[linux-2.6-block.git] / fs / ext4 / ext4_extents.h

// SPDX-License-Identifier: GPL-2.0
/*
 * Copyright (c) 2003-2006, Cluster File Systems, Inc, info@clusterfs.com
 * Written by Alex Tomas <alex@clusterfs.com>
 */

#ifndef _EXT4_EXTENTS
#define _EXT4_EXTENTS

#include "ext4.h"

/*
 * With AGGRESSIVE_TEST defined, the capacity of index/leaf blocks
 * becomes very small, so index split, in-depth growing and
 * other hard changes happen much more often.
 * This is for debug purposes only.
 */
#define AGGRESSIVE_TEST_

/*
 * With EXTENTS_STATS defined, the number of blocks and extents
 * are collected in the truncate path. They'll be shown at
 * umount time.
 */
#define EXTENTS_STATS__

/*
 * If CHECK_BINSEARCH is defined, then the results of the binary search
 * will also be checked by linear search.
 */
#define CHECK_BINSEARCH__

/*
 * If EXT_STATS is defined then stats numbers are collected.
 * These number will be displayed at umount time.
 */
#define EXT_STATS_


/*
 * ext4_inode has i_block array (60 bytes total).
 * The first 12 bytes store ext4_extent_header;
 * the remainder stores an array of ext4_extent.
 * For non-inode extent blocks, ext4_extent_tail
 * follows the array.
 */

/*
 * This is the extent tail on-disk structure.
 * All other extent structures are 12 bytes long.  It turns out that
 * block_size % 12 >= 4 for at least all powers of 2 greater than 512, which
 * covers all valid ext4 block sizes.  Therefore, this tail structure can be
 * crammed into the end of the block without having to rebalance the tree.
 */
struct ext4_extent_tail {
	__le32	et_checksum;	/* crc32c(uuid+inum+extent_block) */
};

/*
 * This is the extent on-disk structure.
 * It's used at the bottom of the tree.
 */
struct ext4_extent {
	__le32	ee_block;	/* first logical block extent covers */
	__le16	ee_len;		/* number of blocks covered by extent */
	__le16	ee_start_hi;	/* high 16 bits of physical block */
	__le32	ee_start_lo;	/* low 32 bits of physical block */
};

/*
 * This is index on-disk structure.
 * It's used at all the levels except the bottom.
 */
struct ext4_extent_idx {
	__le32	ei_block;	/* index covers logical blocks from 'block' */
	__le32	ei_leaf_lo;	/* pointer to the physical block of the next *
				 * level. leaf or next index could be there */
	__le16	ei_leaf_hi;	/* high 16 bits of physical block */
	__u16	ei_unused;
};

/*
 * Each block (leaves and indexes), even inode-stored has header.
 */
struct ext4_extent_header {
	__le16	eh_magic;	/* probably will support different formats */
	__le16	eh_entries;	/* number of valid entries */
	__le16	eh_max;		/* capacity of store in entries */
	__le16	eh_depth;	/* has tree real underlying blocks? */
	__le32	eh_generation;	/* generation of the tree */
};

#define EXT4_EXT_MAGIC		cpu_to_le16(0xf30a)
#define EXT4_MAX_EXTENT_DEPTH 5

#define EXT4_EXTENT_TAIL_OFFSET(hdr) \
	(sizeof(struct ext4_extent_header) + \
	 (sizeof(struct ext4_extent) * le16_to_cpu((hdr)->eh_max)))

static inline struct ext4_extent_tail *
find_ext4_extent_tail(struct ext4_extent_header *eh)
{
	return (struct ext4_extent_tail *)(((void *)eh) +
					   EXT4_EXTENT_TAIL_OFFSET(eh));
}

/*
 * Array of ext4_ext_path contains path to some extent.
 * Creation/lookup routines use it for traversal/splitting/etc.
 * Truncate uses it to simulate recursive walking.
 */
struct ext4_ext_path {
	ext4_fsblk_t			p_block;
	__u16				p_depth;
	__u16				p_maxdepth;
	struct ext4_extent		*p_ext;
	struct ext4_extent_idx		*p_idx;
	struct ext4_extent_header	*p_hdr;
	struct buffer_head		*p_bh;
};

/*
 * Used to record a portion of a cluster found at the beginning or end
 * of an extent while traversing the extent tree during space removal.
 * A partial cluster may be removed if it does not contain blocks shared
 * with extents that aren't being deleted (tofree state).  Otherwise,
 * it cannot be removed (nofree state).
 */
struct partial_cluster {
	ext4_fsblk_t pclu;  /* physical cluster number */
	ext4_lblk_t lblk;   /* logical block number within logical cluster */
	enum {initial, tofree, nofree} state;
};

/*
 * structure for external API
 */

/*
 * EXT_INIT_MAX_LEN is the maximum number of blocks we can have in an
 * initialized extent. This is 2^15 and not (2^16 - 1), since we use the
 * MSB of ee_len field in the extent datastructure to signify if this
 * particular extent is an initialized extent or an unwritten (i.e.
 * preallocated).
 * EXT_UNWRITTEN_MAX_LEN is the maximum number of blocks we can have in an
 * unwritten extent.
 * If ee_len is <= 0x8000, it is an initialized extent. Otherwise, it is an
 * unwritten one. In other words, if MSB of ee_len is set, it is an
 * unwritten extent with only one special scenario when ee_len = 0x8000.
 * In this case we can not have an unwritten extent of zero length and
 * thus we make it as a special case of initialized extent with 0x8000 length.
 * This way we get better extent-to-group alignment for initialized extents.
 * Hence, the maximum number of blocks we can have in an *initialized*
 * extent is 2^15 (32768) and in an *unwritten* extent is 2^15-1 (32767).
 */
#define EXT_INIT_MAX_LEN	(1UL << 15)
#define EXT_UNWRITTEN_MAX_LEN	(EXT_INIT_MAX_LEN - 1)


#define EXT_FIRST_EXTENT(__hdr__) \
	((struct ext4_extent *) (((char *) (__hdr__)) +		\
				 sizeof(struct ext4_extent_header)))
#define EXT_FIRST_INDEX(__hdr__) \
	((struct ext4_extent_idx *) (((char *) (__hdr__)) +	\
				     sizeof(struct ext4_extent_header)))
#define EXT_HAS_FREE_INDEX(__path__) \
	(le16_to_cpu((__path__)->p_hdr->eh_entries) \
				     < le16_to_cpu((__path__)->p_hdr->eh_max))
#define EXT_LAST_EXTENT(__hdr__) \
	(EXT_FIRST_EXTENT((__hdr__)) + le16_to_cpu((__hdr__)->eh_entries) - 1)
#define EXT_LAST_INDEX(__hdr__) \
	(EXT_FIRST_INDEX((__hdr__)) + le16_to_cpu((__hdr__)->eh_entries) - 1)
#define EXT_MAX_EXTENT(__hdr__) \
	(EXT_FIRST_EXTENT((__hdr__)) + le16_to_cpu((__hdr__)->eh_max) - 1)
#define EXT_MAX_INDEX(__hdr__) \
	(EXT_FIRST_INDEX((__hdr__)) + le16_to_cpu((__hdr__)->eh_max) - 1)

static inline struct ext4_extent_header *ext_inode_hdr(struct inode *inode)
{
	return (struct ext4_extent_header *) EXT4_I(inode)->i_data;
}

static inline struct ext4_extent_header *ext_block_hdr(struct buffer_head *bh)
{
	return (struct ext4_extent_header *) bh->b_data;
}

static inline unsigned short ext_depth(struct inode *inode)
{
	return le16_to_cpu(ext_inode_hdr(inode)->eh_depth);
}

static inline void ext4_ext_mark_unwritten(struct ext4_extent *ext)
{
	/* We can not have an unwritten extent of zero length! */
	BUG_ON((le16_to_cpu(ext->ee_len) & ~EXT_INIT_MAX_LEN) == 0);
	ext->ee_len |= cpu_to_le16(EXT_INIT_MAX_LEN);
}

static inline int ext4_ext_is_unwritten(struct ext4_extent *ext)
{
	/* Extent with ee_len of 0x8000 is treated as an initialized extent */
	return (le16_to_cpu(ext->ee_len) > EXT_INIT_MAX_LEN);
}

static inline int ext4_ext_get_actual_len(struct ext4_extent *ext)
{
	return (le16_to_cpu(ext->ee_len) <= EXT_INIT_MAX_LEN ?
		le16_to_cpu(ext->ee_len) :
		(le16_to_cpu(ext->ee_len) - EXT_INIT_MAX_LEN));
}

static inline void ext4_ext_mark_initialized(struct ext4_extent *ext)
{
	ext->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ext));
}

/*
 * ext4_ext_pblock:
 * combine low and high parts of physical block number into ext4_fsblk_t
 */
static inline ext4_fsblk_t ext4_ext_pblock(struct ext4_extent *ex)
{
	ext4_fsblk_t block;

	block = le32_to_cpu(ex->ee_start_lo);
	block |= ((ext4_fsblk_t) le16_to_cpu(ex->ee_start_hi) << 31) << 1;
	return block;
}

/*
 * ext4_idx_pblock:
 * combine low and high parts of a leaf physical block number into ext4_fsblk_t
 */
static inline ext4_fsblk_t ext4_idx_pblock(struct ext4_extent_idx *ix)
{
	ext4_fsblk_t block;

	block = le32_to_cpu(ix->ei_leaf_lo);
	block |= ((ext4_fsblk_t) le16_to_cpu(ix->ei_leaf_hi) << 31) << 1;
	return block;
}

/*
 * ext4_ext_store_pblock:
 * stores a large physical block number into an extent struct,
 * breaking it into parts
 */
static inline void ext4_ext_store_pblock(struct ext4_extent *ex,
					 ext4_fsblk_t pb)
{
	ex->ee_start_lo = cpu_to_le32((unsigned long) (pb & 0xffffffff));
	ex->ee_start_hi = cpu_to_le16((unsigned long) ((pb >> 31) >> 1) &
				      0xffff);
}

/*
 * ext4_idx_store_pblock:
 * stores a large physical block number into an index struct,
 * breaking it into parts
 */
static inline void ext4_idx_store_pblock(struct ext4_extent_idx *ix,
					 ext4_fsblk_t pb)
{
	ix->ei_leaf_lo = cpu_to_le32((unsigned long) (pb & 0xffffffff));
	ix->ei_leaf_hi = cpu_to_le16((unsigned long) ((pb >> 31) >> 1) &
				     0xffff);
}

#define ext4_ext_dirty(handle, inode, path) \
		__ext4_ext_dirty(__func__, __LINE__, (handle), (inode), (path))
int __ext4_ext_dirty(const char *where, unsigned int line, handle_t *handle,
		     struct inode *inode, struct ext4_ext_path *path);

#endif /* _EXT4_EXTENTS */
Commit	Line	Data
f5166768	1	// SPDX-License-Identifier: GPL-2.0
a86c6181 AT	2	/*
	3	* Copyright (c) 2003-2006, Cluster File Systems, Inc, info@clusterfs.com
	4	* Written by Alex Tomas <alex@clusterfs.com>
a86c6181 AT	5	*/
a86c6181 AT	6
3dcf5451 CH	7	#ifndef _EXT4_EXTENTS
3dcf5451 CH	8	#define _EXT4_EXTENTS
a86c6181	9
3dcf5451	10	#include "ext4.h"
a86c6181 AT	11
a86c6181 AT	12	/*
bbf2f9fb	13	* With AGGRESSIVE_TEST defined, the capacity of index/leaf blocks
d0d856e8 RD	14	* becomes very small, so index split, in-depth growing and
	15	* other hard changes happen much more often.
	16	* This is for debug purposes only.
a86c6181	17	*/
bbf2f9fb	18	#define AGGRESSIVE_TEST_
a86c6181 AT	19
a86c6181 AT	20	/*
d0d856e8 RD	21	* With EXTENTS_STATS defined, the number of blocks and extents
	22	* are collected in the truncate path. They'll be shown at
	23	* umount time.
a86c6181 AT	24	*/
	25	#define EXTENTS_STATS__
	26
	27	/*
d0d856e8 RD	28	* If CHECK_BINSEARCH is defined, then the results of the binary search
d0d856e8 RD	29	* will also be checked by linear search.
a86c6181 AT	30	*/
	31	#define CHECK_BINSEARCH__
	32
a86c6181	33	/*
d0d856e8 RD	34	* If EXT_STATS is defined then stats numbers are collected.
d0d856e8 RD	35	* These number will be displayed at umount time.
a86c6181 AT	36	*/
	37	#define EXT_STATS_
	38
	39
	40	/*
d0d856e8 RD	41	* ext4_inode has i_block array (60 bytes total).
	42	* The first 12 bytes store ext4_extent_header;
	43	* the remainder stores an array of ext4_extent.
e6153918 DW	44	* For non-inode extent blocks, ext4_extent_tail
e6153918 DW	45	* follows the array.
a86c6181 AT	46	*/
a86c6181 AT	47
e6153918 DW	48	/*
	49	* This is the extent tail on-disk structure.
	50	* All other extent structures are 12 bytes long. It turns out that
	51	* block_size % 12 >= 4 for at least all powers of 2 greater than 512, which
	52	* covers all valid ext4 block sizes. Therefore, this tail structure can be
	53	* crammed into the end of the block without having to rebalance the tree.
	54	*/
	55	struct ext4_extent_tail {
	56	__le32 et_checksum; /* crc32c(uuid+inum+extent_block) */
	57	};
	58
a86c6181	59	/*
d0d856e8 RD	60	* This is the extent on-disk structure.
d0d856e8 RD	61	* It's used at the bottom of the tree.
a86c6181 AT	62	*/
	63	struct ext4_extent {
	64	__le32 ee_block; /* first logical block extent covers */
	65	__le16 ee_len; /* number of blocks covered by extent */
	66	__le16 ee_start_hi; /* high 16 bits of physical block */
b377611d	67	__le32 ee_start_lo; /* low 32 bits of physical block */
a86c6181 AT	68	};
	69
	70	/*
d0d856e8 RD	71	* This is index on-disk structure.
d0d856e8 RD	72	* It's used at all the levels except the bottom.
a86c6181 AT	73	*/
	74	struct ext4_extent_idx {
	75	__le32 ei_block; /* index covers logical blocks from 'block' */
d8dd0b45	76	__le32 ei_leaf_lo; /* pointer to the physical block of the next *
d0d856e8	77	* level. leaf or next index could be there */
a86c6181 AT	78	__le16 ei_leaf_hi; /* high 16 bits of physical block */
	79	__u16 ei_unused;
	80	};
	81
	82	/*
d0d856e8	83	* Each block (leaves and indexes), even inode-stored has header.
a86c6181 AT	84	*/
	85	struct ext4_extent_header {
	86	__le16 eh_magic; /* probably will support different formats */
	87	__le16 eh_entries; /* number of valid entries */
	88	__le16 eh_max; /* capacity of store in entries */
d0d856e8	89	__le16 eh_depth; /* has tree real underlying blocks? */
a86c6181 AT	90	__le32 eh_generation; /* generation of the tree */
	91	};
	92
	93	#define EXT4_EXT_MAGIC cpu_to_le16(0xf30a)
bc890a60	94	#define EXT4_MAX_EXTENT_DEPTH 5
a86c6181	95
7ac5990d DW	96	#define EXT4_EXTENT_TAIL_OFFSET(hdr) \
	97	(sizeof(struct ext4_extent_header) + \
	98	(sizeof(struct ext4_extent) * le16_to_cpu((hdr)->eh_max)))
	99
	100	static inline struct ext4_extent_tail *
	101	find_ext4_extent_tail(struct ext4_extent_header *eh)
	102	{
	103	return (struct ext4_extent_tail )(((void )eh) +
	104	EXT4_EXTENT_TAIL_OFFSET(eh));
	105	}
	106
a86c6181	107	/*
d0d856e8 RD	108	* Array of ext4_ext_path contains path to some extent.
	109	* Creation/lookup routines use it for traversal/splitting/etc.
	110	* Truncate uses it to simulate recursive walking.
a86c6181 AT	111	*/
a86c6181 AT	112	struct ext4_ext_path {
f65e6fba	113	ext4_fsblk_t p_block;
a86c6181	114	__u16 p_depth;
10809df8	115	__u16 p_maxdepth;
a86c6181 AT	116	struct ext4_extent *p_ext;
	117	struct ext4_extent_idx *p_idx;
	118	struct ext4_extent_header *p_hdr;
	119	struct buffer_head *p_bh;
	120	};
	121
9fe67149 EW	122	/*
	123	* Used to record a portion of a cluster found at the beginning or end
	124	* of an extent while traversing the extent tree during space removal.
	125	* A partial cluster may be removed if it does not contain blocks shared
	126	* with extents that aren't being deleted (tofree state). Otherwise,
	127	* it cannot be removed (nofree state).
	128	*/
	129	struct partial_cluster {
	130	ext4_fsblk_t pclu; /* physical cluster number */
	131	ext4_lblk_t lblk; /* logical block number within logical cluster */
	132	enum {initial, tofree, nofree} state;
	133	};
	134
a86c6181 AT	135	/*
	136	* structure for external API
	137	*/
	138
749269fa AA	139	/*
	140	* EXT_INIT_MAX_LEN is the maximum number of blocks we can have in an
	141	* initialized extent. This is 2^15 and not (2^16 - 1), since we use the
	142	* MSB of ee_len field in the extent datastructure to signify if this
556615dc	143	* particular extent is an initialized extent or an unwritten (i.e.
749269fa	144	* preallocated).
556615dc LC	145	* EXT_UNWRITTEN_MAX_LEN is the maximum number of blocks we can have in an
556615dc LC	146	* unwritten extent.
749269fa	147	* If ee_len is <= 0x8000, it is an initialized extent. Otherwise, it is an
556615dc LC	148	* unwritten one. In other words, if MSB of ee_len is set, it is an
	149	* unwritten extent with only one special scenario when ee_len = 0x8000.
	150	* In this case we can not have an unwritten extent of zero length and
749269fa AA	151	* thus we make it as a special case of initialized extent with 0x8000 length.
	152	* This way we get better extent-to-group alignment for initialized extents.
	153	* Hence, the maximum number of blocks we can have in an initialized
556615dc	154	* extent is 2^15 (32768) and in an unwritten extent is 2^15-1 (32767).
749269fa AA	155	*/
749269fa AA	156	#define EXT_INIT_MAX_LEN (1UL << 15)
556615dc	157	#define EXT_UNWRITTEN_MAX_LEN (EXT_INIT_MAX_LEN - 1)
471d4011	158
a86c6181 AT	159
	160	#define EXT_FIRST_EXTENT(__hdr__) \
	161	((struct ext4_extent ) (((char ) (__hdr__)) + \
	162	sizeof(struct ext4_extent_header)))
	163	#define EXT_FIRST_INDEX(__hdr__) \
	164	((struct ext4_extent_idx ) (((char ) (__hdr__)) + \
	165	sizeof(struct ext4_extent_header)))
	166	#define EXT_HAS_FREE_INDEX(__path__) \
8c55e204 DK	167	(le16_to_cpu((__path__)->p_hdr->eh_entries) \
8c55e204 DK	168	< le16_to_cpu((__path__)->p_hdr->eh_max))
a86c6181 AT	169	#define EXT_LAST_EXTENT(__hdr__) \
	170	(EXT_FIRST_EXTENT((__hdr__)) + le16_to_cpu((__hdr__)->eh_entries) - 1)
	171	#define EXT_LAST_INDEX(__hdr__) \
	172	(EXT_FIRST_INDEX((__hdr__)) + le16_to_cpu((__hdr__)->eh_entries) - 1)
	173	#define EXT_MAX_EXTENT(__hdr__) \
	174	(EXT_FIRST_EXTENT((__hdr__)) + le16_to_cpu((__hdr__)->eh_max) - 1)
	175	#define EXT_MAX_INDEX(__hdr__) \
	176	(EXT_FIRST_INDEX((__hdr__)) + le16_to_cpu((__hdr__)->eh_max) - 1)
	177
	178	static inline struct ext4_extent_header ext_inode_hdr(struct inode inode)
	179	{
	180	return (struct ext4_extent_header *) EXT4_I(inode)->i_data;
	181	}
	182
	183	static inline struct ext4_extent_header ext_block_hdr(struct buffer_head bh)
	184	{
	185	return (struct ext4_extent_header *) bh->b_data;
	186	}
	187
	188	static inline unsigned short ext_depth(struct inode *inode)
	189	{
	190	return le16_to_cpu(ext_inode_hdr(inode)->eh_depth);
	191	}
	192
556615dc	193	static inline void ext4_ext_mark_unwritten(struct ext4_extent *ext)
a2df2a63	194	{
556615dc	195	/* We can not have an unwritten extent of zero length! */
749269fa AA	196	BUG_ON((le16_to_cpu(ext->ee_len) & ~EXT_INIT_MAX_LEN) == 0);
749269fa AA	197	ext->ee_len \|= cpu_to_le16(EXT_INIT_MAX_LEN);
a2df2a63 AA	198	}
a2df2a63 AA	199
556615dc	200	static inline int ext4_ext_is_unwritten(struct ext4_extent *ext)
a2df2a63	201	{
749269fa AA	202	/* Extent with ee_len of 0x8000 is treated as an initialized extent */
749269fa AA	203	return (le16_to_cpu(ext->ee_len) > EXT_INIT_MAX_LEN);
a2df2a63 AA	204	}
	205
	206	static inline int ext4_ext_get_actual_len(struct ext4_extent *ext)
	207	{
749269fa AA	208	return (le16_to_cpu(ext->ee_len) <= EXT_INIT_MAX_LEN ?
	209	le16_to_cpu(ext->ee_len) :
	210	(le16_to_cpu(ext->ee_len) - EXT_INIT_MAX_LEN));
a2df2a63 AA	211	}
a2df2a63 AA	212
0031462b MC	213	static inline void ext4_ext_mark_initialized(struct ext4_extent *ext)
	214	{
	215	ext->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ext));
	216	}
	217
bf89d16f TT	218	/*
	219	* ext4_ext_pblock:
	220	* combine low and high parts of physical block number into ext4_fsblk_t
	221	*/
	222	static inline ext4_fsblk_t ext4_ext_pblock(struct ext4_extent *ex)
	223	{
	224	ext4_fsblk_t block;
	225
	226	block = le32_to_cpu(ex->ee_start_lo);
	227	block \|= ((ext4_fsblk_t) le16_to_cpu(ex->ee_start_hi) << 31) << 1;
	228	return block;
	229	}
	230
	231	/*
	232	* ext4_idx_pblock:
	233	* combine low and high parts of a leaf physical block number into ext4_fsblk_t
	234	*/
	235	static inline ext4_fsblk_t ext4_idx_pblock(struct ext4_extent_idx *ix)
	236	{
	237	ext4_fsblk_t block;
	238
	239	block = le32_to_cpu(ix->ei_leaf_lo);
	240	block \|= ((ext4_fsblk_t) le16_to_cpu(ix->ei_leaf_hi) << 31) << 1;
	241	return block;
	242	}
	243
	244	/*
	245	* ext4_ext_store_pblock:
	246	* stores a large physical block number into an extent struct,
	247	* breaking it into parts
	248	*/
	249	static inline void ext4_ext_store_pblock(struct ext4_extent *ex,
	250	ext4_fsblk_t pb)
	251	{
	252	ex->ee_start_lo = cpu_to_le32((unsigned long) (pb & 0xffffffff));
	253	ex->ee_start_hi = cpu_to_le16((unsigned long) ((pb >> 31) >> 1) &
	254	0xffff);
	255	}
	256
	257	/*
	258	* ext4_idx_store_pblock:
	259	* stores a large physical block number into an index struct,
	260	* breaking it into parts
	261	*/
	262	static inline void ext4_idx_store_pblock(struct ext4_extent_idx *ix,
	263	ext4_fsblk_t pb)
	264	{
	265	ix->ei_leaf_lo = cpu_to_le32((unsigned long) (pb & 0xffffffff));
	266	ix->ei_leaf_hi = cpu_to_le16((unsigned long) ((pb >> 31) >> 1) &
	267	0xffff);
	268	}
	269
2656497b DW	270	#define ext4_ext_dirty(handle, inode, path) \
	271	__ext4_ext_dirty(__func__, __LINE__, (handle), (inode), (path))
	272	int __ext4_ext_dirty(const char where, unsigned int line, handle_t handle,
	273	struct inode inode, struct ext4_ext_path path);
	274
3dcf5451	275	#endif /* _EXT4_EXTENTS */
a86c6181	276