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)
80 #define XFS_BTNUM_RMAP ((xfs_btnum_t)XFS_BTNUM_RMAPi)
83 * For logging record fields.
85 #define XFS_BB_MAGIC (1 << 0)
86 #define XFS_BB_LEVEL (1 << 1)
87 #define XFS_BB_NUMRECS (1 << 2)
88 #define XFS_BB_LEFTSIB (1 << 3)
89 #define XFS_BB_RIGHTSIB (1 << 4)
90 #define XFS_BB_BLKNO (1 << 5)
91 #define XFS_BB_LSN (1 << 6)
92 #define XFS_BB_UUID (1 << 7)
93 #define XFS_BB_OWNER (1 << 8)
94 #define XFS_BB_NUM_BITS 5
95 #define XFS_BB_ALL_BITS ((1 << XFS_BB_NUM_BITS) - 1)
96 #define XFS_BB_NUM_BITS_CRC 9
97 #define XFS_BB_ALL_BITS_CRC ((1 << XFS_BB_NUM_BITS_CRC) - 1)
100 * Generic stats interface
102 #define __XFS_BTREE_STATS_INC(mp, type, stat) \
103 XFS_STATS_INC(mp, xs_ ## type ## _2_ ## stat)
104 #define XFS_BTREE_STATS_INC(cur, stat) \
106 struct xfs_mount *__mp = cur->bc_mp; \
107 switch (cur->bc_btnum) { \
108 case XFS_BTNUM_BNO: __XFS_BTREE_STATS_INC(__mp, abtb, stat); break; \
109 case XFS_BTNUM_CNT: __XFS_BTREE_STATS_INC(__mp, abtc, stat); break; \
110 case XFS_BTNUM_BMAP: __XFS_BTREE_STATS_INC(__mp, bmbt, stat); break; \
111 case XFS_BTNUM_INO: __XFS_BTREE_STATS_INC(__mp, ibt, stat); break; \
112 case XFS_BTNUM_FINO: __XFS_BTREE_STATS_INC(__mp, fibt, stat); break; \
113 case XFS_BTNUM_RMAP: __XFS_BTREE_STATS_INC(__mp, rmap, stat); break; \
114 case XFS_BTNUM_MAX: ASSERT(0); /* fucking gcc */ ; break; \
118 #define __XFS_BTREE_STATS_ADD(mp, type, stat, val) \
119 XFS_STATS_ADD(mp, xs_ ## type ## _2_ ## stat, val)
120 #define XFS_BTREE_STATS_ADD(cur, stat, val) \
122 struct xfs_mount *__mp = cur->bc_mp; \
123 switch (cur->bc_btnum) { \
124 case XFS_BTNUM_BNO: \
125 __XFS_BTREE_STATS_ADD(__mp, abtb, stat, val); break; \
126 case XFS_BTNUM_CNT: \
127 __XFS_BTREE_STATS_ADD(__mp, abtc, stat, val); break; \
128 case XFS_BTNUM_BMAP: \
129 __XFS_BTREE_STATS_ADD(__mp, bmbt, stat, val); break; \
130 case XFS_BTNUM_INO: \
131 __XFS_BTREE_STATS_ADD(__mp, ibt, stat, val); break; \
132 case XFS_BTNUM_FINO: \
133 __XFS_BTREE_STATS_ADD(__mp, fibt, stat, val); break; \
134 case XFS_BTNUM_RMAP: \
135 __XFS_BTREE_STATS_ADD(__mp, rmap, stat, val); break; \
136 case XFS_BTNUM_MAX: ASSERT(0); /* fucking gcc */ ; break; \
140 #define XFS_BTREE_MAXLEVELS 9 /* max of all btrees */
142 struct xfs_btree_ops {
143 /* size of the key and record structures */
147 /* cursor operations */
148 struct xfs_btree_cur *(*dup_cursor)(struct xfs_btree_cur *);
149 void (*update_cursor)(struct xfs_btree_cur *src,
150 struct xfs_btree_cur *dst);
152 /* update btree root pointer */
153 void (*set_root)(struct xfs_btree_cur *cur,
154 union xfs_btree_ptr *nptr, int level_change);
156 /* block allocation / freeing */
157 int (*alloc_block)(struct xfs_btree_cur *cur,
158 union xfs_btree_ptr *start_bno,
159 union xfs_btree_ptr *new_bno,
161 int (*free_block)(struct xfs_btree_cur *cur, struct xfs_buf *bp);
163 /* update last record information */
164 void (*update_lastrec)(struct xfs_btree_cur *cur,
165 struct xfs_btree_block *block,
166 union xfs_btree_rec *rec,
167 int ptr, int reason);
169 /* records in block/level */
170 int (*get_minrecs)(struct xfs_btree_cur *cur, int level);
171 int (*get_maxrecs)(struct xfs_btree_cur *cur, int level);
173 /* records on disk. Matter for the root in inode case. */
174 int (*get_dmaxrecs)(struct xfs_btree_cur *cur, int level);
176 /* init values of btree structures */
177 void (*init_key_from_rec)(union xfs_btree_key *key,
178 union xfs_btree_rec *rec);
179 void (*init_rec_from_cur)(struct xfs_btree_cur *cur,
180 union xfs_btree_rec *rec);
181 void (*init_ptr_from_cur)(struct xfs_btree_cur *cur,
182 union xfs_btree_ptr *ptr);
183 void (*init_high_key_from_rec)(union xfs_btree_key *key,
184 union xfs_btree_rec *rec);
186 /* difference between key value and cursor value */
187 __int64_t (*key_diff)(struct xfs_btree_cur *cur,
188 union xfs_btree_key *key);
191 * Difference between key2 and key1 -- positive if key1 > key2,
192 * negative if key1 < key2, and zero if equal.
194 __int64_t (*diff_two_keys)(struct xfs_btree_cur *cur,
195 union xfs_btree_key *key1,
196 union xfs_btree_key *key2);
198 const struct xfs_buf_ops *buf_ops;
200 #if defined(DEBUG) || defined(XFS_WARN)
201 /* check that k1 is lower than k2 */
202 int (*keys_inorder)(struct xfs_btree_cur *cur,
203 union xfs_btree_key *k1,
204 union xfs_btree_key *k2);
206 /* check that r1 is lower than r2 */
207 int (*recs_inorder)(struct xfs_btree_cur *cur,
208 union xfs_btree_rec *r1,
209 union xfs_btree_rec *r2);
212 /* derive the low & high keys from the records in a leaf block */
213 void (*get_leaf_keys)(struct xfs_btree_cur *cur,
214 struct xfs_btree_block *block,
215 union xfs_btree_key *key);
217 /* derive the low & high keys from the keys in a node block */
218 void (*get_node_keys)(struct xfs_btree_cur *cur,
219 struct xfs_btree_block *block,
220 union xfs_btree_key *key);
222 /* update the parent keys of given btree level */
223 int (*update_keys)(struct xfs_btree_cur *cur, int level);
227 * Reasons for the update_lastrec method to be called.
229 #define LASTREC_UPDATE 0
230 #define LASTREC_INSREC 1
231 #define LASTREC_DELREC 2
234 union xfs_btree_irec {
235 struct xfs_alloc_rec_incore a;
236 struct xfs_bmbt_irec b;
237 struct xfs_inobt_rec_incore i;
241 * Btree cursor structure.
242 * This collects all information needed by the btree code in one place.
244 typedef struct xfs_btree_cur
246 struct xfs_trans *bc_tp; /* transaction we're in, if any */
247 struct xfs_mount *bc_mp; /* file system mount struct */
248 const struct xfs_btree_ops *bc_ops;
249 uint bc_flags; /* btree features - below */
250 union xfs_btree_irec bc_rec; /* current insert/search record value */
251 struct xfs_buf *bc_bufs[XFS_BTREE_MAXLEVELS]; /* buf ptr per level */
252 int bc_ptrs[XFS_BTREE_MAXLEVELS]; /* key/record # */
253 __uint8_t bc_ra[XFS_BTREE_MAXLEVELS]; /* readahead bits */
254 #define XFS_BTCUR_LEFTRA 1 /* left sibling has been read-ahead */
255 #define XFS_BTCUR_RIGHTRA 2 /* right sibling has been read-ahead */
256 __uint8_t bc_nlevels; /* number of levels in the tree */
257 __uint8_t bc_blocklog; /* log2(blocksize) of btree blocks */
258 xfs_btnum_t bc_btnum; /* identifies which btree type */
260 struct { /* needed for BNO, CNT, INO */
261 struct xfs_buf *agbp; /* agf/agi buffer pointer */
262 struct xfs_defer_ops *dfops; /* deferred updates */
263 xfs_agnumber_t agno; /* ag number */
265 struct { /* needed for BMAP */
266 struct xfs_inode *ip; /* pointer to our inode */
267 struct xfs_defer_ops *dfops; /* deferred updates */
268 xfs_fsblock_t firstblock; /* 1st blk allocated */
269 int allocated; /* count of alloced */
270 short forksize; /* fork's inode space */
271 char whichfork; /* data or attr fork */
272 char flags; /* flags */
273 #define XFS_BTCUR_BPRV_WASDEL 1 /* was delayed */
275 } bc_private; /* per-btree type data */
279 #define XFS_BTREE_LONG_PTRS (1<<0) /* pointers are 64bits long */
280 #define XFS_BTREE_ROOT_IN_INODE (1<<1) /* root may be variable size */
281 #define XFS_BTREE_LASTREC_UPDATE (1<<2) /* track last rec externally */
282 #define XFS_BTREE_CRC_BLOCKS (1<<3) /* uses extended btree blocks */
283 #define XFS_BTREE_OVERLAPPING (1<<4) /* overlapping intervals */
286 #define XFS_BTREE_NOERROR 0
287 #define XFS_BTREE_ERROR 1
290 * Convert from buffer to btree block header.
292 #define XFS_BUF_TO_BLOCK(bp) ((struct xfs_btree_block *)((bp)->b_addr))
296 * Check that block header is ok.
299 xfs_btree_check_block(
300 struct xfs_btree_cur *cur, /* btree cursor */
301 struct xfs_btree_block *block, /* generic btree block pointer */
302 int level, /* level of the btree block */
303 struct xfs_buf *bp); /* buffer containing block, if any */
306 * Check that (long) pointer is ok.
308 int /* error (0 or EFSCORRUPTED) */
309 xfs_btree_check_lptr(
310 struct xfs_btree_cur *cur, /* btree cursor */
311 xfs_fsblock_t ptr, /* btree block disk address */
312 int level); /* btree block level */
315 * Delete the btree cursor.
318 xfs_btree_del_cursor(
319 xfs_btree_cur_t *cur, /* btree cursor */
320 int error); /* del because of error */
323 * Duplicate the btree cursor.
324 * Allocate a new one, copy the record, re-get the buffers.
327 xfs_btree_dup_cursor(
328 xfs_btree_cur_t *cur, /* input cursor */
329 xfs_btree_cur_t **ncur);/* output cursor */
332 * Get a buffer for the block, return it with no data read.
333 * Long-form addressing.
335 struct xfs_buf * /* buffer for fsbno */
337 struct xfs_mount *mp, /* file system mount point */
338 struct xfs_trans *tp, /* transaction pointer */
339 xfs_fsblock_t fsbno, /* file system block number */
340 uint lock); /* lock flags for get_buf */
343 * Get a buffer for the block, return it with no data read.
344 * Short-form addressing.
346 struct xfs_buf * /* buffer for agno/agbno */
348 struct xfs_mount *mp, /* file system mount point */
349 struct xfs_trans *tp, /* transaction pointer */
350 xfs_agnumber_t agno, /* allocation group number */
351 xfs_agblock_t agbno, /* allocation group block number */
352 uint lock); /* lock flags for get_buf */
355 * Check for the cursor referring to the last block at the given level.
357 int /* 1=is last block, 0=not last block */
358 xfs_btree_islastblock(
359 xfs_btree_cur_t *cur, /* btree cursor */
360 int level); /* level to check */
363 * Compute first and last byte offsets for the fields given.
364 * Interprets the offsets table, which contains struct field offsets.
368 __int64_t fields, /* bitmask of fields */
369 const short *offsets,/* table of field offsets */
370 int nbits, /* number of bits to inspect */
371 int *first, /* output: first byte offset */
372 int *last); /* output: last byte offset */
375 * Get a buffer for the block, return it read in.
376 * Long-form addressing.
380 struct xfs_mount *mp, /* file system mount point */
381 struct xfs_trans *tp, /* transaction pointer */
382 xfs_fsblock_t fsbno, /* file system block number */
383 uint lock, /* lock flags for read_buf */
384 struct xfs_buf **bpp, /* buffer for fsbno */
385 int refval, /* ref count value for buffer */
386 const struct xfs_buf_ops *ops);
389 * Read-ahead the block, don't wait for it, don't return a buffer.
390 * Long-form addressing.
393 xfs_btree_reada_bufl(
394 struct xfs_mount *mp, /* file system mount point */
395 xfs_fsblock_t fsbno, /* file system block number */
396 xfs_extlen_t count, /* count of filesystem blocks */
397 const struct xfs_buf_ops *ops);
400 * Read-ahead the block, don't wait for it, don't return a buffer.
401 * Short-form addressing.
404 xfs_btree_reada_bufs(
405 struct xfs_mount *mp, /* file system mount point */
406 xfs_agnumber_t agno, /* allocation group number */
407 xfs_agblock_t agbno, /* allocation group block number */
408 xfs_extlen_t count, /* count of filesystem blocks */
409 const struct xfs_buf_ops *ops);
412 * Initialise a new btree block header
415 xfs_btree_init_block(
416 struct xfs_mount *mp,
425 xfs_btree_init_block_int(
426 struct xfs_mount *mp,
427 struct xfs_btree_block *buf,
436 * Common btree core entry points.
438 int xfs_btree_increment(struct xfs_btree_cur *, int, int *);
439 int xfs_btree_decrement(struct xfs_btree_cur *, int, int *);
440 int xfs_btree_lookup(struct xfs_btree_cur *, xfs_lookup_t, int *);
441 int xfs_btree_update(struct xfs_btree_cur *, union xfs_btree_rec *);
442 int xfs_btree_new_iroot(struct xfs_btree_cur *, int *, int *);
443 int xfs_btree_insert(struct xfs_btree_cur *, int *);
444 int xfs_btree_delete(struct xfs_btree_cur *, int *);
445 int xfs_btree_get_rec(struct xfs_btree_cur *, union xfs_btree_rec **, int *);
446 int xfs_btree_change_owner(struct xfs_btree_cur *cur, __uint64_t new_owner,
447 struct list_head *buffer_list);
450 * btree block CRC helpers
452 void xfs_btree_lblock_calc_crc(struct xfs_buf *);
453 bool xfs_btree_lblock_verify_crc(struct xfs_buf *);
454 void xfs_btree_sblock_calc_crc(struct xfs_buf *);
455 bool xfs_btree_sblock_verify_crc(struct xfs_buf *);
458 * Internal btree helpers also used by xfs_bmap.c.
460 void xfs_btree_log_block(struct xfs_btree_cur *, struct xfs_buf *, int);
461 void xfs_btree_log_recs(struct xfs_btree_cur *, struct xfs_buf *, int, int);
466 static inline int xfs_btree_get_numrecs(struct xfs_btree_block *block)
468 return be16_to_cpu(block->bb_numrecs);
471 static inline void xfs_btree_set_numrecs(struct xfs_btree_block *block,
474 block->bb_numrecs = cpu_to_be16(numrecs);
477 static inline int xfs_btree_get_level(struct xfs_btree_block *block)
479 return be16_to_cpu(block->bb_level);
484 * Min and max functions for extlen, agblock, fileoff, and filblks types.
486 #define XFS_EXTLEN_MIN(a,b) min_t(xfs_extlen_t, (a), (b))
487 #define XFS_EXTLEN_MAX(a,b) max_t(xfs_extlen_t, (a), (b))
488 #define XFS_AGBLOCK_MIN(a,b) min_t(xfs_agblock_t, (a), (b))
489 #define XFS_AGBLOCK_MAX(a,b) max_t(xfs_agblock_t, (a), (b))
490 #define XFS_FILEOFF_MIN(a,b) min_t(xfs_fileoff_t, (a), (b))
491 #define XFS_FILEOFF_MAX(a,b) max_t(xfs_fileoff_t, (a), (b))
492 #define XFS_FILBLKS_MIN(a,b) min_t(xfs_filblks_t, (a), (b))
493 #define XFS_FILBLKS_MAX(a,b) max_t(xfs_filblks_t, (a), (b))
495 #define XFS_FSB_SANITY_CHECK(mp,fsb) \
496 (XFS_FSB_TO_AGNO(mp, fsb) < mp->m_sb.sb_agcount && \
497 XFS_FSB_TO_AGBNO(mp, fsb) < mp->m_sb.sb_agblocks)
500 * Trace hooks. Currently not implemented as they need to be ported
501 * over to the generic tracing functionality, which is some effort.
503 * i,j = integer (32 bit)
504 * b = btree block buffer (xfs_buf_t)
509 #define XFS_BTREE_TRACE_ARGBI(c, b, i)
510 #define XFS_BTREE_TRACE_ARGBII(c, b, i, j)
511 #define XFS_BTREE_TRACE_ARGI(c, i)
512 #define XFS_BTREE_TRACE_ARGIPK(c, i, p, s)
513 #define XFS_BTREE_TRACE_ARGIPR(c, i, p, r)
514 #define XFS_BTREE_TRACE_ARGIK(c, i, k)
515 #define XFS_BTREE_TRACE_ARGR(c, r)
516 #define XFS_BTREE_TRACE_CURSOR(c, t)
518 bool xfs_btree_sblock_v5hdr_verify(struct xfs_buf *bp);
519 bool xfs_btree_sblock_verify(struct xfs_buf *bp, unsigned int max_recs);
520 uint xfs_btree_compute_maxlevels(struct xfs_mount *mp, uint *limits,
523 void xfs_btree_get_leaf_keys(struct xfs_btree_cur *cur,
524 struct xfs_btree_block *block, union xfs_btree_key *key);
525 void xfs_btree_get_node_keys(struct xfs_btree_cur *cur,
526 struct xfs_btree_block *block, union xfs_btree_key *key);
527 int xfs_btree_update_keys(struct xfs_btree_cur *cur, int level);
528 void xfs_btree_get_leaf_keys_overlapped(struct xfs_btree_cur *cur,
529 struct xfs_btree_block *block, union xfs_btree_key *key);
530 void xfs_btree_get_node_keys_overlapped(struct xfs_btree_cur *cur,
531 struct xfs_btree_block *block, union xfs_btree_key *key);
532 int xfs_btree_update_keys_overlapped(struct xfs_btree_cur *cur, int level);
535 #define XFS_BTREE_QUERY_RANGE_CONTINUE 0 /* keep iterating */
536 #define XFS_BTREE_QUERY_RANGE_ABORT 1 /* stop iterating */
537 typedef int (*xfs_btree_query_range_fn)(struct xfs_btree_cur *cur,
538 union xfs_btree_rec *rec, void *priv);
540 int xfs_btree_query_range(struct xfs_btree_cur *cur,
541 union xfs_btree_irec *low_rec, union xfs_btree_irec *high_rec,
542 xfs_btree_query_range_fn fn, void *priv);
544 typedef int (*xfs_btree_visit_blocks_fn)(struct xfs_btree_cur *cur, int level,
546 int xfs_btree_visit_blocks(struct xfs_btree_cur *cur,
547 xfs_btree_visit_blocks_fn fn, void *data);
549 #endif /* __XFS_BTREE_H__ */