2 * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as
7 * published by the Free Software Foundation.
9 * This program is distributed in the hope that it would 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 License
15 * along with this program; if not, write the Free Software Foundation,
16 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
18 #ifndef __XFS_BTREE_H__
19 #define __XFS_BTREE_H__
27 extern kmem_zone_t *xfs_btree_cur_zone;
30 * Generic key, ptr and record wrapper structures.
32 * These are disk format structures, and are converted where necessary
33 * by the btree specific code that needs to interpret them.
36 __be32 s; /* short form ptr */
37 __be64 l; /* long form ptr */
42 xfs_bmdr_key_t bmbr; /* bmbt root block */
43 xfs_alloc_key_t alloc;
44 xfs_inobt_key_t inobt;
48 * In-core key that holds both low and high keys for overlapped btrees.
49 * The two keys are packed next to each other on disk, so do the same
50 * in memory. Preserve the existing xfs_btree_key as a single key to
51 * avoid the mental model breakage that would happen if we passed a
52 * bigkey into a function that operates on a single key.
54 union xfs_btree_bigkey {
55 struct xfs_bmbt_key bmbt;
56 xfs_bmdr_key_t bmbr; /* bmbt root block */
57 xfs_alloc_key_t alloc;
58 struct xfs_inobt_key inobt;
63 xfs_bmdr_rec_t bmbr; /* bmbt root block */
64 xfs_alloc_rec_t alloc;
65 xfs_inobt_rec_t inobt;
69 * This nonsense is to make -wlint happy.
71 #define XFS_LOOKUP_EQ ((xfs_lookup_t)XFS_LOOKUP_EQi)
72 #define XFS_LOOKUP_LE ((xfs_lookup_t)XFS_LOOKUP_LEi)
73 #define XFS_LOOKUP_GE ((xfs_lookup_t)XFS_LOOKUP_GEi)
75 #define XFS_BTNUM_BNO ((xfs_btnum_t)XFS_BTNUM_BNOi)
76 #define XFS_BTNUM_CNT ((xfs_btnum_t)XFS_BTNUM_CNTi)
77 #define XFS_BTNUM_BMAP ((xfs_btnum_t)XFS_BTNUM_BMAPi)
78 #define XFS_BTNUM_INO ((xfs_btnum_t)XFS_BTNUM_INOi)
79 #define XFS_BTNUM_FINO ((xfs_btnum_t)XFS_BTNUM_FINOi)
82 * For logging record fields.
84 #define XFS_BB_MAGIC (1 << 0)
85 #define XFS_BB_LEVEL (1 << 1)
86 #define XFS_BB_NUMRECS (1 << 2)
87 #define XFS_BB_LEFTSIB (1 << 3)
88 #define XFS_BB_RIGHTSIB (1 << 4)
89 #define XFS_BB_BLKNO (1 << 5)
90 #define XFS_BB_LSN (1 << 6)
91 #define XFS_BB_UUID (1 << 7)
92 #define XFS_BB_OWNER (1 << 8)
93 #define XFS_BB_NUM_BITS 5
94 #define XFS_BB_ALL_BITS ((1 << XFS_BB_NUM_BITS) - 1)
95 #define XFS_BB_NUM_BITS_CRC 9
96 #define XFS_BB_ALL_BITS_CRC ((1 << XFS_BB_NUM_BITS_CRC) - 1)
99 * Generic stats interface
101 #define __XFS_BTREE_STATS_INC(mp, type, stat) \
102 XFS_STATS_INC(mp, xs_ ## type ## _2_ ## stat)
103 #define XFS_BTREE_STATS_INC(cur, stat) \
105 struct xfs_mount *__mp = cur->bc_mp; \
106 switch (cur->bc_btnum) { \
107 case XFS_BTNUM_BNO: __XFS_BTREE_STATS_INC(__mp, abtb, stat); break; \
108 case XFS_BTNUM_CNT: __XFS_BTREE_STATS_INC(__mp, abtc, stat); break; \
109 case XFS_BTNUM_BMAP: __XFS_BTREE_STATS_INC(__mp, bmbt, stat); break; \
110 case XFS_BTNUM_INO: __XFS_BTREE_STATS_INC(__mp, ibt, stat); break; \
111 case XFS_BTNUM_FINO: __XFS_BTREE_STATS_INC(__mp, fibt, stat); break; \
112 case XFS_BTNUM_MAX: ASSERT(0); /* fucking gcc */ ; break; \
116 #define __XFS_BTREE_STATS_ADD(mp, type, stat, val) \
117 XFS_STATS_ADD(mp, xs_ ## type ## _2_ ## stat, val)
118 #define XFS_BTREE_STATS_ADD(cur, stat, val) \
120 struct xfs_mount *__mp = cur->bc_mp; \
121 switch (cur->bc_btnum) { \
122 case XFS_BTNUM_BNO: \
123 __XFS_BTREE_STATS_ADD(__mp, abtb, stat, val); break; \
124 case XFS_BTNUM_CNT: \
125 __XFS_BTREE_STATS_ADD(__mp, abtc, stat, val); break; \
126 case XFS_BTNUM_BMAP: \
127 __XFS_BTREE_STATS_ADD(__mp, bmbt, stat, val); break; \
128 case XFS_BTNUM_INO: \
129 __XFS_BTREE_STATS_ADD(__mp, ibt, stat, val); break; \
130 case XFS_BTNUM_FINO: \
131 __XFS_BTREE_STATS_ADD(__mp, fibt, stat, val); break; \
132 case XFS_BTNUM_MAX: ASSERT(0); /* fucking gcc */ ; break; \
136 #define XFS_BTREE_MAXLEVELS 8 /* max of all btrees */
138 struct xfs_btree_ops {
139 /* size of the key and record structures */
143 /* cursor operations */
144 struct xfs_btree_cur *(*dup_cursor)(struct xfs_btree_cur *);
145 void (*update_cursor)(struct xfs_btree_cur *src,
146 struct xfs_btree_cur *dst);
148 /* update btree root pointer */
149 void (*set_root)(struct xfs_btree_cur *cur,
150 union xfs_btree_ptr *nptr, int level_change);
152 /* block allocation / freeing */
153 int (*alloc_block)(struct xfs_btree_cur *cur,
154 union xfs_btree_ptr *start_bno,
155 union xfs_btree_ptr *new_bno,
157 int (*free_block)(struct xfs_btree_cur *cur, struct xfs_buf *bp);
159 /* update last record information */
160 void (*update_lastrec)(struct xfs_btree_cur *cur,
161 struct xfs_btree_block *block,
162 union xfs_btree_rec *rec,
163 int ptr, int reason);
165 /* records in block/level */
166 int (*get_minrecs)(struct xfs_btree_cur *cur, int level);
167 int (*get_maxrecs)(struct xfs_btree_cur *cur, int level);
169 /* records on disk. Matter for the root in inode case. */
170 int (*get_dmaxrecs)(struct xfs_btree_cur *cur, int level);
172 /* init values of btree structures */
173 void (*init_key_from_rec)(union xfs_btree_key *key,
174 union xfs_btree_rec *rec);
175 void (*init_rec_from_cur)(struct xfs_btree_cur *cur,
176 union xfs_btree_rec *rec);
177 void (*init_ptr_from_cur)(struct xfs_btree_cur *cur,
178 union xfs_btree_ptr *ptr);
179 void (*init_high_key_from_rec)(union xfs_btree_key *key,
180 union xfs_btree_rec *rec);
182 /* difference between key value and cursor value */
183 __int64_t (*key_diff)(struct xfs_btree_cur *cur,
184 union xfs_btree_key *key);
187 * Difference between key2 and key1 -- positive if key1 > key2,
188 * negative if key1 < key2, and zero if equal.
190 __int64_t (*diff_two_keys)(struct xfs_btree_cur *cur,
191 union xfs_btree_key *key1,
192 union xfs_btree_key *key2);
194 const struct xfs_buf_ops *buf_ops;
196 #if defined(DEBUG) || defined(XFS_WARN)
197 /* check that k1 is lower than k2 */
198 int (*keys_inorder)(struct xfs_btree_cur *cur,
199 union xfs_btree_key *k1,
200 union xfs_btree_key *k2);
202 /* check that r1 is lower than r2 */
203 int (*recs_inorder)(struct xfs_btree_cur *cur,
204 union xfs_btree_rec *r1,
205 union xfs_btree_rec *r2);
208 /* derive the low & high keys from the records in a leaf block */
209 void (*get_leaf_keys)(struct xfs_btree_cur *cur,
210 struct xfs_btree_block *block,
211 union xfs_btree_key *key);
213 /* derive the low & high keys from the keys in a node block */
214 void (*get_node_keys)(struct xfs_btree_cur *cur,
215 struct xfs_btree_block *block,
216 union xfs_btree_key *key);
218 /* update the parent keys of given btree level */
219 int (*update_keys)(struct xfs_btree_cur *cur, int level);
223 * Reasons for the update_lastrec method to be called.
225 #define LASTREC_UPDATE 0
226 #define LASTREC_INSREC 1
227 #define LASTREC_DELREC 2
230 union xfs_btree_irec {
231 struct xfs_alloc_rec_incore a;
232 struct xfs_bmbt_irec b;
233 struct xfs_inobt_rec_incore i;
237 * Btree cursor structure.
238 * This collects all information needed by the btree code in one place.
240 typedef struct xfs_btree_cur
242 struct xfs_trans *bc_tp; /* transaction we're in, if any */
243 struct xfs_mount *bc_mp; /* file system mount struct */
244 const struct xfs_btree_ops *bc_ops;
245 uint bc_flags; /* btree features - below */
246 union xfs_btree_irec bc_rec; /* current insert/search record value */
247 struct xfs_buf *bc_bufs[XFS_BTREE_MAXLEVELS]; /* buf ptr per level */
248 int bc_ptrs[XFS_BTREE_MAXLEVELS]; /* key/record # */
249 __uint8_t bc_ra[XFS_BTREE_MAXLEVELS]; /* readahead bits */
250 #define XFS_BTCUR_LEFTRA 1 /* left sibling has been read-ahead */
251 #define XFS_BTCUR_RIGHTRA 2 /* right sibling has been read-ahead */
252 __uint8_t bc_nlevels; /* number of levels in the tree */
253 __uint8_t bc_blocklog; /* log2(blocksize) of btree blocks */
254 xfs_btnum_t bc_btnum; /* identifies which btree type */
256 struct { /* needed for BNO, CNT, INO */
257 struct xfs_buf *agbp; /* agf/agi buffer pointer */
258 xfs_agnumber_t agno; /* ag number */
260 struct { /* needed for BMAP */
261 struct xfs_inode *ip; /* pointer to our inode */
262 struct xfs_bmap_free *flist; /* list to free after */
263 xfs_fsblock_t firstblock; /* 1st blk allocated */
264 int allocated; /* count of alloced */
265 short forksize; /* fork's inode space */
266 char whichfork; /* data or attr fork */
267 char flags; /* flags */
268 #define XFS_BTCUR_BPRV_WASDEL 1 /* was delayed */
270 } bc_private; /* per-btree type data */
274 #define XFS_BTREE_LONG_PTRS (1<<0) /* pointers are 64bits long */
275 #define XFS_BTREE_ROOT_IN_INODE (1<<1) /* root may be variable size */
276 #define XFS_BTREE_LASTREC_UPDATE (1<<2) /* track last rec externally */
277 #define XFS_BTREE_CRC_BLOCKS (1<<3) /* uses extended btree blocks */
278 #define XFS_BTREE_OVERLAPPING (1<<4) /* overlapping intervals */
281 #define XFS_BTREE_NOERROR 0
282 #define XFS_BTREE_ERROR 1
285 * Convert from buffer to btree block header.
287 #define XFS_BUF_TO_BLOCK(bp) ((struct xfs_btree_block *)((bp)->b_addr))
291 * Check that block header is ok.
294 xfs_btree_check_block(
295 struct xfs_btree_cur *cur, /* btree cursor */
296 struct xfs_btree_block *block, /* generic btree block pointer */
297 int level, /* level of the btree block */
298 struct xfs_buf *bp); /* buffer containing block, if any */
301 * Check that (long) pointer is ok.
303 int /* error (0 or EFSCORRUPTED) */
304 xfs_btree_check_lptr(
305 struct xfs_btree_cur *cur, /* btree cursor */
306 xfs_fsblock_t ptr, /* btree block disk address */
307 int level); /* btree block level */
310 * Delete the btree cursor.
313 xfs_btree_del_cursor(
314 xfs_btree_cur_t *cur, /* btree cursor */
315 int error); /* del because of error */
318 * Duplicate the btree cursor.
319 * Allocate a new one, copy the record, re-get the buffers.
322 xfs_btree_dup_cursor(
323 xfs_btree_cur_t *cur, /* input cursor */
324 xfs_btree_cur_t **ncur);/* output cursor */
327 * Get a buffer for the block, return it with no data read.
328 * Long-form addressing.
330 struct xfs_buf * /* buffer for fsbno */
332 struct xfs_mount *mp, /* file system mount point */
333 struct xfs_trans *tp, /* transaction pointer */
334 xfs_fsblock_t fsbno, /* file system block number */
335 uint lock); /* lock flags for get_buf */
338 * Get a buffer for the block, return it with no data read.
339 * Short-form addressing.
341 struct xfs_buf * /* buffer for agno/agbno */
343 struct xfs_mount *mp, /* file system mount point */
344 struct xfs_trans *tp, /* transaction pointer */
345 xfs_agnumber_t agno, /* allocation group number */
346 xfs_agblock_t agbno, /* allocation group block number */
347 uint lock); /* lock flags for get_buf */
350 * Check for the cursor referring to the last block at the given level.
352 int /* 1=is last block, 0=not last block */
353 xfs_btree_islastblock(
354 xfs_btree_cur_t *cur, /* btree cursor */
355 int level); /* level to check */
358 * Compute first and last byte offsets for the fields given.
359 * Interprets the offsets table, which contains struct field offsets.
363 __int64_t fields, /* bitmask of fields */
364 const short *offsets,/* table of field offsets */
365 int nbits, /* number of bits to inspect */
366 int *first, /* output: first byte offset */
367 int *last); /* output: last byte offset */
370 * Get a buffer for the block, return it read in.
371 * Long-form addressing.
375 struct xfs_mount *mp, /* file system mount point */
376 struct xfs_trans *tp, /* transaction pointer */
377 xfs_fsblock_t fsbno, /* file system block number */
378 uint lock, /* lock flags for read_buf */
379 struct xfs_buf **bpp, /* buffer for fsbno */
380 int refval, /* ref count value for buffer */
381 const struct xfs_buf_ops *ops);
384 * Read-ahead the block, don't wait for it, don't return a buffer.
385 * Long-form addressing.
388 xfs_btree_reada_bufl(
389 struct xfs_mount *mp, /* file system mount point */
390 xfs_fsblock_t fsbno, /* file system block number */
391 xfs_extlen_t count, /* count of filesystem blocks */
392 const struct xfs_buf_ops *ops);
395 * Read-ahead the block, don't wait for it, don't return a buffer.
396 * Short-form addressing.
399 xfs_btree_reada_bufs(
400 struct xfs_mount *mp, /* file system mount point */
401 xfs_agnumber_t agno, /* allocation group number */
402 xfs_agblock_t agbno, /* allocation group block number */
403 xfs_extlen_t count, /* count of filesystem blocks */
404 const struct xfs_buf_ops *ops);
407 * Initialise a new btree block header
410 xfs_btree_init_block(
411 struct xfs_mount *mp,
420 xfs_btree_init_block_int(
421 struct xfs_mount *mp,
422 struct xfs_btree_block *buf,
431 * Common btree core entry points.
433 int xfs_btree_increment(struct xfs_btree_cur *, int, int *);
434 int xfs_btree_decrement(struct xfs_btree_cur *, int, int *);
435 int xfs_btree_lookup(struct xfs_btree_cur *, xfs_lookup_t, int *);
436 int xfs_btree_update(struct xfs_btree_cur *, union xfs_btree_rec *);
437 int xfs_btree_new_iroot(struct xfs_btree_cur *, int *, int *);
438 int xfs_btree_insert(struct xfs_btree_cur *, int *);
439 int xfs_btree_delete(struct xfs_btree_cur *, int *);
440 int xfs_btree_get_rec(struct xfs_btree_cur *, union xfs_btree_rec **, int *);
441 int xfs_btree_change_owner(struct xfs_btree_cur *cur, __uint64_t new_owner,
442 struct list_head *buffer_list);
445 * btree block CRC helpers
447 void xfs_btree_lblock_calc_crc(struct xfs_buf *);
448 bool xfs_btree_lblock_verify_crc(struct xfs_buf *);
449 void xfs_btree_sblock_calc_crc(struct xfs_buf *);
450 bool xfs_btree_sblock_verify_crc(struct xfs_buf *);
453 * Internal btree helpers also used by xfs_bmap.c.
455 void xfs_btree_log_block(struct xfs_btree_cur *, struct xfs_buf *, int);
456 void xfs_btree_log_recs(struct xfs_btree_cur *, struct xfs_buf *, int, int);
461 static inline int xfs_btree_get_numrecs(struct xfs_btree_block *block)
463 return be16_to_cpu(block->bb_numrecs);
466 static inline void xfs_btree_set_numrecs(struct xfs_btree_block *block,
469 block->bb_numrecs = cpu_to_be16(numrecs);
472 static inline int xfs_btree_get_level(struct xfs_btree_block *block)
474 return be16_to_cpu(block->bb_level);
479 * Min and max functions for extlen, agblock, fileoff, and filblks types.
481 #define XFS_EXTLEN_MIN(a,b) min_t(xfs_extlen_t, (a), (b))
482 #define XFS_EXTLEN_MAX(a,b) max_t(xfs_extlen_t, (a), (b))
483 #define XFS_AGBLOCK_MIN(a,b) min_t(xfs_agblock_t, (a), (b))
484 #define XFS_AGBLOCK_MAX(a,b) max_t(xfs_agblock_t, (a), (b))
485 #define XFS_FILEOFF_MIN(a,b) min_t(xfs_fileoff_t, (a), (b))
486 #define XFS_FILEOFF_MAX(a,b) max_t(xfs_fileoff_t, (a), (b))
487 #define XFS_FILBLKS_MIN(a,b) min_t(xfs_filblks_t, (a), (b))
488 #define XFS_FILBLKS_MAX(a,b) max_t(xfs_filblks_t, (a), (b))
490 #define XFS_FSB_SANITY_CHECK(mp,fsb) \
491 (XFS_FSB_TO_AGNO(mp, fsb) < mp->m_sb.sb_agcount && \
492 XFS_FSB_TO_AGBNO(mp, fsb) < mp->m_sb.sb_agblocks)
495 * Trace hooks. Currently not implemented as they need to be ported
496 * over to the generic tracing functionality, which is some effort.
498 * i,j = integer (32 bit)
499 * b = btree block buffer (xfs_buf_t)
504 #define XFS_BTREE_TRACE_ARGBI(c, b, i)
505 #define XFS_BTREE_TRACE_ARGBII(c, b, i, j)
506 #define XFS_BTREE_TRACE_ARGI(c, i)
507 #define XFS_BTREE_TRACE_ARGIPK(c, i, p, s)
508 #define XFS_BTREE_TRACE_ARGIPR(c, i, p, r)
509 #define XFS_BTREE_TRACE_ARGIK(c, i, k)
510 #define XFS_BTREE_TRACE_ARGR(c, r)
511 #define XFS_BTREE_TRACE_CURSOR(c, t)
513 bool xfs_btree_sblock_v5hdr_verify(struct xfs_buf *bp);
514 bool xfs_btree_sblock_verify(struct xfs_buf *bp, unsigned int max_recs);
515 uint xfs_btree_compute_maxlevels(struct xfs_mount *mp, uint *limits,
518 void xfs_btree_get_leaf_keys(struct xfs_btree_cur *cur,
519 struct xfs_btree_block *block, union xfs_btree_key *key);
520 void xfs_btree_get_node_keys(struct xfs_btree_cur *cur,
521 struct xfs_btree_block *block, union xfs_btree_key *key);
522 int xfs_btree_update_keys(struct xfs_btree_cur *cur, int level);
523 void xfs_btree_get_leaf_keys_overlapped(struct xfs_btree_cur *cur,
524 struct xfs_btree_block *block, union xfs_btree_key *key);
525 void xfs_btree_get_node_keys_overlapped(struct xfs_btree_cur *cur,
526 struct xfs_btree_block *block, union xfs_btree_key *key);
527 int xfs_btree_update_keys_overlapped(struct xfs_btree_cur *cur, int level);
530 #define XFS_BTREE_QUERY_RANGE_CONTINUE 0 /* keep iterating */
531 #define XFS_BTREE_QUERY_RANGE_ABORT 1 /* stop iterating */
532 typedef int (*xfs_btree_query_range_fn)(struct xfs_btree_cur *cur,
533 union xfs_btree_rec *rec, void *priv);
535 int xfs_btree_query_range(struct xfs_btree_cur *cur,
536 union xfs_btree_irec *low_rec, union xfs_btree_irec *high_rec,
537 xfs_btree_query_range_fn fn, void *priv);
539 typedef int (*xfs_btree_visit_blocks_fn)(struct xfs_btree_cur *cur, int level,
541 int xfs_btree_visit_blocks(struct xfs_btree_cur *cur,
542 xfs_btree_visit_blocks_fn fn, void *data);
544 #endif /* __XFS_BTREE_H__ */