fs/dlm/lowcomms.c

   1 /******************************************************************************
   2 *******************************************************************************
   3 **
   4 **  Copyright (C) Sistina Software, Inc.  1997-2003  All rights reserved.
   5 **  Copyright (C) 2004-2009 Red Hat, Inc.  All rights reserved.
   6 **
   7 **  This copyrighted material is made available to anyone wishing to use,
   8 **  modify, copy, or redistribute it subject to the terms and conditions
   9 **  of the GNU General Public License v.2.
  10 **
  11 *******************************************************************************
  12 ******************************************************************************/
  13
  14 /*
  15  * lowcomms.c
  16  *
  17  * This is the "low-level" comms layer.
  18  *
  19  * It is responsible for sending/receiving messages
  20  * from other nodes in the cluster.
  21  *
  22  * Cluster nodes are referred to by their nodeids. nodeids are
  23  * simply 32 bit numbers to the locking module - if they need to
  24  * be expanded for the cluster infrastructure then that is its
  25  * responsibility. It is this layer's
  26  * responsibility to resolve these into IP address or
  27  * whatever it needs for inter-node communication.
  28  *
  29  * The comms level is two kernel threads that deal mainly with
  30  * the receiving of messages from other nodes and passing them
  31  * up to the mid-level comms layer (which understands the
  32  * message format) for execution by the locking core, and
  33  * a send thread which does all the setting up of connections
  34  * to remote nodes and the sending of data. Threads are not allowed
  35  * to send their own data because it may cause them to wait in times
  36  * of high load. Also, this way, the sending thread can collect together
  37  * messages bound for one node and send them in one block.
  38  *
  39  * lowcomms will choose to use either TCP or SCTP as its transport layer
  40  * depending on the configuration variable 'protocol'. This should be set
  41  * to 0 (default) for TCP or 1 for SCTP. It should be configured using a
  42  * cluster-wide mechanism as it must be the same on all nodes of the cluster
  43  * for the DLM to function.
  44  *
  45  */
  46
  47 #include <asm/ioctls.h>
  48 #include <net/sock.h>
  49 #include <net/tcp.h>
  50 #include <linux/pagemap.h>
  51 #include <linux/file.h>
  52 #include <linux/mutex.h>
  53 #include <linux/sctp.h>
  54 #include <linux/slab.h>
  55 #include <net/sctp/sctp.h>
  56 #include <net/ipv6.h>
  57
  58 #include "dlm_internal.h"
  59 #include "lowcomms.h"
  60 #include "midcomms.h"
  61 #include "config.h"
  62
  63 #define NEEDED_RMEM (4*1024*1024)
  64 #define CONN_HASH_SIZE 32
  65
  66 /* Number of messages to send before rescheduling */
  67 #define MAX_SEND_MSG_COUNT 25
  68
  69 struct cbuf {
  70         unsigned int base;
  71         unsigned int len;
  72         unsigned int mask;
  73 };
  74
  75 static void cbuf_add(struct cbuf *cb, int n)
  76 {
  77         cb->len += n;
  78 }
  79
  80 static int cbuf_data(struct cbuf *cb)
  81 {
  82         return ((cb->base + cb->len) & cb->mask);
  83 }
  84
  85 static void cbuf_init(struct cbuf *cb, int size)
  86 {
  87         cb->base = cb->len = 0;
  88         cb->mask = size-1;
  89 }
  90
  91 static void cbuf_eat(struct cbuf *cb, int n)
  92 {
  93         cb->len  -= n;
  94         cb->base += n;
  95         cb->base &= cb->mask;
  96 }
  97
  98 static bool cbuf_empty(struct cbuf *cb)
  99 {
 100         return cb->len == 0;
 101 }
 102
 103 struct connection {
 104         struct socket *sock;    /* NULL if not connected */
 105         uint32_t nodeid;        /* So we know who we are in the list */
 106         struct mutex sock_mutex;
 107         unsigned long flags;
 108 #define CF_READ_PENDING 1
 109 #define CF_WRITE_PENDING 2
 110 #define CF_CONNECT_PENDING 3
 111 #define CF_INIT_PENDING 4
 112 #define CF_IS_OTHERCON 5
 113 #define CF_CLOSE 6
 114 #define CF_APP_LIMITED 7
 115         struct list_head writequeue;  /* List of outgoing writequeue_entries */
 116         spinlock_t writequeue_lock;
 117         int (*rx_action) (struct connection *); /* What to do when active */
 118         void (*connect_action) (struct connection *);   /* What to do to connect */
 119         struct page *rx_page;
 120         struct cbuf cb;
 121         int retries;
 122 #define MAX_CONNECT_RETRIES 3
 123         int sctp_assoc;
 124         struct hlist_node list;
 125         struct connection *othercon;
 126         struct work_struct rwork; /* Receive workqueue */
 127         struct work_struct swork; /* Send workqueue */
 128         bool try_new_addr;
 129 };
 130 #define sock2con(x) ((struct connection *)(x)->sk_user_data)
 131
 132 /* An entry waiting to be sent */
 133 struct writequeue_entry {
 134         struct list_head list;
 135         struct page *page;
 136         int offset;
 137         int len;
 138         int end;
 139         int users;
 140         struct connection *con;
 141 };
 142
 143 struct dlm_node_addr {
 144         struct list_head list;
 145         int nodeid;
 146         int addr_count;
 147         int curr_addr_index;
 148         struct sockaddr_storage *addr[DLM_MAX_ADDR_COUNT];
 149 };
 150
 151 static LIST_HEAD(dlm_node_addrs);
 152 static DEFINE_SPINLOCK(dlm_node_addrs_spin);
 153
 154 static struct sockaddr_storage *dlm_local_addr[DLM_MAX_ADDR_COUNT];
 155 static int dlm_local_count;
 156 static int dlm_allow_conn;
 157
 158 /* Work queues */
 159 static struct workqueue_struct *recv_workqueue;
 160 static struct workqueue_struct *send_workqueue;
 161
 162 static struct hlist_head connection_hash[CONN_HASH_SIZE];
 163 static DEFINE_MUTEX(connections_lock);
 164 static struct kmem_cache *con_cache;
 165
 166 static void process_recv_sockets(struct work_struct *work);
 167 static void process_send_sockets(struct work_struct *work);
 168
 169
 170 /* This is deliberately very simple because most clusters have simple
 171    sequential nodeids, so we should be able to go straight to a connection
 172    struct in the array */
 173 static inline int nodeid_hash(int nodeid)
 174 {
 175         return nodeid & (CONN_HASH_SIZE-1);
 176 }
 177
 178 static struct connection *__find_con(int nodeid)
 179 {
 180         int r;
 181         struct connection *con;
 182
 183         r = nodeid_hash(nodeid);
 184
 185         hlist_for_each_entry(con, &connection_hash[r], list) {
 186                 if (con->nodeid == nodeid)
 187                         return con;
 188         }
 189         return NULL;
 190 }
 191
 192 /*
 193  * If 'allocation' is zero then we don't attempt to create a new
 194  * connection structure for this node.
 195  */
 196 static struct connection *__nodeid2con(int nodeid, gfp_t alloc)
 197 {
 198         struct connection *con = NULL;
 199         int r;
 200
 201         con = __find_con(nodeid);
 202         if (con || !alloc)
 203                 return con;
 204
 205         con = kmem_cache_zalloc(con_cache, alloc);
 206         if (!con)
 207                 return NULL;
 208
 209         r = nodeid_hash(nodeid);
 210         hlist_add_head(&con->list, &connection_hash[r]);
 211
 212         con->nodeid = nodeid;
 213         mutex_init(&con->sock_mutex);
 214         INIT_LIST_HEAD(&con->writequeue);
 215         spin_lock_init(&con->writequeue_lock);
 216         INIT_WORK(&con->swork, process_send_sockets);
 217         INIT_WORK(&con->rwork, process_recv_sockets);
 218
 219         /* Setup action pointers for child sockets */
 220         if (con->nodeid) {
 221                 struct connection *zerocon = __find_con(0);
 222
 223                 con->connect_action = zerocon->connect_action;
 224                 if (!con->rx_action)
 225                         con->rx_action = zerocon->rx_action;
 226         }
 227
 228         return con;
 229 }
 230
 231 /* Loop round all connections */
 232 static void foreach_conn(void (*conn_func)(struct connection *c))
 233 {
 234         int i;
 235         struct hlist_node *n;
 236         struct connection *con;
 237
 238         for (i = 0; i < CONN_HASH_SIZE; i++) {
 239                 hlist_for_each_entry_safe(con, n, &connection_hash[i], list)
 240                         conn_func(con);
 241         }
 242 }
 243
 244 static struct connection *nodeid2con(int nodeid, gfp_t allocation)
 245 {
 246         struct connection *con;
 247
 248         mutex_lock(&connections_lock);
 249         con = __nodeid2con(nodeid, allocation);
 250         mutex_unlock(&connections_lock);
 251
 252         return con;
 253 }
 254
 255 /* This is a bit drastic, but only called when things go wrong */
 256 static struct connection *assoc2con(int assoc_id)
 257 {
 258         int i;
 259         struct connection *con;
 260
 261         mutex_lock(&connections_lock);
 262
 263         for (i = 0 ; i < CONN_HASH_SIZE; i++) {
 264                 hlist_for_each_entry(con, &connection_hash[i], list) {
 265                         if (con->sctp_assoc == assoc_id) {
 266                                 mutex_unlock(&connections_lock);
 267                                 return con;
 268                         }
 269                 }
 270         }
 271         mutex_unlock(&connections_lock);
 272         return NULL;
 273 }
 274
 275 static struct dlm_node_addr *find_node_addr(int nodeid)
 276 {
 277         struct dlm_node_addr *na;
 278
 279         list_for_each_entry(na, &dlm_node_addrs, list) {
 280                 if (na->nodeid == nodeid)
 281                         return na;
 282         }
 283         return NULL;
 284 }
 285
 286 static int addr_compare(struct sockaddr_storage *x, struct sockaddr_storage *y)
 287 {
 288         switch (x->ss_family) {
 289         case AF_INET: {
 290                 struct sockaddr_in *sinx = (struct sockaddr_in *)x;
 291                 struct sockaddr_in *siny = (struct sockaddr_in *)y;
 292                 if (sinx->sin_addr.s_addr != siny->sin_addr.s_addr)
 293                         return 0;
 294                 if (sinx->sin_port != siny->sin_port)
 295                         return 0;
 296                 break;
 297         }
 298         case AF_INET6: {
 299                 struct sockaddr_in6 *sinx = (struct sockaddr_in6 *)x;
 300                 struct sockaddr_in6 *siny = (struct sockaddr_in6 *)y;
 301                 if (!ipv6_addr_equal(&sinx->sin6_addr, &siny->sin6_addr))
 302                         return 0;
 303                 if (sinx->sin6_port != siny->sin6_port)
 304                         return 0;
 305                 break;
 306         }
 307         default:
 308                 return 0;
 309         }
 310         return 1;
 311 }
 312
 313 static int nodeid_to_addr(int nodeid, struct sockaddr_storage *sas_out,
 314                           struct sockaddr *sa_out, bool try_new_addr)
 315 {
 316         struct sockaddr_storage sas;
 317         struct dlm_node_addr *na;
 318
 319         if (!dlm_local_count)
 320                 return -1;
 321
 322         spin_lock(&dlm_node_addrs_spin);
 323         na = find_node_addr(nodeid);
 324         if (na && na->addr_count) {
 325                 if (try_new_addr) {
 326                         na->curr_addr_index++;
 327                         if (na->curr_addr_index == na->addr_count)
 328                                 na->curr_addr_index = 0;
 329                 }
 330
 331                 memcpy(&sas, na->addr[na->curr_addr_index ],
 332                         sizeof(struct sockaddr_storage));
 333         }
 334         spin_unlock(&dlm_node_addrs_spin);
 335
 336         if (!na)
 337                 return -EEXIST;
 338
 339         if (!na->addr_count)
 340                 return -ENOENT;
 341
 342         if (sas_out)
 343                 memcpy(sas_out, &sas, sizeof(struct sockaddr_storage));
 344
 345         if (!sa_out)
 346                 return 0;
 347
 348         if (dlm_local_addr[0]->ss_family == AF_INET) {
 349                 struct sockaddr_in *in4  = (struct sockaddr_in *) &sas;
 350                 struct sockaddr_in *ret4 = (struct sockaddr_in *) sa_out;
 351                 ret4->sin_addr.s_addr = in4->sin_addr.s_addr;
 352         } else {
 353                 struct sockaddr_in6 *in6  = (struct sockaddr_in6 *) &sas;
 354                 struct sockaddr_in6 *ret6 = (struct sockaddr_in6 *) sa_out;
 355                 ret6->sin6_addr = in6->sin6_addr;
 356         }
 357
 358         return 0;
 359 }
 360
 361 static int addr_to_nodeid(struct sockaddr_storage *addr, int *nodeid)
 362 {
 363         struct dlm_node_addr *na;
 364         int rv = -EEXIST;
 365         int addr_i;
 366
 367         spin_lock(&dlm_node_addrs_spin);
 368         list_for_each_entry(na, &dlm_node_addrs, list) {
 369                 if (!na->addr_count)
 370                         continue;
 371
 372                 for (addr_i = 0; addr_i < na->addr_count; addr_i++) {
 373                         if (addr_compare(na->addr[addr_i], addr)) {
 374                                 *nodeid = na->nodeid;
 375                                 rv = 0;
 376                                 goto unlock;
 377                         }
 378                 }
 379         }
 380 unlock:
 381         spin_unlock(&dlm_node_addrs_spin);
 382         return rv;
 383 }
 384
 385 int dlm_lowcomms_addr(int nodeid, struct sockaddr_storage *addr, int len)
 386 {
 387         struct sockaddr_storage *new_addr;
 388         struct dlm_node_addr *new_node, *na;
 389
 390         new_node = kzalloc(sizeof(struct dlm_node_addr), GFP_NOFS);
 391         if (!new_node)
 392                 return -ENOMEM;
 393
 394         new_addr = kzalloc(sizeof(struct sockaddr_storage), GFP_NOFS);
 395         if (!new_addr) {
 396                 kfree(new_node);
 397                 return -ENOMEM;
 398         }
 399
 400         memcpy(new_addr, addr, len);
 401
 402         spin_lock(&dlm_node_addrs_spin);
 403         na = find_node_addr(nodeid);
 404         if (!na) {
 405                 new_node->nodeid = nodeid;
 406                 new_node->addr[0] = new_addr;
 407                 new_node->addr_count = 1;
 408                 list_add(&new_node->list, &dlm_node_addrs);
 409                 spin_unlock(&dlm_node_addrs_spin);
 410                 return 0;
 411         }
 412
 413         if (na->addr_count >= DLM_MAX_ADDR_COUNT) {
 414                 spin_unlock(&dlm_node_addrs_spin);
 415                 kfree(new_addr);
 416                 kfree(new_node);
 417                 return -ENOSPC;
 418         }
 419
 420         na->addr[na->addr_count++] = new_addr;
 421         spin_unlock(&dlm_node_addrs_spin);
 422         kfree(new_node);
 423         return 0;
 424 }
 425
 426 /* Data available on socket or listen socket received a connect */
 427 static void lowcomms_data_ready(struct sock *sk, int count_unused)
 428 {
 429         struct connection *con = sock2con(sk);
 430         if (con && !test_and_set_bit(CF_READ_PENDING, &con->flags))
 431                 queue_work(recv_workqueue, &con->rwork);
 432 }
 433
 434 static void lowcomms_write_space(struct sock *sk)
 435 {
 436         struct connection *con = sock2con(sk);
 437
 438         if (!con)
 439                 return;
 440
 441         clear_bit(SOCK_NOSPACE, &con->sock->flags);
 442
 443         if (test_and_clear_bit(CF_APP_LIMITED, &con->flags)) {
 444                 con->sock->sk->sk_write_pending--;
 445                 clear_bit(SOCK_ASYNC_NOSPACE, &con->sock->flags);
 446         }
 447
 448         if (!test_and_set_bit(CF_WRITE_PENDING, &con->flags))
 449                 queue_work(send_workqueue, &con->swork);
 450 }
 451
 452 static inline void lowcomms_connect_sock(struct connection *con)
 453 {
 454         if (test_bit(CF_CLOSE, &con->flags))
 455                 return;
 456         if (!test_and_set_bit(CF_CONNECT_PENDING, &con->flags))
 457                 queue_work(send_workqueue, &con->swork);
 458 }
 459
 460 static void lowcomms_state_change(struct sock *sk)
 461 {
 462         if (sk->sk_state == TCP_ESTABLISHED)
 463                 lowcomms_write_space(sk);
 464 }
 465
 466 int dlm_lowcomms_connect_node(int nodeid)
 467 {
 468         struct connection *con;
 469
 470         /* with sctp there's no connecting without sending */
 471         if (dlm_config.ci_protocol != 0)
 472                 return 0;
 473
 474         if (nodeid == dlm_our_nodeid())
 475                 return 0;
 476
 477         con = nodeid2con(nodeid, GFP_NOFS);
 478         if (!con)
 479                 return -ENOMEM;
 480         lowcomms_connect_sock(con);
 481         return 0;
 482 }
 483
 484 /* Make a socket active */
 485 static void add_sock(struct socket *sock, struct connection *con)
 486 {
 487         con->sock = sock;
 488
 489         /* Install a data_ready callback */
 490         con->sock->sk->sk_data_ready = lowcomms_data_ready;
 491         con->sock->sk->sk_write_space = lowcomms_write_space;
 492         con->sock->sk->sk_state_change = lowcomms_state_change;
 493         con->sock->sk->sk_user_data = con;
 494         con->sock->sk->sk_allocation = GFP_NOFS;
 495 }
 496
 497 /* Add the port number to an IPv6 or 4 sockaddr and return the address
 498    length */
 499 static void make_sockaddr(struct sockaddr_storage *saddr, uint16_t port,
 500                           int *addr_len)
 501 {
 502         saddr->ss_family =  dlm_local_addr[0]->ss_family;
 503         if (saddr->ss_family == AF_INET) {
 504                 struct sockaddr_in *in4_addr = (struct sockaddr_in *)saddr;
 505                 in4_addr->sin_port = cpu_to_be16(port);
 506                 *addr_len = sizeof(struct sockaddr_in);
 507                 memset(&in4_addr->sin_zero, 0, sizeof(in4_addr->sin_zero));
 508         } else {
 509                 struct sockaddr_in6 *in6_addr = (struct sockaddr_in6 *)saddr;
 510                 in6_addr->sin6_port = cpu_to_be16(port);
 511                 *addr_len = sizeof(struct sockaddr_in6);
 512         }
 513         memset((char *)saddr + *addr_len, 0, sizeof(struct sockaddr_storage) - *addr_len);
 514 }
 515
 516 /* Close a remote connection and tidy up */
 517 static void close_connection(struct connection *con, bool and_other)
 518 {
 519         mutex_lock(&con->sock_mutex);
 520
 521         if (con->sock) {
 522                 sock_release(con->sock);
 523                 con->sock = NULL;
 524         }
 525         if (con->othercon && and_other) {
 526                 /* Will only re-enter once. */
 527                 close_connection(con->othercon, false);
 528         }
 529         if (con->rx_page) {
 530                 __free_page(con->rx_page);
 531                 con->rx_page = NULL;
 532         }
 533
 534         con->retries = 0;
 535         mutex_unlock(&con->sock_mutex);
 536 }
 537
 538 /* We only send shutdown messages to nodes that are not part of the cluster */
 539 static void sctp_send_shutdown(sctp_assoc_t associd)
 540 {
 541         static char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))];
 542         struct msghdr outmessage;
 543         struct cmsghdr *cmsg;
 544         struct sctp_sndrcvinfo *sinfo;
 545         int ret;
 546         struct connection *con;
 547
 548         con = nodeid2con(0,0);
 549         BUG_ON(con == NULL);
 550
 551         outmessage.msg_name = NULL;
 552         outmessage.msg_namelen = 0;
 553         outmessage.msg_control = outcmsg;
 554         outmessage.msg_controllen = sizeof(outcmsg);
 555         outmessage.msg_flags = MSG_EOR;
 556
 557         cmsg = CMSG_FIRSTHDR(&outmessage);
 558         cmsg->cmsg_level = IPPROTO_SCTP;
 559         cmsg->cmsg_type = SCTP_SNDRCV;
 560         cmsg->cmsg_len = CMSG_LEN(sizeof(struct sctp_sndrcvinfo));
 561         outmessage.msg_controllen = cmsg->cmsg_len;
 562         sinfo = CMSG_DATA(cmsg);
 563         memset(sinfo, 0x00, sizeof(struct sctp_sndrcvinfo));
 564
 565         sinfo->sinfo_flags |= MSG_EOF;
 566         sinfo->sinfo_assoc_id = associd;
 567
 568         ret = kernel_sendmsg(con->sock, &outmessage, NULL, 0, 0);
 569
 570         if (ret != 0)
 571                 log_print("send EOF to node failed: %d", ret);
 572 }
 573
 574 static void sctp_init_failed_foreach(struct connection *con)
 575 {
 576
 577         /*
 578          * Don't try to recover base con and handle race where the
 579          * other node's assoc init creates a assoc and we get that
 580          * notification, then we get a notification that our attempt
 581          * failed due. This happens when we are still trying the primary
 582          * address, but the other node has already tried secondary addrs
 583          * and found one that worked.
 584          */
 585         if (!con->nodeid || con->sctp_assoc)
 586                 return;
 587
 588         log_print("Retrying SCTP association init for node %d\n", con->nodeid);
 589
 590         con->try_new_addr = true;
 591         con->sctp_assoc = 0;
 592         if (test_and_clear_bit(CF_INIT_PENDING, &con->flags)) {
 593                 if (!test_and_set_bit(CF_WRITE_PENDING, &con->flags))
 594                         queue_work(send_workqueue, &con->swork);
 595         }
 596 }
 597
 598 /* INIT failed but we don't know which node...
 599    restart INIT on all pending nodes */
 600 static void sctp_init_failed(void)
 601 {
 602         mutex_lock(&connections_lock);
 603
 604         foreach_conn(sctp_init_failed_foreach);
 605
 606         mutex_unlock(&connections_lock);
 607 }
 608
 609 static void retry_failed_sctp_send(struct connection *recv_con,
 610                                    struct sctp_send_failed *sn_send_failed,
 611                                    char *buf)
 612 {
 613         int len = sn_send_failed->ssf_length - sizeof(struct sctp_send_failed);
 614         struct dlm_mhandle *mh;
 615         struct connection *con;
 616         char *retry_buf;
 617         int nodeid = sn_send_failed->ssf_info.sinfo_ppid;
 618
 619         log_print("Retry sending %d bytes to node id %d", len, nodeid);
 620
 621         con = nodeid2con(nodeid, 0);
 622         if (!con) {
 623                 log_print("Could not look up con for nodeid %d\n",
 624                           nodeid);
 625                 return;
 626         }
 627
 628         mh = dlm_lowcomms_get_buffer(nodeid, len, GFP_NOFS, &retry_buf);
 629         if (!mh) {
 630                 log_print("Could not allocate buf for retry.");
 631                 return;
 632         }
 633         memcpy(retry_buf, buf + sizeof(struct sctp_send_failed), len);
 634         dlm_lowcomms_commit_buffer(mh);
 635
 636         /*
 637          * If we got a assoc changed event before the send failed event then
 638          * we only need to retry the send.
 639          */
 640         if (con->sctp_assoc) {
 641                 if (!test_and_set_bit(CF_WRITE_PENDING, &con->flags))
 642                         queue_work(send_workqueue, &con->swork);
 643         } else
 644                 sctp_init_failed_foreach(con);
 645 }
 646
 647 /* Something happened to an association */
 648 static void process_sctp_notification(struct connection *con,
 649                                       struct msghdr *msg, char *buf)
 650 {
 651         union sctp_notification *sn = (union sctp_notification *)buf;
 652
 653         switch (sn->sn_header.sn_type) {
 654         case SCTP_SEND_FAILED:
 655                 retry_failed_sctp_send(con, &sn->sn_send_failed, buf);
 656                 break;
 657         case SCTP_ASSOC_CHANGE:
 658                 switch (sn->sn_assoc_change.sac_state) {
 659                 case SCTP_COMM_UP:
 660                 case SCTP_RESTART:
 661                 {
 662                         /* Check that the new node is in the lockspace */
 663                         struct sctp_prim prim;
 664                         int nodeid;
 665                         int prim_len, ret;
 666                         int addr_len;
 667                         struct connection *new_con;
 668
 669                         /*
 670                          * We get this before any data for an association.
 671                          * We verify that the node is in the cluster and
 672                          * then peel off a socket for it.
 673                          */
 674                         if ((int)sn->sn_assoc_change.sac_assoc_id <= 0) {
 675                                 log_print("COMM_UP for invalid assoc ID %d",
 676                                          (int)sn->sn_assoc_change.sac_assoc_id);
 677                                 sctp_init_failed();
 678                                 return;
 679                         }
 680                         memset(&prim, 0, sizeof(struct sctp_prim));
 681                         prim_len = sizeof(struct sctp_prim);
 682                         prim.ssp_assoc_id = sn->sn_assoc_change.sac_assoc_id;
 683
 684                         ret = kernel_getsockopt(con->sock,
 685                                                 IPPROTO_SCTP,
 686                                                 SCTP_PRIMARY_ADDR,
 687                                                 (char*)&prim,
 688                                                 &prim_len);
 689                         if (ret < 0) {
 690                                 log_print("getsockopt/sctp_primary_addr on "
 691                                           "new assoc %d failed : %d",
 692                                           (int)sn->sn_assoc_change.sac_assoc_id,
 693                                           ret);
 694
 695                                 /* Retry INIT later */
 696                                 new_con = assoc2con(sn->sn_assoc_change.sac_assoc_id);
 697                                 if (new_con)
 698                                         clear_bit(CF_CONNECT_PENDING, &con->flags);
 699                                 return;
 700                         }
 701                         make_sockaddr(&prim.ssp_addr, 0, &addr_len);
 702                         if (addr_to_nodeid(&prim.ssp_addr, &nodeid)) {
 703                                 unsigned char *b=(unsigned char *)&prim.ssp_addr;
 704                                 log_print("reject connect from unknown addr");
 705                                 print_hex_dump_bytes("ss: ", DUMP_PREFIX_NONE,
 706                                                      b, sizeof(struct sockaddr_storage));
 707                                 sctp_send_shutdown(prim.ssp_assoc_id);
 708                                 return;
 709                         }
 710
 711                         new_con = nodeid2con(nodeid, GFP_NOFS);
 712                         if (!new_con)
 713                                 return;
 714
 715                         /* Peel off a new sock */
 716                         sctp_lock_sock(con->sock->sk);
 717                         ret = sctp_do_peeloff(con->sock->sk,
 718                                 sn->sn_assoc_change.sac_assoc_id,
 719                                 &new_con->sock);
 720                         sctp_release_sock(con->sock->sk);
 721                         if (ret < 0) {
 722                                 log_print("Can't peel off a socket for "
 723                                           "connection %d to node %d: err=%d",
 724                                           (int)sn->sn_assoc_change.sac_assoc_id,
 725                                           nodeid, ret);
 726                                 return;
 727                         }
 728                         add_sock(new_con->sock, new_con);
 729
 730                         log_print("connecting to %d sctp association %d",
 731                                  nodeid, (int)sn->sn_assoc_change.sac_assoc_id);
 732
 733                         new_con->sctp_assoc = sn->sn_assoc_change.sac_assoc_id;
 734                         new_con->try_new_addr = false;
 735                         /* Send any pending writes */
 736                         clear_bit(CF_CONNECT_PENDING, &new_con->flags);
 737                         clear_bit(CF_INIT_PENDING, &new_con->flags);
 738                         if (!test_and_set_bit(CF_WRITE_PENDING, &new_con->flags)) {
 739                                 queue_work(send_workqueue, &new_con->swork);
 740                         }
 741                         if (!test_and_set_bit(CF_READ_PENDING, &new_con->flags))
 742                                 queue_work(recv_workqueue, &new_con->rwork);
 743                 }
 744                 break;
 745
 746                 case SCTP_COMM_LOST:
 747                 case SCTP_SHUTDOWN_COMP:
 748                 {
 749                         con = assoc2con(sn->sn_assoc_change.sac_assoc_id);
 750                         if (con) {
 751                                 con->sctp_assoc = 0;
 752                         }
 753                 }
 754                 break;
 755
 756                 case SCTP_CANT_STR_ASSOC:
 757                 {
 758                         /* Will retry init when we get the send failed notification */
 759                         log_print("Can't start SCTP association - retrying");
 760                 }
 761                 break;
 762
 763                 default:
 764                         log_print("unexpected SCTP assoc change id=%d state=%d",
 765                                   (int)sn->sn_assoc_change.sac_assoc_id,
 766                                   sn->sn_assoc_change.sac_state);
 767                 }
 768         default:
 769                 ; /* fall through */
 770         }
 771 }
 772
 773 /* Data received from remote end */
 774 static int receive_from_sock(struct connection *con)
 775 {
 776         int ret = 0;
 777         struct msghdr msg = {};
 778         struct kvec iov[2];
 779         unsigned len;
 780         int r;
 781         int call_again_soon = 0;
 782         int nvec;
 783         char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))];
 784
 785         mutex_lock(&con->sock_mutex);
 786
 787         if (con->sock == NULL) {
 788                 ret = -EAGAIN;
 789                 goto out_close;
 790         }
 791
 792         if (con->rx_page == NULL) {
 793                 /*
 794                  * This doesn't need to be atomic, but I think it should
 795                  * improve performance if it is.
 796                  */
 797                 con->rx_page = alloc_page(GFP_ATOMIC);
 798                 if (con->rx_page == NULL)
 799                         goto out_resched;
 800                 cbuf_init(&con->cb, PAGE_CACHE_SIZE);
 801         }
 802
 803         /* Only SCTP needs these really */
 804         memset(&incmsg, 0, sizeof(incmsg));
 805         msg.msg_control = incmsg;
 806         msg.msg_controllen = sizeof(incmsg);
 807
 808         /*
 809          * iov[0] is the bit of the circular buffer between the current end
 810          * point (cb.base + cb.len) and the end of the buffer.
 811          */
 812         iov[0].iov_len = con->cb.base - cbuf_data(&con->cb);
 813         iov[0].iov_base = page_address(con->rx_page) + cbuf_data(&con->cb);
 814         iov[1].iov_len = 0;
 815         nvec = 1;
 816
 817         /*
 818          * iov[1] is the bit of the circular buffer between the start of the
 819          * buffer and the start of the currently used section (cb.base)
 820          */
 821         if (cbuf_data(&con->cb) >= con->cb.base) {
 822                 iov[0].iov_len = PAGE_CACHE_SIZE - cbuf_data(&con->cb);
 823                 iov[1].iov_len = con->cb.base;
 824                 iov[1].iov_base = page_address(con->rx_page);
 825                 nvec = 2;
 826         }
 827         len = iov[0].iov_len + iov[1].iov_len;
 828
 829         r = ret = kernel_recvmsg(con->sock, &msg, iov, nvec, len,
 830                                MSG_DONTWAIT | MSG_NOSIGNAL);
 831         if (ret <= 0)
 832                 goto out_close;
 833
 834         /* Process SCTP notifications */
 835         if (msg.msg_flags & MSG_NOTIFICATION) {
 836                 msg.msg_control = incmsg;
 837                 msg.msg_controllen = sizeof(incmsg);
 838
 839                 process_sctp_notification(con, &msg,
 840                                 page_address(con->rx_page) + con->cb.base);
 841                 mutex_unlock(&con->sock_mutex);
 842                 return 0;
 843         }
 844         BUG_ON(con->nodeid == 0);
 845
 846         if (ret == len)
 847                 call_again_soon = 1;
 848         cbuf_add(&con->cb, ret);
 849         ret = dlm_process_incoming_buffer(con->nodeid,
 850                                           page_address(con->rx_page),
 851                                           con->cb.base, con->cb.len,
 852                                           PAGE_CACHE_SIZE);
 853         if (ret == -EBADMSG) {
 854                 log_print("lowcomms: addr=%p, base=%u, len=%u, "
 855                           "iov_len=%u, iov_base[0]=%p, read=%d",
 856                           page_address(con->rx_page), con->cb.base, con->cb.len,
 857                           len, iov[0].iov_base, r);
 858         }
 859         if (ret < 0)
 860                 goto out_close;
 861         cbuf_eat(&con->cb, ret);
 862
 863         if (cbuf_empty(&con->cb) && !call_again_soon) {
 864                 __free_page(con->rx_page);
 865                 con->rx_page = NULL;
 866         }
 867
 868         if (call_again_soon)
 869                 goto out_resched;
 870         mutex_unlock(&con->sock_mutex);
 871         return 0;
 872
 873 out_resched:
 874         if (!test_and_set_bit(CF_READ_PENDING, &con->flags))
 875                 queue_work(recv_workqueue, &con->rwork);
 876         mutex_unlock(&con->sock_mutex);
 877         return -EAGAIN;
 878
 879 out_close:
 880         mutex_unlock(&con->sock_mutex);
 881         if (ret != -EAGAIN) {
 882                 close_connection(con, false);
 883                 /* Reconnect when there is something to send */
 884         }
 885         /* Don't return success if we really got EOF */
 886         if (ret == 0)
 887                 ret = -EAGAIN;
 888
 889         return ret;
 890 }
 891
 892 /* Listening socket is busy, accept a connection */
 893 static int tcp_accept_from_sock(struct connection *con)
 894 {
 895         int result;
 896         struct sockaddr_storage peeraddr;
 897         struct socket *newsock;
 898         int len;
 899         int nodeid;
 900         struct connection *newcon;
 901         struct connection *addcon;
 902
 903         mutex_lock(&connections_lock);
 904         if (!dlm_allow_conn) {
 905                 mutex_unlock(&connections_lock);
 906                 return -1;
 907         }
 908         mutex_unlock(&connections_lock);
 909
 910         memset(&peeraddr, 0, sizeof(peeraddr));
 911         result = sock_create_kern(dlm_local_addr[0]->ss_family, SOCK_STREAM,
 912                                   IPPROTO_TCP, &newsock);
 913         if (result < 0)
 914                 return -ENOMEM;
 915
 916         mutex_lock_nested(&con->sock_mutex, 0);
 917
 918         result = -ENOTCONN;
 919         if (con->sock == NULL)
 920                 goto accept_err;
 921
 922         newsock->type = con->sock->type;
 923         newsock->ops = con->sock->ops;
 924
 925         result = con->sock->ops->accept(con->sock, newsock, O_NONBLOCK);
 926         if (result < 0)
 927                 goto accept_err;
 928
 929         /* Get the connected socket's peer */
 930         memset(&peeraddr, 0, sizeof(peeraddr));
 931         if (newsock->ops->getname(newsock, (struct sockaddr *)&peeraddr,
 932                                   &len, 2)) {
 933                 result = -ECONNABORTED;
 934                 goto accept_err;
 935         }
 936
 937         /* Get the new node's NODEID */
 938         make_sockaddr(&peeraddr, 0, &len);
 939         if (addr_to_nodeid(&peeraddr, &nodeid)) {
 940                 unsigned char *b=(unsigned char *)&peeraddr;
 941                 log_print("connect from non cluster node");
 942                 print_hex_dump_bytes("ss: ", DUMP_PREFIX_NONE,
 943                                      b, sizeof(struct sockaddr_storage));
 944                 sock_release(newsock);
 945                 mutex_unlock(&con->sock_mutex);
 946                 return -1;
 947         }
 948
 949         log_print("got connection from %d", nodeid);
 950
 951         /*  Check to see if we already have a connection to this node. This
 952          *  could happen if the two nodes initiate a connection at roughly
 953          *  the same time and the connections cross on the wire.
 954          *  In this case we store the incoming one in "othercon"
 955          */
 956         newcon = nodeid2con(nodeid, GFP_NOFS);
 957         if (!newcon) {
 958                 result = -ENOMEM;
 959                 goto accept_err;
 960         }
 961         mutex_lock_nested(&newcon->sock_mutex, 1);
 962         if (newcon->sock) {
 963                 struct connection *othercon = newcon->othercon;
 964
 965                 if (!othercon) {
 966                         othercon = kmem_cache_zalloc(con_cache, GFP_NOFS);
 967                         if (!othercon) {
 968                                 log_print("failed to allocate incoming socket");
 969                                 mutex_unlock(&newcon->sock_mutex);
 970                                 result = -ENOMEM;
 971                                 goto accept_err;
 972                         }
 973                         othercon->nodeid = nodeid;
 974                         othercon->rx_action = receive_from_sock;
 975                         mutex_init(&othercon->sock_mutex);
 976                         INIT_WORK(&othercon->swork, process_send_sockets);
 977                         INIT_WORK(&othercon->rwork, process_recv_sockets);
 978                         set_bit(CF_IS_OTHERCON, &othercon->flags);
 979                 }
 980                 if (!othercon->sock) {
 981                         newcon->othercon = othercon;
 982                         othercon->sock = newsock;
 983                         newsock->sk->sk_user_data = othercon;
 984                         add_sock(newsock, othercon);
 985                         addcon = othercon;
 986                 }
 987                 else {
 988                         printk("Extra connection from node %d attempted\n", nodeid);
 989                         result = -EAGAIN;
 990                         mutex_unlock(&newcon->sock_mutex);
 991                         goto accept_err;
 992                 }
 993         }
 994         else {
 995                 newsock->sk->sk_user_data = newcon;
 996                 newcon->rx_action = receive_from_sock;
 997                 add_sock(newsock, newcon);
 998                 addcon = newcon;
 999         }
1000
1001         mutex_unlock(&newcon->sock_mutex);
1002
1003         /*
1004          * Add it to the active queue in case we got data
1005          * between processing the accept adding the socket
1006          * to the read_sockets list
1007          */
1008         if (!test_and_set_bit(CF_READ_PENDING, &addcon->flags))
1009                 queue_work(recv_workqueue, &addcon->rwork);
1010         mutex_unlock(&con->sock_mutex);
1011
1012         return 0;
1013
1014 accept_err:
1015         mutex_unlock(&con->sock_mutex);
1016         sock_release(newsock);
1017
1018         if (result != -EAGAIN)
1019                 log_print("error accepting connection from node: %d", result);
1020         return result;
1021 }
1022
1023 static void free_entry(struct writequeue_entry *e)
1024 {
1025         __free_page(e->page);
1026         kfree(e);
1027 }
1028
1029 /*
1030  * writequeue_entry_complete - try to delete and free write queue entry
1031  * @e: write queue entry to try to delete
1032  * @completed: bytes completed
1033  *
1034  * writequeue_lock must be held.
1035  */
1036 static void writequeue_entry_complete(struct writequeue_entry *e, int completed)
1037 {
1038         e->offset += completed;
1039         e->len -= completed;
1040
1041         if (e->len == 0 && e->users == 0) {
1042                 list_del(&e->list);
1043                 free_entry(e);
1044         }
1045 }
1046
1047 /* Initiate an SCTP association.
1048    This is a special case of send_to_sock() in that we don't yet have a
1049    peeled-off socket for this association, so we use the listening socket
1050    and add the primary IP address of the remote node.
1051  */
1052 static void sctp_init_assoc(struct connection *con)
1053 {
1054         struct sockaddr_storage rem_addr;
1055         char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))];
1056         struct msghdr outmessage;
1057         struct cmsghdr *cmsg;
1058         struct sctp_sndrcvinfo *sinfo;
1059         struct connection *base_con;
1060         struct writequeue_entry *e;
1061         int len, offset;
1062         int ret;
1063         int addrlen;
1064         struct kvec iov[1];
1065
1066         mutex_lock(&con->sock_mutex);
1067         if (test_and_set_bit(CF_INIT_PENDING, &con->flags))
1068                 goto unlock;
1069
1070         if (nodeid_to_addr(con->nodeid, NULL, (struct sockaddr *)&rem_addr,
1071                            con->try_new_addr)) {
1072                 log_print("no address for nodeid %d", con->nodeid);
1073                 goto unlock;
1074         }
1075         base_con = nodeid2con(0, 0);
1076         BUG_ON(base_con == NULL);
1077
1078         make_sockaddr(&rem_addr, dlm_config.ci_tcp_port, &addrlen);
1079
1080         outmessage.msg_name = &rem_addr;
1081         outmessage.msg_namelen = addrlen;
1082         outmessage.msg_control = outcmsg;
1083         outmessage.msg_controllen = sizeof(outcmsg);
1084         outmessage.msg_flags = MSG_EOR;
1085
1086         spin_lock(&con->writequeue_lock);
1087
1088         if (list_empty(&con->writequeue)) {
1089                 spin_unlock(&con->writequeue_lock);
1090                 log_print("writequeue empty for nodeid %d", con->nodeid);
1091                 goto unlock;
1092         }
1093
1094         e = list_first_entry(&con->writequeue, struct writequeue_entry, list);
1095         len = e->len;
1096         offset = e->offset;
1097
1098         /* Send the first block off the write queue */
1099         iov[0].iov_base = page_address(e->page)+offset;
1100         iov[0].iov_len = len;
1101         spin_unlock(&con->writequeue_lock);
1102
1103         if (rem_addr.ss_family == AF_INET) {
1104                 struct sockaddr_in *sin = (struct sockaddr_in *)&rem_addr;
1105                 log_print("Trying to connect to %pI4", &sin->sin_addr.s_addr);
1106         } else {
1107                 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)&rem_addr;
1108                 log_print("Trying to connect to %pI6", &sin6->sin6_addr);
1109         }
1110
1111         cmsg = CMSG_FIRSTHDR(&outmessage);
1112         cmsg->cmsg_level = IPPROTO_SCTP;
1113         cmsg->cmsg_type = SCTP_SNDRCV;
1114         cmsg->cmsg_len = CMSG_LEN(sizeof(struct sctp_sndrcvinfo));
1115         sinfo = CMSG_DATA(cmsg);
1116         memset(sinfo, 0x00, sizeof(struct sctp_sndrcvinfo));
1117         sinfo->sinfo_ppid = cpu_to_le32(con->nodeid);
1118         outmessage.msg_controllen = cmsg->cmsg_len;
1119         sinfo->sinfo_flags |= SCTP_ADDR_OVER;
1120
1121         ret = kernel_sendmsg(base_con->sock, &outmessage, iov, 1, len);
1122         if (ret < 0) {
1123                 log_print("Send first packet to node %d failed: %d",
1124                           con->nodeid, ret);
1125
1126                 /* Try again later */
1127                 clear_bit(CF_CONNECT_PENDING, &con->flags);
1128                 clear_bit(CF_INIT_PENDING, &con->flags);
1129         }
1130         else {
1131                 spin_lock(&con->writequeue_lock);
1132                 writequeue_entry_complete(e, ret);
1133                 spin_unlock(&con->writequeue_lock);
1134         }
1135
1136 unlock:
1137         mutex_unlock(&con->sock_mutex);
1138 }
1139
1140 /* Connect a new socket to its peer */
1141 static void tcp_connect_to_sock(struct connection *con)
1142 {
1143         struct sockaddr_storage saddr, src_addr;
1144         int addr_len;
1145         struct socket *sock = NULL;
1146         int one = 1;
1147         int result;
1148
1149         if (con->nodeid == 0) {
1150                 log_print("attempt to connect sock 0 foiled");
1151                 return;
1152         }
1153
1154         mutex_lock(&con->sock_mutex);
1155         if (con->retries++ > MAX_CONNECT_RETRIES)
1156                 goto out;
1157
1158         /* Some odd races can cause double-connects, ignore them */
1159         if (con->sock)
1160                 goto out;
1161
1162         /* Create a socket to communicate with */
1163         result = sock_create_kern(dlm_local_addr[0]->ss_family, SOCK_STREAM,
1164                                   IPPROTO_TCP, &sock);
1165         if (result < 0)
1166                 goto out_err;
1167
1168         memset(&saddr, 0, sizeof(saddr));
1169         result = nodeid_to_addr(con->nodeid, &saddr, NULL, false);
1170         if (result < 0) {
1171                 log_print("no address for nodeid %d", con->nodeid);
1172                 goto out_err;
1173         }
1174
1175         sock->sk->sk_user_data = con;
1176         con->rx_action = receive_from_sock;
1177         con->connect_action = tcp_connect_to_sock;
1178         add_sock(sock, con);
1179
1180         /* Bind to our cluster-known address connecting to avoid
1181            routing problems */
1182         memcpy(&src_addr, dlm_local_addr[0], sizeof(src_addr));
1183         make_sockaddr(&src_addr, 0, &addr_len);
1184         result = sock->ops->bind(sock, (struct sockaddr *) &src_addr,
1185                                  addr_len);
1186         if (result < 0) {
1187                 log_print("could not bind for connect: %d", result);
1188                 /* This *may* not indicate a critical error */
1189         }
1190
1191         make_sockaddr(&saddr, dlm_config.ci_tcp_port, &addr_len);
1192
1193         log_print("connecting to %d", con->nodeid);
1194
1195         /* Turn off Nagle's algorithm */
1196         kernel_setsockopt(sock, SOL_TCP, TCP_NODELAY, (char *)&one,
1197                           sizeof(one));
1198
1199         result = sock->ops->connect(sock, (struct sockaddr *)&saddr, addr_len,
1200                                    O_NONBLOCK);
1201         if (result == -EINPROGRESS)
1202                 result = 0;
1203         if (result == 0)
1204                 goto out;
1205
1206 out_err:
1207         if (con->sock) {
1208                 sock_release(con->sock);
1209                 con->sock = NULL;
1210         } else if (sock) {
1211                 sock_release(sock);
1212         }
1213         /*
1214          * Some errors are fatal and this list might need adjusting. For other
1215          * errors we try again until the max number of retries is reached.
1216          */
1217         if (result != -EHOSTUNREACH &&
1218             result != -ENETUNREACH &&
1219             result != -ENETDOWN &&
1220             result != -EINVAL &&
1221             result != -EPROTONOSUPPORT) {
1222                 log_print("connect %d try %d error %d", con->nodeid,
1223                           con->retries, result);
1224                 mutex_unlock(&con->sock_mutex);
1225                 msleep(1000);
1226                 lowcomms_connect_sock(con);
1227                 return;
1228         }
1229 out:
1230         mutex_unlock(&con->sock_mutex);
1231         return;
1232 }
1233
1234 static struct socket *tcp_create_listen_sock(struct connection *con,
1235                                              struct sockaddr_storage *saddr)
1236 {
1237         struct socket *sock = NULL;
1238         int result = 0;
1239         int one = 1;
1240         int addr_len;
1241
1242         if (dlm_local_addr[0]->ss_family == AF_INET)
1243                 addr_len = sizeof(struct sockaddr_in);
1244         else
1245                 addr_len = sizeof(struct sockaddr_in6);
1246
1247         /* Create a socket to communicate with */
1248         result = sock_create_kern(dlm_local_addr[0]->ss_family, SOCK_STREAM,
1249                                   IPPROTO_TCP, &sock);
1250         if (result < 0) {
1251                 log_print("Can't create listening comms socket");
1252                 goto create_out;
1253         }
1254
1255         /* Turn off Nagle's algorithm */
1256         kernel_setsockopt(sock, SOL_TCP, TCP_NODELAY, (char *)&one,
1257                           sizeof(one));
1258
1259         result = kernel_setsockopt(sock, SOL_SOCKET, SO_REUSEADDR,
1260                                    (char *)&one, sizeof(one));
1261
1262         if (result < 0) {
1263                 log_print("Failed to set SO_REUSEADDR on socket: %d", result);
1264         }
1265         con->rx_action = tcp_accept_from_sock;
1266         con->connect_action = tcp_connect_to_sock;
1267
1268         /* Bind to our port */
1269         make_sockaddr(saddr, dlm_config.ci_tcp_port, &addr_len);
1270         result = sock->ops->bind(sock, (struct sockaddr *) saddr, addr_len);
1271         if (result < 0) {
1272                 log_print("Can't bind to port %d", dlm_config.ci_tcp_port);
1273                 sock_release(sock);
1274                 sock = NULL;
1275                 con->sock = NULL;
1276                 goto create_out;
1277         }
1278         result = kernel_setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE,
1279                                  (char *)&one, sizeof(one));
1280         if (result < 0) {
1281                 log_print("Set keepalive failed: %d", result);
1282         }
1283
1284         result = sock->ops->listen(sock, 5);
1285         if (result < 0) {
1286                 log_print("Can't listen on port %d", dlm_config.ci_tcp_port);
1287                 sock_release(sock);
1288                 sock = NULL;
1289                 goto create_out;
1290         }
1291
1292 create_out:
1293         return sock;
1294 }
1295
1296 /* Get local addresses */
1297 static void init_local(void)
1298 {
1299         struct sockaddr_storage sas, *addr;
1300         int i;
1301
1302         dlm_local_count = 0;
1303         for (i = 0; i < DLM_MAX_ADDR_COUNT; i++) {
1304                 if (dlm_our_addr(&sas, i))
1305                         break;
1306
1307                 addr = kmalloc(sizeof(*addr), GFP_NOFS);
1308                 if (!addr)
1309                         break;
1310                 memcpy(addr, &sas, sizeof(*addr));
1311                 dlm_local_addr[dlm_local_count++] = addr;
1312         }
1313 }
1314
1315 /* Bind to an IP address. SCTP allows multiple address so it can do
1316    multi-homing */
1317 static int add_sctp_bind_addr(struct connection *sctp_con,
1318                               struct sockaddr_storage *addr,
1319                               int addr_len, int num)
1320 {
1321         int result = 0;
1322
1323         if (num == 1)
1324                 result = kernel_bind(sctp_con->sock,
1325                                      (struct sockaddr *) addr,
1326                                      addr_len);
1327         else
1328                 result = kernel_setsockopt(sctp_con->sock, SOL_SCTP,
1329                                            SCTP_SOCKOPT_BINDX_ADD,
1330                                            (char *)addr, addr_len);
1331
1332         if (result < 0)
1333                 log_print("Can't bind to port %d addr number %d",
1334                           dlm_config.ci_tcp_port, num);
1335
1336         return result;
1337 }
1338
1339 /* Initialise SCTP socket and bind to all interfaces */
1340 static int sctp_listen_for_all(void)
1341 {
1342         struct socket *sock = NULL;
1343         struct sockaddr_storage localaddr;
1344         struct sctp_event_subscribe subscribe;
1345         int result = -EINVAL, num = 1, i, addr_len;
1346         struct connection *con = nodeid2con(0, GFP_NOFS);
1347         int bufsize = NEEDED_RMEM;
1348         int one = 1;
1349
1350         if (!con)
1351                 return -ENOMEM;
1352
1353         log_print("Using SCTP for communications");
1354
1355         result = sock_create_kern(dlm_local_addr[0]->ss_family, SOCK_SEQPACKET,
1356                                   IPPROTO_SCTP, &sock);
1357         if (result < 0) {
1358                 log_print("Can't create comms socket, check SCTP is loaded");
1359                 goto out;
1360         }
1361
1362         /* Listen for events */
1363         memset(&subscribe, 0, sizeof(subscribe));
1364         subscribe.sctp_data_io_event = 1;
1365         subscribe.sctp_association_event = 1;
1366         subscribe.sctp_send_failure_event = 1;
1367         subscribe.sctp_shutdown_event = 1;
1368         subscribe.sctp_partial_delivery_event = 1;
1369
1370         result = kernel_setsockopt(sock, SOL_SOCKET, SO_RCVBUFFORCE,
1371                                  (char *)&bufsize, sizeof(bufsize));
1372         if (result)
1373                 log_print("Error increasing buffer space on socket %d", result);
1374
1375         result = kernel_setsockopt(sock, SOL_SCTP, SCTP_EVENTS,
1376                                    (char *)&subscribe, sizeof(subscribe));
1377         if (result < 0) {
1378                 log_print("Failed to set SCTP_EVENTS on socket: result=%d",
1379                           result);
1380                 goto create_delsock;
1381         }
1382
1383         result = kernel_setsockopt(sock, SOL_SCTP, SCTP_NODELAY, (char *)&one,
1384                                    sizeof(one));
1385         if (result < 0)
1386                 log_print("Could not set SCTP NODELAY error %d\n", result);
1387
1388         /* Init con struct */
1389         sock->sk->sk_user_data = con;
1390         con->sock = sock;
1391         con->sock->sk->sk_data_ready = lowcomms_data_ready;
1392         con->rx_action = receive_from_sock;
1393         con->connect_action = sctp_init_assoc;
1394
1395         /* Bind to all interfaces. */
1396         for (i = 0; i < dlm_local_count; i++) {
1397                 memcpy(&localaddr, dlm_local_addr[i], sizeof(localaddr));
1398                 make_sockaddr(&localaddr, dlm_config.ci_tcp_port, &addr_len);
1399
1400                 result = add_sctp_bind_addr(con, &localaddr, addr_len, num);
1401                 if (result)
1402                         goto create_delsock;
1403                 ++num;
1404         }
1405
1406         result = sock->ops->listen(sock, 5);
1407         if (result < 0) {
1408                 log_print("Can't set socket listening");
1409                 goto create_delsock;
1410         }
1411
1412         return 0;
1413
1414 create_delsock:
1415         sock_release(sock);
1416         con->sock = NULL;
1417 out:
1418         return result;
1419 }
1420
1421 static int tcp_listen_for_all(void)
1422 {
1423         struct socket *sock = NULL;
1424         struct connection *con = nodeid2con(0, GFP_NOFS);
1425         int result = -EINVAL;
1426
1427         if (!con)
1428                 return -ENOMEM;
1429
1430         /* We don't support multi-homed hosts */
1431         if (dlm_local_addr[1] != NULL) {
1432                 log_print("TCP protocol can't handle multi-homed hosts, "
1433                           "try SCTP");
1434                 return -EINVAL;
1435         }
1436
1437         log_print("Using TCP for communications");
1438
1439         sock = tcp_create_listen_sock(con, dlm_local_addr[0]);
1440         if (sock) {
1441                 add_sock(sock, con);
1442                 result = 0;
1443         }
1444         else {
1445                 result = -EADDRINUSE;
1446         }
1447
1448         return result;
1449 }
1450
1451
1452
1453 static struct writequeue_entry *new_writequeue_entry(struct connection *con,
1454                                                      gfp_t allocation)
1455 {
1456         struct writequeue_entry *entry;
1457
1458         entry = kmalloc(sizeof(struct writequeue_entry), allocation);
1459         if (!entry)
1460                 return NULL;
1461
1462         entry->page = alloc_page(allocation);
1463         if (!entry->page) {
1464                 kfree(entry);
1465                 return NULL;
1466         }
1467
1468         entry->offset = 0;
1469         entry->len = 0;
1470         entry->end = 0;
1471         entry->users = 0;
1472         entry->con = con;
1473
1474         return entry;
1475 }
1476
1477 void *dlm_lowcomms_get_buffer(int nodeid, int len, gfp_t allocation, char **ppc)
1478 {
1479         struct connection *con;
1480         struct writequeue_entry *e;
1481         int offset = 0;
1482
1483         con = nodeid2con(nodeid, allocation);
1484         if (!con)
1485                 return NULL;
1486
1487         spin_lock(&con->writequeue_lock);
1488         e = list_entry(con->writequeue.prev, struct writequeue_entry, list);
1489         if ((&e->list == &con->writequeue) ||
1490             (PAGE_CACHE_SIZE - e->end < len)) {
1491                 e = NULL;
1492         } else {
1493                 offset = e->end;
1494                 e->end += len;
1495                 e->users++;
1496         }
1497         spin_unlock(&con->writequeue_lock);
1498
1499         if (e) {
1500         got_one:
1501                 *ppc = page_address(e->page) + offset;
1502                 return e;
1503         }
1504
1505         e = new_writequeue_entry(con, allocation);
1506         if (e) {
1507                 spin_lock(&con->writequeue_lock);
1508                 offset = e->end;
1509                 e->end += len;
1510                 e->users++;
1511                 list_add_tail(&e->list, &con->writequeue);
1512                 spin_unlock(&con->writequeue_lock);
1513                 goto got_one;
1514         }
1515         return NULL;
1516 }
1517
1518 void dlm_lowcomms_commit_buffer(void *mh)
1519 {
1520         struct writequeue_entry *e = (struct writequeue_entry *)mh;
1521         struct connection *con = e->con;
1522         int users;
1523
1524         spin_lock(&con->writequeue_lock);
1525         users = --e->users;
1526         if (users)
1527                 goto out;
1528         e->len = e->end - e->offset;
1529         spin_unlock(&con->writequeue_lock);
1530
1531         if (!test_and_set_bit(CF_WRITE_PENDING, &con->flags)) {
1532                 queue_work(send_workqueue, &con->swork);
1533         }
1534         return;
1535
1536 out:
1537         spin_unlock(&con->writequeue_lock);
1538         return;
1539 }
1540
1541 /* Send a message */
1542 static void send_to_sock(struct connection *con)
1543 {
1544         int ret = 0;
1545         const int msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL;
1546         struct writequeue_entry *e;
1547         int len, offset;
1548         int count = 0;
1549
1550         mutex_lock(&con->sock_mutex);
1551         if (con->sock == NULL)
1552                 goto out_connect;
1553
1554         spin_lock(&con->writequeue_lock);
1555         for (;;) {
1556                 e = list_entry(con->writequeue.next, struct writequeue_entry,
1557                                list);
1558                 if ((struct list_head *) e == &con->writequeue)
1559                         break;
1560
1561                 len = e->len;
1562                 offset = e->offset;
1563                 BUG_ON(len == 0 && e->users == 0);
1564                 spin_unlock(&con->writequeue_lock);
1565
1566                 ret = 0;
1567                 if (len) {
1568                         ret = kernel_sendpage(con->sock, e->page, offset, len,
1569                                               msg_flags);
1570                         if (ret == -EAGAIN || ret == 0) {
1571                                 if (ret == -EAGAIN &&
1572                                     test_bit(SOCK_ASYNC_NOSPACE, &con->sock->flags) &&
1573                                     !test_and_set_bit(CF_APP_LIMITED, &con->flags)) {
1574                                         /* Notify TCP that we're limited by the
1575                                          * application window size.
1576                                          */
1577                                         set_bit(SOCK_NOSPACE, &con->sock->flags);
1578                                         con->sock->sk->sk_write_pending++;
1579                                 }
1580                                 cond_resched();
1581                                 goto out;
1582                         } else if (ret < 0)
1583                                 goto send_error;
1584                 }
1585
1586                 /* Don't starve people filling buffers */
1587                 if (++count >= MAX_SEND_MSG_COUNT) {
1588                         cond_resched();
1589                         count = 0;
1590                 }
1591
1592                 spin_lock(&con->writequeue_lock);
1593                 writequeue_entry_complete(e, ret);
1594         }
1595         spin_unlock(&con->writequeue_lock);
1596 out:
1597         mutex_unlock(&con->sock_mutex);
1598         return;
1599
1600 send_error:
1601         mutex_unlock(&con->sock_mutex);
1602         close_connection(con, false);
1603         lowcomms_connect_sock(con);
1604         return;
1605
1606 out_connect:
1607         mutex_unlock(&con->sock_mutex);
1608         if (!test_bit(CF_INIT_PENDING, &con->flags))
1609                 lowcomms_connect_sock(con);
1610 }
1611
1612 static void clean_one_writequeue(struct connection *con)
1613 {
1614         struct writequeue_entry *e, *safe;
1615
1616         spin_lock(&con->writequeue_lock);
1617         list_for_each_entry_safe(e, safe, &con->writequeue, list) {
1618                 list_del(&e->list);
1619                 free_entry(e);
1620         }
1621         spin_unlock(&con->writequeue_lock);
1622 }
1623
1624 /* Called from recovery when it knows that a node has
1625    left the cluster */
1626 int dlm_lowcomms_close(int nodeid)
1627 {
1628         struct connection *con;
1629         struct dlm_node_addr *na;
1630
1631         log_print("closing connection to node %d", nodeid);
1632         con = nodeid2con(nodeid, 0);
1633         if (con) {
1634                 clear_bit(CF_CONNECT_PENDING, &con->flags);
1635                 clear_bit(CF_WRITE_PENDING, &con->flags);
1636                 set_bit(CF_CLOSE, &con->flags);
1637                 if (cancel_work_sync(&con->swork))
1638                         log_print("canceled swork for node %d", nodeid);
1639                 if (cancel_work_sync(&con->rwork))
1640                         log_print("canceled rwork for node %d", nodeid);
1641                 clean_one_writequeue(con);
1642                 close_connection(con, true);
1643         }
1644
1645         spin_lock(&dlm_node_addrs_spin);
1646         na = find_node_addr(nodeid);
1647         if (na) {
1648                 list_del(&na->list);
1649                 while (na->addr_count--)
1650                         kfree(na->addr[na->addr_count]);
1651                 kfree(na);
1652         }
1653         spin_unlock(&dlm_node_addrs_spin);
1654
1655         return 0;
1656 }
1657
1658 /* Receive workqueue function */
1659 static void process_recv_sockets(struct work_struct *work)
1660 {
1661         struct connection *con = container_of(work, struct connection, rwork);
1662         int err;
1663
1664         clear_bit(CF_READ_PENDING, &con->flags);
1665         do {
1666                 err = con->rx_action(con);
1667         } while (!err);
1668 }
1669
1670 /* Send workqueue function */
1671 static void process_send_sockets(struct work_struct *work)
1672 {
1673         struct connection *con = container_of(work, struct connection, swork);
1674
1675         if (test_and_clear_bit(CF_CONNECT_PENDING, &con->flags)) {
1676                 con->connect_action(con);
1677                 set_bit(CF_WRITE_PENDING, &con->flags);
1678         }
1679         if (test_and_clear_bit(CF_WRITE_PENDING, &con->flags))
1680                 send_to_sock(con);
1681 }
1682
1683
1684 /* Discard all entries on the write queues */
1685 static void clean_writequeues(void)
1686 {
1687         foreach_conn(clean_one_writequeue);
1688 }
1689
1690 static void work_stop(void)
1691 {
1692         destroy_workqueue(recv_workqueue);
1693         destroy_workqueue(send_workqueue);
1694 }
1695
1696 static int work_start(void)
1697 {
1698         recv_workqueue = alloc_workqueue("dlm_recv",
1699                                          WQ_UNBOUND | WQ_MEM_RECLAIM, 1);
1700         if (!recv_workqueue) {
1701                 log_print("can't start dlm_recv");
1702                 return -ENOMEM;
1703         }
1704
1705         send_workqueue = alloc_workqueue("dlm_send",
1706                                          WQ_UNBOUND | WQ_MEM_RECLAIM, 1);
1707         if (!send_workqueue) {
1708                 log_print("can't start dlm_send");
1709                 destroy_workqueue(recv_workqueue);
1710                 return -ENOMEM;
1711         }
1712
1713         return 0;
1714 }
1715
1716 static void stop_conn(struct connection *con)
1717 {
1718         con->flags |= 0x0F;
1719         if (con->sock && con->sock->sk)
1720                 con->sock->sk->sk_user_data = NULL;
1721 }
1722
1723 static void free_conn(struct connection *con)
1724 {
1725         close_connection(con, true);
1726         if (con->othercon)
1727                 kmem_cache_free(con_cache, con->othercon);
1728         hlist_del(&con->list);
1729         kmem_cache_free(con_cache, con);
1730 }
1731
1732 void dlm_lowcomms_stop(void)
1733 {
1734         /* Set all the flags to prevent any
1735            socket activity.
1736         */
1737         mutex_lock(&connections_lock);
1738         dlm_allow_conn = 0;
1739         foreach_conn(stop_conn);
1740         mutex_unlock(&connections_lock);
1741
1742         work_stop();
1743
1744         mutex_lock(&connections_lock);
1745         clean_writequeues();
1746
1747         foreach_conn(free_conn);
1748
1749         mutex_unlock(&connections_lock);
1750         kmem_cache_destroy(con_cache);
1751 }
1752
1753 int dlm_lowcomms_start(void)
1754 {
1755         int error = -EINVAL;
1756         struct connection *con;
1757         int i;
1758
1759         for (i = 0; i < CONN_HASH_SIZE; i++)
1760                 INIT_HLIST_HEAD(&connection_hash[i]);
1761
1762         init_local();
1763         if (!dlm_local_count) {
1764                 error = -ENOTCONN;
1765                 log_print("no local IP address has been set");
1766                 goto fail;
1767         }
1768
1769         error = -ENOMEM;
1770         con_cache = kmem_cache_create("dlm_conn", sizeof(struct connection),
1771                                       __alignof__(struct connection), 0,
1772                                       NULL);
1773         if (!con_cache)
1774                 goto fail;
1775
1776         error = work_start();
1777         if (error)
1778                 goto fail_destroy;
1779
1780         dlm_allow_conn = 1;
1781
1782         /* Start listening */
1783         if (dlm_config.ci_protocol == 0)
1784                 error = tcp_listen_for_all();
1785         else
1786                 error = sctp_listen_for_all();
1787         if (error)
1788                 goto fail_unlisten;
1789
1790         return 0;
1791
1792 fail_unlisten:
1793         dlm_allow_conn = 0;
1794         con = nodeid2con(0,0);
1795         if (con) {
1796                 close_connection(con, false);
1797                 kmem_cache_free(con_cache, con);
1798         }
1799 fail_destroy:
1800         kmem_cache_destroy(con_cache);
1801 fail:
1802         return error;
1803 }
1804
1805 void dlm_lowcomms_exit(void)
1806 {
1807         struct dlm_node_addr *na, *safe;
1808
1809         spin_lock(&dlm_node_addrs_spin);
1810         list_for_each_entry_safe(na, safe, &dlm_node_addrs, list) {
1811                 list_del(&na->list);
1812                 while (na->addr_count--)
1813                         kfree(na->addr[na->addr_count]);
1814                 kfree(na);
1815         }
1816         spin_unlock(&dlm_node_addrs_spin);
1817 }