lnet/selftest/rpc.c

   1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
   2  * vim:expandtab:shiftwidth=8:tabstop=8:
   3  *
   4  * Copyright (C) 2001, 2002 Cluster File Systems, Inc.
   5  *   Author: Isaac Huang <isaac@clusterfs.com>
   6  *
   7  */
   8
   9 #define DEBUG_SUBSYSTEM S_LNET
  10
  11 #include <libcfs/kp30.h>
  12 #include <libcfs/libcfs.h>
  13 #include <lnet/lib-lnet.h>
  14
  15 #include "selftest.h"
  16
  17
  18 typedef enum {
  19         SRPC_STATE_NONE,
  20         SRPC_STATE_NI_INIT,
  21         SRPC_STATE_EQ_INIT,
  22         SRPC_STATE_WI_INIT,
  23         SRPC_STATE_RUNNING,
  24         SRPC_STATE_STOPPING,
  25 } srpc_state_t;
  26
  27 #define SRPC_PEER_HASH_SIZE       101  /* # peer lists */
  28 #define SRPC_PEER_CREDITS         16   /* >= most LND's default peer credit */
  29
  30 struct smoketest_rpc {
  31         spinlock_t        rpc_glock;     /* global lock */
  32         srpc_service_t   *rpc_services[SRPC_SERVICE_MAX_ID + 1];
  33         struct list_head *rpc_peers;     /* hash table of known peers */
  34         lnet_handle_eq_t  rpc_lnet_eq;   /* _the_ LNet event queue */
  35         srpc_state_t      rpc_state;
  36         srpc_counters_t   rpc_counters;
  37         __u64             rpc_matchbits; /* matchbits counter */
  38 } srpc_data;
  39
  40 /* forward ref's */
  41 int srpc_handle_rpc (swi_workitem_t *wi);
  42
  43 void srpc_get_counters (srpc_counters_t *cnt)
  44 {
  45         spin_lock(&srpc_data.rpc_glock);
  46         *cnt = srpc_data.rpc_counters;
  47         spin_unlock(&srpc_data.rpc_glock);
  48 }
  49
  50 void srpc_set_counters (const srpc_counters_t *cnt)
  51 {
  52         spin_lock(&srpc_data.rpc_glock);
  53         srpc_data.rpc_counters = *cnt;
  54         spin_unlock(&srpc_data.rpc_glock);
  55 }
  56
  57 void
  58 srpc_add_bulk_page (srpc_bulk_t *bk, cfs_page_t *pg, int i)
  59 {
  60         LASSERT (i >= 0 && i < bk->bk_niov);
  61
  62 #ifdef __KERNEL__
  63         bk->bk_iovs[i].kiov_offset = 0;
  64         bk->bk_iovs[i].kiov_page   = pg;
  65         bk->bk_iovs[i].kiov_len    = CFS_PAGE_SIZE;
  66 #else
  67         LASSERT (bk->bk_pages != NULL);
  68
  69         bk->bk_pages[i] = pg;
  70         bk->bk_iovs[i].iov_len  = CFS_PAGE_SIZE;
  71         bk->bk_iovs[i].iov_base = cfs_page_address(pg);
  72 #endif
  73         return;
  74 }
  75
  76 void
  77 srpc_free_bulk (srpc_bulk_t *bk)
  78 {
  79         int         i;
  80         cfs_page_t *pg;
  81
  82         LASSERT (bk != NULL);
  83 #ifndef __KERNEL__
  84         LASSERT (bk->bk_pages != NULL);
  85 #endif
  86
  87         for (i = 0; i < bk->bk_niov; i++) {
  88 #ifdef __KERNEL__
  89                 pg = bk->bk_iovs[i].kiov_page;
  90 #else
  91                 pg = bk->bk_pages[i];
  92 #endif
  93                 if (pg == NULL) break;
  94
  95                 cfs_free_page(pg);
  96         }
  97
  98 #ifndef __KERNEL__
  99         LIBCFS_FREE(bk->bk_pages, sizeof(cfs_page_t *) * bk->bk_niov);
 100 #endif
 101         LIBCFS_FREE(bk, offsetof(srpc_bulk_t, bk_iovs[bk->bk_niov]));
 102         return;
 103 }
 104
 105 srpc_bulk_t *
 106 srpc_alloc_bulk (int npages, int sink)
 107 {
 108         srpc_bulk_t  *bk;
 109         cfs_page_t  **pages;
 110         int           i;
 111
 112         LASSERT (npages > 0 && npages <= LNET_MAX_IOV);
 113
 114         LIBCFS_ALLOC(bk, offsetof(srpc_bulk_t, bk_iovs[npages]));
 115         if (bk == NULL) {
 116                 CERROR ("Can't allocate descriptor for %d pages\n", npages);
 117                 return NULL;
 118         }
 119
 120         memset(bk, 0, offsetof(srpc_bulk_t, bk_iovs[npages]));
 121         bk->bk_sink = sink;
 122         bk->bk_niov = npages;
 123         bk->bk_len  = npages * CFS_PAGE_SIZE;
 124 #ifndef __KERNEL__
 125         LIBCFS_ALLOC(pages, sizeof(cfs_page_t *) * npages);
 126         if (pages == NULL) {
 127                 LIBCFS_FREE(bk, offsetof(srpc_bulk_t, bk_iovs[npages]));
 128                 CERROR ("Can't allocate page array for %d pages\n", npages);
 129                 return NULL;
 130         }
 131
 132         memset(pages, 0, sizeof(cfs_page_t *) * npages);
 133         bk->bk_pages = pages;
 134 #else
 135         UNUSED (pages);
 136 #endif
 137
 138         for (i = 0; i < npages; i++) {
 139                 cfs_page_t *pg = cfs_alloc_page(CFS_ALLOC_STD);
 140
 141                 if (pg == NULL) {
 142                         CERROR ("Can't allocate page %d of %d\n", i, npages);
 143                         srpc_free_bulk(bk);
 144                         return NULL;
 145                 }
 146
 147                 srpc_add_bulk_page(bk, pg, i);
 148         }
 149
 150         return bk;
 151 }
 152
 153
 154 static inline struct list_head *
 155 srpc_nid2peerlist (lnet_nid_t nid)
 156 {
 157         unsigned int hash = ((unsigned int)nid) % SRPC_PEER_HASH_SIZE;
 158
 159         return &srpc_data.rpc_peers[hash];
 160 }
 161
 162 static inline srpc_peer_t *
 163 srpc_create_peer (lnet_nid_t nid)
 164 {
 165         srpc_peer_t *peer;
 166
 167         LASSERT (nid != LNET_NID_ANY);
 168
 169         LIBCFS_ALLOC(peer, sizeof(srpc_peer_t));
 170         if (peer == NULL) {
 171                 CERROR ("Failed to allocate peer structure for %s\n",
 172                         libcfs_nid2str(nid));
 173                 return NULL;
 174         }
 175
 176         memset(peer, 0, sizeof(srpc_peer_t));
 177         peer->stp_nid     = nid;
 178         peer->stp_credits = SRPC_PEER_CREDITS;
 179
 180         spin_lock_init(&peer->stp_lock);
 181         CFS_INIT_LIST_HEAD(&peer->stp_rpcq);
 182         CFS_INIT_LIST_HEAD(&peer->stp_ctl_rpcq);
 183         return peer;
 184 }
 185
 186 srpc_peer_t *
 187 srpc_find_peer_locked (lnet_nid_t nid)
 188 {
 189         struct list_head *peer_list = srpc_nid2peerlist(nid);
 190         srpc_peer_t      *peer;
 191
 192         LASSERT (nid != LNET_NID_ANY);
 193
 194         list_for_each_entry (peer, peer_list, stp_list) {
 195                 if (peer->stp_nid == nid)
 196                         return peer;
 197         }
 198
 199         return NULL;
 200 }
 201
 202 static srpc_peer_t *
 203 srpc_nid2peer (lnet_nid_t nid)
 204 {
 205         srpc_peer_t *peer;
 206         srpc_peer_t *new_peer;
 207
 208         spin_lock(&srpc_data.rpc_glock);
 209         peer = srpc_find_peer_locked(nid);
 210         spin_unlock(&srpc_data.rpc_glock);
 211
 212         if (peer != NULL)
 213                 return peer;
 214
 215         new_peer = srpc_create_peer(nid);
 216
 217         spin_lock(&srpc_data.rpc_glock);
 218
 219         peer = srpc_find_peer_locked(nid);
 220         if (peer != NULL) {
 221                 spin_unlock(&srpc_data.rpc_glock);
 222                 if (new_peer != NULL)
 223                         LIBCFS_FREE(new_peer, sizeof(srpc_peer_t));
 224
 225                 return peer;
 226         }
 227
 228         if (new_peer == NULL) {
 229                 spin_unlock(&srpc_data.rpc_glock);
 230                 return NULL;
 231         }
 232
 233         list_add_tail(&new_peer->stp_list, srpc_nid2peerlist(nid));
 234         spin_unlock(&srpc_data.rpc_glock);
 235         return new_peer;
 236 }
 237
 238 static inline __u64
 239 srpc_next_id (void)
 240 {
 241         __u64 id;
 242
 243         spin_lock(&srpc_data.rpc_glock);
 244         id = srpc_data.rpc_matchbits++;
 245         spin_unlock(&srpc_data.rpc_glock);
 246         return id;
 247 }
 248
 249 void
 250 srpc_init_server_rpc (srpc_server_rpc_t *rpc,
 251                       srpc_service_t *sv, srpc_buffer_t *buffer)
 252 {
 253         memset(rpc, 0, sizeof(*rpc));
 254         swi_init_workitem(&rpc->srpc_wi, rpc, srpc_handle_rpc);
 255
 256         rpc->srpc_ev.ev_fired = 1; /* no event expected now */
 257
 258         rpc->srpc_service  = sv;
 259         rpc->srpc_reqstbuf = buffer;
 260         rpc->srpc_peer     = buffer->buf_peer;
 261         rpc->srpc_self     = buffer->buf_self;
 262         rpc->srpc_replymdh = LNET_INVALID_HANDLE;
 263 }
 264
 265 int
 266 srpc_add_service (srpc_service_t *sv)
 267 {
 268         int                id = sv->sv_id;
 269         int                i;
 270         srpc_server_rpc_t *rpc;
 271
 272         LASSERT (sv->sv_concur > 0);
 273         LASSERT (0 <= id && id <= SRPC_SERVICE_MAX_ID);
 274
 275         spin_lock(&srpc_data.rpc_glock);
 276
 277         LASSERT (srpc_data.rpc_state == SRPC_STATE_RUNNING);
 278
 279         if (srpc_data.rpc_services[id] != NULL) {
 280                 spin_unlock(&srpc_data.rpc_glock);
 281                 return -EBUSY;
 282         }
 283
 284         srpc_data.rpc_services[id] = sv;
 285         spin_unlock(&srpc_data.rpc_glock);
 286
 287         sv->sv_nprune       = 0;
 288         sv->sv_nposted_msg  = 0;
 289         sv->sv_shuttingdown = 0;
 290         spin_lock_init(&sv->sv_lock);
 291         CFS_INIT_LIST_HEAD(&sv->sv_free_rpcq);
 292         CFS_INIT_LIST_HEAD(&sv->sv_active_rpcq);
 293         CFS_INIT_LIST_HEAD(&sv->sv_posted_msgq);
 294         CFS_INIT_LIST_HEAD(&sv->sv_blocked_msgq);
 295
 296         sv->sv_ev.ev_data = sv;
 297         sv->sv_ev.ev_type = SRPC_REQUEST_RCVD;
 298
 299         for (i = 0; i < sv->sv_concur; i++) {
 300                 LIBCFS_ALLOC(rpc, sizeof(*rpc));
 301                 if (rpc == NULL) goto enomem;
 302
 303                 list_add(&rpc->srpc_list, &sv->sv_free_rpcq);
 304         }
 305
 306         CDEBUG (D_NET, "Adding service: id %d, name %s, concurrency %d\n",
 307                 id, sv->sv_name, sv->sv_concur);
 308         return 0;
 309
 310 enomem:
 311         while (!list_empty(&sv->sv_free_rpcq)) {
 312                 rpc = list_entry(sv->sv_free_rpcq.next,
 313                                  srpc_server_rpc_t, srpc_list);
 314                 list_del(&rpc->srpc_list);
 315                 LIBCFS_FREE(rpc, sizeof(*rpc));
 316         }
 317
 318         spin_lock(&srpc_data.rpc_glock);
 319         srpc_data.rpc_services[id] = NULL;
 320         spin_unlock(&srpc_data.rpc_glock);
 321         return -ENOMEM;
 322 }
 323
 324 int
 325 srpc_remove_service (srpc_service_t *sv)
 326 {
 327         int id = sv->sv_id;
 328
 329         spin_lock(&srpc_data.rpc_glock);
 330
 331         if (srpc_data.rpc_services[id] != sv) {
 332                 spin_unlock(&srpc_data.rpc_glock);
 333                 return -ENOENT;
 334         }
 335
 336         srpc_data.rpc_services[id] = NULL;
 337         spin_unlock(&srpc_data.rpc_glock);
 338         return 0;
 339 }
 340
 341 int
 342 srpc_post_passive_rdma(int portal, __u64 matchbits, void *buf,
 343                        int len, int options, lnet_process_id_t peer,
 344                        lnet_handle_md_t *mdh, srpc_event_t *ev)
 345 {
 346         int              rc;
 347         lnet_md_t        md;
 348         lnet_handle_me_t meh;
 349
 350         rc = LNetMEAttach(portal, peer, matchbits, 0,
 351                           LNET_UNLINK, LNET_INS_AFTER, &meh);
 352         if (rc != 0) {
 353                 CERROR ("LNetMEAttach failed: %d\n", rc);
 354                 LASSERT (rc == -ENOMEM);
 355                 return -ENOMEM;
 356         }
 357
 358         md.threshold = 1;
 359         md.user_ptr  = ev;
 360         md.start     = buf;
 361         md.length    = len;
 362         md.options   = options;
 363         md.eq_handle = srpc_data.rpc_lnet_eq;
 364
 365         rc = LNetMDAttach(meh, md, LNET_UNLINK, mdh);
 366         if (rc != 0) {
 367                 CERROR ("LNetMDAttach failed: %d\n", rc);
 368                 LASSERT (rc == -ENOMEM);
 369
 370                 rc = LNetMEUnlink(meh);
 371                 LASSERT (rc == 0);
 372                 return -ENOMEM;
 373         }
 374
 375         CDEBUG (D_NET,
 376                 "Posted passive RDMA: peer %s, portal %d, matchbits "LPX64"\n",
 377                 libcfs_id2str(peer), portal, matchbits);
 378         return 0;
 379 }
 380
 381 int
 382 srpc_post_active_rdma(int portal, __u64 matchbits, void *buf, int len,
 383                       int options, lnet_process_id_t peer, lnet_nid_t self,
 384                       lnet_handle_md_t *mdh, srpc_event_t *ev)
 385 {
 386         int       rc;
 387         lnet_md_t md;
 388
 389         md.user_ptr  = ev;
 390         md.start     = buf;
 391         md.length    = len;
 392         md.eq_handle = srpc_data.rpc_lnet_eq;
 393         md.threshold = ((options & LNET_MD_OP_GET) != 0) ? 2 : 1;
 394         md.options   = options & ~(LNET_MD_OP_PUT | LNET_MD_OP_GET);
 395
 396         rc = LNetMDBind(md, LNET_UNLINK, mdh);
 397         if (rc != 0) {
 398                 CERROR ("LNetMDBind failed: %d\n", rc);
 399                 LASSERT (rc == -ENOMEM);
 400                 return -ENOMEM;
 401         }
 402
 403         /* this is kind of an abuse of the LNET_MD_OP_{PUT,GET} options.
 404          * they're only meaningful for MDs attached to an ME (i.e. passive
 405          * buffers... */
 406         if ((options & LNET_MD_OP_PUT) != 0) {
 407                 rc = LNetPut(self, *mdh, LNET_NOACK_REQ, peer,
 408                              portal, matchbits, 0, 0);
 409         } else {
 410                 LASSERT ((options & LNET_MD_OP_GET) != 0);
 411
 412                 rc = LNetGet(self, *mdh, peer, portal, matchbits, 0);
 413         }
 414
 415         if (rc != 0) {
 416                 CERROR ("LNet%s(%s, %d, "LPD64") failed: %d\n",
 417                         ((options & LNET_MD_OP_PUT) != 0) ? "Put" : "Get",
 418                         libcfs_id2str(peer), portal, matchbits, rc);
 419
 420                 /* The forthcoming unlink event will complete this operation
 421                  * with failure, so fall through and return success here.
 422                  */
 423                 rc = LNetMDUnlink(*mdh);
 424                 LASSERT (rc == 0);
 425         } else {
 426                 CDEBUG (D_NET,
 427                         "Posted active RDMA: peer %s, portal %u, matchbits "LPX64"\n",
 428                         libcfs_id2str(peer), portal, matchbits);
 429         }
 430         return 0;
 431 }
 432
 433 int
 434 srpc_post_active_rqtbuf(lnet_process_id_t peer, int service, void *buf,
 435                         int len, lnet_handle_md_t *mdh, srpc_event_t *ev)
 436 {
 437         int rc;
 438         int portal;
 439
 440         if (service > SRPC_FRAMEWORK_SERVICE_MAX_ID)
 441                 portal = SRPC_REQUEST_PORTAL;
 442         else
 443                 portal = SRPC_FRAMEWORK_REQUEST_PORTAL;
 444
 445         rc = srpc_post_active_rdma(portal, service, buf, len,
 446                                    LNET_MD_OP_PUT, peer,
 447                                    LNET_NID_ANY, mdh, ev);
 448         return rc;
 449 }
 450
 451 int
 452 srpc_post_passive_rqtbuf(int service, void *buf, int len,
 453                          lnet_handle_md_t *mdh, srpc_event_t *ev)
 454 {
 455         int               rc;
 456         int               portal;
 457         lnet_process_id_t any = {.nid = LNET_NID_ANY,
 458                                  .pid = LNET_PID_ANY};
 459
 460         if (service > SRPC_FRAMEWORK_SERVICE_MAX_ID)
 461                 portal = SRPC_REQUEST_PORTAL;
 462         else
 463                 portal = SRPC_FRAMEWORK_REQUEST_PORTAL;
 464
 465         rc = srpc_post_passive_rdma(portal, service, buf, len,
 466                                     LNET_MD_OP_PUT, any, mdh, ev);
 467         return rc;
 468 }
 469
 470 int
 471 srpc_service_post_buffer (srpc_service_t *sv, srpc_buffer_t *buf)
 472 {
 473         srpc_msg_t *msg = &buf->buf_msg;
 474         int         rc;
 475
 476         LASSERT (!sv->sv_shuttingdown);
 477
 478         buf->buf_mdh = LNET_INVALID_HANDLE;
 479         list_add(&buf->buf_list, &sv->sv_posted_msgq);
 480         sv->sv_nposted_msg++;
 481         spin_unlock(&sv->sv_lock);
 482
 483         rc = srpc_post_passive_rqtbuf(sv->sv_id, msg, sizeof(*msg),
 484                                       &buf->buf_mdh, &sv->sv_ev);
 485
 486         /* At this point, a RPC (new or delayed) may have arrived in
 487          * msg and its event handler has been called. So we must add
 488          * buf to sv_posted_msgq _before_ dropping sv_lock */
 489
 490         spin_lock(&sv->sv_lock);
 491
 492         if (rc == 0) {
 493                 if (sv->sv_shuttingdown) {
 494                         spin_unlock(&sv->sv_lock);
 495
 496                         /* srpc_shutdown_service might have tried to unlink me
 497                          * when my buf_mdh was still invalid */
 498                         LNetMDUnlink(buf->buf_mdh);
 499
 500                         spin_lock(&sv->sv_lock);
 501                 }
 502                 return 0;
 503         }
 504
 505         sv->sv_nposted_msg--;
 506         if (sv->sv_shuttingdown) return rc;
 507
 508         list_del(&buf->buf_list);
 509
 510         spin_unlock(&sv->sv_lock);
 511         LIBCFS_FREE(buf, sizeof(*buf));
 512         spin_lock(&sv->sv_lock);
 513         return rc;
 514 }
 515
 516 int
 517 srpc_service_add_buffers (srpc_service_t *sv, int nbuffer)
 518 {
 519         int                rc;
 520         int                posted;
 521         srpc_buffer_t     *buf;
 522
 523         LASSERTF (nbuffer > 0,
 524                   "nbuffer must be positive: %d\n", nbuffer);
 525
 526         for (posted = 0; posted < nbuffer; posted++) {
 527                 LIBCFS_ALLOC(buf, sizeof(*buf));
 528                 if (buf == NULL) break;
 529
 530                 spin_lock(&sv->sv_lock);
 531                 rc = srpc_service_post_buffer(sv, buf);
 532                 spin_unlock(&sv->sv_lock);
 533
 534                 if (rc != 0) break;
 535         }
 536
 537         return posted;
 538 }
 539
 540 void
 541 srpc_service_remove_buffers (srpc_service_t *sv, int nbuffer)
 542 {
 543         LASSERTF (nbuffer > 0,
 544                   "nbuffer must be positive: %d\n", nbuffer);
 545
 546         spin_lock(&sv->sv_lock);
 547
 548         LASSERT (sv->sv_nprune >= 0);
 549         LASSERT (!sv->sv_shuttingdown);
 550
 551         sv->sv_nprune += nbuffer;
 552
 553         spin_unlock(&sv->sv_lock);
 554         return;
 555 }
 556
 557 /* returns 1 if sv has finished, otherwise 0 */
 558 int
 559 srpc_finish_service (srpc_service_t *sv)
 560 {
 561         srpc_server_rpc_t *rpc;
 562         srpc_buffer_t     *buf;
 563
 564         spin_lock(&sv->sv_lock);
 565
 566         LASSERT (sv->sv_shuttingdown); /* srpc_shutdown_service called */
 567
 568         if (sv->sv_nposted_msg != 0 || !list_empty(&sv->sv_active_rpcq)) {
 569                 CDEBUG (D_NET,
 570                         "waiting for %d posted buffers to unlink and "
 571                         "in-flight RPCs to die.\n",
 572                         sv->sv_nposted_msg);
 573
 574                 if (!list_empty(&sv->sv_active_rpcq)) {
 575                         rpc = list_entry(sv->sv_active_rpcq.next,
 576                                          srpc_server_rpc_t, srpc_list);
 577                         CDEBUG (D_NETERROR,
 578                                 "Active RPC on shutdown: sv %s, peer %s, "
 579                                 "wi %s scheduled %d running %d, "
 580                                 "ev fired %d type %d status %d lnet %d\n",
 581                                 sv->sv_name, libcfs_id2str(rpc->srpc_peer),
 582                                 swi_state2str(rpc->srpc_wi.wi_state),
 583                                 rpc->srpc_wi.wi_scheduled,
 584                                 rpc->srpc_wi.wi_running,
 585                                 rpc->srpc_ev.ev_fired,
 586                                 rpc->srpc_ev.ev_type,
 587                                 rpc->srpc_ev.ev_status,
 588                                 rpc->srpc_ev.ev_lnet);
 589                 }
 590
 591                 spin_unlock(&sv->sv_lock);
 592                 return 0;
 593         }
 594
 595         spin_unlock(&sv->sv_lock); /* no lock needed from now on */
 596
 597         for (;;) {
 598                 struct list_head *q;
 599
 600                 if (!list_empty(&sv->sv_posted_msgq))
 601                         q = &sv->sv_posted_msgq;
 602                 else if (!list_empty(&sv->sv_blocked_msgq))
 603                         q = &sv->sv_blocked_msgq;
 604                 else
 605                         break;
 606
 607                 buf = list_entry(q->next, srpc_buffer_t, buf_list);
 608                 list_del(&buf->buf_list);
 609
 610                 LIBCFS_FREE(buf, sizeof(*buf));
 611         }
 612
 613         while (!list_empty(&sv->sv_free_rpcq)) {
 614                 rpc = list_entry(sv->sv_free_rpcq.next,
 615                                  srpc_server_rpc_t, srpc_list);
 616                 list_del(&rpc->srpc_list);
 617                 LIBCFS_FREE(rpc, sizeof(*rpc));
 618         }
 619
 620         return 1;
 621 }
 622
 623 /* called with sv->sv_lock held */
 624 void
 625 srpc_service_recycle_buffer (srpc_service_t *sv, srpc_buffer_t *buf)
 626 {
 627         if (sv->sv_shuttingdown)
 628                 goto free;
 629
 630         if (sv->sv_nprune == 0) {
 631                 if (srpc_service_post_buffer(sv, buf) != 0)
 632                         CWARN ("Failed to post %s buffer\n", sv->sv_name);
 633                 return;
 634         }
 635
 636         sv->sv_nprune--;
 637 free:
 638         spin_unlock(&sv->sv_lock);
 639         LIBCFS_FREE(buf, sizeof(*buf));
 640         spin_lock(&sv->sv_lock);
 641 }
 642
 643 void
 644 srpc_shutdown_service (srpc_service_t *sv)
 645 {
 646         srpc_server_rpc_t *rpc;
 647         srpc_buffer_t     *buf;
 648
 649         spin_lock(&sv->sv_lock);
 650
 651         CDEBUG (D_NET, "Shutting down service: id %d, name %s\n",
 652                 sv->sv_id, sv->sv_name);
 653
 654         sv->sv_shuttingdown = 1; /* i.e. no new active RPC */
 655
 656         /* schedule in-flight RPCs to notice the shutdown */
 657         list_for_each_entry (rpc, &sv->sv_active_rpcq, srpc_list) {
 658                 swi_schedule_workitem(&rpc->srpc_wi);
 659         }
 660
 661         spin_unlock(&sv->sv_lock);
 662
 663         /* OK to traverse sv_posted_msgq without lock, since no one
 664          * touches sv_posted_msgq now */
 665         list_for_each_entry (buf, &sv->sv_posted_msgq, buf_list)
 666                 LNetMDUnlink(buf->buf_mdh);
 667
 668         return;
 669 }
 670
 671 int
 672 srpc_send_request (srpc_client_rpc_t *rpc)
 673 {
 674         srpc_event_t *ev = &rpc->crpc_reqstev;
 675         int           rc;
 676
 677         ev->ev_fired = 0;
 678         ev->ev_data  = rpc;
 679         ev->ev_type  = SRPC_REQUEST_SENT;
 680
 681         rc = srpc_post_active_rqtbuf(rpc->crpc_dest, rpc->crpc_service,
 682                                      &rpc->crpc_reqstmsg, sizeof(srpc_msg_t),
 683                                      &rpc->crpc_reqstmdh, ev);
 684         if (rc != 0) {
 685                 LASSERT (rc == -ENOMEM);
 686                 ev->ev_fired = 1;  /* no more event expected */
 687         }
 688         return rc;
 689 }
 690
 691 int
 692 srpc_prepare_reply (srpc_client_rpc_t *rpc)
 693 {
 694         srpc_event_t *ev = &rpc->crpc_replyev;
 695         __u64        *id = &rpc->crpc_reqstmsg.msg_body.reqst.rpyid;
 696         int           rc;
 697
 698         ev->ev_fired = 0;
 699         ev->ev_data  = rpc;
 700         ev->ev_type  = SRPC_REPLY_RCVD;
 701
 702         *id = srpc_next_id();
 703
 704         rc = srpc_post_passive_rdma(SRPC_RDMA_PORTAL, *id,
 705                                     &rpc->crpc_replymsg, sizeof(srpc_msg_t),
 706                                     LNET_MD_OP_PUT, rpc->crpc_dest,
 707                                     &rpc->crpc_replymdh, ev);
 708         if (rc != 0) {
 709                 LASSERT (rc == -ENOMEM);
 710                 ev->ev_fired = 1;  /* no more event expected */
 711         }
 712         return rc;
 713 }
 714
 715 int
 716 srpc_prepare_bulk (srpc_client_rpc_t *rpc)
 717 {
 718         srpc_bulk_t  *bk = &rpc->crpc_bulk;
 719         srpc_event_t *ev = &rpc->crpc_bulkev;
 720         __u64        *id = &rpc->crpc_reqstmsg.msg_body.reqst.bulkid;
 721         int           rc;
 722         int           opt;
 723
 724         LASSERT (bk->bk_niov <= LNET_MAX_IOV);
 725
 726         if (bk->bk_niov == 0) return 0; /* nothing to do */
 727
 728         opt = bk->bk_sink ? LNET_MD_OP_PUT : LNET_MD_OP_GET;
 729 #ifdef __KERNEL__
 730         opt |= LNET_MD_KIOV;
 731 #else
 732         opt |= LNET_MD_IOVEC;
 733 #endif
 734
 735         ev->ev_fired = 0;
 736         ev->ev_data  = rpc;
 737         ev->ev_type  = SRPC_BULK_REQ_RCVD;
 738
 739         *id = srpc_next_id();
 740
 741         rc = srpc_post_passive_rdma(SRPC_RDMA_PORTAL, *id,
 742                                     &bk->bk_iovs[0], bk->bk_niov, opt,
 743                                     rpc->crpc_dest, &bk->bk_mdh, ev);
 744         if (rc != 0) {
 745                 LASSERT (rc == -ENOMEM);
 746                 ev->ev_fired = 1;  /* no more event expected */
 747         }
 748         return rc;
 749 }
 750
 751 int
 752 srpc_do_bulk (srpc_server_rpc_t *rpc)
 753 {
 754         srpc_event_t  *ev = &rpc->srpc_ev;
 755         srpc_bulk_t   *bk = rpc->srpc_bulk;
 756         __u64          id = rpc->srpc_reqstbuf->buf_msg.msg_body.reqst.bulkid;
 757         int            rc;
 758         int            opt;
 759
 760         LASSERT (bk != NULL);
 761
 762         opt = bk->bk_sink ? LNET_MD_OP_GET : LNET_MD_OP_PUT;
 763 #ifdef __KERNEL__
 764         opt |= LNET_MD_KIOV;
 765 #else
 766         opt |= LNET_MD_IOVEC;
 767 #endif
 768
 769         ev->ev_fired = 0;
 770         ev->ev_data  = rpc;
 771         ev->ev_type  = bk->bk_sink ? SRPC_BULK_GET_RPLD : SRPC_BULK_PUT_SENT;
 772
 773         rc = srpc_post_active_rdma(SRPC_RDMA_PORTAL, id,
 774                                    &bk->bk_iovs[0], bk->bk_niov, opt,
 775                                    rpc->srpc_peer, rpc->srpc_self,
 776                                    &bk->bk_mdh, ev);
 777         if (rc != 0)
 778                 ev->ev_fired = 1;  /* no more event expected */
 779         return rc;
 780 }
 781
 782 /* called with srpc_service_t::sv_lock held */
 783 inline void
 784 srpc_schedule_server_rpc (srpc_server_rpc_t *rpc)
 785 {
 786         srpc_service_t *sv = rpc->srpc_service;
 787
 788         if (sv->sv_id > SRPC_FRAMEWORK_SERVICE_MAX_ID)
 789                 swi_schedule_workitem(&rpc->srpc_wi);
 790         else    /* framework RPCs are handled one by one */
 791                 swi_schedule_serial_workitem(&rpc->srpc_wi);
 792
 793         return;
 794 }
 795
 796 /* only called from srpc_handle_rpc */
 797 void
 798 srpc_server_rpc_done (srpc_server_rpc_t *rpc, int status)
 799 {
 800         srpc_service_t *sv = rpc->srpc_service;
 801         srpc_buffer_t  *buffer;
 802
 803         LASSERT (status != 0 || rpc->srpc_wi.wi_state == SWI_STATE_DONE);
 804
 805         rpc->srpc_status = status;
 806
 807         CDEBUG (status == 0 ? D_NET : D_NETERROR,
 808                 "Server RPC done: service %s, peer %s, status %s:%d\n",
 809                 sv->sv_name, libcfs_id2str(rpc->srpc_peer),
 810                 swi_state2str(rpc->srpc_wi.wi_state), status);
 811
 812         if (status != 0) {
 813                 spin_lock(&srpc_data.rpc_glock);
 814                 srpc_data.rpc_counters.rpcs_dropped++;
 815                 spin_unlock(&srpc_data.rpc_glock);
 816         }
 817
 818         if (rpc->srpc_done != NULL)
 819                 (*rpc->srpc_done) (rpc);
 820         LASSERT (rpc->srpc_bulk == NULL);
 821
 822         spin_lock(&sv->sv_lock);
 823
 824         if (rpc->srpc_reqstbuf != NULL) {
 825                 /* NB might drop sv_lock in srpc_service_recycle_buffer, but
 826                  * sv won't go away for sv_active_rpcq must not be empty */
 827                 srpc_service_recycle_buffer(sv, rpc->srpc_reqstbuf);
 828                 rpc->srpc_reqstbuf = NULL;
 829         }
 830
 831         list_del(&rpc->srpc_list); /* from sv->sv_active_rpcq */
 832
 833         /*
 834          * No one can schedule me now since:
 835          * - I'm not on sv_active_rpcq.
 836          * - all LNet events have been fired.
 837          * Cancel pending schedules and prevent future schedule attempts:
 838          */
 839         LASSERT (rpc->srpc_ev.ev_fired);
 840         swi_kill_workitem(&rpc->srpc_wi);
 841
 842         if (!sv->sv_shuttingdown && !list_empty(&sv->sv_blocked_msgq)) {
 843                 buffer = list_entry(sv->sv_blocked_msgq.next,
 844                                     srpc_buffer_t, buf_list);
 845                 list_del(&buffer->buf_list);
 846
 847                 srpc_init_server_rpc(rpc, sv, buffer);
 848                 list_add_tail(&rpc->srpc_list, &sv->sv_active_rpcq);
 849                 srpc_schedule_server_rpc(rpc);
 850         } else {
 851                 list_add(&rpc->srpc_list, &sv->sv_free_rpcq);
 852         }
 853
 854         spin_unlock(&sv->sv_lock);
 855         return;
 856 }
 857
 858 /* handles an incoming RPC */
 859 int
 860 srpc_handle_rpc (swi_workitem_t *wi)
 861 {
 862         srpc_server_rpc_t *rpc = wi->wi_data;
 863         srpc_service_t    *sv = rpc->srpc_service;
 864         srpc_event_t      *ev = &rpc->srpc_ev;
 865         int                rc = 0;
 866
 867         LASSERT (wi == &rpc->srpc_wi);
 868
 869         spin_lock(&sv->sv_lock);
 870
 871         if (sv->sv_shuttingdown) {
 872                 spin_unlock(&sv->sv_lock);
 873
 874                 if (rpc->srpc_bulk != NULL)
 875                         LNetMDUnlink(rpc->srpc_bulk->bk_mdh);
 876                 LNetMDUnlink(rpc->srpc_replymdh);
 877
 878                 if (ev->ev_fired) { /* no more event, OK to finish */
 879                         srpc_server_rpc_done(rpc, -ESHUTDOWN);
 880                         return 1;
 881                 }
 882                 return 0;
 883         }
 884
 885         spin_unlock(&sv->sv_lock);
 886
 887         switch (wi->wi_state) {
 888         default:
 889                 LBUG ();
 890         case SWI_STATE_NEWBORN: {
 891                 srpc_msg_t           *msg;
 892                 srpc_generic_reply_t *reply;
 893
 894                 msg = &rpc->srpc_reqstbuf->buf_msg;
 895                 reply = &rpc->srpc_replymsg.msg_body.reply;
 896
 897                 if (msg->msg_version != SRPC_MSG_VERSION &&
 898                     msg->msg_version != __swab32(SRPC_MSG_VERSION)) {
 899                         CWARN ("Version mismatch: %u, %u expected, from %s\n",
 900                                msg->msg_version, SRPC_MSG_VERSION,
 901                                libcfs_id2str(rpc->srpc_peer));
 902                         reply->status = EPROTO;
 903                 } else {
 904                         reply->status = 0;
 905                         rc = (*sv->sv_handler) (rpc);
 906                         LASSERT (reply->status == 0 || !rpc->srpc_bulk);
 907                 }
 908
 909                 if (rc != 0) {
 910                         srpc_server_rpc_done(rpc, rc);
 911                         return 1;
 912                 }
 913
 914                 wi->wi_state = SWI_STATE_BULK_STARTED;
 915
 916                 if (rpc->srpc_bulk != NULL) {
 917                         rc = srpc_do_bulk(rpc);
 918                         if (rc == 0)
 919                                 return 0; /* wait for bulk */
 920
 921                         LASSERT (ev->ev_fired);
 922                         ev->ev_status = rc;
 923                 }
 924         }
 925         case SWI_STATE_BULK_STARTED:
 926                 LASSERT (rpc->srpc_bulk == NULL || ev->ev_fired);
 927
 928                 if (rpc->srpc_bulk != NULL) {
 929                         rc = ev->ev_status;
 930
 931                         if (sv->sv_bulk_ready != NULL)
 932                                 rc = (*sv->sv_bulk_ready) (rpc, rc);
 933
 934                         if (rc != 0) {
 935                                 srpc_server_rpc_done(rpc, rc);
 936                                 return 1;
 937                         }
 938                 }
 939
 940                 wi->wi_state = SWI_STATE_REPLY_SUBMITTED;
 941                 rc = srpc_send_reply(rpc);
 942                 if (rc == 0)
 943                         return 0; /* wait for reply */
 944                 srpc_server_rpc_done(rpc, rc);
 945                 return 1;
 946
 947         case SWI_STATE_REPLY_SUBMITTED:
 948                 LASSERT (ev->ev_fired);
 949
 950                 wi->wi_state = SWI_STATE_DONE;
 951                 srpc_server_rpc_done(rpc, ev->ev_status);
 952                 return 1;
 953         }
 954
 955         return 0;
 956 }
 957
 958 void
 959 srpc_client_rpc_expired (void *data)
 960 {
 961         srpc_client_rpc_t *rpc = data;
 962
 963         CWARN ("Client RPC expired: service %d, peer %s, timeout %d.\n",
 964                rpc->crpc_service, libcfs_id2str(rpc->crpc_dest),
 965                rpc->crpc_timeout);
 966
 967         spin_lock(&rpc->crpc_lock);
 968
 969         rpc->crpc_timeout = 0;
 970         srpc_abort_rpc(rpc, -ETIMEDOUT);
 971
 972         spin_unlock(&rpc->crpc_lock);
 973
 974         spin_lock(&srpc_data.rpc_glock);
 975         srpc_data.rpc_counters.rpcs_expired++;
 976         spin_unlock(&srpc_data.rpc_glock);
 977         return;
 978 }
 979
 980 inline void
 981 srpc_add_client_rpc_timer (srpc_client_rpc_t *rpc)
 982 {
 983         stt_timer_t *timer = &rpc->crpc_timer;
 984
 985         if (rpc->crpc_timeout == 0) return;
 986
 987         CFS_INIT_LIST_HEAD(&timer->stt_list);
 988         timer->stt_data    = rpc;
 989         timer->stt_func    = srpc_client_rpc_expired;
 990         timer->stt_expires = cfs_time_add(rpc->crpc_timeout,
 991                                           cfs_time_current_sec());
 992         stt_add_timer(timer);
 993         return;
 994 }
 995
 996 /*
 997  * Called with rpc->crpc_lock held.
 998  *
 999  * Upon exit the RPC expiry timer is not queued and the handler is not
1000  * running on any CPU. */
1001 void
1002 srpc_del_client_rpc_timer (srpc_client_rpc_t *rpc)
1003 {
1004         /* timer not planted or already exploded */
1005         if (rpc->crpc_timeout == 0) return;
1006
1007         /* timer sucessfully defused */
1008         if (stt_del_timer(&rpc->crpc_timer)) return;
1009
1010 #ifdef __KERNEL__
1011         /* timer detonated, wait for it to explode */
1012         while (rpc->crpc_timeout != 0) {
1013                 spin_unlock(&rpc->crpc_lock);
1014
1015                 cfs_schedule();
1016
1017                 spin_lock(&rpc->crpc_lock);
1018         }
1019 #else
1020         LBUG(); /* impossible in single-threaded runtime */
1021 #endif
1022         return;
1023 }
1024
1025 void
1026 srpc_check_sends (srpc_peer_t *peer, int credits)
1027 {
1028         struct list_head  *q;
1029         srpc_client_rpc_t *rpc;
1030
1031         LASSERT (credits >= 0);
1032         LASSERT (srpc_data.rpc_state == SRPC_STATE_RUNNING);
1033
1034         spin_lock(&peer->stp_lock);
1035         peer->stp_credits += credits;
1036
1037         while (peer->stp_credits) {
1038                 if (!list_empty(&peer->stp_ctl_rpcq))
1039                         q = &peer->stp_ctl_rpcq;
1040                 else if (!list_empty(&peer->stp_rpcq))
1041                         q = &peer->stp_rpcq;
1042                 else
1043                         break;
1044
1045                 peer->stp_credits--;
1046
1047                 rpc = list_entry(q->next, srpc_client_rpc_t, crpc_privl);
1048                 list_del_init(&rpc->crpc_privl);
1049                 srpc_client_rpc_decref(rpc);  /* --ref for peer->*rpcq */
1050
1051                 swi_schedule_workitem(&rpc->crpc_wi);
1052         }
1053
1054         spin_unlock(&peer->stp_lock);
1055         return;
1056 }
1057
1058 void
1059 srpc_client_rpc_done (srpc_client_rpc_t *rpc, int status)
1060 {
1061         swi_workitem_t *wi = &rpc->crpc_wi;
1062         srpc_peer_t    *peer = rpc->crpc_peer;
1063
1064         LASSERT (status != 0 || wi->wi_state == SWI_STATE_DONE);
1065
1066         spin_lock(&rpc->crpc_lock);
1067
1068         rpc->crpc_closed = 1;
1069         if (rpc->crpc_status == 0)
1070                 rpc->crpc_status = status;
1071
1072         srpc_del_client_rpc_timer(rpc);
1073
1074         CDEBUG ((status == 0) ? D_NET : D_NETERROR,
1075                 "Client RPC done: service %d, peer %s, status %s:%d:%d\n",
1076                 rpc->crpc_service, libcfs_id2str(rpc->crpc_dest),
1077                 swi_state2str(wi->wi_state), rpc->crpc_aborted, status);
1078
1079         /*
1080          * No one can schedule me now since:
1081          * - RPC timer has been defused.
1082          * - all LNet events have been fired.
1083          * - crpc_closed has been set, preventing srpc_abort_rpc from
1084          *   scheduling me.
1085          * Cancel pending schedules and prevent future schedule attempts:
1086          */
1087         LASSERT (!srpc_event_pending(rpc));
1088         swi_kill_workitem(wi);
1089
1090         spin_unlock(&rpc->crpc_lock);
1091
1092         (*rpc->crpc_done) (rpc);
1093
1094         if (peer != NULL)
1095                 srpc_check_sends(peer, 1);
1096         return;
1097 }
1098
1099 /* sends an outgoing RPC */
1100 int
1101 srpc_send_rpc (swi_workitem_t *wi)
1102 {
1103         int                rc = 0;
1104         srpc_client_rpc_t *rpc = wi->wi_data;
1105         srpc_msg_t        *reply = &rpc->crpc_replymsg;
1106         int                do_bulk = rpc->crpc_bulk.bk_niov > 0;
1107
1108         LASSERT (rpc != NULL);
1109         LASSERT (wi == &rpc->crpc_wi);
1110
1111         spin_lock(&rpc->crpc_lock);
1112
1113         if (rpc->crpc_aborted) {
1114                 spin_unlock(&rpc->crpc_lock);
1115                 goto abort;
1116         }
1117
1118         spin_unlock(&rpc->crpc_lock);
1119
1120         switch (wi->wi_state) {
1121         default:
1122                 LBUG ();
1123         case SWI_STATE_NEWBORN:
1124                 LASSERT (!srpc_event_pending(rpc));
1125
1126                 rc = srpc_prepare_reply(rpc);
1127                 if (rc != 0) {
1128                         srpc_client_rpc_done(rpc, rc);
1129                         return 1;
1130                 }
1131
1132                 rc = srpc_prepare_bulk(rpc);
1133                 if (rc != 0) break;
1134
1135                 wi->wi_state = SWI_STATE_REQUEST_SUBMITTED;
1136                 rc = srpc_send_request(rpc);
1137                 break;
1138
1139         case SWI_STATE_REQUEST_SUBMITTED:
1140                 /* CAVEAT EMPTOR: rqtev, rpyev, and bulkev may come in any
1141                  * order; however, they're processed in a strict order:
1142                  * rqt, rpy, and bulk. */
1143                 if (!rpc->crpc_reqstev.ev_fired) break;
1144
1145                 rc = rpc->crpc_reqstev.ev_status;
1146                 if (rc != 0) break;
1147
1148                 wi->wi_state = SWI_STATE_REQUEST_SENT;
1149                 /* perhaps more events, fall thru */
1150         case SWI_STATE_REQUEST_SENT: {
1151                 srpc_msg_type_t type = srpc_service2reply(rpc->crpc_service);
1152
1153                 if (!rpc->crpc_replyev.ev_fired) break;
1154
1155                 rc = rpc->crpc_replyev.ev_status;
1156                 if (rc != 0) break;
1157
1158                 if ((reply->msg_type != type &&
1159                      reply->msg_type != __swab32(type)) ||
1160                     (reply->msg_magic != SRPC_MSG_MAGIC &&
1161                      reply->msg_magic != __swab32(SRPC_MSG_MAGIC))) {
1162                         CWARN ("Bad message from %s: type %u (%d expected),"
1163                                " magic %u (%d expected).\n",
1164                                libcfs_id2str(rpc->crpc_dest),
1165                                reply->msg_type, type,
1166                                reply->msg_magic, SRPC_MSG_MAGIC);
1167                         rc = -EBADMSG;
1168                         break;
1169                 }
1170
1171                 if (do_bulk && reply->msg_body.reply.status != 0) {
1172                         CWARN ("Remote error %d at %s, unlink bulk buffer in "
1173                                "case peer didn't initiate bulk transfer\n",
1174                                reply->msg_body.reply.status,
1175                                libcfs_id2str(rpc->crpc_dest));
1176                         LNetMDUnlink(rpc->crpc_bulk.bk_mdh);
1177                 }
1178
1179                 wi->wi_state = SWI_STATE_REPLY_RECEIVED;
1180         }
1181         case SWI_STATE_REPLY_RECEIVED:
1182                 if (do_bulk && !rpc->crpc_bulkev.ev_fired) break;
1183
1184                 rc = do_bulk ? rpc->crpc_bulkev.ev_status : 0;
1185
1186                 /* Bulk buffer was unlinked due to remote error. Clear error
1187                  * since reply buffer still contains valid data.
1188                  * NB rpc->crpc_done shouldn't look into bulk data in case of
1189                  * remote error. */
1190                 if (do_bulk && rpc->crpc_bulkev.ev_lnet == LNET_EVENT_UNLINK &&
1191                     rpc->crpc_status == 0 && reply->msg_body.reply.status != 0)
1192                         rc = 0;
1193
1194                 wi->wi_state = SWI_STATE_DONE;
1195                 srpc_client_rpc_done(rpc, rc);
1196                 return 1;
1197         }
1198
1199         if (rc != 0) {
1200                 spin_lock(&rpc->crpc_lock);
1201                 srpc_abort_rpc(rpc, rc);
1202                 spin_unlock(&rpc->crpc_lock);
1203         }
1204
1205 abort:
1206         if (rpc->crpc_aborted) {
1207                 LNetMDUnlink(rpc->crpc_reqstmdh);
1208                 LNetMDUnlink(rpc->crpc_replymdh);
1209                 LNetMDUnlink(rpc->crpc_bulk.bk_mdh);
1210
1211                 if (!srpc_event_pending(rpc)) {
1212                         srpc_client_rpc_done(rpc, -EINTR);
1213                         return 1;
1214                 }
1215         }
1216         return 0;
1217 }
1218
1219 srpc_client_rpc_t *
1220 srpc_create_client_rpc (lnet_process_id_t peer, int service,
1221                         int nbulkiov, int bulklen,
1222                         void (*rpc_done)(srpc_client_rpc_t *),
1223                         void (*rpc_fini)(srpc_client_rpc_t *), void *priv)
1224 {
1225         srpc_client_rpc_t *rpc;
1226
1227         LIBCFS_ALLOC(rpc, offsetof(srpc_client_rpc_t,
1228                                    crpc_bulk.bk_iovs[nbulkiov]));
1229         if (rpc == NULL)
1230                 return NULL;
1231
1232         srpc_init_client_rpc(rpc, peer, service, nbulkiov,
1233                              bulklen, rpc_done, rpc_fini, priv);
1234         return rpc;
1235 }
1236
1237 /* called with rpc->crpc_lock held */
1238 static inline void
1239 srpc_queue_rpc (srpc_peer_t *peer, srpc_client_rpc_t *rpc)
1240 {
1241         int service = rpc->crpc_service;
1242
1243         LASSERT (peer->stp_nid == rpc->crpc_dest.nid);
1244         LASSERT (srpc_data.rpc_state == SRPC_STATE_RUNNING);
1245
1246         rpc->crpc_peer = peer;
1247
1248         spin_lock(&peer->stp_lock);
1249
1250         /* Framework RPCs that alter session state shall take precedence
1251          * over test RPCs and framework query RPCs */
1252         if (service <= SRPC_FRAMEWORK_SERVICE_MAX_ID &&
1253             service != SRPC_SERVICE_DEBUG &&
1254             service != SRPC_SERVICE_QUERY_STAT)
1255                 list_add_tail(&rpc->crpc_privl, &peer->stp_ctl_rpcq);
1256         else
1257                 list_add_tail(&rpc->crpc_privl, &peer->stp_rpcq);
1258
1259         srpc_client_rpc_addref(rpc); /* ++ref for peer->*rpcq */
1260         spin_unlock(&peer->stp_lock);
1261         return;
1262 }
1263
1264 /* called with rpc->crpc_lock held */
1265 void
1266 srpc_abort_rpc (srpc_client_rpc_t *rpc, int why)
1267 {
1268         srpc_peer_t *peer = rpc->crpc_peer;
1269
1270         LASSERT (why != 0);
1271
1272         if (rpc->crpc_aborted || /* already aborted */
1273             rpc->crpc_closed)    /* callback imminent */
1274                 return;
1275
1276         CDEBUG (D_NET,
1277                 "Aborting RPC: service %d, peer %s, state %s, why %d\n",
1278                 rpc->crpc_service, libcfs_id2str(rpc->crpc_dest),
1279                 swi_state2str(rpc->crpc_wi.wi_state), why);
1280
1281         rpc->crpc_aborted = 1;
1282         rpc->crpc_status  = why;
1283
1284         if (peer != NULL) {
1285                 spin_lock(&peer->stp_lock);
1286
1287                 if (!list_empty(&rpc->crpc_privl)) { /* still queued */
1288                         list_del_init(&rpc->crpc_privl);
1289                         srpc_client_rpc_decref(rpc); /* --ref for peer->*rpcq */
1290                         rpc->crpc_peer = NULL;       /* no credit taken */
1291                 }
1292
1293                 spin_unlock(&peer->stp_lock);
1294         }
1295
1296         swi_schedule_workitem(&rpc->crpc_wi);
1297         return;
1298 }
1299
1300 /* called with rpc->crpc_lock held */
1301 void
1302 srpc_post_rpc (srpc_client_rpc_t *rpc)
1303 {
1304         srpc_peer_t *peer;
1305
1306         LASSERT (!rpc->crpc_aborted);
1307         LASSERT (rpc->crpc_peer == NULL);
1308         LASSERT (srpc_data.rpc_state == SRPC_STATE_RUNNING);
1309         LASSERT ((rpc->crpc_bulk.bk_len & ~CFS_PAGE_MASK) == 0);
1310
1311         CDEBUG (D_NET, "Posting RPC: peer %s, service %d, timeout %d\n",
1312                 libcfs_id2str(rpc->crpc_dest), rpc->crpc_service,
1313                 rpc->crpc_timeout);
1314
1315         srpc_add_client_rpc_timer(rpc);
1316
1317         peer = srpc_nid2peer(rpc->crpc_dest.nid);
1318         if (peer == NULL) {
1319                 srpc_abort_rpc(rpc, -ENOMEM);
1320                 return;
1321         }
1322
1323         srpc_queue_rpc(peer, rpc);
1324
1325         spin_unlock(&rpc->crpc_lock);
1326         srpc_check_sends(peer, 0);
1327         spin_lock(&rpc->crpc_lock);
1328         return;
1329 }
1330
1331
1332 int
1333 srpc_send_reply (srpc_server_rpc_t *rpc)
1334 {
1335         srpc_event_t   *ev = &rpc->srpc_ev;
1336         srpc_msg_t     *msg = &rpc->srpc_replymsg;
1337         srpc_buffer_t  *buffer = rpc->srpc_reqstbuf;
1338         srpc_service_t *sv = rpc->srpc_service;
1339         __u64           rpyid;
1340         int             rc;
1341
1342         LASSERT (buffer != NULL);
1343         rpyid = buffer->buf_msg.msg_body.reqst.rpyid;
1344
1345         spin_lock(&sv->sv_lock);
1346
1347         if (!sv->sv_shuttingdown &&
1348             sv->sv_id > SRPC_FRAMEWORK_SERVICE_MAX_ID) {
1349                 /* Repost buffer before replying since test client
1350                  * might send me another RPC once it gets the reply */
1351                 if (srpc_service_post_buffer(sv, buffer) != 0)
1352                         CWARN ("Failed to repost %s buffer\n", sv->sv_name);
1353                 rpc->srpc_reqstbuf = NULL;
1354         }
1355
1356         spin_unlock(&sv->sv_lock);
1357
1358         ev->ev_fired = 0;
1359         ev->ev_data  = rpc;
1360         ev->ev_type  = SRPC_REPLY_SENT;
1361
1362         msg->msg_magic   = SRPC_MSG_MAGIC;
1363         msg->msg_version = SRPC_MSG_VERSION;
1364         msg->msg_type    = srpc_service2reply(sv->sv_id);
1365
1366         rc = srpc_post_active_rdma(SRPC_RDMA_PORTAL, rpyid, msg,
1367                                    sizeof(*msg), LNET_MD_OP_PUT,
1368                                    rpc->srpc_peer, rpc->srpc_self,
1369                                    &rpc->srpc_replymdh, ev);
1370         if (rc != 0)
1371                 ev->ev_fired = 1;  /* no more event expected */
1372         return rc;
1373 }
1374
1375 /* always called with LNET_LOCK() held, and in thread context */
1376 void
1377 srpc_lnet_ev_handler (lnet_event_t *ev)
1378 {
1379         srpc_event_t      *rpcev = ev->md.user_ptr;
1380         srpc_client_rpc_t *crpc;
1381         srpc_server_rpc_t *srpc;
1382         srpc_buffer_t     *buffer;
1383         srpc_service_t    *sv;
1384         srpc_msg_t        *msg;
1385         srpc_msg_type_t    type;
1386
1387         LASSERT (!in_interrupt());
1388
1389         if (ev->status != 0) {
1390                 spin_lock(&srpc_data.rpc_glock);
1391                 srpc_data.rpc_counters.errors++;
1392                 spin_unlock(&srpc_data.rpc_glock);
1393         }
1394
1395         rpcev->ev_lnet = ev->type;
1396
1397         switch (rpcev->ev_type) {
1398         default:
1399                 LBUG ();
1400         case SRPC_REQUEST_SENT:
1401                 if (ev->status == 0 && ev->type != LNET_EVENT_UNLINK) {
1402                         spin_lock(&srpc_data.rpc_glock);
1403                         srpc_data.rpc_counters.rpcs_sent++;
1404                         spin_unlock(&srpc_data.rpc_glock);
1405                 }
1406         case SRPC_REPLY_RCVD:
1407         case SRPC_BULK_REQ_RCVD:
1408                 crpc = rpcev->ev_data;
1409
1410                 LASSERT (rpcev == &crpc->crpc_reqstev ||
1411                          rpcev == &crpc->crpc_replyev ||
1412                          rpcev == &crpc->crpc_bulkev);
1413
1414                 spin_lock(&crpc->crpc_lock);
1415
1416                 LASSERT (rpcev->ev_fired == 0);
1417                 rpcev->ev_fired  = 1;
1418                 rpcev->ev_status = (ev->type == LNET_EVENT_UNLINK) ?
1419                                                 -EINTR : ev->status;
1420                 swi_schedule_workitem(&crpc->crpc_wi);
1421
1422                 spin_unlock(&crpc->crpc_lock);
1423                 break;
1424
1425         case SRPC_REQUEST_RCVD:
1426                 sv = rpcev->ev_data;
1427
1428                 LASSERT (rpcev == &sv->sv_ev);
1429
1430                 spin_lock(&sv->sv_lock);
1431
1432                 LASSERT (ev->unlinked);
1433                 LASSERT (ev->type == LNET_EVENT_PUT ||
1434                          ev->type == LNET_EVENT_UNLINK);
1435                 LASSERT (ev->type != LNET_EVENT_UNLINK ||
1436                          sv->sv_shuttingdown);
1437
1438                 buffer = container_of(ev->md.start, srpc_buffer_t, buf_msg);
1439                 buffer->buf_peer = ev->initiator;
1440                 buffer->buf_self = ev->target.nid;
1441
1442                 sv->sv_nposted_msg--;
1443                 LASSERT (sv->sv_nposted_msg >= 0);
1444
1445                 if (sv->sv_shuttingdown) {
1446                         /* Leave buffer on sv->sv_posted_msgq since
1447                          * srpc_finish_service needs to traverse it. */
1448                         spin_unlock(&sv->sv_lock);
1449                         break;
1450                 }
1451
1452                 list_del(&buffer->buf_list); /* from sv->sv_posted_msgq */
1453                 msg = &buffer->buf_msg;
1454                 type = srpc_service2request(sv->sv_id);
1455
1456                 if (ev->status != 0 || ev->mlength != sizeof(*msg) ||
1457                     (msg->msg_type != type &&
1458                      msg->msg_type != __swab32(type)) ||
1459                     (msg->msg_magic != SRPC_MSG_MAGIC &&
1460                      msg->msg_magic != __swab32(SRPC_MSG_MAGIC))) {
1461                         CERROR ("Dropping RPC (%s) from %s: "
1462                                 "status %d mlength %d type %u magic %u.\n",
1463                                 sv->sv_name, libcfs_id2str(ev->initiator),
1464                                 ev->status, ev->mlength,
1465                                 msg->msg_type, msg->msg_magic);
1466
1467                         /* NB might drop sv_lock in srpc_service_recycle_buffer,
1468                          * sv_nposted_msg++ as an implicit reference to prevent
1469                          * sv from disappearing under me */
1470                         sv->sv_nposted_msg++;
1471                         srpc_service_recycle_buffer(sv, buffer);
1472                         sv->sv_nposted_msg--;
1473                         spin_unlock(&sv->sv_lock);
1474
1475                         if (ev->status == 0) { /* status!=0 counted already */
1476                                 spin_lock(&srpc_data.rpc_glock);
1477                                 srpc_data.rpc_counters.errors++;
1478                                 spin_unlock(&srpc_data.rpc_glock);
1479                         }
1480                         break;
1481                 }
1482
1483                 if (!list_empty(&sv->sv_free_rpcq)) {
1484                         srpc = list_entry(sv->sv_free_rpcq.next,
1485                                           srpc_server_rpc_t, srpc_list);
1486                         list_del(&srpc->srpc_list);
1487
1488                         srpc_init_server_rpc(srpc, sv, buffer);
1489                         list_add_tail(&srpc->srpc_list, &sv->sv_active_rpcq);
1490                         srpc_schedule_server_rpc(srpc);
1491                 } else {
1492                         list_add_tail(&buffer->buf_list, &sv->sv_blocked_msgq);
1493                 }
1494
1495                 spin_unlock(&sv->sv_lock);
1496
1497                 spin_lock(&srpc_data.rpc_glock);
1498                 srpc_data.rpc_counters.rpcs_rcvd++;
1499                 spin_unlock(&srpc_data.rpc_glock);
1500                 break;
1501
1502         case SRPC_BULK_GET_RPLD:
1503                 LASSERT (ev->type == LNET_EVENT_SEND ||
1504                          ev->type == LNET_EVENT_REPLY ||
1505                          ev->type == LNET_EVENT_UNLINK);
1506
1507                 if (ev->type == LNET_EVENT_SEND &&
1508                     ev->status == 0 && !ev->unlinked)
1509                         break; /* wait for the final LNET_EVENT_REPLY */
1510
1511         case SRPC_BULK_PUT_SENT:
1512                 if (ev->status == 0 && ev->type != LNET_EVENT_UNLINK) {
1513                         spin_lock(&srpc_data.rpc_glock);
1514
1515                         if (rpcev->ev_type == SRPC_BULK_GET_RPLD)
1516                                 srpc_data.rpc_counters.bulk_get += ev->mlength;
1517                         else
1518                                 srpc_data.rpc_counters.bulk_put += ev->mlength;
1519
1520                         spin_unlock(&srpc_data.rpc_glock);
1521                 }
1522         case SRPC_REPLY_SENT:
1523                 srpc = rpcev->ev_data;
1524                 sv = srpc->srpc_service;
1525
1526                 LASSERT (rpcev == &srpc->srpc_ev);
1527
1528                 spin_lock(&sv->sv_lock);
1529                 rpcev->ev_fired  = 1;
1530                 rpcev->ev_status = (ev->type == LNET_EVENT_UNLINK) ?
1531                                                 -EINTR : ev->status;
1532                 srpc_schedule_server_rpc(srpc);
1533                 spin_unlock(&sv->sv_lock);
1534                 break;
1535         }
1536
1537         return;
1538 }
1539
1540 #ifndef __KERNEL__
1541
1542 int
1543 srpc_check_event (int timeout)
1544 {
1545         lnet_event_t ev;
1546         int          rc;
1547         int          i;
1548
1549         rc = LNetEQPoll(&srpc_data.rpc_lnet_eq, 1,
1550                         timeout * 1000, &ev, &i);
1551         if (rc == 0)
1552                 return 0;
1553
1554         LASSERT (rc == -EOVERFLOW || rc == 1);
1555
1556         /* We can't affort to miss any events... */
1557         if (rc == -EOVERFLOW) {
1558                 CERROR ("Dropped an event!!!\n");
1559                 abort();
1560         }
1561
1562         srpc_lnet_ev_handler(&ev);
1563         return 1;
1564 }
1565
1566 #endif
1567
1568 int
1569 srpc_startup (void)
1570 {
1571         int i;
1572         int rc;
1573
1574         memset(&srpc_data, 0, sizeof(struct smoketest_rpc));
1575         spin_lock_init(&srpc_data.rpc_glock);
1576
1577         /* 1 second pause to avoid timestamp reuse */
1578         cfs_pause(cfs_time_seconds(1));
1579         srpc_data.rpc_matchbits = ((__u64) cfs_time_current_sec()) << 48;
1580
1581         srpc_data.rpc_state = SRPC_STATE_NONE;
1582
1583         LIBCFS_ALLOC(srpc_data.rpc_peers,
1584                      sizeof(struct list_head) * SRPC_PEER_HASH_SIZE);
1585         if (srpc_data.rpc_peers == NULL) {
1586                 CERROR ("Failed to alloc peer hash.\n");
1587                 return -ENOMEM;
1588         }
1589
1590         for (i = 0; i < SRPC_PEER_HASH_SIZE; i++)
1591                 CFS_INIT_LIST_HEAD(&srpc_data.rpc_peers[i]);
1592
1593 #ifdef __KERNEL__
1594         rc = LNetNIInit(LUSTRE_SRV_LNET_PID);
1595 #else
1596         rc = LNetNIInit(getpid());
1597 #endif
1598         if (rc < 0) {
1599                 CERROR ("LNetNIInit() has failed: %d\n", rc);
1600                 LIBCFS_FREE(srpc_data.rpc_peers,
1601                             sizeof(struct list_head) * SRPC_PEER_HASH_SIZE);
1602                 return rc;
1603         }
1604
1605         srpc_data.rpc_state = SRPC_STATE_NI_INIT;
1606
1607         srpc_data.rpc_lnet_eq = LNET_EQ_NONE;
1608 #ifdef __KERNEL__
1609         rc = LNetEQAlloc(16, srpc_lnet_ev_handler, &srpc_data.rpc_lnet_eq);
1610 #else
1611         rc = LNetEQAlloc(10240, LNET_EQ_HANDLER_NONE, &srpc_data.rpc_lnet_eq);
1612 #endif
1613         if (rc != 0) {
1614                 CERROR("LNetEQAlloc() has failed: %d\n", rc);
1615                 goto bail;
1616         }
1617
1618         rc = LNetSetLazyPortal(SRPC_FRAMEWORK_REQUEST_PORTAL);
1619         LASSERT (rc == 0);
1620
1621         srpc_data.rpc_state = SRPC_STATE_EQ_INIT;
1622
1623         rc = swi_startup();
1624         if (rc != 0)
1625                 goto bail;
1626
1627         srpc_data.rpc_state = SRPC_STATE_WI_INIT;
1628
1629         rc = stt_startup();
1630
1631 bail:
1632         if (rc != 0)
1633                 srpc_shutdown();
1634         else
1635                 srpc_data.rpc_state = SRPC_STATE_RUNNING;
1636
1637         return rc;
1638 }
1639
1640 void
1641 srpc_shutdown (void)
1642 {
1643         int i;
1644         int rc;
1645         int state;
1646
1647         state = srpc_data.rpc_state;
1648         srpc_data.rpc_state = SRPC_STATE_STOPPING;
1649
1650         switch (state) {
1651         default:
1652                 LBUG ();
1653         case SRPC_STATE_RUNNING:
1654                 spin_lock(&srpc_data.rpc_glock);
1655
1656                 for (i = 0; i <= SRPC_SERVICE_MAX_ID; i++) {
1657                         srpc_service_t *sv = srpc_data.rpc_services[i];
1658
1659                         LASSERTF (sv == NULL,
1660                                   "service not empty: id %d, name %s\n",
1661                                   i, sv->sv_name);
1662                 }
1663
1664                 spin_unlock(&srpc_data.rpc_glock);
1665
1666                 stt_shutdown();
1667
1668         case SRPC_STATE_WI_INIT:
1669                 swi_shutdown();
1670
1671         case SRPC_STATE_EQ_INIT:
1672                 rc = LNetClearLazyPortal(SRPC_FRAMEWORK_REQUEST_PORTAL);
1673                 LASSERT (rc == 0);
1674                 rc = LNetEQFree(srpc_data.rpc_lnet_eq);
1675                 LASSERT (rc == 0); /* the EQ should have no user by now */
1676
1677         case SRPC_STATE_NI_INIT:
1678                 LNetNIFini();
1679                 break;
1680         }
1681
1682         /* srpc_peer_t's are kept in hash until shutdown */
1683         for (i = 0; i < SRPC_PEER_HASH_SIZE; i++) {
1684                 srpc_peer_t *peer;
1685
1686                 while (!list_empty(&srpc_data.rpc_peers[i])) {
1687                         peer = list_entry(srpc_data.rpc_peers[i].next,
1688                                           srpc_peer_t, stp_list);
1689                         list_del(&peer->stp_list);
1690
1691                         LASSERT (list_empty(&peer->stp_rpcq));
1692                         LASSERT (list_empty(&peer->stp_ctl_rpcq));
1693                         LASSERT (peer->stp_credits == SRPC_PEER_CREDITS);
1694
1695                         LIBCFS_FREE(peer, sizeof(srpc_peer_t));
1696                 }
1697         }
1698
1699         LIBCFS_FREE(srpc_data.rpc_peers,
1700                     sizeof(struct list_head) * SRPC_PEER_HASH_SIZE);
1701         return;
1702 }