?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
kvm tools, setup: Create private directory
[linux-2.6/next.git] / net / ceph / messenger.c
blob78b55f49de7cba5bbdf0fb872b4b3c0464190f04
1 #include <linux/ceph/ceph_debug.h>
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>
9 #include <linux/slab.h>
10 #include <linux/socket.h>
11 #include <linux/string.h>
12 #include <linux/bio.h>
13 #include <linux/blkdev.h>
14 #include <net/tcp.h>
15
16 #include <linux/ceph/libceph.h>
17 #include <linux/ceph/messenger.h>
18 #include <linux/ceph/decode.h>
19 #include <linux/ceph/pagelist.h>
20
21 /*
22 * Ceph uses the messenger to exchange ceph_msg messages with other
23 * hosts in the system. The messenger provides ordered and reliable
24 * delivery. We tolerate TCP disconnects by reconnecting (with
25 * exponential backoff) in the case of a fault (disconnection, bad
26 * crc, protocol error). Acks allow sent messages to be discarded by
27 * the sender.
28 */
29
30 /* static tag bytes (protocol control messages) */
31 static char tag_msg = CEPH_MSGR_TAG_MSG;
32 static char tag_ack = CEPH_MSGR_TAG_ACK;
33 static char tag_keepalive = CEPH_MSGR_TAG_KEEPALIVE;
34
35 #ifdef CONFIG_LOCKDEP
36 static struct lock_class_key socket_class;
37 #endif
38
39
40 static void queue_con(struct ceph_connection *con);
41 static void con_work(struct work_struct *);
42 static void ceph_fault(struct ceph_connection *con);
43
44 /*
45 * nicely render a sockaddr as a string.
46 */
47 #define MAX_ADDR_STR 20
48 #define MAX_ADDR_STR_LEN 60
49 static char addr_str[MAX_ADDR_STR][MAX_ADDR_STR_LEN];
50 static DEFINE_SPINLOCK(addr_str_lock);
51 static int last_addr_str;
52
53 const char *ceph_pr_addr(const struct sockaddr_storage *ss)
54 {
55 int i;
56 char *s;
57 struct sockaddr_in *in4 = (void *)ss;
58 struct sockaddr_in6 *in6 = (void *)ss;
59
60 spin_lock(&addr_str_lock);
61 i = last_addr_str++;
62 if (last_addr_str == MAX_ADDR_STR)
63 last_addr_str = 0;
64 spin_unlock(&addr_str_lock);
65 s = addr_str[i];
66
67 switch (ss->ss_family) {
68 case AF_INET:
69 snprintf(s, MAX_ADDR_STR_LEN, "%pI4:%u", &in4->sin_addr,
70 (unsigned int)ntohs(in4->sin_port));
71 break;
72
73 case AF_INET6:
74 snprintf(s, MAX_ADDR_STR_LEN, "[%pI6c]:%u", &in6->sin6_addr,
75 (unsigned int)ntohs(in6->sin6_port));
76 break;
77
78 default:
79 snprintf(s, MAX_ADDR_STR_LEN, "(unknown sockaddr family %d)",
80 (int)ss->ss_family);
81 }
82
83 return s;
84 }
85 EXPORT_SYMBOL(ceph_pr_addr);
86
87 static void encode_my_addr(struct ceph_messenger *msgr)
88 {
89 memcpy(&msgr->my_enc_addr, &msgr->inst.addr, sizeof(msgr->my_enc_addr));
90 ceph_encode_addr(&msgr->my_enc_addr);
91 }
92
93 /*
94 * work queue for all reading and writing to/from the socket.
95 */
96 struct workqueue_struct *ceph_msgr_wq;
97
98 int ceph_msgr_init(void)
99 {
100 ceph_msgr_wq = alloc_workqueue("ceph-msgr", WQ_NON_REENTRANT, 0);
101 if (!ceph_msgr_wq) {
102 pr_err("msgr_init failed to create workqueue\n");
103 return -ENOMEM;
104 }
105 return 0;
106 }
107 EXPORT_SYMBOL(ceph_msgr_init);
108
109 void ceph_msgr_exit(void)
110 {
111 destroy_workqueue(ceph_msgr_wq);
112 }
113 EXPORT_SYMBOL(ceph_msgr_exit);
114
115 void ceph_msgr_flush(void)
116 {
117 flush_workqueue(ceph_msgr_wq);
118 }
119 EXPORT_SYMBOL(ceph_msgr_flush);
120
121
122 /*
123 * socket callback functions
124 */
125
126 /* data available on socket, or listen socket received a connect */
127 static void ceph_data_ready(struct sock *sk, int count_unused)
128 {
129 struct ceph_connection *con =
130 (struct ceph_connection *)sk->sk_user_data;
131 if (sk->sk_state != TCP_CLOSE_WAIT) {
132 dout("ceph_data_ready on %p state = %lu, queueing work\n",
133 con, con->state);
134 queue_con(con);
135 }
136 }
137
138 /* socket has buffer space for writing */
139 static void ceph_write_space(struct sock *sk)
140 {
141 struct ceph_connection *con =
142 (struct ceph_connection *)sk->sk_user_data;
143
144 /* only queue to workqueue if there is data we want to write. */
145 if (test_bit(WRITE_PENDING, &con->state)) {
146 dout("ceph_write_space %p queueing write work\n", con);
147 queue_con(con);
148 } else {
149 dout("ceph_write_space %p nothing to write\n", con);
150 }
151
152 /* since we have our own write_space, clear the SOCK_NOSPACE flag */
153 clear_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
154 }
155
156 /* socket's state has changed */
157 static void ceph_state_change(struct sock *sk)
158 {
159 struct ceph_connection *con =
160 (struct ceph_connection *)sk->sk_user_data;
161
162 dout("ceph_state_change %p state = %lu sk_state = %u\n",
163 con, con->state, sk->sk_state);
164
165 if (test_bit(CLOSED, &con->state))
166 return;
167
168 switch (sk->sk_state) {
169 case TCP_CLOSE:
170 dout("ceph_state_change TCP_CLOSE\n");
171 case TCP_CLOSE_WAIT:
172 dout("ceph_state_change TCP_CLOSE_WAIT\n");
173 if (test_and_set_bit(SOCK_CLOSED, &con->state) == 0) {
174 if (test_bit(CONNECTING, &con->state))
175 con->error_msg = "connection failed";
176 else
177 con->error_msg = "socket closed";
178 queue_con(con);
179 }
180 break;
181 case TCP_ESTABLISHED:
182 dout("ceph_state_change TCP_ESTABLISHED\n");
183 queue_con(con);
184 break;
185 }
186 }
187
188 /*
189 * set up socket callbacks
190 */
191 static void set_sock_callbacks(struct socket *sock,
192 struct ceph_connection *con)
193 {
194 struct sock *sk = sock->sk;
195 sk->sk_user_data = (void *)con;
196 sk->sk_data_ready = ceph_data_ready;
197 sk->sk_write_space = ceph_write_space;
198 sk->sk_state_change = ceph_state_change;
199 }
200
201
202 /*
203 * socket helpers
204 */
205
206 /*
207 * initiate connection to a remote socket.
208 */
209 static struct socket *ceph_tcp_connect(struct ceph_connection *con)
210 {
211 struct sockaddr_storage *paddr = &con->peer_addr.in_addr;
212 struct socket *sock;
213 int ret;
214
215 BUG_ON(con->sock);
216 ret = sock_create_kern(con->peer_addr.in_addr.ss_family, SOCK_STREAM,
217 IPPROTO_TCP, &sock);
218 if (ret)
219 return ERR_PTR(ret);
220 con->sock = sock;
221 sock->sk->sk_allocation = GFP_NOFS;
222
223 #ifdef CONFIG_LOCKDEP
224 lockdep_set_class(&sock->sk->sk_lock, &socket_class);
225 #endif
226
227 set_sock_callbacks(sock, con);
228
229 dout("connect %s\n", ceph_pr_addr(&con->peer_addr.in_addr));
230
231 ret = sock->ops->connect(sock, (struct sockaddr *)paddr, sizeof(*paddr),
232 O_NONBLOCK);
233 if (ret == -EINPROGRESS) {
234 dout("connect %s EINPROGRESS sk_state = %u\n",
235 ceph_pr_addr(&con->peer_addr.in_addr),
236 sock->sk->sk_state);
237 ret = 0;
238 }
239 if (ret < 0) {
240 pr_err("connect %s error %d\n",
241 ceph_pr_addr(&con->peer_addr.in_addr), ret);
242 sock_release(sock);
243 con->sock = NULL;
244 con->error_msg = "connect error";
245 }
246
247 if (ret < 0)
248 return ERR_PTR(ret);
249 return sock;
250 }
251
252 static int ceph_tcp_recvmsg(struct socket *sock, void *buf, size_t len)
253 {
254 struct kvec iov = {buf, len};
255 struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
256 int r;
257
258 r = kernel_recvmsg(sock, &msg, &iov, 1, len, msg.msg_flags);
259 if (r == -EAGAIN)
260 r = 0;
261 return r;
262 }
263
264 /*
265 * write something. @more is true if caller will be sending more data
266 * shortly.
267 */
268 static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov,
269 size_t kvlen, size_t len, int more)
270 {
271 struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
272 int r;
273
274 if (more)
275 msg.msg_flags |= MSG_MORE;
276 else
277 msg.msg_flags |= MSG_EOR; /* superfluous, but what the hell */
278
279 r = kernel_sendmsg(sock, &msg, iov, kvlen, len);
280 if (r == -EAGAIN)
281 r = 0;
282 return r;
283 }
284
285
286 /*
287 * Shutdown/close the socket for the given connection.
288 */
289 static int con_close_socket(struct ceph_connection *con)
290 {
291 int rc;
292
293 dout("con_close_socket on %p sock %p\n", con, con->sock);
294 if (!con->sock)
295 return 0;
296 set_bit(SOCK_CLOSED, &con->state);
297 rc = con->sock->ops->shutdown(con->sock, SHUT_RDWR);
298 sock_release(con->sock);
299 con->sock = NULL;
300 clear_bit(SOCK_CLOSED, &con->state);
301 return rc;
302 }
303
304 /*
305 * Reset a connection. Discard all incoming and outgoing messages
306 * and clear *_seq state.
307 */
308 static void ceph_msg_remove(struct ceph_msg *msg)
309 {
310 list_del_init(&msg->list_head);
311 ceph_msg_put(msg);
312 }
313 static void ceph_msg_remove_list(struct list_head *head)
314 {
315 while (!list_empty(head)) {
316 struct ceph_msg *msg = list_first_entry(head, struct ceph_msg,
317 list_head);
318 ceph_msg_remove(msg);
319 }
320 }
321
322 static void reset_connection(struct ceph_connection *con)
323 {
324 /* reset connection, out_queue, msg_ and connect_seq */
325 /* discard existing out_queue and msg_seq */
326 ceph_msg_remove_list(&con->out_queue);
327 ceph_msg_remove_list(&con->out_sent);
328
329 if (con->in_msg) {
330 ceph_msg_put(con->in_msg);
331 con->in_msg = NULL;
332 }
333
334 con->connect_seq = 0;
335 con->out_seq = 0;
336 if (con->out_msg) {
337 ceph_msg_put(con->out_msg);
338 con->out_msg = NULL;
339 }
340 con->in_seq = 0;
341 con->in_seq_acked = 0;
342 }
343
344 /*
345 * mark a peer down. drop any open connections.
346 */
347 void ceph_con_close(struct ceph_connection *con)
348 {
349 dout("con_close %p peer %s\n", con,
350 ceph_pr_addr(&con->peer_addr.in_addr));
351 set_bit(CLOSED, &con->state); /* in case there's queued work */
352 clear_bit(STANDBY, &con->state); /* avoid connect_seq bump */
353 clear_bit(LOSSYTX, &con->state); /* so we retry next connect */
354 clear_bit(KEEPALIVE_PENDING, &con->state);
355 clear_bit(WRITE_PENDING, &con->state);
356 mutex_lock(&con->mutex);
357 reset_connection(con);
358 con->peer_global_seq = 0;
359 cancel_delayed_work(&con->work);
360 mutex_unlock(&con->mutex);
361 queue_con(con);
362 }
363 EXPORT_SYMBOL(ceph_con_close);
364
365 /*
366 * Reopen a closed connection, with a new peer address.
367 */
368 void ceph_con_open(struct ceph_connection *con, struct ceph_entity_addr *addr)
369 {
370 dout("con_open %p %s\n", con, ceph_pr_addr(&addr->in_addr));
371 set_bit(OPENING, &con->state);
372 clear_bit(CLOSED, &con->state);
373 memcpy(&con->peer_addr, addr, sizeof(*addr));
374 con->delay = 0; /* reset backoff memory */
375 queue_con(con);
376 }
377 EXPORT_SYMBOL(ceph_con_open);
378
379 /*
380 * return true if this connection ever successfully opened
381 */
382 bool ceph_con_opened(struct ceph_connection *con)
383 {
384 return con->connect_seq > 0;
385 }
386
387 /*
388 * generic get/put
389 */
390 struct ceph_connection *ceph_con_get(struct ceph_connection *con)
391 {
392 dout("con_get %p nref = %d -> %d\n", con,
393 atomic_read(&con->nref), atomic_read(&con->nref) + 1);
394 if (atomic_inc_not_zero(&con->nref))
395 return con;
396 return NULL;
397 }
398
399 void ceph_con_put(struct ceph_connection *con)
400 {
401 dout("con_put %p nref = %d -> %d\n", con,
402 atomic_read(&con->nref), atomic_read(&con->nref) - 1);
403 BUG_ON(atomic_read(&con->nref) == 0);
404 if (atomic_dec_and_test(&con->nref)) {
405 BUG_ON(con->sock);
406 kfree(con);
407 }
408 }
409
410 /*
411 * initialize a new connection.
412 */
413 void ceph_con_init(struct ceph_messenger *msgr, struct ceph_connection *con)
414 {
415 dout("con_init %p\n", con);
416 memset(con, 0, sizeof(*con));
417 atomic_set(&con->nref, 1);
418 con->msgr = msgr;
419 mutex_init(&con->mutex);
420 INIT_LIST_HEAD(&con->out_queue);
421 INIT_LIST_HEAD(&con->out_sent);
422 INIT_DELAYED_WORK(&con->work, con_work);
423 }
424 EXPORT_SYMBOL(ceph_con_init);
425
426
427 /*
428 * We maintain a global counter to order connection attempts. Get
429 * a unique seq greater than @gt.
430 */
431 static u32 get_global_seq(struct ceph_messenger *msgr, u32 gt)
432 {
433 u32 ret;
434
435 spin_lock(&msgr->global_seq_lock);
436 if (msgr->global_seq < gt)
437 msgr->global_seq = gt;
438 ret = ++msgr->global_seq;
439 spin_unlock(&msgr->global_seq_lock);
440 return ret;
441 }
442
443
444 /*
445 * Prepare footer for currently outgoing message, and finish things
446 * off. Assumes out_kvec* are already valid.. we just add on to the end.
447 */
448 static void prepare_write_message_footer(struct ceph_connection *con, int v)
449 {
450 struct ceph_msg *m = con->out_msg;
451
452 dout("prepare_write_message_footer %p\n", con);
453 con->out_kvec_is_msg = true;
454 con->out_kvec[v].iov_base = &m->footer;
455 con->out_kvec[v].iov_len = sizeof(m->footer);
456 con->out_kvec_bytes += sizeof(m->footer);
457 con->out_kvec_left++;
458 con->out_more = m->more_to_follow;
459 con->out_msg_done = true;
460 }
461
462 /*
463 * Prepare headers for the next outgoing message.
464 */
465 static void prepare_write_message(struct ceph_connection *con)
466 {
467 struct ceph_msg *m;
468 int v = 0;
469
470 con->out_kvec_bytes = 0;
471 con->out_kvec_is_msg = true;
472 con->out_msg_done = false;
473
474 /* Sneak an ack in there first? If we can get it into the same
475 * TCP packet that's a good thing. */
476 if (con->in_seq > con->in_seq_acked) {
477 con->in_seq_acked = con->in_seq;
478 con->out_kvec[v].iov_base = &tag_ack;
479 con->out_kvec[v++].iov_len = 1;
480 con->out_temp_ack = cpu_to_le64(con->in_seq_acked);
481 con->out_kvec[v].iov_base = &con->out_temp_ack;
482 con->out_kvec[v++].iov_len = sizeof(con->out_temp_ack);
483 con->out_kvec_bytes = 1 + sizeof(con->out_temp_ack);
484 }
485
486 m = list_first_entry(&con->out_queue,
487 struct ceph_msg, list_head);
488 con->out_msg = m;
489 if (test_bit(LOSSYTX, &con->state)) {
490 list_del_init(&m->list_head);
491 } else {
492 /* put message on sent list */
493 ceph_msg_get(m);
494 list_move_tail(&m->list_head, &con->out_sent);
495 }
496
497 /*
498 * only assign outgoing seq # if we haven't sent this message
499 * yet. if it is requeued, resend with it's original seq.
500 */
501 if (m->needs_out_seq) {
502 m->hdr.seq = cpu_to_le64(++con->out_seq);
503 m->needs_out_seq = false;
504 }
505
506 dout("prepare_write_message %p seq %lld type %d len %d+%d+%d %d pgs\n",
507 m, con->out_seq, le16_to_cpu(m->hdr.type),
508 le32_to_cpu(m->hdr.front_len), le32_to_cpu(m->hdr.middle_len),
509 le32_to_cpu(m->hdr.data_len),
510 m->nr_pages);
511 BUG_ON(le32_to_cpu(m->hdr.front_len) != m->front.iov_len);
512
513 /* tag + hdr + front + middle */
514 con->out_kvec[v].iov_base = &tag_msg;
515 con->out_kvec[v++].iov_len = 1;
516 con->out_kvec[v].iov_base = &m->hdr;
517 con->out_kvec[v++].iov_len = sizeof(m->hdr);
518 con->out_kvec[v++] = m->front;
519 if (m->middle)
520 con->out_kvec[v++] = m->middle->vec;
521 con->out_kvec_left = v;
522 con->out_kvec_bytes += 1 + sizeof(m->hdr) + m->front.iov_len +
523 (m->middle ? m->middle->vec.iov_len : 0);
524 con->out_kvec_cur = con->out_kvec;
525
526 /* fill in crc (except data pages), footer */
527 con->out_msg->hdr.crc =
528 cpu_to_le32(crc32c(0, (void *)&m->hdr,
529 sizeof(m->hdr) - sizeof(m->hdr.crc)));
530 con->out_msg->footer.flags = CEPH_MSG_FOOTER_COMPLETE;
531 con->out_msg->footer.front_crc =
532 cpu_to_le32(crc32c(0, m->front.iov_base, m->front.iov_len));
533 if (m->middle)
534 con->out_msg->footer.middle_crc =
535 cpu_to_le32(crc32c(0, m->middle->vec.iov_base,
536 m->middle->vec.iov_len));
537 else
538 con->out_msg->footer.middle_crc = 0;
539 con->out_msg->footer.data_crc = 0;
540 dout("prepare_write_message front_crc %u data_crc %u\n",
541 le32_to_cpu(con->out_msg->footer.front_crc),
542 le32_to_cpu(con->out_msg->footer.middle_crc));
543
544 /* is there a data payload? */
545 if (le32_to_cpu(m->hdr.data_len) > 0) {
546 /* initialize page iterator */
547 con->out_msg_pos.page = 0;
548 if (m->pages)
549 con->out_msg_pos.page_pos = m->page_alignment;
550 else
551 con->out_msg_pos.page_pos = 0;
552 con->out_msg_pos.data_pos = 0;
553 con->out_msg_pos.did_page_crc = 0;
554 con->out_more = 1; /* data + footer will follow */
555 } else {
556 /* no, queue up footer too and be done */
557 prepare_write_message_footer(con, v);
558 }
559
560 set_bit(WRITE_PENDING, &con->state);
561 }
562
563 /*
564 * Prepare an ack.
565 */
566 static void prepare_write_ack(struct ceph_connection *con)
567 {
568 dout("prepare_write_ack %p %llu -> %llu\n", con,
569 con->in_seq_acked, con->in_seq);
570 con->in_seq_acked = con->in_seq;
571
572 con->out_kvec[0].iov_base = &tag_ack;
573 con->out_kvec[0].iov_len = 1;
574 con->out_temp_ack = cpu_to_le64(con->in_seq_acked);
575 con->out_kvec[1].iov_base = &con->out_temp_ack;
576 con->out_kvec[1].iov_len = sizeof(con->out_temp_ack);
577 con->out_kvec_left = 2;
578 con->out_kvec_bytes = 1 + sizeof(con->out_temp_ack);
579 con->out_kvec_cur = con->out_kvec;
580 con->out_more = 1; /* more will follow.. eventually.. */
581 set_bit(WRITE_PENDING, &con->state);
582 }
583
584 /*
585 * Prepare to write keepalive byte.
586 */
587 static void prepare_write_keepalive(struct ceph_connection *con)
588 {
589 dout("prepare_write_keepalive %p\n", con);
590 con->out_kvec[0].iov_base = &tag_keepalive;
591 con->out_kvec[0].iov_len = 1;
592 con->out_kvec_left = 1;
593 con->out_kvec_bytes = 1;
594 con->out_kvec_cur = con->out_kvec;
595 set_bit(WRITE_PENDING, &con->state);
596 }
597
598 /*
599 * Connection negotiation.
600 */
601
602 static int prepare_connect_authorizer(struct ceph_connection *con)
603 {
604 void *auth_buf;
605 int auth_len = 0;
606 int auth_protocol = 0;
607
608 mutex_unlock(&con->mutex);
609 if (con->ops->get_authorizer)
610 con->ops->get_authorizer(con, &auth_buf, &auth_len,
611 &auth_protocol, &con->auth_reply_buf,
612 &con->auth_reply_buf_len,
613 con->auth_retry);
614 mutex_lock(&con->mutex);
615
616 if (test_bit(CLOSED, &con->state) ||
617 test_bit(OPENING, &con->state))
618 return -EAGAIN;
619
620 con->out_connect.authorizer_protocol = cpu_to_le32(auth_protocol);
621 con->out_connect.authorizer_len = cpu_to_le32(auth_len);
622
623 if (auth_len) {
624 con->out_kvec[con->out_kvec_left].iov_base = auth_buf;
625 con->out_kvec[con->out_kvec_left].iov_len = auth_len;
626 con->out_kvec_left++;
627 con->out_kvec_bytes += auth_len;
628 }
629 return 0;
630 }
631
632 /*
633 * We connected to a peer and are saying hello.
634 */
635 static void prepare_write_banner(struct ceph_messenger *msgr,
636 struct ceph_connection *con)
637 {
638 int len = strlen(CEPH_BANNER);
639
640 con->out_kvec[0].iov_base = CEPH_BANNER;
641 con->out_kvec[0].iov_len = len;
642 con->out_kvec[1].iov_base = &msgr->my_enc_addr;
643 con->out_kvec[1].iov_len = sizeof(msgr->my_enc_addr);
644 con->out_kvec_left = 2;
645 con->out_kvec_bytes = len + sizeof(msgr->my_enc_addr);
646 con->out_kvec_cur = con->out_kvec;
647 con->out_more = 0;
648 set_bit(WRITE_PENDING, &con->state);
649 }
650
651 static int prepare_write_connect(struct ceph_messenger *msgr,
652 struct ceph_connection *con,
653 int after_banner)
654 {
655 unsigned global_seq = get_global_seq(con->msgr, 0);
656 int proto;
657
658 switch (con->peer_name.type) {
659 case CEPH_ENTITY_TYPE_MON:
660 proto = CEPH_MONC_PROTOCOL;
661 break;
662 case CEPH_ENTITY_TYPE_OSD:
663 proto = CEPH_OSDC_PROTOCOL;
664 break;
665 case CEPH_ENTITY_TYPE_MDS:
666 proto = CEPH_MDSC_PROTOCOL;
667 break;
668 default:
669 BUG();
670 }
671
672 dout("prepare_write_connect %p cseq=%d gseq=%d proto=%d\n", con,
673 con->connect_seq, global_seq, proto);
674
675 con->out_connect.features = cpu_to_le64(msgr->supported_features);
676 con->out_connect.host_type = cpu_to_le32(CEPH_ENTITY_TYPE_CLIENT);
677 con->out_connect.connect_seq = cpu_to_le32(con->connect_seq);
678 con->out_connect.global_seq = cpu_to_le32(global_seq);
679 con->out_connect.protocol_version = cpu_to_le32(proto);
680 con->out_connect.flags = 0;
681
682 if (!after_banner) {
683 con->out_kvec_left = 0;
684 con->out_kvec_bytes = 0;
685 }
686 con->out_kvec[con->out_kvec_left].iov_base = &con->out_connect;
687 con->out_kvec[con->out_kvec_left].iov_len = sizeof(con->out_connect);
688 con->out_kvec_left++;
689 con->out_kvec_bytes += sizeof(con->out_connect);
690 con->out_kvec_cur = con->out_kvec;
691 con->out_more = 0;
692 set_bit(WRITE_PENDING, &con->state);
693
694 return prepare_connect_authorizer(con);
695 }
696
697
698 /*
699 * write as much of pending kvecs to the socket as we can.
700 * 1 -> done
701 * 0 -> socket full, but more to do
702 * <0 -> error
703 */
704 static int write_partial_kvec(struct ceph_connection *con)
705 {
706 int ret;
707
708 dout("write_partial_kvec %p %d left\n", con, con->out_kvec_bytes);
709 while (con->out_kvec_bytes > 0) {
710 ret = ceph_tcp_sendmsg(con->sock, con->out_kvec_cur,
711 con->out_kvec_left, con->out_kvec_bytes,
712 con->out_more);
713 if (ret <= 0)
714 goto out;
715 con->out_kvec_bytes -= ret;
716 if (con->out_kvec_bytes == 0)
717 break; /* done */
718 while (ret > 0) {
719 if (ret >= con->out_kvec_cur->iov_len) {
720 ret -= con->out_kvec_cur->iov_len;
721 con->out_kvec_cur++;
722 con->out_kvec_left--;
723 } else {
724 con->out_kvec_cur->iov_len -= ret;
725 con->out_kvec_cur->iov_base += ret;
726 ret = 0;
727 break;
728 }
729 }
730 }
731 con->out_kvec_left = 0;
732 con->out_kvec_is_msg = false;
733 ret = 1;
734 out:
735 dout("write_partial_kvec %p %d left in %d kvecs ret = %d\n", con,
736 con->out_kvec_bytes, con->out_kvec_left, ret);
737 return ret; /* done! */
738 }
739
740 #ifdef CONFIG_BLOCK
741 static void init_bio_iter(struct bio *bio, struct bio **iter, int *seg)
742 {
743 if (!bio) {
744 *iter = NULL;
745 *seg = 0;
746 return;
747 }
748 *iter = bio;
749 *seg = bio->bi_idx;
750 }
751
752 static void iter_bio_next(struct bio **bio_iter, int *seg)
753 {
754 if (*bio_iter == NULL)
755 return;
756
757 BUG_ON(*seg >= (*bio_iter)->bi_vcnt);
758
759 (*seg)++;
760 if (*seg == (*bio_iter)->bi_vcnt)
761 init_bio_iter((*bio_iter)->bi_next, bio_iter, seg);
762 }
763 #endif
764
765 /*
766 * Write as much message data payload as we can. If we finish, queue
767 * up the footer.
768 * 1 -> done, footer is now queued in out_kvec[].
769 * 0 -> socket full, but more to do
770 * <0 -> error
771 */
772 static int write_partial_msg_pages(struct ceph_connection *con)
773 {
774 struct ceph_msg *msg = con->out_msg;
775 unsigned data_len = le32_to_cpu(msg->hdr.data_len);
776 size_t len;
777 int crc = con->msgr->nocrc;
778 int ret;
779 int total_max_write;
780 int in_trail = 0;
781 size_t trail_len = (msg->trail ? msg->trail->length : 0);
782
783 dout("write_partial_msg_pages %p msg %p page %d/%d offset %d\n",
784 con, con->out_msg, con->out_msg_pos.page, con->out_msg->nr_pages,
785 con->out_msg_pos.page_pos);
786
787 #ifdef CONFIG_BLOCK
788 if (msg->bio && !msg->bio_iter)
789 init_bio_iter(msg->bio, &msg->bio_iter, &msg->bio_seg);
790 #endif
791
792 while (data_len > con->out_msg_pos.data_pos) {
793 struct page *page = NULL;
794 void *kaddr = NULL;
795 int max_write = PAGE_SIZE;
796 int page_shift = 0;
797
798 total_max_write = data_len - trail_len -
799 con->out_msg_pos.data_pos;
800
801 /*
802 * if we are calculating the data crc (the default), we need
803 * to map the page. if our pages[] has been revoked, use the
804 * zero page.
805 */
806
807 /* have we reached the trail part of the data? */
808 if (con->out_msg_pos.data_pos >= data_len - trail_len) {
809 in_trail = 1;
810
811 total_max_write = data_len - con->out_msg_pos.data_pos;
812
813 page = list_first_entry(&msg->trail->head,
814 struct page, lru);
815 if (crc)
816 kaddr = kmap(page);
817 max_write = PAGE_SIZE;
818 } else if (msg->pages) {
819 page = msg->pages[con->out_msg_pos.page];
820 if (crc)
821 kaddr = kmap(page);
822 } else if (msg->pagelist) {
823 page = list_first_entry(&msg->pagelist->head,
824 struct page, lru);
825 if (crc)
826 kaddr = kmap(page);
827 #ifdef CONFIG_BLOCK
828 } else if (msg->bio) {
829 struct bio_vec *bv;
830
831 bv = bio_iovec_idx(msg->bio_iter, msg->bio_seg);
832 page = bv->bv_page;
833 page_shift = bv->bv_offset;
834 if (crc)
835 kaddr = kmap(page) + page_shift;
836 max_write = bv->bv_len;
837 #endif
838 } else {
839 page = con->msgr->zero_page;
840 if (crc)
841 kaddr = page_address(con->msgr->zero_page);
842 }
843 len = min_t(int, max_write - con->out_msg_pos.page_pos,
844 total_max_write);
845
846 if (crc && !con->out_msg_pos.did_page_crc) {
847 void *base = kaddr + con->out_msg_pos.page_pos;
848 u32 tmpcrc = le32_to_cpu(con->out_msg->footer.data_crc);
849
850 BUG_ON(kaddr == NULL);
851 con->out_msg->footer.data_crc =
852 cpu_to_le32(crc32c(tmpcrc, base, len));
853 con->out_msg_pos.did_page_crc = 1;
854 }
855 ret = kernel_sendpage(con->sock, page,
856 con->out_msg_pos.page_pos + page_shift,
857 len,
858 MSG_DONTWAIT | MSG_NOSIGNAL |
859 MSG_MORE);
860
861 if (crc &&
862 (msg->pages || msg->pagelist || msg->bio || in_trail))
863 kunmap(page);
864
865 if (ret == -EAGAIN)
866 ret = 0;
867 if (ret <= 0)
868 goto out;
869
870 con->out_msg_pos.data_pos += ret;
871 con->out_msg_pos.page_pos += ret;
872 if (ret == len) {
873 con->out_msg_pos.page_pos = 0;
874 con->out_msg_pos.page++;
875 con->out_msg_pos.did_page_crc = 0;
876 if (in_trail)
877 list_move_tail(&page->lru,
878 &msg->trail->head);
879 else if (msg->pagelist)
880 list_move_tail(&page->lru,
881 &msg->pagelist->head);
882 #ifdef CONFIG_BLOCK
883 else if (msg->bio)
884 iter_bio_next(&msg->bio_iter, &msg->bio_seg);
885 #endif
886 }
887 }
888
889 dout("write_partial_msg_pages %p msg %p done\n", con, msg);
890
891 /* prepare and queue up footer, too */
892 if (!crc)
893 con->out_msg->footer.flags |= CEPH_MSG_FOOTER_NOCRC;
894 con->out_kvec_bytes = 0;
895 con->out_kvec_left = 0;
896 con->out_kvec_cur = con->out_kvec;
897 prepare_write_message_footer(con, 0);
898 ret = 1;
899 out:
900 return ret;
901 }
902
903 /*
904 * write some zeros
905 */
906 static int write_partial_skip(struct ceph_connection *con)
907 {
908 int ret;
909
910 while (con->out_skip > 0) {
911 struct kvec iov = {
912 .iov_base = page_address(con->msgr->zero_page),
913 .iov_len = min(con->out_skip, (int)PAGE_CACHE_SIZE)
914 };
915
916 ret = ceph_tcp_sendmsg(con->sock, &iov, 1, iov.iov_len, 1);
917 if (ret <= 0)
918 goto out;
919 con->out_skip -= ret;
920 }
921 ret = 1;
922 out:
923 return ret;
924 }
925
926 /*
927 * Prepare to read connection handshake, or an ack.
928 */
929 static void prepare_read_banner(struct ceph_connection *con)
930 {
931 dout("prepare_read_banner %p\n", con);
932 con->in_base_pos = 0;
933 }
934
935 static void prepare_read_connect(struct ceph_connection *con)
936 {
937 dout("prepare_read_connect %p\n", con);
938 con->in_base_pos = 0;
939 }
940
941 static void prepare_read_ack(struct ceph_connection *con)
942 {
943 dout("prepare_read_ack %p\n", con);
944 con->in_base_pos = 0;
945 }
946
947 static void prepare_read_tag(struct ceph_connection *con)
948 {
949 dout("prepare_read_tag %p\n", con);
950 con->in_base_pos = 0;
951 con->in_tag = CEPH_MSGR_TAG_READY;
952 }
953
954 /*
955 * Prepare to read a message.
956 */
957 static int prepare_read_message(struct ceph_connection *con)
958 {
959 dout("prepare_read_message %p\n", con);
960 BUG_ON(con->in_msg != NULL);
961 con->in_base_pos = 0;
962 con->in_front_crc = con->in_middle_crc = con->in_data_crc = 0;
963 return 0;
964 }
965
966
967 static int read_partial(struct ceph_connection *con,
968 int *to, int size, void *object)
969 {
970 *to += size;
971 while (con->in_base_pos < *to) {
972 int left = *to - con->in_base_pos;
973 int have = size - left;
974 int ret = ceph_tcp_recvmsg(con->sock, object + have, left);
975 if (ret <= 0)
976 return ret;
977 con->in_base_pos += ret;
978 }
979 return 1;
980 }
981
982
983 /*
984 * Read all or part of the connect-side handshake on a new connection
985 */
986 static int read_partial_banner(struct ceph_connection *con)
987 {
988 int ret, to = 0;
989
990 dout("read_partial_banner %p at %d\n", con, con->in_base_pos);
991
992 /* peer's banner */
993 ret = read_partial(con, &to, strlen(CEPH_BANNER), con->in_banner);
994 if (ret <= 0)
995 goto out;
996 ret = read_partial(con, &to, sizeof(con->actual_peer_addr),
997 &con->actual_peer_addr);
998 if (ret <= 0)
999 goto out;
1000 ret = read_partial(con, &to, sizeof(con->peer_addr_for_me),
1001 &con->peer_addr_for_me);
1002 if (ret <= 0)
1003 goto out;
1004 out:
1005 return ret;
1006 }
1007
1008 static int read_partial_connect(struct ceph_connection *con)
1009 {
1010 int ret, to = 0;
1011
1012 dout("read_partial_connect %p at %d\n", con, con->in_base_pos);
1013
1014 ret = read_partial(con, &to, sizeof(con->in_reply), &con->in_reply);
1015 if (ret <= 0)
1016 goto out;
1017 ret = read_partial(con, &to, le32_to_cpu(con->in_reply.authorizer_len),
1018 con->auth_reply_buf);
1019 if (ret <= 0)
1020 goto out;
1021
1022 dout("read_partial_connect %p tag %d, con_seq = %u, g_seq = %u\n",
1023 con, (int)con->in_reply.tag,
1024 le32_to_cpu(con->in_reply.connect_seq),
1025 le32_to_cpu(con->in_reply.global_seq));
1026 out:
1027 return ret;
1028
1029 }
1030
1031 /*
1032 * Verify the hello banner looks okay.
1033 */
1034 static int verify_hello(struct ceph_connection *con)
1035 {
1036 if (memcmp(con->in_banner, CEPH_BANNER, strlen(CEPH_BANNER))) {
1037 pr_err("connect to %s got bad banner\n",
1038 ceph_pr_addr(&con->peer_addr.in_addr));
1039 con->error_msg = "protocol error, bad banner";
1040 return -1;
1041 }
1042 return 0;
1043 }
1044
1045 static bool addr_is_blank(struct sockaddr_storage *ss)
1046 {
1047 switch (ss->ss_family) {
1048 case AF_INET:
1049 return ((struct sockaddr_in *)ss)->sin_addr.s_addr == 0;
1050 case AF_INET6:
1051 return
1052 ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[0] == 0 &&
1053 ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[1] == 0 &&
1054 ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[2] == 0 &&
1055 ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[3] == 0;
1056 }
1057 return false;
1058 }
1059
1060 static int addr_port(struct sockaddr_storage *ss)
1061 {
1062 switch (ss->ss_family) {
1063 case AF_INET:
1064 return ntohs(((struct sockaddr_in *)ss)->sin_port);
1065 case AF_INET6:
1066 return ntohs(((struct sockaddr_in6 *)ss)->sin6_port);
1067 }
1068 return 0;
1069 }
1070
1071 static void addr_set_port(struct sockaddr_storage *ss, int p)
1072 {
1073 switch (ss->ss_family) {
1074 case AF_INET:
1075 ((struct sockaddr_in *)ss)->sin_port = htons(p);
1076 break;
1077 case AF_INET6:
1078 ((struct sockaddr_in6 *)ss)->sin6_port = htons(p);
1079 break;
1080 }
1081 }
1082
1083 /*
1084 * Parse an ip[:port] list into an addr array. Use the default
1085 * monitor port if a port isn't specified.
1086 */
1087 int ceph_parse_ips(const char *c, const char *end,
1088 struct ceph_entity_addr *addr,
1089 int max_count, int *count)
1090 {
1091 int i;
1092 const char *p = c;
1093
1094 dout("parse_ips on '%.*s'\n", (int)(end-c), c);
1095 for (i = 0; i < max_count; i++) {
1096 const char *ipend;
1097 struct sockaddr_storage *ss = &addr[i].in_addr;
1098 struct sockaddr_in *in4 = (void *)ss;
1099 struct sockaddr_in6 *in6 = (void *)ss;
1100 int port;
1101 char delim = ',';
1102
1103 if (*p == '[') {
1104 delim = ']';
1105 p++;
1106 }
1107
1108 memset(ss, 0, sizeof(*ss));
1109 if (in4_pton(p, end - p, (u8 *)&in4->sin_addr.s_addr,
1110 delim, &ipend))
1111 ss->ss_family = AF_INET;
1112 else if (in6_pton(p, end - p, (u8 *)&in6->sin6_addr.s6_addr,
1113 delim, &ipend))
1114 ss->ss_family = AF_INET6;
1115 else
1116 goto bad;
1117 p = ipend;
1118
1119 if (delim == ']') {
1120 if (*p != ']') {
1121 dout("missing matching ']'\n");
1122 goto bad;
1123 }
1124 p++;
1125 }
1126
1127 /* port? */
1128 if (p < end && *p == ':') {
1129 port = 0;
1130 p++;
1131 while (p < end && *p >= '0' && *p <= '9') {
1132 port = (port * 10) + (*p - '0');
1133 p++;
1134 }
1135 if (port > 65535 || port == 0)
1136 goto bad;
1137 } else {
1138 port = CEPH_MON_PORT;
1139 }
1140
1141 addr_set_port(ss, port);
1142
1143 dout("parse_ips got %s\n", ceph_pr_addr(ss));
1144
1145 if (p == end)
1146 break;
1147 if (*p != ',')
1148 goto bad;
1149 p++;
1150 }
1151
1152 if (p != end)
1153 goto bad;
1154
1155 if (count)
1156 *count = i + 1;
1157 return 0;
1158
1159 bad:
1160 pr_err("parse_ips bad ip '%.*s'\n", (int)(end - c), c);
1161 return -EINVAL;
1162 }
1163 EXPORT_SYMBOL(ceph_parse_ips);
1164
1165 static int process_banner(struct ceph_connection *con)
1166 {
1167 dout("process_banner on %p\n", con);
1168
1169 if (verify_hello(con) < 0)
1170 return -1;
1171
1172 ceph_decode_addr(&con->actual_peer_addr);
1173 ceph_decode_addr(&con->peer_addr_for_me);
1174
1175 /*
1176 * Make sure the other end is who we wanted. note that the other
1177 * end may not yet know their ip address, so if it's 0.0.0.0, give
1178 * them the benefit of the doubt.
1179 */
1180 if (memcmp(&con->peer_addr, &con->actual_peer_addr,
1181 sizeof(con->peer_addr)) != 0 &&
1182 !(addr_is_blank(&con->actual_peer_addr.in_addr) &&
1183 con->actual_peer_addr.nonce == con->peer_addr.nonce)) {
1184 pr_warning("wrong peer, want %s/%d, got %s/%d\n",
1185 ceph_pr_addr(&con->peer_addr.in_addr),
1186 (int)le32_to_cpu(con->peer_addr.nonce),
1187 ceph_pr_addr(&con->actual_peer_addr.in_addr),
1188 (int)le32_to_cpu(con->actual_peer_addr.nonce));
1189 con->error_msg = "wrong peer at address";
1190 return -1;
1191 }
1192
1193 /*
1194 * did we learn our address?
1195 */
1196 if (addr_is_blank(&con->msgr->inst.addr.in_addr)) {
1197 int port = addr_port(&con->msgr->inst.addr.in_addr);
1198
1199 memcpy(&con->msgr->inst.addr.in_addr,
1200 &con->peer_addr_for_me.in_addr,
1201 sizeof(con->peer_addr_for_me.in_addr));
1202 addr_set_port(&con->msgr->inst.addr.in_addr, port);
1203 encode_my_addr(con->msgr);
1204 dout("process_banner learned my addr is %s\n",
1205 ceph_pr_addr(&con->msgr->inst.addr.in_addr));
1206 }
1207
1208 set_bit(NEGOTIATING, &con->state);
1209 prepare_read_connect(con);
1210 return 0;
1211 }
1212
1213 static void fail_protocol(struct ceph_connection *con)
1214 {
1215 reset_connection(con);
1216 set_bit(CLOSED, &con->state); /* in case there's queued work */
1217
1218 mutex_unlock(&con->mutex);
1219 if (con->ops->bad_proto)
1220 con->ops->bad_proto(con);
1221 mutex_lock(&con->mutex);
1222 }
1223
1224 static int process_connect(struct ceph_connection *con)
1225 {
1226 u64 sup_feat = con->msgr->supported_features;
1227 u64 req_feat = con->msgr->required_features;
1228 u64 server_feat = le64_to_cpu(con->in_reply.features);
1229 int ret;
1230
1231 dout("process_connect on %p tag %d\n", con, (int)con->in_tag);
1232
1233 switch (con->in_reply.tag) {
1234 case CEPH_MSGR_TAG_FEATURES:
1235 pr_err("%s%lld %s feature set mismatch,"
1236 " my %llx < server's %llx, missing %llx\n",
1237 ENTITY_NAME(con->peer_name),
1238 ceph_pr_addr(&con->peer_addr.in_addr),
1239 sup_feat, server_feat, server_feat & ~sup_feat);
1240 con->error_msg = "missing required protocol features";
1241 fail_protocol(con);
1242 return -1;
1243
1244 case CEPH_MSGR_TAG_BADPROTOVER:
1245 pr_err("%s%lld %s protocol version mismatch,"
1246 " my %d != server's %d\n",
1247 ENTITY_NAME(con->peer_name),
1248 ceph_pr_addr(&con->peer_addr.in_addr),
1249 le32_to_cpu(con->out_connect.protocol_version),
1250 le32_to_cpu(con->in_reply.protocol_version));
1251 con->error_msg = "protocol version mismatch";
1252 fail_protocol(con);
1253 return -1;
1254
1255 case CEPH_MSGR_TAG_BADAUTHORIZER:
1256 con->auth_retry++;
1257 dout("process_connect %p got BADAUTHORIZER attempt %d\n", con,
1258 con->auth_retry);
1259 if (con->auth_retry == 2) {
1260 con->error_msg = "connect authorization failure";
1261 return -1;
1262 }
1263 con->auth_retry = 1;
1264 ret = prepare_write_connect(con->msgr, con, 0);
1265 if (ret < 0)
1266 return ret;
1267 prepare_read_connect(con);
1268 break;
1269
1270 case CEPH_MSGR_TAG_RESETSESSION:
1271 /*
1272 * If we connected with a large connect_seq but the peer
1273 * has no record of a session with us (no connection, or
1274 * connect_seq == 0), they will send RESETSESION to indicate
1275 * that they must have reset their session, and may have
1276 * dropped messages.
1277 */
1278 dout("process_connect got RESET peer seq %u\n",
1279 le32_to_cpu(con->in_connect.connect_seq));
1280 pr_err("%s%lld %s connection reset\n",
1281 ENTITY_NAME(con->peer_name),
1282 ceph_pr_addr(&con->peer_addr.in_addr));
1283 reset_connection(con);
1284 prepare_write_connect(con->msgr, con, 0);
1285 prepare_read_connect(con);
1286
1287 /* Tell ceph about it. */
1288 mutex_unlock(&con->mutex);
1289 pr_info("reset on %s%lld\n", ENTITY_NAME(con->peer_name));
1290 if (con->ops->peer_reset)
1291 con->ops->peer_reset(con);
1292 mutex_lock(&con->mutex);
1293 if (test_bit(CLOSED, &con->state) ||
1294 test_bit(OPENING, &con->state))
1295 return -EAGAIN;
1296 break;
1297
1298 case CEPH_MSGR_TAG_RETRY_SESSION:
1299 /*
1300 * If we sent a smaller connect_seq than the peer has, try
1301 * again with a larger value.
1302 */
1303 dout("process_connect got RETRY my seq = %u, peer_seq = %u\n",
1304 le32_to_cpu(con->out_connect.connect_seq),
1305 le32_to_cpu(con->in_connect.connect_seq));
1306 con->connect_seq = le32_to_cpu(con->in_connect.connect_seq);
1307 prepare_write_connect(con->msgr, con, 0);
1308 prepare_read_connect(con);
1309 break;
1310
1311 case CEPH_MSGR_TAG_RETRY_GLOBAL:
1312 /*
1313 * If we sent a smaller global_seq than the peer has, try
1314 * again with a larger value.
1315 */
1316 dout("process_connect got RETRY_GLOBAL my %u peer_gseq %u\n",
1317 con->peer_global_seq,
1318 le32_to_cpu(con->in_connect.global_seq));
1319 get_global_seq(con->msgr,
1320 le32_to_cpu(con->in_connect.global_seq));
1321 prepare_write_connect(con->msgr, con, 0);
1322 prepare_read_connect(con);
1323 break;
1324
1325 case CEPH_MSGR_TAG_READY:
1326 if (req_feat & ~server_feat) {
1327 pr_err("%s%lld %s protocol feature mismatch,"
1328 " my required %llx > server's %llx, need %llx\n",
1329 ENTITY_NAME(con->peer_name),
1330 ceph_pr_addr(&con->peer_addr.in_addr),
1331 req_feat, server_feat, req_feat & ~server_feat);
1332 con->error_msg = "missing required protocol features";
1333 fail_protocol(con);
1334 return -1;
1335 }
1336 clear_bit(CONNECTING, &con->state);
1337 con->peer_global_seq = le32_to_cpu(con->in_reply.global_seq);
1338 con->connect_seq++;
1339 con->peer_features = server_feat;
1340 dout("process_connect got READY gseq %d cseq %d (%d)\n",
1341 con->peer_global_seq,
1342 le32_to_cpu(con->in_reply.connect_seq),
1343 con->connect_seq);
1344 WARN_ON(con->connect_seq !=
1345 le32_to_cpu(con->in_reply.connect_seq));
1346
1347 if (con->in_reply.flags & CEPH_MSG_CONNECT_LOSSY)
1348 set_bit(LOSSYTX, &con->state);
1349
1350 prepare_read_tag(con);
1351 break;
1352
1353 case CEPH_MSGR_TAG_WAIT:
1354 /*
1355 * If there is a connection race (we are opening
1356 * connections to each other), one of us may just have
1357 * to WAIT. This shouldn't happen if we are the
1358 * client.
1359 */
1360 pr_err("process_connect got WAIT as client\n");
1361 con->error_msg = "protocol error, got WAIT as client";
1362 return -1;
1363
1364 default:
1365 pr_err("connect protocol error, will retry\n");
1366 con->error_msg = "protocol error, garbage tag during connect";
1367 return -1;
1368 }
1369 return 0;
1370 }
1371
1372
1373 /*
1374 * read (part of) an ack
1375 */
1376 static int read_partial_ack(struct ceph_connection *con)
1377 {
1378 int to = 0;
1379
1380 return read_partial(con, &to, sizeof(con->in_temp_ack),
1381 &con->in_temp_ack);
1382 }
1383
1384
1385 /*
1386 * We can finally discard anything that's been acked.
1387 */
1388 static void process_ack(struct ceph_connection *con)
1389 {
1390 struct ceph_msg *m;
1391 u64 ack = le64_to_cpu(con->in_temp_ack);
1392 u64 seq;
1393
1394 while (!list_empty(&con->out_sent)) {
1395 m = list_first_entry(&con->out_sent, struct ceph_msg,
1396 list_head);
1397 seq = le64_to_cpu(m->hdr.seq);
1398 if (seq > ack)
1399 break;
1400 dout("got ack for seq %llu type %d at %p\n", seq,
1401 le16_to_cpu(m->hdr.type), m);
1402 ceph_msg_remove(m);
1403 }
1404 prepare_read_tag(con);
1405 }
1406
1407
1408
1409
1410 static int read_partial_message_section(struct ceph_connection *con,
1411 struct kvec *section,
1412 unsigned int sec_len, u32 *crc)
1413 {
1414 int ret, left;
1415
1416 BUG_ON(!section);
1417
1418 while (section->iov_len < sec_len) {
1419 BUG_ON(section->iov_base == NULL);
1420 left = sec_len - section->iov_len;
1421 ret = ceph_tcp_recvmsg(con->sock, (char *)section->iov_base +
1422 section->iov_len, left);
1423 if (ret <= 0)
1424 return ret;
1425 section->iov_len += ret;
1426 if (section->iov_len == sec_len)
1427 *crc = crc32c(0, section->iov_base,
1428 section->iov_len);
1429 }
1430
1431 return 1;
1432 }
1433
1434 static struct ceph_msg *ceph_alloc_msg(struct ceph_connection *con,
1435 struct ceph_msg_header *hdr,
1436 int *skip);
1437
1438
1439 static int read_partial_message_pages(struct ceph_connection *con,
1440 struct page **pages,
1441 unsigned data_len, int datacrc)
1442 {
1443 void *p;
1444 int ret;
1445 int left;
1446
1447 left = min((int)(data_len - con->in_msg_pos.data_pos),
1448 (int)(PAGE_SIZE - con->in_msg_pos.page_pos));
1449 /* (page) data */
1450 BUG_ON(pages == NULL);
1451 p = kmap(pages[con->in_msg_pos.page]);
1452 ret = ceph_tcp_recvmsg(con->sock, p + con->in_msg_pos.page_pos,
1453 left);
1454 if (ret > 0 && datacrc)
1455 con->in_data_crc =
1456 crc32c(con->in_data_crc,
1457 p + con->in_msg_pos.page_pos, ret);
1458 kunmap(pages[con->in_msg_pos.page]);
1459 if (ret <= 0)
1460 return ret;
1461 con->in_msg_pos.data_pos += ret;
1462 con->in_msg_pos.page_pos += ret;
1463 if (con->in_msg_pos.page_pos == PAGE_SIZE) {
1464 con->in_msg_pos.page_pos = 0;
1465 con->in_msg_pos.page++;
1466 }
1467
1468 return ret;
1469 }
1470
1471 #ifdef CONFIG_BLOCK
1472 static int read_partial_message_bio(struct ceph_connection *con,
1473 struct bio **bio_iter, int *bio_seg,
1474 unsigned data_len, int datacrc)
1475 {
1476 struct bio_vec *bv = bio_iovec_idx(*bio_iter, *bio_seg);
1477 void *p;
1478 int ret, left;
1479
1480 if (IS_ERR(bv))
1481 return PTR_ERR(bv);
1482
1483 left = min((int)(data_len - con->in_msg_pos.data_pos),
1484 (int)(bv->bv_len - con->in_msg_pos.page_pos));
1485
1486 p = kmap(bv->bv_page) + bv->bv_offset;
1487
1488 ret = ceph_tcp_recvmsg(con->sock, p + con->in_msg_pos.page_pos,
1489 left);
1490 if (ret > 0 && datacrc)
1491 con->in_data_crc =
1492 crc32c(con->in_data_crc,
1493 p + con->in_msg_pos.page_pos, ret);
1494 kunmap(bv->bv_page);
1495 if (ret <= 0)
1496 return ret;
1497 con->in_msg_pos.data_pos += ret;
1498 con->in_msg_pos.page_pos += ret;
1499 if (con->in_msg_pos.page_pos == bv->bv_len) {
1500 con->in_msg_pos.page_pos = 0;
1501 iter_bio_next(bio_iter, bio_seg);
1502 }
1503
1504 return ret;
1505 }
1506 #endif
1507
1508 /*
1509 * read (part of) a message.
1510 */
1511 static int read_partial_message(struct ceph_connection *con)
1512 {
1513 struct ceph_msg *m = con->in_msg;
1514 int ret;
1515 int to, left;
1516 unsigned front_len, middle_len, data_len;
1517 int datacrc = con->msgr->nocrc;
1518 int skip;
1519 u64 seq;
1520
1521 dout("read_partial_message con %p msg %p\n", con, m);
1522
1523 /* header */
1524 while (con->in_base_pos < sizeof(con->in_hdr)) {
1525 left = sizeof(con->in_hdr) - con->in_base_pos;
1526 ret = ceph_tcp_recvmsg(con->sock,
1527 (char *)&con->in_hdr + con->in_base_pos,
1528 left);
1529 if (ret <= 0)
1530 return ret;
1531 con->in_base_pos += ret;
1532 if (con->in_base_pos == sizeof(con->in_hdr)) {
1533 u32 crc = crc32c(0, (void *)&con->in_hdr,
1534 sizeof(con->in_hdr) - sizeof(con->in_hdr.crc));
1535 if (crc != le32_to_cpu(con->in_hdr.crc)) {
1536 pr_err("read_partial_message bad hdr "
1537 " crc %u != expected %u\n",
1538 crc, con->in_hdr.crc);
1539 return -EBADMSG;
1540 }
1541 }
1542 }
1543 front_len = le32_to_cpu(con->in_hdr.front_len);
1544 if (front_len > CEPH_MSG_MAX_FRONT_LEN)
1545 return -EIO;
1546 middle_len = le32_to_cpu(con->in_hdr.middle_len);
1547 if (middle_len > CEPH_MSG_MAX_DATA_LEN)
1548 return -EIO;
1549 data_len = le32_to_cpu(con->in_hdr.data_len);
1550 if (data_len > CEPH_MSG_MAX_DATA_LEN)
1551 return -EIO;
1552
1553 /* verify seq# */
1554 seq = le64_to_cpu(con->in_hdr.seq);
1555 if ((s64)seq - (s64)con->in_seq < 1) {
1556 pr_info("skipping %s%lld %s seq %lld expected %lld\n",
1557 ENTITY_NAME(con->peer_name),
1558 ceph_pr_addr(&con->peer_addr.in_addr),
1559 seq, con->in_seq + 1);
1560 con->in_base_pos = -front_len - middle_len - data_len -
1561 sizeof(m->footer);
1562 con->in_tag = CEPH_MSGR_TAG_READY;
1563 return 0;
1564 } else if ((s64)seq - (s64)con->in_seq > 1) {
1565 pr_err("read_partial_message bad seq %lld expected %lld\n",
1566 seq, con->in_seq + 1);
1567 con->error_msg = "bad message sequence # for incoming message";
1568 return -EBADMSG;
1569 }
1570
1571 /* allocate message? */
1572 if (!con->in_msg) {
1573 dout("got hdr type %d front %d data %d\n", con->in_hdr.type,
1574 con->in_hdr.front_len, con->in_hdr.data_len);
1575 skip = 0;
1576 con->in_msg = ceph_alloc_msg(con, &con->in_hdr, &skip);
1577 if (skip) {
1578 /* skip this message */
1579 dout("alloc_msg said skip message\n");
1580 BUG_ON(con->in_msg);
1581 con->in_base_pos = -front_len - middle_len - data_len -
1582 sizeof(m->footer);
1583 con->in_tag = CEPH_MSGR_TAG_READY;
1584 con->in_seq++;
1585 return 0;
1586 }
1587 if (!con->in_msg) {
1588 con->error_msg =
1589 "error allocating memory for incoming message";
1590 return -ENOMEM;
1591 }
1592 m = con->in_msg;
1593 m->front.iov_len = 0; /* haven't read it yet */
1594 if (m->middle)
1595 m->middle->vec.iov_len = 0;
1596
1597 con->in_msg_pos.page = 0;
1598 if (m->pages)
1599 con->in_msg_pos.page_pos = m->page_alignment;
1600 else
1601 con->in_msg_pos.page_pos = 0;
1602 con->in_msg_pos.data_pos = 0;
1603 }
1604
1605 /* front */
1606 ret = read_partial_message_section(con, &m->front, front_len,
1607 &con->in_front_crc);
1608 if (ret <= 0)
1609 return ret;
1610
1611 /* middle */
1612 if (m->middle) {
1613 ret = read_partial_message_section(con, &m->middle->vec,
1614 middle_len,
1615 &con->in_middle_crc);
1616 if (ret <= 0)
1617 return ret;
1618 }
1619 #ifdef CONFIG_BLOCK
1620 if (m->bio && !m->bio_iter)
1621 init_bio_iter(m->bio, &m->bio_iter, &m->bio_seg);
1622 #endif
1623
1624 /* (page) data */
1625 while (con->in_msg_pos.data_pos < data_len) {
1626 if (m->pages) {
1627 ret = read_partial_message_pages(con, m->pages,
1628 data_len, datacrc);
1629 if (ret <= 0)
1630 return ret;
1631 #ifdef CONFIG_BLOCK
1632 } else if (m->bio) {
1633
1634 ret = read_partial_message_bio(con,
1635 &m->bio_iter, &m->bio_seg,
1636 data_len, datacrc);
1637 if (ret <= 0)
1638 return ret;
1639 #endif
1640 } else {
1641 BUG_ON(1);
1642 }
1643 }
1644
1645 /* footer */
1646 to = sizeof(m->hdr) + sizeof(m->footer);
1647 while (con->in_base_pos < to) {
1648 left = to - con->in_base_pos;
1649 ret = ceph_tcp_recvmsg(con->sock, (char *)&m->footer +
1650 (con->in_base_pos - sizeof(m->hdr)),
1651 left);
1652 if (ret <= 0)
1653 return ret;
1654 con->in_base_pos += ret;
1655 }
1656 dout("read_partial_message got msg %p %d (%u) + %d (%u) + %d (%u)\n",
1657 m, front_len, m->footer.front_crc, middle_len,
1658 m->footer.middle_crc, data_len, m->footer.data_crc);
1659
1660 /* crc ok? */
1661 if (con->in_front_crc != le32_to_cpu(m->footer.front_crc)) {
1662 pr_err("read_partial_message %p front crc %u != exp. %u\n",
1663 m, con->in_front_crc, m->footer.front_crc);
1664 return -EBADMSG;
1665 }
1666 if (con->in_middle_crc != le32_to_cpu(m->footer.middle_crc)) {
1667 pr_err("read_partial_message %p middle crc %u != exp %u\n",
1668 m, con->in_middle_crc, m->footer.middle_crc);
1669 return -EBADMSG;
1670 }
1671 if (datacrc &&
1672 (m->footer.flags & CEPH_MSG_FOOTER_NOCRC) == 0 &&
1673 con->in_data_crc != le32_to_cpu(m->footer.data_crc)) {
1674 pr_err("read_partial_message %p data crc %u != exp. %u\n", m,
1675 con->in_data_crc, le32_to_cpu(m->footer.data_crc));
1676 return -EBADMSG;
1677 }
1678
1679 return 1; /* done! */
1680 }
1681
1682 /*
1683 * Process message. This happens in the worker thread. The callback should
1684 * be careful not to do anything that waits on other incoming messages or it
1685 * may deadlock.
1686 */
1687 static void process_message(struct ceph_connection *con)
1688 {
1689 struct ceph_msg *msg;
1690
1691 msg = con->in_msg;
1692 con->in_msg = NULL;
1693
1694 /* if first message, set peer_name */
1695 if (con->peer_name.type == 0)
1696 con->peer_name = msg->hdr.src;
1697
1698 con->in_seq++;
1699 mutex_unlock(&con->mutex);
1700
1701 dout("===== %p %llu from %s%lld %d=%s len %d+%d (%u %u %u) =====\n",
1702 msg, le64_to_cpu(msg->hdr.seq),
1703 ENTITY_NAME(msg->hdr.src),
1704 le16_to_cpu(msg->hdr.type),
1705 ceph_msg_type_name(le16_to_cpu(msg->hdr.type)),
1706 le32_to_cpu(msg->hdr.front_len),
1707 le32_to_cpu(msg->hdr.data_len),
1708 con->in_front_crc, con->in_middle_crc, con->in_data_crc);
1709 con->ops->dispatch(con, msg);
1710
1711 mutex_lock(&con->mutex);
1712 prepare_read_tag(con);
1713 }
1714
1715
1716 /*
1717 * Write something to the socket. Called in a worker thread when the
1718 * socket appears to be writeable and we have something ready to send.
1719 */
1720 static int try_write(struct ceph_connection *con)
1721 {
1722 struct ceph_messenger *msgr = con->msgr;
1723 int ret = 1;
1724
1725 dout("try_write start %p state %lu nref %d\n", con, con->state,
1726 atomic_read(&con->nref));
1727
1728 more:
1729 dout("try_write out_kvec_bytes %d\n", con->out_kvec_bytes);
1730
1731 /* open the socket first? */
1732 if (con->sock == NULL) {
1733 prepare_write_banner(msgr, con);
1734 prepare_write_connect(msgr, con, 1);
1735 prepare_read_banner(con);
1736 set_bit(CONNECTING, &con->state);
1737 clear_bit(NEGOTIATING, &con->state);
1738
1739 BUG_ON(con->in_msg);
1740 con->in_tag = CEPH_MSGR_TAG_READY;
1741 dout("try_write initiating connect on %p new state %lu\n",
1742 con, con->state);
1743 con->sock = ceph_tcp_connect(con);
1744 if (IS_ERR(con->sock)) {
1745 con->sock = NULL;
1746 con->error_msg = "connect error";
1747 ret = -1;
1748 goto out;
1749 }
1750 }
1751
1752 more_kvec:
1753 /* kvec data queued? */
1754 if (con->out_skip) {
1755 ret = write_partial_skip(con);
1756 if (ret <= 0)
1757 goto out;
1758 }
1759 if (con->out_kvec_left) {
1760 ret = write_partial_kvec(con);
1761 if (ret <= 0)
1762 goto out;
1763 }
1764
1765 /* msg pages? */
1766 if (con->out_msg) {
1767 if (con->out_msg_done) {
1768 ceph_msg_put(con->out_msg);
1769 con->out_msg = NULL; /* we're done with this one */
1770 goto do_next;
1771 }
1772
1773 ret = write_partial_msg_pages(con);
1774 if (ret == 1)
1775 goto more_kvec; /* we need to send the footer, too! */
1776 if (ret == 0)
1777 goto out;
1778 if (ret < 0) {
1779 dout("try_write write_partial_msg_pages err %d\n",
1780 ret);
1781 goto out;
1782 }
1783 }
1784
1785 do_next:
1786 if (!test_bit(CONNECTING, &con->state)) {
1787 /* is anything else pending? */
1788 if (!list_empty(&con->out_queue)) {
1789 prepare_write_message(con);
1790 goto more;
1791 }
1792 if (con->in_seq > con->in_seq_acked) {
1793 prepare_write_ack(con);
1794 goto more;
1795 }
1796 if (test_and_clear_bit(KEEPALIVE_PENDING, &con->state)) {
1797 prepare_write_keepalive(con);
1798 goto more;
1799 }
1800 }
1801
1802 /* Nothing to do! */
1803 clear_bit(WRITE_PENDING, &con->state);
1804 dout("try_write nothing else to write.\n");
1805 ret = 0;
1806 out:
1807 dout("try_write done on %p ret %d\n", con, ret);
1808 return ret;
1809 }
1810
1811
1812
1813 /*
1814 * Read what we can from the socket.
1815 */
1816 static int try_read(struct ceph_connection *con)
1817 {
1818 int ret = -1;
1819
1820 if (!con->sock)
1821 return 0;
1822
1823 if (test_bit(STANDBY, &con->state))
1824 return 0;
1825
1826 dout("try_read start on %p\n", con);
1827
1828 more:
1829 dout("try_read tag %d in_base_pos %d\n", (int)con->in_tag,
1830 con->in_base_pos);
1831
1832 /*
1833 * process_connect and process_message drop and re-take
1834 * con->mutex. make sure we handle a racing close or reopen.
1835 */
1836 if (test_bit(CLOSED, &con->state) ||
1837 test_bit(OPENING, &con->state)) {
1838 ret = -EAGAIN;
1839 goto out;
1840 }
1841
1842 if (test_bit(CONNECTING, &con->state)) {
1843 if (!test_bit(NEGOTIATING, &con->state)) {
1844 dout("try_read connecting\n");
1845 ret = read_partial_banner(con);
1846 if (ret <= 0)
1847 goto out;
1848 ret = process_banner(con);
1849 if (ret < 0)
1850 goto out;
1851 }
1852 ret = read_partial_connect(con);
1853 if (ret <= 0)
1854 goto out;
1855 ret = process_connect(con);
1856 if (ret < 0)
1857 goto out;
1858 goto more;
1859 }
1860
1861 if (con->in_base_pos < 0) {
1862 /*
1863 * skipping + discarding content.
1864 *
1865 * FIXME: there must be a better way to do this!
1866 */
1867 static char buf[1024];
1868 int skip = min(1024, -con->in_base_pos);
1869 dout("skipping %d / %d bytes\n", skip, -con->in_base_pos);
1870 ret = ceph_tcp_recvmsg(con->sock, buf, skip);
1871 if (ret <= 0)
1872 goto out;
1873 con->in_base_pos += ret;
1874 if (con->in_base_pos)
1875 goto more;
1876 }
1877 if (con->in_tag == CEPH_MSGR_TAG_READY) {
1878 /*
1879 * what's next?
1880 */
1881 ret = ceph_tcp_recvmsg(con->sock, &con->in_tag, 1);
1882 if (ret <= 0)
1883 goto out;
1884 dout("try_read got tag %d\n", (int)con->in_tag);
1885 switch (con->in_tag) {
1886 case CEPH_MSGR_TAG_MSG:
1887 prepare_read_message(con);
1888 break;
1889 case CEPH_MSGR_TAG_ACK:
1890 prepare_read_ack(con);
1891 break;
1892 case CEPH_MSGR_TAG_CLOSE:
1893 set_bit(CLOSED, &con->state); /* fixme */
1894 goto out;
1895 default:
1896 goto bad_tag;
1897 }
1898 }
1899 if (con->in_tag == CEPH_MSGR_TAG_MSG) {
1900 ret = read_partial_message(con);
1901 if (ret <= 0) {
1902 switch (ret) {
1903 case -EBADMSG:
1904 con->error_msg = "bad crc";
1905 ret = -EIO;
1906 break;
1907 case -EIO:
1908 con->error_msg = "io error";
1909 break;
1910 }
1911 goto out;
1912 }
1913 if (con->in_tag == CEPH_MSGR_TAG_READY)
1914 goto more;
1915 process_message(con);
1916 goto more;
1917 }
1918 if (con->in_tag == CEPH_MSGR_TAG_ACK) {
1919 ret = read_partial_ack(con);
1920 if (ret <= 0)
1921 goto out;
1922 process_ack(con);
1923 goto more;
1924 }
1925
1926 out:
1927 dout("try_read done on %p ret %d\n", con, ret);
1928 return ret;
1929
1930 bad_tag:
1931 pr_err("try_read bad con->in_tag = %d\n", (int)con->in_tag);
1932 con->error_msg = "protocol error, garbage tag";
1933 ret = -1;
1934 goto out;
1935 }
1936
1937
1938 /*
1939 * Atomically queue work on a connection. Bump @con reference to
1940 * avoid races with connection teardown.
1941 */
1942 static void queue_con(struct ceph_connection *con)
1943 {
1944 if (test_bit(DEAD, &con->state)) {
1945 dout("queue_con %p ignoring: DEAD\n",
1946 con);
1947 return;
1948 }
1949
1950 if (!con->ops->get(con)) {
1951 dout("queue_con %p ref count 0\n", con);
1952 return;
1953 }
1954
1955 if (!queue_delayed_work(ceph_msgr_wq, &con->work, 0)) {
1956 dout("queue_con %p - already queued\n", con);
1957 con->ops->put(con);
1958 } else {
1959 dout("queue_con %p\n", con);
1960 }
1961 }
1962
1963 /*
1964 * Do some work on a connection. Drop a connection ref when we're done.
1965 */
1966 static void con_work(struct work_struct *work)
1967 {
1968 struct ceph_connection *con = container_of(work, struct ceph_connection,
1969 work.work);
1970 int ret;
1971
1972 mutex_lock(&con->mutex);
1973 restart:
1974 if (test_and_clear_bit(BACKOFF, &con->state)) {
1975 dout("con_work %p backing off\n", con);
1976 if (queue_delayed_work(ceph_msgr_wq, &con->work,
1977 round_jiffies_relative(con->delay))) {
1978 dout("con_work %p backoff %lu\n", con, con->delay);
1979 mutex_unlock(&con->mutex);
1980 return;
1981 } else {
1982 con->ops->put(con);
1983 dout("con_work %p FAILED to back off %lu\n", con,
1984 con->delay);
1985 }
1986 }
1987
1988 if (test_bit(STANDBY, &con->state)) {
1989 dout("con_work %p STANDBY\n", con);
1990 goto done;
1991 }
1992 if (test_bit(CLOSED, &con->state)) { /* e.g. if we are replaced */
1993 dout("con_work CLOSED\n");
1994 con_close_socket(con);
1995 goto done;
1996 }
1997 if (test_and_clear_bit(OPENING, &con->state)) {
1998 /* reopen w/ new peer */
1999 dout("con_work OPENING\n");
2000 con_close_socket(con);
2001 }
2002
2003 if (test_and_clear_bit(SOCK_CLOSED, &con->state))
2004 goto fault;
2005
2006 ret = try_read(con);
2007 if (ret == -EAGAIN)
2008 goto restart;
2009 if (ret < 0)
2010 goto fault;
2011
2012 ret = try_write(con);
2013 if (ret == -EAGAIN)
2014 goto restart;
2015 if (ret < 0)
2016 goto fault;
2017
2018 done:
2019 mutex_unlock(&con->mutex);
2020 done_unlocked:
2021 con->ops->put(con);
2022 return;
2023
2024 fault:
2025 mutex_unlock(&con->mutex);
2026 ceph_fault(con); /* error/fault path */
2027 goto done_unlocked;
2028 }
2029
2030
2031 /*
2032 * Generic error/fault handler. A retry mechanism is used with
2033 * exponential backoff
2034 */
2035 static void ceph_fault(struct ceph_connection *con)
2036 {
2037 pr_err("%s%lld %s %s\n", ENTITY_NAME(con->peer_name),
2038 ceph_pr_addr(&con->peer_addr.in_addr), con->error_msg);
2039 dout("fault %p state %lu to peer %s\n",
2040 con, con->state, ceph_pr_addr(&con->peer_addr.in_addr));
2041
2042 if (test_bit(LOSSYTX, &con->state)) {
2043 dout("fault on LOSSYTX channel\n");
2044 goto out;
2045 }
2046
2047 mutex_lock(&con->mutex);
2048 if (test_bit(CLOSED, &con->state))
2049 goto out_unlock;
2050
2051 con_close_socket(con);
2052
2053 if (con->in_msg) {
2054 ceph_msg_put(con->in_msg);
2055 con->in_msg = NULL;
2056 }
2057
2058 /* Requeue anything that hasn't been acked */
2059 list_splice_init(&con->out_sent, &con->out_queue);
2060
2061 /* If there are no messages queued or keepalive pending, place
2062 * the connection in a STANDBY state */
2063 if (list_empty(&con->out_queue) &&
2064 !test_bit(KEEPALIVE_PENDING, &con->state)) {
2065 dout("fault %p setting STANDBY clearing WRITE_PENDING\n", con);
2066 clear_bit(WRITE_PENDING, &con->state);
2067 set_bit(STANDBY, &con->state);
2068 } else {
2069 /* retry after a delay. */
2070 if (con->delay == 0)
2071 con->delay = BASE_DELAY_INTERVAL;
2072 else if (con->delay < MAX_DELAY_INTERVAL)
2073 con->delay *= 2;
2074 con->ops->get(con);
2075 if (queue_delayed_work(ceph_msgr_wq, &con->work,
2076 round_jiffies_relative(con->delay))) {
2077 dout("fault queued %p delay %lu\n", con, con->delay);
2078 } else {
2079 con->ops->put(con);
2080 dout("fault failed to queue %p delay %lu, backoff\n",
2081 con, con->delay);
2082 /*
2083 * In many cases we see a socket state change
2084 * while con_work is running and end up
2085 * queuing (non-delayed) work, such that we
2086 * can't backoff with a delay. Set a flag so
2087 * that when con_work restarts we schedule the
2088 * delay then.
2089 */
2090 set_bit(BACKOFF, &con->state);
2091 }
2092 }
2093
2094 out_unlock:
2095 mutex_unlock(&con->mutex);
2096 out:
2097 /*
2098 * in case we faulted due to authentication, invalidate our
2099 * current tickets so that we can get new ones.
2100 */
2101 if (con->auth_retry && con->ops->invalidate_authorizer) {
2102 dout("calling invalidate_authorizer()\n");
2103 con->ops->invalidate_authorizer(con);
2104 }
2105
2106 if (con->ops->fault)
2107 con->ops->fault(con);
2108 }
2109
2110
2111
2112 /*
2113 * create a new messenger instance
2114 */
2115 struct ceph_messenger *ceph_messenger_create(struct ceph_entity_addr *myaddr,
2116 u32 supported_features,
2117 u32 required_features)
2118 {
2119 struct ceph_messenger *msgr;
2120
2121 msgr = kzalloc(sizeof(*msgr), GFP_KERNEL);
2122 if (msgr == NULL)
2123 return ERR_PTR(-ENOMEM);
2124
2125 msgr->supported_features = supported_features;
2126 msgr->required_features = required_features;
2127
2128 spin_lock_init(&msgr->global_seq_lock);
2129
2130 /* the zero page is needed if a request is "canceled" while the message
2131 * is being written over the socket */
2132 msgr->zero_page = __page_cache_alloc(GFP_KERNEL | __GFP_ZERO);
2133 if (!msgr->zero_page) {
2134 kfree(msgr);
2135 return ERR_PTR(-ENOMEM);
2136 }
2137 kmap(msgr->zero_page);
2138
2139 if (myaddr)
2140 msgr->inst.addr = *myaddr;
2141
2142 /* select a random nonce */
2143 msgr->inst.addr.type = 0;
2144 get_random_bytes(&msgr->inst.addr.nonce, sizeof(msgr->inst.addr.nonce));
2145 encode_my_addr(msgr);
2146
2147 dout("messenger_create %p\n", msgr);
2148 return msgr;
2149 }
2150 EXPORT_SYMBOL(ceph_messenger_create);
2151
2152 void ceph_messenger_destroy(struct ceph_messenger *msgr)
2153 {
2154 dout("destroy %p\n", msgr);
2155 kunmap(msgr->zero_page);
2156 __free_page(msgr->zero_page);
2157 kfree(msgr);
2158 dout("destroyed messenger %p\n", msgr);
2159 }
2160 EXPORT_SYMBOL(ceph_messenger_destroy);
2161
2162 static void clear_standby(struct ceph_connection *con)
2163 {
2164 /* come back from STANDBY? */
2165 if (test_and_clear_bit(STANDBY, &con->state)) {
2166 mutex_lock(&con->mutex);
2167 dout("clear_standby %p and ++connect_seq\n", con);
2168 con->connect_seq++;
2169 WARN_ON(test_bit(WRITE_PENDING, &con->state));
2170 WARN_ON(test_bit(KEEPALIVE_PENDING, &con->state));
2171 mutex_unlock(&con->mutex);
2172 }
2173 }
2174
2175 /*
2176 * Queue up an outgoing message on the given connection.
2177 */
2178 void ceph_con_send(struct ceph_connection *con, struct ceph_msg *msg)
2179 {
2180 if (test_bit(CLOSED, &con->state)) {
2181 dout("con_send %p closed, dropping %p\n", con, msg);
2182 ceph_msg_put(msg);
2183 return;
2184 }
2185
2186 /* set src+dst */
2187 msg->hdr.src = con->msgr->inst.name;
2188
2189 BUG_ON(msg->front.iov_len != le32_to_cpu(msg->hdr.front_len));
2190
2191 msg->needs_out_seq = true;
2192
2193 /* queue */
2194 mutex_lock(&con->mutex);
2195 BUG_ON(!list_empty(&msg->list_head));
2196 list_add_tail(&msg->list_head, &con->out_queue);
2197 dout("----- %p to %s%lld %d=%s len %d+%d+%d -----\n", msg,
2198 ENTITY_NAME(con->peer_name), le16_to_cpu(msg->hdr.type),
2199 ceph_msg_type_name(le16_to_cpu(msg->hdr.type)),
2200 le32_to_cpu(msg->hdr.front_len),
2201 le32_to_cpu(msg->hdr.middle_len),
2202 le32_to_cpu(msg->hdr.data_len));
2203 mutex_unlock(&con->mutex);
2204
2205 /* if there wasn't anything waiting to send before, queue
2206 * new work */
2207 clear_standby(con);
2208 if (test_and_set_bit(WRITE_PENDING, &con->state) == 0)
2209 queue_con(con);
2210 }
2211 EXPORT_SYMBOL(ceph_con_send);
2212
2213 /*
2214 * Revoke a message that was previously queued for send
2215 */
2216 void ceph_con_revoke(struct ceph_connection *con, struct ceph_msg *msg)
2217 {
2218 mutex_lock(&con->mutex);
2219 if (!list_empty(&msg->list_head)) {
2220 dout("con_revoke %p msg %p - was on queue\n", con, msg);
2221 list_del_init(&msg->list_head);
2222 ceph_msg_put(msg);
2223 msg->hdr.seq = 0;
2224 }
2225 if (con->out_msg == msg) {
2226 dout("con_revoke %p msg %p - was sending\n", con, msg);
2227 con->out_msg = NULL;
2228 if (con->out_kvec_is_msg) {
2229 con->out_skip = con->out_kvec_bytes;
2230 con->out_kvec_is_msg = false;
2231 }
2232 ceph_msg_put(msg);
2233 msg->hdr.seq = 0;
2234 }
2235 mutex_unlock(&con->mutex);
2236 }
2237
2238 /*
2239 * Revoke a message that we may be reading data into
2240 */
2241 void ceph_con_revoke_message(struct ceph_connection *con, struct ceph_msg *msg)
2242 {
2243 mutex_lock(&con->mutex);
2244 if (con->in_msg && con->in_msg == msg) {
2245 unsigned front_len = le32_to_cpu(con->in_hdr.front_len);
2246 unsigned middle_len = le32_to_cpu(con->in_hdr.middle_len);
2247 unsigned data_len = le32_to_cpu(con->in_hdr.data_len);
2248
2249 /* skip rest of message */
2250 dout("con_revoke_pages %p msg %p revoked\n", con, msg);
2251 con->in_base_pos = con->in_base_pos -
2252 sizeof(struct ceph_msg_header) -
2253 front_len -
2254 middle_len -
2255 data_len -
2256 sizeof(struct ceph_msg_footer);
2257 ceph_msg_put(con->in_msg);
2258 con->in_msg = NULL;
2259 con->in_tag = CEPH_MSGR_TAG_READY;
2260 con->in_seq++;
2261 } else {
2262 dout("con_revoke_pages %p msg %p pages %p no-op\n",
2263 con, con->in_msg, msg);
2264 }
2265 mutex_unlock(&con->mutex);
2266 }
2267
2268 /*
2269 * Queue a keepalive byte to ensure the tcp connection is alive.
2270 */
2271 void ceph_con_keepalive(struct ceph_connection *con)
2272 {
2273 dout("con_keepalive %p\n", con);
2274 clear_standby(con);
2275 if (test_and_set_bit(KEEPALIVE_PENDING, &con->state) == 0 &&
2276 test_and_set_bit(WRITE_PENDING, &con->state) == 0)
2277 queue_con(con);
2278 }
2279 EXPORT_SYMBOL(ceph_con_keepalive);
2280
2281
2282 /*
2283 * construct a new message with given type, size
2284 * the new msg has a ref count of 1.
2285 */
2286 struct ceph_msg *ceph_msg_new(int type, int front_len, gfp_t flags)
2287 {
2288 struct ceph_msg *m;
2289
2290 m = kmalloc(sizeof(*m), flags);
2291 if (m == NULL)
2292 goto out;
2293 kref_init(&m->kref);
2294 INIT_LIST_HEAD(&m->list_head);
2295
2296 m->hdr.tid = 0;
2297 m->hdr.type = cpu_to_le16(type);
2298 m->hdr.priority = cpu_to_le16(CEPH_MSG_PRIO_DEFAULT);
2299 m->hdr.version = 0;
2300 m->hdr.front_len = cpu_to_le32(front_len);
2301 m->hdr.middle_len = 0;
2302 m->hdr.data_len = 0;
2303 m->hdr.data_off = 0;
2304 m->hdr.reserved = 0;
2305 m->footer.front_crc = 0;
2306 m->footer.middle_crc = 0;
2307 m->footer.data_crc = 0;
2308 m->footer.flags = 0;
2309 m->front_max = front_len;
2310 m->front_is_vmalloc = false;
2311 m->more_to_follow = false;
2312 m->pool = NULL;
2313
2314 /* middle */
2315 m->middle = NULL;
2316
2317 /* data */
2318 m->nr_pages = 0;
2319 m->page_alignment = 0;
2320 m->pages = NULL;
2321 m->pagelist = NULL;
2322 m->bio = NULL;
2323 m->bio_iter = NULL;
2324 m->bio_seg = 0;
2325 m->trail = NULL;
2326
2327 /* front */
2328 if (front_len) {
2329 if (front_len > PAGE_CACHE_SIZE) {
2330 m->front.iov_base = __vmalloc(front_len, flags,
2331 PAGE_KERNEL);
2332 m->front_is_vmalloc = true;
2333 } else {
2334 m->front.iov_base = kmalloc(front_len, flags);
2335 }
2336 if (m->front.iov_base == NULL) {
2337 pr_err("msg_new can't allocate %d bytes\n",
2338 front_len);
2339 goto out2;
2340 }
2341 } else {
2342 m->front.iov_base = NULL;
2343 }
2344 m->front.iov_len = front_len;
2345
2346 dout("ceph_msg_new %p front %d\n", m, front_len);
2347 return m;
2348
2349 out2:
2350 ceph_msg_put(m);
2351 out:
2352 pr_err("msg_new can't create type %d front %d\n", type, front_len);
2353 return NULL;
2354 }
2355 EXPORT_SYMBOL(ceph_msg_new);
2356
2357 /*
2358 * Allocate "middle" portion of a message, if it is needed and wasn't
2359 * allocated by alloc_msg. This allows us to read a small fixed-size
2360 * per-type header in the front and then gracefully fail (i.e.,
2361 * propagate the error to the caller based on info in the front) when
2362 * the middle is too large.
2363 */
2364 static int ceph_alloc_middle(struct ceph_connection *con, struct ceph_msg *msg)
2365 {
2366 int type = le16_to_cpu(msg->hdr.type);
2367 int middle_len = le32_to_cpu(msg->hdr.middle_len);
2368
2369 dout("alloc_middle %p type %d %s middle_len %d\n", msg, type,
2370 ceph_msg_type_name(type), middle_len);
2371 BUG_ON(!middle_len);
2372 BUG_ON(msg->middle);
2373
2374 msg->middle = ceph_buffer_new(middle_len, GFP_NOFS);
2375 if (!msg->middle)
2376 return -ENOMEM;
2377 return 0;
2378 }
2379
2380 /*
2381 * Generic message allocator, for incoming messages.
2382 */
2383 static struct ceph_msg *ceph_alloc_msg(struct ceph_connection *con,
2384 struct ceph_msg_header *hdr,
2385 int *skip)
2386 {
2387 int type = le16_to_cpu(hdr->type);
2388 int front_len = le32_to_cpu(hdr->front_len);
2389 int middle_len = le32_to_cpu(hdr->middle_len);
2390 struct ceph_msg *msg = NULL;
2391 int ret;
2392
2393 if (con->ops->alloc_msg) {
2394 mutex_unlock(&con->mutex);
2395 msg = con->ops->alloc_msg(con, hdr, skip);
2396 mutex_lock(&con->mutex);
2397 if (!msg || *skip)
2398 return NULL;
2399 }
2400 if (!msg) {
2401 *skip = 0;
2402 msg = ceph_msg_new(type, front_len, GFP_NOFS);
2403 if (!msg) {
2404 pr_err("unable to allocate msg type %d len %d\n",
2405 type, front_len);
2406 return NULL;
2407 }
2408 msg->page_alignment = le16_to_cpu(hdr->data_off);
2409 }
2410 memcpy(&msg->hdr, &con->in_hdr, sizeof(con->in_hdr));
2411
2412 if (middle_len && !msg->middle) {
2413 ret = ceph_alloc_middle(con, msg);
2414 if (ret < 0) {
2415 ceph_msg_put(msg);
2416 return NULL;
2417 }
2418 }
2419
2420 return msg;
2421 }
2422
2423
2424 /*
2425 * Free a generically kmalloc'd message.
2426 */
2427 void ceph_msg_kfree(struct ceph_msg *m)
2428 {
2429 dout("msg_kfree %p\n", m);
2430 if (m->front_is_vmalloc)
2431 vfree(m->front.iov_base);
2432 else
2433 kfree(m->front.iov_base);
2434 kfree(m);
2435 }
2436
2437 /*
2438 * Drop a msg ref. Destroy as needed.
2439 */
2440 void ceph_msg_last_put(struct kref *kref)
2441 {
2442 struct ceph_msg *m = container_of(kref, struct ceph_msg, kref);
2443
2444 dout("ceph_msg_put last one on %p\n", m);
2445 WARN_ON(!list_empty(&m->list_head));
2446
2447 /* drop middle, data, if any */
2448 if (m->middle) {
2449 ceph_buffer_put(m->middle);
2450 m->middle = NULL;
2451 }
2452 m->nr_pages = 0;
2453 m->pages = NULL;
2454
2455 if (m->pagelist) {
2456 ceph_pagelist_release(m->pagelist);
2457 kfree(m->pagelist);
2458 m->pagelist = NULL;
2459 }
2460
2461 m->trail = NULL;
2462
2463 if (m->pool)
2464 ceph_msgpool_put(m->pool, m);
2465 else
2466 ceph_msg_kfree(m);
2467 }
2468 EXPORT_SYMBOL(ceph_msg_last_put);
2469
2470 void ceph_msg_dump(struct ceph_msg *msg)
2471 {
2472 pr_debug("msg_dump %p (front_max %d nr_pages %d)\n", msg,
2473 msg->front_max, msg->nr_pages);
2474 print_hex_dump(KERN_DEBUG, "header: ",
2475 DUMP_PREFIX_OFFSET, 16, 1,
2476 &msg->hdr, sizeof(msg->hdr), true);
2477 print_hex_dump(KERN_DEBUG, " front: ",
2478 DUMP_PREFIX_OFFSET, 16, 1,
2479 msg->front.iov_base, msg->front.iov_len, true);
2480 if (msg->middle)
2481 print_hex_dump(KERN_DEBUG, "middle: ",
2482 DUMP_PREFIX_OFFSET, 16, 1,
2483 msg->middle->vec.iov_base,
2484 msg->middle->vec.iov_len, true);
2485 print_hex_dump(KERN_DEBUG, "footer: ",
2486 DUMP_PREFIX_OFFSET, 16, 1,
2487 &msg->footer, sizeof(msg->footer), true);
2488 }
2489 EXPORT_SYMBOL(ceph_msg_dump);
linux-next