volumes.c 160.8 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */
#include <linux/sched.h>
#include <linux/bio.h>
20
#include <linux/slab.h>
21
#include <linux/buffer_head.h>
22
#include <linux/blkdev.h>
23
#include <linux/random.h>
24
#include <linux/iocontext.h>
25
#include <linux/capability.h>
26
#include <linux/ratelimit.h>
27
#include <linux/kthread.h>
D
David Woodhouse 已提交
28
#include <linux/raid/pq.h>
29
#include <linux/semaphore.h>
D
David Woodhouse 已提交
30
#include <asm/div64.h>
31 32 33 34 35 36
#include "ctree.h"
#include "extent_map.h"
#include "disk-io.h"
#include "transaction.h"
#include "print-tree.h"
#include "volumes.h"
D
David Woodhouse 已提交
37
#include "raid56.h"
38
#include "async-thread.h"
39
#include "check-integrity.h"
40
#include "rcu-string.h"
41
#include "math.h"
42
#include "dev-replace.h"
43

Y
Yan Zheng 已提交
44 45 46 47
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);
48
static void __btrfs_reset_dev_stats(struct btrfs_device *dev);
49
static void btrfs_dev_stat_print_on_error(struct btrfs_device *dev);
50
static void btrfs_dev_stat_print_on_load(struct btrfs_device *device);
Y
Yan Zheng 已提交
51

52 53 54
static DEFINE_MUTEX(uuid_mutex);
static LIST_HEAD(fs_uuids);

55 56 57 58 59 60 61 62 63 64
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);
}

65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106
static struct btrfs_fs_devices *__alloc_fs_devices(void)
{
	struct btrfs_fs_devices *fs_devs;

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

	mutex_init(&fs_devs->device_list_mutex);

	INIT_LIST_HEAD(&fs_devs->devices);
	INIT_LIST_HEAD(&fs_devs->alloc_list);
	INIT_LIST_HEAD(&fs_devs->list);

	return fs_devs;
}

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

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

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

	return fs_devs;
}

Y
Yan Zheng 已提交
107 108 109 110 111 112 113 114
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);
115
		rcu_string_free(device->name);
Y
Yan Zheng 已提交
116 117 118 119 120
		kfree(device);
	}
	kfree(fs_devices);
}

121 122 123 124 125 126 127
static void btrfs_kobject_uevent(struct block_device *bdev,
				 enum kobject_action action)
{
	int ret;

	ret = kobject_uevent(&disk_to_dev(bdev->bd_disk)->kobj, action);
	if (ret)
128
		pr_warn("BTRFS: Sending event '%d' to kobject: '%s' (%p): failed\n",
129 130 131 132 133
			action,
			kobject_name(&disk_to_dev(bdev->bd_disk)->kobj),
			&disk_to_dev(bdev->bd_disk)->kobj);
}

134
void btrfs_cleanup_fs_uuids(void)
135 136 137
{
	struct btrfs_fs_devices *fs_devices;

Y
Yan Zheng 已提交
138 139 140 141
	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
142
		free_fs_devices(fs_devices);
143 144 145
	}
}

146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166
static struct btrfs_device *__alloc_device(void)
{
	struct btrfs_device *dev;

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

	INIT_LIST_HEAD(&dev->dev_list);
	INIT_LIST_HEAD(&dev->dev_alloc_list);

	spin_lock_init(&dev->io_lock);

	spin_lock_init(&dev->reada_lock);
	atomic_set(&dev->reada_in_flight, 0);
	INIT_RADIX_TREE(&dev->reada_zones, GFP_NOFS & ~__GFP_WAIT);
	INIT_RADIX_TREE(&dev->reada_extents, GFP_NOFS & ~__GFP_WAIT);

	return dev;
}

167 168
static noinline struct btrfs_device *__find_device(struct list_head *head,
						   u64 devid, u8 *uuid)
169 170 171
{
	struct btrfs_device *dev;

172
	list_for_each_entry(dev, head, dev_list) {
173
		if (dev->devid == devid &&
174
		    (!uuid || !memcmp(dev->uuid, uuid, BTRFS_UUID_SIZE))) {
175
			return dev;
176
		}
177 178 179 180
	}
	return NULL;
}

181
static noinline struct btrfs_fs_devices *find_fsid(u8 *fsid)
182 183 184
{
	struct btrfs_fs_devices *fs_devices;

185
	list_for_each_entry(fs_devices, &fs_uuids, list) {
186 187 188 189 190 191
		if (memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE) == 0)
			return fs_devices;
	}
	return NULL;
}

192 193 194 195 196 197 198 199 200 201 202
static int
btrfs_get_bdev_and_sb(const char *device_path, fmode_t flags, void *holder,
		      int flush, struct block_device **bdev,
		      struct buffer_head **bh)
{
	int ret;

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

	if (IS_ERR(*bdev)) {
		ret = PTR_ERR(*bdev);
203
		printk(KERN_INFO "BTRFS: open %s failed\n", device_path);
204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229
		goto error;
	}

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

	return 0;

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

230 231 232 233 234 235 236 237 238 239 240 241 242 243
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;
}

244 245 246 247 248 249 250 251 252 253 254
/*
 * 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.
 */
255
static noinline void run_scheduled_bios(struct btrfs_device *device)
256 257 258
{
	struct bio *pending;
	struct backing_dev_info *bdi;
259
	struct btrfs_fs_info *fs_info;
260
	struct btrfs_pending_bios *pending_bios;
261 262 263
	struct bio *tail;
	struct bio *cur;
	int again = 0;
264
	unsigned long num_run;
265
	unsigned long batch_run = 0;
266
	unsigned long limit;
267
	unsigned long last_waited = 0;
268
	int force_reg = 0;
269
	int sync_pending = 0;
270 271 272 273 274 275 276 277 278
	struct blk_plug plug;

	/*
	 * this function runs all the bios we've collected for
	 * a particular device.  We don't want to wander off to
	 * another device without first sending all of these down.
	 * So, setup a plug here and finish it off before we return
	 */
	blk_start_plug(&plug);
279

280
	bdi = blk_get_backing_dev_info(device->bdev);
281 282 283 284
	fs_info = device->dev_root->fs_info;
	limit = btrfs_async_submit_limit(fs_info);
	limit = limit * 2 / 3;

285 286 287
loop:
	spin_lock(&device->io_lock);

288
loop_lock:
289
	num_run = 0;
290

291 292 293 294 295
	/* 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
	 */
296
	if (!force_reg && device->pending_sync_bios.head) {
297
		pending_bios = &device->pending_sync_bios;
298 299
		force_reg = 1;
	} else {
300
		pending_bios = &device->pending_bios;
301 302
		force_reg = 0;
	}
303 304 305

	pending = pending_bios->head;
	tail = pending_bios->tail;
306 307 308 309 310 311 312 313 314 315
	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.
	 */
316 317
	if (device->pending_sync_bios.head == NULL &&
	    device->pending_bios.head == NULL) {
318 319
		again = 0;
		device->running_pending = 0;
320 321 322
	} else {
		again = 1;
		device->running_pending = 1;
323
	}
324 325 326 327

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

328 329
	spin_unlock(&device->io_lock);

330
	while (pending) {
331 332

		rmb();
333 334 335 336 337 338 339 340
		/* 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)) {
341 342 343 344 345
			spin_lock(&device->io_lock);
			requeue_list(pending_bios, pending, tail);
			goto loop_lock;
		}

346 347 348
		cur = pending;
		pending = pending->bi_next;
		cur->bi_next = NULL;
349

350
		if (atomic_dec_return(&fs_info->nr_async_bios) < limit &&
351 352
		    waitqueue_active(&fs_info->async_submit_wait))
			wake_up(&fs_info->async_submit_wait);
353 354

		BUG_ON(atomic_read(&cur->bi_cnt) == 0);
355

356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371
		/*
		 * if we're doing the sync list, record that our
		 * plug has some sync requests on it
		 *
		 * If we're doing the regular list and there are
		 * sync requests sitting around, unplug before
		 * we add more
		 */
		if (pending_bios == &device->pending_sync_bios) {
			sync_pending = 1;
		} else if (sync_pending) {
			blk_finish_plug(&plug);
			blk_start_plug(&plug);
			sync_pending = 0;
		}

372
		btrfsic_submit_bio(cur->bi_rw, cur);
373 374
		num_run++;
		batch_run++;
375
		if (need_resched())
376
			cond_resched();
377 378 379 380 381 382

		/*
		 * we made progress, there is more work to do and the bdi
		 * is now congested.  Back off and let other work structs
		 * run instead
		 */
383
		if (pending && bdi_write_congested(bdi) && batch_run > 8 &&
384
		    fs_info->fs_devices->open_devices > 1) {
385
			struct io_context *ioc;
386

387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408
			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;
409
				if (need_resched())
410
					cond_resched();
411 412
				continue;
			}
413
			spin_lock(&device->io_lock);
414
			requeue_list(pending_bios, pending, tail);
415
			device->running_pending = 1;
416 417

			spin_unlock(&device->io_lock);
418 419
			btrfs_queue_work(fs_info->submit_workers,
					 &device->work);
420 421
			goto done;
		}
422 423 424 425 426 427
		/* unplug every 64 requests just for good measure */
		if (batch_run % 64 == 0) {
			blk_finish_plug(&plug);
			blk_start_plug(&plug);
			sync_pending = 0;
		}
428
	}
429

430 431 432 433 434 435 436 437 438
	cond_resched();
	if (again)
		goto loop;

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

439
done:
440
	blk_finish_plug(&plug);
441 442
}

443
static void pending_bios_fn(struct btrfs_work *work)
444 445 446 447 448 449 450
{
	struct btrfs_device *device;

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

451 452 453 454 455 456 457 458
/*
 * Add new device to list of registered devices
 *
 * Returns:
 * 1   - first time device is seen
 * 0   - device already known
 * < 0 - error
 */
459
static noinline int device_list_add(const char *path,
460 461 462 463 464
			   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;
465
	struct rcu_string *name;
466
	int ret = 0;
467 468 469 470
	u64 found_transid = btrfs_super_generation(disk_super);

	fs_devices = find_fsid(disk_super->fsid);
	if (!fs_devices) {
471 472 473 474
		fs_devices = alloc_fs_devices(disk_super->fsid);
		if (IS_ERR(fs_devices))
			return PTR_ERR(fs_devices);

475 476 477
		list_add(&fs_devices->list, &fs_uuids);
		fs_devices->latest_devid = devid;
		fs_devices->latest_trans = found_transid;
478

479 480
		device = NULL;
	} else {
481 482
		device = __find_device(&fs_devices->devices, devid,
				       disk_super->dev_item.uuid);
483 484
	}
	if (!device) {
Y
Yan Zheng 已提交
485 486 487
		if (fs_devices->opened)
			return -EBUSY;

488 489 490
		device = btrfs_alloc_device(NULL, &devid,
					    disk_super->dev_item.uuid);
		if (IS_ERR(device)) {
491
			/* we can safely leave the fs_devices entry around */
492
			return PTR_ERR(device);
493
		}
494 495 496

		name = rcu_string_strdup(path, GFP_NOFS);
		if (!name) {
497 498 499
			kfree(device);
			return -ENOMEM;
		}
500
		rcu_assign_pointer(device->name, name);
501

502
		mutex_lock(&fs_devices->device_list_mutex);
503
		list_add_rcu(&device->dev_list, &fs_devices->devices);
504
		fs_devices->num_devices++;
505 506
		mutex_unlock(&fs_devices->device_list_mutex);

507
		ret = 1;
Y
Yan Zheng 已提交
508
		device->fs_devices = fs_devices;
509 510
	} else if (!device->name || strcmp(device->name->str, path)) {
		name = rcu_string_strdup(path, GFP_NOFS);
511 512
		if (!name)
			return -ENOMEM;
513 514
		rcu_string_free(device->name);
		rcu_assign_pointer(device->name, name);
515 516 517 518
		if (device->missing) {
			fs_devices->missing_devices--;
			device->missing = 0;
		}
519 520 521 522 523 524 525
	}

	if (found_transid > fs_devices->latest_trans) {
		fs_devices->latest_devid = devid;
		fs_devices->latest_trans = found_transid;
	}
	*fs_devices_ret = fs_devices;
526 527

	return ret;
528 529
}

Y
Yan Zheng 已提交
530 531 532 533 534 535
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;

536 537 538
	fs_devices = alloc_fs_devices(orig->fsid);
	if (IS_ERR(fs_devices))
		return fs_devices;
Y
Yan Zheng 已提交
539 540 541

	fs_devices->latest_devid = orig->latest_devid;
	fs_devices->latest_trans = orig->latest_trans;
542
	fs_devices->total_devices = orig->total_devices;
Y
Yan Zheng 已提交
543

544
	/* We have held the volume lock, it is safe to get the devices. */
Y
Yan Zheng 已提交
545
	list_for_each_entry(orig_dev, &orig->devices, dev_list) {
546 547
		struct rcu_string *name;

548 549 550
		device = btrfs_alloc_device(NULL, &orig_dev->devid,
					    orig_dev->uuid);
		if (IS_ERR(device))
Y
Yan Zheng 已提交
551 552
			goto error;

553 554 555 556 557 558
		/*
		 * This is ok to do without rcu read locked because we hold the
		 * uuid mutex so nothing we touch in here is going to disappear.
		 */
		name = rcu_string_strdup(orig_dev->name->str, GFP_NOFS);
		if (!name) {
559
			kfree(device);
Y
Yan Zheng 已提交
560
			goto error;
561
		}
562
		rcu_assign_pointer(device->name, name);
Y
Yan Zheng 已提交
563 564 565 566 567 568 569 570 571 572 573

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

574 575
void btrfs_close_extra_devices(struct btrfs_fs_info *fs_info,
			       struct btrfs_fs_devices *fs_devices, int step)
576
{
577
	struct btrfs_device *device, *next;
578

579 580 581 582
	struct block_device *latest_bdev = NULL;
	u64 latest_devid = 0;
	u64 latest_transid = 0;

583 584
	mutex_lock(&uuid_mutex);
again:
585
	/* This is the initialized path, it is safe to release the devices. */
586
	list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
587
		if (device->in_fs_metadata) {
588 589 590
			if (!device->is_tgtdev_for_dev_replace &&
			    (!latest_transid ||
			     device->generation > latest_transid)) {
591 592 593 594
				latest_devid = device->devid;
				latest_transid = device->generation;
				latest_bdev = device->bdev;
			}
Y
Yan Zheng 已提交
595
			continue;
596
		}
Y
Yan Zheng 已提交
597

598 599 600 601 602 603 604 605 606 607 608 609 610 611 612
		if (device->devid == BTRFS_DEV_REPLACE_DEVID) {
			/*
			 * In the first step, keep the device which has
			 * the correct fsid and the devid that is used
			 * for the dev_replace procedure.
			 * In the second step, the dev_replace state is
			 * read from the device tree and it is known
			 * whether the procedure is really active or
			 * not, which means whether this device is
			 * used or whether it should be removed.
			 */
			if (step == 0 || device->is_tgtdev_for_dev_replace) {
				continue;
			}
		}
Y
Yan Zheng 已提交
613
		if (device->bdev) {
614
			blkdev_put(device->bdev, device->mode);
Y
Yan Zheng 已提交
615 616 617 618 619 620
			device->bdev = NULL;
			fs_devices->open_devices--;
		}
		if (device->writeable) {
			list_del_init(&device->dev_alloc_list);
			device->writeable = 0;
621 622
			if (!device->is_tgtdev_for_dev_replace)
				fs_devices->rw_devices--;
Y
Yan Zheng 已提交
623
		}
Y
Yan Zheng 已提交
624 625
		list_del_init(&device->dev_list);
		fs_devices->num_devices--;
626
		rcu_string_free(device->name);
Y
Yan Zheng 已提交
627
		kfree(device);
628
	}
Y
Yan Zheng 已提交
629 630 631 632 633 634

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

635 636 637 638
	fs_devices->latest_bdev = latest_bdev;
	fs_devices->latest_devid = latest_devid;
	fs_devices->latest_trans = latest_transid;

639 640
	mutex_unlock(&uuid_mutex);
}
641

642 643 644 645 646 647 648 649 650
static void __free_device(struct work_struct *work)
{
	struct btrfs_device *device;

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

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

651
	rcu_string_free(device->name);
652 653 654 655 656 657 658 659 660 661 662 663 664
	kfree(device);
}

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

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

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

Y
Yan Zheng 已提交
665
static int __btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
666 667
{
	struct btrfs_device *device;
Y
Yan Zheng 已提交
668

Y
Yan Zheng 已提交
669 670
	if (--fs_devices->opened > 0)
		return 0;
671

672
	mutex_lock(&fs_devices->device_list_mutex);
673
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
674
		struct btrfs_device *new_device;
675
		struct rcu_string *name;
676 677

		if (device->bdev)
678
			fs_devices->open_devices--;
679

680 681
		if (device->writeable &&
		    device->devid != BTRFS_DEV_REPLACE_DEVID) {
Y
Yan Zheng 已提交
682 683 684 685
			list_del_init(&device->dev_alloc_list);
			fs_devices->rw_devices--;
		}

686 687
		if (device->can_discard)
			fs_devices->num_can_discard--;
688 689
		if (device->missing)
			fs_devices->missing_devices--;
690

691 692 693
		new_device = btrfs_alloc_device(NULL, &device->devid,
						device->uuid);
		BUG_ON(IS_ERR(new_device)); /* -ENOMEM */
694 695

		/* Safe because we are under uuid_mutex */
696 697
		if (device->name) {
			name = rcu_string_strdup(device->name->str, GFP_NOFS);
698
			BUG_ON(!name); /* -ENOMEM */
699 700
			rcu_assign_pointer(new_device->name, name);
		}
701

702
		list_replace_rcu(&device->dev_list, &new_device->dev_list);
703
		new_device->fs_devices = device->fs_devices;
704 705

		call_rcu(&device->rcu, free_device);
706
	}
707 708
	mutex_unlock(&fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
709 710
	WARN_ON(fs_devices->open_devices);
	WARN_ON(fs_devices->rw_devices);
Y
Yan Zheng 已提交
711 712 713
	fs_devices->opened = 0;
	fs_devices->seeding = 0;

714 715 716
	return 0;
}

Y
Yan Zheng 已提交
717 718
int btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
{
Y
Yan Zheng 已提交
719
	struct btrfs_fs_devices *seed_devices = NULL;
Y
Yan Zheng 已提交
720 721 722 723
	int ret;

	mutex_lock(&uuid_mutex);
	ret = __btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
724 725 726 727
	if (!fs_devices->opened) {
		seed_devices = fs_devices->seed;
		fs_devices->seed = NULL;
	}
Y
Yan Zheng 已提交
728
	mutex_unlock(&uuid_mutex);
Y
Yan Zheng 已提交
729 730 731 732 733 734 735

	while (seed_devices) {
		fs_devices = seed_devices;
		seed_devices = fs_devices->seed;
		__btrfs_close_devices(fs_devices);
		free_fs_devices(fs_devices);
	}
736 737 738 739 740 741
	/*
	 * Wait for rcu kworkers under __btrfs_close_devices
	 * to finish all blkdev_puts so device is really
	 * free when umount is done.
	 */
	rcu_barrier();
Y
Yan Zheng 已提交
742 743 744
	return ret;
}

Y
Yan Zheng 已提交
745 746
static int __btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
				fmode_t flags, void *holder)
747
{
748
	struct request_queue *q;
749 750 751
	struct block_device *bdev;
	struct list_head *head = &fs_devices->devices;
	struct btrfs_device *device;
752 753 754 755 756 757
	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
758
	int seeding = 1;
759
	int ret = 0;
760

761 762
	flags |= FMODE_EXCL;

763
	list_for_each_entry(device, head, dev_list) {
764 765
		if (device->bdev)
			continue;
766 767 768
		if (!device->name)
			continue;

769 770 771
		/* Just open everything we can; ignore failures here */
		if (btrfs_get_bdev_and_sb(device->name->str, flags, holder, 1,
					    &bdev, &bh))
772
			continue;
773 774

		disk_super = (struct btrfs_super_block *)bh->b_data;
775
		devid = btrfs_stack_device_id(&disk_super->dev_item);
776 777 778
		if (devid != device->devid)
			goto error_brelse;

Y
Yan Zheng 已提交
779 780 781 782 783 784
		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) {
785
			latest_devid = devid;
Y
Yan Zheng 已提交
786
			latest_transid = device->generation;
787 788 789
			latest_bdev = bdev;
		}

Y
Yan Zheng 已提交
790 791 792 793 794 795 796
		if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_SEEDING) {
			device->writeable = 0;
		} else {
			device->writeable = !bdev_read_only(bdev);
			seeding = 0;
		}

797 798 799 800 801 802
		q = bdev_get_queue(bdev);
		if (blk_queue_discard(q)) {
			device->can_discard = 1;
			fs_devices->num_can_discard++;
		}

803
		device->bdev = bdev;
804
		device->in_fs_metadata = 0;
805 806
		device->mode = flags;

807 808 809
		if (!blk_queue_nonrot(bdev_get_queue(bdev)))
			fs_devices->rotating = 1;

810
		fs_devices->open_devices++;
811 812
		if (device->writeable &&
		    device->devid != BTRFS_DEV_REPLACE_DEVID) {
Y
Yan Zheng 已提交
813 814 815 816
			fs_devices->rw_devices++;
			list_add(&device->dev_alloc_list,
				 &fs_devices->alloc_list);
		}
817
		brelse(bh);
818
		continue;
819

820 821
error_brelse:
		brelse(bh);
822
		blkdev_put(bdev, flags);
823
		continue;
824
	}
825
	if (fs_devices->open_devices == 0) {
826
		ret = -EINVAL;
827 828
		goto out;
	}
Y
Yan Zheng 已提交
829 830
	fs_devices->seeding = seeding;
	fs_devices->opened = 1;
831 832 833
	fs_devices->latest_bdev = latest_bdev;
	fs_devices->latest_devid = latest_devid;
	fs_devices->latest_trans = latest_transid;
Y
Yan Zheng 已提交
834
	fs_devices->total_rw_bytes = 0;
835
out:
Y
Yan Zheng 已提交
836 837 838 839
	return ret;
}

int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
840
		       fmode_t flags, void *holder)
Y
Yan Zheng 已提交
841 842 843 844 845
{
	int ret;

	mutex_lock(&uuid_mutex);
	if (fs_devices->opened) {
Y
Yan Zheng 已提交
846 847
		fs_devices->opened++;
		ret = 0;
Y
Yan Zheng 已提交
848
	} else {
849
		ret = __btrfs_open_devices(fs_devices, flags, holder);
Y
Yan Zheng 已提交
850
	}
851 852 853 854
	mutex_unlock(&uuid_mutex);
	return ret;
}

855 856 857 858 859
/*
 * Look for a btrfs signature on a device. This may be called out of the mount path
 * and we are not allowed to call set_blocksize during the scan. The superblock
 * is read via pagecache
 */
860
int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
861 862 863 864
			  struct btrfs_fs_devices **fs_devices_ret)
{
	struct btrfs_super_block *disk_super;
	struct block_device *bdev;
865 866 867
	struct page *page;
	void *p;
	int ret = -EINVAL;
868
	u64 devid;
869
	u64 transid;
870
	u64 total_devices;
871 872
	u64 bytenr;
	pgoff_t index;
873

874 875 876 877 878 879 880
	/*
	 * we would like to check all the supers, but that would make
	 * a btrfs mount succeed after a mkfs from a different FS.
	 * So, we need to add a special mount option to scan for
	 * later supers, using BTRFS_SUPER_MIRROR_MAX instead
	 */
	bytenr = btrfs_sb_offset(0);
881
	flags |= FMODE_EXCL;
882
	mutex_lock(&uuid_mutex);
883 884 885 886 887

	bdev = blkdev_get_by_path(path, flags, holder);

	if (IS_ERR(bdev)) {
		ret = PTR_ERR(bdev);
888
		goto error;
889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916
	}

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

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

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

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

	if (IS_ERR_OR_NULL(page))
		goto error_bdev_put;

	p = kmap(page);

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

	if (btrfs_super_bytenr(disk_super) != bytenr ||
917
	    btrfs_super_magic(disk_super) != BTRFS_MAGIC)
918 919
		goto error_unmap;

920
	devid = btrfs_stack_device_id(&disk_super->dev_item);
921
	transid = btrfs_super_generation(disk_super);
922
	total_devices = btrfs_super_num_devices(disk_super);
923

924
	ret = device_list_add(path, disk_super, devid, fs_devices_ret);
925 926 927 928 929 930 931 932 933 934 935 936
	if (ret > 0) {
		if (disk_super->label[0]) {
			if (disk_super->label[BTRFS_LABEL_SIZE - 1])
				disk_super->label[BTRFS_LABEL_SIZE - 1] = '\0';
			printk(KERN_INFO "BTRFS: device label %s ", disk_super->label);
		} else {
			printk(KERN_INFO "BTRFS: device fsid %pU ", disk_super->fsid);
		}

		printk(KERN_CONT "devid %llu transid %llu %s\n", devid, transid, path);
		ret = 0;
	}
937 938
	if (!ret && fs_devices_ret)
		(*fs_devices_ret)->total_devices = total_devices;
939 940 941 942 943 944

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

error_bdev_put:
945
	blkdev_put(bdev, flags);
946
error:
947
	mutex_unlock(&uuid_mutex);
948 949
	return ret;
}
950

951 952 953 954 955 956 957 958 959 960 961 962 963 964 965
/* helper to account the used device space in the range */
int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start,
				   u64 end, u64 *length)
{
	struct btrfs_key key;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_dev_extent *dev_extent;
	struct btrfs_path *path;
	u64 extent_end;
	int ret;
	int slot;
	struct extent_buffer *l;

	*length = 0;

966
	if (start >= device->total_bytes || device->is_tgtdev_for_dev_replace)
967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034
		return 0;

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

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

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

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

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

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

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

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

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

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

1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063
static int contains_pending_extent(struct btrfs_trans_handle *trans,
				   struct btrfs_device *device,
				   u64 *start, u64 len)
{
	struct extent_map *em;
	int ret = 0;

	list_for_each_entry(em, &trans->transaction->pending_chunks, list) {
		struct map_lookup *map;
		int i;

		map = (struct map_lookup *)em->bdev;
		for (i = 0; i < map->num_stripes; i++) {
			if (map->stripes[i].dev != device)
				continue;
			if (map->stripes[i].physical >= *start + len ||
			    map->stripes[i].physical + em->orig_block_len <=
			    *start)
				continue;
			*start = map->stripes[i].physical +
				em->orig_block_len;
			ret = 1;
		}
	}

	return ret;
}


1064
/*
1065 1066 1067 1068 1069 1070 1071
 * find_free_dev_extent - find free space in the specified device
 * @device:	the device which we search the free space in
 * @num_bytes:	the size of the free space that we need
 * @start:	store the start of the free space.
 * @len:	the size of the free space. that we find, or the size of the max
 * 		free space if we don't find suitable free space
 *
1072 1073 1074
 * 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
1075 1076 1077 1078 1079 1080 1081 1082
 *
 * @start is used to store the start of the free space if we find. But if we
 * don't find suitable free space, it will be used to store the start position
 * of the max free space.
 *
 * @len is used to store the size of the free space that we find.
 * But if we don't find suitable free space, it is used to store the size of
 * the max free space.
1083
 */
1084 1085
int find_free_dev_extent(struct btrfs_trans_handle *trans,
			 struct btrfs_device *device, u64 num_bytes,
1086
			 u64 *start, u64 *len)
1087 1088 1089
{
	struct btrfs_key key;
	struct btrfs_root *root = device->dev_root;
1090
	struct btrfs_dev_extent *dev_extent;
Y
Yan Zheng 已提交
1091
	struct btrfs_path *path;
1092 1093 1094 1095 1096
	u64 hole_size;
	u64 max_hole_start;
	u64 max_hole_size;
	u64 extent_end;
	u64 search_start;
1097 1098
	u64 search_end = device->total_bytes;
	int ret;
1099
	int slot;
1100 1101 1102 1103
	struct extent_buffer *l;

	/* FIXME use last free of some kind */

1104 1105 1106
	/* we don't want to overwrite the superblock on the drive,
	 * so we make sure to start at an offset of at least 1MB
	 */
A
Arne Jansen 已提交
1107
	search_start = max(root->fs_info->alloc_start, 1024ull * 1024);
1108

1109 1110 1111 1112
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
again:
1113 1114
	max_hole_start = search_start;
	max_hole_size = 0;
1115
	hole_size = 0;
1116

1117
	if (search_start >= search_end || device->is_tgtdev_for_dev_replace) {
1118
		ret = -ENOSPC;
1119
		goto out;
1120 1121 1122
	}

	path->reada = 2;
1123 1124
	path->search_commit_root = 1;
	path->skip_locking = 1;
1125

1126 1127 1128
	key.objectid = device->devid;
	key.offset = search_start;
	key.type = BTRFS_DEV_EXTENT_KEY;
1129

1130
	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1131
	if (ret < 0)
1132
		goto out;
1133 1134 1135
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid, key.type);
		if (ret < 0)
1136
			goto out;
1137
	}
1138

1139 1140 1141 1142 1143 1144 1145 1146
	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)
1147 1148 1149
				goto out;

			break;
1150 1151 1152 1153 1154 1155 1156
		}
		btrfs_item_key_to_cpu(l, &key, slot);

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

		if (key.objectid > device->devid)
1157
			break;
1158

1159 1160
		if (btrfs_key_type(&key) != BTRFS_DEV_EXTENT_KEY)
			goto next;
1161

1162 1163
		if (key.offset > search_start) {
			hole_size = key.offset - search_start;
1164

1165 1166 1167 1168 1169 1170 1171 1172 1173
			/*
			 * Have to check before we set max_hole_start, otherwise
			 * we could end up sending back this offset anyway.
			 */
			if (contains_pending_extent(trans, device,
						    &search_start,
						    hole_size))
				hole_size = 0;

1174 1175 1176 1177
			if (hole_size > max_hole_size) {
				max_hole_start = search_start;
				max_hole_size = hole_size;
			}
1178

1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190
			/*
			 * If this free space is greater than which we need,
			 * it must be the max free space that we have found
			 * until now, so max_hole_start must point to the start
			 * of this free space and the length of this free space
			 * is stored in max_hole_size. Thus, we return
			 * max_hole_start and max_hole_size and go back to the
			 * caller.
			 */
			if (hole_size >= num_bytes) {
				ret = 0;
				goto out;
1191 1192 1193 1194
			}
		}

		dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
1195 1196 1197 1198
		extent_end = key.offset + btrfs_dev_extent_length(l,
								  dev_extent);
		if (extent_end > search_start)
			search_start = extent_end;
1199 1200 1201 1202 1203
next:
		path->slots[0]++;
		cond_resched();
	}

1204 1205 1206 1207 1208 1209 1210 1211
	/*
	 * At this point, search_start should be the end of
	 * allocated dev extents, and when shrinking the device,
	 * search_end may be smaller than search_start.
	 */
	if (search_end > search_start)
		hole_size = search_end - search_start;

1212 1213 1214
	if (hole_size > max_hole_size) {
		max_hole_start = search_start;
		max_hole_size = hole_size;
1215 1216
	}

1217 1218 1219 1220 1221
	if (contains_pending_extent(trans, device, &search_start, hole_size)) {
		btrfs_release_path(path);
		goto again;
	}

1222 1223 1224 1225 1226 1227 1228
	/* See above. */
	if (hole_size < num_bytes)
		ret = -ENOSPC;
	else
		ret = 0;

out:
Y
Yan Zheng 已提交
1229
	btrfs_free_path(path);
1230
	*start = max_hole_start;
1231
	if (len)
1232
		*len = max_hole_size;
1233 1234 1235
	return ret;
}

1236
static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans,
1237 1238 1239 1240 1241 1242 1243
			  struct btrfs_device *device,
			  u64 start)
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_key key;
1244 1245 1246
	struct btrfs_key found_key;
	struct extent_buffer *leaf = NULL;
	struct btrfs_dev_extent *extent = NULL;
1247 1248 1249 1250 1251 1252 1253 1254

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

	key.objectid = device->devid;
	key.offset = start;
	key.type = BTRFS_DEV_EXTENT_KEY;
1255
again:
1256
	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
1257 1258 1259
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid,
					  BTRFS_DEV_EXTENT_KEY);
1260 1261
		if (ret)
			goto out;
1262 1263 1264 1265 1266 1267
		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);
1268 1269 1270
		key = found_key;
		btrfs_release_path(path);
		goto again;
1271 1272 1273 1274
	} else if (ret == 0) {
		leaf = path->nodes[0];
		extent = btrfs_item_ptr(leaf, path->slots[0],
					struct btrfs_dev_extent);
1275 1276 1277
	} else {
		btrfs_error(root->fs_info, ret, "Slot search failed");
		goto out;
1278
	}
1279

1280 1281 1282 1283 1284 1285 1286
	if (device->bytes_used > 0) {
		u64 len = btrfs_dev_extent_length(leaf, extent);
		device->bytes_used -= len;
		spin_lock(&root->fs_info->free_chunk_lock);
		root->fs_info->free_chunk_space += len;
		spin_unlock(&root->fs_info->free_chunk_lock);
	}
1287
	ret = btrfs_del_item(trans, root, path);
1288 1289 1290 1291
	if (ret) {
		btrfs_error(root->fs_info, ret,
			    "Failed to remove dev extent item");
	}
1292
out:
1293 1294 1295 1296
	btrfs_free_path(path);
	return ret;
}

1297 1298 1299 1300
static int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
				  struct btrfs_device *device,
				  u64 chunk_tree, u64 chunk_objectid,
				  u64 chunk_offset, u64 start, u64 num_bytes)
1301 1302 1303 1304 1305 1306 1307 1308
{
	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;

1309
	WARN_ON(!device->in_fs_metadata);
1310
	WARN_ON(device->is_tgtdev_for_dev_replace);
1311 1312 1313 1314 1315
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = device->devid;
Y
Yan Zheng 已提交
1316
	key.offset = start;
1317 1318 1319
	key.type = BTRFS_DEV_EXTENT_KEY;
	ret = btrfs_insert_empty_item(trans, root, path, &key,
				      sizeof(*extent));
1320 1321
	if (ret)
		goto out;
1322 1323 1324 1325

	leaf = path->nodes[0];
	extent = btrfs_item_ptr(leaf, path->slots[0],
				struct btrfs_dev_extent);
1326 1327 1328 1329 1330
	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,
1331
		    btrfs_dev_extent_chunk_tree_uuid(extent), BTRFS_UUID_SIZE);
1332

1333 1334
	btrfs_set_dev_extent_length(leaf, extent, num_bytes);
	btrfs_mark_buffer_dirty(leaf);
1335
out:
1336 1337 1338 1339
	btrfs_free_path(path);
	return ret;
}

1340
static u64 find_next_chunk(struct btrfs_fs_info *fs_info)
1341
{
1342 1343 1344 1345
	struct extent_map_tree *em_tree;
	struct extent_map *em;
	struct rb_node *n;
	u64 ret = 0;
1346

1347 1348 1349 1350 1351 1352
	em_tree = &fs_info->mapping_tree.map_tree;
	read_lock(&em_tree->lock);
	n = rb_last(&em_tree->map);
	if (n) {
		em = rb_entry(n, struct extent_map, rb_node);
		ret = em->start + em->len;
1353
	}
1354 1355
	read_unlock(&em_tree->lock);

1356 1357 1358
	return ret;
}

1359 1360
static noinline int find_next_devid(struct btrfs_fs_info *fs_info,
				    u64 *devid_ret)
1361 1362 1363 1364
{
	int ret;
	struct btrfs_key key;
	struct btrfs_key found_key;
Y
Yan Zheng 已提交
1365 1366 1367 1368 1369
	struct btrfs_path *path;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
1370 1371 1372 1373 1374

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

1375
	ret = btrfs_search_slot(NULL, fs_info->chunk_root, &key, path, 0, 0);
1376 1377 1378
	if (ret < 0)
		goto error;

1379
	BUG_ON(ret == 0); /* Corruption */
1380

1381 1382
	ret = btrfs_previous_item(fs_info->chunk_root, path,
				  BTRFS_DEV_ITEMS_OBJECTID,
1383 1384
				  BTRFS_DEV_ITEM_KEY);
	if (ret) {
1385
		*devid_ret = 1;
1386 1387 1388
	} else {
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
1389
		*devid_ret = found_key.offset + 1;
1390 1391 1392
	}
	ret = 0;
error:
Y
Yan Zheng 已提交
1393
	btrfs_free_path(path);
1394 1395 1396 1397 1398 1399 1400
	return ret;
}

/*
 * the device information is stored in the chunk root
 * the btrfs_device struct should be fully filled in
 */
1401 1402 1403
static int btrfs_add_device(struct btrfs_trans_handle *trans,
			    struct btrfs_root *root,
			    struct btrfs_device *device)
1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419
{
	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
1420
	key.offset = device->devid;
1421 1422

	ret = btrfs_insert_empty_item(trans, root, path, &key,
1423
				      sizeof(*dev_item));
1424 1425 1426 1427 1428 1429 1430
	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
1431
	btrfs_set_device_generation(leaf, dev_item, 0);
1432 1433 1434 1435 1436 1437
	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);
1438 1439 1440
	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);
1441
	btrfs_set_device_start_offset(leaf, dev_item, 0);
1442

1443
	ptr = btrfs_device_uuid(dev_item);
1444
	write_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
1445
	ptr = btrfs_device_fsid(dev_item);
Y
Yan Zheng 已提交
1446
	write_extent_buffer(leaf, root->fs_info->fsid, ptr, BTRFS_UUID_SIZE);
1447 1448
	btrfs_mark_buffer_dirty(leaf);

Y
Yan Zheng 已提交
1449
	ret = 0;
1450 1451 1452 1453
out:
	btrfs_free_path(path);
	return ret;
}
1454

1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468
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;

1469
	trans = btrfs_start_transaction(root, 0);
1470 1471 1472 1473
	if (IS_ERR(trans)) {
		btrfs_free_path(path);
		return PTR_ERR(trans);
	}
1474 1475 1476
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.type = BTRFS_DEV_ITEM_KEY;
	key.offset = device->devid;
1477
	lock_chunks(root);
1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492

	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);
1493
	unlock_chunks(root);
1494 1495 1496 1497 1498 1499 1500
	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
1501
	struct btrfs_device *next_device;
1502
	struct block_device *bdev;
1503
	struct buffer_head *bh = NULL;
1504
	struct btrfs_super_block *disk_super;
1505
	struct btrfs_fs_devices *cur_devices;
1506 1507
	u64 all_avail;
	u64 devid;
Y
Yan Zheng 已提交
1508 1509
	u64 num_devices;
	u8 *dev_uuid;
1510
	unsigned seq;
1511
	int ret = 0;
1512
	bool clear_super = false;
1513 1514 1515

	mutex_lock(&uuid_mutex);

1516 1517 1518 1519 1520 1521 1522
	do {
		seq = read_seqbegin(&root->fs_info->profiles_lock);

		all_avail = root->fs_info->avail_data_alloc_bits |
			    root->fs_info->avail_system_alloc_bits |
			    root->fs_info->avail_metadata_alloc_bits;
	} while (read_seqretry(&root->fs_info->profiles_lock, seq));
1523

1524 1525 1526 1527 1528 1529 1530 1531 1532
	num_devices = root->fs_info->fs_devices->num_devices;
	btrfs_dev_replace_lock(&root->fs_info->dev_replace);
	if (btrfs_dev_replace_is_ongoing(&root->fs_info->dev_replace)) {
		WARN_ON(num_devices < 1);
		num_devices--;
	}
	btrfs_dev_replace_unlock(&root->fs_info->dev_replace);

	if ((all_avail & BTRFS_BLOCK_GROUP_RAID10) && num_devices <= 4) {
1533
		ret = BTRFS_ERROR_DEV_RAID10_MIN_NOT_MET;
1534 1535 1536
		goto out;
	}

1537
	if ((all_avail & BTRFS_BLOCK_GROUP_RAID1) && num_devices <= 2) {
1538
		ret = BTRFS_ERROR_DEV_RAID1_MIN_NOT_MET;
1539 1540 1541
		goto out;
	}

D
David Woodhouse 已提交
1542 1543
	if ((all_avail & BTRFS_BLOCK_GROUP_RAID5) &&
	    root->fs_info->fs_devices->rw_devices <= 2) {
1544
		ret = BTRFS_ERROR_DEV_RAID5_MIN_NOT_MET;
D
David Woodhouse 已提交
1545 1546 1547 1548
		goto out;
	}
	if ((all_avail & BTRFS_BLOCK_GROUP_RAID6) &&
	    root->fs_info->fs_devices->rw_devices <= 3) {
1549
		ret = BTRFS_ERROR_DEV_RAID6_MIN_NOT_MET;
D
David Woodhouse 已提交
1550 1551 1552
		goto out;
	}

1553 1554 1555
	if (strcmp(device_path, "missing") == 0) {
		struct list_head *devices;
		struct btrfs_device *tmp;
1556

1557 1558
		device = NULL;
		devices = &root->fs_info->fs_devices->devices;
1559 1560 1561 1562
		/*
		 * It is safe to read the devices since the volume_mutex
		 * is held.
		 */
1563
		list_for_each_entry(tmp, devices, dev_list) {
1564 1565 1566
			if (tmp->in_fs_metadata &&
			    !tmp->is_tgtdev_for_dev_replace &&
			    !tmp->bdev) {
1567 1568 1569 1570 1571 1572 1573 1574
				device = tmp;
				break;
			}
		}
		bdev = NULL;
		bh = NULL;
		disk_super = NULL;
		if (!device) {
1575
			ret = BTRFS_ERROR_DEV_MISSING_NOT_FOUND;
1576 1577 1578
			goto out;
		}
	} else {
1579
		ret = btrfs_get_bdev_and_sb(device_path,
1580
					    FMODE_WRITE | FMODE_EXCL,
1581 1582 1583
					    root->fs_info->bdev_holder, 0,
					    &bdev, &bh);
		if (ret)
1584 1585
			goto out;
		disk_super = (struct btrfs_super_block *)bh->b_data;
1586
		devid = btrfs_stack_device_id(&disk_super->dev_item);
Y
Yan Zheng 已提交
1587
		dev_uuid = disk_super->dev_item.uuid;
1588
		device = btrfs_find_device(root->fs_info, devid, dev_uuid,
Y
Yan Zheng 已提交
1589
					   disk_super->fsid);
1590 1591 1592 1593
		if (!device) {
			ret = -ENOENT;
			goto error_brelse;
		}
Y
Yan Zheng 已提交
1594
	}
1595

1596
	if (device->is_tgtdev_for_dev_replace) {
1597
		ret = BTRFS_ERROR_DEV_TGT_REPLACE;
1598 1599 1600
		goto error_brelse;
	}

Y
Yan Zheng 已提交
1601
	if (device->writeable && root->fs_info->fs_devices->rw_devices == 1) {
1602
		ret = BTRFS_ERROR_DEV_ONLY_WRITABLE;
Y
Yan Zheng 已提交
1603 1604 1605 1606
		goto error_brelse;
	}

	if (device->writeable) {
1607
		lock_chunks(root);
Y
Yan Zheng 已提交
1608
		list_del_init(&device->dev_alloc_list);
1609
		unlock_chunks(root);
Y
Yan Zheng 已提交
1610
		root->fs_info->fs_devices->rw_devices--;
1611
		clear_super = true;
1612
	}
1613

1614
	mutex_unlock(&uuid_mutex);
1615
	ret = btrfs_shrink_device(device, 0);
1616
	mutex_lock(&uuid_mutex);
1617
	if (ret)
1618
		goto error_undo;
1619

1620 1621 1622 1623 1624
	/*
	 * TODO: the superblock still includes this device in its num_devices
	 * counter although write_all_supers() is not locked out. This
	 * could give a filesystem state which requires a degraded mount.
	 */
1625 1626
	ret = btrfs_rm_dev_item(root->fs_info->chunk_root, device);
	if (ret)
1627
		goto error_undo;
1628

1629 1630 1631 1632 1633
	spin_lock(&root->fs_info->free_chunk_lock);
	root->fs_info->free_chunk_space = device->total_bytes -
		device->bytes_used;
	spin_unlock(&root->fs_info->free_chunk_lock);

Y
Yan Zheng 已提交
1634
	device->in_fs_metadata = 0;
1635
	btrfs_scrub_cancel_dev(root->fs_info, device);
1636 1637 1638 1639

	/*
	 * the device list mutex makes sure that we don't change
	 * the device list while someone else is writing out all
1640 1641 1642 1643 1644
	 * the device supers. Whoever is writing all supers, should
	 * lock the device list mutex before getting the number of
	 * devices in the super block (super_copy). Conversely,
	 * whoever updates the number of devices in the super block
	 * (super_copy) should hold the device list mutex.
1645
	 */
1646 1647

	cur_devices = device->fs_devices;
1648
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1649
	list_del_rcu(&device->dev_list);
1650

Y
Yan Zheng 已提交
1651
	device->fs_devices->num_devices--;
1652
	device->fs_devices->total_devices--;
Y
Yan Zheng 已提交
1653

1654 1655 1656
	if (device->missing)
		root->fs_info->fs_devices->missing_devices--;

Y
Yan Zheng 已提交
1657 1658 1659 1660 1661 1662 1663
	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;

1664
	if (device->bdev)
Y
Yan Zheng 已提交
1665
		device->fs_devices->open_devices--;
1666 1667

	call_rcu(&device->rcu, free_device);
Y
Yan Zheng 已提交
1668

1669 1670
	num_devices = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
	btrfs_set_super_num_devices(root->fs_info->super_copy, num_devices);
1671
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
1672

1673
	if (cur_devices->open_devices == 0) {
Y
Yan Zheng 已提交
1674 1675 1676
		struct btrfs_fs_devices *fs_devices;
		fs_devices = root->fs_info->fs_devices;
		while (fs_devices) {
1677
			if (fs_devices->seed == cur_devices)
Y
Yan Zheng 已提交
1678 1679
				break;
			fs_devices = fs_devices->seed;
Y
Yan Zheng 已提交
1680
		}
1681 1682
		fs_devices->seed = cur_devices->seed;
		cur_devices->seed = NULL;
1683
		lock_chunks(root);
1684
		__btrfs_close_devices(cur_devices);
1685
		unlock_chunks(root);
1686
		free_fs_devices(cur_devices);
Y
Yan Zheng 已提交
1687 1688
	}

1689 1690 1691
	root->fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(root->fs_info);

Y
Yan Zheng 已提交
1692 1693 1694 1695
	/*
	 * at this point, the device is zero sized.  We want to
	 * remove it from the devices list and zero out the old super
	 */
1696
	if (clear_super && disk_super) {
1697 1698 1699 1700 1701 1702 1703
		/* 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);
	}
1704 1705 1706

	ret = 0;

1707
	/* Notify udev that device has changed */
1708 1709
	if (bdev)
		btrfs_kobject_uevent(bdev, KOBJ_CHANGE);
1710

1711 1712
error_brelse:
	brelse(bh);
1713
	if (bdev)
1714
		blkdev_put(bdev, FMODE_READ | FMODE_EXCL);
1715 1716 1717
out:
	mutex_unlock(&uuid_mutex);
	return ret;
1718 1719
error_undo:
	if (device->writeable) {
1720
		lock_chunks(root);
1721 1722
		list_add(&device->dev_alloc_list,
			 &root->fs_info->fs_devices->alloc_list);
1723
		unlock_chunks(root);
1724 1725 1726
		root->fs_info->fs_devices->rw_devices++;
	}
	goto error_brelse;
1727 1728
}

1729 1730 1731 1732
void btrfs_rm_dev_replace_srcdev(struct btrfs_fs_info *fs_info,
				 struct btrfs_device *srcdev)
{
	WARN_ON(!mutex_is_locked(&fs_info->fs_devices->device_list_mutex));
1733

1734 1735 1736 1737 1738 1739 1740 1741 1742
	list_del_rcu(&srcdev->dev_list);
	list_del_rcu(&srcdev->dev_alloc_list);
	fs_info->fs_devices->num_devices--;
	if (srcdev->missing) {
		fs_info->fs_devices->missing_devices--;
		fs_info->fs_devices->rw_devices++;
	}
	if (srcdev->can_discard)
		fs_info->fs_devices->num_can_discard--;
1743
	if (srcdev->bdev) {
1744 1745
		fs_info->fs_devices->open_devices--;

1746 1747 1748 1749
		/* zero out the old super */
		btrfs_scratch_superblock(srcdev);
	}

1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780
	call_rcu(&srcdev->rcu, free_device);
}

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

	WARN_ON(!tgtdev);
	mutex_lock(&fs_info->fs_devices->device_list_mutex);
	if (tgtdev->bdev) {
		btrfs_scratch_superblock(tgtdev);
		fs_info->fs_devices->open_devices--;
	}
	fs_info->fs_devices->num_devices--;
	if (tgtdev->can_discard)
		fs_info->fs_devices->num_can_discard++;

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

	call_rcu(&tgtdev->rcu, free_device);

	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
}

1781 1782
static int btrfs_find_device_by_path(struct btrfs_root *root, char *device_path,
				     struct btrfs_device **device)
1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798
{
	int ret = 0;
	struct btrfs_super_block *disk_super;
	u64 devid;
	u8 *dev_uuid;
	struct block_device *bdev;
	struct buffer_head *bh;

	*device = NULL;
	ret = btrfs_get_bdev_and_sb(device_path, FMODE_READ,
				    root->fs_info->bdev_holder, 0, &bdev, &bh);
	if (ret)
		return ret;
	disk_super = (struct btrfs_super_block *)bh->b_data;
	devid = btrfs_stack_device_id(&disk_super->dev_item);
	dev_uuid = disk_super->dev_item.uuid;
1799
	*device = btrfs_find_device(root->fs_info, devid, dev_uuid,
1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829
				    disk_super->fsid);
	brelse(bh);
	if (!*device)
		ret = -ENOENT;
	blkdev_put(bdev, FMODE_READ);
	return ret;
}

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

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

		if (!*device) {
1830
			btrfs_err(root->fs_info, "no missing device found");
1831 1832 1833 1834 1835 1836 1837 1838 1839
			return -ENOENT;
		}

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

Y
Yan Zheng 已提交
1840 1841 1842
/*
 * does all the dirty work required for changing file system's UUID.
 */
1843
static int btrfs_prepare_sprout(struct btrfs_root *root)
Y
Yan Zheng 已提交
1844 1845 1846
{
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
	struct btrfs_fs_devices *old_devices;
Y
Yan Zheng 已提交
1847
	struct btrfs_fs_devices *seed_devices;
1848
	struct btrfs_super_block *disk_super = root->fs_info->super_copy;
Y
Yan Zheng 已提交
1849 1850 1851 1852
	struct btrfs_device *device;
	u64 super_flags;

	BUG_ON(!mutex_is_locked(&uuid_mutex));
Y
Yan Zheng 已提交
1853
	if (!fs_devices->seeding)
Y
Yan Zheng 已提交
1854 1855
		return -EINVAL;

1856 1857 1858
	seed_devices = __alloc_fs_devices();
	if (IS_ERR(seed_devices))
		return PTR_ERR(seed_devices);
Y
Yan Zheng 已提交
1859

Y
Yan Zheng 已提交
1860 1861 1862 1863
	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
1864
	}
Y
Yan Zheng 已提交
1865

Y
Yan Zheng 已提交
1866 1867
	list_add(&old_devices->list, &fs_uuids);

Y
Yan Zheng 已提交
1868 1869 1870 1871
	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);
1872
	mutex_init(&seed_devices->device_list_mutex);
1873 1874

	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1875 1876
	list_splice_init_rcu(&fs_devices->devices, &seed_devices->devices,
			      synchronize_rcu);
1877

Y
Yan Zheng 已提交
1878 1879 1880 1881 1882
	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
1883 1884 1885
	fs_devices->seeding = 0;
	fs_devices->num_devices = 0;
	fs_devices->open_devices = 0;
1886
	fs_devices->total_devices = 0;
Y
Yan Zheng 已提交
1887
	fs_devices->seed = seed_devices;
Y
Yan Zheng 已提交
1888 1889 1890 1891

	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);
1892 1893
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940
	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]);
1941
			btrfs_release_path(path);
Y
Yan Zheng 已提交
1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952
			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);
1953
		read_extent_buffer(leaf, dev_uuid, btrfs_device_uuid(dev_item),
Y
Yan Zheng 已提交
1954
				   BTRFS_UUID_SIZE);
1955
		read_extent_buffer(leaf, fs_uuid, btrfs_device_fsid(dev_item),
Y
Yan Zheng 已提交
1956
				   BTRFS_UUID_SIZE);
1957 1958
		device = btrfs_find_device(root->fs_info, devid, dev_uuid,
					   fs_uuid);
1959
		BUG_ON(!device); /* Logic error */
Y
Yan Zheng 已提交
1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975

		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;
}

1976 1977
int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
{
1978
	struct request_queue *q;
1979 1980 1981 1982
	struct btrfs_trans_handle *trans;
	struct btrfs_device *device;
	struct block_device *bdev;
	struct list_head *devices;
Y
Yan Zheng 已提交
1983
	struct super_block *sb = root->fs_info->sb;
1984
	struct rcu_string *name;
1985
	u64 total_bytes;
Y
Yan Zheng 已提交
1986
	int seeding_dev = 0;
1987 1988
	int ret = 0;

Y
Yan Zheng 已提交
1989
	if ((sb->s_flags & MS_RDONLY) && !root->fs_info->fs_devices->seeding)
1990
		return -EROFS;
1991

1992
	bdev = blkdev_get_by_path(device_path, FMODE_WRITE | FMODE_EXCL,
1993
				  root->fs_info->bdev_holder);
1994 1995
	if (IS_ERR(bdev))
		return PTR_ERR(bdev);
1996

Y
Yan Zheng 已提交
1997 1998 1999 2000 2001 2002
	if (root->fs_info->fs_devices->seeding) {
		seeding_dev = 1;
		down_write(&sb->s_umount);
		mutex_lock(&uuid_mutex);
	}

2003
	filemap_write_and_wait(bdev->bd_inode->i_mapping);
2004

2005
	devices = &root->fs_info->fs_devices->devices;
2006 2007

	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
2008
	list_for_each_entry(device, devices, dev_list) {
2009 2010
		if (device->bdev == bdev) {
			ret = -EEXIST;
2011 2012
			mutex_unlock(
				&root->fs_info->fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
2013
			goto error;
2014 2015
		}
	}
2016
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
2017

2018 2019
	device = btrfs_alloc_device(root->fs_info, NULL, NULL);
	if (IS_ERR(device)) {
2020
		/* we can safely leave the fs_devices entry around */
2021
		ret = PTR_ERR(device);
Y
Yan Zheng 已提交
2022
		goto error;
2023 2024
	}

2025 2026
	name = rcu_string_strdup(device_path, GFP_NOFS);
	if (!name) {
2027
		kfree(device);
Y
Yan Zheng 已提交
2028 2029
		ret = -ENOMEM;
		goto error;
2030
	}
2031
	rcu_assign_pointer(device->name, name);
Y
Yan Zheng 已提交
2032

2033
	trans = btrfs_start_transaction(root, 0);
2034
	if (IS_ERR(trans)) {
2035
		rcu_string_free(device->name);
2036 2037 2038 2039 2040
		kfree(device);
		ret = PTR_ERR(trans);
		goto error;
	}

Y
Yan Zheng 已提交
2041 2042
	lock_chunks(root);

2043 2044 2045
	q = bdev_get_queue(bdev);
	if (blk_queue_discard(q))
		device->can_discard = 1;
Y
Yan Zheng 已提交
2046 2047
	device->writeable = 1;
	device->generation = trans->transid;
2048 2049 2050 2051
	device->io_width = root->sectorsize;
	device->io_align = root->sectorsize;
	device->sector_size = root->sectorsize;
	device->total_bytes = i_size_read(bdev->bd_inode);
2052
	device->disk_total_bytes = device->total_bytes;
2053 2054
	device->dev_root = root->fs_info->dev_root;
	device->bdev = bdev;
2055
	device->in_fs_metadata = 1;
2056
	device->is_tgtdev_for_dev_replace = 0;
2057
	device->mode = FMODE_EXCL;
2058
	device->dev_stats_valid = 1;
Y
Yan Zheng 已提交
2059
	set_blocksize(device->bdev, 4096);
2060

Y
Yan Zheng 已提交
2061 2062
	if (seeding_dev) {
		sb->s_flags &= ~MS_RDONLY;
2063
		ret = btrfs_prepare_sprout(root);
2064
		BUG_ON(ret); /* -ENOMEM */
Y
Yan Zheng 已提交
2065
	}
2066

Y
Yan Zheng 已提交
2067
	device->fs_devices = root->fs_info->fs_devices;
2068 2069

	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
2070
	list_add_rcu(&device->dev_list, &root->fs_info->fs_devices->devices);
Y
Yan Zheng 已提交
2071 2072 2073 2074 2075
	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++;
2076
	root->fs_info->fs_devices->total_devices++;
2077 2078
	if (device->can_discard)
		root->fs_info->fs_devices->num_can_discard++;
Y
Yan Zheng 已提交
2079
	root->fs_info->fs_devices->total_rw_bytes += device->total_bytes;
2080

2081 2082 2083 2084
	spin_lock(&root->fs_info->free_chunk_lock);
	root->fs_info->free_chunk_space += device->total_bytes;
	spin_unlock(&root->fs_info->free_chunk_lock);

2085 2086 2087
	if (!blk_queue_nonrot(bdev_get_queue(bdev)))
		root->fs_info->fs_devices->rotating = 1;

2088 2089
	total_bytes = btrfs_super_total_bytes(root->fs_info->super_copy);
	btrfs_set_super_total_bytes(root->fs_info->super_copy,
2090 2091
				    total_bytes + device->total_bytes);

2092 2093
	total_bytes = btrfs_super_num_devices(root->fs_info->super_copy);
	btrfs_set_super_num_devices(root->fs_info->super_copy,
2094
				    total_bytes + 1);
2095
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
2096

Y
Yan Zheng 已提交
2097 2098
	if (seeding_dev) {
		ret = init_first_rw_device(trans, root, device);
2099 2100
		if (ret) {
			btrfs_abort_transaction(trans, root, ret);
2101
			goto error_trans;
2102
		}
Y
Yan Zheng 已提交
2103
		ret = btrfs_finish_sprout(trans, root);
2104 2105
		if (ret) {
			btrfs_abort_transaction(trans, root, ret);
2106
			goto error_trans;
2107
		}
Y
Yan Zheng 已提交
2108 2109
	} else {
		ret = btrfs_add_device(trans, root, device);
2110 2111
		if (ret) {
			btrfs_abort_transaction(trans, root, ret);
2112
			goto error_trans;
2113
		}
Y
Yan Zheng 已提交
2114 2115
	}

2116 2117 2118 2119 2120 2121
	/*
	 * we've got more storage, clear any full flags on the space
	 * infos
	 */
	btrfs_clear_space_info_full(root->fs_info);

2122
	unlock_chunks(root);
2123 2124
	root->fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(root->fs_info);
2125
	ret = btrfs_commit_transaction(trans, root);
2126

Y
Yan Zheng 已提交
2127 2128 2129
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
2130

2131 2132 2133
		if (ret) /* transaction commit */
			return ret;

Y
Yan Zheng 已提交
2134
		ret = btrfs_relocate_sys_chunks(root);
2135 2136 2137 2138 2139
		if (ret < 0)
			btrfs_error(root->fs_info, ret,
				    "Failed to relocate sys chunks after "
				    "device initialization. This can be fixed "
				    "using the \"btrfs balance\" command.");
2140 2141 2142 2143 2144 2145 2146
		trans = btrfs_attach_transaction(root);
		if (IS_ERR(trans)) {
			if (PTR_ERR(trans) == -ENOENT)
				return 0;
			return PTR_ERR(trans);
		}
		ret = btrfs_commit_transaction(trans, root);
Y
Yan Zheng 已提交
2147
	}
2148

Y
Yan Zheng 已提交
2149
	return ret;
2150 2151 2152 2153

error_trans:
	unlock_chunks(root);
	btrfs_end_transaction(trans, root);
2154
	rcu_string_free(device->name);
2155
	kfree(device);
Y
Yan Zheng 已提交
2156
error:
2157
	blkdev_put(bdev, FMODE_EXCL);
Y
Yan Zheng 已提交
2158 2159 2160 2161
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
	}
2162
	return ret;
2163 2164
}

2165 2166 2167 2168 2169 2170 2171 2172 2173
int btrfs_init_dev_replace_tgtdev(struct btrfs_root *root, char *device_path,
				  struct btrfs_device **device_out)
{
	struct request_queue *q;
	struct btrfs_device *device;
	struct block_device *bdev;
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct list_head *devices;
	struct rcu_string *name;
2174
	u64 devid = BTRFS_DEV_REPLACE_DEVID;
2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195
	int ret = 0;

	*device_out = NULL;
	if (fs_info->fs_devices->seeding)
		return -EINVAL;

	bdev = blkdev_get_by_path(device_path, FMODE_WRITE | FMODE_EXCL,
				  fs_info->bdev_holder);
	if (IS_ERR(bdev))
		return PTR_ERR(bdev);

	filemap_write_and_wait(bdev->bd_inode->i_mapping);

	devices = &fs_info->fs_devices->devices;
	list_for_each_entry(device, devices, dev_list) {
		if (device->bdev == bdev) {
			ret = -EEXIST;
			goto error;
		}
	}

2196 2197 2198
	device = btrfs_alloc_device(NULL, &devid, NULL);
	if (IS_ERR(device)) {
		ret = PTR_ERR(device);
2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225
		goto error;
	}

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

	q = bdev_get_queue(bdev);
	if (blk_queue_discard(q))
		device->can_discard = 1;
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
	device->writeable = 1;
	device->generation = 0;
	device->io_width = root->sectorsize;
	device->io_align = root->sectorsize;
	device->sector_size = root->sectorsize;
	device->total_bytes = i_size_read(bdev->bd_inode);
	device->disk_total_bytes = device->total_bytes;
	device->dev_root = fs_info->dev_root;
	device->bdev = bdev;
	device->in_fs_metadata = 1;
	device->is_tgtdev_for_dev_replace = 1;
	device->mode = FMODE_EXCL;
2226
	device->dev_stats_valid = 1;
2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254
	set_blocksize(device->bdev, 4096);
	device->fs_devices = fs_info->fs_devices;
	list_add(&device->dev_list, &fs_info->fs_devices->devices);
	fs_info->fs_devices->num_devices++;
	fs_info->fs_devices->open_devices++;
	if (device->can_discard)
		fs_info->fs_devices->num_can_discard++;
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);

	*device_out = device;
	return ret;

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

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

2255 2256
static noinline int btrfs_update_device(struct btrfs_trans_handle *trans,
					struct btrfs_device *device)
2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291
{
	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);
2292
	btrfs_set_device_total_bytes(leaf, dev_item, device->disk_total_bytes);
2293 2294 2295 2296 2297 2298 2299 2300
	btrfs_set_device_bytes_used(leaf, dev_item, device->bytes_used);
	btrfs_mark_buffer_dirty(leaf);

out:
	btrfs_free_path(path);
	return ret;
}

2301
static int __btrfs_grow_device(struct btrfs_trans_handle *trans,
2302 2303 2304
		      struct btrfs_device *device, u64 new_size)
{
	struct btrfs_super_block *super_copy =
2305
		device->dev_root->fs_info->super_copy;
2306 2307 2308
	u64 old_total = btrfs_super_total_bytes(super_copy);
	u64 diff = new_size - device->total_bytes;

Y
Yan Zheng 已提交
2309 2310
	if (!device->writeable)
		return -EACCES;
2311 2312
	if (new_size <= device->total_bytes ||
	    device->is_tgtdev_for_dev_replace)
Y
Yan Zheng 已提交
2313 2314
		return -EINVAL;

2315
	btrfs_set_super_total_bytes(super_copy, old_total + diff);
Y
Yan Zheng 已提交
2316 2317 2318
	device->fs_devices->total_rw_bytes += diff;

	device->total_bytes = new_size;
2319
	device->disk_total_bytes = new_size;
2320 2321
	btrfs_clear_space_info_full(device->dev_root->fs_info);

2322 2323 2324
	return btrfs_update_device(trans, device);
}

2325 2326 2327 2328 2329 2330 2331 2332 2333 2334
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;
}

2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353
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);
2354 2355 2356 2357 2358 2359 2360 2361
	if (ret < 0)
		goto out;
	else if (ret > 0) { /* Logic error or corruption */
		btrfs_error(root->fs_info, -ENOENT,
			    "Failed lookup while freeing chunk.");
		ret = -ENOENT;
		goto out;
	}
2362 2363

	ret = btrfs_del_item(trans, root, path);
2364 2365 2366 2367
	if (ret < 0)
		btrfs_error(root->fs_info, ret,
			    "Failed to delete chunk item.");
out:
2368
	btrfs_free_path(path);
2369
	return ret;
2370 2371
}

2372
static int btrfs_del_sys_chunk(struct btrfs_root *root, u64 chunk_objectid, u64
2373 2374
			chunk_offset)
{
2375
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417
	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;
}

2418
static int btrfs_relocate_chunk(struct btrfs_root *root,
2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433
			 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;

2434 2435 2436 2437
	ret = btrfs_can_relocate(extent_root, chunk_offset);
	if (ret)
		return -ENOSPC;

2438
	/* step one, relocate all the extents inside this chunk */
2439
	ret = btrfs_relocate_block_group(extent_root, chunk_offset);
2440 2441
	if (ret)
		return ret;
2442

2443
	trans = btrfs_start_transaction(root, 0);
2444 2445 2446 2447 2448
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		btrfs_std_error(root->fs_info, ret);
		return ret;
	}
2449

2450 2451
	lock_chunks(root);

2452 2453 2454 2455
	/*
	 * step two, delete the device extents and the
	 * chunk tree entries
	 */
2456
	read_lock(&em_tree->lock);
2457
	em = lookup_extent_mapping(em_tree, chunk_offset, 1);
2458
	read_unlock(&em_tree->lock);
2459

2460
	BUG_ON(!em || em->start > chunk_offset ||
2461
	       em->start + em->len < chunk_offset);
2462 2463 2464 2465 2466 2467
	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);
2468

2469 2470 2471 2472
		if (map->stripes[i].dev) {
			ret = btrfs_update_device(trans, map->stripes[i].dev);
			BUG_ON(ret);
		}
2473 2474 2475 2476 2477 2478
	}
	ret = btrfs_free_chunk(trans, root, chunk_tree, chunk_objectid,
			       chunk_offset);

	BUG_ON(ret);

2479 2480
	trace_btrfs_chunk_free(root, map, chunk_offset, em->len);

2481 2482 2483 2484 2485
	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
2486 2487 2488
	ret = btrfs_remove_block_group(trans, extent_root, chunk_offset);
	BUG_ON(ret);

2489
	write_lock(&em_tree->lock);
Y
Yan Zheng 已提交
2490
	remove_extent_mapping(em_tree, em);
2491
	write_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515

	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;
2516 2517
	bool retried = false;
	int failed = 0;
Y
Yan Zheng 已提交
2518 2519 2520 2521 2522 2523
	int ret;

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

2524
again:
Y
Yan Zheng 已提交
2525 2526 2527 2528 2529 2530 2531 2532
	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;
2533
		BUG_ON(ret == 0); /* Corruption */
Y
Yan Zheng 已提交
2534 2535 2536 2537 2538 2539 2540

		ret = btrfs_previous_item(chunk_root, path, key.objectid,
					  key.type);
		if (ret < 0)
			goto error;
		if (ret > 0)
			break;
2541

Y
Yan Zheng 已提交
2542 2543
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
2544

Y
Yan Zheng 已提交
2545 2546 2547
		chunk = btrfs_item_ptr(leaf, path->slots[0],
				       struct btrfs_chunk);
		chunk_type = btrfs_chunk_type(leaf, chunk);
2548
		btrfs_release_path(path);
2549

Y
Yan Zheng 已提交
2550 2551 2552 2553
		if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) {
			ret = btrfs_relocate_chunk(chunk_root, chunk_tree,
						   found_key.objectid,
						   found_key.offset);
2554 2555 2556 2557
			if (ret == -ENOSPC)
				failed++;
			else if (ret)
				BUG();
Y
Yan Zheng 已提交
2558
		}
2559

Y
Yan Zheng 已提交
2560 2561 2562 2563 2564
		if (found_key.offset == 0)
			break;
		key.offset = found_key.offset - 1;
	}
	ret = 0;
2565 2566 2567 2568
	if (failed && !retried) {
		failed = 0;
		retried = true;
		goto again;
2569
	} else if (WARN_ON(failed && retried)) {
2570 2571
		ret = -ENOSPC;
	}
Y
Yan Zheng 已提交
2572 2573 2574
error:
	btrfs_free_path(path);
	return ret;
2575 2576
}

2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667
static int insert_balance_item(struct btrfs_root *root,
			       struct btrfs_balance_control *bctl)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_balance_item *item;
	struct btrfs_disk_balance_args disk_bargs;
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_key key;
	int ret, err;

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707
/*
 * This is a heuristic used to reduce the number of chunks balanced on
 * resume after balance was interrupted.
 */
static void update_balance_args(struct btrfs_balance_control *bctl)
{
	/*
	 * Turn on soft mode for chunk types that were being converted.
	 */
	if (bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT)
		bctl->data.flags |= BTRFS_BALANCE_ARGS_SOFT;
	if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT)
		bctl->sys.flags |= BTRFS_BALANCE_ARGS_SOFT;
	if (bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT)
		bctl->meta.flags |= BTRFS_BALANCE_ARGS_SOFT;

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

2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736
/*
 * Should be called with both balance and volume mutexes held to
 * serialize other volume operations (add_dev/rm_dev/resize) with
 * restriper.  Same goes for unset_balance_control.
 */
static void set_balance_control(struct btrfs_balance_control *bctl)
{
	struct btrfs_fs_info *fs_info = bctl->fs_info;

	BUG_ON(fs_info->balance_ctl);

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

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

	BUG_ON(!fs_info->balance_ctl);

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

	kfree(bctl);
}

I
Ilya Dryomov 已提交
2737 2738 2739 2740
/*
 * Balance filters.  Return 1 if chunk should be filtered out
 * (should not be balanced).
 */
2741
static int chunk_profiles_filter(u64 chunk_type,
I
Ilya Dryomov 已提交
2742 2743
				 struct btrfs_balance_args *bargs)
{
2744 2745
	chunk_type = chunk_to_extended(chunk_type) &
				BTRFS_EXTENDED_PROFILE_MASK;
I
Ilya Dryomov 已提交
2746

2747
	if (bargs->profiles & chunk_type)
I
Ilya Dryomov 已提交
2748 2749 2750 2751 2752
		return 0;

	return 1;
}

I
Ilya Dryomov 已提交
2753 2754 2755 2756 2757 2758 2759 2760 2761 2762
static int chunk_usage_filter(struct btrfs_fs_info *fs_info, u64 chunk_offset,
			      struct btrfs_balance_args *bargs)
{
	struct btrfs_block_group_cache *cache;
	u64 chunk_used, user_thresh;
	int ret = 1;

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

2763
	if (bargs->usage == 0)
2764
		user_thresh = 1;
2765 2766 2767 2768 2769 2770
	else if (bargs->usage > 100)
		user_thresh = cache->key.offset;
	else
		user_thresh = div_factor_fine(cache->key.offset,
					      bargs->usage);

I
Ilya Dryomov 已提交
2771 2772 2773 2774 2775 2776 2777
	if (chunk_used < user_thresh)
		ret = 0;

	btrfs_put_block_group(cache);
	return ret;
}

I
Ilya Dryomov 已提交
2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794
static int chunk_devid_filter(struct extent_buffer *leaf,
			      struct btrfs_chunk *chunk,
			      struct btrfs_balance_args *bargs)
{
	struct btrfs_stripe *stripe;
	int num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
	int i;

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

	return 1;
}

I
Ilya Dryomov 已提交
2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811
/* [pstart, pend) */
static int chunk_drange_filter(struct extent_buffer *leaf,
			       struct btrfs_chunk *chunk,
			       u64 chunk_offset,
			       struct btrfs_balance_args *bargs)
{
	struct btrfs_stripe *stripe;
	int num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
	u64 stripe_offset;
	u64 stripe_length;
	int factor;
	int i;

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

	if (btrfs_chunk_type(leaf, chunk) & (BTRFS_BLOCK_GROUP_DUP |
D
David Woodhouse 已提交
2812 2813 2814 2815 2816 2817 2818 2819 2820
	     BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10)) {
		factor = num_stripes / 2;
	} else if (btrfs_chunk_type(leaf, chunk) & BTRFS_BLOCK_GROUP_RAID5) {
		factor = num_stripes - 1;
	} else if (btrfs_chunk_type(leaf, chunk) & BTRFS_BLOCK_GROUP_RAID6) {
		factor = num_stripes - 2;
	} else {
		factor = num_stripes;
	}
I
Ilya Dryomov 已提交
2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838

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

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

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

	return 1;
}

2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852
/* [vstart, vend) */
static int chunk_vrange_filter(struct extent_buffer *leaf,
			       struct btrfs_chunk *chunk,
			       u64 chunk_offset,
			       struct btrfs_balance_args *bargs)
{
	if (chunk_offset < bargs->vend &&
	    chunk_offset + btrfs_chunk_length(leaf, chunk) > bargs->vstart)
		/* at least part of the chunk is inside this vrange */
		return 0;

	return 1;
}

2853
static int chunk_soft_convert_filter(u64 chunk_type,
2854 2855 2856 2857 2858
				     struct btrfs_balance_args *bargs)
{
	if (!(bargs->flags & BTRFS_BALANCE_ARGS_CONVERT))
		return 0;

2859 2860
	chunk_type = chunk_to_extended(chunk_type) &
				BTRFS_EXTENDED_PROFILE_MASK;
2861

2862
	if (bargs->target == chunk_type)
2863 2864 2865 2866 2867
		return 1;

	return 0;
}

2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888
static int should_balance_chunk(struct btrfs_root *root,
				struct extent_buffer *leaf,
				struct btrfs_chunk *chunk, u64 chunk_offset)
{
	struct btrfs_balance_control *bctl = root->fs_info->balance_ctl;
	struct btrfs_balance_args *bargs = NULL;
	u64 chunk_type = btrfs_chunk_type(leaf, chunk);

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

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

I
Ilya Dryomov 已提交
2889 2890 2891 2892
	/* profiles filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_PROFILES) &&
	    chunk_profiles_filter(chunk_type, bargs)) {
		return 0;
I
Ilya Dryomov 已提交
2893 2894 2895 2896 2897 2898
	}

	/* usage filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_USAGE) &&
	    chunk_usage_filter(bctl->fs_info, chunk_offset, bargs)) {
		return 0;
I
Ilya Dryomov 已提交
2899 2900 2901 2902 2903 2904
	}

	/* devid filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_DEVID) &&
	    chunk_devid_filter(leaf, chunk, bargs)) {
		return 0;
I
Ilya Dryomov 已提交
2905 2906 2907 2908 2909 2910
	}

	/* drange filter, makes sense only with devid filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_DRANGE) &&
	    chunk_drange_filter(leaf, chunk, chunk_offset, bargs)) {
		return 0;
2911 2912 2913 2914 2915 2916
	}

	/* vrange filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_VRANGE) &&
	    chunk_vrange_filter(leaf, chunk, chunk_offset, bargs)) {
		return 0;
I
Ilya Dryomov 已提交
2917 2918
	}

2919 2920 2921 2922 2923 2924
	/* soft profile changing mode */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_SOFT) &&
	    chunk_soft_convert_filter(chunk_type, bargs)) {
		return 0;
	}

2925 2926 2927
	return 1;
}

2928
static int __btrfs_balance(struct btrfs_fs_info *fs_info)
2929
{
2930
	struct btrfs_balance_control *bctl = fs_info->balance_ctl;
2931 2932 2933
	struct btrfs_root *chunk_root = fs_info->chunk_root;
	struct btrfs_root *dev_root = fs_info->dev_root;
	struct list_head *devices;
2934 2935 2936
	struct btrfs_device *device;
	u64 old_size;
	u64 size_to_free;
2937
	struct btrfs_chunk *chunk;
2938 2939 2940
	struct btrfs_path *path;
	struct btrfs_key key;
	struct btrfs_key found_key;
2941
	struct btrfs_trans_handle *trans;
2942 2943
	struct extent_buffer *leaf;
	int slot;
2944 2945
	int ret;
	int enospc_errors = 0;
2946
	bool counting = true;
2947 2948

	/* step one make some room on all the devices */
2949
	devices = &fs_info->fs_devices->devices;
2950
	list_for_each_entry(device, devices, dev_list) {
2951 2952 2953
		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
2954
		if (!device->writeable ||
2955 2956
		    device->total_bytes - device->bytes_used > size_to_free ||
		    device->is_tgtdev_for_dev_replace)
2957 2958 2959
			continue;

		ret = btrfs_shrink_device(device, old_size - size_to_free);
2960 2961
		if (ret == -ENOSPC)
			break;
2962 2963
		BUG_ON(ret);

2964
		trans = btrfs_start_transaction(dev_root, 0);
2965
		BUG_ON(IS_ERR(trans));
2966 2967 2968 2969 2970 2971 2972 2973 2974

		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();
2975 2976 2977 2978
	if (!path) {
		ret = -ENOMEM;
		goto error;
	}
2979 2980 2981 2982 2983 2984

	/* zero out stat counters */
	spin_lock(&fs_info->balance_lock);
	memset(&bctl->stat, 0, sizeof(bctl->stat));
	spin_unlock(&fs_info->balance_lock);
again:
2985 2986 2987 2988
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.offset = (u64)-1;
	key.type = BTRFS_CHUNK_ITEM_KEY;

2989
	while (1) {
2990
		if ((!counting && atomic_read(&fs_info->balance_pause_req)) ||
2991
		    atomic_read(&fs_info->balance_cancel_req)) {
2992 2993 2994 2995
			ret = -ECANCELED;
			goto error;
		}

2996 2997 2998 2999 3000 3001 3002 3003 3004
		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)
3005
			BUG(); /* FIXME break ? */
3006 3007 3008

		ret = btrfs_previous_item(chunk_root, path, 0,
					  BTRFS_CHUNK_ITEM_KEY);
3009 3010
		if (ret) {
			ret = 0;
3011
			break;
3012
		}
3013

3014 3015 3016
		leaf = path->nodes[0];
		slot = path->slots[0];
		btrfs_item_key_to_cpu(leaf, &found_key, slot);
3017

3018 3019
		if (found_key.objectid != key.objectid)
			break;
3020

3021 3022
		chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);

3023 3024 3025 3026 3027 3028
		if (!counting) {
			spin_lock(&fs_info->balance_lock);
			bctl->stat.considered++;
			spin_unlock(&fs_info->balance_lock);
		}

3029 3030
		ret = should_balance_chunk(chunk_root, leaf, chunk,
					   found_key.offset);
3031
		btrfs_release_path(path);
3032 3033 3034
		if (!ret)
			goto loop;

3035 3036 3037 3038 3039 3040 3041
		if (counting) {
			spin_lock(&fs_info->balance_lock);
			bctl->stat.expected++;
			spin_unlock(&fs_info->balance_lock);
			goto loop;
		}

3042 3043 3044 3045
		ret = btrfs_relocate_chunk(chunk_root,
					   chunk_root->root_key.objectid,
					   found_key.objectid,
					   found_key.offset);
3046 3047
		if (ret && ret != -ENOSPC)
			goto error;
3048
		if (ret == -ENOSPC) {
3049
			enospc_errors++;
3050 3051 3052 3053 3054
		} else {
			spin_lock(&fs_info->balance_lock);
			bctl->stat.completed++;
			spin_unlock(&fs_info->balance_lock);
		}
3055
loop:
3056 3057
		if (found_key.offset == 0)
			break;
3058
		key.offset = found_key.offset - 1;
3059
	}
3060

3061 3062 3063 3064 3065
	if (counting) {
		btrfs_release_path(path);
		counting = false;
		goto again;
	}
3066 3067
error:
	btrfs_free_path(path);
3068
	if (enospc_errors) {
3069
		btrfs_info(fs_info, "%d enospc errors during balance",
3070 3071 3072 3073 3074
		       enospc_errors);
		if (!ret)
			ret = -ENOSPC;
	}

3075 3076 3077
	return ret;
}

3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101
/**
 * alloc_profile_is_valid - see if a given profile is valid and reduced
 * @flags: profile to validate
 * @extended: if true @flags is treated as an extended profile
 */
static int alloc_profile_is_valid(u64 flags, int extended)
{
	u64 mask = (extended ? BTRFS_EXTENDED_PROFILE_MASK :
			       BTRFS_BLOCK_GROUP_PROFILE_MASK);

	flags &= ~BTRFS_BLOCK_GROUP_TYPE_MASK;

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

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

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

3102 3103
static inline int balance_need_close(struct btrfs_fs_info *fs_info)
{
3104 3105 3106 3107
	/* cancel requested || normal exit path */
	return atomic_read(&fs_info->balance_cancel_req) ||
		(atomic_read(&fs_info->balance_pause_req) == 0 &&
		 atomic_read(&fs_info->balance_cancel_req) == 0);
3108 3109
}

3110 3111
static void __cancel_balance(struct btrfs_fs_info *fs_info)
{
3112 3113
	int ret;

3114
	unset_balance_control(fs_info);
3115
	ret = del_balance_item(fs_info->tree_root);
3116 3117
	if (ret)
		btrfs_std_error(fs_info, ret);
3118 3119

	atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
3120 3121 3122 3123 3124 3125 3126 3127 3128
}

/*
 * Should be called with both balance and volume mutexes held
 */
int btrfs_balance(struct btrfs_balance_control *bctl,
		  struct btrfs_ioctl_balance_args *bargs)
{
	struct btrfs_fs_info *fs_info = bctl->fs_info;
3129
	u64 allowed;
3130
	int mixed = 0;
3131
	int ret;
3132
	u64 num_devices;
3133
	unsigned seq;
3134

3135
	if (btrfs_fs_closing(fs_info) ||
3136 3137
	    atomic_read(&fs_info->balance_pause_req) ||
	    atomic_read(&fs_info->balance_cancel_req)) {
3138 3139 3140 3141
		ret = -EINVAL;
		goto out;
	}

3142 3143 3144 3145
	allowed = btrfs_super_incompat_flags(fs_info->super_copy);
	if (allowed & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS)
		mixed = 1;

3146 3147 3148 3149
	/*
	 * In case of mixed groups both data and meta should be picked,
	 * and identical options should be given for both of them.
	 */
3150 3151
	allowed = BTRFS_BALANCE_DATA | BTRFS_BALANCE_METADATA;
	if (mixed && (bctl->flags & allowed)) {
3152 3153 3154
		if (!(bctl->flags & BTRFS_BALANCE_DATA) ||
		    !(bctl->flags & BTRFS_BALANCE_METADATA) ||
		    memcmp(&bctl->data, &bctl->meta, sizeof(bctl->data))) {
3155 3156
			btrfs_err(fs_info, "with mixed groups data and "
				   "metadata balance options must be the same");
3157 3158 3159 3160 3161
			ret = -EINVAL;
			goto out;
		}
	}

3162 3163 3164 3165 3166 3167 3168
	num_devices = fs_info->fs_devices->num_devices;
	btrfs_dev_replace_lock(&fs_info->dev_replace);
	if (btrfs_dev_replace_is_ongoing(&fs_info->dev_replace)) {
		BUG_ON(num_devices < 1);
		num_devices--;
	}
	btrfs_dev_replace_unlock(&fs_info->dev_replace);
3169
	allowed = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
3170
	if (num_devices == 1)
3171
		allowed |= BTRFS_BLOCK_GROUP_DUP;
3172
	else if (num_devices > 1)
3173
		allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1);
3174 3175 3176 3177 3178
	if (num_devices > 2)
		allowed |= BTRFS_BLOCK_GROUP_RAID5;
	if (num_devices > 3)
		allowed |= (BTRFS_BLOCK_GROUP_RAID10 |
			    BTRFS_BLOCK_GROUP_RAID6);
3179 3180 3181
	if ((bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
	    (!alloc_profile_is_valid(bctl->data.target, 1) ||
	     (bctl->data.target & ~allowed))) {
3182 3183
		btrfs_err(fs_info, "unable to start balance with target "
			   "data profile %llu",
3184
		       bctl->data.target);
3185 3186 3187
		ret = -EINVAL;
		goto out;
	}
3188 3189 3190
	if ((bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
	    (!alloc_profile_is_valid(bctl->meta.target, 1) ||
	     (bctl->meta.target & ~allowed))) {
3191 3192
		btrfs_err(fs_info,
			   "unable to start balance with target metadata profile %llu",
3193
		       bctl->meta.target);
3194 3195 3196
		ret = -EINVAL;
		goto out;
	}
3197 3198 3199
	if ((bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
	    (!alloc_profile_is_valid(bctl->sys.target, 1) ||
	     (bctl->sys.target & ~allowed))) {
3200 3201
		btrfs_err(fs_info,
			   "unable to start balance with target system profile %llu",
3202
		       bctl->sys.target);
3203 3204 3205 3206
		ret = -EINVAL;
		goto out;
	}

3207 3208
	/* allow dup'ed data chunks only in mixed mode */
	if (!mixed && (bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
3209
	    (bctl->data.target & BTRFS_BLOCK_GROUP_DUP)) {
3210
		btrfs_err(fs_info, "dup for data is not allowed");
3211 3212 3213 3214 3215 3216
		ret = -EINVAL;
		goto out;
	}

	/* allow to reduce meta or sys integrity only if force set */
	allowed = BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1 |
D
David Woodhouse 已提交
3217 3218 3219
			BTRFS_BLOCK_GROUP_RAID10 |
			BTRFS_BLOCK_GROUP_RAID5 |
			BTRFS_BLOCK_GROUP_RAID6;
3220 3221 3222 3223 3224 3225 3226 3227 3228 3229
	do {
		seq = read_seqbegin(&fs_info->profiles_lock);

		if (((bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
		     (fs_info->avail_system_alloc_bits & allowed) &&
		     !(bctl->sys.target & allowed)) ||
		    ((bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
		     (fs_info->avail_metadata_alloc_bits & allowed) &&
		     !(bctl->meta.target & allowed))) {
			if (bctl->flags & BTRFS_BALANCE_FORCE) {
3230
				btrfs_info(fs_info, "force reducing metadata integrity");
3231
			} else {
3232 3233
				btrfs_err(fs_info, "balance will reduce metadata "
					   "integrity, use force if you want this");
3234 3235 3236
				ret = -EINVAL;
				goto out;
			}
3237
		}
3238
	} while (read_seqretry(&fs_info->profiles_lock, seq));
3239

3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259
	if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
		int num_tolerated_disk_barrier_failures;
		u64 target = bctl->sys.target;

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

		fs_info->num_tolerated_disk_barrier_failures =
			num_tolerated_disk_barrier_failures;
	}

3260
	ret = insert_balance_item(fs_info->tree_root, bctl);
3261
	if (ret && ret != -EEXIST)
3262 3263
		goto out;

3264 3265 3266 3267 3268 3269 3270 3271 3272
	if (!(bctl->flags & BTRFS_BALANCE_RESUME)) {
		BUG_ON(ret == -EEXIST);
		set_balance_control(bctl);
	} else {
		BUG_ON(ret != -EEXIST);
		spin_lock(&fs_info->balance_lock);
		update_balance_args(bctl);
		spin_unlock(&fs_info->balance_lock);
	}
3273

3274
	atomic_inc(&fs_info->balance_running);
3275 3276 3277 3278 3279
	mutex_unlock(&fs_info->balance_mutex);

	ret = __btrfs_balance(fs_info);

	mutex_lock(&fs_info->balance_mutex);
3280
	atomic_dec(&fs_info->balance_running);
3281

3282 3283 3284 3285 3286
	if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
		fs_info->num_tolerated_disk_barrier_failures =
			btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
	}

3287 3288
	if (bargs) {
		memset(bargs, 0, sizeof(*bargs));
3289
		update_ioctl_balance_args(fs_info, 0, bargs);
3290 3291
	}

3292 3293 3294 3295 3296
	if ((ret && ret != -ECANCELED && ret != -ENOSPC) ||
	    balance_need_close(fs_info)) {
		__cancel_balance(fs_info);
	}

3297
	wake_up(&fs_info->balance_wait_q);
3298 3299 3300

	return ret;
out:
3301 3302
	if (bctl->flags & BTRFS_BALANCE_RESUME)
		__cancel_balance(fs_info);
3303
	else {
3304
		kfree(bctl);
3305 3306
		atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
	}
3307 3308 3309 3310 3311
	return ret;
}

static int balance_kthread(void *data)
{
3312
	struct btrfs_fs_info *fs_info = data;
3313
	int ret = 0;
3314 3315 3316 3317

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

3318
	if (fs_info->balance_ctl) {
3319
		btrfs_info(fs_info, "continuing balance");
3320
		ret = btrfs_balance(fs_info->balance_ctl, NULL);
3321
	}
3322 3323 3324

	mutex_unlock(&fs_info->balance_mutex);
	mutex_unlock(&fs_info->volume_mutex);
3325

3326 3327 3328
	return ret;
}

3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340
int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info)
{
	struct task_struct *tsk;

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

	if (btrfs_test_opt(fs_info->tree_root, SKIP_BALANCE)) {
3341
		btrfs_info(fs_info, "force skipping balance");
3342 3343 3344 3345
		return 0;
	}

	tsk = kthread_run(balance_kthread, fs_info, "btrfs-balance");
3346
	return PTR_ERR_OR_ZERO(tsk);
3347 3348
}

3349
int btrfs_recover_balance(struct btrfs_fs_info *fs_info)
3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366
{
	struct btrfs_balance_control *bctl;
	struct btrfs_balance_item *item;
	struct btrfs_disk_balance_args disk_bargs;
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_key key;
	int ret;

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

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

3367
	ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0);
3368
	if (ret < 0)
3369
		goto out;
3370 3371
	if (ret > 0) { /* ret = -ENOENT; */
		ret = 0;
3372 3373 3374 3375 3376 3377 3378
		goto out;
	}

	bctl = kzalloc(sizeof(*bctl), GFP_NOFS);
	if (!bctl) {
		ret = -ENOMEM;
		goto out;
3379 3380 3381 3382 3383
	}

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

3384 3385 3386
	bctl->fs_info = fs_info;
	bctl->flags = btrfs_balance_flags(leaf, item);
	bctl->flags |= BTRFS_BALANCE_RESUME;
3387 3388 3389 3390 3391 3392 3393 3394

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

3395 3396
	WARN_ON(atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1));

3397 3398
	mutex_lock(&fs_info->volume_mutex);
	mutex_lock(&fs_info->balance_mutex);
3399

3400 3401 3402 3403
	set_balance_control(bctl);

	mutex_unlock(&fs_info->balance_mutex);
	mutex_unlock(&fs_info->volume_mutex);
3404 3405
out:
	btrfs_free_path(path);
3406 3407 3408
	return ret;
}

3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437
int btrfs_pause_balance(struct btrfs_fs_info *fs_info)
{
	int ret = 0;

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

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

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

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

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

3438 3439
int btrfs_cancel_balance(struct btrfs_fs_info *fs_info)
{
3440 3441 3442
	if (fs_info->sb->s_flags & MS_RDONLY)
		return -EROFS;

3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476
	mutex_lock(&fs_info->balance_mutex);
	if (!fs_info->balance_ctl) {
		mutex_unlock(&fs_info->balance_mutex);
		return -ENOTCONN;
	}

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

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

		mutex_unlock(&fs_info->volume_mutex);
	}

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

3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488
static int btrfs_uuid_scan_kthread(void *data)
{
	struct btrfs_fs_info *fs_info = data;
	struct btrfs_root *root = fs_info->tree_root;
	struct btrfs_key key;
	struct btrfs_key max_key;
	struct btrfs_path *path = NULL;
	int ret = 0;
	struct extent_buffer *eb;
	int slot;
	struct btrfs_root_item root_item;
	u32 item_size;
3489
	struct btrfs_trans_handle *trans = NULL;
3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507

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

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

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

	path->keep_locks = 1;

	while (1) {
3508
		ret = btrfs_search_forward(root, &key, path, 0);
3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531
		if (ret) {
			if (ret > 0)
				ret = 0;
			break;
		}

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

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

		read_extent_buffer(eb, &root_item,
				   btrfs_item_ptr_offset(eb, slot),
				   (int)sizeof(root_item));
		if (btrfs_root_refs(&root_item) == 0)
			goto skip;
3532 3533 3534 3535 3536 3537 3538

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

			btrfs_release_path(path);
3539 3540 3541 3542 3543 3544 3545 3546 3547
			/*
			 * 1 - subvol uuid item
			 * 1 - received_subvol uuid item
			 */
			trans = btrfs_start_transaction(fs_info->uuid_root, 2);
			if (IS_ERR(trans)) {
				ret = PTR_ERR(trans);
				break;
			}
3548 3549 3550 3551 3552 3553
			continue;
		} else {
			goto skip;
		}
update_tree:
		if (!btrfs_is_empty_uuid(root_item.uuid)) {
3554 3555 3556 3557 3558
			ret = btrfs_uuid_tree_add(trans, fs_info->uuid_root,
						  root_item.uuid,
						  BTRFS_UUID_KEY_SUBVOL,
						  key.objectid);
			if (ret < 0) {
3559
				btrfs_warn(fs_info, "uuid_tree_add failed %d",
3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570
					ret);
				break;
			}
		}

		if (!btrfs_is_empty_uuid(root_item.received_uuid)) {
			ret = btrfs_uuid_tree_add(trans, fs_info->uuid_root,
						  root_item.received_uuid,
						 BTRFS_UUID_KEY_RECEIVED_SUBVOL,
						  key.objectid);
			if (ret < 0) {
3571
				btrfs_warn(fs_info, "uuid_tree_add failed %d",
3572 3573 3574 3575 3576
					ret);
				break;
			}
		}

3577
skip:
3578 3579
		if (trans) {
			ret = btrfs_end_transaction(trans, fs_info->uuid_root);
3580
			trans = NULL;
3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602
			if (ret)
				break;
		}

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

out:
	btrfs_free_path(path);
3603 3604
	if (trans && !IS_ERR(trans))
		btrfs_end_transaction(trans, fs_info->uuid_root);
3605
	if (ret)
3606
		btrfs_warn(fs_info, "btrfs_uuid_scan_kthread failed %d", ret);
3607 3608
	else
		fs_info->update_uuid_tree_gen = 1;
3609 3610 3611 3612
	up(&fs_info->uuid_tree_rescan_sem);
	return 0;
}

3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669
/*
 * Callback for btrfs_uuid_tree_iterate().
 * returns:
 * 0	check succeeded, the entry is not outdated.
 * < 0	if an error occured.
 * > 0	if the check failed, which means the caller shall remove the entry.
 */
static int btrfs_check_uuid_tree_entry(struct btrfs_fs_info *fs_info,
				       u8 *uuid, u8 type, u64 subid)
{
	struct btrfs_key key;
	int ret = 0;
	struct btrfs_root *subvol_root;

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

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

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

out:
	return ret;
}

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

	/*
	 * 1st step is to iterate through the existing UUID tree and
	 * to delete all entries that contain outdated data.
	 * 2nd step is to add all missing entries to the UUID tree.
	 */
	ret = btrfs_uuid_tree_iterate(fs_info, btrfs_check_uuid_tree_entry);
	if (ret < 0) {
3670
		btrfs_warn(fs_info, "iterating uuid_tree failed %d", ret);
3671 3672 3673 3674 3675 3676
		up(&fs_info->uuid_tree_rescan_sem);
		return ret;
	}
	return btrfs_uuid_scan_kthread(data);
}

3677 3678 3679 3680 3681
int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_root *tree_root = fs_info->tree_root;
	struct btrfs_root *uuid_root;
3682 3683
	struct task_struct *task;
	int ret;
3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702

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

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

	fs_info->uuid_root = uuid_root;

3703 3704 3705 3706 3707 3708 3709
	ret = btrfs_commit_transaction(trans, tree_root);
	if (ret)
		return ret;

	down(&fs_info->uuid_tree_rescan_sem);
	task = kthread_run(btrfs_uuid_scan_kthread, fs_info, "btrfs-uuid");
	if (IS_ERR(task)) {
3710
		/* fs_info->update_uuid_tree_gen remains 0 in all error case */
3711
		btrfs_warn(fs_info, "failed to start uuid_scan task");
3712 3713 3714 3715 3716
		up(&fs_info->uuid_tree_rescan_sem);
		return PTR_ERR(task);
	}

	return 0;
3717
}
3718

3719 3720 3721 3722 3723 3724 3725 3726
int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info)
{
	struct task_struct *task;

	down(&fs_info->uuid_tree_rescan_sem);
	task = kthread_run(btrfs_uuid_rescan_kthread, fs_info, "btrfs-uuid");
	if (IS_ERR(task)) {
		/* fs_info->update_uuid_tree_gen remains 0 in all error case */
3727
		btrfs_warn(fs_info, "failed to start uuid_rescan task");
3728 3729 3730 3731 3732 3733 3734
		up(&fs_info->uuid_tree_rescan_sem);
		return PTR_ERR(task);
	}

	return 0;
}

3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751
/*
 * 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;
3752 3753
	int failed = 0;
	bool retried = false;
3754 3755
	struct extent_buffer *l;
	struct btrfs_key key;
3756
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
3757
	u64 old_total = btrfs_super_total_bytes(super_copy);
3758
	u64 old_size = device->total_bytes;
3759 3760
	u64 diff = device->total_bytes - new_size;

3761 3762 3763
	if (device->is_tgtdev_for_dev_replace)
		return -EINVAL;

3764 3765 3766 3767 3768 3769
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	path->reada = 2;

3770 3771
	lock_chunks(root);

3772
	device->total_bytes = new_size;
3773
	if (device->writeable) {
Y
Yan Zheng 已提交
3774
		device->fs_devices->total_rw_bytes -= diff;
3775 3776 3777 3778
		spin_lock(&root->fs_info->free_chunk_lock);
		root->fs_info->free_chunk_space -= diff;
		spin_unlock(&root->fs_info->free_chunk_lock);
	}
3779
	unlock_chunks(root);
3780

3781
again:
3782 3783 3784 3785
	key.objectid = device->devid;
	key.offset = (u64)-1;
	key.type = BTRFS_DEV_EXTENT_KEY;

3786
	do {
3787 3788 3789 3790 3791 3792 3793 3794 3795
		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;
3796
			btrfs_release_path(path);
3797
			break;
3798 3799 3800 3801 3802 3803
		}

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

3804
		if (key.objectid != device->devid) {
3805
			btrfs_release_path(path);
3806
			break;
3807
		}
3808 3809 3810 3811

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

3812
		if (key.offset + length <= new_size) {
3813
			btrfs_release_path(path);
3814
			break;
3815
		}
3816 3817 3818 3819

		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);
3820
		btrfs_release_path(path);
3821 3822 3823

		ret = btrfs_relocate_chunk(root, chunk_tree, chunk_objectid,
					   chunk_offset);
3824
		if (ret && ret != -ENOSPC)
3825
			goto done;
3826 3827
		if (ret == -ENOSPC)
			failed++;
3828
	} while (key.offset-- > 0);
3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840

	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;
3841 3842 3843
		spin_lock(&root->fs_info->free_chunk_lock);
		root->fs_info->free_chunk_space += diff;
		spin_unlock(&root->fs_info->free_chunk_lock);
3844 3845
		unlock_chunks(root);
		goto done;
3846 3847
	}

3848
	/* Shrinking succeeded, else we would be at "done". */
3849
	trans = btrfs_start_transaction(root, 0);
3850 3851 3852 3853 3854
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto done;
	}

3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868
	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);
3869 3870 3871 3872 3873
done:
	btrfs_free_path(path);
	return ret;
}

3874
static int btrfs_add_system_chunk(struct btrfs_root *root,
3875 3876 3877
			   struct btrfs_key *key,
			   struct btrfs_chunk *chunk, int item_size)
{
3878
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896
	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;
}

3897 3898 3899 3900
/*
 * sort the devices in descending order by max_avail, total_avail
 */
static int btrfs_cmp_device_info(const void *a, const void *b)
3901
{
3902 3903
	const struct btrfs_device_info *di_a = a;
	const struct btrfs_device_info *di_b = b;
3904

3905
	if (di_a->max_avail > di_b->max_avail)
3906
		return -1;
3907
	if (di_a->max_avail < di_b->max_avail)
3908
		return 1;
3909 3910 3911 3912 3913
	if (di_a->total_avail > di_b->total_avail)
		return -1;
	if (di_a->total_avail < di_b->total_avail)
		return 1;
	return 0;
3914
}
3915

3916
static struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = {
3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956
	[BTRFS_RAID_RAID10] = {
		.sub_stripes	= 2,
		.dev_stripes	= 1,
		.devs_max	= 0,	/* 0 == as many as possible */
		.devs_min	= 4,
		.devs_increment	= 2,
		.ncopies	= 2,
	},
	[BTRFS_RAID_RAID1] = {
		.sub_stripes	= 1,
		.dev_stripes	= 1,
		.devs_max	= 2,
		.devs_min	= 2,
		.devs_increment	= 2,
		.ncopies	= 2,
	},
	[BTRFS_RAID_DUP] = {
		.sub_stripes	= 1,
		.dev_stripes	= 2,
		.devs_max	= 1,
		.devs_min	= 1,
		.devs_increment	= 1,
		.ncopies	= 2,
	},
	[BTRFS_RAID_RAID0] = {
		.sub_stripes	= 1,
		.dev_stripes	= 1,
		.devs_max	= 0,
		.devs_min	= 2,
		.devs_increment	= 1,
		.ncopies	= 1,
	},
	[BTRFS_RAID_SINGLE] = {
		.sub_stripes	= 1,
		.dev_stripes	= 1,
		.devs_max	= 1,
		.devs_min	= 1,
		.devs_increment	= 1,
		.ncopies	= 1,
	},
3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972
	[BTRFS_RAID_RAID5] = {
		.sub_stripes	= 1,
		.dev_stripes	= 1,
		.devs_max	= 0,
		.devs_min	= 2,
		.devs_increment	= 1,
		.ncopies	= 2,
	},
	[BTRFS_RAID_RAID6] = {
		.sub_stripes	= 1,
		.dev_stripes	= 1,
		.devs_max	= 0,
		.devs_min	= 3,
		.devs_increment	= 1,
		.ncopies	= 3,
	},
3973 3974
};

D
David Woodhouse 已提交
3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985
static u32 find_raid56_stripe_len(u32 data_devices, u32 dev_stripe_target)
{
	/* TODO allow them to set a preferred stripe size */
	return 64 * 1024;
}

static void check_raid56_incompat_flag(struct btrfs_fs_info *info, u64 type)
{
	if (!(type & (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)))
		return;

3986
	btrfs_set_fs_incompat(info, RAID56);
D
David Woodhouse 已提交
3987 3988
}

3989
static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
3990 3991
			       struct btrfs_root *extent_root, u64 start,
			       u64 type)
3992
{
3993 3994 3995 3996 3997 3998 3999 4000 4001
	struct btrfs_fs_info *info = extent_root->fs_info;
	struct btrfs_fs_devices *fs_devices = info->fs_devices;
	struct list_head *cur;
	struct map_lookup *map = NULL;
	struct extent_map_tree *em_tree;
	struct extent_map *em;
	struct btrfs_device_info *devices_info = NULL;
	u64 total_avail;
	int num_stripes;	/* total number of stripes to allocate */
D
David Woodhouse 已提交
4002 4003
	int data_stripes;	/* number of stripes that count for
				   block group size */
4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014
	int sub_stripes;	/* sub_stripes info for map */
	int dev_stripes;	/* stripes per dev */
	int devs_max;		/* max devs to use */
	int devs_min;		/* min devs needed */
	int devs_increment;	/* ndevs has to be a multiple of this */
	int ncopies;		/* how many copies to data has */
	int ret;
	u64 max_stripe_size;
	u64 max_chunk_size;
	u64 stripe_size;
	u64 num_bytes;
D
David Woodhouse 已提交
4015
	u64 raid_stripe_len = BTRFS_STRIPE_LEN;
4016 4017 4018
	int ndevs;
	int i;
	int j;
4019
	int index;
4020

4021
	BUG_ON(!alloc_profile_is_valid(type, 0));
4022

4023 4024
	if (list_empty(&fs_devices->alloc_list))
		return -ENOSPC;
4025

4026
	index = __get_raid_index(type);
4027

4028 4029 4030 4031 4032 4033
	sub_stripes = btrfs_raid_array[index].sub_stripes;
	dev_stripes = btrfs_raid_array[index].dev_stripes;
	devs_max = btrfs_raid_array[index].devs_max;
	devs_min = btrfs_raid_array[index].devs_min;
	devs_increment = btrfs_raid_array[index].devs_increment;
	ncopies = btrfs_raid_array[index].ncopies;
4034

4035
	if (type & BTRFS_BLOCK_GROUP_DATA) {
4036 4037
		max_stripe_size = 1024 * 1024 * 1024;
		max_chunk_size = 10 * max_stripe_size;
4038
	} else if (type & BTRFS_BLOCK_GROUP_METADATA) {
4039 4040 4041 4042 4043
		/* for larger filesystems, use larger metadata chunks */
		if (fs_devices->total_rw_bytes > 50ULL * 1024 * 1024 * 1024)
			max_stripe_size = 1024 * 1024 * 1024;
		else
			max_stripe_size = 256 * 1024 * 1024;
4044
		max_chunk_size = max_stripe_size;
4045
	} else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
4046
		max_stripe_size = 32 * 1024 * 1024;
4047 4048
		max_chunk_size = 2 * max_stripe_size;
	} else {
4049
		btrfs_err(info, "invalid chunk type 0x%llx requested\n",
4050 4051
		       type);
		BUG_ON(1);
4052 4053
	}

Y
Yan Zheng 已提交
4054 4055 4056
	/* 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);
4057

4058 4059 4060 4061
	devices_info = kzalloc(sizeof(*devices_info) * fs_devices->rw_devices,
			       GFP_NOFS);
	if (!devices_info)
		return -ENOMEM;
4062

4063
	cur = fs_devices->alloc_list.next;
4064

4065
	/*
4066 4067
	 * in the first pass through the devices list, we gather information
	 * about the available holes on each device.
4068
	 */
4069 4070 4071 4072 4073
	ndevs = 0;
	while (cur != &fs_devices->alloc_list) {
		struct btrfs_device *device;
		u64 max_avail;
		u64 dev_offset;
4074

4075
		device = list_entry(cur, struct btrfs_device, dev_alloc_list);
4076

4077
		cur = cur->next;
4078

4079
		if (!device->writeable) {
J
Julia Lawall 已提交
4080
			WARN(1, KERN_ERR
4081
			       "BTRFS: read-only device in alloc_list\n");
4082 4083
			continue;
		}
4084

4085 4086
		if (!device->in_fs_metadata ||
		    device->is_tgtdev_for_dev_replace)
4087
			continue;
4088

4089 4090 4091 4092
		if (device->total_bytes > device->bytes_used)
			total_avail = device->total_bytes - device->bytes_used;
		else
			total_avail = 0;
4093 4094 4095 4096

		/* If there is no space on this device, skip it. */
		if (total_avail == 0)
			continue;
4097

4098
		ret = find_free_dev_extent(trans, device,
4099 4100 4101 4102
					   max_stripe_size * dev_stripes,
					   &dev_offset, &max_avail);
		if (ret && ret != -ENOSPC)
			goto error;
4103

4104 4105
		if (ret == 0)
			max_avail = max_stripe_size * dev_stripes;
4106

4107 4108
		if (max_avail < BTRFS_STRIPE_LEN * dev_stripes)
			continue;
4109

4110 4111 4112 4113 4114
		if (ndevs == fs_devices->rw_devices) {
			WARN(1, "%s: found more than %llu devices\n",
			     __func__, fs_devices->rw_devices);
			break;
		}
4115 4116 4117 4118 4119 4120
		devices_info[ndevs].dev_offset = dev_offset;
		devices_info[ndevs].max_avail = max_avail;
		devices_info[ndevs].total_avail = total_avail;
		devices_info[ndevs].dev = device;
		++ndevs;
	}
4121

4122 4123 4124 4125 4126
	/*
	 * now sort the devices by hole size / available space
	 */
	sort(devices_info, ndevs, sizeof(struct btrfs_device_info),
	     btrfs_cmp_device_info, NULL);
4127

4128 4129
	/* round down to number of usable stripes */
	ndevs -= ndevs % devs_increment;
4130

4131 4132 4133
	if (ndevs < devs_increment * sub_stripes || ndevs < devs_min) {
		ret = -ENOSPC;
		goto error;
4134
	}
4135

4136 4137 4138 4139 4140 4141 4142 4143
	if (devs_max && ndevs > devs_max)
		ndevs = devs_max;
	/*
	 * the primary goal is to maximize the number of stripes, so use as many
	 * devices as possible, even if the stripes are not maximum sized.
	 */
	stripe_size = devices_info[ndevs-1].max_avail;
	num_stripes = ndevs * dev_stripes;
4144

D
David Woodhouse 已提交
4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160
	/*
	 * this will have to be fixed for RAID1 and RAID10 over
	 * more drives
	 */
	data_stripes = num_stripes / ncopies;

	if (type & BTRFS_BLOCK_GROUP_RAID5) {
		raid_stripe_len = find_raid56_stripe_len(ndevs - 1,
				 btrfs_super_stripesize(info->super_copy));
		data_stripes = num_stripes - 1;
	}
	if (type & BTRFS_BLOCK_GROUP_RAID6) {
		raid_stripe_len = find_raid56_stripe_len(ndevs - 2,
				 btrfs_super_stripesize(info->super_copy));
		data_stripes = num_stripes - 2;
	}
4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181

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

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

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

4182
	do_div(stripe_size, dev_stripes);
4183 4184

	/* align to BTRFS_STRIPE_LEN */
D
David Woodhouse 已提交
4185 4186
	do_div(stripe_size, raid_stripe_len);
	stripe_size *= raid_stripe_len;
4187 4188 4189 4190 4191 4192 4193

	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
	if (!map) {
		ret = -ENOMEM;
		goto error;
	}
	map->num_stripes = num_stripes;
4194

4195 4196 4197 4198 4199 4200
	for (i = 0; i < ndevs; ++i) {
		for (j = 0; j < dev_stripes; ++j) {
			int s = i * dev_stripes + j;
			map->stripes[s].dev = devices_info[i].dev;
			map->stripes[s].physical = devices_info[i].dev_offset +
						   j * stripe_size;
4201 4202
		}
	}
Y
Yan Zheng 已提交
4203
	map->sector_size = extent_root->sectorsize;
D
David Woodhouse 已提交
4204 4205 4206
	map->stripe_len = raid_stripe_len;
	map->io_align = raid_stripe_len;
	map->io_width = raid_stripe_len;
Y
Yan Zheng 已提交
4207 4208
	map->type = type;
	map->sub_stripes = sub_stripes;
4209

D
David Woodhouse 已提交
4210
	num_bytes = stripe_size * data_stripes;
4211

4212
	trace_btrfs_chunk_alloc(info->chunk_root, map, start, num_bytes);
4213

4214
	em = alloc_extent_map();
Y
Yan Zheng 已提交
4215
	if (!em) {
4216 4217
		ret = -ENOMEM;
		goto error;
4218
	}
Y
Yan Zheng 已提交
4219 4220
	em->bdev = (struct block_device *)map;
	em->start = start;
4221
	em->len = num_bytes;
Y
Yan Zheng 已提交
4222 4223
	em->block_start = 0;
	em->block_len = em->len;
4224
	em->orig_block_len = stripe_size;
4225

Y
Yan Zheng 已提交
4226
	em_tree = &extent_root->fs_info->mapping_tree.map_tree;
4227
	write_lock(&em_tree->lock);
4228
	ret = add_extent_mapping(em_tree, em, 0);
4229 4230 4231 4232
	if (!ret) {
		list_add_tail(&em->list, &trans->transaction->pending_chunks);
		atomic_inc(&em->refs);
	}
4233
	write_unlock(&em_tree->lock);
4234 4235
	if (ret) {
		free_extent_map(em);
4236
		goto error;
4237
	}
4238

4239 4240 4241
	ret = btrfs_make_block_group(trans, extent_root, 0, type,
				     BTRFS_FIRST_CHUNK_TREE_OBJECTID,
				     start, num_bytes);
4242 4243
	if (ret)
		goto error_del_extent;
Y
Yan Zheng 已提交
4244

4245
	free_extent_map(em);
D
David Woodhouse 已提交
4246 4247
	check_raid56_incompat_flag(extent_root->fs_info, type);

4248
	kfree(devices_info);
Y
Yan Zheng 已提交
4249
	return 0;
4250

4251
error_del_extent:
4252 4253 4254 4255 4256 4257 4258 4259
	write_lock(&em_tree->lock);
	remove_extent_mapping(em_tree, em);
	write_unlock(&em_tree->lock);

	/* One for our allocation */
	free_extent_map(em);
	/* One for the tree reference */
	free_extent_map(em);
4260 4261 4262 4263
error:
	kfree(map);
	kfree(devices_info);
	return ret;
Y
Yan Zheng 已提交
4264 4265
}

4266
int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
Y
Yan Zheng 已提交
4267
				struct btrfs_root *extent_root,
4268
				u64 chunk_offset, u64 chunk_size)
Y
Yan Zheng 已提交
4269 4270 4271 4272 4273 4274
{
	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;
4275 4276 4277 4278 4279 4280 4281
	struct extent_map_tree *em_tree;
	struct extent_map *em;
	struct map_lookup *map;
	size_t item_size;
	u64 dev_offset;
	u64 stripe_size;
	int i = 0;
Y
Yan Zheng 已提交
4282 4283
	int ret;

4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306
	em_tree = &extent_root->fs_info->mapping_tree.map_tree;
	read_lock(&em_tree->lock);
	em = lookup_extent_mapping(em_tree, chunk_offset, chunk_size);
	read_unlock(&em_tree->lock);

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

	if (em->start != chunk_offset || em->len != chunk_size) {
		btrfs_crit(extent_root->fs_info, "found a bad mapping, wanted"
			  " %Lu-%Lu, found %Lu-%Lu\n", chunk_offset,
			  chunk_size, em->start, em->len);
		free_extent_map(em);
		return -EINVAL;
	}

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

Y
Yan Zheng 已提交
4307
	chunk = kzalloc(item_size, GFP_NOFS);
4308 4309 4310 4311 4312 4313 4314 4315
	if (!chunk) {
		ret = -ENOMEM;
		goto out;
	}

	for (i = 0; i < map->num_stripes; i++) {
		device = map->stripes[i].dev;
		dev_offset = map->stripes[i].physical;
Y
Yan Zheng 已提交
4316 4317

		device->bytes_used += stripe_size;
4318
		ret = btrfs_update_device(trans, device);
4319
		if (ret)
4320 4321 4322 4323 4324 4325 4326 4327
			goto out;
		ret = btrfs_alloc_dev_extent(trans, device,
					     chunk_root->root_key.objectid,
					     BTRFS_FIRST_CHUNK_TREE_OBJECTID,
					     chunk_offset, dev_offset,
					     stripe_size);
		if (ret)
			goto out;
Y
Yan Zheng 已提交
4328 4329
	}

4330 4331 4332 4333 4334
	spin_lock(&extent_root->fs_info->free_chunk_lock);
	extent_root->fs_info->free_chunk_space -= (stripe_size *
						   map->num_stripes);
	spin_unlock(&extent_root->fs_info->free_chunk_lock);

Y
Yan Zheng 已提交
4335
	stripe = &chunk->stripe;
4336 4337 4338
	for (i = 0; i < map->num_stripes; i++) {
		device = map->stripes[i].dev;
		dev_offset = map->stripes[i].physical;
4339

4340 4341 4342
		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
4343
		stripe++;
4344 4345
	}

Y
Yan Zheng 已提交
4346
	btrfs_set_stack_chunk_length(chunk, chunk_size);
4347
	btrfs_set_stack_chunk_owner(chunk, extent_root->root_key.objectid);
Y
Yan Zheng 已提交
4348 4349 4350 4351 4352
	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);
4353
	btrfs_set_stack_chunk_sector_size(chunk, extent_root->sectorsize);
Y
Yan Zheng 已提交
4354
	btrfs_set_stack_chunk_sub_stripes(chunk, map->sub_stripes);
4355

Y
Yan Zheng 已提交
4356 4357 4358
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.type = BTRFS_CHUNK_ITEM_KEY;
	key.offset = chunk_offset;
4359

Y
Yan Zheng 已提交
4360
	ret = btrfs_insert_item(trans, chunk_root, &key, chunk, item_size);
4361 4362 4363 4364 4365
	if (ret == 0 && map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
		/*
		 * TODO: Cleanup of inserted chunk root in case of
		 * failure.
		 */
4366
		ret = btrfs_add_system_chunk(chunk_root, &key, chunk,
Y
Yan Zheng 已提交
4367
					     item_size);
4368
	}
4369

4370
out:
4371
	kfree(chunk);
4372
	free_extent_map(em);
4373
	return ret;
Y
Yan Zheng 已提交
4374
}
4375

Y
Yan Zheng 已提交
4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387
/*
 * 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;

4388 4389
	chunk_offset = find_next_chunk(extent_root->fs_info);
	return __btrfs_alloc_chunk(trans, extent_root, chunk_offset, type);
Y
Yan Zheng 已提交
4390 4391
}

4392
static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
Y
Yan Zheng 已提交
4393 4394 4395 4396 4397 4398 4399 4400 4401 4402
					 struct btrfs_root *root,
					 struct btrfs_device *device)
{
	u64 chunk_offset;
	u64 sys_chunk_offset;
	u64 alloc_profile;
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct btrfs_root *extent_root = fs_info->extent_root;
	int ret;

4403
	chunk_offset = find_next_chunk(fs_info);
4404
	alloc_profile = btrfs_get_alloc_profile(extent_root, 0);
4405 4406
	ret = __btrfs_alloc_chunk(trans, extent_root, chunk_offset,
				  alloc_profile);
4407 4408
	if (ret)
		return ret;
Y
Yan Zheng 已提交
4409

4410
	sys_chunk_offset = find_next_chunk(root->fs_info);
4411
	alloc_profile = btrfs_get_alloc_profile(fs_info->chunk_root, 0);
4412 4413
	ret = __btrfs_alloc_chunk(trans, extent_root, sys_chunk_offset,
				  alloc_profile);
4414 4415 4416 4417
	if (ret) {
		btrfs_abort_transaction(trans, root, ret);
		goto out;
	}
Y
Yan Zheng 已提交
4418 4419

	ret = btrfs_add_device(trans, fs_info->chunk_root, device);
4420
	if (ret)
4421 4422
		btrfs_abort_transaction(trans, root, ret);
out:
4423
	return ret;
Y
Yan Zheng 已提交
4424 4425 4426 4427 4428 4429 4430 4431 4432 4433
}

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;

4434
	read_lock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
4435
	em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
4436
	read_unlock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
4437 4438 4439
	if (!em)
		return 1;

4440 4441 4442 4443 4444
	if (btrfs_test_opt(root, DEGRADED)) {
		free_extent_map(em);
		return 0;
	}

Y
Yan Zheng 已提交
4445 4446 4447 4448 4449 4450 4451
	map = (struct map_lookup *)em->bdev;
	for (i = 0; i < map->num_stripes; i++) {
		if (!map->stripes[i].dev->writeable) {
			readonly = 1;
			break;
		}
	}
4452
	free_extent_map(em);
Y
Yan Zheng 已提交
4453
	return readonly;
4454 4455 4456 4457
}

void btrfs_mapping_init(struct btrfs_mapping_tree *tree)
{
4458
	extent_map_tree_init(&tree->map_tree);
4459 4460 4461 4462 4463 4464
}

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

4465
	while (1) {
4466
		write_lock(&tree->map_tree.lock);
4467 4468 4469
		em = lookup_extent_mapping(&tree->map_tree, 0, (u64)-1);
		if (em)
			remove_extent_mapping(&tree->map_tree, em);
4470
		write_unlock(&tree->map_tree.lock);
4471 4472 4473 4474 4475 4476 4477 4478 4479 4480
		if (!em)
			break;
		kfree(em->bdev);
		/* once for us */
		free_extent_map(em);
		/* once for the tree */
		free_extent_map(em);
	}
}

4481
int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len)
4482
{
4483
	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
4484 4485 4486 4487 4488
	struct extent_map *em;
	struct map_lookup *map;
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	int ret;

4489
	read_lock(&em_tree->lock);
4490
	em = lookup_extent_mapping(em_tree, logical, len);
4491
	read_unlock(&em_tree->lock);
4492

4493 4494 4495 4496 4497 4498
	/*
	 * We could return errors for these cases, but that could get ugly and
	 * we'd probably do the same thing which is just not do anything else
	 * and exit, so return 1 so the callers don't try to use other copies.
	 */
	if (!em) {
4499
		btrfs_crit(fs_info, "No mapping for %Lu-%Lu\n", logical,
4500 4501 4502 4503 4504
			    logical+len);
		return 1;
	}

	if (em->start > logical || em->start + em->len < logical) {
4505
		btrfs_crit(fs_info, "Invalid mapping for %Lu-%Lu, got "
4506 4507
			    "%Lu-%Lu\n", logical, logical+len, em->start,
			    em->start + em->len);
4508
		free_extent_map(em);
4509 4510 4511
		return 1;
	}

4512 4513 4514
	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
4515 4516
	else if (map->type & BTRFS_BLOCK_GROUP_RAID10)
		ret = map->sub_stripes;
D
David Woodhouse 已提交
4517 4518 4519 4520
	else if (map->type & BTRFS_BLOCK_GROUP_RAID5)
		ret = 2;
	else if (map->type & BTRFS_BLOCK_GROUP_RAID6)
		ret = 3;
4521 4522 4523
	else
		ret = 1;
	free_extent_map(em);
4524 4525 4526 4527 4528 4529

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

4530 4531 4532
	return ret;
}

D
David Woodhouse 已提交
4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578
unsigned long btrfs_full_stripe_len(struct btrfs_root *root,
				    struct btrfs_mapping_tree *map_tree,
				    u64 logical)
{
	struct extent_map *em;
	struct map_lookup *map;
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	unsigned long len = root->sectorsize;

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

	BUG_ON(em->start > logical || em->start + em->len < logical);
	map = (struct map_lookup *)em->bdev;
	if (map->type & (BTRFS_BLOCK_GROUP_RAID5 |
			 BTRFS_BLOCK_GROUP_RAID6)) {
		len = map->stripe_len * nr_data_stripes(map);
	}
	free_extent_map(em);
	return len;
}

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

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

	BUG_ON(em->start > logical || em->start + em->len < logical);
	map = (struct map_lookup *)em->bdev;
	if (map->type & (BTRFS_BLOCK_GROUP_RAID5 |
			 BTRFS_BLOCK_GROUP_RAID6))
		ret = 1;
	free_extent_map(em);
	return ret;
}

4579 4580 4581
static int find_live_mirror(struct btrfs_fs_info *fs_info,
			    struct map_lookup *map, int first, int num,
			    int optimal, int dev_replace_is_ongoing)
4582 4583
{
	int i;
4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607
	int tolerance;
	struct btrfs_device *srcdev;

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

	/*
	 * try to avoid the drive that is the source drive for a
	 * dev-replace procedure, only choose it if no other non-missing
	 * mirror is available
	 */
	for (tolerance = 0; tolerance < 2; tolerance++) {
		if (map->stripes[optimal].dev->bdev &&
		    (tolerance || map->stripes[optimal].dev != srcdev))
			return optimal;
		for (i = first; i < first + num; i++) {
			if (map->stripes[i].dev->bdev &&
			    (tolerance || map->stripes[i].dev != srcdev))
				return i;
		}
4608
	}
4609

4610 4611 4612 4613 4614 4615
	/* we couldn't find one that doesn't fail.  Just return something
	 * and the io error handling code will clean up eventually
	 */
	return optimal;
}

D
David Woodhouse 已提交
4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644
static inline int parity_smaller(u64 a, u64 b)
{
	return a > b;
}

/* Bubble-sort the stripe set to put the parity/syndrome stripes last */
static void sort_parity_stripes(struct btrfs_bio *bbio, u64 *raid_map)
{
	struct btrfs_bio_stripe s;
	int i;
	u64 l;
	int again = 1;

	while (again) {
		again = 0;
		for (i = 0; i < bbio->num_stripes - 1; i++) {
			if (parity_smaller(raid_map[i], raid_map[i+1])) {
				s = bbio->stripes[i];
				l = raid_map[i];
				bbio->stripes[i] = bbio->stripes[i+1];
				raid_map[i] = raid_map[i+1];
				bbio->stripes[i+1] = s;
				raid_map[i+1] = l;
				again = 1;
			}
		}
	}
}

4645
static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
4646
			     u64 logical, u64 *length,
4647
			     struct btrfs_bio **bbio_ret,
D
David Woodhouse 已提交
4648
			     int mirror_num, u64 **raid_map_ret)
4649 4650 4651
{
	struct extent_map *em;
	struct map_lookup *map;
4652
	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
4653 4654
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	u64 offset;
4655
	u64 stripe_offset;
4656
	u64 stripe_end_offset;
4657
	u64 stripe_nr;
4658 4659
	u64 stripe_nr_orig;
	u64 stripe_nr_end;
D
David Woodhouse 已提交
4660 4661
	u64 stripe_len;
	u64 *raid_map = NULL;
4662
	int stripe_index;
4663
	int i;
4664
	int ret = 0;
4665
	int num_stripes;
4666
	int max_errors = 0;
4667
	struct btrfs_bio *bbio = NULL;
4668 4669 4670
	struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
	int dev_replace_is_ongoing = 0;
	int num_alloc_stripes;
4671 4672
	int patch_the_first_stripe_for_dev_replace = 0;
	u64 physical_to_patch_in_first_stripe = 0;
D
David Woodhouse 已提交
4673
	u64 raid56_full_stripe_start = (u64)-1;
4674

4675
	read_lock(&em_tree->lock);
4676
	em = lookup_extent_mapping(em_tree, logical, *length);
4677
	read_unlock(&em_tree->lock);
4678

4679
	if (!em) {
4680
		btrfs_crit(fs_info, "unable to find logical %llu len %llu",
4681
			logical, *length);
4682 4683 4684 4685 4686 4687 4688
		return -EINVAL;
	}

	if (em->start > logical || em->start + em->len < logical) {
		btrfs_crit(fs_info, "found a bad mapping, wanted %Lu, "
			   "found %Lu-%Lu\n", logical, em->start,
			   em->start + em->len);
4689
		free_extent_map(em);
4690
		return -EINVAL;
4691
	}
4692 4693 4694

	map = (struct map_lookup *)em->bdev;
	offset = logical - em->start;
4695

D
David Woodhouse 已提交
4696
	stripe_len = map->stripe_len;
4697 4698 4699 4700 4701
	stripe_nr = offset;
	/*
	 * stripe_nr counts the total number of stripes we have to stride
	 * to get to this block
	 */
D
David Woodhouse 已提交
4702
	do_div(stripe_nr, stripe_len);
4703

D
David Woodhouse 已提交
4704
	stripe_offset = stripe_nr * stripe_len;
4705 4706 4707 4708 4709
	BUG_ON(offset < stripe_offset);

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

D
David Woodhouse 已提交
4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728
	/* if we're here for raid56, we need to know the stripe aligned start */
	if (map->type & (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)) {
		unsigned long full_stripe_len = stripe_len * nr_data_stripes(map);
		raid56_full_stripe_start = offset;

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

	if (rw & REQ_DISCARD) {
		/* we don't discard raid56 yet */
		if (map->type &
		    (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)) {
			ret = -EOPNOTSUPP;
			goto out;
		}
4729
		*length = min_t(u64, em->len - offset, *length);
D
David Woodhouse 已提交
4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743
	} else if (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK) {
		u64 max_len;
		/* For writes to RAID[56], allow a full stripeset across all disks.
		   For other RAID types and for RAID[56] reads, just allow a single
		   stripe (on a single disk). */
		if (map->type & (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6) &&
		    (rw & REQ_WRITE)) {
			max_len = stripe_len * nr_data_stripes(map) -
				(offset - raid56_full_stripe_start);
		} else {
			/* we limit the length of each bio to what fits in a stripe */
			max_len = stripe_len - stripe_offset;
		}
		*length = min_t(u64, em->len - offset, max_len);
4744 4745 4746
	} else {
		*length = em->len - offset;
	}
4747

D
David Woodhouse 已提交
4748 4749
	/* This is for when we're called from btrfs_merge_bio_hook() and all
	   it cares about is the length */
4750
	if (!bbio_ret)
4751 4752
		goto out;

4753 4754 4755 4756 4757
	btrfs_dev_replace_lock(dev_replace);
	dev_replace_is_ongoing = btrfs_dev_replace_is_ongoing(dev_replace);
	if (!dev_replace_is_ongoing)
		btrfs_dev_replace_unlock(dev_replace);

4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781
	if (dev_replace_is_ongoing && mirror_num == map->num_stripes + 1 &&
	    !(rw & (REQ_WRITE | REQ_DISCARD | REQ_GET_READ_MIRRORS)) &&
	    dev_replace->tgtdev != NULL) {
		/*
		 * in dev-replace case, for repair case (that's the only
		 * case where the mirror is selected explicitly when
		 * calling btrfs_map_block), blocks left of the left cursor
		 * can also be read from the target drive.
		 * For REQ_GET_READ_MIRRORS, the target drive is added as
		 * the last one to the array of stripes. For READ, it also
		 * needs to be supported using the same mirror number.
		 * If the requested block is not left of the left cursor,
		 * EIO is returned. This can happen because btrfs_num_copies()
		 * returns one more in the dev-replace case.
		 */
		u64 tmp_length = *length;
		struct btrfs_bio *tmp_bbio = NULL;
		int tmp_num_stripes;
		u64 srcdev_devid = dev_replace->srcdev->devid;
		int index_srcdev = 0;
		int found = 0;
		u64 physical_of_found = 0;

		ret = __btrfs_map_block(fs_info, REQ_GET_READ_MIRRORS,
D
David Woodhouse 已提交
4782
			     logical, &tmp_length, &tmp_bbio, 0, NULL);
4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839
		if (ret) {
			WARN_ON(tmp_bbio != NULL);
			goto out;
		}

		tmp_num_stripes = tmp_bbio->num_stripes;
		if (mirror_num > tmp_num_stripes) {
			/*
			 * REQ_GET_READ_MIRRORS does not contain this
			 * mirror, that means that the requested area
			 * is not left of the left cursor
			 */
			ret = -EIO;
			kfree(tmp_bbio);
			goto out;
		}

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

		if (found) {
			mirror_num = index_srcdev + 1;
			patch_the_first_stripe_for_dev_replace = 1;
			physical_to_patch_in_first_stripe = physical_of_found;
		} else {
			WARN_ON(1);
			ret = -EIO;
			kfree(tmp_bbio);
			goto out;
		}

		kfree(tmp_bbio);
	} else if (mirror_num > map->num_stripes) {
		mirror_num = 0;
	}

4840
	num_stripes = 1;
4841
	stripe_index = 0;
4842
	stripe_nr_orig = stripe_nr;
4843
	stripe_nr_end = ALIGN(offset + *length, map->stripe_len);
4844 4845 4846
	do_div(stripe_nr_end, map->stripe_len);
	stripe_end_offset = stripe_nr_end * map->stripe_len -
			    (offset + *length);
D
David Woodhouse 已提交
4847

4848 4849 4850 4851 4852 4853
	if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
		if (rw & REQ_DISCARD)
			num_stripes = min_t(u64, map->num_stripes,
					    stripe_nr_end - stripe_nr_orig);
		stripe_index = do_div(stripe_nr, map->num_stripes);
	} else if (map->type & BTRFS_BLOCK_GROUP_RAID1) {
4854
		if (rw & (REQ_WRITE | REQ_DISCARD | REQ_GET_READ_MIRRORS))
4855
			num_stripes = map->num_stripes;
4856
		else if (mirror_num)
4857
			stripe_index = mirror_num - 1;
4858
		else {
4859
			stripe_index = find_live_mirror(fs_info, map, 0,
4860
					    map->num_stripes,
4861 4862
					    current->pid % map->num_stripes,
					    dev_replace_is_ongoing);
4863
			mirror_num = stripe_index + 1;
4864
		}
4865

4866
	} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
4867
		if (rw & (REQ_WRITE | REQ_DISCARD | REQ_GET_READ_MIRRORS)) {
4868
			num_stripes = map->num_stripes;
4869
		} else if (mirror_num) {
4870
			stripe_index = mirror_num - 1;
4871 4872 4873
		} else {
			mirror_num = 1;
		}
4874

C
Chris Mason 已提交
4875 4876 4877 4878 4879 4880
	} 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;

4881
		if (rw & (REQ_WRITE | REQ_GET_READ_MIRRORS))
4882
			num_stripes = map->sub_stripes;
4883 4884 4885 4886
		else if (rw & REQ_DISCARD)
			num_stripes = min_t(u64, map->sub_stripes *
					    (stripe_nr_end - stripe_nr_orig),
					    map->num_stripes);
C
Chris Mason 已提交
4887 4888
		else if (mirror_num)
			stripe_index += mirror_num - 1;
4889
		else {
4890
			int old_stripe_index = stripe_index;
4891 4892
			stripe_index = find_live_mirror(fs_info, map,
					      stripe_index,
4893
					      map->sub_stripes, stripe_index +
4894 4895
					      current->pid % map->sub_stripes,
					      dev_replace_is_ongoing);
4896
			mirror_num = stripe_index - old_stripe_index + 1;
4897
		}
D
David Woodhouse 已提交
4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916

	} else if (map->type & (BTRFS_BLOCK_GROUP_RAID5 |
				BTRFS_BLOCK_GROUP_RAID6)) {
		u64 tmp;

		if (bbio_ret && ((rw & REQ_WRITE) || mirror_num > 1)
		    && raid_map_ret) {
			int i, rot;

			/* push stripe_nr back to the start of the full stripe */
			stripe_nr = raid56_full_stripe_start;
			do_div(stripe_nr, stripe_len);

			stripe_index = do_div(stripe_nr, nr_data_stripes(map));

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

4917
			raid_map = kmalloc_array(num_stripes, sizeof(u64),
D
David Woodhouse 已提交
4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956
					   GFP_NOFS);
			if (!raid_map) {
				ret = -ENOMEM;
				goto out;
			}

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

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

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

			*length = map->stripe_len;
			stripe_index = 0;
			stripe_offset = 0;
		} else {
			/*
			 * Mirror #0 or #1 means the original data block.
			 * Mirror #2 is RAID5 parity block.
			 * Mirror #3 is RAID6 Q block.
			 */
			stripe_index = do_div(stripe_nr, nr_data_stripes(map));
			if (mirror_num > 1)
				stripe_index = nr_data_stripes(map) +
						mirror_num - 2;

			/* We distribute the parity blocks across stripes */
			tmp = stripe_nr + stripe_index;
			stripe_index = do_div(tmp, map->num_stripes);
		}
4957 4958 4959 4960 4961 4962 4963
	} 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);
4964
		mirror_num = stripe_index + 1;
4965
	}
4966
	BUG_ON(stripe_index >= map->num_stripes);
4967

4968
	num_alloc_stripes = num_stripes;
4969 4970 4971 4972 4973 4974
	if (dev_replace_is_ongoing) {
		if (rw & (REQ_WRITE | REQ_DISCARD))
			num_alloc_stripes <<= 1;
		if (rw & REQ_GET_READ_MIRRORS)
			num_alloc_stripes++;
	}
4975
	bbio = kzalloc(btrfs_bio_size(num_alloc_stripes), GFP_NOFS);
4976
	if (!bbio) {
4977
		kfree(raid_map);
4978 4979 4980 4981 4982
		ret = -ENOMEM;
		goto out;
	}
	atomic_set(&bbio->error, 0);

4983
	if (rw & REQ_DISCARD) {
4984 4985 4986 4987
		int factor = 0;
		int sub_stripes = 0;
		u64 stripes_per_dev = 0;
		u32 remaining_stripes = 0;
4988
		u32 last_stripe = 0;
4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001

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

			factor = map->num_stripes / sub_stripes;
			stripes_per_dev = div_u64_rem(stripe_nr_end -
						      stripe_nr_orig,
						      factor,
						      &remaining_stripes);
5002 5003
			div_u64_rem(stripe_nr_end - 1, factor, &last_stripe);
			last_stripe *= sub_stripes;
5004 5005
		}

5006
		for (i = 0; i < num_stripes; i++) {
5007
			bbio->stripes[i].physical =
5008 5009
				map->stripes[stripe_index].physical +
				stripe_offset + stripe_nr * map->stripe_len;
5010
			bbio->stripes[i].dev = map->stripes[stripe_index].dev;
5011

5012 5013 5014 5015
			if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
					 BTRFS_BLOCK_GROUP_RAID10)) {
				bbio->stripes[i].length = stripes_per_dev *
							  map->stripe_len;
5016

5017 5018 5019
				if (i / sub_stripes < remaining_stripes)
					bbio->stripes[i].length +=
						map->stripe_len;
5020 5021 5022 5023 5024 5025 5026 5027 5028

				/*
				 * Special for the first stripe and
				 * the last stripe:
				 *
				 * |-------|...|-------|
				 *     |----------|
				 *    off     end_off
				 */
5029
				if (i < sub_stripes)
5030
					bbio->stripes[i].length -=
5031
						stripe_offset;
5032 5033 5034 5035

				if (stripe_index >= last_stripe &&
				    stripe_index <= (last_stripe +
						     sub_stripes - 1))
5036
					bbio->stripes[i].length -=
5037
						stripe_end_offset;
5038

5039 5040
				if (i == sub_stripes - 1)
					stripe_offset = 0;
5041
			} else
5042
				bbio->stripes[i].length = *length;
5043 5044 5045 5046 5047 5048 5049 5050 5051 5052

			stripe_index++;
			if (stripe_index == map->num_stripes) {
				/* This could only happen for RAID0/10 */
				stripe_index = 0;
				stripe_nr++;
			}
		}
	} else {
		for (i = 0; i < num_stripes; i++) {
5053
			bbio->stripes[i].physical =
5054 5055 5056
				map->stripes[stripe_index].physical +
				stripe_offset +
				stripe_nr * map->stripe_len;
5057
			bbio->stripes[i].dev =
5058
				map->stripes[stripe_index].dev;
5059
			stripe_index++;
5060
		}
5061
	}
5062

5063
	if (rw & (REQ_WRITE | REQ_GET_READ_MIRRORS)) {
5064 5065
		if (map->type & (BTRFS_BLOCK_GROUP_RAID1 |
				 BTRFS_BLOCK_GROUP_RAID10 |
D
David Woodhouse 已提交
5066
				 BTRFS_BLOCK_GROUP_RAID5 |
5067 5068
				 BTRFS_BLOCK_GROUP_DUP)) {
			max_errors = 1;
D
David Woodhouse 已提交
5069 5070
		} else if (map->type & BTRFS_BLOCK_GROUP_RAID6) {
			max_errors = 2;
5071
		}
5072
	}
5073

5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106
	if (dev_replace_is_ongoing && (rw & (REQ_WRITE | REQ_DISCARD)) &&
	    dev_replace->tgtdev != NULL) {
		int index_where_to_add;
		u64 srcdev_devid = dev_replace->srcdev->devid;

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

				new->physical = old->physical;
				new->length = old->length;
				new->dev = dev_replace->tgtdev;
				index_where_to_add++;
				max_errors++;
			}
		}
		num_stripes = index_where_to_add;
5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152
	} else if (dev_replace_is_ongoing && (rw & REQ_GET_READ_MIRRORS) &&
		   dev_replace->tgtdev != NULL) {
		u64 srcdev_devid = dev_replace->srcdev->devid;
		int index_srcdev = 0;
		int found = 0;
		u64 physical_of_found = 0;

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

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

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

				num_stripes++;
			}
		}
5153 5154
	}

5155 5156 5157 5158
	*bbio_ret = bbio;
	bbio->num_stripes = num_stripes;
	bbio->max_errors = max_errors;
	bbio->mirror_num = mirror_num;
5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170

	/*
	 * this is the case that REQ_READ && dev_replace_is_ongoing &&
	 * mirror_num == num_stripes + 1 && dev_replace target drive is
	 * available as a mirror
	 */
	if (patch_the_first_stripe_for_dev_replace && num_stripes > 0) {
		WARN_ON(num_stripes > 1);
		bbio->stripes[0].dev = dev_replace->tgtdev;
		bbio->stripes[0].physical = physical_to_patch_in_first_stripe;
		bbio->mirror_num = map->num_stripes + 1;
	}
D
David Woodhouse 已提交
5171 5172 5173 5174
	if (raid_map) {
		sort_parity_stripes(bbio, raid_map);
		*raid_map_ret = raid_map;
	}
5175
out:
5176 5177
	if (dev_replace_is_ongoing)
		btrfs_dev_replace_unlock(dev_replace);
5178
	free_extent_map(em);
5179
	return ret;
5180 5181
}

5182
int btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
5183
		      u64 logical, u64 *length,
5184
		      struct btrfs_bio **bbio_ret, int mirror_num)
5185
{
5186
	return __btrfs_map_block(fs_info, rw, logical, length, bbio_ret,
D
David Woodhouse 已提交
5187
				 mirror_num, NULL);
5188 5189
}

Y
Yan Zheng 已提交
5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200
int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
		     u64 chunk_start, u64 physical, u64 devid,
		     u64 **logical, int *naddrs, int *stripe_len)
{
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	struct extent_map *em;
	struct map_lookup *map;
	u64 *buf;
	u64 bytenr;
	u64 length;
	u64 stripe_nr;
D
David Woodhouse 已提交
5201
	u64 rmap_len;
Y
Yan Zheng 已提交
5202 5203
	int i, j, nr = 0;

5204
	read_lock(&em_tree->lock);
Y
Yan Zheng 已提交
5205
	em = lookup_extent_mapping(em_tree, chunk_start, 1);
5206
	read_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
5207

5208
	if (!em) {
5209
		printk(KERN_ERR "BTRFS: couldn't find em for chunk %Lu\n",
5210 5211 5212 5213 5214
		       chunk_start);
		return -EIO;
	}

	if (em->start != chunk_start) {
5215
		printk(KERN_ERR "BTRFS: bad chunk start, em=%Lu, wanted=%Lu\n",
5216 5217 5218 5219
		       em->start, chunk_start);
		free_extent_map(em);
		return -EIO;
	}
Y
Yan Zheng 已提交
5220 5221 5222
	map = (struct map_lookup *)em->bdev;

	length = em->len;
D
David Woodhouse 已提交
5223 5224
	rmap_len = map->stripe_len;

Y
Yan Zheng 已提交
5225 5226 5227 5228
	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);
D
David Woodhouse 已提交
5229 5230 5231 5232 5233
	else if (map->type & (BTRFS_BLOCK_GROUP_RAID5 |
			      BTRFS_BLOCK_GROUP_RAID6)) {
		do_div(length, nr_data_stripes(map));
		rmap_len = map->stripe_len * nr_data_stripes(map);
	}
Y
Yan Zheng 已提交
5234 5235

	buf = kzalloc(sizeof(u64) * map->num_stripes, GFP_NOFS);
5236
	BUG_ON(!buf); /* -ENOMEM */
Y
Yan Zheng 已提交
5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252

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

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

		if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
			stripe_nr = stripe_nr * map->num_stripes + i;
			do_div(stripe_nr, map->sub_stripes);
		} else if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
			stripe_nr = stripe_nr * map->num_stripes + i;
D
David Woodhouse 已提交
5253 5254 5255 5256 5257
		} /* else if RAID[56], multiply by nr_data_stripes().
		   * Alternatively, just use rmap_len below instead of
		   * map->stripe_len */

		bytenr = chunk_start + stripe_nr * rmap_len;
5258
		WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
5259 5260 5261 5262
		for (j = 0; j < nr; j++) {
			if (buf[j] == bytenr)
				break;
		}
5263 5264
		if (j == nr) {
			WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
5265
			buf[nr++] = bytenr;
5266
		}
Y
Yan Zheng 已提交
5267 5268 5269 5270
	}

	*logical = buf;
	*naddrs = nr;
D
David Woodhouse 已提交
5271
	*stripe_len = rmap_len;
Y
Yan Zheng 已提交
5272 5273 5274

	free_extent_map(em);
	return 0;
5275 5276
}

5277
static void btrfs_end_bio(struct bio *bio, int err)
5278
{
5279
	struct btrfs_bio *bbio = bio->bi_private;
5280
	struct btrfs_device *dev = bbio->stripes[0].dev;
5281
	int is_orig_bio = 0;
5282

5283
	if (err) {
5284
		atomic_inc(&bbio->error);
5285 5286
		if (err == -EIO || err == -EREMOTEIO) {
			unsigned int stripe_index =
5287
				btrfs_io_bio(bio)->stripe_index;
5288 5289 5290

			BUG_ON(stripe_index >= bbio->num_stripes);
			dev = bbio->stripes[stripe_index].dev;
5291 5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302
			if (dev->bdev) {
				if (bio->bi_rw & WRITE)
					btrfs_dev_stat_inc(dev,
						BTRFS_DEV_STAT_WRITE_ERRS);
				else
					btrfs_dev_stat_inc(dev,
						BTRFS_DEV_STAT_READ_ERRS);
				if ((bio->bi_rw & WRITE_FLUSH) == WRITE_FLUSH)
					btrfs_dev_stat_inc(dev,
						BTRFS_DEV_STAT_FLUSH_ERRS);
				btrfs_dev_stat_print_on_error(dev);
			}
5303 5304
		}
	}
5305

5306
	if (bio == bbio->orig_bio)
5307 5308
		is_orig_bio = 1;

5309 5310
	btrfs_bio_counter_dec(bbio->fs_info);

5311
	if (atomic_dec_and_test(&bbio->stripes_pending)) {
5312 5313
		if (!is_orig_bio) {
			bio_put(bio);
5314
			bio = bbio->orig_bio;
5315
		}
5316 5317
		bio->bi_private = bbio->private;
		bio->bi_end_io = bbio->end_io;
5318
		btrfs_io_bio(bio)->mirror_num = bbio->mirror_num;
5319
		/* only send an error to the higher layers if it is
D
David Woodhouse 已提交
5320
		 * beyond the tolerance of the btrfs bio
5321
		 */
5322
		if (atomic_read(&bbio->error) > bbio->max_errors) {
5323
			err = -EIO;
5324
		} else {
5325 5326 5327 5328 5329
			/*
			 * this bio is actually up to date, we didn't
			 * go over the max number of errors
			 */
			set_bit(BIO_UPTODATE, &bio->bi_flags);
5330
			err = 0;
5331
		}
5332
		kfree(bbio);
5333 5334

		bio_endio(bio, err);
5335
	} else if (!is_orig_bio) {
5336 5337 5338 5339
		bio_put(bio);
	}
}

5340 5341 5342 5343 5344 5345 5346
/*
 * 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.
 */
5347 5348 5349
static noinline void btrfs_schedule_bio(struct btrfs_root *root,
					struct btrfs_device *device,
					int rw, struct bio *bio)
5350 5351
{
	int should_queue = 1;
5352
	struct btrfs_pending_bios *pending_bios;
5353

D
David Woodhouse 已提交
5354 5355 5356 5357 5358
	if (device->missing || !device->bdev) {
		bio_endio(bio, -EIO);
		return;
	}

5359
	/* don't bother with additional async steps for reads, right now */
5360
	if (!(rw & REQ_WRITE)) {
5361
		bio_get(bio);
5362
		btrfsic_submit_bio(rw, bio);
5363
		bio_put(bio);
5364
		return;
5365 5366 5367
	}

	/*
5368
	 * nr_async_bios allows us to reliably return congestion to the
5369 5370 5371 5372
	 * 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
	 */
5373
	atomic_inc(&root->fs_info->nr_async_bios);
5374
	WARN_ON(bio->bi_next);
5375 5376 5377 5378
	bio->bi_next = NULL;
	bio->bi_rw |= rw;

	spin_lock(&device->io_lock);
5379
	if (bio->bi_rw & REQ_SYNC)
5380 5381 5382
		pending_bios = &device->pending_sync_bios;
	else
		pending_bios = &device->pending_bios;
5383

5384 5385
	if (pending_bios->tail)
		pending_bios->tail->bi_next = bio;
5386

5387 5388 5389
	pending_bios->tail = bio;
	if (!pending_bios->head)
		pending_bios->head = bio;
5390 5391 5392 5393 5394 5395
	if (device->running_pending)
		should_queue = 0;

	spin_unlock(&device->io_lock);

	if (should_queue)
5396 5397
		btrfs_queue_work(root->fs_info->submit_workers,
				 &device->work);
5398 5399
}

5400 5401 5402 5403 5404
static int bio_size_ok(struct block_device *bdev, struct bio *bio,
		       sector_t sector)
{
	struct bio_vec *prev;
	struct request_queue *q = bdev_get_queue(bdev);
5405
	unsigned int max_sectors = queue_max_sectors(q);
5406 5407 5408 5409 5410 5411
	struct bvec_merge_data bvm = {
		.bi_bdev = bdev,
		.bi_sector = sector,
		.bi_rw = bio->bi_rw,
	};

5412
	if (WARN_ON(bio->bi_vcnt == 0))
5413 5414 5415
		return 1;

	prev = &bio->bi_io_vec[bio->bi_vcnt - 1];
5416
	if (bio_sectors(bio) > max_sectors)
5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434
		return 0;

	if (!q->merge_bvec_fn)
		return 1;

	bvm.bi_size = bio->bi_size - prev->bv_len;
	if (q->merge_bvec_fn(q, &bvm, prev) < prev->bv_len)
		return 0;
	return 1;
}

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

	bio->bi_private = bbio;
5435
	btrfs_io_bio(bio)->stripe_index = dev_nr;
5436 5437 5438 5439 5440 5441 5442 5443
	bio->bi_end_io = btrfs_end_bio;
	bio->bi_sector = physical >> 9;
#ifdef DEBUG
	{
		struct rcu_string *name;

		rcu_read_lock();
		name = rcu_dereference(dev->name);
5444
		pr_debug("btrfs_map_bio: rw %d, sector=%llu, dev=%lu "
5445 5446 5447 5448 5449 5450 5451
			 "(%s id %llu), size=%u\n", rw,
			 (u64)bio->bi_sector, (u_long)dev->bdev->bd_dev,
			 name->str, dev->devid, bio->bi_size);
		rcu_read_unlock();
	}
#endif
	bio->bi_bdev = dev->bdev;
5452 5453 5454

	btrfs_bio_counter_inc_noblocked(root->fs_info);

5455
	if (async)
D
David Woodhouse 已提交
5456
		btrfs_schedule_bio(root, dev, rw, bio);
5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497 5498
	else
		btrfsic_submit_bio(rw, bio);
}

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

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

	while (bvec <= (first_bio->bi_io_vec + first_bio->bi_vcnt - 1)) {
		if (bio_add_page(bio, bvec->bv_page, bvec->bv_len,
				 bvec->bv_offset) < bvec->bv_len) {
			u64 len = bio->bi_size;

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

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

static void bbio_error(struct btrfs_bio *bbio, struct bio *bio, u64 logical)
{
	atomic_inc(&bbio->error);
	if (atomic_dec_and_test(&bbio->stripes_pending)) {
		bio->bi_private = bbio->private;
		bio->bi_end_io = bbio->end_io;
5499
		btrfs_io_bio(bio)->mirror_num = bbio->mirror_num;
5500 5501 5502 5503 5504 5505
		bio->bi_sector = logical >> 9;
		kfree(bbio);
		bio_endio(bio, -EIO);
	}
}

5506
int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
5507
		  int mirror_num, int async_submit)
5508 5509
{
	struct btrfs_device *dev;
5510
	struct bio *first_bio = bio;
5511
	u64 logical = (u64)bio->bi_sector << 9;
5512 5513
	u64 length = 0;
	u64 map_length;
D
David Woodhouse 已提交
5514
	u64 *raid_map = NULL;
5515
	int ret;
5516 5517
	int dev_nr = 0;
	int total_devs = 1;
5518
	struct btrfs_bio *bbio = NULL;
5519

5520
	length = bio->bi_size;
5521
	map_length = length;
5522

5523
	btrfs_bio_counter_inc_blocked(root->fs_info);
D
David Woodhouse 已提交
5524 5525
	ret = __btrfs_map_block(root->fs_info, rw, logical, &map_length, &bbio,
			      mirror_num, &raid_map);
5526 5527
	if (ret) {
		btrfs_bio_counter_dec(root->fs_info);
5528
		return ret;
5529
	}
5530

5531
	total_devs = bbio->num_stripes;
D
David Woodhouse 已提交
5532 5533 5534
	bbio->orig_bio = first_bio;
	bbio->private = first_bio->bi_private;
	bbio->end_io = first_bio->bi_end_io;
5535
	bbio->fs_info = root->fs_info;
D
David Woodhouse 已提交
5536 5537 5538 5539 5540 5541
	atomic_set(&bbio->stripes_pending, bbio->num_stripes);

	if (raid_map) {
		/* In this case, map_length has been set to the length of
		   a single stripe; not the whole write */
		if (rw & WRITE) {
5542 5543
			ret = raid56_parity_write(root, bio, bbio,
						  raid_map, map_length);
D
David Woodhouse 已提交
5544
		} else {
5545 5546 5547
			ret = raid56_parity_recover(root, bio, bbio,
						    raid_map, map_length,
						    mirror_num);
D
David Woodhouse 已提交
5548
		}
5549 5550 5551 5552 5553 5554
		/*
		 * FIXME, replace dosen't support raid56 yet, please fix
		 * it in the future.
		 */
		btrfs_bio_counter_dec(root->fs_info);
		return ret;
D
David Woodhouse 已提交
5555 5556
	}

5557
	if (map_length < length) {
5558
		btrfs_crit(root->fs_info, "mapping failed logical %llu bio len %llu len %llu",
5559
			logical, length, map_length);
5560 5561
		BUG();
	}
5562

5563
	while (dev_nr < total_devs) {
5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583
		dev = bbio->stripes[dev_nr].dev;
		if (!dev || !dev->bdev || (rw & WRITE && !dev->writeable)) {
			bbio_error(bbio, first_bio, logical);
			dev_nr++;
			continue;
		}

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

5584
		if (dev_nr < total_devs - 1) {
5585
			bio = btrfs_bio_clone(first_bio, GFP_NOFS);
5586
			BUG_ON(!bio); /* -ENOMEM */
5587 5588
		} else {
			bio = first_bio;
5589
		}
5590 5591 5592 5593

		submit_stripe_bio(root, bbio, bio,
				  bbio->stripes[dev_nr].physical, dev_nr, rw,
				  async_submit);
5594 5595
		dev_nr++;
	}
5596
	btrfs_bio_counter_dec(root->fs_info);
5597 5598 5599
	return 0;
}

5600
struct btrfs_device *btrfs_find_device(struct btrfs_fs_info *fs_info, u64 devid,
Y
Yan Zheng 已提交
5601
				       u8 *uuid, u8 *fsid)
5602
{
Y
Yan Zheng 已提交
5603 5604 5605
	struct btrfs_device *device;
	struct btrfs_fs_devices *cur_devices;

5606
	cur_devices = fs_info->fs_devices;
Y
Yan Zheng 已提交
5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617
	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;
5618 5619
}

5620 5621 5622 5623 5624 5625
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;

5626 5627
	device = btrfs_alloc_device(NULL, &devid, dev_uuid);
	if (IS_ERR(device))
5628
		return NULL;
5629 5630

	list_add(&device->dev_list, &fs_devices->devices);
Y
Yan Zheng 已提交
5631
	device->fs_devices = fs_devices;
5632
	fs_devices->num_devices++;
5633 5634

	device->missing = 1;
5635
	fs_devices->missing_devices++;
5636

5637 5638 5639
	return device;
}

5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656 5657 5658 5659
/**
 * btrfs_alloc_device - allocate struct btrfs_device
 * @fs_info:	used only for generating a new devid, can be NULL if
 *		devid is provided (i.e. @devid != NULL).
 * @devid:	a pointer to devid for this device.  If NULL a new devid
 *		is generated.
 * @uuid:	a pointer to UUID for this device.  If NULL a new UUID
 *		is generated.
 *
 * Return: a pointer to a new &struct btrfs_device on success; ERR_PTR()
 * on error.  Returned struct is not linked onto any lists and can be
 * destroyed with kfree() right away.
 */
struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info,
					const u64 *devid,
					const u8 *uuid)
{
	struct btrfs_device *dev;
	u64 tmp;

5660
	if (WARN_ON(!devid && !fs_info))
5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684
		return ERR_PTR(-EINVAL);

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

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

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

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

5685
	btrfs_init_work(&dev->work, pending_bios_fn, NULL, NULL);
5686 5687 5688 5689

	return dev;
}

5690 5691 5692 5693 5694 5695 5696 5697 5698 5699
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;
5700
	u8 uuid[BTRFS_UUID_SIZE];
5701
	int num_stripes;
5702
	int ret;
5703
	int i;
5704

5705 5706
	logical = key->offset;
	length = btrfs_chunk_length(leaf, chunk);
5707

5708
	read_lock(&map_tree->map_tree.lock);
5709
	em = lookup_extent_mapping(&map_tree->map_tree, logical, 1);
5710
	read_unlock(&map_tree->map_tree.lock);
5711 5712 5713 5714 5715 5716 5717 5718 5719

	/* 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);
	}

5720
	em = alloc_extent_map();
5721 5722
	if (!em)
		return -ENOMEM;
5723 5724
	num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
5725 5726 5727 5728 5729 5730 5731 5732
	if (!map) {
		free_extent_map(em);
		return -ENOMEM;
	}

	em->bdev = (struct block_device *)map;
	em->start = logical;
	em->len = length;
5733
	em->orig_start = 0;
5734
	em->block_start = 0;
5735
	em->block_len = em->len;
5736

5737 5738 5739 5740 5741 5742
	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
5743
	map->sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
5744 5745 5746 5747
	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);
5748 5749 5750
		read_extent_buffer(leaf, uuid, (unsigned long)
				   btrfs_stripe_dev_uuid_nr(chunk, i),
				   BTRFS_UUID_SIZE);
5751 5752
		map->stripes[i].dev = btrfs_find_device(root->fs_info, devid,
							uuid, NULL);
5753
		if (!map->stripes[i].dev && !btrfs_test_opt(root, DEGRADED)) {
5754 5755 5756 5757
			kfree(map);
			free_extent_map(em);
			return -EIO;
		}
5758 5759 5760 5761 5762 5763 5764 5765 5766 5767
		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;
5768 5769
	}

5770
	write_lock(&map_tree->map_tree.lock);
5771
	ret = add_extent_mapping(&map_tree->map_tree, em, 0);
5772
	write_unlock(&map_tree->map_tree.lock);
5773
	BUG_ON(ret); /* Tree corruption */
5774 5775 5776 5777 5778
	free_extent_map(em);

	return 0;
}

5779
static void fill_device_from_item(struct extent_buffer *leaf,
5780 5781 5782 5783 5784 5785
				 struct btrfs_dev_item *dev_item,
				 struct btrfs_device *device)
{
	unsigned long ptr;

	device->devid = btrfs_device_id(leaf, dev_item);
5786 5787
	device->disk_total_bytes = btrfs_device_total_bytes(leaf, dev_item);
	device->total_bytes = device->disk_total_bytes;
5788 5789 5790 5791 5792
	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);
5793
	WARN_ON(device->devid == BTRFS_DEV_REPLACE_DEVID);
5794
	device->is_tgtdev_for_dev_replace = 0;
5795

5796
	ptr = btrfs_device_uuid(dev_item);
5797
	read_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
5798 5799
}

Y
Yan Zheng 已提交
5800 5801 5802 5803 5804
static int open_seed_devices(struct btrfs_root *root, u8 *fsid)
{
	struct btrfs_fs_devices *fs_devices;
	int ret;

5805
	BUG_ON(!mutex_is_locked(&uuid_mutex));
Y
Yan Zheng 已提交
5806 5807 5808 5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820

	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
5821 5822 5823 5824

	fs_devices = clone_fs_devices(fs_devices);
	if (IS_ERR(fs_devices)) {
		ret = PTR_ERR(fs_devices);
Y
Yan Zheng 已提交
5825 5826 5827
		goto out;
	}

5828
	ret = __btrfs_open_devices(fs_devices, FMODE_READ,
5829
				   root->fs_info->bdev_holder);
5830 5831
	if (ret) {
		free_fs_devices(fs_devices);
Y
Yan Zheng 已提交
5832
		goto out;
5833
	}
Y
Yan Zheng 已提交
5834 5835 5836

	if (!fs_devices->seeding) {
		__btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
5837
		free_fs_devices(fs_devices);
Y
Yan Zheng 已提交
5838 5839 5840 5841 5842 5843 5844 5845 5846 5847
		ret = -EINVAL;
		goto out;
	}

	fs_devices->seed = root->fs_info->fs_devices->seed;
	root->fs_info->fs_devices->seed = fs_devices;
out:
	return ret;
}

5848
static int read_one_dev(struct btrfs_root *root,
5849 5850 5851 5852 5853 5854
			struct extent_buffer *leaf,
			struct btrfs_dev_item *dev_item)
{
	struct btrfs_device *device;
	u64 devid;
	int ret;
Y
Yan Zheng 已提交
5855
	u8 fs_uuid[BTRFS_UUID_SIZE];
5856 5857
	u8 dev_uuid[BTRFS_UUID_SIZE];

5858
	devid = btrfs_device_id(leaf, dev_item);
5859
	read_extent_buffer(leaf, dev_uuid, btrfs_device_uuid(dev_item),
5860
			   BTRFS_UUID_SIZE);
5861
	read_extent_buffer(leaf, fs_uuid, btrfs_device_fsid(dev_item),
Y
Yan Zheng 已提交
5862 5863 5864 5865
			   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
5866
		if (ret && !btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
5867 5868 5869
			return ret;
	}

5870
	device = btrfs_find_device(root->fs_info, devid, dev_uuid, fs_uuid);
Y
Yan Zheng 已提交
5871
	if (!device || !device->bdev) {
Y
Yan Zheng 已提交
5872
		if (!btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
5873 5874 5875
			return -EIO;

		if (!device) {
5876
			btrfs_warn(root->fs_info, "devid %llu missing", devid);
Y
Yan Zheng 已提交
5877 5878 5879
			device = add_missing_dev(root, devid, dev_uuid);
			if (!device)
				return -ENOMEM;
5880 5881 5882 5883 5884 5885 5886 5887 5888
		} else if (!device->missing) {
			/*
			 * this happens when a device that was properly setup
			 * in the device info lists suddenly goes bad.
			 * device->bdev is NULL, and so we have to set
			 * device->missing to one here
			 */
			root->fs_info->fs_devices->missing_devices++;
			device->missing = 1;
Y
Yan Zheng 已提交
5889 5890 5891 5892 5893 5894 5895 5896
		}
	}

	if (device->fs_devices != root->fs_info->fs_devices) {
		BUG_ON(device->writeable);
		if (device->generation !=
		    btrfs_device_generation(leaf, dev_item))
			return -EINVAL;
5897
	}
5898 5899

	fill_device_from_item(leaf, dev_item, device);
5900
	device->in_fs_metadata = 1;
5901
	if (device->writeable && !device->is_tgtdev_for_dev_replace) {
Y
Yan Zheng 已提交
5902
		device->fs_devices->total_rw_bytes += device->total_bytes;
5903 5904 5905 5906 5907
		spin_lock(&root->fs_info->free_chunk_lock);
		root->fs_info->free_chunk_space += device->total_bytes -
			device->bytes_used;
		spin_unlock(&root->fs_info->free_chunk_lock);
	}
5908 5909 5910 5911
	ret = 0;
	return ret;
}

Y
Yan Zheng 已提交
5912
int btrfs_read_sys_array(struct btrfs_root *root)
5913
{
5914
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
5915
	struct extent_buffer *sb;
5916 5917
	struct btrfs_disk_key *disk_key;
	struct btrfs_chunk *chunk;
5918 5919 5920
	u8 *ptr;
	unsigned long sb_ptr;
	int ret = 0;
5921 5922 5923 5924
	u32 num_stripes;
	u32 array_size;
	u32 len = 0;
	u32 cur;
5925
	struct btrfs_key key;
5926

Y
Yan Zheng 已提交
5927
	sb = btrfs_find_create_tree_block(root, BTRFS_SUPER_INFO_OFFSET,
5928 5929 5930 5931
					  BTRFS_SUPER_INFO_SIZE);
	if (!sb)
		return -ENOMEM;
	btrfs_set_buffer_uptodate(sb);
5932
	btrfs_set_buffer_lockdep_class(root->root_key.objectid, sb, 0);
5933 5934 5935 5936 5937 5938 5939 5940 5941 5942 5943 5944 5945
	/*
	 * The sb extent buffer is artifical and just used to read the system array.
	 * btrfs_set_buffer_uptodate() call does not properly mark all it's
	 * pages up-to-date when the page is larger: extent does not cover the
	 * whole page and consequently check_page_uptodate does not find all
	 * the page's extents up-to-date (the hole beyond sb),
	 * write_extent_buffer then triggers a WARN_ON.
	 *
	 * Regular short extents go through mark_extent_buffer_dirty/writeback cycle,
	 * but sb spans only this function. Add an explicit SetPageUptodate call
	 * to silence the warning eg. on PowerPC 64.
	 */
	if (PAGE_CACHE_SIZE > BTRFS_SUPER_INFO_SIZE)
5946
		SetPageUptodate(sb->pages[0]);
5947

5948
	write_extent_buffer(sb, super_copy, 0, BTRFS_SUPER_INFO_SIZE);
5949 5950 5951 5952 5953 5954 5955 5956 5957 5958
	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);

5959
		len = sizeof(*disk_key); ptr += len;
5960 5961 5962
		sb_ptr += len;
		cur += len;

5963
		if (key.type == BTRFS_CHUNK_ITEM_KEY) {
5964
			chunk = (struct btrfs_chunk *)sb_ptr;
5965
			ret = read_one_chunk(root, &key, sb, chunk);
5966 5967
			if (ret)
				break;
5968 5969 5970
			num_stripes = btrfs_chunk_num_stripes(sb, chunk);
			len = btrfs_chunk_item_size(num_stripes);
		} else {
5971 5972
			ret = -EIO;
			break;
5973 5974 5975 5976 5977
		}
		ptr += len;
		sb_ptr += len;
		cur += len;
	}
5978
	free_extent_buffer(sb);
5979
	return ret;
5980 5981 5982 5983 5984 5985 5986 5987 5988 5989 5990 5991 5992 5993 5994 5995 5996
}

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;

5997 5998 5999
	mutex_lock(&uuid_mutex);
	lock_chunks(root);

6000 6001 6002 6003 6004
	/*
	 * Read all device items, and then all the chunk items. All
	 * device items are found before any chunk item (their object id
	 * is smaller than the lowest possible object id for a chunk
	 * item - BTRFS_FIRST_CHUNK_TREE_OBJECTID).
6005 6006 6007 6008 6009
	 */
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.offset = 0;
	key.type = 0;
	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
6010 6011
	if (ret < 0)
		goto error;
6012
	while (1) {
6013 6014 6015 6016 6017 6018 6019 6020 6021 6022 6023
		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);
6024 6025 6026
		if (found_key.type == BTRFS_DEV_ITEM_KEY) {
			struct btrfs_dev_item *dev_item;
			dev_item = btrfs_item_ptr(leaf, slot,
6027
						  struct btrfs_dev_item);
6028 6029 6030
			ret = read_one_dev(root, leaf, dev_item);
			if (ret)
				goto error;
6031 6032 6033 6034
		} 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
6035 6036
			if (ret)
				goto error;
6037 6038 6039 6040 6041
		}
		path->slots[0]++;
	}
	ret = 0;
error:
6042 6043 6044
	unlock_chunks(root);
	mutex_unlock(&uuid_mutex);

Y
Yan Zheng 已提交
6045
	btrfs_free_path(path);
6046 6047
	return ret;
}
6048

6049 6050 6051 6052 6053 6054 6055 6056 6057 6058 6059
void btrfs_init_devices_late(struct btrfs_fs_info *fs_info)
{
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
	struct btrfs_device *device;

	mutex_lock(&fs_devices->device_list_mutex);
	list_for_each_entry(device, &fs_devices->devices, dev_list)
		device->dev_root = fs_info->dev_root;
	mutex_unlock(&fs_devices->device_list_mutex);
}

6060 6061 6062 6063 6064 6065 6066 6067 6068 6069 6070 6071 6072 6073 6074 6075 6076 6077 6078 6079 6080 6081 6082 6083 6084 6085 6086 6087 6088 6089 6090 6091 6092 6093 6094 6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106 6107 6108 6109 6110 6111 6112 6113 6114 6115 6116 6117 6118 6119 6120 6121 6122 6123 6124 6125 6126 6127 6128 6129 6130 6131 6132 6133 6134 6135 6136 6137 6138 6139 6140 6141 6142 6143 6144 6145 6146 6147
static void __btrfs_reset_dev_stats(struct btrfs_device *dev)
{
	int i;

	for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++)
		btrfs_dev_stat_reset(dev, i);
}

int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info)
{
	struct btrfs_key key;
	struct btrfs_key found_key;
	struct btrfs_root *dev_root = fs_info->dev_root;
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
	struct extent_buffer *eb;
	int slot;
	int ret = 0;
	struct btrfs_device *device;
	struct btrfs_path *path = NULL;
	int i;

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

	mutex_lock(&fs_devices->device_list_mutex);
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
		int item_size;
		struct btrfs_dev_stats_item *ptr;

		key.objectid = 0;
		key.type = BTRFS_DEV_STATS_KEY;
		key.offset = device->devid;
		ret = btrfs_search_slot(NULL, dev_root, &key, path, 0, 0);
		if (ret) {
			__btrfs_reset_dev_stats(device);
			device->dev_stats_valid = 1;
			btrfs_release_path(path);
			continue;
		}
		slot = path->slots[0];
		eb = path->nodes[0];
		btrfs_item_key_to_cpu(eb, &found_key, slot);
		item_size = btrfs_item_size_nr(eb, slot);

		ptr = btrfs_item_ptr(eb, slot,
				     struct btrfs_dev_stats_item);

		for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) {
			if (item_size >= (1 + i) * sizeof(__le64))
				btrfs_dev_stat_set(device, i,
					btrfs_dev_stats_value(eb, ptr, i));
			else
				btrfs_dev_stat_reset(device, i);
		}

		device->dev_stats_valid = 1;
		btrfs_dev_stat_print_on_load(device);
		btrfs_release_path(path);
	}
	mutex_unlock(&fs_devices->device_list_mutex);

out:
	btrfs_free_path(path);
	return ret < 0 ? ret : 0;
}

static int update_dev_stat_item(struct btrfs_trans_handle *trans,
				struct btrfs_root *dev_root,
				struct btrfs_device *device)
{
	struct btrfs_path *path;
	struct btrfs_key key;
	struct extent_buffer *eb;
	struct btrfs_dev_stats_item *ptr;
	int ret;
	int i;

	key.objectid = 0;
	key.type = BTRFS_DEV_STATS_KEY;
	key.offset = device->devid;

	path = btrfs_alloc_path();
	BUG_ON(!path);
	ret = btrfs_search_slot(trans, dev_root, &key, path, -1, 1);
	if (ret < 0) {
6148 6149
		printk_in_rcu(KERN_WARNING "BTRFS: "
			"error %d while searching for dev_stats item for device %s!\n",
6150
			      ret, rcu_str_deref(device->name));
6151 6152 6153 6154 6155 6156 6157 6158
		goto out;
	}

	if (ret == 0 &&
	    btrfs_item_size_nr(path->nodes[0], path->slots[0]) < sizeof(*ptr)) {
		/* need to delete old one and insert a new one */
		ret = btrfs_del_item(trans, dev_root, path);
		if (ret != 0) {
6159 6160
			printk_in_rcu(KERN_WARNING "BTRFS: "
				"delete too small dev_stats item for device %s failed %d!\n",
6161
				      rcu_str_deref(device->name), ret);
6162 6163 6164 6165 6166 6167 6168 6169 6170 6171 6172
			goto out;
		}
		ret = 1;
	}

	if (ret == 1) {
		/* need to insert a new item */
		btrfs_release_path(path);
		ret = btrfs_insert_empty_item(trans, dev_root, path,
					      &key, sizeof(*ptr));
		if (ret < 0) {
6173 6174
			printk_in_rcu(KERN_WARNING "BTRFS: "
					  "insert dev_stats item for device %s failed %d!\n",
6175
				      rcu_str_deref(device->name), ret);
6176 6177 6178 6179 6180 6181 6182 6183 6184 6185 6186 6187 6188 6189 6190 6191 6192 6193 6194 6195 6196 6197 6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209 6210 6211 6212 6213 6214 6215 6216
			goto out;
		}
	}

	eb = path->nodes[0];
	ptr = btrfs_item_ptr(eb, path->slots[0], struct btrfs_dev_stats_item);
	for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++)
		btrfs_set_dev_stats_value(eb, ptr, i,
					  btrfs_dev_stat_read(device, i));
	btrfs_mark_buffer_dirty(eb);

out:
	btrfs_free_path(path);
	return ret;
}

/*
 * called from commit_transaction. Writes all changed device stats to disk.
 */
int btrfs_run_dev_stats(struct btrfs_trans_handle *trans,
			struct btrfs_fs_info *fs_info)
{
	struct btrfs_root *dev_root = fs_info->dev_root;
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
	struct btrfs_device *device;
	int ret = 0;

	mutex_lock(&fs_devices->device_list_mutex);
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
		if (!device->dev_stats_valid || !device->dev_stats_dirty)
			continue;

		ret = update_dev_stat_item(trans, dev_root, device);
		if (!ret)
			device->dev_stats_dirty = 0;
	}
	mutex_unlock(&fs_devices->device_list_mutex);

	return ret;
}

6217 6218 6219 6220 6221 6222
void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index)
{
	btrfs_dev_stat_inc(dev, index);
	btrfs_dev_stat_print_on_error(dev);
}

6223
static void btrfs_dev_stat_print_on_error(struct btrfs_device *dev)
6224
{
6225 6226
	if (!dev->dev_stats_valid)
		return;
6227 6228
	printk_ratelimited_in_rcu(KERN_ERR "BTRFS: "
			   "bdev %s errs: wr %u, rd %u, flush %u, corrupt %u, gen %u\n",
6229
			   rcu_str_deref(dev->name),
6230 6231 6232
			   btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_WRITE_ERRS),
			   btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_READ_ERRS),
			   btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_FLUSH_ERRS),
6233 6234
			   btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_CORRUPTION_ERRS),
			   btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_GENERATION_ERRS));
6235
}
6236

6237 6238
static void btrfs_dev_stat_print_on_load(struct btrfs_device *dev)
{
6239 6240 6241 6242 6243 6244 6245 6246
	int i;

	for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++)
		if (btrfs_dev_stat_read(dev, i) != 0)
			break;
	if (i == BTRFS_DEV_STAT_VALUES_MAX)
		return; /* all values == 0, suppress message */

6247 6248
	printk_in_rcu(KERN_INFO "BTRFS: "
		   "bdev %s errs: wr %u, rd %u, flush %u, corrupt %u, gen %u\n",
6249
	       rcu_str_deref(dev->name),
6250 6251 6252 6253 6254 6255 6256
	       btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_WRITE_ERRS),
	       btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_READ_ERRS),
	       btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_FLUSH_ERRS),
	       btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_CORRUPTION_ERRS),
	       btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_GENERATION_ERRS));
}

6257
int btrfs_get_dev_stats(struct btrfs_root *root,
6258
			struct btrfs_ioctl_get_dev_stats *stats)
6259 6260 6261 6262 6263 6264
{
	struct btrfs_device *dev;
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
	int i;

	mutex_lock(&fs_devices->device_list_mutex);
6265
	dev = btrfs_find_device(root->fs_info, stats->devid, NULL, NULL);
6266 6267 6268
	mutex_unlock(&fs_devices->device_list_mutex);

	if (!dev) {
6269
		btrfs_warn(root->fs_info, "get dev_stats failed, device not found");
6270
		return -ENODEV;
6271
	} else if (!dev->dev_stats_valid) {
6272
		btrfs_warn(root->fs_info, "get dev_stats failed, not yet valid");
6273
		return -ENODEV;
6274
	} else if (stats->flags & BTRFS_DEV_STATS_RESET) {
6275 6276 6277 6278 6279 6280 6281 6282 6283 6284 6285 6286 6287 6288 6289 6290
		for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) {
			if (stats->nr_items > i)
				stats->values[i] =
					btrfs_dev_stat_read_and_reset(dev, i);
			else
				btrfs_dev_stat_reset(dev, i);
		}
	} else {
		for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++)
			if (stats->nr_items > i)
				stats->values[i] = btrfs_dev_stat_read(dev, i);
	}
	if (stats->nr_items > BTRFS_DEV_STAT_VALUES_MAX)
		stats->nr_items = BTRFS_DEV_STAT_VALUES_MAX;
	return 0;
}
6291 6292 6293 6294 6295 6296 6297 6298 6299 6300 6301 6302 6303 6304 6305 6306 6307 6308

int btrfs_scratch_superblock(struct btrfs_device *device)
{
	struct buffer_head *bh;
	struct btrfs_super_block *disk_super;

	bh = btrfs_read_dev_super(device->bdev);
	if (!bh)
		return -EINVAL;
	disk_super = (struct btrfs_super_block *)bh->b_data;

	memset(&disk_super->magic, 0, sizeof(disk_super->magic));
	set_buffer_dirty(bh);
	sync_dirty_buffer(bh);
	brelse(bh);

	return 0;
}
反馈
建议
客服 返回
顶部