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.gnu.org/licenses/gpl-2.0.html
23 * Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved.
24 * Use is subject to license terms.
26 * Copyright (c) 2010, 2017, Intel Corporation.
29 * This file is part of Lustre, http://www.lustre.org/
31 * lustre/ldlm/ldlm_lockd.c
33 * Author: Peter Braam <braam@clusterfs.com>
34 * Author: Phil Schwan <phil@clusterfs.com>
37 #define DEBUG_SUBSYSTEM S_LDLM
39 #include <linux/kthread.h>
40 #include <linux/list.h>
41 #include <libcfs/libcfs.h>
42 #include <libcfs/linux/linux-mem.h>
43 #include <lustre_errno.h>
44 #include <lustre_dlm.h>
45 #include <obd_class.h>
46 #include "ldlm_internal.h"
48 static int ldlm_num_threads;
49 module_param(ldlm_num_threads, int, 0444);
50 MODULE_PARM_DESC(ldlm_num_threads, "number of DLM service threads to start");
52 static unsigned int ldlm_cpu_bind = 1;
53 module_param(ldlm_cpu_bind, uint, 0444);
54 MODULE_PARM_DESC(ldlm_cpu_bind,
55 "bind DLM service threads to particular CPU partitions");
57 static char *ldlm_cpts;
58 module_param(ldlm_cpts, charp, 0444);
59 MODULE_PARM_DESC(ldlm_cpts, "CPU partitions ldlm threads should run on");
61 static DEFINE_MUTEX(ldlm_ref_mutex);
62 static int ldlm_refcount;
64 struct kobject *ldlm_kobj;
65 struct kset *ldlm_ns_kset;
66 struct kset *ldlm_svc_kset;
70 static struct ldlm_state *ldlm_state;
73 * timeout for initial callback (AST) reply (bz10399)
74 * Due to having to send a 32 bit time value over the
75 * wire return it as timeout_t instead of time64_t
77 static inline timeout_t ldlm_get_rq_timeout(void)
80 timeout_t timeout = min(ldlm_timeout, obd_timeout / 3);
82 return timeout < 1 ? 1 : timeout;
89 * blp_prio_list is used for callbacks that should be handled
90 * as a priority. It is used for LDLM_FL_DISCARD_DATA requests.
93 struct list_head blp_prio_list;
96 * blp_list is used for all other callbacks which are likely
97 * to take longer to process.
99 struct list_head blp_list;
101 wait_queue_head_t blp_waitq;
102 struct completion blp_comp;
103 atomic_t blp_num_threads;
104 atomic_t blp_busy_threads;
109 struct ldlm_bl_work_item {
110 struct list_head blwi_entry;
111 struct ldlm_namespace *blwi_ns;
112 struct ldlm_lock_desc blwi_ld;
113 struct ldlm_lock *blwi_lock;
114 struct list_head blwi_head;
116 struct completion blwi_comp;
117 enum ldlm_cancel_flags blwi_flags;
118 int blwi_mem_pressure;
121 #ifdef HAVE_SERVER_SUPPORT
124 * Protects both waiting_locks_list and expired_lock_thread.
126 static DEFINE_SPINLOCK(waiting_locks_spinlock); /* BH lock (timer) */
129 * List for contended locks.
131 * As soon as a lock is contended, it gets placed on this list and
132 * expected time to get a response is filled in the lock. A special
133 * thread walks the list looking for locks that should be released and
134 * schedules client evictions for those that have not been released in
137 * All access to it should be under waiting_locks_spinlock.
139 static LIST_HEAD(waiting_locks_list);
140 static void waiting_locks_callback(TIMER_DATA_TYPE unused);
141 static CFS_DEFINE_TIMER(waiting_locks_timer, waiting_locks_callback, 0, 0);
149 static DECLARE_WAIT_QUEUE_HEAD(expired_lock_wait_queue);
150 static enum elt_state expired_lock_thread_state = ELT_STOPPED;
151 static int expired_lock_dump;
152 static LIST_HEAD(expired_lock_list);
154 static int ldlm_lock_busy(struct ldlm_lock *lock);
155 static int ldlm_add_waiting_lock(struct ldlm_lock *lock, timeout_t timeout);
156 static int __ldlm_add_waiting_lock(struct ldlm_lock *lock, timeout_t timeout);
158 static inline int have_expired_locks(void)
163 spin_lock_bh(&waiting_locks_spinlock);
164 need_to_run = !list_empty(&expired_lock_list);
165 spin_unlock_bh(&waiting_locks_spinlock);
171 * Check expired lock list for expired locks and time them out.
173 static int expired_lock_main(void *arg)
175 struct list_head *expired = &expired_lock_list;
180 expired_lock_thread_state = ELT_READY;
181 wake_up(&expired_lock_wait_queue);
184 wait_event_idle(expired_lock_wait_queue,
185 have_expired_locks() ||
186 expired_lock_thread_state == ELT_TERMINATE);
188 spin_lock_bh(&waiting_locks_spinlock);
189 if (expired_lock_dump) {
190 spin_unlock_bh(&waiting_locks_spinlock);
192 /* from waiting_locks_callback, but not in timer */
193 libcfs_debug_dumplog();
195 spin_lock_bh(&waiting_locks_spinlock);
196 expired_lock_dump = 0;
201 while (!list_empty(expired)) {
202 struct obd_export *export;
203 struct ldlm_lock *lock;
205 lock = list_entry(expired->next, struct ldlm_lock,
207 if ((void *)lock < LP_POISON + PAGE_SIZE &&
208 (void *)lock >= LP_POISON) {
209 spin_unlock_bh(&waiting_locks_spinlock);
210 CERROR("free lock on elt list %p\n", lock);
213 list_del_init(&lock->l_pending_chain);
214 if ((void *)lock->l_export <
215 LP_POISON + PAGE_SIZE &&
216 (void *)lock->l_export >= LP_POISON) {
217 CERROR("lock with free export on elt list %p\n",
219 lock->l_export = NULL;
220 LDLM_ERROR(lock, "free export");
222 * release extra ref grabbed by
223 * ldlm_add_waiting_lock() or
226 LDLM_LOCK_RELEASE(lock);
230 if (ldlm_is_destroyed(lock)) {
232 * release the lock refcount where
233 * waiting_locks_callback() founds
235 LDLM_LOCK_RELEASE(lock);
238 export = class_export_lock_get(lock->l_export, lock);
239 spin_unlock_bh(&waiting_locks_spinlock);
241 /* Check if we need to prolong timeout */
242 if (!OBD_FAIL_CHECK(OBD_FAIL_PTLRPC_HPREQ_TIMEOUT) &&
243 lock->l_callback_timestamp != 0 && /* not AST error */
244 ldlm_lock_busy(lock)) {
245 LDLM_DEBUG(lock, "prolong the busy lock");
246 lock_res_and_lock(lock);
247 ldlm_add_waiting_lock(lock,
248 ldlm_bl_timeout(lock) >> 1);
249 unlock_res_and_lock(lock);
251 spin_lock_bh(&export->exp_bl_list_lock);
252 list_del_init(&lock->l_exp_list);
253 spin_unlock_bh(&export->exp_bl_list_lock);
256 "lock callback timer expired after %llds: evicting client at %s ",
257 ktime_get_seconds() -
259 obd_export_nid2str(export));
260 ldlm_lock_to_ns(lock)->ns_timeouts++;
262 class_fail_export(export);
264 class_export_lock_put(export, lock);
266 * release extra ref grabbed by ldlm_add_waiting_lock()
267 * or ldlm_failed_ast()
269 LDLM_LOCK_RELEASE(lock);
271 spin_lock_bh(&waiting_locks_spinlock);
273 spin_unlock_bh(&waiting_locks_spinlock);
275 if (do_dump && obd_dump_on_eviction) {
276 CERROR("dump the log upon eviction\n");
277 libcfs_debug_dumplog();
280 if (expired_lock_thread_state == ELT_TERMINATE)
284 expired_lock_thread_state = ELT_STOPPED;
285 wake_up(&expired_lock_wait_queue);
290 * Check if there is a request in the export request list
291 * which prevents the lock canceling.
293 static int ldlm_lock_busy(struct ldlm_lock *lock)
295 struct ptlrpc_request *req;
300 if (lock->l_export == NULL)
303 spin_lock(&lock->l_export->exp_rpc_lock);
304 list_for_each_entry(req, &lock->l_export->exp_hp_rpcs,
306 if (req->rq_ops->hpreq_lock_match) {
307 match = req->rq_ops->hpreq_lock_match(req, lock);
312 spin_unlock(&lock->l_export->exp_rpc_lock);
316 /* This is called from within a timer interrupt and cannot schedule */
317 static void waiting_locks_callback(TIMER_DATA_TYPE unused)
319 struct ldlm_lock *lock;
322 spin_lock_bh(&waiting_locks_spinlock);
323 while (!list_empty(&waiting_locks_list)) {
324 lock = list_entry(waiting_locks_list.next, struct ldlm_lock,
326 if (lock->l_callback_timestamp > ktime_get_seconds() ||
327 lock->l_req_mode == LCK_GROUP)
331 * no needs to take an extra ref on the lock since it was in
332 * the waiting_locks_list and ldlm_add_waiting_lock()
333 * already grabbed a ref
335 list_move(&lock->l_pending_chain, &expired_lock_list);
339 if (!list_empty(&expired_lock_list)) {
340 if (obd_dump_on_timeout && need_dump)
341 expired_lock_dump = __LINE__;
343 wake_up(&expired_lock_wait_queue);
347 * Make sure the timer will fire again if we have any locks
350 if (!list_empty(&waiting_locks_list)) {
351 time64_t now = ktime_get_seconds();
354 lock = list_entry(waiting_locks_list.next, struct ldlm_lock,
356 if (lock->l_callback_timestamp - now > 0)
357 delta = lock->l_callback_timestamp - now;
358 mod_timer(&waiting_locks_timer,
359 jiffies + cfs_time_seconds(delta));
361 spin_unlock_bh(&waiting_locks_spinlock);
365 * Add lock to the list of contended locks.
367 * Indicate that we're waiting for a client to call us back cancelling a given
368 * lock. We add it to the pending-callback chain, and schedule the lock-timeout
369 * timer to fire appropriately. (We round up to the next second, to avoid
370 * floods of timer firings during periods of high lock contention and traffic).
371 * As done by ldlm_add_waiting_lock(), the caller must grab a lock reference
372 * if it has been added to the waiting list (1 is returned).
374 * Called with the namespace lock held.
376 static int __ldlm_add_waiting_lock(struct ldlm_lock *lock, timeout_t delay)
378 unsigned long timeout_jiffies = jiffies;
382 lock->l_blast_sent = ktime_get_seconds();
383 if (!list_empty(&lock->l_pending_chain))
386 if (OBD_FAIL_CHECK(OBD_FAIL_PTLRPC_HPREQ_NOTIMEOUT) ||
387 OBD_FAIL_CHECK(OBD_FAIL_PTLRPC_HPREQ_TIMEOUT))
390 deadline = lock->l_blast_sent + delay;
391 if (likely(deadline > lock->l_callback_timestamp))
392 lock->l_callback_timestamp = deadline;
394 timeout = clamp_t(timeout_t,
395 lock->l_callback_timestamp - lock->l_blast_sent,
397 timeout_jiffies += cfs_time_seconds(timeout);
399 if (time_before(timeout_jiffies, waiting_locks_timer.expires) ||
400 !timer_pending(&waiting_locks_timer))
401 mod_timer(&waiting_locks_timer, timeout_jiffies);
404 * if the new lock has a shorter timeout than something earlier on
405 * the list, we'll wait the longer amount of time; no big deal.
408 list_add_tail(&lock->l_pending_chain, &waiting_locks_list);
412 static void ldlm_add_blocked_lock(struct ldlm_lock *lock)
414 spin_lock_bh(&lock->l_export->exp_bl_list_lock);
415 if (list_empty(&lock->l_exp_list)) {
416 if (!ldlm_is_granted(lock))
417 list_add_tail(&lock->l_exp_list,
418 &lock->l_export->exp_bl_list);
420 list_add(&lock->l_exp_list,
421 &lock->l_export->exp_bl_list);
423 spin_unlock_bh(&lock->l_export->exp_bl_list_lock);
426 * A blocked lock is added. Adjust the position in
427 * the stale list if the export is in the list.
428 * If export is stale and not in the list - it is being
429 * processed and will be placed on the right position
430 * on obd_stale_export_put().
432 if (!list_empty(&lock->l_export->exp_stale_list))
433 obd_stale_export_adjust(lock->l_export);
436 static int ldlm_add_waiting_lock(struct ldlm_lock *lock, timeout_t timeout)
440 /* NB: must be called with hold of lock_res_and_lock() */
441 LASSERT(ldlm_is_res_locked(lock));
442 LASSERT(!ldlm_is_cancel_on_block(lock));
445 * Do not put cross-MDT lock in the waiting list, since we
446 * will not evict it due to timeout for now
448 if (lock->l_export != NULL &&
449 (exp_connect_flags(lock->l_export) & OBD_CONNECT_MDS_MDS))
452 spin_lock_bh(&waiting_locks_spinlock);
453 if (ldlm_is_cancel(lock)) {
454 spin_unlock_bh(&waiting_locks_spinlock);
458 if (ldlm_is_destroyed(lock)) {
459 static time64_t next;
461 spin_unlock_bh(&waiting_locks_spinlock);
462 LDLM_ERROR(lock, "not waiting on destroyed lock (b=5653)");
463 if (ktime_get_seconds() > next) {
464 next = ktime_get_seconds() + 14400;
465 libcfs_debug_dumpstack(NULL);
470 ldlm_set_waited(lock);
471 ret = __ldlm_add_waiting_lock(lock, timeout);
474 * grab ref on the lock if it has been added to the
479 spin_unlock_bh(&waiting_locks_spinlock);
482 ldlm_add_blocked_lock(lock);
484 LDLM_DEBUG(lock, "%sadding to wait list(timeout: %d, AT: %s)",
485 ret == 0 ? "not re-" : "", timeout,
486 AT_OFF ? "off" : "on");
491 * Remove a lock from the pending list, likely because it had its cancellation
492 * callback arrive without incident. This adjusts the lock-timeout timer if
493 * needed. Returns 0 if the lock wasn't pending after all, 1 if it was.
494 * As done by ldlm_del_waiting_lock(), the caller must release the lock
495 * reference when the lock is removed from any list (1 is returned).
497 * Called with namespace lock held.
499 static int __ldlm_del_waiting_lock(struct ldlm_lock *lock)
501 struct list_head *list_next;
503 if (list_empty(&lock->l_pending_chain))
506 list_next = lock->l_pending_chain.next;
507 if (lock->l_pending_chain.prev == &waiting_locks_list) {
508 /* Removing the head of the list, adjust timer. */
509 if (list_next == &waiting_locks_list) {
510 /* No more, just cancel. */
511 del_timer(&waiting_locks_timer);
513 time64_t now = ktime_get_seconds();
514 struct ldlm_lock *next;
517 next = list_entry(list_next, struct ldlm_lock,
519 if (next->l_callback_timestamp - now > 0)
520 delta = lock->l_callback_timestamp - now;
522 mod_timer(&waiting_locks_timer,
523 jiffies + cfs_time_seconds(delta));
526 list_del_init(&lock->l_pending_chain);
531 int ldlm_del_waiting_lock(struct ldlm_lock *lock)
535 if (lock->l_export == NULL) {
536 /* We don't have a "waiting locks list" on clients. */
537 CDEBUG(D_DLMTRACE, "Client lock %p : no-op\n", lock);
541 spin_lock_bh(&waiting_locks_spinlock);
542 ret = __ldlm_del_waiting_lock(lock);
543 ldlm_clear_waited(lock);
544 spin_unlock_bh(&waiting_locks_spinlock);
546 /* remove the lock out of export blocking list */
547 spin_lock_bh(&lock->l_export->exp_bl_list_lock);
548 list_del_init(&lock->l_exp_list);
549 spin_unlock_bh(&lock->l_export->exp_bl_list_lock);
553 * release lock ref if it has indeed been removed
556 LDLM_LOCK_RELEASE(lock);
559 LDLM_DEBUG(lock, "%s", ret == 0 ? "wasn't waiting" : "removed");
564 * Prolong the contended lock waiting time.
566 * Called with namespace lock held.
568 int ldlm_refresh_waiting_lock(struct ldlm_lock *lock, timeout_t timeout)
570 if (lock->l_export == NULL) {
571 /* We don't have a "waiting locks list" on clients. */
572 LDLM_DEBUG(lock, "client lock: no-op");
576 if (exp_connect_flags(lock->l_export) & OBD_CONNECT_MDS_MDS) {
577 /* We don't have a "waiting locks list" on OSP. */
578 LDLM_DEBUG(lock, "MDS-MDS lock: no-op");
582 spin_lock_bh(&waiting_locks_spinlock);
584 if (list_empty(&lock->l_pending_chain)) {
585 spin_unlock_bh(&waiting_locks_spinlock);
586 LDLM_DEBUG(lock, "wasn't waiting");
591 * we remove/add the lock to the waiting list, so no needs to
592 * release/take a lock reference
594 __ldlm_del_waiting_lock(lock);
595 __ldlm_add_waiting_lock(lock, timeout);
596 spin_unlock_bh(&waiting_locks_spinlock);
598 LDLM_DEBUG(lock, "refreshed to %ds", timeout);
601 EXPORT_SYMBOL(ldlm_refresh_waiting_lock);
603 #else /* HAVE_SERVER_SUPPORT */
605 int ldlm_del_waiting_lock(struct ldlm_lock *lock)
610 int ldlm_refresh_waiting_lock(struct ldlm_lock *lock, timeout_t timeout)
615 #endif /* !HAVE_SERVER_SUPPORT */
617 #ifdef HAVE_SERVER_SUPPORT
620 * Calculate the per-export Blocking timeout (covering BL AST, data flush,
621 * lock cancel, and their replies). Used for lock callback timeout and AST
624 * \param[in] lock lock which is getting the blocking callback
626 * \retval timeout in seconds to wait for the client reply
628 timeout_t ldlm_bl_timeout(struct ldlm_lock *lock)
633 return obd_timeout / 2;
636 * Since these are non-updating timeouts, we should be conservative.
637 * Take more than usually, 150%
638 * It would be nice to have some kind of "early reply" mechanism for
639 * lock callbacks too...
641 timeout = at_get(&lock->l_export->exp_bl_lock_at);
642 return max_t(timeout_t, timeout + (timeout >> 1),
643 (timeout_t)ldlm_enqueue_min);
645 EXPORT_SYMBOL(ldlm_bl_timeout);
648 * Perform lock cleanup if AST sending failed.
650 static void ldlm_failed_ast(struct ldlm_lock *lock, int rc,
651 const char *ast_type)
653 LCONSOLE_ERROR_MSG(0x138,
654 "%s: A client on nid %s was evicted due to a lock %s callback time out: rc %d\n",
655 lock->l_export->exp_obd->obd_name,
656 obd_export_nid2str(lock->l_export), ast_type, rc);
658 if (obd_dump_on_timeout)
659 libcfs_debug_dumplog();
660 spin_lock_bh(&waiting_locks_spinlock);
661 if (__ldlm_del_waiting_lock(lock) == 0)
663 * the lock was not in any list, grab an extra ref before adding
664 * the lock to the expired list
667 /* differentiate it from expired locks */
668 lock->l_callback_timestamp = 0;
669 list_add(&lock->l_pending_chain, &expired_lock_list);
670 wake_up(&expired_lock_wait_queue);
671 spin_unlock_bh(&waiting_locks_spinlock);
675 * Perform lock cleanup if AST reply came with error.
677 static int ldlm_handle_ast_error(struct ldlm_lock *lock,
678 struct ptlrpc_request *req, int rc,
679 const char *ast_type)
681 struct lnet_process_id peer = req->rq_import->imp_connection->c_peer;
683 if (!req->rq_replied || (rc && rc != -EINVAL)) {
684 if (ldlm_is_cancel(lock)) {
686 "%s AST (req@%p x%llu) timeout from nid %s, but cancel was received (AST reply lost?)",
687 ast_type, req, req->rq_xid,
688 libcfs_nid2str(peer.nid));
689 ldlm_lock_cancel(lock);
691 } else if (rc == -ENODEV || rc == -ESHUTDOWN ||
693 req->rq_import->imp_state == LUSTRE_IMP_CLOSED)) {
695 * Upon umount process the AST fails because cannot be
696 * sent. This shouldn't lead to the client eviction.
697 * -ENODEV error is returned by ptl_send_rpc() for
698 * new request in such import.
699 * -SHUTDOWN is returned by ptlrpc_import_delay_req()
700 * if imp_invalid is set or obd_no_recov.
701 * Meanwhile there is also check for LUSTRE_IMP_CLOSED
702 * in ptlrpc_import_delay_req() as well with -EIO code.
703 * In all such cases errors are ignored.
706 "%s AST can't be sent due to a server %s failure or umount process: rc = %d\n",
708 req->rq_import->imp_obd->obd_name, rc);
711 "client (nid %s) %s %s AST (req@%p x%llu status %d rc %d), evict it",
712 libcfs_nid2str(peer.nid),
713 req->rq_replied ? "returned error from" :
714 "failed to reply to",
715 ast_type, req, req->rq_xid,
716 (req->rq_repmsg != NULL) ?
717 lustre_msg_get_status(req->rq_repmsg) : 0,
719 ldlm_failed_ast(lock, rc, ast_type);
725 struct ldlm_resource *res = lock->l_resource;
728 "client (nid %s) returned %d from %s AST (req@%p x%llu) - normal race",
729 libcfs_nid2str(peer.nid),
731 lustre_msg_get_status(req->rq_repmsg) : -1,
732 ast_type, req, req->rq_xid);
735 * update lvbo to return proper attributes.
738 ldlm_resource_getref(res);
739 ldlm_lvbo_update(res, lock, NULL, 1);
740 ldlm_resource_putref(res);
742 ldlm_lock_cancel(lock);
749 static int ldlm_cb_interpret(const struct lu_env *env,
750 struct ptlrpc_request *req, void *args, int rc)
752 struct ldlm_cb_async_args *ca = args;
753 struct ldlm_lock *lock = ca->ca_lock;
754 struct ldlm_cb_set_arg *arg = ca->ca_set_arg;
758 LASSERT(lock != NULL);
761 case LDLM_GL_CALLBACK:
763 * Update the LVB from disk if the AST failed
764 * (this is a legal race)
766 * - Glimpse callback of local lock just returns
767 * -ELDLM_NO_LOCK_DATA.
768 * - Glimpse callback of remote lock might return
769 * -ELDLM_NO_LOCK_DATA when inode is cleared. LU-274
771 if (unlikely(arg->gl_interpret_reply)) {
772 rc = arg->gl_interpret_reply(NULL, req, args, rc);
773 } else if (rc == -ELDLM_NO_LOCK_DATA) {
775 "lost race - client has a lock but no inode");
776 ldlm_lvbo_update(lock->l_resource, lock, NULL, 1);
777 } else if (rc != 0) {
778 rc = ldlm_handle_ast_error(lock, req, rc, "glimpse");
780 rc = ldlm_lvbo_update(lock->l_resource,
784 case LDLM_BL_CALLBACK:
786 rc = ldlm_handle_ast_error(lock, req, rc, "blocking");
788 case LDLM_CP_CALLBACK:
790 rc = ldlm_handle_ast_error(lock, req, rc, "completion");
793 LDLM_ERROR(lock, "invalid opcode for lock callback %d",
798 /* release extra reference taken in ldlm_ast_fini() */
799 LDLM_LOCK_RELEASE(lock);
802 atomic_inc(&arg->restart);
807 static void ldlm_update_resend(struct ptlrpc_request *req, void *data)
809 struct ldlm_cb_async_args *ca = data;
810 struct ldlm_lock *lock = ca->ca_lock;
812 ldlm_refresh_waiting_lock(lock, ldlm_bl_timeout(lock));
815 static inline int ldlm_ast_fini(struct ptlrpc_request *req,
816 struct ldlm_cb_set_arg *arg,
817 struct ldlm_lock *lock,
824 if (unlikely(instant_cancel)) {
825 rc = ptl_send_rpc(req, 1);
826 ptlrpc_req_finished(req);
828 atomic_inc(&arg->restart);
831 ptlrpc_set_add_req(arg->set, req);
838 * Check if there are requests in the export request list which prevent
839 * the lock canceling and make these requests high priority ones.
841 static void ldlm_lock_reorder_req(struct ldlm_lock *lock)
843 struct ptlrpc_request *req;
847 if (lock->l_export == NULL) {
848 LDLM_DEBUG(lock, "client lock: no-op");
852 spin_lock(&lock->l_export->exp_rpc_lock);
853 list_for_each_entry(req, &lock->l_export->exp_hp_rpcs,
856 * Do not process requests that were not yet added to there
857 * incoming queue or were already removed from there for
858 * processing. We evaluate ptlrpc_nrs_req_can_move() without
859 * holding svcpt->scp_req_lock, and then redo the check with
860 * the lock held once we need to obtain a reliable result.
862 if (ptlrpc_nrs_req_can_move(req) &&
863 req->rq_ops->hpreq_lock_match &&
864 req->rq_ops->hpreq_lock_match(req, lock))
865 ptlrpc_nrs_req_hp_move(req);
867 spin_unlock(&lock->l_export->exp_rpc_lock);
872 * ->l_blocking_ast() method for server-side locks. This is invoked when newly
873 * enqueued server lock conflicts with given one.
875 * Sends blocking AST RPC to the client owning that lock; arms timeout timer
876 * to wait for client response.
878 int ldlm_server_blocking_ast(struct ldlm_lock *lock,
879 struct ldlm_lock_desc *desc,
880 void *data, int flag)
882 struct ldlm_cb_async_args *ca;
883 struct ldlm_cb_set_arg *arg = data;
884 struct ldlm_request *body;
885 struct ptlrpc_request *req;
886 int instant_cancel = 0;
891 if (flag == LDLM_CB_CANCELING)
892 /* Don't need to do anything here. */
895 if (OBD_FAIL_PRECHECK(OBD_FAIL_LDLM_SRV_BL_AST)) {
896 LDLM_DEBUG(lock, "dropping BL AST");
901 LASSERT(data != NULL);
902 if (lock->l_export->exp_obd->obd_recovering != 0)
903 LDLM_ERROR(lock, "BUG 6063: lock collide during recovery");
905 ldlm_lock_reorder_req(lock);
907 req = ptlrpc_request_alloc_pack(lock->l_export->exp_imp_reverse,
908 &RQF_LDLM_BL_CALLBACK,
909 LUSTRE_DLM_VERSION, LDLM_BL_CALLBACK);
913 ca = ptlrpc_req_async_args(ca, req);
914 ca->ca_set_arg = arg;
917 req->rq_interpret_reply = ldlm_cb_interpret;
919 lock_res_and_lock(lock);
920 if (ldlm_is_destroyed(lock)) {
921 /* What's the point? */
922 unlock_res_and_lock(lock);
923 ptlrpc_req_finished(req);
927 if (!ldlm_is_granted(lock)) {
929 * this blocking AST will be communicated as part of the
930 * completion AST instead
932 ldlm_add_blocked_lock(lock);
933 ldlm_set_waited(lock);
934 unlock_res_and_lock(lock);
936 ptlrpc_req_finished(req);
937 LDLM_DEBUG(lock, "lock not granted, not sending blocking AST");
941 if (ldlm_is_cancel_on_block(lock))
944 body = req_capsule_client_get(&req->rq_pill, &RMF_DLM_REQ);
945 body->lock_handle[0] = lock->l_remote_handle;
946 body->lock_handle[1].cookie = lock->l_handle.h_cookie;
947 body->lock_desc = *desc;
948 body->lock_flags |= ldlm_flags_to_wire(lock->l_flags & LDLM_FL_AST_MASK);
950 LDLM_DEBUG(lock, "server preparing blocking AST");
952 ptlrpc_request_set_replen(req);
953 ldlm_set_cbpending(lock);
954 if (instant_cancel) {
955 unlock_res_and_lock(lock);
956 ldlm_lock_cancel(lock);
958 req->rq_no_resend = 1;
960 LASSERT(ldlm_is_granted(lock));
961 ldlm_add_waiting_lock(lock, ldlm_bl_timeout(lock));
962 unlock_res_and_lock(lock);
964 /* Do not resend after lock callback timeout */
965 req->rq_delay_limit = ldlm_bl_timeout(lock);
966 req->rq_resend_cb = ldlm_update_resend;
969 req->rq_send_state = LUSTRE_IMP_FULL;
970 /* ptlrpc_request_alloc_pack already set timeout */
972 req->rq_timeout = ldlm_get_rq_timeout();
974 if (lock->l_export && lock->l_export->exp_nid_stats &&
975 lock->l_export->exp_nid_stats->nid_ldlm_stats)
976 lprocfs_counter_incr(lock->l_export->exp_nid_stats->nid_ldlm_stats,
977 LDLM_BL_CALLBACK - LDLM_FIRST_OPC);
979 rc = ldlm_ast_fini(req, arg, lock, instant_cancel);
985 * ->l_completion_ast callback for a remote lock in server namespace.
987 * Sends AST to the client notifying it of lock granting. If initial
988 * lock response was not sent yet, instead of sending another RPC, just
989 * mark the lock as granted and client will understand
991 int ldlm_server_completion_ast(struct ldlm_lock *lock, __u64 flags, void *data)
993 struct ldlm_cb_set_arg *arg = data;
994 struct ldlm_request *body;
995 struct ptlrpc_request *req;
996 struct ldlm_cb_async_args *ca;
997 int instant_cancel = 0;
1003 LASSERT(lock != NULL);
1004 LASSERT(data != NULL);
1006 if (OBD_FAIL_PRECHECK(OBD_FAIL_LDLM_SRV_CP_AST)) {
1007 LDLM_DEBUG(lock, "dropping CP AST");
1011 req = ptlrpc_request_alloc(lock->l_export->exp_imp_reverse,
1012 &RQF_LDLM_CP_CALLBACK);
1016 /* server namespace, doesn't need lock */
1017 lvb_len = ldlm_lvbo_size(lock);
1019 * LU-3124 & LU-2187: to not return layout in completion AST because
1020 * it may deadlock for LU-2187, or client may not have enough space
1021 * for large layout. The layout will be returned to client with an
1022 * extra RPC to fetch xattr.lov
1024 if (ldlm_has_layout(lock))
1027 req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_CLIENT, lvb_len);
1028 rc = ptlrpc_request_pack(req, LUSTRE_DLM_VERSION, LDLM_CP_CALLBACK);
1030 ptlrpc_request_free(req);
1034 ca = ptlrpc_req_async_args(ca, req);
1035 ca->ca_set_arg = arg;
1038 req->rq_interpret_reply = ldlm_cb_interpret;
1039 body = req_capsule_client_get(&req->rq_pill, &RMF_DLM_REQ);
1041 body->lock_handle[0] = lock->l_remote_handle;
1042 body->lock_handle[1].cookie = lock->l_handle.h_cookie;
1043 body->lock_flags = ldlm_flags_to_wire(flags);
1044 ldlm_lock2desc(lock, &body->lock_desc);
1046 void *lvb = req_capsule_client_get(&req->rq_pill, &RMF_DLM_LVB);
1047 lvb_len = ldlm_lvbo_fill(lock, lvb, &lvb_len);
1050 * We still need to send the RPC to wake up the blocked
1051 * enqueue thread on the client.
1053 * Consider old client, there is no better way to notify
1054 * the failure, just zero-sized the LVB, then the client
1055 * will fail out as "-EPROTO".
1057 req_capsule_shrink(&req->rq_pill, &RMF_DLM_LVB, 0,
1061 req_capsule_shrink(&req->rq_pill, &RMF_DLM_LVB, lvb_len,
1066 LDLM_DEBUG(lock, "server preparing completion AST");
1068 ptlrpc_request_set_replen(req);
1070 req->rq_send_state = LUSTRE_IMP_FULL;
1071 /* ptlrpc_request_pack already set timeout */
1073 req->rq_timeout = ldlm_get_rq_timeout();
1075 /* We only send real blocking ASTs after the lock is granted */
1076 lock_res_and_lock(lock);
1077 if (ldlm_is_ast_sent(lock)) {
1078 body->lock_flags |= ldlm_flags_to_wire(LDLM_FL_AST_SENT);
1079 /* Copy AST flags like LDLM_FL_DISCARD_DATA. */
1080 body->lock_flags |= ldlm_flags_to_wire(lock->l_flags &
1084 * We might get here prior to ldlm_handle_enqueue setting
1085 * LDLM_FL_CANCEL_ON_BLOCK flag. Then we will put this lock
1086 * into waiting list, but this is safe and similar code in
1087 * ldlm_handle_enqueue will call ldlm_lock_cancel() still,
1088 * that would not only cancel the lock, but will also remove
1089 * it from waiting list
1091 if (ldlm_is_cancel_on_block(lock)) {
1092 unlock_res_and_lock(lock);
1093 ldlm_lock_cancel(lock);
1096 req->rq_no_resend = 1;
1098 lock_res_and_lock(lock);
1100 /* start the lock-timeout clock */
1101 ldlm_add_waiting_lock(lock, ldlm_bl_timeout(lock));
1102 /* Do not resend after lock callback timeout */
1103 req->rq_delay_limit = ldlm_bl_timeout(lock);
1104 req->rq_resend_cb = ldlm_update_resend;
1107 unlock_res_and_lock(lock);
1109 if (lock->l_export && lock->l_export->exp_nid_stats &&
1110 lock->l_export->exp_nid_stats->nid_ldlm_stats)
1111 lprocfs_counter_incr(lock->l_export->exp_nid_stats->nid_ldlm_stats,
1112 LDLM_CP_CALLBACK - LDLM_FIRST_OPC);
1114 rc = ldlm_ast_fini(req, arg, lock, instant_cancel);
1116 RETURN(lvb_len < 0 ? lvb_len : rc);
1120 * Server side ->l_glimpse_ast handler for client locks.
1122 * Sends glimpse AST to the client and waits for reply. Then updates
1123 * lvbo with the result.
1125 int ldlm_server_glimpse_ast(struct ldlm_lock *lock, void *data)
1127 struct ldlm_cb_set_arg *arg = data;
1128 struct ldlm_request *body;
1129 struct ptlrpc_request *req;
1130 struct ldlm_cb_async_args *ca;
1132 struct req_format *req_fmt;
1136 LASSERT(lock != NULL);
1138 if (arg->gl_desc != NULL)
1139 /* There is a glimpse descriptor to pack */
1140 req_fmt = &RQF_LDLM_GL_CALLBACK_DESC;
1142 req_fmt = &RQF_LDLM_GL_CALLBACK;
1144 req = ptlrpc_request_alloc_pack(lock->l_export->exp_imp_reverse,
1145 req_fmt, LUSTRE_DLM_VERSION,
1151 if (arg->gl_desc != NULL) {
1152 /* copy the GL descriptor */
1153 union ldlm_gl_desc *desc;
1155 desc = req_capsule_client_get(&req->rq_pill, &RMF_DLM_GL_DESC);
1156 *desc = *arg->gl_desc;
1159 body = req_capsule_client_get(&req->rq_pill, &RMF_DLM_REQ);
1160 body->lock_handle[0] = lock->l_remote_handle;
1161 ldlm_lock2desc(lock, &body->lock_desc);
1163 ca = ptlrpc_req_async_args(ca, req);
1164 ca->ca_set_arg = arg;
1167 /* server namespace, doesn't need lock */
1168 req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER,
1169 ldlm_lvbo_size(lock));
1170 ptlrpc_request_set_replen(req);
1172 req->rq_send_state = LUSTRE_IMP_FULL;
1173 /* ptlrpc_request_alloc_pack already set timeout */
1175 req->rq_timeout = ldlm_get_rq_timeout();
1177 req->rq_interpret_reply = ldlm_cb_interpret;
1179 if (lock->l_export && lock->l_export->exp_nid_stats) {
1180 struct nid_stat *nid_stats = lock->l_export->exp_nid_stats;
1182 lprocfs_counter_incr(nid_stats->nid_ldlm_stats,
1183 LDLM_GL_CALLBACK - LDLM_FIRST_OPC);
1186 rc = ldlm_ast_fini(req, arg, lock, 0);
1190 EXPORT_SYMBOL(ldlm_server_glimpse_ast);
1192 int ldlm_glimpse_locks(struct ldlm_resource *res,
1193 struct list_head *gl_work_list)
1199 rc = ldlm_run_ast_work(ldlm_res_to_ns(res), gl_work_list,
1201 if (rc == -ERESTART)
1202 ldlm_reprocess_all(res, NULL);
1206 EXPORT_SYMBOL(ldlm_glimpse_locks);
1208 /* return LDLM lock associated with a lock callback request */
1209 struct ldlm_lock *ldlm_request_lock(struct ptlrpc_request *req)
1211 struct ldlm_cb_async_args *ca;
1212 struct ldlm_lock *lock;
1216 ca = ptlrpc_req_async_args(ca, req);
1219 RETURN(ERR_PTR(-EFAULT));
1223 EXPORT_SYMBOL(ldlm_request_lock);
1226 * Main server-side entry point into LDLM for enqueue. This is called by ptlrpc
1227 * service threads to carry out client lock enqueueing requests.
1229 int ldlm_handle_enqueue0(struct ldlm_namespace *ns,
1230 struct ptlrpc_request *req,
1231 const struct ldlm_request *dlm_req,
1232 const struct ldlm_callback_suite *cbs)
1234 struct ldlm_reply *dlm_rep;
1236 enum ldlm_error err = ELDLM_OK;
1237 struct ldlm_lock *lock = NULL;
1238 void *cookie = NULL;
1240 struct ldlm_resource *res = NULL;
1241 const struct lu_env *env = req->rq_svc_thread->t_env;
1245 LDLM_DEBUG_NOLOCK("server-side enqueue handler START");
1247 ldlm_request_cancel(req, dlm_req, LDLM_ENQUEUE_CANCEL_OFF, LATF_SKIP);
1248 flags = ldlm_flags_from_wire(dlm_req->lock_flags);
1250 LASSERT(req->rq_export);
1252 /* for intent enqueue the stat will be updated inside intent policy */
1253 if (ptlrpc_req2svc(req)->srv_stats != NULL &&
1254 !(dlm_req->lock_flags & LDLM_FL_HAS_INTENT))
1255 ldlm_svc_get_eopc(dlm_req, ptlrpc_req2svc(req)->srv_stats);
1257 if (req->rq_export && req->rq_export->exp_nid_stats &&
1258 req->rq_export->exp_nid_stats->nid_ldlm_stats)
1259 lprocfs_counter_incr(req->rq_export->exp_nid_stats->nid_ldlm_stats,
1260 LDLM_ENQUEUE - LDLM_FIRST_OPC);
1262 if (unlikely(dlm_req->lock_desc.l_resource.lr_type < LDLM_MIN_TYPE ||
1263 dlm_req->lock_desc.l_resource.lr_type >= LDLM_MAX_TYPE)) {
1264 DEBUG_REQ(D_ERROR, req, "invalid lock request type %d",
1265 dlm_req->lock_desc.l_resource.lr_type);
1266 GOTO(out, rc = -EFAULT);
1269 if (unlikely(dlm_req->lock_desc.l_req_mode <= LCK_MINMODE ||
1270 dlm_req->lock_desc.l_req_mode >= LCK_MAXMODE ||
1271 dlm_req->lock_desc.l_req_mode &
1272 (dlm_req->lock_desc.l_req_mode-1))) {
1273 DEBUG_REQ(D_ERROR, req, "invalid lock request mode %d",
1274 dlm_req->lock_desc.l_req_mode);
1275 GOTO(out, rc = -EFAULT);
1278 if (unlikely((flags & LDLM_FL_REPLAY) ||
1279 (lustre_msg_get_flags(req->rq_reqmsg) & MSG_RESENT))) {
1280 /* Find an existing lock in the per-export lock hash */
1282 * In the function below, .hs_keycmp resolves to
1283 * ldlm_export_lock_keycmp()
1285 /* coverity[overrun-buffer-val] */
1286 lock = cfs_hash_lookup(req->rq_export->exp_lock_hash,
1287 (void *)&dlm_req->lock_handle[0]);
1289 DEBUG_REQ(D_DLMTRACE, req,
1290 "found existing lock cookie %#llx",
1291 lock->l_handle.h_cookie);
1292 flags |= LDLM_FL_RESENT;
1293 GOTO(existing_lock, rc = 0);
1296 if (ldlm_reclaim_full()) {
1297 DEBUG_REQ(D_DLMTRACE, req,
1298 "Too many granted locks, reject current enqueue request and let the client retry later");
1299 GOTO(out, rc = -EINPROGRESS);
1303 /* The lock's callback data might be set in the policy function */
1304 lock = ldlm_lock_create(ns, &dlm_req->lock_desc.l_resource.lr_name,
1305 dlm_req->lock_desc.l_resource.lr_type,
1306 dlm_req->lock_desc.l_req_mode,
1307 cbs, NULL, 0, LVB_T_NONE);
1314 lock->l_remote_handle = dlm_req->lock_handle[0];
1315 LDLM_DEBUG(lock, "server-side enqueue handler, new lock created");
1318 * Initialize resource lvb but not for a lock being replayed since
1319 * Client already got lvb sent in this case.
1320 * This must occur early since some policy methods assume resource
1321 * lvb is available (lr_lvb_data != NULL).
1323 res = lock->l_resource;
1324 if (!(flags & LDLM_FL_REPLAY)) {
1325 /* non-replayed lock, delayed lvb init may need to be done */
1326 rc = ldlm_lvbo_init(res);
1328 LDLM_DEBUG(lock, "delayed lvb init failed (rc %d)", rc);
1333 OBD_FAIL_TIMEOUT(OBD_FAIL_LDLM_ENQUEUE_BLOCKED, obd_timeout * 2);
1335 * Don't enqueue a lock onto the export if it is been disonnected
1336 * due to eviction (b=3822) or server umount (b=24324).
1337 * Cancel it now instead.
1339 if (req->rq_export->exp_disconnected) {
1340 LDLM_ERROR(lock, "lock on disconnected export %p",
1342 GOTO(out, rc = -ENOTCONN);
1345 lock->l_export = class_export_lock_get(req->rq_export, lock);
1346 if (lock->l_export->exp_lock_hash)
1347 cfs_hash_add(lock->l_export->exp_lock_hash,
1348 &lock->l_remote_handle,
1352 * Inherit the enqueue flags before the operation, because we do not
1353 * keep the res lock on return and next operations (BL AST) may proceed
1356 lock->l_flags |= ldlm_flags_from_wire(dlm_req->lock_flags &
1357 LDLM_FL_INHERIT_MASK);
1359 ldlm_convert_policy_to_local(req->rq_export,
1360 dlm_req->lock_desc.l_resource.lr_type,
1361 &dlm_req->lock_desc.l_policy_data,
1362 &lock->l_policy_data);
1363 if (dlm_req->lock_desc.l_resource.lr_type == LDLM_EXTENT) {
1364 lock->l_req_extent = lock->l_policy_data.l_extent;
1365 } else if (dlm_req->lock_desc.l_resource.lr_type == LDLM_IBITS) {
1366 lock->l_policy_data.l_inodebits.try_bits =
1367 dlm_req->lock_desc.l_policy_data.l_inodebits.try_bits;
1368 lock->l_policy_data.l_inodebits.li_gid =
1369 dlm_req->lock_desc.l_policy_data.l_inodebits.li_gid;
1374 if (!(flags & LDLM_FL_HAS_INTENT)) {
1375 /* based on the assumption that lvb size never changes during
1376 * resource life time otherwise it need resource->lr_lock's
1378 req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB,
1379 RCL_SERVER, ldlm_lvbo_size(lock));
1381 if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_ENQUEUE_EXTENT_ERR))
1382 GOTO(out, rc = -ENOMEM);
1384 rc = req_capsule_server_pack(&req->rq_pill);
1389 err = ldlm_lock_enqueue(env, ns, &lock, cookie, &flags);
1396 dlm_rep = req_capsule_server_get(&req->rq_pill, &RMF_DLM_REP);
1398 ldlm_lock2desc(lock, &dlm_rep->lock_desc);
1399 ldlm_lock2handle(lock, &dlm_rep->lock_handle);
1401 if (lock && lock->l_resource->lr_type == LDLM_EXTENT)
1402 OBD_FAIL_TIMEOUT(OBD_FAIL_LDLM_BL_EVICT, 6);
1405 * We never send a blocking AST until the lock is granted, but
1406 * we can tell it right now
1408 lock_res_and_lock(lock);
1411 * Now take into account flags to be inherited from original lock
1412 * request both in reply to client and in our own lock flags.
1414 dlm_rep->lock_flags = ldlm_flags_to_wire(flags);
1415 lock->l_flags |= flags & LDLM_FL_INHERIT_MASK;
1418 * Don't move a pending lock onto the export if it has already been
1419 * disconnected due to eviction (b=5683) or server umount (b=24324).
1420 * Cancel it now instead.
1422 if (unlikely(req->rq_export->exp_disconnected ||
1423 OBD_FAIL_CHECK(OBD_FAIL_LDLM_ENQUEUE_OLD_EXPORT))) {
1424 LDLM_ERROR(lock, "lock on destroyed export %p", req->rq_export);
1426 } else if (ldlm_is_ast_sent(lock)) {
1427 /* fill lock desc for possible lock convert */
1428 if (lock->l_blocking_lock &&
1429 lock->l_resource->lr_type == LDLM_IBITS) {
1430 struct ldlm_lock *bl_lock = lock->l_blocking_lock;
1431 struct ldlm_lock_desc *rep_desc = &dlm_rep->lock_desc;
1434 "save blocking bits %llx in granted lock",
1435 bl_lock->l_policy_data.l_inodebits.bits);
1437 * If lock is blocked then save blocking ibits
1438 * in returned lock policy for the possible lock
1439 * convert on a client.
1441 rep_desc->l_policy_data.l_inodebits.cancel_bits =
1442 bl_lock->l_policy_data.l_inodebits.bits;
1444 dlm_rep->lock_flags |= ldlm_flags_to_wire(LDLM_FL_AST_SENT);
1445 if (ldlm_is_granted(lock)) {
1447 * Only cancel lock if it was granted, because it would
1448 * be destroyed immediately and would never be granted
1449 * in the future, causing timeouts on client. Not
1450 * granted lock will be cancelled immediately after
1451 * sending completion AST.
1453 if (ldlm_is_cancel_on_block(lock)) {
1454 unlock_res_and_lock(lock);
1455 ldlm_lock_cancel(lock);
1456 lock_res_and_lock(lock);
1458 ldlm_add_waiting_lock(lock,
1459 ldlm_bl_timeout(lock));
1463 unlock_res_and_lock(lock);
1467 req->rq_status = rc ?: err; /* return either error - b=11190 */
1468 if (!req->rq_packed_final) {
1469 int rc1 = lustre_pack_reply(req, 1, NULL, NULL);
1475 * The LOCK_CHANGED code in ldlm_lock_enqueue depends on this
1476 * ldlm_reprocess_all. If this moves, revisit that code. -phil
1480 "server-side enqueue handler, sending reply (err=%d, rc=%d)",
1484 req_capsule_has_field(&req->rq_pill, &RMF_DLM_LVB,
1486 ldlm_lvbo_size(lock) > 0) {
1491 buf = req_capsule_server_get(&req->rq_pill,
1493 LASSERTF(buf != NULL, "req %p, lock %p\n", req, lock);
1494 buflen = req_capsule_get_size(&req->rq_pill,
1495 &RMF_DLM_LVB, RCL_SERVER);
1497 * non-replayed lock, delayed lvb init may
1498 * need to be occur now
1500 if ((buflen > 0) && !(flags & LDLM_FL_REPLAY)) {
1503 rc2 = ldlm_lvbo_fill(lock, buf, &buflen);
1505 req_capsule_shrink(&req->rq_pill,
1508 } else if (rc2 == -ERANGE) {
1509 rc2 = req_capsule_server_grow(
1511 &RMF_DLM_LVB, buflen);
1516 * if we can't grow the buffer,
1517 * it's ok to return empty lvb
1528 } else if (flags & LDLM_FL_REPLAY) {
1529 /* no LVB resend upon replay */
1531 req_capsule_shrink(&req->rq_pill,
1541 if (rc != 0 && !(flags & LDLM_FL_RESENT)) {
1542 if (lock->l_export) {
1543 ldlm_lock_cancel(lock);
1545 lock_res_and_lock(lock);
1546 ldlm_resource_unlink_lock(lock);
1547 ldlm_lock_destroy_nolock(lock);
1548 unlock_res_and_lock(lock);
1550 ldlm_reprocess_all(lock->l_resource, lock);
1553 if (!err && !ldlm_is_cbpending(lock) &&
1554 dlm_req->lock_desc.l_resource.lr_type != LDLM_FLOCK)
1555 ldlm_reprocess_all(lock->l_resource, lock);
1557 LDLM_LOCK_RELEASE(lock);
1560 LDLM_DEBUG_NOLOCK("server-side enqueue handler END (lock %p, rc %d)",
1567 * Clear the blocking lock, the race is possible between ldlm_handle_convert0()
1568 * and ldlm_work_bl_ast_lock(), so this is done under lock with check for NULL.
1570 void ldlm_clear_blocking_lock(struct ldlm_lock *lock)
1572 if (lock->l_blocking_lock) {
1573 LDLM_LOCK_RELEASE(lock->l_blocking_lock);
1574 lock->l_blocking_lock = NULL;
1578 /* A lock can be converted to new ibits or mode and should be considered
1579 * as new lock. Clear all states related to a previous blocking AST
1580 * processing so new conflicts will cause new blocking ASTs.
1582 * This is used during lock convert below and lock downgrade to COS mode in
1583 * ldlm_lock_mode_downgrade().
1585 void ldlm_clear_blocking_data(struct ldlm_lock *lock)
1587 ldlm_clear_ast_sent(lock);
1588 lock->l_bl_ast_run = 0;
1589 ldlm_clear_blocking_lock(lock);
1593 * Main LDLM entry point for server code to process lock conversion requests.
1595 int ldlm_handle_convert0(struct ptlrpc_request *req,
1596 const struct ldlm_request *dlm_req)
1598 struct obd_export *exp = req->rq_export;
1599 struct ldlm_reply *dlm_rep;
1600 struct ldlm_lock *lock;
1607 if (exp && exp->exp_nid_stats && exp->exp_nid_stats->nid_ldlm_stats)
1608 lprocfs_counter_incr(exp->exp_nid_stats->nid_ldlm_stats,
1609 LDLM_CONVERT - LDLM_FIRST_OPC);
1611 rc = req_capsule_server_pack(&req->rq_pill);
1615 dlm_rep = req_capsule_server_get(&req->rq_pill, &RMF_DLM_REP);
1616 dlm_rep->lock_flags = dlm_req->lock_flags;
1618 lock = ldlm_handle2lock(&dlm_req->lock_handle[0]);
1620 LDLM_DEBUG_NOLOCK("server lock is canceled already");
1621 req->rq_status = ELDLM_NO_LOCK_DATA;
1625 LDLM_DEBUG(lock, "server-side convert handler START");
1627 lock_res_and_lock(lock);
1628 bits = lock->l_policy_data.l_inodebits.bits;
1629 new_bits = dlm_req->lock_desc.l_policy_data.l_inodebits.bits;
1631 if (ldlm_is_cancel(lock)) {
1632 LDLM_DEBUG(lock, "convert on canceled lock!");
1633 unlock_res_and_lock(lock);
1634 GOTO(out_put, rc = ELDLM_NO_LOCK_DATA);
1637 if (dlm_req->lock_desc.l_req_mode != lock->l_granted_mode) {
1638 LDLM_ERROR(lock, "lock mode differs!");
1639 unlock_res_and_lock(lock);
1640 GOTO(out_put, rc = -EPROTO);
1643 if (bits == new_bits) {
1645 * This can be valid situation if CONVERT RPCs are
1646 * re-ordered. Just finish silently
1648 LDLM_DEBUG(lock, "lock is converted already!");
1649 unlock_res_and_lock(lock);
1651 if (ldlm_is_waited(lock))
1652 ldlm_del_waiting_lock(lock);
1654 ldlm_clear_cbpending(lock);
1655 lock->l_policy_data.l_inodebits.cancel_bits = 0;
1656 ldlm_inodebits_drop(lock, bits & ~new_bits);
1658 ldlm_clear_blocking_data(lock);
1659 unlock_res_and_lock(lock);
1661 ldlm_reprocess_all(lock->l_resource, NULL);
1664 dlm_rep->lock_handle = lock->l_remote_handle;
1665 ldlm_ibits_policy_local_to_wire(&lock->l_policy_data,
1666 &dlm_rep->lock_desc.l_policy_data);
1670 LDLM_DEBUG(lock, "server-side convert handler END, rc = %d", rc);
1671 LDLM_LOCK_PUT(lock);
1672 req->rq_status = rc;
1677 * Cancel all the locks whose handles are packed into ldlm_request
1679 * Called by server code expecting such combined cancel activity
1682 int ldlm_request_cancel(struct ptlrpc_request *req,
1683 const struct ldlm_request *dlm_req,
1684 int first, enum lustre_at_flags flags)
1686 struct ldlm_resource *res, *pres = NULL;
1687 struct ldlm_lock *lock;
1688 int i, count, done = 0;
1693 size = req_capsule_get_size(&req->rq_pill, &RMF_DLM_REQ, RCL_CLIENT);
1694 if (size <= offsetof(struct ldlm_request, lock_handle) ||
1695 (size - offsetof(struct ldlm_request, lock_handle)) /
1696 sizeof(struct lustre_handle) < dlm_req->lock_count)
1699 count = dlm_req->lock_count ? dlm_req->lock_count : 1;
1703 if (count == 1 && dlm_req->lock_handle[0].cookie == 0)
1707 * There is no lock on the server at the replay time,
1708 * skip lock cancelling to make replay tests to pass.
1710 if (lustre_msg_get_flags(req->rq_reqmsg) & MSG_REPLAY)
1713 LDLM_DEBUG_NOLOCK("server-side cancel handler START: %d locks, starting at %d",
1716 for (i = first; i < count; i++) {
1717 lock = ldlm_handle2lock(&dlm_req->lock_handle[i]);
1719 /* below message checked in replay-single.sh test_36 */
1720 LDLM_DEBUG_NOLOCK("server-side cancel handler stale lock (cookie %llu)",
1721 dlm_req->lock_handle[i].cookie);
1725 res = lock->l_resource;
1729 * This code is an optimization to only attempt lock
1730 * granting on the resource (that could be CPU-expensive)
1731 * after we are done cancelling lock in that resource.
1735 ldlm_reprocess_all(pres, NULL);
1736 LDLM_RESOURCE_DELREF(pres);
1737 ldlm_resource_putref(pres);
1740 ldlm_resource_getref(res);
1741 LDLM_RESOURCE_ADDREF(res);
1743 if (!ldlm_is_discard_data(lock))
1744 ldlm_lvbo_update(res, lock,
1750 if ((flags & LATF_STATS) && ldlm_is_ast_sent(lock) &&
1751 lock->l_blast_sent != 0) {
1752 timeout_t delay = 0;
1754 if (ktime_get_seconds() > lock->l_blast_sent)
1755 delay = ktime_get_seconds() -
1758 "server cancels blocked lock after %ds",
1760 at_measured(&lock->l_export->exp_bl_lock_at, delay);
1762 ldlm_lock_cancel(lock);
1763 LDLM_LOCK_PUT(lock);
1766 ldlm_reprocess_all(pres, NULL);
1767 LDLM_RESOURCE_DELREF(pres);
1768 ldlm_resource_putref(pres);
1770 LDLM_DEBUG_NOLOCK("server-side cancel handler END");
1773 EXPORT_SYMBOL(ldlm_request_cancel);
1776 * Main LDLM entry point for server code to cancel locks.
1778 * Typically gets called from service handler on LDLM_CANCEL opc.
1780 int ldlm_handle_cancel(struct ptlrpc_request *req)
1782 struct ldlm_request *dlm_req;
1787 dlm_req = req_capsule_client_get(&req->rq_pill, &RMF_DLM_REQ);
1788 if (dlm_req == NULL) {
1789 CDEBUG(D_INFO, "bad request buffer for cancel\n");
1793 if (req_capsule_get_size(&req->rq_pill, &RMF_DLM_REQ, RCL_CLIENT) <
1794 offsetof(struct ldlm_request, lock_handle[1]))
1797 if (req->rq_export && req->rq_export->exp_nid_stats &&
1798 req->rq_export->exp_nid_stats->nid_ldlm_stats)
1799 lprocfs_counter_incr(req->rq_export->exp_nid_stats->nid_ldlm_stats,
1800 LDLM_CANCEL - LDLM_FIRST_OPC);
1802 rc = req_capsule_server_pack(&req->rq_pill);
1806 if (!ldlm_request_cancel(req, dlm_req, 0, LATF_STATS))
1807 req->rq_status = LUSTRE_ESTALE;
1809 RETURN(ptlrpc_reply(req));
1811 #endif /* HAVE_SERVER_SUPPORT */
1814 * Server may pass additional information about blocking lock.
1815 * For IBITS locks it is conflicting bits which can be used for
1816 * lock convert instead of cancel.
1818 void ldlm_bl_desc2lock(const struct ldlm_lock_desc *ld, struct ldlm_lock *lock)
1820 struct ldlm_namespace *ns = ldlm_lock_to_ns(lock);
1822 check_res_locked(lock->l_resource);
1823 if (ns_is_client(ns) && ld &&
1824 (lock->l_resource->lr_type == LDLM_IBITS)) {
1826 * Lock description contains policy of blocking lock,
1827 * and its cancel_bits is used to pass conflicting bits.
1828 * NOTE: ld can be NULL or can be not NULL but zeroed if
1829 * passed from ldlm_bl_thread_blwi(), check below used bits
1830 * in ld to make sure it is valid description.
1832 * If server may replace lock resource keeping the same cookie,
1833 * never use cancel bits from different resource, full cancel
1836 if (ld->l_policy_data.l_inodebits.cancel_bits &&
1837 ldlm_res_eq(&ld->l_resource.lr_name,
1838 &lock->l_resource->lr_name) &&
1839 !(ldlm_is_cbpending(lock) &&
1840 lock->l_policy_data.l_inodebits.cancel_bits == 0)) {
1841 /* always combine conflicting ibits */
1842 lock->l_policy_data.l_inodebits.cancel_bits |=
1843 ld->l_policy_data.l_inodebits.cancel_bits;
1845 /* If cancel_bits are not obtained or
1846 * if the lock is already CBPENDING and
1847 * has no cancel_bits set
1848 * - the full lock is to be cancelled
1850 lock->l_policy_data.l_inodebits.cancel_bits = 0;
1856 * Callback handler for receiving incoming blocking ASTs.
1858 * This can only happen on client side.
1860 void ldlm_handle_bl_callback(struct ldlm_namespace *ns,
1861 struct ldlm_lock_desc *ld, struct ldlm_lock *lock)
1867 LDLM_DEBUG(lock, "client blocking AST callback handler");
1869 lock_res_and_lock(lock);
1871 /* get extra information from desc if any */
1872 ldlm_bl_desc2lock(ld, lock);
1873 ldlm_set_cbpending(lock);
1875 do_ast = (!lock->l_readers && !lock->l_writers);
1876 unlock_res_and_lock(lock);
1880 "Lock %p already unused, calling callback (%p)\n",
1881 lock, lock->l_blocking_ast);
1882 if (lock->l_blocking_ast != NULL)
1883 lock->l_blocking_ast(lock, ld, lock->l_ast_data,
1887 "Lock %p is referenced, will be cancelled later\n",
1891 LDLM_DEBUG(lock, "client blocking callback handler END");
1892 LDLM_LOCK_RELEASE(lock);
1896 static int ldlm_callback_reply(struct ptlrpc_request *req, int rc)
1898 if (req->rq_no_reply)
1901 req->rq_status = rc;
1902 if (!req->rq_packed_final) {
1903 rc = lustre_pack_reply(req, 1, NULL, NULL);
1907 return ptlrpc_reply(req);
1911 * Callback handler for receiving incoming completion ASTs.
1913 * This only can happen on client side.
1915 static int ldlm_handle_cp_callback(struct ptlrpc_request *req,
1916 struct ldlm_namespace *ns,
1917 struct ldlm_request *dlm_req,
1918 struct ldlm_lock *lock)
1920 LIST_HEAD(ast_list);
1926 LDLM_DEBUG(lock, "client completion callback handler START");
1928 if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_CANCEL_BL_CB_RACE)) {
1929 long to = cfs_time_seconds(1);
1931 ldlm_callback_reply(req, 0);
1934 to = schedule_timeout_interruptible(to);
1935 if (ldlm_is_granted(lock) ||
1936 ldlm_is_destroyed(lock))
1941 lvb_len = req_capsule_get_size(&req->rq_pill, &RMF_DLM_LVB, RCL_CLIENT);
1943 LDLM_ERROR(lock, "Fail to get lvb_len, rc = %d", lvb_len);
1944 GOTO(out, rc = lvb_len);
1945 } else if (lvb_len > 0) {
1946 if (lock->l_lvb_len > 0) {
1947 /* for extent lock, lvb contains ost_lvb{}. */
1948 LASSERT(lock->l_lvb_data != NULL);
1950 if (unlikely(lock->l_lvb_len < lvb_len)) {
1952 "Replied LVB is larger than expectation, expected = %d, replied = %d",
1953 lock->l_lvb_len, lvb_len);
1954 GOTO(out, rc = -EINVAL);
1959 lock_res_and_lock(lock);
1961 if (!ldlm_res_eq(&dlm_req->lock_desc.l_resource.lr_name,
1962 &lock->l_resource->lr_name)) {
1963 ldlm_resource_unlink_lock(lock);
1964 unlock_res_and_lock(lock);
1965 rc = ldlm_lock_change_resource(ns, lock,
1966 &dlm_req->lock_desc.l_resource.lr_name);
1968 LDLM_ERROR(lock, "Failed to allocate resource");
1971 LDLM_DEBUG(lock, "completion AST, new resource");
1972 lock_res_and_lock(lock);
1975 if (ldlm_is_failed(lock)) {
1976 unlock_res_and_lock(lock);
1977 LDLM_LOCK_RELEASE(lock);
1981 if (ldlm_is_destroyed(lock) ||
1982 ldlm_is_granted(lock)) {
1983 /* b=11300: the lock has already been granted */
1984 unlock_res_and_lock(lock);
1985 LDLM_DEBUG(lock, "Double grant race happened");
1990 * If we receive the completion AST before the actual enqueue returned,
1991 * then we might need to switch lock modes, resources, or extents.
1993 if (dlm_req->lock_desc.l_granted_mode != lock->l_req_mode) {
1994 lock->l_req_mode = dlm_req->lock_desc.l_granted_mode;
1995 LDLM_DEBUG(lock, "completion AST, new lock mode");
1998 if (lock->l_resource->lr_type != LDLM_PLAIN) {
1999 ldlm_convert_policy_to_local(req->rq_export,
2000 dlm_req->lock_desc.l_resource.lr_type,
2001 &dlm_req->lock_desc.l_policy_data,
2002 &lock->l_policy_data);
2003 LDLM_DEBUG(lock, "completion AST, new policy data");
2006 ldlm_resource_unlink_lock(lock);
2008 if (dlm_req->lock_flags & LDLM_FL_AST_SENT) {
2010 * BL_AST locks are not needed in LRU.
2011 * Let ldlm_cancel_lru() be fast.
2013 ldlm_lock_remove_from_lru(lock);
2014 ldlm_bl_desc2lock(&dlm_req->lock_desc, lock);
2015 lock->l_flags |= LDLM_FL_CBPENDING | LDLM_FL_BL_AST;
2016 LDLM_DEBUG(lock, "completion AST includes blocking AST");
2019 if (lock->l_lvb_len > 0) {
2020 rc = ldlm_fill_lvb(lock, &req->rq_pill, RCL_CLIENT,
2021 lock->l_lvb_data, lvb_len);
2023 unlock_res_and_lock(lock);
2028 ldlm_grant_lock(lock, &ast_list);
2029 unlock_res_and_lock(lock);
2031 LDLM_DEBUG(lock, "callback handler finished, about to run_ast_work");
2034 * Let Enqueue to call osc_lock_upcall() and initialize
2037 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_ENQ_RACE, 2);
2039 ldlm_run_ast_work(ns, &ast_list, LDLM_WORK_CP_AST);
2041 LDLM_DEBUG_NOLOCK("client completion callback handler END (lock %p)",
2047 lock_res_and_lock(lock);
2048 ldlm_set_failed(lock);
2049 unlock_res_and_lock(lock);
2050 wake_up(&lock->l_waitq);
2052 LDLM_LOCK_RELEASE(lock);
2058 * Callback handler for receiving incoming glimpse ASTs.
2060 * This only can happen on client side. After handling the glimpse AST
2061 * we also consider dropping the lock here if it is unused locally for a
2064 static void ldlm_handle_gl_callback(struct ptlrpc_request *req,
2065 struct ldlm_namespace *ns,
2066 struct ldlm_request *dlm_req,
2067 struct ldlm_lock *lock)
2069 struct ldlm_lock_desc *ld = &dlm_req->lock_desc;
2074 LDLM_DEBUG(lock, "client glimpse AST callback handler");
2076 if (lock->l_glimpse_ast != NULL)
2077 rc = lock->l_glimpse_ast(lock, req);
2079 if (req->rq_repmsg != NULL) {
2082 req->rq_status = rc;
2086 lock_res_and_lock(lock);
2087 if (lock->l_granted_mode == LCK_PW &&
2088 !lock->l_readers && !lock->l_writers &&
2089 ktime_after(ktime_get(),
2090 ktime_add(lock->l_last_used, ns->ns_dirty_age_limit))) {
2091 unlock_res_and_lock(lock);
2093 /* For MDS glimpse it is always DOM lock, set corresponding
2094 * cancel_bits to perform lock convert if needed
2096 if (lock->l_resource->lr_type == LDLM_IBITS)
2097 ld->l_policy_data.l_inodebits.cancel_bits =
2099 if (ldlm_bl_to_thread_lock(ns, ld, lock))
2100 ldlm_handle_bl_callback(ns, ld, lock);
2105 unlock_res_and_lock(lock);
2106 LDLM_LOCK_RELEASE(lock);
2110 static int __ldlm_bl_to_thread(struct ldlm_bl_work_item *blwi,
2111 enum ldlm_cancel_flags cancel_flags)
2113 struct ldlm_bl_pool *blp = ldlm_state->ldlm_bl_pool;
2117 spin_lock(&blp->blp_lock);
2118 if (blwi->blwi_lock &&
2119 ldlm_is_discard_data(blwi->blwi_lock)) {
2120 /* add LDLM_FL_DISCARD_DATA requests to the priority list */
2121 list_add_tail(&blwi->blwi_entry, &blp->blp_prio_list);
2123 /* other blocking callbacks are added to the regular list */
2124 list_add_tail(&blwi->blwi_entry, &blp->blp_list);
2126 spin_unlock(&blp->blp_lock);
2128 wake_up(&blp->blp_waitq);
2131 * can not check blwi->blwi_flags as blwi could be already freed in
2134 if (!(cancel_flags & LCF_ASYNC))
2135 wait_for_completion(&blwi->blwi_comp);
2140 static inline void init_blwi(struct ldlm_bl_work_item *blwi,
2141 struct ldlm_namespace *ns,
2142 struct ldlm_lock_desc *ld,
2143 struct list_head *cancels, int count,
2144 struct ldlm_lock *lock,
2145 enum ldlm_cancel_flags cancel_flags)
2147 init_completion(&blwi->blwi_comp);
2148 INIT_LIST_HEAD(&blwi->blwi_head);
2150 if (current->flags & PF_MEMALLOC)
2151 blwi->blwi_mem_pressure = 1;
2154 blwi->blwi_flags = cancel_flags;
2156 blwi->blwi_ld = *ld;
2158 list_splice_init(cancels, &blwi->blwi_head);
2159 blwi->blwi_count = count;
2161 blwi->blwi_lock = lock;
2166 * Queues a list of locks \a cancels containing \a count locks
2167 * for later processing by a blocking thread. If \a count is zero,
2168 * then the lock referenced as \a lock is queued instead.
2170 * The blocking thread would then call ->l_blocking_ast callback in the lock.
2171 * If list addition fails an error is returned and caller is supposed to
2172 * call ->l_blocking_ast itself.
2174 static int ldlm_bl_to_thread(struct ldlm_namespace *ns,
2175 struct ldlm_lock_desc *ld,
2176 struct ldlm_lock *lock,
2177 struct list_head *cancels, int count,
2178 enum ldlm_cancel_flags cancel_flags)
2182 if (cancels && count == 0)
2185 if (cancel_flags & LCF_ASYNC) {
2186 struct ldlm_bl_work_item *blwi;
2188 OBD_ALLOC(blwi, sizeof(*blwi));
2191 init_blwi(blwi, ns, ld, cancels, count, lock, cancel_flags);
2193 RETURN(__ldlm_bl_to_thread(blwi, cancel_flags));
2196 * if it is synchronous call do minimum mem alloc, as it could
2197 * be triggered from kernel shrinker
2199 struct ldlm_bl_work_item blwi;
2201 memset(&blwi, 0, sizeof(blwi));
2202 init_blwi(&blwi, ns, ld, cancels, count, lock, cancel_flags);
2203 RETURN(__ldlm_bl_to_thread(&blwi, cancel_flags));
2208 int ldlm_bl_to_thread_lock(struct ldlm_namespace *ns, struct ldlm_lock_desc *ld,
2209 struct ldlm_lock *lock)
2211 return ldlm_bl_to_thread(ns, ld, lock, NULL, 0, LCF_ASYNC);
2214 int ldlm_bl_to_thread_list(struct ldlm_namespace *ns, struct ldlm_lock_desc *ld,
2215 struct list_head *cancels, int count,
2216 enum ldlm_cancel_flags cancel_flags)
2218 return ldlm_bl_to_thread(ns, ld, NULL, cancels, count, cancel_flags);
2221 int ldlm_bl_to_thread_ns(struct ldlm_namespace *ns)
2223 return ldlm_bl_to_thread(ns, NULL, NULL, NULL, 0, LCF_ASYNC);
2226 int ldlm_bl_thread_wakeup(void)
2228 wake_up(&ldlm_state->ldlm_bl_pool->blp_waitq);
2232 /* Setinfo coming from Server (eg MDT) to Client (eg MDC)! */
2233 static int ldlm_handle_setinfo(struct ptlrpc_request *req)
2235 struct obd_device *obd = req->rq_export->exp_obd;
2243 DEBUG_REQ(D_HSM, req, "%s: handle setinfo", obd->obd_name);
2245 req_capsule_set(&req->rq_pill, &RQF_OBD_SET_INFO);
2247 key = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
2249 DEBUG_REQ(D_IOCTL, req, "no set_info key");
2252 keylen = req_capsule_get_size(&req->rq_pill, &RMF_SETINFO_KEY,
2254 val = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_VAL);
2256 DEBUG_REQ(D_IOCTL, req, "no set_info val");
2259 vallen = req_capsule_get_size(&req->rq_pill, &RMF_SETINFO_VAL,
2262 /* We are responsible for swabbing contents of val */
2264 if (KEY_IS(KEY_HSM_COPYTOOL_SEND))
2265 /* Pass it on to mdc (the "export" in this case) */
2266 rc = obd_set_info_async(req->rq_svc_thread->t_env,
2268 sizeof(KEY_HSM_COPYTOOL_SEND),
2269 KEY_HSM_COPYTOOL_SEND,
2272 DEBUG_REQ(D_WARNING, req, "ignoring unknown key '%s'", key);
2277 static inline void ldlm_callback_errmsg(struct ptlrpc_request *req,
2278 const char *msg, int rc,
2279 const struct lustre_handle *handle)
2281 DEBUG_REQ((req->rq_no_reply || rc) ? D_WARNING : D_DLMTRACE, req,
2282 "%s, NID=%s lock=%#llx: rc = %d",
2283 msg, libcfs_id2str(req->rq_peer),
2284 handle ? handle->cookie : 0, rc);
2285 if (req->rq_no_reply)
2286 CWARN("No reply was sent, maybe cause b=21636.\n");
2288 CWARN("Send reply failed, maybe cause b=21636.\n");
2291 /* TODO: handle requests in a similar way as MDT: see mdt_handle_common() */
2292 static int ldlm_callback_handler(struct ptlrpc_request *req)
2294 struct ldlm_namespace *ns;
2295 struct ldlm_request *dlm_req;
2296 struct ldlm_lock *lock;
2302 * Requests arrive in sender's byte order. The ptlrpc service
2303 * handler has already checked and, if necessary, byte-swapped the
2304 * incoming request message body, but I am responsible for the
2308 /* do nothing for sec context finalize */
2309 if (lustre_msg_get_opc(req->rq_reqmsg) == SEC_CTX_FINI)
2312 req_capsule_init(&req->rq_pill, req, RCL_SERVER);
2314 if (req->rq_export == NULL) {
2315 rc = ldlm_callback_reply(req, -ENOTCONN);
2316 ldlm_callback_errmsg(req, "Operate on unconnected server",
2321 LASSERT(req->rq_export != NULL);
2322 LASSERT(req->rq_export->exp_obd != NULL);
2324 switch (lustre_msg_get_opc(req->rq_reqmsg)) {
2325 case LDLM_BL_CALLBACK:
2326 if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_BL_CALLBACK_NET)) {
2328 ldlm_callback_reply(req, -(int)cfs_fail_err);
2332 case LDLM_CP_CALLBACK:
2333 if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_CP_CALLBACK_NET))
2336 case LDLM_GL_CALLBACK:
2337 if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_GL_CALLBACK_NET))
2341 rc = ldlm_handle_setinfo(req);
2342 ldlm_callback_reply(req, rc);
2345 CERROR("unknown opcode %u\n",
2346 lustre_msg_get_opc(req->rq_reqmsg));
2347 ldlm_callback_reply(req, -EPROTO);
2351 ns = req->rq_export->exp_obd->obd_namespace;
2352 LASSERT(ns != NULL);
2354 req_capsule_set(&req->rq_pill, &RQF_LDLM_CALLBACK);
2356 dlm_req = req_capsule_client_get(&req->rq_pill, &RMF_DLM_REQ);
2357 if (dlm_req == NULL) {
2358 rc = ldlm_callback_reply(req, -EPROTO);
2359 ldlm_callback_errmsg(req, "Operate without parameter", rc,
2365 * Force a known safe race, send a cancel to the server for a lock
2366 * which the server has already started a blocking callback on.
2368 if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_CANCEL_BL_CB_RACE) &&
2369 lustre_msg_get_opc(req->rq_reqmsg) == LDLM_BL_CALLBACK) {
2370 rc = ldlm_cli_cancel(&dlm_req->lock_handle[0], 0);
2372 CERROR("ldlm_cli_cancel: %d\n", rc);
2375 lock = ldlm_handle2lock_long(&dlm_req->lock_handle[0], 0);
2378 "callback on lock %#llx - lock disappeared\n",
2379 dlm_req->lock_handle[0].cookie);
2380 rc = ldlm_callback_reply(req, -EINVAL);
2381 ldlm_callback_errmsg(req, "Operate with invalid parameter", rc,
2382 &dlm_req->lock_handle[0]);
2386 if (ldlm_is_fail_loc(lock) &&
2387 lustre_msg_get_opc(req->rq_reqmsg) == LDLM_BL_CALLBACK)
2388 OBD_RACE(OBD_FAIL_LDLM_CP_BL_RACE);
2390 /* Copy hints/flags (e.g. LDLM_FL_DISCARD_DATA) from AST. */
2391 lock_res_and_lock(lock);
2392 lock->l_flags |= ldlm_flags_from_wire(dlm_req->lock_flags &
2394 if (lustre_msg_get_opc(req->rq_reqmsg) == LDLM_BL_CALLBACK) {
2396 * If somebody cancels lock and cache is already dropped,
2397 * or lock is failed before cp_ast received on client,
2398 * we can tell the server we have no lock. Otherwise, we
2399 * should send cancel after dropping the cache.
2401 if ((ldlm_is_canceling(lock) && ldlm_is_bl_done(lock)) ||
2402 ldlm_is_failed(lock)) {
2404 "callback on lock %llx - lock disappeared",
2405 dlm_req->lock_handle[0].cookie);
2406 unlock_res_and_lock(lock);
2407 LDLM_LOCK_RELEASE(lock);
2408 rc = ldlm_callback_reply(req, -EINVAL);
2409 ldlm_callback_errmsg(req, "Operate on stale lock", rc,
2410 &dlm_req->lock_handle[0]);
2414 * BL_AST locks are not needed in LRU.
2415 * Let ldlm_cancel_lru() be fast.
2417 ldlm_lock_remove_from_lru(lock);
2418 ldlm_set_bl_ast(lock);
2420 if (lock->l_remote_handle.cookie == 0)
2421 lock->l_remote_handle = dlm_req->lock_handle[1];
2422 unlock_res_and_lock(lock);
2425 * We want the ost thread to get this reply so that it can respond
2426 * to ost requests (write cache writeback) that might be triggered
2429 * But we'd also like to be able to indicate in the reply that we're
2430 * cancelling right now, because it's unused, or have an intent result
2431 * in the reply, so we might have to push the responsibility for sending
2432 * the reply down into the AST handlers, alas.
2435 switch (lustre_msg_get_opc(req->rq_reqmsg)) {
2436 case LDLM_BL_CALLBACK:
2437 CDEBUG(D_INODE, "blocking ast\n");
2438 req_capsule_extend(&req->rq_pill, &RQF_LDLM_BL_CALLBACK);
2439 if (!ldlm_is_cancel_on_block(lock)) {
2440 rc = ldlm_callback_reply(req, 0);
2441 if (req->rq_no_reply || rc)
2442 ldlm_callback_errmsg(req, "Normal process", rc,
2443 &dlm_req->lock_handle[0]);
2445 if (ldlm_bl_to_thread_lock(ns, &dlm_req->lock_desc, lock))
2446 ldlm_handle_bl_callback(ns, &dlm_req->lock_desc, lock);
2448 case LDLM_CP_CALLBACK:
2449 CDEBUG(D_INODE, "completion ast\n");
2450 req_capsule_extend(&req->rq_pill, &RQF_LDLM_CP_CALLBACK);
2451 rc = ldlm_handle_cp_callback(req, ns, dlm_req, lock);
2452 if (!OBD_FAIL_CHECK(OBD_FAIL_LDLM_CANCEL_BL_CB_RACE))
2453 ldlm_callback_reply(req, rc);
2455 case LDLM_GL_CALLBACK:
2456 CDEBUG(D_INODE, "glimpse ast\n");
2457 req_capsule_extend(&req->rq_pill, &RQF_LDLM_GL_CALLBACK);
2458 ldlm_handle_gl_callback(req, ns, dlm_req, lock);
2461 LBUG(); /* checked above */
2467 #ifdef HAVE_SERVER_SUPPORT
2469 * Main handler for canceld thread.
2471 * Separated into its own thread to avoid deadlocks.
2473 static int ldlm_cancel_handler(struct ptlrpc_request *req)
2480 * Requests arrive in sender's byte order. The ptlrpc service
2481 * handler has already checked and, if necessary, byte-swapped the
2482 * incoming request message body, but I am responsible for the
2486 req_capsule_init(&req->rq_pill, req, RCL_SERVER);
2488 if (req->rq_export == NULL) {
2489 struct ldlm_request *dlm_req;
2491 CERROR("%s from %s arrived at %llu with bad export cookie %llu\n",
2492 ll_opcode2str(lustre_msg_get_opc(req->rq_reqmsg)),
2493 libcfs_nid2str(req->rq_peer.nid),
2494 (unsigned long long)req->rq_arrival_time.tv_sec,
2495 lustre_msg_get_handle(req->rq_reqmsg)->cookie);
2497 if (lustre_msg_get_opc(req->rq_reqmsg) == LDLM_CANCEL) {
2498 req_capsule_set(&req->rq_pill, &RQF_LDLM_CALLBACK);
2499 dlm_req = req_capsule_client_get(&req->rq_pill,
2501 if (dlm_req != NULL)
2502 ldlm_lock_dump_handle(D_ERROR,
2503 &dlm_req->lock_handle[0]);
2505 ldlm_callback_reply(req, -ENOTCONN);
2509 switch (lustre_msg_get_opc(req->rq_reqmsg)) {
2510 /* XXX FIXME move this back to mds/handler.c, b=249 */
2512 req_capsule_set(&req->rq_pill, &RQF_LDLM_CANCEL);
2513 CDEBUG(D_INODE, "cancel\n");
2514 if (CFS_FAIL_CHECK(OBD_FAIL_LDLM_CANCEL_NET) ||
2515 CFS_FAIL_CHECK(OBD_FAIL_PTLRPC_CANCEL_RESEND) ||
2516 CFS_FAIL_CHECK(OBD_FAIL_LDLM_BL_EVICT))
2518 rc = ldlm_handle_cancel(req);
2522 struct ldlm_request *dlm_req;
2524 req_capsule_set(&req->rq_pill, &RQF_LDLM_CONVERT);
2525 CDEBUG(D_INODE, "convert\n");
2527 dlm_req = req_capsule_client_get(&req->rq_pill, &RMF_DLM_REQ);
2528 if (dlm_req == NULL) {
2529 CDEBUG(D_INFO, "bad request buffer for cancel\n");
2530 rc = ldlm_callback_reply(req, -EPROTO);
2532 req->rq_status = ldlm_handle_convert0(req, dlm_req);
2533 rc = ptlrpc_reply(req);
2538 CERROR("invalid opcode %d\n",
2539 lustre_msg_get_opc(req->rq_reqmsg));
2540 req_capsule_set(&req->rq_pill, &RQF_LDLM_CALLBACK);
2541 rc = ldlm_callback_reply(req, -EINVAL);
2547 static int ldlm_cancel_hpreq_lock_match(struct ptlrpc_request *req,
2548 struct ldlm_lock *lock)
2550 struct ldlm_request *dlm_req;
2551 struct lustre_handle lockh;
2557 dlm_req = req_capsule_client_get(&req->rq_pill, &RMF_DLM_REQ);
2558 if (dlm_req == NULL)
2561 ldlm_lock2handle(lock, &lockh);
2562 for (i = 0; i < dlm_req->lock_count; i++) {
2563 if (lustre_handle_equal(&dlm_req->lock_handle[i],
2565 DEBUG_REQ(D_RPCTRACE, req,
2566 "Prio raised by lock %#llx", lockh.cookie);
2575 static int ldlm_cancel_hpreq_check(struct ptlrpc_request *req)
2577 struct ldlm_request *dlm_req;
2584 /* no prolong in recovery */
2585 if (lustre_msg_get_flags(req->rq_reqmsg) & MSG_REPLAY)
2588 dlm_req = req_capsule_client_get(&req->rq_pill, &RMF_DLM_REQ);
2589 if (dlm_req == NULL)
2592 size = req_capsule_get_size(&req->rq_pill, &RMF_DLM_REQ, RCL_CLIENT);
2593 if (size <= offsetof(struct ldlm_request, lock_handle) ||
2594 (size - offsetof(struct ldlm_request, lock_handle)) /
2595 sizeof(struct lustre_handle) < dlm_req->lock_count)
2598 for (i = 0; i < dlm_req->lock_count; i++) {
2599 struct ldlm_lock *lock;
2601 lock = ldlm_handle2lock(&dlm_req->lock_handle[i]);
2605 rc = ldlm_is_ast_sent(lock) ? 1 : 0;
2607 LDLM_DEBUG(lock, "hpreq cancel/convert lock");
2608 LDLM_LOCK_PUT(lock);
2617 static struct ptlrpc_hpreq_ops ldlm_cancel_hpreq_ops = {
2618 .hpreq_lock_match = ldlm_cancel_hpreq_lock_match,
2619 .hpreq_check = ldlm_cancel_hpreq_check,
2623 static int ldlm_hpreq_handler(struct ptlrpc_request *req)
2627 req_capsule_init(&req->rq_pill, req, RCL_SERVER);
2629 if (req->rq_export == NULL)
2632 if (LDLM_CANCEL == lustre_msg_get_opc(req->rq_reqmsg)) {
2633 req_capsule_set(&req->rq_pill, &RQF_LDLM_CANCEL);
2634 req->rq_ops = &ldlm_cancel_hpreq_ops;
2635 } else if (LDLM_CONVERT == lustre_msg_get_opc(req->rq_reqmsg)) {
2636 req_capsule_set(&req->rq_pill, &RQF_LDLM_CONVERT);
2637 req->rq_ops = &ldlm_cancel_hpreq_ops;
2642 static int ldlm_revoke_lock_cb(struct cfs_hash *hs, struct cfs_hash_bd *bd,
2643 struct hlist_node *hnode, void *data)
2646 struct list_head *rpc_list = data;
2647 struct ldlm_lock *lock = cfs_hash_object(hs, hnode);
2649 lock_res_and_lock(lock);
2651 if (!ldlm_is_granted(lock)) {
2652 unlock_res_and_lock(lock);
2656 LASSERT(lock->l_resource);
2657 if (lock->l_resource->lr_type != LDLM_IBITS &&
2658 lock->l_resource->lr_type != LDLM_PLAIN) {
2659 unlock_res_and_lock(lock);
2663 if (ldlm_is_ast_sent(lock)) {
2664 unlock_res_and_lock(lock);
2668 LASSERT(lock->l_blocking_ast);
2669 LASSERT(!lock->l_blocking_lock);
2671 ldlm_set_ast_sent(lock);
2672 if (lock->l_export && lock->l_export->exp_lock_hash) {
2674 * NB: it's safe to call cfs_hash_del() even lock isn't
2678 * In the function below, .hs_keycmp resolves to
2679 * ldlm_export_lock_keycmp()
2681 /* coverity[overrun-buffer-val] */
2682 cfs_hash_del(lock->l_export->exp_lock_hash,
2683 &lock->l_remote_handle, &lock->l_exp_hash);
2686 list_add_tail(&lock->l_rk_ast, rpc_list);
2687 LDLM_LOCK_GET(lock);
2689 unlock_res_and_lock(lock);
2693 void ldlm_revoke_export_locks(struct obd_export *exp)
2695 LIST_HEAD(rpc_list);
2699 cfs_hash_for_each_nolock(exp->exp_lock_hash,
2700 ldlm_revoke_lock_cb, &rpc_list, 0);
2701 ldlm_run_ast_work(exp->exp_obd->obd_namespace, &rpc_list,
2702 LDLM_WORK_REVOKE_AST);
2706 EXPORT_SYMBOL(ldlm_revoke_export_locks);
2707 #endif /* HAVE_SERVER_SUPPORT */
2709 static int ldlm_bl_get_work(struct ldlm_bl_pool *blp,
2710 struct ldlm_bl_work_item **p_blwi,
2711 struct obd_export **p_exp)
2713 struct ldlm_bl_work_item *blwi = NULL;
2714 static unsigned int num_bl;
2715 static unsigned int num_stale;
2716 int num_th = atomic_read(&blp->blp_num_threads);
2718 *p_exp = obd_stale_export_get();
2720 spin_lock(&blp->blp_lock);
2721 if (*p_exp != NULL) {
2722 if (num_th == 1 || ++num_stale < num_th) {
2723 spin_unlock(&blp->blp_lock);
2729 /* process a request from the blp_list at least every blp_num_threads */
2730 if (!list_empty(&blp->blp_list) &&
2731 (list_empty(&blp->blp_prio_list) || num_bl == 0))
2732 blwi = list_entry(blp->blp_list.next,
2733 struct ldlm_bl_work_item, blwi_entry);
2735 if (!list_empty(&blp->blp_prio_list))
2736 blwi = list_entry(blp->blp_prio_list.next,
2737 struct ldlm_bl_work_item,
2741 if (++num_bl >= num_th)
2743 list_del(&blwi->blwi_entry);
2745 spin_unlock(&blp->blp_lock);
2748 if (*p_exp != NULL && *p_blwi != NULL) {
2749 obd_stale_export_put(*p_exp);
2753 return (*p_blwi != NULL || *p_exp != NULL) ? 1 : 0;
2756 /* This only contains temporary data until the thread starts */
2757 struct ldlm_bl_thread_data {
2758 struct ldlm_bl_pool *bltd_blp;
2759 struct completion bltd_comp;
2763 static int ldlm_bl_thread_main(void *arg);
2765 static int ldlm_bl_thread_start(struct ldlm_bl_pool *blp, bool check_busy)
2767 struct ldlm_bl_thread_data bltd = { .bltd_blp = blp };
2768 struct task_struct *task;
2770 init_completion(&bltd.bltd_comp);
2772 bltd.bltd_num = atomic_inc_return(&blp->blp_num_threads);
2773 if (bltd.bltd_num >= blp->blp_max_threads) {
2774 atomic_dec(&blp->blp_num_threads);
2778 LASSERTF(bltd.bltd_num > 0, "thread num:%d\n", bltd.bltd_num);
2780 atomic_read(&blp->blp_busy_threads) < (bltd.bltd_num - 1)) {
2781 atomic_dec(&blp->blp_num_threads);
2785 task = kthread_run(ldlm_bl_thread_main, &bltd, "ldlm_bl_%02d",
2788 CERROR("cannot start LDLM thread ldlm_bl_%02d: rc %ld\n",
2789 bltd.bltd_num, PTR_ERR(task));
2790 atomic_dec(&blp->blp_num_threads);
2791 return PTR_ERR(task);
2793 wait_for_completion(&bltd.bltd_comp);
2798 /* Not fatal if racy and have a few too many threads */
2799 static int ldlm_bl_thread_need_create(struct ldlm_bl_pool *blp,
2800 struct ldlm_bl_work_item *blwi)
2802 if (atomic_read(&blp->blp_num_threads) >= blp->blp_max_threads)
2805 if (atomic_read(&blp->blp_busy_threads) <
2806 atomic_read(&blp->blp_num_threads))
2809 if (blwi != NULL && (blwi->blwi_ns == NULL ||
2810 blwi->blwi_mem_pressure))
2816 static int ldlm_bl_thread_blwi(struct ldlm_bl_pool *blp,
2817 struct ldlm_bl_work_item *blwi)
2819 /* '1' for consistency with code that checks !mpflag to restore */
2820 unsigned int mpflags = 1;
2824 if (blwi->blwi_ns == NULL)
2825 /* added by ldlm_cleanup() */
2826 RETURN(LDLM_ITER_STOP);
2828 if (blwi->blwi_mem_pressure)
2829 mpflags = memalloc_noreclaim_save();
2831 OBD_FAIL_TIMEOUT(OBD_FAIL_LDLM_PAUSE_CANCEL2, 4);
2833 if (blwi->blwi_count) {
2836 * The special case when we cancel locks in lru
2837 * asynchronously, we pass the list of locks here.
2838 * Thus locks are marked LDLM_FL_CANCELING, but NOT
2839 * canceled locally yet.
2841 count = ldlm_cli_cancel_list_local(&blwi->blwi_head,
2844 ldlm_cli_cancel_list(&blwi->blwi_head, count, NULL,
2846 } else if (blwi->blwi_lock) {
2847 ldlm_handle_bl_callback(blwi->blwi_ns, &blwi->blwi_ld,
2850 ldlm_pool_recalc(&blwi->blwi_ns->ns_pool, true);
2851 spin_lock(&blwi->blwi_ns->ns_lock);
2852 blwi->blwi_ns->ns_rpc_recalc = 0;
2853 spin_unlock(&blwi->blwi_ns->ns_lock);
2854 ldlm_namespace_put(blwi->blwi_ns);
2857 if (blwi->blwi_mem_pressure)
2858 memalloc_noreclaim_restore(mpflags);
2860 if (blwi->blwi_flags & LCF_ASYNC)
2861 OBD_FREE(blwi, sizeof(*blwi));
2863 complete(&blwi->blwi_comp);
2869 * Cancel stale locks on export. Cancel blocked locks first.
2870 * If the given export has blocked locks, the next in the list may have
2871 * them too, thus cancel not blocked locks only if the current export has
2874 static int ldlm_bl_thread_exports(struct ldlm_bl_pool *blp,
2875 struct obd_export *exp)
2881 OBD_FAIL_TIMEOUT(OBD_FAIL_LDLM_BL_EVICT, 4);
2883 num = ldlm_export_cancel_blocked_locks(exp);
2885 ldlm_export_cancel_locks(exp);
2887 obd_stale_export_put(exp);
2894 * Main blocking requests processing thread.
2896 * Callers put locks into its queue by calling ldlm_bl_to_thread.
2897 * This thread in the end ends up doing actual call to ->l_blocking_ast
2900 static int ldlm_bl_thread_main(void *arg)
2903 struct ldlm_bl_pool *blp;
2904 struct ldlm_bl_thread_data *bltd = arg;
2912 rc = lu_env_init(env, LCT_DT_THREAD);
2915 rc = lu_env_add(env);
2917 GOTO(out_env_fini, rc);
2919 blp = bltd->bltd_blp;
2921 complete(&bltd->bltd_comp);
2922 /* cannot use bltd after this, it is only on caller's stack */
2925 struct ldlm_bl_work_item *blwi = NULL;
2926 struct obd_export *exp = NULL;
2929 rc = ldlm_bl_get_work(blp, &blwi, &exp);
2932 wait_event_idle_exclusive(blp->blp_waitq,
2933 ldlm_bl_get_work(blp, &blwi,
2935 atomic_inc(&blp->blp_busy_threads);
2937 if (ldlm_bl_thread_need_create(blp, blwi))
2938 /* discard the return value, we tried */
2939 ldlm_bl_thread_start(blp, true);
2942 rc = ldlm_bl_thread_exports(blp, exp);
2944 rc = ldlm_bl_thread_blwi(blp, blwi);
2946 atomic_dec(&blp->blp_busy_threads);
2948 if (rc == LDLM_ITER_STOP)
2952 * If there are many namespaces, we will not sleep waiting for
2953 * work, and must do a cond_resched to avoid holding the CPU
2959 atomic_dec(&blp->blp_num_threads);
2960 complete(&blp->blp_comp);
2971 static int ldlm_setup(void);
2972 static int ldlm_cleanup(void);
2974 int ldlm_get_ref(void)
2979 mutex_lock(&ldlm_ref_mutex);
2980 if (++ldlm_refcount == 1) {
2985 mutex_unlock(&ldlm_ref_mutex);
2990 void ldlm_put_ref(void)
2993 mutex_lock(&ldlm_ref_mutex);
2994 if (ldlm_refcount == 1) {
2995 int rc = ldlm_cleanup();
2998 CERROR("ldlm_cleanup failed: %d\n", rc);
3004 mutex_unlock(&ldlm_ref_mutex);
3010 * Export handle<->lock hash operations.
3013 ldlm_export_lock_hash(struct cfs_hash *hs, const void *key, unsigned int mask)
3015 return cfs_hash_u64_hash(((struct lustre_handle *)key)->cookie, mask);
3019 ldlm_export_lock_key(struct hlist_node *hnode)
3021 struct ldlm_lock *lock;
3023 lock = hlist_entry(hnode, struct ldlm_lock, l_exp_hash);
3024 return &lock->l_remote_handle;
3028 ldlm_export_lock_keycpy(struct hlist_node *hnode, void *key)
3030 struct ldlm_lock *lock;
3032 lock = hlist_entry(hnode, struct ldlm_lock, l_exp_hash);
3033 lock->l_remote_handle = *(struct lustre_handle *)key;
3037 ldlm_export_lock_keycmp(const void *key, struct hlist_node *hnode)
3039 return lustre_handle_equal(ldlm_export_lock_key(hnode), key);
3043 ldlm_export_lock_object(struct hlist_node *hnode)
3045 return hlist_entry(hnode, struct ldlm_lock, l_exp_hash);
3049 ldlm_export_lock_get(struct cfs_hash *hs, struct hlist_node *hnode)
3051 struct ldlm_lock *lock;
3053 lock = hlist_entry(hnode, struct ldlm_lock, l_exp_hash);
3054 LDLM_LOCK_GET(lock);
3058 ldlm_export_lock_put(struct cfs_hash *hs, struct hlist_node *hnode)
3060 struct ldlm_lock *lock;
3062 lock = hlist_entry(hnode, struct ldlm_lock, l_exp_hash);
3063 LDLM_LOCK_RELEASE(lock);
3066 static struct cfs_hash_ops ldlm_export_lock_ops = {
3067 .hs_hash = ldlm_export_lock_hash,
3068 .hs_key = ldlm_export_lock_key,
3069 .hs_keycmp = ldlm_export_lock_keycmp,
3070 .hs_keycpy = ldlm_export_lock_keycpy,
3071 .hs_object = ldlm_export_lock_object,
3072 .hs_get = ldlm_export_lock_get,
3073 .hs_put = ldlm_export_lock_put,
3074 .hs_put_locked = ldlm_export_lock_put,
3077 int ldlm_init_export(struct obd_export *exp)
3083 exp->exp_lock_hash =
3084 cfs_hash_create(obd_uuid2str(&exp->exp_client_uuid),
3085 HASH_EXP_LOCK_CUR_BITS,
3086 HASH_EXP_LOCK_MAX_BITS,
3087 HASH_EXP_LOCK_BKT_BITS, 0,
3088 CFS_HASH_MIN_THETA, CFS_HASH_MAX_THETA,
3089 &ldlm_export_lock_ops,
3090 CFS_HASH_DEFAULT | CFS_HASH_REHASH_KEY |
3091 CFS_HASH_NBLK_CHANGE);
3093 if (!exp->exp_lock_hash)
3096 rc = ldlm_init_flock_export(exp);
3102 ldlm_destroy_export(exp);
3105 EXPORT_SYMBOL(ldlm_init_export);
3107 void ldlm_destroy_export(struct obd_export *exp)
3110 cfs_hash_putref(exp->exp_lock_hash);
3111 exp->exp_lock_hash = NULL;
3113 ldlm_destroy_flock_export(exp);
3116 EXPORT_SYMBOL(ldlm_destroy_export);
3118 static ssize_t cancel_unused_locks_before_replay_show(struct kobject *kobj,
3119 struct attribute *attr,
3122 return sprintf(buf, "%d\n", ldlm_cancel_unused_locks_before_replay);
3125 static ssize_t cancel_unused_locks_before_replay_store(struct kobject *kobj,
3126 struct attribute *attr,
3133 rc = kstrtoul(buffer, 10, &val);
3137 ldlm_cancel_unused_locks_before_replay = val;
3141 LUSTRE_RW_ATTR(cancel_unused_locks_before_replay);
3143 static struct attribute *ldlm_attrs[] = {
3144 &lustre_attr_cancel_unused_locks_before_replay.attr,
3148 static struct attribute_group ldlm_attr_group = {
3149 .attrs = ldlm_attrs,
3152 static int ldlm_setup(void)
3154 static struct ptlrpc_service_conf conf;
3155 struct ldlm_bl_pool *blp = NULL;
3156 #ifdef HAVE_SERVER_SUPPORT
3157 struct task_struct *task;
3158 #endif /* HAVE_SERVER_SUPPORT */
3164 if (ldlm_state != NULL)
3167 OBD_ALLOC(ldlm_state, sizeof(*ldlm_state));
3168 if (ldlm_state == NULL)
3171 ldlm_kobj = kobject_create_and_add("ldlm", &lustre_kset->kobj);
3175 rc = sysfs_create_group(ldlm_kobj, &ldlm_attr_group);
3179 ldlm_ns_kset = kset_create_and_add("namespaces", NULL, ldlm_kobj);
3183 ldlm_svc_kset = kset_create_and_add("services", NULL, ldlm_kobj);
3187 rc = ldlm_debugfs_setup();
3191 memset(&conf, 0, sizeof(conf));
3192 conf = (typeof(conf)) {
3193 .psc_name = "ldlm_cbd",
3194 .psc_watchdog_factor = 2,
3196 .bc_nbufs = LDLM_CLIENT_NBUFS,
3197 .bc_buf_size = LDLM_BUFSIZE,
3198 .bc_req_max_size = LDLM_MAXREQSIZE,
3199 .bc_rep_max_size = LDLM_MAXREPSIZE,
3200 .bc_req_portal = LDLM_CB_REQUEST_PORTAL,
3201 .bc_rep_portal = LDLM_CB_REPLY_PORTAL,
3204 .tc_thr_name = "ldlm_cb",
3205 .tc_thr_factor = LDLM_THR_FACTOR,
3206 .tc_nthrs_init = LDLM_NTHRS_INIT,
3207 .tc_nthrs_base = LDLM_NTHRS_BASE,
3208 .tc_nthrs_max = LDLM_NTHRS_MAX,
3209 .tc_nthrs_user = ldlm_num_threads,
3210 .tc_cpu_bind = ldlm_cpu_bind,
3211 .tc_ctx_tags = LCT_MD_THREAD | LCT_DT_THREAD,
3214 .cc_pattern = ldlm_cpts,
3215 .cc_affinity = true,
3218 .so_req_handler = ldlm_callback_handler,
3221 ldlm_state->ldlm_cb_service = \
3222 ptlrpc_register_service(&conf, ldlm_svc_kset,
3223 ldlm_svc_debugfs_dir);
3224 if (IS_ERR(ldlm_state->ldlm_cb_service)) {
3225 CERROR("failed to start service\n");
3226 rc = PTR_ERR(ldlm_state->ldlm_cb_service);
3227 ldlm_state->ldlm_cb_service = NULL;
3231 #ifdef HAVE_SERVER_SUPPORT
3232 memset(&conf, 0, sizeof(conf));
3233 conf = (typeof(conf)) {
3234 .psc_name = "ldlm_canceld",
3235 .psc_watchdog_factor = 6,
3237 .bc_nbufs = LDLM_SERVER_NBUFS,
3238 .bc_buf_size = LDLM_BUFSIZE,
3239 .bc_req_max_size = LDLM_MAXREQSIZE,
3240 .bc_rep_max_size = LDLM_MAXREPSIZE,
3241 .bc_req_portal = LDLM_CANCEL_REQUEST_PORTAL,
3242 .bc_rep_portal = LDLM_CANCEL_REPLY_PORTAL,
3246 .tc_thr_name = "ldlm_cn",
3247 .tc_thr_factor = LDLM_THR_FACTOR,
3248 .tc_nthrs_init = LDLM_NTHRS_INIT,
3249 .tc_nthrs_base = LDLM_NTHRS_BASE,
3250 .tc_nthrs_max = LDLM_NTHRS_MAX,
3251 .tc_nthrs_user = ldlm_num_threads,
3252 .tc_cpu_bind = ldlm_cpu_bind,
3253 .tc_ctx_tags = LCT_MD_THREAD | \
3258 .cc_pattern = ldlm_cpts,
3259 .cc_affinity = true,
3262 .so_req_handler = ldlm_cancel_handler,
3263 .so_hpreq_handler = ldlm_hpreq_handler,
3266 ldlm_state->ldlm_cancel_service = \
3267 ptlrpc_register_service(&conf, ldlm_svc_kset,
3268 ldlm_svc_debugfs_dir);
3269 if (IS_ERR(ldlm_state->ldlm_cancel_service)) {
3270 CERROR("failed to start service\n");
3271 rc = PTR_ERR(ldlm_state->ldlm_cancel_service);
3272 ldlm_state->ldlm_cancel_service = NULL;
3275 #endif /* HAVE_SERVER_SUPPORT */
3277 OBD_ALLOC(blp, sizeof(*blp));
3279 GOTO(out, rc = -ENOMEM);
3280 ldlm_state->ldlm_bl_pool = blp;
3282 spin_lock_init(&blp->blp_lock);
3283 INIT_LIST_HEAD(&blp->blp_list);
3284 INIT_LIST_HEAD(&blp->blp_prio_list);
3285 init_waitqueue_head(&blp->blp_waitq);
3286 atomic_set(&blp->blp_num_threads, 0);
3287 atomic_set(&blp->blp_busy_threads, 0);
3289 if (ldlm_num_threads == 0) {
3290 blp->blp_min_threads = LDLM_NTHRS_INIT;
3291 blp->blp_max_threads = LDLM_NTHRS_MAX;
3293 blp->blp_min_threads = blp->blp_max_threads = \
3294 min_t(int, LDLM_NTHRS_MAX, max_t(int, LDLM_NTHRS_INIT,
3298 for (i = 0; i < blp->blp_min_threads; i++) {
3299 rc = ldlm_bl_thread_start(blp, false);
3304 #ifdef HAVE_SERVER_SUPPORT
3305 task = kthread_run(expired_lock_main, NULL, "ldlm_elt");
3308 CERROR("Cannot start ldlm expired-lock thread: %d\n", rc);
3312 wait_event(expired_lock_wait_queue,
3313 expired_lock_thread_state == ELT_READY);
3314 #endif /* HAVE_SERVER_SUPPORT */
3316 rc = ldlm_pools_init();
3318 CERROR("Failed to initialize LDLM pools: %d\n", rc);
3322 rc = ldlm_reclaim_setup();
3324 CERROR("Failed to setup reclaim thread: rc = %d\n", rc);
3334 static int ldlm_cleanup(void)
3338 if (!list_empty(ldlm_namespace_list(LDLM_NAMESPACE_SERVER)) ||
3339 !list_empty(ldlm_namespace_list(LDLM_NAMESPACE_CLIENT))) {
3340 CERROR("ldlm still has namespaces; clean these up first.\n");
3341 ldlm_dump_all_namespaces(LDLM_NAMESPACE_SERVER, D_DLMTRACE);
3342 ldlm_dump_all_namespaces(LDLM_NAMESPACE_CLIENT, D_DLMTRACE);
3346 ldlm_reclaim_cleanup();
3349 if (ldlm_state->ldlm_bl_pool != NULL) {
3350 struct ldlm_bl_pool *blp = ldlm_state->ldlm_bl_pool;
3352 while (atomic_read(&blp->blp_num_threads) > 0) {
3353 struct ldlm_bl_work_item blwi = { .blwi_ns = NULL };
3355 init_completion(&blp->blp_comp);
3357 spin_lock(&blp->blp_lock);
3358 list_add_tail(&blwi.blwi_entry, &blp->blp_list);
3359 wake_up(&blp->blp_waitq);
3360 spin_unlock(&blp->blp_lock);
3362 wait_for_completion(&blp->blp_comp);
3365 OBD_FREE(blp, sizeof(*blp));
3368 if (ldlm_state->ldlm_cb_service != NULL)
3369 ptlrpc_unregister_service(ldlm_state->ldlm_cb_service);
3370 #ifdef HAVE_SERVER_SUPPORT
3371 if (ldlm_state->ldlm_cancel_service != NULL)
3372 ptlrpc_unregister_service(ldlm_state->ldlm_cancel_service);
3376 kset_unregister(ldlm_ns_kset);
3378 kset_unregister(ldlm_svc_kset);
3380 sysfs_remove_group(ldlm_kobj, &ldlm_attr_group);
3381 kobject_put(ldlm_kobj);
3384 ldlm_debugfs_cleanup();
3386 #ifdef HAVE_SERVER_SUPPORT
3387 if (expired_lock_thread_state != ELT_STOPPED) {
3388 expired_lock_thread_state = ELT_TERMINATE;
3389 wake_up(&expired_lock_wait_queue);
3390 wait_event(expired_lock_wait_queue,
3391 expired_lock_thread_state == ELT_STOPPED);
3395 OBD_FREE(ldlm_state, sizeof(*ldlm_state));
3403 ldlm_resource_slab = kmem_cache_create("ldlm_resources",
3404 sizeof(struct ldlm_resource), 0,
3405 SLAB_HWCACHE_ALIGN, NULL);
3406 if (ldlm_resource_slab == NULL)
3409 ldlm_lock_slab = kmem_cache_create("ldlm_locks",
3410 sizeof(struct ldlm_lock), 0,
3411 SLAB_HWCACHE_ALIGN, NULL);
3412 if (ldlm_lock_slab == NULL)
3415 ldlm_interval_slab = kmem_cache_create("interval_node",
3416 sizeof(struct ldlm_interval),
3417 0, SLAB_HWCACHE_ALIGN, NULL);
3418 if (ldlm_interval_slab == NULL)
3421 ldlm_interval_tree_slab = kmem_cache_create("interval_tree",
3422 sizeof(struct ldlm_interval_tree) * LCK_MODE_NUM,
3423 0, SLAB_HWCACHE_ALIGN, NULL);
3424 if (ldlm_interval_tree_slab == NULL)
3427 #ifdef HAVE_SERVER_SUPPORT
3428 ldlm_inodebits_slab = kmem_cache_create("ldlm_ibits_node",
3429 sizeof(struct ldlm_ibits_node),
3430 0, SLAB_HWCACHE_ALIGN, NULL);
3431 if (ldlm_inodebits_slab == NULL)
3432 goto out_interval_tree;
3434 ldlm_glimpse_work_kmem = kmem_cache_create("ldlm_glimpse_work_kmem",
3435 sizeof(struct ldlm_glimpse_work),
3437 if (ldlm_glimpse_work_kmem == NULL)
3441 #if LUSTRE_TRACKS_LOCK_EXP_REFS
3442 class_export_dump_hook = ldlm_dump_export_locks;
3445 #ifdef HAVE_SERVER_SUPPORT
3447 kmem_cache_destroy(ldlm_inodebits_slab);
3449 kmem_cache_destroy(ldlm_interval_tree_slab);
3452 kmem_cache_destroy(ldlm_interval_slab);
3454 kmem_cache_destroy(ldlm_lock_slab);
3456 kmem_cache_destroy(ldlm_resource_slab);
3461 void ldlm_exit(void)
3464 CERROR("ldlm_refcount is %d in ldlm_exit!\n", ldlm_refcount);
3466 kmem_cache_destroy(ldlm_resource_slab);
3468 * ldlm_lock_put() use RCU to call ldlm_lock_free, so need call
3469 * rcu_barrier() to wait all outstanding RCU callbacks to complete,
3470 * so that ldlm_lock_free() get a chance to be called.
3473 kmem_cache_destroy(ldlm_lock_slab);
3474 kmem_cache_destroy(ldlm_interval_slab);
3475 kmem_cache_destroy(ldlm_interval_tree_slab);
3476 #ifdef HAVE_SERVER_SUPPORT
3477 kmem_cache_destroy(ldlm_inodebits_slab);
3478 kmem_cache_destroy(ldlm_glimpse_work_kmem);