1 // SPDX-License-Identifier: GPL-2.0
3 * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
8 #include "xfs_shared.h"
9 #include "xfs_format.h"
10 #include "xfs_log_format.h"
11 #include "xfs_trans_resv.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"
23 #include "xfs_btree_staging.h"
25 #include "xfs_alloc_btree.h"
26 #include "xfs_ialloc_btree.h"
27 #include "xfs_bmap_btree.h"
28 #include "xfs_rmap_btree.h"
29 #include "xfs_refcount_btree.h"
32 * Btree magic numbers.
34 static const uint32_t xfs_magics[2][XFS_BTNUM_MAX] = {
35 { XFS_ABTB_MAGIC, XFS_ABTC_MAGIC, 0, XFS_BMAP_MAGIC, XFS_IBT_MAGIC,
37 { XFS_ABTB_CRC_MAGIC, XFS_ABTC_CRC_MAGIC, XFS_RMAP_CRC_MAGIC,
38 XFS_BMAP_CRC_MAGIC, XFS_IBT_CRC_MAGIC, XFS_FIBT_CRC_MAGIC,
47 uint32_t magic = xfs_magics[crc][btnum];
49 /* Ensure we asked for crc for crc-only magics. */
55 * These sibling pointer checks are optimised for null sibling pointers. This
56 * happens a lot, and we don't need to byte swap at runtime if the sibling
59 * These are explicitly marked at inline because the cost of calling them as
60 * functions instead of inlining them is about 36 bytes extra code per call site
61 * on x86-64. Yes, gcc-11 fails to inline them, and explicit inlining of these
62 * two sibling check functions reduces the compiled code size by over 300
65 static inline xfs_failaddr_t
66 xfs_btree_check_lblock_siblings(
68 struct xfs_btree_cur *cur,
73 xfs_fsblock_t sibling;
75 if (dsibling == cpu_to_be64(NULLFSBLOCK))
78 sibling = be64_to_cpu(dsibling);
80 return __this_address;
82 if (!xfs_btree_check_lptr(cur, sibling, level + 1))
83 return __this_address;
85 if (!xfs_verify_fsbno(mp, sibling))
86 return __this_address;
92 static inline xfs_failaddr_t
93 xfs_btree_check_sblock_siblings(
94 struct xfs_perag *pag,
95 struct xfs_btree_cur *cur,
100 xfs_agblock_t sibling;
102 if (dsibling == cpu_to_be32(NULLAGBLOCK))
105 sibling = be32_to_cpu(dsibling);
106 if (sibling == agbno)
107 return __this_address;
109 if (!xfs_btree_check_sptr(cur, sibling, level + 1))
110 return __this_address;
112 if (!xfs_verify_agbno(pag, sibling))
113 return __this_address;
119 * Check a long btree block header. Return the address of the failing check,
120 * or NULL if everything is ok.
123 __xfs_btree_check_lblock(
124 struct xfs_btree_cur *cur,
125 struct xfs_btree_block *block,
129 struct xfs_mount *mp = cur->bc_mp;
130 xfs_btnum_t btnum = cur->bc_btnum;
131 int crc = xfs_has_crc(mp);
133 xfs_fsblock_t fsb = NULLFSBLOCK;
136 if (!uuid_equal(&block->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid))
137 return __this_address;
138 if (block->bb_u.l.bb_blkno !=
139 cpu_to_be64(bp ? xfs_buf_daddr(bp) : XFS_BUF_DADDR_NULL))
140 return __this_address;
141 if (block->bb_u.l.bb_pad != cpu_to_be32(0))
142 return __this_address;
145 if (be32_to_cpu(block->bb_magic) != xfs_btree_magic(crc, btnum))
146 return __this_address;
147 if (be16_to_cpu(block->bb_level) != level)
148 return __this_address;
149 if (be16_to_cpu(block->bb_numrecs) >
150 cur->bc_ops->get_maxrecs(cur, level))
151 return __this_address;
154 fsb = XFS_DADDR_TO_FSB(mp, xfs_buf_daddr(bp));
156 fa = xfs_btree_check_lblock_siblings(mp, cur, level, fsb,
157 block->bb_u.l.bb_leftsib);
159 fa = xfs_btree_check_lblock_siblings(mp, cur, level, fsb,
160 block->bb_u.l.bb_rightsib);
164 /* Check a long btree block header. */
166 xfs_btree_check_lblock(
167 struct xfs_btree_cur *cur,
168 struct xfs_btree_block *block,
172 struct xfs_mount *mp = cur->bc_mp;
175 fa = __xfs_btree_check_lblock(cur, block, level, bp);
176 if (XFS_IS_CORRUPT(mp, fa != NULL) ||
177 XFS_TEST_ERROR(false, mp, XFS_ERRTAG_BTREE_CHECK_LBLOCK)) {
179 trace_xfs_btree_corrupt(bp, _RET_IP_);
180 return -EFSCORRUPTED;
186 * Check a short btree block header. Return the address of the failing check,
187 * or NULL if everything is ok.
190 __xfs_btree_check_sblock(
191 struct xfs_btree_cur *cur,
192 struct xfs_btree_block *block,
196 struct xfs_mount *mp = cur->bc_mp;
197 struct xfs_perag *pag = cur->bc_ag.pag;
198 xfs_btnum_t btnum = cur->bc_btnum;
199 int crc = xfs_has_crc(mp);
201 xfs_agblock_t agbno = NULLAGBLOCK;
204 if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid))
205 return __this_address;
206 if (block->bb_u.s.bb_blkno !=
207 cpu_to_be64(bp ? xfs_buf_daddr(bp) : XFS_BUF_DADDR_NULL))
208 return __this_address;
211 if (be32_to_cpu(block->bb_magic) != xfs_btree_magic(crc, btnum))
212 return __this_address;
213 if (be16_to_cpu(block->bb_level) != level)
214 return __this_address;
215 if (be16_to_cpu(block->bb_numrecs) >
216 cur->bc_ops->get_maxrecs(cur, level))
217 return __this_address;
220 agbno = xfs_daddr_to_agbno(mp, xfs_buf_daddr(bp));
222 fa = xfs_btree_check_sblock_siblings(pag, cur, level, agbno,
223 block->bb_u.s.bb_leftsib);
225 fa = xfs_btree_check_sblock_siblings(pag, cur, level, agbno,
226 block->bb_u.s.bb_rightsib);
230 /* Check a short btree block header. */
232 xfs_btree_check_sblock(
233 struct xfs_btree_cur *cur,
234 struct xfs_btree_block *block,
238 struct xfs_mount *mp = cur->bc_mp;
241 fa = __xfs_btree_check_sblock(cur, block, level, bp);
242 if (XFS_IS_CORRUPT(mp, fa != NULL) ||
243 XFS_TEST_ERROR(false, mp, XFS_ERRTAG_BTREE_CHECK_SBLOCK)) {
245 trace_xfs_btree_corrupt(bp, _RET_IP_);
246 return -EFSCORRUPTED;
252 * Debug routine: check that block header is ok.
255 xfs_btree_check_block(
256 struct xfs_btree_cur *cur, /* btree cursor */
257 struct xfs_btree_block *block, /* generic btree block pointer */
258 int level, /* level of the btree block */
259 struct xfs_buf *bp) /* buffer containing block, if any */
261 if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
262 return xfs_btree_check_lblock(cur, block, level, bp);
264 return xfs_btree_check_sblock(cur, block, level, bp);
267 /* Check that this long pointer is valid and points within the fs. */
269 xfs_btree_check_lptr(
270 struct xfs_btree_cur *cur,
276 return xfs_verify_fsbno(cur->bc_mp, fsbno);
279 /* Check that this short pointer is valid and points within the AG. */
281 xfs_btree_check_sptr(
282 struct xfs_btree_cur *cur,
288 return xfs_verify_agbno(cur->bc_ag.pag, agbno);
292 * Check that a given (indexed) btree pointer at a certain level of a
293 * btree is valid and doesn't point past where it should.
297 struct xfs_btree_cur *cur,
298 const union xfs_btree_ptr *ptr,
302 if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
303 if (xfs_btree_check_lptr(cur, be64_to_cpu((&ptr->l)[index]),
307 "Inode %llu fork %d: Corrupt btree %d pointer at level %d index %d.",
308 cur->bc_ino.ip->i_ino,
309 cur->bc_ino.whichfork, cur->bc_btnum,
312 if (xfs_btree_check_sptr(cur, be32_to_cpu((&ptr->s)[index]),
316 "AG %u: Corrupt btree %d pointer at level %d index %d.",
317 cur->bc_ag.pag->pag_agno, cur->bc_btnum,
321 return -EFSCORRUPTED;
325 # define xfs_btree_debug_check_ptr xfs_btree_check_ptr
327 # define xfs_btree_debug_check_ptr(...) (0)
331 * Calculate CRC on the whole btree block and stuff it into the
332 * long-form btree header.
334 * Prior to calculting the CRC, pull the LSN out of the buffer log item and put
335 * it into the buffer so recovery knows what the last modification was that made
339 xfs_btree_lblock_calc_crc(
342 struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
343 struct xfs_buf_log_item *bip = bp->b_log_item;
345 if (!xfs_has_crc(bp->b_mount))
348 block->bb_u.l.bb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
349 xfs_buf_update_cksum(bp, XFS_BTREE_LBLOCK_CRC_OFF);
353 xfs_btree_lblock_verify_crc(
356 struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
357 struct xfs_mount *mp = bp->b_mount;
359 if (xfs_has_crc(mp)) {
360 if (!xfs_log_check_lsn(mp, be64_to_cpu(block->bb_u.l.bb_lsn)))
362 return xfs_buf_verify_cksum(bp, XFS_BTREE_LBLOCK_CRC_OFF);
369 * Calculate CRC on the whole btree block and stuff it into the
370 * short-form btree header.
372 * Prior to calculting the CRC, pull the LSN out of the buffer log item and put
373 * it into the buffer so recovery knows what the last modification was that made
377 xfs_btree_sblock_calc_crc(
380 struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
381 struct xfs_buf_log_item *bip = bp->b_log_item;
383 if (!xfs_has_crc(bp->b_mount))
386 block->bb_u.s.bb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
387 xfs_buf_update_cksum(bp, XFS_BTREE_SBLOCK_CRC_OFF);
391 xfs_btree_sblock_verify_crc(
394 struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
395 struct xfs_mount *mp = bp->b_mount;
397 if (xfs_has_crc(mp)) {
398 if (!xfs_log_check_lsn(mp, be64_to_cpu(block->bb_u.s.bb_lsn)))
400 return xfs_buf_verify_cksum(bp, XFS_BTREE_SBLOCK_CRC_OFF);
407 xfs_btree_free_block(
408 struct xfs_btree_cur *cur,
413 error = cur->bc_ops->free_block(cur, bp);
415 xfs_trans_binval(cur->bc_tp, bp);
416 XFS_BTREE_STATS_INC(cur, free);
422 * Delete the btree cursor.
425 xfs_btree_del_cursor(
426 struct xfs_btree_cur *cur, /* btree cursor */
427 int error) /* del because of error */
429 int i; /* btree level */
432 * Clear the buffer pointers and release the buffers. If we're doing
433 * this because of an error, inspect all of the entries in the bc_bufs
434 * array for buffers to be unlocked. This is because some of the btree
435 * code works from level n down to 0, and if we get an error along the
436 * way we won't have initialized all the entries down to 0.
438 for (i = 0; i < cur->bc_nlevels; i++) {
439 if (cur->bc_levels[i].bp)
440 xfs_trans_brelse(cur->bc_tp, cur->bc_levels[i].bp);
446 * If we are doing a BMBT update, the number of unaccounted blocks
447 * allocated during this cursor life time should be zero. If it's not
448 * zero, then we should be shut down or on our way to shutdown due to
449 * cancelling a dirty transaction on error.
451 ASSERT(cur->bc_btnum != XFS_BTNUM_BMAP || cur->bc_ino.allocated == 0 ||
452 xfs_is_shutdown(cur->bc_mp) || error != 0);
453 if (unlikely(cur->bc_flags & XFS_BTREE_STAGING))
454 kmem_free(cur->bc_ops);
455 if (!(cur->bc_flags & XFS_BTREE_LONG_PTRS) && cur->bc_ag.pag)
456 xfs_perag_put(cur->bc_ag.pag);
457 kmem_cache_free(cur->bc_cache, cur);
461 * Duplicate the btree cursor.
462 * Allocate a new one, copy the record, re-get the buffers.
465 xfs_btree_dup_cursor(
466 struct xfs_btree_cur *cur, /* input cursor */
467 struct xfs_btree_cur **ncur) /* output cursor */
469 struct xfs_buf *bp; /* btree block's buffer pointer */
470 int error; /* error return value */
471 int i; /* level number of btree block */
472 xfs_mount_t *mp; /* mount structure for filesystem */
473 struct xfs_btree_cur *new; /* new cursor value */
474 xfs_trans_t *tp; /* transaction pointer, can be NULL */
480 * Allocate a new cursor like the old one.
482 new = cur->bc_ops->dup_cursor(cur);
485 * Copy the record currently in the cursor.
487 new->bc_rec = cur->bc_rec;
490 * For each level current, re-get the buffer and copy the ptr value.
492 for (i = 0; i < new->bc_nlevels; i++) {
493 new->bc_levels[i].ptr = cur->bc_levels[i].ptr;
494 new->bc_levels[i].ra = cur->bc_levels[i].ra;
495 bp = cur->bc_levels[i].bp;
497 error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
498 xfs_buf_daddr(bp), mp->m_bsize,
500 cur->bc_ops->buf_ops);
502 xfs_btree_del_cursor(new, error);
507 new->bc_levels[i].bp = bp;
514 * XFS btree block layout and addressing:
516 * There are two types of blocks in the btree: leaf and non-leaf blocks.
518 * The leaf record start with a header then followed by records containing
519 * the values. A non-leaf block also starts with the same header, and
520 * then first contains lookup keys followed by an equal number of pointers
521 * to the btree blocks at the previous level.
523 * +--------+-------+-------+-------+-------+-------+-------+
524 * Leaf: | header | rec 1 | rec 2 | rec 3 | rec 4 | rec 5 | rec N |
525 * +--------+-------+-------+-------+-------+-------+-------+
527 * +--------+-------+-------+-------+-------+-------+-------+
528 * Non-Leaf: | header | key 1 | key 2 | key N | ptr 1 | ptr 2 | ptr N |
529 * +--------+-------+-------+-------+-------+-------+-------+
531 * The header is called struct xfs_btree_block for reasons better left unknown
532 * and comes in different versions for short (32bit) and long (64bit) block
533 * pointers. The record and key structures are defined by the btree instances
534 * and opaque to the btree core. The block pointers are simple disk endian
535 * integers, available in a short (32bit) and long (64bit) variant.
537 * The helpers below calculate the offset of a given record, key or pointer
538 * into a btree block (xfs_btree_*_offset) or return a pointer to the given
539 * record, key or pointer (xfs_btree_*_addr). Note that all addressing
540 * inside the btree block is done using indices starting at one, not zero!
542 * If XFS_BTREE_OVERLAPPING is set, then this btree supports keys containing
543 * overlapping intervals. In such a tree, records are still sorted lowest to
544 * highest and indexed by the smallest key value that refers to the record.
545 * However, nodes are different: each pointer has two associated keys -- one
546 * indexing the lowest key available in the block(s) below (the same behavior
547 * as the key in a regular btree) and another indexing the highest key
548 * available in the block(s) below. Because records are /not/ sorted by the
549 * highest key, all leaf block updates require us to compute the highest key
550 * that matches any record in the leaf and to recursively update the high keys
551 * in the nodes going further up in the tree, if necessary. Nodes look like
554 * +--------+-----+-----+-----+-----+-----+-------+-------+-----+
555 * Non-Leaf: | header | lo1 | hi1 | lo2 | hi2 | ... | ptr 1 | ptr 2 | ... |
556 * +--------+-----+-----+-----+-----+-----+-------+-------+-----+
558 * To perform an interval query on an overlapped tree, perform the usual
559 * depth-first search and use the low and high keys to decide if we can skip
560 * that particular node. If a leaf node is reached, return the records that
561 * intersect the interval. Note that an interval query may return numerous
562 * entries. For a non-overlapped tree, simply search for the record associated
563 * with the lowest key and iterate forward until a non-matching record is
564 * found. Section 14.3 ("Interval Trees") of _Introduction to Algorithms_ by
565 * Cormen, Leiserson, Rivest, and Stein (2nd or 3rd ed. only) discuss this in
568 * Why do we care about overlapping intervals? Let's say you have a bunch of
569 * reverse mapping records on a reflink filesystem:
571 * 1: +- file A startblock B offset C length D -----------+
572 * 2: +- file E startblock F offset G length H --------------+
573 * 3: +- file I startblock F offset J length K --+
574 * 4: +- file L... --+
576 * Now say we want to map block (B+D) into file A at offset (C+D). Ideally,
577 * we'd simply increment the length of record 1. But how do we find the record
578 * that ends at (B+D-1) (i.e. record 1)? A LE lookup of (B+D-1) would return
579 * record 3 because the keys are ordered first by startblock. An interval
580 * query would return records 1 and 2 because they both overlap (B+D-1), and
581 * from that we can pick out record 1 as the appropriate left neighbor.
583 * In the non-overlapped case you can do a LE lookup and decrement the cursor
584 * because a record's interval must end before the next record.
588 * Return size of the btree block header for this btree instance.
590 static inline size_t xfs_btree_block_len(struct xfs_btree_cur *cur)
592 if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
593 if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS)
594 return XFS_BTREE_LBLOCK_CRC_LEN;
595 return XFS_BTREE_LBLOCK_LEN;
597 if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS)
598 return XFS_BTREE_SBLOCK_CRC_LEN;
599 return XFS_BTREE_SBLOCK_LEN;
603 * Return size of btree block pointers for this btree instance.
605 static inline size_t xfs_btree_ptr_len(struct xfs_btree_cur *cur)
607 return (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
608 sizeof(__be64) : sizeof(__be32);
612 * Calculate offset of the n-th record in a btree block.
615 xfs_btree_rec_offset(
616 struct xfs_btree_cur *cur,
619 return xfs_btree_block_len(cur) +
620 (n - 1) * cur->bc_ops->rec_len;
624 * Calculate offset of the n-th key in a btree block.
627 xfs_btree_key_offset(
628 struct xfs_btree_cur *cur,
631 return xfs_btree_block_len(cur) +
632 (n - 1) * cur->bc_ops->key_len;
636 * Calculate offset of the n-th high key in a btree block.
639 xfs_btree_high_key_offset(
640 struct xfs_btree_cur *cur,
643 return xfs_btree_block_len(cur) +
644 (n - 1) * cur->bc_ops->key_len + (cur->bc_ops->key_len / 2);
648 * Calculate offset of the n-th block pointer in a btree block.
651 xfs_btree_ptr_offset(
652 struct xfs_btree_cur *cur,
656 return xfs_btree_block_len(cur) +
657 cur->bc_ops->get_maxrecs(cur, level) * cur->bc_ops->key_len +
658 (n - 1) * xfs_btree_ptr_len(cur);
662 * Return a pointer to the n-th record in the btree block.
664 union xfs_btree_rec *
666 struct xfs_btree_cur *cur,
668 struct xfs_btree_block *block)
670 return (union xfs_btree_rec *)
671 ((char *)block + xfs_btree_rec_offset(cur, n));
675 * Return a pointer to the n-th key in the btree block.
677 union xfs_btree_key *
679 struct xfs_btree_cur *cur,
681 struct xfs_btree_block *block)
683 return (union xfs_btree_key *)
684 ((char *)block + xfs_btree_key_offset(cur, n));
688 * Return a pointer to the n-th high key in the btree block.
690 union xfs_btree_key *
691 xfs_btree_high_key_addr(
692 struct xfs_btree_cur *cur,
694 struct xfs_btree_block *block)
696 return (union xfs_btree_key *)
697 ((char *)block + xfs_btree_high_key_offset(cur, n));
701 * Return a pointer to the n-th block pointer in the btree block.
703 union xfs_btree_ptr *
705 struct xfs_btree_cur *cur,
707 struct xfs_btree_block *block)
709 int level = xfs_btree_get_level(block);
711 ASSERT(block->bb_level != 0);
713 return (union xfs_btree_ptr *)
714 ((char *)block + xfs_btree_ptr_offset(cur, n, level));
719 struct xfs_btree_cur *cur)
721 ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
723 if (cur->bc_flags & XFS_BTREE_STAGING)
724 return cur->bc_ino.ifake->if_fork;
725 return xfs_ifork_ptr(cur->bc_ino.ip, cur->bc_ino.whichfork);
729 * Get the root block which is stored in the inode.
731 * For now this btree implementation assumes the btree root is always
732 * stored in the if_broot field of an inode fork.
734 STATIC struct xfs_btree_block *
736 struct xfs_btree_cur *cur)
738 struct xfs_ifork *ifp = xfs_btree_ifork_ptr(cur);
740 return (struct xfs_btree_block *)ifp->if_broot;
744 * Retrieve the block pointer from the cursor at the given level.
745 * This may be an inode btree root or from a buffer.
747 struct xfs_btree_block * /* generic btree block pointer */
749 struct xfs_btree_cur *cur, /* btree cursor */
750 int level, /* level in btree */
751 struct xfs_buf **bpp) /* buffer containing the block */
753 if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
754 (level == cur->bc_nlevels - 1)) {
756 return xfs_btree_get_iroot(cur);
759 *bpp = cur->bc_levels[level].bp;
760 return XFS_BUF_TO_BLOCK(*bpp);
764 * Change the cursor to point to the first record at the given level.
765 * Other levels are unaffected.
767 STATIC int /* success=1, failure=0 */
769 struct xfs_btree_cur *cur, /* btree cursor */
770 int level) /* level to change */
772 struct xfs_btree_block *block; /* generic btree block pointer */
773 struct xfs_buf *bp; /* buffer containing block */
776 * Get the block pointer for this level.
778 block = xfs_btree_get_block(cur, level, &bp);
779 if (xfs_btree_check_block(cur, block, level, bp))
782 * It's empty, there is no such record.
784 if (!block->bb_numrecs)
787 * Set the ptr value to 1, that's the first record/key.
789 cur->bc_levels[level].ptr = 1;
794 * Change the cursor to point to the last record in the current block
795 * at the given level. Other levels are unaffected.
797 STATIC int /* success=1, failure=0 */
799 struct xfs_btree_cur *cur, /* btree cursor */
800 int level) /* level to change */
802 struct xfs_btree_block *block; /* generic btree block pointer */
803 struct xfs_buf *bp; /* buffer containing block */
806 * Get the block pointer for this level.
808 block = xfs_btree_get_block(cur, level, &bp);
809 if (xfs_btree_check_block(cur, block, level, bp))
812 * It's empty, there is no such record.
814 if (!block->bb_numrecs)
817 * Set the ptr value to numrecs, that's the last record/key.
819 cur->bc_levels[level].ptr = be16_to_cpu(block->bb_numrecs);
824 * Compute first and last byte offsets for the fields given.
825 * Interprets the offsets table, which contains struct field offsets.
829 uint32_t fields, /* bitmask of fields */
830 const short *offsets, /* table of field offsets */
831 int nbits, /* number of bits to inspect */
832 int *first, /* output: first byte offset */
833 int *last) /* output: last byte offset */
835 int i; /* current bit number */
836 uint32_t imask; /* mask for current bit number */
840 * Find the lowest bit, so the first byte offset.
842 for (i = 0, imask = 1u; ; i++, imask <<= 1) {
843 if (imask & fields) {
849 * Find the highest bit, so the last byte offset.
851 for (i = nbits - 1, imask = 1u << i; ; i--, imask >>= 1) {
852 if (imask & fields) {
853 *last = offsets[i + 1] - 1;
860 * Get a buffer for the block, return it read in.
861 * Long-form addressing.
865 struct xfs_mount *mp, /* file system mount point */
866 struct xfs_trans *tp, /* transaction pointer */
867 xfs_fsblock_t fsbno, /* file system block number */
868 struct xfs_buf **bpp, /* buffer for fsbno */
869 int refval, /* ref count value for buffer */
870 const struct xfs_buf_ops *ops)
872 struct xfs_buf *bp; /* return value */
873 xfs_daddr_t d; /* real disk block address */
876 if (!xfs_verify_fsbno(mp, fsbno))
877 return -EFSCORRUPTED;
878 d = XFS_FSB_TO_DADDR(mp, fsbno);
879 error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, d,
880 mp->m_bsize, 0, &bp, ops);
884 xfs_buf_set_ref(bp, refval);
890 * Read-ahead the block, don't wait for it, don't return a buffer.
891 * Long-form addressing.
895 xfs_btree_reada_bufl(
896 struct xfs_mount *mp, /* file system mount point */
897 xfs_fsblock_t fsbno, /* file system block number */
898 xfs_extlen_t count, /* count of filesystem blocks */
899 const struct xfs_buf_ops *ops)
903 ASSERT(fsbno != NULLFSBLOCK);
904 d = XFS_FSB_TO_DADDR(mp, fsbno);
905 xfs_buf_readahead(mp->m_ddev_targp, d, mp->m_bsize * count, ops);
909 * Read-ahead the block, don't wait for it, don't return a buffer.
910 * Short-form addressing.
914 xfs_btree_reada_bufs(
915 struct xfs_mount *mp, /* file system mount point */
916 xfs_agnumber_t agno, /* allocation group number */
917 xfs_agblock_t agbno, /* allocation group block number */
918 xfs_extlen_t count, /* count of filesystem blocks */
919 const struct xfs_buf_ops *ops)
923 ASSERT(agno != NULLAGNUMBER);
924 ASSERT(agbno != NULLAGBLOCK);
925 d = XFS_AGB_TO_DADDR(mp, agno, agbno);
926 xfs_buf_readahead(mp->m_ddev_targp, d, mp->m_bsize * count, ops);
930 xfs_btree_readahead_lblock(
931 struct xfs_btree_cur *cur,
933 struct xfs_btree_block *block)
936 xfs_fsblock_t left = be64_to_cpu(block->bb_u.l.bb_leftsib);
937 xfs_fsblock_t right = be64_to_cpu(block->bb_u.l.bb_rightsib);
939 if ((lr & XFS_BTCUR_LEFTRA) && left != NULLFSBLOCK) {
940 xfs_btree_reada_bufl(cur->bc_mp, left, 1,
941 cur->bc_ops->buf_ops);
945 if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLFSBLOCK) {
946 xfs_btree_reada_bufl(cur->bc_mp, right, 1,
947 cur->bc_ops->buf_ops);
955 xfs_btree_readahead_sblock(
956 struct xfs_btree_cur *cur,
958 struct xfs_btree_block *block)
961 xfs_agblock_t left = be32_to_cpu(block->bb_u.s.bb_leftsib);
962 xfs_agblock_t right = be32_to_cpu(block->bb_u.s.bb_rightsib);
965 if ((lr & XFS_BTCUR_LEFTRA) && left != NULLAGBLOCK) {
966 xfs_btree_reada_bufs(cur->bc_mp, cur->bc_ag.pag->pag_agno,
967 left, 1, cur->bc_ops->buf_ops);
971 if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLAGBLOCK) {
972 xfs_btree_reada_bufs(cur->bc_mp, cur->bc_ag.pag->pag_agno,
973 right, 1, cur->bc_ops->buf_ops);
981 * Read-ahead btree blocks, at the given level.
982 * Bits in lr are set from XFS_BTCUR_{LEFT,RIGHT}RA.
986 struct xfs_btree_cur *cur, /* btree cursor */
987 int lev, /* level in btree */
988 int lr) /* left/right bits */
990 struct xfs_btree_block *block;
993 * No readahead needed if we are at the root level and the
994 * btree root is stored in the inode.
996 if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
997 (lev == cur->bc_nlevels - 1))
1000 if ((cur->bc_levels[lev].ra | lr) == cur->bc_levels[lev].ra)
1003 cur->bc_levels[lev].ra |= lr;
1004 block = XFS_BUF_TO_BLOCK(cur->bc_levels[lev].bp);
1006 if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1007 return xfs_btree_readahead_lblock(cur, lr, block);
1008 return xfs_btree_readahead_sblock(cur, lr, block);
1012 xfs_btree_ptr_to_daddr(
1013 struct xfs_btree_cur *cur,
1014 const union xfs_btree_ptr *ptr,
1017 xfs_fsblock_t fsbno;
1018 xfs_agblock_t agbno;
1021 error = xfs_btree_check_ptr(cur, ptr, 0, 1);
1025 if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
1026 fsbno = be64_to_cpu(ptr->l);
1027 *daddr = XFS_FSB_TO_DADDR(cur->bc_mp, fsbno);
1029 agbno = be32_to_cpu(ptr->s);
1030 *daddr = XFS_AGB_TO_DADDR(cur->bc_mp, cur->bc_ag.pag->pag_agno,
1038 * Readahead @count btree blocks at the given @ptr location.
1040 * We don't need to care about long or short form btrees here as we have a
1041 * method of converting the ptr directly to a daddr available to us.
1044 xfs_btree_readahead_ptr(
1045 struct xfs_btree_cur *cur,
1046 union xfs_btree_ptr *ptr,
1051 if (xfs_btree_ptr_to_daddr(cur, ptr, &daddr))
1053 xfs_buf_readahead(cur->bc_mp->m_ddev_targp, daddr,
1054 cur->bc_mp->m_bsize * count, cur->bc_ops->buf_ops);
1058 * Set the buffer for level "lev" in the cursor to bp, releasing
1059 * any previous buffer.
1063 struct xfs_btree_cur *cur, /* btree cursor */
1064 int lev, /* level in btree */
1065 struct xfs_buf *bp) /* new buffer to set */
1067 struct xfs_btree_block *b; /* btree block */
1069 if (cur->bc_levels[lev].bp)
1070 xfs_trans_brelse(cur->bc_tp, cur->bc_levels[lev].bp);
1071 cur->bc_levels[lev].bp = bp;
1072 cur->bc_levels[lev].ra = 0;
1074 b = XFS_BUF_TO_BLOCK(bp);
1075 if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
1076 if (b->bb_u.l.bb_leftsib == cpu_to_be64(NULLFSBLOCK))
1077 cur->bc_levels[lev].ra |= XFS_BTCUR_LEFTRA;
1078 if (b->bb_u.l.bb_rightsib == cpu_to_be64(NULLFSBLOCK))
1079 cur->bc_levels[lev].ra |= XFS_BTCUR_RIGHTRA;
1081 if (b->bb_u.s.bb_leftsib == cpu_to_be32(NULLAGBLOCK))
1082 cur->bc_levels[lev].ra |= XFS_BTCUR_LEFTRA;
1083 if (b->bb_u.s.bb_rightsib == cpu_to_be32(NULLAGBLOCK))
1084 cur->bc_levels[lev].ra |= XFS_BTCUR_RIGHTRA;
1089 xfs_btree_ptr_is_null(
1090 struct xfs_btree_cur *cur,
1091 const union xfs_btree_ptr *ptr)
1093 if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1094 return ptr->l == cpu_to_be64(NULLFSBLOCK);
1096 return ptr->s == cpu_to_be32(NULLAGBLOCK);
1100 xfs_btree_set_ptr_null(
1101 struct xfs_btree_cur *cur,
1102 union xfs_btree_ptr *ptr)
1104 if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1105 ptr->l = cpu_to_be64(NULLFSBLOCK);
1107 ptr->s = cpu_to_be32(NULLAGBLOCK);
1111 * Get/set/init sibling pointers
1114 xfs_btree_get_sibling(
1115 struct xfs_btree_cur *cur,
1116 struct xfs_btree_block *block,
1117 union xfs_btree_ptr *ptr,
1120 ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
1122 if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
1123 if (lr == XFS_BB_RIGHTSIB)
1124 ptr->l = block->bb_u.l.bb_rightsib;
1126 ptr->l = block->bb_u.l.bb_leftsib;
1128 if (lr == XFS_BB_RIGHTSIB)
1129 ptr->s = block->bb_u.s.bb_rightsib;
1131 ptr->s = block->bb_u.s.bb_leftsib;
1136 xfs_btree_set_sibling(
1137 struct xfs_btree_cur *cur,
1138 struct xfs_btree_block *block,
1139 const union xfs_btree_ptr *ptr,
1142 ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
1144 if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
1145 if (lr == XFS_BB_RIGHTSIB)
1146 block->bb_u.l.bb_rightsib = ptr->l;
1148 block->bb_u.l.bb_leftsib = ptr->l;
1150 if (lr == XFS_BB_RIGHTSIB)
1151 block->bb_u.s.bb_rightsib = ptr->s;
1153 block->bb_u.s.bb_leftsib = ptr->s;
1158 xfs_btree_init_block_int(
1159 struct xfs_mount *mp,
1160 struct xfs_btree_block *buf,
1168 int crc = xfs_has_crc(mp);
1169 __u32 magic = xfs_btree_magic(crc, btnum);
1171 buf->bb_magic = cpu_to_be32(magic);
1172 buf->bb_level = cpu_to_be16(level);
1173 buf->bb_numrecs = cpu_to_be16(numrecs);
1175 if (flags & XFS_BTREE_LONG_PTRS) {
1176 buf->bb_u.l.bb_leftsib = cpu_to_be64(NULLFSBLOCK);
1177 buf->bb_u.l.bb_rightsib = cpu_to_be64(NULLFSBLOCK);
1179 buf->bb_u.l.bb_blkno = cpu_to_be64(blkno);
1180 buf->bb_u.l.bb_owner = cpu_to_be64(owner);
1181 uuid_copy(&buf->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid);
1182 buf->bb_u.l.bb_pad = 0;
1183 buf->bb_u.l.bb_lsn = 0;
1186 /* owner is a 32 bit value on short blocks */
1187 __u32 __owner = (__u32)owner;
1189 buf->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
1190 buf->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
1192 buf->bb_u.s.bb_blkno = cpu_to_be64(blkno);
1193 buf->bb_u.s.bb_owner = cpu_to_be32(__owner);
1194 uuid_copy(&buf->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid);
1195 buf->bb_u.s.bb_lsn = 0;
1201 xfs_btree_init_block(
1202 struct xfs_mount *mp,
1209 xfs_btree_init_block_int(mp, XFS_BUF_TO_BLOCK(bp), xfs_buf_daddr(bp),
1210 btnum, level, numrecs, owner, 0);
1214 xfs_btree_init_block_cur(
1215 struct xfs_btree_cur *cur,
1223 * we can pull the owner from the cursor right now as the different
1224 * owners align directly with the pointer size of the btree. This may
1225 * change in future, but is safe for current users of the generic btree
1228 if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1229 owner = cur->bc_ino.ip->i_ino;
1231 owner = cur->bc_ag.pag->pag_agno;
1233 xfs_btree_init_block_int(cur->bc_mp, XFS_BUF_TO_BLOCK(bp),
1234 xfs_buf_daddr(bp), cur->bc_btnum, level,
1235 numrecs, owner, cur->bc_flags);
1239 * Return true if ptr is the last record in the btree and
1240 * we need to track updates to this record. The decision
1241 * will be further refined in the update_lastrec method.
1244 xfs_btree_is_lastrec(
1245 struct xfs_btree_cur *cur,
1246 struct xfs_btree_block *block,
1249 union xfs_btree_ptr ptr;
1253 if (!(cur->bc_flags & XFS_BTREE_LASTREC_UPDATE))
1256 xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
1257 if (!xfs_btree_ptr_is_null(cur, &ptr))
1263 xfs_btree_buf_to_ptr(
1264 struct xfs_btree_cur *cur,
1266 union xfs_btree_ptr *ptr)
1268 if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
1269 ptr->l = cpu_to_be64(XFS_DADDR_TO_FSB(cur->bc_mp,
1270 xfs_buf_daddr(bp)));
1272 ptr->s = cpu_to_be32(xfs_daddr_to_agbno(cur->bc_mp,
1273 xfs_buf_daddr(bp)));
1279 struct xfs_btree_cur *cur,
1282 switch (cur->bc_btnum) {
1285 xfs_buf_set_ref(bp, XFS_ALLOC_BTREE_REF);
1288 case XFS_BTNUM_FINO:
1289 xfs_buf_set_ref(bp, XFS_INO_BTREE_REF);
1291 case XFS_BTNUM_BMAP:
1292 xfs_buf_set_ref(bp, XFS_BMAP_BTREE_REF);
1294 case XFS_BTNUM_RMAP:
1295 xfs_buf_set_ref(bp, XFS_RMAP_BTREE_REF);
1297 case XFS_BTNUM_REFC:
1298 xfs_buf_set_ref(bp, XFS_REFC_BTREE_REF);
1306 xfs_btree_get_buf_block(
1307 struct xfs_btree_cur *cur,
1308 const union xfs_btree_ptr *ptr,
1309 struct xfs_btree_block **block,
1310 struct xfs_buf **bpp)
1312 struct xfs_mount *mp = cur->bc_mp;
1316 error = xfs_btree_ptr_to_daddr(cur, ptr, &d);
1319 error = xfs_trans_get_buf(cur->bc_tp, mp->m_ddev_targp, d, mp->m_bsize,
1324 (*bpp)->b_ops = cur->bc_ops->buf_ops;
1325 *block = XFS_BUF_TO_BLOCK(*bpp);
1330 * Read in the buffer at the given ptr and return the buffer and
1331 * the block pointer within the buffer.
1334 xfs_btree_read_buf_block(
1335 struct xfs_btree_cur *cur,
1336 const union xfs_btree_ptr *ptr,
1338 struct xfs_btree_block **block,
1339 struct xfs_buf **bpp)
1341 struct xfs_mount *mp = cur->bc_mp;
1345 /* need to sort out how callers deal with failures first */
1346 ASSERT(!(flags & XBF_TRYLOCK));
1348 error = xfs_btree_ptr_to_daddr(cur, ptr, &d);
1351 error = xfs_trans_read_buf(mp, cur->bc_tp, mp->m_ddev_targp, d,
1352 mp->m_bsize, flags, bpp,
1353 cur->bc_ops->buf_ops);
1357 xfs_btree_set_refs(cur, *bpp);
1358 *block = XFS_BUF_TO_BLOCK(*bpp);
1363 * Copy keys from one btree block to another.
1366 xfs_btree_copy_keys(
1367 struct xfs_btree_cur *cur,
1368 union xfs_btree_key *dst_key,
1369 const union xfs_btree_key *src_key,
1372 ASSERT(numkeys >= 0);
1373 memcpy(dst_key, src_key, numkeys * cur->bc_ops->key_len);
1377 * Copy records from one btree block to another.
1380 xfs_btree_copy_recs(
1381 struct xfs_btree_cur *cur,
1382 union xfs_btree_rec *dst_rec,
1383 union xfs_btree_rec *src_rec,
1386 ASSERT(numrecs >= 0);
1387 memcpy(dst_rec, src_rec, numrecs * cur->bc_ops->rec_len);
1391 * Copy block pointers from one btree block to another.
1394 xfs_btree_copy_ptrs(
1395 struct xfs_btree_cur *cur,
1396 union xfs_btree_ptr *dst_ptr,
1397 const union xfs_btree_ptr *src_ptr,
1400 ASSERT(numptrs >= 0);
1401 memcpy(dst_ptr, src_ptr, numptrs * xfs_btree_ptr_len(cur));
1405 * Shift keys one index left/right inside a single btree block.
1408 xfs_btree_shift_keys(
1409 struct xfs_btree_cur *cur,
1410 union xfs_btree_key *key,
1416 ASSERT(numkeys >= 0);
1417 ASSERT(dir == 1 || dir == -1);
1419 dst_key = (char *)key + (dir * cur->bc_ops->key_len);
1420 memmove(dst_key, key, numkeys * cur->bc_ops->key_len);
1424 * Shift records one index left/right inside a single btree block.
1427 xfs_btree_shift_recs(
1428 struct xfs_btree_cur *cur,
1429 union xfs_btree_rec *rec,
1435 ASSERT(numrecs >= 0);
1436 ASSERT(dir == 1 || dir == -1);
1438 dst_rec = (char *)rec + (dir * cur->bc_ops->rec_len);
1439 memmove(dst_rec, rec, numrecs * cur->bc_ops->rec_len);
1443 * Shift block pointers one index left/right inside a single btree block.
1446 xfs_btree_shift_ptrs(
1447 struct xfs_btree_cur *cur,
1448 union xfs_btree_ptr *ptr,
1454 ASSERT(numptrs >= 0);
1455 ASSERT(dir == 1 || dir == -1);
1457 dst_ptr = (char *)ptr + (dir * xfs_btree_ptr_len(cur));
1458 memmove(dst_ptr, ptr, numptrs * xfs_btree_ptr_len(cur));
1462 * Log key values from the btree block.
1466 struct xfs_btree_cur *cur,
1473 xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1474 xfs_trans_log_buf(cur->bc_tp, bp,
1475 xfs_btree_key_offset(cur, first),
1476 xfs_btree_key_offset(cur, last + 1) - 1);
1478 xfs_trans_log_inode(cur->bc_tp, cur->bc_ino.ip,
1479 xfs_ilog_fbroot(cur->bc_ino.whichfork));
1484 * Log record values from the btree block.
1488 struct xfs_btree_cur *cur,
1494 xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1495 xfs_trans_log_buf(cur->bc_tp, bp,
1496 xfs_btree_rec_offset(cur, first),
1497 xfs_btree_rec_offset(cur, last + 1) - 1);
1502 * Log block pointer fields from a btree block (nonleaf).
1506 struct xfs_btree_cur *cur, /* btree cursor */
1507 struct xfs_buf *bp, /* buffer containing btree block */
1508 int first, /* index of first pointer to log */
1509 int last) /* index of last pointer to log */
1513 struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
1514 int level = xfs_btree_get_level(block);
1516 xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1517 xfs_trans_log_buf(cur->bc_tp, bp,
1518 xfs_btree_ptr_offset(cur, first, level),
1519 xfs_btree_ptr_offset(cur, last + 1, level) - 1);
1521 xfs_trans_log_inode(cur->bc_tp, cur->bc_ino.ip,
1522 xfs_ilog_fbroot(cur->bc_ino.whichfork));
1528 * Log fields from a btree block header.
1531 xfs_btree_log_block(
1532 struct xfs_btree_cur *cur, /* btree cursor */
1533 struct xfs_buf *bp, /* buffer containing btree block */
1534 uint32_t fields) /* mask of fields: XFS_BB_... */
1536 int first; /* first byte offset logged */
1537 int last; /* last byte offset logged */
1538 static const short soffsets[] = { /* table of offsets (short) */
1539 offsetof(struct xfs_btree_block, bb_magic),
1540 offsetof(struct xfs_btree_block, bb_level),
1541 offsetof(struct xfs_btree_block, bb_numrecs),
1542 offsetof(struct xfs_btree_block, bb_u.s.bb_leftsib),
1543 offsetof(struct xfs_btree_block, bb_u.s.bb_rightsib),
1544 offsetof(struct xfs_btree_block, bb_u.s.bb_blkno),
1545 offsetof(struct xfs_btree_block, bb_u.s.bb_lsn),
1546 offsetof(struct xfs_btree_block, bb_u.s.bb_uuid),
1547 offsetof(struct xfs_btree_block, bb_u.s.bb_owner),
1548 offsetof(struct xfs_btree_block, bb_u.s.bb_crc),
1549 XFS_BTREE_SBLOCK_CRC_LEN
1551 static const short loffsets[] = { /* table of offsets (long) */
1552 offsetof(struct xfs_btree_block, bb_magic),
1553 offsetof(struct xfs_btree_block, bb_level),
1554 offsetof(struct xfs_btree_block, bb_numrecs),
1555 offsetof(struct xfs_btree_block, bb_u.l.bb_leftsib),
1556 offsetof(struct xfs_btree_block, bb_u.l.bb_rightsib),
1557 offsetof(struct xfs_btree_block, bb_u.l.bb_blkno),
1558 offsetof(struct xfs_btree_block, bb_u.l.bb_lsn),
1559 offsetof(struct xfs_btree_block, bb_u.l.bb_uuid),
1560 offsetof(struct xfs_btree_block, bb_u.l.bb_owner),
1561 offsetof(struct xfs_btree_block, bb_u.l.bb_crc),
1562 offsetof(struct xfs_btree_block, bb_u.l.bb_pad),
1563 XFS_BTREE_LBLOCK_CRC_LEN
1569 if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS) {
1571 * We don't log the CRC when updating a btree
1572 * block but instead recreate it during log
1573 * recovery. As the log buffers have checksums
1574 * of their own this is safe and avoids logging a crc
1575 * update in a lot of places.
1577 if (fields == XFS_BB_ALL_BITS)
1578 fields = XFS_BB_ALL_BITS_CRC;
1579 nbits = XFS_BB_NUM_BITS_CRC;
1581 nbits = XFS_BB_NUM_BITS;
1583 xfs_btree_offsets(fields,
1584 (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
1585 loffsets : soffsets,
1586 nbits, &first, &last);
1587 xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1588 xfs_trans_log_buf(cur->bc_tp, bp, first, last);
1590 xfs_trans_log_inode(cur->bc_tp, cur->bc_ino.ip,
1591 xfs_ilog_fbroot(cur->bc_ino.whichfork));
1596 * Increment cursor by one record at the level.
1597 * For nonzero levels the leaf-ward information is untouched.
1600 xfs_btree_increment(
1601 struct xfs_btree_cur *cur,
1603 int *stat) /* success/failure */
1605 struct xfs_btree_block *block;
1606 union xfs_btree_ptr ptr;
1608 int error; /* error return value */
1611 ASSERT(level < cur->bc_nlevels);
1613 /* Read-ahead to the right at this level. */
1614 xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
1616 /* Get a pointer to the btree block. */
1617 block = xfs_btree_get_block(cur, level, &bp);
1620 error = xfs_btree_check_block(cur, block, level, bp);
1625 /* We're done if we remain in the block after the increment. */
1626 if (++cur->bc_levels[level].ptr <= xfs_btree_get_numrecs(block))
1629 /* Fail if we just went off the right edge of the tree. */
1630 xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
1631 if (xfs_btree_ptr_is_null(cur, &ptr))
1634 XFS_BTREE_STATS_INC(cur, increment);
1637 * March up the tree incrementing pointers.
1638 * Stop when we don't go off the right edge of a block.
1640 for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
1641 block = xfs_btree_get_block(cur, lev, &bp);
1644 error = xfs_btree_check_block(cur, block, lev, bp);
1649 if (++cur->bc_levels[lev].ptr <= xfs_btree_get_numrecs(block))
1652 /* Read-ahead the right block for the next loop. */
1653 xfs_btree_readahead(cur, lev, XFS_BTCUR_RIGHTRA);
1657 * If we went off the root then we are either seriously
1658 * confused or have the tree root in an inode.
1660 if (lev == cur->bc_nlevels) {
1661 if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
1664 error = -EFSCORRUPTED;
1667 ASSERT(lev < cur->bc_nlevels);
1670 * Now walk back down the tree, fixing up the cursor's buffer
1671 * pointers and key numbers.
1673 for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
1674 union xfs_btree_ptr *ptrp;
1676 ptrp = xfs_btree_ptr_addr(cur, cur->bc_levels[lev].ptr, block);
1678 error = xfs_btree_read_buf_block(cur, ptrp, 0, &block, &bp);
1682 xfs_btree_setbuf(cur, lev, bp);
1683 cur->bc_levels[lev].ptr = 1;
1698 * Decrement cursor by one record at the level.
1699 * For nonzero levels the leaf-ward information is untouched.
1702 xfs_btree_decrement(
1703 struct xfs_btree_cur *cur,
1705 int *stat) /* success/failure */
1707 struct xfs_btree_block *block;
1709 int error; /* error return value */
1711 union xfs_btree_ptr ptr;
1713 ASSERT(level < cur->bc_nlevels);
1715 /* Read-ahead to the left at this level. */
1716 xfs_btree_readahead(cur, level, XFS_BTCUR_LEFTRA);
1718 /* We're done if we remain in the block after the decrement. */
1719 if (--cur->bc_levels[level].ptr > 0)
1722 /* Get a pointer to the btree block. */
1723 block = xfs_btree_get_block(cur, level, &bp);
1726 error = xfs_btree_check_block(cur, block, level, bp);
1731 /* Fail if we just went off the left edge of the tree. */
1732 xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
1733 if (xfs_btree_ptr_is_null(cur, &ptr))
1736 XFS_BTREE_STATS_INC(cur, decrement);
1739 * March up the tree decrementing pointers.
1740 * Stop when we don't go off the left edge of a block.
1742 for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
1743 if (--cur->bc_levels[lev].ptr > 0)
1745 /* Read-ahead the left block for the next loop. */
1746 xfs_btree_readahead(cur, lev, XFS_BTCUR_LEFTRA);
1750 * If we went off the root then we are seriously confused.
1751 * or the root of the tree is in an inode.
1753 if (lev == cur->bc_nlevels) {
1754 if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
1757 error = -EFSCORRUPTED;
1760 ASSERT(lev < cur->bc_nlevels);
1763 * Now walk back down the tree, fixing up the cursor's buffer
1764 * pointers and key numbers.
1766 for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
1767 union xfs_btree_ptr *ptrp;
1769 ptrp = xfs_btree_ptr_addr(cur, cur->bc_levels[lev].ptr, block);
1771 error = xfs_btree_read_buf_block(cur, ptrp, 0, &block, &bp);
1774 xfs_btree_setbuf(cur, lev, bp);
1775 cur->bc_levels[lev].ptr = xfs_btree_get_numrecs(block);
1790 xfs_btree_lookup_get_block(
1791 struct xfs_btree_cur *cur, /* btree cursor */
1792 int level, /* level in the btree */
1793 const union xfs_btree_ptr *pp, /* ptr to btree block */
1794 struct xfs_btree_block **blkp) /* return btree block */
1796 struct xfs_buf *bp; /* buffer pointer for btree block */
1800 /* special case the root block if in an inode */
1801 if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
1802 (level == cur->bc_nlevels - 1)) {
1803 *blkp = xfs_btree_get_iroot(cur);
1808 * If the old buffer at this level for the disk address we are
1809 * looking for re-use it.
1811 * Otherwise throw it away and get a new one.
1813 bp = cur->bc_levels[level].bp;
1814 error = xfs_btree_ptr_to_daddr(cur, pp, &daddr);
1817 if (bp && xfs_buf_daddr(bp) == daddr) {
1818 *blkp = XFS_BUF_TO_BLOCK(bp);
1822 error = xfs_btree_read_buf_block(cur, pp, 0, blkp, &bp);
1826 /* Check the inode owner since the verifiers don't. */
1827 if (xfs_has_crc(cur->bc_mp) &&
1828 !(cur->bc_ino.flags & XFS_BTCUR_BMBT_INVALID_OWNER) &&
1829 (cur->bc_flags & XFS_BTREE_LONG_PTRS) &&
1830 be64_to_cpu((*blkp)->bb_u.l.bb_owner) !=
1831 cur->bc_ino.ip->i_ino)
1834 /* Did we get the level we were looking for? */
1835 if (be16_to_cpu((*blkp)->bb_level) != level)
1838 /* Check that internal nodes have at least one record. */
1839 if (level != 0 && be16_to_cpu((*blkp)->bb_numrecs) == 0)
1842 xfs_btree_setbuf(cur, level, bp);
1847 xfs_buf_mark_corrupt(bp);
1848 xfs_trans_brelse(cur->bc_tp, bp);
1849 return -EFSCORRUPTED;
1853 * Get current search key. For level 0 we don't actually have a key
1854 * structure so we make one up from the record. For all other levels
1855 * we just return the right key.
1857 STATIC union xfs_btree_key *
1858 xfs_lookup_get_search_key(
1859 struct xfs_btree_cur *cur,
1862 struct xfs_btree_block *block,
1863 union xfs_btree_key *kp)
1866 cur->bc_ops->init_key_from_rec(kp,
1867 xfs_btree_rec_addr(cur, keyno, block));
1871 return xfs_btree_key_addr(cur, keyno, block);
1875 * Lookup the record. The cursor is made to point to it, based on dir.
1876 * stat is set to 0 if can't find any such record, 1 for success.
1880 struct xfs_btree_cur *cur, /* btree cursor */
1881 xfs_lookup_t dir, /* <=, ==, or >= */
1882 int *stat) /* success/failure */
1884 struct xfs_btree_block *block; /* current btree block */
1885 int64_t diff; /* difference for the current key */
1886 int error; /* error return value */
1887 int keyno; /* current key number */
1888 int level; /* level in the btree */
1889 union xfs_btree_ptr *pp; /* ptr to btree block */
1890 union xfs_btree_ptr ptr; /* ptr to btree block */
1892 XFS_BTREE_STATS_INC(cur, lookup);
1894 /* No such thing as a zero-level tree. */
1895 if (XFS_IS_CORRUPT(cur->bc_mp, cur->bc_nlevels == 0))
1896 return -EFSCORRUPTED;
1901 /* initialise start pointer from cursor */
1902 cur->bc_ops->init_ptr_from_cur(cur, &ptr);
1906 * Iterate over each level in the btree, starting at the root.
1907 * For each level above the leaves, find the key we need, based
1908 * on the lookup record, then follow the corresponding block
1909 * pointer down to the next level.
1911 for (level = cur->bc_nlevels - 1, diff = 1; level >= 0; level--) {
1912 /* Get the block we need to do the lookup on. */
1913 error = xfs_btree_lookup_get_block(cur, level, pp, &block);
1919 * If we already had a key match at a higher level, we
1920 * know we need to use the first entry in this block.
1924 /* Otherwise search this block. Do a binary search. */
1926 int high; /* high entry number */
1927 int low; /* low entry number */
1929 /* Set low and high entry numbers, 1-based. */
1931 high = xfs_btree_get_numrecs(block);
1933 /* Block is empty, must be an empty leaf. */
1934 if (level != 0 || cur->bc_nlevels != 1) {
1935 XFS_CORRUPTION_ERROR(__func__,
1939 return -EFSCORRUPTED;
1942 cur->bc_levels[0].ptr = dir != XFS_LOOKUP_LE;
1947 /* Binary search the block. */
1948 while (low <= high) {
1949 union xfs_btree_key key;
1950 union xfs_btree_key *kp;
1952 XFS_BTREE_STATS_INC(cur, compare);
1954 /* keyno is average of low and high. */
1955 keyno = (low + high) >> 1;
1957 /* Get current search key */
1958 kp = xfs_lookup_get_search_key(cur, level,
1959 keyno, block, &key);
1962 * Compute difference to get next direction:
1963 * - less than, move right
1964 * - greater than, move left
1965 * - equal, we're done
1967 diff = cur->bc_ops->key_diff(cur, kp);
1978 * If there are more levels, set up for the next level
1979 * by getting the block number and filling in the cursor.
1983 * If we moved left, need the previous key number,
1984 * unless there isn't one.
1986 if (diff > 0 && --keyno < 1)
1988 pp = xfs_btree_ptr_addr(cur, keyno, block);
1990 error = xfs_btree_debug_check_ptr(cur, pp, 0, level);
1994 cur->bc_levels[level].ptr = keyno;
1998 /* Done with the search. See if we need to adjust the results. */
1999 if (dir != XFS_LOOKUP_LE && diff < 0) {
2002 * If ge search and we went off the end of the block, but it's
2003 * not the last block, we're in the wrong block.
2005 xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
2006 if (dir == XFS_LOOKUP_GE &&
2007 keyno > xfs_btree_get_numrecs(block) &&
2008 !xfs_btree_ptr_is_null(cur, &ptr)) {
2011 cur->bc_levels[0].ptr = keyno;
2012 error = xfs_btree_increment(cur, 0, &i);
2015 if (XFS_IS_CORRUPT(cur->bc_mp, i != 1))
2016 return -EFSCORRUPTED;
2020 } else if (dir == XFS_LOOKUP_LE && diff > 0)
2022 cur->bc_levels[0].ptr = keyno;
2024 /* Return if we succeeded or not. */
2025 if (keyno == 0 || keyno > xfs_btree_get_numrecs(block))
2027 else if (dir != XFS_LOOKUP_EQ || diff == 0)
2037 /* Find the high key storage area from a regular key. */
2038 union xfs_btree_key *
2039 xfs_btree_high_key_from_key(
2040 struct xfs_btree_cur *cur,
2041 union xfs_btree_key *key)
2043 ASSERT(cur->bc_flags & XFS_BTREE_OVERLAPPING);
2044 return (union xfs_btree_key *)((char *)key +
2045 (cur->bc_ops->key_len / 2));
2048 /* Determine the low (and high if overlapped) keys of a leaf block */
2050 xfs_btree_get_leaf_keys(
2051 struct xfs_btree_cur *cur,
2052 struct xfs_btree_block *block,
2053 union xfs_btree_key *key)
2055 union xfs_btree_key max_hkey;
2056 union xfs_btree_key hkey;
2057 union xfs_btree_rec *rec;
2058 union xfs_btree_key *high;
2061 rec = xfs_btree_rec_addr(cur, 1, block);
2062 cur->bc_ops->init_key_from_rec(key, rec);
2064 if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2066 cur->bc_ops->init_high_key_from_rec(&max_hkey, rec);
2067 for (n = 2; n <= xfs_btree_get_numrecs(block); n++) {
2068 rec = xfs_btree_rec_addr(cur, n, block);
2069 cur->bc_ops->init_high_key_from_rec(&hkey, rec);
2070 if (xfs_btree_keycmp_gt(cur, &hkey, &max_hkey))
2074 high = xfs_btree_high_key_from_key(cur, key);
2075 memcpy(high, &max_hkey, cur->bc_ops->key_len / 2);
2079 /* Determine the low (and high if overlapped) keys of a node block */
2081 xfs_btree_get_node_keys(
2082 struct xfs_btree_cur *cur,
2083 struct xfs_btree_block *block,
2084 union xfs_btree_key *key)
2086 union xfs_btree_key *hkey;
2087 union xfs_btree_key *max_hkey;
2088 union xfs_btree_key *high;
2091 if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2092 memcpy(key, xfs_btree_key_addr(cur, 1, block),
2093 cur->bc_ops->key_len / 2);
2095 max_hkey = xfs_btree_high_key_addr(cur, 1, block);
2096 for (n = 2; n <= xfs_btree_get_numrecs(block); n++) {
2097 hkey = xfs_btree_high_key_addr(cur, n, block);
2098 if (xfs_btree_keycmp_gt(cur, hkey, max_hkey))
2102 high = xfs_btree_high_key_from_key(cur, key);
2103 memcpy(high, max_hkey, cur->bc_ops->key_len / 2);
2105 memcpy(key, xfs_btree_key_addr(cur, 1, block),
2106 cur->bc_ops->key_len);
2110 /* Derive the keys for any btree block. */
2113 struct xfs_btree_cur *cur,
2114 struct xfs_btree_block *block,
2115 union xfs_btree_key *key)
2117 if (be16_to_cpu(block->bb_level) == 0)
2118 xfs_btree_get_leaf_keys(cur, block, key);
2120 xfs_btree_get_node_keys(cur, block, key);
2124 * Decide if we need to update the parent keys of a btree block. For
2125 * a standard btree this is only necessary if we're updating the first
2126 * record/key. For an overlapping btree, we must always update the
2127 * keys because the highest key can be in any of the records or keys
2131 xfs_btree_needs_key_update(
2132 struct xfs_btree_cur *cur,
2135 return (cur->bc_flags & XFS_BTREE_OVERLAPPING) || ptr == 1;
2139 * Update the low and high parent keys of the given level, progressing
2140 * towards the root. If force_all is false, stop if the keys for a given
2141 * level do not need updating.
2144 __xfs_btree_updkeys(
2145 struct xfs_btree_cur *cur,
2147 struct xfs_btree_block *block,
2148 struct xfs_buf *bp0,
2151 union xfs_btree_key key; /* keys from current level */
2152 union xfs_btree_key *lkey; /* keys from the next level up */
2153 union xfs_btree_key *hkey;
2154 union xfs_btree_key *nlkey; /* keys from the next level up */
2155 union xfs_btree_key *nhkey;
2159 ASSERT(cur->bc_flags & XFS_BTREE_OVERLAPPING);
2161 /* Exit if there aren't any parent levels to update. */
2162 if (level + 1 >= cur->bc_nlevels)
2165 trace_xfs_btree_updkeys(cur, level, bp0);
2168 hkey = xfs_btree_high_key_from_key(cur, lkey);
2169 xfs_btree_get_keys(cur, block, lkey);
2170 for (level++; level < cur->bc_nlevels; level++) {
2174 block = xfs_btree_get_block(cur, level, &bp);
2175 trace_xfs_btree_updkeys(cur, level, bp);
2177 error = xfs_btree_check_block(cur, block, level, bp);
2181 ptr = cur->bc_levels[level].ptr;
2182 nlkey = xfs_btree_key_addr(cur, ptr, block);
2183 nhkey = xfs_btree_high_key_addr(cur, ptr, block);
2185 xfs_btree_keycmp_eq(cur, nlkey, lkey) &&
2186 xfs_btree_keycmp_eq(cur, nhkey, hkey))
2188 xfs_btree_copy_keys(cur, nlkey, lkey, 1);
2189 xfs_btree_log_keys(cur, bp, ptr, ptr);
2190 if (level + 1 >= cur->bc_nlevels)
2192 xfs_btree_get_node_keys(cur, block, lkey);
2198 /* Update all the keys from some level in cursor back to the root. */
2200 xfs_btree_updkeys_force(
2201 struct xfs_btree_cur *cur,
2205 struct xfs_btree_block *block;
2207 block = xfs_btree_get_block(cur, level, &bp);
2208 return __xfs_btree_updkeys(cur, level, block, bp, true);
2212 * Update the parent keys of the given level, progressing towards the root.
2215 xfs_btree_update_keys(
2216 struct xfs_btree_cur *cur,
2219 struct xfs_btree_block *block;
2221 union xfs_btree_key *kp;
2222 union xfs_btree_key key;
2227 block = xfs_btree_get_block(cur, level, &bp);
2228 if (cur->bc_flags & XFS_BTREE_OVERLAPPING)
2229 return __xfs_btree_updkeys(cur, level, block, bp, false);
2232 * Go up the tree from this level toward the root.
2233 * At each level, update the key value to the value input.
2234 * Stop when we reach a level where the cursor isn't pointing
2235 * at the first entry in the block.
2237 xfs_btree_get_keys(cur, block, &key);
2238 for (level++, ptr = 1; ptr == 1 && level < cur->bc_nlevels; level++) {
2242 block = xfs_btree_get_block(cur, level, &bp);
2244 error = xfs_btree_check_block(cur, block, level, bp);
2248 ptr = cur->bc_levels[level].ptr;
2249 kp = xfs_btree_key_addr(cur, ptr, block);
2250 xfs_btree_copy_keys(cur, kp, &key, 1);
2251 xfs_btree_log_keys(cur, bp, ptr, ptr);
2258 * Update the record referred to by cur to the value in the
2259 * given record. This either works (return 0) or gets an
2260 * EFSCORRUPTED error.
2264 struct xfs_btree_cur *cur,
2265 union xfs_btree_rec *rec)
2267 struct xfs_btree_block *block;
2271 union xfs_btree_rec *rp;
2273 /* Pick up the current block. */
2274 block = xfs_btree_get_block(cur, 0, &bp);
2277 error = xfs_btree_check_block(cur, block, 0, bp);
2281 /* Get the address of the rec to be updated. */
2282 ptr = cur->bc_levels[0].ptr;
2283 rp = xfs_btree_rec_addr(cur, ptr, block);
2285 /* Fill in the new contents and log them. */
2286 xfs_btree_copy_recs(cur, rp, rec, 1);
2287 xfs_btree_log_recs(cur, bp, ptr, ptr);
2290 * If we are tracking the last record in the tree and
2291 * we are at the far right edge of the tree, update it.
2293 if (xfs_btree_is_lastrec(cur, block, 0)) {
2294 cur->bc_ops->update_lastrec(cur, block, rec,
2295 ptr, LASTREC_UPDATE);
2298 /* Pass new key value up to our parent. */
2299 if (xfs_btree_needs_key_update(cur, ptr)) {
2300 error = xfs_btree_update_keys(cur, 0);
2312 * Move 1 record left from cur/level if possible.
2313 * Update cur to reflect the new path.
2315 STATIC int /* error */
2317 struct xfs_btree_cur *cur,
2319 int *stat) /* success/failure */
2321 struct xfs_buf *lbp; /* left buffer pointer */
2322 struct xfs_btree_block *left; /* left btree block */
2323 int lrecs; /* left record count */
2324 struct xfs_buf *rbp; /* right buffer pointer */
2325 struct xfs_btree_block *right; /* right btree block */
2326 struct xfs_btree_cur *tcur; /* temporary btree cursor */
2327 int rrecs; /* right record count */
2328 union xfs_btree_ptr lptr; /* left btree pointer */
2329 union xfs_btree_key *rkp = NULL; /* right btree key */
2330 union xfs_btree_ptr *rpp = NULL; /* right address pointer */
2331 union xfs_btree_rec *rrp = NULL; /* right record pointer */
2332 int error; /* error return value */
2335 if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
2336 level == cur->bc_nlevels - 1)
2339 /* Set up variables for this block as "right". */
2340 right = xfs_btree_get_block(cur, level, &rbp);
2343 error = xfs_btree_check_block(cur, right, level, rbp);
2348 /* If we've got no left sibling then we can't shift an entry left. */
2349 xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
2350 if (xfs_btree_ptr_is_null(cur, &lptr))
2354 * If the cursor entry is the one that would be moved, don't
2355 * do it... it's too complicated.
2357 if (cur->bc_levels[level].ptr <= 1)
2360 /* Set up the left neighbor as "left". */
2361 error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
2365 /* If it's full, it can't take another entry. */
2366 lrecs = xfs_btree_get_numrecs(left);
2367 if (lrecs == cur->bc_ops->get_maxrecs(cur, level))
2370 rrecs = xfs_btree_get_numrecs(right);
2373 * We add one entry to the left side and remove one for the right side.
2374 * Account for it here, the changes will be updated on disk and logged
2380 XFS_BTREE_STATS_INC(cur, lshift);
2381 XFS_BTREE_STATS_ADD(cur, moves, 1);
2384 * If non-leaf, copy a key and a ptr to the left block.
2385 * Log the changes to the left block.
2388 /* It's a non-leaf. Move keys and pointers. */
2389 union xfs_btree_key *lkp; /* left btree key */
2390 union xfs_btree_ptr *lpp; /* left address pointer */
2392 lkp = xfs_btree_key_addr(cur, lrecs, left);
2393 rkp = xfs_btree_key_addr(cur, 1, right);
2395 lpp = xfs_btree_ptr_addr(cur, lrecs, left);
2396 rpp = xfs_btree_ptr_addr(cur, 1, right);
2398 error = xfs_btree_debug_check_ptr(cur, rpp, 0, level);
2402 xfs_btree_copy_keys(cur, lkp, rkp, 1);
2403 xfs_btree_copy_ptrs(cur, lpp, rpp, 1);
2405 xfs_btree_log_keys(cur, lbp, lrecs, lrecs);
2406 xfs_btree_log_ptrs(cur, lbp, lrecs, lrecs);
2408 ASSERT(cur->bc_ops->keys_inorder(cur,
2409 xfs_btree_key_addr(cur, lrecs - 1, left), lkp));
2411 /* It's a leaf. Move records. */
2412 union xfs_btree_rec *lrp; /* left record pointer */
2414 lrp = xfs_btree_rec_addr(cur, lrecs, left);
2415 rrp = xfs_btree_rec_addr(cur, 1, right);
2417 xfs_btree_copy_recs(cur, lrp, rrp, 1);
2418 xfs_btree_log_recs(cur, lbp, lrecs, lrecs);
2420 ASSERT(cur->bc_ops->recs_inorder(cur,
2421 xfs_btree_rec_addr(cur, lrecs - 1, left), lrp));
2424 xfs_btree_set_numrecs(left, lrecs);
2425 xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
2427 xfs_btree_set_numrecs(right, rrecs);
2428 xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
2431 * Slide the contents of right down one entry.
2433 XFS_BTREE_STATS_ADD(cur, moves, rrecs - 1);
2435 /* It's a nonleaf. operate on keys and ptrs */
2436 for (i = 0; i < rrecs; i++) {
2437 error = xfs_btree_debug_check_ptr(cur, rpp, i + 1, level);
2442 xfs_btree_shift_keys(cur,
2443 xfs_btree_key_addr(cur, 2, right),
2445 xfs_btree_shift_ptrs(cur,
2446 xfs_btree_ptr_addr(cur, 2, right),
2449 xfs_btree_log_keys(cur, rbp, 1, rrecs);
2450 xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
2452 /* It's a leaf. operate on records */
2453 xfs_btree_shift_recs(cur,
2454 xfs_btree_rec_addr(cur, 2, right),
2456 xfs_btree_log_recs(cur, rbp, 1, rrecs);
2460 * Using a temporary cursor, update the parent key values of the
2461 * block on the left.
2463 if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2464 error = xfs_btree_dup_cursor(cur, &tcur);
2467 i = xfs_btree_firstrec(tcur, level);
2468 if (XFS_IS_CORRUPT(tcur->bc_mp, i != 1)) {
2469 error = -EFSCORRUPTED;
2473 error = xfs_btree_decrement(tcur, level, &i);
2477 /* Update the parent high keys of the left block, if needed. */
2478 error = xfs_btree_update_keys(tcur, level);
2482 xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
2485 /* Update the parent keys of the right block. */
2486 error = xfs_btree_update_keys(cur, level);
2490 /* Slide the cursor value left one. */
2491 cur->bc_levels[level].ptr--;
2504 xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
2509 * Move 1 record right from cur/level if possible.
2510 * Update cur to reflect the new path.
2512 STATIC int /* error */
2514 struct xfs_btree_cur *cur,
2516 int *stat) /* success/failure */
2518 struct xfs_buf *lbp; /* left buffer pointer */
2519 struct xfs_btree_block *left; /* left btree block */
2520 struct xfs_buf *rbp; /* right buffer pointer */
2521 struct xfs_btree_block *right; /* right btree block */
2522 struct xfs_btree_cur *tcur; /* temporary btree cursor */
2523 union xfs_btree_ptr rptr; /* right block pointer */
2524 union xfs_btree_key *rkp; /* right btree key */
2525 int rrecs; /* right record count */
2526 int lrecs; /* left record count */
2527 int error; /* error return value */
2528 int i; /* loop counter */
2530 if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
2531 (level == cur->bc_nlevels - 1))
2534 /* Set up variables for this block as "left". */
2535 left = xfs_btree_get_block(cur, level, &lbp);
2538 error = xfs_btree_check_block(cur, left, level, lbp);
2543 /* If we've got no right sibling then we can't shift an entry right. */
2544 xfs_btree_get_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
2545 if (xfs_btree_ptr_is_null(cur, &rptr))
2549 * If the cursor entry is the one that would be moved, don't
2550 * do it... it's too complicated.
2552 lrecs = xfs_btree_get_numrecs(left);
2553 if (cur->bc_levels[level].ptr >= lrecs)
2556 /* Set up the right neighbor as "right". */
2557 error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
2561 /* If it's full, it can't take another entry. */
2562 rrecs = xfs_btree_get_numrecs(right);
2563 if (rrecs == cur->bc_ops->get_maxrecs(cur, level))
2566 XFS_BTREE_STATS_INC(cur, rshift);
2567 XFS_BTREE_STATS_ADD(cur, moves, rrecs);
2570 * Make a hole at the start of the right neighbor block, then
2571 * copy the last left block entry to the hole.
2574 /* It's a nonleaf. make a hole in the keys and ptrs */
2575 union xfs_btree_key *lkp;
2576 union xfs_btree_ptr *lpp;
2577 union xfs_btree_ptr *rpp;
2579 lkp = xfs_btree_key_addr(cur, lrecs, left);
2580 lpp = xfs_btree_ptr_addr(cur, lrecs, left);
2581 rkp = xfs_btree_key_addr(cur, 1, right);
2582 rpp = xfs_btree_ptr_addr(cur, 1, right);
2584 for (i = rrecs - 1; i >= 0; i--) {
2585 error = xfs_btree_debug_check_ptr(cur, rpp, i, level);
2590 xfs_btree_shift_keys(cur, rkp, 1, rrecs);
2591 xfs_btree_shift_ptrs(cur, rpp, 1, rrecs);
2593 error = xfs_btree_debug_check_ptr(cur, lpp, 0, level);
2597 /* Now put the new data in, and log it. */
2598 xfs_btree_copy_keys(cur, rkp, lkp, 1);
2599 xfs_btree_copy_ptrs(cur, rpp, lpp, 1);
2601 xfs_btree_log_keys(cur, rbp, 1, rrecs + 1);
2602 xfs_btree_log_ptrs(cur, rbp, 1, rrecs + 1);
2604 ASSERT(cur->bc_ops->keys_inorder(cur, rkp,
2605 xfs_btree_key_addr(cur, 2, right)));
2607 /* It's a leaf. make a hole in the records */
2608 union xfs_btree_rec *lrp;
2609 union xfs_btree_rec *rrp;
2611 lrp = xfs_btree_rec_addr(cur, lrecs, left);
2612 rrp = xfs_btree_rec_addr(cur, 1, right);
2614 xfs_btree_shift_recs(cur, rrp, 1, rrecs);
2616 /* Now put the new data in, and log it. */
2617 xfs_btree_copy_recs(cur, rrp, lrp, 1);
2618 xfs_btree_log_recs(cur, rbp, 1, rrecs + 1);
2622 * Decrement and log left's numrecs, bump and log right's numrecs.
2624 xfs_btree_set_numrecs(left, --lrecs);
2625 xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
2627 xfs_btree_set_numrecs(right, ++rrecs);
2628 xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
2631 * Using a temporary cursor, update the parent key values of the
2632 * block on the right.
2634 error = xfs_btree_dup_cursor(cur, &tcur);
2637 i = xfs_btree_lastrec(tcur, level);
2638 if (XFS_IS_CORRUPT(tcur->bc_mp, i != 1)) {
2639 error = -EFSCORRUPTED;
2643 error = xfs_btree_increment(tcur, level, &i);
2647 /* Update the parent high keys of the left block, if needed. */
2648 if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2649 error = xfs_btree_update_keys(cur, level);
2654 /* Update the parent keys of the right block. */
2655 error = xfs_btree_update_keys(tcur, level);
2659 xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
2672 xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
2677 * Split cur/level block in half.
2678 * Return new block number and the key to its first
2679 * record (to be inserted into parent).
2681 STATIC int /* error */
2683 struct xfs_btree_cur *cur,
2685 union xfs_btree_ptr *ptrp,
2686 union xfs_btree_key *key,
2687 struct xfs_btree_cur **curp,
2688 int *stat) /* success/failure */
2690 union xfs_btree_ptr lptr; /* left sibling block ptr */
2691 struct xfs_buf *lbp; /* left buffer pointer */
2692 struct xfs_btree_block *left; /* left btree block */
2693 union xfs_btree_ptr rptr; /* right sibling block ptr */
2694 struct xfs_buf *rbp; /* right buffer pointer */
2695 struct xfs_btree_block *right; /* right btree block */
2696 union xfs_btree_ptr rrptr; /* right-right sibling ptr */
2697 struct xfs_buf *rrbp; /* right-right buffer pointer */
2698 struct xfs_btree_block *rrblock; /* right-right btree block */
2702 int error; /* error return value */
2705 XFS_BTREE_STATS_INC(cur, split);
2707 /* Set up left block (current one). */
2708 left = xfs_btree_get_block(cur, level, &lbp);
2711 error = xfs_btree_check_block(cur, left, level, lbp);
2716 xfs_btree_buf_to_ptr(cur, lbp, &lptr);
2718 /* Allocate the new block. If we can't do it, we're toast. Give up. */
2719 error = cur->bc_ops->alloc_block(cur, &lptr, &rptr, stat);
2724 XFS_BTREE_STATS_INC(cur, alloc);
2726 /* Set up the new block as "right". */
2727 error = xfs_btree_get_buf_block(cur, &rptr, &right, &rbp);
2731 /* Fill in the btree header for the new right block. */
2732 xfs_btree_init_block_cur(cur, rbp, xfs_btree_get_level(left), 0);
2735 * Split the entries between the old and the new block evenly.
2736 * Make sure that if there's an odd number of entries now, that
2737 * each new block will have the same number of entries.
2739 lrecs = xfs_btree_get_numrecs(left);
2741 if ((lrecs & 1) && cur->bc_levels[level].ptr <= rrecs + 1)
2743 src_index = (lrecs - rrecs + 1);
2745 XFS_BTREE_STATS_ADD(cur, moves, rrecs);
2747 /* Adjust numrecs for the later get_*_keys() calls. */
2749 xfs_btree_set_numrecs(left, lrecs);
2750 xfs_btree_set_numrecs(right, xfs_btree_get_numrecs(right) + rrecs);
2753 * Copy btree block entries from the left block over to the
2754 * new block, the right. Update the right block and log the
2758 /* It's a non-leaf. Move keys and pointers. */
2759 union xfs_btree_key *lkp; /* left btree key */
2760 union xfs_btree_ptr *lpp; /* left address pointer */
2761 union xfs_btree_key *rkp; /* right btree key */
2762 union xfs_btree_ptr *rpp; /* right address pointer */
2764 lkp = xfs_btree_key_addr(cur, src_index, left);
2765 lpp = xfs_btree_ptr_addr(cur, src_index, left);
2766 rkp = xfs_btree_key_addr(cur, 1, right);
2767 rpp = xfs_btree_ptr_addr(cur, 1, right);
2769 for (i = src_index; i < rrecs; i++) {
2770 error = xfs_btree_debug_check_ptr(cur, lpp, i, level);
2775 /* Copy the keys & pointers to the new block. */
2776 xfs_btree_copy_keys(cur, rkp, lkp, rrecs);
2777 xfs_btree_copy_ptrs(cur, rpp, lpp, rrecs);
2779 xfs_btree_log_keys(cur, rbp, 1, rrecs);
2780 xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
2782 /* Stash the keys of the new block for later insertion. */
2783 xfs_btree_get_node_keys(cur, right, key);
2785 /* It's a leaf. Move records. */
2786 union xfs_btree_rec *lrp; /* left record pointer */
2787 union xfs_btree_rec *rrp; /* right record pointer */
2789 lrp = xfs_btree_rec_addr(cur, src_index, left);
2790 rrp = xfs_btree_rec_addr(cur, 1, right);
2792 /* Copy records to the new block. */
2793 xfs_btree_copy_recs(cur, rrp, lrp, rrecs);
2794 xfs_btree_log_recs(cur, rbp, 1, rrecs);
2796 /* Stash the keys of the new block for later insertion. */
2797 xfs_btree_get_leaf_keys(cur, right, key);
2801 * Find the left block number by looking in the buffer.
2802 * Adjust sibling pointers.
2804 xfs_btree_get_sibling(cur, left, &rrptr, XFS_BB_RIGHTSIB);
2805 xfs_btree_set_sibling(cur, right, &rrptr, XFS_BB_RIGHTSIB);
2806 xfs_btree_set_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
2807 xfs_btree_set_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
2809 xfs_btree_log_block(cur, rbp, XFS_BB_ALL_BITS);
2810 xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
2813 * If there's a block to the new block's right, make that block
2814 * point back to right instead of to left.
2816 if (!xfs_btree_ptr_is_null(cur, &rrptr)) {
2817 error = xfs_btree_read_buf_block(cur, &rrptr,
2818 0, &rrblock, &rrbp);
2821 xfs_btree_set_sibling(cur, rrblock, &rptr, XFS_BB_LEFTSIB);
2822 xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
2825 /* Update the parent high keys of the left block, if needed. */
2826 if (cur->bc_flags & XFS_BTREE_OVERLAPPING) {
2827 error = xfs_btree_update_keys(cur, level);
2833 * If the cursor is really in the right block, move it there.
2834 * If it's just pointing past the last entry in left, then we'll
2835 * insert there, so don't change anything in that case.
2837 if (cur->bc_levels[level].ptr > lrecs + 1) {
2838 xfs_btree_setbuf(cur, level, rbp);
2839 cur->bc_levels[level].ptr -= lrecs;
2842 * If there are more levels, we'll need another cursor which refers
2843 * the right block, no matter where this cursor was.
2845 if (level + 1 < cur->bc_nlevels) {
2846 error = xfs_btree_dup_cursor(cur, curp);
2849 (*curp)->bc_levels[level + 1].ptr++;
2863 struct xfs_btree_split_args {
2864 struct xfs_btree_cur *cur;
2866 union xfs_btree_ptr *ptrp;
2867 union xfs_btree_key *key;
2868 struct xfs_btree_cur **curp;
2869 int *stat; /* success/failure */
2871 bool kswapd; /* allocation in kswapd context */
2872 struct completion *done;
2873 struct work_struct work;
2877 * Stack switching interfaces for allocation
2880 xfs_btree_split_worker(
2881 struct work_struct *work)
2883 struct xfs_btree_split_args *args = container_of(work,
2884 struct xfs_btree_split_args, work);
2885 unsigned long pflags;
2886 unsigned long new_pflags = 0;
2889 * we are in a transaction context here, but may also be doing work
2890 * in kswapd context, and hence we may need to inherit that state
2891 * temporarily to ensure that we don't block waiting for memory reclaim
2895 new_pflags |= PF_MEMALLOC | PF_KSWAPD;
2897 current_set_flags_nested(&pflags, new_pflags);
2898 xfs_trans_set_context(args->cur->bc_tp);
2900 args->result = __xfs_btree_split(args->cur, args->level, args->ptrp,
2901 args->key, args->curp, args->stat);
2903 xfs_trans_clear_context(args->cur->bc_tp);
2904 current_restore_flags_nested(&pflags, new_pflags);
2907 * Do not access args after complete() has run here. We don't own args
2908 * and the owner may run and free args before we return here.
2910 complete(args->done);
2915 * BMBT split requests often come in with little stack to work on so we push
2916 * them off to a worker thread so there is lots of stack to use. For the other
2917 * btree types, just call directly to avoid the context switch overhead here.
2919 * Care must be taken here - the work queue rescuer thread introduces potential
2920 * AGF <> worker queue deadlocks if the BMBT block allocation has to lock new
2921 * AGFs to allocate blocks. A task being run by the rescuer could attempt to
2922 * lock an AGF that is already locked by a task queued to run by the rescuer,
2923 * resulting in an ABBA deadlock as the rescuer cannot run the lock holder to
2924 * release it until the current thread it is running gains the lock.
2926 * To avoid this issue, we only ever queue BMBT splits that don't have an AGF
2927 * already locked to allocate from. The only place that doesn't hold an AGF
2928 * locked is unwritten extent conversion at IO completion, but that has already
2929 * been offloaded to a worker thread and hence has no stack consumption issues
2930 * we have to worry about.
2932 STATIC int /* error */
2934 struct xfs_btree_cur *cur,
2936 union xfs_btree_ptr *ptrp,
2937 union xfs_btree_key *key,
2938 struct xfs_btree_cur **curp,
2939 int *stat) /* success/failure */
2941 struct xfs_btree_split_args args;
2942 DECLARE_COMPLETION_ONSTACK(done);
2944 if (cur->bc_btnum != XFS_BTNUM_BMAP ||
2945 cur->bc_tp->t_highest_agno == NULLAGNUMBER)
2946 return __xfs_btree_split(cur, level, ptrp, key, curp, stat);
2955 args.kswapd = current_is_kswapd();
2956 INIT_WORK_ONSTACK(&args.work, xfs_btree_split_worker);
2957 queue_work(xfs_alloc_wq, &args.work);
2958 wait_for_completion(&done);
2959 destroy_work_on_stack(&args.work);
2963 #define xfs_btree_split __xfs_btree_split
2964 #endif /* __KERNEL__ */
2968 * Copy the old inode root contents into a real block and make the
2969 * broot point to it.
2972 xfs_btree_new_iroot(
2973 struct xfs_btree_cur *cur, /* btree cursor */
2974 int *logflags, /* logging flags for inode */
2975 int *stat) /* return status - 0 fail */
2977 struct xfs_buf *cbp; /* buffer for cblock */
2978 struct xfs_btree_block *block; /* btree block */
2979 struct xfs_btree_block *cblock; /* child btree block */
2980 union xfs_btree_key *ckp; /* child key pointer */
2981 union xfs_btree_ptr *cpp; /* child ptr pointer */
2982 union xfs_btree_key *kp; /* pointer to btree key */
2983 union xfs_btree_ptr *pp; /* pointer to block addr */
2984 union xfs_btree_ptr nptr; /* new block addr */
2985 int level; /* btree level */
2986 int error; /* error return code */
2987 int i; /* loop counter */
2989 XFS_BTREE_STATS_INC(cur, newroot);
2991 ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
2993 level = cur->bc_nlevels - 1;
2995 block = xfs_btree_get_iroot(cur);
2996 pp = xfs_btree_ptr_addr(cur, 1, block);
2998 /* Allocate the new block. If we can't do it, we're toast. Give up. */
2999 error = cur->bc_ops->alloc_block(cur, pp, &nptr, stat);
3005 XFS_BTREE_STATS_INC(cur, alloc);
3007 /* Copy the root into a real block. */
3008 error = xfs_btree_get_buf_block(cur, &nptr, &cblock, &cbp);
3013 * we can't just memcpy() the root in for CRC enabled btree blocks.
3014 * In that case have to also ensure the blkno remains correct
3016 memcpy(cblock, block, xfs_btree_block_len(cur));
3017 if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS) {
3018 __be64 bno = cpu_to_be64(xfs_buf_daddr(cbp));
3019 if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
3020 cblock->bb_u.l.bb_blkno = bno;
3022 cblock->bb_u.s.bb_blkno = bno;
3025 be16_add_cpu(&block->bb_level, 1);
3026 xfs_btree_set_numrecs(block, 1);
3028 ASSERT(cur->bc_nlevels <= cur->bc_maxlevels);
3029 cur->bc_levels[level + 1].ptr = 1;
3031 kp = xfs_btree_key_addr(cur, 1, block);
3032 ckp = xfs_btree_key_addr(cur, 1, cblock);
3033 xfs_btree_copy_keys(cur, ckp, kp, xfs_btree_get_numrecs(cblock));
3035 cpp = xfs_btree_ptr_addr(cur, 1, cblock);
3036 for (i = 0; i < be16_to_cpu(cblock->bb_numrecs); i++) {
3037 error = xfs_btree_debug_check_ptr(cur, pp, i, level);
3042 xfs_btree_copy_ptrs(cur, cpp, pp, xfs_btree_get_numrecs(cblock));
3044 error = xfs_btree_debug_check_ptr(cur, &nptr, 0, level);
3048 xfs_btree_copy_ptrs(cur, pp, &nptr, 1);
3050 xfs_iroot_realloc(cur->bc_ino.ip,
3051 1 - xfs_btree_get_numrecs(cblock),
3052 cur->bc_ino.whichfork);
3054 xfs_btree_setbuf(cur, level, cbp);
3057 * Do all this logging at the end so that
3058 * the root is at the right level.
3060 xfs_btree_log_block(cur, cbp, XFS_BB_ALL_BITS);
3061 xfs_btree_log_keys(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
3062 xfs_btree_log_ptrs(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
3065 XFS_ILOG_CORE | xfs_ilog_fbroot(cur->bc_ino.whichfork);
3073 * Allocate a new root block, fill it in.
3075 STATIC int /* error */
3077 struct xfs_btree_cur *cur, /* btree cursor */
3078 int *stat) /* success/failure */
3080 struct xfs_btree_block *block; /* one half of the old root block */
3081 struct xfs_buf *bp; /* buffer containing block */
3082 int error; /* error return value */
3083 struct xfs_buf *lbp; /* left buffer pointer */
3084 struct xfs_btree_block *left; /* left btree block */
3085 struct xfs_buf *nbp; /* new (root) buffer */
3086 struct xfs_btree_block *new; /* new (root) btree block */
3087 int nptr; /* new value for key index, 1 or 2 */
3088 struct xfs_buf *rbp; /* right buffer pointer */
3089 struct xfs_btree_block *right; /* right btree block */
3090 union xfs_btree_ptr rptr;
3091 union xfs_btree_ptr lptr;
3093 XFS_BTREE_STATS_INC(cur, newroot);
3095 /* initialise our start point from the cursor */
3096 cur->bc_ops->init_ptr_from_cur(cur, &rptr);
3098 /* Allocate the new block. If we can't do it, we're toast. Give up. */
3099 error = cur->bc_ops->alloc_block(cur, &rptr, &lptr, stat);
3104 XFS_BTREE_STATS_INC(cur, alloc);
3106 /* Set up the new block. */
3107 error = xfs_btree_get_buf_block(cur, &lptr, &new, &nbp);
3111 /* Set the root in the holding structure increasing the level by 1. */
3112 cur->bc_ops->set_root(cur, &lptr, 1);
3115 * At the previous root level there are now two blocks: the old root,
3116 * and the new block generated when it was split. We don't know which
3117 * one the cursor is pointing at, so we set up variables "left" and
3118 * "right" for each case.
3120 block = xfs_btree_get_block(cur, cur->bc_nlevels - 1, &bp);
3123 error = xfs_btree_check_block(cur, block, cur->bc_nlevels - 1, bp);
3128 xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
3129 if (!xfs_btree_ptr_is_null(cur, &rptr)) {
3130 /* Our block is left, pick up the right block. */
3132 xfs_btree_buf_to_ptr(cur, lbp, &lptr);
3134 error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
3140 /* Our block is right, pick up the left block. */
3142 xfs_btree_buf_to_ptr(cur, rbp, &rptr);
3144 xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
3145 error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
3152 /* Fill in the new block's btree header and log it. */
3153 xfs_btree_init_block_cur(cur, nbp, cur->bc_nlevels, 2);
3154 xfs_btree_log_block(cur, nbp, XFS_BB_ALL_BITS);
3155 ASSERT(!xfs_btree_ptr_is_null(cur, &lptr) &&
3156 !xfs_btree_ptr_is_null(cur, &rptr));
3158 /* Fill in the key data in the new root. */
3159 if (xfs_btree_get_level(left) > 0) {
3161 * Get the keys for the left block's keys and put them directly
3162 * in the parent block. Do the same for the right block.
3164 xfs_btree_get_node_keys(cur, left,
3165 xfs_btree_key_addr(cur, 1, new));
3166 xfs_btree_get_node_keys(cur, right,
3167 xfs_btree_key_addr(cur, 2, new));
3170 * Get the keys for the left block's records and put them
3171 * directly in the parent block. Do the same for the right
3174 xfs_btree_get_leaf_keys(cur, left,
3175 xfs_btree_key_addr(cur, 1, new));
3176 xfs_btree_get_leaf_keys(cur, right,
3177 xfs_btree_key_addr(cur, 2, new));
3179 xfs_btree_log_keys(cur, nbp, 1, 2);
3181 /* Fill in the pointer data in the new root. */
3182 xfs_btree_copy_ptrs(cur,
3183 xfs_btree_ptr_addr(cur, 1, new), &lptr, 1);
3184 xfs_btree_copy_ptrs(cur,
3185 xfs_btree_ptr_addr(cur, 2, new), &rptr, 1);
3186 xfs_btree_log_ptrs(cur, nbp, 1, 2);
3188 /* Fix up the cursor. */
3189 xfs_btree_setbuf(cur, cur->bc_nlevels, nbp);
3190 cur->bc_levels[cur->bc_nlevels].ptr = nptr;
3192 ASSERT(cur->bc_nlevels <= cur->bc_maxlevels);
3203 xfs_btree_make_block_unfull(
3204 struct xfs_btree_cur *cur, /* btree cursor */
3205 int level, /* btree level */
3206 int numrecs,/* # of recs in block */
3207 int *oindex,/* old tree index */
3208 int *index, /* new tree index */
3209 union xfs_btree_ptr *nptr, /* new btree ptr */
3210 struct xfs_btree_cur **ncur, /* new btree cursor */
3211 union xfs_btree_key *key, /* key of new block */
3216 if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
3217 level == cur->bc_nlevels - 1) {
3218 struct xfs_inode *ip = cur->bc_ino.ip;
3220 if (numrecs < cur->bc_ops->get_dmaxrecs(cur, level)) {
3221 /* A root block that can be made bigger. */
3222 xfs_iroot_realloc(ip, 1, cur->bc_ino.whichfork);
3225 /* A root block that needs replacing */
3228 error = xfs_btree_new_iroot(cur, &logflags, stat);
3229 if (error || *stat == 0)
3232 xfs_trans_log_inode(cur->bc_tp, ip, logflags);
3238 /* First, try shifting an entry to the right neighbor. */
3239 error = xfs_btree_rshift(cur, level, stat);
3243 /* Next, try shifting an entry to the left neighbor. */
3244 error = xfs_btree_lshift(cur, level, stat);
3249 *oindex = *index = cur->bc_levels[level].ptr;
3254 * Next, try splitting the current block in half.
3256 * If this works we have to re-set our variables because we
3257 * could be in a different block now.
3259 error = xfs_btree_split(cur, level, nptr, key, ncur, stat);
3260 if (error || *stat == 0)
3264 *index = cur->bc_levels[level].ptr;
3269 * Insert one record/level. Return information to the caller
3270 * allowing the next level up to proceed if necessary.
3274 struct xfs_btree_cur *cur, /* btree cursor */
3275 int level, /* level to insert record at */
3276 union xfs_btree_ptr *ptrp, /* i/o: block number inserted */
3277 union xfs_btree_rec *rec, /* record to insert */
3278 union xfs_btree_key *key, /* i/o: block key for ptrp */
3279 struct xfs_btree_cur **curp, /* output: new cursor replacing cur */
3280 int *stat) /* success/failure */
3282 struct xfs_btree_block *block; /* btree block */
3283 struct xfs_buf *bp; /* buffer for block */
3284 union xfs_btree_ptr nptr; /* new block ptr */
3285 struct xfs_btree_cur *ncur = NULL; /* new btree cursor */
3286 union xfs_btree_key nkey; /* new block key */
3287 union xfs_btree_key *lkey;
3288 int optr; /* old key/record index */
3289 int ptr; /* key/record index */
3290 int numrecs;/* number of records */
3291 int error; /* error return value */
3299 * If we have an external root pointer, and we've made it to the
3300 * root level, allocate a new root block and we're done.
3302 if (!(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
3303 (level >= cur->bc_nlevels)) {
3304 error = xfs_btree_new_root(cur, stat);
3305 xfs_btree_set_ptr_null(cur, ptrp);
3310 /* If we're off the left edge, return failure. */
3311 ptr = cur->bc_levels[level].ptr;
3319 XFS_BTREE_STATS_INC(cur, insrec);
3321 /* Get pointers to the btree buffer and block. */
3322 block = xfs_btree_get_block(cur, level, &bp);
3323 old_bn = bp ? xfs_buf_daddr(bp) : XFS_BUF_DADDR_NULL;
3324 numrecs = xfs_btree_get_numrecs(block);
3327 error = xfs_btree_check_block(cur, block, level, bp);
3331 /* Check that the new entry is being inserted in the right place. */
3332 if (ptr <= numrecs) {
3334 ASSERT(cur->bc_ops->recs_inorder(cur, rec,
3335 xfs_btree_rec_addr(cur, ptr, block)));
3337 ASSERT(cur->bc_ops->keys_inorder(cur, key,
3338 xfs_btree_key_addr(cur, ptr, block)));
3344 * If the block is full, we can't insert the new entry until we
3345 * make the block un-full.
3347 xfs_btree_set_ptr_null(cur, &nptr);
3348 if (numrecs == cur->bc_ops->get_maxrecs(cur, level)) {
3349 error = xfs_btree_make_block_unfull(cur, level, numrecs,
3350 &optr, &ptr, &nptr, &ncur, lkey, stat);
3351 if (error || *stat == 0)
3356 * The current block may have changed if the block was
3357 * previously full and we have just made space in it.
3359 block = xfs_btree_get_block(cur, level, &bp);
3360 numrecs = xfs_btree_get_numrecs(block);
3363 error = xfs_btree_check_block(cur, block, level, bp);
3369 * At this point we know there's room for our new entry in the block
3370 * we're pointing at.
3372 XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr + 1);
3375 /* It's a nonleaf. make a hole in the keys and ptrs */
3376 union xfs_btree_key *kp;
3377 union xfs_btree_ptr *pp;
3379 kp = xfs_btree_key_addr(cur, ptr, block);
3380 pp = xfs_btree_ptr_addr(cur, ptr, block);
3382 for (i = numrecs - ptr; i >= 0; i--) {
3383 error = xfs_btree_debug_check_ptr(cur, pp, i, level);
3388 xfs_btree_shift_keys(cur, kp, 1, numrecs - ptr + 1);
3389 xfs_btree_shift_ptrs(cur, pp, 1, numrecs - ptr + 1);
3391 error = xfs_btree_debug_check_ptr(cur, ptrp, 0, level);
3395 /* Now put the new data in, bump numrecs and log it. */
3396 xfs_btree_copy_keys(cur, kp, key, 1);
3397 xfs_btree_copy_ptrs(cur, pp, ptrp, 1);
3399 xfs_btree_set_numrecs(block, numrecs);
3400 xfs_btree_log_ptrs(cur, bp, ptr, numrecs);
3401 xfs_btree_log_keys(cur, bp, ptr, numrecs);
3403 if (ptr < numrecs) {
3404 ASSERT(cur->bc_ops->keys_inorder(cur, kp,
3405 xfs_btree_key_addr(cur, ptr + 1, block)));
3409 /* It's a leaf. make a hole in the records */
3410 union xfs_btree_rec *rp;
3412 rp = xfs_btree_rec_addr(cur, ptr, block);
3414 xfs_btree_shift_recs(cur, rp, 1, numrecs - ptr + 1);
3416 /* Now put the new data in, bump numrecs and log it. */
3417 xfs_btree_copy_recs(cur, rp, rec, 1);
3418 xfs_btree_set_numrecs(block, ++numrecs);
3419 xfs_btree_log_recs(cur, bp, ptr, numrecs);
3421 if (ptr < numrecs) {
3422 ASSERT(cur->bc_ops->recs_inorder(cur, rp,
3423 xfs_btree_rec_addr(cur, ptr + 1, block)));
3428 /* Log the new number of records in the btree header. */
3429 xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
3432 * If we just inserted into a new tree block, we have to
3433 * recalculate nkey here because nkey is out of date.
3435 * Otherwise we're just updating an existing block (having shoved
3436 * some records into the new tree block), so use the regular key
3439 if (bp && xfs_buf_daddr(bp) != old_bn) {
3440 xfs_btree_get_keys(cur, block, lkey);
3441 } else if (xfs_btree_needs_key_update(cur, optr)) {
3442 error = xfs_btree_update_keys(cur, level);
3448 * If we are tracking the last record in the tree and
3449 * we are at the far right edge of the tree, update it.
3451 if (xfs_btree_is_lastrec(cur, block, level)) {
3452 cur->bc_ops->update_lastrec(cur, block, rec,
3453 ptr, LASTREC_INSREC);
3457 * Return the new block number, if any.
3458 * If there is one, give back a record value and a cursor too.
3461 if (!xfs_btree_ptr_is_null(cur, &nptr)) {
3462 xfs_btree_copy_keys(cur, key, lkey, 1);
3471 xfs_btree_del_cursor(ncur, error);
3476 * Insert the record at the point referenced by cur.
3478 * A multi-level split of the tree on insert will invalidate the original
3479 * cursor. All callers of this function should assume that the cursor is
3480 * no longer valid and revalidate it.
3484 struct xfs_btree_cur *cur,
3487 int error; /* error return value */
3488 int i; /* result value, 0 for failure */
3489 int level; /* current level number in btree */
3490 union xfs_btree_ptr nptr; /* new block number (split result) */
3491 struct xfs_btree_cur *ncur; /* new cursor (split result) */
3492 struct xfs_btree_cur *pcur; /* previous level's cursor */
3493 union xfs_btree_key bkey; /* key of block to insert */
3494 union xfs_btree_key *key;
3495 union xfs_btree_rec rec; /* record to insert */
3502 xfs_btree_set_ptr_null(cur, &nptr);
3504 /* Make a key out of the record data to be inserted, and save it. */
3505 cur->bc_ops->init_rec_from_cur(cur, &rec);
3506 cur->bc_ops->init_key_from_rec(key, &rec);
3509 * Loop going up the tree, starting at the leaf level.
3510 * Stop when we don't get a split block, that must mean that
3511 * the insert is finished with this level.
3515 * Insert nrec/nptr into this level of the tree.
3516 * Note if we fail, nptr will be null.
3518 error = xfs_btree_insrec(pcur, level, &nptr, &rec, key,
3522 xfs_btree_del_cursor(pcur, XFS_BTREE_ERROR);
3526 if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
3527 error = -EFSCORRUPTED;
3533 * See if the cursor we just used is trash.
3534 * Can't trash the caller's cursor, but otherwise we should
3535 * if ncur is a new cursor or we're about to be done.
3538 (ncur || xfs_btree_ptr_is_null(cur, &nptr))) {
3539 /* Save the state from the cursor before we trash it */
3540 if (cur->bc_ops->update_cursor)
3541 cur->bc_ops->update_cursor(pcur, cur);
3542 cur->bc_nlevels = pcur->bc_nlevels;
3543 xfs_btree_del_cursor(pcur, XFS_BTREE_NOERROR);
3545 /* If we got a new cursor, switch to it. */
3550 } while (!xfs_btree_ptr_is_null(cur, &nptr));
3559 * Try to merge a non-leaf block back into the inode root.
3561 * Note: the killroot names comes from the fact that we're effectively
3562 * killing the old root block. But because we can't just delete the
3563 * inode we have to copy the single block it was pointing to into the
3567 xfs_btree_kill_iroot(
3568 struct xfs_btree_cur *cur)
3570 int whichfork = cur->bc_ino.whichfork;
3571 struct xfs_inode *ip = cur->bc_ino.ip;
3572 struct xfs_ifork *ifp = xfs_ifork_ptr(ip, whichfork);
3573 struct xfs_btree_block *block;
3574 struct xfs_btree_block *cblock;
3575 union xfs_btree_key *kp;
3576 union xfs_btree_key *ckp;
3577 union xfs_btree_ptr *pp;
3578 union xfs_btree_ptr *cpp;
3579 struct xfs_buf *cbp;
3585 union xfs_btree_ptr ptr;
3589 ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
3590 ASSERT(cur->bc_nlevels > 1);
3593 * Don't deal with the root block needs to be a leaf case.
3594 * We're just going to turn the thing back into extents anyway.
3596 level = cur->bc_nlevels - 1;
3601 * Give up if the root has multiple children.
3603 block = xfs_btree_get_iroot(cur);
3604 if (xfs_btree_get_numrecs(block) != 1)
3607 cblock = xfs_btree_get_block(cur, level - 1, &cbp);
3608 numrecs = xfs_btree_get_numrecs(cblock);
3611 * Only do this if the next level will fit.
3612 * Then the data must be copied up to the inode,
3613 * instead of freeing the root you free the next level.
3615 if (numrecs > cur->bc_ops->get_dmaxrecs(cur, level))
3618 XFS_BTREE_STATS_INC(cur, killroot);
3621 xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
3622 ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
3623 xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
3624 ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
3627 index = numrecs - cur->bc_ops->get_maxrecs(cur, level);
3629 xfs_iroot_realloc(cur->bc_ino.ip, index,
3630 cur->bc_ino.whichfork);
3631 block = ifp->if_broot;
3634 be16_add_cpu(&block->bb_numrecs, index);
3635 ASSERT(block->bb_numrecs == cblock->bb_numrecs);
3637 kp = xfs_btree_key_addr(cur, 1, block);
3638 ckp = xfs_btree_key_addr(cur, 1, cblock);
3639 xfs_btree_copy_keys(cur, kp, ckp, numrecs);
3641 pp = xfs_btree_ptr_addr(cur, 1, block);
3642 cpp = xfs_btree_ptr_addr(cur, 1, cblock);
3644 for (i = 0; i < numrecs; i++) {
3645 error = xfs_btree_debug_check_ptr(cur, cpp, i, level - 1);
3650 xfs_btree_copy_ptrs(cur, pp, cpp, numrecs);
3652 error = xfs_btree_free_block(cur, cbp);
3656 cur->bc_levels[level - 1].bp = NULL;
3657 be16_add_cpu(&block->bb_level, -1);
3658 xfs_trans_log_inode(cur->bc_tp, ip,
3659 XFS_ILOG_CORE | xfs_ilog_fbroot(cur->bc_ino.whichfork));
3666 * Kill the current root node, and replace it with it's only child node.
3669 xfs_btree_kill_root(
3670 struct xfs_btree_cur *cur,
3673 union xfs_btree_ptr *newroot)
3677 XFS_BTREE_STATS_INC(cur, killroot);
3680 * Update the root pointer, decreasing the level by 1 and then
3681 * free the old root.
3683 cur->bc_ops->set_root(cur, newroot, -1);
3685 error = xfs_btree_free_block(cur, bp);
3689 cur->bc_levels[level].bp = NULL;
3690 cur->bc_levels[level].ra = 0;
3697 xfs_btree_dec_cursor(
3698 struct xfs_btree_cur *cur,
3706 error = xfs_btree_decrement(cur, level, &i);
3716 * Single level of the btree record deletion routine.
3717 * Delete record pointed to by cur/level.
3718 * Remove the record from its block then rebalance the tree.
3719 * Return 0 for error, 1 for done, 2 to go on to the next level.
3721 STATIC int /* error */
3723 struct xfs_btree_cur *cur, /* btree cursor */
3724 int level, /* level removing record from */
3725 int *stat) /* fail/done/go-on */
3727 struct xfs_btree_block *block; /* btree block */
3728 union xfs_btree_ptr cptr; /* current block ptr */
3729 struct xfs_buf *bp; /* buffer for block */
3730 int error; /* error return value */
3731 int i; /* loop counter */
3732 union xfs_btree_ptr lptr; /* left sibling block ptr */
3733 struct xfs_buf *lbp; /* left buffer pointer */
3734 struct xfs_btree_block *left; /* left btree block */
3735 int lrecs = 0; /* left record count */
3736 int ptr; /* key/record index */
3737 union xfs_btree_ptr rptr; /* right sibling block ptr */
3738 struct xfs_buf *rbp; /* right buffer pointer */
3739 struct xfs_btree_block *right; /* right btree block */
3740 struct xfs_btree_block *rrblock; /* right-right btree block */
3741 struct xfs_buf *rrbp; /* right-right buffer pointer */
3742 int rrecs = 0; /* right record count */
3743 struct xfs_btree_cur *tcur; /* temporary btree cursor */
3744 int numrecs; /* temporary numrec count */
3748 /* Get the index of the entry being deleted, check for nothing there. */
3749 ptr = cur->bc_levels[level].ptr;
3755 /* Get the buffer & block containing the record or key/ptr. */
3756 block = xfs_btree_get_block(cur, level, &bp);
3757 numrecs = xfs_btree_get_numrecs(block);
3760 error = xfs_btree_check_block(cur, block, level, bp);
3765 /* Fail if we're off the end of the block. */
3766 if (ptr > numrecs) {
3771 XFS_BTREE_STATS_INC(cur, delrec);
3772 XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr);
3774 /* Excise the entries being deleted. */
3776 /* It's a nonleaf. operate on keys and ptrs */
3777 union xfs_btree_key *lkp;
3778 union xfs_btree_ptr *lpp;
3780 lkp = xfs_btree_key_addr(cur, ptr + 1, block);
3781 lpp = xfs_btree_ptr_addr(cur, ptr + 1, block);
3783 for (i = 0; i < numrecs - ptr; i++) {
3784 error = xfs_btree_debug_check_ptr(cur, lpp, i, level);
3789 if (ptr < numrecs) {
3790 xfs_btree_shift_keys(cur, lkp, -1, numrecs - ptr);
3791 xfs_btree_shift_ptrs(cur, lpp, -1, numrecs - ptr);
3792 xfs_btree_log_keys(cur, bp, ptr, numrecs - 1);
3793 xfs_btree_log_ptrs(cur, bp, ptr, numrecs - 1);
3796 /* It's a leaf. operate on records */
3797 if (ptr < numrecs) {
3798 xfs_btree_shift_recs(cur,
3799 xfs_btree_rec_addr(cur, ptr + 1, block),
3801 xfs_btree_log_recs(cur, bp, ptr, numrecs - 1);
3806 * Decrement and log the number of entries in the block.
3808 xfs_btree_set_numrecs(block, --numrecs);
3809 xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
3812 * If we are tracking the last record in the tree and
3813 * we are at the far right edge of the tree, update it.
3815 if (xfs_btree_is_lastrec(cur, block, level)) {
3816 cur->bc_ops->update_lastrec(cur, block, NULL,
3817 ptr, LASTREC_DELREC);
3821 * We're at the root level. First, shrink the root block in-memory.
3822 * Try to get rid of the next level down. If we can't then there's
3823 * nothing left to do.
3825 if (level == cur->bc_nlevels - 1) {
3826 if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
3827 xfs_iroot_realloc(cur->bc_ino.ip, -1,
3828 cur->bc_ino.whichfork);
3830 error = xfs_btree_kill_iroot(cur);
3834 error = xfs_btree_dec_cursor(cur, level, stat);
3842 * If this is the root level, and there's only one entry left,
3843 * and it's NOT the leaf level, then we can get rid of this
3846 if (numrecs == 1 && level > 0) {
3847 union xfs_btree_ptr *pp;
3849 * pp is still set to the first pointer in the block.
3850 * Make it the new root of the btree.
3852 pp = xfs_btree_ptr_addr(cur, 1, block);
3853 error = xfs_btree_kill_root(cur, bp, level, pp);
3856 } else if (level > 0) {
3857 error = xfs_btree_dec_cursor(cur, level, stat);
3866 * If we deleted the leftmost entry in the block, update the
3867 * key values above us in the tree.
3869 if (xfs_btree_needs_key_update(cur, ptr)) {
3870 error = xfs_btree_update_keys(cur, level);
3876 * If the number of records remaining in the block is at least
3877 * the minimum, we're done.
3879 if (numrecs >= cur->bc_ops->get_minrecs(cur, level)) {
3880 error = xfs_btree_dec_cursor(cur, level, stat);
3887 * Otherwise, we have to move some records around to keep the
3888 * tree balanced. Look at the left and right sibling blocks to
3889 * see if we can re-balance by moving only one record.
3891 xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
3892 xfs_btree_get_sibling(cur, block, &lptr, XFS_BB_LEFTSIB);
3894 if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
3896 * One child of root, need to get a chance to copy its contents
3897 * into the root and delete it. Can't go up to next level,
3898 * there's nothing to delete there.
3900 if (xfs_btree_ptr_is_null(cur, &rptr) &&
3901 xfs_btree_ptr_is_null(cur, &lptr) &&
3902 level == cur->bc_nlevels - 2) {
3903 error = xfs_btree_kill_iroot(cur);
3905 error = xfs_btree_dec_cursor(cur, level, stat);
3912 ASSERT(!xfs_btree_ptr_is_null(cur, &rptr) ||
3913 !xfs_btree_ptr_is_null(cur, &lptr));
3916 * Duplicate the cursor so our btree manipulations here won't
3917 * disrupt the next level up.
3919 error = xfs_btree_dup_cursor(cur, &tcur);
3924 * If there's a right sibling, see if it's ok to shift an entry
3927 if (!xfs_btree_ptr_is_null(cur, &rptr)) {
3929 * Move the temp cursor to the last entry in the next block.
3930 * Actually any entry but the first would suffice.
3932 i = xfs_btree_lastrec(tcur, level);
3933 if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
3934 error = -EFSCORRUPTED;
3938 error = xfs_btree_increment(tcur, level, &i);
3941 if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
3942 error = -EFSCORRUPTED;
3946 i = xfs_btree_lastrec(tcur, level);
3947 if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
3948 error = -EFSCORRUPTED;
3952 /* Grab a pointer to the block. */
3953 right = xfs_btree_get_block(tcur, level, &rbp);
3955 error = xfs_btree_check_block(tcur, right, level, rbp);
3959 /* Grab the current block number, for future use. */
3960 xfs_btree_get_sibling(tcur, right, &cptr, XFS_BB_LEFTSIB);
3963 * If right block is full enough so that removing one entry
3964 * won't make it too empty, and left-shifting an entry out
3965 * of right to us works, we're done.
3967 if (xfs_btree_get_numrecs(right) - 1 >=
3968 cur->bc_ops->get_minrecs(tcur, level)) {
3969 error = xfs_btree_lshift(tcur, level, &i);
3973 ASSERT(xfs_btree_get_numrecs(block) >=
3974 cur->bc_ops->get_minrecs(tcur, level));
3976 xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
3979 error = xfs_btree_dec_cursor(cur, level, stat);
3987 * Otherwise, grab the number of records in right for
3988 * future reference, and fix up the temp cursor to point
3989 * to our block again (last record).
3991 rrecs = xfs_btree_get_numrecs(right);
3992 if (!xfs_btree_ptr_is_null(cur, &lptr)) {
3993 i = xfs_btree_firstrec(tcur, level);
3994 if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
3995 error = -EFSCORRUPTED;
3999 error = xfs_btree_decrement(tcur, level, &i);
4002 if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
4003 error = -EFSCORRUPTED;
4010 * If there's a left sibling, see if it's ok to shift an entry
4013 if (!xfs_btree_ptr_is_null(cur, &lptr)) {
4015 * Move the temp cursor to the first entry in the
4018 i = xfs_btree_firstrec(tcur, level);
4019 if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
4020 error = -EFSCORRUPTED;
4024 error = xfs_btree_decrement(tcur, level, &i);
4027 i = xfs_btree_firstrec(tcur, level);
4028 if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
4029 error = -EFSCORRUPTED;
4033 /* Grab a pointer to the block. */
4034 left = xfs_btree_get_block(tcur, level, &lbp);
4036 error = xfs_btree_check_block(cur, left, level, lbp);
4040 /* Grab the current block number, for future use. */
4041 xfs_btree_get_sibling(tcur, left, &cptr, XFS_BB_RIGHTSIB);
4044 * If left block is full enough so that removing one entry
4045 * won't make it too empty, and right-shifting an entry out
4046 * of left to us works, we're done.
4048 if (xfs_btree_get_numrecs(left) - 1 >=
4049 cur->bc_ops->get_minrecs(tcur, level)) {
4050 error = xfs_btree_rshift(tcur, level, &i);
4054 ASSERT(xfs_btree_get_numrecs(block) >=
4055 cur->bc_ops->get_minrecs(tcur, level));
4056 xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
4059 cur->bc_levels[0].ptr++;
4067 * Otherwise, grab the number of records in right for
4070 lrecs = xfs_btree_get_numrecs(left);
4073 /* Delete the temp cursor, we're done with it. */
4074 xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
4077 /* If here, we need to do a join to keep the tree balanced. */
4078 ASSERT(!xfs_btree_ptr_is_null(cur, &cptr));
4080 if (!xfs_btree_ptr_is_null(cur, &lptr) &&
4081 lrecs + xfs_btree_get_numrecs(block) <=
4082 cur->bc_ops->get_maxrecs(cur, level)) {
4084 * Set "right" to be the starting block,
4085 * "left" to be the left neighbor.
4090 error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
4095 * If that won't work, see if we can join with the right neighbor block.
4097 } else if (!xfs_btree_ptr_is_null(cur, &rptr) &&
4098 rrecs + xfs_btree_get_numrecs(block) <=
4099 cur->bc_ops->get_maxrecs(cur, level)) {
4101 * Set "left" to be the starting block,
4102 * "right" to be the right neighbor.
4107 error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
4112 * Otherwise, we can't fix the imbalance.
4113 * Just return. This is probably a logic error, but it's not fatal.
4116 error = xfs_btree_dec_cursor(cur, level, stat);
4122 rrecs = xfs_btree_get_numrecs(right);
4123 lrecs = xfs_btree_get_numrecs(left);
4126 * We're now going to join "left" and "right" by moving all the stuff
4127 * in "right" to "left" and deleting "right".
4129 XFS_BTREE_STATS_ADD(cur, moves, rrecs);
4131 /* It's a non-leaf. Move keys and pointers. */
4132 union xfs_btree_key *lkp; /* left btree key */
4133 union xfs_btree_ptr *lpp; /* left address pointer */
4134 union xfs_btree_key *rkp; /* right btree key */
4135 union xfs_btree_ptr *rpp; /* right address pointer */
4137 lkp = xfs_btree_key_addr(cur, lrecs + 1, left);
4138 lpp = xfs_btree_ptr_addr(cur, lrecs + 1, left);
4139 rkp = xfs_btree_key_addr(cur, 1, right);
4140 rpp = xfs_btree_ptr_addr(cur, 1, right);
4142 for (i = 1; i < rrecs; i++) {
4143 error = xfs_btree_debug_check_ptr(cur, rpp, i, level);
4148 xfs_btree_copy_keys(cur, lkp, rkp, rrecs);
4149 xfs_btree_copy_ptrs(cur, lpp, rpp, rrecs);
4151 xfs_btree_log_keys(cur, lbp, lrecs + 1, lrecs + rrecs);
4152 xfs_btree_log_ptrs(cur, lbp, lrecs + 1, lrecs + rrecs);
4154 /* It's a leaf. Move records. */
4155 union xfs_btree_rec *lrp; /* left record pointer */
4156 union xfs_btree_rec *rrp; /* right record pointer */
4158 lrp = xfs_btree_rec_addr(cur, lrecs + 1, left);
4159 rrp = xfs_btree_rec_addr(cur, 1, right);
4161 xfs_btree_copy_recs(cur, lrp, rrp, rrecs);
4162 xfs_btree_log_recs(cur, lbp, lrecs + 1, lrecs + rrecs);
4165 XFS_BTREE_STATS_INC(cur, join);
4168 * Fix up the number of records and right block pointer in the
4169 * surviving block, and log it.
4171 xfs_btree_set_numrecs(left, lrecs + rrecs);
4172 xfs_btree_get_sibling(cur, right, &cptr, XFS_BB_RIGHTSIB);
4173 xfs_btree_set_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
4174 xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
4176 /* If there is a right sibling, point it to the remaining block. */
4177 xfs_btree_get_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
4178 if (!xfs_btree_ptr_is_null(cur, &cptr)) {
4179 error = xfs_btree_read_buf_block(cur, &cptr, 0, &rrblock, &rrbp);
4182 xfs_btree_set_sibling(cur, rrblock, &lptr, XFS_BB_LEFTSIB);
4183 xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
4186 /* Free the deleted block. */
4187 error = xfs_btree_free_block(cur, rbp);
4192 * If we joined with the left neighbor, set the buffer in the
4193 * cursor to the left block, and fix up the index.
4196 cur->bc_levels[level].bp = lbp;
4197 cur->bc_levels[level].ptr += lrecs;
4198 cur->bc_levels[level].ra = 0;
4201 * If we joined with the right neighbor and there's a level above
4202 * us, increment the cursor at that level.
4204 else if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) ||
4205 (level + 1 < cur->bc_nlevels)) {
4206 error = xfs_btree_increment(cur, level + 1, &i);
4212 * Readjust the ptr at this level if it's not a leaf, since it's
4213 * still pointing at the deletion point, which makes the cursor
4214 * inconsistent. If this makes the ptr 0, the caller fixes it up.
4215 * We can't use decrement because it would change the next level up.
4218 cur->bc_levels[level].ptr--;
4221 * We combined blocks, so we have to update the parent keys if the
4222 * btree supports overlapped intervals. However,
4223 * bc_levels[level + 1].ptr points to the old block so that the caller
4224 * knows which record to delete. Therefore, the caller must be savvy
4225 * enough to call updkeys for us if we return stat == 2. The other
4226 * exit points from this function don't require deletions further up
4227 * the tree, so they can call updkeys directly.
4230 /* Return value means the next level up has something to do. */
4236 xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
4241 * Delete the record pointed to by cur.
4242 * The cursor refers to the place where the record was (could be inserted)
4243 * when the operation returns.
4247 struct xfs_btree_cur *cur,
4248 int *stat) /* success/failure */
4250 int error; /* error return value */
4253 bool joined = false;
4256 * Go up the tree, starting at leaf level.
4258 * If 2 is returned then a join was done; go to the next level.
4259 * Otherwise we are done.
4261 for (level = 0, i = 2; i == 2; level++) {
4262 error = xfs_btree_delrec(cur, level, &i);
4270 * If we combined blocks as part of deleting the record, delrec won't
4271 * have updated the parent high keys so we have to do that here.
4273 if (joined && (cur->bc_flags & XFS_BTREE_OVERLAPPING)) {
4274 error = xfs_btree_updkeys_force(cur, 0);
4280 for (level = 1; level < cur->bc_nlevels; level++) {
4281 if (cur->bc_levels[level].ptr == 0) {
4282 error = xfs_btree_decrement(cur, level, &i);
4297 * Get the data from the pointed-to record.
4301 struct xfs_btree_cur *cur, /* btree cursor */
4302 union xfs_btree_rec **recp, /* output: btree record */
4303 int *stat) /* output: success/failure */
4305 struct xfs_btree_block *block; /* btree block */
4306 struct xfs_buf *bp; /* buffer pointer */
4307 int ptr; /* record number */
4309 int error; /* error return value */
4312 ptr = cur->bc_levels[0].ptr;
4313 block = xfs_btree_get_block(cur, 0, &bp);
4316 error = xfs_btree_check_block(cur, block, 0, bp);
4322 * Off the right end or left end, return failure.
4324 if (ptr > xfs_btree_get_numrecs(block) || ptr <= 0) {
4330 * Point to the record and extract its data.
4332 *recp = xfs_btree_rec_addr(cur, ptr, block);
4337 /* Visit a block in a btree. */
4339 xfs_btree_visit_block(
4340 struct xfs_btree_cur *cur,
4342 xfs_btree_visit_blocks_fn fn,
4345 struct xfs_btree_block *block;
4347 union xfs_btree_ptr rptr;
4350 /* do right sibling readahead */
4351 xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
4352 block = xfs_btree_get_block(cur, level, &bp);
4354 /* process the block */
4355 error = fn(cur, level, data);
4359 /* now read rh sibling block for next iteration */
4360 xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
4361 if (xfs_btree_ptr_is_null(cur, &rptr))
4365 * We only visit blocks once in this walk, so we have to avoid the
4366 * internal xfs_btree_lookup_get_block() optimisation where it will
4367 * return the same block without checking if the right sibling points
4368 * back to us and creates a cyclic reference in the btree.
4370 if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
4371 if (be64_to_cpu(rptr.l) == XFS_DADDR_TO_FSB(cur->bc_mp,
4373 return -EFSCORRUPTED;
4375 if (be32_to_cpu(rptr.s) == xfs_daddr_to_agbno(cur->bc_mp,
4377 return -EFSCORRUPTED;
4379 return xfs_btree_lookup_get_block(cur, level, &rptr, &block);
4383 /* Visit every block in a btree. */
4385 xfs_btree_visit_blocks(
4386 struct xfs_btree_cur *cur,
4387 xfs_btree_visit_blocks_fn fn,
4391 union xfs_btree_ptr lptr;
4393 struct xfs_btree_block *block = NULL;
4396 cur->bc_ops->init_ptr_from_cur(cur, &lptr);
4398 /* for each level */
4399 for (level = cur->bc_nlevels - 1; level >= 0; level--) {
4400 /* grab the left hand block */
4401 error = xfs_btree_lookup_get_block(cur, level, &lptr, &block);
4405 /* readahead the left most block for the next level down */
4407 union xfs_btree_ptr *ptr;
4409 ptr = xfs_btree_ptr_addr(cur, 1, block);
4410 xfs_btree_readahead_ptr(cur, ptr, 1);
4412 /* save for the next iteration of the loop */
4413 xfs_btree_copy_ptrs(cur, &lptr, ptr, 1);
4415 if (!(flags & XFS_BTREE_VISIT_LEAVES))
4417 } else if (!(flags & XFS_BTREE_VISIT_RECORDS)) {
4421 /* for each buffer in the level */
4423 error = xfs_btree_visit_block(cur, level, fn, data);
4426 if (error != -ENOENT)
4434 * Change the owner of a btree.
4436 * The mechanism we use here is ordered buffer logging. Because we don't know
4437 * how many buffers were are going to need to modify, we don't really want to
4438 * have to make transaction reservations for the worst case of every buffer in a
4439 * full size btree as that may be more space that we can fit in the log....
4441 * We do the btree walk in the most optimal manner possible - we have sibling
4442 * pointers so we can just walk all the blocks on each level from left to right
4443 * in a single pass, and then move to the next level and do the same. We can
4444 * also do readahead on the sibling pointers to get IO moving more quickly,
4445 * though for slow disks this is unlikely to make much difference to performance
4446 * as the amount of CPU work we have to do before moving to the next block is
4449 * For each btree block that we load, modify the owner appropriately, set the
4450 * buffer as an ordered buffer and log it appropriately. We need to ensure that
4451 * we mark the region we change dirty so that if the buffer is relogged in
4452 * a subsequent transaction the changes we make here as an ordered buffer are
4453 * correctly relogged in that transaction. If we are in recovery context, then
4454 * just queue the modified buffer as delayed write buffer so the transaction
4455 * recovery completion writes the changes to disk.
4457 struct xfs_btree_block_change_owner_info {
4459 struct list_head *buffer_list;
4463 xfs_btree_block_change_owner(
4464 struct xfs_btree_cur *cur,
4468 struct xfs_btree_block_change_owner_info *bbcoi = data;
4469 struct xfs_btree_block *block;
4472 /* modify the owner */
4473 block = xfs_btree_get_block(cur, level, &bp);
4474 if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
4475 if (block->bb_u.l.bb_owner == cpu_to_be64(bbcoi->new_owner))
4477 block->bb_u.l.bb_owner = cpu_to_be64(bbcoi->new_owner);
4479 if (block->bb_u.s.bb_owner == cpu_to_be32(bbcoi->new_owner))
4481 block->bb_u.s.bb_owner = cpu_to_be32(bbcoi->new_owner);
4485 * If the block is a root block hosted in an inode, we might not have a
4486 * buffer pointer here and we shouldn't attempt to log the change as the
4487 * information is already held in the inode and discarded when the root
4488 * block is formatted into the on-disk inode fork. We still change it,
4489 * though, so everything is consistent in memory.
4492 ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
4493 ASSERT(level == cur->bc_nlevels - 1);
4498 if (!xfs_trans_ordered_buf(cur->bc_tp, bp)) {
4499 xfs_btree_log_block(cur, bp, XFS_BB_OWNER);
4503 xfs_buf_delwri_queue(bp, bbcoi->buffer_list);
4510 xfs_btree_change_owner(
4511 struct xfs_btree_cur *cur,
4513 struct list_head *buffer_list)
4515 struct xfs_btree_block_change_owner_info bbcoi;
4517 bbcoi.new_owner = new_owner;
4518 bbcoi.buffer_list = buffer_list;
4520 return xfs_btree_visit_blocks(cur, xfs_btree_block_change_owner,
4521 XFS_BTREE_VISIT_ALL, &bbcoi);
4524 /* Verify the v5 fields of a long-format btree block. */
4526 xfs_btree_lblock_v5hdr_verify(
4530 struct xfs_mount *mp = bp->b_mount;
4531 struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
4533 if (!xfs_has_crc(mp))
4534 return __this_address;
4535 if (!uuid_equal(&block->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid))
4536 return __this_address;
4537 if (block->bb_u.l.bb_blkno != cpu_to_be64(xfs_buf_daddr(bp)))
4538 return __this_address;
4539 if (owner != XFS_RMAP_OWN_UNKNOWN &&
4540 be64_to_cpu(block->bb_u.l.bb_owner) != owner)
4541 return __this_address;
4545 /* Verify a long-format btree block. */
4547 xfs_btree_lblock_verify(
4549 unsigned int max_recs)
4551 struct xfs_mount *mp = bp->b_mount;
4552 struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
4556 /* numrecs verification */
4557 if (be16_to_cpu(block->bb_numrecs) > max_recs)
4558 return __this_address;
4560 /* sibling pointer verification */
4561 fsb = XFS_DADDR_TO_FSB(mp, xfs_buf_daddr(bp));
4562 fa = xfs_btree_check_lblock_siblings(mp, NULL, -1, fsb,
4563 block->bb_u.l.bb_leftsib);
4565 fa = xfs_btree_check_lblock_siblings(mp, NULL, -1, fsb,
4566 block->bb_u.l.bb_rightsib);
4571 * xfs_btree_sblock_v5hdr_verify() -- verify the v5 fields of a short-format
4574 * @bp: buffer containing the btree block
4577 xfs_btree_sblock_v5hdr_verify(
4580 struct xfs_mount *mp = bp->b_mount;
4581 struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
4582 struct xfs_perag *pag = bp->b_pag;
4584 if (!xfs_has_crc(mp))
4585 return __this_address;
4586 if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid))
4587 return __this_address;
4588 if (block->bb_u.s.bb_blkno != cpu_to_be64(xfs_buf_daddr(bp)))
4589 return __this_address;
4590 if (pag && be32_to_cpu(block->bb_u.s.bb_owner) != pag->pag_agno)
4591 return __this_address;
4596 * xfs_btree_sblock_verify() -- verify a short-format btree block
4598 * @bp: buffer containing the btree block
4599 * @max_recs: maximum records allowed in this btree node
4602 xfs_btree_sblock_verify(
4604 unsigned int max_recs)
4606 struct xfs_mount *mp = bp->b_mount;
4607 struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
4608 xfs_agblock_t agbno;
4611 /* numrecs verification */
4612 if (be16_to_cpu(block->bb_numrecs) > max_recs)
4613 return __this_address;
4615 /* sibling pointer verification */
4616 agbno = xfs_daddr_to_agbno(mp, xfs_buf_daddr(bp));
4617 fa = xfs_btree_check_sblock_siblings(bp->b_pag, NULL, -1, agbno,
4618 block->bb_u.s.bb_leftsib);
4620 fa = xfs_btree_check_sblock_siblings(bp->b_pag, NULL, -1, agbno,
4621 block->bb_u.s.bb_rightsib);
4626 * For the given limits on leaf and keyptr records per block, calculate the
4627 * height of the tree needed to index the number of leaf records.
4630 xfs_btree_compute_maxlevels(
4631 const unsigned int *limits,
4632 unsigned long long records)
4634 unsigned long long level_blocks = howmany_64(records, limits[0]);
4635 unsigned int height = 1;
4637 while (level_blocks > 1) {
4638 level_blocks = howmany_64(level_blocks, limits[1]);
4646 * For the given limits on leaf and keyptr records per block, calculate the
4647 * number of blocks needed to index the given number of leaf records.
4650 xfs_btree_calc_size(
4651 const unsigned int *limits,
4652 unsigned long long records)
4654 unsigned long long level_blocks = howmany_64(records, limits[0]);
4655 unsigned long long blocks = level_blocks;
4657 while (level_blocks > 1) {
4658 level_blocks = howmany_64(level_blocks, limits[1]);
4659 blocks += level_blocks;
4666 * Given a number of available blocks for the btree to consume with records and
4667 * pointers, calculate the height of the tree needed to index all the records
4668 * that space can hold based on the number of pointers each interior node
4671 * We start by assuming a single level tree consumes a single block, then track
4672 * the number of blocks each node level consumes until we no longer have space
4673 * to store the next node level. At this point, we are indexing all the leaf
4674 * blocks in the space, and there's no more free space to split the tree any
4675 * further. That's our maximum btree height.
4678 xfs_btree_space_to_height(
4679 const unsigned int *limits,
4680 unsigned long long leaf_blocks)
4683 * The root btree block can have fewer than minrecs pointers in it
4684 * because the tree might not be big enough to require that amount of
4685 * fanout. Hence it has a minimum size of 2 pointers, not limits[1].
4687 unsigned long long node_blocks = 2;
4688 unsigned long long blocks_left = leaf_blocks - 1;
4689 unsigned int height = 1;
4691 if (leaf_blocks < 1)
4694 while (node_blocks < blocks_left) {
4695 blocks_left -= node_blocks;
4696 node_blocks *= limits[1];
4704 * Query a regular btree for all records overlapping a given interval.
4705 * Start with a LE lookup of the key of low_rec and return all records
4706 * until we find a record with a key greater than the key of high_rec.
4709 xfs_btree_simple_query_range(
4710 struct xfs_btree_cur *cur,
4711 const union xfs_btree_key *low_key,
4712 const union xfs_btree_key *high_key,
4713 xfs_btree_query_range_fn fn,
4716 union xfs_btree_rec *recp;
4717 union xfs_btree_key rec_key;
4719 bool firstrec = true;
4722 ASSERT(cur->bc_ops->init_high_key_from_rec);
4723 ASSERT(cur->bc_ops->diff_two_keys);
4726 * Find the leftmost record. The btree cursor must be set
4727 * to the low record used to generate low_key.
4730 error = xfs_btree_lookup(cur, XFS_LOOKUP_LE, &stat);
4734 /* Nothing? See if there's anything to the right. */
4736 error = xfs_btree_increment(cur, 0, &stat);
4742 /* Find the record. */
4743 error = xfs_btree_get_rec(cur, &recp, &stat);
4747 /* Skip if low_key > high_key(rec). */
4749 cur->bc_ops->init_high_key_from_rec(&rec_key, recp);
4751 if (xfs_btree_keycmp_gt(cur, low_key, &rec_key))
4755 /* Stop if low_key(rec) > high_key. */
4756 cur->bc_ops->init_key_from_rec(&rec_key, recp);
4757 if (xfs_btree_keycmp_gt(cur, &rec_key, high_key))
4761 error = fn(cur, recp, priv);
4766 /* Move on to the next record. */
4767 error = xfs_btree_increment(cur, 0, &stat);
4777 * Query an overlapped interval btree for all records overlapping a given
4778 * interval. This function roughly follows the algorithm given in
4779 * "Interval Trees" of _Introduction to Algorithms_, which is section
4780 * 14.3 in the 2nd and 3rd editions.
4782 * First, generate keys for the low and high records passed in.
4784 * For any leaf node, generate the high and low keys for the record.
4785 * If the record keys overlap with the query low/high keys, pass the
4786 * record to the function iterator.
4788 * For any internal node, compare the low and high keys of each
4789 * pointer against the query low/high keys. If there's an overlap,
4790 * follow the pointer.
4792 * As an optimization, we stop scanning a block when we find a low key
4793 * that is greater than the query's high key.
4796 xfs_btree_overlapped_query_range(
4797 struct xfs_btree_cur *cur,
4798 const union xfs_btree_key *low_key,
4799 const union xfs_btree_key *high_key,
4800 xfs_btree_query_range_fn fn,
4803 union xfs_btree_ptr ptr;
4804 union xfs_btree_ptr *pp;
4805 union xfs_btree_key rec_key;
4806 union xfs_btree_key rec_hkey;
4807 union xfs_btree_key *lkp;
4808 union xfs_btree_key *hkp;
4809 union xfs_btree_rec *recp;
4810 struct xfs_btree_block *block;
4816 /* Load the root of the btree. */
4817 level = cur->bc_nlevels - 1;
4818 cur->bc_ops->init_ptr_from_cur(cur, &ptr);
4819 error = xfs_btree_lookup_get_block(cur, level, &ptr, &block);
4822 xfs_btree_get_block(cur, level, &bp);
4823 trace_xfs_btree_overlapped_query_range(cur, level, bp);
4825 error = xfs_btree_check_block(cur, block, level, bp);
4829 cur->bc_levels[level].ptr = 1;
4831 while (level < cur->bc_nlevels) {
4832 block = xfs_btree_get_block(cur, level, &bp);
4834 /* End of node, pop back towards the root. */
4835 if (cur->bc_levels[level].ptr >
4836 be16_to_cpu(block->bb_numrecs)) {
4838 if (level < cur->bc_nlevels - 1)
4839 cur->bc_levels[level + 1].ptr++;
4845 /* Handle a leaf node. */
4846 recp = xfs_btree_rec_addr(cur, cur->bc_levels[0].ptr,
4849 cur->bc_ops->init_high_key_from_rec(&rec_hkey, recp);
4850 cur->bc_ops->init_key_from_rec(&rec_key, recp);
4853 * If (query's high key < record's low key), then there
4854 * are no more interesting records in this block. Pop
4855 * up to the leaf level to find more record blocks.
4857 * If (record's high key >= query's low key) and
4858 * (query's high key >= record's low key), then
4859 * this record overlaps the query range; callback.
4861 if (xfs_btree_keycmp_lt(cur, high_key, &rec_key))
4863 if (xfs_btree_keycmp_ge(cur, &rec_hkey, low_key)) {
4864 error = fn(cur, recp, priv);
4868 cur->bc_levels[level].ptr++;
4872 /* Handle an internal node. */
4873 lkp = xfs_btree_key_addr(cur, cur->bc_levels[level].ptr, block);
4874 hkp = xfs_btree_high_key_addr(cur, cur->bc_levels[level].ptr,
4876 pp = xfs_btree_ptr_addr(cur, cur->bc_levels[level].ptr, block);
4879 * If (query's high key < pointer's low key), then there are no
4880 * more interesting keys in this block. Pop up one leaf level
4881 * to continue looking for records.
4883 * If (pointer's high key >= query's low key) and
4884 * (query's high key >= pointer's low key), then
4885 * this record overlaps the query range; follow pointer.
4887 if (xfs_btree_keycmp_lt(cur, high_key, lkp))
4889 if (xfs_btree_keycmp_ge(cur, hkp, low_key)) {
4891 error = xfs_btree_lookup_get_block(cur, level, pp,
4895 xfs_btree_get_block(cur, level, &bp);
4896 trace_xfs_btree_overlapped_query_range(cur, level, bp);
4898 error = xfs_btree_check_block(cur, block, level, bp);
4902 cur->bc_levels[level].ptr = 1;
4905 cur->bc_levels[level].ptr++;
4910 * If we don't end this function with the cursor pointing at a record
4911 * block, a subsequent non-error cursor deletion will not release
4912 * node-level buffers, causing a buffer leak. This is quite possible
4913 * with a zero-results range query, so release the buffers if we
4914 * failed to return any results.
4916 if (cur->bc_levels[0].bp == NULL) {
4917 for (i = 0; i < cur->bc_nlevels; i++) {
4918 if (cur->bc_levels[i].bp) {
4919 xfs_trans_brelse(cur->bc_tp,
4920 cur->bc_levels[i].bp);
4921 cur->bc_levels[i].bp = NULL;
4922 cur->bc_levels[i].ptr = 0;
4923 cur->bc_levels[i].ra = 0;
4932 xfs_btree_key_from_irec(
4933 struct xfs_btree_cur *cur,
4934 union xfs_btree_key *key,
4935 const union xfs_btree_irec *irec)
4937 union xfs_btree_rec rec;
4939 cur->bc_rec = *irec;
4940 cur->bc_ops->init_rec_from_cur(cur, &rec);
4941 cur->bc_ops->init_key_from_rec(key, &rec);
4945 * Query a btree for all records overlapping a given interval of keys. The
4946 * supplied function will be called with each record found; return one of the
4947 * XFS_BTREE_QUERY_RANGE_{CONTINUE,ABORT} values or the usual negative error
4948 * code. This function returns -ECANCELED, zero, or a negative error code.
4951 xfs_btree_query_range(
4952 struct xfs_btree_cur *cur,
4953 const union xfs_btree_irec *low_rec,
4954 const union xfs_btree_irec *high_rec,
4955 xfs_btree_query_range_fn fn,
4958 union xfs_btree_key low_key;
4959 union xfs_btree_key high_key;
4961 /* Find the keys of both ends of the interval. */
4962 xfs_btree_key_from_irec(cur, &high_key, high_rec);
4963 xfs_btree_key_from_irec(cur, &low_key, low_rec);
4965 /* Enforce low key <= high key. */
4966 if (!xfs_btree_keycmp_le(cur, &low_key, &high_key))
4969 if (!(cur->bc_flags & XFS_BTREE_OVERLAPPING))
4970 return xfs_btree_simple_query_range(cur, &low_key,
4971 &high_key, fn, priv);
4972 return xfs_btree_overlapped_query_range(cur, &low_key, &high_key,
4976 /* Query a btree for all records. */
4978 xfs_btree_query_all(
4979 struct xfs_btree_cur *cur,
4980 xfs_btree_query_range_fn fn,
4983 union xfs_btree_key low_key;
4984 union xfs_btree_key high_key;
4986 memset(&cur->bc_rec, 0, sizeof(cur->bc_rec));
4987 memset(&low_key, 0, sizeof(low_key));
4988 memset(&high_key, 0xFF, sizeof(high_key));
4990 return xfs_btree_simple_query_range(cur, &low_key, &high_key, fn, priv);
4994 xfs_btree_count_blocks_helper(
4995 struct xfs_btree_cur *cur,
4999 xfs_extlen_t *blocks = data;
5005 /* Count the blocks in a btree and return the result in *blocks. */
5007 xfs_btree_count_blocks(
5008 struct xfs_btree_cur *cur,
5009 xfs_extlen_t *blocks)
5012 return xfs_btree_visit_blocks(cur, xfs_btree_count_blocks_helper,
5013 XFS_BTREE_VISIT_ALL, blocks);
5016 /* Compare two btree pointers. */
5018 xfs_btree_diff_two_ptrs(
5019 struct xfs_btree_cur *cur,
5020 const union xfs_btree_ptr *a,
5021 const union xfs_btree_ptr *b)
5023 if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
5024 return (int64_t)be64_to_cpu(a->l) - be64_to_cpu(b->l);
5025 return (int64_t)be32_to_cpu(a->s) - be32_to_cpu(b->s);
5028 struct xfs_btree_has_records {
5029 /* Keys for the start and end of the range we want to know about. */
5030 union xfs_btree_key start_key;
5031 union xfs_btree_key end_key;
5033 /* Mask for key comparisons, if desired. */
5034 const union xfs_btree_key *key_mask;
5036 /* Highest record key we've seen so far. */
5037 union xfs_btree_key high_key;
5039 enum xbtree_recpacking outcome;
5043 xfs_btree_has_records_helper(
5044 struct xfs_btree_cur *cur,
5045 const union xfs_btree_rec *rec,
5048 union xfs_btree_key rec_key;
5049 union xfs_btree_key rec_high_key;
5050 struct xfs_btree_has_records *info = priv;
5051 enum xbtree_key_contig key_contig;
5053 cur->bc_ops->init_key_from_rec(&rec_key, rec);
5055 if (info->outcome == XBTREE_RECPACKING_EMPTY) {
5056 info->outcome = XBTREE_RECPACKING_SPARSE;
5059 * If the first record we find does not overlap the start key,
5060 * then there is a hole at the start of the search range.
5061 * Classify this as sparse and stop immediately.
5063 if (xfs_btree_masked_keycmp_lt(cur, &info->start_key, &rec_key,
5068 * If a subsequent record does not overlap with the any record
5069 * we've seen so far, there is a hole in the middle of the
5070 * search range. Classify this as sparse and stop.
5071 * If the keys overlap and this btree does not allow overlap,
5072 * signal corruption.
5074 key_contig = cur->bc_ops->keys_contiguous(cur, &info->high_key,
5075 &rec_key, info->key_mask);
5076 if (key_contig == XBTREE_KEY_OVERLAP &&
5077 !(cur->bc_flags & XFS_BTREE_OVERLAPPING))
5078 return -EFSCORRUPTED;
5079 if (key_contig == XBTREE_KEY_GAP)
5084 * If high_key(rec) is larger than any other high key we've seen,
5085 * remember it for later.
5087 cur->bc_ops->init_high_key_from_rec(&rec_high_key, rec);
5088 if (xfs_btree_masked_keycmp_gt(cur, &rec_high_key, &info->high_key,
5090 info->high_key = rec_high_key; /* struct copy */
5096 * Scan part of the keyspace of a btree and tell us if that keyspace does not
5097 * map to any records; is fully mapped to records; or is partially mapped to
5098 * records. This is the btree record equivalent to determining if a file is
5101 * For most btree types, the record scan should use all available btree key
5102 * fields to compare the keys encountered. These callers should pass NULL for
5103 * @mask. However, some callers (e.g. scanning physical space in the rmapbt)
5104 * want to ignore some part of the btree record keyspace when performing the
5105 * comparison. These callers should pass in a union xfs_btree_key object with
5106 * the fields that *should* be a part of the comparison set to any nonzero
5107 * value, and the rest zeroed.
5110 xfs_btree_has_records(
5111 struct xfs_btree_cur *cur,
5112 const union xfs_btree_irec *low,
5113 const union xfs_btree_irec *high,
5114 const union xfs_btree_key *mask,
5115 enum xbtree_recpacking *outcome)
5117 struct xfs_btree_has_records info = {
5118 .outcome = XBTREE_RECPACKING_EMPTY,
5123 /* Not all btrees support this operation. */
5124 if (!cur->bc_ops->keys_contiguous) {
5129 xfs_btree_key_from_irec(cur, &info.start_key, low);
5130 xfs_btree_key_from_irec(cur, &info.end_key, high);
5132 error = xfs_btree_query_range(cur, low, high,
5133 xfs_btree_has_records_helper, &info);
5134 if (error == -ECANCELED)
5139 if (info.outcome == XBTREE_RECPACKING_EMPTY)
5143 * If the largest high_key(rec) we saw during the walk is greater than
5144 * the end of the search range, classify this as full. Otherwise,
5145 * there is a hole at the end of the search range.
5147 if (xfs_btree_masked_keycmp_ge(cur, &info.high_key, &info.end_key,
5149 info.outcome = XBTREE_RECPACKING_FULL;
5152 *outcome = info.outcome;
5156 /* Are there more records in this btree? */
5158 xfs_btree_has_more_records(
5159 struct xfs_btree_cur *cur)
5161 struct xfs_btree_block *block;
5164 block = xfs_btree_get_block(cur, 0, &bp);
5166 /* There are still records in this block. */
5167 if (cur->bc_levels[0].ptr < xfs_btree_get_numrecs(block))
5170 /* There are more record blocks. */
5171 if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
5172 return block->bb_u.l.bb_rightsib != cpu_to_be64(NULLFSBLOCK);
5174 return block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK);
5177 /* Set up all the btree cursor caches. */
5179 xfs_btree_init_cur_caches(void)
5183 error = xfs_allocbt_init_cur_cache();
5186 error = xfs_inobt_init_cur_cache();
5189 error = xfs_bmbt_init_cur_cache();
5192 error = xfs_rmapbt_init_cur_cache();
5195 error = xfs_refcountbt_init_cur_cache();
5201 xfs_btree_destroy_cur_caches();
5205 /* Destroy all the btree cursor caches, if they've been allocated. */
5207 xfs_btree_destroy_cur_caches(void)
5209 xfs_allocbt_destroy_cur_cache();
5210 xfs_inobt_destroy_cur_cache();
5211 xfs_bmbt_destroy_cur_cache();
5212 xfs_rmapbt_destroy_cur_cache();
5213 xfs_refcountbt_destroy_cur_cache();
5216 /* Move the btree cursor before the first record. */
5218 xfs_btree_goto_left_edge(
5219 struct xfs_btree_cur *cur)
5224 memset(&cur->bc_rec, 0, sizeof(cur->bc_rec));
5225 error = xfs_btree_lookup(cur, XFS_LOOKUP_LE, &stat);
5231 error = xfs_btree_decrement(cur, 0, &stat);
5236 return -EFSCORRUPTED;