volumes.c 91.9 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 <asm/div64.h>
C
Chris Mason 已提交
27
#include "compat.h"
28 29 30 31 32 33
#include "ctree.h"
#include "extent_map.h"
#include "disk-io.h"
#include "transaction.h"
#include "print-tree.h"
#include "volumes.h"
34
#include "async-thread.h"
35

Y
Yan Zheng 已提交
36 37 38 39 40
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);

41 42 43
static DEFINE_MUTEX(uuid_mutex);
static LIST_HEAD(fs_uuids);

44 45 46 47 48 49 50 51 52 53
static void lock_chunks(struct btrfs_root *root)
{
	mutex_lock(&root->fs_info->chunk_mutex);
}

static void unlock_chunks(struct btrfs_root *root)
{
	mutex_unlock(&root->fs_info->chunk_mutex);
}

Y
Yan Zheng 已提交
54 55 56 57 58 59 60 61 62 63 64 65 66 67
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);
		kfree(device->name);
		kfree(device);
	}
	kfree(fs_devices);
}

68 69 70 71
int btrfs_cleanup_fs_uuids(void)
{
	struct btrfs_fs_devices *fs_devices;

Y
Yan Zheng 已提交
72 73 74 75
	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
76
		free_fs_devices(fs_devices);
77 78 79 80
	}
	return 0;
}

81 82
static noinline struct btrfs_device *__find_device(struct list_head *head,
						   u64 devid, u8 *uuid)
83 84 85
{
	struct btrfs_device *dev;

Q
Qinghuang Feng 已提交
86
	list_for_each_entry(dev, head, dev_list) {
87
		if (dev->devid == devid &&
88
		    (!uuid || !memcmp(dev->uuid, uuid, BTRFS_UUID_SIZE))) {
89
			return dev;
90
		}
91 92 93 94
	}
	return NULL;
}

95
static noinline struct btrfs_fs_devices *find_fsid(u8 *fsid)
96 97 98
{
	struct btrfs_fs_devices *fs_devices;

Q
Qinghuang Feng 已提交
99
	list_for_each_entry(fs_devices, &fs_uuids, list) {
100 101 102 103 104 105
		if (memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE) == 0)
			return fs_devices;
	}
	return NULL;
}

106 107 108 109 110 111 112 113 114 115 116 117 118 119
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;
}

120 121 122 123 124 125 126 127 128 129 130
/*
 * 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.
 */
C
Chris Mason 已提交
131
static noinline int run_scheduled_bios(struct btrfs_device *device)
132 133 134
{
	struct bio *pending;
	struct backing_dev_info *bdi;
135
	struct btrfs_fs_info *fs_info;
136
	struct btrfs_pending_bios *pending_bios;
137 138 139
	struct bio *tail;
	struct bio *cur;
	int again = 0;
140
	unsigned long num_run;
141
	unsigned long batch_run = 0;
142
	unsigned long limit;
143
	unsigned long last_waited = 0;
144
	int force_reg = 0;
145 146 147 148 149 150 151 152 153
	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);
154

155
	bdi = blk_get_backing_dev_info(device->bdev);
156 157 158 159
	fs_info = device->dev_root->fs_info;
	limit = btrfs_async_submit_limit(fs_info);
	limit = limit * 2 / 3;

160 161 162
loop:
	spin_lock(&device->io_lock);

163
loop_lock:
164
	num_run = 0;
165

166 167 168 169 170
	/* 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
	 */
171
	if (!force_reg && device->pending_sync_bios.head) {
172
		pending_bios = &device->pending_sync_bios;
173 174
		force_reg = 1;
	} else {
175
		pending_bios = &device->pending_bios;
176 177
		force_reg = 0;
	}
178 179 180

	pending = pending_bios->head;
	tail = pending_bios->tail;
181 182 183 184 185 186 187 188 189 190
	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.
	 */
191 192
	if (device->pending_sync_bios.head == NULL &&
	    device->pending_bios.head == NULL) {
193 194
		again = 0;
		device->running_pending = 0;
195 196 197
	} else {
		again = 1;
		device->running_pending = 1;
198
	}
199 200 201 202

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

203 204
	spin_unlock(&device->io_lock);

C
Chris Mason 已提交
205
	while (pending) {
206 207

		rmb();
208 209 210 211 212 213 214 215
		/* 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)) {
216 217 218 219 220
			spin_lock(&device->io_lock);
			requeue_list(pending_bios, pending, tail);
			goto loop_lock;
		}

221 222 223
		cur = pending;
		pending = pending->bi_next;
		cur->bi_next = NULL;
224 225 226 227 228
		atomic_dec(&fs_info->nr_async_bios);

		if (atomic_read(&fs_info->nr_async_bios) < limit &&
		    waitqueue_active(&fs_info->async_submit_wait))
			wake_up(&fs_info->async_submit_wait);
229 230

		BUG_ON(atomic_read(&cur->bi_cnt) == 0);
231

232 233 234
		submit_bio(cur->bi_rw, cur);
		num_run++;
		batch_run++;
J
Jens Axboe 已提交
235
		if (need_resched())
236
			cond_resched();
237 238 239 240 241 242

		/*
		 * 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
243
		if (pending && bdi_write_congested(bdi) && batch_run > 8 &&
244
		    fs_info->fs_devices->open_devices > 1) {
245
			struct io_context *ioc;
246

247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268
			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;
J
Jens Axboe 已提交
269
				if (need_resched())
270
					cond_resched();
271 272
				continue;
			}
273
			spin_lock(&device->io_lock);
274
			requeue_list(pending_bios, pending, tail);
275
			device->running_pending = 1;
276 277 278 279 280 281

			spin_unlock(&device->io_lock);
			btrfs_requeue_work(&device->work);
			goto done;
		}
	}
282

283 284 285 286 287 288 289 290 291
	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);

292
done:
293
	blk_finish_plug(&plug);
294 295 296
	return 0;
}

297
static void pending_bios_fn(struct btrfs_work *work)
298 299 300 301 302 303 304
{
	struct btrfs_device *device;

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

305
static noinline int device_list_add(const char *path,
306 307 308 309 310 311
			   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;
	u64 found_transid = btrfs_super_generation(disk_super);
312
	char *name;
313 314 315

	fs_devices = find_fsid(disk_super->fsid);
	if (!fs_devices) {
316
		fs_devices = kzalloc(sizeof(*fs_devices), GFP_NOFS);
317 318 319
		if (!fs_devices)
			return -ENOMEM;
		INIT_LIST_HEAD(&fs_devices->devices);
320
		INIT_LIST_HEAD(&fs_devices->alloc_list);
321 322 323 324
		list_add(&fs_devices->list, &fs_uuids);
		memcpy(fs_devices->fsid, disk_super->fsid, BTRFS_FSID_SIZE);
		fs_devices->latest_devid = devid;
		fs_devices->latest_trans = found_transid;
325
		mutex_init(&fs_devices->device_list_mutex);
326 327
		device = NULL;
	} else {
328 329
		device = __find_device(&fs_devices->devices, devid,
				       disk_super->dev_item.uuid);
330 331
	}
	if (!device) {
Y
Yan Zheng 已提交
332 333 334
		if (fs_devices->opened)
			return -EBUSY;

335 336 337 338 339 340
		device = kzalloc(sizeof(*device), GFP_NOFS);
		if (!device) {
			/* we can safely leave the fs_devices entry around */
			return -ENOMEM;
		}
		device->devid = devid;
341
		device->work.func = pending_bios_fn;
342 343
		memcpy(device->uuid, disk_super->dev_item.uuid,
		       BTRFS_UUID_SIZE);
344
		spin_lock_init(&device->io_lock);
345 346 347 348 349
		device->name = kstrdup(path, GFP_NOFS);
		if (!device->name) {
			kfree(device);
			return -ENOMEM;
		}
Y
Yan Zheng 已提交
350
		INIT_LIST_HEAD(&device->dev_alloc_list);
351 352

		mutex_lock(&fs_devices->device_list_mutex);
353
		list_add_rcu(&device->dev_list, &fs_devices->devices);
354 355
		mutex_unlock(&fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
356
		device->fs_devices = fs_devices;
357
		fs_devices->num_devices++;
358
	} else if (!device->name || strcmp(device->name, path)) {
359 360 361 362 363
		name = kstrdup(path, GFP_NOFS);
		if (!name)
			return -ENOMEM;
		kfree(device->name);
		device->name = name;
364 365 366 367
		if (device->missing) {
			fs_devices->missing_devices--;
			device->missing = 0;
		}
368 369 370 371 372 373 374 375 376 377
	}

	if (found_transid > fs_devices->latest_trans) {
		fs_devices->latest_devid = devid;
		fs_devices->latest_trans = found_transid;
	}
	*fs_devices_ret = fs_devices;
	return 0;
}

Y
Yan Zheng 已提交
378 379 380 381 382 383 384 385 386 387 388 389 390
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;

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

	INIT_LIST_HEAD(&fs_devices->devices);
	INIT_LIST_HEAD(&fs_devices->alloc_list);
	INIT_LIST_HEAD(&fs_devices->list);
391
	mutex_init(&fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
392 393 394 395
	fs_devices->latest_devid = orig->latest_devid;
	fs_devices->latest_trans = orig->latest_trans;
	memcpy(fs_devices->fsid, orig->fsid, sizeof(fs_devices->fsid));

396
	/* We have held the volume lock, it is safe to get the devices. */
Y
Yan Zheng 已提交
397 398 399 400 401 402
	list_for_each_entry(orig_dev, &orig->devices, dev_list) {
		device = kzalloc(sizeof(*device), GFP_NOFS);
		if (!device)
			goto error;

		device->name = kstrdup(orig_dev->name, GFP_NOFS);
J
Julia Lawall 已提交
403 404
		if (!device->name) {
			kfree(device);
Y
Yan Zheng 已提交
405
			goto error;
J
Julia Lawall 已提交
406
		}
Y
Yan Zheng 已提交
407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424

		device->devid = orig_dev->devid;
		device->work.func = pending_bios_fn;
		memcpy(device->uuid, orig_dev->uuid, sizeof(device->uuid));
		spin_lock_init(&device->io_lock);
		INIT_LIST_HEAD(&device->dev_list);
		INIT_LIST_HEAD(&device->dev_alloc_list);

		list_add(&device->dev_list, &fs_devices->devices);
		device->fs_devices = fs_devices;
		fs_devices->num_devices++;
	}
	return fs_devices;
error:
	free_fs_devices(fs_devices);
	return ERR_PTR(-ENOMEM);
}

425 426
int btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices)
{
Q
Qinghuang Feng 已提交
427
	struct btrfs_device *device, *next;
428 429 430

	mutex_lock(&uuid_mutex);
again:
431
	/* This is the initialized path, it is safe to release the devices. */
Q
Qinghuang Feng 已提交
432
	list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
Y
Yan Zheng 已提交
433 434 435 436
		if (device->in_fs_metadata)
			continue;

		if (device->bdev) {
437
			blkdev_put(device->bdev, device->mode);
Y
Yan Zheng 已提交
438 439 440 441 442 443 444 445
			device->bdev = NULL;
			fs_devices->open_devices--;
		}
		if (device->writeable) {
			list_del_init(&device->dev_alloc_list);
			device->writeable = 0;
			fs_devices->rw_devices--;
		}
Y
Yan Zheng 已提交
446 447 448 449
		list_del_init(&device->dev_list);
		fs_devices->num_devices--;
		kfree(device->name);
		kfree(device);
450
	}
Y
Yan Zheng 已提交
451 452 453 454 455 456

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

457 458 459
	mutex_unlock(&uuid_mutex);
	return 0;
}
460

461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483
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);

	kfree(device->name);
	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
484
static int __btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
485 486
{
	struct btrfs_device *device;
Y
Yan Zheng 已提交
487

Y
Yan Zheng 已提交
488 489
	if (--fs_devices->opened > 0)
		return 0;
490

491
	mutex_lock(&fs_devices->device_list_mutex);
Q
Qinghuang Feng 已提交
492
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
493 494 495
		struct btrfs_device *new_device;

		if (device->bdev)
496
			fs_devices->open_devices--;
497

Y
Yan Zheng 已提交
498 499 500 501 502
		if (device->writeable) {
			list_del_init(&device->dev_alloc_list);
			fs_devices->rw_devices--;
		}

503 504 505 506 507 508 509 510 511 512 513
		new_device = kmalloc(sizeof(*new_device), GFP_NOFS);
		BUG_ON(!new_device);
		memcpy(new_device, device, sizeof(*new_device));
		new_device->name = kstrdup(device->name, GFP_NOFS);
		BUG_ON(!new_device->name);
		new_device->bdev = NULL;
		new_device->writeable = 0;
		new_device->in_fs_metadata = 0;
		list_replace_rcu(&device->dev_list, &new_device->dev_list);

		call_rcu(&device->rcu, free_device);
514
	}
515 516
	mutex_unlock(&fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
517 518
	WARN_ON(fs_devices->open_devices);
	WARN_ON(fs_devices->rw_devices);
Y
Yan Zheng 已提交
519 520 521
	fs_devices->opened = 0;
	fs_devices->seeding = 0;

522 523 524
	return 0;
}

Y
Yan Zheng 已提交
525 526
int btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
{
Y
Yan Zheng 已提交
527
	struct btrfs_fs_devices *seed_devices = NULL;
Y
Yan Zheng 已提交
528 529 530 531
	int ret;

	mutex_lock(&uuid_mutex);
	ret = __btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
532 533 534 535
	if (!fs_devices->opened) {
		seed_devices = fs_devices->seed;
		fs_devices->seed = NULL;
	}
Y
Yan Zheng 已提交
536
	mutex_unlock(&uuid_mutex);
Y
Yan Zheng 已提交
537 538 539 540 541 542 543

	while (seed_devices) {
		fs_devices = seed_devices;
		seed_devices = fs_devices->seed;
		__btrfs_close_devices(fs_devices);
		free_fs_devices(fs_devices);
	}
Y
Yan Zheng 已提交
544 545 546
	return ret;
}

Y
Yan Zheng 已提交
547 548
static int __btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
				fmode_t flags, void *holder)
549 550 551 552
{
	struct block_device *bdev;
	struct list_head *head = &fs_devices->devices;
	struct btrfs_device *device;
553 554 555 556 557 558
	struct block_device *latest_bdev = NULL;
	struct buffer_head *bh;
	struct btrfs_super_block *disk_super;
	u64 latest_devid = 0;
	u64 latest_transid = 0;
	u64 devid;
Y
Yan Zheng 已提交
559
	int seeding = 1;
560
	int ret = 0;
561

562 563
	flags |= FMODE_EXCL;

Q
Qinghuang Feng 已提交
564
	list_for_each_entry(device, head, dev_list) {
565 566
		if (device->bdev)
			continue;
567 568 569
		if (!device->name)
			continue;

570
		bdev = blkdev_get_by_path(device->name, flags, holder);
571
		if (IS_ERR(bdev)) {
C
Chris Mason 已提交
572
			printk(KERN_INFO "open %s failed\n", device->name);
573
			goto error;
574
		}
575
		set_blocksize(bdev, 4096);
576

Y
Yan Zheng 已提交
577
		bh = btrfs_read_dev_super(bdev);
578 579
		if (!bh) {
			ret = -EINVAL;
580
			goto error_close;
581
		}
582 583

		disk_super = (struct btrfs_super_block *)bh->b_data;
584
		devid = btrfs_stack_device_id(&disk_super->dev_item);
585 586 587
		if (devid != device->devid)
			goto error_brelse;

Y
Yan Zheng 已提交
588 589 590 591 592 593
		if (memcmp(device->uuid, disk_super->dev_item.uuid,
			   BTRFS_UUID_SIZE))
			goto error_brelse;

		device->generation = btrfs_super_generation(disk_super);
		if (!latest_transid || device->generation > latest_transid) {
594
			latest_devid = devid;
Y
Yan Zheng 已提交
595
			latest_transid = device->generation;
596 597 598
			latest_bdev = bdev;
		}

Y
Yan Zheng 已提交
599 600 601 602 603 604 605
		if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_SEEDING) {
			device->writeable = 0;
		} else {
			device->writeable = !bdev_read_only(bdev);
			seeding = 0;
		}

606
		device->bdev = bdev;
607
		device->in_fs_metadata = 0;
608 609
		device->mode = flags;

C
Chris Mason 已提交
610 611 612
		if (!blk_queue_nonrot(bdev_get_queue(bdev)))
			fs_devices->rotating = 1;

613
		fs_devices->open_devices++;
Y
Yan Zheng 已提交
614 615 616 617 618
		if (device->writeable) {
			fs_devices->rw_devices++;
			list_add(&device->dev_alloc_list,
				 &fs_devices->alloc_list);
		}
619
		brelse(bh);
620
		continue;
621

622 623 624
error_brelse:
		brelse(bh);
error_close:
625
		blkdev_put(bdev, flags);
626 627
error:
		continue;
628
	}
629 630 631 632
	if (fs_devices->open_devices == 0) {
		ret = -EIO;
		goto out;
	}
Y
Yan Zheng 已提交
633 634
	fs_devices->seeding = seeding;
	fs_devices->opened = 1;
635 636 637
	fs_devices->latest_bdev = latest_bdev;
	fs_devices->latest_devid = latest_devid;
	fs_devices->latest_trans = latest_transid;
Y
Yan Zheng 已提交
638
	fs_devices->total_rw_bytes = 0;
639
out:
Y
Yan Zheng 已提交
640 641 642 643
	return ret;
}

int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
644
		       fmode_t flags, void *holder)
Y
Yan Zheng 已提交
645 646 647 648 649
{
	int ret;

	mutex_lock(&uuid_mutex);
	if (fs_devices->opened) {
Y
Yan Zheng 已提交
650 651
		fs_devices->opened++;
		ret = 0;
Y
Yan Zheng 已提交
652
	} else {
653
		ret = __btrfs_open_devices(fs_devices, flags, holder);
Y
Yan Zheng 已提交
654
	}
655 656 657 658
	mutex_unlock(&uuid_mutex);
	return ret;
}

659
int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
660 661 662 663 664 665 666
			  struct btrfs_fs_devices **fs_devices_ret)
{
	struct btrfs_super_block *disk_super;
	struct block_device *bdev;
	struct buffer_head *bh;
	int ret;
	u64 devid;
667
	u64 transid;
668 669 670

	mutex_lock(&uuid_mutex);

671 672
	flags |= FMODE_EXCL;
	bdev = blkdev_get_by_path(path, flags, holder);
673 674 675 676 677 678 679 680 681

	if (IS_ERR(bdev)) {
		ret = PTR_ERR(bdev);
		goto error;
	}

	ret = set_blocksize(bdev, 4096);
	if (ret)
		goto error_close;
Y
Yan Zheng 已提交
682
	bh = btrfs_read_dev_super(bdev);
683
	if (!bh) {
684
		ret = -EINVAL;
685 686 687
		goto error_close;
	}
	disk_super = (struct btrfs_super_block *)bh->b_data;
688
	devid = btrfs_stack_device_id(&disk_super->dev_item);
689
	transid = btrfs_super_generation(disk_super);
690
	if (disk_super->label[0])
C
Chris Mason 已提交
691
		printk(KERN_INFO "device label %s ", disk_super->label);
692 693
	else {
		/* FIXME, make a readl uuid parser */
C
Chris Mason 已提交
694
		printk(KERN_INFO "device fsid %llx-%llx ",
695 696 697
		       *(unsigned long long *)disk_super->fsid,
		       *(unsigned long long *)(disk_super->fsid + 8));
	}
698
	printk(KERN_CONT "devid %llu transid %llu %s\n",
C
Chris Mason 已提交
699
	       (unsigned long long)devid, (unsigned long long)transid, path);
700 701 702 703
	ret = device_list_add(path, disk_super, devid, fs_devices_ret);

	brelse(bh);
error_close:
704
	blkdev_put(bdev, flags);
705 706 707 708
error:
	mutex_unlock(&uuid_mutex);
	return ret;
}
709

710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793
/* 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;

	if (start >= device->total_bytes)
		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;

		if (btrfs_key_type(&key) != BTRFS_DEV_EXTENT_KEY)
			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;
}

794
/*
795 796 797 798 799 800 801 802
 * find_free_dev_extent - find free space in the specified device
 * @trans:	transaction handler
 * @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
 *
803 804 805
 * 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
806 807 808 809 810 811 812 813
 *
 * @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.
814
 */
815 816
int find_free_dev_extent(struct btrfs_trans_handle *trans,
			 struct btrfs_device *device, u64 num_bytes,
817
			 u64 *start, u64 *len)
818 819 820
{
	struct btrfs_key key;
	struct btrfs_root *root = device->dev_root;
821
	struct btrfs_dev_extent *dev_extent;
Y
Yan Zheng 已提交
822
	struct btrfs_path *path;
823 824 825 826 827
	u64 hole_size;
	u64 max_hole_start;
	u64 max_hole_size;
	u64 extent_end;
	u64 search_start;
828 829
	u64 search_end = device->total_bytes;
	int ret;
830
	int slot;
831 832 833 834
	struct extent_buffer *l;

	/* FIXME use last free of some kind */

835 836 837
	/* 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
838
	search_start = max(root->fs_info->alloc_start, 1024ull * 1024);
839

840 841 842 843 844 845 846 847 848 849 850 851 852 853 854
	max_hole_start = search_start;
	max_hole_size = 0;

	if (search_start >= search_end) {
		ret = -ENOSPC;
		goto error;
	}

	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto error;
	}
	path->reada = 2;

855 856 857
	key.objectid = device->devid;
	key.offset = search_start;
	key.type = BTRFS_DEV_EXTENT_KEY;
858

859 860
	ret = btrfs_search_slot(trans, root, &key, path, 0, 0);
	if (ret < 0)
861
		goto out;
862 863 864
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid, key.type);
		if (ret < 0)
865
			goto out;
866
	}
867

868 869 870 871 872 873 874 875
	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)
876 877 878
				goto out;

			break;
879 880 881 882 883 884 885
		}
		btrfs_item_key_to_cpu(l, &key, slot);

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

		if (key.objectid > device->devid)
886
			break;
887

888 889
		if (btrfs_key_type(&key) != BTRFS_DEV_EXTENT_KEY)
			goto next;
890

891 892
		if (key.offset > search_start) {
			hole_size = key.offset - search_start;
893

894 895 896 897
			if (hole_size > max_hole_size) {
				max_hole_start = search_start;
				max_hole_size = hole_size;
			}
898

899 900 901 902 903 904 905 906 907 908 909 910
			/*
			 * 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;
911 912 913 914
			}
		}

		dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
915 916 917 918
		extent_end = key.offset + btrfs_dev_extent_length(l,
								  dev_extent);
		if (extent_end > search_start)
			search_start = extent_end;
919 920 921 922 923
next:
		path->slots[0]++;
		cond_resched();
	}

924 925 926 927
	hole_size = search_end- search_start;
	if (hole_size > max_hole_size) {
		max_hole_start = search_start;
		max_hole_size = hole_size;
928 929
	}

930 931 932 933 934 935 936
	/* See above. */
	if (hole_size < num_bytes)
		ret = -ENOSPC;
	else
		ret = 0;

out:
Y
Yan Zheng 已提交
937
	btrfs_free_path(path);
938 939
error:
	*start = max_hole_start;
940
	if (len)
941
		*len = max_hole_size;
942 943 944
	return ret;
}

945
static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans,
946 947 948 949 950 951 952
			  struct btrfs_device *device,
			  u64 start)
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_key key;
953 954 955
	struct btrfs_key found_key;
	struct extent_buffer *leaf = NULL;
	struct btrfs_dev_extent *extent = NULL;
956 957 958 959 960 961 962 963 964 965

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

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

	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
966 967 968
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid,
					  BTRFS_DEV_EXTENT_KEY);
969 970
		if (ret)
			goto out;
971 972 973 974 975 976 977 978 979 980 981
		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);
	} else if (ret == 0) {
		leaf = path->nodes[0];
		extent = btrfs_item_ptr(leaf, path->slots[0],
					struct btrfs_dev_extent);
	}
982 983
	BUG_ON(ret);

984 985
	if (device->bytes_used > 0)
		device->bytes_used -= btrfs_dev_extent_length(leaf, extent);
986 987
	ret = btrfs_del_item(trans, root, path);

988
out:
989 990 991 992
	btrfs_free_path(path);
	return ret;
}

Y
Yan Zheng 已提交
993
int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
994
			   struct btrfs_device *device,
995
			   u64 chunk_tree, u64 chunk_objectid,
Y
Yan Zheng 已提交
996
			   u64 chunk_offset, u64 start, u64 num_bytes)
997 998 999 1000 1001 1002 1003 1004
{
	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;

1005
	WARN_ON(!device->in_fs_metadata);
1006 1007 1008 1009 1010
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = device->devid;
Y
Yan Zheng 已提交
1011
	key.offset = start;
1012 1013 1014 1015 1016 1017 1018 1019
	key.type = BTRFS_DEV_EXTENT_KEY;
	ret = btrfs_insert_empty_item(trans, root, path, &key,
				      sizeof(*extent));
	BUG_ON(ret);

	leaf = path->nodes[0];
	extent = btrfs_item_ptr(leaf, path->slots[0],
				struct btrfs_dev_extent);
1020 1021 1022 1023 1024 1025 1026 1027
	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,
		    (unsigned long)btrfs_dev_extent_chunk_tree_uuid(extent),
		    BTRFS_UUID_SIZE);

1028 1029 1030 1031 1032 1033
	btrfs_set_dev_extent_length(leaf, extent, num_bytes);
	btrfs_mark_buffer_dirty(leaf);
	btrfs_free_path(path);
	return ret;
}

1034 1035
static noinline int find_next_chunk(struct btrfs_root *root,
				    u64 objectid, u64 *offset)
1036 1037 1038 1039
{
	struct btrfs_path *path;
	int ret;
	struct btrfs_key key;
1040
	struct btrfs_chunk *chunk;
1041 1042 1043 1044 1045
	struct btrfs_key found_key;

	path = btrfs_alloc_path();
	BUG_ON(!path);

1046
	key.objectid = objectid;
1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057
	key.offset = (u64)-1;
	key.type = BTRFS_CHUNK_ITEM_KEY;

	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (ret < 0)
		goto error;

	BUG_ON(ret == 0);

	ret = btrfs_previous_item(root, path, 0, BTRFS_CHUNK_ITEM_KEY);
	if (ret) {
1058
		*offset = 0;
1059 1060 1061
	} else {
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
1062 1063 1064 1065 1066 1067 1068 1069
		if (found_key.objectid != objectid)
			*offset = 0;
		else {
			chunk = btrfs_item_ptr(path->nodes[0], path->slots[0],
					       struct btrfs_chunk);
			*offset = found_key.offset +
				btrfs_chunk_length(path->nodes[0], chunk);
		}
1070 1071 1072 1073 1074 1075 1076
	}
	ret = 0;
error:
	btrfs_free_path(path);
	return ret;
}

Y
Yan Zheng 已提交
1077
static noinline int find_next_devid(struct btrfs_root *root, u64 *objectid)
1078 1079 1080 1081
{
	int ret;
	struct btrfs_key key;
	struct btrfs_key found_key;
Y
Yan Zheng 已提交
1082 1083 1084 1085 1086 1087 1088
	struct btrfs_path *path;

	root = root->fs_info->chunk_root;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110

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

	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (ret < 0)
		goto error;

	BUG_ON(ret == 0);

	ret = btrfs_previous_item(root, path, BTRFS_DEV_ITEMS_OBJECTID,
				  BTRFS_DEV_ITEM_KEY);
	if (ret) {
		*objectid = 1;
	} else {
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
		*objectid = found_key.offset + 1;
	}
	ret = 0;
error:
Y
Yan Zheng 已提交
1111
	btrfs_free_path(path);
1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137
	return ret;
}

/*
 * the device information is stored in the chunk root
 * the btrfs_device struct should be fully filled in
 */
int btrfs_add_device(struct btrfs_trans_handle *trans,
		     struct btrfs_root *root,
		     struct btrfs_device *device)
{
	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
1138
	key.offset = device->devid;
1139 1140

	ret = btrfs_insert_empty_item(trans, root, path, &key,
1141
				      sizeof(*dev_item));
1142 1143 1144 1145 1146 1147 1148
	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
1149
	btrfs_set_device_generation(leaf, dev_item, 0);
1150 1151 1152 1153 1154 1155
	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);
	btrfs_set_device_total_bytes(leaf, dev_item, device->total_bytes);
	btrfs_set_device_bytes_used(leaf, dev_item, device->bytes_used);
1156 1157 1158
	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);
1159
	btrfs_set_device_start_offset(leaf, dev_item, 0);
1160 1161

	ptr = (unsigned long)btrfs_device_uuid(dev_item);
1162
	write_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
1163 1164
	ptr = (unsigned long)btrfs_device_fsid(dev_item);
	write_extent_buffer(leaf, root->fs_info->fsid, ptr, BTRFS_UUID_SIZE);
1165 1166
	btrfs_mark_buffer_dirty(leaf);

Y
Yan Zheng 已提交
1167
	ret = 0;
1168 1169 1170 1171
out:
	btrfs_free_path(path);
	return ret;
}
1172

1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186
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;

1187
	trans = btrfs_start_transaction(root, 0);
1188 1189 1190 1191
	if (IS_ERR(trans)) {
		btrfs_free_path(path);
		return PTR_ERR(trans);
	}
1192 1193 1194
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.type = BTRFS_DEV_ITEM_KEY;
	key.offset = device->devid;
1195
	lock_chunks(root);
1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210

	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);
1211
	unlock_chunks(root);
1212 1213 1214 1215 1216 1217 1218
	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
1219
	struct btrfs_device *next_device;
1220
	struct block_device *bdev;
1221
	struct buffer_head *bh = NULL;
1222
	struct btrfs_super_block *disk_super;
1223
	struct btrfs_fs_devices *cur_devices;
1224 1225
	u64 all_avail;
	u64 devid;
Y
Yan Zheng 已提交
1226 1227
	u64 num_devices;
	u8 *dev_uuid;
1228
	int ret = 0;
1229
	bool clear_super = false;
1230 1231

	mutex_lock(&uuid_mutex);
1232
	mutex_lock(&root->fs_info->volume_mutex);
1233 1234 1235 1236 1237 1238

	all_avail = root->fs_info->avail_data_alloc_bits |
		root->fs_info->avail_system_alloc_bits |
		root->fs_info->avail_metadata_alloc_bits;

	if ((all_avail & BTRFS_BLOCK_GROUP_RAID10) &&
1239
	    root->fs_info->fs_devices->num_devices <= 4) {
C
Chris Mason 已提交
1240 1241
		printk(KERN_ERR "btrfs: unable to go below four devices "
		       "on raid10\n");
1242 1243 1244 1245 1246
		ret = -EINVAL;
		goto out;
	}

	if ((all_avail & BTRFS_BLOCK_GROUP_RAID1) &&
1247
	    root->fs_info->fs_devices->num_devices <= 2) {
C
Chris Mason 已提交
1248 1249
		printk(KERN_ERR "btrfs: unable to go below two "
		       "devices on raid1\n");
1250 1251 1252 1253
		ret = -EINVAL;
		goto out;
	}

1254 1255 1256
	if (strcmp(device_path, "missing") == 0) {
		struct list_head *devices;
		struct btrfs_device *tmp;
1257

1258 1259
		device = NULL;
		devices = &root->fs_info->fs_devices->devices;
1260 1261 1262 1263
		/*
		 * It is safe to read the devices since the volume_mutex
		 * is held.
		 */
Q
Qinghuang Feng 已提交
1264
		list_for_each_entry(tmp, devices, dev_list) {
1265 1266 1267 1268 1269 1270 1271 1272 1273
			if (tmp->in_fs_metadata && !tmp->bdev) {
				device = tmp;
				break;
			}
		}
		bdev = NULL;
		bh = NULL;
		disk_super = NULL;
		if (!device) {
C
Chris Mason 已提交
1274 1275
			printk(KERN_ERR "btrfs: no missing devices found to "
			       "remove\n");
1276 1277 1278
			goto out;
		}
	} else {
1279 1280
		bdev = blkdev_get_by_path(device_path, FMODE_READ | FMODE_EXCL,
					  root->fs_info->bdev_holder);
1281 1282 1283 1284
		if (IS_ERR(bdev)) {
			ret = PTR_ERR(bdev);
			goto out;
		}
1285

Y
Yan Zheng 已提交
1286
		set_blocksize(bdev, 4096);
Y
Yan Zheng 已提交
1287
		bh = btrfs_read_dev_super(bdev);
1288
		if (!bh) {
1289
			ret = -EINVAL;
1290 1291 1292
			goto error_close;
		}
		disk_super = (struct btrfs_super_block *)bh->b_data;
1293
		devid = btrfs_stack_device_id(&disk_super->dev_item);
Y
Yan Zheng 已提交
1294 1295 1296
		dev_uuid = disk_super->dev_item.uuid;
		device = btrfs_find_device(root, devid, dev_uuid,
					   disk_super->fsid);
1297 1298 1299 1300
		if (!device) {
			ret = -ENOENT;
			goto error_brelse;
		}
Y
Yan Zheng 已提交
1301
	}
1302

Y
Yan Zheng 已提交
1303
	if (device->writeable && root->fs_info->fs_devices->rw_devices == 1) {
C
Chris Mason 已提交
1304 1305
		printk(KERN_ERR "btrfs: unable to remove the only writeable "
		       "device\n");
Y
Yan Zheng 已提交
1306 1307 1308 1309 1310
		ret = -EINVAL;
		goto error_brelse;
	}

	if (device->writeable) {
1311
		lock_chunks(root);
Y
Yan Zheng 已提交
1312
		list_del_init(&device->dev_alloc_list);
1313
		unlock_chunks(root);
Y
Yan Zheng 已提交
1314
		root->fs_info->fs_devices->rw_devices--;
1315
		clear_super = true;
1316
	}
1317 1318 1319

	ret = btrfs_shrink_device(device, 0);
	if (ret)
1320
		goto error_undo;
1321 1322 1323

	ret = btrfs_rm_dev_item(root->fs_info->chunk_root, device);
	if (ret)
1324
		goto error_undo;
1325

Y
Yan Zheng 已提交
1326
	device->in_fs_metadata = 0;
A
Arne Jansen 已提交
1327
	btrfs_scrub_cancel_dev(root, device);
1328 1329 1330 1331 1332 1333

	/*
	 * the device list mutex makes sure that we don't change
	 * the device list while someone else is writing out all
	 * the device supers.
	 */
1334 1335

	cur_devices = device->fs_devices;
1336
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1337
	list_del_rcu(&device->dev_list);
1338

Y
Yan Zheng 已提交
1339
	device->fs_devices->num_devices--;
Y
Yan Zheng 已提交
1340

1341 1342 1343
	if (device->missing)
		root->fs_info->fs_devices->missing_devices--;

Y
Yan Zheng 已提交
1344 1345 1346 1347 1348 1349 1350
	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;

1351
	if (device->bdev)
Y
Yan Zheng 已提交
1352
		device->fs_devices->open_devices--;
1353 1354 1355

	call_rcu(&device->rcu, free_device);
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
1356

Y
Yan Zheng 已提交
1357 1358 1359
	num_devices = btrfs_super_num_devices(&root->fs_info->super_copy) - 1;
	btrfs_set_super_num_devices(&root->fs_info->super_copy, num_devices);

1360
	if (cur_devices->open_devices == 0) {
Y
Yan Zheng 已提交
1361 1362 1363
		struct btrfs_fs_devices *fs_devices;
		fs_devices = root->fs_info->fs_devices;
		while (fs_devices) {
1364
			if (fs_devices->seed == cur_devices)
Y
Yan Zheng 已提交
1365 1366
				break;
			fs_devices = fs_devices->seed;
Y
Yan Zheng 已提交
1367
		}
1368 1369
		fs_devices->seed = cur_devices->seed;
		cur_devices->seed = NULL;
1370
		lock_chunks(root);
1371
		__btrfs_close_devices(cur_devices);
1372
		unlock_chunks(root);
1373
		free_fs_devices(cur_devices);
Y
Yan Zheng 已提交
1374 1375 1376 1377 1378 1379
	}

	/*
	 * at this point, the device is zero sized.  We want to
	 * remove it from the devices list and zero out the old super
	 */
1380
	if (clear_super) {
1381 1382 1383 1384 1385 1386 1387
		/* 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);
	}
1388 1389 1390 1391 1392 1393

	ret = 0;

error_brelse:
	brelse(bh);
error_close:
1394
	if (bdev)
1395
		blkdev_put(bdev, FMODE_READ | FMODE_EXCL);
1396
out:
1397
	mutex_unlock(&root->fs_info->volume_mutex);
1398 1399
	mutex_unlock(&uuid_mutex);
	return ret;
1400 1401
error_undo:
	if (device->writeable) {
1402
		lock_chunks(root);
1403 1404
		list_add(&device->dev_alloc_list,
			 &root->fs_info->fs_devices->alloc_list);
1405
		unlock_chunks(root);
1406 1407 1408
		root->fs_info->fs_devices->rw_devices++;
	}
	goto error_brelse;
1409 1410
}

Y
Yan Zheng 已提交
1411 1412 1413 1414 1415 1416 1417 1418
/*
 * does all the dirty work required for changing file system's UUID.
 */
static int btrfs_prepare_sprout(struct btrfs_trans_handle *trans,
				struct btrfs_root *root)
{
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
	struct btrfs_fs_devices *old_devices;
Y
Yan Zheng 已提交
1419
	struct btrfs_fs_devices *seed_devices;
Y
Yan Zheng 已提交
1420 1421 1422 1423 1424
	struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
	struct btrfs_device *device;
	u64 super_flags;

	BUG_ON(!mutex_is_locked(&uuid_mutex));
Y
Yan Zheng 已提交
1425
	if (!fs_devices->seeding)
Y
Yan Zheng 已提交
1426 1427
		return -EINVAL;

Y
Yan Zheng 已提交
1428 1429
	seed_devices = kzalloc(sizeof(*fs_devices), GFP_NOFS);
	if (!seed_devices)
Y
Yan Zheng 已提交
1430 1431
		return -ENOMEM;

Y
Yan Zheng 已提交
1432 1433 1434 1435
	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
1436
	}
Y
Yan Zheng 已提交
1437

Y
Yan Zheng 已提交
1438 1439
	list_add(&old_devices->list, &fs_uuids);

Y
Yan Zheng 已提交
1440 1441 1442 1443
	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);
1444
	mutex_init(&seed_devices->device_list_mutex);
1445 1446

	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1447 1448
	list_splice_init_rcu(&fs_devices->devices, &seed_devices->devices,
			      synchronize_rcu);
1449 1450
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
1451 1452 1453 1454 1455
	list_splice_init(&fs_devices->alloc_list, &seed_devices->alloc_list);
	list_for_each_entry(device, &seed_devices->devices, dev_list) {
		device->fs_devices = seed_devices;
	}

Y
Yan Zheng 已提交
1456 1457 1458
	fs_devices->seeding = 0;
	fs_devices->num_devices = 0;
	fs_devices->open_devices = 0;
Y
Yan Zheng 已提交
1459
	fs_devices->seed = seed_devices;
Y
Yan Zheng 已提交
1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510

	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);
	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]);
1511
			btrfs_release_path(path);
Y
Yan Zheng 已提交
1512 1513 1514 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 1540 1541 1542 1543 1544 1545 1546
			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);
		read_extent_buffer(leaf, dev_uuid,
				   (unsigned long)btrfs_device_uuid(dev_item),
				   BTRFS_UUID_SIZE);
		read_extent_buffer(leaf, fs_uuid,
				   (unsigned long)btrfs_device_fsid(dev_item),
				   BTRFS_UUID_SIZE);
		device = btrfs_find_device(root, devid, dev_uuid, fs_uuid);
		BUG_ON(!device);

		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;
}

1547 1548 1549 1550 1551 1552
int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_device *device;
	struct block_device *bdev;
	struct list_head *devices;
Y
Yan Zheng 已提交
1553
	struct super_block *sb = root->fs_info->sb;
1554
	u64 total_bytes;
Y
Yan Zheng 已提交
1555
	int seeding_dev = 0;
1556 1557
	int ret = 0;

Y
Yan Zheng 已提交
1558 1559
	if ((sb->s_flags & MS_RDONLY) && !root->fs_info->fs_devices->seeding)
		return -EINVAL;
1560

1561 1562
	bdev = blkdev_get_by_path(device_path, FMODE_EXCL,
				  root->fs_info->bdev_holder);
1563 1564
	if (IS_ERR(bdev))
		return PTR_ERR(bdev);
1565

Y
Yan Zheng 已提交
1566 1567 1568 1569 1570 1571
	if (root->fs_info->fs_devices->seeding) {
		seeding_dev = 1;
		down_write(&sb->s_umount);
		mutex_lock(&uuid_mutex);
	}

1572
	filemap_write_and_wait(bdev->bd_inode->i_mapping);
1573
	mutex_lock(&root->fs_info->volume_mutex);
1574

1575
	devices = &root->fs_info->fs_devices->devices;
1576 1577 1578 1579
	/*
	 * we have the volume lock, so we don't need the extra
	 * device list mutex while reading the list here.
	 */
Q
Qinghuang Feng 已提交
1580
	list_for_each_entry(device, devices, dev_list) {
1581 1582
		if (device->bdev == bdev) {
			ret = -EEXIST;
Y
Yan Zheng 已提交
1583
			goto error;
1584 1585 1586 1587 1588 1589 1590
		}
	}

	device = kzalloc(sizeof(*device), GFP_NOFS);
	if (!device) {
		/* we can safely leave the fs_devices entry around */
		ret = -ENOMEM;
Y
Yan Zheng 已提交
1591
		goto error;
1592 1593 1594 1595 1596
	}

	device->name = kstrdup(device_path, GFP_NOFS);
	if (!device->name) {
		kfree(device);
Y
Yan Zheng 已提交
1597 1598
		ret = -ENOMEM;
		goto error;
1599
	}
Y
Yan Zheng 已提交
1600 1601 1602

	ret = find_next_devid(root, &device->devid);
	if (ret) {
1603
		kfree(device->name);
Y
Yan Zheng 已提交
1604 1605 1606 1607
		kfree(device);
		goto error;
	}

1608
	trans = btrfs_start_transaction(root, 0);
1609
	if (IS_ERR(trans)) {
1610
		kfree(device->name);
1611 1612 1613 1614 1615
		kfree(device);
		ret = PTR_ERR(trans);
		goto error;
	}

Y
Yan Zheng 已提交
1616 1617 1618 1619 1620 1621 1622
	lock_chunks(root);

	device->writeable = 1;
	device->work.func = pending_bios_fn;
	generate_random_uuid(device->uuid);
	spin_lock_init(&device->io_lock);
	device->generation = trans->transid;
1623 1624 1625 1626
	device->io_width = root->sectorsize;
	device->io_align = root->sectorsize;
	device->sector_size = root->sectorsize;
	device->total_bytes = i_size_read(bdev->bd_inode);
1627
	device->disk_total_bytes = device->total_bytes;
1628 1629
	device->dev_root = root->fs_info->dev_root;
	device->bdev = bdev;
1630
	device->in_fs_metadata = 1;
1631
	device->mode = FMODE_EXCL;
Y
Yan Zheng 已提交
1632
	set_blocksize(device->bdev, 4096);
1633

Y
Yan Zheng 已提交
1634 1635 1636 1637 1638
	if (seeding_dev) {
		sb->s_flags &= ~MS_RDONLY;
		ret = btrfs_prepare_sprout(trans, root);
		BUG_ON(ret);
	}
1639

Y
Yan Zheng 已提交
1640
	device->fs_devices = root->fs_info->fs_devices;
1641 1642 1643 1644 1645 1646

	/*
	 * we don't want write_supers to jump in here with our device
	 * half setup
	 */
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1647
	list_add_rcu(&device->dev_list, &root->fs_info->fs_devices->devices);
Y
Yan Zheng 已提交
1648 1649 1650 1651 1652 1653
	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++;
	root->fs_info->fs_devices->total_rw_bytes += device->total_bytes;
1654

C
Chris Mason 已提交
1655 1656 1657
	if (!blk_queue_nonrot(bdev_get_queue(bdev)))
		root->fs_info->fs_devices->rotating = 1;

1658 1659 1660 1661 1662 1663 1664
	total_bytes = btrfs_super_total_bytes(&root->fs_info->super_copy);
	btrfs_set_super_total_bytes(&root->fs_info->super_copy,
				    total_bytes + device->total_bytes);

	total_bytes = btrfs_super_num_devices(&root->fs_info->super_copy);
	btrfs_set_super_num_devices(&root->fs_info->super_copy,
				    total_bytes + 1);
1665
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
1666

Y
Yan Zheng 已提交
1667 1668 1669 1670 1671 1672 1673 1674 1675
	if (seeding_dev) {
		ret = init_first_rw_device(trans, root, device);
		BUG_ON(ret);
		ret = btrfs_finish_sprout(trans, root);
		BUG_ON(ret);
	} else {
		ret = btrfs_add_device(trans, root, device);
	}

1676 1677 1678 1679 1680 1681
	/*
	 * we've got more storage, clear any full flags on the space
	 * infos
	 */
	btrfs_clear_space_info_full(root->fs_info);

1682
	unlock_chunks(root);
Y
Yan Zheng 已提交
1683
	btrfs_commit_transaction(trans, root);
1684

Y
Yan Zheng 已提交
1685 1686 1687
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
1688

Y
Yan Zheng 已提交
1689 1690 1691 1692 1693 1694 1695
		ret = btrfs_relocate_sys_chunks(root);
		BUG_ON(ret);
	}
out:
	mutex_unlock(&root->fs_info->volume_mutex);
	return ret;
error:
1696
	blkdev_put(bdev, FMODE_EXCL);
Y
Yan Zheng 已提交
1697 1698 1699 1700
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
	}
1701 1702 1703
	goto out;
}

C
Chris Mason 已提交
1704 1705
static noinline int btrfs_update_device(struct btrfs_trans_handle *trans,
					struct btrfs_device *device)
1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740
{
	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);
1741
	btrfs_set_device_total_bytes(leaf, dev_item, device->disk_total_bytes);
1742 1743 1744 1745 1746 1747 1748 1749
	btrfs_set_device_bytes_used(leaf, dev_item, device->bytes_used);
	btrfs_mark_buffer_dirty(leaf);

out:
	btrfs_free_path(path);
	return ret;
}

1750
static int __btrfs_grow_device(struct btrfs_trans_handle *trans,
1751 1752 1753 1754 1755 1756 1757
		      struct btrfs_device *device, u64 new_size)
{
	struct btrfs_super_block *super_copy =
		&device->dev_root->fs_info->super_copy;
	u64 old_total = btrfs_super_total_bytes(super_copy);
	u64 diff = new_size - device->total_bytes;

Y
Yan Zheng 已提交
1758 1759 1760 1761 1762
	if (!device->writeable)
		return -EACCES;
	if (new_size <= device->total_bytes)
		return -EINVAL;

1763
	btrfs_set_super_total_bytes(super_copy, old_total + diff);
Y
Yan Zheng 已提交
1764 1765 1766
	device->fs_devices->total_rw_bytes += diff;

	device->total_bytes = new_size;
1767
	device->disk_total_bytes = new_size;
1768 1769
	btrfs_clear_space_info_full(device->dev_root->fs_info);

1770 1771 1772
	return btrfs_update_device(trans, device);
}

1773 1774 1775 1776 1777 1778 1779 1780 1781 1782
int btrfs_grow_device(struct btrfs_trans_handle *trans,
		      struct btrfs_device *device, u64 new_size)
{
	int ret;
	lock_chunks(device->dev_root);
	ret = __btrfs_grow_device(trans, device, new_size);
	unlock_chunks(device->dev_root);
	return ret;
}

1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806
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);
	BUG_ON(ret);

	ret = btrfs_del_item(trans, root, path);

	btrfs_free_path(path);
1807
	return ret;
1808 1809
}

1810
static int btrfs_del_sys_chunk(struct btrfs_root *root, u64 chunk_objectid, u64
1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855
			chunk_offset)
{
	struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
	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;

	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;
		}
	}
	return ret;
}

1856
static int btrfs_relocate_chunk(struct btrfs_root *root,
1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871
			 u64 chunk_tree, u64 chunk_objectid,
			 u64 chunk_offset)
{
	struct extent_map_tree *em_tree;
	struct btrfs_root *extent_root;
	struct btrfs_trans_handle *trans;
	struct extent_map *em;
	struct map_lookup *map;
	int ret;
	int i;

	root = root->fs_info->chunk_root;
	extent_root = root->fs_info->extent_root;
	em_tree = &root->fs_info->mapping_tree.map_tree;

1872 1873 1874 1875
	ret = btrfs_can_relocate(extent_root, chunk_offset);
	if (ret)
		return -ENOSPC;

1876
	/* step one, relocate all the extents inside this chunk */
Z
Zheng Yan 已提交
1877
	ret = btrfs_relocate_block_group(extent_root, chunk_offset);
1878 1879
	if (ret)
		return ret;
1880

1881
	trans = btrfs_start_transaction(root, 0);
1882
	BUG_ON(IS_ERR(trans));
1883

1884 1885
	lock_chunks(root);

1886 1887 1888 1889
	/*
	 * step two, delete the device extents and the
	 * chunk tree entries
	 */
1890
	read_lock(&em_tree->lock);
1891
	em = lookup_extent_mapping(em_tree, chunk_offset, 1);
1892
	read_unlock(&em_tree->lock);
1893

1894 1895
	BUG_ON(em->start > chunk_offset ||
	       em->start + em->len < chunk_offset);
1896 1897 1898 1899 1900 1901
	map = (struct map_lookup *)em->bdev;

	for (i = 0; i < map->num_stripes; i++) {
		ret = btrfs_free_dev_extent(trans, map->stripes[i].dev,
					    map->stripes[i].physical);
		BUG_ON(ret);
1902

1903 1904 1905 1906
		if (map->stripes[i].dev) {
			ret = btrfs_update_device(trans, map->stripes[i].dev);
			BUG_ON(ret);
		}
1907 1908 1909 1910 1911 1912
	}
	ret = btrfs_free_chunk(trans, root, chunk_tree, chunk_objectid,
			       chunk_offset);

	BUG_ON(ret);

1913 1914
	trace_btrfs_chunk_free(root, map, chunk_offset, em->len);

1915 1916 1917 1918 1919
	if (map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
		ret = btrfs_del_sys_chunk(root, chunk_objectid, chunk_offset);
		BUG_ON(ret);
	}

Y
Yan Zheng 已提交
1920 1921 1922
	ret = btrfs_remove_block_group(trans, extent_root, chunk_offset);
	BUG_ON(ret);

1923
	write_lock(&em_tree->lock);
Y
Yan Zheng 已提交
1924
	remove_extent_mapping(em_tree, em);
1925
	write_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949

	kfree(map);
	em->bdev = NULL;

	/* once for the tree */
	free_extent_map(em);
	/* once for us */
	free_extent_map(em);

	unlock_chunks(root);
	btrfs_end_transaction(trans, root);
	return 0;
}

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;
1950 1951
	bool retried = false;
	int failed = 0;
Y
Yan Zheng 已提交
1952 1953 1954 1955 1956 1957
	int ret;

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

1958
again:
Y
Yan Zheng 已提交
1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974
	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;
		BUG_ON(ret == 0);

		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
1975

Y
Yan Zheng 已提交
1976 1977
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
Z
Zheng Yan 已提交
1978

Y
Yan Zheng 已提交
1979 1980 1981
		chunk = btrfs_item_ptr(leaf, path->slots[0],
				       struct btrfs_chunk);
		chunk_type = btrfs_chunk_type(leaf, chunk);
1982
		btrfs_release_path(path);
1983

Y
Yan Zheng 已提交
1984 1985 1986 1987
		if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) {
			ret = btrfs_relocate_chunk(chunk_root, chunk_tree,
						   found_key.objectid,
						   found_key.offset);
1988 1989 1990 1991
			if (ret == -ENOSPC)
				failed++;
			else if (ret)
				BUG();
Y
Yan Zheng 已提交
1992
		}
1993

Y
Yan Zheng 已提交
1994 1995 1996 1997 1998
		if (found_key.offset == 0)
			break;
		key.offset = found_key.offset - 1;
	}
	ret = 0;
1999 2000 2001 2002 2003 2004 2005 2006
	if (failed && !retried) {
		failed = 0;
		retried = true;
		goto again;
	} else if (failed && retried) {
		WARN_ON(1);
		ret = -ENOSPC;
	}
Y
Yan Zheng 已提交
2007 2008 2009
error:
	btrfs_free_path(path);
	return ret;
2010 2011
}

2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033
static u64 div_factor(u64 num, int factor)
{
	if (factor == 10)
		return num;
	num *= factor;
	do_div(num, 10);
	return num;
}

int btrfs_balance(struct btrfs_root *dev_root)
{
	int ret;
	struct list_head *devices = &dev_root->fs_info->fs_devices->devices;
	struct btrfs_device *device;
	u64 old_size;
	u64 size_to_free;
	struct btrfs_path *path;
	struct btrfs_key key;
	struct btrfs_root *chunk_root = dev_root->fs_info->chunk_root;
	struct btrfs_trans_handle *trans;
	struct btrfs_key found_key;

Y
Yan Zheng 已提交
2034 2035
	if (dev_root->fs_info->sb->s_flags & MS_RDONLY)
		return -EROFS;
2036

2037 2038 2039
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

2040
	mutex_lock(&dev_root->fs_info->volume_mutex);
2041 2042 2043
	dev_root = dev_root->fs_info->dev_root;

	/* step one make some room on all the devices */
Q
Qinghuang Feng 已提交
2044
	list_for_each_entry(device, devices, dev_list) {
2045 2046 2047
		old_size = device->total_bytes;
		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
2048 2049
		if (!device->writeable ||
		    device->total_bytes - device->bytes_used > size_to_free)
2050 2051 2052
			continue;

		ret = btrfs_shrink_device(device, old_size - size_to_free);
2053 2054
		if (ret == -ENOSPC)
			break;
2055 2056
		BUG_ON(ret);

2057
		trans = btrfs_start_transaction(dev_root, 0);
2058
		BUG_ON(IS_ERR(trans));
2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073

		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();
	BUG_ON(!path);

	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.offset = (u64)-1;
	key.type = BTRFS_CHUNK_ITEM_KEY;

C
Chris Mason 已提交
2074
	while (1) {
2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087
		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)
			break;

		ret = btrfs_previous_item(chunk_root, path, 0,
					  BTRFS_CHUNK_ITEM_KEY);
2088
		if (ret)
2089
			break;
2090

2091 2092 2093 2094
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
		if (found_key.objectid != key.objectid)
			break;
2095

2096
		/* chunk zero is special */
2097
		if (found_key.offset == 0)
2098 2099
			break;

2100
		btrfs_release_path(path);
2101 2102 2103 2104
		ret = btrfs_relocate_chunk(chunk_root,
					   chunk_root->root_key.objectid,
					   found_key.objectid,
					   found_key.offset);
2105 2106
		BUG_ON(ret && ret != -ENOSPC);
		key.offset = found_key.offset - 1;
2107 2108 2109 2110
	}
	ret = 0;
error:
	btrfs_free_path(path);
2111
	mutex_unlock(&dev_root->fs_info->volume_mutex);
2112 2113 2114
	return ret;
}

2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131
/*
 * 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;
2132 2133
	int failed = 0;
	bool retried = false;
2134 2135 2136 2137
	struct extent_buffer *l;
	struct btrfs_key key;
	struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
	u64 old_total = btrfs_super_total_bytes(super_copy);
2138
	u64 old_size = device->total_bytes;
2139 2140
	u64 diff = device->total_bytes - new_size;

Y
Yan Zheng 已提交
2141 2142
	if (new_size >= device->total_bytes)
		return -EINVAL;
2143 2144 2145 2146 2147 2148 2149

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

	path->reada = 2;

2150 2151
	lock_chunks(root);

2152
	device->total_bytes = new_size;
Y
Yan Zheng 已提交
2153 2154
	if (device->writeable)
		device->fs_devices->total_rw_bytes -= diff;
2155
	unlock_chunks(root);
2156

2157
again:
2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171
	key.objectid = device->devid;
	key.offset = (u64)-1;
	key.type = BTRFS_DEV_EXTENT_KEY;

	while (1) {
		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;
2172
			btrfs_release_path(path);
2173
			break;
2174 2175 2176 2177 2178 2179
		}

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

2180
		if (key.objectid != device->devid) {
2181
			btrfs_release_path(path);
2182
			break;
2183
		}
2184 2185 2186 2187

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

2188
		if (key.offset + length <= new_size) {
2189
			btrfs_release_path(path);
2190
			break;
2191
		}
2192 2193 2194 2195

		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);
2196
		btrfs_release_path(path);
2197 2198 2199

		ret = btrfs_relocate_chunk(root, chunk_tree, chunk_objectid,
					   chunk_offset);
2200
		if (ret && ret != -ENOSPC)
2201
			goto done;
2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219
		if (ret == -ENOSPC)
			failed++;
		key.offset -= 1;
	}

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

		device->total_bytes = old_size;
		if (device->writeable)
			device->fs_devices->total_rw_bytes += diff;
		unlock_chunks(root);
		goto done;
2220 2221
	}

2222
	/* Shrinking succeeded, else we would be at "done". */
2223
	trans = btrfs_start_transaction(root, 0);
2224 2225 2226 2227 2228
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto done;
	}

2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242
	lock_chunks(root);

	device->disk_total_bytes = new_size;
	/* Now btrfs_update_device() will change the on-disk size. */
	ret = btrfs_update_device(trans, device);
	if (ret) {
		unlock_chunks(root);
		btrfs_end_transaction(trans, root);
		goto done;
	}
	WARN_ON(diff > old_total);
	btrfs_set_super_total_bytes(super_copy, old_total - diff);
	unlock_chunks(root);
	btrfs_end_transaction(trans, root);
2243 2244 2245 2246 2247
done:
	btrfs_free_path(path);
	return ret;
}

2248
static int btrfs_add_system_chunk(struct btrfs_trans_handle *trans,
2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271
			   struct btrfs_root *root,
			   struct btrfs_key *key,
			   struct btrfs_chunk *chunk, int item_size)
{
	struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
	struct btrfs_disk_key disk_key;
	u32 array_size;
	u8 *ptr;

	array_size = btrfs_super_sys_array_size(super_copy);
	if (array_size + item_size > BTRFS_SYSTEM_CHUNK_ARRAY_SIZE)
		return -EFBIG;

	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);
	return 0;
}

2272 2273 2274 2275
/*
 * sort the devices in descending order by max_avail, total_avail
 */
static int btrfs_cmp_device_info(const void *a, const void *b)
2276
{
2277 2278
	const struct btrfs_device_info *di_a = a;
	const struct btrfs_device_info *di_b = b;
2279

2280
	if (di_a->max_avail > di_b->max_avail)
2281
		return -1;
2282
	if (di_a->max_avail < di_b->max_avail)
2283
		return 1;
2284 2285 2286 2287 2288
	if (di_a->total_avail > di_b->total_avail)
		return -1;
	if (di_a->total_avail < di_b->total_avail)
		return 1;
	return 0;
2289
}
2290

2291 2292 2293 2294 2295
static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
			       struct btrfs_root *extent_root,
			       struct map_lookup **map_ret,
			       u64 *num_bytes_out, u64 *stripe_size_out,
			       u64 start, u64 type)
2296
{
2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319
	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 */
	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;
	int ndevs;
	int i;
	int j;
2320

2321 2322 2323 2324
	if ((type & BTRFS_BLOCK_GROUP_RAID1) &&
	    (type & BTRFS_BLOCK_GROUP_DUP)) {
		WARN_ON(1);
		type &= ~BTRFS_BLOCK_GROUP_DUP;
C
Chris Mason 已提交
2325
	}
2326

2327 2328
	if (list_empty(&fs_devices->alloc_list))
		return -ENOSPC;
2329

2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343
	sub_stripes = 1;
	dev_stripes = 1;
	devs_increment = 1;
	ncopies = 1;
	devs_max = 0;	/* 0 == as many as possible */
	devs_min = 1;

	/*
	 * define the properties of each RAID type.
	 * FIXME: move this to a global table and use it in all RAID
	 * calculation code
	 */
	if (type & (BTRFS_BLOCK_GROUP_DUP)) {
		dev_stripes = 2;
2344
		ncopies = 2;
2345 2346 2347 2348 2349
		devs_max = 1;
	} else if (type & (BTRFS_BLOCK_GROUP_RAID0)) {
		devs_min = 2;
	} else if (type & (BTRFS_BLOCK_GROUP_RAID1)) {
		devs_increment = 2;
2350
		ncopies = 2;
2351 2352 2353 2354 2355 2356 2357 2358 2359 2360
		devs_max = 2;
		devs_min = 2;
	} else if (type & (BTRFS_BLOCK_GROUP_RAID10)) {
		sub_stripes = 2;
		devs_increment = 2;
		ncopies = 2;
		devs_min = 4;
	} else {
		devs_max = 1;
	}
2361

2362
	if (type & BTRFS_BLOCK_GROUP_DATA) {
2363 2364
		max_stripe_size = 1024 * 1024 * 1024;
		max_chunk_size = 10 * max_stripe_size;
2365
	} else if (type & BTRFS_BLOCK_GROUP_METADATA) {
2366 2367
		max_stripe_size = 256 * 1024 * 1024;
		max_chunk_size = max_stripe_size;
2368
	} else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
2369 2370 2371 2372 2373 2374
		max_stripe_size = 8 * 1024 * 1024;
		max_chunk_size = 2 * max_stripe_size;
	} else {
		printk(KERN_ERR "btrfs: invalid chunk type 0x%llx requested\n",
		       type);
		BUG_ON(1);
2375 2376
	}

Y
Yan Zheng 已提交
2377 2378 2379
	/* 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);
2380

2381 2382 2383 2384
	devices_info = kzalloc(sizeof(*devices_info) * fs_devices->rw_devices,
			       GFP_NOFS);
	if (!devices_info)
		return -ENOMEM;
2385

2386
	cur = fs_devices->alloc_list.next;
2387

2388
	/*
2389 2390
	 * in the first pass through the devices list, we gather information
	 * about the available holes on each device.
2391
	 */
2392 2393 2394 2395 2396
	ndevs = 0;
	while (cur != &fs_devices->alloc_list) {
		struct btrfs_device *device;
		u64 max_avail;
		u64 dev_offset;
2397

2398
		device = list_entry(cur, struct btrfs_device, dev_alloc_list);
2399

2400
		cur = cur->next;
2401

2402 2403 2404 2405 2406 2407
		if (!device->writeable) {
			printk(KERN_ERR
			       "btrfs: read-only device in alloc_list\n");
			WARN_ON(1);
			continue;
		}
2408

2409 2410
		if (!device->in_fs_metadata)
			continue;
2411

2412 2413 2414 2415 2416 2417
		if (device->total_bytes > device->bytes_used)
			total_avail = device->total_bytes - device->bytes_used;
		else
			total_avail = 0;
		/* avail is off by max(alloc_start, 1MB), but that is the same
		 * for all devices, so it doesn't hurt the sorting later on
2418 2419
		 */

2420 2421 2422 2423 2424
		ret = find_free_dev_extent(trans, device,
					   max_stripe_size * dev_stripes,
					   &dev_offset, &max_avail);
		if (ret && ret != -ENOSPC)
			goto error;
2425

2426 2427
		if (ret == 0)
			max_avail = max_stripe_size * dev_stripes;
2428

2429 2430
		if (max_avail < BTRFS_STRIPE_LEN * dev_stripes)
			continue;
2431

2432 2433 2434 2435 2436 2437
		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;
	}
2438

2439 2440 2441 2442 2443
	/*
	 * now sort the devices by hole size / available space
	 */
	sort(devices_info, ndevs, sizeof(struct btrfs_device_info),
	     btrfs_cmp_device_info, NULL);
2444

2445 2446
	/* round down to number of usable stripes */
	ndevs -= ndevs % devs_increment;
2447

2448 2449 2450
	if (ndevs < devs_increment * sub_stripes || ndevs < devs_min) {
		ret = -ENOSPC;
		goto error;
2451
	}
2452

2453 2454 2455 2456 2457 2458 2459 2460
	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;
2461

2462 2463 2464
	if (stripe_size * num_stripes > max_chunk_size * ncopies) {
		stripe_size = max_chunk_size * ncopies;
		do_div(stripe_size, num_stripes);
2465 2466
	}

2467 2468 2469
	do_div(stripe_size, dev_stripes);
	do_div(stripe_size, BTRFS_STRIPE_LEN);
	stripe_size *= BTRFS_STRIPE_LEN;
2470 2471 2472 2473 2474 2475 2476

	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
	if (!map) {
		ret = -ENOMEM;
		goto error;
	}
	map->num_stripes = num_stripes;
2477

2478 2479 2480 2481 2482 2483
	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;
2484 2485
		}
	}
Y
Yan Zheng 已提交
2486
	map->sector_size = extent_root->sectorsize;
2487 2488 2489
	map->stripe_len = BTRFS_STRIPE_LEN;
	map->io_align = BTRFS_STRIPE_LEN;
	map->io_width = BTRFS_STRIPE_LEN;
Y
Yan Zheng 已提交
2490 2491
	map->type = type;
	map->sub_stripes = sub_stripes;
2492

Y
Yan Zheng 已提交
2493
	*map_ret = map;
2494
	num_bytes = stripe_size * (num_stripes / ncopies);
2495

2496 2497
	*stripe_size_out = stripe_size;
	*num_bytes_out = num_bytes;
2498

2499
	trace_btrfs_chunk_alloc(info->chunk_root, map, start, num_bytes);
2500

2501
	em = alloc_extent_map();
Y
Yan Zheng 已提交
2502
	if (!em) {
2503 2504
		ret = -ENOMEM;
		goto error;
2505
	}
Y
Yan Zheng 已提交
2506 2507
	em->bdev = (struct block_device *)map;
	em->start = start;
2508
	em->len = num_bytes;
Y
Yan Zheng 已提交
2509 2510
	em->block_start = 0;
	em->block_len = em->len;
2511

Y
Yan Zheng 已提交
2512
	em_tree = &extent_root->fs_info->mapping_tree.map_tree;
2513
	write_lock(&em_tree->lock);
Y
Yan Zheng 已提交
2514
	ret = add_extent_mapping(em_tree, em);
2515
	write_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
2516 2517
	BUG_ON(ret);
	free_extent_map(em);
2518

Y
Yan Zheng 已提交
2519 2520
	ret = btrfs_make_block_group(trans, extent_root, 0, type,
				     BTRFS_FIRST_CHUNK_TREE_OBJECTID,
2521
				     start, num_bytes);
Y
Yan Zheng 已提交
2522
	BUG_ON(ret);
2523

2524 2525 2526 2527 2528 2529
	for (i = 0; i < map->num_stripes; ++i) {
		struct btrfs_device *device;
		u64 dev_offset;

		device = map->stripes[i].dev;
		dev_offset = map->stripes[i].physical;
2530 2531

		ret = btrfs_alloc_dev_extent(trans, device,
Y
Yan Zheng 已提交
2532 2533
				info->chunk_root->root_key.objectid,
				BTRFS_FIRST_CHUNK_TREE_OBJECTID,
2534
				start, dev_offset, stripe_size);
2535
		BUG_ON(ret);
Y
Yan Zheng 已提交
2536 2537
	}

2538
	kfree(devices_info);
Y
Yan Zheng 已提交
2539
	return 0;
2540 2541 2542 2543 2544

error:
	kfree(map);
	kfree(devices_info);
	return ret;
Y
Yan Zheng 已提交
2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569
}

static int __finish_chunk_alloc(struct btrfs_trans_handle *trans,
				struct btrfs_root *extent_root,
				struct map_lookup *map, u64 chunk_offset,
				u64 chunk_size, u64 stripe_size)
{
	u64 dev_offset;
	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;
	size_t item_size = btrfs_chunk_item_size(map->num_stripes);
	int index = 0;
	int ret;

	chunk = kzalloc(item_size, GFP_NOFS);
	if (!chunk)
		return -ENOMEM;

	index = 0;
	while (index < map->num_stripes) {
		device = map->stripes[index].dev;
		device->bytes_used += stripe_size;
2570 2571
		ret = btrfs_update_device(trans, device);
		BUG_ON(ret);
Y
Yan Zheng 已提交
2572 2573 2574 2575 2576 2577 2578 2579
		index++;
	}

	index = 0;
	stripe = &chunk->stripe;
	while (index < map->num_stripes) {
		device = map->stripes[index].dev;
		dev_offset = map->stripes[index].physical;
2580

2581 2582 2583
		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
2584
		stripe++;
2585 2586 2587
		index++;
	}

Y
Yan Zheng 已提交
2588
	btrfs_set_stack_chunk_length(chunk, chunk_size);
2589
	btrfs_set_stack_chunk_owner(chunk, extent_root->root_key.objectid);
Y
Yan Zheng 已提交
2590 2591 2592 2593 2594
	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);
2595
	btrfs_set_stack_chunk_sector_size(chunk, extent_root->sectorsize);
Y
Yan Zheng 已提交
2596
	btrfs_set_stack_chunk_sub_stripes(chunk, map->sub_stripes);
2597

Y
Yan Zheng 已提交
2598 2599 2600
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.type = BTRFS_CHUNK_ITEM_KEY;
	key.offset = chunk_offset;
2601

Y
Yan Zheng 已提交
2602 2603
	ret = btrfs_insert_item(trans, chunk_root, &key, chunk, item_size);
	BUG_ON(ret);
2604

Y
Yan Zheng 已提交
2605 2606 2607
	if (map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
		ret = btrfs_add_system_chunk(trans, chunk_root, &key, chunk,
					     item_size);
2608 2609
		BUG_ON(ret);
	}
2610

2611
	kfree(chunk);
Y
Yan Zheng 已提交
2612 2613
	return 0;
}
2614

Y
Yan Zheng 已提交
2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647
/*
 * 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;
	u64 chunk_size;
	u64 stripe_size;
	struct map_lookup *map;
	struct btrfs_root *chunk_root = extent_root->fs_info->chunk_root;
	int ret;

	ret = find_next_chunk(chunk_root, BTRFS_FIRST_CHUNK_TREE_OBJECTID,
			      &chunk_offset);
	if (ret)
		return ret;

	ret = __btrfs_alloc_chunk(trans, extent_root, &map, &chunk_size,
				  &stripe_size, chunk_offset, type);
	if (ret)
		return ret;

	ret = __finish_chunk_alloc(trans, extent_root, map, chunk_offset,
				   chunk_size, stripe_size);
	BUG_ON(ret);
	return 0;
}

C
Chris Mason 已提交
2648
static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
Y
Yan Zheng 已提交
2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 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 2697 2698 2699 2700 2701 2702 2703 2704 2705
					 struct btrfs_root *root,
					 struct btrfs_device *device)
{
	u64 chunk_offset;
	u64 sys_chunk_offset;
	u64 chunk_size;
	u64 sys_chunk_size;
	u64 stripe_size;
	u64 sys_stripe_size;
	u64 alloc_profile;
	struct map_lookup *map;
	struct map_lookup *sys_map;
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct btrfs_root *extent_root = fs_info->extent_root;
	int ret;

	ret = find_next_chunk(fs_info->chunk_root,
			      BTRFS_FIRST_CHUNK_TREE_OBJECTID, &chunk_offset);
	BUG_ON(ret);

	alloc_profile = BTRFS_BLOCK_GROUP_METADATA |
			(fs_info->metadata_alloc_profile &
			 fs_info->avail_metadata_alloc_bits);
	alloc_profile = btrfs_reduce_alloc_profile(root, alloc_profile);

	ret = __btrfs_alloc_chunk(trans, extent_root, &map, &chunk_size,
				  &stripe_size, chunk_offset, alloc_profile);
	BUG_ON(ret);

	sys_chunk_offset = chunk_offset + chunk_size;

	alloc_profile = BTRFS_BLOCK_GROUP_SYSTEM |
			(fs_info->system_alloc_profile &
			 fs_info->avail_system_alloc_bits);
	alloc_profile = btrfs_reduce_alloc_profile(root, alloc_profile);

	ret = __btrfs_alloc_chunk(trans, extent_root, &sys_map,
				  &sys_chunk_size, &sys_stripe_size,
				  sys_chunk_offset, alloc_profile);
	BUG_ON(ret);

	ret = btrfs_add_device(trans, fs_info->chunk_root, device);
	BUG_ON(ret);

	/*
	 * Modifying chunk tree needs allocating new blocks from both
	 * system block group and metadata block group. So we only can
	 * do operations require modifying the chunk tree after both
	 * block groups were created.
	 */
	ret = __finish_chunk_alloc(trans, extent_root, map, chunk_offset,
				   chunk_size, stripe_size);
	BUG_ON(ret);

	ret = __finish_chunk_alloc(trans, extent_root, sys_map,
				   sys_chunk_offset, sys_chunk_size,
				   sys_stripe_size);
2706
	BUG_ON(ret);
Y
Yan Zheng 已提交
2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717
	return 0;
}

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;
	int i;

2718
	read_lock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
2719
	em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
2720
	read_unlock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
2721 2722 2723
	if (!em)
		return 1;

2724 2725 2726 2727 2728
	if (btrfs_test_opt(root, DEGRADED)) {
		free_extent_map(em);
		return 0;
	}

Y
Yan Zheng 已提交
2729 2730 2731 2732 2733 2734 2735
	map = (struct map_lookup *)em->bdev;
	for (i = 0; i < map->num_stripes; i++) {
		if (!map->stripes[i].dev->writeable) {
			readonly = 1;
			break;
		}
	}
2736
	free_extent_map(em);
Y
Yan Zheng 已提交
2737
	return readonly;
2738 2739 2740 2741
}

void btrfs_mapping_init(struct btrfs_mapping_tree *tree)
{
2742
	extent_map_tree_init(&tree->map_tree);
2743 2744 2745 2746 2747 2748
}

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

C
Chris Mason 已提交
2749
	while (1) {
2750
		write_lock(&tree->map_tree.lock);
2751 2752 2753
		em = lookup_extent_mapping(&tree->map_tree, 0, (u64)-1);
		if (em)
			remove_extent_mapping(&tree->map_tree, em);
2754
		write_unlock(&tree->map_tree.lock);
2755 2756 2757 2758 2759 2760 2761 2762 2763 2764
		if (!em)
			break;
		kfree(em->bdev);
		/* once for us */
		free_extent_map(em);
		/* once for the tree */
		free_extent_map(em);
	}
}

2765 2766 2767 2768 2769 2770 2771
int btrfs_num_copies(struct btrfs_mapping_tree *map_tree, u64 logical, u64 len)
{
	struct extent_map *em;
	struct map_lookup *map;
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	int ret;

2772
	read_lock(&em_tree->lock);
2773
	em = lookup_extent_mapping(em_tree, logical, len);
2774
	read_unlock(&em_tree->lock);
2775 2776 2777 2778 2779 2780
	BUG_ON(!em);

	BUG_ON(em->start > logical || em->start + em->len < logical);
	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
2781 2782
	else if (map->type & BTRFS_BLOCK_GROUP_RAID10)
		ret = map->sub_stripes;
2783 2784 2785 2786 2787 2788
	else
		ret = 1;
	free_extent_map(em);
	return ret;
}

2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804
static int find_live_mirror(struct map_lookup *map, int first, int num,
			    int optimal)
{
	int i;
	if (map->stripes[optimal].dev->bdev)
		return optimal;
	for (i = first; i < first + num; i++) {
		if (map->stripes[i].dev->bdev)
			return i;
	}
	/* we couldn't find one that doesn't fail.  Just return something
	 * and the io error handling code will clean up eventually
	 */
	return optimal;
}

2805 2806 2807
static int __btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
			     u64 logical, u64 *length,
			     struct btrfs_multi_bio **multi_ret,
J
Jens Axboe 已提交
2808
			     int mirror_num)
2809 2810 2811 2812 2813
{
	struct extent_map *em;
	struct map_lookup *map;
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	u64 offset;
2814
	u64 stripe_offset;
2815
	u64 stripe_end_offset;
2816
	u64 stripe_nr;
2817 2818
	u64 stripe_nr_orig;
	u64 stripe_nr_end;
2819
	int stripes_allocated = 8;
C
Chris Mason 已提交
2820
	int stripes_required = 1;
2821
	int stripe_index;
2822
	int i;
2823
	int num_stripes;
2824
	int max_errors = 0;
2825
	struct btrfs_multi_bio *multi = NULL;
2826

2827
	if (multi_ret && !(rw & (REQ_WRITE | REQ_DISCARD)))
2828 2829 2830 2831 2832 2833 2834
		stripes_allocated = 1;
again:
	if (multi_ret) {
		multi = kzalloc(btrfs_multi_bio_size(stripes_allocated),
				GFP_NOFS);
		if (!multi)
			return -ENOMEM;
2835 2836

		atomic_set(&multi->error, 0);
2837
	}
2838

2839
	read_lock(&em_tree->lock);
2840
	em = lookup_extent_mapping(em_tree, logical, *length);
2841
	read_unlock(&em_tree->lock);
2842

2843
	if (!em) {
C
Chris Mason 已提交
2844 2845 2846
		printk(KERN_CRIT "unable to find logical %llu len %llu\n",
		       (unsigned long long)logical,
		       (unsigned long long)*length);
2847
		BUG();
2848
	}
2849 2850 2851 2852

	BUG_ON(em->start > logical || em->start + em->len < logical);
	map = (struct map_lookup *)em->bdev;
	offset = logical - em->start;
2853

2854 2855 2856
	if (mirror_num > map->num_stripes)
		mirror_num = 0;

2857
	/* if our multi bio struct is too small, back off and try again */
2858
	if (rw & REQ_WRITE) {
C
Chris Mason 已提交
2859 2860 2861
		if (map->type & (BTRFS_BLOCK_GROUP_RAID1 |
				 BTRFS_BLOCK_GROUP_DUP)) {
			stripes_required = map->num_stripes;
2862
			max_errors = 1;
C
Chris Mason 已提交
2863 2864
		} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
			stripes_required = map->sub_stripes;
2865
			max_errors = 1;
C
Chris Mason 已提交
2866 2867
		}
	}
2868 2869 2870 2871 2872 2873 2874 2875 2876
	if (rw & REQ_DISCARD) {
		if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
				 BTRFS_BLOCK_GROUP_RAID1 |
				 BTRFS_BLOCK_GROUP_DUP |
				 BTRFS_BLOCK_GROUP_RAID10)) {
			stripes_required = map->num_stripes;
		}
	}
	if (multi_ret && (rw & (REQ_WRITE | REQ_DISCARD)) &&
C
Chris Mason 已提交
2877
	    stripes_allocated < stripes_required) {
2878 2879 2880 2881 2882
		stripes_allocated = map->num_stripes;
		free_extent_map(em);
		kfree(multi);
		goto again;
	}
2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895
	stripe_nr = offset;
	/*
	 * stripe_nr counts the total number of stripes we have to stride
	 * to get to this block
	 */
	do_div(stripe_nr, map->stripe_len);

	stripe_offset = stripe_nr * map->stripe_len;
	BUG_ON(offset < stripe_offset);

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

2896 2897 2898 2899 2900 2901
	if (rw & REQ_DISCARD)
		*length = min_t(u64, em->len - offset, *length);
	else if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
			      BTRFS_BLOCK_GROUP_RAID1 |
			      BTRFS_BLOCK_GROUP_RAID10 |
			      BTRFS_BLOCK_GROUP_DUP)) {
2902 2903
		/* we limit the length of each bio to what fits in a stripe */
		*length = min_t(u64, em->len - offset,
2904
				map->stripe_len - stripe_offset);
2905 2906 2907
	} else {
		*length = em->len - offset;
	}
2908

J
Jens Axboe 已提交
2909
	if (!multi_ret)
2910 2911
		goto out;

2912
	num_stripes = 1;
2913
	stripe_index = 0;
2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925
	stripe_nr_orig = stripe_nr;
	stripe_nr_end = (offset + *length + map->stripe_len - 1) &
			(~(map->stripe_len - 1));
	do_div(stripe_nr_end, map->stripe_len);
	stripe_end_offset = stripe_nr_end * map->stripe_len -
			    (offset + *length);
	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);
	} else if (map->type & BTRFS_BLOCK_GROUP_RAID1) {
2926
		if (rw & (REQ_WRITE | REQ_DISCARD))
2927
			num_stripes = map->num_stripes;
2928
		else if (mirror_num)
2929
			stripe_index = mirror_num - 1;
2930 2931 2932 2933 2934
		else {
			stripe_index = find_live_mirror(map, 0,
					    map->num_stripes,
					    current->pid % map->num_stripes);
		}
2935

2936
	} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
2937
		if (rw & (REQ_WRITE | REQ_DISCARD))
2938
			num_stripes = map->num_stripes;
2939 2940
		else if (mirror_num)
			stripe_index = mirror_num - 1;
2941

C
Chris Mason 已提交
2942 2943 2944 2945 2946 2947
	} 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;

J
Jens Axboe 已提交
2948
		if (rw & REQ_WRITE)
2949
			num_stripes = map->sub_stripes;
2950 2951 2952 2953
		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
2954 2955
		else if (mirror_num)
			stripe_index += mirror_num - 1;
2956 2957 2958 2959 2960
		else {
			stripe_index = find_live_mirror(map, stripe_index,
					      map->sub_stripes, stripe_index +
					      current->pid % map->sub_stripes);
		}
2961 2962 2963 2964 2965 2966 2967 2968
	} 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);
	}
2969
	BUG_ON(stripe_index >= map->num_stripes);
2970

2971 2972
	if (rw & REQ_DISCARD) {
		for (i = 0; i < num_stripes; i++) {
2973 2974 2975 2976
			multi->stripes[i].physical =
				map->stripes[stripe_index].physical +
				stripe_offset + stripe_nr * map->stripe_len;
			multi->stripes[i].dev = map->stripes[stripe_index].dev;
2977 2978 2979

			if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
				u64 stripes;
2980
				u32 last_stripe = 0;
2981 2982
				int j;

2983 2984 2985 2986
				div_u64_rem(stripe_nr_end - 1,
					    map->num_stripes,
					    &last_stripe);

2987
				for (j = 0; j < map->num_stripes; j++) {
2988 2989 2990 2991 2992
					u32 test;

					div_u64_rem(stripe_nr_end - 1 - j,
						    map->num_stripes, &test);
					if (test == stripe_index)
2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012
						break;
				}
				stripes = stripe_nr_end - 1 - j;
				do_div(stripes, map->num_stripes);
				multi->stripes[i].length = map->stripe_len *
					(stripes - stripe_nr + 1);

				if (i == 0) {
					multi->stripes[i].length -=
						stripe_offset;
					stripe_offset = 0;
				}
				if (stripe_index == last_stripe)
					multi->stripes[i].length -=
						stripe_end_offset;
			} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
				u64 stripes;
				int j;
				int factor = map->num_stripes /
					     map->sub_stripes;
3013 3014 3015 3016
				u32 last_stripe = 0;

				div_u64_rem(stripe_nr_end - 1,
					    factor, &last_stripe);
3017 3018 3019
				last_stripe *= map->sub_stripes;

				for (j = 0; j < factor; j++) {
3020 3021 3022 3023 3024 3025
					u32 test;

					div_u64_rem(stripe_nr_end - 1 - j,
						    factor, &test);

					if (test ==
3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057
					    stripe_index / map->sub_stripes)
						break;
				}
				stripes = stripe_nr_end - 1 - j;
				do_div(stripes, factor);
				multi->stripes[i].length = map->stripe_len *
					(stripes - stripe_nr + 1);

				if (i < map->sub_stripes) {
					multi->stripes[i].length -=
						stripe_offset;
					if (i == map->sub_stripes - 1)
						stripe_offset = 0;
				}
				if (stripe_index >= last_stripe &&
				    stripe_index <= (last_stripe +
						     map->sub_stripes - 1)) {
					multi->stripes[i].length -=
						stripe_end_offset;
				}
			} else
				multi->stripes[i].length = *length;

			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++) {
3058 3059 3060 3061 3062 3063
			multi->stripes[i].physical =
				map->stripes[stripe_index].physical +
				stripe_offset +
				stripe_nr * map->stripe_len;
			multi->stripes[i].dev =
				map->stripes[stripe_index].dev;
3064
			stripe_index++;
3065
		}
3066
	}
3067 3068 3069
	if (multi_ret) {
		*multi_ret = multi;
		multi->num_stripes = num_stripes;
3070
		multi->max_errors = max_errors;
3071
	}
3072
out:
3073 3074 3075 3076
	free_extent_map(em);
	return 0;
}

3077 3078 3079 3080 3081
int btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
		      u64 logical, u64 *length,
		      struct btrfs_multi_bio **multi_ret, int mirror_num)
{
	return __btrfs_map_block(map_tree, rw, logical, length, multi_ret,
J
Jens Axboe 已提交
3082
				 mirror_num);
3083 3084
}

Y
Yan Zheng 已提交
3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097
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;
	int i, j, nr = 0;

3098
	read_lock(&em_tree->lock);
Y
Yan Zheng 已提交
3099
	em = lookup_extent_mapping(em_tree, chunk_start, 1);
3100
	read_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130

	BUG_ON(!em || em->start != chunk_start);
	map = (struct map_lookup *)em->bdev;

	length = em->len;
	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);

	buf = kzalloc(sizeof(u64) * map->num_stripes, GFP_NOFS);
	BUG_ON(!buf);

	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;
		}
		bytenr = chunk_start + stripe_nr * map->stripe_len;
3131
		WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
3132 3133 3134 3135
		for (j = 0; j < nr; j++) {
			if (buf[j] == bytenr)
				break;
		}
3136 3137
		if (j == nr) {
			WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
3138
			buf[nr++] = bytenr;
3139
		}
Y
Yan Zheng 已提交
3140 3141 3142 3143 3144 3145 3146 3147
	}

	*logical = buf;
	*naddrs = nr;
	*stripe_len = map->stripe_len;

	free_extent_map(em);
	return 0;
3148 3149
}

3150 3151
static void end_bio_multi_stripe(struct bio *bio, int err)
{
3152
	struct btrfs_multi_bio *multi = bio->bi_private;
3153
	int is_orig_bio = 0;
3154 3155

	if (err)
3156
		atomic_inc(&multi->error);
3157

3158 3159 3160
	if (bio == multi->orig_bio)
		is_orig_bio = 1;

3161
	if (atomic_dec_and_test(&multi->stripes_pending)) {
3162 3163 3164 3165
		if (!is_orig_bio) {
			bio_put(bio);
			bio = multi->orig_bio;
		}
3166 3167
		bio->bi_private = multi->private;
		bio->bi_end_io = multi->end_io;
3168 3169 3170
		/* only send an error to the higher layers if it is
		 * beyond the tolerance of the multi-bio
		 */
3171
		if (atomic_read(&multi->error) > multi->max_errors) {
3172
			err = -EIO;
3173 3174 3175 3176 3177 3178
		} else if (err) {
			/*
			 * this bio is actually up to date, we didn't
			 * go over the max number of errors
			 */
			set_bit(BIO_UPTODATE, &bio->bi_flags);
3179
			err = 0;
3180
		}
3181 3182 3183
		kfree(multi);

		bio_endio(bio, err);
3184
	} else if (!is_orig_bio) {
3185 3186 3187 3188
		bio_put(bio);
	}
}

3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202
struct async_sched {
	struct bio *bio;
	int rw;
	struct btrfs_fs_info *info;
	struct btrfs_work work;
};

/*
 * 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.
 */
C
Chris Mason 已提交
3203
static noinline int schedule_bio(struct btrfs_root *root,
3204 3205
				 struct btrfs_device *device,
				 int rw, struct bio *bio)
3206 3207
{
	int should_queue = 1;
3208
	struct btrfs_pending_bios *pending_bios;
3209 3210

	/* don't bother with additional async steps for reads, right now */
3211
	if (!(rw & REQ_WRITE)) {
3212
		bio_get(bio);
3213
		submit_bio(rw, bio);
3214
		bio_put(bio);
3215 3216 3217 3218
		return 0;
	}

	/*
3219
	 * nr_async_bios allows us to reliably return congestion to the
3220 3221 3222 3223
	 * 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
	 */
3224
	atomic_inc(&root->fs_info->nr_async_bios);
3225
	WARN_ON(bio->bi_next);
3226 3227 3228 3229
	bio->bi_next = NULL;
	bio->bi_rw |= rw;

	spin_lock(&device->io_lock);
3230
	if (bio->bi_rw & REQ_SYNC)
3231 3232 3233
		pending_bios = &device->pending_sync_bios;
	else
		pending_bios = &device->pending_bios;
3234

3235 3236
	if (pending_bios->tail)
		pending_bios->tail->bi_next = bio;
3237

3238 3239 3240
	pending_bios->tail = bio;
	if (!pending_bios->head)
		pending_bios->head = bio;
3241 3242 3243 3244 3245 3246
	if (device->running_pending)
		should_queue = 0;

	spin_unlock(&device->io_lock);

	if (should_queue)
3247 3248
		btrfs_queue_worker(&root->fs_info->submit_workers,
				   &device->work);
3249 3250 3251
	return 0;
}

3252
int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
3253
		  int mirror_num, int async_submit)
3254 3255 3256
{
	struct btrfs_mapping_tree *map_tree;
	struct btrfs_device *dev;
3257
	struct bio *first_bio = bio;
3258
	u64 logical = (u64)bio->bi_sector << 9;
3259 3260
	u64 length = 0;
	u64 map_length;
3261
	struct btrfs_multi_bio *multi = NULL;
3262
	int ret;
3263 3264
	int dev_nr = 0;
	int total_devs = 1;
3265

3266
	length = bio->bi_size;
3267 3268
	map_tree = &root->fs_info->mapping_tree;
	map_length = length;
3269

3270 3271
	ret = btrfs_map_block(map_tree, rw, logical, &map_length, &multi,
			      mirror_num);
3272 3273 3274 3275
	BUG_ON(ret);

	total_devs = multi->num_stripes;
	if (map_length < length) {
C
Chris Mason 已提交
3276 3277 3278 3279
		printk(KERN_CRIT "mapping failed logical %llu bio len %llu "
		       "len %llu\n", (unsigned long long)logical,
		       (unsigned long long)length,
		       (unsigned long long)map_length);
3280 3281 3282 3283
		BUG();
	}
	multi->end_io = first_bio->bi_end_io;
	multi->private = first_bio->bi_private;
3284
	multi->orig_bio = first_bio;
3285 3286
	atomic_set(&multi->stripes_pending, multi->num_stripes);

C
Chris Mason 已提交
3287
	while (dev_nr < total_devs) {
3288 3289 3290 3291 3292 3293 3294 3295 3296 3297
		if (total_devs > 1) {
			if (dev_nr < total_devs - 1) {
				bio = bio_clone(first_bio, GFP_NOFS);
				BUG_ON(!bio);
			} else {
				bio = first_bio;
			}
			bio->bi_private = multi;
			bio->bi_end_io = end_bio_multi_stripe;
		}
3298 3299
		bio->bi_sector = multi->stripes[dev_nr].physical >> 9;
		dev = multi->stripes[dev_nr].dev;
3300
		if (dev && dev->bdev && (rw != WRITE || dev->writeable)) {
3301
			bio->bi_bdev = dev->bdev;
3302 3303 3304 3305
			if (async_submit)
				schedule_bio(root, dev, rw, bio);
			else
				submit_bio(rw, bio);
3306 3307 3308 3309 3310
		} else {
			bio->bi_bdev = root->fs_info->fs_devices->latest_bdev;
			bio->bi_sector = logical >> 9;
			bio_endio(bio, -EIO);
		}
3311 3312
		dev_nr++;
	}
3313 3314
	if (total_devs == 1)
		kfree(multi);
3315 3316 3317
	return 0;
}

3318
struct btrfs_device *btrfs_find_device(struct btrfs_root *root, u64 devid,
Y
Yan Zheng 已提交
3319
				       u8 *uuid, u8 *fsid)
3320
{
Y
Yan Zheng 已提交
3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335
	struct btrfs_device *device;
	struct btrfs_fs_devices *cur_devices;

	cur_devices = root->fs_info->fs_devices;
	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;
3336 3337
}

3338 3339 3340 3341 3342 3343 3344
static struct btrfs_device *add_missing_dev(struct btrfs_root *root,
					    u64 devid, u8 *dev_uuid)
{
	struct btrfs_device *device;
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;

	device = kzalloc(sizeof(*device), GFP_NOFS);
3345 3346
	if (!device)
		return NULL;
3347 3348 3349 3350
	list_add(&device->dev_list,
		 &fs_devices->devices);
	device->dev_root = root->fs_info->dev_root;
	device->devid = devid;
3351
	device->work.func = pending_bios_fn;
Y
Yan Zheng 已提交
3352
	device->fs_devices = fs_devices;
3353
	device->missing = 1;
3354
	fs_devices->num_devices++;
3355
	fs_devices->missing_devices++;
3356
	spin_lock_init(&device->io_lock);
3357
	INIT_LIST_HEAD(&device->dev_alloc_list);
3358 3359 3360 3361
	memcpy(device->uuid, dev_uuid, BTRFS_UUID_SIZE);
	return device;
}

3362 3363 3364 3365 3366 3367 3368 3369 3370 3371
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;
3372
	u8 uuid[BTRFS_UUID_SIZE];
3373
	int num_stripes;
3374
	int ret;
3375
	int i;
3376

3377 3378
	logical = key->offset;
	length = btrfs_chunk_length(leaf, chunk);
3379

3380
	read_lock(&map_tree->map_tree.lock);
3381
	em = lookup_extent_mapping(&map_tree->map_tree, logical, 1);
3382
	read_unlock(&map_tree->map_tree.lock);
3383 3384 3385 3386 3387 3388 3389 3390 3391

	/* 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);
	}

3392
	em = alloc_extent_map();
3393 3394
	if (!em)
		return -ENOMEM;
3395 3396
	num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
3397 3398 3399 3400 3401 3402 3403 3404 3405
	if (!map) {
		free_extent_map(em);
		return -ENOMEM;
	}

	em->bdev = (struct block_device *)map;
	em->start = logical;
	em->len = length;
	em->block_start = 0;
C
Chris Mason 已提交
3406
	em->block_len = em->len;
3407

3408 3409 3410 3411 3412 3413
	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
3414
	map->sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
3415 3416 3417 3418
	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);
3419 3420 3421
		read_extent_buffer(leaf, uuid, (unsigned long)
				   btrfs_stripe_dev_uuid_nr(chunk, i),
				   BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3422 3423
		map->stripes[i].dev = btrfs_find_device(root, devid, uuid,
							NULL);
3424
		if (!map->stripes[i].dev && !btrfs_test_opt(root, DEGRADED)) {
3425 3426 3427 3428
			kfree(map);
			free_extent_map(em);
			return -EIO;
		}
3429 3430 3431 3432 3433 3434 3435 3436 3437 3438
		if (!map->stripes[i].dev) {
			map->stripes[i].dev =
				add_missing_dev(root, devid, uuid);
			if (!map->stripes[i].dev) {
				kfree(map);
				free_extent_map(em);
				return -EIO;
			}
		}
		map->stripes[i].dev->in_fs_metadata = 1;
3439 3440
	}

3441
	write_lock(&map_tree->map_tree.lock);
3442
	ret = add_extent_mapping(&map_tree->map_tree, em);
3443
	write_unlock(&map_tree->map_tree.lock);
3444
	BUG_ON(ret);
3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456
	free_extent_map(em);

	return 0;
}

static int fill_device_from_item(struct extent_buffer *leaf,
				 struct btrfs_dev_item *dev_item,
				 struct btrfs_device *device)
{
	unsigned long ptr;

	device->devid = btrfs_device_id(leaf, dev_item);
3457 3458
	device->disk_total_bytes = btrfs_device_total_bytes(leaf, dev_item);
	device->total_bytes = device->disk_total_bytes;
3459 3460 3461 3462 3463 3464 3465
	device->bytes_used = btrfs_device_bytes_used(leaf, dev_item);
	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);

	ptr = (unsigned long)btrfs_device_uuid(dev_item);
3466
	read_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
3467 3468 3469 3470

	return 0;
}

Y
Yan Zheng 已提交
3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491
static int open_seed_devices(struct btrfs_root *root, u8 *fsid)
{
	struct btrfs_fs_devices *fs_devices;
	int ret;

	mutex_lock(&uuid_mutex);

	fs_devices = root->fs_info->fs_devices->seed;
	while (fs_devices) {
		if (!memcmp(fs_devices->fsid, fsid, BTRFS_UUID_SIZE)) {
			ret = 0;
			goto out;
		}
		fs_devices = fs_devices->seed;
	}

	fs_devices = find_fsid(fsid);
	if (!fs_devices) {
		ret = -ENOENT;
		goto out;
	}
Y
Yan Zheng 已提交
3492 3493 3494 3495

	fs_devices = clone_fs_devices(fs_devices);
	if (IS_ERR(fs_devices)) {
		ret = PTR_ERR(fs_devices);
Y
Yan Zheng 已提交
3496 3497 3498
		goto out;
	}

3499
	ret = __btrfs_open_devices(fs_devices, FMODE_READ,
3500
				   root->fs_info->bdev_holder);
Y
Yan Zheng 已提交
3501 3502 3503 3504 3505
	if (ret)
		goto out;

	if (!fs_devices->seeding) {
		__btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
3506
		free_fs_devices(fs_devices);
Y
Yan Zheng 已提交
3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517
		ret = -EINVAL;
		goto out;
	}

	fs_devices->seed = root->fs_info->fs_devices->seed;
	root->fs_info->fs_devices->seed = fs_devices;
out:
	mutex_unlock(&uuid_mutex);
	return ret;
}

3518
static int read_one_dev(struct btrfs_root *root,
3519 3520 3521 3522 3523 3524
			struct extent_buffer *leaf,
			struct btrfs_dev_item *dev_item)
{
	struct btrfs_device *device;
	u64 devid;
	int ret;
Y
Yan Zheng 已提交
3525
	u8 fs_uuid[BTRFS_UUID_SIZE];
3526 3527
	u8 dev_uuid[BTRFS_UUID_SIZE];

3528
	devid = btrfs_device_id(leaf, dev_item);
3529 3530 3531
	read_extent_buffer(leaf, dev_uuid,
			   (unsigned long)btrfs_device_uuid(dev_item),
			   BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3532 3533 3534 3535 3536 3537
	read_extent_buffer(leaf, fs_uuid,
			   (unsigned long)btrfs_device_fsid(dev_item),
			   BTRFS_UUID_SIZE);

	if (memcmp(fs_uuid, root->fs_info->fsid, BTRFS_UUID_SIZE)) {
		ret = open_seed_devices(root, fs_uuid);
Y
Yan Zheng 已提交
3538
		if (ret && !btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
3539 3540 3541 3542 3543
			return ret;
	}

	device = btrfs_find_device(root, devid, dev_uuid, fs_uuid);
	if (!device || !device->bdev) {
Y
Yan Zheng 已提交
3544
		if (!btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
3545 3546 3547
			return -EIO;

		if (!device) {
C
Chris Mason 已提交
3548 3549
			printk(KERN_WARNING "warning devid %llu missing\n",
			       (unsigned long long)devid);
Y
Yan Zheng 已提交
3550 3551 3552
			device = add_missing_dev(root, devid, dev_uuid);
			if (!device)
				return -ENOMEM;
3553 3554 3555 3556 3557 3558 3559 3560 3561
		} else if (!device->missing) {
			/*
			 * 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
			 */
			root->fs_info->fs_devices->missing_devices++;
			device->missing = 1;
Y
Yan Zheng 已提交
3562 3563 3564 3565 3566 3567 3568 3569
		}
	}

	if (device->fs_devices != root->fs_info->fs_devices) {
		BUG_ON(device->writeable);
		if (device->generation !=
		    btrfs_device_generation(leaf, dev_item))
			return -EINVAL;
3570
	}
3571 3572 3573

	fill_device_from_item(leaf, dev_item, device);
	device->dev_root = root->fs_info->dev_root;
3574
	device->in_fs_metadata = 1;
Y
Yan Zheng 已提交
3575 3576
	if (device->writeable)
		device->fs_devices->total_rw_bytes += device->total_bytes;
3577 3578 3579 3580
	ret = 0;
	return ret;
}

Y
Yan Zheng 已提交
3581
int btrfs_read_sys_array(struct btrfs_root *root)
3582 3583
{
	struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
3584
	struct extent_buffer *sb;
3585 3586
	struct btrfs_disk_key *disk_key;
	struct btrfs_chunk *chunk;
3587 3588 3589
	u8 *ptr;
	unsigned long sb_ptr;
	int ret = 0;
3590 3591 3592 3593
	u32 num_stripes;
	u32 array_size;
	u32 len = 0;
	u32 cur;
3594
	struct btrfs_key key;
3595

Y
Yan Zheng 已提交
3596
	sb = btrfs_find_create_tree_block(root, BTRFS_SUPER_INFO_OFFSET,
3597 3598 3599 3600
					  BTRFS_SUPER_INFO_SIZE);
	if (!sb)
		return -ENOMEM;
	btrfs_set_buffer_uptodate(sb);
3601 3602
	btrfs_set_buffer_lockdep_class(sb, 0);

3603
	write_extent_buffer(sb, super_copy, 0, BTRFS_SUPER_INFO_SIZE);
3604 3605 3606 3607 3608 3609 3610 3611 3612 3613
	array_size = btrfs_super_sys_array_size(super_copy);

	ptr = super_copy->sys_chunk_array;
	sb_ptr = offsetof(struct btrfs_super_block, sys_chunk_array);
	cur = 0;

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

3614
		len = sizeof(*disk_key); ptr += len;
3615 3616 3617
		sb_ptr += len;
		cur += len;

3618
		if (key.type == BTRFS_CHUNK_ITEM_KEY) {
3619
			chunk = (struct btrfs_chunk *)sb_ptr;
3620
			ret = read_one_chunk(root, &key, sb, chunk);
3621 3622
			if (ret)
				break;
3623 3624 3625
			num_stripes = btrfs_chunk_num_stripes(sb, chunk);
			len = btrfs_chunk_item_size(num_stripes);
		} else {
3626 3627
			ret = -EIO;
			break;
3628 3629 3630 3631 3632
		}
		ptr += len;
		sb_ptr += len;
		cur += len;
	}
3633
	free_extent_buffer(sb);
3634
	return ret;
3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660
}

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;

	/* first we search for all of the device items, and then we
	 * read in all of the chunk items.  This way we can create chunk
	 * mappings that reference all of the devices that are afound
	 */
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.offset = 0;
	key.type = 0;
again:
	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
3661 3662
	if (ret < 0)
		goto error;
C
Chris Mason 已提交
3663
	while (1) {
3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681
		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);
		if (key.objectid == BTRFS_DEV_ITEMS_OBJECTID) {
			if (found_key.objectid != BTRFS_DEV_ITEMS_OBJECTID)
				break;
			if (found_key.type == BTRFS_DEV_ITEM_KEY) {
				struct btrfs_dev_item *dev_item;
				dev_item = btrfs_item_ptr(leaf, slot,
						  struct btrfs_dev_item);
3682
				ret = read_one_dev(root, leaf, dev_item);
Y
Yan Zheng 已提交
3683 3684
				if (ret)
					goto error;
3685 3686 3687 3688 3689
			}
		} 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
3690 3691
			if (ret)
				goto error;
3692 3693 3694 3695 3696
		}
		path->slots[0]++;
	}
	if (key.objectid == BTRFS_DEV_ITEMS_OBJECTID) {
		key.objectid = 0;
3697
		btrfs_release_path(path);
3698 3699 3700 3701
		goto again;
	}
	ret = 0;
error:
Y
Yan Zheng 已提交
3702
	btrfs_free_path(path);
3703 3704
	return ret;
}