1 /* 2 * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc. 3 * All Rights Reserved. 4 * 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. 8 * 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. 13 * 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 17 */ 18 #ifndef __XFS_BTREE_H__ 19 #define __XFS_BTREE_H__ 20 21 struct xfs_buf; 22 struct xfs_defer_ops; 23 struct xfs_inode; 24 struct xfs_mount; 25 struct xfs_trans; 26 27 extern kmem_zone_t *xfs_btree_cur_zone; 28 29 /* 30 * Generic key, ptr and record wrapper structures. 31 * 32 * These are disk format structures, and are converted where necessary 33 * by the btree specific code that needs to interpret them. 34 */ 35 union xfs_btree_ptr { 36 __be32 s; /* short form ptr */ 37 __be64 l; /* long form ptr */ 38 }; 39 40 union xfs_btree_key { 41 struct xfs_bmbt_key bmbt; 42 xfs_bmdr_key_t bmbr; /* bmbt root block */ 43 xfs_alloc_key_t alloc; 44 struct xfs_inobt_key inobt; 45 struct xfs_rmap_key rmap; 46 }; 47 48 /* 49 * In-core key that holds both low and high keys for overlapped btrees. 50 * The two keys are packed next to each other on disk, so do the same 51 * in memory. Preserve the existing xfs_btree_key as a single key to 52 * avoid the mental model breakage that would happen if we passed a 53 * bigkey into a function that operates on a single key. 54 */ 55 union xfs_btree_bigkey { 56 struct xfs_bmbt_key bmbt; 57 xfs_bmdr_key_t bmbr; /* bmbt root block */ 58 xfs_alloc_key_t alloc; 59 struct xfs_inobt_key inobt; 60 struct { 61 struct xfs_rmap_key rmap; 62 struct xfs_rmap_key rmap_hi; 63 }; 64 }; 65 66 union xfs_btree_rec { 67 struct xfs_bmbt_rec bmbt; 68 xfs_bmdr_rec_t bmbr; /* bmbt root block */ 69 struct xfs_alloc_rec alloc; 70 struct xfs_inobt_rec inobt; 71 struct xfs_rmap_rec rmap; 72 }; 73 74 /* 75 * This nonsense is to make -wlint happy. 76 */ 77 #define XFS_LOOKUP_EQ ((xfs_lookup_t)XFS_LOOKUP_EQi) 78 #define XFS_LOOKUP_LE ((xfs_lookup_t)XFS_LOOKUP_LEi) 79 #define XFS_LOOKUP_GE ((xfs_lookup_t)XFS_LOOKUP_GEi) 80 81 #define XFS_BTNUM_BNO ((xfs_btnum_t)XFS_BTNUM_BNOi) 82 #define XFS_BTNUM_CNT ((xfs_btnum_t)XFS_BTNUM_CNTi) 83 #define XFS_BTNUM_BMAP ((xfs_btnum_t)XFS_BTNUM_BMAPi) 84 #define XFS_BTNUM_INO ((xfs_btnum_t)XFS_BTNUM_INOi) 85 #define XFS_BTNUM_FINO ((xfs_btnum_t)XFS_BTNUM_FINOi) 86 #define XFS_BTNUM_RMAP ((xfs_btnum_t)XFS_BTNUM_RMAPi) 87 88 /* 89 * For logging record fields. 90 */ 91 #define XFS_BB_MAGIC (1 << 0) 92 #define XFS_BB_LEVEL (1 << 1) 93 #define XFS_BB_NUMRECS (1 << 2) 94 #define XFS_BB_LEFTSIB (1 << 3) 95 #define XFS_BB_RIGHTSIB (1 << 4) 96 #define XFS_BB_BLKNO (1 << 5) 97 #define XFS_BB_LSN (1 << 6) 98 #define XFS_BB_UUID (1 << 7) 99 #define XFS_BB_OWNER (1 << 8) 100 #define XFS_BB_NUM_BITS 5 101 #define XFS_BB_ALL_BITS ((1 << XFS_BB_NUM_BITS) - 1) 102 #define XFS_BB_NUM_BITS_CRC 9 103 #define XFS_BB_ALL_BITS_CRC ((1 << XFS_BB_NUM_BITS_CRC) - 1) 104 105 /* 106 * Generic stats interface 107 */ 108 #define __XFS_BTREE_STATS_INC(mp, type, stat) \ 109 XFS_STATS_INC(mp, xs_ ## type ## _2_ ## stat) 110 #define XFS_BTREE_STATS_INC(cur, stat) \ 111 do { \ 112 struct xfs_mount *__mp = cur->bc_mp; \ 113 switch (cur->bc_btnum) { \ 114 case XFS_BTNUM_BNO: __XFS_BTREE_STATS_INC(__mp, abtb, stat); break; \ 115 case XFS_BTNUM_CNT: __XFS_BTREE_STATS_INC(__mp, abtc, stat); break; \ 116 case XFS_BTNUM_BMAP: __XFS_BTREE_STATS_INC(__mp, bmbt, stat); break; \ 117 case XFS_BTNUM_INO: __XFS_BTREE_STATS_INC(__mp, ibt, stat); break; \ 118 case XFS_BTNUM_FINO: __XFS_BTREE_STATS_INC(__mp, fibt, stat); break; \ 119 case XFS_BTNUM_RMAP: __XFS_BTREE_STATS_INC(__mp, rmap, stat); break; \ 120 case XFS_BTNUM_MAX: ASSERT(0); /* fucking gcc */ ; break; \ 121 } \ 122 } while (0) 123 124 #define __XFS_BTREE_STATS_ADD(mp, type, stat, val) \ 125 XFS_STATS_ADD(mp, xs_ ## type ## _2_ ## stat, val) 126 #define XFS_BTREE_STATS_ADD(cur, stat, val) \ 127 do { \ 128 struct xfs_mount *__mp = cur->bc_mp; \ 129 switch (cur->bc_btnum) { \ 130 case XFS_BTNUM_BNO: \ 131 __XFS_BTREE_STATS_ADD(__mp, abtb, stat, val); break; \ 132 case XFS_BTNUM_CNT: \ 133 __XFS_BTREE_STATS_ADD(__mp, abtc, stat, val); break; \ 134 case XFS_BTNUM_BMAP: \ 135 __XFS_BTREE_STATS_ADD(__mp, bmbt, stat, val); break; \ 136 case XFS_BTNUM_INO: \ 137 __XFS_BTREE_STATS_ADD(__mp, ibt, stat, val); break; \ 138 case XFS_BTNUM_FINO: \ 139 __XFS_BTREE_STATS_ADD(__mp, fibt, stat, val); break; \ 140 case XFS_BTNUM_RMAP: \ 141 __XFS_BTREE_STATS_ADD(__mp, rmap, stat, val); break; \ 142 case XFS_BTNUM_MAX: ASSERT(0); /* fucking gcc */ ; break; \ 143 } \ 144 } while (0) 145 146 #define XFS_BTREE_MAXLEVELS 9 /* max of all btrees */ 147 148 struct xfs_btree_ops { 149 /* size of the key and record structures */ 150 size_t key_len; 151 size_t rec_len; 152 153 /* cursor operations */ 154 struct xfs_btree_cur *(*dup_cursor)(struct xfs_btree_cur *); 155 void (*update_cursor)(struct xfs_btree_cur *src, 156 struct xfs_btree_cur *dst); 157 158 /* update btree root pointer */ 159 void (*set_root)(struct xfs_btree_cur *cur, 160 union xfs_btree_ptr *nptr, int level_change); 161 162 /* block allocation / freeing */ 163 int (*alloc_block)(struct xfs_btree_cur *cur, 164 union xfs_btree_ptr *start_bno, 165 union xfs_btree_ptr *new_bno, 166 int *stat); 167 int (*free_block)(struct xfs_btree_cur *cur, struct xfs_buf *bp); 168 169 /* update last record information */ 170 void (*update_lastrec)(struct xfs_btree_cur *cur, 171 struct xfs_btree_block *block, 172 union xfs_btree_rec *rec, 173 int ptr, int reason); 174 175 /* records in block/level */ 176 int (*get_minrecs)(struct xfs_btree_cur *cur, int level); 177 int (*get_maxrecs)(struct xfs_btree_cur *cur, int level); 178 179 /* records on disk. Matter for the root in inode case. */ 180 int (*get_dmaxrecs)(struct xfs_btree_cur *cur, int level); 181 182 /* init values of btree structures */ 183 void (*init_key_from_rec)(union xfs_btree_key *key, 184 union xfs_btree_rec *rec); 185 void (*init_rec_from_cur)(struct xfs_btree_cur *cur, 186 union xfs_btree_rec *rec); 187 void (*init_ptr_from_cur)(struct xfs_btree_cur *cur, 188 union xfs_btree_ptr *ptr); 189 void (*init_high_key_from_rec)(union xfs_btree_key *key, 190 union xfs_btree_rec *rec); 191 192 /* difference between key value and cursor value */ 193 __int64_t (*key_diff)(struct xfs_btree_cur *cur, 194 union xfs_btree_key *key); 195 196 /* 197 * Difference between key2 and key1 -- positive if key1 > key2, 198 * negative if key1 < key2, and zero if equal. 199 */ 200 __int64_t (*diff_two_keys)(struct xfs_btree_cur *cur, 201 union xfs_btree_key *key1, 202 union xfs_btree_key *key2); 203 204 const struct xfs_buf_ops *buf_ops; 205 206 #if defined(DEBUG) || defined(XFS_WARN) 207 /* check that k1 is lower than k2 */ 208 int (*keys_inorder)(struct xfs_btree_cur *cur, 209 union xfs_btree_key *k1, 210 union xfs_btree_key *k2); 211 212 /* check that r1 is lower than r2 */ 213 int (*recs_inorder)(struct xfs_btree_cur *cur, 214 union xfs_btree_rec *r1, 215 union xfs_btree_rec *r2); 216 #endif 217 }; 218 219 /* 220 * Reasons for the update_lastrec method to be called. 221 */ 222 #define LASTREC_UPDATE 0 223 #define LASTREC_INSREC 1 224 #define LASTREC_DELREC 2 225 226 227 union xfs_btree_irec { 228 struct xfs_alloc_rec_incore a; 229 struct xfs_bmbt_irec b; 230 struct xfs_inobt_rec_incore i; 231 struct xfs_rmap_irec r; 232 }; 233 234 /* 235 * Btree cursor structure. 236 * This collects all information needed by the btree code in one place. 237 */ 238 typedef struct xfs_btree_cur 239 { 240 struct xfs_trans *bc_tp; /* transaction we're in, if any */ 241 struct xfs_mount *bc_mp; /* file system mount struct */ 242 const struct xfs_btree_ops *bc_ops; 243 uint bc_flags; /* btree features - below */ 244 union xfs_btree_irec bc_rec; /* current insert/search record value */ 245 struct xfs_buf *bc_bufs[XFS_BTREE_MAXLEVELS]; /* buf ptr per level */ 246 int bc_ptrs[XFS_BTREE_MAXLEVELS]; /* key/record # */ 247 __uint8_t bc_ra[XFS_BTREE_MAXLEVELS]; /* readahead bits */ 248 #define XFS_BTCUR_LEFTRA 1 /* left sibling has been read-ahead */ 249 #define XFS_BTCUR_RIGHTRA 2 /* right sibling has been read-ahead */ 250 __uint8_t bc_nlevels; /* number of levels in the tree */ 251 __uint8_t bc_blocklog; /* log2(blocksize) of btree blocks */ 252 xfs_btnum_t bc_btnum; /* identifies which btree type */ 253 union { 254 struct { /* needed for BNO, CNT, INO */ 255 struct xfs_buf *agbp; /* agf/agi buffer pointer */ 256 struct xfs_defer_ops *dfops; /* deferred updates */ 257 xfs_agnumber_t agno; /* ag number */ 258 } a; 259 struct { /* needed for BMAP */ 260 struct xfs_inode *ip; /* pointer to our inode */ 261 struct xfs_defer_ops *dfops; /* deferred updates */ 262 xfs_fsblock_t firstblock; /* 1st blk allocated */ 263 int allocated; /* count of alloced */ 264 short forksize; /* fork's inode space */ 265 char whichfork; /* data or attr fork */ 266 char flags; /* flags */ 267 #define XFS_BTCUR_BPRV_WASDEL 1 /* was delayed */ 268 } b; 269 } bc_private; /* per-btree type data */ 270 } xfs_btree_cur_t; 271 272 /* cursor flags */ 273 #define XFS_BTREE_LONG_PTRS (1<<0) /* pointers are 64bits long */ 274 #define XFS_BTREE_ROOT_IN_INODE (1<<1) /* root may be variable size */ 275 #define XFS_BTREE_LASTREC_UPDATE (1<<2) /* track last rec externally */ 276 #define XFS_BTREE_CRC_BLOCKS (1<<3) /* uses extended btree blocks */ 277 #define XFS_BTREE_OVERLAPPING (1<<4) /* overlapping intervals */ 278 279 280 #define XFS_BTREE_NOERROR 0 281 #define XFS_BTREE_ERROR 1 282 283 /* 284 * Convert from buffer to btree block header. 285 */ 286 #define XFS_BUF_TO_BLOCK(bp) ((struct xfs_btree_block *)((bp)->b_addr)) 287 288 289 /* 290 * Check that block header is ok. 291 */ 292 int 293 xfs_btree_check_block( 294 struct xfs_btree_cur *cur, /* btree cursor */ 295 struct xfs_btree_block *block, /* generic btree block pointer */ 296 int level, /* level of the btree block */ 297 struct xfs_buf *bp); /* buffer containing block, if any */ 298 299 /* 300 * Check that (long) pointer is ok. 301 */ 302 int /* error (0 or EFSCORRUPTED) */ 303 xfs_btree_check_lptr( 304 struct xfs_btree_cur *cur, /* btree cursor */ 305 xfs_fsblock_t ptr, /* btree block disk address */ 306 int level); /* btree block level */ 307 308 /* 309 * Delete the btree cursor. 310 */ 311 void 312 xfs_btree_del_cursor( 313 xfs_btree_cur_t *cur, /* btree cursor */ 314 int error); /* del because of error */ 315 316 /* 317 * Duplicate the btree cursor. 318 * Allocate a new one, copy the record, re-get the buffers. 319 */ 320 int /* error */ 321 xfs_btree_dup_cursor( 322 xfs_btree_cur_t *cur, /* input cursor */ 323 xfs_btree_cur_t **ncur);/* output cursor */ 324 325 /* 326 * Get a buffer for the block, return it with no data read. 327 * Long-form addressing. 328 */ 329 struct xfs_buf * /* buffer for fsbno */ 330 xfs_btree_get_bufl( 331 struct xfs_mount *mp, /* file system mount point */ 332 struct xfs_trans *tp, /* transaction pointer */ 333 xfs_fsblock_t fsbno, /* file system block number */ 334 uint lock); /* lock flags for get_buf */ 335 336 /* 337 * Get a buffer for the block, return it with no data read. 338 * Short-form addressing. 339 */ 340 struct xfs_buf * /* buffer for agno/agbno */ 341 xfs_btree_get_bufs( 342 struct xfs_mount *mp, /* file system mount point */ 343 struct xfs_trans *tp, /* transaction pointer */ 344 xfs_agnumber_t agno, /* allocation group number */ 345 xfs_agblock_t agbno, /* allocation group block number */ 346 uint lock); /* lock flags for get_buf */ 347 348 /* 349 * Check for the cursor referring to the last block at the given level. 350 */ 351 int /* 1=is last block, 0=not last block */ 352 xfs_btree_islastblock( 353 xfs_btree_cur_t *cur, /* btree cursor */ 354 int level); /* level to check */ 355 356 /* 357 * Compute first and last byte offsets for the fields given. 358 * Interprets the offsets table, which contains struct field offsets. 359 */ 360 void 361 xfs_btree_offsets( 362 __int64_t fields, /* bitmask of fields */ 363 const short *offsets,/* table of field offsets */ 364 int nbits, /* number of bits to inspect */ 365 int *first, /* output: first byte offset */ 366 int *last); /* output: last byte offset */ 367 368 /* 369 * Get a buffer for the block, return it read in. 370 * Long-form addressing. 371 */ 372 int /* error */ 373 xfs_btree_read_bufl( 374 struct xfs_mount *mp, /* file system mount point */ 375 struct xfs_trans *tp, /* transaction pointer */ 376 xfs_fsblock_t fsbno, /* file system block number */ 377 uint lock, /* lock flags for read_buf */ 378 struct xfs_buf **bpp, /* buffer for fsbno */ 379 int refval, /* ref count value for buffer */ 380 const struct xfs_buf_ops *ops); 381 382 /* 383 * Read-ahead the block, don't wait for it, don't return a buffer. 384 * Long-form addressing. 385 */ 386 void /* error */ 387 xfs_btree_reada_bufl( 388 struct xfs_mount *mp, /* file system mount point */ 389 xfs_fsblock_t fsbno, /* file system block number */ 390 xfs_extlen_t count, /* count of filesystem blocks */ 391 const struct xfs_buf_ops *ops); 392 393 /* 394 * Read-ahead the block, don't wait for it, don't return a buffer. 395 * Short-form addressing. 396 */ 397 void /* error */ 398 xfs_btree_reada_bufs( 399 struct xfs_mount *mp, /* file system mount point */ 400 xfs_agnumber_t agno, /* allocation group number */ 401 xfs_agblock_t agbno, /* allocation group block number */ 402 xfs_extlen_t count, /* count of filesystem blocks */ 403 const struct xfs_buf_ops *ops); 404 405 /* 406 * Initialise a new btree block header 407 */ 408 void 409 xfs_btree_init_block( 410 struct xfs_mount *mp, 411 struct xfs_buf *bp, 412 __u32 magic, 413 __u16 level, 414 __u16 numrecs, 415 __u64 owner, 416 unsigned int flags); 417 418 void 419 xfs_btree_init_block_int( 420 struct xfs_mount *mp, 421 struct xfs_btree_block *buf, 422 xfs_daddr_t blkno, 423 __u32 magic, 424 __u16 level, 425 __u16 numrecs, 426 __u64 owner, 427 unsigned int flags); 428 429 /* 430 * Common btree core entry points. 431 */ 432 int xfs_btree_increment(struct xfs_btree_cur *, int, int *); 433 int xfs_btree_decrement(struct xfs_btree_cur *, int, int *); 434 int xfs_btree_lookup(struct xfs_btree_cur *, xfs_lookup_t, int *); 435 int xfs_btree_update(struct xfs_btree_cur *, union xfs_btree_rec *); 436 int xfs_btree_new_iroot(struct xfs_btree_cur *, int *, int *); 437 int xfs_btree_insert(struct xfs_btree_cur *, int *); 438 int xfs_btree_delete(struct xfs_btree_cur *, int *); 439 int xfs_btree_get_rec(struct xfs_btree_cur *, union xfs_btree_rec **, int *); 440 int xfs_btree_change_owner(struct xfs_btree_cur *cur, __uint64_t new_owner, 441 struct list_head *buffer_list); 442 443 /* 444 * btree block CRC helpers 445 */ 446 void xfs_btree_lblock_calc_crc(struct xfs_buf *); 447 bool xfs_btree_lblock_verify_crc(struct xfs_buf *); 448 void xfs_btree_sblock_calc_crc(struct xfs_buf *); 449 bool xfs_btree_sblock_verify_crc(struct xfs_buf *); 450 451 /* 452 * Internal btree helpers also used by xfs_bmap.c. 453 */ 454 void xfs_btree_log_block(struct xfs_btree_cur *, struct xfs_buf *, int); 455 void xfs_btree_log_recs(struct xfs_btree_cur *, struct xfs_buf *, int, int); 456 457 /* 458 * Helpers. 459 */ 460 static inline int xfs_btree_get_numrecs(struct xfs_btree_block *block) 461 { 462 return be16_to_cpu(block->bb_numrecs); 463 } 464 465 static inline void xfs_btree_set_numrecs(struct xfs_btree_block *block, 466 __uint16_t numrecs) 467 { 468 block->bb_numrecs = cpu_to_be16(numrecs); 469 } 470 471 static inline int xfs_btree_get_level(struct xfs_btree_block *block) 472 { 473 return be16_to_cpu(block->bb_level); 474 } 475 476 477 /* 478 * Min and max functions for extlen, agblock, fileoff, and filblks types. 479 */ 480 #define XFS_EXTLEN_MIN(a,b) min_t(xfs_extlen_t, (a), (b)) 481 #define XFS_EXTLEN_MAX(a,b) max_t(xfs_extlen_t, (a), (b)) 482 #define XFS_AGBLOCK_MIN(a,b) min_t(xfs_agblock_t, (a), (b)) 483 #define XFS_AGBLOCK_MAX(a,b) max_t(xfs_agblock_t, (a), (b)) 484 #define XFS_FILEOFF_MIN(a,b) min_t(xfs_fileoff_t, (a), (b)) 485 #define XFS_FILEOFF_MAX(a,b) max_t(xfs_fileoff_t, (a), (b)) 486 #define XFS_FILBLKS_MIN(a,b) min_t(xfs_filblks_t, (a), (b)) 487 #define XFS_FILBLKS_MAX(a,b) max_t(xfs_filblks_t, (a), (b)) 488 489 #define XFS_FSB_SANITY_CHECK(mp,fsb) \ 490 (XFS_FSB_TO_AGNO(mp, fsb) < mp->m_sb.sb_agcount && \ 491 XFS_FSB_TO_AGBNO(mp, fsb) < mp->m_sb.sb_agblocks) 492 493 /* 494 * Trace hooks. Currently not implemented as they need to be ported 495 * over to the generic tracing functionality, which is some effort. 496 * 497 * i,j = integer (32 bit) 498 * b = btree block buffer (xfs_buf_t) 499 * p = btree ptr 500 * r = btree record 501 * k = btree key 502 */ 503 #define XFS_BTREE_TRACE_ARGBI(c, b, i) 504 #define XFS_BTREE_TRACE_ARGBII(c, b, i, j) 505 #define XFS_BTREE_TRACE_ARGI(c, i) 506 #define XFS_BTREE_TRACE_ARGIPK(c, i, p, s) 507 #define XFS_BTREE_TRACE_ARGIPR(c, i, p, r) 508 #define XFS_BTREE_TRACE_ARGIK(c, i, k) 509 #define XFS_BTREE_TRACE_ARGR(c, r) 510 #define XFS_BTREE_TRACE_CURSOR(c, t) 511 512 bool xfs_btree_sblock_v5hdr_verify(struct xfs_buf *bp); 513 bool xfs_btree_sblock_verify(struct xfs_buf *bp, unsigned int max_recs); 514 uint xfs_btree_compute_maxlevels(struct xfs_mount *mp, uint *limits, 515 unsigned long len); 516 517 /* return codes */ 518 #define XFS_BTREE_QUERY_RANGE_CONTINUE 0 /* keep iterating */ 519 #define XFS_BTREE_QUERY_RANGE_ABORT 1 /* stop iterating */ 520 typedef int (*xfs_btree_query_range_fn)(struct xfs_btree_cur *cur, 521 union xfs_btree_rec *rec, void *priv); 522 523 int xfs_btree_query_range(struct xfs_btree_cur *cur, 524 union xfs_btree_irec *low_rec, union xfs_btree_irec *high_rec, 525 xfs_btree_query_range_fn fn, void *priv); 526 527 typedef int (*xfs_btree_visit_blocks_fn)(struct xfs_btree_cur *cur, int level, 528 void *data); 529 int xfs_btree_visit_blocks(struct xfs_btree_cur *cur, 530 xfs_btree_visit_blocks_fn fn, void *data); 531 532 #endif /* __XFS_BTREE_H__ */ 533