lustre/ldlm/ldlm_flock.c

   1 /*
   2  * GPL HEADER START
   3  *
   4  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   5  *
   6  * This program is free software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License version 2 only,
   8  * as published by the Free Software Foundation.
   9  *
  10  * This program is distributed in the hope that it will be useful, but
  11  * WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  * General Public License version 2 for more details (a copy is included
  14  * in the LICENSE file that accompanied this code).
  15  *
  16  * You should have received a copy of the GNU General Public License
  17  * version 2 along with this program; If not, see
  18  * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
  19  *
  20  * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
  21  * CA 95054 USA or visit www.sun.com if you need additional information or
  22  * have any questions.
  23  *
  24  * GPL HEADER END
  25  */
  26 /*
  27  * Copyright (c) 2003 Hewlett-Packard Development Company LP.
  28  * Developed under the sponsorship of the US Government under
  29  * Subcontract No. B514193
  30  *
  31  * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
  32  * Use is subject to license terms.
  33  *
  34  * Copyright (c) 2010, 2012, Intel Corporation.
  35  */
  36 /*
  37  * This file is part of Lustre, http://www.lustre.org/
  38  * Lustre is a trademark of Sun Microsystems, Inc.
  39  */
  40
  41 /**
  42  * This file implements POSIX lock type for Lustre.
  43  * Its policy properties are start and end of extent and PID.
  44  *
  45  * These locks are only done through MDS due to POSIX semantics requiring
  46  * e.g. that locks could be only partially released and as such split into
  47  * two parts, and also that two adjacent locks from the same process may be
  48  * merged into a single wider lock.
  49  *
  50  * Lock modes are mapped like this:
  51  * PR and PW for READ and WRITE locks
  52  * NL to request a releasing of a portion of the lock
  53  *
  54  * These flock locks never timeout.
  55  */
  56
  57 #define DEBUG_SUBSYSTEM S_LDLM
  58
  59 #ifdef __KERNEL__
  60 #include <lustre_dlm.h>
  61 #include <obd_support.h>
  62 #include <obd_class.h>
  63 #include <lustre_lib.h>
  64 #include <libcfs/list.h>
  65 #else
  66 #include <liblustre.h>
  67 #include <obd_class.h>
  68 #endif
  69
  70 #include "ldlm_internal.h"
  71
  72 int ldlm_flock_blocking_ast(struct ldlm_lock *lock, struct ldlm_lock_desc *desc,
  73                             void *data, int flag);
  74
  75 /**
  76  * list_for_remaining_safe - iterate over the remaining entries in a list
  77  *              and safeguard against removal of a list entry.
  78  * \param pos   the &struct list_head to use as a loop counter. pos MUST
  79  *              have been initialized prior to using it in this macro.
  80  * \param n     another &struct list_head to use as temporary storage
  81  * \param head  the head for your list.
  82  */
  83 #define list_for_remaining_safe(pos, n, head) \
  84         for (n = pos->next; pos != (head); pos = n, n = pos->next)
  85
  86 static inline int
  87 ldlm_same_flock_owner(struct ldlm_lock *lock, struct ldlm_lock *new)
  88 {
  89         return((new->l_policy_data.l_flock.owner ==
  90                 lock->l_policy_data.l_flock.owner) &&
  91                (new->l_export == lock->l_export));
  92 }
  93
  94 static inline int
  95 ldlm_flocks_overlap(struct ldlm_lock *lock, struct ldlm_lock *new)
  96 {
  97         return((new->l_policy_data.l_flock.start <=
  98                 lock->l_policy_data.l_flock.end) &&
  99                (new->l_policy_data.l_flock.end >=
 100                 lock->l_policy_data.l_flock.start));
 101 }
 102
 103 static inline int ldlm_flock_blocking_link(struct ldlm_lock *req,
 104                                            struct ldlm_lock *lock)
 105 {
 106         int rc = 0;
 107
 108         /* For server only */
 109         if (req->l_export == NULL)
 110                 return 0;
 111
 112         if (unlikely(req->l_export->exp_flock_hash == NULL)) {
 113                 rc = ldlm_init_flock_export(req->l_export);
 114                 if (rc)
 115                         goto error;
 116         }
 117
 118         LASSERT(cfs_hlist_unhashed(&req->l_exp_flock_hash));
 119
 120         req->l_policy_data.l_flock.blocking_owner =
 121                 lock->l_policy_data.l_flock.owner;
 122         req->l_policy_data.l_flock.blocking_export =
 123                 lock->l_export;
 124         req->l_policy_data.l_flock.blocking_refs = 0;
 125
 126         cfs_hash_add(req->l_export->exp_flock_hash,
 127                      &req->l_policy_data.l_flock.owner,
 128                      &req->l_exp_flock_hash);
 129 error:
 130         return rc;
 131 }
 132
 133 static inline void ldlm_flock_blocking_unlink(struct ldlm_lock *req)
 134 {
 135         /* For server only */
 136         if (req->l_export == NULL)
 137                 return;
 138
 139         check_res_locked(req->l_resource);
 140         if (req->l_export->exp_flock_hash != NULL &&
 141             !cfs_hlist_unhashed(&req->l_exp_flock_hash))
 142                 cfs_hash_del(req->l_export->exp_flock_hash,
 143                              &req->l_policy_data.l_flock.owner,
 144                              &req->l_exp_flock_hash);
 145 }
 146
 147 static inline void
 148 ldlm_flock_destroy(struct ldlm_lock *lock, ldlm_mode_t mode, __u64 flags)
 149 {
 150         ENTRY;
 151
 152         LDLM_DEBUG(lock, "ldlm_flock_destroy(mode: %d, flags: 0x%llx)",
 153                    mode, flags);
 154
 155         /* Safe to not lock here, since it should be empty anyway */
 156         LASSERT(cfs_hlist_unhashed(&lock->l_exp_flock_hash));
 157
 158         cfs_list_del_init(&lock->l_res_link);
 159         if (flags == LDLM_FL_WAIT_NOREPROC &&
 160             !(lock->l_flags & LDLM_FL_FAILED)) {
 161                 /* client side - set a flag to prevent sending a CANCEL */
 162                 lock->l_flags |= LDLM_FL_LOCAL_ONLY | LDLM_FL_CBPENDING;
 163
 164                 /* when reaching here, it is under lock_res_and_lock(). Thus,
 165                    need call the nolock version of ldlm_lock_decref_internal*/
 166                 ldlm_lock_decref_internal_nolock(lock, mode);
 167         }
 168
 169         ldlm_lock_destroy_nolock(lock);
 170         EXIT;
 171 }
 172
 173 /**
 174  * POSIX locks deadlock detection code.
 175  *
 176  * Given a new lock \a req and an existing lock \a bl_lock it conflicts
 177  * with, we need to iterate through all blocked POSIX locks for this
 178  * export and see if there is a deadlock condition arising. (i.e. when
 179  * one client holds a lock on something and want a lock on something
 180  * else and at the same time another client has the opposite situation).
 181  */
 182 static int
 183 ldlm_flock_deadlock(struct ldlm_lock *req, struct ldlm_lock *bl_lock)
 184 {
 185         struct obd_export *req_exp = req->l_export;
 186         struct obd_export *bl_exp = bl_lock->l_export;
 187         __u64 req_owner = req->l_policy_data.l_flock.owner;
 188         __u64 bl_owner = bl_lock->l_policy_data.l_flock.owner;
 189
 190         /* For server only */
 191         if (req_exp == NULL)
 192                 return 0;
 193
 194         class_export_get(bl_exp);
 195         while (1) {
 196                 struct obd_export *bl_exp_new;
 197                 struct ldlm_lock *lock = NULL;
 198                 struct ldlm_flock *flock;
 199
 200                 if (bl_exp->exp_flock_hash != NULL)
 201                         lock = cfs_hash_lookup(bl_exp->exp_flock_hash,
 202                                                &bl_owner);
 203                 if (lock == NULL)
 204                         break;
 205
 206                 flock = &lock->l_policy_data.l_flock;
 207                 LASSERT(flock->owner == bl_owner);
 208                 bl_owner = flock->blocking_owner;
 209                 bl_exp_new = class_export_get(flock->blocking_export);
 210                 class_export_put(bl_exp);
 211
 212                 cfs_hash_put(bl_exp->exp_flock_hash, &lock->l_exp_flock_hash);
 213                 bl_exp = bl_exp_new;
 214
 215                 if (bl_owner == req_owner && bl_exp == req_exp) {
 216                         class_export_put(bl_exp);
 217                         return 1;
 218                 }
 219         }
 220         class_export_put(bl_exp);
 221
 222         return 0;
 223 }
 224
 225 /**
 226  * Process a granting attempt for flock lock.
 227  * Must be called under ns lock held.
 228  *
 229  * This function looks for any conflicts for \a lock in the granted or
 230  * waiting queues. The lock is granted if no conflicts are found in
 231  * either queue.
 232  *
 233  * It is also responsible for splitting a lock if a portion of the lock
 234  * is released.
 235  *
 236  * If \a first_enq is 0 (ie, called from ldlm_reprocess_queue):
 237  *   - blocking ASTs have already been sent
 238  *
 239  * If \a first_enq is 1 (ie, called from ldlm_lock_enqueue):
 240  *   - blocking ASTs have not been sent yet, so list of conflicting locks
 241  *     would be collected and ASTs sent.
 242  */
 243 int
 244 ldlm_process_flock_lock(struct ldlm_lock *req, __u64 *flags, int first_enq,
 245                         ldlm_error_t *err, cfs_list_t *work_list)
 246 {
 247         struct ldlm_resource *res = req->l_resource;
 248         struct ldlm_namespace *ns = ldlm_res_to_ns(res);
 249         cfs_list_t *tmp;
 250         cfs_list_t *ownlocks = NULL;
 251         struct ldlm_lock *lock = NULL;
 252         struct ldlm_lock *new = req;
 253         struct ldlm_lock *new2 = NULL;
 254         ldlm_mode_t mode = req->l_req_mode;
 255         int local = ns_is_client(ns);
 256         int added = (mode == LCK_NL);
 257         int overlaps = 0;
 258         int splitted = 0;
 259         const struct ldlm_callback_suite null_cbs = { NULL };
 260         int rc;
 261         ENTRY;
 262
 263         CDEBUG(D_DLMTRACE, "flags %#llx owner "LPU64" pid %u mode %u start "
 264                LPU64" end "LPU64"\n", *flags,
 265                new->l_policy_data.l_flock.owner,
 266                new->l_policy_data.l_flock.pid, mode,
 267                req->l_policy_data.l_flock.start,
 268                req->l_policy_data.l_flock.end);
 269
 270         *err = ELDLM_OK;
 271
 272         if (local) {
 273                 /* No blocking ASTs are sent to the clients for
 274                  * Posix file & record locks */
 275                 req->l_blocking_ast = NULL;
 276         } else {
 277                 /* Called on the server for lock cancels. */
 278                 req->l_blocking_ast = ldlm_flock_blocking_ast;
 279         }
 280
 281 reprocess:
 282         if ((*flags == LDLM_FL_WAIT_NOREPROC) || (mode == LCK_NL)) {
 283                 /* This loop determines where this processes locks start
 284                  * in the resource lr_granted list. */
 285                 cfs_list_for_each(tmp, &res->lr_granted) {
 286                         lock = cfs_list_entry(tmp, struct ldlm_lock,
 287                                               l_res_link);
 288                         if (ldlm_same_flock_owner(lock, req)) {
 289                                 ownlocks = tmp;
 290                                 break;
 291                         }
 292                 }
 293         } else {
 294                 lockmode_verify(mode);
 295
 296                 /* This loop determines if there are existing locks
 297                  * that conflict with the new lock request. */
 298                 cfs_list_for_each(tmp, &res->lr_granted) {
 299                         lock = cfs_list_entry(tmp, struct ldlm_lock,
 300                                               l_res_link);
 301
 302                         if (ldlm_same_flock_owner(lock, req)) {
 303                                 if (!ownlocks)
 304                                         ownlocks = tmp;
 305                                 continue;
 306                         }
 307
 308                         /* locks are compatible, overlap doesn't matter */
 309                         if (lockmode_compat(lock->l_granted_mode, mode))
 310                                 continue;
 311
 312                         if (!ldlm_flocks_overlap(lock, req))
 313                                 continue;
 314
 315                         if (!first_enq)
 316                                 RETURN(LDLM_ITER_CONTINUE);
 317
 318                         if (*flags & LDLM_FL_BLOCK_NOWAIT) {
 319                                 ldlm_flock_destroy(req, mode, *flags);
 320                                 *err = -EAGAIN;
 321                                 RETURN(LDLM_ITER_STOP);
 322                         }
 323
 324                         if (*flags & LDLM_FL_TEST_LOCK) {
 325                                 ldlm_flock_destroy(req, mode, *flags);
 326                                 req->l_req_mode = lock->l_granted_mode;
 327                                 req->l_policy_data.l_flock.pid =
 328                                         lock->l_policy_data.l_flock.pid;
 329                                 req->l_policy_data.l_flock.start =
 330                                         lock->l_policy_data.l_flock.start;
 331                                 req->l_policy_data.l_flock.end =
 332                                         lock->l_policy_data.l_flock.end;
 333                                 *flags |= LDLM_FL_LOCK_CHANGED;
 334                                 RETURN(LDLM_ITER_STOP);
 335                         }
 336
 337                         if (ldlm_flock_deadlock(req, lock)) {
 338                                 ldlm_flock_destroy(req, mode, *flags);
 339                                 *err = -EDEADLK;
 340                                 RETURN(LDLM_ITER_STOP);
 341                         }
 342
 343                         rc = ldlm_flock_blocking_link(req, lock);
 344                         if (rc) {
 345                                 ldlm_flock_destroy(req, mode, *flags);
 346                                 *err = rc;
 347                                 RETURN(LDLM_ITER_STOP);
 348                         }
 349                         ldlm_resource_add_lock(res, &res->lr_waiting, req);
 350                         *flags |= LDLM_FL_BLOCK_GRANTED;
 351                         RETURN(LDLM_ITER_STOP);
 352                 }
 353         }
 354
 355         if (*flags & LDLM_FL_TEST_LOCK) {
 356                 ldlm_flock_destroy(req, mode, *flags);
 357                 req->l_req_mode = LCK_NL;
 358                 *flags |= LDLM_FL_LOCK_CHANGED;
 359                 RETURN(LDLM_ITER_STOP);
 360         }
 361
 362         /* In case we had slept on this lock request take it off of the
 363          * deadlock detection hash list. */
 364         ldlm_flock_blocking_unlink(req);
 365
 366         /* Scan the locks owned by this process that overlap this request.
 367          * We may have to merge or split existing locks. */
 368
 369         if (!ownlocks)
 370                 ownlocks = &res->lr_granted;
 371
 372         list_for_remaining_safe(ownlocks, tmp, &res->lr_granted) {
 373                 lock = cfs_list_entry(ownlocks, struct ldlm_lock, l_res_link);
 374
 375                 if (!ldlm_same_flock_owner(lock, new))
 376                         break;
 377
 378                 if (lock->l_granted_mode == mode) {
 379                         /* If the modes are the same then we need to process
 380                          * locks that overlap OR adjoin the new lock. The extra
 381                          * logic condition is necessary to deal with arithmetic
 382                          * overflow and underflow. */
 383                         if ((new->l_policy_data.l_flock.start >
 384                              (lock->l_policy_data.l_flock.end + 1))
 385                             && (lock->l_policy_data.l_flock.end !=
 386                                 OBD_OBJECT_EOF))
 387                                 continue;
 388
 389                         if ((new->l_policy_data.l_flock.end <
 390                              (lock->l_policy_data.l_flock.start - 1))
 391                             && (lock->l_policy_data.l_flock.start != 0))
 392                                 break;
 393
 394                         if (new->l_policy_data.l_flock.start <
 395                             lock->l_policy_data.l_flock.start) {
 396                                 lock->l_policy_data.l_flock.start =
 397                                         new->l_policy_data.l_flock.start;
 398                         } else {
 399                                 new->l_policy_data.l_flock.start =
 400                                         lock->l_policy_data.l_flock.start;
 401                         }
 402
 403                         if (new->l_policy_data.l_flock.end >
 404                             lock->l_policy_data.l_flock.end) {
 405                                 lock->l_policy_data.l_flock.end =
 406                                         new->l_policy_data.l_flock.end;
 407                         } else {
 408                                 new->l_policy_data.l_flock.end =
 409                                         lock->l_policy_data.l_flock.end;
 410                         }
 411
 412                         if (added) {
 413                                 ldlm_flock_destroy(lock, mode, *flags);
 414                         } else {
 415                                 new = lock;
 416                                 added = 1;
 417                         }
 418                         continue;
 419                 }
 420
 421                 if (new->l_policy_data.l_flock.start >
 422                     lock->l_policy_data.l_flock.end)
 423                         continue;
 424
 425                 if (new->l_policy_data.l_flock.end <
 426                     lock->l_policy_data.l_flock.start)
 427                         break;
 428
 429                 ++overlaps;
 430
 431                 if (new->l_policy_data.l_flock.start <=
 432                     lock->l_policy_data.l_flock.start) {
 433                         if (new->l_policy_data.l_flock.end <
 434                             lock->l_policy_data.l_flock.end) {
 435                                 lock->l_policy_data.l_flock.start =
 436                                         new->l_policy_data.l_flock.end + 1;
 437                                 break;
 438                         }
 439                         ldlm_flock_destroy(lock, lock->l_req_mode, *flags);
 440                         continue;
 441                 }
 442                 if (new->l_policy_data.l_flock.end >=
 443                     lock->l_policy_data.l_flock.end) {
 444                         lock->l_policy_data.l_flock.end =
 445                                 new->l_policy_data.l_flock.start - 1;
 446                         continue;
 447                 }
 448
 449                 /* split the existing lock into two locks */
 450
 451                 /* if this is an F_UNLCK operation then we could avoid
 452                  * allocating a new lock and use the req lock passed in
 453                  * with the request but this would complicate the reply
 454                  * processing since updates to req get reflected in the
 455                  * reply. The client side replays the lock request so
 456                  * it must see the original lock data in the reply. */
 457
 458                 /* XXX - if ldlm_lock_new() can sleep we should
 459                  * release the lr_lock, allocate the new lock,
 460                  * and restart processing this lock. */
 461                 if (!new2) {
 462                         unlock_res_and_lock(req);
 463                         new2 = ldlm_lock_create(ns, &res->lr_name, LDLM_FLOCK,
 464                                                 lock->l_granted_mode, &null_cbs,
 465                                                 NULL, 0, LVB_T_NONE);
 466                         lock_res_and_lock(req);
 467                         if (!new2) {
 468                                 ldlm_flock_destroy(req, lock->l_granted_mode,
 469                                                    *flags);
 470                                 *err = -ENOLCK;
 471                                 RETURN(LDLM_ITER_STOP);
 472                         }
 473                         goto reprocess;
 474                 }
 475
 476                 splitted = 1;
 477
 478                 new2->l_granted_mode = lock->l_granted_mode;
 479                 new2->l_policy_data.l_flock.pid =
 480                         new->l_policy_data.l_flock.pid;
 481                 new2->l_policy_data.l_flock.owner =
 482                         new->l_policy_data.l_flock.owner;
 483                 new2->l_policy_data.l_flock.start =
 484                         lock->l_policy_data.l_flock.start;
 485                 new2->l_policy_data.l_flock.end =
 486                         new->l_policy_data.l_flock.start - 1;
 487                 lock->l_policy_data.l_flock.start =
 488                         new->l_policy_data.l_flock.end + 1;
 489                 new2->l_conn_export = lock->l_conn_export;
 490                 if (lock->l_export != NULL) {
 491                         new2->l_export = class_export_lock_get(lock->l_export, new2);
 492                         if (new2->l_export->exp_lock_hash &&
 493                             cfs_hlist_unhashed(&new2->l_exp_hash))
 494                                 cfs_hash_add(new2->l_export->exp_lock_hash,
 495                                              &new2->l_remote_handle,
 496                                              &new2->l_exp_hash);
 497                 }
 498                 if (*flags == LDLM_FL_WAIT_NOREPROC)
 499                         ldlm_lock_addref_internal_nolock(new2,
 500                                                          lock->l_granted_mode);
 501
 502                 /* insert new2 at lock */
 503                 ldlm_resource_add_lock(res, ownlocks, new2);
 504                 LDLM_LOCK_RELEASE(new2);
 505                 break;
 506         }
 507
 508         /* if new2 is created but never used, destroy it*/
 509         if (splitted == 0 && new2 != NULL)
 510                 ldlm_lock_destroy_nolock(new2);
 511
 512         /* At this point we're granting the lock request. */
 513         req->l_granted_mode = req->l_req_mode;
 514
 515         /* Add req to the granted queue before calling ldlm_reprocess_all(). */
 516         if (!added) {
 517                 cfs_list_del_init(&req->l_res_link);
 518                 /* insert new lock before ownlocks in list. */
 519                 ldlm_resource_add_lock(res, ownlocks, req);
 520         }
 521
 522         if (*flags != LDLM_FL_WAIT_NOREPROC) {
 523 #ifdef HAVE_SERVER_SUPPORT
 524                 if (first_enq) {
 525                         /* If this is an unlock, reprocess the waitq and
 526                          * send completions ASTs for locks that can now be
 527                          * granted. The only problem with doing this
 528                          * reprocessing here is that the completion ASTs for
 529                          * newly granted locks will be sent before the unlock
 530                          * completion is sent. It shouldn't be an issue. Also
 531                          * note that ldlm_process_flock_lock() will recurse,
 532                          * but only once because first_enq will be false from
 533                          * ldlm_reprocess_queue. */
 534                         if ((mode == LCK_NL) && overlaps) {
 535                                 CFS_LIST_HEAD(rpc_list);
 536                                 int rc;
 537 restart:
 538                                 ldlm_reprocess_queue(res, &res->lr_waiting,
 539                                                      &rpc_list);
 540
 541                                 unlock_res_and_lock(req);
 542                                 rc = ldlm_run_ast_work(ns, &rpc_list,
 543                                                        LDLM_WORK_CP_AST);
 544                                 lock_res_and_lock(req);
 545                                 if (rc == -ERESTART)
 546                                         GOTO(restart, -ERESTART);
 547                        }
 548                 } else {
 549                         LASSERT(req->l_completion_ast);
 550                         ldlm_add_ast_work_item(req, NULL, work_list);
 551                 }
 552 #else /* !HAVE_SERVER_SUPPORT */
 553                 /* The only one possible case for client-side calls flock
 554                  * policy function is ldlm_flock_completion_ast inside which
 555                  * carries LDLM_FL_WAIT_NOREPROC flag. */
 556                 CERROR("Illegal parameter for client-side-only module.\n");
 557                 LBUG();
 558 #endif /* HAVE_SERVER_SUPPORT */
 559         }
 560
 561         /* In case we're reprocessing the requested lock we can't destroy
 562          * it until after calling ldlm_add_ast_work_item() above so that laawi()
 563          * can bump the reference count on \a req. Otherwise \a req
 564          * could be freed before the completion AST can be sent.  */
 565         if (added)
 566                 ldlm_flock_destroy(req, mode, *flags);
 567
 568         ldlm_resource_dump(D_INFO, res);
 569         RETURN(LDLM_ITER_CONTINUE);
 570 }
 571
 572 struct ldlm_flock_wait_data {
 573         struct ldlm_lock *fwd_lock;
 574         int               fwd_generation;
 575 };
 576
 577 static void
 578 ldlm_flock_interrupted_wait(void *data)
 579 {
 580         struct ldlm_lock *lock;
 581         ENTRY;
 582
 583         lock = ((struct ldlm_flock_wait_data *)data)->fwd_lock;
 584
 585         /* take lock off the deadlock detection hash list. */
 586         lock_res_and_lock(lock);
 587         ldlm_flock_blocking_unlink(lock);
 588
 589         /* client side - set flag to prevent lock from being put on LRU list */
 590         lock->l_flags |= LDLM_FL_CBPENDING;
 591         unlock_res_and_lock(lock);
 592
 593         EXIT;
 594 }
 595
 596 /**
 597  * Flock completion callback function.
 598  *
 599  * \param lock [in,out]: A lock to be handled
 600  * \param flags    [in]: flags
 601  * \param *data    [in]: ldlm_work_cp_ast_lock() will use ldlm_cb_set_arg
 602  *
 603  * \retval 0    : success
 604  * \retval <0   : failure
 605  */
 606 int
 607 ldlm_flock_completion_ast(struct ldlm_lock *lock, __u64 flags, void *data)
 608 {
 609         struct file_lock                *getlk = lock->l_ast_data;
 610         struct obd_device              *obd;
 611         struct obd_import              *imp = NULL;
 612         struct ldlm_flock_wait_data     fwd;
 613         struct l_wait_info              lwi;
 614         ldlm_error_t                    err;
 615         int                             rc = 0;
 616         ENTRY;
 617
 618         CDEBUG(D_DLMTRACE, "flags: 0x%llx data: %p getlk: %p\n",
 619                flags, data, getlk);
 620
 621         /* Import invalidation. We need to actually release the lock
 622          * references being held, so that it can go away. No point in
 623          * holding the lock even if app still believes it has it, since
 624          * server already dropped it anyway. Only for granted locks too. */
 625         if ((lock->l_flags & (LDLM_FL_FAILED|LDLM_FL_LOCAL_ONLY)) ==
 626             (LDLM_FL_FAILED|LDLM_FL_LOCAL_ONLY)) {
 627                 if (lock->l_req_mode == lock->l_granted_mode &&
 628                     lock->l_granted_mode != LCK_NL &&
 629                     NULL == data)
 630                         ldlm_lock_decref_internal(lock, lock->l_req_mode);
 631
 632                 /* Need to wake up the waiter if we were evicted */
 633                 cfs_waitq_signal(&lock->l_waitq);
 634                 RETURN(0);
 635         }
 636
 637         LASSERT(flags != LDLM_FL_WAIT_NOREPROC);
 638
 639         if (!(flags & (LDLM_FL_BLOCK_WAIT | LDLM_FL_BLOCK_GRANTED |
 640                        LDLM_FL_BLOCK_CONV))) {
 641                 if (NULL == data)
 642                         /* mds granted the lock in the reply */
 643                         goto granted;
 644                 /* CP AST RPC: lock get granted, wake it up */
 645                 cfs_waitq_signal(&lock->l_waitq);
 646                 RETURN(0);
 647         }
 648
 649         LDLM_DEBUG(lock, "client-side enqueue returned a blocked lock, "
 650                    "sleeping");
 651         fwd.fwd_lock = lock;
 652         obd = class_exp2obd(lock->l_conn_export);
 653
 654         /* if this is a local lock, there is no import */
 655         if (NULL != obd)
 656                 imp = obd->u.cli.cl_import;
 657
 658         if (NULL != imp) {
 659                 spin_lock(&imp->imp_lock);
 660                 fwd.fwd_generation = imp->imp_generation;
 661                 spin_unlock(&imp->imp_lock);
 662         }
 663
 664         lwi = LWI_TIMEOUT_INTR(0, NULL, ldlm_flock_interrupted_wait, &fwd);
 665
 666         /* Go to sleep until the lock is granted. */
 667         rc = l_wait_event(lock->l_waitq, is_granted_or_cancelled(lock), &lwi);
 668
 669         if (rc) {
 670                 LDLM_DEBUG(lock, "client-side enqueue waking up: failed (%d)",
 671                            rc);
 672                 RETURN(rc);
 673         }
 674
 675 granted:
 676         OBD_FAIL_TIMEOUT(OBD_FAIL_LDLM_CP_CB_WAIT, 10);
 677
 678         if (lock->l_flags & LDLM_FL_DESTROYED) {
 679                 LDLM_DEBUG(lock, "client-side enqueue waking up: destroyed");
 680                 RETURN(0);
 681         }
 682
 683         if (lock->l_flags & LDLM_FL_FAILED) {
 684                 LDLM_DEBUG(lock, "client-side enqueue waking up: failed");
 685                 RETURN(-EIO);
 686         }
 687
 688         if (rc) {
 689                 LDLM_DEBUG(lock, "client-side enqueue waking up: failed (%d)",
 690                            rc);
 691                 RETURN(rc);
 692         }
 693
 694         LDLM_DEBUG(lock, "client-side enqueue granted");
 695
 696         lock_res_and_lock(lock);
 697
 698         /* take lock off the deadlock detection hash list. */
 699         ldlm_flock_blocking_unlink(lock);
 700
 701         /* ldlm_lock_enqueue() has already placed lock on the granted list. */
 702         cfs_list_del_init(&lock->l_res_link);
 703
 704         if (flags & LDLM_FL_TEST_LOCK) {
 705                 /* fcntl(F_GETLK) request */
 706                 /* The old mode was saved in getlk->fl_type so that if the mode
 707                  * in the lock changes we can decref the appropriate refcount.*/
 708                 ldlm_flock_destroy(lock, flock_type(getlk),
 709                                    LDLM_FL_WAIT_NOREPROC);
 710                 switch (lock->l_granted_mode) {
 711                 case LCK_PR:
 712                         flock_set_type(getlk, F_RDLCK);
 713                         break;
 714                 case LCK_PW:
 715                         flock_set_type(getlk, F_WRLCK);
 716                         break;
 717                 default:
 718                         flock_set_type(getlk, F_UNLCK);
 719                 }
 720                 flock_set_pid(getlk, (pid_t)lock->l_policy_data.l_flock.pid);
 721                 flock_set_start(getlk,
 722                                 (loff_t)lock->l_policy_data.l_flock.start);
 723                 flock_set_end(getlk,
 724                               (loff_t)lock->l_policy_data.l_flock.end);
 725         } else {
 726                 __u64 noreproc = LDLM_FL_WAIT_NOREPROC;
 727
 728                 /* We need to reprocess the lock to do merges or splits
 729                  * with existing locks owned by this process. */
 730                 ldlm_process_flock_lock(lock, &noreproc, 1, &err, NULL);
 731         }
 732         unlock_res_and_lock(lock);
 733         RETURN(0);
 734 }
 735 EXPORT_SYMBOL(ldlm_flock_completion_ast);
 736
 737 int ldlm_flock_blocking_ast(struct ldlm_lock *lock, struct ldlm_lock_desc *desc,
 738                             void *data, int flag)
 739 {
 740         ENTRY;
 741
 742         LASSERT(lock);
 743         LASSERT(flag == LDLM_CB_CANCELING);
 744
 745         /* take lock off the deadlock detection hash list. */
 746         lock_res_and_lock(lock);
 747         ldlm_flock_blocking_unlink(lock);
 748         unlock_res_and_lock(lock);
 749         RETURN(0);
 750 }
 751
 752 void ldlm_flock_policy_wire18_to_local(const ldlm_wire_policy_data_t *wpolicy,
 753                                        ldlm_policy_data_t *lpolicy)
 754 {
 755         memset(lpolicy, 0, sizeof(*lpolicy));
 756         lpolicy->l_flock.start = wpolicy->l_flock.lfw_start;
 757         lpolicy->l_flock.end = wpolicy->l_flock.lfw_end;
 758         lpolicy->l_flock.pid = wpolicy->l_flock.lfw_pid;
 759         /* Compat code, old clients had no idea about owner field and
 760          * relied solely on pid for ownership. Introduced in LU-104, 2.1,
 761          * April 2011 */
 762         lpolicy->l_flock.owner = wpolicy->l_flock.lfw_pid;
 763 }
 764
 765
 766 void ldlm_flock_policy_wire21_to_local(const ldlm_wire_policy_data_t *wpolicy,
 767                                        ldlm_policy_data_t *lpolicy)
 768 {
 769         memset(lpolicy, 0, sizeof(*lpolicy));
 770         lpolicy->l_flock.start = wpolicy->l_flock.lfw_start;
 771         lpolicy->l_flock.end = wpolicy->l_flock.lfw_end;
 772         lpolicy->l_flock.pid = wpolicy->l_flock.lfw_pid;
 773         lpolicy->l_flock.owner = wpolicy->l_flock.lfw_owner;
 774 }
 775
 776 void ldlm_flock_policy_local_to_wire(const ldlm_policy_data_t *lpolicy,
 777                                      ldlm_wire_policy_data_t *wpolicy)
 778 {
 779         memset(wpolicy, 0, sizeof(*wpolicy));
 780         wpolicy->l_flock.lfw_start = lpolicy->l_flock.start;
 781         wpolicy->l_flock.lfw_end = lpolicy->l_flock.end;
 782         wpolicy->l_flock.lfw_pid = lpolicy->l_flock.pid;
 783         wpolicy->l_flock.lfw_owner = lpolicy->l_flock.owner;
 784 }
 785
 786 /*
 787  * Export handle<->flock hash operations.
 788  */
 789 static unsigned
 790 ldlm_export_flock_hash(cfs_hash_t *hs, const void *key, unsigned mask)
 791 {
 792         return cfs_hash_u64_hash(*(__u64 *)key, mask);
 793 }
 794
 795 static void *
 796 ldlm_export_flock_key(cfs_hlist_node_t *hnode)
 797 {
 798         struct ldlm_lock *lock;
 799
 800         lock = cfs_hlist_entry(hnode, struct ldlm_lock, l_exp_flock_hash);
 801         return &lock->l_policy_data.l_flock.owner;
 802 }
 803
 804 static int
 805 ldlm_export_flock_keycmp(const void *key, cfs_hlist_node_t *hnode)
 806 {
 807         return !memcmp(ldlm_export_flock_key(hnode), key, sizeof(__u64));
 808 }
 809
 810 static void *
 811 ldlm_export_flock_object(cfs_hlist_node_t *hnode)
 812 {
 813         return cfs_hlist_entry(hnode, struct ldlm_lock, l_exp_flock_hash);
 814 }
 815
 816 static void
 817 ldlm_export_flock_get(cfs_hash_t *hs, cfs_hlist_node_t *hnode)
 818 {
 819         struct ldlm_lock *lock;
 820         struct ldlm_flock *flock;
 821
 822         lock = cfs_hlist_entry(hnode, struct ldlm_lock, l_exp_flock_hash);
 823         LDLM_LOCK_GET(lock);
 824
 825         flock = &lock->l_policy_data.l_flock;
 826         LASSERT(flock->blocking_export != NULL);
 827         class_export_get(flock->blocking_export);
 828         flock->blocking_refs++;
 829 }
 830
 831 static void
 832 ldlm_export_flock_put(cfs_hash_t *hs, cfs_hlist_node_t *hnode)
 833 {
 834         struct ldlm_lock *lock;
 835         struct ldlm_flock *flock;
 836
 837         lock = cfs_hlist_entry(hnode, struct ldlm_lock, l_exp_flock_hash);
 838         LDLM_LOCK_RELEASE(lock);
 839
 840         flock = &lock->l_policy_data.l_flock;
 841         LASSERT(flock->blocking_export != NULL);
 842         class_export_put(flock->blocking_export);
 843         if (--flock->blocking_refs == 0) {
 844                 flock->blocking_owner = 0;
 845                 flock->blocking_export = NULL;
 846         }
 847 }
 848
 849 static cfs_hash_ops_t ldlm_export_flock_ops = {
 850         .hs_hash        = ldlm_export_flock_hash,
 851         .hs_key         = ldlm_export_flock_key,
 852         .hs_keycmp      = ldlm_export_flock_keycmp,
 853         .hs_object      = ldlm_export_flock_object,
 854         .hs_get         = ldlm_export_flock_get,
 855         .hs_put         = ldlm_export_flock_put,
 856         .hs_put_locked  = ldlm_export_flock_put,
 857 };
 858
 859 int ldlm_init_flock_export(struct obd_export *exp)
 860 {
 861         exp->exp_flock_hash =
 862                 cfs_hash_create(obd_uuid2str(&exp->exp_client_uuid),
 863                                 HASH_EXP_LOCK_CUR_BITS,
 864                                 HASH_EXP_LOCK_MAX_BITS,
 865                                 HASH_EXP_LOCK_BKT_BITS, 0,
 866                                 CFS_HASH_MIN_THETA, CFS_HASH_MAX_THETA,
 867                                 &ldlm_export_flock_ops,
 868                                 CFS_HASH_DEFAULT | CFS_HASH_NBLK_CHANGE);
 869         if (!exp->exp_flock_hash)
 870                 RETURN(-ENOMEM);
 871
 872         RETURN(0);
 873 }
 874 EXPORT_SYMBOL(ldlm_init_flock_export);
 875
 876 void ldlm_destroy_flock_export(struct obd_export *exp)
 877 {
 878         ENTRY;
 879         if (exp->exp_flock_hash) {
 880                 cfs_hash_putref(exp->exp_flock_hash);
 881                 exp->exp_flock_hash = NULL;
 882         }
 883         EXIT;
 884 }
 885 EXPORT_SYMBOL(ldlm_destroy_flock_export);