fs/ceph/messenger.c

   1 #include "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 <net/tcp.h>
  13
  14 #include "super.h"
  15 #include "messenger.h"
  16 #include "decode.h"
  17 #include "pagelist.h"
  18
  19 /*
  20  * Ceph uses the messenger to exchange ceph_msg messages with other
  21  * hosts in the system.  The messenger provides ordered and reliable
  22  * delivery.  We tolerate TCP disconnects by reconnecting (with
  23  * exponential backoff) in the case of a fault (disconnection, bad
  24  * crc, protocol error).  Acks allow sent messages to be discarded by
  25  * the sender.
  26  */
  27
  28 /* static tag bytes (protocol control messages) */
  29 static char tag_msg = CEPH_MSGR_TAG_MSG;
  30 static char tag_ack = CEPH_MSGR_TAG_ACK;
  31 static char tag_keepalive = CEPH_MSGR_TAG_KEEPALIVE;
  32
  33 #ifdef CONFIG_LOCKDEP
  34 static struct lock_class_key socket_class;
  35 #endif
  36
  37
  38 static void queue_con(struct ceph_connection *con);
  39 static void con_work(struct work_struct *);
  40 static void ceph_fault(struct ceph_connection *con);
  41
  42 const char *ceph_name_type_str(int t)
  43 {
  44         switch (t) {
  45         case CEPH_ENTITY_TYPE_MON: return "mon";
  46         case CEPH_ENTITY_TYPE_MDS: return "mds";
  47         case CEPH_ENTITY_TYPE_OSD: return "osd";
  48         case CEPH_ENTITY_TYPE_CLIENT: return "client";
  49         case CEPH_ENTITY_TYPE_ADMIN: return "admin";
  50         default: return "???";
  51         }
  52 }
  53
  54 /*
  55  * nicely render a sockaddr as a string.
  56  */
  57 #define MAX_ADDR_STR 20
  58 static char addr_str[MAX_ADDR_STR][40];
  59 static DEFINE_SPINLOCK(addr_str_lock);
  60 static int last_addr_str;
  61
  62 const char *pr_addr(const struct sockaddr_storage *ss)
  63 {
  64         int i;
  65         char *s;
  66         struct sockaddr_in *in4 = (void *)ss;
  67         unsigned char *quad = (void *)&in4->sin_addr.s_addr;
  68         struct sockaddr_in6 *in6 = (void *)ss;
  69
  70         spin_lock(&addr_str_lock);
  71         i = last_addr_str++;
  72         if (last_addr_str == MAX_ADDR_STR)
  73                 last_addr_str = 0;
  74         spin_unlock(&addr_str_lock);
  75         s = addr_str[i];
  76
  77         switch (ss->ss_family) {
  78         case AF_INET:
  79                 sprintf(s, "%u.%u.%u.%u:%u",
  80                         (unsigned int)quad[0],
  81                         (unsigned int)quad[1],
  82                         (unsigned int)quad[2],
  83                         (unsigned int)quad[3],
  84                         (unsigned int)ntohs(in4->sin_port));
  85                 break;
  86
  87         case AF_INET6:
  88                 sprintf(s, "%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x:%u",
  89                         in6->sin6_addr.s6_addr16[0],
  90                         in6->sin6_addr.s6_addr16[1],
  91                         in6->sin6_addr.s6_addr16[2],
  92                         in6->sin6_addr.s6_addr16[3],
  93                         in6->sin6_addr.s6_addr16[4],
  94                         in6->sin6_addr.s6_addr16[5],
  95                         in6->sin6_addr.s6_addr16[6],
  96                         in6->sin6_addr.s6_addr16[7],
  97                         (unsigned int)ntohs(in6->sin6_port));
  98                 break;
  99
 100         default:
 101                 sprintf(s, "(unknown sockaddr family %d)", (int)ss->ss_family);
 102         }
 103
 104         return s;
 105 }
 106
 107 static void encode_my_addr(struct ceph_messenger *msgr)
 108 {
 109         memcpy(&msgr->my_enc_addr, &msgr->inst.addr, sizeof(msgr->my_enc_addr));
 110         ceph_encode_addr(&msgr->my_enc_addr);
 111 }
 112
 113 /*
 114  * work queue for all reading and writing to/from the socket.
 115  */
 116 struct workqueue_struct *ceph_msgr_wq;
 117
 118 int __init ceph_msgr_init(void)
 119 {
 120         ceph_msgr_wq = create_workqueue("ceph-msgr");
 121         if (IS_ERR(ceph_msgr_wq)) {
 122                 int ret = PTR_ERR(ceph_msgr_wq);
 123                 pr_err("msgr_init failed to create workqueue: %d\n", ret);
 124                 ceph_msgr_wq = NULL;
 125                 return ret;
 126         }
 127         return 0;
 128 }
 129
 130 void ceph_msgr_exit(void)
 131 {
 132         destroy_workqueue(ceph_msgr_wq);
 133 }
 134
 135 /*
 136  * socket callback functions
 137  */
 138
 139 /* data available on socket, or listen socket received a connect */
 140 static void ceph_data_ready(struct sock *sk, int count_unused)
 141 {
 142         struct ceph_connection *con =
 143                 (struct ceph_connection *)sk->sk_user_data;
 144         if (sk->sk_state != TCP_CLOSE_WAIT) {
 145                 dout("ceph_data_ready on %p state = %lu, queueing work\n",
 146                      con, con->state);
 147                 queue_con(con);
 148         }
 149 }
 150
 151 /* socket has buffer space for writing */
 152 static void ceph_write_space(struct sock *sk)
 153 {
 154         struct ceph_connection *con =
 155                 (struct ceph_connection *)sk->sk_user_data;
 156
 157         /* only queue to workqueue if there is data we want to write. */
 158         if (test_bit(WRITE_PENDING, &con->state)) {
 159                 dout("ceph_write_space %p queueing write work\n", con);
 160                 queue_con(con);
 161         } else {
 162                 dout("ceph_write_space %p nothing to write\n", con);
 163         }
 164
 165         /* since we have our own write_space, clear the SOCK_NOSPACE flag */
 166         clear_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
 167 }
 168
 169 /* socket's state has changed */
 170 static void ceph_state_change(struct sock *sk)
 171 {
 172         struct ceph_connection *con =
 173                 (struct ceph_connection *)sk->sk_user_data;
 174
 175         dout("ceph_state_change %p state = %lu sk_state = %u\n",
 176              con, con->state, sk->sk_state);
 177
 178         if (test_bit(CLOSED, &con->state))
 179                 return;
 180
 181         switch (sk->sk_state) {
 182         case TCP_CLOSE:
 183                 dout("ceph_state_change TCP_CLOSE\n");
 184         case TCP_CLOSE_WAIT:
 185                 dout("ceph_state_change TCP_CLOSE_WAIT\n");
 186                 if (test_and_set_bit(SOCK_CLOSED, &con->state) == 0) {
 187                         if (test_bit(CONNECTING, &con->state))
 188                                 con->error_msg = "connection failed";
 189                         else
 190                                 con->error_msg = "socket closed";
 191                         queue_con(con);
 192                 }
 193                 break;
 194         case TCP_ESTABLISHED:
 195                 dout("ceph_state_change TCP_ESTABLISHED\n");
 196                 queue_con(con);
 197                 break;
 198         }
 199 }
 200
 201 /*
 202  * set up socket callbacks
 203  */
 204 static void set_sock_callbacks(struct socket *sock,
 205                                struct ceph_connection *con)
 206 {
 207         struct sock *sk = sock->sk;
 208         sk->sk_user_data = (void *)con;
 209         sk->sk_data_ready = ceph_data_ready;
 210         sk->sk_write_space = ceph_write_space;
 211         sk->sk_state_change = ceph_state_change;
 212 }
 213
 214
 215 /*
 216  * socket helpers
 217  */
 218
 219 /*
 220  * initiate connection to a remote socket.
 221  */
 222 static struct socket *ceph_tcp_connect(struct ceph_connection *con)
 223 {
 224         struct sockaddr *paddr = (struct sockaddr *)&con->peer_addr.in_addr;
 225         struct socket *sock;
 226         int ret;
 227
 228         BUG_ON(con->sock);
 229         ret = sock_create_kern(AF_INET, SOCK_STREAM, IPPROTO_TCP, &sock);
 230         if (ret)
 231                 return ERR_PTR(ret);
 232         con->sock = sock;
 233         sock->sk->sk_allocation = GFP_NOFS;
 234
 235 #ifdef CONFIG_LOCKDEP
 236         lockdep_set_class(&sock->sk->sk_lock, &socket_class);
 237 #endif
 238
 239         set_sock_callbacks(sock, con);
 240
 241         dout("connect %s\n", pr_addr(&con->peer_addr.in_addr));
 242
 243         ret = sock->ops->connect(sock, paddr, sizeof(*paddr), O_NONBLOCK);
 244         if (ret == -EINPROGRESS) {
 245                 dout("connect %s EINPROGRESS sk_state = %u\n",
 246                      pr_addr(&con->peer_addr.in_addr),
 247                      sock->sk->sk_state);
 248                 ret = 0;
 249         }
 250         if (ret < 0) {
 251                 pr_err("connect %s error %d\n",
 252                        pr_addr(&con->peer_addr.in_addr), ret);
 253                 sock_release(sock);
 254                 con->sock = NULL;
 255                 con->error_msg = "connect error";
 256         }
 257
 258         if (ret < 0)
 259                 return ERR_PTR(ret);
 260         return sock;
 261 }
 262
 263 static int ceph_tcp_recvmsg(struct socket *sock, void *buf, size_t len)
 264 {
 265         struct kvec iov = {buf, len};
 266         struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
 267
 268         return kernel_recvmsg(sock, &msg, &iov, 1, len, msg.msg_flags);
 269 }
 270
 271 /*
 272  * write something.  @more is true if caller will be sending more data
 273  * shortly.
 274  */
 275 static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov,
 276                      size_t kvlen, size_t len, int more)
 277 {
 278         struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
 279
 280         if (more)
 281                 msg.msg_flags |= MSG_MORE;
 282         else
 283                 msg.msg_flags |= MSG_EOR;  /* superfluous, but what the hell */
 284
 285         return kernel_sendmsg(sock, &msg, iov, kvlen, len);
 286 }
 287
 288
 289 /*
 290  * Shutdown/close the socket for the given connection.
 291  */
 292 static int con_close_socket(struct ceph_connection *con)
 293 {
 294         int rc;
 295
 296         dout("con_close_socket on %p sock %p\n", con, con->sock);
 297         if (!con->sock)
 298                 return 0;
 299         set_bit(SOCK_CLOSED, &con->state);
 300         rc = con->sock->ops->shutdown(con->sock, SHUT_RDWR);
 301         sock_release(con->sock);
 302         con->sock = NULL;
 303         clear_bit(SOCK_CLOSED, &con->state);
 304         return rc;
 305 }
 306
 307 /*
 308  * Reset a connection.  Discard all incoming and outgoing messages
 309  * and clear *_seq state.
 310  */
 311 static void ceph_msg_remove(struct ceph_msg *msg)
 312 {
 313         list_del_init(&msg->list_head);
 314         ceph_msg_put(msg);
 315 }
 316 static void ceph_msg_remove_list(struct list_head *head)
 317 {
 318         while (!list_empty(head)) {
 319                 struct ceph_msg *msg = list_first_entry(head, struct ceph_msg,
 320                                                         list_head);
 321                 ceph_msg_remove(msg);
 322         }
 323 }
 324
 325 static void reset_connection(struct ceph_connection *con)
 326 {
 327         /* reset connection, out_queue, msg_ and connect_seq */
 328         /* discard existing out_queue and msg_seq */
 329         ceph_msg_remove_list(&con->out_queue);
 330         ceph_msg_remove_list(&con->out_sent);
 331
 332         if (con->in_msg) {
 333                 ceph_msg_put(con->in_msg);
 334                 con->in_msg = NULL;
 335         }
 336
 337         con->connect_seq = 0;
 338         con->out_seq = 0;
 339         if (con->out_msg) {
 340                 ceph_msg_put(con->out_msg);
 341                 con->out_msg = NULL;
 342         }
 343         con->in_seq = 0;
 344         con->in_seq_acked = 0;
 345 }
 346
 347 /*
 348  * mark a peer down.  drop any open connections.
 349  */
 350 void ceph_con_close(struct ceph_connection *con)
 351 {
 352         dout("con_close %p peer %s\n", con, pr_addr(&con->peer_addr.in_addr));
 353         set_bit(CLOSED, &con->state);  /* in case there's queued work */
 354         clear_bit(STANDBY, &con->state);  /* avoid connect_seq bump */
 355         clear_bit(LOSSYTX, &con->state);  /* so we retry next connect */
 356         clear_bit(KEEPALIVE_PENDING, &con->state);
 357         clear_bit(WRITE_PENDING, &con->state);
 358         mutex_lock(&con->mutex);
 359         reset_connection(con);
 360         cancel_delayed_work(&con->work);
 361         mutex_unlock(&con->mutex);
 362         queue_con(con);
 363 }
 364
 365 /*
 366  * Reopen a closed connection, with a new peer address.
 367  */
 368 void ceph_con_open(struct ceph_connection *con, struct ceph_entity_addr *addr)
 369 {
 370         dout("con_open %p %s\n", con, pr_addr(&addr->in_addr));
 371         set_bit(OPENING, &con->state);
 372         clear_bit(CLOSED, &con->state);
 373         memcpy(&con->peer_addr, addr, sizeof(*addr));
 374         con->delay = 0;      /* reset backoff memory */
 375         queue_con(con);
 376 }
 377
 378 /*
 379  * return true if this connection ever successfully opened
 380  */
 381 bool ceph_con_opened(struct ceph_connection *con)
 382 {
 383         return con->connect_seq > 0;
 384 }
 385
 386 /*
 387  * generic get/put
 388  */
 389 struct ceph_connection *ceph_con_get(struct ceph_connection *con)
 390 {
 391         dout("con_get %p nref = %d -> %d\n", con,
 392              atomic_read(&con->nref), atomic_read(&con->nref) + 1);
 393         if (atomic_inc_not_zero(&con->nref))
 394                 return con;
 395         return NULL;
 396 }
 397
 398 void ceph_con_put(struct ceph_connection *con)
 399 {
 400         dout("con_put %p nref = %d -> %d\n", con,
 401              atomic_read(&con->nref), atomic_read(&con->nref) - 1);
 402         BUG_ON(atomic_read(&con->nref) == 0);
 403         if (atomic_dec_and_test(&con->nref)) {
 404                 BUG_ON(con->sock);
 405                 kfree(con);
 406         }
 407 }
 408
 409 /*
 410  * initialize a new connection.
 411  */
 412 void ceph_con_init(struct ceph_messenger *msgr, struct ceph_connection *con)
 413 {
 414         dout("con_init %p\n", con);
 415         memset(con, 0, sizeof(*con));
 416         atomic_set(&con->nref, 1);
 417         con->msgr = msgr;
 418         mutex_init(&con->mutex);
 419         INIT_LIST_HEAD(&con->out_queue);
 420         INIT_LIST_HEAD(&con->out_sent);
 421         INIT_DELAYED_WORK(&con->work, con_work);
 422 }
 423
 424
 425 /*
 426  * We maintain a global counter to order connection attempts.  Get
 427  * a unique seq greater than @gt.
 428  */
 429 static u32 get_global_seq(struct ceph_messenger *msgr, u32 gt)
 430 {
 431         u32 ret;
 432
 433         spin_lock(&msgr->global_seq_lock);
 434         if (msgr->global_seq < gt)
 435                 msgr->global_seq = gt;
 436         ret = ++msgr->global_seq;
 437         spin_unlock(&msgr->global_seq_lock);
 438         return ret;
 439 }
 440
 441
 442 /*
 443  * Prepare footer for currently outgoing message, and finish things
 444  * off.  Assumes out_kvec* are already valid.. we just add on to the end.
 445  */
 446 static void prepare_write_message_footer(struct ceph_connection *con, int v)
 447 {
 448         struct ceph_msg *m = con->out_msg;
 449
 450         dout("prepare_write_message_footer %p\n", con);
 451         con->out_kvec_is_msg = true;
 452         con->out_kvec[v].iov_base = &m->footer;
 453         con->out_kvec[v].iov_len = sizeof(m->footer);
 454         con->out_kvec_bytes += sizeof(m->footer);
 455         con->out_kvec_left++;
 456         con->out_more = m->more_to_follow;
 457         con->out_msg_done = true;
 458 }
 459
 460 /*
 461  * Prepare headers for the next outgoing message.
 462  */
 463 static void prepare_write_message(struct ceph_connection *con)
 464 {
 465         struct ceph_msg *m;
 466         int v = 0;
 467
 468         con->out_kvec_bytes = 0;
 469         con->out_kvec_is_msg = true;
 470         con->out_msg_done = false;
 471
 472         /* Sneak an ack in there first?  If we can get it into the same
 473          * TCP packet that's a good thing. */
 474         if (con->in_seq > con->in_seq_acked) {
 475                 con->in_seq_acked = con->in_seq;
 476                 con->out_kvec[v].iov_base = &tag_ack;
 477                 con->out_kvec[v++].iov_len = 1;
 478                 con->out_temp_ack = cpu_to_le64(con->in_seq_acked);
 479                 con->out_kvec[v].iov_base = &con->out_temp_ack;
 480                 con->out_kvec[v++].iov_len = sizeof(con->out_temp_ack);
 481                 con->out_kvec_bytes = 1 + sizeof(con->out_temp_ack);
 482         }
 483
 484         m = list_first_entry(&con->out_queue,
 485                        struct ceph_msg, list_head);
 486         con->out_msg = m;
 487         if (test_bit(LOSSYTX, &con->state)) {
 488                 list_del_init(&m->list_head);
 489         } else {
 490                 /* put message on sent list */
 491                 ceph_msg_get(m);
 492                 list_move_tail(&m->list_head, &con->out_sent);
 493         }
 494
 495         /*
 496          * only assign outgoing seq # if we haven't sent this message
 497          * yet.  if it is requeued, resend with it's original seq.
 498          */
 499         if (m->needs_out_seq) {
 500                 m->hdr.seq = cpu_to_le64(++con->out_seq);
 501                 m->needs_out_seq = false;
 502         }
 503
 504         dout("prepare_write_message %p seq %lld type %d len %d+%d+%d %d pgs\n",
 505              m, con->out_seq, le16_to_cpu(m->hdr.type),
 506              le32_to_cpu(m->hdr.front_len), le32_to_cpu(m->hdr.middle_len),
 507              le32_to_cpu(m->hdr.data_len),
 508              m->nr_pages);
 509         BUG_ON(le32_to_cpu(m->hdr.front_len) != m->front.iov_len);
 510
 511         /* tag + hdr + front + middle */
 512         con->out_kvec[v].iov_base = &tag_msg;
 513         con->out_kvec[v++].iov_len = 1;
 514         con->out_kvec[v].iov_base = &m->hdr;
 515         con->out_kvec[v++].iov_len = sizeof(m->hdr);
 516         con->out_kvec[v++] = m->front;
 517         if (m->middle)
 518                 con->out_kvec[v++] = m->middle->vec;
 519         con->out_kvec_left = v;
 520         con->out_kvec_bytes += 1 + sizeof(m->hdr) + m->front.iov_len +
 521                 (m->middle ? m->middle->vec.iov_len : 0);
 522         con->out_kvec_cur = con->out_kvec;
 523
 524         /* fill in crc (except data pages), footer */
 525         con->out_msg->hdr.crc =
 526                 cpu_to_le32(crc32c(0, (void *)&m->hdr,
 527                                       sizeof(m->hdr) - sizeof(m->hdr.crc)));
 528         con->out_msg->footer.flags = CEPH_MSG_FOOTER_COMPLETE;
 529         con->out_msg->footer.front_crc =
 530                 cpu_to_le32(crc32c(0, m->front.iov_base, m->front.iov_len));
 531         if (m->middle)
 532                 con->out_msg->footer.middle_crc =
 533                         cpu_to_le32(crc32c(0, m->middle->vec.iov_base,
 534                                            m->middle->vec.iov_len));
 535         else
 536                 con->out_msg->footer.middle_crc = 0;
 537         con->out_msg->footer.data_crc = 0;
 538         dout("prepare_write_message front_crc %u data_crc %u\n",
 539              le32_to_cpu(con->out_msg->footer.front_crc),
 540              le32_to_cpu(con->out_msg->footer.middle_crc));
 541
 542         /* is there a data payload? */
 543         if (le32_to_cpu(m->hdr.data_len) > 0) {
 544                 /* initialize page iterator */
 545                 con->out_msg_pos.page = 0;
 546                 con->out_msg_pos.page_pos =
 547                         le16_to_cpu(m->hdr.data_off) & ~PAGE_MASK;
 548                 con->out_msg_pos.data_pos = 0;
 549                 con->out_msg_pos.did_page_crc = 0;
 550                 con->out_more = 1;  /* data + footer will follow */
 551         } else {
 552                 /* no, queue up footer too and be done */
 553                 prepare_write_message_footer(con, v);
 554         }
 555
 556         set_bit(WRITE_PENDING, &con->state);
 557 }
 558
 559 /*
 560  * Prepare an ack.
 561  */
 562 static void prepare_write_ack(struct ceph_connection *con)
 563 {
 564         dout("prepare_write_ack %p %llu -> %llu\n", con,
 565              con->in_seq_acked, con->in_seq);
 566         con->in_seq_acked = con->in_seq;
 567
 568         con->out_kvec[0].iov_base = &tag_ack;
 569         con->out_kvec[0].iov_len = 1;
 570         con->out_temp_ack = cpu_to_le64(con->in_seq_acked);
 571         con->out_kvec[1].iov_base = &con->out_temp_ack;
 572         con->out_kvec[1].iov_len = sizeof(con->out_temp_ack);
 573         con->out_kvec_left = 2;
 574         con->out_kvec_bytes = 1 + sizeof(con->out_temp_ack);
 575         con->out_kvec_cur = con->out_kvec;
 576         con->out_more = 1;  /* more will follow.. eventually.. */
 577         set_bit(WRITE_PENDING, &con->state);
 578 }
 579
 580 /*
 581  * Prepare to write keepalive byte.
 582  */
 583 static void prepare_write_keepalive(struct ceph_connection *con)
 584 {
 585         dout("prepare_write_keepalive %p\n", con);
 586         con->out_kvec[0].iov_base = &tag_keepalive;
 587         con->out_kvec[0].iov_len = 1;
 588         con->out_kvec_left = 1;
 589         con->out_kvec_bytes = 1;
 590         con->out_kvec_cur = con->out_kvec;
 591         set_bit(WRITE_PENDING, &con->state);
 592 }
 593
 594 /*
 595  * Connection negotiation.
 596  */
 597
 598 static void prepare_connect_authorizer(struct ceph_connection *con)
 599 {
 600         void *auth_buf;
 601         int auth_len = 0;
 602         int auth_protocol = 0;
 603
 604         mutex_unlock(&con->mutex);
 605         if (con->ops->get_authorizer)
 606                 con->ops->get_authorizer(con, &auth_buf, &auth_len,
 607                                          &auth_protocol, &con->auth_reply_buf,
 608                                          &con->auth_reply_buf_len,
 609                                          con->auth_retry);
 610         mutex_lock(&con->mutex);
 611
 612         con->out_connect.authorizer_protocol = cpu_to_le32(auth_protocol);
 613         con->out_connect.authorizer_len = cpu_to_le32(auth_len);
 614
 615         con->out_kvec[con->out_kvec_left].iov_base = auth_buf;
 616         con->out_kvec[con->out_kvec_left].iov_len = auth_len;
 617         con->out_kvec_left++;
 618         con->out_kvec_bytes += auth_len;
 619 }
 620
 621 /*
 622  * We connected to a peer and are saying hello.
 623  */
 624 static void prepare_write_banner(struct ceph_messenger *msgr,
 625                                  struct ceph_connection *con)
 626 {
 627         int len = strlen(CEPH_BANNER);
 628
 629         con->out_kvec[0].iov_base = CEPH_BANNER;
 630         con->out_kvec[0].iov_len = len;
 631         con->out_kvec[1].iov_base = &msgr->my_enc_addr;
 632         con->out_kvec[1].iov_len = sizeof(msgr->my_enc_addr);
 633         con->out_kvec_left = 2;
 634         con->out_kvec_bytes = len + sizeof(msgr->my_enc_addr);
 635         con->out_kvec_cur = con->out_kvec;
 636         con->out_more = 0;
 637         set_bit(WRITE_PENDING, &con->state);
 638 }
 639
 640 static void prepare_write_connect(struct ceph_messenger *msgr,
 641                                   struct ceph_connection *con,
 642                                   int after_banner)
 643 {
 644         unsigned global_seq = get_global_seq(con->msgr, 0);
 645         int proto;
 646
 647         switch (con->peer_name.type) {
 648         case CEPH_ENTITY_TYPE_MON:
 649                 proto = CEPH_MONC_PROTOCOL;
 650                 break;
 651         case CEPH_ENTITY_TYPE_OSD:
 652                 proto = CEPH_OSDC_PROTOCOL;
 653                 break;
 654         case CEPH_ENTITY_TYPE_MDS:
 655                 proto = CEPH_MDSC_PROTOCOL;
 656                 break;
 657         default:
 658                 BUG();
 659         }
 660
 661         dout("prepare_write_connect %p cseq=%d gseq=%d proto=%d\n", con,
 662              con->connect_seq, global_seq, proto);
 663
 664         con->out_connect.features = CEPH_FEATURE_SUPPORTED;
 665         con->out_connect.host_type = cpu_to_le32(CEPH_ENTITY_TYPE_CLIENT);
 666         con->out_connect.connect_seq = cpu_to_le32(con->connect_seq);
 667         con->out_connect.global_seq = cpu_to_le32(global_seq);
 668         con->out_connect.protocol_version = cpu_to_le32(proto);
 669         con->out_connect.flags = 0;
 670
 671         if (!after_banner) {
 672                 con->out_kvec_left = 0;
 673                 con->out_kvec_bytes = 0;
 674         }
 675         con->out_kvec[con->out_kvec_left].iov_base = &con->out_connect;
 676         con->out_kvec[con->out_kvec_left].iov_len = sizeof(con->out_connect);
 677         con->out_kvec_left++;
 678         con->out_kvec_bytes += sizeof(con->out_connect);
 679         con->out_kvec_cur = con->out_kvec;
 680         con->out_more = 0;
 681         set_bit(WRITE_PENDING, &con->state);
 682
 683         prepare_connect_authorizer(con);
 684 }
 685
 686
 687 /*
 688  * write as much of pending kvecs to the socket as we can.
 689  *  1 -> done
 690  *  0 -> socket full, but more to do
 691  * <0 -> error
 692  */
 693 static int write_partial_kvec(struct ceph_connection *con)
 694 {
 695         int ret;
 696
 697         dout("write_partial_kvec %p %d left\n", con, con->out_kvec_bytes);
 698         while (con->out_kvec_bytes > 0) {
 699                 ret = ceph_tcp_sendmsg(con->sock, con->out_kvec_cur,
 700                                        con->out_kvec_left, con->out_kvec_bytes,
 701                                        con->out_more);
 702                 if (ret <= 0)
 703                         goto out;
 704                 con->out_kvec_bytes -= ret;
 705                 if (con->out_kvec_bytes == 0)
 706                         break;            /* done */
 707                 while (ret > 0) {
 708                         if (ret >= con->out_kvec_cur->iov_len) {
 709                                 ret -= con->out_kvec_cur->iov_len;
 710                                 con->out_kvec_cur++;
 711                                 con->out_kvec_left--;
 712                         } else {
 713                                 con->out_kvec_cur->iov_len -= ret;
 714                                 con->out_kvec_cur->iov_base += ret;
 715                                 ret = 0;
 716                                 break;
 717                         }
 718                 }
 719         }
 720         con->out_kvec_left = 0;
 721         con->out_kvec_is_msg = false;
 722         ret = 1;
 723 out:
 724         dout("write_partial_kvec %p %d left in %d kvecs ret = %d\n", con,
 725              con->out_kvec_bytes, con->out_kvec_left, ret);
 726         return ret;  /* done! */
 727 }
 728
 729 /*
 730  * Write as much message data payload as we can.  If we finish, queue
 731  * up the footer.
 732  *  1 -> done, footer is now queued in out_kvec[].
 733  *  0 -> socket full, but more to do
 734  * <0 -> error
 735  */
 736 static int write_partial_msg_pages(struct ceph_connection *con)
 737 {
 738         struct ceph_msg *msg = con->out_msg;
 739         unsigned data_len = le32_to_cpu(msg->hdr.data_len);
 740         size_t len;
 741         int crc = con->msgr->nocrc;
 742         int ret;
 743
 744         dout("write_partial_msg_pages %p msg %p page %d/%d offset %d\n",
 745              con, con->out_msg, con->out_msg_pos.page, con->out_msg->nr_pages,
 746              con->out_msg_pos.page_pos);
 747
 748         while (con->out_msg_pos.page < con->out_msg->nr_pages) {
 749                 struct page *page = NULL;
 750                 void *kaddr = NULL;
 751
 752                 /*
 753                  * if we are calculating the data crc (the default), we need
 754                  * to map the page.  if our pages[] has been revoked, use the
 755                  * zero page.
 756                  */
 757                 if (msg->pages) {
 758                         page = msg->pages[con->out_msg_pos.page];
 759                         if (crc)
 760                                 kaddr = kmap(page);
 761                 } else if (msg->pagelist) {
 762                         page = list_first_entry(&msg->pagelist->head,
 763                                                 struct page, lru);
 764                         if (crc)
 765                                 kaddr = kmap(page);
 766                 } else {
 767                         page = con->msgr->zero_page;
 768                         if (crc)
 769                                 kaddr = page_address(con->msgr->zero_page);
 770                 }
 771                 len = min((int)(PAGE_SIZE - con->out_msg_pos.page_pos),
 772                           (int)(data_len - con->out_msg_pos.data_pos));
 773                 if (crc && !con->out_msg_pos.did_page_crc) {
 774                         void *base = kaddr + con->out_msg_pos.page_pos;
 775                         u32 tmpcrc = le32_to_cpu(con->out_msg->footer.data_crc);
 776
 777                         BUG_ON(kaddr == NULL);
 778                         con->out_msg->footer.data_crc =
 779                                 cpu_to_le32(crc32c(tmpcrc, base, len));
 780                         con->out_msg_pos.did_page_crc = 1;
 781                 }
 782
 783                 ret = kernel_sendpage(con->sock, page,
 784                                       con->out_msg_pos.page_pos, len,
 785                                       MSG_DONTWAIT | MSG_NOSIGNAL |
 786                                       MSG_MORE);
 787
 788                 if (crc && (msg->pages || msg->pagelist))
 789                         kunmap(page);
 790
 791                 if (ret <= 0)
 792                         goto out;
 793
 794                 con->out_msg_pos.data_pos += ret;
 795                 con->out_msg_pos.page_pos += ret;
 796                 if (ret == len) {
 797                         con->out_msg_pos.page_pos = 0;
 798                         con->out_msg_pos.page++;
 799                         con->out_msg_pos.did_page_crc = 0;
 800                         if (msg->pagelist)
 801                                 list_move_tail(&page->lru,
 802                                                &msg->pagelist->head);
 803                 }
 804         }
 805
 806         dout("write_partial_msg_pages %p msg %p done\n", con, msg);
 807
 808         /* prepare and queue up footer, too */
 809         if (!crc)
 810                 con->out_msg->footer.flags |= CEPH_MSG_FOOTER_NOCRC;
 811         con->out_kvec_bytes = 0;
 812         con->out_kvec_left = 0;
 813         con->out_kvec_cur = con->out_kvec;
 814         prepare_write_message_footer(con, 0);
 815         ret = 1;
 816 out:
 817         return ret;
 818 }
 819
 820 /*
 821  * write some zeros
 822  */
 823 static int write_partial_skip(struct ceph_connection *con)
 824 {
 825         int ret;
 826
 827         while (con->out_skip > 0) {
 828                 struct kvec iov = {
 829                         .iov_base = page_address(con->msgr->zero_page),
 830                         .iov_len = min(con->out_skip, (int)PAGE_CACHE_SIZE)
 831                 };
 832
 833                 ret = ceph_tcp_sendmsg(con->sock, &iov, 1, iov.iov_len, 1);
 834                 if (ret <= 0)
 835                         goto out;
 836                 con->out_skip -= ret;
 837         }
 838         ret = 1;
 839 out:
 840         return ret;
 841 }
 842
 843 /*
 844  * Prepare to read connection handshake, or an ack.
 845  */
 846 static void prepare_read_banner(struct ceph_connection *con)
 847 {
 848         dout("prepare_read_banner %p\n", con);
 849         con->in_base_pos = 0;
 850 }
 851
 852 static void prepare_read_connect(struct ceph_connection *con)
 853 {
 854         dout("prepare_read_connect %p\n", con);
 855         con->in_base_pos = 0;
 856 }
 857
 858 static void prepare_read_ack(struct ceph_connection *con)
 859 {
 860         dout("prepare_read_ack %p\n", con);
 861         con->in_base_pos = 0;
 862 }
 863
 864 static void prepare_read_tag(struct ceph_connection *con)
 865 {
 866         dout("prepare_read_tag %p\n", con);
 867         con->in_base_pos = 0;
 868         con->in_tag = CEPH_MSGR_TAG_READY;
 869 }
 870
 871 /*
 872  * Prepare to read a message.
 873  */
 874 static int prepare_read_message(struct ceph_connection *con)
 875 {
 876         dout("prepare_read_message %p\n", con);
 877         BUG_ON(con->in_msg != NULL);
 878         con->in_base_pos = 0;
 879         con->in_front_crc = con->in_middle_crc = con->in_data_crc = 0;
 880         return 0;
 881 }
 882
 883
 884 static int read_partial(struct ceph_connection *con,
 885                         int *to, int size, void *object)
 886 {
 887         *to += size;
 888         while (con->in_base_pos < *to) {
 889                 int left = *to - con->in_base_pos;
 890                 int have = size - left;
 891                 int ret = ceph_tcp_recvmsg(con->sock, object + have, left);
 892                 if (ret <= 0)
 893                         return ret;
 894                 con->in_base_pos += ret;
 895         }
 896         return 1;
 897 }
 898
 899
 900 /*
 901  * Read all or part of the connect-side handshake on a new connection
 902  */
 903 static int read_partial_banner(struct ceph_connection *con)
 904 {
 905         int ret, to = 0;
 906
 907         dout("read_partial_banner %p at %d\n", con, con->in_base_pos);
 908
 909         /* peer's banner */
 910         ret = read_partial(con, &to, strlen(CEPH_BANNER), con->in_banner);
 911         if (ret <= 0)
 912                 goto out;
 913         ret = read_partial(con, &to, sizeof(con->actual_peer_addr),
 914                            &con->actual_peer_addr);
 915         if (ret <= 0)
 916                 goto out;
 917         ret = read_partial(con, &to, sizeof(con->peer_addr_for_me),
 918                            &con->peer_addr_for_me);
 919         if (ret <= 0)
 920                 goto out;
 921 out:
 922         return ret;
 923 }
 924
 925 static int read_partial_connect(struct ceph_connection *con)
 926 {
 927         int ret, to = 0;
 928
 929         dout("read_partial_connect %p at %d\n", con, con->in_base_pos);
 930
 931         ret = read_partial(con, &to, sizeof(con->in_reply), &con->in_reply);
 932         if (ret <= 0)
 933                 goto out;
 934         ret = read_partial(con, &to, le32_to_cpu(con->in_reply.authorizer_len),
 935                            con->auth_reply_buf);
 936         if (ret <= 0)
 937                 goto out;
 938
 939         dout("read_partial_connect %p tag %d, con_seq = %u, g_seq = %u\n",
 940              con, (int)con->in_reply.tag,
 941              le32_to_cpu(con->in_reply.connect_seq),
 942              le32_to_cpu(con->in_reply.global_seq));
 943 out:
 944         return ret;
 945
 946 }
 947
 948 /*
 949  * Verify the hello banner looks okay.
 950  */
 951 static int verify_hello(struct ceph_connection *con)
 952 {
 953         if (memcmp(con->in_banner, CEPH_BANNER, strlen(CEPH_BANNER))) {
 954                 pr_err("connect to %s got bad banner\n",
 955                        pr_addr(&con->peer_addr.in_addr));
 956                 con->error_msg = "protocol error, bad banner";
 957                 return -1;
 958         }
 959         return 0;
 960 }
 961
 962 static bool addr_is_blank(struct sockaddr_storage *ss)
 963 {
 964         switch (ss->ss_family) {
 965         case AF_INET:
 966                 return ((struct sockaddr_in *)ss)->sin_addr.s_addr == 0;
 967         case AF_INET6:
 968                 return
 969                      ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[0] == 0 &&
 970                      ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[1] == 0 &&
 971                      ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[2] == 0 &&
 972                      ((struct sockaddr_in6 *)ss)->sin6_addr.s6_addr32[3] == 0;
 973         }
 974         return false;
 975 }
 976
 977 static int addr_port(struct sockaddr_storage *ss)
 978 {
 979         switch (ss->ss_family) {
 980         case AF_INET:
 981                 return ntohs(((struct sockaddr_in *)ss)->sin_port);
 982         case AF_INET6:
 983                 return ntohs(((struct sockaddr_in6 *)ss)->sin6_port);
 984         }
 985         return 0;
 986 }
 987
 988 static void addr_set_port(struct sockaddr_storage *ss, int p)
 989 {
 990         switch (ss->ss_family) {
 991         case AF_INET:
 992                 ((struct sockaddr_in *)ss)->sin_port = htons(p);
 993         case AF_INET6:
 994                 ((struct sockaddr_in6 *)ss)->sin6_port = htons(p);
 995         }
 996 }
 997
 998 /*
 999  * Parse an ip[:port] list into an addr array.  Use the default
1000  * monitor port if a port isn't specified.
1001  */
1002 int ceph_parse_ips(const char *c, const char *end,
1003                    struct ceph_entity_addr *addr,
1004                    int max_count, int *count)
1005 {
1006         int i;
1007         const char *p = c;
1008
1009         dout("parse_ips on '%.*s'\n", (int)(end-c), c);
1010         for (i = 0; i < max_count; i++) {
1011                 const char *ipend;
1012                 struct sockaddr_storage *ss = &addr[i].in_addr;
1013                 struct sockaddr_in *in4 = (void *)ss;
1014                 struct sockaddr_in6 *in6 = (void *)ss;
1015                 int port;
1016
1017                 memset(ss, 0, sizeof(*ss));
1018                 if (in4_pton(p, end - p, (u8 *)&in4->sin_addr.s_addr,
1019                              ',', &ipend)) {
1020                         ss->ss_family = AF_INET;
1021                 } else if (in6_pton(p, end - p, (u8 *)&in6->sin6_addr.s6_addr,
1022                                     ',', &ipend)) {
1023                         ss->ss_family = AF_INET6;
1024                 } else {
1025                         goto bad;
1026                 }
1027                 p = ipend;
1028
1029                 /* port? */
1030                 if (p < end && *p == ':') {
1031                         port = 0;
1032                         p++;
1033                         while (p < end && *p >= '0' && *p <= '9') {
1034                                 port = (port * 10) + (*p - '0');
1035                                 p++;
1036                         }
1037                         if (port > 65535 || port == 0)
1038                                 goto bad;
1039                 } else {
1040                         port = CEPH_MON_PORT;
1041                 }
1042
1043                 addr_set_port(ss, port);
1044
1045                 dout("parse_ips got %s\n", pr_addr(ss));
1046
1047                 if (p == end)
1048                         break;
1049                 if (*p != ',')
1050                         goto bad;
1051                 p++;
1052         }
1053
1054         if (p != end)
1055                 goto bad;
1056
1057         if (count)
1058                 *count = i + 1;
1059         return 0;
1060
1061 bad:
1062         pr_err("parse_ips bad ip '%s'\n", c);
1063         return -EINVAL;
1064 }
1065
1066 static int process_banner(struct ceph_connection *con)
1067 {
1068         dout("process_banner on %p\n", con);
1069
1070         if (verify_hello(con) < 0)
1071                 return -1;
1072
1073         ceph_decode_addr(&con->actual_peer_addr);
1074         ceph_decode_addr(&con->peer_addr_for_me);
1075
1076         /*
1077          * Make sure the other end is who we wanted.  note that the other
1078          * end may not yet know their ip address, so if it's 0.0.0.0, give
1079          * them the benefit of the doubt.
1080          */
1081         if (memcmp(&con->peer_addr, &con->actual_peer_addr,
1082                    sizeof(con->peer_addr)) != 0 &&
1083             !(addr_is_blank(&con->actual_peer_addr.in_addr) &&
1084               con->actual_peer_addr.nonce == con->peer_addr.nonce)) {
1085                 pr_warning("wrong peer, want %s/%lld, got %s/%lld\n",
1086                            pr_addr(&con->peer_addr.in_addr),
1087                            le64_to_cpu(con->peer_addr.nonce),
1088                            pr_addr(&con->actual_peer_addr.in_addr),
1089                            le64_to_cpu(con->actual_peer_addr.nonce));
1090                 con->error_msg = "wrong peer at address";
1091                 return -1;
1092         }
1093
1094         /*
1095          * did we learn our address?
1096          */
1097         if (addr_is_blank(&con->msgr->inst.addr.in_addr)) {
1098                 int port = addr_port(&con->msgr->inst.addr.in_addr);
1099
1100                 memcpy(&con->msgr->inst.addr.in_addr,
1101                        &con->peer_addr_for_me.in_addr,
1102                        sizeof(con->peer_addr_for_me.in_addr));
1103                 addr_set_port(&con->msgr->inst.addr.in_addr, port);
1104                 encode_my_addr(con->msgr);
1105                 dout("process_banner learned my addr is %s\n",
1106                      pr_addr(&con->msgr->inst.addr.in_addr));
1107         }
1108
1109         set_bit(NEGOTIATING, &con->state);
1110         prepare_read_connect(con);
1111         return 0;
1112 }
1113
1114 static void fail_protocol(struct ceph_connection *con)
1115 {
1116         reset_connection(con);
1117         set_bit(CLOSED, &con->state);  /* in case there's queued work */
1118
1119         mutex_unlock(&con->mutex);
1120         if (con->ops->bad_proto)
1121                 con->ops->bad_proto(con);
1122         mutex_lock(&con->mutex);
1123 }
1124
1125 static int process_connect(struct ceph_connection *con)
1126 {
1127         u64 sup_feat = CEPH_FEATURE_SUPPORTED;
1128         u64 req_feat = CEPH_FEATURE_REQUIRED;
1129         u64 server_feat = le64_to_cpu(con->in_reply.features);
1130
1131         dout("process_connect on %p tag %d\n", con, (int)con->in_tag);
1132
1133         switch (con->in_reply.tag) {
1134         case CEPH_MSGR_TAG_FEATURES:
1135                 pr_err("%s%lld %s feature set mismatch,"
1136                        " my %llx < server's %llx, missing %llx\n",
1137                        ENTITY_NAME(con->peer_name),
1138                        pr_addr(&con->peer_addr.in_addr),
1139                        sup_feat, server_feat, server_feat & ~sup_feat);
1140                 con->error_msg = "missing required protocol features";
1141                 fail_protocol(con);
1142                 return -1;
1143
1144         case CEPH_MSGR_TAG_BADPROTOVER:
1145                 pr_err("%s%lld %s protocol version mismatch,"
1146                        " my %d != server's %d\n",
1147                        ENTITY_NAME(con->peer_name),
1148                        pr_addr(&con->peer_addr.in_addr),
1149                        le32_to_cpu(con->out_connect.protocol_version),
1150                        le32_to_cpu(con->in_reply.protocol_version));
1151                 con->error_msg = "protocol version mismatch";
1152                 fail_protocol(con);
1153                 return -1;
1154
1155         case CEPH_MSGR_TAG_BADAUTHORIZER:
1156                 con->auth_retry++;
1157                 dout("process_connect %p got BADAUTHORIZER attempt %d\n", con,
1158                      con->auth_retry);
1159                 if (con->auth_retry == 2) {
1160                         con->error_msg = "connect authorization failure";
1161                         reset_connection(con);
1162                         set_bit(CLOSED, &con->state);
1163                         return -1;
1164                 }
1165                 con->auth_retry = 1;
1166                 prepare_write_connect(con->msgr, con, 0);
1167                 prepare_read_connect(con);
1168                 break;
1169
1170         case CEPH_MSGR_TAG_RESETSESSION:
1171                 /*
1172                  * If we connected with a large connect_seq but the peer
1173                  * has no record of a session with us (no connection, or
1174                  * connect_seq == 0), they will send RESETSESION to indicate
1175                  * that they must have reset their session, and may have
1176                  * dropped messages.
1177                  */
1178                 dout("process_connect got RESET peer seq %u\n",
1179                      le32_to_cpu(con->in_connect.connect_seq));
1180                 pr_err("%s%lld %s connection reset\n",
1181                        ENTITY_NAME(con->peer_name),
1182                        pr_addr(&con->peer_addr.in_addr));
1183                 reset_connection(con);
1184                 prepare_write_connect(con->msgr, con, 0);
1185                 prepare_read_connect(con);
1186
1187                 /* Tell ceph about it. */
1188                 mutex_unlock(&con->mutex);
1189                 pr_info("reset on %s%lld\n", ENTITY_NAME(con->peer_name));
1190                 if (con->ops->peer_reset)
1191                         con->ops->peer_reset(con);
1192                 mutex_lock(&con->mutex);
1193                 break;
1194
1195         case CEPH_MSGR_TAG_RETRY_SESSION:
1196                 /*
1197                  * If we sent a smaller connect_seq than the peer has, try
1198                  * again with a larger value.
1199                  */
1200                 dout("process_connect got RETRY my seq = %u, peer_seq = %u\n",
1201                      le32_to_cpu(con->out_connect.connect_seq),
1202                      le32_to_cpu(con->in_connect.connect_seq));
1203                 con->connect_seq = le32_to_cpu(con->in_connect.connect_seq);
1204                 prepare_write_connect(con->msgr, con, 0);
1205                 prepare_read_connect(con);
1206                 break;
1207
1208         case CEPH_MSGR_TAG_RETRY_GLOBAL:
1209                 /*
1210                  * If we sent a smaller global_seq than the peer has, try
1211                  * again with a larger value.
1212                  */
1213                 dout("process_connect got RETRY_GLOBAL my %u peer_gseq %u\n",
1214                      con->peer_global_seq,
1215                      le32_to_cpu(con->in_connect.global_seq));
1216                 get_global_seq(con->msgr,
1217                                le32_to_cpu(con->in_connect.global_seq));
1218                 prepare_write_connect(con->msgr, con, 0);
1219                 prepare_read_connect(con);
1220                 break;
1221
1222         case CEPH_MSGR_TAG_READY:
1223                 if (req_feat & ~server_feat) {
1224                         pr_err("%s%lld %s protocol feature mismatch,"
1225                                " my required %llx > server's %llx, need %llx\n",
1226                                ENTITY_NAME(con->peer_name),
1227                                pr_addr(&con->peer_addr.in_addr),
1228                                req_feat, server_feat, req_feat & ~server_feat);
1229                         con->error_msg = "missing required protocol features";
1230                         fail_protocol(con);
1231                         return -1;
1232                 }
1233                 clear_bit(CONNECTING, &con->state);
1234                 con->peer_global_seq = le32_to_cpu(con->in_reply.global_seq);
1235                 con->connect_seq++;
1236                 dout("process_connect got READY gseq %d cseq %d (%d)\n",
1237                      con->peer_global_seq,
1238                      le32_to_cpu(con->in_reply.connect_seq),
1239                      con->connect_seq);
1240                 WARN_ON(con->connect_seq !=
1241                         le32_to_cpu(con->in_reply.connect_seq));
1242
1243                 if (con->in_reply.flags & CEPH_MSG_CONNECT_LOSSY)
1244                         set_bit(LOSSYTX, &con->state);
1245
1246                 prepare_read_tag(con);
1247                 break;
1248
1249         case CEPH_MSGR_TAG_WAIT:
1250                 /*
1251                  * If there is a connection race (we are opening
1252                  * connections to each other), one of us may just have
1253                  * to WAIT.  This shouldn't happen if we are the
1254                  * client.
1255                  */
1256                 pr_err("process_connect peer connecting WAIT\n");
1257
1258         default:
1259                 pr_err("connect protocol error, will retry\n");
1260                 con->error_msg = "protocol error, garbage tag during connect";
1261                 return -1;
1262         }
1263         return 0;
1264 }
1265
1266
1267 /*
1268  * read (part of) an ack
1269  */
1270 static int read_partial_ack(struct ceph_connection *con)
1271 {
1272         int to = 0;
1273
1274         return read_partial(con, &to, sizeof(con->in_temp_ack),
1275                             &con->in_temp_ack);
1276 }
1277
1278
1279 /*
1280  * We can finally discard anything that's been acked.
1281  */
1282 static void process_ack(struct ceph_connection *con)
1283 {
1284         struct ceph_msg *m;
1285         u64 ack = le64_to_cpu(con->in_temp_ack);
1286         u64 seq;
1287
1288         while (!list_empty(&con->out_sent)) {
1289                 m = list_first_entry(&con->out_sent, struct ceph_msg,
1290                                      list_head);
1291                 seq = le64_to_cpu(m->hdr.seq);
1292                 if (seq > ack)
1293                         break;
1294                 dout("got ack for seq %llu type %d at %p\n", seq,
1295                      le16_to_cpu(m->hdr.type), m);
1296                 ceph_msg_remove(m);
1297         }
1298         prepare_read_tag(con);
1299 }
1300
1301
1302
1303
1304 static int read_partial_message_section(struct ceph_connection *con,
1305                                         struct kvec *section, unsigned int sec_len,
1306                                         u32 *crc)
1307 {
1308         int left;
1309         int ret;
1310
1311         BUG_ON(!section);
1312
1313         while (section->iov_len < sec_len) {
1314                 BUG_ON(section->iov_base == NULL);
1315                 left = sec_len - section->iov_len;
1316                 ret = ceph_tcp_recvmsg(con->sock, (char *)section->iov_base +
1317                                        section->iov_len, left);
1318                 if (ret <= 0)
1319                         return ret;
1320                 section->iov_len += ret;
1321                 if (section->iov_len == sec_len)
1322                         *crc = crc32c(0, section->iov_base,
1323                                       section->iov_len);
1324         }
1325
1326         return 1;
1327 }
1328
1329 static struct ceph_msg *ceph_alloc_msg(struct ceph_connection *con,
1330                                 struct ceph_msg_header *hdr,
1331                                 int *skip);
1332 /*
1333  * read (part of) a message.
1334  */
1335 static int read_partial_message(struct ceph_connection *con)
1336 {
1337         struct ceph_msg *m = con->in_msg;
1338         void *p;
1339         int ret;
1340         int to, left;
1341         unsigned front_len, middle_len, data_len, data_off;
1342         int datacrc = con->msgr->nocrc;
1343         int skip;
1344         u64 seq;
1345
1346         dout("read_partial_message con %p msg %p\n", con, m);
1347
1348         /* header */
1349         while (con->in_base_pos < sizeof(con->in_hdr)) {
1350                 left = sizeof(con->in_hdr) - con->in_base_pos;
1351                 ret = ceph_tcp_recvmsg(con->sock,
1352                                        (char *)&con->in_hdr + con->in_base_pos,
1353                                        left);
1354                 if (ret <= 0)
1355                         return ret;
1356                 con->in_base_pos += ret;
1357                 if (con->in_base_pos == sizeof(con->in_hdr)) {
1358                         u32 crc = crc32c(0, (void *)&con->in_hdr,
1359                                  sizeof(con->in_hdr) - sizeof(con->in_hdr.crc));
1360                         if (crc != le32_to_cpu(con->in_hdr.crc)) {
1361                                 pr_err("read_partial_message bad hdr "
1362                                        " crc %u != expected %u\n",
1363                                        crc, con->in_hdr.crc);
1364                                 return -EBADMSG;
1365                         }
1366                 }
1367         }
1368         front_len = le32_to_cpu(con->in_hdr.front_len);
1369         if (front_len > CEPH_MSG_MAX_FRONT_LEN)
1370                 return -EIO;
1371         middle_len = le32_to_cpu(con->in_hdr.middle_len);
1372         if (middle_len > CEPH_MSG_MAX_DATA_LEN)
1373                 return -EIO;
1374         data_len = le32_to_cpu(con->in_hdr.data_len);
1375         if (data_len > CEPH_MSG_MAX_DATA_LEN)
1376                 return -EIO;
1377         data_off = le16_to_cpu(con->in_hdr.data_off);
1378
1379         /* verify seq# */
1380         seq = le64_to_cpu(con->in_hdr.seq);
1381         if ((s64)seq - (s64)con->in_seq < 1) {
1382                 pr_info("skipping %s%lld %s seq %lld, expected %lld\n",
1383                         ENTITY_NAME(con->peer_name),
1384                         pr_addr(&con->peer_addr.in_addr),
1385                         seq, con->in_seq + 1);
1386                 con->in_base_pos = -front_len - middle_len - data_len -
1387                         sizeof(m->footer);
1388                 con->in_tag = CEPH_MSGR_TAG_READY;
1389                 con->in_seq++;
1390                 return 0;
1391         } else if ((s64)seq - (s64)con->in_seq > 1) {
1392                 pr_err("read_partial_message bad seq %lld expected %lld\n",
1393                        seq, con->in_seq + 1);
1394                 con->error_msg = "bad message sequence # for incoming message";
1395                 return -EBADMSG;
1396         }
1397
1398         /* allocate message? */
1399         if (!con->in_msg) {
1400                 dout("got hdr type %d front %d data %d\n", con->in_hdr.type,
1401                      con->in_hdr.front_len, con->in_hdr.data_len);
1402                 con->in_msg = ceph_alloc_msg(con, &con->in_hdr, &skip);
1403                 if (skip) {
1404                         /* skip this message */
1405                         dout("alloc_msg said skip message\n");
1406                         con->in_base_pos = -front_len - middle_len - data_len -
1407                                 sizeof(m->footer);
1408                         con->in_tag = CEPH_MSGR_TAG_READY;
1409                         con->in_seq++;
1410                         return 0;
1411                 }
1412                 if (!con->in_msg) {
1413                         con->error_msg =
1414                                 "error allocating memory for incoming message";
1415                         return -ENOMEM;
1416                 }
1417                 m = con->in_msg;
1418                 m->front.iov_len = 0;    /* haven't read it yet */
1419                 if (m->middle)
1420                         m->middle->vec.iov_len = 0;
1421
1422                 con->in_msg_pos.page = 0;
1423                 con->in_msg_pos.page_pos = data_off & ~PAGE_MASK;
1424                 con->in_msg_pos.data_pos = 0;
1425         }
1426
1427         /* front */
1428         ret = read_partial_message_section(con, &m->front, front_len,
1429                                            &con->in_front_crc);
1430         if (ret <= 0)
1431                 return ret;
1432
1433         /* middle */
1434         if (m->middle) {
1435                 ret = read_partial_message_section(con, &m->middle->vec, middle_len,
1436                                                    &con->in_middle_crc);
1437                 if (ret <= 0)
1438                         return ret;
1439         }
1440
1441         /* (page) data */
1442         while (con->in_msg_pos.data_pos < data_len) {
1443                 left = min((int)(data_len - con->in_msg_pos.data_pos),
1444                            (int)(PAGE_SIZE - con->in_msg_pos.page_pos));
1445                 BUG_ON(m->pages == NULL);
1446                 p = kmap(m->pages[con->in_msg_pos.page]);
1447                 ret = ceph_tcp_recvmsg(con->sock, p + con->in_msg_pos.page_pos,
1448                                        left);
1449                 if (ret > 0 && datacrc)
1450                         con->in_data_crc =
1451                                 crc32c(con->in_data_crc,
1452                                           p + con->in_msg_pos.page_pos, ret);
1453                 kunmap(m->pages[con->in_msg_pos.page]);
1454                 if (ret <= 0)
1455                         return ret;
1456                 con->in_msg_pos.data_pos += ret;
1457                 con->in_msg_pos.page_pos += ret;
1458                 if (con->in_msg_pos.page_pos == PAGE_SIZE) {
1459                         con->in_msg_pos.page_pos = 0;
1460                         con->in_msg_pos.page++;
1461                 }
1462         }
1463
1464         /* footer */
1465         to = sizeof(m->hdr) + sizeof(m->footer);
1466         while (con->in_base_pos < to) {
1467                 left = to - con->in_base_pos;
1468                 ret = ceph_tcp_recvmsg(con->sock, (char *)&m->footer +
1469                                        (con->in_base_pos - sizeof(m->hdr)),
1470                                        left);
1471                 if (ret <= 0)
1472                         return ret;
1473                 con->in_base_pos += ret;
1474         }
1475         dout("read_partial_message got msg %p %d (%u) + %d (%u) + %d (%u)\n",
1476              m, front_len, m->footer.front_crc, middle_len,
1477              m->footer.middle_crc, data_len, m->footer.data_crc);
1478
1479         /* crc ok? */
1480         if (con->in_front_crc != le32_to_cpu(m->footer.front_crc)) {
1481                 pr_err("read_partial_message %p front crc %u != exp. %u\n",
1482                        m, con->in_front_crc, m->footer.front_crc);
1483                 return -EBADMSG;
1484         }
1485         if (con->in_middle_crc != le32_to_cpu(m->footer.middle_crc)) {
1486                 pr_err("read_partial_message %p middle crc %u != exp %u\n",
1487                        m, con->in_middle_crc, m->footer.middle_crc);
1488                 return -EBADMSG;
1489         }
1490         if (datacrc &&
1491             (m->footer.flags & CEPH_MSG_FOOTER_NOCRC) == 0 &&
1492             con->in_data_crc != le32_to_cpu(m->footer.data_crc)) {
1493                 pr_err("read_partial_message %p data crc %u != exp. %u\n", m,
1494                        con->in_data_crc, le32_to_cpu(m->footer.data_crc));
1495                 return -EBADMSG;
1496         }
1497
1498         return 1; /* done! */
1499 }
1500
1501 /*
1502  * Process message.  This happens in the worker thread.  The callback should
1503  * be careful not to do anything that waits on other incoming messages or it
1504  * may deadlock.
1505  */
1506 static void process_message(struct ceph_connection *con)
1507 {
1508         struct ceph_msg *msg;
1509
1510         msg = con->in_msg;
1511         con->in_msg = NULL;
1512
1513         /* if first message, set peer_name */
1514         if (con->peer_name.type == 0)
1515                 con->peer_name = msg->hdr.src.name;
1516
1517         con->in_seq++;
1518         mutex_unlock(&con->mutex);
1519
1520         dout("===== %p %llu from %s%lld %d=%s len %d+%d (%u %u %u) =====\n",
1521              msg, le64_to_cpu(msg->hdr.seq),
1522              ENTITY_NAME(msg->hdr.src.name),
1523              le16_to_cpu(msg->hdr.type),
1524              ceph_msg_type_name(le16_to_cpu(msg->hdr.type)),
1525              le32_to_cpu(msg->hdr.front_len),
1526              le32_to_cpu(msg->hdr.data_len),
1527              con->in_front_crc, con->in_middle_crc, con->in_data_crc);
1528         con->ops->dispatch(con, msg);
1529
1530         mutex_lock(&con->mutex);
1531         prepare_read_tag(con);
1532 }
1533
1534
1535 /*
1536  * Write something to the socket.  Called in a worker thread when the
1537  * socket appears to be writeable and we have something ready to send.
1538  */
1539 static int try_write(struct ceph_connection *con)
1540 {
1541         struct ceph_messenger *msgr = con->msgr;
1542         int ret = 1;
1543
1544         dout("try_write start %p state %lu nref %d\n", con, con->state,
1545              atomic_read(&con->nref));
1546
1547         mutex_lock(&con->mutex);
1548 more:
1549         dout("try_write out_kvec_bytes %d\n", con->out_kvec_bytes);
1550
1551         /* open the socket first? */
1552         if (con->sock == NULL) {
1553                 /*
1554                  * if we were STANDBY and are reconnecting _this_
1555                  * connection, bump connect_seq now.  Always bump
1556                  * global_seq.
1557                  */
1558                 if (test_and_clear_bit(STANDBY, &con->state))
1559                         con->connect_seq++;
1560
1561                 prepare_write_banner(msgr, con);
1562                 prepare_write_connect(msgr, con, 1);
1563                 prepare_read_banner(con);
1564                 set_bit(CONNECTING, &con->state);
1565                 clear_bit(NEGOTIATING, &con->state);
1566
1567                 BUG_ON(con->in_msg);
1568                 con->in_tag = CEPH_MSGR_TAG_READY;
1569                 dout("try_write initiating connect on %p new state %lu\n",
1570                      con, con->state);
1571                 con->sock = ceph_tcp_connect(con);
1572                 if (IS_ERR(con->sock)) {
1573                         con->sock = NULL;
1574                         con->error_msg = "connect error";
1575                         ret = -1;
1576                         goto out;
1577                 }
1578         }
1579
1580 more_kvec:
1581         /* kvec data queued? */
1582         if (con->out_skip) {
1583                 ret = write_partial_skip(con);
1584                 if (ret <= 0)
1585                         goto done;
1586                 if (ret < 0) {
1587                         dout("try_write write_partial_skip err %d\n", ret);
1588                         goto done;
1589                 }
1590         }
1591         if (con->out_kvec_left) {
1592                 ret = write_partial_kvec(con);
1593                 if (ret <= 0)
1594                         goto done;
1595         }
1596
1597         /* msg pages? */
1598         if (con->out_msg) {
1599                 if (con->out_msg_done) {
1600                         ceph_msg_put(con->out_msg);
1601                         con->out_msg = NULL;   /* we're done with this one */
1602                         goto do_next;
1603                 }
1604
1605                 ret = write_partial_msg_pages(con);
1606                 if (ret == 1)
1607                         goto more_kvec;  /* we need to send the footer, too! */
1608                 if (ret == 0)
1609                         goto done;
1610                 if (ret < 0) {
1611                         dout("try_write write_partial_msg_pages err %d\n",
1612                              ret);
1613                         goto done;
1614                 }
1615         }
1616
1617 do_next:
1618         if (!test_bit(CONNECTING, &con->state)) {
1619                 /* is anything else pending? */
1620                 if (!list_empty(&con->out_queue)) {
1621                         prepare_write_message(con);
1622                         goto more;
1623                 }
1624                 if (con->in_seq > con->in_seq_acked) {
1625                         prepare_write_ack(con);
1626                         goto more;
1627                 }
1628                 if (test_and_clear_bit(KEEPALIVE_PENDING, &con->state)) {
1629                         prepare_write_keepalive(con);
1630                         goto more;
1631                 }
1632         }
1633
1634         /* Nothing to do! */
1635         clear_bit(WRITE_PENDING, &con->state);
1636         dout("try_write nothing else to write.\n");
1637 done:
1638         ret = 0;
1639 out:
1640         mutex_unlock(&con->mutex);
1641         dout("try_write done on %p\n", con);
1642         return ret;
1643 }
1644
1645
1646
1647 /*
1648  * Read what we can from the socket.
1649  */
1650 static int try_read(struct ceph_connection *con)
1651 {
1652         struct ceph_messenger *msgr;
1653         int ret = -1;
1654
1655         if (!con->sock)
1656                 return 0;
1657
1658         if (test_bit(STANDBY, &con->state))
1659                 return 0;
1660
1661         dout("try_read start on %p\n", con);
1662         msgr = con->msgr;
1663
1664         mutex_lock(&con->mutex);
1665
1666 more:
1667         dout("try_read tag %d in_base_pos %d\n", (int)con->in_tag,
1668              con->in_base_pos);
1669         if (test_bit(CONNECTING, &con->state)) {
1670                 if (!test_bit(NEGOTIATING, &con->state)) {
1671                         dout("try_read connecting\n");
1672                         ret = read_partial_banner(con);
1673                         if (ret <= 0)
1674                                 goto done;
1675                         if (process_banner(con) < 0) {
1676                                 ret = -1;
1677                                 goto out;
1678                         }
1679                 }
1680                 ret = read_partial_connect(con);
1681                 if (ret <= 0)
1682                         goto done;
1683                 if (process_connect(con) < 0) {
1684                         ret = -1;
1685                         goto out;
1686                 }
1687                 goto more;
1688         }
1689
1690         if (con->in_base_pos < 0) {
1691                 /*
1692                  * skipping + discarding content.
1693                  *
1694                  * FIXME: there must be a better way to do this!
1695                  */
1696                 static char buf[1024];
1697                 int skip = min(1024, -con->in_base_pos);
1698                 dout("skipping %d / %d bytes\n", skip, -con->in_base_pos);
1699                 ret = ceph_tcp_recvmsg(con->sock, buf, skip);
1700                 if (ret <= 0)
1701                         goto done;
1702                 con->in_base_pos += ret;
1703                 if (con->in_base_pos)
1704                         goto more;
1705         }
1706         if (con->in_tag == CEPH_MSGR_TAG_READY) {
1707                 /*
1708                  * what's next?
1709                  */
1710                 ret = ceph_tcp_recvmsg(con->sock, &con->in_tag, 1);
1711                 if (ret <= 0)
1712                         goto done;
1713                 dout("try_read got tag %d\n", (int)con->in_tag);
1714                 switch (con->in_tag) {
1715                 case CEPH_MSGR_TAG_MSG:
1716                         prepare_read_message(con);
1717                         break;
1718                 case CEPH_MSGR_TAG_ACK:
1719                         prepare_read_ack(con);
1720                         break;
1721                 case CEPH_MSGR_TAG_CLOSE:
1722                         set_bit(CLOSED, &con->state);   /* fixme */
1723                         goto done;
1724                 default:
1725                         goto bad_tag;
1726                 }
1727         }
1728         if (con->in_tag == CEPH_MSGR_TAG_MSG) {
1729                 ret = read_partial_message(con);
1730                 if (ret <= 0) {
1731                         switch (ret) {
1732                         case -EBADMSG:
1733                                 con->error_msg = "bad crc";
1734                                 ret = -EIO;
1735                                 goto out;
1736                         case -EIO:
1737                                 con->error_msg = "io error";
1738                                 goto out;
1739                         default:
1740                                 goto done;
1741                         }
1742                 }
1743                 if (con->in_tag == CEPH_MSGR_TAG_READY)
1744                         goto more;
1745                 process_message(con);
1746                 goto more;
1747         }
1748         if (con->in_tag == CEPH_MSGR_TAG_ACK) {
1749                 ret = read_partial_ack(con);
1750                 if (ret <= 0)
1751                         goto done;
1752                 process_ack(con);
1753                 goto more;
1754         }
1755
1756 done:
1757         ret = 0;
1758 out:
1759         mutex_unlock(&con->mutex);
1760         dout("try_read done on %p\n", con);
1761         return ret;
1762
1763 bad_tag:
1764         pr_err("try_read bad con->in_tag = %d\n", (int)con->in_tag);
1765         con->error_msg = "protocol error, garbage tag";
1766         ret = -1;
1767         goto out;
1768 }
1769
1770
1771 /*
1772  * Atomically queue work on a connection.  Bump @con reference to
1773  * avoid races with connection teardown.
1774  *
1775  * There is some trickery going on with QUEUED and BUSY because we
1776  * only want a _single_ thread operating on each connection at any
1777  * point in time, but we want to use all available CPUs.
1778  *
1779  * The worker thread only proceeds if it can atomically set BUSY.  It
1780  * clears QUEUED and does it's thing.  When it thinks it's done, it
1781  * clears BUSY, then rechecks QUEUED.. if it's set again, it loops
1782  * (tries again to set BUSY).
1783  *
1784  * To queue work, we first set QUEUED, _then_ if BUSY isn't set, we
1785  * try to queue work.  If that fails (work is already queued, or BUSY)
1786  * we give up (work also already being done or is queued) but leave QUEUED
1787  * set so that the worker thread will loop if necessary.
1788  */
1789 static void queue_con(struct ceph_connection *con)
1790 {
1791         if (test_bit(DEAD, &con->state)) {
1792                 dout("queue_con %p ignoring: DEAD\n",
1793                      con);
1794                 return;
1795         }
1796
1797         if (!con->ops->get(con)) {
1798                 dout("queue_con %p ref count 0\n", con);
1799                 return;
1800         }
1801
1802         set_bit(QUEUED, &con->state);
1803         if (test_bit(BUSY, &con->state)) {
1804                 dout("queue_con %p - already BUSY\n", con);
1805                 con->ops->put(con);
1806         } else if (!queue_work(ceph_msgr_wq, &con->work.work)) {
1807                 dout("queue_con %p - already queued\n", con);
1808                 con->ops->put(con);
1809         } else {
1810                 dout("queue_con %p\n", con);
1811         }
1812 }
1813
1814 /*
1815  * Do some work on a connection.  Drop a connection ref when we're done.
1816  */
1817 static void con_work(struct work_struct *work)
1818 {
1819         struct ceph_connection *con = container_of(work, struct ceph_connection,
1820                                                    work.work);
1821         int backoff = 0;
1822
1823 more:
1824         if (test_and_set_bit(BUSY, &con->state) != 0) {
1825                 dout("con_work %p BUSY already set\n", con);
1826                 goto out;
1827         }
1828         dout("con_work %p start, clearing QUEUED\n", con);
1829         clear_bit(QUEUED, &con->state);
1830
1831         if (test_bit(CLOSED, &con->state)) { /* e.g. if we are replaced */
1832                 dout("con_work CLOSED\n");
1833                 con_close_socket(con);
1834                 goto done;
1835         }
1836         if (test_and_clear_bit(OPENING, &con->state)) {
1837                 /* reopen w/ new peer */
1838                 dout("con_work OPENING\n");
1839                 con_close_socket(con);
1840         }
1841
1842         if (test_and_clear_bit(SOCK_CLOSED, &con->state) ||
1843             try_read(con) < 0 ||
1844             try_write(con) < 0) {
1845                 backoff = 1;
1846                 ceph_fault(con);     /* error/fault path */
1847         }
1848
1849 done:
1850         clear_bit(BUSY, &con->state);
1851         dout("con->state=%lu\n", con->state);
1852         if (test_bit(QUEUED, &con->state)) {
1853                 if (!backoff || test_bit(OPENING, &con->state)) {
1854                         dout("con_work %p QUEUED reset, looping\n", con);
1855                         goto more;
1856                 }
1857                 dout("con_work %p QUEUED reset, but just faulted\n", con);
1858                 clear_bit(QUEUED, &con->state);
1859         }
1860         dout("con_work %p done\n", con);
1861
1862 out:
1863         con->ops->put(con);
1864 }
1865
1866
1867 /*
1868  * Generic error/fault handler.  A retry mechanism is used with
1869  * exponential backoff
1870  */
1871 static void ceph_fault(struct ceph_connection *con)
1872 {
1873         pr_err("%s%lld %s %s\n", ENTITY_NAME(con->peer_name),
1874                pr_addr(&con->peer_addr.in_addr), con->error_msg);
1875         dout("fault %p state %lu to peer %s\n",
1876              con, con->state, pr_addr(&con->peer_addr.in_addr));
1877
1878         if (test_bit(LOSSYTX, &con->state)) {
1879                 dout("fault on LOSSYTX channel\n");
1880                 goto out;
1881         }
1882
1883         mutex_lock(&con->mutex);
1884         if (test_bit(CLOSED, &con->state))
1885                 goto out_unlock;
1886
1887         con_close_socket(con);
1888
1889         if (con->in_msg) {
1890                 ceph_msg_put(con->in_msg);
1891                 con->in_msg = NULL;
1892         }
1893
1894         /* Requeue anything that hasn't been acked */
1895         list_splice_init(&con->out_sent, &con->out_queue);
1896
1897         /* If there are no messages in the queue, place the connection
1898          * in a STANDBY state (i.e., don't try to reconnect just yet). */
1899         if (list_empty(&con->out_queue) && !con->out_keepalive_pending) {
1900                 dout("fault setting STANDBY\n");
1901                 set_bit(STANDBY, &con->state);
1902         } else {
1903                 /* retry after a delay. */
1904                 if (con->delay == 0)
1905                         con->delay = BASE_DELAY_INTERVAL;
1906                 else if (con->delay < MAX_DELAY_INTERVAL)
1907                         con->delay *= 2;
1908                 dout("fault queueing %p delay %lu\n", con, con->delay);
1909                 con->ops->get(con);
1910                 if (queue_delayed_work(ceph_msgr_wq, &con->work,
1911                                        round_jiffies_relative(con->delay)) == 0)
1912                         con->ops->put(con);
1913         }
1914
1915 out_unlock:
1916         mutex_unlock(&con->mutex);
1917 out:
1918         /*
1919          * in case we faulted due to authentication, invalidate our
1920          * current tickets so that we can get new ones.
1921          */
1922         if (con->auth_retry && con->ops->invalidate_authorizer) {
1923                 dout("calling invalidate_authorizer()\n");
1924                 con->ops->invalidate_authorizer(con);
1925         }
1926
1927         if (con->ops->fault)
1928                 con->ops->fault(con);
1929 }
1930
1931
1932
1933 /*
1934  * create a new messenger instance
1935  */
1936 struct ceph_messenger *ceph_messenger_create(struct ceph_entity_addr *myaddr)
1937 {
1938         struct ceph_messenger *msgr;
1939
1940         msgr = kzalloc(sizeof(*msgr), GFP_KERNEL);
1941         if (msgr == NULL)
1942                 return ERR_PTR(-ENOMEM);
1943
1944         spin_lock_init(&msgr->global_seq_lock);
1945
1946         /* the zero page is needed if a request is "canceled" while the message
1947          * is being written over the socket */
1948         msgr->zero_page = __page_cache_alloc(GFP_KERNEL | __GFP_ZERO);
1949         if (!msgr->zero_page) {
1950                 kfree(msgr);
1951                 return ERR_PTR(-ENOMEM);
1952         }
1953         kmap(msgr->zero_page);
1954
1955         if (myaddr)
1956                 msgr->inst.addr = *myaddr;
1957
1958         /* select a random nonce */
1959         msgr->inst.addr.type = 0;
1960         get_random_bytes(&msgr->inst.addr.nonce, sizeof(msgr->inst.addr.nonce));
1961         encode_my_addr(msgr);
1962
1963         dout("messenger_create %p\n", msgr);
1964         return msgr;
1965 }
1966
1967 void ceph_messenger_destroy(struct ceph_messenger *msgr)
1968 {
1969         dout("destroy %p\n", msgr);
1970         kunmap(msgr->zero_page);
1971         __free_page(msgr->zero_page);
1972         kfree(msgr);
1973         dout("destroyed messenger %p\n", msgr);
1974 }
1975
1976 /*
1977  * Queue up an outgoing message on the given connection.
1978  */
1979 void ceph_con_send(struct ceph_connection *con, struct ceph_msg *msg)
1980 {
1981         if (test_bit(CLOSED, &con->state)) {
1982                 dout("con_send %p closed, dropping %p\n", con, msg);
1983                 ceph_msg_put(msg);
1984                 return;
1985         }
1986
1987         /* set src+dst */
1988         msg->hdr.src.name = con->msgr->inst.name;
1989         msg->hdr.src.addr = con->msgr->my_enc_addr;
1990         msg->hdr.orig_src = msg->hdr.src;
1991
1992         BUG_ON(msg->front.iov_len != le32_to_cpu(msg->hdr.front_len));
1993
1994         msg->needs_out_seq = true;
1995
1996         /* queue */
1997         mutex_lock(&con->mutex);
1998         BUG_ON(!list_empty(&msg->list_head));
1999         list_add_tail(&msg->list_head, &con->out_queue);
2000         dout("----- %p to %s%lld %d=%s len %d+%d+%d -----\n", msg,
2001              ENTITY_NAME(con->peer_name), le16_to_cpu(msg->hdr.type),
2002              ceph_msg_type_name(le16_to_cpu(msg->hdr.type)),
2003              le32_to_cpu(msg->hdr.front_len),
2004              le32_to_cpu(msg->hdr.middle_len),
2005              le32_to_cpu(msg->hdr.data_len));
2006         mutex_unlock(&con->mutex);
2007
2008         /* if there wasn't anything waiting to send before, queue
2009          * new work */
2010         if (test_and_set_bit(WRITE_PENDING, &con->state) == 0)
2011                 queue_con(con);
2012 }
2013
2014 /*
2015  * Revoke a message that was previously queued for send
2016  */
2017 void ceph_con_revoke(struct ceph_connection *con, struct ceph_msg *msg)
2018 {
2019         mutex_lock(&con->mutex);
2020         if (!list_empty(&msg->list_head)) {
2021                 dout("con_revoke %p msg %p\n", con, msg);
2022                 list_del_init(&msg->list_head);
2023                 ceph_msg_put(msg);
2024                 msg->hdr.seq = 0;
2025                 if (con->out_msg == msg) {
2026                         ceph_msg_put(con->out_msg);
2027                         con->out_msg = NULL;
2028                 }
2029                 if (con->out_kvec_is_msg) {
2030                         con->out_skip = con->out_kvec_bytes;
2031                         con->out_kvec_is_msg = false;
2032                 }
2033         } else {
2034                 dout("con_revoke %p msg %p - not queued (sent?)\n", con, msg);
2035         }
2036         mutex_unlock(&con->mutex);
2037 }
2038
2039 /*
2040  * Revoke a message that we may be reading data into
2041  */
2042 void ceph_con_revoke_message(struct ceph_connection *con, struct ceph_msg *msg)
2043 {
2044         mutex_lock(&con->mutex);
2045         if (con->in_msg && con->in_msg == msg) {
2046                 unsigned front_len = le32_to_cpu(con->in_hdr.front_len);
2047                 unsigned middle_len = le32_to_cpu(con->in_hdr.middle_len);
2048                 unsigned data_len = le32_to_cpu(con->in_hdr.data_len);
2049
2050                 /* skip rest of message */
2051                 dout("con_revoke_pages %p msg %p revoked\n", con, msg);
2052                         con->in_base_pos = con->in_base_pos -
2053                                 sizeof(struct ceph_msg_header) -
2054                                 front_len -
2055                                 middle_len -
2056                                 data_len -
2057                                 sizeof(struct ceph_msg_footer);
2058                 ceph_msg_put(con->in_msg);
2059                 con->in_msg = NULL;
2060                 con->in_tag = CEPH_MSGR_TAG_READY;
2061                 con->in_seq++;
2062         } else {
2063                 dout("con_revoke_pages %p msg %p pages %p no-op\n",
2064                      con, con->in_msg, msg);
2065         }
2066         mutex_unlock(&con->mutex);
2067 }
2068
2069 /*
2070  * Queue a keepalive byte to ensure the tcp connection is alive.
2071  */
2072 void ceph_con_keepalive(struct ceph_connection *con)
2073 {
2074         if (test_and_set_bit(KEEPALIVE_PENDING, &con->state) == 0 &&
2075             test_and_set_bit(WRITE_PENDING, &con->state) == 0)
2076                 queue_con(con);
2077 }
2078
2079
2080 /*
2081  * construct a new message with given type, size
2082  * the new msg has a ref count of 1.
2083  */
2084 struct ceph_msg *ceph_msg_new(int type, int front_len,
2085                               int page_len, int page_off, struct page **pages)
2086 {
2087         struct ceph_msg *m;
2088
2089         m = kmalloc(sizeof(*m), GFP_NOFS);
2090         if (m == NULL)
2091                 goto out;
2092         kref_init(&m->kref);
2093         INIT_LIST_HEAD(&m->list_head);
2094
2095         m->hdr.tid = 0;
2096         m->hdr.type = cpu_to_le16(type);
2097         m->hdr.priority = cpu_to_le16(CEPH_MSG_PRIO_DEFAULT);
2098         m->hdr.version = 0;
2099         m->hdr.front_len = cpu_to_le32(front_len);
2100         m->hdr.middle_len = 0;
2101         m->hdr.data_len = cpu_to_le32(page_len);
2102         m->hdr.data_off = cpu_to_le16(page_off);
2103         m->hdr.reserved = 0;
2104         m->footer.front_crc = 0;
2105         m->footer.middle_crc = 0;
2106         m->footer.data_crc = 0;
2107         m->footer.flags = 0;
2108         m->front_max = front_len;
2109         m->front_is_vmalloc = false;
2110         m->more_to_follow = false;
2111         m->pool = NULL;
2112
2113         /* front */
2114         if (front_len) {
2115                 if (front_len > PAGE_CACHE_SIZE) {
2116                         m->front.iov_base = __vmalloc(front_len, GFP_NOFS,
2117                                                       PAGE_KERNEL);
2118                         m->front_is_vmalloc = true;
2119                 } else {
2120                         m->front.iov_base = kmalloc(front_len, GFP_NOFS);
2121                 }
2122                 if (m->front.iov_base == NULL) {
2123                         pr_err("msg_new can't allocate %d bytes\n",
2124                              front_len);
2125                         goto out2;
2126                 }
2127         } else {
2128                 m->front.iov_base = NULL;
2129         }
2130         m->front.iov_len = front_len;
2131
2132         /* middle */
2133         m->middle = NULL;
2134
2135         /* data */
2136         m->nr_pages = calc_pages_for(page_off, page_len);
2137         m->pages = pages;
2138         m->pagelist = NULL;
2139
2140         dout("ceph_msg_new %p page %d~%d -> %d\n", m, page_off, page_len,
2141              m->nr_pages);
2142         return m;
2143
2144 out2:
2145         ceph_msg_put(m);
2146 out:
2147         pr_err("msg_new can't create type %d len %d\n", type, front_len);
2148         return NULL;
2149 }
2150
2151 /*
2152  * Allocate "middle" portion of a message, if it is needed and wasn't
2153  * allocated by alloc_msg.  This allows us to read a small fixed-size
2154  * per-type header in the front and then gracefully fail (i.e.,
2155  * propagate the error to the caller based on info in the front) when
2156  * the middle is too large.
2157  */
2158 static int ceph_alloc_middle(struct ceph_connection *con, struct ceph_msg *msg)
2159 {
2160         int type = le16_to_cpu(msg->hdr.type);
2161         int middle_len = le32_to_cpu(msg->hdr.middle_len);
2162
2163         dout("alloc_middle %p type %d %s middle_len %d\n", msg, type,
2164              ceph_msg_type_name(type), middle_len);
2165         BUG_ON(!middle_len);
2166         BUG_ON(msg->middle);
2167
2168         msg->middle = ceph_buffer_new(middle_len, GFP_NOFS);
2169         if (!msg->middle)
2170                 return -ENOMEM;
2171         return 0;
2172 }
2173
2174 /*
2175  * Generic message allocator, for incoming messages.
2176  */
2177 static struct ceph_msg *ceph_alloc_msg(struct ceph_connection *con,
2178                                 struct ceph_msg_header *hdr,
2179                                 int *skip)
2180 {
2181         int type = le16_to_cpu(hdr->type);
2182         int front_len = le32_to_cpu(hdr->front_len);
2183         int middle_len = le32_to_cpu(hdr->middle_len);
2184         struct ceph_msg *msg = NULL;
2185         int ret;
2186
2187         if (con->ops->alloc_msg) {
2188                 mutex_unlock(&con->mutex);
2189                 msg = con->ops->alloc_msg(con, hdr, skip);
2190                 mutex_lock(&con->mutex);
2191                 if (!msg || *skip)
2192                         return NULL;
2193         }
2194         if (!msg) {
2195                 *skip = 0;
2196                 msg = ceph_msg_new(type, front_len, 0, 0, NULL);
2197                 if (!msg) {
2198                         pr_err("unable to allocate msg type %d len %d\n",
2199                                type, front_len);
2200                         return NULL;
2201                 }
2202         }
2203         memcpy(&msg->hdr, &con->in_hdr, sizeof(con->in_hdr));
2204
2205         if (middle_len) {
2206                 ret = ceph_alloc_middle(con, msg);
2207                 if (ret < 0) {
2208                         ceph_msg_put(msg);
2209                         return NULL;
2210                 }
2211         }
2212
2213         return msg;
2214 }
2215
2216
2217 /*
2218  * Free a generically kmalloc'd message.
2219  */
2220 void ceph_msg_kfree(struct ceph_msg *m)
2221 {
2222         dout("msg_kfree %p\n", m);
2223         if (m->front_is_vmalloc)
2224                 vfree(m->front.iov_base);
2225         else
2226                 kfree(m->front.iov_base);
2227         kfree(m);
2228 }
2229
2230 /*
2231  * Drop a msg ref.  Destroy as needed.
2232  */
2233 void ceph_msg_last_put(struct kref *kref)
2234 {
2235         struct ceph_msg *m = container_of(kref, struct ceph_msg, kref);
2236
2237         dout("ceph_msg_put last one on %p\n", m);
2238         WARN_ON(!list_empty(&m->list_head));
2239
2240         /* drop middle, data, if any */
2241         if (m->middle) {
2242                 ceph_buffer_put(m->middle);
2243                 m->middle = NULL;
2244         }
2245         m->nr_pages = 0;
2246         m->pages = NULL;
2247
2248         if (m->pagelist) {
2249                 ceph_pagelist_release(m->pagelist);
2250                 kfree(m->pagelist);
2251                 m->pagelist = NULL;
2252         }
2253
2254         if (m->pool)
2255                 ceph_msgpool_put(m->pool, m);
2256         else
2257                 ceph_msg_kfree(m);
2258 }
2259
2260 void ceph_msg_dump(struct ceph_msg *msg)
2261 {
2262         pr_debug("msg_dump %p (front_max %d nr_pages %d)\n", msg,
2263                  msg->front_max, msg->nr_pages);
2264         print_hex_dump(KERN_DEBUG, "header: ",
2265                        DUMP_PREFIX_OFFSET, 16, 1,
2266                        &msg->hdr, sizeof(msg->hdr), true);
2267         print_hex_dump(KERN_DEBUG, " front: ",
2268                        DUMP_PREFIX_OFFSET, 16, 1,
2269                        msg->front.iov_base, msg->front.iov_len, true);
2270         if (msg->middle)
2271                 print_hex_dump(KERN_DEBUG, "middle: ",
2272                                DUMP_PREFIX_OFFSET, 16, 1,
2273                                msg->middle->vec.iov_base,
2274                                msg->middle->vec.iov_len, true);
2275         print_hex_dump(KERN_DEBUG, "footer: ",
2276                        DUMP_PREFIX_OFFSET, 16, 1,
2277                        &msg->footer, sizeof(msg->footer), true);
2278 }