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 <linux/dns_resolver.h>
  15 #include <net/tcp.h>
  16
  17 #include <linux/ceph/libceph.h>
  18 #include <linux/ceph/messenger.h>
  19 #include <linux/ceph/decode.h>
  20 #include <linux/ceph/pagelist.h>
  21 #include <linux/export.h>
  22
  23 /*
  24  * Ceph uses the messenger to exchange ceph_msg messages with other
  25  * hosts in the system.  The messenger provides ordered and reliable
  26  * delivery.  We tolerate TCP disconnects by reconnecting (with
  27  * exponential backoff) in the case of a fault (disconnection, bad
  28  * crc, protocol error).  Acks allow sent messages to be discarded by
  29  * the sender.
  30  */
  31
  32 /* static tag bytes (protocol control messages) */
  33 static char tag_msg = CEPH_MSGR_TAG_MSG;
  34 static char tag_ack = CEPH_MSGR_TAG_ACK;
  35 static char tag_keepalive = CEPH_MSGR_TAG_KEEPALIVE;
  36
  37 #ifdef CONFIG_LOCKDEP
  38 static struct lock_class_key socket_class;
  39 #endif
  40
  41 /*
  42  * When skipping (ignoring) a block of input we read it into a "skip
  43  * buffer," which is this many bytes in size.
  44  */
  45 #define SKIP_BUF_SIZE   1024
  46
  47 static void queue_con(struct ceph_connection *con);
  48 static void con_work(struct work_struct *);
  49 static void ceph_fault(struct ceph_connection *con);
  50
  51 /*
  52  * Nicely render a sockaddr as a string.  An array of formatted
  53  * strings is used, to approximate reentrancy.
  54  */
  55 #define ADDR_STR_COUNT_LOG      5       /* log2(# address strings in array) */
  56 #define ADDR_STR_COUNT          (1 << ADDR_STR_COUNT_LOG)
  57 #define ADDR_STR_COUNT_MASK     (ADDR_STR_COUNT - 1)
  58 #define MAX_ADDR_STR_LEN        64      /* 54 is enough */
  59
  60 static char addr_str[ADDR_STR_COUNT][MAX_ADDR_STR_LEN];
  61 static atomic_t addr_str_seq = ATOMIC_INIT(0);
  62
  63 static struct page *zero_page;          /* used in certain error cases */
  64
  65 const char *ceph_pr_addr(const struct sockaddr_storage *ss)
  66 {
  67         int i;
  68         char *s;
  69         struct sockaddr_in *in4 = (struct sockaddr_in *) ss;
  70         struct sockaddr_in6 *in6 = (struct sockaddr_in6 *) ss;
  71
  72         i = atomic_inc_return(&addr_str_seq) & ADDR_STR_COUNT_MASK;
  73         s = addr_str[i];
  74
  75         switch (ss->ss_family) {
  76         case AF_INET:
  77                 snprintf(s, MAX_ADDR_STR_LEN, "%pI4:%hu", &in4->sin_addr,
  78                          ntohs(in4->sin_port));
  79                 break;
  80
  81         case AF_INET6:
  82                 snprintf(s, MAX_ADDR_STR_LEN, "[%pI6c]:%hu", &in6->sin6_addr,
  83                          ntohs(in6->sin6_port));
  84                 break;
  85
  86         default:
  87                 snprintf(s, MAX_ADDR_STR_LEN, "(unknown sockaddr family %hu)",
  88                          ss->ss_family);
  89         }
  90
  91         return s;
  92 }
  93 EXPORT_SYMBOL(ceph_pr_addr);
  94
  95 static void encode_my_addr(struct ceph_messenger *msgr)
  96 {
  97         memcpy(&msgr->my_enc_addr, &msgr->inst.addr, sizeof(msgr->my_enc_addr));
  98         ceph_encode_addr(&msgr->my_enc_addr);
  99 }
 100
 101 /*
 102  * work queue for all reading and writing to/from the socket.
 103  */
 104 static struct workqueue_struct *ceph_msgr_wq;
 105
 106 void _ceph_msgr_exit(void)
 107 {
 108         if (ceph_msgr_wq) {
 109                 destroy_workqueue(ceph_msgr_wq);
 110                 ceph_msgr_wq = NULL;
 111         }
 112
 113         BUG_ON(zero_page == NULL);
 114         kunmap(zero_page);
 115         page_cache_release(zero_page);
 116         zero_page = NULL;
 117 }
 118
 119 int ceph_msgr_init(void)
 120 {
 121         BUG_ON(zero_page != NULL);
 122         zero_page = ZERO_PAGE(0);
 123         page_cache_get(zero_page);
 124
 125         ceph_msgr_wq = alloc_workqueue("ceph-msgr", WQ_NON_REENTRANT, 0);
 126         if (ceph_msgr_wq)
 127                 return 0;
 128
 129         pr_err("msgr_init failed to create workqueue\n");
 130         _ceph_msgr_exit();
 131
 132         return -ENOMEM;
 133 }
 134 EXPORT_SYMBOL(ceph_msgr_init);
 135
 136 void ceph_msgr_exit(void)
 137 {
 138         BUG_ON(ceph_msgr_wq == NULL);
 139
 140         _ceph_msgr_exit();
 141 }
 142 EXPORT_SYMBOL(ceph_msgr_exit);
 143
 144 void ceph_msgr_flush(void)
 145 {
 146         flush_workqueue(ceph_msgr_wq);
 147 }
 148 EXPORT_SYMBOL(ceph_msgr_flush);
 149
 150
 151 /*
 152  * socket callback functions
 153  */
 154
 155 /* data available on socket, or listen socket received a connect */
 156 static void ceph_data_ready(struct sock *sk, int count_unused)
 157 {
 158         struct ceph_connection *con = sk->sk_user_data;
 159
 160         if (sk->sk_state != TCP_CLOSE_WAIT) {
 161                 dout("ceph_data_ready on %p state = %lu, queueing work\n",
 162                      con, con->state);
 163                 queue_con(con);
 164         }
 165 }
 166
 167 /* socket has buffer space for writing */
 168 static void ceph_write_space(struct sock *sk)
 169 {
 170         struct ceph_connection *con = sk->sk_user_data;
 171
 172         /* only queue to workqueue if there is data we want to write,
 173          * and there is sufficient space in the socket buffer to accept
 174          * more data.  clear SOCK_NOSPACE so that ceph_write_space()
 175          * doesn't get called again until try_write() fills the socket
 176          * buffer. See net/ipv4/tcp_input.c:tcp_check_space()
 177          * and net/core/stream.c:sk_stream_write_space().
 178          */
 179         if (test_bit(WRITE_PENDING, &con->state)) {
 180                 if (sk_stream_wspace(sk) >= sk_stream_min_wspace(sk)) {
 181                         dout("ceph_write_space %p queueing write work\n", con);
 182                         clear_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
 183                         queue_con(con);
 184                 }
 185         } else {
 186                 dout("ceph_write_space %p nothing to write\n", con);
 187         }
 188 }
 189
 190 /* socket's state has changed */
 191 static void ceph_state_change(struct sock *sk)
 192 {
 193         struct ceph_connection *con = sk->sk_user_data;
 194
 195         dout("ceph_state_change %p state = %lu sk_state = %u\n",
 196              con, con->state, sk->sk_state);
 197
 198         if (test_bit(CLOSED, &con->state))
 199                 return;
 200
 201         switch (sk->sk_state) {
 202         case TCP_CLOSE:
 203                 dout("ceph_state_change TCP_CLOSE\n");
 204         case TCP_CLOSE_WAIT:
 205                 dout("ceph_state_change TCP_CLOSE_WAIT\n");
 206                 if (test_and_set_bit(SOCK_CLOSED, &con->state) == 0) {
 207                         if (test_bit(CONNECTING, &con->state))
 208                                 con->error_msg = "connection failed";
 209                         else
 210                                 con->error_msg = "socket closed";
 211                         queue_con(con);
 212                 }
 213                 break;
 214         case TCP_ESTABLISHED:
 215                 dout("ceph_state_change TCP_ESTABLISHED\n");
 216                 queue_con(con);
 217                 break;
 218         default:        /* Everything else is uninteresting */
 219                 break;
 220         }
 221 }
 222
 223 /*
 224  * set up socket callbacks
 225  */
 226 static void set_sock_callbacks(struct socket *sock,
 227                                struct ceph_connection *con)
 228 {
 229         struct sock *sk = sock->sk;
 230         sk->sk_user_data = con;
 231         sk->sk_data_ready = ceph_data_ready;
 232         sk->sk_write_space = ceph_write_space;
 233         sk->sk_state_change = ceph_state_change;
 234 }
 235
 236
 237 /*
 238  * socket helpers
 239  */
 240
 241 /*
 242  * initiate connection to a remote socket.
 243  */
 244 static int ceph_tcp_connect(struct ceph_connection *con)
 245 {
 246         struct sockaddr_storage *paddr = &con->peer_addr.in_addr;
 247         struct socket *sock;
 248         int ret;
 249
 250         BUG_ON(con->sock);
 251         ret = sock_create_kern(con->peer_addr.in_addr.ss_family, SOCK_STREAM,
 252                                IPPROTO_TCP, &sock);
 253         if (ret)
 254                 return ret;
 255         sock->sk->sk_allocation = GFP_NOFS;
 256
 257 #ifdef CONFIG_LOCKDEP
 258         lockdep_set_class(&sock->sk->sk_lock, &socket_class);
 259 #endif
 260
 261         set_sock_callbacks(sock, con);
 262
 263         dout("connect %s\n", ceph_pr_addr(&con->peer_addr.in_addr));
 264
 265         ret = sock->ops->connect(sock, (struct sockaddr *)paddr, sizeof(*paddr),
 266                                  O_NONBLOCK);
 267         if (ret == -EINPROGRESS) {
 268                 dout("connect %s EINPROGRESS sk_state = %u\n",
 269                      ceph_pr_addr(&con->peer_addr.in_addr),
 270                      sock->sk->sk_state);
 271         } else if (ret < 0) {
 272                 pr_err("connect %s error %d\n",
 273                        ceph_pr_addr(&con->peer_addr.in_addr), ret);
 274                 sock_release(sock);
 275                 con->error_msg = "connect error";
 276
 277                 return ret;
 278         }
 279         con->sock = sock;
 280
 281         return 0;
 282 }
 283
 284 static int ceph_tcp_recvmsg(struct socket *sock, void *buf, size_t len)
 285 {
 286         struct kvec iov = {buf, len};
 287         struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
 288         int r;
 289
 290         r = kernel_recvmsg(sock, &msg, &iov, 1, len, msg.msg_flags);
 291         if (r == -EAGAIN)
 292                 r = 0;
 293         return r;
 294 }
 295
 296 /*
 297  * write something.  @more is true if caller will be sending more data
 298  * shortly.
 299  */
 300 static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov,
 301                      size_t kvlen, size_t len, int more)
 302 {
 303         struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
 304         int r;
 305
 306         if (more)
 307                 msg.msg_flags |= MSG_MORE;
 308         else
 309                 msg.msg_flags |= MSG_EOR;  /* superfluous, but what the hell */
 310
 311         r = kernel_sendmsg(sock, &msg, iov, kvlen, len);
 312         if (r == -EAGAIN)
 313                 r = 0;
 314         return r;
 315 }
 316
 317 static int ceph_tcp_sendpage(struct socket *sock, struct page *page,
 318                      int offset, size_t size, int more)
 319 {
 320         int flags = MSG_DONTWAIT | MSG_NOSIGNAL | (more ? MSG_MORE : MSG_EOR);
 321         int ret;
 322
 323         ret = kernel_sendpage(sock, page, offset, size, flags);
 324         if (ret == -EAGAIN)
 325                 ret = 0;
 326
 327         return ret;
 328 }
 329
 330
 331 /*
 332  * Shutdown/close the socket for the given connection.
 333  */
 334 static int con_close_socket(struct ceph_connection *con)
 335 {
 336         int rc;
 337
 338         dout("con_close_socket on %p sock %p\n", con, con->sock);
 339         if (!con->sock)
 340                 return 0;
 341         set_bit(SOCK_CLOSED, &con->state);
 342         rc = con->sock->ops->shutdown(con->sock, SHUT_RDWR);
 343         sock_release(con->sock);
 344         con->sock = NULL;
 345         clear_bit(SOCK_CLOSED, &con->state);
 346         return rc;
 347 }
 348
 349 /*
 350  * Reset a connection.  Discard all incoming and outgoing messages
 351  * and clear *_seq state.
 352  */
 353 static void ceph_msg_remove(struct ceph_msg *msg)
 354 {
 355         list_del_init(&msg->list_head);
 356         ceph_msg_put(msg);
 357 }
 358 static void ceph_msg_remove_list(struct list_head *head)
 359 {
 360         while (!list_empty(head)) {
 361                 struct ceph_msg *msg = list_first_entry(head, struct ceph_msg,
 362                                                         list_head);
 363                 ceph_msg_remove(msg);
 364         }
 365 }
 366
 367 static void reset_connection(struct ceph_connection *con)
 368 {
 369         /* reset connection, out_queue, msg_ and connect_seq */
 370         /* discard existing out_queue and msg_seq */
 371         ceph_msg_remove_list(&con->out_queue);
 372         ceph_msg_remove_list(&con->out_sent);
 373
 374         if (con->in_msg) {
 375                 ceph_msg_put(con->in_msg);
 376                 con->in_msg = NULL;
 377         }
 378
 379         con->connect_seq = 0;
 380         con->out_seq = 0;
 381         if (con->out_msg) {
 382                 ceph_msg_put(con->out_msg);
 383                 con->out_msg = NULL;
 384         }
 385         con->in_seq = 0;
 386         con->in_seq_acked = 0;
 387 }
 388
 389 /*
 390  * mark a peer down.  drop any open connections.
 391  */
 392 void ceph_con_close(struct ceph_connection *con)
 393 {
 394         dout("con_close %p peer %s\n", con,
 395              ceph_pr_addr(&con->peer_addr.in_addr));
 396         set_bit(CLOSED, &con->state);  /* in case there's queued work */
 397         clear_bit(STANDBY, &con->state);  /* avoid connect_seq bump */
 398         clear_bit(LOSSYTX, &con->state);  /* so we retry next connect */
 399         clear_bit(KEEPALIVE_PENDING, &con->state);
 400         clear_bit(WRITE_PENDING, &con->state);
 401         mutex_lock(&con->mutex);
 402         reset_connection(con);
 403         con->peer_global_seq = 0;
 404         cancel_delayed_work(&con->work);
 405         mutex_unlock(&con->mutex);
 406         queue_con(con);
 407 }
 408 EXPORT_SYMBOL(ceph_con_close);
 409
 410 /*
 411  * Reopen a closed connection, with a new peer address.
 412  */
 413 void ceph_con_open(struct ceph_connection *con, struct ceph_entity_addr *addr)
 414 {
 415         dout("con_open %p %s\n", con, ceph_pr_addr(&addr->in_addr));
 416         set_bit(OPENING, &con->state);
 417         clear_bit(CLOSED, &con->state);
 418         memcpy(&con->peer_addr, addr, sizeof(*addr));
 419         con->delay = 0;      /* reset backoff memory */
 420         queue_con(con);
 421 }
 422 EXPORT_SYMBOL(ceph_con_open);
 423
 424 /*
 425  * return true if this connection ever successfully opened
 426  */
 427 bool ceph_con_opened(struct ceph_connection *con)
 428 {
 429         return con->connect_seq > 0;
 430 }
 431
 432 /*
 433  * generic get/put
 434  */
 435 struct ceph_connection *ceph_con_get(struct ceph_connection *con)
 436 {
 437         int nref = __atomic_add_unless(&con->nref, 1, 0);
 438
 439         dout("con_get %p nref = %d -> %d\n", con, nref, nref + 1);
 440
 441         return nref ? con : NULL;
 442 }
 443
 444 void ceph_con_put(struct ceph_connection *con)
 445 {
 446         int nref = atomic_dec_return(&con->nref);
 447
 448         BUG_ON(nref < 0);
 449         if (nref == 0) {
 450                 BUG_ON(con->sock);
 451                 kfree(con);
 452         }
 453         dout("con_put %p nref = %d -> %d\n", con, nref + 1, nref);
 454 }
 455
 456 /*
 457  * initialize a new connection.
 458  */
 459 void ceph_con_init(struct ceph_messenger *msgr, struct ceph_connection *con)
 460 {
 461         dout("con_init %p\n", con);
 462         memset(con, 0, sizeof(*con));
 463         atomic_set(&con->nref, 1);
 464         con->msgr = msgr;
 465         mutex_init(&con->mutex);
 466         INIT_LIST_HEAD(&con->out_queue);
 467         INIT_LIST_HEAD(&con->out_sent);
 468         INIT_DELAYED_WORK(&con->work, con_work);
 469 }
 470 EXPORT_SYMBOL(ceph_con_init);
 471
 472
 473 /*
 474  * We maintain a global counter to order connection attempts.  Get
 475  * a unique seq greater than @gt.
 476  */
 477 static u32 get_global_seq(struct ceph_messenger *msgr, u32 gt)
 478 {
 479         u32 ret;
 480
 481         spin_lock(&msgr->global_seq_lock);
 482         if (msgr->global_seq < gt)
 483                 msgr->global_seq = gt;
 484         ret = ++msgr->global_seq;
 485         spin_unlock(&msgr->global_seq_lock);
 486         return ret;
 487 }
 488
 489 static void ceph_con_out_kvec_reset(struct ceph_connection *con)
 490 {
 491         con->out_kvec_left = 0;
 492         con->out_kvec_bytes = 0;
 493         con->out_kvec_cur = &con->out_kvec[0];
 494 }
 495
 496 static void ceph_con_out_kvec_add(struct ceph_connection *con,
 497                                 size_t size, void *data)
 498 {
 499         int index;
 500
 501         index = con->out_kvec_left;
 502         BUG_ON(index >= ARRAY_SIZE(con->out_kvec));
 503
 504         con->out_kvec[index].iov_len = size;
 505         con->out_kvec[index].iov_base = data;
 506         con->out_kvec_left++;
 507         con->out_kvec_bytes += size;
 508 }
 509
 510 /*
 511  * Prepare footer for currently outgoing message, and finish things
 512  * off.  Assumes out_kvec* are already valid.. we just add on to the end.
 513  */
 514 static void prepare_write_message_footer(struct ceph_connection *con)
 515 {
 516         struct ceph_msg *m = con->out_msg;
 517         int v = con->out_kvec_left;
 518
 519         dout("prepare_write_message_footer %p\n", con);
 520         con->out_kvec_is_msg = true;
 521         con->out_kvec[v].iov_base = &m->footer;
 522         con->out_kvec[v].iov_len = sizeof(m->footer);
 523         con->out_kvec_bytes += sizeof(m->footer);
 524         con->out_kvec_left++;
 525         con->out_more = m->more_to_follow;
 526         con->out_msg_done = true;
 527 }
 528
 529 /*
 530  * Prepare headers for the next outgoing message.
 531  */
 532 static void prepare_write_message(struct ceph_connection *con)
 533 {
 534         struct ceph_msg *m;
 535         u32 crc;
 536
 537         ceph_con_out_kvec_reset(con);
 538         con->out_kvec_is_msg = true;
 539         con->out_msg_done = false;
 540
 541         /* Sneak an ack in there first?  If we can get it into the same
 542          * TCP packet that's a good thing. */
 543         if (con->in_seq > con->in_seq_acked) {
 544                 con->in_seq_acked = con->in_seq;
 545                 ceph_con_out_kvec_add(con, sizeof (tag_ack), &tag_ack);
 546                 con->out_temp_ack = cpu_to_le64(con->in_seq_acked);
 547                 ceph_con_out_kvec_add(con, sizeof (con->out_temp_ack),
 548                         &con->out_temp_ack);
 549         }
 550
 551         m = list_first_entry(&con->out_queue, struct ceph_msg, list_head);
 552         con->out_msg = m;
 553
 554         /* put message on sent list */
 555         ceph_msg_get(m);
 556         list_move_tail(&m->list_head, &con->out_sent);
 557
 558         /*
 559          * only assign outgoing seq # if we haven't sent this message
 560          * yet.  if it is requeued, resend with it's original seq.
 561          */
 562         if (m->needs_out_seq) {
 563                 m->hdr.seq = cpu_to_le64(++con->out_seq);
 564                 m->needs_out_seq = false;
 565         }
 566
 567         dout("prepare_write_message %p seq %lld type %d len %d+%d+%d %d pgs\n",
 568              m, con->out_seq, le16_to_cpu(m->hdr.type),
 569              le32_to_cpu(m->hdr.front_len), le32_to_cpu(m->hdr.middle_len),
 570              le32_to_cpu(m->hdr.data_len),
 571              m->nr_pages);
 572         BUG_ON(le32_to_cpu(m->hdr.front_len) != m->front.iov_len);
 573
 574         /* tag + hdr + front + middle */
 575         ceph_con_out_kvec_add(con, sizeof (tag_msg), &tag_msg);
 576         ceph_con_out_kvec_add(con, sizeof (m->hdr), &m->hdr);
 577         ceph_con_out_kvec_add(con, m->front.iov_len, m->front.iov_base);
 578
 579         if (m->middle)
 580                 ceph_con_out_kvec_add(con, m->middle->vec.iov_len,
 581                         m->middle->vec.iov_base);
 582
 583         /* fill in crc (except data pages), footer */
 584         crc = crc32c(0, &m->hdr, offsetof(struct ceph_msg_header, crc));
 585         con->out_msg->hdr.crc = cpu_to_le32(crc);
 586         con->out_msg->footer.flags = CEPH_MSG_FOOTER_COMPLETE;
 587
 588         crc = crc32c(0, m->front.iov_base, m->front.iov_len);
 589         con->out_msg->footer.front_crc = cpu_to_le32(crc);
 590         if (m->middle) {
 591                 crc = crc32c(0, m->middle->vec.iov_base,
 592                                 m->middle->vec.iov_len);
 593                 con->out_msg->footer.middle_crc = cpu_to_le32(crc);
 594         } else
 595                 con->out_msg->footer.middle_crc = 0;
 596         con->out_msg->footer.data_crc = 0;
 597         dout("prepare_write_message front_crc %u data_crc %u\n",
 598              le32_to_cpu(con->out_msg->footer.front_crc),
 599              le32_to_cpu(con->out_msg->footer.middle_crc));
 600
 601         /* is there a data payload? */
 602         if (le32_to_cpu(m->hdr.data_len) > 0) {
 603                 /* initialize page iterator */
 604                 con->out_msg_pos.page = 0;
 605                 if (m->pages)
 606                         con->out_msg_pos.page_pos = m->page_alignment;
 607                 else
 608                         con->out_msg_pos.page_pos = 0;
 609                 con->out_msg_pos.data_pos = 0;
 610                 con->out_msg_pos.did_page_crc = false;
 611                 con->out_more = 1;  /* data + footer will follow */
 612         } else {
 613                 /* no, queue up footer too and be done */
 614                 prepare_write_message_footer(con);
 615         }
 616
 617         set_bit(WRITE_PENDING, &con->state);
 618 }
 619
 620 /*
 621  * Prepare an ack.
 622  */
 623 static void prepare_write_ack(struct ceph_connection *con)
 624 {
 625         dout("prepare_write_ack %p %llu -> %llu\n", con,
 626              con->in_seq_acked, con->in_seq);
 627         con->in_seq_acked = con->in_seq;
 628
 629         ceph_con_out_kvec_reset(con);
 630
 631         ceph_con_out_kvec_add(con, sizeof (tag_ack), &tag_ack);
 632
 633         con->out_temp_ack = cpu_to_le64(con->in_seq_acked);
 634         ceph_con_out_kvec_add(con, sizeof (con->out_temp_ack),
 635                                 &con->out_temp_ack);
 636
 637         con->out_more = 1;  /* more will follow.. eventually.. */
 638         set_bit(WRITE_PENDING, &con->state);
 639 }
 640
 641 /*
 642  * Prepare to write keepalive byte.
 643  */
 644 static void prepare_write_keepalive(struct ceph_connection *con)
 645 {
 646         dout("prepare_write_keepalive %p\n", con);
 647         ceph_con_out_kvec_reset(con);
 648         ceph_con_out_kvec_add(con, sizeof (tag_keepalive), &tag_keepalive);
 649         set_bit(WRITE_PENDING, &con->state);
 650 }
 651
 652 /*
 653  * Connection negotiation.
 654  */
 655
 656 static struct ceph_auth_handshake *get_connect_authorizer(struct ceph_connection *con,
 657                                                 int *auth_proto)
 658 {
 659         struct ceph_auth_handshake *auth;
 660
 661         if (!con->ops->get_authorizer) {
 662                 con->out_connect.authorizer_protocol = CEPH_AUTH_UNKNOWN;
 663                 con->out_connect.authorizer_len = 0;
 664
 665                 return NULL;
 666         }
 667
 668         /* Can't hold the mutex while getting authorizer */
 669
 670         mutex_unlock(&con->mutex);
 671
 672         auth = con->ops->get_authorizer(con, auth_proto, con->auth_retry);
 673
 674         mutex_lock(&con->mutex);
 675
 676         if (IS_ERR(auth))
 677                 return auth;
 678         if (test_bit(CLOSED, &con->state) || test_bit(OPENING, &con->state))
 679                 return ERR_PTR(-EAGAIN);
 680
 681         con->auth_reply_buf = auth->authorizer_reply_buf;
 682         con->auth_reply_buf_len = auth->authorizer_reply_buf_len;
 683
 684
 685         return auth;
 686 }
 687
 688 /*
 689  * We connected to a peer and are saying hello.
 690  */
 691 static void prepare_write_banner(struct ceph_connection *con)
 692 {
 693         ceph_con_out_kvec_add(con, strlen(CEPH_BANNER), CEPH_BANNER);
 694         ceph_con_out_kvec_add(con, sizeof (con->msgr->my_enc_addr),
 695                                         &con->msgr->my_enc_addr);
 696
 697         con->out_more = 0;
 698         set_bit(WRITE_PENDING, &con->state);
 699 }
 700
 701 static int prepare_write_connect(struct ceph_connection *con)
 702 {
 703         unsigned int global_seq = get_global_seq(con->msgr, 0);
 704         int proto;
 705         int auth_proto;
 706         struct ceph_auth_handshake *auth;
 707
 708         switch (con->peer_name.type) {
 709         case CEPH_ENTITY_TYPE_MON:
 710                 proto = CEPH_MONC_PROTOCOL;
 711                 break;
 712         case CEPH_ENTITY_TYPE_OSD:
 713                 proto = CEPH_OSDC_PROTOCOL;
 714                 break;
 715         case CEPH_ENTITY_TYPE_MDS:
 716                 proto = CEPH_MDSC_PROTOCOL;
 717                 break;
 718         default:
 719                 BUG();
 720         }
 721
 722         dout("prepare_write_connect %p cseq=%d gseq=%d proto=%d\n", con,
 723              con->connect_seq, global_seq, proto);
 724
 725         con->out_connect.features = cpu_to_le64(con->msgr->supported_features);
 726         con->out_connect.host_type = cpu_to_le32(CEPH_ENTITY_TYPE_CLIENT);
 727         con->out_connect.connect_seq = cpu_to_le32(con->connect_seq);
 728         con->out_connect.global_seq = cpu_to_le32(global_seq);
 729         con->out_connect.protocol_version = cpu_to_le32(proto);
 730         con->out_connect.flags = 0;
 731
 732         auth_proto = CEPH_AUTH_UNKNOWN;
 733         auth = get_connect_authorizer(con, &auth_proto);
 734         if (IS_ERR(auth))
 735                 return PTR_ERR(auth);
 736
 737         con->out_connect.authorizer_protocol = cpu_to_le32(auth_proto);
 738         con->out_connect.authorizer_len = auth ?
 739                 cpu_to_le32(auth->authorizer_buf_len) : 0;
 740
 741         ceph_con_out_kvec_add(con, sizeof (con->out_connect),
 742                                         &con->out_connect);
 743         if (auth && auth->authorizer_buf_len)
 744                 ceph_con_out_kvec_add(con, auth->authorizer_buf_len,
 745                                         auth->authorizer_buf);
 746
 747         con->out_more = 0;
 748         set_bit(WRITE_PENDING, &con->state);
 749
 750         return 0;
 751 }
 752
 753 /*
 754  * write as much of pending kvecs to the socket as we can.
 755  *  1 -> done
 756  *  0 -> socket full, but more to do
 757  * <0 -> error
 758  */
 759 static int write_partial_kvec(struct ceph_connection *con)
 760 {
 761         int ret;
 762
 763         dout("write_partial_kvec %p %d left\n", con, con->out_kvec_bytes);
 764         while (con->out_kvec_bytes > 0) {
 765                 ret = ceph_tcp_sendmsg(con->sock, con->out_kvec_cur,
 766                                        con->out_kvec_left, con->out_kvec_bytes,
 767                                        con->out_more);
 768                 if (ret <= 0)
 769                         goto out;
 770                 con->out_kvec_bytes -= ret;
 771                 if (con->out_kvec_bytes == 0)
 772                         break;            /* done */
 773
 774                 /* account for full iov entries consumed */
 775                 while (ret >= con->out_kvec_cur->iov_len) {
 776                         BUG_ON(!con->out_kvec_left);
 777                         ret -= con->out_kvec_cur->iov_len;
 778                         con->out_kvec_cur++;
 779                         con->out_kvec_left--;
 780                 }
 781                 /* and for a partially-consumed entry */
 782                 if (ret) {
 783                         con->out_kvec_cur->iov_len -= ret;
 784                         con->out_kvec_cur->iov_base += ret;
 785                 }
 786         }
 787         con->out_kvec_left = 0;
 788         con->out_kvec_is_msg = false;
 789         ret = 1;
 790 out:
 791         dout("write_partial_kvec %p %d left in %d kvecs ret = %d\n", con,
 792              con->out_kvec_bytes, con->out_kvec_left, ret);
 793         return ret;  /* done! */
 794 }
 795
 796 #ifdef CONFIG_BLOCK
 797 static void init_bio_iter(struct bio *bio, struct bio **iter, int *seg)
 798 {
 799         if (!bio) {
 800                 *iter = NULL;
 801                 *seg = 0;
 802                 return;
 803         }
 804         *iter = bio;
 805         *seg = bio->bi_idx;
 806 }
 807
 808 static void iter_bio_next(struct bio **bio_iter, int *seg)
 809 {
 810         if (*bio_iter == NULL)
 811                 return;
 812
 813         BUG_ON(*seg >= (*bio_iter)->bi_vcnt);
 814
 815         (*seg)++;
 816         if (*seg == (*bio_iter)->bi_vcnt)
 817                 init_bio_iter((*bio_iter)->bi_next, bio_iter, seg);
 818 }
 819 #endif
 820
 821 /*
 822  * Write as much message data payload as we can.  If we finish, queue
 823  * up the footer.
 824  *  1 -> done, footer is now queued in out_kvec[].
 825  *  0 -> socket full, but more to do
 826  * <0 -> error
 827  */
 828 static int write_partial_msg_pages(struct ceph_connection *con)
 829 {
 830         struct ceph_msg *msg = con->out_msg;
 831         unsigned int data_len = le32_to_cpu(msg->hdr.data_len);
 832         size_t len;
 833         bool do_datacrc = !con->msgr->nocrc;
 834         int ret;
 835         int total_max_write;
 836         int in_trail = 0;
 837         size_t trail_len = (msg->trail ? msg->trail->length : 0);
 838
 839         dout("write_partial_msg_pages %p msg %p page %d/%d offset %d\n",
 840              con, con->out_msg, con->out_msg_pos.page, con->out_msg->nr_pages,
 841              con->out_msg_pos.page_pos);
 842
 843 #ifdef CONFIG_BLOCK
 844         if (msg->bio && !msg->bio_iter)
 845                 init_bio_iter(msg->bio, &msg->bio_iter, &msg->bio_seg);
 846 #endif
 847
 848         while (data_len > con->out_msg_pos.data_pos) {
 849                 struct page *page = NULL;
 850                 int max_write = PAGE_SIZE;
 851                 int bio_offset = 0;
 852
 853                 total_max_write = data_len - trail_len -
 854                         con->out_msg_pos.data_pos;
 855
 856                 /*
 857                  * if we are calculating the data crc (the default), we need
 858                  * to map the page.  if our pages[] has been revoked, use the
 859                  * zero page.
 860                  */
 861
 862                 /* have we reached the trail part of the data? */
 863                 if (con->out_msg_pos.data_pos >= data_len - trail_len) {
 864                         in_trail = 1;
 865
 866                         total_max_write = data_len - con->out_msg_pos.data_pos;
 867
 868                         page = list_first_entry(&msg->trail->head,
 869                                                 struct page, lru);
 870                         max_write = PAGE_SIZE;
 871                 } else if (msg->pages) {
 872                         page = msg->pages[con->out_msg_pos.page];
 873                 } else if (msg->pagelist) {
 874                         page = list_first_entry(&msg->pagelist->head,
 875                                                 struct page, lru);
 876 #ifdef CONFIG_BLOCK
 877                 } else if (msg->bio) {
 878                         struct bio_vec *bv;
 879
 880                         bv = bio_iovec_idx(msg->bio_iter, msg->bio_seg);
 881                         page = bv->bv_page;
 882                         bio_offset = bv->bv_offset;
 883                         max_write = bv->bv_len;
 884 #endif
 885                 } else {
 886                         page = zero_page;
 887                 }
 888                 len = min_t(int, max_write - con->out_msg_pos.page_pos,
 889                             total_max_write);
 890
 891                 if (do_datacrc && !con->out_msg_pos.did_page_crc) {
 892                         void *base;
 893                         u32 crc;
 894                         u32 tmpcrc = le32_to_cpu(con->out_msg->footer.data_crc);
 895                         char *kaddr;
 896
 897                         kaddr = kmap(page);
 898                         BUG_ON(kaddr == NULL);
 899                         base = kaddr + con->out_msg_pos.page_pos + bio_offset;
 900                         crc = crc32c(tmpcrc, base, len);
 901                         con->out_msg->footer.data_crc = cpu_to_le32(crc);
 902                         con->out_msg_pos.did_page_crc = true;
 903                 }
 904                 ret = ceph_tcp_sendpage(con->sock, page,
 905                                       con->out_msg_pos.page_pos + bio_offset,
 906                                       len, 1);
 907
 908                 if (do_datacrc)
 909                         kunmap(page);
 910
 911                 if (ret <= 0)
 912                         goto out;
 913
 914                 con->out_msg_pos.data_pos += ret;
 915                 con->out_msg_pos.page_pos += ret;
 916                 if (ret == len) {
 917                         con->out_msg_pos.page_pos = 0;
 918                         con->out_msg_pos.page++;
 919                         con->out_msg_pos.did_page_crc = false;
 920                         if (in_trail)
 921                                 list_move_tail(&page->lru,
 922                                                &msg->trail->head);
 923                         else if (msg->pagelist)
 924                                 list_move_tail(&page->lru,
 925                                                &msg->pagelist->head);
 926 #ifdef CONFIG_BLOCK
 927                         else if (msg->bio)
 928                                 iter_bio_next(&msg->bio_iter, &msg->bio_seg);
 929 #endif
 930                 }
 931         }
 932
 933         dout("write_partial_msg_pages %p msg %p done\n", con, msg);
 934
 935         /* prepare and queue up footer, too */
 936         if (!do_datacrc)
 937                 con->out_msg->footer.flags |= CEPH_MSG_FOOTER_NOCRC;
 938         ceph_con_out_kvec_reset(con);
 939         prepare_write_message_footer(con);
 940         ret = 1;
 941 out:
 942         return ret;
 943 }
 944
 945 /*
 946  * write some zeros
 947  */
 948 static int write_partial_skip(struct ceph_connection *con)
 949 {
 950         int ret;
 951
 952         while (con->out_skip > 0) {
 953                 size_t size = min(con->out_skip, (int) PAGE_CACHE_SIZE);
 954
 955                 ret = ceph_tcp_sendpage(con->sock, zero_page, 0, size, 1);
 956                 if (ret <= 0)
 957                         goto out;
 958                 con->out_skip -= ret;
 959         }
 960         ret = 1;
 961 out:
 962         return ret;
 963 }
 964
 965 /*
 966  * Prepare to read connection handshake, or an ack.
 967  */
 968 static void prepare_read_banner(struct ceph_connection *con)
 969 {
 970         dout("prepare_read_banner %p\n", con);
 971         con->in_base_pos = 0;
 972 }
 973
 974 static void prepare_read_connect(struct ceph_connection *con)
 975 {
 976         dout("prepare_read_connect %p\n", con);
 977         con->in_base_pos = 0;
 978 }
 979
 980 static void prepare_read_ack(struct ceph_connection *con)
 981 {
 982         dout("prepare_read_ack %p\n", con);
 983         con->in_base_pos = 0;
 984 }
 985
 986 static void prepare_read_tag(struct ceph_connection *con)
 987 {
 988         dout("prepare_read_tag %p\n", con);
 989         con->in_base_pos = 0;
 990         con->in_tag = CEPH_MSGR_TAG_READY;
 991 }
 992
 993 /*
 994  * Prepare to read a message.
 995  */
 996 static int prepare_read_message(struct ceph_connection *con)
 997 {
 998         dout("prepare_read_message %p\n", con);
 999         BUG_ON(con->in_msg != NULL);
1000         con->in_base_pos = 0;
1001         con->in_front_crc = con->in_middle_crc = con->in_data_crc = 0;
1002         return 0;
1003 }
1004
1005
1006 static int read_partial(struct ceph_connection *con,
1007                         int end, int size, void *object)
1008 {
1009         while (con->in_base_pos < end) {
1010                 int left = end - con->in_base_pos;
1011                 int have = size - left;
1012                 int ret = ceph_tcp_recvmsg(con->sock, object + have, left);
1013                 if (ret <= 0)
1014                         return ret;
1015                 con->in_base_pos += ret;
1016         }
1017         return 1;
1018 }
1019
1020
1021 /*
1022  * Read all or part of the connect-side handshake on a new connection
1023  */
1024 static int read_partial_banner(struct ceph_connection *con)
1025 {
1026         int size;
1027         int end;
1028         int ret;
1029
1030         dout("read_partial_banner %p at %d\n", con, con->in_base_pos);
1031
1032         /* peer's banner */
1033         size = strlen(CEPH_BANNER);
1034         end = size;
1035         ret = read_partial(con, end, size, con->in_banner);
1036         if (ret <= 0)
1037                 goto out;
1038
1039         size = sizeof (con->actual_peer_addr);
1040         end += size;
1041         ret = read_partial(con, end, size, &con->actual_peer_addr);
1042         if (ret <= 0)
1043                 goto out;
1044
1045         size = sizeof (con->peer_addr_for_me);
1046         end += size;
1047         ret = read_partial(con, end, size, &con->peer_addr_for_me);
1048         if (ret <= 0)
1049                 goto out;
1050
1051 out:
1052         return ret;
1053 }
1054
1055 static int read_partial_connect(struct ceph_connection *con)
1056 {
1057         int size;
1058         int end;
1059         int ret;
1060
1061         dout("read_partial_connect %p at %d\n", con, con->in_base_pos);
1062
1063         size = sizeof (con->in_reply);
1064         end = size;
1065         ret = read_partial(con, end, size, &con->in_reply);
1066         if (ret <= 0)
1067                 goto out;
1068
1069         size = le32_to_cpu(con->in_reply.authorizer_len);
1070         end += size;
1071         ret = read_partial(con, end, size, con->auth_reply_buf);
1072         if (ret <= 0)
1073                 goto out;
1074
1075         dout("read_partial_connect %p tag %d, con_seq = %u, g_seq = %u\n",
1076              con, (int)con->in_reply.tag,
1077              le32_to_cpu(con->in_reply.connect_seq),
1078              le32_to_cpu(con->in_reply.global_seq));
1079 out:
1080         return ret;
1081
1082 }
1083
1084 /*
1085  * Verify the hello banner looks okay.
1086  */
1087 static int verify_hello(struct ceph_connection *con)
1088 {
1089         if (memcmp(con->in_banner, CEPH_BANNER, strlen(CEPH_BANNER))) {
1090                 pr_err("connect to %s got bad banner\n",
1091                        ceph_pr_addr(&con->peer_addr.in_addr));
1092                 con->error_msg = "protocol error, bad banner";
1093                 return -1;
1094         }
1095         return 0;
1096 }
1097
1098 static bool addr_is_blank(struct sockaddr_storage *ss)
1099 {
1100         switch (ss->ss_family) {
1101         case AF_INET:
1102                 return ((struct sockaddr_in *)ss)->sin_addr.s_addr == 0;
1103         case AF_INET6:
1104                 return
1105                      ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[0] == 0 &&
1106                      ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[1] == 0 &&
1107                      ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[2] == 0 &&
1108                      ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[3] == 0;
1109         }
1110         return false;
1111 }
1112
1113 static int addr_port(struct sockaddr_storage *ss)
1114 {
1115         switch (ss->ss_family) {
1116         case AF_INET:
1117                 return ntohs(((struct sockaddr_in *)ss)->sin_port);
1118         case AF_INET6:
1119                 return ntohs(((struct sockaddr_in6 *)ss)->sin6_port);
1120         }
1121         return 0;
1122 }
1123
1124 static void addr_set_port(struct sockaddr_storage *ss, int p)
1125 {
1126         switch (ss->ss_family) {
1127         case AF_INET:
1128                 ((struct sockaddr_in *)ss)->sin_port = htons(p);
1129                 break;
1130         case AF_INET6:
1131                 ((struct sockaddr_in6 *)ss)->sin6_port = htons(p);
1132                 break;
1133         }
1134 }
1135
1136 /*
1137  * Unlike other *_pton function semantics, zero indicates success.
1138  */
1139 static int ceph_pton(const char *str, size_t len, struct sockaddr_storage *ss,
1140                 char delim, const char **ipend)
1141 {
1142         struct sockaddr_in *in4 = (struct sockaddr_in *) ss;
1143         struct sockaddr_in6 *in6 = (struct sockaddr_in6 *) ss;
1144
1145         memset(ss, 0, sizeof(*ss));
1146
1147         if (in4_pton(str, len, (u8 *)&in4->sin_addr.s_addr, delim, ipend)) {
1148                 ss->ss_family = AF_INET;
1149                 return 0;
1150         }
1151
1152         if (in6_pton(str, len, (u8 *)&in6->sin6_addr.s6_addr, delim, ipend)) {
1153                 ss->ss_family = AF_INET6;
1154                 return 0;
1155         }
1156
1157         return -EINVAL;
1158 }
1159
1160 /*
1161  * Extract hostname string and resolve using kernel DNS facility.
1162  */
1163 #ifdef CONFIG_CEPH_LIB_USE_DNS_RESOLVER
1164 static int ceph_dns_resolve_name(const char *name, size_t namelen,
1165                 struct sockaddr_storage *ss, char delim, const char **ipend)
1166 {
1167         const char *end, *delim_p;
1168         char *colon_p, *ip_addr = NULL;
1169         int ip_len, ret;
1170
1171         /*
1172          * The end of the hostname occurs immediately preceding the delimiter or
1173          * the port marker (':') where the delimiter takes precedence.
1174          */
1175         delim_p = memchr(name, delim, namelen);
1176         colon_p = memchr(name, ':', namelen);
1177
1178         if (delim_p && colon_p)
1179                 end = delim_p < colon_p ? delim_p : colon_p;
1180         else if (!delim_p && colon_p)
1181                 end = colon_p;
1182         else {
1183                 end = delim_p;
1184                 if (!end) /* case: hostname:/ */
1185                         end = name + namelen;
1186         }
1187
1188         if (end <= name)
1189                 return -EINVAL;
1190
1191         /* do dns_resolve upcall */
1192         ip_len = dns_query(NULL, name, end - name, NULL, &ip_addr, NULL);
1193         if (ip_len > 0)
1194                 ret = ceph_pton(ip_addr, ip_len, ss, -1, NULL);
1195         else
1196                 ret = -ESRCH;
1197
1198         kfree(ip_addr);
1199
1200         *ipend = end;
1201
1202         pr_info("resolve '%.*s' (ret=%d): %s\n", (int)(end - name), name,
1203                         ret, ret ? "failed" : ceph_pr_addr(ss));
1204
1205         return ret;
1206 }
1207 #else
1208 static inline int ceph_dns_resolve_name(const char *name, size_t namelen,
1209                 struct sockaddr_storage *ss, char delim, const char **ipend)
1210 {
1211         return -EINVAL;
1212 }
1213 #endif
1214
1215 /*
1216  * Parse a server name (IP or hostname). If a valid IP address is not found
1217  * then try to extract a hostname to resolve using userspace DNS upcall.
1218  */
1219 static int ceph_parse_server_name(const char *name, size_t namelen,
1220                         struct sockaddr_storage *ss, char delim, const char **ipend)
1221 {
1222         int ret;
1223
1224         ret = ceph_pton(name, namelen, ss, delim, ipend);
1225         if (ret)
1226                 ret = ceph_dns_resolve_name(name, namelen, ss, delim, ipend);
1227
1228         return ret;
1229 }
1230
1231 /*
1232  * Parse an ip[:port] list into an addr array.  Use the default
1233  * monitor port if a port isn't specified.
1234  */
1235 int ceph_parse_ips(const char *c, const char *end,
1236                    struct ceph_entity_addr *addr,
1237                    int max_count, int *count)
1238 {
1239         int i, ret = -EINVAL;
1240         const char *p = c;
1241
1242         dout("parse_ips on '%.*s'\n", (int)(end-c), c);
1243         for (i = 0; i < max_count; i++) {
1244                 const char *ipend;
1245                 struct sockaddr_storage *ss = &addr[i].in_addr;
1246                 int port;
1247                 char delim = ',';
1248
1249                 if (*p == '[') {
1250                         delim = ']';
1251                         p++;
1252                 }
1253
1254                 ret = ceph_parse_server_name(p, end - p, ss, delim, &ipend);
1255                 if (ret)
1256                         goto bad;
1257                 ret = -EINVAL;
1258
1259                 p = ipend;
1260
1261                 if (delim == ']') {
1262                         if (*p != ']') {
1263                                 dout("missing matching ']'\n");
1264                                 goto bad;
1265                         }
1266                         p++;
1267                 }
1268
1269                 /* port? */
1270                 if (p < end && *p == ':') {
1271                         port = 0;
1272                         p++;
1273                         while (p < end && *p >= '0' && *p <= '9') {
1274                                 port = (port * 10) + (*p - '0');
1275                                 p++;
1276                         }
1277                         if (port > 65535 || port == 0)
1278                                 goto bad;
1279                 } else {
1280                         port = CEPH_MON_PORT;
1281                 }
1282
1283                 addr_set_port(ss, port);
1284
1285                 dout("parse_ips got %s\n", ceph_pr_addr(ss));
1286
1287                 if (p == end)
1288                         break;
1289                 if (*p != ',')
1290                         goto bad;
1291                 p++;
1292         }
1293
1294         if (p != end)
1295                 goto bad;
1296
1297         if (count)
1298                 *count = i + 1;
1299         return 0;
1300
1301 bad:
1302         pr_err("parse_ips bad ip '%.*s'\n", (int)(end - c), c);
1303         return ret;
1304 }
1305 EXPORT_SYMBOL(ceph_parse_ips);
1306
1307 static int process_banner(struct ceph_connection *con)
1308 {
1309         dout("process_banner on %p\n", con);
1310
1311         if (verify_hello(con) < 0)
1312                 return -1;
1313
1314         ceph_decode_addr(&con->actual_peer_addr);
1315         ceph_decode_addr(&con->peer_addr_for_me);
1316
1317         /*
1318          * Make sure the other end is who we wanted.  note that the other
1319          * end may not yet know their ip address, so if it's 0.0.0.0, give
1320          * them the benefit of the doubt.
1321          */
1322         if (memcmp(&con->peer_addr, &con->actual_peer_addr,
1323                    sizeof(con->peer_addr)) != 0 &&
1324             !(addr_is_blank(&con->actual_peer_addr.in_addr) &&
1325               con->actual_peer_addr.nonce == con->peer_addr.nonce)) {
1326                 pr_warning("wrong peer, want %s/%d, got %s/%d\n",
1327                            ceph_pr_addr(&con->peer_addr.in_addr),
1328                            (int)le32_to_cpu(con->peer_addr.nonce),
1329                            ceph_pr_addr(&con->actual_peer_addr.in_addr),
1330                            (int)le32_to_cpu(con->actual_peer_addr.nonce));
1331                 con->error_msg = "wrong peer at address";
1332                 return -1;
1333         }
1334
1335         /*
1336          * did we learn our address?
1337          */
1338         if (addr_is_blank(&con->msgr->inst.addr.in_addr)) {
1339                 int port = addr_port(&con->msgr->inst.addr.in_addr);
1340
1341                 memcpy(&con->msgr->inst.addr.in_addr,
1342                        &con->peer_addr_for_me.in_addr,
1343                        sizeof(con->peer_addr_for_me.in_addr));
1344                 addr_set_port(&con->msgr->inst.addr.in_addr, port);
1345                 encode_my_addr(con->msgr);
1346                 dout("process_banner learned my addr is %s\n",
1347                      ceph_pr_addr(&con->msgr->inst.addr.in_addr));
1348         }
1349
1350         set_bit(NEGOTIATING, &con->state);
1351         prepare_read_connect(con);
1352         return 0;
1353 }
1354
1355 static void fail_protocol(struct ceph_connection *con)
1356 {
1357         reset_connection(con);
1358         set_bit(CLOSED, &con->state);  /* in case there's queued work */
1359
1360         mutex_unlock(&con->mutex);
1361         if (con->ops->bad_proto)
1362                 con->ops->bad_proto(con);
1363         mutex_lock(&con->mutex);
1364 }
1365
1366 static int process_connect(struct ceph_connection *con)
1367 {
1368         u64 sup_feat = con->msgr->supported_features;
1369         u64 req_feat = con->msgr->required_features;
1370         u64 server_feat = le64_to_cpu(con->in_reply.features);
1371         int ret;
1372
1373         dout("process_connect on %p tag %d\n", con, (int)con->in_tag);
1374
1375         switch (con->in_reply.tag) {
1376         case CEPH_MSGR_TAG_FEATURES:
1377                 pr_err("%s%lld %s feature set mismatch,"
1378                        " my %llx < server's %llx, missing %llx\n",
1379                        ENTITY_NAME(con->peer_name),
1380                        ceph_pr_addr(&con->peer_addr.in_addr),
1381                        sup_feat, server_feat, server_feat & ~sup_feat);
1382                 con->error_msg = "missing required protocol features";
1383                 fail_protocol(con);
1384                 return -1;
1385
1386         case CEPH_MSGR_TAG_BADPROTOVER:
1387                 pr_err("%s%lld %s protocol version mismatch,"
1388                        " my %d != server's %d\n",
1389                        ENTITY_NAME(con->peer_name),
1390                        ceph_pr_addr(&con->peer_addr.in_addr),
1391                        le32_to_cpu(con->out_connect.protocol_version),
1392                        le32_to_cpu(con->in_reply.protocol_version));
1393                 con->error_msg = "protocol version mismatch";
1394                 fail_protocol(con);
1395                 return -1;
1396
1397         case CEPH_MSGR_TAG_BADAUTHORIZER:
1398                 con->auth_retry++;
1399                 dout("process_connect %p got BADAUTHORIZER attempt %d\n", con,
1400                      con->auth_retry);
1401                 if (con->auth_retry == 2) {
1402                         con->error_msg = "connect authorization failure";
1403                         return -1;
1404                 }
1405                 con->auth_retry = 1;
1406                 ceph_con_out_kvec_reset(con);
1407                 ret = prepare_write_connect(con);
1408                 if (ret < 0)
1409                         return ret;
1410                 prepare_read_connect(con);
1411                 break;
1412
1413         case CEPH_MSGR_TAG_RESETSESSION:
1414                 /*
1415                  * If we connected with a large connect_seq but the peer
1416                  * has no record of a session with us (no connection, or
1417                  * connect_seq == 0), they will send RESETSESION to indicate
1418                  * that they must have reset their session, and may have
1419                  * dropped messages.
1420                  */
1421                 dout("process_connect got RESET peer seq %u\n",
1422                      le32_to_cpu(con->in_connect.connect_seq));
1423                 pr_err("%s%lld %s connection reset\n",
1424                        ENTITY_NAME(con->peer_name),
1425                        ceph_pr_addr(&con->peer_addr.in_addr));
1426                 reset_connection(con);
1427                 ceph_con_out_kvec_reset(con);
1428                 ret = prepare_write_connect(con);
1429                 if (ret < 0)
1430                         return ret;
1431                 prepare_read_connect(con);
1432
1433                 /* Tell ceph about it. */
1434                 mutex_unlock(&con->mutex);
1435                 pr_info("reset on %s%lld\n", ENTITY_NAME(con->peer_name));
1436                 if (con->ops->peer_reset)
1437                         con->ops->peer_reset(con);
1438                 mutex_lock(&con->mutex);
1439                 if (test_bit(CLOSED, &con->state) ||
1440                     test_bit(OPENING, &con->state))
1441                         return -EAGAIN;
1442                 break;
1443
1444         case CEPH_MSGR_TAG_RETRY_SESSION:
1445                 /*
1446                  * If we sent a smaller connect_seq than the peer has, try
1447                  * again with a larger value.
1448                  */
1449                 dout("process_connect got RETRY my seq = %u, peer_seq = %u\n",
1450                      le32_to_cpu(con->out_connect.connect_seq),
1451                      le32_to_cpu(con->in_connect.connect_seq));
1452                 con->connect_seq = le32_to_cpu(con->in_connect.connect_seq);
1453                 ceph_con_out_kvec_reset(con);
1454                 ret = prepare_write_connect(con);
1455                 if (ret < 0)
1456                         return ret;
1457                 prepare_read_connect(con);
1458                 break;
1459
1460         case CEPH_MSGR_TAG_RETRY_GLOBAL:
1461                 /*
1462                  * If we sent a smaller global_seq than the peer has, try
1463                  * again with a larger value.
1464                  */
1465                 dout("process_connect got RETRY_GLOBAL my %u peer_gseq %u\n",
1466                      con->peer_global_seq,
1467                      le32_to_cpu(con->in_connect.global_seq));
1468                 get_global_seq(con->msgr,
1469                                le32_to_cpu(con->in_connect.global_seq));
1470                 ceph_con_out_kvec_reset(con);
1471                 ret = prepare_write_connect(con);
1472                 if (ret < 0)
1473                         return ret;
1474                 prepare_read_connect(con);
1475                 break;
1476
1477         case CEPH_MSGR_TAG_READY:
1478                 if (req_feat & ~server_feat) {
1479                         pr_err("%s%lld %s protocol feature mismatch,"
1480                                " my required %llx > server's %llx, need %llx\n",
1481                                ENTITY_NAME(con->peer_name),
1482                                ceph_pr_addr(&con->peer_addr.in_addr),
1483                                req_feat, server_feat, req_feat & ~server_feat);
1484                         con->error_msg = "missing required protocol features";
1485                         fail_protocol(con);
1486                         return -1;
1487                 }
1488                 clear_bit(CONNECTING, &con->state);
1489                 con->peer_global_seq = le32_to_cpu(con->in_reply.global_seq);
1490                 con->connect_seq++;
1491                 con->peer_features = server_feat;
1492                 dout("process_connect got READY gseq %d cseq %d (%d)\n",
1493                      con->peer_global_seq,
1494                      le32_to_cpu(con->in_reply.connect_seq),
1495                      con->connect_seq);
1496                 WARN_ON(con->connect_seq !=
1497                         le32_to_cpu(con->in_reply.connect_seq));
1498
1499                 if (con->in_reply.flags & CEPH_MSG_CONNECT_LOSSY)
1500                         set_bit(LOSSYTX, &con->state);
1501
1502                 prepare_read_tag(con);
1503                 break;
1504
1505         case CEPH_MSGR_TAG_WAIT:
1506                 /*
1507                  * If there is a connection race (we are opening
1508                  * connections to each other), one of us may just have
1509                  * to WAIT.  This shouldn't happen if we are the
1510                  * client.
1511                  */
1512                 pr_err("process_connect got WAIT as client\n");
1513                 con->error_msg = "protocol error, got WAIT as client";
1514                 return -1;
1515
1516         default:
1517                 pr_err("connect protocol error, will retry\n");
1518                 con->error_msg = "protocol error, garbage tag during connect";
1519                 return -1;
1520         }
1521         return 0;
1522 }
1523
1524
1525 /*
1526  * read (part of) an ack
1527  */
1528 static int read_partial_ack(struct ceph_connection *con)
1529 {
1530         int size = sizeof (con->in_temp_ack);
1531         int end = size;
1532
1533         return read_partial(con, end, size, &con->in_temp_ack);
1534 }
1535
1536
1537 /*
1538  * We can finally discard anything that's been acked.
1539  */
1540 static void process_ack(struct ceph_connection *con)
1541 {
1542         struct ceph_msg *m;
1543         u64 ack = le64_to_cpu(con->in_temp_ack);
1544         u64 seq;
1545
1546         while (!list_empty(&con->out_sent)) {
1547                 m = list_first_entry(&con->out_sent, struct ceph_msg,
1548                                      list_head);
1549                 seq = le64_to_cpu(m->hdr.seq);
1550                 if (seq > ack)
1551                         break;
1552                 dout("got ack for seq %llu type %d at %p\n", seq,
1553                      le16_to_cpu(m->hdr.type), m);
1554                 m->ack_stamp = jiffies;
1555                 ceph_msg_remove(m);
1556         }
1557         prepare_read_tag(con);
1558 }
1559
1560
1561
1562
1563 static int read_partial_message_section(struct ceph_connection *con,
1564                                         struct kvec *section,
1565                                         unsigned int sec_len, u32 *crc)
1566 {
1567         int ret, left;
1568
1569         BUG_ON(!section);
1570
1571         while (section->iov_len < sec_len) {
1572                 BUG_ON(section->iov_base == NULL);
1573                 left = sec_len - section->iov_len;
1574                 ret = ceph_tcp_recvmsg(con->sock, (char *)section->iov_base +
1575                                        section->iov_len, left);
1576                 if (ret <= 0)
1577                         return ret;
1578                 section->iov_len += ret;
1579         }
1580         if (section->iov_len == sec_len)
1581                 *crc = crc32c(0, section->iov_base, section->iov_len);
1582
1583         return 1;
1584 }
1585
1586 static struct ceph_msg *ceph_alloc_msg(struct ceph_connection *con,
1587                                 struct ceph_msg_header *hdr,
1588                                 int *skip);
1589
1590
1591 static int read_partial_message_pages(struct ceph_connection *con,
1592                                       struct page **pages,
1593                                       unsigned int data_len, bool do_datacrc)
1594 {
1595         void *p;
1596         int ret;
1597         int left;
1598
1599         left = min((int)(data_len - con->in_msg_pos.data_pos),
1600                    (int)(PAGE_SIZE - con->in_msg_pos.page_pos));
1601         /* (page) data */
1602         BUG_ON(pages == NULL);
1603         p = kmap(pages[con->in_msg_pos.page]);
1604         ret = ceph_tcp_recvmsg(con->sock, p + con->in_msg_pos.page_pos,
1605                                left);
1606         if (ret > 0 && do_datacrc)
1607                 con->in_data_crc =
1608                         crc32c(con->in_data_crc,
1609                                   p + con->in_msg_pos.page_pos, ret);
1610         kunmap(pages[con->in_msg_pos.page]);
1611         if (ret <= 0)
1612                 return ret;
1613         con->in_msg_pos.data_pos += ret;
1614         con->in_msg_pos.page_pos += ret;
1615         if (con->in_msg_pos.page_pos == PAGE_SIZE) {
1616                 con->in_msg_pos.page_pos = 0;
1617                 con->in_msg_pos.page++;
1618         }
1619
1620         return ret;
1621 }
1622
1623 #ifdef CONFIG_BLOCK
1624 static int read_partial_message_bio(struct ceph_connection *con,
1625                                     struct bio **bio_iter, int *bio_seg,
1626                                     unsigned int data_len, bool do_datacrc)
1627 {
1628         struct bio_vec *bv = bio_iovec_idx(*bio_iter, *bio_seg);
1629         void *p;
1630         int ret, left;
1631
1632         if (IS_ERR(bv))
1633                 return PTR_ERR(bv);
1634
1635         left = min((int)(data_len - con->in_msg_pos.data_pos),
1636                    (int)(bv->bv_len - con->in_msg_pos.page_pos));
1637
1638         p = kmap(bv->bv_page) + bv->bv_offset;
1639
1640         ret = ceph_tcp_recvmsg(con->sock, p + con->in_msg_pos.page_pos,
1641                                left);
1642         if (ret > 0 && do_datacrc)
1643                 con->in_data_crc =
1644                         crc32c(con->in_data_crc,
1645                                   p + con->in_msg_pos.page_pos, ret);
1646         kunmap(bv->bv_page);
1647         if (ret <= 0)
1648                 return ret;
1649         con->in_msg_pos.data_pos += ret;
1650         con->in_msg_pos.page_pos += ret;
1651         if (con->in_msg_pos.page_pos == bv->bv_len) {
1652                 con->in_msg_pos.page_pos = 0;
1653                 iter_bio_next(bio_iter, bio_seg);
1654         }
1655
1656         return ret;
1657 }
1658 #endif
1659
1660 /*
1661  * read (part of) a message.
1662  */
1663 static int read_partial_message(struct ceph_connection *con)
1664 {
1665         struct ceph_msg *m = con->in_msg;
1666         int size;
1667         int end;
1668         int ret;
1669         unsigned int front_len, middle_len, data_len;
1670         bool do_datacrc = !con->msgr->nocrc;
1671         int skip;
1672         u64 seq;
1673         u32 crc;
1674
1675         dout("read_partial_message con %p msg %p\n", con, m);
1676
1677         /* header */
1678         size = sizeof (con->in_hdr);
1679         end = size;
1680         ret = read_partial(con, end, size, &con->in_hdr);
1681         if (ret <= 0)
1682                 return ret;
1683
1684         crc = crc32c(0, &con->in_hdr, offsetof(struct ceph_msg_header, crc));
1685         if (cpu_to_le32(crc) != con->in_hdr.crc) {
1686                 pr_err("read_partial_message bad hdr "
1687                        " crc %u != expected %u\n",
1688                        crc, con->in_hdr.crc);
1689                 return -EBADMSG;
1690         }
1691
1692         front_len = le32_to_cpu(con->in_hdr.front_len);
1693         if (front_len > CEPH_MSG_MAX_FRONT_LEN)
1694                 return -EIO;
1695         middle_len = le32_to_cpu(con->in_hdr.middle_len);
1696         if (middle_len > CEPH_MSG_MAX_DATA_LEN)
1697                 return -EIO;
1698         data_len = le32_to_cpu(con->in_hdr.data_len);
1699         if (data_len > CEPH_MSG_MAX_DATA_LEN)
1700                 return -EIO;
1701
1702         /* verify seq# */
1703         seq = le64_to_cpu(con->in_hdr.seq);
1704         if ((s64)seq - (s64)con->in_seq < 1) {
1705                 pr_info("skipping %s%lld %s seq %lld expected %lld\n",
1706                         ENTITY_NAME(con->peer_name),
1707                         ceph_pr_addr(&con->peer_addr.in_addr),
1708                         seq, con->in_seq + 1);
1709                 con->in_base_pos = -front_len - middle_len - data_len -
1710                         sizeof(m->footer);
1711                 con->in_tag = CEPH_MSGR_TAG_READY;
1712                 return 0;
1713         } else if ((s64)seq - (s64)con->in_seq > 1) {
1714                 pr_err("read_partial_message bad seq %lld expected %lld\n",
1715                        seq, con->in_seq + 1);
1716                 con->error_msg = "bad message sequence # for incoming message";
1717                 return -EBADMSG;
1718         }
1719
1720         /* allocate message? */
1721         if (!con->in_msg) {
1722                 dout("got hdr type %d front %d data %d\n", con->in_hdr.type,
1723                      con->in_hdr.front_len, con->in_hdr.data_len);
1724                 skip = 0;
1725                 con->in_msg = ceph_alloc_msg(con, &con->in_hdr, &skip);
1726                 if (skip) {
1727                         /* skip this message */
1728                         dout("alloc_msg said skip message\n");
1729                         BUG_ON(con->in_msg);
1730                         con->in_base_pos = -front_len - middle_len - data_len -
1731                                 sizeof(m->footer);
1732                         con->in_tag = CEPH_MSGR_TAG_READY;
1733                         con->in_seq++;
1734                         return 0;
1735                 }
1736                 if (!con->in_msg) {
1737                         con->error_msg =
1738                                 "error allocating memory for incoming message";
1739                         return -ENOMEM;
1740                 }
1741                 m = con->in_msg;
1742                 m->front.iov_len = 0;    /* haven't read it yet */
1743                 if (m->middle)
1744                         m->middle->vec.iov_len = 0;
1745
1746                 con->in_msg_pos.page = 0;
1747                 if (m->pages)
1748                         con->in_msg_pos.page_pos = m->page_alignment;
1749                 else
1750                         con->in_msg_pos.page_pos = 0;
1751                 con->in_msg_pos.data_pos = 0;
1752         }
1753
1754         /* front */
1755         ret = read_partial_message_section(con, &m->front, front_len,
1756                                            &con->in_front_crc);
1757         if (ret <= 0)
1758                 return ret;
1759
1760         /* middle */
1761         if (m->middle) {
1762                 ret = read_partial_message_section(con, &m->middle->vec,
1763                                                    middle_len,
1764                                                    &con->in_middle_crc);
1765                 if (ret <= 0)
1766                         return ret;
1767         }
1768 #ifdef CONFIG_BLOCK
1769         if (m->bio && !m->bio_iter)
1770                 init_bio_iter(m->bio, &m->bio_iter, &m->bio_seg);
1771 #endif
1772
1773         /* (page) data */
1774         while (con->in_msg_pos.data_pos < data_len) {
1775                 if (m->pages) {
1776                         ret = read_partial_message_pages(con, m->pages,
1777                                                  data_len, do_datacrc);
1778                         if (ret <= 0)
1779                                 return ret;
1780 #ifdef CONFIG_BLOCK
1781                 } else if (m->bio) {
1782
1783                         ret = read_partial_message_bio(con,
1784                                                  &m->bio_iter, &m->bio_seg,
1785                                                  data_len, do_datacrc);
1786                         if (ret <= 0)
1787                                 return ret;
1788 #endif
1789                 } else {
1790                         BUG_ON(1);
1791                 }
1792         }
1793
1794         /* footer */
1795         size = sizeof (m->footer);
1796         end += size;
1797         ret = read_partial(con, end, size, &m->footer);
1798         if (ret <= 0)
1799                 return ret;
1800
1801         dout("read_partial_message got msg %p %d (%u) + %d (%u) + %d (%u)\n",
1802              m, front_len, m->footer.front_crc, middle_len,
1803              m->footer.middle_crc, data_len, m->footer.data_crc);
1804
1805         /* crc ok? */
1806         if (con->in_front_crc != le32_to_cpu(m->footer.front_crc)) {
1807                 pr_err("read_partial_message %p front crc %u != exp. %u\n",
1808                        m, con->in_front_crc, m->footer.front_crc);
1809                 return -EBADMSG;
1810         }
1811         if (con->in_middle_crc != le32_to_cpu(m->footer.middle_crc)) {
1812                 pr_err("read_partial_message %p middle crc %u != exp %u\n",
1813                        m, con->in_middle_crc, m->footer.middle_crc);
1814                 return -EBADMSG;
1815         }
1816         if (do_datacrc &&
1817             (m->footer.flags & CEPH_MSG_FOOTER_NOCRC) == 0 &&
1818             con->in_data_crc != le32_to_cpu(m->footer.data_crc)) {
1819                 pr_err("read_partial_message %p data crc %u != exp. %u\n", m,
1820                        con->in_data_crc, le32_to_cpu(m->footer.data_crc));
1821                 return -EBADMSG;
1822         }
1823
1824         return 1; /* done! */
1825 }
1826
1827 /*
1828  * Process message.  This happens in the worker thread.  The callback should
1829  * be careful not to do anything that waits on other incoming messages or it
1830  * may deadlock.
1831  */
1832 static void process_message(struct ceph_connection *con)
1833 {
1834         struct ceph_msg *msg;
1835
1836         msg = con->in_msg;
1837         con->in_msg = NULL;
1838
1839         /* if first message, set peer_name */
1840         if (con->peer_name.type == 0)
1841                 con->peer_name = msg->hdr.src;
1842
1843         con->in_seq++;
1844         mutex_unlock(&con->mutex);
1845
1846         dout("===== %p %llu from %s%lld %d=%s len %d+%d (%u %u %u) =====\n",
1847              msg, le64_to_cpu(msg->hdr.seq),
1848              ENTITY_NAME(msg->hdr.src),
1849              le16_to_cpu(msg->hdr.type),
1850              ceph_msg_type_name(le16_to_cpu(msg->hdr.type)),
1851              le32_to_cpu(msg->hdr.front_len),
1852              le32_to_cpu(msg->hdr.data_len),
1853              con->in_front_crc, con->in_middle_crc, con->in_data_crc);
1854         con->ops->dispatch(con, msg);
1855
1856         mutex_lock(&con->mutex);
1857         prepare_read_tag(con);
1858 }
1859
1860
1861 /*
1862  * Write something to the socket.  Called in a worker thread when the
1863  * socket appears to be writeable and we have something ready to send.
1864  */
1865 static int try_write(struct ceph_connection *con)
1866 {
1867         int ret = 1;
1868
1869         dout("try_write start %p state %lu nref %d\n", con, con->state,
1870              atomic_read(&con->nref));
1871
1872 more:
1873         dout("try_write out_kvec_bytes %d\n", con->out_kvec_bytes);
1874
1875         /* open the socket first? */
1876         if (con->sock == NULL) {
1877                 ceph_con_out_kvec_reset(con);
1878                 prepare_write_banner(con);
1879                 ret = prepare_write_connect(con);
1880                 if (ret < 0)
1881                         goto out;
1882                 prepare_read_banner(con);
1883                 set_bit(CONNECTING, &con->state);
1884                 clear_bit(NEGOTIATING, &con->state);
1885
1886                 BUG_ON(con->in_msg);
1887                 con->in_tag = CEPH_MSGR_TAG_READY;
1888                 dout("try_write initiating connect on %p new state %lu\n",
1889                      con, con->state);
1890                 ret = ceph_tcp_connect(con);
1891                 if (ret < 0) {
1892                         con->error_msg = "connect error";
1893                         goto out;
1894                 }
1895         }
1896
1897 more_kvec:
1898         /* kvec data queued? */
1899         if (con->out_skip) {
1900                 ret = write_partial_skip(con);
1901                 if (ret <= 0)
1902                         goto out;
1903         }
1904         if (con->out_kvec_left) {
1905                 ret = write_partial_kvec(con);
1906                 if (ret <= 0)
1907                         goto out;
1908         }
1909
1910         /* msg pages? */
1911         if (con->out_msg) {
1912                 if (con->out_msg_done) {
1913                         ceph_msg_put(con->out_msg);
1914                         con->out_msg = NULL;   /* we're done with this one */
1915                         goto do_next;
1916                 }
1917
1918                 ret = write_partial_msg_pages(con);
1919                 if (ret == 1)
1920                         goto more_kvec;  /* we need to send the footer, too! */
1921                 if (ret == 0)
1922                         goto out;
1923                 if (ret < 0) {
1924                         dout("try_write write_partial_msg_pages err %d\n",
1925                              ret);
1926                         goto out;
1927                 }
1928         }
1929
1930 do_next:
1931         if (!test_bit(CONNECTING, &con->state)) {
1932                 /* is anything else pending? */
1933                 if (!list_empty(&con->out_queue)) {
1934                         prepare_write_message(con);
1935                         goto more;
1936                 }
1937                 if (con->in_seq > con->in_seq_acked) {
1938                         prepare_write_ack(con);
1939                         goto more;
1940                 }
1941                 if (test_and_clear_bit(KEEPALIVE_PENDING, &con->state)) {
1942                         prepare_write_keepalive(con);
1943                         goto more;
1944                 }
1945         }
1946
1947         /* Nothing to do! */
1948         clear_bit(WRITE_PENDING, &con->state);
1949         dout("try_write nothing else to write.\n");
1950         ret = 0;
1951 out:
1952         dout("try_write done on %p ret %d\n", con, ret);
1953         return ret;
1954 }
1955
1956
1957
1958 /*
1959  * Read what we can from the socket.
1960  */
1961 static int try_read(struct ceph_connection *con)
1962 {
1963         int ret = -1;
1964
1965         if (!con->sock)
1966                 return 0;
1967
1968         if (test_bit(STANDBY, &con->state))
1969                 return 0;
1970
1971         dout("try_read start on %p\n", con);
1972
1973 more:
1974         dout("try_read tag %d in_base_pos %d\n", (int)con->in_tag,
1975              con->in_base_pos);
1976
1977         /*
1978          * process_connect and process_message drop and re-take
1979          * con->mutex.  make sure we handle a racing close or reopen.
1980          */
1981         if (test_bit(CLOSED, &con->state) ||
1982             test_bit(OPENING, &con->state)) {
1983                 ret = -EAGAIN;
1984                 goto out;
1985         }
1986
1987         if (test_bit(CONNECTING, &con->state)) {
1988                 if (!test_bit(NEGOTIATING, &con->state)) {
1989                         dout("try_read connecting\n");
1990                         ret = read_partial_banner(con);
1991                         if (ret <= 0)
1992                                 goto out;
1993                         ret = process_banner(con);
1994                         if (ret < 0)
1995                                 goto out;
1996                 }
1997                 ret = read_partial_connect(con);
1998                 if (ret <= 0)
1999                         goto out;
2000                 ret = process_connect(con);
2001                 if (ret < 0)
2002                         goto out;
2003                 goto more;
2004         }
2005
2006         if (con->in_base_pos < 0) {
2007                 /*
2008                  * skipping + discarding content.
2009                  *
2010                  * FIXME: there must be a better way to do this!
2011                  */
2012                 static char buf[SKIP_BUF_SIZE];
2013                 int skip = min((int) sizeof (buf), -con->in_base_pos);
2014
2015                 dout("skipping %d / %d bytes\n", skip, -con->in_base_pos);
2016                 ret = ceph_tcp_recvmsg(con->sock, buf, skip);
2017                 if (ret <= 0)
2018                         goto out;
2019                 con->in_base_pos += ret;
2020                 if (con->in_base_pos)
2021                         goto more;
2022         }
2023         if (con->in_tag == CEPH_MSGR_TAG_READY) {
2024                 /*
2025                  * what's next?
2026                  */
2027                 ret = ceph_tcp_recvmsg(con->sock, &con->in_tag, 1);
2028                 if (ret <= 0)
2029                         goto out;
2030                 dout("try_read got tag %d\n", (int)con->in_tag);
2031                 switch (con->in_tag) {
2032                 case CEPH_MSGR_TAG_MSG:
2033                         prepare_read_message(con);
2034                         break;
2035                 case CEPH_MSGR_TAG_ACK:
2036                         prepare_read_ack(con);
2037                         break;
2038                 case CEPH_MSGR_TAG_CLOSE:
2039                         set_bit(CLOSED, &con->state);   /* fixme */
2040                         goto out;
2041                 default:
2042                         goto bad_tag;
2043                 }
2044         }
2045         if (con->in_tag == CEPH_MSGR_TAG_MSG) {
2046                 ret = read_partial_message(con);
2047                 if (ret <= 0) {
2048                         switch (ret) {
2049                         case -EBADMSG:
2050                                 con->error_msg = "bad crc";
2051                                 ret = -EIO;
2052                                 break;
2053                         case -EIO:
2054                                 con->error_msg = "io error";
2055                                 break;
2056                         }
2057                         goto out;
2058                 }
2059                 if (con->in_tag == CEPH_MSGR_TAG_READY)
2060                         goto more;
2061                 process_message(con);
2062                 goto more;
2063         }
2064         if (con->in_tag == CEPH_MSGR_TAG_ACK) {
2065                 ret = read_partial_ack(con);
2066                 if (ret <= 0)
2067                         goto out;
2068                 process_ack(con);
2069                 goto more;
2070         }
2071
2072 out:
2073         dout("try_read done on %p ret %d\n", con, ret);
2074         return ret;
2075
2076 bad_tag:
2077         pr_err("try_read bad con->in_tag = %d\n", (int)con->in_tag);
2078         con->error_msg = "protocol error, garbage tag";
2079         ret = -1;
2080         goto out;
2081 }
2082
2083
2084 /*
2085  * Atomically queue work on a connection.  Bump @con reference to
2086  * avoid races with connection teardown.
2087  */
2088 static void queue_con(struct ceph_connection *con)
2089 {
2090         if (test_bit(DEAD, &con->state)) {
2091                 dout("queue_con %p ignoring: DEAD\n",
2092                      con);
2093                 return;
2094         }
2095
2096         if (!con->ops->get(con)) {
2097                 dout("queue_con %p ref count 0\n", con);
2098                 return;
2099         }
2100
2101         if (!queue_delayed_work(ceph_msgr_wq, &con->work, 0)) {
2102                 dout("queue_con %p - already queued\n", con);
2103                 con->ops->put(con);
2104         } else {
2105                 dout("queue_con %p\n", con);
2106         }
2107 }
2108
2109 /*
2110  * Do some work on a connection.  Drop a connection ref when we're done.
2111  */
2112 static void con_work(struct work_struct *work)
2113 {
2114         struct ceph_connection *con = container_of(work, struct ceph_connection,
2115                                                    work.work);
2116         int ret;
2117
2118         mutex_lock(&con->mutex);
2119 restart:
2120         if (test_and_clear_bit(BACKOFF, &con->state)) {
2121                 dout("con_work %p backing off\n", con);
2122                 if (queue_delayed_work(ceph_msgr_wq, &con->work,
2123                                        round_jiffies_relative(con->delay))) {
2124                         dout("con_work %p backoff %lu\n", con, con->delay);
2125                         mutex_unlock(&con->mutex);
2126                         return;
2127                 } else {
2128                         con->ops->put(con);
2129                         dout("con_work %p FAILED to back off %lu\n", con,
2130                              con->delay);
2131                 }
2132         }
2133
2134         if (test_bit(STANDBY, &con->state)) {
2135                 dout("con_work %p STANDBY\n", con);
2136                 goto done;
2137         }
2138         if (test_bit(CLOSED, &con->state)) { /* e.g. if we are replaced */
2139                 dout("con_work CLOSED\n");
2140                 con_close_socket(con);
2141                 goto done;
2142         }
2143         if (test_and_clear_bit(OPENING, &con->state)) {
2144                 /* reopen w/ new peer */
2145                 dout("con_work OPENING\n");
2146                 con_close_socket(con);
2147         }
2148
2149         if (test_and_clear_bit(SOCK_CLOSED, &con->state))
2150                 goto fault;
2151
2152         ret = try_read(con);
2153         if (ret == -EAGAIN)
2154                 goto restart;
2155         if (ret < 0)
2156                 goto fault;
2157
2158         ret = try_write(con);
2159         if (ret == -EAGAIN)
2160                 goto restart;
2161         if (ret < 0)
2162                 goto fault;
2163
2164 done:
2165         mutex_unlock(&con->mutex);
2166 done_unlocked:
2167         con->ops->put(con);
2168         return;
2169
2170 fault:
2171         mutex_unlock(&con->mutex);
2172         ceph_fault(con);     /* error/fault path */
2173         goto done_unlocked;
2174 }
2175
2176
2177 /*
2178  * Generic error/fault handler.  A retry mechanism is used with
2179  * exponential backoff
2180  */
2181 static void ceph_fault(struct ceph_connection *con)
2182 {
2183         pr_err("%s%lld %s %s\n", ENTITY_NAME(con->peer_name),
2184                ceph_pr_addr(&con->peer_addr.in_addr), con->error_msg);
2185         dout("fault %p state %lu to peer %s\n",
2186              con, con->state, ceph_pr_addr(&con->peer_addr.in_addr));
2187
2188         if (test_bit(LOSSYTX, &con->state)) {
2189                 dout("fault on LOSSYTX channel\n");
2190                 goto out;
2191         }
2192
2193         mutex_lock(&con->mutex);
2194         if (test_bit(CLOSED, &con->state))
2195                 goto out_unlock;
2196
2197         con_close_socket(con);
2198
2199         if (con->in_msg) {
2200                 ceph_msg_put(con->in_msg);
2201                 con->in_msg = NULL;
2202         }
2203
2204         /* Requeue anything that hasn't been acked */
2205         list_splice_init(&con->out_sent, &con->out_queue);
2206
2207         /* If there are no messages queued or keepalive pending, place
2208          * the connection in a STANDBY state */
2209         if (list_empty(&con->out_queue) &&
2210             !test_bit(KEEPALIVE_PENDING, &con->state)) {
2211                 dout("fault %p setting STANDBY clearing WRITE_PENDING\n", con);
2212                 clear_bit(WRITE_PENDING, &con->state);
2213                 set_bit(STANDBY, &con->state);
2214         } else {
2215                 /* retry after a delay. */
2216                 if (con->delay == 0)
2217                         con->delay = BASE_DELAY_INTERVAL;
2218                 else if (con->delay < MAX_DELAY_INTERVAL)
2219                         con->delay *= 2;
2220                 con->ops->get(con);
2221                 if (queue_delayed_work(ceph_msgr_wq, &con->work,
2222                                        round_jiffies_relative(con->delay))) {
2223                         dout("fault queued %p delay %lu\n", con, con->delay);
2224                 } else {
2225                         con->ops->put(con);
2226                         dout("fault failed to queue %p delay %lu, backoff\n",
2227                              con, con->delay);
2228                         /*
2229                          * In many cases we see a socket state change
2230                          * while con_work is running and end up
2231                          * queuing (non-delayed) work, such that we
2232                          * can't backoff with a delay.  Set a flag so
2233                          * that when con_work restarts we schedule the
2234                          * delay then.
2235                          */
2236                         set_bit(BACKOFF, &con->state);
2237                 }
2238         }
2239
2240 out_unlock:
2241         mutex_unlock(&con->mutex);
2242 out:
2243         /*
2244          * in case we faulted due to authentication, invalidate our
2245          * current tickets so that we can get new ones.
2246          */
2247         if (con->auth_retry && con->ops->invalidate_authorizer) {
2248                 dout("calling invalidate_authorizer()\n");
2249                 con->ops->invalidate_authorizer(con);
2250         }
2251
2252         if (con->ops->fault)
2253                 con->ops->fault(con);
2254 }
2255
2256
2257
2258 /*
2259  * create a new messenger instance
2260  */
2261 struct ceph_messenger *ceph_messenger_create(struct ceph_entity_addr *myaddr,
2262                                              u32 supported_features,
2263                                              u32 required_features)
2264 {
2265         struct ceph_messenger *msgr;
2266
2267         msgr = kzalloc(sizeof(*msgr), GFP_KERNEL);
2268         if (msgr == NULL)
2269                 return ERR_PTR(-ENOMEM);
2270
2271         msgr->supported_features = supported_features;
2272         msgr->required_features = required_features;
2273
2274         spin_lock_init(&msgr->global_seq_lock);
2275
2276         if (myaddr)
2277                 msgr->inst.addr = *myaddr;
2278
2279         /* select a random nonce */
2280         msgr->inst.addr.type = 0;
2281         get_random_bytes(&msgr->inst.addr.nonce, sizeof(msgr->inst.addr.nonce));
2282         encode_my_addr(msgr);
2283
2284         dout("messenger_create %p\n", msgr);
2285         return msgr;
2286 }
2287 EXPORT_SYMBOL(ceph_messenger_create);
2288
2289 void ceph_messenger_destroy(struct ceph_messenger *msgr)
2290 {
2291         dout("destroy %p\n", msgr);
2292         kfree(msgr);
2293         dout("destroyed messenger %p\n", msgr);
2294 }
2295 EXPORT_SYMBOL(ceph_messenger_destroy);
2296
2297 static void clear_standby(struct ceph_connection *con)
2298 {
2299         /* come back from STANDBY? */
2300         if (test_and_clear_bit(STANDBY, &con->state)) {
2301                 mutex_lock(&con->mutex);
2302                 dout("clear_standby %p and ++connect_seq\n", con);
2303                 con->connect_seq++;
2304                 WARN_ON(test_bit(WRITE_PENDING, &con->state));
2305                 WARN_ON(test_bit(KEEPALIVE_PENDING, &con->state));
2306                 mutex_unlock(&con->mutex);
2307         }
2308 }
2309
2310 /*
2311  * Queue up an outgoing message on the given connection.
2312  */
2313 void ceph_con_send(struct ceph_connection *con, struct ceph_msg *msg)
2314 {
2315         if (test_bit(CLOSED, &con->state)) {
2316                 dout("con_send %p closed, dropping %p\n", con, msg);
2317                 ceph_msg_put(msg);
2318                 return;
2319         }
2320
2321         /* set src+dst */
2322         msg->hdr.src = con->msgr->inst.name;
2323
2324         BUG_ON(msg->front.iov_len != le32_to_cpu(msg->hdr.front_len));
2325
2326         msg->needs_out_seq = true;
2327
2328         /* queue */
2329         mutex_lock(&con->mutex);
2330         BUG_ON(!list_empty(&msg->list_head));
2331         list_add_tail(&msg->list_head, &con->out_queue);
2332         dout("----- %p to %s%lld %d=%s len %d+%d+%d -----\n", msg,
2333              ENTITY_NAME(con->peer_name), le16_to_cpu(msg->hdr.type),
2334              ceph_msg_type_name(le16_to_cpu(msg->hdr.type)),
2335              le32_to_cpu(msg->hdr.front_len),
2336              le32_to_cpu(msg->hdr.middle_len),
2337              le32_to_cpu(msg->hdr.data_len));
2338         mutex_unlock(&con->mutex);
2339
2340         /* if there wasn't anything waiting to send before, queue
2341          * new work */
2342         clear_standby(con);
2343         if (test_and_set_bit(WRITE_PENDING, &con->state) == 0)
2344                 queue_con(con);
2345 }
2346 EXPORT_SYMBOL(ceph_con_send);
2347
2348 /*
2349  * Revoke a message that was previously queued for send
2350  */
2351 void ceph_con_revoke(struct ceph_connection *con, struct ceph_msg *msg)
2352 {
2353         mutex_lock(&con->mutex);
2354         if (!list_empty(&msg->list_head)) {
2355                 dout("con_revoke %p msg %p - was on queue\n", con, msg);
2356                 list_del_init(&msg->list_head);
2357                 ceph_msg_put(msg);
2358                 msg->hdr.seq = 0;
2359         }
2360         if (con->out_msg == msg) {
2361                 dout("con_revoke %p msg %p - was sending\n", con, msg);
2362                 con->out_msg = NULL;
2363                 if (con->out_kvec_is_msg) {
2364                         con->out_skip = con->out_kvec_bytes;
2365                         con->out_kvec_is_msg = false;
2366                 }
2367                 ceph_msg_put(msg);
2368                 msg->hdr.seq = 0;
2369         }
2370         mutex_unlock(&con->mutex);
2371 }
2372
2373 /*
2374  * Revoke a message that we may be reading data into
2375  */
2376 void ceph_con_revoke_message(struct ceph_connection *con, struct ceph_msg *msg)
2377 {
2378         mutex_lock(&con->mutex);
2379         if (con->in_msg && con->in_msg == msg) {
2380                 unsigned int front_len = le32_to_cpu(con->in_hdr.front_len);
2381                 unsigned int middle_len = le32_to_cpu(con->in_hdr.middle_len);
2382                 unsigned int data_len = le32_to_cpu(con->in_hdr.data_len);
2383
2384                 /* skip rest of message */
2385                 dout("con_revoke_pages %p msg %p revoked\n", con, msg);
2386                         con->in_base_pos = con->in_base_pos -
2387                                 sizeof(struct ceph_msg_header) -
2388                                 front_len -
2389                                 middle_len -
2390                                 data_len -
2391                                 sizeof(struct ceph_msg_footer);
2392                 ceph_msg_put(con->in_msg);
2393                 con->in_msg = NULL;
2394                 con->in_tag = CEPH_MSGR_TAG_READY;
2395                 con->in_seq++;
2396         } else {
2397                 dout("con_revoke_pages %p msg %p pages %p no-op\n",
2398                      con, con->in_msg, msg);
2399         }
2400         mutex_unlock(&con->mutex);
2401 }
2402
2403 /*
2404  * Queue a keepalive byte to ensure the tcp connection is alive.
2405  */
2406 void ceph_con_keepalive(struct ceph_connection *con)
2407 {
2408         dout("con_keepalive %p\n", con);
2409         clear_standby(con);
2410         if (test_and_set_bit(KEEPALIVE_PENDING, &con->state) == 0 &&
2411             test_and_set_bit(WRITE_PENDING, &con->state) == 0)
2412                 queue_con(con);
2413 }
2414 EXPORT_SYMBOL(ceph_con_keepalive);
2415
2416
2417 /*
2418  * construct a new message with given type, size
2419  * the new msg has a ref count of 1.
2420  */
2421 struct ceph_msg *ceph_msg_new(int type, int front_len, gfp_t flags,
2422                               bool can_fail)
2423 {
2424         struct ceph_msg *m;
2425
2426         m = kmalloc(sizeof(*m), flags);
2427         if (m == NULL)
2428                 goto out;
2429         kref_init(&m->kref);
2430         INIT_LIST_HEAD(&m->list_head);
2431
2432         m->hdr.tid = 0;
2433         m->hdr.type = cpu_to_le16(type);
2434         m->hdr.priority = cpu_to_le16(CEPH_MSG_PRIO_DEFAULT);
2435         m->hdr.version = 0;
2436         m->hdr.front_len = cpu_to_le32(front_len);
2437         m->hdr.middle_len = 0;
2438         m->hdr.data_len = 0;
2439         m->hdr.data_off = 0;
2440         m->hdr.reserved = 0;
2441         m->footer.front_crc = 0;
2442         m->footer.middle_crc = 0;
2443         m->footer.data_crc = 0;
2444         m->footer.flags = 0;
2445         m->front_max = front_len;
2446         m->front_is_vmalloc = false;
2447         m->more_to_follow = false;
2448         m->ack_stamp = 0;
2449         m->pool = NULL;
2450
2451         /* middle */
2452         m->middle = NULL;
2453
2454         /* data */
2455         m->nr_pages = 0;
2456         m->page_alignment = 0;
2457         m->pages = NULL;
2458         m->pagelist = NULL;
2459         m->bio = NULL;
2460         m->bio_iter = NULL;
2461         m->bio_seg = 0;
2462         m->trail = NULL;
2463
2464         /* front */
2465         if (front_len) {
2466                 if (front_len > PAGE_CACHE_SIZE) {
2467                         m->front.iov_base = __vmalloc(front_len, flags,
2468                                                       PAGE_KERNEL);
2469                         m->front_is_vmalloc = true;
2470                 } else {
2471                         m->front.iov_base = kmalloc(front_len, flags);
2472                 }
2473                 if (m->front.iov_base == NULL) {
2474                         dout("ceph_msg_new can't allocate %d bytes\n",
2475                              front_len);
2476                         goto out2;
2477                 }
2478         } else {
2479                 m->front.iov_base = NULL;
2480         }
2481         m->front.iov_len = front_len;
2482
2483         dout("ceph_msg_new %p front %d\n", m, front_len);
2484         return m;
2485
2486 out2:
2487         ceph_msg_put(m);
2488 out:
2489         if (!can_fail) {
2490                 pr_err("msg_new can't create type %d front %d\n", type,
2491                        front_len);
2492                 WARN_ON(1);
2493         } else {
2494                 dout("msg_new can't create type %d front %d\n", type,
2495                      front_len);
2496         }
2497         return NULL;
2498 }
2499 EXPORT_SYMBOL(ceph_msg_new);
2500
2501 /*
2502  * Allocate "middle" portion of a message, if it is needed and wasn't
2503  * allocated by alloc_msg.  This allows us to read a small fixed-size
2504  * per-type header in the front and then gracefully fail (i.e.,
2505  * propagate the error to the caller based on info in the front) when
2506  * the middle is too large.
2507  */
2508 static int ceph_alloc_middle(struct ceph_connection *con, struct ceph_msg *msg)
2509 {
2510         int type = le16_to_cpu(msg->hdr.type);
2511         int middle_len = le32_to_cpu(msg->hdr.middle_len);
2512
2513         dout("alloc_middle %p type %d %s middle_len %d\n", msg, type,
2514              ceph_msg_type_name(type), middle_len);
2515         BUG_ON(!middle_len);
2516         BUG_ON(msg->middle);
2517
2518         msg->middle = ceph_buffer_new(middle_len, GFP_NOFS);
2519         if (!msg->middle)
2520                 return -ENOMEM;
2521         return 0;
2522 }
2523
2524 /*
2525  * Generic message allocator, for incoming messages.
2526  */
2527 static struct ceph_msg *ceph_alloc_msg(struct ceph_connection *con,
2528                                 struct ceph_msg_header *hdr,
2529                                 int *skip)
2530 {
2531         int type = le16_to_cpu(hdr->type);
2532         int front_len = le32_to_cpu(hdr->front_len);
2533         int middle_len = le32_to_cpu(hdr->middle_len);
2534         struct ceph_msg *msg = NULL;
2535         int ret;
2536
2537         if (con->ops->alloc_msg) {
2538                 mutex_unlock(&con->mutex);
2539                 msg = con->ops->alloc_msg(con, hdr, skip);
2540                 mutex_lock(&con->mutex);
2541                 if (!msg || *skip)
2542                         return NULL;
2543         }
2544         if (!msg) {
2545                 *skip = 0;
2546                 msg = ceph_msg_new(type, front_len, GFP_NOFS, false);
2547                 if (!msg) {
2548                         pr_err("unable to allocate msg type %d len %d\n",
2549                                type, front_len);
2550                         return NULL;
2551                 }
2552                 msg->page_alignment = le16_to_cpu(hdr->data_off);
2553         }
2554         memcpy(&msg->hdr, &con->in_hdr, sizeof(con->in_hdr));
2555
2556         if (middle_len && !msg->middle) {
2557                 ret = ceph_alloc_middle(con, msg);
2558                 if (ret < 0) {
2559                         ceph_msg_put(msg);
2560                         return NULL;
2561                 }
2562         }
2563
2564         return msg;
2565 }
2566
2567
2568 /*
2569  * Free a generically kmalloc'd message.
2570  */
2571 void ceph_msg_kfree(struct ceph_msg *m)
2572 {
2573         dout("msg_kfree %p\n", m);
2574         if (m->front_is_vmalloc)
2575                 vfree(m->front.iov_base);
2576         else
2577                 kfree(m->front.iov_base);
2578         kfree(m);
2579 }
2580
2581 /*
2582  * Drop a msg ref.  Destroy as needed.
2583  */
2584 void ceph_msg_last_put(struct kref *kref)
2585 {
2586         struct ceph_msg *m = container_of(kref, struct ceph_msg, kref);
2587
2588         dout("ceph_msg_put last one on %p\n", m);
2589         WARN_ON(!list_empty(&m->list_head));
2590
2591         /* drop middle, data, if any */
2592         if (m->middle) {
2593                 ceph_buffer_put(m->middle);
2594                 m->middle = NULL;
2595         }
2596         m->nr_pages = 0;
2597         m->pages = NULL;
2598
2599         if (m->pagelist) {
2600                 ceph_pagelist_release(m->pagelist);
2601                 kfree(m->pagelist);
2602                 m->pagelist = NULL;
2603         }
2604
2605         m->trail = NULL;
2606
2607         if (m->pool)
2608                 ceph_msgpool_put(m->pool, m);
2609         else
2610                 ceph_msg_kfree(m);
2611 }
2612 EXPORT_SYMBOL(ceph_msg_last_put);
2613
2614 void ceph_msg_dump(struct ceph_msg *msg)
2615 {
2616         pr_debug("msg_dump %p (front_max %d nr_pages %d)\n", msg,
2617                  msg->front_max, msg->nr_pages);
2618         print_hex_dump(KERN_DEBUG, "header: ",
2619                        DUMP_PREFIX_OFFSET, 16, 1,
2620                        &msg->hdr, sizeof(msg->hdr), true);
2621         print_hex_dump(KERN_DEBUG, " front: ",
2622                        DUMP_PREFIX_OFFSET, 16, 1,
2623                        msg->front.iov_base, msg->front.iov_len, true);
2624         if (msg->middle)
2625                 print_hex_dump(KERN_DEBUG, "middle: ",
2626                                DUMP_PREFIX_OFFSET, 16, 1,
2627                                msg->middle->vec.iov_base,
2628                                msg->middle->vec.iov_len, true);
2629         print_hex_dump(KERN_DEBUG, "footer: ",
2630                        DUMP_PREFIX_OFFSET, 16, 1,
2631                        &msg->footer, sizeof(msg->footer), true);
2632 }
2633 EXPORT_SYMBOL(ceph_msg_dump);