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) 2010, 2013, Intel Corporation.
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
44 #include <libcfs/libcfs.h>
45 #include <obd_class.h>
46 #include "ldlm_internal.h"
49 char *ldlm_lockname[] = {
57 [LCK_GROUP] = "GROUP",
60 EXPORT_SYMBOL(ldlm_lockname);
62 char *ldlm_typename[] = {
64 [LDLM_EXTENT] = "EXT",
68 EXPORT_SYMBOL(ldlm_typename);
70 static ldlm_policy_wire_to_local_t ldlm_policy_wire18_to_local[] = {
71 [LDLM_PLAIN - LDLM_MIN_TYPE] = ldlm_plain_policy_wire_to_local,
72 [LDLM_EXTENT - LDLM_MIN_TYPE] = ldlm_extent_policy_wire_to_local,
73 [LDLM_FLOCK - LDLM_MIN_TYPE] = ldlm_flock_policy_wire18_to_local,
74 [LDLM_IBITS - LDLM_MIN_TYPE] = ldlm_ibits_policy_wire_to_local,
77 static ldlm_policy_wire_to_local_t ldlm_policy_wire21_to_local[] = {
78 [LDLM_PLAIN - LDLM_MIN_TYPE] = ldlm_plain_policy_wire_to_local,
79 [LDLM_EXTENT - LDLM_MIN_TYPE] = ldlm_extent_policy_wire_to_local,
80 [LDLM_FLOCK - LDLM_MIN_TYPE] = ldlm_flock_policy_wire21_to_local,
81 [LDLM_IBITS - LDLM_MIN_TYPE] = ldlm_ibits_policy_wire_to_local,
84 static ldlm_policy_local_to_wire_t ldlm_policy_local_to_wire[] = {
85 [LDLM_PLAIN - LDLM_MIN_TYPE] = ldlm_plain_policy_local_to_wire,
86 [LDLM_EXTENT - LDLM_MIN_TYPE] = ldlm_extent_policy_local_to_wire,
87 [LDLM_FLOCK - LDLM_MIN_TYPE] = ldlm_flock_policy_local_to_wire,
88 [LDLM_IBITS - LDLM_MIN_TYPE] = ldlm_ibits_policy_local_to_wire,
92 * Converts lock policy from local format to on the wire lock_desc format
94 void ldlm_convert_policy_to_wire(ldlm_type_t type,
95 const ldlm_policy_data_t *lpolicy,
96 ldlm_wire_policy_data_t *wpolicy)
98 ldlm_policy_local_to_wire_t convert;
100 convert = ldlm_policy_local_to_wire[type - LDLM_MIN_TYPE];
102 convert(lpolicy, wpolicy);
106 * Converts lock policy from on the wire lock_desc format to local format
108 void ldlm_convert_policy_to_local(struct obd_export *exp, ldlm_type_t type,
109 const ldlm_wire_policy_data_t *wpolicy,
110 ldlm_policy_data_t *lpolicy)
112 ldlm_policy_wire_to_local_t convert;
115 /** some badness for 2.0.0 clients, but 2.0.0 isn't supported */
116 new_client = (exp_connect_flags(exp) & OBD_CONNECT_FULL20) != 0;
118 convert = ldlm_policy_wire21_to_local[type - LDLM_MIN_TYPE];
120 convert = ldlm_policy_wire18_to_local[type - LDLM_MIN_TYPE];
122 convert(wpolicy, lpolicy);
125 char *ldlm_it2str(int it)
132 case (IT_OPEN | IT_CREAT):
147 CERROR("Unknown intent %d\n", it);
151 EXPORT_SYMBOL(ldlm_it2str);
153 extern struct kmem_cache *ldlm_lock_slab;
155 #ifdef HAVE_SERVER_SUPPORT
156 static ldlm_processing_policy ldlm_processing_policy_table[] = {
157 [LDLM_PLAIN] = ldlm_process_plain_lock,
158 [LDLM_EXTENT] = ldlm_process_extent_lock,
159 [LDLM_FLOCK] = ldlm_process_flock_lock,
160 [LDLM_IBITS] = ldlm_process_inodebits_lock,
163 ldlm_processing_policy ldlm_get_processing_policy(struct ldlm_resource *res)
165 return ldlm_processing_policy_table[res->lr_type];
167 EXPORT_SYMBOL(ldlm_get_processing_policy);
168 #endif /* HAVE_SERVER_SUPPORT */
170 void ldlm_register_intent(struct ldlm_namespace *ns, ldlm_res_policy arg)
174 EXPORT_SYMBOL(ldlm_register_intent);
177 * REFCOUNTED LOCK OBJECTS
182 * Get a reference on a lock.
184 * Lock refcounts, during creation:
185 * - one special one for allocation, dec'd only once in destroy
186 * - one for being a lock that's in-use
187 * - one for the addref associated with a new lock
189 struct ldlm_lock *ldlm_lock_get(struct ldlm_lock *lock)
191 atomic_inc(&lock->l_refc);
194 EXPORT_SYMBOL(ldlm_lock_get);
197 * Release lock reference.
199 * Also frees the lock if it was last reference.
201 void ldlm_lock_put(struct ldlm_lock *lock)
205 LASSERT(lock->l_resource != LP_POISON);
206 LASSERT(atomic_read(&lock->l_refc) > 0);
207 if (atomic_dec_and_test(&lock->l_refc)) {
208 struct ldlm_resource *res;
211 "final lock_put on destroyed lock, freeing it.");
213 res = lock->l_resource;
214 LASSERT(ldlm_is_destroyed(lock));
215 LASSERT(list_empty(&lock->l_res_link));
216 LASSERT(list_empty(&lock->l_pending_chain));
218 lprocfs_counter_decr(ldlm_res_to_ns(res)->ns_stats,
220 lu_ref_del(&res->lr_reference, "lock", lock);
221 ldlm_resource_putref(res);
222 lock->l_resource = NULL;
223 if (lock->l_export) {
224 class_export_lock_put(lock->l_export, lock);
225 lock->l_export = NULL;
228 if (lock->l_lvb_data != NULL)
229 OBD_FREE_LARGE(lock->l_lvb_data, lock->l_lvb_len);
231 ldlm_interval_free(ldlm_interval_detach(lock));
232 lu_ref_fini(&lock->l_reference);
233 OBD_FREE_RCU(lock, sizeof(*lock), &lock->l_handle);
238 EXPORT_SYMBOL(ldlm_lock_put);
241 * Removes LDLM lock \a lock from LRU. Assumes LRU is already locked.
243 int ldlm_lock_remove_from_lru_nolock(struct ldlm_lock *lock)
246 if (!list_empty(&lock->l_lru)) {
247 struct ldlm_namespace *ns = ldlm_lock_to_ns(lock);
249 LASSERT(lock->l_resource->lr_type != LDLM_FLOCK);
250 list_del_init(&lock->l_lru);
251 LASSERT(ns->ns_nr_unused > 0);
259 * Removes LDLM lock \a lock from LRU. Obtains the LRU lock first.
261 int ldlm_lock_remove_from_lru(struct ldlm_lock *lock)
263 struct ldlm_namespace *ns = ldlm_lock_to_ns(lock);
267 if (ldlm_is_ns_srv(lock)) {
268 LASSERT(list_empty(&lock->l_lru));
272 spin_lock(&ns->ns_lock);
273 rc = ldlm_lock_remove_from_lru_nolock(lock);
274 spin_unlock(&ns->ns_lock);
280 * Adds LDLM lock \a lock to namespace LRU. Assumes LRU is already locked.
282 void ldlm_lock_add_to_lru_nolock(struct ldlm_lock *lock)
284 struct ldlm_namespace *ns = ldlm_lock_to_ns(lock);
286 lock->l_last_used = cfs_time_current();
287 LASSERT(list_empty(&lock->l_lru));
288 LASSERT(lock->l_resource->lr_type != LDLM_FLOCK);
289 list_add_tail(&lock->l_lru, &ns->ns_unused_list);
290 ldlm_clear_skipped(lock);
291 LASSERT(ns->ns_nr_unused >= 0);
296 * Adds LDLM lock \a lock to namespace LRU. Obtains necessary LRU locks
299 void ldlm_lock_add_to_lru(struct ldlm_lock *lock)
301 struct ldlm_namespace *ns = ldlm_lock_to_ns(lock);
304 spin_lock(&ns->ns_lock);
305 ldlm_lock_add_to_lru_nolock(lock);
306 spin_unlock(&ns->ns_lock);
311 * Moves LDLM lock \a lock that is already in namespace LRU to the tail of
312 * the LRU. Performs necessary LRU locking
314 void ldlm_lock_touch_in_lru(struct ldlm_lock *lock)
316 struct ldlm_namespace *ns = ldlm_lock_to_ns(lock);
319 if (ldlm_is_ns_srv(lock)) {
320 LASSERT(list_empty(&lock->l_lru));
325 spin_lock(&ns->ns_lock);
326 if (!list_empty(&lock->l_lru)) {
327 ldlm_lock_remove_from_lru_nolock(lock);
328 ldlm_lock_add_to_lru_nolock(lock);
330 spin_unlock(&ns->ns_lock);
335 * Helper to destroy a locked lock.
337 * Used by ldlm_lock_destroy and ldlm_lock_destroy_nolock
338 * Must be called with l_lock and lr_lock held.
340 * Does not actually free the lock data, but rather marks the lock as
341 * destroyed by setting l_destroyed field in the lock to 1. Destroys a
342 * handle->lock association too, so that the lock can no longer be found
343 * and removes the lock from LRU list. Actual lock freeing occurs when
344 * last lock reference goes away.
346 * Original comment (of some historical value):
347 * This used to have a 'strict' flag, which recovery would use to mark an
348 * in-use lock as needing-to-die. Lest I am ever tempted to put it back, I
349 * shall explain why it's gone: with the new hash table scheme, once you call
350 * ldlm_lock_destroy, you can never drop your final references on this lock.
351 * Because it's not in the hash table anymore. -phil
353 static int ldlm_lock_destroy_internal(struct ldlm_lock *lock)
357 if (lock->l_readers || lock->l_writers) {
358 LDLM_ERROR(lock, "lock still has references");
362 if (!list_empty(&lock->l_res_link)) {
363 LDLM_ERROR(lock, "lock still on resource");
367 if (ldlm_is_destroyed(lock)) {
368 LASSERT(list_empty(&lock->l_lru));
372 ldlm_set_destroyed(lock);
374 if (lock->l_export && lock->l_export->exp_lock_hash) {
375 /* NB: it's safe to call cfs_hash_del() even lock isn't
376 * in exp_lock_hash. */
377 /* In the function below, .hs_keycmp resolves to
378 * ldlm_export_lock_keycmp() */
379 /* coverity[overrun-buffer-val] */
380 cfs_hash_del(lock->l_export->exp_lock_hash,
381 &lock->l_remote_handle, &lock->l_exp_hash);
384 ldlm_lock_remove_from_lru(lock);
385 class_handle_unhash(&lock->l_handle);
388 /* Wake anyone waiting for this lock */
389 /* FIXME: I should probably add yet another flag, instead of using
390 * l_export to only call this on clients */
392 class_export_put(lock->l_export);
393 lock->l_export = NULL;
394 if (lock->l_export && lock->l_completion_ast)
395 lock->l_completion_ast(lock, 0);
402 * Destroys a LDLM lock \a lock. Performs necessary locking first.
404 void ldlm_lock_destroy(struct ldlm_lock *lock)
408 lock_res_and_lock(lock);
409 first = ldlm_lock_destroy_internal(lock);
410 unlock_res_and_lock(lock);
412 /* drop reference from hashtable only for first destroy */
414 lu_ref_del(&lock->l_reference, "hash", lock);
415 LDLM_LOCK_RELEASE(lock);
421 * Destroys a LDLM lock \a lock that is already locked.
423 void ldlm_lock_destroy_nolock(struct ldlm_lock *lock)
427 first = ldlm_lock_destroy_internal(lock);
428 /* drop reference from hashtable only for first destroy */
430 lu_ref_del(&lock->l_reference, "hash", lock);
431 LDLM_LOCK_RELEASE(lock);
436 /* this is called by portals_handle2object with the handle lock taken */
437 static void lock_handle_addref(void *lock)
439 LDLM_LOCK_GET((struct ldlm_lock *)lock);
442 static void lock_handle_free(void *lock, int size)
444 LASSERT(size == sizeof(struct ldlm_lock));
445 OBD_SLAB_FREE(lock, ldlm_lock_slab, size);
448 static struct portals_handle_ops lock_handle_ops = {
449 .hop_addref = lock_handle_addref,
450 .hop_free = lock_handle_free,
455 * Allocate and initialize new lock structure.
457 * usage: pass in a resource on which you have done ldlm_resource_get
458 * new lock will take over the refcount.
459 * returns: lock with refcount 2 - one for current caller and one for remote
461 static struct ldlm_lock *ldlm_lock_new(struct ldlm_resource *resource)
463 struct ldlm_lock *lock;
466 if (resource == NULL)
469 OBD_SLAB_ALLOC_PTR_GFP(lock, ldlm_lock_slab, GFP_NOFS);
473 spin_lock_init(&lock->l_lock);
474 lock->l_resource = resource;
475 lu_ref_add(&resource->lr_reference, "lock", lock);
477 atomic_set(&lock->l_refc, 2);
478 INIT_LIST_HEAD(&lock->l_res_link);
479 INIT_LIST_HEAD(&lock->l_lru);
480 INIT_LIST_HEAD(&lock->l_pending_chain);
481 INIT_LIST_HEAD(&lock->l_bl_ast);
482 INIT_LIST_HEAD(&lock->l_cp_ast);
483 INIT_LIST_HEAD(&lock->l_rk_ast);
484 init_waitqueue_head(&lock->l_waitq);
485 lock->l_blocking_lock = NULL;
486 INIT_LIST_HEAD(&lock->l_sl_mode);
487 INIT_LIST_HEAD(&lock->l_sl_policy);
488 INIT_HLIST_NODE(&lock->l_exp_hash);
489 INIT_HLIST_NODE(&lock->l_exp_flock_hash);
491 lprocfs_counter_incr(ldlm_res_to_ns(resource)->ns_stats,
493 INIT_LIST_HEAD(&lock->l_handle.h_link);
494 class_handle_hash(&lock->l_handle, &lock_handle_ops);
496 lu_ref_init(&lock->l_reference);
497 lu_ref_add(&lock->l_reference, "hash", lock);
498 lock->l_callback_timeout = 0;
500 #if LUSTRE_TRACKS_LOCK_EXP_REFS
501 INIT_LIST_HEAD(&lock->l_exp_refs_link);
502 lock->l_exp_refs_nr = 0;
503 lock->l_exp_refs_target = NULL;
505 INIT_LIST_HEAD(&lock->l_exp_list);
511 * Moves LDLM lock \a lock to another resource.
512 * This is used on client when server returns some other lock than requested
513 * (typically as a result of intent operation)
515 int ldlm_lock_change_resource(struct ldlm_namespace *ns, struct ldlm_lock *lock,
516 const struct ldlm_res_id *new_resid)
518 struct ldlm_resource *oldres = lock->l_resource;
519 struct ldlm_resource *newres;
523 LASSERT(ns_is_client(ns));
525 lock_res_and_lock(lock);
526 if (memcmp(new_resid, &lock->l_resource->lr_name,
527 sizeof(lock->l_resource->lr_name)) == 0) {
529 unlock_res_and_lock(lock);
533 LASSERT(new_resid->name[0] != 0);
535 /* This function assumes that the lock isn't on any lists */
536 LASSERT(list_empty(&lock->l_res_link));
538 type = oldres->lr_type;
539 unlock_res_and_lock(lock);
541 newres = ldlm_resource_get(ns, NULL, new_resid, type, 1);
543 RETURN(PTR_ERR(newres));
545 lu_ref_add(&newres->lr_reference, "lock", lock);
547 * To flip the lock from the old to the new resource, lock, oldres and
548 * newres have to be locked. Resource spin-locks are nested within
549 * lock->l_lock, and are taken in the memory address order to avoid
552 spin_lock(&lock->l_lock);
553 oldres = lock->l_resource;
554 if (oldres < newres) {
556 lock_res_nested(newres, LRT_NEW);
559 lock_res_nested(oldres, LRT_NEW);
561 LASSERT(memcmp(new_resid, &oldres->lr_name,
562 sizeof oldres->lr_name) != 0);
563 lock->l_resource = newres;
565 unlock_res_and_lock(lock);
567 /* ...and the flowers are still standing! */
568 lu_ref_del(&oldres->lr_reference, "lock", lock);
569 ldlm_resource_putref(oldres);
573 EXPORT_SYMBOL(ldlm_lock_change_resource);
575 /** \defgroup ldlm_handles LDLM HANDLES
576 * Ways to get hold of locks without any addresses.
581 * Fills in handle for LDLM lock \a lock into supplied \a lockh
582 * Does not take any references.
584 void ldlm_lock2handle(const struct ldlm_lock *lock, struct lustre_handle *lockh)
586 lockh->cookie = lock->l_handle.h_cookie;
588 EXPORT_SYMBOL(ldlm_lock2handle);
591 * Obtain a lock reference by handle.
593 * if \a flags: atomically get the lock and set the flags.
594 * Return NULL if flag already set
596 struct ldlm_lock *__ldlm_handle2lock(const struct lustre_handle *handle,
599 struct ldlm_lock *lock;
604 lock = class_handle2object(handle->cookie, NULL);
608 /* It's unlikely but possible that someone marked the lock as
609 * destroyed after we did handle2object on it */
610 if ((flags == 0) && !ldlm_is_destroyed(lock)) {
611 lu_ref_add(&lock->l_reference, "handle", current);
615 lock_res_and_lock(lock);
617 LASSERT(lock->l_resource != NULL);
619 lu_ref_add_atomic(&lock->l_reference, "handle", current);
620 if (unlikely(ldlm_is_destroyed(lock))) {
621 unlock_res_and_lock(lock);
622 CDEBUG(D_INFO, "lock already destroyed: lock %p\n", lock);
627 /* If we're setting flags, make sure none of them are already set. */
629 if ((lock->l_flags & flags) != 0) {
630 unlock_res_and_lock(lock);
635 lock->l_flags |= flags;
638 unlock_res_and_lock(lock);
641 EXPORT_SYMBOL(__ldlm_handle2lock);
642 /** @} ldlm_handles */
645 * Fill in "on the wire" representation for given LDLM lock into supplied
646 * lock descriptor \a desc structure.
648 void ldlm_lock2desc(struct ldlm_lock *lock, struct ldlm_lock_desc *desc)
650 ldlm_res2desc(lock->l_resource, &desc->l_resource);
651 desc->l_req_mode = lock->l_req_mode;
652 desc->l_granted_mode = lock->l_granted_mode;
653 ldlm_convert_policy_to_wire(lock->l_resource->lr_type,
654 &lock->l_policy_data,
655 &desc->l_policy_data);
657 EXPORT_SYMBOL(ldlm_lock2desc);
660 * Add a lock to list of conflicting locks to send AST to.
662 * Only add if we have not sent a blocking AST to the lock yet.
664 static void ldlm_add_bl_work_item(struct ldlm_lock *lock, struct ldlm_lock *new,
665 struct list_head *work_list)
667 if (!ldlm_is_ast_sent(lock)) {
668 LDLM_DEBUG(lock, "lock incompatible; sending blocking AST.");
669 ldlm_set_ast_sent(lock);
670 /* If the enqueuing client said so, tell the AST recipient to
671 * discard dirty data, rather than writing back. */
672 if (ldlm_is_ast_discard_data(new))
673 ldlm_set_discard_data(lock);
674 LASSERT(list_empty(&lock->l_bl_ast));
675 list_add(&lock->l_bl_ast, work_list);
677 LASSERT(lock->l_blocking_lock == NULL);
678 lock->l_blocking_lock = LDLM_LOCK_GET(new);
683 * Add a lock to list of just granted locks to send completion AST to.
685 static void ldlm_add_cp_work_item(struct ldlm_lock *lock,
686 struct list_head *work_list)
688 if (!ldlm_is_cp_reqd(lock)) {
689 ldlm_set_cp_reqd(lock);
690 LDLM_DEBUG(lock, "lock granted; sending completion AST.");
691 LASSERT(list_empty(&lock->l_cp_ast));
692 list_add(&lock->l_cp_ast, work_list);
698 * Aggregator function to add AST work items into a list. Determines
699 * what sort of an AST work needs to be done and calls the proper
701 * Must be called with lr_lock held.
703 void ldlm_add_ast_work_item(struct ldlm_lock *lock, struct ldlm_lock *new,
704 struct list_head *work_list)
707 check_res_locked(lock->l_resource);
709 ldlm_add_bl_work_item(lock, new, work_list);
711 ldlm_add_cp_work_item(lock, work_list);
716 * Add specified reader/writer reference to LDLM lock with handle \a lockh.
717 * r/w reference type is determined by \a mode
718 * Calls ldlm_lock_addref_internal.
720 void ldlm_lock_addref(struct lustre_handle *lockh, __u32 mode)
722 struct ldlm_lock *lock;
724 lock = ldlm_handle2lock(lockh);
725 LASSERTF(lock != NULL, "Non-existing lock: "LPX64"\n", lockh->cookie);
726 ldlm_lock_addref_internal(lock, mode);
729 EXPORT_SYMBOL(ldlm_lock_addref);
733 * Add specified reader/writer reference to LDLM lock \a lock.
734 * r/w reference type is determined by \a mode
735 * Removes lock from LRU if it is there.
736 * Assumes the LDLM lock is already locked.
738 void ldlm_lock_addref_internal_nolock(struct ldlm_lock *lock, __u32 mode)
740 ldlm_lock_remove_from_lru(lock);
741 if (mode & (LCK_NL | LCK_CR | LCK_PR)) {
743 lu_ref_add_atomic(&lock->l_reference, "reader", lock);
745 if (mode & (LCK_EX | LCK_CW | LCK_PW | LCK_GROUP | LCK_COS)) {
747 lu_ref_add_atomic(&lock->l_reference, "writer", lock);
750 lu_ref_add_atomic(&lock->l_reference, "user", lock);
751 LDLM_DEBUG(lock, "ldlm_lock_addref(%s)", ldlm_lockname[mode]);
755 * Attempts to add reader/writer reference to a lock with handle \a lockh, and
756 * fails if lock is already LDLM_FL_CBPENDING or destroyed.
758 * \retval 0 success, lock was addref-ed
760 * \retval -EAGAIN lock is being canceled.
762 int ldlm_lock_addref_try(struct lustre_handle *lockh, __u32 mode)
764 struct ldlm_lock *lock;
768 lock = ldlm_handle2lock(lockh);
770 lock_res_and_lock(lock);
771 if (lock->l_readers != 0 || lock->l_writers != 0 ||
772 !ldlm_is_cbpending(lock)) {
773 ldlm_lock_addref_internal_nolock(lock, mode);
776 unlock_res_and_lock(lock);
781 EXPORT_SYMBOL(ldlm_lock_addref_try);
784 * Add specified reader/writer reference to LDLM lock \a lock.
785 * Locks LDLM lock and calls ldlm_lock_addref_internal_nolock to do the work.
786 * Only called for local locks.
788 void ldlm_lock_addref_internal(struct ldlm_lock *lock, __u32 mode)
790 lock_res_and_lock(lock);
791 ldlm_lock_addref_internal_nolock(lock, mode);
792 unlock_res_and_lock(lock);
796 * Removes reader/writer reference for LDLM lock \a lock.
797 * Assumes LDLM lock is already locked.
798 * only called in ldlm_flock_destroy and for local locks.
799 * Does NOT add lock to LRU if no r/w references left to accomodate flock locks
800 * that cannot be placed in LRU.
802 void ldlm_lock_decref_internal_nolock(struct ldlm_lock *lock, __u32 mode)
804 LDLM_DEBUG(lock, "ldlm_lock_decref(%s)", ldlm_lockname[mode]);
805 if (mode & (LCK_NL | LCK_CR | LCK_PR)) {
806 LASSERT(lock->l_readers > 0);
807 lu_ref_del(&lock->l_reference, "reader", lock);
810 if (mode & (LCK_EX | LCK_CW | LCK_PW | LCK_GROUP | LCK_COS)) {
811 LASSERT(lock->l_writers > 0);
812 lu_ref_del(&lock->l_reference, "writer", lock);
816 lu_ref_del(&lock->l_reference, "user", lock);
817 LDLM_LOCK_RELEASE(lock); /* matches the LDLM_LOCK_GET() in addref */
821 * Removes reader/writer reference for LDLM lock \a lock.
822 * Locks LDLM lock first.
823 * If the lock is determined to be client lock on a client and r/w refcount
824 * drops to zero and the lock is not blocked, the lock is added to LRU lock
826 * For blocked LDLM locks if r/w count drops to zero, blocking_ast is called.
828 void ldlm_lock_decref_internal(struct ldlm_lock *lock, __u32 mode)
830 struct ldlm_namespace *ns;
833 lock_res_and_lock(lock);
835 ns = ldlm_lock_to_ns(lock);
837 ldlm_lock_decref_internal_nolock(lock, mode);
839 if (ldlm_is_local(lock) &&
840 !lock->l_readers && !lock->l_writers) {
841 /* If this is a local lock on a server namespace and this was
842 * the last reference, cancel the lock. */
843 CDEBUG(D_INFO, "forcing cancel of local lock\n");
844 ldlm_set_cbpending(lock);
847 if (!lock->l_readers && !lock->l_writers &&
848 ldlm_is_cbpending(lock)) {
849 /* If we received a blocked AST and this was the last reference,
850 * run the callback. */
851 if (ldlm_is_ns_srv(lock) && lock->l_export)
852 CERROR("FL_CBPENDING set on non-local lock--just a "
855 LDLM_DEBUG(lock, "final decref done on cbpending lock");
857 LDLM_LOCK_GET(lock); /* dropped by bl thread */
858 ldlm_lock_remove_from_lru(lock);
859 unlock_res_and_lock(lock);
861 if (ldlm_is_fail_loc(lock))
862 OBD_RACE(OBD_FAIL_LDLM_CP_BL_RACE);
864 if (ldlm_is_atomic_cb(lock) ||
865 ldlm_bl_to_thread_lock(ns, NULL, lock) != 0)
866 ldlm_handle_bl_callback(ns, NULL, lock);
867 } else if (ns_is_client(ns) &&
868 !lock->l_readers && !lock->l_writers &&
869 !ldlm_is_no_lru(lock) &&
870 !ldlm_is_bl_ast(lock)) {
872 LDLM_DEBUG(lock, "add lock into lru list");
874 /* If this is a client-side namespace and this was the last
875 * reference, put it on the LRU. */
876 ldlm_lock_add_to_lru(lock);
877 unlock_res_and_lock(lock);
879 if (ldlm_is_fail_loc(lock))
880 OBD_RACE(OBD_FAIL_LDLM_CP_BL_RACE);
882 /* Call ldlm_cancel_lru() only if EARLY_CANCEL and LRU RESIZE
883 * are not supported by the server, otherwise, it is done on
885 if (!exp_connect_cancelset(lock->l_conn_export) &&
886 !ns_connect_lru_resize(ns))
887 ldlm_cancel_lru(ns, 0, LCF_ASYNC, 0);
889 LDLM_DEBUG(lock, "do not add lock into lru list");
890 unlock_res_and_lock(lock);
897 * Decrease reader/writer refcount for LDLM lock with handle \a lockh
899 void ldlm_lock_decref(struct lustre_handle *lockh, __u32 mode)
901 struct ldlm_lock *lock = __ldlm_handle2lock(lockh, 0);
902 LASSERTF(lock != NULL, "Non-existing lock: "LPX64"\n", lockh->cookie);
903 ldlm_lock_decref_internal(lock, mode);
906 EXPORT_SYMBOL(ldlm_lock_decref);
909 * Decrease reader/writer refcount for LDLM lock with handle
910 * \a lockh and mark it for subsequent cancellation once r/w refcount
911 * drops to zero instead of putting into LRU.
913 * Typical usage is for GROUP locks which we cannot allow to be cached.
915 void ldlm_lock_decref_and_cancel(struct lustre_handle *lockh, __u32 mode)
917 struct ldlm_lock *lock = __ldlm_handle2lock(lockh, 0);
920 LASSERT(lock != NULL);
922 LDLM_DEBUG(lock, "ldlm_lock_decref(%s)", ldlm_lockname[mode]);
923 lock_res_and_lock(lock);
924 ldlm_set_cbpending(lock);
925 unlock_res_and_lock(lock);
926 ldlm_lock_decref_internal(lock, mode);
929 EXPORT_SYMBOL(ldlm_lock_decref_and_cancel);
931 struct sl_insert_point {
932 struct list_head *res_link;
933 struct list_head *mode_link;
934 struct list_head *policy_link;
938 * Finds a position to insert the new lock into granted lock list.
940 * Used for locks eligible for skiplist optimization.
943 * queue [input]: the granted list where search acts on;
944 * req [input]: the lock whose position to be located;
945 * prev [output]: positions within 3 lists to insert @req to
949 * - ldlm_grant_lock_with_skiplist
951 static void search_granted_lock(struct list_head *queue,
952 struct ldlm_lock *req,
953 struct sl_insert_point *prev)
955 struct list_head *tmp;
956 struct ldlm_lock *lock, *mode_end, *policy_end;
959 list_for_each(tmp, queue) {
960 lock = list_entry(tmp, struct ldlm_lock, l_res_link);
962 mode_end = list_entry(lock->l_sl_mode.prev,
963 struct ldlm_lock, l_sl_mode);
965 if (lock->l_req_mode != req->l_req_mode) {
966 /* jump to last lock of mode group */
967 tmp = &mode_end->l_res_link;
971 /* suitable mode group is found */
972 if (lock->l_resource->lr_type == LDLM_PLAIN) {
973 /* insert point is last lock of the mode group */
974 prev->res_link = &mode_end->l_res_link;
975 prev->mode_link = &mode_end->l_sl_mode;
976 prev->policy_link = &req->l_sl_policy;
979 } else if (lock->l_resource->lr_type == LDLM_IBITS) {
982 list_entry(lock->l_sl_policy.prev,
986 if (lock->l_policy_data.l_inodebits.bits ==
987 req->l_policy_data.l_inodebits.bits) {
988 /* insert point is last lock of
989 * the policy group */
991 &policy_end->l_res_link;
993 &policy_end->l_sl_mode;
995 &policy_end->l_sl_policy;
1000 if (policy_end == mode_end)
1001 /* done with mode group */
1004 /* go to next policy group within mode group */
1005 tmp = policy_end->l_res_link.next;
1006 lock = list_entry(tmp, struct ldlm_lock,
1008 } /* loop over policy groups within the mode group */
1010 /* insert point is last lock of the mode group,
1011 * new policy group is started */
1012 prev->res_link = &mode_end->l_res_link;
1013 prev->mode_link = &mode_end->l_sl_mode;
1014 prev->policy_link = &req->l_sl_policy;
1018 LDLM_ERROR(lock,"is not LDLM_PLAIN or LDLM_IBITS lock");
1023 /* insert point is last lock on the queue,
1024 * new mode group and new policy group are started */
1025 prev->res_link = queue->prev;
1026 prev->mode_link = &req->l_sl_mode;
1027 prev->policy_link = &req->l_sl_policy;
1033 * Add a lock into resource granted list after a position described by
1036 static void ldlm_granted_list_add_lock(struct ldlm_lock *lock,
1037 struct sl_insert_point *prev)
1039 struct ldlm_resource *res = lock->l_resource;
1042 check_res_locked(res);
1044 ldlm_resource_dump(D_INFO, res);
1045 LDLM_DEBUG(lock, "About to add lock:");
1047 if (ldlm_is_destroyed(lock)) {
1048 CDEBUG(D_OTHER, "Lock destroyed, not adding to resource\n");
1052 LASSERT(list_empty(&lock->l_res_link));
1053 LASSERT(list_empty(&lock->l_sl_mode));
1054 LASSERT(list_empty(&lock->l_sl_policy));
1057 * lock->link == prev->link means lock is first starting the group.
1058 * Don't re-add to itself to suppress kernel warnings.
1060 if (&lock->l_res_link != prev->res_link)
1061 list_add(&lock->l_res_link, prev->res_link);
1062 if (&lock->l_sl_mode != prev->mode_link)
1063 list_add(&lock->l_sl_mode, prev->mode_link);
1064 if (&lock->l_sl_policy != prev->policy_link)
1065 list_add(&lock->l_sl_policy, prev->policy_link);
1071 * Add a lock to granted list on a resource maintaining skiplist
1074 static void ldlm_grant_lock_with_skiplist(struct ldlm_lock *lock)
1076 struct sl_insert_point prev;
1079 LASSERT(lock->l_req_mode == lock->l_granted_mode);
1081 search_granted_lock(&lock->l_resource->lr_granted, lock, &prev);
1082 ldlm_granted_list_add_lock(lock, &prev);
1087 * Perform lock granting bookkeeping.
1089 * Includes putting the lock into granted list and updating lock mode.
1091 * - ldlm_lock_enqueue
1092 * - ldlm_reprocess_queue
1093 * - ldlm_lock_convert
1095 * must be called with lr_lock held
1097 void ldlm_grant_lock(struct ldlm_lock *lock, struct list_head *work_list)
1099 struct ldlm_resource *res = lock->l_resource;
1102 check_res_locked(res);
1104 lock->l_granted_mode = lock->l_req_mode;
1106 if (work_list && lock->l_completion_ast != NULL)
1107 ldlm_add_ast_work_item(lock, NULL, work_list);
1109 /* We should not add locks to granted list in the following cases:
1110 * - this is an UNLOCK but not a real lock;
1111 * - this is a TEST lock;
1112 * - this is a F_CANCELLK lock (async flock has req_mode == 0)
1113 * - this is a deadlock (flock cannot be granted) */
1114 if (lock->l_req_mode == 0 ||
1115 lock->l_req_mode == LCK_NL ||
1116 ldlm_is_test_lock(lock) ||
1117 ldlm_is_flock_deadlock(lock))
1120 if (res->lr_type == LDLM_PLAIN || res->lr_type == LDLM_IBITS)
1121 ldlm_grant_lock_with_skiplist(lock);
1122 else if (res->lr_type == LDLM_EXTENT)
1123 ldlm_extent_add_lock(res, lock);
1125 ldlm_resource_add_lock(res, &res->lr_granted, lock);
1127 if (lock->l_granted_mode < res->lr_most_restr)
1128 res->lr_most_restr = lock->l_granted_mode;
1130 ldlm_pool_add(&ldlm_res_to_ns(res)->ns_pool, lock);
1135 * Search for a lock with given properties in a queue.
1137 * \retval a referenced lock or NULL. See the flag descriptions below, in the
1138 * comment above ldlm_lock_match
1140 static struct ldlm_lock *search_queue(struct list_head *queue,
1142 ldlm_policy_data_t *policy,
1143 struct ldlm_lock *old_lock,
1144 __u64 flags, int unref)
1146 struct ldlm_lock *lock;
1147 struct list_head *tmp;
1149 list_for_each(tmp, queue) {
1152 lock = list_entry(tmp, struct ldlm_lock, l_res_link);
1154 if (lock == old_lock)
1157 /* Check if this lock can be matched.
1158 * Used by LU-2919(exclusive open) for open lease lock */
1159 if (ldlm_is_excl(lock))
1162 /* llite sometimes wants to match locks that will be
1163 * canceled when their users drop, but we allow it to match
1164 * if it passes in CBPENDING and the lock still has users.
1165 * this is generally only going to be used by children
1166 * whose parents already hold a lock so forward progress
1167 * can still happen. */
1168 if (ldlm_is_cbpending(lock) &&
1169 !(flags & LDLM_FL_CBPENDING))
1171 if (!unref && ldlm_is_cbpending(lock) &&
1172 lock->l_readers == 0 && lock->l_writers == 0)
1175 if (!(lock->l_req_mode & *mode))
1177 match = lock->l_req_mode;
1179 if (lock->l_resource->lr_type == LDLM_EXTENT &&
1180 (lock->l_policy_data.l_extent.start >
1181 policy->l_extent.start ||
1182 lock->l_policy_data.l_extent.end < policy->l_extent.end))
1185 if (unlikely(match == LCK_GROUP) &&
1186 lock->l_resource->lr_type == LDLM_EXTENT &&
1187 policy->l_extent.gid != LDLM_GID_ANY &&
1188 lock->l_policy_data.l_extent.gid != policy->l_extent.gid)
1191 /* We match if we have existing lock with same or wider set
1193 if (lock->l_resource->lr_type == LDLM_IBITS &&
1194 ((lock->l_policy_data.l_inodebits.bits &
1195 policy->l_inodebits.bits) !=
1196 policy->l_inodebits.bits))
1199 if (!unref && LDLM_HAVE_MASK(lock, GONE))
1202 if ((flags & LDLM_FL_LOCAL_ONLY) &&
1203 !ldlm_is_local(lock))
1206 if (flags & LDLM_FL_TEST_LOCK) {
1207 LDLM_LOCK_GET(lock);
1208 ldlm_lock_touch_in_lru(lock);
1210 ldlm_lock_addref_internal_nolock(lock, match);
1219 void ldlm_lock_fail_match_locked(struct ldlm_lock *lock)
1221 if ((lock->l_flags & LDLM_FL_FAIL_NOTIFIED) == 0) {
1222 lock->l_flags |= LDLM_FL_FAIL_NOTIFIED;
1223 wake_up_all(&lock->l_waitq);
1226 EXPORT_SYMBOL(ldlm_lock_fail_match_locked);
1228 void ldlm_lock_fail_match(struct ldlm_lock *lock)
1230 lock_res_and_lock(lock);
1231 ldlm_lock_fail_match_locked(lock);
1232 unlock_res_and_lock(lock);
1234 EXPORT_SYMBOL(ldlm_lock_fail_match);
1237 * Mark lock as "matchable" by OST.
1239 * Used to prevent certain races in LOV/OSC where the lock is granted, but LVB
1241 * Assumes LDLM lock is already locked.
1243 void ldlm_lock_allow_match_locked(struct ldlm_lock *lock)
1245 ldlm_set_lvb_ready(lock);
1246 wake_up_all(&lock->l_waitq);
1248 EXPORT_SYMBOL(ldlm_lock_allow_match_locked);
1251 * Mark lock as "matchable" by OST.
1252 * Locks the lock and then \see ldlm_lock_allow_match_locked
1254 void ldlm_lock_allow_match(struct ldlm_lock *lock)
1256 lock_res_and_lock(lock);
1257 ldlm_lock_allow_match_locked(lock);
1258 unlock_res_and_lock(lock);
1260 EXPORT_SYMBOL(ldlm_lock_allow_match);
1263 * Attempt to find a lock with specified properties.
1265 * Typically returns a reference to matched lock unless LDLM_FL_TEST_LOCK is
1268 * Can be called in two ways:
1270 * If 'ns' is NULL, then lockh describes an existing lock that we want to look
1271 * for a duplicate of.
1273 * Otherwise, all of the fields must be filled in, to match against.
1275 * If 'flags' contains LDLM_FL_LOCAL_ONLY, then only match local locks on the
1276 * server (ie, connh is NULL)
1277 * If 'flags' contains LDLM_FL_BLOCK_GRANTED, then only locks on the granted
1278 * list will be considered
1279 * If 'flags' contains LDLM_FL_CBPENDING, then locks that have been marked
1280 * to be canceled can still be matched as long as they still have reader
1281 * or writer refernces
1282 * If 'flags' contains LDLM_FL_TEST_LOCK, then don't actually reference a lock,
1283 * just tell us if we would have matched.
1285 * \retval 1 if it finds an already-existing lock that is compatible; in this
1286 * case, lockh is filled in with a addref()ed lock
1288 * We also check security context, and if that fails we simply return 0 (to
1289 * keep caller code unchanged), the context failure will be discovered by
1290 * caller sometime later.
1292 ldlm_mode_t ldlm_lock_match(struct ldlm_namespace *ns, __u64 flags,
1293 const struct ldlm_res_id *res_id, ldlm_type_t type,
1294 ldlm_policy_data_t *policy, ldlm_mode_t mode,
1295 struct lustre_handle *lockh, int unref)
1297 struct ldlm_resource *res;
1298 struct ldlm_lock *lock, *old_lock = NULL;
1303 old_lock = ldlm_handle2lock(lockh);
1306 ns = ldlm_lock_to_ns(old_lock);
1307 res_id = &old_lock->l_resource->lr_name;
1308 type = old_lock->l_resource->lr_type;
1309 mode = old_lock->l_req_mode;
1312 res = ldlm_resource_get(ns, NULL, res_id, type, 0);
1314 LASSERT(old_lock == NULL);
1318 LDLM_RESOURCE_ADDREF(res);
1321 lock = search_queue(&res->lr_granted, &mode, policy, old_lock,
1325 if (flags & LDLM_FL_BLOCK_GRANTED)
1327 lock = search_queue(&res->lr_converting, &mode, policy, old_lock,
1331 lock = search_queue(&res->lr_waiting, &mode, policy, old_lock,
1339 LDLM_RESOURCE_DELREF(res);
1340 ldlm_resource_putref(res);
1343 ldlm_lock2handle(lock, lockh);
1344 if ((flags & LDLM_FL_LVB_READY) &&
1345 (!ldlm_is_lvb_ready(lock))) {
1346 __u64 wait_flags = LDLM_FL_LVB_READY |
1347 LDLM_FL_DESTROYED | LDLM_FL_FAIL_NOTIFIED;
1348 struct l_wait_info lwi;
1349 if (lock->l_completion_ast) {
1350 int err = lock->l_completion_ast(lock,
1351 LDLM_FL_WAIT_NOREPROC,
1354 if (flags & LDLM_FL_TEST_LOCK)
1355 LDLM_LOCK_RELEASE(lock);
1357 ldlm_lock_decref_internal(lock,
1364 lwi = LWI_TIMEOUT_INTR(cfs_time_seconds(obd_timeout),
1365 NULL, LWI_ON_SIGNAL_NOOP, NULL);
1367 /* XXX FIXME see comment on CAN_MATCH in lustre_dlm.h */
1368 l_wait_event(lock->l_waitq,
1369 lock->l_flags & wait_flags,
1371 if (!ldlm_is_lvb_ready(lock)) {
1372 if (flags & LDLM_FL_TEST_LOCK)
1373 LDLM_LOCK_RELEASE(lock);
1375 ldlm_lock_decref_internal(lock, mode);
1382 LDLM_DEBUG(lock, "matched ("LPU64" "LPU64")",
1383 (type == LDLM_PLAIN || type == LDLM_IBITS) ?
1384 res_id->name[2] : policy->l_extent.start,
1385 (type == LDLM_PLAIN || type == LDLM_IBITS) ?
1386 res_id->name[3] : policy->l_extent.end);
1388 /* check user's security context */
1389 if (lock->l_conn_export &&
1390 sptlrpc_import_check_ctx(
1391 class_exp2cliimp(lock->l_conn_export))) {
1392 if (!(flags & LDLM_FL_TEST_LOCK))
1393 ldlm_lock_decref_internal(lock, mode);
1397 if (flags & LDLM_FL_TEST_LOCK)
1398 LDLM_LOCK_RELEASE(lock);
1400 } else if (!(flags & LDLM_FL_TEST_LOCK)) {/*less verbose for test-only*/
1401 LDLM_DEBUG_NOLOCK("not matched ns %p type %u mode %u res "
1402 LPU64"/"LPU64" ("LPU64" "LPU64")", ns,
1403 type, mode, res_id->name[0], res_id->name[1],
1404 (type == LDLM_PLAIN || type == LDLM_IBITS) ?
1405 res_id->name[2] :policy->l_extent.start,
1406 (type == LDLM_PLAIN || type == LDLM_IBITS) ?
1407 res_id->name[3] : policy->l_extent.end);
1410 LDLM_LOCK_PUT(old_lock);
1412 return rc ? mode : 0;
1414 EXPORT_SYMBOL(ldlm_lock_match);
1416 ldlm_mode_t ldlm_revalidate_lock_handle(struct lustre_handle *lockh,
1419 struct ldlm_lock *lock;
1420 ldlm_mode_t mode = 0;
1423 lock = ldlm_handle2lock(lockh);
1425 lock_res_and_lock(lock);
1426 if (LDLM_HAVE_MASK(lock, GONE))
1429 if (ldlm_is_cbpending(lock) &&
1430 lock->l_readers == 0 && lock->l_writers == 0)
1434 *bits = lock->l_policy_data.l_inodebits.bits;
1435 mode = lock->l_granted_mode;
1436 ldlm_lock_addref_internal_nolock(lock, mode);
1443 unlock_res_and_lock(lock);
1444 LDLM_LOCK_PUT(lock);
1448 EXPORT_SYMBOL(ldlm_revalidate_lock_handle);
1450 /** The caller must guarantee that the buffer is large enough. */
1451 int ldlm_fill_lvb(struct ldlm_lock *lock, struct req_capsule *pill,
1452 enum req_location loc, void *data, int size)
1457 LASSERT(data != NULL);
1460 switch (lock->l_lvb_type) {
1462 if (size == sizeof(struct ost_lvb)) {
1463 if (loc == RCL_CLIENT)
1464 lvb = req_capsule_client_swab_get(pill,
1466 lustre_swab_ost_lvb);
1468 lvb = req_capsule_server_swab_get(pill,
1470 lustre_swab_ost_lvb);
1471 if (unlikely(lvb == NULL)) {
1472 LDLM_ERROR(lock, "no LVB");
1476 memcpy(data, lvb, size);
1477 } else if (size == sizeof(struct ost_lvb_v1)) {
1478 struct ost_lvb *olvb = data;
1480 if (loc == RCL_CLIENT)
1481 lvb = req_capsule_client_swab_get(pill,
1483 lustre_swab_ost_lvb_v1);
1485 lvb = req_capsule_server_sized_swab_get(pill,
1487 lustre_swab_ost_lvb_v1);
1488 if (unlikely(lvb == NULL)) {
1489 LDLM_ERROR(lock, "no LVB");
1493 memcpy(data, lvb, size);
1494 olvb->lvb_mtime_ns = 0;
1495 olvb->lvb_atime_ns = 0;
1496 olvb->lvb_ctime_ns = 0;
1498 LDLM_ERROR(lock, "Replied unexpected ost LVB size %d",
1504 if (size == sizeof(struct lquota_lvb)) {
1505 if (loc == RCL_CLIENT)
1506 lvb = req_capsule_client_swab_get(pill,
1508 lustre_swab_lquota_lvb);
1510 lvb = req_capsule_server_swab_get(pill,
1512 lustre_swab_lquota_lvb);
1513 if (unlikely(lvb == NULL)) {
1514 LDLM_ERROR(lock, "no LVB");
1518 memcpy(data, lvb, size);
1520 LDLM_ERROR(lock, "Replied unexpected lquota LVB size %d",
1529 if (loc == RCL_CLIENT)
1530 lvb = req_capsule_client_get(pill, &RMF_DLM_LVB);
1532 lvb = req_capsule_server_get(pill, &RMF_DLM_LVB);
1533 if (unlikely(lvb == NULL)) {
1534 LDLM_ERROR(lock, "no LVB");
1538 memcpy(data, lvb, size);
1541 LDLM_ERROR(lock, "Unknown LVB type: %d\n", lock->l_lvb_type);
1542 libcfs_debug_dumpstack(NULL);
1550 * Create and fill in new LDLM lock with specified properties.
1551 * Returns a referenced lock
1553 struct ldlm_lock *ldlm_lock_create(struct ldlm_namespace *ns,
1554 const struct ldlm_res_id *res_id,
1557 const struct ldlm_callback_suite *cbs,
1558 void *data, __u32 lvb_len,
1559 enum lvb_type lvb_type)
1561 struct ldlm_lock *lock;
1562 struct ldlm_resource *res;
1566 res = ldlm_resource_get(ns, NULL, res_id, type, 1);
1568 RETURN(ERR_CAST(res));
1570 lock = ldlm_lock_new(res);
1572 RETURN(ERR_PTR(-ENOMEM));
1574 lock->l_req_mode = mode;
1575 lock->l_ast_data = data;
1576 lock->l_pid = current_pid();
1577 if (ns_is_server(ns))
1578 ldlm_set_ns_srv(lock);
1580 lock->l_blocking_ast = cbs->lcs_blocking;
1581 lock->l_completion_ast = cbs->lcs_completion;
1582 lock->l_glimpse_ast = cbs->lcs_glimpse;
1585 lock->l_tree_node = NULL;
1586 /* if this is the extent lock, allocate the interval tree node */
1587 if (type == LDLM_EXTENT)
1588 if (ldlm_interval_alloc(lock) == NULL)
1589 GOTO(out, rc = -ENOMEM);
1592 lock->l_lvb_len = lvb_len;
1593 OBD_ALLOC_LARGE(lock->l_lvb_data, lvb_len);
1594 if (lock->l_lvb_data == NULL)
1595 GOTO(out, rc = -ENOMEM);
1598 lock->l_lvb_type = lvb_type;
1599 if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_NEW_LOCK))
1600 GOTO(out, rc = -ENOENT);
1605 ldlm_lock_destroy(lock);
1606 LDLM_LOCK_RELEASE(lock);
1607 RETURN(ERR_PTR(rc));
1611 * Enqueue (request) a lock.
1613 * Does not block. As a result of enqueue the lock would be put
1614 * into granted or waiting list.
1616 * If namespace has intent policy sent and the lock has LDLM_FL_HAS_INTENT flag
1617 * set, skip all the enqueueing and delegate lock processing to intent policy
1620 ldlm_error_t ldlm_lock_enqueue(struct ldlm_namespace *ns,
1621 struct ldlm_lock **lockp,
1622 void *cookie, __u64 *flags)
1624 struct ldlm_lock *lock = *lockp;
1625 struct ldlm_resource *res = lock->l_resource;
1626 int local = ns_is_client(ldlm_res_to_ns(res));
1627 #ifdef HAVE_SERVER_SUPPORT
1628 ldlm_processing_policy policy;
1630 ldlm_error_t rc = ELDLM_OK;
1631 struct ldlm_interval *node = NULL;
1634 /* policies are not executed on the client or during replay */
1635 if ((*flags & (LDLM_FL_HAS_INTENT|LDLM_FL_REPLAY)) == LDLM_FL_HAS_INTENT
1636 && !local && ns->ns_policy) {
1637 rc = ns->ns_policy(ns, lockp, cookie, lock->l_req_mode, *flags,
1639 if (rc == ELDLM_LOCK_REPLACED) {
1640 /* The lock that was returned has already been granted,
1641 * and placed into lockp. If it's not the same as the
1642 * one we passed in, then destroy the old one and our
1643 * work here is done. */
1644 if (lock != *lockp) {
1645 ldlm_lock_destroy(lock);
1646 LDLM_LOCK_RELEASE(lock);
1648 *flags |= LDLM_FL_LOCK_CHANGED;
1650 } else if (rc != ELDLM_OK ||
1651 (rc == ELDLM_OK && (*flags & LDLM_FL_INTENT_ONLY))) {
1652 ldlm_lock_destroy(lock);
1657 if (*flags & LDLM_FL_RESENT) {
1658 /* Reconstruct LDLM_FL_SRV_ENQ_MASK @flags for reply.
1659 * Set LOCK_CHANGED always.
1660 * Check if the lock is granted for BLOCK_GRANTED.
1661 * Take NO_TIMEOUT from the lock as it is inherited through
1662 * LDLM_FL_INHERIT_MASK */
1663 *flags |= LDLM_FL_LOCK_CHANGED;
1664 if (lock->l_req_mode != lock->l_granted_mode)
1665 *flags |= LDLM_FL_BLOCK_GRANTED;
1666 *flags |= lock->l_flags & LDLM_FL_NO_TIMEOUT;
1670 /* For a replaying lock, it might be already in granted list. So
1671 * unlinking the lock will cause the interval node to be freed, we
1672 * have to allocate the interval node early otherwise we can't regrant
1673 * this lock in the future. - jay */
1674 if (!local && (*flags & LDLM_FL_REPLAY) && res->lr_type == LDLM_EXTENT)
1675 OBD_SLAB_ALLOC_PTR_GFP(node, ldlm_interval_slab, GFP_NOFS);
1677 lock_res_and_lock(lock);
1678 if (local && lock->l_req_mode == lock->l_granted_mode) {
1679 /* The server returned a blocked lock, but it was granted
1680 * before we got a chance to actually enqueue it. We don't
1681 * need to do anything else. */
1682 *flags &= ~LDLM_FL_BLOCKED_MASK;
1683 GOTO(out, rc = ELDLM_OK);
1686 ldlm_resource_unlink_lock(lock);
1687 if (res->lr_type == LDLM_EXTENT && lock->l_tree_node == NULL) {
1689 ldlm_lock_destroy_nolock(lock);
1690 GOTO(out, rc = -ENOMEM);
1693 INIT_LIST_HEAD(&node->li_group);
1694 ldlm_interval_attach(node, lock);
1698 /* Some flags from the enqueue want to make it into the AST, via the
1699 * lock's l_flags. */
1700 if (*flags & LDLM_FL_AST_DISCARD_DATA)
1701 ldlm_set_ast_discard_data(lock);
1702 if (*flags & LDLM_FL_TEST_LOCK)
1703 ldlm_set_test_lock(lock);
1705 /* This distinction between local lock trees is very important; a client
1706 * namespace only has information about locks taken by that client, and
1707 * thus doesn't have enough information to decide for itself if it can
1708 * be granted (below). In this case, we do exactly what the server
1709 * tells us to do, as dictated by the 'flags'.
1711 * We do exactly the same thing during recovery, when the server is
1712 * more or less trusting the clients not to lie.
1714 * FIXME (bug 268): Detect obvious lies by checking compatibility in
1715 * granted/converting queues. */
1717 if (*flags & LDLM_FL_BLOCK_CONV)
1718 ldlm_resource_add_lock(res, &res->lr_converting, lock);
1719 else if (*flags & (LDLM_FL_BLOCK_WAIT | LDLM_FL_BLOCK_GRANTED))
1720 ldlm_resource_add_lock(res, &res->lr_waiting, lock);
1722 ldlm_grant_lock(lock, NULL);
1723 GOTO(out, rc = ELDLM_OK);
1724 #ifdef HAVE_SERVER_SUPPORT
1725 } else if (*flags & LDLM_FL_REPLAY) {
1726 if (*flags & LDLM_FL_BLOCK_CONV) {
1727 ldlm_resource_add_lock(res, &res->lr_converting, lock);
1728 GOTO(out, rc = ELDLM_OK);
1729 } else if (*flags & LDLM_FL_BLOCK_WAIT) {
1730 ldlm_resource_add_lock(res, &res->lr_waiting, lock);
1731 GOTO(out, rc = ELDLM_OK);
1732 } else if (*flags & LDLM_FL_BLOCK_GRANTED) {
1733 ldlm_grant_lock(lock, NULL);
1734 GOTO(out, rc = ELDLM_OK);
1736 /* If no flags, fall through to normal enqueue path. */
1739 policy = ldlm_processing_policy_table[res->lr_type];
1740 policy(lock, flags, 1, &rc, NULL);
1744 CERROR("This is client-side-only module, cannot handle "
1745 "LDLM_NAMESPACE_SERVER resource type lock.\n");
1751 unlock_res_and_lock(lock);
1753 OBD_SLAB_FREE(node, ldlm_interval_slab, sizeof(*node));
1757 #ifdef HAVE_SERVER_SUPPORT
1759 * Iterate through all waiting locks on a given resource queue and attempt to
1762 * Must be called with resource lock held.
1764 int ldlm_reprocess_queue(struct ldlm_resource *res, struct list_head *queue,
1765 struct list_head *work_list)
1767 struct list_head *tmp, *pos;
1768 ldlm_processing_policy policy;
1770 int rc = LDLM_ITER_CONTINUE;
1774 check_res_locked(res);
1776 policy = ldlm_processing_policy_table[res->lr_type];
1779 list_for_each_safe(tmp, pos, queue) {
1780 struct ldlm_lock *pending;
1781 pending = list_entry(tmp, struct ldlm_lock, l_res_link);
1783 CDEBUG(D_INFO, "Reprocessing lock %p\n", pending);
1786 rc = policy(pending, &flags, 0, &err, work_list);
1787 if (rc != LDLM_ITER_CONTINUE)
1796 * Process a call to blocking AST callback for a lock in ast_work list
1799 ldlm_work_bl_ast_lock(struct ptlrpc_request_set *rqset, void *opaq)
1801 struct ldlm_cb_set_arg *arg = opaq;
1802 struct ldlm_lock_desc d;
1804 struct ldlm_lock *lock;
1807 if (list_empty(arg->list))
1810 lock = list_entry(arg->list->next, struct ldlm_lock, l_bl_ast);
1812 /* nobody should touch l_bl_ast */
1813 lock_res_and_lock(lock);
1814 list_del_init(&lock->l_bl_ast);
1816 LASSERT(ldlm_is_ast_sent(lock));
1817 LASSERT(lock->l_bl_ast_run == 0);
1818 LASSERT(lock->l_blocking_lock);
1819 lock->l_bl_ast_run++;
1820 unlock_res_and_lock(lock);
1822 ldlm_lock2desc(lock->l_blocking_lock, &d);
1824 rc = lock->l_blocking_ast(lock, &d, (void *)arg, LDLM_CB_BLOCKING);
1825 LDLM_LOCK_RELEASE(lock->l_blocking_lock);
1826 lock->l_blocking_lock = NULL;
1827 LDLM_LOCK_RELEASE(lock);
1833 * Process a call to completion AST callback for a lock in ast_work list
1836 ldlm_work_cp_ast_lock(struct ptlrpc_request_set *rqset, void *opaq)
1838 struct ldlm_cb_set_arg *arg = opaq;
1840 struct ldlm_lock *lock;
1841 ldlm_completion_callback completion_callback;
1844 if (list_empty(arg->list))
1847 lock = list_entry(arg->list->next, struct ldlm_lock, l_cp_ast);
1849 /* It's possible to receive a completion AST before we've set
1850 * the l_completion_ast pointer: either because the AST arrived
1851 * before the reply, or simply because there's a small race
1852 * window between receiving the reply and finishing the local
1853 * enqueue. (bug 842)
1855 * This can't happen with the blocking_ast, however, because we
1856 * will never call the local blocking_ast until we drop our
1857 * reader/writer reference, which we won't do until we get the
1858 * reply and finish enqueueing. */
1860 /* nobody should touch l_cp_ast */
1861 lock_res_and_lock(lock);
1862 list_del_init(&lock->l_cp_ast);
1863 LASSERT(ldlm_is_cp_reqd(lock));
1864 /* save l_completion_ast since it can be changed by
1865 * mds_intent_policy(), see bug 14225 */
1866 completion_callback = lock->l_completion_ast;
1867 ldlm_clear_cp_reqd(lock);
1868 unlock_res_and_lock(lock);
1870 if (completion_callback != NULL)
1871 rc = completion_callback(lock, 0, (void *)arg);
1872 LDLM_LOCK_RELEASE(lock);
1878 * Process a call to revocation AST callback for a lock in ast_work list
1881 ldlm_work_revoke_ast_lock(struct ptlrpc_request_set *rqset, void *opaq)
1883 struct ldlm_cb_set_arg *arg = opaq;
1884 struct ldlm_lock_desc desc;
1886 struct ldlm_lock *lock;
1889 if (list_empty(arg->list))
1892 lock = list_entry(arg->list->next, struct ldlm_lock, l_rk_ast);
1893 list_del_init(&lock->l_rk_ast);
1895 /* the desc just pretend to exclusive */
1896 ldlm_lock2desc(lock, &desc);
1897 desc.l_req_mode = LCK_EX;
1898 desc.l_granted_mode = 0;
1900 rc = lock->l_blocking_ast(lock, &desc, (void*)arg, LDLM_CB_BLOCKING);
1901 LDLM_LOCK_RELEASE(lock);
1907 * Process a call to glimpse AST callback for a lock in ast_work list
1909 int ldlm_work_gl_ast_lock(struct ptlrpc_request_set *rqset, void *opaq)
1911 struct ldlm_cb_set_arg *arg = opaq;
1912 struct ldlm_glimpse_work *gl_work;
1913 struct ldlm_lock *lock;
1917 if (list_empty(arg->list))
1920 gl_work = list_entry(arg->list->next, struct ldlm_glimpse_work,
1922 list_del_init(&gl_work->gl_list);
1924 lock = gl_work->gl_lock;
1926 /* transfer the glimpse descriptor to ldlm_cb_set_arg */
1927 arg->gl_desc = gl_work->gl_desc;
1929 /* invoke the actual glimpse callback */
1930 if (lock->l_glimpse_ast(lock, (void*)arg) == 0)
1933 LDLM_LOCK_RELEASE(lock);
1935 if ((gl_work->gl_flags & LDLM_GL_WORK_NOFREE) == 0)
1936 OBD_FREE_PTR(gl_work);
1942 * Process list of locks in need of ASTs being sent.
1944 * Used on server to send multiple ASTs together instead of sending one by
1947 int ldlm_run_ast_work(struct ldlm_namespace *ns, struct list_head *rpc_list,
1948 ldlm_desc_ast_t ast_type)
1950 struct ldlm_cb_set_arg *arg;
1951 set_producer_func work_ast_lock;
1954 if (list_empty(rpc_list))
1961 atomic_set(&arg->restart, 0);
1962 arg->list = rpc_list;
1965 case LDLM_WORK_BL_AST:
1966 arg->type = LDLM_BL_CALLBACK;
1967 work_ast_lock = ldlm_work_bl_ast_lock;
1969 case LDLM_WORK_CP_AST:
1970 arg->type = LDLM_CP_CALLBACK;
1971 work_ast_lock = ldlm_work_cp_ast_lock;
1973 case LDLM_WORK_REVOKE_AST:
1974 arg->type = LDLM_BL_CALLBACK;
1975 work_ast_lock = ldlm_work_revoke_ast_lock;
1977 case LDLM_WORK_GL_AST:
1978 arg->type = LDLM_GL_CALLBACK;
1979 work_ast_lock = ldlm_work_gl_ast_lock;
1985 /* We create a ptlrpc request set with flow control extension.
1986 * This request set will use the work_ast_lock function to produce new
1987 * requests and will send a new request each time one completes in order
1988 * to keep the number of requests in flight to ns_max_parallel_ast */
1989 arg->set = ptlrpc_prep_fcset(ns->ns_max_parallel_ast ? : UINT_MAX,
1990 work_ast_lock, arg);
1991 if (arg->set == NULL)
1992 GOTO(out, rc = -ENOMEM);
1994 ptlrpc_set_wait(arg->set);
1995 ptlrpc_set_destroy(arg->set);
1997 rc = atomic_read(&arg->restart) ? -ERESTART : 0;
2004 static int reprocess_one_queue(struct ldlm_resource *res, void *closure)
2006 ldlm_reprocess_all(res);
2007 return LDLM_ITER_CONTINUE;
2010 static int ldlm_reprocess_res(cfs_hash_t *hs, cfs_hash_bd_t *bd,
2011 struct hlist_node *hnode, void *arg)
2013 struct ldlm_resource *res = cfs_hash_object(hs, hnode);
2016 rc = reprocess_one_queue(res, arg);
2018 return rc == LDLM_ITER_STOP;
2022 * Iterate through all resources on a namespace attempting to grant waiting
2025 void ldlm_reprocess_all_ns(struct ldlm_namespace *ns)
2030 cfs_hash_for_each_nolock(ns->ns_rs_hash,
2031 ldlm_reprocess_res, NULL);
2035 EXPORT_SYMBOL(ldlm_reprocess_all_ns);
2038 * Try to grant all waiting locks on a resource.
2040 * Calls ldlm_reprocess_queue on converting and waiting queues.
2042 * Typically called after some resource locks are cancelled to see
2043 * if anything could be granted as a result of the cancellation.
2045 void ldlm_reprocess_all(struct ldlm_resource *res)
2047 struct list_head rpc_list;
2048 #ifdef HAVE_SERVER_SUPPORT
2052 INIT_LIST_HEAD(&rpc_list);
2053 /* Local lock trees don't get reprocessed. */
2054 if (ns_is_client(ldlm_res_to_ns(res))) {
2061 rc = ldlm_reprocess_queue(res, &res->lr_converting, &rpc_list);
2062 if (rc == LDLM_ITER_CONTINUE)
2063 ldlm_reprocess_queue(res, &res->lr_waiting, &rpc_list);
2066 rc = ldlm_run_ast_work(ldlm_res_to_ns(res), &rpc_list,
2068 if (rc == -ERESTART) {
2069 LASSERT(list_empty(&rpc_list));
2075 INIT_LIST_HEAD(&rpc_list);
2076 if (!ns_is_client(ldlm_res_to_ns(res))) {
2077 CERROR("This is client-side-only module, cannot handle "
2078 "LDLM_NAMESPACE_SERVER resource type lock.\n");
2084 EXPORT_SYMBOL(ldlm_reprocess_all);
2087 * Helper function to call blocking AST for LDLM lock \a lock in a
2088 * "cancelling" mode.
2090 void ldlm_cancel_callback(struct ldlm_lock *lock)
2092 check_res_locked(lock->l_resource);
2093 if (!ldlm_is_cancel(lock)) {
2094 ldlm_set_cancel(lock);
2095 if (lock->l_blocking_ast) {
2096 unlock_res_and_lock(lock);
2097 lock->l_blocking_ast(lock, NULL, lock->l_ast_data,
2099 lock_res_and_lock(lock);
2101 LDLM_DEBUG(lock, "no blocking ast");
2104 ldlm_set_bl_done(lock);
2108 * Remove skiplist-enabled LDLM lock \a req from granted list
2110 void ldlm_unlink_lock_skiplist(struct ldlm_lock *req)
2112 if (req->l_resource->lr_type != LDLM_PLAIN &&
2113 req->l_resource->lr_type != LDLM_IBITS)
2116 list_del_init(&req->l_sl_policy);
2117 list_del_init(&req->l_sl_mode);
2121 * Attempts to cancel LDLM lock \a lock that has no reader/writer references.
2123 void ldlm_lock_cancel(struct ldlm_lock *lock)
2125 struct ldlm_resource *res;
2126 struct ldlm_namespace *ns;
2129 lock_res_and_lock(lock);
2131 res = lock->l_resource;
2132 ns = ldlm_res_to_ns(res);
2134 /* Please do not, no matter how tempting, remove this LBUG without
2135 * talking to me first. -phik */
2136 if (lock->l_readers || lock->l_writers) {
2137 LDLM_ERROR(lock, "lock still has references");
2141 if (ldlm_is_waited(lock))
2142 ldlm_del_waiting_lock(lock);
2144 /* Releases cancel callback. */
2145 ldlm_cancel_callback(lock);
2147 /* Yes, second time, just in case it was added again while we were
2148 * running with no res lock in ldlm_cancel_callback */
2149 if (ldlm_is_waited(lock))
2150 ldlm_del_waiting_lock(lock);
2152 ldlm_resource_unlink_lock(lock);
2153 ldlm_lock_destroy_nolock(lock);
2155 if (lock->l_granted_mode == lock->l_req_mode)
2156 ldlm_pool_del(&ns->ns_pool, lock);
2158 /* Make sure we will not be called again for same lock what is possible
2159 * if not to zero out lock->l_granted_mode */
2160 lock->l_granted_mode = LCK_MINMODE;
2161 unlock_res_and_lock(lock);
2165 EXPORT_SYMBOL(ldlm_lock_cancel);
2168 * Set opaque data into the lock that only makes sense to upper layer.
2170 int ldlm_lock_set_data(struct lustre_handle *lockh, void *data)
2172 struct ldlm_lock *lock = ldlm_handle2lock(lockh);
2177 if (lock->l_ast_data == NULL)
2178 lock->l_ast_data = data;
2179 if (lock->l_ast_data == data)
2181 LDLM_LOCK_PUT(lock);
2185 EXPORT_SYMBOL(ldlm_lock_set_data);
2187 struct export_cl_data {
2188 struct obd_export *ecl_exp;
2193 * Iterator function for ldlm_cancel_locks_for_export.
2194 * Cancels passed locks.
2196 static int ldlm_cancel_locks_for_export_cb(cfs_hash_t *hs, cfs_hash_bd_t *bd,
2197 struct hlist_node *hnode, void *data)
2200 struct export_cl_data *ecl = (struct export_cl_data *)data;
2201 struct obd_export *exp = ecl->ecl_exp;
2202 struct ldlm_lock *lock = cfs_hash_object(hs, hnode);
2203 struct ldlm_resource *res;
2205 res = ldlm_resource_getref(lock->l_resource);
2206 LDLM_LOCK_GET(lock);
2208 LDLM_DEBUG(lock, "export %p", exp);
2209 ldlm_res_lvbo_update(res, NULL, 1);
2210 ldlm_lock_cancel(lock);
2211 ldlm_reprocess_all(res);
2212 ldlm_resource_putref(res);
2213 LDLM_LOCK_RELEASE(lock);
2216 if ((ecl->ecl_loop & -ecl->ecl_loop) == ecl->ecl_loop) {
2218 "Cancel lock %p for export %p (loop %d), still have "
2219 "%d locks left on hash table.\n",
2220 lock, exp, ecl->ecl_loop,
2221 atomic_read(&hs->hs_count));
2228 * Cancel all locks for given export.
2230 * Typically called on client disconnection/eviction
2232 void ldlm_cancel_locks_for_export(struct obd_export *exp)
2234 struct export_cl_data ecl = {
2239 cfs_hash_for_each_empty(exp->exp_lock_hash,
2240 ldlm_cancel_locks_for_export_cb, &ecl);
2244 * Downgrade an exclusive lock.
2246 * A fast variant of ldlm_lock_convert for convertion of exclusive
2247 * locks. The convertion is always successful.
2248 * Used by Commit on Sharing (COS) code.
2250 * \param lock A lock to convert
2251 * \param new_mode new lock mode
2253 void ldlm_lock_downgrade(struct ldlm_lock *lock, int new_mode)
2257 LASSERT(lock->l_granted_mode & (LCK_PW | LCK_EX));
2258 LASSERT(new_mode == LCK_COS);
2260 lock_res_and_lock(lock);
2261 ldlm_resource_unlink_lock(lock);
2263 * Remove the lock from pool as it will be added again in
2264 * ldlm_grant_lock() called below.
2266 ldlm_pool_del(&ldlm_lock_to_ns(lock)->ns_pool, lock);
2268 lock->l_req_mode = new_mode;
2269 ldlm_grant_lock(lock, NULL);
2270 unlock_res_and_lock(lock);
2271 ldlm_reprocess_all(lock->l_resource);
2275 EXPORT_SYMBOL(ldlm_lock_downgrade);
2278 * Attempt to convert already granted lock to a different mode.
2280 * While lock conversion is not currently used, future client-side
2281 * optimizations could take advantage of it to avoid discarding cached
2284 struct ldlm_resource *ldlm_lock_convert(struct ldlm_lock *lock, int new_mode,
2287 struct list_head rpc_list;
2288 struct ldlm_resource *res;
2289 struct ldlm_namespace *ns;
2291 #ifdef HAVE_SERVER_SUPPORT
2293 struct sl_insert_point prev;
2295 struct ldlm_interval *node;
2298 INIT_LIST_HEAD(&rpc_list);
2299 /* Just return if mode is unchanged. */
2300 if (new_mode == lock->l_granted_mode) {
2301 *flags |= LDLM_FL_BLOCK_GRANTED;
2302 RETURN(lock->l_resource);
2305 /* I can't check the type of lock here because the bitlock of lock
2306 * is not held here, so do the allocation blindly. -jay */
2307 OBD_SLAB_ALLOC_PTR_GFP(node, ldlm_interval_slab, GFP_NOFS);
2308 if (node == NULL) /* Actually, this causes EDEADLOCK to be returned */
2311 LASSERTF((new_mode == LCK_PW && lock->l_granted_mode == LCK_PR),
2312 "new_mode %u, granted %u\n", new_mode, lock->l_granted_mode);
2314 lock_res_and_lock(lock);
2316 res = lock->l_resource;
2317 ns = ldlm_res_to_ns(res);
2319 #ifdef HAVE_SERVER_SUPPORT
2320 old_mode = lock->l_req_mode;
2322 lock->l_req_mode = new_mode;
2323 if (res->lr_type == LDLM_PLAIN || res->lr_type == LDLM_IBITS) {
2324 #ifdef HAVE_SERVER_SUPPORT
2325 /* remember the lock position where the lock might be
2326 * added back to the granted list later and also
2327 * remember the join mode for skiplist fixing. */
2328 prev.res_link = lock->l_res_link.prev;
2329 prev.mode_link = lock->l_sl_mode.prev;
2330 prev.policy_link = lock->l_sl_policy.prev;
2332 ldlm_resource_unlink_lock(lock);
2334 ldlm_resource_unlink_lock(lock);
2335 if (res->lr_type == LDLM_EXTENT) {
2336 /* FIXME: ugly code, I have to attach the lock to a
2337 * interval node again since perhaps it will be granted
2339 INIT_LIST_HEAD(&node->li_group);
2340 ldlm_interval_attach(node, lock);
2346 * Remove old lock from the pool before adding the lock with new
2347 * mode below in ->policy()
2349 ldlm_pool_del(&ns->ns_pool, lock);
2351 /* If this is a local resource, put it on the appropriate list. */
2352 if (ns_is_client(ldlm_res_to_ns(res))) {
2353 if (*flags & (LDLM_FL_BLOCK_CONV | LDLM_FL_BLOCK_GRANTED)) {
2354 ldlm_resource_add_lock(res, &res->lr_converting, lock);
2356 /* This should never happen, because of the way the
2357 * server handles conversions. */
2358 LDLM_ERROR(lock, "Erroneous flags %x on local lock\n",
2362 ldlm_grant_lock(lock, &rpc_list);
2364 /* FIXME: completion handling not with lr_lock held ! */
2365 if (lock->l_completion_ast)
2366 lock->l_completion_ast(lock, 0, NULL);
2368 #ifdef HAVE_SERVER_SUPPORT
2373 ldlm_processing_policy policy;
2374 policy = ldlm_processing_policy_table[res->lr_type];
2375 rc = policy(lock, &pflags, 0, &err, &rpc_list);
2376 if (rc == LDLM_ITER_STOP) {
2377 lock->l_req_mode = old_mode;
2378 if (res->lr_type == LDLM_EXTENT)
2379 ldlm_extent_add_lock(res, lock);
2381 ldlm_granted_list_add_lock(lock, &prev);
2385 *flags |= LDLM_FL_BLOCK_GRANTED;
2391 CERROR("This is client-side-only module, cannot handle "
2392 "LDLM_NAMESPACE_SERVER resource type lock.\n");
2396 unlock_res_and_lock(lock);
2399 ldlm_run_ast_work(ns, &rpc_list, LDLM_WORK_CP_AST);
2401 OBD_SLAB_FREE(node, ldlm_interval_slab, sizeof(*node));
2404 EXPORT_SYMBOL(ldlm_lock_convert);
2407 * Print lock with lock handle \a lockh description into debug log.
2409 * Used when printing all locks on a resource for debug purposes.
2411 void ldlm_lock_dump_handle(int level, struct lustre_handle *lockh)
2413 struct ldlm_lock *lock;
2415 if (!((libcfs_debug | D_ERROR) & level))
2418 lock = ldlm_handle2lock(lockh);
2422 LDLM_DEBUG_LIMIT(level, lock, "###");
2424 LDLM_LOCK_PUT(lock);
2426 EXPORT_SYMBOL(ldlm_lock_dump_handle);
2429 * Print lock information with custom message into debug log.
2432 void _ldlm_lock_debug(struct ldlm_lock *lock,
2433 struct libcfs_debug_msg_data *msgdata,
2434 const char *fmt, ...)
2437 struct obd_export *exp = lock->l_export;
2438 struct ldlm_resource *resource = lock->l_resource;
2439 char *nid = "local";
2441 va_start(args, fmt);
2443 if (exp && exp->exp_connection) {
2444 nid = libcfs_nid2str(exp->exp_connection->c_peer.nid);
2445 } else if (exp && exp->exp_obd != NULL) {
2446 struct obd_import *imp = exp->exp_obd->u.cli.cl_import;
2447 nid = libcfs_nid2str(imp->imp_connection->c_peer.nid);
2450 if (resource == NULL) {
2451 libcfs_debug_vmsg2(msgdata, fmt, args,
2452 " ns: \?\? lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
2453 "res: \?\? rrc=\?\? type: \?\?\? flags: "LPX64" nid: %s "
2454 "remote: "LPX64" expref: %d pid: %u timeout: %lu "
2457 lock->l_handle.h_cookie, atomic_read(&lock->l_refc),
2458 lock->l_readers, lock->l_writers,
2459 ldlm_lockname[lock->l_granted_mode],
2460 ldlm_lockname[lock->l_req_mode],
2461 lock->l_flags, nid, lock->l_remote_handle.cookie,
2462 exp ? atomic_read(&exp->exp_refcount) : -99,
2463 lock->l_pid, lock->l_callback_timeout, lock->l_lvb_type);
2468 switch (resource->lr_type) {
2470 libcfs_debug_vmsg2(msgdata, fmt, args,
2471 " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
2472 "res: "DLDLMRES" rrc: %d type: %s ["LPU64"->"LPU64"] "
2473 "(req "LPU64"->"LPU64") flags: "LPX64" nid: %s remote: "
2474 LPX64" expref: %d pid: %u timeout: %lu lvb_type: %d\n",
2475 ldlm_lock_to_ns_name(lock), lock,
2476 lock->l_handle.h_cookie, atomic_read(&lock->l_refc),
2477 lock->l_readers, lock->l_writers,
2478 ldlm_lockname[lock->l_granted_mode],
2479 ldlm_lockname[lock->l_req_mode],
2481 atomic_read(&resource->lr_refcount),
2482 ldlm_typename[resource->lr_type],
2483 lock->l_policy_data.l_extent.start,
2484 lock->l_policy_data.l_extent.end,
2485 lock->l_req_extent.start, lock->l_req_extent.end,
2486 lock->l_flags, nid, lock->l_remote_handle.cookie,
2487 exp ? atomic_read(&exp->exp_refcount) : -99,
2488 lock->l_pid, lock->l_callback_timeout,
2493 libcfs_debug_vmsg2(msgdata, fmt, args,
2494 " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
2495 "res: "DLDLMRES" rrc: %d type: %s pid: %d "
2496 "["LPU64"->"LPU64"] flags: "LPX64" nid: %s "
2497 "remote: "LPX64" expref: %d pid: %u timeout: %lu\n",
2498 ldlm_lock_to_ns_name(lock), lock,
2499 lock->l_handle.h_cookie, atomic_read(&lock->l_refc),
2500 lock->l_readers, lock->l_writers,
2501 ldlm_lockname[lock->l_granted_mode],
2502 ldlm_lockname[lock->l_req_mode],
2504 atomic_read(&resource->lr_refcount),
2505 ldlm_typename[resource->lr_type],
2506 lock->l_policy_data.l_flock.pid,
2507 lock->l_policy_data.l_flock.start,
2508 lock->l_policy_data.l_flock.end,
2509 lock->l_flags, nid, lock->l_remote_handle.cookie,
2510 exp ? atomic_read(&exp->exp_refcount) : -99,
2511 lock->l_pid, lock->l_callback_timeout);
2515 libcfs_debug_vmsg2(msgdata, fmt, args,
2516 " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
2517 "res: "DLDLMRES" bits "LPX64" rrc: %d type: %s "
2518 "flags: "LPX64" nid: %s remote: "LPX64" expref: %d "
2519 "pid: %u timeout: %lu lvb_type: %d\n",
2520 ldlm_lock_to_ns_name(lock),
2521 lock, lock->l_handle.h_cookie,
2522 atomic_read(&lock->l_refc),
2523 lock->l_readers, lock->l_writers,
2524 ldlm_lockname[lock->l_granted_mode],
2525 ldlm_lockname[lock->l_req_mode],
2527 lock->l_policy_data.l_inodebits.bits,
2528 atomic_read(&resource->lr_refcount),
2529 ldlm_typename[resource->lr_type],
2530 lock->l_flags, nid, lock->l_remote_handle.cookie,
2531 exp ? atomic_read(&exp->exp_refcount) : -99,
2532 lock->l_pid, lock->l_callback_timeout,
2537 libcfs_debug_vmsg2(msgdata, fmt, args,
2538 " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
2539 "res: "DLDLMRES" rrc: %d type: %s flags: "LPX64" "
2540 "nid: %s remote: "LPX64" expref: %d pid: %u "
2541 "timeout: %lu lvb_type: %d\n",
2542 ldlm_lock_to_ns_name(lock),
2543 lock, lock->l_handle.h_cookie,
2544 atomic_read(&lock->l_refc),
2545 lock->l_readers, lock->l_writers,
2546 ldlm_lockname[lock->l_granted_mode],
2547 ldlm_lockname[lock->l_req_mode],
2549 atomic_read(&resource->lr_refcount),
2550 ldlm_typename[resource->lr_type],
2551 lock->l_flags, nid, lock->l_remote_handle.cookie,
2552 exp ? atomic_read(&exp->exp_refcount) : -99,
2553 lock->l_pid, lock->l_callback_timeout,
2559 EXPORT_SYMBOL(_ldlm_lock_debug);