volumes.c 93.3 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

36 37 38 39 40 41 42
struct map_lookup {
	u64 type;
	int io_align;
	int io_width;
	int stripe_len;
	int sector_size;
	int num_stripes;
C
Chris Mason 已提交
43
	int sub_stripes;
44
	struct btrfs_bio_stripe stripes[];
45 46
};

Y
Yan Zheng 已提交
47 48 49 50 51
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);

52
#define map_lookup_size(n) (sizeof(struct map_lookup) + \
53
			    (sizeof(struct btrfs_bio_stripe) * (n)))
54

55 56 57
static DEFINE_MUTEX(uuid_mutex);
static LIST_HEAD(fs_uuids);

58 59 60 61 62 63 64 65 66 67
void btrfs_lock_volumes(void)
{
	mutex_lock(&uuid_mutex);
}

void btrfs_unlock_volumes(void)
{
	mutex_unlock(&uuid_mutex);
}

68 69 70 71 72 73 74 75 76 77
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
78 79 80 81 82 83 84 85 86 87 88 89 90 91
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);
}

92 93 94 95
int btrfs_cleanup_fs_uuids(void)
{
	struct btrfs_fs_devices *fs_devices;

Y
Yan Zheng 已提交
96 97 98 99
	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
100
		free_fs_devices(fs_devices);
101 102 103 104
	}
	return 0;
}

105 106
static noinline struct btrfs_device *__find_device(struct list_head *head,
						   u64 devid, u8 *uuid)
107 108 109
{
	struct btrfs_device *dev;

Q
Qinghuang Feng 已提交
110
	list_for_each_entry(dev, head, dev_list) {
111
		if (dev->devid == devid &&
112
		    (!uuid || !memcmp(dev->uuid, uuid, BTRFS_UUID_SIZE))) {
113
			return dev;
114
		}
115 116 117 118
	}
	return NULL;
}

119
static noinline struct btrfs_fs_devices *find_fsid(u8 *fsid)
120 121 122
{
	struct btrfs_fs_devices *fs_devices;

Q
Qinghuang Feng 已提交
123
	list_for_each_entry(fs_devices, &fs_uuids, list) {
124 125 126 127 128 129
		if (memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE) == 0)
			return fs_devices;
	}
	return NULL;
}

130 131 132 133 134 135 136 137 138 139 140 141 142 143
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;
}

144 145 146 147 148 149 150 151 152 153 154
/*
 * 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
155
static noinline int run_scheduled_bios(struct btrfs_device *device)
156 157 158
{
	struct bio *pending;
	struct backing_dev_info *bdi;
159
	struct btrfs_fs_info *fs_info;
160
	struct btrfs_pending_bios *pending_bios;
161 162 163
	struct bio *tail;
	struct bio *cur;
	int again = 0;
164 165
	unsigned long num_run;
	unsigned long num_sync_run;
166
	unsigned long batch_run = 0;
167
	unsigned long limit;
168
	unsigned long last_waited = 0;
169
	int force_reg = 0;
170

171
	bdi = blk_get_backing_dev_info(device->bdev);
172 173 174 175
	fs_info = device->dev_root->fs_info;
	limit = btrfs_async_submit_limit(fs_info);
	limit = limit * 2 / 3;

176 177 178 179 180
	/* we want to make sure that every time we switch from the sync
	 * list to the normal list, we unplug
	 */
	num_sync_run = 0;

181 182 183
loop:
	spin_lock(&device->io_lock);

184
loop_lock:
185
	num_run = 0;
186

187 188 189 190 191
	/* 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
	 */
192
	if (!force_reg && device->pending_sync_bios.head) {
193
		pending_bios = &device->pending_sync_bios;
194 195
		force_reg = 1;
	} else {
196
		pending_bios = &device->pending_bios;
197 198
		force_reg = 0;
	}
199 200 201

	pending = pending_bios->head;
	tail = pending_bios->tail;
202 203 204 205 206 207 208 209 210 211
	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.
	 */
212 213
	if (device->pending_sync_bios.head == NULL &&
	    device->pending_bios.head == NULL) {
214 215
		again = 0;
		device->running_pending = 0;
216 217 218
	} else {
		again = 1;
		device->running_pending = 1;
219
	}
220 221 222 223

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

224 225
	spin_unlock(&device->io_lock);

226 227 228 229 230 231 232 233 234
	/*
	 * if we're doing the regular priority list, make sure we unplug
	 * for any high prio bios we've sent down
	 */
	if (pending_bios == &device->pending_bios && num_sync_run > 0) {
		num_sync_run = 0;
		blk_run_backing_dev(bdi, NULL);
	}

C
Chris Mason 已提交
235
	while (pending) {
236 237

		rmb();
238 239 240 241 242 243 244 245
		/* 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)) {
246 247 248 249 250
			spin_lock(&device->io_lock);
			requeue_list(pending_bios, pending, tail);
			goto loop_lock;
		}

251 252 253
		cur = pending;
		pending = pending->bi_next;
		cur->bi_next = NULL;
254 255 256 257 258
		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);
259 260

		BUG_ON(atomic_read(&cur->bi_cnt) == 0);
261

262
		if (cur->bi_rw & REQ_SYNC)
263 264
			num_sync_run++;

265 266 267
		submit_bio(cur->bi_rw, cur);
		num_run++;
		batch_run++;
268 269 270 271 272 273 274
		if (need_resched()) {
			if (num_sync_run) {
				blk_run_backing_dev(bdi, NULL);
				num_sync_run = 0;
			}
			cond_resched();
		}
275 276 277 278 279 280

		/*
		 * 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
281
		if (pending && bdi_write_congested(bdi) && batch_run > 8 &&
282
		    fs_info->fs_devices->open_devices > 1) {
283
			struct io_context *ioc;
284

285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306
			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;
307 308 309 310 311 312 313
				if (need_resched()) {
					if (num_sync_run) {
						blk_run_backing_dev(bdi, NULL);
						num_sync_run = 0;
					}
					cond_resched();
				}
314 315
				continue;
			}
316
			spin_lock(&device->io_lock);
317
			requeue_list(pending_bios, pending, tail);
318
			device->running_pending = 1;
319 320 321 322 323 324

			spin_unlock(&device->io_lock);
			btrfs_requeue_work(&device->work);
			goto done;
		}
	}
325 326 327 328 329

	if (num_sync_run) {
		num_sync_run = 0;
		blk_run_backing_dev(bdi, NULL);
	}
330 331 332 333 334 335 336 337 338 339 340
	/*
	 * IO has already been through a long path to get here.  Checksumming,
	 * async helper threads, perhaps compression.  We've done a pretty
	 * good job of collecting a batch of IO and should just unplug
	 * the device right away.
	 *
	 * This will help anyone who is waiting on the IO, they might have
	 * already unplugged, but managed to do so before the bio they
	 * cared about found its way down here.
	 */
	blk_run_backing_dev(bdi, NULL);
341 342 343 344 345 346 347 348 349 350

	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);

351 352 353 354
done:
	return 0;
}

355
static void pending_bios_fn(struct btrfs_work *work)
356 357 358 359 360 361 362
{
	struct btrfs_device *device;

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

363
static noinline int device_list_add(const char *path,
364 365 366 367 368 369
			   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);
370
	char *name;
371 372 373

	fs_devices = find_fsid(disk_super->fsid);
	if (!fs_devices) {
374
		fs_devices = kzalloc(sizeof(*fs_devices), GFP_NOFS);
375 376 377
		if (!fs_devices)
			return -ENOMEM;
		INIT_LIST_HEAD(&fs_devices->devices);
378
		INIT_LIST_HEAD(&fs_devices->alloc_list);
379 380 381 382
		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;
383
		mutex_init(&fs_devices->device_list_mutex);
384 385
		device = NULL;
	} else {
386 387
		device = __find_device(&fs_devices->devices, devid,
				       disk_super->dev_item.uuid);
388 389
	}
	if (!device) {
Y
Yan Zheng 已提交
390 391 392
		if (fs_devices->opened)
			return -EBUSY;

393 394 395 396 397 398
		device = kzalloc(sizeof(*device), GFP_NOFS);
		if (!device) {
			/* we can safely leave the fs_devices entry around */
			return -ENOMEM;
		}
		device->devid = devid;
399
		device->work.func = pending_bios_fn;
400 401
		memcpy(device->uuid, disk_super->dev_item.uuid,
		       BTRFS_UUID_SIZE);
402
		device->barriers = 1;
403
		spin_lock_init(&device->io_lock);
404 405 406 407 408
		device->name = kstrdup(path, GFP_NOFS);
		if (!device->name) {
			kfree(device);
			return -ENOMEM;
		}
Y
Yan Zheng 已提交
409
		INIT_LIST_HEAD(&device->dev_alloc_list);
410 411

		mutex_lock(&fs_devices->device_list_mutex);
412
		list_add(&device->dev_list, &fs_devices->devices);
413 414
		mutex_unlock(&fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
415
		device->fs_devices = fs_devices;
416
		fs_devices->num_devices++;
417
	} else if (!device->name || strcmp(device->name, path)) {
418 419 420 421 422
		name = kstrdup(path, GFP_NOFS);
		if (!name)
			return -ENOMEM;
		kfree(device->name);
		device->name = name;
423 424 425 426
		if (device->missing) {
			fs_devices->missing_devices--;
			device->missing = 0;
		}
427 428 429 430 431 432 433 434 435 436
	}

	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
437 438 439 440 441 442 443 444 445 446 447 448 449
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);
450
	mutex_init(&fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
451 452 453 454
	fs_devices->latest_devid = orig->latest_devid;
	fs_devices->latest_trans = orig->latest_trans;
	memcpy(fs_devices->fsid, orig->fsid, sizeof(fs_devices->fsid));

455
	mutex_lock(&orig->device_list_mutex);
Y
Yan Zheng 已提交
456 457 458 459 460 461
	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
462 463
		if (!device->name) {
			kfree(device);
Y
Yan Zheng 已提交
464
			goto error;
J
Julia Lawall 已提交
465
		}
Y
Yan Zheng 已提交
466 467 468 469 470 471 472 473 474 475 476 477 478

		device->devid = orig_dev->devid;
		device->work.func = pending_bios_fn;
		memcpy(device->uuid, orig_dev->uuid, sizeof(device->uuid));
		device->barriers = 1;
		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++;
	}
479
	mutex_unlock(&orig->device_list_mutex);
Y
Yan Zheng 已提交
480 481
	return fs_devices;
error:
482
	mutex_unlock(&orig->device_list_mutex);
Y
Yan Zheng 已提交
483 484 485 486
	free_fs_devices(fs_devices);
	return ERR_PTR(-ENOMEM);
}

487 488
int btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices)
{
Q
Qinghuang Feng 已提交
489
	struct btrfs_device *device, *next;
490 491 492

	mutex_lock(&uuid_mutex);
again:
493
	mutex_lock(&fs_devices->device_list_mutex);
Q
Qinghuang Feng 已提交
494
	list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
Y
Yan Zheng 已提交
495 496 497 498
		if (device->in_fs_metadata)
			continue;

		if (device->bdev) {
499
			close_bdev_exclusive(device->bdev, device->mode);
Y
Yan Zheng 已提交
500 501 502 503 504 505 506 507
			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
508 509 510 511
		list_del_init(&device->dev_list);
		fs_devices->num_devices--;
		kfree(device->name);
		kfree(device);
512
	}
513
	mutex_unlock(&fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
514 515 516 517 518 519

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

520 521 522
	mutex_unlock(&uuid_mutex);
	return 0;
}
523

Y
Yan Zheng 已提交
524
static int __btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
525 526
{
	struct btrfs_device *device;
Y
Yan Zheng 已提交
527

Y
Yan Zheng 已提交
528 529
	if (--fs_devices->opened > 0)
		return 0;
530

Q
Qinghuang Feng 已提交
531
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
532
		if (device->bdev) {
533
			close_bdev_exclusive(device->bdev, device->mode);
534
			fs_devices->open_devices--;
535
		}
Y
Yan Zheng 已提交
536 537 538 539 540
		if (device->writeable) {
			list_del_init(&device->dev_alloc_list);
			fs_devices->rw_devices--;
		}

541
		device->bdev = NULL;
Y
Yan Zheng 已提交
542
		device->writeable = 0;
543
		device->in_fs_metadata = 0;
544
	}
Y
Yan Zheng 已提交
545 546
	WARN_ON(fs_devices->open_devices);
	WARN_ON(fs_devices->rw_devices);
Y
Yan Zheng 已提交
547 548 549
	fs_devices->opened = 0;
	fs_devices->seeding = 0;

550 551 552
	return 0;
}

Y
Yan Zheng 已提交
553 554
int btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
{
Y
Yan Zheng 已提交
555
	struct btrfs_fs_devices *seed_devices = NULL;
Y
Yan Zheng 已提交
556 557 558 559
	int ret;

	mutex_lock(&uuid_mutex);
	ret = __btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
560 561 562 563
	if (!fs_devices->opened) {
		seed_devices = fs_devices->seed;
		fs_devices->seed = NULL;
	}
Y
Yan Zheng 已提交
564
	mutex_unlock(&uuid_mutex);
Y
Yan Zheng 已提交
565 566 567 568 569 570 571

	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
572 573 574
	return ret;
}

Y
Yan Zheng 已提交
575 576
static int __btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
				fmode_t flags, void *holder)
577 578 579 580
{
	struct block_device *bdev;
	struct list_head *head = &fs_devices->devices;
	struct btrfs_device *device;
581 582 583 584 585 586
	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
587
	int seeding = 1;
588
	int ret = 0;
589

Q
Qinghuang Feng 已提交
590
	list_for_each_entry(device, head, dev_list) {
591 592
		if (device->bdev)
			continue;
593 594 595
		if (!device->name)
			continue;

596
		bdev = open_bdev_exclusive(device->name, flags, holder);
597
		if (IS_ERR(bdev)) {
C
Chris Mason 已提交
598
			printk(KERN_INFO "open %s failed\n", device->name);
599
			goto error;
600
		}
601
		set_blocksize(bdev, 4096);
602

Y
Yan Zheng 已提交
603
		bh = btrfs_read_dev_super(bdev);
604 605
		if (!bh) {
			ret = -EINVAL;
606
			goto error_close;
607
		}
608 609

		disk_super = (struct btrfs_super_block *)bh->b_data;
610
		devid = btrfs_stack_device_id(&disk_super->dev_item);
611 612 613
		if (devid != device->devid)
			goto error_brelse;

Y
Yan Zheng 已提交
614 615 616 617 618 619
		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) {
620
			latest_devid = devid;
Y
Yan Zheng 已提交
621
			latest_transid = device->generation;
622 623 624
			latest_bdev = bdev;
		}

Y
Yan Zheng 已提交
625 626 627 628 629 630 631
		if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_SEEDING) {
			device->writeable = 0;
		} else {
			device->writeable = !bdev_read_only(bdev);
			seeding = 0;
		}

632
		device->bdev = bdev;
633
		device->in_fs_metadata = 0;
634 635
		device->mode = flags;

C
Chris Mason 已提交
636 637 638
		if (!blk_queue_nonrot(bdev_get_queue(bdev)))
			fs_devices->rotating = 1;

639
		fs_devices->open_devices++;
Y
Yan Zheng 已提交
640 641 642 643 644
		if (device->writeable) {
			fs_devices->rw_devices++;
			list_add(&device->dev_alloc_list,
				 &fs_devices->alloc_list);
		}
645
		continue;
646

647 648 649
error_brelse:
		brelse(bh);
error_close:
650
		close_bdev_exclusive(bdev, FMODE_READ);
651 652
error:
		continue;
653
	}
654 655 656 657
	if (fs_devices->open_devices == 0) {
		ret = -EIO;
		goto out;
	}
Y
Yan Zheng 已提交
658 659
	fs_devices->seeding = seeding;
	fs_devices->opened = 1;
660 661 662
	fs_devices->latest_bdev = latest_bdev;
	fs_devices->latest_devid = latest_devid;
	fs_devices->latest_trans = latest_transid;
Y
Yan Zheng 已提交
663
	fs_devices->total_rw_bytes = 0;
664
out:
Y
Yan Zheng 已提交
665 666 667 668
	return ret;
}

int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
669
		       fmode_t flags, void *holder)
Y
Yan Zheng 已提交
670 671 672 673 674
{
	int ret;

	mutex_lock(&uuid_mutex);
	if (fs_devices->opened) {
Y
Yan Zheng 已提交
675 676
		fs_devices->opened++;
		ret = 0;
Y
Yan Zheng 已提交
677
	} else {
678
		ret = __btrfs_open_devices(fs_devices, flags, holder);
Y
Yan Zheng 已提交
679
	}
680 681 682 683
	mutex_unlock(&uuid_mutex);
	return ret;
}

684
int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
685 686 687 688 689 690 691
			  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;
692
	u64 transid;
693 694 695

	mutex_lock(&uuid_mutex);

696
	bdev = open_bdev_exclusive(path, flags, holder);
697 698 699 700 701 702 703 704 705

	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
706
	bh = btrfs_read_dev_super(bdev);
707
	if (!bh) {
708
		ret = -EINVAL;
709 710 711
		goto error_close;
	}
	disk_super = (struct btrfs_super_block *)bh->b_data;
712
	devid = btrfs_stack_device_id(&disk_super->dev_item);
713
	transid = btrfs_super_generation(disk_super);
714
	if (disk_super->label[0])
C
Chris Mason 已提交
715
		printk(KERN_INFO "device label %s ", disk_super->label);
716 717
	else {
		/* FIXME, make a readl uuid parser */
C
Chris Mason 已提交
718
		printk(KERN_INFO "device fsid %llx-%llx ",
719 720 721
		       *(unsigned long long *)disk_super->fsid,
		       *(unsigned long long *)(disk_super->fsid + 8));
	}
722
	printk(KERN_CONT "devid %llu transid %llu %s\n",
C
Chris Mason 已提交
723
	       (unsigned long long)devid, (unsigned long long)transid, path);
724 725 726 727
	ret = device_list_add(path, disk_super, devid, fs_devices_ret);

	brelse(bh);
error_close:
728
	close_bdev_exclusive(bdev, flags);
729 730 731 732
error:
	mutex_unlock(&uuid_mutex);
	return ret;
}
733

734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817
/* 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;
}

818
/*
819 820 821 822 823 824 825 826
 * 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
 *
827 828 829
 * 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
830 831 832 833 834 835 836 837
 *
 * @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.
838
 */
839 840
int find_free_dev_extent(struct btrfs_trans_handle *trans,
			 struct btrfs_device *device, u64 num_bytes,
841
			 u64 *start, u64 *len)
842 843 844
{
	struct btrfs_key key;
	struct btrfs_root *root = device->dev_root;
845
	struct btrfs_dev_extent *dev_extent;
Y
Yan Zheng 已提交
846
	struct btrfs_path *path;
847 848 849 850 851
	u64 hole_size;
	u64 max_hole_start;
	u64 max_hole_size;
	u64 extent_end;
	u64 search_start;
852 853
	u64 search_end = device->total_bytes;
	int ret;
854
	int slot;
855 856 857 858
	struct extent_buffer *l;

	/* FIXME use last free of some kind */

859 860 861
	/* we don't want to overwrite the superblock on the drive,
	 * so we make sure to start at an offset of at least 1MB
	 */
862
	search_start = 1024 * 1024;
863

864
	if (root->fs_info->alloc_start + num_bytes <= search_end)
865 866
		search_start = max(root->fs_info->alloc_start, search_start);

867 868 869 870 871 872 873 874 875 876 877 878 879 880 881
	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;

882 883 884
	key.objectid = device->devid;
	key.offset = search_start;
	key.type = BTRFS_DEV_EXTENT_KEY;
885

886 887
	ret = btrfs_search_slot(trans, root, &key, path, 0, 0);
	if (ret < 0)
888
		goto out;
889 890 891
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid, key.type);
		if (ret < 0)
892
			goto out;
893
	}
894

895 896 897 898 899 900 901 902
	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)
903 904 905
				goto out;

			break;
906 907 908 909 910 911 912
		}
		btrfs_item_key_to_cpu(l, &key, slot);

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

		if (key.objectid > device->devid)
913
			break;
914

915 916
		if (btrfs_key_type(&key) != BTRFS_DEV_EXTENT_KEY)
			goto next;
917

918 919
		if (key.offset > search_start) {
			hole_size = key.offset - search_start;
920

921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937
			if (hole_size > max_hole_size) {
				max_hole_start = search_start;
				max_hole_size = hole_size;
			}

			/*
			 * 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;
938 939 940 941
			}
		}

		dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
942 943 944 945
		extent_end = key.offset + btrfs_dev_extent_length(l,
								  dev_extent);
		if (extent_end > search_start)
			search_start = extent_end;
946 947 948 949 950
next:
		path->slots[0]++;
		cond_resched();
	}

951 952 953 954
	hole_size = search_end- search_start;
	if (hole_size > max_hole_size) {
		max_hole_start = search_start;
		max_hole_size = hole_size;
955 956
	}

957 958 959 960 961 962 963
	/* See above. */
	if (hole_size < num_bytes)
		ret = -ENOSPC;
	else
		ret = 0;

out:
Y
Yan Zheng 已提交
964
	btrfs_free_path(path);
965 966
error:
	*start = max_hole_start;
967
	if (len)
968
		*len = max_hole_size;
969 970 971
	return ret;
}

972
static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans,
973 974 975 976 977 978 979
			  struct btrfs_device *device,
			  u64 start)
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_key key;
980 981 982
	struct btrfs_key found_key;
	struct extent_buffer *leaf = NULL;
	struct btrfs_dev_extent *extent = NULL;
983 984 985 986 987 988 989 990 991 992

	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);
993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid,
					  BTRFS_DEV_EXTENT_KEY);
		BUG_ON(ret);
		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);
		ret = 0;
	} else if (ret == 0) {
		leaf = path->nodes[0];
		extent = btrfs_item_ptr(leaf, path->slots[0],
					struct btrfs_dev_extent);
	}
1009 1010
	BUG_ON(ret);

1011 1012
	if (device->bytes_used > 0)
		device->bytes_used -= btrfs_dev_extent_length(leaf, extent);
1013 1014 1015 1016 1017 1018 1019
	ret = btrfs_del_item(trans, root, path);
	BUG_ON(ret);

	btrfs_free_path(path);
	return ret;
}

Y
Yan Zheng 已提交
1020
int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
1021
			   struct btrfs_device *device,
1022
			   u64 chunk_tree, u64 chunk_objectid,
Y
Yan Zheng 已提交
1023
			   u64 chunk_offset, u64 start, u64 num_bytes)
1024 1025 1026 1027 1028 1029 1030 1031
{
	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;

1032
	WARN_ON(!device->in_fs_metadata);
1033 1034 1035 1036 1037
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = device->devid;
Y
Yan Zheng 已提交
1038
	key.offset = start;
1039 1040 1041 1042 1043 1044 1045 1046
	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);
1047 1048 1049 1050 1051 1052 1053 1054
	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);

1055 1056 1057 1058 1059 1060
	btrfs_set_dev_extent_length(leaf, extent, num_bytes);
	btrfs_mark_buffer_dirty(leaf);
	btrfs_free_path(path);
	return ret;
}

1061 1062
static noinline int find_next_chunk(struct btrfs_root *root,
				    u64 objectid, u64 *offset)
1063 1064 1065 1066
{
	struct btrfs_path *path;
	int ret;
	struct btrfs_key key;
1067
	struct btrfs_chunk *chunk;
1068 1069 1070 1071 1072
	struct btrfs_key found_key;

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

1073
	key.objectid = objectid;
1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084
	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) {
1085
		*offset = 0;
1086 1087 1088
	} else {
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
1089 1090 1091 1092 1093 1094 1095 1096
		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);
		}
1097 1098 1099 1100 1101 1102 1103
	}
	ret = 0;
error:
	btrfs_free_path(path);
	return ret;
}

Y
Yan Zheng 已提交
1104
static noinline int find_next_devid(struct btrfs_root *root, u64 *objectid)
1105 1106 1107 1108
{
	int ret;
	struct btrfs_key key;
	struct btrfs_key found_key;
Y
Yan Zheng 已提交
1109 1110 1111 1112 1113 1114 1115
	struct btrfs_path *path;

	root = root->fs_info->chunk_root;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137

	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
1138
	btrfs_free_path(path);
1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164
	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
1165
	key.offset = device->devid;
1166 1167

	ret = btrfs_insert_empty_item(trans, root, path, &key,
1168
				      sizeof(*dev_item));
1169 1170 1171 1172 1173 1174 1175
	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
1176
	btrfs_set_device_generation(leaf, dev_item, 0);
1177 1178 1179 1180 1181 1182
	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);
1183 1184 1185
	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);
1186
	btrfs_set_device_start_offset(leaf, dev_item, 0);
1187 1188

	ptr = (unsigned long)btrfs_device_uuid(dev_item);
1189
	write_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
1190 1191
	ptr = (unsigned long)btrfs_device_fsid(dev_item);
	write_extent_buffer(leaf, root->fs_info->fsid, ptr, BTRFS_UUID_SIZE);
1192 1193
	btrfs_mark_buffer_dirty(leaf);

Y
Yan Zheng 已提交
1194
	ret = 0;
1195 1196 1197 1198
out:
	btrfs_free_path(path);
	return ret;
}
1199

1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213
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;

1214
	trans = btrfs_start_transaction(root, 0);
1215 1216 1217 1218
	if (IS_ERR(trans)) {
		btrfs_free_path(path);
		return PTR_ERR(trans);
	}
1219 1220 1221
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.type = BTRFS_DEV_ITEM_KEY;
	key.offset = device->devid;
1222
	lock_chunks(root);
1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237

	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);
1238
	unlock_chunks(root);
1239 1240 1241 1242 1243 1244 1245
	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
1246
	struct btrfs_device *next_device;
1247
	struct block_device *bdev;
1248
	struct buffer_head *bh = NULL;
1249 1250 1251
	struct btrfs_super_block *disk_super;
	u64 all_avail;
	u64 devid;
Y
Yan Zheng 已提交
1252 1253
	u64 num_devices;
	u8 *dev_uuid;
1254 1255 1256
	int ret = 0;

	mutex_lock(&uuid_mutex);
1257
	mutex_lock(&root->fs_info->volume_mutex);
1258 1259 1260 1261 1262 1263

	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) &&
1264
	    root->fs_info->fs_devices->num_devices <= 4) {
C
Chris Mason 已提交
1265 1266
		printk(KERN_ERR "btrfs: unable to go below four devices "
		       "on raid10\n");
1267 1268 1269 1270 1271
		ret = -EINVAL;
		goto out;
	}

	if ((all_avail & BTRFS_BLOCK_GROUP_RAID1) &&
1272
	    root->fs_info->fs_devices->num_devices <= 2) {
C
Chris Mason 已提交
1273 1274
		printk(KERN_ERR "btrfs: unable to go below two "
		       "devices on raid1\n");
1275 1276 1277 1278
		ret = -EINVAL;
		goto out;
	}

1279 1280 1281
	if (strcmp(device_path, "missing") == 0) {
		struct list_head *devices;
		struct btrfs_device *tmp;
1282

1283 1284
		device = NULL;
		devices = &root->fs_info->fs_devices->devices;
1285
		mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
Q
Qinghuang Feng 已提交
1286
		list_for_each_entry(tmp, devices, dev_list) {
1287 1288 1289 1290 1291
			if (tmp->in_fs_metadata && !tmp->bdev) {
				device = tmp;
				break;
			}
		}
1292
		mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
1293 1294 1295 1296
		bdev = NULL;
		bh = NULL;
		disk_super = NULL;
		if (!device) {
C
Chris Mason 已提交
1297 1298
			printk(KERN_ERR "btrfs: no missing devices found to "
			       "remove\n");
1299 1300 1301
			goto out;
		}
	} else {
1302
		bdev = open_bdev_exclusive(device_path, FMODE_READ,
1303 1304 1305 1306 1307
				      root->fs_info->bdev_holder);
		if (IS_ERR(bdev)) {
			ret = PTR_ERR(bdev);
			goto out;
		}
1308

Y
Yan Zheng 已提交
1309
		set_blocksize(bdev, 4096);
Y
Yan Zheng 已提交
1310
		bh = btrfs_read_dev_super(bdev);
1311
		if (!bh) {
1312
			ret = -EINVAL;
1313 1314 1315
			goto error_close;
		}
		disk_super = (struct btrfs_super_block *)bh->b_data;
1316
		devid = btrfs_stack_device_id(&disk_super->dev_item);
Y
Yan Zheng 已提交
1317 1318 1319
		dev_uuid = disk_super->dev_item.uuid;
		device = btrfs_find_device(root, devid, dev_uuid,
					   disk_super->fsid);
1320 1321 1322 1323
		if (!device) {
			ret = -ENOENT;
			goto error_brelse;
		}
Y
Yan Zheng 已提交
1324
	}
1325

Y
Yan Zheng 已提交
1326
	if (device->writeable && root->fs_info->fs_devices->rw_devices == 1) {
C
Chris Mason 已提交
1327 1328
		printk(KERN_ERR "btrfs: unable to remove the only writeable "
		       "device\n");
Y
Yan Zheng 已提交
1329 1330 1331 1332 1333 1334 1335
		ret = -EINVAL;
		goto error_brelse;
	}

	if (device->writeable) {
		list_del_init(&device->dev_alloc_list);
		root->fs_info->fs_devices->rw_devices--;
1336
	}
1337 1338 1339 1340 1341 1342 1343 1344 1345

	ret = btrfs_shrink_device(device, 0);
	if (ret)
		goto error_brelse;

	ret = btrfs_rm_dev_item(root->fs_info->chunk_root, device);
	if (ret)
		goto error_brelse;

Y
Yan Zheng 已提交
1346
	device->in_fs_metadata = 0;
1347 1348 1349 1350 1351 1352 1353

	/*
	 * the device list mutex makes sure that we don't change
	 * the device list while someone else is writing out all
	 * the device supers.
	 */
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
1354
	list_del_init(&device->dev_list);
1355 1356
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
1357
	device->fs_devices->num_devices--;
Y
Yan Zheng 已提交
1358

1359 1360 1361
	if (device->missing)
		root->fs_info->fs_devices->missing_devices--;

Y
Yan Zheng 已提交
1362 1363 1364 1365 1366 1367 1368
	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;

Y
Yan Zheng 已提交
1369 1370 1371 1372 1373 1374
	if (device->bdev) {
		close_bdev_exclusive(device->bdev, device->mode);
		device->bdev = NULL;
		device->fs_devices->open_devices--;
	}

Y
Yan Zheng 已提交
1375 1376 1377
	num_devices = btrfs_super_num_devices(&root->fs_info->super_copy) - 1;
	btrfs_set_super_num_devices(&root->fs_info->super_copy, num_devices);

Y
Yan Zheng 已提交
1378 1379 1380 1381 1382 1383 1384
	if (device->fs_devices->open_devices == 0) {
		struct btrfs_fs_devices *fs_devices;
		fs_devices = root->fs_info->fs_devices;
		while (fs_devices) {
			if (fs_devices->seed == device->fs_devices)
				break;
			fs_devices = fs_devices->seed;
Y
Yan Zheng 已提交
1385
		}
Y
Yan Zheng 已提交
1386 1387 1388 1389
		fs_devices->seed = device->fs_devices->seed;
		device->fs_devices->seed = NULL;
		__btrfs_close_devices(device->fs_devices);
		free_fs_devices(device->fs_devices);
Y
Yan Zheng 已提交
1390 1391 1392 1393 1394 1395 1396
	}

	/*
	 * at this point, the device is zero sized.  We want to
	 * remove it from the devices list and zero out the old super
	 */
	if (device->writeable) {
1397 1398 1399 1400 1401 1402 1403
		/* 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);
	}
1404 1405 1406 1407 1408 1409 1410 1411

	kfree(device->name);
	kfree(device);
	ret = 0;

error_brelse:
	brelse(bh);
error_close:
1412
	if (bdev)
1413
		close_bdev_exclusive(bdev, FMODE_READ);
1414
out:
1415
	mutex_unlock(&root->fs_info->volume_mutex);
1416 1417 1418 1419
	mutex_unlock(&uuid_mutex);
	return ret;
}

Y
Yan Zheng 已提交
1420 1421 1422 1423 1424 1425 1426 1427
/*
 * 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
1428
	struct btrfs_fs_devices *seed_devices;
Y
Yan Zheng 已提交
1429 1430 1431 1432 1433
	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
1434
	if (!fs_devices->seeding)
Y
Yan Zheng 已提交
1435 1436
		return -EINVAL;

Y
Yan Zheng 已提交
1437 1438
	seed_devices = kzalloc(sizeof(*fs_devices), GFP_NOFS);
	if (!seed_devices)
Y
Yan Zheng 已提交
1439 1440
		return -ENOMEM;

Y
Yan Zheng 已提交
1441 1442 1443 1444
	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
1445
	}
Y
Yan Zheng 已提交
1446

Y
Yan Zheng 已提交
1447 1448
	list_add(&old_devices->list, &fs_uuids);

Y
Yan Zheng 已提交
1449 1450 1451 1452
	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);
1453
	mutex_init(&seed_devices->device_list_mutex);
Y
Yan Zheng 已提交
1454 1455 1456 1457 1458 1459
	list_splice_init(&fs_devices->devices, &seed_devices->devices);
	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
1460 1461 1462
	fs_devices->seeding = 0;
	fs_devices->num_devices = 0;
	fs_devices->open_devices = 0;
Y
Yan Zheng 已提交
1463
	fs_devices->seed = seed_devices;
Y
Yan Zheng 已提交
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 1511 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 1547 1548 1549 1550

	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]);
			btrfs_release_path(root, path);
			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;
}

1551 1552 1553 1554 1555 1556
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
1557
	struct super_block *sb = root->fs_info->sb;
1558
	u64 total_bytes;
Y
Yan Zheng 已提交
1559
	int seeding_dev = 0;
1560 1561
	int ret = 0;

Y
Yan Zheng 已提交
1562 1563
	if ((sb->s_flags & MS_RDONLY) && !root->fs_info->fs_devices->seeding)
		return -EINVAL;
1564

1565
	bdev = open_bdev_exclusive(device_path, 0, root->fs_info->bdev_holder);
1566 1567
	if (IS_ERR(bdev))
		return PTR_ERR(bdev);
1568

Y
Yan Zheng 已提交
1569 1570 1571 1572 1573 1574
	if (root->fs_info->fs_devices->seeding) {
		seeding_dev = 1;
		down_write(&sb->s_umount);
		mutex_lock(&uuid_mutex);
	}

1575
	filemap_write_and_wait(bdev->bd_inode->i_mapping);
1576
	mutex_lock(&root->fs_info->volume_mutex);
1577

1578
	devices = &root->fs_info->fs_devices->devices;
1579 1580 1581 1582
	/*
	 * 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
1583
	list_for_each_entry(device, devices, dev_list) {
1584 1585
		if (device->bdev == bdev) {
			ret = -EEXIST;
Y
Yan Zheng 已提交
1586
			goto error;
1587 1588 1589 1590 1591 1592 1593
		}
	}

	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
1594
		goto error;
1595 1596 1597 1598 1599
	}

	device->name = kstrdup(device_path, GFP_NOFS);
	if (!device->name) {
		kfree(device);
Y
Yan Zheng 已提交
1600 1601
		ret = -ENOMEM;
		goto error;
1602
	}
Y
Yan Zheng 已提交
1603 1604 1605 1606 1607 1608 1609

	ret = find_next_devid(root, &device->devid);
	if (ret) {
		kfree(device);
		goto error;
	}

1610
	trans = btrfs_start_transaction(root, 0);
1611 1612 1613 1614 1615 1616
	if (IS_ERR(trans)) {
		kfree(device);
		ret = PTR_ERR(trans);
		goto error;
	}

Y
Yan Zheng 已提交
1617 1618 1619 1620 1621 1622 1623 1624
	lock_chunks(root);

	device->barriers = 1;
	device->writeable = 1;
	device->work.func = pending_bios_fn;
	generate_random_uuid(device->uuid);
	spin_lock_init(&device->io_lock);
	device->generation = trans->transid;
1625 1626 1627 1628
	device->io_width = root->sectorsize;
	device->io_align = root->sectorsize;
	device->sector_size = root->sectorsize;
	device->total_bytes = i_size_read(bdev->bd_inode);
1629
	device->disk_total_bytes = device->total_bytes;
1630 1631
	device->dev_root = root->fs_info->dev_root;
	device->bdev = bdev;
1632
	device->in_fs_metadata = 1;
1633
	device->mode = 0;
Y
Yan Zheng 已提交
1634
	set_blocksize(device->bdev, 4096);
1635

Y
Yan Zheng 已提交
1636 1637 1638 1639 1640
	if (seeding_dev) {
		sb->s_flags &= ~MS_RDONLY;
		ret = btrfs_prepare_sprout(trans, root);
		BUG_ON(ret);
	}
1641

Y
Yan Zheng 已提交
1642
	device->fs_devices = root->fs_info->fs_devices;
1643 1644 1645 1646 1647 1648

	/*
	 * 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);
Y
Yan Zheng 已提交
1649 1650 1651 1652 1653 1654 1655
	list_add(&device->dev_list, &root->fs_info->fs_devices->devices);
	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;
1656

C
Chris Mason 已提交
1657 1658 1659
	if (!blk_queue_nonrot(bdev_get_queue(bdev)))
		root->fs_info->fs_devices->rotating = 1;

1660 1661 1662 1663 1664 1665 1666
	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);
1667
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
1668

Y
Yan Zheng 已提交
1669 1670 1671 1672 1673 1674 1675 1676 1677
	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);
	}

1678 1679 1680 1681 1682 1683
	/*
	 * we've got more storage, clear any full flags on the space
	 * infos
	 */
	btrfs_clear_space_info_full(root->fs_info);

1684
	unlock_chunks(root);
Y
Yan Zheng 已提交
1685
	btrfs_commit_transaction(trans, root);
1686

Y
Yan Zheng 已提交
1687 1688 1689
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
1690

Y
Yan Zheng 已提交
1691 1692 1693 1694 1695 1696 1697
		ret = btrfs_relocate_sys_chunks(root);
		BUG_ON(ret);
	}
out:
	mutex_unlock(&root->fs_info->volume_mutex);
	return ret;
error:
1698
	close_bdev_exclusive(bdev, 0);
Y
Yan Zheng 已提交
1699 1700 1701 1702
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
	}
1703 1704 1705
	goto out;
}

C
Chris Mason 已提交
1706 1707
static noinline int btrfs_update_device(struct btrfs_trans_handle *trans,
					struct btrfs_device *device)
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 1741 1742
{
	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);
1743
	btrfs_set_device_total_bytes(leaf, dev_item, device->disk_total_bytes);
1744 1745 1746 1747 1748 1749 1750 1751
	btrfs_set_device_bytes_used(leaf, dev_item, device->bytes_used);
	btrfs_mark_buffer_dirty(leaf);

out:
	btrfs_free_path(path);
	return ret;
}

1752
static int __btrfs_grow_device(struct btrfs_trans_handle *trans,
1753 1754 1755 1756 1757 1758 1759
		      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
1760 1761 1762 1763 1764
	if (!device->writeable)
		return -EACCES;
	if (new_size <= device->total_bytes)
		return -EINVAL;

1765
	btrfs_set_super_total_bytes(super_copy, old_total + diff);
Y
Yan Zheng 已提交
1766 1767 1768
	device->fs_devices->total_rw_bytes += diff;

	device->total_bytes = new_size;
1769
	device->disk_total_bytes = new_size;
1770 1771
	btrfs_clear_space_info_full(device->dev_root->fs_info);

1772 1773 1774
	return btrfs_update_device(trans, device);
}

1775 1776 1777 1778 1779 1780 1781 1782 1783 1784
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;
}

1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812
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);
	BUG_ON(ret);

	btrfs_free_path(path);
	return 0;
}

1813
static int btrfs_del_sys_chunk(struct btrfs_root *root, u64 chunk_objectid, u64
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 1856 1857 1858
			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;
}

1859
static int btrfs_relocate_chunk(struct btrfs_root *root,
1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874
			 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;

1875 1876 1877 1878
	ret = btrfs_can_relocate(extent_root, chunk_offset);
	if (ret)
		return -ENOSPC;

1879
	/* step one, relocate all the extents inside this chunk */
Z
Zheng Yan 已提交
1880
	ret = btrfs_relocate_block_group(extent_root, chunk_offset);
1881 1882
	if (ret)
		return ret;
1883

1884
	trans = btrfs_start_transaction(root, 0);
1885
	BUG_ON(IS_ERR(trans));
1886

1887 1888
	lock_chunks(root);

1889 1890 1891 1892
	/*
	 * step two, delete the device extents and the
	 * chunk tree entries
	 */
1893
	read_lock(&em_tree->lock);
1894
	em = lookup_extent_mapping(em_tree, chunk_offset, 1);
1895
	read_unlock(&em_tree->lock);
1896

1897 1898
	BUG_ON(em->start > chunk_offset ||
	       em->start + em->len < chunk_offset);
1899 1900 1901 1902 1903 1904
	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);
1905

1906 1907 1908 1909
		if (map->stripes[i].dev) {
			ret = btrfs_update_device(trans, map->stripes[i].dev);
			BUG_ON(ret);
		}
1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920
	}
	ret = btrfs_free_chunk(trans, root, chunk_tree, chunk_objectid,
			       chunk_offset);

	BUG_ON(ret);

	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
1921 1922 1923
	ret = btrfs_remove_block_group(trans, extent_root, chunk_offset);
	BUG_ON(ret);

1924
	write_lock(&em_tree->lock);
Y
Yan Zheng 已提交
1925
	remove_extent_mapping(em_tree, em);
1926
	write_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950

	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;
1951 1952
	bool retried = false;
	int failed = 0;
Y
Yan Zheng 已提交
1953 1954 1955 1956 1957 1958
	int ret;

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

1959
again:
Y
Yan Zheng 已提交
1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975
	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
1976

Y
Yan Zheng 已提交
1977 1978
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
Z
Zheng Yan 已提交
1979

Y
Yan Zheng 已提交
1980 1981 1982 1983
		chunk = btrfs_item_ptr(leaf, path->slots[0],
				       struct btrfs_chunk);
		chunk_type = btrfs_chunk_type(leaf, chunk);
		btrfs_release_path(chunk_root, path);
1984

Y
Yan Zheng 已提交
1985 1986 1987 1988
		if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) {
			ret = btrfs_relocate_chunk(chunk_root, chunk_tree,
						   found_key.objectid,
						   found_key.offset);
1989 1990 1991 1992
			if (ret == -ENOSPC)
				failed++;
			else if (ret)
				BUG();
Y
Yan Zheng 已提交
1993
		}
1994

Y
Yan Zheng 已提交
1995 1996 1997 1998 1999
		if (found_key.offset == 0)
			break;
		key.offset = found_key.offset - 1;
	}
	ret = 0;
2000 2001 2002 2003 2004 2005 2006 2007
	if (failed && !retried) {
		failed = 0;
		retried = true;
		goto again;
	} else if (failed && retried) {
		WARN_ON(1);
		ret = -ENOSPC;
	}
Y
Yan Zheng 已提交
2008 2009 2010
error:
	btrfs_free_path(path);
	return ret;
2011 2012
}

2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034
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
2035 2036
	if (dev_root->fs_info->sb->s_flags & MS_RDONLY)
		return -EROFS;
2037

2038 2039 2040
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

2041
	mutex_lock(&dev_root->fs_info->volume_mutex);
2042 2043 2044
	dev_root = dev_root->fs_info->dev_root;

	/* step one make some room on all the devices */
Q
Qinghuang Feng 已提交
2045
	list_for_each_entry(device, devices, dev_list) {
2046 2047 2048
		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
2049 2050
		if (!device->writeable ||
		    device->total_bytes - device->bytes_used > size_to_free)
2051 2052 2053
			continue;

		ret = btrfs_shrink_device(device, old_size - size_to_free);
2054 2055
		if (ret == -ENOSPC)
			break;
2056 2057
		BUG_ON(ret);

2058
		trans = btrfs_start_transaction(dev_root, 0);
2059
		BUG_ON(IS_ERR(trans));
2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074

		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
2075
	while (1) {
2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088
		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);
2089
		if (ret)
2090
			break;
2091

2092 2093 2094 2095
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
		if (found_key.objectid != key.objectid)
			break;
2096

2097
		/* chunk zero is special */
2098
		if (found_key.offset == 0)
2099 2100
			break;

2101
		btrfs_release_path(chunk_root, path);
2102 2103 2104 2105
		ret = btrfs_relocate_chunk(chunk_root,
					   chunk_root->root_key.objectid,
					   found_key.objectid,
					   found_key.offset);
2106 2107
		BUG_ON(ret && ret != -ENOSPC);
		key.offset = found_key.offset - 1;
2108 2109 2110 2111
	}
	ret = 0;
error:
	btrfs_free_path(path);
2112
	mutex_unlock(&dev_root->fs_info->volume_mutex);
2113 2114 2115
	return ret;
}

2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132
/*
 * 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;
2133 2134
	int failed = 0;
	bool retried = false;
2135 2136 2137 2138
	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);
2139
	u64 old_size = device->total_bytes;
2140 2141
	u64 diff = device->total_bytes - new_size;

Y
Yan Zheng 已提交
2142 2143
	if (new_size >= device->total_bytes)
		return -EINVAL;
2144 2145 2146 2147 2148 2149 2150

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

	path->reada = 2;

2151 2152
	lock_chunks(root);

2153
	device->total_bytes = new_size;
Y
Yan Zheng 已提交
2154 2155
	if (device->writeable)
		device->fs_devices->total_rw_bytes -= diff;
2156
	unlock_chunks(root);
2157

2158
again:
2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172
	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;
2173
			btrfs_release_path(root, path);
2174
			break;
2175 2176 2177 2178 2179 2180
		}

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

2181 2182
		if (key.objectid != device->devid) {
			btrfs_release_path(root, path);
2183
			break;
2184
		}
2185 2186 2187 2188

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

2189 2190
		if (key.offset + length <= new_size) {
			btrfs_release_path(root, path);
2191
			break;
2192
		}
2193 2194 2195 2196 2197 2198 2199 2200

		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);
		btrfs_release_path(root, path);

		ret = btrfs_relocate_chunk(root, chunk_tree, chunk_objectid,
					   chunk_offset);
2201
		if (ret && ret != -ENOSPC)
2202
			goto done;
2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220
		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;
2221 2222
	}

2223
	/* Shrinking succeeded, else we would be at "done". */
2224
	trans = btrfs_start_transaction(root, 0);
2225 2226 2227 2228 2229
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto done;
	}

2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243
	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);
2244 2245 2246 2247 2248
done:
	btrfs_free_path(path);
	return ret;
}

2249
static int btrfs_add_system_chunk(struct btrfs_trans_handle *trans,
2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272
			   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;
}

C
Chris Mason 已提交
2273
static noinline u64 chunk_bytes_by_type(u64 type, u64 calc_size,
2274
					int num_stripes, int sub_stripes)
2275 2276 2277 2278 2279 2280 2281 2282 2283
{
	if (type & (BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_DUP))
		return calc_size;
	else if (type & BTRFS_BLOCK_GROUP_RAID10)
		return calc_size * (num_stripes / sub_stripes);
	else
		return calc_size * num_stripes;
}

2284 2285
/* Used to sort the devices by max_avail(descending sort) */
int btrfs_cmp_device_free_bytes(const void *dev_info1, const void *dev_info2)
2286
{
2287 2288 2289 2290 2291 2292 2293 2294 2295
	if (((struct btrfs_device_info *)dev_info1)->max_avail >
	    ((struct btrfs_device_info *)dev_info2)->max_avail)
		return -1;
	else if (((struct btrfs_device_info *)dev_info1)->max_avail <
		 ((struct btrfs_device_info *)dev_info2)->max_avail)
		return 1;
	else
		return 0;
}
2296

2297 2298 2299 2300 2301 2302 2303
static int __btrfs_calc_nstripes(struct btrfs_fs_devices *fs_devices, u64 type,
				 int *num_stripes, int *min_stripes,
				 int *sub_stripes)
{
	*num_stripes = 1;
	*min_stripes = 1;
	*sub_stripes = 0;
2304

2305
	if (type & (BTRFS_BLOCK_GROUP_RAID0)) {
2306 2307
		*num_stripes = fs_devices->rw_devices;
		*min_stripes = 2;
2308 2309
	}
	if (type & (BTRFS_BLOCK_GROUP_DUP)) {
2310 2311
		*num_stripes = 2;
		*min_stripes = 2;
2312
	}
2313
	if (type & (BTRFS_BLOCK_GROUP_RAID1)) {
2314
		if (fs_devices->rw_devices < 2)
2315
			return -ENOSPC;
2316 2317
		*num_stripes = 2;
		*min_stripes = 2;
2318
	}
C
Chris Mason 已提交
2319
	if (type & (BTRFS_BLOCK_GROUP_RAID10)) {
2320 2321
		*num_stripes = fs_devices->rw_devices;
		if (*num_stripes < 4)
C
Chris Mason 已提交
2322
			return -ENOSPC;
2323 2324 2325
		*num_stripes &= ~(u32)1;
		*sub_stripes = 2;
		*min_stripes = 4;
C
Chris Mason 已提交
2326
	}
2327

2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344
	return 0;
}

static u64 __btrfs_calc_stripe_size(struct btrfs_fs_devices *fs_devices,
				    u64 proposed_size, u64 type,
				    int num_stripes, int small_stripe)
{
	int min_stripe_size = 1 * 1024 * 1024;
	u64 calc_size = proposed_size;
	u64 max_chunk_size = calc_size;
	int ncopies = 1;

	if (type & (BTRFS_BLOCK_GROUP_RAID1 |
		    BTRFS_BLOCK_GROUP_DUP |
		    BTRFS_BLOCK_GROUP_RAID10))
		ncopies = 2;

2345 2346
	if (type & BTRFS_BLOCK_GROUP_DATA) {
		max_chunk_size = 10 * calc_size;
2347
		min_stripe_size = 64 * 1024 * 1024;
2348
	} else if (type & BTRFS_BLOCK_GROUP_METADATA) {
J
Josef Bacik 已提交
2349
		max_chunk_size = 256 * 1024 * 1024;
2350 2351 2352 2353 2354
		min_stripe_size = 32 * 1024 * 1024;
	} else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
		calc_size = 8 * 1024 * 1024;
		max_chunk_size = calc_size * 2;
		min_stripe_size = 1 * 1024 * 1024;
2355 2356
	}

Y
Yan Zheng 已提交
2357 2358 2359
	/* 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);
2360

2361 2362
	if (calc_size * num_stripes > max_chunk_size * ncopies) {
		calc_size = max_chunk_size * ncopies;
2363
		do_div(calc_size, num_stripes);
2364 2365
		do_div(calc_size, BTRFS_STRIPE_LEN);
		calc_size *= BTRFS_STRIPE_LEN;
2366
	}
2367

2368
	/* we don't want tiny stripes */
2369
	if (!small_stripe)
2370
		calc_size = max_t(u64, min_stripe_size, calc_size);
2371

2372
	/*
2373
	 * we're about to do_div by the BTRFS_STRIPE_LEN so lets make sure
2374 2375
	 * we end up with something bigger than a stripe
	 */
2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426
	calc_size = max_t(u64, calc_size, BTRFS_STRIPE_LEN);

	do_div(calc_size, BTRFS_STRIPE_LEN);
	calc_size *= BTRFS_STRIPE_LEN;

	return calc_size;
}

static struct map_lookup *__shrink_map_lookup_stripes(struct map_lookup *map,
						      int num_stripes)
{
	struct map_lookup *new;
	size_t len = map_lookup_size(num_stripes);

	BUG_ON(map->num_stripes < num_stripes);

	if (map->num_stripes == num_stripes)
		return map;

	new = kmalloc(len, GFP_NOFS);
	if (!new) {
		/* just change map->num_stripes */
		map->num_stripes = num_stripes;
		return map;
	}

	memcpy(new, map, len);
	new->num_stripes = num_stripes;
	kfree(map);
	return new;
}

/*
 * helper to allocate device space from btrfs_device_info, in which we stored
 * max free space information of every device. It is used when we can not
 * allocate chunks by default size.
 *
 * By this helper, we can allocate a new chunk as larger as possible.
 */
static int __btrfs_alloc_tiny_space(struct btrfs_trans_handle *trans,
				    struct btrfs_fs_devices *fs_devices,
				    struct btrfs_device_info *devices,
				    int nr_device, u64 type,
				    struct map_lookup **map_lookup,
				    int min_stripes, u64 *stripe_size)
{
	int i, index, sort_again = 0;
	int min_devices = min_stripes;
	u64 max_avail, min_free;
	struct map_lookup *map = *map_lookup;
	int ret;
2427

2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548
	if (nr_device < min_stripes)
		return -ENOSPC;

	btrfs_descending_sort_devices(devices, nr_device);

	max_avail = devices[0].max_avail;
	if (!max_avail)
		return -ENOSPC;

	for (i = 0; i < nr_device; i++) {
		/*
		 * if dev_offset = 0, it means the free space of this device
		 * is less than what we need, and we didn't search max avail
		 * extent on this device, so do it now.
		 */
		if (!devices[i].dev_offset) {
			ret = find_free_dev_extent(trans, devices[i].dev,
						   max_avail,
						   &devices[i].dev_offset,
						   &devices[i].max_avail);
			if (ret != 0 && ret != -ENOSPC)
				return ret;
			sort_again = 1;
		}
	}

	/* we update the max avail free extent of each devices, sort again */
	if (sort_again)
		btrfs_descending_sort_devices(devices, nr_device);

	if (type & BTRFS_BLOCK_GROUP_DUP)
		min_devices = 1;

	if (!devices[min_devices - 1].max_avail)
		return -ENOSPC;

	max_avail = devices[min_devices - 1].max_avail;
	if (type & BTRFS_BLOCK_GROUP_DUP)
		do_div(max_avail, 2);

	max_avail = __btrfs_calc_stripe_size(fs_devices, max_avail, type,
					     min_stripes, 1);
	if (type & BTRFS_BLOCK_GROUP_DUP)
		min_free = max_avail * 2;
	else
		min_free = max_avail;

	if (min_free > devices[min_devices - 1].max_avail)
		return -ENOSPC;

	map = __shrink_map_lookup_stripes(map, min_stripes);
	*stripe_size = max_avail;

	index = 0;
	for (i = 0; i < min_stripes; i++) {
		map->stripes[i].dev = devices[index].dev;
		map->stripes[i].physical = devices[index].dev_offset;
		if (type & BTRFS_BLOCK_GROUP_DUP) {
			i++;
			map->stripes[i].dev = devices[index].dev;
			map->stripes[i].physical = devices[index].dev_offset +
						   max_avail;
		}
		index++;
	}
	*map_lookup = map;

	return 0;
}

static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
			       struct btrfs_root *extent_root,
			       struct map_lookup **map_ret,
			       u64 *num_bytes, u64 *stripe_size,
			       u64 start, u64 type)
{
	struct btrfs_fs_info *info = extent_root->fs_info;
	struct btrfs_device *device = NULL;
	struct btrfs_fs_devices *fs_devices = info->fs_devices;
	struct list_head *cur;
	struct map_lookup *map;
	struct extent_map_tree *em_tree;
	struct extent_map *em;
	struct btrfs_device_info *devices_info;
	struct list_head private_devs;
	u64 calc_size = 1024 * 1024 * 1024;
	u64 min_free;
	u64 avail;
	u64 dev_offset;
	int num_stripes;
	int min_stripes;
	int sub_stripes;
	int min_devices;	/* the min number of devices we need */
	int i;
	int ret;
	int index;

	if ((type & BTRFS_BLOCK_GROUP_RAID1) &&
	    (type & BTRFS_BLOCK_GROUP_DUP)) {
		WARN_ON(1);
		type &= ~BTRFS_BLOCK_GROUP_DUP;
	}
	if (list_empty(&fs_devices->alloc_list))
		return -ENOSPC;

	ret = __btrfs_calc_nstripes(fs_devices, type, &num_stripes,
				    &min_stripes, &sub_stripes);
	if (ret)
		return ret;

	devices_info = kzalloc(sizeof(*devices_info) * fs_devices->rw_devices,
			       GFP_NOFS);
	if (!devices_info)
		return -ENOMEM;

	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
	if (!map) {
		ret = -ENOMEM;
		goto error;
	}
	map->num_stripes = num_stripes;
2549

Y
Yan Zheng 已提交
2550
	cur = fs_devices->alloc_list.next;
2551
	index = 0;
2552
	i = 0;
2553

2554 2555 2556 2557
	calc_size = __btrfs_calc_stripe_size(fs_devices, calc_size, type,
					     num_stripes, 0);

	if (type & BTRFS_BLOCK_GROUP_DUP) {
2558
		min_free = calc_size * 2;
2559 2560
		min_devices = 1;
	} else {
2561
		min_free = calc_size;
2562 2563
		min_devices = min_stripes;
	}
2564

Y
Yan Zheng 已提交
2565
	INIT_LIST_HEAD(&private_devs);
C
Chris Mason 已提交
2566
	while (index < num_stripes) {
2567
		device = list_entry(cur, struct btrfs_device, dev_alloc_list);
Y
Yan Zheng 已提交
2568
		BUG_ON(!device->writeable);
2569 2570 2571 2572
		if (device->total_bytes > device->bytes_used)
			avail = device->total_bytes - device->bytes_used;
		else
			avail = 0;
2573
		cur = cur->next;
2574

2575
		if (device->in_fs_metadata && avail >= min_free) {
2576 2577 2578
			ret = find_free_dev_extent(trans, device, min_free,
						   &devices_info[i].dev_offset,
						   &devices_info[i].max_avail);
2579 2580 2581
			if (ret == 0) {
				list_move_tail(&device->dev_alloc_list,
					       &private_devs);
Y
Yan Zheng 已提交
2582
				map->stripes[index].dev = device;
2583 2584
				map->stripes[index].physical =
						devices_info[i].dev_offset;
2585
				index++;
Y
Yan Zheng 已提交
2586 2587 2588
				if (type & BTRFS_BLOCK_GROUP_DUP) {
					map->stripes[index].dev = device;
					map->stripes[index].physical =
2589 2590
						devices_info[i].dev_offset +
						calc_size;
2591
					index++;
Y
Yan Zheng 已提交
2592
				}
2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604
			} else if (ret != -ENOSPC)
				goto error;

			devices_info[i].dev = device;
			i++;
		} else if (device->in_fs_metadata &&
			   avail >= BTRFS_STRIPE_LEN) {
			devices_info[i].dev = device;
			devices_info[i].max_avail = avail;
			i++;
		}

Y
Yan Zheng 已提交
2605
		if (cur == &fs_devices->alloc_list)
2606 2607
			break;
	}
2608

Y
Yan Zheng 已提交
2609
	list_splice(&private_devs, &fs_devices->alloc_list);
2610
	if (index < num_stripes) {
2611 2612 2613 2614 2615 2616
		if (index >= min_stripes) {
			num_stripes = index;
			if (type & (BTRFS_BLOCK_GROUP_RAID10)) {
				num_stripes /= sub_stripes;
				num_stripes *= sub_stripes;
			}
2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628

			map = __shrink_map_lookup_stripes(map, num_stripes);
		} else if (i >= min_devices) {
			ret = __btrfs_alloc_tiny_space(trans, fs_devices,
						       devices_info, i, type,
						       &map, min_stripes,
						       &calc_size);
			if (ret)
				goto error;
		} else {
			ret = -ENOSPC;
			goto error;
2629 2630
		}
	}
Y
Yan Zheng 已提交
2631
	map->sector_size = extent_root->sectorsize;
2632 2633 2634
	map->stripe_len = BTRFS_STRIPE_LEN;
	map->io_align = BTRFS_STRIPE_LEN;
	map->io_width = BTRFS_STRIPE_LEN;
Y
Yan Zheng 已提交
2635 2636
	map->type = type;
	map->sub_stripes = sub_stripes;
2637

Y
Yan Zheng 已提交
2638 2639 2640
	*map_ret = map;
	*stripe_size = calc_size;
	*num_bytes = chunk_bytes_by_type(type, calc_size,
2641
					 map->num_stripes, sub_stripes);
2642

Y
Yan Zheng 已提交
2643 2644
	em = alloc_extent_map(GFP_NOFS);
	if (!em) {
2645 2646
		ret = -ENOMEM;
		goto error;
2647
	}
Y
Yan Zheng 已提交
2648 2649 2650 2651 2652
	em->bdev = (struct block_device *)map;
	em->start = start;
	em->len = *num_bytes;
	em->block_start = 0;
	em->block_len = em->len;
2653

Y
Yan Zheng 已提交
2654
	em_tree = &extent_root->fs_info->mapping_tree.map_tree;
2655
	write_lock(&em_tree->lock);
Y
Yan Zheng 已提交
2656
	ret = add_extent_mapping(em_tree, em);
2657
	write_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
2658 2659
	BUG_ON(ret);
	free_extent_map(em);
2660

Y
Yan Zheng 已提交
2661 2662 2663 2664
	ret = btrfs_make_block_group(trans, extent_root, 0, type,
				     BTRFS_FIRST_CHUNK_TREE_OBJECTID,
				     start, *num_bytes);
	BUG_ON(ret);
2665

Y
Yan Zheng 已提交
2666 2667 2668 2669
	index = 0;
	while (index < map->num_stripes) {
		device = map->stripes[index].dev;
		dev_offset = map->stripes[index].physical;
2670 2671

		ret = btrfs_alloc_dev_extent(trans, device,
Y
Yan Zheng 已提交
2672 2673 2674
				info->chunk_root->root_key.objectid,
				BTRFS_FIRST_CHUNK_TREE_OBJECTID,
				start, dev_offset, calc_size);
2675
		BUG_ON(ret);
Y
Yan Zheng 已提交
2676 2677 2678
		index++;
	}

2679
	kfree(devices_info);
Y
Yan Zheng 已提交
2680
	return 0;
2681 2682 2683 2684 2685

error:
	kfree(map);
	kfree(devices_info);
	return ret;
Y
Yan Zheng 已提交
2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710
}

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;
2711 2712
		ret = btrfs_update_device(trans, device);
		BUG_ON(ret);
Y
Yan Zheng 已提交
2713 2714 2715 2716 2717 2718 2719 2720
		index++;
	}

	index = 0;
	stripe = &chunk->stripe;
	while (index < map->num_stripes) {
		device = map->stripes[index].dev;
		dev_offset = map->stripes[index].physical;
2721

2722 2723 2724
		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
2725
		stripe++;
2726 2727 2728
		index++;
	}

Y
Yan Zheng 已提交
2729
	btrfs_set_stack_chunk_length(chunk, chunk_size);
2730
	btrfs_set_stack_chunk_owner(chunk, extent_root->root_key.objectid);
Y
Yan Zheng 已提交
2731 2732 2733 2734 2735
	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);
2736
	btrfs_set_stack_chunk_sector_size(chunk, extent_root->sectorsize);
Y
Yan Zheng 已提交
2737
	btrfs_set_stack_chunk_sub_stripes(chunk, map->sub_stripes);
2738

Y
Yan Zheng 已提交
2739 2740 2741
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.type = BTRFS_CHUNK_ITEM_KEY;
	key.offset = chunk_offset;
2742

Y
Yan Zheng 已提交
2743 2744
	ret = btrfs_insert_item(trans, chunk_root, &key, chunk, item_size);
	BUG_ON(ret);
2745

Y
Yan Zheng 已提交
2746 2747 2748
	if (map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
		ret = btrfs_add_system_chunk(trans, chunk_root, &key, chunk,
					     item_size);
2749 2750
		BUG_ON(ret);
	}
2751
	kfree(chunk);
Y
Yan Zheng 已提交
2752 2753
	return 0;
}
2754

Y
Yan Zheng 已提交
2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787
/*
 * 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
2788
static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
Y
Yan Zheng 已提交
2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845
					 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);
2846
	BUG_ON(ret);
Y
Yan Zheng 已提交
2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857
	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;

2858
	read_lock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
2859
	em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
2860
	read_unlock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
2861 2862 2863
	if (!em)
		return 1;

2864 2865 2866 2867 2868
	if (btrfs_test_opt(root, DEGRADED)) {
		free_extent_map(em);
		return 0;
	}

Y
Yan Zheng 已提交
2869 2870 2871 2872 2873 2874 2875
	map = (struct map_lookup *)em->bdev;
	for (i = 0; i < map->num_stripes; i++) {
		if (!map->stripes[i].dev->writeable) {
			readonly = 1;
			break;
		}
	}
2876
	free_extent_map(em);
Y
Yan Zheng 已提交
2877
	return readonly;
2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888
}

void btrfs_mapping_init(struct btrfs_mapping_tree *tree)
{
	extent_map_tree_init(&tree->map_tree, GFP_NOFS);
}

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

C
Chris Mason 已提交
2889
	while (1) {
2890
		write_lock(&tree->map_tree.lock);
2891 2892 2893
		em = lookup_extent_mapping(&tree->map_tree, 0, (u64)-1);
		if (em)
			remove_extent_mapping(&tree->map_tree, em);
2894
		write_unlock(&tree->map_tree.lock);
2895 2896 2897 2898 2899 2900 2901 2902 2903 2904
		if (!em)
			break;
		kfree(em->bdev);
		/* once for us */
		free_extent_map(em);
		/* once for the tree */
		free_extent_map(em);
	}
}

2905 2906 2907 2908 2909 2910 2911
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;

2912
	read_lock(&em_tree->lock);
2913
	em = lookup_extent_mapping(em_tree, logical, len);
2914
	read_unlock(&em_tree->lock);
2915 2916 2917 2918 2919 2920
	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
2921 2922
	else if (map->type & BTRFS_BLOCK_GROUP_RAID10)
		ret = map->sub_stripes;
2923 2924 2925 2926 2927 2928
	else
		ret = 1;
	free_extent_map(em);
	return ret;
}

2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944
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;
}

2945 2946 2947 2948
static 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, struct page *unplug_page)
2949 2950 2951 2952 2953
{
	struct extent_map *em;
	struct map_lookup *map;
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	u64 offset;
2954 2955
	u64 stripe_offset;
	u64 stripe_nr;
2956
	int stripes_allocated = 8;
C
Chris Mason 已提交
2957
	int stripes_required = 1;
2958
	int stripe_index;
2959
	int i;
2960
	int num_stripes;
2961
	int max_errors = 0;
2962
	struct btrfs_multi_bio *multi = NULL;
2963

2964
	if (multi_ret && !(rw & REQ_WRITE))
2965 2966 2967 2968 2969 2970 2971
		stripes_allocated = 1;
again:
	if (multi_ret) {
		multi = kzalloc(btrfs_multi_bio_size(stripes_allocated),
				GFP_NOFS);
		if (!multi)
			return -ENOMEM;
2972 2973

		atomic_set(&multi->error, 0);
2974
	}
2975

2976
	read_lock(&em_tree->lock);
2977
	em = lookup_extent_mapping(em_tree, logical, *length);
2978
	read_unlock(&em_tree->lock);
2979

2980 2981
	if (!em && unplug_page) {
		kfree(multi);
2982
		return 0;
2983
	}
2984

2985
	if (!em) {
C
Chris Mason 已提交
2986 2987 2988
		printk(KERN_CRIT "unable to find logical %llu len %llu\n",
		       (unsigned long long)logical,
		       (unsigned long long)*length);
2989
		BUG();
2990
	}
2991 2992 2993 2994

	BUG_ON(em->start > logical || em->start + em->len < logical);
	map = (struct map_lookup *)em->bdev;
	offset = logical - em->start;
2995

2996 2997 2998
	if (mirror_num > map->num_stripes)
		mirror_num = 0;

2999
	/* if our multi bio struct is too small, back off and try again */
3000
	if (rw & REQ_WRITE) {
C
Chris Mason 已提交
3001 3002 3003
		if (map->type & (BTRFS_BLOCK_GROUP_RAID1 |
				 BTRFS_BLOCK_GROUP_DUP)) {
			stripes_required = map->num_stripes;
3004
			max_errors = 1;
C
Chris Mason 已提交
3005 3006
		} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
			stripes_required = map->sub_stripes;
3007
			max_errors = 1;
C
Chris Mason 已提交
3008 3009
		}
	}
3010
	if (multi_ret && (rw & REQ_WRITE) &&
C
Chris Mason 已提交
3011
	    stripes_allocated < stripes_required) {
3012 3013 3014 3015 3016
		stripes_allocated = map->num_stripes;
		free_extent_map(em);
		kfree(multi);
		goto again;
	}
3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029
	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;

3030
	if (map->type & (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1 |
C
Chris Mason 已提交
3031
			 BTRFS_BLOCK_GROUP_RAID10 |
3032 3033 3034 3035 3036 3037 3038
			 BTRFS_BLOCK_GROUP_DUP)) {
		/* we limit the length of each bio to what fits in a stripe */
		*length = min_t(u64, em->len - offset,
			      map->stripe_len - stripe_offset);
	} else {
		*length = em->len - offset;
	}
3039 3040

	if (!multi_ret && !unplug_page)
3041 3042
		goto out;

3043
	num_stripes = 1;
3044
	stripe_index = 0;
3045
	if (map->type & BTRFS_BLOCK_GROUP_RAID1) {
3046
		if (unplug_page || (rw & REQ_WRITE))
3047
			num_stripes = map->num_stripes;
3048
		else if (mirror_num)
3049
			stripe_index = mirror_num - 1;
3050 3051 3052 3053 3054
		else {
			stripe_index = find_live_mirror(map, 0,
					    map->num_stripes,
					    current->pid % map->num_stripes);
		}
3055

3056
	} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
3057
		if (rw & REQ_WRITE)
3058
			num_stripes = map->num_stripes;
3059 3060
		else if (mirror_num)
			stripe_index = mirror_num - 1;
3061

C
Chris Mason 已提交
3062 3063 3064 3065 3066 3067
	} 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;

3068
		if (unplug_page || (rw & REQ_WRITE))
3069
			num_stripes = map->sub_stripes;
C
Chris Mason 已提交
3070 3071
		else if (mirror_num)
			stripe_index += mirror_num - 1;
3072 3073 3074 3075 3076
		else {
			stripe_index = find_live_mirror(map, stripe_index,
					      map->sub_stripes, stripe_index +
					      current->pid % map->sub_stripes);
		}
3077 3078 3079 3080 3081 3082 3083 3084
	} 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);
	}
3085
	BUG_ON(stripe_index >= map->num_stripes);
3086

3087 3088 3089 3090 3091 3092
	for (i = 0; i < num_stripes; i++) {
		if (unplug_page) {
			struct btrfs_device *device;
			struct backing_dev_info *bdi;

			device = map->stripes[stripe_index].dev;
3093 3094
			if (device->bdev) {
				bdi = blk_get_backing_dev_info(device->bdev);
C
Chris Mason 已提交
3095
				if (bdi->unplug_io_fn)
3096
					bdi->unplug_io_fn(bdi, unplug_page);
3097 3098 3099 3100 3101 3102 3103
			}
		} else {
			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;
		}
3104
		stripe_index++;
3105
	}
3106 3107 3108
	if (multi_ret) {
		*multi_ret = multi;
		multi->num_stripes = num_stripes;
3109
		multi->max_errors = max_errors;
3110
	}
3111
out:
3112 3113 3114 3115
	free_extent_map(em);
	return 0;
}

3116 3117 3118 3119 3120 3121 3122 3123
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,
				 mirror_num, NULL);
}

Y
Yan Zheng 已提交
3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136
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;

3137
	read_lock(&em_tree->lock);
Y
Yan Zheng 已提交
3138
	em = lookup_extent_mapping(em_tree, chunk_start, 1);
3139
	read_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169

	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;
3170
		WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
3171 3172 3173 3174
		for (j = 0; j < nr; j++) {
			if (buf[j] == bytenr)
				break;
		}
3175 3176
		if (j == nr) {
			WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
3177
			buf[nr++] = bytenr;
3178
		}
Y
Yan Zheng 已提交
3179 3180 3181 3182 3183 3184 3185 3186 3187 3188
	}

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

	free_extent_map(em);
	return 0;
}

3189 3190 3191 3192 3193 3194 3195 3196
int btrfs_unplug_page(struct btrfs_mapping_tree *map_tree,
		      u64 logical, struct page *page)
{
	u64 length = PAGE_CACHE_SIZE;
	return __btrfs_map_block(map_tree, READ, logical, &length,
				 NULL, 0, page);
}

3197 3198
static void end_bio_multi_stripe(struct bio *bio, int err)
{
3199
	struct btrfs_multi_bio *multi = bio->bi_private;
3200
	int is_orig_bio = 0;
3201 3202

	if (err)
3203
		atomic_inc(&multi->error);
3204

3205 3206 3207
	if (bio == multi->orig_bio)
		is_orig_bio = 1;

3208
	if (atomic_dec_and_test(&multi->stripes_pending)) {
3209 3210 3211 3212
		if (!is_orig_bio) {
			bio_put(bio);
			bio = multi->orig_bio;
		}
3213 3214
		bio->bi_private = multi->private;
		bio->bi_end_io = multi->end_io;
3215 3216 3217
		/* only send an error to the higher layers if it is
		 * beyond the tolerance of the multi-bio
		 */
3218
		if (atomic_read(&multi->error) > multi->max_errors) {
3219
			err = -EIO;
3220 3221 3222 3223 3224 3225
		} 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);
3226
			err = 0;
3227
		}
3228 3229 3230
		kfree(multi);

		bio_endio(bio, err);
3231
	} else if (!is_orig_bio) {
3232 3233 3234 3235
		bio_put(bio);
	}
}

3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249
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
3250
static noinline int schedule_bio(struct btrfs_root *root,
3251 3252
				 struct btrfs_device *device,
				 int rw, struct bio *bio)
3253 3254
{
	int should_queue = 1;
3255
	struct btrfs_pending_bios *pending_bios;
3256 3257

	/* don't bother with additional async steps for reads, right now */
3258
	if (!(rw & REQ_WRITE)) {
3259
		bio_get(bio);
3260
		submit_bio(rw, bio);
3261
		bio_put(bio);
3262 3263 3264 3265
		return 0;
	}

	/*
3266
	 * nr_async_bios allows us to reliably return congestion to the
3267 3268 3269 3270
	 * 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
	 */
3271
	atomic_inc(&root->fs_info->nr_async_bios);
3272
	WARN_ON(bio->bi_next);
3273 3274 3275 3276
	bio->bi_next = NULL;
	bio->bi_rw |= rw;

	spin_lock(&device->io_lock);
3277
	if (bio->bi_rw & REQ_SYNC)
3278 3279 3280
		pending_bios = &device->pending_sync_bios;
	else
		pending_bios = &device->pending_bios;
3281

3282 3283
	if (pending_bios->tail)
		pending_bios->tail->bi_next = bio;
3284

3285 3286 3287
	pending_bios->tail = bio;
	if (!pending_bios->head)
		pending_bios->head = bio;
3288 3289 3290 3291 3292 3293
	if (device->running_pending)
		should_queue = 0;

	spin_unlock(&device->io_lock);

	if (should_queue)
3294 3295
		btrfs_queue_worker(&root->fs_info->submit_workers,
				   &device->work);
3296 3297 3298
	return 0;
}

3299
int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
3300
		  int mirror_num, int async_submit)
3301 3302 3303
{
	struct btrfs_mapping_tree *map_tree;
	struct btrfs_device *dev;
3304
	struct bio *first_bio = bio;
3305
	u64 logical = (u64)bio->bi_sector << 9;
3306 3307
	u64 length = 0;
	u64 map_length;
3308
	struct btrfs_multi_bio *multi = NULL;
3309
	int ret;
3310 3311
	int dev_nr = 0;
	int total_devs = 1;
3312

3313
	length = bio->bi_size;
3314 3315
	map_tree = &root->fs_info->mapping_tree;
	map_length = length;
3316

3317 3318
	ret = btrfs_map_block(map_tree, rw, logical, &map_length, &multi,
			      mirror_num);
3319 3320 3321 3322
	BUG_ON(ret);

	total_devs = multi->num_stripes;
	if (map_length < length) {
C
Chris Mason 已提交
3323 3324 3325 3326
		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);
3327 3328 3329 3330
		BUG();
	}
	multi->end_io = first_bio->bi_end_io;
	multi->private = first_bio->bi_private;
3331
	multi->orig_bio = first_bio;
3332 3333
	atomic_set(&multi->stripes_pending, multi->num_stripes);

C
Chris Mason 已提交
3334
	while (dev_nr < total_devs) {
3335 3336 3337 3338 3339 3340 3341 3342 3343 3344
		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;
		}
3345 3346
		bio->bi_sector = multi->stripes[dev_nr].physical >> 9;
		dev = multi->stripes[dev_nr].dev;
3347
		if (dev && dev->bdev && (rw != WRITE || dev->writeable)) {
3348
			bio->bi_bdev = dev->bdev;
3349 3350 3351 3352
			if (async_submit)
				schedule_bio(root, dev, rw, bio);
			else
				submit_bio(rw, bio);
3353 3354 3355 3356 3357
		} else {
			bio->bi_bdev = root->fs_info->fs_devices->latest_bdev;
			bio->bi_sector = logical >> 9;
			bio_endio(bio, -EIO);
		}
3358 3359
		dev_nr++;
	}
3360 3361
	if (total_devs == 1)
		kfree(multi);
3362 3363 3364
	return 0;
}

3365
struct btrfs_device *btrfs_find_device(struct btrfs_root *root, u64 devid,
Y
Yan Zheng 已提交
3366
				       u8 *uuid, u8 *fsid)
3367
{
Y
Yan Zheng 已提交
3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382
	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;
3383 3384
}

3385 3386 3387 3388 3389 3390 3391
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);
3392 3393
	if (!device)
		return NULL;
3394 3395 3396 3397 3398
	list_add(&device->dev_list,
		 &fs_devices->devices);
	device->barriers = 1;
	device->dev_root = root->fs_info->dev_root;
	device->devid = devid;
3399
	device->work.func = pending_bios_fn;
Y
Yan Zheng 已提交
3400
	device->fs_devices = fs_devices;
3401
	device->missing = 1;
3402
	fs_devices->num_devices++;
3403
	fs_devices->missing_devices++;
3404
	spin_lock_init(&device->io_lock);
3405
	INIT_LIST_HEAD(&device->dev_alloc_list);
3406 3407 3408 3409
	memcpy(device->uuid, dev_uuid, BTRFS_UUID_SIZE);
	return device;
}

3410 3411 3412 3413 3414 3415 3416 3417 3418 3419
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;
3420
	u8 uuid[BTRFS_UUID_SIZE];
3421
	int num_stripes;
3422
	int ret;
3423
	int i;
3424

3425 3426
	logical = key->offset;
	length = btrfs_chunk_length(leaf, chunk);
3427

3428
	read_lock(&map_tree->map_tree.lock);
3429
	em = lookup_extent_mapping(&map_tree->map_tree, logical, 1);
3430
	read_unlock(&map_tree->map_tree.lock);
3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442

	/* 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);
	}

	em = alloc_extent_map(GFP_NOFS);
	if (!em)
		return -ENOMEM;
3443 3444
	num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
3445 3446 3447 3448 3449 3450 3451 3452 3453
	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
3454
	em->block_len = em->len;
3455

3456 3457 3458 3459 3460 3461
	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
3462
	map->sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
3463 3464 3465 3466
	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);
3467 3468 3469
		read_extent_buffer(leaf, uuid, (unsigned long)
				   btrfs_stripe_dev_uuid_nr(chunk, i),
				   BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3470 3471
		map->stripes[i].dev = btrfs_find_device(root, devid, uuid,
							NULL);
3472
		if (!map->stripes[i].dev && !btrfs_test_opt(root, DEGRADED)) {
3473 3474 3475 3476
			kfree(map);
			free_extent_map(em);
			return -EIO;
		}
3477 3478 3479 3480 3481 3482 3483 3484 3485 3486
		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;
3487 3488
	}

3489
	write_lock(&map_tree->map_tree.lock);
3490
	ret = add_extent_mapping(&map_tree->map_tree, em);
3491
	write_unlock(&map_tree->map_tree.lock);
3492
	BUG_ON(ret);
3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504
	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);
3505 3506
	device->disk_total_bytes = btrfs_device_total_bytes(leaf, dev_item);
	device->total_bytes = device->disk_total_bytes;
3507 3508 3509 3510 3511 3512 3513
	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);
3514
	read_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
3515 3516 3517 3518

	return 0;
}

Y
Yan Zheng 已提交
3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539
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
3540 3541 3542 3543

	fs_devices = clone_fs_devices(fs_devices);
	if (IS_ERR(fs_devices)) {
		ret = PTR_ERR(fs_devices);
Y
Yan Zheng 已提交
3544 3545 3546
		goto out;
	}

3547
	ret = __btrfs_open_devices(fs_devices, FMODE_READ,
3548
				   root->fs_info->bdev_holder);
Y
Yan Zheng 已提交
3549 3550 3551 3552 3553
	if (ret)
		goto out;

	if (!fs_devices->seeding) {
		__btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
3554
		free_fs_devices(fs_devices);
Y
Yan Zheng 已提交
3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565
		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;
}

3566
static int read_one_dev(struct btrfs_root *root,
3567 3568 3569 3570 3571 3572
			struct extent_buffer *leaf,
			struct btrfs_dev_item *dev_item)
{
	struct btrfs_device *device;
	u64 devid;
	int ret;
Y
Yan Zheng 已提交
3573
	u8 fs_uuid[BTRFS_UUID_SIZE];
3574 3575
	u8 dev_uuid[BTRFS_UUID_SIZE];

3576
	devid = btrfs_device_id(leaf, dev_item);
3577 3578 3579
	read_extent_buffer(leaf, dev_uuid,
			   (unsigned long)btrfs_device_uuid(dev_item),
			   BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3580 3581 3582 3583 3584 3585
	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
3586
		if (ret && !btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
3587 3588 3589 3590 3591
			return ret;
	}

	device = btrfs_find_device(root, devid, dev_uuid, fs_uuid);
	if (!device || !device->bdev) {
Y
Yan Zheng 已提交
3592
		if (!btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
3593 3594 3595
			return -EIO;

		if (!device) {
C
Chris Mason 已提交
3596 3597
			printk(KERN_WARNING "warning devid %llu missing\n",
			       (unsigned long long)devid);
Y
Yan Zheng 已提交
3598 3599 3600
			device = add_missing_dev(root, devid, dev_uuid);
			if (!device)
				return -ENOMEM;
3601 3602 3603 3604 3605 3606 3607 3608 3609
		} 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
3610 3611 3612 3613 3614 3615 3616 3617
		}
	}

	if (device->fs_devices != root->fs_info->fs_devices) {
		BUG_ON(device->writeable);
		if (device->generation !=
		    btrfs_device_generation(leaf, dev_item))
			return -EINVAL;
3618
	}
3619 3620 3621

	fill_device_from_item(leaf, dev_item, device);
	device->dev_root = root->fs_info->dev_root;
3622
	device->in_fs_metadata = 1;
Y
Yan Zheng 已提交
3623 3624
	if (device->writeable)
		device->fs_devices->total_rw_bytes += device->total_bytes;
3625 3626 3627 3628
	ret = 0;
	return ret;
}

3629 3630 3631 3632 3633 3634 3635 3636 3637
int btrfs_read_super_device(struct btrfs_root *root, struct extent_buffer *buf)
{
	struct btrfs_dev_item *dev_item;

	dev_item = (struct btrfs_dev_item *)offsetof(struct btrfs_super_block,
						     dev_item);
	return read_one_dev(root, buf, dev_item);
}

Y
Yan Zheng 已提交
3638
int btrfs_read_sys_array(struct btrfs_root *root)
3639 3640
{
	struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
3641
	struct extent_buffer *sb;
3642 3643
	struct btrfs_disk_key *disk_key;
	struct btrfs_chunk *chunk;
3644 3645 3646
	u8 *ptr;
	unsigned long sb_ptr;
	int ret = 0;
3647 3648 3649 3650
	u32 num_stripes;
	u32 array_size;
	u32 len = 0;
	u32 cur;
3651
	struct btrfs_key key;
3652

Y
Yan Zheng 已提交
3653
	sb = btrfs_find_create_tree_block(root, BTRFS_SUPER_INFO_OFFSET,
3654 3655 3656 3657
					  BTRFS_SUPER_INFO_SIZE);
	if (!sb)
		return -ENOMEM;
	btrfs_set_buffer_uptodate(sb);
3658 3659
	btrfs_set_buffer_lockdep_class(sb, 0);

3660
	write_extent_buffer(sb, super_copy, 0, BTRFS_SUPER_INFO_SIZE);
3661 3662 3663 3664 3665 3666 3667 3668 3669 3670
	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);

3671
		len = sizeof(*disk_key); ptr += len;
3672 3673 3674
		sb_ptr += len;
		cur += len;

3675
		if (key.type == BTRFS_CHUNK_ITEM_KEY) {
3676
			chunk = (struct btrfs_chunk *)sb_ptr;
3677
			ret = read_one_chunk(root, &key, sb, chunk);
3678 3679
			if (ret)
				break;
3680 3681 3682
			num_stripes = btrfs_chunk_num_stripes(sb, chunk);
			len = btrfs_chunk_item_size(num_stripes);
		} else {
3683 3684
			ret = -EIO;
			break;
3685 3686 3687 3688 3689
		}
		ptr += len;
		sb_ptr += len;
		cur += len;
	}
3690
	free_extent_buffer(sb);
3691
	return ret;
3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717
}

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);
3718 3719
	if (ret < 0)
		goto error;
C
Chris Mason 已提交
3720
	while (1) {
3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738
		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);
3739
				ret = read_one_dev(root, leaf, dev_item);
Y
Yan Zheng 已提交
3740 3741
				if (ret)
					goto error;
3742 3743 3744 3745 3746
			}
		} 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
3747 3748
			if (ret)
				goto error;
3749 3750 3751 3752 3753 3754 3755 3756 3757 3758
		}
		path->slots[0]++;
	}
	if (key.objectid == BTRFS_DEV_ITEMS_OBJECTID) {
		key.objectid = 0;
		btrfs_release_path(root, path);
		goto again;
	}
	ret = 0;
error:
Y
Yan Zheng 已提交
3759
	btrfs_free_path(path);
3760 3761
	return ret;
}