4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
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.
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).
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
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
27 * Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved.
28 * Use is subject to license terms.
30 * Copyright (c) 2011, 2012, Whamcloud, Inc.
33 * This file is part of Lustre, http://www.lustre.org/
34 * Lustre is a trademark of Sun Microsystems, Inc.
36 * lustre/ldlm/ldlm_lock.c
38 * Author: Peter Braam <braam@clusterfs.com>
39 * Author: Phil Schwan <phil@clusterfs.com>
42 #define DEBUG_SUBSYSTEM S_LDLM
45 # include <libcfs/libcfs.h>
46 # include <linux/lustre_intent.h>
48 # include <liblustre.h>
51 #include <obd_class.h>
52 #include "ldlm_internal.h"
55 char *ldlm_lockname[] = {
67 char *ldlm_typename[] = {
74 static ldlm_policy_wire_to_local_t ldlm_policy_wire18_to_local[] = {
75 [LDLM_PLAIN - LDLM_MIN_TYPE] ldlm_plain_policy_wire_to_local,
76 [LDLM_EXTENT - LDLM_MIN_TYPE] ldlm_extent_policy_wire_to_local,
77 [LDLM_FLOCK - LDLM_MIN_TYPE] ldlm_flock_policy_wire18_to_local,
78 [LDLM_IBITS - LDLM_MIN_TYPE] ldlm_ibits_policy_wire_to_local,
81 static ldlm_policy_wire_to_local_t ldlm_policy_wire21_to_local[] = {
82 [LDLM_PLAIN - LDLM_MIN_TYPE] ldlm_plain_policy_wire_to_local,
83 [LDLM_EXTENT - LDLM_MIN_TYPE] ldlm_extent_policy_wire_to_local,
84 [LDLM_FLOCK - LDLM_MIN_TYPE] ldlm_flock_policy_wire21_to_local,
85 [LDLM_IBITS - LDLM_MIN_TYPE] ldlm_ibits_policy_wire_to_local,
88 static ldlm_policy_local_to_wire_t ldlm_policy_local_to_wire[] = {
89 [LDLM_PLAIN - LDLM_MIN_TYPE] ldlm_plain_policy_local_to_wire,
90 [LDLM_EXTENT - LDLM_MIN_TYPE] ldlm_extent_policy_local_to_wire,
91 [LDLM_FLOCK - LDLM_MIN_TYPE] ldlm_flock_policy_local_to_wire,
92 [LDLM_IBITS - LDLM_MIN_TYPE] ldlm_ibits_policy_local_to_wire,
96 * Converts lock policy from local format to on the wire lock_desc format
98 void ldlm_convert_policy_to_wire(ldlm_type_t type,
99 const ldlm_policy_data_t *lpolicy,
100 ldlm_wire_policy_data_t *wpolicy)
102 ldlm_policy_local_to_wire_t convert;
104 convert = ldlm_policy_local_to_wire[type - LDLM_MIN_TYPE];
106 convert(lpolicy, wpolicy);
110 * Converts lock policy from on the wire lock_desc format to local format
112 void ldlm_convert_policy_to_local(struct obd_export *exp, ldlm_type_t type,
113 const ldlm_wire_policy_data_t *wpolicy,
114 ldlm_policy_data_t *lpolicy)
116 ldlm_policy_wire_to_local_t convert;
119 /** some badnes for 2.0.0 clients, but 2.0.0 isn't supported */
120 new_client = (exp->exp_connect_flags & OBD_CONNECT_FULL20) != 0;
122 convert = ldlm_policy_wire21_to_local[type - LDLM_MIN_TYPE];
124 convert = ldlm_policy_wire18_to_local[type - LDLM_MIN_TYPE];
126 convert(wpolicy, lpolicy);
129 char *ldlm_it2str(int it)
136 case (IT_OPEN | IT_CREAT):
151 CERROR("Unknown intent %d\n", it);
156 extern cfs_mem_cache_t *ldlm_lock_slab;
158 #ifdef HAVE_SERVER_SUPPORT
159 static ldlm_processing_policy ldlm_processing_policy_table[] = {
160 [LDLM_PLAIN] ldlm_process_plain_lock,
161 [LDLM_EXTENT] ldlm_process_extent_lock,
163 [LDLM_FLOCK] ldlm_process_flock_lock,
165 [LDLM_IBITS] ldlm_process_inodebits_lock,
168 ldlm_processing_policy ldlm_get_processing_policy(struct ldlm_resource *res)
170 return ldlm_processing_policy_table[res->lr_type];
172 #endif /* HAVE_SERVER_SUPPORT */
174 void ldlm_register_intent(struct ldlm_namespace *ns, ldlm_res_policy arg)
180 * REFCOUNTED LOCK OBJECTS
185 * Lock refcounts, during creation:
186 * - one special one for allocation, dec'd only once in destroy
187 * - one for being a lock that's in-use
188 * - one for the addref associated with a new lock
190 struct ldlm_lock *ldlm_lock_get(struct ldlm_lock *lock)
192 cfs_atomic_inc(&lock->l_refc);
196 void ldlm_lock_put(struct ldlm_lock *lock)
200 LASSERT(lock->l_resource != LP_POISON);
201 LASSERT(cfs_atomic_read(&lock->l_refc) > 0);
202 if (cfs_atomic_dec_and_test(&lock->l_refc)) {
203 struct ldlm_resource *res;
206 "final lock_put on destroyed lock, freeing it.");
208 res = lock->l_resource;
209 LASSERT(lock->l_destroyed);
210 LASSERT(cfs_list_empty(&lock->l_res_link));
211 LASSERT(cfs_list_empty(&lock->l_pending_chain));
213 lprocfs_counter_decr(ldlm_res_to_ns(res)->ns_stats,
215 lu_ref_del(&res->lr_reference, "lock", lock);
216 ldlm_resource_putref(res);
217 lock->l_resource = NULL;
218 if (lock->l_export) {
219 class_export_lock_put(lock->l_export, lock);
220 lock->l_export = NULL;
223 if (lock->l_lvb_data != NULL)
224 OBD_FREE(lock->l_lvb_data, lock->l_lvb_len);
226 ldlm_interval_free(ldlm_interval_detach(lock));
227 lu_ref_fini(&lock->l_reference);
228 OBD_FREE_RCU(lock, sizeof(*lock), &lock->l_handle);
234 int ldlm_lock_remove_from_lru_nolock(struct ldlm_lock *lock)
237 if (!cfs_list_empty(&lock->l_lru)) {
238 struct ldlm_namespace *ns = ldlm_lock_to_ns(lock);
240 LASSERT(lock->l_resource->lr_type != LDLM_FLOCK);
241 cfs_list_del_init(&lock->l_lru);
242 if (lock->l_flags & LDLM_FL_SKIPPED)
243 lock->l_flags &= ~LDLM_FL_SKIPPED;
244 LASSERT(ns->ns_nr_unused > 0);
251 int ldlm_lock_remove_from_lru(struct ldlm_lock *lock)
253 struct ldlm_namespace *ns = ldlm_lock_to_ns(lock);
257 if (lock->l_ns_srv) {
258 LASSERT(cfs_list_empty(&lock->l_lru));
262 cfs_spin_lock(&ns->ns_lock);
263 rc = ldlm_lock_remove_from_lru_nolock(lock);
264 cfs_spin_unlock(&ns->ns_lock);
269 void ldlm_lock_add_to_lru_nolock(struct ldlm_lock *lock)
271 struct ldlm_namespace *ns = ldlm_lock_to_ns(lock);
273 lock->l_last_used = cfs_time_current();
274 LASSERT(cfs_list_empty(&lock->l_lru));
275 LASSERT(lock->l_resource->lr_type != LDLM_FLOCK);
276 cfs_list_add_tail(&lock->l_lru, &ns->ns_unused_list);
277 LASSERT(ns->ns_nr_unused >= 0);
281 void ldlm_lock_add_to_lru(struct ldlm_lock *lock)
283 struct ldlm_namespace *ns = ldlm_lock_to_ns(lock);
286 cfs_spin_lock(&ns->ns_lock);
287 ldlm_lock_add_to_lru_nolock(lock);
288 cfs_spin_unlock(&ns->ns_lock);
292 void ldlm_lock_touch_in_lru(struct ldlm_lock *lock)
294 struct ldlm_namespace *ns = ldlm_lock_to_ns(lock);
297 if (lock->l_ns_srv) {
298 LASSERT(cfs_list_empty(&lock->l_lru));
303 cfs_spin_lock(&ns->ns_lock);
304 if (!cfs_list_empty(&lock->l_lru)) {
305 ldlm_lock_remove_from_lru_nolock(lock);
306 ldlm_lock_add_to_lru_nolock(lock);
308 cfs_spin_unlock(&ns->ns_lock);
312 /* This used to have a 'strict' flag, which recovery would use to mark an
313 * in-use lock as needing-to-die. Lest I am ever tempted to put it back, I
314 * shall explain why it's gone: with the new hash table scheme, once you call
315 * ldlm_lock_destroy, you can never drop your final references on this lock.
316 * Because it's not in the hash table anymore. -phil */
317 int ldlm_lock_destroy_internal(struct ldlm_lock *lock)
321 if (lock->l_readers || lock->l_writers) {
322 LDLM_ERROR(lock, "lock still has references");
326 if (!cfs_list_empty(&lock->l_res_link)) {
327 LDLM_ERROR(lock, "lock still on resource");
331 if (lock->l_destroyed) {
332 LASSERT(cfs_list_empty(&lock->l_lru));
336 lock->l_destroyed = 1;
338 if (lock->l_export && lock->l_export->exp_lock_hash) {
339 /* NB: it's safe to call cfs_hash_del() even lock isn't
340 * in exp_lock_hash. */
341 cfs_hash_del(lock->l_export->exp_lock_hash,
342 &lock->l_remote_handle, &lock->l_exp_hash);
345 ldlm_lock_remove_from_lru(lock);
346 class_handle_unhash(&lock->l_handle);
349 /* Wake anyone waiting for this lock */
350 /* FIXME: I should probably add yet another flag, instead of using
351 * l_export to only call this on clients */
353 class_export_put(lock->l_export);
354 lock->l_export = NULL;
355 if (lock->l_export && lock->l_completion_ast)
356 lock->l_completion_ast(lock, 0);
362 void ldlm_lock_destroy(struct ldlm_lock *lock)
366 lock_res_and_lock(lock);
367 first = ldlm_lock_destroy_internal(lock);
368 unlock_res_and_lock(lock);
370 /* drop reference from hashtable only for first destroy */
372 lu_ref_del(&lock->l_reference, "hash", lock);
373 LDLM_LOCK_RELEASE(lock);
378 void ldlm_lock_destroy_nolock(struct ldlm_lock *lock)
382 first = ldlm_lock_destroy_internal(lock);
383 /* drop reference from hashtable only for first destroy */
385 lu_ref_del(&lock->l_reference, "hash", lock);
386 LDLM_LOCK_RELEASE(lock);
391 /* this is called by portals_handle2object with the handle lock taken */
392 static void lock_handle_addref(void *lock)
394 LDLM_LOCK_GET((struct ldlm_lock *)lock);
397 static void lock_handle_free(void *lock, int size)
399 LASSERT(size == sizeof(struct ldlm_lock));
400 OBD_SLAB_FREE(lock, ldlm_lock_slab, size);
403 struct portals_handle_ops lock_handle_ops = {
404 .hop_addref = lock_handle_addref,
405 .hop_free = lock_handle_free,
409 * usage: pass in a resource on which you have done ldlm_resource_get
410 * new lock will take over the refcount.
411 * returns: lock with refcount 2 - one for current caller and one for remote
413 static struct ldlm_lock *ldlm_lock_new(struct ldlm_resource *resource)
415 struct ldlm_lock *lock;
418 if (resource == NULL)
421 OBD_SLAB_ALLOC_PTR_GFP(lock, ldlm_lock_slab, CFS_ALLOC_IO);
425 cfs_spin_lock_init(&lock->l_lock);
426 lock->l_resource = resource;
427 lu_ref_add(&resource->lr_reference, "lock", lock);
429 cfs_atomic_set(&lock->l_refc, 2);
430 CFS_INIT_LIST_HEAD(&lock->l_res_link);
431 CFS_INIT_LIST_HEAD(&lock->l_lru);
432 CFS_INIT_LIST_HEAD(&lock->l_pending_chain);
433 CFS_INIT_LIST_HEAD(&lock->l_bl_ast);
434 CFS_INIT_LIST_HEAD(&lock->l_cp_ast);
435 CFS_INIT_LIST_HEAD(&lock->l_rk_ast);
436 cfs_waitq_init(&lock->l_waitq);
437 lock->l_blocking_lock = NULL;
438 CFS_INIT_LIST_HEAD(&lock->l_sl_mode);
439 CFS_INIT_LIST_HEAD(&lock->l_sl_policy);
440 CFS_INIT_HLIST_NODE(&lock->l_exp_hash);
442 lprocfs_counter_incr(ldlm_res_to_ns(resource)->ns_stats,
444 CFS_INIT_LIST_HEAD(&lock->l_handle.h_link);
445 class_handle_hash(&lock->l_handle, &lock_handle_ops);
447 lu_ref_init(&lock->l_reference);
448 lu_ref_add(&lock->l_reference, "hash", lock);
449 lock->l_callback_timeout = 0;
451 #if LUSTRE_TRACKS_LOCK_EXP_REFS
452 CFS_INIT_LIST_HEAD(&lock->l_exp_refs_link);
453 lock->l_exp_refs_nr = 0;
454 lock->l_exp_refs_target = NULL;
456 CFS_INIT_LIST_HEAD(&lock->l_exp_list);
461 int ldlm_lock_change_resource(struct ldlm_namespace *ns, struct ldlm_lock *lock,
462 const struct ldlm_res_id *new_resid)
464 struct ldlm_resource *oldres = lock->l_resource;
465 struct ldlm_resource *newres;
469 LASSERT(ns_is_client(ns));
471 lock_res_and_lock(lock);
472 if (memcmp(new_resid, &lock->l_resource->lr_name,
473 sizeof(lock->l_resource->lr_name)) == 0) {
475 unlock_res_and_lock(lock);
479 LASSERT(new_resid->name[0] != 0);
481 /* This function assumes that the lock isn't on any lists */
482 LASSERT(cfs_list_empty(&lock->l_res_link));
484 type = oldres->lr_type;
485 unlock_res_and_lock(lock);
487 newres = ldlm_resource_get(ns, NULL, new_resid, type, 1);
491 lu_ref_add(&newres->lr_reference, "lock", lock);
493 * To flip the lock from the old to the new resource, lock, oldres and
494 * newres have to be locked. Resource spin-locks are nested within
495 * lock->l_lock, and are taken in the memory address order to avoid
498 cfs_spin_lock(&lock->l_lock);
499 oldres = lock->l_resource;
500 if (oldres < newres) {
502 lock_res_nested(newres, LRT_NEW);
505 lock_res_nested(oldres, LRT_NEW);
507 LASSERT(memcmp(new_resid, &oldres->lr_name,
508 sizeof oldres->lr_name) != 0);
509 lock->l_resource = newres;
511 unlock_res_and_lock(lock);
513 /* ...and the flowers are still standing! */
514 lu_ref_del(&oldres->lr_reference, "lock", lock);
515 ldlm_resource_putref(oldres);
524 void ldlm_lock2handle(const struct ldlm_lock *lock, struct lustre_handle *lockh)
526 lockh->cookie = lock->l_handle.h_cookie;
529 /* if flags: atomically get the lock and set the flags.
530 * Return NULL if flag already set
533 struct ldlm_lock *__ldlm_handle2lock(const struct lustre_handle *handle,
536 struct ldlm_lock *lock;
541 lock = class_handle2object(handle->cookie);
545 /* It's unlikely but possible that someone marked the lock as
546 * destroyed after we did handle2object on it */
547 if (flags == 0 && !lock->l_destroyed) {
548 lu_ref_add(&lock->l_reference, "handle", cfs_current());
552 lock_res_and_lock(lock);
554 LASSERT(lock->l_resource != NULL);
556 lu_ref_add_atomic(&lock->l_reference, "handle", cfs_current());
557 if (unlikely(lock->l_destroyed)) {
558 unlock_res_and_lock(lock);
559 CDEBUG(D_INFO, "lock already destroyed: lock %p\n", lock);
564 if (flags && (lock->l_flags & flags)) {
565 unlock_res_and_lock(lock);
571 lock->l_flags |= flags;
573 unlock_res_and_lock(lock);
577 void ldlm_lock2desc(struct ldlm_lock *lock, struct ldlm_lock_desc *desc)
579 struct obd_export *exp = lock->l_export?:lock->l_conn_export;
580 /* INODEBITS_INTEROP: If the other side does not support
581 * inodebits, reply with a plain lock descriptor.
583 if ((lock->l_resource->lr_type == LDLM_IBITS) &&
584 (exp && !(exp->exp_connect_flags & OBD_CONNECT_IBITS))) {
585 /* Make sure all the right bits are set in this lock we
586 are going to pass to client */
587 LASSERTF(lock->l_policy_data.l_inodebits.bits ==
588 (MDS_INODELOCK_LOOKUP | MDS_INODELOCK_UPDATE |
589 MDS_INODELOCK_LAYOUT),
590 "Inappropriate inode lock bits during "
591 "conversion " LPU64 "\n",
592 lock->l_policy_data.l_inodebits.bits);
594 ldlm_res2desc(lock->l_resource, &desc->l_resource);
595 desc->l_resource.lr_type = LDLM_PLAIN;
597 /* Convert "new" lock mode to something old client can
599 if ((lock->l_req_mode == LCK_CR) ||
600 (lock->l_req_mode == LCK_CW))
601 desc->l_req_mode = LCK_PR;
603 desc->l_req_mode = lock->l_req_mode;
604 if ((lock->l_granted_mode == LCK_CR) ||
605 (lock->l_granted_mode == LCK_CW)) {
606 desc->l_granted_mode = LCK_PR;
608 /* We never grant PW/EX locks to clients */
609 LASSERT((lock->l_granted_mode != LCK_PW) &&
610 (lock->l_granted_mode != LCK_EX));
611 desc->l_granted_mode = lock->l_granted_mode;
614 /* We do not copy policy here, because there is no
615 policy for plain locks */
617 ldlm_res2desc(lock->l_resource, &desc->l_resource);
618 desc->l_req_mode = lock->l_req_mode;
619 desc->l_granted_mode = lock->l_granted_mode;
620 ldlm_convert_policy_to_wire(lock->l_resource->lr_type,
621 &lock->l_policy_data,
622 &desc->l_policy_data);
626 void ldlm_add_bl_work_item(struct ldlm_lock *lock, struct ldlm_lock *new,
627 cfs_list_t *work_list)
629 if ((lock->l_flags & LDLM_FL_AST_SENT) == 0) {
630 LDLM_DEBUG(lock, "lock incompatible; sending blocking AST.");
631 lock->l_flags |= LDLM_FL_AST_SENT;
632 /* If the enqueuing client said so, tell the AST recipient to
633 * discard dirty data, rather than writing back. */
634 if (new->l_flags & LDLM_AST_DISCARD_DATA)
635 lock->l_flags |= LDLM_FL_DISCARD_DATA;
636 LASSERT(cfs_list_empty(&lock->l_bl_ast));
637 cfs_list_add(&lock->l_bl_ast, work_list);
639 LASSERT(lock->l_blocking_lock == NULL);
640 lock->l_blocking_lock = LDLM_LOCK_GET(new);
644 void ldlm_add_cp_work_item(struct ldlm_lock *lock, cfs_list_t *work_list)
646 if ((lock->l_flags & LDLM_FL_CP_REQD) == 0) {
647 lock->l_flags |= LDLM_FL_CP_REQD;
648 LDLM_DEBUG(lock, "lock granted; sending completion AST.");
649 LASSERT(cfs_list_empty(&lock->l_cp_ast));
650 cfs_list_add(&lock->l_cp_ast, work_list);
655 /* must be called with lr_lock held */
656 void ldlm_add_ast_work_item(struct ldlm_lock *lock, struct ldlm_lock *new,
657 cfs_list_t *work_list)
660 check_res_locked(lock->l_resource);
662 ldlm_add_bl_work_item(lock, new, work_list);
664 ldlm_add_cp_work_item(lock, work_list);
668 void ldlm_lock_addref(struct lustre_handle *lockh, __u32 mode)
670 struct ldlm_lock *lock;
672 lock = ldlm_handle2lock(lockh);
673 LASSERT(lock != NULL);
674 ldlm_lock_addref_internal(lock, mode);
678 void ldlm_lock_addref_internal_nolock(struct ldlm_lock *lock, __u32 mode)
680 ldlm_lock_remove_from_lru(lock);
681 if (mode & (LCK_NL | LCK_CR | LCK_PR)) {
683 lu_ref_add_atomic(&lock->l_reference, "reader", lock);
685 if (mode & (LCK_EX | LCK_CW | LCK_PW | LCK_GROUP | LCK_COS)) {
687 lu_ref_add_atomic(&lock->l_reference, "writer", lock);
690 lu_ref_add_atomic(&lock->l_reference, "user", lock);
691 LDLM_DEBUG(lock, "ldlm_lock_addref(%s)", ldlm_lockname[mode]);
695 * Attempts to addref a lock, and fails if lock is already LDLM_FL_CBPENDING
698 * \retval 0 success, lock was addref-ed
700 * \retval -EAGAIN lock is being canceled.
702 int ldlm_lock_addref_try(struct lustre_handle *lockh, __u32 mode)
704 struct ldlm_lock *lock;
708 lock = ldlm_handle2lock(lockh);
710 lock_res_and_lock(lock);
711 if (lock->l_readers != 0 || lock->l_writers != 0 ||
712 !(lock->l_flags & LDLM_FL_CBPENDING)) {
713 ldlm_lock_addref_internal_nolock(lock, mode);
716 unlock_res_and_lock(lock);
722 /* only called for local locks */
723 void ldlm_lock_addref_internal(struct ldlm_lock *lock, __u32 mode)
725 lock_res_and_lock(lock);
726 ldlm_lock_addref_internal_nolock(lock, mode);
727 unlock_res_and_lock(lock);
730 /* only called in ldlm_flock_destroy and for local locks.
731 * * for LDLM_FLOCK type locks, l_blocking_ast is null, and
732 * * ldlm_lock_remove_from_lru() does nothing, it is safe
733 * * for ldlm_flock_destroy usage by dropping some code */
734 void ldlm_lock_decref_internal_nolock(struct ldlm_lock *lock, __u32 mode)
736 LDLM_DEBUG(lock, "ldlm_lock_decref(%s)", ldlm_lockname[mode]);
737 if (mode & (LCK_NL | LCK_CR | LCK_PR)) {
738 LASSERT(lock->l_readers > 0);
739 lu_ref_del(&lock->l_reference, "reader", lock);
742 if (mode & (LCK_EX | LCK_CW | LCK_PW | LCK_GROUP | LCK_COS)) {
743 LASSERT(lock->l_writers > 0);
744 lu_ref_del(&lock->l_reference, "writer", lock);
748 lu_ref_del(&lock->l_reference, "user", lock);
749 LDLM_LOCK_RELEASE(lock); /* matches the LDLM_LOCK_GET() in addref */
752 void ldlm_lock_decref_internal(struct ldlm_lock *lock, __u32 mode)
754 struct ldlm_namespace *ns;
757 lock_res_and_lock(lock);
759 ns = ldlm_lock_to_ns(lock);
761 ldlm_lock_decref_internal_nolock(lock, mode);
763 if (lock->l_flags & LDLM_FL_LOCAL &&
764 !lock->l_readers && !lock->l_writers) {
765 /* If this is a local lock on a server namespace and this was
766 * the last reference, cancel the lock. */
767 CDEBUG(D_INFO, "forcing cancel of local lock\n");
768 lock->l_flags |= LDLM_FL_CBPENDING;
771 if (!lock->l_readers && !lock->l_writers &&
772 (lock->l_flags & LDLM_FL_CBPENDING)) {
773 /* If we received a blocked AST and this was the last reference,
774 * run the callback. */
775 if (lock->l_ns_srv && lock->l_export)
776 CERROR("FL_CBPENDING set on non-local lock--just a "
779 LDLM_DEBUG(lock, "final decref done on cbpending lock");
781 LDLM_LOCK_GET(lock); /* dropped by bl thread */
782 ldlm_lock_remove_from_lru(lock);
783 unlock_res_and_lock(lock);
785 if (lock->l_flags & LDLM_FL_FAIL_LOC)
786 OBD_RACE(OBD_FAIL_LDLM_CP_BL_RACE);
788 if ((lock->l_flags & LDLM_FL_ATOMIC_CB) ||
789 ldlm_bl_to_thread_lock(ns, NULL, lock) != 0)
790 ldlm_handle_bl_callback(ns, NULL, lock);
791 } else if (ns_is_client(ns) &&
792 !lock->l_readers && !lock->l_writers &&
793 !(lock->l_flags & LDLM_FL_NO_LRU) &&
794 !(lock->l_flags & LDLM_FL_BL_AST)) {
796 LDLM_DEBUG(lock, "add lock into lru list");
798 /* If this is a client-side namespace and this was the last
799 * reference, put it on the LRU. */
800 ldlm_lock_add_to_lru(lock);
801 unlock_res_and_lock(lock);
803 if (lock->l_flags & LDLM_FL_FAIL_LOC)
804 OBD_RACE(OBD_FAIL_LDLM_CP_BL_RACE);
806 /* Call ldlm_cancel_lru() only if EARLY_CANCEL and LRU RESIZE
807 * are not supported by the server, otherwise, it is done on
809 if (!exp_connect_cancelset(lock->l_conn_export) &&
810 !ns_connect_lru_resize(ns))
811 ldlm_cancel_lru(ns, 0, LDLM_ASYNC, 0);
813 LDLM_DEBUG(lock, "do not add lock into lru list");
814 unlock_res_and_lock(lock);
820 void ldlm_lock_decref(struct lustre_handle *lockh, __u32 mode)
822 struct ldlm_lock *lock = __ldlm_handle2lock(lockh, 0);
823 LASSERTF(lock != NULL, "Non-existing lock: "LPX64"\n", lockh->cookie);
824 ldlm_lock_decref_internal(lock, mode);
828 /* This will drop a lock reference and mark it for destruction, but will not
829 * necessarily cancel the lock before returning. */
830 void ldlm_lock_decref_and_cancel(struct lustre_handle *lockh, __u32 mode)
832 struct ldlm_lock *lock = __ldlm_handle2lock(lockh, 0);
835 LASSERT(lock != NULL);
837 LDLM_DEBUG(lock, "ldlm_lock_decref(%s)", ldlm_lockname[mode]);
838 lock_res_and_lock(lock);
839 lock->l_flags |= LDLM_FL_CBPENDING;
840 unlock_res_and_lock(lock);
841 ldlm_lock_decref_internal(lock, mode);
845 struct sl_insert_point {
846 cfs_list_t *res_link;
847 cfs_list_t *mode_link;
848 cfs_list_t *policy_link;
852 * search_granted_lock
855 * Finds a position to insert the new lock.
857 * queue [input]: the granted list where search acts on;
858 * req [input]: the lock whose position to be located;
859 * prev [output]: positions within 3 lists to insert @req to
863 * - ldlm_grant_lock_with_skiplist
865 static void search_granted_lock(cfs_list_t *queue,
866 struct ldlm_lock *req,
867 struct sl_insert_point *prev)
870 struct ldlm_lock *lock, *mode_end, *policy_end;
873 cfs_list_for_each(tmp, queue) {
874 lock = cfs_list_entry(tmp, struct ldlm_lock, l_res_link);
876 mode_end = cfs_list_entry(lock->l_sl_mode.prev,
877 struct ldlm_lock, l_sl_mode);
879 if (lock->l_req_mode != req->l_req_mode) {
880 /* jump to last lock of mode group */
881 tmp = &mode_end->l_res_link;
885 /* suitable mode group is found */
886 if (lock->l_resource->lr_type == LDLM_PLAIN) {
887 /* insert point is last lock of the mode group */
888 prev->res_link = &mode_end->l_res_link;
889 prev->mode_link = &mode_end->l_sl_mode;
890 prev->policy_link = &req->l_sl_policy;
893 } else if (lock->l_resource->lr_type == LDLM_IBITS) {
896 cfs_list_entry(lock->l_sl_policy.prev,
900 if (lock->l_policy_data.l_inodebits.bits ==
901 req->l_policy_data.l_inodebits.bits) {
902 /* insert point is last lock of
903 * the policy group */
905 &policy_end->l_res_link;
907 &policy_end->l_sl_mode;
909 &policy_end->l_sl_policy;
914 if (policy_end == mode_end)
915 /* done with mode group */
918 /* go to next policy group within mode group */
919 tmp = policy_end->l_res_link.next;
920 lock = cfs_list_entry(tmp, struct ldlm_lock,
922 } /* loop over policy groups within the mode group */
924 /* insert point is last lock of the mode group,
925 * new policy group is started */
926 prev->res_link = &mode_end->l_res_link;
927 prev->mode_link = &mode_end->l_sl_mode;
928 prev->policy_link = &req->l_sl_policy;
932 LDLM_ERROR(lock,"is not LDLM_PLAIN or LDLM_IBITS lock");
937 /* insert point is last lock on the queue,
938 * new mode group and new policy group are started */
939 prev->res_link = queue->prev;
940 prev->mode_link = &req->l_sl_mode;
941 prev->policy_link = &req->l_sl_policy;
946 static void ldlm_granted_list_add_lock(struct ldlm_lock *lock,
947 struct sl_insert_point *prev)
949 struct ldlm_resource *res = lock->l_resource;
952 check_res_locked(res);
954 ldlm_resource_dump(D_INFO, res);
955 LDLM_DEBUG(lock, "About to add lock:");
957 if (lock->l_destroyed) {
958 CDEBUG(D_OTHER, "Lock destroyed, not adding to resource\n");
962 LASSERT(cfs_list_empty(&lock->l_res_link));
963 LASSERT(cfs_list_empty(&lock->l_sl_mode));
964 LASSERT(cfs_list_empty(&lock->l_sl_policy));
966 cfs_list_add(&lock->l_res_link, prev->res_link);
967 cfs_list_add(&lock->l_sl_mode, prev->mode_link);
968 cfs_list_add(&lock->l_sl_policy, prev->policy_link);
973 static void ldlm_grant_lock_with_skiplist(struct ldlm_lock *lock)
975 struct sl_insert_point prev;
978 LASSERT(lock->l_req_mode == lock->l_granted_mode);
980 search_granted_lock(&lock->l_resource->lr_granted, lock, &prev);
981 ldlm_granted_list_add_lock(lock, &prev);
986 * - ldlm_lock_enqueue
987 * - ldlm_reprocess_queue
988 * - ldlm_lock_convert
990 * must be called with lr_lock held
992 void ldlm_grant_lock(struct ldlm_lock *lock, cfs_list_t *work_list)
994 struct ldlm_resource *res = lock->l_resource;
997 check_res_locked(res);
999 lock->l_granted_mode = lock->l_req_mode;
1000 if (res->lr_type == LDLM_PLAIN || res->lr_type == LDLM_IBITS)
1001 ldlm_grant_lock_with_skiplist(lock);
1002 else if (res->lr_type == LDLM_EXTENT)
1003 ldlm_extent_add_lock(res, lock);
1005 ldlm_resource_add_lock(res, &res->lr_granted, lock);
1007 if (lock->l_granted_mode < res->lr_most_restr)
1008 res->lr_most_restr = lock->l_granted_mode;
1010 if (work_list && lock->l_completion_ast != NULL)
1011 ldlm_add_ast_work_item(lock, NULL, work_list);
1013 ldlm_pool_add(&ldlm_res_to_ns(res)->ns_pool, lock);
1017 /* returns a referenced lock or NULL. See the flag descriptions below, in the
1018 * comment above ldlm_lock_match */
1019 static struct ldlm_lock *search_queue(cfs_list_t *queue,
1021 ldlm_policy_data_t *policy,
1022 struct ldlm_lock *old_lock,
1023 int flags, int unref)
1025 struct ldlm_lock *lock;
1028 cfs_list_for_each(tmp, queue) {
1031 lock = cfs_list_entry(tmp, struct ldlm_lock, l_res_link);
1033 if (lock == old_lock)
1036 /* llite sometimes wants to match locks that will be
1037 * canceled when their users drop, but we allow it to match
1038 * if it passes in CBPENDING and the lock still has users.
1039 * this is generally only going to be used by children
1040 * whose parents already hold a lock so forward progress
1041 * can still happen. */
1042 if (lock->l_flags & LDLM_FL_CBPENDING &&
1043 !(flags & LDLM_FL_CBPENDING))
1045 if (!unref && lock->l_flags & LDLM_FL_CBPENDING &&
1046 lock->l_readers == 0 && lock->l_writers == 0)
1049 if (!(lock->l_req_mode & *mode))
1051 match = lock->l_req_mode;
1053 if (lock->l_resource->lr_type == LDLM_EXTENT &&
1054 (lock->l_policy_data.l_extent.start >
1055 policy->l_extent.start ||
1056 lock->l_policy_data.l_extent.end < policy->l_extent.end))
1059 if (unlikely(match == LCK_GROUP) &&
1060 lock->l_resource->lr_type == LDLM_EXTENT &&
1061 lock->l_policy_data.l_extent.gid != policy->l_extent.gid)
1064 /* We match if we have existing lock with same or wider set
1066 if (lock->l_resource->lr_type == LDLM_IBITS &&
1067 ((lock->l_policy_data.l_inodebits.bits &
1068 policy->l_inodebits.bits) !=
1069 policy->l_inodebits.bits))
1073 (lock->l_destroyed || lock->l_flags & LDLM_FL_FAILED ||
1077 if ((flags & LDLM_FL_LOCAL_ONLY) &&
1078 !(lock->l_flags & LDLM_FL_LOCAL))
1081 if (flags & LDLM_FL_TEST_LOCK) {
1082 LDLM_LOCK_GET(lock);
1083 ldlm_lock_touch_in_lru(lock);
1085 ldlm_lock_addref_internal_nolock(lock, match);
1094 void ldlm_lock_fail_match_locked(struct ldlm_lock *lock)
1096 if (!lock->l_failed) {
1098 cfs_waitq_broadcast(&lock->l_waitq);
1101 EXPORT_SYMBOL(ldlm_lock_fail_match_locked);
1103 void ldlm_lock_fail_match(struct ldlm_lock *lock)
1105 lock_res_and_lock(lock);
1106 ldlm_lock_fail_match_locked(lock);
1107 unlock_res_and_lock(lock);
1109 EXPORT_SYMBOL(ldlm_lock_fail_match);
1111 void ldlm_lock_allow_match_locked(struct ldlm_lock *lock)
1113 lock->l_flags |= LDLM_FL_LVB_READY;
1114 cfs_waitq_broadcast(&lock->l_waitq);
1117 void ldlm_lock_allow_match(struct ldlm_lock *lock)
1119 lock_res_and_lock(lock);
1120 ldlm_lock_allow_match_locked(lock);
1121 unlock_res_and_lock(lock);
1124 /* Can be called in two ways:
1126 * If 'ns' is NULL, then lockh describes an existing lock that we want to look
1127 * for a duplicate of.
1129 * Otherwise, all of the fields must be filled in, to match against.
1131 * If 'flags' contains LDLM_FL_LOCAL_ONLY, then only match local locks on the
1132 * server (ie, connh is NULL)
1133 * If 'flags' contains LDLM_FL_BLOCK_GRANTED, then only locks on the granted
1134 * list will be considered
1135 * If 'flags' contains LDLM_FL_CBPENDING, then locks that have been marked
1136 * to be canceled can still be matched as long as they still have reader
1137 * or writer refernces
1138 * If 'flags' contains LDLM_FL_TEST_LOCK, then don't actually reference a lock,
1139 * just tell us if we would have matched.
1141 * Returns 1 if it finds an already-existing lock that is compatible; in this
1142 * case, lockh is filled in with a addref()ed lock
1144 * we also check security context, if that failed we simply return 0 (to keep
1145 * caller code unchanged), the context failure will be discovered by caller
1148 ldlm_mode_t ldlm_lock_match(struct ldlm_namespace *ns, int flags,
1149 const struct ldlm_res_id *res_id, ldlm_type_t type,
1150 ldlm_policy_data_t *policy, ldlm_mode_t mode,
1151 struct lustre_handle *lockh, int unref)
1153 struct ldlm_resource *res;
1154 struct ldlm_lock *lock, *old_lock = NULL;
1159 old_lock = ldlm_handle2lock(lockh);
1162 ns = ldlm_lock_to_ns(old_lock);
1163 res_id = &old_lock->l_resource->lr_name;
1164 type = old_lock->l_resource->lr_type;
1165 mode = old_lock->l_req_mode;
1168 res = ldlm_resource_get(ns, NULL, res_id, type, 0);
1170 LASSERT(old_lock == NULL);
1174 LDLM_RESOURCE_ADDREF(res);
1177 lock = search_queue(&res->lr_granted, &mode, policy, old_lock,
1181 if (flags & LDLM_FL_BLOCK_GRANTED)
1183 lock = search_queue(&res->lr_converting, &mode, policy, old_lock,
1187 lock = search_queue(&res->lr_waiting, &mode, policy, old_lock,
1195 LDLM_RESOURCE_DELREF(res);
1196 ldlm_resource_putref(res);
1199 ldlm_lock2handle(lock, lockh);
1200 if ((flags & LDLM_FL_LVB_READY) &&
1201 (!(lock->l_flags & LDLM_FL_LVB_READY))) {
1202 struct l_wait_info lwi;
1203 if (lock->l_completion_ast) {
1204 int err = lock->l_completion_ast(lock,
1205 LDLM_FL_WAIT_NOREPROC,
1208 if (flags & LDLM_FL_TEST_LOCK)
1209 LDLM_LOCK_RELEASE(lock);
1211 ldlm_lock_decref_internal(lock,
1218 lwi = LWI_TIMEOUT_INTR(cfs_time_seconds(obd_timeout),
1219 NULL, LWI_ON_SIGNAL_NOOP, NULL);
1221 /* XXX FIXME see comment on CAN_MATCH in lustre_dlm.h */
1222 l_wait_event(lock->l_waitq,
1223 lock->l_flags & LDLM_FL_LVB_READY ||
1226 if (!(lock->l_flags & LDLM_FL_LVB_READY)) {
1227 if (flags & LDLM_FL_TEST_LOCK)
1228 LDLM_LOCK_RELEASE(lock);
1230 ldlm_lock_decref_internal(lock, mode);
1237 LDLM_DEBUG(lock, "matched ("LPU64" "LPU64")",
1238 (type == LDLM_PLAIN || type == LDLM_IBITS) ?
1239 res_id->name[2] : policy->l_extent.start,
1240 (type == LDLM_PLAIN || type == LDLM_IBITS) ?
1241 res_id->name[3] : policy->l_extent.end);
1243 /* check user's security context */
1244 if (lock->l_conn_export &&
1245 sptlrpc_import_check_ctx(
1246 class_exp2cliimp(lock->l_conn_export))) {
1247 if (!(flags & LDLM_FL_TEST_LOCK))
1248 ldlm_lock_decref_internal(lock, mode);
1252 if (flags & LDLM_FL_TEST_LOCK)
1253 LDLM_LOCK_RELEASE(lock);
1255 } else if (!(flags & LDLM_FL_TEST_LOCK)) {/*less verbose for test-only*/
1256 LDLM_DEBUG_NOLOCK("not matched ns %p type %u mode %u res "
1257 LPU64"/"LPU64" ("LPU64" "LPU64")", ns,
1258 type, mode, res_id->name[0], res_id->name[1],
1259 (type == LDLM_PLAIN || type == LDLM_IBITS) ?
1260 res_id->name[2] :policy->l_extent.start,
1261 (type == LDLM_PLAIN || type == LDLM_IBITS) ?
1262 res_id->name[3] : policy->l_extent.end);
1265 LDLM_LOCK_PUT(old_lock);
1267 return rc ? mode : 0;
1270 ldlm_mode_t ldlm_revalidate_lock_handle(struct lustre_handle *lockh,
1273 struct ldlm_lock *lock;
1274 ldlm_mode_t mode = 0;
1277 lock = ldlm_handle2lock(lockh);
1279 lock_res_and_lock(lock);
1280 if (lock->l_destroyed || lock->l_flags & LDLM_FL_FAILED ||
1284 if (lock->l_flags & LDLM_FL_CBPENDING &&
1285 lock->l_readers == 0 && lock->l_writers == 0)
1289 *bits = lock->l_policy_data.l_inodebits.bits;
1290 mode = lock->l_granted_mode;
1291 ldlm_lock_addref_internal_nolock(lock, mode);
1298 unlock_res_and_lock(lock);
1299 LDLM_LOCK_PUT(lock);
1303 EXPORT_SYMBOL(ldlm_revalidate_lock_handle);
1305 /* Returns a referenced lock */
1306 struct ldlm_lock *ldlm_lock_create(struct ldlm_namespace *ns,
1307 const struct ldlm_res_id *res_id,
1310 const struct ldlm_callback_suite *cbs,
1311 void *data, __u32 lvb_len)
1313 struct ldlm_lock *lock;
1314 struct ldlm_resource *res;
1317 res = ldlm_resource_get(ns, NULL, res_id, type, 1);
1321 lock = ldlm_lock_new(res);
1326 lock->l_req_mode = mode;
1327 lock->l_ast_data = data;
1328 lock->l_pid = cfs_curproc_pid();
1329 lock->l_ns_srv = !!ns_is_server(ns);
1331 lock->l_blocking_ast = cbs->lcs_blocking;
1332 lock->l_completion_ast = cbs->lcs_completion;
1333 lock->l_glimpse_ast = cbs->lcs_glimpse;
1334 lock->l_weigh_ast = cbs->lcs_weigh;
1337 lock->l_tree_node = NULL;
1338 /* if this is the extent lock, allocate the interval tree node */
1339 if (type == LDLM_EXTENT) {
1340 if (ldlm_interval_alloc(lock) == NULL)
1345 lock->l_lvb_len = lvb_len;
1346 OBD_ALLOC(lock->l_lvb_data, lvb_len);
1347 if (lock->l_lvb_data == NULL)
1351 if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_NEW_LOCK))
1357 ldlm_lock_destroy(lock);
1358 LDLM_LOCK_RELEASE(lock);
1362 ldlm_error_t ldlm_lock_enqueue(struct ldlm_namespace *ns,
1363 struct ldlm_lock **lockp,
1364 void *cookie, int *flags)
1366 struct ldlm_lock *lock = *lockp;
1367 struct ldlm_resource *res = lock->l_resource;
1368 int local = ns_is_client(ldlm_res_to_ns(res));
1369 #ifdef HAVE_SERVER_SUPPORT
1370 ldlm_processing_policy policy;
1372 ldlm_error_t rc = ELDLM_OK;
1373 struct ldlm_interval *node = NULL;
1376 lock->l_last_activity = cfs_time_current_sec();
1377 /* policies are not executed on the client or during replay */
1378 if ((*flags & (LDLM_FL_HAS_INTENT|LDLM_FL_REPLAY)) == LDLM_FL_HAS_INTENT
1379 && !local && ns->ns_policy) {
1380 rc = ns->ns_policy(ns, lockp, cookie, lock->l_req_mode, *flags,
1382 if (rc == ELDLM_LOCK_REPLACED) {
1383 /* The lock that was returned has already been granted,
1384 * and placed into lockp. If it's not the same as the
1385 * one we passed in, then destroy the old one and our
1386 * work here is done. */
1387 if (lock != *lockp) {
1388 ldlm_lock_destroy(lock);
1389 LDLM_LOCK_RELEASE(lock);
1391 *flags |= LDLM_FL_LOCK_CHANGED;
1393 } else if (rc != ELDLM_OK ||
1394 (rc == ELDLM_OK && (*flags & LDLM_FL_INTENT_ONLY))) {
1395 ldlm_lock_destroy(lock);
1400 /* For a replaying lock, it might be already in granted list. So
1401 * unlinking the lock will cause the interval node to be freed, we
1402 * have to allocate the interval node early otherwise we can't regrant
1403 * this lock in the future. - jay */
1404 if (!local && (*flags & LDLM_FL_REPLAY) && res->lr_type == LDLM_EXTENT)
1405 OBD_SLAB_ALLOC_PTR_GFP(node, ldlm_interval_slab, CFS_ALLOC_IO);
1407 lock_res_and_lock(lock);
1408 if (local && lock->l_req_mode == lock->l_granted_mode) {
1409 /* The server returned a blocked lock, but it was granted
1410 * before we got a chance to actually enqueue it. We don't
1411 * need to do anything else. */
1412 *flags &= ~(LDLM_FL_BLOCK_GRANTED |
1413 LDLM_FL_BLOCK_CONV | LDLM_FL_BLOCK_WAIT);
1414 GOTO(out, ELDLM_OK);
1417 ldlm_resource_unlink_lock(lock);
1418 if (res->lr_type == LDLM_EXTENT && lock->l_tree_node == NULL) {
1420 ldlm_lock_destroy_nolock(lock);
1421 GOTO(out, rc = -ENOMEM);
1424 CFS_INIT_LIST_HEAD(&node->li_group);
1425 ldlm_interval_attach(node, lock);
1429 /* Some flags from the enqueue want to make it into the AST, via the
1430 * lock's l_flags. */
1431 lock->l_flags |= *flags & LDLM_AST_DISCARD_DATA;
1433 /* This distinction between local lock trees is very important; a client
1434 * namespace only has information about locks taken by that client, and
1435 * thus doesn't have enough information to decide for itself if it can
1436 * be granted (below). In this case, we do exactly what the server
1437 * tells us to do, as dictated by the 'flags'.
1439 * We do exactly the same thing during recovery, when the server is
1440 * more or less trusting the clients not to lie.
1442 * FIXME (bug 268): Detect obvious lies by checking compatibility in
1443 * granted/converting queues. */
1445 if (*flags & LDLM_FL_BLOCK_CONV)
1446 ldlm_resource_add_lock(res, &res->lr_converting, lock);
1447 else if (*flags & (LDLM_FL_BLOCK_WAIT | LDLM_FL_BLOCK_GRANTED))
1448 ldlm_resource_add_lock(res, &res->lr_waiting, lock);
1450 ldlm_grant_lock(lock, NULL);
1451 GOTO(out, ELDLM_OK);
1452 #ifdef HAVE_SERVER_SUPPORT
1453 } else if (*flags & LDLM_FL_REPLAY) {
1454 if (*flags & LDLM_FL_BLOCK_CONV) {
1455 ldlm_resource_add_lock(res, &res->lr_converting, lock);
1456 GOTO(out, ELDLM_OK);
1457 } else if (*flags & LDLM_FL_BLOCK_WAIT) {
1458 ldlm_resource_add_lock(res, &res->lr_waiting, lock);
1459 GOTO(out, ELDLM_OK);
1460 } else if (*flags & LDLM_FL_BLOCK_GRANTED) {
1461 ldlm_grant_lock(lock, NULL);
1462 GOTO(out, ELDLM_OK);
1464 /* If no flags, fall through to normal enqueue path. */
1467 policy = ldlm_processing_policy_table[res->lr_type];
1468 policy(lock, flags, 1, &rc, NULL);
1472 CERROR("This is client-side-only module, cannot handle "
1473 "LDLM_NAMESPACE_SERVER resource type lock.\n");
1479 unlock_res_and_lock(lock);
1481 OBD_SLAB_FREE(node, ldlm_interval_slab, sizeof(*node));
1485 #ifdef HAVE_SERVER_SUPPORT
1486 /* Must be called with namespace taken: queue is waiting or converting. */
1487 int ldlm_reprocess_queue(struct ldlm_resource *res, cfs_list_t *queue,
1488 cfs_list_t *work_list)
1490 cfs_list_t *tmp, *pos;
1491 ldlm_processing_policy policy;
1493 int rc = LDLM_ITER_CONTINUE;
1497 check_res_locked(res);
1499 policy = ldlm_processing_policy_table[res->lr_type];
1502 cfs_list_for_each_safe(tmp, pos, queue) {
1503 struct ldlm_lock *pending;
1504 pending = cfs_list_entry(tmp, struct ldlm_lock, l_res_link);
1506 CDEBUG(D_INFO, "Reprocessing lock %p\n", pending);
1509 rc = policy(pending, &flags, 0, &err, work_list);
1510 if (rc != LDLM_ITER_CONTINUE)
1519 ldlm_work_bl_ast_lock(struct ptlrpc_request_set *rqset, void *opaq)
1521 struct ldlm_cb_set_arg *arg = opaq;
1522 struct ldlm_lock_desc d;
1524 struct ldlm_lock *lock;
1527 if (cfs_list_empty(arg->list))
1530 lock = cfs_list_entry(arg->list->next, struct ldlm_lock, l_bl_ast);
1532 /* nobody should touch l_bl_ast */
1533 lock_res_and_lock(lock);
1534 cfs_list_del_init(&lock->l_bl_ast);
1536 LASSERT(lock->l_flags & LDLM_FL_AST_SENT);
1537 LASSERT(lock->l_bl_ast_run == 0);
1538 LASSERT(lock->l_blocking_lock);
1539 lock->l_bl_ast_run++;
1540 unlock_res_and_lock(lock);
1542 ldlm_lock2desc(lock->l_blocking_lock, &d);
1544 rc = lock->l_blocking_ast(lock, &d, (void *)arg, LDLM_CB_BLOCKING);
1545 LDLM_LOCK_RELEASE(lock->l_blocking_lock);
1546 lock->l_blocking_lock = NULL;
1547 LDLM_LOCK_RELEASE(lock);
1553 ldlm_work_cp_ast_lock(struct ptlrpc_request_set *rqset, void *opaq)
1555 struct ldlm_cb_set_arg *arg = opaq;
1557 struct ldlm_lock *lock;
1558 ldlm_completion_callback completion_callback;
1561 if (cfs_list_empty(arg->list))
1564 lock = cfs_list_entry(arg->list->next, struct ldlm_lock, l_cp_ast);
1566 /* It's possible to receive a completion AST before we've set
1567 * the l_completion_ast pointer: either because the AST arrived
1568 * before the reply, or simply because there's a small race
1569 * window between receiving the reply and finishing the local
1570 * enqueue. (bug 842)
1572 * This can't happen with the blocking_ast, however, because we
1573 * will never call the local blocking_ast until we drop our
1574 * reader/writer reference, which we won't do until we get the
1575 * reply and finish enqueueing. */
1577 /* nobody should touch l_cp_ast */
1578 lock_res_and_lock(lock);
1579 cfs_list_del_init(&lock->l_cp_ast);
1580 LASSERT(lock->l_flags & LDLM_FL_CP_REQD);
1581 /* save l_completion_ast since it can be changed by
1582 * mds_intent_policy(), see bug 14225 */
1583 completion_callback = lock->l_completion_ast;
1584 lock->l_flags &= ~LDLM_FL_CP_REQD;
1585 unlock_res_and_lock(lock);
1587 if (completion_callback != NULL)
1588 rc = completion_callback(lock, 0, (void *)arg);
1589 LDLM_LOCK_RELEASE(lock);
1595 ldlm_work_revoke_ast_lock(struct ptlrpc_request_set *rqset, void *opaq)
1597 struct ldlm_cb_set_arg *arg = opaq;
1598 struct ldlm_lock_desc desc;
1600 struct ldlm_lock *lock;
1603 if (cfs_list_empty(arg->list))
1606 lock = cfs_list_entry(arg->list->next, struct ldlm_lock, l_rk_ast);
1607 cfs_list_del_init(&lock->l_rk_ast);
1609 /* the desc just pretend to exclusive */
1610 ldlm_lock2desc(lock, &desc);
1611 desc.l_req_mode = LCK_EX;
1612 desc.l_granted_mode = 0;
1614 rc = lock->l_blocking_ast(lock, &desc, (void*)arg, LDLM_CB_BLOCKING);
1615 LDLM_LOCK_RELEASE(lock);
1620 int ldlm_run_ast_work(struct ldlm_namespace *ns, cfs_list_t *rpc_list,
1621 ldlm_desc_ast_t ast_type)
1623 struct ldlm_cb_set_arg *arg;
1624 set_producer_func work_ast_lock;
1627 if (cfs_list_empty(rpc_list))
1634 cfs_atomic_set(&arg->restart, 0);
1635 arg->list = rpc_list;
1638 case LDLM_WORK_BL_AST:
1639 arg->type = LDLM_BL_CALLBACK;
1640 work_ast_lock = ldlm_work_bl_ast_lock;
1642 case LDLM_WORK_CP_AST:
1643 arg->type = LDLM_CP_CALLBACK;
1644 work_ast_lock = ldlm_work_cp_ast_lock;
1646 case LDLM_WORK_REVOKE_AST:
1647 arg->type = LDLM_BL_CALLBACK;
1648 work_ast_lock = ldlm_work_revoke_ast_lock;
1654 /* We create a ptlrpc request set with flow control extension.
1655 * This request set will use the work_ast_lock function to produce new
1656 * requests and will send a new request each time one completes in order
1657 * to keep the number of requests in flight to ns_max_parallel_ast */
1658 arg->set = ptlrpc_prep_fcset(ns->ns_max_parallel_ast ? : UINT_MAX,
1659 work_ast_lock, arg);
1660 if (arg->set == NULL)
1661 GOTO(out, rc = -ENOMEM);
1663 ptlrpc_set_wait(arg->set);
1664 ptlrpc_set_destroy(arg->set);
1666 rc = cfs_atomic_read(&arg->restart) ? -ERESTART : 0;
1673 static int reprocess_one_queue(struct ldlm_resource *res, void *closure)
1675 ldlm_reprocess_all(res);
1676 return LDLM_ITER_CONTINUE;
1679 static int ldlm_reprocess_res(cfs_hash_t *hs, cfs_hash_bd_t *bd,
1680 cfs_hlist_node_t *hnode, void *arg)
1682 struct ldlm_resource *res = cfs_hash_object(hs, hnode);
1685 rc = reprocess_one_queue(res, arg);
1687 return rc == LDLM_ITER_STOP;
1690 void ldlm_reprocess_all_ns(struct ldlm_namespace *ns)
1695 cfs_hash_for_each_nolock(ns->ns_rs_hash,
1696 ldlm_reprocess_res, NULL);
1701 void ldlm_reprocess_all(struct ldlm_resource *res)
1703 CFS_LIST_HEAD(rpc_list);
1705 #ifdef HAVE_SERVER_SUPPORT
1708 /* Local lock trees don't get reprocessed. */
1709 if (ns_is_client(ldlm_res_to_ns(res))) {
1716 rc = ldlm_reprocess_queue(res, &res->lr_converting, &rpc_list);
1717 if (rc == LDLM_ITER_CONTINUE)
1718 ldlm_reprocess_queue(res, &res->lr_waiting, &rpc_list);
1721 rc = ldlm_run_ast_work(ldlm_res_to_ns(res), &rpc_list,
1723 if (rc == -ERESTART) {
1724 LASSERT(cfs_list_empty(&rpc_list));
1729 if (!ns_is_client(ldlm_res_to_ns(res))) {
1730 CERROR("This is client-side-only module, cannot handle "
1731 "LDLM_NAMESPACE_SERVER resource type lock.\n");
1738 void ldlm_cancel_callback(struct ldlm_lock *lock)
1740 check_res_locked(lock->l_resource);
1741 if (!(lock->l_flags & LDLM_FL_CANCEL)) {
1742 lock->l_flags |= LDLM_FL_CANCEL;
1743 if (lock->l_blocking_ast) {
1744 // l_check_no_ns_lock(ns);
1745 unlock_res_and_lock(lock);
1746 lock->l_blocking_ast(lock, NULL, lock->l_ast_data,
1748 lock_res_and_lock(lock);
1750 LDLM_DEBUG(lock, "no blocking ast");
1753 lock->l_flags |= LDLM_FL_BL_DONE;
1756 void ldlm_unlink_lock_skiplist(struct ldlm_lock *req)
1758 if (req->l_resource->lr_type != LDLM_PLAIN &&
1759 req->l_resource->lr_type != LDLM_IBITS)
1762 cfs_list_del_init(&req->l_sl_policy);
1763 cfs_list_del_init(&req->l_sl_mode);
1766 void ldlm_lock_cancel(struct ldlm_lock *lock)
1768 struct ldlm_resource *res;
1769 struct ldlm_namespace *ns;
1772 lock_res_and_lock(lock);
1774 res = lock->l_resource;
1775 ns = ldlm_res_to_ns(res);
1777 /* Please do not, no matter how tempting, remove this LBUG without
1778 * talking to me first. -phik */
1779 if (lock->l_readers || lock->l_writers) {
1780 LDLM_ERROR(lock, "lock still has references");
1784 ldlm_del_waiting_lock(lock);
1786 /* Releases cancel callback. */
1787 ldlm_cancel_callback(lock);
1789 /* Yes, second time, just in case it was added again while we were
1790 running with no res lock in ldlm_cancel_callback */
1791 ldlm_del_waiting_lock(lock);
1792 ldlm_resource_unlink_lock(lock);
1793 ldlm_lock_destroy_nolock(lock);
1795 if (lock->l_granted_mode == lock->l_req_mode)
1796 ldlm_pool_del(&ns->ns_pool, lock);
1798 /* Make sure we will not be called again for same lock what is possible
1799 * if not to zero out lock->l_granted_mode */
1800 lock->l_granted_mode = LCK_MINMODE;
1801 unlock_res_and_lock(lock);
1806 int ldlm_lock_set_data(struct lustre_handle *lockh, void *data)
1808 struct ldlm_lock *lock = ldlm_handle2lock(lockh);
1813 if (lock->l_ast_data == NULL)
1814 lock->l_ast_data = data;
1815 if (lock->l_ast_data == data)
1817 LDLM_LOCK_PUT(lock);
1821 EXPORT_SYMBOL(ldlm_lock_set_data);
1823 struct export_cl_data {
1824 struct obd_export *ecl_exp;
1828 int ldlm_cancel_locks_for_export_cb(cfs_hash_t *hs, cfs_hash_bd_t *bd,
1829 cfs_hlist_node_t *hnode, void *data)
1832 struct export_cl_data *ecl = (struct export_cl_data *)data;
1833 struct obd_export *exp = ecl->ecl_exp;
1834 struct ldlm_lock *lock = cfs_hash_object(hs, hnode);
1835 struct ldlm_resource *res;
1837 res = ldlm_resource_getref(lock->l_resource);
1838 LDLM_LOCK_GET(lock);
1840 LDLM_DEBUG(lock, "export %p", exp);
1841 ldlm_res_lvbo_update(res, NULL, 1);
1842 ldlm_lock_cancel(lock);
1843 ldlm_reprocess_all(res);
1844 ldlm_resource_putref(res);
1845 LDLM_LOCK_RELEASE(lock);
1848 if ((ecl->ecl_loop & -ecl->ecl_loop) == ecl->ecl_loop) {
1850 "Cancel lock %p for export %p (loop %d), still have "
1851 "%d locks left on hash table.\n",
1852 lock, exp, ecl->ecl_loop,
1853 cfs_atomic_read(&hs->hs_count));
1859 void ldlm_cancel_locks_for_export(struct obd_export *exp)
1861 struct export_cl_data ecl = {
1866 cfs_hash_for_each_empty(exp->exp_lock_hash,
1867 ldlm_cancel_locks_for_export_cb, &ecl);
1871 * Downgrade an exclusive lock.
1873 * A fast variant of ldlm_lock_convert for convertion of exclusive
1874 * locks. The convertion is always successful.
1876 * \param lock A lock to convert
1877 * \param new_mode new lock mode
1879 void ldlm_lock_downgrade(struct ldlm_lock *lock, int new_mode)
1883 LASSERT(lock->l_granted_mode & (LCK_PW | LCK_EX));
1884 LASSERT(new_mode == LCK_COS);
1886 lock_res_and_lock(lock);
1887 ldlm_resource_unlink_lock(lock);
1889 * Remove the lock from pool as it will be added again in
1890 * ldlm_grant_lock() called below.
1892 ldlm_pool_del(&ldlm_lock_to_ns(lock)->ns_pool, lock);
1894 lock->l_req_mode = new_mode;
1895 ldlm_grant_lock(lock, NULL);
1896 unlock_res_and_lock(lock);
1897 ldlm_reprocess_all(lock->l_resource);
1902 struct ldlm_resource *ldlm_lock_convert(struct ldlm_lock *lock, int new_mode,
1905 CFS_LIST_HEAD(rpc_list);
1906 struct ldlm_resource *res;
1907 struct ldlm_namespace *ns;
1909 #ifdef HAVE_SERVER_SUPPORT
1911 struct sl_insert_point prev;
1913 struct ldlm_interval *node;
1916 if (new_mode == lock->l_granted_mode) { // No changes? Just return.
1917 *flags |= LDLM_FL_BLOCK_GRANTED;
1918 RETURN(lock->l_resource);
1921 /* I can't check the type of lock here because the bitlock of lock
1922 * is not held here, so do the allocation blindly. -jay */
1923 OBD_SLAB_ALLOC_PTR_GFP(node, ldlm_interval_slab, CFS_ALLOC_IO);
1924 if (node == NULL) /* Actually, this causes EDEADLOCK to be returned */
1927 LASSERTF((new_mode == LCK_PW && lock->l_granted_mode == LCK_PR),
1928 "new_mode %u, granted %u\n", new_mode, lock->l_granted_mode);
1930 lock_res_and_lock(lock);
1932 res = lock->l_resource;
1933 ns = ldlm_res_to_ns(res);
1935 #ifdef HAVE_SERVER_SUPPORT
1936 old_mode = lock->l_req_mode;
1938 lock->l_req_mode = new_mode;
1939 if (res->lr_type == LDLM_PLAIN || res->lr_type == LDLM_IBITS) {
1940 #ifdef HAVE_SERVER_SUPPORT
1941 /* remember the lock position where the lock might be
1942 * added back to the granted list later and also
1943 * remember the join mode for skiplist fixing. */
1944 prev.res_link = lock->l_res_link.prev;
1945 prev.mode_link = lock->l_sl_mode.prev;
1946 prev.policy_link = lock->l_sl_policy.prev;
1948 ldlm_resource_unlink_lock(lock);
1950 ldlm_resource_unlink_lock(lock);
1951 if (res->lr_type == LDLM_EXTENT) {
1952 /* FIXME: ugly code, I have to attach the lock to a
1953 * interval node again since perhaps it will be granted
1955 CFS_INIT_LIST_HEAD(&node->li_group);
1956 ldlm_interval_attach(node, lock);
1962 * Remove old lock from the pool before adding the lock with new
1963 * mode below in ->policy()
1965 ldlm_pool_del(&ns->ns_pool, lock);
1967 /* If this is a local resource, put it on the appropriate list. */
1968 if (ns_is_client(ldlm_res_to_ns(res))) {
1969 if (*flags & (LDLM_FL_BLOCK_CONV | LDLM_FL_BLOCK_GRANTED)) {
1970 ldlm_resource_add_lock(res, &res->lr_converting, lock);
1972 /* This should never happen, because of the way the
1973 * server handles conversions. */
1974 LDLM_ERROR(lock, "Erroneous flags %d on local lock\n",
1978 ldlm_grant_lock(lock, &rpc_list);
1980 /* FIXME: completion handling not with lr_lock held ! */
1981 if (lock->l_completion_ast)
1982 lock->l_completion_ast(lock, 0, NULL);
1984 #ifdef HAVE_SERVER_SUPPORT
1989 ldlm_processing_policy policy;
1990 policy = ldlm_processing_policy_table[res->lr_type];
1991 rc = policy(lock, &pflags, 0, &err, &rpc_list);
1992 if (rc == LDLM_ITER_STOP) {
1993 lock->l_req_mode = old_mode;
1994 if (res->lr_type == LDLM_EXTENT)
1995 ldlm_extent_add_lock(res, lock);
1997 ldlm_granted_list_add_lock(lock, &prev);
2001 *flags |= LDLM_FL_BLOCK_GRANTED;
2007 CERROR("This is client-side-only module, cannot handle "
2008 "LDLM_NAMESPACE_SERVER resource type lock.\n");
2012 unlock_res_and_lock(lock);
2015 ldlm_run_ast_work(ns, &rpc_list, LDLM_WORK_CP_AST);
2017 OBD_SLAB_FREE(node, ldlm_interval_slab, sizeof(*node));
2021 void ldlm_lock_dump_handle(int level, struct lustre_handle *lockh)
2023 struct ldlm_lock *lock;
2025 if (!((libcfs_debug | D_ERROR) & level))
2028 lock = ldlm_handle2lock(lockh);
2032 LDLM_DEBUG_LIMIT(level, lock, "###");
2034 LDLM_LOCK_PUT(lock);
2037 void _ldlm_lock_debug(struct ldlm_lock *lock,
2038 struct libcfs_debug_msg_data *msgdata,
2039 const char *fmt, ...)
2042 struct obd_export *exp = lock->l_export;
2043 struct ldlm_resource *resource = lock->l_resource;
2044 char *nid = "local";
2046 va_start(args, fmt);
2048 if (exp && exp->exp_connection) {
2049 nid = libcfs_nid2str(exp->exp_connection->c_peer.nid);
2050 } else if (exp && exp->exp_obd != NULL) {
2051 struct obd_import *imp = exp->exp_obd->u.cli.cl_import;
2052 nid = libcfs_nid2str(imp->imp_connection->c_peer.nid);
2055 if (resource == NULL) {
2056 libcfs_debug_vmsg2(msgdata, fmt, args,
2057 " ns: \?\? lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
2058 "res: \?\? rrc=\?\? type: \?\?\? flags: "LPX64" nid: %s "
2059 "remote: "LPX64" expref: %d pid: %u timeout: %lu\n",
2061 lock->l_handle.h_cookie, cfs_atomic_read(&lock->l_refc),
2062 lock->l_readers, lock->l_writers,
2063 ldlm_lockname[lock->l_granted_mode],
2064 ldlm_lockname[lock->l_req_mode],
2065 lock->l_flags, nid, lock->l_remote_handle.cookie,
2066 exp ? cfs_atomic_read(&exp->exp_refcount) : -99,
2067 lock->l_pid, lock->l_callback_timeout);
2072 switch (resource->lr_type) {
2074 libcfs_debug_vmsg2(msgdata, fmt, args,
2075 " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
2076 "res: "LPU64"/"LPU64" rrc: %d type: %s ["LPU64"->"LPU64
2077 "] (req "LPU64"->"LPU64") flags: "LPX64" nid: %s remote:"
2078 " "LPX64" expref: %d pid: %u timeout %lu\n",
2079 ldlm_lock_to_ns_name(lock), lock,
2080 lock->l_handle.h_cookie, cfs_atomic_read(&lock->l_refc),
2081 lock->l_readers, lock->l_writers,
2082 ldlm_lockname[lock->l_granted_mode],
2083 ldlm_lockname[lock->l_req_mode],
2084 resource->lr_name.name[0],
2085 resource->lr_name.name[1],
2086 cfs_atomic_read(&resource->lr_refcount),
2087 ldlm_typename[resource->lr_type],
2088 lock->l_policy_data.l_extent.start,
2089 lock->l_policy_data.l_extent.end,
2090 lock->l_req_extent.start, lock->l_req_extent.end,
2091 lock->l_flags, nid, lock->l_remote_handle.cookie,
2092 exp ? cfs_atomic_read(&exp->exp_refcount) : -99,
2093 lock->l_pid, lock->l_callback_timeout);
2097 libcfs_debug_vmsg2(msgdata, fmt, args,
2098 " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
2099 "res: "LPU64"/"LPU64" rrc: %d type: %s pid: %d "
2100 "["LPU64"->"LPU64"] flags: "LPX64" nid: %s remote: "LPX64
2101 " expref: %d pid: %u timeout: %lu\n",
2102 ldlm_lock_to_ns_name(lock), lock,
2103 lock->l_handle.h_cookie, cfs_atomic_read(&lock->l_refc),
2104 lock->l_readers, lock->l_writers,
2105 ldlm_lockname[lock->l_granted_mode],
2106 ldlm_lockname[lock->l_req_mode],
2107 resource->lr_name.name[0],
2108 resource->lr_name.name[1],
2109 cfs_atomic_read(&resource->lr_refcount),
2110 ldlm_typename[resource->lr_type],
2111 lock->l_policy_data.l_flock.pid,
2112 lock->l_policy_data.l_flock.start,
2113 lock->l_policy_data.l_flock.end,
2114 lock->l_flags, nid, lock->l_remote_handle.cookie,
2115 exp ? cfs_atomic_read(&exp->exp_refcount) : -99,
2116 lock->l_pid, lock->l_callback_timeout);
2120 libcfs_debug_vmsg2(msgdata, fmt, args,
2121 " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
2122 "res: "LPU64"/"LPU64" bits "LPX64" rrc: %d type: %s "
2123 "flags: "LPX64" nid: %s remote: "LPX64" expref: %d "
2124 "pid: %u timeout: %lu\n",
2125 ldlm_lock_to_ns_name(lock),
2126 lock, lock->l_handle.h_cookie,
2127 cfs_atomic_read (&lock->l_refc),
2128 lock->l_readers, lock->l_writers,
2129 ldlm_lockname[lock->l_granted_mode],
2130 ldlm_lockname[lock->l_req_mode],
2131 resource->lr_name.name[0],
2132 resource->lr_name.name[1],
2133 lock->l_policy_data.l_inodebits.bits,
2134 cfs_atomic_read(&resource->lr_refcount),
2135 ldlm_typename[resource->lr_type],
2136 lock->l_flags, nid, lock->l_remote_handle.cookie,
2137 exp ? cfs_atomic_read(&exp->exp_refcount) : -99,
2138 lock->l_pid, lock->l_callback_timeout);
2142 libcfs_debug_vmsg2(msgdata, fmt, args,
2143 " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
2144 "res: "LPU64"/"LPU64" rrc: %d type: %s flags: "LPX64" "
2145 "nid: %s remote: "LPX64" expref: %d pid: %u timeout %lu"
2147 ldlm_lock_to_ns_name(lock),
2148 lock, lock->l_handle.h_cookie,
2149 cfs_atomic_read (&lock->l_refc),
2150 lock->l_readers, lock->l_writers,
2151 ldlm_lockname[lock->l_granted_mode],
2152 ldlm_lockname[lock->l_req_mode],
2153 resource->lr_name.name[0],
2154 resource->lr_name.name[1],
2155 cfs_atomic_read(&resource->lr_refcount),
2156 ldlm_typename[resource->lr_type],
2157 lock->l_flags, nid, lock->l_remote_handle.cookie,
2158 exp ? cfs_atomic_read(&exp->exp_refcount) : -99,
2159 lock->l_pid, lock->l_callback_timeout);
2164 EXPORT_SYMBOL(_ldlm_lock_debug);