fs/xfs/libxfs/xfs

/* [<][>][^][v][top][bottom][index][help] */
This source file includes following definitions.
xfs_btree_magic
__xfs_btree_check_lblock
xfs_btree_check_lblock
__xfs_btree_check_sblock
xfs_btree_check_sblock
xfs_btree_check_block
xfs_btree_check_lptr
xfs_btree_check_sptr
xfs_btree_check_ptr
xfs_btree_lblock_calc_crc
xfs_btree_lblock_verify_crc
xfs_btree_sblock_calc_crc
xfs_btree_sblock_verify_crc
xfs_btree_free_block
xfs_btree_del_cursor
xfs_btree_dup_cursor
xfs_btree_block_len
xfs_btree_ptr_len
xfs_btree_rec_offset
xfs_btree_key_offset
xfs_btree_high_key_offset
xfs_btree_ptr_offset
xfs_btree_rec_addr
xfs_btree_key_addr
xfs_btree_high_key_addr
xfs_btree_ptr_addr
xfs_btree_get_iroot
xfs_btree_get_block
xfs_btree_get_bufl
xfs_btree_get_bufs
xfs_btree_islastblock
xfs_btree_firstrec
xfs_btree_lastrec
xfs_btree_offsets
xfs_btree_read_bufl
xfs_btree_reada_bufl
xfs_btree_reada_bufs
xfs_btree_readahead_lblock
xfs_btree_readahead_sblock
xfs_btree_readahead
xfs_btree_ptr_to_daddr
xfs_btree_readahead_ptr
xfs_btree_setbuf
xfs_btree_ptr_is_null
xfs_btree_set_ptr_null
xfs_btree_get_sibling
xfs_btree_set_sibling
xfs_btree_init_block_int
xfs_btree_init_block
xfs_btree_init_block_cur
xfs_btree_is_lastrec
xfs_btree_buf_to_ptr
xfs_btree_set_refs
xfs_btree_get_buf_block
xfs_btree_read_buf_block
xfs_btree_copy_keys
xfs_btree_copy_recs
xfs_btree_copy_ptrs
xfs_btree_shift_keys
xfs_btree_shift_recs
xfs_btree_shift_ptrs
xfs_btree_log_keys
xfs_btree_log_recs
xfs_btree_log_ptrs
xfs_btree_log_block
xfs_btree_increment
xfs_btree_decrement
xfs_btree_lookup_get_block
xfs_lookup_get_search_key
xfs_btree_lookup
xfs_btree_high_key_from_key
xfs_btree_get_leaf_keys
xfs_btree_get_node_keys
xfs_btree_get_keys
xfs_btree_needs_key_update
__xfs_btree_updkeys
xfs_btree_updkeys_force
xfs_btree_update_keys
xfs_btree_update
xfs_btree_lshift
xfs_btree_rshift
__xfs_btree_split
xfs_btree_split_worker
xfs_btree_split
xfs_btree_new_iroot
xfs_btree_new_root
xfs_btree_make_block_unfull
xfs_btree_insrec
xfs_btree_insert
xfs_btree_kill_iroot
xfs_btree_kill_root
xfs_btree_dec_cursor
xfs_btree_delrec
xfs_btree_delete
xfs_btree_get_rec
xfs_btree_visit_block
xfs_btree_visit_blocks
xfs_btree_block_change_owner
xfs_btree_change_owner
xfs_btree_lblock_v5hdr_verify
xfs_btree_lblock_verify
xfs_btree_sblock_v5hdr_verify
xfs_btree_sblock_verify
xfs_btree_compute_maxlevels
xfs_btree_simple_query_range
xfs_btree_overlapped_query_range
xfs_btree_query_range
xfs_btree_query_all
xfs_btree_calc_size
xfs_btree_count_blocks_helper
xfs_btree_count_blocks
xfs_btree_diff_two_ptrs
xfs_btree_has_record_helper
xfs_btree_has_record
xfs_btree_has_more_records
   1 // SPDX-License-Identifier: GPL-2.0
   2 /*
   3  * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
   4  * All Rights Reserved.
   5  */
   6 #include "xfs.h"
   7 #include "xfs_fs.h"
   8 #include "xfs_shared.h"
   9 #include "xfs_format.h"
  10 #include "xfs_log_format.h"
  11 #include "xfs_trans_resv.h"
  12 #include "xfs_bit.h"
  13 #include "xfs_mount.h"
  14 #include "xfs_inode.h"
  15 #include "xfs_trans.h"
  16 #include "xfs_buf_item.h"
  17 #include "xfs_btree.h"
  18 #include "xfs_errortag.h"
  19 #include "xfs_error.h"
  20 #include "xfs_trace.h"
  21 #include "xfs_alloc.h"
  22 #include "xfs_log.h"
  23 
  24 /*
  25  * Cursor allocation zone.
  26  */
  27 kmem_zone_t     *xfs_btree_cur_zone;
  28 
  29 /*
  30  * Btree magic numbers.
  31  */
  32 static const uint32_t xfs_magics[2][XFS_BTNUM_MAX] = {
  33         { XFS_ABTB_MAGIC, XFS_ABTC_MAGIC, 0, XFS_BMAP_MAGIC, XFS_IBT_MAGIC,
  34           XFS_FIBT_MAGIC, 0 },
  35         { XFS_ABTB_CRC_MAGIC, XFS_ABTC_CRC_MAGIC, XFS_RMAP_CRC_MAGIC,
  36           XFS_BMAP_CRC_MAGIC, XFS_IBT_CRC_MAGIC, XFS_FIBT_CRC_MAGIC,
  37           XFS_REFC_CRC_MAGIC }
  38 };
  39 
  40 uint32_t
  41 xfs_btree_magic(
  42         int                     crc,
  43         xfs_btnum_t             btnum)
  44 {
  45         uint32_t                magic = xfs_magics[crc][btnum];
  46 
  47         /* Ensure we asked for crc for crc-only magics. */
  48         ASSERT(magic != 0);
  49         return magic;
  50 }
  51 
  52 /*
  53  * Check a long btree block header.  Return the address of the failing check,
  54  * or NULL if everything is ok.
  55  */
  56 xfs_failaddr_t
  57 __xfs_btree_check_lblock(
  58         struct xfs_btree_cur    *cur,
  59         struct xfs_btree_block  *block,
  60         int                     level,
  61         struct xfs_buf          *bp)
  62 {
  63         struct xfs_mount        *mp = cur->bc_mp;
  64         xfs_btnum_t             btnum = cur->bc_btnum;
  65         int                     crc = xfs_sb_version_hascrc(&mp->m_sb);
  66 
  67         if (crc) {
  68                 if (!uuid_equal(&block->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid))
  69                         return __this_address;
  70                 if (block->bb_u.l.bb_blkno !=
  71                     cpu_to_be64(bp ? bp->b_bn : XFS_BUF_DADDR_NULL))
  72                         return __this_address;
  73                 if (block->bb_u.l.bb_pad != cpu_to_be32(0))
  74                         return __this_address;
  75         }
  76 
  77         if (be32_to_cpu(block->bb_magic) != xfs_btree_magic(crc, btnum))
  78                 return __this_address;
  79         if (be16_to_cpu(block->bb_level) != level)
  80                 return __this_address;
  81         if (be16_to_cpu(block->bb_numrecs) >
  82             cur->bc_ops->get_maxrecs(cur, level))
  83                 return __this_address;
  84         if (block->bb_u.l.bb_leftsib != cpu_to_be64(NULLFSBLOCK) &&
  85             !xfs_btree_check_lptr(cur, be64_to_cpu(block->bb_u.l.bb_leftsib),
  86                         level + 1))
  87                 return __this_address;
  88         if (block->bb_u.l.bb_rightsib != cpu_to_be64(NULLFSBLOCK) &&
  89             !xfs_btree_check_lptr(cur, be64_to_cpu(block->bb_u.l.bb_rightsib),
  90                         level + 1))
  91                 return __this_address;
  92 
  93         return NULL;
  94 }
  95 
  96 /* Check a long btree block header. */
  97 static int
  98 xfs_btree_check_lblock(
  99         struct xfs_btree_cur    *cur,
 100         struct xfs_btree_block  *block,
 101         int                     level,
 102         struct xfs_buf          *bp)
 103 {
 104         struct xfs_mount        *mp = cur->bc_mp;
 105         xfs_failaddr_t          fa;
 106 
 107         fa = __xfs_btree_check_lblock(cur, block, level, bp);
 108         if (unlikely(XFS_TEST_ERROR(fa != NULL, mp,
 109                         XFS_ERRTAG_BTREE_CHECK_LBLOCK))) {
 110                 if (bp)
 111                         trace_xfs_btree_corrupt(bp, _RET_IP_);
 112                 XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, mp);
 113                 return -EFSCORRUPTED;
 114         }
 115         return 0;
 116 }
 117 
 118 /*
 119  * Check a short btree block header.  Return the address of the failing check,
 120  * or NULL if everything is ok.
 121  */
 122 xfs_failaddr_t
 123 __xfs_btree_check_sblock(
 124         struct xfs_btree_cur    *cur,
 125         struct xfs_btree_block  *block,
 126         int                     level,
 127         struct xfs_buf          *bp)
 128 {
 129         struct xfs_mount        *mp = cur->bc_mp;
 130         xfs_btnum_t             btnum = cur->bc_btnum;
 131         int                     crc = xfs_sb_version_hascrc(&mp->m_sb);
 132 
 133         if (crc) {
 134                 if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid))
 135                         return __this_address;
 136                 if (block->bb_u.s.bb_blkno !=
 137                     cpu_to_be64(bp ? bp->b_bn : XFS_BUF_DADDR_NULL))
 138                         return __this_address;
 139         }
 140 
 141         if (be32_to_cpu(block->bb_magic) != xfs_btree_magic(crc, btnum))
 142                 return __this_address;
 143         if (be16_to_cpu(block->bb_level) != level)
 144                 return __this_address;
 145         if (be16_to_cpu(block->bb_numrecs) >
 146             cur->bc_ops->get_maxrecs(cur, level))
 147                 return __this_address;
 148         if (block->bb_u.s.bb_leftsib != cpu_to_be32(NULLAGBLOCK) &&
 149             !xfs_btree_check_sptr(cur, be32_to_cpu(block->bb_u.s.bb_leftsib),
 150                         level + 1))
 151                 return __this_address;
 152         if (block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK) &&
 153             !xfs_btree_check_sptr(cur, be32_to_cpu(block->bb_u.s.bb_rightsib),
 154                         level + 1))
 155                 return __this_address;
 156 
 157         return NULL;
 158 }
 159 
 160 /* Check a short btree block header. */
 161 STATIC int
 162 xfs_btree_check_sblock(
 163         struct xfs_btree_cur    *cur,
 164         struct xfs_btree_block  *block,
 165         int                     level,
 166         struct xfs_buf          *bp)
 167 {
 168         struct xfs_mount        *mp = cur->bc_mp;
 169         xfs_failaddr_t          fa;
 170 
 171         fa = __xfs_btree_check_sblock(cur, block, level, bp);
 172         if (unlikely(XFS_TEST_ERROR(fa != NULL, mp,
 173                         XFS_ERRTAG_BTREE_CHECK_SBLOCK))) {
 174                 if (bp)
 175                         trace_xfs_btree_corrupt(bp, _RET_IP_);
 176                 XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, mp);
 177                 return -EFSCORRUPTED;
 178         }
 179         return 0;
 180 }
 181 
 182 /*
 183  * Debug routine: check that block header is ok.
 184  */
 185 int
 186 xfs_btree_check_block(
 187         struct xfs_btree_cur    *cur,   /* btree cursor */
 188         struct xfs_btree_block  *block, /* generic btree block pointer */
 189         int                     level,  /* level of the btree block */
 190         struct xfs_buf          *bp)    /* buffer containing block, if any */
 191 {
 192         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
 193                 return xfs_btree_check_lblock(cur, block, level, bp);
 194         else
 195                 return xfs_btree_check_sblock(cur, block, level, bp);
 196 }
 197 
 198 /* Check that this long pointer is valid and points within the fs. */
 199 bool
 200 xfs_btree_check_lptr(
 201         struct xfs_btree_cur    *cur,
 202         xfs_fsblock_t           fsbno,
 203         int                     level)
 204 {
 205         if (level <= 0)
 206                 return false;
 207         return xfs_verify_fsbno(cur->bc_mp, fsbno);
 208 }
 209 
 210 /* Check that this short pointer is valid and points within the AG. */
 211 bool
 212 xfs_btree_check_sptr(
 213         struct xfs_btree_cur    *cur,
 214         xfs_agblock_t           agbno,
 215         int                     level)
 216 {
 217         if (level <= 0)
 218                 return false;
 219         return xfs_verify_agbno(cur->bc_mp, cur->bc_private.a.agno, agbno);
 220 }
 221 
 222 /*
 223  * Check that a given (indexed) btree pointer at a certain level of a
 224  * btree is valid and doesn't point past where it should.
 225  */
 226 static int
 227 xfs_btree_check_ptr(
 228         struct xfs_btree_cur    *cur,
 229         union xfs_btree_ptr     *ptr,
 230         int                     index,
 231         int                     level)
 232 {
 233         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
 234                 if (xfs_btree_check_lptr(cur, be64_to_cpu((&ptr->l)[index]),
 235                                 level))
 236                         return 0;
 237                 xfs_err(cur->bc_mp,
 238 "Inode %llu fork %d: Corrupt btree %d pointer at level %d index %d.",
 239                                 cur->bc_private.b.ip->i_ino,
 240                                 cur->bc_private.b.whichfork, cur->bc_btnum,
 241                                 level, index);
 242         } else {
 243                 if (xfs_btree_check_sptr(cur, be32_to_cpu((&ptr->s)[index]),
 244                                 level))
 245                         return 0;
 246                 xfs_err(cur->bc_mp,
 247 "AG %u: Corrupt btree %d pointer at level %d index %d.",
 248                                 cur->bc_private.a.agno, cur->bc_btnum,
 249                                 level, index);
 250         }
 251 
 252         return -EFSCORRUPTED;
 253 }
 254 
 255 #ifdef DEBUG
 256 # define xfs_btree_debug_check_ptr      xfs_btree_check_ptr
 257 #else
 258 # define xfs_btree_debug_check_ptr(...) (0)
 259 #endif
 260 
 261 /*
 262  * Calculate CRC on the whole btree block and stuff it into the
 263  * long-form btree header.
 264  *
 265  * Prior to calculting the CRC, pull the LSN out of the buffer log item and put
 266  * it into the buffer so recovery knows what the last modification was that made
 267  * it to disk.
 268  */
 269 void
 270 xfs_btree_lblock_calc_crc(
 271         struct xfs_buf          *bp)
 272 {
 273         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
 274         struct xfs_buf_log_item *bip = bp->b_log_item;
 275 
 276         if (!xfs_sb_version_hascrc(&bp->b_mount->m_sb))
 277                 return;
 278         if (bip)
 279                 block->bb_u.l.bb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
 280         xfs_buf_update_cksum(bp, XFS_BTREE_LBLOCK_CRC_OFF);
 281 }
 282 
 283 bool
 284 xfs_btree_lblock_verify_crc(
 285         struct xfs_buf          *bp)
 286 {
 287         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
 288         struct xfs_mount        *mp = bp->b_mount;
 289 
 290         if (xfs_sb_version_hascrc(&mp->m_sb)) {
 291                 if (!xfs_log_check_lsn(mp, be64_to_cpu(block->bb_u.l.bb_lsn)))
 292                         return false;
 293                 return xfs_buf_verify_cksum(bp, XFS_BTREE_LBLOCK_CRC_OFF);
 294         }
 295 
 296         return true;
 297 }
 298 
 299 /*
 300  * Calculate CRC on the whole btree block and stuff it into the
 301  * short-form btree header.
 302  *
 303  * Prior to calculting the CRC, pull the LSN out of the buffer log item and put
 304  * it into the buffer so recovery knows what the last modification was that made
 305  * it to disk.
 306  */
 307 void
 308 xfs_btree_sblock_calc_crc(
 309         struct xfs_buf          *bp)
 310 {
 311         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
 312         struct xfs_buf_log_item *bip = bp->b_log_item;
 313 
 314         if (!xfs_sb_version_hascrc(&bp->b_mount->m_sb))
 315                 return;
 316         if (bip)
 317                 block->bb_u.s.bb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
 318         xfs_buf_update_cksum(bp, XFS_BTREE_SBLOCK_CRC_OFF);
 319 }
 320 
 321 bool
 322 xfs_btree_sblock_verify_crc(
 323         struct xfs_buf          *bp)
 324 {
 325         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
 326         struct xfs_mount        *mp = bp->b_mount;
 327 
 328         if (xfs_sb_version_hascrc(&mp->m_sb)) {
 329                 if (!xfs_log_check_lsn(mp, be64_to_cpu(block->bb_u.s.bb_lsn)))
 330                         return false;
 331                 return xfs_buf_verify_cksum(bp, XFS_BTREE_SBLOCK_CRC_OFF);
 332         }
 333 
 334         return true;
 335 }
 336 
 337 static int
 338 xfs_btree_free_block(
 339         struct xfs_btree_cur    *cur,
 340         struct xfs_buf          *bp)
 341 {
 342         int                     error;
 343 
 344         error = cur->bc_ops->free_block(cur, bp);
 345         if (!error) {
 346                 xfs_trans_binval(cur->bc_tp, bp);
 347                 XFS_BTREE_STATS_INC(cur, free);
 348         }
 349         return error;
 350 }
 351 
 352 /*
 353  * Delete the btree cursor.
 354  */
 355 void
 356 xfs_btree_del_cursor(
 357         xfs_btree_cur_t *cur,           /* btree cursor */
 358         int             error)          /* del because of error */
 359 {
 360         int             i;              /* btree level */
 361 
 362         /*
 363          * Clear the buffer pointers, and release the buffers.
 364          * If we're doing this in the face of an error, we
 365          * need to make sure to inspect all of the entries
 366          * in the bc_bufs array for buffers to be unlocked.
 367          * This is because some of the btree code works from
 368          * level n down to 0, and if we get an error along
 369          * the way we won't have initialized all the entries
 370          * down to 0.
 371          */
 372         for (i = 0; i < cur->bc_nlevels; i++) {
 373                 if (cur->bc_bufs[i])
 374                         xfs_trans_brelse(cur->bc_tp, cur->bc_bufs[i]);
 375                 else if (!error)
 376                         break;
 377         }
 378         /*
 379          * Can't free a bmap cursor without having dealt with the
 380          * allocated indirect blocks' accounting.
 381          */
 382         ASSERT(cur->bc_btnum != XFS_BTNUM_BMAP ||
 383                cur->bc_private.b.allocated == 0);
 384         /*
 385          * Free the cursor.
 386          */
 387         kmem_zone_free(xfs_btree_cur_zone, cur);
 388 }
 389 
 390 /*
 391  * Duplicate the btree cursor.
 392  * Allocate a new one, copy the record, re-get the buffers.
 393  */
 394 int                                     /* error */
 395 xfs_btree_dup_cursor(
 396         xfs_btree_cur_t *cur,           /* input cursor */
 397         xfs_btree_cur_t **ncur)         /* output cursor */
 398 {
 399         xfs_buf_t       *bp;            /* btree block's buffer pointer */
 400         int             error;          /* error return value */
 401         int             i;              /* level number of btree block */
 402         xfs_mount_t     *mp;            /* mount structure for filesystem */
 403         xfs_btree_cur_t *new;           /* new cursor value */
 404         xfs_trans_t     *tp;            /* transaction pointer, can be NULL */
 405 
 406         tp = cur->bc_tp;
 407         mp = cur->bc_mp;
 408 
 409         /*
 410          * Allocate a new cursor like the old one.
 411          */
 412         new = cur->bc_ops->dup_cursor(cur);
 413 
 414         /*
 415          * Copy the record currently in the cursor.
 416          */
 417         new->bc_rec = cur->bc_rec;
 418 
 419         /*
 420          * For each level current, re-get the buffer and copy the ptr value.
 421          */
 422         for (i = 0; i < new->bc_nlevels; i++) {
 423                 new->bc_ptrs[i] = cur->bc_ptrs[i];
 424                 new->bc_ra[i] = cur->bc_ra[i];
 425                 bp = cur->bc_bufs[i];
 426                 if (bp) {
 427                         error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
 428                                                    XFS_BUF_ADDR(bp), mp->m_bsize,
 429                                                    0, &bp,
 430                                                    cur->bc_ops->buf_ops);
 431                         if (error) {
 432                                 xfs_btree_del_cursor(new, error);
 433                                 *ncur = NULL;
 434                                 return error;
 435                         }
 436                 }
 437                 new->bc_bufs[i] = bp;
 438         }
 439         *ncur = new;
 440         return 0;
 441 }
 442 
 443 /*
 444  * XFS btree block layout and addressing:
 445  *
 446  * There are two types of blocks in the btree: leaf and non-leaf blocks.
 447  *
 448  * The leaf record start with a header then followed by records containing
 449  * the values.  A non-leaf block also starts with the same header, and
 450  * then first contains lookup keys followed by an equal number of pointers
 451  * to the btree blocks at the previous level.
 452  *
 453  *              +--------+-------+-------+-------+-------+-------+-------+
 454  * Leaf:        | header | rec 1 | rec 2 | rec 3 | rec 4 | rec 5 | rec N |
 455  *              +--------+-------+-------+-------+-------+-------+-------+
 456  *
 457  *              +--------+-------+-------+-------+-------+-------+-------+
 458  * Non-Leaf:    | header | key 1 | key 2 | key N | ptr 1 | ptr 2 | ptr N |
 459  *              +--------+-------+-------+-------+-------+-------+-------+
 460  *
 461  * The header is called struct xfs_btree_block for reasons better left unknown
 462  * and comes in different versions for short (32bit) and long (64bit) block
 463  * pointers.  The record and key structures are defined by the btree instances
 464  * and opaque to the btree core.  The block pointers are simple disk endian
 465  * integers, available in a short (32bit) and long (64bit) variant.
 466  *
 467  * The helpers below calculate the offset of a given record, key or pointer
 468  * into a btree block (xfs_btree_*_offset) or return a pointer to the given
 469  * record, key or pointer (xfs_btree_*_addr).  Note that all addressing
 470  * inside the btree block is done using indices starting at one, not zero!
 471  *
 472  * If XFS_BTREE_OVERLAPPING is set, then this btree supports keys containing
 473  * overlapping intervals.  In such a tree, records are still sorted lowest to
 474  * highest and indexed by the smallest key value that refers to the record.
 475  * However, nodes are different: each pointer has two associated keys -- one
 476  * indexing the lowest key available in the block(s) below (the same behavior
 477  * as the key in a regular btree) and another indexing the highest key
 478  * available in the block(s) below.  Because records are /not/ sorted by the
 479  * highest key, all leaf block updates require us to compute the highest key
 480  * that matches any record in the leaf and to recursively update the high keys
 481  * in the nodes going further up in the tree, if necessary.  Nodes look like
 482  * this:
 483  *
 484  *              +--------+-----+-----+-----+-----+-----+-------+-------+-----+
 485  * Non-Leaf:    | header | lo1 | hi1 | lo2 | hi2 | ... | ptr 1 | ptr 2 | ... |
 486  *              +--------+-----+-----+-----+-----+-----+-------+-------+-----+
 487  *
 488  * To perform an interval query on an overlapped tree, perform the usual
 489  * depth-first search and use the low and high keys to decide if we can skip
 490  * that particular node.  If a leaf node is reached, return the records that
 491  * intersect the interval.  Note that an interval query may return numerous
 492  * entries.  For a non-overlapped tree, simply search for the record associated
 493  * with the lowest key and iterate forward until a non-matching record is
 494  * found.  Section 14.3 ("Interval Trees") of _Introduction to Algorithms_ by
 495  * Cormen, Leiserson, Rivest, and Stein (2nd or 3rd ed. only) discuss this in
 496  * more detail.
 497  *
 498  * Why do we care about overlapping intervals?  Let's say you have a bunch of
 499  * reverse mapping records on a reflink filesystem:
 500  *
 501  * 1: +- file A startblock B offset C length D -----------+
 502  * 2:      +- file E startblock F offset G length H --------------+
 503  * 3:      +- file I startblock F offset J length K --+
 504  * 4:                                                        +- file L... --+
 505  *
 506  * Now say we want to map block (B+D) into file A at offset (C+D).  Ideally,
 507  * we'd simply increment the length of record 1.  But how do we find the record
 508  * that ends at (B+D-1) (i.e. record 1)?  A LE lookup of (B+D-1) would return
 509  * record 3 because the keys are ordered first by startblock.  An interval
 510  * query would return records 1 and 2 because they both overlap (B+D-1), and
 511  * from that we can pick out record 1 as the appropriate left neighbor.
 512  *
 513  * In the non-overlapped case you can do a LE lookup and decrement the cursor
 514  * because a record's interval must end before the next record.
 515  */
 516 
 517 /*
 518  * Return size of the btree block header for this btree instance.
 519  */
 520 static inline size_t xfs_btree_block_len(struct xfs_btree_cur *cur)
 521 {
 522         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
 523                 if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS)
 524                         return XFS_BTREE_LBLOCK_CRC_LEN;
 525                 return XFS_BTREE_LBLOCK_LEN;
 526         }
 527         if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS)
 528                 return XFS_BTREE_SBLOCK_CRC_LEN;
 529         return XFS_BTREE_SBLOCK_LEN;
 530 }
 531 
 532 /*
 533  * Return size of btree block pointers for this btree instance.
 534  */
 535 static inline size_t xfs_btree_ptr_len(struct xfs_btree_cur *cur)
 536 {
 537         return (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
 538                 sizeof(__be64) : sizeof(__be32);
 539 }
 540 
 541 /*
 542  * Calculate offset of the n-th record in a btree block.
 543  */
 544 STATIC size_t
 545 xfs_btree_rec_offset(
 546         struct xfs_btree_cur    *cur,
 547         int                     n)
 548 {
 549         return xfs_btree_block_len(cur) +
 550                 (n - 1) * cur->bc_ops->rec_len;
 551 }
 552 
 553 /*
 554  * Calculate offset of the n-th key in a btree block.
 555  */
 556 STATIC size_t
 557 xfs_btree_key_offset(
 558         struct xfs_btree_cur    *cur,
 559         int                     n)
 560 {
 561         return xfs_btree_block_len(cur) +
 562                 (n - 1) * cur->bc_ops->key_len;
 563 }
 564 
 565 /*
 566  * Calculate offset of the n-th high key in a btree block.
 567  */
 568 STATIC size_t
 569 xfs_btree_high_key_offset(
 570         struct xfs_btree_cur    *cur,
 571         int                     n)
 572 {
 573         return xfs_btree_block_len(cur) +
 574                 (n - 1) * cur->bc_ops->key_len + (cur->bc_ops->key_len / 2);
 575 }
 576 
 577 /*
 578  * Calculate offset of the n-th block pointer in a btree block.
 579  */
 580 STATIC size_t
 581 xfs_btree_ptr_offset(
 582         struct xfs_btree_cur    *cur,
 583         int                     n,
 584         int                     level)
 585 {
 586         return xfs_btree_block_len(cur) +
 587                 cur->bc_ops->get_maxrecs(cur, level) * cur->bc_ops->key_len +
 588                 (n - 1) * xfs_btree_ptr_len(cur);
 589 }
 590 
 591 /*
 592  * Return a pointer to the n-th record in the btree block.
 593  */
 594 union xfs_btree_rec *
 595 xfs_btree_rec_addr(
 596         struct xfs_btree_cur    *cur,
 597         int                     n,
 598         struct xfs_btree_block  *block)
 599 {
 600         return (union xfs_btree_rec *)
 601                 ((char *)block + xfs_btree_rec_offset(cur, n));
 602 }
 603 
 604 /*
 605  * Return a pointer to the n-th key in the btree block.
 606  */
 607 union xfs_btree_key *
 608 xfs_btree_key_addr(
 609         struct xfs_btree_cur    *cur,
 610         int                     n,
 611         struct xfs_btree_block  *block)
 612 {
 613         return (union xfs_btree_key *)
 614                 ((char *)block + xfs_btree_key_offset(cur, n));
 615 }
 616 
 617 /*
 618  * Return a pointer to the n-th high key in the btree block.
 619  */
 620 union xfs_btree_key *
 621 xfs_btree_high_key_addr(
 622         struct xfs_btree_cur    *cur,
 623         int                     n,
 624         struct xfs_btree_block  *block)
 625 {
 626         return (union xfs_btree_key *)
 627                 ((char *)block + xfs_btree_high_key_offset(cur, n));
 628 }
 629 
 630 /*
 631  * Return a pointer to the n-th block pointer in the btree block.
 632  */
 633 union xfs_btree_ptr *
 634 xfs_btree_ptr_addr(
 635         struct xfs_btree_cur    *cur,
 636         int                     n,
 637         struct xfs_btree_block  *block)
 638 {
 639         int                     level = xfs_btree_get_level(block);
 640 
 641         ASSERT(block->bb_level != 0);
 642 
 643         return (union xfs_btree_ptr *)
 644                 ((char *)block + xfs_btree_ptr_offset(cur, n, level));
 645 }
 646 
 647 /*
 648  * Get the root block which is stored in the inode.
 649  *
 650  * For now this btree implementation assumes the btree root is always
 651  * stored in the if_broot field of an inode fork.
 652  */
 653 STATIC struct xfs_btree_block *
 654 xfs_btree_get_iroot(
 655         struct xfs_btree_cur    *cur)
 656 {
 657         struct xfs_ifork        *ifp;
 658 
 659         ifp = XFS_IFORK_PTR(cur->bc_private.b.ip, cur->bc_private.b.whichfork);
 660         return (struct xfs_btree_block *)ifp->if_broot;
 661 }
 662 
 663 /*
 664  * Retrieve the block pointer from the cursor at the given level.
 665  * This may be an inode btree root or from a buffer.
 666  */
 667 struct xfs_btree_block *                /* generic btree block pointer */
 668 xfs_btree_get_block(
 669         struct xfs_btree_cur    *cur,   /* btree cursor */
 670         int                     level,  /* level in btree */
 671         struct xfs_buf          **bpp)  /* buffer containing the block */
 672 {
 673         if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
 674             (level == cur->bc_nlevels - 1)) {
 675                 *bpp = NULL;
 676                 return xfs_btree_get_iroot(cur);
 677         }
 678 
 679         *bpp = cur->bc_bufs[level];
 680         return XFS_BUF_TO_BLOCK(*bpp);
 681 }
 682 
 683 /*
 684  * Get a buffer for the block, return it with no data read.
 685  * Long-form addressing.
 686  */
 687 xfs_buf_t *                             /* buffer for fsbno */
 688 xfs_btree_get_bufl(
 689         xfs_mount_t     *mp,            /* file system mount point */
 690         xfs_trans_t     *tp,            /* transaction pointer */
 691         xfs_fsblock_t   fsbno)          /* file system block number */
 692 {
 693         xfs_daddr_t             d;              /* real disk block address */
 694 
 695         ASSERT(fsbno != NULLFSBLOCK);
 696         d = XFS_FSB_TO_DADDR(mp, fsbno);
 697         return xfs_trans_get_buf(tp, mp->m_ddev_targp, d, mp->m_bsize, 0);
 698 }
 699 
 700 /*
 701  * Get a buffer for the block, return it with no data read.
 702  * Short-form addressing.
 703  */
 704 xfs_buf_t *                             /* buffer for agno/agbno */
 705 xfs_btree_get_bufs(
 706         xfs_mount_t     *mp,            /* file system mount point */
 707         xfs_trans_t     *tp,            /* transaction pointer */
 708         xfs_agnumber_t  agno,           /* allocation group number */
 709         xfs_agblock_t   agbno)          /* allocation group block number */
 710 {
 711         xfs_daddr_t             d;              /* real disk block address */
 712 
 713         ASSERT(agno != NULLAGNUMBER);
 714         ASSERT(agbno != NULLAGBLOCK);
 715         d = XFS_AGB_TO_DADDR(mp, agno, agbno);
 716         return xfs_trans_get_buf(tp, mp->m_ddev_targp, d, mp->m_bsize, 0);
 717 }
 718 
 719 /*
 720  * Check for the cursor referring to the last block at the given level.
 721  */
 722 int                                     /* 1=is last block, 0=not last block */
 723 xfs_btree_islastblock(
 724         xfs_btree_cur_t         *cur,   /* btree cursor */
 725         int                     level)  /* level to check */
 726 {
 727         struct xfs_btree_block  *block; /* generic btree block pointer */
 728         xfs_buf_t               *bp;    /* buffer containing block */
 729 
 730         block = xfs_btree_get_block(cur, level, &bp);
 731         xfs_btree_check_block(cur, block, level, bp);
 732         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
 733                 return block->bb_u.l.bb_rightsib == cpu_to_be64(NULLFSBLOCK);
 734         else
 735                 return block->bb_u.s.bb_rightsib == cpu_to_be32(NULLAGBLOCK);
 736 }
 737 
 738 /*
 739  * Change the cursor to point to the first record at the given level.
 740  * Other levels are unaffected.
 741  */
 742 STATIC int                              /* success=1, failure=0 */
 743 xfs_btree_firstrec(
 744         xfs_btree_cur_t         *cur,   /* btree cursor */
 745         int                     level)  /* level to change */
 746 {
 747         struct xfs_btree_block  *block; /* generic btree block pointer */
 748         xfs_buf_t               *bp;    /* buffer containing block */
 749 
 750         /*
 751          * Get the block pointer for this level.
 752          */
 753         block = xfs_btree_get_block(cur, level, &bp);
 754         if (xfs_btree_check_block(cur, block, level, bp))
 755                 return 0;
 756         /*
 757          * It's empty, there is no such record.
 758          */
 759         if (!block->bb_numrecs)
 760                 return 0;
 761         /*
 762          * Set the ptr value to 1, that's the first record/key.
 763          */
 764         cur->bc_ptrs[level] = 1;
 765         return 1;
 766 }
 767 
 768 /*
 769  * Change the cursor to point to the last record in the current block
 770  * at the given level.  Other levels are unaffected.
 771  */
 772 STATIC int                              /* success=1, failure=0 */
 773 xfs_btree_lastrec(
 774         xfs_btree_cur_t         *cur,   /* btree cursor */
 775         int                     level)  /* level to change */
 776 {
 777         struct xfs_btree_block  *block; /* generic btree block pointer */
 778         xfs_buf_t               *bp;    /* buffer containing block */
 779 
 780         /*
 781          * Get the block pointer for this level.
 782          */
 783         block = xfs_btree_get_block(cur, level, &bp);
 784         if (xfs_btree_check_block(cur, block, level, bp))
 785                 return 0;
 786         /*
 787          * It's empty, there is no such record.
 788          */
 789         if (!block->bb_numrecs)
 790                 return 0;
 791         /*
 792          * Set the ptr value to numrecs, that's the last record/key.
 793          */
 794         cur->bc_ptrs[level] = be16_to_cpu(block->bb_numrecs);
 795         return 1;
 796 }
 797 
 798 /*
 799  * Compute first and last byte offsets for the fields given.
 800  * Interprets the offsets table, which contains struct field offsets.
 801  */
 802 void
 803 xfs_btree_offsets(
 804         int64_t         fields,         /* bitmask of fields */
 805         const short     *offsets,       /* table of field offsets */
 806         int             nbits,          /* number of bits to inspect */
 807         int             *first,         /* output: first byte offset */
 808         int             *last)          /* output: last byte offset */
 809 {
 810         int             i;              /* current bit number */
 811         int64_t         imask;          /* mask for current bit number */
 812 
 813         ASSERT(fields != 0);
 814         /*
 815          * Find the lowest bit, so the first byte offset.
 816          */
 817         for (i = 0, imask = 1LL; ; i++, imask <<= 1) {
 818                 if (imask & fields) {
 819                         *first = offsets[i];
 820                         break;
 821                 }
 822         }
 823         /*
 824          * Find the highest bit, so the last byte offset.
 825          */
 826         for (i = nbits - 1, imask = 1LL << i; ; i--, imask >>= 1) {
 827                 if (imask & fields) {
 828                         *last = offsets[i + 1] - 1;
 829                         break;
 830                 }
 831         }
 832 }
 833 
 834 /*
 835  * Get a buffer for the block, return it read in.
 836  * Long-form addressing.
 837  */
 838 int
 839 xfs_btree_read_bufl(
 840         struct xfs_mount        *mp,            /* file system mount point */
 841         struct xfs_trans        *tp,            /* transaction pointer */
 842         xfs_fsblock_t           fsbno,          /* file system block number */
 843         struct xfs_buf          **bpp,          /* buffer for fsbno */
 844         int                     refval,         /* ref count value for buffer */
 845         const struct xfs_buf_ops *ops)
 846 {
 847         struct xfs_buf          *bp;            /* return value */
 848         xfs_daddr_t             d;              /* real disk block address */
 849         int                     error;
 850 
 851         if (!xfs_verify_fsbno(mp, fsbno))
 852                 return -EFSCORRUPTED;
 853         d = XFS_FSB_TO_DADDR(mp, fsbno);
 854         error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, d,
 855                                    mp->m_bsize, 0, &bp, ops);
 856         if (error)
 857                 return error;
 858         if (bp)
 859                 xfs_buf_set_ref(bp, refval);
 860         *bpp = bp;
 861         return 0;
 862 }
 863 
 864 /*
 865  * Read-ahead the block, don't wait for it, don't return a buffer.
 866  * Long-form addressing.
 867  */
 868 /* ARGSUSED */
 869 void
 870 xfs_btree_reada_bufl(
 871         struct xfs_mount        *mp,            /* file system mount point */
 872         xfs_fsblock_t           fsbno,          /* file system block number */
 873         xfs_extlen_t            count,          /* count of filesystem blocks */
 874         const struct xfs_buf_ops *ops)
 875 {
 876         xfs_daddr_t             d;
 877 
 878         ASSERT(fsbno != NULLFSBLOCK);
 879         d = XFS_FSB_TO_DADDR(mp, fsbno);
 880         xfs_buf_readahead(mp->m_ddev_targp, d, mp->m_bsize * count, ops);
 881 }
 882 
 883 /*
 884  * Read-ahead the block, don't wait for it, don't return a buffer.
 885  * Short-form addressing.
 886  */
 887 /* ARGSUSED */
 888 void
 889 xfs_btree_reada_bufs(
 890         struct xfs_mount        *mp,            /* file system mount point */
 891         xfs_agnumber_t          agno,           /* allocation group number */
 892         xfs_agblock_t           agbno,          /* allocation group block number */
 893         xfs_extlen_t            count,          /* count of filesystem blocks */
 894         const struct xfs_buf_ops *ops)
 895 {
 896         xfs_daddr_t             d;
 897 
 898         ASSERT(agno != NULLAGNUMBER);
 899         ASSERT(agbno != NULLAGBLOCK);
 900         d = XFS_AGB_TO_DADDR(mp, agno, agbno);
 901         xfs_buf_readahead(mp->m_ddev_targp, d, mp->m_bsize * count, ops);
 902 }
 903 
 904 STATIC int
 905 xfs_btree_readahead_lblock(
 906         struct xfs_btree_cur    *cur,
 907         int                     lr,
 908         struct xfs_btree_block  *block)
 909 {
 910         int                     rval = 0;
 911         xfs_fsblock_t           left = be64_to_cpu(block->bb_u.l.bb_leftsib);
 912         xfs_fsblock_t           right = be64_to_cpu(block->bb_u.l.bb_rightsib);
 913 
 914         if ((lr & XFS_BTCUR_LEFTRA) && left != NULLFSBLOCK) {
 915                 xfs_btree_reada_bufl(cur->bc_mp, left, 1,
 916                                      cur->bc_ops->buf_ops);
 917                 rval++;
 918         }
 919 
 920         if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLFSBLOCK) {
 921                 xfs_btree_reada_bufl(cur->bc_mp, right, 1,
 922                                      cur->bc_ops->buf_ops);
 923                 rval++;
 924         }
 925 
 926         return rval;
 927 }
 928 
 929 STATIC int
 930 xfs_btree_readahead_sblock(
 931         struct xfs_btree_cur    *cur,
 932         int                     lr,
 933         struct xfs_btree_block *block)
 934 {
 935         int                     rval = 0;
 936         xfs_agblock_t           left = be32_to_cpu(block->bb_u.s.bb_leftsib);
 937         xfs_agblock_t           right = be32_to_cpu(block->bb_u.s.bb_rightsib);
 938 
 939 
 940         if ((lr & XFS_BTCUR_LEFTRA) && left != NULLAGBLOCK) {
 941                 xfs_btree_reada_bufs(cur->bc_mp, cur->bc_private.a.agno,
 942                                      left, 1, cur->bc_ops->buf_ops);
 943                 rval++;
 944         }
 945 
 946         if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLAGBLOCK) {
 947                 xfs_btree_reada_bufs(cur->bc_mp, cur->bc_private.a.agno,
 948                                      right, 1, cur->bc_ops->buf_ops);
 949                 rval++;
 950         }
 951 
 952         return rval;
 953 }
 954 
 955 /*
 956  * Read-ahead btree blocks, at the given level.
 957  * Bits in lr are set from XFS_BTCUR_{LEFT,RIGHT}RA.
 958  */
 959 STATIC int
 960 xfs_btree_readahead(
 961         struct xfs_btree_cur    *cur,           /* btree cursor */
 962         int                     lev,            /* level in btree */
 963         int                     lr)             /* left/right bits */
 964 {
 965         struct xfs_btree_block  *block;
 966 
 967         /*
 968          * No readahead needed if we are at the root level and the
 969          * btree root is stored in the inode.
 970          */
 971         if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
 972             (lev == cur->bc_nlevels - 1))
 973                 return 0;
 974 
 975         if ((cur->bc_ra[lev] | lr) == cur->bc_ra[lev])
 976                 return 0;
 977 
 978         cur->bc_ra[lev] |= lr;
 979         block = XFS_BUF_TO_BLOCK(cur->bc_bufs[lev]);
 980 
 981         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
 982                 return xfs_btree_readahead_lblock(cur, lr, block);
 983         return xfs_btree_readahead_sblock(cur, lr, block);
 984 }
 985 
 986 STATIC int
 987 xfs_btree_ptr_to_daddr(
 988         struct xfs_btree_cur    *cur,
 989         union xfs_btree_ptr     *ptr,
 990         xfs_daddr_t             *daddr)
 991 {
 992         xfs_fsblock_t           fsbno;
 993         xfs_agblock_t           agbno;
 994         int                     error;
 995 
 996         error = xfs_btree_check_ptr(cur, ptr, 0, 1);
 997         if (error)
 998                 return error;
 999 
1000         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
1001                 fsbno = be64_to_cpu(ptr->l);
1002                 *daddr = XFS_FSB_TO_DADDR(cur->bc_mp, fsbno);
1003         } else {
1004                 agbno = be32_to_cpu(ptr->s);
1005                 *daddr = XFS_AGB_TO_DADDR(cur->bc_mp, cur->bc_private.a.agno,
1006                                 agbno);
1007         }
1008 
1009         return 0;
1010 }
1011 
1012 /*
1013  * Readahead @count btree blocks at the given @ptr location.
1014  *
1015  * We don't need to care about long or short form btrees here as we have a
1016  * method of converting the ptr directly to a daddr available to us.
1017  */
1018 STATIC void
1019 xfs_btree_readahead_ptr(
1020         struct xfs_btree_cur    *cur,
1021         union xfs_btree_ptr     *ptr,
1022         xfs_extlen_t            count)
1023 {
1024         xfs_daddr_t             daddr;
1025 
1026         if (xfs_btree_ptr_to_daddr(cur, ptr, &daddr))
1027                 return;
1028         xfs_buf_readahead(cur->bc_mp->m_ddev_targp, daddr,
1029                           cur->bc_mp->m_bsize * count, cur->bc_ops->buf_ops);
1030 }
1031 
1032 /*
1033  * Set the buffer for level "lev" in the cursor to bp, releasing
1034  * any previous buffer.
1035  */
1036 STATIC void
1037 xfs_btree_setbuf(
1038         xfs_btree_cur_t         *cur,   /* btree cursor */
1039         int                     lev,    /* level in btree */
1040         xfs_buf_t               *bp)    /* new buffer to set */
1041 {
1042         struct xfs_btree_block  *b;     /* btree block */
1043 
1044         if (cur->bc_bufs[lev])
1045                 xfs_trans_brelse(cur->bc_tp, cur->bc_bufs[lev]);
1046         cur->bc_bufs[lev] = bp;
1047         cur->bc_ra[lev] = 0;
1048 
1049         b = XFS_BUF_TO_BLOCK(bp);
1050         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
1051                 if (b->bb_u.l.bb_leftsib == cpu_to_be64(NULLFSBLOCK))
1052                         cur->bc_ra[lev] |= XFS_BTCUR_LEFTRA;
1053                 if (b->bb_u.l.bb_rightsib == cpu_to_be64(NULLFSBLOCK))
1054                         cur->bc_ra[lev] |= XFS_BTCUR_RIGHTRA;
1055         } else {
1056                 if (b->bb_u.s.bb_leftsib == cpu_to_be32(NULLAGBLOCK))
1057                         cur->bc_ra[lev] |= XFS_BTCUR_LEFTRA;
1058                 if (b->bb_u.s.bb_rightsib == cpu_to_be32(NULLAGBLOCK))
1059                         cur->bc_ra[lev] |= XFS_BTCUR_RIGHTRA;
1060         }
1061 }
1062 
1063 bool
1064 xfs_btree_ptr_is_null(
1065         struct xfs_btree_cur    *cur,
1066         union xfs_btree_ptr     *ptr)
1067 {
1068         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1069                 return ptr->l == cpu_to_be64(NULLFSBLOCK);
1070         else
1071                 return ptr->s == cpu_to_be32(NULLAGBLOCK);
1072 }
1073 
1074 STATIC void
1075 xfs_btree_set_ptr_null(
1076         struct xfs_btree_cur    *cur,
1077         union xfs_btree_ptr     *ptr)
1078 {
1079         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1080                 ptr->l = cpu_to_be64(NULLFSBLOCK);
1081         else
1082                 ptr->s = cpu_to_be32(NULLAGBLOCK);
1083 }
1084 
1085 /*
1086  * Get/set/init sibling pointers
1087  */
1088 void
1089 xfs_btree_get_sibling(
1090         struct xfs_btree_cur    *cur,
1091         struct xfs_btree_block  *block,
1092         union xfs_btree_ptr     *ptr,
1093         int                     lr)
1094 {
1095         ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
1096 
1097         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
1098                 if (lr == XFS_BB_RIGHTSIB)
1099                         ptr->l = block->bb_u.l.bb_rightsib;
1100                 else
1101                         ptr->l = block->bb_u.l.bb_leftsib;
1102         } else {
1103                 if (lr == XFS_BB_RIGHTSIB)
1104                         ptr->s = block->bb_u.s.bb_rightsib;
1105                 else
1106                         ptr->s = block->bb_u.s.bb_leftsib;
1107         }
1108 }
1109 
1110 STATIC void
1111 xfs_btree_set_sibling(
1112         struct xfs_btree_cur    *cur,
1113         struct xfs_btree_block  *block,
1114         union xfs_btree_ptr     *ptr,
1115         int                     lr)
1116 {
1117         ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
1118 
1119         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
1120                 if (lr == XFS_BB_RIGHTSIB)
1121                         block->bb_u.l.bb_rightsib = ptr->l;
1122                 else
1123                         block->bb_u.l.bb_leftsib = ptr->l;
1124         } else {
1125                 if (lr == XFS_BB_RIGHTSIB)
1126                         block->bb_u.s.bb_rightsib = ptr->s;
1127                 else
1128                         block->bb_u.s.bb_leftsib = ptr->s;
1129         }
1130 }
1131 
1132 void
1133 xfs_btree_init_block_int(
1134         struct xfs_mount        *mp,
1135         struct xfs_btree_block  *buf,
1136         xfs_daddr_t             blkno,
1137         xfs_btnum_t             btnum,
1138         __u16                   level,
1139         __u16                   numrecs,
1140         __u64                   owner,
1141         unsigned int            flags)
1142 {
1143         int                     crc = xfs_sb_version_hascrc(&mp->m_sb);
1144         __u32                   magic = xfs_btree_magic(crc, btnum);
1145 
1146         buf->bb_magic = cpu_to_be32(magic);
1147         buf->bb_level = cpu_to_be16(level);
1148         buf->bb_numrecs = cpu_to_be16(numrecs);
1149 
1150         if (flags & XFS_BTREE_LONG_PTRS) {
1151                 buf->bb_u.l.bb_leftsib = cpu_to_be64(NULLFSBLOCK);
1152                 buf->bb_u.l.bb_rightsib = cpu_to_be64(NULLFSBLOCK);
1153                 if (crc) {
1154                         buf->bb_u.l.bb_blkno = cpu_to_be64(blkno);
1155                         buf->bb_u.l.bb_owner = cpu_to_be64(owner);
1156                         uuid_copy(&buf->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid);
1157                         buf->bb_u.l.bb_pad = 0;
1158                         buf->bb_u.l.bb_lsn = 0;
1159                 }
1160         } else {
1161                 /* owner is a 32 bit value on short blocks */
1162                 __u32 __owner = (__u32)owner;
1163 
1164                 buf->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
1165                 buf->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
1166                 if (crc) {
1167                         buf->bb_u.s.bb_blkno = cpu_to_be64(blkno);
1168                         buf->bb_u.s.bb_owner = cpu_to_be32(__owner);
1169                         uuid_copy(&buf->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid);
1170                         buf->bb_u.s.bb_lsn = 0;
1171                 }
1172         }
1173 }
1174 
1175 void
1176 xfs_btree_init_block(
1177         struct xfs_mount *mp,
1178         struct xfs_buf  *bp,
1179         xfs_btnum_t     btnum,
1180         __u16           level,
1181         __u16           numrecs,
1182         __u64           owner)
1183 {
1184         xfs_btree_init_block_int(mp, XFS_BUF_TO_BLOCK(bp), bp->b_bn,
1185                                  btnum, level, numrecs, owner, 0);
1186 }
1187 
1188 STATIC void
1189 xfs_btree_init_block_cur(
1190         struct xfs_btree_cur    *cur,
1191         struct xfs_buf          *bp,
1192         int                     level,
1193         int                     numrecs)
1194 {
1195         __u64                   owner;
1196 
1197         /*
1198          * we can pull the owner from the cursor right now as the different
1199          * owners align directly with the pointer size of the btree. This may
1200          * change in future, but is safe for current users of the generic btree
1201          * code.
1202          */
1203         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1204                 owner = cur->bc_private.b.ip->i_ino;
1205         else
1206                 owner = cur->bc_private.a.agno;
1207 
1208         xfs_btree_init_block_int(cur->bc_mp, XFS_BUF_TO_BLOCK(bp), bp->b_bn,
1209                                  cur->bc_btnum, level, numrecs,
1210                                  owner, cur->bc_flags);
1211 }
1212 
1213 /*
1214  * Return true if ptr is the last record in the btree and
1215  * we need to track updates to this record.  The decision
1216  * will be further refined in the update_lastrec method.
1217  */
1218 STATIC int
1219 xfs_btree_is_lastrec(
1220         struct xfs_btree_cur    *cur,
1221         struct xfs_btree_block  *block,
1222         int                     level)
1223 {
1224         union xfs_btree_ptr     ptr;
1225 
1226         if (level > 0)
1227                 return 0;
1228         if (!(cur->bc_flags & XFS_BTREE_LASTREC_UPDATE))
1229                 return 0;
1230 
1231         xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
1232         if (!xfs_btree_ptr_is_null(cur, &ptr))
1233                 return 0;
1234         return 1;
1235 }
1236 
1237 STATIC void
1238 xfs_btree_buf_to_ptr(
1239         struct xfs_btree_cur    *cur,
1240         struct xfs_buf          *bp,
1241         union xfs_btree_ptr     *ptr)
1242 {
1243         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1244                 ptr->l = cpu_to_be64(XFS_DADDR_TO_FSB(cur->bc_mp,
1245                                         XFS_BUF_ADDR(bp)));
1246         else {
1247                 ptr->s = cpu_to_be32(xfs_daddr_to_agbno(cur->bc_mp,
1248                                         XFS_BUF_ADDR(bp)));
1249         }
1250 }
1251 
1252 STATIC void
1253 xfs_btree_set_refs(
1254         struct xfs_btree_cur    *cur,
1255         struct xfs_buf          *bp)
1256 {
1257         switch (cur->bc_btnum) {
1258         case XFS_BTNUM_BNO:
1259         case XFS_BTNUM_CNT:
1260                 xfs_buf_set_ref(bp, XFS_ALLOC_BTREE_REF);
1261                 break;
1262         case XFS_BTNUM_INO:
1263         case XFS_BTNUM_FINO:
1264                 xfs_buf_set_ref(bp, XFS_INO_BTREE_REF);
1265                 break;
1266         case XFS_BTNUM_BMAP:
1267                 xfs_buf_set_ref(bp, XFS_BMAP_BTREE_REF);
1268                 break;
1269         case XFS_BTNUM_RMAP:
1270                 xfs_buf_set_ref(bp, XFS_RMAP_BTREE_REF);
1271                 break;
1272         case XFS_BTNUM_REFC:
1273                 xfs_buf_set_ref(bp, XFS_REFC_BTREE_REF);
1274                 break;
1275         default:
1276                 ASSERT(0);
1277         }
1278 }
1279 
1280 STATIC int
1281 xfs_btree_get_buf_block(
1282         struct xfs_btree_cur    *cur,
1283         union xfs_btree_ptr     *ptr,
1284         struct xfs_btree_block  **block,
1285         struct xfs_buf          **bpp)
1286 {
1287         struct xfs_mount        *mp = cur->bc_mp;
1288         xfs_daddr_t             d;
1289         int                     error;
1290 
1291         error = xfs_btree_ptr_to_daddr(cur, ptr, &d);
1292         if (error)
1293                 return error;
1294         *bpp = xfs_trans_get_buf(cur->bc_tp, mp->m_ddev_targp, d,
1295                                  mp->m_bsize, 0);
1296 
1297         if (!*bpp)
1298                 return -ENOMEM;
1299 
1300         (*bpp)->b_ops = cur->bc_ops->buf_ops;
1301         *block = XFS_BUF_TO_BLOCK(*bpp);
1302         return 0;
1303 }
1304 
1305 /*
1306  * Read in the buffer at the given ptr and return the buffer and
1307  * the block pointer within the buffer.
1308  */
1309 STATIC int
1310 xfs_btree_read_buf_block(
1311         struct xfs_btree_cur    *cur,
1312         union xfs_btree_ptr     *ptr,
1313         int                     flags,
1314         struct xfs_btree_block  **block,
1315         struct xfs_buf          **bpp)
1316 {
1317         struct xfs_mount        *mp = cur->bc_mp;
1318         xfs_daddr_t             d;
1319         int                     error;
1320 
1321         /* need to sort out how callers deal with failures first */
1322         ASSERT(!(flags & XBF_TRYLOCK));
1323 
1324         error = xfs_btree_ptr_to_daddr(cur, ptr, &d);
1325         if (error)
1326                 return error;
1327         error = xfs_trans_read_buf(mp, cur->bc_tp, mp->m_ddev_targp, d,
1328                                    mp->m_bsize, flags, bpp,
1329                                    cur->bc_ops->buf_ops);
1330         if (error)
1331                 return error;
1332 
1333         xfs_btree_set_refs(cur, *bpp);
1334         *block = XFS_BUF_TO_BLOCK(*bpp);
1335         return 0;
1336 }
1337 
1338 /*
1339  * Copy keys from one btree block to another.
1340  */
1341 STATIC void
1342 xfs_btree_copy_keys(
1343         struct xfs_btree_cur    *cur,
1344         union xfs_btree_key     *dst_key,
1345         union xfs_btree_key     *src_key,
1346         int                     numkeys)
1347 {
1348         ASSERT(numkeys >= 0);
1349         memcpy(dst_key, src_key, numkeys * cur->bc_ops->key_len);
1350 }
1351 
1352 /*
1353  * Copy records from one btree block to another.
1354  */
1355 STATIC void
1356 xfs_btree_copy_recs(
1357         struct xfs_btree_cur    *cur,
1358         union xfs_btree_rec     *dst_rec,
1359         union xfs_btree_rec     *src_rec,
1360         int                     numrecs)
1361 {
1362         ASSERT(numrecs >= 0);
1363         memcpy(dst_rec, src_rec, numrecs * cur->bc_ops->rec_len);
1364 }
1365 
1366 /*
1367  * Copy block pointers from one btree block to another.
1368  */
1369 STATIC void
1370 xfs_btree_copy_ptrs(
1371         struct xfs_btree_cur    *cur,
1372         union xfs_btree_ptr     *dst_ptr,
1373         union xfs_btree_ptr     *src_ptr,
1374         int                     numptrs)
1375 {
1376         ASSERT(numptrs >= 0);
1377         memcpy(dst_ptr, src_ptr, numptrs * xfs_btree_ptr_len(cur));
1378 }
1379 
1380 /*
1381  * Shift keys one index left/right inside a single btree block.
1382  */
1383 STATIC void
1384 xfs_btree_shift_keys(
1385         struct xfs_btree_cur    *cur,
1386         union xfs_btree_key     *key,
1387         int                     dir,
1388         int                     numkeys)
1389 {
1390         char                    *dst_key;
1391 
1392         ASSERT(numkeys >= 0);
1393         ASSERT(dir == 1 || dir == -1);
1394 
1395         dst_key = (char *)key + (dir * cur->bc_ops->key_len);
1396         memmove(dst_key, key, numkeys * cur->bc_ops->key_len);
1397 }
1398 
1399 /*
1400  * Shift records one index left/right inside a single btree block.
1401  */
1402 STATIC void
1403 xfs_btree_shift_recs(
1404         struct xfs_btree_cur    *cur,
1405         union xfs_btree_rec     *rec,
1406         int                     dir,
1407         int                     numrecs)
1408 {
1409         char                    *dst_rec;
1410 
1411         ASSERT(numrecs >= 0);
1412         ASSERT(dir == 1 || dir == -1);
1413 
1414         dst_rec = (char *)rec + (dir * cur->bc_ops->rec_len);
1415         memmove(dst_rec, rec, numrecs * cur->bc_ops->rec_len);
1416 }
1417 
1418 /*
1419  * Shift block pointers one index left/right inside a single btree block.
1420  */
1421 STATIC void
1422 xfs_btree_shift_ptrs(
1423         struct xfs_btree_cur    *cur,
1424         union xfs_btree_ptr     *ptr,
1425         int                     dir,
1426         int                     numptrs)
1427 {
1428         char                    *dst_ptr;
1429 
1430         ASSERT(numptrs >= 0);
1431         ASSERT(dir == 1 || dir == -1);
1432 
1433         dst_ptr = (char *)ptr + (dir * xfs_btree_ptr_len(cur));
1434         memmove(dst_ptr, ptr, numptrs * xfs_btree_ptr_len(cur));
1435 }
1436 
1437 /*
1438  * Log key values from the btree block.
1439  */
1440 STATIC void
1441 xfs_btree_log_keys(
1442         struct xfs_btree_cur    *cur,
1443         struct xfs_buf          *bp,
1444         int                     first,
1445         int                     last)
1446 {
1447 
1448         if (bp) {
1449                 xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1450                 xfs_trans_log_buf(cur->bc_tp, bp,
1451                                   xfs_btree_key_offset(cur, first),
1452                                   xfs_btree_key_offset(cur, last + 1) - 1);
1453         } else {
1454                 xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
1455                                 xfs_ilog_fbroot(cur->bc_private.b.whichfork));
1456         }
1457 }
1458 
1459 /*
1460  * Log record values from the btree block.
1461  */
1462 void
1463 xfs_btree_log_recs(
1464         struct xfs_btree_cur    *cur,
1465         struct xfs_buf          *bp,
1466         int                     first,
1467         int                     last)
1468 {
1469 
1470         xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1471         xfs_trans_log_buf(cur->bc_tp, bp,
1472                           xfs_btree_rec_offset(cur, first),
1473                           xfs_btree_rec_offset(cur, last + 1) - 1);
1474 
1475 }
1476 
1477 /*
1478  * Log block pointer fields from a btree block (nonleaf).
1479  */
1480 STATIC void
1481 xfs_btree_log_ptrs(
1482         struct xfs_btree_cur    *cur,   /* btree cursor */
1483         struct xfs_buf          *bp,    /* buffer containing btree block */
1484         int                     first,  /* index of first pointer to log */
1485         int                     last)   /* index of last pointer to log */
1486 {
1487 
1488         if (bp) {
1489                 struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
1490                 int                     level = xfs_btree_get_level(block);
1491 
1492                 xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1493                 xfs_trans_log_buf(cur->bc_tp, bp,
1494                                 xfs_btree_ptr_offset(cur, first, level),
1495                                 xfs_btree_ptr_offset(cur, last + 1, level) - 1);
1496         } else {
1497                 xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
1498                         xfs_ilog_fbroot(cur->bc_private.b.whichfork));
1499         }
1500 
1501 }
1502 
1503 /*
1504  * Log fields from a btree block header.
1505  */
1506 void
1507 xfs_btree_log_block(
1508         struct xfs_btree_cur    *cur,   /* btree cursor */
1509         struct xfs_buf          *bp,    /* buffer containing btree block */
1510         int                     fields) /* mask of fields: XFS_BB_... */
1511 {
1512         int                     first;  /* first byte offset logged */
1513         int                     last;   /* last byte offset logged */
1514         static const short      soffsets[] = {  /* table of offsets (short) */
1515                 offsetof(struct xfs_btree_block, bb_magic),
1516                 offsetof(struct xfs_btree_block, bb_level),
1517                 offsetof(struct xfs_btree_block, bb_numrecs),
1518                 offsetof(struct xfs_btree_block, bb_u.s.bb_leftsib),
1519                 offsetof(struct xfs_btree_block, bb_u.s.bb_rightsib),
1520                 offsetof(struct xfs_btree_block, bb_u.s.bb_blkno),
1521                 offsetof(struct xfs_btree_block, bb_u.s.bb_lsn),
1522                 offsetof(struct xfs_btree_block, bb_u.s.bb_uuid),
1523                 offsetof(struct xfs_btree_block, bb_u.s.bb_owner),
1524                 offsetof(struct xfs_btree_block, bb_u.s.bb_crc),
1525                 XFS_BTREE_SBLOCK_CRC_LEN
1526         };
1527         static const short      loffsets[] = {  /* table of offsets (long) */
1528                 offsetof(struct xfs_btree_block, bb_magic),
1529                 offsetof(struct xfs_btree_block, bb_level),
1530                 offsetof(struct xfs_btree_block, bb_numrecs),
1531                 offsetof(struct xfs_btree_block, bb_u.l.bb_leftsib),
1532                 offsetof(struct xfs_btree_block, bb_u.l.bb_rightsib),
1533                 offsetof(struct xfs_btree_block, bb_u.l.bb_blkno),
1534                 offsetof(struct xfs_btree_block, bb_u.l.bb_lsn),
1535                 offsetof(struct xfs_btree_block, bb_u.l.bb_uuid),
1536                 offsetof(struct xfs_btree_block, bb_u.l.bb_owner),
1537                 offsetof(struct xfs_btree_block, bb_u.l.bb_crc),
1538                 offsetof(struct xfs_btree_block, bb_u.l.bb_pad),
1539                 XFS_BTREE_LBLOCK_CRC_LEN
1540         };
1541 
1542         if (bp) {
1543                 int nbits;
1544 
1545                 if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS) {
1546                         /*
1547                          * We don't log the CRC when updating a btree
1548                          * block but instead recreate it during log
1549                          * recovery.  As the log buffers have checksums
1550                          * of their own this is safe and avoids logging a crc
1551                          * update in a lot of places.
1552                          */
1553                         if (fields == XFS_BB_ALL_BITS)
1554                                 fields = XFS_BB_ALL_BITS_CRC;
1555                         nbits = XFS_BB_NUM_BITS_CRC;
1556                 } else {
1557                         nbits = XFS_BB_NUM_BITS;
1558                 }
1559                 xfs_btree_offsets(fields,
1560                                   (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
1561                                         loffsets : soffsets,
1562                                   nbits, &first, &last);
1563                 xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1564                 xfs_trans_log_buf(cur->bc_tp, bp, first, last);
1565         } else {
1566                 xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
1567                         xfs_ilog_fbroot(cur->bc_private.b.whichfork));
1568         }
1569 }
1570 
1571 /*
1572  * Increment cursor by one record at the level.
1573  * For nonzero levels the leaf-ward information is untouched.
1574  */
1575 int                                             /* error */
1576 xfs_btree_increment(
1577         struct xfs_btree_cur    *cur,
1578         int                     level,
1579         int                     *stat)          /* success/failure */
1580 {
1581         struct xfs_btree_block  *block;
1582         union xfs_btree_ptr     ptr;
1583         struct xfs_buf          *bp;
1584         int                     error;          /* error return value */
1585         int                     lev;
1586 
1587         ASSERT(level < cur->bc_nlevels);
1588 
1589         /* Read-ahead to the right at this level. */
1590         xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
1591 
1592         /* Get a pointer to the btree block. */
1593         block = xfs_btree_get_block(cur, level, &bp);
1594 
1595 #ifdef DEBUG
1596         error = xfs_btree_check_block(cur, block, level, bp);
1597         if (error)
1598                 goto error0;
1599 #endif
1600 
1601         /* We're done if we remain in the block after the increment. */
1602         if (++cur->bc_ptrs[level] <= xfs_btree_get_numrecs(block))
1603                 goto out1;
1604 
1605         /* Fail if we just went off the right edge of the tree. */
1606         xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
1607         if (xfs_btree_ptr_is_null(cur, &ptr))
1608                 goto out0;
1609 
1610         XFS_BTREE_STATS_INC(cur, increment);
1611 
1612         /*
1613          * March up the tree incrementing pointers.
1614          * Stop when we don't go off the right edge of a block.
1615          */
1616         for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
1617                 block = xfs_btree_get_block(cur, lev, &bp);
1618 
1619 #ifdef DEBUG
1620                 error = xfs_btree_check_block(cur, block, lev, bp);
1621                 if (error)
1622                         goto error0;
1623 #endif
1624 
1625                 if (++cur->bc_ptrs[lev] <= xfs_btree_get_numrecs(block))
1626                         break;
1627 
1628                 /* Read-ahead the right block for the next loop. */
1629                 xfs_btree_readahead(cur, lev, XFS_BTCUR_RIGHTRA);
1630         }
1631 
1632         /*
1633          * If we went off the root then we are either seriously
1634          * confused or have the tree root in an inode.
1635          */
1636         if (lev == cur->bc_nlevels) {
1637                 if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
1638                         goto out0;
1639                 ASSERT(0);
1640                 error = -EFSCORRUPTED;
1641                 goto error0;
1642         }
1643         ASSERT(lev < cur->bc_nlevels);
1644 
1645         /*
1646          * Now walk back down the tree, fixing up the cursor's buffer
1647          * pointers and key numbers.
1648          */
1649         for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
1650                 union xfs_btree_ptr     *ptrp;
1651 
1652                 ptrp = xfs_btree_ptr_addr(cur, cur->bc_ptrs[lev], block);
1653                 --lev;
1654                 error = xfs_btree_read_buf_block(cur, ptrp, 0, &block, &bp);
1655                 if (error)
1656                         goto error0;
1657 
1658                 xfs_btree_setbuf(cur, lev, bp);
1659                 cur->bc_ptrs[lev] = 1;
1660         }
1661 out1:
1662         *stat = 1;
1663         return 0;
1664 
1665 out0:
1666         *stat = 0;
1667         return 0;
1668 
1669 error0:
1670         return error;
1671 }
1672 
1673 /*
1674  * Decrement cursor by one record at the level.
1675  * For nonzero levels the leaf-ward information is untouched.
1676  */
1677 int                                             /* error */
1678 xfs_btree_decrement(
1679         struct xfs_btree_cur    *cur,
1680         int                     level,
1681         int                     *stat)          /* success/failure */
1682 {
1683         struct xfs_btree_block  *block;
1684         xfs_buf_t               *bp;
1685         int                     error;          /* error return value */
1686         int                     lev;
1687         union xfs_btree_ptr     ptr;
1688 
1689         ASSERT(level < cur->bc_nlevels);
1690 
1691         /* Read-ahead to the left at this level. */
1692         xfs_btree_readahead(cur, level, XFS_BTCUR_LEFTRA);
1693 
1694         /* We're done if we remain in the block after the decrement. */
1695         if (--cur->bc_ptrs[level] > 0)
1696                 goto out1;
1697 
1698         /* Get a pointer to the btree block. */
1699         block = xfs_btree_get_block(cur, level, &bp);
1700 
1701 #ifdef DEBUG
1702         error = xfs_btree_check_block(cur, block, level, bp);
1703         if (error)
1704                 goto error0;
1705 #endif
1706 
1707         /* Fail if we just went off the left edge of the tree. */
1708         xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
1709         if (xfs_btree_ptr_is_null(cur, &ptr))
1710                 goto out0;
1711 
1712         XFS_BTREE_STATS_INC(cur, decrement);
1713 
1714         /*
1715          * March up the tree decrementing pointers.
1716          * Stop when we don't go off the left edge of a block.
1717          */
1718         for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
1719                 if (--cur->bc_ptrs[lev] > 0)
1720                         break;
1721                 /* Read-ahead the left block for the next loop. */
1722                 xfs_btree_readahead(cur, lev, XFS_BTCUR_LEFTRA);
1723         }
1724 
1725         /*
1726          * If we went off the root then we are seriously confused.
1727          * or the root of the tree is in an inode.
1728          */
1729         if (lev == cur->bc_nlevels) {
1730                 if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
1731                         goto out0;
1732                 ASSERT(0);
1733                 error = -EFSCORRUPTED;
1734                 goto error0;
1735         }
1736         ASSERT(lev < cur->bc_nlevels);
1737 
1738         /*
1739          * Now walk back down the tree, fixing up the cursor's buffer
1740          * pointers and key numbers.
1741          */
1742         for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
1743                 union xfs_btree_ptr     *ptrp;
1744 
1745                 ptrp = xfs_btree_ptr_addr(cur, cur->bc_ptrs[lev], block);
1746                 --lev;
1747                 error = xfs_btree_read_buf_block(cur, ptrp, 0, &block, &bp);
1748                 if (error)
1749                         goto error0;
1750                 xfs_btree_setbuf(cur, lev, bp);
1751                 cur->bc_ptrs[lev] = xfs_btree_get_numrecs(block);
1752         }
1753 out1:
1754         *stat = 1;
1755         return 0;
1756 
1757 out0:
1758         *stat = 0;
1759         return 0;
1760 
1761 error0:
1762         return error;
1763 }
1764 
1765 int
1766 xfs_btree_lookup_get_block(
1767         struct xfs_btree_cur    *cur,   /* btree cursor */
1768         int                     level,  /* level in the btree */
1769         union xfs_btree_ptr     *pp,    /* ptr to btree block */
1770         struct xfs_btree_block  **blkp) /* return btree block */
1771 {
1772         struct xfs_buf          *bp;    /* buffer pointer for btree block */
1773         xfs_daddr_t             daddr;
1774         int                     error = 0;
1775 
1776         /* special case the root block if in an inode */
1777         if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
1778             (level == cur->bc_nlevels - 1)) {
1779                 *blkp = xfs_btree_get_iroot(cur);
1780                 return 0;
1781         }
1782 
1783         /*
1784          * If the old buffer at this level for the disk address we are
1785          * looking for re-use it.
1786          *
1787          * Otherwise throw it away and get a new one.
1788          */
1789         bp = cur->bc_bufs[level];
1790         error = xfs_btree_ptr_to_daddr(cur, pp, &daddr);
1791         if (error)
1792                 return error;
1793         if (bp && XFS_BUF_ADDR(bp) == daddr) {
1794                 *blkp = XFS_BUF_TO_BLOCK(bp);
1795                 return 0;
1796         }
1797 
1798         error = xfs_btree_read_buf_block(cur, pp, 0, blkp, &bp);
1799         if (error)
1800                 return error;
1801 
1802         /* Check the inode owner since the verifiers don't. */
1803         if (xfs_sb_version_hascrc(&cur->bc_mp->m_sb) &&
1804             !(cur->bc_private.b.flags & XFS_BTCUR_BPRV_INVALID_OWNER) &&
1805             (cur->bc_flags & XFS_BTREE_LONG_PTRS) &&
1806             be64_to_cpu((*blkp)->bb_u.l.bb_owner) !=
1807                         cur->bc_private.b.ip->i_ino)
1808                 goto out_bad;
1809 
1810         /* Did we get the level we were looking for? */
1811         if (be16_to_cpu((*blkp)->bb_level) != level)
1812                 goto out_bad;
1813 
1814         /* Check that internal nodes have at least one record. */
1815         if (level != 0 && be16_to_cpu((*blkp)->bb_numrecs) == 0)
1816                 goto out_bad;
1817 
1818         xfs_btree_setbuf(cur, level, bp);
1819         return 0;
1820 
1821 out_bad:
1822         *blkp = NULL;
1823         xfs_trans_brelse(cur->bc_tp, bp);
1824         return -EFSCORRUPTED;
1825 }
1826 
1827 /*
1828  * Get current search key.  For level 0 we don't actually have a key
1829  * structure so we make one up from the record.  For all other levels
1830  * we just return the right key.
1831  */
1832 STATIC union xfs_btree_key *
1833 xfs_lookup_get_search_key(
1834         struct xfs_btree_cur    *cur,
1835         int                     level,
1836         int                     keyno,
1837         struct xfs_btree_block  *block,
1838         union xfs_btree_key     *kp)
1839 {
1840         if (level == 0) {
1841                 cur->bc_ops->init_key_from_rec(kp,
1842                                 xfs_btree_rec_addr(cur, keyno, block));
1843                 return kp;
1844         }
1845 
1846         return xfs_btree_key_addr(cur, keyno, block);
1847 }
1848 
1849 /*
1850  * Lookup the record.  The cursor is made to point to it, based on dir.
1851  * stat is set to 0 if can't find any such record, 1 for success.
1852  */
1853 int                                     /* error */
1854 xfs_btree_lookup(
1855         struct xfs_btree_cur    *cur,   /* btree cursor */
1856         xfs_lookup_t            dir,    /* <=, ==, or >= */
1857         int                     *stat)  /* success/failure */
1858 {
1859         struct xfs_btree_block  *block; /* current btree block */
1860         int64_t                 diff;   /* difference for the current key */
1861         int                     error;  /* error return value */
1862         int                     keyno;  /* current key number */
1863         int                     level;  /* level in the btree */
1864         union xfs_btree_ptr     *pp;    /* ptr to btree block */
1865         union xfs_btree_ptr     ptr;    /* ptr to btree block */
1866 
1867         XFS_BTREE_STATS_INC(cur, lookup);
1868 
1869         /* No such thing as a zero-level tree. */
1870         if (cur->bc_nlevels == 0)
1871                 return -EFSCORRUPTED;
1872 
1873         block = NULL;
1874         keyno = 0;
1875 
1876         /* initialise start pointer from cursor */
1877         cur->bc_ops->init_ptr_from_cur(cur, &ptr);
1878         pp = &ptr;
1879 
1880         /*
1881          * Iterate over each level in the btree, starting at the root.
1882          * For each level above the leaves, find the key we need, based
1883          * on the lookup record, then follow the corresponding block
1884          * pointer down to the next level.
1885          */
1886         for (level = cur->bc_nlevels - 1, diff = 1; level >= 0; level--) {
1887                 /* Get the block we need to do the lookup on. */
1888                 error = xfs_btree_lookup_get_block(cur, level, pp, &block);
1889                 if (error)
1890                         goto error0;
1891 
1892                 if (diff == 0) {
1893                         /*
1894                          * If we already had a key match at a higher level, we
1895                          * know we need to use the first entry in this block.
1896                          */
1897                         keyno = 1;
1898                 } else {
1899                         /* Otherwise search this block. Do a binary search. */
1900 
1901                         int     high;   /* high entry number */
1902                         int     low;    /* low entry number */
1903 
1904                         /* Set low and high entry numbers, 1-based. */
1905                         low = 1;
1906                         high = xfs_btree_get_numrecs(block);
1907                         if (!high) {
1908                                 /* Block is empty, must be an empty leaf. */
1909                                 if (level != 0 || cur->bc_nlevels != 1) {
1910                                         XFS_CORRUPTION_ERROR(__func__,
1911                                                         XFS_ERRLEVEL_LOW,
1912                                                         cur->bc_mp, block,
1913                                                         sizeof(*block));
1914                                         return -EFSCORRUPTED;
1915                                 }
1916 
1917                                 cur->bc_ptrs[0] = dir != XFS_LOOKUP_LE;
1918                                 *stat = 0;
1919                                 return 0;
1920                         }
1921 
1922                         /* Binary search the block. */
1923                         while (low <= high) {
1924                                 union xfs_btree_key     key;
1925                                 union xfs_btree_key     *kp;
1926 
1927                                 XFS_BTREE_STATS_INC(cur, compare);
1928 
1929                                 /* keyno is average of low and high. */
1930                                 keyno = (low + high) >> 1;
1931 
1932                                 /* Get current search key */
1933                                 kp = xfs_lookup_get_search_key(cur, level,
1934                                                 keyno, block, &key);
1935 
1936                                 /*
1937                                  * Compute difference to get next direction:
1938                                  *  - less than, move right
1939                                  *  - greater than, move left
1940                                  *  - equal, we're done
1941                                  */
1942                                 diff = cur->bc_ops->key_diff(cur, kp);
1943                                 if (diff < 0)
1944                                         low = keyno + 1;
1945                                 else if (diff > 0)
1946                                         high = keyno - 1;
1947                                 else
1948                                         break;
1949                         }
1950                 }
1951 
1952                 /*
1953                  * If there are more levels, set up for the next level
1954                  * by getting the block number and filling in the cursor.
1955                  */
1956                 if (level > 0) {
1957                         /*
1958                          * If we moved left, need the previous key number,
1959                          * unless there isn't one.
1960                          */
1961                         if (diff > 0 && --keyno < 1)
1962                                 keyno = 1;
1963                         pp = xfs_btree_ptr_addr(cur, keyno, block);
1964 
1965                         error = xfs_btree_debug_check_ptr(cur, pp, 0, level);
1966                         if (error)
1967                                 goto error0;
1968 
1969                         cur->bc_ptrs[level] = keyno;
1970                 }
1971         }
1972 
1973         /* Done with the search. See if we need to adjust the results. */
1974         if (dir != XFS_LOOKUP_LE && diff < 0) {
1975                 keyno++;
1976                 /*
1977                  * If ge search and we went off the end of the block, but it's
1978                  * not the last block, we're in the wrong block.
1979                  */
1980                 xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
1981                 if (dir == XFS_LOOKUP_GE &&
1982                     keyno > xfs_btree_get_numrecs(block) &&
1983                     !xfs_btree_ptr_is_null(cur, &ptr)) {
1984                         int     i;
1985 
1986                         cur->bc_ptrs[0] = keyno;
1987                         error = xfs_btree_increment(cur, 0, &i);
1988                         if (error)
1989                                 goto error0;
1990                         XFS_WANT_CORRUPTED_RETURN(cur->bc_mp, i == 1);
1991                         *stat = 1;
1992                         return 0;
1993                 }
1994         } else if (dir == XFS_LOOKUP_LE && diff > 0)
1995                 keyno--;
1996         cur->bc_ptrs[0] = keyno;
1997 
1998         /* Return if we succeeded or not. */
1999         if (keyno == 0 || keyno > xfs_btree_get_numrecs(block))
2000                 *stat = 0;
2001         else if (dir != XFS_LOOKUP_EQ || diff == 0)
2002                 *stat = 1;
2003         else
2004                 *stat = 0;
2005         return 0;
2006 
2007 error0:
2008         return error;
2009 }
2010 
2011 /* Find the high key storage area from a regular key. */
2012 union xfs_btree_key *
2013 xfs_btree_high_key_from_key(
2014         struct xfs_btree_cur    *cur,
2015         union xfs_btree_key     *key)
2016 {
2017         ASSERT(cur->bc_flags & XFS_BTREE_OVERLAPPING);
2018         return (union xfs_btree_key *)((char *)key +
2019                         (cur->bc_ops->key_len / 2));
2020 }
2021 
2022 /* Determine the low (and high if overlapped) keys of a leaf block */
2023 STATIC void
2024 xfs_btree_get_leaf_keys(
2025         struct xfs_btree_cur    *cur,
2026         struct xfs_btree_block  *block,
2027         union xfs_btree_key     *key)
2028 {
2029         union xfs_btree_key     max_hkey;
2030         union xfs_btree_key     hkey;
2031         union xfs_btree_rec     *rec;
2032         union xfs_btree_key     *high;
2033         int                     n;
2034 
2035         rec = xfs_btree_rec_addr(cur, 1, block);
2036         cur->bc_ops->init_key_from_rec(key, rec);
2037 
2038         if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2039 
2040                 cur->bc_ops->init_high_key_from_rec(&max_hkey, rec);
2041                 for (n = 2; n <= xfs_btree_get_numrecs(block); n++) {
2042                         rec = xfs_btree_rec_addr(cur, n, block);
2043                         cur->bc_ops->init_high_key_from_rec(&hkey, rec);
2044                         if (cur->bc_ops->diff_two_keys(cur, &hkey, &max_hkey)
2045                                         > 0)
2046                                 max_hkey = hkey;
2047                 }
2048 
2049                 high = xfs_btree_high_key_from_key(cur, key);
2050                 memcpy(high, &max_hkey, cur->bc_ops->key_len / 2);
2051         }
2052 }
2053 
2054 /* Determine the low (and high if overlapped) keys of a node block */
2055 STATIC void
2056 xfs_btree_get_node_keys(
2057         struct xfs_btree_cur    *cur,
2058         struct xfs_btree_block  *block,
2059         union xfs_btree_key     *key)
2060 {
2061         union xfs_btree_key     *hkey;
2062         union xfs_btree_key     *max_hkey;
2063         union xfs_btree_key     *high;
2064         int                     n;
2065 
2066         if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2067                 memcpy(key, xfs_btree_key_addr(cur, 1, block),
2068                                 cur->bc_ops->key_len / 2);
2069 
2070                 max_hkey = xfs_btree_high_key_addr(cur, 1, block);
2071                 for (n = 2; n <= xfs_btree_get_numrecs(block); n++) {
2072                         hkey = xfs_btree_high_key_addr(cur, n, block);
2073                         if (cur->bc_ops->diff_two_keys(cur, hkey, max_hkey) > 0)
2074                                 max_hkey = hkey;
2075                 }
2076 
2077                 high = xfs_btree_high_key_from_key(cur, key);
2078                 memcpy(high, max_hkey, cur->bc_ops->key_len / 2);
2079         } else {
2080                 memcpy(key, xfs_btree_key_addr(cur, 1, block),
2081                                 cur->bc_ops->key_len);
2082         }
2083 }
2084 
2085 /* Derive the keys for any btree block. */
2086 void
2087 xfs_btree_get_keys(
2088         struct xfs_btree_cur    *cur,
2089         struct xfs_btree_block  *block,
2090         union xfs_btree_key     *key)
2091 {
2092         if (be16_to_cpu(block->bb_level) == 0)
2093                 xfs_btree_get_leaf_keys(cur, block, key);
2094         else
2095                 xfs_btree_get_node_keys(cur, block, key);
2096 }
2097 
2098 /*
2099  * Decide if we need to update the parent keys of a btree block.  For
2100  * a standard btree this is only necessary if we're updating the first
2101  * record/key.  For an overlapping btree, we must always update the
2102  * keys because the highest key can be in any of the records or keys
2103  * in the block.
2104  */
2105 static inline bool
2106 xfs_btree_needs_key_update(
2107         struct xfs_btree_cur    *cur,
2108         int                     ptr)
2109 {
2110         return (cur->bc_flags & XFS_BTREE_OVERLAPPING) || ptr == 1;
2111 }
2112 
2113 /*
2114  * Update the low and high parent keys of the given level, progressing
2115  * towards the root.  If force_all is false, stop if the keys for a given
2116  * level do not need updating.
2117  */
2118 STATIC int
2119 __xfs_btree_updkeys(
2120         struct xfs_btree_cur    *cur,
2121         int                     level,
2122         struct xfs_btree_block  *block,
2123         struct xfs_buf          *bp0,
2124         bool                    force_all)
2125 {
2126         union xfs_btree_key     key;    /* keys from current level */
2127         union xfs_btree_key     *lkey;  /* keys from the next level up */
2128         union xfs_btree_key     *hkey;
2129         union xfs_btree_key     *nlkey; /* keys from the next level up */
2130         union xfs_btree_key     *nhkey;
2131         struct xfs_buf          *bp;
2132         int                     ptr;
2133 
2134         ASSERT(cur->bc_flags & XFS_BTREE_OVERLAPPING);
2135 
2136         /* Exit if there aren't any parent levels to update. */
2137         if (level + 1 >= cur->bc_nlevels)
2138                 return 0;
2139 
2140         trace_xfs_btree_updkeys(cur, level, bp0);
2141 
2142         lkey = &key;
2143         hkey = xfs_btree_high_key_from_key(cur, lkey);
2144         xfs_btree_get_keys(cur, block, lkey);
2145         for (level++; level < cur->bc_nlevels; level++) {
2146 #ifdef DEBUG
2147                 int             error;
2148 #endif
2149                 block = xfs_btree_get_block(cur, level, &bp);
2150                 trace_xfs_btree_updkeys(cur, level, bp);
2151 #ifdef DEBUG
2152                 error = xfs_btree_check_block(cur, block, level, bp);
2153                 if (error)
2154                         return error;
2155 #endif
2156                 ptr = cur->bc_ptrs[level];
2157                 nlkey = xfs_btree_key_addr(cur, ptr, block);
2158                 nhkey = xfs_btree_high_key_addr(cur, ptr, block);
2159                 if (!force_all &&
2160                     !(cur->bc_ops->diff_two_keys(cur, nlkey, lkey) != 0 ||
2161                       cur->bc_ops->diff_two_keys(cur, nhkey, hkey) != 0))
2162                         break;
2163                 xfs_btree_copy_keys(cur, nlkey, lkey, 1);
2164                 xfs_btree_log_keys(cur, bp, ptr, ptr);
2165                 if (level + 1 >= cur->bc_nlevels)
2166                         break;
2167                 xfs_btree_get_node_keys(cur, block, lkey);
2168         }
2169 
2170         return 0;
2171 }
2172 
2173 /* Update all the keys from some level in cursor back to the root. */
2174 STATIC int
2175 xfs_btree_updkeys_force(
2176         struct xfs_btree_cur    *cur,
2177         int                     level)
2178 {
2179         struct xfs_buf          *bp;
2180         struct xfs_btree_block  *block;
2181 
2182         block = xfs_btree_get_block(cur, level, &bp);
2183         return __xfs_btree_updkeys(cur, level, block, bp, true);
2184 }
2185 
2186 /*
2187  * Update the parent keys of the given level, progressing towards the root.
2188  */
2189 STATIC int
2190 xfs_btree_update_keys(
2191         struct xfs_btree_cur    *cur,
2192         int                     level)
2193 {
2194         struct xfs_btree_block  *block;
2195         struct xfs_buf          *bp;
2196         union xfs_btree_key     *kp;
2197         union xfs_btree_key     key;
2198         int                     ptr;
2199 
2200         ASSERT(level >= 0);
2201 
2202         block = xfs_btree_get_block(cur, level, &bp);
2203         if (cur->bc_flags & XFS_BTREE_OVERLAPPING)
2204                 return __xfs_btree_updkeys(cur, level, block, bp, false);
2205 
2206         /*
2207          * Go up the tree from this level toward the root.
2208          * At each level, update the key value to the value input.
2209          * Stop when we reach a level where the cursor isn't pointing
2210          * at the first entry in the block.
2211          */
2212         xfs_btree_get_keys(cur, block, &key);
2213         for (level++, ptr = 1; ptr == 1 && level < cur->bc_nlevels; level++) {
2214 #ifdef DEBUG
2215                 int             error;
2216 #endif
2217                 block = xfs_btree_get_block(cur, level, &bp);
2218 #ifdef DEBUG
2219                 error = xfs_btree_check_block(cur, block, level, bp);
2220                 if (error)
2221                         return error;
2222 #endif
2223                 ptr = cur->bc_ptrs[level];
2224                 kp = xfs_btree_key_addr(cur, ptr, block);
2225                 xfs_btree_copy_keys(cur, kp, &key, 1);
2226                 xfs_btree_log_keys(cur, bp, ptr, ptr);
2227         }
2228 
2229         return 0;
2230 }
2231 
2232 /*
2233  * Update the record referred to by cur to the value in the
2234  * given record. This either works (return 0) or gets an
2235  * EFSCORRUPTED error.
2236  */
2237 int
2238 xfs_btree_update(
2239         struct xfs_btree_cur    *cur,
2240         union xfs_btree_rec     *rec)
2241 {
2242         struct xfs_btree_block  *block;
2243         struct xfs_buf          *bp;
2244         int                     error;
2245         int                     ptr;
2246         union xfs_btree_rec     *rp;
2247 
2248         /* Pick up the current block. */
2249         block = xfs_btree_get_block(cur, 0, &bp);
2250 
2251 #ifdef DEBUG
2252         error = xfs_btree_check_block(cur, block, 0, bp);
2253         if (error)
2254                 goto error0;
2255 #endif
2256         /* Get the address of the rec to be updated. */
2257         ptr = cur->bc_ptrs[0];
2258         rp = xfs_btree_rec_addr(cur, ptr, block);
2259 
2260         /* Fill in the new contents and log them. */
2261         xfs_btree_copy_recs(cur, rp, rec, 1);
2262         xfs_btree_log_recs(cur, bp, ptr, ptr);
2263 
2264         /*
2265          * If we are tracking the last record in the tree and
2266          * we are at the far right edge of the tree, update it.
2267          */
2268         if (xfs_btree_is_lastrec(cur, block, 0)) {
2269                 cur->bc_ops->update_lastrec(cur, block, rec,
2270                                             ptr, LASTREC_UPDATE);
2271         }
2272 
2273         /* Pass new key value up to our parent. */
2274         if (xfs_btree_needs_key_update(cur, ptr)) {
2275                 error = xfs_btree_update_keys(cur, 0);
2276                 if (error)
2277                         goto error0;
2278         }
2279 
2280         return 0;
2281 
2282 error0:
2283         return error;
2284 }
2285 
2286 /*
2287  * Move 1 record left from cur/level if possible.
2288  * Update cur to reflect the new path.
2289  */
2290 STATIC int                                      /* error */
2291 xfs_btree_lshift(
2292         struct xfs_btree_cur    *cur,
2293         int                     level,
2294         int                     *stat)          /* success/failure */
2295 {
2296         struct xfs_buf          *lbp;           /* left buffer pointer */
2297         struct xfs_btree_block  *left;          /* left btree block */
2298         int                     lrecs;          /* left record count */
2299         struct xfs_buf          *rbp;           /* right buffer pointer */
2300         struct xfs_btree_block  *right;         /* right btree block */
2301         struct xfs_btree_cur    *tcur;          /* temporary btree cursor */
2302         int                     rrecs;          /* right record count */
2303         union xfs_btree_ptr     lptr;           /* left btree pointer */
2304         union xfs_btree_key     *rkp = NULL;    /* right btree key */
2305         union xfs_btree_ptr     *rpp = NULL;    /* right address pointer */
2306         union xfs_btree_rec     *rrp = NULL;    /* right record pointer */
2307         int                     error;          /* error return value */
2308         int                     i;
2309 
2310         if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
2311             level == cur->bc_nlevels - 1)
2312                 goto out0;
2313 
2314         /* Set up variables for this block as "right". */
2315         right = xfs_btree_get_block(cur, level, &rbp);
2316 
2317 #ifdef DEBUG
2318         error = xfs_btree_check_block(cur, right, level, rbp);
2319         if (error)
2320                 goto error0;
2321 #endif
2322 
2323         /* If we've got no left sibling then we can't shift an entry left. */
2324         xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
2325         if (xfs_btree_ptr_is_null(cur, &lptr))
2326                 goto out0;
2327 
2328         /*
2329          * If the cursor entry is the one that would be moved, don't
2330          * do it... it's too complicated.
2331          */
2332         if (cur->bc_ptrs[level] <= 1)
2333                 goto out0;
2334 
2335         /* Set up the left neighbor as "left". */
2336         error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
2337         if (error)
2338                 goto error0;
2339 
2340         /* If it's full, it can't take another entry. */
2341         lrecs = xfs_btree_get_numrecs(left);
2342         if (lrecs == cur->bc_ops->get_maxrecs(cur, level))
2343                 goto out0;
2344 
2345         rrecs = xfs_btree_get_numrecs(right);
2346 
2347         /*
2348          * We add one entry to the left side and remove one for the right side.
2349          * Account for it here, the changes will be updated on disk and logged
2350          * later.
2351          */
2352         lrecs++;
2353         rrecs--;
2354 
2355         XFS_BTREE_STATS_INC(cur, lshift);
2356         XFS_BTREE_STATS_ADD(cur, moves, 1);
2357 
2358         /*
2359          * If non-leaf, copy a key and a ptr to the left block.
2360          * Log the changes to the left block.
2361          */
2362         if (level > 0) {
2363                 /* It's a non-leaf.  Move keys and pointers. */
2364                 union xfs_btree_key     *lkp;   /* left btree key */
2365                 union xfs_btree_ptr     *lpp;   /* left address pointer */
2366 
2367                 lkp = xfs_btree_key_addr(cur, lrecs, left);
2368                 rkp = xfs_btree_key_addr(cur, 1, right);
2369 
2370                 lpp = xfs_btree_ptr_addr(cur, lrecs, left);
2371                 rpp = xfs_btree_ptr_addr(cur, 1, right);
2372 
2373                 error = xfs_btree_debug_check_ptr(cur, rpp, 0, level);
2374                 if (error)
2375                         goto error0;
2376 
2377                 xfs_btree_copy_keys(cur, lkp, rkp, 1);
2378                 xfs_btree_copy_ptrs(cur, lpp, rpp, 1);
2379 
2380                 xfs_btree_log_keys(cur, lbp, lrecs, lrecs);
2381                 xfs_btree_log_ptrs(cur, lbp, lrecs, lrecs);
2382 
2383                 ASSERT(cur->bc_ops->keys_inorder(cur,
2384                         xfs_btree_key_addr(cur, lrecs - 1, left), lkp));
2385         } else {
2386                 /* It's a leaf.  Move records.  */
2387                 union xfs_btree_rec     *lrp;   /* left record pointer */
2388 
2389                 lrp = xfs_btree_rec_addr(cur, lrecs, left);
2390                 rrp = xfs_btree_rec_addr(cur, 1, right);
2391 
2392                 xfs_btree_copy_recs(cur, lrp, rrp, 1);
2393                 xfs_btree_log_recs(cur, lbp, lrecs, lrecs);
2394 
2395                 ASSERT(cur->bc_ops->recs_inorder(cur,
2396                         xfs_btree_rec_addr(cur, lrecs - 1, left), lrp));
2397         }
2398 
2399         xfs_btree_set_numrecs(left, lrecs);
2400         xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
2401 
2402         xfs_btree_set_numrecs(right, rrecs);
2403         xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
2404 
2405         /*
2406          * Slide the contents of right down one entry.
2407          */
2408         XFS_BTREE_STATS_ADD(cur, moves, rrecs - 1);
2409         if (level > 0) {
2410                 /* It's a nonleaf. operate on keys and ptrs */
2411                 int                     i;              /* loop index */
2412 
2413                 for (i = 0; i < rrecs; i++) {
2414                         error = xfs_btree_debug_check_ptr(cur, rpp, i + 1, level);
2415                         if (error)
2416                                 goto error0;
2417                 }
2418 
2419                 xfs_btree_shift_keys(cur,
2420                                 xfs_btree_key_addr(cur, 2, right),
2421                                 -1, rrecs);
2422                 xfs_btree_shift_ptrs(cur,
2423                                 xfs_btree_ptr_addr(cur, 2, right),
2424                                 -1, rrecs);
2425 
2426                 xfs_btree_log_keys(cur, rbp, 1, rrecs);
2427                 xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
2428         } else {
2429                 /* It's a leaf. operate on records */
2430                 xfs_btree_shift_recs(cur,
2431                         xfs_btree_rec_addr(cur, 2, right),
2432                         -1, rrecs);
2433                 xfs_btree_log_recs(cur, rbp, 1, rrecs);
2434         }
2435 
2436         /*
2437          * Using a temporary cursor, update the parent key values of the
2438          * block on the left.
2439          */
2440         if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2441                 error = xfs_btree_dup_cursor(cur, &tcur);
2442                 if (error)
2443                         goto error0;
2444                 i = xfs_btree_firstrec(tcur, level);
2445                 XFS_WANT_CORRUPTED_GOTO(tcur->bc_mp, i == 1, error0);
2446 
2447                 error = xfs_btree_decrement(tcur, level, &i);
2448                 if (error)
2449                         goto error1;
2450 
2451                 /* Update the parent high keys of the left block, if needed. */
2452                 error = xfs_btree_update_keys(tcur, level);
2453                 if (error)
2454                         goto error1;
2455 
2456                 xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
2457         }
2458 
2459         /* Update the parent keys of the right block. */
2460         error = xfs_btree_update_keys(cur, level);
2461         if (error)
2462                 goto error0;
2463 
2464         /* Slide the cursor value left one. */
2465         cur->bc_ptrs[level]--;
2466 
2467         *stat = 1;
2468         return 0;
2469 
2470 out0:
2471         *stat = 0;
2472         return 0;
2473 
2474 error0:
2475         return error;
2476 
2477 error1:
2478         xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
2479         return error;
2480 }
2481 
2482 /*
2483  * Move 1 record right from cur/level if possible.
2484  * Update cur to reflect the new path.
2485  */
2486 STATIC int                                      /* error */
2487 xfs_btree_rshift(
2488         struct xfs_btree_cur    *cur,
2489         int                     level,
2490         int                     *stat)          /* success/failure */
2491 {
2492         struct xfs_buf          *lbp;           /* left buffer pointer */
2493         struct xfs_btree_block  *left;          /* left btree block */
2494         struct xfs_buf          *rbp;           /* right buffer pointer */
2495         struct xfs_btree_block  *right;         /* right btree block */
2496         struct xfs_btree_cur    *tcur;          /* temporary btree cursor */
2497         union xfs_btree_ptr     rptr;           /* right block pointer */
2498         union xfs_btree_key     *rkp;           /* right btree key */
2499         int                     rrecs;          /* right record count */
2500         int                     lrecs;          /* left record count */
2501         int                     error;          /* error return value */
2502         int                     i;              /* loop counter */
2503 
2504         if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
2505             (level == cur->bc_nlevels - 1))
2506                 goto out0;
2507 
2508         /* Set up variables for this block as "left". */
2509         left = xfs_btree_get_block(cur, level, &lbp);
2510 
2511 #ifdef DEBUG
2512         error = xfs_btree_check_block(cur, left, level, lbp);
2513         if (error)
2514                 goto error0;
2515 #endif
2516 
2517         /* If we've got no right sibling then we can't shift an entry right. */
2518         xfs_btree_get_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
2519         if (xfs_btree_ptr_is_null(cur, &rptr))
2520                 goto out0;
2521 
2522         /*
2523          * If the cursor entry is the one that would be moved, don't
2524          * do it... it's too complicated.
2525          */
2526         lrecs = xfs_btree_get_numrecs(left);
2527         if (cur->bc_ptrs[level] >= lrecs)
2528                 goto out0;
2529 
2530         /* Set up the right neighbor as "right". */
2531         error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
2532         if (error)
2533                 goto error0;
2534 
2535         /* If it's full, it can't take another entry. */
2536         rrecs = xfs_btree_get_numrecs(right);
2537         if (rrecs == cur->bc_ops->get_maxrecs(cur, level))
2538                 goto out0;
2539 
2540         XFS_BTREE_STATS_INC(cur, rshift);
2541         XFS_BTREE_STATS_ADD(cur, moves, rrecs);
2542 
2543         /*
2544          * Make a hole at the start of the right neighbor block, then
2545          * copy the last left block entry to the hole.
2546          */
2547         if (level > 0) {
2548                 /* It's a nonleaf. make a hole in the keys and ptrs */
2549                 union xfs_btree_key     *lkp;
2550                 union xfs_btree_ptr     *lpp;
2551                 union xfs_btree_ptr     *rpp;
2552 
2553                 lkp = xfs_btree_key_addr(cur, lrecs, left);
2554                 lpp = xfs_btree_ptr_addr(cur, lrecs, left);
2555                 rkp = xfs_btree_key_addr(cur, 1, right);
2556                 rpp = xfs_btree_ptr_addr(cur, 1, right);
2557 
2558                 for (i = rrecs - 1; i >= 0; i--) {
2559                         error = xfs_btree_debug_check_ptr(cur, rpp, i, level);
2560                         if (error)
2561                                 goto error0;
2562                 }
2563 
2564                 xfs_btree_shift_keys(cur, rkp, 1, rrecs);
2565                 xfs_btree_shift_ptrs(cur, rpp, 1, rrecs);
2566 
2567                 error = xfs_btree_debug_check_ptr(cur, lpp, 0, level);
2568                 if (error)
2569                         goto error0;
2570 
2571                 /* Now put the new data in, and log it. */
2572                 xfs_btree_copy_keys(cur, rkp, lkp, 1);
2573                 xfs_btree_copy_ptrs(cur, rpp, lpp, 1);
2574 
2575                 xfs_btree_log_keys(cur, rbp, 1, rrecs + 1);
2576                 xfs_btree_log_ptrs(cur, rbp, 1, rrecs + 1);
2577 
2578                 ASSERT(cur->bc_ops->keys_inorder(cur, rkp,
2579                         xfs_btree_key_addr(cur, 2, right)));
2580         } else {
2581                 /* It's a leaf. make a hole in the records */
2582                 union xfs_btree_rec     *lrp;
2583                 union xfs_btree_rec     *rrp;
2584 
2585                 lrp = xfs_btree_rec_addr(cur, lrecs, left);
2586                 rrp = xfs_btree_rec_addr(cur, 1, right);
2587 
2588                 xfs_btree_shift_recs(cur, rrp, 1, rrecs);
2589 
2590                 /* Now put the new data in, and log it. */
2591                 xfs_btree_copy_recs(cur, rrp, lrp, 1);
2592                 xfs_btree_log_recs(cur, rbp, 1, rrecs + 1);
2593         }
2594 
2595         /*
2596          * Decrement and log left's numrecs, bump and log right's numrecs.
2597          */
2598         xfs_btree_set_numrecs(left, --lrecs);
2599         xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
2600 
2601         xfs_btree_set_numrecs(right, ++rrecs);
2602         xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
2603 
2604         /*
2605          * Using a temporary cursor, update the parent key values of the
2606          * block on the right.
2607          */
2608         error = xfs_btree_dup_cursor(cur, &tcur);
2609         if (error)
2610                 goto error0;
2611         i = xfs_btree_lastrec(tcur, level);
2612         XFS_WANT_CORRUPTED_GOTO(tcur->bc_mp, i == 1, error0);
2613 
2614         error = xfs_btree_increment(tcur, level, &i);
2615         if (error)
2616                 goto error1;
2617 
2618         /* Update the parent high keys of the left block, if needed. */
2619         if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2620                 error = xfs_btree_update_keys(cur, level);
2621                 if (error)
2622                         goto error1;
2623         }
2624 
2625         /* Update the parent keys of the right block. */
2626         error = xfs_btree_update_keys(tcur, level);
2627         if (error)
2628                 goto error1;
2629 
2630         xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
2631 
2632         *stat = 1;
2633         return 0;
2634 
2635 out0:
2636         *stat = 0;
2637         return 0;
2638 
2639 error0:
2640         return error;
2641 
2642 error1:
2643         xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
2644         return error;
2645 }
2646 
2647 /*
2648  * Split cur/level block in half.
2649  * Return new block number and the key to its first
2650  * record (to be inserted into parent).
2651  */
2652 STATIC int                                      /* error */
2653 __xfs_btree_split(
2654         struct xfs_btree_cur    *cur,
2655         int                     level,
2656         union xfs_btree_ptr     *ptrp,
2657         union xfs_btree_key     *key,
2658         struct xfs_btree_cur    **curp,
2659         int                     *stat)          /* success/failure */
2660 {
2661         union xfs_btree_ptr     lptr;           /* left sibling block ptr */
2662         struct xfs_buf          *lbp;           /* left buffer pointer */
2663         struct xfs_btree_block  *left;          /* left btree block */
2664         union xfs_btree_ptr     rptr;           /* right sibling block ptr */
2665         struct xfs_buf          *rbp;           /* right buffer pointer */
2666         struct xfs_btree_block  *right;         /* right btree block */
2667         union xfs_btree_ptr     rrptr;          /* right-right sibling ptr */
2668         struct xfs_buf          *rrbp;          /* right-right buffer pointer */
2669         struct xfs_btree_block  *rrblock;       /* right-right btree block */
2670         int                     lrecs;
2671         int                     rrecs;
2672         int                     src_index;
2673         int                     error;          /* error return value */
2674         int                     i;
2675 
2676         XFS_BTREE_STATS_INC(cur, split);
2677 
2678         /* Set up left block (current one). */
2679         left = xfs_btree_get_block(cur, level, &lbp);
2680 
2681 #ifdef DEBUG
2682         error = xfs_btree_check_block(cur, left, level, lbp);
2683         if (error)
2684                 goto error0;
2685 #endif
2686 
2687         xfs_btree_buf_to_ptr(cur, lbp, &lptr);
2688 
2689         /* Allocate the new block. If we can't do it, we're toast. Give up. */
2690         error = cur->bc_ops->alloc_block(cur, &lptr, &rptr, stat);
2691         if (error)
2692                 goto error0;
2693         if (*stat == 0)
2694                 goto out0;
2695         XFS_BTREE_STATS_INC(cur, alloc);
2696 
2697         /* Set up the new block as "right". */
2698         error = xfs_btree_get_buf_block(cur, &rptr, &right, &rbp);
2699         if (error)
2700                 goto error0;
2701 
2702         /* Fill in the btree header for the new right block. */
2703         xfs_btree_init_block_cur(cur, rbp, xfs_btree_get_level(left), 0);
2704 
2705         /*
2706          * Split the entries between the old and the new block evenly.
2707          * Make sure that if there's an odd number of entries now, that
2708          * each new block will have the same number of entries.
2709          */
2710         lrecs = xfs_btree_get_numrecs(left);
2711         rrecs = lrecs / 2;
2712         if ((lrecs & 1) && cur->bc_ptrs[level] <= rrecs + 1)
2713                 rrecs++;
2714         src_index = (lrecs - rrecs + 1);
2715 
2716         XFS_BTREE_STATS_ADD(cur, moves, rrecs);
2717 
2718         /* Adjust numrecs for the later get_*_keys() calls. */
2719         lrecs -= rrecs;
2720         xfs_btree_set_numrecs(left, lrecs);
2721         xfs_btree_set_numrecs(right, xfs_btree_get_numrecs(right) + rrecs);
2722 
2723         /*
2724          * Copy btree block entries from the left block over to the
2725          * new block, the right. Update the right block and log the
2726          * changes.
2727          */
2728         if (level > 0) {
2729                 /* It's a non-leaf.  Move keys and pointers. */
2730                 union xfs_btree_key     *lkp;   /* left btree key */
2731                 union xfs_btree_ptr     *lpp;   /* left address pointer */
2732                 union xfs_btree_key     *rkp;   /* right btree key */
2733                 union xfs_btree_ptr     *rpp;   /* right address pointer */
2734 
2735                 lkp = xfs_btree_key_addr(cur, src_index, left);
2736                 lpp = xfs_btree_ptr_addr(cur, src_index, left);
2737                 rkp = xfs_btree_key_addr(cur, 1, right);
2738                 rpp = xfs_btree_ptr_addr(cur, 1, right);
2739 
2740                 for (i = src_index; i < rrecs; i++) {
2741                         error = xfs_btree_debug_check_ptr(cur, lpp, i, level);
2742                         if (error)
2743                                 goto error0;
2744                 }
2745 
2746                 /* Copy the keys & pointers to the new block. */
2747                 xfs_btree_copy_keys(cur, rkp, lkp, rrecs);
2748                 xfs_btree_copy_ptrs(cur, rpp, lpp, rrecs);
2749 
2750                 xfs_btree_log_keys(cur, rbp, 1, rrecs);
2751                 xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
2752 
2753                 /* Stash the keys of the new block for later insertion. */
2754                 xfs_btree_get_node_keys(cur, right, key);
2755         } else {
2756                 /* It's a leaf.  Move records.  */
2757                 union xfs_btree_rec     *lrp;   /* left record pointer */
2758                 union xfs_btree_rec     *rrp;   /* right record pointer */
2759 
2760                 lrp = xfs_btree_rec_addr(cur, src_index, left);
2761                 rrp = xfs_btree_rec_addr(cur, 1, right);
2762 
2763                 /* Copy records to the new block. */
2764                 xfs_btree_copy_recs(cur, rrp, lrp, rrecs);
2765                 xfs_btree_log_recs(cur, rbp, 1, rrecs);
2766 
2767                 /* Stash the keys of the new block for later insertion. */
2768                 xfs_btree_get_leaf_keys(cur, right, key);
2769         }
2770 
2771         /*
2772          * Find the left block number by looking in the buffer.
2773          * Adjust sibling pointers.
2774          */
2775         xfs_btree_get_sibling(cur, left, &rrptr, XFS_BB_RIGHTSIB);
2776         xfs_btree_set_sibling(cur, right, &rrptr, XFS_BB_RIGHTSIB);
2777         xfs_btree_set_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
2778         xfs_btree_set_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
2779 
2780         xfs_btree_log_block(cur, rbp, XFS_BB_ALL_BITS);
2781         xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
2782 
2783         /*
2784          * If there's a block to the new block's right, make that block
2785          * point back to right instead of to left.
2786          */
2787         if (!xfs_btree_ptr_is_null(cur, &rrptr)) {
2788                 error = xfs_btree_read_buf_block(cur, &rrptr,
2789                                                         0, &rrblock, &rrbp);
2790                 if (error)
2791                         goto error0;
2792                 xfs_btree_set_sibling(cur, rrblock, &rptr, XFS_BB_LEFTSIB);
2793                 xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
2794         }
2795 
2796         /* Update the parent high keys of the left block, if needed. */
2797         if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2798                 error = xfs_btree_update_keys(cur, level);
2799                 if (error)
2800                         goto error0;
2801         }
2802 
2803         /*
2804          * If the cursor is really in the right block, move it there.
2805          * If it's just pointing past the last entry in left, then we'll
2806          * insert there, so don't change anything in that case.
2807          */
2808         if (cur->bc_ptrs[level] > lrecs + 1) {
2809                 xfs_btree_setbuf(cur, level, rbp);
2810                 cur->bc_ptrs[level] -= lrecs;
2811         }
2812         /*
2813          * If there are more levels, we'll need another cursor which refers
2814          * the right block, no matter where this cursor was.
2815          */
2816         if (level + 1 < cur->bc_nlevels) {
2817                 error = xfs_btree_dup_cursor(cur, curp);
2818                 if (error)
2819                         goto error0;
2820                 (*curp)->bc_ptrs[level + 1]++;
2821         }
2822         *ptrp = rptr;
2823         *stat = 1;
2824         return 0;
2825 out0:
2826         *stat = 0;
2827         return 0;
2828 
2829 error0:
2830         return error;
2831 }
2832 
2833 struct xfs_btree_split_args {
2834         struct xfs_btree_cur    *cur;
2835         int                     level;
2836         union xfs_btree_ptr     *ptrp;
2837         union xfs_btree_key     *key;
2838         struct xfs_btree_cur    **curp;
2839         int                     *stat;          /* success/failure */
2840         int                     result;
2841         bool                    kswapd; /* allocation in kswapd context */
2842         struct completion       *done;
2843         struct work_struct      work;
2844 };
2845 
2846 /*
2847  * Stack switching interfaces for allocation
2848  */
2849 static void
2850 xfs_btree_split_worker(
2851         struct work_struct      *work)
2852 {
2853         struct xfs_btree_split_args     *args = container_of(work,
2854                                                 struct xfs_btree_split_args, work);
2855         unsigned long           pflags;
2856         unsigned long           new_pflags = PF_MEMALLOC_NOFS;
2857 
2858         /*
2859          * we are in a transaction context here, but may also be doing work
2860          * in kswapd context, and hence we may need to inherit that state
2861          * temporarily to ensure that we don't block waiting for memory reclaim
2862          * in any way.
2863          */
2864         if (args->kswapd)
2865                 new_pflags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD;
2866 
2867         current_set_flags_nested(&pflags, new_pflags);
2868 
2869         args->result = __xfs_btree_split(args->cur, args->level, args->ptrp,
2870                                          args->key, args->curp, args->stat);
2871         complete(args->done);
2872 
2873         current_restore_flags_nested(&pflags, new_pflags);
2874 }
2875 
2876 /*
2877  * BMBT split requests often come in with little stack to work on. Push
2878  * them off to a worker thread so there is lots of stack to use. For the other
2879  * btree types, just call directly to avoid the context switch overhead here.
2880  */
2881 STATIC int                                      /* error */
2882 xfs_btree_split(
2883         struct xfs_btree_cur    *cur,
2884         int                     level,
2885         union xfs_btree_ptr     *ptrp,
2886         union xfs_btree_key     *key,
2887         struct xfs_btree_cur    **curp,
2888         int                     *stat)          /* success/failure */
2889 {
2890         struct xfs_btree_split_args     args;
2891         DECLARE_COMPLETION_ONSTACK(done);
2892 
2893         if (cur->bc_btnum != XFS_BTNUM_BMAP)
2894                 return __xfs_btree_split(cur, level, ptrp, key, curp, stat);
2895 
2896         args.cur = cur;
2897         args.level = level;
2898         args.ptrp = ptrp;
2899         args.key = key;
2900         args.curp = curp;
2901         args.stat = stat;
2902         args.done = &done;
2903         args.kswapd = current_is_kswapd();
2904         INIT_WORK_ONSTACK(&args.work, xfs_btree_split_worker);
2905         queue_work(xfs_alloc_wq, &args.work);
2906         wait_for_completion(&done);
2907         destroy_work_on_stack(&args.work);
2908         return args.result;
2909 }
2910 
2911 
2912 /*
2913  * Copy the old inode root contents into a real block and make the
2914  * broot point to it.
2915  */
2916 int                                             /* error */
2917 xfs_btree_new_iroot(
2918         struct xfs_btree_cur    *cur,           /* btree cursor */
2919         int                     *logflags,      /* logging flags for inode */
2920         int                     *stat)          /* return status - 0 fail */
2921 {
2922         struct xfs_buf          *cbp;           /* buffer for cblock */
2923         struct xfs_btree_block  *block;         /* btree block */
2924         struct xfs_btree_block  *cblock;        /* child btree block */
2925         union xfs_btree_key     *ckp;           /* child key pointer */
2926         union xfs_btree_ptr     *cpp;           /* child ptr pointer */
2927         union xfs_btree_key     *kp;            /* pointer to btree key */
2928         union xfs_btree_ptr     *pp;            /* pointer to block addr */
2929         union xfs_btree_ptr     nptr;           /* new block addr */
2930         int                     level;          /* btree level */
2931         int                     error;          /* error return code */
2932         int                     i;              /* loop counter */
2933 
2934         XFS_BTREE_STATS_INC(cur, newroot);
2935 
2936         ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
2937 
2938         level = cur->bc_nlevels - 1;
2939 
2940         block = xfs_btree_get_iroot(cur);
2941         pp = xfs_btree_ptr_addr(cur, 1, block);
2942 
2943         /* Allocate the new block. If we can't do it, we're toast. Give up. */
2944         error = cur->bc_ops->alloc_block(cur, pp, &nptr, stat);
2945         if (error)
2946                 goto error0;
2947         if (*stat == 0)
2948                 return 0;
2949 
2950         XFS_BTREE_STATS_INC(cur, alloc);
2951 
2952         /* Copy the root into a real block. */
2953         error = xfs_btree_get_buf_block(cur, &nptr, &cblock, &cbp);
2954         if (error)
2955                 goto error0;
2956 
2957         /*
2958          * we can't just memcpy() the root in for CRC enabled btree blocks.
2959          * In that case have to also ensure the blkno remains correct
2960          */
2961         memcpy(cblock, block, xfs_btree_block_len(cur));
2962         if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS) {
2963                 if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
2964                         cblock->bb_u.l.bb_blkno = cpu_to_be64(cbp->b_bn);
2965                 else
2966                         cblock->bb_u.s.bb_blkno = cpu_to_be64(cbp->b_bn);
2967         }
2968 
2969         be16_add_cpu(&block->bb_level, 1);
2970         xfs_btree_set_numrecs(block, 1);
2971         cur->bc_nlevels++;
2972         cur->bc_ptrs[level + 1] = 1;
2973 
2974         kp = xfs_btree_key_addr(cur, 1, block);
2975         ckp = xfs_btree_key_addr(cur, 1, cblock);
2976         xfs_btree_copy_keys(cur, ckp, kp, xfs_btree_get_numrecs(cblock));
2977 
2978         cpp = xfs_btree_ptr_addr(cur, 1, cblock);
2979         for (i = 0; i < be16_to_cpu(cblock->bb_numrecs); i++) {
2980                 error = xfs_btree_debug_check_ptr(cur, pp, i, level);
2981                 if (error)
2982                         goto error0;
2983         }
2984 
2985         xfs_btree_copy_ptrs(cur, cpp, pp, xfs_btree_get_numrecs(cblock));
2986 
2987         error = xfs_btree_debug_check_ptr(cur, &nptr, 0, level);
2988         if (error)
2989                 goto error0;
2990 
2991         xfs_btree_copy_ptrs(cur, pp, &nptr, 1);
2992 
2993         xfs_iroot_realloc(cur->bc_private.b.ip,
2994                           1 - xfs_btree_get_numrecs(cblock),
2995                           cur->bc_private.b.whichfork);
2996 
2997         xfs_btree_setbuf(cur, level, cbp);
2998 
2999         /*
3000          * Do all this logging at the end so that
3001          * the root is at the right level.
3002          */
3003         xfs_btree_log_block(cur, cbp, XFS_BB_ALL_BITS);
3004         xfs_btree_log_keys(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
3005         xfs_btree_log_ptrs(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
3006 
3007         *logflags |=
3008                 XFS_ILOG_CORE | xfs_ilog_fbroot(cur->bc_private.b.whichfork);
3009         *stat = 1;
3010         return 0;
3011 error0:
3012         return error;
3013 }
3014 
3015 /*
3016  * Allocate a new root block, fill it in.
3017  */
3018 STATIC int                              /* error */
3019 xfs_btree_new_root(
3020         struct xfs_btree_cur    *cur,   /* btree cursor */
3021         int                     *stat)  /* success/failure */
3022 {
3023         struct xfs_btree_block  *block; /* one half of the old root block */
3024         struct xfs_buf          *bp;    /* buffer containing block */
3025         int                     error;  /* error return value */
3026         struct xfs_buf          *lbp;   /* left buffer pointer */
3027         struct xfs_btree_block  *left;  /* left btree block */
3028         struct xfs_buf          *nbp;   /* new (root) buffer */
3029         struct xfs_btree_block  *new;   /* new (root) btree block */
3030         int                     nptr;   /* new value for key index, 1 or 2 */
3031         struct xfs_buf          *rbp;   /* right buffer pointer */
3032         struct xfs_btree_block  *right; /* right btree block */
3033         union xfs_btree_ptr     rptr;
3034         union xfs_btree_ptr     lptr;
3035 
3036         XFS_BTREE_STATS_INC(cur, newroot);
3037 
3038         /* initialise our start point from the cursor */
3039         cur->bc_ops->init_ptr_from_cur(cur, &rptr);
3040 
3041         /* Allocate the new block. If we can't do it, we're toast. Give up. */
3042         error = cur->bc_ops->alloc_block(cur, &rptr, &lptr, stat);
3043         if (error)
3044                 goto error0;
3045         if (*stat == 0)
3046                 goto out0;
3047         XFS_BTREE_STATS_INC(cur, alloc);
3048 
3049         /* Set up the new block. */
3050         error = xfs_btree_get_buf_block(cur, &lptr, &new, &nbp);
3051         if (error)
3052                 goto error0;
3053 
3054         /* Set the root in the holding structure  increasing the level by 1. */
3055         cur->bc_ops->set_root(cur, &lptr, 1);
3056 
3057         /*
3058          * At the previous root level there are now two blocks: the old root,
3059          * and the new block generated when it was split.  We don't know which
3060          * one the cursor is pointing at, so we set up variables "left" and
3061          * "right" for each case.
3062          */
3063         block = xfs_btree_get_block(cur, cur->bc_nlevels - 1, &bp);
3064 
3065 #ifdef DEBUG
3066         error = xfs_btree_check_block(cur, block, cur->bc_nlevels - 1, bp);
3067         if (error)
3068                 goto error0;
3069 #endif
3070 
3071         xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
3072         if (!xfs_btree_ptr_is_null(cur, &rptr)) {
3073                 /* Our block is left, pick up the right block. */
3074                 lbp = bp;
3075                 xfs_btree_buf_to_ptr(cur, lbp, &lptr);
3076                 left = block;
3077                 error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
3078                 if (error)
3079                         goto error0;
3080                 bp = rbp;
3081                 nptr = 1;
3082         } else {
3083                 /* Our block is right, pick up the left block. */
3084                 rbp = bp;
3085                 xfs_btree_buf_to_ptr(cur, rbp, &rptr);
3086                 right = block;
3087                 xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
3088                 error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
3089                 if (error)
3090                         goto error0;
3091                 bp = lbp;
3092                 nptr = 2;
3093         }
3094 
3095         /* Fill in the new block's btree header and log it. */
3096         xfs_btree_init_block_cur(cur, nbp, cur->bc_nlevels, 2);
3097         xfs_btree_log_block(cur, nbp, XFS_BB_ALL_BITS);
3098         ASSERT(!xfs_btree_ptr_is_null(cur, &lptr) &&
3099                         !xfs_btree_ptr_is_null(cur, &rptr));
3100 
3101         /* Fill in the key data in the new root. */
3102         if (xfs_btree_get_level(left) > 0) {
3103                 /*
3104                  * Get the keys for the left block's keys and put them directly
3105                  * in the parent block.  Do the same for the right block.
3106                  */
3107                 xfs_btree_get_node_keys(cur, left,
3108                                 xfs_btree_key_addr(cur, 1, new));
3109                 xfs_btree_get_node_keys(cur, right,
3110                                 xfs_btree_key_addr(cur, 2, new));
3111         } else {
3112                 /*
3113                  * Get the keys for the left block's records and put them
3114                  * directly in the parent block.  Do the same for the right
3115                  * block.
3116                  */
3117                 xfs_btree_get_leaf_keys(cur, left,
3118                         xfs_btree_key_addr(cur, 1, new));
3119                 xfs_btree_get_leaf_keys(cur, right,
3120                         xfs_btree_key_addr(cur, 2, new));
3121         }
3122         xfs_btree_log_keys(cur, nbp, 1, 2);
3123 
3124         /* Fill in the pointer data in the new root. */
3125         xfs_btree_copy_ptrs(cur,
3126                 xfs_btree_ptr_addr(cur, 1, new), &lptr, 1);
3127         xfs_btree_copy_ptrs(cur,
3128                 xfs_btree_ptr_addr(cur, 2, new), &rptr, 1);
3129         xfs_btree_log_ptrs(cur, nbp, 1, 2);
3130 
3131         /* Fix up the cursor. */
3132         xfs_btree_setbuf(cur, cur->bc_nlevels, nbp);
3133         cur->bc_ptrs[cur->bc_nlevels] = nptr;
3134         cur->bc_nlevels++;
3135         *stat = 1;
3136         return 0;
3137 error0:
3138         return error;
3139 out0:
3140         *stat = 0;
3141         return 0;
3142 }
3143 
3144 STATIC int
3145 xfs_btree_make_block_unfull(
3146         struct xfs_btree_cur    *cur,   /* btree cursor */
3147         int                     level,  /* btree level */
3148         int                     numrecs,/* # of recs in block */
3149         int                     *oindex,/* old tree index */
3150         int                     *index, /* new tree index */
3151         union xfs_btree_ptr     *nptr,  /* new btree ptr */
3152         struct xfs_btree_cur    **ncur, /* new btree cursor */
3153         union xfs_btree_key     *key,   /* key of new block */
3154         int                     *stat)
3155 {
3156         int                     error = 0;
3157 
3158         if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
3159             level == cur->bc_nlevels - 1) {
3160                 struct xfs_inode *ip = cur->bc_private.b.ip;
3161 
3162                 if (numrecs < cur->bc_ops->get_dmaxrecs(cur, level)) {
3163                         /* A root block that can be made bigger. */
3164                         xfs_iroot_realloc(ip, 1, cur->bc_private.b.whichfork);
3165                         *stat = 1;
3166                 } else {
3167                         /* A root block that needs replacing */
3168                         int     logflags = 0;
3169 
3170                         error = xfs_btree_new_iroot(cur, &logflags, stat);
3171                         if (error || *stat == 0)
3172                                 return error;
3173 
3174                         xfs_trans_log_inode(cur->bc_tp, ip, logflags);
3175                 }
3176 
3177                 return 0;
3178         }
3179 
3180         /* First, try shifting an entry to the right neighbor. */
3181         error = xfs_btree_rshift(cur, level, stat);
3182         if (error || *stat)
3183                 return error;
3184 
3185         /* Next, try shifting an entry to the left neighbor. */
3186         error = xfs_btree_lshift(cur, level, stat);
3187         if (error)
3188                 return error;
3189 
3190         if (*stat) {
3191                 *oindex = *index = cur->bc_ptrs[level];
3192                 return 0;
3193         }
3194 
3195         /*
3196          * Next, try splitting the current block in half.
3197          *
3198          * If this works we have to re-set our variables because we
3199          * could be in a different block now.
3200          */
3201         error = xfs_btree_split(cur, level, nptr, key, ncur, stat);
3202         if (error || *stat == 0)
3203                 return error;
3204 
3205 
3206         *index = cur->bc_ptrs[level];
3207         return 0;
3208 }
3209 
3210 /*
3211  * Insert one record/level.  Return information to the caller
3212  * allowing the next level up to proceed if necessary.
3213  */
3214 STATIC int
3215 xfs_btree_insrec(
3216         struct xfs_btree_cur    *cur,   /* btree cursor */
3217         int                     level,  /* level to insert record at */
3218         union xfs_btree_ptr     *ptrp,  /* i/o: block number inserted */
3219         union xfs_btree_rec     *rec,   /* record to insert */
3220         union xfs_btree_key     *key,   /* i/o: block key for ptrp */
3221         struct xfs_btree_cur    **curp, /* output: new cursor replacing cur */
3222         int                     *stat)  /* success/failure */
3223 {
3224         struct xfs_btree_block  *block; /* btree block */
3225         struct xfs_buf          *bp;    /* buffer for block */
3226         union xfs_btree_ptr     nptr;   /* new block ptr */
3227         struct xfs_btree_cur    *ncur;  /* new btree cursor */
3228         union xfs_btree_key     nkey;   /* new block key */
3229         union xfs_btree_key     *lkey;
3230         int                     optr;   /* old key/record index */
3231         int                     ptr;    /* key/record index */
3232         int                     numrecs;/* number of records */
3233         int                     error;  /* error return value */
3234         int                     i;
3235         xfs_daddr_t             old_bn;
3236 
3237         ncur = NULL;
3238         lkey = &nkey;
3239 
3240         /*
3241          * If we have an external root pointer, and we've made it to the
3242          * root level, allocate a new root block and we're done.
3243          */
3244         if (!(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
3245             (level >= cur->bc_nlevels)) {
3246                 error = xfs_btree_new_root(cur, stat);
3247                 xfs_btree_set_ptr_null(cur, ptrp);
3248 
3249                 return error;
3250         }
3251 
3252         /* If we're off the left edge, return failure. */
3253         ptr = cur->bc_ptrs[level];
3254         if (ptr == 0) {
3255                 *stat = 0;
3256                 return 0;
3257         }
3258 
3259         optr = ptr;
3260 
3261         XFS_BTREE_STATS_INC(cur, insrec);
3262 
3263         /* Get pointers to the btree buffer and block. */
3264         block = xfs_btree_get_block(cur, level, &bp);
3265         old_bn = bp ? bp->b_bn : XFS_BUF_DADDR_NULL;
3266         numrecs = xfs_btree_get_numrecs(block);
3267 
3268 #ifdef DEBUG
3269         error = xfs_btree_check_block(cur, block, level, bp);
3270         if (error)
3271                 goto error0;
3272 
3273         /* Check that the new entry is being inserted in the right place. */
3274         if (ptr <= numrecs) {
3275                 if (level == 0) {
3276                         ASSERT(cur->bc_ops->recs_inorder(cur, rec,
3277                                 xfs_btree_rec_addr(cur, ptr, block)));
3278                 } else {
3279                         ASSERT(cur->bc_ops->keys_inorder(cur, key,
3280                                 xfs_btree_key_addr(cur, ptr, block)));
3281                 }
3282         }
3283 #endif
3284 
3285         /*
3286          * If the block is full, we can't insert the new entry until we
3287          * make the block un-full.
3288          */
3289         xfs_btree_set_ptr_null(cur, &nptr);
3290         if (numrecs == cur->bc_ops->get_maxrecs(cur, level)) {
3291                 error = xfs_btree_make_block_unfull(cur, level, numrecs,
3292                                         &optr, &ptr, &nptr, &ncur, lkey, stat);
3293                 if (error || *stat == 0)
3294                         goto error0;
3295         }
3296 
3297         /*
3298          * The current block may have changed if the block was
3299          * previously full and we have just made space in it.
3300          */
3301         block = xfs_btree_get_block(cur, level, &bp);
3302         numrecs = xfs_btree_get_numrecs(block);
3303 
3304 #ifdef DEBUG
3305         error = xfs_btree_check_block(cur, block, level, bp);
3306         if (error)
3307                 return error;
3308 #endif
3309 
3310         /*
3311          * At this point we know there's room for our new entry in the block
3312          * we're pointing at.
3313          */
3314         XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr + 1);
3315 
3316         if (level > 0) {
3317                 /* It's a nonleaf. make a hole in the keys and ptrs */
3318                 union xfs_btree_key     *kp;
3319                 union xfs_btree_ptr     *pp;
3320 
3321                 kp = xfs_btree_key_addr(cur, ptr, block);
3322                 pp = xfs_btree_ptr_addr(cur, ptr, block);
3323 
3324                 for (i = numrecs - ptr; i >= 0; i--) {
3325                         error = xfs_btree_debug_check_ptr(cur, pp, i, level);
3326                         if (error)
3327                                 return error;
3328                 }
3329 
3330                 xfs_btree_shift_keys(cur, kp, 1, numrecs - ptr + 1);
3331                 xfs_btree_shift_ptrs(cur, pp, 1, numrecs - ptr + 1);
3332 
3333                 error = xfs_btree_debug_check_ptr(cur, ptrp, 0, level);
3334                 if (error)
3335                         goto error0;
3336 
3337                 /* Now put the new data in, bump numrecs and log it. */
3338                 xfs_btree_copy_keys(cur, kp, key, 1);
3339                 xfs_btree_copy_ptrs(cur, pp, ptrp, 1);
3340                 numrecs++;
3341                 xfs_btree_set_numrecs(block, numrecs);
3342                 xfs_btree_log_ptrs(cur, bp, ptr, numrecs);
3343                 xfs_btree_log_keys(cur, bp, ptr, numrecs);
3344 #ifdef DEBUG
3345                 if (ptr < numrecs) {
3346                         ASSERT(cur->bc_ops->keys_inorder(cur, kp,
3347                                 xfs_btree_key_addr(cur, ptr + 1, block)));
3348                 }
3349 #endif
3350         } else {
3351                 /* It's a leaf. make a hole in the records */
3352                 union xfs_btree_rec             *rp;
3353 
3354                 rp = xfs_btree_rec_addr(cur, ptr, block);
3355 
3356                 xfs_btree_shift_recs(cur, rp, 1, numrecs - ptr + 1);
3357 
3358                 /* Now put the new data in, bump numrecs and log it. */
3359                 xfs_btree_copy_recs(cur, rp, rec, 1);
3360                 xfs_btree_set_numrecs(block, ++numrecs);
3361                 xfs_btree_log_recs(cur, bp, ptr, numrecs);
3362 #ifdef DEBUG
3363                 if (ptr < numrecs) {
3364                         ASSERT(cur->bc_ops->recs_inorder(cur, rp,
3365                                 xfs_btree_rec_addr(cur, ptr + 1, block)));
3366                 }
3367 #endif
3368         }
3369 
3370         /* Log the new number of records in the btree header. */
3371         xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
3372 
3373         /*
3374          * If we just inserted into a new tree block, we have to
3375          * recalculate nkey here because nkey is out of date.
3376          *
3377          * Otherwise we're just updating an existing block (having shoved
3378          * some records into the new tree block), so use the regular key
3379          * update mechanism.
3380          */
3381         if (bp && bp->b_bn != old_bn) {
3382                 xfs_btree_get_keys(cur, block, lkey);
3383         } else if (xfs_btree_needs_key_update(cur, optr)) {
3384                 error = xfs_btree_update_keys(cur, level);
3385                 if (error)
3386                         goto error0;
3387         }
3388 
3389         /*
3390          * If we are tracking the last record in the tree and
3391          * we are at the far right edge of the tree, update it.
3392          */
3393         if (xfs_btree_is_lastrec(cur, block, level)) {
3394                 cur->bc_ops->update_lastrec(cur, block, rec,
3395                                             ptr, LASTREC_INSREC);
3396         }
3397 
3398         /*
3399          * Return the new block number, if any.
3400          * If there is one, give back a record value and a cursor too.
3401          */
3402         *ptrp = nptr;
3403         if (!xfs_btree_ptr_is_null(cur, &nptr)) {
3404                 xfs_btree_copy_keys(cur, key, lkey, 1);
3405                 *curp = ncur;
3406         }
3407 
3408         *stat = 1;
3409         return 0;
3410 
3411 error0:
3412         return error;
3413 }
3414 
3415 /*
3416  * Insert the record at the point referenced by cur.
3417  *
3418  * A multi-level split of the tree on insert will invalidate the original
3419  * cursor.  All callers of this function should assume that the cursor is
3420  * no longer valid and revalidate it.
3421  */
3422 int
3423 xfs_btree_insert(
3424         struct xfs_btree_cur    *cur,
3425         int                     *stat)
3426 {
3427         int                     error;  /* error return value */
3428         int                     i;      /* result value, 0 for failure */
3429         int                     level;  /* current level number in btree */
3430         union xfs_btree_ptr     nptr;   /* new block number (split result) */
3431         struct xfs_btree_cur    *ncur;  /* new cursor (split result) */
3432         struct xfs_btree_cur    *pcur;  /* previous level's cursor */
3433         union xfs_btree_key     bkey;   /* key of block to insert */
3434         union xfs_btree_key     *key;
3435         union xfs_btree_rec     rec;    /* record to insert */
3436 
3437         level = 0;
3438         ncur = NULL;
3439         pcur = cur;
3440         key = &bkey;
3441 
3442         xfs_btree_set_ptr_null(cur, &nptr);
3443 
3444         /* Make a key out of the record data to be inserted, and save it. */
3445         cur->bc_ops->init_rec_from_cur(cur, &rec);
3446         cur->bc_ops->init_key_from_rec(key, &rec);
3447 
3448         /*
3449          * Loop going up the tree, starting at the leaf level.
3450          * Stop when we don't get a split block, that must mean that
3451          * the insert is finished with this level.
3452          */
3453         do {
3454                 /*
3455                  * Insert nrec/nptr into this level of the tree.
3456                  * Note if we fail, nptr will be null.
3457                  */
3458                 error = xfs_btree_insrec(pcur, level, &nptr, &rec, key,
3459                                 &ncur, &i);
3460                 if (error) {
3461                         if (pcur != cur)
3462                                 xfs_btree_del_cursor(pcur, XFS_BTREE_ERROR);
3463                         goto error0;
3464                 }
3465 
3466                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
3467                 level++;
3468 
3469                 /*
3470                  * See if the cursor we just used is trash.
3471                  * Can't trash the caller's cursor, but otherwise we should
3472                  * if ncur is a new cursor or we're about to be done.
3473                  */
3474                 if (pcur != cur &&
3475                     (ncur || xfs_btree_ptr_is_null(cur, &nptr))) {
3476                         /* Save the state from the cursor before we trash it */
3477                         if (cur->bc_ops->update_cursor)
3478                                 cur->bc_ops->update_cursor(pcur, cur);
3479                         cur->bc_nlevels = pcur->bc_nlevels;
3480                         xfs_btree_del_cursor(pcur, XFS_BTREE_NOERROR);
3481                 }
3482                 /* If we got a new cursor, switch to it. */
3483                 if (ncur) {
3484                         pcur = ncur;
3485                         ncur = NULL;
3486                 }
3487         } while (!xfs_btree_ptr_is_null(cur, &nptr));
3488 
3489         *stat = i;
3490         return 0;
3491 error0:
3492         return error;
3493 }
3494 
3495 /*
3496  * Try to merge a non-leaf block back into the inode root.
3497  *
3498  * Note: the killroot names comes from the fact that we're effectively
3499  * killing the old root block.  But because we can't just delete the
3500  * inode we have to copy the single block it was pointing to into the
3501  * inode.
3502  */
3503 STATIC int
3504 xfs_btree_kill_iroot(
3505         struct xfs_btree_cur    *cur)
3506 {
3507         int                     whichfork = cur->bc_private.b.whichfork;
3508         struct xfs_inode        *ip = cur->bc_private.b.ip;
3509         struct xfs_ifork        *ifp = XFS_IFORK_PTR(ip, whichfork);
3510         struct xfs_btree_block  *block;
3511         struct xfs_btree_block  *cblock;
3512         union xfs_btree_key     *kp;
3513         union xfs_btree_key     *ckp;
3514         union xfs_btree_ptr     *pp;
3515         union xfs_btree_ptr     *cpp;
3516         struct xfs_buf          *cbp;
3517         int                     level;
3518         int                     index;
3519         int                     numrecs;
3520         int                     error;
3521 #ifdef DEBUG
3522         union xfs_btree_ptr     ptr;
3523 #endif
3524         int                     i;
3525 
3526         ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
3527         ASSERT(cur->bc_nlevels > 1);
3528 
3529         /*
3530          * Don't deal with the root block needs to be a leaf case.
3531          * We're just going to turn the thing back into extents anyway.
3532          */
3533         level = cur->bc_nlevels - 1;
3534         if (level == 1)
3535                 goto out0;
3536 
3537         /*
3538          * Give up if the root has multiple children.
3539          */
3540         block = xfs_btree_get_iroot(cur);
3541         if (xfs_btree_get_numrecs(block) != 1)
3542                 goto out0;
3543 
3544         cblock = xfs_btree_get_block(cur, level - 1, &cbp);
3545         numrecs = xfs_btree_get_numrecs(cblock);
3546 
3547         /*
3548          * Only do this if the next level will fit.
3549          * Then the data must be copied up to the inode,
3550          * instead of freeing the root you free the next level.
3551          */
3552         if (numrecs > cur->bc_ops->get_dmaxrecs(cur, level))
3553                 goto out0;
3554 
3555         XFS_BTREE_STATS_INC(cur, killroot);
3556 
3557 #ifdef DEBUG
3558         xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
3559         ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
3560         xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
3561         ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
3562 #endif
3563 
3564         index = numrecs - cur->bc_ops->get_maxrecs(cur, level);
3565         if (index) {
3566                 xfs_iroot_realloc(cur->bc_private.b.ip, index,
3567                                   cur->bc_private.b.whichfork);
3568                 block = ifp->if_broot;
3569         }
3570 
3571         be16_add_cpu(&block->bb_numrecs, index);
3572         ASSERT(block->bb_numrecs == cblock->bb_numrecs);
3573 
3574         kp = xfs_btree_key_addr(cur, 1, block);
3575         ckp = xfs_btree_key_addr(cur, 1, cblock);
3576         xfs_btree_copy_keys(cur, kp, ckp, numrecs);
3577 
3578         pp = xfs_btree_ptr_addr(cur, 1, block);
3579         cpp = xfs_btree_ptr_addr(cur, 1, cblock);
3580 
3581         for (i = 0; i < numrecs; i++) {
3582                 error = xfs_btree_debug_check_ptr(cur, cpp, i, level - 1);
3583                 if (error)
3584                         return error;
3585         }
3586 
3587         xfs_btree_copy_ptrs(cur, pp, cpp, numrecs);
3588 
3589         error = xfs_btree_free_block(cur, cbp);
3590         if (error)
3591                 return error;
3592 
3593         cur->bc_bufs[level - 1] = NULL;
3594         be16_add_cpu(&block->bb_level, -1);
3595         xfs_trans_log_inode(cur->bc_tp, ip,
3596                 XFS_ILOG_CORE | xfs_ilog_fbroot(cur->bc_private.b.whichfork));
3597         cur->bc_nlevels--;
3598 out0:
3599         return 0;
3600 }
3601 
3602 /*
3603  * Kill the current root node, and replace it with it's only child node.
3604  */
3605 STATIC int
3606 xfs_btree_kill_root(
3607         struct xfs_btree_cur    *cur,
3608         struct xfs_buf          *bp,
3609         int                     level,
3610         union xfs_btree_ptr     *newroot)
3611 {
3612         int                     error;
3613 
3614         XFS_BTREE_STATS_INC(cur, killroot);
3615 
3616         /*
3617          * Update the root pointer, decreasing the level by 1 and then
3618          * free the old root.
3619          */
3620         cur->bc_ops->set_root(cur, newroot, -1);
3621 
3622         error = xfs_btree_free_block(cur, bp);
3623         if (error)
3624                 return error;
3625 
3626         cur->bc_bufs[level] = NULL;
3627         cur->bc_ra[level] = 0;
3628         cur->bc_nlevels--;
3629 
3630         return 0;
3631 }
3632 
3633 STATIC int
3634 xfs_btree_dec_cursor(
3635         struct xfs_btree_cur    *cur,
3636         int                     level,
3637         int                     *stat)
3638 {
3639         int                     error;
3640         int                     i;
3641 
3642         if (level > 0) {
3643                 error = xfs_btree_decrement(cur, level, &i);
3644                 if (error)
3645                         return error;
3646         }
3647 
3648         *stat = 1;
3649         return 0;
3650 }
3651 
3652 /*
3653  * Single level of the btree record deletion routine.
3654  * Delete record pointed to by cur/level.
3655  * Remove the record from its block then rebalance the tree.
3656  * Return 0 for error, 1 for done, 2 to go on to the next level.
3657  */
3658 STATIC int                                      /* error */
3659 xfs_btree_delrec(
3660         struct xfs_btree_cur    *cur,           /* btree cursor */
3661         int                     level,          /* level removing record from */
3662         int                     *stat)          /* fail/done/go-on */
3663 {
3664         struct xfs_btree_block  *block;         /* btree block */
3665         union xfs_btree_ptr     cptr;           /* current block ptr */
3666         struct xfs_buf          *bp;            /* buffer for block */
3667         int                     error;          /* error return value */
3668         int                     i;              /* loop counter */
3669         union xfs_btree_ptr     lptr;           /* left sibling block ptr */
3670         struct xfs_buf          *lbp;           /* left buffer pointer */
3671         struct xfs_btree_block  *left;          /* left btree block */
3672         int                     lrecs = 0;      /* left record count */
3673         int                     ptr;            /* key/record index */
3674         union xfs_btree_ptr     rptr;           /* right sibling block ptr */
3675         struct xfs_buf          *rbp;           /* right buffer pointer */
3676         struct xfs_btree_block  *right;         /* right btree block */
3677         struct xfs_btree_block  *rrblock;       /* right-right btree block */
3678         struct xfs_buf          *rrbp;          /* right-right buffer pointer */
3679         int                     rrecs = 0;      /* right record count */
3680         struct xfs_btree_cur    *tcur;          /* temporary btree cursor */
3681         int                     numrecs;        /* temporary numrec count */
3682 
3683         tcur = NULL;
3684 
3685         /* Get the index of the entry being deleted, check for nothing there. */
3686         ptr = cur->bc_ptrs[level];
3687         if (ptr == 0) {
3688                 *stat = 0;
3689                 return 0;
3690         }
3691 
3692         /* Get the buffer & block containing the record or key/ptr. */
3693         block = xfs_btree_get_block(cur, level, &bp);
3694         numrecs = xfs_btree_get_numrecs(block);
3695 
3696 #ifdef DEBUG
3697         error = xfs_btree_check_block(cur, block, level, bp);
3698         if (error)
3699                 goto error0;
3700 #endif
3701 
3702         /* Fail if we're off the end of the block. */
3703         if (ptr > numrecs) {
3704                 *stat = 0;
3705                 return 0;
3706         }
3707 
3708         XFS_BTREE_STATS_INC(cur, delrec);
3709         XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr);
3710 
3711         /* Excise the entries being deleted. */
3712         if (level > 0) {
3713                 /* It's a nonleaf. operate on keys and ptrs */
3714                 union xfs_btree_key     *lkp;
3715                 union xfs_btree_ptr     *lpp;
3716 
3717                 lkp = xfs_btree_key_addr(cur, ptr + 1, block);
3718                 lpp = xfs_btree_ptr_addr(cur, ptr + 1, block);
3719 
3720                 for (i = 0; i < numrecs - ptr; i++) {
3721                         error = xfs_btree_debug_check_ptr(cur, lpp, i, level);
3722                         if (error)
3723                                 goto error0;
3724                 }
3725 
3726                 if (ptr < numrecs) {
3727                         xfs_btree_shift_keys(cur, lkp, -1, numrecs - ptr);
3728                         xfs_btree_shift_ptrs(cur, lpp, -1, numrecs - ptr);
3729                         xfs_btree_log_keys(cur, bp, ptr, numrecs - 1);
3730                         xfs_btree_log_ptrs(cur, bp, ptr, numrecs - 1);
3731                 }
3732         } else {
3733                 /* It's a leaf. operate on records */
3734                 if (ptr < numrecs) {
3735                         xfs_btree_shift_recs(cur,
3736                                 xfs_btree_rec_addr(cur, ptr + 1, block),
3737                                 -1, numrecs - ptr);
3738                         xfs_btree_log_recs(cur, bp, ptr, numrecs - 1);
3739                 }
3740         }
3741 
3742         /*
3743          * Decrement and log the number of entries in the block.
3744          */
3745         xfs_btree_set_numrecs(block, --numrecs);
3746         xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
3747 
3748         /*
3749          * If we are tracking the last record in the tree and
3750          * we are at the far right edge of the tree, update it.
3751          */
3752         if (xfs_btree_is_lastrec(cur, block, level)) {
3753                 cur->bc_ops->update_lastrec(cur, block, NULL,
3754                                             ptr, LASTREC_DELREC);
3755         }
3756 
3757         /*
3758          * We're at the root level.  First, shrink the root block in-memory.
3759          * Try to get rid of the next level down.  If we can't then there's
3760          * nothing left to do.
3761          */
3762         if (level == cur->bc_nlevels - 1) {
3763                 if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
3764                         xfs_iroot_realloc(cur->bc_private.b.ip, -1,
3765                                           cur->bc_private.b.whichfork);
3766 
3767                         error = xfs_btree_kill_iroot(cur);
3768                         if (error)
3769                                 goto error0;
3770 
3771                         error = xfs_btree_dec_cursor(cur, level, stat);
3772                         if (error)
3773                                 goto error0;
3774                         *stat = 1;
3775                         return 0;
3776                 }
3777 
3778                 /*
3779                  * If this is the root level, and there's only one entry left,
3780                  * and it's NOT the leaf level, then we can get rid of this
3781                  * level.
3782                  */
3783                 if (numrecs == 1 && level > 0) {
3784                         union xfs_btree_ptr     *pp;
3785                         /*
3786                          * pp is still set to the first pointer in the block.
3787                          * Make it the new root of the btree.
3788                          */
3789                         pp = xfs_btree_ptr_addr(cur, 1, block);
3790                         error = xfs_btree_kill_root(cur, bp, level, pp);
3791                         if (error)
3792                                 goto error0;
3793                 } else if (level > 0) {
3794                         error = xfs_btree_dec_cursor(cur, level, stat);
3795                         if (error)
3796                                 goto error0;
3797                 }
3798                 *stat = 1;
3799                 return 0;
3800         }
3801 
3802         /*
3803          * If we deleted the leftmost entry in the block, update the
3804          * key values above us in the tree.
3805          */
3806         if (xfs_btree_needs_key_update(cur, ptr)) {
3807                 error = xfs_btree_update_keys(cur, level);
3808                 if (error)
3809                         goto error0;
3810         }
3811 
3812         /*
3813          * If the number of records remaining in the block is at least
3814          * the minimum, we're done.
3815          */
3816         if (numrecs >= cur->bc_ops->get_minrecs(cur, level)) {
3817                 error = xfs_btree_dec_cursor(cur, level, stat);
3818                 if (error)
3819                         goto error0;
3820                 return 0;
3821         }
3822 
3823         /*
3824          * Otherwise, we have to move some records around to keep the
3825          * tree balanced.  Look at the left and right sibling blocks to
3826          * see if we can re-balance by moving only one record.
3827          */
3828         xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
3829         xfs_btree_get_sibling(cur, block, &lptr, XFS_BB_LEFTSIB);
3830 
3831         if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
3832                 /*
3833                  * One child of root, need to get a chance to copy its contents
3834                  * into the root and delete it. Can't go up to next level,
3835                  * there's nothing to delete there.
3836                  */
3837                 if (xfs_btree_ptr_is_null(cur, &rptr) &&
3838                     xfs_btree_ptr_is_null(cur, &lptr) &&
3839                     level == cur->bc_nlevels - 2) {
3840                         error = xfs_btree_kill_iroot(cur);
3841                         if (!error)
3842                                 error = xfs_btree_dec_cursor(cur, level, stat);
3843                         if (error)
3844                                 goto error0;
3845                         return 0;
3846                 }
3847         }
3848 
3849         ASSERT(!xfs_btree_ptr_is_null(cur, &rptr) ||
3850                !xfs_btree_ptr_is_null(cur, &lptr));
3851 
3852         /*
3853          * Duplicate the cursor so our btree manipulations here won't
3854          * disrupt the next level up.
3855          */
3856         error = xfs_btree_dup_cursor(cur, &tcur);
3857         if (error)
3858                 goto error0;
3859 
3860         /*
3861          * If there's a right sibling, see if it's ok to shift an entry
3862          * out of it.
3863          */
3864         if (!xfs_btree_ptr_is_null(cur, &rptr)) {
3865                 /*
3866                  * Move the temp cursor to the last entry in the next block.
3867                  * Actually any entry but the first would suffice.
3868                  */
3869                 i = xfs_btree_lastrec(tcur, level);
3870                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
3871 
3872                 error = xfs_btree_increment(tcur, level, &i);
3873                 if (error)
3874                         goto error0;
3875                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
3876 
3877                 i = xfs_btree_lastrec(tcur, level);
3878                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
3879 
3880                 /* Grab a pointer to the block. */
3881                 right = xfs_btree_get_block(tcur, level, &rbp);
3882 #ifdef DEBUG
3883                 error = xfs_btree_check_block(tcur, right, level, rbp);
3884                 if (error)
3885                         goto error0;
3886 #endif
3887                 /* Grab the current block number, for future use. */
3888                 xfs_btree_get_sibling(tcur, right, &cptr, XFS_BB_LEFTSIB);
3889 
3890                 /*
3891                  * If right block is full enough so that removing one entry
3892                  * won't make it too empty, and left-shifting an entry out
3893                  * of right to us works, we're done.
3894                  */
3895                 if (xfs_btree_get_numrecs(right) - 1 >=
3896                     cur->bc_ops->get_minrecs(tcur, level)) {
3897                         error = xfs_btree_lshift(tcur, level, &i);
3898                         if (error)
3899                                 goto error0;
3900                         if (i) {
3901                                 ASSERT(xfs_btree_get_numrecs(block) >=
3902                                        cur->bc_ops->get_minrecs(tcur, level));
3903 
3904                                 xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
3905                                 tcur = NULL;
3906 
3907                                 error = xfs_btree_dec_cursor(cur, level, stat);
3908                                 if (error)
3909                                         goto error0;
3910                                 return 0;
3911                         }
3912                 }
3913 
3914                 /*
3915                  * Otherwise, grab the number of records in right for
3916                  * future reference, and fix up the temp cursor to point
3917                  * to our block again (last record).
3918                  */
3919                 rrecs = xfs_btree_get_numrecs(right);
3920                 if (!xfs_btree_ptr_is_null(cur, &lptr)) {
3921                         i = xfs_btree_firstrec(tcur, level);
3922                         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
3923 
3924                         error = xfs_btree_decrement(tcur, level, &i);
3925                         if (error)
3926                                 goto error0;
3927                         XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
3928                 }
3929         }
3930 
3931         /*
3932          * If there's a left sibling, see if it's ok to shift an entry
3933          * out of it.
3934          */
3935         if (!xfs_btree_ptr_is_null(cur, &lptr)) {
3936                 /*
3937                  * Move the temp cursor to the first entry in the
3938                  * previous block.
3939                  */
3940                 i = xfs_btree_firstrec(tcur, level);
3941                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
3942 
3943                 error = xfs_btree_decrement(tcur, level, &i);
3944                 if (error)
3945                         goto error0;
3946                 i = xfs_btree_firstrec(tcur, level);
3947                 XFS_WANT_CORRUPTED_GOTO(cur->bc_mp, i == 1, error0);
3948 
3949                 /* Grab a pointer to the block. */
3950                 left = xfs_btree_get_block(tcur, level, &lbp);
3951 #ifdef DEBUG
3952                 error = xfs_btree_check_block(cur, left, level, lbp);
3953                 if (error)
3954                         goto error0;
3955 #endif
3956                 /* Grab the current block number, for future use. */
3957                 xfs_btree_get_sibling(tcur, left, &cptr, XFS_BB_RIGHTSIB);
3958 
3959                 /*
3960                  * If left block is full enough so that removing one entry
3961                  * won't make it too empty, and right-shifting an entry out
3962                  * of left to us works, we're done.
3963                  */
3964                 if (xfs_btree_get_numrecs(left) - 1 >=
3965                     cur->bc_ops->get_minrecs(tcur, level)) {
3966                         error = xfs_btree_rshift(tcur, level, &i);
3967                         if (error)
3968                                 goto error0;
3969                         if (i) {
3970                                 ASSERT(xfs_btree_get_numrecs(block) >=
3971                                        cur->bc_ops->get_minrecs(tcur, level));
3972                                 xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
3973                                 tcur = NULL;
3974                                 if (level == 0)
3975                                         cur->bc_ptrs[0]++;
3976 
3977                                 *stat = 1;
3978                                 return 0;
3979                         }
3980                 }
3981 
3982                 /*
3983                  * Otherwise, grab the number of records in right for
3984                  * future reference.
3985                  */
3986                 lrecs = xfs_btree_get_numrecs(left);
3987         }
3988 
3989         /* Delete the temp cursor, we're done with it. */
3990         xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
3991         tcur = NULL;
3992 
3993         /* If here, we need to do a join to keep the tree balanced. */
3994         ASSERT(!xfs_btree_ptr_is_null(cur, &cptr));
3995 
3996         if (!xfs_btree_ptr_is_null(cur, &lptr) &&
3997             lrecs + xfs_btree_get_numrecs(block) <=
3998                         cur->bc_ops->get_maxrecs(cur, level)) {
3999                 /*
4000                  * Set "right" to be the starting block,
4001                  * "left" to be the left neighbor.
4002                  */
4003                 rptr = cptr;
4004                 right = block;
4005                 rbp = bp;
4006                 error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
4007                 if (error)
4008                         goto error0;
4009 
4010         /*
4011          * If that won't work, see if we can join with the right neighbor block.
4012          */
4013         } else if (!xfs_btree_ptr_is_null(cur, &rptr) &&
4014                    rrecs + xfs_btree_get_numrecs(block) <=
4015                         cur->bc_ops->get_maxrecs(cur, level)) {
4016                 /*
4017                  * Set "left" to be the starting block,
4018                  * "right" to be the right neighbor.
4019                  */
4020                 lptr = cptr;
4021                 left = block;
4022                 lbp = bp;
4023                 error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
4024                 if (error)
4025                         goto error0;
4026 
4027         /*
4028          * Otherwise, we can't fix the imbalance.
4029          * Just return.  This is probably a logic error, but it's not fatal.
4030          */
4031         } else {
4032                 error = xfs_btree_dec_cursor(cur, level, stat);
4033                 if (error)
4034                         goto error0;
4035                 return 0;
4036         }
4037 
4038         rrecs = xfs_btree_get_numrecs(right);
4039         lrecs = xfs_btree_get_numrecs(left);
4040 
4041         /*
4042          * We're now going to join "left" and "right" by moving all the stuff
4043          * in "right" to "left" and deleting "right".
4044          */
4045         XFS_BTREE_STATS_ADD(cur, moves, rrecs);
4046         if (level > 0) {
4047                 /* It's a non-leaf.  Move keys and pointers. */
4048                 union xfs_btree_key     *lkp;   /* left btree key */
4049                 union xfs_btree_ptr     *lpp;   /* left address pointer */
4050                 union xfs_btree_key     *rkp;   /* right btree key */
4051                 union xfs_btree_ptr     *rpp;   /* right address pointer */
4052 
4053                 lkp = xfs_btree_key_addr(cur, lrecs + 1, left);
4054                 lpp = xfs_btree_ptr_addr(cur, lrecs + 1, left);
4055                 rkp = xfs_btree_key_addr(cur, 1, right);
4056                 rpp = xfs_btree_ptr_addr(cur, 1, right);
4057 
4058                 for (i = 1; i < rrecs; i++) {
4059                         error = xfs_btree_debug_check_ptr(cur, rpp, i, level);
4060                         if (error)
4061                                 goto error0;
4062                 }
4063 
4064                 xfs_btree_copy_keys(cur, lkp, rkp, rrecs);
4065                 xfs_btree_copy_ptrs(cur, lpp, rpp, rrecs);
4066 
4067                 xfs_btree_log_keys(cur, lbp, lrecs + 1, lrecs + rrecs);
4068                 xfs_btree_log_ptrs(cur, lbp, lrecs + 1, lrecs + rrecs);
4069         } else {
4070                 /* It's a leaf.  Move records.  */
4071                 union xfs_btree_rec     *lrp;   /* left record pointer */
4072                 union xfs_btree_rec     *rrp;   /* right record pointer */
4073 
4074                 lrp = xfs_btree_rec_addr(cur, lrecs + 1, left);
4075                 rrp = xfs_btree_rec_addr(cur, 1, right);
4076 
4077                 xfs_btree_copy_recs(cur, lrp, rrp, rrecs);
4078                 xfs_btree_log_recs(cur, lbp, lrecs + 1, lrecs + rrecs);
4079         }
4080 
4081         XFS_BTREE_STATS_INC(cur, join);
4082 
4083         /*
4084          * Fix up the number of records and right block pointer in the
4085          * surviving block, and log it.
4086          */
4087         xfs_btree_set_numrecs(left, lrecs + rrecs);
4088         xfs_btree_get_sibling(cur, right, &cptr, XFS_BB_RIGHTSIB),
4089         xfs_btree_set_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
4090         xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
4091 
4092         /* If there is a right sibling, point it to the remaining block. */
4093         xfs_btree_get_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
4094         if (!xfs_btree_ptr_is_null(cur, &cptr)) {
4095                 error = xfs_btree_read_buf_block(cur, &cptr, 0, &rrblock, &rrbp);
4096                 if (error)
4097                         goto error0;
4098                 xfs_btree_set_sibling(cur, rrblock, &lptr, XFS_BB_LEFTSIB);
4099                 xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
4100         }
4101 
4102         /* Free the deleted block. */
4103         error = xfs_btree_free_block(cur, rbp);
4104         if (error)
4105                 goto error0;
4106 
4107         /*
4108          * If we joined with the left neighbor, set the buffer in the
4109          * cursor to the left block, and fix up the index.
4110          */
4111         if (bp != lbp) {
4112                 cur->bc_bufs[level] = lbp;
4113                 cur->bc_ptrs[level] += lrecs;
4114                 cur->bc_ra[level] = 0;
4115         }
4116         /*
4117          * If we joined with the right neighbor and there's a level above
4118          * us, increment the cursor at that level.
4119          */
4120         else if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) ||
4121                    (level + 1 < cur->bc_nlevels)) {
4122                 error = xfs_btree_increment(cur, level + 1, &i);
4123                 if (error)
4124                         goto error0;
4125         }
4126 
4127         /*
4128          * Readjust the ptr at this level if it's not a leaf, since it's
4129          * still pointing at the deletion point, which makes the cursor
4130          * inconsistent.  If this makes the ptr 0, the caller fixes it up.
4131          * We can't use decrement because it would change the next level up.
4132          */
4133         if (level > 0)
4134                 cur->bc_ptrs[level]--;
4135 
4136         /*
4137          * We combined blocks, so we have to update the parent keys if the
4138          * btree supports overlapped intervals.  However, bc_ptrs[level + 1]
4139          * points to the old block so that the caller knows which record to
4140          * delete.  Therefore, the caller must be savvy enough to call updkeys
4141          * for us if we return stat == 2.  The other exit points from this
4142          * function don't require deletions further up the tree, so they can
4143          * call updkeys directly.
4144          */
4145 
4146         /* Return value means the next level up has something to do. */
4147         *stat = 2;
4148         return 0;
4149 
4150 error0:
4151         if (tcur)
4152                 xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
4153         return error;
4154 }
4155 
4156 /*
4157  * Delete the record pointed to by cur.
4158  * The cursor refers to the place where the record was (could be inserted)
4159  * when the operation returns.
4160  */
4161 int                                     /* error */
4162 xfs_btree_delete(
4163         struct xfs_btree_cur    *cur,
4164         int                     *stat)  /* success/failure */
4165 {
4166         int                     error;  /* error return value */
4167         int                     level;
4168         int                     i;
4169         bool                    joined = false;
4170 
4171         /*
4172          * Go up the tree, starting at leaf level.
4173          *
4174          * If 2 is returned then a join was done; go to the next level.
4175          * Otherwise we are done.
4176          */
4177         for (level = 0, i = 2; i == 2; level++) {
4178                 error = xfs_btree_delrec(cur, level, &i);
4179                 if (error)
4180                         goto error0;
4181                 if (i == 2)
4182                         joined = true;
4183         }
4184 
4185         /*
4186          * If we combined blocks as part of deleting the record, delrec won't
4187          * have updated the parent high keys so we have to do that here.
4188          */
4189         if (joined && (cur->bc_flags & XFS_BTREE_OVERLAPPING)) {
4190                 error = xfs_btree_updkeys_force(cur, 0);
4191                 if (error)
4192                         goto error0;
4193         }
4194 
4195         if (i == 0) {
4196                 for (level = 1; level < cur->bc_nlevels; level++) {
4197                         if (cur->bc_ptrs[level] == 0) {
4198                                 error = xfs_btree_decrement(cur, level, &i);
4199                                 if (error)
4200                                         goto error0;
4201                                 break;
4202                         }
4203                 }
4204         }
4205 
4206         *stat = i;
4207         return 0;
4208 error0:
4209         return error;
4210 }
4211 
4212 /*
4213  * Get the data from the pointed-to record.
4214  */
4215 int                                     /* error */
4216 xfs_btree_get_rec(
4217         struct xfs_btree_cur    *cur,   /* btree cursor */
4218         union xfs_btree_rec     **recp, /* output: btree record */
4219         int                     *stat)  /* output: success/failure */
4220 {
4221         struct xfs_btree_block  *block; /* btree block */
4222         struct xfs_buf          *bp;    /* buffer pointer */
4223         int                     ptr;    /* record number */
4224 #ifdef DEBUG
4225         int                     error;  /* error return value */
4226 #endif
4227 
4228         ptr = cur->bc_ptrs[0];
4229         block = xfs_btree_get_block(cur, 0, &bp);
4230 
4231 #ifdef DEBUG
4232         error = xfs_btree_check_block(cur, block, 0, bp);
4233         if (error)
4234                 return error;
4235 #endif
4236 
4237         /*
4238          * Off the right end or left end, return failure.
4239          */
4240         if (ptr > xfs_btree_get_numrecs(block) || ptr <= 0) {
4241                 *stat = 0;
4242                 return 0;
4243         }
4244 
4245         /*
4246          * Point to the record and extract its data.
4247          */
4248         *recp = xfs_btree_rec_addr(cur, ptr, block);
4249         *stat = 1;
4250         return 0;
4251 }
4252 
4253 /* Visit a block in a btree. */
4254 STATIC int
4255 xfs_btree_visit_block(
4256         struct xfs_btree_cur            *cur,
4257         int                             level,
4258         xfs_btree_visit_blocks_fn       fn,
4259         void                            *data)
4260 {
4261         struct xfs_btree_block          *block;
4262         struct xfs_buf                  *bp;
4263         union xfs_btree_ptr             rptr;
4264         int                             error;
4265 
4266         /* do right sibling readahead */
4267         xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
4268         block = xfs_btree_get_block(cur, level, &bp);
4269 
4270         /* process the block */
4271         error = fn(cur, level, data);
4272         if (error)
4273                 return error;
4274 
4275         /* now read rh sibling block for next iteration */
4276         xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
4277         if (xfs_btree_ptr_is_null(cur, &rptr))
4278                 return -ENOENT;
4279 
4280         return xfs_btree_lookup_get_block(cur, level, &rptr, &block);
4281 }
4282 
4283 
4284 /* Visit every block in a btree. */
4285 int
4286 xfs_btree_visit_blocks(
4287         struct xfs_btree_cur            *cur,
4288         xfs_btree_visit_blocks_fn       fn,
4289         void                            *data)
4290 {
4291         union xfs_btree_ptr             lptr;
4292         int                             level;
4293         struct xfs_btree_block          *block = NULL;
4294         int                             error = 0;
4295 
4296         cur->bc_ops->init_ptr_from_cur(cur, &lptr);
4297 
4298         /* for each level */
4299         for (level = cur->bc_nlevels - 1; level >= 0; level--) {
4300                 /* grab the left hand block */
4301                 error = xfs_btree_lookup_get_block(cur, level, &lptr, &block);
4302                 if (error)
4303                         return error;
4304 
4305                 /* readahead the left most block for the next level down */
4306                 if (level > 0) {
4307                         union xfs_btree_ptr     *ptr;
4308 
4309                         ptr = xfs_btree_ptr_addr(cur, 1, block);
4310                         xfs_btree_readahead_ptr(cur, ptr, 1);
4311 
4312                         /* save for the next iteration of the loop */
4313                         xfs_btree_copy_ptrs(cur, &lptr, ptr, 1);
4314                 }
4315 
4316                 /* for each buffer in the level */
4317                 do {
4318                         error = xfs_btree_visit_block(cur, level, fn, data);
4319                 } while (!error);
4320 
4321                 if (error != -ENOENT)
4322                         return error;
4323         }
4324 
4325         return 0;
4326 }
4327 
4328 /*
4329  * Change the owner of a btree.
4330  *
4331  * The mechanism we use here is ordered buffer logging. Because we don't know
4332  * how many buffers were are going to need to modify, we don't really want to
4333  * have to make transaction reservations for the worst case of every buffer in a
4334  * full size btree as that may be more space that we can fit in the log....
4335  *
4336  * We do the btree walk in the most optimal manner possible - we have sibling
4337  * pointers so we can just walk all the blocks on each level from left to right
4338  * in a single pass, and then move to the next level and do the same. We can
4339  * also do readahead on the sibling pointers to get IO moving more quickly,
4340  * though for slow disks this is unlikely to make much difference to performance
4341  * as the amount of CPU work we have to do before moving to the next block is
4342  * relatively small.
4343  *
4344  * For each btree block that we load, modify the owner appropriately, set the
4345  * buffer as an ordered buffer and log it appropriately. We need to ensure that
4346  * we mark the region we change dirty so that if the buffer is relogged in
4347  * a subsequent transaction the changes we make here as an ordered buffer are
4348  * correctly relogged in that transaction.  If we are in recovery context, then
4349  * just queue the modified buffer as delayed write buffer so the transaction
4350  * recovery completion writes the changes to disk.
4351  */
4352 struct xfs_btree_block_change_owner_info {
4353         uint64_t                new_owner;
4354         struct list_head        *buffer_list;
4355 };
4356 
4357 static int
4358 xfs_btree_block_change_owner(
4359         struct xfs_btree_cur    *cur,
4360         int                     level,
4361         void                    *data)
4362 {
4363         struct xfs_btree_block_change_owner_info        *bbcoi = data;
4364         struct xfs_btree_block  *block;
4365         struct xfs_buf          *bp;
4366 
4367         /* modify the owner */
4368         block = xfs_btree_get_block(cur, level, &bp);
4369         if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
4370                 if (block->bb_u.l.bb_owner == cpu_to_be64(bbcoi->new_owner))
4371                         return 0;
4372                 block->bb_u.l.bb_owner = cpu_to_be64(bbcoi->new_owner);
4373         } else {
4374                 if (block->bb_u.s.bb_owner == cpu_to_be32(bbcoi->new_owner))
4375                         return 0;
4376                 block->bb_u.s.bb_owner = cpu_to_be32(bbcoi->new_owner);
4377         }
4378 
4379         /*
4380          * If the block is a root block hosted in an inode, we might not have a
4381          * buffer pointer here and we shouldn't attempt to log the change as the
4382          * information is already held in the inode and discarded when the root
4383          * block is formatted into the on-disk inode fork. We still change it,
4384          * though, so everything is consistent in memory.
4385          */
4386         if (!bp) {
4387                 ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
4388                 ASSERT(level == cur->bc_nlevels - 1);
4389                 return 0;
4390         }
4391 
4392         if (cur->bc_tp) {
4393                 if (!xfs_trans_ordered_buf(cur->bc_tp, bp)) {
4394                         xfs_btree_log_block(cur, bp, XFS_BB_OWNER);
4395                         return -EAGAIN;
4396                 }
4397         } else {
4398                 xfs_buf_delwri_queue(bp, bbcoi->buffer_list);
4399         }
4400 
4401         return 0;
4402 }
4403 
4404 int
4405 xfs_btree_change_owner(
4406         struct xfs_btree_cur    *cur,
4407         uint64_t                new_owner,
4408         struct list_head        *buffer_list)
4409 {
4410         struct xfs_btree_block_change_owner_info        bbcoi;
4411 
4412         bbcoi.new_owner = new_owner;
4413         bbcoi.buffer_list = buffer_list;
4414 
4415         return xfs_btree_visit_blocks(cur, xfs_btree_block_change_owner,
4416                         &bbcoi);
4417 }
4418 
4419 /* Verify the v5 fields of a long-format btree block. */
4420 xfs_failaddr_t
4421 xfs_btree_lblock_v5hdr_verify(
4422         struct xfs_buf          *bp,
4423         uint64_t                owner)
4424 {
4425         struct xfs_mount        *mp = bp->b_mount;
4426         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
4427 
4428         if (!xfs_sb_version_hascrc(&mp->m_sb))
4429                 return __this_address;
4430         if (!uuid_equal(&block->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid))
4431                 return __this_address;
4432         if (block->bb_u.l.bb_blkno != cpu_to_be64(bp->b_bn))
4433                 return __this_address;
4434         if (owner != XFS_RMAP_OWN_UNKNOWN &&
4435             be64_to_cpu(block->bb_u.l.bb_owner) != owner)
4436                 return __this_address;
4437         return NULL;
4438 }
4439 
4440 /* Verify a long-format btree block. */
4441 xfs_failaddr_t
4442 xfs_btree_lblock_verify(
4443         struct xfs_buf          *bp,
4444         unsigned int            max_recs)
4445 {
4446         struct xfs_mount        *mp = bp->b_mount;
4447         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
4448 
4449         /* numrecs verification */
4450         if (be16_to_cpu(block->bb_numrecs) > max_recs)
4451                 return __this_address;
4452 
4453         /* sibling pointer verification */
4454         if (block->bb_u.l.bb_leftsib != cpu_to_be64(NULLFSBLOCK) &&
4455             !xfs_verify_fsbno(mp, be64_to_cpu(block->bb_u.l.bb_leftsib)))
4456                 return __this_address;
4457         if (block->bb_u.l.bb_rightsib != cpu_to_be64(NULLFSBLOCK) &&
4458             !xfs_verify_fsbno(mp, be64_to_cpu(block->bb_u.l.bb_rightsib)))
4459                 return __this_address;
4460 
4461         return NULL;
4462 }
4463 
4464 /**
4465  * xfs_btree_sblock_v5hdr_verify() -- verify the v5 fields of a short-format
4466  *                                    btree block
4467  *
4468  * @bp: buffer containing the btree block
4469  */
4470 xfs_failaddr_t
4471 xfs_btree_sblock_v5hdr_verify(
4472         struct xfs_buf          *bp)
4473 {
4474         struct xfs_mount        *mp = bp->b_mount;
4475         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
4476         struct xfs_perag        *pag = bp->b_pag;
4477 
4478         if (!xfs_sb_version_hascrc(&mp->m_sb))
4479                 return __this_address;
4480         if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid))
4481                 return __this_address;
4482         if (block->bb_u.s.bb_blkno != cpu_to_be64(bp->b_bn))
4483                 return __this_address;
4484         if (pag && be32_to_cpu(block->bb_u.s.bb_owner) != pag->pag_agno)
4485                 return __this_address;
4486         return NULL;
4487 }
4488 
4489 /**
4490  * xfs_btree_sblock_verify() -- verify a short-format btree block
4491  *
4492  * @bp: buffer containing the btree block
4493  * @max_recs: maximum records allowed in this btree node
4494  */
4495 xfs_failaddr_t
4496 xfs_btree_sblock_verify(
4497         struct xfs_buf          *bp,
4498         unsigned int            max_recs)
4499 {
4500         struct xfs_mount        *mp = bp->b_mount;
4501         struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
4502         xfs_agblock_t           agno;
4503 
4504         /* numrecs verification */
4505         if (be16_to_cpu(block->bb_numrecs) > max_recs)
4506                 return __this_address;
4507 
4508         /* sibling pointer verification */
4509         agno = xfs_daddr_to_agno(mp, XFS_BUF_ADDR(bp));
4510         if (block->bb_u.s.bb_leftsib != cpu_to_be32(NULLAGBLOCK) &&
4511             !xfs_verify_agbno(mp, agno, be32_to_cpu(block->bb_u.s.bb_leftsib)))
4512                 return __this_address;
4513         if (block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK) &&
4514             !xfs_verify_agbno(mp, agno, be32_to_cpu(block->bb_u.s.bb_rightsib)))
4515                 return __this_address;
4516 
4517         return NULL;
4518 }
4519 
4520 /*
4521  * Calculate the number of btree levels needed to store a given number of
4522  * records in a short-format btree.
4523  */
4524 uint
4525 xfs_btree_compute_maxlevels(
4526         uint                    *limits,
4527         unsigned long           len)
4528 {
4529         uint                    level;
4530         unsigned long           maxblocks;
4531 
4532         maxblocks = (len + limits[0] - 1) / limits[0];
4533         for (level = 1; maxblocks > 1; level++)
4534                 maxblocks = (maxblocks + limits[1] - 1) / limits[1];
4535         return level;
4536 }
4537 
4538 /*
4539  * Query a regular btree for all records overlapping a given interval.
4540  * Start with a LE lookup of the key of low_rec and return all records
4541  * until we find a record with a key greater than the key of high_rec.
4542  */
4543 STATIC int
4544 xfs_btree_simple_query_range(
4545         struct xfs_btree_cur            *cur,
4546         union xfs_btree_key             *low_key,
4547         union xfs_btree_key             *high_key,
4548         xfs_btree_query_range_fn        fn,
4549         void                            *priv)
4550 {
4551         union xfs_btree_rec             *recp;
4552         union xfs_btree_key             rec_key;
4553         int64_t                         diff;
4554         int                             stat;
4555         bool                            firstrec = true;
4556         int                             error;
4557 
4558         ASSERT(cur->bc_ops->init_high_key_from_rec);
4559         ASSERT(cur->bc_ops->diff_two_keys);
4560 
4561         /*
4562          * Find the leftmost record.  The btree cursor must be set
4563          * to the low record used to generate low_key.
4564          */
4565         stat = 0;
4566         error = xfs_btree_lookup(cur, XFS_LOOKUP_LE, &stat);
4567         if (error)
4568                 goto out;
4569 
4570         /* Nothing?  See if there's anything to the right. */
4571         if (!stat) {
4572                 error = xfs_btree_increment(cur, 0, &stat);
4573                 if (error)
4574                         goto out;
4575         }
4576 
4577         while (stat) {
4578                 /* Find the record. */
4579                 error = xfs_btree_get_rec(cur, &recp, &stat);
4580                 if (error || !stat)
4581                         break;
4582 
4583                 /* Skip if high_key(rec) < low_key. */
4584                 if (firstrec) {
4585                         cur->bc_ops->init_high_key_from_rec(&rec_key, recp);
4586                         firstrec = false;
4587                         diff = cur->bc_ops->diff_two_keys(cur, low_key,
4588                                         &rec_key);
4589                         if (diff > 0)
4590                                 goto advloop;
4591                 }
4592 
4593                 /* Stop if high_key < low_key(rec). */
4594                 cur->bc_ops->init_key_from_rec(&rec_key, recp);
4595                 diff = cur->bc_ops->diff_two_keys(cur, &rec_key, high_key);
4596                 if (diff > 0)
4597                         break;
4598 
4599                 /* Callback */
4600                 error = fn(cur, recp, priv);
4601                 if (error)
4602                         break;
4603 
4604 advloop:
4605                 /* Move on to the next record. */
4606                 error = xfs_btree_increment(cur, 0, &stat);
4607                 if (error)
4608                         break;
4609         }
4610 
4611 out:
4612         return error;
4613 }
4614 
4615 /*
4616  * Query an overlapped interval btree for all records overlapping a given
4617  * interval.  This function roughly follows the algorithm given in
4618  * "Interval Trees" of _Introduction to Algorithms_, which is section
4619  * 14.3 in the 2nd and 3rd editions.
4620  *
4621  * First, generate keys for the low and high records passed in.
4622  *
4623  * For any leaf node, generate the high and low keys for the record.
4624  * If the record keys overlap with the query low/high keys, pass the
4625  * record to the function iterator.
4626  *
4627  * For any internal node, compare the low and high keys of each
4628  * pointer against the query low/high keys.  If there's an overlap,
4629  * follow the pointer.
4630  *
4631  * As an optimization, we stop scanning a block when we find a low key
4632  * that is greater than the query's high key.
4633  */
4634 STATIC int
4635 xfs_btree_overlapped_query_range(
4636         struct xfs_btree_cur            *cur,
4637         union xfs_btree_key             *low_key,
4638         union xfs_btree_key             *high_key,
4639         xfs_btree_query_range_fn        fn,
4640         void                            *priv)
4641 {
4642         union xfs_btree_ptr             ptr;
4643         union xfs_btree_ptr             *pp;
4644         union xfs_btree_key             rec_key;
4645         union xfs_btree_key             rec_hkey;
4646         union xfs_btree_key             *lkp;
4647         union xfs_btree_key             *hkp;
4648         union xfs_btree_rec             *recp;
4649         struct xfs_btree_block          *block;
4650         int64_t                         ldiff;
4651         int64_t                         hdiff;
4652         int                             level;
4653         struct xfs_buf                  *bp;
4654         int                             i;
4655         int                             error;
4656 
4657         /* Load the root of the btree. */
4658         level = cur->bc_nlevels - 1;
4659         cur->bc_ops->init_ptr_from_cur(cur, &ptr);
4660         error = xfs_btree_lookup_get_block(cur, level, &ptr, &block);
4661         if (error)
4662                 return error;
4663         xfs_btree_get_block(cur, level, &bp);
4664         trace_xfs_btree_overlapped_query_range(cur, level, bp);
4665 #ifdef DEBUG
4666         error = xfs_btree_check_block(cur, block, level, bp);
4667         if (error)
4668                 goto out;
4669 #endif
4670         cur->bc_ptrs[level] = 1;
4671 
4672         while (level < cur->bc_nlevels) {
4673                 block = xfs_btree_get_block(cur, level, &bp);
4674 
4675                 /* End of node, pop back towards the root. */
4676                 if (cur->bc_ptrs[level] > be16_to_cpu(block->bb_numrecs)) {
4677 pop_up:
4678                         if (level < cur->bc_nlevels - 1)
4679                                 cur->bc_ptrs[level + 1]++;
4680                         level++;
4681                         continue;
4682                 }
4683 
4684                 if (level == 0) {
4685                         /* Handle a leaf node. */
4686                         recp = xfs_btree_rec_addr(cur, cur->bc_ptrs[0], block);
4687 
4688                         cur->bc_ops->init_high_key_from_rec(&rec_hkey, recp);
4689                         ldiff = cur->bc_ops->diff_two_keys(cur, &rec_hkey,
4690                                         low_key);
4691 
4692                         cur->bc_ops->init_key_from_rec(&rec_key, recp);
4693                         hdiff = cur->bc_ops->diff_two_keys(cur, high_key,
4694                                         &rec_key);
4695 
4696                         /*
4697                          * If (record's high key >= query's low key) and
4698                          *    (query's high key >= record's low key), then
4699                          * this record overlaps the query range; callback.
4700                          */
4701                         if (ldiff >= 0 && hdiff >= 0) {
4702                                 error = fn(cur, recp, priv);
4703                                 if (error)
4704                                         break;
4705                         } else if (hdiff < 0) {
4706                                 /* Record is larger than high key; pop. */
4707                                 goto pop_up;
4708                         }
4709                         cur->bc_ptrs[level]++;
4710                         continue;
4711                 }
4712 
4713                 /* Handle an internal node. */
4714                 lkp = xfs_btree_key_addr(cur, cur->bc_ptrs[level], block);
4715                 hkp = xfs_btree_high_key_addr(cur, cur->bc_ptrs[level], block);
4716                 pp = xfs_btree_ptr_addr(cur, cur->bc_ptrs[level], block);
4717 
4718                 ldiff = cur->bc_ops->diff_two_keys(cur, hkp, low_key);
4719                 hdiff = cur->bc_ops->diff_two_keys(cur, high_key, lkp);
4720 
4721                 /*
4722                  * If (pointer's high key >= query's low key) and
4723                  *    (query's high key >= pointer's low key), then
4724                  * this record overlaps the query range; follow pointer.
4725                  */
4726                 if (ldiff >= 0 && hdiff >= 0) {
4727                         level--;
4728                         error = xfs_btree_lookup_get_block(cur, level, pp,
4729                                         &block);
4730                         if (error)
4731                                 goto out;
4732                         xfs_btree_get_block(cur, level, &bp);
4733                         trace_xfs_btree_overlapped_query_range(cur, level, bp);
4734 #ifdef DEBUG
4735                         error = xfs_btree_check_block(cur, block, level, bp);
4736                         if (error)
4737                                 goto out;
4738 #endif
4739                         cur->bc_ptrs[level] = 1;
4740                         continue;
4741                 } else if (hdiff < 0) {
4742                         /* The low key is larger than the upper range; pop. */
4743                         goto pop_up;
4744                 }
4745                 cur->bc_ptrs[level]++;
4746         }
4747 
4748 out:
4749         /*
4750          * If we don't end this function with the cursor pointing at a record
4751          * block, a subsequent non-error cursor deletion will not release
4752          * node-level buffers, causing a buffer leak.  This is quite possible
4753          * with a zero-results range query, so release the buffers if we
4754          * failed to return any results.
4755          */
4756         if (cur->bc_bufs[0] == NULL) {
4757                 for (i = 0; i < cur->bc_nlevels; i++) {
4758                         if (cur->bc_bufs[i]) {
4759                                 xfs_trans_brelse(cur->bc_tp, cur->bc_bufs[i]);
4760                                 cur->bc_bufs[i] = NULL;
4761                                 cur->bc_ptrs[i] = 0;
4762                                 cur->bc_ra[i] = 0;
4763                         }
4764                 }
4765         }
4766 
4767         return error;
4768 }
4769 
4770 /*
4771  * Query a btree for all records overlapping a given interval of keys.  The
4772  * supplied function will be called with each record found; return one of the
4773  * XFS_BTREE_QUERY_RANGE_{CONTINUE,ABORT} values or the usual negative error
4774  * code.  This function returns -ECANCELED, zero, or a negative error code.
4775  */
4776 int
4777 xfs_btree_query_range(
4778         struct xfs_btree_cur            *cur,
4779         union xfs_btree_irec            *low_rec,
4780         union xfs_btree_irec            *high_rec,
4781         xfs_btree_query_range_fn        fn,
4782         void                            *priv)
4783 {
4784         union xfs_btree_rec             rec;
4785         union xfs_btree_key             low_key;
4786         union xfs_btree_key             high_key;
4787 
4788         /* Find the keys of both ends of the interval. */
4789         cur->bc_rec = *high_rec;
4790         cur->bc_ops->init_rec_from_cur(cur, &rec);
4791         cur->bc_ops->init_key_from_rec(&high_key, &rec);
4792 
4793         cur->bc_rec = *low_rec;
4794         cur->bc_ops->init_rec_from_cur(cur, &rec);
4795         cur->bc_ops->init_key_from_rec(&low_key, &rec);
4796 
4797         /* Enforce low key < high key. */
4798         if (cur->bc_ops->diff_two_keys(cur, &low_key, &high_key) > 0)
4799                 return -EINVAL;
4800 
4801         if (!(cur->bc_flags & XFS_BTREE_OVERLAPPING))
4802                 return xfs_btree_simple_query_range(cur, &low_key,
4803                                 &high_key, fn, priv);
4804         return xfs_btree_overlapped_query_range(cur, &low_key, &high_key,
4805                         fn, priv);
4806 }
4807 
4808 /* Query a btree for all records. */
4809 int
4810 xfs_btree_query_all(
4811         struct xfs_btree_cur            *cur,
4812         xfs_btree_query_range_fn        fn,
4813         void                            *priv)
4814 {
4815         union xfs_btree_key             low_key;
4816         union xfs_btree_key             high_key;
4817 
4818         memset(&cur->bc_rec, 0, sizeof(cur->bc_rec));
4819         memset(&low_key, 0, sizeof(low_key));
4820         memset(&high_key, 0xFF, sizeof(high_key));
4821 
4822         return xfs_btree_simple_query_range(cur, &low_key, &high_key, fn, priv);
4823 }
4824 
4825 /*
4826  * Calculate the number of blocks needed to store a given number of records
4827  * in a short-format (per-AG metadata) btree.
4828  */
4829 unsigned long long
4830 xfs_btree_calc_size(
4831         uint                    *limits,
4832         unsigned long long      len)
4833 {
4834         int                     level;
4835         int                     maxrecs;
4836         unsigned long long      rval;
4837 
4838         maxrecs = limits[0];
4839         for (level = 0, rval = 0; len > 1; level++) {
4840                 len += maxrecs - 1;
4841                 do_div(len, maxrecs);
4842                 maxrecs = limits[1];
4843                 rval += len;
4844         }
4845         return rval;
4846 }
4847 
4848 static int
4849 xfs_btree_count_blocks_helper(
4850         struct xfs_btree_cur    *cur,
4851         int                     level,
4852         void                    *data)
4853 {
4854         xfs_extlen_t            *blocks = data;
4855         (*blocks)++;
4856 
4857         return 0;
4858 }
4859 
4860 /* Count the blocks in a btree and return the result in *blocks. */
4861 int
4862 xfs_btree_count_blocks(
4863         struct xfs_btree_cur    *cur,
4864         xfs_extlen_t            *blocks)
4865 {
4866         *blocks = 0;
4867         return xfs_btree_visit_blocks(cur, xfs_btree_count_blocks_helper,
4868                         blocks);
4869 }
4870 
4871 /* Compare two btree pointers. */
4872 int64_t
4873 xfs_btree_diff_two_ptrs(
4874         struct xfs_btree_cur            *cur,
4875         const union xfs_btree_ptr       *a,
4876         const union xfs_btree_ptr       *b)
4877 {
4878         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
4879                 return (int64_t)be64_to_cpu(a->l) - be64_to_cpu(b->l);
4880         return (int64_t)be32_to_cpu(a->s) - be32_to_cpu(b->s);
4881 }
4882 
4883 /* If there's an extent, we're done. */
4884 STATIC int
4885 xfs_btree_has_record_helper(
4886         struct xfs_btree_cur            *cur,
4887         union xfs_btree_rec             *rec,
4888         void                            *priv)
4889 {
4890         return -ECANCELED;
4891 }
4892 
4893 /* Is there a record covering a given range of keys? */
4894 int
4895 xfs_btree_has_record(
4896         struct xfs_btree_cur    *cur,
4897         union xfs_btree_irec    *low,
4898         union xfs_btree_irec    *high,
4899         bool                    *exists)
4900 {
4901         int                     error;
4902 
4903         error = xfs_btree_query_range(cur, low, high,
4904                         &xfs_btree_has_record_helper, NULL);
4905         if (error == -ECANCELED) {
4906                 *exists = true;
4907                 return 0;
4908         }
4909         *exists = false;
4910         return error;
4911 }
4912 
4913 /* Are there more records in this btree? */
4914 bool
4915 xfs_btree_has_more_records(
4916         struct xfs_btree_cur    *cur)
4917 {
4918         struct xfs_btree_block  *block;
4919         struct xfs_buf          *bp;
4920 
4921         block = xfs_btree_get_block(cur, 0, &bp);
4922 
4923         /* There are still records in this block. */
4924         if (cur->bc_ptrs[0] < xfs_btree_get_numrecs(block))
4925                 return true;
4926 
4927         /* There are more record blocks. */
4928         if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
4929                 return block->bb_u.l.bb_rightsib != cpu_to_be64(NULLFSBLOCK);
4930         else
4931                 return block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK);
4932 }
/* [<][>][^][v][top][bottom][index][help] */
root/fs/xfs/libxfs/xfs_btree.c

DEFINITIONS
