fs/dlm/recover.c

   1 /******************************************************************************
   2 *******************************************************************************
   3 **
   4 **  Copyright (C) Sistina Software, Inc.  1997-2003  All rights reserved.
   5 **  Copyright (C) 2004-2005 Red Hat, Inc.  All rights reserved.
   6 **
   7 **  This copyrighted material is made available to anyone wishing to use,
   8 **  modify, copy, or redistribute it subject to the terms and conditions
   9 **  of the GNU General Public License v.2.
  10 **
  11 *******************************************************************************
  12 ******************************************************************************/
  13
  14 #include "dlm_internal.h"
  15 #include "lockspace.h"
  16 #include "dir.h"
  17 #include "config.h"
  18 #include "ast.h"
  19 #include "memory.h"
  20 #include "rcom.h"
  21 #include "lock.h"
  22 #include "lowcomms.h"
  23 #include "member.h"
  24 #include "recover.h"
  25
  26
  27 /*
  28  * Recovery waiting routines: these functions wait for a particular reply from
  29  * a remote node, or for the remote node to report a certain status.  They need
  30  * to abort if the lockspace is stopped indicating a node has failed (perhaps
  31  * the one being waited for).
  32  */
  33
  34 /*
  35  * Wait until given function returns non-zero or lockspace is stopped
  36  * (LS_RECOVERY_STOP set due to failure of a node in ls_nodes).  When another
  37  * function thinks it could have completed the waited-on task, they should wake
  38  * up ls_wait_general to get an immediate response rather than waiting for the
  39  * timeout.  This uses a timeout so it can check periodically if the wait
  40  * should abort due to node failure (which doesn't cause a wake_up).
  41  * This should only be called by the dlm_recoverd thread.
  42  */
  43
  44 int dlm_wait_function(struct dlm_ls *ls, int (*testfn) (struct dlm_ls *ls))
  45 {
  46         int error = 0;
  47         int rv;
  48
  49         while (1) {
  50                 rv = wait_event_timeout(ls->ls_wait_general,
  51                                         testfn(ls) || dlm_recovery_stopped(ls),
  52                                         dlm_config.ci_recover_timer * HZ);
  53                 if (rv)
  54                         break;
  55                 if (test_bit(LSFL_RCOM_WAIT, &ls->ls_flags)) {
  56                         log_debug(ls, "dlm_wait_function timed out");
  57                         return -ETIMEDOUT;
  58                 }
  59         }
  60
  61         if (dlm_recovery_stopped(ls)) {
  62                 log_debug(ls, "dlm_wait_function aborted");
  63                 error = -EINTR;
  64         }
  65         return error;
  66 }
  67
  68 /*
  69  * An efficient way for all nodes to wait for all others to have a certain
  70  * status.  The node with the lowest nodeid polls all the others for their
  71  * status (wait_status_all) and all the others poll the node with the low id
  72  * for its accumulated result (wait_status_low).  When all nodes have set
  73  * status flag X, then status flag X_ALL will be set on the low nodeid.
  74  */
  75
  76 uint32_t dlm_recover_status(struct dlm_ls *ls)
  77 {
  78         uint32_t status;
  79         spin_lock(&ls->ls_recover_lock);
  80         status = ls->ls_recover_status;
  81         spin_unlock(&ls->ls_recover_lock);
  82         return status;
  83 }
  84
  85 static void _set_recover_status(struct dlm_ls *ls, uint32_t status)
  86 {
  87         ls->ls_recover_status |= status;
  88 }
  89
  90 void dlm_set_recover_status(struct dlm_ls *ls, uint32_t status)
  91 {
  92         spin_lock(&ls->ls_recover_lock);
  93         _set_recover_status(ls, status);
  94         spin_unlock(&ls->ls_recover_lock);
  95 }
  96
  97 static int wait_status_all(struct dlm_ls *ls, uint32_t wait_status,
  98                            int save_slots)
  99 {
 100         struct dlm_rcom *rc = ls->ls_recover_buf;
 101         struct dlm_member *memb;
 102         int error = 0, delay;
 103
 104         list_for_each_entry(memb, &ls->ls_nodes, list) {
 105                 delay = 0;
 106                 for (;;) {
 107                         if (dlm_recovery_stopped(ls)) {
 108                                 error = -EINTR;
 109                                 goto out;
 110                         }
 111
 112                         error = dlm_rcom_status(ls, memb->nodeid, 0);
 113                         if (error)
 114                                 goto out;
 115
 116                         if (save_slots)
 117                                 dlm_slot_save(ls, rc, memb);
 118
 119                         if (rc->rc_result & wait_status)
 120                                 break;
 121                         if (delay < 1000)
 122                                 delay += 20;
 123                         msleep(delay);
 124                 }
 125         }
 126  out:
 127         return error;
 128 }
 129
 130 static int wait_status_low(struct dlm_ls *ls, uint32_t wait_status,
 131                            uint32_t status_flags)
 132 {
 133         struct dlm_rcom *rc = ls->ls_recover_buf;
 134         int error = 0, delay = 0, nodeid = ls->ls_low_nodeid;
 135
 136         for (;;) {
 137                 if (dlm_recovery_stopped(ls)) {
 138                         error = -EINTR;
 139                         goto out;
 140                 }
 141
 142                 error = dlm_rcom_status(ls, nodeid, status_flags);
 143                 if (error)
 144                         break;
 145
 146                 if (rc->rc_result & wait_status)
 147                         break;
 148                 if (delay < 1000)
 149                         delay += 20;
 150                 msleep(delay);
 151         }
 152  out:
 153         return error;
 154 }
 155
 156 static int wait_status(struct dlm_ls *ls, uint32_t status)
 157 {
 158         uint32_t status_all = status << 1;
 159         int error;
 160
 161         if (ls->ls_low_nodeid == dlm_our_nodeid()) {
 162                 error = wait_status_all(ls, status, 0);
 163                 if (!error)
 164                         dlm_set_recover_status(ls, status_all);
 165         } else
 166                 error = wait_status_low(ls, status_all, 0);
 167
 168         return error;
 169 }
 170
 171 int dlm_recover_members_wait(struct dlm_ls *ls)
 172 {
 173         struct dlm_member *memb;
 174         struct dlm_slot *slots;
 175         int num_slots, slots_size;
 176         int error, rv;
 177         uint32_t gen;
 178
 179         list_for_each_entry(memb, &ls->ls_nodes, list) {
 180                 memb->slot = -1;
 181                 memb->generation = 0;
 182         }
 183
 184         if (ls->ls_low_nodeid == dlm_our_nodeid()) {
 185                 error = wait_status_all(ls, DLM_RS_NODES, 1);
 186                 if (error)
 187                         goto out;
 188
 189                 /* slots array is sparse, slots_size may be > num_slots */
 190
 191                 rv = dlm_slots_assign(ls, &num_slots, &slots_size, &slots, &gen);
 192                 if (!rv) {
 193                         spin_lock(&ls->ls_recover_lock);
 194                         _set_recover_status(ls, DLM_RS_NODES_ALL);
 195                         ls->ls_num_slots = num_slots;
 196                         ls->ls_slots_size = slots_size;
 197                         ls->ls_slots = slots;
 198                         ls->ls_generation = gen;
 199                         spin_unlock(&ls->ls_recover_lock);
 200                 } else {
 201                         dlm_set_recover_status(ls, DLM_RS_NODES_ALL);
 202                 }
 203         } else {
 204                 error = wait_status_low(ls, DLM_RS_NODES_ALL, DLM_RSF_NEED_SLOTS);
 205                 if (error)
 206                         goto out;
 207
 208                 dlm_slots_copy_in(ls);
 209         }
 210  out:
 211         return error;
 212 }
 213
 214 int dlm_recover_directory_wait(struct dlm_ls *ls)
 215 {
 216         return wait_status(ls, DLM_RS_DIR);
 217 }
 218
 219 int dlm_recover_locks_wait(struct dlm_ls *ls)
 220 {
 221         return wait_status(ls, DLM_RS_LOCKS);
 222 }
 223
 224 int dlm_recover_done_wait(struct dlm_ls *ls)
 225 {
 226         return wait_status(ls, DLM_RS_DONE);
 227 }
 228
 229 /*
 230  * The recover_list contains all the rsb's for which we've requested the new
 231  * master nodeid.  As replies are returned from the resource directories the
 232  * rsb's are removed from the list.  When the list is empty we're done.
 233  *
 234  * The recover_list is later similarly used for all rsb's for which we've sent
 235  * new lkb's and need to receive new corresponding lkid's.
 236  *
 237  * We use the address of the rsb struct as a simple local identifier for the
 238  * rsb so we can match an rcom reply with the rsb it was sent for.
 239  */
 240
 241 static int recover_list_empty(struct dlm_ls *ls)
 242 {
 243         int empty;
 244
 245         spin_lock(&ls->ls_recover_list_lock);
 246         empty = list_empty(&ls->ls_recover_list);
 247         spin_unlock(&ls->ls_recover_list_lock);
 248
 249         return empty;
 250 }
 251
 252 static void recover_list_add(struct dlm_rsb *r)
 253 {
 254         struct dlm_ls *ls = r->res_ls;
 255
 256         spin_lock(&ls->ls_recover_list_lock);
 257         if (list_empty(&r->res_recover_list)) {
 258                 list_add_tail(&r->res_recover_list, &ls->ls_recover_list);
 259                 ls->ls_recover_list_count++;
 260                 dlm_hold_rsb(r);
 261         }
 262         spin_unlock(&ls->ls_recover_list_lock);
 263 }
 264
 265 static void recover_list_del(struct dlm_rsb *r)
 266 {
 267         struct dlm_ls *ls = r->res_ls;
 268
 269         spin_lock(&ls->ls_recover_list_lock);
 270         list_del_init(&r->res_recover_list);
 271         ls->ls_recover_list_count--;
 272         spin_unlock(&ls->ls_recover_list_lock);
 273
 274         dlm_put_rsb(r);
 275 }
 276
 277 static void recover_list_clear(struct dlm_ls *ls)
 278 {
 279         struct dlm_rsb *r, *s;
 280
 281         spin_lock(&ls->ls_recover_list_lock);
 282         list_for_each_entry_safe(r, s, &ls->ls_recover_list, res_recover_list) {
 283                 list_del_init(&r->res_recover_list);
 284                 r->res_recover_locks_count = 0;
 285                 dlm_put_rsb(r);
 286                 ls->ls_recover_list_count--;
 287         }
 288
 289         if (ls->ls_recover_list_count != 0) {
 290                 log_error(ls, "warning: recover_list_count %d",
 291                           ls->ls_recover_list_count);
 292                 ls->ls_recover_list_count = 0;
 293         }
 294         spin_unlock(&ls->ls_recover_list_lock);
 295 }
 296
 297 static int recover_idr_empty(struct dlm_ls *ls)
 298 {
 299         int empty = 1;
 300
 301         spin_lock(&ls->ls_recover_idr_lock);
 302         if (ls->ls_recover_list_count)
 303                 empty = 0;
 304         spin_unlock(&ls->ls_recover_idr_lock);
 305
 306         return empty;
 307 }
 308
 309 static int recover_idr_add(struct dlm_rsb *r)
 310 {
 311         struct dlm_ls *ls = r->res_ls;
 312         int rv;
 313
 314         idr_preload(GFP_NOFS);
 315         spin_lock(&ls->ls_recover_idr_lock);
 316         if (r->res_id) {
 317                 rv = -1;
 318                 goto out_unlock;
 319         }
 320         rv = idr_alloc(&ls->ls_recover_idr, r, 1, 0, GFP_NOWAIT);
 321         if (rv < 0)
 322                 goto out_unlock;
 323
 324         r->res_id = rv;
 325         ls->ls_recover_list_count++;
 326         dlm_hold_rsb(r);
 327         rv = 0;
 328 out_unlock:
 329         spin_unlock(&ls->ls_recover_idr_lock);
 330         idr_preload_end();
 331         return rv;
 332 }
 333
 334 static void recover_idr_del(struct dlm_rsb *r)
 335 {
 336         struct dlm_ls *ls = r->res_ls;
 337
 338         spin_lock(&ls->ls_recover_idr_lock);
 339         idr_remove(&ls->ls_recover_idr, r->res_id);
 340         r->res_id = 0;
 341         ls->ls_recover_list_count--;
 342         spin_unlock(&ls->ls_recover_idr_lock);
 343
 344         dlm_put_rsb(r);
 345 }
 346
 347 static struct dlm_rsb *recover_idr_find(struct dlm_ls *ls, uint64_t id)
 348 {
 349         struct dlm_rsb *r;
 350
 351         spin_lock(&ls->ls_recover_idr_lock);
 352         r = idr_find(&ls->ls_recover_idr, (int)id);
 353         spin_unlock(&ls->ls_recover_idr_lock);
 354         return r;
 355 }
 356
 357 static void recover_idr_clear(struct dlm_ls *ls)
 358 {
 359         struct dlm_rsb *r;
 360         int id;
 361
 362         spin_lock(&ls->ls_recover_idr_lock);
 363
 364         idr_for_each_entry(&ls->ls_recover_idr, r, id) {
 365                 idr_remove(&ls->ls_recover_idr, id);
 366                 r->res_id = 0;
 367                 r->res_recover_locks_count = 0;
 368                 ls->ls_recover_list_count--;
 369
 370                 dlm_put_rsb(r);
 371         }
 372
 373         if (ls->ls_recover_list_count != 0) {
 374                 log_error(ls, "warning: recover_list_count %d",
 375                           ls->ls_recover_list_count);
 376                 ls->ls_recover_list_count = 0;
 377         }
 378         spin_unlock(&ls->ls_recover_idr_lock);
 379 }
 380
 381
 382 /* Master recovery: find new master node for rsb's that were
 383    mastered on nodes that have been removed.
 384
 385    dlm_recover_masters
 386    recover_master
 387    dlm_send_rcom_lookup            ->  receive_rcom_lookup
 388                                        dlm_dir_lookup
 389    receive_rcom_lookup_reply       <-
 390    dlm_recover_master_reply
 391    set_new_master
 392    set_master_lkbs
 393    set_lock_master
 394 */
 395
 396 /*
 397  * Set the lock master for all LKBs in a lock queue
 398  * If we are the new master of the rsb, we may have received new
 399  * MSTCPY locks from other nodes already which we need to ignore
 400  * when setting the new nodeid.
 401  */
 402
 403 static void set_lock_master(struct list_head *queue, int nodeid)
 404 {
 405         struct dlm_lkb *lkb;
 406
 407         list_for_each_entry(lkb, queue, lkb_statequeue) {
 408                 if (!(lkb->lkb_flags & DLM_IFL_MSTCPY)) {
 409                         lkb->lkb_nodeid = nodeid;
 410                         lkb->lkb_remid = 0;
 411                 }
 412         }
 413 }
 414
 415 static void set_master_lkbs(struct dlm_rsb *r)
 416 {
 417         set_lock_master(&r->res_grantqueue, r->res_nodeid);
 418         set_lock_master(&r->res_convertqueue, r->res_nodeid);
 419         set_lock_master(&r->res_waitqueue, r->res_nodeid);
 420 }
 421
 422 /*
 423  * Propagate the new master nodeid to locks
 424  * The NEW_MASTER flag tells dlm_recover_locks() which rsb's to consider.
 425  * The NEW_MASTER2 flag tells recover_lvb() and recover_grant() which
 426  * rsb's to consider.
 427  */
 428
 429 static void set_new_master(struct dlm_rsb *r)
 430 {
 431         set_master_lkbs(r);
 432         rsb_set_flag(r, RSB_NEW_MASTER);
 433         rsb_set_flag(r, RSB_NEW_MASTER2);
 434 }
 435
 436 /*
 437  * We do async lookups on rsb's that need new masters.  The rsb's
 438  * waiting for a lookup reply are kept on the recover_list.
 439  *
 440  * Another node recovering the master may have sent us a rcom lookup,
 441  * and our dlm_master_lookup() set it as the new master, along with
 442  * NEW_MASTER so that we'll recover it here (this implies dir_nodeid
 443  * equals our_nodeid below).
 444  */
 445
 446 static int recover_master(struct dlm_rsb *r, unsigned int *count)
 447 {
 448         struct dlm_ls *ls = r->res_ls;
 449         int our_nodeid, dir_nodeid;
 450         int is_removed = 0;
 451         int error;
 452
 453         if (is_master(r))
 454                 return 0;
 455
 456         is_removed = dlm_is_removed(ls, r->res_nodeid);
 457
 458         if (!is_removed && !rsb_flag(r, RSB_NEW_MASTER))
 459                 return 0;
 460
 461         our_nodeid = dlm_our_nodeid();
 462         dir_nodeid = dlm_dir_nodeid(r);
 463
 464         if (dir_nodeid == our_nodeid) {
 465                 if (is_removed) {
 466                         r->res_master_nodeid = our_nodeid;
 467                         r->res_nodeid = 0;
 468                 }
 469
 470                 /* set master of lkbs to ourself when is_removed, or to
 471                    another new master which we set along with NEW_MASTER
 472                    in dlm_master_lookup */
 473                 set_new_master(r);
 474                 error = 0;
 475         } else {
 476                 recover_idr_add(r);
 477                 error = dlm_send_rcom_lookup(r, dir_nodeid);
 478         }
 479
 480         (*count)++;
 481         return error;
 482 }
 483
 484 /*
 485  * All MSTCPY locks are purged and rebuilt, even if the master stayed the same.
 486  * This is necessary because recovery can be started, aborted and restarted,
 487  * causing the master nodeid to briefly change during the aborted recovery, and
 488  * change back to the original value in the second recovery.  The MSTCPY locks
 489  * may or may not have been purged during the aborted recovery.  Another node
 490  * with an outstanding request in waiters list and a request reply saved in the
 491  * requestqueue, cannot know whether it should ignore the reply and resend the
 492  * request, or accept the reply and complete the request.  It must do the
 493  * former if the remote node purged MSTCPY locks, and it must do the later if
 494  * the remote node did not.  This is solved by always purging MSTCPY locks, in
 495  * which case, the request reply would always be ignored and the request
 496  * resent.
 497  */
 498
 499 static int recover_master_static(struct dlm_rsb *r, unsigned int *count)
 500 {
 501         int dir_nodeid = dlm_dir_nodeid(r);
 502         int new_master = dir_nodeid;
 503
 504         if (dir_nodeid == dlm_our_nodeid())
 505                 new_master = 0;
 506
 507         dlm_purge_mstcpy_locks(r);
 508         r->res_master_nodeid = dir_nodeid;
 509         r->res_nodeid = new_master;
 510         set_new_master(r);
 511         (*count)++;
 512         return 0;
 513 }
 514
 515 /*
 516  * Go through local root resources and for each rsb which has a master which
 517  * has departed, get the new master nodeid from the directory.  The dir will
 518  * assign mastery to the first node to look up the new master.  That means
 519  * we'll discover in this lookup if we're the new master of any rsb's.
 520  *
 521  * We fire off all the dir lookup requests individually and asynchronously to
 522  * the correct dir node.
 523  */
 524
 525 int dlm_recover_masters(struct dlm_ls *ls)
 526 {
 527         struct dlm_rsb *r;
 528         unsigned int total = 0;
 529         unsigned int count = 0;
 530         int nodir = dlm_no_directory(ls);
 531         int error;
 532
 533         log_rinfo(ls, "dlm_recover_masters");
 534
 535         down_read(&ls->ls_root_sem);
 536         list_for_each_entry(r, &ls->ls_root_list, res_root_list) {
 537                 if (dlm_recovery_stopped(ls)) {
 538                         up_read(&ls->ls_root_sem);
 539                         error = -EINTR;
 540                         goto out;
 541                 }
 542
 543                 lock_rsb(r);
 544                 if (nodir)
 545                         error = recover_master_static(r, &count);
 546                 else
 547                         error = recover_master(r, &count);
 548                 unlock_rsb(r);
 549                 cond_resched();
 550                 total++;
 551
 552                 if (error) {
 553                         up_read(&ls->ls_root_sem);
 554                         goto out;
 555                 }
 556         }
 557         up_read(&ls->ls_root_sem);
 558
 559         log_rinfo(ls, "dlm_recover_masters %u of %u", count, total);
 560
 561         error = dlm_wait_function(ls, &recover_idr_empty);
 562  out:
 563         if (error)
 564                 recover_idr_clear(ls);
 565         return error;
 566 }
 567
 568 int dlm_recover_master_reply(struct dlm_ls *ls, struct dlm_rcom *rc)
 569 {
 570         struct dlm_rsb *r;
 571         int ret_nodeid, new_master;
 572
 573         r = recover_idr_find(ls, rc->rc_id);
 574         if (!r) {
 575                 log_error(ls, "dlm_recover_master_reply no id %llx",
 576                           (unsigned long long)rc->rc_id);
 577                 goto out;
 578         }
 579
 580         ret_nodeid = rc->rc_result;
 581
 582         if (ret_nodeid == dlm_our_nodeid())
 583                 new_master = 0;
 584         else
 585                 new_master = ret_nodeid;
 586
 587         lock_rsb(r);
 588         r->res_master_nodeid = ret_nodeid;
 589         r->res_nodeid = new_master;
 590         set_new_master(r);
 591         unlock_rsb(r);
 592         recover_idr_del(r);
 593
 594         if (recover_idr_empty(ls))
 595                 wake_up(&ls->ls_wait_general);
 596  out:
 597         return 0;
 598 }
 599
 600
 601 /* Lock recovery: rebuild the process-copy locks we hold on a
 602    remastered rsb on the new rsb master.
 603
 604    dlm_recover_locks
 605    recover_locks
 606    recover_locks_queue
 607    dlm_send_rcom_lock              ->  receive_rcom_lock
 608                                        dlm_recover_master_copy
 609    receive_rcom_lock_reply         <-
 610    dlm_recover_process_copy
 611 */
 612
 613
 614 /*
 615  * keep a count of the number of lkb's we send to the new master; when we get
 616  * an equal number of replies then recovery for the rsb is done
 617  */
 618
 619 static int recover_locks_queue(struct dlm_rsb *r, struct list_head *head)
 620 {
 621         struct dlm_lkb *lkb;
 622         int error = 0;
 623
 624         list_for_each_entry(lkb, head, lkb_statequeue) {
 625                 error = dlm_send_rcom_lock(r, lkb);
 626                 if (error)
 627                         break;
 628                 r->res_recover_locks_count++;
 629         }
 630
 631         return error;
 632 }
 633
 634 static int recover_locks(struct dlm_rsb *r)
 635 {
 636         int error = 0;
 637
 638         lock_rsb(r);
 639
 640         DLM_ASSERT(!r->res_recover_locks_count, dlm_dump_rsb(r););
 641
 642         error = recover_locks_queue(r, &r->res_grantqueue);
 643         if (error)
 644                 goto out;
 645         error = recover_locks_queue(r, &r->res_convertqueue);
 646         if (error)
 647                 goto out;
 648         error = recover_locks_queue(r, &r->res_waitqueue);
 649         if (error)
 650                 goto out;
 651
 652         if (r->res_recover_locks_count)
 653                 recover_list_add(r);
 654         else
 655                 rsb_clear_flag(r, RSB_NEW_MASTER);
 656  out:
 657         unlock_rsb(r);
 658         return error;
 659 }
 660
 661 int dlm_recover_locks(struct dlm_ls *ls)
 662 {
 663         struct dlm_rsb *r;
 664         int error, count = 0;
 665
 666         down_read(&ls->ls_root_sem);
 667         list_for_each_entry(r, &ls->ls_root_list, res_root_list) {
 668                 if (is_master(r)) {
 669                         rsb_clear_flag(r, RSB_NEW_MASTER);
 670                         continue;
 671                 }
 672
 673                 if (!rsb_flag(r, RSB_NEW_MASTER))
 674                         continue;
 675
 676                 if (dlm_recovery_stopped(ls)) {
 677                         error = -EINTR;
 678                         up_read(&ls->ls_root_sem);
 679                         goto out;
 680                 }
 681
 682                 error = recover_locks(r);
 683                 if (error) {
 684                         up_read(&ls->ls_root_sem);
 685                         goto out;
 686                 }
 687
 688                 count += r->res_recover_locks_count;
 689         }
 690         up_read(&ls->ls_root_sem);
 691
 692         log_rinfo(ls, "dlm_recover_locks %d out", count);
 693
 694         error = dlm_wait_function(ls, &recover_list_empty);
 695  out:
 696         if (error)
 697                 recover_list_clear(ls);
 698         return error;
 699 }
 700
 701 void dlm_recovered_lock(struct dlm_rsb *r)
 702 {
 703         DLM_ASSERT(rsb_flag(r, RSB_NEW_MASTER), dlm_dump_rsb(r););
 704
 705         r->res_recover_locks_count--;
 706         if (!r->res_recover_locks_count) {
 707                 rsb_clear_flag(r, RSB_NEW_MASTER);
 708                 recover_list_del(r);
 709         }
 710
 711         if (recover_list_empty(r->res_ls))
 712                 wake_up(&r->res_ls->ls_wait_general);
 713 }
 714
 715 /*
 716  * The lvb needs to be recovered on all master rsb's.  This includes setting
 717  * the VALNOTVALID flag if necessary, and determining the correct lvb contents
 718  * based on the lvb's of the locks held on the rsb.
 719  *
 720  * RSB_VALNOTVALID is set in two cases:
 721  *
 722  * 1. we are master, but not new, and we purged an EX/PW lock held by a
 723  * failed node (in dlm_recover_purge which set RSB_RECOVER_LVB_INVAL)
 724  *
 725  * 2. we are a new master, and there are only NL/CR locks left.
 726  * (We could probably improve this by only invaliding in this way when
 727  * the previous master left uncleanly.  VMS docs mention that.)
 728  *
 729  * The LVB contents are only considered for changing when this is a new master
 730  * of the rsb (NEW_MASTER2).  Then, the rsb's lvb is taken from any lkb with
 731  * mode > CR.  If no lkb's exist with mode above CR, the lvb contents are taken
 732  * from the lkb with the largest lvb sequence number.
 733  */
 734
 735 static void recover_lvb(struct dlm_rsb *r)
 736 {
 737         struct dlm_lkb *lkb, *high_lkb = NULL;
 738         uint32_t high_seq = 0;
 739         int lock_lvb_exists = 0;
 740         int big_lock_exists = 0;
 741         int lvblen = r->res_ls->ls_lvblen;
 742
 743         if (!rsb_flag(r, RSB_NEW_MASTER2) &&
 744             rsb_flag(r, RSB_RECOVER_LVB_INVAL)) {
 745                 /* case 1 above */
 746                 rsb_set_flag(r, RSB_VALNOTVALID);
 747                 return;
 748         }
 749
 750         if (!rsb_flag(r, RSB_NEW_MASTER2))
 751                 return;
 752
 753         /* we are the new master, so figure out if VALNOTVALID should
 754            be set, and set the rsb lvb from the best lkb available. */
 755
 756         list_for_each_entry(lkb, &r->res_grantqueue, lkb_statequeue) {
 757                 if (!(lkb->lkb_exflags & DLM_LKF_VALBLK))
 758                         continue;
 759
 760                 lock_lvb_exists = 1;
 761
 762                 if (lkb->lkb_grmode > DLM_LOCK_CR) {
 763                         big_lock_exists = 1;
 764                         goto setflag;
 765                 }
 766
 767                 if (((int)lkb->lkb_lvbseq - (int)high_seq) >= 0) {
 768                         high_lkb = lkb;
 769                         high_seq = lkb->lkb_lvbseq;
 770                 }
 771         }
 772
 773         list_for_each_entry(lkb, &r->res_convertqueue, lkb_statequeue) {
 774                 if (!(lkb->lkb_exflags & DLM_LKF_VALBLK))
 775                         continue;
 776
 777                 lock_lvb_exists = 1;
 778
 779                 if (lkb->lkb_grmode > DLM_LOCK_CR) {
 780                         big_lock_exists = 1;
 781                         goto setflag;
 782                 }
 783
 784                 if (((int)lkb->lkb_lvbseq - (int)high_seq) >= 0) {
 785                         high_lkb = lkb;
 786                         high_seq = lkb->lkb_lvbseq;
 787                 }
 788         }
 789
 790  setflag:
 791         if (!lock_lvb_exists)
 792                 goto out;
 793
 794         /* lvb is invalidated if only NL/CR locks remain */
 795         if (!big_lock_exists)
 796                 rsb_set_flag(r, RSB_VALNOTVALID);
 797
 798         if (!r->res_lvbptr) {
 799                 r->res_lvbptr = dlm_allocate_lvb(r->res_ls);
 800                 if (!r->res_lvbptr)
 801                         goto out;
 802         }
 803
 804         if (big_lock_exists) {
 805                 r->res_lvbseq = lkb->lkb_lvbseq;
 806                 memcpy(r->res_lvbptr, lkb->lkb_lvbptr, lvblen);
 807         } else if (high_lkb) {
 808                 r->res_lvbseq = high_lkb->lkb_lvbseq;
 809                 memcpy(r->res_lvbptr, high_lkb->lkb_lvbptr, lvblen);
 810         } else {
 811                 r->res_lvbseq = 0;
 812                 memset(r->res_lvbptr, 0, lvblen);
 813         }
 814  out:
 815         return;
 816 }
 817
 818 /* All master rsb's flagged RECOVER_CONVERT need to be looked at.  The locks
 819    converting PR->CW or CW->PR need to have their lkb_grmode set. */
 820
 821 static void recover_conversion(struct dlm_rsb *r)
 822 {
 823         struct dlm_ls *ls = r->res_ls;
 824         struct dlm_lkb *lkb;
 825         int grmode = -1;
 826
 827         list_for_each_entry(lkb, &r->res_grantqueue, lkb_statequeue) {
 828                 if (lkb->lkb_grmode == DLM_LOCK_PR ||
 829                     lkb->lkb_grmode == DLM_LOCK_CW) {
 830                         grmode = lkb->lkb_grmode;
 831                         break;
 832                 }
 833         }
 834
 835         list_for_each_entry(lkb, &r->res_convertqueue, lkb_statequeue) {
 836                 if (lkb->lkb_grmode != DLM_LOCK_IV)
 837                         continue;
 838                 if (grmode == -1) {
 839                         log_debug(ls, "recover_conversion %x set gr to rq %d",
 840                                   lkb->lkb_id, lkb->lkb_rqmode);
 841                         lkb->lkb_grmode = lkb->lkb_rqmode;
 842                 } else {
 843                         log_debug(ls, "recover_conversion %x set gr %d",
 844                                   lkb->lkb_id, grmode);
 845                         lkb->lkb_grmode = grmode;
 846                 }
 847         }
 848 }
 849
 850 /* We've become the new master for this rsb and waiting/converting locks may
 851    need to be granted in dlm_recover_grant() due to locks that may have
 852    existed from a removed node. */
 853
 854 static void recover_grant(struct dlm_rsb *r)
 855 {
 856         if (!list_empty(&r->res_waitqueue) || !list_empty(&r->res_convertqueue))
 857                 rsb_set_flag(r, RSB_RECOVER_GRANT);
 858 }
 859
 860 void dlm_recover_rsbs(struct dlm_ls *ls)
 861 {
 862         struct dlm_rsb *r;
 863         unsigned int count = 0;
 864
 865         down_read(&ls->ls_root_sem);
 866         list_for_each_entry(r, &ls->ls_root_list, res_root_list) {
 867                 lock_rsb(r);
 868                 if (is_master(r)) {
 869                         if (rsb_flag(r, RSB_RECOVER_CONVERT))
 870                                 recover_conversion(r);
 871
 872                         /* recover lvb before granting locks so the updated
 873                            lvb/VALNOTVALID is presented in the completion */
 874                         recover_lvb(r);
 875
 876                         if (rsb_flag(r, RSB_NEW_MASTER2))
 877                                 recover_grant(r);
 878                         count++;
 879                 } else {
 880                         rsb_clear_flag(r, RSB_VALNOTVALID);
 881                 }
 882                 rsb_clear_flag(r, RSB_RECOVER_CONVERT);
 883                 rsb_clear_flag(r, RSB_RECOVER_LVB_INVAL);
 884                 rsb_clear_flag(r, RSB_NEW_MASTER2);
 885                 unlock_rsb(r);
 886         }
 887         up_read(&ls->ls_root_sem);
 888
 889         if (count)
 890                 log_rinfo(ls, "dlm_recover_rsbs %d done", count);
 891 }
 892
 893 /* Create a single list of all root rsb's to be used during recovery */
 894
 895 int dlm_create_root_list(struct dlm_ls *ls)
 896 {
 897         struct rb_node *n;
 898         struct dlm_rsb *r;
 899         int i, error = 0;
 900
 901         down_write(&ls->ls_root_sem);
 902         if (!list_empty(&ls->ls_root_list)) {
 903                 log_error(ls, "root list not empty");
 904                 error = -EINVAL;
 905                 goto out;
 906         }
 907
 908         for (i = 0; i < ls->ls_rsbtbl_size; i++) {
 909                 spin_lock(&ls->ls_rsbtbl[i].lock);
 910                 for (n = rb_first(&ls->ls_rsbtbl[i].keep); n; n = rb_next(n)) {
 911                         r = rb_entry(n, struct dlm_rsb, res_hashnode);
 912                         list_add(&r->res_root_list, &ls->ls_root_list);
 913                         dlm_hold_rsb(r);
 914                 }
 915
 916                 if (!RB_EMPTY_ROOT(&ls->ls_rsbtbl[i].toss))
 917                         log_error(ls, "dlm_create_root_list toss not empty");
 918                 spin_unlock(&ls->ls_rsbtbl[i].lock);
 919         }
 920  out:
 921         up_write(&ls->ls_root_sem);
 922         return error;
 923 }
 924
 925 void dlm_release_root_list(struct dlm_ls *ls)
 926 {
 927         struct dlm_rsb *r, *safe;
 928
 929         down_write(&ls->ls_root_sem);
 930         list_for_each_entry_safe(r, safe, &ls->ls_root_list, res_root_list) {
 931                 list_del_init(&r->res_root_list);
 932                 dlm_put_rsb(r);
 933         }
 934         up_write(&ls->ls_root_sem);
 935 }
 936
 937 void dlm_clear_toss(struct dlm_ls *ls)
 938 {
 939         struct rb_node *n, *next;
 940         struct dlm_rsb *r;
 941         unsigned int count = 0;
 942         int i;
 943
 944         for (i = 0; i < ls->ls_rsbtbl_size; i++) {
 945                 spin_lock(&ls->ls_rsbtbl[i].lock);
 946                 for (n = rb_first(&ls->ls_rsbtbl[i].toss); n; n = next) {
 947                         next = rb_next(n);
 948                         r = rb_entry(n, struct dlm_rsb, res_hashnode);
 949                         rb_erase(n, &ls->ls_rsbtbl[i].toss);
 950                         dlm_free_rsb(r);
 951                         count++;
 952                 }
 953                 spin_unlock(&ls->ls_rsbtbl[i].lock);
 954         }
 955
 956         if (count)
 957                 log_rinfo(ls, "dlm_clear_toss %u done", count);
 958 }
 959