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

/*
 * Copyright (C) 2014 Facebook.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

#ifndef __BTRFS_QGROUP__
#define __BTRFS_QGROUP__

#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.
 */

/*
 * 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;
};

/*
 * 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
	 */
	u64 reserved;

	/*
	 * 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_trans_handle *trans,
		       struct btrfs_fs_info *fs_info);
int btrfs_quota_disable(struct btrfs_trans_handle *trans,
			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,
			      struct btrfs_fs_info *fs_info, u64 src, u64 dst);
int btrfs_del_qgroup_relation(struct btrfs_trans_handle *trans,
			      struct btrfs_fs_info *fs_info, u64 src, u64 dst);
int btrfs_create_qgroup(struct btrfs_trans_handle *trans,
			struct btrfs_fs_info *fs_info, u64 qgroupid);
int btrfs_remove_qgroup(struct btrfs_trans_handle *trans,
			      struct btrfs_fs_info *fs_info, u64 qgroupid);
int btrfs_limit_qgroup(struct btrfs_trans_handle *trans,
		       struct btrfs_fs_info *fs_info, 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;
int btrfs_qgroup_prepare_account_extents(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info);
/*
 * 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,
		struct btrfs_fs_info *fs_info, 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 btrfs_fs_info *fs_info,
				  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 btrfs_root *root,
			       struct extent_buffer *root_eb,
			       u64 root_gen, int root_level);
int
btrfs_qgroup_account_extent(struct btrfs_trans_handle *trans,
			    struct btrfs_fs_info *fs_info,
			    u64 bytenr, u64 num_bytes,
			    struct ulist *old_roots, struct ulist *new_roots);
int btrfs_qgroup_account_extents(struct btrfs_trans_handle *trans,
				 struct btrfs_fs_info *fs_info);
int btrfs_run_qgroups(struct btrfs_trans_handle *trans,
		      struct btrfs_fs_info *fs_info);
int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans,
			 struct btrfs_fs_info *fs_info, 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);
static inline void btrfs_qgroup_free_delayed_ref(struct btrfs_fs_info *fs_info,
						 u64 ref_root, u64 num_bytes)
{
	trace_btrfs_qgroup_free_delayed_ref(fs_info, ref_root, num_bytes);
	btrfs_qgroup_free_refroot(fs_info, ref_root, num_bytes);
}

#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, u64 start, u64 len);
int btrfs_qgroup_release_data(struct inode *inode, u64 start, u64 len);
int btrfs_qgroup_free_data(struct inode *inode, u64 start, u64 len);

int btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
			      bool enforce);
void btrfs_qgroup_free_meta_all(struct btrfs_root *root);
void btrfs_qgroup_free_meta(struct btrfs_root *root, int num_bytes);
void btrfs_qgroup_check_reserved_leak(struct inode *inode);
#endif /* __BTRFS_QGROUP__ */
Commit	Line	Data
fcebe456 JB	1	/*
	2	* Copyright (C) 2014 Facebook. All rights reserved.
	3	*
	4	* This program is free software; you can redistribute it and/or
	5	* modify it under the terms of the GNU General Public
	6	* License v2 as published by the Free Software Foundation.
	7	*
	8	* This program is distributed in the hope that it will be useful,
	9	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	10	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	11	* General Public License for more details.
	12	*
	13	* You should have received a copy of the GNU General Public
	14	* License along with this program; if not, write to the
	15	* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
	16	* Boston, MA 021110-1307, USA.
	17	*/
	18
	19	#ifndef __BTRFS_QGROUP__
	20	#define __BTRFS_QGROUP__
	21
3368d001 QW	22	#include "ulist.h"
	23	#include "delayed-ref.h"
	24
1d2beaa9 QW	25	/*
	26	* Btrfs qgroup overview
	27	*
	28	* Btrfs qgroup splits into 3 main part:
	29	* 1) Reserve
	30	* Reserve metadata/data space for incoming operations
	31	* Affect how qgroup limit works
	32	*
	33	* 2) Trace
	34	* Tell btrfs qgroup to trace dirty extents.
	35	*
	36	* Dirty extents including:
	37	* - Newly allocated extents
	38	* - Extents going to be deleted (in this trans)
	39	* - Extents whose owner is going to be modified
	40	*
	41	* This is the main part affects whether qgroup numbers will stay
	42	* consistent.
	43	* Btrfs qgroup can trace clean extents and won't cause any problem,
	44	* but it will consume extra CPU time, it should be avoided if possible.
	45	*
	46	* 3) Account
	47	* Btrfs qgroup will updates its numbers, based on dirty extents traced
	48	* in previous step.
	49	*
	50	* Normally at qgroup rescan and transaction commit time.
	51	*/
	52
3368d001 QW	53	/*
	54	* Record a dirty extent, and info qgroup to update quota on it
	55	* TODO: Use kmem cache to alloc it.
	56	*/
	57	struct btrfs_qgroup_extent_record {
	58	struct rb_node node;
	59	u64 bytenr;
	60	u64 num_bytes;
	61	struct ulist *old_roots;
	62	};
	63
3159fe7b QW	64	/*
	65	* one struct for each qgroup, organized in fs_info->qgroup_tree.
	66	*/
	67	struct btrfs_qgroup {
	68	u64 qgroupid;
	69
	70	/*
	71	* state
	72	*/
	73	u64 rfer; /* referenced */
	74	u64 rfer_cmpr; /* referenced compressed */
	75	u64 excl; /* exclusive */
	76	u64 excl_cmpr; /* exclusive compressed */
	77
	78	/*
	79	* limits
	80	*/
	81	u64 lim_flags; /* which limits are set */
	82	u64 max_rfer;
	83	u64 max_excl;
	84	u64 rsv_rfer;
	85	u64 rsv_excl;
	86
	87	/*
	88	* reservation tracking
	89	*/
	90	u64 reserved;
	91
	92	/*
	93	* lists
	94	*/
	95	struct list_head groups; /* groups this group is member of */
	96	struct list_head members; /* groups that are members of this group */
	97	struct list_head dirty; /* dirty groups */
	98	struct rb_node node; /* tree of qgroups */
	99
	100	/*
	101	* temp variables for accounting operations
	102	* Refer to qgroup_shared_accounting() for details.
	103	*/
	104	u64 old_refcnt;
	105	u64 new_refcnt;
	106	};
	107
81fb6f77 QW	108	/*
	109	* For qgroup event trace points only
	110	*/
	111	#define QGROUP_RESERVE (1<<0)
	112	#define QGROUP_RELEASE (1<<1)
	113	#define QGROUP_FREE (1<<2)
	114
fcebe456 JB	115	int btrfs_quota_enable(struct btrfs_trans_handle *trans,
	116	struct btrfs_fs_info *fs_info);
	117	int btrfs_quota_disable(struct btrfs_trans_handle *trans,
	118	struct btrfs_fs_info *fs_info);
	119	int btrfs_qgroup_rescan(struct btrfs_fs_info *fs_info);
	120	void btrfs_qgroup_rescan_resume(struct btrfs_fs_info *fs_info);
d06f23d6 JM	121	int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info,
d06f23d6 JM	122	bool interruptible);
fcebe456 JB	123	int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans,
	124	struct btrfs_fs_info *fs_info, u64 src, u64 dst);
	125	int btrfs_del_qgroup_relation(struct btrfs_trans_handle *trans,
	126	struct btrfs_fs_info *fs_info, u64 src, u64 dst);
	127	int btrfs_create_qgroup(struct btrfs_trans_handle *trans,
4087cf24	128	struct btrfs_fs_info *fs_info, u64 qgroupid);
fcebe456 JB	129	int btrfs_remove_qgroup(struct btrfs_trans_handle *trans,
	130	struct btrfs_fs_info *fs_info, u64 qgroupid);
	131	int btrfs_limit_qgroup(struct btrfs_trans_handle *trans,
	132	struct btrfs_fs_info *fs_info, u64 qgroupid,
	133	struct btrfs_qgroup_limit *limit);
	134	int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info);
	135	void btrfs_free_qgroup_config(struct btrfs_fs_info *fs_info);
	136	struct btrfs_delayed_extent_op;
3b7d00f9 QW	137	int btrfs_qgroup_prepare_account_extents(struct btrfs_trans_handle *trans,
3b7d00f9 QW	138	struct btrfs_fs_info *fs_info);
cb93b52c	139	/*
50b3e040	140	* Inform qgroup to trace one dirty extent, its info is recorded in @record.
fb235dc0	141	* So qgroup can account it at transaction committing time.
cb93b52c	142	*
fb235dc0 QW	143	* No lock version, caller must acquire delayed ref lock and allocated memory,
fb235dc0 QW	144	* then call btrfs_qgroup_trace_extent_post() after exiting lock context.
cb93b52c QW	145	*
	146	* Return 0 for success insert
	147	* Return >0 for existing record, caller can free @record safely.
	148	* Error is not possible
	149	*/
50b3e040	150	int btrfs_qgroup_trace_extent_nolock(
cb93b52c QW	151	struct btrfs_fs_info *fs_info,
	152	struct btrfs_delayed_ref_root *delayed_refs,
	153	struct btrfs_qgroup_extent_record *record);
	154
fb235dc0 QW	155	/*
	156	* Post handler after qgroup_trace_extent_nolock().
	157	*
	158	* NOTE: Current qgroup does the expensive backref walk at transaction
	159	* committing time with TRANS_STATE_COMMIT_DOING, this blocks incoming
	160	* new transaction.
	161	* This is designed to allow btrfs_find_all_roots() to get correct new_roots
	162	* result.
	163	*
	164	* However for old_roots there is no need to do backref walk at that time,
	165	* since we search commit roots to walk backref and result will always be
	166	* correct.
	167	*
	168	* Due to the nature of no lock version, we can't do backref there.
	169	* So we must call btrfs_qgroup_trace_extent_post() after exiting
	170	* spinlock context.
	171	*
	172	* TODO: If we can fix and prove btrfs_find_all_roots() can get correct result
	173	* using current root, then we can move all expensive backref walk out of
	174	* transaction committing, but not now as qgroup accounting will be wrong again.
	175	*/
	176	int btrfs_qgroup_trace_extent_post(struct btrfs_fs_info *fs_info,
	177	struct btrfs_qgroup_extent_record *qrecord);
	178
cb93b52c	179	/*
50b3e040 QW	180	* Inform qgroup to trace one dirty extent, specified by @bytenr and
	181	* @num_bytes.
	182	* So qgroup can account it at commit trans time.
cb93b52c	183	*
fb235dc0 QW	184	* Better encapsulated version, with memory allocation and backref walk for
	185	* commit roots.
	186	* So this can sleep.
cb93b52c QW	187	*
	188	* Return 0 if the operation is done.
	189	* Return <0 for error, like memory allocation failure or invalid parameter
	190	* (NULL trans)
	191	*/
50b3e040	192	int btrfs_qgroup_trace_extent(struct btrfs_trans_handle *trans,
cb93b52c QW	193	struct btrfs_fs_info *fs_info, u64 bytenr, u64 num_bytes,
	194	gfp_t gfp_flag);
	195
33d1f05c QW	196	/*
	197	* Inform qgroup to trace all leaf items of data
	198	*
	199	* Return 0 for success
	200	* Return <0 for error(ENOMEM)
	201	*/
	202	int btrfs_qgroup_trace_leaf_items(struct btrfs_trans_handle *trans,
2ff7e61e	203	struct btrfs_fs_info *fs_info,
33d1f05c QW	204	struct extent_buffer *eb);
	205	/*
	206	* Inform qgroup to trace a whole subtree, including all its child tree
	207	* blocks and data.
	208	* The root tree block is specified by @root_eb.
	209	*
	210	* Normally used by relocation(tree block swap) and subvolume deletion.
	211	*
	212	* Return 0 for success
	213	* Return <0 for error(ENOMEM or tree search error)
	214	*/
	215	int btrfs_qgroup_trace_subtree(struct btrfs_trans_handle *trans,
	216	struct btrfs_root *root,
	217	struct extent_buffer *root_eb,
	218	u64 root_gen, int root_level);
442244c9 QW	219	int
	220	btrfs_qgroup_account_extent(struct btrfs_trans_handle *trans,
	221	struct btrfs_fs_info *fs_info,
	222	u64 bytenr, u64 num_bytes,
	223	struct ulist old_roots, struct ulist new_roots);
550d7a2e QW	224	int btrfs_qgroup_account_extents(struct btrfs_trans_handle *trans,
550d7a2e QW	225	struct btrfs_fs_info *fs_info);
fcebe456 JB	226	int btrfs_run_qgroups(struct btrfs_trans_handle *trans,
	227	struct btrfs_fs_info *fs_info);
	228	int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans,
	229	struct btrfs_fs_info *fs_info, u64 srcid, u64 objectid,
	230	struct btrfs_qgroup_inherit *inherit);
297d750b QW	231	void btrfs_qgroup_free_refroot(struct btrfs_fs_info *fs_info,
297d750b QW	232	u64 ref_root, u64 num_bytes);
297d750b QW	233	static inline void btrfs_qgroup_free_delayed_ref(struct btrfs_fs_info *fs_info,
	234	u64 ref_root, u64 num_bytes)
	235	{
bc074524	236	trace_btrfs_qgroup_free_delayed_ref(fs_info, ref_root, num_bytes);
d51ea5dd	237	btrfs_qgroup_free_refroot(fs_info, ref_root, num_bytes);
297d750b	238	}
fcebe456 JB	239
	240	#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
	241	int btrfs_verify_qgroup_counts(struct btrfs_fs_info *fs_info, u64 qgroupid,
	242	u64 rfer, u64 excl);
	243	#endif
	244
52472553 QW	245	/* New io_tree based accurate qgroup reserve API */
52472553 QW	246	int btrfs_qgroup_reserve_data(struct inode *inode, u64 start, u64 len);
f695fdce QW	247	int btrfs_qgroup_release_data(struct inode *inode, u64 start, u64 len);
f695fdce QW	248	int btrfs_qgroup_free_data(struct inode *inode, u64 start, u64 len);
55eeaf05	249
003d7c59 JM	250	int btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
003d7c59 JM	251	bool enforce);
55eeaf05 QW	252	void btrfs_qgroup_free_meta_all(struct btrfs_root *root);
55eeaf05 QW	253	void btrfs_qgroup_free_meta(struct btrfs_root *root, int num_bytes);
56fa9d07	254	void btrfs_qgroup_check_reserved_leak(struct inode *inode);
fcebe456	255	#endif /* __BTRFS_QGROUP__ */