2 * Copyright (c) 2003-2006, Cluster File Systems, Inc, info@clusterfs.com
3 * Written by Alex Tomas <alex@clusterfs.com>
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License version 2 as
7 * published by the Free Software Foundation.
9 * This program is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
14 * You should have received a copy of the GNU General Public Licens
15 * along with this program; if not, write to the Free Software
16 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-
25 * With AGGRESSIVE_TEST defined, the capacity of index/leaf blocks
26 * becomes very small, so index split, in-depth growing and
27 * other hard changes happen much more often.
28 * This is for debug purposes only.
30 #define AGGRESSIVE_TEST_
33 * With EXTENTS_STATS defined, the number of blocks and extents
34 * are collected in the truncate path. They'll be shown at
37 #define EXTENTS_STATS__
40 * If CHECK_BINSEARCH is defined, then the results of the binary search
41 * will also be checked by linear search.
43 #define CHECK_BINSEARCH__
46 * Turn on EXT_DEBUG to get lots of info about extents operations.
50 #define ext_debug(a...) printk(a)
52 #define ext_debug(a...)
56 * If EXT_STATS is defined then stats numbers are collected.
57 * These number will be displayed at umount time.
63 * ext4_inode has i_block array (60 bytes total).
64 * The first 12 bytes store ext4_extent_header;
65 * the remainder stores an array of ext4_extent.
69 * This is the extent on-disk structure.
70 * It's used at the bottom of the tree.
73 __le32 ee_block
; /* first logical block extent covers */
74 __le16 ee_len
; /* number of blocks covered by extent */
75 __le16 ee_start_hi
; /* high 16 bits of physical block */
76 __le32 ee_start_lo
; /* low 32 bits of physical block */
80 * This is index on-disk structure.
81 * It's used at all the levels except the bottom.
83 struct ext4_extent_idx
{
84 __le32 ei_block
; /* index covers logical blocks from 'block' */
85 __le32 ei_leaf_lo
; /* pointer to the physical block of the next *
86 * level. leaf or next index could be there */
87 __le16 ei_leaf_hi
; /* high 16 bits of physical block */
92 * Each block (leaves and indexes), even inode-stored has header.
94 struct ext4_extent_header
{
95 __le16 eh_magic
; /* probably will support different formats */
96 __le16 eh_entries
; /* number of valid entries */
97 __le16 eh_max
; /* capacity of store in entries */
98 __le16 eh_depth
; /* has tree real underlying blocks? */
99 __le32 eh_generation
; /* generation of the tree */
102 #define EXT4_EXT_MAGIC cpu_to_le16(0xf30a)
105 * Array of ext4_ext_path contains path to some extent.
106 * Creation/lookup routines use it for traversal/splitting/etc.
107 * Truncate uses it to simulate recursive walking.
109 struct ext4_ext_path
{
110 ext4_fsblk_t p_block
;
112 struct ext4_extent
*p_ext
;
113 struct ext4_extent_idx
*p_idx
;
114 struct ext4_extent_header
*p_hdr
;
115 struct buffer_head
*p_bh
;
119 * structure for external API
123 * to be called by ext4_ext_walk_space()
124 * negative retcode - error
125 * positive retcode - signal for ext4_ext_walk_space(), see below
126 * callback must return valid extent (passed or newly created)
128 typedef int (*ext_prepare_callback
)(struct inode
*, struct ext4_ext_path
*,
129 struct ext4_ext_cache
*,
130 struct ext4_extent
*, void *);
132 #define EXT_CONTINUE 0
136 /* Maximum logical block in a file; ext4_extent's ee_block is __le32 */
137 #define EXT_MAX_BLOCK 0xffffffff
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
143 * particular extent is an initialized extent or an uninitialized (i.e.
145 * EXT_UNINIT_MAX_LEN is the maximum number of blocks we can have in an
146 * uninitialized extent.
147 * If ee_len is <= 0x8000, it is an initialized extent. Otherwise, it is an
148 * uninitialized one. In other words, if MSB of ee_len is set, it is an
149 * uninitialized extent with only one special scenario when ee_len = 0x8000.
150 * In this case we can not have an uninitialized extent of zero length and
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*
154 * extent is 2^15 (32768) and in an *uninitialized* extent is 2^15-1 (32767).
156 #define EXT_INIT_MAX_LEN (1UL << 15)
157 #define EXT_UNINIT_MAX_LEN (EXT_INIT_MAX_LEN - 1)
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__) \
167 (le16_to_cpu((__path__)->p_hdr->eh_entries) \
168 < le16_to_cpu((__path__)->p_hdr->eh_max))
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)
178 static inline struct ext4_extent_header
*ext_inode_hdr(struct inode
*inode
)
180 return (struct ext4_extent_header
*) EXT4_I(inode
)->i_data
;
183 static inline struct ext4_extent_header
*ext_block_hdr(struct buffer_head
*bh
)
185 return (struct ext4_extent_header
*) bh
->b_data
;
188 static inline unsigned short ext_depth(struct inode
*inode
)
190 return le16_to_cpu(ext_inode_hdr(inode
)->eh_depth
);
194 ext4_ext_invalidate_cache(struct inode
*inode
)
196 EXT4_I(inode
)->i_cached_extent
.ec_len
= 0;
199 static inline void ext4_ext_mark_uninitialized(struct ext4_extent
*ext
)
201 /* We can not have an uninitialized extent of zero length! */
202 BUG_ON((le16_to_cpu(ext
->ee_len
) & ~EXT_INIT_MAX_LEN
) == 0);
203 ext
->ee_len
|= cpu_to_le16(EXT_INIT_MAX_LEN
);
206 static inline int ext4_ext_is_uninitialized(struct ext4_extent
*ext
)
208 /* Extent with ee_len of 0x8000 is treated as an initialized extent */
209 return (le16_to_cpu(ext
->ee_len
) > EXT_INIT_MAX_LEN
);
212 static inline int ext4_ext_get_actual_len(struct ext4_extent
*ext
)
214 return (le16_to_cpu(ext
->ee_len
) <= EXT_INIT_MAX_LEN
?
215 le16_to_cpu(ext
->ee_len
) :
216 (le16_to_cpu(ext
->ee_len
) - EXT_INIT_MAX_LEN
));
219 static inline void ext4_ext_mark_initialized(struct ext4_extent
*ext
)
221 ext
->ee_len
= cpu_to_le16(ext4_ext_get_actual_len(ext
));
226 * combine low and high parts of physical block number into ext4_fsblk_t
228 static inline ext4_fsblk_t
ext4_ext_pblock(struct ext4_extent
*ex
)
232 block
= le32_to_cpu(ex
->ee_start_lo
);
233 block
|= ((ext4_fsblk_t
) le16_to_cpu(ex
->ee_start_hi
) << 31) << 1;
239 * combine low and high parts of a leaf physical block number into ext4_fsblk_t
241 static inline ext4_fsblk_t
ext4_idx_pblock(struct ext4_extent_idx
*ix
)
245 block
= le32_to_cpu(ix
->ei_leaf_lo
);
246 block
|= ((ext4_fsblk_t
) le16_to_cpu(ix
->ei_leaf_hi
) << 31) << 1;
251 * ext4_ext_store_pblock:
252 * stores a large physical block number into an extent struct,
253 * breaking it into parts
255 static inline void ext4_ext_store_pblock(struct ext4_extent
*ex
,
258 ex
->ee_start_lo
= cpu_to_le32((unsigned long) (pb
& 0xffffffff));
259 ex
->ee_start_hi
= cpu_to_le16((unsigned long) ((pb
>> 31) >> 1) &
264 * ext4_idx_store_pblock:
265 * stores a large physical block number into an index struct,
266 * breaking it into parts
268 static inline void ext4_idx_store_pblock(struct ext4_extent_idx
*ix
,
271 ix
->ei_leaf_lo
= cpu_to_le32((unsigned long) (pb
& 0xffffffff));
272 ix
->ei_leaf_hi
= cpu_to_le16((unsigned long) ((pb
>> 31) >> 1) &
276 extern int ext4_ext_calc_metadata_amount(struct inode
*inode
,
277 ext4_lblk_t lblocks
);
278 extern int ext4_extent_tree_init(handle_t
*, struct inode
*);
279 extern int ext4_ext_calc_credits_for_single_extent(struct inode
*inode
,
281 struct ext4_ext_path
*path
);
282 extern int ext4_can_extents_be_merged(struct inode
*inode
,
283 struct ext4_extent
*ex1
,
284 struct ext4_extent
*ex2
);
285 extern int ext4_ext_insert_extent(handle_t
*, struct inode
*, struct ext4_ext_path
*, struct ext4_extent
*, int);
286 extern struct ext4_ext_path
*ext4_ext_find_extent(struct inode
*, ext4_lblk_t
,
287 struct ext4_ext_path
*);
288 extern void ext4_ext_drop_refs(struct ext4_ext_path
*);
289 extern int ext4_ext_check_inode(struct inode
*inode
);
290 #endif /* _EXT4_EXTENTS */