lustre/ptlrpc/events.c

   1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
   2  * vim:expandtab:shiftwidth=8:tabstop=8:
   3  *
   4  * GPL HEADER START
   5  *
   6  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License version 2 only,
  10  * as published by the Free Software Foundation.
  11  *
  12  * This program is distributed in the hope that it will be useful, but
  13  * WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * General Public License version 2 for more details (a copy is included
  16  * in the LICENSE file that accompanied this code).
  17  *
  18  * You should have received a copy of the GNU General Public License
  19  * version 2 along with this program; If not, see
  20  * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
  21  *
  22  * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
  23  * CA 95054 USA or visit www.sun.com if you need additional information or
  24  * have any questions.
  25  *
  26  * GPL HEADER END
  27  */
  28 /*
  29  * Copyright  2008 Sun Microsystems, Inc. All rights reserved
  30  * Use is subject to license terms.
  31  */
  32 /*
  33  * This file is part of Lustre, http://www.lustre.org/
  34  * Lustre is a trademark of Sun Microsystems, Inc.
  35  */
  36
  37 #define DEBUG_SUBSYSTEM S_RPC
  38
  39 #ifndef __KERNEL__
  40 # include <liblustre.h>
  41 #else
  42 # include <libcfs/libcfs.h>
  43 # ifdef __mips64__
  44 #  include <linux/kernel.h>
  45 # endif
  46 #endif
  47
  48 #include <obd_class.h>
  49 #include <lustre_net.h>
  50 #include <lustre_sec.h>
  51 #include "ptlrpc_internal.h"
  52
  53 lnet_handle_eq_t   ptlrpc_eq_h;
  54
  55 /*
  56  *  Client's outgoing request callback
  57  */
  58 void request_out_callback(lnet_event_t *ev)
  59 {
  60         struct ptlrpc_cb_id   *cbid = ev->md.user_ptr;
  61         struct ptlrpc_request *req = cbid->cbid_arg;
  62         ENTRY;
  63
  64         LASSERT (ev->type == LNET_EVENT_SEND ||
  65                  ev->type == LNET_EVENT_UNLINK);
  66         LASSERT (ev->unlinked);
  67
  68         DEBUG_REQ((ev->status == 0) ? D_NET : D_ERROR, req,
  69                   "type %d, status %d", ev->type, ev->status);
  70
  71         sptlrpc_request_out_callback(req);
  72
  73         if (ev->type == LNET_EVENT_UNLINK || ev->status != 0) {
  74
  75                 /* Failed send: make it seem like the reply timed out, just
  76                  * like failing sends in client.c does currently...  */
  77
  78                 spin_lock(&req->rq_lock);
  79                 req->rq_net_err = 1;
  80                 spin_unlock(&req->rq_lock);
  81
  82                 ptlrpc_client_wake_req(req);
  83         }
  84
  85         ptlrpc_req_finished(req);
  86
  87         EXIT;
  88 }
  89
  90 /*
  91  * Client's incoming reply callback
  92  */
  93 void reply_in_callback(lnet_event_t *ev)
  94 {
  95         struct ptlrpc_cb_id   *cbid = ev->md.user_ptr;
  96         struct ptlrpc_request *req = cbid->cbid_arg;
  97         ENTRY;
  98
  99         DEBUG_REQ((ev->status == 0) ? D_NET : D_ERROR, req,
 100                   "type %d, status %d", ev->type, ev->status);
 101
 102         LASSERT (ev->type == LNET_EVENT_PUT || ev->type == LNET_EVENT_UNLINK);
 103         LASSERT (ev->md.start == req->rq_repbuf);
 104         LASSERT (ev->mlength <= req->rq_repbuf_len);
 105         /* We've set LNET_MD_MANAGE_REMOTE for all outgoing requests
 106            for adaptive timeouts' early reply. */
 107         LASSERT((ev->md.options & LNET_MD_MANAGE_REMOTE) != 0);
 108
 109         spin_lock(&req->rq_lock);
 110
 111         req->rq_receiving_reply = 0;
 112         req->rq_early = 0;
 113         if (ev->unlinked)
 114                 req->rq_must_unlink = 0;
 115
 116         if (ev->status)
 117                 goto out_wake;
 118         if (ev->type == LNET_EVENT_UNLINK) {
 119                 LASSERT(ev->unlinked);
 120                 DEBUG_REQ(D_RPCTRACE, req, "unlink");
 121                 goto out_wake;
 122         }
 123
 124         if ((ev->offset == 0) &&
 125             ((lustre_msghdr_get_flags(req->rq_reqmsg) & MSGHDR_AT_SUPPORT))) {
 126                 /* Early reply */
 127                 DEBUG_REQ(D_ADAPTTO, req,
 128                           "Early reply received: mlen=%u offset=%d replen=%d "
 129                           "replied=%d unlinked=%d", ev->mlength, ev->offset,
 130                           req->rq_replen, req->rq_replied, ev->unlinked);
 131
 132                 req->rq_early_count++; /* number received, client side */
 133
 134                 if (req->rq_replied)   /* already got the real reply */
 135                         goto out_wake;
 136
 137                 req->rq_early = 1;
 138                 req->rq_reply_off = ev->offset;
 139                 req->rq_nob_received = ev->mlength;
 140                 /* And we're still receiving */
 141                 req->rq_receiving_reply = 1;
 142         } else {
 143                 /* Real reply */
 144                 req->rq_replied = 1;
 145                 req->rq_reply_off = ev->offset;
 146                 req->rq_nob_received = ev->mlength;
 147                 /* LNetMDUnlink can't be called under the LNET_LOCK,
 148                    so we must unlink in ptlrpc_unregister_reply */
 149                 DEBUG_REQ(D_INFO, req,
 150                           "reply in flags=%x mlen=%u offset=%d replen=%d",
 151                           lustre_msg_get_flags(req->rq_reqmsg),
 152                           ev->mlength, ev->offset, req->rq_replen);
 153         }
 154
 155         req->rq_import->imp_last_reply_time = cfs_time_current_sec();
 156
 157 out_wake:
 158         /* NB don't unlock till after wakeup; req can disappear under us
 159          * since we don't have our own ref */
 160         ptlrpc_client_wake_req(req);
 161         spin_unlock(&req->rq_lock);
 162         EXIT;
 163 }
 164
 165 /*
 166  * Client's bulk has been written/read
 167  */
 168 void client_bulk_callback (lnet_event_t *ev)
 169 {
 170         struct ptlrpc_cb_id     *cbid = ev->md.user_ptr;
 171         struct ptlrpc_bulk_desc *desc = cbid->cbid_arg;
 172         ENTRY;
 173
 174         LASSERT ((desc->bd_type == BULK_PUT_SINK &&
 175                   ev->type == LNET_EVENT_PUT) ||
 176                  (desc->bd_type == BULK_GET_SOURCE &&
 177                   ev->type == LNET_EVENT_GET) ||
 178                  ev->type == LNET_EVENT_UNLINK);
 179         LASSERT (ev->unlinked);
 180
 181         CDEBUG((ev->status == 0) ? D_NET : D_ERROR,
 182                "event type %d, status %d, desc %p\n",
 183                ev->type, ev->status, desc);
 184
 185         spin_lock(&desc->bd_lock);
 186
 187         LASSERT(desc->bd_network_rw);
 188         desc->bd_network_rw = 0;
 189
 190         if (ev->type != LNET_EVENT_UNLINK && ev->status == 0) {
 191                 desc->bd_success = 1;
 192                 desc->bd_nob_transferred = ev->mlength;
 193                 desc->bd_sender = ev->sender;
 194         }
 195
 196         sptlrpc_enc_pool_put_pages(desc);
 197
 198         /* NB don't unlock till after wakeup; desc can disappear under us
 199          * otherwise */
 200         ptlrpc_client_wake_req(desc->bd_req);
 201
 202         spin_unlock(&desc->bd_lock);
 203         EXIT;
 204 }
 205
 206 /*
 207  * Server's incoming request callback
 208  */
 209 void request_in_callback(lnet_event_t *ev)
 210 {
 211         struct ptlrpc_cb_id               *cbid = ev->md.user_ptr;
 212         struct ptlrpc_request_buffer_desc *rqbd = cbid->cbid_arg;
 213         struct ptlrpc_service             *service = rqbd->rqbd_service;
 214         struct ptlrpc_request             *req;
 215         ENTRY;
 216
 217         LASSERT (ev->type == LNET_EVENT_PUT ||
 218                  ev->type == LNET_EVENT_UNLINK);
 219         LASSERT ((char *)ev->md.start >= rqbd->rqbd_buffer);
 220         LASSERT ((char *)ev->md.start + ev->offset + ev->mlength <=
 221                  rqbd->rqbd_buffer + service->srv_buf_size);
 222
 223         CDEBUG((ev->status == 0) ? D_NET : D_ERROR,
 224                "event type %d, status %d, service %s\n",
 225                ev->type, ev->status, service->srv_name);
 226
 227         if (ev->unlinked) {
 228                 /* If this is the last request message to fit in the
 229                  * request buffer we can use the request object embedded in
 230                  * rqbd.  Note that if we failed to allocate a request,
 231                  * we'd have to re-post the rqbd, which we can't do in this
 232                  * context. */
 233                 req = &rqbd->rqbd_req;
 234                 memset(req, 0, sizeof (*req));
 235         } else {
 236                 LASSERT (ev->type == LNET_EVENT_PUT);
 237                 if (ev->status != 0) {
 238                         /* We moaned above already... */
 239                         return;
 240                 }
 241                 OBD_ALLOC_GFP(req, sizeof(*req), CFS_ALLOC_ATOMIC_TRY);
 242                 if (req == NULL) {
 243                         CERROR("Can't allocate incoming request descriptor: "
 244                                "Dropping %s RPC from %s\n",
 245                                service->srv_name,
 246                                libcfs_id2str(ev->initiator));
 247                         return;
 248                 }
 249         }
 250
 251         /* NB we ABSOLUTELY RELY on req being zeroed, so pointers are NULL,
 252          * flags are reset and scalars are zero.  We only set the message
 253          * size to non-zero if this was a successful receive. */
 254         req->rq_xid = ev->match_bits;
 255         req->rq_reqbuf = ev->md.start + ev->offset;
 256         if (ev->type == LNET_EVENT_PUT && ev->status == 0)
 257                 req->rq_reqdata_len = ev->mlength;
 258         do_gettimeofday(&req->rq_arrival_time);
 259         req->rq_peer = ev->initiator;
 260         req->rq_self = ev->target.nid;
 261         req->rq_rqbd = rqbd;
 262         req->rq_phase = RQ_PHASE_NEW;
 263 #ifdef CRAY_XT3
 264         req->rq_uid = ev->uid;
 265 #endif
 266         spin_lock_init(&req->rq_lock);
 267         CFS_INIT_LIST_HEAD(&req->rq_timed_list);
 268         atomic_set(&req->rq_refcount, 1);
 269         if (ev->type == LNET_EVENT_PUT)
 270                 DEBUG_REQ(D_RPCTRACE, req, "incoming req");
 271
 272         CDEBUG(D_RPCTRACE, "peer: %s\n", libcfs_id2str(req->rq_peer));
 273
 274         spin_lock(&service->srv_lock);
 275
 276         req->rq_history_seq = service->srv_request_seq++;
 277         list_add_tail(&req->rq_history_list, &service->srv_request_history);
 278
 279         if (ev->unlinked) {
 280                 service->srv_nrqbd_receiving--;
 281                 CDEBUG(D_INFO, "Buffer complete: %d buffers still posted\n",
 282                        service->srv_nrqbd_receiving);
 283
 284                 /* Normally, don't complain about 0 buffers posted; LNET won't
 285                  * drop incoming reqs since we set the portal lazy */
 286                 if (test_req_buffer_pressure &&
 287                     ev->type != LNET_EVENT_UNLINK &&
 288                     service->srv_nrqbd_receiving == 0)
 289                         CWARN("All %s request buffers busy\n",
 290                               service->srv_name);
 291
 292                 /* req takes over the network's ref on rqbd */
 293         } else {
 294                 /* req takes a ref on rqbd */
 295                 rqbd->rqbd_refcount++;
 296         }
 297
 298         list_add_tail(&req->rq_list, &service->srv_req_in_queue);
 299         service->srv_n_queued_reqs++;
 300
 301         /* NB everything can disappear under us once the request
 302          * has been queued and we unlock, so do the wake now... */
 303         cfs_waitq_signal(&service->srv_waitq);
 304
 305         spin_unlock(&service->srv_lock);
 306         EXIT;
 307 }
 308
 309 /*
 310  *  Server's outgoing reply callback
 311  */
 312 void reply_out_callback(lnet_event_t *ev)
 313 {
 314         struct ptlrpc_cb_id       *cbid = ev->md.user_ptr;
 315         struct ptlrpc_reply_state *rs = cbid->cbid_arg;
 316         struct ptlrpc_service     *svc = rs->rs_service;
 317         ENTRY;
 318
 319         LASSERT (ev->type == LNET_EVENT_SEND ||
 320                  ev->type == LNET_EVENT_ACK ||
 321                  ev->type == LNET_EVENT_UNLINK);
 322
 323         if (!rs->rs_difficult) {
 324                 /* 'Easy' replies have no further processing so I drop the
 325                  * net's ref on 'rs' */
 326                 LASSERT (ev->unlinked);
 327                 ptlrpc_rs_decref(rs);
 328                 atomic_dec (&svc->srv_outstanding_replies);
 329                 EXIT;
 330                 return;
 331         }
 332
 333         LASSERT (rs->rs_on_net);
 334
 335         if (ev->unlinked) {
 336                 /* Last network callback.  The net's ref on 'rs' stays put
 337                  * until ptlrpc_server_handle_reply() is done with it */
 338                 spin_lock(&svc->srv_lock);
 339                 spin_lock(&rs->rs_lock);
 340                 rs->rs_on_net = 0;
 341                 if (!rs->rs_no_ack ||
 342                     rs->rs_transno <= rs->rs_export->exp_obd->obd_last_committed)
 343                         ptlrpc_schedule_difficult_reply (rs);
 344                 spin_unlock(&rs->rs_lock);
 345                 spin_unlock(&svc->srv_lock);
 346         }
 347
 348         EXIT;
 349 }
 350
 351 /*
 352  * Server's bulk completion callback
 353  */
 354 void server_bulk_callback (lnet_event_t *ev)
 355 {
 356         struct ptlrpc_cb_id     *cbid = ev->md.user_ptr;
 357         struct ptlrpc_bulk_desc *desc = cbid->cbid_arg;
 358         ENTRY;
 359
 360         LASSERT (ev->type == LNET_EVENT_SEND ||
 361                  ev->type == LNET_EVENT_UNLINK ||
 362                  (desc->bd_type == BULK_PUT_SOURCE &&
 363                   ev->type == LNET_EVENT_ACK) ||
 364                  (desc->bd_type == BULK_GET_SINK &&
 365                   ev->type == LNET_EVENT_REPLY));
 366
 367         CDEBUG((ev->status == 0) ? D_NET : D_ERROR,
 368                "event type %d, status %d, desc %p\n",
 369                ev->type, ev->status, desc);
 370
 371         spin_lock(&desc->bd_lock);
 372
 373         if ((ev->type == LNET_EVENT_ACK ||
 374              ev->type == LNET_EVENT_REPLY) &&
 375             ev->status == 0) {
 376                 /* We heard back from the peer, so even if we get this
 377                  * before the SENT event (oh yes we can), we know we
 378                  * read/wrote the peer buffer and how much... */
 379                 desc->bd_success = 1;
 380                 desc->bd_nob_transferred = ev->mlength;
 381                 desc->bd_sender = ev->sender;
 382         }
 383
 384         if (ev->unlinked) {
 385                 /* This is the last callback no matter what... */
 386                 desc->bd_network_rw = 0;
 387                 cfs_waitq_signal(&desc->bd_waitq);
 388         }
 389
 390         spin_unlock(&desc->bd_lock);
 391         EXIT;
 392 }
 393
 394 static void ptlrpc_master_callback(lnet_event_t *ev)
 395 {
 396         struct ptlrpc_cb_id *cbid = ev->md.user_ptr;
 397         void (*callback)(lnet_event_t *ev) = cbid->cbid_fn;
 398
 399         /* Honestly, it's best to find out early. */
 400         LASSERT (cbid->cbid_arg != LP_POISON);
 401         LASSERT (callback == request_out_callback ||
 402                  callback == reply_in_callback ||
 403                  callback == client_bulk_callback ||
 404                  callback == request_in_callback ||
 405                  callback == reply_out_callback ||
 406                  callback == server_bulk_callback);
 407
 408         callback (ev);
 409 }
 410
 411 int ptlrpc_uuid_to_peer (struct obd_uuid *uuid,
 412                          lnet_process_id_t *peer, lnet_nid_t *self)
 413 {
 414         int               best_dist = 0;
 415         __u32             best_order = 0;
 416         int               count = 0;
 417         int               rc = -ENOENT;
 418         int               portals_compatibility;
 419         int               dist;
 420         __u32             order;
 421         lnet_nid_t        dst_nid;
 422         lnet_nid_t        src_nid;
 423
 424         portals_compatibility = LNetCtl(IOC_LIBCFS_PORTALS_COMPATIBILITY, NULL);
 425
 426         peer->pid = LUSTRE_SRV_LNET_PID;
 427
 428         /* Choose the matching UUID that's closest */
 429         while (lustre_uuid_to_peer(uuid->uuid, &dst_nid, count++) == 0) {
 430                 dist = LNetDist(dst_nid, &src_nid, &order);
 431                 if (dist < 0)
 432                         continue;
 433
 434                 if (dist == 0) {                /* local! use loopback LND */
 435                         peer->nid = *self = LNET_MKNID(LNET_MKNET(LOLND, 0), 0);
 436                         rc = 0;
 437                         break;
 438                 }
 439
 440                 if (rc < 0 ||
 441                     dist < best_dist ||
 442                     (dist == best_dist && order < best_order)) {
 443                         best_dist = dist;
 444                         best_order = order;
 445
 446                         if (portals_compatibility > 1) {
 447                                 /* Strong portals compatibility: Zero the nid's
 448                                  * NET, so if I'm reading new config logs, or
 449                                  * getting configured by (new) lconf I can
 450                                  * still talk to old servers. */
 451                                 dst_nid = LNET_MKNID(0, LNET_NIDADDR(dst_nid));
 452                                 src_nid = LNET_MKNID(0, LNET_NIDADDR(src_nid));
 453                         }
 454                         peer->nid = dst_nid;
 455                         *self = src_nid;
 456                         rc = 0;
 457                 }
 458         }
 459
 460         CDEBUG(D_NET,"%s->%s\n", uuid->uuid, libcfs_id2str(*peer));
 461         if (rc != 0)
 462                 CERROR("No NID found for %s\n", uuid->uuid);
 463         return rc;
 464 }
 465
 466 void ptlrpc_ni_fini(void)
 467 {
 468         cfs_waitq_t         waitq;
 469         struct l_wait_info  lwi;
 470         int                 rc;
 471         int                 retries;
 472
 473         /* Wait for the event queue to become idle since there may still be
 474          * messages in flight with pending events (i.e. the fire-and-forget
 475          * messages == client requests and "non-difficult" server
 476          * replies */
 477
 478         for (retries = 0;; retries++) {
 479                 rc = LNetEQFree(ptlrpc_eq_h);
 480                 switch (rc) {
 481                 default:
 482                         LBUG();
 483
 484                 case 0:
 485                         LNetNIFini();
 486                         return;
 487
 488                 case -EBUSY:
 489                         if (retries != 0)
 490                                 CWARN("Event queue still busy\n");
 491
 492                         /* Wait for a bit */
 493                         cfs_waitq_init(&waitq);
 494                         lwi = LWI_TIMEOUT(cfs_time_seconds(2), NULL, NULL);
 495                         l_wait_event(waitq, 0, &lwi);
 496                         break;
 497                 }
 498         }
 499         /* notreached */
 500 }
 501
 502 lnet_pid_t ptl_get_pid(void)
 503 {
 504         lnet_pid_t        pid;
 505
 506 #ifndef  __KERNEL__
 507         pid = getpid();
 508 #else
 509         pid = LUSTRE_SRV_LNET_PID;
 510 #endif
 511         return pid;
 512 }
 513
 514 int ptlrpc_ni_init(void)
 515 {
 516         int              rc;
 517         lnet_pid_t       pid;
 518
 519         pid = ptl_get_pid();
 520         CDEBUG(D_NET, "My pid is: %x\n", pid);
 521
 522         /* We're not passing any limits yet... */
 523         rc = LNetNIInit(pid);
 524         if (rc < 0) {
 525                 CDEBUG (D_NET, "Can't init network interface: %d\n", rc);
 526                 return (-ENOENT);
 527         }
 528
 529         /* CAVEAT EMPTOR: how we process portals events is _radically_
 530          * different depending on... */
 531 #ifdef __KERNEL__
 532         /* kernel portals calls our master callback when events are added to
 533          * the event queue.  In fact lustre never pulls events off this queue,
 534          * so it's only sized for some debug history. */
 535         rc = LNetEQAlloc(1024, ptlrpc_master_callback, &ptlrpc_eq_h);
 536 #else
 537         /* liblustre calls the master callback when it removes events from the
 538          * event queue.  The event queue has to be big enough not to drop
 539          * anything */
 540         rc = LNetEQAlloc(10240, LNET_EQ_HANDLER_NONE, &ptlrpc_eq_h);
 541 #endif
 542         if (rc == 0)
 543                 return 0;
 544
 545         CERROR ("Failed to allocate event queue: %d\n", rc);
 546         LNetNIFini();
 547
 548         return (-ENOMEM);
 549 }
 550
 551 #ifndef __KERNEL__
 552 CFS_LIST_HEAD(liblustre_wait_callbacks);
 553 CFS_LIST_HEAD(liblustre_idle_callbacks);
 554 void *liblustre_services_callback;
 555
 556 void *
 557 liblustre_register_waitidle_callback (struct list_head *callback_list,
 558                                       const char *name,
 559                                       int (*fn)(void *arg), void *arg)
 560 {
 561         struct liblustre_wait_callback *llwc;
 562
 563         OBD_ALLOC(llwc, sizeof(*llwc));
 564         LASSERT (llwc != NULL);
 565
 566         llwc->llwc_name = name;
 567         llwc->llwc_fn = fn;
 568         llwc->llwc_arg = arg;
 569         list_add_tail(&llwc->llwc_list, callback_list);
 570
 571         return (llwc);
 572 }
 573
 574 void
 575 liblustre_deregister_waitidle_callback (void *opaque)
 576 {
 577         struct liblustre_wait_callback *llwc = opaque;
 578
 579         list_del(&llwc->llwc_list);
 580         OBD_FREE(llwc, sizeof(*llwc));
 581 }
 582
 583 void *
 584 liblustre_register_wait_callback (const char *name,
 585                                   int (*fn)(void *arg), void *arg)
 586 {
 587         return liblustre_register_waitidle_callback(&liblustre_wait_callbacks,
 588                                                     name, fn, arg);
 589 }
 590
 591 void
 592 liblustre_deregister_wait_callback (void *opaque)
 593 {
 594         liblustre_deregister_waitidle_callback(opaque);
 595 }
 596
 597 void *
 598 liblustre_register_idle_callback (const char *name,
 599                                   int (*fn)(void *arg), void *arg)
 600 {
 601         return liblustre_register_waitidle_callback(&liblustre_idle_callbacks,
 602                                                     name, fn, arg);
 603 }
 604
 605 void
 606 liblustre_deregister_idle_callback (void *opaque)
 607 {
 608         liblustre_deregister_waitidle_callback(opaque);
 609 }
 610
 611 int
 612 liblustre_check_events (int timeout)
 613 {
 614         lnet_event_t ev;
 615         int         rc;
 616         int         i;
 617         ENTRY;
 618
 619         rc = LNetEQPoll(&ptlrpc_eq_h, 1, timeout * 1000, &ev, &i);
 620         if (rc == 0)
 621                 RETURN(0);
 622
 623         LASSERT (rc == -EOVERFLOW || rc == 1);
 624
 625         /* liblustre: no asynch callback so we can't affort to miss any
 626          * events... */
 627         if (rc == -EOVERFLOW) {
 628                 CERROR ("Dropped an event!!!\n");
 629                 abort();
 630         }
 631
 632         ptlrpc_master_callback (&ev);
 633         RETURN(1);
 634 }
 635
 636 int liblustre_waiting = 0;
 637
 638 int
 639 liblustre_wait_event (int timeout)
 640 {
 641         struct list_head               *tmp;
 642         struct liblustre_wait_callback *llwc;
 643         int                             found_something = 0;
 644
 645         /* single threaded recursion check... */
 646         liblustre_waiting = 1;
 647
 648         for (;;) {
 649                 /* Deal with all pending events */
 650                 while (liblustre_check_events(0))
 651                         found_something = 1;
 652
 653                 /* Give all registered callbacks a bite at the cherry */
 654                 list_for_each(tmp, &liblustre_wait_callbacks) {
 655                         llwc = list_entry(tmp, struct liblustre_wait_callback,
 656                                           llwc_list);
 657
 658                         if (llwc->llwc_fn(llwc->llwc_arg))
 659                                 found_something = 1;
 660                 }
 661
 662                 if (found_something || timeout == 0)
 663                         break;
 664
 665                 /* Nothing so far, but I'm allowed to block... */
 666                 found_something = liblustre_check_events(timeout);
 667                 if (!found_something)           /* still nothing */
 668                         break;                  /* I timed out */
 669         }
 670
 671         liblustre_waiting = 0;
 672
 673         return found_something;
 674 }
 675
 676 void
 677 liblustre_wait_idle(void)
 678 {
 679         static int recursed = 0;
 680
 681         struct list_head               *tmp;
 682         struct liblustre_wait_callback *llwc;
 683         int                             idle = 0;
 684
 685         LASSERT(!recursed);
 686         recursed = 1;
 687
 688         do {
 689                 liblustre_wait_event(0);
 690
 691                 idle = 1;
 692
 693                 list_for_each(tmp, &liblustre_idle_callbacks) {
 694                         llwc = list_entry(tmp, struct liblustre_wait_callback,
 695                                           llwc_list);
 696
 697                         if (!llwc->llwc_fn(llwc->llwc_arg)) {
 698                                 idle = 0;
 699                                 break;
 700                         }
 701                 }
 702
 703         } while (!idle);
 704
 705         recursed = 0;
 706 }
 707
 708 #endif /* __KERNEL__ */
 709
 710 int ptlrpc_init_portals(void)
 711 {
 712         int   rc = ptlrpc_ni_init();
 713
 714         if (rc != 0) {
 715                 CERROR("network initialisation failed\n");
 716                 return -EIO;
 717         }
 718 #ifndef __KERNEL__
 719         liblustre_services_callback =
 720                 liblustre_register_wait_callback("liblustre_check_services",
 721                                                  &liblustre_check_services,
 722                                                  NULL);
 723         init_completion_module(liblustre_wait_event);
 724 #endif
 725         rc = ptlrpcd_addref();
 726         if (rc == 0)
 727                 return 0;
 728
 729         CERROR("rpcd initialisation failed\n");
 730 #ifndef __KERNEL__
 731         liblustre_deregister_wait_callback(liblustre_services_callback);
 732 #endif
 733         ptlrpc_ni_fini();
 734         return rc;
 735 }
 736
 737 void ptlrpc_exit_portals(void)
 738 {
 739 #ifndef __KERNEL__
 740         liblustre_deregister_wait_callback(liblustre_services_callback);
 741 #endif
 742         ptlrpcd_decref();
 743         ptlrpc_ni_fini();
 744 }