1 #include <linux/ceph/ceph_debug.h>
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>
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 <linux/dns_resolver.h>
17 #include <linux/ceph/libceph.h>
18 #include <linux/ceph/messenger.h>
19 #include <linux/ceph/decode.h>
20 #include <linux/ceph/pagelist.h>
21 #include <linux/export.h>
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
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;
38 static struct lock_class_key socket_class;
42 * When skipping (ignoring) a block of input we read it into a "skip
43 * buffer," which is this many bytes in size.
45 #define SKIP_BUF_SIZE 1024
47 static void queue_con(struct ceph_connection *con);
48 static void con_work(struct work_struct *);
49 static void ceph_fault(struct ceph_connection *con);
52 * Nicely render a sockaddr as a string. An array of formatted
53 * strings is used, to approximate reentrancy.
55 #define ADDR_STR_COUNT_LOG 5 /* log2(# address strings in array) */
56 #define ADDR_STR_COUNT (1 << ADDR_STR_COUNT_LOG)
57 #define ADDR_STR_COUNT_MASK (ADDR_STR_COUNT - 1)
58 #define MAX_ADDR_STR_LEN 64 /* 54 is enough */
60 static char addr_str[ADDR_STR_COUNT][MAX_ADDR_STR_LEN];
61 static atomic_t addr_str_seq = ATOMIC_INIT(0);
63 static struct page *zero_page; /* used in certain error cases */
65 const char *ceph_pr_addr(const struct sockaddr_storage *ss)
69 struct sockaddr_in *in4 = (struct sockaddr_in *) ss;
70 struct sockaddr_in6 *in6 = (struct sockaddr_in6 *) ss;
72 i = atomic_inc_return(&addr_str_seq) & ADDR_STR_COUNT_MASK;
75 switch (ss->ss_family) {
77 snprintf(s, MAX_ADDR_STR_LEN, "%pI4:%hu", &in4->sin_addr,
78 ntohs(in4->sin_port));
82 snprintf(s, MAX_ADDR_STR_LEN, "[%pI6c]:%hu", &in6->sin6_addr,
83 ntohs(in6->sin6_port));
87 snprintf(s, MAX_ADDR_STR_LEN, "(unknown sockaddr family %hu)",
93 EXPORT_SYMBOL(ceph_pr_addr);
95 static void encode_my_addr(struct ceph_messenger *msgr)
97 memcpy(&msgr->my_enc_addr, &msgr->inst.addr, sizeof(msgr->my_enc_addr));
98 ceph_encode_addr(&msgr->my_enc_addr);
102 * work queue for all reading and writing to/from the socket.
104 static struct workqueue_struct *ceph_msgr_wq;
106 void _ceph_msgr_exit(void)
109 destroy_workqueue(ceph_msgr_wq);
113 BUG_ON(zero_page == NULL);
115 page_cache_release(zero_page);
119 int ceph_msgr_init(void)
121 BUG_ON(zero_page != NULL);
122 zero_page = ZERO_PAGE(0);
123 page_cache_get(zero_page);
125 ceph_msgr_wq = alloc_workqueue("ceph-msgr", WQ_NON_REENTRANT, 0);
129 pr_err("msgr_init failed to create workqueue\n");
134 EXPORT_SYMBOL(ceph_msgr_init);
136 void ceph_msgr_exit(void)
138 BUG_ON(ceph_msgr_wq == NULL);
142 EXPORT_SYMBOL(ceph_msgr_exit);
144 void ceph_msgr_flush(void)
146 flush_workqueue(ceph_msgr_wq);
148 EXPORT_SYMBOL(ceph_msgr_flush);
152 * socket callback functions
155 /* data available on socket, or listen socket received a connect */
156 static void ceph_data_ready(struct sock *sk, int count_unused)
158 struct ceph_connection *con = sk->sk_user_data;
160 if (sk->sk_state != TCP_CLOSE_WAIT) {
161 dout("ceph_data_ready on %p state = %lu, queueing work\n",
167 /* socket has buffer space for writing */
168 static void ceph_write_space(struct sock *sk)
170 struct ceph_connection *con = sk->sk_user_data;
172 /* only queue to workqueue if there is data we want to write,
173 * and there is sufficient space in the socket buffer to accept
174 * more data. clear SOCK_NOSPACE so that ceph_write_space()
175 * doesn't get called again until try_write() fills the socket
176 * buffer. See net/ipv4/tcp_input.c:tcp_check_space()
177 * and net/core/stream.c:sk_stream_write_space().
179 if (test_bit(WRITE_PENDING, &con->state)) {
180 if (sk_stream_wspace(sk) >= sk_stream_min_wspace(sk)) {
181 dout("ceph_write_space %p queueing write work\n", con);
182 clear_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
186 dout("ceph_write_space %p nothing to write\n", con);
190 /* socket's state has changed */
191 static void ceph_state_change(struct sock *sk)
193 struct ceph_connection *con = sk->sk_user_data;
195 dout("ceph_state_change %p state = %lu sk_state = %u\n",
196 con, con->state, sk->sk_state);
198 if (test_bit(CLOSED, &con->state))
201 switch (sk->sk_state) {
203 dout("ceph_state_change TCP_CLOSE\n");
205 dout("ceph_state_change TCP_CLOSE_WAIT\n");
206 if (test_and_set_bit(SOCK_CLOSED, &con->state) == 0) {
207 if (test_bit(CONNECTING, &con->state))
208 con->error_msg = "connection failed";
210 con->error_msg = "socket closed";
214 case TCP_ESTABLISHED:
215 dout("ceph_state_change TCP_ESTABLISHED\n");
218 default: /* Everything else is uninteresting */
224 * set up socket callbacks
226 static void set_sock_callbacks(struct socket *sock,
227 struct ceph_connection *con)
229 struct sock *sk = sock->sk;
230 sk->sk_user_data = con;
231 sk->sk_data_ready = ceph_data_ready;
232 sk->sk_write_space = ceph_write_space;
233 sk->sk_state_change = ceph_state_change;
242 * initiate connection to a remote socket.
244 static int ceph_tcp_connect(struct ceph_connection *con)
246 struct sockaddr_storage *paddr = &con->peer_addr.in_addr;
251 ret = sock_create_kern(con->peer_addr.in_addr.ss_family, SOCK_STREAM,
255 sock->sk->sk_allocation = GFP_NOFS;
257 #ifdef CONFIG_LOCKDEP
258 lockdep_set_class(&sock->sk->sk_lock, &socket_class);
261 set_sock_callbacks(sock, con);
263 dout("connect %s\n", ceph_pr_addr(&con->peer_addr.in_addr));
265 ret = sock->ops->connect(sock, (struct sockaddr *)paddr, sizeof(*paddr),
267 if (ret == -EINPROGRESS) {
268 dout("connect %s EINPROGRESS sk_state = %u\n",
269 ceph_pr_addr(&con->peer_addr.in_addr),
271 } else if (ret < 0) {
272 pr_err("connect %s error %d\n",
273 ceph_pr_addr(&con->peer_addr.in_addr), ret);
275 con->error_msg = "connect error";
284 static int ceph_tcp_recvmsg(struct socket *sock, void *buf, size_t len)
286 struct kvec iov = {buf, len};
287 struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
290 r = kernel_recvmsg(sock, &msg, &iov, 1, len, msg.msg_flags);
297 * write something. @more is true if caller will be sending more data
300 static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov,
301 size_t kvlen, size_t len, int more)
303 struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
307 msg.msg_flags |= MSG_MORE;
309 msg.msg_flags |= MSG_EOR; /* superfluous, but what the hell */
311 r = kernel_sendmsg(sock, &msg, iov, kvlen, len);
317 static int ceph_tcp_sendpage(struct socket *sock, struct page *page,
318 int offset, size_t size, int more)
320 int flags = MSG_DONTWAIT | MSG_NOSIGNAL | (more ? MSG_MORE : MSG_EOR);
323 ret = kernel_sendpage(sock, page, offset, size, flags);
332 * Shutdown/close the socket for the given connection.
334 static int con_close_socket(struct ceph_connection *con)
338 dout("con_close_socket on %p sock %p\n", con, con->sock);
341 set_bit(SOCK_CLOSED, &con->state);
342 rc = con->sock->ops->shutdown(con->sock, SHUT_RDWR);
343 sock_release(con->sock);
345 clear_bit(SOCK_CLOSED, &con->state);
350 * Reset a connection. Discard all incoming and outgoing messages
351 * and clear *_seq state.
353 static void ceph_msg_remove(struct ceph_msg *msg)
355 list_del_init(&msg->list_head);
358 static void ceph_msg_remove_list(struct list_head *head)
360 while (!list_empty(head)) {
361 struct ceph_msg *msg = list_first_entry(head, struct ceph_msg,
363 ceph_msg_remove(msg);
367 static void reset_connection(struct ceph_connection *con)
369 /* reset connection, out_queue, msg_ and connect_seq */
370 /* discard existing out_queue and msg_seq */
371 ceph_msg_remove_list(&con->out_queue);
372 ceph_msg_remove_list(&con->out_sent);
375 ceph_msg_put(con->in_msg);
379 con->connect_seq = 0;
382 ceph_msg_put(con->out_msg);
386 con->in_seq_acked = 0;
390 * mark a peer down. drop any open connections.
392 void ceph_con_close(struct ceph_connection *con)
394 dout("con_close %p peer %s\n", con,
395 ceph_pr_addr(&con->peer_addr.in_addr));
396 set_bit(CLOSED, &con->state); /* in case there's queued work */
397 clear_bit(STANDBY, &con->state); /* avoid connect_seq bump */
398 clear_bit(LOSSYTX, &con->state); /* so we retry next connect */
399 clear_bit(KEEPALIVE_PENDING, &con->state);
400 clear_bit(WRITE_PENDING, &con->state);
401 mutex_lock(&con->mutex);
402 reset_connection(con);
403 con->peer_global_seq = 0;
404 cancel_delayed_work(&con->work);
405 mutex_unlock(&con->mutex);
408 EXPORT_SYMBOL(ceph_con_close);
411 * Reopen a closed connection, with a new peer address.
413 void ceph_con_open(struct ceph_connection *con, struct ceph_entity_addr *addr)
415 dout("con_open %p %s\n", con, ceph_pr_addr(&addr->in_addr));
416 set_bit(OPENING, &con->state);
417 clear_bit(CLOSED, &con->state);
418 memcpy(&con->peer_addr, addr, sizeof(*addr));
419 con->delay = 0; /* reset backoff memory */
422 EXPORT_SYMBOL(ceph_con_open);
425 * return true if this connection ever successfully opened
427 bool ceph_con_opened(struct ceph_connection *con)
429 return con->connect_seq > 0;
435 struct ceph_connection *ceph_con_get(struct ceph_connection *con)
437 int nref = __atomic_add_unless(&con->nref, 1, 0);
439 dout("con_get %p nref = %d -> %d\n", con, nref, nref + 1);
441 return nref ? con : NULL;
444 void ceph_con_put(struct ceph_connection *con)
446 int nref = atomic_dec_return(&con->nref);
453 dout("con_put %p nref = %d -> %d\n", con, nref + 1, nref);
457 * initialize a new connection.
459 void ceph_con_init(struct ceph_messenger *msgr, struct ceph_connection *con)
461 dout("con_init %p\n", con);
462 memset(con, 0, sizeof(*con));
463 atomic_set(&con->nref, 1);
465 mutex_init(&con->mutex);
466 INIT_LIST_HEAD(&con->out_queue);
467 INIT_LIST_HEAD(&con->out_sent);
468 INIT_DELAYED_WORK(&con->work, con_work);
470 EXPORT_SYMBOL(ceph_con_init);
474 * We maintain a global counter to order connection attempts. Get
475 * a unique seq greater than @gt.
477 static u32 get_global_seq(struct ceph_messenger *msgr, u32 gt)
481 spin_lock(&msgr->global_seq_lock);
482 if (msgr->global_seq < gt)
483 msgr->global_seq = gt;
484 ret = ++msgr->global_seq;
485 spin_unlock(&msgr->global_seq_lock);
489 static void ceph_con_out_kvec_reset(struct ceph_connection *con)
491 con->out_kvec_left = 0;
492 con->out_kvec_bytes = 0;
493 con->out_kvec_cur = &con->out_kvec[0];
496 static void ceph_con_out_kvec_add(struct ceph_connection *con,
497 size_t size, void *data)
501 index = con->out_kvec_left;
502 BUG_ON(index >= ARRAY_SIZE(con->out_kvec));
504 con->out_kvec[index].iov_len = size;
505 con->out_kvec[index].iov_base = data;
506 con->out_kvec_left++;
507 con->out_kvec_bytes += size;
511 * Prepare footer for currently outgoing message, and finish things
512 * off. Assumes out_kvec* are already valid.. we just add on to the end.
514 static void prepare_write_message_footer(struct ceph_connection *con)
516 struct ceph_msg *m = con->out_msg;
517 int v = con->out_kvec_left;
519 dout("prepare_write_message_footer %p\n", con);
520 con->out_kvec_is_msg = true;
521 con->out_kvec[v].iov_base = &m->footer;
522 con->out_kvec[v].iov_len = sizeof(m->footer);
523 con->out_kvec_bytes += sizeof(m->footer);
524 con->out_kvec_left++;
525 con->out_more = m->more_to_follow;
526 con->out_msg_done = true;
530 * Prepare headers for the next outgoing message.
532 static void prepare_write_message(struct ceph_connection *con)
537 ceph_con_out_kvec_reset(con);
538 con->out_kvec_is_msg = true;
539 con->out_msg_done = false;
541 /* Sneak an ack in there first? If we can get it into the same
542 * TCP packet that's a good thing. */
543 if (con->in_seq > con->in_seq_acked) {
544 con->in_seq_acked = con->in_seq;
545 ceph_con_out_kvec_add(con, sizeof (tag_ack), &tag_ack);
546 con->out_temp_ack = cpu_to_le64(con->in_seq_acked);
547 ceph_con_out_kvec_add(con, sizeof (con->out_temp_ack),
551 m = list_first_entry(&con->out_queue, struct ceph_msg, list_head);
554 /* put message on sent list */
556 list_move_tail(&m->list_head, &con->out_sent);
559 * only assign outgoing seq # if we haven't sent this message
560 * yet. if it is requeued, resend with it's original seq.
562 if (m->needs_out_seq) {
563 m->hdr.seq = cpu_to_le64(++con->out_seq);
564 m->needs_out_seq = false;
571 dout("prepare_write_message %p seq %lld type %d len %d+%d+%d %d pgs\n",
572 m, con->out_seq, le16_to_cpu(m->hdr.type),
573 le32_to_cpu(m->hdr.front_len), le32_to_cpu(m->hdr.middle_len),
574 le32_to_cpu(m->hdr.data_len),
576 BUG_ON(le32_to_cpu(m->hdr.front_len) != m->front.iov_len);
578 /* tag + hdr + front + middle */
579 ceph_con_out_kvec_add(con, sizeof (tag_msg), &tag_msg);
580 ceph_con_out_kvec_add(con, sizeof (m->hdr), &m->hdr);
581 ceph_con_out_kvec_add(con, m->front.iov_len, m->front.iov_base);
584 ceph_con_out_kvec_add(con, m->middle->vec.iov_len,
585 m->middle->vec.iov_base);
587 /* fill in crc (except data pages), footer */
588 crc = crc32c(0, &m->hdr, offsetof(struct ceph_msg_header, crc));
589 con->out_msg->hdr.crc = cpu_to_le32(crc);
590 con->out_msg->footer.flags = CEPH_MSG_FOOTER_COMPLETE;
592 crc = crc32c(0, m->front.iov_base, m->front.iov_len);
593 con->out_msg->footer.front_crc = cpu_to_le32(crc);
595 crc = crc32c(0, m->middle->vec.iov_base,
596 m->middle->vec.iov_len);
597 con->out_msg->footer.middle_crc = cpu_to_le32(crc);
599 con->out_msg->footer.middle_crc = 0;
600 con->out_msg->footer.data_crc = 0;
601 dout("prepare_write_message front_crc %u data_crc %u\n",
602 le32_to_cpu(con->out_msg->footer.front_crc),
603 le32_to_cpu(con->out_msg->footer.middle_crc));
605 /* is there a data payload? */
606 if (le32_to_cpu(m->hdr.data_len) > 0) {
607 /* initialize page iterator */
608 con->out_msg_pos.page = 0;
610 con->out_msg_pos.page_pos = m->page_alignment;
612 con->out_msg_pos.page_pos = 0;
613 con->out_msg_pos.data_pos = 0;
614 con->out_msg_pos.did_page_crc = false;
615 con->out_more = 1; /* data + footer will follow */
617 /* no, queue up footer too and be done */
618 prepare_write_message_footer(con);
621 set_bit(WRITE_PENDING, &con->state);
627 static void prepare_write_ack(struct ceph_connection *con)
629 dout("prepare_write_ack %p %llu -> %llu\n", con,
630 con->in_seq_acked, con->in_seq);
631 con->in_seq_acked = con->in_seq;
633 ceph_con_out_kvec_reset(con);
635 ceph_con_out_kvec_add(con, sizeof (tag_ack), &tag_ack);
637 con->out_temp_ack = cpu_to_le64(con->in_seq_acked);
638 ceph_con_out_kvec_add(con, sizeof (con->out_temp_ack),
641 con->out_more = 1; /* more will follow.. eventually.. */
642 set_bit(WRITE_PENDING, &con->state);
646 * Prepare to write keepalive byte.
648 static void prepare_write_keepalive(struct ceph_connection *con)
650 dout("prepare_write_keepalive %p\n", con);
651 ceph_con_out_kvec_reset(con);
652 ceph_con_out_kvec_add(con, sizeof (tag_keepalive), &tag_keepalive);
653 set_bit(WRITE_PENDING, &con->state);
657 * Connection negotiation.
660 static struct ceph_auth_handshake *get_connect_authorizer(struct ceph_connection *con,
663 struct ceph_auth_handshake *auth;
665 if (!con->ops->get_authorizer) {
666 con->out_connect.authorizer_protocol = CEPH_AUTH_UNKNOWN;
667 con->out_connect.authorizer_len = 0;
672 /* Can't hold the mutex while getting authorizer */
674 mutex_unlock(&con->mutex);
676 auth = con->ops->get_authorizer(con, auth_proto, con->auth_retry);
678 mutex_lock(&con->mutex);
682 if (test_bit(CLOSED, &con->state) || test_bit(OPENING, &con->state))
683 return ERR_PTR(-EAGAIN);
685 con->auth_reply_buf = auth->authorizer_reply_buf;
686 con->auth_reply_buf_len = auth->authorizer_reply_buf_len;
693 * We connected to a peer and are saying hello.
695 static void prepare_write_banner(struct ceph_connection *con)
697 ceph_con_out_kvec_add(con, strlen(CEPH_BANNER), CEPH_BANNER);
698 ceph_con_out_kvec_add(con, sizeof (con->msgr->my_enc_addr),
699 &con->msgr->my_enc_addr);
702 set_bit(WRITE_PENDING, &con->state);
705 static int prepare_write_connect(struct ceph_connection *con)
707 unsigned int global_seq = get_global_seq(con->msgr, 0);
710 struct ceph_auth_handshake *auth;
712 switch (con->peer_name.type) {
713 case CEPH_ENTITY_TYPE_MON:
714 proto = CEPH_MONC_PROTOCOL;
716 case CEPH_ENTITY_TYPE_OSD:
717 proto = CEPH_OSDC_PROTOCOL;
719 case CEPH_ENTITY_TYPE_MDS:
720 proto = CEPH_MDSC_PROTOCOL;
726 dout("prepare_write_connect %p cseq=%d gseq=%d proto=%d\n", con,
727 con->connect_seq, global_seq, proto);
729 con->out_connect.features = cpu_to_le64(con->msgr->supported_features);
730 con->out_connect.host_type = cpu_to_le32(CEPH_ENTITY_TYPE_CLIENT);
731 con->out_connect.connect_seq = cpu_to_le32(con->connect_seq);
732 con->out_connect.global_seq = cpu_to_le32(global_seq);
733 con->out_connect.protocol_version = cpu_to_le32(proto);
734 con->out_connect.flags = 0;
736 auth_proto = CEPH_AUTH_UNKNOWN;
737 auth = get_connect_authorizer(con, &auth_proto);
739 return PTR_ERR(auth);
741 con->out_connect.authorizer_protocol = cpu_to_le32(auth_proto);
742 con->out_connect.authorizer_len = auth ?
743 cpu_to_le32(auth->authorizer_buf_len) : 0;
745 ceph_con_out_kvec_add(con, sizeof (con->out_connect),
747 if (auth && auth->authorizer_buf_len)
748 ceph_con_out_kvec_add(con, auth->authorizer_buf_len,
749 auth->authorizer_buf);
752 set_bit(WRITE_PENDING, &con->state);
758 * write as much of pending kvecs to the socket as we can.
760 * 0 -> socket full, but more to do
763 static int write_partial_kvec(struct ceph_connection *con)
767 dout("write_partial_kvec %p %d left\n", con, con->out_kvec_bytes);
768 while (con->out_kvec_bytes > 0) {
769 ret = ceph_tcp_sendmsg(con->sock, con->out_kvec_cur,
770 con->out_kvec_left, con->out_kvec_bytes,
774 con->out_kvec_bytes -= ret;
775 if (con->out_kvec_bytes == 0)
778 /* account for full iov entries consumed */
779 while (ret >= con->out_kvec_cur->iov_len) {
780 BUG_ON(!con->out_kvec_left);
781 ret -= con->out_kvec_cur->iov_len;
783 con->out_kvec_left--;
785 /* and for a partially-consumed entry */
787 con->out_kvec_cur->iov_len -= ret;
788 con->out_kvec_cur->iov_base += ret;
791 con->out_kvec_left = 0;
792 con->out_kvec_is_msg = false;
795 dout("write_partial_kvec %p %d left in %d kvecs ret = %d\n", con,
796 con->out_kvec_bytes, con->out_kvec_left, ret);
797 return ret; /* done! */
801 static void init_bio_iter(struct bio *bio, struct bio **iter, int *seg)
812 static void iter_bio_next(struct bio **bio_iter, int *seg)
814 if (*bio_iter == NULL)
817 BUG_ON(*seg >= (*bio_iter)->bi_vcnt);
820 if (*seg == (*bio_iter)->bi_vcnt)
821 init_bio_iter((*bio_iter)->bi_next, bio_iter, seg);
826 * Write as much message data payload as we can. If we finish, queue
828 * 1 -> done, footer is now queued in out_kvec[].
829 * 0 -> socket full, but more to do
832 static int write_partial_msg_pages(struct ceph_connection *con)
834 struct ceph_msg *msg = con->out_msg;
835 unsigned int data_len = le32_to_cpu(msg->hdr.data_len);
837 bool do_datacrc = !con->msgr->nocrc;
841 size_t trail_len = (msg->trail ? msg->trail->length : 0);
843 dout("write_partial_msg_pages %p msg %p page %d/%d offset %d\n",
844 con, con->out_msg, con->out_msg_pos.page, con->out_msg->nr_pages,
845 con->out_msg_pos.page_pos);
848 if (msg->bio && !msg->bio_iter)
849 init_bio_iter(msg->bio, &msg->bio_iter, &msg->bio_seg);
852 while (data_len > con->out_msg_pos.data_pos) {
853 struct page *page = NULL;
854 int max_write = PAGE_SIZE;
857 total_max_write = data_len - trail_len -
858 con->out_msg_pos.data_pos;
861 * if we are calculating the data crc (the default), we need
862 * to map the page. if our pages[] has been revoked, use the
866 /* have we reached the trail part of the data? */
867 if (con->out_msg_pos.data_pos >= data_len - trail_len) {
870 total_max_write = data_len - con->out_msg_pos.data_pos;
872 page = list_first_entry(&msg->trail->head,
874 max_write = PAGE_SIZE;
875 } else if (msg->pages) {
876 page = msg->pages[con->out_msg_pos.page];
877 } else if (msg->pagelist) {
878 page = list_first_entry(&msg->pagelist->head,
881 } else if (msg->bio) {
884 bv = bio_iovec_idx(msg->bio_iter, msg->bio_seg);
886 bio_offset = bv->bv_offset;
887 max_write = bv->bv_len;
892 len = min_t(int, max_write - con->out_msg_pos.page_pos,
895 if (do_datacrc && !con->out_msg_pos.did_page_crc) {
898 u32 tmpcrc = le32_to_cpu(con->out_msg->footer.data_crc);
902 BUG_ON(kaddr == NULL);
903 base = kaddr + con->out_msg_pos.page_pos + bio_offset;
904 crc = crc32c(tmpcrc, base, len);
905 con->out_msg->footer.data_crc = cpu_to_le32(crc);
906 con->out_msg_pos.did_page_crc = true;
908 ret = ceph_tcp_sendpage(con->sock, page,
909 con->out_msg_pos.page_pos + bio_offset,
918 con->out_msg_pos.data_pos += ret;
919 con->out_msg_pos.page_pos += ret;
921 con->out_msg_pos.page_pos = 0;
922 con->out_msg_pos.page++;
923 con->out_msg_pos.did_page_crc = false;
925 list_move_tail(&page->lru,
927 else if (msg->pagelist)
928 list_move_tail(&page->lru,
929 &msg->pagelist->head);
932 iter_bio_next(&msg->bio_iter, &msg->bio_seg);
937 dout("write_partial_msg_pages %p msg %p done\n", con, msg);
939 /* prepare and queue up footer, too */
941 con->out_msg->footer.flags |= CEPH_MSG_FOOTER_NOCRC;
942 ceph_con_out_kvec_reset(con);
943 prepare_write_message_footer(con);
952 static int write_partial_skip(struct ceph_connection *con)
956 while (con->out_skip > 0) {
957 size_t size = min(con->out_skip, (int) PAGE_CACHE_SIZE);
959 ret = ceph_tcp_sendpage(con->sock, zero_page, 0, size, 1);
962 con->out_skip -= ret;
970 * Prepare to read connection handshake, or an ack.
972 static void prepare_read_banner(struct ceph_connection *con)
974 dout("prepare_read_banner %p\n", con);
975 con->in_base_pos = 0;
978 static void prepare_read_connect(struct ceph_connection *con)
980 dout("prepare_read_connect %p\n", con);
981 con->in_base_pos = 0;
984 static void prepare_read_ack(struct ceph_connection *con)
986 dout("prepare_read_ack %p\n", con);
987 con->in_base_pos = 0;
990 static void prepare_read_tag(struct ceph_connection *con)
992 dout("prepare_read_tag %p\n", con);
993 con->in_base_pos = 0;
994 con->in_tag = CEPH_MSGR_TAG_READY;
998 * Prepare to read a message.
1000 static int prepare_read_message(struct ceph_connection *con)
1002 dout("prepare_read_message %p\n", con);
1003 BUG_ON(con->in_msg != NULL);
1004 con->in_base_pos = 0;
1005 con->in_front_crc = con->in_middle_crc = con->in_data_crc = 0;
1010 static int read_partial(struct ceph_connection *con,
1011 int end, int size, void *object)
1013 while (con->in_base_pos < end) {
1014 int left = end - con->in_base_pos;
1015 int have = size - left;
1016 int ret = ceph_tcp_recvmsg(con->sock, object + have, left);
1019 con->in_base_pos += ret;
1026 * Read all or part of the connect-side handshake on a new connection
1028 static int read_partial_banner(struct ceph_connection *con)
1034 dout("read_partial_banner %p at %d\n", con, con->in_base_pos);
1037 size = strlen(CEPH_BANNER);
1039 ret = read_partial(con, end, size, con->in_banner);
1043 size = sizeof (con->actual_peer_addr);
1045 ret = read_partial(con, end, size, &con->actual_peer_addr);
1049 size = sizeof (con->peer_addr_for_me);
1051 ret = read_partial(con, end, size, &con->peer_addr_for_me);
1059 static int read_partial_connect(struct ceph_connection *con)
1065 dout("read_partial_connect %p at %d\n", con, con->in_base_pos);
1067 size = sizeof (con->in_reply);
1069 ret = read_partial(con, end, size, &con->in_reply);
1073 size = le32_to_cpu(con->in_reply.authorizer_len);
1075 ret = read_partial(con, end, size, con->auth_reply_buf);
1079 dout("read_partial_connect %p tag %d, con_seq = %u, g_seq = %u\n",
1080 con, (int)con->in_reply.tag,
1081 le32_to_cpu(con->in_reply.connect_seq),
1082 le32_to_cpu(con->in_reply.global_seq));
1089 * Verify the hello banner looks okay.
1091 static int verify_hello(struct ceph_connection *con)
1093 if (memcmp(con->in_banner, CEPH_BANNER, strlen(CEPH_BANNER))) {
1094 pr_err("connect to %s got bad banner\n",
1095 ceph_pr_addr(&con->peer_addr.in_addr));
1096 con->error_msg = "protocol error, bad banner";
1102 static bool addr_is_blank(struct sockaddr_storage *ss)
1104 switch (ss->ss_family) {
1106 return ((struct sockaddr_in *)ss)->sin_addr.s_addr == 0;
1109 ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[0] == 0 &&
1110 ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[1] == 0 &&
1111 ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[2] == 0 &&
1112 ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[3] == 0;
1117 static int addr_port(struct sockaddr_storage *ss)
1119 switch (ss->ss_family) {
1121 return ntohs(((struct sockaddr_in *)ss)->sin_port);
1123 return ntohs(((struct sockaddr_in6 *)ss)->sin6_port);
1128 static void addr_set_port(struct sockaddr_storage *ss, int p)
1130 switch (ss->ss_family) {
1132 ((struct sockaddr_in *)ss)->sin_port = htons(p);
1135 ((struct sockaddr_in6 *)ss)->sin6_port = htons(p);
1141 * Unlike other *_pton function semantics, zero indicates success.
1143 static int ceph_pton(const char *str, size_t len, struct sockaddr_storage *ss,
1144 char delim, const char **ipend)
1146 struct sockaddr_in *in4 = (struct sockaddr_in *) ss;
1147 struct sockaddr_in6 *in6 = (struct sockaddr_in6 *) ss;
1149 memset(ss, 0, sizeof(*ss));
1151 if (in4_pton(str, len, (u8 *)&in4->sin_addr.s_addr, delim, ipend)) {
1152 ss->ss_family = AF_INET;
1156 if (in6_pton(str, len, (u8 *)&in6->sin6_addr.s6_addr, delim, ipend)) {
1157 ss->ss_family = AF_INET6;
1165 * Extract hostname string and resolve using kernel DNS facility.
1167 #ifdef CONFIG_CEPH_LIB_USE_DNS_RESOLVER
1168 static int ceph_dns_resolve_name(const char *name, size_t namelen,
1169 struct sockaddr_storage *ss, char delim, const char **ipend)
1171 const char *end, *delim_p;
1172 char *colon_p, *ip_addr = NULL;
1176 * The end of the hostname occurs immediately preceding the delimiter or
1177 * the port marker (':') where the delimiter takes precedence.
1179 delim_p = memchr(name, delim, namelen);
1180 colon_p = memchr(name, ':', namelen);
1182 if (delim_p && colon_p)
1183 end = delim_p < colon_p ? delim_p : colon_p;
1184 else if (!delim_p && colon_p)
1188 if (!end) /* case: hostname:/ */
1189 end = name + namelen;
1195 /* do dns_resolve upcall */
1196 ip_len = dns_query(NULL, name, end - name, NULL, &ip_addr, NULL);
1198 ret = ceph_pton(ip_addr, ip_len, ss, -1, NULL);
1206 pr_info("resolve '%.*s' (ret=%d): %s\n", (int)(end - name), name,
1207 ret, ret ? "failed" : ceph_pr_addr(ss));
1212 static inline int ceph_dns_resolve_name(const char *name, size_t namelen,
1213 struct sockaddr_storage *ss, char delim, const char **ipend)
1220 * Parse a server name (IP or hostname). If a valid IP address is not found
1221 * then try to extract a hostname to resolve using userspace DNS upcall.
1223 static int ceph_parse_server_name(const char *name, size_t namelen,
1224 struct sockaddr_storage *ss, char delim, const char **ipend)
1228 ret = ceph_pton(name, namelen, ss, delim, ipend);
1230 ret = ceph_dns_resolve_name(name, namelen, ss, delim, ipend);
1236 * Parse an ip[:port] list into an addr array. Use the default
1237 * monitor port if a port isn't specified.
1239 int ceph_parse_ips(const char *c, const char *end,
1240 struct ceph_entity_addr *addr,
1241 int max_count, int *count)
1243 int i, ret = -EINVAL;
1246 dout("parse_ips on '%.*s'\n", (int)(end-c), c);
1247 for (i = 0; i < max_count; i++) {
1249 struct sockaddr_storage *ss = &addr[i].in_addr;
1258 ret = ceph_parse_server_name(p, end - p, ss, delim, &ipend);
1267 dout("missing matching ']'\n");
1274 if (p < end && *p == ':') {
1277 while (p < end && *p >= '0' && *p <= '9') {
1278 port = (port * 10) + (*p - '0');
1281 if (port > 65535 || port == 0)
1284 port = CEPH_MON_PORT;
1287 addr_set_port(ss, port);
1289 dout("parse_ips got %s\n", ceph_pr_addr(ss));
1306 pr_err("parse_ips bad ip '%.*s'\n", (int)(end - c), c);
1309 EXPORT_SYMBOL(ceph_parse_ips);
1311 static int process_banner(struct ceph_connection *con)
1313 dout("process_banner on %p\n", con);
1315 if (verify_hello(con) < 0)
1318 ceph_decode_addr(&con->actual_peer_addr);
1319 ceph_decode_addr(&con->peer_addr_for_me);
1322 * Make sure the other end is who we wanted. note that the other
1323 * end may not yet know their ip address, so if it's 0.0.0.0, give
1324 * them the benefit of the doubt.
1326 if (memcmp(&con->peer_addr, &con->actual_peer_addr,
1327 sizeof(con->peer_addr)) != 0 &&
1328 !(addr_is_blank(&con->actual_peer_addr.in_addr) &&
1329 con->actual_peer_addr.nonce == con->peer_addr.nonce)) {
1330 pr_warning("wrong peer, want %s/%d, got %s/%d\n",
1331 ceph_pr_addr(&con->peer_addr.in_addr),
1332 (int)le32_to_cpu(con->peer_addr.nonce),
1333 ceph_pr_addr(&con->actual_peer_addr.in_addr),
1334 (int)le32_to_cpu(con->actual_peer_addr.nonce));
1335 con->error_msg = "wrong peer at address";
1340 * did we learn our address?
1342 if (addr_is_blank(&con->msgr->inst.addr.in_addr)) {
1343 int port = addr_port(&con->msgr->inst.addr.in_addr);
1345 memcpy(&con->msgr->inst.addr.in_addr,
1346 &con->peer_addr_for_me.in_addr,
1347 sizeof(con->peer_addr_for_me.in_addr));
1348 addr_set_port(&con->msgr->inst.addr.in_addr, port);
1349 encode_my_addr(con->msgr);
1350 dout("process_banner learned my addr is %s\n",
1351 ceph_pr_addr(&con->msgr->inst.addr.in_addr));
1354 set_bit(NEGOTIATING, &con->state);
1355 prepare_read_connect(con);
1359 static void fail_protocol(struct ceph_connection *con)
1361 reset_connection(con);
1362 set_bit(CLOSED, &con->state); /* in case there's queued work */
1364 mutex_unlock(&con->mutex);
1365 if (con->ops->bad_proto)
1366 con->ops->bad_proto(con);
1367 mutex_lock(&con->mutex);
1370 static int process_connect(struct ceph_connection *con)
1372 u64 sup_feat = con->msgr->supported_features;
1373 u64 req_feat = con->msgr->required_features;
1374 u64 server_feat = le64_to_cpu(con->in_reply.features);
1377 dout("process_connect on %p tag %d\n", con, (int)con->in_tag);
1379 switch (con->in_reply.tag) {
1380 case CEPH_MSGR_TAG_FEATURES:
1381 pr_err("%s%lld %s feature set mismatch,"
1382 " my %llx < server's %llx, missing %llx\n",
1383 ENTITY_NAME(con->peer_name),
1384 ceph_pr_addr(&con->peer_addr.in_addr),
1385 sup_feat, server_feat, server_feat & ~sup_feat);
1386 con->error_msg = "missing required protocol features";
1390 case CEPH_MSGR_TAG_BADPROTOVER:
1391 pr_err("%s%lld %s protocol version mismatch,"
1392 " my %d != server's %d\n",
1393 ENTITY_NAME(con->peer_name),
1394 ceph_pr_addr(&con->peer_addr.in_addr),
1395 le32_to_cpu(con->out_connect.protocol_version),
1396 le32_to_cpu(con->in_reply.protocol_version));
1397 con->error_msg = "protocol version mismatch";
1401 case CEPH_MSGR_TAG_BADAUTHORIZER:
1403 dout("process_connect %p got BADAUTHORIZER attempt %d\n", con,
1405 if (con->auth_retry == 2) {
1406 con->error_msg = "connect authorization failure";
1409 con->auth_retry = 1;
1410 ceph_con_out_kvec_reset(con);
1411 ret = prepare_write_connect(con);
1414 prepare_read_connect(con);
1417 case CEPH_MSGR_TAG_RESETSESSION:
1419 * If we connected with a large connect_seq but the peer
1420 * has no record of a session with us (no connection, or
1421 * connect_seq == 0), they will send RESETSESION to indicate
1422 * that they must have reset their session, and may have
1425 dout("process_connect got RESET peer seq %u\n",
1426 le32_to_cpu(con->in_reply.connect_seq));
1427 pr_err("%s%lld %s connection reset\n",
1428 ENTITY_NAME(con->peer_name),
1429 ceph_pr_addr(&con->peer_addr.in_addr));
1430 reset_connection(con);
1431 ceph_con_out_kvec_reset(con);
1432 ret = prepare_write_connect(con);
1435 prepare_read_connect(con);
1437 /* Tell ceph about it. */
1438 mutex_unlock(&con->mutex);
1439 pr_info("reset on %s%lld\n", ENTITY_NAME(con->peer_name));
1440 if (con->ops->peer_reset)
1441 con->ops->peer_reset(con);
1442 mutex_lock(&con->mutex);
1443 if (test_bit(CLOSED, &con->state) ||
1444 test_bit(OPENING, &con->state))
1448 case CEPH_MSGR_TAG_RETRY_SESSION:
1450 * If we sent a smaller connect_seq than the peer has, try
1451 * again with a larger value.
1453 dout("process_connect got RETRY_SESSION my seq %u, peer %u\n",
1454 le32_to_cpu(con->out_connect.connect_seq),
1455 le32_to_cpu(con->in_reply.connect_seq));
1456 con->connect_seq = le32_to_cpu(con->in_reply.connect_seq);
1457 ceph_con_out_kvec_reset(con);
1458 ret = prepare_write_connect(con);
1461 prepare_read_connect(con);
1464 case CEPH_MSGR_TAG_RETRY_GLOBAL:
1466 * If we sent a smaller global_seq than the peer has, try
1467 * again with a larger value.
1469 dout("process_connect got RETRY_GLOBAL my %u peer_gseq %u\n",
1470 con->peer_global_seq,
1471 le32_to_cpu(con->in_reply.global_seq));
1472 get_global_seq(con->msgr,
1473 le32_to_cpu(con->in_reply.global_seq));
1474 ceph_con_out_kvec_reset(con);
1475 ret = prepare_write_connect(con);
1478 prepare_read_connect(con);
1481 case CEPH_MSGR_TAG_READY:
1482 if (req_feat & ~server_feat) {
1483 pr_err("%s%lld %s protocol feature mismatch,"
1484 " my required %llx > server's %llx, need %llx\n",
1485 ENTITY_NAME(con->peer_name),
1486 ceph_pr_addr(&con->peer_addr.in_addr),
1487 req_feat, server_feat, req_feat & ~server_feat);
1488 con->error_msg = "missing required protocol features";
1492 clear_bit(CONNECTING, &con->state);
1493 con->peer_global_seq = le32_to_cpu(con->in_reply.global_seq);
1495 con->peer_features = server_feat;
1496 dout("process_connect got READY gseq %d cseq %d (%d)\n",
1497 con->peer_global_seq,
1498 le32_to_cpu(con->in_reply.connect_seq),
1500 WARN_ON(con->connect_seq !=
1501 le32_to_cpu(con->in_reply.connect_seq));
1503 if (con->in_reply.flags & CEPH_MSG_CONNECT_LOSSY)
1504 set_bit(LOSSYTX, &con->state);
1506 prepare_read_tag(con);
1509 case CEPH_MSGR_TAG_WAIT:
1511 * If there is a connection race (we are opening
1512 * connections to each other), one of us may just have
1513 * to WAIT. This shouldn't happen if we are the
1516 pr_err("process_connect got WAIT as client\n");
1517 con->error_msg = "protocol error, got WAIT as client";
1521 pr_err("connect protocol error, will retry\n");
1522 con->error_msg = "protocol error, garbage tag during connect";
1530 * read (part of) an ack
1532 static int read_partial_ack(struct ceph_connection *con)
1534 int size = sizeof (con->in_temp_ack);
1537 return read_partial(con, end, size, &con->in_temp_ack);
1542 * We can finally discard anything that's been acked.
1544 static void process_ack(struct ceph_connection *con)
1547 u64 ack = le64_to_cpu(con->in_temp_ack);
1550 while (!list_empty(&con->out_sent)) {
1551 m = list_first_entry(&con->out_sent, struct ceph_msg,
1553 seq = le64_to_cpu(m->hdr.seq);
1556 dout("got ack for seq %llu type %d at %p\n", seq,
1557 le16_to_cpu(m->hdr.type), m);
1558 m->ack_stamp = jiffies;
1561 prepare_read_tag(con);
1567 static int read_partial_message_section(struct ceph_connection *con,
1568 struct kvec *section,
1569 unsigned int sec_len, u32 *crc)
1575 while (section->iov_len < sec_len) {
1576 BUG_ON(section->iov_base == NULL);
1577 left = sec_len - section->iov_len;
1578 ret = ceph_tcp_recvmsg(con->sock, (char *)section->iov_base +
1579 section->iov_len, left);
1582 section->iov_len += ret;
1584 if (section->iov_len == sec_len)
1585 *crc = crc32c(0, section->iov_base, section->iov_len);
1590 static struct ceph_msg *ceph_alloc_msg(struct ceph_connection *con,
1591 struct ceph_msg_header *hdr,
1595 static int read_partial_message_pages(struct ceph_connection *con,
1596 struct page **pages,
1597 unsigned int data_len, bool do_datacrc)
1603 left = min((int)(data_len - con->in_msg_pos.data_pos),
1604 (int)(PAGE_SIZE - con->in_msg_pos.page_pos));
1606 BUG_ON(pages == NULL);
1607 p = kmap(pages[con->in_msg_pos.page]);
1608 ret = ceph_tcp_recvmsg(con->sock, p + con->in_msg_pos.page_pos,
1610 if (ret > 0 && do_datacrc)
1612 crc32c(con->in_data_crc,
1613 p + con->in_msg_pos.page_pos, ret);
1614 kunmap(pages[con->in_msg_pos.page]);
1617 con->in_msg_pos.data_pos += ret;
1618 con->in_msg_pos.page_pos += ret;
1619 if (con->in_msg_pos.page_pos == PAGE_SIZE) {
1620 con->in_msg_pos.page_pos = 0;
1621 con->in_msg_pos.page++;
1628 static int read_partial_message_bio(struct ceph_connection *con,
1629 struct bio **bio_iter, int *bio_seg,
1630 unsigned int data_len, bool do_datacrc)
1632 struct bio_vec *bv = bio_iovec_idx(*bio_iter, *bio_seg);
1639 left = min((int)(data_len - con->in_msg_pos.data_pos),
1640 (int)(bv->bv_len - con->in_msg_pos.page_pos));
1642 p = kmap(bv->bv_page) + bv->bv_offset;
1644 ret = ceph_tcp_recvmsg(con->sock, p + con->in_msg_pos.page_pos,
1646 if (ret > 0 && do_datacrc)
1648 crc32c(con->in_data_crc,
1649 p + con->in_msg_pos.page_pos, ret);
1650 kunmap(bv->bv_page);
1653 con->in_msg_pos.data_pos += ret;
1654 con->in_msg_pos.page_pos += ret;
1655 if (con->in_msg_pos.page_pos == bv->bv_len) {
1656 con->in_msg_pos.page_pos = 0;
1657 iter_bio_next(bio_iter, bio_seg);
1665 * read (part of) a message.
1667 static int read_partial_message(struct ceph_connection *con)
1669 struct ceph_msg *m = con->in_msg;
1673 unsigned int front_len, middle_len, data_len;
1674 bool do_datacrc = !con->msgr->nocrc;
1679 dout("read_partial_message con %p msg %p\n", con, m);
1682 size = sizeof (con->in_hdr);
1684 ret = read_partial(con, end, size, &con->in_hdr);
1688 crc = crc32c(0, &con->in_hdr, offsetof(struct ceph_msg_header, crc));
1689 if (cpu_to_le32(crc) != con->in_hdr.crc) {
1690 pr_err("read_partial_message bad hdr "
1691 " crc %u != expected %u\n",
1692 crc, con->in_hdr.crc);
1696 front_len = le32_to_cpu(con->in_hdr.front_len);
1697 if (front_len > CEPH_MSG_MAX_FRONT_LEN)
1699 middle_len = le32_to_cpu(con->in_hdr.middle_len);
1700 if (middle_len > CEPH_MSG_MAX_DATA_LEN)
1702 data_len = le32_to_cpu(con->in_hdr.data_len);
1703 if (data_len > CEPH_MSG_MAX_DATA_LEN)
1707 seq = le64_to_cpu(con->in_hdr.seq);
1708 if ((s64)seq - (s64)con->in_seq < 1) {
1709 pr_info("skipping %s%lld %s seq %lld expected %lld\n",
1710 ENTITY_NAME(con->peer_name),
1711 ceph_pr_addr(&con->peer_addr.in_addr),
1712 seq, con->in_seq + 1);
1713 con->in_base_pos = -front_len - middle_len - data_len -
1715 con->in_tag = CEPH_MSGR_TAG_READY;
1717 } else if ((s64)seq - (s64)con->in_seq > 1) {
1718 pr_err("read_partial_message bad seq %lld expected %lld\n",
1719 seq, con->in_seq + 1);
1720 con->error_msg = "bad message sequence # for incoming message";
1724 /* allocate message? */
1726 dout("got hdr type %d front %d data %d\n", con->in_hdr.type,
1727 con->in_hdr.front_len, con->in_hdr.data_len);
1729 con->in_msg = ceph_alloc_msg(con, &con->in_hdr, &skip);
1731 /* skip this message */
1732 dout("alloc_msg said skip message\n");
1733 BUG_ON(con->in_msg);
1734 con->in_base_pos = -front_len - middle_len - data_len -
1736 con->in_tag = CEPH_MSGR_TAG_READY;
1742 "error allocating memory for incoming message";
1746 m->front.iov_len = 0; /* haven't read it yet */
1748 m->middle->vec.iov_len = 0;
1750 con->in_msg_pos.page = 0;
1752 con->in_msg_pos.page_pos = m->page_alignment;
1754 con->in_msg_pos.page_pos = 0;
1755 con->in_msg_pos.data_pos = 0;
1759 ret = read_partial_message_section(con, &m->front, front_len,
1760 &con->in_front_crc);
1766 ret = read_partial_message_section(con, &m->middle->vec,
1768 &con->in_middle_crc);
1773 if (m->bio && !m->bio_iter)
1774 init_bio_iter(m->bio, &m->bio_iter, &m->bio_seg);
1778 while (con->in_msg_pos.data_pos < data_len) {
1780 ret = read_partial_message_pages(con, m->pages,
1781 data_len, do_datacrc);
1785 } else if (m->bio) {
1787 ret = read_partial_message_bio(con,
1788 &m->bio_iter, &m->bio_seg,
1789 data_len, do_datacrc);
1799 size = sizeof (m->footer);
1801 ret = read_partial(con, end, size, &m->footer);
1805 dout("read_partial_message got msg %p %d (%u) + %d (%u) + %d (%u)\n",
1806 m, front_len, m->footer.front_crc, middle_len,
1807 m->footer.middle_crc, data_len, m->footer.data_crc);
1810 if (con->in_front_crc != le32_to_cpu(m->footer.front_crc)) {
1811 pr_err("read_partial_message %p front crc %u != exp. %u\n",
1812 m, con->in_front_crc, m->footer.front_crc);
1815 if (con->in_middle_crc != le32_to_cpu(m->footer.middle_crc)) {
1816 pr_err("read_partial_message %p middle crc %u != exp %u\n",
1817 m, con->in_middle_crc, m->footer.middle_crc);
1821 (m->footer.flags & CEPH_MSG_FOOTER_NOCRC) == 0 &&
1822 con->in_data_crc != le32_to_cpu(m->footer.data_crc)) {
1823 pr_err("read_partial_message %p data crc %u != exp. %u\n", m,
1824 con->in_data_crc, le32_to_cpu(m->footer.data_crc));
1828 return 1; /* done! */
1832 * Process message. This happens in the worker thread. The callback should
1833 * be careful not to do anything that waits on other incoming messages or it
1836 static void process_message(struct ceph_connection *con)
1838 struct ceph_msg *msg;
1843 /* if first message, set peer_name */
1844 if (con->peer_name.type == 0)
1845 con->peer_name = msg->hdr.src;
1848 mutex_unlock(&con->mutex);
1850 dout("===== %p %llu from %s%lld %d=%s len %d+%d (%u %u %u) =====\n",
1851 msg, le64_to_cpu(msg->hdr.seq),
1852 ENTITY_NAME(msg->hdr.src),
1853 le16_to_cpu(msg->hdr.type),
1854 ceph_msg_type_name(le16_to_cpu(msg->hdr.type)),
1855 le32_to_cpu(msg->hdr.front_len),
1856 le32_to_cpu(msg->hdr.data_len),
1857 con->in_front_crc, con->in_middle_crc, con->in_data_crc);
1858 con->ops->dispatch(con, msg);
1860 mutex_lock(&con->mutex);
1861 prepare_read_tag(con);
1866 * Write something to the socket. Called in a worker thread when the
1867 * socket appears to be writeable and we have something ready to send.
1869 static int try_write(struct ceph_connection *con)
1873 dout("try_write start %p state %lu nref %d\n", con, con->state,
1874 atomic_read(&con->nref));
1877 dout("try_write out_kvec_bytes %d\n", con->out_kvec_bytes);
1879 /* open the socket first? */
1880 if (con->sock == NULL) {
1881 ceph_con_out_kvec_reset(con);
1882 prepare_write_banner(con);
1883 ret = prepare_write_connect(con);
1886 prepare_read_banner(con);
1887 set_bit(CONNECTING, &con->state);
1888 clear_bit(NEGOTIATING, &con->state);
1890 BUG_ON(con->in_msg);
1891 con->in_tag = CEPH_MSGR_TAG_READY;
1892 dout("try_write initiating connect on %p new state %lu\n",
1894 ret = ceph_tcp_connect(con);
1896 con->error_msg = "connect error";
1902 /* kvec data queued? */
1903 if (con->out_skip) {
1904 ret = write_partial_skip(con);
1908 if (con->out_kvec_left) {
1909 ret = write_partial_kvec(con);
1916 if (con->out_msg_done) {
1917 ceph_msg_put(con->out_msg);
1918 con->out_msg = NULL; /* we're done with this one */
1922 ret = write_partial_msg_pages(con);
1924 goto more_kvec; /* we need to send the footer, too! */
1928 dout("try_write write_partial_msg_pages err %d\n",
1935 if (!test_bit(CONNECTING, &con->state)) {
1936 /* is anything else pending? */
1937 if (!list_empty(&con->out_queue)) {
1938 prepare_write_message(con);
1941 if (con->in_seq > con->in_seq_acked) {
1942 prepare_write_ack(con);
1945 if (test_and_clear_bit(KEEPALIVE_PENDING, &con->state)) {
1946 prepare_write_keepalive(con);
1951 /* Nothing to do! */
1952 clear_bit(WRITE_PENDING, &con->state);
1953 dout("try_write nothing else to write.\n");
1956 dout("try_write done on %p ret %d\n", con, ret);
1963 * Read what we can from the socket.
1965 static int try_read(struct ceph_connection *con)
1972 if (test_bit(STANDBY, &con->state))
1975 dout("try_read start on %p\n", con);
1978 dout("try_read tag %d in_base_pos %d\n", (int)con->in_tag,
1982 * process_connect and process_message drop and re-take
1983 * con->mutex. make sure we handle a racing close or reopen.
1985 if (test_bit(CLOSED, &con->state) ||
1986 test_bit(OPENING, &con->state)) {
1991 if (test_bit(CONNECTING, &con->state)) {
1992 if (!test_bit(NEGOTIATING, &con->state)) {
1993 dout("try_read connecting\n");
1994 ret = read_partial_banner(con);
1997 ret = process_banner(con);
2001 ret = read_partial_connect(con);
2004 ret = process_connect(con);
2010 if (con->in_base_pos < 0) {
2012 * skipping + discarding content.
2014 * FIXME: there must be a better way to do this!
2016 static char buf[SKIP_BUF_SIZE];
2017 int skip = min((int) sizeof (buf), -con->in_base_pos);
2019 dout("skipping %d / %d bytes\n", skip, -con->in_base_pos);
2020 ret = ceph_tcp_recvmsg(con->sock, buf, skip);
2023 con->in_base_pos += ret;
2024 if (con->in_base_pos)
2027 if (con->in_tag == CEPH_MSGR_TAG_READY) {
2031 ret = ceph_tcp_recvmsg(con->sock, &con->in_tag, 1);
2034 dout("try_read got tag %d\n", (int)con->in_tag);
2035 switch (con->in_tag) {
2036 case CEPH_MSGR_TAG_MSG:
2037 prepare_read_message(con);
2039 case CEPH_MSGR_TAG_ACK:
2040 prepare_read_ack(con);
2042 case CEPH_MSGR_TAG_CLOSE:
2043 set_bit(CLOSED, &con->state); /* fixme */
2049 if (con->in_tag == CEPH_MSGR_TAG_MSG) {
2050 ret = read_partial_message(con);
2054 con->error_msg = "bad crc";
2058 con->error_msg = "io error";
2063 if (con->in_tag == CEPH_MSGR_TAG_READY)
2065 process_message(con);
2068 if (con->in_tag == CEPH_MSGR_TAG_ACK) {
2069 ret = read_partial_ack(con);
2077 dout("try_read done on %p ret %d\n", con, ret);
2081 pr_err("try_read bad con->in_tag = %d\n", (int)con->in_tag);
2082 con->error_msg = "protocol error, garbage tag";
2089 * Atomically queue work on a connection. Bump @con reference to
2090 * avoid races with connection teardown.
2092 static void queue_con(struct ceph_connection *con)
2094 if (test_bit(DEAD, &con->state)) {
2095 dout("queue_con %p ignoring: DEAD\n",
2100 if (!con->ops->get(con)) {
2101 dout("queue_con %p ref count 0\n", con);
2105 if (!queue_delayed_work(ceph_msgr_wq, &con->work, 0)) {
2106 dout("queue_con %p - already queued\n", con);
2109 dout("queue_con %p\n", con);
2114 * Do some work on a connection. Drop a connection ref when we're done.
2116 static void con_work(struct work_struct *work)
2118 struct ceph_connection *con = container_of(work, struct ceph_connection,
2122 mutex_lock(&con->mutex);
2124 if (test_and_clear_bit(BACKOFF, &con->state)) {
2125 dout("con_work %p backing off\n", con);
2126 if (queue_delayed_work(ceph_msgr_wq, &con->work,
2127 round_jiffies_relative(con->delay))) {
2128 dout("con_work %p backoff %lu\n", con, con->delay);
2129 mutex_unlock(&con->mutex);
2133 dout("con_work %p FAILED to back off %lu\n", con,
2138 if (test_bit(STANDBY, &con->state)) {
2139 dout("con_work %p STANDBY\n", con);
2142 if (test_bit(CLOSED, &con->state)) { /* e.g. if we are replaced */
2143 dout("con_work CLOSED\n");
2144 con_close_socket(con);
2147 if (test_and_clear_bit(OPENING, &con->state)) {
2148 /* reopen w/ new peer */
2149 dout("con_work OPENING\n");
2150 con_close_socket(con);
2153 if (test_and_clear_bit(SOCK_CLOSED, &con->state))
2156 ret = try_read(con);
2162 ret = try_write(con);
2169 mutex_unlock(&con->mutex);
2175 mutex_unlock(&con->mutex);
2176 ceph_fault(con); /* error/fault path */
2182 * Generic error/fault handler. A retry mechanism is used with
2183 * exponential backoff
2185 static void ceph_fault(struct ceph_connection *con)
2187 pr_err("%s%lld %s %s\n", ENTITY_NAME(con->peer_name),
2188 ceph_pr_addr(&con->peer_addr.in_addr), con->error_msg);
2189 dout("fault %p state %lu to peer %s\n",
2190 con, con->state, ceph_pr_addr(&con->peer_addr.in_addr));
2192 if (test_bit(LOSSYTX, &con->state)) {
2193 dout("fault on LOSSYTX channel\n");
2197 mutex_lock(&con->mutex);
2198 if (test_bit(CLOSED, &con->state))
2201 con_close_socket(con);
2204 ceph_msg_put(con->in_msg);
2208 /* Requeue anything that hasn't been acked */
2209 list_splice_init(&con->out_sent, &con->out_queue);
2211 /* If there are no messages queued or keepalive pending, place
2212 * the connection in a STANDBY state */
2213 if (list_empty(&con->out_queue) &&
2214 !test_bit(KEEPALIVE_PENDING, &con->state)) {
2215 dout("fault %p setting STANDBY clearing WRITE_PENDING\n", con);
2216 clear_bit(WRITE_PENDING, &con->state);
2217 set_bit(STANDBY, &con->state);
2219 /* retry after a delay. */
2220 if (con->delay == 0)
2221 con->delay = BASE_DELAY_INTERVAL;
2222 else if (con->delay < MAX_DELAY_INTERVAL)
2225 if (queue_delayed_work(ceph_msgr_wq, &con->work,
2226 round_jiffies_relative(con->delay))) {
2227 dout("fault queued %p delay %lu\n", con, con->delay);
2230 dout("fault failed to queue %p delay %lu, backoff\n",
2233 * In many cases we see a socket state change
2234 * while con_work is running and end up
2235 * queuing (non-delayed) work, such that we
2236 * can't backoff with a delay. Set a flag so
2237 * that when con_work restarts we schedule the
2240 set_bit(BACKOFF, &con->state);
2245 mutex_unlock(&con->mutex);
2248 * in case we faulted due to authentication, invalidate our
2249 * current tickets so that we can get new ones.
2251 if (con->auth_retry && con->ops->invalidate_authorizer) {
2252 dout("calling invalidate_authorizer()\n");
2253 con->ops->invalidate_authorizer(con);
2256 if (con->ops->fault)
2257 con->ops->fault(con);
2263 * create a new messenger instance
2265 struct ceph_messenger *ceph_messenger_create(struct ceph_entity_addr *myaddr,
2266 u32 supported_features,
2267 u32 required_features)
2269 struct ceph_messenger *msgr;
2271 msgr = kzalloc(sizeof(*msgr), GFP_KERNEL);
2273 return ERR_PTR(-ENOMEM);
2275 msgr->supported_features = supported_features;
2276 msgr->required_features = required_features;
2278 spin_lock_init(&msgr->global_seq_lock);
2281 msgr->inst.addr = *myaddr;
2283 /* select a random nonce */
2284 msgr->inst.addr.type = 0;
2285 get_random_bytes(&msgr->inst.addr.nonce, sizeof(msgr->inst.addr.nonce));
2286 encode_my_addr(msgr);
2288 dout("messenger_create %p\n", msgr);
2291 EXPORT_SYMBOL(ceph_messenger_create);
2293 void ceph_messenger_destroy(struct ceph_messenger *msgr)
2295 dout("destroy %p\n", msgr);
2297 dout("destroyed messenger %p\n", msgr);
2299 EXPORT_SYMBOL(ceph_messenger_destroy);
2301 static void clear_standby(struct ceph_connection *con)
2303 /* come back from STANDBY? */
2304 if (test_and_clear_bit(STANDBY, &con->state)) {
2305 mutex_lock(&con->mutex);
2306 dout("clear_standby %p and ++connect_seq\n", con);
2308 WARN_ON(test_bit(WRITE_PENDING, &con->state));
2309 WARN_ON(test_bit(KEEPALIVE_PENDING, &con->state));
2310 mutex_unlock(&con->mutex);
2315 * Queue up an outgoing message on the given connection.
2317 void ceph_con_send(struct ceph_connection *con, struct ceph_msg *msg)
2319 if (test_bit(CLOSED, &con->state)) {
2320 dout("con_send %p closed, dropping %p\n", con, msg);
2326 msg->hdr.src = con->msgr->inst.name;
2328 BUG_ON(msg->front.iov_len != le32_to_cpu(msg->hdr.front_len));
2330 msg->needs_out_seq = true;
2333 mutex_lock(&con->mutex);
2334 BUG_ON(!list_empty(&msg->list_head));
2335 list_add_tail(&msg->list_head, &con->out_queue);
2336 dout("----- %p to %s%lld %d=%s len %d+%d+%d -----\n", msg,
2337 ENTITY_NAME(con->peer_name), le16_to_cpu(msg->hdr.type),
2338 ceph_msg_type_name(le16_to_cpu(msg->hdr.type)),
2339 le32_to_cpu(msg->hdr.front_len),
2340 le32_to_cpu(msg->hdr.middle_len),
2341 le32_to_cpu(msg->hdr.data_len));
2342 mutex_unlock(&con->mutex);
2344 /* if there wasn't anything waiting to send before, queue
2347 if (test_and_set_bit(WRITE_PENDING, &con->state) == 0)
2350 EXPORT_SYMBOL(ceph_con_send);
2353 * Revoke a message that was previously queued for send
2355 void ceph_con_revoke(struct ceph_connection *con, struct ceph_msg *msg)
2357 mutex_lock(&con->mutex);
2358 if (!list_empty(&msg->list_head)) {
2359 dout("con_revoke %p msg %p - was on queue\n", con, msg);
2360 list_del_init(&msg->list_head);
2364 if (con->out_msg == msg) {
2365 dout("con_revoke %p msg %p - was sending\n", con, msg);
2366 con->out_msg = NULL;
2367 if (con->out_kvec_is_msg) {
2368 con->out_skip = con->out_kvec_bytes;
2369 con->out_kvec_is_msg = false;
2374 mutex_unlock(&con->mutex);
2378 * Revoke a message that we may be reading data into
2380 void ceph_con_revoke_message(struct ceph_connection *con, struct ceph_msg *msg)
2382 mutex_lock(&con->mutex);
2383 if (con->in_msg && con->in_msg == msg) {
2384 unsigned int front_len = le32_to_cpu(con->in_hdr.front_len);
2385 unsigned int middle_len = le32_to_cpu(con->in_hdr.middle_len);
2386 unsigned int data_len = le32_to_cpu(con->in_hdr.data_len);
2388 /* skip rest of message */
2389 dout("con_revoke_pages %p msg %p revoked\n", con, msg);
2390 con->in_base_pos = con->in_base_pos -
2391 sizeof(struct ceph_msg_header) -
2395 sizeof(struct ceph_msg_footer);
2396 ceph_msg_put(con->in_msg);
2398 con->in_tag = CEPH_MSGR_TAG_READY;
2401 dout("con_revoke_pages %p msg %p pages %p no-op\n",
2402 con, con->in_msg, msg);
2404 mutex_unlock(&con->mutex);
2408 * Queue a keepalive byte to ensure the tcp connection is alive.
2410 void ceph_con_keepalive(struct ceph_connection *con)
2412 dout("con_keepalive %p\n", con);
2414 if (test_and_set_bit(KEEPALIVE_PENDING, &con->state) == 0 &&
2415 test_and_set_bit(WRITE_PENDING, &con->state) == 0)
2418 EXPORT_SYMBOL(ceph_con_keepalive);
2422 * construct a new message with given type, size
2423 * the new msg has a ref count of 1.
2425 struct ceph_msg *ceph_msg_new(int type, int front_len, gfp_t flags,
2430 m = kmalloc(sizeof(*m), flags);
2433 kref_init(&m->kref);
2434 INIT_LIST_HEAD(&m->list_head);
2437 m->hdr.type = cpu_to_le16(type);
2438 m->hdr.priority = cpu_to_le16(CEPH_MSG_PRIO_DEFAULT);
2440 m->hdr.front_len = cpu_to_le32(front_len);
2441 m->hdr.middle_len = 0;
2442 m->hdr.data_len = 0;
2443 m->hdr.data_off = 0;
2444 m->hdr.reserved = 0;
2445 m->footer.front_crc = 0;
2446 m->footer.middle_crc = 0;
2447 m->footer.data_crc = 0;
2448 m->footer.flags = 0;
2449 m->front_max = front_len;
2450 m->front_is_vmalloc = false;
2451 m->more_to_follow = false;
2460 m->page_alignment = 0;
2470 if (front_len > PAGE_CACHE_SIZE) {
2471 m->front.iov_base = __vmalloc(front_len, flags,
2473 m->front_is_vmalloc = true;
2475 m->front.iov_base = kmalloc(front_len, flags);
2477 if (m->front.iov_base == NULL) {
2478 dout("ceph_msg_new can't allocate %d bytes\n",
2483 m->front.iov_base = NULL;
2485 m->front.iov_len = front_len;
2487 dout("ceph_msg_new %p front %d\n", m, front_len);
2494 pr_err("msg_new can't create type %d front %d\n", type,
2498 dout("msg_new can't create type %d front %d\n", type,
2503 EXPORT_SYMBOL(ceph_msg_new);
2506 * Allocate "middle" portion of a message, if it is needed and wasn't
2507 * allocated by alloc_msg. This allows us to read a small fixed-size
2508 * per-type header in the front and then gracefully fail (i.e.,
2509 * propagate the error to the caller based on info in the front) when
2510 * the middle is too large.
2512 static int ceph_alloc_middle(struct ceph_connection *con, struct ceph_msg *msg)
2514 int type = le16_to_cpu(msg->hdr.type);
2515 int middle_len = le32_to_cpu(msg->hdr.middle_len);
2517 dout("alloc_middle %p type %d %s middle_len %d\n", msg, type,
2518 ceph_msg_type_name(type), middle_len);
2519 BUG_ON(!middle_len);
2520 BUG_ON(msg->middle);
2522 msg->middle = ceph_buffer_new(middle_len, GFP_NOFS);
2529 * Generic message allocator, for incoming messages.
2531 static struct ceph_msg *ceph_alloc_msg(struct ceph_connection *con,
2532 struct ceph_msg_header *hdr,
2535 int type = le16_to_cpu(hdr->type);
2536 int front_len = le32_to_cpu(hdr->front_len);
2537 int middle_len = le32_to_cpu(hdr->middle_len);
2538 struct ceph_msg *msg = NULL;
2541 if (con->ops->alloc_msg) {
2542 mutex_unlock(&con->mutex);
2543 msg = con->ops->alloc_msg(con, hdr, skip);
2544 mutex_lock(&con->mutex);
2550 msg = ceph_msg_new(type, front_len, GFP_NOFS, false);
2552 pr_err("unable to allocate msg type %d len %d\n",
2556 msg->page_alignment = le16_to_cpu(hdr->data_off);
2558 memcpy(&msg->hdr, &con->in_hdr, sizeof(con->in_hdr));
2560 if (middle_len && !msg->middle) {
2561 ret = ceph_alloc_middle(con, msg);
2573 * Free a generically kmalloc'd message.
2575 void ceph_msg_kfree(struct ceph_msg *m)
2577 dout("msg_kfree %p\n", m);
2578 if (m->front_is_vmalloc)
2579 vfree(m->front.iov_base);
2581 kfree(m->front.iov_base);
2586 * Drop a msg ref. Destroy as needed.
2588 void ceph_msg_last_put(struct kref *kref)
2590 struct ceph_msg *m = container_of(kref, struct ceph_msg, kref);
2592 dout("ceph_msg_put last one on %p\n", m);
2593 WARN_ON(!list_empty(&m->list_head));
2595 /* drop middle, data, if any */
2597 ceph_buffer_put(m->middle);
2604 ceph_pagelist_release(m->pagelist);
2612 ceph_msgpool_put(m->pool, m);
2616 EXPORT_SYMBOL(ceph_msg_last_put);
2618 void ceph_msg_dump(struct ceph_msg *msg)
2620 pr_debug("msg_dump %p (front_max %d nr_pages %d)\n", msg,
2621 msg->front_max, msg->nr_pages);
2622 print_hex_dump(KERN_DEBUG, "header: ",
2623 DUMP_PREFIX_OFFSET, 16, 1,
2624 &msg->hdr, sizeof(msg->hdr), true);
2625 print_hex_dump(KERN_DEBUG, " front: ",
2626 DUMP_PREFIX_OFFSET, 16, 1,
2627 msg->front.iov_base, msg->front.iov_len, true);
2629 print_hex_dump(KERN_DEBUG, "middle: ",
2630 DUMP_PREFIX_OFFSET, 16, 1,
2631 msg->middle->vec.iov_base,
2632 msg->middle->vec.iov_len, true);
2633 print_hex_dump(KERN_DEBUG, "footer: ",
2634 DUMP_PREFIX_OFFSET, 16, 1,
2635 &msg->footer, sizeof(msg->footer), true);
2637 EXPORT_SYMBOL(ceph_msg_dump);