volumes.c 112.5 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/kthread.h>
27
#include <asm/div64.h>
C
Chris Mason 已提交
28
#include "compat.h"
29 30 31 32 33 34
#include "ctree.h"
#include "extent_map.h"
#include "disk-io.h"
#include "transaction.h"
#include "print-tree.h"
#include "volumes.h"
35
#include "async-thread.h"
36

Y
Yan Zheng 已提交
37 38 39 40 41
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);

42 43 44
static DEFINE_MUTEX(uuid_mutex);
static LIST_HEAD(fs_uuids);

45 46 47 48 49 50 51 52 53 54
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
55 56 57 58 59 60 61 62 63 64 65 66 67 68
static void free_fs_devices(struct btrfs_fs_devices *fs_devices)
{
	struct btrfs_device *device;
	WARN_ON(fs_devices->opened);
	while (!list_empty(&fs_devices->devices)) {
		device = list_entry(fs_devices->devices.next,
				    struct btrfs_device, dev_list);
		list_del(&device->dev_list);
		kfree(device->name);
		kfree(device);
	}
	kfree(fs_devices);
}

69 70 71 72
int btrfs_cleanup_fs_uuids(void)
{
	struct btrfs_fs_devices *fs_devices;

Y
Yan Zheng 已提交
73 74 75 76
	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
77
		free_fs_devices(fs_devices);
78 79 80 81
	}
	return 0;
}

82 83
static noinline struct btrfs_device *__find_device(struct list_head *head,
						   u64 devid, u8 *uuid)
84 85 86
{
	struct btrfs_device *dev;

87
	list_for_each_entry(dev, head, dev_list) {
88
		if (dev->devid == devid &&
89
		    (!uuid || !memcmp(dev->uuid, uuid, BTRFS_UUID_SIZE))) {
90
			return dev;
91
		}
92 93 94 95
	}
	return NULL;
}

96
static noinline struct btrfs_fs_devices *find_fsid(u8 *fsid)
97 98 99
{
	struct btrfs_fs_devices *fs_devices;

100
	list_for_each_entry(fs_devices, &fs_uuids, list) {
101 102 103 104 105 106
		if (memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE) == 0)
			return fs_devices;
	}
	return NULL;
}

107 108 109 110 111 112 113 114 115 116 117 118 119 120
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;
}

121 122 123 124 125 126 127 128 129 130 131
/*
 * 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.
 */
132
static noinline int run_scheduled_bios(struct btrfs_device *device)
133 134 135
{
	struct bio *pending;
	struct backing_dev_info *bdi;
136
	struct btrfs_fs_info *fs_info;
137
	struct btrfs_pending_bios *pending_bios;
138 139 140
	struct bio *tail;
	struct bio *cur;
	int again = 0;
141
	unsigned long num_run;
142
	unsigned long batch_run = 0;
143
	unsigned long limit;
144
	unsigned long last_waited = 0;
145
	int force_reg = 0;
146
	int sync_pending = 0;
147 148 149 150 151 152 153 154 155
	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);
156

157
	bdi = blk_get_backing_dev_info(device->bdev);
158 159 160 161
	fs_info = device->dev_root->fs_info;
	limit = btrfs_async_submit_limit(fs_info);
	limit = limit * 2 / 3;

162 163 164
loop:
	spin_lock(&device->io_lock);

165
loop_lock:
166
	num_run = 0;
167

168 169 170 171 172
	/* 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
	 */
173
	if (!force_reg && device->pending_sync_bios.head) {
174
		pending_bios = &device->pending_sync_bios;
175 176
		force_reg = 1;
	} else {
177
		pending_bios = &device->pending_bios;
178 179
		force_reg = 0;
	}
180 181 182

	pending = pending_bios->head;
	tail = pending_bios->tail;
183 184 185 186 187 188 189 190 191 192
	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.
	 */
193 194
	if (device->pending_sync_bios.head == NULL &&
	    device->pending_bios.head == NULL) {
195 196
		again = 0;
		device->running_pending = 0;
197 198 199
	} else {
		again = 1;
		device->running_pending = 1;
200
	}
201 202 203 204

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

205 206
	spin_unlock(&device->io_lock);

207
	while (pending) {
208 209

		rmb();
210 211 212 213 214 215 216 217
		/* 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)) {
218 219 220 221 222
			spin_lock(&device->io_lock);
			requeue_list(pending_bios, pending, tail);
			goto loop_lock;
		}

223 224 225
		cur = pending;
		pending = pending->bi_next;
		cur->bi_next = NULL;
226 227 228 229 230
		atomic_dec(&fs_info->nr_async_bios);

		if (atomic_read(&fs_info->nr_async_bios) < limit &&
		    waitqueue_active(&fs_info->async_submit_wait))
			wake_up(&fs_info->async_submit_wait);
231 232

		BUG_ON(atomic_read(&cur->bi_cnt) == 0);
233

234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249
		/*
		 * 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;
		}

250 251 252
		submit_bio(cur->bi_rw, cur);
		num_run++;
		batch_run++;
253
		if (need_resched())
254
			cond_resched();
255 256 257 258 259 260

		/*
		 * we made progress, there is more work to do and the bdi
		 * is now congested.  Back off and let other work structs
		 * run instead
		 */
261
		if (pending && bdi_write_congested(bdi) && batch_run > 8 &&
262
		    fs_info->fs_devices->open_devices > 1) {
263
			struct io_context *ioc;
264

265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286
			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;
287
				if (need_resched())
288
					cond_resched();
289 290
				continue;
			}
291
			spin_lock(&device->io_lock);
292
			requeue_list(pending_bios, pending, tail);
293
			device->running_pending = 1;
294 295 296 297 298

			spin_unlock(&device->io_lock);
			btrfs_requeue_work(&device->work);
			goto done;
		}
299 300 301 302 303 304
		/* unplug every 64 requests just for good measure */
		if (batch_run % 64 == 0) {
			blk_finish_plug(&plug);
			blk_start_plug(&plug);
			sync_pending = 0;
		}
305
	}
306

307 308 309 310 311 312 313 314 315
	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);

316
done:
317
	blk_finish_plug(&plug);
318 319 320
	return 0;
}

321
static void pending_bios_fn(struct btrfs_work *work)
322 323 324 325 326 327 328
{
	struct btrfs_device *device;

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

329
static noinline int device_list_add(const char *path,
330 331 332 333 334 335
			   struct btrfs_super_block *disk_super,
			   u64 devid, struct btrfs_fs_devices **fs_devices_ret)
{
	struct btrfs_device *device;
	struct btrfs_fs_devices *fs_devices;
	u64 found_transid = btrfs_super_generation(disk_super);
336
	char *name;
337 338 339

	fs_devices = find_fsid(disk_super->fsid);
	if (!fs_devices) {
340
		fs_devices = kzalloc(sizeof(*fs_devices), GFP_NOFS);
341 342 343
		if (!fs_devices)
			return -ENOMEM;
		INIT_LIST_HEAD(&fs_devices->devices);
344
		INIT_LIST_HEAD(&fs_devices->alloc_list);
345 346 347 348
		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;
349
		mutex_init(&fs_devices->device_list_mutex);
350 351
		device = NULL;
	} else {
352 353
		device = __find_device(&fs_devices->devices, devid,
				       disk_super->dev_item.uuid);
354 355
	}
	if (!device) {
Y
Yan Zheng 已提交
356 357 358
		if (fs_devices->opened)
			return -EBUSY;

359 360 361 362 363 364
		device = kzalloc(sizeof(*device), GFP_NOFS);
		if (!device) {
			/* we can safely leave the fs_devices entry around */
			return -ENOMEM;
		}
		device->devid = devid;
365
		device->work.func = pending_bios_fn;
366 367
		memcpy(device->uuid, disk_super->dev_item.uuid,
		       BTRFS_UUID_SIZE);
368
		spin_lock_init(&device->io_lock);
369 370 371 372 373
		device->name = kstrdup(path, GFP_NOFS);
		if (!device->name) {
			kfree(device);
			return -ENOMEM;
		}
Y
Yan Zheng 已提交
374
		INIT_LIST_HEAD(&device->dev_alloc_list);
375

376 377 378 379 380 381 382 383
		/* 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);

384
		mutex_lock(&fs_devices->device_list_mutex);
385
		list_add_rcu(&device->dev_list, &fs_devices->devices);
386 387
		mutex_unlock(&fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
388
		device->fs_devices = fs_devices;
389
		fs_devices->num_devices++;
390
	} else if (!device->name || strcmp(device->name, path)) {
391 392 393 394 395
		name = kstrdup(path, GFP_NOFS);
		if (!name)
			return -ENOMEM;
		kfree(device->name);
		device->name = name;
396 397 398 399
		if (device->missing) {
			fs_devices->missing_devices--;
			device->missing = 0;
		}
400 401 402 403 404 405 406 407 408 409
	}

	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
410 411 412 413 414 415 416 417 418 419 420 421 422
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);
423
	mutex_init(&fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
424 425 426 427
	fs_devices->latest_devid = orig->latest_devid;
	fs_devices->latest_trans = orig->latest_trans;
	memcpy(fs_devices->fsid, orig->fsid, sizeof(fs_devices->fsid));

428
	/* We have held the volume lock, it is safe to get the devices. */
Y
Yan Zheng 已提交
429 430 431 432 433 434
	list_for_each_entry(orig_dev, &orig->devices, dev_list) {
		device = kzalloc(sizeof(*device), GFP_NOFS);
		if (!device)
			goto error;

		device->name = kstrdup(orig_dev->name, GFP_NOFS);
435 436
		if (!device->name) {
			kfree(device);
Y
Yan Zheng 已提交
437
			goto error;
438
		}
Y
Yan Zheng 已提交
439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456

		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);
}

457 458
int btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices)
{
459
	struct btrfs_device *device, *next;
460 461 462

	mutex_lock(&uuid_mutex);
again:
463
	/* This is the initialized path, it is safe to release the devices. */
464
	list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
Y
Yan Zheng 已提交
465 466 467 468
		if (device->in_fs_metadata)
			continue;

		if (device->bdev) {
469
			blkdev_put(device->bdev, device->mode);
Y
Yan Zheng 已提交
470 471 472 473 474 475 476 477
			device->bdev = NULL;
			fs_devices->open_devices--;
		}
		if (device->writeable) {
			list_del_init(&device->dev_alloc_list);
			device->writeable = 0;
			fs_devices->rw_devices--;
		}
Y
Yan Zheng 已提交
478 479 480 481
		list_del_init(&device->dev_list);
		fs_devices->num_devices--;
		kfree(device->name);
		kfree(device);
482
	}
Y
Yan Zheng 已提交
483 484 485 486 487 488

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

489 490 491
	mutex_unlock(&uuid_mutex);
	return 0;
}
492

493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515
static void __free_device(struct work_struct *work)
{
	struct btrfs_device *device;

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

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

	kfree(device->name);
	kfree(device);
}

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

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

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

Y
Yan Zheng 已提交
516
static int __btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
517 518
{
	struct btrfs_device *device;
Y
Yan Zheng 已提交
519

Y
Yan Zheng 已提交
520 521
	if (--fs_devices->opened > 0)
		return 0;
522

523
	mutex_lock(&fs_devices->device_list_mutex);
524
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
525 526 527
		struct btrfs_device *new_device;

		if (device->bdev)
528
			fs_devices->open_devices--;
529

Y
Yan Zheng 已提交
530 531 532 533 534
		if (device->writeable) {
			list_del_init(&device->dev_alloc_list);
			fs_devices->rw_devices--;
		}

535 536 537
		if (device->can_discard)
			fs_devices->num_can_discard--;

538 539 540 541
		new_device = kmalloc(sizeof(*new_device), GFP_NOFS);
		BUG_ON(!new_device);
		memcpy(new_device, device, sizeof(*new_device));
		new_device->name = kstrdup(device->name, GFP_NOFS);
542
		BUG_ON(device->name && !new_device->name);
543 544 545
		new_device->bdev = NULL;
		new_device->writeable = 0;
		new_device->in_fs_metadata = 0;
546
		new_device->can_discard = 0;
547 548 549
		list_replace_rcu(&device->dev_list, &new_device->dev_list);

		call_rcu(&device->rcu, free_device);
550
	}
551 552
	mutex_unlock(&fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
553 554
	WARN_ON(fs_devices->open_devices);
	WARN_ON(fs_devices->rw_devices);
Y
Yan Zheng 已提交
555 556 557
	fs_devices->opened = 0;
	fs_devices->seeding = 0;

558 559 560
	return 0;
}

Y
Yan Zheng 已提交
561 562
int btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
{
Y
Yan Zheng 已提交
563
	struct btrfs_fs_devices *seed_devices = NULL;
Y
Yan Zheng 已提交
564 565 566 567
	int ret;

	mutex_lock(&uuid_mutex);
	ret = __btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
568 569 570 571
	if (!fs_devices->opened) {
		seed_devices = fs_devices->seed;
		fs_devices->seed = NULL;
	}
Y
Yan Zheng 已提交
572
	mutex_unlock(&uuid_mutex);
Y
Yan Zheng 已提交
573 574 575 576 577 578 579

	while (seed_devices) {
		fs_devices = seed_devices;
		seed_devices = fs_devices->seed;
		__btrfs_close_devices(fs_devices);
		free_fs_devices(fs_devices);
	}
Y
Yan Zheng 已提交
580 581 582
	return ret;
}

Y
Yan Zheng 已提交
583 584
static int __btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
				fmode_t flags, void *holder)
585
{
586
	struct request_queue *q;
587 588 589
	struct block_device *bdev;
	struct list_head *head = &fs_devices->devices;
	struct btrfs_device *device;
590 591 592 593 594 595
	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
596
	int seeding = 1;
597
	int ret = 0;
598

599 600
	flags |= FMODE_EXCL;

601
	list_for_each_entry(device, head, dev_list) {
602 603
		if (device->bdev)
			continue;
604 605 606
		if (!device->name)
			continue;

607
		bdev = blkdev_get_by_path(device->name, flags, holder);
608
		if (IS_ERR(bdev)) {
609
			printk(KERN_INFO "open %s failed\n", device->name);
610
			goto error;
611
		}
612
		set_blocksize(bdev, 4096);
613

Y
Yan Zheng 已提交
614
		bh = btrfs_read_dev_super(bdev);
615
		if (!bh)
616 617 618
			goto error_close;

		disk_super = (struct btrfs_super_block *)bh->b_data;
619
		devid = btrfs_stack_device_id(&disk_super->dev_item);
620 621 622
		if (devid != device->devid)
			goto error_brelse;

Y
Yan Zheng 已提交
623 624 625 626 627 628
		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) {
629
			latest_devid = devid;
Y
Yan Zheng 已提交
630
			latest_transid = device->generation;
631 632 633
			latest_bdev = bdev;
		}

Y
Yan Zheng 已提交
634 635 636 637 638 639 640
		if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_SEEDING) {
			device->writeable = 0;
		} else {
			device->writeable = !bdev_read_only(bdev);
			seeding = 0;
		}

641 642 643 644 645 646
		q = bdev_get_queue(bdev);
		if (blk_queue_discard(q)) {
			device->can_discard = 1;
			fs_devices->num_can_discard++;
		}

647
		device->bdev = bdev;
648
		device->in_fs_metadata = 0;
649 650
		device->mode = flags;

651 652 653
		if (!blk_queue_nonrot(bdev_get_queue(bdev)))
			fs_devices->rotating = 1;

654
		fs_devices->open_devices++;
Y
Yan Zheng 已提交
655 656 657 658 659
		if (device->writeable) {
			fs_devices->rw_devices++;
			list_add(&device->dev_alloc_list,
				 &fs_devices->alloc_list);
		}
660
		brelse(bh);
661
		continue;
662

663 664 665
error_brelse:
		brelse(bh);
error_close:
666
		blkdev_put(bdev, flags);
667 668
error:
		continue;
669
	}
670
	if (fs_devices->open_devices == 0) {
671
		ret = -EINVAL;
672 673
		goto out;
	}
Y
Yan Zheng 已提交
674 675
	fs_devices->seeding = seeding;
	fs_devices->opened = 1;
676 677 678
	fs_devices->latest_bdev = latest_bdev;
	fs_devices->latest_devid = latest_devid;
	fs_devices->latest_trans = latest_transid;
Y
Yan Zheng 已提交
679
	fs_devices->total_rw_bytes = 0;
680
out:
Y
Yan Zheng 已提交
681 682 683 684
	return ret;
}

int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
685
		       fmode_t flags, void *holder)
Y
Yan Zheng 已提交
686 687 688 689 690
{
	int ret;

	mutex_lock(&uuid_mutex);
	if (fs_devices->opened) {
Y
Yan Zheng 已提交
691 692
		fs_devices->opened++;
		ret = 0;
Y
Yan Zheng 已提交
693
	} else {
694
		ret = __btrfs_open_devices(fs_devices, flags, holder);
Y
Yan Zheng 已提交
695
	}
696 697 698 699
	mutex_unlock(&uuid_mutex);
	return ret;
}

700
int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
701 702 703 704 705 706 707
			  struct btrfs_fs_devices **fs_devices_ret)
{
	struct btrfs_super_block *disk_super;
	struct block_device *bdev;
	struct buffer_head *bh;
	int ret;
	u64 devid;
708
	u64 transid;
709 710 711

	mutex_lock(&uuid_mutex);

712 713
	flags |= FMODE_EXCL;
	bdev = blkdev_get_by_path(path, flags, holder);
714 715 716 717 718 719 720 721 722

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

	ret = set_blocksize(bdev, 4096);
	if (ret)
		goto error_close;
Y
Yan Zheng 已提交
723
	bh = btrfs_read_dev_super(bdev);
724
	if (!bh) {
725
		ret = -EINVAL;
726 727 728
		goto error_close;
	}
	disk_super = (struct btrfs_super_block *)bh->b_data;
729
	devid = btrfs_stack_device_id(&disk_super->dev_item);
730
	transid = btrfs_super_generation(disk_super);
731
	if (disk_super->label[0])
732
		printk(KERN_INFO "device label %s ", disk_super->label);
733 734
	else
		printk(KERN_INFO "device fsid %pU ", disk_super->fsid);
735
	printk(KERN_CONT "devid %llu transid %llu %s\n",
736
	       (unsigned long long)devid, (unsigned long long)transid, path);
737 738 739 740
	ret = device_list_add(path, disk_super, devid, fs_devices_ret);

	brelse(bh);
error_close:
741
	blkdev_put(bdev, flags);
742 743 744 745
error:
	mutex_unlock(&uuid_mutex);
	return ret;
}
746

747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830
/* helper to account the used device space in the range */
int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start,
				   u64 end, u64 *length)
{
	struct btrfs_key key;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_dev_extent *dev_extent;
	struct btrfs_path *path;
	u64 extent_end;
	int ret;
	int slot;
	struct extent_buffer *l;

	*length = 0;

	if (start >= device->total_bytes)
		return 0;

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

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

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

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

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

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

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

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

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

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

831
/*
832 833 834 835 836 837 838 839
 * find_free_dev_extent - find free space in the specified device
 * @trans:	transaction handler
 * @device:	the device which we search the free space in
 * @num_bytes:	the size of the free space that we need
 * @start:	store the start of the free space.
 * @len:	the size of the free space. that we find, or the size of the max
 * 		free space if we don't find suitable free space
 *
840 841 842
 * 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
843 844 845 846 847 848 849 850
 *
 * @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.
851
 */
852 853
int find_free_dev_extent(struct btrfs_trans_handle *trans,
			 struct btrfs_device *device, u64 num_bytes,
854
			 u64 *start, u64 *len)
855 856 857
{
	struct btrfs_key key;
	struct btrfs_root *root = device->dev_root;
858
	struct btrfs_dev_extent *dev_extent;
Y
Yan Zheng 已提交
859
	struct btrfs_path *path;
860 861 862 863 864
	u64 hole_size;
	u64 max_hole_start;
	u64 max_hole_size;
	u64 extent_end;
	u64 search_start;
865 866
	u64 search_end = device->total_bytes;
	int ret;
867
	int slot;
868 869 870 871
	struct extent_buffer *l;

	/* FIXME use last free of some kind */

872 873 874
	/* 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
875
	search_start = max(root->fs_info->alloc_start, 1024ull * 1024);
876

877 878
	max_hole_start = search_start;
	max_hole_size = 0;
879
	hole_size = 0;
880 881 882 883 884 885 886 887 888 889 890 891 892

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

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

893 894 895
	key.objectid = device->devid;
	key.offset = search_start;
	key.type = BTRFS_DEV_EXTENT_KEY;
896

897 898
	ret = btrfs_search_slot(trans, root, &key, path, 0, 0);
	if (ret < 0)
899
		goto out;
900 901 902
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid, key.type);
		if (ret < 0)
903
			goto out;
904
	}
905

906 907 908 909 910 911 912 913
	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)
914 915 916
				goto out;

			break;
917 918 919 920 921 922 923
		}
		btrfs_item_key_to_cpu(l, &key, slot);

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

		if (key.objectid > device->devid)
924
			break;
925

926 927
		if (btrfs_key_type(&key) != BTRFS_DEV_EXTENT_KEY)
			goto next;
928

929 930
		if (key.offset > search_start) {
			hole_size = key.offset - search_start;
931

932 933 934 935
			if (hole_size > max_hole_size) {
				max_hole_start = search_start;
				max_hole_size = hole_size;
			}
936

937 938 939 940 941 942 943 944 945 946 947 948
			/*
			 * 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;
949 950 951 952
			}
		}

		dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
953 954 955 956
		extent_end = key.offset + btrfs_dev_extent_length(l,
								  dev_extent);
		if (extent_end > search_start)
			search_start = extent_end;
957 958 959 960 961
next:
		path->slots[0]++;
		cond_resched();
	}

962 963 964 965 966 967 968 969
	/*
	 * 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;

970 971 972
	if (hole_size > max_hole_size) {
		max_hole_start = search_start;
		max_hole_size = hole_size;
973 974
	}

975 976 977 978 979 980 981
	/* See above. */
	if (hole_size < num_bytes)
		ret = -ENOSPC;
	else
		ret = 0;

out:
Y
Yan Zheng 已提交
982
	btrfs_free_path(path);
983 984
error:
	*start = max_hole_start;
985
	if (len)
986
		*len = max_hole_size;
987 988 989
	return ret;
}

990
static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans,
991 992 993 994 995 996 997
			  struct btrfs_device *device,
			  u64 start)
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_root *root = device->dev_root;
	struct btrfs_key key;
998 999 1000
	struct btrfs_key found_key;
	struct extent_buffer *leaf = NULL;
	struct btrfs_dev_extent *extent = NULL;
1001 1002 1003 1004 1005 1006 1007 1008

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

	key.objectid = device->devid;
	key.offset = start;
	key.type = BTRFS_DEV_EXTENT_KEY;
1009
again:
1010
	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
1011 1012 1013
	if (ret > 0) {
		ret = btrfs_previous_item(root, path, key.objectid,
					  BTRFS_DEV_EXTENT_KEY);
1014 1015
		if (ret)
			goto out;
1016 1017 1018 1019 1020 1021
		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);
1022 1023 1024
		key = found_key;
		btrfs_release_path(path);
		goto again;
1025 1026 1027 1028 1029
	} else if (ret == 0) {
		leaf = path->nodes[0];
		extent = btrfs_item_ptr(leaf, path->slots[0],
					struct btrfs_dev_extent);
	}
1030 1031
	BUG_ON(ret);

1032 1033 1034 1035 1036 1037 1038
	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);
	}
1039 1040
	ret = btrfs_del_item(trans, root, path);

1041
out:
1042 1043 1044 1045
	btrfs_free_path(path);
	return ret;
}

Y
Yan Zheng 已提交
1046
int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
1047
			   struct btrfs_device *device,
1048
			   u64 chunk_tree, u64 chunk_objectid,
Y
Yan Zheng 已提交
1049
			   u64 chunk_offset, u64 start, u64 num_bytes)
1050 1051 1052 1053 1054 1055 1056 1057
{
	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;

1058
	WARN_ON(!device->in_fs_metadata);
1059 1060 1061 1062 1063
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = device->devid;
Y
Yan Zheng 已提交
1064
	key.offset = start;
1065 1066 1067 1068 1069 1070 1071 1072
	key.type = BTRFS_DEV_EXTENT_KEY;
	ret = btrfs_insert_empty_item(trans, root, path, &key,
				      sizeof(*extent));
	BUG_ON(ret);

	leaf = path->nodes[0];
	extent = btrfs_item_ptr(leaf, path->slots[0],
				struct btrfs_dev_extent);
1073 1074 1075 1076 1077 1078 1079 1080
	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);

1081 1082 1083 1084 1085 1086
	btrfs_set_dev_extent_length(leaf, extent, num_bytes);
	btrfs_mark_buffer_dirty(leaf);
	btrfs_free_path(path);
	return ret;
}

1087 1088
static noinline int find_next_chunk(struct btrfs_root *root,
				    u64 objectid, u64 *offset)
1089 1090 1091 1092
{
	struct btrfs_path *path;
	int ret;
	struct btrfs_key key;
1093
	struct btrfs_chunk *chunk;
1094 1095 1096
	struct btrfs_key found_key;

	path = btrfs_alloc_path();
1097 1098
	if (!path)
		return -ENOMEM;
1099

1100
	key.objectid = objectid;
1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111
	key.offset = (u64)-1;
	key.type = BTRFS_CHUNK_ITEM_KEY;

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

	BUG_ON(ret == 0);

	ret = btrfs_previous_item(root, path, 0, BTRFS_CHUNK_ITEM_KEY);
	if (ret) {
1112
		*offset = 0;
1113 1114 1115
	} else {
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
1116 1117 1118 1119 1120 1121 1122 1123
		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);
		}
1124 1125 1126 1127 1128 1129 1130
	}
	ret = 0;
error:
	btrfs_free_path(path);
	return ret;
}

Y
Yan Zheng 已提交
1131
static noinline int find_next_devid(struct btrfs_root *root, u64 *objectid)
1132 1133 1134 1135
{
	int ret;
	struct btrfs_key key;
	struct btrfs_key found_key;
Y
Yan Zheng 已提交
1136 1137 1138 1139 1140 1141 1142
	struct btrfs_path *path;

	root = root->fs_info->chunk_root;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164

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

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

	BUG_ON(ret == 0);

	ret = btrfs_previous_item(root, path, BTRFS_DEV_ITEMS_OBJECTID,
				  BTRFS_DEV_ITEM_KEY);
	if (ret) {
		*objectid = 1;
	} else {
		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
				      path->slots[0]);
		*objectid = found_key.offset + 1;
	}
	ret = 0;
error:
Y
Yan Zheng 已提交
1165
	btrfs_free_path(path);
1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191
	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
1192
	key.offset = device->devid;
1193 1194

	ret = btrfs_insert_empty_item(trans, root, path, &key,
1195
				      sizeof(*dev_item));
1196 1197 1198 1199 1200 1201 1202
	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
1203
	btrfs_set_device_generation(leaf, dev_item, 0);
1204 1205 1206 1207 1208 1209
	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);
1210 1211 1212
	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);
1213
	btrfs_set_device_start_offset(leaf, dev_item, 0);
1214 1215

	ptr = (unsigned long)btrfs_device_uuid(dev_item);
1216
	write_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
1217 1218
	ptr = (unsigned long)btrfs_device_fsid(dev_item);
	write_extent_buffer(leaf, root->fs_info->fsid, ptr, BTRFS_UUID_SIZE);
1219 1220
	btrfs_mark_buffer_dirty(leaf);

Y
Yan Zheng 已提交
1221
	ret = 0;
1222 1223 1224 1225
out:
	btrfs_free_path(path);
	return ret;
}
1226

1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240
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;

1241
	trans = btrfs_start_transaction(root, 0);
1242 1243 1244 1245
	if (IS_ERR(trans)) {
		btrfs_free_path(path);
		return PTR_ERR(trans);
	}
1246 1247 1248
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.type = BTRFS_DEV_ITEM_KEY;
	key.offset = device->devid;
1249
	lock_chunks(root);
1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264

	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);
1265
	unlock_chunks(root);
1266 1267 1268 1269 1270 1271 1272
	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
1273
	struct btrfs_device *next_device;
1274
	struct block_device *bdev;
1275
	struct buffer_head *bh = NULL;
1276
	struct btrfs_super_block *disk_super;
1277
	struct btrfs_fs_devices *cur_devices;
1278 1279
	u64 all_avail;
	u64 devid;
Y
Yan Zheng 已提交
1280 1281
	u64 num_devices;
	u8 *dev_uuid;
1282
	int ret = 0;
1283
	bool clear_super = false;
1284 1285 1286 1287 1288 1289 1290 1291

	mutex_lock(&uuid_mutex);

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

	if ((all_avail & BTRFS_BLOCK_GROUP_RAID10) &&
1292
	    root->fs_info->fs_devices->num_devices <= 4) {
1293 1294
		printk(KERN_ERR "btrfs: unable to go below four devices "
		       "on raid10\n");
1295 1296 1297 1298 1299
		ret = -EINVAL;
		goto out;
	}

	if ((all_avail & BTRFS_BLOCK_GROUP_RAID1) &&
1300
	    root->fs_info->fs_devices->num_devices <= 2) {
1301 1302
		printk(KERN_ERR "btrfs: unable to go below two "
		       "devices on raid1\n");
1303 1304 1305 1306
		ret = -EINVAL;
		goto out;
	}

1307 1308 1309
	if (strcmp(device_path, "missing") == 0) {
		struct list_head *devices;
		struct btrfs_device *tmp;
1310

1311 1312
		device = NULL;
		devices = &root->fs_info->fs_devices->devices;
1313 1314 1315 1316
		/*
		 * It is safe to read the devices since the volume_mutex
		 * is held.
		 */
1317
		list_for_each_entry(tmp, devices, dev_list) {
1318 1319 1320 1321 1322 1323 1324 1325 1326
			if (tmp->in_fs_metadata && !tmp->bdev) {
				device = tmp;
				break;
			}
		}
		bdev = NULL;
		bh = NULL;
		disk_super = NULL;
		if (!device) {
1327 1328
			printk(KERN_ERR "btrfs: no missing devices found to "
			       "remove\n");
1329 1330 1331
			goto out;
		}
	} else {
1332 1333
		bdev = blkdev_get_by_path(device_path, FMODE_READ | FMODE_EXCL,
					  root->fs_info->bdev_holder);
1334 1335 1336 1337
		if (IS_ERR(bdev)) {
			ret = PTR_ERR(bdev);
			goto out;
		}
1338

Y
Yan Zheng 已提交
1339
		set_blocksize(bdev, 4096);
Y
Yan Zheng 已提交
1340
		bh = btrfs_read_dev_super(bdev);
1341
		if (!bh) {
1342
			ret = -EINVAL;
1343 1344 1345
			goto error_close;
		}
		disk_super = (struct btrfs_super_block *)bh->b_data;
1346
		devid = btrfs_stack_device_id(&disk_super->dev_item);
Y
Yan Zheng 已提交
1347 1348 1349
		dev_uuid = disk_super->dev_item.uuid;
		device = btrfs_find_device(root, devid, dev_uuid,
					   disk_super->fsid);
1350 1351 1352 1353
		if (!device) {
			ret = -ENOENT;
			goto error_brelse;
		}
Y
Yan Zheng 已提交
1354
	}
1355

Y
Yan Zheng 已提交
1356
	if (device->writeable && root->fs_info->fs_devices->rw_devices == 1) {
1357 1358
		printk(KERN_ERR "btrfs: unable to remove the only writeable "
		       "device\n");
Y
Yan Zheng 已提交
1359 1360 1361 1362 1363
		ret = -EINVAL;
		goto error_brelse;
	}

	if (device->writeable) {
1364
		lock_chunks(root);
Y
Yan Zheng 已提交
1365
		list_del_init(&device->dev_alloc_list);
1366
		unlock_chunks(root);
Y
Yan Zheng 已提交
1367
		root->fs_info->fs_devices->rw_devices--;
1368
		clear_super = true;
1369
	}
1370 1371 1372

	ret = btrfs_shrink_device(device, 0);
	if (ret)
1373
		goto error_undo;
1374 1375 1376

	ret = btrfs_rm_dev_item(root->fs_info->chunk_root, device);
	if (ret)
1377
		goto error_undo;
1378

1379 1380 1381 1382 1383
	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
1384
	device->in_fs_metadata = 0;
A
Arne Jansen 已提交
1385
	btrfs_scrub_cancel_dev(root, device);
1386 1387 1388 1389 1390 1391

	/*
	 * the device list mutex makes sure that we don't change
	 * the device list while someone else is writing out all
	 * the device supers.
	 */
1392 1393

	cur_devices = device->fs_devices;
1394
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1395
	list_del_rcu(&device->dev_list);
1396

Y
Yan Zheng 已提交
1397
	device->fs_devices->num_devices--;
Y
Yan Zheng 已提交
1398

1399 1400 1401
	if (device->missing)
		root->fs_info->fs_devices->missing_devices--;

Y
Yan Zheng 已提交
1402 1403 1404 1405 1406 1407 1408
	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;

1409
	if (device->bdev)
Y
Yan Zheng 已提交
1410
		device->fs_devices->open_devices--;
1411 1412 1413

	call_rcu(&device->rcu, free_device);
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
Y
Yan Zheng 已提交
1414

1415 1416
	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
1417

1418
	if (cur_devices->open_devices == 0) {
Y
Yan Zheng 已提交
1419 1420 1421
		struct btrfs_fs_devices *fs_devices;
		fs_devices = root->fs_info->fs_devices;
		while (fs_devices) {
1422
			if (fs_devices->seed == cur_devices)
Y
Yan Zheng 已提交
1423 1424
				break;
			fs_devices = fs_devices->seed;
Y
Yan Zheng 已提交
1425
		}
1426 1427
		fs_devices->seed = cur_devices->seed;
		cur_devices->seed = NULL;
1428
		lock_chunks(root);
1429
		__btrfs_close_devices(cur_devices);
1430
		unlock_chunks(root);
1431
		free_fs_devices(cur_devices);
Y
Yan Zheng 已提交
1432 1433 1434 1435 1436 1437
	}

	/*
	 * at this point, the device is zero sized.  We want to
	 * remove it from the devices list and zero out the old super
	 */
1438
	if (clear_super) {
1439 1440 1441 1442 1443 1444 1445
		/* 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);
	}
1446 1447 1448 1449 1450 1451

	ret = 0;

error_brelse:
	brelse(bh);
error_close:
1452
	if (bdev)
1453
		blkdev_put(bdev, FMODE_READ | FMODE_EXCL);
1454 1455 1456
out:
	mutex_unlock(&uuid_mutex);
	return ret;
1457 1458
error_undo:
	if (device->writeable) {
1459
		lock_chunks(root);
1460 1461
		list_add(&device->dev_alloc_list,
			 &root->fs_info->fs_devices->alloc_list);
1462
		unlock_chunks(root);
1463 1464 1465
		root->fs_info->fs_devices->rw_devices++;
	}
	goto error_brelse;
1466 1467
}

Y
Yan Zheng 已提交
1468 1469 1470 1471 1472 1473 1474 1475
/*
 * does all the dirty work required for changing file system's UUID.
 */
static int btrfs_prepare_sprout(struct btrfs_trans_handle *trans,
				struct btrfs_root *root)
{
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
	struct btrfs_fs_devices *old_devices;
Y
Yan Zheng 已提交
1476
	struct btrfs_fs_devices *seed_devices;
1477
	struct btrfs_super_block *disk_super = root->fs_info->super_copy;
Y
Yan Zheng 已提交
1478 1479 1480 1481
	struct btrfs_device *device;
	u64 super_flags;

	BUG_ON(!mutex_is_locked(&uuid_mutex));
Y
Yan Zheng 已提交
1482
	if (!fs_devices->seeding)
Y
Yan Zheng 已提交
1483 1484
		return -EINVAL;

Y
Yan Zheng 已提交
1485 1486
	seed_devices = kzalloc(sizeof(*fs_devices), GFP_NOFS);
	if (!seed_devices)
Y
Yan Zheng 已提交
1487 1488
		return -ENOMEM;

Y
Yan Zheng 已提交
1489 1490 1491 1492
	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
1493
	}
Y
Yan Zheng 已提交
1494

Y
Yan Zheng 已提交
1495 1496
	list_add(&old_devices->list, &fs_uuids);

Y
Yan Zheng 已提交
1497 1498 1499 1500
	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);
1501
	mutex_init(&seed_devices->device_list_mutex);
1502 1503

	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1504 1505
	list_splice_init_rcu(&fs_devices->devices, &seed_devices->devices,
			      synchronize_rcu);
1506 1507
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);

Y
Yan Zheng 已提交
1508 1509 1510 1511 1512
	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
1513 1514 1515
	fs_devices->seeding = 0;
	fs_devices->num_devices = 0;
	fs_devices->open_devices = 0;
Y
Yan Zheng 已提交
1516
	fs_devices->seed = seed_devices;
Y
Yan Zheng 已提交
1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567

	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]);
1568
			btrfs_release_path(path);
Y
Yan Zheng 已提交
1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603
			continue;
		}

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

		dev_item = btrfs_item_ptr(leaf, path->slots[0],
					  struct btrfs_dev_item);
		devid = btrfs_device_id(leaf, dev_item);
		read_extent_buffer(leaf, dev_uuid,
				   (unsigned long)btrfs_device_uuid(dev_item),
				   BTRFS_UUID_SIZE);
		read_extent_buffer(leaf, fs_uuid,
				   (unsigned long)btrfs_device_fsid(dev_item),
				   BTRFS_UUID_SIZE);
		device = btrfs_find_device(root, devid, dev_uuid, fs_uuid);
		BUG_ON(!device);

		if (device->fs_devices->seeding) {
			btrfs_set_device_generation(leaf, dev_item,
						    device->generation);
			btrfs_mark_buffer_dirty(leaf);
		}

		path->slots[0]++;
		goto next_slot;
	}
	ret = 0;
error:
	btrfs_free_path(path);
	return ret;
}

1604 1605
int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
{
1606
	struct request_queue *q;
1607 1608 1609 1610
	struct btrfs_trans_handle *trans;
	struct btrfs_device *device;
	struct block_device *bdev;
	struct list_head *devices;
Y
Yan Zheng 已提交
1611
	struct super_block *sb = root->fs_info->sb;
1612
	u64 total_bytes;
Y
Yan Zheng 已提交
1613
	int seeding_dev = 0;
1614 1615
	int ret = 0;

Y
Yan Zheng 已提交
1616 1617
	if ((sb->s_flags & MS_RDONLY) && !root->fs_info->fs_devices->seeding)
		return -EINVAL;
1618

1619
	bdev = blkdev_get_by_path(device_path, FMODE_WRITE | FMODE_EXCL,
1620
				  root->fs_info->bdev_holder);
1621 1622
	if (IS_ERR(bdev))
		return PTR_ERR(bdev);
1623

Y
Yan Zheng 已提交
1624 1625 1626 1627 1628 1629
	if (root->fs_info->fs_devices->seeding) {
		seeding_dev = 1;
		down_write(&sb->s_umount);
		mutex_lock(&uuid_mutex);
	}

1630
	filemap_write_and_wait(bdev->bd_inode->i_mapping);
1631

1632
	devices = &root->fs_info->fs_devices->devices;
1633 1634 1635 1636
	/*
	 * we have the volume lock, so we don't need the extra
	 * device list mutex while reading the list here.
	 */
1637
	list_for_each_entry(device, devices, dev_list) {
1638 1639
		if (device->bdev == bdev) {
			ret = -EEXIST;
Y
Yan Zheng 已提交
1640
			goto error;
1641 1642 1643 1644 1645 1646 1647
		}
	}

	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
1648
		goto error;
1649 1650 1651 1652 1653
	}

	device->name = kstrdup(device_path, GFP_NOFS);
	if (!device->name) {
		kfree(device);
Y
Yan Zheng 已提交
1654 1655
		ret = -ENOMEM;
		goto error;
1656
	}
Y
Yan Zheng 已提交
1657 1658 1659

	ret = find_next_devid(root, &device->devid);
	if (ret) {
1660
		kfree(device->name);
Y
Yan Zheng 已提交
1661 1662 1663 1664
		kfree(device);
		goto error;
	}

1665
	trans = btrfs_start_transaction(root, 0);
1666
	if (IS_ERR(trans)) {
1667
		kfree(device->name);
1668 1669 1670 1671 1672
		kfree(device);
		ret = PTR_ERR(trans);
		goto error;
	}

Y
Yan Zheng 已提交
1673 1674
	lock_chunks(root);

1675 1676 1677
	q = bdev_get_queue(bdev);
	if (blk_queue_discard(q))
		device->can_discard = 1;
Y
Yan Zheng 已提交
1678 1679 1680 1681 1682
	device->writeable = 1;
	device->work.func = pending_bios_fn;
	generate_random_uuid(device->uuid);
	spin_lock_init(&device->io_lock);
	device->generation = trans->transid;
1683 1684 1685 1686
	device->io_width = root->sectorsize;
	device->io_align = root->sectorsize;
	device->sector_size = root->sectorsize;
	device->total_bytes = i_size_read(bdev->bd_inode);
1687
	device->disk_total_bytes = device->total_bytes;
1688 1689
	device->dev_root = root->fs_info->dev_root;
	device->bdev = bdev;
1690
	device->in_fs_metadata = 1;
1691
	device->mode = FMODE_EXCL;
Y
Yan Zheng 已提交
1692
	set_blocksize(device->bdev, 4096);
1693

Y
Yan Zheng 已提交
1694 1695 1696 1697 1698
	if (seeding_dev) {
		sb->s_flags &= ~MS_RDONLY;
		ret = btrfs_prepare_sprout(trans, root);
		BUG_ON(ret);
	}
1699

Y
Yan Zheng 已提交
1700
	device->fs_devices = root->fs_info->fs_devices;
1701 1702 1703 1704 1705 1706

	/*
	 * we don't want write_supers to jump in here with our device
	 * half setup
	 */
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
1707
	list_add_rcu(&device->dev_list, &root->fs_info->fs_devices->devices);
Y
Yan Zheng 已提交
1708 1709 1710 1711 1712
	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++;
1713 1714
	if (device->can_discard)
		root->fs_info->fs_devices->num_can_discard++;
Y
Yan Zheng 已提交
1715
	root->fs_info->fs_devices->total_rw_bytes += device->total_bytes;
1716

1717 1718 1719 1720
	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);

1721 1722 1723
	if (!blk_queue_nonrot(bdev_get_queue(bdev)))
		root->fs_info->fs_devices->rotating = 1;

1724 1725
	total_bytes = btrfs_super_total_bytes(root->fs_info->super_copy);
	btrfs_set_super_total_bytes(root->fs_info->super_copy,
1726 1727
				    total_bytes + device->total_bytes);

1728 1729
	total_bytes = btrfs_super_num_devices(root->fs_info->super_copy);
	btrfs_set_super_num_devices(root->fs_info->super_copy,
1730
				    total_bytes + 1);
1731
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
1732

Y
Yan Zheng 已提交
1733 1734 1735 1736 1737 1738 1739 1740 1741
	if (seeding_dev) {
		ret = init_first_rw_device(trans, root, device);
		BUG_ON(ret);
		ret = btrfs_finish_sprout(trans, root);
		BUG_ON(ret);
	} else {
		ret = btrfs_add_device(trans, root, device);
	}

1742 1743 1744 1745 1746 1747
	/*
	 * we've got more storage, clear any full flags on the space
	 * infos
	 */
	btrfs_clear_space_info_full(root->fs_info);

1748
	unlock_chunks(root);
Y
Yan Zheng 已提交
1749
	btrfs_commit_transaction(trans, root);
1750

Y
Yan Zheng 已提交
1751 1752 1753
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
1754

Y
Yan Zheng 已提交
1755 1756 1757
		ret = btrfs_relocate_sys_chunks(root);
		BUG_ON(ret);
	}
1758

Y
Yan Zheng 已提交
1759 1760
	return ret;
error:
1761
	blkdev_put(bdev, FMODE_EXCL);
Y
Yan Zheng 已提交
1762 1763 1764 1765
	if (seeding_dev) {
		mutex_unlock(&uuid_mutex);
		up_write(&sb->s_umount);
	}
1766
	return ret;
1767 1768
}

1769 1770
static noinline int btrfs_update_device(struct btrfs_trans_handle *trans,
					struct btrfs_device *device)
1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805
{
	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);
1806
	btrfs_set_device_total_bytes(leaf, dev_item, device->disk_total_bytes);
1807 1808 1809 1810 1811 1812 1813 1814
	btrfs_set_device_bytes_used(leaf, dev_item, device->bytes_used);
	btrfs_mark_buffer_dirty(leaf);

out:
	btrfs_free_path(path);
	return ret;
}

1815
static int __btrfs_grow_device(struct btrfs_trans_handle *trans,
1816 1817 1818
		      struct btrfs_device *device, u64 new_size)
{
	struct btrfs_super_block *super_copy =
1819
		device->dev_root->fs_info->super_copy;
1820 1821 1822
	u64 old_total = btrfs_super_total_bytes(super_copy);
	u64 diff = new_size - device->total_bytes;

Y
Yan Zheng 已提交
1823 1824 1825 1826 1827
	if (!device->writeable)
		return -EACCES;
	if (new_size <= device->total_bytes)
		return -EINVAL;

1828
	btrfs_set_super_total_bytes(super_copy, old_total + diff);
Y
Yan Zheng 已提交
1829 1830 1831
	device->fs_devices->total_rw_bytes += diff;

	device->total_bytes = new_size;
1832
	device->disk_total_bytes = new_size;
1833 1834
	btrfs_clear_space_info_full(device->dev_root->fs_info);

1835 1836 1837
	return btrfs_update_device(trans, device);
}

1838 1839 1840 1841 1842 1843 1844 1845 1846 1847
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;
}

1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871
static int btrfs_free_chunk(struct btrfs_trans_handle *trans,
			    struct btrfs_root *root,
			    u64 chunk_tree, u64 chunk_objectid,
			    u64 chunk_offset)
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_key key;

	root = root->fs_info->chunk_root;
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	key.objectid = chunk_objectid;
	key.offset = chunk_offset;
	key.type = BTRFS_CHUNK_ITEM_KEY;

	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
	BUG_ON(ret);

	ret = btrfs_del_item(trans, root, path);

	btrfs_free_path(path);
1872
	return ret;
1873 1874
}

1875
static int btrfs_del_sys_chunk(struct btrfs_root *root, u64 chunk_objectid, u64
1876 1877
			chunk_offset)
{
1878
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920
	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;
}

1921
static int btrfs_relocate_chunk(struct btrfs_root *root,
1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936
			 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;

1937 1938 1939 1940
	ret = btrfs_can_relocate(extent_root, chunk_offset);
	if (ret)
		return -ENOSPC;

1941
	/* step one, relocate all the extents inside this chunk */
1942
	ret = btrfs_relocate_block_group(extent_root, chunk_offset);
1943 1944
	if (ret)
		return ret;
1945

1946
	trans = btrfs_start_transaction(root, 0);
1947
	BUG_ON(IS_ERR(trans));
1948

1949 1950
	lock_chunks(root);

1951 1952 1953 1954
	/*
	 * step two, delete the device extents and the
	 * chunk tree entries
	 */
1955
	read_lock(&em_tree->lock);
1956
	em = lookup_extent_mapping(em_tree, chunk_offset, 1);
1957
	read_unlock(&em_tree->lock);
1958

1959 1960
	BUG_ON(em->start > chunk_offset ||
	       em->start + em->len < chunk_offset);
1961 1962 1963 1964 1965 1966
	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);
1967

1968 1969 1970 1971
		if (map->stripes[i].dev) {
			ret = btrfs_update_device(trans, map->stripes[i].dev);
			BUG_ON(ret);
		}
1972 1973 1974 1975 1976 1977
	}
	ret = btrfs_free_chunk(trans, root, chunk_tree, chunk_objectid,
			       chunk_offset);

	BUG_ON(ret);

1978 1979
	trace_btrfs_chunk_free(root, map, chunk_offset, em->len);

1980 1981 1982 1983 1984
	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
1985 1986 1987
	ret = btrfs_remove_block_group(trans, extent_root, chunk_offset);
	BUG_ON(ret);

1988
	write_lock(&em_tree->lock);
Y
Yan Zheng 已提交
1989
	remove_extent_mapping(em_tree, em);
1990
	write_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

	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;
2015 2016
	bool retried = false;
	int failed = 0;
Y
Yan Zheng 已提交
2017 2018 2019 2020 2021 2022
	int ret;

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

2023
again:
Y
Yan Zheng 已提交
2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.offset = (u64)-1;
	key.type = BTRFS_CHUNK_ITEM_KEY;

	while (1) {
		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
		if (ret < 0)
			goto error;
		BUG_ON(ret == 0);

		ret = btrfs_previous_item(chunk_root, path, key.objectid,
					  key.type);
		if (ret < 0)
			goto error;
		if (ret > 0)
			break;
2040

Y
Yan Zheng 已提交
2041 2042
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
2043

Y
Yan Zheng 已提交
2044 2045 2046
		chunk = btrfs_item_ptr(leaf, path->slots[0],
				       struct btrfs_chunk);
		chunk_type = btrfs_chunk_type(leaf, chunk);
2047
		btrfs_release_path(path);
2048

Y
Yan Zheng 已提交
2049 2050 2051 2052
		if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) {
			ret = btrfs_relocate_chunk(chunk_root, chunk_tree,
						   found_key.objectid,
						   found_key.offset);
2053 2054 2055 2056
			if (ret == -ENOSPC)
				failed++;
			else if (ret)
				BUG();
Y
Yan Zheng 已提交
2057
		}
2058

Y
Yan Zheng 已提交
2059 2060 2061 2062 2063
		if (found_key.offset == 0)
			break;
		key.offset = found_key.offset - 1;
	}
	ret = 0;
2064 2065 2066 2067 2068 2069 2070 2071
	if (failed && !retried) {
		failed = 0;
		retried = true;
		goto again;
	} else if (failed && retried) {
		WARN_ON(1);
		ret = -ENOSPC;
	}
Y
Yan Zheng 已提交
2072 2073 2074
error:
	btrfs_free_path(path);
	return ret;
2075 2076
}

2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 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
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;
}

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 2199 2200 2201 2202 2203 2204 2205 2206 2207
/*
 * 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;
	}
}

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 2236
/*
 * 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
2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254
/*
 * Balance filters.  Return 1 if chunk should be filtered out
 * (should not be balanced).
 */
static int chunk_profiles_filter(u64 chunk_profile,
				 struct btrfs_balance_args *bargs)
{
	chunk_profile &= BTRFS_BLOCK_GROUP_PROFILE_MASK;

	if (chunk_profile == 0)
		chunk_profile = BTRFS_AVAIL_ALLOC_BIT_SINGLE;

	if (bargs->profiles & chunk_profile)
		return 0;

	return 1;
}

I
Ilya Dryomov 已提交
2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284
static u64 div_factor_fine(u64 num, int factor)
{
	if (factor <= 0)
		return 0;
	if (factor >= 100)
		return num;

	num *= factor;
	do_div(num, 100);
	return num;
}

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);

	user_thresh = div_factor_fine(cache->key.offset, bargs->usage);
	if (chunk_used < user_thresh)
		ret = 0;

	btrfs_put_block_group(cache);
	return ret;
}

I
Ilya Dryomov 已提交
2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301
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
2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341
/* [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 |
	     BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10))
		factor = 2;
	else
		factor = 1;
	factor = num_stripes / factor;

	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;
}

2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355
/* [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;
}

2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372
static int chunk_soft_convert_filter(u64 chunk_profile,
				     struct btrfs_balance_args *bargs)
{
	if (!(bargs->flags & BTRFS_BALANCE_ARGS_CONVERT))
		return 0;

	chunk_profile &= BTRFS_BLOCK_GROUP_PROFILE_MASK;

	if (chunk_profile == 0)
		chunk_profile = BTRFS_AVAIL_ALLOC_BIT_SINGLE;

	if (bargs->target & chunk_profile)
		return 1;

	return 0;
}

2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393
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
2394 2395 2396 2397
	/* profiles filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_PROFILES) &&
	    chunk_profiles_filter(chunk_type, bargs)) {
		return 0;
I
Ilya Dryomov 已提交
2398 2399 2400 2401 2402 2403
	}

	/* 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
2404 2405 2406 2407 2408 2409
	}

	/* devid filter */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_DEVID) &&
	    chunk_devid_filter(leaf, chunk, bargs)) {
		return 0;
I
Ilya Dryomov 已提交
2410 2411 2412 2413 2414 2415
	}

	/* 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;
2416 2417 2418 2419 2420 2421
	}

	/* 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
2422 2423
	}

2424 2425 2426 2427 2428 2429
	/* soft profile changing mode */
	if ((bargs->flags & BTRFS_BALANCE_ARGS_SOFT) &&
	    chunk_soft_convert_filter(chunk_type, bargs)) {
		return 0;
	}

2430 2431 2432
	return 1;
}

2433 2434 2435 2436 2437 2438 2439 2440 2441
static u64 div_factor(u64 num, int factor)
{
	if (factor == 10)
		return num;
	num *= factor;
	do_div(num, 10);
	return num;
}

2442
static int __btrfs_balance(struct btrfs_fs_info *fs_info)
2443
{
2444
	struct btrfs_balance_control *bctl = fs_info->balance_ctl;
2445 2446 2447
	struct btrfs_root *chunk_root = fs_info->chunk_root;
	struct btrfs_root *dev_root = fs_info->dev_root;
	struct list_head *devices;
2448 2449 2450
	struct btrfs_device *device;
	u64 old_size;
	u64 size_to_free;
2451
	struct btrfs_chunk *chunk;
2452 2453 2454
	struct btrfs_path *path;
	struct btrfs_key key;
	struct btrfs_key found_key;
2455
	struct btrfs_trans_handle *trans;
2456 2457
	struct extent_buffer *leaf;
	int slot;
2458 2459
	int ret;
	int enospc_errors = 0;
2460
	bool counting = true;
2461 2462

	/* step one make some room on all the devices */
2463
	devices = &fs_info->fs_devices->devices;
2464
	list_for_each_entry(device, devices, dev_list) {
2465 2466 2467
		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
2468 2469
		if (!device->writeable ||
		    device->total_bytes - device->bytes_used > size_to_free)
2470 2471 2472
			continue;

		ret = btrfs_shrink_device(device, old_size - size_to_free);
2473 2474
		if (ret == -ENOSPC)
			break;
2475 2476
		BUG_ON(ret);

2477
		trans = btrfs_start_transaction(dev_root, 0);
2478
		BUG_ON(IS_ERR(trans));
2479 2480 2481 2482 2483 2484 2485 2486 2487

		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();
2488 2489 2490 2491
	if (!path) {
		ret = -ENOMEM;
		goto error;
	}
2492 2493 2494 2495 2496 2497

	/* zero out stat counters */
	spin_lock(&fs_info->balance_lock);
	memset(&bctl->stat, 0, sizeof(bctl->stat));
	spin_unlock(&fs_info->balance_lock);
again:
2498 2499 2500 2501
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.offset = (u64)-1;
	key.type = BTRFS_CHUNK_ITEM_KEY;

2502
	while (1) {
2503
		if ((!counting && atomic_read(&fs_info->balance_pause_req)) ||
2504
		    atomic_read(&fs_info->balance_cancel_req)) {
2505 2506 2507 2508
			ret = -ECANCELED;
			goto error;
		}

2509 2510 2511 2512 2513 2514 2515 2516 2517
		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)
2518
			BUG(); /* FIXME break ? */
2519 2520 2521

		ret = btrfs_previous_item(chunk_root, path, 0,
					  BTRFS_CHUNK_ITEM_KEY);
2522 2523
		if (ret) {
			ret = 0;
2524
			break;
2525
		}
2526

2527 2528 2529 2530
		leaf = path->nodes[0];
		slot = path->slots[0];
		btrfs_item_key_to_cpu(leaf, &found_key, slot);

2531 2532
		if (found_key.objectid != key.objectid)
			break;
2533

2534
		/* chunk zero is special */
2535
		if (found_key.offset == 0)
2536 2537
			break;

2538 2539
		chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);

2540 2541 2542 2543 2544 2545
		if (!counting) {
			spin_lock(&fs_info->balance_lock);
			bctl->stat.considered++;
			spin_unlock(&fs_info->balance_lock);
		}

2546 2547
		ret = should_balance_chunk(chunk_root, leaf, chunk,
					   found_key.offset);
2548
		btrfs_release_path(path);
2549 2550 2551
		if (!ret)
			goto loop;

2552 2553 2554 2555 2556 2557 2558
		if (counting) {
			spin_lock(&fs_info->balance_lock);
			bctl->stat.expected++;
			spin_unlock(&fs_info->balance_lock);
			goto loop;
		}

2559 2560 2561 2562
		ret = btrfs_relocate_chunk(chunk_root,
					   chunk_root->root_key.objectid,
					   found_key.objectid,
					   found_key.offset);
2563 2564
		if (ret && ret != -ENOSPC)
			goto error;
2565
		if (ret == -ENOSPC) {
2566
			enospc_errors++;
2567 2568 2569 2570 2571
		} else {
			spin_lock(&fs_info->balance_lock);
			bctl->stat.completed++;
			spin_unlock(&fs_info->balance_lock);
		}
2572
loop:
2573
		key.offset = found_key.offset - 1;
2574
	}
2575

2576 2577 2578 2579 2580
	if (counting) {
		btrfs_release_path(path);
		counting = false;
		goto again;
	}
2581 2582
error:
	btrfs_free_path(path);
2583 2584 2585 2586 2587 2588 2589 2590 2591 2592
	if (enospc_errors) {
		printk(KERN_INFO "btrfs: %d enospc errors during balance\n",
		       enospc_errors);
		if (!ret)
			ret = -ENOSPC;
	}

	return ret;
}

2593 2594
static inline int balance_need_close(struct btrfs_fs_info *fs_info)
{
2595 2596 2597 2598
	/* 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);
2599 2600
}

2601 2602
static void __cancel_balance(struct btrfs_fs_info *fs_info)
{
2603 2604
	int ret;

2605
	unset_balance_control(fs_info);
2606 2607
	ret = del_balance_item(fs_info->tree_root);
	BUG_ON(ret);
2608 2609
}

2610
void update_ioctl_balance_args(struct btrfs_fs_info *fs_info, int lock,
2611 2612 2613 2614 2615 2616 2617 2618 2619
			       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;
2620
	u64 allowed;
2621 2622
	int ret;

2623
	if (btrfs_fs_closing(fs_info) ||
2624 2625
	    atomic_read(&fs_info->balance_pause_req) ||
	    atomic_read(&fs_info->balance_cancel_req)) {
2626 2627 2628 2629
		ret = -EINVAL;
		goto out;
	}

2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646
	/*
	 * In case of mixed groups both data and meta should be picked,
	 * and identical options should be given for both of them.
	 */
	allowed = btrfs_super_incompat_flags(fs_info->super_copy);
	if ((allowed & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) &&
	    (bctl->flags & (BTRFS_BALANCE_DATA | BTRFS_BALANCE_METADATA))) {
		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;
		}
	}

2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715
	/*
	 * Profile changing sanity checks.  Skip them if a simple
	 * balance is requested.
	 */
	if (!((bctl->data.flags | bctl->sys.flags | bctl->meta.flags) &
	      BTRFS_BALANCE_ARGS_CONVERT))
		goto do_balance;

	allowed = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
	if (fs_info->fs_devices->num_devices == 1)
		allowed |= BTRFS_BLOCK_GROUP_DUP;
	else if (fs_info->fs_devices->num_devices < 4)
		allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1);
	else
		allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1 |
				BTRFS_BLOCK_GROUP_RAID10);

	if (!profile_is_valid(bctl->data.target, 1) ||
	    bctl->data.target & ~allowed) {
		printk(KERN_ERR "btrfs: unable to start balance with target "
		       "data profile %llu\n",
		       (unsigned long long)bctl->data.target);
		ret = -EINVAL;
		goto out;
	}
	if (!profile_is_valid(bctl->meta.target, 1) ||
	    bctl->meta.target & ~allowed) {
		printk(KERN_ERR "btrfs: unable to start balance with target "
		       "metadata profile %llu\n",
		       (unsigned long long)bctl->meta.target);
		ret = -EINVAL;
		goto out;
	}
	if (!profile_is_valid(bctl->sys.target, 1) ||
	    bctl->sys.target & ~allowed) {
		printk(KERN_ERR "btrfs: unable to start balance with target "
		       "system profile %llu\n",
		       (unsigned long long)bctl->sys.target);
		ret = -EINVAL;
		goto out;
	}

	if (bctl->data.target & BTRFS_BLOCK_GROUP_DUP) {
		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 |
			BTRFS_BLOCK_GROUP_RAID10;
	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;
		}
	}

do_balance:
2716
	ret = insert_balance_item(fs_info->tree_root, bctl);
2717
	if (ret && ret != -EEXIST)
2718 2719
		goto out;

2720 2721 2722 2723 2724 2725 2726 2727 2728
	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);
	}
2729

2730
	atomic_inc(&fs_info->balance_running);
2731 2732 2733 2734 2735
	mutex_unlock(&fs_info->balance_mutex);

	ret = __btrfs_balance(fs_info);

	mutex_lock(&fs_info->balance_mutex);
2736
	atomic_dec(&fs_info->balance_running);
2737 2738 2739

	if (bargs) {
		memset(bargs, 0, sizeof(*bargs));
2740
		update_ioctl_balance_args(fs_info, 0, bargs);
2741 2742
	}

2743 2744 2745 2746 2747 2748
	if ((ret && ret != -ECANCELED && ret != -ENOSPC) ||
	    balance_need_close(fs_info)) {
		__cancel_balance(fs_info);
	}

	wake_up(&fs_info->balance_wait_q);
2749 2750 2751

	return ret;
out:
2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763
	if (bctl->flags & BTRFS_BALANCE_RESUME)
		__cancel_balance(fs_info);
	else
		kfree(bctl);
	return ret;
}

static int balance_kthread(void *data)
{
	struct btrfs_balance_control *bctl =
			(struct btrfs_balance_control *)data;
	struct btrfs_fs_info *fs_info = bctl->fs_info;
2764
	int ret = 0;
2765 2766 2767 2768 2769 2770

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

	set_balance_control(bctl);

2771 2772 2773 2774 2775 2776
	if (btrfs_test_opt(fs_info->tree_root, SKIP_BALANCE)) {
		printk(KERN_INFO "btrfs: force skipping balance\n");
	} else {
		printk(KERN_INFO "btrfs: continuing balance\n");
		ret = btrfs_balance(bctl, NULL);
	}
2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835

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

int btrfs_recover_balance(struct btrfs_root *tree_root)
{
	struct task_struct *tsk;
	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;

	bctl = kzalloc(sizeof(*bctl), GFP_NOFS);
	if (!bctl) {
		ret = -ENOMEM;
		goto out;
	}

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

	ret = btrfs_search_slot(NULL, tree_root, &key, path, 0, 0);
	if (ret < 0)
		goto out_bctl;
	if (ret > 0) { /* ret = -ENOENT; */
		ret = 0;
		goto out_bctl;
	}

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

	bctl->fs_info = tree_root->fs_info;
	bctl->flags = btrfs_balance_flags(leaf, item) | BTRFS_BALANCE_RESUME;

	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);

	tsk = kthread_run(balance_kthread, bctl, "btrfs-balance");
	if (IS_ERR(tsk))
		ret = PTR_ERR(tsk);
	else
		goto out;

out_bctl:
2836
	kfree(bctl);
2837 2838
out:
	btrfs_free_path(path);
2839 2840 2841
	return ret;
}

2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870
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;
}

2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906
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;
}

2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923
/*
 * 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;
2924 2925
	int failed = 0;
	bool retried = false;
2926 2927
	struct extent_buffer *l;
	struct btrfs_key key;
2928
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
2929
	u64 old_total = btrfs_super_total_bytes(super_copy);
2930
	u64 old_size = device->total_bytes;
2931 2932
	u64 diff = device->total_bytes - new_size;

Y
Yan Zheng 已提交
2933 2934
	if (new_size >= device->total_bytes)
		return -EINVAL;
2935 2936 2937 2938 2939 2940 2941

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

	path->reada = 2;

2942 2943
	lock_chunks(root);

2944
	device->total_bytes = new_size;
2945
	if (device->writeable) {
Y
Yan Zheng 已提交
2946
		device->fs_devices->total_rw_bytes -= diff;
2947 2948 2949 2950
		spin_lock(&root->fs_info->free_chunk_lock);
		root->fs_info->free_chunk_space -= diff;
		spin_unlock(&root->fs_info->free_chunk_lock);
	}
2951
	unlock_chunks(root);
2952

2953
again:
2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967
	key.objectid = device->devid;
	key.offset = (u64)-1;
	key.type = BTRFS_DEV_EXTENT_KEY;

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

		ret = btrfs_previous_item(root, path, 0, key.type);
		if (ret < 0)
			goto done;
		if (ret) {
			ret = 0;
2968
			btrfs_release_path(path);
2969
			break;
2970 2971 2972 2973 2974 2975
		}

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

2976
		if (key.objectid != device->devid) {
2977
			btrfs_release_path(path);
2978
			break;
2979
		}
2980 2981 2982 2983

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

2984
		if (key.offset + length <= new_size) {
2985
			btrfs_release_path(path);
2986
			break;
2987
		}
2988 2989 2990 2991

		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);
2992
		btrfs_release_path(path);
2993 2994 2995

		ret = btrfs_relocate_chunk(root, chunk_tree, chunk_objectid,
					   chunk_offset);
2996
		if (ret && ret != -ENOSPC)
2997
			goto done;
2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013
		if (ret == -ENOSPC)
			failed++;
		key.offset -= 1;
	}

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

		device->total_bytes = old_size;
		if (device->writeable)
			device->fs_devices->total_rw_bytes += diff;
3014 3015 3016
		spin_lock(&root->fs_info->free_chunk_lock);
		root->fs_info->free_chunk_space += diff;
		spin_unlock(&root->fs_info->free_chunk_lock);
3017 3018
		unlock_chunks(root);
		goto done;
3019 3020
	}

3021
	/* Shrinking succeeded, else we would be at "done". */
3022
	trans = btrfs_start_transaction(root, 0);
3023 3024 3025 3026 3027
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto done;
	}

3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041
	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);
3042 3043 3044 3045 3046
done:
	btrfs_free_path(path);
	return ret;
}

3047
static int btrfs_add_system_chunk(struct btrfs_trans_handle *trans,
3048 3049 3050 3051
			   struct btrfs_root *root,
			   struct btrfs_key *key,
			   struct btrfs_chunk *chunk, int item_size)
{
3052
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070
	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;
}

3071 3072 3073 3074
/*
 * sort the devices in descending order by max_avail, total_avail
 */
static int btrfs_cmp_device_info(const void *a, const void *b)
3075
{
3076 3077
	const struct btrfs_device_info *di_a = a;
	const struct btrfs_device_info *di_b = b;
3078

3079
	if (di_a->max_avail > di_b->max_avail)
3080
		return -1;
3081
	if (di_a->max_avail < di_b->max_avail)
3082
		return 1;
3083 3084 3085 3086 3087
	if (di_a->total_avail > di_b->total_avail)
		return -1;
	if (di_a->total_avail < di_b->total_avail)
		return 1;
	return 0;
3088
}
3089

3090 3091 3092 3093 3094
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)
3095
{
3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118
	struct btrfs_fs_info *info = extent_root->fs_info;
	struct btrfs_fs_devices *fs_devices = info->fs_devices;
	struct list_head *cur;
	struct map_lookup *map = NULL;
	struct extent_map_tree *em_tree;
	struct extent_map *em;
	struct btrfs_device_info *devices_info = NULL;
	u64 total_avail;
	int num_stripes;	/* total number of stripes to allocate */
	int sub_stripes;	/* sub_stripes info for map */
	int dev_stripes;	/* stripes per dev */
	int devs_max;		/* max devs to use */
	int devs_min;		/* min devs needed */
	int devs_increment;	/* ndevs has to be a multiple of this */
	int ncopies;		/* how many copies to data has */
	int ret;
	u64 max_stripe_size;
	u64 max_chunk_size;
	u64 stripe_size;
	u64 num_bytes;
	int ndevs;
	int i;
	int j;
3119

3120 3121 3122 3123
	if ((type & BTRFS_BLOCK_GROUP_RAID1) &&
	    (type & BTRFS_BLOCK_GROUP_DUP)) {
		WARN_ON(1);
		type &= ~BTRFS_BLOCK_GROUP_DUP;
C
Chris Mason 已提交
3124
	}
3125

3126 3127
	if (list_empty(&fs_devices->alloc_list))
		return -ENOSPC;
3128

3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142
	sub_stripes = 1;
	dev_stripes = 1;
	devs_increment = 1;
	ncopies = 1;
	devs_max = 0;	/* 0 == as many as possible */
	devs_min = 1;

	/*
	 * define the properties of each RAID type.
	 * FIXME: move this to a global table and use it in all RAID
	 * calculation code
	 */
	if (type & (BTRFS_BLOCK_GROUP_DUP)) {
		dev_stripes = 2;
3143
		ncopies = 2;
3144 3145 3146 3147 3148
		devs_max = 1;
	} else if (type & (BTRFS_BLOCK_GROUP_RAID0)) {
		devs_min = 2;
	} else if (type & (BTRFS_BLOCK_GROUP_RAID1)) {
		devs_increment = 2;
3149
		ncopies = 2;
3150 3151 3152 3153 3154 3155 3156 3157 3158 3159
		devs_max = 2;
		devs_min = 2;
	} else if (type & (BTRFS_BLOCK_GROUP_RAID10)) {
		sub_stripes = 2;
		devs_increment = 2;
		ncopies = 2;
		devs_min = 4;
	} else {
		devs_max = 1;
	}
3160

3161
	if (type & BTRFS_BLOCK_GROUP_DATA) {
3162 3163
		max_stripe_size = 1024 * 1024 * 1024;
		max_chunk_size = 10 * max_stripe_size;
3164
	} else if (type & BTRFS_BLOCK_GROUP_METADATA) {
3165 3166
		max_stripe_size = 256 * 1024 * 1024;
		max_chunk_size = max_stripe_size;
3167
	} else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
3168 3169 3170 3171 3172 3173
		max_stripe_size = 8 * 1024 * 1024;
		max_chunk_size = 2 * max_stripe_size;
	} else {
		printk(KERN_ERR "btrfs: invalid chunk type 0x%llx requested\n",
		       type);
		BUG_ON(1);
3174 3175
	}

Y
Yan Zheng 已提交
3176 3177 3178
	/* 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);
3179

3180 3181 3182 3183
	devices_info = kzalloc(sizeof(*devices_info) * fs_devices->rw_devices,
			       GFP_NOFS);
	if (!devices_info)
		return -ENOMEM;
3184

3185
	cur = fs_devices->alloc_list.next;
3186

3187
	/*
3188 3189
	 * in the first pass through the devices list, we gather information
	 * about the available holes on each device.
3190
	 */
3191 3192 3193 3194 3195
	ndevs = 0;
	while (cur != &fs_devices->alloc_list) {
		struct btrfs_device *device;
		u64 max_avail;
		u64 dev_offset;
3196

3197
		device = list_entry(cur, struct btrfs_device, dev_alloc_list);
3198

3199
		cur = cur->next;
3200

3201 3202 3203 3204 3205 3206
		if (!device->writeable) {
			printk(KERN_ERR
			       "btrfs: read-only device in alloc_list\n");
			WARN_ON(1);
			continue;
		}
3207

3208 3209
		if (!device->in_fs_metadata)
			continue;
3210

3211 3212 3213 3214
		if (device->total_bytes > device->bytes_used)
			total_avail = device->total_bytes - device->bytes_used;
		else
			total_avail = 0;
3215 3216 3217 3218

		/* If there is no space on this device, skip it. */
		if (total_avail == 0)
			continue;
3219

3220 3221 3222 3223 3224
		ret = find_free_dev_extent(trans, device,
					   max_stripe_size * dev_stripes,
					   &dev_offset, &max_avail);
		if (ret && ret != -ENOSPC)
			goto error;
3225

3226 3227
		if (ret == 0)
			max_avail = max_stripe_size * dev_stripes;
3228

3229 3230
		if (max_avail < BTRFS_STRIPE_LEN * dev_stripes)
			continue;
3231

3232 3233 3234 3235 3236 3237
		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;
	}
3238

3239 3240 3241 3242 3243
	/*
	 * now sort the devices by hole size / available space
	 */
	sort(devices_info, ndevs, sizeof(struct btrfs_device_info),
	     btrfs_cmp_device_info, NULL);
3244

3245 3246
	/* round down to number of usable stripes */
	ndevs -= ndevs % devs_increment;
3247

3248 3249 3250
	if (ndevs < devs_increment * sub_stripes || ndevs < devs_min) {
		ret = -ENOSPC;
		goto error;
3251
	}
3252

3253 3254 3255 3256 3257 3258 3259 3260
	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;
3261

3262 3263 3264
	if (stripe_size * num_stripes > max_chunk_size * ncopies) {
		stripe_size = max_chunk_size * ncopies;
		do_div(stripe_size, num_stripes);
3265 3266
	}

3267 3268 3269
	do_div(stripe_size, dev_stripes);
	do_div(stripe_size, BTRFS_STRIPE_LEN);
	stripe_size *= BTRFS_STRIPE_LEN;
3270 3271 3272 3273 3274 3275 3276

	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
	if (!map) {
		ret = -ENOMEM;
		goto error;
	}
	map->num_stripes = num_stripes;
3277

3278 3279 3280 3281 3282 3283
	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;
3284 3285
		}
	}
Y
Yan Zheng 已提交
3286
	map->sector_size = extent_root->sectorsize;
3287 3288 3289
	map->stripe_len = BTRFS_STRIPE_LEN;
	map->io_align = BTRFS_STRIPE_LEN;
	map->io_width = BTRFS_STRIPE_LEN;
Y
Yan Zheng 已提交
3290 3291
	map->type = type;
	map->sub_stripes = sub_stripes;
3292

Y
Yan Zheng 已提交
3293
	*map_ret = map;
3294
	num_bytes = stripe_size * (num_stripes / ncopies);
3295

3296 3297
	*stripe_size_out = stripe_size;
	*num_bytes_out = num_bytes;
3298

3299
	trace_btrfs_chunk_alloc(info->chunk_root, map, start, num_bytes);
3300

3301
	em = alloc_extent_map();
Y
Yan Zheng 已提交
3302
	if (!em) {
3303 3304
		ret = -ENOMEM;
		goto error;
3305
	}
Y
Yan Zheng 已提交
3306 3307
	em->bdev = (struct block_device *)map;
	em->start = start;
3308
	em->len = num_bytes;
Y
Yan Zheng 已提交
3309 3310
	em->block_start = 0;
	em->block_len = em->len;
3311

Y
Yan Zheng 已提交
3312
	em_tree = &extent_root->fs_info->mapping_tree.map_tree;
3313
	write_lock(&em_tree->lock);
Y
Yan Zheng 已提交
3314
	ret = add_extent_mapping(em_tree, em);
3315
	write_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
3316 3317
	BUG_ON(ret);
	free_extent_map(em);
3318

Y
Yan Zheng 已提交
3319 3320
	ret = btrfs_make_block_group(trans, extent_root, 0, type,
				     BTRFS_FIRST_CHUNK_TREE_OBJECTID,
3321
				     start, num_bytes);
Y
Yan Zheng 已提交
3322
	BUG_ON(ret);
3323

3324 3325 3326 3327 3328 3329
	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;
3330 3331

		ret = btrfs_alloc_dev_extent(trans, device,
Y
Yan Zheng 已提交
3332 3333
				info->chunk_root->root_key.objectid,
				BTRFS_FIRST_CHUNK_TREE_OBJECTID,
3334
				start, dev_offset, stripe_size);
3335
		BUG_ON(ret);
Y
Yan Zheng 已提交
3336 3337
	}

3338
	kfree(devices_info);
Y
Yan Zheng 已提交
3339
	return 0;
3340 3341 3342 3343 3344

error:
	kfree(map);
	kfree(devices_info);
	return ret;
Y
Yan Zheng 已提交
3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369
}

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;
3370 3371
		ret = btrfs_update_device(trans, device);
		BUG_ON(ret);
Y
Yan Zheng 已提交
3372 3373 3374
		index++;
	}

3375 3376 3377 3378 3379
	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
3380 3381 3382 3383 3384
	index = 0;
	stripe = &chunk->stripe;
	while (index < map->num_stripes) {
		device = map->stripes[index].dev;
		dev_offset = map->stripes[index].physical;
3385

3386 3387 3388
		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
3389
		stripe++;
3390 3391 3392
		index++;
	}

Y
Yan Zheng 已提交
3393
	btrfs_set_stack_chunk_length(chunk, chunk_size);
3394
	btrfs_set_stack_chunk_owner(chunk, extent_root->root_key.objectid);
Y
Yan Zheng 已提交
3395 3396 3397 3398 3399
	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);
3400
	btrfs_set_stack_chunk_sector_size(chunk, extent_root->sectorsize);
Y
Yan Zheng 已提交
3401
	btrfs_set_stack_chunk_sub_stripes(chunk, map->sub_stripes);
3402

Y
Yan Zheng 已提交
3403 3404 3405
	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
	key.type = BTRFS_CHUNK_ITEM_KEY;
	key.offset = chunk_offset;
3406

Y
Yan Zheng 已提交
3407 3408
	ret = btrfs_insert_item(trans, chunk_root, &key, chunk, item_size);
	BUG_ON(ret);
3409

Y
Yan Zheng 已提交
3410 3411 3412
	if (map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
		ret = btrfs_add_system_chunk(trans, chunk_root, &key, chunk,
					     item_size);
3413 3414
		BUG_ON(ret);
	}
3415

3416
	kfree(chunk);
Y
Yan Zheng 已提交
3417 3418
	return 0;
}
3419

Y
Yan Zheng 已提交
3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452
/*
 * Chunk allocation falls into two parts. The first part does works
 * that make the new allocated chunk useable, but not do any operation
 * that modifies the chunk tree. The second part does the works that
 * require modifying the chunk tree. This division is important for the
 * bootstrap process of adding storage to a seed btrfs.
 */
int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
		      struct btrfs_root *extent_root, u64 type)
{
	u64 chunk_offset;
	u64 chunk_size;
	u64 stripe_size;
	struct map_lookup *map;
	struct btrfs_root *chunk_root = extent_root->fs_info->chunk_root;
	int ret;

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

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

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

3453
static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
Y
Yan Zheng 已提交
3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471
					 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);
3472 3473
	if (ret)
		return ret;
Y
Yan Zheng 已提交
3474 3475

	alloc_profile = BTRFS_BLOCK_GROUP_METADATA |
3476
				fs_info->avail_metadata_alloc_bits;
Y
Yan Zheng 已提交
3477 3478 3479 3480 3481 3482 3483 3484 3485
	alloc_profile = btrfs_reduce_alloc_profile(root, alloc_profile);

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

	sys_chunk_offset = chunk_offset + chunk_size;

	alloc_profile = BTRFS_BLOCK_GROUP_SYSTEM |
3486
				fs_info->avail_system_alloc_bits;
Y
Yan Zheng 已提交
3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509
	alloc_profile = btrfs_reduce_alloc_profile(root, alloc_profile);

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

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

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

	ret = __finish_chunk_alloc(trans, extent_root, sys_map,
				   sys_chunk_offset, sys_chunk_size,
				   sys_stripe_size);
3510
	BUG_ON(ret);
Y
Yan Zheng 已提交
3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521
	return 0;
}

int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset)
{
	struct extent_map *em;
	struct map_lookup *map;
	struct btrfs_mapping_tree *map_tree = &root->fs_info->mapping_tree;
	int readonly = 0;
	int i;

3522
	read_lock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
3523
	em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
3524
	read_unlock(&map_tree->map_tree.lock);
Y
Yan Zheng 已提交
3525 3526 3527
	if (!em)
		return 1;

3528 3529 3530 3531 3532
	if (btrfs_test_opt(root, DEGRADED)) {
		free_extent_map(em);
		return 0;
	}

Y
Yan Zheng 已提交
3533 3534 3535 3536 3537 3538 3539
	map = (struct map_lookup *)em->bdev;
	for (i = 0; i < map->num_stripes; i++) {
		if (!map->stripes[i].dev->writeable) {
			readonly = 1;
			break;
		}
	}
3540
	free_extent_map(em);
Y
Yan Zheng 已提交
3541
	return readonly;
3542 3543 3544 3545
}

void btrfs_mapping_init(struct btrfs_mapping_tree *tree)
{
3546
	extent_map_tree_init(&tree->map_tree);
3547 3548 3549 3550 3551 3552
}

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

3553
	while (1) {
3554
		write_lock(&tree->map_tree.lock);
3555 3556 3557
		em = lookup_extent_mapping(&tree->map_tree, 0, (u64)-1);
		if (em)
			remove_extent_mapping(&tree->map_tree, em);
3558
		write_unlock(&tree->map_tree.lock);
3559 3560 3561 3562 3563 3564 3565 3566 3567 3568
		if (!em)
			break;
		kfree(em->bdev);
		/* once for us */
		free_extent_map(em);
		/* once for the tree */
		free_extent_map(em);
	}
}

3569 3570 3571 3572 3573 3574 3575
int btrfs_num_copies(struct btrfs_mapping_tree *map_tree, u64 logical, u64 len)
{
	struct extent_map *em;
	struct map_lookup *map;
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	int ret;

3576
	read_lock(&em_tree->lock);
3577
	em = lookup_extent_mapping(em_tree, logical, len);
3578
	read_unlock(&em_tree->lock);
3579 3580 3581 3582 3583 3584
	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
3585 3586
	else if (map->type & BTRFS_BLOCK_GROUP_RAID10)
		ret = map->sub_stripes;
3587 3588 3589 3590 3591 3592
	else
		ret = 1;
	free_extent_map(em);
	return ret;
}

3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608
static int find_live_mirror(struct map_lookup *map, int first, int num,
			    int optimal)
{
	int i;
	if (map->stripes[optimal].dev->bdev)
		return optimal;
	for (i = first; i < first + num; i++) {
		if (map->stripes[i].dev->bdev)
			return i;
	}
	/* we couldn't find one that doesn't fail.  Just return something
	 * and the io error handling code will clean up eventually
	 */
	return optimal;
}

3609 3610
static int __btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
			     u64 logical, u64 *length,
3611
			     struct btrfs_bio **bbio_ret,
3612
			     int mirror_num)
3613 3614 3615 3616 3617
{
	struct extent_map *em;
	struct map_lookup *map;
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	u64 offset;
3618
	u64 stripe_offset;
3619
	u64 stripe_end_offset;
3620
	u64 stripe_nr;
3621 3622
	u64 stripe_nr_orig;
	u64 stripe_nr_end;
3623
	int stripes_allocated = 8;
C
Chris Mason 已提交
3624
	int stripes_required = 1;
3625
	int stripe_index;
3626
	int i;
3627
	int num_stripes;
3628
	int max_errors = 0;
3629
	struct btrfs_bio *bbio = NULL;
3630

3631
	if (bbio_ret && !(rw & (REQ_WRITE | REQ_DISCARD)))
3632 3633
		stripes_allocated = 1;
again:
3634 3635
	if (bbio_ret) {
		bbio = kzalloc(btrfs_bio_size(stripes_allocated),
3636
				GFP_NOFS);
3637
		if (!bbio)
3638
			return -ENOMEM;
3639

3640
		atomic_set(&bbio->error, 0);
3641
	}
3642

3643
	read_lock(&em_tree->lock);
3644
	em = lookup_extent_mapping(em_tree, logical, *length);
3645
	read_unlock(&em_tree->lock);
3646

3647
	if (!em) {
3648 3649 3650
		printk(KERN_CRIT "unable to find logical %llu len %llu\n",
		       (unsigned long long)logical,
		       (unsigned long long)*length);
3651
		BUG();
3652
	}
3653 3654 3655 3656

	BUG_ON(em->start > logical || em->start + em->len < logical);
	map = (struct map_lookup *)em->bdev;
	offset = logical - em->start;
3657

3658 3659 3660
	if (mirror_num > map->num_stripes)
		mirror_num = 0;

3661
	/* if our btrfs_bio struct is too small, back off and try again */
3662
	if (rw & REQ_WRITE) {
C
Chris Mason 已提交
3663 3664 3665
		if (map->type & (BTRFS_BLOCK_GROUP_RAID1 |
				 BTRFS_BLOCK_GROUP_DUP)) {
			stripes_required = map->num_stripes;
3666
			max_errors = 1;
C
Chris Mason 已提交
3667 3668
		} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
			stripes_required = map->sub_stripes;
3669
			max_errors = 1;
C
Chris Mason 已提交
3670 3671
		}
	}
3672
	if (rw & REQ_DISCARD) {
3673
		if (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK)
3674 3675
			stripes_required = map->num_stripes;
	}
3676
	if (bbio_ret && (rw & (REQ_WRITE | REQ_DISCARD)) &&
C
Chris Mason 已提交
3677
	    stripes_allocated < stripes_required) {
3678 3679
		stripes_allocated = map->num_stripes;
		free_extent_map(em);
3680
		kfree(bbio);
3681 3682
		goto again;
	}
3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695
	stripe_nr = offset;
	/*
	 * stripe_nr counts the total number of stripes we have to stride
	 * to get to this block
	 */
	do_div(stripe_nr, map->stripe_len);

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

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

3696 3697
	if (rw & REQ_DISCARD)
		*length = min_t(u64, em->len - offset, *length);
3698
	else if (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK) {
3699 3700
		/* we limit the length of each bio to what fits in a stripe */
		*length = min_t(u64, em->len - offset,
3701
				map->stripe_len - stripe_offset);
3702 3703 3704
	} else {
		*length = em->len - offset;
	}
3705

3706
	if (!bbio_ret)
3707 3708
		goto out;

3709
	num_stripes = 1;
3710
	stripe_index = 0;
3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722
	stripe_nr_orig = stripe_nr;
	stripe_nr_end = (offset + *length + map->stripe_len - 1) &
			(~(map->stripe_len - 1));
	do_div(stripe_nr_end, map->stripe_len);
	stripe_end_offset = stripe_nr_end * map->stripe_len -
			    (offset + *length);
	if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
		if (rw & REQ_DISCARD)
			num_stripes = min_t(u64, map->num_stripes,
					    stripe_nr_end - stripe_nr_orig);
		stripe_index = do_div(stripe_nr, map->num_stripes);
	} else if (map->type & BTRFS_BLOCK_GROUP_RAID1) {
3723
		if (rw & (REQ_WRITE | REQ_DISCARD))
3724
			num_stripes = map->num_stripes;
3725
		else if (mirror_num)
3726
			stripe_index = mirror_num - 1;
3727 3728 3729 3730
		else {
			stripe_index = find_live_mirror(map, 0,
					    map->num_stripes,
					    current->pid % map->num_stripes);
3731
			mirror_num = stripe_index + 1;
3732
		}
3733

3734
	} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
3735
		if (rw & (REQ_WRITE | REQ_DISCARD)) {
3736
			num_stripes = map->num_stripes;
3737
		} else if (mirror_num) {
3738
			stripe_index = mirror_num - 1;
3739 3740 3741
		} else {
			mirror_num = 1;
		}
3742

C
Chris Mason 已提交
3743 3744 3745 3746 3747 3748
	} 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;

3749
		if (rw & REQ_WRITE)
3750
			num_stripes = map->sub_stripes;
3751 3752 3753 3754
		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
3755 3756
		else if (mirror_num)
			stripe_index += mirror_num - 1;
3757 3758 3759 3760
		else {
			stripe_index = find_live_mirror(map, stripe_index,
					      map->sub_stripes, stripe_index +
					      current->pid % map->sub_stripes);
3761
			mirror_num = stripe_index + 1;
3762
		}
3763 3764 3765 3766 3767 3768 3769
	} 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);
3770
		mirror_num = stripe_index + 1;
3771
	}
3772
	BUG_ON(stripe_index >= map->num_stripes);
3773

3774 3775
	if (rw & REQ_DISCARD) {
		for (i = 0; i < num_stripes; i++) {
3776
			bbio->stripes[i].physical =
3777 3778
				map->stripes[stripe_index].physical +
				stripe_offset + stripe_nr * map->stripe_len;
3779
			bbio->stripes[i].dev = map->stripes[stripe_index].dev;
3780 3781 3782

			if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
				u64 stripes;
3783
				u32 last_stripe = 0;
3784 3785
				int j;

3786 3787 3788 3789
				div_u64_rem(stripe_nr_end - 1,
					    map->num_stripes,
					    &last_stripe);

3790
				for (j = 0; j < map->num_stripes; j++) {
3791 3792 3793 3794 3795
					u32 test;

					div_u64_rem(stripe_nr_end - 1 - j,
						    map->num_stripes, &test);
					if (test == stripe_index)
3796 3797 3798 3799
						break;
				}
				stripes = stripe_nr_end - 1 - j;
				do_div(stripes, map->num_stripes);
3800
				bbio->stripes[i].length = map->stripe_len *
3801 3802 3803
					(stripes - stripe_nr + 1);

				if (i == 0) {
3804
					bbio->stripes[i].length -=
3805 3806 3807 3808
						stripe_offset;
					stripe_offset = 0;
				}
				if (stripe_index == last_stripe)
3809
					bbio->stripes[i].length -=
3810 3811 3812 3813 3814 3815
						stripe_end_offset;
			} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
				u64 stripes;
				int j;
				int factor = map->num_stripes /
					     map->sub_stripes;
3816 3817 3818 3819
				u32 last_stripe = 0;

				div_u64_rem(stripe_nr_end - 1,
					    factor, &last_stripe);
3820 3821 3822
				last_stripe *= map->sub_stripes;

				for (j = 0; j < factor; j++) {
3823 3824 3825 3826 3827 3828
					u32 test;

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

					if (test ==
3829 3830 3831 3832 3833
					    stripe_index / map->sub_stripes)
						break;
				}
				stripes = stripe_nr_end - 1 - j;
				do_div(stripes, factor);
3834
				bbio->stripes[i].length = map->stripe_len *
3835 3836 3837
					(stripes - stripe_nr + 1);

				if (i < map->sub_stripes) {
3838
					bbio->stripes[i].length -=
3839 3840 3841 3842 3843 3844 3845
						stripe_offset;
					if (i == map->sub_stripes - 1)
						stripe_offset = 0;
				}
				if (stripe_index >= last_stripe &&
				    stripe_index <= (last_stripe +
						     map->sub_stripes - 1)) {
3846
					bbio->stripes[i].length -=
3847 3848 3849
						stripe_end_offset;
				}
			} else
3850
				bbio->stripes[i].length = *length;
3851 3852 3853 3854 3855 3856 3857 3858 3859 3860

			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++) {
3861
			bbio->stripes[i].physical =
3862 3863 3864
				map->stripes[stripe_index].physical +
				stripe_offset +
				stripe_nr * map->stripe_len;
3865
			bbio->stripes[i].dev =
3866
				map->stripes[stripe_index].dev;
3867
			stripe_index++;
3868
		}
3869
	}
3870 3871 3872 3873 3874
	if (bbio_ret) {
		*bbio_ret = bbio;
		bbio->num_stripes = num_stripes;
		bbio->max_errors = max_errors;
		bbio->mirror_num = mirror_num;
3875
	}
3876
out:
3877 3878 3879 3880
	free_extent_map(em);
	return 0;
}

3881 3882
int btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
		      u64 logical, u64 *length,
3883
		      struct btrfs_bio **bbio_ret, int mirror_num)
3884
{
3885
	return __btrfs_map_block(map_tree, rw, logical, length, bbio_ret,
3886
				 mirror_num);
3887 3888
}

Y
Yan Zheng 已提交
3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901
int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
		     u64 chunk_start, u64 physical, u64 devid,
		     u64 **logical, int *naddrs, int *stripe_len)
{
	struct extent_map_tree *em_tree = &map_tree->map_tree;
	struct extent_map *em;
	struct map_lookup *map;
	u64 *buf;
	u64 bytenr;
	u64 length;
	u64 stripe_nr;
	int i, j, nr = 0;

3902
	read_lock(&em_tree->lock);
Y
Yan Zheng 已提交
3903
	em = lookup_extent_mapping(em_tree, chunk_start, 1);
3904
	read_unlock(&em_tree->lock);
Y
Yan Zheng 已提交
3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934

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

	length = em->len;
	if (map->type & BTRFS_BLOCK_GROUP_RAID10)
		do_div(length, map->num_stripes / map->sub_stripes);
	else if (map->type & BTRFS_BLOCK_GROUP_RAID0)
		do_div(length, map->num_stripes);

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

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

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

		if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
			stripe_nr = stripe_nr * map->num_stripes + i;
			do_div(stripe_nr, map->sub_stripes);
		} else if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
			stripe_nr = stripe_nr * map->num_stripes + i;
		}
		bytenr = chunk_start + stripe_nr * map->stripe_len;
3935
		WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
3936 3937 3938 3939
		for (j = 0; j < nr; j++) {
			if (buf[j] == bytenr)
				break;
		}
3940 3941
		if (j == nr) {
			WARN_ON(nr >= map->num_stripes);
Y
Yan Zheng 已提交
3942
			buf[nr++] = bytenr;
3943
		}
Y
Yan Zheng 已提交
3944 3945 3946 3947 3948 3949 3950 3951
	}

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

	free_extent_map(em);
	return 0;
3952 3953
}

3954
static void btrfs_end_bio(struct bio *bio, int err)
3955
{
3956
	struct btrfs_bio *bbio = bio->bi_private;
3957
	int is_orig_bio = 0;
3958 3959

	if (err)
3960
		atomic_inc(&bbio->error);
3961

3962
	if (bio == bbio->orig_bio)
3963 3964
		is_orig_bio = 1;

3965
	if (atomic_dec_and_test(&bbio->stripes_pending)) {
3966 3967
		if (!is_orig_bio) {
			bio_put(bio);
3968
			bio = bbio->orig_bio;
3969
		}
3970 3971
		bio->bi_private = bbio->private;
		bio->bi_end_io = bbio->end_io;
3972 3973
		bio->bi_bdev = (struct block_device *)
					(unsigned long)bbio->mirror_num;
3974 3975 3976
		/* only send an error to the higher layers if it is
		 * beyond the tolerance of the multi-bio
		 */
3977
		if (atomic_read(&bbio->error) > bbio->max_errors) {
3978
			err = -EIO;
3979
		} else {
3980 3981 3982 3983 3984
			/*
			 * this bio is actually up to date, we didn't
			 * go over the max number of errors
			 */
			set_bit(BIO_UPTODATE, &bio->bi_flags);
3985
			err = 0;
3986
		}
3987
		kfree(bbio);
3988 3989

		bio_endio(bio, err);
3990
	} else if (!is_orig_bio) {
3991 3992 3993 3994
		bio_put(bio);
	}
}

3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008
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.
 */
4009
static noinline int schedule_bio(struct btrfs_root *root,
4010 4011
				 struct btrfs_device *device,
				 int rw, struct bio *bio)
4012 4013
{
	int should_queue = 1;
4014
	struct btrfs_pending_bios *pending_bios;
4015 4016

	/* don't bother with additional async steps for reads, right now */
4017
	if (!(rw & REQ_WRITE)) {
4018
		bio_get(bio);
4019
		submit_bio(rw, bio);
4020
		bio_put(bio);
4021 4022 4023 4024
		return 0;
	}

	/*
4025
	 * nr_async_bios allows us to reliably return congestion to the
4026 4027 4028 4029
	 * 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
	 */
4030
	atomic_inc(&root->fs_info->nr_async_bios);
4031
	WARN_ON(bio->bi_next);
4032 4033 4034 4035
	bio->bi_next = NULL;
	bio->bi_rw |= rw;

	spin_lock(&device->io_lock);
4036
	if (bio->bi_rw & REQ_SYNC)
4037 4038 4039
		pending_bios = &device->pending_sync_bios;
	else
		pending_bios = &device->pending_bios;
4040

4041 4042
	if (pending_bios->tail)
		pending_bios->tail->bi_next = bio;
4043

4044 4045 4046
	pending_bios->tail = bio;
	if (!pending_bios->head)
		pending_bios->head = bio;
4047 4048 4049 4050 4051 4052
	if (device->running_pending)
		should_queue = 0;

	spin_unlock(&device->io_lock);

	if (should_queue)
4053 4054
		btrfs_queue_worker(&root->fs_info->submit_workers,
				   &device->work);
4055 4056 4057
	return 0;
}

4058
int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
4059
		  int mirror_num, int async_submit)
4060 4061 4062
{
	struct btrfs_mapping_tree *map_tree;
	struct btrfs_device *dev;
4063
	struct bio *first_bio = bio;
4064
	u64 logical = (u64)bio->bi_sector << 9;
4065 4066 4067
	u64 length = 0;
	u64 map_length;
	int ret;
4068 4069
	int dev_nr = 0;
	int total_devs = 1;
4070
	struct btrfs_bio *bbio = NULL;
4071

4072
	length = bio->bi_size;
4073 4074
	map_tree = &root->fs_info->mapping_tree;
	map_length = length;
4075

4076
	ret = btrfs_map_block(map_tree, rw, logical, &map_length, &bbio,
4077
			      mirror_num);
4078 4079
	BUG_ON(ret);

4080
	total_devs = bbio->num_stripes;
4081
	if (map_length < length) {
4082 4083 4084 4085
		printk(KERN_CRIT "mapping failed logical %llu bio len %llu "
		       "len %llu\n", (unsigned long long)logical,
		       (unsigned long long)length,
		       (unsigned long long)map_length);
4086 4087
		BUG();
	}
4088 4089 4090 4091 4092

	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);
4093

4094
	while (dev_nr < total_devs) {
4095 4096 4097 4098 4099
		if (dev_nr < total_devs - 1) {
			bio = bio_clone(first_bio, GFP_NOFS);
			BUG_ON(!bio);
		} else {
			bio = first_bio;
4100
		}
4101 4102 4103 4104
		bio->bi_private = bbio;
		bio->bi_end_io = btrfs_end_bio;
		bio->bi_sector = bbio->stripes[dev_nr].physical >> 9;
		dev = bbio->stripes[dev_nr].dev;
4105
		if (dev && dev->bdev && (rw != WRITE || dev->writeable)) {
4106 4107 4108 4109
			pr_debug("btrfs_map_bio: rw %d, secor=%llu, dev=%lu "
				 "(%s id %llu), size=%u\n", rw,
				 (u64)bio->bi_sector, (u_long)dev->bdev->bd_dev,
				 dev->name, dev->devid, bio->bi_size);
4110
			bio->bi_bdev = dev->bdev;
4111 4112 4113 4114
			if (async_submit)
				schedule_bio(root, dev, rw, bio);
			else
				submit_bio(rw, bio);
4115 4116 4117 4118 4119
		} else {
			bio->bi_bdev = root->fs_info->fs_devices->latest_bdev;
			bio->bi_sector = logical >> 9;
			bio_endio(bio, -EIO);
		}
4120 4121
		dev_nr++;
	}
4122 4123 4124
	return 0;
}

4125
struct btrfs_device *btrfs_find_device(struct btrfs_root *root, u64 devid,
Y
Yan Zheng 已提交
4126
				       u8 *uuid, u8 *fsid)
4127
{
Y
Yan Zheng 已提交
4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142
	struct btrfs_device *device;
	struct btrfs_fs_devices *cur_devices;

	cur_devices = root->fs_info->fs_devices;
	while (cur_devices) {
		if (!fsid ||
		    !memcmp(cur_devices->fsid, fsid, BTRFS_UUID_SIZE)) {
			device = __find_device(&cur_devices->devices,
					       devid, uuid);
			if (device)
				return device;
		}
		cur_devices = cur_devices->seed;
	}
	return NULL;
4143 4144
}

4145 4146 4147 4148 4149 4150 4151
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);
4152 4153
	if (!device)
		return NULL;
4154 4155 4156 4157
	list_add(&device->dev_list,
		 &fs_devices->devices);
	device->dev_root = root->fs_info->dev_root;
	device->devid = devid;
4158
	device->work.func = pending_bios_fn;
Y
Yan Zheng 已提交
4159
	device->fs_devices = fs_devices;
4160
	device->missing = 1;
4161
	fs_devices->num_devices++;
4162
	fs_devices->missing_devices++;
4163
	spin_lock_init(&device->io_lock);
4164
	INIT_LIST_HEAD(&device->dev_alloc_list);
4165 4166 4167 4168
	memcpy(device->uuid, dev_uuid, BTRFS_UUID_SIZE);
	return device;
}

4169 4170 4171 4172 4173 4174 4175 4176 4177 4178
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;
4179
	u8 uuid[BTRFS_UUID_SIZE];
4180
	int num_stripes;
4181
	int ret;
4182
	int i;
4183

4184 4185
	logical = key->offset;
	length = btrfs_chunk_length(leaf, chunk);
4186

4187
	read_lock(&map_tree->map_tree.lock);
4188
	em = lookup_extent_mapping(&map_tree->map_tree, logical, 1);
4189
	read_unlock(&map_tree->map_tree.lock);
4190 4191 4192 4193 4194 4195 4196 4197 4198

	/* 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);
	}

4199
	em = alloc_extent_map();
4200 4201
	if (!em)
		return -ENOMEM;
4202 4203
	num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
4204 4205 4206 4207 4208 4209 4210 4211 4212
	if (!map) {
		free_extent_map(em);
		return -ENOMEM;
	}

	em->bdev = (struct block_device *)map;
	em->start = logical;
	em->len = length;
	em->block_start = 0;
4213
	em->block_len = em->len;
4214

4215 4216 4217 4218 4219 4220
	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
4221
	map->sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
4222 4223 4224 4225
	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);
4226 4227 4228
		read_extent_buffer(leaf, uuid, (unsigned long)
				   btrfs_stripe_dev_uuid_nr(chunk, i),
				   BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
4229 4230
		map->stripes[i].dev = btrfs_find_device(root, devid, uuid,
							NULL);
4231
		if (!map->stripes[i].dev && !btrfs_test_opt(root, DEGRADED)) {
4232 4233 4234 4235
			kfree(map);
			free_extent_map(em);
			return -EIO;
		}
4236 4237 4238 4239 4240 4241 4242 4243 4244 4245
		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;
4246 4247
	}

4248
	write_lock(&map_tree->map_tree.lock);
4249
	ret = add_extent_mapping(&map_tree->map_tree, em);
4250
	write_unlock(&map_tree->map_tree.lock);
4251
	BUG_ON(ret);
4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263
	free_extent_map(em);

	return 0;
}

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

	device->devid = btrfs_device_id(leaf, dev_item);
4264 4265
	device->disk_total_bytes = btrfs_device_total_bytes(leaf, dev_item);
	device->total_bytes = device->disk_total_bytes;
4266 4267 4268 4269 4270 4271 4272
	device->bytes_used = btrfs_device_bytes_used(leaf, dev_item);
	device->type = btrfs_device_type(leaf, dev_item);
	device->io_align = btrfs_device_io_align(leaf, dev_item);
	device->io_width = btrfs_device_io_width(leaf, dev_item);
	device->sector_size = btrfs_device_sector_size(leaf, dev_item);

	ptr = (unsigned long)btrfs_device_uuid(dev_item);
4273
	read_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
4274 4275 4276 4277

	return 0;
}

Y
Yan Zheng 已提交
4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298
static int open_seed_devices(struct btrfs_root *root, u8 *fsid)
{
	struct btrfs_fs_devices *fs_devices;
	int ret;

	mutex_lock(&uuid_mutex);

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

	fs_devices = find_fsid(fsid);
	if (!fs_devices) {
		ret = -ENOENT;
		goto out;
	}
Y
Yan Zheng 已提交
4299 4300 4301 4302

	fs_devices = clone_fs_devices(fs_devices);
	if (IS_ERR(fs_devices)) {
		ret = PTR_ERR(fs_devices);
Y
Yan Zheng 已提交
4303 4304 4305
		goto out;
	}

4306
	ret = __btrfs_open_devices(fs_devices, FMODE_READ,
4307
				   root->fs_info->bdev_holder);
Y
Yan Zheng 已提交
4308 4309 4310 4311 4312
	if (ret)
		goto out;

	if (!fs_devices->seeding) {
		__btrfs_close_devices(fs_devices);
Y
Yan Zheng 已提交
4313
		free_fs_devices(fs_devices);
Y
Yan Zheng 已提交
4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324
		ret = -EINVAL;
		goto out;
	}

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

4325
static int read_one_dev(struct btrfs_root *root,
4326 4327 4328 4329 4330 4331
			struct extent_buffer *leaf,
			struct btrfs_dev_item *dev_item)
{
	struct btrfs_device *device;
	u64 devid;
	int ret;
Y
Yan Zheng 已提交
4332
	u8 fs_uuid[BTRFS_UUID_SIZE];
4333 4334
	u8 dev_uuid[BTRFS_UUID_SIZE];

4335
	devid = btrfs_device_id(leaf, dev_item);
4336 4337 4338
	read_extent_buffer(leaf, dev_uuid,
			   (unsigned long)btrfs_device_uuid(dev_item),
			   BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
4339 4340 4341 4342 4343 4344
	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
4345
		if (ret && !btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
4346 4347 4348 4349 4350
			return ret;
	}

	device = btrfs_find_device(root, devid, dev_uuid, fs_uuid);
	if (!device || !device->bdev) {
Y
Yan Zheng 已提交
4351
		if (!btrfs_test_opt(root, DEGRADED))
Y
Yan Zheng 已提交
4352 4353 4354
			return -EIO;

		if (!device) {
4355 4356
			printk(KERN_WARNING "warning devid %llu missing\n",
			       (unsigned long long)devid);
Y
Yan Zheng 已提交
4357 4358 4359
			device = add_missing_dev(root, devid, dev_uuid);
			if (!device)
				return -ENOMEM;
4360 4361 4362 4363 4364 4365 4366 4367 4368
		} 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
4369 4370 4371 4372 4373 4374 4375 4376
		}
	}

	if (device->fs_devices != root->fs_info->fs_devices) {
		BUG_ON(device->writeable);
		if (device->generation !=
		    btrfs_device_generation(leaf, dev_item))
			return -EINVAL;
4377
	}
4378 4379 4380

	fill_device_from_item(leaf, dev_item, device);
	device->dev_root = root->fs_info->dev_root;
4381
	device->in_fs_metadata = 1;
4382
	if (device->writeable) {
Y
Yan Zheng 已提交
4383
		device->fs_devices->total_rw_bytes += device->total_bytes;
4384 4385 4386 4387 4388
		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);
	}
4389 4390 4391 4392
	ret = 0;
	return ret;
}

Y
Yan Zheng 已提交
4393
int btrfs_read_sys_array(struct btrfs_root *root)
4394
{
4395
	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
4396
	struct extent_buffer *sb;
4397 4398
	struct btrfs_disk_key *disk_key;
	struct btrfs_chunk *chunk;
4399 4400 4401
	u8 *ptr;
	unsigned long sb_ptr;
	int ret = 0;
4402 4403 4404 4405
	u32 num_stripes;
	u32 array_size;
	u32 len = 0;
	u32 cur;
4406
	struct btrfs_key key;
4407

Y
Yan Zheng 已提交
4408
	sb = btrfs_find_create_tree_block(root, BTRFS_SUPER_INFO_OFFSET,
4409 4410 4411 4412
					  BTRFS_SUPER_INFO_SIZE);
	if (!sb)
		return -ENOMEM;
	btrfs_set_buffer_uptodate(sb);
4413
	btrfs_set_buffer_lockdep_class(root->root_key.objectid, sb, 0);
4414

4415
	write_extent_buffer(sb, super_copy, 0, BTRFS_SUPER_INFO_SIZE);
4416 4417 4418 4419 4420 4421 4422 4423 4424 4425
	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);

4426
		len = sizeof(*disk_key); ptr += len;
4427 4428 4429
		sb_ptr += len;
		cur += len;

4430
		if (key.type == BTRFS_CHUNK_ITEM_KEY) {
4431
			chunk = (struct btrfs_chunk *)sb_ptr;
4432
			ret = read_one_chunk(root, &key, sb, chunk);
4433 4434
			if (ret)
				break;
4435 4436 4437
			num_stripes = btrfs_chunk_num_stripes(sb, chunk);
			len = btrfs_chunk_item_size(num_stripes);
		} else {
4438 4439
			ret = -EIO;
			break;
4440 4441 4442 4443 4444
		}
		ptr += len;
		sb_ptr += len;
		cur += len;
	}
4445
	free_extent_buffer(sb);
4446
	return ret;
4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472
}

int btrfs_read_chunk_tree(struct btrfs_root *root)
{
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_key key;
	struct btrfs_key found_key;
	int ret;
	int slot;

	root = root->fs_info->chunk_root;

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

	/* first we search for all of the device items, and then we
	 * read in all of the chunk items.  This way we can create chunk
	 * mappings that reference all of the devices that are afound
	 */
	key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
	key.offset = 0;
	key.type = 0;
again:
	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
4473 4474
	if (ret < 0)
		goto error;
4475
	while (1) {
4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493
		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);
4494
				ret = read_one_dev(root, leaf, dev_item);
Y
Yan Zheng 已提交
4495 4496
				if (ret)
					goto error;
4497 4498 4499 4500 4501
			}
		} 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
4502 4503
			if (ret)
				goto error;
4504 4505 4506 4507 4508
		}
		path->slots[0]++;
	}
	if (key.objectid == BTRFS_DEV_ITEMS_OBJECTID) {
		key.objectid = 0;
4509
		btrfs_release_path(path);
4510 4511 4512 4513
		goto again;
	}
	ret = 0;
error:
Y
Yan Zheng 已提交
4514
	btrfs_free_path(path);
4515 4516
	return ret;
}
反馈
建议
客服 返回
顶部