volumes.c 85.4 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/buffer_head.h>
21
#include <linux/blkdev.h>
22
#include <linux/random.h>
23
#include <linux/iocontext.h>
24
#include <asm/div64.h>
C
Chris Mason 已提交
25
#include "compat.h"
26 27 28 29 30 31
#include "ctree.h"
#include "extent_map.h"
#include "disk-io.h"
#include "transaction.h"
#include "print-tree.h"
#include "volumes.h"
32
#include "async-thread.h"
33

34 35 36 37 38 39 40
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
41
	int sub_stripes;
42
	struct btrfs_bio_stripe stripes[];
43 44
};

Y
Yan Zheng 已提交
45 46 47 48 49
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);

50
#define map_lookup_size(n) (sizeof(struct map_lookup) + \
51
			    (sizeof(struct btrfs_bio_stripe) * (n)))
52

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

56 57 58 59 60 61 62 63 64 65
void btrfs_lock_volumes(void)
{
	mutex_lock(&uuid_mutex);
}

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

66 67 68 69 70 71 72 73 74 75
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
76 77 78 79 80 81 82 83 84 85 86 87 88 89
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);
}

90 91 92 93
int btrfs_cleanup_fs_uuids(void)
{
	struct btrfs_fs_devices *fs_devices;

Y
Yan Zheng 已提交
94 95 96 97
	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
98
		free_fs_devices(fs_devices);
99 100 101 102
	}
	return 0;
}

103 104
static noinline struct btrfs_device *__find_device(struct list_head *head,
						   u64 devid, u8 *uuid)
105 106 107
{
	struct btrfs_device *dev;

Q
Qinghuang Feng 已提交
108
	list_for_each_entry(dev, head, dev_list) {
109
		if (dev->devid == devid &&
110
		    (!uuid || !memcmp(dev->uuid, uuid, BTRFS_UUID_SIZE))) {
111
			return dev;
112
		}
113 114 115 116
	}
	return NULL;
}

117
static noinline struct btrfs_fs_devices *find_fsid(u8 *fsid)
118 119 120
{
	struct btrfs_fs_devices *fs_devices;

Q
Qinghuang Feng 已提交
121
	list_for_each_entry(fs_devices, &fs_uuids, list) {
122 123 124 125 126 127
		if (memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE) == 0)
			return fs_devices;
	}
	return NULL;
}

128 129 130 131 132 133 134 135 136 137 138 139 140 141
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;
}

142 143 144 145 146 147 148 149 150 151 152
/*
 * 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
153
static noinline int run_scheduled_bios(struct btrfs_device *device)
154 155 156
{
	struct bio *pending;
	struct backing_dev_info *bdi;
157
	struct btrfs_fs_info *fs_info;
158
	struct btrfs_pending_bios *pending_bios;
159 160 161
	struct bio *tail;
	struct bio *cur;
	int again = 0;
162 163
	unsigned long num_run;
	unsigned long num_sync_run;
164
	unsigned long batch_run = 0;
165
	unsigned long limit;
166
	unsigned long last_waited = 0;
167
	int force_reg = 0;
168

169
	bdi = blk_get_backing_dev_info(device->bdev);
170 171 172 173
	fs_info = device->dev_root->fs_info;
	limit = btrfs_async_submit_limit(fs_info);
	limit = limit * 2 / 3;

174 175 176 177 178
	/* we want to make sure that every time we switch from the sync
	 * list to the normal list, we unplug
	 */
	num_sync_run = 0;

179 180 181
loop:
	spin_lock(&device->io_lock);

182
loop_lock:
183
	num_run = 0;
184

185 186 187 188 189
	/* 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
	 */
190
	if (!force_reg && device->pending_sync_bios.head) {
191
		pending_bios = &device->pending_sync_bios;
192 193
		force_reg = 1;
	} else {
194
		pending_bios = &device->pending_bios;
195 196
		force_reg = 0;
	}
197 198 199

	pending = pending_bios->head;
	tail = pending_bios->tail;
200 201 202 203 204 205 206 207 208 209
	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.
	 */
210 211
	if (device->pending_sync_bios.head == NULL &&
	    device->pending_bios.head == NULL) {
212 213
		again = 0;
		device->running_pending = 0;
214 215 216
	} else {
		again = 1;
		device->running_pending = 1;
217
	}
218 219 220 221

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

222 223
	spin_unlock(&device->io_lock);

224 225 226 227 228 229 230 231 232
	/*
	 * 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
233
	while (pending) {
234 235

		rmb();
236 237 238 239 240 241 242 243
		/* 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)) {
244 245 246 247 248
			spin_lock(&device->io_lock);
			requeue_list(pending_bios, pending, tail);
			goto loop_lock;
		}

249 250 251
		cur = pending;
		pending = pending->bi_next;
		cur->bi_next = NULL;
252 253 254 255 256
		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);
257 258

		BUG_ON(atomic_read(&cur->bi_cnt) == 0);
259 260
		submit_bio(cur->bi_rw, cur);
		num_run++;
261 262
		batch_run++;

263
		if (bio_rw_flagged(cur, BIO_RW_SYNCIO))
264 265 266 267 268 269 270 271 272
			num_sync_run++;

		if (need_resched()) {
			if (num_sync_run) {
				blk_run_backing_dev(bdi, NULL);
				num_sync_run = 0;
			}
			cond_resched();
		}
273 274 275 276 277 278

		/*
		 * 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
279
		if (pending && bdi_write_congested(bdi) && batch_run > 8 &&
280
		    fs_info->fs_devices->open_devices > 1) {
281
			struct io_context *ioc;
282

283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304
			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;
305 306 307 308 309 310 311
				if (need_resched()) {
					if (num_sync_run) {
						blk_run_backing_dev(bdi, NULL);
						num_sync_run = 0;
					}
					cond_resched();
				}
312 313
				continue;
			}
314
			spin_lock(&device->io_lock);
315
			requeue_list(pending_bios, pending, tail);
316
			device->running_pending = 1;
317 318 319 320 321 322

			spin_unlock(&device->io_lock);
			btrfs_requeue_work(&device->work);
			goto done;
		}
	}
323 324 325 326 327 328 329

	if (num_sync_run) {
		num_sync_run = 0;
		blk_run_backing_dev(bdi, NULL);
	}

	cond_resched();
330 331
	if (again)
		goto loop;
332 333

	spin_lock(&device->io_lock);
334
	if (device->pending_bios.head || device->pending_sync_bios.head)
335 336
		goto loop_lock;
	spin_unlock(&device->io_lock);
337 338 339 340 341 342 343 344 345 346 347 348

	/*
	 * 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);
349 350 351 352
done:
	return 0;
}

353
static void pending_bios_fn(struct btrfs_work *work)
354 355 356 357 358 359 360
{
	struct btrfs_device *device;

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

361
static noinline int device_list_add(const char *path,
362 363 364 365 366 367 368 369 370
			   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);

	fs_devices = find_fsid(disk_super->fsid);
	if (!fs_devices) {
371
		fs_devices = kzalloc(sizeof(*fs_devices), GFP_NOFS);
372 373 374
		if (!fs_devices)
			return -ENOMEM;
		INIT_LIST_HEAD(&fs_devices->devices);
375
		INIT_LIST_HEAD(&fs_devices->alloc_list);
376 377 378 379
		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;
380
		mutex_init(&fs_devices->device_list_mutex);
381 382
		device = NULL;
	} else {
383 384
		device = __find_device(&fs_devices->devices, devid,
				       disk_super->dev_item.uuid);
385 386
	}
	if (!device) {
Y
Yan Zheng 已提交
387 388 389
		if (fs_devices->opened)
			return -EBUSY;

390 391 392 393 394 395
		device = kzalloc(sizeof(*device), GFP_NOFS);
		if (!device) {
			/* we can safely leave the fs_devices entry around */
			return -ENOMEM;
		}
		device->devid = devid;
396
		device->work.func = pending_bios_fn;
397 398
		memcpy(device->uuid, disk_super->dev_item.uuid,
		       BTRFS_UUID_SIZE);
399
		device->barriers = 1;
400
		spin_lock_init(&device->io_lock);
401 402 403 404 405
		device->name = kstrdup(path, GFP_NOFS);
		if (!device->name) {
			kfree(device);
			return -ENOMEM;
		}
Y
Yan Zheng 已提交
406
		INIT_LIST_HEAD(&device->dev_alloc_list);
407 408

		mutex_lock(&fs_devices->device_list_mutex);
409
		list_add(&device->dev_list, &fs_devices->devices);
410 411
		mutex_unlock(&fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
412
		device->fs_devices = fs_devices;
413 414 415 416 417 418 419 420 421 422 423
		fs_devices->num_devices++;
	}

	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
424 425 426 427 428 429 430 431 432 433 434 435 436
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);
437
	mutex_init(&fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
438 439 440 441
	fs_devices->latest_devid = orig->latest_devid;
	fs_devices->latest_trans = orig->latest_trans;
	memcpy(fs_devices->fsid, orig->fsid, sizeof(fs_devices->fsid));

442
	mutex_lock(&orig->device_list_mutex);
Y
Yan Zheng 已提交
443 444 445 446 447 448
	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
449 450
		if (!device->name) {
			kfree(device);
Y
Yan Zheng 已提交
451
			goto error;
J
Julia Lawall 已提交
452
		}
Y
Yan Zheng 已提交
453 454 455 456 457 458 459 460 461 462 463 464 465

		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++;
	}
466
	mutex_unlock(&orig->device_list_mutex);
Y
Yan Zheng 已提交
467 468
	return fs_devices;
error:
469
	mutex_unlock(&orig->device_list_mutex);
Y
Yan Zheng 已提交
470 471 472 473
	free_fs_devices(fs_devices);
	return ERR_PTR(-ENOMEM);
}

474 475
int btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices)
{
Q
Qinghuang Feng 已提交
476
	struct btrfs_device *device, *next;
477 478 479

	mutex_lock(&uuid_mutex);
again:
480
	mutex_lock(&fs_devices->device_list_mutex);
Q
Qinghuang Feng 已提交
481
	list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
Y
Yan Zheng 已提交
482 483 484 485
		if (device->in_fs_metadata)
			continue;

		if (device->bdev) {
486
			close_bdev_exclusive(device->bdev, device->mode);
Y
Yan Zheng 已提交
487 488 489 490 491 492 493 494
			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
495 496 497 498
		list_del_init(&device->dev_list);
		fs_devices->num_devices--;
		kfree(device->name);
		kfree(device);
499
	}
500
	mutex_unlock(&fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
501 502 503 504 505 506

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

507 508 509
	mutex_unlock(&uuid_mutex);
	return 0;
}
510

Y
Yan Zheng 已提交
511
static int __btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
512 513
{
	struct btrfs_device *device;
Y
Yan Zheng 已提交
514

Y
Yan Zheng 已提交
515 516
	if (--fs_devices->opened > 0)
		return 0;
517

Q
Qinghuang Feng 已提交
518
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
519
		if (device->bdev) {
520
			close_bdev_exclusive(device->bdev, device->mode);
521
			fs_devices->open_devices--;
522
		}
Y
Yan Zheng 已提交
523 524 525 526 527
		if (device->writeable) {
			list_del_init(&device->dev_alloc_list);
			fs_devices->rw_devices--;
		}

528
		device->bdev = NULL;
Y
Yan Zheng 已提交
529
		device->writeable = 0;
530
		device->in_fs_metadata = 0;
531
	}
Y
Yan Zheng 已提交
532 533
	WARN_ON(fs_devices->open_devices);
	WARN_ON(fs_devices->rw_devices);
Y
Yan Zheng 已提交
534 535 536
	fs_devices->opened = 0;
	fs_devices->seeding = 0;

537 538 539
	return 0;
}

Y
Yan Zheng 已提交
540 541
int btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
{
Y
Yan Zheng 已提交
542
	struct btrfs_fs_devices *seed_devices = NULL;
Y
Yan Zheng 已提交
543 544 545 546
	int ret;

	mutex_lock(&uuid_mutex);
	ret = __btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
547 548 549 550
	if (!fs_devices->opened) {
		seed_devices = fs_devices->seed;
		fs_devices->seed = NULL;
	}
Y
Yan Zheng 已提交
551
	mutex_unlock(&uuid_mutex);
Y
Yan Zheng 已提交
552 553 554 555 556 557 558

	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
559 560 561
	return ret;
}

Y
Yan Zheng 已提交
562 563
static int __btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
				fmode_t flags, void *holder)
564 565 566 567
{
	struct block_device *bdev;
	struct list_head *head = &fs_devices->devices;
	struct btrfs_device *device;
568 569 570 571 572 573
	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
574
	int seeding = 1;
575
	int ret = 0;
576

Q
Qinghuang Feng 已提交
577
	list_for_each_entry(device, head, dev_list) {
578 579
		if (device->bdev)
			continue;
580 581 582
		if (!device->name)
			continue;

583
		bdev = open_bdev_exclusive(device->name, flags, holder);
584
		if (IS_ERR(bdev)) {
C
Chris Mason 已提交
585
			printk(KERN_INFO "open %s failed\n", device->name);
586
			goto error;
587
		}
588
		set_blocksize(bdev, 4096);
589

Y
Yan Zheng 已提交
590
		bh = btrfs_read_dev_super(bdev);
591 592 593 594 595 596 597 598
		if (!bh)
			goto error_close;

		disk_super = (struct btrfs_super_block *)bh->b_data;
		devid = le64_to_cpu(disk_super->dev_item.devid);
		if (devid != device->devid)
			goto error_brelse;

Y
Yan Zheng 已提交
599 600 601 602 603 604
		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) {
605
			latest_devid = devid;
Y
Yan Zheng 已提交
606
			latest_transid = device->generation;
607 608 609
			latest_bdev = bdev;
		}

Y
Yan Zheng 已提交
610 611 612 613 614 615 616
		if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_SEEDING) {
			device->writeable = 0;
		} else {
			device->writeable = !bdev_read_only(bdev);
			seeding = 0;
		}

617
		device->bdev = bdev;
618
		device->in_fs_metadata = 0;
619 620
		device->mode = flags;

C
Chris Mason 已提交
621 622 623
		if (!blk_queue_nonrot(bdev_get_queue(bdev)))
			fs_devices->rotating = 1;

624
		fs_devices->open_devices++;
Y
Yan Zheng 已提交
625 626 627 628 629
		if (device->writeable) {
			fs_devices->rw_devices++;
			list_add(&device->dev_alloc_list,
				 &fs_devices->alloc_list);
		}
630
		continue;
631

632 633 634
error_brelse:
		brelse(bh);
error_close:
635
		close_bdev_exclusive(bdev, FMODE_READ);
636 637
error:
		continue;
638
	}
639 640 641 642
	if (fs_devices->open_devices == 0) {
		ret = -EIO;
		goto out;
	}
Y
Yan Zheng 已提交
643 644
	fs_devices->seeding = seeding;
	fs_devices->opened = 1;
645 646 647
	fs_devices->latest_bdev = latest_bdev;
	fs_devices->latest_devid = latest_devid;
	fs_devices->latest_trans = latest_transid;
Y
Yan Zheng 已提交
648
	fs_devices->total_rw_bytes = 0;
649
out:
Y
Yan Zheng 已提交
650 651 652 653
	return ret;
}

int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
654
		       fmode_t flags, void *holder)
Y
Yan Zheng 已提交
655 656 657 658 659
{
	int ret;

	mutex_lock(&uuid_mutex);
	if (fs_devices->opened) {
Y
Yan Zheng 已提交
660 661
		fs_devices->opened++;
		ret = 0;
Y
Yan Zheng 已提交
662
	} else {
663
		ret = __btrfs_open_devices(fs_devices, flags, holder);
Y
Yan Zheng 已提交
664
	}
665 666 667 668
	mutex_unlock(&uuid_mutex);
	return ret;
}

669
int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
670 671 672 673 674 675 676
			  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;
677
	u64 transid;
678 679 680

	mutex_lock(&uuid_mutex);

681
	bdev = open_bdev_exclusive(path, flags, holder);
682 683 684 685 686 687 688 689 690

	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
691
	bh = btrfs_read_dev_super(bdev);
692 693 694 695 696 697
	if (!bh) {
		ret = -EIO;
		goto error_close;
	}
	disk_super = (struct btrfs_super_block *)bh->b_data;
	devid = le64_to_cpu(disk_super->dev_item.devid);
698
	transid = btrfs_super_generation(disk_super);
699
	if (disk_super->label[0])
C
Chris Mason 已提交
700
		printk(KERN_INFO "device label %s ", disk_super->label);
701 702
	else {
		/* FIXME, make a readl uuid parser */
C
Chris Mason 已提交
703
		printk(KERN_INFO "device fsid %llx-%llx ",
704 705 706
		       *(unsigned long long *)disk_super->fsid,
		       *(unsigned long long *)(disk_super->fsid + 8));
	}
707
	printk(KERN_CONT "devid %llu transid %llu %s\n",
C
Chris Mason 已提交
708
	       (unsigned long long)devid, (unsigned long long)transid, path);
709 710 711 712
	ret = device_list_add(path, disk_super, devid, fs_devices_ret);

	brelse(bh);
error_close:
713
	close_bdev_exclusive(bdev, flags);
714 715 716 717
error:
	mutex_unlock(&uuid_mutex);
	return ret;
}
718 719 720 721 722 723

/*
 * 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
 */
724 725 726
int find_free_dev_extent(struct btrfs_trans_handle *trans,
			 struct btrfs_device *device, u64 num_bytes,
			 u64 *start, u64 *max_avail)
727 728 729 730
{
	struct btrfs_key key;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_dev_extent *dev_extent = NULL;
Y
Yan Zheng 已提交
731
	struct btrfs_path *path;
732 733 734 735 736 737 738 739 740
	u64 hole_size = 0;
	u64 last_byte = 0;
	u64 search_start = 0;
	u64 search_end = device->total_bytes;
	int ret;
	int slot = 0;
	int start_found;
	struct extent_buffer *l;

Y
Yan Zheng 已提交
741 742 743
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
744
	path->reada = 2;
Y
Yan Zheng 已提交
745
	start_found = 0;
746 747 748

	/* FIXME use last free of some kind */

749 750 751 752
	/* we don't want to overwrite the superblock on the drive,
	 * so we make sure to start at an offset of at least 1MB
	 */
	search_start = max((u64)1024 * 1024, search_start);
753 754 755 756

	if (root->fs_info->alloc_start + num_bytes <= device->total_bytes)
		search_start = max(root->fs_info->alloc_start, search_start);

757 758 759 760 761 762
	key.objectid = device->devid;
	key.offset = search_start;
	key.type = BTRFS_DEV_EXTENT_KEY;
	ret = btrfs_search_slot(trans, root, &key, path, 0, 0);
	if (ret < 0)
		goto error;
763 764 765 766 767 768 769
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid, key.type);
		if (ret < 0)
			goto error;
		if (ret > 0)
			start_found = 1;
	}
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
	l = path->nodes[0];
	btrfs_item_key_to_cpu(l, &key, path->slots[0]);
	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 error;
no_more_items:
			if (!start_found) {
				if (search_start >= search_end) {
					ret = -ENOSPC;
					goto error;
				}
				*start = search_start;
				start_found = 1;
				goto check_pending;
			}
			*start = last_byte > search_start ?
				last_byte : search_start;
			if (search_end <= *start) {
				ret = -ENOSPC;
				goto error;
			}
			goto check_pending;
		}
		btrfs_item_key_to_cpu(l, &key, slot);

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

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

		if (key.offset >= search_start && key.offset > last_byte &&
		    start_found) {
			if (last_byte < search_start)
				last_byte = search_start;
			hole_size = key.offset - last_byte;
812 813 814 815

			if (hole_size > *max_avail)
				*max_avail = hole_size;

816 817 818 819 820 821
			if (key.offset > last_byte &&
			    hole_size >= num_bytes) {
				*start = last_byte;
				goto check_pending;
			}
		}
C
Chris Mason 已提交
822
		if (btrfs_key_type(&key) != BTRFS_DEV_EXTENT_KEY)
823 824 825 826 827 828 829 830 831 832 833 834 835 836 837
			goto next;

		start_found = 1;
		dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
		last_byte = key.offset + btrfs_dev_extent_length(l, dev_extent);
next:
		path->slots[0]++;
		cond_resched();
	}
check_pending:
	/* we have to make sure we didn't find an extent that has already
	 * been allocated by the map tree or the original allocation
	 */
	BUG_ON(*start < search_start);

838
	if (*start + num_bytes > search_end) {
839 840 841 842
		ret = -ENOSPC;
		goto error;
	}
	/* check for pending inserts here */
Y
Yan Zheng 已提交
843
	ret = 0;
844 845

error:
Y
Yan Zheng 已提交
846
	btrfs_free_path(path);
847 848 849
	return ret;
}

850
static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans,
851 852 853 854 855 856 857
			  struct btrfs_device *device,
			  u64 start)
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_key key;
858 859 860
	struct btrfs_key found_key;
	struct extent_buffer *leaf = NULL;
	struct btrfs_dev_extent *extent = NULL;
861 862 863 864 865 866 867 868 869 870

	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);
871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886
	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);
	}
887 888
	BUG_ON(ret);

889 890
	if (device->bytes_used > 0)
		device->bytes_used -= btrfs_dev_extent_length(leaf, extent);
891 892 893 894 895 896 897
	ret = btrfs_del_item(trans, root, path);
	BUG_ON(ret);

	btrfs_free_path(path);
	return ret;
}

Y
Yan Zheng 已提交
898
int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
899
			   struct btrfs_device *device,
900
			   u64 chunk_tree, u64 chunk_objectid,
Y
Yan Zheng 已提交
901
			   u64 chunk_offset, u64 start, u64 num_bytes)
902 903 904 905 906 907 908 909
{
	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;

910
	WARN_ON(!device->in_fs_metadata);
911 912 913 914 915
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = device->devid;
Y
Yan Zheng 已提交
916
	key.offset = start;
917 918 919 920 921 922 923 924
	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);
925 926 927 928 929 930 931 932
	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);

933 934 935 936 937 938
	btrfs_set_dev_extent_length(leaf, extent, num_bytes);
	btrfs_mark_buffer_dirty(leaf);
	btrfs_free_path(path);
	return ret;
}

939 940
static noinline int find_next_chunk(struct btrfs_root *root,
				    u64 objectid, u64 *offset)
941 942 943 944
{
	struct btrfs_path *path;
	int ret;
	struct btrfs_key key;
945
	struct btrfs_chunk *chunk;
946 947 948 949 950
	struct btrfs_key found_key;

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

951
	key.objectid = objectid;
952 953 954 955 956 957 958 959 960 961 962
	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) {
963
		*offset = 0;
964 965 966
	} else {
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
967 968 969 970 971 972 973 974
		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);
		}
975 976 977 978 979 980 981
	}
	ret = 0;
error:
	btrfs_free_path(path);
	return ret;
}

Y
Yan Zheng 已提交
982
static noinline int find_next_devid(struct btrfs_root *root, u64 *objectid)
983 984 985 986
{
	int ret;
	struct btrfs_key key;
	struct btrfs_key found_key;
Y
Yan Zheng 已提交
987 988 989 990 991 992 993
	struct btrfs_path *path;

	root = root->fs_info->chunk_root;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015

	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
1016
	btrfs_free_path(path);
1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042
	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
1043
	key.offset = device->devid;
1044 1045

	ret = btrfs_insert_empty_item(trans, root, path, &key,
1046
				      sizeof(*dev_item));
1047 1048 1049 1050 1051 1052 1053
	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
1054
	btrfs_set_device_generation(leaf, dev_item, 0);
1055 1056 1057 1058 1059 1060
	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);
1061 1062 1063
	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);
1064
	btrfs_set_device_start_offset(leaf, dev_item, 0);
1065 1066

	ptr = (unsigned long)btrfs_device_uuid(dev_item);
1067
	write_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
1068 1069
	ptr = (unsigned long)btrfs_device_fsid(dev_item);
	write_extent_buffer(leaf, root->fs_info->fsid, ptr, BTRFS_UUID_SIZE);
1070 1071
	btrfs_mark_buffer_dirty(leaf);

Y
Yan Zheng 已提交
1072
	ret = 0;
1073 1074 1075 1076
out:
	btrfs_free_path(path);
	return ret;
}
1077

1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095
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;

	trans = btrfs_start_transaction(root, 1);
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.type = BTRFS_DEV_ITEM_KEY;
	key.offset = device->devid;
1096
	lock_chunks(root);
1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111

	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);
1112
	unlock_chunks(root);
1113 1114 1115 1116 1117 1118 1119
	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
1120
	struct btrfs_device *next_device;
1121
	struct block_device *bdev;
1122
	struct buffer_head *bh = NULL;
1123 1124 1125
	struct btrfs_super_block *disk_super;
	u64 all_avail;
	u64 devid;
Y
Yan Zheng 已提交
1126 1127
	u64 num_devices;
	u8 *dev_uuid;
1128 1129 1130
	int ret = 0;

	mutex_lock(&uuid_mutex);
1131
	mutex_lock(&root->fs_info->volume_mutex);
1132 1133 1134 1135 1136 1137

	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) &&
Y
Yan Zheng 已提交
1138
	    root->fs_info->fs_devices->rw_devices <= 4) {
C
Chris Mason 已提交
1139 1140
		printk(KERN_ERR "btrfs: unable to go below four devices "
		       "on raid10\n");
1141 1142 1143 1144 1145
		ret = -EINVAL;
		goto out;
	}

	if ((all_avail & BTRFS_BLOCK_GROUP_RAID1) &&
Y
Yan Zheng 已提交
1146
	    root->fs_info->fs_devices->rw_devices <= 2) {
C
Chris Mason 已提交
1147 1148
		printk(KERN_ERR "btrfs: unable to go below two "
		       "devices on raid1\n");
1149 1150 1151 1152
		ret = -EINVAL;
		goto out;
	}

1153 1154 1155
	if (strcmp(device_path, "missing") == 0) {
		struct list_head *devices;
		struct btrfs_device *tmp;
1156

1157 1158
		device = NULL;
		devices = &root->fs_info->fs_devices->devices;
1159
		mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
Q
Qinghuang Feng 已提交
1160
		list_for_each_entry(tmp, devices, dev_list) {
1161 1162 1163 1164 1165
			if (tmp->in_fs_metadata && !tmp->bdev) {
				device = tmp;
				break;
			}
		}
1166
		mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
1167 1168 1169 1170
		bdev = NULL;
		bh = NULL;
		disk_super = NULL;
		if (!device) {
C
Chris Mason 已提交
1171 1172
			printk(KERN_ERR "btrfs: no missing devices found to "
			       "remove\n");
1173 1174 1175
			goto out;
		}
	} else {
1176
		bdev = open_bdev_exclusive(device_path, FMODE_READ,
1177 1178 1179 1180 1181
				      root->fs_info->bdev_holder);
		if (IS_ERR(bdev)) {
			ret = PTR_ERR(bdev);
			goto out;
		}
1182

Y
Yan Zheng 已提交
1183
		set_blocksize(bdev, 4096);
Y
Yan Zheng 已提交
1184
		bh = btrfs_read_dev_super(bdev);
1185 1186 1187 1188 1189 1190
		if (!bh) {
			ret = -EIO;
			goto error_close;
		}
		disk_super = (struct btrfs_super_block *)bh->b_data;
		devid = le64_to_cpu(disk_super->dev_item.devid);
Y
Yan Zheng 已提交
1191 1192 1193
		dev_uuid = disk_super->dev_item.uuid;
		device = btrfs_find_device(root, devid, dev_uuid,
					   disk_super->fsid);
1194 1195 1196 1197
		if (!device) {
			ret = -ENOENT;
			goto error_brelse;
		}
Y
Yan Zheng 已提交
1198
	}
1199

Y
Yan Zheng 已提交
1200
	if (device->writeable && root->fs_info->fs_devices->rw_devices == 1) {
C
Chris Mason 已提交
1201 1202
		printk(KERN_ERR "btrfs: unable to remove the only writeable "
		       "device\n");
Y
Yan Zheng 已提交
1203 1204 1205 1206 1207 1208 1209
		ret = -EINVAL;
		goto error_brelse;
	}

	if (device->writeable) {
		list_del_init(&device->dev_alloc_list);
		root->fs_info->fs_devices->rw_devices--;
1210
	}
1211 1212 1213 1214 1215 1216 1217 1218 1219

	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
1220
	device->in_fs_metadata = 0;
1221 1222 1223 1224 1225 1226 1227

	/*
	 * 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
1228
	list_del_init(&device->dev_list);
1229 1230
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
1231
	device->fs_devices->num_devices--;
Y
Yan Zheng 已提交
1232 1233 1234 1235 1236 1237 1238 1239

	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
1240 1241 1242 1243 1244 1245
	if (device->bdev) {
		close_bdev_exclusive(device->bdev, device->mode);
		device->bdev = NULL;
		device->fs_devices->open_devices--;
	}

Y
Yan Zheng 已提交
1246 1247 1248
	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
1249 1250 1251 1252 1253 1254 1255
	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
1256
		}
Y
Yan Zheng 已提交
1257 1258 1259 1260
		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
1261 1262 1263 1264 1265 1266 1267
	}

	/*
	 * 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) {
1268 1269 1270 1271 1272 1273 1274
		/* 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);
	}
1275 1276 1277 1278 1279 1280 1281 1282

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

error_brelse:
	brelse(bh);
error_close:
1283
	if (bdev)
1284
		close_bdev_exclusive(bdev, FMODE_READ);
1285
out:
1286
	mutex_unlock(&root->fs_info->volume_mutex);
1287 1288 1289 1290
	mutex_unlock(&uuid_mutex);
	return ret;
}

Y
Yan Zheng 已提交
1291 1292 1293 1294 1295 1296 1297 1298
/*
 * 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
1299
	struct btrfs_fs_devices *seed_devices;
Y
Yan Zheng 已提交
1300 1301 1302 1303 1304
	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
1305
	if (!fs_devices->seeding)
Y
Yan Zheng 已提交
1306 1307
		return -EINVAL;

Y
Yan Zheng 已提交
1308 1309
	seed_devices = kzalloc(sizeof(*fs_devices), GFP_NOFS);
	if (!seed_devices)
Y
Yan Zheng 已提交
1310 1311
		return -ENOMEM;

Y
Yan Zheng 已提交
1312 1313 1314 1315
	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
1316
	}
Y
Yan Zheng 已提交
1317

Y
Yan Zheng 已提交
1318 1319
	list_add(&old_devices->list, &fs_uuids);

Y
Yan Zheng 已提交
1320 1321 1322 1323
	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);
1324
	mutex_init(&seed_devices->device_list_mutex);
Y
Yan Zheng 已提交
1325 1326 1327 1328 1329 1330
	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
1331 1332 1333
	fs_devices->seeding = 0;
	fs_devices->num_devices = 0;
	fs_devices->open_devices = 0;
Y
Yan Zheng 已提交
1334
	fs_devices->seed = seed_devices;
Y
Yan Zheng 已提交
1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421

	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;
}

1422 1423 1424 1425 1426 1427
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
1428
	struct super_block *sb = root->fs_info->sb;
1429
	u64 total_bytes;
Y
Yan Zheng 已提交
1430
	int seeding_dev = 0;
1431 1432
	int ret = 0;

Y
Yan Zheng 已提交
1433 1434
	if ((sb->s_flags & MS_RDONLY) && !root->fs_info->fs_devices->seeding)
		return -EINVAL;
1435

1436
	bdev = open_bdev_exclusive(device_path, 0, root->fs_info->bdev_holder);
1437 1438
	if (IS_ERR(bdev))
		return PTR_ERR(bdev);
1439

Y
Yan Zheng 已提交
1440 1441 1442 1443 1444 1445
	if (root->fs_info->fs_devices->seeding) {
		seeding_dev = 1;
		down_write(&sb->s_umount);
		mutex_lock(&uuid_mutex);
	}

1446
	filemap_write_and_wait(bdev->bd_inode->i_mapping);
1447
	mutex_lock(&root->fs_info->volume_mutex);
1448

1449
	devices = &root->fs_info->fs_devices->devices;
1450 1451 1452 1453
	/*
	 * 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
1454
	list_for_each_entry(device, devices, dev_list) {
1455 1456
		if (device->bdev == bdev) {
			ret = -EEXIST;
Y
Yan Zheng 已提交
1457
			goto error;
1458 1459 1460 1461 1462 1463 1464
		}
	}

	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
1465
		goto error;
1466 1467 1468 1469 1470
	}

	device->name = kstrdup(device_path, GFP_NOFS);
	if (!device->name) {
		kfree(device);
Y
Yan Zheng 已提交
1471 1472
		ret = -ENOMEM;
		goto error;
1473
	}
Y
Yan Zheng 已提交
1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489

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

	trans = btrfs_start_transaction(root, 1);
	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;
1490 1491 1492 1493
	device->io_width = root->sectorsize;
	device->io_align = root->sectorsize;
	device->sector_size = root->sectorsize;
	device->total_bytes = i_size_read(bdev->bd_inode);
1494
	device->disk_total_bytes = device->total_bytes;
1495 1496
	device->dev_root = root->fs_info->dev_root;
	device->bdev = bdev;
1497
	device->in_fs_metadata = 1;
1498
	device->mode = 0;
Y
Yan Zheng 已提交
1499
	set_blocksize(device->bdev, 4096);
1500

Y
Yan Zheng 已提交
1501 1502 1503 1504 1505
	if (seeding_dev) {
		sb->s_flags &= ~MS_RDONLY;
		ret = btrfs_prepare_sprout(trans, root);
		BUG_ON(ret);
	}
1506

Y
Yan Zheng 已提交
1507
	device->fs_devices = root->fs_info->fs_devices;
1508 1509 1510 1511 1512 1513

	/*
	 * 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
1514 1515 1516 1517 1518 1519 1520
	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;
1521

C
Chris Mason 已提交
1522 1523 1524
	if (!blk_queue_nonrot(bdev_get_queue(bdev)))
		root->fs_info->fs_devices->rotating = 1;

1525 1526 1527 1528 1529 1530 1531
	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);
1532
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
1533

Y
Yan Zheng 已提交
1534 1535 1536 1537 1538 1539 1540 1541 1542
	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);
	}

1543 1544 1545 1546 1547 1548
	/*
	 * we've got more storage, clear any full flags on the space
	 * infos
	 */
	btrfs_clear_space_info_full(root->fs_info);

1549
	unlock_chunks(root);
Y
Yan Zheng 已提交
1550
	btrfs_commit_transaction(trans, root);
1551

Y
Yan Zheng 已提交
1552 1553 1554
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
1555

Y
Yan Zheng 已提交
1556 1557 1558 1559 1560 1561 1562
		ret = btrfs_relocate_sys_chunks(root);
		BUG_ON(ret);
	}
out:
	mutex_unlock(&root->fs_info->volume_mutex);
	return ret;
error:
1563
	close_bdev_exclusive(bdev, 0);
Y
Yan Zheng 已提交
1564 1565 1566 1567
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
	}
1568 1569 1570
	goto out;
}

C
Chris Mason 已提交
1571 1572
static noinline int btrfs_update_device(struct btrfs_trans_handle *trans,
					struct btrfs_device *device)
1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607
{
	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);
1608
	btrfs_set_device_total_bytes(leaf, dev_item, device->disk_total_bytes);
1609 1610 1611 1612 1613 1614 1615 1616
	btrfs_set_device_bytes_used(leaf, dev_item, device->bytes_used);
	btrfs_mark_buffer_dirty(leaf);

out:
	btrfs_free_path(path);
	return ret;
}

1617
static int __btrfs_grow_device(struct btrfs_trans_handle *trans,
1618 1619 1620 1621 1622 1623 1624
		      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
1625 1626 1627 1628 1629
	if (!device->writeable)
		return -EACCES;
	if (new_size <= device->total_bytes)
		return -EINVAL;

1630
	btrfs_set_super_total_bytes(super_copy, old_total + diff);
Y
Yan Zheng 已提交
1631 1632 1633
	device->fs_devices->total_rw_bytes += diff;

	device->total_bytes = new_size;
1634
	device->disk_total_bytes = new_size;
1635 1636
	btrfs_clear_space_info_full(device->dev_root->fs_info);

1637 1638 1639
	return btrfs_update_device(trans, device);
}

1640 1641 1642 1643 1644 1645 1646 1647 1648 1649
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;
}

1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677
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;
}

1678
static int btrfs_del_sys_chunk(struct btrfs_root *root, u64 chunk_objectid, u64
1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723
			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;
}

1724
static int btrfs_relocate_chunk(struct btrfs_root *root,
1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739
			 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;

1740 1741 1742 1743
	ret = btrfs_can_relocate(extent_root, chunk_offset);
	if (ret)
		return -ENOSPC;

1744
	/* step one, relocate all the extents inside this chunk */
Z
Zheng Yan 已提交
1745
	ret = btrfs_relocate_block_group(extent_root, chunk_offset);
1746 1747 1748 1749 1750
	BUG_ON(ret);

	trans = btrfs_start_transaction(root, 1);
	BUG_ON(!trans);

1751 1752
	lock_chunks(root);

1753 1754 1755 1756
	/*
	 * step two, delete the device extents and the
	 * chunk tree entries
	 */
1757
	read_lock(&em_tree->lock);
1758
	em = lookup_extent_mapping(em_tree, chunk_offset, 1);
1759
	read_unlock(&em_tree->lock);
1760

1761 1762
	BUG_ON(em->start > chunk_offset ||
	       em->start + em->len < chunk_offset);
1763 1764 1765 1766 1767 1768
	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);
1769

1770 1771 1772 1773
		if (map->stripes[i].dev) {
			ret = btrfs_update_device(trans, map->stripes[i].dev);
			BUG_ON(ret);
		}
1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784
	}
	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
1785 1786 1787
	ret = btrfs_remove_block_group(trans, extent_root, chunk_offset);
	BUG_ON(ret);

1788
	write_lock(&em_tree->lock);
Y
Yan Zheng 已提交
1789
	remove_extent_mapping(em_tree, em);
1790
	write_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814

	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;
1815 1816
	bool retried = false;
	int failed = 0;
Y
Yan Zheng 已提交
1817 1818 1819 1820 1821 1822
	int ret;

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

1823
again:
Y
Yan Zheng 已提交
1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839
	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
1840

Y
Yan Zheng 已提交
1841 1842
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
Z
Zheng Yan 已提交
1843

Y
Yan Zheng 已提交
1844 1845 1846 1847
		chunk = btrfs_item_ptr(leaf, path->slots[0],
				       struct btrfs_chunk);
		chunk_type = btrfs_chunk_type(leaf, chunk);
		btrfs_release_path(chunk_root, path);
1848

Y
Yan Zheng 已提交
1849 1850 1851 1852
		if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) {
			ret = btrfs_relocate_chunk(chunk_root, chunk_tree,
						   found_key.objectid,
						   found_key.offset);
1853 1854 1855 1856
			if (ret == -ENOSPC)
				failed++;
			else if (ret)
				BUG();
Y
Yan Zheng 已提交
1857
		}
1858

Y
Yan Zheng 已提交
1859 1860 1861 1862 1863
		if (found_key.offset == 0)
			break;
		key.offset = found_key.offset - 1;
	}
	ret = 0;
1864 1865 1866 1867 1868 1869 1870 1871
	if (failed && !retried) {
		failed = 0;
		retried = true;
		goto again;
	} else if (failed && retried) {
		WARN_ON(1);
		ret = -ENOSPC;
	}
Y
Yan Zheng 已提交
1872 1873 1874
error:
	btrfs_free_path(path);
	return ret;
1875 1876
}

1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899
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_chunk *chunk;
	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
1900 1901
	if (dev_root->fs_info->sb->s_flags & MS_RDONLY)
		return -EROFS;
1902

1903
	mutex_lock(&dev_root->fs_info->volume_mutex);
1904 1905 1906
	dev_root = dev_root->fs_info->dev_root;

	/* step one make some room on all the devices */
Q
Qinghuang Feng 已提交
1907
	list_for_each_entry(device, devices, dev_list) {
1908 1909 1910
		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
1911 1912
		if (!device->writeable ||
		    device->total_bytes - device->bytes_used > size_to_free)
1913 1914 1915
			continue;

		ret = btrfs_shrink_device(device, old_size - size_to_free);
1916 1917
		if (ret == -ENOSPC)
			break;
1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936
		BUG_ON(ret);

		trans = btrfs_start_transaction(dev_root, 1);
		BUG_ON(!trans);

		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
1937
	while (1) {
1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950
		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);
1951
		if (ret)
1952
			break;
1953

1954 1955 1956 1957
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
		if (found_key.objectid != key.objectid)
			break;
1958

1959 1960 1961 1962
		chunk = btrfs_item_ptr(path->nodes[0],
				       path->slots[0],
				       struct btrfs_chunk);
		/* chunk zero is special */
1963
		if (found_key.offset == 0)
1964 1965
			break;

1966
		btrfs_release_path(chunk_root, path);
1967 1968 1969 1970
		ret = btrfs_relocate_chunk(chunk_root,
					   chunk_root->root_key.objectid,
					   found_key.objectid,
					   found_key.offset);
1971 1972
		BUG_ON(ret && ret != -ENOSPC);
		key.offset = found_key.offset - 1;
1973 1974 1975 1976
	}
	ret = 0;
error:
	btrfs_free_path(path);
1977
	mutex_unlock(&dev_root->fs_info->volume_mutex);
1978 1979 1980
	return ret;
}

1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997
/*
 * 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;
1998 1999
	int failed = 0;
	bool retried = false;
2000 2001 2002 2003
	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);
2004
	u64 old_size = device->total_bytes;
2005 2006
	u64 diff = device->total_bytes - new_size;

Y
Yan Zheng 已提交
2007 2008
	if (new_size >= device->total_bytes)
		return -EINVAL;
2009 2010 2011 2012 2013 2014 2015

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

	path->reada = 2;

2016 2017
	lock_chunks(root);

2018
	device->total_bytes = new_size;
Y
Yan Zheng 已提交
2019 2020
	if (device->writeable)
		device->fs_devices->total_rw_bytes -= diff;
2021
	unlock_chunks(root);
2022

2023
again:
2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037
	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;
2038
			btrfs_release_path(root, path);
2039
			break;
2040 2041 2042 2043 2044 2045
		}

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

2046 2047
		if (key.objectid != device->devid) {
			btrfs_release_path(root, path);
2048
			break;
2049
		}
2050 2051 2052 2053

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

2054 2055
		if (key.offset + length <= new_size) {
			btrfs_release_path(root, path);
2056
			break;
2057
		}
2058 2059 2060 2061 2062 2063 2064 2065

		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);
2066
		if (ret && ret != -ENOSPC)
2067
			goto done;
2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085
		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;
2086 2087
	}

2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107
	/* Shrinking succeeded, else we would be at "done". */
	trans = btrfs_start_transaction(root, 1);
	if (!trans) {
		ret = -ENOMEM;
		goto done;
	}
	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);
2108 2109 2110 2111 2112
done:
	btrfs_free_path(path);
	return ret;
}

2113
static int btrfs_add_system_chunk(struct btrfs_trans_handle *trans,
2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136
			   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
2137
static noinline u64 chunk_bytes_by_type(u64 type, u64 calc_size,
2138
					int num_stripes, int sub_stripes)
2139 2140 2141 2142 2143 2144 2145 2146 2147
{
	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;
}

Y
Yan Zheng 已提交
2148 2149 2150 2151 2152
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)
2153
{
2154
	struct btrfs_fs_info *info = extent_root->fs_info;
2155
	struct btrfs_device *device = NULL;
Y
Yan Zheng 已提交
2156
	struct btrfs_fs_devices *fs_devices = info->fs_devices;
2157
	struct list_head *cur;
Y
Yan Zheng 已提交
2158
	struct map_lookup *map = NULL;
2159 2160
	struct extent_map_tree *em_tree;
	struct extent_map *em;
Y
Yan Zheng 已提交
2161
	struct list_head private_devs;
2162
	int min_stripe_size = 1 * 1024 * 1024;
2163
	u64 calc_size = 1024 * 1024 * 1024;
2164 2165
	u64 max_chunk_size = calc_size;
	u64 min_free;
2166 2167
	u64 avail;
	u64 max_avail = 0;
Y
Yan Zheng 已提交
2168
	u64 dev_offset;
2169
	int num_stripes = 1;
2170
	int min_stripes = 1;
C
Chris Mason 已提交
2171
	int sub_stripes = 0;
2172
	int looped = 0;
2173
	int ret;
2174
	int index;
2175
	int stripe_len = 64 * 1024;
2176

2177 2178 2179 2180 2181
	if ((type & BTRFS_BLOCK_GROUP_RAID1) &&
	    (type & BTRFS_BLOCK_GROUP_DUP)) {
		WARN_ON(1);
		type &= ~BTRFS_BLOCK_GROUP_DUP;
	}
Y
Yan Zheng 已提交
2182
	if (list_empty(&fs_devices->alloc_list))
2183
		return -ENOSPC;
2184

2185
	if (type & (BTRFS_BLOCK_GROUP_RAID0)) {
Y
Yan Zheng 已提交
2186
		num_stripes = fs_devices->rw_devices;
2187 2188 2189
		min_stripes = 2;
	}
	if (type & (BTRFS_BLOCK_GROUP_DUP)) {
2190
		num_stripes = 2;
2191 2192
		min_stripes = 2;
	}
2193
	if (type & (BTRFS_BLOCK_GROUP_RAID1)) {
Y
Yan Zheng 已提交
2194
		num_stripes = min_t(u64, 2, fs_devices->rw_devices);
2195 2196
		if (num_stripes < 2)
			return -ENOSPC;
2197
		min_stripes = 2;
2198
	}
C
Chris Mason 已提交
2199
	if (type & (BTRFS_BLOCK_GROUP_RAID10)) {
Y
Yan Zheng 已提交
2200
		num_stripes = fs_devices->rw_devices;
C
Chris Mason 已提交
2201 2202 2203 2204
		if (num_stripes < 4)
			return -ENOSPC;
		num_stripes &= ~(u32)1;
		sub_stripes = 2;
2205
		min_stripes = 4;
C
Chris Mason 已提交
2206
	}
2207 2208 2209

	if (type & BTRFS_BLOCK_GROUP_DATA) {
		max_chunk_size = 10 * calc_size;
2210
		min_stripe_size = 64 * 1024 * 1024;
2211
	} else if (type & BTRFS_BLOCK_GROUP_METADATA) {
J
Josef Bacik 已提交
2212
		max_chunk_size = 256 * 1024 * 1024;
2213 2214 2215 2216 2217
		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;
2218 2219
	}

Y
Yan Zheng 已提交
2220 2221 2222
	/* 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);
2223

2224
again:
2225
	max_avail = 0;
Y
Yan Zheng 已提交
2226 2227 2228 2229 2230 2231 2232 2233
	if (!map || map->num_stripes != num_stripes) {
		kfree(map);
		map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
		if (!map)
			return -ENOMEM;
		map->num_stripes = num_stripes;
	}

2234 2235 2236 2237 2238 2239 2240
	if (calc_size * num_stripes > max_chunk_size) {
		calc_size = max_chunk_size;
		do_div(calc_size, num_stripes);
		do_div(calc_size, stripe_len);
		calc_size *= stripe_len;
	}
	/* we don't want tiny stripes */
2241
	calc_size = max_t(u64, min_stripe_size, calc_size);
2242 2243 2244 2245

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

Y
Yan Zheng 已提交
2246
	cur = fs_devices->alloc_list.next;
2247
	index = 0;
2248 2249 2250

	if (type & BTRFS_BLOCK_GROUP_DUP)
		min_free = calc_size * 2;
2251 2252
	else
		min_free = calc_size;
2253

J
Josef Bacik 已提交
2254 2255 2256 2257 2258 2259 2260
	/*
	 * we add 1MB because we never use the first 1MB of the device, unless
	 * we've looped, then we are likely allocating the maximum amount of
	 * space left already
	 */
	if (!looped)
		min_free += 1024 * 1024;
2261

Y
Yan Zheng 已提交
2262
	INIT_LIST_HEAD(&private_devs);
C
Chris Mason 已提交
2263
	while (index < num_stripes) {
2264
		device = list_entry(cur, struct btrfs_device, dev_alloc_list);
Y
Yan Zheng 已提交
2265
		BUG_ON(!device->writeable);
2266 2267 2268 2269
		if (device->total_bytes > device->bytes_used)
			avail = device->total_bytes - device->bytes_used;
		else
			avail = 0;
2270
		cur = cur->next;
2271

2272
		if (device->in_fs_metadata && avail >= min_free) {
Y
Yan Zheng 已提交
2273
			ret = find_free_dev_extent(trans, device,
2274 2275
						   min_free, &dev_offset,
						   &max_avail);
2276 2277 2278
			if (ret == 0) {
				list_move_tail(&device->dev_alloc_list,
					       &private_devs);
Y
Yan Zheng 已提交
2279 2280
				map->stripes[index].dev = device;
				map->stripes[index].physical = dev_offset;
2281
				index++;
Y
Yan Zheng 已提交
2282 2283 2284 2285
				if (type & BTRFS_BLOCK_GROUP_DUP) {
					map->stripes[index].dev = device;
					map->stripes[index].physical =
						dev_offset + calc_size;
2286
					index++;
Y
Yan Zheng 已提交
2287
				}
2288
			}
2289
		} else if (device->in_fs_metadata && avail > max_avail)
2290
			max_avail = avail;
Y
Yan Zheng 已提交
2291
		if (cur == &fs_devices->alloc_list)
2292 2293
			break;
	}
Y
Yan Zheng 已提交
2294
	list_splice(&private_devs, &fs_devices->alloc_list);
2295
	if (index < num_stripes) {
2296 2297 2298 2299 2300 2301 2302 2303 2304
		if (index >= min_stripes) {
			num_stripes = index;
			if (type & (BTRFS_BLOCK_GROUP_RAID10)) {
				num_stripes /= sub_stripes;
				num_stripes *= sub_stripes;
			}
			looped = 1;
			goto again;
		}
2305 2306 2307 2308 2309
		if (!looped && max_avail > 0) {
			looped = 1;
			calc_size = max_avail;
			goto again;
		}
Y
Yan Zheng 已提交
2310
		kfree(map);
2311 2312
		return -ENOSPC;
	}
Y
Yan Zheng 已提交
2313 2314 2315 2316 2317 2318 2319
	map->sector_size = extent_root->sectorsize;
	map->stripe_len = stripe_len;
	map->io_align = stripe_len;
	map->io_width = stripe_len;
	map->type = type;
	map->num_stripes = num_stripes;
	map->sub_stripes = sub_stripes;
2320

Y
Yan Zheng 已提交
2321 2322 2323 2324
	*map_ret = map;
	*stripe_size = calc_size;
	*num_bytes = chunk_bytes_by_type(type, calc_size,
					 num_stripes, sub_stripes);
2325

Y
Yan Zheng 已提交
2326 2327 2328
	em = alloc_extent_map(GFP_NOFS);
	if (!em) {
		kfree(map);
2329 2330
		return -ENOMEM;
	}
Y
Yan Zheng 已提交
2331 2332 2333 2334 2335
	em->bdev = (struct block_device *)map;
	em->start = start;
	em->len = *num_bytes;
	em->block_start = 0;
	em->block_len = em->len;
2336

Y
Yan Zheng 已提交
2337
	em_tree = &extent_root->fs_info->mapping_tree.map_tree;
2338
	write_lock(&em_tree->lock);
Y
Yan Zheng 已提交
2339
	ret = add_extent_mapping(em_tree, em);
2340
	write_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
2341 2342
	BUG_ON(ret);
	free_extent_map(em);
2343

Y
Yan Zheng 已提交
2344 2345 2346 2347
	ret = btrfs_make_block_group(trans, extent_root, 0, type,
				     BTRFS_FIRST_CHUNK_TREE_OBJECTID,
				     start, *num_bytes);
	BUG_ON(ret);
2348

Y
Yan Zheng 已提交
2349 2350 2351 2352
	index = 0;
	while (index < map->num_stripes) {
		device = map->stripes[index].dev;
		dev_offset = map->stripes[index].physical;
2353 2354

		ret = btrfs_alloc_dev_extent(trans, device,
Y
Yan Zheng 已提交
2355 2356 2357
				info->chunk_root->root_key.objectid,
				BTRFS_FIRST_CHUNK_TREE_OBJECTID,
				start, dev_offset, calc_size);
2358
		BUG_ON(ret);
Y
Yan Zheng 已提交
2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387
		index++;
	}

	return 0;
}

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;
2388 2389
		ret = btrfs_update_device(trans, device);
		BUG_ON(ret);
Y
Yan Zheng 已提交
2390 2391 2392 2393 2394 2395 2396 2397
		index++;
	}

	index = 0;
	stripe = &chunk->stripe;
	while (index < map->num_stripes) {
		device = map->stripes[index].dev;
		dev_offset = map->stripes[index].physical;
2398

2399 2400 2401
		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
2402
		stripe++;
2403 2404 2405
		index++;
	}

Y
Yan Zheng 已提交
2406
	btrfs_set_stack_chunk_length(chunk, chunk_size);
2407
	btrfs_set_stack_chunk_owner(chunk, extent_root->root_key.objectid);
Y
Yan Zheng 已提交
2408 2409 2410 2411 2412
	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);
2413
	btrfs_set_stack_chunk_sector_size(chunk, extent_root->sectorsize);
Y
Yan Zheng 已提交
2414
	btrfs_set_stack_chunk_sub_stripes(chunk, map->sub_stripes);
2415

Y
Yan Zheng 已提交
2416 2417 2418
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.type = BTRFS_CHUNK_ITEM_KEY;
	key.offset = chunk_offset;
2419

Y
Yan Zheng 已提交
2420 2421
	ret = btrfs_insert_item(trans, chunk_root, &key, chunk, item_size);
	BUG_ON(ret);
2422

Y
Yan Zheng 已提交
2423 2424 2425
	if (map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
		ret = btrfs_add_system_chunk(trans, chunk_root, &key, chunk,
					     item_size);
2426 2427
		BUG_ON(ret);
	}
2428
	kfree(chunk);
Y
Yan Zheng 已提交
2429 2430
	return 0;
}
2431

Y
Yan Zheng 已提交
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
/*
 * 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
2465
static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
Y
Yan Zheng 已提交
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
					 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);
2523
	BUG_ON(ret);
Y
Yan Zheng 已提交
2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534
	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;

2535
	read_lock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
2536
	em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
2537
	read_unlock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
2538 2539 2540
	if (!em)
		return 1;

2541 2542 2543 2544 2545
	if (btrfs_test_opt(root, DEGRADED)) {
		free_extent_map(em);
		return 0;
	}

Y
Yan Zheng 已提交
2546 2547 2548 2549 2550 2551 2552
	map = (struct map_lookup *)em->bdev;
	for (i = 0; i < map->num_stripes; i++) {
		if (!map->stripes[i].dev->writeable) {
			readonly = 1;
			break;
		}
	}
2553
	free_extent_map(em);
Y
Yan Zheng 已提交
2554
	return readonly;
2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565
}

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
2566
	while (1) {
2567
		write_lock(&tree->map_tree.lock);
2568 2569 2570
		em = lookup_extent_mapping(&tree->map_tree, 0, (u64)-1);
		if (em)
			remove_extent_mapping(&tree->map_tree, em);
2571
		write_unlock(&tree->map_tree.lock);
2572 2573 2574 2575 2576 2577 2578 2579 2580 2581
		if (!em)
			break;
		kfree(em->bdev);
		/* once for us */
		free_extent_map(em);
		/* once for the tree */
		free_extent_map(em);
	}
}

2582 2583 2584 2585 2586 2587 2588
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;

2589
	read_lock(&em_tree->lock);
2590
	em = lookup_extent_mapping(em_tree, logical, len);
2591
	read_unlock(&em_tree->lock);
2592 2593 2594 2595 2596 2597
	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
2598 2599
	else if (map->type & BTRFS_BLOCK_GROUP_RAID10)
		ret = map->sub_stripes;
2600 2601 2602 2603 2604 2605
	else
		ret = 1;
	free_extent_map(em);
	return ret;
}

2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621
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;
}

2622 2623 2624 2625
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)
2626 2627 2628 2629 2630
{
	struct extent_map *em;
	struct map_lookup *map;
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	u64 offset;
2631 2632
	u64 stripe_offset;
	u64 stripe_nr;
2633
	int stripes_allocated = 8;
C
Chris Mason 已提交
2634
	int stripes_required = 1;
2635
	int stripe_index;
2636
	int i;
2637
	int num_stripes;
2638
	int max_errors = 0;
2639
	struct btrfs_multi_bio *multi = NULL;
2640

C
Chris Mason 已提交
2641
	if (multi_ret && !(rw & (1 << BIO_RW)))
2642 2643 2644 2645 2646 2647 2648
		stripes_allocated = 1;
again:
	if (multi_ret) {
		multi = kzalloc(btrfs_multi_bio_size(stripes_allocated),
				GFP_NOFS);
		if (!multi)
			return -ENOMEM;
2649 2650

		atomic_set(&multi->error, 0);
2651
	}
2652

2653
	read_lock(&em_tree->lock);
2654
	em = lookup_extent_mapping(em_tree, logical, *length);
2655
	read_unlock(&em_tree->lock);
2656

2657 2658
	if (!em && unplug_page) {
		kfree(multi);
2659
		return 0;
2660
	}
2661

2662
	if (!em) {
C
Chris Mason 已提交
2663 2664 2665
		printk(KERN_CRIT "unable to find logical %llu len %llu\n",
		       (unsigned long long)logical,
		       (unsigned long long)*length);
2666
		BUG();
2667
	}
2668 2669 2670 2671

	BUG_ON(em->start > logical || em->start + em->len < logical);
	map = (struct map_lookup *)em->bdev;
	offset = logical - em->start;
2672

2673 2674 2675
	if (mirror_num > map->num_stripes)
		mirror_num = 0;

2676
	/* if our multi bio struct is too small, back off and try again */
C
Chris Mason 已提交
2677 2678 2679 2680
	if (rw & (1 << BIO_RW)) {
		if (map->type & (BTRFS_BLOCK_GROUP_RAID1 |
				 BTRFS_BLOCK_GROUP_DUP)) {
			stripes_required = map->num_stripes;
2681
			max_errors = 1;
C
Chris Mason 已提交
2682 2683
		} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
			stripes_required = map->sub_stripes;
2684
			max_errors = 1;
C
Chris Mason 已提交
2685 2686
		}
	}
2687
	if (multi_ret && (rw & (1 << BIO_RW)) &&
C
Chris Mason 已提交
2688
	    stripes_allocated < stripes_required) {
2689 2690 2691 2692 2693
		stripes_allocated = map->num_stripes;
		free_extent_map(em);
		kfree(multi);
		goto again;
	}
2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706
	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;

2707
	if (map->type & (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1 |
C
Chris Mason 已提交
2708
			 BTRFS_BLOCK_GROUP_RAID10 |
2709 2710 2711 2712 2713 2714 2715
			 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;
	}
2716 2717

	if (!multi_ret && !unplug_page)
2718 2719
		goto out;

2720
	num_stripes = 1;
2721
	stripe_index = 0;
2722
	if (map->type & BTRFS_BLOCK_GROUP_RAID1) {
2723 2724
		if (unplug_page || (rw & (1 << BIO_RW)))
			num_stripes = map->num_stripes;
2725
		else if (mirror_num)
2726
			stripe_index = mirror_num - 1;
2727 2728 2729 2730 2731
		else {
			stripe_index = find_live_mirror(map, 0,
					    map->num_stripes,
					    current->pid % map->num_stripes);
		}
2732

2733
	} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
2734
		if (rw & (1 << BIO_RW))
2735
			num_stripes = map->num_stripes;
2736 2737
		else if (mirror_num)
			stripe_index = mirror_num - 1;
2738

C
Chris Mason 已提交
2739 2740 2741 2742 2743 2744
	} 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;

2745 2746
		if (unplug_page || (rw & (1 << BIO_RW)))
			num_stripes = map->sub_stripes;
C
Chris Mason 已提交
2747 2748
		else if (mirror_num)
			stripe_index += mirror_num - 1;
2749 2750 2751 2752 2753
		else {
			stripe_index = find_live_mirror(map, stripe_index,
					      map->sub_stripes, stripe_index +
					      current->pid % map->sub_stripes);
		}
2754 2755 2756 2757 2758 2759 2760 2761
	} 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);
	}
2762
	BUG_ON(stripe_index >= map->num_stripes);
2763

2764 2765 2766 2767 2768 2769
	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;
2770 2771
			if (device->bdev) {
				bdi = blk_get_backing_dev_info(device->bdev);
C
Chris Mason 已提交
2772
				if (bdi->unplug_io_fn)
2773
					bdi->unplug_io_fn(bdi, unplug_page);
2774 2775 2776 2777 2778 2779 2780
			}
		} 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;
		}
2781
		stripe_index++;
2782
	}
2783 2784 2785
	if (multi_ret) {
		*multi_ret = multi;
		multi->num_stripes = num_stripes;
2786
		multi->max_errors = max_errors;
2787
	}
2788
out:
2789 2790 2791 2792
	free_extent_map(em);
	return 0;
}

2793 2794 2795 2796 2797 2798 2799 2800
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
2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813
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;

2814
	read_lock(&em_tree->lock);
Y
Yan Zheng 已提交
2815
	em = lookup_extent_mapping(em_tree, chunk_start, 1);
2816
	read_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846

	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;
2847
		WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
2848 2849 2850 2851
		for (j = 0; j < nr; j++) {
			if (buf[j] == bytenr)
				break;
		}
2852 2853
		if (j == nr) {
			WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
2854
			buf[nr++] = bytenr;
2855
		}
Y
Yan Zheng 已提交
2856 2857 2858 2859 2860 2861 2862 2863 2864 2865
	}

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

	free_extent_map(em);
	return 0;
}

2866 2867 2868 2869 2870 2871 2872 2873
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);
}

2874 2875
static void end_bio_multi_stripe(struct bio *bio, int err)
{
2876
	struct btrfs_multi_bio *multi = bio->bi_private;
2877
	int is_orig_bio = 0;
2878 2879

	if (err)
2880
		atomic_inc(&multi->error);
2881

2882 2883 2884
	if (bio == multi->orig_bio)
		is_orig_bio = 1;

2885
	if (atomic_dec_and_test(&multi->stripes_pending)) {
2886 2887 2888 2889
		if (!is_orig_bio) {
			bio_put(bio);
			bio = multi->orig_bio;
		}
2890 2891
		bio->bi_private = multi->private;
		bio->bi_end_io = multi->end_io;
2892 2893 2894
		/* only send an error to the higher layers if it is
		 * beyond the tolerance of the multi-bio
		 */
2895
		if (atomic_read(&multi->error) > multi->max_errors) {
2896
			err = -EIO;
2897 2898 2899 2900 2901 2902
		} 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);
2903
			err = 0;
2904
		}
2905 2906 2907
		kfree(multi);

		bio_endio(bio, err);
2908
	} else if (!is_orig_bio) {
2909 2910 2911 2912
		bio_put(bio);
	}
}

2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926
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
2927
static noinline int schedule_bio(struct btrfs_root *root,
2928 2929
				 struct btrfs_device *device,
				 int rw, struct bio *bio)
2930 2931
{
	int should_queue = 1;
2932
	struct btrfs_pending_bios *pending_bios;
2933 2934 2935

	/* don't bother with additional async steps for reads, right now */
	if (!(rw & (1 << BIO_RW))) {
2936
		bio_get(bio);
2937
		submit_bio(rw, bio);
2938
		bio_put(bio);
2939 2940 2941 2942
		return 0;
	}

	/*
2943
	 * nr_async_bios allows us to reliably return congestion to the
2944 2945 2946 2947
	 * 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
	 */
2948
	atomic_inc(&root->fs_info->nr_async_bios);
2949
	WARN_ON(bio->bi_next);
2950 2951 2952 2953
	bio->bi_next = NULL;
	bio->bi_rw |= rw;

	spin_lock(&device->io_lock);
2954
	if (bio_rw_flagged(bio, BIO_RW_SYNCIO))
2955 2956 2957
		pending_bios = &device->pending_sync_bios;
	else
		pending_bios = &device->pending_bios;
2958

2959 2960
	if (pending_bios->tail)
		pending_bios->tail->bi_next = bio;
2961

2962 2963 2964
	pending_bios->tail = bio;
	if (!pending_bios->head)
		pending_bios->head = bio;
2965 2966 2967 2968 2969 2970
	if (device->running_pending)
		should_queue = 0;

	spin_unlock(&device->io_lock);

	if (should_queue)
2971 2972
		btrfs_queue_worker(&root->fs_info->submit_workers,
				   &device->work);
2973 2974 2975
	return 0;
}

2976
int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
2977
		  int mirror_num, int async_submit)
2978 2979 2980
{
	struct btrfs_mapping_tree *map_tree;
	struct btrfs_device *dev;
2981
	struct bio *first_bio = bio;
2982
	u64 logical = (u64)bio->bi_sector << 9;
2983 2984
	u64 length = 0;
	u64 map_length;
2985
	struct btrfs_multi_bio *multi = NULL;
2986
	int ret;
2987 2988
	int dev_nr = 0;
	int total_devs = 1;
2989

2990
	length = bio->bi_size;
2991 2992
	map_tree = &root->fs_info->mapping_tree;
	map_length = length;
2993

2994 2995
	ret = btrfs_map_block(map_tree, rw, logical, &map_length, &multi,
			      mirror_num);
2996 2997 2998 2999
	BUG_ON(ret);

	total_devs = multi->num_stripes;
	if (map_length < length) {
C
Chris Mason 已提交
3000 3001 3002 3003
		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);
3004 3005 3006 3007
		BUG();
	}
	multi->end_io = first_bio->bi_end_io;
	multi->private = first_bio->bi_private;
3008
	multi->orig_bio = first_bio;
3009 3010
	atomic_set(&multi->stripes_pending, multi->num_stripes);

C
Chris Mason 已提交
3011
	while (dev_nr < total_devs) {
3012 3013 3014 3015 3016 3017 3018 3019 3020 3021
		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;
		}
3022 3023
		bio->bi_sector = multi->stripes[dev_nr].physical >> 9;
		dev = multi->stripes[dev_nr].dev;
Y
Yan Zheng 已提交
3024
		BUG_ON(rw == WRITE && !dev->writeable);
3025 3026
		if (dev && dev->bdev) {
			bio->bi_bdev = dev->bdev;
3027 3028 3029 3030
			if (async_submit)
				schedule_bio(root, dev, rw, bio);
			else
				submit_bio(rw, bio);
3031 3032 3033 3034 3035
		} else {
			bio->bi_bdev = root->fs_info->fs_devices->latest_bdev;
			bio->bi_sector = logical >> 9;
			bio_endio(bio, -EIO);
		}
3036 3037
		dev_nr++;
	}
3038 3039
	if (total_devs == 1)
		kfree(multi);
3040 3041 3042
	return 0;
}

3043
struct btrfs_device *btrfs_find_device(struct btrfs_root *root, u64 devid,
Y
Yan Zheng 已提交
3044
				       u8 *uuid, u8 *fsid)
3045
{
Y
Yan Zheng 已提交
3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060
	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;
3061 3062
}

3063 3064 3065 3066 3067 3068 3069
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);
3070 3071
	if (!device)
		return NULL;
3072 3073 3074 3075 3076
	list_add(&device->dev_list,
		 &fs_devices->devices);
	device->barriers = 1;
	device->dev_root = root->fs_info->dev_root;
	device->devid = devid;
3077
	device->work.func = pending_bios_fn;
Y
Yan Zheng 已提交
3078
	device->fs_devices = fs_devices;
3079 3080
	fs_devices->num_devices++;
	spin_lock_init(&device->io_lock);
3081
	INIT_LIST_HEAD(&device->dev_alloc_list);
3082 3083 3084 3085
	memcpy(device->uuid, dev_uuid, BTRFS_UUID_SIZE);
	return device;
}

3086 3087 3088 3089 3090 3091 3092 3093 3094 3095
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;
3096
	u8 uuid[BTRFS_UUID_SIZE];
3097
	int num_stripes;
3098
	int ret;
3099
	int i;
3100

3101 3102
	logical = key->offset;
	length = btrfs_chunk_length(leaf, chunk);
3103

3104
	read_lock(&map_tree->map_tree.lock);
3105
	em = lookup_extent_mapping(&map_tree->map_tree, logical, 1);
3106
	read_unlock(&map_tree->map_tree.lock);
3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118

	/* 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;
3119 3120
	num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
3121 3122 3123 3124 3125 3126 3127 3128 3129
	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
3130
	em->block_len = em->len;
3131

3132 3133 3134 3135 3136 3137
	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
3138
	map->sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
3139 3140 3141 3142
	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);
3143 3144 3145
		read_extent_buffer(leaf, uuid, (unsigned long)
				   btrfs_stripe_dev_uuid_nr(chunk, i),
				   BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3146 3147
		map->stripes[i].dev = btrfs_find_device(root, devid, uuid,
							NULL);
3148
		if (!map->stripes[i].dev && !btrfs_test_opt(root, DEGRADED)) {
3149 3150 3151 3152
			kfree(map);
			free_extent_map(em);
			return -EIO;
		}
3153 3154 3155 3156 3157 3158 3159 3160 3161 3162
		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;
3163 3164
	}

3165
	write_lock(&map_tree->map_tree.lock);
3166
	ret = add_extent_mapping(&map_tree->map_tree, em);
3167
	write_unlock(&map_tree->map_tree.lock);
3168
	BUG_ON(ret);
3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180
	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);
3181 3182
	device->disk_total_bytes = btrfs_device_total_bytes(leaf, dev_item);
	device->total_bytes = device->disk_total_bytes;
3183 3184 3185 3186 3187 3188 3189
	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);
3190
	read_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
3191 3192 3193 3194

	return 0;
}

Y
Yan Zheng 已提交
3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215
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
3216 3217 3218 3219

	fs_devices = clone_fs_devices(fs_devices);
	if (IS_ERR(fs_devices)) {
		ret = PTR_ERR(fs_devices);
Y
Yan Zheng 已提交
3220 3221 3222
		goto out;
	}

3223
	ret = __btrfs_open_devices(fs_devices, FMODE_READ,
3224
				   root->fs_info->bdev_holder);
Y
Yan Zheng 已提交
3225 3226 3227 3228 3229
	if (ret)
		goto out;

	if (!fs_devices->seeding) {
		__btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
3230
		free_fs_devices(fs_devices);
Y
Yan Zheng 已提交
3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241
		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;
}

3242
static int read_one_dev(struct btrfs_root *root,
3243 3244 3245 3246 3247 3248
			struct extent_buffer *leaf,
			struct btrfs_dev_item *dev_item)
{
	struct btrfs_device *device;
	u64 devid;
	int ret;
Y
Yan Zheng 已提交
3249
	u8 fs_uuid[BTRFS_UUID_SIZE];
3250 3251
	u8 dev_uuid[BTRFS_UUID_SIZE];

3252
	devid = btrfs_device_id(leaf, dev_item);
3253 3254 3255
	read_extent_buffer(leaf, dev_uuid,
			   (unsigned long)btrfs_device_uuid(dev_item),
			   BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3256 3257 3258 3259 3260 3261
	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
3262
		if (ret && !btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
3263 3264 3265 3266 3267
			return ret;
	}

	device = btrfs_find_device(root, devid, dev_uuid, fs_uuid);
	if (!device || !device->bdev) {
Y
Yan Zheng 已提交
3268
		if (!btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
3269 3270 3271
			return -EIO;

		if (!device) {
C
Chris Mason 已提交
3272 3273
			printk(KERN_WARNING "warning devid %llu missing\n",
			       (unsigned long long)devid);
Y
Yan Zheng 已提交
3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284
			device = add_missing_dev(root, devid, dev_uuid);
			if (!device)
				return -ENOMEM;
		}
	}

	if (device->fs_devices != root->fs_info->fs_devices) {
		BUG_ON(device->writeable);
		if (device->generation !=
		    btrfs_device_generation(leaf, dev_item))
			return -EINVAL;
3285
	}
3286 3287 3288

	fill_device_from_item(leaf, dev_item, device);
	device->dev_root = root->fs_info->dev_root;
3289
	device->in_fs_metadata = 1;
Y
Yan Zheng 已提交
3290 3291
	if (device->writeable)
		device->fs_devices->total_rw_bytes += device->total_bytes;
3292 3293 3294 3295
	ret = 0;
	return ret;
}

3296 3297 3298 3299 3300 3301 3302 3303 3304
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
3305
int btrfs_read_sys_array(struct btrfs_root *root)
3306 3307
{
	struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
3308
	struct extent_buffer *sb;
3309 3310
	struct btrfs_disk_key *disk_key;
	struct btrfs_chunk *chunk;
3311 3312 3313
	u8 *ptr;
	unsigned long sb_ptr;
	int ret = 0;
3314 3315 3316 3317
	u32 num_stripes;
	u32 array_size;
	u32 len = 0;
	u32 cur;
3318
	struct btrfs_key key;
3319

Y
Yan Zheng 已提交
3320
	sb = btrfs_find_create_tree_block(root, BTRFS_SUPER_INFO_OFFSET,
3321 3322 3323 3324
					  BTRFS_SUPER_INFO_SIZE);
	if (!sb)
		return -ENOMEM;
	btrfs_set_buffer_uptodate(sb);
3325 3326
	btrfs_set_buffer_lockdep_class(sb, 0);

3327
	write_extent_buffer(sb, super_copy, 0, BTRFS_SUPER_INFO_SIZE);
3328 3329 3330 3331 3332 3333 3334 3335 3336 3337
	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);

3338
		len = sizeof(*disk_key); ptr += len;
3339 3340 3341
		sb_ptr += len;
		cur += len;

3342
		if (key.type == BTRFS_CHUNK_ITEM_KEY) {
3343
			chunk = (struct btrfs_chunk *)sb_ptr;
3344
			ret = read_one_chunk(root, &key, sb, chunk);
3345 3346
			if (ret)
				break;
3347 3348 3349
			num_stripes = btrfs_chunk_num_stripes(sb, chunk);
			len = btrfs_chunk_item_size(num_stripes);
		} else {
3350 3351
			ret = -EIO;
			break;
3352 3353 3354 3355 3356
		}
		ptr += len;
		sb_ptr += len;
		cur += len;
	}
3357
	free_extent_buffer(sb);
3358
	return ret;
3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384
}

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);
C
Chris Mason 已提交
3385
	while (1) {
3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403
		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);
3404
				ret = read_one_dev(root, leaf, dev_item);
Y
Yan Zheng 已提交
3405 3406
				if (ret)
					goto error;
3407 3408 3409 3410 3411
			}
		} 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
3412 3413
			if (ret)
				goto error;
3414 3415 3416 3417 3418 3419 3420 3421 3422 3423
		}
		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
3424
	btrfs_free_path(path);
3425 3426
	return ret;
}