4 This file is part of DRBD by Philipp Reisner and Lars Ellenberg.
6 Copyright (C) 2001-2008, LINBIT Information Technologies GmbH.
7 Copyright (C) 1999-2008, Philipp Reisner <philipp.reisner@linbit.com>.
8 Copyright (C) 2002-2008, Lars Ellenberg <lars.ellenberg@linbit.com>.
10 Thanks to Carter Burden, Bart Grantham and Gennadiy Nerubayev
11 from Logicworks, Inc. for making SDP replication support possible.
13 drbd is free software; you can redistribute it and/or modify
14 it under the terms of the GNU General Public License as published by
15 the Free Software Foundation; either version 2, or (at your option)
18 drbd is distributed in the hope that it will be useful,
19 but WITHOUT ANY WARRANTY; without even the implied warranty of
20 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
21 GNU General Public License for more details.
23 You should have received a copy of the GNU General Public License
24 along with drbd; see the file COPYING. If not, write to
25 the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
29 #include <linux/module.h>
30 #include <linux/drbd.h>
31 #include <asm/uaccess.h>
32 #include <asm/types.h>
34 #include <linux/ctype.h>
35 #include <linux/mutex.h>
37 #include <linux/file.h>
38 #include <linux/proc_fs.h>
39 #include <linux/init.h>
41 #include <linux/memcontrol.h>
42 #include <linux/mm_inline.h>
43 #include <linux/slab.h>
44 #include <linux/random.h>
45 #include <linux/reboot.h>
46 #include <linux/notifier.h>
47 #include <linux/kthread.h>
49 #define __KERNEL_SYSCALLS__
50 #include <linux/unistd.h>
51 #include <linux/vmalloc.h>
53 #include <linux/drbd_limits.h>
55 #include "drbd_req.h" /* only for _req_mod in tl_release and tl_clear */
59 struct after_state_chg_work {
63 enum chg_state_flags flags;
64 struct completion *done;
67 static DEFINE_MUTEX(drbd_main_mutex);
68 int drbdd_init(struct drbd_thread *);
69 int drbd_worker(struct drbd_thread *);
70 int drbd_asender(struct drbd_thread *);
73 static int drbd_open(struct block_device *bdev, fmode_t mode);
74 static int drbd_release(struct gendisk *gd, fmode_t mode);
75 static int w_after_state_ch(struct drbd_conf *mdev, struct drbd_work *w, int unused);
76 static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
77 union drbd_state ns, enum chg_state_flags flags);
78 static int w_md_sync(struct drbd_conf *mdev, struct drbd_work *w, int unused);
79 static void md_sync_timer_fn(unsigned long data);
80 static int w_bitmap_io(struct drbd_conf *mdev, struct drbd_work *w, int unused);
81 static int w_go_diskless(struct drbd_conf *mdev, struct drbd_work *w, int unused);
83 MODULE_AUTHOR("Philipp Reisner <phil@linbit.com>, "
84 "Lars Ellenberg <lars@linbit.com>");
85 MODULE_DESCRIPTION("drbd - Distributed Replicated Block Device v" REL_VERSION);
86 MODULE_VERSION(REL_VERSION);
87 MODULE_LICENSE("GPL");
88 MODULE_PARM_DESC(minor_count, "Maximum number of drbd devices (1-255)");
89 MODULE_ALIAS_BLOCKDEV_MAJOR(DRBD_MAJOR);
91 #include <linux/moduleparam.h>
92 /* allow_open_on_secondary */
93 MODULE_PARM_DESC(allow_oos, "DONT USE!");
94 /* thanks to these macros, if compiled into the kernel (not-module),
95 * this becomes the boot parameter drbd.minor_count */
96 module_param(minor_count, uint, 0444);
97 module_param(disable_sendpage, bool, 0644);
98 module_param(allow_oos, bool, 0);
99 module_param(cn_idx, uint, 0444);
100 module_param(proc_details, int, 0644);
102 #ifdef CONFIG_DRBD_FAULT_INJECTION
105 static int fault_count;
107 /* bitmap of enabled faults */
108 module_param(enable_faults, int, 0664);
109 /* fault rate % value - applies to all enabled faults */
110 module_param(fault_rate, int, 0664);
111 /* count of faults inserted */
112 module_param(fault_count, int, 0664);
113 /* bitmap of devices to insert faults on */
114 module_param(fault_devs, int, 0644);
117 /* module parameter, defined */
118 unsigned int minor_count = 32;
119 int disable_sendpage;
121 unsigned int cn_idx = CN_IDX_DRBD;
122 int proc_details; /* Detail level in proc drbd*/
124 /* Module parameter for setting the user mode helper program
125 * to run. Default is /sbin/drbdadm */
126 char usermode_helper[80] = "/sbin/drbdadm";
128 module_param_string(usermode_helper, usermode_helper, sizeof(usermode_helper), 0644);
130 /* in 2.6.x, our device mapping and config info contains our virtual gendisks
131 * as member "struct gendisk *vdisk;"
133 struct drbd_conf **minor_table;
135 struct kmem_cache *drbd_request_cache;
136 struct kmem_cache *drbd_ee_cache; /* epoch entries */
137 struct kmem_cache *drbd_bm_ext_cache; /* bitmap extents */
138 struct kmem_cache *drbd_al_ext_cache; /* activity log extents */
139 mempool_t *drbd_request_mempool;
140 mempool_t *drbd_ee_mempool;
142 /* I do not use a standard mempool, because:
143 1) I want to hand out the pre-allocated objects first.
144 2) I want to be able to interrupt sleeping allocation with a signal.
145 Note: This is a single linked list, the next pointer is the private
146 member of struct page.
148 struct page *drbd_pp_pool;
149 spinlock_t drbd_pp_lock;
151 wait_queue_head_t drbd_pp_wait;
153 DEFINE_RATELIMIT_STATE(drbd_ratelimit_state, 5 * HZ, 5);
155 static const struct block_device_operations drbd_ops = {
156 .owner = THIS_MODULE,
158 .release = drbd_release,
161 #define ARRY_SIZE(A) (sizeof(A)/sizeof(A[0]))
164 /* When checking with sparse, and this is an inline function, sparse will
165 give tons of false positives. When this is a real functions sparse works.
167 int _get_ldev_if_state(struct drbd_conf *mdev, enum drbd_disk_state mins)
171 atomic_inc(&mdev->local_cnt);
172 io_allowed = (mdev->state.disk >= mins);
174 if (atomic_dec_and_test(&mdev->local_cnt))
175 wake_up(&mdev->misc_wait);
183 * DOC: The transfer log
185 * The transfer log is a single linked list of &struct drbd_tl_epoch objects.
186 * mdev->newest_tle points to the head, mdev->oldest_tle points to the tail
187 * of the list. There is always at least one &struct drbd_tl_epoch object.
189 * Each &struct drbd_tl_epoch has a circular double linked list of requests
192 static int tl_init(struct drbd_conf *mdev)
194 struct drbd_tl_epoch *b;
196 /* during device minor initialization, we may well use GFP_KERNEL */
197 b = kmalloc(sizeof(struct drbd_tl_epoch), GFP_KERNEL);
200 INIT_LIST_HEAD(&b->requests);
201 INIT_LIST_HEAD(&b->w.list);
205 b->w.cb = NULL; /* if this is != NULL, we need to dec_ap_pending in tl_clear */
207 mdev->oldest_tle = b;
208 mdev->newest_tle = b;
209 INIT_LIST_HEAD(&mdev->out_of_sequence_requests);
211 mdev->tl_hash = NULL;
217 static void tl_cleanup(struct drbd_conf *mdev)
219 D_ASSERT(mdev->oldest_tle == mdev->newest_tle);
220 D_ASSERT(list_empty(&mdev->out_of_sequence_requests));
221 kfree(mdev->oldest_tle);
222 mdev->oldest_tle = NULL;
223 kfree(mdev->unused_spare_tle);
224 mdev->unused_spare_tle = NULL;
225 kfree(mdev->tl_hash);
226 mdev->tl_hash = NULL;
231 * _tl_add_barrier() - Adds a barrier to the transfer log
232 * @mdev: DRBD device.
233 * @new: Barrier to be added before the current head of the TL.
235 * The caller must hold the req_lock.
237 void _tl_add_barrier(struct drbd_conf *mdev, struct drbd_tl_epoch *new)
239 struct drbd_tl_epoch *newest_before;
241 INIT_LIST_HEAD(&new->requests);
242 INIT_LIST_HEAD(&new->w.list);
243 new->w.cb = NULL; /* if this is != NULL, we need to dec_ap_pending in tl_clear */
247 newest_before = mdev->newest_tle;
248 /* never send a barrier number == 0, because that is special-cased
249 * when using TCQ for our write ordering code */
250 new->br_number = (newest_before->br_number+1) ?: 1;
251 if (mdev->newest_tle != new) {
252 mdev->newest_tle->next = new;
253 mdev->newest_tle = new;
258 * tl_release() - Free or recycle the oldest &struct drbd_tl_epoch object of the TL
259 * @mdev: DRBD device.
260 * @barrier_nr: Expected identifier of the DRBD write barrier packet.
261 * @set_size: Expected number of requests before that barrier.
263 * In case the passed barrier_nr or set_size does not match the oldest
264 * &struct drbd_tl_epoch objects this function will cause a termination
267 void tl_release(struct drbd_conf *mdev, unsigned int barrier_nr,
268 unsigned int set_size)
270 struct drbd_tl_epoch *b, *nob; /* next old barrier */
271 struct list_head *le, *tle;
272 struct drbd_request *r;
274 spin_lock_irq(&mdev->req_lock);
276 b = mdev->oldest_tle;
278 /* first some paranoia code */
280 dev_err(DEV, "BAD! BarrierAck #%u received, but no epoch in tl!?\n",
284 if (b->br_number != barrier_nr) {
285 dev_err(DEV, "BAD! BarrierAck #%u received, expected #%u!\n",
286 barrier_nr, b->br_number);
289 if (b->n_writes != set_size) {
290 dev_err(DEV, "BAD! BarrierAck #%u received with n_writes=%u, expected n_writes=%u!\n",
291 barrier_nr, set_size, b->n_writes);
295 /* Clean up list of requests processed during current epoch */
296 list_for_each_safe(le, tle, &b->requests) {
297 r = list_entry(le, struct drbd_request, tl_requests);
298 _req_mod(r, barrier_acked);
300 /* There could be requests on the list waiting for completion
301 of the write to the local disk. To avoid corruptions of
302 slab's data structures we have to remove the lists head.
304 Also there could have been a barrier ack out of sequence, overtaking
305 the write acks - which would be a bug and violating write ordering.
306 To not deadlock in case we lose connection while such requests are
307 still pending, we need some way to find them for the
308 _req_mode(connection_lost_while_pending).
310 These have been list_move'd to the out_of_sequence_requests list in
311 _req_mod(, barrier_acked) above.
313 list_del_init(&b->requests);
316 if (test_and_clear_bit(CREATE_BARRIER, &mdev->flags)) {
317 _tl_add_barrier(mdev, b);
319 mdev->oldest_tle = nob;
320 /* if nob == NULL b was the only barrier, and becomes the new
321 barrier. Therefore mdev->oldest_tle points already to b */
323 D_ASSERT(nob != NULL);
324 mdev->oldest_tle = nob;
328 spin_unlock_irq(&mdev->req_lock);
329 dec_ap_pending(mdev);
334 spin_unlock_irq(&mdev->req_lock);
335 drbd_force_state(mdev, NS(conn, C_PROTOCOL_ERROR));
339 * _tl_restart() - Walks the transfer log, and applies an action to all requests
340 * @mdev: DRBD device.
341 * @what: The action/event to perform with all request objects
343 * @what might be one of connection_lost_while_pending, resend, fail_frozen_disk_io,
344 * restart_frozen_disk_io.
346 static void _tl_restart(struct drbd_conf *mdev, enum drbd_req_event what)
348 struct drbd_tl_epoch *b, *tmp, **pn;
349 struct list_head *le, *tle, carry_reads;
350 struct drbd_request *req;
351 int rv, n_writes, n_reads;
353 b = mdev->oldest_tle;
354 pn = &mdev->oldest_tle;
358 INIT_LIST_HEAD(&carry_reads);
359 list_for_each_safe(le, tle, &b->requests) {
360 req = list_entry(le, struct drbd_request, tl_requests);
361 rv = _req_mod(req, what);
363 n_writes += (rv & MR_WRITE) >> MR_WRITE_SHIFT;
364 n_reads += (rv & MR_READ) >> MR_READ_SHIFT;
369 if (what == resend) {
370 b->n_writes = n_writes;
371 if (b->w.cb == NULL) {
372 b->w.cb = w_send_barrier;
373 inc_ap_pending(mdev);
374 set_bit(CREATE_BARRIER, &mdev->flags);
377 drbd_queue_work(&mdev->data.work, &b->w);
382 list_add(&carry_reads, &b->requests);
383 /* there could still be requests on that ring list,
384 * in case local io is still pending */
385 list_del(&b->requests);
387 /* dec_ap_pending corresponding to queue_barrier.
388 * the newest barrier may not have been queued yet,
389 * in which case w.cb is still NULL. */
391 dec_ap_pending(mdev);
393 if (b == mdev->newest_tle) {
394 /* recycle, but reinit! */
395 D_ASSERT(tmp == NULL);
396 INIT_LIST_HEAD(&b->requests);
397 list_splice(&carry_reads, &b->requests);
398 INIT_LIST_HEAD(&b->w.list);
400 b->br_number = net_random();
410 list_splice(&carry_reads, &b->requests);
416 * tl_clear() - Clears all requests and &struct drbd_tl_epoch objects out of the TL
417 * @mdev: DRBD device.
419 * This is called after the connection to the peer was lost. The storage covered
420 * by the requests on the transfer gets marked as our of sync. Called from the
421 * receiver thread and the worker thread.
423 void tl_clear(struct drbd_conf *mdev)
425 struct list_head *le, *tle;
426 struct drbd_request *r;
428 spin_lock_irq(&mdev->req_lock);
430 _tl_restart(mdev, connection_lost_while_pending);
432 /* we expect this list to be empty. */
433 D_ASSERT(list_empty(&mdev->out_of_sequence_requests));
435 /* but just in case, clean it up anyways! */
436 list_for_each_safe(le, tle, &mdev->out_of_sequence_requests) {
437 r = list_entry(le, struct drbd_request, tl_requests);
438 /* It would be nice to complete outside of spinlock.
439 * But this is easier for now. */
440 _req_mod(r, connection_lost_while_pending);
443 /* ensure bit indicating barrier is required is clear */
444 clear_bit(CREATE_BARRIER, &mdev->flags);
446 memset(mdev->app_reads_hash, 0, APP_R_HSIZE*sizeof(void *));
448 spin_unlock_irq(&mdev->req_lock);
451 void tl_restart(struct drbd_conf *mdev, enum drbd_req_event what)
453 spin_lock_irq(&mdev->req_lock);
454 _tl_restart(mdev, what);
455 spin_unlock_irq(&mdev->req_lock);
459 * cl_wide_st_chg() - TRUE if the state change is a cluster wide one
460 * @mdev: DRBD device.
461 * @os: old (current) state.
462 * @ns: new (wanted) state.
464 static int cl_wide_st_chg(struct drbd_conf *mdev,
465 union drbd_state os, union drbd_state ns)
467 return (os.conn >= C_CONNECTED && ns.conn >= C_CONNECTED &&
468 ((os.role != R_PRIMARY && ns.role == R_PRIMARY) ||
469 (os.conn != C_STARTING_SYNC_T && ns.conn == C_STARTING_SYNC_T) ||
470 (os.conn != C_STARTING_SYNC_S && ns.conn == C_STARTING_SYNC_S) ||
471 (os.disk != D_DISKLESS && ns.disk == D_DISKLESS))) ||
472 (os.conn >= C_CONNECTED && ns.conn == C_DISCONNECTING) ||
473 (os.conn == C_CONNECTED && ns.conn == C_VERIFY_S);
476 int drbd_change_state(struct drbd_conf *mdev, enum chg_state_flags f,
477 union drbd_state mask, union drbd_state val)
480 union drbd_state os, ns;
483 spin_lock_irqsave(&mdev->req_lock, flags);
485 ns.i = (os.i & ~mask.i) | val.i;
486 rv = _drbd_set_state(mdev, ns, f, NULL);
488 spin_unlock_irqrestore(&mdev->req_lock, flags);
494 * drbd_force_state() - Impose a change which happens outside our control on our state
495 * @mdev: DRBD device.
496 * @mask: mask of state bits to change.
497 * @val: value of new state bits.
499 void drbd_force_state(struct drbd_conf *mdev,
500 union drbd_state mask, union drbd_state val)
502 drbd_change_state(mdev, CS_HARD, mask, val);
505 static int is_valid_state(struct drbd_conf *mdev, union drbd_state ns);
506 static int is_valid_state_transition(struct drbd_conf *,
507 union drbd_state, union drbd_state);
508 static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state os,
509 union drbd_state ns, const char **warn_sync_abort);
510 int drbd_send_state_req(struct drbd_conf *,
511 union drbd_state, union drbd_state);
513 static enum drbd_state_ret_codes _req_st_cond(struct drbd_conf *mdev,
514 union drbd_state mask, union drbd_state val)
516 union drbd_state os, ns;
520 if (test_and_clear_bit(CL_ST_CHG_SUCCESS, &mdev->flags))
521 return SS_CW_SUCCESS;
523 if (test_and_clear_bit(CL_ST_CHG_FAIL, &mdev->flags))
524 return SS_CW_FAILED_BY_PEER;
527 spin_lock_irqsave(&mdev->req_lock, flags);
529 ns.i = (os.i & ~mask.i) | val.i;
530 ns = sanitize_state(mdev, os, ns, NULL);
532 if (!cl_wide_st_chg(mdev, os, ns))
535 rv = is_valid_state(mdev, ns);
536 if (rv == SS_SUCCESS) {
537 rv = is_valid_state_transition(mdev, ns, os);
538 if (rv == SS_SUCCESS)
539 rv = 0; /* cont waiting, otherwise fail. */
542 spin_unlock_irqrestore(&mdev->req_lock, flags);
548 * drbd_req_state() - Perform an eventually cluster wide state change
549 * @mdev: DRBD device.
550 * @mask: mask of state bits to change.
551 * @val: value of new state bits.
554 * Should not be called directly, use drbd_request_state() or
555 * _drbd_request_state().
557 static int drbd_req_state(struct drbd_conf *mdev,
558 union drbd_state mask, union drbd_state val,
559 enum chg_state_flags f)
561 struct completion done;
563 union drbd_state os, ns;
566 init_completion(&done);
568 if (f & CS_SERIALIZE)
569 mutex_lock(&mdev->state_mutex);
571 spin_lock_irqsave(&mdev->req_lock, flags);
573 ns.i = (os.i & ~mask.i) | val.i;
574 ns = sanitize_state(mdev, os, ns, NULL);
576 if (cl_wide_st_chg(mdev, os, ns)) {
577 rv = is_valid_state(mdev, ns);
578 if (rv == SS_SUCCESS)
579 rv = is_valid_state_transition(mdev, ns, os);
580 spin_unlock_irqrestore(&mdev->req_lock, flags);
582 if (rv < SS_SUCCESS) {
584 print_st_err(mdev, os, ns, rv);
588 drbd_state_lock(mdev);
589 if (!drbd_send_state_req(mdev, mask, val)) {
590 drbd_state_unlock(mdev);
591 rv = SS_CW_FAILED_BY_PEER;
593 print_st_err(mdev, os, ns, rv);
597 wait_event(mdev->state_wait,
598 (rv = _req_st_cond(mdev, mask, val)));
600 if (rv < SS_SUCCESS) {
601 drbd_state_unlock(mdev);
603 print_st_err(mdev, os, ns, rv);
606 spin_lock_irqsave(&mdev->req_lock, flags);
608 ns.i = (os.i & ~mask.i) | val.i;
609 rv = _drbd_set_state(mdev, ns, f, &done);
610 drbd_state_unlock(mdev);
612 rv = _drbd_set_state(mdev, ns, f, &done);
615 spin_unlock_irqrestore(&mdev->req_lock, flags);
617 if (f & CS_WAIT_COMPLETE && rv == SS_SUCCESS) {
618 D_ASSERT(current != mdev->worker.task);
619 wait_for_completion(&done);
623 if (f & CS_SERIALIZE)
624 mutex_unlock(&mdev->state_mutex);
630 * _drbd_request_state() - Request a state change (with flags)
631 * @mdev: DRBD device.
632 * @mask: mask of state bits to change.
633 * @val: value of new state bits.
636 * Cousin of drbd_request_state(), useful with the CS_WAIT_COMPLETE
637 * flag, or when logging of failed state change requests is not desired.
639 int _drbd_request_state(struct drbd_conf *mdev, union drbd_state mask,
640 union drbd_state val, enum chg_state_flags f)
644 wait_event(mdev->state_wait,
645 (rv = drbd_req_state(mdev, mask, val, f)) != SS_IN_TRANSIENT_STATE);
650 static void print_st(struct drbd_conf *mdev, char *name, union drbd_state ns)
652 dev_err(DEV, " %s = { cs:%s ro:%s/%s ds:%s/%s %c%c%c%c }\n",
654 drbd_conn_str(ns.conn),
655 drbd_role_str(ns.role),
656 drbd_role_str(ns.peer),
657 drbd_disk_str(ns.disk),
658 drbd_disk_str(ns.pdsk),
659 is_susp(ns) ? 's' : 'r',
660 ns.aftr_isp ? 'a' : '-',
661 ns.peer_isp ? 'p' : '-',
662 ns.user_isp ? 'u' : '-'
666 void print_st_err(struct drbd_conf *mdev,
667 union drbd_state os, union drbd_state ns, int err)
669 if (err == SS_IN_TRANSIENT_STATE)
671 dev_err(DEV, "State change failed: %s\n", drbd_set_st_err_str(err));
672 print_st(mdev, " state", os);
673 print_st(mdev, "wanted", ns);
677 #define drbd_peer_str drbd_role_str
678 #define drbd_pdsk_str drbd_disk_str
680 #define drbd_susp_str(A) ((A) ? "1" : "0")
681 #define drbd_aftr_isp_str(A) ((A) ? "1" : "0")
682 #define drbd_peer_isp_str(A) ((A) ? "1" : "0")
683 #define drbd_user_isp_str(A) ((A) ? "1" : "0")
686 ({ if (ns.A != os.A) { \
687 pbp += sprintf(pbp, #A "( %s -> %s ) ", \
688 drbd_##A##_str(os.A), \
689 drbd_##A##_str(ns.A)); \
693 * is_valid_state() - Returns an SS_ error code if ns is not valid
694 * @mdev: DRBD device.
695 * @ns: State to consider.
697 static int is_valid_state(struct drbd_conf *mdev, union drbd_state ns)
699 /* See drbd_state_sw_errors in drbd_strings.c */
701 enum drbd_fencing_p fp;
705 if (get_ldev(mdev)) {
706 fp = mdev->ldev->dc.fencing;
710 if (get_net_conf(mdev)) {
711 if (!mdev->net_conf->two_primaries &&
712 ns.role == R_PRIMARY && ns.peer == R_PRIMARY)
713 rv = SS_TWO_PRIMARIES;
718 /* already found a reason to abort */;
719 else if (ns.role == R_SECONDARY && mdev->open_cnt)
720 rv = SS_DEVICE_IN_USE;
722 else if (ns.role == R_PRIMARY && ns.conn < C_CONNECTED && ns.disk < D_UP_TO_DATE)
723 rv = SS_NO_UP_TO_DATE_DISK;
725 else if (fp >= FP_RESOURCE &&
726 ns.role == R_PRIMARY && ns.conn < C_CONNECTED && ns.pdsk >= D_UNKNOWN)
729 else if (ns.role == R_PRIMARY && ns.disk <= D_INCONSISTENT && ns.pdsk <= D_INCONSISTENT)
730 rv = SS_NO_UP_TO_DATE_DISK;
732 else if (ns.conn > C_CONNECTED && ns.disk < D_INCONSISTENT)
733 rv = SS_NO_LOCAL_DISK;
735 else if (ns.conn > C_CONNECTED && ns.pdsk < D_INCONSISTENT)
736 rv = SS_NO_REMOTE_DISK;
738 else if (ns.conn > C_CONNECTED && ns.disk < D_UP_TO_DATE && ns.pdsk < D_UP_TO_DATE)
739 rv = SS_NO_UP_TO_DATE_DISK;
741 else if ((ns.conn == C_CONNECTED ||
742 ns.conn == C_WF_BITMAP_S ||
743 ns.conn == C_SYNC_SOURCE ||
744 ns.conn == C_PAUSED_SYNC_S) &&
745 ns.disk == D_OUTDATED)
746 rv = SS_CONNECTED_OUTDATES;
748 else if ((ns.conn == C_VERIFY_S || ns.conn == C_VERIFY_T) &&
749 (mdev->sync_conf.verify_alg[0] == 0))
750 rv = SS_NO_VERIFY_ALG;
752 else if ((ns.conn == C_VERIFY_S || ns.conn == C_VERIFY_T) &&
753 mdev->agreed_pro_version < 88)
754 rv = SS_NOT_SUPPORTED;
760 * is_valid_state_transition() - Returns an SS_ error code if the state transition is not possible
761 * @mdev: DRBD device.
765 static int is_valid_state_transition(struct drbd_conf *mdev,
766 union drbd_state ns, union drbd_state os)
770 if ((ns.conn == C_STARTING_SYNC_T || ns.conn == C_STARTING_SYNC_S) &&
771 os.conn > C_CONNECTED)
772 rv = SS_RESYNC_RUNNING;
774 if (ns.conn == C_DISCONNECTING && os.conn == C_STANDALONE)
775 rv = SS_ALREADY_STANDALONE;
777 if (ns.disk > D_ATTACHING && os.disk == D_DISKLESS)
780 if (ns.conn == C_WF_CONNECTION && os.conn < C_UNCONNECTED)
781 rv = SS_NO_NET_CONFIG;
783 if (ns.disk == D_OUTDATED && os.disk < D_OUTDATED && os.disk != D_ATTACHING)
784 rv = SS_LOWER_THAN_OUTDATED;
786 if (ns.conn == C_DISCONNECTING && os.conn == C_UNCONNECTED)
787 rv = SS_IN_TRANSIENT_STATE;
789 if (ns.conn == os.conn && ns.conn == C_WF_REPORT_PARAMS)
790 rv = SS_IN_TRANSIENT_STATE;
792 if ((ns.conn == C_VERIFY_S || ns.conn == C_VERIFY_T) && os.conn < C_CONNECTED)
793 rv = SS_NEED_CONNECTION;
795 if ((ns.conn == C_VERIFY_S || ns.conn == C_VERIFY_T) &&
796 ns.conn != os.conn && os.conn > C_CONNECTED)
797 rv = SS_RESYNC_RUNNING;
799 if ((ns.conn == C_STARTING_SYNC_S || ns.conn == C_STARTING_SYNC_T) &&
800 os.conn < C_CONNECTED)
801 rv = SS_NEED_CONNECTION;
807 * sanitize_state() - Resolves implicitly necessary additional changes to a state transition
808 * @mdev: DRBD device.
813 * When we loose connection, we have to set the state of the peers disk (pdsk)
814 * to D_UNKNOWN. This rule and many more along those lines are in this function.
816 static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state os,
817 union drbd_state ns, const char **warn_sync_abort)
819 enum drbd_fencing_p fp;
822 if (get_ldev(mdev)) {
823 fp = mdev->ldev->dc.fencing;
827 /* Disallow Network errors to configure a device's network part */
828 if ((ns.conn >= C_TIMEOUT && ns.conn <= C_TEAR_DOWN) &&
829 os.conn <= C_DISCONNECTING)
832 /* After a network error (+C_TEAR_DOWN) only C_UNCONNECTED or C_DISCONNECTING can follow.
833 * If you try to go into some Sync* state, that shall fail (elsewhere). */
834 if (os.conn >= C_TIMEOUT && os.conn <= C_TEAR_DOWN &&
835 ns.conn != C_UNCONNECTED && ns.conn != C_DISCONNECTING && ns.conn <= C_TEAR_DOWN)
838 /* After C_DISCONNECTING only C_STANDALONE may follow */
839 if (os.conn == C_DISCONNECTING && ns.conn != C_STANDALONE)
842 if (ns.conn < C_CONNECTED) {
845 if (ns.pdsk > D_UNKNOWN || ns.pdsk < D_INCONSISTENT)
849 /* Clear the aftr_isp when becoming unconfigured */
850 if (ns.conn == C_STANDALONE && ns.disk == D_DISKLESS && ns.role == R_SECONDARY)
853 /* Abort resync if a disk fails/detaches */
854 if (os.conn > C_CONNECTED && ns.conn > C_CONNECTED &&
855 (ns.disk <= D_FAILED || ns.pdsk <= D_FAILED)) {
858 os.conn == C_VERIFY_S || os.conn == C_VERIFY_T ?
859 "Online-verify" : "Resync";
860 ns.conn = C_CONNECTED;
863 if (ns.conn >= C_CONNECTED &&
864 ((ns.disk == D_CONSISTENT || ns.disk == D_OUTDATED) ||
865 (ns.disk == D_NEGOTIATING && ns.conn == C_WF_BITMAP_T))) {
868 case C_PAUSED_SYNC_T:
869 ns.disk = D_OUTDATED;
874 case C_PAUSED_SYNC_S:
875 ns.disk = D_UP_TO_DATE;
878 ns.disk = D_INCONSISTENT;
879 dev_warn(DEV, "Implicitly set disk state Inconsistent!\n");
882 if (os.disk == D_OUTDATED && ns.disk == D_UP_TO_DATE)
883 dev_warn(DEV, "Implicitly set disk from Outdated to UpToDate\n");
886 if (ns.conn >= C_CONNECTED &&
887 (ns.pdsk == D_CONSISTENT || ns.pdsk == D_OUTDATED)) {
891 case C_PAUSED_SYNC_T:
893 ns.pdsk = D_UP_TO_DATE;
896 case C_PAUSED_SYNC_S:
897 /* remap any consistent state to D_OUTDATED,
898 * but disallow "upgrade" of not even consistent states.
901 (D_DISKLESS < os.pdsk && os.pdsk < D_OUTDATED)
902 ? os.pdsk : D_OUTDATED;
905 ns.pdsk = D_INCONSISTENT;
906 dev_warn(DEV, "Implicitly set pdsk Inconsistent!\n");
909 if (os.pdsk == D_OUTDATED && ns.pdsk == D_UP_TO_DATE)
910 dev_warn(DEV, "Implicitly set pdsk from Outdated to UpToDate\n");
913 /* Connection breaks down before we finished "Negotiating" */
914 if (ns.conn < C_CONNECTED && ns.disk == D_NEGOTIATING &&
915 get_ldev_if_state(mdev, D_NEGOTIATING)) {
916 if (mdev->ed_uuid == mdev->ldev->md.uuid[UI_CURRENT]) {
917 ns.disk = mdev->new_state_tmp.disk;
918 ns.pdsk = mdev->new_state_tmp.pdsk;
920 dev_alert(DEV, "Connection lost while negotiating, no data!\n");
921 ns.disk = D_DISKLESS;
927 if (fp == FP_STONITH &&
928 (ns.role == R_PRIMARY && ns.conn < C_CONNECTED && ns.pdsk > D_OUTDATED) &&
929 !(os.role == R_PRIMARY && os.conn < C_CONNECTED && os.pdsk > D_OUTDATED))
930 ns.susp_fen = 1; /* Suspend IO while fence-peer handler runs (peer lost) */
932 if (mdev->sync_conf.on_no_data == OND_SUSPEND_IO &&
933 (ns.role == R_PRIMARY && ns.disk < D_UP_TO_DATE && ns.pdsk < D_UP_TO_DATE) &&
934 !(os.role == R_PRIMARY && os.disk < D_UP_TO_DATE && os.pdsk < D_UP_TO_DATE))
935 ns.susp_nod = 1; /* Suspend IO while no data available (no accessible data available) */
937 if (ns.aftr_isp || ns.peer_isp || ns.user_isp) {
938 if (ns.conn == C_SYNC_SOURCE)
939 ns.conn = C_PAUSED_SYNC_S;
940 if (ns.conn == C_SYNC_TARGET)
941 ns.conn = C_PAUSED_SYNC_T;
943 if (ns.conn == C_PAUSED_SYNC_S)
944 ns.conn = C_SYNC_SOURCE;
945 if (ns.conn == C_PAUSED_SYNC_T)
946 ns.conn = C_SYNC_TARGET;
952 /* helper for __drbd_set_state */
953 static void set_ov_position(struct drbd_conf *mdev, enum drbd_conns cs)
955 if (cs == C_VERIFY_T) {
956 /* starting online verify from an arbitrary position
957 * does not fit well into the existing protocol.
958 * on C_VERIFY_T, we initialize ov_left and friends
959 * implicitly in receive_DataRequest once the
960 * first P_OV_REQUEST is received */
961 mdev->ov_start_sector = ~(sector_t)0;
963 unsigned long bit = BM_SECT_TO_BIT(mdev->ov_start_sector);
964 if (bit >= mdev->rs_total)
965 mdev->ov_start_sector =
966 BM_BIT_TO_SECT(mdev->rs_total - 1);
967 mdev->ov_position = mdev->ov_start_sector;
971 static void drbd_resume_al(struct drbd_conf *mdev)
973 if (test_and_clear_bit(AL_SUSPENDED, &mdev->flags))
974 dev_info(DEV, "Resumed AL updates\n");
978 * __drbd_set_state() - Set a new DRBD state
979 * @mdev: DRBD device.
982 * @done: Optional completion, that will get completed after the after_state_ch() finished
984 * Caller needs to hold req_lock, and global_state_lock. Do not call directly.
986 int __drbd_set_state(struct drbd_conf *mdev,
987 union drbd_state ns, enum chg_state_flags flags,
988 struct completion *done)
992 const char *warn_sync_abort = NULL;
993 struct after_state_chg_work *ascw;
997 ns = sanitize_state(mdev, os, ns, &warn_sync_abort);
1000 return SS_NOTHING_TO_DO;
1002 if (!(flags & CS_HARD)) {
1003 /* pre-state-change checks ; only look at ns */
1004 /* See drbd_state_sw_errors in drbd_strings.c */
1006 rv = is_valid_state(mdev, ns);
1007 if (rv < SS_SUCCESS) {
1008 /* If the old state was illegal as well, then let
1011 if (is_valid_state(mdev, os) == rv)
1012 rv = is_valid_state_transition(mdev, ns, os);
1014 rv = is_valid_state_transition(mdev, ns, os);
1017 if (rv < SS_SUCCESS) {
1018 if (flags & CS_VERBOSE)
1019 print_st_err(mdev, os, ns, rv);
1023 if (warn_sync_abort)
1024 dev_warn(DEV, "%s aborted.\n", warn_sync_abort);
1035 if (is_susp(ns) != is_susp(os))
1036 pbp += sprintf(pbp, "susp( %s -> %s ) ",
1037 drbd_susp_str(is_susp(os)),
1038 drbd_susp_str(is_susp(ns)));
1042 dev_info(DEV, "%s\n", pb);
1045 /* solve the race between becoming unconfigured,
1046 * worker doing the cleanup, and
1047 * admin reconfiguring us:
1048 * on (re)configure, first set CONFIG_PENDING,
1049 * then wait for a potentially exiting worker,
1050 * start the worker, and schedule one no_op.
1051 * then proceed with configuration.
1053 if (ns.disk == D_DISKLESS &&
1054 ns.conn == C_STANDALONE &&
1055 ns.role == R_SECONDARY &&
1056 !test_and_set_bit(CONFIG_PENDING, &mdev->flags))
1057 set_bit(DEVICE_DYING, &mdev->flags);
1059 mdev->state.i = ns.i;
1060 wake_up(&mdev->misc_wait);
1061 wake_up(&mdev->state_wait);
1063 /* aborted verify run. log the last position */
1064 if ((os.conn == C_VERIFY_S || os.conn == C_VERIFY_T) &&
1065 ns.conn < C_CONNECTED) {
1066 mdev->ov_start_sector =
1067 BM_BIT_TO_SECT(mdev->rs_total - mdev->ov_left);
1068 dev_info(DEV, "Online Verify reached sector %llu\n",
1069 (unsigned long long)mdev->ov_start_sector);
1072 if ((os.conn == C_PAUSED_SYNC_T || os.conn == C_PAUSED_SYNC_S) &&
1073 (ns.conn == C_SYNC_TARGET || ns.conn == C_SYNC_SOURCE)) {
1074 dev_info(DEV, "Syncer continues.\n");
1075 mdev->rs_paused += (long)jiffies
1076 -(long)mdev->rs_mark_time[mdev->rs_last_mark];
1077 if (ns.conn == C_SYNC_TARGET)
1078 mod_timer(&mdev->resync_timer, jiffies);
1081 if ((os.conn == C_SYNC_TARGET || os.conn == C_SYNC_SOURCE) &&
1082 (ns.conn == C_PAUSED_SYNC_T || ns.conn == C_PAUSED_SYNC_S)) {
1083 dev_info(DEV, "Resync suspended\n");
1084 mdev->rs_mark_time[mdev->rs_last_mark] = jiffies;
1087 if (os.conn == C_CONNECTED &&
1088 (ns.conn == C_VERIFY_S || ns.conn == C_VERIFY_T)) {
1089 unsigned long now = jiffies;
1092 mdev->ov_position = 0;
1093 mdev->rs_total = drbd_bm_bits(mdev);
1094 if (mdev->agreed_pro_version >= 90)
1095 set_ov_position(mdev, ns.conn);
1097 mdev->ov_start_sector = 0;
1098 mdev->ov_left = mdev->rs_total
1099 - BM_SECT_TO_BIT(mdev->ov_position);
1100 mdev->rs_start = now;
1101 mdev->rs_last_events = 0;
1102 mdev->rs_last_sect_ev = 0;
1103 mdev->ov_last_oos_size = 0;
1104 mdev->ov_last_oos_start = 0;
1106 for (i = 0; i < DRBD_SYNC_MARKS; i++) {
1107 mdev->rs_mark_left[i] = mdev->rs_total;
1108 mdev->rs_mark_time[i] = now;
1111 if (ns.conn == C_VERIFY_S) {
1112 dev_info(DEV, "Starting Online Verify from sector %llu\n",
1113 (unsigned long long)mdev->ov_position);
1114 mod_timer(&mdev->resync_timer, jiffies);
1118 if (get_ldev(mdev)) {
1119 u32 mdf = mdev->ldev->md.flags & ~(MDF_CONSISTENT|MDF_PRIMARY_IND|
1120 MDF_CONNECTED_IND|MDF_WAS_UP_TO_DATE|
1121 MDF_PEER_OUT_DATED|MDF_CRASHED_PRIMARY);
1123 if (test_bit(CRASHED_PRIMARY, &mdev->flags))
1124 mdf |= MDF_CRASHED_PRIMARY;
1125 if (mdev->state.role == R_PRIMARY ||
1126 (mdev->state.pdsk < D_INCONSISTENT && mdev->state.peer == R_PRIMARY))
1127 mdf |= MDF_PRIMARY_IND;
1128 if (mdev->state.conn > C_WF_REPORT_PARAMS)
1129 mdf |= MDF_CONNECTED_IND;
1130 if (mdev->state.disk > D_INCONSISTENT)
1131 mdf |= MDF_CONSISTENT;
1132 if (mdev->state.disk > D_OUTDATED)
1133 mdf |= MDF_WAS_UP_TO_DATE;
1134 if (mdev->state.pdsk <= D_OUTDATED && mdev->state.pdsk >= D_INCONSISTENT)
1135 mdf |= MDF_PEER_OUT_DATED;
1136 if (mdf != mdev->ldev->md.flags) {
1137 mdev->ldev->md.flags = mdf;
1138 drbd_md_mark_dirty(mdev);
1140 if (os.disk < D_CONSISTENT && ns.disk >= D_CONSISTENT)
1141 drbd_set_ed_uuid(mdev, mdev->ldev->md.uuid[UI_CURRENT]);
1145 /* Peer was forced D_UP_TO_DATE & R_PRIMARY, consider to resync */
1146 if (os.disk == D_INCONSISTENT && os.pdsk == D_INCONSISTENT &&
1147 os.peer == R_SECONDARY && ns.peer == R_PRIMARY)
1148 set_bit(CONSIDER_RESYNC, &mdev->flags);
1150 /* Receiver should clean up itself */
1151 if (os.conn != C_DISCONNECTING && ns.conn == C_DISCONNECTING)
1152 drbd_thread_stop_nowait(&mdev->receiver);
1154 /* Now the receiver finished cleaning up itself, it should die */
1155 if (os.conn != C_STANDALONE && ns.conn == C_STANDALONE)
1156 drbd_thread_stop_nowait(&mdev->receiver);
1158 /* Upon network failure, we need to restart the receiver. */
1159 if (os.conn > C_TEAR_DOWN &&
1160 ns.conn <= C_TEAR_DOWN && ns.conn >= C_TIMEOUT)
1161 drbd_thread_restart_nowait(&mdev->receiver);
1163 /* Resume AL writing if we get a connection */
1164 if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED)
1165 drbd_resume_al(mdev);
1167 ascw = kmalloc(sizeof(*ascw), GFP_ATOMIC);
1171 ascw->flags = flags;
1172 ascw->w.cb = w_after_state_ch;
1174 drbd_queue_work(&mdev->data.work, &ascw->w);
1176 dev_warn(DEV, "Could not kmalloc an ascw\n");
1182 static int w_after_state_ch(struct drbd_conf *mdev, struct drbd_work *w, int unused)
1184 struct after_state_chg_work *ascw =
1185 container_of(w, struct after_state_chg_work, w);
1186 after_state_ch(mdev, ascw->os, ascw->ns, ascw->flags);
1187 if (ascw->flags & CS_WAIT_COMPLETE) {
1188 D_ASSERT(ascw->done != NULL);
1189 complete(ascw->done);
1196 static void abw_start_sync(struct drbd_conf *mdev, int rv)
1199 dev_err(DEV, "Writing the bitmap failed not starting resync.\n");
1200 _drbd_request_state(mdev, NS(conn, C_CONNECTED), CS_VERBOSE);
1204 switch (mdev->state.conn) {
1205 case C_STARTING_SYNC_T:
1206 _drbd_request_state(mdev, NS(conn, C_WF_SYNC_UUID), CS_VERBOSE);
1208 case C_STARTING_SYNC_S:
1209 drbd_start_resync(mdev, C_SYNC_SOURCE);
1215 * after_state_ch() - Perform after state change actions that may sleep
1216 * @mdev: DRBD device.
1221 static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
1222 union drbd_state ns, enum chg_state_flags flags)
1224 enum drbd_fencing_p fp;
1225 enum drbd_req_event what = nothing;
1226 union drbd_state nsm = (union drbd_state){ .i = -1 };
1228 if (os.conn != C_CONNECTED && ns.conn == C_CONNECTED) {
1229 clear_bit(CRASHED_PRIMARY, &mdev->flags);
1231 mdev->p_uuid[UI_FLAGS] &= ~((u64)2);
1235 if (get_ldev(mdev)) {
1236 fp = mdev->ldev->dc.fencing;
1240 /* Inform userspace about the change... */
1241 drbd_bcast_state(mdev, ns);
1243 if (!(os.role == R_PRIMARY && os.disk < D_UP_TO_DATE && os.pdsk < D_UP_TO_DATE) &&
1244 (ns.role == R_PRIMARY && ns.disk < D_UP_TO_DATE && ns.pdsk < D_UP_TO_DATE))
1245 drbd_khelper(mdev, "pri-on-incon-degr");
1247 /* Here we have the actions that are performed after a
1248 state change. This function might sleep */
1252 if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED) {
1253 if (ns.conn == C_CONNECTED)
1254 what = resend, nsm.susp_nod = 0;
1255 else /* ns.conn > C_CONNECTED */
1256 dev_err(DEV, "Unexpected Resynd going on!\n");
1259 if (os.disk == D_ATTACHING && ns.disk > D_ATTACHING)
1260 what = restart_frozen_disk_io, nsm.susp_nod = 0;
1265 /* case1: The outdate peer handler is successful: */
1266 if (os.pdsk > D_OUTDATED && ns.pdsk <= D_OUTDATED) {
1268 if (test_bit(NEW_CUR_UUID, &mdev->flags)) {
1269 drbd_uuid_new_current(mdev);
1270 clear_bit(NEW_CUR_UUID, &mdev->flags);
1273 spin_lock_irq(&mdev->req_lock);
1274 _drbd_set_state(_NS(mdev, susp_fen, 0), CS_VERBOSE, NULL);
1275 spin_unlock_irq(&mdev->req_lock);
1277 /* case2: The connection was established again: */
1278 if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED) {
1279 clear_bit(NEW_CUR_UUID, &mdev->flags);
1285 if (what != nothing) {
1286 spin_lock_irq(&mdev->req_lock);
1287 _tl_restart(mdev, what);
1288 nsm.i &= mdev->state.i;
1289 _drbd_set_state(mdev, nsm, CS_VERBOSE, NULL);
1290 spin_unlock_irq(&mdev->req_lock);
1293 /* Do not change the order of the if above and the two below... */
1294 if (os.pdsk == D_DISKLESS && ns.pdsk > D_DISKLESS) { /* attach on the peer */
1295 drbd_send_uuids(mdev);
1296 drbd_send_state(mdev);
1298 if (os.conn != C_WF_BITMAP_S && ns.conn == C_WF_BITMAP_S)
1299 drbd_queue_bitmap_io(mdev, &drbd_send_bitmap, NULL, "send_bitmap (WFBitMapS)");
1301 /* Lost contact to peer's copy of the data */
1302 if ((os.pdsk >= D_INCONSISTENT &&
1303 os.pdsk != D_UNKNOWN &&
1304 os.pdsk != D_OUTDATED)
1305 && (ns.pdsk < D_INCONSISTENT ||
1306 ns.pdsk == D_UNKNOWN ||
1307 ns.pdsk == D_OUTDATED)) {
1308 if (get_ldev(mdev)) {
1309 if ((ns.role == R_PRIMARY || ns.peer == R_PRIMARY) &&
1310 mdev->ldev->md.uuid[UI_BITMAP] == 0 && ns.disk >= D_UP_TO_DATE) {
1311 if (is_susp(mdev->state)) {
1312 set_bit(NEW_CUR_UUID, &mdev->flags);
1314 drbd_uuid_new_current(mdev);
1315 drbd_send_uuids(mdev);
1322 if (ns.pdsk < D_INCONSISTENT && get_ldev(mdev)) {
1323 if (ns.peer == R_PRIMARY && mdev->ldev->md.uuid[UI_BITMAP] == 0) {
1324 drbd_uuid_new_current(mdev);
1325 drbd_send_uuids(mdev);
1328 /* D_DISKLESS Peer becomes secondary */
1329 if (os.peer == R_PRIMARY && ns.peer == R_SECONDARY)
1330 drbd_al_to_on_disk_bm(mdev);
1334 /* Last part of the attaching process ... */
1335 if (ns.conn >= C_CONNECTED &&
1336 os.disk == D_ATTACHING && ns.disk == D_NEGOTIATING) {
1337 drbd_send_sizes(mdev, 0, 0); /* to start sync... */
1338 drbd_send_uuids(mdev);
1339 drbd_send_state(mdev);
1342 /* We want to pause/continue resync, tell peer. */
1343 if (ns.conn >= C_CONNECTED &&
1344 ((os.aftr_isp != ns.aftr_isp) ||
1345 (os.user_isp != ns.user_isp)))
1346 drbd_send_state(mdev);
1348 /* In case one of the isp bits got set, suspend other devices. */
1349 if ((!os.aftr_isp && !os.peer_isp && !os.user_isp) &&
1350 (ns.aftr_isp || ns.peer_isp || ns.user_isp))
1351 suspend_other_sg(mdev);
1353 /* Make sure the peer gets informed about eventual state
1354 changes (ISP bits) while we were in WFReportParams. */
1355 if (os.conn == C_WF_REPORT_PARAMS && ns.conn >= C_CONNECTED)
1356 drbd_send_state(mdev);
1358 /* We are in the progress to start a full sync... */
1359 if ((os.conn != C_STARTING_SYNC_T && ns.conn == C_STARTING_SYNC_T) ||
1360 (os.conn != C_STARTING_SYNC_S && ns.conn == C_STARTING_SYNC_S))
1361 drbd_queue_bitmap_io(mdev, &drbd_bmio_set_n_write, &abw_start_sync, "set_n_write from StartingSync");
1363 /* We are invalidating our self... */
1364 if (os.conn < C_CONNECTED && ns.conn < C_CONNECTED &&
1365 os.disk > D_INCONSISTENT && ns.disk == D_INCONSISTENT)
1366 drbd_queue_bitmap_io(mdev, &drbd_bmio_set_n_write, NULL, "set_n_write from invalidate");
1368 /* first half of local IO error */
1369 if (os.disk > D_FAILED && ns.disk == D_FAILED) {
1370 enum drbd_io_error_p eh = EP_PASS_ON;
1372 if (drbd_send_state(mdev))
1373 dev_warn(DEV, "Notified peer that my disk is broken.\n");
1375 dev_err(DEV, "Sending state for drbd_io_error() failed\n");
1377 drbd_rs_cancel_all(mdev);
1379 if (get_ldev_if_state(mdev, D_FAILED)) {
1380 eh = mdev->ldev->dc.on_io_error;
1383 if (eh == EP_CALL_HELPER)
1384 drbd_khelper(mdev, "local-io-error");
1388 /* second half of local IO error handling,
1389 * after local_cnt references have reached zero: */
1390 if (os.disk == D_FAILED && ns.disk == D_DISKLESS) {
1392 mdev->rs_failed = 0;
1393 atomic_set(&mdev->rs_pending_cnt, 0);
1396 if (os.disk > D_DISKLESS && ns.disk == D_DISKLESS) {
1397 /* We must still be diskless,
1398 * re-attach has to be serialized with this! */
1399 if (mdev->state.disk != D_DISKLESS)
1401 "ASSERT FAILED: disk is %s while going diskless\n",
1402 drbd_disk_str(mdev->state.disk));
1404 /* we cannot assert local_cnt == 0 here, as get_ldev_if_state
1405 * will inc/dec it frequently. Since we became D_DISKLESS, no
1406 * one has touched the protected members anymore, though, so we
1407 * are safe to free them here. */
1408 if (drbd_send_state(mdev))
1409 dev_warn(DEV, "Notified peer that I detached my disk.\n");
1411 dev_err(DEV, "Sending state for detach failed\n");
1413 lc_destroy(mdev->resync);
1414 mdev->resync = NULL;
1415 lc_destroy(mdev->act_log);
1416 mdev->act_log = NULL;
1418 drbd_free_bc(mdev->ldev);
1419 mdev->ldev = NULL;);
1421 if (mdev->md_io_tmpp) {
1422 __free_page(mdev->md_io_tmpp);
1423 mdev->md_io_tmpp = NULL;
1427 /* Disks got bigger while they were detached */
1428 if (ns.disk > D_NEGOTIATING && ns.pdsk > D_NEGOTIATING &&
1429 test_and_clear_bit(RESYNC_AFTER_NEG, &mdev->flags)) {
1430 if (ns.conn == C_CONNECTED)
1431 resync_after_online_grow(mdev);
1434 /* A resync finished or aborted, wake paused devices... */
1435 if ((os.conn > C_CONNECTED && ns.conn <= C_CONNECTED) ||
1436 (os.peer_isp && !ns.peer_isp) ||
1437 (os.user_isp && !ns.user_isp))
1438 resume_next_sg(mdev);
1440 /* sync target done with resync. Explicitly notify peer, even though
1441 * it should (at least for non-empty resyncs) already know itself. */
1442 if (os.disk < D_UP_TO_DATE && os.conn >= C_SYNC_SOURCE && ns.conn == C_CONNECTED)
1443 drbd_send_state(mdev);
1445 /* free tl_hash if we Got thawed and are C_STANDALONE */
1446 if (ns.conn == C_STANDALONE && !is_susp(ns) && mdev->tl_hash)
1447 drbd_free_tl_hash(mdev);
1449 /* Upon network connection, we need to start the receiver */
1450 if (os.conn == C_STANDALONE && ns.conn == C_UNCONNECTED)
1451 drbd_thread_start(&mdev->receiver);
1453 /* Terminate worker thread if we are unconfigured - it will be
1454 restarted as needed... */
1455 if (ns.disk == D_DISKLESS &&
1456 ns.conn == C_STANDALONE &&
1457 ns.role == R_SECONDARY) {
1458 if (os.aftr_isp != ns.aftr_isp)
1459 resume_next_sg(mdev);
1460 /* set in __drbd_set_state, unless CONFIG_PENDING was set */
1461 if (test_bit(DEVICE_DYING, &mdev->flags))
1462 drbd_thread_stop_nowait(&mdev->worker);
1469 static int drbd_thread_setup(void *arg)
1471 struct drbd_thread *thi = (struct drbd_thread *) arg;
1472 struct drbd_conf *mdev = thi->mdev;
1473 unsigned long flags;
1477 retval = thi->function(thi);
1479 spin_lock_irqsave(&thi->t_lock, flags);
1481 /* if the receiver has been "Exiting", the last thing it did
1482 * was set the conn state to "StandAlone",
1483 * if now a re-connect request comes in, conn state goes C_UNCONNECTED,
1484 * and receiver thread will be "started".
1485 * drbd_thread_start needs to set "Restarting" in that case.
1486 * t_state check and assignment needs to be within the same spinlock,
1487 * so either thread_start sees Exiting, and can remap to Restarting,
1488 * or thread_start see None, and can proceed as normal.
1491 if (thi->t_state == Restarting) {
1492 dev_info(DEV, "Restarting %s\n", current->comm);
1493 thi->t_state = Running;
1494 spin_unlock_irqrestore(&thi->t_lock, flags);
1499 thi->t_state = None;
1501 complete(&thi->stop);
1502 spin_unlock_irqrestore(&thi->t_lock, flags);
1504 dev_info(DEV, "Terminating %s\n", current->comm);
1506 /* Release mod reference taken when thread was started */
1507 module_put(THIS_MODULE);
1511 static void drbd_thread_init(struct drbd_conf *mdev, struct drbd_thread *thi,
1512 int (*func) (struct drbd_thread *))
1514 spin_lock_init(&thi->t_lock);
1516 thi->t_state = None;
1517 thi->function = func;
1521 int drbd_thread_start(struct drbd_thread *thi)
1523 struct drbd_conf *mdev = thi->mdev;
1524 struct task_struct *nt;
1525 unsigned long flags;
1528 thi == &mdev->receiver ? "receiver" :
1529 thi == &mdev->asender ? "asender" :
1530 thi == &mdev->worker ? "worker" : "NONSENSE";
1532 /* is used from state engine doing drbd_thread_stop_nowait,
1533 * while holding the req lock irqsave */
1534 spin_lock_irqsave(&thi->t_lock, flags);
1536 switch (thi->t_state) {
1538 dev_info(DEV, "Starting %s thread (from %s [%d])\n",
1539 me, current->comm, current->pid);
1541 /* Get ref on module for thread - this is released when thread exits */
1542 if (!try_module_get(THIS_MODULE)) {
1543 dev_err(DEV, "Failed to get module reference in drbd_thread_start\n");
1544 spin_unlock_irqrestore(&thi->t_lock, flags);
1548 init_completion(&thi->stop);
1549 D_ASSERT(thi->task == NULL);
1550 thi->reset_cpu_mask = 1;
1551 thi->t_state = Running;
1552 spin_unlock_irqrestore(&thi->t_lock, flags);
1553 flush_signals(current); /* otherw. may get -ERESTARTNOINTR */
1555 nt = kthread_create(drbd_thread_setup, (void *) thi,
1556 "drbd%d_%s", mdev_to_minor(mdev), me);
1559 dev_err(DEV, "Couldn't start thread\n");
1561 module_put(THIS_MODULE);
1564 spin_lock_irqsave(&thi->t_lock, flags);
1566 thi->t_state = Running;
1567 spin_unlock_irqrestore(&thi->t_lock, flags);
1568 wake_up_process(nt);
1571 thi->t_state = Restarting;
1572 dev_info(DEV, "Restarting %s thread (from %s [%d])\n",
1573 me, current->comm, current->pid);
1578 spin_unlock_irqrestore(&thi->t_lock, flags);
1586 void _drbd_thread_stop(struct drbd_thread *thi, int restart, int wait)
1588 unsigned long flags;
1590 enum drbd_thread_state ns = restart ? Restarting : Exiting;
1592 /* may be called from state engine, holding the req lock irqsave */
1593 spin_lock_irqsave(&thi->t_lock, flags);
1595 if (thi->t_state == None) {
1596 spin_unlock_irqrestore(&thi->t_lock, flags);
1598 drbd_thread_start(thi);
1602 if (thi->t_state != ns) {
1603 if (thi->task == NULL) {
1604 spin_unlock_irqrestore(&thi->t_lock, flags);
1610 init_completion(&thi->stop);
1611 if (thi->task != current)
1612 force_sig(DRBD_SIGKILL, thi->task);
1616 spin_unlock_irqrestore(&thi->t_lock, flags);
1619 wait_for_completion(&thi->stop);
1624 * drbd_calc_cpu_mask() - Generate CPU masks, spread over all CPUs
1625 * @mdev: DRBD device.
1627 * Forces all threads of a device onto the same CPU. This is beneficial for
1628 * DRBD's performance. May be overwritten by user's configuration.
1630 void drbd_calc_cpu_mask(struct drbd_conf *mdev)
1634 /* user override. */
1635 if (cpumask_weight(mdev->cpu_mask))
1638 ord = mdev_to_minor(mdev) % cpumask_weight(cpu_online_mask);
1639 for_each_online_cpu(cpu) {
1641 cpumask_set_cpu(cpu, mdev->cpu_mask);
1645 /* should not be reached */
1646 cpumask_setall(mdev->cpu_mask);
1650 * drbd_thread_current_set_cpu() - modifies the cpu mask of the _current_ thread
1651 * @mdev: DRBD device.
1653 * call in the "main loop" of _all_ threads, no need for any mutex, current won't die
1656 void drbd_thread_current_set_cpu(struct drbd_conf *mdev)
1658 struct task_struct *p = current;
1659 struct drbd_thread *thi =
1660 p == mdev->asender.task ? &mdev->asender :
1661 p == mdev->receiver.task ? &mdev->receiver :
1662 p == mdev->worker.task ? &mdev->worker :
1666 if (!thi->reset_cpu_mask)
1668 thi->reset_cpu_mask = 0;
1669 set_cpus_allowed_ptr(p, mdev->cpu_mask);
1673 /* the appropriate socket mutex must be held already */
1674 int _drbd_send_cmd(struct drbd_conf *mdev, struct socket *sock,
1675 enum drbd_packets cmd, struct p_header80 *h,
1676 size_t size, unsigned msg_flags)
1680 ERR_IF(!h) return FALSE;
1681 ERR_IF(!size) return FALSE;
1683 h->magic = BE_DRBD_MAGIC;
1684 h->command = cpu_to_be16(cmd);
1685 h->length = cpu_to_be16(size-sizeof(struct p_header80));
1687 sent = drbd_send(mdev, sock, h, size, msg_flags);
1689 ok = (sent == size);
1691 dev_err(DEV, "short sent %s size=%d sent=%d\n",
1692 cmdname(cmd), (int)size, sent);
1696 /* don't pass the socket. we may only look at it
1697 * when we hold the appropriate socket mutex.
1699 int drbd_send_cmd(struct drbd_conf *mdev, int use_data_socket,
1700 enum drbd_packets cmd, struct p_header80 *h, size_t size)
1703 struct socket *sock;
1705 if (use_data_socket) {
1706 mutex_lock(&mdev->data.mutex);
1707 sock = mdev->data.socket;
1709 mutex_lock(&mdev->meta.mutex);
1710 sock = mdev->meta.socket;
1713 /* drbd_disconnect() could have called drbd_free_sock()
1714 * while we were waiting in down()... */
1715 if (likely(sock != NULL))
1716 ok = _drbd_send_cmd(mdev, sock, cmd, h, size, 0);
1718 if (use_data_socket)
1719 mutex_unlock(&mdev->data.mutex);
1721 mutex_unlock(&mdev->meta.mutex);
1725 int drbd_send_cmd2(struct drbd_conf *mdev, enum drbd_packets cmd, char *data,
1728 struct p_header80 h;
1731 h.magic = BE_DRBD_MAGIC;
1732 h.command = cpu_to_be16(cmd);
1733 h.length = cpu_to_be16(size);
1735 if (!drbd_get_data_sock(mdev))
1739 drbd_send(mdev, mdev->data.socket, &h, sizeof(h), 0));
1741 drbd_send(mdev, mdev->data.socket, data, size, 0));
1743 drbd_put_data_sock(mdev);
1748 int drbd_send_sync_param(struct drbd_conf *mdev, struct syncer_conf *sc)
1750 struct p_rs_param_95 *p;
1751 struct socket *sock;
1753 const int apv = mdev->agreed_pro_version;
1755 size = apv <= 87 ? sizeof(struct p_rs_param)
1756 : apv == 88 ? sizeof(struct p_rs_param)
1757 + strlen(mdev->sync_conf.verify_alg) + 1
1758 : apv <= 94 ? sizeof(struct p_rs_param_89)
1759 : /* apv >= 95 */ sizeof(struct p_rs_param_95);
1761 /* used from admin command context and receiver/worker context.
1762 * to avoid kmalloc, grab the socket right here,
1763 * then use the pre-allocated sbuf there */
1764 mutex_lock(&mdev->data.mutex);
1765 sock = mdev->data.socket;
1767 if (likely(sock != NULL)) {
1768 enum drbd_packets cmd = apv >= 89 ? P_SYNC_PARAM89 : P_SYNC_PARAM;
1770 p = &mdev->data.sbuf.rs_param_95;
1772 /* initialize verify_alg and csums_alg */
1773 memset(p->verify_alg, 0, 2 * SHARED_SECRET_MAX);
1775 p->rate = cpu_to_be32(sc->rate);
1776 p->c_plan_ahead = cpu_to_be32(sc->c_plan_ahead);
1777 p->c_delay_target = cpu_to_be32(sc->c_delay_target);
1778 p->c_fill_target = cpu_to_be32(sc->c_fill_target);
1779 p->c_max_rate = cpu_to_be32(sc->c_max_rate);
1782 strcpy(p->verify_alg, mdev->sync_conf.verify_alg);
1784 strcpy(p->csums_alg, mdev->sync_conf.csums_alg);
1786 rv = _drbd_send_cmd(mdev, sock, cmd, &p->head, size, 0);
1788 rv = 0; /* not ok */
1790 mutex_unlock(&mdev->data.mutex);
1795 int drbd_send_protocol(struct drbd_conf *mdev)
1797 struct p_protocol *p;
1800 size = sizeof(struct p_protocol);
1802 if (mdev->agreed_pro_version >= 87)
1803 size += strlen(mdev->net_conf->integrity_alg) + 1;
1805 /* we must not recurse into our own queue,
1806 * as that is blocked during handshake */
1807 p = kmalloc(size, GFP_NOIO);
1811 p->protocol = cpu_to_be32(mdev->net_conf->wire_protocol);
1812 p->after_sb_0p = cpu_to_be32(mdev->net_conf->after_sb_0p);
1813 p->after_sb_1p = cpu_to_be32(mdev->net_conf->after_sb_1p);
1814 p->after_sb_2p = cpu_to_be32(mdev->net_conf->after_sb_2p);
1815 p->two_primaries = cpu_to_be32(mdev->net_conf->two_primaries);
1818 if (mdev->net_conf->want_lose)
1820 if (mdev->net_conf->dry_run) {
1821 if (mdev->agreed_pro_version >= 92)
1824 dev_err(DEV, "--dry-run is not supported by peer");
1829 p->conn_flags = cpu_to_be32(cf);
1831 if (mdev->agreed_pro_version >= 87)
1832 strcpy(p->integrity_alg, mdev->net_conf->integrity_alg);
1834 rv = drbd_send_cmd(mdev, USE_DATA_SOCKET, P_PROTOCOL,
1835 (struct p_header80 *)p, size);
1840 int _drbd_send_uuids(struct drbd_conf *mdev, u64 uuid_flags)
1845 if (!get_ldev_if_state(mdev, D_NEGOTIATING))
1848 for (i = UI_CURRENT; i < UI_SIZE; i++)
1849 p.uuid[i] = mdev->ldev ? cpu_to_be64(mdev->ldev->md.uuid[i]) : 0;
1851 mdev->comm_bm_set = drbd_bm_total_weight(mdev);
1852 p.uuid[UI_SIZE] = cpu_to_be64(mdev->comm_bm_set);
1853 uuid_flags |= mdev->net_conf->want_lose ? 1 : 0;
1854 uuid_flags |= test_bit(CRASHED_PRIMARY, &mdev->flags) ? 2 : 0;
1855 uuid_flags |= mdev->new_state_tmp.disk == D_INCONSISTENT ? 4 : 0;
1856 p.uuid[UI_FLAGS] = cpu_to_be64(uuid_flags);
1860 return drbd_send_cmd(mdev, USE_DATA_SOCKET, P_UUIDS,
1861 (struct p_header80 *)&p, sizeof(p));
1864 int drbd_send_uuids(struct drbd_conf *mdev)
1866 return _drbd_send_uuids(mdev, 0);
1869 int drbd_send_uuids_skip_initial_sync(struct drbd_conf *mdev)
1871 return _drbd_send_uuids(mdev, 8);
1875 int drbd_send_sync_uuid(struct drbd_conf *mdev, u64 val)
1879 p.uuid = cpu_to_be64(val);
1881 return drbd_send_cmd(mdev, USE_DATA_SOCKET, P_SYNC_UUID,
1882 (struct p_header80 *)&p, sizeof(p));
1885 int drbd_send_sizes(struct drbd_conf *mdev, int trigger_reply, enum dds_flags flags)
1888 sector_t d_size, u_size;
1892 if (get_ldev_if_state(mdev, D_NEGOTIATING)) {
1893 D_ASSERT(mdev->ldev->backing_bdev);
1894 d_size = drbd_get_max_capacity(mdev->ldev);
1895 u_size = mdev->ldev->dc.disk_size;
1896 q_order_type = drbd_queue_order_type(mdev);
1901 q_order_type = QUEUE_ORDERED_NONE;
1904 p.d_size = cpu_to_be64(d_size);
1905 p.u_size = cpu_to_be64(u_size);
1906 p.c_size = cpu_to_be64(trigger_reply ? 0 : drbd_get_capacity(mdev->this_bdev));
1907 p.max_segment_size = cpu_to_be32(queue_max_segment_size(mdev->rq_queue));
1908 p.queue_order_type = cpu_to_be16(q_order_type);
1909 p.dds_flags = cpu_to_be16(flags);
1911 ok = drbd_send_cmd(mdev, USE_DATA_SOCKET, P_SIZES,
1912 (struct p_header80 *)&p, sizeof(p));
1917 * drbd_send_state() - Sends the drbd state to the peer
1918 * @mdev: DRBD device.
1920 int drbd_send_state(struct drbd_conf *mdev)
1922 struct socket *sock;
1926 /* Grab state lock so we wont send state if we're in the middle
1927 * of a cluster wide state change on another thread */
1928 drbd_state_lock(mdev);
1930 mutex_lock(&mdev->data.mutex);
1932 p.state = cpu_to_be32(mdev->state.i); /* Within the send mutex */
1933 sock = mdev->data.socket;
1935 if (likely(sock != NULL)) {
1936 ok = _drbd_send_cmd(mdev, sock, P_STATE,
1937 (struct p_header80 *)&p, sizeof(p), 0);
1940 mutex_unlock(&mdev->data.mutex);
1942 drbd_state_unlock(mdev);
1946 int drbd_send_state_req(struct drbd_conf *mdev,
1947 union drbd_state mask, union drbd_state val)
1949 struct p_req_state p;
1951 p.mask = cpu_to_be32(mask.i);
1952 p.val = cpu_to_be32(val.i);
1954 return drbd_send_cmd(mdev, USE_DATA_SOCKET, P_STATE_CHG_REQ,
1955 (struct p_header80 *)&p, sizeof(p));
1958 int drbd_send_sr_reply(struct drbd_conf *mdev, int retcode)
1960 struct p_req_state_reply p;
1962 p.retcode = cpu_to_be32(retcode);
1964 return drbd_send_cmd(mdev, USE_META_SOCKET, P_STATE_CHG_REPLY,
1965 (struct p_header80 *)&p, sizeof(p));
1968 int fill_bitmap_rle_bits(struct drbd_conf *mdev,
1969 struct p_compressed_bm *p,
1970 struct bm_xfer_ctx *c)
1972 struct bitstream bs;
1973 unsigned long plain_bits;
1980 /* may we use this feature? */
1981 if ((mdev->sync_conf.use_rle == 0) ||
1982 (mdev->agreed_pro_version < 90))
1985 if (c->bit_offset >= c->bm_bits)
1986 return 0; /* nothing to do. */
1988 /* use at most thus many bytes */
1989 bitstream_init(&bs, p->code, BM_PACKET_VLI_BYTES_MAX, 0);
1990 memset(p->code, 0, BM_PACKET_VLI_BYTES_MAX);
1991 /* plain bits covered in this code string */
1994 /* p->encoding & 0x80 stores whether the first run length is set.
1995 * bit offset is implicit.
1996 * start with toggle == 2 to be able to tell the first iteration */
1999 /* see how much plain bits we can stuff into one packet
2000 * using RLE and VLI. */
2002 tmp = (toggle == 0) ? _drbd_bm_find_next_zero(mdev, c->bit_offset)
2003 : _drbd_bm_find_next(mdev, c->bit_offset);
2006 rl = tmp - c->bit_offset;
2008 if (toggle == 2) { /* first iteration */
2010 /* the first checked bit was set,
2011 * store start value, */
2012 DCBP_set_start(p, 1);
2013 /* but skip encoding of zero run length */
2017 DCBP_set_start(p, 0);
2020 /* paranoia: catch zero runlength.
2021 * can only happen if bitmap is modified while we scan it. */
2023 dev_err(DEV, "unexpected zero runlength while encoding bitmap "
2024 "t:%u bo:%lu\n", toggle, c->bit_offset);
2028 bits = vli_encode_bits(&bs, rl);
2029 if (bits == -ENOBUFS) /* buffer full */
2032 dev_err(DEV, "error while encoding bitmap: %d\n", bits);
2038 c->bit_offset = tmp;
2039 } while (c->bit_offset < c->bm_bits);
2041 len = bs.cur.b - p->code + !!bs.cur.bit;
2043 if (plain_bits < (len << 3)) {
2044 /* incompressible with this method.
2045 * we need to rewind both word and bit position. */
2046 c->bit_offset -= plain_bits;
2047 bm_xfer_ctx_bit_to_word_offset(c);
2048 c->bit_offset = c->word_offset * BITS_PER_LONG;
2052 /* RLE + VLI was able to compress it just fine.
2053 * update c->word_offset. */
2054 bm_xfer_ctx_bit_to_word_offset(c);
2056 /* store pad_bits */
2057 DCBP_set_pad_bits(p, (8 - bs.cur.bit) & 0x7);
2062 enum { OK, FAILED, DONE }
2063 send_bitmap_rle_or_plain(struct drbd_conf *mdev,
2064 struct p_header80 *h, struct bm_xfer_ctx *c)
2066 struct p_compressed_bm *p = (void*)h;
2067 unsigned long num_words;
2071 len = fill_bitmap_rle_bits(mdev, p, c);
2077 DCBP_set_code(p, RLE_VLI_Bits);
2078 ok = _drbd_send_cmd(mdev, mdev->data.socket, P_COMPRESSED_BITMAP, h,
2079 sizeof(*p) + len, 0);
2082 c->bytes[0] += sizeof(*p) + len;
2084 if (c->bit_offset >= c->bm_bits)
2087 /* was not compressible.
2088 * send a buffer full of plain text bits instead. */
2089 num_words = min_t(size_t, BM_PACKET_WORDS, c->bm_words - c->word_offset);
2090 len = num_words * sizeof(long);
2092 drbd_bm_get_lel(mdev, c->word_offset, num_words, (unsigned long*)h->payload);
2093 ok = _drbd_send_cmd(mdev, mdev->data.socket, P_BITMAP,
2094 h, sizeof(struct p_header80) + len, 0);
2095 c->word_offset += num_words;
2096 c->bit_offset = c->word_offset * BITS_PER_LONG;
2099 c->bytes[1] += sizeof(struct p_header80) + len;
2101 if (c->bit_offset > c->bm_bits)
2102 c->bit_offset = c->bm_bits;
2104 ok = ok ? ((len == 0) ? DONE : OK) : FAILED;
2107 INFO_bm_xfer_stats(mdev, "send", c);
2111 /* See the comment at receive_bitmap() */
2112 int _drbd_send_bitmap(struct drbd_conf *mdev)
2114 struct bm_xfer_ctx c;
2115 struct p_header80 *p;
2118 ERR_IF(!mdev->bitmap) return FALSE;
2120 /* maybe we should use some per thread scratch page,
2121 * and allocate that during initial device creation? */
2122 p = (struct p_header80 *) __get_free_page(GFP_NOIO);
2124 dev_err(DEV, "failed to allocate one page buffer in %s\n", __func__);
2128 if (get_ldev(mdev)) {
2129 if (drbd_md_test_flag(mdev->ldev, MDF_FULL_SYNC)) {
2130 dev_info(DEV, "Writing the whole bitmap, MDF_FullSync was set.\n");
2131 drbd_bm_set_all(mdev);
2132 if (drbd_bm_write(mdev)) {
2133 /* write_bm did fail! Leave full sync flag set in Meta P_DATA
2134 * but otherwise process as per normal - need to tell other
2135 * side that a full resync is required! */
2136 dev_err(DEV, "Failed to write bitmap to disk!\n");
2138 drbd_md_clear_flag(mdev, MDF_FULL_SYNC);
2145 c = (struct bm_xfer_ctx) {
2146 .bm_bits = drbd_bm_bits(mdev),
2147 .bm_words = drbd_bm_words(mdev),
2151 ret = send_bitmap_rle_or_plain(mdev, p, &c);
2152 } while (ret == OK);
2154 free_page((unsigned long) p);
2155 return (ret == DONE);
2158 int drbd_send_bitmap(struct drbd_conf *mdev)
2162 if (!drbd_get_data_sock(mdev))
2164 err = !_drbd_send_bitmap(mdev);
2165 drbd_put_data_sock(mdev);
2169 int drbd_send_b_ack(struct drbd_conf *mdev, u32 barrier_nr, u32 set_size)
2172 struct p_barrier_ack p;
2174 p.barrier = barrier_nr;
2175 p.set_size = cpu_to_be32(set_size);
2177 if (mdev->state.conn < C_CONNECTED)
2179 ok = drbd_send_cmd(mdev, USE_META_SOCKET, P_BARRIER_ACK,
2180 (struct p_header80 *)&p, sizeof(p));
2185 * _drbd_send_ack() - Sends an ack packet
2186 * @mdev: DRBD device.
2187 * @cmd: Packet command code.
2188 * @sector: sector, needs to be in big endian byte order
2189 * @blksize: size in byte, needs to be in big endian byte order
2190 * @block_id: Id, big endian byte order
2192 static int _drbd_send_ack(struct drbd_conf *mdev, enum drbd_packets cmd,
2198 struct p_block_ack p;
2201 p.block_id = block_id;
2202 p.blksize = blksize;
2203 p.seq_num = cpu_to_be32(atomic_add_return(1, &mdev->packet_seq));
2205 if (!mdev->meta.socket || mdev->state.conn < C_CONNECTED)
2207 ok = drbd_send_cmd(mdev, USE_META_SOCKET, cmd,
2208 (struct p_header80 *)&p, sizeof(p));
2212 /* dp->sector and dp->block_id already/still in network byte order,
2213 * data_size is payload size according to dp->head,
2214 * and may need to be corrected for digest size. */
2215 int drbd_send_ack_dp(struct drbd_conf *mdev, enum drbd_packets cmd,
2216 struct p_data *dp, int data_size)
2218 data_size -= (mdev->agreed_pro_version >= 87 && mdev->integrity_r_tfm) ?
2219 crypto_hash_digestsize(mdev->integrity_r_tfm) : 0;
2220 return _drbd_send_ack(mdev, cmd, dp->sector, cpu_to_be32(data_size),
2224 int drbd_send_ack_rp(struct drbd_conf *mdev, enum drbd_packets cmd,
2225 struct p_block_req *rp)
2227 return _drbd_send_ack(mdev, cmd, rp->sector, rp->blksize, rp->block_id);
2231 * drbd_send_ack() - Sends an ack packet
2232 * @mdev: DRBD device.
2233 * @cmd: Packet command code.
2236 int drbd_send_ack(struct drbd_conf *mdev,
2237 enum drbd_packets cmd, struct drbd_epoch_entry *e)
2239 return _drbd_send_ack(mdev, cmd,
2240 cpu_to_be64(e->sector),
2241 cpu_to_be32(e->size),
2245 /* This function misuses the block_id field to signal if the blocks
2246 * are is sync or not. */
2247 int drbd_send_ack_ex(struct drbd_conf *mdev, enum drbd_packets cmd,
2248 sector_t sector, int blksize, u64 block_id)
2250 return _drbd_send_ack(mdev, cmd,
2251 cpu_to_be64(sector),
2252 cpu_to_be32(blksize),
2253 cpu_to_be64(block_id));
2256 int drbd_send_drequest(struct drbd_conf *mdev, int cmd,
2257 sector_t sector, int size, u64 block_id)
2260 struct p_block_req p;
2262 p.sector = cpu_to_be64(sector);
2263 p.block_id = block_id;
2264 p.blksize = cpu_to_be32(size);
2266 ok = drbd_send_cmd(mdev, USE_DATA_SOCKET, cmd,
2267 (struct p_header80 *)&p, sizeof(p));
2271 int drbd_send_drequest_csum(struct drbd_conf *mdev,
2272 sector_t sector, int size,
2273 void *digest, int digest_size,
2274 enum drbd_packets cmd)
2277 struct p_block_req p;
2279 p.sector = cpu_to_be64(sector);
2280 p.block_id = BE_DRBD_MAGIC + 0xbeef;
2281 p.blksize = cpu_to_be32(size);
2283 p.head.magic = BE_DRBD_MAGIC;
2284 p.head.command = cpu_to_be16(cmd);
2285 p.head.length = cpu_to_be16(sizeof(p) - sizeof(struct p_header80) + digest_size);
2287 mutex_lock(&mdev->data.mutex);
2289 ok = (sizeof(p) == drbd_send(mdev, mdev->data.socket, &p, sizeof(p), 0));
2290 ok = ok && (digest_size == drbd_send(mdev, mdev->data.socket, digest, digest_size, 0));
2292 mutex_unlock(&mdev->data.mutex);
2297 int drbd_send_ov_request(struct drbd_conf *mdev, sector_t sector, int size)
2300 struct p_block_req p;
2302 p.sector = cpu_to_be64(sector);
2303 p.block_id = BE_DRBD_MAGIC + 0xbabe;
2304 p.blksize = cpu_to_be32(size);
2306 ok = drbd_send_cmd(mdev, USE_DATA_SOCKET, P_OV_REQUEST,
2307 (struct p_header80 *)&p, sizeof(p));
2311 /* called on sndtimeo
2312 * returns FALSE if we should retry,
2313 * TRUE if we think connection is dead
2315 static int we_should_drop_the_connection(struct drbd_conf *mdev, struct socket *sock)
2318 /* long elapsed = (long)(jiffies - mdev->last_received); */
2320 drop_it = mdev->meta.socket == sock
2321 || !mdev->asender.task
2322 || get_t_state(&mdev->asender) != Running
2323 || mdev->state.conn < C_CONNECTED;
2328 drop_it = !--mdev->ko_count;
2330 dev_err(DEV, "[%s/%d] sock_sendmsg time expired, ko = %u\n",
2331 current->comm, current->pid, mdev->ko_count);
2335 return drop_it; /* && (mdev->state == R_PRIMARY) */;
2338 /* The idea of sendpage seems to be to put some kind of reference
2339 * to the page into the skb, and to hand it over to the NIC. In
2340 * this process get_page() gets called.
2342 * As soon as the page was really sent over the network put_page()
2343 * gets called by some part of the network layer. [ NIC driver? ]
2345 * [ get_page() / put_page() increment/decrement the count. If count
2346 * reaches 0 the page will be freed. ]
2348 * This works nicely with pages from FSs.
2349 * But this means that in protocol A we might signal IO completion too early!
2351 * In order not to corrupt data during a resync we must make sure
2352 * that we do not reuse our own buffer pages (EEs) to early, therefore
2353 * we have the net_ee list.
2355 * XFS seems to have problems, still, it submits pages with page_count == 0!
2356 * As a workaround, we disable sendpage on pages
2357 * with page_count == 0 or PageSlab.
2359 static int _drbd_no_send_page(struct drbd_conf *mdev, struct page *page,
2360 int offset, size_t size, unsigned msg_flags)
2362 int sent = drbd_send(mdev, mdev->data.socket, kmap(page) + offset, size, msg_flags);
2365 mdev->send_cnt += size>>9;
2366 return sent == size;
2369 static int _drbd_send_page(struct drbd_conf *mdev, struct page *page,
2370 int offset, size_t size, unsigned msg_flags)
2372 mm_segment_t oldfs = get_fs();
2376 /* e.g. XFS meta- & log-data is in slab pages, which have a
2377 * page_count of 0 and/or have PageSlab() set.
2378 * we cannot use send_page for those, as that does get_page();
2379 * put_page(); and would cause either a VM_BUG directly, or
2380 * __page_cache_release a page that would actually still be referenced
2381 * by someone, leading to some obscure delayed Oops somewhere else. */
2382 if (disable_sendpage || (page_count(page) < 1) || PageSlab(page))
2383 return _drbd_no_send_page(mdev, page, offset, size, msg_flags);
2385 msg_flags |= MSG_NOSIGNAL;
2386 drbd_update_congested(mdev);
2389 sent = mdev->data.socket->ops->sendpage(mdev->data.socket, page,
2392 if (sent == -EAGAIN) {
2393 if (we_should_drop_the_connection(mdev,
2400 dev_warn(DEV, "%s: size=%d len=%d sent=%d\n",
2401 __func__, (int)size, len, sent);
2406 } while (len > 0 /* THINK && mdev->cstate >= C_CONNECTED*/);
2408 clear_bit(NET_CONGESTED, &mdev->flags);
2412 mdev->send_cnt += size>>9;
2416 static int _drbd_send_bio(struct drbd_conf *mdev, struct bio *bio)
2418 struct bio_vec *bvec;
2420 /* hint all but last page with MSG_MORE */
2421 __bio_for_each_segment(bvec, bio, i, 0) {
2422 if (!_drbd_no_send_page(mdev, bvec->bv_page,
2423 bvec->bv_offset, bvec->bv_len,
2424 i == bio->bi_vcnt -1 ? 0 : MSG_MORE))
2430 static int _drbd_send_zc_bio(struct drbd_conf *mdev, struct bio *bio)
2432 struct bio_vec *bvec;
2434 /* hint all but last page with MSG_MORE */
2435 __bio_for_each_segment(bvec, bio, i, 0) {
2436 if (!_drbd_send_page(mdev, bvec->bv_page,
2437 bvec->bv_offset, bvec->bv_len,
2438 i == bio->bi_vcnt -1 ? 0 : MSG_MORE))
2444 static int _drbd_send_zc_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e)
2446 struct page *page = e->pages;
2447 unsigned len = e->size;
2448 /* hint all but last page with MSG_MORE */
2449 page_chain_for_each(page) {
2450 unsigned l = min_t(unsigned, len, PAGE_SIZE);
2451 if (!_drbd_send_page(mdev, page, 0, l,
2452 page_chain_next(page) ? MSG_MORE : 0))
2459 static u32 bio_flags_to_wire(struct drbd_conf *mdev, unsigned long bi_rw)
2461 if (mdev->agreed_pro_version >= 95)
2462 return (bi_rw & REQ_SYNC ? DP_RW_SYNC : 0) |
2463 (bi_rw & REQ_UNPLUG ? DP_UNPLUG : 0) |
2464 (bi_rw & REQ_FUA ? DP_FUA : 0) |
2465 (bi_rw & REQ_FLUSH ? DP_FLUSH : 0) |
2466 (bi_rw & REQ_DISCARD ? DP_DISCARD : 0);
2468 return bi_rw & (REQ_SYNC | REQ_UNPLUG) ? DP_RW_SYNC : 0;
2471 /* Used to send write requests
2472 * R_PRIMARY -> Peer (P_DATA)
2474 int drbd_send_dblock(struct drbd_conf *mdev, struct drbd_request *req)
2478 unsigned int dp_flags = 0;
2482 if (!drbd_get_data_sock(mdev))
2485 dgs = (mdev->agreed_pro_version >= 87 && mdev->integrity_w_tfm) ?
2486 crypto_hash_digestsize(mdev->integrity_w_tfm) : 0;
2488 if (req->size <= DRBD_MAX_SIZE_H80_PACKET) {
2489 p.head.h80.magic = BE_DRBD_MAGIC;
2490 p.head.h80.command = cpu_to_be16(P_DATA);
2492 cpu_to_be16(sizeof(p) - sizeof(union p_header) + dgs + req->size);
2494 p.head.h95.magic = BE_DRBD_MAGIC_BIG;
2495 p.head.h95.command = cpu_to_be16(P_DATA);
2497 cpu_to_be32(sizeof(p) - sizeof(union p_header) + dgs + req->size);
2500 p.sector = cpu_to_be64(req->sector);
2501 p.block_id = (unsigned long)req;
2502 p.seq_num = cpu_to_be32(req->seq_num =
2503 atomic_add_return(1, &mdev->packet_seq));
2505 dp_flags = bio_flags_to_wire(mdev, req->master_bio->bi_rw);
2507 if (mdev->state.conn >= C_SYNC_SOURCE &&
2508 mdev->state.conn <= C_PAUSED_SYNC_T)
2509 dp_flags |= DP_MAY_SET_IN_SYNC;
2511 p.dp_flags = cpu_to_be32(dp_flags);
2512 set_bit(UNPLUG_REMOTE, &mdev->flags);
2514 drbd_send(mdev, mdev->data.socket, &p, sizeof(p), dgs ? MSG_MORE : 0));
2516 dgb = mdev->int_dig_out;
2517 drbd_csum_bio(mdev, mdev->integrity_w_tfm, req->master_bio, dgb);
2518 ok = drbd_send(mdev, mdev->data.socket, dgb, dgs, 0);
2521 if (mdev->net_conf->wire_protocol == DRBD_PROT_A)
2522 ok = _drbd_send_bio(mdev, req->master_bio);
2524 ok = _drbd_send_zc_bio(mdev, req->master_bio);
2527 drbd_put_data_sock(mdev);
2532 /* answer packet, used to send data back for read requests:
2533 * Peer -> (diskless) R_PRIMARY (P_DATA_REPLY)
2534 * C_SYNC_SOURCE -> C_SYNC_TARGET (P_RS_DATA_REPLY)
2536 int drbd_send_block(struct drbd_conf *mdev, enum drbd_packets cmd,
2537 struct drbd_epoch_entry *e)
2544 dgs = (mdev->agreed_pro_version >= 87 && mdev->integrity_w_tfm) ?
2545 crypto_hash_digestsize(mdev->integrity_w_tfm) : 0;
2547 if (e->size <= DRBD_MAX_SIZE_H80_PACKET) {
2548 p.head.h80.magic = BE_DRBD_MAGIC;
2549 p.head.h80.command = cpu_to_be16(cmd);
2551 cpu_to_be16(sizeof(p) - sizeof(struct p_header80) + dgs + e->size);
2553 p.head.h95.magic = BE_DRBD_MAGIC_BIG;
2554 p.head.h95.command = cpu_to_be16(cmd);
2556 cpu_to_be32(sizeof(p) - sizeof(struct p_header80) + dgs + e->size);
2559 p.sector = cpu_to_be64(e->sector);
2560 p.block_id = e->block_id;
2561 /* p.seq_num = 0; No sequence numbers here.. */
2563 /* Only called by our kernel thread.
2564 * This one may be interrupted by DRBD_SIG and/or DRBD_SIGKILL
2565 * in response to admin command or module unload.
2567 if (!drbd_get_data_sock(mdev))
2570 ok = sizeof(p) == drbd_send(mdev, mdev->data.socket, &p, sizeof(p), dgs ? MSG_MORE : 0);
2572 dgb = mdev->int_dig_out;
2573 drbd_csum_ee(mdev, mdev->integrity_w_tfm, e, dgb);
2574 ok = drbd_send(mdev, mdev->data.socket, dgb, dgs, 0);
2577 ok = _drbd_send_zc_ee(mdev, e);
2579 drbd_put_data_sock(mdev);
2585 drbd_send distinguishes two cases:
2587 Packets sent via the data socket "sock"
2588 and packets sent via the meta data socket "msock"
2591 -----------------+-------------------------+------------------------------
2592 timeout conf.timeout / 2 conf.timeout / 2
2593 timeout action send a ping via msock Abort communication
2594 and close all sockets
2598 * you must have down()ed the appropriate [m]sock_mutex elsewhere!
2600 int drbd_send(struct drbd_conf *mdev, struct socket *sock,
2601 void *buf, size_t size, unsigned msg_flags)
2610 /* THINK if (signal_pending) return ... ? */
2615 msg.msg_name = NULL;
2616 msg.msg_namelen = 0;
2617 msg.msg_control = NULL;
2618 msg.msg_controllen = 0;
2619 msg.msg_flags = msg_flags | MSG_NOSIGNAL;
2621 if (sock == mdev->data.socket) {
2622 mdev->ko_count = mdev->net_conf->ko_count;
2623 drbd_update_congested(mdev);
2627 * tcp_sendmsg does _not_ use its size parameter at all ?
2629 * -EAGAIN on timeout, -EINTR on signal.
2632 * do we need to block DRBD_SIG if sock == &meta.socket ??
2633 * otherwise wake_asender() might interrupt some send_*Ack !
2635 rv = kernel_sendmsg(sock, &msg, &iov, 1, size);
2636 if (rv == -EAGAIN) {
2637 if (we_should_drop_the_connection(mdev, sock))
2644 flush_signals(current);
2652 } while (sent < size);
2654 if (sock == mdev->data.socket)
2655 clear_bit(NET_CONGESTED, &mdev->flags);
2658 if (rv != -EAGAIN) {
2659 dev_err(DEV, "%s_sendmsg returned %d\n",
2660 sock == mdev->meta.socket ? "msock" : "sock",
2662 drbd_force_state(mdev, NS(conn, C_BROKEN_PIPE));
2664 drbd_force_state(mdev, NS(conn, C_TIMEOUT));
2670 static int drbd_open(struct block_device *bdev, fmode_t mode)
2672 struct drbd_conf *mdev = bdev->bd_disk->private_data;
2673 unsigned long flags;
2676 mutex_lock(&drbd_main_mutex);
2677 spin_lock_irqsave(&mdev->req_lock, flags);
2678 /* to have a stable mdev->state.role
2679 * and no race with updating open_cnt */
2681 if (mdev->state.role != R_PRIMARY) {
2682 if (mode & FMODE_WRITE)
2684 else if (!allow_oos)
2690 spin_unlock_irqrestore(&mdev->req_lock, flags);
2691 mutex_unlock(&drbd_main_mutex);
2696 static int drbd_release(struct gendisk *gd, fmode_t mode)
2698 struct drbd_conf *mdev = gd->private_data;
2699 mutex_lock(&drbd_main_mutex);
2701 mutex_unlock(&drbd_main_mutex);
2705 static void drbd_unplug_fn(struct request_queue *q)
2707 struct drbd_conf *mdev = q->queuedata;
2710 spin_lock_irq(q->queue_lock);
2712 spin_unlock_irq(q->queue_lock);
2714 /* only if connected */
2715 spin_lock_irq(&mdev->req_lock);
2716 if (mdev->state.pdsk >= D_INCONSISTENT && mdev->state.conn >= C_CONNECTED) {
2717 D_ASSERT(mdev->state.role == R_PRIMARY);
2718 if (test_and_clear_bit(UNPLUG_REMOTE, &mdev->flags)) {
2719 /* add to the data.work queue,
2720 * unless already queued.
2721 * XXX this might be a good addition to drbd_queue_work
2722 * anyways, to detect "double queuing" ... */
2723 if (list_empty(&mdev->unplug_work.list))
2724 drbd_queue_work(&mdev->data.work,
2725 &mdev->unplug_work);
2728 spin_unlock_irq(&mdev->req_lock);
2730 if (mdev->state.disk >= D_INCONSISTENT)
2734 static void drbd_set_defaults(struct drbd_conf *mdev)
2736 /* This way we get a compile error when sync_conf grows,
2737 and we forgot to initialize it here */
2738 mdev->sync_conf = (struct syncer_conf) {
2739 /* .rate = */ DRBD_RATE_DEF,
2740 /* .after = */ DRBD_AFTER_DEF,
2741 /* .al_extents = */ DRBD_AL_EXTENTS_DEF,
2742 /* .verify_alg = */ {}, 0,
2743 /* .cpu_mask = */ {}, 0,
2744 /* .csums_alg = */ {}, 0,
2746 /* .on_no_data = */ DRBD_ON_NO_DATA_DEF,
2747 /* .c_plan_ahead = */ DRBD_C_PLAN_AHEAD_DEF,
2748 /* .c_delay_target = */ DRBD_C_DELAY_TARGET_DEF,
2749 /* .c_fill_target = */ DRBD_C_FILL_TARGET_DEF,
2750 /* .c_max_rate = */ DRBD_C_MAX_RATE_DEF,
2751 /* .c_min_rate = */ DRBD_C_MIN_RATE_DEF
2754 /* Have to use that way, because the layout differs between
2755 big endian and little endian */
2756 mdev->state = (union drbd_state) {
2757 { .role = R_SECONDARY,
2759 .conn = C_STANDALONE,
2768 void drbd_init_set_defaults(struct drbd_conf *mdev)
2770 /* the memset(,0,) did most of this.
2771 * note: only assignments, no allocation in here */
2773 drbd_set_defaults(mdev);
2775 /* for now, we do NOT yet support it,
2776 * even though we start some framework
2777 * to eventually support barriers */
2778 set_bit(NO_BARRIER_SUPP, &mdev->flags);
2780 atomic_set(&mdev->ap_bio_cnt, 0);
2781 atomic_set(&mdev->ap_pending_cnt, 0);
2782 atomic_set(&mdev->rs_pending_cnt, 0);
2783 atomic_set(&mdev->unacked_cnt, 0);
2784 atomic_set(&mdev->local_cnt, 0);
2785 atomic_set(&mdev->net_cnt, 0);
2786 atomic_set(&mdev->packet_seq, 0);
2787 atomic_set(&mdev->pp_in_use, 0);
2788 atomic_set(&mdev->pp_in_use_by_net, 0);
2789 atomic_set(&mdev->rs_sect_in, 0);
2790 atomic_set(&mdev->rs_sect_ev, 0);
2792 mutex_init(&mdev->md_io_mutex);
2793 mutex_init(&mdev->data.mutex);
2794 mutex_init(&mdev->meta.mutex);
2795 sema_init(&mdev->data.work.s, 0);
2796 sema_init(&mdev->meta.work.s, 0);
2797 mutex_init(&mdev->state_mutex);
2799 spin_lock_init(&mdev->data.work.q_lock);
2800 spin_lock_init(&mdev->meta.work.q_lock);
2802 spin_lock_init(&mdev->al_lock);
2803 spin_lock_init(&mdev->req_lock);
2804 spin_lock_init(&mdev->peer_seq_lock);
2805 spin_lock_init(&mdev->epoch_lock);
2807 INIT_LIST_HEAD(&mdev->active_ee);
2808 INIT_LIST_HEAD(&mdev->sync_ee);
2809 INIT_LIST_HEAD(&mdev->done_ee);
2810 INIT_LIST_HEAD(&mdev->read_ee);
2811 INIT_LIST_HEAD(&mdev->net_ee);
2812 INIT_LIST_HEAD(&mdev->resync_reads);
2813 INIT_LIST_HEAD(&mdev->data.work.q);
2814 INIT_LIST_HEAD(&mdev->meta.work.q);
2815 INIT_LIST_HEAD(&mdev->resync_work.list);
2816 INIT_LIST_HEAD(&mdev->unplug_work.list);
2817 INIT_LIST_HEAD(&mdev->go_diskless.list);
2818 INIT_LIST_HEAD(&mdev->md_sync_work.list);
2819 INIT_LIST_HEAD(&mdev->bm_io_work.w.list);
2821 mdev->resync_work.cb = w_resync_inactive;
2822 mdev->unplug_work.cb = w_send_write_hint;
2823 mdev->go_diskless.cb = w_go_diskless;
2824 mdev->md_sync_work.cb = w_md_sync;
2825 mdev->bm_io_work.w.cb = w_bitmap_io;
2826 init_timer(&mdev->resync_timer);
2827 init_timer(&mdev->md_sync_timer);
2828 mdev->resync_timer.function = resync_timer_fn;
2829 mdev->resync_timer.data = (unsigned long) mdev;
2830 mdev->md_sync_timer.function = md_sync_timer_fn;
2831 mdev->md_sync_timer.data = (unsigned long) mdev;
2833 init_waitqueue_head(&mdev->misc_wait);
2834 init_waitqueue_head(&mdev->state_wait);
2835 init_waitqueue_head(&mdev->net_cnt_wait);
2836 init_waitqueue_head(&mdev->ee_wait);
2837 init_waitqueue_head(&mdev->al_wait);
2838 init_waitqueue_head(&mdev->seq_wait);
2840 drbd_thread_init(mdev, &mdev->receiver, drbdd_init);
2841 drbd_thread_init(mdev, &mdev->worker, drbd_worker);
2842 drbd_thread_init(mdev, &mdev->asender, drbd_asender);
2844 mdev->agreed_pro_version = PRO_VERSION_MAX;
2845 mdev->write_ordering = WO_bio_barrier;
2846 mdev->resync_wenr = LC_FREE;
2849 void drbd_mdev_cleanup(struct drbd_conf *mdev)
2852 if (mdev->receiver.t_state != None)
2853 dev_err(DEV, "ASSERT FAILED: receiver t_state == %d expected 0.\n",
2854 mdev->receiver.t_state);
2856 /* no need to lock it, I'm the only thread alive */
2857 if (atomic_read(&mdev->current_epoch->epoch_size) != 0)
2858 dev_err(DEV, "epoch_size:%d\n", atomic_read(&mdev->current_epoch->epoch_size));
2868 mdev->rs_failed = 0;
2869 mdev->rs_last_events = 0;
2870 mdev->rs_last_sect_ev = 0;
2871 for (i = 0; i < DRBD_SYNC_MARKS; i++) {
2872 mdev->rs_mark_left[i] = 0;
2873 mdev->rs_mark_time[i] = 0;
2875 D_ASSERT(mdev->net_conf == NULL);
2877 drbd_set_my_capacity(mdev, 0);
2879 /* maybe never allocated. */
2880 drbd_bm_resize(mdev, 0, 1);
2881 drbd_bm_cleanup(mdev);
2884 drbd_free_resources(mdev);
2885 clear_bit(AL_SUSPENDED, &mdev->flags);
2888 * currently we drbd_init_ee only on module load, so
2889 * we may do drbd_release_ee only on module unload!
2891 D_ASSERT(list_empty(&mdev->active_ee));
2892 D_ASSERT(list_empty(&mdev->sync_ee));
2893 D_ASSERT(list_empty(&mdev->done_ee));
2894 D_ASSERT(list_empty(&mdev->read_ee));
2895 D_ASSERT(list_empty(&mdev->net_ee));
2896 D_ASSERT(list_empty(&mdev->resync_reads));
2897 D_ASSERT(list_empty(&mdev->data.work.q));
2898 D_ASSERT(list_empty(&mdev->meta.work.q));
2899 D_ASSERT(list_empty(&mdev->resync_work.list));
2900 D_ASSERT(list_empty(&mdev->unplug_work.list));
2901 D_ASSERT(list_empty(&mdev->go_diskless.list));
2906 static void drbd_destroy_mempools(void)
2910 while (drbd_pp_pool) {
2911 page = drbd_pp_pool;
2912 drbd_pp_pool = (struct page *)page_private(page);
2917 /* D_ASSERT(atomic_read(&drbd_pp_vacant)==0); */
2919 if (drbd_ee_mempool)
2920 mempool_destroy(drbd_ee_mempool);
2921 if (drbd_request_mempool)
2922 mempool_destroy(drbd_request_mempool);
2924 kmem_cache_destroy(drbd_ee_cache);
2925 if (drbd_request_cache)
2926 kmem_cache_destroy(drbd_request_cache);
2927 if (drbd_bm_ext_cache)
2928 kmem_cache_destroy(drbd_bm_ext_cache);
2929 if (drbd_al_ext_cache)
2930 kmem_cache_destroy(drbd_al_ext_cache);
2932 drbd_ee_mempool = NULL;
2933 drbd_request_mempool = NULL;
2934 drbd_ee_cache = NULL;
2935 drbd_request_cache = NULL;
2936 drbd_bm_ext_cache = NULL;
2937 drbd_al_ext_cache = NULL;
2942 static int drbd_create_mempools(void)
2945 const int number = (DRBD_MAX_SEGMENT_SIZE/PAGE_SIZE) * minor_count;
2948 /* prepare our caches and mempools */
2949 drbd_request_mempool = NULL;
2950 drbd_ee_cache = NULL;
2951 drbd_request_cache = NULL;
2952 drbd_bm_ext_cache = NULL;
2953 drbd_al_ext_cache = NULL;
2954 drbd_pp_pool = NULL;
2957 drbd_request_cache = kmem_cache_create(
2958 "drbd_req", sizeof(struct drbd_request), 0, 0, NULL);
2959 if (drbd_request_cache == NULL)
2962 drbd_ee_cache = kmem_cache_create(
2963 "drbd_ee", sizeof(struct drbd_epoch_entry), 0, 0, NULL);
2964 if (drbd_ee_cache == NULL)
2967 drbd_bm_ext_cache = kmem_cache_create(
2968 "drbd_bm", sizeof(struct bm_extent), 0, 0, NULL);
2969 if (drbd_bm_ext_cache == NULL)
2972 drbd_al_ext_cache = kmem_cache_create(
2973 "drbd_al", sizeof(struct lc_element), 0, 0, NULL);
2974 if (drbd_al_ext_cache == NULL)
2978 drbd_request_mempool = mempool_create(number,
2979 mempool_alloc_slab, mempool_free_slab, drbd_request_cache);
2980 if (drbd_request_mempool == NULL)
2983 drbd_ee_mempool = mempool_create(number,
2984 mempool_alloc_slab, mempool_free_slab, drbd_ee_cache);
2985 if (drbd_ee_mempool == NULL)
2988 /* drbd's page pool */
2989 spin_lock_init(&drbd_pp_lock);
2991 for (i = 0; i < number; i++) {
2992 page = alloc_page(GFP_HIGHUSER);
2995 set_page_private(page, (unsigned long)drbd_pp_pool);
2996 drbd_pp_pool = page;
2998 drbd_pp_vacant = number;
3003 drbd_destroy_mempools(); /* in case we allocated some */
3007 static int drbd_notify_sys(struct notifier_block *this, unsigned long code,
3010 /* just so we have it. you never know what interesting things we
3011 * might want to do here some day...
3017 static struct notifier_block drbd_notifier = {
3018 .notifier_call = drbd_notify_sys,
3021 static void drbd_release_ee_lists(struct drbd_conf *mdev)
3025 rr = drbd_release_ee(mdev, &mdev->active_ee);
3027 dev_err(DEV, "%d EEs in active list found!\n", rr);
3029 rr = drbd_release_ee(mdev, &mdev->sync_ee);
3031 dev_err(DEV, "%d EEs in sync list found!\n", rr);
3033 rr = drbd_release_ee(mdev, &mdev->read_ee);
3035 dev_err(DEV, "%d EEs in read list found!\n", rr);
3037 rr = drbd_release_ee(mdev, &mdev->done_ee);
3039 dev_err(DEV, "%d EEs in done list found!\n", rr);
3041 rr = drbd_release_ee(mdev, &mdev->net_ee);
3043 dev_err(DEV, "%d EEs in net list found!\n", rr);
3046 /* caution. no locking.
3047 * currently only used from module cleanup code. */
3048 static void drbd_delete_device(unsigned int minor)
3050 struct drbd_conf *mdev = minor_to_mdev(minor);
3055 /* paranoia asserts */
3056 if (mdev->open_cnt != 0)
3057 dev_err(DEV, "open_cnt = %d in %s:%u", mdev->open_cnt,
3058 __FILE__ , __LINE__);
3060 ERR_IF (!list_empty(&mdev->data.work.q)) {
3061 struct list_head *lp;
3062 list_for_each(lp, &mdev->data.work.q) {
3063 dev_err(DEV, "lp = %p\n", lp);
3066 /* end paranoia asserts */
3068 del_gendisk(mdev->vdisk);
3070 /* cleanup stuff that may have been allocated during
3071 * device (re-)configuration or state changes */
3073 if (mdev->this_bdev)
3074 bdput(mdev->this_bdev);
3076 drbd_free_resources(mdev);
3078 drbd_release_ee_lists(mdev);
3080 /* should be free'd on disconnect? */
3081 kfree(mdev->ee_hash);
3083 mdev->ee_hash_s = 0;
3084 mdev->ee_hash = NULL;
3087 lc_destroy(mdev->act_log);
3088 lc_destroy(mdev->resync);
3090 kfree(mdev->p_uuid);
3091 /* mdev->p_uuid = NULL; */
3093 kfree(mdev->int_dig_out);
3094 kfree(mdev->int_dig_in);
3095 kfree(mdev->int_dig_vv);
3097 /* cleanup the rest that has been
3098 * allocated from drbd_new_device
3099 * and actually free the mdev itself */
3100 drbd_free_mdev(mdev);
3103 static void drbd_cleanup(void)
3107 unregister_reboot_notifier(&drbd_notifier);
3113 remove_proc_entry("drbd", NULL);
3116 drbd_delete_device(i);
3117 drbd_destroy_mempools();
3122 unregister_blkdev(DRBD_MAJOR, "drbd");
3124 printk(KERN_INFO "drbd: module cleanup done.\n");
3128 * drbd_congested() - Callback for pdflush
3129 * @congested_data: User data
3130 * @bdi_bits: Bits pdflush is currently interested in
3132 * Returns 1<<BDI_async_congested and/or 1<<BDI_sync_congested if we are congested.
3134 static int drbd_congested(void *congested_data, int bdi_bits)
3136 struct drbd_conf *mdev = congested_data;
3137 struct request_queue *q;
3141 if (!__inc_ap_bio_cond(mdev)) {
3142 /* DRBD has frozen IO */
3148 if (get_ldev(mdev)) {
3149 q = bdev_get_queue(mdev->ldev->backing_bdev);
3150 r = bdi_congested(&q->backing_dev_info, bdi_bits);
3156 if (bdi_bits & (1 << BDI_async_congested) && test_bit(NET_CONGESTED, &mdev->flags)) {
3157 r |= (1 << BDI_async_congested);
3158 reason = reason == 'b' ? 'a' : 'n';
3162 mdev->congestion_reason = reason;
3166 struct drbd_conf *drbd_new_device(unsigned int minor)
3168 struct drbd_conf *mdev;
3169 struct gendisk *disk;
3170 struct request_queue *q;
3172 /* GFP_KERNEL, we are outside of all write-out paths */
3173 mdev = kzalloc(sizeof(struct drbd_conf), GFP_KERNEL);
3176 if (!zalloc_cpumask_var(&mdev->cpu_mask, GFP_KERNEL))
3177 goto out_no_cpumask;
3179 mdev->minor = minor;
3181 drbd_init_set_defaults(mdev);
3183 q = blk_alloc_queue(GFP_KERNEL);
3187 q->queuedata = mdev;
3189 disk = alloc_disk(1);
3194 set_disk_ro(disk, TRUE);
3197 disk->major = DRBD_MAJOR;
3198 disk->first_minor = minor;
3199 disk->fops = &drbd_ops;
3200 sprintf(disk->disk_name, "drbd%d", minor);
3201 disk->private_data = mdev;
3203 mdev->this_bdev = bdget(MKDEV(DRBD_MAJOR, minor));
3204 /* we have no partitions. we contain only ourselves. */
3205 mdev->this_bdev->bd_contains = mdev->this_bdev;
3207 q->backing_dev_info.congested_fn = drbd_congested;
3208 q->backing_dev_info.congested_data = mdev;
3210 blk_queue_make_request(q, drbd_make_request_26);
3211 blk_queue_max_segment_size(q, DRBD_MAX_SEGMENT_SIZE);
3212 blk_queue_bounce_limit(q, BLK_BOUNCE_ANY);
3213 blk_queue_merge_bvec(q, drbd_merge_bvec);
3214 q->queue_lock = &mdev->req_lock; /* needed since we use */
3215 /* plugging on a queue, that actually has no requests! */
3216 q->unplug_fn = drbd_unplug_fn;
3218 mdev->md_io_page = alloc_page(GFP_KERNEL);
3219 if (!mdev->md_io_page)
3220 goto out_no_io_page;
3222 if (drbd_bm_init(mdev))
3224 /* no need to lock access, we are still initializing this minor device. */
3228 mdev->app_reads_hash = kzalloc(APP_R_HSIZE*sizeof(void *), GFP_KERNEL);
3229 if (!mdev->app_reads_hash)
3230 goto out_no_app_reads;
3232 mdev->current_epoch = kzalloc(sizeof(struct drbd_epoch), GFP_KERNEL);
3233 if (!mdev->current_epoch)
3236 INIT_LIST_HEAD(&mdev->current_epoch->list);
3241 /* out_whatever_else:
3242 kfree(mdev->current_epoch); */
3244 kfree(mdev->app_reads_hash);
3248 drbd_bm_cleanup(mdev);
3250 __free_page(mdev->md_io_page);
3254 blk_cleanup_queue(q);
3256 free_cpumask_var(mdev->cpu_mask);
3262 /* counterpart of drbd_new_device.
3263 * last part of drbd_delete_device. */
3264 void drbd_free_mdev(struct drbd_conf *mdev)
3266 kfree(mdev->current_epoch);
3267 kfree(mdev->app_reads_hash);
3269 if (mdev->bitmap) /* should no longer be there. */
3270 drbd_bm_cleanup(mdev);
3271 __free_page(mdev->md_io_page);
3272 put_disk(mdev->vdisk);
3273 blk_cleanup_queue(mdev->rq_queue);
3274 free_cpumask_var(mdev->cpu_mask);
3279 int __init drbd_init(void)
3283 if (sizeof(struct p_handshake) != 80) {
3285 "drbd: never change the size or layout "
3286 "of the HandShake packet.\n");
3290 if (1 > minor_count || minor_count > 255) {
3292 "drbd: invalid minor_count (%d)\n", minor_count);
3300 err = drbd_nl_init();
3304 err = register_blkdev(DRBD_MAJOR, "drbd");
3307 "drbd: unable to register block device major %d\n",
3312 register_reboot_notifier(&drbd_notifier);
3315 * allocate all necessary structs
3319 init_waitqueue_head(&drbd_pp_wait);
3321 drbd_proc = NULL; /* play safe for drbd_cleanup */
3322 minor_table = kzalloc(sizeof(struct drbd_conf *)*minor_count,
3327 err = drbd_create_mempools();
3331 drbd_proc = proc_create_data("drbd", S_IFREG | S_IRUGO , NULL, &drbd_proc_fops, NULL);
3333 printk(KERN_ERR "drbd: unable to register proc file\n");
3337 rwlock_init(&global_state_lock);
3339 printk(KERN_INFO "drbd: initialized. "
3340 "Version: " REL_VERSION " (api:%d/proto:%d-%d)\n",
3341 API_VERSION, PRO_VERSION_MIN, PRO_VERSION_MAX);
3342 printk(KERN_INFO "drbd: %s\n", drbd_buildtag());
3343 printk(KERN_INFO "drbd: registered as block device major %d\n",
3345 printk(KERN_INFO "drbd: minor_table @ 0x%p\n", minor_table);
3347 return 0; /* Success! */
3352 /* currently always the case */
3353 printk(KERN_ERR "drbd: ran out of memory\n");
3355 printk(KERN_ERR "drbd: initialization failure\n");
3359 void drbd_free_bc(struct drbd_backing_dev *ldev)
3364 bd_release(ldev->backing_bdev);
3365 bd_release(ldev->md_bdev);
3367 fput(ldev->lo_file);
3368 fput(ldev->md_file);
3373 void drbd_free_sock(struct drbd_conf *mdev)
3375 if (mdev->data.socket) {
3376 mutex_lock(&mdev->data.mutex);
3377 kernel_sock_shutdown(mdev->data.socket, SHUT_RDWR);
3378 sock_release(mdev->data.socket);
3379 mdev->data.socket = NULL;
3380 mutex_unlock(&mdev->data.mutex);
3382 if (mdev->meta.socket) {
3383 mutex_lock(&mdev->meta.mutex);
3384 kernel_sock_shutdown(mdev->meta.socket, SHUT_RDWR);
3385 sock_release(mdev->meta.socket);
3386 mdev->meta.socket = NULL;
3387 mutex_unlock(&mdev->meta.mutex);
3392 void drbd_free_resources(struct drbd_conf *mdev)
3394 crypto_free_hash(mdev->csums_tfm);
3395 mdev->csums_tfm = NULL;
3396 crypto_free_hash(mdev->verify_tfm);
3397 mdev->verify_tfm = NULL;
3398 crypto_free_hash(mdev->cram_hmac_tfm);
3399 mdev->cram_hmac_tfm = NULL;
3400 crypto_free_hash(mdev->integrity_w_tfm);
3401 mdev->integrity_w_tfm = NULL;
3402 crypto_free_hash(mdev->integrity_r_tfm);
3403 mdev->integrity_r_tfm = NULL;
3405 drbd_free_sock(mdev);
3408 drbd_free_bc(mdev->ldev);
3409 mdev->ldev = NULL;);
3412 /* meta data management */
3414 struct meta_data_on_disk {
3415 u64 la_size; /* last agreed size. */
3416 u64 uuid[UI_SIZE]; /* UUIDs. */
3419 u32 flags; /* MDF */
3422 u32 al_offset; /* offset to this block */
3423 u32 al_nr_extents; /* important for restoring the AL */
3424 /* `-- act_log->nr_elements <-- sync_conf.al_extents */
3425 u32 bm_offset; /* offset to the bitmap, from here */
3426 u32 bm_bytes_per_bit; /* BM_BLOCK_SIZE */
3427 u32 reserved_u32[4];
3432 * drbd_md_sync() - Writes the meta data super block if the MD_DIRTY flag bit is set
3433 * @mdev: DRBD device.
3435 void drbd_md_sync(struct drbd_conf *mdev)
3437 struct meta_data_on_disk *buffer;
3441 del_timer(&mdev->md_sync_timer);
3442 /* timer may be rearmed by drbd_md_mark_dirty() now. */
3443 if (!test_and_clear_bit(MD_DIRTY, &mdev->flags))
3446 /* We use here D_FAILED and not D_ATTACHING because we try to write
3447 * metadata even if we detach due to a disk failure! */
3448 if (!get_ldev_if_state(mdev, D_FAILED))
3451 mutex_lock(&mdev->md_io_mutex);
3452 buffer = (struct meta_data_on_disk *)page_address(mdev->md_io_page);
3453 memset(buffer, 0, 512);
3455 buffer->la_size = cpu_to_be64(drbd_get_capacity(mdev->this_bdev));
3456 for (i = UI_CURRENT; i < UI_SIZE; i++)
3457 buffer->uuid[i] = cpu_to_be64(mdev->ldev->md.uuid[i]);
3458 buffer->flags = cpu_to_be32(mdev->ldev->md.flags);
3459 buffer->magic = cpu_to_be32(DRBD_MD_MAGIC);
3461 buffer->md_size_sect = cpu_to_be32(mdev->ldev->md.md_size_sect);
3462 buffer->al_offset = cpu_to_be32(mdev->ldev->md.al_offset);
3463 buffer->al_nr_extents = cpu_to_be32(mdev->act_log->nr_elements);
3464 buffer->bm_bytes_per_bit = cpu_to_be32(BM_BLOCK_SIZE);
3465 buffer->device_uuid = cpu_to_be64(mdev->ldev->md.device_uuid);
3467 buffer->bm_offset = cpu_to_be32(mdev->ldev->md.bm_offset);
3469 D_ASSERT(drbd_md_ss__(mdev, mdev->ldev) == mdev->ldev->md.md_offset);
3470 sector = mdev->ldev->md.md_offset;
3472 if (!drbd_md_sync_page_io(mdev, mdev->ldev, sector, WRITE)) {
3473 /* this was a try anyways ... */
3474 dev_err(DEV, "meta data update failed!\n");
3475 drbd_chk_io_error(mdev, 1, TRUE);
3478 /* Update mdev->ldev->md.la_size_sect,
3479 * since we updated it on metadata. */
3480 mdev->ldev->md.la_size_sect = drbd_get_capacity(mdev->this_bdev);
3482 mutex_unlock(&mdev->md_io_mutex);
3487 * drbd_md_read() - Reads in the meta data super block
3488 * @mdev: DRBD device.
3489 * @bdev: Device from which the meta data should be read in.
3491 * Return 0 (NO_ERROR) on success, and an enum drbd_ret_codes in case
3492 * something goes wrong. Currently only: ERR_IO_MD_DISK, ERR_MD_INVALID.
3494 int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev)
3496 struct meta_data_on_disk *buffer;
3497 int i, rv = NO_ERROR;
3499 if (!get_ldev_if_state(mdev, D_ATTACHING))
3500 return ERR_IO_MD_DISK;
3502 mutex_lock(&mdev->md_io_mutex);
3503 buffer = (struct meta_data_on_disk *)page_address(mdev->md_io_page);
3505 if (!drbd_md_sync_page_io(mdev, bdev, bdev->md.md_offset, READ)) {
3506 /* NOTE: cant do normal error processing here as this is
3507 called BEFORE disk is attached */
3508 dev_err(DEV, "Error while reading metadata.\n");
3509 rv = ERR_IO_MD_DISK;
3513 if (be32_to_cpu(buffer->magic) != DRBD_MD_MAGIC) {
3514 dev_err(DEV, "Error while reading metadata, magic not found.\n");
3515 rv = ERR_MD_INVALID;
3518 if (be32_to_cpu(buffer->al_offset) != bdev->md.al_offset) {
3519 dev_err(DEV, "unexpected al_offset: %d (expected %d)\n",
3520 be32_to_cpu(buffer->al_offset), bdev->md.al_offset);
3521 rv = ERR_MD_INVALID;
3524 if (be32_to_cpu(buffer->bm_offset) != bdev->md.bm_offset) {
3525 dev_err(DEV, "unexpected bm_offset: %d (expected %d)\n",
3526 be32_to_cpu(buffer->bm_offset), bdev->md.bm_offset);
3527 rv = ERR_MD_INVALID;
3530 if (be32_to_cpu(buffer->md_size_sect) != bdev->md.md_size_sect) {
3531 dev_err(DEV, "unexpected md_size: %u (expected %u)\n",
3532 be32_to_cpu(buffer->md_size_sect), bdev->md.md_size_sect);
3533 rv = ERR_MD_INVALID;
3537 if (be32_to_cpu(buffer->bm_bytes_per_bit) != BM_BLOCK_SIZE) {
3538 dev_err(DEV, "unexpected bm_bytes_per_bit: %u (expected %u)\n",
3539 be32_to_cpu(buffer->bm_bytes_per_bit), BM_BLOCK_SIZE);
3540 rv = ERR_MD_INVALID;
3544 bdev->md.la_size_sect = be64_to_cpu(buffer->la_size);
3545 for (i = UI_CURRENT; i < UI_SIZE; i++)
3546 bdev->md.uuid[i] = be64_to_cpu(buffer->uuid[i]);
3547 bdev->md.flags = be32_to_cpu(buffer->flags);
3548 mdev->sync_conf.al_extents = be32_to_cpu(buffer->al_nr_extents);
3549 bdev->md.device_uuid = be64_to_cpu(buffer->device_uuid);
3551 if (mdev->sync_conf.al_extents < 7)
3552 mdev->sync_conf.al_extents = 127;
3555 mutex_unlock(&mdev->md_io_mutex);
3561 static void debug_drbd_uuid(struct drbd_conf *mdev, enum drbd_uuid_index index)
3563 static char *uuid_str[UI_EXTENDED_SIZE] = {
3564 [UI_CURRENT] = "CURRENT",
3565 [UI_BITMAP] = "BITMAP",
3566 [UI_HISTORY_START] = "HISTORY_START",
3567 [UI_HISTORY_END] = "HISTORY_END",
3569 [UI_FLAGS] = "FLAGS",
3572 if (index >= UI_EXTENDED_SIZE) {
3573 dev_warn(DEV, " uuid_index >= EXTENDED_SIZE\n");
3577 dynamic_dev_dbg(DEV, " uuid[%s] now %016llX\n",
3579 (unsigned long long)mdev->ldev->md.uuid[index]);
3584 * drbd_md_mark_dirty() - Mark meta data super block as dirty
3585 * @mdev: DRBD device.
3587 * Call this function if you change anything that should be written to
3588 * the meta-data super block. This function sets MD_DIRTY, and starts a
3589 * timer that ensures that within five seconds you have to call drbd_md_sync().
3592 void drbd_md_mark_dirty_(struct drbd_conf *mdev, unsigned int line, const char *func)
3594 if (!test_and_set_bit(MD_DIRTY, &mdev->flags)) {
3595 mod_timer(&mdev->md_sync_timer, jiffies + HZ);
3596 mdev->last_md_mark_dirty.line = line;
3597 mdev->last_md_mark_dirty.func = func;
3601 void drbd_md_mark_dirty(struct drbd_conf *mdev)
3603 if (!test_and_set_bit(MD_DIRTY, &mdev->flags))
3604 mod_timer(&mdev->md_sync_timer, jiffies + 5*HZ);
3608 static void drbd_uuid_move_history(struct drbd_conf *mdev) __must_hold(local)
3612 for (i = UI_HISTORY_START; i < UI_HISTORY_END; i++) {
3613 mdev->ldev->md.uuid[i+1] = mdev->ldev->md.uuid[i];
3614 debug_drbd_uuid(mdev, i+1);
3618 void _drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local)
3620 if (idx == UI_CURRENT) {
3621 if (mdev->state.role == R_PRIMARY)
3626 drbd_set_ed_uuid(mdev, val);
3629 mdev->ldev->md.uuid[idx] = val;
3630 debug_drbd_uuid(mdev, idx);
3631 drbd_md_mark_dirty(mdev);
3635 void drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local)
3637 if (mdev->ldev->md.uuid[idx]) {
3638 drbd_uuid_move_history(mdev);
3639 mdev->ldev->md.uuid[UI_HISTORY_START] = mdev->ldev->md.uuid[idx];
3640 debug_drbd_uuid(mdev, UI_HISTORY_START);
3642 _drbd_uuid_set(mdev, idx, val);
3646 * drbd_uuid_new_current() - Creates a new current UUID
3647 * @mdev: DRBD device.
3649 * Creates a new current UUID, and rotates the old current UUID into
3650 * the bitmap slot. Causes an incremental resync upon next connect.
3652 void drbd_uuid_new_current(struct drbd_conf *mdev) __must_hold(local)
3656 dev_info(DEV, "Creating new current UUID\n");
3657 D_ASSERT(mdev->ldev->md.uuid[UI_BITMAP] == 0);
3658 mdev->ldev->md.uuid[UI_BITMAP] = mdev->ldev->md.uuid[UI_CURRENT];
3659 debug_drbd_uuid(mdev, UI_BITMAP);
3661 get_random_bytes(&val, sizeof(u64));
3662 _drbd_uuid_set(mdev, UI_CURRENT, val);
3665 void drbd_uuid_set_bm(struct drbd_conf *mdev, u64 val) __must_hold(local)
3667 if (mdev->ldev->md.uuid[UI_BITMAP] == 0 && val == 0)
3671 drbd_uuid_move_history(mdev);
3672 mdev->ldev->md.uuid[UI_HISTORY_START] = mdev->ldev->md.uuid[UI_BITMAP];
3673 mdev->ldev->md.uuid[UI_BITMAP] = 0;
3674 debug_drbd_uuid(mdev, UI_HISTORY_START);
3675 debug_drbd_uuid(mdev, UI_BITMAP);
3677 if (mdev->ldev->md.uuid[UI_BITMAP])
3678 dev_warn(DEV, "bm UUID already set");
3680 mdev->ldev->md.uuid[UI_BITMAP] = val;
3681 mdev->ldev->md.uuid[UI_BITMAP] &= ~((u64)1);
3683 debug_drbd_uuid(mdev, UI_BITMAP);
3685 drbd_md_mark_dirty(mdev);
3689 * drbd_bmio_set_n_write() - io_fn for drbd_queue_bitmap_io() or drbd_bitmap_io()
3690 * @mdev: DRBD device.
3692 * Sets all bits in the bitmap and writes the whole bitmap to stable storage.
3694 int drbd_bmio_set_n_write(struct drbd_conf *mdev)
3698 if (get_ldev_if_state(mdev, D_ATTACHING)) {
3699 drbd_md_set_flag(mdev, MDF_FULL_SYNC);
3701 drbd_bm_set_all(mdev);
3703 rv = drbd_bm_write(mdev);
3706 drbd_md_clear_flag(mdev, MDF_FULL_SYNC);
3717 * drbd_bmio_clear_n_write() - io_fn for drbd_queue_bitmap_io() or drbd_bitmap_io()
3718 * @mdev: DRBD device.
3720 * Clears all bits in the bitmap and writes the whole bitmap to stable storage.
3722 int drbd_bmio_clear_n_write(struct drbd_conf *mdev)
3726 drbd_resume_al(mdev);
3727 if (get_ldev_if_state(mdev, D_ATTACHING)) {
3728 drbd_bm_clear_all(mdev);
3729 rv = drbd_bm_write(mdev);
3736 static int w_bitmap_io(struct drbd_conf *mdev, struct drbd_work *w, int unused)
3738 struct bm_io_work *work = container_of(w, struct bm_io_work, w);
3741 D_ASSERT(atomic_read(&mdev->ap_bio_cnt) == 0);
3743 drbd_bm_lock(mdev, work->why);
3744 rv = work->io_fn(mdev);
3745 drbd_bm_unlock(mdev);
3747 clear_bit(BITMAP_IO, &mdev->flags);
3748 wake_up(&mdev->misc_wait);
3751 work->done(mdev, rv);
3753 clear_bit(BITMAP_IO_QUEUED, &mdev->flags);
3759 static int w_go_diskless(struct drbd_conf *mdev, struct drbd_work *w, int unused)
3761 D_ASSERT(mdev->state.disk == D_FAILED);
3762 /* we cannot assert local_cnt == 0 here, as get_ldev_if_state will
3763 * inc/dec it frequently. Once we are D_DISKLESS, no one will touch
3764 * the protected members anymore, though, so in the after_state_ch work
3765 * it will be safe to free them. */
3766 drbd_force_state(mdev, NS(disk, D_DISKLESS));
3767 /* We need to wait for return of references checked out while we still
3768 * have been D_FAILED, though (drbd_md_sync, bitmap io). */
3769 wait_event(mdev->misc_wait, !atomic_read(&mdev->local_cnt));
3771 clear_bit(GO_DISKLESS, &mdev->flags);
3775 void drbd_go_diskless(struct drbd_conf *mdev)
3777 D_ASSERT(mdev->state.disk == D_FAILED);
3778 if (!test_and_set_bit(GO_DISKLESS, &mdev->flags))
3779 drbd_queue_work(&mdev->data.work, &mdev->go_diskless);
3780 /* don't drbd_queue_work_front,
3781 * we need to serialize with the after_state_ch work
3782 * of the -> D_FAILED transition. */
3786 * drbd_queue_bitmap_io() - Queues an IO operation on the whole bitmap
3787 * @mdev: DRBD device.
3788 * @io_fn: IO callback to be called when bitmap IO is possible
3789 * @done: callback to be called after the bitmap IO was performed
3790 * @why: Descriptive text of the reason for doing the IO
3792 * While IO on the bitmap happens we freeze application IO thus we ensure
3793 * that drbd_set_out_of_sync() can not be called. This function MAY ONLY be
3794 * called from worker context. It MUST NOT be used while a previous such
3795 * work is still pending!
3797 void drbd_queue_bitmap_io(struct drbd_conf *mdev,
3798 int (*io_fn)(struct drbd_conf *),
3799 void (*done)(struct drbd_conf *, int),
3802 D_ASSERT(current == mdev->worker.task);
3804 D_ASSERT(!test_bit(BITMAP_IO_QUEUED, &mdev->flags));
3805 D_ASSERT(!test_bit(BITMAP_IO, &mdev->flags));
3806 D_ASSERT(list_empty(&mdev->bm_io_work.w.list));
3807 if (mdev->bm_io_work.why)
3808 dev_err(DEV, "FIXME going to queue '%s' but '%s' still pending?\n",
3809 why, mdev->bm_io_work.why);
3811 mdev->bm_io_work.io_fn = io_fn;
3812 mdev->bm_io_work.done = done;
3813 mdev->bm_io_work.why = why;
3815 set_bit(BITMAP_IO, &mdev->flags);
3816 if (atomic_read(&mdev->ap_bio_cnt) == 0) {
3817 if (list_empty(&mdev->bm_io_work.w.list)) {
3818 set_bit(BITMAP_IO_QUEUED, &mdev->flags);
3819 drbd_queue_work(&mdev->data.work, &mdev->bm_io_work.w);
3821 dev_err(DEV, "FIXME avoided double queuing bm_io_work\n");
3826 * drbd_bitmap_io() - Does an IO operation on the whole bitmap
3827 * @mdev: DRBD device.
3828 * @io_fn: IO callback to be called when bitmap IO is possible
3829 * @why: Descriptive text of the reason for doing the IO
3831 * freezes application IO while that the actual IO operations runs. This
3832 * functions MAY NOT be called from worker context.
3834 int drbd_bitmap_io(struct drbd_conf *mdev, int (*io_fn)(struct drbd_conf *), char *why)
3838 D_ASSERT(current != mdev->worker.task);
3840 drbd_suspend_io(mdev);
3842 drbd_bm_lock(mdev, why);
3844 drbd_bm_unlock(mdev);
3846 drbd_resume_io(mdev);
3851 void drbd_md_set_flag(struct drbd_conf *mdev, int flag) __must_hold(local)
3853 if ((mdev->ldev->md.flags & flag) != flag) {
3854 drbd_md_mark_dirty(mdev);
3855 mdev->ldev->md.flags |= flag;
3859 void drbd_md_clear_flag(struct drbd_conf *mdev, int flag) __must_hold(local)
3861 if ((mdev->ldev->md.flags & flag) != 0) {
3862 drbd_md_mark_dirty(mdev);
3863 mdev->ldev->md.flags &= ~flag;
3866 int drbd_md_test_flag(struct drbd_backing_dev *bdev, int flag)
3868 return (bdev->md.flags & flag) != 0;
3871 static void md_sync_timer_fn(unsigned long data)
3873 struct drbd_conf *mdev = (struct drbd_conf *) data;
3875 drbd_queue_work_front(&mdev->data.work, &mdev->md_sync_work);
3878 static int w_md_sync(struct drbd_conf *mdev, struct drbd_work *w, int unused)
3880 dev_warn(DEV, "md_sync_timer expired! Worker calls drbd_md_sync().\n");
3882 dev_warn(DEV, "last md_mark_dirty: %s:%u\n",
3883 mdev->last_md_mark_dirty.func, mdev->last_md_mark_dirty.line);
3889 #ifdef CONFIG_DRBD_FAULT_INJECTION
3890 /* Fault insertion support including random number generator shamelessly
3891 * stolen from kernel/rcutorture.c */
3892 struct fault_random_state {
3893 unsigned long state;
3894 unsigned long count;
3897 #define FAULT_RANDOM_MULT 39916801 /* prime */
3898 #define FAULT_RANDOM_ADD 479001701 /* prime */
3899 #define FAULT_RANDOM_REFRESH 10000
3902 * Crude but fast random-number generator. Uses a linear congruential
3903 * generator, with occasional help from get_random_bytes().
3905 static unsigned long
3906 _drbd_fault_random(struct fault_random_state *rsp)
3910 if (!rsp->count--) {
3911 get_random_bytes(&refresh, sizeof(refresh));
3912 rsp->state += refresh;
3913 rsp->count = FAULT_RANDOM_REFRESH;
3915 rsp->state = rsp->state * FAULT_RANDOM_MULT + FAULT_RANDOM_ADD;
3916 return swahw32(rsp->state);
3920 _drbd_fault_str(unsigned int type) {
3921 static char *_faults[] = {
3922 [DRBD_FAULT_MD_WR] = "Meta-data write",
3923 [DRBD_FAULT_MD_RD] = "Meta-data read",
3924 [DRBD_FAULT_RS_WR] = "Resync write",
3925 [DRBD_FAULT_RS_RD] = "Resync read",
3926 [DRBD_FAULT_DT_WR] = "Data write",
3927 [DRBD_FAULT_DT_RD] = "Data read",
3928 [DRBD_FAULT_DT_RA] = "Data read ahead",
3929 [DRBD_FAULT_BM_ALLOC] = "BM allocation",
3930 [DRBD_FAULT_AL_EE] = "EE allocation",
3931 [DRBD_FAULT_RECEIVE] = "receive data corruption",
3934 return (type < DRBD_FAULT_MAX) ? _faults[type] : "**Unknown**";
3938 _drbd_insert_fault(struct drbd_conf *mdev, unsigned int type)
3940 static struct fault_random_state rrs = {0, 0};
3942 unsigned int ret = (
3944 ((1 << mdev_to_minor(mdev)) & fault_devs) != 0) &&
3945 (((_drbd_fault_random(&rrs) % 100) + 1) <= fault_rate));
3950 if (__ratelimit(&drbd_ratelimit_state))
3951 dev_warn(DEV, "***Simulating %s failure\n",
3952 _drbd_fault_str(type));
3959 const char *drbd_buildtag(void)
3961 /* DRBD built from external sources has here a reference to the
3962 git hash of the source code. */
3964 static char buildtag[38] = "\0uilt-in";
3966 if (buildtag[0] == 0) {
3967 #ifdef CONFIG_MODULES
3968 if (THIS_MODULE != NULL)
3969 sprintf(buildtag, "srcversion: %-24s", THIS_MODULE->srcversion);
3978 module_init(drbd_init)
3979 module_exit(drbd_cleanup)
3981 EXPORT_SYMBOL(drbd_conn_str);
3982 EXPORT_SYMBOL(drbd_role_str);
3983 EXPORT_SYMBOL(drbd_disk_str);
3984 EXPORT_SYMBOL(drbd_set_st_err_str);