drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd_cb.c

   1 /*
   2  * GPL HEADER START
   3  *
   4  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   5  *
   6  * This program is free software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License version 2 only,
   8  * as published by the Free Software Foundation.
   9  *
  10  * This program is distributed in the hope that it will be useful, but
  11  * WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  * General Public License version 2 for more details (a copy is included
  14  * in the LICENSE file that accompanied this code).
  15  *
  16  * You should have received a copy of the GNU General Public License
  17  * version 2 along with this program; If not, see
  18  * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
  19  *
  20  * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
  21  * CA 95054 USA or visit www.sun.com if you need additional information or
  22  * have any questions.
  23  *
  24  * GPL HEADER END
  25  */
  26 /*
  27  * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
  28  * Use is subject to license terms.
  29  *
  30  * Copyright (c) 2012, Intel Corporation.
  31  */
  32 /*
  33  * This file is part of Lustre, http://www.lustre.org/
  34  * Lustre is a trademark of Sun Microsystems, Inc.
  35  *
  36  * lnet/klnds/o2iblnd/o2iblnd_cb.c
  37  *
  38  * Author: Eric Barton <eric@bartonsoftware.com>
  39  */
  40
  41 #include "o2iblnd.h"
  42
  43 void
  44 kiblnd_tx_done (lnet_ni_t *ni, kib_tx_t *tx)
  45 {
  46         lnet_msg_t *lntmsg[2];
  47         kib_net_t  *net = ni->ni_data;
  48         int      rc;
  49         int      i;
  50
  51         LASSERT (net != NULL);
  52         LASSERT (!in_interrupt());
  53         LASSERT (!tx->tx_queued);              /* mustn't be queued for sending */
  54         LASSERT (tx->tx_sending == 0);    /* mustn't be awaiting sent callback */
  55         LASSERT (!tx->tx_waiting);            /* mustn't be awaiting peer response */
  56         LASSERT (tx->tx_pool != NULL);
  57
  58         kiblnd_unmap_tx(ni, tx);
  59
  60         /* tx may have up to 2 lnet msgs to finalise */
  61         lntmsg[0] = tx->tx_lntmsg[0]; tx->tx_lntmsg[0] = NULL;
  62         lntmsg[1] = tx->tx_lntmsg[1]; tx->tx_lntmsg[1] = NULL;
  63         rc = tx->tx_status;
  64
  65         if (tx->tx_conn != NULL) {
  66                 LASSERT (ni == tx->tx_conn->ibc_peer->ibp_ni);
  67
  68                 kiblnd_conn_decref(tx->tx_conn);
  69                 tx->tx_conn = NULL;
  70         }
  71
  72         tx->tx_nwrq = 0;
  73         tx->tx_status = 0;
  74
  75         kiblnd_pool_free_node(&tx->tx_pool->tpo_pool, &tx->tx_list);
  76
  77         /* delay finalize until my descs have been freed */
  78         for (i = 0; i < 2; i++) {
  79                 if (lntmsg[i] == NULL)
  80                         continue;
  81
  82                 lnet_finalize(ni, lntmsg[i], rc);
  83         }
  84 }
  85
  86 void
  87 kiblnd_txlist_done (lnet_ni_t *ni, struct list_head *txlist, int status)
  88 {
  89         kib_tx_t *tx;
  90
  91         while (!list_empty (txlist)) {
  92                 tx = list_entry (txlist->next, kib_tx_t, tx_list);
  93
  94                 list_del(&tx->tx_list);
  95                 /* complete now */
  96                 tx->tx_waiting = 0;
  97                 tx->tx_status = status;
  98                 kiblnd_tx_done(ni, tx);
  99         }
 100 }
 101
 102 kib_tx_t *
 103 kiblnd_get_idle_tx(lnet_ni_t *ni, lnet_nid_t target)
 104 {
 105         kib_net_t               *net = (kib_net_t *)ni->ni_data;
 106         struct list_head                *node;
 107         kib_tx_t                *tx;
 108         kib_tx_poolset_t        *tps;
 109
 110         tps = net->ibn_tx_ps[lnet_cpt_of_nid(target)];
 111         node = kiblnd_pool_alloc_node(&tps->tps_poolset);
 112         if (node == NULL)
 113                 return NULL;
 114         tx = container_of(node, kib_tx_t, tx_list);
 115
 116         LASSERT (tx->tx_nwrq == 0);
 117         LASSERT (!tx->tx_queued);
 118         LASSERT (tx->tx_sending == 0);
 119         LASSERT (!tx->tx_waiting);
 120         LASSERT (tx->tx_status == 0);
 121         LASSERT (tx->tx_conn == NULL);
 122         LASSERT (tx->tx_lntmsg[0] == NULL);
 123         LASSERT (tx->tx_lntmsg[1] == NULL);
 124         LASSERT (tx->tx_u.pmr == NULL);
 125         LASSERT (tx->tx_nfrags == 0);
 126
 127         return tx;
 128 }
 129
 130 void
 131 kiblnd_drop_rx(kib_rx_t *rx)
 132 {
 133         kib_conn_t              *conn   = rx->rx_conn;
 134         struct kib_sched_info   *sched  = conn->ibc_sched;
 135         unsigned long           flags;
 136
 137         spin_lock_irqsave(&sched->ibs_lock, flags);
 138         LASSERT(conn->ibc_nrx > 0);
 139         conn->ibc_nrx--;
 140         spin_unlock_irqrestore(&sched->ibs_lock, flags);
 141
 142         kiblnd_conn_decref(conn);
 143 }
 144
 145 int
 146 kiblnd_post_rx (kib_rx_t *rx, int credit)
 147 {
 148         kib_conn_t       *conn = rx->rx_conn;
 149         kib_net_t         *net = conn->ibc_peer->ibp_ni->ni_data;
 150         struct ib_recv_wr  *bad_wrq = NULL;
 151         struct ib_mr       *mr;
 152         int              rc;
 153
 154         LASSERT (net != NULL);
 155         LASSERT (!in_interrupt());
 156         LASSERT (credit == IBLND_POSTRX_NO_CREDIT ||
 157                  credit == IBLND_POSTRX_PEER_CREDIT ||
 158                  credit == IBLND_POSTRX_RSRVD_CREDIT);
 159
 160         mr = kiblnd_find_dma_mr(conn->ibc_hdev, rx->rx_msgaddr, IBLND_MSG_SIZE);
 161         LASSERT (mr != NULL);
 162
 163         rx->rx_sge.lkey   = mr->lkey;
 164         rx->rx_sge.addr   = rx->rx_msgaddr;
 165         rx->rx_sge.length = IBLND_MSG_SIZE;
 166
 167         rx->rx_wrq.next = NULL;
 168         rx->rx_wrq.sg_list = &rx->rx_sge;
 169         rx->rx_wrq.num_sge = 1;
 170         rx->rx_wrq.wr_id = kiblnd_ptr2wreqid(rx, IBLND_WID_RX);
 171
 172         LASSERT (conn->ibc_state >= IBLND_CONN_INIT);
 173         LASSERT (rx->rx_nob >= 0);            /* not posted */
 174
 175         if (conn->ibc_state > IBLND_CONN_ESTABLISHED) {
 176                 kiblnd_drop_rx(rx);          /* No more posts for this rx */
 177                 return 0;
 178         }
 179
 180         rx->rx_nob = -1;                        /* flag posted */
 181
 182         rc = ib_post_recv(conn->ibc_cmid->qp, &rx->rx_wrq, &bad_wrq);
 183         if (rc != 0) {
 184                 CERROR("Can't post rx for %s: %d, bad_wrq: %p\n",
 185                        libcfs_nid2str(conn->ibc_peer->ibp_nid), rc, bad_wrq);
 186                 rx->rx_nob = 0;
 187         }
 188
 189         if (conn->ibc_state < IBLND_CONN_ESTABLISHED) /* Initial post */
 190                 return rc;
 191
 192         if (rc != 0) {
 193                 kiblnd_close_conn(conn, rc);
 194                 kiblnd_drop_rx(rx);          /* No more posts for this rx */
 195                 return rc;
 196         }
 197
 198         if (credit == IBLND_POSTRX_NO_CREDIT)
 199                 return 0;
 200
 201         spin_lock(&conn->ibc_lock);
 202         if (credit == IBLND_POSTRX_PEER_CREDIT)
 203                 conn->ibc_outstanding_credits++;
 204         else
 205                 conn->ibc_reserved_credits++;
 206         spin_unlock(&conn->ibc_lock);
 207
 208         kiblnd_check_sends(conn);
 209         return 0;
 210 }
 211
 212 kib_tx_t *
 213 kiblnd_find_waiting_tx_locked(kib_conn_t *conn, int txtype, __u64 cookie)
 214 {
 215         struct list_head   *tmp;
 216
 217         list_for_each(tmp, &conn->ibc_active_txs) {
 218                 kib_tx_t *tx = list_entry(tmp, kib_tx_t, tx_list);
 219
 220                 LASSERT (!tx->tx_queued);
 221                 LASSERT (tx->tx_sending != 0 || tx->tx_waiting);
 222
 223                 if (tx->tx_cookie != cookie)
 224                         continue;
 225
 226                 if (tx->tx_waiting &&
 227                     tx->tx_msg->ibm_type == txtype)
 228                         return tx;
 229
 230                 CWARN("Bad completion: %swaiting, type %x (wanted %x)\n",
 231                       tx->tx_waiting ? "" : "NOT ",
 232                       tx->tx_msg->ibm_type, txtype);
 233         }
 234         return NULL;
 235 }
 236
 237 void
 238 kiblnd_handle_completion(kib_conn_t *conn, int txtype, int status, __u64 cookie)
 239 {
 240         kib_tx_t    *tx;
 241         lnet_ni_t   *ni = conn->ibc_peer->ibp_ni;
 242         int       idle;
 243
 244         spin_lock(&conn->ibc_lock);
 245
 246         tx = kiblnd_find_waiting_tx_locked(conn, txtype, cookie);
 247         if (tx == NULL) {
 248                 spin_unlock(&conn->ibc_lock);
 249
 250                 CWARN("Unmatched completion type %x cookie "LPX64" from %s\n",
 251                       txtype, cookie, libcfs_nid2str(conn->ibc_peer->ibp_nid));
 252                 kiblnd_close_conn(conn, -EPROTO);
 253                 return;
 254         }
 255
 256         if (tx->tx_status == 0) {              /* success so far */
 257                 if (status < 0) {              /* failed? */
 258                         tx->tx_status = status;
 259                 } else if (txtype == IBLND_MSG_GET_REQ) {
 260                         lnet_set_reply_msg_len(ni, tx->tx_lntmsg[1], status);
 261                 }
 262         }
 263
 264         tx->tx_waiting = 0;
 265
 266         idle = !tx->tx_queued && (tx->tx_sending == 0);
 267         if (idle)
 268                 list_del(&tx->tx_list);
 269
 270         spin_unlock(&conn->ibc_lock);
 271
 272         if (idle)
 273                 kiblnd_tx_done(ni, tx);
 274 }
 275
 276 void
 277 kiblnd_send_completion(kib_conn_t *conn, int type, int status, __u64 cookie)
 278 {
 279         lnet_ni_t   *ni = conn->ibc_peer->ibp_ni;
 280         kib_tx_t    *tx = kiblnd_get_idle_tx(ni, conn->ibc_peer->ibp_nid);
 281
 282         if (tx == NULL) {
 283                 CERROR("Can't get tx for completion %x for %s\n",
 284                        type, libcfs_nid2str(conn->ibc_peer->ibp_nid));
 285                 return;
 286         }
 287
 288         tx->tx_msg->ibm_u.completion.ibcm_status = status;
 289         tx->tx_msg->ibm_u.completion.ibcm_cookie = cookie;
 290         kiblnd_init_tx_msg(ni, tx, type, sizeof(kib_completion_msg_t));
 291
 292         kiblnd_queue_tx(tx, conn);
 293 }
 294
 295 void
 296 kiblnd_handle_rx (kib_rx_t *rx)
 297 {
 298         kib_msg_t    *msg = rx->rx_msg;
 299         kib_conn_t   *conn = rx->rx_conn;
 300         lnet_ni_t    *ni = conn->ibc_peer->ibp_ni;
 301         int        credits = msg->ibm_credits;
 302         kib_tx_t     *tx;
 303         int        rc = 0;
 304         int        rc2;
 305         int        post_credit;
 306
 307         LASSERT (conn->ibc_state >= IBLND_CONN_ESTABLISHED);
 308
 309         CDEBUG (D_NET, "Received %x[%d] from %s\n",
 310                 msg->ibm_type, credits,
 311                 libcfs_nid2str(conn->ibc_peer->ibp_nid));
 312
 313         if (credits != 0) {
 314                 /* Have I received credits that will let me send? */
 315                 spin_lock(&conn->ibc_lock);
 316
 317                 if (conn->ibc_credits + credits >
 318                     IBLND_MSG_QUEUE_SIZE(conn->ibc_version)) {
 319                         rc2 = conn->ibc_credits;
 320                         spin_unlock(&conn->ibc_lock);
 321
 322                         CERROR("Bad credits from %s: %d + %d > %d\n",
 323                                libcfs_nid2str(conn->ibc_peer->ibp_nid),
 324                                rc2, credits,
 325                                IBLND_MSG_QUEUE_SIZE(conn->ibc_version));
 326
 327                         kiblnd_close_conn(conn, -EPROTO);
 328                         kiblnd_post_rx(rx, IBLND_POSTRX_NO_CREDIT);
 329                         return;
 330                 }
 331
 332                 conn->ibc_credits += credits;
 333
 334                 /* This ensures the credit taken by NOOP can be returned */
 335                 if (msg->ibm_type == IBLND_MSG_NOOP &&
 336                     !IBLND_OOB_CAPABLE(conn->ibc_version)) /* v1 only */
 337                         conn->ibc_outstanding_credits++;
 338
 339                 spin_unlock(&conn->ibc_lock);
 340                 kiblnd_check_sends(conn);
 341         }
 342
 343         switch (msg->ibm_type) {
 344         default:
 345                 CERROR("Bad IBLND message type %x from %s\n",
 346                        msg->ibm_type, libcfs_nid2str(conn->ibc_peer->ibp_nid));
 347                 post_credit = IBLND_POSTRX_NO_CREDIT;
 348                 rc = -EPROTO;
 349                 break;
 350
 351         case IBLND_MSG_NOOP:
 352                 if (IBLND_OOB_CAPABLE(conn->ibc_version)) {
 353                         post_credit = IBLND_POSTRX_NO_CREDIT;
 354                         break;
 355                 }
 356
 357                 if (credits != 0) /* credit already posted */
 358                         post_credit = IBLND_POSTRX_NO_CREDIT;
 359                 else          /* a keepalive NOOP */
 360                         post_credit = IBLND_POSTRX_PEER_CREDIT;
 361                 break;
 362
 363         case IBLND_MSG_IMMEDIATE:
 364                 post_credit = IBLND_POSTRX_DONT_POST;
 365                 rc = lnet_parse(ni, &msg->ibm_u.immediate.ibim_hdr,
 366                                 msg->ibm_srcnid, rx, 0);
 367                 if (rc < 0)                  /* repost on error */
 368                         post_credit = IBLND_POSTRX_PEER_CREDIT;
 369                 break;
 370
 371         case IBLND_MSG_PUT_REQ:
 372                 post_credit = IBLND_POSTRX_DONT_POST;
 373                 rc = lnet_parse(ni, &msg->ibm_u.putreq.ibprm_hdr,
 374                                 msg->ibm_srcnid, rx, 1);
 375                 if (rc < 0)                  /* repost on error */
 376                         post_credit = IBLND_POSTRX_PEER_CREDIT;
 377                 break;
 378
 379         case IBLND_MSG_PUT_NAK:
 380                 CWARN ("PUT_NACK from %s\n",
 381                        libcfs_nid2str(conn->ibc_peer->ibp_nid));
 382                 post_credit = IBLND_POSTRX_RSRVD_CREDIT;
 383                 kiblnd_handle_completion(conn, IBLND_MSG_PUT_REQ,
 384                                          msg->ibm_u.completion.ibcm_status,
 385                                          msg->ibm_u.completion.ibcm_cookie);
 386                 break;
 387
 388         case IBLND_MSG_PUT_ACK:
 389                 post_credit = IBLND_POSTRX_RSRVD_CREDIT;
 390
 391                 spin_lock(&conn->ibc_lock);
 392                 tx = kiblnd_find_waiting_tx_locked(conn, IBLND_MSG_PUT_REQ,
 393                                         msg->ibm_u.putack.ibpam_src_cookie);
 394                 if (tx != NULL)
 395                         list_del(&tx->tx_list);
 396                 spin_unlock(&conn->ibc_lock);
 397
 398                 if (tx == NULL) {
 399                         CERROR("Unmatched PUT_ACK from %s\n",
 400                                libcfs_nid2str(conn->ibc_peer->ibp_nid));
 401                         rc = -EPROTO;
 402                         break;
 403                 }
 404
 405                 LASSERT (tx->tx_waiting);
 406                 /* CAVEAT EMPTOR: I could be racing with tx_complete, but...
 407                  * (a) I can overwrite tx_msg since my peer has received it!
 408                  * (b) tx_waiting set tells tx_complete() it's not done. */
 409
 410                 tx->tx_nwrq = 0;                /* overwrite PUT_REQ */
 411
 412                 rc2 = kiblnd_init_rdma(conn, tx, IBLND_MSG_PUT_DONE,
 413                                        kiblnd_rd_size(&msg->ibm_u.putack.ibpam_rd),
 414                                        &msg->ibm_u.putack.ibpam_rd,
 415                                        msg->ibm_u.putack.ibpam_dst_cookie);
 416                 if (rc2 < 0)
 417                         CERROR("Can't setup rdma for PUT to %s: %d\n",
 418                                libcfs_nid2str(conn->ibc_peer->ibp_nid), rc2);
 419
 420                 spin_lock(&conn->ibc_lock);
 421                 tx->tx_waiting = 0;     /* clear waiting and queue atomically */
 422                 kiblnd_queue_tx_locked(tx, conn);
 423                 spin_unlock(&conn->ibc_lock);
 424                 break;
 425
 426         case IBLND_MSG_PUT_DONE:
 427                 post_credit = IBLND_POSTRX_PEER_CREDIT;
 428                 kiblnd_handle_completion(conn, IBLND_MSG_PUT_ACK,
 429                                          msg->ibm_u.completion.ibcm_status,
 430                                          msg->ibm_u.completion.ibcm_cookie);
 431                 break;
 432
 433         case IBLND_MSG_GET_REQ:
 434                 post_credit = IBLND_POSTRX_DONT_POST;
 435                 rc = lnet_parse(ni, &msg->ibm_u.get.ibgm_hdr,
 436                                 msg->ibm_srcnid, rx, 1);
 437                 if (rc < 0)                  /* repost on error */
 438                         post_credit = IBLND_POSTRX_PEER_CREDIT;
 439                 break;
 440
 441         case IBLND_MSG_GET_DONE:
 442                 post_credit = IBLND_POSTRX_RSRVD_CREDIT;
 443                 kiblnd_handle_completion(conn, IBLND_MSG_GET_REQ,
 444                                          msg->ibm_u.completion.ibcm_status,
 445                                          msg->ibm_u.completion.ibcm_cookie);
 446                 break;
 447         }
 448
 449         if (rc < 0)                          /* protocol error */
 450                 kiblnd_close_conn(conn, rc);
 451
 452         if (post_credit != IBLND_POSTRX_DONT_POST)
 453                 kiblnd_post_rx(rx, post_credit);
 454 }
 455
 456 void
 457 kiblnd_rx_complete (kib_rx_t *rx, int status, int nob)
 458 {
 459         kib_msg_t    *msg = rx->rx_msg;
 460         kib_conn_t   *conn = rx->rx_conn;
 461         lnet_ni_t    *ni = conn->ibc_peer->ibp_ni;
 462         kib_net_t    *net = ni->ni_data;
 463         int        rc;
 464         int        err = -EIO;
 465
 466         LASSERT (net != NULL);
 467         LASSERT (rx->rx_nob < 0);              /* was posted */
 468         rx->rx_nob = 0;                  /* isn't now */
 469
 470         if (conn->ibc_state > IBLND_CONN_ESTABLISHED)
 471                 goto ignore;
 472
 473         if (status != IB_WC_SUCCESS) {
 474                 CNETERR("Rx from %s failed: %d\n",
 475                         libcfs_nid2str(conn->ibc_peer->ibp_nid), status);
 476                 goto failed;
 477         }
 478
 479         LASSERT (nob >= 0);
 480         rx->rx_nob = nob;
 481
 482         rc = kiblnd_unpack_msg(msg, rx->rx_nob);
 483         if (rc != 0) {
 484                 CERROR ("Error %d unpacking rx from %s\n",
 485                         rc, libcfs_nid2str(conn->ibc_peer->ibp_nid));
 486                 goto failed;
 487         }
 488
 489         if (msg->ibm_srcnid != conn->ibc_peer->ibp_nid ||
 490             msg->ibm_dstnid != ni->ni_nid ||
 491             msg->ibm_srcstamp != conn->ibc_incarnation ||
 492             msg->ibm_dststamp != net->ibn_incarnation) {
 493                 CERROR ("Stale rx from %s\n",
 494                         libcfs_nid2str(conn->ibc_peer->ibp_nid));
 495                 err = -ESTALE;
 496                 goto failed;
 497         }
 498
 499         /* set time last known alive */
 500         kiblnd_peer_alive(conn->ibc_peer);
 501
 502         /* racing with connection establishment/teardown! */
 503
 504         if (conn->ibc_state < IBLND_CONN_ESTABLISHED) {
 505                 rwlock_t  *g_lock = &kiblnd_data.kib_global_lock;
 506                 unsigned long  flags;
 507
 508                 write_lock_irqsave(g_lock, flags);
 509                 /* must check holding global lock to eliminate race */
 510                 if (conn->ibc_state < IBLND_CONN_ESTABLISHED) {
 511                         list_add_tail(&rx->rx_list, &conn->ibc_early_rxs);
 512                         write_unlock_irqrestore(g_lock, flags);
 513                         return;
 514                 }
 515                 write_unlock_irqrestore(g_lock, flags);
 516         }
 517         kiblnd_handle_rx(rx);
 518         return;
 519
 520  failed:
 521         CDEBUG(D_NET, "rx %p conn %p\n", rx, conn);
 522         kiblnd_close_conn(conn, err);
 523  ignore:
 524         kiblnd_drop_rx(rx);                  /* Don't re-post rx. */
 525 }
 526
 527 struct page *
 528 kiblnd_kvaddr_to_page (unsigned long vaddr)
 529 {
 530         struct page *page;
 531
 532         if (vaddr >= VMALLOC_START &&
 533             vaddr < VMALLOC_END) {
 534                 page = vmalloc_to_page ((void *)vaddr);
 535                 LASSERT (page != NULL);
 536                 return page;
 537         }
 538 #ifdef CONFIG_HIGHMEM
 539         if (vaddr >= PKMAP_BASE &&
 540             vaddr < (PKMAP_BASE + LAST_PKMAP * PAGE_SIZE)) {
 541                 /* No highmem pages only used for bulk (kiov) I/O */
 542                 CERROR("find page for address in highmem\n");
 543                 LBUG();
 544         }
 545 #endif
 546         page = virt_to_page (vaddr);
 547         LASSERT (page != NULL);
 548         return page;
 549 }
 550
 551 static int
 552 kiblnd_fmr_map_tx(kib_net_t *net, kib_tx_t *tx, kib_rdma_desc_t *rd, int nob)
 553 {
 554         kib_hca_dev_t           *hdev;
 555         __u64                   *pages = tx->tx_pages;
 556         kib_fmr_poolset_t       *fps;
 557         int                     npages;
 558         int                     size;
 559         int                     cpt;
 560         int                     rc;
 561         int                     i;
 562
 563         LASSERT(tx->tx_pool != NULL);
 564         LASSERT(tx->tx_pool->tpo_pool.po_owner != NULL);
 565
 566         hdev  = tx->tx_pool->tpo_hdev;
 567
 568         for (i = 0, npages = 0; i < rd->rd_nfrags; i++) {
 569                 for (size = 0; size <  rd->rd_frags[i].rf_nob;
 570                                size += hdev->ibh_page_size) {
 571                         pages[npages ++] = (rd->rd_frags[i].rf_addr &
 572                                             hdev->ibh_page_mask) + size;
 573                 }
 574         }
 575
 576         cpt = tx->tx_pool->tpo_pool.po_owner->ps_cpt;
 577
 578         fps = net->ibn_fmr_ps[cpt];
 579         rc = kiblnd_fmr_pool_map(fps, pages, npages, 0, &tx->tx_u.fmr);
 580         if (rc != 0) {
 581                 CERROR ("Can't map %d pages: %d\n", npages, rc);
 582                 return rc;
 583         }
 584
 585         /* If rd is not tx_rd, it's going to get sent to a peer, who will need
 586          * the rkey */
 587         rd->rd_key = (rd != tx->tx_rd) ? tx->tx_u.fmr.fmr_pfmr->fmr->rkey :
 588                                          tx->tx_u.fmr.fmr_pfmr->fmr->lkey;
 589         rd->rd_frags[0].rf_addr &= ~hdev->ibh_page_mask;
 590         rd->rd_frags[0].rf_nob   = nob;
 591         rd->rd_nfrags = 1;
 592
 593         return 0;
 594 }
 595
 596 static int
 597 kiblnd_pmr_map_tx(kib_net_t *net, kib_tx_t *tx, kib_rdma_desc_t *rd, int nob)
 598 {
 599         kib_hca_dev_t           *hdev;
 600         kib_pmr_poolset_t       *pps;
 601         __u64                   iova;
 602         int                     cpt;
 603         int                     rc;
 604
 605         LASSERT(tx->tx_pool != NULL);
 606         LASSERT(tx->tx_pool->tpo_pool.po_owner != NULL);
 607
 608         hdev = tx->tx_pool->tpo_hdev;
 609
 610         iova = rd->rd_frags[0].rf_addr & ~hdev->ibh_page_mask;
 611
 612         cpt = tx->tx_pool->tpo_pool.po_owner->ps_cpt;
 613
 614         pps = net->ibn_pmr_ps[cpt];
 615         rc = kiblnd_pmr_pool_map(pps, hdev, rd, &iova, &tx->tx_u.pmr);
 616         if (rc != 0) {
 617                 CERROR("Failed to create MR by phybuf: %d\n", rc);
 618                 return rc;
 619         }
 620
 621         /* If rd is not tx_rd, it's going to get sent to a peer, who will need
 622          * the rkey */
 623         rd->rd_key = (rd != tx->tx_rd) ? tx->tx_u.pmr->pmr_mr->rkey :
 624                                          tx->tx_u.pmr->pmr_mr->lkey;
 625         rd->rd_nfrags = 1;
 626         rd->rd_frags[0].rf_addr = iova;
 627         rd->rd_frags[0].rf_nob  = nob;
 628
 629         return 0;
 630 }
 631
 632 void
 633 kiblnd_unmap_tx(lnet_ni_t *ni, kib_tx_t *tx)
 634 {
 635         kib_net_t  *net = ni->ni_data;
 636
 637         LASSERT(net != NULL);
 638
 639         if (net->ibn_fmr_ps != NULL && tx->tx_u.fmr.fmr_pfmr != NULL) {
 640                 kiblnd_fmr_pool_unmap(&tx->tx_u.fmr, tx->tx_status);
 641                 tx->tx_u.fmr.fmr_pfmr = NULL;
 642
 643         } else if (net->ibn_pmr_ps != NULL && tx->tx_u.pmr != NULL) {
 644                 kiblnd_pmr_pool_unmap(tx->tx_u.pmr);
 645                 tx->tx_u.pmr = NULL;
 646         }
 647
 648         if (tx->tx_nfrags != 0) {
 649                 kiblnd_dma_unmap_sg(tx->tx_pool->tpo_hdev->ibh_ibdev,
 650                                     tx->tx_frags, tx->tx_nfrags, tx->tx_dmadir);
 651                 tx->tx_nfrags = 0;
 652         }
 653 }
 654
 655 int
 656 kiblnd_map_tx(lnet_ni_t *ni, kib_tx_t *tx,
 657               kib_rdma_desc_t *rd, int nfrags)
 658 {
 659         kib_hca_dev_t      *hdev  = tx->tx_pool->tpo_hdev;
 660         kib_net_t         *net   = ni->ni_data;
 661         struct ib_mr       *mr    = NULL;
 662         __u32          nob;
 663         int              i;
 664
 665         /* If rd is not tx_rd, it's going to get sent to a peer and I'm the
 666          * RDMA sink */
 667         tx->tx_dmadir = (rd != tx->tx_rd) ? DMA_FROM_DEVICE : DMA_TO_DEVICE;
 668         tx->tx_nfrags = nfrags;
 669
 670         rd->rd_nfrags =
 671                 kiblnd_dma_map_sg(hdev->ibh_ibdev,
 672                                   tx->tx_frags, tx->tx_nfrags, tx->tx_dmadir);
 673
 674         for (i = 0, nob = 0; i < rd->rd_nfrags; i++) {
 675                 rd->rd_frags[i].rf_nob  = kiblnd_sg_dma_len(
 676                         hdev->ibh_ibdev, &tx->tx_frags[i]);
 677                 rd->rd_frags[i].rf_addr = kiblnd_sg_dma_address(
 678                         hdev->ibh_ibdev, &tx->tx_frags[i]);
 679                 nob += rd->rd_frags[i].rf_nob;
 680         }
 681
 682         /* looking for pre-mapping MR */
 683         mr = kiblnd_find_rd_dma_mr(hdev, rd);
 684         if (mr != NULL) {
 685                 /* found pre-mapping MR */
 686                 rd->rd_key = (rd != tx->tx_rd) ? mr->rkey : mr->lkey;
 687                 return 0;
 688         }
 689
 690         if (net->ibn_fmr_ps != NULL)
 691                 return kiblnd_fmr_map_tx(net, tx, rd, nob);
 692         else if (net->ibn_pmr_ps != NULL)
 693                 return kiblnd_pmr_map_tx(net, tx, rd, nob);
 694
 695         return -EINVAL;
 696 }
 697
 698
 699 int
 700 kiblnd_setup_rd_iov(lnet_ni_t *ni, kib_tx_t *tx, kib_rdma_desc_t *rd,
 701                     unsigned int niov, struct iovec *iov, int offset, int nob)
 702 {
 703         kib_net_t         *net = ni->ni_data;
 704         struct page     *page;
 705         struct scatterlist *sg;
 706         unsigned long       vaddr;
 707         int              fragnob;
 708         int              page_offset;
 709
 710         LASSERT (nob > 0);
 711         LASSERT (niov > 0);
 712         LASSERT (net != NULL);
 713
 714         while (offset >= iov->iov_len) {
 715                 offset -= iov->iov_len;
 716                 niov--;
 717                 iov++;
 718                 LASSERT (niov > 0);
 719         }
 720
 721         sg = tx->tx_frags;
 722         do {
 723                 LASSERT (niov > 0);
 724
 725                 vaddr = ((unsigned long)iov->iov_base) + offset;
 726                 page_offset = vaddr & (PAGE_SIZE - 1);
 727                 page = kiblnd_kvaddr_to_page(vaddr);
 728                 if (page == NULL) {
 729                         CERROR ("Can't find page\n");
 730                         return -EFAULT;
 731                 }
 732
 733                 fragnob = min((int)(iov->iov_len - offset), nob);
 734                 fragnob = min(fragnob, (int)PAGE_SIZE - page_offset);
 735
 736                 sg_set_page(sg, page, fragnob, page_offset);
 737                 sg++;
 738
 739                 if (offset + fragnob < iov->iov_len) {
 740                         offset += fragnob;
 741                 } else {
 742                         offset = 0;
 743                         iov++;
 744                         niov--;
 745                 }
 746                 nob -= fragnob;
 747         } while (nob > 0);
 748
 749         return kiblnd_map_tx(ni, tx, rd, sg - tx->tx_frags);
 750 }
 751
 752 int
 753 kiblnd_setup_rd_kiov (lnet_ni_t *ni, kib_tx_t *tx, kib_rdma_desc_t *rd,
 754                       int nkiov, lnet_kiov_t *kiov, int offset, int nob)
 755 {
 756         kib_net_t         *net = ni->ni_data;
 757         struct scatterlist *sg;
 758         int              fragnob;
 759
 760         CDEBUG(D_NET, "niov %d offset %d nob %d\n", nkiov, offset, nob);
 761
 762         LASSERT (nob > 0);
 763         LASSERT (nkiov > 0);
 764         LASSERT (net != NULL);
 765
 766         while (offset >= kiov->kiov_len) {
 767                 offset -= kiov->kiov_len;
 768                 nkiov--;
 769                 kiov++;
 770                 LASSERT (nkiov > 0);
 771         }
 772
 773         sg = tx->tx_frags;
 774         do {
 775                 LASSERT (nkiov > 0);
 776
 777                 fragnob = min((int)(kiov->kiov_len - offset), nob);
 778
 779                 sg_set_page(sg, kiov->kiov_page, fragnob,
 780                             kiov->kiov_offset + offset);
 781                 sg++;
 782
 783                 offset = 0;
 784                 kiov++;
 785                 nkiov--;
 786                 nob -= fragnob;
 787         } while (nob > 0);
 788
 789         return kiblnd_map_tx(ni, tx, rd, sg - tx->tx_frags);
 790 }
 791
 792 int
 793 kiblnd_post_tx_locked (kib_conn_t *conn, kib_tx_t *tx, int credit)
 794 {
 795         kib_msg_t        *msg = tx->tx_msg;
 796         kib_peer_t      *peer = conn->ibc_peer;
 797         int             ver = conn->ibc_version;
 798         int             rc;
 799         int             done;
 800         struct ib_send_wr *bad_wrq;
 801
 802         LASSERT (tx->tx_queued);
 803         /* We rely on this for QP sizing */
 804         LASSERT (tx->tx_nwrq > 0);
 805         LASSERT (tx->tx_nwrq <= 1 + IBLND_RDMA_FRAGS(ver));
 806
 807         LASSERT (credit == 0 || credit == 1);
 808         LASSERT (conn->ibc_outstanding_credits >= 0);
 809         LASSERT (conn->ibc_outstanding_credits <= IBLND_MSG_QUEUE_SIZE(ver));
 810         LASSERT (conn->ibc_credits >= 0);
 811         LASSERT (conn->ibc_credits <= IBLND_MSG_QUEUE_SIZE(ver));
 812
 813         if (conn->ibc_nsends_posted == IBLND_CONCURRENT_SENDS(ver)) {
 814                 /* tx completions outstanding... */
 815                 CDEBUG(D_NET, "%s: posted enough\n",
 816                        libcfs_nid2str(peer->ibp_nid));
 817                 return -EAGAIN;
 818         }
 819
 820         if (credit != 0 && conn->ibc_credits == 0) {   /* no credits */
 821                 CDEBUG(D_NET, "%s: no credits\n",
 822                        libcfs_nid2str(peer->ibp_nid));
 823                 return -EAGAIN;
 824         }
 825
 826         if (credit != 0 && !IBLND_OOB_CAPABLE(ver) &&
 827             conn->ibc_credits == 1 &&   /* last credit reserved */
 828             msg->ibm_type != IBLND_MSG_NOOP) {      /* for NOOP */
 829                 CDEBUG(D_NET, "%s: not using last credit\n",
 830                        libcfs_nid2str(peer->ibp_nid));
 831                 return -EAGAIN;
 832         }
 833
 834         /* NB don't drop ibc_lock before bumping tx_sending */
 835         list_del(&tx->tx_list);
 836         tx->tx_queued = 0;
 837
 838         if (msg->ibm_type == IBLND_MSG_NOOP &&
 839             (!kiblnd_need_noop(conn) ||     /* redundant NOOP */
 840              (IBLND_OOB_CAPABLE(ver) && /* posted enough NOOP */
 841               conn->ibc_noops_posted == IBLND_OOB_MSGS(ver)))) {
 842                 /* OK to drop when posted enough NOOPs, since
 843                  * kiblnd_check_sends will queue NOOP again when
 844                  * posted NOOPs complete */
 845                 spin_unlock(&conn->ibc_lock);
 846                 kiblnd_tx_done(peer->ibp_ni, tx);
 847                 spin_lock(&conn->ibc_lock);
 848                 CDEBUG(D_NET, "%s(%d): redundant or enough NOOP\n",
 849                        libcfs_nid2str(peer->ibp_nid),
 850                        conn->ibc_noops_posted);
 851                 return 0;
 852         }
 853
 854         kiblnd_pack_msg(peer->ibp_ni, msg, ver, conn->ibc_outstanding_credits,
 855                         peer->ibp_nid, conn->ibc_incarnation);
 856
 857         conn->ibc_credits -= credit;
 858         conn->ibc_outstanding_credits = 0;
 859         conn->ibc_nsends_posted++;
 860         if (msg->ibm_type == IBLND_MSG_NOOP)
 861                 conn->ibc_noops_posted++;
 862
 863         /* CAVEAT EMPTOR!  This tx could be the PUT_DONE of an RDMA
 864          * PUT.  If so, it was first queued here as a PUT_REQ, sent and
 865          * stashed on ibc_active_txs, matched by an incoming PUT_ACK,
 866          * and then re-queued here.  It's (just) possible that
 867          * tx_sending is non-zero if we've not done the tx_complete()
 868          * from the first send; hence the ++ rather than = below. */
 869         tx->tx_sending++;
 870         list_add(&tx->tx_list, &conn->ibc_active_txs);
 871
 872         /* I'm still holding ibc_lock! */
 873         if (conn->ibc_state != IBLND_CONN_ESTABLISHED) {
 874                 rc = -ECONNABORTED;
 875         } else if (tx->tx_pool->tpo_pool.po_failed ||
 876                  conn->ibc_hdev != tx->tx_pool->tpo_hdev) {
 877                 /* close_conn will launch failover */
 878                 rc = -ENETDOWN;
 879         } else {
 880                 rc = ib_post_send(conn->ibc_cmid->qp,
 881                                   tx->tx_wrq, &bad_wrq);
 882         }
 883
 884         conn->ibc_last_send = jiffies;
 885
 886         if (rc == 0)
 887                 return 0;
 888
 889         /* NB credits are transferred in the actual
 890          * message, which can only be the last work item */
 891         conn->ibc_credits += credit;
 892         conn->ibc_outstanding_credits += msg->ibm_credits;
 893         conn->ibc_nsends_posted--;
 894         if (msg->ibm_type == IBLND_MSG_NOOP)
 895                 conn->ibc_noops_posted--;
 896
 897         tx->tx_status = rc;
 898         tx->tx_waiting = 0;
 899         tx->tx_sending--;
 900
 901         done = (tx->tx_sending == 0);
 902         if (done)
 903                 list_del(&tx->tx_list);
 904
 905         spin_unlock(&conn->ibc_lock);
 906
 907         if (conn->ibc_state == IBLND_CONN_ESTABLISHED)
 908                 CERROR("Error %d posting transmit to %s\n",
 909                        rc, libcfs_nid2str(peer->ibp_nid));
 910         else
 911                 CDEBUG(D_NET, "Error %d posting transmit to %s\n",
 912                        rc, libcfs_nid2str(peer->ibp_nid));
 913
 914         kiblnd_close_conn(conn, rc);
 915
 916         if (done)
 917                 kiblnd_tx_done(peer->ibp_ni, tx);
 918
 919         spin_lock(&conn->ibc_lock);
 920
 921         return -EIO;
 922 }
 923
 924 void
 925 kiblnd_check_sends (kib_conn_t *conn)
 926 {
 927         int     ver = conn->ibc_version;
 928         lnet_ni_t *ni = conn->ibc_peer->ibp_ni;
 929         kib_tx_t  *tx;
 930
 931         /* Don't send anything until after the connection is established */
 932         if (conn->ibc_state < IBLND_CONN_ESTABLISHED) {
 933                 CDEBUG(D_NET, "%s too soon\n",
 934                        libcfs_nid2str(conn->ibc_peer->ibp_nid));
 935                 return;
 936         }
 937
 938         spin_lock(&conn->ibc_lock);
 939
 940         LASSERT (conn->ibc_nsends_posted <= IBLND_CONCURRENT_SENDS(ver));
 941         LASSERT (!IBLND_OOB_CAPABLE(ver) ||
 942                  conn->ibc_noops_posted <= IBLND_OOB_MSGS(ver));
 943         LASSERT (conn->ibc_reserved_credits >= 0);
 944
 945         while (conn->ibc_reserved_credits > 0 &&
 946                !list_empty(&conn->ibc_tx_queue_rsrvd)) {
 947                 tx = list_entry(conn->ibc_tx_queue_rsrvd.next,
 948                                     kib_tx_t, tx_list);
 949                 list_del(&tx->tx_list);
 950                 list_add_tail(&tx->tx_list, &conn->ibc_tx_queue);
 951                 conn->ibc_reserved_credits--;
 952         }
 953
 954         if (kiblnd_need_noop(conn)) {
 955                 spin_unlock(&conn->ibc_lock);
 956
 957                 tx = kiblnd_get_idle_tx(ni, conn->ibc_peer->ibp_nid);
 958                 if (tx != NULL)
 959                         kiblnd_init_tx_msg(ni, tx, IBLND_MSG_NOOP, 0);
 960
 961                 spin_lock(&conn->ibc_lock);
 962                 if (tx != NULL)
 963                         kiblnd_queue_tx_locked(tx, conn);
 964         }
 965
 966         kiblnd_conn_addref(conn); /* 1 ref for me.... (see b21911) */
 967
 968         for (;;) {
 969                 int credit;
 970
 971                 if (!list_empty(&conn->ibc_tx_queue_nocred)) {
 972                         credit = 0;
 973                         tx = list_entry(conn->ibc_tx_queue_nocred.next,
 974                                             kib_tx_t, tx_list);
 975                 } else if (!list_empty(&conn->ibc_tx_noops)) {
 976                         LASSERT (!IBLND_OOB_CAPABLE(ver));
 977                         credit = 1;
 978                         tx = list_entry(conn->ibc_tx_noops.next,
 979                                         kib_tx_t, tx_list);
 980                 } else if (!list_empty(&conn->ibc_tx_queue)) {
 981                         credit = 1;
 982                         tx = list_entry(conn->ibc_tx_queue.next,
 983                                             kib_tx_t, tx_list);
 984                 } else
 985                         break;
 986
 987                 if (kiblnd_post_tx_locked(conn, tx, credit) != 0)
 988                         break;
 989         }
 990
 991         spin_unlock(&conn->ibc_lock);
 992
 993         kiblnd_conn_decref(conn); /* ...until here */
 994 }
 995
 996 void
 997 kiblnd_tx_complete (kib_tx_t *tx, int status)
 998 {
 999         int        failed = (status != IB_WC_SUCCESS);
1000         kib_conn_t   *conn = tx->tx_conn;
1001         int        idle;
1002
1003         LASSERT (tx->tx_sending > 0);
1004
1005         if (failed) {
1006                 if (conn->ibc_state == IBLND_CONN_ESTABLISHED)
1007                         CNETERR("Tx -> %s cookie "LPX64
1008                                 " sending %d waiting %d: failed %d\n",
1009                                 libcfs_nid2str(conn->ibc_peer->ibp_nid),
1010                                 tx->tx_cookie, tx->tx_sending, tx->tx_waiting,
1011                                 status);
1012
1013                 kiblnd_close_conn(conn, -EIO);
1014         } else {
1015                 kiblnd_peer_alive(conn->ibc_peer);
1016         }
1017
1018         spin_lock(&conn->ibc_lock);
1019
1020         /* I could be racing with rdma completion.  Whoever makes 'tx' idle
1021          * gets to free it, which also drops its ref on 'conn'. */
1022
1023         tx->tx_sending--;
1024         conn->ibc_nsends_posted--;
1025         if (tx->tx_msg->ibm_type == IBLND_MSG_NOOP)
1026                 conn->ibc_noops_posted--;
1027
1028         if (failed) {
1029                 tx->tx_waiting = 0;          /* don't wait for peer */
1030                 tx->tx_status = -EIO;
1031         }
1032
1033         idle = (tx->tx_sending == 0) &&  /* This is the final callback */
1034                !tx->tx_waiting &&              /* Not waiting for peer */
1035                !tx->tx_queued;            /* Not re-queued (PUT_DONE) */
1036         if (idle)
1037                 list_del(&tx->tx_list);
1038
1039         kiblnd_conn_addref(conn);              /* 1 ref for me.... */
1040
1041         spin_unlock(&conn->ibc_lock);
1042
1043         if (idle)
1044                 kiblnd_tx_done(conn->ibc_peer->ibp_ni, tx);
1045
1046         kiblnd_check_sends(conn);
1047
1048         kiblnd_conn_decref(conn);              /* ...until here */
1049 }
1050
1051 void
1052 kiblnd_init_tx_msg (lnet_ni_t *ni, kib_tx_t *tx, int type, int body_nob)
1053 {
1054         kib_hca_dev_t     *hdev = tx->tx_pool->tpo_hdev;
1055         struct ib_sge     *sge = &tx->tx_sge[tx->tx_nwrq];
1056         struct ib_send_wr *wrq = &tx->tx_wrq[tx->tx_nwrq];
1057         int             nob = offsetof (kib_msg_t, ibm_u) + body_nob;
1058         struct ib_mr      *mr;
1059
1060         LASSERT (tx->tx_nwrq >= 0);
1061         LASSERT (tx->tx_nwrq < IBLND_MAX_RDMA_FRAGS + 1);
1062         LASSERT (nob <= IBLND_MSG_SIZE);
1063
1064         kiblnd_init_msg(tx->tx_msg, type, body_nob);
1065
1066         mr = kiblnd_find_dma_mr(hdev, tx->tx_msgaddr, nob);
1067         LASSERT (mr != NULL);
1068
1069         sge->lkey   = mr->lkey;
1070         sge->addr   = tx->tx_msgaddr;
1071         sge->length = nob;
1072
1073         memset(wrq, 0, sizeof(*wrq));
1074
1075         wrq->next       = NULL;
1076         wrq->wr_id      = kiblnd_ptr2wreqid(tx, IBLND_WID_TX);
1077         wrq->sg_list    = sge;
1078         wrq->num_sge    = 1;
1079         wrq->opcode     = IB_WR_SEND;
1080         wrq->send_flags = IB_SEND_SIGNALED;
1081
1082         tx->tx_nwrq++;
1083 }
1084
1085 int
1086 kiblnd_init_rdma (kib_conn_t *conn, kib_tx_t *tx, int type,
1087                   int resid, kib_rdma_desc_t *dstrd, __u64 dstcookie)
1088 {
1089         kib_msg_t        *ibmsg = tx->tx_msg;
1090         kib_rdma_desc_t   *srcrd = tx->tx_rd;
1091         struct ib_sge     *sge = &tx->tx_sge[0];
1092         struct ib_send_wr *wrq = &tx->tx_wrq[0];
1093         int             rc  = resid;
1094         int             srcidx;
1095         int             dstidx;
1096         int             wrknob;
1097
1098         LASSERT (!in_interrupt());
1099         LASSERT (tx->tx_nwrq == 0);
1100         LASSERT (type == IBLND_MSG_GET_DONE ||
1101                  type == IBLND_MSG_PUT_DONE);
1102
1103         srcidx = dstidx = 0;
1104
1105         while (resid > 0) {
1106                 if (srcidx >= srcrd->rd_nfrags) {
1107                         CERROR("Src buffer exhausted: %d frags\n", srcidx);
1108                         rc = -EPROTO;
1109                         break;
1110                 }
1111
1112                 if (dstidx == dstrd->rd_nfrags) {
1113                         CERROR("Dst buffer exhausted: %d frags\n", dstidx);
1114                         rc = -EPROTO;
1115                         break;
1116                 }
1117
1118                 if (tx->tx_nwrq == IBLND_RDMA_FRAGS(conn->ibc_version)) {
1119                         CERROR("RDMA too fragmented for %s (%d): "
1120                                "%d/%d src %d/%d dst frags\n",
1121                                libcfs_nid2str(conn->ibc_peer->ibp_nid),
1122                                IBLND_RDMA_FRAGS(conn->ibc_version),
1123                                srcidx, srcrd->rd_nfrags,
1124                                dstidx, dstrd->rd_nfrags);
1125                         rc = -EMSGSIZE;
1126                         break;
1127                 }
1128
1129                 wrknob = MIN(MIN(kiblnd_rd_frag_size(srcrd, srcidx),
1130                                  kiblnd_rd_frag_size(dstrd, dstidx)), resid);
1131
1132                 sge = &tx->tx_sge[tx->tx_nwrq];
1133                 sge->addr   = kiblnd_rd_frag_addr(srcrd, srcidx);
1134                 sge->lkey   = kiblnd_rd_frag_key(srcrd, srcidx);
1135                 sge->length = wrknob;
1136
1137                 wrq = &tx->tx_wrq[tx->tx_nwrq];
1138
1139                 wrq->next       = wrq + 1;
1140                 wrq->wr_id      = kiblnd_ptr2wreqid(tx, IBLND_WID_RDMA);
1141                 wrq->sg_list    = sge;
1142                 wrq->num_sge    = 1;
1143                 wrq->opcode     = IB_WR_RDMA_WRITE;
1144                 wrq->send_flags = 0;
1145
1146                 wrq->wr.rdma.remote_addr = kiblnd_rd_frag_addr(dstrd, dstidx);
1147                 wrq->wr.rdma.rkey       = kiblnd_rd_frag_key(dstrd, dstidx);
1148
1149                 srcidx = kiblnd_rd_consume_frag(srcrd, srcidx, wrknob);
1150                 dstidx = kiblnd_rd_consume_frag(dstrd, dstidx, wrknob);
1151
1152                 resid -= wrknob;
1153
1154                 tx->tx_nwrq++;
1155                 wrq++;
1156                 sge++;
1157         }
1158
1159         if (rc < 0)                          /* no RDMA if completing with failure */
1160                 tx->tx_nwrq = 0;
1161
1162         ibmsg->ibm_u.completion.ibcm_status = rc;
1163         ibmsg->ibm_u.completion.ibcm_cookie = dstcookie;
1164         kiblnd_init_tx_msg(conn->ibc_peer->ibp_ni, tx,
1165                            type, sizeof (kib_completion_msg_t));
1166
1167         return rc;
1168 }
1169
1170 void
1171 kiblnd_queue_tx_locked (kib_tx_t *tx, kib_conn_t *conn)
1172 {
1173         struct list_head   *q;
1174
1175         LASSERT (tx->tx_nwrq > 0);            /* work items set up */
1176         LASSERT (!tx->tx_queued);              /* not queued for sending already */
1177         LASSERT (conn->ibc_state >= IBLND_CONN_ESTABLISHED);
1178
1179         tx->tx_queued = 1;
1180         tx->tx_deadline = jiffies + (*kiblnd_tunables.kib_timeout * HZ);
1181
1182         if (tx->tx_conn == NULL) {
1183                 kiblnd_conn_addref(conn);
1184                 tx->tx_conn = conn;
1185                 LASSERT (tx->tx_msg->ibm_type != IBLND_MSG_PUT_DONE);
1186         } else {
1187                 /* PUT_DONE first attached to conn as a PUT_REQ */
1188                 LASSERT (tx->tx_conn == conn);
1189                 LASSERT (tx->tx_msg->ibm_type == IBLND_MSG_PUT_DONE);
1190         }
1191
1192         switch (tx->tx_msg->ibm_type) {
1193         default:
1194                 LBUG();
1195
1196         case IBLND_MSG_PUT_REQ:
1197         case IBLND_MSG_GET_REQ:
1198                 q = &conn->ibc_tx_queue_rsrvd;
1199                 break;
1200
1201         case IBLND_MSG_PUT_NAK:
1202         case IBLND_MSG_PUT_ACK:
1203         case IBLND_MSG_PUT_DONE:
1204         case IBLND_MSG_GET_DONE:
1205                 q = &conn->ibc_tx_queue_nocred;
1206                 break;
1207
1208         case IBLND_MSG_NOOP:
1209                 if (IBLND_OOB_CAPABLE(conn->ibc_version))
1210                         q = &conn->ibc_tx_queue_nocred;
1211                 else
1212                         q = &conn->ibc_tx_noops;
1213                 break;
1214
1215         case IBLND_MSG_IMMEDIATE:
1216                 q = &conn->ibc_tx_queue;
1217                 break;
1218         }
1219
1220         list_add_tail(&tx->tx_list, q);
1221 }
1222
1223 void
1224 kiblnd_queue_tx (kib_tx_t *tx, kib_conn_t *conn)
1225 {
1226         spin_lock(&conn->ibc_lock);
1227         kiblnd_queue_tx_locked(tx, conn);
1228         spin_unlock(&conn->ibc_lock);
1229
1230         kiblnd_check_sends(conn);
1231 }
1232
1233 static int kiblnd_resolve_addr(struct rdma_cm_id *cmid,
1234                                struct sockaddr_in *srcaddr,
1235                                struct sockaddr_in *dstaddr,
1236                                int timeout_ms)
1237 {
1238         unsigned short port;
1239         int rc;
1240
1241         /* allow the port to be reused */
1242         rc = rdma_set_reuseaddr(cmid, 1);
1243         if (rc != 0) {
1244                 CERROR("Unable to set reuse on cmid: %d\n", rc);
1245                 return rc;
1246         }
1247
1248         /* look for a free privileged port */
1249         for (port = PROT_SOCK-1; port > 0; port--) {
1250                 srcaddr->sin_port = htons(port);
1251                 rc = rdma_resolve_addr(cmid,
1252                                        (struct sockaddr *)srcaddr,
1253                                        (struct sockaddr *)dstaddr,
1254                                        timeout_ms);
1255                 if (rc == 0) {
1256                         CDEBUG(D_NET, "bound to port %hu\n", port);
1257                         return 0;
1258                 } else if (rc == -EADDRINUSE || rc == -EADDRNOTAVAIL) {
1259                         CDEBUG(D_NET, "bind to port %hu failed: %d\n",
1260                                port, rc);
1261                 } else {
1262                         return rc;
1263                 }
1264         }
1265
1266         CERROR("Failed to bind to a free privileged port\n");
1267         return rc;
1268 }
1269
1270 void
1271 kiblnd_connect_peer (kib_peer_t *peer)
1272 {
1273         struct rdma_cm_id *cmid;
1274         kib_dev_t        *dev;
1275         kib_net_t        *net = peer->ibp_ni->ni_data;
1276         struct sockaddr_in srcaddr;
1277         struct sockaddr_in dstaddr;
1278         int             rc;
1279
1280         LASSERT (net != NULL);
1281         LASSERT (peer->ibp_connecting > 0);
1282
1283         cmid = kiblnd_rdma_create_id(kiblnd_cm_callback, peer, RDMA_PS_TCP,
1284                                      IB_QPT_RC);
1285
1286         if (IS_ERR(cmid)) {
1287                 CERROR("Can't create CMID for %s: %ld\n",
1288                        libcfs_nid2str(peer->ibp_nid), PTR_ERR(cmid));
1289                 rc = PTR_ERR(cmid);
1290                 goto failed;
1291         }
1292
1293         dev = net->ibn_dev;
1294         memset(&srcaddr, 0, sizeof(srcaddr));
1295         srcaddr.sin_family = AF_INET;
1296         srcaddr.sin_addr.s_addr = htonl(dev->ibd_ifip);
1297
1298         memset(&dstaddr, 0, sizeof(dstaddr));
1299         dstaddr.sin_family = AF_INET;
1300         dstaddr.sin_port = htons(*kiblnd_tunables.kib_service);
1301         dstaddr.sin_addr.s_addr = htonl(LNET_NIDADDR(peer->ibp_nid));
1302
1303         kiblnd_peer_addref(peer);              /* cmid's ref */
1304
1305         if (*kiblnd_tunables.kib_use_priv_port) {
1306                 rc = kiblnd_resolve_addr(cmid, &srcaddr, &dstaddr,
1307                                          *kiblnd_tunables.kib_timeout * 1000);
1308         } else {
1309                 rc = rdma_resolve_addr(cmid,
1310                                        (struct sockaddr *)&srcaddr,
1311                                        (struct sockaddr *)&dstaddr,
1312                                        *kiblnd_tunables.kib_timeout * 1000);
1313         }
1314         if (rc != 0) {
1315                 /* Can't initiate address resolution:  */
1316                 CERROR("Can't resolve addr for %s: %d\n",
1317                        libcfs_nid2str(peer->ibp_nid), rc);
1318                 goto failed2;
1319         }
1320
1321         LASSERT (cmid->device != NULL);
1322         CDEBUG(D_NET, "%s: connection bound to %s:%u.%u.%u.%u:%s\n",
1323                libcfs_nid2str(peer->ibp_nid), dev->ibd_ifname,
1324                HIPQUAD(dev->ibd_ifip), cmid->device->name);
1325
1326         return;
1327
1328  failed2:
1329         kiblnd_peer_decref(peer);              /* cmid's ref */
1330         rdma_destroy_id(cmid);
1331  failed:
1332         kiblnd_peer_connect_failed(peer, 1, rc);
1333 }
1334
1335 void
1336 kiblnd_launch_tx (lnet_ni_t *ni, kib_tx_t *tx, lnet_nid_t nid)
1337 {
1338         kib_peer_t      *peer;
1339         kib_peer_t      *peer2;
1340         kib_conn_t      *conn;
1341         rwlock_t        *g_lock = &kiblnd_data.kib_global_lock;
1342         unsigned long      flags;
1343         int             rc;
1344
1345         /* If I get here, I've committed to send, so I complete the tx with
1346          * failure on any problems */
1347
1348         LASSERT (tx == NULL || tx->tx_conn == NULL); /* only set when assigned a conn */
1349         LASSERT (tx == NULL || tx->tx_nwrq > 0);     /* work items have been set up */
1350
1351         /* First time, just use a read lock since I expect to find my peer
1352          * connected */
1353         read_lock_irqsave(g_lock, flags);
1354
1355         peer = kiblnd_find_peer_locked(nid);
1356         if (peer != NULL && !list_empty(&peer->ibp_conns)) {
1357                 /* Found a peer with an established connection */
1358                 conn = kiblnd_get_conn_locked(peer);
1359                 kiblnd_conn_addref(conn); /* 1 ref for me... */
1360
1361                 read_unlock_irqrestore(g_lock, flags);
1362
1363                 if (tx != NULL)
1364                         kiblnd_queue_tx(tx, conn);
1365                 kiblnd_conn_decref(conn); /* ...to here */
1366                 return;
1367         }
1368
1369         read_unlock(g_lock);
1370         /* Re-try with a write lock */
1371         write_lock(g_lock);
1372
1373         peer = kiblnd_find_peer_locked(nid);
1374         if (peer != NULL) {
1375                 if (list_empty(&peer->ibp_conns)) {
1376                         /* found a peer, but it's still connecting... */
1377                         LASSERT (peer->ibp_connecting != 0 ||
1378                                  peer->ibp_accepting != 0);
1379                         if (tx != NULL)
1380                                 list_add_tail(&tx->tx_list,
1381                                                   &peer->ibp_tx_queue);
1382                         write_unlock_irqrestore(g_lock, flags);
1383                 } else {
1384                         conn = kiblnd_get_conn_locked(peer);
1385                         kiblnd_conn_addref(conn); /* 1 ref for me... */
1386
1387                         write_unlock_irqrestore(g_lock, flags);
1388
1389                         if (tx != NULL)
1390                                 kiblnd_queue_tx(tx, conn);
1391                         kiblnd_conn_decref(conn); /* ...to here */
1392                 }
1393                 return;
1394         }
1395
1396         write_unlock_irqrestore(g_lock, flags);
1397
1398         /* Allocate a peer ready to add to the peer table and retry */
1399         rc = kiblnd_create_peer(ni, &peer, nid);
1400         if (rc != 0) {
1401                 CERROR("Can't create peer %s\n", libcfs_nid2str(nid));
1402                 if (tx != NULL) {
1403                         tx->tx_status = -EHOSTUNREACH;
1404                         tx->tx_waiting = 0;
1405                         kiblnd_tx_done(ni, tx);
1406                 }
1407                 return;
1408         }
1409
1410         write_lock_irqsave(g_lock, flags);
1411
1412         peer2 = kiblnd_find_peer_locked(nid);
1413         if (peer2 != NULL) {
1414                 if (list_empty(&peer2->ibp_conns)) {
1415                         /* found a peer, but it's still connecting... */
1416                         LASSERT (peer2->ibp_connecting != 0 ||
1417                                  peer2->ibp_accepting != 0);
1418                         if (tx != NULL)
1419                                 list_add_tail(&tx->tx_list,
1420                                                   &peer2->ibp_tx_queue);
1421                         write_unlock_irqrestore(g_lock, flags);
1422                 } else {
1423                         conn = kiblnd_get_conn_locked(peer2);
1424                         kiblnd_conn_addref(conn); /* 1 ref for me... */
1425
1426                         write_unlock_irqrestore(g_lock, flags);
1427
1428                         if (tx != NULL)
1429                                 kiblnd_queue_tx(tx, conn);
1430                         kiblnd_conn_decref(conn); /* ...to here */
1431                 }
1432
1433                 kiblnd_peer_decref(peer);
1434                 return;
1435         }
1436
1437         /* Brand new peer */
1438         LASSERT (peer->ibp_connecting == 0);
1439         peer->ibp_connecting = 1;
1440
1441         /* always called with a ref on ni, which prevents ni being shutdown */
1442         LASSERT (((kib_net_t *)ni->ni_data)->ibn_shutdown == 0);
1443
1444         if (tx != NULL)
1445                 list_add_tail(&tx->tx_list, &peer->ibp_tx_queue);
1446
1447         kiblnd_peer_addref(peer);
1448         list_add_tail(&peer->ibp_list, kiblnd_nid2peerlist(nid));
1449
1450         write_unlock_irqrestore(g_lock, flags);
1451
1452         kiblnd_connect_peer(peer);
1453         kiblnd_peer_decref(peer);
1454 }
1455
1456 int
1457 kiblnd_send (lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg)
1458 {
1459         lnet_hdr_t       *hdr = &lntmsg->msg_hdr;
1460         int            type = lntmsg->msg_type;
1461         lnet_process_id_t target = lntmsg->msg_target;
1462         int            target_is_router = lntmsg->msg_target_is_router;
1463         int            routing = lntmsg->msg_routing;
1464         unsigned int      payload_niov = lntmsg->msg_niov;
1465         struct iovec     *payload_iov = lntmsg->msg_iov;
1466         lnet_kiov_t      *payload_kiov = lntmsg->msg_kiov;
1467         unsigned int      payload_offset = lntmsg->msg_offset;
1468         unsigned int      payload_nob = lntmsg->msg_len;
1469         kib_msg_t       *ibmsg;
1470         kib_tx_t         *tx;
1471         int            nob;
1472         int            rc;
1473
1474         /* NB 'private' is different depending on what we're sending.... */
1475
1476         CDEBUG(D_NET, "sending %d bytes in %d frags to %s\n",
1477                payload_nob, payload_niov, libcfs_id2str(target));
1478
1479         LASSERT (payload_nob == 0 || payload_niov > 0);
1480         LASSERT (payload_niov <= LNET_MAX_IOV);
1481
1482         /* Thread context */
1483         LASSERT (!in_interrupt());
1484         /* payload is either all vaddrs or all pages */
1485         LASSERT (!(payload_kiov != NULL && payload_iov != NULL));
1486
1487         switch (type) {
1488         default:
1489                 LBUG();
1490                 return (-EIO);
1491
1492         case LNET_MSG_ACK:
1493                 LASSERT (payload_nob == 0);
1494                 break;
1495
1496         case LNET_MSG_GET:
1497                 if (routing || target_is_router)
1498                         break;            /* send IMMEDIATE */
1499
1500                 /* is the REPLY message too small for RDMA? */
1501                 nob = offsetof(kib_msg_t, ibm_u.immediate.ibim_payload[lntmsg->msg_md->md_length]);
1502                 if (nob <= IBLND_MSG_SIZE)
1503                         break;            /* send IMMEDIATE */
1504
1505                 tx = kiblnd_get_idle_tx(ni, target.nid);
1506                 if (tx == NULL) {
1507                         CERROR("Can't allocate txd for GET to %s\n",
1508                                libcfs_nid2str(target.nid));
1509                         return -ENOMEM;
1510                 }
1511
1512                 ibmsg = tx->tx_msg;
1513
1514                 if ((lntmsg->msg_md->md_options & LNET_MD_KIOV) == 0)
1515                         rc = kiblnd_setup_rd_iov(ni, tx,
1516                                                  &ibmsg->ibm_u.get.ibgm_rd,
1517                                                  lntmsg->msg_md->md_niov,
1518                                                  lntmsg->msg_md->md_iov.iov,
1519                                                  0, lntmsg->msg_md->md_length);
1520                 else
1521                         rc = kiblnd_setup_rd_kiov(ni, tx,
1522                                                   &ibmsg->ibm_u.get.ibgm_rd,
1523                                                   lntmsg->msg_md->md_niov,
1524                                                   lntmsg->msg_md->md_iov.kiov,
1525                                                   0, lntmsg->msg_md->md_length);
1526                 if (rc != 0) {
1527                         CERROR("Can't setup GET sink for %s: %d\n",
1528                                libcfs_nid2str(target.nid), rc);
1529                         kiblnd_tx_done(ni, tx);
1530                         return -EIO;
1531                 }
1532
1533                 nob = offsetof(kib_get_msg_t, ibgm_rd.rd_frags[tx->tx_nfrags]);
1534                 ibmsg->ibm_u.get.ibgm_cookie = tx->tx_cookie;
1535                 ibmsg->ibm_u.get.ibgm_hdr = *hdr;
1536
1537                 kiblnd_init_tx_msg(ni, tx, IBLND_MSG_GET_REQ, nob);
1538
1539                 tx->tx_lntmsg[1] = lnet_create_reply_msg(ni, lntmsg);
1540                 if (tx->tx_lntmsg[1] == NULL) {
1541                         CERROR("Can't create reply for GET -> %s\n",
1542                                libcfs_nid2str(target.nid));
1543                         kiblnd_tx_done(ni, tx);
1544                         return -EIO;
1545                 }
1546
1547                 tx->tx_lntmsg[0] = lntmsg;      /* finalise lntmsg[0,1] on completion */
1548                 tx->tx_waiting = 1;          /* waiting for GET_DONE */
1549                 kiblnd_launch_tx(ni, tx, target.nid);
1550                 return 0;
1551
1552         case LNET_MSG_REPLY:
1553         case LNET_MSG_PUT:
1554                 /* Is the payload small enough not to need RDMA? */
1555                 nob = offsetof(kib_msg_t, ibm_u.immediate.ibim_payload[payload_nob]);
1556                 if (nob <= IBLND_MSG_SIZE)
1557                         break;            /* send IMMEDIATE */
1558
1559                 tx = kiblnd_get_idle_tx(ni, target.nid);
1560                 if (tx == NULL) {
1561                         CERROR("Can't allocate %s txd for %s\n",
1562                                type == LNET_MSG_PUT ? "PUT" : "REPLY",
1563                                libcfs_nid2str(target.nid));
1564                         return -ENOMEM;
1565                 }
1566
1567                 if (payload_kiov == NULL)
1568                         rc = kiblnd_setup_rd_iov(ni, tx, tx->tx_rd,
1569                                                  payload_niov, payload_iov,
1570                                                  payload_offset, payload_nob);
1571                 else
1572                         rc = kiblnd_setup_rd_kiov(ni, tx, tx->tx_rd,
1573                                                   payload_niov, payload_kiov,
1574                                                   payload_offset, payload_nob);
1575                 if (rc != 0) {
1576                         CERROR("Can't setup PUT src for %s: %d\n",
1577                                libcfs_nid2str(target.nid), rc);
1578                         kiblnd_tx_done(ni, tx);
1579                         return -EIO;
1580                 }
1581
1582                 ibmsg = tx->tx_msg;
1583                 ibmsg->ibm_u.putreq.ibprm_hdr = *hdr;
1584                 ibmsg->ibm_u.putreq.ibprm_cookie = tx->tx_cookie;
1585                 kiblnd_init_tx_msg(ni, tx, IBLND_MSG_PUT_REQ, sizeof(kib_putreq_msg_t));
1586
1587                 tx->tx_lntmsg[0] = lntmsg;      /* finalise lntmsg on completion */
1588                 tx->tx_waiting = 1;          /* waiting for PUT_{ACK,NAK} */
1589                 kiblnd_launch_tx(ni, tx, target.nid);
1590                 return 0;
1591         }
1592
1593         /* send IMMEDIATE */
1594
1595         LASSERT (offsetof(kib_msg_t, ibm_u.immediate.ibim_payload[payload_nob])
1596                  <= IBLND_MSG_SIZE);
1597
1598         tx = kiblnd_get_idle_tx(ni, target.nid);
1599         if (tx == NULL) {
1600                 CERROR ("Can't send %d to %s: tx descs exhausted\n",
1601                         type, libcfs_nid2str(target.nid));
1602                 return -ENOMEM;
1603         }
1604
1605         ibmsg = tx->tx_msg;
1606         ibmsg->ibm_u.immediate.ibim_hdr = *hdr;
1607
1608         if (payload_kiov != NULL)
1609                 lnet_copy_kiov2flat(IBLND_MSG_SIZE, ibmsg,
1610                                     offsetof(kib_msg_t, ibm_u.immediate.ibim_payload),
1611                                     payload_niov, payload_kiov,
1612                                     payload_offset, payload_nob);
1613         else
1614                 lnet_copy_iov2flat(IBLND_MSG_SIZE, ibmsg,
1615                                    offsetof(kib_msg_t, ibm_u.immediate.ibim_payload),
1616                                    payload_niov, payload_iov,
1617                                    payload_offset, payload_nob);
1618
1619         nob = offsetof(kib_immediate_msg_t, ibim_payload[payload_nob]);
1620         kiblnd_init_tx_msg(ni, tx, IBLND_MSG_IMMEDIATE, nob);
1621
1622         tx->tx_lntmsg[0] = lntmsg;            /* finalise lntmsg on completion */
1623         kiblnd_launch_tx(ni, tx, target.nid);
1624         return 0;
1625 }
1626
1627 void
1628 kiblnd_reply (lnet_ni_t *ni, kib_rx_t *rx, lnet_msg_t *lntmsg)
1629 {
1630         lnet_process_id_t target = lntmsg->msg_target;
1631         unsigned int      niov = lntmsg->msg_niov;
1632         struct iovec     *iov = lntmsg->msg_iov;
1633         lnet_kiov_t      *kiov = lntmsg->msg_kiov;
1634         unsigned int      offset = lntmsg->msg_offset;
1635         unsigned int      nob = lntmsg->msg_len;
1636         kib_tx_t         *tx;
1637         int            rc;
1638
1639         tx = kiblnd_get_idle_tx(ni, rx->rx_conn->ibc_peer->ibp_nid);
1640         if (tx == NULL) {
1641                 CERROR("Can't get tx for REPLY to %s\n",
1642                        libcfs_nid2str(target.nid));
1643                 goto failed_0;
1644         }
1645
1646         if (nob == 0)
1647                 rc = 0;
1648         else if (kiov == NULL)
1649                 rc = kiblnd_setup_rd_iov(ni, tx, tx->tx_rd,
1650                                          niov, iov, offset, nob);
1651         else
1652                 rc = kiblnd_setup_rd_kiov(ni, tx, tx->tx_rd,
1653                                           niov, kiov, offset, nob);
1654
1655         if (rc != 0) {
1656                 CERROR("Can't setup GET src for %s: %d\n",
1657                        libcfs_nid2str(target.nid), rc);
1658                 goto failed_1;
1659         }
1660
1661         rc = kiblnd_init_rdma(rx->rx_conn, tx,
1662                               IBLND_MSG_GET_DONE, nob,
1663                               &rx->rx_msg->ibm_u.get.ibgm_rd,
1664                               rx->rx_msg->ibm_u.get.ibgm_cookie);
1665         if (rc < 0) {
1666                 CERROR("Can't setup rdma for GET from %s: %d\n",
1667                        libcfs_nid2str(target.nid), rc);
1668                 goto failed_1;
1669         }
1670
1671         if (nob == 0) {
1672                 /* No RDMA: local completion may happen now! */
1673                 lnet_finalize(ni, lntmsg, 0);
1674         } else {
1675                 /* RDMA: lnet_finalize(lntmsg) when it
1676                  * completes */
1677                 tx->tx_lntmsg[0] = lntmsg;
1678         }
1679
1680         kiblnd_queue_tx(tx, rx->rx_conn);
1681         return;
1682
1683  failed_1:
1684         kiblnd_tx_done(ni, tx);
1685  failed_0:
1686         lnet_finalize(ni, lntmsg, -EIO);
1687 }
1688
1689 int
1690 kiblnd_recv (lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg, int delayed,
1691              unsigned int niov, struct iovec *iov, lnet_kiov_t *kiov,
1692              unsigned int offset, unsigned int mlen, unsigned int rlen)
1693 {
1694         kib_rx_t    *rx = private;
1695         kib_msg_t   *rxmsg = rx->rx_msg;
1696         kib_conn_t  *conn = rx->rx_conn;
1697         kib_tx_t    *tx;
1698         kib_msg_t   *txmsg;
1699         int       nob;
1700         int       post_credit = IBLND_POSTRX_PEER_CREDIT;
1701         int       rc = 0;
1702
1703         LASSERT (mlen <= rlen);
1704         LASSERT (!in_interrupt());
1705         /* Either all pages or all vaddrs */
1706         LASSERT (!(kiov != NULL && iov != NULL));
1707
1708         switch (rxmsg->ibm_type) {
1709         default:
1710                 LBUG();
1711
1712         case IBLND_MSG_IMMEDIATE:
1713                 nob = offsetof(kib_msg_t, ibm_u.immediate.ibim_payload[rlen]);
1714                 if (nob > rx->rx_nob) {
1715                         CERROR ("Immediate message from %s too big: %d(%d)\n",
1716                                 libcfs_nid2str(rxmsg->ibm_u.immediate.ibim_hdr.src_nid),
1717                                 nob, rx->rx_nob);
1718                         rc = -EPROTO;
1719                         break;
1720                 }
1721
1722                 if (kiov != NULL)
1723                         lnet_copy_flat2kiov(niov, kiov, offset,
1724                                             IBLND_MSG_SIZE, rxmsg,
1725                                             offsetof(kib_msg_t, ibm_u.immediate.ibim_payload),
1726                                             mlen);
1727                 else
1728                         lnet_copy_flat2iov(niov, iov, offset,
1729                                            IBLND_MSG_SIZE, rxmsg,
1730                                            offsetof(kib_msg_t, ibm_u.immediate.ibim_payload),
1731                                            mlen);
1732                 lnet_finalize (ni, lntmsg, 0);
1733                 break;
1734
1735         case IBLND_MSG_PUT_REQ:
1736                 if (mlen == 0) {
1737                         lnet_finalize(ni, lntmsg, 0);
1738                         kiblnd_send_completion(rx->rx_conn, IBLND_MSG_PUT_NAK, 0,
1739                                                rxmsg->ibm_u.putreq.ibprm_cookie);
1740                         break;
1741                 }
1742
1743                 tx = kiblnd_get_idle_tx(ni, conn->ibc_peer->ibp_nid);
1744                 if (tx == NULL) {
1745                         CERROR("Can't allocate tx for %s\n",
1746                                libcfs_nid2str(conn->ibc_peer->ibp_nid));
1747                         /* Not replying will break the connection */
1748                         rc = -ENOMEM;
1749                         break;
1750                 }
1751
1752                 txmsg = tx->tx_msg;
1753                 if (kiov == NULL)
1754                         rc = kiblnd_setup_rd_iov(ni, tx,
1755                                                  &txmsg->ibm_u.putack.ibpam_rd,
1756                                                  niov, iov, offset, mlen);
1757                 else
1758                         rc = kiblnd_setup_rd_kiov(ni, tx,
1759                                                   &txmsg->ibm_u.putack.ibpam_rd,
1760                                                   niov, kiov, offset, mlen);
1761                 if (rc != 0) {
1762                         CERROR("Can't setup PUT sink for %s: %d\n",
1763                                libcfs_nid2str(conn->ibc_peer->ibp_nid), rc);
1764                         kiblnd_tx_done(ni, tx);
1765                         /* tell peer it's over */
1766                         kiblnd_send_completion(rx->rx_conn, IBLND_MSG_PUT_NAK, rc,
1767                                                rxmsg->ibm_u.putreq.ibprm_cookie);
1768                         break;
1769                 }
1770
1771                 nob = offsetof(kib_putack_msg_t, ibpam_rd.rd_frags[tx->tx_nfrags]);
1772                 txmsg->ibm_u.putack.ibpam_src_cookie = rxmsg->ibm_u.putreq.ibprm_cookie;
1773                 txmsg->ibm_u.putack.ibpam_dst_cookie = tx->tx_cookie;
1774
1775                 kiblnd_init_tx_msg(ni, tx, IBLND_MSG_PUT_ACK, nob);
1776
1777                 tx->tx_lntmsg[0] = lntmsg;      /* finalise lntmsg on completion */
1778                 tx->tx_waiting = 1;          /* waiting for PUT_DONE */
1779                 kiblnd_queue_tx(tx, conn);
1780
1781                 /* reposted buffer reserved for PUT_DONE */
1782                 post_credit = IBLND_POSTRX_NO_CREDIT;
1783                 break;
1784
1785         case IBLND_MSG_GET_REQ:
1786                 if (lntmsg != NULL) {
1787                         /* Optimized GET; RDMA lntmsg's payload */
1788                         kiblnd_reply(ni, rx, lntmsg);
1789                 } else {
1790                         /* GET didn't match anything */
1791                         kiblnd_send_completion(rx->rx_conn, IBLND_MSG_GET_DONE,
1792                                                -ENODATA,
1793                                                rxmsg->ibm_u.get.ibgm_cookie);
1794                 }
1795                 break;
1796         }
1797
1798         kiblnd_post_rx(rx, post_credit);
1799         return rc;
1800 }
1801
1802 int
1803 kiblnd_thread_start(int (*fn)(void *arg), void *arg, char *name)
1804 {
1805         task_t *task = kthread_run(fn, arg, name);
1806
1807         if (IS_ERR(task))
1808                 return PTR_ERR(task);
1809
1810         atomic_inc(&kiblnd_data.kib_nthreads);
1811         return 0;
1812 }
1813
1814 void
1815 kiblnd_thread_fini (void)
1816 {
1817         atomic_dec (&kiblnd_data.kib_nthreads);
1818 }
1819
1820 void
1821 kiblnd_peer_alive (kib_peer_t *peer)
1822 {
1823         /* This is racy, but everyone's only writing cfs_time_current() */
1824         peer->ibp_last_alive = cfs_time_current();
1825         mb();
1826 }
1827
1828 void
1829 kiblnd_peer_notify (kib_peer_t *peer)
1830 {
1831         int        error = 0;
1832         cfs_time_t    last_alive = 0;
1833         unsigned long flags;
1834
1835         read_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
1836
1837         if (list_empty(&peer->ibp_conns) &&
1838             peer->ibp_accepting == 0 &&
1839             peer->ibp_connecting == 0 &&
1840             peer->ibp_error != 0) {
1841                 error = peer->ibp_error;
1842                 peer->ibp_error = 0;
1843
1844                 last_alive = peer->ibp_last_alive;
1845         }
1846
1847         read_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
1848
1849         if (error != 0)
1850                 lnet_notify(peer->ibp_ni,
1851                             peer->ibp_nid, 0, last_alive);
1852 }
1853
1854 void
1855 kiblnd_close_conn_locked (kib_conn_t *conn, int error)
1856 {
1857         /* This just does the immediate housekeeping.  'error' is zero for a
1858          * normal shutdown which can happen only after the connection has been
1859          * established.  If the connection is established, schedule the
1860          * connection to be finished off by the connd.  Otherwise the connd is
1861          * already dealing with it (either to set it up or tear it down).
1862          * Caller holds kib_global_lock exclusively in irq context */
1863         kib_peer_t       *peer = conn->ibc_peer;
1864         kib_dev_t       *dev;
1865         unsigned long     flags;
1866
1867         LASSERT (error != 0 || conn->ibc_state >= IBLND_CONN_ESTABLISHED);
1868
1869         if (error != 0 && conn->ibc_comms_error == 0)
1870                 conn->ibc_comms_error = error;
1871
1872         if (conn->ibc_state != IBLND_CONN_ESTABLISHED)
1873                 return; /* already being handled  */
1874
1875         if (error == 0 &&
1876             list_empty(&conn->ibc_tx_noops) &&
1877             list_empty(&conn->ibc_tx_queue) &&
1878             list_empty(&conn->ibc_tx_queue_rsrvd) &&
1879             list_empty(&conn->ibc_tx_queue_nocred) &&
1880             list_empty(&conn->ibc_active_txs)) {
1881                 CDEBUG(D_NET, "closing conn to %s\n",
1882                        libcfs_nid2str(peer->ibp_nid));
1883         } else {
1884                 CNETERR("Closing conn to %s: error %d%s%s%s%s%s\n",
1885                        libcfs_nid2str(peer->ibp_nid), error,
1886                        list_empty(&conn->ibc_tx_queue) ? "" : "(sending)",
1887                        list_empty(&conn->ibc_tx_noops) ? "" : "(sending_noops)",
1888                        list_empty(&conn->ibc_tx_queue_rsrvd) ? "" : "(sending_rsrvd)",
1889                        list_empty(&conn->ibc_tx_queue_nocred) ? "" : "(sending_nocred)",
1890                        list_empty(&conn->ibc_active_txs) ? "" : "(waiting)");
1891         }
1892
1893         dev = ((kib_net_t *)peer->ibp_ni->ni_data)->ibn_dev;
1894         list_del(&conn->ibc_list);
1895         /* connd (see below) takes over ibc_list's ref */
1896
1897         if (list_empty (&peer->ibp_conns) &&    /* no more conns */
1898             kiblnd_peer_active(peer)) {  /* still in peer table */
1899                 kiblnd_unlink_peer_locked(peer);
1900
1901                 /* set/clear error on last conn */
1902                 peer->ibp_error = conn->ibc_comms_error;
1903         }
1904
1905         kiblnd_set_conn_state(conn, IBLND_CONN_CLOSING);
1906
1907         if (error != 0 &&
1908             kiblnd_dev_can_failover(dev)) {
1909                 list_add_tail(&dev->ibd_fail_list,
1910                               &kiblnd_data.kib_failed_devs);
1911                 wake_up(&kiblnd_data.kib_failover_waitq);
1912         }
1913
1914         spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags);
1915
1916         list_add_tail(&conn->ibc_list, &kiblnd_data.kib_connd_conns);
1917         wake_up(&kiblnd_data.kib_connd_waitq);
1918
1919         spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock, flags);
1920 }
1921
1922 void
1923 kiblnd_close_conn(kib_conn_t *conn, int error)
1924 {
1925         unsigned long flags;
1926
1927         write_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
1928
1929         kiblnd_close_conn_locked(conn, error);
1930
1931         write_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
1932 }
1933
1934 void
1935 kiblnd_handle_early_rxs(kib_conn_t *conn)
1936 {
1937         unsigned long    flags;
1938         kib_rx_t        *rx;
1939
1940         LASSERT(!in_interrupt());
1941         LASSERT(conn->ibc_state >= IBLND_CONN_ESTABLISHED);
1942
1943         write_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
1944         while (!list_empty(&conn->ibc_early_rxs)) {
1945                 rx = list_entry(conn->ibc_early_rxs.next,
1946                                     kib_rx_t, rx_list);
1947                 list_del(&rx->rx_list);
1948                 write_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
1949
1950                 kiblnd_handle_rx(rx);
1951
1952                 write_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
1953         }
1954         write_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
1955 }
1956
1957 void
1958 kiblnd_abort_txs(kib_conn_t *conn, struct list_head *txs)
1959 {
1960         LIST_HEAD       (zombies);
1961         struct list_head          *tmp;
1962         struct list_head          *nxt;
1963         kib_tx_t            *tx;
1964
1965         spin_lock(&conn->ibc_lock);
1966
1967         list_for_each_safe (tmp, nxt, txs) {
1968                 tx = list_entry (tmp, kib_tx_t, tx_list);
1969
1970                 if (txs == &conn->ibc_active_txs) {
1971                         LASSERT (!tx->tx_queued);
1972                         LASSERT (tx->tx_waiting ||
1973                                  tx->tx_sending != 0);
1974                 } else {
1975                         LASSERT (tx->tx_queued);
1976                 }
1977
1978                 tx->tx_status = -ECONNABORTED;
1979                 tx->tx_waiting = 0;
1980
1981                 if (tx->tx_sending == 0) {
1982                         tx->tx_queued = 0;
1983                         list_del (&tx->tx_list);
1984                         list_add (&tx->tx_list, &zombies);
1985                 }
1986         }
1987
1988         spin_unlock(&conn->ibc_lock);
1989
1990         kiblnd_txlist_done(conn->ibc_peer->ibp_ni, &zombies, -ECONNABORTED);
1991 }
1992
1993 void
1994 kiblnd_finalise_conn (kib_conn_t *conn)
1995 {
1996         LASSERT (!in_interrupt());
1997         LASSERT (conn->ibc_state > IBLND_CONN_INIT);
1998
1999         kiblnd_set_conn_state(conn, IBLND_CONN_DISCONNECTED);
2000
2001         /* abort_receives moves QP state to IB_QPS_ERR.  This is only required
2002          * for connections that didn't get as far as being connected, because
2003          * rdma_disconnect() does this for free. */
2004         kiblnd_abort_receives(conn);
2005
2006         /* Complete all tx descs not waiting for sends to complete.
2007          * NB we should be safe from RDMA now that the QP has changed state */
2008
2009         kiblnd_abort_txs(conn, &conn->ibc_tx_noops);
2010         kiblnd_abort_txs(conn, &conn->ibc_tx_queue);
2011         kiblnd_abort_txs(conn, &conn->ibc_tx_queue_rsrvd);
2012         kiblnd_abort_txs(conn, &conn->ibc_tx_queue_nocred);
2013         kiblnd_abort_txs(conn, &conn->ibc_active_txs);
2014
2015         kiblnd_handle_early_rxs(conn);
2016 }
2017
2018 void
2019 kiblnd_peer_connect_failed (kib_peer_t *peer, int active, int error)
2020 {
2021         LIST_HEAD    (zombies);
2022         unsigned long     flags;
2023
2024         LASSERT (error != 0);
2025         LASSERT (!in_interrupt());
2026
2027         write_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
2028
2029         if (active) {
2030                 LASSERT (peer->ibp_connecting > 0);
2031                 peer->ibp_connecting--;
2032         } else {
2033                 LASSERT (peer->ibp_accepting > 0);
2034                 peer->ibp_accepting--;
2035         }
2036
2037         if (peer->ibp_connecting != 0 ||
2038             peer->ibp_accepting != 0) {
2039                 /* another connection attempt under way... */
2040                 write_unlock_irqrestore(&kiblnd_data.kib_global_lock,
2041                                             flags);
2042                 return;
2043         }
2044
2045         if (list_empty(&peer->ibp_conns)) {
2046                 /* Take peer's blocked transmits to complete with error */
2047                 list_add(&zombies, &peer->ibp_tx_queue);
2048                 list_del_init(&peer->ibp_tx_queue);
2049
2050                 if (kiblnd_peer_active(peer))
2051                         kiblnd_unlink_peer_locked(peer);
2052
2053                 peer->ibp_error = error;
2054         } else {
2055                 /* Can't have blocked transmits if there are connections */
2056                 LASSERT (list_empty(&peer->ibp_tx_queue));
2057         }
2058
2059         write_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
2060
2061         kiblnd_peer_notify(peer);
2062
2063         if (list_empty (&zombies))
2064                 return;
2065
2066         CNETERR("Deleting messages for %s: connection failed\n",
2067                 libcfs_nid2str(peer->ibp_nid));
2068
2069         kiblnd_txlist_done(peer->ibp_ni, &zombies, -EHOSTUNREACH);
2070 }
2071
2072 void
2073 kiblnd_connreq_done(kib_conn_t *conn, int status)
2074 {
2075         kib_peer_t      *peer = conn->ibc_peer;
2076         kib_tx_t          *tx;
2077         struct list_head         txs;
2078         unsigned long      flags;
2079         int             active;
2080
2081         active = (conn->ibc_state == IBLND_CONN_ACTIVE_CONNECT);
2082
2083         CDEBUG(D_NET,"%s: active(%d), version(%x), status(%d)\n",
2084                libcfs_nid2str(peer->ibp_nid), active,
2085                conn->ibc_version, status);
2086
2087         LASSERT (!in_interrupt());
2088         LASSERT ((conn->ibc_state == IBLND_CONN_ACTIVE_CONNECT &&
2089                   peer->ibp_connecting > 0) ||
2090                  (conn->ibc_state == IBLND_CONN_PASSIVE_WAIT &&
2091                   peer->ibp_accepting > 0));
2092
2093         LIBCFS_FREE(conn->ibc_connvars, sizeof(*conn->ibc_connvars));
2094         conn->ibc_connvars = NULL;
2095
2096         if (status != 0) {
2097                 /* failed to establish connection */
2098                 kiblnd_peer_connect_failed(peer, active, status);
2099                 kiblnd_finalise_conn(conn);
2100                 return;
2101         }
2102
2103         /* connection established */
2104         write_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
2105
2106         conn->ibc_last_send = jiffies;
2107         kiblnd_set_conn_state(conn, IBLND_CONN_ESTABLISHED);
2108         kiblnd_peer_alive(peer);
2109
2110         /* Add conn to peer's list and nuke any dangling conns from a different
2111          * peer instance... */
2112         kiblnd_conn_addref(conn);              /* +1 ref for ibc_list */
2113         list_add(&conn->ibc_list, &peer->ibp_conns);
2114         if (active)
2115                 peer->ibp_connecting--;
2116         else
2117                 peer->ibp_accepting--;
2118
2119         if (peer->ibp_version == 0) {
2120                 peer->ibp_version     = conn->ibc_version;
2121                 peer->ibp_incarnation = conn->ibc_incarnation;
2122         }
2123
2124         if (peer->ibp_version     != conn->ibc_version ||
2125             peer->ibp_incarnation != conn->ibc_incarnation) {
2126                 kiblnd_close_stale_conns_locked(peer, conn->ibc_version,
2127                                                 conn->ibc_incarnation);
2128                 peer->ibp_version     = conn->ibc_version;
2129                 peer->ibp_incarnation = conn->ibc_incarnation;
2130         }
2131
2132         /* grab pending txs while I have the lock */
2133         list_add(&txs, &peer->ibp_tx_queue);
2134         list_del_init(&peer->ibp_tx_queue);
2135
2136         if (!kiblnd_peer_active(peer) ||        /* peer has been deleted */
2137             conn->ibc_comms_error != 0) {       /* error has happened already */
2138                 lnet_ni_t *ni = peer->ibp_ni;
2139
2140                 /* start to shut down connection */
2141                 kiblnd_close_conn_locked(conn, -ECONNABORTED);
2142                 write_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
2143
2144                 kiblnd_txlist_done(ni, &txs, -ECONNABORTED);
2145
2146                 return;
2147         }
2148
2149         write_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
2150
2151         /* Schedule blocked txs */
2152         spin_lock(&conn->ibc_lock);
2153         while (!list_empty(&txs)) {
2154                 tx = list_entry(txs.next, kib_tx_t, tx_list);
2155                 list_del(&tx->tx_list);
2156
2157                 kiblnd_queue_tx_locked(tx, conn);
2158         }
2159         spin_unlock(&conn->ibc_lock);
2160
2161         kiblnd_check_sends(conn);
2162
2163         /* schedule blocked rxs */
2164         kiblnd_handle_early_rxs(conn);
2165 }
2166
2167 void
2168 kiblnd_reject(struct rdma_cm_id *cmid, kib_rej_t *rej)
2169 {
2170         int       rc;
2171
2172         rc = rdma_reject(cmid, rej, sizeof(*rej));
2173
2174         if (rc != 0)
2175                 CWARN("Error %d sending reject\n", rc);
2176 }
2177
2178 int
2179 kiblnd_passive_connect (struct rdma_cm_id *cmid, void *priv, int priv_nob)
2180 {
2181         rwlock_t                *g_lock = &kiblnd_data.kib_global_lock;
2182         kib_msg_t            *reqmsg = priv;
2183         kib_msg_t            *ackmsg;
2184         kib_dev_t            *ibdev;
2185         kib_peer_t          *peer;
2186         kib_peer_t          *peer2;
2187         kib_conn_t          *conn;
2188         lnet_ni_t            *ni  = NULL;
2189         kib_net_t            *net = NULL;
2190         lnet_nid_t           nid;
2191         struct rdma_conn_param cp;
2192         kib_rej_t             rej;
2193         int                 version = IBLND_MSG_VERSION;
2194         unsigned long     flags;
2195         int                 rc;
2196         struct sockaddr_in    *peer_addr;
2197         LASSERT (!in_interrupt());
2198
2199         /* cmid inherits 'context' from the corresponding listener id */
2200         ibdev = (kib_dev_t *)cmid->context;
2201         LASSERT (ibdev != NULL);
2202
2203         memset(&rej, 0, sizeof(rej));
2204         rej.ibr_magic           = IBLND_MSG_MAGIC;
2205         rej.ibr_why               = IBLND_REJECT_FATAL;
2206         rej.ibr_cp.ibcp_max_msg_size = IBLND_MSG_SIZE;
2207
2208         peer_addr = (struct sockaddr_in *)&(cmid->route.addr.dst_addr);
2209         if (*kiblnd_tunables.kib_require_priv_port &&
2210             ntohs(peer_addr->sin_port) >= PROT_SOCK) {
2211                 __u32 ip = ntohl(peer_addr->sin_addr.s_addr);
2212                 CERROR("Peer's port (%u.%u.%u.%u:%hu) is not privileged\n",
2213                        HIPQUAD(ip), ntohs(peer_addr->sin_port));
2214                 goto failed;
2215         }
2216
2217         if (priv_nob < offsetof(kib_msg_t, ibm_type)) {
2218                 CERROR("Short connection request\n");
2219                 goto failed;
2220         }
2221
2222         /* Future protocol version compatibility support!  If the
2223          * o2iblnd-specific protocol changes, or when LNET unifies
2224          * protocols over all LNDs, the initial connection will
2225          * negotiate a protocol version.  I trap this here to avoid
2226          * console errors; the reject tells the peer which protocol I
2227          * speak. */
2228         if (reqmsg->ibm_magic == LNET_PROTO_MAGIC ||
2229             reqmsg->ibm_magic == __swab32(LNET_PROTO_MAGIC))
2230                 goto failed;
2231         if (reqmsg->ibm_magic == IBLND_MSG_MAGIC &&
2232             reqmsg->ibm_version != IBLND_MSG_VERSION &&
2233             reqmsg->ibm_version != IBLND_MSG_VERSION_1)
2234                 goto failed;
2235         if (reqmsg->ibm_magic == __swab32(IBLND_MSG_MAGIC) &&
2236             reqmsg->ibm_version != __swab16(IBLND_MSG_VERSION) &&
2237             reqmsg->ibm_version != __swab16(IBLND_MSG_VERSION_1))
2238                 goto failed;
2239
2240         rc = kiblnd_unpack_msg(reqmsg, priv_nob);
2241         if (rc != 0) {
2242                 CERROR("Can't parse connection request: %d\n", rc);
2243                 goto failed;
2244         }
2245
2246         nid = reqmsg->ibm_srcnid;
2247         ni  = lnet_net2ni(LNET_NIDNET(reqmsg->ibm_dstnid));
2248
2249         if (ni != NULL) {
2250                 net = (kib_net_t *)ni->ni_data;
2251                 rej.ibr_incarnation = net->ibn_incarnation;
2252         }
2253
2254         if (ni == NULL ||                        /* no matching net */
2255             ni->ni_nid != reqmsg->ibm_dstnid ||   /* right NET, wrong NID! */
2256             net->ibn_dev != ibdev) {          /* wrong device */
2257                 CERROR("Can't accept %s on %s (%s:%d:%u.%u.%u.%u): "
2258                        "bad dst nid %s\n", libcfs_nid2str(nid),
2259                        ni == NULL ? "NA" : libcfs_nid2str(ni->ni_nid),
2260                        ibdev->ibd_ifname, ibdev->ibd_nnets,
2261                        HIPQUAD(ibdev->ibd_ifip),
2262                        libcfs_nid2str(reqmsg->ibm_dstnid));
2263
2264                 goto failed;
2265         }
2266
2267        /* check time stamp as soon as possible */
2268         if (reqmsg->ibm_dststamp != 0 &&
2269             reqmsg->ibm_dststamp != net->ibn_incarnation) {
2270                 CWARN("Stale connection request\n");
2271                 rej.ibr_why = IBLND_REJECT_CONN_STALE;
2272                 goto failed;
2273         }
2274
2275         /* I can accept peer's version */
2276         version = reqmsg->ibm_version;
2277
2278         if (reqmsg->ibm_type != IBLND_MSG_CONNREQ) {
2279                 CERROR("Unexpected connreq msg type: %x from %s\n",
2280                        reqmsg->ibm_type, libcfs_nid2str(nid));
2281                 goto failed;
2282         }
2283
2284         if (reqmsg->ibm_u.connparams.ibcp_queue_depth !=
2285             IBLND_MSG_QUEUE_SIZE(version)) {
2286                 CERROR("Can't accept %s: incompatible queue depth %d (%d wanted)\n",
2287                        libcfs_nid2str(nid), reqmsg->ibm_u.connparams.ibcp_queue_depth,
2288                        IBLND_MSG_QUEUE_SIZE(version));
2289
2290                 if (version == IBLND_MSG_VERSION)
2291                         rej.ibr_why = IBLND_REJECT_MSG_QUEUE_SIZE;
2292
2293                 goto failed;
2294         }
2295
2296         if (reqmsg->ibm_u.connparams.ibcp_max_frags !=
2297             IBLND_RDMA_FRAGS(version)) {
2298                 CERROR("Can't accept %s(version %x): "
2299                        "incompatible max_frags %d (%d wanted)\n",
2300                        libcfs_nid2str(nid), version,
2301                        reqmsg->ibm_u.connparams.ibcp_max_frags,
2302                        IBLND_RDMA_FRAGS(version));
2303
2304                 if (version == IBLND_MSG_VERSION)
2305                         rej.ibr_why = IBLND_REJECT_RDMA_FRAGS;
2306
2307                 goto failed;
2308
2309         }
2310
2311         if (reqmsg->ibm_u.connparams.ibcp_max_msg_size > IBLND_MSG_SIZE) {
2312                 CERROR("Can't accept %s: message size %d too big (%d max)\n",
2313                        libcfs_nid2str(nid),
2314                        reqmsg->ibm_u.connparams.ibcp_max_msg_size,
2315                        IBLND_MSG_SIZE);
2316                 goto failed;
2317         }
2318
2319         /* assume 'nid' is a new peer; create  */
2320         rc = kiblnd_create_peer(ni, &peer, nid);
2321         if (rc != 0) {
2322                 CERROR("Can't create peer for %s\n", libcfs_nid2str(nid));
2323                 rej.ibr_why = IBLND_REJECT_NO_RESOURCES;
2324                 goto failed;
2325         }
2326
2327         write_lock_irqsave(g_lock, flags);
2328
2329         peer2 = kiblnd_find_peer_locked(nid);
2330         if (peer2 != NULL) {
2331                 if (peer2->ibp_version == 0) {
2332                         peer2->ibp_version     = version;
2333                         peer2->ibp_incarnation = reqmsg->ibm_srcstamp;
2334                 }
2335
2336                 /* not the guy I've talked with */
2337                 if (peer2->ibp_incarnation != reqmsg->ibm_srcstamp ||
2338                     peer2->ibp_version     != version) {
2339                         kiblnd_close_peer_conns_locked(peer2, -ESTALE);
2340                         write_unlock_irqrestore(g_lock, flags);
2341
2342                         CWARN("Conn stale %s [old ver: %x, new ver: %x]\n",
2343                               libcfs_nid2str(nid), peer2->ibp_version, version);
2344
2345                         kiblnd_peer_decref(peer);
2346                         rej.ibr_why = IBLND_REJECT_CONN_STALE;
2347                         goto failed;
2348                 }
2349
2350                 /* tie-break connection race in favour of the higher NID */
2351                 if (peer2->ibp_connecting != 0 &&
2352                     nid < ni->ni_nid) {
2353                         write_unlock_irqrestore(g_lock, flags);
2354
2355                         CWARN("Conn race %s\n", libcfs_nid2str(peer2->ibp_nid));
2356
2357                         kiblnd_peer_decref(peer);
2358                         rej.ibr_why = IBLND_REJECT_CONN_RACE;
2359                         goto failed;
2360                 }
2361
2362                 peer2->ibp_accepting++;
2363                 kiblnd_peer_addref(peer2);
2364
2365                 write_unlock_irqrestore(g_lock, flags);
2366                 kiblnd_peer_decref(peer);
2367                 peer = peer2;
2368         } else {
2369                 /* Brand new peer */
2370                 LASSERT (peer->ibp_accepting == 0);
2371                 LASSERT (peer->ibp_version == 0 &&
2372                          peer->ibp_incarnation == 0);
2373
2374                 peer->ibp_accepting   = 1;
2375                 peer->ibp_version     = version;
2376                 peer->ibp_incarnation = reqmsg->ibm_srcstamp;
2377
2378                 /* I have a ref on ni that prevents it being shutdown */
2379                 LASSERT (net->ibn_shutdown == 0);
2380
2381                 kiblnd_peer_addref(peer);
2382                 list_add_tail(&peer->ibp_list, kiblnd_nid2peerlist(nid));
2383
2384                 write_unlock_irqrestore(g_lock, flags);
2385         }
2386
2387         conn = kiblnd_create_conn(peer, cmid, IBLND_CONN_PASSIVE_WAIT, version);
2388         if (conn == NULL) {
2389                 kiblnd_peer_connect_failed(peer, 0, -ENOMEM);
2390                 kiblnd_peer_decref(peer);
2391                 rej.ibr_why = IBLND_REJECT_NO_RESOURCES;
2392                 goto failed;
2393         }
2394
2395         /* conn now "owns" cmid, so I return success from here on to ensure the
2396          * CM callback doesn't destroy cmid. */
2397
2398         conn->ibc_incarnation      = reqmsg->ibm_srcstamp;
2399         conn->ibc_credits         = IBLND_MSG_QUEUE_SIZE(version);
2400         conn->ibc_reserved_credits = IBLND_MSG_QUEUE_SIZE(version);
2401         LASSERT (conn->ibc_credits + conn->ibc_reserved_credits + IBLND_OOB_MSGS(version)
2402                  <= IBLND_RX_MSGS(version));
2403
2404         ackmsg = &conn->ibc_connvars->cv_msg;
2405         memset(ackmsg, 0, sizeof(*ackmsg));
2406
2407         kiblnd_init_msg(ackmsg, IBLND_MSG_CONNACK,
2408                         sizeof(ackmsg->ibm_u.connparams));
2409         ackmsg->ibm_u.connparams.ibcp_queue_depth  = IBLND_MSG_QUEUE_SIZE(version);
2410         ackmsg->ibm_u.connparams.ibcp_max_msg_size = IBLND_MSG_SIZE;
2411         ackmsg->ibm_u.connparams.ibcp_max_frags    = IBLND_RDMA_FRAGS(version);
2412
2413         kiblnd_pack_msg(ni, ackmsg, version, 0, nid, reqmsg->ibm_srcstamp);
2414
2415         memset(&cp, 0, sizeof(cp));
2416         cp.private_data = ackmsg;
2417         cp.private_data_len    = ackmsg->ibm_nob;
2418         cp.responder_resources = 0;          /* No atomic ops or RDMA reads */
2419         cp.initiator_depth     = 0;
2420         cp.flow_control = 1;
2421         cp.retry_count   = *kiblnd_tunables.kib_retry_count;
2422         cp.rnr_retry_count     = *kiblnd_tunables.kib_rnr_retry_count;
2423
2424         CDEBUG(D_NET, "Accept %s\n", libcfs_nid2str(nid));
2425
2426         rc = rdma_accept(cmid, &cp);
2427         if (rc != 0) {
2428                 CERROR("Can't accept %s: %d\n", libcfs_nid2str(nid), rc);
2429                 rej.ibr_version = version;
2430                 rej.ibr_why     = IBLND_REJECT_FATAL;
2431
2432                 kiblnd_reject(cmid, &rej);
2433                 kiblnd_connreq_done(conn, rc);
2434                 kiblnd_conn_decref(conn);
2435         }
2436
2437         lnet_ni_decref(ni);
2438         return 0;
2439
2440  failed:
2441         if (ni != NULL)
2442                 lnet_ni_decref(ni);
2443
2444         rej.ibr_version = version;
2445         rej.ibr_cp.ibcp_queue_depth = IBLND_MSG_QUEUE_SIZE(version);
2446         rej.ibr_cp.ibcp_max_frags   = IBLND_RDMA_FRAGS(version);
2447         kiblnd_reject(cmid, &rej);
2448
2449         return -ECONNREFUSED;
2450 }
2451
2452 void
2453 kiblnd_reconnect (kib_conn_t *conn, int version,
2454                   __u64 incarnation, int why, kib_connparams_t *cp)
2455 {
2456         kib_peer_t    *peer = conn->ibc_peer;
2457         char      *reason;
2458         int         retry = 0;
2459         unsigned long  flags;
2460
2461         LASSERT (conn->ibc_state == IBLND_CONN_ACTIVE_CONNECT);
2462         LASSERT (peer->ibp_connecting > 0);     /* 'conn' at least */
2463
2464         write_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
2465
2466         /* retry connection if it's still needed and no other connection
2467          * attempts (active or passive) are in progress
2468          * NB: reconnect is still needed even when ibp_tx_queue is
2469          * empty if ibp_version != version because reconnect may be
2470          * initiated by kiblnd_query() */
2471         if ((!list_empty(&peer->ibp_tx_queue) ||
2472              peer->ibp_version != version) &&
2473             peer->ibp_connecting == 1 &&
2474             peer->ibp_accepting == 0) {
2475                 retry = 1;
2476                 peer->ibp_connecting++;
2477
2478                 peer->ibp_version     = version;
2479                 peer->ibp_incarnation = incarnation;
2480         }
2481
2482         write_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
2483
2484         if (!retry)
2485                 return;
2486
2487         switch (why) {
2488         default:
2489                 reason = "Unknown";
2490                 break;
2491
2492         case IBLND_REJECT_CONN_STALE:
2493                 reason = "stale";
2494                 break;
2495
2496         case IBLND_REJECT_CONN_RACE:
2497                 reason = "conn race";
2498                 break;
2499
2500         case IBLND_REJECT_CONN_UNCOMPAT:
2501                 reason = "version negotiation";
2502                 break;
2503         }
2504
2505         CNETERR("%s: retrying (%s), %x, %x, "
2506                 "queue_dep: %d, max_frag: %d, msg_size: %d\n",
2507                 libcfs_nid2str(peer->ibp_nid),
2508                 reason, IBLND_MSG_VERSION, version,
2509                 cp != NULL? cp->ibcp_queue_depth :IBLND_MSG_QUEUE_SIZE(version),
2510                 cp != NULL? cp->ibcp_max_frags   : IBLND_RDMA_FRAGS(version),
2511                 cp != NULL? cp->ibcp_max_msg_size: IBLND_MSG_SIZE);
2512
2513         kiblnd_connect_peer(peer);
2514 }
2515
2516 void
2517 kiblnd_rejected (kib_conn_t *conn, int reason, void *priv, int priv_nob)
2518 {
2519         kib_peer_t    *peer = conn->ibc_peer;
2520
2521         LASSERT (!in_interrupt());
2522         LASSERT (conn->ibc_state == IBLND_CONN_ACTIVE_CONNECT);
2523
2524         switch (reason) {
2525         case IB_CM_REJ_STALE_CONN:
2526                 kiblnd_reconnect(conn, IBLND_MSG_VERSION, 0,
2527                                  IBLND_REJECT_CONN_STALE, NULL);
2528                 break;
2529
2530         case IB_CM_REJ_INVALID_SERVICE_ID:
2531                 CNETERR("%s rejected: no listener at %d\n",
2532                         libcfs_nid2str(peer->ibp_nid),
2533                         *kiblnd_tunables.kib_service);
2534                 break;
2535
2536         case IB_CM_REJ_CONSUMER_DEFINED:
2537                 if (priv_nob >= offsetof(kib_rej_t, ibr_padding)) {
2538                         kib_rej_t       *rej     = priv;
2539                         kib_connparams_t *cp      = NULL;
2540                         int            flip     = 0;
2541                         __u64        incarnation = -1;
2542
2543                         /* NB. default incarnation is -1 because:
2544                          * a) V1 will ignore dst incarnation in connreq.
2545                          * b) V2 will provide incarnation while rejecting me,
2546                          *    -1 will be overwrote.
2547                          *
2548                          * if I try to connect to a V1 peer with V2 protocol,
2549                          * it rejected me then upgrade to V2, I have no idea
2550                          * about the upgrading and try to reconnect with V1,
2551                          * in this case upgraded V2 can find out I'm trying to
2552                          * talk to the old guy and reject me(incarnation is -1).
2553                          */
2554
2555                         if (rej->ibr_magic == __swab32(IBLND_MSG_MAGIC) ||
2556                             rej->ibr_magic == __swab32(LNET_PROTO_MAGIC)) {
2557                                 __swab32s(&rej->ibr_magic);
2558                                 __swab16s(&rej->ibr_version);
2559                                 flip = 1;
2560                         }
2561
2562                         if (priv_nob >= sizeof(kib_rej_t) &&
2563                             rej->ibr_version > IBLND_MSG_VERSION_1) {
2564                                 /* priv_nob is always 148 in current version
2565                                  * of OFED, so we still need to check version.
2566                                  * (define of IB_CM_REJ_PRIVATE_DATA_SIZE) */
2567                                 cp = &rej->ibr_cp;
2568
2569                                 if (flip) {
2570                                         __swab64s(&rej->ibr_incarnation);
2571                                         __swab16s(&cp->ibcp_queue_depth);
2572                                         __swab16s(&cp->ibcp_max_frags);
2573                                         __swab32s(&cp->ibcp_max_msg_size);
2574                                 }
2575
2576                                 incarnation = rej->ibr_incarnation;
2577                         }
2578
2579                         if (rej->ibr_magic != IBLND_MSG_MAGIC &&
2580                             rej->ibr_magic != LNET_PROTO_MAGIC) {
2581                                 CERROR("%s rejected: consumer defined fatal error\n",
2582                                        libcfs_nid2str(peer->ibp_nid));
2583                                 break;
2584                         }
2585
2586                         if (rej->ibr_version != IBLND_MSG_VERSION &&
2587                             rej->ibr_version != IBLND_MSG_VERSION_1) {
2588                                 CERROR("%s rejected: o2iblnd version %x error\n",
2589                                        libcfs_nid2str(peer->ibp_nid),
2590                                        rej->ibr_version);
2591                                 break;
2592                         }
2593
2594                         if (rej->ibr_why     == IBLND_REJECT_FATAL &&
2595                             rej->ibr_version == IBLND_MSG_VERSION_1) {
2596                                 CDEBUG(D_NET, "rejected by old version peer %s: %x\n",
2597                                        libcfs_nid2str(peer->ibp_nid), rej->ibr_version);
2598
2599                                 if (conn->ibc_version != IBLND_MSG_VERSION_1)
2600                                         rej->ibr_why = IBLND_REJECT_CONN_UNCOMPAT;
2601                         }
2602
2603                         switch (rej->ibr_why) {
2604                         case IBLND_REJECT_CONN_RACE:
2605                         case IBLND_REJECT_CONN_STALE:
2606                         case IBLND_REJECT_CONN_UNCOMPAT:
2607                                 kiblnd_reconnect(conn, rej->ibr_version,
2608                                                  incarnation, rej->ibr_why, cp);
2609                                 break;
2610
2611                         case IBLND_REJECT_MSG_QUEUE_SIZE:
2612                                 CERROR("%s rejected: incompatible message queue depth %d, %d\n",
2613                                        libcfs_nid2str(peer->ibp_nid), cp->ibcp_queue_depth,
2614                                        IBLND_MSG_QUEUE_SIZE(conn->ibc_version));
2615                                 break;
2616
2617                         case IBLND_REJECT_RDMA_FRAGS:
2618                                 CERROR("%s rejected: incompatible # of RDMA fragments %d, %d\n",
2619                                        libcfs_nid2str(peer->ibp_nid), cp->ibcp_max_frags,
2620                                        IBLND_RDMA_FRAGS(conn->ibc_version));
2621                                 break;
2622
2623                         case IBLND_REJECT_NO_RESOURCES:
2624                                 CERROR("%s rejected: o2iblnd no resources\n",
2625                                        libcfs_nid2str(peer->ibp_nid));
2626                                 break;
2627
2628                         case IBLND_REJECT_FATAL:
2629                                 CERROR("%s rejected: o2iblnd fatal error\n",
2630                                        libcfs_nid2str(peer->ibp_nid));
2631                                 break;
2632
2633                         default:
2634                                 CERROR("%s rejected: o2iblnd reason %d\n",
2635                                        libcfs_nid2str(peer->ibp_nid),
2636                                        rej->ibr_why);
2637                                 break;
2638                         }
2639                         break;
2640                 }
2641                 /* fall through */
2642         default:
2643                 CNETERR("%s rejected: reason %d, size %d\n",
2644                         libcfs_nid2str(peer->ibp_nid), reason, priv_nob);
2645                 break;
2646         }
2647
2648         kiblnd_connreq_done(conn, -ECONNREFUSED);
2649 }
2650
2651 void
2652 kiblnd_check_connreply (kib_conn_t *conn, void *priv, int priv_nob)
2653 {
2654         kib_peer_t    *peer = conn->ibc_peer;
2655         lnet_ni_t     *ni   = peer->ibp_ni;
2656         kib_net_t     *net  = ni->ni_data;
2657         kib_msg_t     *msg  = priv;
2658         int         ver  = conn->ibc_version;
2659         int         rc   = kiblnd_unpack_msg(msg, priv_nob);
2660         unsigned long  flags;
2661
2662         LASSERT (net != NULL);
2663
2664         if (rc != 0) {
2665                 CERROR("Can't unpack connack from %s: %d\n",
2666                        libcfs_nid2str(peer->ibp_nid), rc);
2667                 goto failed;
2668         }
2669
2670         if (msg->ibm_type != IBLND_MSG_CONNACK) {
2671                 CERROR("Unexpected message %d from %s\n",
2672                        msg->ibm_type, libcfs_nid2str(peer->ibp_nid));
2673                 rc = -EPROTO;
2674                 goto failed;
2675         }
2676
2677         if (ver != msg->ibm_version) {
2678                 CERROR("%s replied version %x is different with "
2679                        "requested version %x\n",
2680                        libcfs_nid2str(peer->ibp_nid), msg->ibm_version, ver);
2681                 rc = -EPROTO;
2682                 goto failed;
2683         }
2684
2685         if (msg->ibm_u.connparams.ibcp_queue_depth !=
2686             IBLND_MSG_QUEUE_SIZE(ver)) {
2687                 CERROR("%s has incompatible queue depth %d(%d wanted)\n",
2688                        libcfs_nid2str(peer->ibp_nid),
2689                        msg->ibm_u.connparams.ibcp_queue_depth,
2690                        IBLND_MSG_QUEUE_SIZE(ver));
2691                 rc = -EPROTO;
2692                 goto failed;
2693         }
2694
2695         if (msg->ibm_u.connparams.ibcp_max_frags !=
2696             IBLND_RDMA_FRAGS(ver)) {
2697                 CERROR("%s has incompatible max_frags %d (%d wanted)\n",
2698                        libcfs_nid2str(peer->ibp_nid),
2699                        msg->ibm_u.connparams.ibcp_max_frags,
2700                        IBLND_RDMA_FRAGS(ver));
2701                 rc = -EPROTO;
2702                 goto failed;
2703         }
2704
2705         if (msg->ibm_u.connparams.ibcp_max_msg_size > IBLND_MSG_SIZE) {
2706                 CERROR("%s max message size %d too big (%d max)\n",
2707                        libcfs_nid2str(peer->ibp_nid),
2708                        msg->ibm_u.connparams.ibcp_max_msg_size,
2709                        IBLND_MSG_SIZE);
2710                 rc = -EPROTO;
2711                 goto failed;
2712         }
2713
2714         read_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
2715         if (msg->ibm_dstnid == ni->ni_nid &&
2716             msg->ibm_dststamp == net->ibn_incarnation)
2717                 rc = 0;
2718         else
2719                 rc = -ESTALE;
2720         read_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
2721
2722         if (rc != 0) {
2723                 CERROR("Bad connection reply from %s, rc = %d, "
2724                        "version: %x max_frags: %d\n",
2725                        libcfs_nid2str(peer->ibp_nid), rc,
2726                        msg->ibm_version, msg->ibm_u.connparams.ibcp_max_frags);
2727                 goto failed;
2728         }
2729
2730         conn->ibc_incarnation      = msg->ibm_srcstamp;
2731         conn->ibc_credits         =
2732         conn->ibc_reserved_credits = IBLND_MSG_QUEUE_SIZE(ver);
2733         LASSERT (conn->ibc_credits + conn->ibc_reserved_credits + IBLND_OOB_MSGS(ver)
2734                  <= IBLND_RX_MSGS(ver));
2735
2736         kiblnd_connreq_done(conn, 0);
2737         return;
2738
2739  failed:
2740         /* NB My QP has already established itself, so I handle anything going
2741          * wrong here by setting ibc_comms_error.
2742          * kiblnd_connreq_done(0) moves the conn state to ESTABLISHED, but then
2743          * immediately tears it down. */
2744
2745         LASSERT (rc != 0);
2746         conn->ibc_comms_error = rc;
2747         kiblnd_connreq_done(conn, 0);
2748 }
2749
2750 int
2751 kiblnd_active_connect (struct rdma_cm_id *cmid)
2752 {
2753         kib_peer_t            *peer = (kib_peer_t *)cmid->context;
2754         kib_conn_t            *conn;
2755         kib_msg_t              *msg;
2756         struct rdma_conn_param   cp;
2757         int                   version;
2758         __u64               incarnation;
2759         unsigned long       flags;
2760         int                   rc;
2761
2762         read_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
2763
2764         incarnation = peer->ibp_incarnation;
2765         version     = (peer->ibp_version == 0) ? IBLND_MSG_VERSION :
2766                                                  peer->ibp_version;
2767
2768         read_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
2769
2770         conn = kiblnd_create_conn(peer, cmid, IBLND_CONN_ACTIVE_CONNECT, version);
2771         if (conn == NULL) {
2772                 kiblnd_peer_connect_failed(peer, 1, -ENOMEM);
2773                 kiblnd_peer_decref(peer); /* lose cmid's ref */
2774                 return -ENOMEM;
2775         }
2776
2777         /* conn "owns" cmid now, so I return success from here on to ensure the
2778          * CM callback doesn't destroy cmid. conn also takes over cmid's ref
2779          * on peer */
2780
2781         msg = &conn->ibc_connvars->cv_msg;
2782
2783         memset(msg, 0, sizeof(*msg));
2784         kiblnd_init_msg(msg, IBLND_MSG_CONNREQ, sizeof(msg->ibm_u.connparams));
2785         msg->ibm_u.connparams.ibcp_queue_depth  = IBLND_MSG_QUEUE_SIZE(version);
2786         msg->ibm_u.connparams.ibcp_max_frags    = IBLND_RDMA_FRAGS(version);
2787         msg->ibm_u.connparams.ibcp_max_msg_size = IBLND_MSG_SIZE;
2788
2789         kiblnd_pack_msg(peer->ibp_ni, msg, version,
2790                         0, peer->ibp_nid, incarnation);
2791
2792         memset(&cp, 0, sizeof(cp));
2793         cp.private_data = msg;
2794         cp.private_data_len    = msg->ibm_nob;
2795         cp.responder_resources = 0;          /* No atomic ops or RDMA reads */
2796         cp.initiator_depth     = 0;
2797         cp.flow_control = 1;
2798         cp.retry_count   = *kiblnd_tunables.kib_retry_count;
2799         cp.rnr_retry_count     = *kiblnd_tunables.kib_rnr_retry_count;
2800
2801         LASSERT(cmid->context == (void *)conn);
2802         LASSERT(conn->ibc_cmid == cmid);
2803
2804         rc = rdma_connect(cmid, &cp);
2805         if (rc != 0) {
2806                 CERROR("Can't connect to %s: %d\n",
2807                        libcfs_nid2str(peer->ibp_nid), rc);
2808                 kiblnd_connreq_done(conn, rc);
2809                 kiblnd_conn_decref(conn);
2810         }
2811
2812         return 0;
2813 }
2814
2815 int
2816 kiblnd_cm_callback(struct rdma_cm_id *cmid, struct rdma_cm_event *event)
2817 {
2818         kib_peer_t  *peer;
2819         kib_conn_t  *conn;
2820         int       rc;
2821
2822         switch (event->event) {
2823         default:
2824                 CERROR("Unexpected event: %d, status: %d\n",
2825                        event->event, event->status);
2826                 LBUG();
2827
2828         case RDMA_CM_EVENT_CONNECT_REQUEST:
2829                 /* destroy cmid on failure */
2830                 rc = kiblnd_passive_connect(cmid,
2831                                             (void *)KIBLND_CONN_PARAM(event),
2832                                             KIBLND_CONN_PARAM_LEN(event));
2833                 CDEBUG(D_NET, "connreq: %d\n", rc);
2834                 return rc;
2835
2836         case RDMA_CM_EVENT_ADDR_ERROR:
2837                 peer = (kib_peer_t *)cmid->context;
2838                 CNETERR("%s: ADDR ERROR %d\n",
2839                        libcfs_nid2str(peer->ibp_nid), event->status);
2840                 kiblnd_peer_connect_failed(peer, 1, -EHOSTUNREACH);
2841                 kiblnd_peer_decref(peer);
2842                 return -EHOSTUNREACH;      /* rc != 0 destroys cmid */
2843
2844         case RDMA_CM_EVENT_ADDR_RESOLVED:
2845                 peer = (kib_peer_t *)cmid->context;
2846
2847                 CDEBUG(D_NET,"%s Addr resolved: %d\n",
2848                        libcfs_nid2str(peer->ibp_nid), event->status);
2849
2850                 if (event->status != 0) {
2851                         CNETERR("Can't resolve address for %s: %d\n",
2852                                 libcfs_nid2str(peer->ibp_nid), event->status);
2853                         rc = event->status;
2854                 } else {
2855                         rc = rdma_resolve_route(
2856                                 cmid, *kiblnd_tunables.kib_timeout * 1000);
2857                         if (rc == 0)
2858                                 return 0;
2859                         /* Can't initiate route resolution */
2860                         CERROR("Can't resolve route for %s: %d\n",
2861                                libcfs_nid2str(peer->ibp_nid), rc);
2862                 }
2863                 kiblnd_peer_connect_failed(peer, 1, rc);
2864                 kiblnd_peer_decref(peer);
2865                 return rc;                    /* rc != 0 destroys cmid */
2866
2867         case RDMA_CM_EVENT_ROUTE_ERROR:
2868                 peer = (kib_peer_t *)cmid->context;
2869                 CNETERR("%s: ROUTE ERROR %d\n",
2870                         libcfs_nid2str(peer->ibp_nid), event->status);
2871                 kiblnd_peer_connect_failed(peer, 1, -EHOSTUNREACH);
2872                 kiblnd_peer_decref(peer);
2873                 return -EHOSTUNREACH;      /* rc != 0 destroys cmid */
2874
2875         case RDMA_CM_EVENT_ROUTE_RESOLVED:
2876                 peer = (kib_peer_t *)cmid->context;
2877                 CDEBUG(D_NET,"%s Route resolved: %d\n",
2878                        libcfs_nid2str(peer->ibp_nid), event->status);
2879
2880                 if (event->status == 0)
2881                         return kiblnd_active_connect(cmid);
2882
2883                 CNETERR("Can't resolve route for %s: %d\n",
2884                        libcfs_nid2str(peer->ibp_nid), event->status);
2885                 kiblnd_peer_connect_failed(peer, 1, event->status);
2886                 kiblnd_peer_decref(peer);
2887                 return event->status;      /* rc != 0 destroys cmid */
2888
2889         case RDMA_CM_EVENT_UNREACHABLE:
2890                 conn = (kib_conn_t *)cmid->context;
2891                 LASSERT(conn->ibc_state == IBLND_CONN_ACTIVE_CONNECT ||
2892                         conn->ibc_state == IBLND_CONN_PASSIVE_WAIT);
2893                 CNETERR("%s: UNREACHABLE %d\n",
2894                        libcfs_nid2str(conn->ibc_peer->ibp_nid), event->status);
2895                 kiblnd_connreq_done(conn, -ENETDOWN);
2896                 kiblnd_conn_decref(conn);
2897                 return 0;
2898
2899         case RDMA_CM_EVENT_CONNECT_ERROR:
2900                 conn = (kib_conn_t *)cmid->context;
2901                 LASSERT(conn->ibc_state == IBLND_CONN_ACTIVE_CONNECT ||
2902                         conn->ibc_state == IBLND_CONN_PASSIVE_WAIT);
2903                 CNETERR("%s: CONNECT ERROR %d\n",
2904                         libcfs_nid2str(conn->ibc_peer->ibp_nid), event->status);
2905                 kiblnd_connreq_done(conn, -ENOTCONN);
2906                 kiblnd_conn_decref(conn);
2907                 return 0;
2908
2909         case RDMA_CM_EVENT_REJECTED:
2910                 conn = (kib_conn_t *)cmid->context;
2911                 switch (conn->ibc_state) {
2912                 default:
2913                         LBUG();
2914
2915                 case IBLND_CONN_PASSIVE_WAIT:
2916                         CERROR ("%s: REJECTED %d\n",
2917                                 libcfs_nid2str(conn->ibc_peer->ibp_nid),
2918                                 event->status);
2919                         kiblnd_connreq_done(conn, -ECONNRESET);
2920                         break;
2921
2922                 case IBLND_CONN_ACTIVE_CONNECT:
2923                         kiblnd_rejected(conn, event->status,
2924                                         (void *)KIBLND_CONN_PARAM(event),
2925                                         KIBLND_CONN_PARAM_LEN(event));
2926                         break;
2927                 }
2928                 kiblnd_conn_decref(conn);
2929                 return 0;
2930
2931         case RDMA_CM_EVENT_ESTABLISHED:
2932                 conn = (kib_conn_t *)cmid->context;
2933                 switch (conn->ibc_state) {
2934                 default:
2935                         LBUG();
2936
2937                 case IBLND_CONN_PASSIVE_WAIT:
2938                         CDEBUG(D_NET, "ESTABLISHED (passive): %s\n",
2939                                libcfs_nid2str(conn->ibc_peer->ibp_nid));
2940                         kiblnd_connreq_done(conn, 0);
2941                         break;
2942
2943                 case IBLND_CONN_ACTIVE_CONNECT:
2944                         CDEBUG(D_NET, "ESTABLISHED(active): %s\n",
2945                                libcfs_nid2str(conn->ibc_peer->ibp_nid));
2946                         kiblnd_check_connreply(conn,
2947                                                (void *)KIBLND_CONN_PARAM(event),
2948                                                KIBLND_CONN_PARAM_LEN(event));
2949                         break;
2950                 }
2951                 /* net keeps its ref on conn! */
2952                 return 0;
2953
2954         case RDMA_CM_EVENT_TIMEWAIT_EXIT:
2955                 CDEBUG(D_NET, "Ignore TIMEWAIT_EXIT event\n");
2956                 return 0;
2957         case RDMA_CM_EVENT_DISCONNECTED:
2958                 conn = (kib_conn_t *)cmid->context;
2959                 if (conn->ibc_state < IBLND_CONN_ESTABLISHED) {
2960                         CERROR("%s DISCONNECTED\n",
2961                                libcfs_nid2str(conn->ibc_peer->ibp_nid));
2962                         kiblnd_connreq_done(conn, -ECONNRESET);
2963                 } else {
2964                         kiblnd_close_conn(conn, 0);
2965                 }
2966                 kiblnd_conn_decref(conn);
2967                 cmid->context = NULL;
2968                 return 0;
2969
2970         case RDMA_CM_EVENT_DEVICE_REMOVAL:
2971                 LCONSOLE_ERROR_MSG(0x131,
2972                                    "Received notification of device removal\n"
2973                                    "Please shutdown LNET to allow this to proceed\n");
2974                 /* Can't remove network from underneath LNET for now, so I have
2975                  * to ignore this */
2976                 return 0;
2977
2978         case RDMA_CM_EVENT_ADDR_CHANGE:
2979                 LCONSOLE_INFO("Physical link changed (eg hca/port)\n");
2980                 return 0;
2981         }
2982 }
2983
2984 static int
2985 kiblnd_check_txs_locked(kib_conn_t *conn, struct list_head *txs)
2986 {
2987         kib_tx_t          *tx;
2988         struct list_head        *ttmp;
2989
2990         list_for_each (ttmp, txs) {
2991                 tx = list_entry (ttmp, kib_tx_t, tx_list);
2992
2993                 if (txs != &conn->ibc_active_txs) {
2994                         LASSERT (tx->tx_queued);
2995                 } else {
2996                         LASSERT (!tx->tx_queued);
2997                         LASSERT (tx->tx_waiting || tx->tx_sending != 0);
2998                 }
2999
3000                 if (cfs_time_aftereq (jiffies, tx->tx_deadline)) {
3001                         CERROR("Timed out tx: %s, %lu seconds\n",
3002                                kiblnd_queue2str(conn, txs),
3003                                cfs_duration_sec(jiffies - tx->tx_deadline));
3004                         return 1;
3005                 }
3006         }
3007
3008         return 0;
3009 }
3010
3011 static int
3012 kiblnd_conn_timed_out_locked(kib_conn_t *conn)
3013 {
3014         return  kiblnd_check_txs_locked(conn, &conn->ibc_tx_queue) ||
3015                 kiblnd_check_txs_locked(conn, &conn->ibc_tx_noops) ||
3016                 kiblnd_check_txs_locked(conn, &conn->ibc_tx_queue_rsrvd) ||
3017                 kiblnd_check_txs_locked(conn, &conn->ibc_tx_queue_nocred) ||
3018                 kiblnd_check_txs_locked(conn, &conn->ibc_active_txs);
3019 }
3020
3021 void
3022 kiblnd_check_conns (int idx)
3023 {
3024         LIST_HEAD (closes);
3025         LIST_HEAD (checksends);
3026         struct list_head    *peers = &kiblnd_data.kib_peers[idx];
3027         struct list_head    *ptmp;
3028         kib_peer_t    *peer;
3029         kib_conn_t    *conn;
3030         struct list_head    *ctmp;
3031         unsigned long  flags;
3032
3033         /* NB. We expect to have a look at all the peers and not find any
3034          * RDMAs to time out, so we just use a shared lock while we
3035          * take a look... */
3036         read_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
3037
3038         list_for_each (ptmp, peers) {
3039                 peer = list_entry (ptmp, kib_peer_t, ibp_list);
3040
3041                 list_for_each (ctmp, &peer->ibp_conns) {
3042                         int timedout;
3043                         int sendnoop;
3044
3045                         conn = list_entry(ctmp, kib_conn_t, ibc_list);
3046
3047                         LASSERT (conn->ibc_state == IBLND_CONN_ESTABLISHED);
3048
3049                         spin_lock(&conn->ibc_lock);
3050
3051                         sendnoop = kiblnd_need_noop(conn);
3052                         timedout = kiblnd_conn_timed_out_locked(conn);
3053                         if (!sendnoop && !timedout) {
3054                                 spin_unlock(&conn->ibc_lock);
3055                                 continue;
3056                         }
3057
3058                         if (timedout) {
3059                                 CERROR("Timed out RDMA with %s (%lu): "
3060                                        "c: %u, oc: %u, rc: %u\n",
3061                                        libcfs_nid2str(peer->ibp_nid),
3062                                        cfs_duration_sec(cfs_time_current() -
3063                                                         peer->ibp_last_alive),
3064                                        conn->ibc_credits,
3065                                        conn->ibc_outstanding_credits,
3066                                        conn->ibc_reserved_credits);
3067                                 list_add(&conn->ibc_connd_list, &closes);
3068                         } else {
3069                                 list_add(&conn->ibc_connd_list,
3070                                              &checksends);
3071                         }
3072                         /* +ref for 'closes' or 'checksends' */
3073                         kiblnd_conn_addref(conn);
3074
3075                         spin_unlock(&conn->ibc_lock);
3076                 }
3077         }
3078
3079         read_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags);
3080
3081         /* Handle timeout by closing the whole
3082          * connection. We can only be sure RDMA activity
3083          * has ceased once the QP has been modified. */
3084         while (!list_empty(&closes)) {
3085                 conn = list_entry(closes.next,
3086                                       kib_conn_t, ibc_connd_list);
3087                 list_del(&conn->ibc_connd_list);
3088                 kiblnd_close_conn(conn, -ETIMEDOUT);
3089                 kiblnd_conn_decref(conn);
3090         }
3091
3092         /* In case we have enough credits to return via a
3093          * NOOP, but there were no non-blocking tx descs
3094          * free to do it last time... */
3095         while (!list_empty(&checksends)) {
3096                 conn = list_entry(checksends.next,
3097                                       kib_conn_t, ibc_connd_list);
3098                 list_del(&conn->ibc_connd_list);
3099                 kiblnd_check_sends(conn);
3100                 kiblnd_conn_decref(conn);
3101         }
3102 }
3103
3104 void
3105 kiblnd_disconnect_conn (kib_conn_t *conn)
3106 {
3107         LASSERT (!in_interrupt());
3108         LASSERT (current == kiblnd_data.kib_connd);
3109         LASSERT (conn->ibc_state == IBLND_CONN_CLOSING);
3110
3111         rdma_disconnect(conn->ibc_cmid);
3112         kiblnd_finalise_conn(conn);
3113
3114         kiblnd_peer_notify(conn->ibc_peer);
3115 }
3116
3117 int
3118 kiblnd_connd (void *arg)
3119 {
3120         wait_queue_t     wait;
3121         unsigned long      flags;
3122         kib_conn_t      *conn;
3123         int             timeout;
3124         int             i;
3125         int             dropped_lock;
3126         int             peer_index = 0;
3127         unsigned long      deadline = jiffies;
3128
3129         cfs_block_allsigs ();
3130
3131         init_waitqueue_entry_current (&wait);
3132         kiblnd_data.kib_connd = current;
3133
3134         spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags);
3135
3136         while (!kiblnd_data.kib_shutdown) {
3137
3138                 dropped_lock = 0;
3139
3140                 if (!list_empty (&kiblnd_data.kib_connd_zombies)) {
3141                         conn = list_entry(kiblnd_data. \
3142                                               kib_connd_zombies.next,
3143                                               kib_conn_t, ibc_list);
3144                         list_del(&conn->ibc_list);
3145
3146                         spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock,
3147                                                flags);
3148                         dropped_lock = 1;
3149
3150                         kiblnd_destroy_conn(conn);
3151
3152                         spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags);
3153                 }
3154
3155                 if (!list_empty(&kiblnd_data.kib_connd_conns)) {
3156                         conn = list_entry(kiblnd_data.kib_connd_conns.next,
3157                                               kib_conn_t, ibc_list);
3158                         list_del(&conn->ibc_list);
3159
3160                         spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock,
3161                                                flags);
3162                         dropped_lock = 1;
3163
3164                         kiblnd_disconnect_conn(conn);
3165                         kiblnd_conn_decref(conn);
3166
3167                         spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags);
3168                 }
3169
3170                 /* careful with the jiffy wrap... */
3171                 timeout = (int)(deadline - jiffies);
3172                 if (timeout <= 0) {
3173                         const int n = 4;
3174                         const int p = 1;
3175                         int       chunk = kiblnd_data.kib_peer_hash_size;
3176
3177                         spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock, flags);
3178                         dropped_lock = 1;
3179
3180                         /* Time to check for RDMA timeouts on a few more
3181                          * peers: I do checks every 'p' seconds on a
3182                          * proportion of the peer table and I need to check
3183                          * every connection 'n' times within a timeout
3184                          * interval, to ensure I detect a timeout on any
3185                          * connection within (n+1)/n times the timeout
3186                          * interval. */
3187
3188                         if (*kiblnd_tunables.kib_timeout > n * p)
3189                                 chunk = (chunk * n * p) /
3190                                         *kiblnd_tunables.kib_timeout;
3191                         if (chunk == 0)
3192                                 chunk = 1;
3193
3194                         for (i = 0; i < chunk; i++) {
3195                                 kiblnd_check_conns(peer_index);
3196                                 peer_index = (peer_index + 1) %
3197                                              kiblnd_data.kib_peer_hash_size;
3198                         }
3199
3200                         deadline += p * HZ;
3201                         spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags);
3202                 }
3203
3204                 if (dropped_lock)
3205                         continue;
3206
3207                 /* Nothing to do for 'timeout'  */
3208                 set_current_state(TASK_INTERRUPTIBLE);
3209                 add_wait_queue(&kiblnd_data.kib_connd_waitq, &wait);
3210                 spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock, flags);
3211
3212                 waitq_timedwait(&wait, TASK_INTERRUPTIBLE, timeout);
3213
3214                 set_current_state(TASK_RUNNING);
3215                 remove_wait_queue(&kiblnd_data.kib_connd_waitq, &wait);
3216                 spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags);
3217         }
3218
3219         spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock, flags);
3220
3221         kiblnd_thread_fini();
3222         return 0;
3223 }
3224
3225 void
3226 kiblnd_qp_event(struct ib_event *event, void *arg)
3227 {
3228         kib_conn_t *conn = arg;
3229
3230         switch (event->event) {
3231         case IB_EVENT_COMM_EST:
3232                 CDEBUG(D_NET, "%s established\n",
3233                        libcfs_nid2str(conn->ibc_peer->ibp_nid));
3234                 return;
3235
3236         default:
3237                 CERROR("%s: Async QP event type %d\n",
3238                        libcfs_nid2str(conn->ibc_peer->ibp_nid), event->event);
3239                 return;
3240         }
3241 }
3242
3243 void
3244 kiblnd_complete (struct ib_wc *wc)
3245 {
3246         switch (kiblnd_wreqid2type(wc->wr_id)) {
3247         default:
3248                 LBUG();
3249
3250         case IBLND_WID_RDMA:
3251                 /* We only get RDMA completion notification if it fails.  All
3252                  * subsequent work items, including the final SEND will fail
3253                  * too.  However we can't print out any more info about the
3254                  * failing RDMA because 'tx' might be back on the idle list or
3255                  * even reused already if we didn't manage to post all our work
3256                  * items */
3257                 CNETERR("RDMA (tx: %p) failed: %d\n",
3258                         kiblnd_wreqid2ptr(wc->wr_id), wc->status);
3259                 return;
3260
3261         case IBLND_WID_TX:
3262                 kiblnd_tx_complete(kiblnd_wreqid2ptr(wc->wr_id), wc->status);
3263                 return;
3264
3265         case IBLND_WID_RX:
3266                 kiblnd_rx_complete(kiblnd_wreqid2ptr(wc->wr_id), wc->status,
3267                                    wc->byte_len);
3268                 return;
3269         }
3270 }
3271
3272 void
3273 kiblnd_cq_completion(struct ib_cq *cq, void *arg)
3274 {
3275         /* NB I'm not allowed to schedule this conn once its refcount has
3276          * reached 0.  Since fundamentally I'm racing with scheduler threads
3277          * consuming my CQ I could be called after all completions have
3278          * occurred.  But in this case, ibc_nrx == 0 && ibc_nsends_posted == 0
3279          * and this CQ is about to be destroyed so I NOOP. */
3280         kib_conn_t              *conn = (kib_conn_t *)arg;
3281         struct kib_sched_info   *sched = conn->ibc_sched;
3282         unsigned long           flags;
3283
3284         LASSERT(cq == conn->ibc_cq);
3285
3286         spin_lock_irqsave(&sched->ibs_lock, flags);
3287
3288         conn->ibc_ready = 1;
3289
3290         if (!conn->ibc_scheduled &&
3291             (conn->ibc_nrx > 0 ||
3292              conn->ibc_nsends_posted > 0)) {
3293                 kiblnd_conn_addref(conn); /* +1 ref for sched_conns */
3294                 conn->ibc_scheduled = 1;
3295                 list_add_tail(&conn->ibc_sched_list, &sched->ibs_conns);
3296
3297                 if (waitqueue_active(&sched->ibs_waitq))
3298                         wake_up(&sched->ibs_waitq);
3299         }
3300
3301         spin_unlock_irqrestore(&sched->ibs_lock, flags);
3302 }
3303
3304 void
3305 kiblnd_cq_event(struct ib_event *event, void *arg)
3306 {
3307         kib_conn_t *conn = arg;
3308
3309         CERROR("%s: async CQ event type %d\n",
3310                libcfs_nid2str(conn->ibc_peer->ibp_nid), event->event);
3311 }
3312
3313 int
3314 kiblnd_scheduler(void *arg)
3315 {
3316         long                    id = (long)arg;
3317         struct kib_sched_info   *sched;
3318         kib_conn_t              *conn;
3319         wait_queue_t            wait;
3320         unsigned long           flags;
3321         struct ib_wc            wc;
3322         int                     did_something;
3323         int                     busy_loops = 0;
3324         int                     rc;
3325
3326         cfs_block_allsigs();
3327
3328         init_waitqueue_entry_current(&wait);
3329
3330         sched = kiblnd_data.kib_scheds[KIB_THREAD_CPT(id)];
3331
3332         rc = cfs_cpt_bind(lnet_cpt_table(), sched->ibs_cpt);
3333         if (rc != 0) {
3334                 CWARN("Failed to bind on CPT %d, please verify whether "
3335                       "all CPUs are healthy and reload modules if necessary, "
3336                       "otherwise your system might under risk of low "
3337                       "performance\n", sched->ibs_cpt);
3338         }
3339
3340         spin_lock_irqsave(&sched->ibs_lock, flags);
3341
3342         while (!kiblnd_data.kib_shutdown) {
3343                 if (busy_loops++ >= IBLND_RESCHED) {
3344                         spin_unlock_irqrestore(&sched->ibs_lock, flags);
3345
3346                         cond_resched();
3347                         busy_loops = 0;
3348
3349                         spin_lock_irqsave(&sched->ibs_lock, flags);
3350                 }
3351
3352                 did_something = 0;
3353
3354                 if (!list_empty(&sched->ibs_conns)) {
3355                         conn = list_entry(sched->ibs_conns.next,
3356                                               kib_conn_t, ibc_sched_list);
3357                         /* take over kib_sched_conns' ref on conn... */
3358                         LASSERT(conn->ibc_scheduled);
3359                         list_del(&conn->ibc_sched_list);
3360                         conn->ibc_ready = 0;
3361
3362                         spin_unlock_irqrestore(&sched->ibs_lock, flags);
3363
3364                         rc = ib_poll_cq(conn->ibc_cq, 1, &wc);
3365                         if (rc == 0) {
3366                                 rc = ib_req_notify_cq(conn->ibc_cq,
3367                                                       IB_CQ_NEXT_COMP);
3368                                 if (rc < 0) {
3369                                         CWARN("%s: ib_req_notify_cq failed: %d, "
3370                                               "closing connection\n",
3371                                               libcfs_nid2str(conn->ibc_peer->ibp_nid), rc);
3372                                         kiblnd_close_conn(conn, -EIO);
3373                                         kiblnd_conn_decref(conn);
3374                                         spin_lock_irqsave(&sched->ibs_lock,
3375                                                               flags);
3376                                         continue;
3377                                 }
3378
3379                                 rc = ib_poll_cq(conn->ibc_cq, 1, &wc);
3380                         }
3381
3382                         if (rc < 0) {
3383                                 CWARN("%s: ib_poll_cq failed: %d, "
3384                                       "closing connection\n",
3385                                       libcfs_nid2str(conn->ibc_peer->ibp_nid),
3386                                       rc);
3387                                 kiblnd_close_conn(conn, -EIO);
3388                                 kiblnd_conn_decref(conn);
3389                                 spin_lock_irqsave(&sched->ibs_lock, flags);
3390                                 continue;
3391                         }
3392
3393                         spin_lock_irqsave(&sched->ibs_lock, flags);
3394
3395                         if (rc != 0 || conn->ibc_ready) {
3396                                 /* There may be another completion waiting; get
3397                                  * another scheduler to check while I handle
3398                                  * this one... */
3399                                 /* +1 ref for sched_conns */
3400                                 kiblnd_conn_addref(conn);
3401                                 list_add_tail(&conn->ibc_sched_list,
3402                                                   &sched->ibs_conns);
3403                                 if (waitqueue_active(&sched->ibs_waitq))
3404                                         wake_up(&sched->ibs_waitq);
3405                         } else {
3406                                 conn->ibc_scheduled = 0;
3407                         }
3408
3409                         if (rc != 0) {
3410                                 spin_unlock_irqrestore(&sched->ibs_lock, flags);
3411                                 kiblnd_complete(&wc);
3412
3413                                 spin_lock_irqsave(&sched->ibs_lock, flags);
3414                         }
3415
3416                         kiblnd_conn_decref(conn); /* ...drop my ref from above */
3417                         did_something = 1;
3418                 }
3419
3420                 if (did_something)
3421                         continue;
3422
3423                 set_current_state(TASK_INTERRUPTIBLE);
3424                 add_wait_queue_exclusive(&sched->ibs_waitq, &wait);
3425                 spin_unlock_irqrestore(&sched->ibs_lock, flags);
3426
3427                 waitq_wait(&wait, TASK_INTERRUPTIBLE);
3428                 busy_loops = 0;
3429
3430                 remove_wait_queue(&sched->ibs_waitq, &wait);
3431                 set_current_state(TASK_RUNNING);
3432                 spin_lock_irqsave(&sched->ibs_lock, flags);
3433         }
3434
3435         spin_unlock_irqrestore(&sched->ibs_lock, flags);
3436
3437         kiblnd_thread_fini();
3438         return 0;
3439 }
3440
3441 int
3442 kiblnd_failover_thread(void *arg)
3443 {
3444         rwlock_t                *glock = &kiblnd_data.kib_global_lock;
3445         kib_dev_t        *dev;
3446         wait_queue_t     wait;
3447         unsigned long      flags;
3448         int             rc;
3449
3450         LASSERT (*kiblnd_tunables.kib_dev_failover != 0);
3451
3452         cfs_block_allsigs ();
3453
3454         init_waitqueue_entry_current(&wait);
3455         write_lock_irqsave(glock, flags);
3456
3457         while (!kiblnd_data.kib_shutdown) {
3458                 int     do_failover = 0;
3459                 int     long_sleep;
3460
3461                 list_for_each_entry(dev, &kiblnd_data.kib_failed_devs,
3462                                     ibd_fail_list) {
3463                         if (cfs_time_before(cfs_time_current(),
3464                                             dev->ibd_next_failover))
3465                                 continue;
3466                         do_failover = 1;
3467                         break;
3468                 }
3469
3470                 if (do_failover) {
3471                         list_del_init(&dev->ibd_fail_list);
3472                         dev->ibd_failover = 1;
3473                         write_unlock_irqrestore(glock, flags);
3474
3475                         rc = kiblnd_dev_failover(dev);
3476
3477                         write_lock_irqsave(glock, flags);
3478
3479                         LASSERT (dev->ibd_failover);
3480                         dev->ibd_failover = 0;
3481                         if (rc >= 0) { /* Device is OK or failover succeed */
3482                                 dev->ibd_next_failover = cfs_time_shift(3);
3483                                 continue;
3484                         }
3485
3486                         /* failed to failover, retry later */
3487                         dev->ibd_next_failover =
3488                                 cfs_time_shift(min(dev->ibd_failed_failover, 10));
3489                         if (kiblnd_dev_can_failover(dev)) {
3490                                 list_add_tail(&dev->ibd_fail_list,
3491                                               &kiblnd_data.kib_failed_devs);
3492                         }
3493
3494                         continue;
3495                 }
3496
3497                 /* long sleep if no more pending failover */
3498                 long_sleep = list_empty(&kiblnd_data.kib_failed_devs);
3499
3500                 set_current_state(TASK_INTERRUPTIBLE);
3501                 add_wait_queue(&kiblnd_data.kib_failover_waitq, &wait);
3502                 write_unlock_irqrestore(glock, flags);
3503
3504                 rc = schedule_timeout(long_sleep ? cfs_time_seconds(10) :
3505                                                    cfs_time_seconds(1));
3506                 set_current_state(TASK_RUNNING);
3507                 remove_wait_queue(&kiblnd_data.kib_failover_waitq, &wait);
3508                 write_lock_irqsave(glock, flags);
3509
3510                 if (!long_sleep || rc != 0)
3511                         continue;
3512
3513                 /* have a long sleep, routine check all active devices,
3514                  * we need checking like this because if there is not active
3515                  * connection on the dev and no SEND from local, we may listen
3516                  * on wrong HCA for ever while there is a bonding failover */
3517                 list_for_each_entry(dev, &kiblnd_data.kib_devs, ibd_list) {
3518                         if (kiblnd_dev_can_failover(dev)) {
3519                                 list_add_tail(&dev->ibd_fail_list,
3520                                               &kiblnd_data.kib_failed_devs);
3521                         }
3522                 }
3523         }
3524
3525         write_unlock_irqrestore(glock, flags);
3526
3527         kiblnd_thread_fini();
3528         return 0;
3529 }