net/ceph/messenger.c

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