volumes.c 154.6 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */
#include <linux/sched.h>
#include <linux/bio.h>
20
#include <linux/slab.h>
21
#include <linux/buffer_head.h>
22
#include <linux/blkdev.h>
23
#include <linux/random.h>
24
#include <linux/iocontext.h>
25
#include <linux/capability.h>
26
#include <linux/ratelimit.h>
I
Ilya Dryomov 已提交
27
#include <linux/kthread.h>
D
David Woodhouse 已提交
28 29
#include <linux/raid/pq.h>
#include <asm/div64.h>
C
Chris Mason 已提交
30
#include "compat.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 49
static void __btrfs_reset_dev_stats(struct btrfs_device *dev);
static void btrfs_dev_stat_print_on_load(struct btrfs_device *device);
Y
Yan Zheng 已提交
50

51 52 53
static DEFINE_MUTEX(uuid_mutex);
static LIST_HEAD(fs_uuids);

54 55 56 57 58 59 60 61 62 63
static void lock_chunks(struct btrfs_root *root)
{
	mutex_lock(&root->fs_info->chunk_mutex);
}

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

Y
Yan Zheng 已提交
64 65 66 67 68 69 70 71
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);
72
		rcu_string_free(device->name);
Y
Yan Zheng 已提交
73 74 75 76 77
		kfree(device);
	}
	kfree(fs_devices);
}

78 79 80 81 82 83 84 85 86 87 88 89 90
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)
		pr_warn("Sending event '%d' to kobject: '%s' (%p): failed\n",
			action,
			kobject_name(&disk_to_dev(bdev->bd_disk)->kobj),
			&disk_to_dev(bdev->bd_disk)->kobj);
}

91
void btrfs_cleanup_fs_uuids(void)
92 93 94
{
	struct btrfs_fs_devices *fs_devices;

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

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

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

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

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

128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165
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);
		printk(KERN_INFO "btrfs: open %s failed\n", device_path);
		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;
}

166 167 168 169 170 171 172 173 174 175 176 177 178 179
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;
}

180 181 182 183 184 185 186 187 188 189 190
/*
 * 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.
 */
191
static noinline void run_scheduled_bios(struct btrfs_device *device)
192 193 194
{
	struct bio *pending;
	struct backing_dev_info *bdi;
195
	struct btrfs_fs_info *fs_info;
196
	struct btrfs_pending_bios *pending_bios;
197 198 199
	struct bio *tail;
	struct bio *cur;
	int again = 0;
200
	unsigned long num_run;
201
	unsigned long batch_run = 0;
202
	unsigned long limit;
203
	unsigned long last_waited = 0;
204
	int force_reg = 0;
M
Miao Xie 已提交
205
	int sync_pending = 0;
206 207 208 209 210 211 212 213 214
	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);
215

216
	bdi = blk_get_backing_dev_info(device->bdev);
217 218 219 220
	fs_info = device->dev_root->fs_info;
	limit = btrfs_async_submit_limit(fs_info);
	limit = limit * 2 / 3;

221 222 223
loop:
	spin_lock(&device->io_lock);

224
loop_lock:
225
	num_run = 0;
226

227 228 229 230 231
	/* 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
	 */
232
	if (!force_reg && device->pending_sync_bios.head) {
233
		pending_bios = &device->pending_sync_bios;
234 235
		force_reg = 1;
	} else {
236
		pending_bios = &device->pending_bios;
237 238
		force_reg = 0;
	}
239 240 241

	pending = pending_bios->head;
	tail = pending_bios->tail;
242 243 244 245 246 247 248 249 250 251
	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.
	 */
252 253
	if (device->pending_sync_bios.head == NULL &&
	    device->pending_bios.head == NULL) {
254 255
		again = 0;
		device->running_pending = 0;
256 257 258
	} else {
		again = 1;
		device->running_pending = 1;
259
	}
260 261 262 263

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

264 265
	spin_unlock(&device->io_lock);

C
Chris Mason 已提交
266
	while (pending) {
267 268

		rmb();
269 270 271 272 273 274 275 276
		/* 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)) {
277 278 279 280 281
			spin_lock(&device->io_lock);
			requeue_list(pending_bios, pending, tail);
			goto loop_lock;
		}

282 283 284
		cur = pending;
		pending = pending->bi_next;
		cur->bi_next = NULL;
285

286
		if (atomic_dec_return(&fs_info->nr_async_bios) < limit &&
287 288
		    waitqueue_active(&fs_info->async_submit_wait))
			wake_up(&fs_info->async_submit_wait);
289 290

		BUG_ON(atomic_read(&cur->bi_cnt) == 0);
291

292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307
		/*
		 * 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;
		}

308
		btrfsic_submit_bio(cur->bi_rw, cur);
309 310
		num_run++;
		batch_run++;
J
Jens Axboe 已提交
311
		if (need_resched())
312
			cond_resched();
313 314 315 316 317 318

		/*
		 * we made progress, there is more work to do and the bdi
		 * is now congested.  Back off and let other work structs
		 * run instead
		 */
C
Chris Mason 已提交
319
		if (pending && bdi_write_congested(bdi) && batch_run > 8 &&
320
		    fs_info->fs_devices->open_devices > 1) {
321
			struct io_context *ioc;
322

323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344
			ioc = current->io_context;

			/*
			 * the main goal here is that we don't want to
			 * block if we're going to be able to submit
			 * more requests without blocking.
			 *
			 * This code does two great things, it pokes into
			 * the elevator code from a filesystem _and_
			 * it makes assumptions about how batching works.
			 */
			if (ioc && ioc->nr_batch_requests > 0 &&
			    time_before(jiffies, ioc->last_waited + HZ/50UL) &&
			    (last_waited == 0 ||
			     ioc->last_waited == last_waited)) {
				/*
				 * we want to go through our batch of
				 * requests and stop.  So, we copy out
				 * the ioc->last_waited time and test
				 * against it before looping
				 */
				last_waited = ioc->last_waited;
J
Jens Axboe 已提交
345
				if (need_resched())
346
					cond_resched();
347 348
				continue;
			}
349
			spin_lock(&device->io_lock);
350
			requeue_list(pending_bios, pending, tail);
351
			device->running_pending = 1;
352 353 354 355 356

			spin_unlock(&device->io_lock);
			btrfs_requeue_work(&device->work);
			goto done;
		}
C
Chris Mason 已提交
357 358 359 360 361 362
		/* unplug every 64 requests just for good measure */
		if (batch_run % 64 == 0) {
			blk_finish_plug(&plug);
			blk_start_plug(&plug);
			sync_pending = 0;
		}
363
	}
364

365 366 367 368 369 370 371 372 373
	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);

374
done:
375
	blk_finish_plug(&plug);
376 377
}

378
static void pending_bios_fn(struct btrfs_work *work)
379 380 381 382 383 384 385
{
	struct btrfs_device *device;

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

386
static noinline int device_list_add(const char *path,
387 388 389 390 391
			   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;
392
	struct rcu_string *name;
393 394 395 396
	u64 found_transid = btrfs_super_generation(disk_super);

	fs_devices = find_fsid(disk_super->fsid);
	if (!fs_devices) {
397
		fs_devices = kzalloc(sizeof(*fs_devices), GFP_NOFS);
398 399 400
		if (!fs_devices)
			return -ENOMEM;
		INIT_LIST_HEAD(&fs_devices->devices);
401
		INIT_LIST_HEAD(&fs_devices->alloc_list);
402 403 404 405
		list_add(&fs_devices->list, &fs_uuids);
		memcpy(fs_devices->fsid, disk_super->fsid, BTRFS_FSID_SIZE);
		fs_devices->latest_devid = devid;
		fs_devices->latest_trans = found_transid;
406
		mutex_init(&fs_devices->device_list_mutex);
407 408
		device = NULL;
	} else {
409 410
		device = __find_device(&fs_devices->devices, devid,
				       disk_super->dev_item.uuid);
411 412
	}
	if (!device) {
Y
Yan Zheng 已提交
413 414 415
		if (fs_devices->opened)
			return -EBUSY;

416 417 418 419 420 421
		device = kzalloc(sizeof(*device), GFP_NOFS);
		if (!device) {
			/* we can safely leave the fs_devices entry around */
			return -ENOMEM;
		}
		device->devid = devid;
422
		device->dev_stats_valid = 0;
423
		device->work.func = pending_bios_fn;
424 425
		memcpy(device->uuid, disk_super->dev_item.uuid,
		       BTRFS_UUID_SIZE);
426
		spin_lock_init(&device->io_lock);
427 428 429

		name = rcu_string_strdup(path, GFP_NOFS);
		if (!name) {
430 431 432
			kfree(device);
			return -ENOMEM;
		}
433
		rcu_assign_pointer(device->name, name);
Y
Yan Zheng 已提交
434
		INIT_LIST_HEAD(&device->dev_alloc_list);
435

436 437 438 439 440 441 442 443
		/* init readahead state */
		spin_lock_init(&device->reada_lock);
		device->reada_curr_zone = NULL;
		atomic_set(&device->reada_in_flight, 0);
		device->reada_next = 0;
		INIT_RADIX_TREE(&device->reada_zones, GFP_NOFS & ~__GFP_WAIT);
		INIT_RADIX_TREE(&device->reada_extents, GFP_NOFS & ~__GFP_WAIT);

444
		mutex_lock(&fs_devices->device_list_mutex);
445
		list_add_rcu(&device->dev_list, &fs_devices->devices);
446 447
		mutex_unlock(&fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
448
		device->fs_devices = fs_devices;
449
		fs_devices->num_devices++;
450 451
	} else if (!device->name || strcmp(device->name->str, path)) {
		name = rcu_string_strdup(path, GFP_NOFS);
452 453
		if (!name)
			return -ENOMEM;
454 455
		rcu_string_free(device->name);
		rcu_assign_pointer(device->name, name);
456 457 458 459
		if (device->missing) {
			fs_devices->missing_devices--;
			device->missing = 0;
		}
460 461 462 463 464 465 466 467 468 469
	}

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

Y
Yan Zheng 已提交
470 471 472 473 474 475 476 477 478 479 480 481 482
static struct btrfs_fs_devices *clone_fs_devices(struct btrfs_fs_devices *orig)
{
	struct btrfs_fs_devices *fs_devices;
	struct btrfs_device *device;
	struct btrfs_device *orig_dev;

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

	INIT_LIST_HEAD(&fs_devices->devices);
	INIT_LIST_HEAD(&fs_devices->alloc_list);
	INIT_LIST_HEAD(&fs_devices->list);
483
	mutex_init(&fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
484 485
	fs_devices->latest_devid = orig->latest_devid;
	fs_devices->latest_trans = orig->latest_trans;
J
Josef Bacik 已提交
486
	fs_devices->total_devices = orig->total_devices;
Y
Yan Zheng 已提交
487 488
	memcpy(fs_devices->fsid, orig->fsid, sizeof(fs_devices->fsid));

489
	/* We have held the volume lock, it is safe to get the devices. */
Y
Yan Zheng 已提交
490
	list_for_each_entry(orig_dev, &orig->devices, dev_list) {
491 492
		struct rcu_string *name;

Y
Yan Zheng 已提交
493 494 495 496
		device = kzalloc(sizeof(*device), GFP_NOFS);
		if (!device)
			goto error;

497 498 499 500 501 502
		/*
		 * 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) {
J
Julia Lawall 已提交
503
			kfree(device);
Y
Yan Zheng 已提交
504
			goto error;
J
Julia Lawall 已提交
505
		}
506
		rcu_assign_pointer(device->name, name);
Y
Yan Zheng 已提交
507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524

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

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

525 526
void btrfs_close_extra_devices(struct btrfs_fs_info *fs_info,
			       struct btrfs_fs_devices *fs_devices, int step)
527
{
Q
Qinghuang Feng 已提交
528
	struct btrfs_device *device, *next;
529

530 531 532 533
	struct block_device *latest_bdev = NULL;
	u64 latest_devid = 0;
	u64 latest_transid = 0;

534 535
	mutex_lock(&uuid_mutex);
again:
536
	/* This is the initialized path, it is safe to release the devices. */
Q
Qinghuang Feng 已提交
537
	list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
538
		if (device->in_fs_metadata) {
539 540 541
			if (!device->is_tgtdev_for_dev_replace &&
			    (!latest_transid ||
			     device->generation > latest_transid)) {
542 543 544 545
				latest_devid = device->devid;
				latest_transid = device->generation;
				latest_bdev = device->bdev;
			}
Y
Yan Zheng 已提交
546
			continue;
547
		}
Y
Yan Zheng 已提交
548

549 550 551 552 553 554 555 556 557 558 559 560 561 562 563
		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
564
		if (device->bdev) {
565
			blkdev_put(device->bdev, device->mode);
Y
Yan Zheng 已提交
566 567 568 569 570 571
			device->bdev = NULL;
			fs_devices->open_devices--;
		}
		if (device->writeable) {
			list_del_init(&device->dev_alloc_list);
			device->writeable = 0;
572 573
			if (!device->is_tgtdev_for_dev_replace)
				fs_devices->rw_devices--;
Y
Yan Zheng 已提交
574
		}
Y
Yan Zheng 已提交
575 576
		list_del_init(&device->dev_list);
		fs_devices->num_devices--;
577
		rcu_string_free(device->name);
Y
Yan Zheng 已提交
578
		kfree(device);
579
	}
Y
Yan Zheng 已提交
580 581 582 583 584 585

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

586 587 588 589
	fs_devices->latest_bdev = latest_bdev;
	fs_devices->latest_devid = latest_devid;
	fs_devices->latest_trans = latest_transid;

590 591
	mutex_unlock(&uuid_mutex);
}
592

593 594 595 596 597 598 599 600 601
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);

602
	rcu_string_free(device->name);
603 604 605 606 607 608 609 610 611 612 613 614 615
	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
616
static int __btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
617 618
{
	struct btrfs_device *device;
Y
Yan Zheng 已提交
619

Y
Yan Zheng 已提交
620 621
	if (--fs_devices->opened > 0)
		return 0;
622

623
	mutex_lock(&fs_devices->device_list_mutex);
Q
Qinghuang Feng 已提交
624
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
625
		struct btrfs_device *new_device;
626
		struct rcu_string *name;
627 628

		if (device->bdev)
629
			fs_devices->open_devices--;
630

631
		if (device->writeable && !device->is_tgtdev_for_dev_replace) {
Y
Yan Zheng 已提交
632 633 634 635
			list_del_init(&device->dev_alloc_list);
			fs_devices->rw_devices--;
		}

636 637 638
		if (device->can_discard)
			fs_devices->num_can_discard--;

639
		new_device = kmalloc(sizeof(*new_device), GFP_NOFS);
640
		BUG_ON(!new_device); /* -ENOMEM */
641
		memcpy(new_device, device, sizeof(*new_device));
642 643

		/* Safe because we are under uuid_mutex */
644 645 646 647 648
		if (device->name) {
			name = rcu_string_strdup(device->name->str, GFP_NOFS);
			BUG_ON(device->name && !name); /* -ENOMEM */
			rcu_assign_pointer(new_device->name, name);
		}
649 650 651
		new_device->bdev = NULL;
		new_device->writeable = 0;
		new_device->in_fs_metadata = 0;
652
		new_device->can_discard = 0;
653
		spin_lock_init(&new_device->io_lock);
654 655 656
		list_replace_rcu(&device->dev_list, &new_device->dev_list);

		call_rcu(&device->rcu, free_device);
657
	}
658 659
	mutex_unlock(&fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
660 661
	WARN_ON(fs_devices->open_devices);
	WARN_ON(fs_devices->rw_devices);
Y
Yan Zheng 已提交
662 663 664
	fs_devices->opened = 0;
	fs_devices->seeding = 0;

665 666 667
	return 0;
}

Y
Yan Zheng 已提交
668 669
int btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
{
Y
Yan Zheng 已提交
670
	struct btrfs_fs_devices *seed_devices = NULL;
Y
Yan Zheng 已提交
671 672 673 674
	int ret;

	mutex_lock(&uuid_mutex);
	ret = __btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
675 676 677 678
	if (!fs_devices->opened) {
		seed_devices = fs_devices->seed;
		fs_devices->seed = NULL;
	}
Y
Yan Zheng 已提交
679
	mutex_unlock(&uuid_mutex);
Y
Yan Zheng 已提交
680 681 682 683 684 685 686

	while (seed_devices) {
		fs_devices = seed_devices;
		seed_devices = fs_devices->seed;
		__btrfs_close_devices(fs_devices);
		free_fs_devices(fs_devices);
	}
687 688 689 690 691 692
	/*
	 * 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
693 694 695
	return ret;
}

Y
Yan Zheng 已提交
696 697
static int __btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
				fmode_t flags, void *holder)
698
{
699
	struct request_queue *q;
700 701 702
	struct block_device *bdev;
	struct list_head *head = &fs_devices->devices;
	struct btrfs_device *device;
703 704 705 706 707 708
	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
709
	int seeding = 1;
710
	int ret = 0;
711

712 713
	flags |= FMODE_EXCL;

Q
Qinghuang Feng 已提交
714
	list_for_each_entry(device, head, dev_list) {
715 716
		if (device->bdev)
			continue;
717 718 719
		if (!device->name)
			continue;

720 721 722 723
		ret = btrfs_get_bdev_and_sb(device->name->str, flags, holder, 1,
					    &bdev, &bh);
		if (ret)
			continue;
724 725

		disk_super = (struct btrfs_super_block *)bh->b_data;
726
		devid = btrfs_stack_device_id(&disk_super->dev_item);
727 728 729
		if (devid != device->devid)
			goto error_brelse;

Y
Yan Zheng 已提交
730 731 732 733 734 735
		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) {
736
			latest_devid = devid;
Y
Yan Zheng 已提交
737
			latest_transid = device->generation;
738 739 740
			latest_bdev = bdev;
		}

Y
Yan Zheng 已提交
741 742 743 744 745 746 747
		if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_SEEDING) {
			device->writeable = 0;
		} else {
			device->writeable = !bdev_read_only(bdev);
			seeding = 0;
		}

748 749 750 751 752 753
		q = bdev_get_queue(bdev);
		if (blk_queue_discard(q)) {
			device->can_discard = 1;
			fs_devices->num_can_discard++;
		}

754
		device->bdev = bdev;
755
		device->in_fs_metadata = 0;
756 757
		device->mode = flags;

C
Chris Mason 已提交
758 759 760
		if (!blk_queue_nonrot(bdev_get_queue(bdev)))
			fs_devices->rotating = 1;

761
		fs_devices->open_devices++;
762
		if (device->writeable && !device->is_tgtdev_for_dev_replace) {
Y
Yan Zheng 已提交
763 764 765 766
			fs_devices->rw_devices++;
			list_add(&device->dev_alloc_list,
				 &fs_devices->alloc_list);
		}
767
		brelse(bh);
768
		continue;
769

770 771
error_brelse:
		brelse(bh);
772
		blkdev_put(bdev, flags);
773
		continue;
774
	}
775
	if (fs_devices->open_devices == 0) {
776
		ret = -EINVAL;
777 778
		goto out;
	}
Y
Yan Zheng 已提交
779 780
	fs_devices->seeding = seeding;
	fs_devices->opened = 1;
781 782 783
	fs_devices->latest_bdev = latest_bdev;
	fs_devices->latest_devid = latest_devid;
	fs_devices->latest_trans = latest_transid;
Y
Yan Zheng 已提交
784
	fs_devices->total_rw_bytes = 0;
785
out:
Y
Yan Zheng 已提交
786 787 788 789
	return ret;
}

int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
790
		       fmode_t flags, void *holder)
Y
Yan Zheng 已提交
791 792 793 794 795
{
	int ret;

	mutex_lock(&uuid_mutex);
	if (fs_devices->opened) {
Y
Yan Zheng 已提交
796 797
		fs_devices->opened++;
		ret = 0;
Y
Yan Zheng 已提交
798
	} else {
799
		ret = __btrfs_open_devices(fs_devices, flags, holder);
Y
Yan Zheng 已提交
800
	}
801 802 803 804
	mutex_unlock(&uuid_mutex);
	return ret;
}

805 806 807 808 809
/*
 * 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
 */
810
int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
811 812 813 814
			  struct btrfs_fs_devices **fs_devices_ret)
{
	struct btrfs_super_block *disk_super;
	struct block_device *bdev;
815 816 817
	struct page *page;
	void *p;
	int ret = -EINVAL;
818
	u64 devid;
819
	u64 transid;
J
Josef Bacik 已提交
820
	u64 total_devices;
821 822
	u64 bytenr;
	pgoff_t index;
823

824 825 826 827 828 829 830
	/*
	 * 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);
831
	flags |= FMODE_EXCL;
832
	mutex_lock(&uuid_mutex);
833 834 835 836 837

	bdev = blkdev_get_by_path(path, flags, holder);

	if (IS_ERR(bdev)) {
		ret = PTR_ERR(bdev);
838
		goto error;
839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866
	}

	/* 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 ||
867
	    disk_super->magic != cpu_to_le64(BTRFS_MAGIC))
868 869
		goto error_unmap;

870
	devid = btrfs_stack_device_id(&disk_super->dev_item);
871
	transid = btrfs_super_generation(disk_super);
J
Josef Bacik 已提交
872
	total_devices = btrfs_super_num_devices(disk_super);
873

874 875 876
	if (disk_super->label[0]) {
		if (disk_super->label[BTRFS_LABEL_SIZE - 1])
			disk_super->label[BTRFS_LABEL_SIZE - 1] = '\0';
C
Chris Mason 已提交
877
		printk(KERN_INFO "device label %s ", disk_super->label);
878
	} else {
I
Ilya Dryomov 已提交
879
		printk(KERN_INFO "device fsid %pU ", disk_super->fsid);
880
	}
881

882
	printk(KERN_CONT "devid %llu transid %llu %s\n",
C
Chris Mason 已提交
883
	       (unsigned long long)devid, (unsigned long long)transid, path);
884

885
	ret = device_list_add(path, disk_super, devid, fs_devices_ret);
J
Josef Bacik 已提交
886 887
	if (!ret && fs_devices_ret)
		(*fs_devices_ret)->total_devices = total_devices;
888 889 890 891 892 893

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

error_bdev_put:
894
	blkdev_put(bdev, flags);
895
error:
896
	mutex_unlock(&uuid_mutex);
897 898
	return ret;
}
899

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

915
	if (start >= device->total_bytes || device->is_tgtdev_for_dev_replace)
916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983
		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;
}

984
/*
985 986 987 988 989 990 991
 * 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
 *
992 993 994
 * 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
995 996 997 998 999 1000 1001 1002
 *
 * @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.
1003
 */
1004
int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes,
1005
			 u64 *start, u64 *len)
1006 1007 1008
{
	struct btrfs_key key;
	struct btrfs_root *root = device->dev_root;
1009
	struct btrfs_dev_extent *dev_extent;
Y
Yan Zheng 已提交
1010
	struct btrfs_path *path;
1011 1012 1013 1014 1015
	u64 hole_size;
	u64 max_hole_start;
	u64 max_hole_size;
	u64 extent_end;
	u64 search_start;
1016 1017
	u64 search_end = device->total_bytes;
	int ret;
1018
	int slot;
1019 1020 1021 1022
	struct extent_buffer *l;

	/* FIXME use last free of some kind */

1023 1024 1025
	/* 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
1026
	search_start = max(root->fs_info->alloc_start, 1024ull * 1024);
1027

1028 1029
	max_hole_start = search_start;
	max_hole_size = 0;
1030
	hole_size = 0;
1031

1032
	if (search_start >= search_end || device->is_tgtdev_for_dev_replace) {
1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043
		ret = -ENOSPC;
		goto error;
	}

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

1044 1045 1046
	key.objectid = device->devid;
	key.offset = search_start;
	key.type = BTRFS_DEV_EXTENT_KEY;
1047

1048
	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1049
	if (ret < 0)
1050
		goto out;
1051 1052 1053
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid, key.type);
		if (ret < 0)
1054
			goto out;
1055
	}
1056

1057 1058 1059 1060 1061 1062 1063 1064
	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)
1065 1066 1067
				goto out;

			break;
1068 1069 1070 1071 1072 1073 1074
		}
		btrfs_item_key_to_cpu(l, &key, slot);

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

		if (key.objectid > device->devid)
1075
			break;
1076

1077 1078
		if (btrfs_key_type(&key) != BTRFS_DEV_EXTENT_KEY)
			goto next;
1079

1080 1081
		if (key.offset > search_start) {
			hole_size = key.offset - search_start;
1082

1083 1084 1085 1086
			if (hole_size > max_hole_size) {
				max_hole_start = search_start;
				max_hole_size = hole_size;
			}
1087

1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
			/*
			 * 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;
1100 1101 1102 1103
			}
		}

		dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
1104 1105 1106 1107
		extent_end = key.offset + btrfs_dev_extent_length(l,
								  dev_extent);
		if (extent_end > search_start)
			search_start = extent_end;
1108 1109 1110 1111 1112
next:
		path->slots[0]++;
		cond_resched();
	}

1113 1114 1115 1116 1117 1118 1119 1120
	/*
	 * 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;

1121 1122 1123
	if (hole_size > max_hole_size) {
		max_hole_start = search_start;
		max_hole_size = hole_size;
1124 1125
	}

1126 1127 1128 1129 1130 1131 1132
	/* See above. */
	if (hole_size < num_bytes)
		ret = -ENOSPC;
	else
		ret = 0;

out:
Y
Yan Zheng 已提交
1133
	btrfs_free_path(path);
1134 1135
error:
	*start = max_hole_start;
1136
	if (len)
1137
		*len = max_hole_size;
1138 1139 1140
	return ret;
}

1141
static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans,
1142 1143 1144 1145 1146 1147 1148
			  struct btrfs_device *device,
			  u64 start)
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_key key;
1149 1150 1151
	struct btrfs_key found_key;
	struct extent_buffer *leaf = NULL;
	struct btrfs_dev_extent *extent = NULL;
1152 1153 1154 1155 1156 1157 1158 1159

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

	key.objectid = device->devid;
	key.offset = start;
	key.type = BTRFS_DEV_EXTENT_KEY;
M
Miao Xie 已提交
1160
again:
1161
	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
1162 1163 1164
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid,
					  BTRFS_DEV_EXTENT_KEY);
1165 1166
		if (ret)
			goto out;
1167 1168 1169 1170 1171 1172
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
		extent = btrfs_item_ptr(leaf, path->slots[0],
					struct btrfs_dev_extent);
		BUG_ON(found_key.offset > start || found_key.offset +
		       btrfs_dev_extent_length(leaf, extent) < start);
M
Miao Xie 已提交
1173 1174 1175
		key = found_key;
		btrfs_release_path(path);
		goto again;
1176 1177 1178 1179
	} else if (ret == 0) {
		leaf = path->nodes[0];
		extent = btrfs_item_ptr(leaf, path->slots[0],
					struct btrfs_dev_extent);
1180 1181 1182
	} else {
		btrfs_error(root->fs_info, ret, "Slot search failed");
		goto out;
1183
	}
1184

1185 1186 1187 1188 1189 1190 1191
	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);
	}
1192
	ret = btrfs_del_item(trans, root, path);
1193 1194 1195 1196
	if (ret) {
		btrfs_error(root->fs_info, ret,
			    "Failed to remove dev extent item");
	}
1197
out:
1198 1199 1200 1201
	btrfs_free_path(path);
	return ret;
}

Y
Yan Zheng 已提交
1202
int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
1203
			   struct btrfs_device *device,
1204
			   u64 chunk_tree, u64 chunk_objectid,
Y
Yan Zheng 已提交
1205
			   u64 chunk_offset, u64 start, u64 num_bytes)
1206 1207 1208 1209 1210 1211 1212 1213
{
	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;

1214
	WARN_ON(!device->in_fs_metadata);
1215
	WARN_ON(device->is_tgtdev_for_dev_replace);
1216 1217 1218 1219 1220
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = device->devid;
Y
Yan Zheng 已提交
1221
	key.offset = start;
1222 1223 1224
	key.type = BTRFS_DEV_EXTENT_KEY;
	ret = btrfs_insert_empty_item(trans, root, path, &key,
				      sizeof(*extent));
1225 1226
	if (ret)
		goto out;
1227 1228 1229 1230

	leaf = path->nodes[0];
	extent = btrfs_item_ptr(leaf, path->slots[0],
				struct btrfs_dev_extent);
1231 1232 1233 1234 1235 1236 1237 1238
	btrfs_set_dev_extent_chunk_tree(leaf, extent, chunk_tree);
	btrfs_set_dev_extent_chunk_objectid(leaf, extent, chunk_objectid);
	btrfs_set_dev_extent_chunk_offset(leaf, extent, chunk_offset);

	write_extent_buffer(leaf, root->fs_info->chunk_tree_uuid,
		    (unsigned long)btrfs_dev_extent_chunk_tree_uuid(extent),
		    BTRFS_UUID_SIZE);

1239 1240
	btrfs_set_dev_extent_length(leaf, extent, num_bytes);
	btrfs_mark_buffer_dirty(leaf);
1241
out:
1242 1243 1244 1245
	btrfs_free_path(path);
	return ret;
}

1246 1247
static noinline int find_next_chunk(struct btrfs_root *root,
				    u64 objectid, u64 *offset)
1248 1249 1250 1251
{
	struct btrfs_path *path;
	int ret;
	struct btrfs_key key;
1252
	struct btrfs_chunk *chunk;
1253 1254 1255
	struct btrfs_key found_key;

	path = btrfs_alloc_path();
1256 1257
	if (!path)
		return -ENOMEM;
1258

1259
	key.objectid = objectid;
1260 1261 1262 1263 1264 1265 1266
	key.offset = (u64)-1;
	key.type = BTRFS_CHUNK_ITEM_KEY;

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

1267
	BUG_ON(ret == 0); /* Corruption */
1268 1269 1270

	ret = btrfs_previous_item(root, path, 0, BTRFS_CHUNK_ITEM_KEY);
	if (ret) {
1271
		*offset = 0;
1272 1273 1274
	} else {
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
1275 1276 1277 1278 1279 1280 1281 1282
		if (found_key.objectid != objectid)
			*offset = 0;
		else {
			chunk = btrfs_item_ptr(path->nodes[0], path->slots[0],
					       struct btrfs_chunk);
			*offset = found_key.offset +
				btrfs_chunk_length(path->nodes[0], chunk);
		}
1283 1284 1285 1286 1287 1288 1289
	}
	ret = 0;
error:
	btrfs_free_path(path);
	return ret;
}

Y
Yan Zheng 已提交
1290
static noinline int find_next_devid(struct btrfs_root *root, u64 *objectid)
1291 1292 1293 1294
{
	int ret;
	struct btrfs_key key;
	struct btrfs_key found_key;
Y
Yan Zheng 已提交
1295 1296 1297 1298 1299 1300 1301
	struct btrfs_path *path;

	root = root->fs_info->chunk_root;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
1302 1303 1304 1305 1306 1307 1308 1309 1310

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

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

1311
	BUG_ON(ret == 0); /* Corruption */
1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323

	ret = btrfs_previous_item(root, path, BTRFS_DEV_ITEMS_OBJECTID,
				  BTRFS_DEV_ITEM_KEY);
	if (ret) {
		*objectid = 1;
	} else {
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
		*objectid = found_key.offset + 1;
	}
	ret = 0;
error:
Y
Yan Zheng 已提交
1324
	btrfs_free_path(path);
1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350
	return ret;
}

/*
 * the device information is stored in the chunk root
 * the btrfs_device struct should be fully filled in
 */
int btrfs_add_device(struct btrfs_trans_handle *trans,
		     struct btrfs_root *root,
		     struct btrfs_device *device)
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_dev_item *dev_item;
	struct extent_buffer *leaf;
	struct btrfs_key key;
	unsigned long ptr;

	root = root->fs_info->chunk_root;

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

	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.type = BTRFS_DEV_ITEM_KEY;
Y
Yan Zheng 已提交
1351
	key.offset = device->devid;
1352 1353

	ret = btrfs_insert_empty_item(trans, root, path, &key,
1354
				      sizeof(*dev_item));
1355 1356 1357 1358 1359 1360 1361
	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
1362
	btrfs_set_device_generation(leaf, dev_item, 0);
1363 1364 1365 1366 1367 1368
	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);
1369 1370 1371
	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);
1372
	btrfs_set_device_start_offset(leaf, dev_item, 0);
1373 1374

	ptr = (unsigned long)btrfs_device_uuid(dev_item);
1375
	write_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
1376 1377
	ptr = (unsigned long)btrfs_device_fsid(dev_item);
	write_extent_buffer(leaf, root->fs_info->fsid, ptr, BTRFS_UUID_SIZE);
1378 1379
	btrfs_mark_buffer_dirty(leaf);

Y
Yan Zheng 已提交
1380
	ret = 0;
1381 1382 1383 1384
out:
	btrfs_free_path(path);
	return ret;
}
1385

1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399
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;

1400
	trans = btrfs_start_transaction(root, 0);
1401 1402 1403 1404
	if (IS_ERR(trans)) {
		btrfs_free_path(path);
		return PTR_ERR(trans);
	}
1405 1406 1407
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.type = BTRFS_DEV_ITEM_KEY;
	key.offset = device->devid;
1408
	lock_chunks(root);
1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423

	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);
1424
	unlock_chunks(root);
1425 1426 1427 1428 1429 1430 1431
	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
1432
	struct btrfs_device *next_device;
1433
	struct block_device *bdev;
1434
	struct buffer_head *bh = NULL;
1435
	struct btrfs_super_block *disk_super;
1436
	struct btrfs_fs_devices *cur_devices;
1437 1438
	u64 all_avail;
	u64 devid;
Y
Yan Zheng 已提交
1439 1440
	u64 num_devices;
	u8 *dev_uuid;
1441
	unsigned seq;
1442
	int ret = 0;
1443
	bool clear_super = false;
1444 1445 1446

	mutex_lock(&uuid_mutex);

1447 1448 1449 1450 1451 1452 1453
	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));
1454

1455 1456 1457 1458 1459 1460 1461 1462 1463
	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) {
C
Chris Mason 已提交
1464 1465
		printk(KERN_ERR "btrfs: unable to go below four devices "
		       "on raid10\n");
1466 1467 1468 1469
		ret = -EINVAL;
		goto out;
	}

1470
	if ((all_avail & BTRFS_BLOCK_GROUP_RAID1) && num_devices <= 2) {
C
Chris Mason 已提交
1471 1472
		printk(KERN_ERR "btrfs: unable to go below two "
		       "devices on raid1\n");
1473 1474 1475 1476
		ret = -EINVAL;
		goto out;
	}

D
David Woodhouse 已提交
1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491
	if ((all_avail & BTRFS_BLOCK_GROUP_RAID5) &&
	    root->fs_info->fs_devices->rw_devices <= 2) {
		printk(KERN_ERR "btrfs: unable to go below two "
		       "devices on raid5\n");
		ret = -EINVAL;
		goto out;
	}
	if ((all_avail & BTRFS_BLOCK_GROUP_RAID6) &&
	    root->fs_info->fs_devices->rw_devices <= 3) {
		printk(KERN_ERR "btrfs: unable to go below three "
		       "devices on raid6\n");
		ret = -EINVAL;
		goto out;
	}

1492 1493 1494
	if (strcmp(device_path, "missing") == 0) {
		struct list_head *devices;
		struct btrfs_device *tmp;
1495

1496 1497
		device = NULL;
		devices = &root->fs_info->fs_devices->devices;
1498 1499 1500 1501
		/*
		 * It is safe to read the devices since the volume_mutex
		 * is held.
		 */
Q
Qinghuang Feng 已提交
1502
		list_for_each_entry(tmp, devices, dev_list) {
1503 1504 1505
			if (tmp->in_fs_metadata &&
			    !tmp->is_tgtdev_for_dev_replace &&
			    !tmp->bdev) {
1506 1507 1508 1509 1510 1511 1512 1513
				device = tmp;
				break;
			}
		}
		bdev = NULL;
		bh = NULL;
		disk_super = NULL;
		if (!device) {
C
Chris Mason 已提交
1514 1515
			printk(KERN_ERR "btrfs: no missing devices found to "
			       "remove\n");
1516 1517 1518
			goto out;
		}
	} else {
1519
		ret = btrfs_get_bdev_and_sb(device_path,
1520
					    FMODE_WRITE | FMODE_EXCL,
1521 1522 1523
					    root->fs_info->bdev_holder, 0,
					    &bdev, &bh);
		if (ret)
1524 1525
			goto out;
		disk_super = (struct btrfs_super_block *)bh->b_data;
1526
		devid = btrfs_stack_device_id(&disk_super->dev_item);
Y
Yan Zheng 已提交
1527
		dev_uuid = disk_super->dev_item.uuid;
1528
		device = btrfs_find_device(root->fs_info, devid, dev_uuid,
Y
Yan Zheng 已提交
1529
					   disk_super->fsid);
1530 1531 1532 1533
		if (!device) {
			ret = -ENOENT;
			goto error_brelse;
		}
Y
Yan Zheng 已提交
1534
	}
1535

1536 1537 1538 1539 1540 1541
	if (device->is_tgtdev_for_dev_replace) {
		pr_err("btrfs: unable to remove the dev_replace target dev\n");
		ret = -EINVAL;
		goto error_brelse;
	}

Y
Yan Zheng 已提交
1542
	if (device->writeable && root->fs_info->fs_devices->rw_devices == 1) {
C
Chris Mason 已提交
1543 1544
		printk(KERN_ERR "btrfs: unable to remove the only writeable "
		       "device\n");
Y
Yan Zheng 已提交
1545 1546 1547 1548 1549
		ret = -EINVAL;
		goto error_brelse;
	}

	if (device->writeable) {
1550
		lock_chunks(root);
Y
Yan Zheng 已提交
1551
		list_del_init(&device->dev_alloc_list);
1552
		unlock_chunks(root);
Y
Yan Zheng 已提交
1553
		root->fs_info->fs_devices->rw_devices--;
1554
		clear_super = true;
1555
	}
1556 1557 1558

	ret = btrfs_shrink_device(device, 0);
	if (ret)
1559
		goto error_undo;
1560

1561 1562 1563 1564 1565
	/*
	 * 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.
	 */
1566 1567
	ret = btrfs_rm_dev_item(root->fs_info->chunk_root, device);
	if (ret)
1568
		goto error_undo;
1569

1570 1571 1572 1573 1574
	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
1575
	device->in_fs_metadata = 0;
1576
	btrfs_scrub_cancel_dev(root->fs_info, device);
1577 1578 1579 1580 1581 1582

	/*
	 * the device list mutex makes sure that we don't change
	 * the device list while someone else is writing out all
	 * the device supers.
	 */
1583 1584

	cur_devices = device->fs_devices;
1585
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1586
	list_del_rcu(&device->dev_list);
1587

Y
Yan Zheng 已提交
1588
	device->fs_devices->num_devices--;
J
Josef Bacik 已提交
1589
	device->fs_devices->total_devices--;
Y
Yan Zheng 已提交
1590

1591 1592 1593
	if (device->missing)
		root->fs_info->fs_devices->missing_devices--;

Y
Yan Zheng 已提交
1594 1595 1596 1597 1598 1599 1600
	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;

1601
	if (device->bdev)
Y
Yan Zheng 已提交
1602
		device->fs_devices->open_devices--;
1603 1604 1605

	call_rcu(&device->rcu, free_device);
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
1606

1607 1608
	num_devices = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
	btrfs_set_super_num_devices(root->fs_info->super_copy, num_devices);
Y
Yan Zheng 已提交
1609

1610
	if (cur_devices->open_devices == 0) {
Y
Yan Zheng 已提交
1611 1612 1613
		struct btrfs_fs_devices *fs_devices;
		fs_devices = root->fs_info->fs_devices;
		while (fs_devices) {
1614
			if (fs_devices->seed == cur_devices)
Y
Yan Zheng 已提交
1615 1616
				break;
			fs_devices = fs_devices->seed;
Y
Yan Zheng 已提交
1617
		}
1618 1619
		fs_devices->seed = cur_devices->seed;
		cur_devices->seed = NULL;
1620
		lock_chunks(root);
1621
		__btrfs_close_devices(cur_devices);
1622
		unlock_chunks(root);
1623
		free_fs_devices(cur_devices);
Y
Yan Zheng 已提交
1624 1625
	}

1626 1627 1628
	root->fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(root->fs_info);

Y
Yan Zheng 已提交
1629 1630 1631 1632
	/*
	 * at this point, the device is zero sized.  We want to
	 * remove it from the devices list and zero out the old super
	 */
1633
	if (clear_super && disk_super) {
1634 1635 1636 1637 1638 1639 1640
		/* 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);
	}
1641 1642 1643

	ret = 0;

1644
	/* Notify udev that device has changed */
1645 1646
	if (bdev)
		btrfs_kobject_uevent(bdev, KOBJ_CHANGE);
1647

1648 1649
error_brelse:
	brelse(bh);
1650
	if (bdev)
1651
		blkdev_put(bdev, FMODE_READ | FMODE_EXCL);
1652 1653 1654
out:
	mutex_unlock(&uuid_mutex);
	return ret;
1655 1656
error_undo:
	if (device->writeable) {
1657
		lock_chunks(root);
1658 1659
		list_add(&device->dev_alloc_list,
			 &root->fs_info->fs_devices->alloc_list);
1660
		unlock_chunks(root);
1661 1662 1663
		root->fs_info->fs_devices->rw_devices++;
	}
	goto error_brelse;
1664 1665
}

1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712
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));
	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--;
	if (srcdev->bdev)
		fs_info->fs_devices->open_devices--;

	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);
}

1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730
int btrfs_find_device_by_path(struct btrfs_root *root, char *device_path,
			      struct btrfs_device **device)
{
	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;
1731
	*device = btrfs_find_device(root->fs_info, devid, dev_uuid,
1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771
				    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) {
			pr_err("btrfs: no missing device found\n");
			return -ENOENT;
		}

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

Y
Yan Zheng 已提交
1772 1773 1774
/*
 * does all the dirty work required for changing file system's UUID.
 */
1775
static int btrfs_prepare_sprout(struct btrfs_root *root)
Y
Yan Zheng 已提交
1776 1777 1778
{
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
	struct btrfs_fs_devices *old_devices;
Y
Yan Zheng 已提交
1779
	struct btrfs_fs_devices *seed_devices;
1780
	struct btrfs_super_block *disk_super = root->fs_info->super_copy;
Y
Yan Zheng 已提交
1781 1782 1783 1784
	struct btrfs_device *device;
	u64 super_flags;

	BUG_ON(!mutex_is_locked(&uuid_mutex));
Y
Yan Zheng 已提交
1785
	if (!fs_devices->seeding)
Y
Yan Zheng 已提交
1786 1787
		return -EINVAL;

Y
Yan Zheng 已提交
1788 1789
	seed_devices = kzalloc(sizeof(*fs_devices), GFP_NOFS);
	if (!seed_devices)
Y
Yan Zheng 已提交
1790 1791
		return -ENOMEM;

Y
Yan Zheng 已提交
1792 1793 1794 1795
	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
1796
	}
Y
Yan Zheng 已提交
1797

Y
Yan Zheng 已提交
1798 1799
	list_add(&old_devices->list, &fs_uuids);

Y
Yan Zheng 已提交
1800 1801 1802 1803
	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);
1804
	mutex_init(&seed_devices->device_list_mutex);
1805 1806

	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1807 1808
	list_splice_init_rcu(&fs_devices->devices, &seed_devices->devices,
			      synchronize_rcu);
1809 1810
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
1811 1812 1813 1814 1815
	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
1816 1817 1818
	fs_devices->seeding = 0;
	fs_devices->num_devices = 0;
	fs_devices->open_devices = 0;
J
Josef Bacik 已提交
1819
	fs_devices->total_devices = 0;
Y
Yan Zheng 已提交
1820
	fs_devices->seed = seed_devices;
Y
Yan Zheng 已提交
1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871

	generate_random_uuid(fs_devices->fsid);
	memcpy(root->fs_info->fsid, fs_devices->fsid, BTRFS_FSID_SIZE);
	memcpy(disk_super->fsid, fs_devices->fsid, BTRFS_FSID_SIZE);
	super_flags = btrfs_super_flags(disk_super) &
		      ~BTRFS_SUPER_FLAG_SEEDING;
	btrfs_set_super_flags(disk_super, super_flags);

	return 0;
}

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

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

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

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

		leaf = path->nodes[0];
next_slot:
		if (path->slots[0] >= btrfs_header_nritems(leaf)) {
			ret = btrfs_next_leaf(root, path);
			if (ret > 0)
				break;
			if (ret < 0)
				goto error;
			leaf = path->nodes[0];
			btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
1872
			btrfs_release_path(path);
Y
Yan Zheng 已提交
1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889
			continue;
		}

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

		dev_item = btrfs_item_ptr(leaf, path->slots[0],
					  struct btrfs_dev_item);
		devid = btrfs_device_id(leaf, dev_item);
		read_extent_buffer(leaf, dev_uuid,
				   (unsigned long)btrfs_device_uuid(dev_item),
				   BTRFS_UUID_SIZE);
		read_extent_buffer(leaf, fs_uuid,
				   (unsigned long)btrfs_device_fsid(dev_item),
				   BTRFS_UUID_SIZE);
1890 1891
		device = btrfs_find_device(root->fs_info, devid, dev_uuid,
					   fs_uuid);
1892
		BUG_ON(!device); /* Logic error */
Y
Yan Zheng 已提交
1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908

		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;
}

1909 1910
int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
{
1911
	struct request_queue *q;
1912 1913 1914 1915
	struct btrfs_trans_handle *trans;
	struct btrfs_device *device;
	struct block_device *bdev;
	struct list_head *devices;
Y
Yan Zheng 已提交
1916
	struct super_block *sb = root->fs_info->sb;
1917
	struct rcu_string *name;
1918
	u64 total_bytes;
Y
Yan Zheng 已提交
1919
	int seeding_dev = 0;
1920 1921
	int ret = 0;

Y
Yan Zheng 已提交
1922
	if ((sb->s_flags & MS_RDONLY) && !root->fs_info->fs_devices->seeding)
1923
		return -EROFS;
1924

1925
	bdev = blkdev_get_by_path(device_path, FMODE_WRITE | FMODE_EXCL,
1926
				  root->fs_info->bdev_holder);
1927 1928
	if (IS_ERR(bdev))
		return PTR_ERR(bdev);
1929

Y
Yan Zheng 已提交
1930 1931 1932 1933 1934 1935
	if (root->fs_info->fs_devices->seeding) {
		seeding_dev = 1;
		down_write(&sb->s_umount);
		mutex_lock(&uuid_mutex);
	}

1936
	filemap_write_and_wait(bdev->bd_inode->i_mapping);
1937

1938
	devices = &root->fs_info->fs_devices->devices;
1939 1940

	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
Q
Qinghuang Feng 已提交
1941
	list_for_each_entry(device, devices, dev_list) {
1942 1943
		if (device->bdev == bdev) {
			ret = -EEXIST;
1944 1945
			mutex_unlock(
				&root->fs_info->fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
1946
			goto error;
1947 1948
		}
	}
1949
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
1950 1951 1952 1953 1954

	device = kzalloc(sizeof(*device), GFP_NOFS);
	if (!device) {
		/* we can safely leave the fs_devices entry around */
		ret = -ENOMEM;
Y
Yan Zheng 已提交
1955
		goto error;
1956 1957
	}

1958 1959
	name = rcu_string_strdup(device_path, GFP_NOFS);
	if (!name) {
1960
		kfree(device);
Y
Yan Zheng 已提交
1961 1962
		ret = -ENOMEM;
		goto error;
1963
	}
1964
	rcu_assign_pointer(device->name, name);
Y
Yan Zheng 已提交
1965 1966 1967

	ret = find_next_devid(root, &device->devid);
	if (ret) {
1968
		rcu_string_free(device->name);
Y
Yan Zheng 已提交
1969 1970 1971 1972
		kfree(device);
		goto error;
	}

1973
	trans = btrfs_start_transaction(root, 0);
1974
	if (IS_ERR(trans)) {
1975
		rcu_string_free(device->name);
1976 1977 1978 1979 1980
		kfree(device);
		ret = PTR_ERR(trans);
		goto error;
	}

Y
Yan Zheng 已提交
1981 1982
	lock_chunks(root);

1983 1984 1985
	q = bdev_get_queue(bdev);
	if (blk_queue_discard(q))
		device->can_discard = 1;
Y
Yan Zheng 已提交
1986 1987 1988 1989 1990
	device->writeable = 1;
	device->work.func = pending_bios_fn;
	generate_random_uuid(device->uuid);
	spin_lock_init(&device->io_lock);
	device->generation = trans->transid;
1991 1992 1993 1994
	device->io_width = root->sectorsize;
	device->io_align = root->sectorsize;
	device->sector_size = root->sectorsize;
	device->total_bytes = i_size_read(bdev->bd_inode);
1995
	device->disk_total_bytes = device->total_bytes;
1996 1997
	device->dev_root = root->fs_info->dev_root;
	device->bdev = bdev;
1998
	device->in_fs_metadata = 1;
1999
	device->is_tgtdev_for_dev_replace = 0;
2000
	device->mode = FMODE_EXCL;
Y
Yan Zheng 已提交
2001
	set_blocksize(device->bdev, 4096);
2002

Y
Yan Zheng 已提交
2003 2004
	if (seeding_dev) {
		sb->s_flags &= ~MS_RDONLY;
2005
		ret = btrfs_prepare_sprout(root);
2006
		BUG_ON(ret); /* -ENOMEM */
Y
Yan Zheng 已提交
2007
	}
2008

Y
Yan Zheng 已提交
2009
	device->fs_devices = root->fs_info->fs_devices;
2010 2011

	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
2012
	list_add_rcu(&device->dev_list, &root->fs_info->fs_devices->devices);
Y
Yan Zheng 已提交
2013 2014 2015 2016 2017
	list_add(&device->dev_alloc_list,
		 &root->fs_info->fs_devices->alloc_list);
	root->fs_info->fs_devices->num_devices++;
	root->fs_info->fs_devices->open_devices++;
	root->fs_info->fs_devices->rw_devices++;
J
Josef Bacik 已提交
2018
	root->fs_info->fs_devices->total_devices++;
2019 2020
	if (device->can_discard)
		root->fs_info->fs_devices->num_can_discard++;
Y
Yan Zheng 已提交
2021
	root->fs_info->fs_devices->total_rw_bytes += device->total_bytes;
2022

2023 2024 2025 2026
	spin_lock(&root->fs_info->free_chunk_lock);
	root->fs_info->free_chunk_space += device->total_bytes;
	spin_unlock(&root->fs_info->free_chunk_lock);

C
Chris Mason 已提交
2027 2028 2029
	if (!blk_queue_nonrot(bdev_get_queue(bdev)))
		root->fs_info->fs_devices->rotating = 1;

2030 2031
	total_bytes = btrfs_super_total_bytes(root->fs_info->super_copy);
	btrfs_set_super_total_bytes(root->fs_info->super_copy,
2032 2033
				    total_bytes + device->total_bytes);

2034 2035
	total_bytes = btrfs_super_num_devices(root->fs_info->super_copy);
	btrfs_set_super_num_devices(root->fs_info->super_copy,
2036
				    total_bytes + 1);
2037
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
2038

Y
Yan Zheng 已提交
2039 2040
	if (seeding_dev) {
		ret = init_first_rw_device(trans, root, device);
2041 2042
		if (ret) {
			btrfs_abort_transaction(trans, root, ret);
2043
			goto error_trans;
2044
		}
Y
Yan Zheng 已提交
2045
		ret = btrfs_finish_sprout(trans, root);
2046 2047
		if (ret) {
			btrfs_abort_transaction(trans, root, ret);
2048
			goto error_trans;
2049
		}
Y
Yan Zheng 已提交
2050 2051
	} else {
		ret = btrfs_add_device(trans, root, device);
2052 2053
		if (ret) {
			btrfs_abort_transaction(trans, root, ret);
2054
			goto error_trans;
2055
		}
Y
Yan Zheng 已提交
2056 2057
	}

2058 2059 2060 2061 2062 2063
	/*
	 * we've got more storage, clear any full flags on the space
	 * infos
	 */
	btrfs_clear_space_info_full(root->fs_info);

2064
	unlock_chunks(root);
2065 2066
	root->fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(root->fs_info);
2067
	ret = btrfs_commit_transaction(trans, root);
2068

Y
Yan Zheng 已提交
2069 2070 2071
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
2072

2073 2074 2075
		if (ret) /* transaction commit */
			return ret;

Y
Yan Zheng 已提交
2076
		ret = btrfs_relocate_sys_chunks(root);
2077 2078 2079 2080 2081
		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.");
2082 2083 2084 2085 2086 2087 2088
		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
2089
	}
2090

Y
Yan Zheng 已提交
2091
	return ret;
2092 2093 2094 2095

error_trans:
	unlock_chunks(root);
	btrfs_end_transaction(trans, root);
2096
	rcu_string_free(device->name);
2097
	kfree(device);
Y
Yan Zheng 已提交
2098
error:
2099
	blkdev_put(bdev, FMODE_EXCL);
Y
Yan Zheng 已提交
2100 2101 2102 2103
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
	}
2104
	return ret;
2105 2106
}

2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198
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;
	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;
		}
	}

	device = kzalloc(sizeof(*device), GFP_NOFS);
	if (!device) {
		ret = -ENOMEM;
		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->work.func = pending_bios_fn;
	generate_random_uuid(device->uuid);
	device->devid = BTRFS_DEV_REPLACE_DEVID;
	spin_lock_init(&device->io_lock);
	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;
	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;
}

C
Chris Mason 已提交
2199 2200
static noinline int btrfs_update_device(struct btrfs_trans_handle *trans,
					struct btrfs_device *device)
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 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235
{
	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);
2236
	btrfs_set_device_total_bytes(leaf, dev_item, device->disk_total_bytes);
2237 2238 2239 2240 2241 2242 2243 2244
	btrfs_set_device_bytes_used(leaf, dev_item, device->bytes_used);
	btrfs_mark_buffer_dirty(leaf);

out:
	btrfs_free_path(path);
	return ret;
}

2245
static int __btrfs_grow_device(struct btrfs_trans_handle *trans,
2246 2247 2248
		      struct btrfs_device *device, u64 new_size)
{
	struct btrfs_super_block *super_copy =
2249
		device->dev_root->fs_info->super_copy;
2250 2251 2252
	u64 old_total = btrfs_super_total_bytes(super_copy);
	u64 diff = new_size - device->total_bytes;

Y
Yan Zheng 已提交
2253 2254
	if (!device->writeable)
		return -EACCES;
2255 2256
	if (new_size <= device->total_bytes ||
	    device->is_tgtdev_for_dev_replace)
Y
Yan Zheng 已提交
2257 2258
		return -EINVAL;

2259
	btrfs_set_super_total_bytes(super_copy, old_total + diff);
Y
Yan Zheng 已提交
2260 2261 2262
	device->fs_devices->total_rw_bytes += diff;

	device->total_bytes = new_size;
2263
	device->disk_total_bytes = new_size;
2264 2265
	btrfs_clear_space_info_full(device->dev_root->fs_info);

2266 2267 2268
	return btrfs_update_device(trans, device);
}

2269 2270 2271 2272 2273 2274 2275 2276 2277 2278
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;
}

2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297
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);
2298 2299 2300 2301 2302 2303 2304 2305
	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;
	}
2306 2307

	ret = btrfs_del_item(trans, root, path);
2308 2309 2310 2311
	if (ret < 0)
		btrfs_error(root->fs_info, ret,
			    "Failed to delete chunk item.");
out:
2312
	btrfs_free_path(path);
2313
	return ret;
2314 2315
}

2316
static int btrfs_del_sys_chunk(struct btrfs_root *root, u64 chunk_objectid, u64
2317 2318
			chunk_offset)
{
2319
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361
	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;
}

2362
static int btrfs_relocate_chunk(struct btrfs_root *root,
2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377
			 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;

2378 2379 2380 2381
	ret = btrfs_can_relocate(extent_root, chunk_offset);
	if (ret)
		return -ENOSPC;

2382
	/* step one, relocate all the extents inside this chunk */
Z
Zheng Yan 已提交
2383
	ret = btrfs_relocate_block_group(extent_root, chunk_offset);
2384 2385
	if (ret)
		return ret;
2386

2387
	trans = btrfs_start_transaction(root, 0);
2388 2389 2390 2391 2392
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		btrfs_std_error(root->fs_info, ret);
		return ret;
	}
2393

2394 2395
	lock_chunks(root);

2396 2397 2398 2399
	/*
	 * step two, delete the device extents and the
	 * chunk tree entries
	 */
2400
	read_lock(&em_tree->lock);
2401
	em = lookup_extent_mapping(em_tree, chunk_offset, 1);
2402
	read_unlock(&em_tree->lock);
2403

2404
	BUG_ON(!em || em->start > chunk_offset ||
2405
	       em->start + em->len < chunk_offset);
2406 2407 2408 2409 2410 2411
	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);
2412

2413 2414 2415 2416
		if (map->stripes[i].dev) {
			ret = btrfs_update_device(trans, map->stripes[i].dev);
			BUG_ON(ret);
		}
2417 2418 2419 2420 2421 2422
	}
	ret = btrfs_free_chunk(trans, root, chunk_tree, chunk_objectid,
			       chunk_offset);

	BUG_ON(ret);

2423 2424
	trace_btrfs_chunk_free(root, map, chunk_offset, em->len);

2425 2426 2427 2428 2429
	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
2430 2431 2432
	ret = btrfs_remove_block_group(trans, extent_root, chunk_offset);
	BUG_ON(ret);

2433
	write_lock(&em_tree->lock);
Y
Yan Zheng 已提交
2434
	remove_extent_mapping(em_tree, em);
2435
	write_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459

	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;
2460 2461
	bool retried = false;
	int failed = 0;
Y
Yan Zheng 已提交
2462 2463 2464 2465 2466 2467
	int ret;

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

2468
again:
Y
Yan Zheng 已提交
2469 2470 2471 2472 2473 2474 2475 2476
	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;
2477
		BUG_ON(ret == 0); /* Corruption */
Y
Yan Zheng 已提交
2478 2479 2480 2481 2482 2483 2484

		ret = btrfs_previous_item(chunk_root, path, key.objectid,
					  key.type);
		if (ret < 0)
			goto error;
		if (ret > 0)
			break;
Z
Zheng Yan 已提交
2485

Y
Yan Zheng 已提交
2486 2487
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
Z
Zheng Yan 已提交
2488

Y
Yan Zheng 已提交
2489 2490 2491
		chunk = btrfs_item_ptr(leaf, path->slots[0],
				       struct btrfs_chunk);
		chunk_type = btrfs_chunk_type(leaf, chunk);
2492
		btrfs_release_path(path);
2493

Y
Yan Zheng 已提交
2494 2495 2496 2497
		if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) {
			ret = btrfs_relocate_chunk(chunk_root, chunk_tree,
						   found_key.objectid,
						   found_key.offset);
2498 2499 2500 2501
			if (ret == -ENOSPC)
				failed++;
			else if (ret)
				BUG();
Y
Yan Zheng 已提交
2502
		}
2503

Y
Yan Zheng 已提交
2504 2505 2506 2507 2508
		if (found_key.offset == 0)
			break;
		key.offset = found_key.offset - 1;
	}
	ret = 0;
2509 2510 2511 2512 2513 2514 2515 2516
	if (failed && !retried) {
		failed = 0;
		retried = true;
		goto again;
	} else if (failed && retried) {
		WARN_ON(1);
		ret = -ENOSPC;
	}
Y
Yan Zheng 已提交
2517 2518 2519
error:
	btrfs_free_path(path);
	return ret;
2520 2521
}

2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 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
static int insert_balance_item(struct btrfs_root *root,
			       struct btrfs_balance_control *bctl)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_balance_item *item;
	struct btrfs_disk_balance_args disk_bargs;
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_key key;
	int ret, err;

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

I
Ilya Dryomov 已提交
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
/*
 * 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;
	}
}

2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681
/*
 * 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
2682 2683 2684 2685
/*
 * Balance filters.  Return 1 if chunk should be filtered out
 * (should not be balanced).
 */
2686
static int chunk_profiles_filter(u64 chunk_type,
I
Ilya Dryomov 已提交
2687 2688
				 struct btrfs_balance_args *bargs)
{
2689 2690
	chunk_type = chunk_to_extended(chunk_type) &
				BTRFS_EXTENDED_PROFILE_MASK;
I
Ilya Dryomov 已提交
2691

2692
	if (bargs->profiles & chunk_type)
I
Ilya Dryomov 已提交
2693 2694 2695 2696 2697
		return 0;

	return 1;
}

I
Ilya Dryomov 已提交
2698 2699 2700 2701 2702 2703 2704 2705 2706 2707
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);

2708
	if (bargs->usage == 0)
2709
		user_thresh = 1;
2710 2711 2712 2713 2714 2715
	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
2716 2717 2718 2719 2720 2721 2722
	if (chunk_used < user_thresh)
		ret = 0;

	btrfs_put_block_group(cache);
	return ret;
}

I
Ilya Dryomov 已提交
2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739
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
2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756
/* [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
2757 2758 2759 2760 2761 2762 2763 2764 2765
	     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
2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783

	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;
}

2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797
/* [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;
}

2798
static int chunk_soft_convert_filter(u64 chunk_type,
2799 2800 2801 2802 2803
				     struct btrfs_balance_args *bargs)
{
	if (!(bargs->flags & BTRFS_BALANCE_ARGS_CONVERT))
		return 0;

2804 2805
	chunk_type = chunk_to_extended(chunk_type) &
				BTRFS_EXTENDED_PROFILE_MASK;
2806

2807
	if (bargs->target == chunk_type)
2808 2809 2810 2811 2812
		return 1;

	return 0;
}

2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833
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
2834 2835 2836 2837
	/* profiles filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_PROFILES) &&
	    chunk_profiles_filter(chunk_type, bargs)) {
		return 0;
I
Ilya Dryomov 已提交
2838 2839 2840 2841 2842 2843
	}

	/* 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
2844 2845 2846 2847 2848 2849
	}

	/* devid filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_DEVID) &&
	    chunk_devid_filter(leaf, chunk, bargs)) {
		return 0;
I
Ilya Dryomov 已提交
2850 2851 2852 2853 2854 2855
	}

	/* 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;
2856 2857 2858 2859 2860 2861
	}

	/* 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
2862 2863
	}

2864 2865 2866 2867 2868 2869
	/* soft profile changing mode */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_SOFT) &&
	    chunk_soft_convert_filter(chunk_type, bargs)) {
		return 0;
	}

2870 2871 2872
	return 1;
}

2873
static int __btrfs_balance(struct btrfs_fs_info *fs_info)
2874
{
2875
	struct btrfs_balance_control *bctl = fs_info->balance_ctl;
2876 2877 2878
	struct btrfs_root *chunk_root = fs_info->chunk_root;
	struct btrfs_root *dev_root = fs_info->dev_root;
	struct list_head *devices;
2879 2880 2881
	struct btrfs_device *device;
	u64 old_size;
	u64 size_to_free;
2882
	struct btrfs_chunk *chunk;
2883 2884 2885
	struct btrfs_path *path;
	struct btrfs_key key;
	struct btrfs_key found_key;
2886
	struct btrfs_trans_handle *trans;
2887 2888
	struct extent_buffer *leaf;
	int slot;
2889 2890
	int ret;
	int enospc_errors = 0;
2891
	bool counting = true;
2892 2893

	/* step one make some room on all the devices */
2894
	devices = &fs_info->fs_devices->devices;
Q
Qinghuang Feng 已提交
2895
	list_for_each_entry(device, devices, dev_list) {
2896 2897 2898
		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
2899
		if (!device->writeable ||
2900 2901
		    device->total_bytes - device->bytes_used > size_to_free ||
		    device->is_tgtdev_for_dev_replace)
2902 2903 2904
			continue;

		ret = btrfs_shrink_device(device, old_size - size_to_free);
2905 2906
		if (ret == -ENOSPC)
			break;
2907 2908
		BUG_ON(ret);

2909
		trans = btrfs_start_transaction(dev_root, 0);
2910
		BUG_ON(IS_ERR(trans));
2911 2912 2913 2914 2915 2916 2917 2918 2919

		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();
2920 2921 2922 2923
	if (!path) {
		ret = -ENOMEM;
		goto error;
	}
2924 2925 2926 2927 2928 2929

	/* zero out stat counters */
	spin_lock(&fs_info->balance_lock);
	memset(&bctl->stat, 0, sizeof(bctl->stat));
	spin_unlock(&fs_info->balance_lock);
again:
2930 2931 2932 2933
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.offset = (u64)-1;
	key.type = BTRFS_CHUNK_ITEM_KEY;

C
Chris Mason 已提交
2934
	while (1) {
2935
		if ((!counting && atomic_read(&fs_info->balance_pause_req)) ||
2936
		    atomic_read(&fs_info->balance_cancel_req)) {
2937 2938 2939 2940
			ret = -ECANCELED;
			goto error;
		}

2941 2942 2943 2944 2945 2946 2947 2948 2949
		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)
2950
			BUG(); /* FIXME break ? */
2951 2952 2953

		ret = btrfs_previous_item(chunk_root, path, 0,
					  BTRFS_CHUNK_ITEM_KEY);
2954 2955
		if (ret) {
			ret = 0;
2956
			break;
2957
		}
2958

2959 2960 2961
		leaf = path->nodes[0];
		slot = path->slots[0];
		btrfs_item_key_to_cpu(leaf, &found_key, slot);
2962

2963 2964
		if (found_key.objectid != key.objectid)
			break;
2965

2966
		/* chunk zero is special */
2967
		if (found_key.offset == 0)
2968 2969
			break;

2970 2971
		chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);

2972 2973 2974 2975 2976 2977
		if (!counting) {
			spin_lock(&fs_info->balance_lock);
			bctl->stat.considered++;
			spin_unlock(&fs_info->balance_lock);
		}

2978 2979
		ret = should_balance_chunk(chunk_root, leaf, chunk,
					   found_key.offset);
2980
		btrfs_release_path(path);
2981 2982 2983
		if (!ret)
			goto loop;

2984 2985 2986 2987 2988 2989 2990
		if (counting) {
			spin_lock(&fs_info->balance_lock);
			bctl->stat.expected++;
			spin_unlock(&fs_info->balance_lock);
			goto loop;
		}

2991 2992 2993 2994
		ret = btrfs_relocate_chunk(chunk_root,
					   chunk_root->root_key.objectid,
					   found_key.objectid,
					   found_key.offset);
2995 2996
		if (ret && ret != -ENOSPC)
			goto error;
2997
		if (ret == -ENOSPC) {
2998
			enospc_errors++;
2999 3000 3001 3002 3003
		} else {
			spin_lock(&fs_info->balance_lock);
			bctl->stat.completed++;
			spin_unlock(&fs_info->balance_lock);
		}
3004
loop:
3005
		key.offset = found_key.offset - 1;
3006
	}
3007

3008 3009 3010 3011 3012
	if (counting) {
		btrfs_release_path(path);
		counting = false;
		goto again;
	}
3013 3014
error:
	btrfs_free_path(path);
3015 3016 3017 3018 3019 3020 3021
	if (enospc_errors) {
		printk(KERN_INFO "btrfs: %d enospc errors during balance\n",
		       enospc_errors);
		if (!ret)
			ret = -ENOSPC;
	}

3022 3023 3024
	return ret;
}

3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048
/**
 * 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;
}

3049 3050
static inline int balance_need_close(struct btrfs_fs_info *fs_info)
{
3051 3052 3053 3054
	/* 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);
3055 3056
}

3057 3058
static void __cancel_balance(struct btrfs_fs_info *fs_info)
{
3059 3060
	int ret;

3061
	unset_balance_control(fs_info);
3062
	ret = del_balance_item(fs_info->tree_root);
3063 3064
	if (ret)
		btrfs_std_error(fs_info, ret);
3065 3066

	atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
3067 3068
}

3069
void update_ioctl_balance_args(struct btrfs_fs_info *fs_info, int lock,
3070 3071 3072 3073 3074 3075 3076 3077 3078
			       struct btrfs_ioctl_balance_args *bargs);

/*
 * 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;
3079
	u64 allowed;
3080
	int mixed = 0;
3081
	int ret;
3082
	u64 num_devices;
3083
	unsigned seq;
3084

3085
	if (btrfs_fs_closing(fs_info) ||
3086 3087
	    atomic_read(&fs_info->balance_pause_req) ||
	    atomic_read(&fs_info->balance_cancel_req)) {
3088 3089 3090 3091
		ret = -EINVAL;
		goto out;
	}

3092 3093 3094 3095
	allowed = btrfs_super_incompat_flags(fs_info->super_copy);
	if (allowed & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS)
		mixed = 1;

3096 3097 3098 3099
	/*
	 * In case of mixed groups both data and meta should be picked,
	 * and identical options should be given for both of them.
	 */
3100 3101
	allowed = BTRFS_BALANCE_DATA | BTRFS_BALANCE_METADATA;
	if (mixed && (bctl->flags & allowed)) {
3102 3103 3104 3105 3106 3107 3108 3109 3110 3111
		if (!(bctl->flags & BTRFS_BALANCE_DATA) ||
		    !(bctl->flags & BTRFS_BALANCE_METADATA) ||
		    memcmp(&bctl->data, &bctl->meta, sizeof(bctl->data))) {
			printk(KERN_ERR "btrfs: with mixed groups data and "
			       "metadata balance options must be the same\n");
			ret = -EINVAL;
			goto out;
		}
	}

3112 3113 3114 3115 3116 3117 3118
	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);
3119
	allowed = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
3120
	if (num_devices == 1)
3121
		allowed |= BTRFS_BLOCK_GROUP_DUP;
3122
	else if (num_devices < 4)
3123 3124 3125
		allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1);
	else
		allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1 |
D
David Woodhouse 已提交
3126 3127 3128
				BTRFS_BLOCK_GROUP_RAID10 |
				BTRFS_BLOCK_GROUP_RAID5 |
				BTRFS_BLOCK_GROUP_RAID6);
3129

3130 3131 3132
	if ((bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
	    (!alloc_profile_is_valid(bctl->data.target, 1) ||
	     (bctl->data.target & ~allowed))) {
3133 3134 3135 3136 3137 3138
		printk(KERN_ERR "btrfs: unable to start balance with target "
		       "data profile %llu\n",
		       (unsigned long long)bctl->data.target);
		ret = -EINVAL;
		goto out;
	}
3139 3140 3141
	if ((bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
	    (!alloc_profile_is_valid(bctl->meta.target, 1) ||
	     (bctl->meta.target & ~allowed))) {
3142 3143 3144 3145 3146 3147
		printk(KERN_ERR "btrfs: unable to start balance with target "
		       "metadata profile %llu\n",
		       (unsigned long long)bctl->meta.target);
		ret = -EINVAL;
		goto out;
	}
3148 3149 3150
	if ((bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
	    (!alloc_profile_is_valid(bctl->sys.target, 1) ||
	     (bctl->sys.target & ~allowed))) {
3151 3152 3153 3154 3155 3156 3157
		printk(KERN_ERR "btrfs: unable to start balance with target "
		       "system profile %llu\n",
		       (unsigned long long)bctl->sys.target);
		ret = -EINVAL;
		goto out;
	}

3158 3159
	/* allow dup'ed data chunks only in mixed mode */
	if (!mixed && (bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
3160
	    (bctl->data.target & BTRFS_BLOCK_GROUP_DUP)) {
3161 3162 3163 3164 3165 3166 3167
		printk(KERN_ERR "btrfs: dup for data is not allowed\n");
		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
3168 3169 3170
			BTRFS_BLOCK_GROUP_RAID10 |
			BTRFS_BLOCK_GROUP_RAID5 |
			BTRFS_BLOCK_GROUP_RAID6;
3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188
	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) {
				printk(KERN_INFO "btrfs: force reducing metadata "
				       "integrity\n");
			} else {
				printk(KERN_ERR "btrfs: balance will reduce metadata "
				       "integrity, use force if you want this\n");
				ret = -EINVAL;
				goto out;
			}
3189
		}
3190
	} while (read_seqretry(&fs_info->profiles_lock, seq));
3191

3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211
	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;
	}

3212
	ret = insert_balance_item(fs_info->tree_root, bctl);
I
Ilya Dryomov 已提交
3213
	if (ret && ret != -EEXIST)
3214 3215
		goto out;

I
Ilya Dryomov 已提交
3216 3217 3218 3219 3220 3221 3222 3223 3224
	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);
	}
3225

3226
	atomic_inc(&fs_info->balance_running);
3227 3228 3229 3230 3231
	mutex_unlock(&fs_info->balance_mutex);

	ret = __btrfs_balance(fs_info);

	mutex_lock(&fs_info->balance_mutex);
3232
	atomic_dec(&fs_info->balance_running);
3233

3234 3235 3236 3237 3238
	if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
		fs_info->num_tolerated_disk_barrier_failures =
			btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
	}

3239 3240
	if (bargs) {
		memset(bargs, 0, sizeof(*bargs));
3241
		update_ioctl_balance_args(fs_info, 0, bargs);
3242 3243
	}

3244 3245 3246 3247 3248
	if ((ret && ret != -ECANCELED && ret != -ENOSPC) ||
	    balance_need_close(fs_info)) {
		__cancel_balance(fs_info);
	}

3249
	wake_up(&fs_info->balance_wait_q);
3250 3251 3252

	return ret;
out:
I
Ilya Dryomov 已提交
3253 3254
	if (bctl->flags & BTRFS_BALANCE_RESUME)
		__cancel_balance(fs_info);
3255
	else {
I
Ilya Dryomov 已提交
3256
		kfree(bctl);
3257 3258
		atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
	}
I
Ilya Dryomov 已提交
3259 3260 3261 3262 3263
	return ret;
}

static int balance_kthread(void *data)
{
3264
	struct btrfs_fs_info *fs_info = data;
3265
	int ret = 0;
I
Ilya Dryomov 已提交
3266 3267 3268 3269

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

3270
	if (fs_info->balance_ctl) {
3271
		printk(KERN_INFO "btrfs: continuing balance\n");
3272
		ret = btrfs_balance(fs_info->balance_ctl, NULL);
3273
	}
I
Ilya Dryomov 已提交
3274 3275 3276

	mutex_unlock(&fs_info->balance_mutex);
	mutex_unlock(&fs_info->volume_mutex);
3277

I
Ilya Dryomov 已提交
3278 3279 3280
	return ret;
}

3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303
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)) {
		printk(KERN_INFO "btrfs: force skipping balance\n");
		return 0;
	}

	tsk = kthread_run(balance_kthread, fs_info, "btrfs-balance");
	if (IS_ERR(tsk))
		return PTR_ERR(tsk);

	return 0;
}

3304
int btrfs_recover_balance(struct btrfs_fs_info *fs_info)
I
Ilya Dryomov 已提交
3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321
{
	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;

3322
	ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0);
I
Ilya Dryomov 已提交
3323
	if (ret < 0)
3324
		goto out;
I
Ilya Dryomov 已提交
3325 3326
	if (ret > 0) { /* ret = -ENOENT; */
		ret = 0;
3327 3328 3329 3330 3331 3332 3333
		goto out;
	}

	bctl = kzalloc(sizeof(*bctl), GFP_NOFS);
	if (!bctl) {
		ret = -ENOMEM;
		goto out;
I
Ilya Dryomov 已提交
3334 3335 3336 3337 3338
	}

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

3339 3340 3341
	bctl->fs_info = fs_info;
	bctl->flags = btrfs_balance_flags(leaf, item);
	bctl->flags |= BTRFS_BALANCE_RESUME;
I
Ilya Dryomov 已提交
3342 3343 3344 3345 3346 3347 3348 3349

	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);

3350 3351
	WARN_ON(atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1));

3352 3353
	mutex_lock(&fs_info->volume_mutex);
	mutex_lock(&fs_info->balance_mutex);
I
Ilya Dryomov 已提交
3354

3355 3356 3357 3358
	set_balance_control(bctl);

	mutex_unlock(&fs_info->balance_mutex);
	mutex_unlock(&fs_info->volume_mutex);
I
Ilya Dryomov 已提交
3359 3360
out:
	btrfs_free_path(path);
3361 3362 3363
	return ret;
}

3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392
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;
}

3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428
int btrfs_cancel_balance(struct btrfs_fs_info *fs_info)
{
	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;
}

3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445
/*
 * 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;
3446 3447
	int failed = 0;
	bool retried = false;
3448 3449
	struct extent_buffer *l;
	struct btrfs_key key;
3450
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
3451
	u64 old_total = btrfs_super_total_bytes(super_copy);
3452
	u64 old_size = device->total_bytes;
3453 3454
	u64 diff = device->total_bytes - new_size;

3455 3456 3457
	if (device->is_tgtdev_for_dev_replace)
		return -EINVAL;

3458 3459 3460 3461 3462 3463
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	path->reada = 2;

3464 3465
	lock_chunks(root);

3466
	device->total_bytes = new_size;
3467
	if (device->writeable) {
Y
Yan Zheng 已提交
3468
		device->fs_devices->total_rw_bytes -= diff;
3469 3470 3471 3472
		spin_lock(&root->fs_info->free_chunk_lock);
		root->fs_info->free_chunk_space -= diff;
		spin_unlock(&root->fs_info->free_chunk_lock);
	}
3473
	unlock_chunks(root);
3474

3475
again:
3476 3477 3478 3479
	key.objectid = device->devid;
	key.offset = (u64)-1;
	key.type = BTRFS_DEV_EXTENT_KEY;

3480
	do {
3481 3482 3483 3484 3485 3486 3487 3488 3489
		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;
3490
			btrfs_release_path(path);
3491
			break;
3492 3493 3494 3495 3496 3497
		}

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

3498
		if (key.objectid != device->devid) {
3499
			btrfs_release_path(path);
3500
			break;
3501
		}
3502 3503 3504 3505

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

3506
		if (key.offset + length <= new_size) {
3507
			btrfs_release_path(path);
3508
			break;
3509
		}
3510 3511 3512 3513

		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);
3514
		btrfs_release_path(path);
3515 3516 3517

		ret = btrfs_relocate_chunk(root, chunk_tree, chunk_objectid,
					   chunk_offset);
3518
		if (ret && ret != -ENOSPC)
3519
			goto done;
3520 3521
		if (ret == -ENOSPC)
			failed++;
3522
	} while (key.offset-- > 0);
3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534

	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;
3535 3536 3537
		spin_lock(&root->fs_info->free_chunk_lock);
		root->fs_info->free_chunk_space += diff;
		spin_unlock(&root->fs_info->free_chunk_lock);
3538 3539
		unlock_chunks(root);
		goto done;
3540 3541
	}

3542
	/* Shrinking succeeded, else we would be at "done". */
3543
	trans = btrfs_start_transaction(root, 0);
3544 3545 3546 3547 3548
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto done;
	}

3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562
	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);
3563 3564 3565 3566 3567
done:
	btrfs_free_path(path);
	return ret;
}

3568
static int btrfs_add_system_chunk(struct btrfs_root *root,
3569 3570 3571
			   struct btrfs_key *key,
			   struct btrfs_chunk *chunk, int item_size)
{
3572
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590
	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;
}

3591 3592 3593 3594
/*
 * sort the devices in descending order by max_avail, total_avail
 */
static int btrfs_cmp_device_info(const void *a, const void *b)
3595
{
3596 3597
	const struct btrfs_device_info *di_a = a;
	const struct btrfs_device_info *di_b = b;
3598

3599
	if (di_a->max_avail > di_b->max_avail)
3600
		return -1;
3601
	if (di_a->max_avail < di_b->max_avail)
3602
		return 1;
3603 3604 3605 3606 3607
	if (di_a->total_avail > di_b->total_avail)
		return -1;
	if (di_a->total_avail < di_b->total_avail)
		return 1;
	return 0;
3608
}
3609

3610
struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = {
3611 3612 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
	[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,
	},
3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666
	[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,
	},
3667 3668
};

D
David Woodhouse 已提交
3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690
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)
{
	u64 features;

	if (!(type & (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)))
		return;

	features = btrfs_super_incompat_flags(info->super_copy);
	if (features & BTRFS_FEATURE_INCOMPAT_RAID56)
		return;

	features |= BTRFS_FEATURE_INCOMPAT_RAID56;
	btrfs_set_super_incompat_flags(info->super_copy, features);
	printk(KERN_INFO "btrfs: setting RAID5/6 feature flag\n");
}

3691 3692 3693 3694 3695
static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
			       struct btrfs_root *extent_root,
			       struct map_lookup **map_ret,
			       u64 *num_bytes_out, u64 *stripe_size_out,
			       u64 start, u64 type)
3696
{
3697 3698 3699 3700 3701 3702 3703 3704 3705
	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
3706 3707
	int data_stripes;	/* number of stripes that count for
				   block group size */
3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718
	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
3719
	u64 raid_stripe_len = BTRFS_STRIPE_LEN;
3720 3721 3722
	int ndevs;
	int i;
	int j;
3723
	int index;
3724

3725
	BUG_ON(!alloc_profile_is_valid(type, 0));
3726

3727 3728
	if (list_empty(&fs_devices->alloc_list))
		return -ENOSPC;
3729

3730
	index = __get_raid_index(type);
3731

3732 3733 3734 3735 3736 3737
	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;
3738

3739
	if (type & BTRFS_BLOCK_GROUP_DATA) {
3740 3741
		max_stripe_size = 1024 * 1024 * 1024;
		max_chunk_size = 10 * max_stripe_size;
3742
	} else if (type & BTRFS_BLOCK_GROUP_METADATA) {
3743 3744 3745 3746 3747
		/* 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;
3748
		max_chunk_size = max_stripe_size;
3749
	} else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
C
Chris Mason 已提交
3750
		max_stripe_size = 32 * 1024 * 1024;
3751 3752 3753 3754 3755
		max_chunk_size = 2 * max_stripe_size;
	} else {
		printk(KERN_ERR "btrfs: invalid chunk type 0x%llx requested\n",
		       type);
		BUG_ON(1);
3756 3757
	}

Y
Yan Zheng 已提交
3758 3759 3760
	/* 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);
3761

3762 3763 3764 3765
	devices_info = kzalloc(sizeof(*devices_info) * fs_devices->rw_devices,
			       GFP_NOFS);
	if (!devices_info)
		return -ENOMEM;
3766

3767
	cur = fs_devices->alloc_list.next;
3768

3769
	/*
3770 3771
	 * in the first pass through the devices list, we gather information
	 * about the available holes on each device.
3772
	 */
3773 3774 3775 3776 3777
	ndevs = 0;
	while (cur != &fs_devices->alloc_list) {
		struct btrfs_device *device;
		u64 max_avail;
		u64 dev_offset;
3778

3779
		device = list_entry(cur, struct btrfs_device, dev_alloc_list);
3780

3781
		cur = cur->next;
3782

3783
		if (!device->writeable) {
J
Julia Lawall 已提交
3784
			WARN(1, KERN_ERR
3785 3786 3787
			       "btrfs: read-only device in alloc_list\n");
			continue;
		}
3788

3789 3790
		if (!device->in_fs_metadata ||
		    device->is_tgtdev_for_dev_replace)
3791
			continue;
3792

3793 3794 3795 3796
		if (device->total_bytes > device->bytes_used)
			total_avail = device->total_bytes - device->bytes_used;
		else
			total_avail = 0;
3797 3798 3799 3800

		/* If there is no space on this device, skip it. */
		if (total_avail == 0)
			continue;
3801

3802
		ret = find_free_dev_extent(device,
3803 3804 3805 3806
					   max_stripe_size * dev_stripes,
					   &dev_offset, &max_avail);
		if (ret && ret != -ENOSPC)
			goto error;
3807

3808 3809
		if (ret == 0)
			max_avail = max_stripe_size * dev_stripes;
3810

3811 3812
		if (max_avail < BTRFS_STRIPE_LEN * dev_stripes)
			continue;
3813

3814 3815 3816 3817 3818
		if (ndevs == fs_devices->rw_devices) {
			WARN(1, "%s: found more than %llu devices\n",
			     __func__, fs_devices->rw_devices);
			break;
		}
3819 3820 3821 3822 3823 3824
		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;
	}
3825

3826 3827 3828 3829 3830
	/*
	 * now sort the devices by hole size / available space
	 */
	sort(devices_info, ndevs, sizeof(struct btrfs_device_info),
	     btrfs_cmp_device_info, NULL);
3831

3832 3833
	/* round down to number of usable stripes */
	ndevs -= ndevs % devs_increment;
3834

3835 3836 3837
	if (ndevs < devs_increment * sub_stripes || ndevs < devs_min) {
		ret = -ENOSPC;
		goto error;
3838
	}
3839

3840 3841 3842 3843 3844 3845 3846 3847
	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;
3848

D
David Woodhouse 已提交
3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864
	/*
	 * 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;
	}
3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885

	/*
	 * 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;
	}

3886
	do_div(stripe_size, dev_stripes);
3887 3888

	/* align to BTRFS_STRIPE_LEN */
D
David Woodhouse 已提交
3889 3890
	do_div(stripe_size, raid_stripe_len);
	stripe_size *= raid_stripe_len;
3891 3892 3893 3894 3895 3896 3897

	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
	if (!map) {
		ret = -ENOMEM;
		goto error;
	}
	map->num_stripes = num_stripes;
3898

3899 3900 3901 3902 3903 3904
	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;
3905 3906
		}
	}
Y
Yan Zheng 已提交
3907
	map->sector_size = extent_root->sectorsize;
D
David Woodhouse 已提交
3908 3909 3910
	map->stripe_len = raid_stripe_len;
	map->io_align = raid_stripe_len;
	map->io_width = raid_stripe_len;
Y
Yan Zheng 已提交
3911 3912
	map->type = type;
	map->sub_stripes = sub_stripes;
3913

Y
Yan Zheng 已提交
3914
	*map_ret = map;
D
David Woodhouse 已提交
3915
	num_bytes = stripe_size * data_stripes;
3916

3917 3918
	*stripe_size_out = stripe_size;
	*num_bytes_out = num_bytes;
3919

3920
	trace_btrfs_chunk_alloc(info->chunk_root, map, start, num_bytes);
3921

3922
	em = alloc_extent_map();
Y
Yan Zheng 已提交
3923
	if (!em) {
3924 3925
		ret = -ENOMEM;
		goto error;
3926
	}
Y
Yan Zheng 已提交
3927 3928
	em->bdev = (struct block_device *)map;
	em->start = start;
3929
	em->len = num_bytes;
Y
Yan Zheng 已提交
3930 3931
	em->block_start = 0;
	em->block_len = em->len;
3932

Y
Yan Zheng 已提交
3933
	em_tree = &extent_root->fs_info->mapping_tree.map_tree;
3934
	write_lock(&em_tree->lock);
Y
Yan Zheng 已提交
3935
	ret = add_extent_mapping(em_tree, em);
3936
	write_unlock(&em_tree->lock);
3937 3938
	if (ret) {
		free_extent_map(em);
3939
		goto error;
3940
	}
3941

3942 3943 3944 3945 3946 3947
	for (i = 0; i < map->num_stripes; ++i) {
		struct btrfs_device *device;
		u64 dev_offset;

		device = map->stripes[i].dev;
		dev_offset = map->stripes[i].physical;
3948 3949

		ret = btrfs_alloc_dev_extent(trans, device,
Y
Yan Zheng 已提交
3950 3951
				info->chunk_root->root_key.objectid,
				BTRFS_FIRST_CHUNK_TREE_OBJECTID,
3952
				start, dev_offset, stripe_size);
3953 3954 3955 3956 3957 3958 3959 3960 3961 3962
		if (ret)
			goto error_dev_extent;
	}

	ret = btrfs_make_block_group(trans, extent_root, 0, type,
				     BTRFS_FIRST_CHUNK_TREE_OBJECTID,
				     start, num_bytes);
	if (ret) {
		i = map->num_stripes - 1;
		goto error_dev_extent;
Y
Yan Zheng 已提交
3963 3964
	}

3965
	free_extent_map(em);
D
David Woodhouse 已提交
3966 3967
	check_raid56_incompat_flag(extent_root->fs_info, type);

3968
	kfree(devices_info);
Y
Yan Zheng 已提交
3969
	return 0;
3970

3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982
error_dev_extent:
	for (; i >= 0; i--) {
		struct btrfs_device *device;
		int err;

		device = map->stripes[i].dev;
		err = btrfs_free_dev_extent(trans, device, start);
		if (err) {
			btrfs_abort_transaction(trans, extent_root, err);
			break;
		}
	}
3983 3984 3985 3986 3987 3988 3989 3990
	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);
3991 3992 3993 3994
error:
	kfree(map);
	kfree(devices_info);
	return ret;
Y
Yan Zheng 已提交
3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019
}

static int __finish_chunk_alloc(struct btrfs_trans_handle *trans,
				struct btrfs_root *extent_root,
				struct map_lookup *map, u64 chunk_offset,
				u64 chunk_size, u64 stripe_size)
{
	u64 dev_offset;
	struct btrfs_key key;
	struct btrfs_root *chunk_root = extent_root->fs_info->chunk_root;
	struct btrfs_device *device;
	struct btrfs_chunk *chunk;
	struct btrfs_stripe *stripe;
	size_t item_size = btrfs_chunk_item_size(map->num_stripes);
	int index = 0;
	int ret;

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

	index = 0;
	while (index < map->num_stripes) {
		device = map->stripes[index].dev;
		device->bytes_used += stripe_size;
4020
		ret = btrfs_update_device(trans, device);
4021 4022
		if (ret)
			goto out_free;
Y
Yan Zheng 已提交
4023 4024 4025
		index++;
	}

4026 4027 4028 4029 4030
	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
4031 4032 4033 4034 4035
	index = 0;
	stripe = &chunk->stripe;
	while (index < map->num_stripes) {
		device = map->stripes[index].dev;
		dev_offset = map->stripes[index].physical;
4036

4037 4038 4039
		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
4040
		stripe++;
4041 4042 4043
		index++;
	}

Y
Yan Zheng 已提交
4044
	btrfs_set_stack_chunk_length(chunk, chunk_size);
4045
	btrfs_set_stack_chunk_owner(chunk, extent_root->root_key.objectid);
Y
Yan Zheng 已提交
4046 4047 4048 4049 4050
	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);
4051
	btrfs_set_stack_chunk_sector_size(chunk, extent_root->sectorsize);
Y
Yan Zheng 已提交
4052
	btrfs_set_stack_chunk_sub_stripes(chunk, map->sub_stripes);
4053

Y
Yan Zheng 已提交
4054 4055 4056
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.type = BTRFS_CHUNK_ITEM_KEY;
	key.offset = chunk_offset;
4057

Y
Yan Zheng 已提交
4058
	ret = btrfs_insert_item(trans, chunk_root, &key, chunk, item_size);
4059

4060 4061 4062 4063 4064
	if (ret == 0 && map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
		/*
		 * TODO: Cleanup of inserted chunk root in case of
		 * failure.
		 */
4065
		ret = btrfs_add_system_chunk(chunk_root, &key, chunk,
Y
Yan Zheng 已提交
4066
					     item_size);
4067
	}
4068

4069
out_free:
4070
	kfree(chunk);
4071
	return ret;
Y
Yan Zheng 已提交
4072
}
4073

Y
Yan Zheng 已提交
4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102
/*
 * Chunk allocation falls into two parts. The first part does works
 * that make the new allocated chunk useable, but not do any operation
 * that modifies the chunk tree. The second part does the works that
 * require modifying the chunk tree. This division is important for the
 * bootstrap process of adding storage to a seed btrfs.
 */
int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
		      struct btrfs_root *extent_root, u64 type)
{
	u64 chunk_offset;
	u64 chunk_size;
	u64 stripe_size;
	struct map_lookup *map;
	struct btrfs_root *chunk_root = extent_root->fs_info->chunk_root;
	int ret;

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

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

	ret = __finish_chunk_alloc(trans, extent_root, map, chunk_offset,
				   chunk_size, stripe_size);
4103 4104
	if (ret)
		return ret;
Y
Yan Zheng 已提交
4105 4106 4107
	return 0;
}

C
Chris Mason 已提交
4108
static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
Y
Yan Zheng 已提交
4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126
					 struct btrfs_root *root,
					 struct btrfs_device *device)
{
	u64 chunk_offset;
	u64 sys_chunk_offset;
	u64 chunk_size;
	u64 sys_chunk_size;
	u64 stripe_size;
	u64 sys_stripe_size;
	u64 alloc_profile;
	struct map_lookup *map;
	struct map_lookup *sys_map;
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct btrfs_root *extent_root = fs_info->extent_root;
	int ret;

	ret = find_next_chunk(fs_info->chunk_root,
			      BTRFS_FIRST_CHUNK_TREE_OBJECTID, &chunk_offset);
4127 4128
	if (ret)
		return ret;
Y
Yan Zheng 已提交
4129

4130
	alloc_profile = btrfs_get_alloc_profile(extent_root, 0);
Y
Yan Zheng 已提交
4131 4132
	ret = __btrfs_alloc_chunk(trans, extent_root, &map, &chunk_size,
				  &stripe_size, chunk_offset, alloc_profile);
4133 4134
	if (ret)
		return ret;
Y
Yan Zheng 已提交
4135 4136 4137

	sys_chunk_offset = chunk_offset + chunk_size;

4138
	alloc_profile = btrfs_get_alloc_profile(fs_info->chunk_root, 0);
Y
Yan Zheng 已提交
4139 4140 4141
	ret = __btrfs_alloc_chunk(trans, extent_root, &sys_map,
				  &sys_chunk_size, &sys_stripe_size,
				  sys_chunk_offset, alloc_profile);
4142 4143 4144 4145
	if (ret) {
		btrfs_abort_transaction(trans, root, ret);
		goto out;
	}
Y
Yan Zheng 已提交
4146 4147

	ret = btrfs_add_device(trans, fs_info->chunk_root, device);
4148 4149 4150 4151
	if (ret) {
		btrfs_abort_transaction(trans, root, ret);
		goto out;
	}
Y
Yan Zheng 已提交
4152 4153 4154 4155 4156 4157 4158 4159 4160

	/*
	 * Modifying chunk tree needs allocating new blocks from both
	 * system block group and metadata block group. So we only can
	 * do operations require modifying the chunk tree after both
	 * block groups were created.
	 */
	ret = __finish_chunk_alloc(trans, extent_root, map, chunk_offset,
				   chunk_size, stripe_size);
4161 4162 4163 4164
	if (ret) {
		btrfs_abort_transaction(trans, root, ret);
		goto out;
	}
Y
Yan Zheng 已提交
4165 4166 4167 4168

	ret = __finish_chunk_alloc(trans, extent_root, sys_map,
				   sys_chunk_offset, sys_chunk_size,
				   sys_stripe_size);
4169
	if (ret)
4170
		btrfs_abort_transaction(trans, root, ret);
4171

4172
out:
4173 4174

	return ret;
Y
Yan Zheng 已提交
4175 4176 4177 4178 4179 4180 4181 4182 4183 4184
}

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;

4185
	read_lock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
4186
	em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
4187
	read_unlock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
4188 4189 4190
	if (!em)
		return 1;

4191 4192 4193 4194 4195
	if (btrfs_test_opt(root, DEGRADED)) {
		free_extent_map(em);
		return 0;
	}

Y
Yan Zheng 已提交
4196 4197 4198 4199 4200 4201 4202
	map = (struct map_lookup *)em->bdev;
	for (i = 0; i < map->num_stripes; i++) {
		if (!map->stripes[i].dev->writeable) {
			readonly = 1;
			break;
		}
	}
4203
	free_extent_map(em);
Y
Yan Zheng 已提交
4204
	return readonly;
4205 4206 4207 4208
}

void btrfs_mapping_init(struct btrfs_mapping_tree *tree)
{
4209
	extent_map_tree_init(&tree->map_tree);
4210 4211 4212 4213 4214 4215
}

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

C
Chris Mason 已提交
4216
	while (1) {
4217
		write_lock(&tree->map_tree.lock);
4218 4219 4220
		em = lookup_extent_mapping(&tree->map_tree, 0, (u64)-1);
		if (em)
			remove_extent_mapping(&tree->map_tree, em);
4221
		write_unlock(&tree->map_tree.lock);
4222 4223 4224 4225 4226 4227 4228 4229 4230 4231
		if (!em)
			break;
		kfree(em->bdev);
		/* once for us */
		free_extent_map(em);
		/* once for the tree */
		free_extent_map(em);
	}
}

4232
int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len)
4233
{
4234
	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
4235 4236 4237 4238 4239
	struct extent_map *em;
	struct map_lookup *map;
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	int ret;

4240
	read_lock(&em_tree->lock);
4241
	em = lookup_extent_mapping(em_tree, logical, len);
4242
	read_unlock(&em_tree->lock);
4243 4244 4245 4246 4247 4248
	BUG_ON(!em);

	BUG_ON(em->start > logical || em->start + em->len < logical);
	map = (struct map_lookup *)em->bdev;
	if (map->type & (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1))
		ret = map->num_stripes;
C
Chris Mason 已提交
4249 4250
	else if (map->type & BTRFS_BLOCK_GROUP_RAID10)
		ret = map->sub_stripes;
D
David Woodhouse 已提交
4251 4252 4253 4254
	else if (map->type & BTRFS_BLOCK_GROUP_RAID5)
		ret = 2;
	else if (map->type & BTRFS_BLOCK_GROUP_RAID6)
		ret = 3;
4255 4256 4257
	else
		ret = 1;
	free_extent_map(em);
4258 4259 4260 4261 4262 4263

	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);

4264 4265 4266
	return ret;
}

D
David Woodhouse 已提交
4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312
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;
}

4313 4314 4315
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)
4316 4317
{
	int i;
4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341
	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;
		}
4342
	}
4343

4344 4345 4346 4347 4348 4349
	/* 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
4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378
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;
			}
		}
	}
}

4379
static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
4380
			     u64 logical, u64 *length,
4381
			     struct btrfs_bio **bbio_ret,
D
David Woodhouse 已提交
4382
			     int mirror_num, u64 **raid_map_ret)
4383 4384 4385
{
	struct extent_map *em;
	struct map_lookup *map;
4386
	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
4387 4388
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	u64 offset;
4389
	u64 stripe_offset;
4390
	u64 stripe_end_offset;
4391
	u64 stripe_nr;
4392 4393
	u64 stripe_nr_orig;
	u64 stripe_nr_end;
D
David Woodhouse 已提交
4394 4395
	u64 stripe_len;
	u64 *raid_map = NULL;
4396
	int stripe_index;
4397
	int i;
L
Li Zefan 已提交
4398
	int ret = 0;
4399
	int num_stripes;
4400
	int max_errors = 0;
4401
	struct btrfs_bio *bbio = NULL;
4402 4403 4404
	struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
	int dev_replace_is_ongoing = 0;
	int num_alloc_stripes;
4405 4406
	int patch_the_first_stripe_for_dev_replace = 0;
	u64 physical_to_patch_in_first_stripe = 0;
D
David Woodhouse 已提交
4407
	u64 raid56_full_stripe_start = (u64)-1;
4408

4409
	read_lock(&em_tree->lock);
4410
	em = lookup_extent_mapping(em_tree, logical, *length);
4411
	read_unlock(&em_tree->lock);
4412

4413
	if (!em) {
D
Daniel J Blueman 已提交
4414
		printk(KERN_CRIT "btrfs: unable to find logical %llu len %llu\n",
C
Chris Mason 已提交
4415 4416
		       (unsigned long long)logical,
		       (unsigned long long)*length);
4417
		BUG();
4418
	}
4419 4420 4421 4422

	BUG_ON(em->start > logical || em->start + em->len < logical);
	map = (struct map_lookup *)em->bdev;
	offset = logical - em->start;
4423

D
David Woodhouse 已提交
4424 4425 4426 4427
	if (mirror_num > map->num_stripes)
		mirror_num = 0;

	stripe_len = map->stripe_len;
4428 4429 4430 4431 4432
	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
4433
	do_div(stripe_nr, stripe_len);
4434

D
David Woodhouse 已提交
4435
	stripe_offset = stripe_nr * stripe_len;
4436 4437 4438 4439 4440
	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
4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459
	/* 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;
		}
4460
		*length = min_t(u64, em->len - offset, *length);
D
David Woodhouse 已提交
4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474
	} 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);
4475 4476 4477
	} else {
		*length = em->len - offset;
	}
4478

D
David Woodhouse 已提交
4479 4480
	/* This is for when we're called from btrfs_merge_bio_hook() and all
	   it cares about is the length */
4481
	if (!bbio_ret)
4482 4483
		goto out;

4484 4485 4486 4487 4488
	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);

4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512
	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
4513
			     logical, &tmp_length, &tmp_bbio, 0, NULL);
4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 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
		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;
	}

4571
	num_stripes = 1;
4572
	stripe_index = 0;
4573
	stripe_nr_orig = stripe_nr;
4574
	stripe_nr_end = ALIGN(offset + *length, map->stripe_len);
4575 4576 4577
	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
4578

4579 4580 4581 4582 4583 4584
	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) {
4585
		if (rw & (REQ_WRITE | REQ_DISCARD | REQ_GET_READ_MIRRORS))
4586
			num_stripes = map->num_stripes;
4587
		else if (mirror_num)
4588
			stripe_index = mirror_num - 1;
4589
		else {
4590
			stripe_index = find_live_mirror(fs_info, map, 0,
4591
					    map->num_stripes,
4592 4593
					    current->pid % map->num_stripes,
					    dev_replace_is_ongoing);
4594
			mirror_num = stripe_index + 1;
4595
		}
4596

4597
	} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
4598
		if (rw & (REQ_WRITE | REQ_DISCARD | REQ_GET_READ_MIRRORS)) {
4599
			num_stripes = map->num_stripes;
4600
		} else if (mirror_num) {
4601
			stripe_index = mirror_num - 1;
4602 4603 4604
		} else {
			mirror_num = 1;
		}
4605

C
Chris Mason 已提交
4606 4607 4608 4609 4610 4611
	} 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;

4612
		if (rw & (REQ_WRITE | REQ_GET_READ_MIRRORS))
4613
			num_stripes = map->sub_stripes;
4614 4615 4616 4617
		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
4618 4619
		else if (mirror_num)
			stripe_index += mirror_num - 1;
4620
		else {
J
Jan Schmidt 已提交
4621
			int old_stripe_index = stripe_index;
4622 4623
			stripe_index = find_live_mirror(fs_info, map,
					      stripe_index,
4624
					      map->sub_stripes, stripe_index +
4625 4626
					      current->pid % map->sub_stripes,
					      dev_replace_is_ongoing);
J
Jan Schmidt 已提交
4627
			mirror_num = stripe_index - old_stripe_index + 1;
4628
		}
D
David Woodhouse 已提交
4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687

	} 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);

			raid_map = kmalloc(sizeof(u64) * num_stripes,
					   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);
		}
4688 4689 4690 4691 4692 4693 4694
	} 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);
4695
		mirror_num = stripe_index + 1;
4696
	}
4697
	BUG_ON(stripe_index >= map->num_stripes);
4698

4699
	num_alloc_stripes = num_stripes;
4700 4701 4702 4703 4704 4705
	if (dev_replace_is_ongoing) {
		if (rw & (REQ_WRITE | REQ_DISCARD))
			num_alloc_stripes <<= 1;
		if (rw & REQ_GET_READ_MIRRORS)
			num_alloc_stripes++;
	}
4706
	bbio = kzalloc(btrfs_bio_size(num_alloc_stripes), GFP_NOFS);
L
Li Zefan 已提交
4707 4708 4709 4710 4711 4712
	if (!bbio) {
		ret = -ENOMEM;
		goto out;
	}
	atomic_set(&bbio->error, 0);

4713
	if (rw & REQ_DISCARD) {
4714 4715 4716 4717
		int factor = 0;
		int sub_stripes = 0;
		u64 stripes_per_dev = 0;
		u32 remaining_stripes = 0;
L
Liu Bo 已提交
4718
		u32 last_stripe = 0;
4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731

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

			factor = map->num_stripes / sub_stripes;
			stripes_per_dev = div_u64_rem(stripe_nr_end -
						      stripe_nr_orig,
						      factor,
						      &remaining_stripes);
L
Liu Bo 已提交
4732 4733
			div_u64_rem(stripe_nr_end - 1, factor, &last_stripe);
			last_stripe *= sub_stripes;
4734 4735
		}

4736
		for (i = 0; i < num_stripes; i++) {
4737
			bbio->stripes[i].physical =
4738 4739
				map->stripes[stripe_index].physical +
				stripe_offset + stripe_nr * map->stripe_len;
4740
			bbio->stripes[i].dev = map->stripes[stripe_index].dev;
4741

4742 4743 4744 4745
			if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
					 BTRFS_BLOCK_GROUP_RAID10)) {
				bbio->stripes[i].length = stripes_per_dev *
							  map->stripe_len;
L
Liu Bo 已提交
4746

4747 4748 4749
				if (i / sub_stripes < remaining_stripes)
					bbio->stripes[i].length +=
						map->stripe_len;
L
Liu Bo 已提交
4750 4751 4752 4753 4754 4755 4756 4757 4758

				/*
				 * Special for the first stripe and
				 * the last stripe:
				 *
				 * |-------|...|-------|
				 *     |----------|
				 *    off     end_off
				 */
4759
				if (i < sub_stripes)
4760
					bbio->stripes[i].length -=
4761
						stripe_offset;
L
Liu Bo 已提交
4762 4763 4764 4765

				if (stripe_index >= last_stripe &&
				    stripe_index <= (last_stripe +
						     sub_stripes - 1))
4766
					bbio->stripes[i].length -=
4767
						stripe_end_offset;
L
Liu Bo 已提交
4768

4769 4770
				if (i == sub_stripes - 1)
					stripe_offset = 0;
4771
			} else
4772
				bbio->stripes[i].length = *length;
4773 4774 4775 4776 4777 4778 4779 4780 4781 4782

			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++) {
4783
			bbio->stripes[i].physical =
4784 4785 4786
				map->stripes[stripe_index].physical +
				stripe_offset +
				stripe_nr * map->stripe_len;
4787
			bbio->stripes[i].dev =
4788
				map->stripes[stripe_index].dev;
4789
			stripe_index++;
4790
		}
4791
	}
L
Li Zefan 已提交
4792

4793
	if (rw & (REQ_WRITE | REQ_GET_READ_MIRRORS)) {
L
Li Zefan 已提交
4794 4795
		if (map->type & (BTRFS_BLOCK_GROUP_RAID1 |
				 BTRFS_BLOCK_GROUP_RAID10 |
D
David Woodhouse 已提交
4796
				 BTRFS_BLOCK_GROUP_RAID5 |
L
Li Zefan 已提交
4797 4798
				 BTRFS_BLOCK_GROUP_DUP)) {
			max_errors = 1;
D
David Woodhouse 已提交
4799 4800
		} else if (map->type & BTRFS_BLOCK_GROUP_RAID6) {
			max_errors = 2;
L
Li Zefan 已提交
4801
		}
4802
	}
L
Li Zefan 已提交
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
	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;
4837 4838 4839 4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882
	} 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++;
			}
		}
4883 4884
	}

L
Li Zefan 已提交
4885 4886 4887 4888
	*bbio_ret = bbio;
	bbio->num_stripes = num_stripes;
	bbio->max_errors = max_errors;
	bbio->mirror_num = mirror_num;
4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900

	/*
	 * 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
4901 4902 4903 4904
	if (raid_map) {
		sort_parity_stripes(bbio, raid_map);
		*raid_map_ret = raid_map;
	}
4905
out:
4906 4907
	if (dev_replace_is_ongoing)
		btrfs_dev_replace_unlock(dev_replace);
4908
	free_extent_map(em);
L
Li Zefan 已提交
4909
	return ret;
4910 4911
}

4912
int btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
4913
		      u64 logical, u64 *length,
4914
		      struct btrfs_bio **bbio_ret, int mirror_num)
4915
{
4916
	return __btrfs_map_block(fs_info, rw, logical, length, bbio_ret,
D
David Woodhouse 已提交
4917
				 mirror_num, NULL);
4918 4919
}

Y
Yan Zheng 已提交
4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930
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
4931
	u64 rmap_len;
Y
Yan Zheng 已提交
4932 4933
	int i, j, nr = 0;

4934
	read_lock(&em_tree->lock);
Y
Yan Zheng 已提交
4935
	em = lookup_extent_mapping(em_tree, chunk_start, 1);
4936
	read_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
4937 4938 4939 4940 4941

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

	length = em->len;
D
David Woodhouse 已提交
4942 4943
	rmap_len = map->stripe_len;

Y
Yan Zheng 已提交
4944 4945 4946 4947
	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
4948 4949 4950 4951 4952
	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
4953 4954

	buf = kzalloc(sizeof(u64) * map->num_stripes, GFP_NOFS);
4955
	BUG_ON(!buf); /* -ENOMEM */
Y
Yan Zheng 已提交
4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971

	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
4972 4973 4974 4975 4976
		} /* 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;
4977
		WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
4978 4979 4980 4981
		for (j = 0; j < nr; j++) {
			if (buf[j] == bytenr)
				break;
		}
4982 4983
		if (j == nr) {
			WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
4984
			buf[nr++] = bytenr;
4985
		}
Y
Yan Zheng 已提交
4986 4987 4988 4989
	}

	*logical = buf;
	*naddrs = nr;
D
David Woodhouse 已提交
4990
	*stripe_len = rmap_len;
Y
Yan Zheng 已提交
4991 4992 4993

	free_extent_map(em);
	return 0;
4994 4995
}

4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020
static void *merge_stripe_index_into_bio_private(void *bi_private,
						 unsigned int stripe_index)
{
	/*
	 * with single, dup, RAID0, RAID1 and RAID10, stripe_index is
	 * at most 1.
	 * The alternative solution (instead of stealing bits from the
	 * pointer) would be to allocate an intermediate structure
	 * that contains the old private pointer plus the stripe_index.
	 */
	BUG_ON((((uintptr_t)bi_private) & 3) != 0);
	BUG_ON(stripe_index > 3);
	return (void *)(((uintptr_t)bi_private) | stripe_index);
}

static struct btrfs_bio *extract_bbio_from_bio_private(void *bi_private)
{
	return (struct btrfs_bio *)(((uintptr_t)bi_private) & ~((uintptr_t)3));
}

static unsigned int extract_stripe_index_from_bio_private(void *bi_private)
{
	return (unsigned int)((uintptr_t)bi_private) & 3;
}

5021
static void btrfs_end_bio(struct bio *bio, int err)
5022
{
5023
	struct btrfs_bio *bbio = extract_bbio_from_bio_private(bio->bi_private);
5024
	int is_orig_bio = 0;
5025

5026
	if (err) {
5027
		atomic_inc(&bbio->error);
5028 5029 5030 5031 5032 5033 5034 5035
		if (err == -EIO || err == -EREMOTEIO) {
			unsigned int stripe_index =
				extract_stripe_index_from_bio_private(
					bio->bi_private);
			struct btrfs_device *dev;

			BUG_ON(stripe_index >= bbio->num_stripes);
			dev = bbio->stripes[stripe_index].dev;
5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047
			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);
			}
5048 5049
		}
	}
5050

5051
	if (bio == bbio->orig_bio)
5052 5053
		is_orig_bio = 1;

5054
	if (atomic_dec_and_test(&bbio->stripes_pending)) {
5055 5056
		if (!is_orig_bio) {
			bio_put(bio);
5057
			bio = bbio->orig_bio;
5058
		}
5059 5060
		bio->bi_private = bbio->private;
		bio->bi_end_io = bbio->end_io;
J
Jan Schmidt 已提交
5061 5062
		bio->bi_bdev = (struct block_device *)
					(unsigned long)bbio->mirror_num;
5063
		/* only send an error to the higher layers if it is
D
David Woodhouse 已提交
5064
		 * beyond the tolerance of the btrfs bio
5065
		 */
5066
		if (atomic_read(&bbio->error) > bbio->max_errors) {
5067
			err = -EIO;
5068
		} else {
5069 5070 5071 5072 5073
			/*
			 * this bio is actually up to date, we didn't
			 * go over the max number of errors
			 */
			set_bit(BIO_UPTODATE, &bio->bi_flags);
5074
			err = 0;
5075
		}
5076
		kfree(bbio);
5077 5078

		bio_endio(bio, err);
5079
	} else if (!is_orig_bio) {
5080 5081 5082 5083
		bio_put(bio);
	}
}

5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097
struct async_sched {
	struct bio *bio;
	int rw;
	struct btrfs_fs_info *info;
	struct btrfs_work work;
};

/*
 * see run_scheduled_bios for a description of why bios are collected for
 * async submit.
 *
 * This will add one bio to the pending list for a device and make sure
 * the work struct is scheduled.
 */
D
David Woodhouse 已提交
5098
noinline void btrfs_schedule_bio(struct btrfs_root *root,
5099 5100
				 struct btrfs_device *device,
				 int rw, struct bio *bio)
5101 5102
{
	int should_queue = 1;
5103
	struct btrfs_pending_bios *pending_bios;
5104

D
David Woodhouse 已提交
5105 5106 5107 5108 5109
	if (device->missing || !device->bdev) {
		bio_endio(bio, -EIO);
		return;
	}

5110
	/* don't bother with additional async steps for reads, right now */
5111
	if (!(rw & REQ_WRITE)) {
5112
		bio_get(bio);
5113
		btrfsic_submit_bio(rw, bio);
5114
		bio_put(bio);
5115
		return;
5116 5117 5118
	}

	/*
5119
	 * nr_async_bios allows us to reliably return congestion to the
5120 5121 5122 5123
	 * 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
	 */
5124
	atomic_inc(&root->fs_info->nr_async_bios);
5125
	WARN_ON(bio->bi_next);
5126 5127 5128 5129
	bio->bi_next = NULL;
	bio->bi_rw |= rw;

	spin_lock(&device->io_lock);
5130
	if (bio->bi_rw & REQ_SYNC)
5131 5132 5133
		pending_bios = &device->pending_sync_bios;
	else
		pending_bios = &device->pending_bios;
5134

5135 5136
	if (pending_bios->tail)
		pending_bios->tail->bi_next = bio;
5137

5138 5139 5140
	pending_bios->tail = bio;
	if (!pending_bios->head)
		pending_bios->head = bio;
5141 5142 5143 5144 5145 5146
	if (device->running_pending)
		should_queue = 0;

	spin_unlock(&device->io_lock);

	if (should_queue)
5147 5148
		btrfs_queue_worker(&root->fs_info->submit_workers,
				   &device->work);
5149 5150
}

5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197
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);
	unsigned short max_sectors = queue_max_sectors(q);
	struct bvec_merge_data bvm = {
		.bi_bdev = bdev,
		.bi_sector = sector,
		.bi_rw = bio->bi_rw,
	};

	if (bio->bi_vcnt == 0) {
		WARN_ON(1);
		return 1;
	}

	prev = &bio->bi_io_vec[bio->bi_vcnt - 1];
	if ((bio->bi_size >> 9) > max_sectors)
		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;
	bio->bi_private = merge_stripe_index_into_bio_private(
			bio->bi_private, (unsigned int)dev_nr);
	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);
M
Masanari Iida 已提交
5198
		pr_debug("btrfs_map_bio: rw %d, sector=%llu, dev=%lu "
5199 5200 5201 5202 5203 5204 5205 5206
			 "(%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;
	if (async)
D
David Woodhouse 已提交
5207
		btrfs_schedule_bio(root, dev, rw, bio);
5208 5209 5210 5211 5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257
	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;
		bio->bi_bdev = (struct block_device *)
			(unsigned long)bbio->mirror_num;
		bio->bi_sector = logical >> 9;
		kfree(bbio);
		bio_endio(bio, -EIO);
	}
}

5258
int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
5259
		  int mirror_num, int async_submit)
5260 5261
{
	struct btrfs_device *dev;
5262
	struct bio *first_bio = bio;
5263
	u64 logical = (u64)bio->bi_sector << 9;
5264 5265
	u64 length = 0;
	u64 map_length;
D
David Woodhouse 已提交
5266
	u64 *raid_map = NULL;
5267
	int ret;
5268 5269
	int dev_nr = 0;
	int total_devs = 1;
5270
	struct btrfs_bio *bbio = NULL;
5271

5272
	length = bio->bi_size;
5273
	map_length = length;
5274

D
David Woodhouse 已提交
5275 5276 5277
	ret = __btrfs_map_block(root->fs_info, rw, logical, &map_length, &bbio,
			      mirror_num, &raid_map);
	if (ret) /* -ENOMEM */
5278
		return ret;
5279

5280
	total_devs = bbio->num_stripes;
D
David Woodhouse 已提交
5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297 5298
	bbio->orig_bio = first_bio;
	bbio->private = first_bio->bi_private;
	bbio->end_io = first_bio->bi_end_io;
	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) {
			return raid56_parity_write(root, bio, bbio,
						   raid_map, map_length);
		} else {
			return raid56_parity_recover(root, bio, bbio,
						     raid_map, map_length,
						     mirror_num);
		}
	}

5299
	if (map_length < length) {
D
Daniel J Blueman 已提交
5300
		printk(KERN_CRIT "btrfs: mapping failed logical %llu bio len %llu "
C
Chris Mason 已提交
5301 5302 5303
		       "len %llu\n", (unsigned long long)logical,
		       (unsigned long long)length,
		       (unsigned long long)map_length);
5304 5305
		BUG();
	}
5306

C
Chris Mason 已提交
5307
	while (dev_nr < total_devs) {
5308 5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327
		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;
		}

5328 5329
		if (dev_nr < total_devs - 1) {
			bio = bio_clone(first_bio, GFP_NOFS);
5330
			BUG_ON(!bio); /* -ENOMEM */
5331 5332
		} else {
			bio = first_bio;
5333
		}
5334 5335 5336 5337

		submit_stripe_bio(root, bbio, bio,
				  bbio->stripes[dev_nr].physical, dev_nr, rw,
				  async_submit);
5338 5339
		dev_nr++;
	}
5340 5341 5342
	return 0;
}

5343
struct btrfs_device *btrfs_find_device(struct btrfs_fs_info *fs_info, u64 devid,
Y
Yan Zheng 已提交
5344
				       u8 *uuid, u8 *fsid)
5345
{
Y
Yan Zheng 已提交
5346 5347 5348
	struct btrfs_device *device;
	struct btrfs_fs_devices *cur_devices;

5349
	cur_devices = fs_info->fs_devices;
Y
Yan Zheng 已提交
5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360
	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;
5361 5362
}

5363 5364 5365 5366 5367 5368 5369
static struct btrfs_device *add_missing_dev(struct btrfs_root *root,
					    u64 devid, u8 *dev_uuid)
{
	struct btrfs_device *device;
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;

	device = kzalloc(sizeof(*device), GFP_NOFS);
5370 5371
	if (!device)
		return NULL;
5372 5373 5374 5375
	list_add(&device->dev_list,
		 &fs_devices->devices);
	device->dev_root = root->fs_info->dev_root;
	device->devid = devid;
5376
	device->work.func = pending_bios_fn;
Y
Yan Zheng 已提交
5377
	device->fs_devices = fs_devices;
5378
	device->missing = 1;
5379
	fs_devices->num_devices++;
5380
	fs_devices->missing_devices++;
5381
	spin_lock_init(&device->io_lock);
5382
	INIT_LIST_HEAD(&device->dev_alloc_list);
5383 5384 5385 5386
	memcpy(device->uuid, dev_uuid, BTRFS_UUID_SIZE);
	return device;
}

5387 5388 5389 5390 5391 5392 5393 5394 5395 5396
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;
5397
	u8 uuid[BTRFS_UUID_SIZE];
5398
	int num_stripes;
5399
	int ret;
5400
	int i;
5401

5402 5403
	logical = key->offset;
	length = btrfs_chunk_length(leaf, chunk);
5404

5405
	read_lock(&map_tree->map_tree.lock);
5406
	em = lookup_extent_mapping(&map_tree->map_tree, logical, 1);
5407
	read_unlock(&map_tree->map_tree.lock);
5408 5409 5410 5411 5412 5413 5414 5415 5416

	/* 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);
	}

5417
	em = alloc_extent_map();
5418 5419
	if (!em)
		return -ENOMEM;
5420 5421
	num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
5422 5423 5424 5425 5426 5427 5428 5429
	if (!map) {
		free_extent_map(em);
		return -ENOMEM;
	}

	em->bdev = (struct block_device *)map;
	em->start = logical;
	em->len = length;
5430
	em->orig_start = 0;
5431
	em->block_start = 0;
C
Chris Mason 已提交
5432
	em->block_len = em->len;
5433

5434 5435 5436 5437 5438 5439
	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
5440
	map->sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
5441 5442 5443 5444
	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);
5445 5446 5447
		read_extent_buffer(leaf, uuid, (unsigned long)
				   btrfs_stripe_dev_uuid_nr(chunk, i),
				   BTRFS_UUID_SIZE);
5448 5449
		map->stripes[i].dev = btrfs_find_device(root->fs_info, devid,
							uuid, NULL);
5450
		if (!map->stripes[i].dev && !btrfs_test_opt(root, DEGRADED)) {
5451 5452 5453 5454
			kfree(map);
			free_extent_map(em);
			return -EIO;
		}
5455 5456 5457 5458 5459 5460 5461 5462 5463 5464
		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;
5465 5466
	}

5467
	write_lock(&map_tree->map_tree.lock);
5468
	ret = add_extent_mapping(&map_tree->map_tree, em);
5469
	write_unlock(&map_tree->map_tree.lock);
5470
	BUG_ON(ret); /* Tree corruption */
5471 5472 5473 5474 5475
	free_extent_map(em);

	return 0;
}

5476
static void fill_device_from_item(struct extent_buffer *leaf,
5477 5478 5479 5480 5481 5482
				 struct btrfs_dev_item *dev_item,
				 struct btrfs_device *device)
{
	unsigned long ptr;

	device->devid = btrfs_device_id(leaf, dev_item);
5483 5484
	device->disk_total_bytes = btrfs_device_total_bytes(leaf, dev_item);
	device->total_bytes = device->disk_total_bytes;
5485 5486 5487 5488 5489
	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);
5490
	WARN_ON(device->devid == BTRFS_DEV_REPLACE_DEVID);
5491
	device->is_tgtdev_for_dev_replace = 0;
5492 5493

	ptr = (unsigned long)btrfs_device_uuid(dev_item);
5494
	read_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
5495 5496
}

Y
Yan Zheng 已提交
5497 5498 5499 5500 5501
static int open_seed_devices(struct btrfs_root *root, u8 *fsid)
{
	struct btrfs_fs_devices *fs_devices;
	int ret;

5502
	BUG_ON(!mutex_is_locked(&uuid_mutex));
Y
Yan Zheng 已提交
5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517

	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
5518 5519 5520 5521

	fs_devices = clone_fs_devices(fs_devices);
	if (IS_ERR(fs_devices)) {
		ret = PTR_ERR(fs_devices);
Y
Yan Zheng 已提交
5522 5523 5524
		goto out;
	}

5525
	ret = __btrfs_open_devices(fs_devices, FMODE_READ,
5526
				   root->fs_info->bdev_holder);
5527 5528
	if (ret) {
		free_fs_devices(fs_devices);
Y
Yan Zheng 已提交
5529
		goto out;
5530
	}
Y
Yan Zheng 已提交
5531 5532 5533

	if (!fs_devices->seeding) {
		__btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
5534
		free_fs_devices(fs_devices);
Y
Yan Zheng 已提交
5535 5536 5537 5538 5539 5540 5541 5542 5543 5544
		ret = -EINVAL;
		goto out;
	}

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

5545
static int read_one_dev(struct btrfs_root *root,
5546 5547 5548 5549 5550 5551
			struct extent_buffer *leaf,
			struct btrfs_dev_item *dev_item)
{
	struct btrfs_device *device;
	u64 devid;
	int ret;
Y
Yan Zheng 已提交
5552
	u8 fs_uuid[BTRFS_UUID_SIZE];
5553 5554
	u8 dev_uuid[BTRFS_UUID_SIZE];

5555
	devid = btrfs_device_id(leaf, dev_item);
5556 5557 5558
	read_extent_buffer(leaf, dev_uuid,
			   (unsigned long)btrfs_device_uuid(dev_item),
			   BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
5559 5560 5561 5562 5563 5564
	read_extent_buffer(leaf, fs_uuid,
			   (unsigned long)btrfs_device_fsid(dev_item),
			   BTRFS_UUID_SIZE);

	if (memcmp(fs_uuid, root->fs_info->fsid, BTRFS_UUID_SIZE)) {
		ret = open_seed_devices(root, fs_uuid);
Y
Yan Zheng 已提交
5565
		if (ret && !btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
5566 5567 5568
			return ret;
	}

5569
	device = btrfs_find_device(root->fs_info, devid, dev_uuid, fs_uuid);
Y
Yan Zheng 已提交
5570
	if (!device || !device->bdev) {
Y
Yan Zheng 已提交
5571
		if (!btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
5572 5573 5574
			return -EIO;

		if (!device) {
C
Chris Mason 已提交
5575 5576
			printk(KERN_WARNING "warning devid %llu missing\n",
			       (unsigned long long)devid);
Y
Yan Zheng 已提交
5577 5578 5579
			device = add_missing_dev(root, devid, dev_uuid);
			if (!device)
				return -ENOMEM;
5580 5581 5582 5583 5584 5585 5586 5587 5588
		} 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
5589 5590 5591 5592 5593 5594 5595 5596
		}
	}

	if (device->fs_devices != root->fs_info->fs_devices) {
		BUG_ON(device->writeable);
		if (device->generation !=
		    btrfs_device_generation(leaf, dev_item))
			return -EINVAL;
5597
	}
5598 5599 5600

	fill_device_from_item(leaf, dev_item, device);
	device->dev_root = root->fs_info->dev_root;
5601
	device->in_fs_metadata = 1;
5602
	if (device->writeable && !device->is_tgtdev_for_dev_replace) {
Y
Yan Zheng 已提交
5603
		device->fs_devices->total_rw_bytes += device->total_bytes;
5604 5605 5606 5607 5608
		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);
	}
5609 5610 5611 5612
	ret = 0;
	return ret;
}

Y
Yan Zheng 已提交
5613
int btrfs_read_sys_array(struct btrfs_root *root)
5614
{
5615
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
5616
	struct extent_buffer *sb;
5617 5618
	struct btrfs_disk_key *disk_key;
	struct btrfs_chunk *chunk;
5619 5620 5621
	u8 *ptr;
	unsigned long sb_ptr;
	int ret = 0;
5622 5623 5624 5625
	u32 num_stripes;
	u32 array_size;
	u32 len = 0;
	u32 cur;
5626
	struct btrfs_key key;
5627

Y
Yan Zheng 已提交
5628
	sb = btrfs_find_create_tree_block(root, BTRFS_SUPER_INFO_OFFSET,
5629 5630 5631 5632
					  BTRFS_SUPER_INFO_SIZE);
	if (!sb)
		return -ENOMEM;
	btrfs_set_buffer_uptodate(sb);
5633
	btrfs_set_buffer_lockdep_class(root->root_key.objectid, sb, 0);
5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646
	/*
	 * 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)
5647
		SetPageUptodate(sb->pages[0]);
5648

5649
	write_extent_buffer(sb, super_copy, 0, BTRFS_SUPER_INFO_SIZE);
5650 5651 5652 5653 5654 5655 5656 5657 5658 5659
	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);

5660
		len = sizeof(*disk_key); ptr += len;
5661 5662 5663
		sb_ptr += len;
		cur += len;

5664
		if (key.type == BTRFS_CHUNK_ITEM_KEY) {
5665
			chunk = (struct btrfs_chunk *)sb_ptr;
5666
			ret = read_one_chunk(root, &key, sb, chunk);
5667 5668
			if (ret)
				break;
5669 5670 5671
			num_stripes = btrfs_chunk_num_stripes(sb, chunk);
			len = btrfs_chunk_item_size(num_stripes);
		} else {
5672 5673
			ret = -EIO;
			break;
5674 5675 5676 5677 5678
		}
		ptr += len;
		sb_ptr += len;
		cur += len;
	}
5679
	free_extent_buffer(sb);
5680
	return ret;
5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697
}

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;

5698 5699 5700
	mutex_lock(&uuid_mutex);
	lock_chunks(root);

5701 5702 5703 5704 5705 5706 5707 5708 5709
	/* first we search for all of the device items, and then we
	 * read in all of the chunk items.  This way we can create chunk
	 * mappings that reference all of the devices that are afound
	 */
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.offset = 0;
	key.type = 0;
again:
	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
5710 5711
	if (ret < 0)
		goto error;
C
Chris Mason 已提交
5712
	while (1) {
5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730
		leaf = path->nodes[0];
		slot = path->slots[0];
		if (slot >= btrfs_header_nritems(leaf)) {
			ret = btrfs_next_leaf(root, path);
			if (ret == 0)
				continue;
			if (ret < 0)
				goto error;
			break;
		}
		btrfs_item_key_to_cpu(leaf, &found_key, slot);
		if (key.objectid == BTRFS_DEV_ITEMS_OBJECTID) {
			if (found_key.objectid != BTRFS_DEV_ITEMS_OBJECTID)
				break;
			if (found_key.type == BTRFS_DEV_ITEM_KEY) {
				struct btrfs_dev_item *dev_item;
				dev_item = btrfs_item_ptr(leaf, slot,
						  struct btrfs_dev_item);
5731
				ret = read_one_dev(root, leaf, dev_item);
Y
Yan Zheng 已提交
5732 5733
				if (ret)
					goto error;
5734 5735 5736 5737 5738
			}
		} 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
5739 5740
			if (ret)
				goto error;
5741 5742 5743 5744 5745
		}
		path->slots[0]++;
	}
	if (key.objectid == BTRFS_DEV_ITEMS_OBJECTID) {
		key.objectid = 0;
5746
		btrfs_release_path(path);
5747 5748 5749 5750
		goto again;
	}
	ret = 0;
error:
5751 5752 5753
	unlock_chunks(root);
	mutex_unlock(&uuid_mutex);

Y
Yan Zheng 已提交
5754
	btrfs_free_path(path);
5755 5756
	return ret;
}
5757

5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796 5797 5798 5799 5800 5801 5802 5803 5804 5805 5806 5807 5808 5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828 5829 5830 5831 5832 5833 5834 5835 5836 5837 5838 5839 5840 5841 5842 5843 5844 5845
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) {
5846 5847
		printk_in_rcu(KERN_WARNING "btrfs: error %d while searching for dev_stats item for device %s!\n",
			      ret, rcu_str_deref(device->name));
5848 5849 5850 5851 5852 5853 5854 5855
		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) {
5856 5857
			printk_in_rcu(KERN_WARNING "btrfs: delete too small dev_stats item for device %s failed %d!\n",
				      rcu_str_deref(device->name), ret);
5858 5859 5860 5861 5862 5863 5864 5865 5866 5867 5868
			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) {
5869 5870
			printk_in_rcu(KERN_WARNING "btrfs: insert dev_stats item for device %s failed %d!\n",
				      rcu_str_deref(device->name), ret);
5871 5872 5873 5874 5875 5876 5877 5878 5879 5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898 5899 5900 5901 5902 5903 5904 5905 5906 5907 5908 5909 5910 5911
			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;
}

5912 5913 5914 5915 5916 5917 5918 5919
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);
}

void btrfs_dev_stat_print_on_error(struct btrfs_device *dev)
{
5920 5921
	if (!dev->dev_stats_valid)
		return;
5922
	printk_ratelimited_in_rcu(KERN_ERR
5923
			   "btrfs: bdev %s errs: wr %u, rd %u, flush %u, corrupt %u, gen %u\n",
5924
			   rcu_str_deref(dev->name),
5925 5926 5927 5928 5929 5930 5931 5932
			   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));
}
5933

5934 5935
static void btrfs_dev_stat_print_on_load(struct btrfs_device *dev)
{
5936 5937 5938 5939 5940 5941 5942 5943
	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 */

5944 5945
	printk_in_rcu(KERN_INFO "btrfs: bdev %s errs: wr %u, rd %u, flush %u, corrupt %u, gen %u\n",
	       rcu_str_deref(dev->name),
5946 5947 5948 5949 5950 5951 5952
	       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));
}

5953
int btrfs_get_dev_stats(struct btrfs_root *root,
5954
			struct btrfs_ioctl_get_dev_stats *stats)
5955 5956 5957 5958 5959 5960
{
	struct btrfs_device *dev;
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
	int i;

	mutex_lock(&fs_devices->device_list_mutex);
5961
	dev = btrfs_find_device(root->fs_info, stats->devid, NULL, NULL);
5962 5963 5964 5965 5966 5967
	mutex_unlock(&fs_devices->device_list_mutex);

	if (!dev) {
		printk(KERN_WARNING
		       "btrfs: get dev_stats failed, device not found\n");
		return -ENODEV;
5968 5969 5970 5971
	} else if (!dev->dev_stats_valid) {
		printk(KERN_WARNING
		       "btrfs: get dev_stats failed, not yet valid\n");
		return -ENODEV;
5972
	} else if (stats->flags & BTRFS_DEV_STATS_RESET) {
5973 5974 5975 5976 5977 5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988
		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;
}
5989 5990 5991 5992 5993 5994 5995 5996 5997 5998 5999 6000 6001 6002 6003 6004 6005 6006

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;
}