kernel/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)
 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         if (!nodeid) {
 622                 log_print("Shouldn't resend data via listening connection.");
 623                 return;
 624         }
 625
 626         con = nodeid2con(nodeid, 0);
 627         if (!con) {
 628                 log_print("Could not look up con for nodeid %d\n",
 629                           nodeid);
 630                 return;
 631         }
 632
 633         mh = dlm_lowcomms_get_buffer(nodeid, len, GFP_NOFS, &retry_buf);
 634         if (!mh) {
 635                 log_print("Could not allocate buf for retry.");
 636                 return;
 637         }
 638         memcpy(retry_buf, buf + sizeof(struct sctp_send_failed), len);
 639         dlm_lowcomms_commit_buffer(mh);
 640
 641         /*
 642          * If we got a assoc changed event before the send failed event then
 643          * we only need to retry the send.
 644          */
 645         if (con->sctp_assoc) {
 646                 if (!test_and_set_bit(CF_WRITE_PENDING, &con->flags))
 647                         queue_work(send_workqueue, &con->swork);
 648         } else
 649                 sctp_init_failed_foreach(con);
 650 }
 651
 652 /* Something happened to an association */
 653 static void process_sctp_notification(struct connection *con,
 654                                       struct msghdr *msg, char *buf)
 655 {
 656         union sctp_notification *sn = (union sctp_notification *)buf;
 657         struct linger linger;
 658
 659         switch (sn->sn_header.sn_type) {
 660         case SCTP_SEND_FAILED:
 661                 retry_failed_sctp_send(con, &sn->sn_send_failed, buf);
 662                 break;
 663         case SCTP_ASSOC_CHANGE:
 664                 switch (sn->sn_assoc_change.sac_state) {
 665                 case SCTP_COMM_UP:
 666                 case SCTP_RESTART:
 667                 {
 668                         /* Check that the new node is in the lockspace */
 669                         struct sctp_prim prim;
 670                         int nodeid;
 671                         int prim_len, ret;
 672                         int addr_len;
 673                         struct connection *new_con;
 674
 675                         /*
 676                          * We get this before any data for an association.
 677                          * We verify that the node is in the cluster and
 678                          * then peel off a socket for it.
 679                          */
 680                         if ((int)sn->sn_assoc_change.sac_assoc_id <= 0) {
 681                                 log_print("COMM_UP for invalid assoc ID %d",
 682                                          (int)sn->sn_assoc_change.sac_assoc_id);
 683                                 sctp_init_failed();
 684                                 return;
 685                         }
 686                         memset(&prim, 0, sizeof(struct sctp_prim));
 687                         prim_len = sizeof(struct sctp_prim);
 688                         prim.ssp_assoc_id = sn->sn_assoc_change.sac_assoc_id;
 689
 690                         ret = kernel_getsockopt(con->sock,
 691                                                 IPPROTO_SCTP,
 692                                                 SCTP_PRIMARY_ADDR,
 693                                                 (char*)&prim,
 694                                                 &prim_len);
 695                         if (ret < 0) {
 696                                 log_print("getsockopt/sctp_primary_addr on "
 697                                           "new assoc %d failed : %d",
 698                                           (int)sn->sn_assoc_change.sac_assoc_id,
 699                                           ret);
 700
 701                                 /* Retry INIT later */
 702                                 new_con = assoc2con(sn->sn_assoc_change.sac_assoc_id);
 703                                 if (new_con)
 704                                         clear_bit(CF_CONNECT_PENDING, &con->flags);
 705                                 return;
 706                         }
 707                         make_sockaddr(&prim.ssp_addr, 0, &addr_len);
 708                         if (addr_to_nodeid(&prim.ssp_addr, &nodeid)) {
 709                                 unsigned char *b=(unsigned char *)&prim.ssp_addr;
 710                                 log_print("reject connect from unknown addr");
 711                                 print_hex_dump_bytes("ss: ", DUMP_PREFIX_NONE,
 712                                                      b, sizeof(struct sockaddr_storage));
 713                                 sctp_send_shutdown(prim.ssp_assoc_id);
 714                                 return;
 715                         }
 716
 717                         new_con = nodeid2con(nodeid, GFP_NOFS);
 718                         if (!new_con)
 719                                 return;
 720
 721                         /* Peel off a new sock */
 722                         lock_sock(con->sock->sk);
 723                         ret = sctp_do_peeloff(con->sock->sk,
 724                                 sn->sn_assoc_change.sac_assoc_id,
 725                                 &new_con->sock);
 726                         release_sock(con->sock->sk);
 727                         if (ret < 0) {
 728                                 log_print("Can't peel off a socket for "
 729                                           "connection %d to node %d: err=%d",
 730                                           (int)sn->sn_assoc_change.sac_assoc_id,
 731                                           nodeid, ret);
 732                                 return;
 733                         }
 734                         add_sock(new_con->sock, new_con);
 735
 736                         linger.l_onoff = 1;
 737                         linger.l_linger = 0;
 738                         ret = kernel_setsockopt(new_con->sock, SOL_SOCKET, SO_LINGER,
 739                                                 (char *)&linger, sizeof(linger));
 740                         if (ret < 0)
 741                                 log_print("set socket option SO_LINGER failed");
 742
 743                         log_print("connecting to %d sctp association %d",
 744                                  nodeid, (int)sn->sn_assoc_change.sac_assoc_id);
 745
 746                         new_con->sctp_assoc = sn->sn_assoc_change.sac_assoc_id;
 747                         new_con->try_new_addr = false;
 748                         /* Send any pending writes */
 749                         clear_bit(CF_CONNECT_PENDING, &new_con->flags);
 750                         clear_bit(CF_INIT_PENDING, &new_con->flags);
 751                         if (!test_and_set_bit(CF_WRITE_PENDING, &new_con->flags)) {
 752                                 queue_work(send_workqueue, &new_con->swork);
 753                         }
 754                         if (!test_and_set_bit(CF_READ_PENDING, &new_con->flags))
 755                                 queue_work(recv_workqueue, &new_con->rwork);
 756                 }
 757                 break;
 758
 759                 case SCTP_COMM_LOST:
 760                 case SCTP_SHUTDOWN_COMP:
 761                 {
 762                         con = assoc2con(sn->sn_assoc_change.sac_assoc_id);
 763                         if (con) {
 764                                 con->sctp_assoc = 0;
 765                         }
 766                 }
 767                 break;
 768
 769                 case SCTP_CANT_STR_ASSOC:
 770                 {
 771                         /* Will retry init when we get the send failed notification */
 772                         log_print("Can't start SCTP association - retrying");
 773                 }
 774                 break;
 775
 776                 default:
 777                         log_print("unexpected SCTP assoc change id=%d state=%d",
 778                                   (int)sn->sn_assoc_change.sac_assoc_id,
 779                                   sn->sn_assoc_change.sac_state);
 780                 }
 781         default:
 782                 ; /* fall through */
 783         }
 784 }
 785
 786 /* Data received from remote end */
 787 static int receive_from_sock(struct connection *con)
 788 {
 789         int ret = 0;
 790         struct msghdr msg = {};
 791         struct kvec iov[2];
 792         unsigned len;
 793         int r;
 794         int call_again_soon = 0;
 795         int nvec;
 796         char incmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))];
 797
 798         mutex_lock(&con->sock_mutex);
 799
 800         if (con->sock == NULL) {
 801                 ret = -EAGAIN;
 802                 goto out_close;
 803         }
 804
 805         if (con->rx_page == NULL) {
 806                 /*
 807                  * This doesn't need to be atomic, but I think it should
 808                  * improve performance if it is.
 809                  */
 810                 con->rx_page = alloc_page(GFP_ATOMIC);
 811                 if (con->rx_page == NULL)
 812                         goto out_resched;
 813                 cbuf_init(&con->cb, PAGE_CACHE_SIZE);
 814         }
 815
 816         /* Only SCTP needs these really */
 817         memset(&incmsg, 0, sizeof(incmsg));
 818         msg.msg_control = incmsg;
 819         msg.msg_controllen = sizeof(incmsg);
 820
 821         /*
 822          * iov[0] is the bit of the circular buffer between the current end
 823          * point (cb.base + cb.len) and the end of the buffer.
 824          */
 825         iov[0].iov_len = con->cb.base - cbuf_data(&con->cb);
 826         iov[0].iov_base = page_address(con->rx_page) + cbuf_data(&con->cb);
 827         iov[1].iov_len = 0;
 828         nvec = 1;
 829
 830         /*
 831          * iov[1] is the bit of the circular buffer between the start of the
 832          * buffer and the start of the currently used section (cb.base)
 833          */
 834         if (cbuf_data(&con->cb) >= con->cb.base) {
 835                 iov[0].iov_len = PAGE_CACHE_SIZE - cbuf_data(&con->cb);
 836                 iov[1].iov_len = con->cb.base;
 837                 iov[1].iov_base = page_address(con->rx_page);
 838                 nvec = 2;
 839         }
 840         len = iov[0].iov_len + iov[1].iov_len;
 841
 842         r = ret = kernel_recvmsg(con->sock, &msg, iov, nvec, len,
 843                                MSG_DONTWAIT | MSG_NOSIGNAL);
 844         if (ret <= 0)
 845                 goto out_close;
 846
 847         /* Process SCTP notifications */
 848         if (msg.msg_flags & MSG_NOTIFICATION) {
 849                 msg.msg_control = incmsg;
 850                 msg.msg_controllen = sizeof(incmsg);
 851
 852                 process_sctp_notification(con, &msg,
 853                                 page_address(con->rx_page) + con->cb.base);
 854                 mutex_unlock(&con->sock_mutex);
 855                 return 0;
 856         }
 857         BUG_ON(con->nodeid == 0);
 858
 859         if (ret == len)
 860                 call_again_soon = 1;
 861         cbuf_add(&con->cb, ret);
 862         ret = dlm_process_incoming_buffer(con->nodeid,
 863                                           page_address(con->rx_page),
 864                                           con->cb.base, con->cb.len,
 865                                           PAGE_CACHE_SIZE);
 866         if (ret == -EBADMSG) {
 867                 log_print("lowcomms: addr=%p, base=%u, len=%u, "
 868                           "iov_len=%u, iov_base[0]=%p, read=%d",
 869                           page_address(con->rx_page), con->cb.base, con->cb.len,
 870                           len, iov[0].iov_base, r);
 871         }
 872         if (ret < 0)
 873                 goto out_close;
 874         cbuf_eat(&con->cb, ret);
 875
 876         if (cbuf_empty(&con->cb) && !call_again_soon) {
 877                 __free_page(con->rx_page);
 878                 con->rx_page = NULL;
 879         }
 880
 881         if (call_again_soon)
 882                 goto out_resched;
 883         mutex_unlock(&con->sock_mutex);
 884         return 0;
 885
 886 out_resched:
 887         if (!test_and_set_bit(CF_READ_PENDING, &con->flags))
 888                 queue_work(recv_workqueue, &con->rwork);
 889         mutex_unlock(&con->sock_mutex);
 890         return -EAGAIN;
 891
 892 out_close:
 893         mutex_unlock(&con->sock_mutex);
 894         if (ret != -EAGAIN) {
 895                 close_connection(con, false);
 896                 /* Reconnect when there is something to send */
 897         }
 898         /* Don't return success if we really got EOF */
 899         if (ret == 0)
 900                 ret = -EAGAIN;
 901
 902         return ret;
 903 }
 904
 905 /* Listening socket is busy, accept a connection */
 906 static int tcp_accept_from_sock(struct connection *con)
 907 {
 908         int result;
 909         struct sockaddr_storage peeraddr;
 910         struct socket *newsock;
 911         int len;
 912         int nodeid;
 913         struct connection *newcon;
 914         struct connection *addcon;
 915
 916         mutex_lock(&connections_lock);
 917         if (!dlm_allow_conn) {
 918                 mutex_unlock(&connections_lock);
 919                 return -1;
 920         }
 921         mutex_unlock(&connections_lock);
 922
 923         memset(&peeraddr, 0, sizeof(peeraddr));
 924         result = sock_create_kern(dlm_local_addr[0]->ss_family, SOCK_STREAM,
 925                                   IPPROTO_TCP, &newsock);
 926         if (result < 0)
 927                 return -ENOMEM;
 928
 929         mutex_lock_nested(&con->sock_mutex, 0);
 930
 931         result = -ENOTCONN;
 932         if (con->sock == NULL)
 933                 goto accept_err;
 934
 935         newsock->type = con->sock->type;
 936         newsock->ops = con->sock->ops;
 937
 938         result = con->sock->ops->accept(con->sock, newsock, O_NONBLOCK);
 939         if (result < 0)
 940                 goto accept_err;
 941
 942         /* Get the connected socket's peer */
 943         memset(&peeraddr, 0, sizeof(peeraddr));
 944         if (newsock->ops->getname(newsock, (struct sockaddr *)&peeraddr,
 945                                   &len, 2)) {
 946                 result = -ECONNABORTED;
 947                 goto accept_err;
 948         }
 949
 950         /* Get the new node's NODEID */
 951         make_sockaddr(&peeraddr, 0, &len);
 952         if (addr_to_nodeid(&peeraddr, &nodeid)) {
 953                 unsigned char *b=(unsigned char *)&peeraddr;
 954                 log_print("connect from non cluster node");
 955                 print_hex_dump_bytes("ss: ", DUMP_PREFIX_NONE,
 956                                      b, sizeof(struct sockaddr_storage));
 957                 sock_release(newsock);
 958                 mutex_unlock(&con->sock_mutex);
 959                 return -1;
 960         }
 961
 962         log_print("got connection from %d", nodeid);
 963
 964         /*  Check to see if we already have a connection to this node. This
 965          *  could happen if the two nodes initiate a connection at roughly
 966          *  the same time and the connections cross on the wire.
 967          *  In this case we store the incoming one in "othercon"
 968          */
 969         newcon = nodeid2con(nodeid, GFP_NOFS);
 970         if (!newcon) {
 971                 result = -ENOMEM;
 972                 goto accept_err;
 973         }
 974         mutex_lock_nested(&newcon->sock_mutex, 1);
 975         if (newcon->sock) {
 976                 struct connection *othercon = newcon->othercon;
 977
 978                 if (!othercon) {
 979                         othercon = kmem_cache_zalloc(con_cache, GFP_NOFS);
 980                         if (!othercon) {
 981                                 log_print("failed to allocate incoming socket");
 982                                 mutex_unlock(&newcon->sock_mutex);
 983                                 result = -ENOMEM;
 984                                 goto accept_err;
 985                         }
 986                         othercon->nodeid = nodeid;
 987                         othercon->rx_action = receive_from_sock;
 988                         mutex_init(&othercon->sock_mutex);
 989                         INIT_WORK(&othercon->swork, process_send_sockets);
 990                         INIT_WORK(&othercon->rwork, process_recv_sockets);
 991                         set_bit(CF_IS_OTHERCON, &othercon->flags);
 992                 }
 993                 if (!othercon->sock) {
 994                         newcon->othercon = othercon;
 995                         othercon->sock = newsock;
 996                         newsock->sk->sk_user_data = othercon;
 997                         add_sock(newsock, othercon);
 998                         addcon = othercon;
 999                 }
1000                 else {
1001                         printk("Extra connection from node %d attempted\n", nodeid);
1002                         result = -EAGAIN;
1003                         mutex_unlock(&newcon->sock_mutex);
1004                         goto accept_err;
1005                 }
1006         }
1007         else {
1008                 newsock->sk->sk_user_data = newcon;
1009                 newcon->rx_action = receive_from_sock;
1010                 add_sock(newsock, newcon);
1011                 addcon = newcon;
1012         }
1013
1014         mutex_unlock(&newcon->sock_mutex);
1015
1016         /*
1017          * Add it to the active queue in case we got data
1018          * between processing the accept adding the socket
1019          * to the read_sockets list
1020          */
1021         if (!test_and_set_bit(CF_READ_PENDING, &addcon->flags))
1022                 queue_work(recv_workqueue, &addcon->rwork);
1023         mutex_unlock(&con->sock_mutex);
1024
1025         return 0;
1026
1027 accept_err:
1028         mutex_unlock(&con->sock_mutex);
1029         sock_release(newsock);
1030
1031         if (result != -EAGAIN)
1032                 log_print("error accepting connection from node: %d", result);
1033         return result;
1034 }
1035
1036 static void free_entry(struct writequeue_entry *e)
1037 {
1038         __free_page(e->page);
1039         kfree(e);
1040 }
1041
1042 /*
1043  * writequeue_entry_complete - try to delete and free write queue entry
1044  * @e: write queue entry to try to delete
1045  * @completed: bytes completed
1046  *
1047  * writequeue_lock must be held.
1048  */
1049 static void writequeue_entry_complete(struct writequeue_entry *e, int completed)
1050 {
1051         e->offset += completed;
1052         e->len -= completed;
1053
1054         if (e->len == 0 && e->users == 0) {
1055                 list_del(&e->list);
1056                 free_entry(e);
1057         }
1058 }
1059
1060 /* Initiate an SCTP association.
1061    This is a special case of send_to_sock() in that we don't yet have a
1062    peeled-off socket for this association, so we use the listening socket
1063    and add the primary IP address of the remote node.
1064  */
1065 static void sctp_init_assoc(struct connection *con)
1066 {
1067         struct sockaddr_storage rem_addr;
1068         char outcmsg[CMSG_SPACE(sizeof(struct sctp_sndrcvinfo))];
1069         struct msghdr outmessage;
1070         struct cmsghdr *cmsg;
1071         struct sctp_sndrcvinfo *sinfo;
1072         struct connection *base_con;
1073         struct writequeue_entry *e;
1074         int len, offset;
1075         int ret;
1076         int addrlen;
1077         struct kvec iov[1];
1078
1079         mutex_lock(&con->sock_mutex);
1080         if (test_and_set_bit(CF_INIT_PENDING, &con->flags))
1081                 goto unlock;
1082
1083         if (nodeid_to_addr(con->nodeid, NULL, (struct sockaddr *)&rem_addr,
1084                            con->try_new_addr)) {
1085                 log_print("no address for nodeid %d", con->nodeid);
1086                 goto unlock;
1087         }
1088         base_con = nodeid2con(0, 0);
1089         BUG_ON(base_con == NULL);
1090
1091         make_sockaddr(&rem_addr, dlm_config.ci_tcp_port, &addrlen);
1092
1093         outmessage.msg_name = &rem_addr;
1094         outmessage.msg_namelen = addrlen;
1095         outmessage.msg_control = outcmsg;
1096         outmessage.msg_controllen = sizeof(outcmsg);
1097         outmessage.msg_flags = MSG_EOR;
1098
1099         spin_lock(&con->writequeue_lock);
1100
1101         if (list_empty(&con->writequeue)) {
1102                 spin_unlock(&con->writequeue_lock);
1103                 log_print("writequeue empty for nodeid %d", con->nodeid);
1104                 goto unlock;
1105         }
1106
1107         e = list_first_entry(&con->writequeue, struct writequeue_entry, list);
1108         len = e->len;
1109         offset = e->offset;
1110
1111         /* Send the first block off the write queue */
1112         iov[0].iov_base = page_address(e->page)+offset;
1113         iov[0].iov_len = len;
1114         spin_unlock(&con->writequeue_lock);
1115
1116         if (rem_addr.ss_family == AF_INET) {
1117                 struct sockaddr_in *sin = (struct sockaddr_in *)&rem_addr;
1118                 log_print("Trying to connect to %pI4", &sin->sin_addr.s_addr);
1119         } else {
1120                 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)&rem_addr;
1121                 log_print("Trying to connect to %pI6", &sin6->sin6_addr);
1122         }
1123
1124         cmsg = CMSG_FIRSTHDR(&outmessage);
1125         cmsg->cmsg_level = IPPROTO_SCTP;
1126         cmsg->cmsg_type = SCTP_SNDRCV;
1127         cmsg->cmsg_len = CMSG_LEN(sizeof(struct sctp_sndrcvinfo));
1128         sinfo = CMSG_DATA(cmsg);
1129         memset(sinfo, 0x00, sizeof(struct sctp_sndrcvinfo));
1130         sinfo->sinfo_ppid = cpu_to_le32(con->nodeid);
1131         outmessage.msg_controllen = cmsg->cmsg_len;
1132         sinfo->sinfo_flags |= SCTP_ADDR_OVER;
1133
1134         ret = kernel_sendmsg(base_con->sock, &outmessage, iov, 1, len);
1135         if (ret < 0) {
1136                 log_print("Send first packet to node %d failed: %d",
1137                           con->nodeid, ret);
1138
1139                 /* Try again later */
1140                 clear_bit(CF_CONNECT_PENDING, &con->flags);
1141                 clear_bit(CF_INIT_PENDING, &con->flags);
1142         }
1143         else {
1144                 spin_lock(&con->writequeue_lock);
1145                 writequeue_entry_complete(e, ret);
1146                 spin_unlock(&con->writequeue_lock);
1147         }
1148
1149 unlock:
1150         mutex_unlock(&con->sock_mutex);
1151 }
1152
1153 /* Connect a new socket to its peer */
1154 static void tcp_connect_to_sock(struct connection *con)
1155 {
1156         struct sockaddr_storage saddr, src_addr;
1157         int addr_len;
1158         struct socket *sock = NULL;
1159         int one = 1;
1160         int result;
1161
1162         if (con->nodeid == 0) {
1163                 log_print("attempt to connect sock 0 foiled");
1164                 return;
1165         }
1166
1167         mutex_lock(&con->sock_mutex);
1168         if (con->retries++ > MAX_CONNECT_RETRIES)
1169                 goto out;
1170
1171         /* Some odd races can cause double-connects, ignore them */
1172         if (con->sock)
1173                 goto out;
1174
1175         /* Create a socket to communicate with */
1176         result = sock_create_kern(dlm_local_addr[0]->ss_family, SOCK_STREAM,
1177                                   IPPROTO_TCP, &sock);
1178         if (result < 0)
1179                 goto out_err;
1180
1181         memset(&saddr, 0, sizeof(saddr));
1182         result = nodeid_to_addr(con->nodeid, &saddr, NULL, false);
1183         if (result < 0) {
1184                 log_print("no address for nodeid %d", con->nodeid);
1185                 goto out_err;
1186         }
1187
1188         sock->sk->sk_user_data = con;
1189         con->rx_action = receive_from_sock;
1190         con->connect_action = tcp_connect_to_sock;
1191         add_sock(sock, con);
1192
1193         /* Bind to our cluster-known address connecting to avoid
1194            routing problems */
1195         memcpy(&src_addr, dlm_local_addr[0], sizeof(src_addr));
1196         make_sockaddr(&src_addr, 0, &addr_len);
1197         result = sock->ops->bind(sock, (struct sockaddr *) &src_addr,
1198                                  addr_len);
1199         if (result < 0) {
1200                 log_print("could not bind for connect: %d", result);
1201                 /* This *may* not indicate a critical error */
1202         }
1203
1204         make_sockaddr(&saddr, dlm_config.ci_tcp_port, &addr_len);
1205
1206         log_print("connecting to %d", con->nodeid);
1207
1208         /* Turn off Nagle's algorithm */
1209         kernel_setsockopt(sock, SOL_TCP, TCP_NODELAY, (char *)&one,
1210                           sizeof(one));
1211
1212         result = sock->ops->connect(sock, (struct sockaddr *)&saddr, addr_len,
1213                                    O_NONBLOCK);
1214         if (result == -EINPROGRESS)
1215                 result = 0;
1216         if (result == 0)
1217                 goto out;
1218
1219 out_err:
1220         if (con->sock) {
1221                 sock_release(con->sock);
1222                 con->sock = NULL;
1223         } else if (sock) {
1224                 sock_release(sock);
1225         }
1226         /*
1227          * Some errors are fatal and this list might need adjusting. For other
1228          * errors we try again until the max number of retries is reached.
1229          */
1230         if (result != -EHOSTUNREACH &&
1231             result != -ENETUNREACH &&
1232             result != -ENETDOWN &&
1233             result != -EINVAL &&
1234             result != -EPROTONOSUPPORT) {
1235                 log_print("connect %d try %d error %d", con->nodeid,
1236                           con->retries, result);
1237                 mutex_unlock(&con->sock_mutex);
1238                 msleep(1000);
1239                 lowcomms_connect_sock(con);
1240                 return;
1241         }
1242 out:
1243         mutex_unlock(&con->sock_mutex);
1244         return;
1245 }
1246
1247 static struct socket *tcp_create_listen_sock(struct connection *con,
1248                                              struct sockaddr_storage *saddr)
1249 {
1250         struct socket *sock = NULL;
1251         int result = 0;
1252         int one = 1;
1253         int addr_len;
1254
1255         if (dlm_local_addr[0]->ss_family == AF_INET)
1256                 addr_len = sizeof(struct sockaddr_in);
1257         else
1258                 addr_len = sizeof(struct sockaddr_in6);
1259
1260         /* Create a socket to communicate with */
1261         result = sock_create_kern(dlm_local_addr[0]->ss_family, SOCK_STREAM,
1262                                   IPPROTO_TCP, &sock);
1263         if (result < 0) {
1264                 log_print("Can't create listening comms socket");
1265                 goto create_out;
1266         }
1267
1268         /* Turn off Nagle's algorithm */
1269         kernel_setsockopt(sock, SOL_TCP, TCP_NODELAY, (char *)&one,
1270                           sizeof(one));
1271
1272         result = kernel_setsockopt(sock, SOL_SOCKET, SO_REUSEADDR,
1273                                    (char *)&one, sizeof(one));
1274
1275         if (result < 0) {
1276                 log_print("Failed to set SO_REUSEADDR on socket: %d", result);
1277         }
1278         con->rx_action = tcp_accept_from_sock;
1279         con->connect_action = tcp_connect_to_sock;
1280
1281         /* Bind to our port */
1282         make_sockaddr(saddr, dlm_config.ci_tcp_port, &addr_len);
1283         result = sock->ops->bind(sock, (struct sockaddr *) saddr, addr_len);
1284         if (result < 0) {
1285                 log_print("Can't bind to port %d", dlm_config.ci_tcp_port);
1286                 sock_release(sock);
1287                 sock = NULL;
1288                 con->sock = NULL;
1289                 goto create_out;
1290         }
1291         result = kernel_setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE,
1292                                  (char *)&one, sizeof(one));
1293         if (result < 0) {
1294                 log_print("Set keepalive failed: %d", result);
1295         }
1296
1297         result = sock->ops->listen(sock, 5);
1298         if (result < 0) {
1299                 log_print("Can't listen on port %d", dlm_config.ci_tcp_port);
1300                 sock_release(sock);
1301                 sock = NULL;
1302                 goto create_out;
1303         }
1304
1305 create_out:
1306         return sock;
1307 }
1308
1309 /* Get local addresses */
1310 static void init_local(void)
1311 {
1312         struct sockaddr_storage sas, *addr;
1313         int i;
1314
1315         dlm_local_count = 0;
1316         for (i = 0; i < DLM_MAX_ADDR_COUNT; i++) {
1317                 if (dlm_our_addr(&sas, i))
1318                         break;
1319
1320                 addr = kmalloc(sizeof(*addr), GFP_NOFS);
1321                 if (!addr)
1322                         break;
1323                 memcpy(addr, &sas, sizeof(*addr));
1324                 dlm_local_addr[dlm_local_count++] = addr;
1325         }
1326 }
1327
1328 /* Bind to an IP address. SCTP allows multiple address so it can do
1329    multi-homing */
1330 static int add_sctp_bind_addr(struct connection *sctp_con,
1331                               struct sockaddr_storage *addr,
1332                               int addr_len, int num)
1333 {
1334         int result = 0;
1335
1336         if (num == 1)
1337                 result = kernel_bind(sctp_con->sock,
1338                                      (struct sockaddr *) addr,
1339                                      addr_len);
1340         else
1341                 result = kernel_setsockopt(sctp_con->sock, SOL_SCTP,
1342                                            SCTP_SOCKOPT_BINDX_ADD,
1343                                            (char *)addr, addr_len);
1344
1345         if (result < 0)
1346                 log_print("Can't bind to port %d addr number %d",
1347                           dlm_config.ci_tcp_port, num);
1348
1349         return result;
1350 }
1351
1352 /* Initialise SCTP socket and bind to all interfaces */
1353 static int sctp_listen_for_all(void)
1354 {
1355         struct socket *sock = NULL;
1356         struct sockaddr_storage localaddr;
1357         struct sctp_event_subscribe subscribe;
1358         int result = -EINVAL, num = 1, i, addr_len;
1359         struct connection *con = nodeid2con(0, GFP_NOFS);
1360         int bufsize = NEEDED_RMEM;
1361         int one = 1;
1362
1363         if (!con)
1364                 return -ENOMEM;
1365
1366         log_print("Using SCTP for communications");
1367
1368         result = sock_create_kern(dlm_local_addr[0]->ss_family, SOCK_SEQPACKET,
1369                                   IPPROTO_SCTP, &sock);
1370         if (result < 0) {
1371                 log_print("Can't create comms socket, check SCTP is loaded");
1372                 goto out;
1373         }
1374
1375         /* Listen for events */
1376         memset(&subscribe, 0, sizeof(subscribe));
1377         subscribe.sctp_data_io_event = 1;
1378         subscribe.sctp_association_event = 1;
1379         subscribe.sctp_send_failure_event = 1;
1380         subscribe.sctp_shutdown_event = 1;
1381         subscribe.sctp_partial_delivery_event = 1;
1382
1383         result = kernel_setsockopt(sock, SOL_SOCKET, SO_RCVBUFFORCE,
1384                                  (char *)&bufsize, sizeof(bufsize));
1385         if (result)
1386                 log_print("Error increasing buffer space on socket %d", result);
1387
1388         result = kernel_setsockopt(sock, SOL_SCTP, SCTP_EVENTS,
1389                                    (char *)&subscribe, sizeof(subscribe));
1390         if (result < 0) {
1391                 log_print("Failed to set SCTP_EVENTS on socket: result=%d",
1392                           result);
1393                 goto create_delsock;
1394         }
1395
1396         result = kernel_setsockopt(sock, SOL_SCTP, SCTP_NODELAY, (char *)&one,
1397                                    sizeof(one));
1398         if (result < 0)
1399                 log_print("Could not set SCTP NODELAY error %d\n", result);
1400
1401         /* Init con struct */
1402         sock->sk->sk_user_data = con;
1403         con->sock = sock;
1404         con->sock->sk->sk_data_ready = lowcomms_data_ready;
1405         con->rx_action = receive_from_sock;
1406         con->connect_action = sctp_init_assoc;
1407
1408         /* Bind to all interfaces. */
1409         for (i = 0; i < dlm_local_count; i++) {
1410                 memcpy(&localaddr, dlm_local_addr[i], sizeof(localaddr));
1411                 make_sockaddr(&localaddr, dlm_config.ci_tcp_port, &addr_len);
1412
1413                 result = add_sctp_bind_addr(con, &localaddr, addr_len, num);
1414                 if (result)
1415                         goto create_delsock;
1416                 ++num;
1417         }
1418
1419         result = sock->ops->listen(sock, 5);
1420         if (result < 0) {
1421                 log_print("Can't set socket listening");
1422                 goto create_delsock;
1423         }
1424
1425         return 0;
1426
1427 create_delsock:
1428         sock_release(sock);
1429         con->sock = NULL;
1430 out:
1431         return result;
1432 }
1433
1434 static int tcp_listen_for_all(void)
1435 {
1436         struct socket *sock = NULL;
1437         struct connection *con = nodeid2con(0, GFP_NOFS);
1438         int result = -EINVAL;
1439
1440         if (!con)
1441                 return -ENOMEM;
1442
1443         /* We don't support multi-homed hosts */
1444         if (dlm_local_addr[1] != NULL) {
1445                 log_print("TCP protocol can't handle multi-homed hosts, "
1446                           "try SCTP");
1447                 return -EINVAL;
1448         }
1449
1450         log_print("Using TCP for communications");
1451
1452         sock = tcp_create_listen_sock(con, dlm_local_addr[0]);
1453         if (sock) {
1454                 add_sock(sock, con);
1455                 result = 0;
1456         }
1457         else {
1458                 result = -EADDRINUSE;
1459         }
1460
1461         return result;
1462 }
1463
1464
1465
1466 static struct writequeue_entry *new_writequeue_entry(struct connection *con,
1467                                                      gfp_t allocation)
1468 {
1469         struct writequeue_entry *entry;
1470
1471         entry = kmalloc(sizeof(struct writequeue_entry), allocation);
1472         if (!entry)
1473                 return NULL;
1474
1475         entry->page = alloc_page(allocation);
1476         if (!entry->page) {
1477                 kfree(entry);
1478                 return NULL;
1479         }
1480
1481         entry->offset = 0;
1482         entry->len = 0;
1483         entry->end = 0;
1484         entry->users = 0;
1485         entry->con = con;
1486
1487         return entry;
1488 }
1489
1490 void *dlm_lowcomms_get_buffer(int nodeid, int len, gfp_t allocation, char **ppc)
1491 {
1492         struct connection *con;
1493         struct writequeue_entry *e;
1494         int offset = 0;
1495
1496         con = nodeid2con(nodeid, allocation);
1497         if (!con)
1498                 return NULL;
1499
1500         spin_lock(&con->writequeue_lock);
1501         e = list_entry(con->writequeue.prev, struct writequeue_entry, list);
1502         if ((&e->list == &con->writequeue) ||
1503             (PAGE_CACHE_SIZE - e->end < len)) {
1504                 e = NULL;
1505         } else {
1506                 offset = e->end;
1507                 e->end += len;
1508                 e->users++;
1509         }
1510         spin_unlock(&con->writequeue_lock);
1511
1512         if (e) {
1513         got_one:
1514                 *ppc = page_address(e->page) + offset;
1515                 return e;
1516         }
1517
1518         e = new_writequeue_entry(con, allocation);
1519         if (e) {
1520                 spin_lock(&con->writequeue_lock);
1521                 offset = e->end;
1522                 e->end += len;
1523                 e->users++;
1524                 list_add_tail(&e->list, &con->writequeue);
1525                 spin_unlock(&con->writequeue_lock);
1526                 goto got_one;
1527         }
1528         return NULL;
1529 }
1530
1531 void dlm_lowcomms_commit_buffer(void *mh)
1532 {
1533         struct writequeue_entry *e = (struct writequeue_entry *)mh;
1534         struct connection *con = e->con;
1535         int users;
1536
1537         spin_lock(&con->writequeue_lock);
1538         users = --e->users;
1539         if (users)
1540                 goto out;
1541         e->len = e->end - e->offset;
1542         spin_unlock(&con->writequeue_lock);
1543
1544         if (!test_and_set_bit(CF_WRITE_PENDING, &con->flags)) {
1545                 queue_work(send_workqueue, &con->swork);
1546         }
1547         return;
1548
1549 out:
1550         spin_unlock(&con->writequeue_lock);
1551         return;
1552 }
1553
1554 /* Send a message */
1555 static void send_to_sock(struct connection *con)
1556 {
1557         int ret = 0;
1558         const int msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL;
1559         struct writequeue_entry *e;
1560         int len, offset;
1561         int count = 0;
1562
1563         mutex_lock(&con->sock_mutex);
1564         if (con->sock == NULL)
1565                 goto out_connect;
1566
1567         spin_lock(&con->writequeue_lock);
1568         for (;;) {
1569                 e = list_entry(con->writequeue.next, struct writequeue_entry,
1570                                list);
1571                 if ((struct list_head *) e == &con->writequeue)
1572                         break;
1573
1574                 len = e->len;
1575                 offset = e->offset;
1576                 BUG_ON(len == 0 && e->users == 0);
1577                 spin_unlock(&con->writequeue_lock);
1578
1579                 ret = 0;
1580                 if (len) {
1581                         ret = kernel_sendpage(con->sock, e->page, offset, len,
1582                                               msg_flags);
1583                         if (ret == -EAGAIN || ret == 0) {
1584                                 if (ret == -EAGAIN &&
1585                                     test_bit(SOCK_ASYNC_NOSPACE, &con->sock->flags) &&
1586                                     !test_and_set_bit(CF_APP_LIMITED, &con->flags)) {
1587                                         /* Notify TCP that we're limited by the
1588                                          * application window size.
1589                                          */
1590                                         set_bit(SOCK_NOSPACE, &con->sock->flags);
1591                                         con->sock->sk->sk_write_pending++;
1592                                 }
1593                                 cond_resched();
1594                                 goto out;
1595                         } else if (ret < 0)
1596                                 goto send_error;
1597                 }
1598
1599                 /* Don't starve people filling buffers */
1600                 if (++count >= MAX_SEND_MSG_COUNT) {
1601                         cond_resched();
1602                         count = 0;
1603                 }
1604
1605                 spin_lock(&con->writequeue_lock);
1606                 writequeue_entry_complete(e, ret);
1607         }
1608         spin_unlock(&con->writequeue_lock);
1609 out:
1610         mutex_unlock(&con->sock_mutex);
1611         return;
1612
1613 send_error:
1614         mutex_unlock(&con->sock_mutex);
1615         close_connection(con, false);
1616         lowcomms_connect_sock(con);
1617         return;
1618
1619 out_connect:
1620         mutex_unlock(&con->sock_mutex);
1621         if (!test_bit(CF_INIT_PENDING, &con->flags))
1622                 lowcomms_connect_sock(con);
1623 }
1624
1625 static void clean_one_writequeue(struct connection *con)
1626 {
1627         struct writequeue_entry *e, *safe;
1628
1629         spin_lock(&con->writequeue_lock);
1630         list_for_each_entry_safe(e, safe, &con->writequeue, list) {
1631                 list_del(&e->list);
1632                 free_entry(e);
1633         }
1634         spin_unlock(&con->writequeue_lock);
1635 }
1636
1637 /* Called from recovery when it knows that a node has
1638    left the cluster */
1639 int dlm_lowcomms_close(int nodeid)
1640 {
1641         struct connection *con;
1642         struct dlm_node_addr *na;
1643
1644         log_print("closing connection to node %d", nodeid);
1645         con = nodeid2con(nodeid, 0);
1646         if (con) {
1647                 clear_bit(CF_CONNECT_PENDING, &con->flags);
1648                 clear_bit(CF_WRITE_PENDING, &con->flags);
1649                 set_bit(CF_CLOSE, &con->flags);
1650                 if (cancel_work_sync(&con->swork))
1651                         log_print("canceled swork for node %d", nodeid);
1652                 if (cancel_work_sync(&con->rwork))
1653                         log_print("canceled rwork for node %d", nodeid);
1654                 clean_one_writequeue(con);
1655                 close_connection(con, true);
1656         }
1657
1658         spin_lock(&dlm_node_addrs_spin);
1659         na = find_node_addr(nodeid);
1660         if (na) {
1661                 list_del(&na->list);
1662                 while (na->addr_count--)
1663                         kfree(na->addr[na->addr_count]);
1664                 kfree(na);
1665         }
1666         spin_unlock(&dlm_node_addrs_spin);
1667
1668         return 0;
1669 }
1670
1671 /* Receive workqueue function */
1672 static void process_recv_sockets(struct work_struct *work)
1673 {
1674         struct connection *con = container_of(work, struct connection, rwork);
1675         int err;
1676
1677         clear_bit(CF_READ_PENDING, &con->flags);
1678         do {
1679                 err = con->rx_action(con);
1680         } while (!err);
1681 }
1682
1683 /* Send workqueue function */
1684 static void process_send_sockets(struct work_struct *work)
1685 {
1686         struct connection *con = container_of(work, struct connection, swork);
1687
1688         if (test_and_clear_bit(CF_CONNECT_PENDING, &con->flags)) {
1689                 con->connect_action(con);
1690                 set_bit(CF_WRITE_PENDING, &con->flags);
1691         }
1692         if (test_and_clear_bit(CF_WRITE_PENDING, &con->flags))
1693                 send_to_sock(con);
1694 }
1695
1696
1697 /* Discard all entries on the write queues */
1698 static void clean_writequeues(void)
1699 {
1700         foreach_conn(clean_one_writequeue);
1701 }
1702
1703 static void work_stop(void)
1704 {
1705         destroy_workqueue(recv_workqueue);
1706         destroy_workqueue(send_workqueue);
1707 }
1708
1709 static int work_start(void)
1710 {
1711         recv_workqueue = alloc_workqueue("dlm_recv",
1712                                          WQ_UNBOUND | WQ_MEM_RECLAIM, 1);
1713         if (!recv_workqueue) {
1714                 log_print("can't start dlm_recv");
1715                 return -ENOMEM;
1716         }
1717
1718         send_workqueue = alloc_workqueue("dlm_send",
1719                                          WQ_UNBOUND | WQ_MEM_RECLAIM, 1);
1720         if (!send_workqueue) {
1721                 log_print("can't start dlm_send");
1722                 destroy_workqueue(recv_workqueue);
1723                 return -ENOMEM;
1724         }
1725
1726         return 0;
1727 }
1728
1729 static void stop_conn(struct connection *con)
1730 {
1731         con->flags |= 0x0F;
1732         if (con->sock && con->sock->sk)
1733                 con->sock->sk->sk_user_data = NULL;
1734 }
1735
1736 static void free_conn(struct connection *con)
1737 {
1738         close_connection(con, true);
1739         if (con->othercon)
1740                 kmem_cache_free(con_cache, con->othercon);
1741         hlist_del(&con->list);
1742         kmem_cache_free(con_cache, con);
1743 }
1744
1745 void dlm_lowcomms_stop(void)
1746 {
1747         /* Set all the flags to prevent any
1748            socket activity.
1749         */
1750         mutex_lock(&connections_lock);
1751         dlm_allow_conn = 0;
1752         foreach_conn(stop_conn);
1753         mutex_unlock(&connections_lock);
1754
1755         work_stop();
1756
1757         mutex_lock(&connections_lock);
1758         clean_writequeues();
1759
1760         foreach_conn(free_conn);
1761
1762         mutex_unlock(&connections_lock);
1763         kmem_cache_destroy(con_cache);
1764 }
1765
1766 int dlm_lowcomms_start(void)
1767 {
1768         int error = -EINVAL;
1769         struct connection *con;
1770         int i;
1771
1772         for (i = 0; i < CONN_HASH_SIZE; i++)
1773                 INIT_HLIST_HEAD(&connection_hash[i]);
1774
1775         init_local();
1776         if (!dlm_local_count) {
1777                 error = -ENOTCONN;
1778                 log_print("no local IP address has been set");
1779                 goto fail;
1780         }
1781
1782         error = -ENOMEM;
1783         con_cache = kmem_cache_create("dlm_conn", sizeof(struct connection),
1784                                       __alignof__(struct connection), 0,
1785                                       NULL);
1786         if (!con_cache)
1787                 goto fail;
1788
1789         error = work_start();
1790         if (error)
1791                 goto fail_destroy;
1792
1793         dlm_allow_conn = 1;
1794
1795         /* Start listening */
1796         if (dlm_config.ci_protocol == 0)
1797                 error = tcp_listen_for_all();
1798         else
1799                 error = sctp_listen_for_all();
1800         if (error)
1801                 goto fail_unlisten;
1802
1803         return 0;
1804
1805 fail_unlisten:
1806         dlm_allow_conn = 0;
1807         con = nodeid2con(0,0);
1808         if (con) {
1809                 close_connection(con, false);
1810                 kmem_cache_free(con_cache, con);
1811         }
1812 fail_destroy:
1813         kmem_cache_destroy(con_cache);
1814 fail:
1815         return error;
1816 }
1817
1818 void dlm_lowcomms_exit(void)
1819 {
1820         struct dlm_node_addr *na, *safe;
1821
1822         spin_lock(&dlm_node_addrs_spin);
1823         list_for_each_entry_safe(na, safe, &dlm_node_addrs, list) {
1824                 list_del(&na->list);
1825                 while (na->addr_count--)
1826                         kfree(na->addr[na->addr_count]);
1827                 kfree(na);
1828         }
1829         spin_unlock(&dlm_node_addrs_spin);
1830 }