md-cluster.c 34.2 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12
/*
 * Copyright (C) 2015, SUSE
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2, or (at your option)
 * any later version.
 *
 */


#include <linux/module.h>
13 14
#include <linux/dlm.h>
#include <linux/sched.h>
15
#include <linux/raid/md_p.h>
16
#include "md.h"
17
#include "bitmap.h"
18
#include "md-cluster.h"
19 20

#define LVB_SIZE	64
21
#define NEW_DEV_TIMEOUT 5000
22 23 24 25 26 27 28

struct dlm_lock_resource {
	dlm_lockspace_t *ls;
	struct dlm_lksb lksb;
	char *name; /* lock name. */
	uint32_t flags; /* flags to pass to dlm_lock() */
	struct completion completion; /* completion for synchronized locking */
G
Goldwyn Rodrigues 已提交
29 30
	void (*bast)(void *arg, int mode); /* blocking AST function pointer*/
	struct mddev *mddev; /* pointing back to mddev. */
31
	int mode;
G
Goldwyn Rodrigues 已提交
32 33
};

34 35 36 37 38 39 40 41 42 43 44 45
struct suspend_info {
	int slot;
	sector_t lo;
	sector_t hi;
	struct list_head list;
};

struct resync_info {
	__le64 lo;
	__le64 hi;
};

46 47
/* md_cluster_info flags */
#define		MD_CLUSTER_WAITING_FOR_NEWDISK		1
48
#define		MD_CLUSTER_SUSPEND_READ_BALANCING	2
49
#define		MD_CLUSTER_BEGIN_JOIN_CLUSTER		3
50

51 52 53 54 55
/* Lock the send communication. This is done through
 * bit manipulation as opposed to a mutex in order to
 * accomodate lock and hold. See next comment.
 */
#define		MD_CLUSTER_SEND_LOCK			4
56 57 58 59 60 61
/* If cluster operations (such as adding a disk) must lock the
 * communication channel, so as to perform extra operations
 * (update metadata) and no other operation is allowed on the
 * MD. Token needs to be locked and held until the operation
 * completes witha md_update_sb(), which would eventually release
 * the lock.
62 63
 */
#define		MD_CLUSTER_SEND_LOCKED_ALREADY		5
64 65 66 67
/* We should receive message after node joined cluster and
 * set up all the related infos such as bitmap and personality */
#define		MD_CLUSTER_ALREADY_IN_CLUSTER		6
#define		MD_CLUSTER_PENDING_RECV_EVENT		7
68

69

G
Goldwyn Rodrigues 已提交
70 71 72
struct md_cluster_info {
	/* dlm lock space and resources for clustered raid. */
	dlm_lockspace_t *lockspace;
G
Goldwyn Rodrigues 已提交
73 74
	int slot_number;
	struct completion completion;
75
	struct mutex recv_mutex;
76
	struct dlm_lock_resource *bitmap_lockres;
77
	struct dlm_lock_resource **other_bitmap_lockres;
78
	struct dlm_lock_resource *resync_lockres;
79 80
	struct list_head suspend_list;
	spinlock_t suspend_lock;
81 82
	struct md_thread *recovery_thread;
	unsigned long recovery_map;
83 84 85 86
	/* communication loc resources */
	struct dlm_lock_resource *ack_lockres;
	struct dlm_lock_resource *message_lockres;
	struct dlm_lock_resource *token_lockres;
87
	struct dlm_lock_resource *no_new_dev_lockres;
88
	struct md_thread *recv_thread;
89
	struct completion newdisk_completion;
90
	wait_queue_head_t wait;
91
	unsigned long state;
92 93 94
	/* record the region in RESYNCING message */
	sector_t sync_low;
	sector_t sync_hi;
95 96 97 98 99
};

enum msg_type {
	METADATA_UPDATED = 0,
	RESYNCING,
100
	NEWDISK,
101
	REMOVE,
102
	RE_ADD,
103
	BITMAP_NEEDS_SYNC,
104 105 106
};

struct cluster_msg {
107 108
	__le32 type;
	__le32 slot;
109
	/* TODO: Unionize this for smaller footprint */
110 111
	__le64 low;
	__le64 high;
112
	char uuid[16];
113
	__le32 raid_slot;
114 115 116 117 118 119
};

static void sync_ast(void *arg)
{
	struct dlm_lock_resource *res;

120
	res = arg;
121 122 123 124 125 126 127 128 129 130 131 132 133
	complete(&res->completion);
}

static int dlm_lock_sync(struct dlm_lock_resource *res, int mode)
{
	int ret = 0;

	ret = dlm_lock(res->ls, mode, &res->lksb,
			res->flags, res->name, strlen(res->name),
			0, sync_ast, res, res->bast);
	if (ret)
		return ret;
	wait_for_completion(&res->completion);
134 135
	if (res->lksb.sb_status == 0)
		res->mode = mode;
136 137 138 139 140 141 142 143
	return res->lksb.sb_status;
}

static int dlm_unlock_sync(struct dlm_lock_resource *res)
{
	return dlm_lock_sync(res, DLM_LOCK_NL);
}

G
Goldwyn Rodrigues 已提交
144
static struct dlm_lock_resource *lockres_init(struct mddev *mddev,
145 146 147 148
		char *name, void (*bastfn)(void *arg, int mode), int with_lvb)
{
	struct dlm_lock_resource *res = NULL;
	int ret, namelen;
G
Goldwyn Rodrigues 已提交
149
	struct md_cluster_info *cinfo = mddev->cluster_info;
150 151 152 153

	res = kzalloc(sizeof(struct dlm_lock_resource), GFP_KERNEL);
	if (!res)
		return NULL;
154
	init_completion(&res->completion);
G
Goldwyn Rodrigues 已提交
155 156
	res->ls = cinfo->lockspace;
	res->mddev = mddev;
157
	res->mode = DLM_LOCK_IV;
158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196
	namelen = strlen(name);
	res->name = kzalloc(namelen + 1, GFP_KERNEL);
	if (!res->name) {
		pr_err("md-cluster: Unable to allocate resource name for resource %s\n", name);
		goto out_err;
	}
	strlcpy(res->name, name, namelen + 1);
	if (with_lvb) {
		res->lksb.sb_lvbptr = kzalloc(LVB_SIZE, GFP_KERNEL);
		if (!res->lksb.sb_lvbptr) {
			pr_err("md-cluster: Unable to allocate LVB for resource %s\n", name);
			goto out_err;
		}
		res->flags = DLM_LKF_VALBLK;
	}

	if (bastfn)
		res->bast = bastfn;

	res->flags |= DLM_LKF_EXPEDITE;

	ret = dlm_lock_sync(res, DLM_LOCK_NL);
	if (ret) {
		pr_err("md-cluster: Unable to lock NL on new lock resource %s\n", name);
		goto out_err;
	}
	res->flags &= ~DLM_LKF_EXPEDITE;
	res->flags |= DLM_LKF_CONVERT;

	return res;
out_err:
	kfree(res->lksb.sb_lvbptr);
	kfree(res->name);
	kfree(res);
	return NULL;
}

static void lockres_free(struct dlm_lock_resource *res)
{
197 198
	int ret;

199 200 201
	if (!res)
		return;

202 203 204 205 206 207 208 209 210 211 212 213 214
	/* cancel a lock request or a conversion request that is blocked */
	res->flags |= DLM_LKF_CANCEL;
retry:
	ret = dlm_unlock(res->ls, res->lksb.sb_lkid, 0, &res->lksb, res);
	if (unlikely(ret != 0)) {
		pr_info("%s: failed to unlock %s return %d\n", __func__, res->name, ret);

		/* if a lock conversion is cancelled, then the lock is put
		 * back to grant queue, need to ensure it is unlocked */
		if (ret == -DLM_ECANCEL)
			goto retry;
	}
	res->flags &= ~DLM_LKF_CANCEL;
215 216 217 218 219 220
	wait_for_completion(&res->completion);

	kfree(res->name);
	kfree(res->lksb.sb_lvbptr);
	kfree(res);
}
221

222 223
static void add_resync_info(struct dlm_lock_resource *lockres,
			    sector_t lo, sector_t hi)
224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240
{
	struct resync_info *ri;

	ri = (struct resync_info *)lockres->lksb.sb_lvbptr;
	ri->lo = cpu_to_le64(lo);
	ri->hi = cpu_to_le64(hi);
}

static struct suspend_info *read_resync_info(struct mddev *mddev, struct dlm_lock_resource *lockres)
{
	struct resync_info ri;
	struct suspend_info *s = NULL;
	sector_t hi = 0;

	dlm_lock_sync(lockres, DLM_LOCK_CR);
	memcpy(&ri, lockres->lksb.sb_lvbptr, sizeof(struct resync_info));
	hi = le64_to_cpu(ri.hi);
241
	if (hi > 0) {
242 243 244 245 246 247 248 249 250 251 252
		s = kzalloc(sizeof(struct suspend_info), GFP_KERNEL);
		if (!s)
			goto out;
		s->hi = hi;
		s->lo = le64_to_cpu(ri.lo);
	}
	dlm_unlock_sync(lockres);
out:
	return s;
}

253
static void recover_bitmaps(struct md_thread *thread)
254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287
{
	struct mddev *mddev = thread->mddev;
	struct md_cluster_info *cinfo = mddev->cluster_info;
	struct dlm_lock_resource *bm_lockres;
	char str[64];
	int slot, ret;
	struct suspend_info *s, *tmp;
	sector_t lo, hi;

	while (cinfo->recovery_map) {
		slot = fls64((u64)cinfo->recovery_map) - 1;

		/* Clear suspend_area associated with the bitmap */
		spin_lock_irq(&cinfo->suspend_lock);
		list_for_each_entry_safe(s, tmp, &cinfo->suspend_list, list)
			if (slot == s->slot) {
				list_del(&s->list);
				kfree(s);
			}
		spin_unlock_irq(&cinfo->suspend_lock);

		snprintf(str, 64, "bitmap%04d", slot);
		bm_lockres = lockres_init(mddev, str, NULL, 1);
		if (!bm_lockres) {
			pr_err("md-cluster: Cannot initialize bitmaps\n");
			goto clear_bit;
		}

		ret = dlm_lock_sync(bm_lockres, DLM_LOCK_PW);
		if (ret) {
			pr_err("md-cluster: Could not DLM lock %s: %d\n",
					str, ret);
			goto clear_bit;
		}
288
		ret = bitmap_copy_from_slot(mddev, slot, &lo, &hi, true);
289
		if (ret) {
290
			pr_err("md-cluster: Could not copy data from bitmap %d\n", slot);
291 292 293 294 295
			goto dlm_unlock;
		}
		if (hi > 0) {
			if (lo < mddev->recovery_cp)
				mddev->recovery_cp = lo;
296 297 298 299 300 301
			/* wake up thread to continue resync in case resync
			 * is not finished */
			if (mddev->recovery_cp != MaxSector) {
			    set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
			    md_wakeup_thread(mddev->thread);
			}
302 303
		}
dlm_unlock:
304 305
		dlm_unlock_sync(bm_lockres);
clear_bit:
S
Shaohua Li 已提交
306
		lockres_free(bm_lockres);
307 308 309 310
		clear_bit(slot, &cinfo->recovery_map);
	}
}

G
Goldwyn Rodrigues 已提交
311 312
static void recover_prep(void *arg)
{
313 314 315
	struct mddev *mddev = arg;
	struct md_cluster_info *cinfo = mddev->cluster_info;
	set_bit(MD_CLUSTER_SUSPEND_READ_BALANCING, &cinfo->state);
G
Goldwyn Rodrigues 已提交
316 317
}

318
static void __recover_slot(struct mddev *mddev, int slot)
G
Goldwyn Rodrigues 已提交
319 320 321
{
	struct md_cluster_info *cinfo = mddev->cluster_info;

322
	set_bit(slot, &cinfo->recovery_map);
323 324 325 326 327 328 329 330 331
	if (!cinfo->recovery_thread) {
		cinfo->recovery_thread = md_register_thread(recover_bitmaps,
				mddev, "recover");
		if (!cinfo->recovery_thread) {
			pr_warn("md-cluster: Could not create recovery thread\n");
			return;
		}
	}
	md_wakeup_thread(cinfo->recovery_thread);
G
Goldwyn Rodrigues 已提交
332 333
}

334 335 336 337 338 339 340 341 342 343 344 345 346 347
static void recover_slot(void *arg, struct dlm_slot *slot)
{
	struct mddev *mddev = arg;
	struct md_cluster_info *cinfo = mddev->cluster_info;

	pr_info("md-cluster: %s Node %d/%d down. My slot: %d. Initiating recovery.\n",
			mddev->bitmap_info.cluster_name,
			slot->nodeid, slot->slot,
			cinfo->slot_number);
	/* deduct one since dlm slot starts from one while the num of
	 * cluster-md begins with 0 */
	__recover_slot(mddev, slot->slot - 1);
}

G
Goldwyn Rodrigues 已提交
348 349 350 351 352 353 354 355
static void recover_done(void *arg, struct dlm_slot *slots,
		int num_slots, int our_slot,
		uint32_t generation)
{
	struct mddev *mddev = arg;
	struct md_cluster_info *cinfo = mddev->cluster_info;

	cinfo->slot_number = our_slot;
356 357 358 359 360 361
	/* completion is only need to be complete when node join cluster,
	 * it doesn't need to run during another node's failure */
	if (test_bit(MD_CLUSTER_BEGIN_JOIN_CLUSTER, &cinfo->state)) {
		complete(&cinfo->completion);
		clear_bit(MD_CLUSTER_BEGIN_JOIN_CLUSTER, &cinfo->state);
	}
362
	clear_bit(MD_CLUSTER_SUSPEND_READ_BALANCING, &cinfo->state);
G
Goldwyn Rodrigues 已提交
363 364
}

365 366
/* the ops is called when node join the cluster, and do lock recovery
 * if node failure occurs */
G
Goldwyn Rodrigues 已提交
367 368 369 370 371 372
static const struct dlm_lockspace_ops md_ls_ops = {
	.recover_prep = recover_prep,
	.recover_slot = recover_slot,
	.recover_done = recover_done,
};

373 374 375 376 377 378 379
/*
 * The BAST function for the ack lock resource
 * This function wakes up the receive thread in
 * order to receive and process the message.
 */
static void ack_bast(void *arg, int mode)
{
380
	struct dlm_lock_resource *res = arg;
381 382
	struct md_cluster_info *cinfo = res->mddev->cluster_info;

383 384 385 386 387 388
	if (mode == DLM_LOCK_EX) {
		if (test_bit(MD_CLUSTER_ALREADY_IN_CLUSTER, &cinfo->state))
			md_wakeup_thread(cinfo->recv_thread);
		else
			set_bit(MD_CLUSTER_PENDING_RECV_EVENT, &cinfo->state);
	}
389 390
}

G
Goldwyn Rodrigues 已提交
391 392 393 394 395 396 397 398 399 400 401 402
static void __remove_suspend_info(struct md_cluster_info *cinfo, int slot)
{
	struct suspend_info *s, *tmp;

	list_for_each_entry_safe(s, tmp, &cinfo->suspend_list, list)
		if (slot == s->slot) {
			list_del(&s->list);
			kfree(s);
			break;
		}
}

403
static void remove_suspend_info(struct mddev *mddev, int slot)
G
Goldwyn Rodrigues 已提交
404
{
405
	struct md_cluster_info *cinfo = mddev->cluster_info;
G
Goldwyn Rodrigues 已提交
406 407 408
	spin_lock_irq(&cinfo->suspend_lock);
	__remove_suspend_info(cinfo, slot);
	spin_unlock_irq(&cinfo->suspend_lock);
409
	mddev->pers->quiesce(mddev, 2);
G
Goldwyn Rodrigues 已提交
410 411 412
}


413
static void process_suspend_info(struct mddev *mddev,
G
Goldwyn Rodrigues 已提交
414 415
		int slot, sector_t lo, sector_t hi)
{
416
	struct md_cluster_info *cinfo = mddev->cluster_info;
G
Goldwyn Rodrigues 已提交
417 418 419
	struct suspend_info *s;

	if (!hi) {
420
		remove_suspend_info(mddev, slot);
421 422
		set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
		md_wakeup_thread(mddev->thread);
G
Goldwyn Rodrigues 已提交
423 424
		return;
	}
425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448

	/*
	 * The bitmaps are not same for different nodes
	 * if RESYNCING is happening in one node, then
	 * the node which received the RESYNCING message
	 * probably will perform resync with the region
	 * [lo, hi] again, so we could reduce resync time
	 * a lot if we can ensure that the bitmaps among
	 * different nodes are match up well.
	 *
	 * sync_low/hi is used to record the region which
	 * arrived in the previous RESYNCING message,
	 *
	 * Call bitmap_sync_with_cluster to clear
	 * NEEDED_MASK and set RESYNC_MASK since
	 * resync thread is running in another node,
	 * so we don't need to do the resync again
	 * with the same section */
	bitmap_sync_with_cluster(mddev, cinfo->sync_low,
					cinfo->sync_hi,
					lo, hi);
	cinfo->sync_low = lo;
	cinfo->sync_hi = hi;

G
Goldwyn Rodrigues 已提交
449 450 451 452 453 454
	s = kzalloc(sizeof(struct suspend_info), GFP_KERNEL);
	if (!s)
		return;
	s->slot = slot;
	s->lo = lo;
	s->hi = hi;
455 456
	mddev->pers->quiesce(mddev, 1);
	mddev->pers->quiesce(mddev, 0);
G
Goldwyn Rodrigues 已提交
457 458 459 460 461
	spin_lock_irq(&cinfo->suspend_lock);
	/* Remove existing entry (if exists) before adding */
	__remove_suspend_info(cinfo, slot);
	list_add(&s->list, &cinfo->suspend_list);
	spin_unlock_irq(&cinfo->suspend_lock);
462
	mddev->pers->quiesce(mddev, 2);
G
Goldwyn Rodrigues 已提交
463 464
}

465 466 467 468 469 470 471 472 473 474
static void process_add_new_disk(struct mddev *mddev, struct cluster_msg *cmsg)
{
	char disk_uuid[64];
	struct md_cluster_info *cinfo = mddev->cluster_info;
	char event_name[] = "EVENT=ADD_DEVICE";
	char raid_slot[16];
	char *envp[] = {event_name, disk_uuid, raid_slot, NULL};
	int len;

	len = snprintf(disk_uuid, 64, "DEVICE_UUID=");
475
	sprintf(disk_uuid + len, "%pU", cmsg->uuid);
476
	snprintf(raid_slot, 16, "RAID_DISK=%d", le32_to_cpu(cmsg->raid_slot));
477 478
	pr_info("%s:%d Sending kobject change with %s and %s\n", __func__, __LINE__, disk_uuid, raid_slot);
	init_completion(&cinfo->newdisk_completion);
479
	set_bit(MD_CLUSTER_WAITING_FOR_NEWDISK, &cinfo->state);
480 481 482
	kobject_uevent_env(&disk_to_dev(mddev->gendisk)->kobj, KOBJ_CHANGE, envp);
	wait_for_completion_timeout(&cinfo->newdisk_completion,
			NEW_DEV_TIMEOUT);
483
	clear_bit(MD_CLUSTER_WAITING_FOR_NEWDISK, &cinfo->state);
484 485 486 487 488 489
}


static void process_metadata_update(struct mddev *mddev, struct cluster_msg *msg)
{
	struct md_cluster_info *cinfo = mddev->cluster_info;
490 491
	mddev->good_device_nr = le32_to_cpu(msg->raid_slot);
	set_bit(MD_RELOAD_SB, &mddev->flags);
492
	dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR);
493
	md_wakeup_thread(mddev->thread);
494 495
}

496 497
static void process_remove_disk(struct mddev *mddev, struct cluster_msg *msg)
{
498 499
	struct md_rdev *rdev = md_find_rdev_nr_rcu(mddev,
						   le32_to_cpu(msg->raid_slot));
500

501 502 503 504 505
	if (rdev) {
		set_bit(ClusterRemove, &rdev->flags);
		set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
		md_wakeup_thread(mddev->thread);
	}
506
	else
507 508
		pr_warn("%s: %d Could not find disk(%d) to REMOVE\n",
			__func__, __LINE__, le32_to_cpu(msg->raid_slot));
509 510
}

511 512
static void process_readd_disk(struct mddev *mddev, struct cluster_msg *msg)
{
513 514
	struct md_rdev *rdev = md_find_rdev_nr_rcu(mddev,
						   le32_to_cpu(msg->raid_slot));
515 516 517 518

	if (rdev && test_bit(Faulty, &rdev->flags))
		clear_bit(Faulty, &rdev->flags);
	else
519 520
		pr_warn("%s: %d Could not find disk(%d) which is faulty",
			__func__, __LINE__, le32_to_cpu(msg->raid_slot));
521 522
}

523 524
static void process_recvd_msg(struct mddev *mddev, struct cluster_msg *msg)
{
525 526 527
	if (WARN(mddev->cluster_info->slot_number - 1 == le32_to_cpu(msg->slot),
		"node %d received it's own msg\n", le32_to_cpu(msg->slot)))
		return;
528
	switch (le32_to_cpu(msg->type)) {
529
	case METADATA_UPDATED:
530
		process_metadata_update(mddev, msg);
531 532
		break;
	case RESYNCING:
533 534 535
		process_suspend_info(mddev, le32_to_cpu(msg->slot),
				     le64_to_cpu(msg->low),
				     le64_to_cpu(msg->high));
536
		break;
537 538
	case NEWDISK:
		process_add_new_disk(mddev, msg);
539 540 541 542
		break;
	case REMOVE:
		process_remove_disk(mddev, msg);
		break;
543 544 545
	case RE_ADD:
		process_readd_disk(mddev, msg);
		break;
546
	case BITMAP_NEEDS_SYNC:
547
		__recover_slot(mddev, le32_to_cpu(msg->slot));
548
		break;
549 550 551
	default:
		pr_warn("%s:%d Received unknown message from %d\n",
			__func__, __LINE__, msg->slot);
552
	}
553 554 555 556 557 558 559 560 561 562 563
}

/*
 * thread for receiving message
 */
static void recv_daemon(struct md_thread *thread)
{
	struct md_cluster_info *cinfo = thread->mddev->cluster_info;
	struct dlm_lock_resource *ack_lockres = cinfo->ack_lockres;
	struct dlm_lock_resource *message_lockres = cinfo->message_lockres;
	struct cluster_msg msg;
564
	int ret;
565

566
	mutex_lock(&cinfo->recv_mutex);
567 568 569
	/*get CR on Message*/
	if (dlm_lock_sync(message_lockres, DLM_LOCK_CR)) {
		pr_err("md/raid1:failed to get CR on MESSAGE\n");
570
		mutex_unlock(&cinfo->recv_mutex);
571 572 573 574 575 576 577 578
		return;
	}

	/* read lvb and wake up thread to process this message_lockres */
	memcpy(&msg, message_lockres->lksb.sb_lvbptr, sizeof(struct cluster_msg));
	process_recvd_msg(thread->mddev, &msg);

	/*release CR on ack_lockres*/
579 580 581
	ret = dlm_unlock_sync(ack_lockres);
	if (unlikely(ret != 0))
		pr_info("unlock ack failed return %d\n", ret);
582
	/*up-convert to PR on message_lockres*/
583 584 585
	ret = dlm_lock_sync(message_lockres, DLM_LOCK_PR);
	if (unlikely(ret != 0))
		pr_info("lock PR on msg failed return %d\n", ret);
586
	/*get CR on ack_lockres again*/
587 588 589
	ret = dlm_lock_sync(ack_lockres, DLM_LOCK_CR);
	if (unlikely(ret != 0))
		pr_info("lock CR on ack failed return %d\n", ret);
590
	/*release CR on message_lockres*/
591 592 593
	ret = dlm_unlock_sync(message_lockres);
	if (unlikely(ret != 0))
		pr_info("unlock msg failed return %d\n", ret);
594
	mutex_unlock(&cinfo->recv_mutex);
595 596
}

597
/* lock_token()
598 599 600
 * Takes the lock on the TOKEN lock resource so no other
 * node can communicate while the operation is underway.
 */
601
static int lock_token(struct md_cluster_info *cinfo)
602 603 604 605 606 607 608
{
	int error;

	error = dlm_lock_sync(cinfo->token_lockres, DLM_LOCK_EX);
	if (error)
		pr_err("md-cluster(%s:%d): failed to get EX on TOKEN (%d)\n",
				__func__, __LINE__, error);
609 610 611

	/* Lock the receive sequence */
	mutex_lock(&cinfo->recv_mutex);
612 613 614
	return error;
}

615 616 617 618 619 620 621 622 623 624 625
/* lock_comm()
 * Sets the MD_CLUSTER_SEND_LOCK bit to lock the send channel.
 */
static int lock_comm(struct md_cluster_info *cinfo)
{
	wait_event(cinfo->wait,
		   !test_and_set_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state));

	return lock_token(cinfo);
}

626 627
static void unlock_comm(struct md_cluster_info *cinfo)
{
628
	WARN_ON(cinfo->token_lockres->mode != DLM_LOCK_EX);
629
	mutex_unlock(&cinfo->recv_mutex);
630
	dlm_unlock_sync(cinfo->token_lockres);
631 632
	clear_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state);
	wake_up(&cinfo->wait);
633 634 635 636 637 638 639 640
}

/* __sendmsg()
 * This function performs the actual sending of the message. This function is
 * usually called after performing the encompassing operation
 * The function:
 * 1. Grabs the message lockresource in EX mode
 * 2. Copies the message to the message LVB
641
 * 3. Downconverts message lockresource to CW
642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661
 * 4. Upconverts ack lock resource from CR to EX. This forces the BAST on other nodes
 *    and the other nodes read the message. The thread will wait here until all other
 *    nodes have released ack lock resource.
 * 5. Downconvert ack lockresource to CR
 */
static int __sendmsg(struct md_cluster_info *cinfo, struct cluster_msg *cmsg)
{
	int error;
	int slot = cinfo->slot_number - 1;

	cmsg->slot = cpu_to_le32(slot);
	/*get EX on Message*/
	error = dlm_lock_sync(cinfo->message_lockres, DLM_LOCK_EX);
	if (error) {
		pr_err("md-cluster: failed to get EX on MESSAGE (%d)\n", error);
		goto failed_message;
	}

	memcpy(cinfo->message_lockres->lksb.sb_lvbptr, (void *)cmsg,
			sizeof(struct cluster_msg));
662 663
	/*down-convert EX to CW on Message*/
	error = dlm_lock_sync(cinfo->message_lockres, DLM_LOCK_CW);
664
	if (error) {
665
		pr_err("md-cluster: failed to convert EX to CW on MESSAGE(%d)\n",
666
				error);
667
		goto failed_ack;
668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686
	}

	/*up-convert CR to EX on Ack*/
	error = dlm_lock_sync(cinfo->ack_lockres, DLM_LOCK_EX);
	if (error) {
		pr_err("md-cluster: failed to convert CR to EX on ACK(%d)\n",
				error);
		goto failed_ack;
	}

	/*down-convert EX to CR on Ack*/
	error = dlm_lock_sync(cinfo->ack_lockres, DLM_LOCK_CR);
	if (error) {
		pr_err("md-cluster: failed to convert EX to CR on ACK(%d)\n",
				error);
		goto failed_ack;
	}

failed_ack:
687 688 689 690 691 692 693
	error = dlm_unlock_sync(cinfo->message_lockres);
	if (unlikely(error != 0)) {
		pr_err("md-cluster: failed convert to NL on MESSAGE(%d)\n",
			error);
		/* in case the message can't be released due to some reason */
		goto failed_ack;
	}
694 695 696 697 698 699 700 701 702 703 704 705 706 707
failed_message:
	return error;
}

static int sendmsg(struct md_cluster_info *cinfo, struct cluster_msg *cmsg)
{
	int ret;

	lock_comm(cinfo);
	ret = __sendmsg(cinfo, cmsg);
	unlock_comm(cinfo);
	return ret;
}

708 709 710 711 712 713 714
static int gather_all_resync_info(struct mddev *mddev, int total_slots)
{
	struct md_cluster_info *cinfo = mddev->cluster_info;
	int i, ret = 0;
	struct dlm_lock_resource *bm_lockres;
	struct suspend_info *s;
	char str[64];
715
	sector_t lo, hi;
716 717 718 719 720 721 722 723


	for (i = 0; i < total_slots; i++) {
		memset(str, '\0', 64);
		snprintf(str, 64, "bitmap%04d", i);
		bm_lockres = lockres_init(mddev, str, NULL, 1);
		if (!bm_lockres)
			return -ENOMEM;
S
Shaohua Li 已提交
724 725
		if (i == (cinfo->slot_number - 1)) {
			lockres_free(bm_lockres);
726
			continue;
S
Shaohua Li 已提交
727
		}
728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747

		bm_lockres->flags |= DLM_LKF_NOQUEUE;
		ret = dlm_lock_sync(bm_lockres, DLM_LOCK_PW);
		if (ret == -EAGAIN) {
			memset(bm_lockres->lksb.sb_lvbptr, '\0', LVB_SIZE);
			s = read_resync_info(mddev, bm_lockres);
			if (s) {
				pr_info("%s:%d Resync[%llu..%llu] in progress on %d\n",
						__func__, __LINE__,
						(unsigned long long) s->lo,
						(unsigned long long) s->hi, i);
				spin_lock_irq(&cinfo->suspend_lock);
				s->slot = i;
				list_add(&s->list, &cinfo->suspend_list);
				spin_unlock_irq(&cinfo->suspend_lock);
			}
			ret = 0;
			lockres_free(bm_lockres);
			continue;
		}
748 749
		if (ret) {
			lockres_free(bm_lockres);
750
			goto out;
751
		}
752 753 754 755 756 757 758 759 760 761 762 763 764 765

		/* Read the disk bitmap sb and check if it needs recovery */
		ret = bitmap_copy_from_slot(mddev, i, &lo, &hi, false);
		if (ret) {
			pr_warn("md-cluster: Could not gather bitmaps from slot %d", i);
			lockres_free(bm_lockres);
			continue;
		}
		if ((hi > 0) && (lo < mddev->recovery_cp)) {
			set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
			mddev->recovery_cp = lo;
			md_check_recovery(mddev);
		}

766 767 768 769 770 771 772
		dlm_unlock_sync(bm_lockres);
		lockres_free(bm_lockres);
	}
out:
	return ret;
}

773 774
static int join(struct mddev *mddev, int nodes)
{
G
Goldwyn Rodrigues 已提交
775
	struct md_cluster_info *cinfo;
G
Goldwyn Rodrigues 已提交
776
	int ret, ops_rv;
G
Goldwyn Rodrigues 已提交
777 778 779 780 781 782
	char str[64];

	cinfo = kzalloc(sizeof(struct md_cluster_info), GFP_KERNEL);
	if (!cinfo)
		return -ENOMEM;

783 784
	INIT_LIST_HEAD(&cinfo->suspend_list);
	spin_lock_init(&cinfo->suspend_lock);
G
Goldwyn Rodrigues 已提交
785
	init_completion(&cinfo->completion);
786
	set_bit(MD_CLUSTER_BEGIN_JOIN_CLUSTER, &cinfo->state);
787 788
	init_waitqueue_head(&cinfo->wait);
	mutex_init(&cinfo->recv_mutex);
G
Goldwyn Rodrigues 已提交
789 790 791

	mddev->cluster_info = cinfo;

G
Goldwyn Rodrigues 已提交
792
	memset(str, 0, 64);
793
	sprintf(str, "%pU", mddev->uuid);
G
Goldwyn Rodrigues 已提交
794 795 796
	ret = dlm_new_lockspace(str, mddev->bitmap_info.cluster_name,
				DLM_LSFL_FS, LVB_SIZE,
				&md_ls_ops, mddev, &ops_rv, &cinfo->lockspace);
G
Goldwyn Rodrigues 已提交
797 798
	if (ret)
		goto err;
G
Goldwyn Rodrigues 已提交
799
	wait_for_completion(&cinfo->completion);
800 801 802
	if (nodes < cinfo->slot_number) {
		pr_err("md-cluster: Slot allotted(%d) is greater than available slots(%d).",
			cinfo->slot_number, nodes);
803 804 805
		ret = -ERANGE;
		goto err;
	}
806 807 808 809 810 811 812 813 814 815 816 817 818
	/* Initiate the communication resources */
	ret = -ENOMEM;
	cinfo->recv_thread = md_register_thread(recv_daemon, mddev, "cluster_recv");
	if (!cinfo->recv_thread) {
		pr_err("md-cluster: cannot allocate memory for recv_thread!\n");
		goto err;
	}
	cinfo->message_lockres = lockres_init(mddev, "message", NULL, 1);
	if (!cinfo->message_lockres)
		goto err;
	cinfo->token_lockres = lockres_init(mddev, "token", NULL, 0);
	if (!cinfo->token_lockres)
		goto err;
819 820 821 822
	cinfo->no_new_dev_lockres = lockres_init(mddev, "no-new-dev", NULL, 0);
	if (!cinfo->no_new_dev_lockres)
		goto err;

823 824 825 826 827 828 829 830 831
	ret = dlm_lock_sync(cinfo->token_lockres, DLM_LOCK_EX);
	if (ret) {
		ret = -EAGAIN;
		pr_err("md-cluster: can't join cluster to avoid lock issue\n");
		goto err;
	}
	cinfo->ack_lockres = lockres_init(mddev, "ack", ack_bast, 0);
	if (!cinfo->ack_lockres)
		goto err;
832 833 834 835
	/* get sync CR lock on ACK. */
	if (dlm_lock_sync(cinfo->ack_lockres, DLM_LOCK_CR))
		pr_err("md-cluster: failed to get a sync CR lock on ACK!(%d)\n",
				ret);
836
	dlm_unlock_sync(cinfo->token_lockres);
837 838 839 840
	/* get sync CR lock on no-new-dev. */
	if (dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR))
		pr_err("md-cluster: failed to get a sync CR lock on no-new-dev!(%d)\n", ret);

841 842 843 844 845 846 847 848 849 850 851 852

	pr_info("md-cluster: Joined cluster %s slot %d\n", str, cinfo->slot_number);
	snprintf(str, 64, "bitmap%04d", cinfo->slot_number - 1);
	cinfo->bitmap_lockres = lockres_init(mddev, str, NULL, 1);
	if (!cinfo->bitmap_lockres)
		goto err;
	if (dlm_lock_sync(cinfo->bitmap_lockres, DLM_LOCK_PW)) {
		pr_err("Failed to get bitmap lock\n");
		ret = -EINVAL;
		goto err;
	}

853 854 855 856
	cinfo->resync_lockres = lockres_init(mddev, "resync", NULL, 0);
	if (!cinfo->resync_lockres)
		goto err;

857
	return 0;
G
Goldwyn Rodrigues 已提交
858
err:
859 860
	md_unregister_thread(&cinfo->recovery_thread);
	md_unregister_thread(&cinfo->recv_thread);
861 862 863
	lockres_free(cinfo->message_lockres);
	lockres_free(cinfo->token_lockres);
	lockres_free(cinfo->ack_lockres);
864
	lockres_free(cinfo->no_new_dev_lockres);
865
	lockres_free(cinfo->resync_lockres);
866
	lockres_free(cinfo->bitmap_lockres);
G
Goldwyn Rodrigues 已提交
867 868
	if (cinfo->lockspace)
		dlm_release_lockspace(cinfo->lockspace, 2);
G
Goldwyn Rodrigues 已提交
869
	mddev->cluster_info = NULL;
G
Goldwyn Rodrigues 已提交
870 871
	kfree(cinfo);
	return ret;
872 873
}

874 875 876 877 878 879 880 881 882 883 884 885 886
static void load_bitmaps(struct mddev *mddev, int total_slots)
{
	struct md_cluster_info *cinfo = mddev->cluster_info;

	/* load all the node's bitmap info for resync */
	if (gather_all_resync_info(mddev, total_slots))
		pr_err("md-cluster: failed to gather all resyn infos\n");
	set_bit(MD_CLUSTER_ALREADY_IN_CLUSTER, &cinfo->state);
	/* wake up recv thread in case something need to be handled */
	if (test_and_clear_bit(MD_CLUSTER_PENDING_RECV_EVENT, &cinfo->state))
		md_wakeup_thread(cinfo->recv_thread);
}

887 888 889 890 891 892 893 894 895 896 897 898 899
static void resync_bitmap(struct mddev *mddev)
{
	struct md_cluster_info *cinfo = mddev->cluster_info;
	struct cluster_msg cmsg = {0};
	int err;

	cmsg.type = cpu_to_le32(BITMAP_NEEDS_SYNC);
	err = sendmsg(cinfo, &cmsg);
	if (err)
		pr_err("%s:%d: failed to send BITMAP_NEEDS_SYNC message (%d)\n",
			__func__, __LINE__, err);
}

900
static void unlock_all_bitmaps(struct mddev *mddev);
901 902
static int leave(struct mddev *mddev)
{
G
Goldwyn Rodrigues 已提交
903 904 905 906
	struct md_cluster_info *cinfo = mddev->cluster_info;

	if (!cinfo)
		return 0;
907 908 909 910 911 912 913

	/* BITMAP_NEEDS_SYNC message should be sent when node
	 * is leaving the cluster with dirty bitmap, also we
	 * can only deliver it when dlm connection is available */
	if (cinfo->slot_number > 0 && mddev->recovery_cp != MaxSector)
		resync_bitmap(mddev);

914
	md_unregister_thread(&cinfo->recovery_thread);
915 916 917 918
	md_unregister_thread(&cinfo->recv_thread);
	lockres_free(cinfo->message_lockres);
	lockres_free(cinfo->token_lockres);
	lockres_free(cinfo->ack_lockres);
919
	lockres_free(cinfo->no_new_dev_lockres);
S
Shaohua Li 已提交
920
	lockres_free(cinfo->resync_lockres);
921
	lockres_free(cinfo->bitmap_lockres);
922
	unlock_all_bitmaps(mddev);
G
Goldwyn Rodrigues 已提交
923
	dlm_release_lockspace(cinfo->lockspace, 2);
924 925 926
	return 0;
}

G
Goldwyn Rodrigues 已提交
927 928 929 930 931 932 933 934 935 936 937
/* slot_number(): Returns the MD slot number to use
 * DLM starts the slot numbers from 1, wheras cluster-md
 * wants the number to be from zero, so we deduct one
 */
static int slot_number(struct mddev *mddev)
{
	struct md_cluster_info *cinfo = mddev->cluster_info;

	return cinfo->slot_number - 1;
}

938 939 940 941 942 943
/*
 * Check if the communication is already locked, else lock the communication
 * channel.
 * If it is already locked, token is in EX mode, and hence lock_token()
 * should not be called.
 */
944 945
static int metadata_update_start(struct mddev *mddev)
{
946 947 948 949 950 951 952 953 954 955 956
	struct md_cluster_info *cinfo = mddev->cluster_info;

	wait_event(cinfo->wait,
		   !test_and_set_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state) ||
		   test_and_clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state));

	/* If token is already locked, return 0 */
	if (cinfo->token_lockres->mode == DLM_LOCK_EX)
		return 0;

	return lock_token(cinfo);
957 958 959 960 961 962
}

static int metadata_update_finish(struct mddev *mddev)
{
	struct md_cluster_info *cinfo = mddev->cluster_info;
	struct cluster_msg cmsg;
963 964
	struct md_rdev *rdev;
	int ret = 0;
965
	int raid_slot = -1;
966 967 968

	memset(&cmsg, 0, sizeof(cmsg));
	cmsg.type = cpu_to_le32(METADATA_UPDATED);
969 970 971 972
	/* Pick up a good active device number to send.
	 */
	rdev_for_each(rdev, mddev)
		if (rdev->raid_disk > -1 && !test_bit(Faulty, &rdev->flags)) {
973
			raid_slot = rdev->desc_nr;
974 975
			break;
		}
976 977
	if (raid_slot >= 0) {
		cmsg.raid_slot = cpu_to_le32(raid_slot);
978
		ret = __sendmsg(cinfo, &cmsg);
979
	} else
980
		pr_warn("md-cluster: No good device id found to send\n");
981
	clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
982 983 984 985
	unlock_comm(cinfo);
	return ret;
}

986
static void metadata_update_cancel(struct mddev *mddev)
987 988
{
	struct md_cluster_info *cinfo = mddev->cluster_info;
989
	clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
990
	unlock_comm(cinfo);
991 992
}

993 994 995 996 997 998
static int resync_start(struct mddev *mddev)
{
	struct md_cluster_info *cinfo = mddev->cluster_info;
	return dlm_lock_sync(cinfo->resync_lockres, DLM_LOCK_EX);
}

999
static int resync_info_update(struct mddev *mddev, sector_t lo, sector_t hi)
1000 1001
{
	struct md_cluster_info *cinfo = mddev->cluster_info;
1002
	struct resync_info ri;
1003
	struct cluster_msg cmsg = {0};
1004

1005 1006 1007 1008 1009 1010 1011
	/* do not send zero again, if we have sent before */
	if (hi == 0) {
		memcpy(&ri, cinfo->bitmap_lockres->lksb.sb_lvbptr, sizeof(struct resync_info));
		if (le64_to_cpu(ri.hi) == 0)
			return 0;
	}

1012
	add_resync_info(cinfo->bitmap_lockres, lo, hi);
1013 1014 1015
	/* Re-acquire the lock to refresh LVB */
	dlm_lock_sync(cinfo->bitmap_lockres, DLM_LOCK_PW);
	cmsg.type = cpu_to_le32(RESYNCING);
1016 1017
	cmsg.low = cpu_to_le64(lo);
	cmsg.high = cpu_to_le64(hi);
1018

1019 1020 1021
	return sendmsg(cinfo, &cmsg);
}

1022 1023 1024 1025 1026 1027 1028
static int resync_finish(struct mddev *mddev)
{
	struct md_cluster_info *cinfo = mddev->cluster_info;
	dlm_unlock_sync(cinfo->resync_lockres);
	return resync_info_update(mddev, 0, 0);
}

1029 1030
static int area_resyncing(struct mddev *mddev, int direction,
		sector_t lo, sector_t hi)
1031 1032 1033 1034 1035
{
	struct md_cluster_info *cinfo = mddev->cluster_info;
	int ret = 0;
	struct suspend_info *s;

1036 1037 1038 1039
	if ((direction == READ) &&
		test_bit(MD_CLUSTER_SUSPEND_READ_BALANCING, &cinfo->state))
		return 1;

1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052
	spin_lock_irq(&cinfo->suspend_lock);
	if (list_empty(&cinfo->suspend_list))
		goto out;
	list_for_each_entry(s, &cinfo->suspend_list, list)
		if (hi > s->lo && lo < s->hi) {
			ret = 1;
			break;
		}
out:
	spin_unlock_irq(&cinfo->suspend_lock);
	return ret;
}

1053 1054 1055 1056 1057
/* add_new_disk() - initiates a disk add
 * However, if this fails before writing md_update_sb(),
 * add_new_disk_cancel() must be called to release token lock
 */
static int add_new_disk(struct mddev *mddev, struct md_rdev *rdev)
1058 1059 1060 1061 1062 1063 1064 1065 1066 1067
{
	struct md_cluster_info *cinfo = mddev->cluster_info;
	struct cluster_msg cmsg;
	int ret = 0;
	struct mdp_superblock_1 *sb = page_address(rdev->sb_page);
	char *uuid = sb->device_uuid;

	memset(&cmsg, 0, sizeof(cmsg));
	cmsg.type = cpu_to_le32(NEWDISK);
	memcpy(cmsg.uuid, uuid, 16);
1068
	cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
1069 1070 1071 1072 1073 1074 1075 1076 1077 1078
	lock_comm(cinfo);
	ret = __sendmsg(cinfo, &cmsg);
	if (ret)
		return ret;
	cinfo->no_new_dev_lockres->flags |= DLM_LKF_NOQUEUE;
	ret = dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_EX);
	cinfo->no_new_dev_lockres->flags &= ~DLM_LKF_NOQUEUE;
	/* Some node does not "see" the device */
	if (ret == -EAGAIN)
		ret = -ENOENT;
1079 1080
	if (ret)
		unlock_comm(cinfo);
1081
	else {
1082
		dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR);
1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094
		/* Since MD_CHANGE_DEVS will be set in add_bound_rdev which
		 * will run soon after add_new_disk, the below path will be
		 * invoked:
		 *   md_wakeup_thread(mddev->thread)
		 *	-> conf->thread (raid1d)
		 *	-> md_check_recovery -> md_update_sb
		 *	-> metadata_update_start/finish
		 * MD_CLUSTER_SEND_LOCKED_ALREADY will be cleared eventually.
		 *
		 * For other failure cases, metadata_update_cancel and
		 * add_new_disk_cancel also clear below bit as well.
		 * */
1095 1096 1097
		set_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
		wake_up(&cinfo->wait);
	}
1098 1099 1100
	return ret;
}

1101
static void add_new_disk_cancel(struct mddev *mddev)
1102
{
1103
	struct md_cluster_info *cinfo = mddev->cluster_info;
1104
	clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state);
1105
	unlock_comm(cinfo);
1106 1107
}

1108
static int new_disk_ack(struct mddev *mddev, bool ack)
1109 1110 1111
{
	struct md_cluster_info *cinfo = mddev->cluster_info;

1112 1113 1114 1115 1116
	if (!test_bit(MD_CLUSTER_WAITING_FOR_NEWDISK, &cinfo->state)) {
		pr_warn("md-cluster(%s): Spurious cluster confirmation\n", mdname(mddev));
		return -EINVAL;
	}

1117 1118 1119
	if (ack)
		dlm_unlock_sync(cinfo->no_new_dev_lockres);
	complete(&cinfo->newdisk_completion);
1120
	return 0;
1121 1122
}

1123 1124
static int remove_disk(struct mddev *mddev, struct md_rdev *rdev)
{
1125
	struct cluster_msg cmsg = {0};
1126
	struct md_cluster_info *cinfo = mddev->cluster_info;
1127 1128
	cmsg.type = cpu_to_le32(REMOVE);
	cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
1129
	return sendmsg(cinfo, &cmsg);
1130 1131
}

1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183
static int lock_all_bitmaps(struct mddev *mddev)
{
	int slot, my_slot, ret, held = 1, i = 0;
	char str[64];
	struct md_cluster_info *cinfo = mddev->cluster_info;

	cinfo->other_bitmap_lockres = kzalloc((mddev->bitmap_info.nodes - 1) *
					     sizeof(struct dlm_lock_resource *),
					     GFP_KERNEL);
	if (!cinfo->other_bitmap_lockres) {
		pr_err("md: can't alloc mem for other bitmap locks\n");
		return 0;
	}

	my_slot = slot_number(mddev);
	for (slot = 0; slot < mddev->bitmap_info.nodes; slot++) {
		if (slot == my_slot)
			continue;

		memset(str, '\0', 64);
		snprintf(str, 64, "bitmap%04d", slot);
		cinfo->other_bitmap_lockres[i] = lockres_init(mddev, str, NULL, 1);
		if (!cinfo->other_bitmap_lockres[i])
			return -ENOMEM;

		cinfo->other_bitmap_lockres[i]->flags |= DLM_LKF_NOQUEUE;
		ret = dlm_lock_sync(cinfo->other_bitmap_lockres[i], DLM_LOCK_PW);
		if (ret)
			held = -1;
		i++;
	}

	return held;
}

static void unlock_all_bitmaps(struct mddev *mddev)
{
	struct md_cluster_info *cinfo = mddev->cluster_info;
	int i;

	/* release other node's bitmap lock if they are existed */
	if (cinfo->other_bitmap_lockres) {
		for (i = 0; i < mddev->bitmap_info.nodes - 1; i++) {
			if (cinfo->other_bitmap_lockres[i]) {
				dlm_unlock_sync(cinfo->other_bitmap_lockres[i]);
				lockres_free(cinfo->other_bitmap_lockres[i]);
			}
		}
		kfree(cinfo->other_bitmap_lockres);
	}
}

1184 1185 1186 1187
static int gather_bitmaps(struct md_rdev *rdev)
{
	int sn, err;
	sector_t lo, hi;
1188
	struct cluster_msg cmsg = {0};
1189 1190 1191
	struct mddev *mddev = rdev->mddev;
	struct md_cluster_info *cinfo = mddev->cluster_info;

1192 1193
	cmsg.type = cpu_to_le32(RE_ADD);
	cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212
	err = sendmsg(cinfo, &cmsg);
	if (err)
		goto out;

	for (sn = 0; sn < mddev->bitmap_info.nodes; sn++) {
		if (sn == (cinfo->slot_number - 1))
			continue;
		err = bitmap_copy_from_slot(mddev, sn, &lo, &hi, false);
		if (err) {
			pr_warn("md-cluster: Could not gather bitmaps from slot %d", sn);
			goto out;
		}
		if ((hi > 0) && (lo < mddev->recovery_cp))
			mddev->recovery_cp = lo;
	}
out:
	return err;
}

1213 1214 1215
static struct md_cluster_operations cluster_ops = {
	.join   = join,
	.leave  = leave,
G
Goldwyn Rodrigues 已提交
1216
	.slot_number = slot_number,
1217 1218
	.resync_start = resync_start,
	.resync_finish = resync_finish,
1219
	.resync_info_update = resync_info_update,
1220 1221 1222
	.metadata_update_start = metadata_update_start,
	.metadata_update_finish = metadata_update_finish,
	.metadata_update_cancel = metadata_update_cancel,
1223
	.area_resyncing = area_resyncing,
1224 1225
	.add_new_disk = add_new_disk,
	.add_new_disk_cancel = add_new_disk_cancel,
1226
	.new_disk_ack = new_disk_ack,
1227
	.remove_disk = remove_disk,
1228
	.load_bitmaps = load_bitmaps,
1229
	.gather_bitmaps = gather_bitmaps,
1230 1231
	.lock_all_bitmaps = lock_all_bitmaps,
	.unlock_all_bitmaps = unlock_all_bitmaps,
1232 1233
};

1234 1235 1236 1237
static int __init cluster_init(void)
{
	pr_warn("md-cluster: EXPERIMENTAL. Use with caution\n");
	pr_info("Registering Cluster MD functions\n");
1238
	register_md_cluster_operations(&cluster_ops, THIS_MODULE);
1239 1240 1241 1242 1243
	return 0;
}

static void cluster_exit(void)
{
1244
	unregister_md_cluster_operations();
1245 1246 1247 1248
}

module_init(cluster_init);
module_exit(cluster_exit);
1249
MODULE_AUTHOR("SUSE");
1250 1251
MODULE_LICENSE("GPL");
MODULE_DESCRIPTION("Clustering support for MD");