Loading...
1// SPDX-License-Identifier: GPL-2.0-only
2/******************************************************************************
3*******************************************************************************
4**
5** Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
6** Copyright (C) 2004-2011 Red Hat, Inc. All rights reserved.
7**
8**
9*******************************************************************************
10******************************************************************************/
11
12#include <linux/module.h>
13
14#include "dlm_internal.h"
15#include "lockspace.h"
16#include "member.h"
17#include "recoverd.h"
18#include "dir.h"
19#include "midcomms.h"
20#include "lowcomms.h"
21#include "config.h"
22#include "memory.h"
23#include "lock.h"
24#include "recover.h"
25#include "requestqueue.h"
26#include "user.h"
27#include "ast.h"
28
29static int ls_count;
30static struct mutex ls_lock;
31static struct list_head lslist;
32static spinlock_t lslist_lock;
33static struct task_struct * scand_task;
34
35
36static ssize_t dlm_control_store(struct dlm_ls *ls, const char *buf, size_t len)
37{
38 ssize_t ret = len;
39 int n;
40 int rc = kstrtoint(buf, 0, &n);
41
42 if (rc)
43 return rc;
44 ls = dlm_find_lockspace_local(ls->ls_local_handle);
45 if (!ls)
46 return -EINVAL;
47
48 switch (n) {
49 case 0:
50 dlm_ls_stop(ls);
51 break;
52 case 1:
53 dlm_ls_start(ls);
54 break;
55 default:
56 ret = -EINVAL;
57 }
58 dlm_put_lockspace(ls);
59 return ret;
60}
61
62static ssize_t dlm_event_store(struct dlm_ls *ls, const char *buf, size_t len)
63{
64 int rc = kstrtoint(buf, 0, &ls->ls_uevent_result);
65
66 if (rc)
67 return rc;
68 set_bit(LSFL_UEVENT_WAIT, &ls->ls_flags);
69 wake_up(&ls->ls_uevent_wait);
70 return len;
71}
72
73static ssize_t dlm_id_show(struct dlm_ls *ls, char *buf)
74{
75 return snprintf(buf, PAGE_SIZE, "%u\n", ls->ls_global_id);
76}
77
78static ssize_t dlm_id_store(struct dlm_ls *ls, const char *buf, size_t len)
79{
80 int rc = kstrtouint(buf, 0, &ls->ls_global_id);
81
82 if (rc)
83 return rc;
84 return len;
85}
86
87static ssize_t dlm_nodir_show(struct dlm_ls *ls, char *buf)
88{
89 return snprintf(buf, PAGE_SIZE, "%u\n", dlm_no_directory(ls));
90}
91
92static ssize_t dlm_nodir_store(struct dlm_ls *ls, const char *buf, size_t len)
93{
94 int val;
95 int rc = kstrtoint(buf, 0, &val);
96
97 if (rc)
98 return rc;
99 if (val == 1)
100 set_bit(LSFL_NODIR, &ls->ls_flags);
101 return len;
102}
103
104static ssize_t dlm_recover_status_show(struct dlm_ls *ls, char *buf)
105{
106 uint32_t status = dlm_recover_status(ls);
107 return snprintf(buf, PAGE_SIZE, "%x\n", status);
108}
109
110static ssize_t dlm_recover_nodeid_show(struct dlm_ls *ls, char *buf)
111{
112 return snprintf(buf, PAGE_SIZE, "%d\n", ls->ls_recover_nodeid);
113}
114
115struct dlm_attr {
116 struct attribute attr;
117 ssize_t (*show)(struct dlm_ls *, char *);
118 ssize_t (*store)(struct dlm_ls *, const char *, size_t);
119};
120
121static struct dlm_attr dlm_attr_control = {
122 .attr = {.name = "control", .mode = S_IWUSR},
123 .store = dlm_control_store
124};
125
126static struct dlm_attr dlm_attr_event = {
127 .attr = {.name = "event_done", .mode = S_IWUSR},
128 .store = dlm_event_store
129};
130
131static struct dlm_attr dlm_attr_id = {
132 .attr = {.name = "id", .mode = S_IRUGO | S_IWUSR},
133 .show = dlm_id_show,
134 .store = dlm_id_store
135};
136
137static struct dlm_attr dlm_attr_nodir = {
138 .attr = {.name = "nodir", .mode = S_IRUGO | S_IWUSR},
139 .show = dlm_nodir_show,
140 .store = dlm_nodir_store
141};
142
143static struct dlm_attr dlm_attr_recover_status = {
144 .attr = {.name = "recover_status", .mode = S_IRUGO},
145 .show = dlm_recover_status_show
146};
147
148static struct dlm_attr dlm_attr_recover_nodeid = {
149 .attr = {.name = "recover_nodeid", .mode = S_IRUGO},
150 .show = dlm_recover_nodeid_show
151};
152
153static struct attribute *dlm_attrs[] = {
154 &dlm_attr_control.attr,
155 &dlm_attr_event.attr,
156 &dlm_attr_id.attr,
157 &dlm_attr_nodir.attr,
158 &dlm_attr_recover_status.attr,
159 &dlm_attr_recover_nodeid.attr,
160 NULL,
161};
162ATTRIBUTE_GROUPS(dlm);
163
164static ssize_t dlm_attr_show(struct kobject *kobj, struct attribute *attr,
165 char *buf)
166{
167 struct dlm_ls *ls = container_of(kobj, struct dlm_ls, ls_kobj);
168 struct dlm_attr *a = container_of(attr, struct dlm_attr, attr);
169 return a->show ? a->show(ls, buf) : 0;
170}
171
172static ssize_t dlm_attr_store(struct kobject *kobj, struct attribute *attr,
173 const char *buf, size_t len)
174{
175 struct dlm_ls *ls = container_of(kobj, struct dlm_ls, ls_kobj);
176 struct dlm_attr *a = container_of(attr, struct dlm_attr, attr);
177 return a->store ? a->store(ls, buf, len) : len;
178}
179
180static void lockspace_kobj_release(struct kobject *k)
181{
182 struct dlm_ls *ls = container_of(k, struct dlm_ls, ls_kobj);
183 kfree(ls);
184}
185
186static const struct sysfs_ops dlm_attr_ops = {
187 .show = dlm_attr_show,
188 .store = dlm_attr_store,
189};
190
191static struct kobj_type dlm_ktype = {
192 .default_groups = dlm_groups,
193 .sysfs_ops = &dlm_attr_ops,
194 .release = lockspace_kobj_release,
195};
196
197static struct kset *dlm_kset;
198
199static int do_uevent(struct dlm_ls *ls, int in)
200{
201 if (in)
202 kobject_uevent(&ls->ls_kobj, KOBJ_ONLINE);
203 else
204 kobject_uevent(&ls->ls_kobj, KOBJ_OFFLINE);
205
206 log_rinfo(ls, "%s the lockspace group...", in ? "joining" : "leaving");
207
208 /* dlm_controld will see the uevent, do the necessary group management
209 and then write to sysfs to wake us */
210
211 wait_event(ls->ls_uevent_wait,
212 test_and_clear_bit(LSFL_UEVENT_WAIT, &ls->ls_flags));
213
214 log_rinfo(ls, "group event done %d", ls->ls_uevent_result);
215
216 return ls->ls_uevent_result;
217}
218
219static int dlm_uevent(struct kset *kset, struct kobject *kobj,
220 struct kobj_uevent_env *env)
221{
222 struct dlm_ls *ls = container_of(kobj, struct dlm_ls, ls_kobj);
223
224 add_uevent_var(env, "LOCKSPACE=%s", ls->ls_name);
225 return 0;
226}
227
228static const struct kset_uevent_ops dlm_uevent_ops = {
229 .uevent = dlm_uevent,
230};
231
232int __init dlm_lockspace_init(void)
233{
234 ls_count = 0;
235 mutex_init(&ls_lock);
236 INIT_LIST_HEAD(&lslist);
237 spin_lock_init(&lslist_lock);
238
239 dlm_kset = kset_create_and_add("dlm", &dlm_uevent_ops, kernel_kobj);
240 if (!dlm_kset) {
241 printk(KERN_WARNING "%s: can not create kset\n", __func__);
242 return -ENOMEM;
243 }
244 return 0;
245}
246
247void dlm_lockspace_exit(void)
248{
249 kset_unregister(dlm_kset);
250}
251
252static struct dlm_ls *find_ls_to_scan(void)
253{
254 struct dlm_ls *ls;
255
256 spin_lock(&lslist_lock);
257 list_for_each_entry(ls, &lslist, ls_list) {
258 if (time_after_eq(jiffies, ls->ls_scan_time +
259 dlm_config.ci_scan_secs * HZ)) {
260 spin_unlock(&lslist_lock);
261 return ls;
262 }
263 }
264 spin_unlock(&lslist_lock);
265 return NULL;
266}
267
268static int dlm_scand(void *data)
269{
270 struct dlm_ls *ls;
271
272 while (!kthread_should_stop()) {
273 ls = find_ls_to_scan();
274 if (ls) {
275 if (dlm_lock_recovery_try(ls)) {
276 ls->ls_scan_time = jiffies;
277 dlm_scan_rsbs(ls);
278 dlm_scan_timeout(ls);
279 dlm_scan_waiters(ls);
280 dlm_unlock_recovery(ls);
281 } else {
282 ls->ls_scan_time += HZ;
283 }
284 continue;
285 }
286 schedule_timeout_interruptible(dlm_config.ci_scan_secs * HZ);
287 }
288 return 0;
289}
290
291static int dlm_scand_start(void)
292{
293 struct task_struct *p;
294 int error = 0;
295
296 p = kthread_run(dlm_scand, NULL, "dlm_scand");
297 if (IS_ERR(p))
298 error = PTR_ERR(p);
299 else
300 scand_task = p;
301 return error;
302}
303
304static void dlm_scand_stop(void)
305{
306 kthread_stop(scand_task);
307}
308
309struct dlm_ls *dlm_find_lockspace_global(uint32_t id)
310{
311 struct dlm_ls *ls;
312
313 spin_lock(&lslist_lock);
314
315 list_for_each_entry(ls, &lslist, ls_list) {
316 if (ls->ls_global_id == id) {
317 ls->ls_count++;
318 goto out;
319 }
320 }
321 ls = NULL;
322 out:
323 spin_unlock(&lslist_lock);
324 return ls;
325}
326
327struct dlm_ls *dlm_find_lockspace_local(dlm_lockspace_t *lockspace)
328{
329 struct dlm_ls *ls;
330
331 spin_lock(&lslist_lock);
332 list_for_each_entry(ls, &lslist, ls_list) {
333 if (ls->ls_local_handle == lockspace) {
334 ls->ls_count++;
335 goto out;
336 }
337 }
338 ls = NULL;
339 out:
340 spin_unlock(&lslist_lock);
341 return ls;
342}
343
344struct dlm_ls *dlm_find_lockspace_device(int minor)
345{
346 struct dlm_ls *ls;
347
348 spin_lock(&lslist_lock);
349 list_for_each_entry(ls, &lslist, ls_list) {
350 if (ls->ls_device.minor == minor) {
351 ls->ls_count++;
352 goto out;
353 }
354 }
355 ls = NULL;
356 out:
357 spin_unlock(&lslist_lock);
358 return ls;
359}
360
361void dlm_put_lockspace(struct dlm_ls *ls)
362{
363 spin_lock(&lslist_lock);
364 ls->ls_count--;
365 spin_unlock(&lslist_lock);
366}
367
368static void remove_lockspace(struct dlm_ls *ls)
369{
370 for (;;) {
371 spin_lock(&lslist_lock);
372 if (ls->ls_count == 0) {
373 WARN_ON(ls->ls_create_count != 0);
374 list_del(&ls->ls_list);
375 spin_unlock(&lslist_lock);
376 return;
377 }
378 spin_unlock(&lslist_lock);
379 ssleep(1);
380 }
381}
382
383static int threads_start(void)
384{
385 int error;
386
387 error = dlm_scand_start();
388 if (error) {
389 log_print("cannot start dlm_scand thread %d", error);
390 goto fail;
391 }
392
393 /* Thread for sending/receiving messages for all lockspace's */
394 error = dlm_midcomms_start();
395 if (error) {
396 log_print("cannot start dlm lowcomms %d", error);
397 goto scand_fail;
398 }
399
400 return 0;
401
402 scand_fail:
403 dlm_scand_stop();
404 fail:
405 return error;
406}
407
408static int new_lockspace(const char *name, const char *cluster,
409 uint32_t flags, int lvblen,
410 const struct dlm_lockspace_ops *ops, void *ops_arg,
411 int *ops_result, dlm_lockspace_t **lockspace)
412{
413 struct dlm_ls *ls;
414 int i, size, error;
415 int do_unreg = 0;
416 int namelen = strlen(name);
417
418 if (namelen > DLM_LOCKSPACE_LEN || namelen == 0)
419 return -EINVAL;
420
421 if (!lvblen || (lvblen % 8))
422 return -EINVAL;
423
424 if (!try_module_get(THIS_MODULE))
425 return -EINVAL;
426
427 if (!dlm_user_daemon_available()) {
428 log_print("dlm user daemon not available");
429 error = -EUNATCH;
430 goto out;
431 }
432
433 if (ops && ops_result) {
434 if (!dlm_config.ci_recover_callbacks)
435 *ops_result = -EOPNOTSUPP;
436 else
437 *ops_result = 0;
438 }
439
440 if (!cluster)
441 log_print("dlm cluster name '%s' is being used without an application provided cluster name",
442 dlm_config.ci_cluster_name);
443
444 if (dlm_config.ci_recover_callbacks && cluster &&
445 strncmp(cluster, dlm_config.ci_cluster_name, DLM_LOCKSPACE_LEN)) {
446 log_print("dlm cluster name '%s' does not match "
447 "the application cluster name '%s'",
448 dlm_config.ci_cluster_name, cluster);
449 error = -EBADR;
450 goto out;
451 }
452
453 error = 0;
454
455 spin_lock(&lslist_lock);
456 list_for_each_entry(ls, &lslist, ls_list) {
457 WARN_ON(ls->ls_create_count <= 0);
458 if (ls->ls_namelen != namelen)
459 continue;
460 if (memcmp(ls->ls_name, name, namelen))
461 continue;
462 if (flags & DLM_LSFL_NEWEXCL) {
463 error = -EEXIST;
464 break;
465 }
466 ls->ls_create_count++;
467 *lockspace = ls;
468 error = 1;
469 break;
470 }
471 spin_unlock(&lslist_lock);
472
473 if (error)
474 goto out;
475
476 error = -ENOMEM;
477
478 ls = kzalloc(sizeof(struct dlm_ls) + namelen, GFP_NOFS);
479 if (!ls)
480 goto out;
481 memcpy(ls->ls_name, name, namelen);
482 ls->ls_namelen = namelen;
483 ls->ls_lvblen = lvblen;
484 ls->ls_count = 0;
485 ls->ls_flags = 0;
486 ls->ls_scan_time = jiffies;
487
488 if (ops && dlm_config.ci_recover_callbacks) {
489 ls->ls_ops = ops;
490 ls->ls_ops_arg = ops_arg;
491 }
492
493 if (flags & DLM_LSFL_TIMEWARN)
494 set_bit(LSFL_TIMEWARN, &ls->ls_flags);
495
496 /* ls_exflags are forced to match among nodes, and we don't
497 need to require all nodes to have some flags set */
498 ls->ls_exflags = (flags & ~(DLM_LSFL_TIMEWARN | DLM_LSFL_FS |
499 DLM_LSFL_NEWEXCL));
500
501 size = dlm_config.ci_rsbtbl_size;
502 ls->ls_rsbtbl_size = size;
503
504 ls->ls_rsbtbl = vmalloc(array_size(size, sizeof(struct dlm_rsbtable)));
505 if (!ls->ls_rsbtbl)
506 goto out_lsfree;
507 for (i = 0; i < size; i++) {
508 ls->ls_rsbtbl[i].keep.rb_node = NULL;
509 ls->ls_rsbtbl[i].toss.rb_node = NULL;
510 spin_lock_init(&ls->ls_rsbtbl[i].lock);
511 }
512
513 spin_lock_init(&ls->ls_remove_spin);
514
515 for (i = 0; i < DLM_REMOVE_NAMES_MAX; i++) {
516 ls->ls_remove_names[i] = kzalloc(DLM_RESNAME_MAXLEN+1,
517 GFP_KERNEL);
518 if (!ls->ls_remove_names[i])
519 goto out_rsbtbl;
520 }
521
522 idr_init(&ls->ls_lkbidr);
523 spin_lock_init(&ls->ls_lkbidr_spin);
524
525 INIT_LIST_HEAD(&ls->ls_waiters);
526 mutex_init(&ls->ls_waiters_mutex);
527 INIT_LIST_HEAD(&ls->ls_orphans);
528 mutex_init(&ls->ls_orphans_mutex);
529 INIT_LIST_HEAD(&ls->ls_timeout);
530 mutex_init(&ls->ls_timeout_mutex);
531
532 INIT_LIST_HEAD(&ls->ls_new_rsb);
533 spin_lock_init(&ls->ls_new_rsb_spin);
534
535 INIT_LIST_HEAD(&ls->ls_nodes);
536 INIT_LIST_HEAD(&ls->ls_nodes_gone);
537 ls->ls_num_nodes = 0;
538 ls->ls_low_nodeid = 0;
539 ls->ls_total_weight = 0;
540 ls->ls_node_array = NULL;
541
542 memset(&ls->ls_stub_rsb, 0, sizeof(struct dlm_rsb));
543 ls->ls_stub_rsb.res_ls = ls;
544
545 ls->ls_debug_rsb_dentry = NULL;
546 ls->ls_debug_waiters_dentry = NULL;
547
548 init_waitqueue_head(&ls->ls_uevent_wait);
549 ls->ls_uevent_result = 0;
550 init_completion(&ls->ls_members_done);
551 ls->ls_members_result = -1;
552
553 mutex_init(&ls->ls_cb_mutex);
554 INIT_LIST_HEAD(&ls->ls_cb_delay);
555
556 ls->ls_recoverd_task = NULL;
557 mutex_init(&ls->ls_recoverd_active);
558 spin_lock_init(&ls->ls_recover_lock);
559 spin_lock_init(&ls->ls_rcom_spin);
560 get_random_bytes(&ls->ls_rcom_seq, sizeof(uint64_t));
561 ls->ls_recover_status = 0;
562 ls->ls_recover_seq = 0;
563 ls->ls_recover_args = NULL;
564 init_rwsem(&ls->ls_in_recovery);
565 init_rwsem(&ls->ls_recv_active);
566 INIT_LIST_HEAD(&ls->ls_requestqueue);
567 mutex_init(&ls->ls_requestqueue_mutex);
568 mutex_init(&ls->ls_clear_proc_locks);
569
570 /* Due backwards compatibility with 3.1 we need to use maximum
571 * possible dlm message size to be sure the message will fit and
572 * not having out of bounds issues. However on sending side 3.2
573 * might send less.
574 */
575 ls->ls_recover_buf = kmalloc(DLM_MAX_SOCKET_BUFSIZE, GFP_NOFS);
576 if (!ls->ls_recover_buf)
577 goto out_lkbidr;
578
579 ls->ls_slot = 0;
580 ls->ls_num_slots = 0;
581 ls->ls_slots_size = 0;
582 ls->ls_slots = NULL;
583
584 INIT_LIST_HEAD(&ls->ls_recover_list);
585 spin_lock_init(&ls->ls_recover_list_lock);
586 idr_init(&ls->ls_recover_idr);
587 spin_lock_init(&ls->ls_recover_idr_lock);
588 ls->ls_recover_list_count = 0;
589 ls->ls_local_handle = ls;
590 init_waitqueue_head(&ls->ls_wait_general);
591 INIT_LIST_HEAD(&ls->ls_root_list);
592 init_rwsem(&ls->ls_root_sem);
593
594 spin_lock(&lslist_lock);
595 ls->ls_create_count = 1;
596 list_add(&ls->ls_list, &lslist);
597 spin_unlock(&lslist_lock);
598
599 if (flags & DLM_LSFL_FS) {
600 error = dlm_callback_start(ls);
601 if (error) {
602 log_error(ls, "can't start dlm_callback %d", error);
603 goto out_delist;
604 }
605 }
606
607 init_waitqueue_head(&ls->ls_recover_lock_wait);
608
609 /*
610 * Once started, dlm_recoverd first looks for ls in lslist, then
611 * initializes ls_in_recovery as locked in "down" mode. We need
612 * to wait for the wakeup from dlm_recoverd because in_recovery
613 * has to start out in down mode.
614 */
615
616 error = dlm_recoverd_start(ls);
617 if (error) {
618 log_error(ls, "can't start dlm_recoverd %d", error);
619 goto out_callback;
620 }
621
622 wait_event(ls->ls_recover_lock_wait,
623 test_bit(LSFL_RECOVER_LOCK, &ls->ls_flags));
624
625 /* let kobject handle freeing of ls if there's an error */
626 do_unreg = 1;
627
628 ls->ls_kobj.kset = dlm_kset;
629 error = kobject_init_and_add(&ls->ls_kobj, &dlm_ktype, NULL,
630 "%s", ls->ls_name);
631 if (error)
632 goto out_recoverd;
633 kobject_uevent(&ls->ls_kobj, KOBJ_ADD);
634
635 /* This uevent triggers dlm_controld in userspace to add us to the
636 group of nodes that are members of this lockspace (managed by the
637 cluster infrastructure.) Once it's done that, it tells us who the
638 current lockspace members are (via configfs) and then tells the
639 lockspace to start running (via sysfs) in dlm_ls_start(). */
640
641 error = do_uevent(ls, 1);
642 if (error)
643 goto out_recoverd;
644
645 wait_for_completion(&ls->ls_members_done);
646 error = ls->ls_members_result;
647 if (error)
648 goto out_members;
649
650 dlm_create_debug_file(ls);
651
652 log_rinfo(ls, "join complete");
653 *lockspace = ls;
654 return 0;
655
656 out_members:
657 do_uevent(ls, 0);
658 dlm_clear_members(ls);
659 kfree(ls->ls_node_array);
660 out_recoverd:
661 dlm_recoverd_stop(ls);
662 out_callback:
663 dlm_callback_stop(ls);
664 out_delist:
665 spin_lock(&lslist_lock);
666 list_del(&ls->ls_list);
667 spin_unlock(&lslist_lock);
668 idr_destroy(&ls->ls_recover_idr);
669 kfree(ls->ls_recover_buf);
670 out_lkbidr:
671 idr_destroy(&ls->ls_lkbidr);
672 out_rsbtbl:
673 for (i = 0; i < DLM_REMOVE_NAMES_MAX; i++)
674 kfree(ls->ls_remove_names[i]);
675 vfree(ls->ls_rsbtbl);
676 out_lsfree:
677 if (do_unreg)
678 kobject_put(&ls->ls_kobj);
679 else
680 kfree(ls);
681 out:
682 module_put(THIS_MODULE);
683 return error;
684}
685
686int dlm_new_lockspace(const char *name, const char *cluster,
687 uint32_t flags, int lvblen,
688 const struct dlm_lockspace_ops *ops, void *ops_arg,
689 int *ops_result, dlm_lockspace_t **lockspace)
690{
691 int error = 0;
692
693 mutex_lock(&ls_lock);
694 if (!ls_count)
695 error = threads_start();
696 if (error)
697 goto out;
698
699 error = new_lockspace(name, cluster, flags, lvblen, ops, ops_arg,
700 ops_result, lockspace);
701 if (!error)
702 ls_count++;
703 if (error > 0)
704 error = 0;
705 if (!ls_count) {
706 dlm_scand_stop();
707 dlm_midcomms_shutdown();
708 dlm_lowcomms_stop();
709 }
710 out:
711 mutex_unlock(&ls_lock);
712 return error;
713}
714
715static int lkb_idr_is_local(int id, void *p, void *data)
716{
717 struct dlm_lkb *lkb = p;
718
719 return lkb->lkb_nodeid == 0 && lkb->lkb_grmode != DLM_LOCK_IV;
720}
721
722static int lkb_idr_is_any(int id, void *p, void *data)
723{
724 return 1;
725}
726
727static int lkb_idr_free(int id, void *p, void *data)
728{
729 struct dlm_lkb *lkb = p;
730
731 if (lkb->lkb_lvbptr && lkb->lkb_flags & DLM_IFL_MSTCPY)
732 dlm_free_lvb(lkb->lkb_lvbptr);
733
734 dlm_free_lkb(lkb);
735 return 0;
736}
737
738/* NOTE: We check the lkbidr here rather than the resource table.
739 This is because there may be LKBs queued as ASTs that have been unlinked
740 from their RSBs and are pending deletion once the AST has been delivered */
741
742static int lockspace_busy(struct dlm_ls *ls, int force)
743{
744 int rv;
745
746 spin_lock(&ls->ls_lkbidr_spin);
747 if (force == 0) {
748 rv = idr_for_each(&ls->ls_lkbidr, lkb_idr_is_any, ls);
749 } else if (force == 1) {
750 rv = idr_for_each(&ls->ls_lkbidr, lkb_idr_is_local, ls);
751 } else {
752 rv = 0;
753 }
754 spin_unlock(&ls->ls_lkbidr_spin);
755 return rv;
756}
757
758static int release_lockspace(struct dlm_ls *ls, int force)
759{
760 struct dlm_rsb *rsb;
761 struct rb_node *n;
762 int i, busy, rv;
763
764 busy = lockspace_busy(ls, force);
765
766 spin_lock(&lslist_lock);
767 if (ls->ls_create_count == 1) {
768 if (busy) {
769 rv = -EBUSY;
770 } else {
771 /* remove_lockspace takes ls off lslist */
772 ls->ls_create_count = 0;
773 rv = 0;
774 }
775 } else if (ls->ls_create_count > 1) {
776 rv = --ls->ls_create_count;
777 } else {
778 rv = -EINVAL;
779 }
780 spin_unlock(&lslist_lock);
781
782 if (rv) {
783 log_debug(ls, "release_lockspace no remove %d", rv);
784 return rv;
785 }
786
787 dlm_device_deregister(ls);
788
789 if (force < 3 && dlm_user_daemon_available())
790 do_uevent(ls, 0);
791
792 dlm_recoverd_stop(ls);
793
794 if (ls_count == 1) {
795 dlm_scand_stop();
796 dlm_midcomms_shutdown();
797 }
798
799 dlm_callback_stop(ls);
800
801 remove_lockspace(ls);
802
803 dlm_delete_debug_file(ls);
804
805 idr_destroy(&ls->ls_recover_idr);
806 kfree(ls->ls_recover_buf);
807
808 /*
809 * Free all lkb's in idr
810 */
811
812 idr_for_each(&ls->ls_lkbidr, lkb_idr_free, ls);
813 idr_destroy(&ls->ls_lkbidr);
814
815 /*
816 * Free all rsb's on rsbtbl[] lists
817 */
818
819 for (i = 0; i < ls->ls_rsbtbl_size; i++) {
820 while ((n = rb_first(&ls->ls_rsbtbl[i].keep))) {
821 rsb = rb_entry(n, struct dlm_rsb, res_hashnode);
822 rb_erase(n, &ls->ls_rsbtbl[i].keep);
823 dlm_free_rsb(rsb);
824 }
825
826 while ((n = rb_first(&ls->ls_rsbtbl[i].toss))) {
827 rsb = rb_entry(n, struct dlm_rsb, res_hashnode);
828 rb_erase(n, &ls->ls_rsbtbl[i].toss);
829 dlm_free_rsb(rsb);
830 }
831 }
832
833 vfree(ls->ls_rsbtbl);
834
835 for (i = 0; i < DLM_REMOVE_NAMES_MAX; i++)
836 kfree(ls->ls_remove_names[i]);
837
838 while (!list_empty(&ls->ls_new_rsb)) {
839 rsb = list_first_entry(&ls->ls_new_rsb, struct dlm_rsb,
840 res_hashchain);
841 list_del(&rsb->res_hashchain);
842 dlm_free_rsb(rsb);
843 }
844
845 /*
846 * Free structures on any other lists
847 */
848
849 dlm_purge_requestqueue(ls);
850 kfree(ls->ls_recover_args);
851 dlm_clear_members(ls);
852 dlm_clear_members_gone(ls);
853 kfree(ls->ls_node_array);
854 log_rinfo(ls, "release_lockspace final free");
855 kobject_put(&ls->ls_kobj);
856 /* The ls structure will be freed when the kobject is done with */
857
858 module_put(THIS_MODULE);
859 return 0;
860}
861
862/*
863 * Called when a system has released all its locks and is not going to use the
864 * lockspace any longer. We free everything we're managing for this lockspace.
865 * Remaining nodes will go through the recovery process as if we'd died. The
866 * lockspace must continue to function as usual, participating in recoveries,
867 * until this returns.
868 *
869 * Force has 4 possible values:
870 * 0 - don't destroy locksapce if it has any LKBs
871 * 1 - destroy lockspace if it has remote LKBs but not if it has local LKBs
872 * 2 - destroy lockspace regardless of LKBs
873 * 3 - destroy lockspace as part of a forced shutdown
874 */
875
876int dlm_release_lockspace(void *lockspace, int force)
877{
878 struct dlm_ls *ls;
879 int error;
880
881 ls = dlm_find_lockspace_local(lockspace);
882 if (!ls)
883 return -EINVAL;
884 dlm_put_lockspace(ls);
885
886 mutex_lock(&ls_lock);
887 error = release_lockspace(ls, force);
888 if (!error)
889 ls_count--;
890 if (!ls_count)
891 dlm_lowcomms_stop();
892 mutex_unlock(&ls_lock);
893
894 return error;
895}
896
897void dlm_stop_lockspaces(void)
898{
899 struct dlm_ls *ls;
900 int count;
901
902 restart:
903 count = 0;
904 spin_lock(&lslist_lock);
905 list_for_each_entry(ls, &lslist, ls_list) {
906 if (!test_bit(LSFL_RUNNING, &ls->ls_flags)) {
907 count++;
908 continue;
909 }
910 spin_unlock(&lslist_lock);
911 log_error(ls, "no userland control daemon, stopping lockspace");
912 dlm_ls_stop(ls);
913 goto restart;
914 }
915 spin_unlock(&lslist_lock);
916
917 if (count)
918 log_print("dlm user daemon left %d lockspaces", count);
919}
920
1// SPDX-License-Identifier: GPL-2.0-only
2/******************************************************************************
3*******************************************************************************
4**
5** Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
6** Copyright (C) 2004-2011 Red Hat, Inc. All rights reserved.
7**
8**
9*******************************************************************************
10******************************************************************************/
11
12#include <linux/module.h>
13
14#include "dlm_internal.h"
15#include "lockspace.h"
16#include "member.h"
17#include "recoverd.h"
18#include "dir.h"
19#include "lowcomms.h"
20#include "config.h"
21#include "memory.h"
22#include "lock.h"
23#include "recover.h"
24#include "requestqueue.h"
25#include "user.h"
26#include "ast.h"
27
28static int ls_count;
29static struct mutex ls_lock;
30static struct list_head lslist;
31static spinlock_t lslist_lock;
32static struct task_struct * scand_task;
33
34
35static ssize_t dlm_control_store(struct dlm_ls *ls, const char *buf, size_t len)
36{
37 ssize_t ret = len;
38 int n;
39 int rc = kstrtoint(buf, 0, &n);
40
41 if (rc)
42 return rc;
43 ls = dlm_find_lockspace_local(ls->ls_local_handle);
44 if (!ls)
45 return -EINVAL;
46
47 switch (n) {
48 case 0:
49 dlm_ls_stop(ls);
50 break;
51 case 1:
52 dlm_ls_start(ls);
53 break;
54 default:
55 ret = -EINVAL;
56 }
57 dlm_put_lockspace(ls);
58 return ret;
59}
60
61static ssize_t dlm_event_store(struct dlm_ls *ls, const char *buf, size_t len)
62{
63 int rc = kstrtoint(buf, 0, &ls->ls_uevent_result);
64
65 if (rc)
66 return rc;
67 set_bit(LSFL_UEVENT_WAIT, &ls->ls_flags);
68 wake_up(&ls->ls_uevent_wait);
69 return len;
70}
71
72static ssize_t dlm_id_show(struct dlm_ls *ls, char *buf)
73{
74 return snprintf(buf, PAGE_SIZE, "%u\n", ls->ls_global_id);
75}
76
77static ssize_t dlm_id_store(struct dlm_ls *ls, const char *buf, size_t len)
78{
79 int rc = kstrtouint(buf, 0, &ls->ls_global_id);
80
81 if (rc)
82 return rc;
83 return len;
84}
85
86static ssize_t dlm_nodir_show(struct dlm_ls *ls, char *buf)
87{
88 return snprintf(buf, PAGE_SIZE, "%u\n", dlm_no_directory(ls));
89}
90
91static ssize_t dlm_nodir_store(struct dlm_ls *ls, const char *buf, size_t len)
92{
93 int val;
94 int rc = kstrtoint(buf, 0, &val);
95
96 if (rc)
97 return rc;
98 if (val == 1)
99 set_bit(LSFL_NODIR, &ls->ls_flags);
100 return len;
101}
102
103static ssize_t dlm_recover_status_show(struct dlm_ls *ls, char *buf)
104{
105 uint32_t status = dlm_recover_status(ls);
106 return snprintf(buf, PAGE_SIZE, "%x\n", status);
107}
108
109static ssize_t dlm_recover_nodeid_show(struct dlm_ls *ls, char *buf)
110{
111 return snprintf(buf, PAGE_SIZE, "%d\n", ls->ls_recover_nodeid);
112}
113
114struct dlm_attr {
115 struct attribute attr;
116 ssize_t (*show)(struct dlm_ls *, char *);
117 ssize_t (*store)(struct dlm_ls *, const char *, size_t);
118};
119
120static struct dlm_attr dlm_attr_control = {
121 .attr = {.name = "control", .mode = S_IWUSR},
122 .store = dlm_control_store
123};
124
125static struct dlm_attr dlm_attr_event = {
126 .attr = {.name = "event_done", .mode = S_IWUSR},
127 .store = dlm_event_store
128};
129
130static struct dlm_attr dlm_attr_id = {
131 .attr = {.name = "id", .mode = S_IRUGO | S_IWUSR},
132 .show = dlm_id_show,
133 .store = dlm_id_store
134};
135
136static struct dlm_attr dlm_attr_nodir = {
137 .attr = {.name = "nodir", .mode = S_IRUGO | S_IWUSR},
138 .show = dlm_nodir_show,
139 .store = dlm_nodir_store
140};
141
142static struct dlm_attr dlm_attr_recover_status = {
143 .attr = {.name = "recover_status", .mode = S_IRUGO},
144 .show = dlm_recover_status_show
145};
146
147static struct dlm_attr dlm_attr_recover_nodeid = {
148 .attr = {.name = "recover_nodeid", .mode = S_IRUGO},
149 .show = dlm_recover_nodeid_show
150};
151
152static struct attribute *dlm_attrs[] = {
153 &dlm_attr_control.attr,
154 &dlm_attr_event.attr,
155 &dlm_attr_id.attr,
156 &dlm_attr_nodir.attr,
157 &dlm_attr_recover_status.attr,
158 &dlm_attr_recover_nodeid.attr,
159 NULL,
160};
161ATTRIBUTE_GROUPS(dlm);
162
163static ssize_t dlm_attr_show(struct kobject *kobj, struct attribute *attr,
164 char *buf)
165{
166 struct dlm_ls *ls = container_of(kobj, struct dlm_ls, ls_kobj);
167 struct dlm_attr *a = container_of(attr, struct dlm_attr, attr);
168 return a->show ? a->show(ls, buf) : 0;
169}
170
171static ssize_t dlm_attr_store(struct kobject *kobj, struct attribute *attr,
172 const char *buf, size_t len)
173{
174 struct dlm_ls *ls = container_of(kobj, struct dlm_ls, ls_kobj);
175 struct dlm_attr *a = container_of(attr, struct dlm_attr, attr);
176 return a->store ? a->store(ls, buf, len) : len;
177}
178
179static void lockspace_kobj_release(struct kobject *k)
180{
181 struct dlm_ls *ls = container_of(k, struct dlm_ls, ls_kobj);
182 kfree(ls);
183}
184
185static const struct sysfs_ops dlm_attr_ops = {
186 .show = dlm_attr_show,
187 .store = dlm_attr_store,
188};
189
190static struct kobj_type dlm_ktype = {
191 .default_groups = dlm_groups,
192 .sysfs_ops = &dlm_attr_ops,
193 .release = lockspace_kobj_release,
194};
195
196static struct kset *dlm_kset;
197
198static int do_uevent(struct dlm_ls *ls, int in)
199{
200 int error;
201
202 if (in)
203 kobject_uevent(&ls->ls_kobj, KOBJ_ONLINE);
204 else
205 kobject_uevent(&ls->ls_kobj, KOBJ_OFFLINE);
206
207 log_rinfo(ls, "%s the lockspace group...", in ? "joining" : "leaving");
208
209 /* dlm_controld will see the uevent, do the necessary group management
210 and then write to sysfs to wake us */
211
212 error = wait_event_interruptible(ls->ls_uevent_wait,
213 test_and_clear_bit(LSFL_UEVENT_WAIT, &ls->ls_flags));
214
215 log_rinfo(ls, "group event done %d %d", error, ls->ls_uevent_result);
216
217 if (error)
218 goto out;
219
220 error = ls->ls_uevent_result;
221 out:
222 if (error)
223 log_error(ls, "group %s failed %d %d", in ? "join" : "leave",
224 error, ls->ls_uevent_result);
225 return error;
226}
227
228static int dlm_uevent(struct kset *kset, struct kobject *kobj,
229 struct kobj_uevent_env *env)
230{
231 struct dlm_ls *ls = container_of(kobj, struct dlm_ls, ls_kobj);
232
233 add_uevent_var(env, "LOCKSPACE=%s", ls->ls_name);
234 return 0;
235}
236
237static const struct kset_uevent_ops dlm_uevent_ops = {
238 .uevent = dlm_uevent,
239};
240
241int __init dlm_lockspace_init(void)
242{
243 ls_count = 0;
244 mutex_init(&ls_lock);
245 INIT_LIST_HEAD(&lslist);
246 spin_lock_init(&lslist_lock);
247
248 dlm_kset = kset_create_and_add("dlm", &dlm_uevent_ops, kernel_kobj);
249 if (!dlm_kset) {
250 printk(KERN_WARNING "%s: can not create kset\n", __func__);
251 return -ENOMEM;
252 }
253 return 0;
254}
255
256void dlm_lockspace_exit(void)
257{
258 kset_unregister(dlm_kset);
259}
260
261static struct dlm_ls *find_ls_to_scan(void)
262{
263 struct dlm_ls *ls;
264
265 spin_lock(&lslist_lock);
266 list_for_each_entry(ls, &lslist, ls_list) {
267 if (time_after_eq(jiffies, ls->ls_scan_time +
268 dlm_config.ci_scan_secs * HZ)) {
269 spin_unlock(&lslist_lock);
270 return ls;
271 }
272 }
273 spin_unlock(&lslist_lock);
274 return NULL;
275}
276
277static int dlm_scand(void *data)
278{
279 struct dlm_ls *ls;
280
281 while (!kthread_should_stop()) {
282 ls = find_ls_to_scan();
283 if (ls) {
284 if (dlm_lock_recovery_try(ls)) {
285 ls->ls_scan_time = jiffies;
286 dlm_scan_rsbs(ls);
287 dlm_scan_timeout(ls);
288 dlm_scan_waiters(ls);
289 dlm_unlock_recovery(ls);
290 } else {
291 ls->ls_scan_time += HZ;
292 }
293 continue;
294 }
295 schedule_timeout_interruptible(dlm_config.ci_scan_secs * HZ);
296 }
297 return 0;
298}
299
300static int dlm_scand_start(void)
301{
302 struct task_struct *p;
303 int error = 0;
304
305 p = kthread_run(dlm_scand, NULL, "dlm_scand");
306 if (IS_ERR(p))
307 error = PTR_ERR(p);
308 else
309 scand_task = p;
310 return error;
311}
312
313static void dlm_scand_stop(void)
314{
315 kthread_stop(scand_task);
316}
317
318struct dlm_ls *dlm_find_lockspace_global(uint32_t id)
319{
320 struct dlm_ls *ls;
321
322 spin_lock(&lslist_lock);
323
324 list_for_each_entry(ls, &lslist, ls_list) {
325 if (ls->ls_global_id == id) {
326 ls->ls_count++;
327 goto out;
328 }
329 }
330 ls = NULL;
331 out:
332 spin_unlock(&lslist_lock);
333 return ls;
334}
335
336struct dlm_ls *dlm_find_lockspace_local(dlm_lockspace_t *lockspace)
337{
338 struct dlm_ls *ls;
339
340 spin_lock(&lslist_lock);
341 list_for_each_entry(ls, &lslist, ls_list) {
342 if (ls->ls_local_handle == lockspace) {
343 ls->ls_count++;
344 goto out;
345 }
346 }
347 ls = NULL;
348 out:
349 spin_unlock(&lslist_lock);
350 return ls;
351}
352
353struct dlm_ls *dlm_find_lockspace_device(int minor)
354{
355 struct dlm_ls *ls;
356
357 spin_lock(&lslist_lock);
358 list_for_each_entry(ls, &lslist, ls_list) {
359 if (ls->ls_device.minor == minor) {
360 ls->ls_count++;
361 goto out;
362 }
363 }
364 ls = NULL;
365 out:
366 spin_unlock(&lslist_lock);
367 return ls;
368}
369
370void dlm_put_lockspace(struct dlm_ls *ls)
371{
372 spin_lock(&lslist_lock);
373 ls->ls_count--;
374 spin_unlock(&lslist_lock);
375}
376
377static void remove_lockspace(struct dlm_ls *ls)
378{
379 for (;;) {
380 spin_lock(&lslist_lock);
381 if (ls->ls_count == 0) {
382 WARN_ON(ls->ls_create_count != 0);
383 list_del(&ls->ls_list);
384 spin_unlock(&lslist_lock);
385 return;
386 }
387 spin_unlock(&lslist_lock);
388 ssleep(1);
389 }
390}
391
392static int threads_start(void)
393{
394 int error;
395
396 error = dlm_scand_start();
397 if (error) {
398 log_print("cannot start dlm_scand thread %d", error);
399 goto fail;
400 }
401
402 /* Thread for sending/receiving messages for all lockspace's */
403 error = dlm_lowcomms_start();
404 if (error) {
405 log_print("cannot start dlm lowcomms %d", error);
406 goto scand_fail;
407 }
408
409 return 0;
410
411 scand_fail:
412 dlm_scand_stop();
413 fail:
414 return error;
415}
416
417static void threads_stop(void)
418{
419 dlm_scand_stop();
420 dlm_lowcomms_stop();
421}
422
423static int new_lockspace(const char *name, const char *cluster,
424 uint32_t flags, int lvblen,
425 const struct dlm_lockspace_ops *ops, void *ops_arg,
426 int *ops_result, dlm_lockspace_t **lockspace)
427{
428 struct dlm_ls *ls;
429 int i, size, error;
430 int do_unreg = 0;
431 int namelen = strlen(name);
432
433 if (namelen > DLM_LOCKSPACE_LEN || namelen == 0)
434 return -EINVAL;
435
436 if (!lvblen || (lvblen % 8))
437 return -EINVAL;
438
439 if (!try_module_get(THIS_MODULE))
440 return -EINVAL;
441
442 if (!dlm_user_daemon_available()) {
443 log_print("dlm user daemon not available");
444 error = -EUNATCH;
445 goto out;
446 }
447
448 if (ops && ops_result) {
449 if (!dlm_config.ci_recover_callbacks)
450 *ops_result = -EOPNOTSUPP;
451 else
452 *ops_result = 0;
453 }
454
455 if (!cluster)
456 log_print("dlm cluster name '%s' is being used without an application provided cluster name",
457 dlm_config.ci_cluster_name);
458
459 if (dlm_config.ci_recover_callbacks && cluster &&
460 strncmp(cluster, dlm_config.ci_cluster_name, DLM_LOCKSPACE_LEN)) {
461 log_print("dlm cluster name '%s' does not match "
462 "the application cluster name '%s'",
463 dlm_config.ci_cluster_name, cluster);
464 error = -EBADR;
465 goto out;
466 }
467
468 error = 0;
469
470 spin_lock(&lslist_lock);
471 list_for_each_entry(ls, &lslist, ls_list) {
472 WARN_ON(ls->ls_create_count <= 0);
473 if (ls->ls_namelen != namelen)
474 continue;
475 if (memcmp(ls->ls_name, name, namelen))
476 continue;
477 if (flags & DLM_LSFL_NEWEXCL) {
478 error = -EEXIST;
479 break;
480 }
481 ls->ls_create_count++;
482 *lockspace = ls;
483 error = 1;
484 break;
485 }
486 spin_unlock(&lslist_lock);
487
488 if (error)
489 goto out;
490
491 error = -ENOMEM;
492
493 ls = kzalloc(sizeof(struct dlm_ls) + namelen, GFP_NOFS);
494 if (!ls)
495 goto out;
496 memcpy(ls->ls_name, name, namelen);
497 ls->ls_namelen = namelen;
498 ls->ls_lvblen = lvblen;
499 ls->ls_count = 0;
500 ls->ls_flags = 0;
501 ls->ls_scan_time = jiffies;
502
503 if (ops && dlm_config.ci_recover_callbacks) {
504 ls->ls_ops = ops;
505 ls->ls_ops_arg = ops_arg;
506 }
507
508 if (flags & DLM_LSFL_TIMEWARN)
509 set_bit(LSFL_TIMEWARN, &ls->ls_flags);
510
511 /* ls_exflags are forced to match among nodes, and we don't
512 need to require all nodes to have some flags set */
513 ls->ls_exflags = (flags & ~(DLM_LSFL_TIMEWARN | DLM_LSFL_FS |
514 DLM_LSFL_NEWEXCL));
515
516 size = dlm_config.ci_rsbtbl_size;
517 ls->ls_rsbtbl_size = size;
518
519 ls->ls_rsbtbl = vmalloc(array_size(size, sizeof(struct dlm_rsbtable)));
520 if (!ls->ls_rsbtbl)
521 goto out_lsfree;
522 for (i = 0; i < size; i++) {
523 ls->ls_rsbtbl[i].keep.rb_node = NULL;
524 ls->ls_rsbtbl[i].toss.rb_node = NULL;
525 spin_lock_init(&ls->ls_rsbtbl[i].lock);
526 }
527
528 spin_lock_init(&ls->ls_remove_spin);
529
530 for (i = 0; i < DLM_REMOVE_NAMES_MAX; i++) {
531 ls->ls_remove_names[i] = kzalloc(DLM_RESNAME_MAXLEN+1,
532 GFP_KERNEL);
533 if (!ls->ls_remove_names[i])
534 goto out_rsbtbl;
535 }
536
537 idr_init(&ls->ls_lkbidr);
538 spin_lock_init(&ls->ls_lkbidr_spin);
539
540 INIT_LIST_HEAD(&ls->ls_waiters);
541 mutex_init(&ls->ls_waiters_mutex);
542 INIT_LIST_HEAD(&ls->ls_orphans);
543 mutex_init(&ls->ls_orphans_mutex);
544 INIT_LIST_HEAD(&ls->ls_timeout);
545 mutex_init(&ls->ls_timeout_mutex);
546
547 INIT_LIST_HEAD(&ls->ls_new_rsb);
548 spin_lock_init(&ls->ls_new_rsb_spin);
549
550 INIT_LIST_HEAD(&ls->ls_nodes);
551 INIT_LIST_HEAD(&ls->ls_nodes_gone);
552 ls->ls_num_nodes = 0;
553 ls->ls_low_nodeid = 0;
554 ls->ls_total_weight = 0;
555 ls->ls_node_array = NULL;
556
557 memset(&ls->ls_stub_rsb, 0, sizeof(struct dlm_rsb));
558 ls->ls_stub_rsb.res_ls = ls;
559
560 ls->ls_debug_rsb_dentry = NULL;
561 ls->ls_debug_waiters_dentry = NULL;
562
563 init_waitqueue_head(&ls->ls_uevent_wait);
564 ls->ls_uevent_result = 0;
565 init_completion(&ls->ls_members_done);
566 ls->ls_members_result = -1;
567
568 mutex_init(&ls->ls_cb_mutex);
569 INIT_LIST_HEAD(&ls->ls_cb_delay);
570
571 ls->ls_recoverd_task = NULL;
572 mutex_init(&ls->ls_recoverd_active);
573 spin_lock_init(&ls->ls_recover_lock);
574 spin_lock_init(&ls->ls_rcom_spin);
575 get_random_bytes(&ls->ls_rcom_seq, sizeof(uint64_t));
576 ls->ls_recover_status = 0;
577 ls->ls_recover_seq = 0;
578 ls->ls_recover_args = NULL;
579 init_rwsem(&ls->ls_in_recovery);
580 init_rwsem(&ls->ls_recv_active);
581 INIT_LIST_HEAD(&ls->ls_requestqueue);
582 mutex_init(&ls->ls_requestqueue_mutex);
583 mutex_init(&ls->ls_clear_proc_locks);
584
585 ls->ls_recover_buf = kmalloc(dlm_config.ci_buffer_size, GFP_NOFS);
586 if (!ls->ls_recover_buf)
587 goto out_lkbidr;
588
589 ls->ls_slot = 0;
590 ls->ls_num_slots = 0;
591 ls->ls_slots_size = 0;
592 ls->ls_slots = NULL;
593
594 INIT_LIST_HEAD(&ls->ls_recover_list);
595 spin_lock_init(&ls->ls_recover_list_lock);
596 idr_init(&ls->ls_recover_idr);
597 spin_lock_init(&ls->ls_recover_idr_lock);
598 ls->ls_recover_list_count = 0;
599 ls->ls_local_handle = ls;
600 init_waitqueue_head(&ls->ls_wait_general);
601 INIT_LIST_HEAD(&ls->ls_root_list);
602 init_rwsem(&ls->ls_root_sem);
603
604 spin_lock(&lslist_lock);
605 ls->ls_create_count = 1;
606 list_add(&ls->ls_list, &lslist);
607 spin_unlock(&lslist_lock);
608
609 if (flags & DLM_LSFL_FS) {
610 error = dlm_callback_start(ls);
611 if (error) {
612 log_error(ls, "can't start dlm_callback %d", error);
613 goto out_delist;
614 }
615 }
616
617 init_waitqueue_head(&ls->ls_recover_lock_wait);
618
619 /*
620 * Once started, dlm_recoverd first looks for ls in lslist, then
621 * initializes ls_in_recovery as locked in "down" mode. We need
622 * to wait for the wakeup from dlm_recoverd because in_recovery
623 * has to start out in down mode.
624 */
625
626 error = dlm_recoverd_start(ls);
627 if (error) {
628 log_error(ls, "can't start dlm_recoverd %d", error);
629 goto out_callback;
630 }
631
632 wait_event(ls->ls_recover_lock_wait,
633 test_bit(LSFL_RECOVER_LOCK, &ls->ls_flags));
634
635 ls->ls_kobj.kset = dlm_kset;
636 error = kobject_init_and_add(&ls->ls_kobj, &dlm_ktype, NULL,
637 "%s", ls->ls_name);
638 if (error)
639 goto out_recoverd;
640 kobject_uevent(&ls->ls_kobj, KOBJ_ADD);
641
642 /* let kobject handle freeing of ls if there's an error */
643 do_unreg = 1;
644
645 /* This uevent triggers dlm_controld in userspace to add us to the
646 group of nodes that are members of this lockspace (managed by the
647 cluster infrastructure.) Once it's done that, it tells us who the
648 current lockspace members are (via configfs) and then tells the
649 lockspace to start running (via sysfs) in dlm_ls_start(). */
650
651 error = do_uevent(ls, 1);
652 if (error)
653 goto out_recoverd;
654
655 wait_for_completion(&ls->ls_members_done);
656 error = ls->ls_members_result;
657 if (error)
658 goto out_members;
659
660 dlm_create_debug_file(ls);
661
662 log_rinfo(ls, "join complete");
663 *lockspace = ls;
664 return 0;
665
666 out_members:
667 do_uevent(ls, 0);
668 dlm_clear_members(ls);
669 kfree(ls->ls_node_array);
670 out_recoverd:
671 dlm_recoverd_stop(ls);
672 out_callback:
673 dlm_callback_stop(ls);
674 out_delist:
675 spin_lock(&lslist_lock);
676 list_del(&ls->ls_list);
677 spin_unlock(&lslist_lock);
678 idr_destroy(&ls->ls_recover_idr);
679 kfree(ls->ls_recover_buf);
680 out_lkbidr:
681 idr_destroy(&ls->ls_lkbidr);
682 out_rsbtbl:
683 for (i = 0; i < DLM_REMOVE_NAMES_MAX; i++)
684 kfree(ls->ls_remove_names[i]);
685 vfree(ls->ls_rsbtbl);
686 out_lsfree:
687 if (do_unreg)
688 kobject_put(&ls->ls_kobj);
689 else
690 kfree(ls);
691 out:
692 module_put(THIS_MODULE);
693 return error;
694}
695
696int dlm_new_lockspace(const char *name, const char *cluster,
697 uint32_t flags, int lvblen,
698 const struct dlm_lockspace_ops *ops, void *ops_arg,
699 int *ops_result, dlm_lockspace_t **lockspace)
700{
701 int error = 0;
702
703 mutex_lock(&ls_lock);
704 if (!ls_count)
705 error = threads_start();
706 if (error)
707 goto out;
708
709 error = new_lockspace(name, cluster, flags, lvblen, ops, ops_arg,
710 ops_result, lockspace);
711 if (!error)
712 ls_count++;
713 if (error > 0)
714 error = 0;
715 if (!ls_count)
716 threads_stop();
717 out:
718 mutex_unlock(&ls_lock);
719 return error;
720}
721
722static int lkb_idr_is_local(int id, void *p, void *data)
723{
724 struct dlm_lkb *lkb = p;
725
726 return lkb->lkb_nodeid == 0 && lkb->lkb_grmode != DLM_LOCK_IV;
727}
728
729static int lkb_idr_is_any(int id, void *p, void *data)
730{
731 return 1;
732}
733
734static int lkb_idr_free(int id, void *p, void *data)
735{
736 struct dlm_lkb *lkb = p;
737
738 if (lkb->lkb_lvbptr && lkb->lkb_flags & DLM_IFL_MSTCPY)
739 dlm_free_lvb(lkb->lkb_lvbptr);
740
741 dlm_free_lkb(lkb);
742 return 0;
743}
744
745/* NOTE: We check the lkbidr here rather than the resource table.
746 This is because there may be LKBs queued as ASTs that have been unlinked
747 from their RSBs and are pending deletion once the AST has been delivered */
748
749static int lockspace_busy(struct dlm_ls *ls, int force)
750{
751 int rv;
752
753 spin_lock(&ls->ls_lkbidr_spin);
754 if (force == 0) {
755 rv = idr_for_each(&ls->ls_lkbidr, lkb_idr_is_any, ls);
756 } else if (force == 1) {
757 rv = idr_for_each(&ls->ls_lkbidr, lkb_idr_is_local, ls);
758 } else {
759 rv = 0;
760 }
761 spin_unlock(&ls->ls_lkbidr_spin);
762 return rv;
763}
764
765static int release_lockspace(struct dlm_ls *ls, int force)
766{
767 struct dlm_rsb *rsb;
768 struct rb_node *n;
769 int i, busy, rv;
770
771 busy = lockspace_busy(ls, force);
772
773 spin_lock(&lslist_lock);
774 if (ls->ls_create_count == 1) {
775 if (busy) {
776 rv = -EBUSY;
777 } else {
778 /* remove_lockspace takes ls off lslist */
779 ls->ls_create_count = 0;
780 rv = 0;
781 }
782 } else if (ls->ls_create_count > 1) {
783 rv = --ls->ls_create_count;
784 } else {
785 rv = -EINVAL;
786 }
787 spin_unlock(&lslist_lock);
788
789 if (rv) {
790 log_debug(ls, "release_lockspace no remove %d", rv);
791 return rv;
792 }
793
794 dlm_device_deregister(ls);
795
796 if (force < 3 && dlm_user_daemon_available())
797 do_uevent(ls, 0);
798
799 dlm_recoverd_stop(ls);
800
801 dlm_callback_stop(ls);
802
803 remove_lockspace(ls);
804
805 dlm_delete_debug_file(ls);
806
807 idr_destroy(&ls->ls_recover_idr);
808 kfree(ls->ls_recover_buf);
809
810 /*
811 * Free all lkb's in idr
812 */
813
814 idr_for_each(&ls->ls_lkbidr, lkb_idr_free, ls);
815 idr_destroy(&ls->ls_lkbidr);
816
817 /*
818 * Free all rsb's on rsbtbl[] lists
819 */
820
821 for (i = 0; i < ls->ls_rsbtbl_size; i++) {
822 while ((n = rb_first(&ls->ls_rsbtbl[i].keep))) {
823 rsb = rb_entry(n, struct dlm_rsb, res_hashnode);
824 rb_erase(n, &ls->ls_rsbtbl[i].keep);
825 dlm_free_rsb(rsb);
826 }
827
828 while ((n = rb_first(&ls->ls_rsbtbl[i].toss))) {
829 rsb = rb_entry(n, struct dlm_rsb, res_hashnode);
830 rb_erase(n, &ls->ls_rsbtbl[i].toss);
831 dlm_free_rsb(rsb);
832 }
833 }
834
835 vfree(ls->ls_rsbtbl);
836
837 for (i = 0; i < DLM_REMOVE_NAMES_MAX; i++)
838 kfree(ls->ls_remove_names[i]);
839
840 while (!list_empty(&ls->ls_new_rsb)) {
841 rsb = list_first_entry(&ls->ls_new_rsb, struct dlm_rsb,
842 res_hashchain);
843 list_del(&rsb->res_hashchain);
844 dlm_free_rsb(rsb);
845 }
846
847 /*
848 * Free structures on any other lists
849 */
850
851 dlm_purge_requestqueue(ls);
852 kfree(ls->ls_recover_args);
853 dlm_clear_members(ls);
854 dlm_clear_members_gone(ls);
855 kfree(ls->ls_node_array);
856 log_rinfo(ls, "release_lockspace final free");
857 kobject_put(&ls->ls_kobj);
858 /* The ls structure will be freed when the kobject is done with */
859
860 module_put(THIS_MODULE);
861 return 0;
862}
863
864/*
865 * Called when a system has released all its locks and is not going to use the
866 * lockspace any longer. We free everything we're managing for this lockspace.
867 * Remaining nodes will go through the recovery process as if we'd died. The
868 * lockspace must continue to function as usual, participating in recoveries,
869 * until this returns.
870 *
871 * Force has 4 possible values:
872 * 0 - don't destroy locksapce if it has any LKBs
873 * 1 - destroy lockspace if it has remote LKBs but not if it has local LKBs
874 * 2 - destroy lockspace regardless of LKBs
875 * 3 - destroy lockspace as part of a forced shutdown
876 */
877
878int dlm_release_lockspace(void *lockspace, int force)
879{
880 struct dlm_ls *ls;
881 int error;
882
883 ls = dlm_find_lockspace_local(lockspace);
884 if (!ls)
885 return -EINVAL;
886 dlm_put_lockspace(ls);
887
888 mutex_lock(&ls_lock);
889 error = release_lockspace(ls, force);
890 if (!error)
891 ls_count--;
892 if (!ls_count)
893 threads_stop();
894 mutex_unlock(&ls_lock);
895
896 return error;
897}
898
899void dlm_stop_lockspaces(void)
900{
901 struct dlm_ls *ls;
902 int count;
903
904 restart:
905 count = 0;
906 spin_lock(&lslist_lock);
907 list_for_each_entry(ls, &lslist, ls_list) {
908 if (!test_bit(LSFL_RUNNING, &ls->ls_flags)) {
909 count++;
910 continue;
911 }
912 spin_unlock(&lslist_lock);
913 log_error(ls, "no userland control daemon, stopping lockspace");
914 dlm_ls_stop(ls);
915 goto restart;
916 }
917 spin_unlock(&lslist_lock);
918
919 if (count)
920 log_print("dlm user daemon left %d lockspaces", count);
921}
922