1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
2 * vim:expandtab:shiftwidth=8:tabstop=8:
6 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License version 2 only,
10 * as published by the Free Software Foundation.
12 * This program is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * General Public License version 2 for more details (a copy is included
16 * in the LICENSE file that accompanied this code).
18 * You should have received a copy of the GNU General Public License
19 * version 2 along with this program; If not, see
20 * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
22 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
23 * CA 95054 USA or visit www.sun.com if you need additional information or
29 * Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved.
30 * Use is subject to license terms.
32 * Copyright (c) 2011, 2012, Whamcloud, Inc.
35 * This file is part of Lustre, http://www.lustre.org/
36 * Lustre is a trademark of Sun Microsystems, Inc.
38 * Implementation of cl_lock for OSC layer.
40 * Author: Nikita Danilov <nikita.danilov@sun.com>
43 #define DEBUG_SUBSYSTEM S_OSC
46 # include <libcfs/libcfs.h>
48 # include <liblustre.h>
50 /* fid_build_reg_res_name() */
51 #include <lustre_fid.h>
53 #include "osc_cl_internal.h"
59 #define _PAGEREF_MAGIC (-10000000)
61 /*****************************************************************************
67 static const struct cl_lock_operations osc_lock_ops;
68 static const struct cl_lock_operations osc_lock_lockless_ops;
69 static void osc_lock_to_lockless(const struct lu_env *env,
70 struct osc_lock *ols, int force);
71 static int osc_lock_has_pages(struct osc_lock *olck);
73 int osc_lock_is_lockless(const struct osc_lock *olck)
75 return (olck->ols_cl.cls_ops == &osc_lock_lockless_ops);
79 * Returns a weak pointer to the ldlm lock identified by a handle. Returned
80 * pointer cannot be dereferenced, as lock is not protected from concurrent
81 * reclaim. This function is a helper for osc_lock_invariant().
83 static struct ldlm_lock *osc_handle_ptr(struct lustre_handle *handle)
85 struct ldlm_lock *lock;
87 lock = ldlm_handle2lock(handle);
94 * Invariant that has to be true all of the time.
96 static int osc_lock_invariant(struct osc_lock *ols)
98 struct ldlm_lock *lock = osc_handle_ptr(&ols->ols_handle);
99 struct ldlm_lock *olock = ols->ols_lock;
100 int handle_used = lustre_handle_is_used(&ols->ols_handle);
103 ergo(osc_lock_is_lockless(ols),
104 ols->ols_locklessable && ols->ols_lock == NULL) ||
105 (ergo(olock != NULL, handle_used) &&
107 olock->l_handle.h_cookie == ols->ols_handle.cookie) &&
109 * Check that ->ols_handle and ->ols_lock are consistent, but
110 * take into account that they are set at the different time.
113 ergo(lock != NULL && olock != NULL, lock == olock) &&
114 ergo(lock == NULL, olock == NULL)) &&
115 ergo(ols->ols_state == OLS_CANCELLED,
116 olock == NULL && !handle_used) &&
118 * DLM lock is destroyed only after we have seen cancellation
121 ergo(olock != NULL && ols->ols_state < OLS_CANCELLED,
122 !olock->l_destroyed) &&
123 ergo(ols->ols_state == OLS_GRANTED,
125 olock->l_req_mode == olock->l_granted_mode &&
129 /*****************************************************************************
136 * Breaks a link between osc_lock and dlm_lock.
138 static void osc_lock_detach(const struct lu_env *env, struct osc_lock *olck)
140 struct ldlm_lock *dlmlock;
142 cfs_spin_lock(&osc_ast_guard);
143 dlmlock = olck->ols_lock;
144 if (dlmlock == NULL) {
145 cfs_spin_unlock(&osc_ast_guard);
149 olck->ols_lock = NULL;
150 /* wb(); --- for all who checks (ols->ols_lock != NULL) before
151 * call to osc_lock_detach() */
152 dlmlock->l_ast_data = NULL;
153 olck->ols_handle.cookie = 0ULL;
154 cfs_spin_unlock(&osc_ast_guard);
156 lock_res_and_lock(dlmlock);
157 if (dlmlock->l_granted_mode == dlmlock->l_req_mode) {
158 struct cl_object *obj = olck->ols_cl.cls_obj;
159 struct cl_attr *attr = &osc_env_info(env)->oti_attr;
162 cl_object_attr_lock(obj);
163 /* Must get the value under the lock to avoid possible races. */
164 old_kms = cl2osc(obj)->oo_oinfo->loi_kms;
165 /* Update the kms. Need to loop all granted locks.
166 * Not a problem for the client */
167 attr->cat_kms = ldlm_extent_shift_kms(dlmlock, old_kms);
169 cl_object_attr_set(env, obj, attr, CAT_KMS);
170 cl_object_attr_unlock(obj);
172 unlock_res_and_lock(dlmlock);
174 /* release a reference taken in osc_lock_upcall0(). */
175 LASSERT(olck->ols_has_ref);
176 lu_ref_del(&dlmlock->l_reference, "osc_lock", olck);
177 LDLM_LOCK_RELEASE(dlmlock);
178 olck->ols_has_ref = 0;
181 static int osc_lock_unhold(struct osc_lock *ols)
187 result = osc_cancel_base(&ols->ols_handle,
188 ols->ols_einfo.ei_mode);
193 static int osc_lock_unuse(const struct lu_env *env,
194 const struct cl_lock_slice *slice)
196 struct osc_lock *ols = cl2osc_lock(slice);
198 LINVRNT(osc_lock_invariant(ols));
200 switch (ols->ols_state) {
202 LASSERT(!ols->ols_hold);
203 LASSERT(ols->ols_agl);
205 case OLS_UPCALL_RECEIVED:
206 LASSERT(!ols->ols_hold);
207 ols->ols_state = OLS_NEW;
210 LASSERT(!ols->ols_glimpse);
211 LASSERT(ols->ols_hold);
213 * Move lock into OLS_RELEASED state before calling
214 * osc_cancel_base() so that possible synchronous cancellation
215 * (that always happens e.g., for liblustre) sees that lock is
218 ols->ols_state = OLS_RELEASED;
219 return osc_lock_unhold(ols);
221 CERROR("Impossible state: %d\n", ols->ols_state);
226 static void osc_lock_fini(const struct lu_env *env,
227 struct cl_lock_slice *slice)
229 struct osc_lock *ols = cl2osc_lock(slice);
231 LINVRNT(osc_lock_invariant(ols));
233 * ->ols_hold can still be true at this point if, for example, a
234 * thread that requested a lock was killed (and released a reference
235 * to the lock), before reply from a server was received. In this case
236 * lock is destroyed immediately after upcall.
238 osc_lock_unhold(ols);
239 LASSERT(ols->ols_lock == NULL);
240 LASSERT(cfs_atomic_read(&ols->ols_pageref) == 0 ||
241 cfs_atomic_read(&ols->ols_pageref) == _PAGEREF_MAGIC);
243 OBD_SLAB_FREE_PTR(ols, osc_lock_kmem);
246 void osc_lock_build_res(const struct lu_env *env, const struct osc_object *obj,
247 struct ldlm_res_id *resname)
249 const struct lu_fid *fid = lu_object_fid(&obj->oo_cl.co_lu);
252 * In the perfect world of the future, where ost servers talk
255 fid_build_reg_res_name(fid, resname);
258 * In reality, where ost server expects ->lsm_object_id and
259 * ->lsm_object_seq in rename.
261 osc_build_res_name(obj->oo_oinfo->loi_id, obj->oo_oinfo->loi_seq,
266 static void osc_lock_build_policy(const struct lu_env *env,
267 const struct cl_lock *lock,
268 ldlm_policy_data_t *policy)
270 const struct cl_lock_descr *d = &lock->cll_descr;
272 osc_index2policy(policy, d->cld_obj, d->cld_start, d->cld_end);
273 policy->l_extent.gid = d->cld_gid;
276 static int osc_enq2ldlm_flags(__u32 enqflags)
280 LASSERT((enqflags & ~CEF_MASK) == 0);
282 if (enqflags & CEF_NONBLOCK)
283 result |= LDLM_FL_BLOCK_NOWAIT;
284 if (enqflags & CEF_ASYNC)
285 result |= LDLM_FL_HAS_INTENT;
286 if (enqflags & CEF_DISCARD_DATA)
287 result |= LDLM_AST_DISCARD_DATA;
292 * Global spin-lock protecting consistency of ldlm_lock::l_ast_data
293 * pointers. Initialized in osc_init().
295 cfs_spinlock_t osc_ast_guard;
297 static struct osc_lock *osc_ast_data_get(struct ldlm_lock *dlm_lock)
299 struct osc_lock *olck;
301 lock_res_and_lock(dlm_lock);
302 cfs_spin_lock(&osc_ast_guard);
303 olck = dlm_lock->l_ast_data;
305 struct cl_lock *lock = olck->ols_cl.cls_lock;
307 * If osc_lock holds a reference on ldlm lock, return it even
308 * when cl_lock is in CLS_FREEING state. This way
310 * osc_ast_data_get(dlmlock) == NULL
312 * guarantees that all osc references on dlmlock were
313 * released. osc_dlm_blocking_ast0() relies on that.
315 if (lock->cll_state < CLS_FREEING || olck->ols_has_ref) {
316 cl_lock_get_trust(lock);
317 lu_ref_add_atomic(&lock->cll_reference,
318 "ast", cfs_current());
322 cfs_spin_unlock(&osc_ast_guard);
323 unlock_res_and_lock(dlm_lock);
327 static void osc_ast_data_put(const struct lu_env *env, struct osc_lock *olck)
329 struct cl_lock *lock;
331 lock = olck->ols_cl.cls_lock;
332 lu_ref_del(&lock->cll_reference, "ast", cfs_current());
333 cl_lock_put(env, lock);
337 * Updates object attributes from a lock value block (lvb) received together
338 * with the DLM lock reply from the server. Copy of osc_update_enqueue()
341 * This can be optimized to not update attributes when lock is a result of a
344 * Called under lock and resource spin-locks.
346 static void osc_lock_lvb_update(const struct lu_env *env, struct osc_lock *olck,
350 struct cl_object *obj;
351 struct lov_oinfo *oinfo;
352 struct cl_attr *attr;
357 if (!(olck->ols_flags & LDLM_FL_LVB_READY))
360 lvb = &olck->ols_lvb;
361 obj = olck->ols_cl.cls_obj;
362 oinfo = cl2osc(obj)->oo_oinfo;
363 attr = &osc_env_info(env)->oti_attr;
364 valid = CAT_BLOCKS | CAT_ATIME | CAT_CTIME | CAT_MTIME | CAT_SIZE;
365 cl_lvb2attr(attr, lvb);
367 cl_object_attr_lock(obj);
369 struct ldlm_lock *dlmlock;
372 dlmlock = olck->ols_lock;
373 LASSERT(dlmlock != NULL);
375 /* re-grab LVB from a dlm lock under DLM spin-locks. */
376 *lvb = *(struct ost_lvb *)dlmlock->l_lvb_data;
377 size = lvb->lvb_size;
378 /* Extend KMS up to the end of this lock and no further
379 * A lock on [x,y] means a KMS of up to y + 1 bytes! */
380 if (size > dlmlock->l_policy_data.l_extent.end)
381 size = dlmlock->l_policy_data.l_extent.end + 1;
382 if (size >= oinfo->loi_kms) {
383 LDLM_DEBUG(dlmlock, "lock acquired, setting rss="LPU64
384 ", kms="LPU64, lvb->lvb_size, size);
386 attr->cat_kms = size;
388 LDLM_DEBUG(dlmlock, "lock acquired, setting rss="
389 LPU64"; leaving kms="LPU64", end="LPU64,
390 lvb->lvb_size, oinfo->loi_kms,
391 dlmlock->l_policy_data.l_extent.end);
393 ldlm_lock_allow_match_locked(dlmlock);
394 } else if (rc == -ENAVAIL && olck->ols_glimpse) {
395 CDEBUG(D_INODE, "glimpsed, setting rss="LPU64"; leaving"
396 " kms="LPU64"\n", lvb->lvb_size, oinfo->loi_kms);
401 cl_object_attr_set(env, obj, attr, valid);
403 cl_object_attr_unlock(obj);
409 * Called when a lock is granted, from an upcall (when server returned a
410 * granted lock), or from completion AST, when server returned a blocked lock.
412 * Called under lock and resource spin-locks, that are released temporarily
415 static void osc_lock_granted(const struct lu_env *env, struct osc_lock *olck,
416 struct ldlm_lock *dlmlock, int rc)
418 struct ldlm_extent *ext;
419 struct cl_lock *lock;
420 struct cl_lock_descr *descr;
422 LASSERT(dlmlock->l_granted_mode == dlmlock->l_req_mode);
425 if (olck->ols_state < OLS_GRANTED) {
426 lock = olck->ols_cl.cls_lock;
427 ext = &dlmlock->l_policy_data.l_extent;
428 descr = &osc_env_info(env)->oti_descr;
429 descr->cld_obj = lock->cll_descr.cld_obj;
431 /* XXX check that ->l_granted_mode is valid. */
432 descr->cld_mode = osc_ldlm2cl_lock(dlmlock->l_granted_mode);
433 descr->cld_start = cl_index(descr->cld_obj, ext->start);
434 descr->cld_end = cl_index(descr->cld_obj, ext->end);
435 descr->cld_gid = ext->gid;
437 * tell upper layers the extent of the lock that was actually
440 olck->ols_state = OLS_GRANTED;
441 osc_lock_lvb_update(env, olck, rc);
443 /* release DLM spin-locks to allow cl_lock_{modify,signal}()
444 * to take a semaphore on a parent lock. This is safe, because
445 * spin-locks are needed to protect consistency of
446 * dlmlock->l_*_mode and LVB, and we have finished processing
448 unlock_res_and_lock(dlmlock);
449 cl_lock_modify(env, lock, descr);
450 cl_lock_signal(env, lock);
451 LINVRNT(osc_lock_invariant(olck));
452 lock_res_and_lock(dlmlock);
457 static void osc_lock_upcall0(const struct lu_env *env, struct osc_lock *olck)
460 struct ldlm_lock *dlmlock;
464 dlmlock = ldlm_handle2lock_long(&olck->ols_handle, 0);
465 LASSERT(dlmlock != NULL);
467 lock_res_and_lock(dlmlock);
468 cfs_spin_lock(&osc_ast_guard);
469 LASSERT(dlmlock->l_ast_data == olck);
470 LASSERT(olck->ols_lock == NULL);
471 olck->ols_lock = dlmlock;
472 cfs_spin_unlock(&osc_ast_guard);
475 * Lock might be not yet granted. In this case, completion ast
476 * (osc_ldlm_completion_ast()) comes later and finishes lock
479 if (dlmlock->l_granted_mode == dlmlock->l_req_mode)
480 osc_lock_granted(env, olck, dlmlock, 0);
481 unlock_res_and_lock(dlmlock);
484 * osc_enqueue_interpret() decrefs asynchronous locks, counter
487 ldlm_lock_addref(&olck->ols_handle, olck->ols_einfo.ei_mode);
490 /* lock reference taken by ldlm_handle2lock_long() is owned by
491 * osc_lock and released in osc_lock_detach() */
492 lu_ref_add(&dlmlock->l_reference, "osc_lock", olck);
493 olck->ols_has_ref = 1;
497 * Lock upcall function that is executed either when a reply to ENQUEUE rpc is
498 * received from a server, or after osc_enqueue_base() matched a local DLM
501 static int osc_lock_upcall(void *cookie, int errcode)
503 struct osc_lock *olck = cookie;
504 struct cl_lock_slice *slice = &olck->ols_cl;
505 struct cl_lock *lock = slice->cls_lock;
507 struct cl_env_nest nest;
510 env = cl_env_nested_get(&nest);
514 cl_lock_mutex_get(env, lock);
516 LASSERT(lock->cll_state >= CLS_QUEUING);
517 if (olck->ols_state == OLS_ENQUEUED) {
518 olck->ols_state = OLS_UPCALL_RECEIVED;
519 rc = ldlm_error2errno(errcode);
520 } else if (olck->ols_state == OLS_CANCELLED) {
523 CERROR("Impossible state: %d\n", olck->ols_state);
527 struct ldlm_lock *dlmlock;
529 dlmlock = ldlm_handle2lock(&olck->ols_handle);
530 if (dlmlock != NULL) {
531 lock_res_and_lock(dlmlock);
532 cfs_spin_lock(&osc_ast_guard);
533 LASSERT(olck->ols_lock == NULL);
534 dlmlock->l_ast_data = NULL;
535 olck->ols_handle.cookie = 0ULL;
536 cfs_spin_unlock(&osc_ast_guard);
537 ldlm_lock_fail_match_locked(dlmlock);
538 unlock_res_and_lock(dlmlock);
539 LDLM_LOCK_PUT(dlmlock);
542 if (olck->ols_glimpse)
543 olck->ols_glimpse = 0;
544 osc_lock_upcall0(env, olck);
547 /* Error handling, some errors are tolerable. */
548 if (olck->ols_locklessable && rc == -EUSERS) {
549 /* This is a tolerable error, turn this lock into
552 osc_object_set_contended(cl2osc(slice->cls_obj));
553 LASSERT(slice->cls_ops == &osc_lock_ops);
555 /* Change this lock to ldlmlock-less lock. */
556 osc_lock_to_lockless(env, olck, 1);
557 olck->ols_state = OLS_GRANTED;
559 } else if (olck->ols_glimpse && rc == -ENAVAIL) {
560 osc_lock_lvb_update(env, olck, rc);
561 cl_lock_delete(env, lock);
562 /* Hide the error. */
567 /* For AGL case, the RPC sponsor may exits the cl_lock
568 * processing without wait() called before related OSC
569 * lock upcall(). So update the lock status according
570 * to the enqueue result inside AGL upcall(). */
572 lock->cll_flags |= CLF_FROM_UPCALL;
573 cl_wait_try(env, lock);
574 lock->cll_flags &= ~CLF_FROM_UPCALL;
575 if (!olck->ols_glimpse)
578 cl_lock_signal(env, lock);
579 /* del user for lock upcall cookie */
580 cl_unuse_try(env, lock);
582 /* del user for lock upcall cookie */
583 cl_lock_user_del(env, lock);
584 cl_lock_error(env, lock, rc);
587 cl_lock_mutex_put(env, lock);
589 /* release cookie reference, acquired by osc_lock_enqueue() */
590 lu_ref_del(&lock->cll_reference, "upcall", lock);
591 cl_lock_put(env, lock);
593 cl_env_nested_put(&nest, env);
595 /* should never happen, similar to osc_ldlm_blocking_ast(). */
601 * Core of osc_dlm_blocking_ast() logic.
603 static void osc_lock_blocking(const struct lu_env *env,
604 struct ldlm_lock *dlmlock,
605 struct osc_lock *olck, int blocking)
607 struct cl_lock *lock = olck->ols_cl.cls_lock;
609 LASSERT(olck->ols_lock == dlmlock);
610 CLASSERT(OLS_BLOCKED < OLS_CANCELLED);
611 LASSERT(!osc_lock_is_lockless(olck));
614 * Lock might be still addref-ed here, if e.g., blocking ast
615 * is sent for a failed lock.
617 osc_lock_unhold(olck);
619 if (blocking && olck->ols_state < OLS_BLOCKED)
621 * Move osc_lock into OLS_BLOCKED before canceling the lock,
622 * because it recursively re-enters osc_lock_blocking(), with
623 * the state set to OLS_CANCELLED.
625 olck->ols_state = OLS_BLOCKED;
627 * cancel and destroy lock at least once no matter how blocking ast is
628 * entered (see comment above osc_ldlm_blocking_ast() for use
629 * cases). cl_lock_cancel() and cl_lock_delete() are idempotent.
631 cl_lock_cancel(env, lock);
632 cl_lock_delete(env, lock);
636 * Helper for osc_dlm_blocking_ast() handling discrepancies between cl_lock
637 * and ldlm_lock caches.
639 static int osc_dlm_blocking_ast0(const struct lu_env *env,
640 struct ldlm_lock *dlmlock,
641 void *data, int flag)
643 struct osc_lock *olck;
644 struct cl_lock *lock;
648 LASSERT(flag == LDLM_CB_BLOCKING || flag == LDLM_CB_CANCELING);
651 olck = osc_ast_data_get(dlmlock);
653 lock = olck->ols_cl.cls_lock;
654 cl_lock_mutex_get(env, lock);
655 LINVRNT(osc_lock_invariant(olck));
656 if (olck->ols_ast_wait) {
657 /* wake up osc_lock_use() */
658 cl_lock_signal(env, lock);
659 olck->ols_ast_wait = 0;
662 * Lock might have been canceled while this thread was
663 * sleeping for lock mutex, but olck is pinned in memory.
665 if (olck == dlmlock->l_ast_data) {
667 * NOTE: DLM sends blocking AST's for failed locks
668 * (that are still in pre-OLS_GRANTED state)
669 * too, and they have to be canceled otherwise
670 * DLM lock is never destroyed and stuck in
673 * Alternatively, ldlm_cli_cancel() can be
674 * called here directly for osc_locks with
675 * ols_state < OLS_GRANTED to maintain an
676 * invariant that ->clo_cancel() is only called
677 * for locks that were granted.
679 LASSERT(data == olck);
680 osc_lock_blocking(env, dlmlock,
681 olck, flag == LDLM_CB_BLOCKING);
684 cl_lock_mutex_put(env, lock);
685 osc_ast_data_put(env, olck);
688 * DLM lock exists, but there is no cl_lock attached to it.
689 * This is a `normal' race. cl_object and its cl_lock's can be
690 * removed by memory pressure, together with all pages.
692 cancel = (flag == LDLM_CB_BLOCKING);
695 struct lustre_handle *lockh;
697 lockh = &osc_env_info(env)->oti_handle;
698 ldlm_lock2handle(dlmlock, lockh);
699 result = ldlm_cli_cancel(lockh);
706 * Blocking ast invoked by ldlm when dlm lock is either blocking progress of
707 * some other lock, or is canceled. This function is installed as a
708 * ldlm_lock::l_blocking_ast() for client extent locks.
710 * Control flow is tricky, because ldlm uses the same call-back
711 * (ldlm_lock::l_blocking_ast()) for both blocking and cancellation ast's.
713 * \param dlmlock lock for which ast occurred.
715 * \param new description of a conflicting lock in case of blocking ast.
717 * \param data value of dlmlock->l_ast_data
719 * \param flag LDLM_CB_BLOCKING or LDLM_CB_CANCELING. Used to distinguish
720 * cancellation and blocking ast's.
722 * Possible use cases:
724 * - ldlm calls dlmlock->l_blocking_ast(..., LDLM_CB_CANCELING) to cancel
725 * lock due to lock lru pressure, or explicit user request to purge
728 * - ldlm calls dlmlock->l_blocking_ast(..., LDLM_CB_BLOCKING) to notify
729 * us that dlmlock conflicts with another lock that some client is
730 * enqueing. Lock is canceled.
732 * - cl_lock_cancel() is called. osc_lock_cancel() calls
733 * ldlm_cli_cancel() that calls
735 * dlmlock->l_blocking_ast(..., LDLM_CB_CANCELING)
737 * recursively entering osc_ldlm_blocking_ast().
739 * - client cancels lock voluntary (e.g., as a part of early cancellation):
742 * osc_lock_cancel()->
743 * ldlm_cli_cancel()->
744 * dlmlock->l_blocking_ast(..., LDLM_CB_CANCELING)
747 static int osc_ldlm_blocking_ast(struct ldlm_lock *dlmlock,
748 struct ldlm_lock_desc *new, void *data,
752 struct cl_env_nest nest;
756 * This can be called in the context of outer IO, e.g.,
759 * ->osc_enqueue_base()->...
760 * ->ldlm_prep_elc_req()->...
761 * ->ldlm_cancel_callback()->...
762 * ->osc_ldlm_blocking_ast()
764 * new environment has to be created to not corrupt outer context.
766 env = cl_env_nested_get(&nest);
768 result = osc_dlm_blocking_ast0(env, dlmlock, data, flag);
769 cl_env_nested_put(&nest, env);
771 result = PTR_ERR(env);
773 * XXX This should never happen, as cl_lock is
774 * stuck. Pre-allocated environment a la vvp_inode_fini_env
780 if (result == -ENODATA)
783 CERROR("BAST failed: %d\n", result);
788 static int osc_ldlm_completion_ast(struct ldlm_lock *dlmlock,
789 int flags, void *data)
791 struct cl_env_nest nest;
793 struct osc_lock *olck;
794 struct cl_lock *lock;
798 /* first, do dlm part of the work */
799 dlmrc = ldlm_completion_ast_async(dlmlock, flags, data);
800 /* then, notify cl_lock */
801 env = cl_env_nested_get(&nest);
803 olck = osc_ast_data_get(dlmlock);
805 lock = olck->ols_cl.cls_lock;
806 cl_lock_mutex_get(env, lock);
808 * ldlm_handle_cp_callback() copied LVB from request
809 * to lock->l_lvb_data, store it in osc_lock.
811 LASSERT(dlmlock->l_lvb_data != NULL);
812 lock_res_and_lock(dlmlock);
813 olck->ols_lvb = *(struct ost_lvb *)dlmlock->l_lvb_data;
814 if (olck->ols_lock == NULL) {
816 * upcall (osc_lock_upcall()) hasn't yet been
817 * called. Do nothing now, upcall will bind
818 * olck to dlmlock and signal the waiters.
820 * This maintains an invariant that osc_lock
821 * and ldlm_lock are always bound when
822 * osc_lock is in OLS_GRANTED state.
824 } else if (dlmlock->l_granted_mode ==
825 dlmlock->l_req_mode) {
826 osc_lock_granted(env, olck, dlmlock, dlmrc);
828 unlock_res_and_lock(dlmlock);
831 CL_LOCK_DEBUG(D_ERROR, env, lock,
832 "dlmlock returned %d\n", dlmrc);
833 cl_lock_error(env, lock, dlmrc);
835 cl_lock_mutex_put(env, lock);
836 osc_ast_data_put(env, olck);
839 result = -ELDLM_NO_LOCK_DATA;
840 cl_env_nested_put(&nest, env);
842 result = PTR_ERR(env);
843 return dlmrc ?: result;
846 static int osc_ldlm_glimpse_ast(struct ldlm_lock *dlmlock, void *data)
848 struct ptlrpc_request *req = data;
849 struct osc_lock *olck;
850 struct cl_lock *lock;
851 struct cl_object *obj;
852 struct cl_env_nest nest;
855 struct req_capsule *cap;
858 LASSERT(lustre_msg_get_opc(req->rq_reqmsg) == LDLM_GL_CALLBACK);
860 env = cl_env_nested_get(&nest);
863 * osc_ast_data_get() has to go after environment is
864 * allocated, because osc_ast_data() acquires a
865 * reference to a lock, and it can only be released in
868 olck = osc_ast_data_get(dlmlock);
870 lock = olck->ols_cl.cls_lock;
871 cl_lock_mutex_get(env, lock);
873 req_capsule_extend(cap, &RQF_LDLM_GL_CALLBACK);
874 req_capsule_set_size(cap, &RMF_DLM_LVB, RCL_SERVER,
876 result = req_capsule_server_pack(cap);
878 lvb = req_capsule_server_get(cap, &RMF_DLM_LVB);
879 obj = lock->cll_descr.cld_obj;
880 result = cl_object_glimpse(env, obj, lvb);
882 cl_lock_mutex_put(env, lock);
883 osc_ast_data_put(env, olck);
886 * These errors are normal races, so we don't want to
887 * fill the console with messages by calling
890 lustre_pack_reply(req, 1, NULL, NULL);
891 result = -ELDLM_NO_LOCK_DATA;
893 cl_env_nested_put(&nest, env);
895 result = PTR_ERR(env);
896 req->rq_status = result;
900 static unsigned long osc_lock_weigh(const struct lu_env *env,
901 const struct cl_lock_slice *slice)
904 * don't need to grab coh_page_guard since we don't care the exact #
907 return cl_object_header(slice->cls_obj)->coh_pages;
911 * Get the weight of dlm lock for early cancellation.
913 * XXX: it should return the pages covered by this \a dlmlock.
915 static unsigned long osc_ldlm_weigh_ast(struct ldlm_lock *dlmlock)
917 struct cl_env_nest nest;
919 struct osc_lock *lock;
921 unsigned long weight;
926 * osc_ldlm_weigh_ast has a complex context since it might be called
927 * because of lock canceling, or from user's input. We have to make
928 * a new environment for it. Probably it is implementation safe to use
929 * the upper context because cl_lock_put don't modify environment
930 * variables. But in case of ..
932 env = cl_env_nested_get(&nest);
934 /* Mostly because lack of memory, tend to eliminate this lock*/
937 LASSERT(dlmlock->l_resource->lr_type == LDLM_EXTENT);
938 lock = osc_ast_data_get(dlmlock);
940 /* cl_lock was destroyed because of memory pressure.
941 * It is much reasonable to assign this type of lock
944 GOTO(out, weight = 0);
947 cll = lock->ols_cl.cls_lock;
948 cl_lock_mutex_get(env, cll);
949 weight = cl_lock_weigh(env, cll);
950 cl_lock_mutex_put(env, cll);
951 osc_ast_data_put(env, lock);
955 cl_env_nested_put(&nest, env);
959 static void osc_lock_build_einfo(const struct lu_env *env,
960 const struct cl_lock *clock,
961 struct osc_lock *lock,
962 struct ldlm_enqueue_info *einfo)
964 enum cl_lock_mode mode;
966 mode = clock->cll_descr.cld_mode;
967 if (mode == CLM_PHANTOM)
969 * For now, enqueue all glimpse locks in read mode. In the
970 * future, client might choose to enqueue LCK_PW lock for
971 * glimpse on a file opened for write.
975 einfo->ei_type = LDLM_EXTENT;
976 einfo->ei_mode = osc_cl_lock2ldlm(mode);
977 einfo->ei_cb_bl = osc_ldlm_blocking_ast;
978 einfo->ei_cb_cp = osc_ldlm_completion_ast;
979 einfo->ei_cb_gl = osc_ldlm_glimpse_ast;
980 einfo->ei_cb_wg = osc_ldlm_weigh_ast;
981 einfo->ei_cbdata = lock; /* value to be put into ->l_ast_data */
985 * Determine if the lock should be converted into a lockless lock.
988 * - if the lock has an explicite requirment for a non-lockless lock;
989 * - if the io lock request type ci_lockreq;
990 * - send the enqueue rpc to ost to make the further decision;
991 * - special treat to truncate lockless lock
993 * Additional policy can be implemented here, e.g., never do lockless-io
996 static void osc_lock_to_lockless(const struct lu_env *env,
997 struct osc_lock *ols, int force)
999 struct cl_lock_slice *slice = &ols->ols_cl;
1000 struct cl_lock *lock = slice->cls_lock;
1002 LASSERT(ols->ols_state == OLS_NEW ||
1003 ols->ols_state == OLS_UPCALL_RECEIVED);
1006 ols->ols_locklessable = 1;
1007 LASSERT(cl_lock_is_mutexed(lock));
1008 slice->cls_ops = &osc_lock_lockless_ops;
1010 struct osc_io *oio = osc_env_io(env);
1011 struct cl_io *io = oio->oi_cl.cis_io;
1012 struct cl_object *obj = slice->cls_obj;
1013 struct osc_object *oob = cl2osc(obj);
1014 const struct osc_device *osd = lu2osc_dev(obj->co_lu.lo_dev);
1015 struct obd_connect_data *ocd;
1017 LASSERT(io->ci_lockreq == CILR_MANDATORY ||
1018 io->ci_lockreq == CILR_MAYBE ||
1019 io->ci_lockreq == CILR_NEVER);
1021 ocd = &class_exp2cliimp(osc_export(oob))->imp_connect_data;
1022 ols->ols_locklessable = (io->ci_type != CIT_SETATTR) &&
1023 (io->ci_lockreq == CILR_MAYBE) &&
1024 (ocd->ocd_connect_flags & OBD_CONNECT_SRVLOCK);
1025 if (io->ci_lockreq == CILR_NEVER ||
1027 (ols->ols_locklessable && osc_object_is_contended(oob)) ||
1028 /* lockless truncate */
1029 (cl_io_is_trunc(io) &&
1030 (ocd->ocd_connect_flags & OBD_CONNECT_TRUNCLOCK) &&
1031 osd->od_lockless_truncate)) {
1032 ols->ols_locklessable = 1;
1033 slice->cls_ops = &osc_lock_lockless_ops;
1036 LASSERT(ergo(ols->ols_glimpse, !osc_lock_is_lockless(ols)));
1039 static int osc_lock_compatible(const struct osc_lock *qing,
1040 const struct osc_lock *qed)
1042 enum cl_lock_mode qing_mode;
1043 enum cl_lock_mode qed_mode;
1045 qing_mode = qing->ols_cl.cls_lock->cll_descr.cld_mode;
1046 if (qed->ols_glimpse &&
1047 (qed->ols_state >= OLS_UPCALL_RECEIVED || qing_mode == CLM_READ))
1050 qed_mode = qed->ols_cl.cls_lock->cll_descr.cld_mode;
1051 return ((qing_mode == CLM_READ) && (qed_mode == CLM_READ));
1055 * Cancel all conflicting locks and wait for them to be destroyed.
1057 * This function is used for two purposes:
1059 * - early cancel all conflicting locks before starting IO, and
1061 * - guarantee that pages added to the page cache by lockless IO are never
1062 * covered by locks other than lockless IO lock, and, hence, are not
1063 * visible to other threads.
1065 static int osc_lock_enqueue_wait(const struct lu_env *env,
1066 const struct osc_lock *olck)
1068 struct cl_lock *lock = olck->ols_cl.cls_lock;
1069 struct cl_lock_descr *descr = &lock->cll_descr;
1070 struct cl_object_header *hdr = cl_object_header(descr->cld_obj);
1071 struct cl_lock *scan;
1072 struct cl_lock *conflict= NULL;
1073 int lockless = osc_lock_is_lockless(olck);
1077 LASSERT(cl_lock_is_mutexed(lock));
1079 /* make it enqueue anyway for glimpse lock, because we actually
1080 * don't need to cancel any conflicting locks. */
1081 if (olck->ols_glimpse)
1084 cfs_spin_lock(&hdr->coh_lock_guard);
1085 cfs_list_for_each_entry(scan, &hdr->coh_locks, cll_linkage) {
1086 struct cl_lock_descr *cld = &scan->cll_descr;
1087 const struct osc_lock *scan_ols;
1092 if (scan->cll_state < CLS_QUEUING ||
1093 scan->cll_state == CLS_FREEING ||
1094 cld->cld_start > descr->cld_end ||
1095 cld->cld_end < descr->cld_start)
1098 /* overlapped and living locks. */
1100 /* We're not supposed to give up group lock. */
1101 if (scan->cll_descr.cld_mode == CLM_GROUP) {
1102 LASSERT(descr->cld_mode != CLM_GROUP ||
1103 descr->cld_gid != scan->cll_descr.cld_gid);
1107 scan_ols = osc_lock_at(scan);
1109 /* We need to cancel the compatible locks if we're enqueuing
1110 * a lockless lock, for example:
1111 * imagine that client has PR lock on [0, 1000], and thread T0
1112 * is doing lockless IO in [500, 1500] region. Concurrent
1113 * thread T1 can see lockless data in [500, 1000], which is
1114 * wrong, because these data are possibly stale. */
1115 if (!lockless && osc_lock_compatible(olck, scan_ols))
1118 /* Now @scan is conflicting with @lock, this means current
1119 * thread have to sleep for @scan being destroyed. */
1120 if (scan_ols->ols_owner == osc_env_io(env)) {
1121 CERROR("DEADLOCK POSSIBLE!\n");
1122 CL_LOCK_DEBUG(D_ERROR, env, scan, "queued.\n");
1123 CL_LOCK_DEBUG(D_ERROR, env, lock, "queuing.\n");
1124 libcfs_debug_dumpstack(NULL);
1126 cl_lock_get_trust(scan);
1130 cfs_spin_unlock(&hdr->coh_lock_guard);
1133 if (lock->cll_descr.cld_mode == CLM_GROUP) {
1134 /* we want a group lock but a previous lock request
1135 * conflicts, we do not wait but return 0 so the
1136 * request is send to the server
1138 CDEBUG(D_DLMTRACE, "group lock %p is conflicted "
1139 "with %p, no wait, send to server\n",
1141 cl_lock_put(env, conflict);
1144 CDEBUG(D_DLMTRACE, "lock %p is conflicted with %p, "
1147 LASSERT(lock->cll_conflict == NULL);
1148 lu_ref_add(&conflict->cll_reference, "cancel-wait",
1150 lock->cll_conflict = conflict;
1158 * Implementation of cl_lock_operations::clo_enqueue() method for osc
1159 * layer. This initiates ldlm enqueue:
1161 * - cancels conflicting locks early (osc_lock_enqueue_wait());
1163 * - calls osc_enqueue_base() to do actual enqueue.
1165 * osc_enqueue_base() is supplied with an upcall function that is executed
1166 * when lock is received either after a local cached ldlm lock is matched, or
1167 * when a reply from the server is received.
1169 * This function does not wait for the network communication to complete.
1171 static int osc_lock_enqueue(const struct lu_env *env,
1172 const struct cl_lock_slice *slice,
1173 struct cl_io *unused, __u32 enqflags)
1175 struct osc_lock *ols = cl2osc_lock(slice);
1176 struct cl_lock *lock = ols->ols_cl.cls_lock;
1180 LASSERT(cl_lock_is_mutexed(lock));
1181 LASSERTF(ols->ols_state == OLS_NEW,
1182 "Impossible state: %d\n", ols->ols_state);
1184 ols->ols_flags = osc_enq2ldlm_flags(enqflags);
1185 if (enqflags & CEF_AGL) {
1186 ols->ols_flags |= LDLM_FL_BLOCK_NOWAIT;
1189 if (ols->ols_flags & LDLM_FL_HAS_INTENT)
1190 ols->ols_glimpse = 1;
1191 if (!osc_lock_is_lockless(ols) && !(enqflags & CEF_MUST))
1192 /* try to convert this lock to a lockless lock */
1193 osc_lock_to_lockless(env, ols, (enqflags & CEF_NEVER));
1195 result = osc_lock_enqueue_wait(env, ols);
1197 if (!osc_lock_is_lockless(ols)) {
1198 struct osc_object *obj = cl2osc(slice->cls_obj);
1199 struct osc_thread_info *info = osc_env_info(env);
1200 struct ldlm_res_id *resname = &info->oti_resname;
1201 ldlm_policy_data_t *policy = &info->oti_policy;
1202 struct ldlm_enqueue_info *einfo = &ols->ols_einfo;
1204 if (ols->ols_locklessable)
1205 ols->ols_flags |= LDLM_FL_DENY_ON_CONTENTION;
1207 /* a reference for lock, passed as an upcall cookie */
1209 lu_ref_add(&lock->cll_reference, "upcall", lock);
1210 /* a user for lock also */
1211 cl_lock_user_add(env, lock);
1212 ols->ols_state = OLS_ENQUEUED;
1215 * XXX: this is possible blocking point as
1216 * ldlm_lock_match(LDLM_FL_LVB_READY) waits for
1219 osc_lock_build_res(env, obj, resname);
1220 osc_lock_build_policy(env, lock, policy);
1221 result = osc_enqueue_base(osc_export(obj), resname,
1222 &ols->ols_flags, policy,
1224 obj->oo_oinfo->loi_kms_valid,
1226 ols, einfo, &ols->ols_handle,
1227 PTLRPCD_SET, 1, ols->ols_agl);
1229 cl_lock_user_del(env, lock);
1230 lu_ref_del(&lock->cll_reference,
1232 cl_lock_put(env, lock);
1233 if (unlikely(result == -ECANCELED)) {
1234 ols->ols_state = OLS_NEW;
1239 ols->ols_state = OLS_GRANTED;
1240 ols->ols_owner = osc_env_io(env);
1243 LASSERT(ergo(ols->ols_glimpse, !osc_lock_is_lockless(ols)));
1247 static int osc_lock_wait(const struct lu_env *env,
1248 const struct cl_lock_slice *slice)
1250 struct osc_lock *olck = cl2osc_lock(slice);
1251 struct cl_lock *lock = olck->ols_cl.cls_lock;
1253 LINVRNT(osc_lock_invariant(olck));
1255 if (olck->ols_glimpse && olck->ols_state >= OLS_UPCALL_RECEIVED) {
1256 if (olck->ols_flags & LDLM_FL_LVB_READY) {
1258 } else if (olck->ols_agl) {
1259 if (lock->cll_flags & CLF_FROM_UPCALL)
1260 /* It is from enqueue RPC reply upcall for
1261 * updating state. Do not re-enqueue. */
1264 olck->ols_state = OLS_NEW;
1266 LASSERT(lock->cll_error);
1267 return lock->cll_error;
1271 if (olck->ols_state == OLS_NEW) {
1274 LASSERT(olck->ols_agl);
1276 rc = osc_lock_enqueue(env, slice, NULL, CEF_ASYNC | CEF_MUST);
1280 return CLO_REENQUEUED;
1283 LASSERT(equi(olck->ols_state >= OLS_UPCALL_RECEIVED &&
1284 lock->cll_error == 0, olck->ols_lock != NULL));
1286 return lock->cll_error ?: olck->ols_state >= OLS_GRANTED ? 0 : CLO_WAIT;
1290 * An implementation of cl_lock_operations::clo_use() method that pins cached
1293 static int osc_lock_use(const struct lu_env *env,
1294 const struct cl_lock_slice *slice)
1296 struct osc_lock *olck = cl2osc_lock(slice);
1299 LASSERT(!olck->ols_hold);
1302 * Atomically check for LDLM_FL_CBPENDING and addref a lock if this
1303 * flag is not set. This protects us from a concurrent blocking ast.
1305 rc = ldlm_lock_addref_try(&olck->ols_handle, olck->ols_einfo.ei_mode);
1308 olck->ols_state = OLS_GRANTED;
1310 struct cl_lock *lock;
1313 * Lock is being cancelled somewhere within
1314 * ldlm_handle_bl_callback(): LDLM_FL_CBPENDING is already
1315 * set, but osc_ldlm_blocking_ast() hasn't yet acquired
1318 lock = slice->cls_lock;
1319 LASSERT(lock->cll_state == CLS_INTRANSIT);
1320 LASSERT(lock->cll_users > 0);
1321 /* set a flag for osc_dlm_blocking_ast0() to signal the
1323 olck->ols_ast_wait = 1;
1329 static int osc_lock_flush(struct osc_lock *ols, int discard)
1331 struct cl_lock *lock = ols->ols_cl.cls_lock;
1332 struct cl_env_nest nest;
1336 env = cl_env_nested_get(&nest);
1338 result = cl_lock_page_out(env, lock, discard);
1339 cl_env_nested_put(&nest, env);
1341 result = PTR_ERR(env);
1344 LINVRNT(!osc_lock_has_pages(ols));
1350 * Implements cl_lock_operations::clo_cancel() method for osc layer. This is
1351 * called (as part of cl_lock_cancel()) when lock is canceled either voluntary
1352 * (LRU pressure, early cancellation, umount, etc.) or due to the conflict
1353 * with some other lock some where in the cluster. This function does the
1356 * - invalidates all pages protected by this lock (after sending dirty
1357 * ones to the server, as necessary);
1359 * - decref's underlying ldlm lock;
1361 * - cancels ldlm lock (ldlm_cli_cancel()).
1363 static void osc_lock_cancel(const struct lu_env *env,
1364 const struct cl_lock_slice *slice)
1366 struct cl_lock *lock = slice->cls_lock;
1367 struct osc_lock *olck = cl2osc_lock(slice);
1368 struct ldlm_lock *dlmlock = olck->ols_lock;
1372 LASSERT(cl_lock_is_mutexed(lock));
1373 LINVRNT(osc_lock_invariant(olck));
1375 if (dlmlock != NULL) {
1378 discard = !!(dlmlock->l_flags & LDLM_FL_DISCARD_DATA);
1379 result = osc_lock_flush(olck, discard);
1380 osc_lock_unhold(olck);
1382 lock_res_and_lock(dlmlock);
1383 /* Now that we're the only user of dlm read/write reference,
1384 * mostly the ->l_readers + ->l_writers should be zero.
1385 * However, there is a corner case.
1386 * See bug 18829 for details.*/
1387 do_cancel = (dlmlock->l_readers == 0 &&
1388 dlmlock->l_writers == 0);
1389 dlmlock->l_flags |= LDLM_FL_CBPENDING;
1390 unlock_res_and_lock(dlmlock);
1392 result = ldlm_cli_cancel(&olck->ols_handle);
1394 CL_LOCK_DEBUG(D_ERROR, env, lock,
1395 "lock %p cancel failure with error(%d)\n",
1398 olck->ols_state = OLS_CANCELLED;
1399 olck->ols_flags &= ~LDLM_FL_LVB_READY;
1400 osc_lock_detach(env, olck);
1403 #ifdef INVARIANT_CHECK
1404 static int check_cb(const struct lu_env *env, struct cl_io *io,
1405 struct cl_page *page, void *cbdata)
1407 struct cl_lock *lock = cbdata;
1409 if (lock->cll_descr.cld_mode == CLM_READ) {
1410 struct cl_lock *tmp;
1411 tmp = cl_lock_at_page(env, lock->cll_descr.cld_obj,
1414 cl_lock_put(env, tmp);
1415 return CLP_GANG_OKAY;
1419 CL_LOCK_DEBUG(D_ERROR, env, lock, "still has pages\n");
1420 CL_PAGE_DEBUG(D_ERROR, env, page, "\n");
1421 return CLP_GANG_ABORT;
1425 * Returns true iff there are pages under \a olck not protected by other
1428 static int osc_lock_has_pages(struct osc_lock *olck)
1430 struct cl_lock *lock;
1431 struct cl_lock_descr *descr;
1432 struct cl_object *obj;
1433 struct osc_object *oob;
1434 struct cl_env_nest nest;
1439 env = cl_env_nested_get(&nest);
1443 obj = olck->ols_cl.cls_obj;
1445 io = &oob->oo_debug_io;
1446 lock = olck->ols_cl.cls_lock;
1447 descr = &lock->cll_descr;
1449 cfs_mutex_lock(&oob->oo_debug_mutex);
1451 io->ci_obj = cl_object_top(obj);
1452 cl_io_init(env, io, CIT_MISC, io->ci_obj);
1454 result = cl_page_gang_lookup(env, obj, io,
1455 descr->cld_start, descr->cld_end,
1456 check_cb, (void *)lock);
1457 if (result == CLP_GANG_ABORT)
1459 if (result == CLP_GANG_RESCHED)
1461 } while (result != CLP_GANG_OKAY);
1462 cl_io_fini(env, io);
1463 cfs_mutex_unlock(&oob->oo_debug_mutex);
1464 cl_env_nested_put(&nest, env);
1466 return (result == CLP_GANG_ABORT);
1469 static int osc_lock_has_pages(struct osc_lock *olck)
1473 #endif /* INVARIANT_CHECK */
1475 static void osc_lock_delete(const struct lu_env *env,
1476 const struct cl_lock_slice *slice)
1478 struct osc_lock *olck;
1480 olck = cl2osc_lock(slice);
1481 if (olck->ols_glimpse) {
1482 LASSERT(!olck->ols_hold);
1483 LASSERT(!olck->ols_lock);
1487 LINVRNT(osc_lock_invariant(olck));
1488 LINVRNT(!osc_lock_has_pages(olck));
1490 osc_lock_unhold(olck);
1491 osc_lock_detach(env, olck);
1495 * Implements cl_lock_operations::clo_state() method for osc layer.
1497 * Maintains osc_lock::ols_owner field.
1499 * This assumes that lock always enters CLS_HELD (from some other state) in
1500 * the same IO context as one that requested the lock. This should not be a
1501 * problem, because context is by definition shared by all activity pertaining
1502 * to the same high-level IO.
1504 static void osc_lock_state(const struct lu_env *env,
1505 const struct cl_lock_slice *slice,
1506 enum cl_lock_state state)
1508 struct osc_lock *lock = cl2osc_lock(slice);
1511 * XXX multiple io contexts can use the lock at the same time.
1513 LINVRNT(osc_lock_invariant(lock));
1514 if (state == CLS_HELD && slice->cls_lock->cll_state != CLS_HELD) {
1515 struct osc_io *oio = osc_env_io(env);
1517 LASSERT(lock->ols_owner == NULL);
1518 lock->ols_owner = oio;
1519 } else if (state != CLS_HELD)
1520 lock->ols_owner = NULL;
1523 static int osc_lock_print(const struct lu_env *env, void *cookie,
1524 lu_printer_t p, const struct cl_lock_slice *slice)
1526 struct osc_lock *lock = cl2osc_lock(slice);
1529 * XXX print ldlm lock and einfo properly.
1531 (*p)(env, cookie, "%p %08x "LPX64" %d %p ",
1532 lock->ols_lock, lock->ols_flags, lock->ols_handle.cookie,
1533 lock->ols_state, lock->ols_owner);
1534 osc_lvb_print(env, cookie, p, &lock->ols_lvb);
1538 static int osc_lock_fits_into(const struct lu_env *env,
1539 const struct cl_lock_slice *slice,
1540 const struct cl_lock_descr *need,
1541 const struct cl_io *io)
1543 struct osc_lock *ols = cl2osc_lock(slice);
1545 if (need->cld_enq_flags & CEF_NEVER)
1548 if (need->cld_mode == CLM_PHANTOM) {
1550 return !(ols->ols_state > OLS_RELEASED);
1553 * Note: the QUEUED lock can't be matched here, otherwise
1554 * it might cause the deadlocks.
1556 * P1: enqueued read lock, create sublock1
1557 * P2: enqueued write lock, create sublock2(conflicted
1559 * P1: Grant read lock.
1560 * P1: enqueued glimpse lock(with holding sublock1_read),
1561 * matched with sublock2, waiting sublock2 to be granted.
1562 * But sublock2 can not be granted, because P1
1563 * will not release sublock1. Bang!
1565 if (ols->ols_state < OLS_GRANTED ||
1566 ols->ols_state > OLS_RELEASED)
1568 } else if (need->cld_enq_flags & CEF_MUST) {
1570 * If the lock hasn't ever enqueued, it can't be matched
1571 * because enqueue process brings in many information
1572 * which can be used to determine things such as lockless,
1575 if (ols->ols_state < OLS_UPCALL_RECEIVED &&
1576 ols->ols_locklessable)
1582 static const struct cl_lock_operations osc_lock_ops = {
1583 .clo_fini = osc_lock_fini,
1584 .clo_enqueue = osc_lock_enqueue,
1585 .clo_wait = osc_lock_wait,
1586 .clo_unuse = osc_lock_unuse,
1587 .clo_use = osc_lock_use,
1588 .clo_delete = osc_lock_delete,
1589 .clo_state = osc_lock_state,
1590 .clo_cancel = osc_lock_cancel,
1591 .clo_weigh = osc_lock_weigh,
1592 .clo_print = osc_lock_print,
1593 .clo_fits_into = osc_lock_fits_into,
1596 static int osc_lock_lockless_unuse(const struct lu_env *env,
1597 const struct cl_lock_slice *slice)
1599 struct osc_lock *ols = cl2osc_lock(slice);
1600 struct cl_lock *lock = slice->cls_lock;
1602 LASSERT(ols->ols_state == OLS_GRANTED);
1603 LINVRNT(osc_lock_invariant(ols));
1605 cl_lock_cancel(env, lock);
1606 cl_lock_delete(env, lock);
1610 static void osc_lock_lockless_cancel(const struct lu_env *env,
1611 const struct cl_lock_slice *slice)
1613 struct osc_lock *ols = cl2osc_lock(slice);
1616 result = osc_lock_flush(ols, 0);
1618 CERROR("Pages for lockless lock %p were not purged(%d)\n",
1620 ols->ols_state = OLS_CANCELLED;
1623 static int osc_lock_lockless_wait(const struct lu_env *env,
1624 const struct cl_lock_slice *slice)
1626 struct osc_lock *olck = cl2osc_lock(slice);
1627 struct cl_lock *lock = olck->ols_cl.cls_lock;
1629 LINVRNT(osc_lock_invariant(olck));
1630 LASSERT(olck->ols_state >= OLS_UPCALL_RECEIVED);
1632 return lock->cll_error;
1635 static void osc_lock_lockless_state(const struct lu_env *env,
1636 const struct cl_lock_slice *slice,
1637 enum cl_lock_state state)
1639 struct osc_lock *lock = cl2osc_lock(slice);
1641 LINVRNT(osc_lock_invariant(lock));
1642 if (state == CLS_HELD) {
1643 struct osc_io *oio = osc_env_io(env);
1645 LASSERT(ergo(lock->ols_owner, lock->ols_owner == oio));
1646 lock->ols_owner = oio;
1648 /* set the io to be lockless if this lock is for io's
1650 if (cl_object_same(oio->oi_cl.cis_obj, slice->cls_obj))
1651 oio->oi_lockless = 1;
1655 static int osc_lock_lockless_fits_into(const struct lu_env *env,
1656 const struct cl_lock_slice *slice,
1657 const struct cl_lock_descr *need,
1658 const struct cl_io *io)
1660 struct osc_lock *lock = cl2osc_lock(slice);
1662 if (!(need->cld_enq_flags & CEF_NEVER))
1665 /* lockless lock should only be used by its owning io. b22147 */
1666 return (lock->ols_owner == osc_env_io(env));
1669 static const struct cl_lock_operations osc_lock_lockless_ops = {
1670 .clo_fini = osc_lock_fini,
1671 .clo_enqueue = osc_lock_enqueue,
1672 .clo_wait = osc_lock_lockless_wait,
1673 .clo_unuse = osc_lock_lockless_unuse,
1674 .clo_state = osc_lock_lockless_state,
1675 .clo_fits_into = osc_lock_lockless_fits_into,
1676 .clo_cancel = osc_lock_lockless_cancel,
1677 .clo_print = osc_lock_print
1680 int osc_lock_init(const struct lu_env *env,
1681 struct cl_object *obj, struct cl_lock *lock,
1682 const struct cl_io *unused)
1684 struct osc_lock *clk;
1687 OBD_SLAB_ALLOC_PTR_GFP(clk, osc_lock_kmem, CFS_ALLOC_IO);
1689 osc_lock_build_einfo(env, lock, clk, &clk->ols_einfo);
1690 cfs_atomic_set(&clk->ols_pageref, 0);
1691 clk->ols_state = OLS_NEW;
1692 cl_lock_slice_add(lock, &clk->ols_cl, obj, &osc_lock_ops);
1699 int osc_dlm_lock_pageref(struct ldlm_lock *dlm)
1701 struct osc_lock *olock;
1704 cfs_spin_lock(&osc_ast_guard);
1705 olock = dlm->l_ast_data;
1707 * there's a very rare race with osc_page_addref_lock(), but that
1708 * doesn't matter because in the worst case we don't cancel a lock
1709 * which we actually can, that's no harm.
1711 if (olock != NULL &&
1712 cfs_atomic_add_return(_PAGEREF_MAGIC,
1713 &olock->ols_pageref) != _PAGEREF_MAGIC) {
1714 cfs_atomic_sub(_PAGEREF_MAGIC, &olock->ols_pageref);
1717 cfs_spin_unlock(&osc_ast_guard);