qgroup.c

// SPDX-License-Identifier: GPL-2.0
/*
 * Copyright (C) 2011 STRATO.  All rights reserved.
 */

#include <linux/sched.h>
#include <linux/pagemap.h>
#include <linux/writeback.h>
#include <linux/blkdev.h>
#include <linux/rbtree.h>
#include <linux/slab.h>
#include <linux/workqueue.h>
#include <linux/btrfs.h>
#include <linux/sizes.h>

#include "ctree.h"
#include "transaction.h"
#include "disk-io.h"
#include "locking.h"
#include "ulist.h"
#include "backref.h"
#include "extent_io.h"
#include "qgroup.h"
#include "block-group.h"

/* TODO XXX FIXME
 *  - subvol delete -> delete when ref goes to 0? delete limits also?
 *  - reorganize keys
 *  - compressed
 *  - sync
 *  - copy also limits on subvol creation
 *  - limit
 *  - caches for ulists
 *  - performance benchmarks
 *  - check all ioctl parameters
 */

/*
 * Helpers to access qgroup reservation
 *
 * Callers should ensure the lock context and type are valid
 */

static u64 qgroup_rsv_total(const struct btrfs_qgroup *qgroup)
{
	u64 ret = 0;
	int i;

	for (i = 0; i < BTRFS_QGROUP_RSV_LAST; i++)
		ret += qgroup->rsv.values[i];

	return ret;
}

#ifdef CONFIG_BTRFS_DEBUG
static const char *qgroup_rsv_type_str(enum btrfs_qgroup_rsv_type type)
{
	if (type == BTRFS_QGROUP_RSV_DATA)
		return "data";
	if (type == BTRFS_QGROUP_RSV_META_PERTRANS)
		return "meta_pertrans";
	if (type == BTRFS_QGROUP_RSV_META_PREALLOC)
		return "meta_prealloc";
	return NULL;
}
#endif

static void qgroup_rsv_add(struct btrfs_fs_info *fs_info,
			   struct btrfs_qgroup *qgroup, u64 num_bytes,
			   enum btrfs_qgroup_rsv_type type)
{
	trace_qgroup_update_reserve(fs_info, qgroup, num_bytes, type);
	qgroup->rsv.values[type] += num_bytes;
}

static void qgroup_rsv_release(struct btrfs_fs_info *fs_info,
			       struct btrfs_qgroup *qgroup, u64 num_bytes,
			       enum btrfs_qgroup_rsv_type type)
{
	trace_qgroup_update_reserve(fs_info, qgroup, -(s64)num_bytes, type);
	if (qgroup->rsv.values[type] >= num_bytes) {
		qgroup->rsv.values[type] -= num_bytes;
		return;
	}
#ifdef CONFIG_BTRFS_DEBUG
	WARN_RATELIMIT(1,
		"qgroup %llu %s reserved space underflow, have %llu to free %llu",
		qgroup->qgroupid, qgroup_rsv_type_str(type),
		qgroup->rsv.values[type], num_bytes);
#endif
	qgroup->rsv.values[type] = 0;
}

static void qgroup_rsv_add_by_qgroup(struct btrfs_fs_info *fs_info,
				     struct btrfs_qgroup *dest,
				     struct btrfs_qgroup *src)
{
	int i;

	for (i = 0; i < BTRFS_QGROUP_RSV_LAST; i++)
		qgroup_rsv_add(fs_info, dest, src->rsv.values[i], i);
}

static void qgroup_rsv_release_by_qgroup(struct btrfs_fs_info *fs_info,
					 struct btrfs_qgroup *dest,
					  struct btrfs_qgroup *src)
{
	int i;

	for (i = 0; i < BTRFS_QGROUP_RSV_LAST; i++)
		qgroup_rsv_release(fs_info, dest, src->rsv.values[i], i);
}

static void btrfs_qgroup_update_old_refcnt(struct btrfs_qgroup *qg, u64 seq,
					   int mod)
{
	if (qg->old_refcnt < seq)
		qg->old_refcnt = seq;
	qg->old_refcnt += mod;
}

static void btrfs_qgroup_update_new_refcnt(struct btrfs_qgroup *qg, u64 seq,
					   int mod)
{
	if (qg->new_refcnt < seq)
		qg->new_refcnt = seq;
	qg->new_refcnt += mod;
}

static inline u64 btrfs_qgroup_get_old_refcnt(struct btrfs_qgroup *qg, u64 seq)
{
	if (qg->old_refcnt < seq)
		return 0;
	return qg->old_refcnt - seq;
}

static inline u64 btrfs_qgroup_get_new_refcnt(struct btrfs_qgroup *qg, u64 seq)
{
	if (qg->new_refcnt < seq)
		return 0;
	return qg->new_refcnt - seq;
}

/*
 * glue structure to represent the relations between qgroups.
 */
struct btrfs_qgroup_list {
	struct list_head next_group;
	struct list_head next_member;
	struct btrfs_qgroup *group;
	struct btrfs_qgroup *member;
};

static inline u64 qgroup_to_aux(struct btrfs_qgroup *qg)
{
	return (u64)(uintptr_t)qg;
}

static inline struct btrfs_qgroup* unode_aux_to_qgroup(struct ulist_node *n)
{
	return (struct btrfs_qgroup *)(uintptr_t)n->aux;
}

static int
qgroup_rescan_init(struct btrfs_fs_info *fs_info, u64 progress_objectid,
		   int init_flags);
static void qgroup_rescan_zero_tracking(struct btrfs_fs_info *fs_info);

/* must be called with qgroup_ioctl_lock held */
static struct btrfs_qgroup *find_qgroup_rb(struct btrfs_fs_info *fs_info,
					   u64 qgroupid)
{
	struct rb_node *n = fs_info->qgroup_tree.rb_node;
	struct btrfs_qgroup *qgroup;

	while (n) {
		qgroup = rb_entry(n, struct btrfs_qgroup, node);
		if (qgroup->qgroupid < qgroupid)
			n = n->rb_left;
		else if (qgroup->qgroupid > qgroupid)
			n = n->rb_right;
		else
			return qgroup;
	}
	return NULL;
}

/* must be called with qgroup_lock held */
static struct btrfs_qgroup *add_qgroup_rb(struct btrfs_fs_info *fs_info,
					  u64 qgroupid)
{
	struct rb_node **p = &fs_info->qgroup_tree.rb_node;
	struct rb_node *parent = NULL;
	struct btrfs_qgroup *qgroup;

	while (*p) {
		parent = *p;
		qgroup = rb_entry(parent, struct btrfs_qgroup, node);

		if (qgroup->qgroupid < qgroupid)
			p = &(*p)->rb_left;
		else if (qgroup->qgroupid > qgroupid)
			p = &(*p)->rb_right;
		else
			return qgroup;
	}

	qgroup = kzalloc(sizeof(*qgroup), GFP_ATOMIC);
	if (!qgroup)
		return ERR_PTR(-ENOMEM);

	qgroup->qgroupid = qgroupid;
	INIT_LIST_HEAD(&qgroup->groups);
	INIT_LIST_HEAD(&qgroup->members);
	INIT_LIST_HEAD(&qgroup->dirty);

	rb_link_node(&qgroup->node, parent, p);
	rb_insert_color(&qgroup->node, &fs_info->qgroup_tree);

	return qgroup;
}

static void __del_qgroup_rb(struct btrfs_qgroup *qgroup)
{
	struct btrfs_qgroup_list *list;

	list_del(&qgroup->dirty);
	while (!list_empty(&qgroup->groups)) {
		list = list_first_entry(&qgroup->groups,
					struct btrfs_qgroup_list, next_group);
		list_del(&list->next_group);
		list_del(&list->next_member);
		kfree(list);
	}

	while (!list_empty(&qgroup->members)) {
		list = list_first_entry(&qgroup->members,
					struct btrfs_qgroup_list, next_member);
		list_del(&list->next_group);
		list_del(&list->next_member);
		kfree(list);
	}
	kfree(qgroup);
}

/* must be called with qgroup_lock held */
static int del_qgroup_rb(struct btrfs_fs_info *fs_info, u64 qgroupid)
{
	struct btrfs_qgroup *qgroup = find_qgroup_rb(fs_info, qgroupid);

	if (!qgroup)
		return -ENOENT;

	rb_erase(&qgroup->node, &fs_info->qgroup_tree);
	__del_qgroup_rb(qgroup);
	return 0;
}

/* must be called with qgroup_lock held */
static int add_relation_rb(struct btrfs_fs_info *fs_info,
			   u64 memberid, u64 parentid)
{
	struct btrfs_qgroup *member;
	struct btrfs_qgroup *parent;
	struct btrfs_qgroup_list *list;

	member = find_qgroup_rb(fs_info, memberid);
	parent = find_qgroup_rb(fs_info, parentid);
	if (!member || !parent)
		return -ENOENT;

	list = kzalloc(sizeof(*list), GFP_ATOMIC);
	if (!list)
		return -ENOMEM;

	list->group = parent;
	list->member = member;
	list_add_tail(&list->next_group, &member->groups);
	list_add_tail(&list->next_member, &parent->members);

	return 0;
}

/* must be called with qgroup_lock held */
static int del_relation_rb(struct btrfs_fs_info *fs_info,
			   u64 memberid, u64 parentid)
{
	struct btrfs_qgroup *member;
	struct btrfs_qgroup *parent;
	struct btrfs_qgroup_list *list;

	member = find_qgroup_rb(fs_info, memberid);
	parent = find_qgroup_rb(fs_info, parentid);
	if (!member || !parent)
		return -ENOENT;

	list_for_each_entry(list, &member->groups, next_group) {
		if (list->group == parent) {
			list_del(&list->next_group);
			list_del(&list->next_member);
			kfree(list);
			return 0;
		}
	}
	return -ENOENT;
}

#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
int btrfs_verify_qgroup_counts(struct btrfs_fs_info *fs_info, u64 qgroupid,
			       u64 rfer, u64 excl)
{
	struct btrfs_qgroup *qgroup;

	qgroup = find_qgroup_rb(fs_info, qgroupid);
	if (!qgroup)
		return -EINVAL;
	if (qgroup->rfer != rfer || qgroup->excl != excl)
		return -EINVAL;
	return 0;
}
#endif

/*
 * The full config is read in one go, only called from open_ctree()
 * It doesn't use any locking, as at this point we're still single-threaded
 */
int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info)
{
	struct btrfs_key key;
	struct btrfs_key found_key;
	struct btrfs_root *quota_root = fs_info->quota_root;
	struct btrfs_path *path = NULL;
	struct extent_buffer *l;
	int slot;
	int ret = 0;
	u64 flags = 0;
	u64 rescan_progress = 0;

	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
		return 0;

	fs_info->qgroup_ulist = ulist_alloc(GFP_KERNEL);
	if (!fs_info->qgroup_ulist) {
		ret = -ENOMEM;
		goto out;
	}

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

	/* default this to quota off, in case no status key is found */
	fs_info->qgroup_flags = 0;

	/*
	 * pass 1: read status, all qgroup infos and limits
	 */
	key.objectid = 0;
	key.type = 0;
	key.offset = 0;
	ret = btrfs_search_slot_for_read(quota_root, &key, path, 1, 1);
	if (ret)
		goto out;

	while (1) {
		struct btrfs_qgroup *qgroup;

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

		if (found_key.type == BTRFS_QGROUP_STATUS_KEY) {
			struct btrfs_qgroup_status_item *ptr;

			ptr = btrfs_item_ptr(l, slot,
					     struct btrfs_qgroup_status_item);

			if (btrfs_qgroup_status_version(l, ptr) !=
			    BTRFS_QGROUP_STATUS_VERSION) {
				btrfs_err(fs_info,
				 "old qgroup version, quota disabled");
				goto out;
			}
			if (btrfs_qgroup_status_generation(l, ptr) !=
			    fs_info->generation) {
				flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
				btrfs_err(fs_info,
					"qgroup generation mismatch, marked as inconsistent");
			}
			fs_info->qgroup_flags = btrfs_qgroup_status_flags(l,
									  ptr);
			rescan_progress = btrfs_qgroup_status_rescan(l, ptr);
			goto next1;
		}

		if (found_key.type != BTRFS_QGROUP_INFO_KEY &&
		    found_key.type != BTRFS_QGROUP_LIMIT_KEY)
			goto next1;

		qgroup = find_qgroup_rb(fs_info, found_key.offset);
		if ((qgroup && found_key.type == BTRFS_QGROUP_INFO_KEY) ||
		    (!qgroup && found_key.type == BTRFS_QGROUP_LIMIT_KEY)) {
			btrfs_err(fs_info, "inconsistent qgroup config");
			flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
		}
		if (!qgroup) {
			qgroup = add_qgroup_rb(fs_info, found_key.offset);
			if (IS_ERR(qgroup)) {
				ret = PTR_ERR(qgroup);
				goto out;
			}
		}
		switch (found_key.type) {
		case BTRFS_QGROUP_INFO_KEY: {
			struct btrfs_qgroup_info_item *ptr;

			ptr = btrfs_item_ptr(l, slot,
					     struct btrfs_qgroup_info_item);
			qgroup->rfer = btrfs_qgroup_info_rfer(l, ptr);
			qgroup->rfer_cmpr = btrfs_qgroup_info_rfer_cmpr(l, ptr);
			qgroup->excl = btrfs_qgroup_info_excl(l, ptr);
			qgroup->excl_cmpr = btrfs_qgroup_info_excl_cmpr(l, ptr);
			/* generation currently unused */
			break;
		}
		case BTRFS_QGROUP_LIMIT_KEY: {
			struct btrfs_qgroup_limit_item *ptr;

			ptr = btrfs_item_ptr(l, slot,
					     struct btrfs_qgroup_limit_item);
			qgroup->lim_flags = btrfs_qgroup_limit_flags(l, ptr);
			qgroup->max_rfer = btrfs_qgroup_limit_max_rfer(l, ptr);
			qgroup->max_excl = btrfs_qgroup_limit_max_excl(l, ptr);
			qgroup->rsv_rfer = btrfs_qgroup_limit_rsv_rfer(l, ptr);
			qgroup->rsv_excl = btrfs_qgroup_limit_rsv_excl(l, ptr);
			break;
		}
		}
next1:
		ret = btrfs_next_item(quota_root, path);
		if (ret < 0)
			goto out;
		if (ret)
			break;
	}
	btrfs_release_path(path);

	/*
	 * pass 2: read all qgroup relations
	 */
	key.objectid = 0;
	key.type = BTRFS_QGROUP_RELATION_KEY;
	key.offset = 0;
	ret = btrfs_search_slot_for_read(quota_root, &key, path, 1, 0);
	if (ret)
		goto out;
	while (1) {
		slot = path->slots[0];
		l = path->nodes[0];
		btrfs_item_key_to_cpu(l, &found_key, slot);

		if (found_key.type != BTRFS_QGROUP_RELATION_KEY)
			goto next2;

		if (found_key.objectid > found_key.offset) {
			/* parent <- member, not needed to build config */
			/* FIXME should we omit the key completely? */
			goto next2;
		}

		ret = add_relation_rb(fs_info, found_key.objectid,
				      found_key.offset);
		if (ret == -ENOENT) {
			btrfs_warn(fs_info,
				"orphan qgroup relation 0x%llx->0x%llx",
				found_key.objectid, found_key.offset);
			ret = 0;	/* ignore the error */
		}
		if (ret)
			goto out;
next2:
		ret = btrfs_next_item(quota_root, path);
		if (ret < 0)
			goto out;
		if (ret)
			break;
	}
out:
	fs_info->qgroup_flags |= flags;
	if (!(fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_ON))
		clear_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
	else if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN &&
		 ret >= 0)
		ret = qgroup_rescan_init(fs_info, rescan_progress, 0);
	btrfs_free_path(path);

	if (ret < 0) {
		ulist_free(fs_info->qgroup_ulist);
		fs_info->qgroup_ulist = NULL;
		fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_RESCAN;
	}

	return ret < 0 ? ret : 0;
}

/*
 * This is called from close_ctree() or open_ctree() or btrfs_quota_disable(),
 * first two are in single-threaded paths.And for the third one, we have set
 * quota_root to be null with qgroup_lock held before, so it is safe to clean
 * up the in-memory structures without qgroup_lock held.
 */
void btrfs_free_qgroup_config(struct btrfs_fs_info *fs_info)
{
	struct rb_node *n;
	struct btrfs_qgroup *qgroup;

	while ((n = rb_first(&fs_info->qgroup_tree))) {
		qgroup = rb_entry(n, struct btrfs_qgroup, node);
		rb_erase(n, &fs_info->qgroup_tree);
		__del_qgroup_rb(qgroup);
	}
	/*
	 * We call btrfs_free_qgroup_config() when unmounting
	 * filesystem and disabling quota, so we set qgroup_ulist
	 * to be null here to avoid double free.
	 */
	ulist_free(fs_info->qgroup_ulist);
	fs_info->qgroup_ulist = NULL;
}

static int add_qgroup_relation_item(struct btrfs_trans_handle *trans, u64 src,
				    u64 dst)
{
	int ret;
	struct btrfs_root *quota_root = trans->fs_info->quota_root;
	struct btrfs_path *path;
	struct btrfs_key key;

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

	key.objectid = src;
	key.type = BTRFS_QGROUP_RELATION_KEY;
	key.offset = dst;

	ret = btrfs_insert_empty_item(trans, quota_root, path, &key, 0);

	btrfs_mark_buffer_dirty(path->nodes[0]);

	btrfs_free_path(path);
	return ret;
}

static int del_qgroup_relation_item(struct btrfs_trans_handle *trans, u64 src,
				    u64 dst)
{
	int ret;
	struct btrfs_root *quota_root = trans->fs_info->quota_root;
	struct btrfs_path *path;
	struct btrfs_key key;

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

	key.objectid = src;
	key.type = BTRFS_QGROUP_RELATION_KEY;
	key.offset = dst;

	ret = btrfs_search_slot(trans, quota_root, &key, path, -1, 1);
	if (ret < 0)
		goto out;

	if (ret > 0) {
		ret = -ENOENT;
		goto out;
	}

	ret = btrfs_del_item(trans, quota_root, path);
out:
	btrfs_free_path(path);
	return ret;
}

static int add_qgroup_item(struct btrfs_trans_handle *trans,
			   struct btrfs_root *quota_root, u64 qgroupid)
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_qgroup_info_item *qgroup_info;
	struct btrfs_qgroup_limit_item *qgroup_limit;
	struct extent_buffer *leaf;
	struct btrfs_key key;

	if (btrfs_is_testing(quota_root->fs_info))
		return 0;

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

	key.objectid = 0;
	key.type = BTRFS_QGROUP_INFO_KEY;
	key.offset = qgroupid;

	/*
	 * Avoid a transaction abort by catching -EEXIST here. In that
	 * case, we proceed by re-initializing the existing structure
	 * on disk.
	 */

	ret = btrfs_insert_empty_item(trans, quota_root, path, &key,
				      sizeof(*qgroup_info));
	if (ret && ret != -EEXIST)
		goto out;

	leaf = path->nodes[0];
	qgroup_info = btrfs_item_ptr(leaf, path->slots[0],
				 struct btrfs_qgroup_info_item);
	btrfs_set_qgroup_info_generation(leaf, qgroup_info, trans->transid);
	btrfs_set_qgroup_info_rfer(leaf, qgroup_info, 0);
	btrfs_set_qgroup_info_rfer_cmpr(leaf, qgroup_info, 0);
	btrfs_set_qgroup_info_excl(leaf, qgroup_info, 0);
	btrfs_set_qgroup_info_excl_cmpr(leaf, qgroup_info, 0);

	btrfs_mark_buffer_dirty(leaf);

	btrfs_release_path(path);

	key.type = BTRFS_QGROUP_LIMIT_KEY;
	ret = btrfs_insert_empty_item(trans, quota_root, path, &key,
				      sizeof(*qgroup_limit));
	if (ret && ret != -EEXIST)
		goto out;

	leaf = path->nodes[0];
	qgroup_limit = btrfs_item_ptr(leaf, path->slots[0],
				  struct btrfs_qgroup_limit_item);
	btrfs_set_qgroup_limit_flags(leaf, qgroup_limit, 0);
	btrfs_set_qgroup_limit_max_rfer(leaf, qgroup_limit, 0);
	btrfs_set_qgroup_limit_max_excl(leaf, qgroup_limit, 0);
	btrfs_set_qgroup_limit_rsv_rfer(leaf, qgroup_limit, 0);
	btrfs_set_qgroup_limit_rsv_excl(leaf, qgroup_limit, 0);

	btrfs_mark_buffer_dirty(leaf);

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

static int del_qgroup_item(struct btrfs_trans_handle *trans, u64 qgroupid)
{
	int ret;
	struct btrfs_root *quota_root = trans->fs_info->quota_root;
	struct btrfs_path *path;
	struct btrfs_key key;

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

	key.objectid = 0;
	key.type = BTRFS_QGROUP_INFO_KEY;
	key.offset = qgroupid;
	ret = btrfs_search_slot(trans, quota_root, &key, path, -1, 1);
	if (ret < 0)
		goto out;

	if (ret > 0) {
		ret = -ENOENT;
		goto out;
	}

	ret = btrfs_del_item(trans, quota_root, path);
	if (ret)
		goto out;

	btrfs_release_path(path);

	key.type = BTRFS_QGROUP_LIMIT_KEY;
	ret = btrfs_search_slot(trans, quota_root, &key, path, -1, 1);
	if (ret < 0)
		goto out;

	if (ret > 0) {
		ret = -ENOENT;
		goto out;
	}

	ret = btrfs_del_item(trans, quota_root, path);

out:
	btrfs_free_path(path);
	return ret;
}

static int update_qgroup_limit_item(struct btrfs_trans_handle *trans,
				    struct btrfs_qgroup *qgroup)
{
	struct btrfs_root *quota_root = trans->fs_info->quota_root;
	struct btrfs_path *path;
	struct btrfs_key key;
	struct extent_buffer *l;
	struct btrfs_qgroup_limit_item *qgroup_limit;
	int ret;
	int slot;

	key.objectid = 0;
	key.type = BTRFS_QGROUP_LIMIT_KEY;
	key.offset = qgroup->qgroupid;

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

	ret = btrfs_search_slot(trans, quota_root, &key, path, 0, 1);
	if (ret > 0)
		ret = -ENOENT;

	if (ret)
		goto out;

	l = path->nodes[0];
	slot = path->slots[0];
	qgroup_limit = btrfs_item_ptr(l, slot, struct btrfs_qgroup_limit_item);
	btrfs_set_qgroup_limit_flags(l, qgroup_limit, qgroup->lim_flags);
	btrfs_set_qgroup_limit_max_rfer(l, qgroup_limit, qgroup->max_rfer);
	btrfs_set_qgroup_limit_max_excl(l, qgroup_limit, qgroup->max_excl);
	btrfs_set_qgroup_limit_rsv_rfer(l, qgroup_limit, qgroup->rsv_rfer);
	btrfs_set_qgroup_limit_rsv_excl(l, qgroup_limit, qgroup->rsv_excl);

	btrfs_mark_buffer_dirty(l);

out:
	btrfs_free_path(path);
	return ret;
}

static int update_qgroup_info_item(struct btrfs_trans_handle *trans,
				   struct btrfs_qgroup *qgroup)
{
	struct btrfs_fs_info *fs_info = trans->fs_info;
	struct btrfs_root *quota_root = fs_info->quota_root;
	struct btrfs_path *path;
	struct btrfs_key key;
	struct extent_buffer *l;
	struct btrfs_qgroup_info_item *qgroup_info;
	int ret;
	int slot;

	if (btrfs_is_testing(fs_info))
		return 0;

	key.objectid = 0;
	key.type = BTRFS_QGROUP_INFO_KEY;
	key.offset = qgroup->qgroupid;

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

	ret = btrfs_search_slot(trans, quota_root, &key, path, 0, 1);
	if (ret > 0)
		ret = -ENOENT;

	if (ret)
		goto out;

	l = path->nodes[0];
	slot = path->slots[0];
	qgroup_info = btrfs_item_ptr(l, slot, struct btrfs_qgroup_info_item);
	btrfs_set_qgroup_info_generation(l, qgroup_info, trans->transid);
	btrfs_set_qgroup_info_rfer(l, qgroup_info, qgroup->rfer);
	btrfs_set_qgroup_info_rfer_cmpr(l, qgroup_info, qgroup->rfer_cmpr);
	btrfs_set_qgroup_info_excl(l, qgroup_info, qgroup->excl);
	btrfs_set_qgroup_info_excl_cmpr(l, qgroup_info, qgroup->excl_cmpr);

	btrfs_mark_buffer_dirty(l);

out:
	btrfs_free_path(path);
	return ret;
}

static int update_qgroup_status_item(struct btrfs_trans_handle *trans)
{
	struct btrfs_fs_info *fs_info = trans->fs_info;
	struct btrfs_root *quota_root = fs_info->quota_root;
	struct btrfs_path *path;
	struct btrfs_key key;
	struct extent_buffer *l;
	struct btrfs_qgroup_status_item *ptr;
	int ret;
	int slot;

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

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

	ret = btrfs_search_slot(trans, quota_root, &key, path, 0, 1);
	if (ret > 0)
		ret = -ENOENT;

	if (ret)
		goto out;

	l = path->nodes[0];
	slot = path->slots[0];
	ptr = btrfs_item_ptr(l, slot, struct btrfs_qgroup_status_item);
	btrfs_set_qgroup_status_flags(l, ptr, fs_info->qgroup_flags);
	btrfs_set_qgroup_status_generation(l, ptr, trans->transid);
	btrfs_set_qgroup_status_rescan(l, ptr,
				fs_info->qgroup_rescan_progress.objectid);

	btrfs_mark_buffer_dirty(l);

out:
	btrfs_free_path(path);
	return ret;
}

/*
 * called with qgroup_lock held
 */
static int btrfs_clean_quota_tree(struct btrfs_trans_handle *trans,
				  struct btrfs_root *root)
{
	struct btrfs_path *path;
	struct btrfs_key key;
	struct extent_buffer *leaf = NULL;
	int ret;
	int nr = 0;

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

	path->leave_spinning = 1;

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

	while (1) {
		ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
		if (ret < 0)
			goto out;
		leaf = path->nodes[0];
		nr = btrfs_header_nritems(leaf);
		if (!nr)
			break;
		/*
		 * delete the leaf one by one
		 * since the whole tree is going
		 * to be deleted.
		 */
		path->slots[0] = 0;
		ret = btrfs_del_items(trans, root, path, 0, nr);
		if (ret)
			goto out;

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

int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
{
	struct btrfs_root *quota_root;
	struct btrfs_root *tree_root = fs_info->tree_root;
	struct btrfs_path *path = NULL;
	struct btrfs_qgroup_status_item *ptr;
	struct extent_buffer *leaf;
	struct btrfs_key key;
	struct btrfs_key found_key;
	struct btrfs_qgroup *qgroup = NULL;
	struct btrfs_trans_handle *trans = NULL;
	int ret = 0;
	int slot;

	mutex_lock(&fs_info->qgroup_ioctl_lock);
	if (fs_info->quota_root)
		goto out;

	fs_info->qgroup_ulist = ulist_alloc(GFP_KERNEL);
	if (!fs_info->qgroup_ulist) {
		ret = -ENOMEM;
		goto out;
	}

	/*
	 * 1 for quota root item
	 * 1 for BTRFS_QGROUP_STATUS item
	 *
	 * Yet we also need 2*n items for a QGROUP_INFO/QGROUP_LIMIT items
	 * per subvolume. However those are not currently reserved since it
	 * would be a lot of overkill.
	 */
	trans = btrfs_start_transaction(tree_root, 2);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		trans = NULL;
		goto out;
	}

	/*
	 * initially create the quota tree
	 */
	quota_root = btrfs_create_tree(trans, BTRFS_QUOTA_TREE_OBJECTID);
	if (IS_ERR(quota_root)) {
		ret =  PTR_ERR(quota_root);
		btrfs_abort_transaction(trans, ret);
		goto out;
	}

	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		btrfs_abort_transaction(trans, ret);
		goto out_free_root;
	}

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

	ret = btrfs_insert_empty_item(trans, quota_root, path, &key,
				      sizeof(*ptr));
	if (ret) {
		btrfs_abort_transaction(trans, ret);
		goto out_free_path;
	}

	leaf = path->nodes[0];
	ptr = btrfs_item_ptr(leaf, path->slots[0],
				 struct btrfs_qgroup_status_item);
	btrfs_set_qgroup_status_generation(leaf, ptr, trans->transid);
	btrfs_set_qgroup_status_version(leaf, ptr, BTRFS_QGROUP_STATUS_VERSION);
	fs_info->qgroup_flags = BTRFS_QGROUP_STATUS_FLAG_ON |
				BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
	btrfs_set_qgroup_status_flags(leaf, ptr, fs_info->qgroup_flags);
	btrfs_set_qgroup_status_rescan(leaf, ptr, 0);

	btrfs_mark_buffer_dirty(leaf);

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

	btrfs_release_path(path);
	ret = btrfs_search_slot_for_read(tree_root, &key, path, 1, 0);
	if (ret > 0)
		goto out_add_root;
	if (ret < 0) {
		btrfs_abort_transaction(trans, ret);
		goto out_free_path;
	}

	while (1) {
		slot = path->slots[0];
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &found_key, slot);

		if (found_key.type == BTRFS_ROOT_REF_KEY) {
			ret = add_qgroup_item(trans, quota_root,
					      found_key.offset);
			if (ret) {
				btrfs_abort_transaction(trans, ret);
				goto out_free_path;
			}

			qgroup = add_qgroup_rb(fs_info, found_key.offset);
			if (IS_ERR(qgroup)) {
				ret = PTR_ERR(qgroup);
				btrfs_abort_transaction(trans, ret);
				goto out_free_path;
			}
		}
		ret = btrfs_next_item(tree_root, path);
		if (ret < 0) {
			btrfs_abort_transaction(trans, ret);
			goto out_free_path;
		}
		if (ret)
			break;
	}

out_add_root:
	btrfs_release_path(path);
	ret = add_qgroup_item(trans, quota_root, BTRFS_FS_TREE_OBJECTID);
	if (ret) {
		btrfs_abort_transaction(trans, ret);
		goto out_free_path;
	}

	qgroup = add_qgroup_rb(fs_info, BTRFS_FS_TREE_OBJECTID);
	if (IS_ERR(qgroup)) {
		ret = PTR_ERR(qgroup);
		btrfs_abort_transaction(trans, ret);
		goto out_free_path;
	}

	ret = btrfs_commit_transaction(trans);
	trans = NULL;
	if (ret)
		goto out_free_path;

	/*
	 * Set quota enabled flag after committing the transaction, to avoid
	 * deadlocks on fs_info->qgroup_ioctl_lock with concurrent snapshot
	 * creation.
	 */
	spin_lock(&fs_info->qgroup_lock);
	fs_info->quota_root = quota_root;
	set_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
	spin_unlock(&fs_info->qgroup_lock);

	ret = qgroup_rescan_init(fs_info, 0, 1);
	if (!ret) {
	        qgroup_rescan_zero_tracking(fs_info);
	        btrfs_queue_work(fs_info->qgroup_rescan_workers,
	                         &fs_info->qgroup_rescan_work);
	}

out_free_path:
	btrfs_free_path(path);
out_free_root:
	if (ret) {
		free_extent_buffer(quota_root->node);
		free_extent_buffer(quota_root->commit_root);
		kfree(quota_root);
	}
out:
	if (ret) {
		ulist_free(fs_info->qgroup_ulist);
		fs_info->qgroup_ulist = NULL;
		if (trans)
			btrfs_end_transaction(trans);
	}
	mutex_unlock(&fs_info->qgroup_ioctl_lock);
	return ret;
}

int btrfs_quota_disable(struct btrfs_fs_info *fs_info)
{
	struct btrfs_root *quota_root;
	struct btrfs_trans_handle *trans = NULL;
	int ret = 0;

	mutex_lock(&fs_info->qgroup_ioctl_lock);
	if (!fs_info->quota_root)
		goto out;

	/*
	 * 1 For the root item
	 *
	 * We should also reserve enough items for the quota tree deletion in
	 * btrfs_clean_quota_tree but this is not done.
	 */
	trans = btrfs_start_transaction(fs_info->tree_root, 1);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out;
	}

	clear_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
	btrfs_qgroup_wait_for_completion(fs_info, false);
	spin_lock(&fs_info->qgroup_lock);
	quota_root = fs_info->quota_root;
	fs_info->quota_root = NULL;
	fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_ON;
	spin_unlock(&fs_info->qgroup_lock);

	btrfs_free_qgroup_config(fs_info);

	ret = btrfs_clean_quota_tree(trans, quota_root);
	if (ret) {
		btrfs_abort_transaction(trans, ret);
		goto end_trans;
	}

	ret = btrfs_del_root(trans, &quota_root->root_key);
	if (ret) {
		btrfs_abort_transaction(trans, ret);
		goto end_trans;
	}

	list_del(&quota_root->dirty_list);

	btrfs_tree_lock(quota_root->node);
	btrfs_clean_tree_block(quota_root->node);
	btrfs_tree_unlock(quota_root->node);
	btrfs_free_tree_block(trans, quota_root, quota_root->node, 0, 1);

	free_extent_buffer(quota_root->node);
	free_extent_buffer(quota_root->commit_root);
	kfree(quota_root);

end_trans:
	ret = btrfs_end_transaction(trans);
out:
	mutex_unlock(&fs_info->qgroup_ioctl_lock);
	return ret;
}

static void qgroup_dirty(struct btrfs_fs_info *fs_info,
			 struct btrfs_qgroup *qgroup)
{
	if (list_empty(&qgroup->dirty))
		list_add(&qgroup->dirty, &fs_info->dirty_qgroups);
}

/*
 * The easy accounting, we're updating qgroup relationship whose child qgroup
 * only has exclusive extents.
 *
 * In this case, all exclusive extents will also be exclusive for parent, so
 * excl/rfer just get added/removed.
 *
 * So is qgroup reservation space, which should also be added/removed to
 * parent.
 * Or when child tries to release reservation space, parent will underflow its
 * reservation (for relationship adding case).
 *
 * Caller should hold fs_info->qgroup_lock.
 */
static int __qgroup_excl_accounting(struct btrfs_fs_info *fs_info,
				    struct ulist *tmp, u64 ref_root,
				    struct btrfs_qgroup *src, int sign)
{
	struct btrfs_qgroup *qgroup;
	struct btrfs_qgroup_list *glist;
	struct ulist_node *unode;
	struct ulist_iterator uiter;
	u64 num_bytes = src->excl;
	int ret = 0;

	qgroup = find_qgroup_rb(fs_info, ref_root);
	if (!qgroup)
		goto out;

	qgroup->rfer += sign * num_bytes;
	qgroup->rfer_cmpr += sign * num_bytes;

	WARN_ON(sign < 0 && qgroup->excl < num_bytes);
	qgroup->excl += sign * num_bytes;
	qgroup->excl_cmpr += sign * num_bytes;

	if (sign > 0)
		qgroup_rsv_add_by_qgroup(fs_info, qgroup, src);
	else
		qgroup_rsv_release_by_qgroup(fs_info, qgroup, src);

	qgroup_dirty(fs_info, qgroup);

	/* Get all of the parent groups that contain this qgroup */
	list_for_each_entry(glist, &qgroup->groups, next_group) {
		ret = ulist_add(tmp, glist->group->qgroupid,
				qgroup_to_aux(glist->group), GFP_ATOMIC);
		if (ret < 0)
			goto out;
	}

	/* Iterate all of the parents and adjust their reference counts */
	ULIST_ITER_INIT(&uiter);
	while ((unode = ulist_next(tmp, &uiter))) {
		qgroup = unode_aux_to_qgroup(unode);
		qgroup->rfer += sign * num_bytes;
		qgroup->rfer_cmpr += sign * num_bytes;
		WARN_ON(sign < 0 && qgroup->excl < num_bytes);
		qgroup->excl += sign * num_bytes;
		if (sign > 0)
			qgroup_rsv_add_by_qgroup(fs_info, qgroup, src);
		else
			qgroup_rsv_release_by_qgroup(fs_info, qgroup, src);
		qgroup->excl_cmpr += sign * num_bytes;
		qgroup_dirty(fs_info, qgroup);

		/* Add any parents of the parents */
		list_for_each_entry(glist, &qgroup->groups, next_group) {
			ret = ulist_add(tmp, glist->group->qgroupid,
					qgroup_to_aux(glist->group), GFP_ATOMIC);
			if (ret < 0)
				goto out;
		}
	}
	ret = 0;
out:
	return ret;
}


/*
 * Quick path for updating qgroup with only excl refs.
 *
 * In that case, just update all parent will be enough.
 * Or we needs to do a full rescan.
 * Caller should also hold fs_info->qgroup_lock.
 *
 * Return 0 for quick update, return >0 for need to full rescan
 * and mark INCONSISTENT flag.
 * Return < 0 for other error.
 */
static int quick_update_accounting(struct btrfs_fs_info *fs_info,
				   struct ulist *tmp, u64 src, u64 dst,
				   int sign)
{
	struct btrfs_qgroup *qgroup;
	int ret = 1;
	int err = 0;

	qgroup = find_qgroup_rb(fs_info, src);
	if (!qgroup)
		goto out;
	if (qgroup->excl == qgroup->rfer) {
		ret = 0;
		err = __qgroup_excl_accounting(fs_info, tmp, dst,
					       qgroup, sign);
		if (err < 0) {
			ret = err;
			goto out;
		}
	}
out:
	if (ret)
		fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
	return ret;
}

int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
			      u64 dst)
{
	struct btrfs_fs_info *fs_info = trans->fs_info;
	struct btrfs_root *quota_root;
	struct btrfs_qgroup *parent;
	struct btrfs_qgroup *member;
	struct btrfs_qgroup_list *list;
	struct ulist *tmp;
	int ret = 0;

	/* Check the level of src and dst first */
	if (btrfs_qgroup_level(src) >= btrfs_qgroup_level(dst))
		return -EINVAL;

	tmp = ulist_alloc(GFP_KERNEL);
	if (!tmp)
		return -ENOMEM;

	mutex_lock(&fs_info->qgroup_ioctl_lock);
	quota_root = fs_info->quota_root;
	if (!quota_root) {
		ret = -EINVAL;
		goto out;
	}
	member = find_qgroup_rb(fs_info, src);
	parent = find_qgroup_rb(fs_info, dst);
	if (!member || !parent) {
		ret = -EINVAL;
		goto out;
	}

	/* check if such qgroup relation exist firstly */
	list_for_each_entry(list, &member->groups, next_group) {
		if (list->group == parent) {
			ret = -EEXIST;
			goto out;
		}
	}

	ret = add_qgroup_relation_item(trans, src, dst);
	if (ret)
		goto out;

	ret = add_qgroup_relation_item(trans, dst, src);
	if (ret) {
		del_qgroup_relation_item(trans, src, dst);
		goto out;
	}

	spin_lock(&fs_info->qgroup_lock);
	ret = add_relation_rb(fs_info, src, dst);
	if (ret < 0) {
		spin_unlock(&fs_info->qgroup_lock);
		goto out;
	}
	ret = quick_update_accounting(fs_info, tmp, src, dst, 1);
	spin_unlock(&fs_info->qgroup_lock);
out:
	mutex_unlock(&fs_info->qgroup_ioctl_lock);
	ulist_free(tmp);
	return ret;
}

static int __del_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
				 u64 dst)
{
	struct btrfs_fs_info *fs_info = trans->fs_info;
	struct btrfs_root *quota_root;
	struct btrfs_qgroup *parent;
	struct btrfs_qgroup *member;
	struct btrfs_qgroup_list *list;
	struct ulist *tmp;
	bool found = false;
	int ret = 0;
	int ret2;

	tmp = ulist_alloc(GFP_KERNEL);
	if (!tmp)
		return -ENOMEM;

	quota_root = fs_info->quota_root;
	if (!quota_root) {
		ret = -EINVAL;
		goto out;
	}

	member = find_qgroup_rb(fs_info, src);
	parent = find_qgroup_rb(fs_info, dst);
	/*
	 * The parent/member pair doesn't exist, then try to delete the dead
	 * relation items only.
	 */
	if (!member || !parent)
		goto delete_item;

	/* check if such qgroup relation exist firstly */
	list_for_each_entry(list, &member->groups, next_group) {
		if (list->group == parent) {
			found = true;
			break;
		}
	}

delete_item:
	ret = del_qgroup_relation_item(trans, src, dst);
	if (ret < 0 && ret != -ENOENT)
		goto out;
	ret2 = del_qgroup_relation_item(trans, dst, src);
	if (ret2 < 0 && ret2 != -ENOENT)
		goto out;

	/* At least one deletion succeeded, return 0 */
	if (!ret || !ret2)
		ret = 0;

	if (found) {
		spin_lock(&fs_info->qgroup_lock);
		del_relation_rb(fs_info, src, dst);
		ret = quick_update_accounting(fs_info, tmp, src, dst, -1);
		spin_unlock(&fs_info->qgroup_lock);
	}
out:
	ulist_free(tmp);
	return ret;
}

int btrfs_del_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
			      u64 dst)
{
	struct btrfs_fs_info *fs_info = trans->fs_info;
	int ret = 0;

	mutex_lock(&fs_info->qgroup_ioctl_lock);
	ret = __del_qgroup_relation(trans, src, dst);
	mutex_unlock(&fs_info->qgroup_ioctl_lock);

	return ret;
}

int btrfs_create_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid)
{
	struct btrfs_fs_info *fs_info = trans->fs_info;
	struct btrfs_root *quota_root;
	struct btrfs_qgroup *qgroup;
	int ret = 0;

	mutex_lock(&fs_info->qgroup_ioctl_lock);
	quota_root = fs_info->quota_root;
	if (!quota_root) {
		ret = -EINVAL;
		goto out;
	}
	qgroup = find_qgroup_rb(fs_info, qgroupid);
	if (qgroup) {
		ret = -EEXIST;
		goto out;
	}

	ret = add_qgroup_item<