[linux.git] / fs / btrfs / qgroup.h

/* SPDX-License-Identifier: GPL-2.0 */
/*
 * Copyright (C) 2014 Facebook.  All rights reserved.
 */

#ifndef BTRFS_QGROUP_H
#define BTRFS_QGROUP_H

#include <linux/spinlock.h>
#include <linux/rbtree.h>
#include "ulist.h"
#include "delayed-ref.h"

/*
 * Btrfs qgroup overview
 *
 * Btrfs qgroup splits into 3 main part:
 * 1) Reserve
 *    Reserve metadata/data space for incoming operations
 *    Affect how qgroup limit works
 *
 * 2) Trace
 *    Tell btrfs qgroup to trace dirty extents.
 *
 *    Dirty extents including:
 *    - Newly allocated extents
 *    - Extents going to be deleted (in this trans)
 *    - Extents whose owner is going to be modified
 *
 *    This is the main part affects whether qgroup numbers will stay
 *    consistent.
 *    Btrfs qgroup can trace clean extents and won't cause any problem,
 *    but it will consume extra CPU time, it should be avoided if possible.
 *
 * 3) Account
 *    Btrfs qgroup will updates its numbers, based on dirty extents traced
 *    in previous step.
 *
 *    Normally at qgroup rescan and transaction commit time.
 */

/*
 * Special performance optimization for balance.
 *
 * For balance, we need to swap subtree of subvolume and reloc trees.
 * In theory, we need to trace all subtree blocks of both subvolume and reloc
 * trees, since their owner has changed during such swap.
 *
 * However since balance has ensured that both subtrees are containing the
 * same contents and have the same tree structures, such swap won't cause
 * qgroup number change.
 *
 * But there is a race window between subtree swap and transaction commit,
 * during that window, if we increase/decrease tree level or merge/split tree
 * blocks, we still need to trace the original subtrees.
 *
 * So for balance, we use a delayed subtree tracing, whose workflow is:
 *
 * 1) Record the subtree root block get swapped.
 *
 *    During subtree swap:
 *    O = Old tree blocks
 *    N = New tree blocks
 *          reloc tree                     subvolume tree X
 *             Root                               Root
 *            /    \                             /    \
 *          NA     OB                          OA      OB
 *        /  |     |  \                      /  |      |  \
 *      NC  ND     OE  OF                   OC  OD     OE  OF
 *
 *   In this case, NA and OA are going to be swapped, record (NA, OA) into
 *   subvolume tree X.
 *
 * 2) After subtree swap.
 *          reloc tree                     subvolume tree X
 *             Root                               Root
 *            /    \                             /    \
 *          OA     OB                          NA      OB
 *        /  |     |  \                      /  |      |  \
 *      OC  OD     OE  OF                   NC  ND     OE  OF
 *
 * 3a) COW happens for OB
 *     If we are going to COW tree block OB, we check OB's bytenr against
 *     tree X's swapped_blocks structure.
 *     If it doesn't fit any, nothing will happen.
 *
 * 3b) COW happens for NA
 *     Check NA's bytenr against tree X's swapped_blocks, and get a hit.
 *     Then we do subtree scan on both subtrees OA and NA.
 *     Resulting 6 tree blocks to be scanned (OA, OC, OD, NA, NC, ND).
 *
 *     Then no matter what we do to subvolume tree X, qgroup numbers will
 *     still be correct.
 *     Then NA's record gets removed from X's swapped_blocks.
 *
 * 4)  Transaction commit
 *     Any record in X's swapped_blocks gets removed, since there is no
 *     modification to the swapped subtrees, no need to trigger heavy qgroup
 *     subtree rescan for them.
 */

/*
 * Record a dirty extent, and info qgroup to update quota on it
 * TODO: Use kmem cache to alloc it.
 */
struct btrfs_qgroup_extent_record {
	struct rb_node node;
	u64 bytenr;
	u64 num_bytes;
	struct ulist *old_roots;
};

struct btrfs_qgroup_swapped_block {
	struct rb_node node;

	int level;
	bool trace_leaf;

	/* bytenr/generation of the tree block in subvolume tree after swap */
	u64 subvol_bytenr;
	u64 subvol_generation;

	/* bytenr/generation of the tree block in reloc tree after swap */
	u64 reloc_bytenr;
	u64 reloc_generation;

	u64 last_snapshot;
	struct btrfs_key first_key;
};

/*
 * Qgroup reservation types:
 *
 * DATA:
 *	space reserved for data
 *
 * META_PERTRANS:
 * 	Space reserved for metadata (per-transaction)
 * 	Due to the fact that qgroup data is only updated at transaction commit
 * 	time, reserved space for metadata must be kept until transaction
 * 	commits.
 * 	Any metadata reserved that are used in btrfs_start_transaction() should
 * 	be of this type.
 *
 * META_PREALLOC:
 *	There are cases where metadata space is reserved before starting
 *	transaction, and then btrfs_join_transaction() to get a trans handle.
 *	Any metadata reserved for such usage should be of this type.
 *	And after join_transaction() part (or all) of such reservation should
 *	be converted into META_PERTRANS.
 */
enum btrfs_qgroup_rsv_type {
	BTRFS_QGROUP_RSV_DATA,
	BTRFS_QGROUP_RSV_META_PERTRANS,
	BTRFS_QGROUP_RSV_META_PREALLOC,
	BTRFS_QGROUP_RSV_LAST,
};

/*
 * Represents how many bytes we have reserved for this qgroup.
 *
 * Each type should have different reservation behavior.
 * E.g, data follows its io_tree flag modification, while
 * *currently* meta is just reserve-and-clear during transaction.
 *
 * TODO: Add new type for reservation which can survive transaction commit.
 * Current metadata reservation behavior is not suitable for such case.
 */
struct btrfs_qgroup_rsv {
	u64 values[BTRFS_QGROUP_RSV_LAST];
};

/*
 * one struct for each qgroup, organized in fs_info->qgroup_tree.
 */
struct btrfs_qgroup {
	u64 qgroupid;

	/*
	 * state
	 */
	u64 rfer;	/* referenced */
	u64 rfer_cmpr;	/* referenced compressed */
	u64 excl;	/* exclusive */
	u64 excl_cmpr;	/* exclusive compressed */

	/*
	 * limits
	 */
	u64 lim_flags;	/* which limits are set */
	u64 max_rfer;
	u64 max_excl;
	u64 rsv_rfer;
	u64 rsv_excl;

	/*
	 * reservation tracking
	 */
	struct btrfs_qgroup_rsv rsv;

	/*
	 * lists
	 */
	struct list_head groups;  /* groups this group is member of */
	struct list_head members; /* groups that are members of this group */
	struct list_head dirty;   /* dirty groups */
	struct rb_node node;	  /* tree of qgroups */

	/*
	 * temp variables for accounting operations
	 * Refer to qgroup_shared_accounting() for details.
	 */
	u64 old_refcnt;
	u64 new_refcnt;
};

/*
 * For qgroup event trace points only
 */
#define QGROUP_RESERVE		(1<<0)
#define QGROUP_RELEASE		(1<<1)
#define QGROUP_FREE		(1<<2)

int btrfs_quota_enable(struct btrfs_fs_info *fs_info);
int btrfs_quota_disable(struct btrfs_fs_info *fs_info);
int btrfs_qgroup_rescan(struct btrfs_fs_info *fs_info);
void btrfs_qgroup_rescan_resume(struct btrfs_fs_info *fs_info);
int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info,
				     bool interruptible);
int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
			      u64 dst);
int btrfs_del_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
			      u64 dst);
int btrfs_create_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid);
int btrfs_remove_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid);
int btrfs_limit_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid,
		       struct btrfs_qgroup_limit *limit);
int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info);
void btrfs_free_qgroup_config(struct btrfs_fs_info *fs_info);
struct btrfs_delayed_extent_op;

/*
 * Inform qgroup to trace one dirty extent, its info is recorded in @record.
 * So qgroup can account it at transaction committing time.
 *
 * No lock version, caller must acquire delayed ref lock and allocated memory,
 * then call btrfs_qgroup_trace_extent_post() after exiting lock context.
 *
 * Return 0 for success insert
 * Return >0 for existing record, caller can free @record safely.
 * Error is not possible
 */
int btrfs_qgroup_trace_extent_nolock(
		struct btrfs_fs_info *fs_info,
		struct btrfs_delayed_ref_root *delayed_refs,
		struct btrfs_qgroup_extent_record *record);

/*
 * Post handler after qgroup_trace_extent_nolock().
 *
 * NOTE: Current qgroup does the expensive backref walk at transaction
 * committing time with TRANS_STATE_COMMIT_DOING, this blocks incoming
 * new transaction.
 * This is designed to allow btrfs_find_all_roots() to get correct new_roots
 * result.
 *
 * However for old_roots there is no need to do backref walk at that time,
 * since we search commit roots to walk backref and result will always be
 * correct.
 *
 * Due to the nature of no lock version, we can't do backref there.
 * So we must call btrfs_qgroup_trace_extent_post() after exiting
 * spinlock context.
 *
 * TODO: If we can fix and prove btrfs_find_all_roots() can get correct result
 * using current root, then we can move all expensive backref walk out of
 * transaction committing, but not now as qgroup accounting will be wrong again.
 */
int btrfs_qgroup_trace_extent_post(struct btrfs_fs_info *fs_info,
				   struct btrfs_qgroup_extent_record *qrecord);

/*
 * Inform qgroup to trace one dirty extent, specified by @bytenr and
 * @num_bytes.
 * So qgroup can account it at commit trans time.
 *
 * Better encapsulated version, with memory allocation and backref walk for
 * commit roots.
 * So this can sleep.
 *
 * Return 0 if the operation is done.
 * Return <0 for error, like memory allocation failure or invalid parameter
 * (NULL trans)
 */
int btrfs_qgroup_trace_extent(struct btrfs_trans_handle *trans, u64 bytenr,
			      u64 num_bytes, gfp_t gfp_flag);

/*
 * Inform qgroup to trace all leaf items of data
 *
 * Return 0 for success
 * Return <0 for error(ENOMEM)
 */
int btrfs_qgroup_trace_leaf_items(struct btrfs_trans_handle *trans,
				  struct extent_buffer *eb);
/*
 * Inform qgroup to trace a whole subtree, including all its child tree
 * blocks and data.
 * The root tree block is specified by @root_eb.
 *
 * Normally used by relocation(tree block swap) and subvolume deletion.
 *
 * Return 0 for success
 * Return <0 for error(ENOMEM or tree search error)
 */
int btrfs_qgroup_trace_subtree(struct btrfs_trans_handle *trans,
			       struct extent_buffer *root_eb,
			       u64 root_gen, int root_level);

int btrfs_qgroup_trace_subtree_swap(struct btrfs_trans_handle *trans,
				struct btrfs_block_group_cache *bg_cache,
				struct extent_buffer *src_parent, int src_slot,
				struct extent_buffer *dst_parent, int dst_slot,
				u64 last_snapshot);
int btrfs_qgroup_account_extent(struct btrfs_trans_handle *trans, u64 bytenr,
				u64 num_bytes, struct ulist *old_roots,
				struct ulist *new_roots);
int btrfs_qgroup_account_extents(struct btrfs_trans_handle *trans);
int btrfs_run_qgroups(struct btrfs_trans_handle *trans);
int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
			 u64 objectid, struct btrfs_qgroup_inherit *inherit);
void btrfs_qgroup_free_refroot(struct btrfs_fs_info *fs_info,
			       u64 ref_root, u64 num_bytes,
			       enum btrfs_qgroup_rsv_type type);
static inline void btrfs_qgroup_free_delayed_ref(struct btrfs_fs_info *fs_info,
						 u64 ref_root, u64 num_bytes)
{
	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
		return;
	trace_btrfs_qgroup_free_delayed_ref(fs_info, ref_root, num_bytes);
	btrfs_qgroup_free_refroot(fs_info, ref_root, num_bytes,
				  BTRFS_QGROUP_RSV_DATA);
}

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

/* New io_tree based accurate qgroup reserve API */
int btrfs_qgroup_reserve_data(struct inode *inode,
			struct extent_changeset **reserved, u64 start, u64 len);
int btrfs_qgroup_release_data(struct inode *inode, u64 start, u64 len);
int btrfs_qgroup_free_data(struct inode *inode,
			struct extent_changeset *reserved, u64 start, u64 len);

int __btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
				enum btrfs_qgroup_rsv_type type, bool enforce);
/* Reserve metadata space for pertrans and prealloc type */
static inline int btrfs_qgroup_reserve_meta_pertrans(struct btrfs_root *root,
				int num_bytes, bool enforce)
{
	return __btrfs_qgroup_reserve_meta(root, num_bytes,
			BTRFS_QGROUP_RSV_META_PERTRANS, enforce);
}
static inline int btrfs_qgroup_reserve_meta_prealloc(struct btrfs_root *root,
				int num_bytes, bool enforce)
{
	return __btrfs_qgroup_reserve_meta(root, num_bytes,
			BTRFS_QGROUP_RSV_META_PREALLOC, enforce);
}

void __btrfs_qgroup_free_meta(struct btrfs_root *root, int num_bytes,
			     enum btrfs_qgroup_rsv_type type);

/* Free per-transaction meta reservation for error handling */
static inline void btrfs_qgroup_free_meta_pertrans(struct btrfs_root *root,
						   int num_bytes)
{
	__btrfs_qgroup_free_meta(root, num_bytes,
			BTRFS_QGROUP_RSV_META_PERTRANS);
}

/* Pre-allocated meta reservation can be freed at need */
static inline void btrfs_qgroup_free_meta_prealloc(struct btrfs_root *root,
						   int num_bytes)
{
	__btrfs_qgroup_free_meta(root, num_bytes,
			BTRFS_QGROUP_RSV_META_PREALLOC);
}

/*
 * Per-transaction meta reservation should be all freed at transaction commit
 * time
 */
void btrfs_qgroup_free_meta_all_pertrans(struct btrfs_root *root);

/*
 * Convert @num_bytes of META_PREALLOCATED reservation to META_PERTRANS.
 *
 * This is called when preallocated meta reservation needs to be used.
 * Normally after btrfs_join_transaction() call.
 */
void btrfs_qgroup_convert_reserved_meta(struct btrfs_root *root, int num_bytes);

void btrfs_qgroup_check_reserved_leak(struct inode *inode);

/* btrfs_qgroup_swapped_blocks related functions */
void btrfs_qgroup_init_swapped_blocks(
	struct btrfs_qgroup_swapped_blocks *swapped_blocks);

void btrfs_qgroup_clean_swapped_blocks(struct btrfs_root *root);
int btrfs_qgroup_add_swapped_blocks(struct btrfs_trans_handle *trans,
		struct btrfs_root *subvol_root,
		struct btrfs_block_group_cache *bg,
		struct extent_buffer *subvol_parent, int subvol_slot,
		struct extent_buffer *reloc_parent, int reloc_slot,
		u64 last_snapshot);
int btrfs_qgroup_trace_subtree_after_cow(struct btrfs_trans_handle *trans,
		struct btrfs_root *root, struct extent_buffer *eb);

#endif
Commit	Line	Data
9888c340	1	/* SPDX-License-Identifier: GPL-2.0 */
fcebe456 JB	2	/*
fcebe456 JB	3	* Copyright (C) 2014 Facebook. All rights reserved.
fcebe456 JB	4	*/
fcebe456 JB	5
9888c340 DS	6	#ifndef BTRFS_QGROUP_H
9888c340 DS	7	#define BTRFS_QGROUP_H
fcebe456	8
370a11b8 QW	9	#include <linux/spinlock.h>
370a11b8 QW	10	#include <linux/rbtree.h>
3368d001 QW	11	#include "ulist.h"
	12	#include "delayed-ref.h"
	13
1d2beaa9 QW	14	/*
	15	* Btrfs qgroup overview
	16	*
	17	* Btrfs qgroup splits into 3 main part:
	18	* 1) Reserve
	19	* Reserve metadata/data space for incoming operations
	20	* Affect how qgroup limit works
	21	*
	22	* 2) Trace
	23	* Tell btrfs qgroup to trace dirty extents.
	24	*
	25	* Dirty extents including:
	26	* - Newly allocated extents
	27	* - Extents going to be deleted (in this trans)
	28	* - Extents whose owner is going to be modified
	29	*
	30	* This is the main part affects whether qgroup numbers will stay
	31	* consistent.
	32	* Btrfs qgroup can trace clean extents and won't cause any problem,
	33	* but it will consume extra CPU time, it should be avoided if possible.
	34	*
	35	* 3) Account
	36	* Btrfs qgroup will updates its numbers, based on dirty extents traced
	37	* in previous step.
	38	*
	39	* Normally at qgroup rescan and transaction commit time.
	40	*/
	41
370a11b8 QW	42	/*
	43	* Special performance optimization for balance.
	44	*
	45	* For balance, we need to swap subtree of subvolume and reloc trees.
	46	* In theory, we need to trace all subtree blocks of both subvolume and reloc
	47	* trees, since their owner has changed during such swap.
	48	*
	49	* However since balance has ensured that both subtrees are containing the
	50	* same contents and have the same tree structures, such swap won't cause
	51	* qgroup number change.
	52	*
	53	* But there is a race window between subtree swap and transaction commit,
	54	* during that window, if we increase/decrease tree level or merge/split tree
	55	* blocks, we still need to trace the original subtrees.
	56	*
	57	* So for balance, we use a delayed subtree tracing, whose workflow is:
	58	*
	59	* 1) Record the subtree root block get swapped.
	60	*
	61	* During subtree swap:
	62	* O = Old tree blocks
	63	* N = New tree blocks
	64	* reloc tree subvolume tree X
	65	* Root Root
	66	* / \ / \
	67	* NA OB OA OB
	68	* / \| \| \ / \| \| \
	69	* NC ND OE OF OC OD OE OF
	70	*
	71	* In this case, NA and OA are going to be swapped, record (NA, OA) into
	72	* subvolume tree X.
	73	*
	74	* 2) After subtree swap.
	75	* reloc tree subvolume tree X
	76	* Root Root
	77	* / \ / \
	78	* OA OB NA OB
	79	* / \| \| \ / \| \| \
	80	* OC OD OE OF NC ND OE OF
	81	*
	82	* 3a) COW happens for OB
	83	* If we are going to COW tree block OB, we check OB's bytenr against
	84	* tree X's swapped_blocks structure.
	85	* If it doesn't fit any, nothing will happen.
	86	*
	87	* 3b) COW happens for NA
	88	* Check NA's bytenr against tree X's swapped_blocks, and get a hit.
	89	* Then we do subtree scan on both subtrees OA and NA.
	90	* Resulting 6 tree blocks to be scanned (OA, OC, OD, NA, NC, ND).
	91	*
	92	* Then no matter what we do to subvolume tree X, qgroup numbers will
	93	* still be correct.
	94	* Then NA's record gets removed from X's swapped_blocks.
	95	*
	96	* 4) Transaction commit
	97	* Any record in X's swapped_blocks gets removed, since there is no
	98	* modification to the swapped subtrees, no need to trigger heavy qgroup
	99	* subtree rescan for them.
	100	*/
	101
3368d001 QW	102	/*
	103	* Record a dirty extent, and info qgroup to update quota on it
	104	* TODO: Use kmem cache to alloc it.
	105	*/
	106	struct btrfs_qgroup_extent_record {
	107	struct rb_node node;
	108	u64 bytenr;
	109	u64 num_bytes;
	110	struct ulist *old_roots;
	111	};
	112
370a11b8 QW	113	struct btrfs_qgroup_swapped_block {
	114	struct rb_node node;
	115
	116	int level;
	117	bool trace_leaf;
	118
	119	/* bytenr/generation of the tree block in subvolume tree after swap */
	120	u64 subvol_bytenr;
	121	u64 subvol_generation;
	122
	123	/* bytenr/generation of the tree block in reloc tree after swap */
	124	u64 reloc_bytenr;
	125	u64 reloc_generation;
	126
	127	u64 last_snapshot;
	128	struct btrfs_key first_key;
	129	};
	130
733e03a0 QW	131	/*
	132	* Qgroup reservation types:
	133	*
	134	* DATA:
	135	* space reserved for data
	136	*
	137	* META_PERTRANS:
	138	* Space reserved for metadata (per-transaction)
	139	* Due to the fact that qgroup data is only updated at transaction commit
	140	* time, reserved space for metadata must be kept until transaction
	141	* commits.
	142	* Any metadata reserved that are used in btrfs_start_transaction() should
	143	* be of this type.
	144	*
	145	* META_PREALLOC:
	146	* There are cases where metadata space is reserved before starting
	147	* transaction, and then btrfs_join_transaction() to get a trans handle.
	148	* Any metadata reserved for such usage should be of this type.
	149	* And after join_transaction() part (or all) of such reservation should
	150	* be converted into META_PERTRANS.
	151	*/
d4e5c920	152	enum btrfs_qgroup_rsv_type {
bbe339cc	153	BTRFS_QGROUP_RSV_DATA,
733e03a0 QW	154	BTRFS_QGROUP_RSV_META_PERTRANS,
733e03a0 QW	155	BTRFS_QGROUP_RSV_META_PREALLOC,
d4e5c920 QW	156	BTRFS_QGROUP_RSV_LAST,
	157	};
	158
	159	/*
	160	* Represents how many bytes we have reserved for this qgroup.
	161	*
	162	* Each type should have different reservation behavior.
	163	* E.g, data follows its io_tree flag modification, while
52042d8e	164	* currently meta is just reserve-and-clear during transaction.
d4e5c920 QW	165	*
d4e5c920 QW	166	* TODO: Add new type for reservation which can survive transaction commit.
52042d8e	167	* Current metadata reservation behavior is not suitable for such case.
d4e5c920 QW	168	*/
	169	struct btrfs_qgroup_rsv {
	170	u64 values[BTRFS_QGROUP_RSV_LAST];
	171	};
	172
3159fe7b QW	173	/*
	174	* one struct for each qgroup, organized in fs_info->qgroup_tree.
	175	*/
	176	struct btrfs_qgroup {
	177	u64 qgroupid;
	178
	179	/*
	180	* state
	181	*/
	182	u64 rfer; /* referenced */
	183	u64 rfer_cmpr; /* referenced compressed */
	184	u64 excl; /* exclusive */
	185	u64 excl_cmpr; /* exclusive compressed */
	186
	187	/*
	188	* limits
	189	*/
	190	u64 lim_flags; /* which limits are set */
	191	u64 max_rfer;
	192	u64 max_excl;
	193	u64 rsv_rfer;
	194	u64 rsv_excl;
	195
	196	/*
	197	* reservation tracking
	198	*/
d4e5c920	199	struct btrfs_qgroup_rsv rsv;
3159fe7b QW	200
	201	/*
	202	* lists
	203	*/
	204	struct list_head groups; /* groups this group is member of */
	205	struct list_head members; /* groups that are members of this group */
	206	struct list_head dirty; /* dirty groups */
	207	struct rb_node node; /* tree of qgroups */
	208
	209	/*
	210	* temp variables for accounting operations
	211	* Refer to qgroup_shared_accounting() for details.
	212	*/
	213	u64 old_refcnt;
	214	u64 new_refcnt;
	215	};
	216
81fb6f77 QW	217	/*
	218	* For qgroup event trace points only
	219	*/
	220	#define QGROUP_RESERVE (1<<0)
	221	#define QGROUP_RELEASE (1<<1)
	222	#define QGROUP_FREE (1<<2)
	223
340f1aa2 NB	224	int btrfs_quota_enable(struct btrfs_fs_info *fs_info);
340f1aa2 NB	225	int btrfs_quota_disable(struct btrfs_fs_info *fs_info);
fcebe456 JB	226	int btrfs_qgroup_rescan(struct btrfs_fs_info *fs_info);
fcebe456 JB	227	void btrfs_qgroup_rescan_resume(struct btrfs_fs_info *fs_info);
d06f23d6 JM	228	int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info,
d06f23d6 JM	229	bool interruptible);
9f8a6ce6 LF	230	int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
9f8a6ce6 LF	231	u64 dst);
39616c27 LF	232	int btrfs_del_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
39616c27 LF	233	u64 dst);
49a05ecd	234	int btrfs_create_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid);
3efbee1d	235	int btrfs_remove_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid);
f0042d5e	236	int btrfs_limit_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid,
fcebe456 JB	237	struct btrfs_qgroup_limit *limit);
	238	int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info);
	239	void btrfs_free_qgroup_config(struct btrfs_fs_info *fs_info);
	240	struct btrfs_delayed_extent_op;
d1b8b94a	241
cb93b52c	242	/*
50b3e040	243	* Inform qgroup to trace one dirty extent, its info is recorded in @record.
fb235dc0	244	* So qgroup can account it at transaction committing time.
cb93b52c	245	*
fb235dc0 QW	246	* No lock version, caller must acquire delayed ref lock and allocated memory,
fb235dc0 QW	247	* then call btrfs_qgroup_trace_extent_post() after exiting lock context.
cb93b52c QW	248	*
	249	* Return 0 for success insert
	250	* Return >0 for existing record, caller can free @record safely.
	251	* Error is not possible
	252	*/
50b3e040	253	int btrfs_qgroup_trace_extent_nolock(
cb93b52c QW	254	struct btrfs_fs_info *fs_info,
	255	struct btrfs_delayed_ref_root *delayed_refs,
	256	struct btrfs_qgroup_extent_record *record);
	257
fb235dc0 QW	258	/*
	259	* Post handler after qgroup_trace_extent_nolock().
	260	*
	261	* NOTE: Current qgroup does the expensive backref walk at transaction
	262	* committing time with TRANS_STATE_COMMIT_DOING, this blocks incoming
	263	* new transaction.
	264	* This is designed to allow btrfs_find_all_roots() to get correct new_roots
	265	* result.
	266	*
	267	* However for old_roots there is no need to do backref walk at that time,
	268	* since we search commit roots to walk backref and result will always be
	269	* correct.
	270	*
	271	* Due to the nature of no lock version, we can't do backref there.
	272	* So we must call btrfs_qgroup_trace_extent_post() after exiting
	273	* spinlock context.
	274	*
	275	* TODO: If we can fix and prove btrfs_find_all_roots() can get correct result
	276	* using current root, then we can move all expensive backref walk out of
	277	* transaction committing, but not now as qgroup accounting will be wrong again.
	278	*/
	279	int btrfs_qgroup_trace_extent_post(struct btrfs_fs_info *fs_info,
	280	struct btrfs_qgroup_extent_record *qrecord);
	281
cb93b52c	282	/*
50b3e040 QW	283	* Inform qgroup to trace one dirty extent, specified by @bytenr and
	284	* @num_bytes.
	285	* So qgroup can account it at commit trans time.
cb93b52c	286	*
fb235dc0 QW	287	* Better encapsulated version, with memory allocation and backref walk for
	288	* commit roots.
	289	* So this can sleep.
cb93b52c QW	290	*
	291	* Return 0 if the operation is done.
	292	* Return <0 for error, like memory allocation failure or invalid parameter
	293	* (NULL trans)
	294	*/
a95f3aaf LF	295	int btrfs_qgroup_trace_extent(struct btrfs_trans_handle *trans, u64 bytenr,
a95f3aaf LF	296	u64 num_bytes, gfp_t gfp_flag);
cb93b52c	297
33d1f05c QW	298	/*
	299	* Inform qgroup to trace all leaf items of data
	300	*
	301	* Return 0 for success
	302	* Return <0 for error(ENOMEM)
	303	*/
	304	int btrfs_qgroup_trace_leaf_items(struct btrfs_trans_handle *trans,
33d1f05c QW	305	struct extent_buffer *eb);
	306	/*
	307	* Inform qgroup to trace a whole subtree, including all its child tree
	308	* blocks and data.
	309	* The root tree block is specified by @root_eb.
	310	*
	311	* Normally used by relocation(tree block swap) and subvolume deletion.
	312	*
	313	* Return 0 for success
	314	* Return <0 for error(ENOMEM or tree search error)
	315	*/
	316	int btrfs_qgroup_trace_subtree(struct btrfs_trans_handle *trans,
33d1f05c QW	317	struct extent_buffer *root_eb,
33d1f05c QW	318	u64 root_gen, int root_level);
5f527822 QW	319
5f527822 QW	320	int btrfs_qgroup_trace_subtree_swap(struct btrfs_trans_handle *trans,
3d0174f7	321	struct btrfs_block_group_cache *bg_cache,
5f527822 QW	322	struct extent_buffer *src_parent, int src_slot,
	323	struct extent_buffer *dst_parent, int dst_slot,
	324	u64 last_snapshot);
8696d760 LF	325	int btrfs_qgroup_account_extent(struct btrfs_trans_handle *trans, u64 bytenr,
	326	u64 num_bytes, struct ulist *old_roots,
	327	struct ulist *new_roots);
460fb20a	328	int btrfs_qgroup_account_extents(struct btrfs_trans_handle *trans);
280f8bd2	329	int btrfs_run_qgroups(struct btrfs_trans_handle *trans);
a9377422 LF	330	int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
a9377422 LF	331	u64 objectid, struct btrfs_qgroup_inherit *inherit);
297d750b	332	void btrfs_qgroup_free_refroot(struct btrfs_fs_info *fs_info,
d4e5c920 QW	333	u64 ref_root, u64 num_bytes,
d4e5c920 QW	334	enum btrfs_qgroup_rsv_type type);
297d750b QW	335	static inline void btrfs_qgroup_free_delayed_ref(struct btrfs_fs_info *fs_info,
	336	u64 ref_root, u64 num_bytes)
	337	{
3628b4ca QW	338	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
3628b4ca QW	339	return;
bc074524	340	trace_btrfs_qgroup_free_delayed_ref(fs_info, ref_root, num_bytes);
d4e5c920 QW	341	btrfs_qgroup_free_refroot(fs_info, ref_root, num_bytes,
d4e5c920 QW	342	BTRFS_QGROUP_RSV_DATA);
297d750b	343	}
fcebe456 JB	344
	345	#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
	346	int btrfs_verify_qgroup_counts(struct btrfs_fs_info *fs_info, u64 qgroupid,
	347	u64 rfer, u64 excl);
	348	#endif
	349
52472553	350	/* New io_tree based accurate qgroup reserve API */
364ecf36 QW	351	int btrfs_qgroup_reserve_data(struct inode *inode,
364ecf36 QW	352	struct extent_changeset **reserved, u64 start, u64 len);
f695fdce	353	int btrfs_qgroup_release_data(struct inode *inode, u64 start, u64 len);
bc42bda2 QW	354	int btrfs_qgroup_free_data(struct inode *inode,
bc42bda2 QW	355	struct extent_changeset *reserved, u64 start, u64 len);
55eeaf05	356
733e03a0 QW	357	int __btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
	358	enum btrfs_qgroup_rsv_type type, bool enforce);
	359	/* Reserve metadata space for pertrans and prealloc type */
	360	static inline int btrfs_qgroup_reserve_meta_pertrans(struct btrfs_root *root,
	361	int num_bytes, bool enforce)
	362	{
	363	return __btrfs_qgroup_reserve_meta(root, num_bytes,
	364	BTRFS_QGROUP_RSV_META_PERTRANS, enforce);
	365	}
	366	static inline int btrfs_qgroup_reserve_meta_prealloc(struct btrfs_root *root,
	367	int num_bytes, bool enforce)
	368	{
	369	return __btrfs_qgroup_reserve_meta(root, num_bytes,
	370	BTRFS_QGROUP_RSV_META_PREALLOC, enforce);
	371	}
	372
	373	void __btrfs_qgroup_free_meta(struct btrfs_root *root, int num_bytes,
	374	enum btrfs_qgroup_rsv_type type);
	375
	376	/* Free per-transaction meta reservation for error handling */
	377	static inline void btrfs_qgroup_free_meta_pertrans(struct btrfs_root *root,
	378	int num_bytes)
	379	{
	380	__btrfs_qgroup_free_meta(root, num_bytes,
	381	BTRFS_QGROUP_RSV_META_PERTRANS);
	382	}
	383
	384	/* Pre-allocated meta reservation can be freed at need */
	385	static inline void btrfs_qgroup_free_meta_prealloc(struct btrfs_root *root,
	386	int num_bytes)
	387	{
	388	__btrfs_qgroup_free_meta(root, num_bytes,
	389	BTRFS_QGROUP_RSV_META_PREALLOC);
	390	}
	391
	392	/*
	393	* Per-transaction meta reservation should be all freed at transaction commit
	394	* time
	395	*/
	396	void btrfs_qgroup_free_meta_all_pertrans(struct btrfs_root *root);
	397
64cfaef6 QW	398	/*
	399	* Convert @num_bytes of META_PREALLOCATED reservation to META_PERTRANS.
	400	*
	401	* This is called when preallocated meta reservation needs to be used.
	402	* Normally after btrfs_join_transaction() call.
	403	*/
	404	void btrfs_qgroup_convert_reserved_meta(struct btrfs_root *root, int num_bytes);
	405
56fa9d07	406	void btrfs_qgroup_check_reserved_leak(struct inode *inode);
9888c340	407
370a11b8 QW	408	/* btrfs_qgroup_swapped_blocks related functions */
	409	void btrfs_qgroup_init_swapped_blocks(
	410	struct btrfs_qgroup_swapped_blocks *swapped_blocks);
	411
	412	void btrfs_qgroup_clean_swapped_blocks(struct btrfs_root *root);
	413	int btrfs_qgroup_add_swapped_blocks(struct btrfs_trans_handle *trans,
	414	struct btrfs_root *subvol_root,
	415	struct btrfs_block_group_cache *bg,
	416	struct extent_buffer *subvol_parent, int subvol_slot,
	417	struct extent_buffer *reloc_parent, int reloc_slot,
	418	u64 last_snapshot);
f616f5cd QW	419	int btrfs_qgroup_trace_subtree_after_cow(struct btrfs_trans_handle *trans,
f616f5cd QW	420	struct btrfs_root root, struct extent_buffer eb);
370a11b8	421
9888c340	422	#endif