBY, &con->state); /* avoid connect_seq bump */ | |
1679f876 SW |
361 | clear_bit(LOSSYTX, &con->state); /* so we retry next connect */ |
362 | clear_bit(KEEPALIVE_PENDING, &con->state); | |
363 | clear_bit(WRITE_PENDING, &con->state); | |
ec302645 | 364 | mutex_lock(&con->mutex); |
31b8006e | 365 | reset_connection(con); |
6f2bc3ff | 366 | con->peer_global_seq = 0; |
91e45ce3 | 367 | cancel_delayed_work(&con->work); |
ec302645 | 368 | mutex_unlock(&con->mutex); |
31b8006e SW |
369 | queue_con(con); |
370 | } | |
3d14c5d2 | 371 | EXPORT_SYMBOL(ceph_con_close); |
31b8006e | 372 | |
31b8006e SW |
373 | /* |
374 | * Reopen a closed connection, with a new peer address. | |
375 | */ | |
376 | void ceph_con_open(struct ceph_connection *con, struct ceph_entity_addr *addr) | |
377 | { | |
3d14c5d2 | 378 | dout("con_open %p %s\n", con, ceph_pr_addr(&addr->in_addr)); |
31b8006e SW |
379 | set_bit(OPENING, &con->state); |
380 | clear_bit(CLOSED, &con->state); | |
381 | memcpy(&con->peer_addr, addr, sizeof(*addr)); | |
03c677e1 | 382 | con->delay = 0; /* reset backoff memory */ |
31b8006e SW |
383 | queue_con(con); |
384 | } | |
3d14c5d2 | 385 | EXPORT_SYMBOL(ceph_con_open); |
31b8006e | 386 | |
87b315a5 SW |
387 | /* |
388 | * return true if this connection ever successfully opened | |
389 | */ | |
390 | bool ceph_con_opened(struct ceph_connection *con) | |
391 | { | |
392 | return con->connect_seq > 0; | |
393 | } | |
394 | ||
31b8006e SW |
395 | /* |
396 | * generic get/put | |
397 | */ | |
398 | struct ceph_connection *ceph_con_get(struct ceph_connection *con) | |
399 | { | |
400 | dout("con_get %p nref = %d -> %d\n", con, | |
401 | atomic_read(&con->nref), atomic_read(&con->nref) + 1); | |
402 | if (atomic_inc_not_zero(&con->nref)) | |
403 | return con; | |
404 | return NULL; | |
405 | } | |
406 | ||
407 | void ceph_con_put(struct ceph_connection *con) | |
408 | { | |
409 | dout("con_put %p nref = %d -> %d\n", con, | |
410 | atomic_read(&con->nref), atomic_read(&con->nref) - 1); | |
411 | BUG_ON(atomic_read(&con->nref) == 0); | |
412 | if (atomic_dec_and_test(&con->nref)) { | |
71ececda | 413 | BUG_ON(con->sock); |
31b8006e SW |
414 | kfree(con); |
415 | } | |
416 | } | |
417 | ||
418 | /* | |
419 | * initialize a new connection. | |
420 | */ | |
421 | void ceph_con_init(struct ceph_messenger *msgr, struct ceph_connection *con) | |
422 | { | |
423 | dout("con_init %p\n", con); | |
424 | memset(con, 0, sizeof(*con)); | |
425 | atomic_set(&con->nref, 1); | |
426 | con->msgr = msgr; | |
ec302645 | 427 | mutex_init(&con->mutex); |
31b8006e SW |
428 | INIT_LIST_HEAD(&con->out_queue); |
429 | INIT_LIST_HEAD(&con->out_sent); | |
430 | INIT_DELAYED_WORK(&con->work, con_work); | |
431 | } | |
3d14c5d2 | 432 | EXPORT_SYMBOL(ceph_con_init); |
31b8006e SW |
433 | |
434 | ||
435 | /* | |
436 | * We maintain a global counter to order connection attempts. Get | |
437 | * a unique seq greater than @gt. | |
438 | */ | |
439 | static u32 get_global_seq(struct ceph_messenger *msgr, u32 gt) | |
440 | { | |
441 | u32 ret; | |
442 | ||
443 | spin_lock(&msgr->global_seq_lock); | |
444 | if (msgr->global_seq < gt) | |
445 | msgr->global_seq = gt; | |
446 | ret = ++msgr->global_seq; | |
447 | spin_unlock(&msgr->global_seq_lock); | |
448 | return ret; | |
449 | } | |
450 | ||
451 | ||
452 | /* | |
453 | * Prepare footer for currently outgoing message, and finish things | |
454 | * off. Assumes out_kvec* are already valid.. we just add on to the end. | |
455 | */ | |
456 | static void prepare_write_message_footer(struct ceph_connection *con, int v) | |
457 | { | |
458 | struct ceph_msg *m = con->out_msg; | |
459 | ||
460 | dout("prepare_write_message_footer %p\n", con); | |
461 | con->out_kvec_is_msg = true; | |
462 | con->out_kvec[v].iov_base = &m->footer; | |
463 | con->out_kvec[v].iov_len = sizeof(m->footer); | |
464 | con->out_kvec_bytes += sizeof(m->footer); | |
465 | con->out_kvec_left++; | |
466 | con->out_more = m->more_to_follow; | |
c86a2930 | 467 | con->out_msg_done = true; |
31b8006e SW |
468 | } |
469 | ||
470 | /* | |
471 | * Prepare headers for the next outgoing message. | |
472 | */ | |
473 | static void prepare_write_message(struct ceph_connection *con) | |
474 | { | |
475 | struct ceph_msg *m; | |
476 | int v = 0; | |
477 | ||
478 | con->out_kvec_bytes = 0; | |
479 | con->out_kvec_is_msg = true; | |
c86a2930 | 480 | con->out_msg_done = false; |
31b8006e SW |
481 | |
482 | /* Sneak an ack in there first? If we can get it into the same | |
483 | * TCP packet that's a good thing. */ | |
484 | if (con->in_seq > con->in_seq_acked) { | |
485 | con->in_seq_acked = con->in_seq; | |
486 | con->out_kvec[v].iov_base = &tag_ack; | |
487 | con->out_kvec[v++].iov_len = 1; | |
488 | con->out_temp_ack = cpu_to_le64(con->in_seq_acked); | |
489 | con->out_kvec[v].iov_base = &con->out_temp_ack; | |
490 | con->out_kvec[v++].iov_len = sizeof(con->out_temp_ack); | |
491 | con->out_kvec_bytes = 1 + sizeof(con->out_temp_ack); | |
492 | } | |
493 | ||
31b8006e SW |
494 | m = list_first_entry(&con->out_queue, |
495 | struct ceph_msg, list_head); | |
c86a2930 | 496 | con->out_msg = m; |
4cf9d544 SW |
497 | |
498 | /* put message on sent list */ | |
499 | ceph_msg_get(m); | |
500 | list_move_tail(&m->list_head, &con->out_sent); | |
31b8006e | 501 | |
e84346b7 SW |
502 | /* |
503 | * only assign outgoing seq # if we haven't sent this message | |
504 | * yet. if it is requeued, resend with it's original seq. | |
505 | */ | |
506 | if (m->needs_out_seq) { | |
507 | m->hdr.seq = cpu_to_le64(++con->out_seq); | |
508 | m->needs_out_seq = false; | |
509 | } | |
31b8006e SW |
510 | |
511 | dout("prepare_write_message %p seq %lld type %d len %d+%d+%d %d pgs\n", | |
512 | m, con->out_seq, le16_to_cpu(m->hdr.type), | |
513 | le32_to_cpu(m->hdr.front_len), le32_to_cpu(m->hdr.middle_len), | |
514 | le32_to_cpu(m->hdr.data_len), | |
515 | m->nr_pages); | |
516 | BUG_ON(le32_to_cpu(m->hdr.front_len) != m->front.iov_len); | |
517 | ||
518 | /* tag + hdr + front + middle */ | |
519 | con->out_kvec[v].iov_base = &tag_msg; | |
520 | con->out_kvec[v++].iov_len = 1; | |
521 | con->out_kvec[v].iov_base = &m->hdr; | |
522 | con->out_kvec[v++].iov_len = sizeof(m->hdr); | |
523 | con->out_kvec[v++] = m->front; | |
524 | if (m->middle) | |
525 | con->out_kvec[v++] = m->middle->vec; | |
526 | con->out_kvec_left = v; | |
527 | con->out_kvec_bytes += 1 + sizeof(m->hdr) + m->front.iov_len + | |
528 | (m->middle ? m->middle->vec.iov_len : 0); | |
529 | con->out_kvec_cur = con->out_kvec; | |
530 | ||
531 | /* fill in crc (except data pages), footer */ | |
532 | con->out_msg->hdr.crc = | |
533 | cpu_to_le32(crc32c(0, (void *)&m->hdr, | |
534 | sizeof(m->hdr) - sizeof(m->hdr.crc))); | |
535 | con->out_msg->footer.flags = CEPH_MSG_FOOTER_COMPLETE; | |
536 | con->out_msg->footer.front_crc = | |
537 | cpu_to_le32(crc32c(0, m->front.iov_base, m->front.iov_len)); | |
538 | if (m->middle) | |
539 | con->out_msg->footer.middle_crc = | |
540 | cpu_to_le32(crc32c(0, m->middle->vec.iov_base, | |
541 | m->middle->vec.iov_len)); | |
542 | else | |
543 | con->out_msg->footer.middle_crc = 0; | |
544 | con->out_msg->footer.data_crc = 0; | |
545 | dout("prepare_write_message front_crc %u data_crc %u\n", | |
546 | le32_to_cpu(con->out_msg->footer.front_crc), | |
547 | le32_to_cpu(con->out_msg->footer.middle_crc)); | |
548 | ||
549 | /* is there a data payload? */ | |
550 | if (le32_to_cpu(m->hdr.data_len) > 0) { | |
551 | /* initialize page iterator */ | |
552 | con->out_msg_pos.page = 0; | |
68b4476b | 553 | if (m->pages) |
c5c6b19d | 554 | con->out_msg_pos.page_pos = m->page_alignment; |
68b4476b YS |
555 | else |
556 | con->out_msg_pos.page_pos = 0; | |
31b8006e SW |
557 | con->out_msg_pos.data_pos = 0; |
558 | con->out_msg_pos.did_page_crc = 0; | |
559 | con->out_more = 1; /* data + footer will follow */ | |
560 | } else { | |
561 | /* no, queue up footer too and be done */ | |
562 | prepare_write_message_footer(con, v); | |
563 | } | |
564 | ||
565 | set_bit(WRITE_PENDING, &con->state); | |
566 | } | |
567 | ||
568 | /* | |
569 | * Prepare an ack. | |
570 | */ | |
571 | static void prepare_write_ack(struct ceph_connection *con) | |
572 | { | |
573 | dout("prepare_write_ack %p %llu -> %llu\n", con, | |
574 | con->in_seq_acked, con->in_seq); | |
575 | con->in_seq_acked = con->in_seq; | |
576 | ||
577 | con->out_kvec[0].iov_base = &tag_ack; | |
578 | con->out_kvec[0].iov_len = 1; | |
579 | con->out_temp_ack = cpu_to_le64(con->in_seq_acked); | |
580 | con->out_kvec[1].iov_base = &con->out_temp_ack; | |
581 | con->out_kvec[1].iov_len = sizeof(con->out_temp_ack); | |
582 | con->out_kvec_left = 2; | |
583 | con->out_kvec_bytes = 1 + sizeof(con->out_temp_ack); | |
584 | con->out_kvec_cur = con->out_kvec; | |
585 | con->out_more = 1; /* more will follow.. eventually.. */ | |
586 | set_bit(WRITE_PENDING, &con->state); | |
587 | } | |
588 | ||
589 | /* | |
590 | * Prepare to write keepalive byte. | |
591 | */ | |
592 | static void prepare_write_keepalive(struct ceph_connection *con) | |
593 | { | |
594 | dout("prepare_write_keepalive %p\n", con); | |
595 | con->out_kvec[0].iov_base = &tag_keepalive; | |
596 | con->out_kvec[0].iov_len = 1; | |
597 | con->out_kvec_left = 1; | |
598 | con->out_kvec_bytes = 1; | |
599 | con->out_kvec_cur = con->out_kvec; | |
600 | set_bit(WRITE_PENDING, &con->state); | |
601 | } | |
602 | ||
603 | /* | |
604 | * Connection negotiation. | |
605 | */ | |
606 | ||
0da5d703 | 607 | static int prepare_connect_authorizer(struct ceph_connection *con) |
4e7a5dcd SW |
608 | { |
609 | void *auth_buf; | |
610 | int auth_len = 0; | |
611 | int auth_protocol = 0; | |
612 | ||
ec302645 | 613 | mutex_unlock(&con->mutex); |
4e7a5dcd SW |
614 | if (con->ops->get_authorizer) |
615 | con->ops->get_authorizer(con, &auth_buf, &auth_len, | |
616 | &auth_protocol, &con->auth_reply_buf, | |
617 | &con->auth_reply_buf_len, | |
618 | con->auth_retry); | |
ec302645 | 619 | mutex_lock(&con->mutex); |
4e7a5dcd | 620 | |
0da5d703 SW |
621 | if (test_bit(CLOSED, &con->state) || |
622 | test_bit(OPENING, &con->state)) | |
623 | return -EAGAIN; | |
624 | ||
4e7a5dcd SW |
625 | con->out_connect.authorizer_protocol = cpu_to_le32(auth_protocol); |
626 | con->out_connect.authorizer_len = cpu_to_le32(auth_len); | |
627 | ||
e8f54ce1 SW |
628 | if (auth_len) { |
629 | con->out_kvec[con->out_kvec_left].iov_base = auth_buf; | |
630 | con->out_kvec[con->out_kvec_left].iov_len = auth_len; | |
631 | con->out_kvec_left++; | |
632 | con->out_kvec_bytes += auth_len; | |
633 | } | |
0da5d703 | 634 | return 0; |
4e7a5dcd SW |
635 | } |
636 | ||
31b8006e SW |
637 | /* |
638 | * We connected to a peer and are saying hello. | |
639 | */ | |
eed0ef2c SW |
640 | static void prepare_write_banner(struct ceph_messenger *msgr, |
641 | struct ceph_connection *con) | |
31b8006e SW |
642 | { |
643 | int len = strlen(CEPH_BANNER); | |
eed0ef2c SW |
644 | |
645 | con->out_kvec[0].iov_base = CEPH_BANNER; | |
646 | con->out_kvec[0].iov_len = len; | |
647 | con->out_kvec[1].iov_base = &msgr->my_enc_addr; | |
648 | con->out_kvec[1].iov_len = sizeof(msgr->my_enc_addr); | |
649 | con->out_kvec_left = 2; | |
650 | con->out_kvec_bytes = len + sizeof(msgr->my_enc_addr); | |
651 | con->out_kvec_cur = con->out_kvec; | |
652 | con->out_more = 0; | |
653 | set_bit(WRITE_PENDING, &con->state); | |
654 | } | |
655 | ||
0da5d703 SW |
656 | static int prepare_write_connect(struct ceph_messenger *msgr, |
657 | struct ceph_connection *con, | |
658 | int after_banner) | |
eed0ef2c | 659 | { |
31b8006e SW |
660 | unsigned global_seq = get_global_seq(con->msgr, 0); |
661 | int proto; | |
662 | ||
663 | switch (con->peer_name.type) { | |
664 | case CEPH_ENTITY_TYPE_MON: | |
665 | proto = CEPH_MONC_PROTOCOL; | |
666 | break; | |
667 | case CEPH_ENTITY_TYPE_OSD: | |
668 | proto = CEPH_OSDC_PROTOCOL; | |
669 | break; | |
670 | case CEPH_ENTITY_TYPE_MDS: | |
671 | proto = CEPH_MDSC_PROTOCOL; | |
672 | break; | |
673 | default: | |
674 | BUG(); | |
675 | } | |
676 | ||
677 | dout("prepare_write_connect %p cseq=%d gseq=%d proto=%d\n", con, | |
678 | con->connect_seq, global_seq, proto); | |
4e7a5dcd | 679 | |
3d14c5d2 | 680 | con->out_connect.features = cpu_to_le64(msgr->supported_features); |
31b8006e SW |
681 | con->out_connect.host_type = cpu_to_le32(CEPH_ENTITY_TYPE_CLIENT); |
682 | con->out_connect.connect_seq = cpu_to_le32(con->connect_seq); | |
683 | con->out_connect.global_seq = cpu_to_le32(global_seq); | |
684 | con->out_connect.protocol_version = cpu_to_le32(proto); | |
685 | con->out_connect.flags = 0; | |
31b8006e | 686 | |
eed0ef2c SW |
687 | if (!after_banner) { |
688 | con->out_kvec_left = 0; | |
689 | con->out_kvec_bytes = 0; | |
690 | } | |
691 | con->out_kvec[con->out_kvec_left].iov_base = &con->out_connect; | |
692 | con->out_kvec[con->out_kvec_left].iov_len = sizeof(con->out_connect); | |
693 | con->out_kvec_left++; | |
694 | con->out_kvec_bytes += sizeof(con->out_connect); | |
31b8006e SW |
695 | con->out_kvec_cur = con->out_kvec; |
696 | con->out_more = 0; | |
697 | set_bit(WRITE_PENDING, &con->state); | |
4e7a5dcd | 698 | |
0da5d703 | 699 | return prepare_connect_authorizer(con); |
31b8006e SW |
700 | } |
701 | ||
702 | ||
703 | /* | |
704 | * write as much of pending kvecs to the socket as we can. | |
705 | * 1 -> done | |
706 | * 0 -> socket full, but more to do | |
707 | * <0 -> error | |
708 | */ | |
709 | static int write_partial_kvec(struct ceph_connection *con) | |
710 | { | |
711 | int ret; | |
712 | ||
713 | dout("write_partial_kvec %p %d left\n", con, con->out_kvec_bytes); | |
714 | while (con->out_kvec_bytes > 0) { | |
715 | ret = ceph_tcp_sendmsg(con->sock, con->out_kvec_cur, | |
716 | con->out_kvec_left, con->out_kvec_bytes, | |
717 | con->out_more); | |
718 | if (ret <= 0) | |
719 | goto out; | |
720 | con->out_kvec_bytes -= ret; | |
721 | if (con->out_kvec_bytes == 0) | |
722 | break; /* done */ | |
723 | while (ret > 0) { | |
724 | if (ret >= con->out_kvec_cur->iov_len) { | |
725 | ret -= con->out_kvec_cur->iov_len; | |
726 | con->out_kvec_cur++; | |
727 | con->out_kvec_left--; | |
728 | } else { | |
729 | con->out_kvec_cur->iov_len -= ret; | |
730 | con->out_kvec_cur->iov_base += ret; | |
731 | ret = 0; | |
732 | break; | |
733 | } | |
734 | } | |
735 | } | |
736 | con->out_kvec_left = 0; | |
737 | con->out_kvec_is_msg = false; | |
738 | ret = 1; | |
739 | out: | |
740 | dout("write_partial_kvec %p %d left in %d kvecs ret = %d\n", con, | |
741 | con->out_kvec_bytes, con->out_kvec_left, ret); | |
742 | return ret; /* done! */ | |
743 | } | |
744 | ||
68b4476b YS |
745 | #ifdef CONFIG_BLOCK |
746 | static void init_bio_iter(struct bio *bio, struct bio **iter, int *seg) | |
747 | { | |
748 | if (!bio) { | |
749 | *iter = NULL; | |
750 | *seg = 0; | |
751 | return; | |
752 | } | |
753 | *iter = bio; | |
754 | *seg = bio->bi_idx; | |
755 | } | |
756 | ||
757 | static void iter_bio_next(struct bio **bio_iter, int *seg) | |
758 | { | |
759 | if (*bio_iter == NULL) | |
760 | return; | |
761 | ||
762 | BUG_ON(*seg >= (*bio_iter)->bi_vcnt); | |
763 | ||
764 | (*seg)++; | |
765 | if (*seg == (*bio_iter)->bi_vcnt) | |
766 | init_bio_iter((*bio_iter)->bi_next, bio_iter, seg); | |
767 | } | |
768 | #endif | |
769 | ||
31b8006e SW |
770 | /* |
771 | * Write as much message data payload as we can. If we finish, queue | |
772 | * up the footer. | |
773 | * 1 -> done, footer is now queued in out_kvec[]. | |
774 | * 0 -> socket full, but more to do | |
775 | * <0 -> error | |
776 | */ | |
777 | static int write_partial_msg_pages(struct ceph_connection *con) | |
778 | { | |
779 | struct ceph_msg *msg = con->out_msg; | |
780 | unsigned data_len = le32_to_cpu(msg->hdr.data_len); | |
781 | size_t len; | |
782 | int crc = con->msgr->nocrc; | |
783 | int ret; | |
68b4476b YS |
784 | int total_max_write; |
785 | int in_trail = 0; | |
786 | size_t trail_len = (msg->trail ? msg->trail->length : 0); | |
31b8006e SW |
787 | |
788 | dout("write_partial_msg_pages %p msg %p page %d/%d offset %d\n", | |
789 | con, con->out_msg, con->out_msg_pos.page, con->out_msg->nr_pages, | |
790 | con->out_msg_pos.page_pos); | |
791 | ||
68b4476b YS |
792 | #ifdef CONFIG_BLOCK |
793 | if (msg->bio && !msg->bio_iter) | |
794 | init_bio_iter(msg->bio, &msg->bio_iter, &msg->bio_seg); | |
795 | #endif | |
796 | ||
797 | while (data_len > con->out_msg_pos.data_pos) { | |
31b8006e SW |
798 | struct page *page = NULL; |
799 | void *kaddr = NULL; | |
68b4476b YS |
800 | int max_write = PAGE_SIZE; |
801 | int page_shift = 0; | |
802 | ||
803 | total_max_write = data_len - trail_len - | |
804 | con->out_msg_pos.data_pos; | |
31b8006e SW |
805 | |
806 | /* | |
807 | * if we are calculating the data crc (the default), we need | |
808 | * to map the page. if our pages[] has been revoked, use the | |
809 | * zero page. | |
810 | */ | |
68b4476b YS |
811 | |
812 | /* have we reached the trail part of the data? */ | |
813 | if (con->out_msg_pos.data_pos >= data_len - trail_len) { | |
814 | in_trail = 1; | |
815 | ||
816 | total_max_write = data_len - con->out_msg_pos.data_pos; | |
817 | ||
818 | page = list_first_entry(&msg->trail->head, | |
819 | struct page, lru); | |
820 | if (crc) | |
821 | kaddr = kmap(page); | |
822 | max_write = PAGE_SIZE; | |
823 | } else if (msg->pages) { | |
31b8006e SW |
824 | page = msg->pages[con->out_msg_pos.page]; |
825 | if (crc) | |
826 | kaddr = kmap(page); | |
58bb3b37 SW |
827 | } else if (msg->pagelist) { |
828 | page = list_first_entry(&msg->pagelist->head, | |
829 | struct page, lru); | |
830 | if (crc) | |
831 | kaddr = kmap(page); | |
68b4476b YS |
832 | #ifdef CONFIG_BLOCK |
833 | } else if (msg->bio) { | |
834 | struct bio_vec *bv; | |
835 | ||
836 | bv = bio_iovec_idx(msg->bio_iter, msg->bio_seg); | |
837 | page = bv->bv_page; | |
838 | page_shift = bv->bv_offset; | |
839 | if (crc) | |
840 | kaddr = kmap(page) + page_shift; | |
841 | max_write = bv->bv_len; | |
842 | #endif | |
31b8006e SW |
843 | } else { |
844 | page = con->msgr->zero_page; | |
845 | if (crc) | |
846 | kaddr = page_address(con->msgr->zero_page); | |
847 | } | |
68b4476b YS |
848 | len = min_t(int, max_write - con->out_msg_pos.page_pos, |
849 | total_max_write); | |
850 | ||
31b8006e SW |
851 | if (crc && !con->out_msg_pos.did_page_crc) { |
852 | void *base = kaddr + con->out_msg_pos.page_pos; | |
853 | u32 tmpcrc = le32_to_cpu(con->out_msg->footer.data_crc); | |
854 | ||
855 | BUG_ON(kaddr == NULL); | |
856 | con->out_msg->footer.data_crc = | |
857 | cpu_to_le32(crc32c(tmpcrc, base, len)); | |
858 | con->out_msg_pos.did_page_crc = 1; | |
859 | } | |
31b8006e | 860 | ret = kernel_sendpage(con->sock, page, |
68b4476b YS |
861 | con->out_msg_pos.page_pos + page_shift, |
862 | len, | |
31b8006e SW |
863 | MSG_DONTWAIT | MSG_NOSIGNAL | |
864 | MSG_MORE); | |
865 | ||
68b4476b YS |
866 | if (crc && |
867 | (msg->pages || msg->pagelist || msg->bio || in_trail)) | |
31b8006e SW |
868 | kunmap(page); |
869 | ||
42961d23 SW |
870 | if (ret == -EAGAIN) |
871 | ret = 0; | |
31b8006e SW |
872 | if (ret <= 0) |
873 | goto out; | |
874 | ||
875 | con->out_msg_pos.data_pos += ret; | |
876 | con->out_msg_pos.page_pos += ret; | |
877 | if (ret == len) { | |
878 | con->out_msg_pos.page_pos = 0; | |
879 | con->out_msg_pos.page++; | |
880 | con->out_msg_pos.did_page_crc = 0; | |
68b4476b YS |
881 | if (in_trail) |
882 | list_move_tail(&page->lru, | |
883 | &msg->trail->head); | |
884 | else if (msg->pagelist) | |
58bb3b37 SW |
885 | list_move_tail(&page->lru, |
886 | &msg->pagelist->head); | |
68b4476b YS |
887 | #ifdef CONFIG_BLOCK |
888 | else if (msg->bio) | |
889 | iter_bio_next(&msg->bio_iter, &msg->bio_seg); | |
890 | #endif | |
31b8006e SW |
891 | } |
892 | } | |
893 | ||
894 | dout("write_partial_msg_pages %p msg %p done\n", con, msg); | |
895 | ||
896 | /* prepare and queue up footer, too */ | |
897 | if (!crc) | |
898 | con->out_msg->footer.flags |= CEPH_MSG_FOOTER_NOCRC; | |
899 | con->out_kvec_bytes = 0; | |
900 | con->out_kvec_left = 0; | |
901 | con->out_kvec_cur = con->out_kvec; | |
902 | prepare_write_message_footer(con, 0); | |
903 | ret = 1; | |
904 | out: | |
905 | return ret; | |
906 | } | |
907 | ||
908 | /* | |
909 | * write some zeros | |
910 | */ | |
911 | static int write_partial_skip(struct ceph_connection *con) | |
912 | { | |
913 | int ret; | |
914 | ||
915 | while (con->out_skip > 0) { | |
916 | struct kvec iov = { | |
917 | .iov_base = page_address(con->msgr->zero_page), | |
918 | .iov_len = min(con->out_skip, (int)PAGE_CACHE_SIZE) | |
919 | }; | |
920 | ||
921 | ret = ceph_tcp_sendmsg(con->sock, &iov, 1, iov.iov_len, 1); | |
922 | if (ret <= 0) | |
923 | goto out; | |
924 | con->out_skip -= ret; | |
925 | } | |
926 | ret = 1; | |
927 | out: | |
928 | return ret; | |
929 | } | |
930 | ||
931 | /* | |
932 | * Prepare to read connection handshake, or an ack. | |
933 | */ | |
eed0ef2c SW |
934 | static void prepare_read_banner(struct ceph_connection *con) |
935 | { | |
936 | dout("prepare_read_banner %p\n", con); | |
937 | con->in_base_pos = 0; | |
938 | } | |
939 | ||
31b8006e SW |
940 | static void prepare_read_connect(struct ceph_connection *con) |
941 | { | |
942 | dout("prepare_read_connect %p\n", con); | |
943 | con->in_base_pos = 0; | |
944 | } | |
945 | ||
946 | static void prepare_read_ack(struct ceph_connection *con) | |
947 | { | |
948 | dout("prepare_read_ack %p\n", con); | |
949 | con->in_base_pos = 0; | |
950 | } | |
951 | ||
952 | static void prepare_read_tag(struct ceph_connection *con) | |
953 | { | |
954 | dout("prepare_read_tag %p\n", con); | |
955 | con->in_base_pos = 0; | |
956 | con->in_tag = CEPH_MSGR_TAG_READY; | |
957 | } | |
958 | ||
959 | /* | |
960 | * Prepare to read a message. | |
961 | */ | |
962 | static int prepare_read_message(struct ceph_connection *con) | |
963 | { | |
964 | dout("prepare_read_message %p\n", con); | |
965 | BUG_ON(con->in_msg != NULL); | |
966 | con->in_base_pos = 0; | |
967 | con->in_front_crc = con->in_middle_crc = con->in_data_crc = 0; | |
968 | return 0; | |
969 | } | |
970 | ||
971 | ||
972 | static int read_partial(struct ceph_connection *con, | |
973 | int *to, int size, void *object) | |
974 | { | |
975 | *to += size; | |
976 | while (con->in_base_pos < *to) { | |
977 | int left = *to - con->in_base_pos; | |
978 | int have = size - left; | |
979 | int ret = ceph_tcp_recvmsg(con->sock, object + have, left); | |
980 | if (ret <= 0) | |
981 | return ret; | |
982 | con->in_base_pos += ret; | |
983 | } | |
984 | return 1; | |
985 | } | |
986 | ||
987 | ||
988 | /* | |
989 | * Read all or part of the connect-side handshake on a new connection | |
990 | */ | |
eed0ef2c | 991 | static int read_partial_banner(struct ceph_connection *con) |
31b8006e SW |
992 | { |
993 | int ret, to = 0; | |
994 | ||
eed0ef2c | 995 | dout("read_partial_banner %p at %d\n", con, con->in_base_pos); |
31b8006e SW |
996 | |
997 | /* peer's banner */ | |
998 | ret = read_partial(con, &to, strlen(CEPH_BANNER), con->in_banner); | |
999 | if (ret <= 0) | |
1000 | goto out; | |
1001 | ret = read_partial(con, &to, sizeof(con->actual_peer_addr), | |
1002 | &con->actual_peer_addr); | |
1003 | if (ret <= 0) | |
1004 | goto out; | |
1005 | ret = read_partial(con, &to, sizeof(con->peer_addr_for_me), | |
1006 | &con->peer_addr_for_me); | |
1007 | if (ret <= 0) | |
1008 | goto out; | |
eed0ef2c SW |
1009 | out: |
1010 | return ret; | |
1011 | } | |
1012 | ||
1013 | static int read_partial_connect(struct ceph_connection *con) | |
1014 | { | |
1015 | int ret, to = 0; | |
1016 | ||
1017 | dout("read_partial_connect %p at %d\n", con, con->in_base_pos); | |
1018 | ||
31b8006e SW |
1019 | ret = read_partial(con, &to, sizeof(con->in_reply), &con->in_reply); |
1020 | if (ret <= 0) | |
1021 | goto out; | |
4e7a5dcd SW |
1022 | ret = read_partial(con, &to, le32_to_cpu(con->in_reply.authorizer_len), |
1023 | con->auth_reply_buf); | |
1024 | if (ret <= 0) | |
1025 | goto out; | |
31b8006e | 1026 | |
4e7a5dcd SW |
1027 | dout("read_partial_connect %p tag %d, con_seq = %u, g_seq = %u\n", |
1028 | con, (int)con->in_reply.tag, | |
1029 | le32_to_cpu(con->in_reply.connect_seq), | |
31b8006e SW |
1030 | le32_to_cpu(con->in_reply.global_seq)); |
1031 | out: | |
1032 | return ret; | |
eed0ef2c | 1033 | |
31b8006e SW |
1034 | } |
1035 | ||
1036 | /* | |
1037 | * Verify the hello banner looks okay. | |
1038 | */ | |
1039 | static int verify_hello(struct ceph_connection *con) | |
1040 | { | |
1041 | if (memcmp(con->in_banner, CEPH_BANNER, strlen(CEPH_BANNER))) { | |
13e38c8a | 1042 | pr_err("connect to %s got bad banner\n", |
3d14c5d2 | 1043 | ceph_pr_addr(&con->peer_addr.in_addr)); |
31b8006e SW |
1044 | con->error_msg = "protocol error, bad banner"; |
1045 | return -1; | |
1046 | } | |
1047 | return 0; | |
1048 | } | |
1049 | ||
1050 | static bool addr_is_blank(struct sockaddr_storage *ss) | |
1051 | { | |
1052 | switch (ss->ss_family) { | |
1053 | case AF_INET: | |
1054 | return ((struct sockaddr_in *)ss)->sin_addr.s_addr == 0; | |
1055 | case AF_INET6: | |
1056 | return | |
1057 | ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[0] == 0 && | |
1058 | ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[1] == 0 && | |
1059 | ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[2] == 0 && | |
1060 | ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[3] == 0; | |
1061 | } | |
1062 | return false; | |
1063 | } | |
1064 | ||
1065 | static int addr_port(struct sockaddr_storage *ss) | |
1066 | { | |
1067 | switch (ss->ss_family) { | |
1068 | case AF_INET: | |
f28bcfbe | 1069 | return ntohs(((struct sockaddr_in *)ss)->sin_port); |
31b8006e | 1070 | case AF_INET6: |
f28bcfbe | 1071 | return ntohs(((struct sockaddr_in6 *)ss)->sin6_port); |
31b8006e SW |
1072 | } |
1073 | return 0; | |
1074 | } | |
1075 | ||
1076 | static void addr_set_port(struct sockaddr_storage *ss, int p) | |
1077 | { | |
1078 | switch (ss->ss_family) { | |
1079 | case AF_INET: | |
1080 | ((struct sockaddr_in *)ss)->sin_port = htons(p); | |
a2a79609 | 1081 | break; |
31b8006e SW |
1082 | case AF_INET6: |
1083 | ((struct sockaddr_in6 *)ss)->sin6_port = htons(p); | |
a2a79609 | 1084 | break; |
31b8006e SW |
1085 | } |
1086 | } | |
1087 | ||
ee3b56f2 NW |
1088 | /* |
1089 | * Unlike other *_pton function semantics, zero indicates success. | |
1090 | */ | |
1091 | static int ceph_pton(const char *str, size_t len, struct sockaddr_storage *ss, | |
1092 | char delim, const char **ipend) | |
1093 | { | |
1094 | struct sockaddr_in *in4 = (void *)ss; | |
1095 | struct sockaddr_in6 *in6 = (void *)ss; | |
1096 | ||
1097 | memset(ss, 0, sizeof(*ss)); | |
1098 | ||
1099 | if (in4_pton(str, len, (u8 *)&in4->sin_addr.s_addr, delim, ipend)) { | |
1100 | ss->ss_family = AF_INET; | |
1101 | return 0; | |
1102 | } | |
1103 | ||
1104 | if (in6_pton(str, len, (u8 *)&in6->sin6_addr.s6_addr, delim, ipend)) { | |
1105 | ss->ss_family = AF_INET6; | |
1106 | return 0; | |
1107 | } | |
1108 | ||
1109 | return -EINVAL; | |
1110 | } | |
1111 | ||
1112 | /* | |
1113 | * Extract hostname string and resolve using kernel DNS facility. | |
1114 | */ | |
1115 | #ifdef CONFIG_CEPH_LIB_USE_DNS_RESOLVER | |
1116 | static int ceph_dns_resolve_name(const char *name, size_t namelen, | |
1117 | struct sockaddr_storage *ss, char delim, const char **ipend) | |
1118 | { | |
1119 | const char *end, *delim_p; | |
1120 | char *colon_p, *ip_addr = NULL; | |
1121 | int ip_len, ret; | |
1122 | ||
1123 | /* | |
1124 | * The end of the hostname occurs immediately preceding the delimiter or | |
1125 | * the port marker (':') where the delimiter takes precedence. | |
1126 | */ | |
1127 | delim_p = memchr(name, delim, namelen); | |
1128 | colon_p = memchr(name, ':', namelen); | |
1129 | ||
1130 | if (delim_p && colon_p) | |
1131 | end = delim_p < colon_p ? delim_p : colon_p; | |
1132 | else if (!delim_p && colon_p) | |
1133 | end = colon_p; | |
1134 | else { | |
1135 | end = delim_p; | |
1136 | if (!end) /* case: hostname:/ */ | |
1137 | end = name + namelen; | |
1138 | } | |
1139 | ||
1140 | if (end <= name) | |
1141 | return -EINVAL; | |
1142 | ||
1143 | /* do dns_resolve upcall */ | |
1144 | ip_len = dns_query(NULL, name, end - name, NULL, &ip_addr, NULL); | |
1145 | if (ip_len > 0) | |
1146 | ret = ceph_pton(ip_addr, ip_len, ss, -1, NULL); | |
1147 | else | |
1148 | ret = -ESRCH; | |
1149 | ||
1150 | kfree(ip_addr); | |
1151 | ||
1152 | *ipend = end; | |
1153 | ||
1154 | pr_info("resolve '%.*s' (ret=%d): %s\n", (int)(end - name), name, | |
1155 | ret, ret ? "failed" : ceph_pr_addr(ss)); | |
1156 | ||
1157 | return ret; | |
1158 | } | |
1159 | #else | |
1160 | static inline int ceph_dns_resolve_name(const char *name, size_t namelen, | |
1161 | struct sockaddr_storage *ss, char delim, const char **ipend) | |
1162 | { | |
1163 | return -EINVAL; | |
1164 | } | |
1165 | #endif | |
1166 | ||
1167 | /* | |
1168 | * Parse a server name (IP or hostname). If a valid IP address is not found | |
1169 | * then try to extract a hostname to resolve using userspace DNS upcall. | |
1170 | */ | |
1171 | static int ceph_parse_server_name(const char *name, size_t namelen, | |
1172 | struct sockaddr_storage *ss, char delim, const char **ipend) | |
1173 | { | |
1174 | int ret; | |
1175 | ||
1176 | ret = ceph_pton(name, namelen, ss, delim, ipend); | |
1177 | if (ret) | |
1178 | ret = ceph_dns_resolve_name(name, namelen, ss, delim, ipend); | |
1179 | ||
1180 | return ret; | |
1181 | } | |
1182 | ||
31b8006e SW |
1183 | /* |
1184 | * Parse an ip[:port] list into an addr array. Use the default | |
1185 | * monitor port if a port isn't specified. | |
1186 | */ | |
1187 | int ceph_parse_ips(const char *c, const char *end, | |
1188 | struct ceph_entity_addr *addr, | |
1189 | int max_count, int *count) | |
1190 | { | |
ee3b56f2 | 1191 | int i, ret = -EINVAL; |
31b8006e SW |
1192 | const char *p = c; |
1193 | ||
1194 | dout("parse_ips on '%.*s'\n", (int)(end-c), c); | |
1195 | for (i = 0; i < max_count; i++) { | |
1196 | const char *ipend; | |
1197 | struct sockaddr_storage *ss = &addr[i].in_addr; | |
31b8006e | 1198 | int port; |
39139f64 SW |
1199 | char delim = ','; |
1200 | ||
1201 | if (*p == '[') { | |
1202 | delim = ']'; | |
1203 | p++; | |
1204 | } | |
31b8006e | 1205 | |
ee3b56f2 NW |
1206 | ret = ceph_parse_server_name(p, end - p, ss, delim, &ipend); |
1207 | if (ret) | |
31b8006e | 1208 | goto bad; |
ee3b56f2 NW |
1209 | ret = -EINVAL; |
1210 | ||
31b8006e SW |
1211 | p = ipend; |
1212 | ||
39139f64 SW |
1213 | if (delim == ']') { |
1214 | if (*p != ']') { | |
1215 | dout("missing matching ']'\n"); | |
1216 | goto bad; | |
1217 | } | |
1218 | p++; | |
1219 | } | |
1220 | ||
31b8006e SW |
1221 | /* port? */ |
1222 | if (p < end && *p == ':') { | |
1223 | port = 0; | |
1224 | p++; | |
1225 | while (p < end && *p >= '0' && *p <= '9') { | |
1226 | port = (port * 10) + (*p - '0'); | |
1227 | p++; | |
1228 | } | |
1229 | if (port > 65535 || port == 0) | |
1230 | goto bad; | |
1231 | } else { | |
1232 | port = CEPH_MON_PORT; | |
1233 | } | |
1234 | ||
1235 | addr_set_port(ss, port); | |
1236 | ||
3d14c5d2 | 1237 | dout("parse_ips got %s\n", ceph_pr_addr(ss)); |
31b8006e SW |
1238 | |
1239 | if (p == end) | |
1240 | break; | |
1241 | if (*p != ',') | |
1242 | goto bad; | |
1243 | p++; | |
1244 | } | |
1245 | ||
1246 | if (p != end) | |
1247 | goto bad; | |
1248 | ||
1249 | if (count) | |
1250 | *count = i + 1; | |
1251 | return 0; | |
1252 | ||
1253 | bad: | |
39139f64 | 1254 | pr_err("parse_ips bad ip '%.*s'\n", (int)(end - c), c); |
ee3b56f2 | 1255 | return ret; |
31b8006e | 1256 | } |
3d14c5d2 | 1257 | EXPORT_SYMBOL(ceph_parse_ips); |
31b8006e | 1258 | |
eed0ef2c | 1259 | static int process_banner(struct ceph_connection *con) |
31b8006e | 1260 | { |
eed0ef2c | 1261 | dout("process_banner on %p\n", con); |
31b8006e SW |
1262 | |
1263 | if (verify_hello(con) < 0) | |
1264 | return -1; | |
1265 | ||
63f2d211 SW |
1266 | ceph_decode_addr(&con->actual_peer_addr); |
1267 | ceph_decode_addr(&con->peer_addr_for_me); | |
1268 | ||
31b8006e SW |
1269 | /* |
1270 | * Make sure the other end is who we wanted. note that the other | |
1271 | * end may not yet know their ip address, so if it's 0.0.0.0, give | |
1272 | * them the benefit of the doubt. | |
1273 | */ | |
103e2d3a SW |
1274 | if (memcmp(&con->peer_addr, &con->actual_peer_addr, |
1275 | sizeof(con->peer_addr)) != 0 && | |
31b8006e SW |
1276 | !(addr_is_blank(&con->actual_peer_addr.in_addr) && |
1277 | con->actual_peer_addr.nonce == con->peer_addr.nonce)) { | |
cd84db6e | 1278 | pr_warning("wrong peer, want %s/%d, got %s/%d\n", |
3d14c5d2 | 1279 | ceph_pr_addr(&con->peer_addr.in_addr), |
cd84db6e | 1280 | (int)le32_to_cpu(con->peer_addr.nonce), |
3d14c5d2 | 1281 | ceph_pr_addr(&con->actual_peer_addr.in_addr), |
cd84db6e | 1282 | (int)le32_to_cpu(con->actual_peer_addr.nonce)); |
58bb3b37 | 1283 | con->error_msg = "wrong peer at address"; |
31b8006e SW |
1284 | return -1; |
1285 | } | |
1286 | ||
1287 | /* | |
1288 | * did we learn our address? | |
1289 | */ | |
1290 | if (addr_is_blank(&con->msgr->inst.addr.in_addr)) { | |
1291 | int port = addr_port(&con->msgr->inst.addr.in_addr); | |
1292 | ||
1293 | memcpy(&con->msgr->inst.addr.in_addr, | |
1294 | &con->peer_addr_for_me.in_addr, | |
1295 | sizeof(con->peer_addr_for_me.in_addr)); | |
1296 | addr_set_port(&con->msgr->inst.addr.in_addr, port); | |
63f2d211 | 1297 | encode_my_addr(con->msgr); |
eed0ef2c | 1298 | dout("process_banner learned my addr is %s\n", |
3d14c5d2 | 1299 | ceph_pr_addr(&con->msgr->inst.addr.in_addr)); |
31b8006e SW |
1300 | } |
1301 | ||
eed0ef2c SW |
1302 | set_bit(NEGOTIATING, &con->state); |
1303 | prepare_read_connect(con); | |
1304 | return 0; | |
1305 | } | |
1306 | ||
04a419f9 SW |
1307 | static void fail_protocol(struct ceph_connection *con) |
1308 | { | |
1309 | reset_connection(con); | |
1310 | set_bit(CLOSED, &con->state); /* in case there's queued work */ | |
1311 | ||
1312 | mutex_unlock(&con->mutex); | |
1313 | if (con->ops->bad_proto) | |
1314 | con->ops->bad_proto(con); | |
1315 | mutex_lock(&con->mutex); | |
1316 | } | |
1317 | ||
eed0ef2c SW |
1318 | static int process_connect(struct ceph_connection *con) |
1319 | { | |
3d14c5d2 YS |
1320 | u64 sup_feat = con->msgr->supported_features; |
1321 | u64 req_feat = con->msgr->required_features; | |
04a419f9 | 1322 | u64 server_feat = le64_to_cpu(con->in_reply.features); |
0da5d703 | 1323 | int ret; |
04a419f9 | 1324 | |
eed0ef2c SW |
1325 | dout("process_connect on %p tag %d\n", con, (int)con->in_tag); |
1326 | ||
31b8006e | 1327 | switch (con->in_reply.tag) { |
04a419f9 SW |
1328 | case CEPH_MSGR_TAG_FEATURES: |
1329 | pr_err("%s%lld %s feature set mismatch," | |
1330 | " my %llx < server's %llx, missing %llx\n", | |
1331 | ENTITY_NAME(con->peer_name), | |
3d14c5d2 | 1332 | ceph_pr_addr(&con->peer_addr.in_addr), |
04a419f9 SW |
1333 | sup_feat, server_feat, server_feat & ~sup_feat); |
1334 | con->error_msg = "missing required protocol features"; | |
1335 | fail_protocol(con); | |
1336 | return -1; | |
1337 | ||
31b8006e | 1338 | case CEPH_MSGR_TAG_BADPROTOVER: |
31b8006e SW |
1339 | pr_err("%s%lld %s protocol version mismatch," |
1340 | " my %d != server's %d\n", | |
1341 | ENTITY_NAME(con->peer_name), | |
3d14c5d2 | 1342 | ceph_pr_addr(&con->peer_addr.in_addr), |
31b8006e SW |
1343 | le32_to_cpu(con->out_connect.protocol_version), |
1344 | le32_to_cpu(con->in_reply.protocol_version)); | |
1345 | con->error_msg = "protocol version mismatch"; | |
04a419f9 | 1346 | fail_protocol(con); |
31b8006e SW |
1347 | return -1; |
1348 | ||
4e7a5dcd SW |
1349 | case CEPH_MSGR_TAG_BADAUTHORIZER: |
1350 | con->auth_retry++; | |
1351 | dout("process_connect %p got BADAUTHORIZER attempt %d\n", con, | |
1352 | con->auth_retry); | |
1353 | if (con->auth_retry == 2) { | |
1354 | con->error_msg = "connect authorization failure"; | |
4e7a5dcd SW |
1355 | return -1; |
1356 | } | |
1357 | con->auth_retry = 1; | |
0da5d703 SW |
1358 | ret = prepare_write_connect(con->msgr, con, 0); |
1359 | if (ret < 0) | |
1360 | return ret; | |
63733a0f | 1361 | prepare_read_connect(con); |
4e7a5dcd | 1362 | break; |
31b8006e SW |
1363 | |
1364 | case CEPH_MSGR_TAG_RESETSESSION: | |
1365 | /* | |
1366 | * If we connected with a large connect_seq but the peer | |
1367 | * has no record of a session with us (no connection, or | |
1368 | * connect_seq == 0), they will send RESETSESION to indicate | |
1369 | * that they must have reset their session, and may have | |
1370 | * dropped messages. | |
1371 | */ | |
1372 | dout("process_connect got RESET peer seq %u\n", | |
1373 | le32_to_cpu(con->in_connect.connect_seq)); | |
1374 | pr_err("%s%lld %s connection reset\n", | |
1375 | ENTITY_NAME(con->peer_name), | |
3d14c5d2 | 1376 | ceph_pr_addr(&con->peer_addr.in_addr)); |
31b8006e | 1377 | reset_connection(con); |
eed0ef2c | 1378 | prepare_write_connect(con->msgr, con, 0); |
31b8006e SW |
1379 | prepare_read_connect(con); |
1380 | ||
1381 | /* Tell ceph about it. */ | |
ec302645 | 1382 | mutex_unlock(&con->mutex); |
31b8006e SW |
1383 | pr_info("reset on %s%lld\n", ENTITY_NAME(con->peer_name)); |
1384 | if (con->ops->peer_reset) | |
1385 | con->ops->peer_reset(con); | |
ec302645 | 1386 | mutex_lock(&con->mutex); |
0da5d703 SW |
1387 | if (test_bit(CLOSED, &con->state) || |
1388 | test_bit(OPENING, &con->state)) | |
1389 | return -EAGAIN; | |
31b8006e SW |
1390 | break; |
1391 | ||
1392 | case CEPH_MSGR_TAG_RETRY_SESSION: | |
1393 | /* | |
1394 | * If we sent a smaller connect_seq than the peer has, try | |
1395 | * again with a larger value. | |
1396 | */ | |
1397 | dout("process_connect got RETRY my seq = %u, peer_seq = %u\n", | |
1398 | le32_to_cpu(con->out_connect.connect_seq), | |
1399 | le32_to_cpu(con->in_connect.connect_seq)); | |
1400 | con->connect_seq = le32_to_cpu(con->in_connect.connect_seq); | |
eed0ef2c | 1401 | prepare_write_connect(con->msgr, con, 0); |
31b8006e SW |
1402 | prepare_read_connect(con); |
1403 | break; | |
1404 | ||
1405 | case CEPH_MSGR_TAG_RETRY_GLOBAL: | |
1406 | /* | |
1407 | * If we sent a smaller global_seq than the peer has, try | |
1408 | * again with a larger value. | |
1409 | */ | |
eed0ef2c | 1410 | dout("process_connect got RETRY_GLOBAL my %u peer_gseq %u\n", |
31b8006e SW |
1411 | con->peer_global_seq, |
1412 | le32_to_cpu(con->in_connect.global_seq)); | |
1413 | get_global_seq(con->msgr, | |
1414 | le32_to_cpu(con->in_connect.global_seq)); | |
eed0ef2c | 1415 | prepare_write_connect(con->msgr, con, 0); |
31b8006e SW |
1416 | prepare_read_connect(con); |
1417 | break; | |
1418 | ||
1419 | case CEPH_MSGR_TAG_READY: | |
04a419f9 SW |
1420 | if (req_feat & ~server_feat) { |
1421 | pr_err("%s%lld %s protocol feature mismatch," | |
1422 | " my required %llx > server's %llx, need %llx\n", | |
1423 | ENTITY_NAME(con->peer_name), | |
3d14c5d2 | 1424 | ceph_pr_addr(&con->peer_addr.in_addr), |
04a419f9 SW |
1425 | req_feat, server_feat, req_feat & ~server_feat); |
1426 | con->error_msg = "missing required protocol features"; | |
1427 | fail_protocol(con); | |
1428 | return -1; | |
1429 | } | |
31b8006e | 1430 | clear_bit(CONNECTING, &con->state); |
31b8006e SW |
1431 | con->peer_global_seq = le32_to_cpu(con->in_reply.global_seq); |
1432 | con->connect_seq++; | |
aba558e2 | 1433 | con->peer_features = server_feat; |
31b8006e SW |
1434 | dout("process_connect got READY gseq %d cseq %d (%d)\n", |
1435 | con->peer_global_seq, | |
1436 | le32_to_cpu(con->in_reply.connect_seq), | |
1437 | con->connect_seq); | |
1438 | WARN_ON(con->connect_seq != | |
1439 | le32_to_cpu(con->in_reply.connect_seq)); | |
92ac41d0 SW |
1440 | |
1441 | if (con->in_reply.flags & CEPH_MSG_CONNECT_LOSSY) | |
1442 | set_bit(LOSSYTX, &con->state); | |
1443 | ||
31b8006e SW |
1444 | prepare_read_tag(con); |
1445 | break; | |
1446 | ||
1447 | case CEPH_MSGR_TAG_WAIT: | |
1448 | /* | |
1449 | * If there is a connection race (we are opening | |
1450 | * connections to each other), one of us may just have | |
1451 | * to WAIT. This shouldn't happen if we are the | |
1452 | * client. | |
1453 | */ | |
04177882 SW |
1454 | pr_err("process_connect got WAIT as client\n"); |
1455 | con->error_msg = "protocol error, got WAIT as client"; | |
1456 | return -1; | |
31b8006e SW |
1457 | |
1458 | default: | |
1459 | pr_err("connect protocol error, will retry\n"); | |
1460 | con->error_msg = "protocol error, garbage tag during connect"; | |
1461 | return -1; | |
1462 | } | |
1463 | return 0; | |
1464 | } | |
1465 | ||
1466 | ||
1467 | /* | |
1468 | * read (part of) an ack | |
1469 | */ | |
1470 | static int read_partial_ack(struct ceph_connection *con) | |
1471 | { | |
1472 | int to = 0; | |
1473 | ||
1474 | return read_partial(con, &to, sizeof(con->in_temp_ack), | |
1475 | &con->in_temp_ack); | |
1476 | } | |
1477 | ||
1478 | ||
1479 | /* | |
1480 | * We can finally discard anything that's been acked. | |
1481 | */ | |
1482 | static void process_ack(struct ceph_connection *con) | |
1483 | { | |
1484 | struct ceph_msg *m; | |
1485 | u64 ack = le64_to_cpu(con->in_temp_ack); | |
1486 | u64 seq; | |
1487 | ||
31b8006e SW |
1488 | while (!list_empty(&con->out_sent)) { |
1489 | m = list_first_entry(&con->out_sent, struct ceph_msg, | |
1490 | list_head); | |
1491 | seq = le64_to_cpu(m->hdr.seq); | |
1492 | if (seq > ack) | |
1493 | break; | |
1494 | dout("got ack for seq %llu type %d at %p\n", seq, | |
1495 | le16_to_cpu(m->hdr.type), m); | |
4cf9d544 | 1496 | m->ack_stamp = jiffies; |
31b8006e SW |
1497 | ceph_msg_remove(m); |
1498 | } | |
31b8006e SW |
1499 | prepare_read_tag(con); |
1500 | } | |
1501 | ||
1502 | ||
1503 | ||
1504 | ||
2450418c | 1505 | static int read_partial_message_section(struct ceph_connection *con, |
213c99ee SW |
1506 | struct kvec *section, |
1507 | unsigned int sec_len, u32 *crc) | |
2450418c | 1508 | { |
68b4476b | 1509 | int ret, left; |
2450418c YS |
1510 | |
1511 | BUG_ON(!section); | |
1512 | ||
1513 | while (section->iov_len < sec_len) { | |
1514 | BUG_ON(section->iov_base == NULL); | |
1515 | left = sec_len - section->iov_len; | |
1516 | ret = ceph_tcp_recvmsg(con->sock, (char *)section->iov_base + | |
1517 | section->iov_len, left); | |
1518 | if (ret <= 0) | |
1519 | return ret; | |
1520 | section->iov_len += ret; | |
1521 | if (section->iov_len == sec_len) | |
1522 | *crc = crc32c(0, section->iov_base, | |
1523 | section->iov_len); | |
1524 | } | |
31b8006e | 1525 | |
2450418c YS |
1526 | return 1; |
1527 | } | |
31b8006e | 1528 | |
2450418c YS |
1529 | static struct ceph_msg *ceph_alloc_msg(struct ceph_connection *con, |
1530 | struct ceph_msg_header *hdr, | |
1531 | int *skip); | |
68b4476b YS |
1532 | |
1533 | ||
1534 | static int read_partial_message_pages(struct ceph_connection *con, | |
1535 | struct page **pages, | |
1536 | unsigned data_len, int datacrc) | |
1537 | { | |
1538 | void *p; | |
1539 | int ret; | |
1540 | int left; | |
1541 | ||
1542 | left = min((int)(data_len - con->in_msg_pos.data_pos), | |
1543 | (int)(PAGE_SIZE - con->in_msg_pos.page_pos)); | |
1544 | /* (page) data */ | |
1545 | BUG_ON(pages == NULL); | |
1546 | p = kmap(pages[con->in_msg_pos.page]); | |
1547 | ret = ceph_tcp_recvmsg(con->sock, p + con->in_msg_pos.page_pos, | |
1548 | left); | |
1549 | if (ret > 0 && datacrc) | |
1550 | con->in_data_crc = | |
1551 | crc32c(con->in_data_crc, | |
1552 | p + con->in_msg_pos.page_pos, ret); | |
1553 | kunmap(pages[con->in_msg_pos.page]); | |
1554 | if (ret <= 0) | |
1555 | return ret; | |
1556 | con->in_msg_pos.data_pos += ret; | |
1557 | con->in_msg_pos.page_pos += ret; | |
1558 | if (con->in_msg_pos.page_pos == PAGE_SIZE) { | |
1559 | con->in_msg_pos.page_pos = 0; | |
1560 | con->in_msg_pos.page++; | |
1561 | } | |
1562 | ||
1563 | return ret; | |
1564 | } | |
1565 | ||
1566 | #ifdef CONFIG_BLOCK | |
1567 | static int read_partial_message_bio(struct ceph_connection *con, | |
1568 | struct bio **bio_iter, int *bio_seg, | |
1569 | unsigned data_len, int datacrc) | |
1570 | { | |
1571 | struct bio_vec *bv = bio_iovec_idx(*bio_iter, *bio_seg); | |
1572 | void *p; | |
1573 | int ret, left; | |
1574 | ||
1575 | if (IS_ERR(bv)) | |
1576 | return PTR_ERR(bv); | |
1577 | ||
1578 | left = min((int)(data_len - con->in_msg_pos.data_pos), | |
1579 | (int)(bv->bv_len - con->in_msg_pos.page_pos)); | |
1580 | ||
1581 | p = kmap(bv->bv_page) + bv->bv_offset; | |
1582 | ||
1583 | ret = ceph_tcp_recvmsg(con->sock, p + con->in_msg_pos.page_pos, | |
1584 | left); | |
1585 | if (ret > 0 && datacrc) | |
1586 | con->in_data_crc = | |
1587 | crc32c(con->in_data_crc, | |
1588 | p + con->in_msg_pos.page_pos, ret); | |
1589 | kunmap(bv->bv_page); | |
1590 | if (ret <= 0) | |
1591 | return ret; | |
1592 | con->in_msg_pos.data_pos += ret; | |
1593 | con->in_msg_pos.page_pos += ret; | |
1594 | if (con->in_msg_pos.page_pos == bv->bv_len) { | |
1595 | con->in_msg_pos.page_pos = 0; | |
1596 | iter_bio_next(bio_iter, bio_seg); | |
1597 | } | |
1598 | ||
1599 | return ret; | |
1600 | } | |
1601 | #endif | |
1602 | ||
31b8006e SW |
1603 | /* |
1604 | * read (part of) a message. | |
1605 | */ | |
1606 | static int read_partial_message(struct ceph_connection *con) | |
1607 | { | |
1608 | struct ceph_msg *m = con->in_msg; | |
31b8006e | 1609 | int ret; |
9d7f0f13 | 1610 | int to, left; |
c5c6b19d | 1611 | unsigned front_len, middle_len, data_len; |
31b8006e | 1612 | int datacrc = con->msgr->nocrc; |
2450418c | 1613 | int skip; |
ae18756b | 1614 | u64 seq; |
31b8006e SW |
1615 | |
1616 | dout("read_partial_message con %p msg %p\n", con, m); | |
1617 | ||
1618 | /* header */ | |
1619 | while (con->in_base_pos < sizeof(con->in_hdr)) { | |
1620 | left = sizeof(con->in_hdr) - con->in_base_pos; | |
1621 | ret = ceph_tcp_recvmsg(con->sock, | |
1622 | (char *)&con->in_hdr + con->in_base_pos, | |
1623 | left); | |
1624 | if (ret <= 0) | |
1625 | return ret; | |
1626 | con->in_base_pos += ret; | |
1627 | if (con->in_base_pos == sizeof(con->in_hdr)) { | |
1628 | u32 crc = crc32c(0, (void *)&con->in_hdr, | |
1629 | sizeof(con->in_hdr) - sizeof(con->in_hdr.crc)); | |
1630 | if (crc != le32_to_cpu(con->in_hdr.crc)) { | |
1631 | pr_err("read_partial_message bad hdr " | |
1632 | " crc %u != expected %u\n", | |
1633 | crc, con->in_hdr.crc); | |
1634 | return -EBADMSG; | |
1635 | } | |
1636 | } | |
1637 | } | |
31b8006e SW |
1638 | front_len = le32_to_cpu(con->in_hdr.front_len); |
1639 | if (front_len > CEPH_MSG_MAX_FRONT_LEN) | |
1640 | return -EIO; | |
1641 | middle_len = le32_to_cpu(con->in_hdr.middle_len); | |
1642 | if (middle_len > CEPH_MSG_MAX_DATA_LEN) | |
1643 | return -EIO; | |
1644 | data_len = le32_to_cpu(con->in_hdr.data_len); | |
1645 | if (data_len > CEPH_MSG_MAX_DATA_LEN) | |
1646 | return -EIO; | |
1647 | ||
ae18756b SW |
1648 | /* verify seq# */ |
1649 | seq = le64_to_cpu(con->in_hdr.seq); | |
1650 | if ((s64)seq - (s64)con->in_seq < 1) { | |
df9f86fa | 1651 | pr_info("skipping %s%lld %s seq %lld expected %lld\n", |
ae18756b | 1652 | ENTITY_NAME(con->peer_name), |
3d14c5d2 | 1653 | ceph_pr_addr(&con->peer_addr.in_addr), |
ae18756b SW |
1654 | seq, con->in_seq + 1); |
1655 | con->in_base_pos = -front_len - middle_len - data_len - | |
1656 | sizeof(m->footer); | |
1657 | con->in_tag = CEPH_MSGR_TAG_READY; | |
ae18756b SW |
1658 | return 0; |
1659 | } else if ((s64)seq - (s64)con->in_seq > 1) { | |
1660 | pr_err("read_partial_message bad seq %lld expected %lld\n", | |
1661 | seq, con->in_seq + 1); | |
1662 | con->error_msg = "bad message sequence # for incoming message"; | |
1663 | return -EBADMSG; | |
1664 | } | |
1665 | ||
31b8006e SW |
1666 | /* allocate message? */ |
1667 | if (!con->in_msg) { | |
1668 | dout("got hdr type %d front %d data %d\n", con->in_hdr.type, | |
1669 | con->in_hdr.front_len, con->in_hdr.data_len); | |
ae32be31 | 1670 | skip = 0; |
2450418c YS |
1671 | con->in_msg = ceph_alloc_msg(con, &con->in_hdr, &skip); |
1672 | if (skip) { | |
31b8006e | 1673 | /* skip this message */ |
a79832f2 | 1674 | dout("alloc_msg said skip message\n"); |
ae32be31 | 1675 | BUG_ON(con->in_msg); |
31b8006e SW |
1676 | con->in_base_pos = -front_len - middle_len - data_len - |
1677 | sizeof(m->footer); | |
1678 | con->in_tag = CEPH_MSGR_TAG_READY; | |
684be25c | 1679 | con->in_seq++; |
31b8006e SW |
1680 | return 0; |
1681 | } | |
a79832f2 | 1682 | if (!con->in_msg) { |
5b3a4db3 SW |
1683 | con->error_msg = |
1684 | "error allocating memory for incoming message"; | |
a79832f2 | 1685 | return -ENOMEM; |
31b8006e SW |
1686 | } |
1687 | m = con->in_msg; | |
1688 | m->front.iov_len = 0; /* haven't read it yet */ | |
2450418c YS |
1689 | if (m->middle) |
1690 | m->middle->vec.iov_len = 0; | |
9d7f0f13 YS |
1691 | |
1692 | con->in_msg_pos.page = 0; | |
68b4476b | 1693 | if (m->pages) |
c5c6b19d | 1694 | con->in_msg_pos.page_pos = m->page_alignment; |
68b4476b YS |
1695 | else |
1696 | con->in_msg_pos.page_pos = 0; | |
9d7f0f13 | 1697 | con->in_msg_pos.data_pos = 0; |
31b8006e SW |
1698 | } |
1699 | ||
1700 | /* front */ | |
2450418c YS |
1701 | ret = read_partial_message_section(con, &m->front, front_len, |
1702 | &con->in_front_crc); | |
1703 | if (ret <= 0) | |
1704 | return ret; | |
31b8006e SW |
1705 | |
1706 | /* middle */ | |
2450418c | 1707 | if (m->middle) { |
213c99ee SW |
1708 | ret = read_partial_message_section(con, &m->middle->vec, |
1709 | middle_len, | |
2450418c | 1710 | &con->in_middle_crc); |
31b8006e SW |
1711 | if (ret <= 0) |
1712 | return ret; | |
31b8006e | 1713 | } |
68b4476b YS |
1714 | #ifdef CONFIG_BLOCK |
1715 | if (m->bio && !m->bio_iter) | |
1716 | init_bio_iter(m->bio, &m->bio_iter, &m->bio_seg); | |
1717 | #endif | |
31b8006e SW |
1718 | |
1719 | /* (page) data */ | |
31b8006e | 1720 | while (con->in_msg_pos.data_pos < data_len) { |
68b4476b YS |
1721 | if (m->pages) { |
1722 | ret = read_partial_message_pages(con, m->pages, | |
1723 | data_len, datacrc); | |
1724 | if (ret <= 0) | |
1725 | return ret; | |
1726 | #ifdef CONFIG_BLOCK | |
1727 | } else if (m->bio) { | |
1728 | ||
1729 | ret = read_partial_message_bio(con, | |
1730 | &m->bio_iter, &m->bio_seg, | |
1731 | data_len, datacrc); | |
1732 | if (ret <= 0) | |
1733 | return ret; | |
1734 | #endif | |
1735 | } else { | |
1736 | BUG_ON(1); | |
31b8006e SW |
1737 | } |
1738 | } | |
1739 | ||
31b8006e SW |
1740 | /* footer */ |
1741 | to = sizeof(m->hdr) + sizeof(m->footer); | |
1742 | while (con->in_base_pos < to) { | |
1743 | left = to - con->in_base_pos; | |
1744 | ret = ceph_tcp_recvmsg(con->sock, (char *)&m->footer + | |
1745 | (con->in_base_pos - sizeof(m->hdr)), | |
1746 | left); | |
1747 | if (ret <= 0) | |
1748 | return ret; | |
1749 | con->in_base_pos += ret; | |
1750 | } | |
1751 | dout("read_partial_message got msg %p %d (%u) + %d (%u) + %d (%u)\n", | |
1752 | m, front_len, m->footer.front_crc, middle_len, | |
1753 | m->footer.middle_crc, data_len, m->footer.data_crc); | |
1754 | ||
1755 | /* crc ok? */ | |
1756 | if (con->in_front_crc != le32_to_cpu(m->footer.front_crc)) { | |
1757 | pr_err("read_partial_message %p front crc %u != exp. %u\n", | |
1758 | m, con->in_front_crc, m->footer.front_crc); | |
1759 | return -EBADMSG; | |
1760 | } | |
1761 | if (con->in_middle_crc != le32_to_cpu(m->footer.middle_crc)) { | |
1762 | pr_err("read_partial_message %p middle crc %u != exp %u\n", | |
1763 | m, con->in_middle_crc, m->footer.middle_crc); | |
1764 | return -EBADMSG; | |
1765 | } | |
1766 | if (datacrc && | |
1767 | (m->footer.flags & CEPH_MSG_FOOTER_NOCRC) == 0 && | |
1768 | con->in_data_crc != le32_to_cpu(m->footer.data_crc)) { | |
1769 | pr_err("read_partial_message %p data crc %u != exp. %u\n", m, | |
1770 | con->in_data_crc, le32_to_cpu(m->footer.data_crc)); | |
1771 | return -EBADMSG; | |
1772 | } | |
1773 | ||
1774 | return 1; /* done! */ | |
1775 | } | |
1776 | ||
1777 | /* | |
1778 | * Process message. This happens in the worker thread. The callback should | |
1779 | * be careful not to do anything that waits on other incoming messages or it | |
1780 | * may deadlock. | |
1781 | */ | |
1782 | static void process_message(struct ceph_connection *con) | |
1783 | { | |
5e095e8b | 1784 | struct ceph_msg *msg; |
31b8006e | 1785 | |
5e095e8b | 1786 | msg = con->in_msg; |
31b8006e SW |
1787 | con->in_msg = NULL; |
1788 | ||
1789 | /* if first message, set peer_name */ | |
1790 | if (con->peer_name.type == 0) | |
dbad185d | 1791 | con->peer_name = msg->hdr.src; |
31b8006e | 1792 | |
31b8006e | 1793 | con->in_seq++; |
ec302645 | 1794 | mutex_unlock(&con->mutex); |
31b8006e SW |
1795 | |
1796 | dout("===== %p %llu from %s%lld %d=%s len %d+%d (%u %u %u) =====\n", | |
1797 | msg, le64_to_cpu(msg->hdr.seq), | |
dbad185d | 1798 | ENTITY_NAME(msg->hdr.src), |
31b8006e SW |
1799 | le16_to_cpu(msg->hdr.type), |
1800 | ceph_msg_type_name(le16_to_cpu(msg->hdr.type)), | |
1801 | le32_to_cpu(msg->hdr.front_len), | |
1802 | le32_to_cpu(msg->hdr.data_len), | |
1803 | con->in_front_crc, con->in_middle_crc, con->in_data_crc); | |
1804 | con->ops->dispatch(con, msg); | |
ec302645 SW |
1805 | |
1806 | mutex_lock(&con->mutex); | |
31b8006e SW |
1807 | prepare_read_tag(con); |
1808 | } | |
1809 | ||
1810 | ||
1811 | /* | |
1812 | * Write something to the socket. Called in a worker thread when the | |
1813 | * socket appears to be writeable and we have something ready to send. | |
1814 | */ | |
1815 | static int try_write(struct ceph_connection *con) | |
1816 | { | |
1817 | struct ceph_messenger *msgr = con->msgr; | |
1818 | int ret = 1; | |
1819 | ||
1820 | dout("try_write start %p state %lu nref %d\n", con, con->state, | |
1821 | atomic_read(&con->nref)); | |
1822 | ||
31b8006e SW |
1823 | more: |
1824 | dout("try_write out_kvec_bytes %d\n", con->out_kvec_bytes); | |
1825 | ||
1826 | /* open the socket first? */ | |
1827 | if (con->sock == NULL) { | |
eed0ef2c SW |
1828 | prepare_write_banner(msgr, con); |
1829 | prepare_write_connect(msgr, con, 1); | |
1830 | prepare_read_banner(con); | |
31b8006e | 1831 | set_bit(CONNECTING, &con->state); |
eed0ef2c | 1832 | clear_bit(NEGOTIATING, &con->state); |
31b8006e | 1833 | |
cf3e5c40 | 1834 | BUG_ON(con->in_msg); |
31b8006e SW |
1835 | con->in_tag = CEPH_MSGR_TAG_READY; |
1836 | dout("try_write initiating connect on %p new state %lu\n", | |
1837 | con, con->state); | |
1838 | con->sock = ceph_tcp_connect(con); | |
1839 | if (IS_ERR(con->sock)) { | |
1840 | con->sock = NULL; | |
1841 | con->error_msg = "connect error"; | |
1842 | ret = -1; | |
1843 | goto out; | |
1844 | } | |
1845 | } | |
1846 | ||
1847 | more_kvec: | |
1848 | /* kvec data queued? */ | |
1849 | if (con->out_skip) { | |
1850 | ret = write_partial_skip(con); | |
1851 | if (ret <= 0) | |
42961d23 | 1852 | goto out; |
31b8006e SW |
1853 | } |
1854 | if (con->out_kvec_left) { | |
1855 | ret = write_partial_kvec(con); | |
1856 | if (ret <= 0) | |
42961d23 | 1857 | goto out; |
31b8006e SW |
1858 | } |
1859 | ||
1860 | /* msg pages? */ | |
1861 | if (con->out_msg) { | |
c86a2930 SW |
1862 | if (con->out_msg_done) { |
1863 | ceph_msg_put(con->out_msg); | |
1864 | con->out_msg = NULL; /* we're done with this one */ | |
1865 | goto do_next; | |
1866 | } | |
1867 | ||
31b8006e SW |
1868 | ret = write_partial_msg_pages(con); |
1869 | if (ret == 1) | |
1870 | goto more_kvec; /* we need to send the footer, too! */ | |
1871 | if (ret == 0) | |
42961d23 | 1872 | goto out; |
31b8006e SW |
1873 | if (ret < 0) { |
1874 | dout("try_write write_partial_msg_pages err %d\n", | |
1875 | ret); | |
42961d23 | 1876 | goto out; |
31b8006e SW |
1877 | } |
1878 | } | |
1879 | ||
c86a2930 | 1880 | do_next: |
31b8006e SW |
1881 | if (!test_bit(CONNECTING, &con->state)) { |
1882 | /* is anything else pending? */ | |
1883 | if (!list_empty(&con->out_queue)) { | |
1884 | prepare_write_message(con); | |
1885 | goto more; | |
1886 | } | |
1887 | if (con->in_seq > con->in_seq_acked) { | |
1888 | prepare_write_ack(con); | |
1889 | goto more; | |
1890 | } | |
1891 | if (test_and_clear_bit(KEEPALIVE_PENDING, &con->state)) { | |
1892 | prepare_write_keepalive(con); | |
1893 | goto more; | |
1894 | } | |
1895 | } | |
1896 | ||
1897 | /* Nothing to do! */ | |
1898 | clear_bit(WRITE_PENDING, &con->state); | |
1899 | dout("try_write nothing else to write.\n"); | |
31b8006e SW |
1900 | ret = 0; |
1901 | out: | |
42961d23 | 1902 | dout("try_write done on %p ret %d\n", con, ret); |
31b8006e SW |
1903 | return ret; |
1904 | } | |
1905 | ||
1906 | ||
1907 | ||
1908 | /* | |
1909 | * Read what we can from the socket. | |
1910 | */ | |
1911 | static int try_read(struct ceph_connection *con) | |
1912 | { | |
31b8006e SW |
1913 | int ret = -1; |
1914 | ||
1915 | if (!con->sock) | |
1916 | return 0; | |
1917 | ||
1918 | if (test_bit(STANDBY, &con->state)) | |
1919 | return 0; | |
1920 | ||
1921 | dout("try_read start on %p\n", con); | |
ec302645 | 1922 | |
31b8006e SW |
1923 | more: |
1924 | dout("try_read tag %d in_base_pos %d\n", (int)con->in_tag, | |
1925 | con->in_base_pos); | |
0da5d703 SW |
1926 | |
1927 | /* | |
1928 | * process_connect and process_message drop and re-take | |
1929 | * con->mutex. make sure we handle a racing close or reopen. | |
1930 | */ | |
1931 | if (test_bit(CLOSED, &con->state) || | |
1932 | test_bit(OPENING, &con->state)) { | |
1933 | ret = -EAGAIN; | |
1934 | goto out; | |
1935 | } | |
1936 | ||
31b8006e | 1937 | if (test_bit(CONNECTING, &con->state)) { |
eed0ef2c SW |
1938 | if (!test_bit(NEGOTIATING, &con->state)) { |
1939 | dout("try_read connecting\n"); | |
1940 | ret = read_partial_banner(con); | |
1941 | if (ret <= 0) | |
eed0ef2c | 1942 | goto out; |
98bdb0aa SW |
1943 | ret = process_banner(con); |
1944 | if (ret < 0) | |
1945 | goto out; | |
eed0ef2c | 1946 | } |
31b8006e SW |
1947 | ret = read_partial_connect(con); |
1948 | if (ret <= 0) | |
31b8006e | 1949 | goto out; |
98bdb0aa SW |
1950 | ret = process_connect(con); |
1951 | if (ret < 0) | |
1952 | goto out; | |
31b8006e SW |
1953 | goto more; |
1954 | } | |
1955 | ||
1956 | if (con->in_base_pos < 0) { | |
1957 | /* | |
1958 | * skipping + discarding content. | |
1959 | * | |
1960 | * FIXME: there must be a better way to do this! | |
1961 | */ | |
1962 | static char buf[1024]; | |
1963 | int skip = min(1024, -con->in_base_pos); | |
1964 | dout("skipping %d / %d bytes\n", skip, -con->in_base_pos); | |
1965 | ret = ceph_tcp_recvmsg(con->sock, buf, skip); | |
1966 | if (ret <= 0) | |
98bdb0aa | 1967 | goto out; |
31b8006e SW |
1968 | con->in_base_pos += ret; |
1969 | if (con->in_base_pos) | |
1970 | goto more; | |
1971 | } | |
1972 | if (con->in_tag == CEPH_MSGR_TAG_READY) { | |
1973 | /* | |
1974 | * what's next? | |
1975 | */ | |
1976 | ret = ceph_tcp_recvmsg(con->sock, &con->in_tag, 1); | |
1977 | if (ret <= 0) | |
98bdb0aa | 1978 | goto out; |
31b8006e SW |
1979 | dout("try_read got tag %d\n", (int)con->in_tag); |
1980 | switch (con->in_tag) { | |
1981 | case CEPH_MSGR_TAG_MSG: | |
1982 | prepare_read_message(con); | |
1983 | break; | |
1984 | case CEPH_MSGR_TAG_ACK: | |
1985 | prepare_read_ack(con); | |
1986 | break; | |
1987 | case CEPH_MSGR_TAG_CLOSE: | |
1988 | set_bit(CLOSED, &con->state); /* fixme */ | |
98bdb0aa | 1989 | goto out; |
31b8006e SW |
1990 | default: |
1991 | goto bad_tag; | |
1992 | } | |
1993 | } | |
1994 | if (con->in_tag == CEPH_MSGR_TAG_MSG) { | |
1995 | ret = read_partial_message(con); | |
1996 | if (ret <= 0) { | |
1997 | switch (ret) { | |
1998 | case -EBADMSG: | |
1999 | con->error_msg = "bad crc"; | |
2000 | ret = -EIO; | |
98bdb0aa | 2001 | break; |
31b8006e SW |
2002 | case -EIO: |
2003 | con->error_msg = "io error"; | |
98bdb0aa | 2004 | break; |
31b8006e | 2005 | } |
98bdb0aa | 2006 | goto out; |
31b8006e SW |
2007 | } |
2008 | if (con->in_tag == CEPH_MSGR_TAG_READY) | |
2009 | goto more; | |
2010 | process_message(con); | |
2011 | goto more; | |
2012 | } | |
2013 | if (con->in_tag == CEPH_MSGR_TAG_ACK) { | |
2014 | ret = read_partial_ack(con); | |
2015 | if (ret <= 0) | |
98bdb0aa | 2016 | goto out; |
31b8006e SW |
2017 | process_ack(con); |
2018 | goto more; | |
2019 | } | |
2020 | ||
31b8006e | 2021 | out: |
98bdb0aa | 2022 | dout("try_read done on %p ret %d\n", con, ret); |
31b8006e SW |
2023 | return ret; |
2024 | ||
2025 | bad_tag: | |
2026 | pr_err("try_read bad con->in_tag = %d\n", (int)con->in_tag); | |
2027 | con->error_msg = "protocol error, garbage tag"; | |
2028 | ret = -1; | |
2029 | goto out; | |
2030 | } | |
2031 | ||
2032 | ||
2033 | /* | |
2034 | * Atomically queue work on a connection. Bump @con reference to | |
2035 | * avoid races with connection teardown. | |
31b8006e SW |
2036 | */ |
2037 | static void queue_con(struct ceph_connection *con) | |
2038 | { | |
2039 | if (test_bit(DEAD, &con->state)) { | |
2040 | dout("queue_con %p ignoring: DEAD\n", | |
2041 | con); | |
2042 | return; | |
2043 | } | |
2044 | ||
2045 | if (!con->ops->get(con)) { | |
2046 | dout("queue_con %p ref count 0\n", con); | |
2047 | return; | |
2048 | } | |
2049 | ||
f363e45f | 2050 | if (!queue_delayed_work(ceph_msgr_wq, &con->work, 0)) { |
31b8006e SW |
2051 | dout("queue_con %p - already queued\n", con); |
2052 | con->ops->put(con); | |
2053 | } else { | |
2054 | dout("queue_con %p\n", con); | |
2055 | } | |
2056 | } | |
2057 | ||
2058 | /* | |
2059 | * Do some work on a connection. Drop a connection ref when we're done. | |
2060 | */ | |
2061 | static void con_work(struct work_struct *work) | |
2062 | { | |
2063 | struct ceph_connection *con = container_of(work, struct ceph_connection, | |
2064 | work.work); | |
0da5d703 | 2065 | int ret; |
31b8006e | 2066 | |
9dd4658d | 2067 | mutex_lock(&con->mutex); |
0da5d703 | 2068 | restart: |
60bf8bf8 SW |
2069 | if (test_and_clear_bit(BACKOFF, &con->state)) { |
2070 | dout("con_work %p backing off\n", con); | |
2071 | if (queue_delayed_work(ceph_msgr_wq, &con->work, | |
2072 | round_jiffies_relative(con->delay))) { | |
2073 | dout("con_work %p backoff %lu\n", con, con->delay); | |
2074 | mutex_unlock(&con->mutex); | |
2075 | return; | |
2076 | } else { | |
2077 | con->ops->put(con); | |
2078 | dout("con_work %p FAILED to back off %lu\n", con, | |
2079 | con->delay); | |
2080 | } | |
2081 | } | |
9dd4658d | 2082 | |
e00de341 SW |
2083 | if (test_bit(STANDBY, &con->state)) { |
2084 | dout("con_work %p STANDBY\n", con); | |
2085 | goto done; | |
2086 | } | |
31b8006e SW |
2087 | if (test_bit(CLOSED, &con->state)) { /* e.g. if we are replaced */ |
2088 | dout("con_work CLOSED\n"); | |
2089 | con_close_socket(con); | |
2090 | goto done; | |
2091 | } | |
2092 | if (test_and_clear_bit(OPENING, &con->state)) { | |
2093 | /* reopen w/ new peer */ | |
2094 | dout("con_work OPENING\n"); | |
2095 | con_close_socket(con); | |
2096 | } | |
2097 | ||
0da5d703 SW |
2098 | if (test_and_clear_bit(SOCK_CLOSED, &con->state)) |
2099 | goto fault; | |
2100 | ||
2101 | ret = try_read(con); | |
2102 | if (ret == -EAGAIN) | |
2103 | goto restart; | |
2104 | if (ret < 0) | |
2105 | goto fault; | |
2106 | ||
2107 | ret = try_write(con); | |
2108 | if (ret == -EAGAIN) | |
2109 | goto restart; | |
2110 | if (ret < 0) | |
2111 | goto fault; | |
31b8006e SW |
2112 | |
2113 | done: | |
9dd4658d | 2114 | mutex_unlock(&con->mutex); |
9dd4658d | 2115 | done_unlocked: |
31b8006e | 2116 | con->ops->put(con); |
0da5d703 SW |
2117 | return; |
2118 | ||
2119 | fault: | |
2120 | mutex_unlock(&con->mutex); | |
2121 | ceph_fault(con); /* error/fault path */ | |
2122 | goto done_unlocked; | |
31b8006e SW |
2123 | } |
2124 | ||
2125 | ||
2126 | /* | |
2127 | * Generic error/fault handler. A retry mechanism is used with | |
2128 | * exponential backoff | |
2129 | */ | |
2130 | static void ceph_fault(struct ceph_connection *con) | |
2131 | { | |
2132 | pr_err("%s%lld %s %s\n", ENTITY_NAME(con->peer_name), | |
3d14c5d2 | 2133 | ceph_pr_addr(&con->peer_addr.in_addr), con->error_msg); |
31b8006e | 2134 | dout("fault %p state %lu to peer %s\n", |
3d14c5d2 | 2135 | con, con->state, ceph_pr_addr(&con->peer_addr.in_addr)); |
31b8006e SW |
2136 | |
2137 | if (test_bit(LOSSYTX, &con->state)) { | |
2138 | dout("fault on LOSSYTX channel\n"); | |
2139 | goto out; | |
2140 | } | |
2141 | ||
ec302645 | 2142 | mutex_lock(&con->mutex); |
91e45ce3 SW |
2143 | if (test_bit(CLOSED, &con->state)) |
2144 | goto out_unlock; | |
ec302645 | 2145 | |
31b8006e | 2146 | con_close_socket(con); |
5e095e8b SW |
2147 | |
2148 | if (con->in_msg) { | |
2149 | ceph_msg_put(con->in_msg); | |
2150 | con->in_msg = NULL; | |
2151 | } | |
31b8006e | 2152 | |
e80a52d1 SW |
2153 | /* Requeue anything that hasn't been acked */ |
2154 | list_splice_init(&con->out_sent, &con->out_queue); | |
9bd2e6f8 | 2155 | |
e76661d0 SW |
2156 | /* If there are no messages queued or keepalive pending, place |
2157 | * the connection in a STANDBY state */ | |
2158 | if (list_empty(&con->out_queue) && | |
2159 | !test_bit(KEEPALIVE_PENDING, &con->state)) { | |
e00de341 SW |
2160 | dout("fault %p setting STANDBY clearing WRITE_PENDING\n", con); |
2161 | clear_bit(WRITE_PENDING, &con->state); | |
31b8006e | 2162 | set_bit(STANDBY, &con->state); |
e80a52d1 SW |
2163 | } else { |
2164 | /* retry after a delay. */ | |
2165 | if (con->delay == 0) | |
2166 | con->delay = BASE_DELAY_INTERVAL; | |
2167 | else if (con->delay < MAX_DELAY_INTERVAL) | |
2168 | con->delay *= 2; | |
e80a52d1 SW |
2169 | con->ops->get(con); |
2170 | if (queue_delayed_work(ceph_msgr_wq, &con->work, | |
60bf8bf8 SW |
2171 | round_jiffies_relative(con->delay))) { |
2172 | dout("fault queued %p delay %lu\n", con, con->delay); | |
2173 | } else { | |
e80a52d1 | 2174 | con->ops->put(con); |
60bf8bf8 SW |
2175 | dout("fault failed to queue %p delay %lu, backoff\n", |
2176 | con, con->delay); | |
2177 | /* | |
2178 | * In many cases we see a socket state change | |
2179 | * while con_work is running and end up | |
2180 | * queuing (non-delayed) work, such that we | |
2181 | * can't backoff with a delay. Set a flag so | |
2182 | * that when con_work restarts we schedule the | |
2183 | * delay then. | |
2184 | */ | |
2185 | set_bit(BACKOFF, &con->state); | |
2186 | } | |
31b8006e SW |
2187 | } |
2188 | ||
91e45ce3 SW |
2189 | out_unlock: |
2190 | mutex_unlock(&con->mutex); | |
31b8006e | 2191 | out: |
161fd65a SW |
2192 | /* |
2193 | * in case we faulted due to authentication, invalidate our | |
2194 | * current tickets so that we can get new ones. | |
213c99ee | 2195 | */ |
161fd65a SW |
2196 | if (con->auth_retry && con->ops->invalidate_authorizer) { |
2197 | dout("calling invalidate_authorizer()\n"); | |
2198 | con->ops->invalidate_authorizer(con); | |
2199 | } | |
2200 | ||
31b8006e SW |
2201 | if (con->ops->fault) |
2202 | con->ops->fault(con); | |
2203 | } | |
2204 | ||
2205 | ||
2206 | ||
2207 | /* | |
2208 | * create a new messenger instance | |
2209 | */ | |
3d14c5d2 YS |
2210 | struct ceph_messenger *ceph_messenger_create(struct ceph_entity_addr *myaddr, |
2211 | u32 supported_features, | |
2212 | u32 required_features) | |
31b8006e SW |
2213 | { |
2214 | struct ceph_messenger *msgr; | |
2215 | ||
2216 | msgr = kzalloc(sizeof(*msgr), GFP_KERNEL); | |
2217 | if (msgr == NULL) | |
2218 | return ERR_PTR(-ENOMEM); | |
2219 | ||
3d14c5d2 YS |
2220 | msgr->supported_features = supported_features; |
2221 | msgr->required_features = required_features; | |
2222 | ||
31b8006e SW |
2223 | spin_lock_init(&msgr->global_seq_lock); |
2224 | ||
2225 | /* the zero page is needed if a request is "canceled" while the message | |
2226 | * is being written over the socket */ | |
31459fe4 | 2227 | msgr->zero_page = __page_cache_alloc(GFP_KERNEL | __GFP_ZERO); |
31b8006e SW |
2228 | if (!msgr->zero_page) { |
2229 | kfree(msgr); | |
2230 | return ERR_PTR(-ENOMEM); | |
2231 | } | |
2232 | kmap(msgr->zero_page); | |
2233 | ||
2234 | if (myaddr) | |
2235 | msgr->inst.addr = *myaddr; | |
2236 | ||
2237 | /* select a random nonce */ | |
ac8839d7 | 2238 | msgr->inst.addr.type = 0; |
103e2d3a | 2239 | get_random_bytes(&msgr->inst.addr.nonce, sizeof(msgr->inst.addr.nonce)); |
63f2d211 | 2240 | encode_my_addr(msgr); |
31b8006e SW |
2241 | |
2242 | dout("messenger_create %p\n", msgr); | |
2243 | return msgr; | |
2244 | } | |
3d14c5d2 | 2245 | EXPORT_SYMBOL(ceph_messenger_create); |
31b8006e SW |
2246 | |
2247 | void ceph_messenger_destroy(struct ceph_messenger *msgr) | |
2248 | { | |
2249 | dout("destroy %p\n", msgr); | |
2250 | kunmap(msgr->zero_page); | |
2251 | __free_page(msgr->zero_page); | |
2252 | kfree(msgr); | |
2253 | dout("destroyed messenger %p\n", msgr); | |
2254 | } | |
3d14c5d2 | 2255 | EXPORT_SYMBOL(ceph_messenger_destroy); |
31b8006e | 2256 | |
e00de341 SW |
2257 | static void clear_standby(struct ceph_connection *con) |
2258 | { | |
2259 | /* come back from STANDBY? */ | |
2260 | if (test_and_clear_bit(STANDBY, &con->state)) { | |
2261 | mutex_lock(&con->mutex); | |
2262 | dout("clear_standby %p and ++connect_seq\n", con); | |
2263 | con->connect_seq++; | |
2264 | WARN_ON(test_bit(WRITE_PENDING, &con->state)); | |
2265 | WARN_ON(test_bit(KEEPALIVE_PENDING, &con->state)); | |
2266 | mutex_unlock(&con->mutex); | |
2267 | } | |
2268 | } | |
2269 | ||
31b8006e SW |
2270 | /* |
2271 | * Queue up an outgoing message on the given connection. | |
2272 | */ | |
2273 | void ceph_con_send(struct ceph_connection *con, struct ceph_msg *msg) | |
2274 | { | |
2275 | if (test_bit(CLOSED, &con->state)) { | |
2276 | dout("con_send %p closed, dropping %p\n", con, msg); | |
2277 | ceph_msg_put(msg); | |
2278 | return; | |
2279 | } | |
2280 | ||
2281 | /* set src+dst */ | |
dbad185d | 2282 | msg->hdr.src = con->msgr->inst.name; |
31b8006e | 2283 | |
3ca02ef9 SW |
2284 | BUG_ON(msg->front.iov_len != le32_to_cpu(msg->hdr.front_len)); |
2285 | ||
e84346b7 SW |
2286 | msg->needs_out_seq = true; |
2287 | ||
31b8006e | 2288 | /* queue */ |
ec302645 | 2289 | mutex_lock(&con->mutex); |
31b8006e SW |
2290 | BUG_ON(!list_empty(&msg->list_head)); |
2291 | list_add_tail(&msg->list_head, &con->out_queue); | |
2292 | dout("----- %p to %s%lld %d=%s len %d+%d+%d -----\n", msg, | |
2293 | ENTITY_NAME(con->peer_name), le16_to_cpu(msg->hdr.type), | |
2294 | ceph_msg_type_name(le16_to_cpu(msg->hdr.type)), | |
2295 | le32_to_cpu(msg->hdr.front_len), | |
2296 | le32_to_cpu(msg->hdr.middle_len), | |
2297 | le32_to_cpu(msg->hdr.data_len)); | |
ec302645 | 2298 | mutex_unlock(&con->mutex); |
31b8006e SW |
2299 | |
2300 | /* if there wasn't anything waiting to send before, queue | |
2301 | * new work */ | |
e00de341 | 2302 | clear_standby(con); |
31b8006e SW |
2303 | if (test_and_set_bit(WRITE_PENDING, &con->state) == 0) |
2304 | queue_con(con); | |
2305 | } | |
3d14c5d2 | 2306 | EXPORT_SYMBOL(ceph_con_send); |
31b8006e SW |
2307 | |
2308 | /* | |
2309 | * Revoke a message that was previously queued for send | |
2310 | */ | |
2311 | void ceph_con_revoke(struct ceph_connection *con, struct ceph_msg *msg) | |
2312 | { | |
ec302645 | 2313 | mutex_lock(&con->mutex); |
31b8006e | 2314 | if (!list_empty(&msg->list_head)) { |
ed98adad | 2315 | dout("con_revoke %p msg %p - was on queue\n", con, msg); |
31b8006e SW |
2316 | list_del_init(&msg->list_head); |
2317 | ceph_msg_put(msg); | |
2318 | msg->hdr.seq = 0; | |
ed98adad SW |
2319 | } |
2320 | if (con->out_msg == msg) { | |
2321 | dout("con_revoke %p msg %p - was sending\n", con, msg); | |
2322 | con->out_msg = NULL; | |
31b8006e SW |
2323 | if (con->out_kvec_is_msg) { |
2324 | con->out_skip = con->out_kvec_bytes; | |
2325 | con->out_kvec_is_msg = false; | |
2326 | } | |
ed98adad SW |
2327 | ceph_msg_put(msg); |
2328 | msg->hdr.seq = 0; | |
31b8006e | 2329 | } |
ec302645 | 2330 | mutex_unlock(&con->mutex); |
31b8006e SW |
2331 | } |
2332 | ||
350b1c32 | 2333 | /* |
0d59ab81 | 2334 | * Revoke a message that we may be reading data into |
350b1c32 | 2335 | */ |
0d59ab81 | 2336 | void ceph_con_revoke_message(struct ceph_connection *con, struct ceph_msg *msg) |
350b1c32 SW |
2337 | { |
2338 | mutex_lock(&con->mutex); | |
0d59ab81 YS |
2339 | if (con->in_msg && con->in_msg == msg) { |
2340 | unsigned front_len = le32_to_cpu(con->in_hdr.front_len); | |
2341 | unsigned middle_len = le32_to_cpu(con->in_hdr.middle_len); | |
350b1c32 SW |
2342 | unsigned data_len = le32_to_cpu(con->in_hdr.data_len); |
2343 | ||
2344 | /* skip rest of message */ | |
0d59ab81 | 2345 | dout("con_revoke_pages %p msg %p revoked\n", con, msg); |
350b1c32 SW |
2346 | con->in_base_pos = con->in_base_pos - |
2347 | sizeof(struct ceph_msg_header) - | |
0d59ab81 YS |
2348 | front_len - |
2349 | middle_len - | |
2350 | data_len - | |
350b1c32 | 2351 | sizeof(struct ceph_msg_footer); |
350b1c32 SW |
2352 | ceph_msg_put(con->in_msg); |
2353 | con->in_msg = NULL; | |
2354 | con->in_tag = CEPH_MSGR_TAG_READY; | |
684be25c | 2355 | con->in_seq++; |
350b1c32 SW |
2356 | } else { |
2357 | dout("con_revoke_pages %p msg %p pages %p no-op\n", | |
0d59ab81 | 2358 | con, con->in_msg, msg); |
350b1c32 SW |
2359 | } |
2360 | mutex_unlock(&con->mutex); | |
2361 | } | |
2362 | ||
31b8006e SW |
2363 | /* |
2364 | * Queue a keepalive byte to ensure the tcp connection is alive. | |
2365 | */ | |
2366 | void ceph_con_keepalive(struct ceph_connection *con) | |
2367 | { | |
e00de341 SW |
2368 | dout("con_keepalive %p\n", con); |
2369 | clear_standby(con); | |
31b8006e SW |
2370 | if (test_and_set_bit(KEEPALIVE_PENDING, &con->state) == 0 && |
2371 | test_and_set_bit(WRITE_PENDING, &con->state) == 0) | |
2372 | queue_con(con); | |
2373 | } | |
3d14c5d2 | 2374 | EXPORT_SYMBOL(ceph_con_keepalive); |
31b8006e SW |
2375 | |
2376 | ||
2377 | /* | |
2378 | * construct a new message with given type, size | |
2379 | * the new msg has a ref count of 1. | |
2380 | */ | |
b61c2763 SW |
2381 | struct ceph_msg *ceph_msg_new(int type, int front_len, gfp_t flags, |
2382 | bool can_fail) | |
31b8006e SW |
2383 | { |
2384 | struct ceph_msg *m; | |
2385 | ||
34d23762 | 2386 | m = kmalloc(sizeof(*m), flags); |
31b8006e SW |
2387 | if (m == NULL) |
2388 | goto out; | |
c2e552e7 | 2389 | kref_init(&m->kref); |
31b8006e SW |
2390 | INIT_LIST_HEAD(&m->list_head); |
2391 | ||
45c6ceb5 | 2392 | m->hdr.tid = 0; |
31b8006e | 2393 | m->hdr.type = cpu_to_le16(type); |
45c6ceb5 SW |
2394 | m->hdr.priority = cpu_to_le16(CEPH_MSG_PRIO_DEFAULT); |
2395 | m->hdr.version = 0; | |
31b8006e SW |
2396 | m->hdr.front_len = cpu_to_le32(front_len); |
2397 | m->hdr.middle_len = 0; | |
bb257664 SW |
2398 | m->hdr.data_len = 0; |
2399 | m->hdr.data_off = 0; | |
45c6ceb5 | 2400 | m->hdr.reserved = 0; |
31b8006e SW |
2401 | m->footer.front_crc = 0; |
2402 | m->footer.middle_crc = 0; | |
2403 | m->footer.data_crc = 0; | |
45c6ceb5 | 2404 | m->footer.flags = 0; |
31b8006e SW |
2405 | m->front_max = front_len; |
2406 | m->front_is_vmalloc = false; | |
2407 | m->more_to_follow = false; | |
c0d5f9db | 2408 | m->ack_stamp = 0; |
31b8006e SW |
2409 | m->pool = NULL; |
2410 | ||
ca20892d HC |
2411 | /* middle */ |
2412 | m->middle = NULL; | |
2413 | ||
2414 | /* data */ | |
2415 | m->nr_pages = 0; | |
2416 | m->page_alignment = 0; | |
2417 | m->pages = NULL; | |
2418 | m->pagelist = NULL; | |
2419 | m->bio = NULL; | |
2420 | m->bio_iter = NULL; | |
2421 | m->bio_seg = 0; | |
2422 | m->trail = NULL; | |
2423 | ||
31b8006e SW |
2424 | /* front */ |
2425 | if (front_len) { | |
2426 | if (front_len > PAGE_CACHE_SIZE) { | |
34d23762 | 2427 | m->front.iov_base = __vmalloc(front_len, flags, |
31b8006e SW |
2428 | PAGE_KERNEL); |
2429 | m->front_is_vmalloc = true; | |
2430 | } else { | |
34d23762 | 2431 | m->front.iov_base = kmalloc(front_len, flags); |
31b8006e SW |
2432 | } |
2433 | if (m->front.iov_base == NULL) { | |
b61c2763 | 2434 | dout("ceph_msg_new can't allocate %d bytes\n", |
31b8006e SW |
2435 | front_len); |
2436 | goto out2; | |
2437 | } | |
2438 | } else { | |
2439 | m->front.iov_base = NULL; | |
2440 | } | |
2441 | m->front.iov_len = front_len; | |
2442 | ||
bb257664 | 2443 | dout("ceph_msg_new %p front %d\n", m, front_len); |
31b8006e SW |
2444 | return m; |
2445 | ||
2446 | out2: | |
2447 | ceph_msg_put(m); | |
2448 | out: | |
b61c2763 SW |
2449 | if (!can_fail) { |
2450 | pr_err("msg_new can't create type %d front %d\n", type, | |
2451 | front_len); | |
f0ed1b7c | 2452 | WARN_ON(1); |
b61c2763 SW |
2453 | } else { |
2454 | dout("msg_new can't create type %d front %d\n", type, | |
2455 | front_len); | |
2456 | } | |
a79832f2 | 2457 | return NULL; |
31b8006e | 2458 | } |
3d14c5d2 | 2459 | EXPORT_SYMBOL(ceph_msg_new); |
31b8006e | 2460 | |
31b8006e SW |
2461 | /* |
2462 | * Allocate "middle" portion of a message, if it is needed and wasn't | |
2463 | * allocated by alloc_msg. This allows us to read a small fixed-size | |
2464 | * per-type header in the front and then gracefully fail (i.e., | |
2465 | * propagate the error to the caller based on info in the front) when | |
2466 | * the middle is too large. | |
2467 | */ | |
2450418c | 2468 | static int ceph_alloc_middle(struct ceph_connection *con, struct ceph_msg *msg) |
31b8006e SW |
2469 | { |
2470 | int type = le16_to_cpu(msg->hdr.type); | |
2471 | int middle_len = le32_to_cpu(msg->hdr.middle_len); | |
2472 | ||
2473 | dout("alloc_middle %p type %d %s middle_len %d\n", msg, type, | |
2474 | ceph_msg_type_name(type), middle_len); | |
2475 | BUG_ON(!middle_len); | |
2476 | BUG_ON(msg->middle); | |
2477 | ||
b6c1d5b8 | 2478 | msg->middle = ceph_buffer_new(middle_len, GFP_NOFS); |
31b8006e SW |
2479 | if (!msg->middle) |
2480 | return -ENOMEM; | |
2481 | return 0; | |
2482 | } | |
2483 | ||
2450418c YS |
2484 | /* |
2485 | * Generic message allocator, for incoming messages. | |
2486 | */ | |
2487 | static struct ceph_msg *ceph_alloc_msg(struct ceph_connection *con, | |
2488 | struct ceph_msg_header *hdr, | |
2489 | int *skip) | |
2490 | { | |
2491 | int type = le16_to_cpu(hdr->type); | |
2492 | int front_len = le32_to_cpu(hdr->front_len); | |
2493 | int middle_len = le32_to_cpu(hdr->middle_len); | |
2494 | struct ceph_msg *msg = NULL; | |
2495 | int ret; | |
2496 | ||
2497 | if (con->ops->alloc_msg) { | |
0547a9b3 | 2498 | mutex_unlock(&con->mutex); |
2450418c | 2499 | msg = con->ops->alloc_msg(con, hdr, skip); |
0547a9b3 | 2500 | mutex_lock(&con->mutex); |
a79832f2 | 2501 | if (!msg || *skip) |
2450418c YS |
2502 | return NULL; |
2503 | } | |
2504 | if (!msg) { | |
2505 | *skip = 0; | |
b61c2763 | 2506 | msg = ceph_msg_new(type, front_len, GFP_NOFS, false); |
2450418c YS |
2507 | if (!msg) { |
2508 | pr_err("unable to allocate msg type %d len %d\n", | |
2509 | type, front_len); | |
a79832f2 | 2510 | return NULL; |
2450418c | 2511 | } |
c5c6b19d | 2512 | msg->page_alignment = le16_to_cpu(hdr->data_off); |
2450418c | 2513 | } |
9d7f0f13 | 2514 | memcpy(&msg->hdr, &con->in_hdr, sizeof(con->in_hdr)); |
2450418c | 2515 | |
bb257664 | 2516 | if (middle_len && !msg->middle) { |
2450418c | 2517 | ret = ceph_alloc_middle(con, msg); |
2450418c YS |
2518 | if (ret < 0) { |
2519 | ceph_msg_put(msg); | |
a79832f2 | 2520 | return NULL; |
2450418c YS |
2521 | } |
2522 | } | |
9d7f0f13 | 2523 | |
2450418c YS |
2524 | return msg; |
2525 | } | |
2526 | ||
31b8006e SW |
2527 | |
2528 | /* | |
2529 | * Free a generically kmalloc'd message. | |
2530 | */ | |
2531 | void ceph_msg_kfree(struct ceph_msg *m) | |
2532 | { | |
2533 | dout("msg_kfree %p\n", m); | |
2534 | if (m->front_is_vmalloc) | |
2535 | vfree(m->front.iov_base); | |
2536 | else | |
2537 | kfree(m->front.iov_base); | |
2538 | kfree(m); | |
2539 | } | |
2540 | ||
2541 | /* | |
2542 | * Drop a msg ref. Destroy as needed. | |
2543 | */ | |
c2e552e7 SW |
2544 | void ceph_msg_last_put(struct kref *kref) |
2545 | { | |
2546 | struct ceph_msg *m = container_of(kref, struct ceph_msg, kref); | |
31b8006e | 2547 | |
c2e552e7 SW |
2548 | dout("ceph_msg_put last one on %p\n", m); |
2549 | WARN_ON(!list_empty(&m->list_head)); | |
2550 | ||
2551 | /* drop middle, data, if any */ | |
2552 | if (m->middle) { | |
2553 | ceph_buffer_put(m->middle); | |
2554 | m->middle = NULL; | |
31b8006e | 2555 | } |
c2e552e7 SW |
2556 | m->nr_pages = 0; |
2557 | m->pages = NULL; | |
2558 | ||
58bb3b37 SW |
2559 | if (m->pagelist) { |
2560 | ceph_pagelist_release(m->pagelist); | |
2561 | kfree(m->pagelist); | |
2562 | m->pagelist = NULL; | |
2563 | } | |
2564 | ||
68b4476b YS |
2565 | m->trail = NULL; |
2566 | ||
c2e552e7 SW |
2567 | if (m->pool) |
2568 | ceph_msgpool_put(m->pool, m); | |
2569 | else | |
2570 | ceph_msg_kfree(m); | |
31b8006e | 2571 | } |
3d14c5d2 | 2572 | EXPORT_SYMBOL(ceph_msg_last_put); |
9ec7cab1 SW |
2573 | |
2574 | void ceph_msg_dump(struct ceph_msg *msg) | |
2575 | { | |
2576 | pr_debug("msg_dump %p (front_max %d nr_pages %d)\n", msg, | |
2577 | msg->front_max, msg->nr_pages); | |
2578 | print_hex_dump(KERN_DEBUG, "header: ", | |
2579 | DUMP_PREFIX_OFFSET, 16, 1, | |
2580 | &msg->hdr, sizeof(msg->hdr), true); | |
2581 | print_hex_dump(KERN_DEBUG, " front: ", | |
2582 | DUMP_PREFIX_OFFSET, 16, 1, | |
2583 | msg->front.iov_base, msg->front.iov_len, true); | |
2584 | if (msg->middle) | |
2585 | print_hex_dump(KERN_DEBUG, "middle: ", | |
2586 | DUMP_PREFIX_OFFSET, 16, 1, | |
2587 | msg->middle->vec.iov_base, | |
2588 | msg->middle->vec.iov_len, true); | |
2589 | print_hex_dump(KERN_DEBUG, "footer: ", | |
2590 | DUMP_PREFIX_OFFSET, 16, 1, | |
2591 | &msg->footer, sizeof(msg->footer), true); | |
2592 | } | |
3d14c5d2 | 2593 | EXPORT_SYMBOL(ceph_msg_dump); |