[sfrench/cifs-2.6.git] / fs / btrfs / volumes.c

// SPDX-License-Identifier: GPL-2.0
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 */

#include <linux/sched.h>
#include <linux/bio.h>
#include <linux/slab.h>
#include <linux/buffer_head.h>
#include <linux/blkdev.h>
#include <linux/iocontext.h>
#include <linux/capability.h>
#include <linux/ratelimit.h>
#include <linux/kthread.h>
#include <linux/raid/pq.h>
#include <linux/semaphore.h>
#include <linux/uuid.h>
#include <linux/list_sort.h>
#include <asm/div64.h>
#include "ctree.h"
#include "extent_map.h"
#include "disk-io.h"
#include "transaction.h"
#include "print-tree.h"
#include "volumes.h"
#include "raid56.h"
#include "async-thread.h"
#include "check-integrity.h"
#include "rcu-string.h"
#include "math.h"
#include "dev-replace.h"
#include "sysfs.h"

const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = {
        [BTRFS_RAID_RAID10] = {
                .sub_stripes    = 2,
                .dev_stripes    = 1,
                .devs_max       = 0,    /* 0 == as many as possible */
                .devs_min       = 4,
                .tolerated_failures = 1,
                .devs_increment = 2,
                .ncopies        = 2,
                .raid_name      = "raid10",
                .bg_flag        = BTRFS_BLOCK_GROUP_RAID10,
                .mindev_error   = BTRFS_ERROR_DEV_RAID10_MIN_NOT_MET,
        },
        [BTRFS_RAID_RAID1] = {
                .sub_stripes    = 1,
                .dev_stripes    = 1,
                .devs_max       = 2,
                .devs_min       = 2,
                .tolerated_failures = 1,
                .devs_increment = 2,
                .ncopies        = 2,
                .raid_name      = "raid1",
                .bg_flag        = BTRFS_BLOCK_GROUP_RAID1,
                .mindev_error   = BTRFS_ERROR_DEV_RAID1_MIN_NOT_MET,
        },
        [BTRFS_RAID_DUP] = {
                .sub_stripes    = 1,
                .dev_stripes    = 2,
                .devs_max       = 1,
                .devs_min       = 1,
                .tolerated_failures = 0,
                .devs_increment = 1,
                .ncopies        = 2,
                .raid_name      = "dup",
                .bg_flag        = BTRFS_BLOCK_GROUP_DUP,
                .mindev_error   = 0,
        },
        [BTRFS_RAID_RAID0] = {
                .sub_stripes    = 1,
                .dev_stripes    = 1,
                .devs_max       = 0,
                .devs_min       = 2,
                .tolerated_failures = 0,
                .devs_increment = 1,
                .ncopies        = 1,
                .raid_name      = "raid0",
                .bg_flag        = BTRFS_BLOCK_GROUP_RAID0,
                .mindev_error   = 0,
        },
        [BTRFS_RAID_SINGLE] = {
                .sub_stripes    = 1,
                .dev_stripes    = 1,
                .devs_max       = 1,
                .devs_min       = 1,
                .tolerated_failures = 0,
                .devs_increment = 1,
                .ncopies        = 1,
                .raid_name      = "single",
                .bg_flag        = 0,
                .mindev_error   = 0,
        },
        [BTRFS_RAID_RAID5] = {
                .sub_stripes    = 1,
                .dev_stripes    = 1,
                .devs_max       = 0,
                .devs_min       = 2,
                .tolerated_failures = 1,
                .devs_increment = 1,
                .ncopies        = 2,
                .raid_name      = "raid5",
                .bg_flag        = BTRFS_BLOCK_GROUP_RAID5,
                .mindev_error   = BTRFS_ERROR_DEV_RAID5_MIN_NOT_MET,
        },
        [BTRFS_RAID_RAID6] = {
                .sub_stripes    = 1,
                .dev_stripes    = 1,
                .devs_max       = 0,
                .devs_min       = 3,
                .tolerated_failures = 2,
                .devs_increment = 1,
                .ncopies        = 3,
                .raid_name      = "raid6",
                .bg_flag        = BTRFS_BLOCK_GROUP_RAID6,
                .mindev_error   = BTRFS_ERROR_DEV_RAID6_MIN_NOT_MET,
        },
};

const char *get_raid_name(enum btrfs_raid_types type)
{
        if (type >= BTRFS_NR_RAID_TYPES)
                return NULL;

        return btrfs_raid_array[type].raid_name;
}

static int init_first_rw_device(struct btrfs_trans_handle *trans,
                                struct btrfs_fs_info *fs_info);
static int btrfs_relocate_sys_chunks(struct btrfs_fs_info *fs_info);
static void __btrfs_reset_dev_stats(struct btrfs_device *dev);
static void btrfs_dev_stat_print_on_error(struct btrfs_device *dev);
static void btrfs_dev_stat_print_on_load(struct btrfs_device *device);
static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
                             enum btrfs_map_op op,
                             u64 logical, u64 *length,
                             struct btrfs_bio **bbio_ret,
                             int mirror_num, int need_raid_map);

/*
 * Device locking
 * ==============
 *
 * There are several mutexes that protect manipulation of devices and low-level
 * structures like chunks but not block groups, extents or files
 *
 * uuid_mutex (global lock)
 * ------------------------
 * protects the fs_uuids list that tracks all per-fs fs_devices, resulting from
 * the SCAN_DEV ioctl registration or from mount either implicitly (the first
 * device) or requested by the device= mount option
 *
 * the mutex can be very coarse and can cover long-running operations
 *
 * protects: updates to fs_devices counters like missing devices, rw devices,
 * seeding, structure cloning, openning/closing devices at mount/umount time
 *
 * global::fs_devs - add, remove, updates to the global list
 *
 * does not protect: manipulation of the fs_devices::devices list!
 *
 * btrfs_device::name - renames (write side), read is RCU
 *
 * fs_devices::device_list_mutex (per-fs, with RCU)
 * ------------------------------------------------
 * protects updates to fs_devices::devices, ie. adding and deleting
 *
 * simple list traversal with read-only actions can be done with RCU protection
 *
 * may be used to exclude some operations from running concurrently without any
 * modifications to the list (see write_all_supers)
 *
 * balance_mutex
 * -------------
 * protects balance structures (status, state) and context accessed from
 * several places (internally, ioctl)
 *
 * chunk_mutex
 * -----------
 * protects chunks, adding or removing during allocation, trim or when a new
 * device is added/removed
 *
 * cleaner_mutex
 * -------------
 * a big lock that is held by the cleaner thread and prevents running subvolume
 * cleaning together with relocation or delayed iputs
 *
 *
 * Lock nesting
 * ============
 *
 * uuid_mutex
 *   volume_mutex
 *     device_list_mutex
 *       chunk_mutex
 *     balance_mutex
 *
 *
 * Exclusive operations, BTRFS_FS_EXCL_OP
 * ======================================
 *
 * Maintains the exclusivity of the following operations that apply to the
 * whole filesystem and cannot run in parallel.
 *
 * - Balance (*)
 * - Device add
 * - Device remove
 * - Device replace (*)
 * - Resize
 *
 * The device operations (as above) can be in one of the following states:
 *
 * - Running state
 * - Paused state
 * - Completed state
 *
 * Only device operations marked with (*) can go into the Paused state for the
 * following reasons:
 *
 * - ioctl (only Balance can be Paused through ioctl)
 * - filesystem remounted as read-only
 * - filesystem unmounted and mounted as read-only
 * - system power-cycle and filesystem mounted as read-only
 * - filesystem or device errors leading to forced read-only
 *
 * BTRFS_FS_EXCL_OP flag is set and cleared using atomic operations.
 * During the course of Paused state, the BTRFS_FS_EXCL_OP remains set.
 * A device operation in Paused or Running state can be canceled or resumed
 * either by ioctl (Balance only) or when remounted as read-write.
 * BTRFS_FS_EXCL_OP flag is cleared when the device operation is canceled or
 * completed.
 */

DEFINE_MUTEX(uuid_mutex);
static LIST_HEAD(fs_uuids);
struct list_head *btrfs_get_fs_uuids(void)
{
        return &fs_uuids;
}

/*
 * alloc_fs_devices - allocate struct btrfs_fs_devices
 * @fsid:       if not NULL, copy the uuid to fs_devices::fsid
 *
 * Return a pointer to a new struct btrfs_fs_devices on success, or ERR_PTR().
 * The returned struct is not linked onto any lists and can be destroyed with
 * kfree() right away.
 */
static struct btrfs_fs_devices *alloc_fs_devices(const u8 *fsid)
{
        struct btrfs_fs_devices *fs_devs;

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

        mutex_init(&fs_devs->device_list_mutex);

        INIT_LIST_HEAD(&fs_devs->devices);
        INIT_LIST_HEAD(&fs_devs->resized_devices);
        INIT_LIST_HEAD(&fs_devs->alloc_list);
        INIT_LIST_HEAD(&fs_devs->fs_list);
        if (fsid)
                memcpy(fs_devs->fsid, fsid, BTRFS_FSID_SIZE);

        return fs_devs;
}

void btrfs_free_device(struct btrfs_device *device)
{
        rcu_string_free(device->name);
        bio_put(device->flush_bio);
        kfree(device);
}

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);
                btrfs_free_device(device);
        }
        kfree(fs_devices);
}

static void btrfs_kobject_uevent(struct block_device *bdev,
                                 enum kobject_action action)
{
        int ret;

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

void __exit btrfs_cleanup_fs_uuids(void)
{
        struct btrfs_fs_devices *fs_devices;

        while (!list_empty(&fs_uuids)) {
                fs_devices = list_entry(fs_uuids.next,
                                        struct btrfs_fs_devices, fs_list);
                list_del(&fs_devices->fs_list);
                free_fs_devices(fs_devices);
        }
}

/*
 * Returns a pointer to a new btrfs_device on success; ERR_PTR() on error.
 * Returned struct is not linked onto any lists and must be destroyed using
 * btrfs_free_device.
 */
static struct btrfs_device *__alloc_device(void)
{
        struct btrfs_device *dev;

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

        /*
         * Preallocate a bio that's always going to be used for flushing device
         * barriers and matches the device lifespan
         */
        dev->flush_bio = bio_alloc_bioset(GFP_KERNEL, 0, NULL);
        if (!dev->flush_bio) {
                kfree(dev);
                return ERR_PTR(-ENOMEM);
        }

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

        spin_lock_init(&dev->io_lock);

        atomic_set(&dev->reada_in_flight, 0);
        atomic_set(&dev->dev_stats_ccnt, 0);
        btrfs_device_data_ordered_init(dev);
        INIT_RADIX_TREE(&dev->reada_zones, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
        INIT_RADIX_TREE(&dev->reada_extents, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);

        return dev;
}

/*
 * Find a device specified by @devid or @uuid in the list of @fs_devices, or
 * return NULL.
 *
 * If devid and uuid are both specified, the match must be exact, otherwise
 * only devid is used.
 */
static struct btrfs_device *find_device(struct btrfs_fs_devices *fs_devices,
                u64 devid, const u8 *uuid)
{
        struct btrfs_device *dev;

        list_for_each_entry(dev, &fs_devices->devices, dev_list) {
                if (dev->devid == devid &&
                    (!uuid || !memcmp(dev->uuid, uuid, BTRFS_UUID_SIZE))) {
                        return dev;
                }
        }
        return NULL;
}

static noinline struct btrfs_fs_devices *find_fsid(u8 *fsid)
{
        struct btrfs_fs_devices *fs_devices;

        list_for_each_entry(fs_devices, &fs_uuids, fs_list) {
                if (memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE) == 0)
                        return fs_devices;
        }
        return NULL;
}

static int
btrfs_get_bdev_and_sb(const char *device_path, fmode_t flags, void *holder,
                      int flush, struct block_device **bdev,
                      struct buffer_head **bh)
{
        int ret;

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

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

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

        return 0;

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

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;
}

/*
 * 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.
 */
static noinline void run_scheduled_bios(struct btrfs_device *device)
{
        struct btrfs_fs_info *fs_info = device->fs_info;
        struct bio *pending;
        struct backing_dev_info *bdi;
        struct btrfs_pending_bios *pending_bios;
        struct bio *tail;
        struct bio *cur;
        int again = 0;
        unsigned long num_run;
        unsigned long batch_run = 0;
        unsigned long last_waited = 0;
        int force_reg = 0;
        int sync_pending = 0;
        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);

        bdi = device->bdev->bd_bdi;

loop:
        spin_lock(&device->io_lock);

loop_lock:
        num_run = 0;

        /* 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
         */
        if (!force_reg && device->pending_sync_bios.head) {
                pending_bios = &device->pending_sync_bios;
                force_reg = 1;
        } else {
                pending_bios = &device->pending_bios;
                force_reg = 0;
        }

        pending = pending_bios->head;
        tail = pending_bios->tail;
        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.
         */
        if (device->pending_sync_bios.head == NULL &&
            device->pending_bios.head == NULL) {
                again = 0;
                device->running_pending = 0;
        } else {
                again = 1;
                device->running_pending = 1;
        }

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

        spin_unlock(&device->io_lock);

        while (pending) {

                rmb();
                /* 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)) {
                        spin_lock(&device->io_lock);
                        requeue_list(pending_bios, pending, tail);
                        goto loop_lock;
                }

                cur = pending;
                pending = pending->bi_next;
                cur->bi_next = NULL;

                BUG_ON(atomic_read(&cur->__bi_cnt) == 0);

                /*
                 * 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;
                }

                btrfsic_submit_bio(cur);
                num_run++;
                batch_run++;

                cond_resched();

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

                        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;
                                cond_resched();
                                continue;
                        }
                        spin_lock(&device->io_lock);
                        requeue_list(pending_bios, pending, tail);
                        device->running_pending = 1;

                        spin_unlock(&device->io_lock);
                        btrfs_queue_work(fs_info->submit_workers,
                                         &device->work);
                        goto done;
                }
        }

        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);

done:
        blk_finish_plug(&plug);
}

static void pending_bios_fn(struct btrfs_work *work)
{
        struct btrfs_device *device;

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

/*
 *  Search and remove all stale (devices which are not mounted) devices.
 *  When both inputs are NULL, it will search and release all stale devices.
 *  path:       Optional. When provided will it release all unmounted devices
 *              matching this path only.
 *  skip_dev:   Optional. Will skip this device when searching for the stale
 *              devices.
 */
static void btrfs_free_stale_devices(const char *path,
                                     struct btrfs_device *skip_dev)
{
        struct btrfs_fs_devices *fs_devs, *tmp_fs_devs;
        struct btrfs_device *dev, *tmp_dev;

        list_for_each_entry_safe(fs_devs, tmp_fs_devs, &fs_uuids, fs_list) {

                if (fs_devs->opened)
                        continue;

                list_for_each_entry_safe(dev, tmp_dev,
                                         &fs_devs->devices, dev_list) {
                        int not_found = 0;

                        if (skip_dev && skip_dev == dev)
                                continue;
                        if (path && !dev->name)
                                continue;

                        rcu_read_lock();
                        if (path)
                                not_found = strcmp(rcu_str_deref(dev->name),
                                                   path);
                        rcu_read_unlock();
                        if (not_found)
                                continue;

                        /* delete the stale device */
                        if (fs_devs->num_devices == 1) {
                                btrfs_sysfs_remove_fsid(fs_devs);
                                list_del(&fs_devs->fs_list);
                                free_fs_devices(fs_devs);
                                break;
                        } else {
                                fs_devs->num_devices--;
                                list_del(&dev->dev_list);
                                btrfs_free_device(dev);
                        }
                }
        }
}

static int btrfs_open_one_device(struct btrfs_fs_devices *fs_devices,
                        struct btrfs_device *device, fmode_t flags,
                        void *holder)
{
        struct request_queue *q;
        struct block_device *bdev;
        struct buffer_head *bh;
        struct btrfs_super_block *disk_super;
        u64 devid;
        int ret;

        if (device->bdev)
                return -EINVAL;
        if (!device->name)
                return -EINVAL;

        ret = btrfs_get_bdev_and_sb(device->name->str, flags, holder, 1,
                                    &bdev, &bh);
        if (ret)
                return ret;

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

        if (memcmp(device->uuid, disk_super->dev_item.uuid, BTRFS_UUID_SIZE))
                goto error_brelse;

        device->generation = btrfs_super_generation(disk_super);

        if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_SEEDING) {
                clear_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state);
                fs_devices->seeding = 1;
        } else {
                if (bdev_read_only(bdev))
                        clear_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state);
                else
                        set_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state);
        }

        q = bdev_get_queue(bdev);
        if (!blk_queue_nonrot(q))
                fs_devices->rotating = 1;

        device->bdev = bdev;
        clear_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state);
        device->mode = flags;

        fs_devices->open_devices++;
        if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state) &&
            device->devid != BTRFS_DEV_REPLACE_DEVID) {
                fs_devices->rw_devices++;
                list_add_tail(&device->dev_alloc_list, &fs_devices->alloc_list);
        }
        brelse(bh);

        return 0;

error_brelse:
        brelse(bh);
        blkdev_put(bdev, flags);

        return -EINVAL;
}

/*
 * Add new device to list of registered devices
 *
 * Returns:
 * device pointer which was just added or updated when successful
 * error pointer when failed
 */
static noinline struct btrfs_device *device_list_add(const char *path,
                           struct btrfs_super_block *disk_super)
{
        struct btrfs_device *device;
        struct btrfs_fs_devices *fs_devices;
        struct rcu_string *name;
        u64 found_transid = btrfs_super_generation(disk_super);
        u64 devid = btrfs_stack_device_id(&disk_super->dev_item);

        fs_devices = find_fsid(disk_super->fsid);
        if (!fs_devices) {
                fs_devices = alloc_fs_devices(disk_super->fsid);
                if (IS_ERR(fs_devices))
                        return ERR_CAST(fs_devices);

                list_add(&fs_devices->fs_list, &fs_uuids);

                device = NULL;
        } else {
                device = find_device(fs_devices, devid,
                                disk_super->dev_item.uuid);
        }

        if (!device) {
                if (fs_devices->opened)
                        return ERR_PTR(-EBUSY);

                device = btrfs_alloc_device(NULL, &devid,
                                            disk_super->dev_item.uuid);
                if (IS_ERR(device)) {
                        /* we can safely leave the fs_devices entry around */
                        return device;
                }

                name = rcu_string_strdup(path, GFP_NOFS);
                if (!name) {
                        btrfs_free_device(device);
                        return ERR_PTR(-ENOMEM);
                }
                rcu_assign_pointer(device->name, name);

                mutex_lock(&fs_devices->device_list_mutex);
                list_add_rcu(&device->dev_list, &fs_devices->devices);
                fs_devices->num_devices++;
                mutex_unlock(&fs_devices->device_list_mutex);

                device->fs_devices = fs_devices;
                btrfs_free_stale_devices(path, device);

                if (disk_super->label[0])
                        pr_info("BTRFS: device label %s devid %llu transid %llu %s\n",
                                disk_super->label, devid, found_transid, path);
                else
                        pr_info("BTRFS: device fsid %pU devid %llu transid %llu %s\n",
                                disk_super->fsid, devid, found_transid, path);

        } else if (!device->name || strcmp(device->name->str, path)) {
                /*
                 * When FS is already mounted.
                 * 1. If you are here and if the device->name is NULL that
                 *    means this device was missing at time of FS mount.
                 * 2. If you are here and if the device->name is different
                 *    from 'path' that means either
                 *      a. The same device disappeared and reappeared with
                 *         different name. or
                 *      b. The missing-disk-which-was-replaced, has
                 *         reappeared now.
                 *
                 * We must allow 1 and 2a above. But 2b would be a spurious
                 * and unintentional.
                 *
                 * Further in case of 1 and 2a above, the disk at 'path'
                 * would have missed some transaction when it was away and
                 * in case of 2a the stale bdev has to be updated as well.
                 * 2b must not be allowed at all time.
                 */

                /*
                 * For now, we do allow update to btrfs_fs_device through the
                 * btrfs dev scan cli after FS has been mounted.  We're still
                 * tracking a problem where systems fail mount by subvolume id
                 * when we reject replacement on a mounted FS.
                 */
                if (!fs_devices->opened && found_transid < device->generation) {
                        /*
                         * That is if the FS is _not_ mounted and if you
                         * are here, that means there is more than one
                         * disk with same uuid and devid.We keep the one
                         * with larger generation number or the last-in if
                         * generation are equal.
                         */
                        return ERR_PTR(-EEXIST);
                }

                name = rcu_string_strdup(path, GFP_NOFS);
                if (!name)
                        return ERR_PTR(-ENOMEM);
                rcu_string_free(device->name);
                rcu_assign_pointer(device->name, name);
                if (test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state)) {
                        fs_devices->missing_devices--;
                        clear_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state);
                }
        }

        /*
         * Unmount does not free the btrfs_device struct but would zero
         * generation along with most of the other members. So just update
         * it back. We need it to pick the disk with largest generation
         * (as above).
         */
        if (!fs_devices->opened)
                device->generation = found_transid;

        fs_devices->total_devices = btrfs_super_num_devices(disk_super);

        return device;
}

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 = alloc_fs_devices(orig->fsid);
        if (IS_ERR(fs_devices))
                return fs_devices;

        mutex_lock(&orig->device_list_mutex);
        fs_devices->total_devices = orig->total_devices;

        /* We have held the volume lock, it is safe to get the devices. */
        list_for_each_entry(orig_dev, &orig->devices, dev_list) {
                struct rcu_string *name;

                device = btrfs_alloc_device(NULL, &orig_dev->devid,
                                            orig_dev->uuid);
                if (IS_ERR(device))
                        goto error;

                /*
                 * This is ok to do without rcu read locked because we hold the
                 * uuid mutex so nothing we touch in here is going to disappear.
                 */
                if (orig_dev->name) {
                        name = rcu_string_strdup(orig_dev->name->str,
                                        GFP_KERNEL);
                        if (!name) {
                                btrfs_free_device(device);
                                goto error;
                        }
                        rcu_assign_pointer(device->name, name);
                }

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

/*
 * After we have read the system tree and know devids belonging to
 * this filesystem, remove the device which does not belong there.
 */
void btrfs_free_extra_devids(struct btrfs_fs_devices *fs_devices, int step)
{
        struct btrfs_device *device, *next;
        struct btrfs_device *latest_dev = NULL;

        mutex_lock(&uuid_mutex);
again:
        /* This is the initialized path, it is safe to release the devices. */
        list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
                if (test_bit(BTRFS_DEV_STATE_IN_FS_METADATA,
                                                        &device->dev_state)) {
                        if (!test_bit(BTRFS_DEV_STATE_REPLACE_TGT,
                             &device->dev_state) &&
                             (!latest_dev ||
                              device->generation > latest_dev->generation)) {
                                latest_dev = device;
                        }
                        continue;
                }

                if (device->devid == BTRFS_DEV_REPLACE_DEVID) {
                        /*
                         * In the first step, keep the device which has
                         * the correct fsid and the devid that is used
                         * for the dev_replace procedure.
                         * In the second step, the dev_replace state is
                         * read from the device tree and it is known
                         * whether the procedure is really active or
                         * not, which means whether this device is
                         * used or whether it should be removed.
                         */
                        if (step == 0 || test_bit(BTRFS_DEV_STATE_REPLACE_TGT,
                                                  &device->dev_state)) {
                                continue;
                        }
                }
                if (device->bdev) {
                        blkdev_put(device->bdev, device->mode);
                        device->bdev = NULL;
                        fs_devices->open_devices--;
                }
                if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) {
                        list_del_init(&device->dev_alloc_list);
                        clear_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state);
                        if (!test_bit(BTRFS_DEV_STATE_REPLACE_TGT,
                                      &device->dev_state))
                                fs_devices->rw_devices--;
                }
                list_del_init(&device->dev_list);
                fs_devices->num_devices--;
                btrfs_free_device(device);
        }

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

        fs_devices->latest_bdev = latest_dev->bdev;

        mutex_unlock(&uuid_mutex);
}

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

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

static void btrfs_close_bdev(struct btrfs_device *device)
{
        if (!device->bdev)
                return;

        if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) {
                sync_blockdev(device->bdev);
                invalidate_bdev(device->bdev);
        }

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

static void btrfs_prepare_close_one_device(struct btrfs_device *device)
{
        struct btrfs_fs_devices *fs_devices = device->fs_devices;
        struct btrfs_device *new_device;
        struct rcu_string *name;

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

        if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state) &&
            device->devid != BTRFS_DEV_REPLACE_DEVID) {
                list_del_init(&device->dev_alloc_list);
                fs_devices->rw_devices--;
        }

        if (test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state))
                fs_devices->missing_devices--;

        new_device = btrfs_alloc_device(NULL, &device->devid,
                                        device->uuid);
        BUG_ON(IS_ERR(new_device)); /* -ENOMEM */

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

        list_replace_rcu(&device->dev_list, &new_device->dev_list);
        new_device->fs_devices = device->fs_devices;
}

static int close_fs_devices(struct btrfs_fs_devices *fs_devices)
{
        struct btrfs_device *device, *tmp;
        struct list_head pending_put;

        INIT_LIST_HEAD(&pending_put);

        if (--fs_devices->opened > 0)
                return 0;

        mutex_lock(&fs_devices->device_list_mutex);
        list_for_each_entry_safe(device, tmp, &fs_devices->devices, dev_list) {
                btrfs_prepare_close_one_device(device);
                list_add(&device->dev_list, &pending_put);
        }
        mutex_unlock(&fs_devices->device_list_mutex);

        /*
         * btrfs_show_devname() is using the device_list_mutex,
         * sometimes call to blkdev_put() leads vfs calling
         * into this func. So do put outside of device_list_mutex,
         * as of now.
         */
        while (!list_empty(&pending_put)) {
                device = list_first_entry(&pending_put,
                                struct btrfs_device, dev_list);
                list_del(&device->dev_list);
                btrfs_close_bdev(device);
                call_rcu(&device->rcu, free_device_rcu);
        }

        WARN_ON(fs_devices->open_devices);
        WARN_ON(fs_devices->rw_devices);
        fs_devices->opened = 0;
        fs_devices->seeding = 0;

        return 0;
}

int btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
{
        struct btrfs_fs_devices *seed_devices = NULL;
        int ret;

        mutex_lock(&uuid_mutex);
        ret = close_fs_devices(fs_devices);
        if (!fs_devices->opened) {
                seed_devices = fs_devices->seed;
                fs_devices->seed = NULL;
        }
        mutex_unlock(&uuid_mutex);

        while (seed_devices) {
                fs_devices = seed_devices;
                seed_devices = fs_devices->seed;
                close_fs_devices(fs_devices);
                free_fs_devices(fs_devices);
        }
        return ret;
}

static int open_fs_devices(struct btrfs_fs_devices *fs_devices,
                                fmode_t flags, void *holder)
{
        struct btrfs_device *device;
        struct btrfs_device *latest_dev = NULL;
        int ret = 0;

        flags |= FMODE_EXCL;

        list_for_each_entry(device, &fs_devices->devices, dev_list) {
                /* Just open everything we can; ignore failures here */
                if (btrfs_open_one_device(fs_devices, device, flags, holder))
                        continue;

                if (!latest_dev ||
                    device->generation > latest_dev->generation)
                        latest_dev = device;
        }
        if (fs_devices->open_devices == 0) {
                ret = -EINVAL;
                goto out;
        }
        fs_devices->opened = 1;
        fs_devices->latest_bdev = latest_dev->bdev;
        fs_devices->total_rw_bytes = 0;
out:
        return ret;
}

static int devid_cmp(void *priv, struct list_head *a, struct list_head *b)
{
        struct btrfs_device *dev1, *dev2;

        dev1 = list_entry(a, struct btrfs_device, dev_list);
        dev2 = list_entry(b, struct btrfs_device, dev_list);

        if (dev1->devid < dev2->devid)
                return -1;
        else if (dev1->devid > dev2->devid)
                return 1;
        return 0;
}

int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
                       fmode_t flags, void *holder)
{
        int ret;

        mutex_lock(&uuid_mutex);
        mutex_lock(&fs_devices->device_list_mutex);
        if (fs_devices->opened) {
                fs_devices->opened++;
                ret = 0;
        } else {
                list_sort(NULL, &fs_devices->devices, devid_cmp);
                ret = open_fs_devices(fs_devices, flags, holder);
        }
        mutex_unlock(&fs_devices->device_list_mutex);
        mutex_unlock(&uuid_mutex);

        return ret;
}

static void btrfs_release_disk_super(struct page *page)
{
        kunmap(page);
        put_page(page);
}

static int btrfs_read_disk_super(struct block_device *bdev, u64 bytenr,
                                 struct page **page,
                                 struct btrfs_super_block **disk_super)
{
        void *p;
        pgoff_t index;

        /* make sure our super fits in the device */
        if (bytenr + PAGE_SIZE >= i_size_read(bdev->bd_inode))
                return 1;

        /* make sure our super fits in the page */
        if (sizeof(**disk_super) > PAGE_SIZE)
                return 1;

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

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

        if (IS_ERR_OR_NULL(*page))
                return 1;

        p = kmap(*page);

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

        if (btrfs_super_bytenr(*disk_super) != bytenr ||
            btrfs_super_magic(*disk_super) != BTRFS_MAGIC) {
                btrfs_release_disk_super(*page);
                return 1;
        }

        if ((*disk_super)->label[0] &&
                (*disk_super)->label[BTRFS_LABEL_SIZE - 1])
                (*disk_super)->label[BTRFS_LABEL_SIZE - 1] = '\0';

        return 0;
}

/*
 * Look for a btrfs signature on a device. This may be called out of the mount path
 * and we are not allowed to call set_blocksize during the scan. The superblock
 * is read via pagecache
 */
int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
                          struct btrfs_fs_devices **fs_devices_ret)
{
        struct btrfs_super_block *disk_super;
        struct btrfs_device *device;
        struct block_device *bdev;
        struct page *page;
        int ret = 0;
        u64 bytenr;

        /*
         * we would like to check all the supers, but that would make
         * a btrfs mount succeed after a mkfs from a different FS.
         * So, we need to add a special mount option to scan for
         * later supers, using BTRFS_SUPER_MIRROR_MAX instead
         */
        bytenr = btrfs_sb_offset(0);
        flags |= FMODE_EXCL;

        bdev = blkdev_get_by_path(path, flags, holder);
        if (IS_ERR(bdev))
                return PTR_ERR(bdev);

        if (btrfs_read_disk_super(bdev, bytenr, &page, &disk_super)) {
                ret = -EINVAL;
                goto error_bdev_put;
        }

        mutex_lock(&uuid_mutex);
        device = device_list_add(path, disk_super);
        if (IS_ERR(device))
                ret = PTR_ERR(device);
        else
                *fs_devices_ret = device->fs_devices;
        mutex_unlock(&uuid_mutex);

        btrfs_release_disk_super(page);

error_bdev_put:
        blkdev_put(bdev, flags);

        return ret;
}

/* 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->fs_info->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 ||
                test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state))
                return 0;

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

        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 (key.type != 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;
}

static int contains_pending_extent(struct btrfs_transaction *transaction,
                                   struct btrfs_device *device,
                                   u64 *start, u64 len)
{
        struct btrfs_fs_info *fs_info = device->fs_info;
        struct extent_map *em;
        struct list_head *search_list = &fs_info->pinned_chunks;
        int ret = 0;
        u64 physical_start = *start;

        if (transaction)
                search_list = &transaction->pending_chunks;
again:
        list_for_each_entry(em, search_list, list) {
                struct map_lookup *map;
                int i;

                map = em->map_lookup;
                for (i = 0; i < map->num_stripes; i++) {
                        u64 end;

                        if (map->stripes[i].dev != device)
                                continue;
                        if (map->stripes[i].physical >= physical_start + len ||
                            map->stripes[i].physical + em->orig_block_len <=
                            physical_start)
                                continue;
                        /*
                         * Make sure that while processing the pinned list we do
                         * not override our *start with a lower value, because
                         * we can have pinned chunks that fall within this
                         * device hole and that have lower physical addresses
                         * than the pending chunks we processed before. If we
                         * do not take this special care we can end up getting
                         * 2 pending chunks that start at the same physical
                         * device offsets because the end offset of a pinned
                         * chunk can be equal to the start offset of some
                         * pending chunk.
                         */
                        end = map->stripes[i].physical + em->orig_block_len;
                        if (end > *start) {
                                *start = end;
                                ret = 1;
                        }
                }
        }
        if (search_list != &fs_info->pinned_chunks) {
                search_list = &fs_info->pinned_chunks;
                goto again;
        }

        return ret;
}


/*
 * find_free_dev_extent_start - find free space in the specified device
 * @device:       the device which we search the free space in
 * @num_bytes:    the size of the free space that we need
 * @search_start: the position from which to begin the search
 * @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
 *
 * 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
 *
 * @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.
 */
int find_free_dev_extent_start(struct btrfs_transaction *transaction,
                               struct btrfs_device *device, u64 num_bytes,
                               u64 search_start, u64 *start, u64 *len)
{
        struct btrfs_fs_info *fs_info = device->fs_info;
        struct btrfs_root *root = fs_info->dev_root;
        struct btrfs_key key;
        struct btrfs_dev_extent *dev_extent;
        struct btrfs_path *path;
        u64 hole_size;
        u64 max_hole_start;
        u64 max_hole_size;
        u64 extent_end;
        u64 search_end = device->total_bytes;
        int ret;
        int slot;
        struct extent_buffer *l;

        /*
         * We don't want to overwrite the superblock on the drive nor any area
         * used by the boot loader (grub for example), so we make sure to start
         * at an offset of at least 1MB.
         */
        search_start = max_t(u64, search_start, SZ_1M);

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

        max_hole_start = search_start;
        max_hole_size = 0;

again:
        if (search_start >= search_end ||
                test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) {
                ret = -ENOSPC;
                goto out;
        }

        path->reada = READA_FORWARD;
        path->search_commit_root = 1;
        path->skip_locking = 1;

        key.objectid = device->devid;
        key.offset = search_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 (key.type != BTRFS_DEV_EXTENT_KEY)
                        goto next;

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

                        /*
                         * Have to check before we set max_hole_start, otherwise
                         * we could end up sending back this offset anyway.
                         */
                        if (contains_pending_extent(transaction, device,
                                                    &search_start,
                                                    hole_size)) {
                                if (key.offset >= search_start) {
                                        hole_size = key.offset - search_start;
                                } else {
                                        WARN_ON_ONCE(1);
                                        hole_size = 0;
                                }
                        }

                        if (hole_size > max_hole_size) {
                                max_hole_start = search_start;
                                max_hole_size = hole_size;
                        }

                        /*
                         * If this free space is greater than which we need,
                         * it must be the max free space that we have found
                         * until now, so max_hole_start must point to the start
                         * of this free space and the length of this free space
                         * is stored in max_hole_size. Thus, we return
                         * max_hole_start and max_hole_size and go back to the
                         * caller.
                         */
                        if (hole_size >= num_bytes) {
                                ret = 0;
                                goto out;
                        }
                }

                dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
                extent_end = key.offset + btrfs_dev_extent_length(l,
                                                                  dev_extent);
                if (extent_end > search_start)
                        search_start = extent_end;
next:
                path->slots[0]++;
                cond_resched();
        }

        /*
         * 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;

                if (contains_pending_extent(transaction, device, &search_start,
                                            hole_size)) {
                        btrfs_release_path(path);
                        goto again;
                }

                if (hole_size > max_hole_size) {
                        max_hole_start = search_start;
                        max_hole_size = hole_size;
                }
        }

        /* See above. */
        if (max_hole_size < num_bytes)
                ret = -ENOSPC;
        else
                ret = 0;

out:
        btrfs_free_path(path);
        *start = max_hole_start;
        if (len)
                *len = max_hole_size;
        return ret;
}

int find_free_dev_extent(struct btrfs_trans_handle *trans,
                         struct btrfs_device *device, u64 num_bytes,
                         u64 *start, u64 *len)
{
        /* FIXME use last free of some kind */
        return find_free_dev_extent_start(trans->transaction, device,
                                          num_bytes, 0, start, len);
}

static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans,
                          struct btrfs_device *device,
                          u64 start, u64 *dev_extent_len)
{
        struct btrfs_fs_info *fs_info = device->fs_info;
        struct btrfs_root *root = fs_info->dev_root;
        int ret;
        struct btrfs_path *path;
        struct btrfs_key key;
        struct btrfs_key found_key;
        struct extent_buffer *leaf = NULL;
        struct btrfs_dev_extent *extent = NULL;

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

        key.objectid = device->devid;
        key.offset = start;
        key.type = BTRFS_DEV_EXTENT_KEY;
again:
        ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
        if (ret > 0) {
                ret = btrfs_previous_item(root, path, key.objectid,
                                          BTRFS_DEV_EXTENT_KEY);
                if (ret)
                        goto out;
                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);
                key = found_key;
                btrfs_release_path(path);
                goto again;
        } else if (ret == 0) {
                leaf = path->nodes[0];
                extent = btrfs_item_ptr(leaf, path->slots[0],
                                        struct btrfs_dev_extent);
        } else {
                btrfs_handle_fs_error(fs_info, ret, "Slot search failed");
                goto out;
        }

        *dev_extent_len = btrfs_dev_extent_length(leaf, extent);

        ret = btrfs_del_item(trans, root, path);
        if (ret) {
                btrfs_handle_fs_error(fs_info, ret,
                                      "Failed to remove dev extent item");
        } else {
                set_bit(BTRFS_TRANS_HAVE_FREE_BGS, &trans->transaction->flags);
        }
out:
        btrfs_free_path(path);
        return ret;
}

static int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
                                  struct btrfs_device *device,
                                  u64 chunk_offset, u64 start, u64 num_bytes)
{
        int ret;
        struct btrfs_path *path;
        struct btrfs_fs_info *fs_info = device->fs_info;
        struct btrfs_root *root = fs_info->dev_root;
        struct btrfs_dev_extent *extent;
        struct extent_buffer *leaf;
        struct btrfs_key key;

        WARN_ON(!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state));
        WARN_ON(test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state));
        path = btrfs_alloc_path();
        if (!path)
                return -ENOMEM;

        key.objectid = device->devid;
        key.offset = start;
        key.type = BTRFS_DEV_EXTENT_KEY;
        ret = btrfs_insert_empty_item(trans, root, path, &key,
                                      sizeof(*extent));
        if (ret)
                goto out;

        leaf = path->nodes[0];
        extent = btrfs_item_ptr(leaf, path->slots[0],
                                struct btrfs_dev_extent);
        btrfs_set_dev_extent_chunk_tree(leaf, extent,
                                        BTRFS_CHUNK_TREE_OBJECTID);
        btrfs_set_dev_extent_chunk_objectid(leaf, extent,
                                            BTRFS_FIRST_CHUNK_TREE_OBJECTID);
        btrfs_set_dev_extent_chunk_offset(leaf, extent, chunk_offset);

        btrfs_set_dev_extent_length(leaf, extent, num_bytes);
        btrfs_mark_buffer_dirty(leaf);
out:
        btrfs_free_path(path);
        return ret;
}

static u64 find_next_chunk(struct btrfs_fs_info *fs_info)
{
        struct extent_map_tree *em_tree;
        struct extent_map *em;
        struct rb_node *n;
        u64 ret = 0;

        em_tree = &fs_info->mapping_tree.map_tree;
        read_lock(&em_tree->lock);
        n = rb_last(&em_tree->map);
        if (n) {
                em = rb_entry(n, struct extent_map, rb_node);
                ret = em->start + em->len;
        }
        read_unlock(&em_tree->lock);

        return ret;
}

static noinline int find_next_devid(struct btrfs_fs_info *fs_info,
                                    u64 *devid_ret)
{
        int ret;
        struct btrfs_key key;
        struct btrfs_key found_key;
        struct btrfs_path *path;

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

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

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

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

        ret = btrfs_previous_item(fs_info->chunk_root, path,
                                  BTRFS_DEV_ITEMS_OBJECTID,
                                  BTRFS_DEV_ITEM_KEY);
        if (ret) {
                *devid_ret = 1;
        } else {
                btrfs_item_key_to_cpu(path->nodes[0], &found_key,
                                      path->slots[0]);
                *devid_ret = found_key.offset + 1;
        }
        ret = 0;
error:
        btrfs_free_path(path);
        return ret;
}

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

        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_insert_empty_item(trans, root, path, &key,
                                      sizeof(*dev_item));
        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);
        btrfs_set_device_generation(leaf, dev_item, 0);
        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,
                                     btrfs_device_get_disk_total_bytes(device));
        btrfs_set_device_bytes_used(leaf, dev_item,
                                    btrfs_device_get_bytes_used(device));
        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);
        btrfs_set_device_start_offset(leaf, dev_item, 0);

        ptr = btrfs_device_uuid(dev_item);
        write_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
        ptr = btrfs_device_fsid(dev_item);
        write_extent_buffer(leaf, fs_info->fsid, ptr, BTRFS_FSID_SIZE);
        btrfs_mark_buffer_dirty(leaf);

        ret = 0;
out:
        btrfs_free_path(path);
        return ret;
}

/*
 * Function to update ctime/mtime for a given device path.
 * Mainly used for ctime/mtime based probe like libblkid.
 */
static void update_dev_time(const char *path_name)
{
        struct file *filp;

        filp = filp_open(path_name, O_RDWR, 0);
        if (IS_ERR(filp))
                return;
        file_update_time(filp);
        filp_close(filp, NULL);
}

static int btrfs_rm_dev_item(struct btrfs_fs_info *fs_info,
                             struct btrfs_device *device)
{
        struct btrfs_root *root = fs_info->chunk_root;
        int ret;
        struct btrfs_path *path;
        struct btrfs_key key;
        struct btrfs_trans_handle *trans;

        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_DEV_ITEMS_OBJECTID;
        key.type = BTRFS_DEV_ITEM_KEY;
        key.offset = device->devid;

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

        ret = btrfs_del_item(trans, root, path);
        if (ret) {
                btrfs_abort_transaction(trans, ret);
                btrfs_end_transaction(trans);
        }

out:
        btrfs_free_path(path);
        if (!ret)
                ret = btrfs_commit_transaction(trans);
        return ret;
}

/*
 * Verify that @num_devices satisfies the RAID profile constraints in the whole
 * filesystem. It's up to the caller to adjust that number regarding eg. device
 * replace.
 */
static int btrfs_check_raid_min_devices(struct btrfs_fs_info *fs_info,
                u64 num_devices)
{
        u64 all_avail;
        unsigned seq;
        int i;

        do {
                seq = read_seqbegin(&fs_info->profiles_lock);

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

        for (i = 0; i < BTRFS_NR_RAID_TYPES; i++) {
                if (!(all_avail & btrfs_raid_array[i].bg_flag))
                        continue;

                if (num_devices < btrfs_raid_array[i].devs_min) {
                        int ret = btrfs_raid_array[i].mindev_error;

                        if (ret)
                                return ret;
                }
        }

        return 0;
}

static struct btrfs_device * btrfs_find_next_active_device(
                struct btrfs_fs_devices *fs_devs, struct btrfs_device *device)
{
        struct btrfs_device *next_device;

        list_for_each_entry(next_device, &fs_devs->devices, dev_list) {
                if (next_device != device &&
                    !test_bit(BTRFS_DEV_STATE_MISSING, &next_device->dev_state)
                    && next_device->bdev)
                        return next_device;
        }

        return NULL;
}

/*
 * Helper function to check if the given device is part of s_bdev / latest_bdev
 * and replace it with the provided or the next active device, in the context
 * where this function called, there should be always be another device (or
 * this_dev) which is active.
 */
void btrfs_assign_next_active_device(struct btrfs_fs_info *fs_info,
                struct btrfs_device *device, struct btrfs_device *this_dev)
{
        struct btrfs_device *next_device;

        if (this_dev)
                next_device = this_dev;
        else
                next_device = btrfs_find_next_active_device(fs_info->fs_devices,
                                                                device);
        ASSERT(next_device);

        if (fs_info->sb->s_bdev &&
                        (fs_info->sb->s_bdev == device->bdev))
                fs_info->sb->s_bdev = next_device->bdev;

        if (fs_info->fs_devices->latest_bdev == device->bdev)
                fs_info->fs_devices->latest_bdev = next_device->bdev;
}

int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path,
                u64 devid)
{
        struct btrfs_device *device;
        struct btrfs_fs_devices *cur_devices;
        struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
        u64 num_devices;
        int ret = 0;

        mutex_lock(&uuid_mutex);

        num_devices = fs_devices->num_devices;
        btrfs_dev_replace_read_lock(&fs_info->dev_replace);
        if (btrfs_dev_replace_is_ongoing(&fs_info->dev_replace)) {
                WARN_ON(num_devices < 1);
                num_devices--;
        }
        btrfs_dev_replace_read_unlock(&fs_info->dev_replace);

        ret = btrfs_check_raid_min_devices(fs_info, num_devices - 1);
        if (ret)
                goto out;

        ret = btrfs_find_device_by_devspec(fs_info, devid, device_path,
                                           &device);
        if (ret)
                goto out;

        if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) {
                ret = BTRFS_ERROR_DEV_TGT_REPLACE;
                goto out;
        }

        if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state) &&
            fs_info->fs_devices->rw_devices == 1) {
                ret = BTRFS_ERROR_DEV_ONLY_WRITABLE;
                goto out;
        }

        if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) {
                mutex_lock(&fs_info->chunk_mutex);
                list_del_init(&device->dev_alloc_list);
                device->fs_devices->rw_devices--;
                mutex_unlock(&fs_info->chunk_mutex);
        }

        mutex_unlock(&uuid_mutex);
        ret = btrfs_shrink_device(device, 0);
        mutex_lock(&uuid_mutex);
        if (ret)
                goto error_undo;

        /*
         * TODO: the superblock still includes this device in its num_devices
         * counter although write_all_supers() is not locked out. This
         * could give a filesystem state which requires a degraded mount.
         */
        ret = btrfs_rm_dev_item(fs_info, device);
        if (ret)
                goto error_undo;

        clear_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state);
        btrfs_scrub_cancel_dev(fs_info, device);

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

        /*
         * In normal cases the cur_devices == fs_devices. But in case
         * of deleting a seed device, the cur_devices should point to
         * its own fs_devices listed under the fs_devices->seed.
         */
        cur_devices = device->fs_devices;
        mutex_lock(&fs_devices->device_list_mutex);
        list_del_rcu(&device->dev_list);

        cur_devices->num_devices--;
        cur_devices->total_devices--;

        if (test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state))
                cur_devices->missing_devices--;

        btrfs_assign_next_active_device(fs_info, device, NULL);

        if (device->bdev) {
                cur_devices->open_devices--;
                /* remove sysfs entry */
                btrfs_sysfs_rm_device_link(fs_devices, device);
        }

        num_devices = btrfs_super_num_devices(fs_info->super_copy) - 1;
        btrfs_set_super_num_devices(fs_info->super_copy, num_devices);
        mutex_unlock(&fs_devices->device_list_mutex);

        /*
         * at this point, the device is zero sized and detached from
         * the devices list.  All that's left is to zero out the old
         * supers and free the device.
         */
        if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state))
                btrfs_scratch_superblocks(device->bdev, device->name->str);

        btrfs_close_bdev(device);
        call_rcu(&device->rcu, free_device_rcu);

        if (cur_devices->open_devices == 0) {
                while (fs_devices) {
                        if (fs_devices->seed == cur_devices) {
                                fs_devices->seed = cur_devices->seed;
                                break;
                        }
                        fs_devices = fs_devices->seed;
                }
                cur_devices->seed = NULL;
                close_fs_devices(cur_devices);
                free_fs_devices(cur_devices);
        }

out:
        mutex_unlock(&uuid_mutex);
        return ret;

error_undo:
        if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) {
                mutex_lock(&fs_info->chunk_mutex);
                list_add(&device->dev_alloc_list,
                         &fs_devices->alloc_list);
                device->fs_devices->rw_devices++;
                mutex_unlock(&fs_info->chunk_mutex);
        }
        goto out;
}

void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_fs_info *fs_info,
                                        struct btrfs_device *srcdev)
{
        struct btrfs_fs_devices *fs_devices;

        lockdep_assert_held(&fs_info->fs_devices->device_list_mutex);

        /*
         * in case of fs with no seed, srcdev->fs_devices will point
         * to fs_devices of fs_info. However when the dev being replaced is
         * a seed dev it will point to the seed's local fs_devices. In short
         * srcdev will have its correct fs_devices in both the cases.
         */
        fs_devices = srcdev->fs_devices;

        list_del_rcu(&srcdev->dev_list);
        list_del(&srcdev->dev_alloc_list);
        fs_devices->num_devices--;
        if (test_bit(BTRFS_DEV_STATE_MISSING, &srcdev->dev_state))
                fs_devices->missing_devices--;

        if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &srcdev->dev_state))
                fs_devices->rw_devices--;

        if (srcdev->bdev)
                fs_devices->open_devices--;
}

void btrfs_rm_dev_replace_free_srcdev(struct btrfs_fs_info *fs_info,
                                      struct btrfs_device *srcdev)
{
        struct btrfs_fs_devices *fs_devices = srcdev->fs_devices;

        if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &srcdev->dev_state)) {
                /* zero out the old super if it is writable */
                btrfs_scratch_superblocks(srcdev->bdev, srcdev->name->str);
        }

        btrfs_close_bdev(srcdev);
        call_rcu(&srcdev->rcu, free_device_rcu);

        /* if this is no devs we rather delete the fs_devices */
        if (!fs_devices->num_devices) {
                struct btrfs_fs_devices *tmp_fs_devices;

                /*
                 * On a mounted FS, num_devices can't be zero unless it's a
                 * seed. In case of a seed device being replaced, the replace
                 * target added to the sprout FS, so there will be no more
                 * device left under the seed FS.
                 */
                ASSERT(fs_devices->seeding);

                tmp_fs_devices = fs_info->fs_devices;
                while (tmp_fs_devices) {
                        if (tmp_fs_devices->seed == fs_devices) {
                                tmp_fs_devices->seed = fs_devices->seed;
                                break;
                        }
                        tmp_fs_devices = tmp_fs_devices->seed;
                }
                fs_devices->seed = NULL;
                close_fs_devices(fs_devices);
                free_fs_devices(fs_devices);
        }
}

void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
                                      struct btrfs_device *tgtdev)
{
        struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;

        WARN_ON(!tgtdev);
        mutex_lock(&fs_devices->device_list_mutex);

        btrfs_sysfs_rm_device_link(fs_devices, tgtdev);

        if (tgtdev->bdev)
                fs_devices->open_devices--;

        fs_devices->num_devices--;

        btrfs_assign_next_active_device(fs_info, tgtdev, NULL);

        list_del_rcu(&tgtdev->dev_list);

        mutex_unlock(&fs_devices->device_list_mutex);

        /*
         * The update_dev_time() with in btrfs_scratch_superblocks()
         * may lead to a call to btrfs_show_devname() which will try
         * to hold device_list_mutex. And here this device
         * is already out of device list, so we don't have to hold
         * the device_list_mutex lock.
         */
        btrfs_scratch_superblocks(tgtdev->bdev, tgtdev->name->str);

        btrfs_close_bdev(tgtdev);
        call_rcu(&tgtdev->rcu, free_device_rcu);
}

static int btrfs_find_device_by_path(struct btrfs_fs_info *fs_info,
                                     const char *device_path,
                                     struct btrfs_device **device)
{
        int ret = 0;
        struct btrfs_super_block *disk_super;
        u64 devid;
        u8 *dev_uuid;
        struct block_device *bdev;
        struct buffer_head *bh;

        *device = NULL;
        ret = btrfs_get_bdev_and_sb(device_path, FMODE_READ,
                                    fs_info->bdev_holder, 0, &bdev, &bh);
        if (ret)
                return ret;
        disk_super = (struct btrfs_super_block *)bh->b_data;
        devid = btrfs_stack_device_id(&disk_super->dev_item);
        dev_uuid = disk_super->dev_item.uuid;
        *device = btrfs_find_device(fs_info, devid, dev_uuid, disk_super->fsid);
        brelse(bh);
        if (!*device)
                ret = -ENOENT;
        blkdev_put(bdev, FMODE_READ);
        return ret;
}

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

                devices = &fs_info->fs_devices->devices;
                list_for_each_entry(tmp, devices, dev_list) {
                        if (test_bit(BTRFS_DEV_STATE_IN_FS_METADATA,
                                        &tmp->dev_state) && !tmp->bdev) {
                                *device = tmp;
                                break;
                        }
                }

                if (!*device)
                        return BTRFS_ERROR_DEV_MISSING_NOT_FOUND;

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

/*
 * Lookup a device given by device id, or the path if the id is 0.
 */
int btrfs_find_device_by_devspec(struct btrfs_fs_info *fs_info, u64 devid,
                                 const char *devpath,
                                 struct btrfs_device **device)
{
        int ret;

        if (devid) {
                ret = 0;
                *device = btrfs_find_device(fs_info, devid, NULL, NULL);
                if (!*device)
                        ret = -ENOENT;
        } else {
                if (!devpath || !devpath[0])
                        return -EINVAL;

                ret = btrfs_find_device_missing_or_by_path(fs_info, devpath,
                                                           device);
        }
        return ret;
}

/*
 * does all the dirty work required for changing file system's UUID.
 */
static int btrfs_prepare_sprout(struct btrfs_fs_info *fs_info)
{
        struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
        struct btrfs_fs_devices *old_devices;
        struct btrfs_fs_devices *seed_devices;
        struct btrfs_super_block *disk_super = fs_info->super_copy;
        struct btrfs_device *device;
        u64 super_flags;

        lockdep_assert_held(&uuid_mutex);
        if (!fs_devices->seeding)
                return -EINVAL;

        seed_devices = alloc_fs_devices(NULL);
        if (IS_ERR(seed_devices))
                return PTR_ERR(seed_devices);

        old_devices = clone_fs_devices(fs_devices);
        if (IS_ERR(old_devices)) {
                kfree(seed_devices);
                return PTR_ERR(old_devices);
        }

        list_add(&old_devices->fs_list, &fs_uuids);

        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);
        mutex_init(&seed_devices->device_list_mutex);

        mutex_lock(&fs_info->fs_devices->device_list_mutex);
        list_splice_init_rcu(&fs_devices->devices, &seed_devices->devices,
                              synchronize_rcu);
        list_for_each_entry(device, &seed_devices->devices, dev_list)
                device->fs_devices = seed_devices;

        mutex_lock(&fs_info->chunk_mutex);
        list_splice_init(&fs_devices->alloc_list, &seed_devices->alloc_list);
        mutex_unlock(&fs_info->chunk_mutex);

        fs_devices->seeding = 0;
        fs_devices->num_devices = 0;
        fs_devices->open_devices = 0;
        fs_devices->missing_devices = 0;
        fs_devices->rotating = 0;
        fs_devices->seed = seed_devices;

        generate_random_uuid(fs_devices->fsid);
        memcpy(fs_info->fsid, fs_devices->fsid, BTRFS_FSID_SIZE);
        memcpy(disk_super->fsid, fs_devices->fsid, BTRFS_FSID_SIZE);
        mutex_unlock(&fs_info->fs_devices->device_list_mutex);

        super_flags = btrfs_super_flags(disk_super) &
                      ~BTRFS_SUPER_FLAG_SEEDING;
        btrfs_set_super_flags(disk_super, super_flags);

        return 0;
}

/*
 * Store the expected generation for seed devices in device items.
 */
static int btrfs_finish_sprout(struct btrfs_trans_handle *trans,
                               struct btrfs_fs_info *fs_info)
{
        struct btrfs_root *root = fs_info->chunk_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_FSID_SIZE];
        u8 dev_uuid[BTRFS_UUID_SIZE];
        u64 devid;
        int ret;

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

        key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
        key.offset = 0;
        key.type = BTRFS_DEV_ITEM_KEY;

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

                leaf = path->nodes[0];
next_slot:
                if (path->slots[0] >= btrfs_header_nritems(leaf)) {
                        ret = btrfs_next_leaf(root, path);
                        if (ret > 0)
                                break;
                        if (ret < 0)
                                goto error;
                        leaf = path->nodes[0];
                        btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
                        btrfs_release_path(path);
                        continue;
                }

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

                dev_item = btrfs_item_ptr(leaf, path->slots[0],
                                          struct btrfs_dev_item);
                devid = btrfs_device_id(leaf, dev_item);
                read_extent_buffer(leaf, dev_uuid, btrfs_device_uuid(dev_item),
                                   BTRFS_UUID_SIZE);
                read_extent_buffer(leaf, fs_uuid, btrfs_device_fsid(dev_item),
                                   BTRFS_FSID_SIZE);
                device = btrfs_find_device(fs_info, devid, dev_uuid, fs_uuid);
                BUG_ON(!device); /* Logic error */

                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;
}

int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path)
{
        struct btrfs_root *root = fs_info->dev_root;
        struct request_queue *q;
        struct btrfs_trans_handle *trans;
        struct btrfs_device *device;
        struct block_device *bdev;
        struct list_head *devices;
        struct super_block *sb = fs_info->sb;
        struct rcu_string *name;
        u64 tmp;
        int seeding_dev = 0;
        int ret = 0;
        bool unlocked = false;

        if (sb_rdonly(sb) && !fs_info->fs_devices->seeding)
                return -EROFS;

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

        if (fs_info->fs_devices->seeding) {
                seeding_dev = 1;
                down_write(&sb->s_umount);
                mutex_lock(&uuid_mutex);
        }

        filemap_write_and_wait(bdev->bd_inode->i_mapping);

        devices = &fs_info->fs_devices->devices;

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

        device = btrfs_alloc_device(fs_info, NULL, NULL);
        if (IS_ERR(device)) {
                /* we can safely leave the fs_devices entry around */
                ret = PTR_ERR(device);
                goto error;
        }

        name = rcu_string_strdup(device_path, GFP_KERNEL);
        if (!name) {
                ret = -ENOMEM;
                goto error_free_device;
        }
        rcu_assign_pointer(device->name, name);

        trans = btrfs_start_transaction(root, 0);
        if (IS_ERR(trans)) {
                ret = PTR_ERR(trans);
                goto error_free_device;
        }

        q = bdev_get_queue(bdev);
        set_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state);
        device->generation = trans->transid;
        device->io_width = fs_info->sectorsize;
        device->io_align = fs_info->sectorsize;
        device->sector_size = fs_info->sectorsize;
        device->total_bytes = round_down(i_size_read(bdev->bd_inode),
                                         fs_info->sectorsize);
        device->disk_total_bytes = device->total_bytes;
        device->commit_total_bytes = device->total_bytes;
        device->fs_info = fs_info;
        device->bdev = bdev;
        set_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state);
        clear_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state);
        device->mode = FMODE_EXCL;
        device->dev_stats_valid = 1;
        set_blocksize(device->bdev, BTRFS_BDEV_BLOCKSIZE);

        if (seeding_dev) {
                sb->s_flags &= ~SB_RDONLY;
                ret = btrfs_prepare_sprout(fs_info);
                if (ret) {
                        btrfs_abort_transaction(trans, ret);
                        goto error_trans;
                }
        }

        device->fs_devices = fs_info->fs_devices;

        mutex_lock(&fs_info->fs_devices->device_list_mutex);
        mutex_lock(&fs_info->chunk_mutex);
        list_add_rcu(&device->dev_list, &fs_info->fs_devices->devices);
        list_add(&device->dev_alloc_list,
                 &fs_info->fs_devices->alloc_list);
        fs_info->fs_devices->num_devices++;
        fs_info->fs_devices->open_devices++;
        fs_info->fs_devices->rw_devices++;
        fs_info->fs_devices->total_devices++;
        fs_info->fs_devices->total_rw_bytes += device->total_bytes;

        atomic64_add(device->total_bytes, &fs_info->free_chunk_space);

        if (!blk_queue_nonrot(q))
                fs_info->fs_devices->rotating = 1;

        tmp = btrfs_super_total_bytes(fs_info->super_copy);
        btrfs_set_super_total_bytes(fs_info->super_copy,
                round_down(tmp + device->total_bytes, fs_info->sectorsize));

        tmp = btrfs_super_num_devices(fs_info->super_copy);
        btrfs_set_super_num_devices(fs_info->super_copy, tmp + 1);

        /* add sysfs device entry */
        btrfs_sysfs_add_device_link(fs_info->fs_devices, device);

        /*
         * we've got more storage, clear any full flags on the space
         * infos
         */
        btrfs_clear_space_info_full(fs_info);

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

        if (seeding_dev) {
                mutex_lock(&fs_info->chunk_mutex);
                ret = init_first_rw_device(trans, fs_info);
                mutex_unlock(&fs_info->chunk_mutex);
                if (ret) {
                        btrfs_abort_transaction(trans, ret);
                        goto error_sysfs;
                }
        }

        ret = btrfs_add_dev_item(trans, fs_info, device);
        if (ret) {
                btrfs_abort_transaction(trans, ret);
                goto error_sysfs;
        }

        if (seeding_dev) {
                char fsid_buf[BTRFS_UUID_UNPARSED_SIZE];

                ret = btrfs_finish_sprout(trans, fs_info);
                if (ret) {
                        btrfs_abort_transaction(trans, ret);
                        goto error_sysfs;
                }

                /* Sprouting would change fsid of the mounted root,
                 * so rename the fsid on the sysfs
                 */
                snprintf(fsid_buf, BTRFS_UUID_UNPARSED_SIZE, "%pU",
                                                fs_info->fsid);
                if (kobject_rename(&fs_info->fs_devices->fsid_kobj, fsid_buf))
                        btrfs_warn(fs_info,
                                   "sysfs: failed to create fsid for sprout");
        }

        ret = btrfs_commit_transaction(trans);

        if (seeding_dev) {
                mutex_unlock(&uuid_mutex);
                up_write(&sb->s_umount);
                unlocked = true;

                if (ret) /* transaction commit */
                        return ret;

                ret = btrfs_relocate_sys_chunks(fs_info);
                if (ret < 0)
                        btrfs_handle_fs_error(fs_info, ret,
                                    "Failed to relocate sys chunks after device initialization. This can be fixed using the \"btrfs balance\" command.");
                trans = btrfs_attach_transaction(root);
                if (IS_ERR(trans)) {
                        if (PTR_ERR(trans) == -ENOENT)
                                return 0;
                        ret = PTR_ERR(trans);
                        trans = NULL;
                        goto error_sysfs;
                }
                ret = btrfs_commit_transaction(trans);
        }

        /* Update ctime/mtime for libblkid */
        update_dev_time(device_path);
        return ret;

error_sysfs:
        btrfs_sysfs_rm_device_link(fs_info->fs_devices, device);
error_trans:
        if (seeding_dev)
                sb->s_flags |= SB_RDONLY;
        if (trans)
                btrfs_end_transaction(trans);
error_free_device:
        btrfs_free_device(device);
error:
        blkdev_put(bdev, FMODE_EXCL);
        if (seeding_dev && !unlocked) {
                mutex_unlock(&uuid_mutex);
                up_write(&sb->s_umount);
        }
        return ret;
}

static noinline int btrfs_update_device(struct btrfs_trans_handle *trans,
                                        struct btrfs_device *device)
{
        int ret;
        struct btrfs_path *path;
        struct btrfs_root *root = device->fs_info->chunk_root;
        struct btrfs_dev_item *dev_item;
        struct extent_buffer *leaf;
        struct btrfs_key key;

        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);
        btrfs_set_device_total_bytes(leaf, dev_item,
                                     btrfs_device_get_disk_total_bytes(device));
        btrfs_set_device_bytes_used(leaf, dev_item,
                                    btrfs_device_get_bytes_used(device));
        btrfs_mark_buffer_dirty(leaf);

out:
        btrfs_free_path(path);
        return ret;
}

int btrfs_grow_device(struct btrfs_trans_handle *trans,
                      struct btrfs_device *device, u64 new_size)
{
        struct btrfs_fs_info *fs_info = device->fs_info;
        struct btrfs_super_block *super_copy = fs_info->super_copy;
        struct btrfs_fs_devices *fs_devices;
        u64 old_total;
        u64 diff;

        if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state))
                return -EACCES;

        new_size = round_down(new_size, fs_info->sectorsize);

        mutex_lock(&fs_info->chunk_mutex);
        old_total = btrfs_super_total_bytes(super_copy);
        diff = round_down(new_size - device->total_bytes, fs_info->sectorsize);

        if (new_size <= device->total_bytes ||
            test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) {
                mutex_unlock(&fs_info->chunk_mutex);
                return -EINVAL;
        }

        fs_devices = fs_info->fs_devices;

        btrfs_set_super_total_bytes(super_copy,
                        round_down(old_total + diff, fs_info->sectorsize));
        device->fs_devices->total_rw_bytes += diff;

        btrfs_device_set_total_bytes(device, new_size);
        btrfs_device_set_disk_total_bytes(device, new_size);
        btrfs_clear_space_info_full(device->fs_info);
        if (list_empty(&device->resized_list))
                list_add_tail(&device->resized_list,
                              &fs_devices->resized_devices);
        mutex_unlock(&fs_info->chunk_mutex);

        return btrfs_update_device(trans, device);
}

static int btrfs_free_chunk(struct btrfs_trans_handle *trans,
                            struct btrfs_fs_info *fs_info, u64 chunk_offset)
{
        struct btrfs_root *root = fs_info->chunk_root;
        int ret;
        struct btrfs_path *path;
        struct btrfs_key key;

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

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

        ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
        if (ret < 0)
                goto out;
        else if (ret > 0) { /* Logic error or corruption */
                btrfs_handle_fs_error(fs_info, -ENOENT,
                                      "Failed lookup while freeing chunk.");
                ret = -ENOENT;
                goto out;
        }

        ret = btrfs_del_item(trans, root, path);
        if (ret < 0)
                btrfs_handle_fs_error(fs_info, ret,
                                      "Failed to delete chunk item.");
out:
        btrfs_free_path(path);
        return ret;
}

static int btrfs_del_sys_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset)
{
        struct btrfs_super_block *super_copy = fs_info->super_copy;
        struct btrfs_disk_key *disk_key;
        struct btrfs_chunk *chunk;
        u8 *ptr;
        int ret = 0;
        u32 num_stripes;
        u32 array_size;
        u32 len = 0;
        u32 cur;
        struct btrfs_key key;

        mutex_lock(&fs_info->chunk_mutex);
        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 == BTRFS_FIRST_CHUNK_TREE_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;
                }
        }
        mutex_unlock(&fs_info->chunk_mutex);
        return ret;
}

static struct extent_map *get_chunk_map(struct btrfs_fs_info *fs_info,
                                        u64 logical, u64 length)
{
        struct extent_map_tree *em_tree;
        struct extent_map *em;

        em_tree = &fs_info->mapping_tree.map_tree;
        read_lock(&em_tree->lock);
        em = lookup_extent_mapping(em_tree, logical, length);
        read_unlock(&em_tree->lock);

        if (!em) {
                btrfs_crit(fs_info, "unable to find logical %llu length %llu",
                           logical, length);
                return ERR_PTR(-EINVAL);
        }

        if (em->start > logical || em->start + em->len < logical) {
                btrfs_crit(fs_info,
                           "found a bad mapping, wanted %llu-%llu, found %llu-%llu",
                           logical, length, em->start, em->start + em->len);
                free_extent_map(em);
                return ERR_PTR(-EINVAL);
        }

        /* callers are responsible for dropping em's ref. */
        return em;
}

int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
                       struct btrfs_fs_info *fs_info, u64 chunk_offset)
{
        struct extent_map *em;
        struct map_lookup *map;
        u64 dev_extent_len = 0;
        int i, ret = 0;
        struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;

        em = get_chunk_map(fs_info, chunk_offset, 1);
        if (IS_ERR(em)) {
                /*
                 * This is a logic error, but we don't want to just rely on the
                 * user having built with ASSERT enabled, so if ASSERT doesn't
                 * do anything we still error out.
                 */
                ASSERT(0);
                return PTR_ERR(em);
        }
        map = em->map_lookup;
        mutex_lock(&fs_info->chunk_mutex);
        check_system_chunk(trans, fs_info, map->type);
        mutex_unlock(&fs_info->chunk_mutex);

        /*
         * Take the device list mutex to prevent races with the final phase of
         * a device replace operation that replaces the device object associated
         * with map stripes (dev-replace.c:btrfs_dev_replace_finishing()).
         */
        mutex_lock(&fs_devices->device_list_mutex);
        for (i = 0; i < map->num_stripes; i++) {
                struct btrfs_device *device = map->stripes[i].dev;
                ret = btrfs_free_dev_extent(trans, device,
                                            map->stripes[i].physical,
                                            &dev_extent_len);
                if (ret) {
                        mutex_unlock(&fs_devices->device_list_mutex);
                        btrfs_abort_transaction(trans, ret);
                        goto out;
                }

                if (device->bytes_used > 0) {
                        mutex_lock(&fs_info->chunk_mutex);
                        btrfs_device_set_bytes_used(device,
                                        device->bytes_used - dev_extent_len);
                        atomic64_add(dev_extent_len, &fs_info->free_chunk_space);
                        btrfs_clear_space_info_full(fs_info);
                        mutex_unlock(&fs_info->chunk_mutex);
                }

                if (map->stripes[i].dev) {
                        ret = btrfs_update_device(trans, map->stripes[i].dev);
                        if (ret) {
                                mutex_unlock(&fs_devices->device_list_mutex);
                                btrfs_abort_transaction(trans, ret);
                                goto out;
                        }
                }
        }
        mutex_unlock(&fs_devices->device_list_mutex);

        ret = btrfs_free_chunk(trans, fs_info, chunk_offset);
        if (ret) {
                btrfs_abort_transaction(trans, ret);
                goto out;
        }

        trace_btrfs_chunk_free(fs_info, map, chunk_offset, em->len);

        if (map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
                ret = btrfs_del_sys_chunk(fs_info, chunk_offset);
                if (ret) {
                        btrfs_abort_transaction(trans, ret);
                        goto out;
                }
        }

        ret = btrfs_remove_block_group(trans, fs_info, chunk_offset, em);
        if (ret) {
                btrfs_abort_transaction(trans, ret);
                goto out;
        }

out:
        /* once for us */
        free_extent_map(em);
        return ret;
}

static int btrfs_relocate_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset)
{
        struct btrfs_root *root = fs_info->chunk_root;
        struct btrfs_trans_handle *trans;
        int ret;

        /*
         * Prevent races with automatic removal of unused block groups.
         * After we relocate and before we remove the chunk with offset
         * chunk_offset, automatic removal of the block group can kick in,
         * resulting in a failure when calling btrfs_remove_chunk() below.
         *
         * Make sure to acquire this mutex before doing a tree search (dev
         * or chunk trees) to find chunks. Otherwise the cleaner kthread might
         * call btrfs_remove_chunk() (through btrfs_delete_unused_bgs()) after
         * we release the path used to search the chunk/dev tree and before
         * the current task acquires this mutex and calls us.
         */
        lockdep_assert_held(&fs_info->delete_unused_bgs_mutex);

        ret = btrfs_can_relocate(fs_info, chunk_offset);
        if (ret)
                return -ENOSPC;

        /* step one, relocate all the extents inside this chunk */
        btrfs_scrub_pause(fs_info);
        ret = btrfs_relocate_block_group(fs_info, chunk_offset);
        btrfs_scrub_continue(fs_info);
        if (ret)
                return ret;

        /*
         * We add the kobjects here (and after forcing data chunk creation)
         * since relocation is the only place we'll create chunks of a new
         * type at runtime.  The only place where we'll remove the last
         * chunk of a type is the call immediately below this one.  Even
         * so, we're protected against races with the cleaner thread since
         * we're covered by the delete_unused_bgs_mutex.
         */
        btrfs_add_raid_kobjects(fs_info);

        trans = btrfs_start_trans_remove_block_group(root->fs_info,
                                                     chunk_offset);
        if (IS_ERR(trans)) {
                ret = PTR_ERR(trans);
                btrfs_handle_fs_error(root->fs_info, ret, NULL);
                return ret;
        }

        /*
         * step two, delete the device extents and the
         * chunk tree entries
         */
        ret = btrfs_remove_chunk(trans, fs_info, chunk_offset);
        btrfs_end_transaction(trans);
        return ret;
}

static int btrfs_relocate_sys_chunks(struct btrfs_fs_info *fs_info)
{
        struct btrfs_root *chunk_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_type;
        bool retried = false;
        int failed = 0;
        int ret;

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

again:
        key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
        key.offset = (u64)-1;
        key.type = BTRFS_CHUNK_ITEM_KEY;

        while (1) {
                mutex_lock(&fs_info->delete_unused_bgs_mutex);
                ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
                if (ret < 0) {
                        mutex_unlock(&fs_info->delete_unused_bgs_mutex);
                        goto error;
                }
                BUG_ON(ret == 0); /* Corruption */

                ret = btrfs_previous_item(chunk_root, path, key.objectid,
                                          key.type);
                if (ret)
                        mutex_unlock(&fs_info->delete_unused_bgs_mutex);
                if (ret < 0)
                        goto error;
                if (ret > 0)
                        break;

                leaf = path->nodes[0];
                btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);

                chunk = btrfs_item_ptr(leaf, path->slots[0],
                                       struct btrfs_chunk);
                chunk_type = btrfs_chunk_type(leaf, chunk);
                btrfs_release_path(path);

                if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) {
                        ret = btrfs_relocate_chunk(fs_info, found_key.offset);
                        if (ret == -ENOSPC)
                                failed++;
                        else
                                BUG_ON(ret);
                }
                mutex_unlock(&fs_info->delete_unused_bgs_mutex);

                if (found_key.offset == 0)
                        break;
                key.offset = found_key.offset - 1;
        }
        ret = 0;
        if (failed && !retried) {
                failed = 0;
                retried = true;
                goto again;
        } else if (WARN_ON(failed && retried)) {
                ret = -ENOSPC;
        }
error:
        btrfs_free_path(path);
        return ret;
}

/*
 * return 1 : allocate a data chunk successfully,
 * return <0: errors during allocating a data chunk,
 * return 0 : no need to allocate a data chunk.
 */
static int btrfs_may_alloc_data_chunk(struct btrfs_fs_info *fs_info,
                                      u64 chunk_offset)
{
        struct btrfs_block_group_cache *cache;
        u64 bytes_used;
        u64 chunk_type;

        cache = btrfs_lookup_block_group(fs_info, chunk_offset);
        ASSERT(cache);
        chunk_type = cache->flags;
        btrfs_put_block_group(cache);

        if (chunk_type & BTRFS_BLOCK_GROUP_DATA) {
                spin_lock(&fs_info->data_sinfo->lock);
                bytes_used = fs_info->data_sinfo->bytes_used;
                spin_unlock(&fs_info->data_sinfo->lock);

                if (!bytes_used) {
                        struct btrfs_trans_handle *trans;
                        int ret;

                        trans = btrfs_join_transaction(fs_info->tree_root);
                        if (IS_ERR(trans))
                                return PTR_ERR(trans);

                        ret = btrfs_force_chunk_alloc(trans, fs_info,
                                                      BTRFS_BLOCK_GROUP_DATA);
                        btrfs_end_transaction(trans);
                        if (ret < 0)
                                return ret;

                        btrfs_add_raid_kobjects(fs_info);

                        return 1;
                }
        }
        return 0;
}

static int insert_balance_item(struct btrfs_fs_info *fs_info,
                               struct btrfs_balance_control *bctl)
{
        struct btrfs_root *root = fs_info->tree_root;
        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_TEMPORARY_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);

        memzero_extent_buffer(leaf, (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);
        if (err && !ret)
                ret = err;
        return ret;
}

static int del_balance_item(struct btrfs_fs_info *fs_info)
{
        struct btrfs_root *root = fs_info->tree_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_TEMPORARY_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);
        if (err && !ret)
                ret = err;
        return ret;
}

/*
 * 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_USAGE_RANGE) &&
            !(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_USAGE_RANGE) &&
            !(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_USAGE_RANGE) &&
            !(bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT)) {
                bctl->meta.flags |= BTRFS_BALANCE_ARGS_USAGE;
                bctl->meta.usage = 90;
        }
}

/*
 * Clear the balance status in fs_info and delete the balance item from disk.
 */
static void reset_balance_state(struct btrfs_fs_info *fs_info)
{
        struct btrfs_balance_control *bctl = fs_info->balance_ctl;
        int ret;

        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);
        ret = del_balance_item(fs_info);
        if (ret)
                btrfs_handle_fs_error(fs_info, ret, NULL);
}

/*
 * Balance filters.  Return 1 if chunk should be filtered out
 * (should not be balanced).
 */
static int chunk_profiles_filter(u64 chunk_type,
                                 struct btrfs_balance_args *bargs)
{
        chunk_type = chunk_to_extended(chunk_type) &
                                BTRFS_EXTENDED_PROFILE_MASK;

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

        return 1;
}

static int chunk_usage_range_filter(struct btrfs_fs_info *fs_info, u64 chunk_offset,
                              struct btrfs_balance_args *bargs)
{
        struct btrfs_block_group_cache *cache;
        u64 chunk_used;
        u64 user_thresh_min;
        u64 user_thresh_max;
        int ret = 1;

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

        if (bargs->usage_min == 0)
                user_thresh_min = 0;
        else
                user_thresh_min = div_factor_fine(cache->key.offset,
                                        bargs->usage_min);

        if (bargs->usage_max == 0)
                user_thresh_max = 1;
        else if (bargs->usage_max > 100)
                user_thresh_max = cache->key.offset;
        else
                user_thresh_max = div_factor_fine(cache->key.offset,
                                        bargs->usage_max);

        if (user_thresh_min <= chunk_used && chunk_used < user_thresh_max)
                ret = 0;

        btrfs_put_block_group(cache);
        return ret;
}

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);

        if (bargs->usage_min == 0)
                user_thresh = 1;
        else if (bargs->usage > 100)
                user_thresh = cache->key.offset;
        else
                user_thresh = div_factor_fine(cache->key.offset,
                                              bargs->usage);

        if (chunk_used < user_thresh)
                ret = 0;

        btrfs_put_block_group(cache);
        return ret;
}

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;
}

/* [pstart, pend) */
static int chunk_drange_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);
        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 = num_stripes / 2;
        } else if (btrfs_chunk_type(leaf, chunk) & BTRFS_BLOCK_GROUP_RAID5) {
                factor = num_stripes - 1;
        } else if (btrfs_chunk_type(leaf, chunk) & BTRFS_BLOCK_GROUP_RAID6) {
                factor = num_stripes - 2;
        } else {
                factor = num_stripes;
        }

        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);
                stripe_length = div_u64(stripe_length, factor);

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

        return 1;
}

/* [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;
}

static int chunk_stripes_range_filter(struct extent_buffer *leaf,
                               struct btrfs_chunk *chunk,
                               struct btrfs_balance_args *bargs)
{
        int num_stripes = btrfs_chunk_num_stripes(leaf, chunk);

        if (bargs->stripes_min <= num_stripes
                        && num_stripes <= bargs->stripes_max)
                return 0;

        return 1;
}

static int chunk_soft_convert_filter(u64 chunk_type,
                                     struct btrfs_balance_args *bargs)
{
        if (!(bargs->flags & BTRFS_BALANCE_ARGS_CONVERT))
                return 0;

        chunk_type = chunk_to_extended(chunk_type) &
                                BTRFS_EXTENDED_PROFILE_MASK;

        if (bargs->target == chunk_type)
                return 1;

        return 0;
}

static int should_balance_chunk(struct btrfs_fs_info *fs_info,
                                struct extent_buffer *leaf,
                                struct btrfs_chunk *chunk, u64 chunk_offset)
{
        struct btrfs_balance_control *bctl = 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;

        /* profiles filter */
        if ((bargs->flags & BTRFS_BALANCE_ARGS_PROFILES) &&
            chunk_profiles_filter(chunk_type, bargs)) {
                return 0;
        }

        /* usage filter */
        if ((bargs->flags & BTRFS_BALANCE_ARGS_USAGE) &&
            chunk_usage_filter(fs_info, chunk_offset, bargs)) {
                return 0;
        } else if ((bargs->flags & BTRFS_BALANCE_ARGS_USAGE_RANGE) &&
            chunk_usage_range_filter(fs_info, chunk_offset, bargs)) {
                return 0;
        }

        /* devid filter */
        if ((bargs->flags & BTRFS_BALANCE_ARGS_DEVID) &&
            chunk_devid_filter(leaf, chunk, bargs)) {
                return 0;
        }

        /* drange filter, makes sense only with devid filter */
        if ((bargs->flags & BTRFS_BALANCE_ARGS_DRANGE) &&
            chunk_drange_filter(leaf, chunk, bargs)) {
                return 0;
        }

        /* vrange filter */
        if ((bargs->flags & BTRFS_BALANCE_ARGS_VRANGE) &&
            chunk_vrange_filter(leaf, chunk, chunk_offset, bargs)) {
                return 0;
        }

        /* stripes filter */
        if ((bargs->flags & BTRFS_BALANCE_ARGS_STRIPES_RANGE) &&
            chunk_stripes_range_filter(leaf, chunk, bargs)) {
                return 0;
        }

        /* soft profile changing mode */
        if ((bargs->flags & BTRFS_BALANCE_ARGS_SOFT) &&
            chunk_soft_convert_filter(chunk_type, bargs)) {
                return 0;
        }

        /*
         * limited by count, must be the last filter
         */
        if ((bargs->flags & BTRFS_BALANCE_ARGS_LIMIT)) {
                if (bargs->limit == 0)
                        return 0;
                else
                        bargs->limit--;
        } else if ((bargs->flags & BTRFS_BALANCE_ARGS_LIMIT_RANGE)) {
                /*
                 * Same logic as the 'limit' filter; the minimum cannot be
                 * determined here because we do not have the global information
                 * about the count of all chunks that satisfy the filters.
                 */
                if (bargs->limit_max == 0)
                        return 0;
                else
                        bargs->limit_max--;
        }

        return 1;
}

static int __btrfs_balance(struct btrfs_fs_info *fs_info)
{
        struct btrfs_balance_control *bctl = fs_info->balance_ctl;
        struct btrfs_root *chunk_root = fs_info->chunk_root;
        struct btrfs_root *dev_root = fs_info->dev_root;
        struct list_head *devices;
        struct btrfs_device *device;
        u64 old_size;
        u64 size_to_free;
        u64 chunk_type;
        struct btrfs_chunk *chunk;
        struct btrfs_path *path = NULL;
        struct btrfs_key key;
        struct btrfs_key found_key;
        struct btrfs_trans_handle *trans;
        struct extent_buffer *leaf;
        int slot;
        int ret;
        int enospc_errors = 0;
        bool counting = true;
        /* The single value limit and min/max limits use the same bytes in the */
        u64 limit_data = bctl->data.limit;
        u64 limit_meta = bctl->meta.limit;
        u64 limit_sys = bctl->sys.limit;
        u32 count_data = 0;
        u32 count_meta = 0;
        u32 count_sys = 0;
        int chunk_reserved = 0;

        /* step one make some room on all the devices */
        devices = &fs_info->fs_devices->devices;
        list_for_each_entry(device, devices, dev_list) {
                old_size = btrfs_device_get_total_bytes(device);
                size_to_free = div_factor(old_size, 1);
                size_to_free = min_t(u64, size_to_free, SZ_1M);
                if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state) ||
                    btrfs_device_get_total_bytes(device) -
                    btrfs_device_get_bytes_used(device) > size_to_free ||
                    test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state))
                        continue;

                ret = btrfs_shrink_device(device, old_size - size_to_free);
                if (ret == -ENOSPC)
                        break;
                if (ret) {
                        /* btrfs_shrink_device never returns ret > 0 */
                        WARN_ON(ret > 0);
                        goto error;
                }

                trans = btrfs_start_transaction(dev_root, 0);
                if (IS_ERR(trans)) {
                        ret = PTR_ERR(trans);
                        btrfs_info_in_rcu(fs_info,
                 "resize: unable to start transaction after shrinking device %s (error %d), old size %llu, new size %llu",
                                          rcu_str_deref(device->name), ret,
                                          old_size, old_size - size_to_free);
                        goto error;
                }

                ret = btrfs_grow_device(trans, device, old_size);
                if (ret) {
                        btrfs_end_transaction(trans);
                        /* btrfs_grow_device never returns ret > 0 */
                        WARN_ON(ret > 0);
                        btrfs_info_in_rcu(fs_info,
                 "resize: unable to grow device after shrinking device %s (error %d), old size %llu, new size %llu",
                                          rcu_str_deref(device->name), ret,
                                          old_size, old_size - size_to_free);
                        goto error;
                }

                btrfs_end_transaction(trans);
        }

        /* step two, relocate all the chunks */
        path = btrfs_alloc_path();
        if (!path) {
                ret = -ENOMEM;
                goto error;
        }

        /* zero out stat counters */
        spin_lock(&fs_info->balance_lock);
        memset(&bctl->stat, 0, sizeof(bctl->stat));
        spin_unlock(&fs_info->balance_lock);
again:
        if (!counting) {
                /*
                 * The single value limit and min/max limits use the same bytes
                 * in the
                 */
                bctl->data.limit = limit_data;
                bctl->meta.limit = limit_meta;
                bctl->sys.limit = limit_sys;
        }
        key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
        key.offset = (u64)-1;
        key.type = BTRFS_CHUNK_ITEM_KEY;

        while (1) {
                if ((!counting && atomic_read(&fs_info->balance_pause_req)) ||
                    atomic_read(&fs_info->balance_cancel_req)) {
                        ret = -ECANCELED;
                        goto error;
                }

                mutex_lock(&fs_info->delete_unused_bgs_mutex);
                ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
                if (ret < 0) {
                        mutex_unlock(&fs_info->delete_unused_bgs_mutex);
                        goto error;
                }

                /*
                 * this shouldn't happen, it means the last relocate
                 * failed
                 */
                if (ret == 0)
                        BUG(); /* FIXME break ? */

                ret = btrfs_previous_item(chunk_root, path, 0,
                                          BTRFS_CHUNK_ITEM_KEY);
                if (ret) {
                        mutex_unlock(&fs_info->delete_unused_bgs_mutex);
                        ret = 0;
                        break;
                }

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

                if (found_key.objectid != key.objectid) {
                        mutex_unlock(&fs_info->delete_unused_bgs_mutex);
                        break;
                }

                chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);
                chunk_type = btrfs_chunk_type(leaf, chunk);

                if (!counting) {
                        spin_lock(&fs_info->balance_lock);
                        bctl->stat.considered++;
                        spin_unlock(&fs_info->balance_lock);
                }

                ret = should_balance_chunk(fs_info, leaf, chunk,
                                           found_key.offset);

                btrfs_release_path(path);
                if (!ret) {
                        mutex_unlock(&fs_info->delete_unused_bgs_mutex);
                        goto loop;
                }

                if (counting) {
                        mutex_unlock(&fs_info->delete_unused_bgs_mutex);
                        spin_lock(&fs_info->balance_lock);
                        bctl->stat.expected++;
                        spin_unlock(&fs_info->balance_lock);

                        if (chunk_type & BTRFS_BLOCK_GROUP_DATA)
                                count_data++;
                        else if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM)
                                count_sys++;
                        else if (chunk_type & BTRFS_BLOCK_GROUP_METADATA)
                                count_meta++;

                        goto loop;
                }

                /*
                 * Apply limit_min filter, no need to check if the LIMITS
                 * filter is used, limit_min is 0 by default
                 */
                if (((chunk_type & BTRFS_BLOCK_GROUP_DATA) &&
                                        count_data < bctl->data.limit_min)
                                || ((chunk_type & BTRFS_BLOCK_GROUP_METADATA) &&
                                        count_meta < bctl->meta.limit_min)
                                || ((chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) &&
                                        count_sys < bctl->sys.limit_min)) {
                        mutex_unlock(&fs_info->delete_unused_bgs_mutex);
                        goto loop;
                }

                if (!chunk_reserved) {
                        /*
                         * We may be relocating the only data chunk we have,
                         * which could potentially end up with losing data's
                         * raid profile, so lets allocate an empty one in
                         * advance.
                         */
                        ret = btrfs_may_alloc_data_chunk(fs_info,
                                                         found_key.offset);
                        if (ret < 0) {
                                mutex_unlock(&fs_info->delete_unused_bgs_mutex);
                                goto error;
                        } else if (ret == 1) {
                                chunk_reserved = 1;
                        }
                }

                ret = btrfs_relocate_chunk(fs_info, found_key.offset);
                mutex_unlock(&fs_info->delete_unused_bgs_mutex);
                if (ret && ret != -ENOSPC)
                        goto error;
                if (ret == -ENOSPC) {
                        enospc_errors++;
                } else {
                        spin_lock(&fs_info->balance_lock);
                        bctl->stat.completed++;
                        spin_unlock(&fs_info->balance_lock);
                }
loop:
                if (found_key.offset == 0)
                        break;
                key.offset = found_key.offset - 1;
        }

        if (counting) {
                btrfs_release_path(path);
                counting = false;
                goto again;
        }
error:
        btrfs_free_path(path);
        if (enospc_errors) {
                btrfs_info(fs_info, "%d enospc errors during balance",
                           enospc_errors);
                if (!ret)
                        ret = -ENOSPC;
        }

        return ret;
}

/**
 * alloc_profile_is_valid - see if a given profile is valid and reduced
 * @flags: profile to validate
 * @extended: if true @flags is treated as an extended profile
 */