lustre/ptlrpc/client.c

   1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
   2  * vim:expandtab:shiftwidth=8:tabstop=8:
   3  *
   4  *  Copyright (C) 2002 Cluster File Systems, Inc.
   5  *
   6  *   This file is part of Lustre, http://www.lustre.org.
   7  *
   8  *   Lustre is free software; you can redistribute it and/or
   9  *   modify it under the terms of version 2 of the GNU General Public
  10  *   License as published by the Free Software Foundation.
  11  *
  12  *   Lustre is distributed in the hope that it will be useful,
  13  *   but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  *   GNU General Public License for more details.
  16  *
  17  *   You should have received a copy of the GNU General Public License
  18  *   along with Lustre; if not, write to the Free Software
  19  *   Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  20  *
  21  */
  22
  23 #define DEBUG_SUBSYSTEM S_RPC
  24
  25 #include <linux/obd_support.h>
  26 #include <linux/lustre_lib.h>
  27 #include <linux/lustre_ha.h>
  28
  29 void ptlrpc_init_client(struct recovd_obd *recovd,
  30                         int (*recover)(struct ptlrpc_client *recover),
  31                         int req_portal,
  32                         int rep_portal, struct ptlrpc_client *cl)
  33 {
  34         memset(cl, 0, sizeof(*cl));
  35         cl->cli_recovd = recovd;
  36         cl->cli_recover = recover;
  37         if (recovd)
  38                 recovd_cli_manage(recovd, cl);
  39         cl->cli_obd = NULL;
  40         cl->cli_request_portal = req_portal;
  41         cl->cli_reply_portal = rep_portal;
  42         INIT_LIST_HEAD(&cl->cli_delayed_head);
  43         INIT_LIST_HEAD(&cl->cli_sending_head);
  44         INIT_LIST_HEAD(&cl->cli_dying_head);
  45         spin_lock_init(&cl->cli_lock);
  46         sema_init(&cl->cli_rpc_sem, 32);
  47 }
  48
  49 __u8 *ptlrpc_req_to_uuid(struct ptlrpc_request *req)
  50 {
  51         return req->rq_connection->c_remote_uuid;
  52 }
  53
  54 struct ptlrpc_connection *ptlrpc_uuid_to_connection(char *uuid)
  55 {
  56         struct ptlrpc_connection *c;
  57         struct lustre_peer peer;
  58         int err;
  59
  60         err = kportal_uuid_to_peer(uuid, &peer);
  61         if (err != 0) {
  62                 CERROR("cannot find peer %s!\n", uuid);
  63                 return NULL;
  64         }
  65
  66         c = ptlrpc_get_connection(&peer);
  67         if (c) {
  68                 memcpy(c->c_remote_uuid, uuid, sizeof(c->c_remote_uuid));
  69                 c->c_epoch++;
  70         }
  71
  72         return c;
  73 }
  74
  75 void ptlrpc_readdress_connection(struct ptlrpc_connection *conn, char *uuid)
  76 {
  77         struct lustre_peer peer;
  78         int err;
  79
  80         err = kportal_uuid_to_peer(uuid, &peer);
  81         if (err != 0) {
  82                 CERROR("cannot find peer %s!\n", uuid);
  83                 return;
  84         }
  85
  86         memcpy(&conn->c_peer, &peer, sizeof(peer));
  87         return;
  88 }
  89
  90 struct ptlrpc_bulk_desc *ptlrpc_prep_bulk(struct ptlrpc_connection *conn)
  91 {
  92         struct ptlrpc_bulk_desc *bulk;
  93
  94         OBD_ALLOC(bulk, sizeof(*bulk));
  95         if (bulk != NULL) {
  96                 bulk->b_connection = ptlrpc_connection_addref(conn);
  97                 atomic_set(&bulk->b_pages_remaining, 0);
  98                 init_waitqueue_head(&bulk->b_waitq);
  99                 INIT_LIST_HEAD(&bulk->b_page_list);
 100         }
 101
 102         return bulk;
 103 }
 104
 105 struct ptlrpc_bulk_page *ptlrpc_prep_bulk_page(struct ptlrpc_bulk_desc *desc)
 106 {
 107         struct ptlrpc_bulk_page *bulk;
 108
 109         OBD_ALLOC(bulk, sizeof(*bulk));
 110         if (bulk != NULL) {
 111                 bulk->b_desc = desc;
 112                 ptl_set_inv_handle(&bulk->b_md_h);
 113                 ptl_set_inv_handle(&bulk->b_me_h);
 114                 list_add_tail(&bulk->b_link, &desc->b_page_list);
 115                 desc->b_page_count++;
 116                 atomic_inc(&desc->b_pages_remaining);
 117         }
 118         return bulk;
 119 }
 120
 121 void ptlrpc_free_bulk(struct ptlrpc_bulk_desc *desc)
 122 {
 123         struct list_head *tmp, *next;
 124         ENTRY;
 125         if (desc == NULL) {
 126                 EXIT;
 127                 return;
 128         }
 129
 130         list_for_each_safe(tmp, next, &desc->b_page_list) {
 131                 struct ptlrpc_bulk_page *bulk;
 132                 bulk = list_entry(tmp, struct ptlrpc_bulk_page, b_link);
 133                 ptlrpc_free_bulk_page(bulk);
 134         }
 135
 136         ptlrpc_put_connection(desc->b_connection);
 137
 138         OBD_FREE(desc, sizeof(*desc));
 139         EXIT;
 140 }
 141
 142 void ptlrpc_free_bulk_page(struct ptlrpc_bulk_page *bulk)
 143 {
 144         ENTRY;
 145         if (bulk == NULL) {
 146                 EXIT;
 147                 return;
 148         }
 149
 150         list_del(&bulk->b_link);
 151         bulk->b_desc->b_page_count--;
 152         OBD_FREE(bulk, sizeof(*bulk));
 153         EXIT;
 154 }
 155
 156 struct ptlrpc_request *ptlrpc_prep_req(struct ptlrpc_client *cl,
 157                                        struct ptlrpc_connection *conn,
 158                                        int opcode, int count, int *lengths,
 159                                        char **bufs)
 160 {
 161         struct ptlrpc_request *request;
 162         int rc;
 163         ENTRY;
 164
 165         OBD_ALLOC(request, sizeof(*request));
 166         if (!request) {
 167                 CERROR("request allocation out of memory\n");
 168                 RETURN(NULL);
 169         }
 170
 171         rc = lustre_pack_msg(count, lengths, bufs,
 172                              &request->rq_reqlen, &request->rq_reqmsg);
 173         if (rc) {
 174                 CERROR("cannot pack request %d\n", rc);
 175                 OBD_FREE(request, sizeof(*request));
 176                 RETURN(NULL);
 177         }
 178
 179         request->rq_level = LUSTRE_CONN_FULL;
 180         request->rq_type = PTL_RPC_TYPE_REQUEST;
 181         request->rq_client = cl;
 182         request->rq_connection = ptlrpc_connection_addref(conn);
 183
 184         INIT_LIST_HEAD(&request->rq_list);
 185         INIT_LIST_HEAD(&request->rq_multi);
 186         /* this will be dec()d once in req_finished, once in free_committed */
 187         atomic_set(&request->rq_refcount, 2);
 188
 189         spin_lock(&conn->c_lock);
 190         request->rq_xid = HTON__u32(++conn->c_xid_out);
 191         spin_unlock(&conn->c_lock);
 192
 193         request->rq_reqmsg->magic = PTLRPC_MSG_MAGIC;
 194         request->rq_reqmsg->version = PTLRPC_MSG_VERSION;
 195         request->rq_reqmsg->opc = HTON__u32(opcode);
 196         request->rq_reqmsg->type = HTON__u32(PTL_RPC_MSG_REQUEST);
 197
 198         RETURN(request);
 199 }
 200 struct ptlrpc_request *ptlrpc_prep_req2(struct ptlrpc_client *cl,
 201                                         struct ptlrpc_connection *conn,
 202                                         struct lustre_handle *handle,
 203                                        int opcode, int count, int *lengths,
 204                                        char **bufs)
 205 {
 206         struct ptlrpc_request *req;
 207         req = ptlrpc_prep_req(cl, conn, opcode, count, lengths, bufs);
 208         ptlrpc_hdl2req(req, handle);
 209         return req;
 210 }
 211
 212 void ptlrpc_req_finished(struct ptlrpc_request *request)
 213 {
 214         if (request == NULL)
 215                 return;
 216
 217         if (request->rq_repmsg != NULL) {
 218                 OBD_FREE(request->rq_repmsg, request->rq_replen);
 219                 request->rq_repmsg = NULL;
 220                 request->rq_reply_md.start = NULL;
 221         }
 222
 223         if (atomic_dec_and_test(&request->rq_refcount))
 224                 ptlrpc_free_req(request);
 225 }
 226
 227 void ptlrpc_free_req(struct ptlrpc_request *request)
 228 {
 229         ENTRY;
 230         if (request == NULL) {
 231                 EXIT;
 232                 return;
 233         }
 234
 235         if (request->rq_repmsg != NULL)
 236                 OBD_FREE(request->rq_repmsg, request->rq_replen);
 237         if (request->rq_reqmsg != NULL)
 238                 OBD_FREE(request->rq_reqmsg, request->rq_reqlen);
 239
 240         if (request->rq_client) {
 241                 spin_lock(&request->rq_client->cli_lock);
 242                 list_del_init(&request->rq_list);
 243                 spin_unlock(&request->rq_client->cli_lock);
 244         }
 245
 246         ptlrpc_put_connection(request->rq_connection);
 247         list_del(&request->rq_multi);
 248         OBD_FREE(request, sizeof(*request));
 249         EXIT;
 250 }
 251
 252 static int ptlrpc_check_reply(struct ptlrpc_request *req)
 253 {
 254         int rc = 0;
 255
 256         if (req->rq_repmsg != NULL) {
 257                 req->rq_transno = NTOH__u64(req->rq_repmsg->transno);
 258                 req->rq_flags |= PTL_RPC_FL_REPLIED;
 259                 GOTO(out, rc = 1);
 260         }
 261
 262         if (req->rq_flags & PTL_RPC_FL_RESEND) {
 263                 CERROR("-- RESEND --\n");
 264                 GOTO(out, rc = 1);
 265         }
 266
 267         if (req->rq_flags & PTL_RPC_FL_RECOVERY) {
 268                 CERROR("-- RESTART --\n");
 269                 GOTO(out, rc = 1);
 270         }
 271
 272
 273         if (CURRENT_TIME - req->rq_time >= req->rq_timeout) {
 274                 CERROR("-- REQ TIMEOUT ON CONNID %d XID %Ld --\n",
 275                        req->rq_connid, (unsigned long long)req->rq_xid);
 276                 /* clear the timeout */
 277                 req->rq_timeout = 0;
 278                 req->rq_connection->c_level = LUSTRE_CONN_RECOVD;
 279                 req->rq_flags |= PTL_RPC_FL_TIMEOUT;
 280                 if (req->rq_client && req->rq_client->cli_recovd)
 281                         recovd_cli_fail(req->rq_client);
 282                 if (req->rq_level < LUSTRE_CONN_FULL)
 283                         rc = 1;
 284                 else
 285                         rc = 0;
 286                 GOTO(out, rc);
 287         }
 288
 289         if (req->rq_timeout) {
 290                 schedule_timeout(req->rq_timeout * HZ);
 291         }
 292
 293         if (l_killable_pending(current)) {
 294                 req->rq_flags |= PTL_RPC_FL_INTR;
 295                 GOTO(out, rc = 1);
 296         }
 297
 298  out:
 299         CDEBUG(D_NET, "req = %p, rc = %d\n", req, rc);
 300         return rc;
 301 }
 302
 303 int ptlrpc_check_status(struct ptlrpc_request *req, int err)
 304 {
 305         ENTRY;
 306
 307         if (err != 0) {
 308                 CERROR("err is %d\n", err);
 309                 RETURN(err);
 310         }
 311
 312         if (req == NULL) {
 313                 CERROR("req == NULL\n");
 314                 RETURN(-ENOMEM);
 315         }
 316
 317         if (req->rq_repmsg == NULL) {
 318                 CERROR("req->rq_repmsg == NULL\n");
 319                 RETURN(-ENOMEM);
 320         }
 321
 322         if (req->rq_repmsg->type == NTOH__u32(PTL_RPC_MSG_ERR)) {
 323                 CERROR("req->rq_repmsg->type == PTL_RPC_MSG_ERR\n");
 324                 RETURN(-EINVAL);
 325         }
 326
 327         if (req->rq_repmsg->status != 0) {
 328                 if (req->rq_repmsg->status < 0)
 329                         CERROR("req->rq_repmsg->status is %d\n",
 330                                req->rq_repmsg->status);
 331                 else
 332                         CDEBUG(D_INFO, "req->rq_repmsg->status is %d\n",
 333                                req->rq_repmsg->status);
 334                 /* XXX: translate this error from net to host */
 335                 RETURN(req->rq_repmsg->status);
 336         }
 337
 338         RETURN(0);
 339 }
 340
 341 static void ptlrpc_cleanup_request_buf(struct ptlrpc_request *request)
 342 {
 343         OBD_FREE(request->rq_reqmsg, request->rq_reqlen);
 344         request->rq_reqmsg = NULL;
 345         request->rq_reqlen = 0;
 346 }
 347
 348 /* Abort this request and cleanup any resources associated with it. */
 349 static int ptlrpc_abort(struct ptlrpc_request *request)
 350 {
 351         /* First remove the ME for the reply; in theory, this means
 352          * that we can tear down the buffer safely. */
 353         PtlMEUnlink(request->rq_reply_me_h);
 354         OBD_FREE(request->rq_reply_md.start, request->rq_replen);
 355         request->rq_repmsg = NULL;
 356         request->rq_replen = 0;
 357         return 0;
 358 }
 359
 360 /* caller must lock cli */
 361 void ptlrpc_free_committed(struct ptlrpc_client *cli)
 362 {
 363         struct list_head *tmp, *saved;
 364         struct ptlrpc_request *req;
 365
 366         list_for_each_safe(tmp, saved, &cli->cli_sending_head) {
 367                 req = list_entry(tmp, struct ptlrpc_request, rq_list);
 368
 369                 if ( (req->rq_flags & PTL_RPC_FL_REPLAY) ) {
 370                         CDEBUG(D_INFO, "Retaining request %Ld for replay\n",
 371                                req->rq_xid);
 372                         continue;
 373                 }
 374
 375                 /* not yet committed */
 376                 if (req->rq_transno > cli->cli_last_committed)
 377                         break;
 378
 379                 CDEBUG(D_INFO, "Marking request %Ld as committed ("
 380                        "transno=%Lu, last_committed=%Lu\n",
 381                        req->rq_xid, req->rq_transno,
 382                        cli->cli_last_committed);
 383                 if (atomic_dec_and_test(&req->rq_refcount)) {
 384                         /* we do this to prevent free_req deadlock */
 385                         list_del_init(&req->rq_list);
 386                         req->rq_client = NULL;
 387                         ptlrpc_free_req(req);
 388                 } else {
 389                         list_del_init(&req->rq_list);
 390                         list_add(&req->rq_list, &cli->cli_dying_head);
 391                 }
 392         }
 393
 394         EXIT;
 395         return;
 396 }
 397
 398 void ptlrpc_cleanup_client(struct ptlrpc_client *cli)
 399 {
 400         struct list_head *tmp, *saved;
 401         struct ptlrpc_request *req;
 402         ENTRY;
 403
 404         spin_lock(&cli->cli_lock);
 405         list_for_each_safe(tmp, saved, &cli->cli_sending_head) {
 406                 req = list_entry(tmp, struct ptlrpc_request, rq_list);
 407                 CDEBUG(D_INFO, "Cleaning req %p from sending list.\n", req);
 408                 list_del_init(&req->rq_list);
 409                 req->rq_client = NULL;
 410                 ptlrpc_free_req(req);
 411         }
 412         list_for_each_safe(tmp, saved, &cli->cli_dying_head) {
 413                 req = list_entry(tmp, struct ptlrpc_request, rq_list);
 414                 CERROR("Request %p is on the dying list at cleanup!\n", req);
 415                 list_del_init(&req->rq_list);
 416                 req->rq_client = NULL;
 417                 ptlrpc_free_req(req);
 418         }
 419         spin_unlock(&cli->cli_lock);
 420
 421         EXIT;
 422         return;
 423 }
 424
 425 void ptlrpc_continue_req(struct ptlrpc_request *req)
 426 {
 427         ENTRY;
 428         CDEBUG(D_INODE, "continue delayed request %Ld opc %d\n",
 429                req->rq_xid, req->rq_reqmsg->opc);
 430         wake_up(&req->rq_wait_for_rep);
 431         EXIT;
 432 }
 433
 434 void ptlrpc_resend_req(struct ptlrpc_request *req)
 435 {
 436         ENTRY;
 437         CDEBUG(D_INODE, "resend request %Ld, opc %d\n",
 438                req->rq_xid, req->rq_reqmsg->opc);
 439         req->rq_status = -EAGAIN;
 440         req->rq_level = LUSTRE_CONN_RECOVD;
 441         req->rq_flags |= PTL_RPC_FL_RESEND;
 442         req->rq_flags &= ~PTL_RPC_FL_TIMEOUT;
 443         wake_up(&req->rq_wait_for_rep);
 444         EXIT;
 445 }
 446
 447 void ptlrpc_restart_req(struct ptlrpc_request *req)
 448 {
 449         ENTRY;
 450         CDEBUG(D_INODE, "restart completed request %Ld, opc %d\n",
 451                req->rq_xid, req->rq_reqmsg->opc);
 452         req->rq_status = -ERESTARTSYS;
 453         req->rq_flags |= PTL_RPC_FL_RECOVERY;
 454         req->rq_flags &= ~PTL_RPC_FL_TIMEOUT;
 455         wake_up(&req->rq_wait_for_rep);
 456         EXIT;
 457 }
 458
 459 int ptlrpc_queue_wait(struct ptlrpc_request *req)
 460 {
 461         int rc = 0;
 462         struct ptlrpc_client *cli = req->rq_client;
 463         ENTRY;
 464
 465         init_waitqueue_head(&req->rq_wait_for_rep);
 466         CDEBUG(D_NET, "subsys: %s req %Ld opc %d level %d, conn level %d\n",
 467                cli->cli_name, req->rq_xid, req->rq_reqmsg->opc, req->rq_level,
 468                req->rq_connection->c_level);
 469
 470         /* XXX probably both an import and connection level are needed */
 471         if (req->rq_level > req->rq_connection->c_level) {
 472                 CERROR("process %d waiting for recovery (%d > %d)\n",
 473                        current->pid, req->rq_level, req->rq_connection->c_level);
 474                 spin_lock(&cli->cli_lock);
 475                 list_del_init(&req->rq_list);
 476                 list_add(&req->rq_list, cli->cli_delayed_head.prev);
 477                 spin_unlock(&cli->cli_lock);
 478                 l_wait_event_killable
 479                         (req->rq_wait_for_rep,
 480                          req->rq_level <= req->rq_connection->c_level);
 481                 spin_lock(&cli->cli_lock);
 482                 list_del_init(&req->rq_list);
 483                 spin_unlock(&cli->cli_lock);
 484                 CERROR("process %d resumed\n", current->pid);
 485         }
 486  resend:
 487         req->rq_time = CURRENT_TIME;
 488         req->rq_timeout = 100;
 489         rc = ptl_send_rpc(req);
 490         if (rc) {
 491                 CERROR("error %d, opcode %d\n", rc, req->rq_reqmsg->opc);
 492                 if ( rc > 0 )
 493                         rc = -rc;
 494                 ptlrpc_cleanup_request_buf(req);
 495                 up(&cli->cli_rpc_sem);
 496                 RETURN(-rc);
 497         }
 498
 499         spin_lock(&cli->cli_lock);
 500         list_del_init(&req->rq_list);
 501         list_add_tail(&req->rq_list, &cli->cli_sending_head);
 502         spin_unlock(&cli->cli_lock);
 503
 504         CDEBUG(D_OTHER, "-- sleeping\n");
 505         l_wait_event_killable(req->rq_wait_for_rep, ptlrpc_check_reply(req));
 506         CDEBUG(D_OTHER, "-- done\n");
 507
 508         if (req->rq_flags & PTL_RPC_FL_RESEND) {
 509                 req->rq_flags &= ~PTL_RPC_FL_RESEND;
 510                 goto resend;
 511         }
 512
 513         up(&cli->cli_rpc_sem);
 514         if (req->rq_flags & PTL_RPC_FL_TIMEOUT)
 515                 GOTO(out, rc = -ETIMEDOUT);
 516
 517         if (req->rq_flags & PTL_RPC_FL_INTR) {
 518                 /* Clean up the dangling reply buffers */
 519                 ptlrpc_abort(req);
 520                 GOTO(out, rc = -EINTR);
 521         }
 522
 523         if (!(req->rq_flags & PTL_RPC_FL_REPLIED))
 524                 GOTO(out, rc = req->rq_status);
 525
 526         rc = lustre_unpack_msg(req->rq_repmsg, req->rq_replen);
 527         if (rc) {
 528                 CERROR("unpack_rep failed: %d\n", rc);
 529                 GOTO(out, rc);
 530         }
 531         CDEBUG(D_NET, "got rep %Ld\n", req->rq_xid);
 532         if (req->rq_repmsg->status == 0)
 533                 CDEBUG(D_NET, "--> buf %p len %d status %d\n", req->rq_repmsg,
 534                        req->rq_replen, req->rq_repmsg->status);
 535
 536         spin_lock(&cli->cli_lock);
 537         cli->cli_last_rcvd = req->rq_repmsg->last_rcvd;
 538         cli->cli_last_committed = req->rq_repmsg->last_committed;
 539         ptlrpc_free_committed(cli);
 540         spin_unlock(&cli->cli_lock);
 541
 542         EXIT;
 543  out:
 544         return rc;
 545 }
 546
 547 int ptlrpc_replay_req(struct ptlrpc_request *req)
 548 {
 549         int rc = 0;
 550         struct ptlrpc_client *cli = req->rq_client;
 551         ENTRY;
 552
 553         init_waitqueue_head(&req->rq_wait_for_rep);
 554         CDEBUG(D_NET, "req %Ld opc %d level %d, conn level %d\n",
 555                req->rq_xid, req->rq_reqmsg->opc, req->rq_level,
 556                req->rq_connection->c_level);
 557
 558         req->rq_time = CURRENT_TIME;
 559         req->rq_timeout = 100;
 560         rc = ptl_send_rpc(req);
 561         if (rc) {
 562                 CERROR("error %d, opcode %d\n", rc, req->rq_reqmsg->opc);
 563                 ptlrpc_cleanup_request_buf(req);
 564                 up(&cli->cli_rpc_sem);
 565                 RETURN(-rc);
 566         }
 567
 568         CDEBUG(D_OTHER, "-- sleeping\n");
 569         l_wait_event_killable(req->rq_wait_for_rep, ptlrpc_check_reply(req));
 570         CDEBUG(D_OTHER, "-- done\n");
 571
 572         up(&cli->cli_rpc_sem);
 573
 574         if (!(req->rq_flags & PTL_RPC_FL_REPLIED)) {
 575                 CERROR("Unknown reason for wakeup\n");
 576                 /* XXX Phil - I end up here when I kill obdctl */
 577                 ptlrpc_abort(req);
 578                 GOTO(out, rc = -EINTR);
 579         }
 580
 581         rc = lustre_unpack_msg(req->rq_repmsg, req->rq_replen);
 582         if (rc) {
 583                 CERROR("unpack_rep failed: %d\n", rc);
 584                 GOTO(out, rc);
 585         }
 586
 587         CDEBUG(D_NET, "got rep %Ld\n", req->rq_xid);
 588         if (req->rq_repmsg->status == 0)
 589                 CDEBUG(D_NET, "--> buf %p len %d status %d\n", req->rq_repmsg,
 590                        req->rq_replen, req->rq_repmsg->status);
 591         else {
 592                 CERROR("recovery failed: ");
 593                 CERROR("req %Ld opc %d level %d, conn level %d\n",
 594                        req->rq_xid, req->rq_reqmsg->opc, req->rq_level,
 595                        req->rq_connection->c_level);
 596                 LBUG();
 597         }
 598
 599  out:
 600         RETURN(rc);
 601 }