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#ifdef	CONFIG_BLOCK
13#include <linux/bio.h>
14#endif	/* CONFIG_BLOCK */
15#include <linux/dns_resolver.h>
16#include <net/tcp.h>
17
18#include <linux/ceph/ceph_features.h>
19#include <linux/ceph/libceph.h>
20#include <linux/ceph/messenger.h>
21#include <linux/ceph/decode.h>
22#include <linux/ceph/pagelist.h>
23#include <linux/export.h>
24
25#define list_entry_next(pos, member)					\
26	list_entry(pos->member.next, typeof(*pos), member)
27
28/*
29 * Ceph uses the messenger to exchange ceph_msg messages with other
30 * hosts in the system.  The messenger provides ordered and reliable
31 * delivery.  We tolerate TCP disconnects by reconnecting (with
32 * exponential backoff) in the case of a fault (disconnection, bad
33 * crc, protocol error).  Acks allow sent messages to be discarded by
34 * the sender.
35 */
36
37/*
38 * We track the state of the socket on a given connection using
39 * values defined below.  The transition to a new socket state is
40 * handled by a function which verifies we aren't coming from an
41 * unexpected state.
42 *
43 *      --------
44 *      | NEW* |  transient initial state
45 *      --------
46 *          | con_sock_state_init()
47 *          v
48 *      ----------
49 *      | CLOSED |  initialized, but no socket (and no
50 *      ----------  TCP connection)
51 *       ^      \
52 *       |       \ con_sock_state_connecting()
53 *       |        ----------------------
54 *       |                              \
55 *       + con_sock_state_closed()       \
56 *       |+---------------------------    \
57 *       | \                          \    \
58 *       |  -----------                \    \
59 *       |  | CLOSING |  socket event;  \    \
60 *       |  -----------  await close     \    \
61 *       |       ^                        \   |
62 *       |       |                         \  |
63 *       |       + con_sock_state_closing() \ |
64 *       |      / \                         | |
65 *       |     /   ---------------          | |
66 *       |    /                   \         v v
67 *       |   /                    --------------
68 *       |  /    -----------------| CONNECTING |  socket created, TCP
69 *       |  |   /                 --------------  connect initiated
70 *       |  |   | con_sock_state_connected()
71 *       |  |   v
72 *      -------------
73 *      | CONNECTED |  TCP connection established
74 *      -------------
75 *
76 * State values for ceph_connection->sock_state; NEW is assumed to be 0.
77 */
78
79#define CON_SOCK_STATE_NEW		0	/* -> CLOSED */
80#define CON_SOCK_STATE_CLOSED		1	/* -> CONNECTING */
81#define CON_SOCK_STATE_CONNECTING	2	/* -> CONNECTED or -> CLOSING */
82#define CON_SOCK_STATE_CONNECTED	3	/* -> CLOSING or -> CLOSED */
83#define CON_SOCK_STATE_CLOSING		4	/* -> CLOSED */
84
85/*
86 * connection states
87 */
88#define CON_STATE_CLOSED        1  /* -> PREOPEN */
89#define CON_STATE_PREOPEN       2  /* -> CONNECTING, CLOSED */
90#define CON_STATE_CONNECTING    3  /* -> NEGOTIATING, CLOSED */
91#define CON_STATE_NEGOTIATING   4  /* -> OPEN, CLOSED */
92#define CON_STATE_OPEN          5  /* -> STANDBY, CLOSED */
93#define CON_STATE_STANDBY       6  /* -> PREOPEN, CLOSED */
94
95/*
96 * ceph_connection flag bits
97 */
98#define CON_FLAG_LOSSYTX           0  /* we can close channel or drop
99				       * messages on errors */
100#define CON_FLAG_KEEPALIVE_PENDING 1  /* we need to send a keepalive */
101#define CON_FLAG_WRITE_PENDING	   2  /* we have data ready to send */
102#define CON_FLAG_SOCK_CLOSED	   3  /* socket state changed to closed */
103#define CON_FLAG_BACKOFF           4  /* need to retry queuing delayed work */
104
105static bool con_flag_valid(unsigned long con_flag)
106{
107	switch (con_flag) {
108	case CON_FLAG_LOSSYTX:
109	case CON_FLAG_KEEPALIVE_PENDING:
110	case CON_FLAG_WRITE_PENDING:
111	case CON_FLAG_SOCK_CLOSED:
112	case CON_FLAG_BACKOFF:
113		return true;
114	default:
115		return false;
116	}
117}
118
119static void con_flag_clear(struct ceph_connection *con, unsigned long con_flag)
120{
121	BUG_ON(!con_flag_valid(con_flag));
122
123	clear_bit(con_flag, &con->flags);
124}
125
126static void con_flag_set(struct ceph_connection *con, unsigned long con_flag)
127{
128	BUG_ON(!con_flag_valid(con_flag));
129
130	set_bit(con_flag, &con->flags);
131}
132
133static bool con_flag_test(struct ceph_connection *con, unsigned long con_flag)
134{
135	BUG_ON(!con_flag_valid(con_flag));
136
137	return test_bit(con_flag, &con->flags);
138}
139
140static bool con_flag_test_and_clear(struct ceph_connection *con,
141					unsigned long con_flag)
142{
143	BUG_ON(!con_flag_valid(con_flag));
144
145	return test_and_clear_bit(con_flag, &con->flags);
146}
147
148static bool con_flag_test_and_set(struct ceph_connection *con,
149					unsigned long con_flag)
150{
151	BUG_ON(!con_flag_valid(con_flag));
152
153	return test_and_set_bit(con_flag, &con->flags);
154}
155
156/* Slab caches for frequently-allocated structures */
157
158static struct kmem_cache	*ceph_msg_cache;
159static struct kmem_cache	*ceph_msg_data_cache;
160
161/* static tag bytes (protocol control messages) */
162static char tag_msg = CEPH_MSGR_TAG_MSG;
163static char tag_ack = CEPH_MSGR_TAG_ACK;
164static char tag_keepalive = CEPH_MSGR_TAG_KEEPALIVE;
165
166#ifdef CONFIG_LOCKDEP
167static struct lock_class_key socket_class;
168#endif
169
170/*
171 * When skipping (ignoring) a block of input we read it into a "skip
172 * buffer," which is this many bytes in size.
173 */
174#define SKIP_BUF_SIZE	1024
175
176static void queue_con(struct ceph_connection *con);
177static void cancel_con(struct ceph_connection *con);
178static void con_work(struct work_struct *);
179static void con_fault(struct ceph_connection *con);
180
181/*
182 * Nicely render a sockaddr as a string.  An array of formatted
183 * strings is used, to approximate reentrancy.
184 */
185#define ADDR_STR_COUNT_LOG	5	/* log2(# address strings in array) */
186#define ADDR_STR_COUNT		(1 << ADDR_STR_COUNT_LOG)
187#define ADDR_STR_COUNT_MASK	(ADDR_STR_COUNT - 1)
188#define MAX_ADDR_STR_LEN	64	/* 54 is enough */
189
190static char addr_str[ADDR_STR_COUNT][MAX_ADDR_STR_LEN];
191static atomic_t addr_str_seq = ATOMIC_INIT(0);
192
193static struct page *zero_page;		/* used in certain error cases */
194
195const char *ceph_pr_addr(const struct sockaddr_storage *ss)
196{
197	int i;
198	char *s;
199	struct sockaddr_in *in4 = (struct sockaddr_in *) ss;
200	struct sockaddr_in6 *in6 = (struct sockaddr_in6 *) ss;
201
202	i = atomic_inc_return(&addr_str_seq) & ADDR_STR_COUNT_MASK;
203	s = addr_str[i];
204
205	switch (ss->ss_family) {
206	case AF_INET:
207		snprintf(s, MAX_ADDR_STR_LEN, "%pI4:%hu", &in4->sin_addr,
208			 ntohs(in4->sin_port));
209		break;
210
211	case AF_INET6:
212		snprintf(s, MAX_ADDR_STR_LEN, "[%pI6c]:%hu", &in6->sin6_addr,
213			 ntohs(in6->sin6_port));
214		break;
215
216	default:
217		snprintf(s, MAX_ADDR_STR_LEN, "(unknown sockaddr family %hu)",
218			 ss->ss_family);
219	}
220
221	return s;
222}
223EXPORT_SYMBOL(ceph_pr_addr);
224
225static void encode_my_addr(struct ceph_messenger *msgr)
226{
227	memcpy(&msgr->my_enc_addr, &msgr->inst.addr, sizeof(msgr->my_enc_addr));
228	ceph_encode_addr(&msgr->my_enc_addr);
229}
230
231/*
232 * work queue for all reading and writing to/from the socket.
233 */
234static struct workqueue_struct *ceph_msgr_wq;
235
236static int ceph_msgr_slab_init(void)
237{
238	BUG_ON(ceph_msg_cache);
239	ceph_msg_cache = kmem_cache_create("ceph_msg",
240					sizeof (struct ceph_msg),
241					__alignof__(struct ceph_msg), 0, NULL);
242
243	if (!ceph_msg_cache)
244		return -ENOMEM;
245
246	BUG_ON(ceph_msg_data_cache);
247	ceph_msg_data_cache = kmem_cache_create("ceph_msg_data",
248					sizeof (struct ceph_msg_data),
249					__alignof__(struct ceph_msg_data),
250					0, NULL);
251	if (ceph_msg_data_cache)
252		return 0;
253
254	kmem_cache_destroy(ceph_msg_cache);
255	ceph_msg_cache = NULL;
256
257	return -ENOMEM;
258}
259
260static void ceph_msgr_slab_exit(void)
261{
262	BUG_ON(!ceph_msg_data_cache);
263	kmem_cache_destroy(ceph_msg_data_cache);
264	ceph_msg_data_cache = NULL;
265
266	BUG_ON(!ceph_msg_cache);
267	kmem_cache_destroy(ceph_msg_cache);
268	ceph_msg_cache = NULL;
269}
270
271static void _ceph_msgr_exit(void)
272{
273	if (ceph_msgr_wq) {
274		destroy_workqueue(ceph_msgr_wq);
275		ceph_msgr_wq = NULL;
276	}
277
278	ceph_msgr_slab_exit();
279
280	BUG_ON(zero_page == NULL);
281	kunmap(zero_page);
282	page_cache_release(zero_page);
283	zero_page = NULL;
284}
285
286int ceph_msgr_init(void)
287{
288	BUG_ON(zero_page != NULL);
289	zero_page = ZERO_PAGE(0);
290	page_cache_get(zero_page);
291
292	if (ceph_msgr_slab_init())
293		return -ENOMEM;
294
295	/*
296	 * The number of active work items is limited by the number of
297	 * connections, so leave @max_active at default.
298	 */
299	ceph_msgr_wq = alloc_workqueue("ceph-msgr", WQ_MEM_RECLAIM, 0);
300	if (ceph_msgr_wq)
301		return 0;
302
303	pr_err("msgr_init failed to create workqueue\n");
304	_ceph_msgr_exit();
305
306	return -ENOMEM;
307}
308EXPORT_SYMBOL(ceph_msgr_init);
309
310void ceph_msgr_exit(void)
311{
312	BUG_ON(ceph_msgr_wq == NULL);
313
314	_ceph_msgr_exit();
315}
316EXPORT_SYMBOL(ceph_msgr_exit);
317
318void ceph_msgr_flush(void)
319{
320	flush_workqueue(ceph_msgr_wq);
321}
322EXPORT_SYMBOL(ceph_msgr_flush);
323
324/* Connection socket state transition functions */
325
326static void con_sock_state_init(struct ceph_connection *con)
327{
328	int old_state;
329
330	old_state = atomic_xchg(&con->sock_state, CON_SOCK_STATE_CLOSED);
331	if (WARN_ON(old_state != CON_SOCK_STATE_NEW))
332		printk("%s: unexpected old state %d\n", __func__, old_state);
333	dout("%s con %p sock %d -> %d\n", __func__, con, old_state,
334	     CON_SOCK_STATE_CLOSED);
335}
336
337static void con_sock_state_connecting(struct ceph_connection *con)
338{
339	int old_state;
340
341	old_state = atomic_xchg(&con->sock_state, CON_SOCK_STATE_CONNECTING);
342	if (WARN_ON(old_state != CON_SOCK_STATE_CLOSED))
343		printk("%s: unexpected old state %d\n", __func__, old_state);
344	dout("%s con %p sock %d -> %d\n", __func__, con, old_state,
345	     CON_SOCK_STATE_CONNECTING);
346}
347
348static void con_sock_state_connected(struct ceph_connection *con)
349{
350	int old_state;
351
352	old_state = atomic_xchg(&con->sock_state, CON_SOCK_STATE_CONNECTED);
353	if (WARN_ON(old_state != CON_SOCK_STATE_CONNECTING))
354		printk("%s: unexpected old state %d\n", __func__, old_state);
355	dout("%s con %p sock %d -> %d\n", __func__, con, old_state,
356	     CON_SOCK_STATE_CONNECTED);
357}
358
359static void con_sock_state_closing(struct ceph_connection *con)
360{
361	int old_state;
362
363	old_state = atomic_xchg(&con->sock_state, CON_SOCK_STATE_CLOSING);
364	if (WARN_ON(old_state != CON_SOCK_STATE_CONNECTING &&
365			old_state != CON_SOCK_STATE_CONNECTED &&
366			old_state != CON_SOCK_STATE_CLOSING))
367		printk("%s: unexpected old state %d\n", __func__, old_state);
368	dout("%s con %p sock %d -> %d\n", __func__, con, old_state,
369	     CON_SOCK_STATE_CLOSING);
370}
371
372static void con_sock_state_closed(struct ceph_connection *con)
373{
374	int old_state;
375
376	old_state = atomic_xchg(&con->sock_state, CON_SOCK_STATE_CLOSED);
377	if (WARN_ON(old_state != CON_SOCK_STATE_CONNECTED &&
378		    old_state != CON_SOCK_STATE_CLOSING &&
379		    old_state != CON_SOCK_STATE_CONNECTING &&
380		    old_state != CON_SOCK_STATE_CLOSED))
381		printk("%s: unexpected old state %d\n", __func__, old_state);
382	dout("%s con %p sock %d -> %d\n", __func__, con, old_state,
383	     CON_SOCK_STATE_CLOSED);
384}
385
386/*
387 * socket callback functions
388 */
389
390/* data available on socket, or listen socket received a connect */
391static void ceph_sock_data_ready(struct sock *sk)
392{
393	struct ceph_connection *con = sk->sk_user_data;
394	if (atomic_read(&con->msgr->stopping)) {
395		return;
396	}
397
398	if (sk->sk_state != TCP_CLOSE_WAIT) {
399		dout("%s on %p state = %lu, queueing work\n", __func__,
400		     con, con->state);
401		queue_con(con);
402	}
403}
404
405/* socket has buffer space for writing */
406static void ceph_sock_write_space(struct sock *sk)
407{
408	struct ceph_connection *con = sk->sk_user_data;
409
410	/* only queue to workqueue if there is data we want to write,
411	 * and there is sufficient space in the socket buffer to accept
412	 * more data.  clear SOCK_NOSPACE so that ceph_sock_write_space()
413	 * doesn't get called again until try_write() fills the socket
414	 * buffer. See net/ipv4/tcp_input.c:tcp_check_space()
415	 * and net/core/stream.c:sk_stream_write_space().
416	 */
417	if (con_flag_test(con, CON_FLAG_WRITE_PENDING)) {
418		if (sk_stream_is_writeable(sk)) {
419			dout("%s %p queueing write work\n", __func__, con);
420			clear_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
421			queue_con(con);
422		}
423	} else {
424		dout("%s %p nothing to write\n", __func__, con);
425	}
426}
427
428/* socket's state has changed */
429static void ceph_sock_state_change(struct sock *sk)
430{
431	struct ceph_connection *con = sk->sk_user_data;
432
433	dout("%s %p state = %lu sk_state = %u\n", __func__,
434	     con, con->state, sk->sk_state);
435
436	switch (sk->sk_state) {
437	case TCP_CLOSE:
438		dout("%s TCP_CLOSE\n", __func__);
439	case TCP_CLOSE_WAIT:
440		dout("%s TCP_CLOSE_WAIT\n", __func__);
441		con_sock_state_closing(con);
442		con_flag_set(con, CON_FLAG_SOCK_CLOSED);
443		queue_con(con);
444		break;
445	case TCP_ESTABLISHED:
446		dout("%s TCP_ESTABLISHED\n", __func__);
447		con_sock_state_connected(con);
448		queue_con(con);
449		break;
450	default:	/* Everything else is uninteresting */
451		break;
452	}
453}
454
455/*
456 * set up socket callbacks
457 */
458static void set_sock_callbacks(struct socket *sock,
459			       struct ceph_connection *con)
460{
461	struct sock *sk = sock->sk;
462	sk->sk_user_data = con;
463	sk->sk_data_ready = ceph_sock_data_ready;
464	sk->sk_write_space = ceph_sock_write_space;
465	sk->sk_state_change = ceph_sock_state_change;
466}
467
468
469/*
470 * socket helpers
471 */
472
473/*
474 * initiate connection to a remote socket.
475 */
476static int ceph_tcp_connect(struct ceph_connection *con)
477{
478	struct sockaddr_storage *paddr = &con->peer_addr.in_addr;
479	struct socket *sock;
480	int ret;
481
482	BUG_ON(con->sock);
483	ret = sock_create_kern(con->peer_addr.in_addr.ss_family, SOCK_STREAM,
484			       IPPROTO_TCP, &sock);
485	if (ret)
486		return ret;
487	sock->sk->sk_allocation = GFP_NOFS;
488
489#ifdef CONFIG_LOCKDEP
490	lockdep_set_class(&sock->sk->sk_lock, &socket_class);
491#endif
492
493	set_sock_callbacks(sock, con);
494
495	dout("connect %s\n", ceph_pr_addr(&con->peer_addr.in_addr));
496
497	con_sock_state_connecting(con);
498	ret = sock->ops->connect(sock, (struct sockaddr *)paddr, sizeof(*paddr),
499				 O_NONBLOCK);
500	if (ret == -EINPROGRESS) {
501		dout("connect %s EINPROGRESS sk_state = %u\n",
502		     ceph_pr_addr(&con->peer_addr.in_addr),
503		     sock->sk->sk_state);
504	} else if (ret < 0) {
505		pr_err("connect %s error %d\n",
506		       ceph_pr_addr(&con->peer_addr.in_addr), ret);
507		sock_release(sock);
508		return ret;
509	}
510
511	if (con->msgr->tcp_nodelay) {
512		int optval = 1;
513
514		ret = kernel_setsockopt(sock, SOL_TCP, TCP_NODELAY,
515					(char *)&optval, sizeof(optval));
516		if (ret)
517			pr_err("kernel_setsockopt(TCP_NODELAY) failed: %d",
518			       ret);
519	}
520
521	con->sock = sock;
522	return 0;
523}
524
525static int ceph_tcp_recvmsg(struct socket *sock, void *buf, size_t len)
526{
527	struct kvec iov = {buf, len};
528	struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
529	int r;
530
531	r = kernel_recvmsg(sock, &msg, &iov, 1, len, msg.msg_flags);
532	if (r == -EAGAIN)
533		r = 0;
534	return r;
535}
536
537static int ceph_tcp_recvpage(struct socket *sock, struct page *page,
538		     int page_offset, size_t length)
539{
540	void *kaddr;
541	int ret;
542
543	BUG_ON(page_offset + length > PAGE_SIZE);
544
545	kaddr = kmap(page);
546	BUG_ON(!kaddr);
547	ret = ceph_tcp_recvmsg(sock, kaddr + page_offset, length);
548	kunmap(page);
549
550	return ret;
551}
552
553/*
554 * write something.  @more is true if caller will be sending more data
555 * shortly.
556 */
557static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov,
558		     size_t kvlen, size_t len, int more)
559{
560	struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
561	int r;
562
563	if (more)
564		msg.msg_flags |= MSG_MORE;
565	else
566		msg.msg_flags |= MSG_EOR;  /* superfluous, but what the hell */
567
568	r = kernel_sendmsg(sock, &msg, iov, kvlen, len);
569	if (r == -EAGAIN)
570		r = 0;
571	return r;
572}
573
574static int __ceph_tcp_sendpage(struct socket *sock, struct page *page,
575		     int offset, size_t size, bool more)
576{
577	int flags = MSG_DONTWAIT | MSG_NOSIGNAL | (more ? MSG_MORE : MSG_EOR);
578	int ret;
579
580	ret = kernel_sendpage(sock, page, offset, size, flags);
581	if (ret == -EAGAIN)
582		ret = 0;
583
584	return ret;
585}
586
587static int ceph_tcp_sendpage(struct socket *sock, struct page *page,
588		     int offset, size_t size, bool more)
589{
590	int ret;
591	struct kvec iov;
592
593	/* sendpage cannot properly handle pages with page_count == 0,
594	 * we need to fallback to sendmsg if that's the case */
595	if (page_count(page) >= 1)
596		return __ceph_tcp_sendpage(sock, page, offset, size, more);
597
598	iov.iov_base = kmap(page) + offset;
599	iov.iov_len = size;
600	ret = ceph_tcp_sendmsg(sock, &iov, 1, size, more);
601	kunmap(page);
602
603	return ret;
604}
605
606/*
607 * Shutdown/close the socket for the given connection.
608 */
609static int con_close_socket(struct ceph_connection *con)
610{
611	int rc = 0;
612
613	dout("con_close_socket on %p sock %p\n", con, con->sock);
614	if (con->sock) {
615		rc = con->sock->ops->shutdown(con->sock, SHUT_RDWR);
616		sock_release(con->sock);
617		con->sock = NULL;
618	}
619
620	/*
621	 * Forcibly clear the SOCK_CLOSED flag.  It gets set
622	 * independent of the connection mutex, and we could have
623	 * received a socket close event before we had the chance to
624	 * shut the socket down.
625	 */
626	con_flag_clear(con, CON_FLAG_SOCK_CLOSED);
627
628	con_sock_state_closed(con);
629	return rc;
630}
631
632/*
633 * Reset a connection.  Discard all incoming and outgoing messages
634 * and clear *_seq state.
635 */
636static void ceph_msg_remove(struct ceph_msg *msg)
637{
638	list_del_init(&msg->list_head);
639	BUG_ON(msg->con == NULL);
640	msg->con->ops->put(msg->con);
641	msg->con = NULL;
642
643	ceph_msg_put(msg);
644}
645static void ceph_msg_remove_list(struct list_head *head)
646{
647	while (!list_empty(head)) {
648		struct ceph_msg *msg = list_first_entry(head, struct ceph_msg,
649							list_head);
650		ceph_msg_remove(msg);
651	}
652}
653
654static void reset_connection(struct ceph_connection *con)
655{
656	/* reset connection, out_queue, msg_ and connect_seq */
657	/* discard existing out_queue and msg_seq */
658	dout("reset_connection %p\n", con);
659	ceph_msg_remove_list(&con->out_queue);
660	ceph_msg_remove_list(&con->out_sent);
661
662	if (con->in_msg) {
663		BUG_ON(con->in_msg->con != con);
664		con->in_msg->con = NULL;
665		ceph_msg_put(con->in_msg);
666		con->in_msg = NULL;
667		con->ops->put(con);
668	}
669
670	con->connect_seq = 0;
671	con->out_seq = 0;
672	if (con->out_msg) {
673		ceph_msg_put(con->out_msg);
674		con->out_msg = NULL;
675	}
676	con->in_seq = 0;
677	con->in_seq_acked = 0;
678
679	con->out_skip = 0;
680}
681
682/*
683 * mark a peer down.  drop any open connections.
684 */
685void ceph_con_close(struct ceph_connection *con)
686{
687	mutex_lock(&con->mutex);
688	dout("con_close %p peer %s\n", con,
689	     ceph_pr_addr(&con->peer_addr.in_addr));
690	con->state = CON_STATE_CLOSED;
691
692	con_flag_clear(con, CON_FLAG_LOSSYTX);	/* so we retry next connect */
693	con_flag_clear(con, CON_FLAG_KEEPALIVE_PENDING);
694	con_flag_clear(con, CON_FLAG_WRITE_PENDING);
695	con_flag_clear(con, CON_FLAG_BACKOFF);
696
697	reset_connection(con);
698	con->peer_global_seq = 0;
699	cancel_con(con);
700	con_close_socket(con);
701	mutex_unlock(&con->mutex);
702}
703EXPORT_SYMBOL(ceph_con_close);
704
705/*
706 * Reopen a closed connection, with a new peer address.
707 */
708void ceph_con_open(struct ceph_connection *con,
709		   __u8 entity_type, __u64 entity_num,
710		   struct ceph_entity_addr *addr)
711{
712	mutex_lock(&con->mutex);
713	dout("con_open %p %s\n", con, ceph_pr_addr(&addr->in_addr));
714
715	WARN_ON(con->state != CON_STATE_CLOSED);
716	con->state = CON_STATE_PREOPEN;
717
718	con->peer_name.type = (__u8) entity_type;
719	con->peer_name.num = cpu_to_le64(entity_num);
720
721	memcpy(&con->peer_addr, addr, sizeof(*addr));
722	con->delay = 0;      /* reset backoff memory */
723	mutex_unlock(&con->mutex);
724	queue_con(con);
725}
726EXPORT_SYMBOL(ceph_con_open);
727
728/*
729 * return true if this connection ever successfully opened
730 */
731bool ceph_con_opened(struct ceph_connection *con)
732{
733	return con->connect_seq > 0;
734}
735
736/*
737 * initialize a new connection.
738 */
739void ceph_con_init(struct ceph_connection *con, void *private,
740	const struct ceph_connection_operations *ops,
741	struct ceph_messenger *msgr)
742{
743	dout("con_init %p\n", con);
744	memset(con, 0, sizeof(*con));
745	con->private = private;
746	con->ops = ops;
747	con->msgr = msgr;
748
749	con_sock_state_init(con);
750
751	mutex_init(&con->mutex);
752	INIT_LIST_HEAD(&con->out_queue);
753	INIT_LIST_HEAD(&con->out_sent);
754	INIT_DELAYED_WORK(&con->work, con_work);
755
756	con->state = CON_STATE_CLOSED;
757}
758EXPORT_SYMBOL(ceph_con_init);
759
760
761/*
762 * We maintain a global counter to order connection attempts.  Get
763 * a unique seq greater than @gt.
764 */
765static u32 get_global_seq(struct ceph_messenger *msgr, u32 gt)
766{
767	u32 ret;
768
769	spin_lock(&msgr->global_seq_lock);
770	if (msgr->global_seq < gt)
771		msgr->global_seq = gt;
772	ret = ++msgr->global_seq;
773	spin_unlock(&msgr->global_seq_lock);
774	return ret;
775}
776
777static void con_out_kvec_reset(struct ceph_connection *con)
778{
779	BUG_ON(con->out_skip);
780
781	con->out_kvec_left = 0;
782	con->out_kvec_bytes = 0;
783	con->out_kvec_cur = &con->out_kvec[0];
784}
785
786static void con_out_kvec_add(struct ceph_connection *con,
787				size_t size, void *data)
788{
789	int index = con->out_kvec_left;
790
791	BUG_ON(con->out_skip);
792	BUG_ON(index >= ARRAY_SIZE(con->out_kvec));
793
794	con->out_kvec[index].iov_len = size;
795	con->out_kvec[index].iov_base = data;
796	con->out_kvec_left++;
797	con->out_kvec_bytes += size;
798}
799
800/*
801 * Chop off a kvec from the end.  Return residual number of bytes for
802 * that kvec, i.e. how many bytes would have been written if the kvec
803 * hadn't been nuked.
804 */
805static int con_out_kvec_skip(struct ceph_connection *con)
806{
807	int off = con->out_kvec_cur - con->out_kvec;
808	int skip = 0;
809
810	if (con->out_kvec_bytes > 0) {
811		skip = con->out_kvec[off + con->out_kvec_left - 1].iov_len;
812		BUG_ON(con->out_kvec_bytes < skip);
813		BUG_ON(!con->out_kvec_left);
814		con->out_kvec_bytes -= skip;
815		con->out_kvec_left--;
816	}
817
818	return skip;
819}
820
821#ifdef CONFIG_BLOCK
822
823/*
824 * For a bio data item, a piece is whatever remains of the next
825 * entry in the current bio iovec, or the first entry in the next
826 * bio in the list.
827 */
828static void ceph_msg_data_bio_cursor_init(struct ceph_msg_data_cursor *cursor,
829					size_t length)
830{
831	struct ceph_msg_data *data = cursor->data;
832	struct bio *bio;
833
834	BUG_ON(data->type != CEPH_MSG_DATA_BIO);
835
836	bio = data->bio;
837	BUG_ON(!bio);
838
839	cursor->resid = min(length, data->bio_length);
840	cursor->bio = bio;
841	cursor->bvec_iter = bio->bi_iter;
842	cursor->last_piece =
843		cursor->resid <= bio_iter_len(bio, cursor->bvec_iter);
844}
845
846static struct page *ceph_msg_data_bio_next(struct ceph_msg_data_cursor *cursor,
847						size_t *page_offset,
848						size_t *length)
849{
850	struct ceph_msg_data *data = cursor->data;
851	struct bio *bio;
852	struct bio_vec bio_vec;
853
854	BUG_ON(data->type != CEPH_MSG_DATA_BIO);
855
856	bio = cursor->bio;
857	BUG_ON(!bio);
858
859	bio_vec = bio_iter_iovec(bio, cursor->bvec_iter);
860
861	*page_offset = (size_t) bio_vec.bv_offset;
862	BUG_ON(*page_offset >= PAGE_SIZE);
863	if (cursor->last_piece) /* pagelist offset is always 0 */
864		*length = cursor->resid;
865	else
866		*length = (size_t) bio_vec.bv_len;
867	BUG_ON(*length > cursor->resid);
868	BUG_ON(*page_offset + *length > PAGE_SIZE);
869
870	return bio_vec.bv_page;
871}
872
873static bool ceph_msg_data_bio_advance(struct ceph_msg_data_cursor *cursor,
874					size_t bytes)
875{
876	struct bio *bio;
877	struct bio_vec bio_vec;
878
879	BUG_ON(cursor->data->type != CEPH_MSG_DATA_BIO);
880
881	bio = cursor->bio;
882	BUG_ON(!bio);
883
884	bio_vec = bio_iter_iovec(bio, cursor->bvec_iter);
885
886	/* Advance the cursor offset */
887
888	BUG_ON(cursor->resid < bytes);
889	cursor->resid -= bytes;
890
891	bio_advance_iter(bio, &cursor->bvec_iter, bytes);
892
893	if (bytes < bio_vec.bv_len)
894		return false;	/* more bytes to process in this segment */
895
896	/* Move on to the next segment, and possibly the next bio */
897
898	if (!cursor->bvec_iter.bi_size) {
899		bio = bio->bi_next;
900		cursor->bio = bio;
901		if (bio)
902			cursor->bvec_iter = bio->bi_iter;
903		else
904			memset(&cursor->bvec_iter, 0,
905			       sizeof(cursor->bvec_iter));
906	}
907
908	if (!cursor->last_piece) {
909		BUG_ON(!cursor->resid);
910		BUG_ON(!bio);
911		/* A short read is OK, so use <= rather than == */
912		if (cursor->resid <= bio_iter_len(bio, cursor->bvec_iter))
913			cursor->last_piece = true;
914	}
915
916	return true;
917}
918#endif /* CONFIG_BLOCK */
919
920/*
921 * For a page array, a piece comes from the first page in the array
922 * that has not already been fully consumed.
923 */
924static void ceph_msg_data_pages_cursor_init(struct ceph_msg_data_cursor *cursor,
925					size_t length)
926{
927	struct ceph_msg_data *data = cursor->data;
928	int page_count;
929
930	BUG_ON(data->type != CEPH_MSG_DATA_PAGES);
931
932	BUG_ON(!data->pages);
933	BUG_ON(!data->length);
934
935	cursor->resid = min(length, data->length);
936	page_count = calc_pages_for(data->alignment, (u64)data->length);
937	cursor->page_offset = data->alignment & ~PAGE_MASK;
938	cursor->page_index = 0;
939	BUG_ON(page_count > (int)USHRT_MAX);
940	cursor->page_count = (unsigned short)page_count;
941	BUG_ON(length > SIZE_MAX - cursor->page_offset);
942	cursor->last_piece = cursor->page_offset + cursor->resid <= PAGE_SIZE;
943}
944
945static struct page *
946ceph_msg_data_pages_next(struct ceph_msg_data_cursor *cursor,
947					size_t *page_offset, size_t *length)
948{
949	struct ceph_msg_data *data = cursor->data;
950
951	BUG_ON(data->type != CEPH_MSG_DATA_PAGES);
952
953	BUG_ON(cursor->page_index >= cursor->page_count);
954	BUG_ON(cursor->page_offset >= PAGE_SIZE);
955
956	*page_offset = cursor->page_offset;
957	if (cursor->last_piece)
958		*length = cursor->resid;
959	else
960		*length = PAGE_SIZE - *page_offset;
961
962	return data->pages[cursor->page_index];
963}
964
965static bool ceph_msg_data_pages_advance(struct ceph_msg_data_cursor *cursor,
966						size_t bytes)
967{
968	BUG_ON(cursor->data->type != CEPH_MSG_DATA_PAGES);
969
970	BUG_ON(cursor->page_offset + bytes > PAGE_SIZE);
971
972	/* Advance the cursor page offset */
973
974	cursor->resid -= bytes;
975	cursor->page_offset = (cursor->page_offset + bytes) & ~PAGE_MASK;
976	if (!bytes || cursor->page_offset)
977		return false;	/* more bytes to process in the current page */
978
979	if (!cursor->resid)
980		return false;   /* no more data */
981
982	/* Move on to the next page; offset is already at 0 */
983
984	BUG_ON(cursor->page_index >= cursor->page_count);
985	cursor->page_index++;
986	cursor->last_piece = cursor->resid <= PAGE_SIZE;
987
988	return true;
989}
990
991/*
992 * For a pagelist, a piece is whatever remains to be consumed in the
993 * first page in the list, or the front of the next page.
994 */
995static void
996ceph_msg_data_pagelist_cursor_init(struct ceph_msg_data_cursor *cursor,
997					size_t length)
998{
999	struct ceph_msg_data *data = cursor->data;
1000	struct ceph_pagelist *pagelist;
1001	struct page *page;
1002
1003	BUG_ON(data->type != CEPH_MSG_DATA_PAGELIST);
1004
1005	pagelist = data->pagelist;
1006	BUG_ON(!pagelist);
1007
1008	if (!length)
1009		return;		/* pagelist can be assigned but empty */
1010
1011	BUG_ON(list_empty(&pagelist->head));
1012	page = list_first_entry(&pagelist->head, struct page, lru);
1013
1014	cursor->resid = min(length, pagelist->length);
1015	cursor->page = page;
1016	cursor->offset = 0;
1017	cursor->last_piece = cursor->resid <= PAGE_SIZE;
1018}
1019
1020static struct page *
1021ceph_msg_data_pagelist_next(struct ceph_msg_data_cursor *cursor,
1022				size_t *page_offset, size_t *length)
1023{
1024	struct ceph_msg_data *data = cursor->data;
1025	struct ceph_pagelist *pagelist;
1026
1027	BUG_ON(data->type != CEPH_MSG_DATA_PAGELIST);
1028
1029	pagelist = data->pagelist;
1030	BUG_ON(!pagelist);
1031
1032	BUG_ON(!cursor->page);
1033	BUG_ON(cursor->offset + cursor->resid != pagelist->length);
1034
1035	/* offset of first page in pagelist is always 0 */
1036	*page_offset = cursor->offset & ~PAGE_MASK;
1037	if (cursor->last_piece)
1038		*length = cursor->resid;
1039	else
1040		*length = PAGE_SIZE - *page_offset;
1041
1042	return cursor->page;
1043}
1044
1045static bool ceph_msg_data_pagelist_advance(struct ceph_msg_data_cursor *cursor,
1046						size_t bytes)
1047{
1048	struct ceph_msg_data *data = cursor->data;
1049	struct ceph_pagelist *pagelist;
1050
1051	BUG_ON(data->type != CEPH_MSG_DATA_PAGELIST);
1052
1053	pagelist = data->pagelist;
1054	BUG_ON(!pagelist);
1055
1056	BUG_ON(cursor->offset + cursor->resid != pagelist->length);
1057	BUG_ON((cursor->offset & ~PAGE_MASK) + bytes > PAGE_SIZE);
1058
1059	/* Advance the cursor offset */
1060
1061	cursor->resid -= bytes;
1062	cursor->offset += bytes;
1063	/* offset of first page in pagelist is always 0 */
1064	if (!bytes || cursor->offset & ~PAGE_MASK)
1065		return false;	/* more bytes to process in the current page */
1066
1067	if (!cursor->resid)
1068		return false;   /* no more data */
1069
1070	/* Move on to the next page */
1071
1072	BUG_ON(list_is_last(&cursor->page->lru, &pagelist->head));
1073	cursor->page = list_entry_next(cursor->page, lru);
1074	cursor->last_piece = cursor->resid <= PAGE_SIZE;
1075
1076	return true;
1077}
1078
1079/*
1080 * Message data is handled (sent or received) in pieces, where each
1081 * piece resides on a single page.  The network layer might not
1082 * consume an entire piece at once.  A data item's cursor keeps
1083 * track of which piece is next to process and how much remains to
1084 * be processed in that piece.  It also tracks whether the current
1085 * piece is the last one in the data item.
1086 */
1087static void __ceph_msg_data_cursor_init(struct ceph_msg_data_cursor *cursor)
1088{
1089	size_t length = cursor->total_resid;
1090
1091	switch (cursor->data->type) {
1092	case CEPH_MSG_DATA_PAGELIST:
1093		ceph_msg_data_pagelist_cursor_init(cursor, length);
1094		break;
1095	case CEPH_MSG_DATA_PAGES:
1096		ceph_msg_data_pages_cursor_init(cursor, length);
1097		break;
1098#ifdef CONFIG_BLOCK
1099	case CEPH_MSG_DATA_BIO:
1100		ceph_msg_data_bio_cursor_init(cursor, length);
1101		break;
1102#endif /* CONFIG_BLOCK */
1103	case CEPH_MSG_DATA_NONE:
1104	default:
1105		/* BUG(); */
1106		break;
1107	}
1108	cursor->need_crc = true;
1109}
1110
1111static void ceph_msg_data_cursor_init(struct ceph_msg *msg, size_t length)
1112{
1113	struct ceph_msg_data_cursor *cursor = &msg->cursor;
1114	struct ceph_msg_data *data;
1115
1116	BUG_ON(!length);
1117	BUG_ON(length > msg->data_length);
1118	BUG_ON(list_empty(&msg->data));
1119
1120	cursor->data_head = &msg->data;
1121	cursor->total_resid = length;
1122	data = list_first_entry(&msg->data, struct ceph_msg_data, links);
1123	cursor->data = data;
1124
1125	__ceph_msg_data_cursor_init(cursor);
1126}
1127
1128/*
1129 * Return the page containing the next piece to process for a given
1130 * data item, and supply the page offset and length of that piece.
1131 * Indicate whether this is the last piece in this data item.
1132 */
1133static struct page *ceph_msg_data_next(struct ceph_msg_data_cursor *cursor,
1134					size_t *page_offset, size_t *length,
1135					bool *last_piece)
1136{
1137	struct page *page;
1138
1139	switch (cursor->data->type) {
1140	case CEPH_MSG_DATA_PAGELIST:
1141		page = ceph_msg_data_pagelist_next(cursor, page_offset, length);
1142		break;
1143	case CEPH_MSG_DATA_PAGES:
1144		page = ceph_msg_data_pages_next(cursor, page_offset, length);
1145		break;
1146#ifdef CONFIG_BLOCK
1147	case CEPH_MSG_DATA_BIO:
1148		page = ceph_msg_data_bio_next(cursor, page_offset, length);
1149		break;
1150#endif /* CONFIG_BLOCK */
1151	case CEPH_MSG_DATA_NONE:
1152	default:
1153		page = NULL;
1154		break;
1155	}
1156	BUG_ON(!page);
1157	BUG_ON(*page_offset + *length > PAGE_SIZE);
1158	BUG_ON(!*length);
1159	if (last_piece)
1160		*last_piece = cursor->last_piece;
1161
1162	return page;
1163}
1164
1165/*
1166 * Returns true if the result moves the cursor on to the next piece
1167 * of the data item.
1168 */
1169static bool ceph_msg_data_advance(struct ceph_msg_data_cursor *cursor,
1170				size_t bytes)
1171{
1172	bool new_piece;
1173
1174	BUG_ON(bytes > cursor->resid);
1175	switch (cursor->data->type) {
1176	case CEPH_MSG_DATA_PAGELIST:
1177		new_piece = ceph_msg_data_pagelist_advance(cursor, bytes);
1178		break;
1179	case CEPH_MSG_DATA_PAGES:
1180		new_piece = ceph_msg_data_pages_advance(cursor, bytes);
1181		break;
1182#ifdef CONFIG_BLOCK
1183	case CEPH_MSG_DATA_BIO:
1184		new_piece = ceph_msg_data_bio_advance(cursor, bytes);
1185		break;
1186#endif /* CONFIG_BLOCK */
1187	case CEPH_MSG_DATA_NONE:
1188	default:
1189		BUG();
1190		break;
1191	}
1192	cursor->total_resid -= bytes;
1193
1194	if (!cursor->resid && cursor->total_resid) {
1195		WARN_ON(!cursor->last_piece);
1196		BUG_ON(list_is_last(&cursor->data->links, cursor->data_head));
1197		cursor->data = list_entry_next(cursor->data, links);
1198		__ceph_msg_data_cursor_init(cursor);
1199		new_piece = true;
1200	}
1201	cursor->need_crc = new_piece;
1202
1203	return new_piece;
1204}
1205
1206static size_t sizeof_footer(struct ceph_connection *con)
1207{
1208	return (con->peer_features & CEPH_FEATURE_MSG_AUTH) ?
1209	    sizeof(struct ceph_msg_footer) :
1210	    sizeof(struct ceph_msg_footer_old);
1211}
1212
1213static void prepare_message_data(struct ceph_msg *msg, u32 data_len)
1214{
1215	BUG_ON(!msg);
1216	BUG_ON(!data_len);
1217
1218	/* Initialize data cursor */
1219
1220	ceph_msg_data_cursor_init(msg, (size_t)data_len);
1221}
1222
1223/*
1224 * Prepare footer for currently outgoing message, and finish things
1225 * off.  Assumes out_kvec* are already valid.. we just add on to the end.
1226 */
1227static void prepare_write_message_footer(struct ceph_connection *con)
1228{
1229	struct ceph_msg *m = con->out_msg;
1230	int v = con->out_kvec_left;
1231
1232	m->footer.flags |= CEPH_MSG_FOOTER_COMPLETE;
1233
1234	dout("prepare_write_message_footer %p\n", con);
1235	con->out_kvec[v].iov_base = &m->footer;
1236	if (con->peer_features & CEPH_FEATURE_MSG_AUTH) {
1237		if (con->ops->sign_message)
1238			con->ops->sign_message(con, m);
1239		else
1240			m->footer.sig = 0;
1241		con->out_kvec[v].iov_len = sizeof(m->footer);
1242		con->out_kvec_bytes += sizeof(m->footer);
1243	} else {
1244		m->old_footer.flags = m->footer.flags;
1245		con->out_kvec[v].iov_len = sizeof(m->old_footer);
1246		con->out_kvec_bytes += sizeof(m->old_footer);
1247	}
1248	con->out_kvec_left++;
1249	con->out_more = m->more_to_follow;
1250	con->out_msg_done = true;
1251}
1252
1253/*
1254 * Prepare headers for the next outgoing message.
1255 */
1256static void prepare_write_message(struct ceph_connection *con)
1257{
1258	struct ceph_msg *m;
1259	u32 crc;
1260
1261	con_out_kvec_reset(con);
1262	con->out_msg_done = false;
1263
1264	/* Sneak an ack in there first?  If we can get it into the same
1265	 * TCP packet that's a good thing. */
1266	if (con->in_seq > con->in_seq_acked) {
1267		con->in_seq_acked = con->in_seq;
1268		con_out_kvec_add(con, sizeof (tag_ack), &tag_ack);
1269		con->out_temp_ack = cpu_to_le64(con->in_seq_acked);
1270		con_out_kvec_add(con, sizeof (con->out_temp_ack),
1271			&con->out_temp_ack);
1272	}
1273
1274	BUG_ON(list_empty(&con->out_queue));
1275	m = list_first_entry(&con->out_queue, struct ceph_msg, list_head);
1276	con->out_msg = m;
1277	BUG_ON(m->con != con);
1278
1279	/* put message on sent list */
1280	ceph_msg_get(m);
1281	list_move_tail(&m->list_head, &con->out_sent);
1282
1283	/*
1284	 * only assign outgoing seq # if we haven't sent this message
1285	 * yet.  if it is requeued, resend with it's original seq.
1286	 */
1287	if (m->needs_out_seq) {
1288		m->hdr.seq = cpu_to_le64(++con->out_seq);
1289		m->needs_out_seq = false;
1290	}
1291	WARN_ON(m->data_length != le32_to_cpu(m->hdr.data_len));
1292
1293	dout("prepare_write_message %p seq %lld type %d len %d+%d+%zd\n",
1294	     m, con->out_seq, le16_to_cpu(m->hdr.type),
1295	     le32_to_cpu(m->hdr.front_len), le32_to_cpu(m->hdr.middle_len),
1296	     m->data_length);
1297	BUG_ON(le32_to_cpu(m->hdr.front_len) != m->front.iov_len);
1298
1299	/* tag + hdr + front + middle */
1300	con_out_kvec_add(con, sizeof (tag_msg), &tag_msg);
1301	con_out_kvec_add(con, sizeof(con->out_hdr), &con->out_hdr);
1302	con_out_kvec_add(con, m->front.iov_len, m->front.iov_base);
1303
1304	if (m->middle)
1305		con_out_kvec_add(con, m->middle->vec.iov_len,
1306			m->middle->vec.iov_base);
1307
1308	/* fill in hdr crc and finalize hdr */
1309	crc = crc32c(0, &m->hdr, offsetof(struct ceph_msg_header, crc));
1310	con->out_msg->hdr.crc = cpu_to_le32(crc);
1311	memcpy(&con->out_hdr, &con->out_msg->hdr, sizeof(con->out_hdr));
1312
1313	/* fill in front and middle crc, footer */
1314	crc = crc32c(0, m->front.iov_base, m->front.iov_len);
1315	con->out_msg->footer.front_crc = cpu_to_le32(crc);
1316	if (m->middle) {
1317		crc = crc32c(0, m->middle->vec.iov_base,
1318				m->middle->vec.iov_len);
1319		con->out_msg->footer.middle_crc = cpu_to_le32(crc);
1320	} else
1321		con->out_msg->footer.middle_crc = 0;
1322	dout("%s front_crc %u middle_crc %u\n", __func__,
1323	     le32_to_cpu(con->out_msg->footer.front_crc),
1324	     le32_to_cpu(con->out_msg->footer.middle_crc));
1325	con->out_msg->footer.flags = 0;
1326
1327	/* is there a data payload? */
1328	con->out_msg->footer.data_crc = 0;
1329	if (m->data_length) {
1330		prepare_message_data(con->out_msg, m->data_length);
1331		con->out_more = 1;  /* data + footer will follow */
1332	} else {
1333		/* no, queue up footer too and be done */
1334		prepare_write_message_footer(con);
1335	}
1336
1337	con_flag_set(con, CON_FLAG_WRITE_PENDING);
1338}
1339
1340/*
1341 * Prepare an ack.
1342 */
1343static void prepare_write_ack(struct ceph_connection *con)
1344{
1345	dout("prepare_write_ack %p %llu -> %llu\n", con,
1346	     con->in_seq_acked, con->in_seq);
1347	con->in_seq_acked = con->in_seq;
1348
1349	con_out_kvec_reset(con);
1350
1351	con_out_kvec_add(con, sizeof (tag_ack), &tag_ack);
1352
1353	con->out_temp_ack = cpu_to_le64(con->in_seq_acked);
1354	con_out_kvec_add(con, sizeof (con->out_temp_ack),
1355				&con->out_temp_ack);
1356
1357	con->out_more = 1;  /* more will follow.. eventually.. */
1358	con_flag_set(con, CON_FLAG_WRITE_PENDING);
1359}
1360
1361/*
1362 * Prepare to share the seq during handshake
1363 */
1364static void prepare_write_seq(struct ceph_connection *con)
1365{
1366	dout("prepare_write_seq %p %llu -> %llu\n", con,
1367	     con->in_seq_acked, con->in_seq);
1368	con->in_seq_acked = con->in_seq;
1369
1370	con_out_kvec_reset(con);
1371
1372	con->out_temp_ack = cpu_to_le64(con->in_seq_acked);
1373	con_out_kvec_add(con, sizeof (con->out_temp_ack),
1374			 &con->out_temp_ack);
1375
1376	con_flag_set(con, CON_FLAG_WRITE_PENDING);
1377}
1378
1379/*
1380 * Prepare to write keepalive byte.
1381 */
1382static void prepare_write_keepalive(struct ceph_connection *con)
1383{
1384	dout("prepare_write_keepalive %p\n", con);
1385	con_out_kvec_reset(con);
1386	con_out_kvec_add(con, sizeof (tag_keepalive), &tag_keepalive);
1387	con_flag_set(con, CON_FLAG_WRITE_PENDING);
1388}
1389
1390/*
1391 * Connection negotiation.
1392 */
1393
1394static struct ceph_auth_handshake *get_connect_authorizer(struct ceph_connection *con,
1395						int *auth_proto)
1396{
1397	struct ceph_auth_handshake *auth;
1398
1399	if (!con->ops->get_authorizer) {
1400		con->out_connect.authorizer_protocol = CEPH_AUTH_UNKNOWN;
1401		con->out_connect.authorizer_len = 0;
1402		return NULL;
1403	}
1404
1405	/* Can't hold the mutex while getting authorizer */
1406	mutex_unlock(&con->mutex);
1407	auth = con->ops->get_authorizer(con, auth_proto, con->auth_retry);
1408	mutex_lock(&con->mutex);
1409
1410	if (IS_ERR(auth))
1411		return auth;
1412	if (con->state != CON_STATE_NEGOTIATING)
1413		return ERR_PTR(-EAGAIN);
1414
1415	con->auth_reply_buf = auth->authorizer_reply_buf;
1416	con->auth_reply_buf_len = auth->authorizer_reply_buf_len;
1417	return auth;
1418}
1419
1420/*
1421 * We connected to a peer and are saying hello.
1422 */
1423static void prepare_write_banner(struct ceph_connection *con)
1424{
1425	con_out_kvec_add(con, strlen(CEPH_BANNER), CEPH_BANNER);
1426	con_out_kvec_add(con, sizeof (con->msgr->my_enc_addr),
1427					&con->msgr->my_enc_addr);
1428
1429	con->out_more = 0;
1430	con_flag_set(con, CON_FLAG_WRITE_PENDING);
1431}
1432
1433static int prepare_write_connect(struct ceph_connection *con)
1434{
1435	unsigned int global_seq = get_global_seq(con->msgr, 0);
1436	int proto;
1437	int auth_proto;
1438	struct ceph_auth_handshake *auth;
1439
1440	switch (con->peer_name.type) {
1441	case CEPH_ENTITY_TYPE_MON:
1442		proto = CEPH_MONC_PROTOCOL;
1443		break;
1444	case CEPH_ENTITY_TYPE_OSD:
1445		proto = CEPH_OSDC_PROTOCOL;
1446		break;
1447	case CEPH_ENTITY_TYPE_MDS:
1448		proto = CEPH_MDSC_PROTOCOL;
1449		break;
1450	default:
1451		BUG();
1452	}
1453
1454	dout("prepare_write_connect %p cseq=%d gseq=%d proto=%d\n", con,
1455	     con->connect_seq, global_seq, proto);
1456
1457	con->out_connect.features = cpu_to_le64(con->msgr->supported_features);
1458	con->out_connect.host_type = cpu_to_le32(CEPH_ENTITY_TYPE_CLIENT);
1459	con->out_connect.connect_seq = cpu_to_le32(con->connect_seq);
1460	con->out_connect.global_seq = cpu_to_le32(global_seq);
1461	con->out_connect.protocol_version = cpu_to_le32(proto);
1462	con->out_connect.flags = 0;
1463
1464	auth_proto = CEPH_AUTH_UNKNOWN;
1465	auth = get_connect_authorizer(con, &auth_proto);
1466	if (IS_ERR(auth))
1467		return PTR_ERR(auth);
1468
1469	con->out_connect.authorizer_protocol = cpu_to_le32(auth_proto);
1470	con->out_connect.authorizer_len = auth ?
1471		cpu_to_le32(auth->authorizer_buf_len) : 0;
1472
1473	con_out_kvec_add(con, sizeof (con->out_connect),
1474					&con->out_connect);
1475	if (auth && auth->authorizer_buf_len)
1476		con_out_kvec_add(con, auth->authorizer_buf_len,
1477					auth->authorizer_buf);
1478
1479	con->out_more = 0;
1480	con_flag_set(con, CON_FLAG_WRITE_PENDING);
1481
1482	return 0;
1483}
1484
1485/*
1486 * write as much of pending kvecs to the socket as we can.
1487 *  1 -> done
1488 *  0 -> socket full, but more to do
1489 * <0 -> error
1490 */
1491static int write_partial_kvec(struct ceph_connection *con)
1492{
1493	int ret;
1494
1495	dout("write_partial_kvec %p %d left\n", con, con->out_kvec_bytes);
1496	while (con->out_kvec_bytes > 0) {
1497		ret = ceph_tcp_sendmsg(con->sock, con->out_kvec_cur,
1498				       con->out_kvec_left, con->out_kvec_bytes,
1499				       con->out_more);
1500		if (ret <= 0)
1501			goto out;
1502		con->out_kvec_bytes -= ret;
1503		if (con->out_kvec_bytes == 0)
1504			break;            /* done */
1505
1506		/* account for full iov entries consumed */
1507		while (ret >= con->out_kvec_cur->iov_len) {
1508			BUG_ON(!con->out_kvec_left);
1509			ret -= con->out_kvec_cur->iov_len;
1510			con->out_kvec_cur++;
1511			con->out_kvec_left--;
1512		}
1513		/* and for a partially-consumed entry */
1514		if (ret) {
1515			con->out_kvec_cur->iov_len -= ret;
1516			con->out_kvec_cur->iov_base += ret;
1517		}
1518	}
1519	con->out_kvec_left = 0;
1520	ret = 1;
1521out:
1522	dout("write_partial_kvec %p %d left in %d kvecs ret = %d\n", con,
1523	     con->out_kvec_bytes, con->out_kvec_left, ret);
1524	return ret;  /* done! */
1525}
1526
1527static u32 ceph_crc32c_page(u32 crc, struct page *page,
1528				unsigned int page_offset,
1529				unsigned int length)
1530{
1531	char *kaddr;
1532
1533	kaddr = kmap(page);
1534	BUG_ON(kaddr == NULL);
1535	crc = crc32c(crc, kaddr + page_offset, length);
1536	kunmap(page);
1537
1538	return crc;
1539}
1540/*
1541 * Write as much message data payload as we can.  If we finish, queue
1542 * up the footer.
1543 *  1 -> done, footer is now queued in out_kvec[].
1544 *  0 -> socket full, but more to do
1545 * <0 -> error
1546 */
1547static int write_partial_message_data(struct ceph_connection *con)
1548{
1549	struct ceph_msg *msg = con->out_msg;
1550	struct ceph_msg_data_cursor *cursor = &msg->cursor;
1551	bool do_datacrc = !con->msgr->nocrc;
1552	u32 crc;
1553
1554	dout("%s %p msg %p\n", __func__, con, msg);
1555
1556	if (list_empty(&msg->data))
1557		return -EINVAL;
1558
1559	/*
1560	 * Iterate through each page that contains data to be
1561	 * written, and send as much as possible for each.
1562	 *
1563	 * If we are calculating the data crc (the default), we will
1564	 * need to map the page.  If we have no pages, they have
1565	 * been revoked, so use the zero page.
1566	 */
1567	crc = do_datacrc ? le32_to_cpu(msg->footer.data_crc) : 0;
1568	while (cursor->resid) {
1569		struct page *page;
1570		size_t page_offset;
1571		size_t length;
1572		bool last_piece;
1573		bool need_crc;
1574		int ret;
1575
1576		page = ceph_msg_data_next(&msg->cursor, &page_offset, &length,
1577							&last_piece);
1578		ret = ceph_tcp_sendpage(con->sock, page, page_offset,
1579				      length, last_piece);
1580		if (ret <= 0) {
1581			if (do_datacrc)
1582				msg->footer.data_crc = cpu_to_le32(crc);
1583
1584			return ret;
1585		}
1586		if (do_datacrc && cursor->need_crc)
1587			crc = ceph_crc32c_page(crc, page, page_offset, length);
1588		need_crc = ceph_msg_data_advance(&msg->cursor, (size_t)ret);
1589	}
1590
1591	dout("%s %p msg %p done\n", __func__, con, msg);
1592
1593	/* prepare and queue up footer, too */
1594	if (do_datacrc)
1595		msg->footer.data_crc = cpu_to_le32(crc);
1596	else
1597		msg->footer.flags |= CEPH_MSG_FOOTER_NOCRC;
1598	con_out_kvec_reset(con);
1599	prepare_write_message_footer(con);
1600
1601	return 1;	/* must return > 0 to indicate success */
1602}
1603
1604/*
1605 * write some zeros
1606 */
1607static int write_partial_skip(struct ceph_connection *con)
1608{
1609	int ret;
1610
1611	dout("%s %p %d left\n", __func__, con, con->out_skip);
1612	while (con->out_skip > 0) {
1613		size_t size = min(con->out_skip, (int) PAGE_CACHE_SIZE);
1614
1615		ret = ceph_tcp_sendpage(con->sock, zero_page, 0, size, true);
1616		if (ret <= 0)
1617			goto out;
1618		con->out_skip -= ret;
1619	}
1620	ret = 1;
1621out:
1622	return ret;
1623}
1624
1625/*
1626 * Prepare to read connection handshake, or an ack.
1627 */
1628static void prepare_read_banner(struct ceph_connection *con)
1629{
1630	dout("prepare_read_banner %p\n", con);
1631	con->in_base_pos = 0;
1632}
1633
1634static void prepare_read_connect(struct ceph_connection *con)
1635{
1636	dout("prepare_read_connect %p\n", con);
1637	con->in_base_pos = 0;
1638}
1639
1640static void prepare_read_ack(struct ceph_connection *con)
1641{
1642	dout("prepare_read_ack %p\n", con);
1643	con->in_base_pos = 0;
1644}
1645
1646static void prepare_read_seq(struct ceph_connection *con)
1647{
1648	dout("prepare_read_seq %p\n", con);
1649	con->in_base_pos = 0;
1650	con->in_tag = CEPH_MSGR_TAG_SEQ;
1651}
1652
1653static void prepare_read_tag(struct ceph_connection *con)
1654{
1655	dout("prepare_read_tag %p\n", con);
1656	con->in_base_pos = 0;
1657	con->in_tag = CEPH_MSGR_TAG_READY;
1658}
1659
1660/*
1661 * Prepare to read a message.
1662 */
1663static int prepare_read_message(struct ceph_connection *con)
1664{
1665	dout("prepare_read_message %p\n", con);
1666	BUG_ON(con->in_msg != NULL);
1667	con->in_base_pos = 0;
1668	con->in_front_crc = con->in_middle_crc = con->in_data_crc = 0;
1669	return 0;
1670}
1671
1672
1673static int read_partial(struct ceph_connection *con,
1674			int end, int size, void *object)
1675{
1676	while (con->in_base_pos < end) {
1677		int left = end - con->in_base_pos;
1678		int have = size - left;
1679		int ret = ceph_tcp_recvmsg(con->sock, object + have, left);
1680		if (ret <= 0)
1681			return ret;
1682		con->in_base_pos += ret;
1683	}
1684	return 1;
1685}
1686
1687
1688/*
1689 * Read all or part of the connect-side handshake on a new connection
1690 */
1691static int read_partial_banner(struct ceph_connection *con)
1692{
1693	int size;
1694	int end;
1695	int ret;
1696
1697	dout("read_partial_banner %p at %d\n", con, con->in_base_pos);
1698
1699	/* peer's banner */
1700	size = strlen(CEPH_BANNER);
1701	end = size;
1702	ret = read_partial(con, end, size, con->in_banner);
1703	if (ret <= 0)
1704		goto out;
1705
1706	size = sizeof (con->actual_peer_addr);
1707	end += size;
1708	ret = read_partial(con, end, size, &con->actual_peer_addr);
1709	if (ret <= 0)
1710		goto out;
1711
1712	size = sizeof (con->peer_addr_for_me);
1713	end += size;
1714	ret = read_partial(con, end, size, &con->peer_addr_for_me);
1715	if (ret <= 0)
1716		goto out;
1717
1718out:
1719	return ret;
1720}
1721
1722static int read_partial_connect(struct ceph_connection *con)
1723{
1724	int size;
1725	int end;
1726	int ret;
1727
1728	dout("read_partial_connect %p at %d\n", con, con->in_base_pos);
1729
1730	size = sizeof (con->in_reply);
1731	end = size;
1732	ret = read_partial(con, end, size, &con->in_reply);
1733	if (ret <= 0)
1734		goto out;
1735
1736	size = le32_to_cpu(con->in_reply.authorizer_len);
1737	end += size;
1738	ret = read_partial(con, end, size, con->auth_reply_buf);
1739	if (ret <= 0)
1740		goto out;
1741
1742	dout("read_partial_connect %p tag %d, con_seq = %u, g_seq = %u\n",
1743	     con, (int)con->in_reply.tag,
1744	     le32_to_cpu(con->in_reply.connect_seq),
1745	     le32_to_cpu(con->in_reply.global_seq));
1746out:
1747	return ret;
1748
1749}
1750
1751/*
1752 * Verify the hello banner looks okay.
1753 */
1754static int verify_hello(struct ceph_connection *con)
1755{
1756	if (memcmp(con->in_banner, CEPH_BANNER, strlen(CEPH_BANNER))) {
1757		pr_err("connect to %s got bad banner\n",
1758		       ceph_pr_addr(&con->peer_addr.in_addr));
1759		con->error_msg = "protocol error, bad banner";
1760		return -1;
1761	}
1762	return 0;
1763}
1764
1765static bool addr_is_blank(struct sockaddr_storage *ss)
1766{
1767	switch (ss->ss_family) {
1768	case AF_INET:
1769		return ((struct sockaddr_in *)ss)->sin_addr.s_addr == 0;
1770	case AF_INET6:
1771		return
1772		     ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[0] == 0 &&
1773		     ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[1] == 0 &&
1774		     ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[2] == 0 &&
1775		     ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[3] == 0;
1776	}
1777	return false;
1778}
1779
1780static int addr_port(struct sockaddr_storage *ss)
1781{
1782	switch (ss->ss_family) {
1783	case AF_INET:
1784		return ntohs(((struct sockaddr_in *)ss)->sin_port);
1785	case AF_INET6:
1786		return ntohs(((struct sockaddr_in6 *)ss)->sin6_port);
1787	}
1788	return 0;
1789}
1790
1791static void addr_set_port(struct sockaddr_storage *ss, int p)
1792{
1793	switch (ss->ss_family) {
1794	case AF_INET:
1795		((struct sockaddr_in *)ss)->sin_port = htons(p);
1796		break;
1797	case AF_INET6:
1798		((struct sockaddr_in6 *)ss)->sin6_port = htons(p);
1799		break;
1800	}
1801}
1802
1803/*
1804 * Unlike other *_pton function semantics, zero indicates success.
1805 */
1806static int ceph_pton(const char *str, size_t len, struct sockaddr_storage *ss,
1807		char delim, const char **ipend)
1808{
1809	struct sockaddr_in *in4 = (struct sockaddr_in *) ss;
1810	struct sockaddr_in6 *in6 = (struct sockaddr_in6 *) ss;
1811
1812	memset(ss, 0, sizeof(*ss));
1813
1814	if (in4_pton(str, len, (u8 *)&in4->sin_addr.s_addr, delim, ipend)) {
1815		ss->ss_family = AF_INET;
1816		return 0;
1817	}
1818
1819	if (in6_pton(str, len, (u8 *)&in6->sin6_addr.s6_addr, delim, ipend)) {
1820		ss->ss_family = AF_INET6;
1821		return 0;
1822	}
1823
1824	return -EINVAL;
1825}
1826
1827/*
1828 * Extract hostname string and resolve using kernel DNS facility.
1829 */
1830#ifdef CONFIG_CEPH_LIB_USE_DNS_RESOLVER
1831static int ceph_dns_resolve_name(const char *name, size_t namelen,
1832		struct sockaddr_storage *ss, char delim, const char **ipend)
1833{
1834	const char *end, *delim_p;
1835	char *colon_p, *ip_addr = NULL;
1836	int ip_len, ret;
1837
1838	/*
1839	 * The end of the hostname occurs immediately preceding the delimiter or
1840	 * the port marker (':') where the delimiter takes precedence.
1841	 */
1842	delim_p = memchr(name, delim, namelen);
1843	colon_p = memchr(name, ':', namelen);
1844
1845	if (delim_p && colon_p)
1846		end = delim_p < colon_p ? delim_p : colon_p;
1847	else if (!delim_p && colon_p)
1848		end = colon_p;
1849	else {
1850		end = delim_p;
1851		if (!end) /* case: hostname:/ */
1852			end = name + namelen;
1853	}
1854
1855	if (end <= name)
1856		return -EINVAL;
1857
1858	/* do dns_resolve upcall */
1859	ip_len = dns_query(NULL, name, end - name, NULL, &ip_addr, NULL);
1860	if (ip_len > 0)
1861		ret = ceph_pton(ip_addr, ip_len, ss, -1, NULL);
1862	else
1863		ret = -ESRCH;
1864
1865	kfree(ip_addr);
1866
1867	*ipend = end;
1868
1869	pr_info("resolve '%.*s' (ret=%d): %s\n", (int)(end - name), name,
1870			ret, ret ? "failed" : ceph_pr_addr(ss));
1871
1872	return ret;
1873}
1874#else
1875static inline int ceph_dns_resolve_name(const char *name, size_t namelen,
1876		struct sockaddr_storage *ss, char delim, const char **ipend)
1877{
1878	return -EINVAL;
1879}
1880#endif
1881
1882/*
1883 * Parse a server name (IP or hostname). If a valid IP address is not found
1884 * then try to extract a hostname to resolve using userspace DNS upcall.
1885 */
1886static int ceph_parse_server_name(const char *name, size_t namelen,
1887			struct sockaddr_storage *ss, char delim, const char **ipend)
1888{
1889	int ret;
1890
1891	ret = ceph_pton(name, namelen, ss, delim, ipend);
1892	if (ret)
1893		ret = ceph_dns_resolve_name(name, namelen, ss, delim, ipend);
1894
1895	return ret;
1896}
1897
1898/*
1899 * Parse an ip[:port] list into an addr array.  Use the default
1900 * monitor port if a port isn't specified.
1901 */
1902int ceph_parse_ips(const char *c, const char *end,
1903		   struct ceph_entity_addr *addr,
1904		   int max_count, int *count)
1905{
1906	int i, ret = -EINVAL;
1907	const char *p = c;
1908
1909	dout("parse_ips on '%.*s'\n", (int)(end-c), c);
1910	for (i = 0; i < max_count; i++) {
1911		const char *ipend;
1912		struct sockaddr_storage *ss = &addr[i].in_addr;
1913		int port;
1914		char delim = ',';
1915
1916		if (*p == '[') {
1917			delim = ']';
1918			p++;
1919		}
1920
1921		ret = ceph_parse_server_name(p, end - p, ss, delim, &ipend);
1922		if (ret)
1923			goto bad;
1924		ret = -EINVAL;
1925
1926		p = ipend;
1927
1928		if (delim == ']') {
1929			if (*p != ']') {
1930				dout("missing matching ']'\n");
1931				goto bad;
1932			}
1933			p++;
1934		}
1935
1936		/* port? */
1937		if (p < end && *p == ':') {
1938			port = 0;
1939			p++;
1940			while (p < end && *p >= '0' && *p <= '9') {
1941				port = (port * 10) + (*p - '0');
1942				p++;
1943			}
1944			if (port == 0)
1945				port = CEPH_MON_PORT;
1946			else if (port > 65535)
1947				goto bad;
1948		} else {
1949			port = CEPH_MON_PORT;
1950		}
1951
1952		addr_set_port(ss, port);
1953
1954		dout("parse_ips got %s\n", ceph_pr_addr(ss));
1955
1956		if (p == end)
1957			break;
1958		if (*p != ',')
1959			goto bad;
1960		p++;
1961	}
1962
1963	if (p != end)
1964		goto bad;
1965
1966	if (count)
1967		*count = i + 1;
1968	return 0;
1969
1970bad:
1971	pr_err("parse_ips bad ip '%.*s'\n", (int)(end - c), c);
1972	return ret;
1973}
1974EXPORT_SYMBOL(ceph_parse_ips);
1975
1976static int process_banner(struct ceph_connection *con)
1977{
1978	dout("process_banner on %p\n", con);
1979
1980	if (verify_hello(con) < 0)
1981		return -1;
1982
1983	ceph_decode_addr(&con->actual_peer_addr);
1984	ceph_decode_addr(&con->peer_addr_for_me);
1985
1986	/*
1987	 * Make sure the other end is who we wanted.  note that the other
1988	 * end may not yet know their ip address, so if it's 0.0.0.0, give
1989	 * them the benefit of the doubt.
1990	 */
1991	if (memcmp(&con->peer_addr, &con->actual_peer_addr,
1992		   sizeof(con->peer_addr)) != 0 &&
1993	    !(addr_is_blank(&con->actual_peer_addr.in_addr) &&
1994	      con->actual_peer_addr.nonce == con->peer_addr.nonce)) {
1995		pr_warn("wrong peer, want %s/%d, got %s/%d\n",
1996			ceph_pr_addr(&con->peer_addr.in_addr),
1997			(int)le32_to_cpu(con->peer_addr.nonce),
1998			ceph_pr_addr(&con->actual_peer_addr.in_addr),
1999			(int)le32_to_cpu(con->actual_peer_addr.nonce));
2000		con->error_msg = "wrong peer at address";
2001		return -1;
2002	}
2003
2004	/*
2005	 * did we learn our address?
2006	 */
2007	if (addr_is_blank(&con->msgr->inst.addr.in_addr)) {
2008		int port = addr_port(&con->msgr->inst.addr.in_addr);
2009
2010		memcpy(&con->msgr->inst.addr.in_addr,
2011		       &con->peer_addr_for_me.in_addr,
2012		       sizeof(con->peer_addr_for_me.in_addr));
2013		addr_set_port(&con->msgr->inst.addr.in_addr, port);
2014		encode_my_addr(con->msgr);
2015		dout("process_banner learned my addr is %s\n",
2016		     ceph_pr_addr(&con->msgr->inst.addr.in_addr));
2017	}
2018
2019	return 0;
2020}
2021
2022static int process_connect(struct ceph_connection *con)
2023{
2024	u64 sup_feat = con->msgr->supported_features;
2025	u64 req_feat = con->msgr->required_features;
2026	u64 server_feat = ceph_sanitize_features(
2027				le64_to_cpu(con->in_reply.features));
2028	int ret;
2029
2030	dout("process_connect on %p tag %d\n", con, (int)con->in_tag);
2031
2032	switch (con->in_reply.tag) {
2033	case CEPH_MSGR_TAG_FEATURES:
2034		pr_err("%s%lld %s feature set mismatch,"
2035		       " my %llx < server's %llx, missing %llx\n",
2036		       ENTITY_NAME(con->peer_name),
2037		       ceph_pr_addr(&con->peer_addr.in_addr),
2038		       sup_feat, server_feat, server_feat & ~sup_feat);
2039		con->error_msg = "missing required protocol features";
2040		reset_connection(con);
2041		return -1;
2042
2043	case CEPH_MSGR_TAG_BADPROTOVER:
2044		pr_err("%s%lld %s protocol version mismatch,"
2045		       " my %d != server's %d\n",
2046		       ENTITY_NAME(con->peer_name),
2047		       ceph_pr_addr(&con->peer_addr.in_addr),
2048		       le32_to_cpu(con->out_connect.protocol_version),
2049		       le32_to_cpu(con->in_reply.protocol_version));
2050		con->error_msg = "protocol version mismatch";
2051		reset_connection(con);
2052		return -1;
2053
2054	case CEPH_MSGR_TAG_BADAUTHORIZER:
2055		con->auth_retry++;
2056		dout("process_connect %p got BADAUTHORIZER attempt %d\n", con,
2057		     con->auth_retry);
2058		if (con->auth_retry == 2) {
2059			con->error_msg = "connect authorization failure";
2060			return -1;
2061		}
2062		con_out_kvec_reset(con);
2063		ret = prepare_write_connect(con);
2064		if (ret < 0)
2065			return ret;
2066		prepare_read_connect(con);
2067		break;
2068
2069	case CEPH_MSGR_TAG_RESETSESSION:
2070		/*
2071		 * If we connected with a large connect_seq but the peer
2072		 * has no record of a session with us (no connection, or
2073		 * connect_seq == 0), they will send RESETSESION to indicate
2074		 * that they must have reset their session, and may have
2075		 * dropped messages.
2076		 */
2077		dout("process_connect got RESET peer seq %u\n",
2078		     le32_to_cpu(con->in_reply.connect_seq));
2079		pr_err("%s%lld %s connection reset\n",
2080		       ENTITY_NAME(con->peer_name),
2081		       ceph_pr_addr(&con->peer_addr.in_addr));
2082		reset_connection(con);
2083		con_out_kvec_reset(con);
2084		ret = prepare_write_connect(con);
2085		if (ret < 0)
2086			return ret;
2087		prepare_read_connect(con);
2088
2089		/* Tell ceph about it. */
2090		mutex_unlock(&con->mutex);
2091		pr_info("reset on %s%lld\n", ENTITY_NAME(con->peer_name));
2092		if (con->ops->peer_reset)
2093			con->ops->peer_reset(con);
2094		mutex_lock(&con->mutex);
2095		if (con->state != CON_STATE_NEGOTIATING)
2096			return -EAGAIN;
2097		break;
2098
2099	case CEPH_MSGR_TAG_RETRY_SESSION:
2100		/*
2101		 * If we sent a smaller connect_seq than the peer has, try
2102		 * again with a larger value.
2103		 */
2104		dout("process_connect got RETRY_SESSION my seq %u, peer %u\n",
2105		     le32_to_cpu(con->out_connect.connect_seq),
2106		     le32_to_cpu(con->in_reply.connect_seq));
2107		con->connect_seq = le32_to_cpu(con->in_reply.connect_seq);
2108		con_out_kvec_reset(con);
2109		ret = prepare_write_connect(con);
2110		if (ret < 0)
2111			return ret;
2112		prepare_read_connect(con);
2113		break;
2114
2115	case CEPH_MSGR_TAG_RETRY_GLOBAL:
2116		/*
2117		 * If we sent a smaller global_seq than the peer has, try
2118		 * again with a larger value.
2119		 */
2120		dout("process_connect got RETRY_GLOBAL my %u peer_gseq %u\n",
2121		     con->peer_global_seq,
2122		     le32_to_cpu(con->in_reply.global_seq));
2123		get_global_seq(con->msgr,
2124			       le32_to_cpu(con->in_reply.global_seq));
2125		con_out_kvec_reset(con);
2126		ret = prepare_write_connect(con);
2127		if (ret < 0)
2128			return ret;
2129		prepare_read_connect(con);
2130		break;
2131
2132	case CEPH_MSGR_TAG_SEQ:
2133	case CEPH_MSGR_TAG_READY:
2134		if (req_feat & ~server_feat) {
2135			pr_err("%s%lld %s protocol feature mismatch,"
2136			       " my required %llx > server's %llx, need %llx\n",
2137			       ENTITY_NAME(con->peer_name),
2138			       ceph_pr_addr(&con->peer_addr.in_addr),
2139			       req_feat, server_feat, req_feat & ~server_feat);
2140			con->error_msg = "missing required protocol features";
2141			reset_connection(con);
2142			return -1;
2143		}
2144
2145		WARN_ON(con->state != CON_STATE_NEGOTIATING);
2146		con->state = CON_STATE_OPEN;
2147		con->auth_retry = 0;    /* we authenticated; clear flag */
2148		con->peer_global_seq = le32_to_cpu(con->in_reply.global_seq);
2149		con->connect_seq++;
2150		con->peer_features = server_feat;
2151		dout("process_connect got READY gseq %d cseq %d (%d)\n",
2152		     con->peer_global_seq,
2153		     le32_to_cpu(con->in_reply.connect_seq),
2154		     con->connect_seq);
2155		WARN_ON(con->connect_seq !=
2156			le32_to_cpu(con->in_reply.connect_seq));
2157
2158		if (con->in_reply.flags & CEPH_MSG_CONNECT_LOSSY)
2159			con_flag_set(con, CON_FLAG_LOSSYTX);
2160
2161		con->delay = 0;      /* reset backoff memory */
2162
2163		if (con->in_reply.tag == CEPH_MSGR_TAG_SEQ) {
2164			prepare_write_seq(con);
2165			prepare_read_seq(con);
2166		} else {
2167			prepare_read_tag(con);
2168		}
2169		break;
2170
2171	case CEPH_MSGR_TAG_WAIT:
2172		/*
2173		 * If there is a connection race (we are opening
2174		 * connections to each other), one of us may just have
2175		 * to WAIT.  This shouldn't happen if we are the
2176		 * client.
2177		 */
2178		con->error_msg = "protocol error, got WAIT as client";
2179		return -1;
2180
2181	default:
2182		con->error_msg = "protocol error, garbage tag during connect";
2183		return -1;
2184	}
2185	return 0;
2186}
2187
2188
2189/*
2190 * read (part of) an ack
2191 */
2192static int read_partial_ack(struct ceph_connection *con)
2193{
2194	int size = sizeof (con->in_temp_ack);
2195	int end = size;
2196
2197	return read_partial(con, end, size, &con->in_temp_ack);
2198}
2199
2200/*
2201 * We can finally discard anything that's been acked.
2202 */
2203static void process_ack(struct ceph_connection *con)
2204{
2205	struct ceph_msg *m;
2206	u64 ack = le64_to_cpu(con->in_temp_ack);
2207	u64 seq;
2208
2209	while (!list_empty(&con->out_sent)) {
2210		m = list_first_entry(&con->out_sent, struct ceph_msg,
2211				     list_head);
2212		seq = le64_to_cpu(m->hdr.seq);
2213		if (seq > ack)
2214			break;
2215		dout("got ack for seq %llu type %d at %p\n", seq,
2216		     le16_to_cpu(m->hdr.type), m);
2217		m->ack_stamp = jiffies;
2218		ceph_msg_remove(m);
2219	}
2220	prepare_read_tag(con);
2221}
2222
2223
2224static int read_partial_message_section(struct ceph_connection *con,
2225					struct kvec *section,
2226					unsigned int sec_len, u32 *crc)
2227{
2228	int ret, left;
2229
2230	BUG_ON(!section);
2231
2232	while (section->iov_len < sec_len) {
2233		BUG_ON(section->iov_base == NULL);
2234		left = sec_len - section->iov_len;
2235		ret = ceph_tcp_recvmsg(con->sock, (char *)section->iov_base +
2236				       section->iov_len, left);
2237		if (ret <= 0)
2238			return ret;
2239		section->iov_len += ret;
2240	}
2241	if (section->iov_len == sec_len)
2242		*crc = crc32c(0, section->iov_base, section->iov_len);
2243
2244	return 1;
2245}
2246
2247static int read_partial_msg_data(struct ceph_connection *con)
2248{
2249	struct ceph_msg *msg = con->in_msg;
2250	struct ceph_msg_data_cursor *cursor = &msg->cursor;
2251	const bool do_datacrc = !con->msgr->nocrc;
2252	struct page *page;
2253	size_t page_offset;
2254	size_t length;
2255	u32 crc = 0;
2256	int ret;
2257
2258	BUG_ON(!msg);
2259	if (list_empty(&msg->data))
2260		return -EIO;
2261
2262	if (do_datacrc)
2263		crc = con->in_data_crc;
2264	while (cursor->resid) {
2265		page = ceph_msg_data_next(&msg->cursor, &page_offset, &length,
2266							NULL);
2267		ret = ceph_tcp_recvpage(con->sock, page, page_offset, length);
2268		if (ret <= 0) {
2269			if (do_datacrc)
2270				con->in_data_crc = crc;
2271
2272			return ret;
2273		}
2274
2275		if (do_datacrc)
2276			crc = ceph_crc32c_page(crc, page, page_offset, ret);
2277		(void) ceph_msg_data_advance(&msg->cursor, (size_t)ret);
2278	}
2279	if (do_datacrc)
2280		con->in_data_crc = crc;
2281
2282	return 1;	/* must return > 0 to indicate success */
2283}
2284
2285/*
2286 * read (part of) a message.
2287 */
2288static int ceph_con_in_msg_alloc(struct ceph_connection *con, int *skip);
2289
2290static int read_partial_message(struct ceph_connection *con)
2291{
2292	struct ceph_msg *m = con->in_msg;
2293	int size;
2294	int end;
2295	int ret;
2296	unsigned int front_len, middle_len, data_len;
2297	bool do_datacrc = !con->msgr->nocrc;
2298	bool need_sign = (con->peer_features & CEPH_FEATURE_MSG_AUTH);
2299	u64 seq;
2300	u32 crc;
2301
2302	dout("read_partial_message con %p msg %p\n", con, m);
2303
2304	/* header */
2305	size = sizeof (con->in_hdr);
2306	end = size;
2307	ret = read_partial(con, end, size, &con->in_hdr);
2308	if (ret <= 0)
2309		return ret;
2310
2311	crc = crc32c(0, &con->in_hdr, offsetof(struct ceph_msg_header, crc));
2312	if (cpu_to_le32(crc) != con->in_hdr.crc) {
2313		pr_err("read_partial_message bad hdr crc %u != expected %u\n",
2314		       crc, con->in_hdr.crc);
2315		return -EBADMSG;
2316	}
2317
2318	front_len = le32_to_cpu(con->in_hdr.front_len);
2319	if (front_len > CEPH_MSG_MAX_FRONT_LEN)
2320		return -EIO;
2321	middle_len = le32_to_cpu(con->in_hdr.middle_len);
2322	if (middle_len > CEPH_MSG_MAX_MIDDLE_LEN)
2323		return -EIO;
2324	data_len = le32_to_cpu(con->in_hdr.data_len);
2325	if (data_len > CEPH_MSG_MAX_DATA_LEN)
2326		return -EIO;
2327
2328	/* verify seq# */
2329	seq = le64_to_cpu(con->in_hdr.seq);
2330	if ((s64)seq - (s64)con->in_seq < 1) {
2331		pr_info("skipping %s%lld %s seq %lld expected %lld\n",
2332			ENTITY_NAME(con->peer_name),
2333			ceph_pr_addr(&con->peer_addr.in_addr),
2334			seq, con->in_seq + 1);
2335		con->in_base_pos = -front_len - middle_len - data_len -
2336			sizeof_footer(con);
2337		con->in_tag = CEPH_MSGR_TAG_READY;
2338		return 1;
2339	} else if ((s64)seq - (s64)con->in_seq > 1) {
2340		pr_err("read_partial_message bad seq %lld expected %lld\n",
2341		       seq, con->in_seq + 1);
2342		con->error_msg = "bad message sequence # for incoming message";
2343		return -EBADE;
2344	}
2345
2346	/* allocate message? */
2347	if (!con->in_msg) {
2348		int skip = 0;
2349
2350		dout("got hdr type %d front %d data %d\n", con->in_hdr.type,
2351		     front_len, data_len);
2352		ret = ceph_con_in_msg_alloc(con, &skip);
2353		if (ret < 0)
2354			return ret;
2355
2356		BUG_ON(!con->in_msg ^ skip);
2357		if (con->in_msg && data_len > con->in_msg->data_length) {
2358			pr_warn("%s skipping long message (%u > %zd)\n",
2359				__func__, data_len, con->in_msg->data_length);
2360			ceph_msg_put(con->in_msg);
2361			con->in_msg = NULL;
2362			skip = 1;
2363		}
2364		if (skip) {
2365			/* skip this message */
2366			dout("alloc_msg said skip message\n");
2367			con->in_base_pos = -front_len - middle_len - data_len -
2368				sizeof_footer(con);
2369			con->in_tag = CEPH_MSGR_TAG_READY;
2370			con->in_seq++;
2371			return 1;
2372		}
2373
2374		BUG_ON(!con->in_msg);
2375		BUG_ON(con->in_msg->con != con);
2376		m = con->in_msg;
2377		m->front.iov_len = 0;    /* haven't read it yet */
2378		if (m->middle)
2379			m->middle->vec.iov_len = 0;
2380
2381		/* prepare for data payload, if any */
2382
2383		if (data_len)
2384			prepare_message_data(con->in_msg, data_len);
2385	}
2386
2387	/* front */
2388	ret = read_partial_message_section(con, &m->front, front_len,
2389					   &con->in_front_crc);
2390	if (ret <= 0)
2391		return ret;
2392
2393	/* middle */
2394	if (m->middle) {
2395		ret = read_partial_message_section(con, &m->middle->vec,
2396						   middle_len,
2397						   &con->in_middle_crc);
2398		if (ret <= 0)
2399			return ret;
2400	}
2401
2402	/* (page) data */
2403	if (data_len) {
2404		ret = read_partial_msg_data(con);
2405		if (ret <= 0)
2406			return ret;
2407	}
2408
2409	/* footer */
2410	if (need_sign)
2411		size = sizeof(m->footer);
2412	else
2413		size = sizeof(m->old_footer);
2414
2415	end += size;
2416	ret = read_partial(con, end, size, &m->footer);
2417	if (ret <= 0)
2418		return ret;
2419
2420	if (!need_sign) {
2421		m->footer.flags = m->old_footer.flags;
2422		m->footer.sig = 0;
2423	}
2424
2425	dout("read_partial_message got msg %p %d (%u) + %d (%u) + %d (%u)\n",
2426	     m, front_len, m->footer.front_crc, middle_len,
2427	     m->footer.middle_crc, data_len, m->footer.data_crc);
2428
2429	/* crc ok? */
2430	if (con->in_front_crc != le32_to_cpu(m->footer.front_crc)) {
2431		pr_err("read_partial_message %p front crc %u != exp. %u\n",
2432		       m, con->in_front_crc, m->footer.front_crc);
2433		return -EBADMSG;
2434	}
2435	if (con->in_middle_crc != le32_to_cpu(m->footer.middle_crc)) {
2436		pr_err("read_partial_message %p middle crc %u != exp %u\n",
2437		       m, con->in_middle_crc, m->footer.middle_crc);
2438		return -EBADMSG;
2439	}
2440	if (do_datacrc &&
2441	    (m->footer.flags & CEPH_MSG_FOOTER_NOCRC) == 0 &&
2442	    con->in_data_crc != le32_to_cpu(m->footer.data_crc)) {
2443		pr_err("read_partial_message %p data crc %u != exp. %u\n", m,
2444		       con->in_data_crc, le32_to_cpu(m->footer.data_crc));
2445		return -EBADMSG;
2446	}
2447
2448	if (need_sign && con->ops->check_message_signature &&
2449	    con->ops->check_message_signature(con, m)) {
2450		pr_err("read_partial_message %p signature check failed\n", m);
2451		return -EBADMSG;
2452	}
2453
2454	return 1; /* done! */
2455}
2456
2457/*
2458 * Process message.  This happens in the worker thread.  The callback should
2459 * be careful not to do anything that waits on other incoming messages or it
2460 * may deadlock.
2461 */
2462static void process_message(struct ceph_connection *con)
2463{
2464	struct ceph_msg *msg;
2465
2466	BUG_ON(con->in_msg->con != con);
2467	con->in_msg->con = NULL;
2468	msg = con->in_msg;
2469	con->in_msg = NULL;
2470	con->ops->put(con);
2471
2472	/* if first message, set peer_name */
2473	if (con->peer_name.type == 0)
2474		con->peer_name = msg->hdr.src;
2475
2476	con->in_seq++;
2477	mutex_unlock(&con->mutex);
2478
2479	dout("===== %p %llu from %s%lld %d=%s len %d+%d (%u %u %u) =====\n",
2480	     msg, le64_to_cpu(msg->hdr.seq),
2481	     ENTITY_NAME(msg->hdr.src),
2482	     le16_to_cpu(msg->hdr.type),
2483	     ceph_msg_type_name(le16_to_cpu(msg->hdr.type)),
2484	     le32_to_cpu(msg->hdr.front_len),
2485	     le32_to_cpu(msg->hdr.data_len),
2486	     con->in_front_crc, con->in_middle_crc, con->in_data_crc);
2487	con->ops->dispatch(con, msg);
2488
2489	mutex_lock(&con->mutex);
2490}
2491
2492
2493/*
2494 * Write something to the socket.  Called in a worker thread when the
2495 * socket appears to be writeable and we have something ready to send.
2496 */
2497static int try_write(struct ceph_connection *con)
2498{
2499	int ret = 1;
2500
2501	dout("try_write start %p state %lu\n", con, con->state);
2502
2503more:
2504	dout("try_write out_kvec_bytes %d\n", con->out_kvec_bytes);
2505
2506	/* open the socket first? */
2507	if (con->state == CON_STATE_PREOPEN) {
2508		BUG_ON(con->sock);
2509		con->state = CON_STATE_CONNECTING;
2510
2511		con_out_kvec_reset(con);
2512		prepare_write_banner(con);
2513		prepare_read_banner(con);
2514
2515		BUG_ON(con->in_msg);
2516		con->in_tag = CEPH_MSGR_TAG_READY;
2517		dout("try_write initiating connect on %p new state %lu\n",
2518		     con, con->state);
2519		ret = ceph_tcp_connect(con);
2520		if (ret < 0) {
2521			con->error_msg = "connect error";
2522			goto out;
2523		}
2524	}
2525
2526more_kvec:
2527	/* kvec data queued? */
2528	if (con->out_kvec_left) {
2529		ret = write_partial_kvec(con);
2530		if (ret <= 0)
2531			goto out;
2532	}
2533	if (con->out_skip) {
2534		ret = write_partial_skip(con);
2535		if (ret <= 0)
2536			goto out;
2537	}
2538
2539	/* msg pages? */
2540	if (con->out_msg) {
2541		if (con->out_msg_done) {
2542			ceph_msg_put(con->out_msg);
2543			con->out_msg = NULL;   /* we're done with this one */
2544			goto do_next;
2545		}
2546
2547		ret = write_partial_message_data(con);
2548		if (ret == 1)
2549			goto more_kvec;  /* we need to send the footer, too! */
2550		if (ret == 0)
2551			goto out;
2552		if (ret < 0) {
2553			dout("try_write write_partial_message_data err %d\n",
2554			     ret);
2555			goto out;
2556		}
2557	}
2558
2559do_next:
2560	if (con->state == CON_STATE_OPEN) {
2561		/* is anything else pending? */
2562		if (!list_empty(&con->out_queue)) {
2563			prepare_write_message(con);
2564			goto more;
2565		}
2566		if (con->in_seq > con->in_seq_acked) {
2567			prepare_write_ack(con);
2568			goto more;
2569		}
2570		if (con_flag_test_and_clear(con, CON_FLAG_KEEPALIVE_PENDING)) {
2571			prepare_write_keepalive(con);
2572			goto more;
2573		}
2574	}
2575
2576	/* Nothing to do! */
2577	con_flag_clear(con, CON_FLAG_WRITE_PENDING);
2578	dout("try_write nothing else to write.\n");
2579	ret = 0;
2580out:
2581	dout("try_write done on %p ret %d\n", con, ret);
2582	return ret;
2583}
2584
2585
2586
2587/*
2588 * Read what we can from the socket.
2589 */
2590static int try_read(struct ceph_connection *con)
2591{
2592	int ret = -1;
2593
2594more:
2595	dout("try_read start on %p state %lu\n", con, con->state);
2596	if (con->state != CON_STATE_CONNECTING &&
2597	    con->state != CON_STATE_NEGOTIATING &&
2598	    con->state != CON_STATE_OPEN)
2599		return 0;
2600
2601	BUG_ON(!con->sock);
2602
2603	dout("try_read tag %d in_base_pos %d\n", (int)con->in_tag,
2604	     con->in_base_pos);
2605
2606	if (con->state == CON_STATE_CONNECTING) {
2607		dout("try_read connecting\n");
2608		ret = read_partial_banner(con);
2609		if (ret <= 0)
2610			goto out;
2611		ret = process_banner(con);
2612		if (ret < 0)
2613			goto out;
2614
2615		con->state = CON_STATE_NEGOTIATING;
2616
2617		/*
2618		 * Received banner is good, exchange connection info.
2619		 * Do not reset out_kvec, as sending our banner raced
2620		 * with receiving peer banner after connect completed.
2621		 */
2622		ret = prepare_write_connect(con);
2623		if (ret < 0)
2624			goto out;
2625		prepare_read_connect(con);
2626
2627		/* Send connection info before awaiting response */
2628		goto out;
2629	}
2630
2631	if (con->state == CON_STATE_NEGOTIATING) {
2632		dout("try_read negotiating\n");
2633		ret = read_partial_connect(con);
2634		if (ret <= 0)
2635			goto out;
2636		ret = process_connect(con);
2637		if (ret < 0)
2638			goto out;
2639		goto more;
2640	}
2641
2642	WARN_ON(con->state != CON_STATE_OPEN);
2643
2644	if (con->in_base_pos < 0) {
2645		/*
2646		 * skipping + discarding content.
2647		 *
2648		 * FIXME: there must be a better way to do this!
2649		 */
2650		static char buf[SKIP_BUF_SIZE];
2651		int skip = min((int) sizeof (buf), -con->in_base_pos);
2652
2653		dout("skipping %d / %d bytes\n", skip, -con->in_base_pos);
2654		ret = ceph_tcp_recvmsg(con->sock, buf, skip);
2655		if (ret <= 0)
2656			goto out;
2657		con->in_base_pos += ret;
2658		if (con->in_base_pos)
2659			goto more;
2660	}
2661	if (con->in_tag == CEPH_MSGR_TAG_READY) {
2662		/*
2663		 * what's next?
2664		 */
2665		ret = ceph_tcp_recvmsg(con->sock, &con->in_tag, 1);
2666		if (ret <= 0)
2667			goto out;
2668		dout("try_read got tag %d\n", (int)con->in_tag);
2669		switch (con->in_tag) {
2670		case CEPH_MSGR_TAG_MSG:
2671			prepare_read_message(con);
2672			break;
2673		case CEPH_MSGR_TAG_ACK:
2674			prepare_read_ack(con);
2675			break;
2676		case CEPH_MSGR_TAG_CLOSE:
2677			con_close_socket(con);
2678			con->state = CON_STATE_CLOSED;
2679			goto out;
2680		default:
2681			goto bad_tag;
2682		}
2683	}
2684	if (con->in_tag == CEPH_MSGR_TAG_MSG) {
2685		ret = read_partial_message(con);
2686		if (ret <= 0) {
2687			switch (ret) {
2688			case -EBADMSG:
2689				con->error_msg = "bad crc";
2690				/* fall through */
2691			case -EBADE:
2692				ret = -EIO;
2693				break;
2694			case -EIO:
2695				con->error_msg = "io error";
2696				break;
2697			}
2698			goto out;
2699		}
2700		if (con->in_tag == CEPH_MSGR_TAG_READY)
2701			goto more;
2702		process_message(con);
2703		if (con->state == CON_STATE_OPEN)
2704			prepare_read_tag(con);
2705		goto more;
2706	}
2707	if (con->in_tag == CEPH_MSGR_TAG_ACK ||
2708	    con->in_tag == CEPH_MSGR_TAG_SEQ) {
2709		/*
2710		 * the final handshake seq exchange is semantically
2711		 * equivalent to an ACK
2712		 */
2713		ret = read_partial_ack(con);
2714		if (ret <= 0)
2715			goto out;
2716		process_ack(con);
2717		goto more;
2718	}
2719
2720out:
2721	dout("try_read done on %p ret %d\n", con, ret);
2722	return ret;
2723
2724bad_tag:
2725	pr_err("try_read bad con->in_tag = %d\n", (int)con->in_tag);
2726	con->error_msg = "protocol error, garbage tag";
2727	ret = -1;
2728	goto out;
2729}
2730
2731
2732/*
2733 * Atomically queue work on a connection after the specified delay.
2734 * Bump @con reference to avoid races with connection teardown.
2735 * Returns 0 if work was queued, or an error code otherwise.
2736 */
2737static int queue_con_delay(struct ceph_connection *con, unsigned long delay)
2738{
2739	if (!con->ops->get(con)) {
2740		dout("%s %p ref count 0\n", __func__, con);
2741		return -ENOENT;
2742	}
2743
2744	if (!queue_delayed_work(ceph_msgr_wq, &con->work, delay)) {
2745		dout("%s %p - already queued\n", __func__, con);
2746		con->ops->put(con);
2747		return -EBUSY;
2748	}
2749
2750	dout("%s %p %lu\n", __func__, con, delay);
2751	return 0;
2752}
2753
2754static void queue_con(struct ceph_connection *con)
2755{
2756	(void) queue_con_delay(con, 0);
2757}
2758
2759static void cancel_con(struct ceph_connection *con)
2760{
2761	if (cancel_delayed_work(&con->work)) {
2762		dout("%s %p\n", __func__, con);
2763		con->ops->put(con);
2764	}
2765}
2766
2767static bool con_sock_closed(struct ceph_connection *con)
2768{
2769	if (!con_flag_test_and_clear(con, CON_FLAG_SOCK_CLOSED))
2770		return false;
2771
2772#define CASE(x)								\
2773	case CON_STATE_ ## x:						\
2774		con->error_msg = "socket closed (con state " #x ")";	\
2775		break;
2776
2777	switch (con->state) {
2778	CASE(CLOSED);
2779	CASE(PREOPEN);
2780	CASE(CONNECTING);
2781	CASE(NEGOTIATING);
2782	CASE(OPEN);
2783	CASE(STANDBY);
2784	default:
2785		pr_warn("%s con %p unrecognized state %lu\n",
2786			__func__, con, con->state);
2787		con->error_msg = "unrecognized con state";
2788		BUG();
2789		break;
2790	}
2791#undef CASE
2792
2793	return true;
2794}
2795
2796static bool con_backoff(struct ceph_connection *con)
2797{
2798	int ret;
2799
2800	if (!con_flag_test_and_clear(con, CON_FLAG_BACKOFF))
2801		return false;
2802
2803	ret = queue_con_delay(con, round_jiffies_relative(con->delay));
2804	if (ret) {
2805		dout("%s: con %p FAILED to back off %lu\n", __func__,
2806			con, con->delay);
2807		BUG_ON(ret == -ENOENT);
2808		con_flag_set(con, CON_FLAG_BACKOFF);
2809	}
2810
2811	return true;
2812}
2813
2814/* Finish fault handling; con->mutex must *not* be held here */
2815
2816static void con_fault_finish(struct ceph_connection *con)
2817{
2818	/*
2819	 * in case we faulted due to authentication, invalidate our
2820	 * current tickets so that we can get new ones.
2821	 */
2822	if (con->auth_retry && con->ops->invalidate_authorizer) {
2823		dout("calling invalidate_authorizer()\n");
2824		con->ops->invalidate_authorizer(con);
2825	}
2826
2827	if (con->ops->fault)
2828		con->ops->fault(con);
2829}
2830
2831/*
2832 * Do some work on a connection.  Drop a connection ref when we're done.
2833 */
2834static void con_work(struct work_struct *work)
2835{
2836	struct ceph_connection *con = container_of(work, struct ceph_connection,
2837						   work.work);
2838	bool fault;
2839
2840	mutex_lock(&con->mutex);
2841	while (true) {
2842		int ret;
2843
2844		if ((fault = con_sock_closed(con))) {
2845			dout("%s: con %p SOCK_CLOSED\n", __func__, con);
2846			break;
2847		}
2848		if (con_backoff(con)) {
2849			dout("%s: con %p BACKOFF\n", __func__, con);
2850			break;
2851		}
2852		if (con->state == CON_STATE_STANDBY) {
2853			dout("%s: con %p STANDBY\n", __func__, con);
2854			break;
2855		}
2856		if (con->state == CON_STATE_CLOSED) {
2857			dout("%s: con %p CLOSED\n", __func__, con);
2858			BUG_ON(con->sock);
2859			break;
2860		}
2861		if (con->state == CON_STATE_PREOPEN) {
2862			dout("%s: con %p PREOPEN\n", __func__, con);
2863			BUG_ON(con->sock);
2864		}
2865
2866		ret = try_read(con);
2867		if (ret < 0) {
2868			if (ret == -EAGAIN)
2869				continue;
2870			if (!con->error_msg)
2871				con->error_msg = "socket error on read";
2872			fault = true;
2873			break;
2874		}
2875
2876		ret = try_write(con);
2877		if (ret < 0) {
2878			if (ret == -EAGAIN)
2879				continue;
2880			if (!con->error_msg)
2881				con->error_msg = "socket error on write";
2882			fault = true;
2883		}
2884
2885		break;	/* If we make it to here, we're done */
2886	}
2887	if (fault)
2888		con_fault(con);
2889	mutex_unlock(&con->mutex);
2890
2891	if (fault)
2892		con_fault_finish(con);
2893
2894	con->ops->put(con);
2895}
2896
2897/*
2898 * Generic error/fault handler.  A retry mechanism is used with
2899 * exponential backoff
2900 */
2901static void con_fault(struct ceph_connection *con)
2902{
2903	dout("fault %p state %lu to peer %s\n",
2904	     con, con->state, ceph_pr_addr(&con->peer_addr.in_addr));
2905
2906	pr_warn("%s%lld %s %s\n", ENTITY_NAME(con->peer_name),
2907		ceph_pr_addr(&con->peer_addr.in_addr), con->error_msg);
2908	con->error_msg = NULL;
2909
2910	WARN_ON(con->state != CON_STATE_CONNECTING &&
2911	       con->state != CON_STATE_NEGOTIATING &&
2912	       con->state != CON_STATE_OPEN);
2913
2914	con_close_socket(con);
2915
2916	if (con_flag_test(con, CON_FLAG_LOSSYTX)) {
2917		dout("fault on LOSSYTX channel, marking CLOSED\n");
2918		con->state = CON_STATE_CLOSED;
2919		return;
2920	}
2921
2922	if (con->in_msg) {
2923		BUG_ON(con->in_msg->con != con);
2924		con->in_msg->con = NULL;
2925		ceph_msg_put(con->in_msg);
2926		con->in_msg = NULL;
2927		con->ops->put(con);
2928	}
2929
2930	/* Requeue anything that hasn't been acked */
2931	list_splice_init(&con->out_sent, &con->out_queue);
2932
2933	/* If there are no messages queued or keepalive pending, place
2934	 * the connection in a STANDBY state */
2935	if (list_empty(&con->out_queue) &&
2936	    !con_flag_test(con, CON_FLAG_KEEPALIVE_PENDING)) {
2937		dout("fault %p setting STANDBY clearing WRITE_PENDING\n", con);
2938		con_flag_clear(con, CON_FLAG_WRITE_PENDING);
2939		con->state = CON_STATE_STANDBY;
2940	} else {
2941		/* retry after a delay. */
2942		con->state = CON_STATE_PREOPEN;
2943		if (con->delay == 0)
2944			con->delay = BASE_DELAY_INTERVAL;
2945		else if (con->delay < MAX_DELAY_INTERVAL)
2946			con->delay *= 2;
2947		con_flag_set(con, CON_FLAG_BACKOFF);
2948		queue_con(con);
2949	}
2950}
2951
2952
2953
2954/*
2955 * initialize a new messenger instance
2956 */
2957void ceph_messenger_init(struct ceph_messenger *msgr,
2958			struct ceph_entity_addr *myaddr,
2959			u64 supported_features,
2960			u64 required_features,
2961			bool nocrc,
2962			bool tcp_nodelay)
2963{
2964	msgr->supported_features = supported_features;
2965	msgr->required_features = required_features;
2966
2967	spin_lock_init(&msgr->global_seq_lock);
2968
2969	if (myaddr)
2970		msgr->inst.addr = *myaddr;
2971
2972	/* select a random nonce */
2973	msgr->inst.addr.type = 0;
2974	get_random_bytes(&msgr->inst.addr.nonce, sizeof(msgr->inst.addr.nonce));
2975	encode_my_addr(msgr);
2976	msgr->nocrc = nocrc;
2977	msgr->tcp_nodelay = tcp_nodelay;
2978
2979	atomic_set(&msgr->stopping, 0);
2980
2981	dout("%s %p\n", __func__, msgr);
2982}
2983EXPORT_SYMBOL(ceph_messenger_init);
2984
2985static void clear_standby(struct ceph_connection *con)
2986{
2987	/* come back from STANDBY? */
2988	if (con->state == CON_STATE_STANDBY) {
2989		dout("clear_standby %p and ++connect_seq\n", con);
2990		con->state = CON_STATE_PREOPEN;
2991		con->connect_seq++;
2992		WARN_ON(con_flag_test(con, CON_FLAG_WRITE_PENDING));
2993		WARN_ON(con_flag_test(con, CON_FLAG_KEEPALIVE_PENDING));
2994	}
2995}
2996
2997/*
2998 * Queue up an outgoing message on the given connection.
2999 */
3000void ceph_con_send(struct ceph_connection *con, struct ceph_msg *msg)
3001{
3002	/* set src+dst */
3003	msg->hdr.src = con->msgr->inst.name;
3004	BUG_ON(msg->front.iov_len != le32_to_cpu(msg->hdr.front_len));
3005	msg->needs_out_seq = true;
3006
3007	mutex_lock(&con->mutex);
3008
3009	if (con->state == CON_STATE_CLOSED) {
3010		dout("con_send %p closed, dropping %p\n", con, msg);
3011		ceph_msg_put(msg);
3012		mutex_unlock(&con->mutex);
3013		return;
3014	}
3015
3016	BUG_ON(msg->con != NULL);
3017	msg->con = con->ops->get(con);
3018	BUG_ON(msg->con == NULL);
3019
3020	BUG_ON(!list_empty(&msg->list_head));
3021	list_add_tail(&msg->list_head, &con->out_queue);
3022	dout("----- %p to %s%lld %d=%s len %d+%d+%d -----\n", msg,
3023	     ENTITY_NAME(con->peer_name), le16_to_cpu(msg->hdr.type),
3024	     ceph_msg_type_name(le16_to_cpu(msg->hdr.type)),
3025	     le32_to_cpu(msg->hdr.front_len),
3026	     le32_to_cpu(msg->hdr.middle_len),
3027	     le32_to_cpu(msg->hdr.data_len));
3028
3029	clear_standby(con);
3030	mutex_unlock(&con->mutex);
3031
3032	/* if there wasn't anything waiting to send before, queue
3033	 * new work */
3034	if (con_flag_test_and_set(con, CON_FLAG_WRITE_PENDING) == 0)
3035		queue_con(con);
3036}
3037EXPORT_SYMBOL(ceph_con_send);
3038
3039/*
3040 * Revoke a message that was previously queued for send
3041 */
3042void ceph_msg_revoke(struct ceph_msg *msg)
3043{
3044	struct ceph_connection *con = msg->con;
3045
3046	if (!con)
3047		return;		/* Message not in our possession */
3048
3049	mutex_lock(&con->mutex);
3050	if (!list_empty(&msg->list_head)) {
3051		dout("%s %p msg %p - was on queue\n", __func__, con, msg);
3052		list_del_init(&msg->list_head);
3053		BUG_ON(msg->con == NULL);
3054		msg->con->ops->put(msg->con);
3055		msg->con = NULL;
3056		msg->hdr.seq = 0;
3057
3058		ceph_msg_put(msg);
3059	}
3060	if (con->out_msg == msg) {
3061		BUG_ON(con->out_skip);
3062		/* footer */
3063		if (con->out_msg_done) {
3064			con->out_skip += con_out_kvec_skip(con);
3065		} else {
3066			BUG_ON(!msg->data_length);
3067			if (con->peer_features & CEPH_FEATURE_MSG_AUTH)
3068				con->out_skip += sizeof(msg->footer);
3069			else
3070				con->out_skip += sizeof(msg->old_footer);
3071		}
3072		/* data, middle, front */
3073		if (msg->data_length)
3074			con->out_skip += msg->cursor.total_resid;
3075		if (msg->middle)
3076			con->out_skip += con_out_kvec_skip(con);
3077		con->out_skip += con_out_kvec_skip(con);
3078
3079		dout("%s %p msg %p - was sending, will write %d skip %d\n",
3080		     __func__, con, msg, con->out_kvec_bytes, con->out_skip);
3081		msg->hdr.seq = 0;
3082		con->out_msg = NULL;
3083		ceph_msg_put(msg);
3084	}
3085
3086	mutex_unlock(&con->mutex);
3087}
3088
3089/*
3090 * Revoke a message that we may be reading data into
3091 */
3092void ceph_msg_revoke_incoming(struct ceph_msg *msg)
3093{
3094	struct ceph_connection *con;
3095
3096	BUG_ON(msg == NULL);
3097	if (!msg->con) {
3098		dout("%s msg %p null con\n", __func__, msg);
3099
3100		return;		/* Message not in our possession */
3101	}
3102
3103	con = msg->con;
3104	mutex_lock(&con->mutex);
3105	if (con->in_msg == msg) {
3106		unsigned int front_len = le32_to_cpu(con->in_hdr.front_len);
3107		unsigned int middle_len = le32_to_cpu(con->in_hdr.middle_len);
3108		unsigned int data_len = le32_to_cpu(con->in_hdr.data_len);
3109
3110		/* skip rest of message */
3111		dout("%s %p msg %p revoked\n", __func__, con, msg);
3112		con->in_base_pos = con->in_base_pos -
3113				sizeof(struct ceph_msg_header) -
3114				front_len -
3115				middle_len -
3116				data_len -
3117				sizeof(struct ceph_msg_footer);
3118		ceph_msg_put(con->in_msg);
3119		con->in_msg = NULL;
3120		con->in_tag = CEPH_MSGR_TAG_READY;
3121		con->in_seq++;
3122	} else {
3123		dout("%s %p in_msg %p msg %p no-op\n",
3124		     __func__, con, con->in_msg, msg);
3125	}
3126	mutex_unlock(&con->mutex);
3127}
3128
3129/*
3130 * Queue a keepalive byte to ensure the tcp connection is alive.
3131 */
3132void ceph_con_keepalive(struct ceph_connection *con)
3133{
3134	dout("con_keepalive %p\n", con);
3135	mutex_lock(&con->mutex);
3136	clear_standby(con);
3137	mutex_unlock(&con->mutex);
3138	if (con_flag_test_and_set(con, CON_FLAG_KEEPALIVE_PENDING) == 0 &&
3139	    con_flag_test_and_set(con, CON_FLAG_WRITE_PENDING) == 0)
3140		queue_con(con);
3141}
3142EXPORT_SYMBOL(ceph_con_keepalive);
3143
3144static struct ceph_msg_data *ceph_msg_data_create(enum ceph_msg_data_type type)
3145{
3146	struct ceph_msg_data *data;
3147
3148	if (WARN_ON(!ceph_msg_data_type_valid(type)))
3149		return NULL;
3150
3151	data = kmem_cache_zalloc(ceph_msg_data_cache, GFP_NOFS);
3152	if (data)
3153		data->type = type;
3154	INIT_LIST_HEAD(&data->links);
3155
3156	return data;
3157}
3158
3159static void ceph_msg_data_destroy(struct ceph_msg_data *data)
3160{
3161	if (!data)
3162		return;
3163
3164	WARN_ON(!list_empty(&data->links));
3165	if (data->type == CEPH_MSG_DATA_PAGELIST)
3166		ceph_pagelist_release(data->pagelist);
3167	kmem_cache_free(ceph_msg_data_cache, data);
3168}
3169
3170void ceph_msg_data_add_pages(struct ceph_msg *msg, struct page **pages,
3171		size_t length, size_t alignment)
3172{
3173	struct ceph_msg_data *data;
3174
3175	BUG_ON(!pages);
3176	BUG_ON(!length);
3177
3178	data = ceph_msg_data_create(CEPH_MSG_DATA_PAGES);
3179	BUG_ON(!data);
3180	data->pages = pages;
3181	data->length = length;
3182	data->alignment = alignment & ~PAGE_MASK;
3183
3184	list_add_tail(&data->links, &msg->data);
3185	msg->data_length += length;
3186}
3187EXPORT_SYMBOL(ceph_msg_data_add_pages);
3188
3189void ceph_msg_data_add_pagelist(struct ceph_msg *msg,
3190				struct ceph_pagelist *pagelist)
3191{
3192	struct ceph_msg_data *data;
3193
3194	BUG_ON(!pagelist);
3195	BUG_ON(!pagelist->length);
3196
3197	data = ceph_msg_data_create(CEPH_MSG_DATA_PAGELIST);
3198	BUG_ON(!data);
3199	data->pagelist = pagelist;
3200
3201	list_add_tail(&data->links, &msg->data);
3202	msg->data_length += pagelist->length;
3203}
3204EXPORT_SYMBOL(ceph_msg_data_add_pagelist);
3205
3206#ifdef	CONFIG_BLOCK
3207void ceph_msg_data_add_bio(struct ceph_msg *msg, struct bio *bio,
3208		size_t length)
3209{
3210	struct ceph_msg_data *data;
3211
3212	BUG_ON(!bio);
3213
3214	data = ceph_msg_data_create(CEPH_MSG_DATA_BIO);
3215	BUG_ON(!data);
3216	data->bio = bio;
3217	data->bio_length = length;
3218
3219	list_add_tail(&data->links, &msg->data);
3220	msg->data_length += length;
3221}
3222EXPORT_SYMBOL(ceph_msg_data_add_bio);
3223#endif	/* CONFIG_BLOCK */
3224
3225/*
3226 * construct a new message with given type, size
3227 * the new msg has a ref count of 1.
3228 */
3229struct ceph_msg *ceph_msg_new(int type, int front_len, gfp_t flags,
3230			      bool can_fail)
3231{
3232	struct ceph_msg *m;
3233
3234	m = kmem_cache_zalloc(ceph_msg_cache, flags);
3235	if (m == NULL)
3236		goto out;
3237
3238	m->hdr.type = cpu_to_le16(type);
3239	m->hdr.priority = cpu_to_le16(CEPH_MSG_PRIO_DEFAULT);
3240	m->hdr.front_len = cpu_to_le32(front_len);
3241
3242	INIT_LIST_HEAD(&m->list_head);
3243	kref_init(&m->kref);
3244	INIT_LIST_HEAD(&m->data);
3245
3246	/* front */
3247	if (front_len) {
3248		m->front.iov_base = ceph_kvmalloc(front_len, flags);
3249		if (m->front.iov_base == NULL) {
3250			dout("ceph_msg_new can't allocate %d bytes\n",
3251			     front_len);
3252			goto out2;
3253		}
3254	} else {
3255		m->front.iov_base = NULL;
3256	}
3257	m->front_alloc_len = m->front.iov_len = front_len;
3258
3259	dout("ceph_msg_new %p front %d\n", m, front_len);
3260	return m;
3261
3262out2:
3263	ceph_msg_put(m);
3264out:
3265	if (!can_fail) {
3266		pr_err("msg_new can't create type %d front %d\n", type,
3267		       front_len);
3268		WARN_ON(1);
3269	} else {
3270		dout("msg_new can't create type %d front %d\n", type,
3271		     front_len);
3272	}
3273	return NULL;
3274}
3275EXPORT_SYMBOL(ceph_msg_new);
3276
3277/*
3278 * Allocate "middle" portion of a message, if it is needed and wasn't
3279 * allocated by alloc_msg.  This allows us to read a small fixed-size
3280 * per-type header in the front and then gracefully fail (i.e.,
3281 * propagate the error to the caller based on info in the front) when
3282 * the middle is too large.
3283 */
3284static int ceph_alloc_middle(struct ceph_connection *con, struct ceph_msg *msg)
3285{
3286	int type = le16_to_cpu(msg->hdr.type);
3287	int middle_len = le32_to_cpu(msg->hdr.middle_len);
3288
3289	dout("alloc_middle %p type %d %s middle_len %d\n", msg, type,
3290	     ceph_msg_type_name(type), middle_len);
3291	BUG_ON(!middle_len);
3292	BUG_ON(msg->middle);
3293
3294	msg->middle = ceph_buffer_new(middle_len, GFP_NOFS);
3295	if (!msg->middle)
3296		return -ENOMEM;
3297	return 0;
3298}
3299
3300/*
3301 * Allocate a message for receiving an incoming message on a
3302 * connection, and save the result in con->in_msg.  Uses the
3303 * connection's private alloc_msg op if available.
3304 *
3305 * Returns 0 on success, or a negative error code.
3306 *
3307 * On success, if we set *skip = 1:
3308 *  - the next message should be skipped and ignored.
3309 *  - con->in_msg == NULL
3310 * or if we set *skip = 0:
3311 *  - con->in_msg is non-null.
3312 * On error (ENOMEM, EAGAIN, ...),
3313 *  - con->in_msg == NULL
3314 */
3315static int ceph_con_in_msg_alloc(struct ceph_connection *con, int *skip)
3316{
3317	struct ceph_msg_header *hdr = &con->in_hdr;
3318	int middle_len = le32_to_cpu(hdr->middle_len);
3319	struct ceph_msg *msg;
3320	int ret = 0;
3321
3322	BUG_ON(con->in_msg != NULL);
3323	BUG_ON(!con->ops->alloc_msg);
3324
3325	mutex_unlock(&con->mutex);
3326	msg = con->ops->alloc_msg(con, hdr, skip);
3327	mutex_lock(&con->mutex);
3328	if (con->state != CON_STATE_OPEN) {
3329		if (msg)
3330			ceph_msg_put(msg);
3331		return -EAGAIN;
3332	}
3333	if (msg) {
3334		BUG_ON(*skip);
3335		con->in_msg = msg;
3336		con->in_msg->con = con->ops->get(con);
3337		BUG_ON(con->in_msg->con == NULL);
3338	} else {
3339		/*
3340		 * Null message pointer means either we should skip
3341		 * this message or we couldn't allocate memory.  The
3342		 * former is not an error.
3343		 */
3344		if (*skip)
3345			return 0;
3346
3347		con->error_msg = "error allocating memory for incoming message";
3348		return -ENOMEM;
3349	}
3350	memcpy(&con->in_msg->hdr, &con->in_hdr, sizeof(con->in_hdr));
3351
3352	if (middle_len && !con->in_msg->middle) {
3353		ret = ceph_alloc_middle(con, con->in_msg);
3354		if (ret < 0) {
3355			ceph_msg_put(con->in_msg);
3356			con->in_msg = NULL;
3357		}
3358	}
3359
3360	return ret;
3361}
3362
3363
3364/*
3365 * Free a generically kmalloc'd message.
3366 */
3367static void ceph_msg_free(struct ceph_msg *m)
3368{
3369	dout("%s %p\n", __func__, m);
3370	kvfree(m->front.iov_base);
3371	kmem_cache_free(ceph_msg_cache, m);
3372}
3373
3374static void ceph_msg_release(struct kref *kref)
3375{
3376	struct ceph_msg *m = container_of(kref, struct ceph_msg, kref);
3377	LIST_HEAD(data);
3378	struct list_head *links;
3379	struct list_head *next;
3380
3381	dout("%s %p\n", __func__, m);
3382	WARN_ON(!list_empty(&m->list_head));
3383
3384	/* drop middle, data, if any */
3385	if (m->middle) {
3386		ceph_buffer_put(m->middle);
3387		m->middle = NULL;
3388	}
3389
3390	list_splice_init(&m->data, &data);
3391	list_for_each_safe(links, next, &data) {
3392		struct ceph_msg_data *data;
3393
3394		data = list_entry(links, struct ceph_msg_data, links);
3395		list_del_init(links);
3396		ceph_msg_data_destroy(data);
3397	}
3398	m->data_length = 0;
3399
3400	if (m->pool)
3401		ceph_msgpool_put(m->pool, m);
3402	else
3403		ceph_msg_free(m);
3404}
3405
3406struct ceph_msg *ceph_msg_get(struct ceph_msg *msg)
3407{
3408	dout("%s %p (was %d)\n", __func__, msg,
3409	     atomic_read(&msg->kref.refcount));
3410	kref_get(&msg->kref);
3411	return msg;
3412}
3413EXPORT_SYMBOL(ceph_msg_get);
3414
3415void ceph_msg_put(struct ceph_msg *msg)
3416{
3417	dout("%s %p (was %d)\n", __func__, msg,
3418	     atomic_read(&msg->kref.refcount));
3419	kref_put(&msg->kref, ceph_msg_release);
3420}
3421EXPORT_SYMBOL(ceph_msg_put);
3422
3423void ceph_msg_dump(struct ceph_msg *msg)
3424{
3425	pr_debug("msg_dump %p (front_alloc_len %d length %zd)\n", msg,
3426		 msg->front_alloc_len, msg->data_length);
3427	print_hex_dump(KERN_DEBUG, "header: ",
3428		       DUMP_PREFIX_OFFSET, 16, 1,
3429		       &msg->hdr, sizeof(msg->hdr), true);
3430	print_hex_dump(KERN_DEBUG, " front: ",
3431		       DUMP_PREFIX_OFFSET, 16, 1,
3432		       msg->front.iov_base, msg->front.iov_len, true);
3433	if (msg->middle)
3434		print_hex_dump(KERN_DEBUG, "middle: ",
3435			       DUMP_PREFIX_OFFSET, 16, 1,
3436			       msg->middle->vec.iov_base,
3437			       msg->middle->vec.iov_len, true);
3438	print_hex_dump(KERN_DEBUG, "footer: ",
3439		       DUMP_PREFIX_OFFSET, 16, 1,
3440		       &msg->footer, sizeof(msg->footer), true);
3441}
3442EXPORT_SYMBOL(ceph_msg_dump);
3443