fs/xfs/libxfs/xfs_btree.c

   1 /*
   2  * Copyright (c) 2000-2002,2005 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 #include "xfs.h"
  19 #include "xfs_fs.h"
  20 #include "xfs_shared.h"
  21 #include "xfs_format.h"
  22 #include "xfs_log_format.h"
  23 #include "xfs_trans_resv.h"
  24 #include "xfs_bit.h"
  25 #include "xfs_mount.h"
  26 #include "xfs_defer.h"
  27 #include "xfs_inode.h"
  28 #include "xfs_trans.h"
  29 #include "xfs_inode_item.h"
  30 #include "xfs_buf_item.h"
  31 #include "xfs_btree.h"
  32 #include "xfs_error.h"
  33 #include "xfs_trace.h"
  34 #include "xfs_cksum.h"
  35 #include "xfs_alloc.h"
  36 #include "xfs_log.h"
  37
  38 /*
  39  * Cursor allocation zone.
  40  */
  41 kmem_zone_t     *xfs_btree_cur_zone;
  42
  43 /*
  44  * Btree magic numbers.
  45  */
  46 static const uint32_t xfs_magics[2][XFS_BTNUM_MAX] = {
  47         { XFS_ABTB_MAGIC, XFS_ABTC_MAGIC, 0, XFS_BMAP_MAGIC, XFS_IBT_MAGIC,
  48           XFS_FIBT_MAGIC, 0 },
  49         { XFS_ABTB_CRC_MAGIC, XFS_ABTC_CRC_MAGIC, XFS_RMAP_CRC_MAGIC,
  50           XFS_BMAP_CRC_MAGIC, XFS_IBT_CRC_MAGIC, XFS_FIBT_CRC_MAGIC,
  51           XFS_REFC_CRC_MAGIC }
  52 };
  53
  54 uint32_t
  55 xfs_btree_magic(
  56         int                     crc,
  57         xfs_btnum_t             btnum)
  58 {
  59         uint32_t                magic = xfs_magics[crc][btnum];
  60
  61         /* Ensure we asked for crc for crc-only magics. */
  62         ASSERT(magic != 0);
  63         return magic;
  64 }
  65
  66 STATIC int                              /* error (0 or EFSCORRUPTED) */
  67 xfs_btree_check_lblock(
  68         struct xfs_btree_cur    *cur,   /* btree cursor */
  69         struct xfs_btree_block  *block, /* btree long form block pointer */
  70         int                     level,  /* level of the btree block */
  71         struct xfs_buf          *bp)    /* buffer for block, if any */
  72 {
  73         int                     lblock_ok = 1; /* block passes checks */
  74         struct xfs_mount        *mp;    /* file system mount point */
  75         xfs_btnum_t             btnum = cur->bc_btnum;
  76         int                     crc;
  77
  78         mp = cur->bc_mp;
  79         crc = xfs_sb_version_hascrc(&mp->m_sb);
  80
  81         if (crc) {
  82                 lblock_ok = lblock_ok &&
  83                         uuid_equal(&block->bb_u.l.bb_uuid,
  84                                    &mp->m_sb.sb_meta_uuid) &&
  85                         block->bb_u.l.bb_blkno == cpu_to_be64(
  86                                 bp ? bp->b_bn : XFS_BUF_DADDR_NULL);
  87         }
  88
  89         lblock_ok = lblock_ok &&
  90                 be32_to_cpu(block->bb_magic) == xfs_btree_magic(crc, btnum) &&
  91                 be16_to_cpu(block->bb_level) == level &&
  92                 be16_to_cpu(block->bb_numrecs) <=
  93                         cur->bc_ops->get_maxrecs(cur, level) &&
  94                 block->bb_u.l.bb_leftsib &&
  95                 (block->bb_u.l.bb_leftsib == cpu_to_be64(NULLFSBLOCK) ||
  96                  XFS_FSB_SANITY_CHECK(mp,
  97                         be64_to_cpu(block->bb_u.l.bb_leftsib))) &&
  98                 block->bb_u.l.bb_rightsib &&
  99                 (block->bb_u.l.bb_rightsib == cpu_to_be64(NULLFSBLOCK) ||
 100                  XFS_FSB_SANITY_CHECK(mp,
 101                         be64_to_cpu(block->bb_u.l.bb_rightsib)));
 102
 103         if (unlikely(XFS_TEST_ERROR(!lblock_ok, mp,
 104                         XFS_ERRTAG_BTREE_CHECK_LBLOCK))) {
 105                 if (bp)
 106                         trace_xfs_btree_corrupt(bp, _RET_IP_);
 107                 XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, mp);
 108                 return -EFSCORRUPTED;
 109         }
 110         return 0;
 111 }
 112
 113 STATIC int                              /* error (0 or EFSCORRUPTED) */
 114 xfs_btree_check_sblock(
 115         struct xfs_btree_cur    *cur,   /* btree cursor */
 116         struct xfs_btree_block  *block, /* btree short form block pointer */
 117         int                     level,  /* level of the btree block */
 118         struct xfs_buf          *bp)    /* buffer containing block */
 119 {
 120         struct xfs_mount        *mp;    /* file system mount point */
 121         struct xfs_buf          *agbp;  /* buffer for ag. freespace struct */
 122         struct xfs_agf          *agf;   /* ag. freespace structure */
 123         xfs_agblock_t           agflen; /* native ag. freespace length */
 124         int                     sblock_ok = 1; /* block passes checks */
 125         xfs_btnum_t             btnum = cur->bc_btnum;
 126         int                     crc;
 127
 128         mp = cur->bc_mp;
 129         crc = xfs_sb_version_hascrc(&mp->m_sb);
 130         agbp = cur->bc_private.a.agbp;
 131         agf = XFS_BUF_TO_AGF(agbp);
 132         agflen = be32_to_cpu(agf->agf_length);
 133
 134         if (crc) {
 135                 sblock_ok = sblock_ok &&
 136                         uuid_equal(&block->bb_u.s.bb_uuid,
 137                                    &mp->m_sb.sb_meta_uuid) &&
 138                         block->bb_u.s.bb_blkno == cpu_to_be64(
 139                                 bp ? bp->b_bn : XFS_BUF_DADDR_NULL);
 140         }
 141
 142         sblock_ok = sblock_ok &&
 143                 be32_to_cpu(block->bb_magic) == xfs_btree_magic(crc, btnum) &&
 144                 be16_to_cpu(block->bb_level) == level &&
 145                 be16_to_cpu(block->bb_numrecs) <=
 146                         cur->bc_ops->get_maxrecs(cur, level) &&
 147                 (block->bb_u.s.bb_leftsib == cpu_to_be32(NULLAGBLOCK) ||
 148                  be32_to_cpu(block->bb_u.s.bb_leftsib) < agflen) &&
 149                 block->bb_u.s.bb_leftsib &&
 150                 (block->bb_u.s.bb_rightsib == cpu_to_be32(NULLAGBLOCK) ||
 151                  be32_to_cpu(block->bb_u.s.bb_rightsib) < agflen) &&
 152                 block->bb_u.s.bb_rightsib;
 153
 154         if (unlikely(XFS_TEST_ERROR(!sblock_ok, mp,
 155                         XFS_ERRTAG_BTREE_CHECK_SBLOCK))) {
 156                 if (bp)
 157                         trace_xfs_btree_corrupt(bp, _RET_IP_);
 158                 XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, mp);
 159                 return -EFSCORRUPTED;
 160         }
 161         return 0;
 162 }
 163
 164 /*
 165  * Debug routine: check that block header is ok.
 166  */
 167 int
 168 xfs_btree_check_block(
 169         struct xfs_btree_cur    *cur,   /* btree cursor */
 170         struct xfs_btree_block  *block, /* generic btree block pointer */
 171         int                     level,  /* level of the btree block */
 172         struct xfs_buf          *bp)    /* buffer containing block, if any */
 173 {
 174         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
 175                 return xfs_btree_check_lblock(cur, block, level, bp);
 176         else
 177                 return xfs_btree_check_sblock(cur, block, level, bp);
 178 }
 179
 180 /*
 181  * Check that (long) pointer is ok.
 182  */
 183 int                                     /* error (0 or EFSCORRUPTED) */
 184 xfs_btree_check_lptr(
 185         struct xfs_btree_cur    *cur,   /* btree cursor */
 186         xfs_fsblock_t           bno,    /* btree block disk address */
 187         int                     level)  /* btree block level */
 188 {
 189         XFS_WANT_CORRUPTED_RETURN(cur->bc_mp,
 190                 level > 0 &&
 191                 bno != NULLFSBLOCK &&
 192                 XFS_FSB_SANITY_CHECK(cur->bc_mp, bno));
 193         return 0;
 194 }
 195
 196 #ifdef DEBUG
 197 /*
 198  * Check that (short) pointer is ok.
 199  */
 200 STATIC int                              /* error (0 or EFSCORRUPTED) */
 201 xfs_btree_check_sptr(
 202         struct xfs_btree_cur    *cur,   /* btree cursor */
 203         xfs_agblock_t           bno,    /* btree block disk address */
 204         int                     level)  /* btree block level */
 205 {
 206         xfs_agblock_t           agblocks = cur->bc_mp->m_sb.sb_agblocks;
 207
 208         XFS_WANT_CORRUPTED_RETURN(cur->bc_mp,
 209                 level > 0 &&
 210                 bno != NULLAGBLOCK &&
 211                 bno != 0 &&
 212                 bno < agblocks);
 213         return 0;
 214 }
 215
 216 /*
 217  * Check that block ptr is ok.
 218  */
 219 STATIC int                              /* error (0 or EFSCORRUPTED) */
 220 xfs_btree_check_ptr(
 221         struct xfs_btree_cur    *cur,   /* btree cursor */
 222         union xfs_btree_ptr     *ptr,   /* btree block disk address */
 223         int                     index,  /* offset from ptr to check */
 224         int                     level)  /* btree block level */
 225 {
 226         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
 227                 return xfs_btree_check_lptr(cur,
 228                                 be64_to_cpu((&ptr->l)[index]), level);
 229         } else {
 230                 return xfs_btree_check_sptr(cur,
 231                                 be32_to_cpu((&ptr->s)[index]), level);
 232         }
 233 }
 234 #endif
 235
 236 /*
 237  * Calculate CRC on the whole btree block and stuff it into the
 238  * long-form btree header.
 239  *
 240  * Prior to calculting the CRC, pull the LSN out of the buffer log item and put
 241  * it into the buffer so recovery knows what the last modification was that made
 242  * it to disk.
 243  */
 244 void
 245 xfs_btree_lblock_calc_crc(
 246         struct xfs_buf          *bp)
 247 {
 248         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
 249         struct xfs_buf_log_item *bip = bp->b_fspriv;
 250
 251         if (!xfs_sb_version_hascrc(&bp->b_target->bt_mount->m_sb))
 252                 return;
 253         if (bip)
 254                 block->bb_u.l.bb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
 255         xfs_buf_update_cksum(bp, XFS_BTREE_LBLOCK_CRC_OFF);
 256 }
 257
 258 bool
 259 xfs_btree_lblock_verify_crc(
 260         struct xfs_buf          *bp)
 261 {
 262         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
 263         struct xfs_mount        *mp = bp->b_target->bt_mount;
 264
 265         if (xfs_sb_version_hascrc(&mp->m_sb)) {
 266                 if (!xfs_log_check_lsn(mp, be64_to_cpu(block->bb_u.l.bb_lsn)))
 267                         return false;
 268                 return xfs_buf_verify_cksum(bp, XFS_BTREE_LBLOCK_CRC_OFF);
 269         }
 270
 271         return true;
 272 }
 273
 274 /*
 275  * Calculate CRC on the whole btree block and stuff it into the
 276  * short-form btree header.
 277  *
 278  * Prior to calculting the CRC, pull the LSN out of the buffer log item and put
 279  * it into the buffer so recovery knows what the last modification was that made
 280  * it to disk.
 281  */
 282 void
 283 xfs_btree_sblock_calc_crc(
 284         struct xfs_buf          *bp)
 285 {
 286         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
 287         struct xfs_buf_log_item *bip = bp->b_fspriv;
 288
 289         if (!xfs_sb_version_hascrc(&bp->b_target->bt_mount->m_sb))
 290                 return;
 291         if (bip)
 292                 block->bb_u.s.bb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
 293         xfs_buf_update_cksum(bp, XFS_BTREE_SBLOCK_CRC_OFF);
 294 }
 295
 296 bool
 297 xfs_btree_sblock_verify_crc(
 298         struct xfs_buf          *bp)
 299 {
 300         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
 301         struct xfs_mount        *mp = bp->b_target->bt_mount;
 302
 303         if (xfs_sb_version_hascrc(&mp->m_sb)) {
 304                 if (!xfs_log_check_lsn(mp, be64_to_cpu(block->bb_u.s.bb_lsn)))
 305                         return false;
 306                 return xfs_buf_verify_cksum(bp, XFS_BTREE_SBLOCK_CRC_OFF);
 307         }
 308
 309         return true;
 310 }
 311
 312 static int
 313 xfs_btree_free_block(
 314         struct xfs_btree_cur    *cur,
 315         struct xfs_buf          *bp)
 316 {
 317         int                     error;
 318
 319         error = cur->bc_ops->free_block(cur, bp);
 320         if (!error) {
 321                 xfs_trans_binval(cur->bc_tp, bp);
 322                 XFS_BTREE_STATS_INC(cur, free);
 323         }
 324         return error;
 325 }
 326
 327 /*
 328  * Delete the btree cursor.
 329  */
 330 void
 331 xfs_btree_del_cursor(
 332         xfs_btree_cur_t *cur,           /* btree cursor */
 333         int             error)          /* del because of error */
 334 {
 335         int             i;              /* btree level */
 336
 337         /*
 338          * Clear the buffer pointers, and release the buffers.
 339          * If we're doing this in the face of an error, we
 340          * need to make sure to inspect all of the entries
 341          * in the bc_bufs array for buffers to be unlocked.
 342          * This is because some of the btree code works from
 343          * level n down to 0, and if we get an error along
 344          * the way we won't have initialized all the entries
 345          * down to 0.
 346          */
 347         for (i = 0; i < cur->bc_nlevels; i++) {
 348                 if (cur->bc_bufs[i])
 349                         xfs_trans_brelse(cur->bc_tp, cur->bc_bufs[i]);
 350                 else if (!error)
 351                         break;
 352         }
 353         /*
 354          * Can't free a bmap cursor without having dealt with the
 355          * allocated indirect blocks' accounting.
 356          */
 357         ASSERT(cur->bc_btnum != XFS_BTNUM_BMAP ||
 358                cur->bc_private.b.allocated == 0);
 359         /*
 360          * Free the cursor.
 361          */
 362         kmem_zone_free(xfs_btree_cur_zone, cur);
 363 }
 364
 365 /*
 366  * Duplicate the btree cursor.
 367  * Allocate a new one, copy the record, re-get the buffers.
 368  */
 369 int                                     /* error */
 370 xfs_btree_dup_cursor(
 371         xfs_btree_cur_t *cur,           /* input cursor */
 372         xfs_btree_cur_t **ncur)         /* output cursor */
 373 {
 374         xfs_buf_t       *bp;            /* btree block's buffer pointer */
 375         int             error;          /* error return value */
 376         int             i;              /* level number of btree block */
 377         xfs_mount_t     *mp;            /* mount structure for filesystem */
 378         xfs_btree_cur_t *new;           /* new cursor value */
 379         xfs_trans_t     *tp;            /* transaction pointer, can be NULL */
 380
 381         tp = cur->bc_tp;
 382         mp = cur->bc_mp;
 383
 384         /*
 385          * Allocate a new cursor like the old one.
 386          */
 387         new = cur->bc_ops->dup_cursor(cur);
 388
 389         /*
 390          * Copy the record currently in the cursor.
 391          */
 392         new->bc_rec = cur->bc_rec;
 393
 394         /*
 395          * For each level current, re-get the buffer and copy the ptr value.
 396          */
 397         for (i = 0; i < new->bc_nlevels; i++) {
 398                 new->bc_ptrs[i] = cur->bc_ptrs[i];
 399                 new->bc_ra[i] = cur->bc_ra[i];
 400                 bp = cur->bc_bufs[i];
 401                 if (bp) {
 402                         error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
 403                                                    XFS_BUF_ADDR(bp), mp->m_bsize,
 404                                                    0, &bp,
 405                                                    cur->bc_ops->buf_ops);
 406                         if (error) {
 407                                 xfs_btree_del_cursor(new, error);
 408                                 *ncur = NULL;
 409                                 return error;
 410                         }
 411                 }
 412                 new->bc_bufs[i] = bp;
 413         }
 414         *ncur = new;
 415         return 0;
 416 }
 417
 418 /*
 419  * XFS btree block layout and addressing:
 420  *
 421  * There are two types of blocks in the btree: leaf and non-leaf blocks.
 422  *
 423  * The leaf record start with a header then followed by records containing
 424  * the values.  A non-leaf block also starts with the same header, and
 425  * then first contains lookup keys followed by an equal number of pointers
 426  * to the btree blocks at the previous level.
 427  *
 428  *              +--------+-------+-------+-------+-------+-------+-------+
 429  * Leaf:        | header | rec 1 | rec 2 | rec 3 | rec 4 | rec 5 | rec N |
 430  *              +--------+-------+-------+-------+-------+-------+-------+
 431  *
 432  *              +--------+-------+-------+-------+-------+-------+-------+
 433  * Non-Leaf:    | header | key 1 | key 2 | key N | ptr 1 | ptr 2 | ptr N |
 434  *              +--------+-------+-------+-------+-------+-------+-------+
 435  *
 436  * The header is called struct xfs_btree_block for reasons better left unknown
 437  * and comes in different versions for short (32bit) and long (64bit) block
 438  * pointers.  The record and key structures are defined by the btree instances
 439  * and opaque to the btree core.  The block pointers are simple disk endian
 440  * integers, available in a short (32bit) and long (64bit) variant.
 441  *
 442  * The helpers below calculate the offset of a given record, key or pointer
 443  * into a btree block (xfs_btree_*_offset) or return a pointer to the given
 444  * record, key or pointer (xfs_btree_*_addr).  Note that all addressing
 445  * inside the btree block is done using indices starting at one, not zero!
 446  *
 447  * If XFS_BTREE_OVERLAPPING is set, then this btree supports keys containing
 448  * overlapping intervals.  In such a tree, records are still sorted lowest to
 449  * highest and indexed by the smallest key value that refers to the record.
 450  * However, nodes are different: each pointer has two associated keys -- one
 451  * indexing the lowest key available in the block(s) below (the same behavior
 452  * as the key in a regular btree) and another indexing the highest key
 453  * available in the block(s) below.  Because records are /not/ sorted by the
 454  * highest key, all leaf block updates require us to compute the highest key
 455  * that matches any record in the leaf and to recursively update the high keys
 456  * in the nodes going further up in the tree, if necessary.  Nodes look like
 457  * this:
 458  *
 459  *              +--------+-----+-----+-----+-----+-----+-------+-------+-----+
 460  * Non-Leaf:    | header | lo1 | hi1 | lo2 | hi2 | ... | ptr 1 | ptr 2 | ... |
 461  *              +--------+-----+-----+-----+-----+-----+-------+-------+-----+
 462  *
 463  * To perform an interval query on an overlapped tree, perform the usual
 464  * depth-first search and use the low and high keys to decide if we can skip
 465  * that particular node.  If a leaf node is reached, return the records that
 466  * intersect the interval.  Note that an interval query may return numerous
 467  * entries.  For a non-overlapped tree, simply search for the record associated
 468  * with the lowest key and iterate forward until a non-matching record is
 469  * found.  Section 14.3 ("Interval Trees") of _Introduction to Algorithms_ by
 470  * Cormen, Leiserson, Rivest, and Stein (2nd or 3rd ed. only) discuss this in
 471  * more detail.
 472  *
 473  * Why do we care about overlapping intervals?  Let's say you have a bunch of
 474  * reverse mapping records on a reflink filesystem:
 475  *
 476  * 1: +- file A startblock B offset C length D -----------+
 477  * 2:      +- file E startblock F offset G length H --------------+
 478  * 3:      +- file I startblock F offset J length K --+
 479  * 4:                                                        +- file L... --+
 480  *
 481  * Now say we want to map block (B+D) into file A at offset (C+D).  Ideally,
 482  * we'd simply increment the length of record 1.  But how do we find the record
 483  * that ends at (B+D-1) (i.e. record 1)?  A LE lookup of (B+D-1) would return
 484  * record 3 because the keys are ordered first by startblock.  An interval
 485  * query would return records 1 and 2 because they both overlap (B+D-1), and
 486  * from that we can pick out record 1 as the appropriate left neighbor.
 487  *
 488  * In the non-overlapped case you can do a LE lookup and decrement the cursor
 489  * because a record's interval must end before the next record.
 490  */
 491
 492 /*
 493  * Return size of the btree block header for this btree instance.
 494  */
 495 static inline size_t xfs_btree_block_len(struct xfs_btree_cur *cur)
 496 {
 497         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
 498                 if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS)
 499                         return XFS_BTREE_LBLOCK_CRC_LEN;
 500                 return XFS_BTREE_LBLOCK_LEN;
 501         }
 502         if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS)
 503                 return XFS_BTREE_SBLOCK_CRC_LEN;
 504         return XFS_BTREE_SBLOCK_LEN;
 505 }
 506
 507 /*
 508  * Return size of btree block pointers for this btree instance.
 509  */
 510 static inline size_t xfs_btree_ptr_len(struct xfs_btree_cur *cur)
 511 {
 512         return (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
 513                 sizeof(__be64) : sizeof(__be32);
 514 }
 515
 516 /*
 517  * Calculate offset of the n-th record in a btree block.
 518  */
 519 STATIC size_t
 520 xfs_btree_rec_offset(
 521         struct xfs_btree_cur    *cur,
 522         int                     n)
 523 {
 524         return xfs_btree_block_len(cur) +
 525                 (n - 1) * cur->bc_ops->rec_len;
 526 }
 527
 528 /*
 529  * Calculate offset of the n-th key in a btree block.
 530  */
 531 STATIC size_t
 532 xfs_btree_key_offset(
 533         struct xfs_btree_cur    *cur,
 534         int                     n)
 535 {
 536         return xfs_btree_block_len(cur) +
 537                 (n - 1) * cur->bc_ops->key_len;
 538 }
 539
 540 /*
 541  * Calculate offset of the n-th high key in a btree block.
 542  */
 543 STATIC size_t
 544 xfs_btree_high_key_offset(
 545         struct xfs_btree_cur    *cur,
 546         int                     n)
 547 {
 548         return xfs_btree_block_len(cur) +
 549                 (n - 1) * cur->bc_ops->key_len + (cur->bc_ops->key_len / 2);
 550 }
 551
 552 /*
 553  * Calculate offset of the n-th block pointer in a btree block.
 554  */
 555 STATIC size_t
 556 xfs_btree_ptr_offset(
 557         struct xfs_btree_cur    *cur,
 558         int                     n,
 559         int                     level)
 560 {
 561         return xfs_btree_block_len(cur) +
 562                 cur->bc_ops->get_maxrecs(cur, level) * cur->bc_ops->key_len +
 563                 (n - 1) * xfs_btree_ptr_len(cur);
 564 }
 565
 566 /*
 567  * Return a pointer to the n-th record in the btree block.
 568  */
 569 union xfs_btree_rec *
 570 xfs_btree_rec_addr(
 571         struct xfs_btree_cur    *cur,
 572         int                     n,
 573         struct xfs_btree_block  *block)
 574 {
 575         return (union xfs_btree_rec *)
 576                 ((char *)block + xfs_btree_rec_offset(cur, n));
 577 }
 578
 579 /*
 580  * Return a pointer to the n-th key in the btree block.
 581  */
 582 union xfs_btree_key *
 583 xfs_btree_key_addr(
 584         struct xfs_btree_cur    *cur,
 585         int                     n,
 586         struct xfs_btree_block  *block)
 587 {
 588         return (union xfs_btree_key *)
 589                 ((char *)block + xfs_btree_key_offset(cur, n));
 590 }
 591
 592 /*
 593  * Return a pointer to the n-th high key in the btree block.
 594  */
 595 union xfs_btree_key *
 596 xfs_btree_high_key_addr(
 597         struct xfs_btree_cur    *cur,
 598         int                     n,
 599         struct xfs_btree_block  *block)
 600 {
 601         return (union xfs_btree_key *)
 602                 ((char *)block + xfs_btree_high_key_offset(cur, n));
 603 }
 604
 605 /*
 606  * Return a pointer to the n-th block pointer in the btree block.
 607  */
 608 union xfs_btree_ptr *
 609 xfs_btree_ptr_addr(
 610         struct xfs_btree_cur    *cur,
 611         int                     n,
 612         struct xfs_btree_block  *block)
 613 {
 614         int                     level = xfs_btree_get_level(block);
 615
 616         ASSERT(block->bb_level != 0);
 617
 618         return (union xfs_btree_ptr *)
 619                 ((char *)block + xfs_btree_ptr_offset(cur, n, level));
 620 }
 621
 622 /*
 623  * Get the root block which is stored in the inode.
 624  *
 625  * For now this btree implementation assumes the btree root is always
 626  * stored in the if_broot field of an inode fork.
 627  */
 628 STATIC struct xfs_btree_block *
 629 xfs_btree_get_iroot(
 630         struct xfs_btree_cur    *cur)
 631 {
 632         struct xfs_ifork        *ifp;
 633
 634         ifp = XFS_IFORK_PTR(cur->bc_private.b.ip, cur->bc_private.b.whichfork);
 635         return (struct xfs_btree_block *)ifp->if_broot;
 636 }
 637
 638 /*
 639  * Retrieve the block pointer from the cursor at the given level.
 640  * This may be an inode btree root or from a buffer.
 641  */
 642 struct xfs_btree_block *                /* generic btree block pointer */
 643 xfs_btree_get_block(
 644         struct xfs_btree_cur    *cur,   /* btree cursor */
 645         int                     level,  /* level in btree */
 646         struct xfs_buf          **bpp)  /* buffer containing the block */
 647 {
 648         if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
 649             (level == cur->bc_nlevels - 1)) {
 650                 *bpp = NULL;
 651                 return xfs_btree_get_iroot(cur);
 652         }
 653
 654         *bpp = cur->bc_bufs[level];
 655         return XFS_BUF_TO_BLOCK(*bpp);
 656 }
 657
 658 /*
 659  * Get a buffer for the block, return it with no data read.
 660  * Long-form addressing.
 661  */
 662 xfs_buf_t *                             /* buffer for fsbno */
 663 xfs_btree_get_bufl(
 664         xfs_mount_t     *mp,            /* file system mount point */
 665         xfs_trans_t     *tp,            /* transaction pointer */
 666         xfs_fsblock_t   fsbno,          /* file system block number */
 667         uint            lock)           /* lock flags for get_buf */
 668 {
 669         xfs_daddr_t             d;              /* real disk block address */
 670
 671         ASSERT(fsbno != NULLFSBLOCK);
 672         d = XFS_FSB_TO_DADDR(mp, fsbno);
 673         return xfs_trans_get_buf(tp, mp->m_ddev_targp, d, mp->m_bsize, lock);
 674 }
 675
 676 /*
 677  * Get a buffer for the block, return it with no data read.
 678  * Short-form addressing.
 679  */
 680 xfs_buf_t *                             /* buffer for agno/agbno */
 681 xfs_btree_get_bufs(
 682         xfs_mount_t     *mp,            /* file system mount point */
 683         xfs_trans_t     *tp,            /* transaction pointer */
 684         xfs_agnumber_t  agno,           /* allocation group number */
 685         xfs_agblock_t   agbno,          /* allocation group block number */
 686         uint            lock)           /* lock flags for get_buf */
 687 {
 688         xfs_daddr_t             d;              /* real disk block address */
 689
 690         ASSERT(agno != NULLAGNUMBER);
 691         ASSERT(agbno != NULLAGBLOCK);
 692         d = XFS_AGB_TO_DADDR(mp, agno, agbno);
 693         return xfs_trans_get_buf(tp, mp->m_ddev_targp, d, mp->m_bsize, lock);
 694 }
 695
 696 /*
 697  * Check for the cursor referring to the last block at the given level.
 698  */
 699 int                                     /* 1=is last block, 0=not last block */
 700 xfs_btree_islastblock(
 701         xfs_btree_cur_t         *cur,   /* btree cursor */
 702         int                     level)  /* level to check */
 703 {
 704         struct xfs_btree_block  *block; /* generic btree block pointer */
 705         xfs_buf_t               *bp;    /* buffer containing block */
 706
 707         block = xfs_btree_get_block(cur, level, &bp);
 708         xfs_btree_check_block(cur, block, level, bp);
 709         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
 710                 return block->bb_u.l.bb_rightsib == cpu_to_be64(NULLFSBLOCK);
 711         else
 712                 return block->bb_u.s.bb_rightsib == cpu_to_be32(NULLAGBLOCK);
 713 }
 714
 715 /*
 716  * Change the cursor to point to the first record at the given level.
 717  * Other levels are unaffected.
 718  */
 719 STATIC int                              /* success=1, failure=0 */
 720 xfs_btree_firstrec(
 721         xfs_btree_cur_t         *cur,   /* btree cursor */
 722         int                     level)  /* level to change */
 723 {
 724         struct xfs_btree_block  *block; /* generic btree block pointer */
 725         xfs_buf_t               *bp;    /* buffer containing block */
 726
 727         /*
 728          * Get the block pointer for this level.
 729          */
 730         block = xfs_btree_get_block(cur, level, &bp);
 731         if (xfs_btree_check_block(cur, block, level, bp))
 732                 return 0;
 733         /*
 734          * It's empty, there is no such record.
 735          */
 736         if (!block->bb_numrecs)
 737                 return 0;
 738         /*
 739          * Set the ptr value to 1, that's the first record/key.
 740          */
 741         cur->bc_ptrs[level] = 1;
 742         return 1;
 743 }
 744
 745 /*
 746  * Change the cursor to point to the last record in the current block
 747  * at the given level.  Other levels are unaffected.
 748  */
 749 STATIC int                              /* success=1, failure=0 */
 750 xfs_btree_lastrec(
 751         xfs_btree_cur_t         *cur,   /* btree cursor */
 752         int                     level)  /* level to change */
 753 {
 754         struct xfs_btree_block  *block; /* generic btree block pointer */
 755         xfs_buf_t               *bp;    /* buffer containing block */
 756
 757         /*
 758          * Get the block pointer for this level.
 759          */
 760         block = xfs_btree_get_block(cur, level, &bp);
 761         if (xfs_btree_check_block(cur, block, level, bp))
 762                 return 0;
 763         /*
 764          * It's empty, there is no such record.
 765          */
 766         if (!block->bb_numrecs)
 767                 return 0;
 768         /*
 769          * Set the ptr value to numrecs, that's the last record/key.
 770          */
 771         cur->bc_ptrs[level] = be16_to_cpu(block->bb_numrecs);
 772         return 1;
 773 }
 774
 775 /*
 776  * Compute first and last byte offsets for the fields given.
 777  * Interprets the offsets table, which contains struct field offsets.
 778  */
 779 void
 780 xfs_btree_offsets(
 781         int64_t         fields,         /* bitmask of fields */
 782         const short     *offsets,       /* table of field offsets */
 783         int             nbits,          /* number of bits to inspect */
 784         int             *first,         /* output: first byte offset */
 785         int             *last)          /* output: last byte offset */
 786 {
 787         int             i;              /* current bit number */
 788         int64_t         imask;          /* mask for current bit number */
 789
 790         ASSERT(fields != 0);
 791         /*
 792          * Find the lowest bit, so the first byte offset.
 793          */
 794         for (i = 0, imask = 1LL; ; i++, imask <<= 1) {
 795                 if (imask & fields) {
 796                         *first = offsets[i];
 797                         break;
 798                 }
 799         }
 800         /*
 801          * Find the highest bit, so the last byte offset.
 802          */
 803         for (i = nbits - 1, imask = 1LL << i; ; i--, imask >>= 1) {
 804                 if (imask & fields) {
 805                         *last = offsets[i + 1] - 1;
 806                         break;
 807                 }
 808         }
 809 }
 810
 811 /*
 812  * Get a buffer for the block, return it read in.
 813  * Long-form addressing.
 814  */
 815 int
 816 xfs_btree_read_bufl(
 817         struct xfs_mount        *mp,            /* file system mount point */
 818         struct xfs_trans        *tp,            /* transaction pointer */
 819         xfs_fsblock_t           fsbno,          /* file system block number */
 820         uint                    lock,           /* lock flags for read_buf */
 821         struct xfs_buf          **bpp,          /* buffer for fsbno */
 822         int                     refval,         /* ref count value for buffer */
 823         const struct xfs_buf_ops *ops)
 824 {
 825         struct xfs_buf          *bp;            /* return value */
 826         xfs_daddr_t             d;              /* real disk block address */
 827         int                     error;
 828
 829         if (!XFS_FSB_SANITY_CHECK(mp, fsbno))
 830                 return -EFSCORRUPTED;
 831         d = XFS_FSB_TO_DADDR(mp, fsbno);
 832         error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, d,
 833                                    mp->m_bsize, lock, &bp, ops);
 834         if (error)
 835                 return error;
 836         if (bp)
 837                 xfs_buf_set_ref(bp, refval);
 838         *bpp = bp;
 839         return 0;
 840 }
 841
 842 /*
 843  * Read-ahead the block, don't wait for it, don't return a buffer.
 844  * Long-form addressing.
 845  */
 846 /* ARGSUSED */
 847 void
 848 xfs_btree_reada_bufl(
 849         struct xfs_mount        *mp,            /* file system mount point */
 850         xfs_fsblock_t           fsbno,          /* file system block number */
 851         xfs_extlen_t            count,          /* count of filesystem blocks */
 852         const struct xfs_buf_ops *ops)
 853 {
 854         xfs_daddr_t             d;
 855
 856         ASSERT(fsbno != NULLFSBLOCK);
 857         d = XFS_FSB_TO_DADDR(mp, fsbno);
 858         xfs_buf_readahead(mp->m_ddev_targp, d, mp->m_bsize * count, ops);
 859 }
 860
 861 /*
 862  * Read-ahead the block, don't wait for it, don't return a buffer.
 863  * Short-form addressing.
 864  */
 865 /* ARGSUSED */
 866 void
 867 xfs_btree_reada_bufs(
 868         struct xfs_mount        *mp,            /* file system mount point */
 869         xfs_agnumber_t          agno,           /* allocation group number */
 870         xfs_agblock_t           agbno,          /* allocation group block number */
 871         xfs_extlen_t            count,          /* count of filesystem blocks */
 872         const struct xfs_buf_ops *ops)
 873 {
 874         xfs_daddr_t             d;
 875
 876         ASSERT(agno != NULLAGNUMBER);
 877         ASSERT(agbno != NULLAGBLOCK);
 878         d = XFS_AGB_TO_DADDR(mp, agno, agbno);
 879         xfs_buf_readahead(mp->m_ddev_targp, d, mp->m_bsize * count, ops);
 880 }
 881
 882 STATIC int
 883 xfs_btree_readahead_lblock(
 884         struct xfs_btree_cur    *cur,
 885         int                     lr,
 886         struct xfs_btree_block  *block)
 887 {
 888         int                     rval = 0;
 889         xfs_fsblock_t           left = be64_to_cpu(block->bb_u.l.bb_leftsib);
 890         xfs_fsblock_t           right = be64_to_cpu(block->bb_u.l.bb_rightsib);
 891
 892         if ((lr & XFS_BTCUR_LEFTRA) && left != NULLFSBLOCK) {
 893                 xfs_btree_reada_bufl(cur->bc_mp, left, 1,
 894                                      cur->bc_ops->buf_ops);
 895                 rval++;
 896         }
 897
 898         if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLFSBLOCK) {
 899                 xfs_btree_reada_bufl(cur->bc_mp, right, 1,
 900                                      cur->bc_ops->buf_ops);
 901                 rval++;
 902         }
 903
 904         return rval;
 905 }
 906
 907 STATIC int
 908 xfs_btree_readahead_sblock(
 909         struct xfs_btree_cur    *cur,
 910         int                     lr,
 911         struct xfs_btree_block *block)
 912 {
 913         int                     rval = 0;
 914         xfs_agblock_t           left = be32_to_cpu(block->bb_u.s.bb_leftsib);
 915         xfs_agblock_t           right = be32_to_cpu(block->bb_u.s.bb_rightsib);
 916
 917
 918         if ((lr & XFS_BTCUR_LEFTRA) && left != NULLAGBLOCK) {
 919                 xfs_btree_reada_bufs(cur->bc_mp, cur->bc_private.a.agno,
 920                                      left, 1, cur->bc_ops->buf_ops);
 921                 rval++;
 922         }
 923
 924         if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLAGBLOCK) {
 925                 xfs_btree_reada_bufs(cur->bc_mp, cur->bc_private.a.agno,
 926                                      right, 1, cur->bc_ops->buf_ops);
 927                 rval++;
 928         }
 929
 930         return rval;
 931 }
 932
 933 /*
 934  * Read-ahead btree blocks, at the given level.
 935  * Bits in lr are set from XFS_BTCUR_{LEFT,RIGHT}RA.
 936  */
 937 STATIC int
 938 xfs_btree_readahead(
 939         struct xfs_btree_cur    *cur,           /* btree cursor */
 940         int                     lev,            /* level in btree */
 941         int                     lr)             /* left/right bits */
 942 {
 943         struct xfs_btree_block  *block;
 944
 945         /*
 946          * No readahead needed if we are at the root level and the
 947          * btree root is stored in the inode.
 948          */
 949         if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
 950             (lev == cur->bc_nlevels - 1))
 951                 return 0;
 952
 953         if ((cur->bc_ra[lev] | lr) == cur->bc_ra[lev])
 954                 return 0;
 955
 956         cur->bc_ra[lev] |= lr;
 957         block = XFS_BUF_TO_BLOCK(cur->bc_bufs[lev]);
 958
 959         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
 960                 return xfs_btree_readahead_lblock(cur, lr, block);
 961         return xfs_btree_readahead_sblock(cur, lr, block);
 962 }
 963
 964 STATIC xfs_daddr_t
 965 xfs_btree_ptr_to_daddr(
 966         struct xfs_btree_cur    *cur,
 967         union xfs_btree_ptr     *ptr)
 968 {
 969         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
 970                 ASSERT(ptr->l != cpu_to_be64(NULLFSBLOCK));
 971
 972                 return XFS_FSB_TO_DADDR(cur->bc_mp, be64_to_cpu(ptr->l));
 973         } else {
 974                 ASSERT(cur->bc_private.a.agno != NULLAGNUMBER);
 975                 ASSERT(ptr->s != cpu_to_be32(NULLAGBLOCK));
 976
 977                 return XFS_AGB_TO_DADDR(cur->bc_mp, cur->bc_private.a.agno,
 978                                         be32_to_cpu(ptr->s));
 979         }
 980 }
 981
 982 /*
 983  * Readahead @count btree blocks at the given @ptr location.
 984  *
 985  * We don't need to care about long or short form btrees here as we have a
 986  * method of converting the ptr directly to a daddr available to us.
 987  */
 988 STATIC void
 989 xfs_btree_readahead_ptr(
 990         struct xfs_btree_cur    *cur,
 991         union xfs_btree_ptr     *ptr,
 992         xfs_extlen_t            count)
 993 {
 994         xfs_buf_readahead(cur->bc_mp->m_ddev_targp,
 995                           xfs_btree_ptr_to_daddr(cur, ptr),
 996                           cur->bc_mp->m_bsize * count, cur->bc_ops->buf_ops);
 997 }
 998
 999 /*
1000  * Set the buffer for level "lev" in the cursor to bp, releasing
1001  * any previous buffer.
1002  */
1003 STATIC void
1004 xfs_btree_setbuf(
1005         xfs_btree_cur_t         *cur,   /* btree cursor */
1006         int                     lev,    /* level in btree */
1007         xfs_buf_t               *bp)    /* new buffer to set */
1008 {
1009         struct xfs_btree_block  *b;     /* btree block */
1010
1011         if (cur->bc_bufs[lev])
1012                 xfs_trans_brelse(cur->bc_tp, cur->bc_bufs[lev]);
1013         cur->bc_bufs[lev] = bp;
1014         cur->bc_ra[lev] = 0;
1015
1016         b = XFS_BUF_TO_BLOCK(bp);
1017         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
1018                 if (b->bb_u.l.bb_leftsib == cpu_to_be64(NULLFSBLOCK))
1019                         cur->bc_ra[lev] |= XFS_BTCUR_LEFTRA;
1020                 if (b->bb_u.l.bb_rightsib == cpu_to_be64(NULLFSBLOCK))
1021                         cur->bc_ra[lev] |= XFS_BTCUR_RIGHTRA;
1022         } else {
1023                 if (b->bb_u.s.bb_leftsib == cpu_to_be32(NULLAGBLOCK))
1024                         cur->bc_ra[lev] |= XFS_BTCUR_LEFTRA;
1025                 if (b->bb_u.s.bb_rightsib == cpu_to_be32(NULLAGBLOCK))
1026                         cur->bc_ra[lev] |= XFS_BTCUR_RIGHTRA;
1027         }
1028 }
1029
1030 STATIC int
1031 xfs_btree_ptr_is_null(
1032         struct xfs_btree_cur    *cur,
1033         union xfs_btree_ptr     *ptr)
1034 {
1035         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1036                 return ptr->l == cpu_to_be64(NULLFSBLOCK);
1037         else
1038                 return ptr->s == cpu_to_be32(NULLAGBLOCK);
1039 }
1040
1041 STATIC void
1042 xfs_btree_set_ptr_null(
1043         struct xfs_btree_cur    *cur,
1044         union xfs_btree_ptr     *ptr)
1045 {
1046         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1047                 ptr->l = cpu_to_be64(NULLFSBLOCK);
1048         else
1049                 ptr->s = cpu_to_be32(NULLAGBLOCK);
1050 }
1051
1052 /*
1053  * Get/set/init sibling pointers
1054  */
1055 STATIC void
1056 xfs_btree_get_sibling(
1057         struct xfs_btree_cur    *cur,
1058         struct xfs_btree_block  *block,
1059         union xfs_btree_ptr     *ptr,
1060         int                     lr)
1061 {
1062         ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
1063
1064         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
1065                 if (lr == XFS_BB_RIGHTSIB)
1066                         ptr->l = block->bb_u.l.bb_rightsib;
1067                 else
1068                         ptr->l = block->bb_u.l.bb_leftsib;
1069         } else {
1070                 if (lr == XFS_BB_RIGHTSIB)
1071                         ptr->s = block->bb_u.s.bb_rightsib;
1072                 else
1073                         ptr->s = block->bb_u.s.bb_leftsib;
1074         }
1075 }
1076
1077 STATIC void
1078 xfs_btree_set_sibling(
1079         struct xfs_btree_cur    *cur,
1080         struct xfs_btree_block  *block,
1081         union xfs_btree_ptr     *ptr,
1082         int                     lr)
1083 {
1084         ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
1085
1086         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
1087                 if (lr == XFS_BB_RIGHTSIB)
1088                         block->bb_u.l.bb_rightsib = ptr->l;
1089                 else
1090                         block->bb_u.l.bb_leftsib = ptr->l;
1091         } else {
1092                 if (lr == XFS_BB_RIGHTSIB)
1093                         block->bb_u.s.bb_rightsib = ptr->s;
1094                 else
1095                         block->bb_u.s.bb_leftsib = ptr->s;
1096         }
1097 }
1098
1099 void
1100 xfs_btree_init_block_int(
1101         struct xfs_mount        *mp,
1102         struct xfs_btree_block  *buf,
1103         xfs_daddr_t             blkno,
1104         xfs_btnum_t             btnum,
1105         __u16                   level,
1106         __u16                   numrecs,
1107         __u64                   owner,
1108         unsigned int            flags)
1109 {
1110         int                     crc = xfs_sb_version_hascrc(&mp->m_sb);
1111         __u32                   magic = xfs_btree_magic(crc, btnum);
1112
1113         buf->bb_magic = cpu_to_be32(magic);
1114         buf->bb_level = cpu_to_be16(level);
1115         buf->bb_numrecs = cpu_to_be16(numrecs);
1116
1117         if (flags & XFS_BTREE_LONG_PTRS) {
1118                 buf->bb_u.l.bb_leftsib = cpu_to_be64(NULLFSBLOCK);
1119                 buf->bb_u.l.bb_rightsib = cpu_to_be64(NULLFSBLOCK);
1120                 if (crc) {
1121                         buf->bb_u.l.bb_blkno = cpu_to_be64(blkno);
1122                         buf->bb_u.l.bb_owner = cpu_to_be64(owner);
1123                         uuid_copy(&buf->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid);
1124                         buf->bb_u.l.bb_pad = 0;
1125                         buf->bb_u.l.bb_lsn = 0;
1126                 }
1127         } else {
1128                 /* owner is a 32 bit value on short blocks */
1129                 __u32 __owner = (__u32)owner;
1130
1131                 buf->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
1132                 buf->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
1133                 if (crc) {
1134                         buf->bb_u.s.bb_blkno = cpu_to_be64(blkno);
1135                         buf->bb_u.s.bb_owner = cpu_to_be32(__owner);
1136                         uuid_copy(&buf->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid);
1137                         buf->bb_u.s.bb_lsn = 0;
1138                 }
1139         }
1140 }
1141
1142 void
1143 xfs_btree_init_block(
1144         struct xfs_mount *mp,
1145         struct xfs_buf  *bp,
1146         xfs_btnum_t     btnum,
1147         __u16           level,
1148         __u16           numrecs,
1149         __u64           owner,
1150         unsigned int    flags)
1151 {
1152         xfs_btree_init_block_int(mp, XFS_BUF_TO_BLOCK(bp), bp->b_bn,
1153                                  btnum, level, numrecs, owner, flags);
1154 }
1155
1156 STATIC void
1157 xfs_btree_init_block_cur(
1158         struct xfs_btree_cur    *cur,
1159         struct xfs_buf          *bp,
1160         int                     level,
1161         int                     numrecs)
1162 {
1163         __u64                   owner;
1164
1165         /*
1166          * we can pull the owner from the cursor right now as the different
1167          * owners align directly with the pointer size of the btree. This may
1168          * change in future, but is safe for current users of the generic btree
1169          * code.
1170          */
1171         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1172                 owner = cur->bc_private.b.ip->i_ino;
1173         else
1174                 owner = cur->bc_private.a.agno;
1175
1176         xfs_btree_init_block_int(cur->bc_mp, XFS_BUF_TO_BLOCK(bp), bp->b_bn,
1177                                  cur->bc_btnum, level, numrecs,
1178                                  owner, cur->bc_flags);
1179 }
1180
1181 /*
1182  * Return true if ptr is the last record in the btree and
1183  * we need to track updates to this record.  The decision
1184  * will be further refined in the update_lastrec method.
1185  */
1186 STATIC int
1187 xfs_btree_is_lastrec(
1188         struct xfs_btree_cur    *cur,
1189         struct xfs_btree_block  *block,
1190         int                     level)
1191 {
1192         union xfs_btree_ptr     ptr;
1193
1194         if (level > 0)
1195                 return 0;
1196         if (!(cur->bc_flags & XFS_BTREE_LASTREC_UPDATE))
1197                 return 0;
1198
1199         xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
1200         if (!xfs_btree_ptr_is_null(cur, &ptr))
1201                 return 0;
1202         return 1;
1203 }
1204
1205 STATIC void
1206 xfs_btree_buf_to_ptr(
1207         struct xfs_btree_cur    *cur,
1208         struct xfs_buf          *bp,
1209         union xfs_btree_ptr     *ptr)
1210 {
1211         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1212                 ptr->l = cpu_to_be64(XFS_DADDR_TO_FSB(cur->bc_mp,
1213                                         XFS_BUF_ADDR(bp)));
1214         else {
1215                 ptr->s = cpu_to_be32(xfs_daddr_to_agbno(cur->bc_mp,
1216                                         XFS_BUF_ADDR(bp)));
1217         }
1218 }
1219
1220 STATIC void
1221 xfs_btree_set_refs(
1222         struct xfs_btree_cur    *cur,
1223         struct xfs_buf          *bp)
1224 {
1225         switch (cur->bc_btnum) {
1226         case XFS_BTNUM_BNO:
1227         case XFS_BTNUM_CNT:
1228                 xfs_buf_set_ref(bp, XFS_ALLOC_BTREE_REF);
1229                 break;
1230         case XFS_BTNUM_INO:
1231         case XFS_BTNUM_FINO:
1232                 xfs_buf_set_ref(bp, XFS_INO_BTREE_REF);
1233                 break;
1234         case XFS_BTNUM_BMAP:
1235                 xfs_buf_set_ref(bp, XFS_BMAP_BTREE_REF);
1236                 break;
1237         case XFS_BTNUM_RMAP:
1238                 xfs_buf_set_ref(bp, XFS_RMAP_BTREE_REF);
1239                 break;
1240         case XFS_BTNUM_REFC:
1241                 xfs_buf_set_ref(bp, XFS_REFC_BTREE_REF);
1242                 break;
1243         default:
1244                 ASSERT(0);
1245         }
1246 }
1247
1248 STATIC int
1249 xfs_btree_get_buf_block(
1250         struct xfs_btree_cur    *cur,
1251         union xfs_btree_ptr     *ptr,
1252         int                     flags,
1253         struct xfs_btree_block  **block,
1254         struct xfs_buf          **bpp)
1255 {
1256         struct xfs_mount        *mp = cur->bc_mp;
1257         xfs_daddr_t             d;
1258
1259         /* need to sort out how callers deal with failures first */
1260         ASSERT(!(flags & XBF_TRYLOCK));
1261
1262         d = xfs_btree_ptr_to_daddr(cur, ptr);
1263         *bpp = xfs_trans_get_buf(cur->bc_tp, mp->m_ddev_targp, d,
1264                                  mp->m_bsize, flags);
1265
1266         if (!*bpp)
1267                 return -ENOMEM;
1268
1269         (*bpp)->b_ops = cur->bc_ops->buf_ops;
1270         *block = XFS_BUF_TO_BLOCK(*bpp);
1271         return 0;
1272 }
1273
1274 /*
1275  * Read in the buffer at the given ptr and return the buffer and
1276  * the block pointer within the buffer.
1277  */
1278 STATIC int
1279 xfs_btree_read_buf_block(
1280         struct xfs_btree_cur    *cur,
1281         union xfs_btree_ptr     *ptr,
1282         int                     flags,
1283         struct xfs_btree_block  **block,
1284         struct xfs_buf          **bpp)
1285 {
1286         struct xfs_mount        *mp = cur->bc_mp;
1287         xfs_daddr_t             d;
1288         int                     error;
1289
1290         /* need to sort out how callers deal with failures first */
1291         ASSERT(!(flags & XBF_TRYLOCK));
1292
1293         d = xfs_btree_ptr_to_daddr(cur, ptr);
1294         error = xfs_trans_read_buf(mp, cur->bc_tp, mp->m_ddev_targp, d,
1295                                    mp->m_bsize, flags, bpp,
1296                                    cur->bc_ops->buf_ops);
1297         if (error)
1298                 return error;
1299
1300         xfs_btree_set_refs(cur, *bpp);
1301         *block = XFS_BUF_TO_BLOCK(*bpp);
1302         return 0;
1303 }
1304
1305 /*
1306  * Copy keys from one btree block to another.
1307  */
1308 STATIC void
1309 xfs_btree_copy_keys(
1310         struct xfs_btree_cur    *cur,
1311         union xfs_btree_key     *dst_key,
1312         union xfs_btree_key     *src_key,
1313         int                     numkeys)
1314 {
1315         ASSERT(numkeys >= 0);
1316         memcpy(dst_key, src_key, numkeys * cur->bc_ops->key_len);
1317 }
1318
1319 /*
1320  * Copy records from one btree block to another.
1321  */
1322 STATIC void
1323 xfs_btree_copy_recs(
1324         struct xfs_btree_cur    *cur,
1325         union xfs_btree_rec     *dst_rec,
1326         union xfs_btree_rec     *src_rec,
1327         int                     numrecs)
1328 {
1329         ASSERT(numrecs >= 0);
1330         memcpy(dst_rec, src_rec, numrecs * cur->bc_ops->rec_len);
1331 }
1332
1333 /*
1334  * Copy block pointers from one btree block to another.
1335  */
1336 STATIC void
1337 xfs_btree_copy_ptrs(
1338         struct xfs_btree_cur    *cur,
1339         union xfs_btree_ptr     *dst_ptr,
1340         union xfs_btree_ptr     *src_ptr,
1341         int                     numptrs)
1342 {
1343         ASSERT(numptrs >= 0);
1344         memcpy(dst_ptr, src_ptr, numptrs * xfs_btree_ptr_len(cur));
1345 }
1346
1347 /*
1348  * Shift keys one index left/right inside a single btree block.
1349  */
1350 STATIC void
1351 xfs_btree_shift_keys(
1352         struct xfs_btree_cur    *cur,
1353         union xfs_btree_key     *key,
1354         int                     dir,
1355         int                     numkeys)
1356 {
1357         char                    *dst_key;
1358
1359         ASSERT(numkeys >= 0);
1360         ASSERT(dir == 1 || dir == -1);
1361
1362         dst_key = (char *)key + (dir * cur->bc_ops->key_len);
1363         memmove(dst_key, key, numkeys * cur->bc_ops->key_len);
1364 }
1365
1366 /*
1367  * Shift records one index left/right inside a single btree block.
1368  */
1369 STATIC void
1370 xfs_btree_shift_recs(
1371         struct xfs_btree_cur    *cur,
1372         union xfs_btree_rec     *rec,
1373         int                     dir,
1374         int                     numrecs)
1375 {
1376         char                    *dst_rec;
1377
1378         ASSERT(numrecs >= 0);
1379         ASSERT(dir == 1 || dir == -1);
1380
1381         dst_rec = (char *)rec + (dir * cur->bc_ops->rec_len);
1382         memmove(dst_rec, rec, numrecs * cur->bc_ops->rec_len);
1383 }
1384
1385 /*
1386  * Shift block pointers one index left/right inside a single btree block.
1387  */
1388 STATIC void
1389 xfs_btree_shift_ptrs(
1390         struct xfs_btree_cur    *cur,
1391         union xfs_btree_ptr     *ptr,
1392         int                     dir,
1393         int                     numptrs)
1394 {
1395         char                    *dst_ptr;
1396
1397         ASSERT(numptrs >= 0);
1398         ASSERT(dir == 1 || dir == -1);
1399
1400         dst_ptr = (char *)ptr + (dir * xfs_btree_ptr_len(cur));
1401         memmove(dst_ptr, ptr, numptrs * xfs_btree_ptr_len(cur));
1402 }
1403
1404 /*
1405  * Log key values from the btree block.
1406  */
1407 STATIC void
1408 xfs_btree_log_keys(
1409         struct xfs_btree_cur    *cur,
1410         struct xfs_buf          *bp,
1411         int                     first,
1412         int                     last)
1413 {
1414         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
1415         XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
1416
1417         if (bp) {
1418                 xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1419                 xfs_trans_log_buf(cur->bc_tp, bp,
1420                                   xfs_btree_key_offset(cur, first),
1421                                   xfs_btree_key_offset(cur, last + 1) - 1);
1422         } else {
1423                 xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
1424                                 xfs_ilog_fbroot(cur->bc_private.b.whichfork));
1425         }
1426
1427         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
1428 }
1429
1430 /*
1431  * Log record values from the btree block.
1432  */
1433 void
1434 xfs_btree_log_recs(
1435         struct xfs_btree_cur    *cur,
1436         struct xfs_buf          *bp,
1437         int                     first,
1438         int                     last)
1439 {
1440         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
1441         XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
1442
1443         xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1444         xfs_trans_log_buf(cur->bc_tp, bp,
1445                           xfs_btree_rec_offset(cur, first),
1446                           xfs_btree_rec_offset(cur, last + 1) - 1);
1447
1448         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
1449 }
1450
1451 /*
1452  * Log block pointer fields from a btree block (nonleaf).
1453  */
1454 STATIC void
1455 xfs_btree_log_ptrs(
1456         struct xfs_btree_cur    *cur,   /* btree cursor */
1457         struct xfs_buf          *bp,    /* buffer containing btree block */
1458         int                     first,  /* index of first pointer to log */
1459         int                     last)   /* index of last pointer to log */
1460 {
1461         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
1462         XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
1463
1464         if (bp) {
1465                 struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
1466                 int                     level = xfs_btree_get_level(block);
1467
1468                 xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1469                 xfs_trans_log_buf(cur->bc_tp, bp,
1470                                 xfs_btree_ptr_offset(cur, first, level),
1471                                 xfs_btree_ptr_offset(cur, last + 1, level) - 1);
1472         } else {
1473                 xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
1474                         xfs_ilog_fbroot(cur->bc_private.b.whichfork));
1475         }
1476
1477         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
1478 }
1479
1480 /*
1481  * Log fields from a btree block header.
1482  */
1483 void
1484 xfs_btree_log_block(
1485         struct xfs_btree_cur    *cur,   /* btree cursor */
1486         struct xfs_buf          *bp,    /* buffer containing btree block */
1487         int                     fields) /* mask of fields: XFS_BB_... */
1488 {
1489         int                     first;  /* first byte offset logged */
1490         int                     last;   /* last byte offset logged */
1491         static const short      soffsets[] = {  /* table of offsets (short) */
1492                 offsetof(struct xfs_btree_block, bb_magic),
1493                 offsetof(struct xfs_btree_block, bb_level),
1494                 offsetof(struct xfs_btree_block, bb_numrecs),
1495                 offsetof(struct xfs_btree_block, bb_u.s.bb_leftsib),
1496                 offsetof(struct xfs_btree_block, bb_u.s.bb_rightsib),
1497                 offsetof(struct xfs_btree_block, bb_u.s.bb_blkno),
1498                 offsetof(struct xfs_btree_block, bb_u.s.bb_lsn),
1499                 offsetof(struct xfs_btree_block, bb_u.s.bb_uuid),
1500                 offsetof(struct xfs_btree_block, bb_u.s.bb_owner),
1501                 offsetof(struct xfs_btree_block, bb_u.s.bb_crc),
1502                 XFS_BTREE_SBLOCK_CRC_LEN
1503         };
1504         static const short      loffsets[] = {  /* table of offsets (long) */
1505                 offsetof(struct xfs_btree_block, bb_magic),
1506                 offsetof(struct xfs_btree_block, bb_level),
1507                 offsetof(struct xfs_btree_block, bb_numrecs),
1508                 offsetof(struct xfs_btree_block, bb_u.l.bb_leftsib),
1509                 offsetof(struct xfs_btree_block, bb_u.l.bb_rightsib),
1510                 offsetof(struct xfs_btree_block, bb_u.l.bb_blkno),
1511                 offsetof(struct xfs_btree_block, bb_u.l.bb_lsn),
1512                 offsetof(struct xfs_btree_block, bb_u.l.bb_uuid),
1513                 offsetof(struct xfs_btree_block, bb_u.l.bb_owner),
1514                 offsetof(struct xfs_btree_block, bb_u.l.bb_crc),
1515                 offsetof(struct xfs_btree_block, bb_u.l.bb_pad),
1516                 XFS_BTREE_LBLOCK_CRC_LEN
1517         };
1518
1519         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
1520         XFS_BTREE_TRACE_ARGBI(cur, bp, fields);
1521
1522         if (bp) {
1523                 int nbits;
1524
1525                 if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS) {
1526                         /*
1527                          * We don't log the CRC when updating a btree
1528                          * block but instead recreate it during log
1529                          * recovery.  As the log buffers have checksums
1530                          * of their own this is safe and avoids logging a crc
1531                          * update in a lot of places.
1532                          */
1533                         if (fields == XFS_BB_ALL_BITS)
1534                                 fields = XFS_BB_ALL_BITS_CRC;
1535                         nbits = XFS_BB_NUM_BITS_CRC;
1536                 } else {
1537                         nbits = XFS_BB_NUM_BITS;
1538                 }
1539                 xfs_btree_offsets(fields,
1540                                   (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
1541                                         loffsets : soffsets,
1542                                   nbits, &first, &last);
1543                 xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1544                 xfs_trans_log_buf(cur->bc_tp, bp, first, last);
1545         } else {
1546                 xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
1547                         xfs_ilog_fbroot(cur->bc_private.b.whichfork));
1548         }
1549
1550         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
1551 }
1552
1553 /*
1554  * Increment cursor by one record at the level.
1555  * For nonzero levels the leaf-ward information is untouched.
1556  */
1557 int                                             /* error */
1558 xfs_btree_increment(
1559         struct xfs_btree_cur    *cur,
1560         int                     level,
1561         int                     *stat)          /* success/failure */
1562 {
1563         struct xfs_btree_block  *block;
1564         union xfs_btree_ptr     ptr;
1565         struct xfs_buf          *bp;
1566         int                     error;          /* error return value */
1567         int                     lev;
1568
1569         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
1570         XFS_BTREE_TRACE_ARGI(cur, level);
1571
1572         ASSERT(level < cur->bc_nlevels);
1573
1574         /* Read-ahead to the right at this level. */
1575         xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
1576
1577         /* Get a pointer to the btree block. */
1578         block = xfs_btree_get_block(cur, level, &bp);
1579
1580 #ifdef DEBUG
1581         error = xfs_btree_check_block(cur, block, level, bp);
1582         if (error)
1583                 goto error0;
1584 #endif
1585
1586         /* We're done if we remain in the block after the increment. */
1587         if (++cur->bc_ptrs[level] <= xfs_btree_get_numrecs(block))
1588                 goto out1;
1589
1590         /* Fail if we just went off the right edge of the tree. */
1591         xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
1592         if (xfs_btree_ptr_is_null(cur, &ptr))
1593                 goto out0;
1594
1595         XFS_BTREE_STATS_INC(cur, increment);
1596
1597         /*
1598          * March up the tree incrementing pointers.
1599          * Stop when we don't go off the right edge of a block.
1600          */
1601         for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
1602                 block = xfs_btree_get_block(cur, lev, &bp);
1603
1604 #ifdef DEBUG
1605                 error = xfs_btree_check_block(cur, block, lev, bp);
1606                 if (error)
1607                         goto error0;
1608 #endif
1609
1610                 if (++cur->bc_ptrs[lev] <= xfs_btree_get_numrecs(block))
1611                         break;
1612
1613                 /* Read-ahead the right block for the next loop. */
1614                 xfs_btree_readahead(cur, lev, XFS_BTCUR_RIGHTRA);
1615         }
1616
1617         /*
1618          * If we went off the root then we are either seriously
1619          * confused or have the tree root in an inode.
1620          */
1621         if (lev == cur->bc_nlevels) {
1622                 if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
1623                         goto out0;
1624                 ASSERT(0);
1625                 error = -EFSCORRUPTED;
1626                 goto error0;
1627         }
1628         ASSERT(lev < cur->bc_nlevels);
1629
1630         /*
1631          * Now walk back down the tree, fixing up the cursor's buffer
1632          * pointers and key numbers.
1633          */
1634         for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
1635                 union xfs_btree_ptr     *ptrp;
1636
1637                 ptrp = xfs_btree_ptr_addr(cur, cur->bc_ptrs[lev], block);
1638                 --lev;
1639                 error = xfs_btree_read_buf_block(cur, ptrp, 0, &block, &bp);
1640                 if (error)
1641                         goto error0;
1642
1643                 xfs_btree_setbuf(cur, lev, bp);
1644                 cur->bc_ptrs[lev] = 1;
1645         }
1646 out1:
1647         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
1648         *stat = 1;
1649         return 0;
1650
1651 out0:
1652         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
1653         *stat = 0;
1654         return 0;
1655
1656 error0:
1657         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
1658         return error;
1659 }
1660
1661 /*
1662  * Decrement cursor by one record at the level.
1663  * For nonzero levels the leaf-ward information is untouched.
1664  */
1665 int                                             /* error */
1666 xfs_btree_decrement(
1667         struct xfs_btree_cur    *cur,
1668         int                     level,
1669         int                     *stat)          /* success/failure */
1670 {
1671         struct xfs_btree_block  *block;
1672         xfs_buf_t               *bp;
1673         int                     error;          /* error return value */
1674         int                     lev;
1675         union xfs_btree_ptr     ptr;
1676
1677         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
1678         XFS_BTREE_TRACE_ARGI(cur, level);
1679
1680         ASSERT(level < cur->bc_nlevels);
1681
1682         /* Read-ahead to the left at this level. */
1683         xfs_btree_readahead(cur, level, XFS_BTCUR_LEFTRA);
1684
1685         /* We're done if we remain in the block after the decrement. */
1686         if (--cur->bc_ptrs[level] > 0)
1687                 goto out1;
1688
1689         /* Get a pointer to the btree block. */
1690         block = xfs_btree_get_block(cur, level, &bp);
1691
1692 #ifdef DEBUG
1693         error = xfs_btree_check_block(cur, block, level, bp);
1694         if (error)
1695                 goto error0;
1696 #endif
1697
1698         /* Fail if we just went off the left edge of the tree. */
1699         xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
1700         if (xfs_btree_ptr_is_null(cur, &ptr))
1701                 goto out0;
1702
1703         XFS_BTREE_STATS_INC(cur, decrement);
1704
1705         /*
1706          * March up the tree decrementing pointers.
1707          * Stop when we don't go off the left edge of a block.
1708          */
1709         for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
1710                 if (--cur->bc_ptrs[lev] > 0)
1711                         break;
1712                 /* Read-ahead the left block for the next loop. */
1713                 xfs_btree_readahead(cur, lev, XFS_BTCUR_LEFTRA);
1714         }
1715
1716         /*
1717          * If we went off the root then we are seriously confused.
1718          * or the root of the tree is in an inode.
1719          */
1720         if (lev == cur->bc_nlevels) {
1721                 if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
1722                         goto out0;
1723                 ASSERT(0);
1724                 error = -EFSCORRUPTED;
1725                 goto error0;
1726         }
1727         ASSERT(lev < cur->bc_nlevels);
1728
1729         /*
1730          * Now walk back down the tree, fixing up the cursor's buffer
1731          * pointers and key numbers.
1732          */
1733         for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
1734                 union xfs_btree_ptr     *ptrp;
1735
1736                 ptrp = xfs_btree_ptr_addr(cur, cur->bc_ptrs[lev], block);
1737                 --lev;
1738                 error = xfs_btree_read_buf_block(cur, ptrp, 0, &block, &bp);
1739                 if (error)
1740                         goto error0;
1741                 xfs_btree_setbuf(cur, lev, bp);
1742                 cur->bc_ptrs[lev] = xfs_btree_get_numrecs(block);
1743         }
1744 out1:
1745         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
1746         *stat = 1;
1747         return 0;
1748
1749 out0:
1750         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
1751         *stat = 0;
1752         return 0;
1753
1754 error0:
1755         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
1756         return error;
1757 }
1758
1759 int
1760 xfs_btree_lookup_get_block(
1761         struct xfs_btree_cur    *cur,   /* btree cursor */
1762         int                     level,  /* level in the btree */
1763         union xfs_btree_ptr     *pp,    /* ptr to btree block */
1764         struct xfs_btree_block  **blkp) /* return btree block */
1765 {
1766         struct xfs_buf          *bp;    /* buffer pointer for btree block */
1767         int                     error = 0;
1768
1769         /* special case the root block if in an inode */
1770         if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
1771             (level == cur->bc_nlevels - 1)) {
1772                 *blkp = xfs_btree_get_iroot(cur);
1773                 return 0;
1774         }
1775
1776         /*
1777          * If the old buffer at this level for the disk address we are
1778          * looking for re-use it.
1779          *
1780          * Otherwise throw it away and get a new one.
1781          */
1782         bp = cur->bc_bufs[level];
1783         if (bp && XFS_BUF_ADDR(bp) == xfs_btree_ptr_to_daddr(cur, pp)) {
1784                 *blkp = XFS_BUF_TO_BLOCK(bp);
1785                 return 0;
1786         }
1787
1788         error = xfs_btree_read_buf_block(cur, pp, 0, blkp, &bp);
1789         if (error)
1790                 return error;
1791
1792         /* Check the inode owner since the verifiers don't. */
1793         if (xfs_sb_version_hascrc(&cur->bc_mp->m_sb) &&
1794             (cur->bc_flags & XFS_BTREE_LONG_PTRS) &&
1795             be64_to_cpu((*blkp)->bb_u.l.bb_owner) !=
1796                         cur->bc_private.b.ip->i_ino)
1797                 goto out_bad;
1798
1799         /* Did we get the level we were looking for? */
1800         if (be16_to_cpu((*blkp)->bb_level) != level)
1801                 goto out_bad;
1802
1803         /* Check that internal nodes have at least one record. */
1804         if (level != 0 && be16_to_cpu((*blkp)->bb_numrecs) == 0)
1805                 goto out_bad;
1806
1807         xfs_btree_setbuf(cur, level, bp);
1808         return 0;
1809
1810 out_bad:
1811         *blkp = NULL;
1812         xfs_trans_brelse(cur->bc_tp, bp);
1813         return -EFSCORRUPTED;
1814 }
1815
1816 /*
1817  * Get current search key.  For level 0 we don't actually have a key
1818  * structure so we make one up from the record.  For all other levels
1819  * we just return the right key.
1820  */
1821 STATIC union xfs_btree_key *
1822 xfs_lookup_get_search_key(
1823         struct xfs_btree_cur    *cur,
1824         int                     level,
1825         int                     keyno,
1826         struct xfs_btree_block  *block,
1827         union xfs_btree_key     *kp)
1828 {
1829         if (level == 0) {
1830                 cur->bc_ops->init_key_from_rec(kp,
1831                                 xfs_btree_rec_addr(cur, keyno, block));
1832                 return kp;
1833         }
1834
1835         return xfs_btree_key_addr(cur, keyno, block);
1836 }
1837
1838 /*
1839  * Lookup the record.  The cursor is made to point to it, based on dir.
1840  * stat is set to 0 if can't find any such record, 1 for success.
1841  */
1842 int                                     /* error */
1843 xfs_btree_lookup(
1844         struct xfs_btree_cur    *cur,   /* btree cursor */
1845         xfs_lookup_t            dir,    /* <=, ==, or >= */
1846         int                     *stat)  /* success/failure */
1847 {
1848         struct xfs_btree_block  *block; /* current btree block */
1849         int64_t                 diff;   /* difference for the current key */
1850         int                     error;  /* error return value */
1851         int                     keyno;  /* current key number */
1852         int                     level;  /* level in the btree */
1853         union xfs_btree_ptr     *pp;    /* ptr to btree block */
1854         union xfs_btree_ptr     ptr;    /* ptr to btree block */
1855
1856         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
1857         XFS_BTREE_TRACE_ARGI(cur, dir);
1858
1859         XFS_BTREE_STATS_INC(cur, lookup);
1860
1861         /* No such thing as a zero-level tree. */
1862         if (cur->bc_nlevels == 0)
1863                 return -EFSCORRUPTED;
1864
1865         block = NULL;
1866         keyno = 0;
1867
1868         /* initialise start pointer from cursor */
1869         cur->bc_ops->init_ptr_from_cur(cur, &ptr);
1870         pp = &ptr;
1871
1872         /*
1873          * Iterate over each level in the btree, starting at the root.
1874          * For each level above the leaves, find the key we need, based
1875          * on the lookup record, then follow the corresponding block
1876          * pointer down to the next level.
1877          */
1878         for (level = cur->bc_nlevels - 1, diff = 1; level >= 0; level--) {
1879                 /* Get the block we need to do the lookup on. */
1880                 error = xfs_btree_lookup_get_block(cur, level, pp, &block);
1881                 if (error)
1882                         goto error0;
1883
1884                 if (diff == 0) {
1885                         /*
1886                          * If we already had a key match at a higher level, we
1887                          * know we need to use the first entry in this block.
1888                          */
1889                         keyno = 1;
1890                 } else {
1891                         /* Otherwise search this block. Do a binary search. */
1892
1893                         int     high;   /* high entry number */
1894                         int     low;    /* low entry number */
1895
1896                         /* Set low and high entry numbers, 1-based. */
1897                         low = 1;
1898                         high = xfs_btree_get_numrecs(block);
1899                         if (!high) {
1900                                 /* Block is empty, must be an empty leaf. */
1901                                 ASSERT(level == 0 && cur->bc_nlevels == 1);
1902
1903                                 cur->bc_ptrs[0] = dir != XFS_LOOKUP_LE;
1904                                 XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
1905                                 *stat = 0;
1906                                 return 0;
1907                         }
1908
1909                         /* Binary search the block. */
1910                         while (low <= high) {
1911                                 union xfs_btree_key     key;
1912                                 union xfs_btree_key     *kp;
1913
1914                                 XFS_BTREE_STATS_INC(cur, compare);
1915
1916                                 /* keyno is average of low and high. */
1917                                 keyno = (low + high) >> 1;
1918
1919                                 /* Get current search key */
1920                                 kp = xfs_lookup_get_search_key(cur, level,
1921                                                 keyno, block, &key);
1922
1923                                 /*
1924                                  * Compute difference to get next direction:
1925                                  *  - less than, move right
1926                                  *  - greater than, move left
1927                                  *  - equal, we're done
1928                                  */
1929                                 diff = cur->bc_ops->key_diff(cur, kp);
1930                                 if (diff < 0)
1931                                         low = keyno + 1;
1932                                 else if (diff > 0)
1933                                         high = keyno - 1;
1934                                 else
1935                                         break;
1936                         }
1937                 }
1938
1939                 /*
1940                  * If there are more levels, set up for the next level
1941                  * by getting the block number and filling in the cursor.
1942                  */
1943                 if (level > 0) {
1944                         /*
1945                          * If we moved left, need the previous key number,
1946                          * unless there isn't one.
1947                          */
1948                         if (diff > 0 && --keyno < 1)
1949                                 keyno = 1;
1950                         pp = xfs_btree_ptr_addr(cur, keyno, block);
1951
1952 #ifdef DEBUG
1953                         error = xfs_btree_check_ptr(cur, pp, 0, level);
1954                         if (error)
1955                                 goto error0;
1956 #endif
1957                         cur->bc_ptrs[level] = keyno;
1958                 }
1959         }
1960
1961         /* Done with the search. See if we need to adjust the results. */
1962         if (dir != XFS_LOOKUP_LE && diff < 0) {
1963                 keyno++;
1964                 /*
1965                  * If ge search and we went off the end of the block, but it's
1966                  * not the last block, we're in the wrong block.
1967                  */
1968                 xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
1969                 if (dir == XFS_LOOKUP_GE &&
1970                     keyno > xfs_btree_get_numrecs(block) &&
1971                     !xfs_btree_ptr_is_null(cur, &ptr)) {
1972                         int     i;
1973
1974                         cur->bc_ptrs[0] = keyno;
1975                         error = xfs_btree_increment(cur, 0, &i);
1976                         if (error)
1977                                 goto error0;
1978                         XFS_WANT_CORRUPTED_RETURN(cur->bc_mp, i == 1);
1979                         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
1980                         *stat = 1;
1981                         return 0;
1982                 }
1983         } else if (dir == XFS_LOOKUP_LE && diff > 0)
1984                 keyno--;
1985         cur->bc_ptrs[0] = keyno;
1986
1987         /* Return if we succeeded or not. */
1988         if (keyno == 0 || keyno > xfs_btree_get_numrecs(block))
1989                 *stat = 0;
1990         else if (dir != XFS_LOOKUP_EQ || diff == 0)
1991                 *stat = 1;
1992         else
1993                 *stat = 0;
1994         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
1995         return 0;
1996
1997 error0:
1998         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
1999         return error;
2000 }
2001
2002 /* Find the high key storage area from a regular key. */
2003 STATIC union xfs_btree_key *
2004 xfs_btree_high_key_from_key(
2005         struct xfs_btree_cur    *cur,
2006         union xfs_btree_key     *key)
2007 {
2008         ASSERT(cur->bc_flags & XFS_BTREE_OVERLAPPING);
2009         return (union xfs_btree_key *)((char *)key +
2010                         (cur->bc_ops->key_len / 2));
2011 }
2012
2013 /* Determine the low (and high if overlapped) keys of a leaf block */
2014 STATIC void
2015 xfs_btree_get_leaf_keys(
2016         struct xfs_btree_cur    *cur,
2017         struct xfs_btree_block  *block,
2018         union xfs_btree_key     *key)
2019 {
2020         union xfs_btree_key     max_hkey;
2021         union xfs_btree_key     hkey;
2022         union xfs_btree_rec     *rec;
2023         union xfs_btree_key     *high;
2024         int                     n;
2025
2026         rec = xfs_btree_rec_addr(cur, 1, block);
2027         cur->bc_ops->init_key_from_rec(key, rec);
2028
2029         if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2030
2031                 cur->bc_ops->init_high_key_from_rec(&max_hkey, rec);
2032                 for (n = 2; n <= xfs_btree_get_numrecs(block); n++) {
2033                         rec = xfs_btree_rec_addr(cur, n, block);
2034                         cur->bc_ops->init_high_key_from_rec(&hkey, rec);
2035                         if (cur->bc_ops->diff_two_keys(cur, &hkey, &max_hkey)
2036                                         > 0)
2037                                 max_hkey = hkey;
2038                 }
2039
2040                 high = xfs_btree_high_key_from_key(cur, key);
2041                 memcpy(high, &max_hkey, cur->bc_ops->key_len / 2);
2042         }
2043 }
2044
2045 /* Determine the low (and high if overlapped) keys of a node block */
2046 STATIC void
2047 xfs_btree_get_node_keys(
2048         struct xfs_btree_cur    *cur,
2049         struct xfs_btree_block  *block,
2050         union xfs_btree_key     *key)
2051 {
2052         union xfs_btree_key     *hkey;
2053         union xfs_btree_key     *max_hkey;
2054         union xfs_btree_key     *high;
2055         int                     n;
2056
2057         if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2058                 memcpy(key, xfs_btree_key_addr(cur, 1, block),
2059                                 cur->bc_ops->key_len / 2);
2060
2061                 max_hkey = xfs_btree_high_key_addr(cur, 1, block);
2062                 for (n = 2; n <= xfs_btree_get_numrecs(block); n++) {
2063                         hkey = xfs_btree_high_key_addr(cur, n, block);
2064                         if (cur->bc_ops->diff_two_keys(cur, hkey, max_hkey) > 0)
2065                                 max_hkey = hkey;
2066                 }
2067
2068                 high = xfs_btree_high_key_from_key(cur, key);
2069                 memcpy(high, max_hkey, cur->bc_ops->key_len / 2);
2070         } else {
2071                 memcpy(key, xfs_btree_key_addr(cur, 1, block),
2072                                 cur->bc_ops->key_len);
2073         }
2074 }
2075
2076 /* Derive the keys for any btree block. */
2077 STATIC void
2078 xfs_btree_get_keys(
2079         struct xfs_btree_cur    *cur,
2080         struct xfs_btree_block  *block,
2081         union xfs_btree_key     *key)
2082 {
2083         if (be16_to_cpu(block->bb_level) == 0)
2084                 xfs_btree_get_leaf_keys(cur, block, key);
2085         else
2086                 xfs_btree_get_node_keys(cur, block, key);
2087 }
2088
2089 /*
2090  * Decide if we need to update the parent keys of a btree block.  For
2091  * a standard btree this is only necessary if we're updating the first
2092  * record/key.  For an overlapping btree, we must always update the
2093  * keys because the highest key can be in any of the records or keys
2094  * in the block.
2095  */
2096 static inline bool
2097 xfs_btree_needs_key_update(
2098         struct xfs_btree_cur    *cur,
2099         int                     ptr)
2100 {
2101         return (cur->bc_flags & XFS_BTREE_OVERLAPPING) || ptr == 1;
2102 }
2103
2104 /*
2105  * Update the low and high parent keys of the given level, progressing
2106  * towards the root.  If force_all is false, stop if the keys for a given
2107  * level do not need updating.
2108  */
2109 STATIC int
2110 __xfs_btree_updkeys(
2111         struct xfs_btree_cur    *cur,
2112         int                     level,
2113         struct xfs_btree_block  *block,
2114         struct xfs_buf          *bp0,
2115         bool                    force_all)
2116 {
2117         union xfs_btree_key     key;    /* keys from current level */
2118         union xfs_btree_key     *lkey;  /* keys from the next level up */
2119         union xfs_btree_key     *hkey;
2120         union xfs_btree_key     *nlkey; /* keys from the next level up */
2121         union xfs_btree_key     *nhkey;
2122         struct xfs_buf          *bp;
2123         int                     ptr;
2124
2125         ASSERT(cur->bc_flags & XFS_BTREE_OVERLAPPING);
2126
2127         /* Exit if there aren't any parent levels to update. */
2128         if (level + 1 >= cur->bc_nlevels)
2129                 return 0;
2130
2131         trace_xfs_btree_updkeys(cur, level, bp0);
2132
2133         lkey = &key;
2134         hkey = xfs_btree_high_key_from_key(cur, lkey);
2135         xfs_btree_get_keys(cur, block, lkey);
2136         for (level++; level < cur->bc_nlevels; level++) {
2137 #ifdef DEBUG
2138                 int             error;
2139 #endif
2140                 block = xfs_btree_get_block(cur, level, &bp);
2141                 trace_xfs_btree_updkeys(cur, level, bp);
2142 #ifdef DEBUG
2143                 error = xfs_btree_check_block(cur, block, level, bp);
2144                 if (error) {
2145                         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
2146                         return error;
2147                 }
2148 #endif
2149                 ptr = cur->bc_ptrs[level];
2150                 nlkey = xfs_btree_key_addr(cur, ptr, block);
2151                 nhkey = xfs_btree_high_key_addr(cur, ptr, block);
2152                 if (!force_all &&
2153                     !(cur->bc_ops->diff_two_keys(cur, nlkey, lkey) != 0 ||
2154                       cur->bc_ops->diff_two_keys(cur, nhkey, hkey) != 0))
2155                         break;
2156                 xfs_btree_copy_keys(cur, nlkey, lkey, 1);
2157                 xfs_btree_log_keys(cur, bp, ptr, ptr);
2158                 if (level + 1 >= cur->bc_nlevels)
2159                         break;
2160                 xfs_btree_get_node_keys(cur, block, lkey);
2161         }
2162
2163         return 0;
2164 }
2165
2166 /* Update all the keys from some level in cursor back to the root. */
2167 STATIC int
2168 xfs_btree_updkeys_force(
2169         struct xfs_btree_cur    *cur,
2170         int                     level)
2171 {
2172         struct xfs_buf          *bp;
2173         struct xfs_btree_block  *block;
2174
2175         block = xfs_btree_get_block(cur, level, &bp);
2176         return __xfs_btree_updkeys(cur, level, block, bp, true);
2177 }
2178
2179 /*
2180  * Update the parent keys of the given level, progressing towards the root.
2181  */
2182 STATIC int
2183 xfs_btree_update_keys(
2184         struct xfs_btree_cur    *cur,
2185         int                     level)
2186 {
2187         struct xfs_btree_block  *block;
2188         struct xfs_buf          *bp;
2189         union xfs_btree_key     *kp;
2190         union xfs_btree_key     key;
2191         int                     ptr;
2192
2193         ASSERT(level >= 0);
2194
2195         block = xfs_btree_get_block(cur, level, &bp);
2196         if (cur->bc_flags & XFS_BTREE_OVERLAPPING)
2197                 return __xfs_btree_updkeys(cur, level, block, bp, false);
2198
2199         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
2200         XFS_BTREE_TRACE_ARGIK(cur, level, keyp);
2201
2202         /*
2203          * Go up the tree from this level toward the root.
2204          * At each level, update the key value to the value input.
2205          * Stop when we reach a level where the cursor isn't pointing
2206          * at the first entry in the block.
2207          */
2208         xfs_btree_get_keys(cur, block, &key);
2209         for (level++, ptr = 1; ptr == 1 && level < cur->bc_nlevels; level++) {
2210 #ifdef DEBUG
2211                 int             error;
2212 #endif
2213                 block = xfs_btree_get_block(cur, level, &bp);
2214 #ifdef DEBUG
2215                 error = xfs_btree_check_block(cur, block, level, bp);
2216                 if (error) {
2217                         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
2218                         return error;
2219                 }
2220 #endif
2221                 ptr = cur->bc_ptrs[level];
2222                 kp = xfs_btree_key_addr(cur, ptr, block);
2223                 xfs_btree_copy_keys(cur, kp, &key, 1);
2224                 xfs_btree_log_keys(cur, bp, ptr, ptr);
2225         }
2226
2227         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
2228         return 0;
2229 }
2230
2231 /*
2232  * Update the record referred to by cur to the value in the
2233  * given record. This either works (return 0) or gets an
2234  * EFSCORRUPTED error.
2235  */
2236 int
2237 xfs_btree_update(
2238         struct xfs_btree_cur    *cur,
2239         union xfs_btree_rec     *rec)
2240 {
2241         struct xfs_btree_block  *block;
2242         struct xfs_buf          *bp;
2243         int                     error;
2244         int                     ptr;
2245         union xfs_btree_rec     *rp;
2246
2247         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
2248         XFS_BTREE_TRACE_ARGR(cur, rec);
2249
2250         /* Pick up the current block. */
2251         block = xfs_btree_get_block(cur, 0, &bp);
2252
2253 #ifdef DEBUG
2254         error = xfs_btree_check_block(cur, block, 0, bp);
2255         if (error)
2256                 goto error0;
2257 #endif
2258         /* Get the address of the rec to be updated. */
2259         ptr = cur->bc_ptrs[0];
2260         rp = xfs_btree_rec_addr(cur, ptr, block);
2261
2262         /* Fill in the new contents and log them. */
2263         xfs_btree_copy_recs(cur, rp, rec, 1);
2264         xfs_btree_log_recs(cur, bp, ptr, ptr);
2265
2266         /*
2267          * If we are tracking the last record in the tree and
2268          * we are at the far right edge of the tree, update it.
2269          */
2270         if (xfs_btree_is_lastrec(cur, block, 0)) {
2271                 cur->bc_ops->update_lastrec(cur, block, rec,
2272                                             ptr, LASTREC_UPDATE);
2273         }
2274
2275         /* Pass new key value up to our parent. */
2276         if (xfs_btree_needs_key_update(cur, ptr)) {
2277                 error = xfs_btree_update_keys(cur, 0);
2278                 if (error)
2279                         goto error0;
2280         }
2281
2282         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
2283         return 0;
2284
2285 error0:
2286         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
2287         return error;
2288 }
2289
2290 /*
2291  * Move 1 record left from cur/level if possible.
2292  * Update cur to reflect the new path.
2293  */
2294 STATIC int                                      /* error */
2295 xfs_btree_lshift(
2296         struct xfs_btree_cur    *cur,
2297         int                     level,
2298         int                     *stat)          /* success/failure */
2299 {
2300         struct xfs_buf          *lbp;           /* left buffer pointer */
2301         struct xfs_btree_block  *left;          /* left btree block */
2302         int                     lrecs;          /* left record count */
2303         struct xfs_buf          *rbp;           /* right buffer pointer */
2304         struct xfs_btree_block  *right;         /* right btree block */
2305         struct xfs_btree_cur    *tcur;          /* temporary btree cursor */
2306         int                     rrecs;          /* right record count */
2307         union xfs_btree_ptr     lptr;           /* left btree pointer */
2308         union xfs_btree_key     *rkp = NULL;    /* right btree key */
2309         union xfs_btree_ptr     *rpp = NULL;    /* right address pointer */
2310         union xfs_btree_rec     *rrp = NULL;    /* right record pointer */
2311         int                     error;          /* error return value */
2312         int                     i;
2313
2314         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
2315         XFS_BTREE_TRACE_ARGI(cur, level);
2316
2317         if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
2318             level == cur->bc_nlevels - 1)
2319                 goto out0;
2320
2321         /* Set up variables for this block as "right". */
2322         right = xfs_btree_get_block(cur, level, &rbp);
2323
2324 #ifdef DEBUG
2325         error = xfs_btree_check_block(cur, right, level, rbp);
2326         if (error)
2327                 goto error0;
2328 #endif
2329
2330         /* If we've got no left sibling then we can't shift an entry left. */
2331         xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
2332         if (xfs_btree_ptr_is_null(cur, &lptr))
2333                 goto out0;
2334
2335         /*
2336          * If the cursor entry is the one that would be moved, don't
2337          * do it... it's too complicated.
2338          */
2339         if (cur->bc_ptrs[level] <= 1)
2340                 goto out0;
2341
2342         /* Set up the left neighbor as "left". */
2343         error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
2344         if (error)
2345                 goto error0;
2346
2347         /* If it's full, it can't take another entry. */
2348         lrecs = xfs_btree_get_numrecs(left);
2349         if (lrecs == cur->bc_ops->get_maxrecs(cur, level))
2350                 goto out0;
2351
2352         rrecs = xfs_btree_get_numrecs(right);
2353
2354         /*
2355          * We add one entry to the left side and remove one for the right side.
2356          * Account for it here, the changes will be updated on disk and logged
2357          * later.
2358          */
2359         lrecs++;
2360         rrecs--;
2361
2362         XFS_BTREE_STATS_INC(cur, lshift);
2363         XFS_BTREE_STATS_ADD(cur, moves, 1);
2364
2365         /*
2366          * If non-leaf, copy a key and a ptr to the left block.
2367          * Log the changes to the left block.
2368          */
2369         if (level > 0) {
2370                 /* It's a non-leaf.  Move keys and pointers. */
2371                 union xfs_btree_key     *lkp;   /* left btree key */
2372                 union xfs_btree_ptr     *lpp;   /* left address pointer */
2373
2374                 lkp = xfs_btree_key_addr(cur, lrecs, left);
2375                 rkp = xfs_btree_key_addr(cur, 1, right);
2376
2377                 lpp = xfs_btree_ptr_addr(cur, lrecs, left);
2378                 rpp = xfs_btree_ptr_addr(cur, 1, right);
2379 #ifdef DEBUG
2380                 error = xfs_btree_check_ptr(cur, rpp, 0, level);
2381                 if (error)
2382                         goto error0;
2383 #endif
2384                 xfs_btree_copy_keys(cur, lkp, rkp, 1);
2385                 xfs_btree_copy_ptrs(cur, lpp, rpp, 1);
2386
2387                 xfs_btree_log_keys(cur, lbp, lrecs, lrecs);
2388                 xfs_btree_log_ptrs(cur, lbp, lrecs, lrecs);
2389
2390                 ASSERT(cur->bc_ops->keys_inorder(cur,
2391                         xfs_btree_key_addr(cur, lrecs - 1, left), lkp));
2392         } else {
2393                 /* It's a leaf.  Move records.  */
2394                 union xfs_btree_rec     *lrp;   /* left record pointer */
2395
2396                 lrp = xfs_btree_rec_addr(cur, lrecs, left);
2397                 rrp = xfs_btree_rec_addr(cur, 1, right);
2398
2399                 xfs_btree_copy_recs(cur, lrp, rrp, 1);
2400                 xfs_btree_log_recs(cur, lbp, lrecs, lrecs);
2401
2402                 ASSERT(cur->bc_ops->recs_inorder(cur,
2403                         xfs_btree_rec_addr(cur, lrecs - 1, left), lrp));
2404         }
2405
2406         xfs_btree_set_numrecs(left, lrecs);
2407         xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
2408
2409         xfs_btree_set_numrecs(right, rrecs);
2410         xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
2411
2412         /*
2413          * Slide the contents of right down one entry.
2414          */
2415         XFS_BTREE_STATS_ADD(cur, moves, rrecs - 1);
2416         if (level > 0) {
2417                 /* It's a nonleaf. operate on keys and ptrs */
2418 #ifdef DEBUG
2419                 int                     i;              /* loop index */
2420
2421                 for (i = 0; i < rrecs; i++) {
2422                         error = xfs_btree_check_ptr(cur, rpp, i + 1, level);
2423                         if (error)
2424                                 goto error0;
2425                 }
2426 #endif
2427                 xfs_btree_shift_keys(cur,
2428                                 xfs_btree_key_addr(cur, 2, right),
2429                                 -1, rrecs);
2430                 xfs_btree_shift_ptrs(cur,
2431                                 xfs_btree_ptr_addr(cur, 2, right),
2432                                 -1, rrecs);
2433
2434                 xfs_btree_log_keys(cur, rbp, 1, rrecs);
2435                 xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
2436         } else {
2437                 /* It's a leaf. operate on records */
2438                 xfs_btree_shift_recs(cur,
2439                         xfs_btree_rec_addr(cur, 2, right),
2440                         -1, rrecs);
2441                 xfs_btree_log_recs(cur, rbp, 1, rrecs);
2442         }
2443
2444         /*
2445          * Using a temporary cursor, update the parent key values of the
2446          * block on the left.
2447          */
2448         if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2449                 error = xfs_btree_dup_cursor(cur, &tcur);
2450                 if (error)
2451                         goto error0;
2452                 i = xfs_btree_firstrec(tcur, level);
2453                 XFS_WANT_CORRUPTED_GOTO(tcur->bc_mp, i == 1, error0);
2454
2455                 error = xfs_btree_decrement(tcur, level, &i);
2456                 if (error)
2457                         goto error1;
2458
2459                 /* Update the parent high keys of the left block, if needed. */
2460                 error = xfs_btree_update_keys(tcur, level);
2461                 if (error)
2462                         goto error1;
2463
2464                 xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
2465         }
2466
2467         /* Update the parent keys of the right block. */
2468         error = xfs_btree_update_keys(cur, level);
2469         if (error)
2470                 goto error0;
2471
2472         /* Slide the cursor value left one. */
2473         cur->bc_ptrs[level]--;
2474
2475         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
2476         *stat = 1;
2477         return 0;
2478
2479 out0:
2480         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
2481         *stat = 0;
2482         return 0;
2483
2484 error0:
2485         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
2486         return error;
2487
2488 error1:
2489         XFS_BTREE_TRACE_CURSOR(tcur, XBT_ERROR);
2490         xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
2491         return error;
2492 }
2493
2494 /*
2495  * Move 1 record right from cur/level if possible.
2496  * Update cur to reflect the new path.
2497  */
2498 STATIC int                                      /* error */
2499 xfs_btree_rshift(
2500         struct xfs_btree_cur    *cur,
2501         int                     level,
2502         int                     *stat)          /* success/failure */
2503 {
2504         struct xfs_buf          *lbp;           /* left buffer pointer */
2505         struct xfs_btree_block  *left;          /* left btree block */
2506         struct xfs_buf          *rbp;           /* right buffer pointer */
2507         struct xfs_btree_block  *right;         /* right btree block */
2508         struct xfs_btree_cur    *tcur;          /* temporary btree cursor */
2509         union xfs_btree_ptr     rptr;           /* right block pointer */
2510         union xfs_btree_key     *rkp;           /* right btree key */
2511         int                     rrecs;          /* right record count */
2512         int                     lrecs;          /* left record count */
2513         int                     error;          /* error return value */
2514         int                     i;              /* loop counter */
2515
2516         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
2517         XFS_BTREE_TRACE_ARGI(cur, level);
2518
2519         if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
2520             (level == cur->bc_nlevels - 1))
2521                 goto out0;
2522
2523         /* Set up variables for this block as "left". */
2524         left = xfs_btree_get_block(cur, level, &lbp);
2525
2526 #ifdef DEBUG
2527         error = xfs_btree_check_block(cur, left, level, lbp);
2528         if (error)
2529                 goto error0;
2530 #endif
2531
2532         /* If we've got no right sibling then we can't shift an entry right. */
2533         xfs_btree_get_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
2534         if (xfs_btree_ptr_is_null(cur, &rptr))
2535                 goto out0;
2536
2537         /*
2538          * If the cursor entry is the one that would be moved, don't
2539          * do it... it's too complicated.
2540          */
2541         lrecs = xfs_btree_get_numrecs(left);
2542         if (cur->bc_ptrs[level] >= lrecs)
2543                 goto out0;
2544
2545         /* Set up the right neighbor as "right". */
2546         error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
2547         if (error)
2548                 goto error0;
2549
2550         /* If it's full, it can't take another entry. */
2551         rrecs = xfs_btree_get_numrecs(right);
2552         if (rrecs == cur->bc_ops->get_maxrecs(cur, level))
2553                 goto out0;
2554
2555         XFS_BTREE_STATS_INC(cur, rshift);
2556         XFS_BTREE_STATS_ADD(cur, moves, rrecs);
2557
2558         /*
2559          * Make a hole at the start of the right neighbor block, then
2560          * copy the last left block entry to the hole.
2561          */
2562         if (level > 0) {
2563                 /* It's a nonleaf. make a hole in the keys and ptrs */
2564                 union xfs_btree_key     *lkp;
2565                 union xfs_btree_ptr     *lpp;
2566                 union xfs_btree_ptr     *rpp;
2567
2568                 lkp = xfs_btree_key_addr(cur, lrecs, left);
2569                 lpp = xfs_btree_ptr_addr(cur, lrecs, left);
2570                 rkp = xfs_btree_key_addr(cur, 1, right);
2571                 rpp = xfs_btree_ptr_addr(cur, 1, right);
2572
2573 #ifdef DEBUG
2574                 for (i = rrecs - 1; i >= 0; i--) {
2575                         error = xfs_btree_check_ptr(cur, rpp, i, level);
2576                         if (error)
2577                                 goto error0;
2578                 }
2579 #endif
2580
2581                 xfs_btree_shift_keys(cur, rkp, 1, rrecs);
2582                 xfs_btree_shift_ptrs(cur, rpp, 1, rrecs);
2583
2584 #ifdef DEBUG
2585                 error = xfs_btree_check_ptr(cur, lpp, 0, level);
2586                 if (error)
2587                         goto error0;
2588 #endif
2589
2590                 /* Now put the new data in, and log it. */
2591                 xfs_btree_copy_keys(cur, rkp, lkp, 1);
2592                 xfs_btree_copy_ptrs(cur, rpp, lpp, 1);
2593
2594                 xfs_btree_log_keys(cur, rbp, 1, rrecs + 1);
2595                 xfs_btree_log_ptrs(cur, rbp, 1, rrecs + 1);
2596
2597                 ASSERT(cur->bc_ops->keys_inorder(cur, rkp,
2598                         xfs_btree_key_addr(cur, 2, right)));
2599         } else {
2600                 /* It's a leaf. make a hole in the records */
2601                 union xfs_btree_rec     *lrp;
2602                 union xfs_btree_rec     *rrp;
2603
2604                 lrp = xfs_btree_rec_addr(cur, lrecs, left);
2605                 rrp = xfs_btree_rec_addr(cur, 1, right);
2606
2607                 xfs_btree_shift_recs(cur, rrp, 1, rrecs);
2608
2609                 /* Now put the new data in, and log it. */
2610                 xfs_btree_copy_recs(cur, rrp, lrp, 1);
2611                 xfs_btree_log_recs(cur, rbp, 1, rrecs + 1);
2612         }
2613
2614         /*
2615          * Decrement and log left's numrecs, bump and log right's numrecs.
2616          */
2617         xfs_btree_set_numrecs(left, --lrecs);
2618         xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
2619
2620         xfs_btree_set_numrecs(right, ++rrecs);
2621         xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
2622
2623         /*
2624          * Using a temporary cursor, update the parent key values of the
2625          * block on the right.
2626          */
2627         error = xfs_btree_dup_cursor(cur, &tcur);
2628         if (error)
2629                 goto error0;
2630         i = xfs_btree_lastrec(tcur, level);
2631         XFS_WANT_CORRUPTED_GOTO(tcur->bc_mp, i == 1, error0);
2632
2633         error = xfs_btree_increment(tcur, level, &i);
2634         if (error)
2635                 goto error1;
2636
2637         /* Update the parent high keys of the left block, if needed. */
2638         if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2639                 error = xfs_btree_update_keys(cur, level);
2640                 if (error)
2641                         goto error1;
2642         }
2643
2644         /* Update the parent keys of the right block. */
2645         error = xfs_btree_update_keys(tcur, level);
2646         if (error)
2647                 goto error1;
2648
2649         xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
2650
2651         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
2652         *stat = 1;
2653         return 0;
2654
2655 out0:
2656         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
2657         *stat = 0;
2658         return 0;
2659
2660 error0:
2661         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
2662         return error;
2663
2664 error1:
2665         XFS_BTREE_TRACE_CURSOR(tcur, XBT_ERROR);
2666         xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
2667         return error;
2668 }
2669
2670 /*
2671  * Split cur/level block in half.
2672  * Return new block number and the key to its first
2673  * record (to be inserted into parent).
2674  */
2675 STATIC int                                      /* error */
2676 __xfs_btree_split(
2677         struct xfs_btree_cur    *cur,
2678         int                     level,
2679         union xfs_btree_ptr     *ptrp,
2680         union xfs_btree_key     *key,
2681         struct xfs_btree_cur    **curp,
2682         int                     *stat)          /* success/failure */
2683 {
2684         union xfs_btree_ptr     lptr;           /* left sibling block ptr */
2685         struct xfs_buf          *lbp;           /* left buffer pointer */
2686         struct xfs_btree_block  *left;          /* left btree block */
2687         union xfs_btree_ptr     rptr;           /* right sibling block ptr */
2688         struct xfs_buf          *rbp;           /* right buffer pointer */
2689         struct xfs_btree_block  *right;         /* right btree block */
2690         union xfs_btree_ptr     rrptr;          /* right-right sibling ptr */
2691         struct xfs_buf          *rrbp;          /* right-right buffer pointer */
2692         struct xfs_btree_block  *rrblock;       /* right-right btree block */
2693         int                     lrecs;
2694         int                     rrecs;
2695         int                     src_index;
2696         int                     error;          /* error return value */
2697 #ifdef DEBUG
2698         int                     i;
2699 #endif
2700
2701         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
2702         XFS_BTREE_TRACE_ARGIPK(cur, level, *ptrp, key);
2703
2704         XFS_BTREE_STATS_INC(cur, split);
2705
2706         /* Set up left block (current one). */
2707         left = xfs_btree_get_block(cur, level, &lbp);
2708
2709 #ifdef DEBUG
2710         error = xfs_btree_check_block(cur, left, level, lbp);
2711         if (error)
2712                 goto error0;
2713 #endif
2714
2715         xfs_btree_buf_to_ptr(cur, lbp, &lptr);
2716
2717         /* Allocate the new block. If we can't do it, we're toast. Give up. */
2718         error = cur->bc_ops->alloc_block(cur, &lptr, &rptr, stat);
2719         if (error)
2720                 goto error0;
2721         if (*stat == 0)
2722                 goto out0;
2723         XFS_BTREE_STATS_INC(cur, alloc);
2724
2725         /* Set up the new block as "right". */
2726         error = xfs_btree_get_buf_block(cur, &rptr, 0, &right, &rbp);
2727         if (error)
2728                 goto error0;
2729
2730         /* Fill in the btree header for the new right block. */
2731         xfs_btree_init_block_cur(cur, rbp, xfs_btree_get_level(left), 0);
2732
2733         /*
2734          * Split the entries between the old and the new block evenly.
2735          * Make sure that if there's an odd number of entries now, that
2736          * each new block will have the same number of entries.
2737          */
2738         lrecs = xfs_btree_get_numrecs(left);
2739         rrecs = lrecs / 2;
2740         if ((lrecs & 1) && cur->bc_ptrs[level] <= rrecs + 1)
2741                 rrecs++;
2742         src_index = (lrecs - rrecs + 1);
2743
2744         XFS_BTREE_STATS_ADD(cur, moves, rrecs);
2745
2746         /* Adjust numrecs for the later get_*_keys() calls. */
2747         lrecs -= rrecs;
2748         xfs_btree_set_numrecs(left, lrecs);
2749         xfs_btree_set_numrecs(right, xfs_btree_get_numrecs(right) + rrecs);
2750
2751         /*
2752          * Copy btree block entries from the left block over to the
2753          * new block, the right. Update the right block and log the
2754          * changes.
2755          */
2756         if (level > 0) {
2757                 /* It's a non-leaf.  Move keys and pointers. */
2758                 union xfs_btree_key     *lkp;   /* left btree key */
2759                 union xfs_btree_ptr     *lpp;   /* left address pointer */
2760                 union xfs_btree_key     *rkp;   /* right btree key */
2761                 union xfs_btree_ptr     *rpp;   /* right address pointer */
2762
2763                 lkp = xfs_btree_key_addr(cur, src_index, left);
2764                 lpp = xfs_btree_ptr_addr(cur, src_index, left);
2765                 rkp = xfs_btree_key_addr(cur, 1, right);
2766                 rpp = xfs_btree_ptr_addr(cur, 1, right);
2767
2768 #ifdef DEBUG
2769                 for (i = src_index; i < rrecs; i++) {
2770                         error = xfs_btree_check_ptr(cur, lpp, i, level);
2771                         if (error)
2772                                 goto error0;
2773                 }
2774 #endif
2775
2776                 /* Copy the keys & pointers to the new block. */
2777                 xfs_btree_copy_keys(cur, rkp, lkp, rrecs);
2778                 xfs_btree_copy_ptrs(cur, rpp, lpp, rrecs);
2779
2780                 xfs_btree_log_keys(cur, rbp, 1, rrecs);
2781                 xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
2782
2783                 /* Stash the keys of the new block for later insertion. */
2784                 xfs_btree_get_node_keys(cur, right, key);
2785         } else {
2786                 /* It's a leaf.  Move records.  */
2787                 union xfs_btree_rec     *lrp;   /* left record pointer */
2788                 union xfs_btree_rec     *rrp;   /* right record pointer */
2789
2790                 lrp = xfs_btree_rec_addr(cur, src_index, left);
2791                 rrp = xfs_btree_rec_addr(cur, 1, right);
2792
2793                 /* Copy records to the new block. */
2794                 xfs_btree_copy_recs(cur, rrp, lrp, rrecs);
2795                 xfs_btree_log_recs(cur, rbp, 1, rrecs);
2796
2797                 /* Stash the keys of the new block for later insertion. */
2798                 xfs_btree_get_leaf_keys(cur, right, key);
2799         }
2800
2801         /*
2802          * Find the left block number by looking in the buffer.
2803          * Adjust sibling pointers.
2804          */
2805         xfs_btree_get_sibling(cur, left, &rrptr, XFS_BB_RIGHTSIB);
2806         xfs_btree_set_sibling(cur, right, &rrptr, XFS_BB_RIGHTSIB);
2807         xfs_btree_set_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
2808         xfs_btree_set_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
2809
2810         xfs_btree_log_block(cur, rbp, XFS_BB_ALL_BITS);
2811         xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
2812
2813         /*
2814          * If there's a block to the new block's right, make that block
2815          * point back to right instead of to left.
2816          */
2817         if (!xfs_btree_ptr_is_null(cur, &rrptr)) {
2818                 error = xfs_btree_read_buf_block(cur, &rrptr,
2819                                                         0, &rrblock, &rrbp);
2820                 if (error)
2821                         goto error0;
2822                 xfs_btree_set_sibling(cur, rrblock, &rptr, XFS_BB_LEFTSIB);
2823                 xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
2824         }
2825
2826         /* Update the parent high keys of the left block, if needed. */
2827         if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2828                 error = xfs_btree_update_keys(cur, level);
2829                 if (error)
2830                         goto error0;
2831         }
2832
2833         /*
2834          * If the cursor is really in the right block, move it there.
2835          * If it's just pointing past the last entry in left, then we'll
2836          * insert there, so don't change anything in that case.
2837          */
2838         if (cur->bc_ptrs[level] > lrecs + 1) {
2839                 xfs_btree_setbuf(cur, level, rbp);
2840                 cur->bc_ptrs[level] -= lrecs;
2841         }
2842         /*
2843          * If there are more levels, we'll need another cursor which refers
2844          * the right block, no matter where this cursor was.
2845          */
2846         if (level + 1 < cur->bc_nlevels) {
2847                 error = xfs_btree_dup_cursor(cur, curp);
2848                 if (error)
2849                         goto error0;
2850                 (*curp)->bc_ptrs[level + 1]++;
2851         }
2852         *ptrp = rptr;
2853         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
2854         *stat = 1;
2855         return 0;
2856 out0:
2857         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
2858         *stat = 0;
2859         return 0;
2860
2861 error0:
2862         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
2863         return error;
2864 }
2865
2866 struct xfs_btree_split_args {
2867         struct xfs_btree_cur    *cur;
2868         int                     level;
2869         union xfs_btree_ptr     *ptrp;
2870         union xfs_btree_key     *key;
2871         struct xfs_btree_cur    **curp;
2872         int                     *stat;          /* success/failure */
2873         int                     result;
2874         bool                    kswapd; /* allocation in kswapd context */
2875         struct completion       *done;
2876         struct work_struct      work;
2877 };
2878
2879 /*
2880  * Stack switching interfaces for allocation
2881  */
2882 static void
2883 xfs_btree_split_worker(
2884         struct work_struct      *work)
2885 {
2886         struct xfs_btree_split_args     *args = container_of(work,
2887                                                 struct xfs_btree_split_args, work);
2888         unsigned long           pflags;
2889         unsigned long           new_pflags = PF_MEMALLOC_NOFS;
2890
2891         /*
2892          * we are in a transaction context here, but may also be doing work
2893          * in kswapd context, and hence we may need to inherit that state
2894          * temporarily to ensure that we don't block waiting for memory reclaim
2895          * in any way.
2896          */
2897         if (args->kswapd)
2898                 new_pflags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD;
2899
2900         current_set_flags_nested(&pflags, new_pflags);
2901
2902         args->result = __xfs_btree_split(args->cur, args->level, args->ptrp,
2903                                          args->key, args->curp, args->stat);
2904         complete(args->done);
2905
2906         current_restore_flags_nested(&pflags, new_pflags);
2907 }
2908
2909 /*
2910  * BMBT split requests often come in with little stack to work on. Push
2911  * them off to a worker thread so there is lots of stack to use. For the other
2912  * btree types, just call directly to avoid the context switch overhead here.
2913  */
2914 STATIC int                                      /* error */
2915 xfs_btree_split(
2916         struct xfs_btree_cur    *cur,
2917         int                     level,
2918         union xfs_btree_ptr     *ptrp,
2919         union xfs_btree_key     *key,
2920         struct xfs_btree_cur    **curp,
2921         int                     *stat)          /* success/failure */
2922 {
2923         struct xfs_btree_split_args     args;
2924         DECLARE_COMPLETION_ONSTACK(done);
2925
2926         if (cur->bc_btnum != XFS_BTNUM_BMAP)
2927                 return __xfs_btree_split(cur, level, ptrp, key, curp, stat);
2928
2929         args.cur = cur;
2930         args.level = level;
2931         args.ptrp = ptrp;
2932         args.key = key;
2933         args.curp = curp;
2934         args.stat = stat;
2935         args.done = &done;
2936         args.kswapd = current_is_kswapd();
2937         INIT_WORK_ONSTACK(&args.work, xfs_btree_split_worker);
2938         queue_work(xfs_alloc_wq, &args.work);
2939         wait_for_completion(&done);
2940         destroy_work_on_stack(&args.work);
2941         return args.result;
2942 }
2943
2944
2945 /*
2946  * Copy the old inode root contents into a real block and make the
2947  * broot point to it.
2948  */
2949 int                                             /* error */
2950 xfs_btree_new_iroot(
2951         struct xfs_btree_cur    *cur,           /* btree cursor */
2952         int                     *logflags,      /* logging flags for inode */
2953         int                     *stat)          /* return status - 0 fail */
2954 {
2955         struct xfs_buf          *cbp;           /* buffer for cblock */
2956         struct xfs_btree_block  *block;         /* btree block */
2957         struct xfs_btree_block  *cblock;        /* child btree block */
2958         union xfs_btree_key     *ckp;           /* child key pointer */
2959         union xfs_btree_ptr     *cpp;           /* child ptr pointer */
2960         union xfs_btree_key     *kp;            /* pointer to btree key */
2961         union xfs_btree_ptr     *pp;            /* pointer to block addr */
2962         union xfs_btree_ptr     nptr;           /* new block addr */
2963         int                     level;          /* btree level */
2964         int                     error;          /* error return code */
2965 #ifdef DEBUG
2966         int                     i;              /* loop counter */
2967 #endif
2968
2969         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
2970         XFS_BTREE_STATS_INC(cur, newroot);
2971
2972         ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
2973
2974         level = cur->bc_nlevels - 1;
2975
2976         block = xfs_btree_get_iroot(cur);
2977         pp = xfs_btree_ptr_addr(cur, 1, block);
2978
2979         /* Allocate the new block. If we can't do it, we're toast. Give up. */
2980         error = cur->bc_ops->alloc_block(cur, pp, &nptr, stat);
2981         if (error)
2982                 goto error0;
2983         if (*stat == 0) {
2984                 XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
2985                 return 0;
2986         }
2987         XFS_BTREE_STATS_INC(cur, alloc);
2988
2989         /* Copy the root into a real block. */
2990         error = xfs_btree_get_buf_block(cur, &nptr, 0, &cblock, &cbp);
2991         if (error)
2992                 goto error0;
2993
2994         /*
2995          * we can't just memcpy() the root in for CRC enabled btree blocks.
2996          * In that case have to also ensure the blkno remains correct
2997          */
2998         memcpy(cblock, block, xfs_btree_block_len(cur));
2999         if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS) {
3000                 if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
3001                         cblock->bb_u.l.bb_blkno = cpu_to_be64(cbp->b_bn);
3002                 else
3003                         cblock->bb_u.s.bb_blkno = cpu_to_be64(cbp->b_bn);
3004         }
3005
3006         be16_add_cpu(&block->bb_level, 1);
3007         xfs_btree_set_numrecs(block, 1);
3008         cur->bc_nlevels++;
3009         cur->bc_ptrs[level + 1] = 1;
3010
3011         kp = xfs_btree_key_addr(cur, 1, block);
3012         ckp = xfs_btree_key_addr(cur, 1, cblock);
3013         xfs_btree_copy_keys(cur, ckp, kp, xfs_btree_get_numrecs(cblock));
3014
3015         cpp = xfs_btree_ptr_addr(cur, 1, cblock);
3016 #ifdef DEBUG
3017         for (i = 0; i < be16_to_cpu(cblock->bb_numrecs); i++) {
3018                 error = xfs_btree_check_ptr(cur, pp, i, level);
3019                 if (error)
3020                         goto error0;
3021         }
3022 #endif
3023         xfs_btree_copy_ptrs(cur, cpp, pp, xfs_btree_get_numrecs(cblock));
3024
3025 #ifdef DEBUG
3026         error = xfs_btree_check_ptr(cur, &nptr, 0, level);
3027         if (error)
3028                 goto error0;
3029 #endif
3030         xfs_btree_copy_ptrs(cur, pp, &nptr, 1);
3031
3032         xfs_iroot_realloc(cur->bc_private.b.ip,
3033                           1 - xfs_btree_get_numrecs(cblock),
3034                           cur->bc_private.b.whichfork);
3035
3036         xfs_btree_setbuf(cur, level, cbp);
3037
3038         /*
3039          * Do all this logging at the end so that
3040          * the root is at the right level.
3041          */
3042         xfs_btree_log_block(cur, cbp, XFS_BB_ALL_BITS);
3043         xfs_btree_log_keys(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
3044         xfs_btree_log_ptrs(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
3045
3046         *logflags |=
3047                 XFS_ILOG_CORE | xfs_ilog_fbroot(cur->bc_private.b.whichfork);
3048         *stat = 1;
3049         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
3050         return 0;
3051 error0:
3052         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
3053         return error;
3054 }
3055
3056 /*
3057  * Allocate a new root block, fill it in.
3058  */
3059 STATIC int                              /* error */
3060 xfs_btree_new_root(
3061         struct xfs_btree_cur    *cur,   /* btree cursor */
3062         int                     *stat)  /* success/failure */
3063 {
3064         struct xfs_btree_block  *block; /* one half of the old root block */
3065         struct xfs_buf          *bp;    /* buffer containing block */
3066         int                     error;  /* error return value */
3067         struct xfs_buf          *lbp;   /* left buffer pointer */
3068         struct xfs_btree_block  *left;  /* left btree block */
3069         struct xfs_buf          *nbp;   /* new (root) buffer */
3070         struct xfs_btree_block  *new;   /* new (root) btree block */
3071         int                     nptr;   /* new value for key index, 1 or 2 */
3072         struct xfs_buf          *rbp;   /* right buffer pointer */
3073         struct xfs_btree_block  *right; /* right btree block */
3074         union xfs_btree_ptr     rptr;
3075         union xfs_btree_ptr     lptr;
3076
3077         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
3078         XFS_BTREE_STATS_INC(cur, newroot);
3079
3080         /* initialise our start point from the cursor */
3081         cur->bc_ops->init_ptr_from_cur(cur, &rptr);
3082
3083         /* Allocate the new block. If we can't do it, we're toast. Give up. */
3084         error = cur->bc_ops->alloc_block(cur, &rptr, &lptr, stat);
3085         if (error)
3086                 goto error0;
3087         if (*stat == 0)
3088                 goto out0;
3089         XFS_BTREE_STATS_INC(cur, alloc);
3090
3091         /* Set up the new block. */
3092         error = xfs_btree_get_buf_block(cur, &lptr, 0, &new, &nbp);
3093         if (error)
3094                 goto error0;
3095
3096         /* Set the root in the holding structure  increasing the level by 1. */
3097         cur->bc_ops->set_root(cur, &lptr, 1);
3098
3099         /*
3100          * At the previous root level there are now two blocks: the old root,
3101          * and the new block generated when it was split.  We don't know which
3102          * one the cursor is pointing at, so we set up variables "left" and
3103          * "right" for each case.
3104          */
3105         block = xfs_btree_get_block(cur, cur->bc_nlevels - 1, &bp);
3106
3107 #ifdef DEBUG
3108         error = xfs_btree_check_block(cur, block, cur->bc_nlevels - 1, bp);
3109         if (error)
3110                 goto error0;
3111 #endif
3112
3113         xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
3114         if (!xfs_btree_ptr_is_null(cur, &rptr)) {
3115                 /* Our block is left, pick up the right block. */
3116                 lbp = bp;
3117                 xfs_btree_buf_to_ptr(cur, lbp, &lptr);
3118                 left = block;
3119                 error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
3120                 if (error)
3121                         goto error0;
3122                 bp = rbp;
3123                 nptr = 1;
3124         } else {
3125                 /* Our block is right, pick up the left block. */
3126                 rbp = bp;
3127                 xfs_btree_buf_to_ptr(cur, rbp, &rptr);
3128                 right = block;
3129                 xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
3130                 error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
3131                 if (error)
3132                         goto error0;
3133                 bp = lbp;
3134                 nptr = 2;
3135         }
3136
3137         /* Fill in the new block's btree header and log it. */
3138         xfs_btree_init_block_cur(cur, nbp, cur->bc_nlevels, 2);
3139         xfs_btree_log_block(cur, nbp, XFS_BB_ALL_BITS);
3140         ASSERT(!xfs_btree_ptr_is_null(cur, &lptr) &&
3141                         !xfs_btree_ptr_is_null(cur, &rptr));
3142
3143         /* Fill in the key data in the new root. */
3144         if (xfs_btree_get_level(left) > 0) {
3145                 /*
3146                  * Get the keys for the left block's keys and put them directly
3147                  * in the parent block.  Do the same for the right block.
3148                  */
3149                 xfs_btree_get_node_keys(cur, left,
3150                                 xfs_btree_key_addr(cur, 1, new));
3151                 xfs_btree_get_node_keys(cur, right,
3152                                 xfs_btree_key_addr(cur, 2, new));
3153         } else {
3154                 /*
3155                  * Get the keys for the left block's records and put them
3156                  * directly in the parent block.  Do the same for the right
3157                  * block.
3158                  */
3159                 xfs_btree_get_leaf_keys(cur, left,
3160                         xfs_btree_key_addr(cur, 1, new));
3161                 xfs_btree_get_leaf_keys(cur, right,
3162                         xfs_btree_key_addr(cur, 2, new));
3163         }
3164         xfs_btree_log_keys(cur, nbp, 1, 2);
3165
3166         /* Fill in the pointer data in the new root. */
3167         xfs_btree_copy_ptrs(cur,
3168                 xfs_btree_ptr_addr(cur, 1, new), &lptr, 1);
3169         xfs_btree_copy_ptrs(cur,
3170                 xfs_btree_ptr_addr(cur, 2, new), &rptr, 1);
3171         xfs_btree_log_ptrs(cur, nbp, 1, 2);
3172
3173         /* Fix up the cursor. */
3174         xfs_btree_setbuf(cur, cur->bc_nlevels, nbp);
3175         cur->bc_ptrs[cur->bc_nlevels] = nptr;
3176         cur->bc_nlevels++;
3177         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
3178         *stat = 1;
3179         return 0;
3180 error0:
3181         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
3182         return error;
3183 out0:
3184         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
3185         *stat = 0;
3186         return 0;
3187 }
3188
3189 STATIC int
3190 xfs_btree_make_block_unfull(
3191         struct xfs_btree_cur    *cur,   /* btree cursor */
3192         int                     level,  /* btree level */
3193         int                     numrecs,/* # of recs in block */
3194         int                     *oindex,/* old tree index */
3195         int                     *index, /* new tree index */
3196         union xfs_btree_ptr     *nptr,  /* new btree ptr */
3197         struct xfs_btree_cur    **ncur, /* new btree cursor */
3198         union xfs_btree_key     *key,   /* key of new block */
3199         int                     *stat)
3200 {
3201         int                     error = 0;
3202
3203         if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
3204             level == cur->bc_nlevels - 1) {
3205                 struct xfs_inode *ip = cur->bc_private.b.ip;
3206
3207                 if (numrecs < cur->bc_ops->get_dmaxrecs(cur, level)) {
3208                         /* A root block that can be made bigger. */
3209                         xfs_iroot_realloc(ip, 1, cur->bc_private.b.whichfork);
3210                         *stat = 1;
3211                 } else {
3212                         /* A root block that needs replacing */
3213                         int     logflags = 0;
3214
3215                         error = xfs_btree_new_iroot(cur, &logflags, stat);
3216                         if (error || *stat == 0)
3217                                 return error;
3218
3219                         xfs_trans_log_inode(cur->bc_tp, ip, logflags);
3220                 }
3221
3222                 return 0;
3223         }
3224
3225         /* First, try shifting an entry to the right neighbor. */
3226         error = xfs_btree_rshift(cur, level, stat);
3227         if (error || *stat)
3228                 return error;
3229
3230         /* Next, try shifting an entry to the left neighbor. */
3231         error = xfs_btree_lshift(cur, level, stat);
3232         if (error)
3233                 return error;
3234
3235         if (*stat) {
3236                 *oindex = *index = cur->bc_ptrs[level];
3237                 return 0;
3238         }
3239
3240         /*
3241          * Next, try splitting the current block in half.
3242          *
3243          * If this works we have to re-set our variables because we
3244          * could be in a different block now.
3245          */
3246         error = xfs_btree_split(cur, level, nptr, key, ncur, stat);
3247         if (error || *stat == 0)
3248                 return error;
3249
3250
3251         *index = cur->bc_ptrs[level];
3252         return 0;
3253 }
3254
3255 /*
3256  * Insert one record/level.  Return information to the caller
3257  * allowing the next level up to proceed if necessary.
3258  */
3259 STATIC int
3260 xfs_btree_insrec(
3261         struct xfs_btree_cur    *cur,   /* btree cursor */
3262         int                     level,  /* level to insert record at */
3263         union xfs_btree_ptr     *ptrp,  /* i/o: block number inserted */
3264         union xfs_btree_rec     *rec,   /* record to insert */
3265         union xfs_btree_key     *key,   /* i/o: block key for ptrp */
3266         struct xfs_btree_cur    **curp, /* output: new cursor replacing cur */
3267         int                     *stat)  /* success/failure */
3268 {
3269         struct xfs_btree_block  *block; /* btree block */
3270         struct xfs_buf          *bp;    /* buffer for block */
3271         union xfs_btree_ptr     nptr;   /* new block ptr */
3272         struct xfs_btree_cur    *ncur;  /* new btree cursor */
3273         union xfs_btree_key     nkey;   /* new block key */
3274         union xfs_btree_key     *lkey;
3275         int                     optr;   /* old key/record index */
3276         int                     ptr;    /* key/record index */
3277         int                     numrecs;/* number of records */
3278         int                     error;  /* error return value */
3279 #ifdef DEBUG
3280         int                     i;
3281 #endif
3282         xfs_daddr_t             old_bn;
3283
3284         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
3285         XFS_BTREE_TRACE_ARGIPR(cur, level, *ptrp, &rec);
3286
3287         ncur = NULL;
3288         lkey = &nkey;
3289
3290         /*
3291          * If we have an external root pointer, and we've made it to the
3292          * root level, allocate a new root block and we're done.
3293          */
3294         if (!(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
3295             (level >= cur->bc_nlevels)) {
3296                 error = xfs_btree_new_root(cur, stat);
3297                 xfs_btree_set_ptr_null(cur, ptrp);
3298
3299                 XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
3300                 return error;
3301         }
3302
3303         /* If we're off the left edge, return failure. */
3304         ptr = cur->bc_ptrs[level];
3305         if (ptr == 0) {
3306                 XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
3307                 *stat = 0;
3308                 return 0;
3309         }
3310
3311         optr = ptr;
3312
3313         XFS_BTREE_STATS_INC(cur, insrec);
3314
3315         /* Get pointers to the btree buffer and block. */
3316         block = xfs_btree_get_block(cur, level, &bp);
3317         old_bn = bp ? bp->b_bn : XFS_BUF_DADDR_NULL;
3318         numrecs = xfs_btree_get_numrecs(block);
3319
3320 #ifdef DEBUG
3321         error = xfs_btree_check_block(cur, block, level, bp);
3322         if (error)
3323                 goto error0;
3324
3325         /* Check that the new entry is being inserted in the right place. */
3326         if (ptr <= numrecs) {
3327                 if (level == 0) {
3328                         ASSERT(cur->bc_ops->recs_inorder(cur, rec,
3329                                 xfs_btree_rec_addr(cur, ptr, block)));
3330                 } else {
3331                         ASSERT(cur->bc_ops->keys_inorder(cur, key,
3332                                 xfs_btree_key_addr(cur, ptr, block)));
3333                 }
3334         }
3335 #endif
3336
3337         /*
3338          * If the block is full, we can't insert the new entry until we
3339          * make the block un-full.
3340          */
3341         xfs_btree_set_ptr_null(cur, &nptr);
3342         if (numrecs == cur->bc_ops->get_maxrecs(cur, level)) {
3343                 error = xfs_btree_make_block_unfull(cur, level, numrecs,
3344                                         &optr, &ptr, &nptr, &ncur, lkey, stat);
3345                 if (error || *stat == 0)
3346                         goto error0;
3347         }
3348
3349         /*
3350          * The current block may have changed if the block was
3351          * previously full and we have just made space in it.
3352          */
3353         block = xfs_btree_get_block(cur, level, &bp);
3354         numrecs = xfs_btree_get_numrecs(block);
3355
3356 #ifdef DEBUG
3357         error = xfs_btree_check_block(cur, block, level, bp);
3358         if (error)
3359                 return error;
3360 #endif
3361
3362         /*
3363          * At this point we know there's room for our new entry in the block
3364          * we're pointing at.
3365          */
3366         XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr + 1);
3367
3368         if (level > 0) {
3369                 /* It's a nonleaf. make a hole in the keys and ptrs */
3370                 union xfs_btree_key     *kp;
3371                 union xfs_btree_ptr     *pp;
3372
3373                 kp = xfs_btree_key_addr(cur, ptr, block);
3374                 pp = xfs_btree_ptr_addr(cur, ptr, block);
3375
3376 #ifdef DEBUG
3377                 for (i = numrecs - ptr; i >= 0; i--) {
3378                         error = xfs_btree_check_ptr(cur, pp, i, level);
3379                         if (error)
3380                                 return error;
3381                 }
3382 #endif
3383
3384                 xfs_btree_shift_keys(cur, kp, 1, numrecs - ptr + 1);
3385                 xfs_btree_shift_ptrs(cur, pp, 1, numrecs - ptr + 1);
3386
3387 #ifdef DEBUG
3388                 error = xfs_btree_check_ptr(cur, ptrp, 0, level);
3389                 if (error)
3390                         goto error0;
3391 #endif
3392
3393                 /* Now put the new data in, bump numrecs and log it. */
3394                 xfs_btree_copy_keys(cur, kp, key, 1);
3395                 xfs_btree_copy_ptrs(cur, pp, ptrp, 1);
3396                 numrecs++;
3397                 xfs_btree_set_numrecs(block, numrecs);
3398                 xfs_btree_log_ptrs(cur, bp, ptr, numrecs);
3399                 xfs_btree_log_keys(cur, bp, ptr, numrecs);
3400 #ifdef DEBUG
3401                 if (ptr < numrecs) {
3402                         ASSERT(cur->bc_ops->keys_inorder(cur, kp,
3403                                 xfs_btree_key_addr(cur, ptr + 1, block)));
3404                 }
3405 #endif
3406         } else {
3407                 /* It's a leaf. make a hole in the records */
3408                 union xfs_btree_rec             *rp;
3409
3410                 rp = xfs_btree_rec_addr(cur, ptr, block);
3411
3412                 xfs_btree_shift_recs(cur, rp, 1, numrecs - ptr + 1);
3413
3414                 /* Now put the new data in, bump numrecs and log it. */
3415                 xfs_btree_copy_recs(cur, rp, rec, 1);
3416                 xfs_btree_set_numrecs(block, ++numrecs);
3417                 xfs_btree_log_recs(cur, bp, ptr, numrecs);
3418 #ifdef DEBUG
3419                 if (ptr < numrecs) {
3420                         ASSERT(cur->bc_ops->recs_inorder(cur, rp,
3421                                 xfs_btree_rec_addr(cur, ptr + 1, block)));
3422                 }
3423 #endif
3424         }
3425
3426         /* Log the new number of records in the btree header. */
3427         xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
3428
3429         /*
3430          * If we just inserted into a new tree block, we have to
3431          * recalculate nkey here because nkey is out of date.
3432          *
3433          * Otherwise we're just updating an existing block (having shoved
3434          * some records into the new tree block), so use the regular key
3435          * update mechanism.
3436          */
3437         if (bp && bp->b_bn != old_bn) {
3438                 xfs_btree_get_keys(cur, block, lkey);
3439         } else if (xfs_btree_needs_key_update(cur, optr)) {
3440                 error = xfs_btree_update_keys(cur, level);
3441                 if (error)
3442                         goto error0;
3443         }
3444
3445         /*
3446          * If we are tracking the last record in the tree and
3447          * we are at the far right edge of the tree, update it.
3448          */
3449         if (xfs_btree_is_lastrec(cur, block, level)) {
3450                 cur->bc_ops->update_lastrec(cur, block, rec,
3451                                             ptr, LASTREC_INSREC);
3452         }
3453
3454         /*
3455          * Return the new block number, if any.
3456          * If there is one, give back a record value and a cursor too.
3457          */
3458         *ptrp = nptr;
3459         if (!xfs_btree_ptr_is_null(cur, &nptr)) {
3460                 xfs_btree_copy_keys(cur, key, lkey, 1);
3461                 *curp = ncur;
3462         }
3463
3464         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
3465         *stat = 1;
3466         return 0;
3467
3468 error0:
3469         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
3470         return error;
3471 }
3472
3473 /*
3474  * Insert the record at the point referenced by cur.
3475  *
3476  * A multi-level split of the tree on insert will invalidate the original
3477  * cursor.  All callers of this function should assume that the cursor is
3478  * no longer valid and revalidate it.
3479  */
3480 int
3481 xfs_btree_insert(
3482         struct xfs_btree_cur    *cur,
3483         int                     *stat)
3484 {
3485         int                     error;  /* error return value */
3486         int                     i;      /* result value, 0 for failure */
3487         int                     level;  /* current level number in btree */
3488         union xfs_btree_ptr     nptr;   /* new block number (split result) */
3489         struct xfs_btree_cur    *ncur;  /* new cursor (split result) */
3490         struct xfs_btree_cur    *pcur;  /* previous level's cursor */
3491         union xfs_btree_key     bkey;   /* key of block to insert */
3492         union xfs_btree_key     *key;
3493         union xfs_btree_rec     rec;    /* record to insert */
3494
3495         level = 0;
3496         ncur = NULL;
3497         pcur = cur;
3498         key = &bkey;
3499
3500         xfs_btree_set_ptr_null(cur, &nptr);
3501
3502         /* Make a key out of the record data to be inserted, and save it. */
3503         cur->bc_ops->init_rec_from_cur(cur, &rec);
3504         cur->bc_ops->init_key_from_rec(key, &rec);
3505
3506         /*
3507          * Loop going up the tree, starting at the leaf level.
3508          * Stop when we don't get a split block, that must mean that
3509          * the insert is finished with this level.
3510          */
3511         do {
3512                 /*
3513                  * Insert nrec/nptr into this level of the tree.
3514                  * Note if we fail, nptr will be null.
3515                  */
3516                 error = xfs_btree_insrec(pcur, level, &nptr, &rec, key,
3517                                 &ncur, &i);
3518                 if (error) {
3519                         if (pcur != cur)
3520                                 xfs_btree_del_cursor(pcur, XFS_BTREE_ERROR);
3521                         goto error0;
3522                 }
3523
3524                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
3525                 level++;
3526
3527                 /*
3528                  * See if the cursor we just used is trash.
3529                  * Can't trash the caller's cursor, but otherwise we should
3530                  * if ncur is a new cursor or we're about to be done.
3531                  */
3532                 if (pcur != cur &&
3533                     (ncur || xfs_btree_ptr_is_null(cur, &nptr))) {
3534                         /* Save the state from the cursor before we trash it */
3535                         if (cur->bc_ops->update_cursor)
3536                                 cur->bc_ops->update_cursor(pcur, cur);
3537                         cur->bc_nlevels = pcur->bc_nlevels;
3538                         xfs_btree_del_cursor(pcur, XFS_BTREE_NOERROR);
3539                 }
3540                 /* If we got a new cursor, switch to it. */
3541                 if (ncur) {
3542                         pcur = ncur;
3543                         ncur = NULL;
3544                 }
3545         } while (!xfs_btree_ptr_is_null(cur, &nptr));
3546
3547         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
3548         *stat = i;
3549         return 0;
3550 error0:
3551         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
3552         return error;
3553 }
3554
3555 /*
3556  * Try to merge a non-leaf block back into the inode root.
3557  *
3558  * Note: the killroot names comes from the fact that we're effectively
3559  * killing the old root block.  But because we can't just delete the
3560  * inode we have to copy the single block it was pointing to into the
3561  * inode.
3562  */
3563 STATIC int
3564 xfs_btree_kill_iroot(
3565         struct xfs_btree_cur    *cur)
3566 {
3567         int                     whichfork = cur->bc_private.b.whichfork;
3568         struct xfs_inode        *ip = cur->bc_private.b.ip;
3569         struct xfs_ifork        *ifp = XFS_IFORK_PTR(ip, whichfork);
3570         struct xfs_btree_block  *block;
3571         struct xfs_btree_block  *cblock;
3572         union xfs_btree_key     *kp;
3573         union xfs_btree_key     *ckp;
3574         union xfs_btree_ptr     *pp;
3575         union xfs_btree_ptr     *cpp;
3576         struct xfs_buf          *cbp;
3577         int                     level;
3578         int                     index;
3579         int                     numrecs;
3580         int                     error;
3581 #ifdef DEBUG
3582         union xfs_btree_ptr     ptr;
3583         int                     i;
3584 #endif
3585
3586         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
3587
3588         ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
3589         ASSERT(cur->bc_nlevels > 1);
3590
3591         /*
3592          * Don't deal with the root block needs to be a leaf case.
3593          * We're just going to turn the thing back into extents anyway.
3594          */
3595         level = cur->bc_nlevels - 1;
3596         if (level == 1)
3597                 goto out0;
3598
3599         /*
3600          * Give up if the root has multiple children.
3601          */
3602         block = xfs_btree_get_iroot(cur);
3603         if (xfs_btree_get_numrecs(block) != 1)
3604                 goto out0;
3605
3606         cblock = xfs_btree_get_block(cur, level - 1, &cbp);
3607         numrecs = xfs_btree_get_numrecs(cblock);
3608
3609         /*
3610          * Only do this if the next level will fit.
3611          * Then the data must be copied up to the inode,
3612          * instead of freeing the root you free the next level.
3613          */
3614         if (numrecs > cur->bc_ops->get_dmaxrecs(cur, level))
3615                 goto out0;
3616
3617         XFS_BTREE_STATS_INC(cur, killroot);
3618
3619 #ifdef DEBUG
3620         xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
3621         ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
3622         xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
3623         ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
3624 #endif
3625
3626         index = numrecs - cur->bc_ops->get_maxrecs(cur, level);
3627         if (index) {
3628                 xfs_iroot_realloc(cur->bc_private.b.ip, index,
3629                                   cur->bc_private.b.whichfork);
3630                 block = ifp->if_broot;
3631         }
3632
3633         be16_add_cpu(&block->bb_numrecs, index);
3634         ASSERT(block->bb_numrecs == cblock->bb_numrecs);
3635
3636         kp = xfs_btree_key_addr(cur, 1, block);
3637         ckp = xfs_btree_key_addr(cur, 1, cblock);
3638         xfs_btree_copy_keys(cur, kp, ckp, numrecs);
3639
3640         pp = xfs_btree_ptr_addr(cur, 1, block);
3641         cpp = xfs_btree_ptr_addr(cur, 1, cblock);
3642 #ifdef DEBUG
3643         for (i = 0; i < numrecs; i++) {
3644                 error = xfs_btree_check_ptr(cur, cpp, i, level - 1);
3645                 if (error) {
3646                         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
3647                         return error;
3648                 }
3649         }
3650 #endif
3651         xfs_btree_copy_ptrs(cur, pp, cpp, numrecs);
3652
3653         error = xfs_btree_free_block(cur, cbp);
3654         if (error) {
3655                 XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
3656                 return error;
3657         }
3658
3659         cur->bc_bufs[level - 1] = NULL;
3660         be16_add_cpu(&block->bb_level, -1);
3661         xfs_trans_log_inode(cur->bc_tp, ip,
3662                 XFS_ILOG_CORE | xfs_ilog_fbroot(cur->bc_private.b.whichfork));
3663         cur->bc_nlevels--;
3664 out0:
3665         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
3666         return 0;
3667 }
3668
3669 /*
3670  * Kill the current root node, and replace it with it's only child node.
3671  */
3672 STATIC int
3673 xfs_btree_kill_root(
3674         struct xfs_btree_cur    *cur,
3675         struct xfs_buf          *bp,
3676         int                     level,
3677         union xfs_btree_ptr     *newroot)
3678 {
3679         int                     error;
3680
3681         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
3682         XFS_BTREE_STATS_INC(cur, killroot);
3683
3684         /*
3685          * Update the root pointer, decreasing the level by 1 and then
3686          * free the old root.
3687          */
3688         cur->bc_ops->set_root(cur, newroot, -1);
3689
3690         error = xfs_btree_free_block(cur, bp);
3691         if (error) {
3692                 XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
3693                 return error;
3694         }
3695
3696         cur->bc_bufs[level] = NULL;
3697         cur->bc_ra[level] = 0;
3698         cur->bc_nlevels--;
3699
3700         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
3701         return 0;
3702 }
3703
3704 STATIC int
3705 xfs_btree_dec_cursor(
3706         struct xfs_btree_cur    *cur,
3707         int                     level,
3708         int                     *stat)
3709 {
3710         int                     error;
3711         int                     i;
3712
3713         if (level > 0) {
3714                 error = xfs_btree_decrement(cur, level, &i);
3715                 if (error)
3716                         return error;
3717         }
3718
3719         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
3720         *stat = 1;
3721         return 0;
3722 }
3723
3724 /*
3725  * Single level of the btree record deletion routine.
3726  * Delete record pointed to by cur/level.
3727  * Remove the record from its block then rebalance the tree.
3728  * Return 0 for error, 1 for done, 2 to go on to the next level.
3729  */
3730 STATIC int                                      /* error */
3731 xfs_btree_delrec(
3732         struct xfs_btree_cur    *cur,           /* btree cursor */
3733         int                     level,          /* level removing record from */
3734         int                     *stat)          /* fail/done/go-on */
3735 {
3736         struct xfs_btree_block  *block;         /* btree block */
3737         union xfs_btree_ptr     cptr;           /* current block ptr */
3738         struct xfs_buf          *bp;            /* buffer for block */
3739         int                     error;          /* error return value */
3740         int                     i;              /* loop counter */
3741         union xfs_btree_ptr     lptr;           /* left sibling block ptr */
3742         struct xfs_buf          *lbp;           /* left buffer pointer */
3743         struct xfs_btree_block  *left;          /* left btree block */
3744         int                     lrecs = 0;      /* left record count */
3745         int                     ptr;            /* key/record index */
3746         union xfs_btree_ptr     rptr;           /* right sibling block ptr */
3747         struct xfs_buf          *rbp;           /* right buffer pointer */
3748         struct xfs_btree_block  *right;         /* right btree block */
3749         struct xfs_btree_block  *rrblock;       /* right-right btree block */
3750         struct xfs_buf          *rrbp;          /* right-right buffer pointer */
3751         int                     rrecs = 0;      /* right record count */
3752         struct xfs_btree_cur    *tcur;          /* temporary btree cursor */
3753         int                     numrecs;        /* temporary numrec count */
3754
3755         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
3756         XFS_BTREE_TRACE_ARGI(cur, level);
3757
3758         tcur = NULL;
3759
3760         /* Get the index of the entry being deleted, check for nothing there. */
3761         ptr = cur->bc_ptrs[level];
3762         if (ptr == 0) {
3763                 XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
3764                 *stat = 0;
3765                 return 0;
3766         }
3767
3768         /* Get the buffer & block containing the record or key/ptr. */
3769         block = xfs_btree_get_block(cur, level, &bp);
3770         numrecs = xfs_btree_get_numrecs(block);
3771
3772 #ifdef DEBUG
3773         error = xfs_btree_check_block(cur, block, level, bp);
3774         if (error)
3775                 goto error0;
3776 #endif
3777
3778         /* Fail if we're off the end of the block. */
3779         if (ptr > numrecs) {
3780                 XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
3781                 *stat = 0;
3782                 return 0;
3783         }
3784
3785         XFS_BTREE_STATS_INC(cur, delrec);
3786         XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr);
3787
3788         /* Excise the entries being deleted. */
3789         if (level > 0) {
3790                 /* It's a nonleaf. operate on keys and ptrs */
3791                 union xfs_btree_key     *lkp;
3792                 union xfs_btree_ptr     *lpp;
3793
3794                 lkp = xfs_btree_key_addr(cur, ptr + 1, block);
3795                 lpp = xfs_btree_ptr_addr(cur, ptr + 1, block);
3796
3797 #ifdef DEBUG
3798                 for (i = 0; i < numrecs - ptr; i++) {
3799                         error = xfs_btree_check_ptr(cur, lpp, i, level);
3800                         if (error)
3801                                 goto error0;
3802                 }
3803 #endif
3804
3805                 if (ptr < numrecs) {
3806                         xfs_btree_shift_keys(cur, lkp, -1, numrecs - ptr);
3807                         xfs_btree_shift_ptrs(cur, lpp, -1, numrecs - ptr);
3808                         xfs_btree_log_keys(cur, bp, ptr, numrecs - 1);
3809                         xfs_btree_log_ptrs(cur, bp, ptr, numrecs - 1);
3810                 }
3811         } else {
3812                 /* It's a leaf. operate on records */
3813                 if (ptr < numrecs) {
3814                         xfs_btree_shift_recs(cur,
3815                                 xfs_btree_rec_addr(cur, ptr + 1, block),
3816                                 -1, numrecs - ptr);
3817                         xfs_btree_log_recs(cur, bp, ptr, numrecs - 1);
3818                 }
3819         }
3820
3821         /*
3822          * Decrement and log the number of entries in the block.
3823          */
3824         xfs_btree_set_numrecs(block, --numrecs);
3825         xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
3826
3827         /*
3828          * If we are tracking the last record in the tree and
3829          * we are at the far right edge of the tree, update it.
3830          */
3831         if (xfs_btree_is_lastrec(cur, block, level)) {
3832                 cur->bc_ops->update_lastrec(cur, block, NULL,
3833                                             ptr, LASTREC_DELREC);
3834         }
3835
3836         /*
3837          * We're at the root level.  First, shrink the root block in-memory.
3838          * Try to get rid of the next level down.  If we can't then there's
3839          * nothing left to do.
3840          */
3841         if (level == cur->bc_nlevels - 1) {
3842                 if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
3843                         xfs_iroot_realloc(cur->bc_private.b.ip, -1,
3844                                           cur->bc_private.b.whichfork);
3845
3846                         error = xfs_btree_kill_iroot(cur);
3847                         if (error)
3848                                 goto error0;
3849
3850                         error = xfs_btree_dec_cursor(cur, level, stat);
3851                         if (error)
3852                                 goto error0;
3853                         *stat = 1;
3854                         return 0;
3855                 }
3856
3857                 /*
3858                  * If this is the root level, and there's only one entry left,
3859                  * and it's NOT the leaf level, then we can get rid of this
3860                  * level.
3861                  */
3862                 if (numrecs == 1 && level > 0) {
3863                         union xfs_btree_ptr     *pp;
3864                         /*
3865                          * pp is still set to the first pointer in the block.
3866                          * Make it the new root of the btree.
3867                          */
3868                         pp = xfs_btree_ptr_addr(cur, 1, block);
3869                         error = xfs_btree_kill_root(cur, bp, level, pp);
3870                         if (error)
3871                                 goto error0;
3872                 } else if (level > 0) {
3873                         error = xfs_btree_dec_cursor(cur, level, stat);
3874                         if (error)
3875                                 goto error0;
3876                 }
3877                 *stat = 1;
3878                 return 0;
3879         }
3880
3881         /*
3882          * If we deleted the leftmost entry in the block, update the
3883          * key values above us in the tree.
3884          */
3885         if (xfs_btree_needs_key_update(cur, ptr)) {
3886                 error = xfs_btree_update_keys(cur, level);
3887                 if (error)
3888                         goto error0;
3889         }
3890
3891         /*
3892          * If the number of records remaining in the block is at least
3893          * the minimum, we're done.
3894          */
3895         if (numrecs >= cur->bc_ops->get_minrecs(cur, level)) {
3896                 error = xfs_btree_dec_cursor(cur, level, stat);
3897                 if (error)
3898                         goto error0;
3899                 return 0;
3900         }
3901
3902         /*
3903          * Otherwise, we have to move some records around to keep the
3904          * tree balanced.  Look at the left and right sibling blocks to
3905          * see if we can re-balance by moving only one record.
3906          */
3907         xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
3908         xfs_btree_get_sibling(cur, block, &lptr, XFS_BB_LEFTSIB);
3909
3910         if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
3911                 /*
3912                  * One child of root, need to get a chance to copy its contents
3913                  * into the root and delete it. Can't go up to next level,
3914                  * there's nothing to delete there.
3915                  */
3916                 if (xfs_btree_ptr_is_null(cur, &rptr) &&
3917                     xfs_btree_ptr_is_null(cur, &lptr) &&
3918                     level == cur->bc_nlevels - 2) {
3919                         error = xfs_btree_kill_iroot(cur);
3920                         if (!error)
3921                                 error = xfs_btree_dec_cursor(cur, level, stat);
3922                         if (error)
3923                                 goto error0;
3924                         return 0;
3925                 }
3926         }
3927
3928         ASSERT(!xfs_btree_ptr_is_null(cur, &rptr) ||
3929                !xfs_btree_ptr_is_null(cur, &lptr));
3930
3931         /*
3932          * Duplicate the cursor so our btree manipulations here won't
3933          * disrupt the next level up.
3934          */
3935         error = xfs_btree_dup_cursor(cur, &tcur);
3936         if (error)
3937                 goto error0;
3938
3939         /*
3940          * If there's a right sibling, see if it's ok to shift an entry
3941          * out of it.
3942          */
3943         if (!xfs_btree_ptr_is_null(cur, &rptr)) {
3944                 /*
3945                  * Move the temp cursor to the last entry in the next block.
3946                  * Actually any entry but the first would suffice.
3947                  */
3948                 i = xfs_btree_lastrec(tcur, level);
3949                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
3950
3951                 error = xfs_btree_increment(tcur, level, &i);
3952                 if (error)
3953                         goto error0;
3954                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
3955
3956                 i = xfs_btree_lastrec(tcur, level);
3957                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
3958
3959                 /* Grab a pointer to the block. */
3960                 right = xfs_btree_get_block(tcur, level, &rbp);
3961 #ifdef DEBUG
3962                 error = xfs_btree_check_block(tcur, right, level, rbp);
3963                 if (error)
3964                         goto error0;
3965 #endif
3966                 /* Grab the current block number, for future use. */
3967                 xfs_btree_get_sibling(tcur, right, &cptr, XFS_BB_LEFTSIB);
3968
3969                 /*
3970                  * If right block is full enough so that removing one entry
3971                  * won't make it too empty, and left-shifting an entry out
3972                  * of right to us works, we're done.
3973                  */
3974                 if (xfs_btree_get_numrecs(right) - 1 >=
3975                     cur->bc_ops->get_minrecs(tcur, level)) {
3976                         error = xfs_btree_lshift(tcur, level, &i);
3977                         if (error)
3978                                 goto error0;
3979                         if (i) {
3980                                 ASSERT(xfs_btree_get_numrecs(block) >=
3981                                        cur->bc_ops->get_minrecs(tcur, level));
3982
3983                                 xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
3984                                 tcur = NULL;
3985
3986                                 error = xfs_btree_dec_cursor(cur, level, stat);
3987                                 if (error)
3988                                         goto error0;
3989                                 return 0;
3990                         }
3991                 }
3992
3993                 /*
3994                  * Otherwise, grab the number of records in right for
3995                  * future reference, and fix up the temp cursor to point
3996                  * to our block again (last record).
3997                  */
3998                 rrecs = xfs_btree_get_numrecs(right);
3999                 if (!xfs_btree_ptr_is_null(cur, &lptr)) {
4000                         i = xfs_btree_firstrec(tcur, level);
4001                         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
4002
4003                         error = xfs_btree_decrement(tcur, level, &i);
4004                         if (error)
4005                                 goto error0;
4006                         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
4007                 }
4008         }
4009
4010         /*
4011          * If there's a left sibling, see if it's ok to shift an entry
4012          * out of it.
4013          */
4014         if (!xfs_btree_ptr_is_null(cur, &lptr)) {
4015                 /*
4016                  * Move the temp cursor to the first entry in the
4017                  * previous block.
4018                  */
4019                 i = xfs_btree_firstrec(tcur, level);
4020                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
4021
4022                 error = xfs_btree_decrement(tcur, level, &i);
4023                 if (error)
4024                         goto error0;
4025                 i = xfs_btree_firstrec(tcur, level);
4026                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
4027
4028                 /* Grab a pointer to the block. */
4029                 left = xfs_btree_get_block(tcur, level, &lbp);
4030 #ifdef DEBUG
4031                 error = xfs_btree_check_block(cur, left, level, lbp);
4032                 if (error)
4033                         goto error0;
4034 #endif
4035                 /* Grab the current block number, for future use. */
4036                 xfs_btree_get_sibling(tcur, left, &cptr, XFS_BB_RIGHTSIB);
4037
4038                 /*
4039                  * If left block is full enough so that removing one entry
4040                  * won't make it too empty, and right-shifting an entry out
4041                  * of left to us works, we're done.
4042                  */
4043                 if (xfs_btree_get_numrecs(left) - 1 >=
4044                     cur->bc_ops->get_minrecs(tcur, level)) {
4045                         error = xfs_btree_rshift(tcur, level, &i);
4046                         if (error)
4047                                 goto error0;
4048                         if (i) {
4049                                 ASSERT(xfs_btree_get_numrecs(block) >=
4050                                        cur->bc_ops->get_minrecs(tcur, level));
4051                                 xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
4052                                 tcur = NULL;
4053                                 if (level == 0)
4054                                         cur->bc_ptrs[0]++;
4055                                 XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
4056                                 *stat = 1;
4057                                 return 0;
4058                         }
4059                 }
4060
4061                 /*
4062                  * Otherwise, grab the number of records in right for
4063                  * future reference.
4064                  */
4065                 lrecs = xfs_btree_get_numrecs(left);
4066         }
4067
4068         /* Delete the temp cursor, we're done with it. */
4069         xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
4070         tcur = NULL;
4071
4072         /* If here, we need to do a join to keep the tree balanced. */
4073         ASSERT(!xfs_btree_ptr_is_null(cur, &cptr));
4074
4075         if (!xfs_btree_ptr_is_null(cur, &lptr) &&
4076             lrecs + xfs_btree_get_numrecs(block) <=
4077                         cur->bc_ops->get_maxrecs(cur, level)) {
4078                 /*
4079                  * Set "right" to be the starting block,
4080                  * "left" to be the left neighbor.
4081                  */
4082                 rptr = cptr;
4083                 right = block;
4084                 rbp = bp;
4085                 error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
4086                 if (error)
4087                         goto error0;
4088
4089         /*
4090          * If that won't work, see if we can join with the right neighbor block.
4091          */
4092         } else if (!xfs_btree_ptr_is_null(cur, &rptr) &&
4093                    rrecs + xfs_btree_get_numrecs(block) <=
4094                         cur->bc_ops->get_maxrecs(cur, level)) {
4095                 /*
4096                  * Set "left" to be the starting block,
4097                  * "right" to be the right neighbor.
4098                  */
4099                 lptr = cptr;
4100                 left = block;
4101                 lbp = bp;
4102                 error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
4103                 if (error)
4104                         goto error0;
4105
4106         /*
4107          * Otherwise, we can't fix the imbalance.
4108          * Just return.  This is probably a logic error, but it's not fatal.
4109          */
4110         } else {
4111                 error = xfs_btree_dec_cursor(cur, level, stat);
4112                 if (error)
4113                         goto error0;
4114                 return 0;
4115         }
4116
4117         rrecs = xfs_btree_get_numrecs(right);
4118         lrecs = xfs_btree_get_numrecs(left);
4119
4120         /*
4121          * We're now going to join "left" and "right" by moving all the stuff
4122          * in "right" to "left" and deleting "right".
4123          */
4124         XFS_BTREE_STATS_ADD(cur, moves, rrecs);
4125         if (level > 0) {
4126                 /* It's a non-leaf.  Move keys and pointers. */
4127                 union xfs_btree_key     *lkp;   /* left btree key */
4128                 union xfs_btree_ptr     *lpp;   /* left address pointer */
4129                 union xfs_btree_key     *rkp;   /* right btree key */
4130                 union xfs_btree_ptr     *rpp;   /* right address pointer */
4131
4132                 lkp = xfs_btree_key_addr(cur, lrecs + 1, left);
4133                 lpp = xfs_btree_ptr_addr(cur, lrecs + 1, left);
4134                 rkp = xfs_btree_key_addr(cur, 1, right);
4135                 rpp = xfs_btree_ptr_addr(cur, 1, right);
4136 #ifdef DEBUG
4137                 for (i = 1; i < rrecs; i++) {
4138                         error = xfs_btree_check_ptr(cur, rpp, i, level);
4139                         if (error)
4140                                 goto error0;
4141                 }
4142 #endif
4143                 xfs_btree_copy_keys(cur, lkp, rkp, rrecs);
4144                 xfs_btree_copy_ptrs(cur, lpp, rpp, rrecs);
4145
4146                 xfs_btree_log_keys(cur, lbp, lrecs + 1, lrecs + rrecs);
4147                 xfs_btree_log_ptrs(cur, lbp, lrecs + 1, lrecs + rrecs);
4148         } else {
4149                 /* It's a leaf.  Move records.  */
4150                 union xfs_btree_rec     *lrp;   /* left record pointer */
4151                 union xfs_btree_rec     *rrp;   /* right record pointer */
4152
4153                 lrp = xfs_btree_rec_addr(cur, lrecs + 1, left);
4154                 rrp = xfs_btree_rec_addr(cur, 1, right);
4155
4156                 xfs_btree_copy_recs(cur, lrp, rrp, rrecs);
4157                 xfs_btree_log_recs(cur, lbp, lrecs + 1, lrecs + rrecs);
4158         }
4159
4160         XFS_BTREE_STATS_INC(cur, join);
4161
4162         /*
4163          * Fix up the number of records and right block pointer in the
4164          * surviving block, and log it.
4165          */
4166         xfs_btree_set_numrecs(left, lrecs + rrecs);
4167         xfs_btree_get_sibling(cur, right, &cptr, XFS_BB_RIGHTSIB),
4168         xfs_btree_set_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
4169         xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
4170
4171         /* If there is a right sibling, point it to the remaining block. */
4172         xfs_btree_get_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
4173         if (!xfs_btree_ptr_is_null(cur, &cptr)) {
4174                 error = xfs_btree_read_buf_block(cur, &cptr, 0, &rrblock, &rrbp);
4175                 if (error)
4176                         goto error0;
4177                 xfs_btree_set_sibling(cur, rrblock, &lptr, XFS_BB_LEFTSIB);
4178                 xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
4179         }
4180
4181         /* Free the deleted block. */
4182         error = xfs_btree_free_block(cur, rbp);
4183         if (error)
4184                 goto error0;
4185
4186         /*
4187          * If we joined with the left neighbor, set the buffer in the
4188          * cursor to the left block, and fix up the index.
4189          */
4190         if (bp != lbp) {
4191                 cur->bc_bufs[level] = lbp;
4192                 cur->bc_ptrs[level] += lrecs;
4193                 cur->bc_ra[level] = 0;
4194         }
4195         /*
4196          * If we joined with the right neighbor and there's a level above
4197          * us, increment the cursor at that level.
4198          */
4199         else if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) ||
4200                    (level + 1 < cur->bc_nlevels)) {
4201                 error = xfs_btree_increment(cur, level + 1, &i);
4202                 if (error)
4203                         goto error0;
4204         }
4205
4206         /*
4207          * Readjust the ptr at this level if it's not a leaf, since it's
4208          * still pointing at the deletion point, which makes the cursor
4209          * inconsistent.  If this makes the ptr 0, the caller fixes it up.
4210          * We can't use decrement because it would change the next level up.
4211          */
4212         if (level > 0)
4213                 cur->bc_ptrs[level]--;
4214
4215         /*
4216          * We combined blocks, so we have to update the parent keys if the
4217          * btree supports overlapped intervals.  However, bc_ptrs[level + 1]
4218          * points to the old block so that the caller knows which record to
4219          * delete.  Therefore, the caller must be savvy enough to call updkeys
4220          * for us if we return stat == 2.  The other exit points from this
4221          * function don't require deletions further up the tree, so they can
4222          * call updkeys directly.
4223          */
4224
4225         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
4226         /* Return value means the next level up has something to do. */
4227         *stat = 2;
4228         return 0;
4229
4230 error0:
4231         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
4232         if (tcur)
4233                 xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
4234         return error;
4235 }
4236
4237 /*
4238  * Delete the record pointed to by cur.
4239  * The cursor refers to the place where the record was (could be inserted)
4240  * when the operation returns.
4241  */
4242 int                                     /* error */
4243 xfs_btree_delete(
4244         struct xfs_btree_cur    *cur,
4245         int                     *stat)  /* success/failure */
4246 {
4247         int                     error;  /* error return value */
4248         int                     level;
4249         int                     i;
4250         bool                    joined = false;
4251
4252         XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
4253
4254         /*
4255          * Go up the tree, starting at leaf level.
4256          *
4257          * If 2 is returned then a join was done; go to the next level.
4258          * Otherwise we are done.
4259          */
4260         for (level = 0, i = 2; i == 2; level++) {
4261                 error = xfs_btree_delrec(cur, level, &i);
4262                 if (error)
4263                         goto error0;
4264                 if (i == 2)
4265                         joined = true;
4266         }
4267
4268         /*
4269          * If we combined blocks as part of deleting the record, delrec won't
4270          * have updated the parent high keys so we have to do that here.
4271          */
4272         if (joined && (cur->bc_flags & XFS_BTREE_OVERLAPPING)) {
4273                 error = xfs_btree_updkeys_force(cur, 0);
4274                 if (error)
4275                         goto error0;
4276         }
4277
4278         if (i == 0) {
4279                 for (level = 1; level < cur->bc_nlevels; level++) {
4280                         if (cur->bc_ptrs[level] == 0) {
4281                                 error = xfs_btree_decrement(cur, level, &i);
4282                                 if (error)
4283                                         goto error0;
4284                                 break;
4285                         }
4286                 }
4287         }
4288
4289         XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
4290         *stat = i;
4291         return 0;
4292 error0:
4293         XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
4294         return error;
4295 }
4296
4297 /*
4298  * Get the data from the pointed-to record.
4299  */
4300 int                                     /* error */
4301 xfs_btree_get_rec(
4302         struct xfs_btree_cur    *cur,   /* btree cursor */
4303         union xfs_btree_rec     **recp, /* output: btree record */
4304         int                     *stat)  /* output: success/failure */
4305 {
4306         struct xfs_btree_block  *block; /* btree block */
4307         struct xfs_buf          *bp;    /* buffer pointer */
4308         int                     ptr;    /* record number */
4309 #ifdef DEBUG
4310         int                     error;  /* error return value */
4311 #endif
4312
4313         ptr = cur->bc_ptrs[0];
4314         block = xfs_btree_get_block(cur, 0, &bp);
4315
4316 #ifdef DEBUG
4317         error = xfs_btree_check_block(cur, block, 0, bp);
4318         if (error)
4319                 return error;
4320 #endif
4321
4322         /*
4323          * Off the right end or left end, return failure.
4324          */
4325         if (ptr > xfs_btree_get_numrecs(block) || ptr <= 0) {
4326                 *stat = 0;
4327                 return 0;
4328         }
4329
4330         /*
4331          * Point to the record and extract its data.
4332          */
4333         *recp = xfs_btree_rec_addr(cur, ptr, block);
4334         *stat = 1;
4335         return 0;
4336 }
4337
4338 /* Visit a block in a btree. */
4339 STATIC int
4340 xfs_btree_visit_block(
4341         struct xfs_btree_cur            *cur,
4342         int                             level,
4343         xfs_btree_visit_blocks_fn       fn,
4344         void                            *data)
4345 {
4346         struct xfs_btree_block          *block;
4347         struct xfs_buf                  *bp;
4348         union xfs_btree_ptr             rptr;
4349         int                             error;
4350
4351         /* do right sibling readahead */
4352         xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
4353         block = xfs_btree_get_block(cur, level, &bp);
4354
4355         /* process the block */
4356         error = fn(cur, level, data);
4357         if (error)
4358                 return error;
4359
4360         /* now read rh sibling block for next iteration */
4361         xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
4362         if (xfs_btree_ptr_is_null(cur, &rptr))
4363                 return -ENOENT;
4364
4365         return xfs_btree_lookup_get_block(cur, level, &rptr, &block);
4366 }
4367
4368
4369 /* Visit every block in a btree. */
4370 int
4371 xfs_btree_visit_blocks(
4372         struct xfs_btree_cur            *cur,
4373         xfs_btree_visit_blocks_fn       fn,
4374         void                            *data)
4375 {
4376         union xfs_btree_ptr             lptr;
4377         int                             level;
4378         struct xfs_btree_block          *block = NULL;
4379         int                             error = 0;
4380
4381         cur->bc_ops->init_ptr_from_cur(cur, &lptr);
4382
4383         /* for each level */
4384         for (level = cur->bc_nlevels - 1; level >= 0; level--) {
4385                 /* grab the left hand block */
4386                 error = xfs_btree_lookup_get_block(cur, level, &lptr, &block);
4387                 if (error)
4388                         return error;
4389
4390                 /* readahead the left most block for the next level down */
4391                 if (level > 0) {
4392                         union xfs_btree_ptr     *ptr;
4393
4394                         ptr = xfs_btree_ptr_addr(cur, 1, block);
4395                         xfs_btree_readahead_ptr(cur, ptr, 1);
4396
4397                         /* save for the next iteration of the loop */
4398                         xfs_btree_copy_ptrs(cur, &lptr, ptr, 1);
4399                 }
4400
4401                 /* for each buffer in the level */
4402                 do {
4403                         error = xfs_btree_visit_block(cur, level, fn, data);
4404                 } while (!error);
4405
4406                 if (error != -ENOENT)
4407                         return error;
4408         }
4409
4410         return 0;
4411 }
4412
4413 /*
4414  * Change the owner of a btree.
4415  *
4416  * The mechanism we use here is ordered buffer logging. Because we don't know
4417  * how many buffers were are going to need to modify, we don't really want to
4418  * have to make transaction reservations for the worst case of every buffer in a
4419  * full size btree as that may be more space that we can fit in the log....
4420  *
4421  * We do the btree walk in the most optimal manner possible - we have sibling
4422  * pointers so we can just walk all the blocks on each level from left to right
4423  * in a single pass, and then move to the next level and do the same. We can
4424  * also do readahead on the sibling pointers to get IO moving more quickly,
4425  * though for slow disks this is unlikely to make much difference to performance
4426  * as the amount of CPU work we have to do before moving to the next block is
4427  * relatively small.
4428  *
4429  * For each btree block that we load, modify the owner appropriately, set the
4430  * buffer as an ordered buffer and log it appropriately. We need to ensure that
4431  * we mark the region we change dirty so that if the buffer is relogged in
4432  * a subsequent transaction the changes we make here as an ordered buffer are
4433  * correctly relogged in that transaction.  If we are in recovery context, then
4434  * just queue the modified buffer as delayed write buffer so the transaction
4435  * recovery completion writes the changes to disk.
4436  */
4437 struct xfs_btree_block_change_owner_info {
4438         uint64_t                new_owner;
4439         struct list_head        *buffer_list;
4440 };
4441
4442 static int
4443 xfs_btree_block_change_owner(
4444         struct xfs_btree_cur    *cur,
4445         int                     level,
4446         void                    *data)
4447 {
4448         struct xfs_btree_block_change_owner_info        *bbcoi = data;
4449         struct xfs_btree_block  *block;
4450         struct xfs_buf          *bp;
4451
4452         /* modify the owner */
4453         block = xfs_btree_get_block(cur, level, &bp);
4454         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
4455                 block->bb_u.l.bb_owner = cpu_to_be64(bbcoi->new_owner);
4456         else
4457                 block->bb_u.s.bb_owner = cpu_to_be32(bbcoi->new_owner);
4458
4459         /*
4460          * If the block is a root block hosted in an inode, we might not have a
4461          * buffer pointer here and we shouldn't attempt to log the change as the
4462          * information is already held in the inode and discarded when the root
4463          * block is formatted into the on-disk inode fork. We still change it,
4464          * though, so everything is consistent in memory.
4465          */
4466         if (bp) {
4467                 if (cur->bc_tp) {
4468                         xfs_trans_ordered_buf(cur->bc_tp, bp);
4469                         xfs_btree_log_block(cur, bp, XFS_BB_OWNER);
4470                 } else {
4471                         xfs_buf_delwri_queue(bp, bbcoi->buffer_list);
4472                 }
4473         } else {
4474                 ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
4475                 ASSERT(level == cur->bc_nlevels - 1);
4476         }
4477
4478         return 0;
4479 }
4480
4481 int
4482 xfs_btree_change_owner(
4483         struct xfs_btree_cur    *cur,
4484         uint64_t                new_owner,
4485         struct list_head        *buffer_list)
4486 {
4487         struct xfs_btree_block_change_owner_info        bbcoi;
4488
4489         bbcoi.new_owner = new_owner;
4490         bbcoi.buffer_list = buffer_list;
4491
4492         return xfs_btree_visit_blocks(cur, xfs_btree_block_change_owner,
4493                         &bbcoi);
4494 }
4495
4496 /**
4497  * xfs_btree_sblock_v5hdr_verify() -- verify the v5 fields of a short-format
4498  *                                    btree block
4499  *
4500  * @bp: buffer containing the btree block
4501  * @max_recs: pointer to the m_*_mxr max records field in the xfs mount
4502  * @pag_max_level: pointer to the per-ag max level field
4503  */
4504 bool
4505 xfs_btree_sblock_v5hdr_verify(
4506         struct xfs_buf          *bp)
4507 {
4508         struct xfs_mount        *mp = bp->b_target->bt_mount;
4509         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
4510         struct xfs_perag        *pag = bp->b_pag;
4511
4512         if (!xfs_sb_version_hascrc(&mp->m_sb))
4513                 return false;
4514         if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid))
4515                 return false;
4516         if (block->bb_u.s.bb_blkno != cpu_to_be64(bp->b_bn))
4517                 return false;
4518         if (pag && be32_to_cpu(block->bb_u.s.bb_owner) != pag->pag_agno)
4519                 return false;
4520         return true;
4521 }
4522
4523 /**
4524  * xfs_btree_sblock_verify() -- verify a short-format btree block
4525  *
4526  * @bp: buffer containing the btree block
4527  * @max_recs: maximum records allowed in this btree node
4528  */
4529 bool
4530 xfs_btree_sblock_verify(
4531         struct xfs_buf          *bp,
4532         unsigned int            max_recs)
4533 {
4534         struct xfs_mount        *mp = bp->b_target->bt_mount;
4535         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
4536
4537         /* numrecs verification */
4538         if (be16_to_cpu(block->bb_numrecs) > max_recs)
4539                 return false;
4540
4541         /* sibling pointer verification */
4542         if (!block->bb_u.s.bb_leftsib ||
4543             (be32_to_cpu(block->bb_u.s.bb_leftsib) >= mp->m_sb.sb_agblocks &&
4544              block->bb_u.s.bb_leftsib != cpu_to_be32(NULLAGBLOCK)))
4545                 return false;
4546         if (!block->bb_u.s.bb_rightsib ||
4547             (be32_to_cpu(block->bb_u.s.bb_rightsib) >= mp->m_sb.sb_agblocks &&
4548              block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK)))
4549                 return false;
4550
4551         return true;
4552 }
4553
4554 /*
4555  * Calculate the number of btree levels needed to store a given number of
4556  * records in a short-format btree.
4557  */
4558 uint
4559 xfs_btree_compute_maxlevels(
4560         struct xfs_mount        *mp,
4561         uint                    *limits,
4562         unsigned long           len)
4563 {
4564         uint                    level;
4565         unsigned long           maxblocks;
4566
4567         maxblocks = (len + limits[0] - 1) / limits[0];
4568         for (level = 1; maxblocks > 1; level++)
4569                 maxblocks = (maxblocks + limits[1] - 1) / limits[1];
4570         return level;
4571 }
4572
4573 /*
4574  * Query a regular btree for all records overlapping a given interval.
4575  * Start with a LE lookup of the key of low_rec and return all records
4576  * until we find a record with a key greater than the key of high_rec.
4577  */
4578 STATIC int
4579 xfs_btree_simple_query_range(
4580         struct xfs_btree_cur            *cur,
4581         union xfs_btree_key             *low_key,
4582         union xfs_btree_key             *high_key,
4583         xfs_btree_query_range_fn        fn,
4584         void                            *priv)
4585 {
4586         union xfs_btree_rec             *recp;
4587         union xfs_btree_key             rec_key;
4588         int64_t                         diff;
4589         int                             stat;
4590         bool                            firstrec = true;
4591         int                             error;
4592
4593         ASSERT(cur->bc_ops->init_high_key_from_rec);
4594         ASSERT(cur->bc_ops->diff_two_keys);
4595
4596         /*
4597          * Find the leftmost record.  The btree cursor must be set
4598          * to the low record used to generate low_key.
4599          */
4600         stat = 0;
4601         error = xfs_btree_lookup(cur, XFS_LOOKUP_LE, &stat);
4602         if (error)
4603                 goto out;
4604
4605         /* Nothing?  See if there's anything to the right. */
4606         if (!stat) {
4607                 error = xfs_btree_increment(cur, 0, &stat);
4608                 if (error)
4609                         goto out;
4610         }
4611
4612         while (stat) {
4613                 /* Find the record. */
4614                 error = xfs_btree_get_rec(cur, &recp, &stat);
4615                 if (error || !stat)
4616                         break;
4617
4618                 /* Skip if high_key(rec) < low_key. */
4619                 if (firstrec) {
4620                         cur->bc_ops->init_high_key_from_rec(&rec_key, recp);
4621                         firstrec = false;
4622                         diff = cur->bc_ops->diff_two_keys(cur, low_key,
4623                                         &rec_key);
4624                         if (diff > 0)
4625                                 goto advloop;
4626                 }
4627
4628                 /* Stop if high_key < low_key(rec). */
4629                 cur->bc_ops->init_key_from_rec(&rec_key, recp);
4630                 diff = cur->bc_ops->diff_two_keys(cur, &rec_key, high_key);
4631                 if (diff > 0)
4632                         break;
4633
4634                 /* Callback */
4635                 error = fn(cur, recp, priv);
4636                 if (error < 0 || error == XFS_BTREE_QUERY_RANGE_ABORT)
4637                         break;
4638
4639 advloop:
4640                 /* Move on to the next record. */
4641                 error = xfs_btree_increment(cur, 0, &stat);
4642                 if (error)
4643                         break;
4644         }
4645
4646 out:
4647         return error;
4648 }
4649
4650 /*
4651  * Query an overlapped interval btree for all records overlapping a given
4652  * interval.  This function roughly follows the algorithm given in
4653  * "Interval Trees" of _Introduction to Algorithms_, which is section
4654  * 14.3 in the 2nd and 3rd editions.
4655  *
4656  * First, generate keys for the low and high records passed in.
4657  *
4658  * For any leaf node, generate the high and low keys for the record.
4659  * If the record keys overlap with the query low/high keys, pass the
4660  * record to the function iterator.
4661  *
4662  * For any internal node, compare the low and high keys of each
4663  * pointer against the query low/high keys.  If there's an overlap,
4664  * follow the pointer.
4665  *
4666  * As an optimization, we stop scanning a block when we find a low key
4667  * that is greater than the query's high key.
4668  */
4669 STATIC int
4670 xfs_btree_overlapped_query_range(
4671         struct xfs_btree_cur            *cur,
4672         union xfs_btree_key             *low_key,
4673         union xfs_btree_key             *high_key,
4674         xfs_btree_query_range_fn        fn,
4675         void                            *priv)
4676 {
4677         union xfs_btree_ptr             ptr;
4678         union xfs_btree_ptr             *pp;
4679         union xfs_btree_key             rec_key;
4680         union xfs_btree_key             rec_hkey;
4681         union xfs_btree_key             *lkp;
4682         union xfs_btree_key             *hkp;
4683         union xfs_btree_rec             *recp;
4684         struct xfs_btree_block          *block;
4685         int64_t                         ldiff;
4686         int64_t                         hdiff;
4687         int                             level;
4688         struct xfs_buf                  *bp;
4689         int                             i;
4690         int                             error;
4691
4692         /* Load the root of the btree. */
4693         level = cur->bc_nlevels - 1;
4694         cur->bc_ops->init_ptr_from_cur(cur, &ptr);
4695         error = xfs_btree_lookup_get_block(cur, level, &ptr, &block);
4696         if (error)
4697                 return error;
4698         xfs_btree_get_block(cur, level, &bp);
4699         trace_xfs_btree_overlapped_query_range(cur, level, bp);
4700 #ifdef DEBUG
4701         error = xfs_btree_check_block(cur, block, level, bp);
4702         if (error)
4703                 goto out;
4704 #endif
4705         cur->bc_ptrs[level] = 1;
4706
4707         while (level < cur->bc_nlevels) {
4708                 block = xfs_btree_get_block(cur, level, &bp);
4709
4710                 /* End of node, pop back towards the root. */
4711                 if (cur->bc_ptrs[level] > be16_to_cpu(block->bb_numrecs)) {
4712 pop_up:
4713                         if (level < cur->bc_nlevels - 1)
4714                                 cur->bc_ptrs[level + 1]++;
4715                         level++;
4716                         continue;
4717                 }
4718
4719                 if (level == 0) {
4720                         /* Handle a leaf node. */
4721                         recp = xfs_btree_rec_addr(cur, cur->bc_ptrs[0], block);
4722
4723                         cur->bc_ops->init_high_key_from_rec(&rec_hkey, recp);
4724                         ldiff = cur->bc_ops->diff_two_keys(cur, &rec_hkey,
4725                                         low_key);
4726
4727                         cur->bc_ops->init_key_from_rec(&rec_key, recp);
4728                         hdiff = cur->bc_ops->diff_two_keys(cur, high_key,
4729                                         &rec_key);
4730
4731                         /*
4732                          * If (record's high key >= query's low key) and
4733                          *    (query's high key >= record's low key), then
4734                          * this record overlaps the query range; callback.
4735                          */
4736                         if (ldiff >= 0 && hdiff >= 0) {
4737                                 error = fn(cur, recp, priv);
4738                                 if (error < 0 ||
4739                                     error == XFS_BTREE_QUERY_RANGE_ABORT)
4740                                         break;
4741                         } else if (hdiff < 0) {
4742                                 /* Record is larger than high key; pop. */
4743                                 goto pop_up;
4744                         }
4745                         cur->bc_ptrs[level]++;
4746                         continue;
4747                 }
4748
4749                 /* Handle an internal node. */
4750                 lkp = xfs_btree_key_addr(cur, cur->bc_ptrs[level], block);
4751                 hkp = xfs_btree_high_key_addr(cur, cur->bc_ptrs[level], block);
4752                 pp = xfs_btree_ptr_addr(cur, cur->bc_ptrs[level], block);
4753
4754                 ldiff = cur->bc_ops->diff_two_keys(cur, hkp, low_key);
4755                 hdiff = cur->bc_ops->diff_two_keys(cur, high_key, lkp);
4756
4757                 /*
4758                  * If (pointer's high key >= query's low key) and
4759                  *    (query's high key >= pointer's low key), then
4760                  * this record overlaps the query range; follow pointer.
4761                  */
4762                 if (ldiff >= 0 && hdiff >= 0) {
4763                         level--;
4764                         error = xfs_btree_lookup_get_block(cur, level, pp,
4765                                         &block);
4766                         if (error)
4767                                 goto out;
4768                         xfs_btree_get_block(cur, level, &bp);
4769                         trace_xfs_btree_overlapped_query_range(cur, level, bp);
4770 #ifdef DEBUG
4771                         error = xfs_btree_check_block(cur, block, level, bp);
4772                         if (error)
4773                                 goto out;
4774 #endif
4775                         cur->bc_ptrs[level] = 1;
4776                         continue;
4777                 } else if (hdiff < 0) {
4778                         /* The low key is larger than the upper range; pop. */
4779                         goto pop_up;
4780                 }
4781                 cur->bc_ptrs[level]++;
4782         }
4783
4784 out:
4785         /*
4786          * If we don't end this function with the cursor pointing at a record
4787          * block, a subsequent non-error cursor deletion will not release
4788          * node-level buffers, causing a buffer leak.  This is quite possible
4789          * with a zero-results range query, so release the buffers if we
4790          * failed to return any results.
4791          */
4792         if (cur->bc_bufs[0] == NULL) {
4793                 for (i = 0; i < cur->bc_nlevels; i++) {
4794                         if (cur->bc_bufs[i]) {
4795                                 xfs_trans_brelse(cur->bc_tp, cur->bc_bufs[i]);
4796                                 cur->bc_bufs[i] = NULL;
4797                                 cur->bc_ptrs[i] = 0;
4798                                 cur->bc_ra[i] = 0;
4799                         }
4800                 }
4801         }
4802
4803         return error;
4804 }
4805
4806 /*
4807  * Query a btree for all records overlapping a given interval of keys.  The
4808  * supplied function will be called with each record found; return one of the
4809  * XFS_BTREE_QUERY_RANGE_{CONTINUE,ABORT} values or the usual negative error
4810  * code.  This function returns XFS_BTREE_QUERY_RANGE_ABORT, zero, or a
4811  * negative error code.
4812  */
4813 int
4814 xfs_btree_query_range(
4815         struct xfs_btree_cur            *cur,
4816         union xfs_btree_irec            *low_rec,
4817         union xfs_btree_irec            *high_rec,
4818         xfs_btree_query_range_fn        fn,
4819         void                            *priv)
4820 {
4821         union xfs_btree_rec             rec;
4822         union xfs_btree_key             low_key;
4823         union xfs_btree_key             high_key;
4824
4825         /* Find the keys of both ends of the interval. */
4826         cur->bc_rec = *high_rec;
4827         cur->bc_ops->init_rec_from_cur(cur, &rec);
4828         cur->bc_ops->init_key_from_rec(&high_key, &rec);
4829
4830         cur->bc_rec = *low_rec;
4831         cur->bc_ops->init_rec_from_cur(cur, &rec);
4832         cur->bc_ops->init_key_from_rec(&low_key, &rec);
4833
4834         /* Enforce low key < high key. */
4835         if (cur->bc_ops->diff_two_keys(cur, &low_key, &high_key) > 0)
4836                 return -EINVAL;
4837
4838         if (!(cur->bc_flags & XFS_BTREE_OVERLAPPING))
4839                 return xfs_btree_simple_query_range(cur, &low_key,
4840                                 &high_key, fn, priv);
4841         return xfs_btree_overlapped_query_range(cur, &low_key, &high_key,
4842                         fn, priv);
4843 }
4844
4845 /* Query a btree for all records. */
4846 int
4847 xfs_btree_query_all(
4848         struct xfs_btree_cur            *cur,
4849         xfs_btree_query_range_fn        fn,
4850         void                            *priv)
4851 {
4852         union xfs_btree_key             low_key;
4853         union xfs_btree_key             high_key;
4854
4855         memset(&cur->bc_rec, 0, sizeof(cur->bc_rec));
4856         memset(&low_key, 0, sizeof(low_key));
4857         memset(&high_key, 0xFF, sizeof(high_key));
4858
4859         return xfs_btree_simple_query_range(cur, &low_key, &high_key, fn, priv);
4860 }
4861
4862 /*
4863  * Calculate the number of blocks needed to store a given number of records
4864  * in a short-format (per-AG metadata) btree.
4865  */
4866 xfs_extlen_t
4867 xfs_btree_calc_size(
4868         struct xfs_mount        *mp,
4869         uint                    *limits,
4870         unsigned long long      len)
4871 {
4872         int                     level;
4873         int                     maxrecs;
4874         xfs_extlen_t            rval;
4875
4876         maxrecs = limits[0];
4877         for (level = 0, rval = 0; len > 1; level++) {
4878                 len += maxrecs - 1;
4879                 do_div(len, maxrecs);
4880                 maxrecs = limits[1];
4881                 rval += len;
4882         }
4883         return rval;
4884 }
4885
4886 static int
4887 xfs_btree_count_blocks_helper(
4888         struct xfs_btree_cur    *cur,
4889         int                     level,
4890         void                    *data)
4891 {
4892         xfs_extlen_t            *blocks = data;
4893         (*blocks)++;
4894
4895         return 0;
4896 }
4897
4898 /* Count the blocks in a btree and return the result in *blocks. */
4899 int
4900 xfs_btree_count_blocks(
4901         struct xfs_btree_cur    *cur,
4902         xfs_extlen_t            *blocks)
4903 {
4904         *blocks = 0;
4905         return xfs_btree_visit_blocks(cur, xfs_btree_count_blocks_helper,
4906                         blocks);
4907 }