volumes.c 172.8 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */
#include <linux/sched.h>
#include <linux/bio.h>
20
#include <linux/slab.h>
21
#include <linux/buffer_head.h>
22
#include <linux/blkdev.h>
23
#include <linux/random.h>
24
#include <linux/iocontext.h>
25
#include <linux/capability.h>
26
#include <linux/ratelimit.h>
I
Ilya Dryomov 已提交
27
#include <linux/kthread.h>
D
David Woodhouse 已提交
28
#include <linux/raid/pq.h>
S
Stefan Behrens 已提交
29
#include <linux/semaphore.h>
D
David Woodhouse 已提交
30
#include <asm/div64.h>
31 32 33 34 35 36
#include "ctree.h"
#include "extent_map.h"
#include "disk-io.h"
#include "transaction.h"
#include "print-tree.h"
#include "volumes.h"
D
David Woodhouse 已提交
37
#include "raid56.h"
38
#include "async-thread.h"
39
#include "check-integrity.h"
40
#include "rcu-string.h"
41
#include "math.h"
42
#include "dev-replace.h"
43
#include "sysfs.h"
44

Y
Yan Zheng 已提交
45 46 47 48
static int init_first_rw_device(struct btrfs_trans_handle *trans,
				struct btrfs_root *root,
				struct btrfs_device *device);
static int btrfs_relocate_sys_chunks(struct btrfs_root *root);
49
static void __btrfs_reset_dev_stats(struct btrfs_device *dev);
50
static void btrfs_dev_stat_print_on_error(struct btrfs_device *dev);
51
static void btrfs_dev_stat_print_on_load(struct btrfs_device *device);
Y
Yan Zheng 已提交
52

53
DEFINE_MUTEX(uuid_mutex);
54 55
static LIST_HEAD(fs_uuids);

56 57 58 59 60 61 62 63 64 65 66
static struct btrfs_fs_devices *__alloc_fs_devices(void)
{
	struct btrfs_fs_devices *fs_devs;

	fs_devs = kzalloc(sizeof(*fs_devs), GFP_NOFS);
	if (!fs_devs)
		return ERR_PTR(-ENOMEM);

	mutex_init(&fs_devs->device_list_mutex);

	INIT_LIST_HEAD(&fs_devs->devices);
67
	INIT_LIST_HEAD(&fs_devs->resized_devices);
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98
	INIT_LIST_HEAD(&fs_devs->alloc_list);
	INIT_LIST_HEAD(&fs_devs->list);

	return fs_devs;
}

/**
 * alloc_fs_devices - allocate struct btrfs_fs_devices
 * @fsid:	a pointer to UUID for this FS.  If NULL a new UUID is
 *		generated.
 *
 * Return: a pointer to a new &struct btrfs_fs_devices on success;
 * ERR_PTR() on error.  Returned struct is not linked onto any lists and
 * can be destroyed with kfree() right away.
 */
static struct btrfs_fs_devices *alloc_fs_devices(const u8 *fsid)
{
	struct btrfs_fs_devices *fs_devs;

	fs_devs = __alloc_fs_devices();
	if (IS_ERR(fs_devs))
		return fs_devs;

	if (fsid)
		memcpy(fs_devs->fsid, fsid, BTRFS_FSID_SIZE);
	else
		generate_random_uuid(fs_devs->fsid);

	return fs_devs;
}

Y
Yan Zheng 已提交
99 100 101 102 103 104 105 106
static void free_fs_devices(struct btrfs_fs_devices *fs_devices)
{
	struct btrfs_device *device;
	WARN_ON(fs_devices->opened);
	while (!list_empty(&fs_devices->devices)) {
		device = list_entry(fs_devices->devices.next,
				    struct btrfs_device, dev_list);
		list_del(&device->dev_list);
107
		rcu_string_free(device->name);
Y
Yan Zheng 已提交
108 109 110 111 112
		kfree(device);
	}
	kfree(fs_devices);
}

113 114 115 116 117 118 119
static void btrfs_kobject_uevent(struct block_device *bdev,
				 enum kobject_action action)
{
	int ret;

	ret = kobject_uevent(&disk_to_dev(bdev->bd_disk)->kobj, action);
	if (ret)
120
		pr_warn("BTRFS: Sending event '%d' to kobject: '%s' (%p): failed\n",
121 122 123 124 125
			action,
			kobject_name(&disk_to_dev(bdev->bd_disk)->kobj),
			&disk_to_dev(bdev->bd_disk)->kobj);
}

126
void btrfs_cleanup_fs_uuids(void)
127 128 129
{
	struct btrfs_fs_devices *fs_devices;

Y
Yan Zheng 已提交
130 131 132 133
	while (!list_empty(&fs_uuids)) {
		fs_devices = list_entry(fs_uuids.next,
					struct btrfs_fs_devices, list);
		list_del(&fs_devices->list);
Y
Yan Zheng 已提交
134
		free_fs_devices(fs_devices);
135 136 137
	}
}

138 139 140 141 142 143 144 145 146 147
static struct btrfs_device *__alloc_device(void)
{
	struct btrfs_device *dev;

	dev = kzalloc(sizeof(*dev), GFP_NOFS);
	if (!dev)
		return ERR_PTR(-ENOMEM);

	INIT_LIST_HEAD(&dev->dev_list);
	INIT_LIST_HEAD(&dev->dev_alloc_list);
148
	INIT_LIST_HEAD(&dev->resized_list);
149 150 151 152 153

	spin_lock_init(&dev->io_lock);

	spin_lock_init(&dev->reada_lock);
	atomic_set(&dev->reada_in_flight, 0);
154
	atomic_set(&dev->dev_stats_ccnt, 0);
155 156 157 158 159 160
	INIT_RADIX_TREE(&dev->reada_zones, GFP_NOFS & ~__GFP_WAIT);
	INIT_RADIX_TREE(&dev->reada_extents, GFP_NOFS & ~__GFP_WAIT);

	return dev;
}

161 162
static noinline struct btrfs_device *__find_device(struct list_head *head,
						   u64 devid, u8 *uuid)
163 164 165
{
	struct btrfs_device *dev;

Q
Qinghuang Feng 已提交
166
	list_for_each_entry(dev, head, dev_list) {
167
		if (dev->devid == devid &&
168
		    (!uuid || !memcmp(dev->uuid, uuid, BTRFS_UUID_SIZE))) {
169
			return dev;
170
		}
171 172 173 174
	}
	return NULL;
}

175
static noinline struct btrfs_fs_devices *find_fsid(u8 *fsid)
176 177 178
{
	struct btrfs_fs_devices *fs_devices;

Q
Qinghuang Feng 已提交
179
	list_for_each_entry(fs_devices, &fs_uuids, list) {
180 181 182 183 184 185
		if (memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE) == 0)
			return fs_devices;
	}
	return NULL;
}

186 187 188 189 190 191 192 193 194 195 196
static int
btrfs_get_bdev_and_sb(const char *device_path, fmode_t flags, void *holder,
		      int flush, struct block_device **bdev,
		      struct buffer_head **bh)
{
	int ret;

	*bdev = blkdev_get_by_path(device_path, flags, holder);

	if (IS_ERR(*bdev)) {
		ret = PTR_ERR(*bdev);
197
		printk(KERN_INFO "BTRFS: open %s failed\n", device_path);
198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223
		goto error;
	}

	if (flush)
		filemap_write_and_wait((*bdev)->bd_inode->i_mapping);
	ret = set_blocksize(*bdev, 4096);
	if (ret) {
		blkdev_put(*bdev, flags);
		goto error;
	}
	invalidate_bdev(*bdev);
	*bh = btrfs_read_dev_super(*bdev);
	if (!*bh) {
		ret = -EINVAL;
		blkdev_put(*bdev, flags);
		goto error;
	}

	return 0;

error:
	*bdev = NULL;
	*bh = NULL;
	return ret;
}

224 225 226 227 228 229 230 231 232 233 234 235 236 237
static void requeue_list(struct btrfs_pending_bios *pending_bios,
			struct bio *head, struct bio *tail)
{

	struct bio *old_head;

	old_head = pending_bios->head;
	pending_bios->head = head;
	if (pending_bios->tail)
		tail->bi_next = old_head;
	else
		pending_bios->tail = tail;
}

238 239 240 241 242 243 244 245 246 247 248
/*
 * we try to collect pending bios for a device so we don't get a large
 * number of procs sending bios down to the same device.  This greatly
 * improves the schedulers ability to collect and merge the bios.
 *
 * But, it also turns into a long list of bios to process and that is sure
 * to eventually make the worker thread block.  The solution here is to
 * make some progress and then put this work struct back at the end of
 * the list if the block device is congested.  This way, multiple devices
 * can make progress from a single worker thread.
 */
249
static noinline void run_scheduled_bios(struct btrfs_device *device)
250 251 252
{
	struct bio *pending;
	struct backing_dev_info *bdi;
253
	struct btrfs_fs_info *fs_info;
254
	struct btrfs_pending_bios *pending_bios;
255 256 257
	struct bio *tail;
	struct bio *cur;
	int again = 0;
258
	unsigned long num_run;
259
	unsigned long batch_run = 0;
260
	unsigned long limit;
261
	unsigned long last_waited = 0;
262
	int force_reg = 0;
M
Miao Xie 已提交
263
	int sync_pending = 0;
264 265 266 267 268 269 270 271 272
	struct blk_plug plug;

	/*
	 * this function runs all the bios we've collected for
	 * a particular device.  We don't want to wander off to
	 * another device without first sending all of these down.
	 * So, setup a plug here and finish it off before we return
	 */
	blk_start_plug(&plug);
273

274
	bdi = blk_get_backing_dev_info(device->bdev);
275 276 277 278
	fs_info = device->dev_root->fs_info;
	limit = btrfs_async_submit_limit(fs_info);
	limit = limit * 2 / 3;

279 280 281
loop:
	spin_lock(&device->io_lock);

282
loop_lock:
283
	num_run = 0;
284

285 286 287 288 289
	/* take all the bios off the list at once and process them
	 * later on (without the lock held).  But, remember the
	 * tail and other pointers so the bios can be properly reinserted
	 * into the list if we hit congestion
	 */
290
	if (!force_reg && device->pending_sync_bios.head) {
291
		pending_bios = &device->pending_sync_bios;
292 293
		force_reg = 1;
	} else {
294
		pending_bios = &device->pending_bios;
295 296
		force_reg = 0;
	}
297 298 299

	pending = pending_bios->head;
	tail = pending_bios->tail;
300 301 302 303 304 305 306 307 308 309
	WARN_ON(pending && !tail);

	/*
	 * if pending was null this time around, no bios need processing
	 * at all and we can stop.  Otherwise it'll loop back up again
	 * and do an additional check so no bios are missed.
	 *
	 * device->running_pending is used to synchronize with the
	 * schedule_bio code.
	 */
310 311
	if (device->pending_sync_bios.head == NULL &&
	    device->pending_bios.head == NULL) {
312 313
		again = 0;
		device->running_pending = 0;
314 315 316
	} else {
		again = 1;
		device->running_pending = 1;
317
	}
318 319 320 321

	pending_bios->head = NULL;
	pending_bios->tail = NULL;

322 323
	spin_unlock(&device->io_lock);

C
Chris Mason 已提交
324
	while (pending) {
325 326

		rmb();
327 328 329 330 331 332 333 334
		/* we want to work on both lists, but do more bios on the
		 * sync list than the regular list
		 */
		if ((num_run > 32 &&
		    pending_bios != &device->pending_sync_bios &&
		    device->pending_sync_bios.head) ||
		   (num_run > 64 && pending_bios == &device->pending_sync_bios &&
		    device->pending_bios.head)) {
335 336 337 338 339
			spin_lock(&device->io_lock);
			requeue_list(pending_bios, pending, tail);
			goto loop_lock;
		}

340 341 342
		cur = pending;
		pending = pending->bi_next;
		cur->bi_next = NULL;
343

344
		if (atomic_dec_return(&fs_info->nr_async_bios) < limit &&
345 346
		    waitqueue_active(&fs_info->async_submit_wait))
			wake_up(&fs_info->async_submit_wait);
347 348

		BUG_ON(atomic_read(&cur->bi_cnt) == 0);
349

350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365
		/*
		 * if we're doing the sync list, record that our
		 * plug has some sync requests on it
		 *
		 * If we're doing the regular list and there are
		 * sync requests sitting around, unplug before
		 * we add more
		 */
		if (pending_bios == &device->pending_sync_bios) {
			sync_pending = 1;
		} else if (sync_pending) {
			blk_finish_plug(&plug);
			blk_start_plug(&plug);
			sync_pending = 0;
		}

366
		btrfsic_submit_bio(cur->bi_rw, cur);
367 368
		num_run++;
		batch_run++;
369 370

		cond_resched();
371 372 373 374 375 376

		/*
		 * we made progress, there is more work to do and the bdi
		 * is now congested.  Back off and let other work structs
		 * run instead
		 */
C
Chris Mason 已提交
377
		if (pending && bdi_write_congested(bdi) && batch_run > 8 &&
378
		    fs_info->fs_devices->open_devices > 1) {
379
			struct io_context *ioc;
380

381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402
			ioc = current->io_context;

			/*
			 * the main goal here is that we don't want to
			 * block if we're going to be able to submit
			 * more requests without blocking.
			 *
			 * This code does two great things, it pokes into
			 * the elevator code from a filesystem _and_
			 * it makes assumptions about how batching works.
			 */
			if (ioc && ioc->nr_batch_requests > 0 &&
			    time_before(jiffies, ioc->last_waited + HZ/50UL) &&
			    (last_waited == 0 ||
			     ioc->last_waited == last_waited)) {
				/*
				 * we want to go through our batch of
				 * requests and stop.  So, we copy out
				 * the ioc->last_waited time and test
				 * against it before looping
				 */
				last_waited = ioc->last_waited;
403
				cond_resched();
404 405
				continue;
			}
406
			spin_lock(&device->io_lock);
407
			requeue_list(pending_bios, pending, tail);
408
			device->running_pending = 1;
409 410

			spin_unlock(&device->io_lock);
411 412
			btrfs_queue_work(fs_info->submit_workers,
					 &device->work);
413 414
			goto done;
		}
C
Chris Mason 已提交
415 416 417 418 419 420
		/* unplug every 64 requests just for good measure */
		if (batch_run % 64 == 0) {
			blk_finish_plug(&plug);
			blk_start_plug(&plug);
			sync_pending = 0;
		}
421
	}
422

423 424 425 426 427 428 429 430 431
	cond_resched();
	if (again)
		goto loop;

	spin_lock(&device->io_lock);
	if (device->pending_bios.head || device->pending_sync_bios.head)
		goto loop_lock;
	spin_unlock(&device->io_lock);

432
done:
433
	blk_finish_plug(&plug);
434 435
}

436
static void pending_bios_fn(struct btrfs_work *work)
437 438 439 440 441 442 443
{
	struct btrfs_device *device;

	device = container_of(work, struct btrfs_device, work);
	run_scheduled_bios(device);
}

444 445 446 447 448 449 450 451
/*
 * Add new device to list of registered devices
 *
 * Returns:
 * 1   - first time device is seen
 * 0   - device already known
 * < 0 - error
 */
452
static noinline int device_list_add(const char *path,
453 454 455 456 457
			   struct btrfs_super_block *disk_super,
			   u64 devid, struct btrfs_fs_devices **fs_devices_ret)
{
	struct btrfs_device *device;
	struct btrfs_fs_devices *fs_devices;
458
	struct rcu_string *name;
459
	int ret = 0;
460 461 462 463
	u64 found_transid = btrfs_super_generation(disk_super);

	fs_devices = find_fsid(disk_super->fsid);
	if (!fs_devices) {
464 465 466 467
		fs_devices = alloc_fs_devices(disk_super->fsid);
		if (IS_ERR(fs_devices))
			return PTR_ERR(fs_devices);

468
		list_add(&fs_devices->list, &fs_uuids);
469

470 471
		device = NULL;
	} else {
472 473
		device = __find_device(&fs_devices->devices, devid,
				       disk_super->dev_item.uuid);
474
	}
475

476
	if (!device) {
Y
Yan Zheng 已提交
477 478 479
		if (fs_devices->opened)
			return -EBUSY;

480 481 482
		device = btrfs_alloc_device(NULL, &devid,
					    disk_super->dev_item.uuid);
		if (IS_ERR(device)) {
483
			/* we can safely leave the fs_devices entry around */
484
			return PTR_ERR(device);
485
		}
486 487 488

		name = rcu_string_strdup(path, GFP_NOFS);
		if (!name) {
489 490 491
			kfree(device);
			return -ENOMEM;
		}
492
		rcu_assign_pointer(device->name, name);
493

494
		mutex_lock(&fs_devices->device_list_mutex);
495
		list_add_rcu(&device->dev_list, &fs_devices->devices);
496
		fs_devices->num_devices++;
497 498
		mutex_unlock(&fs_devices->device_list_mutex);

499
		ret = 1;
Y
Yan Zheng 已提交
500
		device->fs_devices = fs_devices;
501
	} else if (!device->name || strcmp(device->name->str, path)) {
502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522
		/*
		 * When FS is already mounted.
		 * 1. If you are here and if the device->name is NULL that
		 *    means this device was missing at time of FS mount.
		 * 2. If you are here and if the device->name is different
		 *    from 'path' that means either
		 *      a. The same device disappeared and reappeared with
		 *         different name. or
		 *      b. The missing-disk-which-was-replaced, has
		 *         reappeared now.
		 *
		 * We must allow 1 and 2a above. But 2b would be a spurious
		 * and unintentional.
		 *
		 * Further in case of 1 and 2a above, the disk at 'path'
		 * would have missed some transaction when it was away and
		 * in case of 2a the stale bdev has to be updated as well.
		 * 2b must not be allowed at all time.
		 */

		/*
523 524 525 526
		 * For now, we do allow update to btrfs_fs_device through the
		 * btrfs dev scan cli after FS has been mounted.  We're still
		 * tracking a problem where systems fail mount by subvolume id
		 * when we reject replacement on a mounted FS.
527
		 */
528
		if (!fs_devices->opened && found_transid < device->generation) {
529 530 531 532 533 534 535
			/*
			 * That is if the FS is _not_ mounted and if you
			 * are here, that means there is more than one
			 * disk with same uuid and devid.We keep the one
			 * with larger generation number or the last-in if
			 * generation are equal.
			 */
536
			return -EEXIST;
537
		}
538

539
		name = rcu_string_strdup(path, GFP_NOFS);
540 541
		if (!name)
			return -ENOMEM;
542 543
		rcu_string_free(device->name);
		rcu_assign_pointer(device->name, name);
544 545 546 547
		if (device->missing) {
			fs_devices->missing_devices--;
			device->missing = 0;
		}
548 549
	}

550 551 552 553 554 555 556 557 558
	/*
	 * Unmount does not free the btrfs_device struct but would zero
	 * generation along with most of the other members. So just update
	 * it back. We need it to pick the disk with largest generation
	 * (as above).
	 */
	if (!fs_devices->opened)
		device->generation = found_transid;

559
	*fs_devices_ret = fs_devices;
560 561

	return ret;
562 563
}

Y
Yan Zheng 已提交
564 565 566 567 568 569
static struct btrfs_fs_devices *clone_fs_devices(struct btrfs_fs_devices *orig)
{
	struct btrfs_fs_devices *fs_devices;
	struct btrfs_device *device;
	struct btrfs_device *orig_dev;

570 571 572
	fs_devices = alloc_fs_devices(orig->fsid);
	if (IS_ERR(fs_devices))
		return fs_devices;
Y
Yan Zheng 已提交
573

574
	mutex_lock(&orig->device_list_mutex);
J
Josef Bacik 已提交
575
	fs_devices->total_devices = orig->total_devices;
Y
Yan Zheng 已提交
576

577
	/* We have held the volume lock, it is safe to get the devices. */
Y
Yan Zheng 已提交
578
	list_for_each_entry(orig_dev, &orig->devices, dev_list) {
579 580
		struct rcu_string *name;

581 582 583
		device = btrfs_alloc_device(NULL, &orig_dev->devid,
					    orig_dev->uuid);
		if (IS_ERR(device))
Y
Yan Zheng 已提交
584 585
			goto error;

586 587 588 589
		/*
		 * This is ok to do without rcu read locked because we hold the
		 * uuid mutex so nothing we touch in here is going to disappear.
		 */
590 591 592 593 594 595 596
		if (orig_dev->name) {
			name = rcu_string_strdup(orig_dev->name->str, GFP_NOFS);
			if (!name) {
				kfree(device);
				goto error;
			}
			rcu_assign_pointer(device->name, name);
J
Julia Lawall 已提交
597
		}
Y
Yan Zheng 已提交
598 599 600 601 602

		list_add(&device->dev_list, &fs_devices->devices);
		device->fs_devices = fs_devices;
		fs_devices->num_devices++;
	}
603
	mutex_unlock(&orig->device_list_mutex);
Y
Yan Zheng 已提交
604 605
	return fs_devices;
error:
606
	mutex_unlock(&orig->device_list_mutex);
Y
Yan Zheng 已提交
607 608 609 610
	free_fs_devices(fs_devices);
	return ERR_PTR(-ENOMEM);
}

611 612
void btrfs_close_extra_devices(struct btrfs_fs_info *fs_info,
			       struct btrfs_fs_devices *fs_devices, int step)
613
{
Q
Qinghuang Feng 已提交
614
	struct btrfs_device *device, *next;
615
	struct btrfs_device *latest_dev = NULL;
616

617 618
	mutex_lock(&uuid_mutex);
again:
619
	/* This is the initialized path, it is safe to release the devices. */
Q
Qinghuang Feng 已提交
620
	list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
621
		if (device->in_fs_metadata) {
622
			if (!device->is_tgtdev_for_dev_replace &&
623 624 625
			    (!latest_dev ||
			     device->generation > latest_dev->generation)) {
				latest_dev = device;
626
			}
Y
Yan Zheng 已提交
627
			continue;
628
		}
Y
Yan Zheng 已提交
629

630 631 632 633 634 635 636 637 638 639 640 641 642 643 644
		if (device->devid == BTRFS_DEV_REPLACE_DEVID) {
			/*
			 * In the first step, keep the device which has
			 * the correct fsid and the devid that is used
			 * for the dev_replace procedure.
			 * In the second step, the dev_replace state is
			 * read from the device tree and it is known
			 * whether the procedure is really active or
			 * not, which means whether this device is
			 * used or whether it should be removed.
			 */
			if (step == 0 || device->is_tgtdev_for_dev_replace) {
				continue;
			}
		}
Y
Yan Zheng 已提交
645
		if (device->bdev) {
646
			blkdev_put(device->bdev, device->mode);
Y
Yan Zheng 已提交
647 648 649 650 651 652
			device->bdev = NULL;
			fs_devices->open_devices--;
		}
		if (device->writeable) {
			list_del_init(&device->dev_alloc_list);
			device->writeable = 0;
653 654
			if (!device->is_tgtdev_for_dev_replace)
				fs_devices->rw_devices--;
Y
Yan Zheng 已提交
655
		}
Y
Yan Zheng 已提交
656 657
		list_del_init(&device->dev_list);
		fs_devices->num_devices--;
658
		rcu_string_free(device->name);
Y
Yan Zheng 已提交
659
		kfree(device);
660
	}
Y
Yan Zheng 已提交
661 662 663 664 665 666

	if (fs_devices->seed) {
		fs_devices = fs_devices->seed;
		goto again;
	}

667
	fs_devices->latest_bdev = latest_dev->bdev;
668

669 670
	mutex_unlock(&uuid_mutex);
}
671

672 673 674 675 676 677 678 679 680
static void __free_device(struct work_struct *work)
{
	struct btrfs_device *device;

	device = container_of(work, struct btrfs_device, rcu_work);

	if (device->bdev)
		blkdev_put(device->bdev, device->mode);

681
	rcu_string_free(device->name);
682 683 684 685 686 687 688 689 690 691 692 693 694
	kfree(device);
}

static void free_device(struct rcu_head *head)
{
	struct btrfs_device *device;

	device = container_of(head, struct btrfs_device, rcu);

	INIT_WORK(&device->rcu_work, __free_device);
	schedule_work(&device->rcu_work);
}

Y
Yan Zheng 已提交
695
static int __btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
696 697
{
	struct btrfs_device *device;
Y
Yan Zheng 已提交
698

Y
Yan Zheng 已提交
699 700
	if (--fs_devices->opened > 0)
		return 0;
701

702
	mutex_lock(&fs_devices->device_list_mutex);
Q
Qinghuang Feng 已提交
703
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
704
		struct btrfs_device *new_device;
705
		struct rcu_string *name;
706 707

		if (device->bdev)
708
			fs_devices->open_devices--;
709

710 711
		if (device->writeable &&
		    device->devid != BTRFS_DEV_REPLACE_DEVID) {
Y
Yan Zheng 已提交
712 713 714 715
			list_del_init(&device->dev_alloc_list);
			fs_devices->rw_devices--;
		}

716 717
		if (device->missing)
			fs_devices->missing_devices--;
718

719 720 721
		new_device = btrfs_alloc_device(NULL, &device->devid,
						device->uuid);
		BUG_ON(IS_ERR(new_device)); /* -ENOMEM */
722 723

		/* Safe because we are under uuid_mutex */
724 725
		if (device->name) {
			name = rcu_string_strdup(device->name->str, GFP_NOFS);
726
			BUG_ON(!name); /* -ENOMEM */
727 728
			rcu_assign_pointer(new_device->name, name);
		}
729

730
		list_replace_rcu(&device->dev_list, &new_device->dev_list);
731
		new_device->fs_devices = device->fs_devices;
732 733

		call_rcu(&device->rcu, free_device);
734
	}
735 736
	mutex_unlock(&fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
737 738
	WARN_ON(fs_devices->open_devices);
	WARN_ON(fs_devices->rw_devices);
Y
Yan Zheng 已提交
739 740 741
	fs_devices->opened = 0;
	fs_devices->seeding = 0;

742 743 744
	return 0;
}

Y
Yan Zheng 已提交
745 746
int btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
{
Y
Yan Zheng 已提交
747
	struct btrfs_fs_devices *seed_devices = NULL;
Y
Yan Zheng 已提交
748 749 750 751
	int ret;

	mutex_lock(&uuid_mutex);
	ret = __btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
752 753 754 755
	if (!fs_devices->opened) {
		seed_devices = fs_devices->seed;
		fs_devices->seed = NULL;
	}
Y
Yan Zheng 已提交
756
	mutex_unlock(&uuid_mutex);
Y
Yan Zheng 已提交
757 758 759 760 761 762 763

	while (seed_devices) {
		fs_devices = seed_devices;
		seed_devices = fs_devices->seed;
		__btrfs_close_devices(fs_devices);
		free_fs_devices(fs_devices);
	}
764 765 766 767 768 769
	/*
	 * Wait for rcu kworkers under __btrfs_close_devices
	 * to finish all blkdev_puts so device is really
	 * free when umount is done.
	 */
	rcu_barrier();
Y
Yan Zheng 已提交
770 771 772
	return ret;
}

Y
Yan Zheng 已提交
773 774
static int __btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
				fmode_t flags, void *holder)
775
{
776
	struct request_queue *q;
777 778 779
	struct block_device *bdev;
	struct list_head *head = &fs_devices->devices;
	struct btrfs_device *device;
780
	struct btrfs_device *latest_dev = NULL;
781 782 783
	struct buffer_head *bh;
	struct btrfs_super_block *disk_super;
	u64 devid;
Y
Yan Zheng 已提交
784
	int seeding = 1;
785
	int ret = 0;
786

787 788
	flags |= FMODE_EXCL;

Q
Qinghuang Feng 已提交
789
	list_for_each_entry(device, head, dev_list) {
790 791
		if (device->bdev)
			continue;
792 793 794
		if (!device->name)
			continue;

795 796 797
		/* Just open everything we can; ignore failures here */
		if (btrfs_get_bdev_and_sb(device->name->str, flags, holder, 1,
					    &bdev, &bh))
798
			continue;
799 800

		disk_super = (struct btrfs_super_block *)bh->b_data;
801
		devid = btrfs_stack_device_id(&disk_super->dev_item);
802 803 804
		if (devid != device->devid)
			goto error_brelse;

Y
Yan Zheng 已提交
805 806 807 808 809
		if (memcmp(device->uuid, disk_super->dev_item.uuid,
			   BTRFS_UUID_SIZE))
			goto error_brelse;

		device->generation = btrfs_super_generation(disk_super);
810 811 812
		if (!latest_dev ||
		    device->generation > latest_dev->generation)
			latest_dev = device;
813

Y
Yan Zheng 已提交
814 815 816 817 818 819 820
		if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_SEEDING) {
			device->writeable = 0;
		} else {
			device->writeable = !bdev_read_only(bdev);
			seeding = 0;
		}

821
		q = bdev_get_queue(bdev);
822
		if (blk_queue_discard(q))
823 824
			device->can_discard = 1;

825
		device->bdev = bdev;
826
		device->in_fs_metadata = 0;
827 828
		device->mode = flags;

C
Chris Mason 已提交
829 830 831
		if (!blk_queue_nonrot(bdev_get_queue(bdev)))
			fs_devices->rotating = 1;

832
		fs_devices->open_devices++;
833 834
		if (device->writeable &&
		    device->devid != BTRFS_DEV_REPLACE_DEVID) {
Y
Yan Zheng 已提交
835 836 837 838
			fs_devices->rw_devices++;
			list_add(&device->dev_alloc_list,
				 &fs_devices->alloc_list);
		}
839
		brelse(bh);
840
		continue;
841

842 843
error_brelse:
		brelse(bh);
844
		blkdev_put(bdev, flags);
845
		continue;
846
	}
847
	if (fs_devices->open_devices == 0) {
848
		ret = -EINVAL;
849 850
		goto out;
	}
Y
Yan Zheng 已提交
851 852
	fs_devices->seeding = seeding;
	fs_devices->opened = 1;
853
	fs_devices->latest_bdev = latest_dev->bdev;
Y
Yan Zheng 已提交
854
	fs_devices->total_rw_bytes = 0;
855
out:
Y
Yan Zheng 已提交
856 857 858 859
	return ret;
}

int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
860
		       fmode_t flags, void *holder)
Y
Yan Zheng 已提交
861 862 863 864 865
{
	int ret;

	mutex_lock(&uuid_mutex);
	if (fs_devices->opened) {
Y
Yan Zheng 已提交
866 867
		fs_devices->opened++;
		ret = 0;
Y
Yan Zheng 已提交
868
	} else {
869
		ret = __btrfs_open_devices(fs_devices, flags, holder);
Y
Yan Zheng 已提交
870
	}
871 872 873 874
	mutex_unlock(&uuid_mutex);
	return ret;
}

875 876 877 878 879
/*
 * Look for a btrfs signature on a device. This may be called out of the mount path
 * and we are not allowed to call set_blocksize during the scan. The superblock
 * is read via pagecache
 */
880
int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
881 882 883 884
			  struct btrfs_fs_devices **fs_devices_ret)
{
	struct btrfs_super_block *disk_super;
	struct block_device *bdev;
885 886 887
	struct page *page;
	void *p;
	int ret = -EINVAL;
888
	u64 devid;
889
	u64 transid;
J
Josef Bacik 已提交
890
	u64 total_devices;
891 892
	u64 bytenr;
	pgoff_t index;
893

894 895 896 897 898 899 900
	/*
	 * we would like to check all the supers, but that would make
	 * a btrfs mount succeed after a mkfs from a different FS.
	 * So, we need to add a special mount option to scan for
	 * later supers, using BTRFS_SUPER_MIRROR_MAX instead
	 */
	bytenr = btrfs_sb_offset(0);
901
	flags |= FMODE_EXCL;
902
	mutex_lock(&uuid_mutex);
903 904 905 906 907

	bdev = blkdev_get_by_path(path, flags, holder);

	if (IS_ERR(bdev)) {
		ret = PTR_ERR(bdev);
908
		goto error;
909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936
	}

	/* make sure our super fits in the device */
	if (bytenr + PAGE_CACHE_SIZE >= i_size_read(bdev->bd_inode))
		goto error_bdev_put;

	/* make sure our super fits in the page */
	if (sizeof(*disk_super) > PAGE_CACHE_SIZE)
		goto error_bdev_put;

	/* make sure our super doesn't straddle pages on disk */
	index = bytenr >> PAGE_CACHE_SHIFT;
	if ((bytenr + sizeof(*disk_super) - 1) >> PAGE_CACHE_SHIFT != index)
		goto error_bdev_put;

	/* pull in the page with our super */
	page = read_cache_page_gfp(bdev->bd_inode->i_mapping,
				   index, GFP_NOFS);

	if (IS_ERR_OR_NULL(page))
		goto error_bdev_put;

	p = kmap(page);

	/* align our pointer to the offset of the super block */
	disk_super = p + (bytenr & ~PAGE_CACHE_MASK);

	if (btrfs_super_bytenr(disk_super) != bytenr ||
937
	    btrfs_super_magic(disk_super) != BTRFS_MAGIC)
938 939
		goto error_unmap;

940
	devid = btrfs_stack_device_id(&disk_super->dev_item);
941
	transid = btrfs_super_generation(disk_super);
J
Josef Bacik 已提交
942
	total_devices = btrfs_super_num_devices(disk_super);
943

944
	ret = device_list_add(path, disk_super, devid, fs_devices_ret);
945 946 947 948 949 950 951 952 953 954 955 956
	if (ret > 0) {
		if (disk_super->label[0]) {
			if (disk_super->label[BTRFS_LABEL_SIZE - 1])
				disk_super->label[BTRFS_LABEL_SIZE - 1] = '\0';
			printk(KERN_INFO "BTRFS: device label %s ", disk_super->label);
		} else {
			printk(KERN_INFO "BTRFS: device fsid %pU ", disk_super->fsid);
		}

		printk(KERN_CONT "devid %llu transid %llu %s\n", devid, transid, path);
		ret = 0;
	}
J
Josef Bacik 已提交
957 958
	if (!ret && fs_devices_ret)
		(*fs_devices_ret)->total_devices = total_devices;
959 960 961 962 963 964

error_unmap:
	kunmap(page);
	page_cache_release(page);

error_bdev_put:
965
	blkdev_put(bdev, flags);
966
error:
967
	mutex_unlock(&uuid_mutex);
968 969
	return ret;
}
970

971 972 973 974 975 976 977 978 979 980 981 982 983 984 985
/* helper to account the used device space in the range */
int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start,
				   u64 end, u64 *length)
{
	struct btrfs_key key;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_dev_extent *dev_extent;
	struct btrfs_path *path;
	u64 extent_end;
	int ret;
	int slot;
	struct extent_buffer *l;

	*length = 0;

986
	if (start >= device->total_bytes || device->is_tgtdev_for_dev_replace)
987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026
		return 0;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
	path->reada = 2;

	key.objectid = device->devid;
	key.offset = start;
	key.type = BTRFS_DEV_EXTENT_KEY;

	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (ret < 0)
		goto out;
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid, key.type);
		if (ret < 0)
			goto out;
	}

	while (1) {
		l = path->nodes[0];
		slot = path->slots[0];
		if (slot >= btrfs_header_nritems(l)) {
			ret = btrfs_next_leaf(root, path);
			if (ret == 0)
				continue;
			if (ret < 0)
				goto out;

			break;
		}
		btrfs_item_key_to_cpu(l, &key, slot);

		if (key.objectid < device->devid)
			goto next;

		if (key.objectid > device->devid)
			break;

1027
		if (key.type != BTRFS_DEV_EXTENT_KEY)
1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
			goto next;

		dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
		extent_end = key.offset + btrfs_dev_extent_length(l,
								  dev_extent);
		if (key.offset <= start && extent_end > end) {
			*length = end - start + 1;
			break;
		} else if (key.offset <= start && extent_end > start)
			*length += extent_end - start;
		else if (key.offset > start && extent_end <= end)
			*length += extent_end - key.offset;
		else if (key.offset > start && key.offset <= end) {
			*length += end - key.offset + 1;
			break;
		} else if (key.offset > end)
			break;

next:
		path->slots[0]++;
	}
	ret = 0;
out:
	btrfs_free_path(path);
	return ret;
}

1055 1056 1057 1058 1059
static int contains_pending_extent(struct btrfs_trans_handle *trans,
				   struct btrfs_device *device,
				   u64 *start, u64 len)
{
	struct extent_map *em;
1060
	struct list_head *search_list = &trans->transaction->pending_chunks;
1061 1062
	int ret = 0;

1063 1064
again:
	list_for_each_entry(em, search_list, list) {
1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080
		struct map_lookup *map;
		int i;

		map = (struct map_lookup *)em->bdev;
		for (i = 0; i < map->num_stripes; i++) {
			if (map->stripes[i].dev != device)
				continue;
			if (map->stripes[i].physical >= *start + len ||
			    map->stripes[i].physical + em->orig_block_len <=
			    *start)
				continue;
			*start = map->stripes[i].physical +
				em->orig_block_len;
			ret = 1;
		}
	}
1081 1082 1083 1084
	if (search_list == &trans->transaction->pending_chunks) {
		search_list = &trans->root->fs_info->pinned_chunks;
		goto again;
	}
1085 1086 1087 1088 1089

	return ret;
}


1090
/*
1091 1092 1093 1094 1095 1096 1097
 * find_free_dev_extent - find free space in the specified device
 * @device:	the device which we search the free space in
 * @num_bytes:	the size of the free space that we need
 * @start:	store the start of the free space.
 * @len:	the size of the free space. that we find, or the size of the max
 * 		free space if we don't find suitable free space
 *
1098 1099 1100
 * this uses a pretty simple search, the expectation is that it is
 * called very infrequently and that a given device has a small number
 * of extents
1101 1102 1103 1104 1105 1106 1107 1108
 *
 * @start is used to store the start of the free space if we find. But if we
 * don't find suitable free space, it will be used to store the start position
 * of the max free space.
 *
 * @len is used to store the size of the free space that we find.
 * But if we don't find suitable free space, it is used to store the size of
 * the max free space.
1109
 */
1110 1111
int find_free_dev_extent(struct btrfs_trans_handle *trans,
			 struct btrfs_device *device, u64 num_bytes,
1112
			 u64 *start, u64 *len)
1113 1114 1115
{
	struct btrfs_key key;
	struct btrfs_root *root = device->dev_root;
1116
	struct btrfs_dev_extent *dev_extent;
Y
Yan Zheng 已提交
1117
	struct btrfs_path *path;
1118 1119 1120 1121 1122
	u64 hole_size;
	u64 max_hole_start;
	u64 max_hole_size;
	u64 extent_end;
	u64 search_start;
1123 1124
	u64 search_end = device->total_bytes;
	int ret;
1125
	int slot;
1126 1127 1128 1129
	struct extent_buffer *l;

	/* FIXME use last free of some kind */

1130 1131 1132
	/* we don't want to overwrite the superblock on the drive,
	 * so we make sure to start at an offset of at least 1MB
	 */
A
Arne Jansen 已提交
1133
	search_start = max(root->fs_info->alloc_start, 1024ull * 1024);
1134

1135 1136 1137 1138
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
again:
1139 1140
	max_hole_start = search_start;
	max_hole_size = 0;
1141
	hole_size = 0;
1142

1143
	if (search_start >= search_end || device->is_tgtdev_for_dev_replace) {
1144
		ret = -ENOSPC;
1145
		goto out;
1146 1147 1148
	}

	path->reada = 2;
1149 1150
	path->search_commit_root = 1;
	path->skip_locking = 1;
1151

1152 1153 1154
	key.objectid = device->devid;
	key.offset = search_start;
	key.type = BTRFS_DEV_EXTENT_KEY;
1155

1156
	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1157
	if (ret < 0)
1158
		goto out;
1159 1160 1161
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid, key.type);
		if (ret < 0)
1162
			goto out;
1163
	}
1164

1165 1166 1167 1168 1169 1170 1171 1172
	while (1) {
		l = path->nodes[0];
		slot = path->slots[0];
		if (slot >= btrfs_header_nritems(l)) {
			ret = btrfs_next_leaf(root, path);
			if (ret == 0)
				continue;
			if (ret < 0)
1173 1174 1175
				goto out;

			break;
1176 1177 1178 1179 1180 1181 1182
		}
		btrfs_item_key_to_cpu(l, &key, slot);

		if (key.objectid < device->devid)
			goto next;

		if (key.objectid > device->devid)
1183
			break;
1184

1185
		if (key.type != BTRFS_DEV_EXTENT_KEY)
1186
			goto next;
1187

1188 1189
		if (key.offset > search_start) {
			hole_size = key.offset - search_start;
1190

1191 1192 1193 1194 1195 1196 1197 1198 1199
			/*
			 * Have to check before we set max_hole_start, otherwise
			 * we could end up sending back this offset anyway.
			 */
			if (contains_pending_extent(trans, device,
						    &search_start,
						    hole_size))
				hole_size = 0;

1200 1201 1202 1203
			if (hole_size > max_hole_size) {
				max_hole_start = search_start;
				max_hole_size = hole_size;
			}
1204

1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216
			/*
			 * If this free space is greater than which we need,
			 * it must be the max free space that we have found
			 * until now, so max_hole_start must point to the start
			 * of this free space and the length of this free space
			 * is stored in max_hole_size. Thus, we return
			 * max_hole_start and max_hole_size and go back to the
			 * caller.
			 */
			if (hole_size >= num_bytes) {
				ret = 0;
				goto out;
1217 1218 1219 1220
			}
		}

		dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
1221 1222 1223 1224
		extent_end = key.offset + btrfs_dev_extent_length(l,
								  dev_extent);
		if (extent_end > search_start)
			search_start = extent_end;
1225 1226 1227 1228 1229
next:
		path->slots[0]++;
		cond_resched();
	}

1230 1231 1232 1233 1234 1235 1236 1237
	/*
	 * At this point, search_start should be the end of
	 * allocated dev extents, and when shrinking the device,
	 * search_end may be smaller than search_start.
	 */
	if (search_end > search_start)
		hole_size = search_end - search_start;

1238 1239 1240
	if (hole_size > max_hole_size) {
		max_hole_start = search_start;
		max_hole_size = hole_size;
1241 1242
	}

1243 1244 1245 1246 1247
	if (contains_pending_extent(trans, device, &search_start, hole_size)) {
		btrfs_release_path(path);
		goto again;
	}

1248 1249 1250 1251 1252 1253 1254
	/* See above. */
	if (hole_size < num_bytes)
		ret = -ENOSPC;
	else
		ret = 0;

out:
Y
Yan Zheng 已提交
1255
	btrfs_free_path(path);
1256
	*start = max_hole_start;
1257
	if (len)
1258
		*len = max_hole_size;
1259 1260 1261
	return ret;
}

1262
static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans,
1263
			  struct btrfs_device *device,
M
Miao Xie 已提交
1264
			  u64 start, u64 *dev_extent_len)
1265 1266 1267 1268 1269
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_key key;
1270 1271 1272
	struct btrfs_key found_key;
	struct extent_buffer *leaf = NULL;
	struct btrfs_dev_extent *extent = NULL;
1273 1274 1275 1276 1277 1278 1279 1280

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = device->devid;
	key.offset = start;
	key.type = BTRFS_DEV_EXTENT_KEY;
M
Miao Xie 已提交
1281
again:
1282
	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
1283 1284 1285
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid,
					  BTRFS_DEV_EXTENT_KEY);
1286 1287
		if (ret)
			goto out;
1288 1289 1290 1291 1292 1293
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
		extent = btrfs_item_ptr(leaf, path->slots[0],
					struct btrfs_dev_extent);
		BUG_ON(found_key.offset > start || found_key.offset +
		       btrfs_dev_extent_length(leaf, extent) < start);
M
Miao Xie 已提交
1294 1295 1296
		key = found_key;
		btrfs_release_path(path);
		goto again;
1297 1298 1299 1300
	} else if (ret == 0) {
		leaf = path->nodes[0];
		extent = btrfs_item_ptr(leaf, path->slots[0],
					struct btrfs_dev_extent);
1301 1302 1303
	} else {
		btrfs_error(root->fs_info, ret, "Slot search failed");
		goto out;
1304
	}
1305

M
Miao Xie 已提交
1306 1307
	*dev_extent_len = btrfs_dev_extent_length(leaf, extent);

1308
	ret = btrfs_del_item(trans, root, path);
1309 1310 1311
	if (ret) {
		btrfs_error(root->fs_info, ret,
			    "Failed to remove dev extent item");
Z
Zhao Lei 已提交
1312 1313
	} else {
		trans->transaction->have_free_bgs = 1;
1314
	}
1315
out:
1316 1317 1318 1319
	btrfs_free_path(path);
	return ret;
}

1320 1321 1322 1323
static int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
				  struct btrfs_device *device,
				  u64 chunk_tree, u64 chunk_objectid,
				  u64 chunk_offset, u64 start, u64 num_bytes)
1324 1325 1326 1327 1328 1329 1330 1331
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_dev_extent *extent;
	struct extent_buffer *leaf;
	struct btrfs_key key;

1332
	WARN_ON(!device->in_fs_metadata);
1333
	WARN_ON(device->is_tgtdev_for_dev_replace);
1334 1335 1336 1337 1338
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = device->devid;
Y
Yan Zheng 已提交
1339
	key.offset = start;
1340 1341 1342
	key.type = BTRFS_DEV_EXTENT_KEY;
	ret = btrfs_insert_empty_item(trans, root, path, &key,
				      sizeof(*extent));
1343 1344
	if (ret)
		goto out;
1345 1346 1347 1348

	leaf = path->nodes[0];
	extent = btrfs_item_ptr(leaf, path->slots[0],
				struct btrfs_dev_extent);
1349 1350 1351 1352 1353
	btrfs_set_dev_extent_chunk_tree(leaf, extent, chunk_tree);
	btrfs_set_dev_extent_chunk_objectid(leaf, extent, chunk_objectid);
	btrfs_set_dev_extent_chunk_offset(leaf, extent, chunk_offset);

	write_extent_buffer(leaf, root->fs_info->chunk_tree_uuid,
1354
		    btrfs_dev_extent_chunk_tree_uuid(extent), BTRFS_UUID_SIZE);
1355

1356 1357
	btrfs_set_dev_extent_length(leaf, extent, num_bytes);
	btrfs_mark_buffer_dirty(leaf);
1358
out:
1359 1360 1361 1362
	btrfs_free_path(path);
	return ret;
}

1363
static u64 find_next_chunk(struct btrfs_fs_info *fs_info)
1364
{
1365 1366 1367 1368
	struct extent_map_tree *em_tree;
	struct extent_map *em;
	struct rb_node *n;
	u64 ret = 0;
1369

1370 1371 1372 1373 1374 1375
	em_tree = &fs_info->mapping_tree.map_tree;
	read_lock(&em_tree->lock);
	n = rb_last(&em_tree->map);
	if (n) {
		em = rb_entry(n, struct extent_map, rb_node);
		ret = em->start + em->len;
1376
	}
1377 1378
	read_unlock(&em_tree->lock);

1379 1380 1381
	return ret;
}

1382 1383
static noinline int find_next_devid(struct btrfs_fs_info *fs_info,
				    u64 *devid_ret)
1384 1385 1386 1387
{
	int ret;
	struct btrfs_key key;
	struct btrfs_key found_key;
Y
Yan Zheng 已提交
1388 1389 1390 1391 1392
	struct btrfs_path *path;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
1393 1394 1395 1396 1397

	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.type = BTRFS_DEV_ITEM_KEY;
	key.offset = (u64)-1;

1398
	ret = btrfs_search_slot(NULL, fs_info->chunk_root, &key, path, 0, 0);
1399 1400 1401
	if (ret < 0)
		goto error;

1402
	BUG_ON(ret == 0); /* Corruption */
1403

1404 1405
	ret = btrfs_previous_item(fs_info->chunk_root, path,
				  BTRFS_DEV_ITEMS_OBJECTID,
1406 1407
				  BTRFS_DEV_ITEM_KEY);
	if (ret) {
1408
		*devid_ret = 1;
1409 1410 1411
	} else {
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
1412
		*devid_ret = found_key.offset + 1;
1413 1414 1415
	}
	ret = 0;
error:
Y
Yan Zheng 已提交
1416
	btrfs_free_path(path);
1417 1418 1419 1420 1421 1422 1423
	return ret;
}

/*
 * the device information is stored in the chunk root
 * the btrfs_device struct should be fully filled in
 */
1424 1425 1426
static int btrfs_add_device(struct btrfs_trans_handle *trans,
			    struct btrfs_root *root,
			    struct btrfs_device *device)
1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_dev_item *dev_item;
	struct extent_buffer *leaf;
	struct btrfs_key key;
	unsigned long ptr;

	root = root->fs_info->chunk_root;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.type = BTRFS_DEV_ITEM_KEY;
Y
Yan Zheng 已提交
1443
	key.offset = device->devid;
1444 1445

	ret = btrfs_insert_empty_item(trans, root, path, &key,
1446
				      sizeof(*dev_item));
1447 1448 1449 1450 1451 1452 1453
	if (ret)
		goto out;

	leaf = path->nodes[0];
	dev_item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dev_item);

	btrfs_set_device_id(leaf, dev_item, device->devid);
Y
Yan Zheng 已提交
1454
	btrfs_set_device_generation(leaf, dev_item, 0);
1455 1456 1457 1458
	btrfs_set_device_type(leaf, dev_item, device->type);
	btrfs_set_device_io_align(leaf, dev_item, device->io_align);
	btrfs_set_device_io_width(leaf, dev_item, device->io_width);
	btrfs_set_device_sector_size(leaf, dev_item, device->sector_size);
1459 1460 1461 1462
	btrfs_set_device_total_bytes(leaf, dev_item,
				     btrfs_device_get_disk_total_bytes(device));
	btrfs_set_device_bytes_used(leaf, dev_item,
				    btrfs_device_get_bytes_used(device));
1463 1464 1465
	btrfs_set_device_group(leaf, dev_item, 0);
	btrfs_set_device_seek_speed(leaf, dev_item, 0);
	btrfs_set_device_bandwidth(leaf, dev_item, 0);
1466
	btrfs_set_device_start_offset(leaf, dev_item, 0);
1467

1468
	ptr = btrfs_device_uuid(dev_item);
1469
	write_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
1470
	ptr = btrfs_device_fsid(dev_item);
Y
Yan Zheng 已提交
1471
	write_extent_buffer(leaf, root->fs_info->fsid, ptr, BTRFS_UUID_SIZE);
1472 1473
	btrfs_mark_buffer_dirty(leaf);

Y
Yan Zheng 已提交
1474
	ret = 0;
1475 1476 1477 1478
out:
	btrfs_free_path(path);
	return ret;
}
1479

1480 1481 1482 1483 1484 1485 1486 1487 1488
/*
 * Function to update ctime/mtime for a given device path.
 * Mainly used for ctime/mtime based probe like libblkid.
 */
static void update_dev_time(char *path_name)
{
	struct file *filp;

	filp = filp_open(path_name, O_RDWR, 0);
1489
	if (IS_ERR(filp))
1490 1491 1492 1493 1494 1495
		return;
	file_update_time(filp);
	filp_close(filp, NULL);
	return;
}

1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509
static int btrfs_rm_dev_item(struct btrfs_root *root,
			     struct btrfs_device *device)
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_key key;
	struct btrfs_trans_handle *trans;

	root = root->fs_info->chunk_root;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

1510
	trans = btrfs_start_transaction(root, 0);
1511 1512 1513 1514
	if (IS_ERR(trans)) {
		btrfs_free_path(path);
		return PTR_ERR(trans);
	}
1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.type = BTRFS_DEV_ITEM_KEY;
	key.offset = device->devid;

	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
	if (ret < 0)
		goto out;

	if (ret > 0) {
		ret = -ENOENT;
		goto out;
	}

	ret = btrfs_del_item(trans, root, path);
	if (ret)
		goto out;
out:
	btrfs_free_path(path);
	btrfs_commit_transaction(trans, root);
	return ret;
}

int btrfs_rm_device(struct btrfs_root *root, char *device_path)
{
	struct btrfs_device *device;
Y
Yan Zheng 已提交
1540
	struct btrfs_device *next_device;
1541
	struct block_device *bdev;
1542
	struct buffer_head *bh = NULL;
1543
	struct btrfs_super_block *disk_super;
1544
	struct btrfs_fs_devices *cur_devices;
1545 1546
	u64 all_avail;
	u64 devid;
Y
Yan Zheng 已提交
1547 1548
	u64 num_devices;
	u8 *dev_uuid;
1549
	unsigned seq;
1550
	int ret = 0;
1551
	bool clear_super = false;
1552 1553 1554

	mutex_lock(&uuid_mutex);

1555 1556 1557 1558 1559 1560 1561
	do {
		seq = read_seqbegin(&root->fs_info->profiles_lock);

		all_avail = root->fs_info->avail_data_alloc_bits |
			    root->fs_info->avail_system_alloc_bits |
			    root->fs_info->avail_metadata_alloc_bits;
	} while (read_seqretry(&root->fs_info->profiles_lock, seq));
1562

1563 1564 1565 1566 1567 1568 1569 1570 1571
	num_devices = root->fs_info->fs_devices->num_devices;
	btrfs_dev_replace_lock(&root->fs_info->dev_replace);
	if (btrfs_dev_replace_is_ongoing(&root->fs_info->dev_replace)) {
		WARN_ON(num_devices < 1);
		num_devices--;
	}
	btrfs_dev_replace_unlock(&root->fs_info->dev_replace);

	if ((all_avail & BTRFS_BLOCK_GROUP_RAID10) && num_devices <= 4) {
1572
		ret = BTRFS_ERROR_DEV_RAID10_MIN_NOT_MET;
1573 1574 1575
		goto out;
	}

1576
	if ((all_avail & BTRFS_BLOCK_GROUP_RAID1) && num_devices <= 2) {
1577
		ret = BTRFS_ERROR_DEV_RAID1_MIN_NOT_MET;
1578 1579 1580
		goto out;
	}

D
David Woodhouse 已提交
1581 1582
	if ((all_avail & BTRFS_BLOCK_GROUP_RAID5) &&
	    root->fs_info->fs_devices->rw_devices <= 2) {
1583
		ret = BTRFS_ERROR_DEV_RAID5_MIN_NOT_MET;
D
David Woodhouse 已提交
1584 1585 1586 1587
		goto out;
	}
	if ((all_avail & BTRFS_BLOCK_GROUP_RAID6) &&
	    root->fs_info->fs_devices->rw_devices <= 3) {
1588
		ret = BTRFS_ERROR_DEV_RAID6_MIN_NOT_MET;
D
David Woodhouse 已提交
1589 1590 1591
		goto out;
	}

1592 1593 1594
	if (strcmp(device_path, "missing") == 0) {
		struct list_head *devices;
		struct btrfs_device *tmp;
1595

1596 1597
		device = NULL;
		devices = &root->fs_info->fs_devices->devices;
1598 1599 1600 1601
		/*
		 * It is safe to read the devices since the volume_mutex
		 * is held.
		 */
Q
Qinghuang Feng 已提交
1602
		list_for_each_entry(tmp, devices, dev_list) {
1603 1604 1605
			if (tmp->in_fs_metadata &&
			    !tmp->is_tgtdev_for_dev_replace &&
			    !tmp->bdev) {
1606 1607 1608 1609 1610 1611 1612 1613
				device = tmp;
				break;
			}
		}
		bdev = NULL;
		bh = NULL;
		disk_super = NULL;
		if (!device) {
1614
			ret = BTRFS_ERROR_DEV_MISSING_NOT_FOUND;
1615 1616 1617
			goto out;
		}
	} else {
1618
		ret = btrfs_get_bdev_and_sb(device_path,
1619
					    FMODE_WRITE | FMODE_EXCL,
1620 1621 1622
					    root->fs_info->bdev_holder, 0,
					    &bdev, &bh);
		if (ret)
1623 1624
			goto out;
		disk_super = (struct btrfs_super_block *)bh->b_data;
1625
		devid = btrfs_stack_device_id(&disk_super->dev_item);
Y
Yan Zheng 已提交
1626
		dev_uuid = disk_super->dev_item.uuid;
1627
		device = btrfs_find_device(root->fs_info, devid, dev_uuid,
Y
Yan Zheng 已提交
1628
					   disk_super->fsid);
1629 1630 1631 1632
		if (!device) {
			ret = -ENOENT;
			goto error_brelse;
		}
Y
Yan Zheng 已提交
1633
	}
1634

1635
	if (device->is_tgtdev_for_dev_replace) {
1636
		ret = BTRFS_ERROR_DEV_TGT_REPLACE;
1637 1638 1639
		goto error_brelse;
	}

Y
Yan Zheng 已提交
1640
	if (device->writeable && root->fs_info->fs_devices->rw_devices == 1) {
1641
		ret = BTRFS_ERROR_DEV_ONLY_WRITABLE;
Y
Yan Zheng 已提交
1642 1643 1644 1645
		goto error_brelse;
	}

	if (device->writeable) {
1646
		lock_chunks(root);
Y
Yan Zheng 已提交
1647
		list_del_init(&device->dev_alloc_list);
1648
		device->fs_devices->rw_devices--;
1649
		unlock_chunks(root);
1650
		clear_super = true;
1651
	}
1652

1653
	mutex_unlock(&uuid_mutex);
1654
	ret = btrfs_shrink_device(device, 0);
1655
	mutex_lock(&uuid_mutex);
1656
	if (ret)
1657
		goto error_undo;
1658

1659 1660 1661 1662 1663
	/*
	 * TODO: the superblock still includes this device in its num_devices
	 * counter although write_all_supers() is not locked out. This
	 * could give a filesystem state which requires a degraded mount.
	 */
1664 1665
	ret = btrfs_rm_dev_item(root->fs_info->chunk_root, device);
	if (ret)
1666
		goto error_undo;
1667

Y
Yan Zheng 已提交
1668
	device->in_fs_metadata = 0;
1669
	btrfs_scrub_cancel_dev(root->fs_info, device);
1670 1671 1672 1673

	/*
	 * the device list mutex makes sure that we don't change
	 * the device list while someone else is writing out all
1674 1675 1676 1677 1678
	 * the device supers. Whoever is writing all supers, should
	 * lock the device list mutex before getting the number of
	 * devices in the super block (super_copy). Conversely,
	 * whoever updates the number of devices in the super block
	 * (super_copy) should hold the device list mutex.
1679
	 */
1680 1681

	cur_devices = device->fs_devices;
1682
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1683
	list_del_rcu(&device->dev_list);
1684

Y
Yan Zheng 已提交
1685
	device->fs_devices->num_devices--;
J
Josef Bacik 已提交
1686
	device->fs_devices->total_devices--;
Y
Yan Zheng 已提交
1687

1688
	if (device->missing)
1689
		device->fs_devices->missing_devices--;
1690

Y
Yan Zheng 已提交
1691 1692 1693 1694 1695 1696 1697
	next_device = list_entry(root->fs_info->fs_devices->devices.next,
				 struct btrfs_device, dev_list);
	if (device->bdev == root->fs_info->sb->s_bdev)
		root->fs_info->sb->s_bdev = next_device->bdev;
	if (device->bdev == root->fs_info->fs_devices->latest_bdev)
		root->fs_info->fs_devices->latest_bdev = next_device->bdev;

1698
	if (device->bdev) {
Y
Yan Zheng 已提交
1699
		device->fs_devices->open_devices--;
1700 1701 1702
		/* remove sysfs entry */
		btrfs_kobj_rm_device(root->fs_info, device);
	}
1703

1704
	call_rcu(&device->rcu, free_device);
Y
Yan Zheng 已提交
1705

1706 1707
	num_devices = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
	btrfs_set_super_num_devices(root->fs_info->super_copy, num_devices);
1708
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
1709

1710
	if (cur_devices->open_devices == 0) {
Y
Yan Zheng 已提交
1711 1712 1713
		struct btrfs_fs_devices *fs_devices;
		fs_devices = root->fs_info->fs_devices;
		while (fs_devices) {
1714 1715
			if (fs_devices->seed == cur_devices) {
				fs_devices->seed = cur_devices->seed;
Y
Yan Zheng 已提交
1716
				break;
1717
			}
Y
Yan Zheng 已提交
1718
			fs_devices = fs_devices->seed;
Y
Yan Zheng 已提交
1719
		}
1720 1721 1722
		cur_devices->seed = NULL;
		__btrfs_close_devices(cur_devices);
		free_fs_devices(cur_devices);
Y
Yan Zheng 已提交
1723 1724
	}

1725 1726 1727
	root->fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(root->fs_info);

Y
Yan Zheng 已提交
1728 1729 1730 1731
	/*
	 * at this point, the device is zero sized.  We want to
	 * remove it from the devices list and zero out the old super
	 */
1732
	if (clear_super && disk_super) {
1733 1734 1735
		u64 bytenr;
		int i;

1736 1737 1738 1739 1740 1741
		/* make sure this device isn't detected as part of
		 * the FS anymore
		 */
		memset(&disk_super->magic, 0, sizeof(disk_super->magic));
		set_buffer_dirty(bh);
		sync_dirty_buffer(bh);
1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769

		/* clear the mirror copies of super block on the disk
		 * being removed, 0th copy is been taken care above and
		 * the below would take of the rest
		 */
		for (i = 1; i < BTRFS_SUPER_MIRROR_MAX; i++) {
			bytenr = btrfs_sb_offset(i);
			if (bytenr + BTRFS_SUPER_INFO_SIZE >=
					i_size_read(bdev->bd_inode))
				break;

			brelse(bh);
			bh = __bread(bdev, bytenr / 4096,
					BTRFS_SUPER_INFO_SIZE);
			if (!bh)
				continue;

			disk_super = (struct btrfs_super_block *)bh->b_data;

			if (btrfs_super_bytenr(disk_super) != bytenr ||
				btrfs_super_magic(disk_super) != BTRFS_MAGIC) {
				continue;
			}
			memset(&disk_super->magic, 0,
						sizeof(disk_super->magic));
			set_buffer_dirty(bh);
			sync_dirty_buffer(bh);
		}
1770
	}
1771 1772 1773

	ret = 0;

1774 1775
	if (bdev) {
		/* Notify udev that device has changed */
1776
		btrfs_kobject_uevent(bdev, KOBJ_CHANGE);
1777

1778 1779 1780 1781
		/* Update ctime/mtime for device path for libblkid */
		update_dev_time(device_path);
	}

1782 1783
error_brelse:
	brelse(bh);
1784
	if (bdev)
1785
		blkdev_put(bdev, FMODE_READ | FMODE_EXCL);
1786 1787 1788
out:
	mutex_unlock(&uuid_mutex);
	return ret;
1789 1790
error_undo:
	if (device->writeable) {
1791
		lock_chunks(root);
1792 1793
		list_add(&device->dev_alloc_list,
			 &root->fs_info->fs_devices->alloc_list);
1794
		device->fs_devices->rw_devices++;
1795
		unlock_chunks(root);
1796 1797
	}
	goto error_brelse;
1798 1799
}

1800 1801
void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_fs_info *fs_info,
					struct btrfs_device *srcdev)
1802
{
1803 1804
	struct btrfs_fs_devices *fs_devices;

1805
	WARN_ON(!mutex_is_locked(&fs_info->fs_devices->device_list_mutex));
1806

1807 1808 1809 1810 1811 1812 1813
	/*
	 * in case of fs with no seed, srcdev->fs_devices will point
	 * to fs_devices of fs_info. However when the dev being replaced is
	 * a seed dev it will point to the seed's local fs_devices. In short
	 * srcdev will have its correct fs_devices in both the cases.
	 */
	fs_devices = srcdev->fs_devices;
1814

1815 1816
	list_del_rcu(&srcdev->dev_list);
	list_del_rcu(&srcdev->dev_alloc_list);
1817
	fs_devices->num_devices--;
1818
	if (srcdev->missing)
1819
		fs_devices->missing_devices--;
1820

1821 1822 1823 1824
	if (srcdev->writeable) {
		fs_devices->rw_devices--;
		/* zero out the old super if it is writable */
		btrfs_scratch_superblock(srcdev);
1825 1826
	}

1827
	if (srcdev->bdev)
1828
		fs_devices->open_devices--;
1829 1830 1831 1832 1833 1834
}

void btrfs_rm_dev_replace_free_srcdev(struct btrfs_fs_info *fs_info,
				      struct btrfs_device *srcdev)
{
	struct btrfs_fs_devices *fs_devices = srcdev->fs_devices;
1835 1836

	call_rcu(&srcdev->rcu, free_device);
1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856

	/*
	 * unless fs_devices is seed fs, num_devices shouldn't go
	 * zero
	 */
	BUG_ON(!fs_devices->num_devices && !fs_devices->seeding);

	/* if this is no devs we rather delete the fs_devices */
	if (!fs_devices->num_devices) {
		struct btrfs_fs_devices *tmp_fs_devices;

		tmp_fs_devices = fs_info->fs_devices;
		while (tmp_fs_devices) {
			if (tmp_fs_devices->seed == fs_devices) {
				tmp_fs_devices->seed = fs_devices->seed;
				break;
			}
			tmp_fs_devices = tmp_fs_devices->seed;
		}
		fs_devices->seed = NULL;
1857 1858
		__btrfs_close_devices(fs_devices);
		free_fs_devices(fs_devices);
1859
	}
1860 1861 1862 1863 1864 1865 1866
}

void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
				      struct btrfs_device *tgtdev)
{
	struct btrfs_device *next_device;

1867
	mutex_lock(&uuid_mutex);
1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886
	WARN_ON(!tgtdev);
	mutex_lock(&fs_info->fs_devices->device_list_mutex);
	if (tgtdev->bdev) {
		btrfs_scratch_superblock(tgtdev);
		fs_info->fs_devices->open_devices--;
	}
	fs_info->fs_devices->num_devices--;

	next_device = list_entry(fs_info->fs_devices->devices.next,
				 struct btrfs_device, dev_list);
	if (tgtdev->bdev == fs_info->sb->s_bdev)
		fs_info->sb->s_bdev = next_device->bdev;
	if (tgtdev->bdev == fs_info->fs_devices->latest_bdev)
		fs_info->fs_devices->latest_bdev = next_device->bdev;
	list_del_rcu(&tgtdev->dev_list);

	call_rcu(&tgtdev->rcu, free_device);

	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
1887
	mutex_unlock(&uuid_mutex);
1888 1889
}

1890 1891
static int btrfs_find_device_by_path(struct btrfs_root *root, char *device_path,
				     struct btrfs_device **device)
1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907
{
	int ret = 0;
	struct btrfs_super_block *disk_super;
	u64 devid;
	u8 *dev_uuid;
	struct block_device *bdev;
	struct buffer_head *bh;

	*device = NULL;
	ret = btrfs_get_bdev_and_sb(device_path, FMODE_READ,
				    root->fs_info->bdev_holder, 0, &bdev, &bh);
	if (ret)
		return ret;
	disk_super = (struct btrfs_super_block *)bh->b_data;
	devid = btrfs_stack_device_id(&disk_super->dev_item);
	dev_uuid = disk_super->dev_item.uuid;
1908
	*device = btrfs_find_device(root->fs_info, devid, dev_uuid,
1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938
				    disk_super->fsid);
	brelse(bh);
	if (!*device)
		ret = -ENOENT;
	blkdev_put(bdev, FMODE_READ);
	return ret;
}

int btrfs_find_device_missing_or_by_path(struct btrfs_root *root,
					 char *device_path,
					 struct btrfs_device **device)
{
	*device = NULL;
	if (strcmp(device_path, "missing") == 0) {
		struct list_head *devices;
		struct btrfs_device *tmp;

		devices = &root->fs_info->fs_devices->devices;
		/*
		 * It is safe to read the devices since the volume_mutex
		 * is held by the caller.
		 */
		list_for_each_entry(tmp, devices, dev_list) {
			if (tmp->in_fs_metadata && !tmp->bdev) {
				*device = tmp;
				break;
			}
		}

		if (!*device) {
1939
			btrfs_err(root->fs_info, "no missing device found");
1940 1941 1942 1943 1944 1945 1946 1947 1948
			return -ENOENT;
		}

		return 0;
	} else {
		return btrfs_find_device_by_path(root, device_path, device);
	}
}

Y
Yan Zheng 已提交
1949 1950 1951
/*
 * does all the dirty work required for changing file system's UUID.
 */
1952
static int btrfs_prepare_sprout(struct btrfs_root *root)
Y
Yan Zheng 已提交
1953 1954 1955
{
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
	struct btrfs_fs_devices *old_devices;
Y
Yan Zheng 已提交
1956
	struct btrfs_fs_devices *seed_devices;
1957
	struct btrfs_super_block *disk_super = root->fs_info->super_copy;
Y
Yan Zheng 已提交
1958 1959 1960 1961
	struct btrfs_device *device;
	u64 super_flags;

	BUG_ON(!mutex_is_locked(&uuid_mutex));
Y
Yan Zheng 已提交
1962
	if (!fs_devices->seeding)
Y
Yan Zheng 已提交
1963 1964
		return -EINVAL;

1965 1966 1967
	seed_devices = __alloc_fs_devices();
	if (IS_ERR(seed_devices))
		return PTR_ERR(seed_devices);
Y
Yan Zheng 已提交
1968

Y
Yan Zheng 已提交
1969 1970 1971 1972
	old_devices = clone_fs_devices(fs_devices);
	if (IS_ERR(old_devices)) {
		kfree(seed_devices);
		return PTR_ERR(old_devices);
Y
Yan Zheng 已提交
1973
	}
Y
Yan Zheng 已提交
1974

Y
Yan Zheng 已提交
1975 1976
	list_add(&old_devices->list, &fs_uuids);

Y
Yan Zheng 已提交
1977 1978 1979 1980
	memcpy(seed_devices, fs_devices, sizeof(*seed_devices));
	seed_devices->opened = 1;
	INIT_LIST_HEAD(&seed_devices->devices);
	INIT_LIST_HEAD(&seed_devices->alloc_list);
1981
	mutex_init(&seed_devices->device_list_mutex);
1982 1983

	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1984 1985
	list_splice_init_rcu(&fs_devices->devices, &seed_devices->devices,
			      synchronize_rcu);
M
Miao Xie 已提交
1986 1987
	list_for_each_entry(device, &seed_devices->devices, dev_list)
		device->fs_devices = seed_devices;
1988

M
Miao Xie 已提交
1989
	lock_chunks(root);
Y
Yan Zheng 已提交
1990
	list_splice_init(&fs_devices->alloc_list, &seed_devices->alloc_list);
M
Miao Xie 已提交
1991
	unlock_chunks(root);
Y
Yan Zheng 已提交
1992

Y
Yan Zheng 已提交
1993 1994 1995
	fs_devices->seeding = 0;
	fs_devices->num_devices = 0;
	fs_devices->open_devices = 0;
1996 1997
	fs_devices->missing_devices = 0;
	fs_devices->rotating = 0;
Y
Yan Zheng 已提交
1998
	fs_devices->seed = seed_devices;
Y
Yan Zheng 已提交
1999 2000 2001 2002

	generate_random_uuid(fs_devices->fsid);
	memcpy(root->fs_info->fsid, fs_devices->fsid, BTRFS_FSID_SIZE);
	memcpy(disk_super->fsid, fs_devices->fsid, BTRFS_FSID_SIZE);
2003 2004
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051
	super_flags = btrfs_super_flags(disk_super) &
		      ~BTRFS_SUPER_FLAG_SEEDING;
	btrfs_set_super_flags(disk_super, super_flags);

	return 0;
}

/*
 * strore the expected generation for seed devices in device items.
 */
static int btrfs_finish_sprout(struct btrfs_trans_handle *trans,
			       struct btrfs_root *root)
{
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_dev_item *dev_item;
	struct btrfs_device *device;
	struct btrfs_key key;
	u8 fs_uuid[BTRFS_UUID_SIZE];
	u8 dev_uuid[BTRFS_UUID_SIZE];
	u64 devid;
	int ret;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	root = root->fs_info->chunk_root;
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.offset = 0;
	key.type = BTRFS_DEV_ITEM_KEY;

	while (1) {
		ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
		if (ret < 0)
			goto error;

		leaf = path->nodes[0];
next_slot:
		if (path->slots[0] >= btrfs_header_nritems(leaf)) {
			ret = btrfs_next_leaf(root, path);
			if (ret > 0)
				break;
			if (ret < 0)
				goto error;
			leaf = path->nodes[0];
			btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
2052
			btrfs_release_path(path);
Y
Yan Zheng 已提交
2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063
			continue;
		}

		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
		if (key.objectid != BTRFS_DEV_ITEMS_OBJECTID ||
		    key.type != BTRFS_DEV_ITEM_KEY)
			break;

		dev_item = btrfs_item_ptr(leaf, path->slots[0],
					  struct btrfs_dev_item);
		devid = btrfs_device_id(leaf, dev_item);
2064
		read_extent_buffer(leaf, dev_uuid, btrfs_device_uuid(dev_item),
Y
Yan Zheng 已提交
2065
				   BTRFS_UUID_SIZE);
2066
		read_extent_buffer(leaf, fs_uuid, btrfs_device_fsid(dev_item),
Y
Yan Zheng 已提交
2067
				   BTRFS_UUID_SIZE);
2068 2069
		device = btrfs_find_device(root->fs_info, devid, dev_uuid,
					   fs_uuid);
2070
		BUG_ON(!device); /* Logic error */
Y
Yan Zheng 已提交
2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086

		if (device->fs_devices->seeding) {
			btrfs_set_device_generation(leaf, dev_item,
						    device->generation);
			btrfs_mark_buffer_dirty(leaf);
		}

		path->slots[0]++;
		goto next_slot;
	}
	ret = 0;
error:
	btrfs_free_path(path);
	return ret;
}

2087 2088
int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
{
2089
	struct request_queue *q;
2090 2091 2092 2093
	struct btrfs_trans_handle *trans;
	struct btrfs_device *device;
	struct block_device *bdev;
	struct list_head *devices;
Y
Yan Zheng 已提交
2094
	struct super_block *sb = root->fs_info->sb;
2095
	struct rcu_string *name;
2096
	u64 tmp;
Y
Yan Zheng 已提交
2097
	int seeding_dev = 0;
2098 2099
	int ret = 0;

Y
Yan Zheng 已提交
2100
	if ((sb->s_flags & MS_RDONLY) && !root->fs_info->fs_devices->seeding)
2101
		return -EROFS;
2102

2103
	bdev = blkdev_get_by_path(device_path, FMODE_WRITE | FMODE_EXCL,
2104
				  root->fs_info->bdev_holder);
2105 2106
	if (IS_ERR(bdev))
		return PTR_ERR(bdev);
2107

Y
Yan Zheng 已提交
2108 2109 2110 2111 2112 2113
	if (root->fs_info->fs_devices->seeding) {
		seeding_dev = 1;
		down_write(&sb->s_umount);
		mutex_lock(&uuid_mutex);
	}

2114
	filemap_write_and_wait(bdev->bd_inode->i_mapping);
2115

2116
	devices = &root->fs_info->fs_devices->devices;
2117 2118

	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
Q
Qinghuang Feng 已提交
2119
	list_for_each_entry(device, devices, dev_list) {
2120 2121
		if (device->bdev == bdev) {
			ret = -EEXIST;
2122 2123
			mutex_unlock(
				&root->fs_info->fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
2124
			goto error;
2125 2126
		}
	}
2127
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
2128

2129 2130
	device = btrfs_alloc_device(root->fs_info, NULL, NULL);
	if (IS_ERR(device)) {
2131
		/* we can safely leave the fs_devices entry around */
2132
		ret = PTR_ERR(device);
Y
Yan Zheng 已提交
2133
		goto error;
2134 2135
	}

2136 2137
	name = rcu_string_strdup(device_path, GFP_NOFS);
	if (!name) {
2138
		kfree(device);
Y
Yan Zheng 已提交
2139 2140
		ret = -ENOMEM;
		goto error;
2141
	}
2142
	rcu_assign_pointer(device->name, name);
Y
Yan Zheng 已提交
2143

2144
	trans = btrfs_start_transaction(root, 0);
2145
	if (IS_ERR(trans)) {
2146
		rcu_string_free(device->name);
2147 2148 2149 2150 2151
		kfree(device);
		ret = PTR_ERR(trans);
		goto error;
	}

2152 2153 2154
	q = bdev_get_queue(bdev);
	if (blk_queue_discard(q))
		device->can_discard = 1;
Y
Yan Zheng 已提交
2155 2156
	device->writeable = 1;
	device->generation = trans->transid;
2157 2158 2159 2160
	device->io_width = root->sectorsize;
	device->io_align = root->sectorsize;
	device->sector_size = root->sectorsize;
	device->total_bytes = i_size_read(bdev->bd_inode);
2161
	device->disk_total_bytes = device->total_bytes;
2162
	device->commit_total_bytes = device->total_bytes;
2163 2164
	device->dev_root = root->fs_info->dev_root;
	device->bdev = bdev;
2165
	device->in_fs_metadata = 1;
2166
	device->is_tgtdev_for_dev_replace = 0;
2167
	device->mode = FMODE_EXCL;
2168
	device->dev_stats_valid = 1;
Y
Yan Zheng 已提交
2169
	set_blocksize(device->bdev, 4096);
2170

Y
Yan Zheng 已提交
2171 2172
	if (seeding_dev) {
		sb->s_flags &= ~MS_RDONLY;
2173
		ret = btrfs_prepare_sprout(root);
2174
		BUG_ON(ret); /* -ENOMEM */
Y
Yan Zheng 已提交
2175
	}
2176

Y
Yan Zheng 已提交
2177
	device->fs_devices = root->fs_info->fs_devices;
2178 2179

	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
M
Miao Xie 已提交
2180
	lock_chunks(root);
2181
	list_add_rcu(&device->dev_list, &root->fs_info->fs_devices->devices);
Y
Yan Zheng 已提交
2182 2183 2184 2185 2186
	list_add(&device->dev_alloc_list,
		 &root->fs_info->fs_devices->alloc_list);
	root->fs_info->fs_devices->num_devices++;
	root->fs_info->fs_devices->open_devices++;
	root->fs_info->fs_devices->rw_devices++;
J
Josef Bacik 已提交
2187
	root->fs_info->fs_devices->total_devices++;
Y
Yan Zheng 已提交
2188
	root->fs_info->fs_devices->total_rw_bytes += device->total_bytes;
2189

2190 2191 2192 2193
	spin_lock(&root->fs_info->free_chunk_lock);
	root->fs_info->free_chunk_space += device->total_bytes;
	spin_unlock(&root->fs_info->free_chunk_lock);

C
Chris Mason 已提交
2194 2195 2196
	if (!blk_queue_nonrot(bdev_get_queue(bdev)))
		root->fs_info->fs_devices->rotating = 1;

2197
	tmp = btrfs_super_total_bytes(root->fs_info->super_copy);
2198
	btrfs_set_super_total_bytes(root->fs_info->super_copy,
2199
				    tmp + device->total_bytes);
2200

2201
	tmp = btrfs_super_num_devices(root->fs_info->super_copy);
2202
	btrfs_set_super_num_devices(root->fs_info->super_copy,
2203
				    tmp + 1);
2204 2205 2206 2207

	/* add sysfs device entry */
	btrfs_kobj_add_device(root->fs_info, device);

M
Miao Xie 已提交
2208 2209 2210 2211 2212 2213 2214
	/*
	 * we've got more storage, clear any full flags on the space
	 * infos
	 */
	btrfs_clear_space_info_full(root->fs_info);

	unlock_chunks(root);
2215
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
2216

Y
Yan Zheng 已提交
2217
	if (seeding_dev) {
M
Miao Xie 已提交
2218
		lock_chunks(root);
Y
Yan Zheng 已提交
2219
		ret = init_first_rw_device(trans, root, device);
M
Miao Xie 已提交
2220
		unlock_chunks(root);
2221 2222
		if (ret) {
			btrfs_abort_transaction(trans, root, ret);
2223
			goto error_trans;
2224
		}
M
Miao Xie 已提交
2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235
	}

	ret = btrfs_add_device(trans, root, device);
	if (ret) {
		btrfs_abort_transaction(trans, root, ret);
		goto error_trans;
	}

	if (seeding_dev) {
		char fsid_buf[BTRFS_UUID_UNPARSED_SIZE];

Y
Yan Zheng 已提交
2236
		ret = btrfs_finish_sprout(trans, root);
2237 2238
		if (ret) {
			btrfs_abort_transaction(trans, root, ret);
2239
			goto error_trans;
2240
		}
2241 2242 2243 2244 2245 2246 2247 2248

		/* Sprouting would change fsid of the mounted root,
		 * so rename the fsid on the sysfs
		 */
		snprintf(fsid_buf, BTRFS_UUID_UNPARSED_SIZE, "%pU",
						root->fs_info->fsid);
		if (kobject_rename(&root->fs_info->super_kobj, fsid_buf))
			goto error_trans;
Y
Yan Zheng 已提交
2249 2250
	}

2251 2252
	root->fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(root->fs_info);
2253
	ret = btrfs_commit_transaction(trans, root);
2254

Y
Yan Zheng 已提交
2255 2256 2257
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
2258

2259 2260 2261
		if (ret) /* transaction commit */
			return ret;

Y
Yan Zheng 已提交
2262
		ret = btrfs_relocate_sys_chunks(root);
2263 2264 2265 2266 2267
		if (ret < 0)
			btrfs_error(root->fs_info, ret,
				    "Failed to relocate sys chunks after "
				    "device initialization. This can be fixed "
				    "using the \"btrfs balance\" command.");
2268 2269 2270 2271 2272 2273 2274
		trans = btrfs_attach_transaction(root);
		if (IS_ERR(trans)) {
			if (PTR_ERR(trans) == -ENOENT)
				return 0;
			return PTR_ERR(trans);
		}
		ret = btrfs_commit_transaction(trans, root);
Y
Yan Zheng 已提交
2275
	}
2276

2277 2278
	/* Update ctime/mtime for libblkid */
	update_dev_time(device_path);
Y
Yan Zheng 已提交
2279
	return ret;
2280 2281 2282

error_trans:
	btrfs_end_transaction(trans, root);
2283
	rcu_string_free(device->name);
2284
	btrfs_kobj_rm_device(root->fs_info, device);
2285
	kfree(device);
Y
Yan Zheng 已提交
2286
error:
2287
	blkdev_put(bdev, FMODE_EXCL);
Y
Yan Zheng 已提交
2288 2289 2290 2291
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
	}
2292
	return ret;
2293 2294
}

2295
int btrfs_init_dev_replace_tgtdev(struct btrfs_root *root, char *device_path,
2296
				  struct btrfs_device *srcdev,
2297 2298 2299 2300 2301 2302 2303 2304
				  struct btrfs_device **device_out)
{
	struct request_queue *q;
	struct btrfs_device *device;
	struct block_device *bdev;
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct list_head *devices;
	struct rcu_string *name;
2305
	u64 devid = BTRFS_DEV_REPLACE_DEVID;
2306 2307 2308
	int ret = 0;

	*device_out = NULL;
2309 2310
	if (fs_info->fs_devices->seeding) {
		btrfs_err(fs_info, "the filesystem is a seed filesystem!");
2311
		return -EINVAL;
2312
	}
2313 2314 2315

	bdev = blkdev_get_by_path(device_path, FMODE_WRITE | FMODE_EXCL,
				  fs_info->bdev_holder);
2316 2317
	if (IS_ERR(bdev)) {
		btrfs_err(fs_info, "target device %s is invalid!", device_path);
2318
		return PTR_ERR(bdev);
2319
	}
2320 2321 2322 2323 2324 2325

	filemap_write_and_wait(bdev->bd_inode->i_mapping);

	devices = &fs_info->fs_devices->devices;
	list_for_each_entry(device, devices, dev_list) {
		if (device->bdev == bdev) {
2326
			btrfs_err(fs_info, "target device is in the filesystem!");
2327 2328 2329 2330 2331
			ret = -EEXIST;
			goto error;
		}
	}

2332

2333 2334
	if (i_size_read(bdev->bd_inode) <
	    btrfs_device_get_total_bytes(srcdev)) {
2335 2336 2337 2338 2339 2340
		btrfs_err(fs_info, "target device is smaller than source device!");
		ret = -EINVAL;
		goto error;
	}


2341 2342 2343
	device = btrfs_alloc_device(NULL, &devid, NULL);
	if (IS_ERR(device)) {
		ret = PTR_ERR(device);
2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363
		goto error;
	}

	name = rcu_string_strdup(device_path, GFP_NOFS);
	if (!name) {
		kfree(device);
		ret = -ENOMEM;
		goto error;
	}
	rcu_assign_pointer(device->name, name);

	q = bdev_get_queue(bdev);
	if (blk_queue_discard(q))
		device->can_discard = 1;
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
	device->writeable = 1;
	device->generation = 0;
	device->io_width = root->sectorsize;
	device->io_align = root->sectorsize;
	device->sector_size = root->sectorsize;
2364 2365 2366
	device->total_bytes = btrfs_device_get_total_bytes(srcdev);
	device->disk_total_bytes = btrfs_device_get_disk_total_bytes(srcdev);
	device->bytes_used = btrfs_device_get_bytes_used(srcdev);
2367 2368
	ASSERT(list_empty(&srcdev->resized_list));
	device->commit_total_bytes = srcdev->commit_total_bytes;
2369
	device->commit_bytes_used = device->bytes_used;
2370 2371 2372 2373 2374
	device->dev_root = fs_info->dev_root;
	device->bdev = bdev;
	device->in_fs_metadata = 1;
	device->is_tgtdev_for_dev_replace = 1;
	device->mode = FMODE_EXCL;
2375
	device->dev_stats_valid = 1;
2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401
	set_blocksize(device->bdev, 4096);
	device->fs_devices = fs_info->fs_devices;
	list_add(&device->dev_list, &fs_info->fs_devices->devices);
	fs_info->fs_devices->num_devices++;
	fs_info->fs_devices->open_devices++;
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);

	*device_out = device;
	return ret;

error:
	blkdev_put(bdev, FMODE_EXCL);
	return ret;
}

void btrfs_init_dev_replace_tgtdev_for_resume(struct btrfs_fs_info *fs_info,
					      struct btrfs_device *tgtdev)
{
	WARN_ON(fs_info->fs_devices->rw_devices == 0);
	tgtdev->io_width = fs_info->dev_root->sectorsize;
	tgtdev->io_align = fs_info->dev_root->sectorsize;
	tgtdev->sector_size = fs_info->dev_root->sectorsize;
	tgtdev->dev_root = fs_info->dev_root;
	tgtdev->in_fs_metadata = 1;
}

C
Chris Mason 已提交
2402 2403
static noinline int btrfs_update_device(struct btrfs_trans_handle *trans,
					struct btrfs_device *device)
2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_root *root;
	struct btrfs_dev_item *dev_item;
	struct extent_buffer *leaf;
	struct btrfs_key key;

	root = device->dev_root->fs_info->chunk_root;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.type = BTRFS_DEV_ITEM_KEY;
	key.offset = device->devid;

	ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
	if (ret < 0)
		goto out;

	if (ret > 0) {
		ret = -ENOENT;
		goto out;
	}

	leaf = path->nodes[0];
	dev_item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dev_item);

	btrfs_set_device_id(leaf, dev_item, device->devid);
	btrfs_set_device_type(leaf, dev_item, device->type);
	btrfs_set_device_io_align(leaf, dev_item, device->io_align);
	btrfs_set_device_io_width(leaf, dev_item, device->io_width);
	btrfs_set_device_sector_size(leaf, dev_item, device->sector_size);
2439 2440 2441 2442
	btrfs_set_device_total_bytes(leaf, dev_item,
				     btrfs_device_get_disk_total_bytes(device));
	btrfs_set_device_bytes_used(leaf, dev_item,
				    btrfs_device_get_bytes_used(device));
2443 2444 2445 2446 2447 2448 2449
	btrfs_mark_buffer_dirty(leaf);

out:
	btrfs_free_path(path);
	return ret;
}

M
Miao Xie 已提交
2450
int btrfs_grow_device(struct btrfs_trans_handle *trans,
2451 2452 2453
		      struct btrfs_device *device, u64 new_size)
{
	struct btrfs_super_block *super_copy =
2454
		device->dev_root->fs_info->super_copy;
2455
	struct btrfs_fs_devices *fs_devices;
M
Miao Xie 已提交
2456 2457
	u64 old_total;
	u64 diff;
2458

Y
Yan Zheng 已提交
2459 2460
	if (!device->writeable)
		return -EACCES;
M
Miao Xie 已提交
2461 2462 2463 2464 2465

	lock_chunks(device->dev_root);
	old_total = btrfs_super_total_bytes(super_copy);
	diff = new_size - device->total_bytes;

2466
	if (new_size <= device->total_bytes ||
M
Miao Xie 已提交
2467 2468
	    device->is_tgtdev_for_dev_replace) {
		unlock_chunks(device->dev_root);
Y
Yan Zheng 已提交
2469
		return -EINVAL;
M
Miao Xie 已提交
2470
	}
Y
Yan Zheng 已提交
2471

2472
	fs_devices = device->dev_root->fs_info->fs_devices;
Y
Yan Zheng 已提交
2473

2474
	btrfs_set_super_total_bytes(super_copy, old_total + diff);
Y
Yan Zheng 已提交
2475 2476
	device->fs_devices->total_rw_bytes += diff;

2477 2478
	btrfs_device_set_total_bytes(device, new_size);
	btrfs_device_set_disk_total_bytes(device, new_size);
2479
	btrfs_clear_space_info_full(device->dev_root->fs_info);
2480 2481 2482
	if (list_empty(&device->resized_list))
		list_add_tail(&device->resized_list,
			      &fs_devices->resized_devices);
M
Miao Xie 已提交
2483
	unlock_chunks(device->dev_root);
2484

2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506
	return btrfs_update_device(trans, device);
}

static int btrfs_free_chunk(struct btrfs_trans_handle *trans,
			    struct btrfs_root *root,
			    u64 chunk_tree, u64 chunk_objectid,
			    u64 chunk_offset)
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_key key;

	root = root->fs_info->chunk_root;
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = chunk_objectid;
	key.offset = chunk_offset;
	key.type = BTRFS_CHUNK_ITEM_KEY;

	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
2507 2508 2509 2510 2511 2512 2513 2514
	if (ret < 0)
		goto out;
	else if (ret > 0) { /* Logic error or corruption */
		btrfs_error(root->fs_info, -ENOENT,
			    "Failed lookup while freeing chunk.");
		ret = -ENOENT;
		goto out;
	}
2515 2516

	ret = btrfs_del_item(trans, root, path);
2517 2518 2519 2520
	if (ret < 0)
		btrfs_error(root->fs_info, ret,
			    "Failed to delete chunk item.");
out:
2521
	btrfs_free_path(path);
2522
	return ret;
2523 2524
}

2525
static int btrfs_del_sys_chunk(struct btrfs_root *root, u64 chunk_objectid, u64
2526 2527
			chunk_offset)
{
2528
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
2529 2530 2531 2532 2533 2534 2535 2536 2537 2538
	struct btrfs_disk_key *disk_key;
	struct btrfs_chunk *chunk;
	u8 *ptr;
	int ret = 0;
	u32 num_stripes;
	u32 array_size;
	u32 len = 0;
	u32 cur;
	struct btrfs_key key;

M
Miao Xie 已提交
2539
	lock_chunks(root);
2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568
	array_size = btrfs_super_sys_array_size(super_copy);

	ptr = super_copy->sys_chunk_array;
	cur = 0;

	while (cur < array_size) {
		disk_key = (struct btrfs_disk_key *)ptr;
		btrfs_disk_key_to_cpu(&key, disk_key);

		len = sizeof(*disk_key);

		if (key.type == BTRFS_CHUNK_ITEM_KEY) {
			chunk = (struct btrfs_chunk *)(ptr + len);
			num_stripes = btrfs_stack_chunk_num_stripes(chunk);
			len += btrfs_chunk_item_size(num_stripes);
		} else {
			ret = -EIO;
			break;
		}
		if (key.objectid == chunk_objectid &&
		    key.offset == chunk_offset) {
			memmove(ptr, ptr + len, array_size - (cur + len));
			array_size -= len;
			btrfs_set_super_sys_array_size(super_copy, array_size);
		} else {
			ptr += len;
			cur += len;
		}
	}
M
Miao Xie 已提交
2569
	unlock_chunks(root);
2570 2571 2572
	return ret;
}

2573 2574
int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
		       struct btrfs_root *root, u64 chunk_offset)
2575 2576 2577
{
	struct extent_map_tree *em_tree;
	struct extent_map *em;
2578
	struct btrfs_root *extent_root = root->fs_info->extent_root;
2579
	struct map_lookup *map;
M
Miao Xie 已提交
2580
	u64 dev_extent_len = 0;
2581 2582 2583
	u64 chunk_objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	u64 chunk_tree = root->fs_info->chunk_root->objectid;
	int i, ret = 0;
2584

2585
	/* Just in case */
2586 2587 2588
	root = root->fs_info->chunk_root;
	em_tree = &root->fs_info->mapping_tree.map_tree;

2589
	read_lock(&em_tree->lock);
2590
	em = lookup_extent_mapping(em_tree, chunk_offset, 1);
2591
	read_unlock(&em_tree->lock);
2592

2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604
	if (!em || em->start > chunk_offset ||
	    em->start + em->len < chunk_offset) {
		/*
		 * This is a logic error, but we don't want to just rely on the
		 * user having built with ASSERT enabled, so if ASSERT doens't
		 * do anything we still error out.
		 */
		ASSERT(0);
		if (em)
			free_extent_map(em);
		return -EINVAL;
	}
2605 2606 2607
	map = (struct map_lookup *)em->bdev;

	for (i = 0; i < map->num_stripes; i++) {
2608
		struct btrfs_device *device = map->stripes[i].dev;
M
Miao Xie 已提交
2609 2610 2611
		ret = btrfs_free_dev_extent(trans, device,
					    map->stripes[i].physical,
					    &dev_extent_len);
2612 2613 2614 2615
		if (ret) {
			btrfs_abort_transaction(trans, root, ret);
			goto out;
		}
2616

M
Miao Xie 已提交
2617 2618 2619 2620 2621 2622 2623 2624 2625 2626
		if (device->bytes_used > 0) {
			lock_chunks(root);
			btrfs_device_set_bytes_used(device,
					device->bytes_used - dev_extent_len);
			spin_lock(&root->fs_info->free_chunk_lock);
			root->fs_info->free_chunk_space += dev_extent_len;
			spin_unlock(&root->fs_info->free_chunk_lock);
			btrfs_clear_space_info_full(root->fs_info);
			unlock_chunks(root);
		}
2627

2628 2629
		if (map->stripes[i].dev) {
			ret = btrfs_update_device(trans, map->stripes[i].dev);
2630 2631 2632 2633
			if (ret) {
				btrfs_abort_transaction(trans, root, ret);
				goto out;
			}
2634
		}
2635 2636 2637
	}
	ret = btrfs_free_chunk(trans, root, chunk_tree, chunk_objectid,
			       chunk_offset);
2638 2639 2640 2641
	if (ret) {
		btrfs_abort_transaction(trans, root, ret);
		goto out;
	}
2642

2643 2644
	trace_btrfs_chunk_free(root, map, chunk_offset, em->len);

2645 2646
	if (map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
		ret = btrfs_del_sys_chunk(root, chunk_objectid, chunk_offset);
2647 2648 2649 2650
		if (ret) {
			btrfs_abort_transaction(trans, root, ret);
			goto out;
		}
2651 2652
	}

2653
	ret = btrfs_remove_block_group(trans, extent_root, chunk_offset, em);
2654 2655 2656 2657
	if (ret) {
		btrfs_abort_transaction(trans, extent_root, ret);
		goto out;
	}
Y
Yan Zheng 已提交
2658

2659
out:
Y
Yan Zheng 已提交
2660 2661
	/* once for us */
	free_extent_map(em);
2662 2663
	return ret;
}
Y
Yan Zheng 已提交
2664

2665 2666 2667 2668 2669 2670 2671
static int btrfs_relocate_chunk(struct btrfs_root *root,
			 u64 chunk_tree, u64 chunk_objectid,
			 u64 chunk_offset)
{
	struct btrfs_root *extent_root;
	struct btrfs_trans_handle *trans;
	int ret;
Y
Yan Zheng 已提交
2672

2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696
	root = root->fs_info->chunk_root;
	extent_root = root->fs_info->extent_root;

	ret = btrfs_can_relocate(extent_root, chunk_offset);
	if (ret)
		return -ENOSPC;

	/* step one, relocate all the extents inside this chunk */
	ret = btrfs_relocate_block_group(extent_root, chunk_offset);
	if (ret)
		return ret;

	trans = btrfs_start_transaction(root, 0);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		btrfs_std_error(root->fs_info, ret);
		return ret;
	}

	/*
	 * step two, delete the device extents and the
	 * chunk tree entries
	 */
	ret = btrfs_remove_chunk(trans, root, chunk_offset);
Y
Yan Zheng 已提交
2697
	btrfs_end_transaction(trans, root);
2698
	return ret;
Y
Yan Zheng 已提交
2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710
}

static int btrfs_relocate_sys_chunks(struct btrfs_root *root)
{
	struct btrfs_root *chunk_root = root->fs_info->chunk_root;
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_chunk *chunk;
	struct btrfs_key key;
	struct btrfs_key found_key;
	u64 chunk_tree = chunk_root->root_key.objectid;
	u64 chunk_type;
2711 2712
	bool retried = false;
	int failed = 0;
Y
Yan Zheng 已提交
2713 2714 2715 2716 2717 2718
	int ret;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

2719
again:
Y
Yan Zheng 已提交
2720 2721 2722 2723 2724 2725 2726 2727
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.offset = (u64)-1;
	key.type = BTRFS_CHUNK_ITEM_KEY;

	while (1) {
		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
		if (ret < 0)
			goto error;
2728
		BUG_ON(ret == 0); /* Corruption */
Y
Yan Zheng 已提交
2729 2730 2731 2732 2733 2734 2735

		ret = btrfs_previous_item(chunk_root, path, key.objectid,
					  key.type);
		if (ret < 0)
			goto error;
		if (ret > 0)
			break;
Z
Zheng Yan 已提交
2736

Y
Yan Zheng 已提交
2737 2738
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
Z
Zheng Yan 已提交
2739

Y
Yan Zheng 已提交
2740 2741 2742
		chunk = btrfs_item_ptr(leaf, path->slots[0],
				       struct btrfs_chunk);
		chunk_type = btrfs_chunk_type(leaf, chunk);
2743
		btrfs_release_path(path);
2744

Y
Yan Zheng 已提交
2745 2746 2747 2748
		if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) {
			ret = btrfs_relocate_chunk(chunk_root, chunk_tree,
						   found_key.objectid,
						   found_key.offset);
2749 2750
			if (ret == -ENOSPC)
				failed++;
H
HIMANGI SARAOGI 已提交
2751 2752
			else
				BUG_ON(ret);
Y
Yan Zheng 已提交
2753
		}
2754

Y
Yan Zheng 已提交
2755 2756 2757 2758 2759
		if (found_key.offset == 0)
			break;
		key.offset = found_key.offset - 1;
	}
	ret = 0;
2760 2761 2762 2763
	if (failed && !retried) {
		failed = 0;
		retried = true;
		goto again;
2764
	} else if (WARN_ON(failed && retried)) {
2765 2766
		ret = -ENOSPC;
	}
Y
Yan Zheng 已提交
2767 2768 2769
error:
	btrfs_free_path(path);
	return ret;
2770 2771
}

2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862
static int insert_balance_item(struct btrfs_root *root,
			       struct btrfs_balance_control *bctl)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_balance_item *item;
	struct btrfs_disk_balance_args disk_bargs;
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_key key;
	int ret, err;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	trans = btrfs_start_transaction(root, 0);
	if (IS_ERR(trans)) {
		btrfs_free_path(path);
		return PTR_ERR(trans);
	}

	key.objectid = BTRFS_BALANCE_OBJECTID;
	key.type = BTRFS_BALANCE_ITEM_KEY;
	key.offset = 0;

	ret = btrfs_insert_empty_item(trans, root, path, &key,
				      sizeof(*item));
	if (ret)
		goto out;

	leaf = path->nodes[0];
	item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_balance_item);

	memset_extent_buffer(leaf, 0, (unsigned long)item, sizeof(*item));

	btrfs_cpu_balance_args_to_disk(&disk_bargs, &bctl->data);
	btrfs_set_balance_data(leaf, item, &disk_bargs);
	btrfs_cpu_balance_args_to_disk(&disk_bargs, &bctl->meta);
	btrfs_set_balance_meta(leaf, item, &disk_bargs);
	btrfs_cpu_balance_args_to_disk(&disk_bargs, &bctl->sys);
	btrfs_set_balance_sys(leaf, item, &disk_bargs);

	btrfs_set_balance_flags(leaf, item, bctl->flags);

	btrfs_mark_buffer_dirty(leaf);
out:
	btrfs_free_path(path);
	err = btrfs_commit_transaction(trans, root);
	if (err && !ret)
		ret = err;
	return ret;
}

static int del_balance_item(struct btrfs_root *root)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_path *path;
	struct btrfs_key key;
	int ret, err;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	trans = btrfs_start_transaction(root, 0);
	if (IS_ERR(trans)) {
		btrfs_free_path(path);
		return PTR_ERR(trans);
	}

	key.objectid = BTRFS_BALANCE_OBJECTID;
	key.type = BTRFS_BALANCE_ITEM_KEY;
	key.offset = 0;

	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
	if (ret < 0)
		goto out;
	if (ret > 0) {
		ret = -ENOENT;
		goto out;
	}

	ret = btrfs_del_item(trans, root, path);
out:
	btrfs_free_path(path);
	err = btrfs_commit_transaction(trans, root);
	if (err && !ret)
		ret = err;
	return ret;
}

I
Ilya Dryomov 已提交
2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902
/*
 * This is a heuristic used to reduce the number of chunks balanced on
 * resume after balance was interrupted.
 */
static void update_balance_args(struct btrfs_balance_control *bctl)
{
	/*
	 * Turn on soft mode for chunk types that were being converted.
	 */
	if (bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT)
		bctl->data.flags |= BTRFS_BALANCE_ARGS_SOFT;
	if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT)
		bctl->sys.flags |= BTRFS_BALANCE_ARGS_SOFT;
	if (bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT)
		bctl->meta.flags |= BTRFS_BALANCE_ARGS_SOFT;

	/*
	 * Turn on usage filter if is not already used.  The idea is
	 * that chunks that we have already balanced should be
	 * reasonably full.  Don't do it for chunks that are being
	 * converted - that will keep us from relocating unconverted
	 * (albeit full) chunks.
	 */
	if (!(bctl->data.flags & BTRFS_BALANCE_ARGS_USAGE) &&
	    !(bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT)) {
		bctl->data.flags |= BTRFS_BALANCE_ARGS_USAGE;
		bctl->data.usage = 90;
	}
	if (!(bctl->sys.flags & BTRFS_BALANCE_ARGS_USAGE) &&
	    !(bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT)) {
		bctl->sys.flags |= BTRFS_BALANCE_ARGS_USAGE;
		bctl->sys.usage = 90;
	}
	if (!(bctl->meta.flags & BTRFS_BALANCE_ARGS_USAGE) &&
	    !(bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT)) {
		bctl->meta.flags |= BTRFS_BALANCE_ARGS_USAGE;
		bctl->meta.usage = 90;
	}
}

2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931
/*
 * Should be called with both balance and volume mutexes held to
 * serialize other volume operations (add_dev/rm_dev/resize) with
 * restriper.  Same goes for unset_balance_control.
 */
static void set_balance_control(struct btrfs_balance_control *bctl)
{
	struct btrfs_fs_info *fs_info = bctl->fs_info;

	BUG_ON(fs_info->balance_ctl);

	spin_lock(&fs_info->balance_lock);
	fs_info->balance_ctl = bctl;
	spin_unlock(&fs_info->balance_lock);
}

static void unset_balance_control(struct btrfs_fs_info *fs_info)
{
	struct btrfs_balance_control *bctl = fs_info->balance_ctl;

	BUG_ON(!fs_info->balance_ctl);

	spin_lock(&fs_info->balance_lock);
	fs_info->balance_ctl = NULL;
	spin_unlock(&fs_info->balance_lock);

	kfree(bctl);
}

I
Ilya Dryomov 已提交
2932 2933 2934 2935
/*
 * Balance filters.  Return 1 if chunk should be filtered out
 * (should not be balanced).
 */
2936
static int chunk_profiles_filter(u64 chunk_type,
I
Ilya Dryomov 已提交
2937 2938
				 struct btrfs_balance_args *bargs)
{
2939 2940
	chunk_type = chunk_to_extended(chunk_type) &
				BTRFS_EXTENDED_PROFILE_MASK;
I
Ilya Dryomov 已提交
2941

2942
	if (bargs->profiles & chunk_type)
I
Ilya Dryomov 已提交
2943 2944 2945 2946 2947
		return 0;

	return 1;
}

I
Ilya Dryomov 已提交
2948 2949 2950 2951 2952 2953 2954 2955 2956 2957
static int chunk_usage_filter(struct btrfs_fs_info *fs_info, u64 chunk_offset,
			      struct btrfs_balance_args *bargs)
{
	struct btrfs_block_group_cache *cache;
	u64 chunk_used, user_thresh;
	int ret = 1;

	cache = btrfs_lookup_block_group(fs_info, chunk_offset);
	chunk_used = btrfs_block_group_used(&cache->item);

2958
	if (bargs->usage == 0)
2959
		user_thresh = 1;
2960 2961 2962 2963 2964 2965
	else if (bargs->usage > 100)
		user_thresh = cache->key.offset;
	else
		user_thresh = div_factor_fine(cache->key.offset,
					      bargs->usage);

I
Ilya Dryomov 已提交
2966 2967 2968 2969 2970 2971 2972
	if (chunk_used < user_thresh)
		ret = 0;

	btrfs_put_block_group(cache);
	return ret;
}

I
Ilya Dryomov 已提交
2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989
static int chunk_devid_filter(struct extent_buffer *leaf,
			      struct btrfs_chunk *chunk,
			      struct btrfs_balance_args *bargs)
{
	struct btrfs_stripe *stripe;
	int num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
	int i;

	for (i = 0; i < num_stripes; i++) {
		stripe = btrfs_stripe_nr(chunk, i);
		if (btrfs_stripe_devid(leaf, stripe) == bargs->devid)
			return 0;
	}

	return 1;
}

I
Ilya Dryomov 已提交
2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006
/* [pstart, pend) */
static int chunk_drange_filter(struct extent_buffer *leaf,
			       struct btrfs_chunk *chunk,
			       u64 chunk_offset,
			       struct btrfs_balance_args *bargs)
{
	struct btrfs_stripe *stripe;
	int num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
	u64 stripe_offset;
	u64 stripe_length;
	int factor;
	int i;

	if (!(bargs->flags & BTRFS_BALANCE_ARGS_DEVID))
		return 0;

	if (btrfs_chunk_type(leaf, chunk) & (BTRFS_BLOCK_GROUP_DUP |
D
David Woodhouse 已提交
3007 3008 3009 3010 3011 3012 3013 3014 3015
	     BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10)) {
		factor = num_stripes / 2;
	} else if (btrfs_chunk_type(leaf, chunk) & BTRFS_BLOCK_GROUP_RAID5) {
		factor = num_stripes - 1;
	} else if (btrfs_chunk_type(leaf, chunk) & BTRFS_BLOCK_GROUP_RAID6) {
		factor = num_stripes - 2;
	} else {
		factor = num_stripes;
	}
I
Ilya Dryomov 已提交
3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033

	for (i = 0; i < num_stripes; i++) {
		stripe = btrfs_stripe_nr(chunk, i);
		if (btrfs_stripe_devid(leaf, stripe) != bargs->devid)
			continue;

		stripe_offset = btrfs_stripe_offset(leaf, stripe);
		stripe_length = btrfs_chunk_length(leaf, chunk);
		do_div(stripe_length, factor);

		if (stripe_offset < bargs->pend &&
		    stripe_offset + stripe_length > bargs->pstart)
			return 0;
	}

	return 1;
}

3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047
/* [vstart, vend) */
static int chunk_vrange_filter(struct extent_buffer *leaf,
			       struct btrfs_chunk *chunk,
			       u64 chunk_offset,
			       struct btrfs_balance_args *bargs)
{
	if (chunk_offset < bargs->vend &&
	    chunk_offset + btrfs_chunk_length(leaf, chunk) > bargs->vstart)
		/* at least part of the chunk is inside this vrange */
		return 0;

	return 1;
}

3048
static int chunk_soft_convert_filter(u64 chunk_type,
3049 3050 3051 3052 3053
				     struct btrfs_balance_args *bargs)
{
	if (!(bargs->flags & BTRFS_BALANCE_ARGS_CONVERT))
		return 0;

3054 3055
	chunk_type = chunk_to_extended(chunk_type) &
				BTRFS_EXTENDED_PROFILE_MASK;
3056

3057
	if (bargs->target == chunk_type)
3058 3059 3060 3061 3062
		return 1;

	return 0;
}

3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083
static int should_balance_chunk(struct btrfs_root *root,
				struct extent_buffer *leaf,
				struct btrfs_chunk *chunk, u64 chunk_offset)
{
	struct btrfs_balance_control *bctl = root->fs_info->balance_ctl;
	struct btrfs_balance_args *bargs = NULL;
	u64 chunk_type = btrfs_chunk_type(leaf, chunk);

	/* type filter */
	if (!((chunk_type & BTRFS_BLOCK_GROUP_TYPE_MASK) &
	      (bctl->flags & BTRFS_BALANCE_TYPE_MASK))) {
		return 0;
	}

	if (chunk_type & BTRFS_BLOCK_GROUP_DATA)
		bargs = &bctl->data;
	else if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM)
		bargs = &bctl->sys;
	else if (chunk_type & BTRFS_BLOCK_GROUP_METADATA)
		bargs = &bctl->meta;

I
Ilya Dryomov 已提交
3084 3085 3086 3087
	/* profiles filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_PROFILES) &&
	    chunk_profiles_filter(chunk_type, bargs)) {
		return 0;
I
Ilya Dryomov 已提交
3088 3089 3090 3091 3092 3093
	}

	/* usage filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_USAGE) &&
	    chunk_usage_filter(bctl->fs_info, chunk_offset, bargs)) {
		return 0;
I
Ilya Dryomov 已提交
3094 3095 3096 3097 3098 3099
	}

	/* devid filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_DEVID) &&
	    chunk_devid_filter(leaf, chunk, bargs)) {
		return 0;
I
Ilya Dryomov 已提交
3100 3101 3102 3103 3104 3105
	}

	/* drange filter, makes sense only with devid filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_DRANGE) &&
	    chunk_drange_filter(leaf, chunk, chunk_offset, bargs)) {
		return 0;
3106 3107 3108 3109 3110 3111
	}

	/* vrange filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_VRANGE) &&
	    chunk_vrange_filter(leaf, chunk, chunk_offset, bargs)) {
		return 0;
I
Ilya Dryomov 已提交
3112 3113
	}

3114 3115 3116 3117 3118 3119
	/* soft profile changing mode */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_SOFT) &&
	    chunk_soft_convert_filter(chunk_type, bargs)) {
		return 0;
	}

3120 3121 3122 3123 3124 3125 3126 3127 3128 3129
	/*
	 * limited by count, must be the last filter
	 */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_LIMIT)) {
		if (bargs->limit == 0)
			return 0;
		else
			bargs->limit--;
	}

3130 3131 3132
	return 1;
}

3133
static int __btrfs_balance(struct btrfs_fs_info *fs_info)
3134
{
3135
	struct btrfs_balance_control *bctl = fs_info->balance_ctl;
3136 3137 3138
	struct btrfs_root *chunk_root = fs_info->chunk_root;
	struct btrfs_root *dev_root = fs_info->dev_root;
	struct list_head *devices;
3139 3140 3141
	struct btrfs_device *device;
	u64 old_size;
	u64 size_to_free;
3142
	struct btrfs_chunk *chunk;
3143 3144 3145
	struct btrfs_path *path;
	struct btrfs_key key;
	struct btrfs_key found_key;
3146
	struct btrfs_trans_handle *trans;
3147 3148
	struct extent_buffer *leaf;
	int slot;
3149 3150
	int ret;
	int enospc_errors = 0;
3151
	bool counting = true;
3152 3153 3154
	u64 limit_data = bctl->data.limit;
	u64 limit_meta = bctl->meta.limit;
	u64 limit_sys = bctl->sys.limit;
3155 3156

	/* step one make some room on all the devices */
3157
	devices = &fs_info->fs_devices->devices;
Q
Qinghuang Feng 已提交
3158
	list_for_each_entry(device, devices, dev_list) {
3159
		old_size = btrfs_device_get_total_bytes(device);
3160 3161
		size_to_free = div_factor(old_size, 1);
		size_to_free = min(size_to_free, (u64)1 * 1024 * 1024);
Y
Yan Zheng 已提交
3162
		if (!device->writeable ||
3163 3164
		    btrfs_device_get_total_bytes(device) -
		    btrfs_device_get_bytes_used(device) > size_to_free ||
3165
		    device->is_tgtdev_for_dev_replace)
3166 3167 3168
			continue;

		ret = btrfs_shrink_device(device, old_size - size_to_free);
3169 3170
		if (ret == -ENOSPC)
			break;
3171 3172
		BUG_ON(ret);

3173
		trans = btrfs_start_transaction(dev_root, 0);
3174
		BUG_ON(IS_ERR(trans));
3175 3176 3177 3178 3179 3180 3181 3182 3183

		ret = btrfs_grow_device(trans, device, old_size);
		BUG_ON(ret);

		btrfs_end_transaction(trans, dev_root);
	}

	/* step two, relocate all the chunks */
	path = btrfs_alloc_path();
3184 3185 3186 3187
	if (!path) {
		ret = -ENOMEM;
		goto error;
	}
3188 3189 3190 3191 3192 3193

	/* zero out stat counters */
	spin_lock(&fs_info->balance_lock);
	memset(&bctl->stat, 0, sizeof(bctl->stat));
	spin_unlock(&fs_info->balance_lock);
again:
3194 3195 3196 3197 3198
	if (!counting) {
		bctl->data.limit = limit_data;
		bctl->meta.limit = limit_meta;
		bctl->sys.limit = limit_sys;
	}
3199 3200 3201 3202
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.offset = (u64)-1;
	key.type = BTRFS_CHUNK_ITEM_KEY;

C
Chris Mason 已提交
3203
	while (1) {
3204
		if ((!counting && atomic_read(&fs_info->balance_pause_req)) ||
3205
		    atomic_read(&fs_info->balance_cancel_req)) {
3206 3207 3208 3209
			ret = -ECANCELED;
			goto error;
		}

3210 3211 3212 3213 3214 3215 3216 3217 3218
		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
		if (ret < 0)
			goto error;

		/*
		 * this shouldn't happen, it means the last relocate
		 * failed
		 */
		if (ret == 0)
3219
			BUG(); /* FIXME break ? */
3220 3221 3222

		ret = btrfs_previous_item(chunk_root, path, 0,
					  BTRFS_CHUNK_ITEM_KEY);
3223 3224
		if (ret) {
			ret = 0;
3225
			break;
3226
		}
3227

3228 3229 3230
		leaf = path->nodes[0];
		slot = path->slots[0];
		btrfs_item_key_to_cpu(leaf, &found_key, slot);
3231

3232 3233
		if (found_key.objectid != key.objectid)
			break;
3234

3235 3236
		chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);

3237 3238 3239 3240 3241 3242
		if (!counting) {
			spin_lock(&fs_info->balance_lock);
			bctl->stat.considered++;
			spin_unlock(&fs_info->balance_lock);
		}

3243 3244
		ret = should_balance_chunk(chunk_root, leaf, chunk,
					   found_key.offset);
3245
		btrfs_release_path(path);
3246 3247 3248
		if (!ret)
			goto loop;

3249 3250 3251 3252 3253 3254 3255
		if (counting) {
			spin_lock(&fs_info->balance_lock);
			bctl->stat.expected++;
			spin_unlock(&fs_info->balance_lock);
			goto loop;
		}

3256 3257 3258 3259
		ret = btrfs_relocate_chunk(chunk_root,
					   chunk_root->root_key.objectid,
					   found_key.objectid,
					   found_key.offset);
3260 3261
		if (ret && ret != -ENOSPC)
			goto error;
3262
		if (ret == -ENOSPC) {
3263
			enospc_errors++;
3264 3265 3266 3267 3268
		} else {
			spin_lock(&fs_info->balance_lock);
			bctl->stat.completed++;
			spin_unlock(&fs_info->balance_lock);
		}
3269
loop:
3270 3271
		if (found_key.offset == 0)
			break;
3272
		key.offset = found_key.offset - 1;
3273
	}
3274

3275 3276 3277 3278 3279
	if (counting) {
		btrfs_release_path(path);
		counting = false;
		goto again;
	}
3280 3281
error:
	btrfs_free_path(path);
3282
	if (enospc_errors) {
3283
		btrfs_info(fs_info, "%d enospc errors during balance",
3284 3285 3286 3287 3288
		       enospc_errors);
		if (!ret)
			ret = -ENOSPC;
	}

3289 3290 3291
	return ret;
}

3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315
/**
 * alloc_profile_is_valid - see if a given profile is valid and reduced
 * @flags: profile to validate
 * @extended: if true @flags is treated as an extended profile
 */
static int alloc_profile_is_valid(u64 flags, int extended)
{
	u64 mask = (extended ? BTRFS_EXTENDED_PROFILE_MASK :
			       BTRFS_BLOCK_GROUP_PROFILE_MASK);

	flags &= ~BTRFS_BLOCK_GROUP_TYPE_MASK;

	/* 1) check that all other bits are zeroed */
	if (flags & ~mask)
		return 0;

	/* 2) see if profile is reduced */
	if (flags == 0)
		return !extended; /* "0" is valid for usual profiles */

	/* true if exactly one bit set */
	return (flags & (flags - 1)) == 0;
}

3316 3317
static inline int balance_need_close(struct btrfs_fs_info *fs_info)
{
3318 3319 3320 3321
	/* cancel requested || normal exit path */
	return atomic_read(&fs_info->balance_cancel_req) ||
		(atomic_read(&fs_info->balance_pause_req) == 0 &&
		 atomic_read(&fs_info->balance_cancel_req) == 0);
3322 3323
}

3324 3325
static void __cancel_balance(struct btrfs_fs_info *fs_info)
{
3326 3327
	int ret;

3328
	unset_balance_control(fs_info);
3329
	ret = del_balance_item(fs_info->tree_root);
3330 3331
	if (ret)
		btrfs_std_error(fs_info, ret);
3332 3333

	atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
3334 3335 3336 3337 3338 3339 3340 3341 3342
}

/*
 * Should be called with both balance and volume mutexes held
 */
int btrfs_balance(struct btrfs_balance_control *bctl,
		  struct btrfs_ioctl_balance_args *bargs)
{
	struct btrfs_fs_info *fs_info = bctl->fs_info;
3343
	u64 allowed;
3344
	int mixed = 0;
3345
	int ret;
3346
	u64 num_devices;
3347
	unsigned seq;
3348

3349
	if (btrfs_fs_closing(fs_info) ||
3350 3351
	    atomic_read(&fs_info->balance_pause_req) ||
	    atomic_read(&fs_info->balance_cancel_req)) {
3352 3353 3354 3355
		ret = -EINVAL;
		goto out;
	}

3356 3357 3358 3359
	allowed = btrfs_super_incompat_flags(fs_info->super_copy);
	if (allowed & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS)
		mixed = 1;

3360 3361 3362 3363
	/*
	 * In case of mixed groups both data and meta should be picked,
	 * and identical options should be given for both of them.
	 */
3364 3365
	allowed = BTRFS_BALANCE_DATA | BTRFS_BALANCE_METADATA;
	if (mixed && (bctl->flags & allowed)) {
3366 3367 3368
		if (!(bctl->flags & BTRFS_BALANCE_DATA) ||
		    !(bctl->flags & BTRFS_BALANCE_METADATA) ||
		    memcmp(&bctl->data, &bctl->meta, sizeof(bctl->data))) {
3369 3370
			btrfs_err(fs_info, "with mixed groups data and "
				   "metadata balance options must be the same");
3371 3372 3373 3374 3375
			ret = -EINVAL;
			goto out;
		}
	}

3376 3377 3378 3379 3380 3381 3382
	num_devices = fs_info->fs_devices->num_devices;
	btrfs_dev_replace_lock(&fs_info->dev_replace);
	if (btrfs_dev_replace_is_ongoing(&fs_info->dev_replace)) {
		BUG_ON(num_devices < 1);
		num_devices--;
	}
	btrfs_dev_replace_unlock(&fs_info->dev_replace);
3383
	allowed = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
3384
	if (num_devices == 1)
3385
		allowed |= BTRFS_BLOCK_GROUP_DUP;
3386
	else if (num_devices > 1)
3387
		allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1);
3388 3389 3390 3391 3392
	if (num_devices > 2)
		allowed |= BTRFS_BLOCK_GROUP_RAID5;
	if (num_devices > 3)
		allowed |= (BTRFS_BLOCK_GROUP_RAID10 |
			    BTRFS_BLOCK_GROUP_RAID6);
3393 3394 3395
	if ((bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
	    (!alloc_profile_is_valid(bctl->data.target, 1) ||
	     (bctl->data.target & ~allowed))) {
3396 3397
		btrfs_err(fs_info, "unable to start balance with target "
			   "data profile %llu",
3398
		       bctl->data.target);
3399 3400 3401
		ret = -EINVAL;
		goto out;
	}
3402 3403 3404
	if ((bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
	    (!alloc_profile_is_valid(bctl->meta.target, 1) ||
	     (bctl->meta.target & ~allowed))) {
3405 3406
		btrfs_err(fs_info,
			   "unable to start balance with target metadata profile %llu",
3407
		       bctl->meta.target);
3408 3409 3410
		ret = -EINVAL;
		goto out;
	}
3411 3412 3413
	if ((bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
	    (!alloc_profile_is_valid(bctl->sys.target, 1) ||
	     (bctl->sys.target & ~allowed))) {
3414 3415
		btrfs_err(fs_info,
			   "unable to start balance with target system profile %llu",
3416
		       bctl->sys.target);
3417 3418 3419 3420
		ret = -EINVAL;
		goto out;
	}

3421 3422
	/* allow dup'ed data chunks only in mixed mode */
	if (!mixed && (bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
3423
	    (bctl->data.target & BTRFS_BLOCK_GROUP_DUP)) {
3424
		btrfs_err(fs_info, "dup for data is not allowed");
3425 3426 3427 3428 3429 3430
		ret = -EINVAL;
		goto out;
	}

	/* allow to reduce meta or sys integrity only if force set */
	allowed = BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1 |
D
David Woodhouse 已提交
3431 3432 3433
			BTRFS_BLOCK_GROUP_RAID10 |
			BTRFS_BLOCK_GROUP_RAID5 |
			BTRFS_BLOCK_GROUP_RAID6;
3434 3435 3436 3437 3438 3439 3440 3441 3442 3443
	do {
		seq = read_seqbegin(&fs_info->profiles_lock);

		if (((bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
		     (fs_info->avail_system_alloc_bits & allowed) &&
		     !(bctl->sys.target & allowed)) ||
		    ((bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
		     (fs_info->avail_metadata_alloc_bits & allowed) &&
		     !(bctl->meta.target & allowed))) {
			if (bctl->flags & BTRFS_BALANCE_FORCE) {
3444
				btrfs_info(fs_info, "force reducing metadata integrity");
3445
			} else {
3446 3447
				btrfs_err(fs_info, "balance will reduce metadata "
					   "integrity, use force if you want this");
3448 3449 3450
				ret = -EINVAL;
				goto out;
			}
3451
		}
3452
	} while (read_seqretry(&fs_info->profiles_lock, seq));
3453

3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473
	if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
		int num_tolerated_disk_barrier_failures;
		u64 target = bctl->sys.target;

		num_tolerated_disk_barrier_failures =
			btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
		if (num_tolerated_disk_barrier_failures > 0 &&
		    (target &
		     (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID0 |
		      BTRFS_AVAIL_ALLOC_BIT_SINGLE)))
			num_tolerated_disk_barrier_failures = 0;
		else if (num_tolerated_disk_barrier_failures > 1 &&
			 (target &
			  (BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10)))
			num_tolerated_disk_barrier_failures = 1;

		fs_info->num_tolerated_disk_barrier_failures =
			num_tolerated_disk_barrier_failures;
	}

3474
	ret = insert_balance_item(fs_info->tree_root, bctl);
I
Ilya Dryomov 已提交
3475
	if (ret && ret != -EEXIST)
3476 3477
		goto out;

I
Ilya Dryomov 已提交
3478 3479 3480 3481 3482 3483 3484 3485 3486
	if (!(bctl->flags & BTRFS_BALANCE_RESUME)) {
		BUG_ON(ret == -EEXIST);
		set_balance_control(bctl);
	} else {
		BUG_ON(ret != -EEXIST);
		spin_lock(&fs_info->balance_lock);
		update_balance_args(bctl);
		spin_unlock(&fs_info->balance_lock);
	}
3487

3488
	atomic_inc(&fs_info->balance_running);
3489 3490 3491 3492 3493
	mutex_unlock(&fs_info->balance_mutex);

	ret = __btrfs_balance(fs_info);

	mutex_lock(&fs_info->balance_mutex);
3494
	atomic_dec(&fs_info->balance_running);
3495

3496 3497 3498 3499 3500
	if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
		fs_info->num_tolerated_disk_barrier_failures =
			btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
	}

3501 3502
	if (bargs) {
		memset(bargs, 0, sizeof(*bargs));
3503
		update_ioctl_balance_args(fs_info, 0, bargs);
3504 3505
	}

3506 3507 3508 3509 3510
	if ((ret && ret != -ECANCELED && ret != -ENOSPC) ||
	    balance_need_close(fs_info)) {
		__cancel_balance(fs_info);
	}

3511
	wake_up(&fs_info->balance_wait_q);
3512 3513 3514

	return ret;
out:
I
Ilya Dryomov 已提交
3515 3516
	if (bctl->flags & BTRFS_BALANCE_RESUME)
		__cancel_balance(fs_info);
3517
	else {
I
Ilya Dryomov 已提交
3518
		kfree(bctl);
3519 3520
		atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
	}
I
Ilya Dryomov 已提交
3521 3522 3523 3524 3525
	return ret;
}

static int balance_kthread(void *data)
{
3526
	struct btrfs_fs_info *fs_info = data;
3527
	int ret = 0;
I
Ilya Dryomov 已提交
3528 3529 3530 3531

	mutex_lock(&fs_info->volume_mutex);
	mutex_lock(&fs_info->balance_mutex);

3532
	if (fs_info->balance_ctl) {
3533
		btrfs_info(fs_info, "continuing balance");
3534
		ret = btrfs_balance(fs_info->balance_ctl, NULL);
3535
	}
I
Ilya Dryomov 已提交
3536 3537 3538

	mutex_unlock(&fs_info->balance_mutex);
	mutex_unlock(&fs_info->volume_mutex);
3539

I
Ilya Dryomov 已提交
3540 3541 3542
	return ret;
}

3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554
int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info)
{
	struct task_struct *tsk;

	spin_lock(&fs_info->balance_lock);
	if (!fs_info->balance_ctl) {
		spin_unlock(&fs_info->balance_lock);
		return 0;
	}
	spin_unlock(&fs_info->balance_lock);

	if (btrfs_test_opt(fs_info->tree_root, SKIP_BALANCE)) {
3555
		btrfs_info(fs_info, "force skipping balance");
3556 3557 3558 3559
		return 0;
	}

	tsk = kthread_run(balance_kthread, fs_info, "btrfs-balance");
3560
	return PTR_ERR_OR_ZERO(tsk);
3561 3562
}

3563
int btrfs_recover_balance(struct btrfs_fs_info *fs_info)
I
Ilya Dryomov 已提交
3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580
{
	struct btrfs_balance_control *bctl;
	struct btrfs_balance_item *item;
	struct btrfs_disk_balance_args disk_bargs;
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_key key;
	int ret;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = BTRFS_BALANCE_OBJECTID;
	key.type = BTRFS_BALANCE_ITEM_KEY;
	key.offset = 0;

3581
	ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0);
I
Ilya Dryomov 已提交
3582
	if (ret < 0)
3583
		goto out;
I
Ilya Dryomov 已提交
3584 3585
	if (ret > 0) { /* ret = -ENOENT; */
		ret = 0;
3586 3587 3588 3589 3590 3591 3592
		goto out;
	}

	bctl = kzalloc(sizeof(*bctl), GFP_NOFS);
	if (!bctl) {
		ret = -ENOMEM;
		goto out;
I
Ilya Dryomov 已提交
3593 3594 3595 3596 3597
	}

	leaf = path->nodes[0];
	item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_balance_item);

3598 3599 3600
	bctl->fs_info = fs_info;
	bctl->flags = btrfs_balance_flags(leaf, item);
	bctl->flags |= BTRFS_BALANCE_RESUME;
I
Ilya Dryomov 已提交
3601 3602 3603 3604 3605 3606 3607 3608

	btrfs_balance_data(leaf, item, &disk_bargs);
	btrfs_disk_balance_args_to_cpu(&bctl->data, &disk_bargs);
	btrfs_balance_meta(leaf, item, &disk_bargs);
	btrfs_disk_balance_args_to_cpu(&bctl->meta, &disk_bargs);
	btrfs_balance_sys(leaf, item, &disk_bargs);
	btrfs_disk_balance_args_to_cpu(&bctl->sys, &disk_bargs);

3609 3610
	WARN_ON(atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1));

3611 3612
	mutex_lock(&fs_info->volume_mutex);
	mutex_lock(&fs_info->balance_mutex);
I
Ilya Dryomov 已提交
3613

3614 3615 3616 3617
	set_balance_control(bctl);

	mutex_unlock(&fs_info->balance_mutex);
	mutex_unlock(&fs_info->volume_mutex);
I
Ilya Dryomov 已提交
3618 3619
out:
	btrfs_free_path(path);
3620 3621 3622
	return ret;
}

3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651
int btrfs_pause_balance(struct btrfs_fs_info *fs_info)
{
	int ret = 0;

	mutex_lock(&fs_info->balance_mutex);
	if (!fs_info->balance_ctl) {
		mutex_unlock(&fs_info->balance_mutex);
		return -ENOTCONN;
	}

	if (atomic_read(&fs_info->balance_running)) {
		atomic_inc(&fs_info->balance_pause_req);
		mutex_unlock(&fs_info->balance_mutex);

		wait_event(fs_info->balance_wait_q,
			   atomic_read(&fs_info->balance_running) == 0);

		mutex_lock(&fs_info->balance_mutex);
		/* we are good with balance_ctl ripped off from under us */
		BUG_ON(atomic_read(&fs_info->balance_running));
		atomic_dec(&fs_info->balance_pause_req);
	} else {
		ret = -ENOTCONN;
	}

	mutex_unlock(&fs_info->balance_mutex);
	return ret;
}

3652 3653
int btrfs_cancel_balance(struct btrfs_fs_info *fs_info)
{
3654 3655 3656
	if (fs_info->sb->s_flags & MS_RDONLY)
		return -EROFS;

3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690
	mutex_lock(&fs_info->balance_mutex);
	if (!fs_info->balance_ctl) {
		mutex_unlock(&fs_info->balance_mutex);
		return -ENOTCONN;
	}

	atomic_inc(&fs_info->balance_cancel_req);
	/*
	 * if we are running just wait and return, balance item is
	 * deleted in btrfs_balance in this case
	 */
	if (atomic_read(&fs_info->balance_running)) {
		mutex_unlock(&fs_info->balance_mutex);
		wait_event(fs_info->balance_wait_q,
			   atomic_read(&fs_info->balance_running) == 0);
		mutex_lock(&fs_info->balance_mutex);
	} else {
		/* __cancel_balance needs volume_mutex */
		mutex_unlock(&fs_info->balance_mutex);
		mutex_lock(&fs_info->volume_mutex);
		mutex_lock(&fs_info->balance_mutex);

		if (fs_info->balance_ctl)
			__cancel_balance(fs_info);

		mutex_unlock(&fs_info->volume_mutex);
	}

	BUG_ON(fs_info->balance_ctl || atomic_read(&fs_info->balance_running));
	atomic_dec(&fs_info->balance_cancel_req);
	mutex_unlock(&fs_info->balance_mutex);
	return 0;
}

S
Stefan Behrens 已提交
3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702
static int btrfs_uuid_scan_kthread(void *data)
{
	struct btrfs_fs_info *fs_info = data;
	struct btrfs_root *root = fs_info->tree_root;
	struct btrfs_key key;
	struct btrfs_key max_key;
	struct btrfs_path *path = NULL;
	int ret = 0;
	struct extent_buffer *eb;
	int slot;
	struct btrfs_root_item root_item;
	u32 item_size;
3703
	struct btrfs_trans_handle *trans = NULL;
S
Stefan Behrens 已提交
3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719

	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto out;
	}

	key.objectid = 0;
	key.type = BTRFS_ROOT_ITEM_KEY;
	key.offset = 0;

	max_key.objectid = (u64)-1;
	max_key.type = BTRFS_ROOT_ITEM_KEY;
	max_key.offset = (u64)-1;

	while (1) {
3720
		ret = btrfs_search_forward(root, &key, path, 0);
S
Stefan Behrens 已提交
3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743
		if (ret) {
			if (ret > 0)
				ret = 0;
			break;
		}

		if (key.type != BTRFS_ROOT_ITEM_KEY ||
		    (key.objectid < BTRFS_FIRST_FREE_OBJECTID &&
		     key.objectid != BTRFS_FS_TREE_OBJECTID) ||
		    key.objectid > BTRFS_LAST_FREE_OBJECTID)
			goto skip;

		eb = path->nodes[0];
		slot = path->slots[0];
		item_size = btrfs_item_size_nr(eb, slot);
		if (item_size < sizeof(root_item))
			goto skip;

		read_extent_buffer(eb, &root_item,
				   btrfs_item_ptr_offset(eb, slot),
				   (int)sizeof(root_item));
		if (btrfs_root_refs(&root_item) == 0)
			goto skip;
3744 3745 3746 3747 3748 3749 3750

		if (!btrfs_is_empty_uuid(root_item.uuid) ||
		    !btrfs_is_empty_uuid(root_item.received_uuid)) {
			if (trans)
				goto update_tree;

			btrfs_release_path(path);
S
Stefan Behrens 已提交
3751 3752 3753 3754 3755 3756 3757 3758 3759
			/*
			 * 1 - subvol uuid item
			 * 1 - received_subvol uuid item
			 */
			trans = btrfs_start_transaction(fs_info->uuid_root, 2);
			if (IS_ERR(trans)) {
				ret = PTR_ERR(trans);
				break;
			}
3760 3761 3762 3763 3764 3765
			continue;
		} else {
			goto skip;
		}
update_tree:
		if (!btrfs_is_empty_uuid(root_item.uuid)) {
S
Stefan Behrens 已提交
3766 3767 3768 3769 3770
			ret = btrfs_uuid_tree_add(trans, fs_info->uuid_root,
						  root_item.uuid,
						  BTRFS_UUID_KEY_SUBVOL,
						  key.objectid);
			if (ret < 0) {
3771
				btrfs_warn(fs_info, "uuid_tree_add failed %d",
S
Stefan Behrens 已提交
3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782
					ret);
				break;
			}
		}

		if (!btrfs_is_empty_uuid(root_item.received_uuid)) {
			ret = btrfs_uuid_tree_add(trans, fs_info->uuid_root,
						  root_item.received_uuid,
						 BTRFS_UUID_KEY_RECEIVED_SUBVOL,
						  key.objectid);
			if (ret < 0) {
3783
				btrfs_warn(fs_info, "uuid_tree_add failed %d",
S
Stefan Behrens 已提交
3784 3785 3786 3787 3788
					ret);
				break;
			}
		}

3789
skip:
S
Stefan Behrens 已提交
3790 3791
		if (trans) {
			ret = btrfs_end_transaction(trans, fs_info->uuid_root);
3792
			trans = NULL;
S
Stefan Behrens 已提交
3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814
			if (ret)
				break;
		}

		btrfs_release_path(path);
		if (key.offset < (u64)-1) {
			key.offset++;
		} else if (key.type < BTRFS_ROOT_ITEM_KEY) {
			key.offset = 0;
			key.type = BTRFS_ROOT_ITEM_KEY;
		} else if (key.objectid < (u64)-1) {
			key.offset = 0;
			key.type = BTRFS_ROOT_ITEM_KEY;
			key.objectid++;
		} else {
			break;
		}
		cond_resched();
	}

out:
	btrfs_free_path(path);
3815 3816
	if (trans && !IS_ERR(trans))
		btrfs_end_transaction(trans, fs_info->uuid_root);
S
Stefan Behrens 已提交
3817
	if (ret)
3818
		btrfs_warn(fs_info, "btrfs_uuid_scan_kthread failed %d", ret);
3819 3820
	else
		fs_info->update_uuid_tree_gen = 1;
S
Stefan Behrens 已提交
3821 3822 3823 3824
	up(&fs_info->uuid_tree_rescan_sem);
	return 0;
}

3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881
/*
 * Callback for btrfs_uuid_tree_iterate().
 * returns:
 * 0	check succeeded, the entry is not outdated.
 * < 0	if an error occured.
 * > 0	if the check failed, which means the caller shall remove the entry.
 */
static int btrfs_check_uuid_tree_entry(struct btrfs_fs_info *fs_info,
				       u8 *uuid, u8 type, u64 subid)
{
	struct btrfs_key key;
	int ret = 0;
	struct btrfs_root *subvol_root;

	if (type != BTRFS_UUID_KEY_SUBVOL &&
	    type != BTRFS_UUID_KEY_RECEIVED_SUBVOL)
		goto out;

	key.objectid = subid;
	key.type = BTRFS_ROOT_ITEM_KEY;
	key.offset = (u64)-1;
	subvol_root = btrfs_read_fs_root_no_name(fs_info, &key);
	if (IS_ERR(subvol_root)) {
		ret = PTR_ERR(subvol_root);
		if (ret == -ENOENT)
			ret = 1;
		goto out;
	}

	switch (type) {
	case BTRFS_UUID_KEY_SUBVOL:
		if (memcmp(uuid, subvol_root->root_item.uuid, BTRFS_UUID_SIZE))
			ret = 1;
		break;
	case BTRFS_UUID_KEY_RECEIVED_SUBVOL:
		if (memcmp(uuid, subvol_root->root_item.received_uuid,
			   BTRFS_UUID_SIZE))
			ret = 1;
		break;
	}

out:
	return ret;
}

static int btrfs_uuid_rescan_kthread(void *data)
{
	struct btrfs_fs_info *fs_info = (struct btrfs_fs_info *)data;
	int ret;

	/*
	 * 1st step is to iterate through the existing UUID tree and
	 * to delete all entries that contain outdated data.
	 * 2nd step is to add all missing entries to the UUID tree.
	 */
	ret = btrfs_uuid_tree_iterate(fs_info, btrfs_check_uuid_tree_entry);
	if (ret < 0) {
3882
		btrfs_warn(fs_info, "iterating uuid_tree failed %d", ret);
3883 3884 3885 3886 3887 3888
		up(&fs_info->uuid_tree_rescan_sem);
		return ret;
	}
	return btrfs_uuid_scan_kthread(data);
}

3889 3890 3891 3892 3893
int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_root *tree_root = fs_info->tree_root;
	struct btrfs_root *uuid_root;
S
Stefan Behrens 已提交
3894 3895
	struct task_struct *task;
	int ret;
3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914

	/*
	 * 1 - root node
	 * 1 - root item
	 */
	trans = btrfs_start_transaction(tree_root, 2);
	if (IS_ERR(trans))
		return PTR_ERR(trans);

	uuid_root = btrfs_create_tree(trans, fs_info,
				      BTRFS_UUID_TREE_OBJECTID);
	if (IS_ERR(uuid_root)) {
		btrfs_abort_transaction(trans, tree_root,
					PTR_ERR(uuid_root));
		return PTR_ERR(uuid_root);
	}

	fs_info->uuid_root = uuid_root;

S
Stefan Behrens 已提交
3915 3916 3917 3918 3919 3920 3921
	ret = btrfs_commit_transaction(trans, tree_root);
	if (ret)
		return ret;

	down(&fs_info->uuid_tree_rescan_sem);
	task = kthread_run(btrfs_uuid_scan_kthread, fs_info, "btrfs-uuid");
	if (IS_ERR(task)) {
3922
		/* fs_info->update_uuid_tree_gen remains 0 in all error case */
3923
		btrfs_warn(fs_info, "failed to start uuid_scan task");
S
Stefan Behrens 已提交
3924 3925 3926 3927 3928
		up(&fs_info->uuid_tree_rescan_sem);
		return PTR_ERR(task);
	}

	return 0;
3929
}
S
Stefan Behrens 已提交
3930

3931 3932 3933 3934 3935 3936 3937 3938
int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info)
{
	struct task_struct *task;

	down(&fs_info->uuid_tree_rescan_sem);
	task = kthread_run(btrfs_uuid_rescan_kthread, fs_info, "btrfs-uuid");
	if (IS_ERR(task)) {
		/* fs_info->update_uuid_tree_gen remains 0 in all error case */
3939
		btrfs_warn(fs_info, "failed to start uuid_rescan task");
3940 3941 3942 3943 3944 3945 3946
		up(&fs_info->uuid_tree_rescan_sem);
		return PTR_ERR(task);
	}

	return 0;
}

3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963
/*
 * shrinking a device means finding all of the device extents past
 * the new size, and then following the back refs to the chunks.
 * The chunk relocation code actually frees the device extent
 */
int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_dev_extent *dev_extent = NULL;
	struct btrfs_path *path;
	u64 length;
	u64 chunk_tree;
	u64 chunk_objectid;
	u64 chunk_offset;
	int ret;
	int slot;
3964 3965
	int failed = 0;
	bool retried = false;
3966 3967
	struct extent_buffer *l;
	struct btrfs_key key;
3968
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
3969
	u64 old_total = btrfs_super_total_bytes(super_copy);
3970 3971
	u64 old_size = btrfs_device_get_total_bytes(device);
	u64 diff = old_size - new_size;
3972

3973 3974 3975
	if (device->is_tgtdev_for_dev_replace)
		return -EINVAL;

3976 3977 3978 3979 3980 3981
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	path->reada = 2;

3982 3983
	lock_chunks(root);

3984
	btrfs_device_set_total_bytes(device, new_size);
3985
	if (device->writeable) {
Y
Yan Zheng 已提交
3986
		device->fs_devices->total_rw_bytes -= diff;
3987 3988 3989 3990
		spin_lock(&root->fs_info->free_chunk_lock);
		root->fs_info->free_chunk_space -= diff;
		spin_unlock(&root->fs_info->free_chunk_lock);
	}
3991
	unlock_chunks(root);
3992

3993
again:
3994 3995 3996 3997
	key.objectid = device->devid;
	key.offset = (u64)-1;
	key.type = BTRFS_DEV_EXTENT_KEY;

3998
	do {
3999 4000 4001 4002 4003 4004 4005 4006 4007
		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
		if (ret < 0)
			goto done;

		ret = btrfs_previous_item(root, path, 0, key.type);
		if (ret < 0)
			goto done;
		if (ret) {
			ret = 0;
4008
			btrfs_release_path(path);
4009
			break;
4010 4011 4012 4013 4014 4015
		}

		l = path->nodes[0];
		slot = path->slots[0];
		btrfs_item_key_to_cpu(l, &key, path->slots[0]);

4016
		if (key.objectid != device->devid) {
4017
			btrfs_release_path(path);
4018
			break;
4019
		}
4020 4021 4022 4023

		dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
		length = btrfs_dev_extent_length(l, dev_extent);

4024
		if (key.offset + length <= new_size) {
4025
			btrfs_release_path(path);
4026
			break;
4027
		}
4028 4029 4030 4031

		chunk_tree = btrfs_dev_extent_chunk_tree(l, dev_extent);
		chunk_objectid = btrfs_dev_extent_chunk_objectid(l, dev_extent);
		chunk_offset = btrfs_dev_extent_chunk_offset(l, dev_extent);
4032
		btrfs_release_path(path);
4033 4034 4035

		ret = btrfs_relocate_chunk(root, chunk_tree, chunk_objectid,
					   chunk_offset);
4036
		if (ret && ret != -ENOSPC)
4037
			goto done;
4038 4039
		if (ret == -ENOSPC)
			failed++;
4040
	} while (key.offset-- > 0);
4041 4042 4043 4044 4045 4046 4047 4048 4049

	if (failed && !retried) {
		failed = 0;
		retried = true;
		goto again;
	} else if (failed && retried) {
		ret = -ENOSPC;
		lock_chunks(root);

4050
		btrfs_device_set_total_bytes(device, old_size);
4051 4052
		if (device->writeable)
			device->fs_devices->total_rw_bytes += diff;
4053 4054 4055
		spin_lock(&root->fs_info->free_chunk_lock);
		root->fs_info->free_chunk_space += diff;
		spin_unlock(&root->fs_info->free_chunk_lock);
4056 4057
		unlock_chunks(root);
		goto done;
4058 4059
	}

4060
	/* Shrinking succeeded, else we would be at "done". */
4061
	trans = btrfs_start_transaction(root, 0);
4062 4063 4064 4065 4066
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto done;
	}

4067
	lock_chunks(root);
4068
	btrfs_device_set_disk_total_bytes(device, new_size);
4069 4070 4071
	if (list_empty(&device->resized_list))
		list_add_tail(&device->resized_list,
			      &root->fs_info->fs_devices->resized_devices);
4072 4073 4074 4075

	WARN_ON(diff > old_total);
	btrfs_set_super_total_bytes(super_copy, old_total - diff);
	unlock_chunks(root);
M
Miao Xie 已提交
4076 4077 4078

	/* Now btrfs_update_device() will change the on-disk size. */
	ret = btrfs_update_device(trans, device);
4079
	btrfs_end_transaction(trans, root);
4080 4081 4082 4083 4084
done:
	btrfs_free_path(path);
	return ret;
}

4085
static int btrfs_add_system_chunk(struct btrfs_root *root,
4086 4087 4088
			   struct btrfs_key *key,
			   struct btrfs_chunk *chunk, int item_size)
{
4089
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
4090 4091 4092 4093
	struct btrfs_disk_key disk_key;
	u32 array_size;
	u8 *ptr;

4094
	lock_chunks(root);
4095
	array_size = btrfs_super_sys_array_size(super_copy);
4096
	if (array_size + item_size + sizeof(disk_key)
4097 4098
			> BTRFS_SYSTEM_CHUNK_ARRAY_SIZE) {
		unlock_chunks(root);
4099
		return -EFBIG;
4100
	}
4101 4102 4103 4104 4105 4106 4107 4108

	ptr = super_copy->sys_chunk_array + array_size;
	btrfs_cpu_key_to_disk(&disk_key, key);
	memcpy(ptr, &disk_key, sizeof(disk_key));
	ptr += sizeof(disk_key);
	memcpy(ptr, chunk, item_size);
	item_size += sizeof(disk_key);
	btrfs_set_super_sys_array_size(super_copy, array_size + item_size);
4109 4110
	unlock_chunks(root);

4111 4112 4113
	return 0;
}

4114 4115 4116 4117
/*
 * sort the devices in descending order by max_avail, total_avail
 */
static int btrfs_cmp_device_info(const void *a, const void *b)
4118
{
4119 4120
	const struct btrfs_device_info *di_a = a;
	const struct btrfs_device_info *di_b = b;
4121

4122
	if (di_a->max_avail > di_b->max_avail)
4123
		return -1;
4124
	if (di_a->max_avail < di_b->max_avail)
4125
		return 1;
4126 4127 4128 4129 4130
	if (di_a->total_avail > di_b->total_avail)
		return -1;
	if (di_a->total_avail < di_b->total_avail)
		return 1;
	return 0;
4131
}
4132

4133
static struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = {
4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173
	[BTRFS_RAID_RAID10] = {
		.sub_stripes	= 2,
		.dev_stripes	= 1,
		.devs_max	= 0,	/* 0 == as many as possible */
		.devs_min	= 4,
		.devs_increment	= 2,
		.ncopies	= 2,
	},
	[BTRFS_RAID_RAID1] = {
		.sub_stripes	= 1,
		.dev_stripes	= 1,
		.devs_max	= 2,
		.devs_min	= 2,
		.devs_increment	= 2,
		.ncopies	= 2,
	},
	[BTRFS_RAID_DUP] = {
		.sub_stripes	= 1,
		.dev_stripes	= 2,
		.devs_max	= 1,
		.devs_min	= 1,
		.devs_increment	= 1,
		.ncopies	= 2,
	},
	[BTRFS_RAID_RAID0] = {
		.sub_stripes	= 1,
		.dev_stripes	= 1,
		.devs_max	= 0,
		.devs_min	= 2,
		.devs_increment	= 1,
		.ncopies	= 1,
	},
	[BTRFS_RAID_SINGLE] = {
		.sub_stripes	= 1,
		.dev_stripes	= 1,
		.devs_max	= 1,
		.devs_min	= 1,
		.devs_increment	= 1,
		.ncopies	= 1,
	},
4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189
	[BTRFS_RAID_RAID5] = {
		.sub_stripes	= 1,
		.dev_stripes	= 1,
		.devs_max	= 0,
		.devs_min	= 2,
		.devs_increment	= 1,
		.ncopies	= 2,
	},
	[BTRFS_RAID_RAID6] = {
		.sub_stripes	= 1,
		.dev_stripes	= 1,
		.devs_max	= 0,
		.devs_min	= 3,
		.devs_increment	= 1,
		.ncopies	= 3,
	},
4190 4191
};

D
David Woodhouse 已提交
4192 4193 4194 4195 4196 4197 4198 4199
static u32 find_raid56_stripe_len(u32 data_devices, u32 dev_stripe_target)
{
	/* TODO allow them to set a preferred stripe size */
	return 64 * 1024;
}

static void check_raid56_incompat_flag(struct btrfs_fs_info *info, u64 type)
{
4200
	if (!(type & BTRFS_BLOCK_GROUP_RAID56_MASK))
D
David Woodhouse 已提交
4201 4202
		return;

4203
	btrfs_set_fs_incompat(info, RAID56);
D
David Woodhouse 已提交
4204 4205
}

4206 4207 4208 4209 4210 4211 4212 4213 4214 4215
#define BTRFS_MAX_DEVS(r) ((BTRFS_LEAF_DATA_SIZE(r)		\
			- sizeof(struct btrfs_item)		\
			- sizeof(struct btrfs_chunk))		\
			/ sizeof(struct btrfs_stripe) + 1)

#define BTRFS_MAX_DEVS_SYS_CHUNK ((BTRFS_SYSTEM_CHUNK_ARRAY_SIZE	\
				- 2 * sizeof(struct btrfs_disk_key)	\
				- 2 * sizeof(struct btrfs_chunk))	\
				/ sizeof(struct btrfs_stripe) + 1)

4216
static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
4217 4218
			       struct btrfs_root *extent_root, u64 start,
			       u64 type)
4219
{
4220 4221 4222 4223 4224 4225 4226 4227 4228
	struct btrfs_fs_info *info = extent_root->fs_info;
	struct btrfs_fs_devices *fs_devices = info->fs_devices;
	struct list_head *cur;
	struct map_lookup *map = NULL;
	struct extent_map_tree *em_tree;
	struct extent_map *em;
	struct btrfs_device_info *devices_info = NULL;
	u64 total_avail;
	int num_stripes;	/* total number of stripes to allocate */
D
David Woodhouse 已提交
4229 4230
	int data_stripes;	/* number of stripes that count for
				   block group size */
4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241
	int sub_stripes;	/* sub_stripes info for map */
	int dev_stripes;	/* stripes per dev */
	int devs_max;		/* max devs to use */
	int devs_min;		/* min devs needed */
	int devs_increment;	/* ndevs has to be a multiple of this */
	int ncopies;		/* how many copies to data has */
	int ret;
	u64 max_stripe_size;
	u64 max_chunk_size;
	u64 stripe_size;
	u64 num_bytes;
D
David Woodhouse 已提交
4242
	u64 raid_stripe_len = BTRFS_STRIPE_LEN;
4243 4244 4245
	int ndevs;
	int i;
	int j;
4246
	int index;
4247

4248
	BUG_ON(!alloc_profile_is_valid(type, 0));
4249

4250 4251
	if (list_empty(&fs_devices->alloc_list))
		return -ENOSPC;
4252

4253
	index = __get_raid_index(type);
4254

4255 4256 4257 4258 4259 4260
	sub_stripes = btrfs_raid_array[index].sub_stripes;
	dev_stripes = btrfs_raid_array[index].dev_stripes;
	devs_max = btrfs_raid_array[index].devs_max;
	devs_min = btrfs_raid_array[index].devs_min;
	devs_increment = btrfs_raid_array[index].devs_increment;
	ncopies = btrfs_raid_array[index].ncopies;
4261

4262
	if (type & BTRFS_BLOCK_GROUP_DATA) {
4263 4264
		max_stripe_size = 1024 * 1024 * 1024;
		max_chunk_size = 10 * max_stripe_size;
4265 4266
		if (!devs_max)
			devs_max = BTRFS_MAX_DEVS(info->chunk_root);
4267
	} else if (type & BTRFS_BLOCK_GROUP_METADATA) {
4268 4269 4270 4271 4272
		/* for larger filesystems, use larger metadata chunks */
		if (fs_devices->total_rw_bytes > 50ULL * 1024 * 1024 * 1024)
			max_stripe_size = 1024 * 1024 * 1024;
		else
			max_stripe_size = 256 * 1024 * 1024;
4273
		max_chunk_size = max_stripe_size;
4274 4275
		if (!devs_max)
			devs_max = BTRFS_MAX_DEVS(info->chunk_root);
4276
	} else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
C
Chris Mason 已提交
4277
		max_stripe_size = 32 * 1024 * 1024;
4278
		max_chunk_size = 2 * max_stripe_size;
4279 4280
		if (!devs_max)
			devs_max = BTRFS_MAX_DEVS_SYS_CHUNK;
4281
	} else {
4282
		btrfs_err(info, "invalid chunk type 0x%llx requested",
4283 4284
		       type);
		BUG_ON(1);
4285 4286
	}

Y
Yan Zheng 已提交
4287 4288 4289
	/* we don't want a chunk larger than 10% of writeable space */
	max_chunk_size = min(div_factor(fs_devices->total_rw_bytes, 1),
			     max_chunk_size);
4290

4291
	devices_info = kcalloc(fs_devices->rw_devices, sizeof(*devices_info),
4292 4293 4294
			       GFP_NOFS);
	if (!devices_info)
		return -ENOMEM;
4295

4296
	cur = fs_devices->alloc_list.next;
4297

4298
	/*
4299 4300
	 * in the first pass through the devices list, we gather information
	 * about the available holes on each device.
4301
	 */
4302 4303 4304 4305 4306
	ndevs = 0;
	while (cur != &fs_devices->alloc_list) {
		struct btrfs_device *device;
		u64 max_avail;
		u64 dev_offset;
4307

4308
		device = list_entry(cur, struct btrfs_device, dev_alloc_list);
4309

4310
		cur = cur->next;
4311

4312
		if (!device->writeable) {
J
Julia Lawall 已提交
4313
			WARN(1, KERN_ERR
4314
			       "BTRFS: read-only device in alloc_list\n");
4315 4316
			continue;
		}
4317

4318 4319
		if (!device->in_fs_metadata ||
		    device->is_tgtdev_for_dev_replace)
4320
			continue;
4321

4322 4323 4324 4325
		if (device->total_bytes > device->bytes_used)
			total_avail = device->total_bytes - device->bytes_used;
		else
			total_avail = 0;
4326 4327 4328 4329

		/* If there is no space on this device, skip it. */
		if (total_avail == 0)
			continue;
4330

4331
		ret = find_free_dev_extent(trans, device,
4332 4333 4334 4335
					   max_stripe_size * dev_stripes,
					   &dev_offset, &max_avail);
		if (ret && ret != -ENOSPC)
			goto error;
4336

4337 4338
		if (ret == 0)
			max_avail = max_stripe_size * dev_stripes;
4339

4340 4341
		if (max_avail < BTRFS_STRIPE_LEN * dev_stripes)
			continue;
4342

4343 4344 4345 4346 4347
		if (ndevs == fs_devices->rw_devices) {
			WARN(1, "%s: found more than %llu devices\n",
			     __func__, fs_devices->rw_devices);
			break;
		}
4348 4349 4350 4351 4352 4353
		devices_info[ndevs].dev_offset = dev_offset;
		devices_info[ndevs].max_avail = max_avail;
		devices_info[ndevs].total_avail = total_avail;
		devices_info[ndevs].dev = device;
		++ndevs;
	}
4354

4355 4356 4357 4358 4359
	/*
	 * now sort the devices by hole size / available space
	 */
	sort(devices_info, ndevs, sizeof(struct btrfs_device_info),
	     btrfs_cmp_device_info, NULL);
4360

4361 4362
	/* round down to number of usable stripes */
	ndevs -= ndevs % devs_increment;
4363

4364 4365 4366
	if (ndevs < devs_increment * sub_stripes || ndevs < devs_min) {
		ret = -ENOSPC;
		goto error;
4367
	}
4368

4369 4370 4371 4372 4373 4374 4375 4376
	if (devs_max && ndevs > devs_max)
		ndevs = devs_max;
	/*
	 * the primary goal is to maximize the number of stripes, so use as many
	 * devices as possible, even if the stripes are not maximum sized.
	 */
	stripe_size = devices_info[ndevs-1].max_avail;
	num_stripes = ndevs * dev_stripes;
4377

D
David Woodhouse 已提交
4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393
	/*
	 * this will have to be fixed for RAID1 and RAID10 over
	 * more drives
	 */
	data_stripes = num_stripes / ncopies;

	if (type & BTRFS_BLOCK_GROUP_RAID5) {
		raid_stripe_len = find_raid56_stripe_len(ndevs - 1,
				 btrfs_super_stripesize(info->super_copy));
		data_stripes = num_stripes - 1;
	}
	if (type & BTRFS_BLOCK_GROUP_RAID6) {
		raid_stripe_len = find_raid56_stripe_len(ndevs - 2,
				 btrfs_super_stripesize(info->super_copy));
		data_stripes = num_stripes - 2;
	}
4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414

	/*
	 * Use the number of data stripes to figure out how big this chunk
	 * is really going to be in terms of logical address space,
	 * and compare that answer with the max chunk size
	 */
	if (stripe_size * data_stripes > max_chunk_size) {
		u64 mask = (1ULL << 24) - 1;
		stripe_size = max_chunk_size;
		do_div(stripe_size, data_stripes);

		/* bump the answer up to a 16MB boundary */
		stripe_size = (stripe_size + mask) & ~mask;

		/* but don't go higher than the limits we found
		 * while searching for free extents
		 */
		if (stripe_size > devices_info[ndevs-1].max_avail)
			stripe_size = devices_info[ndevs-1].max_avail;
	}

4415
	do_div(stripe_size, dev_stripes);
4416 4417

	/* align to BTRFS_STRIPE_LEN */
D
David Woodhouse 已提交
4418 4419
	do_div(stripe_size, raid_stripe_len);
	stripe_size *= raid_stripe_len;
4420 4421 4422 4423 4424 4425 4426

	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
	if (!map) {
		ret = -ENOMEM;
		goto error;
	}
	map->num_stripes = num_stripes;
4427

4428 4429 4430 4431 4432 4433
	for (i = 0; i < ndevs; ++i) {
		for (j = 0; j < dev_stripes; ++j) {
			int s = i * dev_stripes + j;
			map->stripes[s].dev = devices_info[i].dev;
			map->stripes[s].physical = devices_info[i].dev_offset +
						   j * stripe_size;
4434 4435
		}
	}
Y
Yan Zheng 已提交
4436
	map->sector_size = extent_root->sectorsize;
D
David Woodhouse 已提交
4437 4438 4439
	map->stripe_len = raid_stripe_len;
	map->io_align = raid_stripe_len;
	map->io_width = raid_stripe_len;
Y
Yan Zheng 已提交
4440 4441
	map->type = type;
	map->sub_stripes = sub_stripes;
4442

D
David Woodhouse 已提交
4443
	num_bytes = stripe_size * data_stripes;
4444

4445
	trace_btrfs_chunk_alloc(info->chunk_root, map, start, num_bytes);
4446

4447
	em = alloc_extent_map();
Y
Yan Zheng 已提交
4448
	if (!em) {
4449
		kfree(map);
4450 4451
		ret = -ENOMEM;
		goto error;
4452
	}
4453
	set_bit(EXTENT_FLAG_FS_MAPPING, &em->flags);
Y
Yan Zheng 已提交
4454 4455
	em->bdev = (struct block_device *)map;
	em->start = start;
4456
	em->len = num_bytes;
Y
Yan Zheng 已提交
4457 4458
	em->block_start = 0;
	em->block_len = em->len;
4459
	em->orig_block_len = stripe_size;
4460

Y
Yan Zheng 已提交
4461
	em_tree = &extent_root->fs_info->mapping_tree.map_tree;
4462
	write_lock(&em_tree->lock);
J
Josef Bacik 已提交
4463
	ret = add_extent_mapping(em_tree, em, 0);
4464 4465 4466 4467
	if (!ret) {
		list_add_tail(&em->list, &trans->transaction->pending_chunks);
		atomic_inc(&em->refs);
	}
4468
	write_unlock(&em_tree->lock);
4469 4470
	if (ret) {
		free_extent_map(em);
4471
		goto error;
4472
	}
4473

4474 4475 4476
	ret = btrfs_make_block_group(trans, extent_root, 0, type,
				     BTRFS_FIRST_CHUNK_TREE_OBJECTID,
				     start, num_bytes);
4477 4478
	if (ret)
		goto error_del_extent;
Y
Yan Zheng 已提交
4479

4480 4481 4482 4483
	for (i = 0; i < map->num_stripes; i++) {
		num_bytes = map->stripes[i].dev->bytes_used + stripe_size;
		btrfs_device_set_bytes_used(map->stripes[i].dev, num_bytes);
	}
4484

4485 4486 4487 4488 4489
	spin_lock(&extent_root->fs_info->free_chunk_lock);
	extent_root->fs_info->free_chunk_space -= (stripe_size *
						   map->num_stripes);
	spin_unlock(&extent_root->fs_info->free_chunk_lock);

4490
	free_extent_map(em);
D
David Woodhouse 已提交
4491 4492
	check_raid56_incompat_flag(extent_root->fs_info, type);

4493
	kfree(devices_info);
Y
Yan Zheng 已提交
4494
	return 0;
4495

4496
error_del_extent:
4497 4498 4499 4500 4501 4502 4503 4504
	write_lock(&em_tree->lock);
	remove_extent_mapping(em_tree, em);
	write_unlock(&em_tree->lock);

	/* One for our allocation */
	free_extent_map(em);
	/* One for the tree reference */
	free_extent_map(em);
4505 4506
	/* One for the pending_chunks list reference */
	free_extent_map(em);
4507 4508 4509
error:
	kfree(devices_info);
	return ret;
Y
Yan Zheng 已提交
4510 4511
}

4512
int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
Y
Yan Zheng 已提交
4513
				struct btrfs_root *extent_root,
4514
				u64 chunk_offset, u64 chunk_size)
Y
Yan Zheng 已提交
4515 4516 4517 4518 4519 4520
{
	struct btrfs_key key;
	struct btrfs_root *chunk_root = extent_root->fs_info->chunk_root;
	struct btrfs_device *device;
	struct btrfs_chunk *chunk;
	struct btrfs_stripe *stripe;
4521 4522 4523 4524 4525 4526 4527
	struct extent_map_tree *em_tree;
	struct extent_map *em;
	struct map_lookup *map;
	size_t item_size;
	u64 dev_offset;
	u64 stripe_size;
	int i = 0;
Y
Yan Zheng 已提交
4528 4529
	int ret;

4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542
	em_tree = &extent_root->fs_info->mapping_tree.map_tree;
	read_lock(&em_tree->lock);
	em = lookup_extent_mapping(em_tree, chunk_offset, chunk_size);
	read_unlock(&em_tree->lock);

	if (!em) {
		btrfs_crit(extent_root->fs_info, "unable to find logical "
			   "%Lu len %Lu", chunk_offset, chunk_size);
		return -EINVAL;
	}

	if (em->start != chunk_offset || em->len != chunk_size) {
		btrfs_crit(extent_root->fs_info, "found a bad mapping, wanted"
4543
			  " %Lu-%Lu, found %Lu-%Lu", chunk_offset,
4544 4545 4546 4547 4548 4549 4550 4551 4552
			  chunk_size, em->start, em->len);
		free_extent_map(em);
		return -EINVAL;
	}

	map = (struct map_lookup *)em->bdev;
	item_size = btrfs_chunk_item_size(map->num_stripes);
	stripe_size = em->orig_block_len;

Y
Yan Zheng 已提交
4553
	chunk = kzalloc(item_size, GFP_NOFS);
4554 4555 4556 4557 4558 4559 4560 4561
	if (!chunk) {
		ret = -ENOMEM;
		goto out;
	}

	for (i = 0; i < map->num_stripes; i++) {
		device = map->stripes[i].dev;
		dev_offset = map->stripes[i].physical;
Y
Yan Zheng 已提交
4562

4563
		ret = btrfs_update_device(trans, device);
4564
		if (ret)
4565 4566 4567 4568 4569 4570 4571 4572
			goto out;
		ret = btrfs_alloc_dev_extent(trans, device,
					     chunk_root->root_key.objectid,
					     BTRFS_FIRST_CHUNK_TREE_OBJECTID,
					     chunk_offset, dev_offset,
					     stripe_size);
		if (ret)
			goto out;
Y
Yan Zheng 已提交
4573 4574 4575
	}

	stripe = &chunk->stripe;
4576 4577 4578
	for (i = 0; i < map->num_stripes; i++) {
		device = map->stripes[i].dev;
		dev_offset = map->stripes[i].physical;
4579

4580 4581 4582
		btrfs_set_stack_stripe_devid(stripe, device->devid);
		btrfs_set_stack_stripe_offset(stripe, dev_offset);
		memcpy(stripe->dev_uuid, device->uuid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
4583
		stripe++;
4584 4585
	}

Y
Yan Zheng 已提交
4586
	btrfs_set_stack_chunk_length(chunk, chunk_size);
4587
	btrfs_set_stack_chunk_owner(chunk, extent_root->root_key.objectid);
Y
Yan Zheng 已提交
4588 4589 4590 4591 4592
	btrfs_set_stack_chunk_stripe_len(chunk, map->stripe_len);
	btrfs_set_stack_chunk_type(chunk, map->type);
	btrfs_set_stack_chunk_num_stripes(chunk, map->num_stripes);
	btrfs_set_stack_chunk_io_align(chunk, map->stripe_len);
	btrfs_set_stack_chunk_io_width(chunk, map->stripe_len);
4593
	btrfs_set_stack_chunk_sector_size(chunk, extent_root->sectorsize);
Y
Yan Zheng 已提交
4594
	btrfs_set_stack_chunk_sub_stripes(chunk, map->sub_stripes);
4595

Y
Yan Zheng 已提交
4596 4597 4598
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.type = BTRFS_CHUNK_ITEM_KEY;
	key.offset = chunk_offset;
4599

Y
Yan Zheng 已提交
4600
	ret = btrfs_insert_item(trans, chunk_root, &key, chunk, item_size);
4601 4602 4603 4604 4605
	if (ret == 0 && map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
		/*
		 * TODO: Cleanup of inserted chunk root in case of
		 * failure.
		 */
4606
		ret = btrfs_add_system_chunk(chunk_root, &key, chunk,
Y
Yan Zheng 已提交
4607
					     item_size);
4608
	}
4609

4610
out:
4611
	kfree(chunk);
4612
	free_extent_map(em);
4613
	return ret;
Y
Yan Zheng 已提交
4614
}
4615

Y
Yan Zheng 已提交
4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627
/*
 * Chunk allocation falls into two parts. The first part does works
 * that make the new allocated chunk useable, but not do any operation
 * that modifies the chunk tree. The second part does the works that
 * require modifying the chunk tree. This division is important for the
 * bootstrap process of adding storage to a seed btrfs.
 */
int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
		      struct btrfs_root *extent_root, u64 type)
{
	u64 chunk_offset;

4628 4629
	chunk_offset = find_next_chunk(extent_root->fs_info);
	return __btrfs_alloc_chunk(trans, extent_root, chunk_offset, type);
Y
Yan Zheng 已提交
4630 4631
}

C
Chris Mason 已提交
4632
static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
Y
Yan Zheng 已提交
4633 4634 4635 4636 4637 4638 4639 4640 4641 4642
					 struct btrfs_root *root,
					 struct btrfs_device *device)
{
	u64 chunk_offset;
	u64 sys_chunk_offset;
	u64 alloc_profile;
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct btrfs_root *extent_root = fs_info->extent_root;
	int ret;

4643
	chunk_offset = find_next_chunk(fs_info);
4644
	alloc_profile = btrfs_get_alloc_profile(extent_root, 0);
4645 4646
	ret = __btrfs_alloc_chunk(trans, extent_root, chunk_offset,
				  alloc_profile);
4647 4648
	if (ret)
		return ret;
Y
Yan Zheng 已提交
4649

4650
	sys_chunk_offset = find_next_chunk(root->fs_info);
4651
	alloc_profile = btrfs_get_alloc_profile(fs_info->chunk_root, 0);
4652 4653
	ret = __btrfs_alloc_chunk(trans, extent_root, sys_chunk_offset,
				  alloc_profile);
4654
	return ret;
Y
Yan Zheng 已提交
4655 4656
}

4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669
static inline int btrfs_chunk_max_errors(struct map_lookup *map)
{
	int max_errors;

	if (map->type & (BTRFS_BLOCK_GROUP_RAID1 |
			 BTRFS_BLOCK_GROUP_RAID10 |
			 BTRFS_BLOCK_GROUP_RAID5 |
			 BTRFS_BLOCK_GROUP_DUP)) {
		max_errors = 1;
	} else if (map->type & BTRFS_BLOCK_GROUP_RAID6) {
		max_errors = 2;
	} else {
		max_errors = 0;
4670
	}
Y
Yan Zheng 已提交
4671

4672
	return max_errors;
Y
Yan Zheng 已提交
4673 4674 4675 4676 4677 4678 4679 4680
}

int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset)
{
	struct extent_map *em;
	struct map_lookup *map;
	struct btrfs_mapping_tree *map_tree = &root->fs_info->mapping_tree;
	int readonly = 0;
4681
	int miss_ndevs = 0;
Y
Yan Zheng 已提交
4682 4683
	int i;

4684
	read_lock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
4685
	em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
4686
	read_unlock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
4687 4688 4689 4690 4691
	if (!em)
		return 1;

	map = (struct map_lookup *)em->bdev;
	for (i = 0; i < map->num_stripes; i++) {
4692 4693 4694 4695 4696
		if (map->stripes[i].dev->missing) {
			miss_ndevs++;
			continue;
		}

Y
Yan Zheng 已提交
4697 4698
		if (!map->stripes[i].dev->writeable) {
			readonly = 1;
4699
			goto end;
Y
Yan Zheng 已提交
4700 4701
		}
	}
4702 4703 4704 4705 4706 4707 4708 4709 4710

	/*
	 * If the number of missing devices is larger than max errors,
	 * we can not write the data into that chunk successfully, so
	 * set it readonly.
	 */
	if (miss_ndevs > btrfs_chunk_max_errors(map))
		readonly = 1;
end:
4711
	free_extent_map(em);
Y
Yan Zheng 已提交
4712
	return readonly;
4713 4714 4715 4716
}

void btrfs_mapping_init(struct btrfs_mapping_tree *tree)
{
4717
	extent_map_tree_init(&tree->map_tree);
4718 4719 4720 4721 4722 4723
}

void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree)
{
	struct extent_map *em;

C
Chris Mason 已提交
4724
	while (1) {
4725
		write_lock(&tree->map_tree.lock);
4726 4727 4728
		em = lookup_extent_mapping(&tree->map_tree, 0, (u64)-1);
		if (em)
			remove_extent_mapping(&tree->map_tree, em);
4729
		write_unlock(&tree->map_tree.lock);
4730 4731 4732 4733 4734 4735 4736 4737 4738
		if (!em)
			break;
		/* once for us */
		free_extent_map(em);
		/* once for the tree */
		free_extent_map(em);
	}
}

4739
int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len)
4740
{
4741
	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
4742 4743 4744 4745 4746
	struct extent_map *em;
	struct map_lookup *map;
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	int ret;

4747
	read_lock(&em_tree->lock);
4748
	em = lookup_extent_mapping(em_tree, logical, len);
4749
	read_unlock(&em_tree->lock);
4750

4751 4752 4753 4754 4755 4756
	/*
	 * We could return errors for these cases, but that could get ugly and
	 * we'd probably do the same thing which is just not do anything else
	 * and exit, so return 1 so the callers don't try to use other copies.
	 */
	if (!em) {
4757
		btrfs_crit(fs_info, "No mapping for %Lu-%Lu", logical,
4758 4759 4760 4761 4762
			    logical+len);
		return 1;
	}

	if (em->start > logical || em->start + em->len < logical) {
4763
		btrfs_crit(fs_info, "Invalid mapping for %Lu-%Lu, got "
4764
			    "%Lu-%Lu", logical, logical+len, em->start,
4765
			    em->start + em->len);
4766
		free_extent_map(em);
4767 4768 4769
		return 1;
	}

4770 4771 4772
	map = (struct map_lookup *)em->bdev;
	if (map->type & (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1))
		ret = map->num_stripes;
C
Chris Mason 已提交
4773 4774
	else if (map->type & BTRFS_BLOCK_GROUP_RAID10)
		ret = map->sub_stripes;
D
David Woodhouse 已提交
4775 4776 4777 4778
	else if (map->type & BTRFS_BLOCK_GROUP_RAID5)
		ret = 2;
	else if (map->type & BTRFS_BLOCK_GROUP_RAID6)
		ret = 3;
4779 4780 4781
	else
		ret = 1;
	free_extent_map(em);
4782 4783 4784 4785 4786 4787

	btrfs_dev_replace_lock(&fs_info->dev_replace);
	if (btrfs_dev_replace_is_ongoing(&fs_info->dev_replace))
		ret++;
	btrfs_dev_replace_unlock(&fs_info->dev_replace);

4788 4789 4790
	return ret;
}

D
David Woodhouse 已提交
4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806
unsigned long btrfs_full_stripe_len(struct btrfs_root *root,
				    struct btrfs_mapping_tree *map_tree,
				    u64 logical)
{
	struct extent_map *em;
	struct map_lookup *map;
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	unsigned long len = root->sectorsize;

	read_lock(&em_tree->lock);
	em = lookup_extent_mapping(em_tree, logical, len);
	read_unlock(&em_tree->lock);
	BUG_ON(!em);

	BUG_ON(em->start > logical || em->start + em->len < logical);
	map = (struct map_lookup *)em->bdev;
4807
	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK)
D
David Woodhouse 已提交
4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827
		len = map->stripe_len * nr_data_stripes(map);
	free_extent_map(em);
	return len;
}

int btrfs_is_parity_mirror(struct btrfs_mapping_tree *map_tree,
			   u64 logical, u64 len, int mirror_num)
{
	struct extent_map *em;
	struct map_lookup *map;
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	int ret = 0;

	read_lock(&em_tree->lock);
	em = lookup_extent_mapping(em_tree, logical, len);
	read_unlock(&em_tree->lock);
	BUG_ON(!em);

	BUG_ON(em->start > logical || em->start + em->len < logical);
	map = (struct map_lookup *)em->bdev;
4828
	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK)
D
David Woodhouse 已提交
4829 4830 4831 4832 4833
		ret = 1;
	free_extent_map(em);
	return ret;
}

4834 4835 4836
static int find_live_mirror(struct btrfs_fs_info *fs_info,
			    struct map_lookup *map, int first, int num,
			    int optimal, int dev_replace_is_ongoing)
4837 4838
{
	int i;
4839 4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862
	int tolerance;
	struct btrfs_device *srcdev;

	if (dev_replace_is_ongoing &&
	    fs_info->dev_replace.cont_reading_from_srcdev_mode ==
	     BTRFS_DEV_REPLACE_ITEM_CONT_READING_FROM_SRCDEV_MODE_AVOID)
		srcdev = fs_info->dev_replace.srcdev;
	else
		srcdev = NULL;

	/*
	 * try to avoid the drive that is the source drive for a
	 * dev-replace procedure, only choose it if no other non-missing
	 * mirror is available
	 */
	for (tolerance = 0; tolerance < 2; tolerance++) {
		if (map->stripes[optimal].dev->bdev &&
		    (tolerance || map->stripes[optimal].dev != srcdev))
			return optimal;
		for (i = first; i < first + num; i++) {
			if (map->stripes[i].dev->bdev &&
			    (tolerance || map->stripes[i].dev != srcdev))
				return i;
		}
4863
	}
4864

4865 4866 4867 4868 4869 4870
	/* we couldn't find one that doesn't fail.  Just return something
	 * and the io error handling code will clean up eventually
	 */
	return optimal;
}

D
David Woodhouse 已提交
4871 4872 4873 4874 4875 4876
static inline int parity_smaller(u64 a, u64 b)
{
	return a > b;
}

/* Bubble-sort the stripe set to put the parity/syndrome stripes last */
4877
static void sort_parity_stripes(struct btrfs_bio *bbio, int num_stripes)
D
David Woodhouse 已提交
4878 4879 4880 4881 4882 4883 4884 4885
{
	struct btrfs_bio_stripe s;
	int i;
	u64 l;
	int again = 1;

	while (again) {
		again = 0;
4886
		for (i = 0; i < num_stripes - 1; i++) {
4887 4888
			if (parity_smaller(bbio->raid_map[i],
					   bbio->raid_map[i+1])) {
D
David Woodhouse 已提交
4889
				s = bbio->stripes[i];
4890
				l = bbio->raid_map[i];
D
David Woodhouse 已提交
4891
				bbio->stripes[i] = bbio->stripes[i+1];
4892
				bbio->raid_map[i] = bbio->raid_map[i+1];
D
David Woodhouse 已提交
4893
				bbio->stripes[i+1] = s;
4894
				bbio->raid_map[i+1] = l;
4895

D
David Woodhouse 已提交
4896 4897 4898 4899 4900 4901
				again = 1;
			}
		}
	}
}

4902 4903 4904
static struct btrfs_bio *alloc_btrfs_bio(int total_stripes, int real_stripes)
{
	struct btrfs_bio *bbio = kzalloc(
4905
		 /* the size of the btrfs_bio */
4906
		sizeof(struct btrfs_bio) +
4907
		/* plus the variable array for the stripes */
4908
		sizeof(struct btrfs_bio_stripe) * (total_stripes) +
4909
		/* plus the variable array for the tgt dev */
4910
		sizeof(int) * (real_stripes) +
4911 4912 4913 4914 4915
		/*
		 * plus the raid_map, which includes both the tgt dev
		 * and the stripes
		 */
		sizeof(u64) * (total_stripes),
4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939
		GFP_NOFS);
	if (!bbio)
		return NULL;

	atomic_set(&bbio->error, 0);
	atomic_set(&bbio->refs, 1);

	return bbio;
}

void btrfs_get_bbio(struct btrfs_bio *bbio)
{
	WARN_ON(!atomic_read(&bbio->refs));
	atomic_inc(&bbio->refs);
}

void btrfs_put_bbio(struct btrfs_bio *bbio)
{
	if (!bbio)
		return;
	if (atomic_dec_and_test(&bbio->refs))
		kfree(bbio);
}

4940
static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
4941
			     u64 logical, u64 *length,
4942
			     struct btrfs_bio **bbio_ret,
4943
			     int mirror_num, int need_raid_map)
4944 4945 4946
{
	struct extent_map *em;
	struct map_lookup *map;
4947
	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
4948 4949
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	u64 offset;
4950
	u64 stripe_offset;
4951
	u64 stripe_end_offset;
4952
	u64 stripe_nr;
4953 4954
	u64 stripe_nr_orig;
	u64 stripe_nr_end;
D
David Woodhouse 已提交
4955
	u64 stripe_len;
4956
	int stripe_index;
4957
	int i;
L
Li Zefan 已提交
4958
	int ret = 0;
4959
	int num_stripes;
4960
	int max_errors = 0;
4961
	int tgtdev_indexes = 0;
4962
	struct btrfs_bio *bbio = NULL;
4963 4964 4965
	struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
	int dev_replace_is_ongoing = 0;
	int num_alloc_stripes;
4966 4967
	int patch_the_first_stripe_for_dev_replace = 0;
	u64 physical_to_patch_in_first_stripe = 0;
D
David Woodhouse 已提交
4968
	u64 raid56_full_stripe_start = (u64)-1;
4969

4970
	read_lock(&em_tree->lock);
4971
	em = lookup_extent_mapping(em_tree, logical, *length);
4972
	read_unlock(&em_tree->lock);
4973

4974
	if (!em) {
4975
		btrfs_crit(fs_info, "unable to find logical %llu len %llu",
4976
			logical, *length);
4977 4978 4979 4980 4981
		return -EINVAL;
	}

	if (em->start > logical || em->start + em->len < logical) {
		btrfs_crit(fs_info, "found a bad mapping, wanted %Lu, "
4982
			   "found %Lu-%Lu", logical, em->start,
4983
			   em->start + em->len);
4984
		free_extent_map(em);
4985
		return -EINVAL;
4986
	}
4987 4988 4989

	map = (struct map_lookup *)em->bdev;
	offset = logical - em->start;
4990

D
David Woodhouse 已提交
4991
	stripe_len = map->stripe_len;
4992 4993 4994 4995 4996
	stripe_nr = offset;
	/*
	 * stripe_nr counts the total number of stripes we have to stride
	 * to get to this block
	 */
D
David Woodhouse 已提交
4997
	do_div(stripe_nr, stripe_len);
4998

D
David Woodhouse 已提交
4999
	stripe_offset = stripe_nr * stripe_len;
5000 5001 5002 5003 5004
	BUG_ON(offset < stripe_offset);

	/* stripe_offset is the offset of this block in its stripe*/
	stripe_offset = offset - stripe_offset;

D
David Woodhouse 已提交
5005
	/* if we're here for raid56, we need to know the stripe aligned start */
5006
	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
D
David Woodhouse 已提交
5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018
		unsigned long full_stripe_len = stripe_len * nr_data_stripes(map);
		raid56_full_stripe_start = offset;

		/* allow a write of a full stripe, but make sure we don't
		 * allow straddling of stripes
		 */
		do_div(raid56_full_stripe_start, full_stripe_len);
		raid56_full_stripe_start *= full_stripe_len;
	}

	if (rw & REQ_DISCARD) {
		/* we don't discard raid56 yet */
5019
		if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
D
David Woodhouse 已提交
5020 5021 5022
			ret = -EOPNOTSUPP;
			goto out;
		}
5023
		*length = min_t(u64, em->len - offset, *length);
D
David Woodhouse 已提交
5024 5025 5026 5027 5028
	} else if (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK) {
		u64 max_len;
		/* For writes to RAID[56], allow a full stripeset across all disks.
		   For other RAID types and for RAID[56] reads, just allow a single
		   stripe (on a single disk). */
5029
		if ((map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) &&
D
David Woodhouse 已提交
5030 5031 5032 5033 5034 5035 5036 5037
		    (rw & REQ_WRITE)) {
			max_len = stripe_len * nr_data_stripes(map) -
				(offset - raid56_full_stripe_start);
		} else {
			/* we limit the length of each bio to what fits in a stripe */
			max_len = stripe_len - stripe_offset;
		}
		*length = min_t(u64, em->len - offset, max_len);
5038 5039 5040
	} else {
		*length = em->len - offset;
	}
5041

D
David Woodhouse 已提交
5042 5043
	/* This is for when we're called from btrfs_merge_bio_hook() and all
	   it cares about is the length */
5044
	if (!bbio_ret)
5045 5046
		goto out;

5047 5048 5049 5050 5051
	btrfs_dev_replace_lock(dev_replace);
	dev_replace_is_ongoing = btrfs_dev_replace_is_ongoing(dev_replace);
	if (!dev_replace_is_ongoing)
		btrfs_dev_replace_unlock(dev_replace);

5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075
	if (dev_replace_is_ongoing && mirror_num == map->num_stripes + 1 &&
	    !(rw & (REQ_WRITE | REQ_DISCARD | REQ_GET_READ_MIRRORS)) &&
	    dev_replace->tgtdev != NULL) {
		/*
		 * in dev-replace case, for repair case (that's the only
		 * case where the mirror is selected explicitly when
		 * calling btrfs_map_block), blocks left of the left cursor
		 * can also be read from the target drive.
		 * For REQ_GET_READ_MIRRORS, the target drive is added as
		 * the last one to the array of stripes. For READ, it also
		 * needs to be supported using the same mirror number.
		 * If the requested block is not left of the left cursor,
		 * EIO is returned. This can happen because btrfs_num_copies()
		 * returns one more in the dev-replace case.
		 */
		u64 tmp_length = *length;
		struct btrfs_bio *tmp_bbio = NULL;
		int tmp_num_stripes;
		u64 srcdev_devid = dev_replace->srcdev->devid;
		int index_srcdev = 0;
		int found = 0;
		u64 physical_of_found = 0;

		ret = __btrfs_map_block(fs_info, REQ_GET_READ_MIRRORS,
5076
			     logical, &tmp_length, &tmp_bbio, 0, 0);
5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089
		if (ret) {
			WARN_ON(tmp_bbio != NULL);
			goto out;
		}

		tmp_num_stripes = tmp_bbio->num_stripes;
		if (mirror_num > tmp_num_stripes) {
			/*
			 * REQ_GET_READ_MIRRORS does not contain this
			 * mirror, that means that the requested area
			 * is not left of the left cursor
			 */
			ret = -EIO;
5090
			btrfs_put_bbio(tmp_bbio);
5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124
			goto out;
		}

		/*
		 * process the rest of the function using the mirror_num
		 * of the source drive. Therefore look it up first.
		 * At the end, patch the device pointer to the one of the
		 * target drive.
		 */
		for (i = 0; i < tmp_num_stripes; i++) {
			if (tmp_bbio->stripes[i].dev->devid == srcdev_devid) {
				/*
				 * In case of DUP, in order to keep it
				 * simple, only add the mirror with the
				 * lowest physical address
				 */
				if (found &&
				    physical_of_found <=
				     tmp_bbio->stripes[i].physical)
					continue;
				index_srcdev = i;
				found = 1;
				physical_of_found =
					tmp_bbio->stripes[i].physical;
			}
		}

		if (found) {
			mirror_num = index_srcdev + 1;
			patch_the_first_stripe_for_dev_replace = 1;
			physical_to_patch_in_first_stripe = physical_of_found;
		} else {
			WARN_ON(1);
			ret = -EIO;
5125
			btrfs_put_bbio(tmp_bbio);
5126 5127 5128
			goto out;
		}

5129
		btrfs_put_bbio(tmp_bbio);
5130 5131 5132 5133
	} else if (mirror_num > map->num_stripes) {
		mirror_num = 0;
	}

5134
	num_stripes = 1;
5135
	stripe_index = 0;
5136
	stripe_nr_orig = stripe_nr;
5137
	stripe_nr_end = ALIGN(offset + *length, map->stripe_len);
5138 5139 5140
	do_div(stripe_nr_end, map->stripe_len);
	stripe_end_offset = stripe_nr_end * map->stripe_len -
			    (offset + *length);
D
David Woodhouse 已提交
5141

5142 5143 5144 5145 5146
	if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
		if (rw & REQ_DISCARD)
			num_stripes = min_t(u64, map->num_stripes,
					    stripe_nr_end - stripe_nr_orig);
		stripe_index = do_div(stripe_nr, map->num_stripes);
5147 5148
		if (!(rw & (REQ_WRITE | REQ_DISCARD | REQ_GET_READ_MIRRORS)))
			mirror_num = 1;
5149
	} else if (map->type & BTRFS_BLOCK_GROUP_RAID1) {
5150
		if (rw & (REQ_WRITE | REQ_DISCARD | REQ_GET_READ_MIRRORS))
5151
			num_stripes = map->num_stripes;
5152
		else if (mirror_num)
5153
			stripe_index = mirror_num - 1;
5154
		else {
5155
			stripe_index = find_live_mirror(fs_info, map, 0,
5156
					    map->num_stripes,
5157 5158
					    current->pid % map->num_stripes,
					    dev_replace_is_ongoing);
5159
			mirror_num = stripe_index + 1;
5160
		}
5161

5162
	} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
5163
		if (rw & (REQ_WRITE | REQ_DISCARD | REQ_GET_READ_MIRRORS)) {
5164
			num_stripes = map->num_stripes;
5165
		} else if (mirror_num) {
5166
			stripe_index = mirror_num - 1;
5167 5168 5169
		} else {
			mirror_num = 1;
		}
5170

C
Chris Mason 已提交
5171 5172 5173 5174 5175 5176
	} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
		int factor = map->num_stripes / map->sub_stripes;

		stripe_index = do_div(stripe_nr, factor);
		stripe_index *= map->sub_stripes;

5177
		if (rw & (REQ_WRITE | REQ_GET_READ_MIRRORS))
5178
			num_stripes = map->sub_stripes;
5179 5180 5181 5182
		else if (rw & REQ_DISCARD)
			num_stripes = min_t(u64, map->sub_stripes *
					    (stripe_nr_end - stripe_nr_orig),
					    map->num_stripes);
C
Chris Mason 已提交
5183 5184
		else if (mirror_num)
			stripe_index += mirror_num - 1;
5185
		else {
J
Jan Schmidt 已提交
5186
			int old_stripe_index = stripe_index;
5187 5188
			stripe_index = find_live_mirror(fs_info, map,
					      stripe_index,
5189
					      map->sub_stripes, stripe_index +
5190 5191
					      current->pid % map->sub_stripes,
					      dev_replace_is_ongoing);
J
Jan Schmidt 已提交
5192
			mirror_num = stripe_index - old_stripe_index + 1;
5193
		}
D
David Woodhouse 已提交
5194

5195
	} else if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
5196
		if (need_raid_map &&
5197 5198
		    ((rw & (REQ_WRITE | REQ_GET_READ_MIRRORS)) ||
		     mirror_num > 1)) {
D
David Woodhouse 已提交
5199 5200
			/* push stripe_nr back to the start of the full stripe */
			stripe_nr = raid56_full_stripe_start;
5201
			do_div(stripe_nr, stripe_len * nr_data_stripes(map));
D
David Woodhouse 已提交
5202 5203 5204 5205 5206 5207 5208 5209 5210

			/* RAID[56] write or recovery. Return all stripes */
			num_stripes = map->num_stripes;
			max_errors = nr_parity_stripes(map);

			*length = map->stripe_len;
			stripe_index = 0;
			stripe_offset = 0;
		} else {
5211 5212
			u64 tmp;

D
David Woodhouse 已提交
5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225
			/*
			 * Mirror #0 or #1 means the original data block.
			 * Mirror #2 is RAID5 parity block.
			 * Mirror #3 is RAID6 Q block.
			 */
			stripe_index = do_div(stripe_nr, nr_data_stripes(map));
			if (mirror_num > 1)
				stripe_index = nr_data_stripes(map) +
						mirror_num - 2;

			/* We distribute the parity blocks across stripes */
			tmp = stripe_nr + stripe_index;
			stripe_index = do_div(tmp, map->num_stripes);
5226 5227 5228
			if (!(rw & (REQ_WRITE | REQ_DISCARD |
				    REQ_GET_READ_MIRRORS)) && mirror_num <= 1)
				mirror_num = 1;
D
David Woodhouse 已提交
5229
		}
5230 5231 5232 5233 5234 5235 5236
	} else {
		/*
		 * after this do_div call, stripe_nr is the number of stripes
		 * on this device we have to walk to find the data, and
		 * stripe_index is the number of our device in the stripe array
		 */
		stripe_index = do_div(stripe_nr, map->num_stripes);
5237
		mirror_num = stripe_index + 1;
5238
	}
5239
	BUG_ON(stripe_index >= map->num_stripes);
5240

5241
	num_alloc_stripes = num_stripes;
5242 5243 5244 5245 5246
	if (dev_replace_is_ongoing) {
		if (rw & (REQ_WRITE | REQ_DISCARD))
			num_alloc_stripes <<= 1;
		if (rw & REQ_GET_READ_MIRRORS)
			num_alloc_stripes++;
5247
		tgtdev_indexes = num_stripes;
5248
	}
5249

5250
	bbio = alloc_btrfs_bio(num_alloc_stripes, tgtdev_indexes);
L
Li Zefan 已提交
5251 5252 5253 5254
	if (!bbio) {
		ret = -ENOMEM;
		goto out;
	}
5255 5256
	if (dev_replace_is_ongoing)
		bbio->tgtdev_map = (int *)(bbio->stripes + num_alloc_stripes);
L
Li Zefan 已提交
5257

5258
	/* build raid_map */
5259
	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK &&
5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285
	    need_raid_map && ((rw & (REQ_WRITE | REQ_GET_READ_MIRRORS)) ||
	    mirror_num > 1)) {
		u64 tmp;
		int i, rot;

		bbio->raid_map = (u64 *)((void *)bbio->stripes +
				 sizeof(struct btrfs_bio_stripe) *
				 num_alloc_stripes +
				 sizeof(int) * tgtdev_indexes);

		/* Work out the disk rotation on this stripe-set */
		tmp = stripe_nr;
		rot = do_div(tmp, num_stripes);

		/* Fill in the logical address of each stripe */
		tmp = stripe_nr * nr_data_stripes(map);
		for (i = 0; i < nr_data_stripes(map); i++)
			bbio->raid_map[(i+rot) % num_stripes] =
				em->start + (tmp + i) * map->stripe_len;

		bbio->raid_map[(i+rot) % map->num_stripes] = RAID5_P_STRIPE;
		if (map->type & BTRFS_BLOCK_GROUP_RAID6)
			bbio->raid_map[(i+rot+1) % num_stripes] =
				RAID6_Q_STRIPE;
	}

5286
	if (rw & REQ_DISCARD) {
5287 5288 5289 5290
		int factor = 0;
		int sub_stripes = 0;
		u64 stripes_per_dev = 0;
		u32 remaining_stripes = 0;
L
Liu Bo 已提交
5291
		u32 last_stripe = 0;
5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304

		if (map->type &
		    (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID10)) {
			if (map->type & BTRFS_BLOCK_GROUP_RAID0)
				sub_stripes = 1;
			else
				sub_stripes = map->sub_stripes;

			factor = map->num_stripes / sub_stripes;
			stripes_per_dev = div_u64_rem(stripe_nr_end -
						      stripe_nr_orig,
						      factor,
						      &remaining_stripes);
L
Liu Bo 已提交
5305 5306
			div_u64_rem(stripe_nr_end - 1, factor, &last_stripe);
			last_stripe *= sub_stripes;
5307 5308
		}

5309
		for (i = 0; i < num_stripes; i++) {
5310
			bbio->stripes[i].physical =
5311 5312
				map->stripes[stripe_index].physical +
				stripe_offset + stripe_nr * map->stripe_len;
5313
			bbio->stripes[i].dev = map->stripes[stripe_index].dev;
5314

5315 5316 5317 5318
			if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
					 BTRFS_BLOCK_GROUP_RAID10)) {
				bbio->stripes[i].length = stripes_per_dev *
							  map->stripe_len;
L
Liu Bo 已提交
5319

5320 5321 5322
				if (i / sub_stripes < remaining_stripes)
					bbio->stripes[i].length +=
						map->stripe_len;
L
Liu Bo 已提交
5323 5324 5325 5326 5327 5328 5329 5330 5331

				/*
				 * Special for the first stripe and
				 * the last stripe:
				 *
				 * |-------|...|-------|
				 *     |----------|
				 *    off     end_off
				 */
5332
				if (i < sub_stripes)
5333
					bbio->stripes[i].length -=
5334
						stripe_offset;
L
Liu Bo 已提交
5335 5336 5337 5338

				if (stripe_index >= last_stripe &&
				    stripe_index <= (last_stripe +
						     sub_stripes - 1))
5339
					bbio->stripes[i].length -=
5340
						stripe_end_offset;
L
Liu Bo 已提交
5341

5342 5343
				if (i == sub_stripes - 1)
					stripe_offset = 0;
5344
			} else
5345
				bbio->stripes[i].length = *length;
5346 5347 5348 5349 5350 5351 5352 5353 5354 5355

			stripe_index++;
			if (stripe_index == map->num_stripes) {
				/* This could only happen for RAID0/10 */
				stripe_index = 0;
				stripe_nr++;
			}
		}
	} else {
		for (i = 0; i < num_stripes; i++) {
5356
			bbio->stripes[i].physical =
5357 5358 5359
				map->stripes[stripe_index].physical +
				stripe_offset +
				stripe_nr * map->stripe_len;
5360
			bbio->stripes[i].dev =
5361
				map->stripes[stripe_index].dev;
5362
			stripe_index++;
5363
		}
5364
	}
L
Li Zefan 已提交
5365

5366 5367
	if (rw & (REQ_WRITE | REQ_GET_READ_MIRRORS))
		max_errors = btrfs_chunk_max_errors(map);
L
Li Zefan 已提交
5368

5369 5370
	if (bbio->raid_map)
		sort_parity_stripes(bbio, num_stripes);
5371

5372
	tgtdev_indexes = 0;
5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400
	if (dev_replace_is_ongoing && (rw & (REQ_WRITE | REQ_DISCARD)) &&
	    dev_replace->tgtdev != NULL) {
		int index_where_to_add;
		u64 srcdev_devid = dev_replace->srcdev->devid;

		/*
		 * duplicate the write operations while the dev replace
		 * procedure is running. Since the copying of the old disk
		 * to the new disk takes place at run time while the
		 * filesystem is mounted writable, the regular write
		 * operations to the old disk have to be duplicated to go
		 * to the new disk as well.
		 * Note that device->missing is handled by the caller, and
		 * that the write to the old disk is already set up in the
		 * stripes array.
		 */
		index_where_to_add = num_stripes;
		for (i = 0; i < num_stripes; i++) {
			if (bbio->stripes[i].dev->devid == srcdev_devid) {
				/* write to new disk, too */
				struct btrfs_bio_stripe *new =
					bbio->stripes + index_where_to_add;
				struct btrfs_bio_stripe *old =
					bbio->stripes + i;

				new->physical = old->physical;
				new->length = old->length;
				new->dev = dev_replace->tgtdev;
5401
				bbio->tgtdev_map[i] = index_where_to_add;
5402 5403
				index_where_to_add++;
				max_errors++;
5404
				tgtdev_indexes++;
5405 5406 5407
			}
		}
		num_stripes = index_where_to_add;
5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449
	} else if (dev_replace_is_ongoing && (rw & REQ_GET_READ_MIRRORS) &&
		   dev_replace->tgtdev != NULL) {
		u64 srcdev_devid = dev_replace->srcdev->devid;
		int index_srcdev = 0;
		int found = 0;
		u64 physical_of_found = 0;

		/*
		 * During the dev-replace procedure, the target drive can
		 * also be used to read data in case it is needed to repair
		 * a corrupt block elsewhere. This is possible if the
		 * requested area is left of the left cursor. In this area,
		 * the target drive is a full copy of the source drive.
		 */
		for (i = 0; i < num_stripes; i++) {
			if (bbio->stripes[i].dev->devid == srcdev_devid) {
				/*
				 * In case of DUP, in order to keep it
				 * simple, only add the mirror with the
				 * lowest physical address
				 */
				if (found &&
				    physical_of_found <=
				     bbio->stripes[i].physical)
					continue;
				index_srcdev = i;
				found = 1;
				physical_of_found = bbio->stripes[i].physical;
			}
		}
		if (found) {
			u64 length = map->stripe_len;

			if (physical_of_found + length <=
			    dev_replace->cursor_left) {
				struct btrfs_bio_stripe *tgtdev_stripe =
					bbio->stripes + num_stripes;

				tgtdev_stripe->physical = physical_of_found;
				tgtdev_stripe->length =
					bbio->stripes[index_srcdev].length;
				tgtdev_stripe->dev = dev_replace->tgtdev;
5450
				bbio->tgtdev_map[index_srcdev] = num_stripes;
5451

5452
				tgtdev_indexes++;
5453 5454 5455
				num_stripes++;
			}
		}
5456 5457
	}

L
Li Zefan 已提交
5458
	*bbio_ret = bbio;
Z
Zhao Lei 已提交
5459
	bbio->map_type = map->type;
L
Li Zefan 已提交
5460 5461 5462
	bbio->num_stripes = num_stripes;
	bbio->max_errors = max_errors;
	bbio->mirror_num = mirror_num;
5463
	bbio->num_tgtdevs = tgtdev_indexes;
5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475

	/*
	 * this is the case that REQ_READ && dev_replace_is_ongoing &&
	 * mirror_num == num_stripes + 1 && dev_replace target drive is
	 * available as a mirror
	 */
	if (patch_the_first_stripe_for_dev_replace && num_stripes > 0) {
		WARN_ON(num_stripes > 1);
		bbio->stripes[0].dev = dev_replace->tgtdev;
		bbio->stripes[0].physical = physical_to_patch_in_first_stripe;
		bbio->mirror_num = map->num_stripes + 1;
	}
5476
out:
5477 5478
	if (dev_replace_is_ongoing)
		btrfs_dev_replace_unlock(dev_replace);
5479
	free_extent_map(em);
L
Li Zefan 已提交
5480
	return ret;
5481 5482
}

5483
int btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
5484
		      u64 logical, u64 *length,
5485
		      struct btrfs_bio **bbio_ret, int mirror_num)
5486
{
5487
	return __btrfs_map_block(fs_info, rw, logical, length, bbio_ret,
5488
				 mirror_num, 0);
5489 5490
}

5491 5492 5493 5494
/* For Scrub/replace */
int btrfs_map_sblock(struct btrfs_fs_info *fs_info, int rw,
		     u64 logical, u64 *length,
		     struct btrfs_bio **bbio_ret, int mirror_num,
5495
		     int need_raid_map)
5496 5497
{
	return __btrfs_map_block(fs_info, rw, logical, length, bbio_ret,
5498
				 mirror_num, need_raid_map);
5499 5500
}

Y
Yan Zheng 已提交
5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511
int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
		     u64 chunk_start, u64 physical, u64 devid,
		     u64 **logical, int *naddrs, int *stripe_len)
{
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	struct extent_map *em;
	struct map_lookup *map;
	u64 *buf;
	u64 bytenr;
	u64 length;
	u64 stripe_nr;
D
David Woodhouse 已提交
5512
	u64 rmap_len;
Y
Yan Zheng 已提交
5513 5514
	int i, j, nr = 0;

5515
	read_lock(&em_tree->lock);
Y
Yan Zheng 已提交
5516
	em = lookup_extent_mapping(em_tree, chunk_start, 1);
5517
	read_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
5518

5519
	if (!em) {
5520
		printk(KERN_ERR "BTRFS: couldn't find em for chunk %Lu\n",
5521 5522 5523 5524 5525
		       chunk_start);
		return -EIO;
	}

	if (em->start != chunk_start) {
5526
		printk(KERN_ERR "BTRFS: bad chunk start, em=%Lu, wanted=%Lu\n",
5527 5528 5529 5530
		       em->start, chunk_start);
		free_extent_map(em);
		return -EIO;
	}
Y
Yan Zheng 已提交
5531 5532 5533
	map = (struct map_lookup *)em->bdev;

	length = em->len;
D
David Woodhouse 已提交
5534 5535
	rmap_len = map->stripe_len;

Y
Yan Zheng 已提交
5536 5537 5538 5539
	if (map->type & BTRFS_BLOCK_GROUP_RAID10)
		do_div(length, map->num_stripes / map->sub_stripes);
	else if (map->type & BTRFS_BLOCK_GROUP_RAID0)
		do_div(length, map->num_stripes);
5540
	else if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
D
David Woodhouse 已提交
5541 5542 5543
		do_div(length, nr_data_stripes(map));
		rmap_len = map->stripe_len * nr_data_stripes(map);
	}
Y
Yan Zheng 已提交
5544

5545
	buf = kcalloc(map->num_stripes, sizeof(u64), GFP_NOFS);
5546
	BUG_ON(!buf); /* -ENOMEM */
Y
Yan Zheng 已提交
5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562

	for (i = 0; i < map->num_stripes; i++) {
		if (devid && map->stripes[i].dev->devid != devid)
			continue;
		if (map->stripes[i].physical > physical ||
		    map->stripes[i].physical + length <= physical)
			continue;

		stripe_nr = physical - map->stripes[i].physical;
		do_div(stripe_nr, map->stripe_len);

		if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
			stripe_nr = stripe_nr * map->num_stripes + i;
			do_div(stripe_nr, map->sub_stripes);
		} else if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
			stripe_nr = stripe_nr * map->num_stripes + i;
D
David Woodhouse 已提交
5563 5564 5565 5566 5567
		} /* else if RAID[56], multiply by nr_data_stripes().
		   * Alternatively, just use rmap_len below instead of
		   * map->stripe_len */

		bytenr = chunk_start + stripe_nr * rmap_len;
5568
		WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
5569 5570 5571 5572
		for (j = 0; j < nr; j++) {
			if (buf[j] == bytenr)
				break;
		}
5573 5574
		if (j == nr) {
			WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
5575
			buf[nr++] = bytenr;
5576
		}
Y
Yan Zheng 已提交
5577 5578 5579 5580
	}

	*logical = buf;
	*naddrs = nr;
D
David Woodhouse 已提交
5581
	*stripe_len = rmap_len;
Y
Yan Zheng 已提交
5582 5583 5584

	free_extent_map(em);
	return 0;
5585 5586
}

5587 5588 5589 5590 5591 5592
static inline void btrfs_end_bbio(struct btrfs_bio *bbio, struct bio *bio, int err)
{
	if (likely(bbio->flags & BTRFS_BIO_ORIG_BIO_SUBMITTED))
		bio_endio_nodec(bio, err);
	else
		bio_endio(bio, err);
5593
	btrfs_put_bbio(bbio);
5594 5595
}

5596
static void btrfs_end_bio(struct bio *bio, int err)
5597
{
5598
	struct btrfs_bio *bbio = bio->bi_private;
5599
	struct btrfs_device *dev = bbio->stripes[0].dev;
5600
	int is_orig_bio = 0;
5601

5602
	if (err) {
5603
		atomic_inc(&bbio->error);
5604 5605
		if (err == -EIO || err == -EREMOTEIO) {
			unsigned int stripe_index =
5606
				btrfs_io_bio(bio)->stripe_index;
5607 5608 5609

			BUG_ON(stripe_index >= bbio->num_stripes);
			dev = bbio->stripes[stripe_index].dev;
5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621
			if (dev->bdev) {
				if (bio->bi_rw & WRITE)
					btrfs_dev_stat_inc(dev,
						BTRFS_DEV_STAT_WRITE_ERRS);
				else
					btrfs_dev_stat_inc(dev,
						BTRFS_DEV_STAT_READ_ERRS);
				if ((bio->bi_rw & WRITE_FLUSH) == WRITE_FLUSH)
					btrfs_dev_stat_inc(dev,
						BTRFS_DEV_STAT_FLUSH_ERRS);
				btrfs_dev_stat_print_on_error(dev);
			}
5622 5623
		}
	}
5624

5625
	if (bio == bbio->orig_bio)
5626 5627
		is_orig_bio = 1;

5628 5629
	btrfs_bio_counter_dec(bbio->fs_info);

5630
	if (atomic_dec_and_test(&bbio->stripes_pending)) {
5631 5632
		if (!is_orig_bio) {
			bio_put(bio);
5633
			bio = bbio->orig_bio;
5634
		}
5635

5636 5637
		bio->bi_private = bbio->private;
		bio->bi_end_io = bbio->end_io;
5638
		btrfs_io_bio(bio)->mirror_num = bbio->mirror_num;
5639
		/* only send an error to the higher layers if it is
D
David Woodhouse 已提交
5640
		 * beyond the tolerance of the btrfs bio
5641
		 */
5642
		if (atomic_read(&bbio->error) > bbio->max_errors) {
5643
			err = -EIO;
5644
		} else {
5645 5646 5647 5648 5649
			/*
			 * this bio is actually up to date, we didn't
			 * go over the max number of errors
			 */
			set_bit(BIO_UPTODATE, &bio->bi_flags);
5650
			err = 0;
5651
		}
5652

5653
		btrfs_end_bbio(bbio, bio, err);
5654
	} else if (!is_orig_bio) {
5655 5656 5657 5658
		bio_put(bio);
	}
}

5659 5660 5661 5662 5663 5664 5665
/*
 * see run_scheduled_bios for a description of why bios are collected for
 * async submit.
 *
 * This will add one bio to the pending list for a device and make sure
 * the work struct is scheduled.
 */
5666 5667 5668
static noinline void btrfs_schedule_bio(struct btrfs_root *root,
					struct btrfs_device *device,
					int rw, struct bio *bio)
5669 5670
{
	int should_queue = 1;
5671
	struct btrfs_pending_bios *pending_bios;
5672

D
David Woodhouse 已提交
5673 5674 5675 5676 5677
	if (device->missing || !device->bdev) {
		bio_endio(bio, -EIO);
		return;
	}

5678
	/* don't bother with additional async steps for reads, right now */
5679
	if (!(rw & REQ_WRITE)) {
5680
		bio_get(bio);
5681
		btrfsic_submit_bio(rw, bio);
5682
		bio_put(bio);
5683
		return;
5684 5685 5686
	}

	/*
5687
	 * nr_async_bios allows us to reliably return congestion to the
5688 5689 5690 5691
	 * higher layers.  Otherwise, the async bio makes it appear we have
	 * made progress against dirty pages when we've really just put it
	 * on a queue for later
	 */
5692
	atomic_inc(&root->fs_info->nr_async_bios);
5693
	WARN_ON(bio->bi_next);
5694 5695 5696 5697
	bio->bi_next = NULL;
	bio->bi_rw |= rw;

	spin_lock(&device->io_lock);
5698
	if (bio->bi_rw & REQ_SYNC)
5699 5700 5701
		pending_bios = &device->pending_sync_bios;
	else
		pending_bios = &device->pending_bios;
5702

5703 5704
	if (pending_bios->tail)
		pending_bios->tail->bi_next = bio;
5705

5706 5707 5708
	pending_bios->tail = bio;
	if (!pending_bios->head)
		pending_bios->head = bio;
5709 5710 5711 5712 5713 5714
	if (device->running_pending)
		should_queue = 0;

	spin_unlock(&device->io_lock);

	if (should_queue)
5715 5716
		btrfs_queue_work(root->fs_info->submit_workers,
				 &device->work);
5717 5718
}

5719 5720 5721 5722 5723
static int bio_size_ok(struct block_device *bdev, struct bio *bio,
		       sector_t sector)
{
	struct bio_vec *prev;
	struct request_queue *q = bdev_get_queue(bdev);
5724
	unsigned int max_sectors = queue_max_sectors(q);
5725 5726 5727 5728 5729 5730
	struct bvec_merge_data bvm = {
		.bi_bdev = bdev,
		.bi_sector = sector,
		.bi_rw = bio->bi_rw,
	};

5731
	if (WARN_ON(bio->bi_vcnt == 0))
5732 5733 5734
		return 1;

	prev = &bio->bi_io_vec[bio->bi_vcnt - 1];
5735
	if (bio_sectors(bio) > max_sectors)
5736 5737 5738 5739 5740
		return 0;

	if (!q->merge_bvec_fn)
		return 1;

5741
	bvm.bi_size = bio->bi_iter.bi_size - prev->bv_len;
5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753
	if (q->merge_bvec_fn(q, &bvm, prev) < prev->bv_len)
		return 0;
	return 1;
}

static void submit_stripe_bio(struct btrfs_root *root, struct btrfs_bio *bbio,
			      struct bio *bio, u64 physical, int dev_nr,
			      int rw, int async)
{
	struct btrfs_device *dev = bbio->stripes[dev_nr].dev;

	bio->bi_private = bbio;
5754
	btrfs_io_bio(bio)->stripe_index = dev_nr;
5755
	bio->bi_end_io = btrfs_end_bio;
5756
	bio->bi_iter.bi_sector = physical >> 9;
5757 5758 5759 5760 5761 5762
#ifdef DEBUG
	{
		struct rcu_string *name;

		rcu_read_lock();
		name = rcu_dereference(dev->name);
M
Masanari Iida 已提交
5763
		pr_debug("btrfs_map_bio: rw %d, sector=%llu, dev=%lu "
5764
			 "(%s id %llu), size=%u\n", rw,
5765 5766
			 (u64)bio->bi_iter.bi_sector, (u_long)dev->bdev->bd_dev,
			 name->str, dev->devid, bio->bi_iter.bi_size);
5767 5768 5769 5770
		rcu_read_unlock();
	}
#endif
	bio->bi_bdev = dev->bdev;
5771 5772 5773

	btrfs_bio_counter_inc_noblocked(root->fs_info);

5774
	if (async)
D
David Woodhouse 已提交
5775
		btrfs_schedule_bio(root, dev, rw, bio);
5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796
	else
		btrfsic_submit_bio(rw, bio);
}

static int breakup_stripe_bio(struct btrfs_root *root, struct btrfs_bio *bbio,
			      struct bio *first_bio, struct btrfs_device *dev,
			      int dev_nr, int rw, int async)
{
	struct bio_vec *bvec = first_bio->bi_io_vec;
	struct bio *bio;
	int nr_vecs = bio_get_nr_vecs(dev->bdev);
	u64 physical = bbio->stripes[dev_nr].physical;

again:
	bio = btrfs_bio_alloc(dev->bdev, physical >> 9, nr_vecs, GFP_NOFS);
	if (!bio)
		return -ENOMEM;

	while (bvec <= (first_bio->bi_io_vec + first_bio->bi_vcnt - 1)) {
		if (bio_add_page(bio, bvec->bv_page, bvec->bv_len,
				 bvec->bv_offset) < bvec->bv_len) {
5797
			u64 len = bio->bi_iter.bi_size;
5798 5799 5800 5801 5802 5803 5804 5805 5806 5807 5808 5809 5810 5811 5812 5813 5814 5815

			atomic_inc(&bbio->stripes_pending);
			submit_stripe_bio(root, bbio, bio, physical, dev_nr,
					  rw, async);
			physical += len;
			goto again;
		}
		bvec++;
	}

	submit_stripe_bio(root, bbio, bio, physical, dev_nr, rw, async);
	return 0;
}

static void bbio_error(struct btrfs_bio *bbio, struct bio *bio, u64 logical)
{
	atomic_inc(&bbio->error);
	if (atomic_dec_and_test(&bbio->stripes_pending)) {
5816 5817 5818
		/* Shoud be the original bio. */
		WARN_ON(bio != bbio->orig_bio);

5819 5820
		bio->bi_private = bbio->private;
		bio->bi_end_io = bbio->end_io;
5821
		btrfs_io_bio(bio)->mirror_num = bbio->mirror_num;
5822
		bio->bi_iter.bi_sector = logical >> 9;
5823 5824

		btrfs_end_bbio(bbio, bio, -EIO);
5825 5826 5827
	}
}

5828
int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
5829
		  int mirror_num, int async_submit)
5830 5831
{
	struct btrfs_device *dev;
5832
	struct bio *first_bio = bio;
5833
	u64 logical = (u64)bio->bi_iter.bi_sector << 9;
5834 5835 5836
	u64 length = 0;
	u64 map_length;
	int ret;
5837 5838
	int dev_nr = 0;
	int total_devs = 1;
5839
	struct btrfs_bio *bbio = NULL;
5840

5841
	length = bio->bi_iter.bi_size;
5842
	map_length = length;
5843

5844
	btrfs_bio_counter_inc_blocked(root->fs_info);
D
David Woodhouse 已提交
5845
	ret = __btrfs_map_block(root->fs_info, rw, logical, &map_length, &bbio,
5846
			      mirror_num, 1);
5847 5848
	if (ret) {
		btrfs_bio_counter_dec(root->fs_info);
5849
		return ret;
5850
	}
5851

5852
	total_devs = bbio->num_stripes;
D
David Woodhouse 已提交
5853 5854 5855
	bbio->orig_bio = first_bio;
	bbio->private = first_bio->bi_private;
	bbio->end_io = first_bio->bi_end_io;
5856
	bbio->fs_info = root->fs_info;
D
David Woodhouse 已提交
5857 5858
	atomic_set(&bbio->stripes_pending, bbio->num_stripes);

5859
	if (bbio->raid_map) {
D
David Woodhouse 已提交
5860 5861 5862
		/* In this case, map_length has been set to the length of
		   a single stripe; not the whole write */
		if (rw & WRITE) {
5863
			ret = raid56_parity_write(root, bio, bbio, map_length);
D
David Woodhouse 已提交
5864
		} else {
5865
			ret = raid56_parity_recover(root, bio, bbio, map_length,
5866
						    mirror_num, 1);
D
David Woodhouse 已提交
5867
		}
5868

5869 5870
		btrfs_bio_counter_dec(root->fs_info);
		return ret;
D
David Woodhouse 已提交
5871 5872
	}

5873
	if (map_length < length) {
5874
		btrfs_crit(root->fs_info, "mapping failed logical %llu bio len %llu len %llu",
5875
			logical, length, map_length);
5876 5877
		BUG();
	}
5878

C
Chris Mason 已提交
5879
	while (dev_nr < total_devs) {
5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898 5899
		dev = bbio->stripes[dev_nr].dev;
		if (!dev || !dev->bdev || (rw & WRITE && !dev->writeable)) {
			bbio_error(bbio, first_bio, logical);
			dev_nr++;
			continue;
		}

		/*
		 * Check and see if we're ok with this bio based on it's size
		 * and offset with the given device.
		 */
		if (!bio_size_ok(dev->bdev, first_bio,
				 bbio->stripes[dev_nr].physical >> 9)) {
			ret = breakup_stripe_bio(root, bbio, first_bio, dev,
						 dev_nr, rw, async_submit);
			BUG_ON(ret);
			dev_nr++;
			continue;
		}

5900
		if (dev_nr < total_devs - 1) {
5901
			bio = btrfs_bio_clone(first_bio, GFP_NOFS);
5902
			BUG_ON(!bio); /* -ENOMEM */
5903 5904
		} else {
			bio = first_bio;
5905
			bbio->flags |= BTRFS_BIO_ORIG_BIO_SUBMITTED;
5906
		}
5907 5908 5909 5910

		submit_stripe_bio(root, bbio, bio,
				  bbio->stripes[dev_nr].physical, dev_nr, rw,
				  async_submit);
5911 5912
		dev_nr++;
	}
5913
	btrfs_bio_counter_dec(root->fs_info);
5914 5915 5916
	return 0;
}

5917
struct btrfs_device *btrfs_find_device(struct btrfs_fs_info *fs_info, u64 devid,
Y
Yan Zheng 已提交
5918
				       u8 *uuid, u8 *fsid)
5919
{
Y
Yan Zheng 已提交
5920 5921 5922
	struct btrfs_device *device;
	struct btrfs_fs_devices *cur_devices;

5923
	cur_devices = fs_info->fs_devices;
Y
Yan Zheng 已提交
5924 5925 5926 5927 5928 5929 5930 5931 5932 5933 5934
	while (cur_devices) {
		if (!fsid ||
		    !memcmp(cur_devices->fsid, fsid, BTRFS_UUID_SIZE)) {
			device = __find_device(&cur_devices->devices,
					       devid, uuid);
			if (device)
				return device;
		}
		cur_devices = cur_devices->seed;
	}
	return NULL;
5935 5936
}

5937
static struct btrfs_device *add_missing_dev(struct btrfs_root *root,
5938
					    struct btrfs_fs_devices *fs_devices,
5939 5940 5941 5942
					    u64 devid, u8 *dev_uuid)
{
	struct btrfs_device *device;

5943 5944
	device = btrfs_alloc_device(NULL, &devid, dev_uuid);
	if (IS_ERR(device))
5945
		return NULL;
5946 5947

	list_add(&device->dev_list, &fs_devices->devices);
Y
Yan Zheng 已提交
5948
	device->fs_devices = fs_devices;
5949
	fs_devices->num_devices++;
5950 5951

	device->missing = 1;
5952
	fs_devices->missing_devices++;
5953

5954 5955 5956
	return device;
}

5957 5958 5959 5960 5961 5962 5963 5964 5965 5966 5967 5968 5969 5970 5971 5972 5973 5974 5975 5976
/**
 * btrfs_alloc_device - allocate struct btrfs_device
 * @fs_info:	used only for generating a new devid, can be NULL if
 *		devid is provided (i.e. @devid != NULL).
 * @devid:	a pointer to devid for this device.  If NULL a new devid
 *		is generated.
 * @uuid:	a pointer to UUID for this device.  If NULL a new UUID
 *		is generated.
 *
 * Return: a pointer to a new &struct btrfs_device on success; ERR_PTR()
 * on error.  Returned struct is not linked onto any lists and can be
 * destroyed with kfree() right away.
 */
struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info,
					const u64 *devid,
					const u8 *uuid)
{
	struct btrfs_device *dev;
	u64 tmp;

5977
	if (WARN_ON(!devid && !fs_info))
5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988 5989 5990 5991 5992 5993 5994 5995 5996 5997 5998 5999 6000 6001
		return ERR_PTR(-EINVAL);

	dev = __alloc_device();
	if (IS_ERR(dev))
		return dev;

	if (devid)
		tmp = *devid;
	else {
		int ret;

		ret = find_next_devid(fs_info, &tmp);
		if (ret) {
			kfree(dev);
			return ERR_PTR(ret);
		}
	}
	dev->devid = tmp;

	if (uuid)
		memcpy(dev->uuid, uuid, BTRFS_UUID_SIZE);
	else
		generate_random_uuid(dev->uuid);

6002 6003
	btrfs_init_work(&dev->work, btrfs_submit_helper,
			pending_bios_fn, NULL, NULL);
6004 6005 6006 6007

	return dev;
}

6008 6009 6010 6011 6012 6013 6014 6015 6016 6017
static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key,
			  struct extent_buffer *leaf,
			  struct btrfs_chunk *chunk)
{
	struct btrfs_mapping_tree *map_tree = &root->fs_info->mapping_tree;
	struct map_lookup *map;
	struct extent_map *em;
	u64 logical;
	u64 length;
	u64 devid;
6018
	u8 uuid[BTRFS_UUID_SIZE];
6019
	int num_stripes;
6020
	int ret;
6021
	int i;
6022

6023 6024
	logical = key->offset;
	length = btrfs_chunk_length(leaf, chunk);
6025

6026
	read_lock(&map_tree->map_tree.lock);
6027
	em = lookup_extent_mapping(&map_tree->map_tree, logical, 1);
6028
	read_unlock(&map_tree->map_tree.lock);
6029 6030 6031 6032 6033 6034 6035 6036 6037

	/* already mapped? */
	if (em && em->start <= logical && em->start + em->len > logical) {
		free_extent_map(em);
		return 0;
	} else if (em) {
		free_extent_map(em);
	}

6038
	em = alloc_extent_map();
6039 6040
	if (!em)
		return -ENOMEM;
6041 6042
	num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
6043 6044 6045 6046 6047
	if (!map) {
		free_extent_map(em);
		return -ENOMEM;
	}

6048
	set_bit(EXTENT_FLAG_FS_MAPPING, &em->flags);
6049 6050 6051
	em->bdev = (struct block_device *)map;
	em->start = logical;
	em->len = length;
6052
	em->orig_start = 0;
6053
	em->block_start = 0;
C
Chris Mason 已提交
6054
	em->block_len = em->len;
6055

6056 6057 6058 6059 6060 6061
	map->num_stripes = num_stripes;
	map->io_width = btrfs_chunk_io_width(leaf, chunk);
	map->io_align = btrfs_chunk_io_align(leaf, chunk);
	map->sector_size = btrfs_chunk_sector_size(leaf, chunk);
	map->stripe_len = btrfs_chunk_stripe_len(leaf, chunk);
	map->type = btrfs_chunk_type(leaf, chunk);
C
Chris Mason 已提交
6062
	map->sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
6063 6064 6065 6066
	for (i = 0; i < num_stripes; i++) {
		map->stripes[i].physical =
			btrfs_stripe_offset_nr(leaf, chunk, i);
		devid = btrfs_stripe_devid_nr(leaf, chunk, i);
6067 6068 6069
		read_extent_buffer(leaf, uuid, (unsigned long)
				   btrfs_stripe_dev_uuid_nr(chunk, i),
				   BTRFS_UUID_SIZE);
6070 6071
		map->stripes[i].dev = btrfs_find_device(root->fs_info, devid,
							uuid, NULL);
6072
		if (!map->stripes[i].dev && !btrfs_test_opt(root, DEGRADED)) {
6073 6074 6075
			free_extent_map(em);
			return -EIO;
		}
6076 6077
		if (!map->stripes[i].dev) {
			map->stripes[i].dev =
6078 6079
				add_missing_dev(root, root->fs_info->fs_devices,
						devid, uuid);
6080 6081 6082 6083 6084 6085
			if (!map->stripes[i].dev) {
				free_extent_map(em);
				return -EIO;
			}
		}
		map->stripes[i].dev->in_fs_metadata = 1;
6086 6087
	}

6088
	write_lock(&map_tree->map_tree.lock);
J
Josef Bacik 已提交
6089
	ret = add_extent_mapping(&map_tree->map_tree, em, 0);
6090
	write_unlock(&map_tree->map_tree.lock);
6091
	BUG_ON(ret); /* Tree corruption */
6092 6093 6094 6095 6096
	free_extent_map(em);

	return 0;
}

6097
static void fill_device_from_item(struct extent_buffer *leaf,
6098 6099 6100 6101 6102 6103
				 struct btrfs_dev_item *dev_item,
				 struct btrfs_device *device)
{
	unsigned long ptr;

	device->devid = btrfs_device_id(leaf, dev_item);
6104 6105
	device->disk_total_bytes = btrfs_device_total_bytes(leaf, dev_item);
	device->total_bytes = device->disk_total_bytes;
6106
	device->commit_total_bytes = device->disk_total_bytes;
6107
	device->bytes_used = btrfs_device_bytes_used(leaf, dev_item);
6108
	device->commit_bytes_used = device->bytes_used;
6109 6110 6111 6112
	device->type = btrfs_device_type(leaf, dev_item);
	device->io_align = btrfs_device_io_align(leaf, dev_item);
	device->io_width = btrfs_device_io_width(leaf, dev_item);
	device->sector_size = btrfs_device_sector_size(leaf, dev_item);
6113
	WARN_ON(device->devid == BTRFS_DEV_REPLACE_DEVID);
6114
	device->is_tgtdev_for_dev_replace = 0;
6115

6116
	ptr = btrfs_device_uuid(dev_item);
6117
	read_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
6118 6119
}

6120 6121
static struct btrfs_fs_devices *open_seed_devices(struct btrfs_root *root,
						  u8 *fsid)
Y
Yan Zheng 已提交
6122 6123 6124 6125
{
	struct btrfs_fs_devices *fs_devices;
	int ret;

6126
	BUG_ON(!mutex_is_locked(&uuid_mutex));
Y
Yan Zheng 已提交
6127 6128 6129

	fs_devices = root->fs_info->fs_devices->seed;
	while (fs_devices) {
6130 6131 6132
		if (!memcmp(fs_devices->fsid, fsid, BTRFS_UUID_SIZE))
			return fs_devices;

Y
Yan Zheng 已提交
6133 6134 6135 6136 6137
		fs_devices = fs_devices->seed;
	}

	fs_devices = find_fsid(fsid);
	if (!fs_devices) {
6138 6139 6140 6141 6142 6143 6144 6145 6146 6147
		if (!btrfs_test_opt(root, DEGRADED))
			return ERR_PTR(-ENOENT);

		fs_devices = alloc_fs_devices(fsid);
		if (IS_ERR(fs_devices))
			return fs_devices;

		fs_devices->seeding = 1;
		fs_devices->opened = 1;
		return fs_devices;
Y
Yan Zheng 已提交
6148
	}
Y
Yan Zheng 已提交
6149 6150

	fs_devices = clone_fs_devices(fs_devices);
6151 6152
	if (IS_ERR(fs_devices))
		return fs_devices;
Y
Yan Zheng 已提交
6153

6154
	ret = __btrfs_open_devices(fs_devices, FMODE_READ,
6155
				   root->fs_info->bdev_holder);
6156 6157
	if (ret) {
		free_fs_devices(fs_devices);
6158
		fs_devices = ERR_PTR(ret);
Y
Yan Zheng 已提交
6159
		goto out;
6160
	}
Y
Yan Zheng 已提交
6161 6162 6163

	if (!fs_devices->seeding) {
		__btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
6164
		free_fs_devices(fs_devices);
6165
		fs_devices = ERR_PTR(-EINVAL);
Y
Yan Zheng 已提交
6166 6167 6168 6169 6170 6171
		goto out;
	}

	fs_devices->seed = root->fs_info->fs_devices->seed;
	root->fs_info->fs_devices->seed = fs_devices;
out:
6172
	return fs_devices;
Y
Yan Zheng 已提交
6173 6174
}

6175
static int read_one_dev(struct btrfs_root *root,
6176 6177 6178
			struct extent_buffer *leaf,
			struct btrfs_dev_item *dev_item)
{
6179
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
6180 6181 6182
	struct btrfs_device *device;
	u64 devid;
	int ret;
Y
Yan Zheng 已提交
6183
	u8 fs_uuid[BTRFS_UUID_SIZE];
6184 6185
	u8 dev_uuid[BTRFS_UUID_SIZE];

6186
	devid = btrfs_device_id(leaf, dev_item);
6187
	read_extent_buffer(leaf, dev_uuid, btrfs_device_uuid(dev_item),
6188
			   BTRFS_UUID_SIZE);
6189
	read_extent_buffer(leaf, fs_uuid, btrfs_device_fsid(dev_item),
Y
Yan Zheng 已提交
6190 6191 6192
			   BTRFS_UUID_SIZE);

	if (memcmp(fs_uuid, root->fs_info->fsid, BTRFS_UUID_SIZE)) {
6193 6194 6195
		fs_devices = open_seed_devices(root, fs_uuid);
		if (IS_ERR(fs_devices))
			return PTR_ERR(fs_devices);
Y
Yan Zheng 已提交
6196 6197
	}

6198
	device = btrfs_find_device(root->fs_info, devid, dev_uuid, fs_uuid);
6199
	if (!device) {
Y
Yan Zheng 已提交
6200
		if (!btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
6201 6202
			return -EIO;

6203 6204 6205 6206 6207 6208 6209 6210 6211
		btrfs_warn(root->fs_info, "devid %llu missing", devid);
		device = add_missing_dev(root, fs_devices, devid, dev_uuid);
		if (!device)
			return -ENOMEM;
	} else {
		if (!device->bdev && !btrfs_test_opt(root, DEGRADED))
			return -EIO;

		if(!device->bdev && !device->missing) {
6212 6213 6214 6215 6216 6217
			/*
			 * this happens when a device that was properly setup
			 * in the device info lists suddenly goes bad.
			 * device->bdev is NULL, and so we have to set
			 * device->missing to one here
			 */
6218
			device->fs_devices->missing_devices++;
6219
			device->missing = 1;
Y
Yan Zheng 已提交
6220
		}
6221 6222 6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234

		/* Move the device to its own fs_devices */
		if (device->fs_devices != fs_devices) {
			ASSERT(device->missing);

			list_move(&device->dev_list, &fs_devices->devices);
			device->fs_devices->num_devices--;
			fs_devices->num_devices++;

			device->fs_devices->missing_devices--;
			fs_devices->missing_devices++;

			device->fs_devices = fs_devices;
		}
Y
Yan Zheng 已提交
6235 6236 6237 6238 6239 6240 6241
	}

	if (device->fs_devices != root->fs_info->fs_devices) {
		BUG_ON(device->writeable);
		if (device->generation !=
		    btrfs_device_generation(leaf, dev_item))
			return -EINVAL;
6242
	}
6243 6244

	fill_device_from_item(leaf, dev_item, device);
6245
	device->in_fs_metadata = 1;
6246
	if (device->writeable && !device->is_tgtdev_for_dev_replace) {
Y
Yan Zheng 已提交
6247
		device->fs_devices->total_rw_bytes += device->total_bytes;
6248 6249 6250 6251 6252
		spin_lock(&root->fs_info->free_chunk_lock);
		root->fs_info->free_chunk_space += device->total_bytes -
			device->bytes_used;
		spin_unlock(&root->fs_info->free_chunk_lock);
	}
6253 6254 6255 6256
	ret = 0;
	return ret;
}

Y
Yan Zheng 已提交
6257
int btrfs_read_sys_array(struct btrfs_root *root)
6258
{
6259
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
6260
	struct extent_buffer *sb;
6261 6262
	struct btrfs_disk_key *disk_key;
	struct btrfs_chunk *chunk;
6263 6264
	u8 *array_ptr;
	unsigned long sb_array_offset;
6265
	int ret = 0;
6266 6267 6268
	u32 num_stripes;
	u32 array_size;
	u32 len = 0;
6269
	u32 cur_offset;
6270
	struct btrfs_key key;
6271

6272 6273 6274 6275 6276 6277 6278
	ASSERT(BTRFS_SUPER_INFO_SIZE <= root->nodesize);
	/*
	 * This will create extent buffer of nodesize, superblock size is
	 * fixed to BTRFS_SUPER_INFO_SIZE. If nodesize > sb size, this will
	 * overallocate but we can keep it as-is, only the first page is used.
	 */
	sb = btrfs_find_create_tree_block(root, BTRFS_SUPER_INFO_OFFSET);
6279 6280 6281
	if (!sb)
		return -ENOMEM;
	btrfs_set_buffer_uptodate(sb);
6282
	btrfs_set_buffer_lockdep_class(root->root_key.objectid, sb, 0);
6283 6284 6285 6286 6287 6288 6289 6290 6291 6292 6293 6294 6295
	/*
	 * The sb extent buffer is artifical and just used to read the system array.
	 * btrfs_set_buffer_uptodate() call does not properly mark all it's
	 * pages up-to-date when the page is larger: extent does not cover the
	 * whole page and consequently check_page_uptodate does not find all
	 * the page's extents up-to-date (the hole beyond sb),
	 * write_extent_buffer then triggers a WARN_ON.
	 *
	 * Regular short extents go through mark_extent_buffer_dirty/writeback cycle,
	 * but sb spans only this function. Add an explicit SetPageUptodate call
	 * to silence the warning eg. on PowerPC 64.
	 */
	if (PAGE_CACHE_SIZE > BTRFS_SUPER_INFO_SIZE)
6296
		SetPageUptodate(sb->pages[0]);
6297

6298
	write_extent_buffer(sb, super_copy, 0, BTRFS_SUPER_INFO_SIZE);
6299 6300
	array_size = btrfs_super_sys_array_size(super_copy);

6301 6302 6303
	array_ptr = super_copy->sys_chunk_array;
	sb_array_offset = offsetof(struct btrfs_super_block, sys_chunk_array);
	cur_offset = 0;
6304

6305 6306
	while (cur_offset < array_size) {
		disk_key = (struct btrfs_disk_key *)array_ptr;
6307 6308 6309 6310
		len = sizeof(*disk_key);
		if (cur_offset + len > array_size)
			goto out_short_read;

6311 6312
		btrfs_disk_key_to_cpu(&key, disk_key);

6313 6314 6315
		array_ptr += len;
		sb_array_offset += len;
		cur_offset += len;
6316

6317
		if (key.type == BTRFS_CHUNK_ITEM_KEY) {
6318
			chunk = (struct btrfs_chunk *)sb_array_offset;
6319 6320 6321 6322 6323 6324 6325 6326 6327 6328 6329 6330 6331
			/*
			 * At least one btrfs_chunk with one stripe must be
			 * present, exact stripe count check comes afterwards
			 */
			len = btrfs_chunk_item_size(1);
			if (cur_offset + len > array_size)
				goto out_short_read;

			num_stripes = btrfs_chunk_num_stripes(sb, chunk);
			len = btrfs_chunk_item_size(num_stripes);
			if (cur_offset + len > array_size)
				goto out_short_read;

6332
			ret = read_one_chunk(root, &key, sb, chunk);
6333 6334
			if (ret)
				break;
6335
		} else {
6336 6337
			ret = -EIO;
			break;
6338
		}
6339 6340 6341
		array_ptr += len;
		sb_array_offset += len;
		cur_offset += len;
6342
	}
6343
	free_extent_buffer(sb);
6344
	return ret;
6345 6346 6347 6348 6349 6350

out_short_read:
	printk(KERN_ERR "BTRFS: sys_array too short to read %u bytes at offset %u\n",
			len, cur_offset);
	free_extent_buffer(sb);
	return -EIO;
6351 6352 6353 6354 6355 6356 6357 6358 6359 6360 6361 6362 6363 6364 6365 6366 6367
}

int btrfs_read_chunk_tree(struct btrfs_root *root)
{
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_key key;
	struct btrfs_key found_key;
	int ret;
	int slot;

	root = root->fs_info->chunk_root;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

6368 6369 6370
	mutex_lock(&uuid_mutex);
	lock_chunks(root);

6371 6372 6373 6374 6375
	/*
	 * Read all device items, and then all the chunk items. All
	 * device items are found before any chunk item (their object id
	 * is smaller than the lowest possible object id for a chunk
	 * item - BTRFS_FIRST_CHUNK_TREE_OBJECTID).
6376 6377 6378 6379 6380
	 */
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.offset = 0;
	key.type = 0;
	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
6381 6382
	if (ret < 0)
		goto error;
C
Chris Mason 已提交
6383
	while (1) {
6384 6385 6386 6387 6388 6389 6390 6391 6392 6393 6394
		leaf = path->nodes[0];
		slot = path->slots[0];
		if (slot >= btrfs_header_nritems(leaf)) {
			ret = btrfs_next_leaf(root, path);
			if (ret == 0)
				continue;
			if (ret < 0)
				goto error;
			break;
		}
		btrfs_item_key_to_cpu(leaf, &found_key, slot);
6395 6396 6397
		if (found_key.type == BTRFS_DEV_ITEM_KEY) {
			struct btrfs_dev_item *dev_item;
			dev_item = btrfs_item_ptr(leaf, slot,
6398
						  struct btrfs_dev_item);
6399 6400 6401
			ret = read_one_dev(root, leaf, dev_item);
			if (ret)
				goto error;
6402 6403 6404 6405
		} else if (found_key.type == BTRFS_CHUNK_ITEM_KEY) {
			struct btrfs_chunk *chunk;
			chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);
			ret = read_one_chunk(root, &found_key, leaf, chunk);
Y
Yan Zheng 已提交
6406 6407
			if (ret)
				goto error;
6408 6409 6410 6411 6412
		}
		path->slots[0]++;
	}
	ret = 0;
error:
6413 6414 6415
	unlock_chunks(root);
	mutex_unlock(&uuid_mutex);

Y
Yan Zheng 已提交
6416
	btrfs_free_path(path);
6417 6418
	return ret;
}
6419

6420 6421 6422 6423 6424
void btrfs_init_devices_late(struct btrfs_fs_info *fs_info)
{
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
	struct btrfs_device *device;

6425 6426 6427 6428 6429 6430 6431 6432
	while (fs_devices) {
		mutex_lock(&fs_devices->device_list_mutex);
		list_for_each_entry(device, &fs_devices->devices, dev_list)
			device->dev_root = fs_info->dev_root;
		mutex_unlock(&fs_devices->device_list_mutex);

		fs_devices = fs_devices->seed;
	}
6433 6434
}

6435 6436 6437 6438 6439 6440 6441 6442 6443 6444 6445 6446 6447 6448 6449 6450 6451 6452 6453 6454 6455 6456 6457 6458 6459 6460 6461 6462 6463 6464 6465 6466 6467 6468 6469 6470 6471 6472 6473 6474 6475 6476 6477 6478 6479 6480 6481 6482 6483 6484 6485 6486 6487 6488 6489 6490 6491 6492 6493 6494 6495 6496 6497 6498 6499 6500 6501 6502 6503 6504 6505 6506 6507 6508 6509 6510 6511 6512 6513 6514 6515 6516 6517 6518 6519 6520 6521 6522
static void __btrfs_reset_dev_stats(struct btrfs_device *dev)
{
	int i;

	for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++)
		btrfs_dev_stat_reset(dev, i);
}

int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info)
{
	struct btrfs_key key;
	struct btrfs_key found_key;
	struct btrfs_root *dev_root = fs_info->dev_root;
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
	struct extent_buffer *eb;
	int slot;
	int ret = 0;
	struct btrfs_device *device;
	struct btrfs_path *path = NULL;
	int i;

	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto out;
	}

	mutex_lock(&fs_devices->device_list_mutex);
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
		int item_size;
		struct btrfs_dev_stats_item *ptr;

		key.objectid = 0;
		key.type = BTRFS_DEV_STATS_KEY;
		key.offset = device->devid;
		ret = btrfs_search_slot(NULL, dev_root, &key, path, 0, 0);
		if (ret) {
			__btrfs_reset_dev_stats(device);
			device->dev_stats_valid = 1;
			btrfs_release_path(path);
			continue;
		}
		slot = path->slots[0];
		eb = path->nodes[0];
		btrfs_item_key_to_cpu(eb, &found_key, slot);
		item_size = btrfs_item_size_nr(eb, slot);

		ptr = btrfs_item_ptr(eb, slot,
				     struct btrfs_dev_stats_item);

		for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) {
			if (item_size >= (1 + i) * sizeof(__le64))
				btrfs_dev_stat_set(device, i,
					btrfs_dev_stats_value(eb, ptr, i));
			else
				btrfs_dev_stat_reset(device, i);
		}

		device->dev_stats_valid = 1;
		btrfs_dev_stat_print_on_load(device);
		btrfs_release_path(path);
	}
	mutex_unlock(&fs_devices->device_list_mutex);

out:
	btrfs_free_path(path);
	return ret < 0 ? ret : 0;
}

static int update_dev_stat_item(struct btrfs_trans_handle *trans,
				struct btrfs_root *dev_root,
				struct btrfs_device *device)
{
	struct btrfs_path *path;
	struct btrfs_key key;
	struct extent_buffer *eb;
	struct btrfs_dev_stats_item *ptr;
	int ret;
	int i;

	key.objectid = 0;
	key.type = BTRFS_DEV_STATS_KEY;
	key.offset = device->devid;

	path = btrfs_alloc_path();
	BUG_ON(!path);
	ret = btrfs_search_slot(trans, dev_root, &key, path, -1, 1);
	if (ret < 0) {
6523 6524
		printk_in_rcu(KERN_WARNING "BTRFS: "
			"error %d while searching for dev_stats item for device %s!\n",
6525
			      ret, rcu_str_deref(device->name));
6526 6527 6528 6529 6530 6531 6532 6533
		goto out;
	}

	if (ret == 0 &&
	    btrfs_item_size_nr(path->nodes[0], path->slots[0]) < sizeof(*ptr)) {
		/* need to delete old one and insert a new one */
		ret = btrfs_del_item(trans, dev_root, path);
		if (ret != 0) {
6534 6535
			printk_in_rcu(KERN_WARNING "BTRFS: "
				"delete too small dev_stats item for device %s failed %d!\n",
6536
				      rcu_str_deref(device->name), ret);
6537 6538 6539 6540 6541 6542 6543 6544 6545 6546 6547
			goto out;
		}
		ret = 1;
	}

	if (ret == 1) {
		/* need to insert a new item */
		btrfs_release_path(path);
		ret = btrfs_insert_empty_item(trans, dev_root, path,
					      &key, sizeof(*ptr));
		if (ret < 0) {
6548 6549
			printk_in_rcu(KERN_WARNING "BTRFS: "
					  "insert dev_stats item for device %s failed %d!\n",
6550
				      rcu_str_deref(device->name), ret);
6551 6552 6553 6554 6555 6556 6557 6558 6559 6560 6561 6562 6563 6564 6565 6566 6567 6568 6569 6570 6571 6572 6573 6574 6575
			goto out;
		}
	}

	eb = path->nodes[0];
	ptr = btrfs_item_ptr(eb, path->slots[0], struct btrfs_dev_stats_item);
	for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++)
		btrfs_set_dev_stats_value(eb, ptr, i,
					  btrfs_dev_stat_read(device, i));
	btrfs_mark_buffer_dirty(eb);

out:
	btrfs_free_path(path);
	return ret;
}

/*
 * called from commit_transaction. Writes all changed device stats to disk.
 */
int btrfs_run_dev_stats(struct btrfs_trans_handle *trans,
			struct btrfs_fs_info *fs_info)
{
	struct btrfs_root *dev_root = fs_info->dev_root;
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
	struct btrfs_device *device;
6576
	int stats_cnt;
6577 6578 6579 6580
	int ret = 0;

	mutex_lock(&fs_devices->device_list_mutex);
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
6581
		if (!device->dev_stats_valid || !btrfs_dev_stats_dirty(device))
6582 6583
			continue;

6584
		stats_cnt = atomic_read(&device->dev_stats_ccnt);
6585 6586
		ret = update_dev_stat_item(trans, dev_root, device);
		if (!ret)
6587
			atomic_sub(stats_cnt, &device->dev_stats_ccnt);
6588 6589 6590 6591 6592 6593
	}
	mutex_unlock(&fs_devices->device_list_mutex);

	return ret;
}

6594 6595 6596 6597 6598 6599
void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index)
{
	btrfs_dev_stat_inc(dev, index);
	btrfs_dev_stat_print_on_error(dev);
}

6600
static void btrfs_dev_stat_print_on_error(struct btrfs_device *dev)
6601
{
6602 6603
	if (!dev->dev_stats_valid)
		return;
6604 6605
	printk_ratelimited_in_rcu(KERN_ERR "BTRFS: "
			   "bdev %s errs: wr %u, rd %u, flush %u, corrupt %u, gen %u\n",
6606
			   rcu_str_deref(dev->name),
6607 6608 6609
			   btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_WRITE_ERRS),
			   btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_READ_ERRS),
			   btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_FLUSH_ERRS),
6610 6611
			   btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_CORRUPTION_ERRS),
			   btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_GENERATION_ERRS));
6612
}
6613

6614 6615
static void btrfs_dev_stat_print_on_load(struct btrfs_device *dev)
{
6616 6617 6618 6619 6620 6621 6622 6623
	int i;

	for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++)
		if (btrfs_dev_stat_read(dev, i) != 0)
			break;
	if (i == BTRFS_DEV_STAT_VALUES_MAX)
		return; /* all values == 0, suppress message */

6624 6625
	printk_in_rcu(KERN_INFO "BTRFS: "
		   "bdev %s errs: wr %u, rd %u, flush %u, corrupt %u, gen %u\n",
6626
	       rcu_str_deref(dev->name),
6627 6628 6629 6630 6631 6632 6633
	       btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_WRITE_ERRS),
	       btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_READ_ERRS),
	       btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_FLUSH_ERRS),
	       btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_CORRUPTION_ERRS),
	       btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_GENERATION_ERRS));
}

6634
int btrfs_get_dev_stats(struct btrfs_root *root,
6635
			struct btrfs_ioctl_get_dev_stats *stats)
6636 6637 6638 6639 6640 6641
{
	struct btrfs_device *dev;
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
	int i;

	mutex_lock(&fs_devices->device_list_mutex);
6642
	dev = btrfs_find_device(root->fs_info, stats->devid, NULL, NULL);
6643 6644 6645
	mutex_unlock(&fs_devices->device_list_mutex);

	if (!dev) {
6646
		btrfs_warn(root->fs_info, "get dev_stats failed, device not found");
6647
		return -ENODEV;
6648
	} else if (!dev->dev_stats_valid) {
6649
		btrfs_warn(root->fs_info, "get dev_stats failed, not yet valid");
6650
		return -ENODEV;
6651
	} else if (stats->flags & BTRFS_DEV_STATS_RESET) {
6652 6653 6654 6655 6656 6657 6658 6659 6660 6661 6662 6663 6664 6665 6666 6667
		for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) {
			if (stats->nr_items > i)
				stats->values[i] =
					btrfs_dev_stat_read_and_reset(dev, i);
			else
				btrfs_dev_stat_reset(dev, i);
		}
	} else {
		for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++)
			if (stats->nr_items > i)
				stats->values[i] = btrfs_dev_stat_read(dev, i);
	}
	if (stats->nr_items > BTRFS_DEV_STAT_VALUES_MAX)
		stats->nr_items = BTRFS_DEV_STAT_VALUES_MAX;
	return 0;
}
6668 6669 6670 6671 6672 6673 6674 6675 6676 6677 6678 6679 6680 6681 6682 6683 6684 6685

int btrfs_scratch_superblock(struct btrfs_device *device)
{
	struct buffer_head *bh;
	struct btrfs_super_block *disk_super;

	bh = btrfs_read_dev_super(device->bdev);
	if (!bh)
		return -EINVAL;
	disk_super = (struct btrfs_super_block *)bh->b_data;

	memset(&disk_super->magic, 0, sizeof(disk_super->magic));
	set_buffer_dirty(bh);
	sync_dirty_buffer(bh);
	brelse(bh);

	return 0;
}
6686 6687 6688 6689 6690 6691 6692 6693 6694 6695 6696 6697 6698 6699 6700 6701 6702 6703 6704 6705 6706 6707 6708

/*
 * Update the size of all devices, which is used for writing out the
 * super blocks.
 */
void btrfs_update_commit_device_size(struct btrfs_fs_info *fs_info)
{
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
	struct btrfs_device *curr, *next;

	if (list_empty(&fs_devices->resized_devices))
		return;

	mutex_lock(&fs_devices->device_list_mutex);
	lock_chunks(fs_info->dev_root);
	list_for_each_entry_safe(curr, next, &fs_devices->resized_devices,
				 resized_list) {
		list_del_init(&curr->resized_list);
		curr->commit_total_bytes = curr->disk_total_bytes;
	}
	unlock_chunks(fs_info->dev_root);
	mutex_unlock(&fs_devices->device_list_mutex);
}
6709 6710 6711 6712 6713 6714 6715 6716 6717 6718 6719 6720 6721 6722 6723 6724 6725 6726 6727 6728 6729 6730 6731 6732 6733

/* Must be invoked during the transaction commit */
void btrfs_update_commit_device_bytes_used(struct btrfs_root *root,
					struct btrfs_transaction *transaction)
{
	struct extent_map *em;
	struct map_lookup *map;
	struct btrfs_device *dev;
	int i;

	if (list_empty(&transaction->pending_chunks))
		return;

	/* In order to kick the device replace finish process */
	lock_chunks(root);
	list_for_each_entry(em, &transaction->pending_chunks, list) {
		map = (struct map_lookup *)em->bdev;

		for (i = 0; i < map->num_stripes; i++) {
			dev = map->stripes[i].dev;
			dev->commit_bytes_used = dev->bytes_used;
		}
	}
	unlock_chunks(root);
}