xref: /linux/fs/xfs/libxfs/xfs_btree.c (revision a1ff5a7d78a036d6c2178ee5acd6ba4946243800)
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
4  * All Rights Reserved.
5  */
6 #include "xfs.h"
7 #include "xfs_fs.h"
8 #include "xfs_shared.h"
9 #include "xfs_format.h"
10 #include "xfs_log_format.h"
11 #include "xfs_trans_resv.h"
12 #include "xfs_bit.h"
13 #include "xfs_mount.h"
14 #include "xfs_inode.h"
15 #include "xfs_trans.h"
16 #include "xfs_buf_item.h"
17 #include "xfs_btree.h"
18 #include "xfs_errortag.h"
19 #include "xfs_error.h"
20 #include "xfs_trace.h"
21 #include "xfs_alloc.h"
22 #include "xfs_log.h"
23 #include "xfs_btree_staging.h"
24 #include "xfs_ag.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"
30 #include "xfs_health.h"
31 #include "xfs_buf_mem.h"
32 #include "xfs_btree_mem.h"
33 
34 /*
35  * Btree magic numbers.
36  */
37 uint32_t
xfs_btree_magic(struct xfs_mount * mp,const struct xfs_btree_ops * ops)38 xfs_btree_magic(
39 	struct xfs_mount		*mp,
40 	const struct xfs_btree_ops	*ops)
41 {
42 	int				idx = xfs_has_crc(mp) ? 1 : 0;
43 	__be32				magic = ops->buf_ops->magic[idx];
44 
45 	/* Ensure we asked for crc for crc-only magics. */
46 	ASSERT(magic != 0);
47 	return be32_to_cpu(magic);
48 }
49 
50 /*
51  * These sibling pointer checks are optimised for null sibling pointers. This
52  * happens a lot, and we don't need to byte swap at runtime if the sibling
53  * pointer is NULL.
54  *
55  * These are explicitly marked at inline because the cost of calling them as
56  * functions instead of inlining them is about 36 bytes extra code per call site
57  * on x86-64. Yes, gcc-11 fails to inline them, and explicit inlining of these
58  * two sibling check functions reduces the compiled code size by over 300
59  * bytes.
60  */
61 static inline xfs_failaddr_t
xfs_btree_check_fsblock_siblings(struct xfs_mount * mp,xfs_fsblock_t fsb,__be64 dsibling)62 xfs_btree_check_fsblock_siblings(
63 	struct xfs_mount	*mp,
64 	xfs_fsblock_t		fsb,
65 	__be64			dsibling)
66 {
67 	xfs_fsblock_t		sibling;
68 
69 	if (dsibling == cpu_to_be64(NULLFSBLOCK))
70 		return NULL;
71 
72 	sibling = be64_to_cpu(dsibling);
73 	if (sibling == fsb)
74 		return __this_address;
75 	if (!xfs_verify_fsbno(mp, sibling))
76 		return __this_address;
77 	return NULL;
78 }
79 
80 static inline xfs_failaddr_t
xfs_btree_check_memblock_siblings(struct xfs_buftarg * btp,xfbno_t bno,__be64 dsibling)81 xfs_btree_check_memblock_siblings(
82 	struct xfs_buftarg	*btp,
83 	xfbno_t			bno,
84 	__be64			dsibling)
85 {
86 	xfbno_t			sibling;
87 
88 	if (dsibling == cpu_to_be64(NULLFSBLOCK))
89 		return NULL;
90 
91 	sibling = be64_to_cpu(dsibling);
92 	if (sibling == bno)
93 		return __this_address;
94 	if (!xmbuf_verify_daddr(btp, xfbno_to_daddr(sibling)))
95 		return __this_address;
96 	return NULL;
97 }
98 
99 static inline xfs_failaddr_t
xfs_btree_check_agblock_siblings(struct xfs_perag * pag,xfs_agblock_t agbno,__be32 dsibling)100 xfs_btree_check_agblock_siblings(
101 	struct xfs_perag	*pag,
102 	xfs_agblock_t		agbno,
103 	__be32			dsibling)
104 {
105 	xfs_agblock_t		sibling;
106 
107 	if (dsibling == cpu_to_be32(NULLAGBLOCK))
108 		return NULL;
109 
110 	sibling = be32_to_cpu(dsibling);
111 	if (sibling == agbno)
112 		return __this_address;
113 	if (!xfs_verify_agbno(pag, sibling))
114 		return __this_address;
115 	return NULL;
116 }
117 
118 static xfs_failaddr_t
__xfs_btree_check_lblock_hdr(struct xfs_btree_cur * cur,struct xfs_btree_block * block,int level,struct xfs_buf * bp)119 __xfs_btree_check_lblock_hdr(
120 	struct xfs_btree_cur	*cur,
121 	struct xfs_btree_block	*block,
122 	int			level,
123 	struct xfs_buf		*bp)
124 {
125 	struct xfs_mount	*mp = cur->bc_mp;
126 
127 	if (xfs_has_crc(mp)) {
128 		if (!uuid_equal(&block->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid))
129 			return __this_address;
130 		if (block->bb_u.l.bb_blkno !=
131 		    cpu_to_be64(bp ? xfs_buf_daddr(bp) : XFS_BUF_DADDR_NULL))
132 			return __this_address;
133 		if (block->bb_u.l.bb_pad != cpu_to_be32(0))
134 			return __this_address;
135 	}
136 
137 	if (be32_to_cpu(block->bb_magic) != xfs_btree_magic(mp, cur->bc_ops))
138 		return __this_address;
139 	if (be16_to_cpu(block->bb_level) != level)
140 		return __this_address;
141 	if (be16_to_cpu(block->bb_numrecs) >
142 	    cur->bc_ops->get_maxrecs(cur, level))
143 		return __this_address;
144 
145 	return NULL;
146 }
147 
148 /*
149  * Check a long btree block header.  Return the address of the failing check,
150  * or NULL if everything is ok.
151  */
152 static xfs_failaddr_t
__xfs_btree_check_fsblock(struct xfs_btree_cur * cur,struct xfs_btree_block * block,int level,struct xfs_buf * bp)153 __xfs_btree_check_fsblock(
154 	struct xfs_btree_cur	*cur,
155 	struct xfs_btree_block	*block,
156 	int			level,
157 	struct xfs_buf		*bp)
158 {
159 	struct xfs_mount	*mp = cur->bc_mp;
160 	xfs_failaddr_t		fa;
161 	xfs_fsblock_t		fsb;
162 
163 	fa = __xfs_btree_check_lblock_hdr(cur, block, level, bp);
164 	if (fa)
165 		return fa;
166 
167 	/*
168 	 * For inode-rooted btrees, the root block sits in the inode fork.  In
169 	 * that case bp is NULL, and the block must not have any siblings.
170 	 */
171 	if (!bp) {
172 		if (block->bb_u.l.bb_leftsib != cpu_to_be64(NULLFSBLOCK))
173 			return __this_address;
174 		if (block->bb_u.l.bb_rightsib != cpu_to_be64(NULLFSBLOCK))
175 			return __this_address;
176 		return NULL;
177 	}
178 
179 	fsb = XFS_DADDR_TO_FSB(mp, xfs_buf_daddr(bp));
180 	fa = xfs_btree_check_fsblock_siblings(mp, fsb,
181 			block->bb_u.l.bb_leftsib);
182 	if (!fa)
183 		fa = xfs_btree_check_fsblock_siblings(mp, fsb,
184 				block->bb_u.l.bb_rightsib);
185 	return fa;
186 }
187 
188 /*
189  * Check an in-memory btree block header.  Return the address of the failing
190  * check, or NULL if everything is ok.
191  */
192 static xfs_failaddr_t
__xfs_btree_check_memblock(struct xfs_btree_cur * cur,struct xfs_btree_block * block,int level,struct xfs_buf * bp)193 __xfs_btree_check_memblock(
194 	struct xfs_btree_cur	*cur,
195 	struct xfs_btree_block	*block,
196 	int			level,
197 	struct xfs_buf		*bp)
198 {
199 	struct xfs_buftarg	*btp = cur->bc_mem.xfbtree->target;
200 	xfs_failaddr_t		fa;
201 	xfbno_t			bno;
202 
203 	fa = __xfs_btree_check_lblock_hdr(cur, block, level, bp);
204 	if (fa)
205 		return fa;
206 
207 	bno = xfs_daddr_to_xfbno(xfs_buf_daddr(bp));
208 	fa = xfs_btree_check_memblock_siblings(btp, bno,
209 			block->bb_u.l.bb_leftsib);
210 	if (!fa)
211 		fa = xfs_btree_check_memblock_siblings(btp, bno,
212 				block->bb_u.l.bb_rightsib);
213 	return fa;
214 }
215 
216 /*
217  * Check a short btree block header.  Return the address of the failing check,
218  * or NULL if everything is ok.
219  */
220 static xfs_failaddr_t
__xfs_btree_check_agblock(struct xfs_btree_cur * cur,struct xfs_btree_block * block,int level,struct xfs_buf * bp)221 __xfs_btree_check_agblock(
222 	struct xfs_btree_cur	*cur,
223 	struct xfs_btree_block	*block,
224 	int			level,
225 	struct xfs_buf		*bp)
226 {
227 	struct xfs_mount	*mp = cur->bc_mp;
228 	struct xfs_perag	*pag = cur->bc_ag.pag;
229 	xfs_failaddr_t		fa;
230 	xfs_agblock_t		agbno;
231 
232 	if (xfs_has_crc(mp)) {
233 		if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid))
234 			return __this_address;
235 		if (block->bb_u.s.bb_blkno != cpu_to_be64(xfs_buf_daddr(bp)))
236 			return __this_address;
237 	}
238 
239 	if (be32_to_cpu(block->bb_magic) != xfs_btree_magic(mp, cur->bc_ops))
240 		return __this_address;
241 	if (be16_to_cpu(block->bb_level) != level)
242 		return __this_address;
243 	if (be16_to_cpu(block->bb_numrecs) >
244 	    cur->bc_ops->get_maxrecs(cur, level))
245 		return __this_address;
246 
247 	agbno = xfs_daddr_to_agbno(mp, xfs_buf_daddr(bp));
248 	fa = xfs_btree_check_agblock_siblings(pag, agbno,
249 			block->bb_u.s.bb_leftsib);
250 	if (!fa)
251 		fa = xfs_btree_check_agblock_siblings(pag, agbno,
252 				block->bb_u.s.bb_rightsib);
253 	return fa;
254 }
255 
256 /*
257  * Internal btree block check.
258  *
259  * Return NULL if the block is ok or the address of the failed check otherwise.
260  */
261 xfs_failaddr_t
__xfs_btree_check_block(struct xfs_btree_cur * cur,struct xfs_btree_block * block,int level,struct xfs_buf * bp)262 __xfs_btree_check_block(
263 	struct xfs_btree_cur	*cur,
264 	struct xfs_btree_block	*block,
265 	int			level,
266 	struct xfs_buf		*bp)
267 {
268 	switch (cur->bc_ops->type) {
269 	case XFS_BTREE_TYPE_MEM:
270 		return __xfs_btree_check_memblock(cur, block, level, bp);
271 	case XFS_BTREE_TYPE_AG:
272 		return __xfs_btree_check_agblock(cur, block, level, bp);
273 	case XFS_BTREE_TYPE_INODE:
274 		return __xfs_btree_check_fsblock(cur, block, level, bp);
275 	default:
276 		ASSERT(0);
277 		return __this_address;
278 	}
279 }
280 
xfs_btree_block_errtag(struct xfs_btree_cur * cur)281 static inline unsigned int xfs_btree_block_errtag(struct xfs_btree_cur *cur)
282 {
283 	if (cur->bc_ops->ptr_len == XFS_BTREE_SHORT_PTR_LEN)
284 		return XFS_ERRTAG_BTREE_CHECK_SBLOCK;
285 	return XFS_ERRTAG_BTREE_CHECK_LBLOCK;
286 }
287 
288 /*
289  * Debug routine: check that block header is ok.
290  */
291 int
xfs_btree_check_block(struct xfs_btree_cur * cur,struct xfs_btree_block * block,int level,struct xfs_buf * bp)292 xfs_btree_check_block(
293 	struct xfs_btree_cur	*cur,	/* btree cursor */
294 	struct xfs_btree_block	*block,	/* generic btree block pointer */
295 	int			level,	/* level of the btree block */
296 	struct xfs_buf		*bp)	/* buffer containing block, if any */
297 {
298 	struct xfs_mount	*mp = cur->bc_mp;
299 	xfs_failaddr_t		fa;
300 
301 	fa = __xfs_btree_check_block(cur, block, level, bp);
302 	if (XFS_IS_CORRUPT(mp, fa != NULL) ||
303 	    XFS_TEST_ERROR(false, mp, xfs_btree_block_errtag(cur))) {
304 		if (bp)
305 			trace_xfs_btree_corrupt(bp, _RET_IP_);
306 		xfs_btree_mark_sick(cur);
307 		return -EFSCORRUPTED;
308 	}
309 	return 0;
310 }
311 
312 int
__xfs_btree_check_ptr(struct xfs_btree_cur * cur,const union xfs_btree_ptr * ptr,int index,int level)313 __xfs_btree_check_ptr(
314 	struct xfs_btree_cur		*cur,
315 	const union xfs_btree_ptr	*ptr,
316 	int				index,
317 	int				level)
318 {
319 	if (level <= 0)
320 		return -EFSCORRUPTED;
321 
322 	switch (cur->bc_ops->type) {
323 	case XFS_BTREE_TYPE_MEM:
324 		if (!xfbtree_verify_bno(cur->bc_mem.xfbtree,
325 				be64_to_cpu((&ptr->l)[index])))
326 			return -EFSCORRUPTED;
327 		break;
328 	case XFS_BTREE_TYPE_INODE:
329 		if (!xfs_verify_fsbno(cur->bc_mp,
330 				be64_to_cpu((&ptr->l)[index])))
331 			return -EFSCORRUPTED;
332 		break;
333 	case XFS_BTREE_TYPE_AG:
334 		if (!xfs_verify_agbno(cur->bc_ag.pag,
335 				be32_to_cpu((&ptr->s)[index])))
336 			return -EFSCORRUPTED;
337 		break;
338 	}
339 
340 	return 0;
341 }
342 
343 /*
344  * Check that a given (indexed) btree pointer at a certain level of a
345  * btree is valid and doesn't point past where it should.
346  */
347 static int
xfs_btree_check_ptr(struct xfs_btree_cur * cur,const union xfs_btree_ptr * ptr,int index,int level)348 xfs_btree_check_ptr(
349 	struct xfs_btree_cur		*cur,
350 	const union xfs_btree_ptr	*ptr,
351 	int				index,
352 	int				level)
353 {
354 	int				error;
355 
356 	error = __xfs_btree_check_ptr(cur, ptr, index, level);
357 	if (error) {
358 		switch (cur->bc_ops->type) {
359 		case XFS_BTREE_TYPE_MEM:
360 			xfs_err(cur->bc_mp,
361 "In-memory: Corrupt %sbt flags 0x%x pointer at level %d index %d fa %pS.",
362 				cur->bc_ops->name, cur->bc_flags, level, index,
363 				__this_address);
364 			break;
365 		case XFS_BTREE_TYPE_INODE:
366 			xfs_err(cur->bc_mp,
367 "Inode %llu fork %d: Corrupt %sbt pointer at level %d index %d.",
368 				cur->bc_ino.ip->i_ino,
369 				cur->bc_ino.whichfork, cur->bc_ops->name,
370 				level, index);
371 			break;
372 		case XFS_BTREE_TYPE_AG:
373 			xfs_err(cur->bc_mp,
374 "AG %u: Corrupt %sbt pointer at level %d index %d.",
375 				cur->bc_ag.pag->pag_agno, cur->bc_ops->name,
376 				level, index);
377 			break;
378 		}
379 		xfs_btree_mark_sick(cur);
380 	}
381 
382 	return error;
383 }
384 
385 #ifdef DEBUG
386 # define xfs_btree_debug_check_ptr	xfs_btree_check_ptr
387 #else
388 # define xfs_btree_debug_check_ptr(...)	(0)
389 #endif
390 
391 /*
392  * Calculate CRC on the whole btree block and stuff it into the
393  * long-form btree header.
394  *
395  * Prior to calculting the CRC, pull the LSN out of the buffer log item and put
396  * it into the buffer so recovery knows what the last modification was that made
397  * it to disk.
398  */
399 void
xfs_btree_fsblock_calc_crc(struct xfs_buf * bp)400 xfs_btree_fsblock_calc_crc(
401 	struct xfs_buf		*bp)
402 {
403 	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
404 	struct xfs_buf_log_item	*bip = bp->b_log_item;
405 
406 	if (!xfs_has_crc(bp->b_mount))
407 		return;
408 	if (bip)
409 		block->bb_u.l.bb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
410 	xfs_buf_update_cksum(bp, XFS_BTREE_LBLOCK_CRC_OFF);
411 }
412 
413 bool
xfs_btree_fsblock_verify_crc(struct xfs_buf * bp)414 xfs_btree_fsblock_verify_crc(
415 	struct xfs_buf		*bp)
416 {
417 	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
418 	struct xfs_mount	*mp = bp->b_mount;
419 
420 	if (xfs_has_crc(mp)) {
421 		if (!xfs_log_check_lsn(mp, be64_to_cpu(block->bb_u.l.bb_lsn)))
422 			return false;
423 		return xfs_buf_verify_cksum(bp, XFS_BTREE_LBLOCK_CRC_OFF);
424 	}
425 
426 	return true;
427 }
428 
429 /*
430  * Calculate CRC on the whole btree block and stuff it into the
431  * short-form btree header.
432  *
433  * Prior to calculting the CRC, pull the LSN out of the buffer log item and put
434  * it into the buffer so recovery knows what the last modification was that made
435  * it to disk.
436  */
437 void
xfs_btree_agblock_calc_crc(struct xfs_buf * bp)438 xfs_btree_agblock_calc_crc(
439 	struct xfs_buf		*bp)
440 {
441 	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
442 	struct xfs_buf_log_item	*bip = bp->b_log_item;
443 
444 	if (!xfs_has_crc(bp->b_mount))
445 		return;
446 	if (bip)
447 		block->bb_u.s.bb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
448 	xfs_buf_update_cksum(bp, XFS_BTREE_SBLOCK_CRC_OFF);
449 }
450 
451 bool
xfs_btree_agblock_verify_crc(struct xfs_buf * bp)452 xfs_btree_agblock_verify_crc(
453 	struct xfs_buf		*bp)
454 {
455 	struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
456 	struct xfs_mount	*mp = bp->b_mount;
457 
458 	if (xfs_has_crc(mp)) {
459 		if (!xfs_log_check_lsn(mp, be64_to_cpu(block->bb_u.s.bb_lsn)))
460 			return false;
461 		return xfs_buf_verify_cksum(bp, XFS_BTREE_SBLOCK_CRC_OFF);
462 	}
463 
464 	return true;
465 }
466 
467 static int
xfs_btree_free_block(struct xfs_btree_cur * cur,struct xfs_buf * bp)468 xfs_btree_free_block(
469 	struct xfs_btree_cur	*cur,
470 	struct xfs_buf		*bp)
471 {
472 	int			error;
473 
474 	trace_xfs_btree_free_block(cur, bp);
475 
476 	/*
477 	 * Don't allow block freeing for a staging cursor, because staging
478 	 * cursors do not support regular btree modifications.
479 	 */
480 	if (unlikely(cur->bc_flags & XFS_BTREE_STAGING)) {
481 		ASSERT(0);
482 		return -EFSCORRUPTED;
483 	}
484 
485 	error = cur->bc_ops->free_block(cur, bp);
486 	if (!error) {
487 		xfs_trans_binval(cur->bc_tp, bp);
488 		XFS_BTREE_STATS_INC(cur, free);
489 	}
490 	return error;
491 }
492 
493 /*
494  * Delete the btree cursor.
495  */
496 void
xfs_btree_del_cursor(struct xfs_btree_cur * cur,int error)497 xfs_btree_del_cursor(
498 	struct xfs_btree_cur	*cur,		/* btree cursor */
499 	int			error)		/* del because of error */
500 {
501 	int			i;		/* btree level */
502 
503 	/*
504 	 * Clear the buffer pointers and release the buffers. If we're doing
505 	 * this because of an error, inspect all of the entries in the bc_bufs
506 	 * array for buffers to be unlocked. This is because some of the btree
507 	 * code works from level n down to 0, and if we get an error along the
508 	 * way we won't have initialized all the entries down to 0.
509 	 */
510 	for (i = 0; i < cur->bc_nlevels; i++) {
511 		if (cur->bc_levels[i].bp)
512 			xfs_trans_brelse(cur->bc_tp, cur->bc_levels[i].bp);
513 		else if (!error)
514 			break;
515 	}
516 
517 	/*
518 	 * If we are doing a BMBT update, the number of unaccounted blocks
519 	 * allocated during this cursor life time should be zero. If it's not
520 	 * zero, then we should be shut down or on our way to shutdown due to
521 	 * cancelling a dirty transaction on error.
522 	 */
523 	ASSERT(!xfs_btree_is_bmap(cur->bc_ops) || cur->bc_bmap.allocated == 0 ||
524 	       xfs_is_shutdown(cur->bc_mp) || error != 0);
525 
526 	switch (cur->bc_ops->type) {
527 	case XFS_BTREE_TYPE_AG:
528 		if (cur->bc_ag.pag)
529 			xfs_perag_put(cur->bc_ag.pag);
530 		break;
531 	case XFS_BTREE_TYPE_INODE:
532 		/* nothing to do */
533 		break;
534 	case XFS_BTREE_TYPE_MEM:
535 		if (cur->bc_mem.pag)
536 			xfs_perag_put(cur->bc_mem.pag);
537 		break;
538 	}
539 
540 	kmem_cache_free(cur->bc_cache, cur);
541 }
542 
543 /* Return the buffer target for this btree's buffer. */
544 static inline struct xfs_buftarg *
xfs_btree_buftarg(struct xfs_btree_cur * cur)545 xfs_btree_buftarg(
546 	struct xfs_btree_cur	*cur)
547 {
548 	if (cur->bc_ops->type == XFS_BTREE_TYPE_MEM)
549 		return cur->bc_mem.xfbtree->target;
550 	return cur->bc_mp->m_ddev_targp;
551 }
552 
553 /* Return the block size (in units of 512b sectors) for this btree. */
554 static inline unsigned int
xfs_btree_bbsize(struct xfs_btree_cur * cur)555 xfs_btree_bbsize(
556 	struct xfs_btree_cur	*cur)
557 {
558 	if (cur->bc_ops->type == XFS_BTREE_TYPE_MEM)
559 		return XFBNO_BBSIZE;
560 	return cur->bc_mp->m_bsize;
561 }
562 
563 /*
564  * Duplicate the btree cursor.
565  * Allocate a new one, copy the record, re-get the buffers.
566  */
567 int						/* error */
xfs_btree_dup_cursor(struct xfs_btree_cur * cur,struct xfs_btree_cur ** ncur)568 xfs_btree_dup_cursor(
569 	struct xfs_btree_cur	*cur,		/* input cursor */
570 	struct xfs_btree_cur	**ncur)		/* output cursor */
571 {
572 	struct xfs_mount	*mp = cur->bc_mp;
573 	struct xfs_trans	*tp = cur->bc_tp;
574 	struct xfs_buf		*bp;
575 	struct xfs_btree_cur	*new;
576 	int			error;
577 	int			i;
578 
579 	/*
580 	 * Don't allow staging cursors to be duplicated because they're supposed
581 	 * to be kept private to a single thread.
582 	 */
583 	if (unlikely(cur->bc_flags & XFS_BTREE_STAGING)) {
584 		ASSERT(0);
585 		return -EFSCORRUPTED;
586 	}
587 
588 	/*
589 	 * Allocate a new cursor like the old one.
590 	 */
591 	new = cur->bc_ops->dup_cursor(cur);
592 
593 	/*
594 	 * Copy the record currently in the cursor.
595 	 */
596 	new->bc_rec = cur->bc_rec;
597 
598 	/*
599 	 * For each level current, re-get the buffer and copy the ptr value.
600 	 */
601 	for (i = 0; i < new->bc_nlevels; i++) {
602 		new->bc_levels[i].ptr = cur->bc_levels[i].ptr;
603 		new->bc_levels[i].ra = cur->bc_levels[i].ra;
604 		bp = cur->bc_levels[i].bp;
605 		if (bp) {
606 			error = xfs_trans_read_buf(mp, tp,
607 					xfs_btree_buftarg(cur),
608 					xfs_buf_daddr(bp),
609 					xfs_btree_bbsize(cur), 0, &bp,
610 					cur->bc_ops->buf_ops);
611 			if (xfs_metadata_is_sick(error))
612 				xfs_btree_mark_sick(new);
613 			if (error) {
614 				xfs_btree_del_cursor(new, error);
615 				*ncur = NULL;
616 				return error;
617 			}
618 		}
619 		new->bc_levels[i].bp = bp;
620 	}
621 	*ncur = new;
622 	return 0;
623 }
624 
625 /*
626  * XFS btree block layout and addressing:
627  *
628  * There are two types of blocks in the btree: leaf and non-leaf blocks.
629  *
630  * The leaf record start with a header then followed by records containing
631  * the values.  A non-leaf block also starts with the same header, and
632  * then first contains lookup keys followed by an equal number of pointers
633  * to the btree blocks at the previous level.
634  *
635  *		+--------+-------+-------+-------+-------+-------+-------+
636  * Leaf:	| header | rec 1 | rec 2 | rec 3 | rec 4 | rec 5 | rec N |
637  *		+--------+-------+-------+-------+-------+-------+-------+
638  *
639  *		+--------+-------+-------+-------+-------+-------+-------+
640  * Non-Leaf:	| header | key 1 | key 2 | key N | ptr 1 | ptr 2 | ptr N |
641  *		+--------+-------+-------+-------+-------+-------+-------+
642  *
643  * The header is called struct xfs_btree_block for reasons better left unknown
644  * and comes in different versions for short (32bit) and long (64bit) block
645  * pointers.  The record and key structures are defined by the btree instances
646  * and opaque to the btree core.  The block pointers are simple disk endian
647  * integers, available in a short (32bit) and long (64bit) variant.
648  *
649  * The helpers below calculate the offset of a given record, key or pointer
650  * into a btree block (xfs_btree_*_offset) or return a pointer to the given
651  * record, key or pointer (xfs_btree_*_addr).  Note that all addressing
652  * inside the btree block is done using indices starting at one, not zero!
653  *
654  * If XFS_BTGEO_OVERLAPPING is set, then this btree supports keys containing
655  * overlapping intervals.  In such a tree, records are still sorted lowest to
656  * highest and indexed by the smallest key value that refers to the record.
657  * However, nodes are different: each pointer has two associated keys -- one
658  * indexing the lowest key available in the block(s) below (the same behavior
659  * as the key in a regular btree) and another indexing the highest key
660  * available in the block(s) below.  Because records are /not/ sorted by the
661  * highest key, all leaf block updates require us to compute the highest key
662  * that matches any record in the leaf and to recursively update the high keys
663  * in the nodes going further up in the tree, if necessary.  Nodes look like
664  * this:
665  *
666  *		+--------+-----+-----+-----+-----+-----+-------+-------+-----+
667  * Non-Leaf:	| header | lo1 | hi1 | lo2 | hi2 | ... | ptr 1 | ptr 2 | ... |
668  *		+--------+-----+-----+-----+-----+-----+-------+-------+-----+
669  *
670  * To perform an interval query on an overlapped tree, perform the usual
671  * depth-first search and use the low and high keys to decide if we can skip
672  * that particular node.  If a leaf node is reached, return the records that
673  * intersect the interval.  Note that an interval query may return numerous
674  * entries.  For a non-overlapped tree, simply search for the record associated
675  * with the lowest key and iterate forward until a non-matching record is
676  * found.  Section 14.3 ("Interval Trees") of _Introduction to Algorithms_ by
677  * Cormen, Leiserson, Rivest, and Stein (2nd or 3rd ed. only) discuss this in
678  * more detail.
679  *
680  * Why do we care about overlapping intervals?  Let's say you have a bunch of
681  * reverse mapping records on a reflink filesystem:
682  *
683  * 1: +- file A startblock B offset C length D -----------+
684  * 2:      +- file E startblock F offset G length H --------------+
685  * 3:      +- file I startblock F offset J length K --+
686  * 4:                                                        +- file L... --+
687  *
688  * Now say we want to map block (B+D) into file A at offset (C+D).  Ideally,
689  * we'd simply increment the length of record 1.  But how do we find the record
690  * that ends at (B+D-1) (i.e. record 1)?  A LE lookup of (B+D-1) would return
691  * record 3 because the keys are ordered first by startblock.  An interval
692  * query would return records 1 and 2 because they both overlap (B+D-1), and
693  * from that we can pick out record 1 as the appropriate left neighbor.
694  *
695  * In the non-overlapped case you can do a LE lookup and decrement the cursor
696  * because a record's interval must end before the next record.
697  */
698 
699 /*
700  * Return size of the btree block header for this btree instance.
701  */
xfs_btree_block_len(struct xfs_btree_cur * cur)702 static inline size_t xfs_btree_block_len(struct xfs_btree_cur *cur)
703 {
704 	if (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN) {
705 		if (xfs_has_crc(cur->bc_mp))
706 			return XFS_BTREE_LBLOCK_CRC_LEN;
707 		return XFS_BTREE_LBLOCK_LEN;
708 	}
709 	if (xfs_has_crc(cur->bc_mp))
710 		return XFS_BTREE_SBLOCK_CRC_LEN;
711 	return XFS_BTREE_SBLOCK_LEN;
712 }
713 
714 /*
715  * Calculate offset of the n-th record in a btree block.
716  */
717 STATIC size_t
xfs_btree_rec_offset(struct xfs_btree_cur * cur,int n)718 xfs_btree_rec_offset(
719 	struct xfs_btree_cur	*cur,
720 	int			n)
721 {
722 	return xfs_btree_block_len(cur) +
723 		(n - 1) * cur->bc_ops->rec_len;
724 }
725 
726 /*
727  * Calculate offset of the n-th key in a btree block.
728  */
729 STATIC size_t
xfs_btree_key_offset(struct xfs_btree_cur * cur,int n)730 xfs_btree_key_offset(
731 	struct xfs_btree_cur	*cur,
732 	int			n)
733 {
734 	return xfs_btree_block_len(cur) +
735 		(n - 1) * cur->bc_ops->key_len;
736 }
737 
738 /*
739  * Calculate offset of the n-th high key in a btree block.
740  */
741 STATIC size_t
xfs_btree_high_key_offset(struct xfs_btree_cur * cur,int n)742 xfs_btree_high_key_offset(
743 	struct xfs_btree_cur	*cur,
744 	int			n)
745 {
746 	return xfs_btree_block_len(cur) +
747 		(n - 1) * cur->bc_ops->key_len + (cur->bc_ops->key_len / 2);
748 }
749 
750 /*
751  * Calculate offset of the n-th block pointer in a btree block.
752  */
753 STATIC size_t
xfs_btree_ptr_offset(struct xfs_btree_cur * cur,int n,int level)754 xfs_btree_ptr_offset(
755 	struct xfs_btree_cur	*cur,
756 	int			n,
757 	int			level)
758 {
759 	return xfs_btree_block_len(cur) +
760 		cur->bc_ops->get_maxrecs(cur, level) * cur->bc_ops->key_len +
761 		(n - 1) * cur->bc_ops->ptr_len;
762 }
763 
764 /*
765  * Return a pointer to the n-th record in the btree block.
766  */
767 union xfs_btree_rec *
xfs_btree_rec_addr(struct xfs_btree_cur * cur,int n,struct xfs_btree_block * block)768 xfs_btree_rec_addr(
769 	struct xfs_btree_cur	*cur,
770 	int			n,
771 	struct xfs_btree_block	*block)
772 {
773 	return (union xfs_btree_rec *)
774 		((char *)block + xfs_btree_rec_offset(cur, n));
775 }
776 
777 /*
778  * Return a pointer to the n-th key in the btree block.
779  */
780 union xfs_btree_key *
xfs_btree_key_addr(struct xfs_btree_cur * cur,int n,struct xfs_btree_block * block)781 xfs_btree_key_addr(
782 	struct xfs_btree_cur	*cur,
783 	int			n,
784 	struct xfs_btree_block	*block)
785 {
786 	return (union xfs_btree_key *)
787 		((char *)block + xfs_btree_key_offset(cur, n));
788 }
789 
790 /*
791  * Return a pointer to the n-th high key in the btree block.
792  */
793 union xfs_btree_key *
xfs_btree_high_key_addr(struct xfs_btree_cur * cur,int n,struct xfs_btree_block * block)794 xfs_btree_high_key_addr(
795 	struct xfs_btree_cur	*cur,
796 	int			n,
797 	struct xfs_btree_block	*block)
798 {
799 	return (union xfs_btree_key *)
800 		((char *)block + xfs_btree_high_key_offset(cur, n));
801 }
802 
803 /*
804  * Return a pointer to the n-th block pointer in the btree block.
805  */
806 union xfs_btree_ptr *
xfs_btree_ptr_addr(struct xfs_btree_cur * cur,int n,struct xfs_btree_block * block)807 xfs_btree_ptr_addr(
808 	struct xfs_btree_cur	*cur,
809 	int			n,
810 	struct xfs_btree_block	*block)
811 {
812 	int			level = xfs_btree_get_level(block);
813 
814 	ASSERT(block->bb_level != 0);
815 
816 	return (union xfs_btree_ptr *)
817 		((char *)block + xfs_btree_ptr_offset(cur, n, level));
818 }
819 
820 struct xfs_ifork *
xfs_btree_ifork_ptr(struct xfs_btree_cur * cur)821 xfs_btree_ifork_ptr(
822 	struct xfs_btree_cur	*cur)
823 {
824 	ASSERT(cur->bc_ops->type == XFS_BTREE_TYPE_INODE);
825 
826 	if (cur->bc_flags & XFS_BTREE_STAGING)
827 		return cur->bc_ino.ifake->if_fork;
828 	return xfs_ifork_ptr(cur->bc_ino.ip, cur->bc_ino.whichfork);
829 }
830 
831 /*
832  * Get the root block which is stored in the inode.
833  *
834  * For now this btree implementation assumes the btree root is always
835  * stored in the if_broot field of an inode fork.
836  */
837 STATIC struct xfs_btree_block *
xfs_btree_get_iroot(struct xfs_btree_cur * cur)838 xfs_btree_get_iroot(
839 	struct xfs_btree_cur	*cur)
840 {
841 	struct xfs_ifork	*ifp = xfs_btree_ifork_ptr(cur);
842 
843 	return (struct xfs_btree_block *)ifp->if_broot;
844 }
845 
846 /*
847  * Retrieve the block pointer from the cursor at the given level.
848  * This may be an inode btree root or from a buffer.
849  */
850 struct xfs_btree_block *		/* generic btree block pointer */
xfs_btree_get_block(struct xfs_btree_cur * cur,int level,struct xfs_buf ** bpp)851 xfs_btree_get_block(
852 	struct xfs_btree_cur	*cur,	/* btree cursor */
853 	int			level,	/* level in btree */
854 	struct xfs_buf		**bpp)	/* buffer containing the block */
855 {
856 	if (xfs_btree_at_iroot(cur, level)) {
857 		*bpp = NULL;
858 		return xfs_btree_get_iroot(cur);
859 	}
860 
861 	*bpp = cur->bc_levels[level].bp;
862 	return XFS_BUF_TO_BLOCK(*bpp);
863 }
864 
865 /*
866  * Change the cursor to point to the first record at the given level.
867  * Other levels are unaffected.
868  */
869 STATIC int				/* success=1, failure=0 */
xfs_btree_firstrec(struct xfs_btree_cur * cur,int level)870 xfs_btree_firstrec(
871 	struct xfs_btree_cur	*cur,	/* btree cursor */
872 	int			level)	/* level to change */
873 {
874 	struct xfs_btree_block	*block;	/* generic btree block pointer */
875 	struct xfs_buf		*bp;	/* buffer containing block */
876 
877 	/*
878 	 * Get the block pointer for this level.
879 	 */
880 	block = xfs_btree_get_block(cur, level, &bp);
881 	if (xfs_btree_check_block(cur, block, level, bp))
882 		return 0;
883 	/*
884 	 * It's empty, there is no such record.
885 	 */
886 	if (!block->bb_numrecs)
887 		return 0;
888 	/*
889 	 * Set the ptr value to 1, that's the first record/key.
890 	 */
891 	cur->bc_levels[level].ptr = 1;
892 	return 1;
893 }
894 
895 /*
896  * Change the cursor to point to the last record in the current block
897  * at the given level.  Other levels are unaffected.
898  */
899 STATIC int				/* success=1, failure=0 */
xfs_btree_lastrec(struct xfs_btree_cur * cur,int level)900 xfs_btree_lastrec(
901 	struct xfs_btree_cur	*cur,	/* btree cursor */
902 	int			level)	/* level to change */
903 {
904 	struct xfs_btree_block	*block;	/* generic btree block pointer */
905 	struct xfs_buf		*bp;	/* buffer containing block */
906 
907 	/*
908 	 * Get the block pointer for this level.
909 	 */
910 	block = xfs_btree_get_block(cur, level, &bp);
911 	if (xfs_btree_check_block(cur, block, level, bp))
912 		return 0;
913 	/*
914 	 * It's empty, there is no such record.
915 	 */
916 	if (!block->bb_numrecs)
917 		return 0;
918 	/*
919 	 * Set the ptr value to numrecs, that's the last record/key.
920 	 */
921 	cur->bc_levels[level].ptr = be16_to_cpu(block->bb_numrecs);
922 	return 1;
923 }
924 
925 /*
926  * Compute first and last byte offsets for the fields given.
927  * Interprets the offsets table, which contains struct field offsets.
928  */
929 void
xfs_btree_offsets(uint32_t fields,const short * offsets,int nbits,int * first,int * last)930 xfs_btree_offsets(
931 	uint32_t	fields,		/* bitmask of fields */
932 	const short	*offsets,	/* table of field offsets */
933 	int		nbits,		/* number of bits to inspect */
934 	int		*first,		/* output: first byte offset */
935 	int		*last)		/* output: last byte offset */
936 {
937 	int		i;		/* current bit number */
938 	uint32_t	imask;		/* mask for current bit number */
939 
940 	ASSERT(fields != 0);
941 	/*
942 	 * Find the lowest bit, so the first byte offset.
943 	 */
944 	for (i = 0, imask = 1u; ; i++, imask <<= 1) {
945 		if (imask & fields) {
946 			*first = offsets[i];
947 			break;
948 		}
949 	}
950 	/*
951 	 * Find the highest bit, so the last byte offset.
952 	 */
953 	for (i = nbits - 1, imask = 1u << i; ; i--, imask >>= 1) {
954 		if (imask & fields) {
955 			*last = offsets[i + 1] - 1;
956 			break;
957 		}
958 	}
959 }
960 
961 STATIC int
xfs_btree_readahead_fsblock(struct xfs_btree_cur * cur,int lr,struct xfs_btree_block * block)962 xfs_btree_readahead_fsblock(
963 	struct xfs_btree_cur	*cur,
964 	int			lr,
965 	struct xfs_btree_block	*block)
966 {
967 	struct xfs_mount	*mp = cur->bc_mp;
968 	xfs_fsblock_t		left = be64_to_cpu(block->bb_u.l.bb_leftsib);
969 	xfs_fsblock_t		right = be64_to_cpu(block->bb_u.l.bb_rightsib);
970 	int			rval = 0;
971 
972 	if ((lr & XFS_BTCUR_LEFTRA) && left != NULLFSBLOCK) {
973 		xfs_buf_readahead(mp->m_ddev_targp, XFS_FSB_TO_DADDR(mp, left),
974 				mp->m_bsize, cur->bc_ops->buf_ops);
975 		rval++;
976 	}
977 
978 	if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLFSBLOCK) {
979 		xfs_buf_readahead(mp->m_ddev_targp, XFS_FSB_TO_DADDR(mp, right),
980 				mp->m_bsize, cur->bc_ops->buf_ops);
981 		rval++;
982 	}
983 
984 	return rval;
985 }
986 
987 STATIC int
xfs_btree_readahead_memblock(struct xfs_btree_cur * cur,int lr,struct xfs_btree_block * block)988 xfs_btree_readahead_memblock(
989 	struct xfs_btree_cur	*cur,
990 	int			lr,
991 	struct xfs_btree_block	*block)
992 {
993 	struct xfs_buftarg	*btp = cur->bc_mem.xfbtree->target;
994 	xfbno_t			left = be64_to_cpu(block->bb_u.l.bb_leftsib);
995 	xfbno_t			right = be64_to_cpu(block->bb_u.l.bb_rightsib);
996 	int			rval = 0;
997 
998 	if ((lr & XFS_BTCUR_LEFTRA) && left != NULLFSBLOCK) {
999 		xfs_buf_readahead(btp, xfbno_to_daddr(left), XFBNO_BBSIZE,
1000 				cur->bc_ops->buf_ops);
1001 		rval++;
1002 	}
1003 
1004 	if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLFSBLOCK) {
1005 		xfs_buf_readahead(btp, xfbno_to_daddr(right), XFBNO_BBSIZE,
1006 				cur->bc_ops->buf_ops);
1007 		rval++;
1008 	}
1009 
1010 	return rval;
1011 }
1012 
1013 STATIC int
xfs_btree_readahead_agblock(struct xfs_btree_cur * cur,int lr,struct xfs_btree_block * block)1014 xfs_btree_readahead_agblock(
1015 	struct xfs_btree_cur	*cur,
1016 	int			lr,
1017 	struct xfs_btree_block	*block)
1018 {
1019 	struct xfs_mount	*mp = cur->bc_mp;
1020 	xfs_agnumber_t		agno = cur->bc_ag.pag->pag_agno;
1021 	xfs_agblock_t		left = be32_to_cpu(block->bb_u.s.bb_leftsib);
1022 	xfs_agblock_t		right = be32_to_cpu(block->bb_u.s.bb_rightsib);
1023 	int			rval = 0;
1024 
1025 	if ((lr & XFS_BTCUR_LEFTRA) && left != NULLAGBLOCK) {
1026 		xfs_buf_readahead(mp->m_ddev_targp,
1027 				XFS_AGB_TO_DADDR(mp, agno, left),
1028 				mp->m_bsize, cur->bc_ops->buf_ops);
1029 		rval++;
1030 	}
1031 
1032 	if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLAGBLOCK) {
1033 		xfs_buf_readahead(mp->m_ddev_targp,
1034 				XFS_AGB_TO_DADDR(mp, agno, right),
1035 				mp->m_bsize, cur->bc_ops->buf_ops);
1036 		rval++;
1037 	}
1038 
1039 	return rval;
1040 }
1041 
1042 /*
1043  * Read-ahead btree blocks, at the given level.
1044  * Bits in lr are set from XFS_BTCUR_{LEFT,RIGHT}RA.
1045  */
1046 STATIC int
xfs_btree_readahead(struct xfs_btree_cur * cur,int lev,int lr)1047 xfs_btree_readahead(
1048 	struct xfs_btree_cur	*cur,		/* btree cursor */
1049 	int			lev,		/* level in btree */
1050 	int			lr)		/* left/right bits */
1051 {
1052 	struct xfs_btree_block	*block;
1053 
1054 	/*
1055 	 * No readahead needed if we are at the root level and the
1056 	 * btree root is stored in the inode.
1057 	 */
1058 	if (xfs_btree_at_iroot(cur, lev))
1059 		return 0;
1060 
1061 	if ((cur->bc_levels[lev].ra | lr) == cur->bc_levels[lev].ra)
1062 		return 0;
1063 
1064 	cur->bc_levels[lev].ra |= lr;
1065 	block = XFS_BUF_TO_BLOCK(cur->bc_levels[lev].bp);
1066 
1067 	switch (cur->bc_ops->type) {
1068 	case XFS_BTREE_TYPE_AG:
1069 		return xfs_btree_readahead_agblock(cur, lr, block);
1070 	case XFS_BTREE_TYPE_INODE:
1071 		return xfs_btree_readahead_fsblock(cur, lr, block);
1072 	case XFS_BTREE_TYPE_MEM:
1073 		return xfs_btree_readahead_memblock(cur, lr, block);
1074 	default:
1075 		ASSERT(0);
1076 		return 0;
1077 	}
1078 }
1079 
1080 STATIC int
xfs_btree_ptr_to_daddr(struct xfs_btree_cur * cur,const union xfs_btree_ptr * ptr,xfs_daddr_t * daddr)1081 xfs_btree_ptr_to_daddr(
1082 	struct xfs_btree_cur		*cur,
1083 	const union xfs_btree_ptr	*ptr,
1084 	xfs_daddr_t			*daddr)
1085 {
1086 	int			error;
1087 
1088 	error = xfs_btree_check_ptr(cur, ptr, 0, 1);
1089 	if (error)
1090 		return error;
1091 
1092 	switch (cur->bc_ops->type) {
1093 	case XFS_BTREE_TYPE_AG:
1094 		*daddr = XFS_AGB_TO_DADDR(cur->bc_mp, cur->bc_ag.pag->pag_agno,
1095 				be32_to_cpu(ptr->s));
1096 		break;
1097 	case XFS_BTREE_TYPE_INODE:
1098 		*daddr = XFS_FSB_TO_DADDR(cur->bc_mp, be64_to_cpu(ptr->l));
1099 		break;
1100 	case XFS_BTREE_TYPE_MEM:
1101 		*daddr = xfbno_to_daddr(be64_to_cpu(ptr->l));
1102 		break;
1103 	}
1104 	return 0;
1105 }
1106 
1107 /*
1108  * Readahead @count btree blocks at the given @ptr location.
1109  *
1110  * We don't need to care about long or short form btrees here as we have a
1111  * method of converting the ptr directly to a daddr available to us.
1112  */
1113 STATIC void
xfs_btree_readahead_ptr(struct xfs_btree_cur * cur,union xfs_btree_ptr * ptr,xfs_extlen_t count)1114 xfs_btree_readahead_ptr(
1115 	struct xfs_btree_cur	*cur,
1116 	union xfs_btree_ptr	*ptr,
1117 	xfs_extlen_t		count)
1118 {
1119 	xfs_daddr_t		daddr;
1120 
1121 	if (xfs_btree_ptr_to_daddr(cur, ptr, &daddr))
1122 		return;
1123 	xfs_buf_readahead(xfs_btree_buftarg(cur), daddr,
1124 			xfs_btree_bbsize(cur) * count,
1125 			cur->bc_ops->buf_ops);
1126 }
1127 
1128 /*
1129  * Set the buffer for level "lev" in the cursor to bp, releasing
1130  * any previous buffer.
1131  */
1132 STATIC void
xfs_btree_setbuf(struct xfs_btree_cur * cur,int lev,struct xfs_buf * bp)1133 xfs_btree_setbuf(
1134 	struct xfs_btree_cur	*cur,	/* btree cursor */
1135 	int			lev,	/* level in btree */
1136 	struct xfs_buf		*bp)	/* new buffer to set */
1137 {
1138 	struct xfs_btree_block	*b;	/* btree block */
1139 
1140 	if (cur->bc_levels[lev].bp)
1141 		xfs_trans_brelse(cur->bc_tp, cur->bc_levels[lev].bp);
1142 	cur->bc_levels[lev].bp = bp;
1143 	cur->bc_levels[lev].ra = 0;
1144 
1145 	b = XFS_BUF_TO_BLOCK(bp);
1146 	if (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN) {
1147 		if (b->bb_u.l.bb_leftsib == cpu_to_be64(NULLFSBLOCK))
1148 			cur->bc_levels[lev].ra |= XFS_BTCUR_LEFTRA;
1149 		if (b->bb_u.l.bb_rightsib == cpu_to_be64(NULLFSBLOCK))
1150 			cur->bc_levels[lev].ra |= XFS_BTCUR_RIGHTRA;
1151 	} else {
1152 		if (b->bb_u.s.bb_leftsib == cpu_to_be32(NULLAGBLOCK))
1153 			cur->bc_levels[lev].ra |= XFS_BTCUR_LEFTRA;
1154 		if (b->bb_u.s.bb_rightsib == cpu_to_be32(NULLAGBLOCK))
1155 			cur->bc_levels[lev].ra |= XFS_BTCUR_RIGHTRA;
1156 	}
1157 }
1158 
1159 bool
xfs_btree_ptr_is_null(struct xfs_btree_cur * cur,const union xfs_btree_ptr * ptr)1160 xfs_btree_ptr_is_null(
1161 	struct xfs_btree_cur		*cur,
1162 	const union xfs_btree_ptr	*ptr)
1163 {
1164 	if (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN)
1165 		return ptr->l == cpu_to_be64(NULLFSBLOCK);
1166 	else
1167 		return ptr->s == cpu_to_be32(NULLAGBLOCK);
1168 }
1169 
1170 void
xfs_btree_set_ptr_null(struct xfs_btree_cur * cur,union xfs_btree_ptr * ptr)1171 xfs_btree_set_ptr_null(
1172 	struct xfs_btree_cur	*cur,
1173 	union xfs_btree_ptr	*ptr)
1174 {
1175 	if (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN)
1176 		ptr->l = cpu_to_be64(NULLFSBLOCK);
1177 	else
1178 		ptr->s = cpu_to_be32(NULLAGBLOCK);
1179 }
1180 
1181 static inline bool
xfs_btree_ptrs_equal(struct xfs_btree_cur * cur,union xfs_btree_ptr * ptr1,union xfs_btree_ptr * ptr2)1182 xfs_btree_ptrs_equal(
1183 	struct xfs_btree_cur		*cur,
1184 	union xfs_btree_ptr		*ptr1,
1185 	union xfs_btree_ptr		*ptr2)
1186 {
1187 	if (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN)
1188 		return ptr1->l == ptr2->l;
1189 	return ptr1->s == ptr2->s;
1190 }
1191 
1192 /*
1193  * Get/set/init sibling pointers
1194  */
1195 void
xfs_btree_get_sibling(struct xfs_btree_cur * cur,struct xfs_btree_block * block,union xfs_btree_ptr * ptr,int lr)1196 xfs_btree_get_sibling(
1197 	struct xfs_btree_cur	*cur,
1198 	struct xfs_btree_block	*block,
1199 	union xfs_btree_ptr	*ptr,
1200 	int			lr)
1201 {
1202 	ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
1203 
1204 	if (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN) {
1205 		if (lr == XFS_BB_RIGHTSIB)
1206 			ptr->l = block->bb_u.l.bb_rightsib;
1207 		else
1208 			ptr->l = block->bb_u.l.bb_leftsib;
1209 	} else {
1210 		if (lr == XFS_BB_RIGHTSIB)
1211 			ptr->s = block->bb_u.s.bb_rightsib;
1212 		else
1213 			ptr->s = block->bb_u.s.bb_leftsib;
1214 	}
1215 }
1216 
1217 void
xfs_btree_set_sibling(struct xfs_btree_cur * cur,struct xfs_btree_block * block,const union xfs_btree_ptr * ptr,int lr)1218 xfs_btree_set_sibling(
1219 	struct xfs_btree_cur		*cur,
1220 	struct xfs_btree_block		*block,
1221 	const union xfs_btree_ptr	*ptr,
1222 	int				lr)
1223 {
1224 	ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
1225 
1226 	if (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN) {
1227 		if (lr == XFS_BB_RIGHTSIB)
1228 			block->bb_u.l.bb_rightsib = ptr->l;
1229 		else
1230 			block->bb_u.l.bb_leftsib = ptr->l;
1231 	} else {
1232 		if (lr == XFS_BB_RIGHTSIB)
1233 			block->bb_u.s.bb_rightsib = ptr->s;
1234 		else
1235 			block->bb_u.s.bb_leftsib = ptr->s;
1236 	}
1237 }
1238 
1239 static void
__xfs_btree_init_block(struct xfs_mount * mp,struct xfs_btree_block * buf,const struct xfs_btree_ops * ops,xfs_daddr_t blkno,__u16 level,__u16 numrecs,__u64 owner)1240 __xfs_btree_init_block(
1241 	struct xfs_mount	*mp,
1242 	struct xfs_btree_block	*buf,
1243 	const struct xfs_btree_ops *ops,
1244 	xfs_daddr_t		blkno,
1245 	__u16			level,
1246 	__u16			numrecs,
1247 	__u64			owner)
1248 {
1249 	bool			crc = xfs_has_crc(mp);
1250 	__u32			magic = xfs_btree_magic(mp, ops);
1251 
1252 	buf->bb_magic = cpu_to_be32(magic);
1253 	buf->bb_level = cpu_to_be16(level);
1254 	buf->bb_numrecs = cpu_to_be16(numrecs);
1255 
1256 	if (ops->ptr_len == XFS_BTREE_LONG_PTR_LEN) {
1257 		buf->bb_u.l.bb_leftsib = cpu_to_be64(NULLFSBLOCK);
1258 		buf->bb_u.l.bb_rightsib = cpu_to_be64(NULLFSBLOCK);
1259 		if (crc) {
1260 			buf->bb_u.l.bb_blkno = cpu_to_be64(blkno);
1261 			buf->bb_u.l.bb_owner = cpu_to_be64(owner);
1262 			uuid_copy(&buf->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid);
1263 			buf->bb_u.l.bb_pad = 0;
1264 			buf->bb_u.l.bb_lsn = 0;
1265 		}
1266 	} else {
1267 		buf->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
1268 		buf->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
1269 		if (crc) {
1270 			buf->bb_u.s.bb_blkno = cpu_to_be64(blkno);
1271 			/* owner is a 32 bit value on short blocks */
1272 			buf->bb_u.s.bb_owner = cpu_to_be32((__u32)owner);
1273 			uuid_copy(&buf->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid);
1274 			buf->bb_u.s.bb_lsn = 0;
1275 		}
1276 	}
1277 }
1278 
1279 void
xfs_btree_init_block(struct xfs_mount * mp,struct xfs_btree_block * block,const struct xfs_btree_ops * ops,__u16 level,__u16 numrecs,__u64 owner)1280 xfs_btree_init_block(
1281 	struct xfs_mount	*mp,
1282 	struct xfs_btree_block	*block,
1283 	const struct xfs_btree_ops *ops,
1284 	__u16			level,
1285 	__u16			numrecs,
1286 	__u64			owner)
1287 {
1288 	__xfs_btree_init_block(mp, block, ops, XFS_BUF_DADDR_NULL, level,
1289 			numrecs, owner);
1290 }
1291 
1292 void
xfs_btree_init_buf(struct xfs_mount * mp,struct xfs_buf * bp,const struct xfs_btree_ops * ops,__u16 level,__u16 numrecs,__u64 owner)1293 xfs_btree_init_buf(
1294 	struct xfs_mount		*mp,
1295 	struct xfs_buf			*bp,
1296 	const struct xfs_btree_ops	*ops,
1297 	__u16				level,
1298 	__u16				numrecs,
1299 	__u64				owner)
1300 {
1301 	__xfs_btree_init_block(mp, XFS_BUF_TO_BLOCK(bp), ops,
1302 			xfs_buf_daddr(bp), level, numrecs, owner);
1303 	bp->b_ops = ops->buf_ops;
1304 }
1305 
1306 static inline __u64
xfs_btree_owner(struct xfs_btree_cur * cur)1307 xfs_btree_owner(
1308 	struct xfs_btree_cur    *cur)
1309 {
1310 	switch (cur->bc_ops->type) {
1311 	case XFS_BTREE_TYPE_MEM:
1312 		return cur->bc_mem.xfbtree->owner;
1313 	case XFS_BTREE_TYPE_INODE:
1314 		return cur->bc_ino.ip->i_ino;
1315 	case XFS_BTREE_TYPE_AG:
1316 		return cur->bc_ag.pag->pag_agno;
1317 	default:
1318 		ASSERT(0);
1319 		return 0;
1320 	}
1321 }
1322 
1323 void
xfs_btree_init_block_cur(struct xfs_btree_cur * cur,struct xfs_buf * bp,int level,int numrecs)1324 xfs_btree_init_block_cur(
1325 	struct xfs_btree_cur	*cur,
1326 	struct xfs_buf		*bp,
1327 	int			level,
1328 	int			numrecs)
1329 {
1330 	xfs_btree_init_buf(cur->bc_mp, bp, cur->bc_ops, level, numrecs,
1331 			xfs_btree_owner(cur));
1332 }
1333 
1334 STATIC void
xfs_btree_buf_to_ptr(struct xfs_btree_cur * cur,struct xfs_buf * bp,union xfs_btree_ptr * ptr)1335 xfs_btree_buf_to_ptr(
1336 	struct xfs_btree_cur	*cur,
1337 	struct xfs_buf		*bp,
1338 	union xfs_btree_ptr	*ptr)
1339 {
1340 	switch (cur->bc_ops->type) {
1341 	case XFS_BTREE_TYPE_AG:
1342 		ptr->s = cpu_to_be32(xfs_daddr_to_agbno(cur->bc_mp,
1343 					xfs_buf_daddr(bp)));
1344 		break;
1345 	case XFS_BTREE_TYPE_INODE:
1346 		ptr->l = cpu_to_be64(XFS_DADDR_TO_FSB(cur->bc_mp,
1347 					xfs_buf_daddr(bp)));
1348 		break;
1349 	case XFS_BTREE_TYPE_MEM:
1350 		ptr->l = cpu_to_be64(xfs_daddr_to_xfbno(xfs_buf_daddr(bp)));
1351 		break;
1352 	}
1353 }
1354 
1355 static inline void
xfs_btree_set_refs(struct xfs_btree_cur * cur,struct xfs_buf * bp)1356 xfs_btree_set_refs(
1357 	struct xfs_btree_cur	*cur,
1358 	struct xfs_buf		*bp)
1359 {
1360 	xfs_buf_set_ref(bp, cur->bc_ops->lru_refs);
1361 }
1362 
1363 int
xfs_btree_get_buf_block(struct xfs_btree_cur * cur,const union xfs_btree_ptr * ptr,struct xfs_btree_block ** block,struct xfs_buf ** bpp)1364 xfs_btree_get_buf_block(
1365 	struct xfs_btree_cur		*cur,
1366 	const union xfs_btree_ptr	*ptr,
1367 	struct xfs_btree_block		**block,
1368 	struct xfs_buf			**bpp)
1369 {
1370 	xfs_daddr_t			d;
1371 	int				error;
1372 
1373 	error = xfs_btree_ptr_to_daddr(cur, ptr, &d);
1374 	if (error)
1375 		return error;
1376 	error = xfs_trans_get_buf(cur->bc_tp, xfs_btree_buftarg(cur), d,
1377 			xfs_btree_bbsize(cur), 0, bpp);
1378 	if (error)
1379 		return error;
1380 
1381 	(*bpp)->b_ops = cur->bc_ops->buf_ops;
1382 	*block = XFS_BUF_TO_BLOCK(*bpp);
1383 	return 0;
1384 }
1385 
1386 /*
1387  * Read in the buffer at the given ptr and return the buffer and
1388  * the block pointer within the buffer.
1389  */
1390 int
xfs_btree_read_buf_block(struct xfs_btree_cur * cur,const union xfs_btree_ptr * ptr,int flags,struct xfs_btree_block ** block,struct xfs_buf ** bpp)1391 xfs_btree_read_buf_block(
1392 	struct xfs_btree_cur		*cur,
1393 	const union xfs_btree_ptr	*ptr,
1394 	int				flags,
1395 	struct xfs_btree_block		**block,
1396 	struct xfs_buf			**bpp)
1397 {
1398 	struct xfs_mount	*mp = cur->bc_mp;
1399 	xfs_daddr_t		d;
1400 	int			error;
1401 
1402 	/* need to sort out how callers deal with failures first */
1403 	ASSERT(!(flags & XBF_TRYLOCK));
1404 
1405 	error = xfs_btree_ptr_to_daddr(cur, ptr, &d);
1406 	if (error)
1407 		return error;
1408 	error = xfs_trans_read_buf(mp, cur->bc_tp, xfs_btree_buftarg(cur), d,
1409 			xfs_btree_bbsize(cur), flags, bpp,
1410 			cur->bc_ops->buf_ops);
1411 	if (xfs_metadata_is_sick(error))
1412 		xfs_btree_mark_sick(cur);
1413 	if (error)
1414 		return error;
1415 
1416 	xfs_btree_set_refs(cur, *bpp);
1417 	*block = XFS_BUF_TO_BLOCK(*bpp);
1418 	return 0;
1419 }
1420 
1421 /*
1422  * Copy keys from one btree block to another.
1423  */
1424 void
xfs_btree_copy_keys(struct xfs_btree_cur * cur,union xfs_btree_key * dst_key,const union xfs_btree_key * src_key,int numkeys)1425 xfs_btree_copy_keys(
1426 	struct xfs_btree_cur		*cur,
1427 	union xfs_btree_key		*dst_key,
1428 	const union xfs_btree_key	*src_key,
1429 	int				numkeys)
1430 {
1431 	ASSERT(numkeys >= 0);
1432 	memcpy(dst_key, src_key, numkeys * cur->bc_ops->key_len);
1433 }
1434 
1435 /*
1436  * Copy records from one btree block to another.
1437  */
1438 STATIC void
xfs_btree_copy_recs(struct xfs_btree_cur * cur,union xfs_btree_rec * dst_rec,union xfs_btree_rec * src_rec,int numrecs)1439 xfs_btree_copy_recs(
1440 	struct xfs_btree_cur	*cur,
1441 	union xfs_btree_rec	*dst_rec,
1442 	union xfs_btree_rec	*src_rec,
1443 	int			numrecs)
1444 {
1445 	ASSERT(numrecs >= 0);
1446 	memcpy(dst_rec, src_rec, numrecs * cur->bc_ops->rec_len);
1447 }
1448 
1449 /*
1450  * Copy block pointers from one btree block to another.
1451  */
1452 void
xfs_btree_copy_ptrs(struct xfs_btree_cur * cur,union xfs_btree_ptr * dst_ptr,const union xfs_btree_ptr * src_ptr,int numptrs)1453 xfs_btree_copy_ptrs(
1454 	struct xfs_btree_cur	*cur,
1455 	union xfs_btree_ptr	*dst_ptr,
1456 	const union xfs_btree_ptr *src_ptr,
1457 	int			numptrs)
1458 {
1459 	ASSERT(numptrs >= 0);
1460 	memcpy(dst_ptr, src_ptr, numptrs * cur->bc_ops->ptr_len);
1461 }
1462 
1463 /*
1464  * Shift keys one index left/right inside a single btree block.
1465  */
1466 STATIC void
xfs_btree_shift_keys(struct xfs_btree_cur * cur,union xfs_btree_key * key,int dir,int numkeys)1467 xfs_btree_shift_keys(
1468 	struct xfs_btree_cur	*cur,
1469 	union xfs_btree_key	*key,
1470 	int			dir,
1471 	int			numkeys)
1472 {
1473 	char			*dst_key;
1474 
1475 	ASSERT(numkeys >= 0);
1476 	ASSERT(dir == 1 || dir == -1);
1477 
1478 	dst_key = (char *)key + (dir * cur->bc_ops->key_len);
1479 	memmove(dst_key, key, numkeys * cur->bc_ops->key_len);
1480 }
1481 
1482 /*
1483  * Shift records one index left/right inside a single btree block.
1484  */
1485 STATIC void
xfs_btree_shift_recs(struct xfs_btree_cur * cur,union xfs_btree_rec * rec,int dir,int numrecs)1486 xfs_btree_shift_recs(
1487 	struct xfs_btree_cur	*cur,
1488 	union xfs_btree_rec	*rec,
1489 	int			dir,
1490 	int			numrecs)
1491 {
1492 	char			*dst_rec;
1493 
1494 	ASSERT(numrecs >= 0);
1495 	ASSERT(dir == 1 || dir == -1);
1496 
1497 	dst_rec = (char *)rec + (dir * cur->bc_ops->rec_len);
1498 	memmove(dst_rec, rec, numrecs * cur->bc_ops->rec_len);
1499 }
1500 
1501 /*
1502  * Shift block pointers one index left/right inside a single btree block.
1503  */
1504 STATIC void
xfs_btree_shift_ptrs(struct xfs_btree_cur * cur,union xfs_btree_ptr * ptr,int dir,int numptrs)1505 xfs_btree_shift_ptrs(
1506 	struct xfs_btree_cur	*cur,
1507 	union xfs_btree_ptr	*ptr,
1508 	int			dir,
1509 	int			numptrs)
1510 {
1511 	char			*dst_ptr;
1512 
1513 	ASSERT(numptrs >= 0);
1514 	ASSERT(dir == 1 || dir == -1);
1515 
1516 	dst_ptr = (char *)ptr + (dir * cur->bc_ops->ptr_len);
1517 	memmove(dst_ptr, ptr, numptrs * cur->bc_ops->ptr_len);
1518 }
1519 
1520 /*
1521  * Log key values from the btree block.
1522  */
1523 STATIC void
xfs_btree_log_keys(struct xfs_btree_cur * cur,struct xfs_buf * bp,int first,int last)1524 xfs_btree_log_keys(
1525 	struct xfs_btree_cur	*cur,
1526 	struct xfs_buf		*bp,
1527 	int			first,
1528 	int			last)
1529 {
1530 
1531 	if (bp) {
1532 		xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1533 		xfs_trans_log_buf(cur->bc_tp, bp,
1534 				  xfs_btree_key_offset(cur, first),
1535 				  xfs_btree_key_offset(cur, last + 1) - 1);
1536 	} else {
1537 		xfs_trans_log_inode(cur->bc_tp, cur->bc_ino.ip,
1538 				xfs_ilog_fbroot(cur->bc_ino.whichfork));
1539 	}
1540 }
1541 
1542 /*
1543  * Log record values from the btree block.
1544  */
1545 void
xfs_btree_log_recs(struct xfs_btree_cur * cur,struct xfs_buf * bp,int first,int last)1546 xfs_btree_log_recs(
1547 	struct xfs_btree_cur	*cur,
1548 	struct xfs_buf		*bp,
1549 	int			first,
1550 	int			last)
1551 {
1552 
1553 	xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1554 	xfs_trans_log_buf(cur->bc_tp, bp,
1555 			  xfs_btree_rec_offset(cur, first),
1556 			  xfs_btree_rec_offset(cur, last + 1) - 1);
1557 
1558 }
1559 
1560 /*
1561  * Log block pointer fields from a btree block (nonleaf).
1562  */
1563 STATIC void
xfs_btree_log_ptrs(struct xfs_btree_cur * cur,struct xfs_buf * bp,int first,int last)1564 xfs_btree_log_ptrs(
1565 	struct xfs_btree_cur	*cur,	/* btree cursor */
1566 	struct xfs_buf		*bp,	/* buffer containing btree block */
1567 	int			first,	/* index of first pointer to log */
1568 	int			last)	/* index of last pointer to log */
1569 {
1570 
1571 	if (bp) {
1572 		struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
1573 		int			level = xfs_btree_get_level(block);
1574 
1575 		xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1576 		xfs_trans_log_buf(cur->bc_tp, bp,
1577 				xfs_btree_ptr_offset(cur, first, level),
1578 				xfs_btree_ptr_offset(cur, last + 1, level) - 1);
1579 	} else {
1580 		xfs_trans_log_inode(cur->bc_tp, cur->bc_ino.ip,
1581 			xfs_ilog_fbroot(cur->bc_ino.whichfork));
1582 	}
1583 
1584 }
1585 
1586 /*
1587  * Log fields from a btree block header.
1588  */
1589 void
xfs_btree_log_block(struct xfs_btree_cur * cur,struct xfs_buf * bp,uint32_t fields)1590 xfs_btree_log_block(
1591 	struct xfs_btree_cur	*cur,	/* btree cursor */
1592 	struct xfs_buf		*bp,	/* buffer containing btree block */
1593 	uint32_t		fields)	/* mask of fields: XFS_BB_... */
1594 {
1595 	int			first;	/* first byte offset logged */
1596 	int			last;	/* last byte offset logged */
1597 	static const short	soffsets[] = {	/* table of offsets (short) */
1598 		offsetof(struct xfs_btree_block, bb_magic),
1599 		offsetof(struct xfs_btree_block, bb_level),
1600 		offsetof(struct xfs_btree_block, bb_numrecs),
1601 		offsetof(struct xfs_btree_block, bb_u.s.bb_leftsib),
1602 		offsetof(struct xfs_btree_block, bb_u.s.bb_rightsib),
1603 		offsetof(struct xfs_btree_block, bb_u.s.bb_blkno),
1604 		offsetof(struct xfs_btree_block, bb_u.s.bb_lsn),
1605 		offsetof(struct xfs_btree_block, bb_u.s.bb_uuid),
1606 		offsetof(struct xfs_btree_block, bb_u.s.bb_owner),
1607 		offsetof(struct xfs_btree_block, bb_u.s.bb_crc),
1608 		XFS_BTREE_SBLOCK_CRC_LEN
1609 	};
1610 	static const short	loffsets[] = {	/* table of offsets (long) */
1611 		offsetof(struct xfs_btree_block, bb_magic),
1612 		offsetof(struct xfs_btree_block, bb_level),
1613 		offsetof(struct xfs_btree_block, bb_numrecs),
1614 		offsetof(struct xfs_btree_block, bb_u.l.bb_leftsib),
1615 		offsetof(struct xfs_btree_block, bb_u.l.bb_rightsib),
1616 		offsetof(struct xfs_btree_block, bb_u.l.bb_blkno),
1617 		offsetof(struct xfs_btree_block, bb_u.l.bb_lsn),
1618 		offsetof(struct xfs_btree_block, bb_u.l.bb_uuid),
1619 		offsetof(struct xfs_btree_block, bb_u.l.bb_owner),
1620 		offsetof(struct xfs_btree_block, bb_u.l.bb_crc),
1621 		offsetof(struct xfs_btree_block, bb_u.l.bb_pad),
1622 		XFS_BTREE_LBLOCK_CRC_LEN
1623 	};
1624 
1625 	if (bp) {
1626 		int nbits;
1627 
1628 		if (xfs_has_crc(cur->bc_mp)) {
1629 			/*
1630 			 * We don't log the CRC when updating a btree
1631 			 * block but instead recreate it during log
1632 			 * recovery.  As the log buffers have checksums
1633 			 * of their own this is safe and avoids logging a crc
1634 			 * update in a lot of places.
1635 			 */
1636 			if (fields == XFS_BB_ALL_BITS)
1637 				fields = XFS_BB_ALL_BITS_CRC;
1638 			nbits = XFS_BB_NUM_BITS_CRC;
1639 		} else {
1640 			nbits = XFS_BB_NUM_BITS;
1641 		}
1642 		xfs_btree_offsets(fields,
1643 				  (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN) ?
1644 					loffsets : soffsets,
1645 				  nbits, &first, &last);
1646 		xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
1647 		xfs_trans_log_buf(cur->bc_tp, bp, first, last);
1648 	} else {
1649 		xfs_trans_log_inode(cur->bc_tp, cur->bc_ino.ip,
1650 			xfs_ilog_fbroot(cur->bc_ino.whichfork));
1651 	}
1652 }
1653 
1654 /*
1655  * Increment cursor by one record at the level.
1656  * For nonzero levels the leaf-ward information is untouched.
1657  */
1658 int						/* error */
xfs_btree_increment(struct xfs_btree_cur * cur,int level,int * stat)1659 xfs_btree_increment(
1660 	struct xfs_btree_cur	*cur,
1661 	int			level,
1662 	int			*stat)		/* success/failure */
1663 {
1664 	struct xfs_btree_block	*block;
1665 	union xfs_btree_ptr	ptr;
1666 	struct xfs_buf		*bp;
1667 	int			error;		/* error return value */
1668 	int			lev;
1669 
1670 	ASSERT(level < cur->bc_nlevels);
1671 
1672 	/* Read-ahead to the right at this level. */
1673 	xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
1674 
1675 	/* Get a pointer to the btree block. */
1676 	block = xfs_btree_get_block(cur, level, &bp);
1677 
1678 #ifdef DEBUG
1679 	error = xfs_btree_check_block(cur, block, level, bp);
1680 	if (error)
1681 		goto error0;
1682 #endif
1683 
1684 	/* We're done if we remain in the block after the increment. */
1685 	if (++cur->bc_levels[level].ptr <= xfs_btree_get_numrecs(block))
1686 		goto out1;
1687 
1688 	/* Fail if we just went off the right edge of the tree. */
1689 	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
1690 	if (xfs_btree_ptr_is_null(cur, &ptr))
1691 		goto out0;
1692 
1693 	XFS_BTREE_STATS_INC(cur, increment);
1694 
1695 	/*
1696 	 * March up the tree incrementing pointers.
1697 	 * Stop when we don't go off the right edge of a block.
1698 	 */
1699 	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
1700 		block = xfs_btree_get_block(cur, lev, &bp);
1701 
1702 #ifdef DEBUG
1703 		error = xfs_btree_check_block(cur, block, lev, bp);
1704 		if (error)
1705 			goto error0;
1706 #endif
1707 
1708 		if (++cur->bc_levels[lev].ptr <= xfs_btree_get_numrecs(block))
1709 			break;
1710 
1711 		/* Read-ahead the right block for the next loop. */
1712 		xfs_btree_readahead(cur, lev, XFS_BTCUR_RIGHTRA);
1713 	}
1714 
1715 	/*
1716 	 * If we went off the root then we are either seriously
1717 	 * confused or have the tree root in an inode.
1718 	 */
1719 	if (lev == cur->bc_nlevels) {
1720 		if (cur->bc_ops->type == XFS_BTREE_TYPE_INODE)
1721 			goto out0;
1722 		ASSERT(0);
1723 		xfs_btree_mark_sick(cur);
1724 		error = -EFSCORRUPTED;
1725 		goto error0;
1726 	}
1727 	ASSERT(lev < cur->bc_nlevels);
1728 
1729 	/*
1730 	 * Now walk back down the tree, fixing up the cursor's buffer
1731 	 * pointers and key numbers.
1732 	 */
1733 	for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
1734 		union xfs_btree_ptr	*ptrp;
1735 
1736 		ptrp = xfs_btree_ptr_addr(cur, cur->bc_levels[lev].ptr, block);
1737 		--lev;
1738 		error = xfs_btree_read_buf_block(cur, ptrp, 0, &block, &bp);
1739 		if (error)
1740 			goto error0;
1741 
1742 		xfs_btree_setbuf(cur, lev, bp);
1743 		cur->bc_levels[lev].ptr = 1;
1744 	}
1745 out1:
1746 	*stat = 1;
1747 	return 0;
1748 
1749 out0:
1750 	*stat = 0;
1751 	return 0;
1752 
1753 error0:
1754 	return error;
1755 }
1756 
1757 /*
1758  * Decrement cursor by one record at the level.
1759  * For nonzero levels the leaf-ward information is untouched.
1760  */
1761 int						/* error */
xfs_btree_decrement(struct xfs_btree_cur * cur,int level,int * stat)1762 xfs_btree_decrement(
1763 	struct xfs_btree_cur	*cur,
1764 	int			level,
1765 	int			*stat)		/* success/failure */
1766 {
1767 	struct xfs_btree_block	*block;
1768 	struct xfs_buf		*bp;
1769 	int			error;		/* error return value */
1770 	int			lev;
1771 	union xfs_btree_ptr	ptr;
1772 
1773 	ASSERT(level < cur->bc_nlevels);
1774 
1775 	/* Read-ahead to the left at this level. */
1776 	xfs_btree_readahead(cur, level, XFS_BTCUR_LEFTRA);
1777 
1778 	/* We're done if we remain in the block after the decrement. */
1779 	if (--cur->bc_levels[level].ptr > 0)
1780 		goto out1;
1781 
1782 	/* Get a pointer to the btree block. */
1783 	block = xfs_btree_get_block(cur, level, &bp);
1784 
1785 #ifdef DEBUG
1786 	error = xfs_btree_check_block(cur, block, level, bp);
1787 	if (error)
1788 		goto error0;
1789 #endif
1790 
1791 	/* Fail if we just went off the left edge of the tree. */
1792 	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
1793 	if (xfs_btree_ptr_is_null(cur, &ptr))
1794 		goto out0;
1795 
1796 	XFS_BTREE_STATS_INC(cur, decrement);
1797 
1798 	/*
1799 	 * March up the tree decrementing pointers.
1800 	 * Stop when we don't go off the left edge of a block.
1801 	 */
1802 	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
1803 		if (--cur->bc_levels[lev].ptr > 0)
1804 			break;
1805 		/* Read-ahead the left block for the next loop. */
1806 		xfs_btree_readahead(cur, lev, XFS_BTCUR_LEFTRA);
1807 	}
1808 
1809 	/*
1810 	 * If we went off the root then we are seriously confused.
1811 	 * or the root of the tree is in an inode.
1812 	 */
1813 	if (lev == cur->bc_nlevels) {
1814 		if (cur->bc_ops->type == XFS_BTREE_TYPE_INODE)
1815 			goto out0;
1816 		ASSERT(0);
1817 		xfs_btree_mark_sick(cur);
1818 		error = -EFSCORRUPTED;
1819 		goto error0;
1820 	}
1821 	ASSERT(lev < cur->bc_nlevels);
1822 
1823 	/*
1824 	 * Now walk back down the tree, fixing up the cursor's buffer
1825 	 * pointers and key numbers.
1826 	 */
1827 	for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
1828 		union xfs_btree_ptr	*ptrp;
1829 
1830 		ptrp = xfs_btree_ptr_addr(cur, cur->bc_levels[lev].ptr, block);
1831 		--lev;
1832 		error = xfs_btree_read_buf_block(cur, ptrp, 0, &block, &bp);
1833 		if (error)
1834 			goto error0;
1835 		xfs_btree_setbuf(cur, lev, bp);
1836 		cur->bc_levels[lev].ptr = xfs_btree_get_numrecs(block);
1837 	}
1838 out1:
1839 	*stat = 1;
1840 	return 0;
1841 
1842 out0:
1843 	*stat = 0;
1844 	return 0;
1845 
1846 error0:
1847 	return error;
1848 }
1849 
1850 /*
1851  * Check the btree block owner now that we have the context to know who the
1852  * real owner is.
1853  */
1854 static inline xfs_failaddr_t
xfs_btree_check_block_owner(struct xfs_btree_cur * cur,struct xfs_btree_block * block)1855 xfs_btree_check_block_owner(
1856 	struct xfs_btree_cur	*cur,
1857 	struct xfs_btree_block	*block)
1858 {
1859 	__u64			owner;
1860 
1861 	if (!xfs_has_crc(cur->bc_mp) ||
1862 	    (cur->bc_flags & XFS_BTREE_BMBT_INVALID_OWNER))
1863 		return NULL;
1864 
1865 	owner = xfs_btree_owner(cur);
1866 	if (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN) {
1867 		if (be64_to_cpu(block->bb_u.l.bb_owner) != owner)
1868 			return __this_address;
1869 	} else {
1870 		if (be32_to_cpu(block->bb_u.s.bb_owner) != owner)
1871 			return __this_address;
1872 	}
1873 
1874 	return NULL;
1875 }
1876 
1877 int
xfs_btree_lookup_get_block(struct xfs_btree_cur * cur,int level,const union xfs_btree_ptr * pp,struct xfs_btree_block ** blkp)1878 xfs_btree_lookup_get_block(
1879 	struct xfs_btree_cur		*cur,	/* btree cursor */
1880 	int				level,	/* level in the btree */
1881 	const union xfs_btree_ptr	*pp,	/* ptr to btree block */
1882 	struct xfs_btree_block		**blkp) /* return btree block */
1883 {
1884 	struct xfs_buf		*bp;	/* buffer pointer for btree block */
1885 	xfs_daddr_t		daddr;
1886 	int			error = 0;
1887 
1888 	/* special case the root block if in an inode */
1889 	if (xfs_btree_at_iroot(cur, level)) {
1890 		*blkp = xfs_btree_get_iroot(cur);
1891 		return 0;
1892 	}
1893 
1894 	/*
1895 	 * If the old buffer at this level for the disk address we are
1896 	 * looking for re-use it.
1897 	 *
1898 	 * Otherwise throw it away and get a new one.
1899 	 */
1900 	bp = cur->bc_levels[level].bp;
1901 	error = xfs_btree_ptr_to_daddr(cur, pp, &daddr);
1902 	if (error)
1903 		return error;
1904 	if (bp && xfs_buf_daddr(bp) == daddr) {
1905 		*blkp = XFS_BUF_TO_BLOCK(bp);
1906 		return 0;
1907 	}
1908 
1909 	error = xfs_btree_read_buf_block(cur, pp, 0, blkp, &bp);
1910 	if (error)
1911 		return error;
1912 
1913 	/* Check the inode owner since the verifiers don't. */
1914 	if (xfs_btree_check_block_owner(cur, *blkp) != NULL)
1915 		goto out_bad;
1916 
1917 	/* Did we get the level we were looking for? */
1918 	if (be16_to_cpu((*blkp)->bb_level) != level)
1919 		goto out_bad;
1920 
1921 	/* Check that internal nodes have at least one record. */
1922 	if (level != 0 && be16_to_cpu((*blkp)->bb_numrecs) == 0)
1923 		goto out_bad;
1924 
1925 	xfs_btree_setbuf(cur, level, bp);
1926 	return 0;
1927 
1928 out_bad:
1929 	*blkp = NULL;
1930 	xfs_buf_mark_corrupt(bp);
1931 	xfs_trans_brelse(cur->bc_tp, bp);
1932 	xfs_btree_mark_sick(cur);
1933 	return -EFSCORRUPTED;
1934 }
1935 
1936 /*
1937  * Get current search key.  For level 0 we don't actually have a key
1938  * structure so we make one up from the record.  For all other levels
1939  * we just return the right key.
1940  */
1941 STATIC union xfs_btree_key *
xfs_lookup_get_search_key(struct xfs_btree_cur * cur,int level,int keyno,struct xfs_btree_block * block,union xfs_btree_key * kp)1942 xfs_lookup_get_search_key(
1943 	struct xfs_btree_cur	*cur,
1944 	int			level,
1945 	int			keyno,
1946 	struct xfs_btree_block	*block,
1947 	union xfs_btree_key	*kp)
1948 {
1949 	if (level == 0) {
1950 		cur->bc_ops->init_key_from_rec(kp,
1951 				xfs_btree_rec_addr(cur, keyno, block));
1952 		return kp;
1953 	}
1954 
1955 	return xfs_btree_key_addr(cur, keyno, block);
1956 }
1957 
1958 /*
1959  * Initialize a pointer to the root block.
1960  */
1961 void
xfs_btree_init_ptr_from_cur(struct xfs_btree_cur * cur,union xfs_btree_ptr * ptr)1962 xfs_btree_init_ptr_from_cur(
1963 	struct xfs_btree_cur	*cur,
1964 	union xfs_btree_ptr	*ptr)
1965 {
1966 	if (cur->bc_ops->type == XFS_BTREE_TYPE_INODE) {
1967 		/*
1968 		 * Inode-rooted btrees call xfs_btree_get_iroot to find the root
1969 		 * in xfs_btree_lookup_get_block and don't need a pointer here.
1970 		 */
1971 		ptr->l = 0;
1972 	} else if (cur->bc_flags & XFS_BTREE_STAGING) {
1973 		ptr->s = cpu_to_be32(cur->bc_ag.afake->af_root);
1974 	} else {
1975 		cur->bc_ops->init_ptr_from_cur(cur, ptr);
1976 	}
1977 }
1978 
1979 /*
1980  * Lookup the record.  The cursor is made to point to it, based on dir.
1981  * stat is set to 0 if can't find any such record, 1 for success.
1982  */
1983 int					/* error */
xfs_btree_lookup(struct xfs_btree_cur * cur,xfs_lookup_t dir,int * stat)1984 xfs_btree_lookup(
1985 	struct xfs_btree_cur	*cur,	/* btree cursor */
1986 	xfs_lookup_t		dir,	/* <=, ==, or >= */
1987 	int			*stat)	/* success/failure */
1988 {
1989 	struct xfs_btree_block	*block;	/* current btree block */
1990 	int64_t			diff;	/* difference for the current key */
1991 	int			error;	/* error return value */
1992 	int			keyno;	/* current key number */
1993 	int			level;	/* level in the btree */
1994 	union xfs_btree_ptr	*pp;	/* ptr to btree block */
1995 	union xfs_btree_ptr	ptr;	/* ptr to btree block */
1996 
1997 	XFS_BTREE_STATS_INC(cur, lookup);
1998 
1999 	/* No such thing as a zero-level tree. */
2000 	if (XFS_IS_CORRUPT(cur->bc_mp, cur->bc_nlevels == 0)) {
2001 		xfs_btree_mark_sick(cur);
2002 		return -EFSCORRUPTED;
2003 	}
2004 
2005 	block = NULL;
2006 	keyno = 0;
2007 
2008 	/* initialise start pointer from cursor */
2009 	xfs_btree_init_ptr_from_cur(cur, &ptr);
2010 	pp = &ptr;
2011 
2012 	/*
2013 	 * Iterate over each level in the btree, starting at the root.
2014 	 * For each level above the leaves, find the key we need, based
2015 	 * on the lookup record, then follow the corresponding block
2016 	 * pointer down to the next level.
2017 	 */
2018 	for (level = cur->bc_nlevels - 1, diff = 1; level >= 0; level--) {
2019 		/* Get the block we need to do the lookup on. */
2020 		error = xfs_btree_lookup_get_block(cur, level, pp, &block);
2021 		if (error)
2022 			goto error0;
2023 
2024 		if (diff == 0) {
2025 			/*
2026 			 * If we already had a key match at a higher level, we
2027 			 * know we need to use the first entry in this block.
2028 			 */
2029 			keyno = 1;
2030 		} else {
2031 			/* Otherwise search this block. Do a binary search. */
2032 
2033 			int	high;	/* high entry number */
2034 			int	low;	/* low entry number */
2035 
2036 			/* Set low and high entry numbers, 1-based. */
2037 			low = 1;
2038 			high = xfs_btree_get_numrecs(block);
2039 			if (!high) {
2040 				/* Block is empty, must be an empty leaf. */
2041 				if (level != 0 || cur->bc_nlevels != 1) {
2042 					XFS_CORRUPTION_ERROR(__func__,
2043 							XFS_ERRLEVEL_LOW,
2044 							cur->bc_mp, block,
2045 							sizeof(*block));
2046 					xfs_btree_mark_sick(cur);
2047 					return -EFSCORRUPTED;
2048 				}
2049 
2050 				cur->bc_levels[0].ptr = dir != XFS_LOOKUP_LE;
2051 				*stat = 0;
2052 				return 0;
2053 			}
2054 
2055 			/* Binary search the block. */
2056 			while (low <= high) {
2057 				union xfs_btree_key	key;
2058 				union xfs_btree_key	*kp;
2059 
2060 				XFS_BTREE_STATS_INC(cur, compare);
2061 
2062 				/* keyno is average of low and high. */
2063 				keyno = (low + high) >> 1;
2064 
2065 				/* Get current search key */
2066 				kp = xfs_lookup_get_search_key(cur, level,
2067 						keyno, block, &key);
2068 
2069 				/*
2070 				 * Compute difference to get next direction:
2071 				 *  - less than, move right
2072 				 *  - greater than, move left
2073 				 *  - equal, we're done
2074 				 */
2075 				diff = cur->bc_ops->key_diff(cur, kp);
2076 				if (diff < 0)
2077 					low = keyno + 1;
2078 				else if (diff > 0)
2079 					high = keyno - 1;
2080 				else
2081 					break;
2082 			}
2083 		}
2084 
2085 		/*
2086 		 * If there are more levels, set up for the next level
2087 		 * by getting the block number and filling in the cursor.
2088 		 */
2089 		if (level > 0) {
2090 			/*
2091 			 * If we moved left, need the previous key number,
2092 			 * unless there isn't one.
2093 			 */
2094 			if (diff > 0 && --keyno < 1)
2095 				keyno = 1;
2096 			pp = xfs_btree_ptr_addr(cur, keyno, block);
2097 
2098 			error = xfs_btree_debug_check_ptr(cur, pp, 0, level);
2099 			if (error)
2100 				goto error0;
2101 
2102 			cur->bc_levels[level].ptr = keyno;
2103 		}
2104 	}
2105 
2106 	/* Done with the search. See if we need to adjust the results. */
2107 	if (dir != XFS_LOOKUP_LE && diff < 0) {
2108 		keyno++;
2109 		/*
2110 		 * If ge search and we went off the end of the block, but it's
2111 		 * not the last block, we're in the wrong block.
2112 		 */
2113 		xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
2114 		if (dir == XFS_LOOKUP_GE &&
2115 		    keyno > xfs_btree_get_numrecs(block) &&
2116 		    !xfs_btree_ptr_is_null(cur, &ptr)) {
2117 			int	i;
2118 
2119 			cur->bc_levels[0].ptr = keyno;
2120 			error = xfs_btree_increment(cur, 0, &i);
2121 			if (error)
2122 				goto error0;
2123 			if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
2124 				xfs_btree_mark_sick(cur);
2125 				return -EFSCORRUPTED;
2126 			}
2127 			*stat = 1;
2128 			return 0;
2129 		}
2130 	} else if (dir == XFS_LOOKUP_LE && diff > 0)
2131 		keyno--;
2132 	cur->bc_levels[0].ptr = keyno;
2133 
2134 	/* Return if we succeeded or not. */
2135 	if (keyno == 0 || keyno > xfs_btree_get_numrecs(block))
2136 		*stat = 0;
2137 	else if (dir != XFS_LOOKUP_EQ || diff == 0)
2138 		*stat = 1;
2139 	else
2140 		*stat = 0;
2141 	return 0;
2142 
2143 error0:
2144 	return error;
2145 }
2146 
2147 /* Find the high key storage area from a regular key. */
2148 union xfs_btree_key *
xfs_btree_high_key_from_key(struct xfs_btree_cur * cur,union xfs_btree_key * key)2149 xfs_btree_high_key_from_key(
2150 	struct xfs_btree_cur	*cur,
2151 	union xfs_btree_key	*key)
2152 {
2153 	ASSERT(cur->bc_ops->geom_flags & XFS_BTGEO_OVERLAPPING);
2154 	return (union xfs_btree_key *)((char *)key +
2155 			(cur->bc_ops->key_len / 2));
2156 }
2157 
2158 /* Determine the low (and high if overlapped) keys of a leaf block */
2159 STATIC void
xfs_btree_get_leaf_keys(struct xfs_btree_cur * cur,struct xfs_btree_block * block,union xfs_btree_key * key)2160 xfs_btree_get_leaf_keys(
2161 	struct xfs_btree_cur	*cur,
2162 	struct xfs_btree_block	*block,
2163 	union xfs_btree_key	*key)
2164 {
2165 	union xfs_btree_key	max_hkey;
2166 	union xfs_btree_key	hkey;
2167 	union xfs_btree_rec	*rec;
2168 	union xfs_btree_key	*high;
2169 	int			n;
2170 
2171 	rec = xfs_btree_rec_addr(cur, 1, block);
2172 	cur->bc_ops->init_key_from_rec(key, rec);
2173 
2174 	if (cur->bc_ops->geom_flags & XFS_BTGEO_OVERLAPPING) {
2175 
2176 		cur->bc_ops->init_high_key_from_rec(&max_hkey, rec);
2177 		for (n = 2; n <= xfs_btree_get_numrecs(block); n++) {
2178 			rec = xfs_btree_rec_addr(cur, n, block);
2179 			cur->bc_ops->init_high_key_from_rec(&hkey, rec);
2180 			if (xfs_btree_keycmp_gt(cur, &hkey, &max_hkey))
2181 				max_hkey = hkey;
2182 		}
2183 
2184 		high = xfs_btree_high_key_from_key(cur, key);
2185 		memcpy(high, &max_hkey, cur->bc_ops->key_len / 2);
2186 	}
2187 }
2188 
2189 /* Determine the low (and high if overlapped) keys of a node block */
2190 STATIC void
xfs_btree_get_node_keys(struct xfs_btree_cur * cur,struct xfs_btree_block * block,union xfs_btree_key * key)2191 xfs_btree_get_node_keys(
2192 	struct xfs_btree_cur	*cur,
2193 	struct xfs_btree_block	*block,
2194 	union xfs_btree_key	*key)
2195 {
2196 	union xfs_btree_key	*hkey;
2197 	union xfs_btree_key	*max_hkey;
2198 	union xfs_btree_key	*high;
2199 	int			n;
2200 
2201 	if (cur->bc_ops->geom_flags & XFS_BTGEO_OVERLAPPING) {
2202 		memcpy(key, xfs_btree_key_addr(cur, 1, block),
2203 				cur->bc_ops->key_len / 2);
2204 
2205 		max_hkey = xfs_btree_high_key_addr(cur, 1, block);
2206 		for (n = 2; n <= xfs_btree_get_numrecs(block); n++) {
2207 			hkey = xfs_btree_high_key_addr(cur, n, block);
2208 			if (xfs_btree_keycmp_gt(cur, hkey, max_hkey))
2209 				max_hkey = hkey;
2210 		}
2211 
2212 		high = xfs_btree_high_key_from_key(cur, key);
2213 		memcpy(high, max_hkey, cur->bc_ops->key_len / 2);
2214 	} else {
2215 		memcpy(key, xfs_btree_key_addr(cur, 1, block),
2216 				cur->bc_ops->key_len);
2217 	}
2218 }
2219 
2220 /* Derive the keys for any btree block. */
2221 void
xfs_btree_get_keys(struct xfs_btree_cur * cur,struct xfs_btree_block * block,union xfs_btree_key * key)2222 xfs_btree_get_keys(
2223 	struct xfs_btree_cur	*cur,
2224 	struct xfs_btree_block	*block,
2225 	union xfs_btree_key	*key)
2226 {
2227 	if (be16_to_cpu(block->bb_level) == 0)
2228 		xfs_btree_get_leaf_keys(cur, block, key);
2229 	else
2230 		xfs_btree_get_node_keys(cur, block, key);
2231 }
2232 
2233 /*
2234  * Decide if we need to update the parent keys of a btree block.  For
2235  * a standard btree this is only necessary if we're updating the first
2236  * record/key.  For an overlapping btree, we must always update the
2237  * keys because the highest key can be in any of the records or keys
2238  * in the block.
2239  */
2240 static inline bool
xfs_btree_needs_key_update(struct xfs_btree_cur * cur,int ptr)2241 xfs_btree_needs_key_update(
2242 	struct xfs_btree_cur	*cur,
2243 	int			ptr)
2244 {
2245 	return (cur->bc_ops->geom_flags & XFS_BTGEO_OVERLAPPING) || ptr == 1;
2246 }
2247 
2248 /*
2249  * Update the low and high parent keys of the given level, progressing
2250  * towards the root.  If force_all is false, stop if the keys for a given
2251  * level do not need updating.
2252  */
2253 STATIC int
__xfs_btree_updkeys(struct xfs_btree_cur * cur,int level,struct xfs_btree_block * block,struct xfs_buf * bp0,bool force_all)2254 __xfs_btree_updkeys(
2255 	struct xfs_btree_cur	*cur,
2256 	int			level,
2257 	struct xfs_btree_block	*block,
2258 	struct xfs_buf		*bp0,
2259 	bool			force_all)
2260 {
2261 	union xfs_btree_key	key;	/* keys from current level */
2262 	union xfs_btree_key	*lkey;	/* keys from the next level up */
2263 	union xfs_btree_key	*hkey;
2264 	union xfs_btree_key	*nlkey;	/* keys from the next level up */
2265 	union xfs_btree_key	*nhkey;
2266 	struct xfs_buf		*bp;
2267 	int			ptr;
2268 
2269 	ASSERT(cur->bc_ops->geom_flags & XFS_BTGEO_OVERLAPPING);
2270 
2271 	/* Exit if there aren't any parent levels to update. */
2272 	if (level + 1 >= cur->bc_nlevels)
2273 		return 0;
2274 
2275 	trace_xfs_btree_updkeys(cur, level, bp0);
2276 
2277 	lkey = &key;
2278 	hkey = xfs_btree_high_key_from_key(cur, lkey);
2279 	xfs_btree_get_keys(cur, block, lkey);
2280 	for (level++; level < cur->bc_nlevels; level++) {
2281 #ifdef DEBUG
2282 		int		error;
2283 #endif
2284 		block = xfs_btree_get_block(cur, level, &bp);
2285 		trace_xfs_btree_updkeys(cur, level, bp);
2286 #ifdef DEBUG
2287 		error = xfs_btree_check_block(cur, block, level, bp);
2288 		if (error)
2289 			return error;
2290 #endif
2291 		ptr = cur->bc_levels[level].ptr;
2292 		nlkey = xfs_btree_key_addr(cur, ptr, block);
2293 		nhkey = xfs_btree_high_key_addr(cur, ptr, block);
2294 		if (!force_all &&
2295 		    xfs_btree_keycmp_eq(cur, nlkey, lkey) &&
2296 		    xfs_btree_keycmp_eq(cur, nhkey, hkey))
2297 			break;
2298 		xfs_btree_copy_keys(cur, nlkey, lkey, 1);
2299 		xfs_btree_log_keys(cur, bp, ptr, ptr);
2300 		if (level + 1 >= cur->bc_nlevels)
2301 			break;
2302 		xfs_btree_get_node_keys(cur, block, lkey);
2303 	}
2304 
2305 	return 0;
2306 }
2307 
2308 /* Update all the keys from some level in cursor back to the root. */
2309 STATIC int
xfs_btree_updkeys_force(struct xfs_btree_cur * cur,int level)2310 xfs_btree_updkeys_force(
2311 	struct xfs_btree_cur	*cur,
2312 	int			level)
2313 {
2314 	struct xfs_buf		*bp;
2315 	struct xfs_btree_block	*block;
2316 
2317 	block = xfs_btree_get_block(cur, level, &bp);
2318 	return __xfs_btree_updkeys(cur, level, block, bp, true);
2319 }
2320 
2321 /*
2322  * Update the parent keys of the given level, progressing towards the root.
2323  */
2324 STATIC int
xfs_btree_update_keys(struct xfs_btree_cur * cur,int level)2325 xfs_btree_update_keys(
2326 	struct xfs_btree_cur	*cur,
2327 	int			level)
2328 {
2329 	struct xfs_btree_block	*block;
2330 	struct xfs_buf		*bp;
2331 	union xfs_btree_key	*kp;
2332 	union xfs_btree_key	key;
2333 	int			ptr;
2334 
2335 	ASSERT(level >= 0);
2336 
2337 	block = xfs_btree_get_block(cur, level, &bp);
2338 	if (cur->bc_ops->geom_flags & XFS_BTGEO_OVERLAPPING)
2339 		return __xfs_btree_updkeys(cur, level, block, bp, false);
2340 
2341 	/*
2342 	 * Go up the tree from this level toward the root.
2343 	 * At each level, update the key value to the value input.
2344 	 * Stop when we reach a level where the cursor isn't pointing
2345 	 * at the first entry in the block.
2346 	 */
2347 	xfs_btree_get_keys(cur, block, &key);
2348 	for (level++, ptr = 1; ptr == 1 && level < cur->bc_nlevels; level++) {
2349 #ifdef DEBUG
2350 		int		error;
2351 #endif
2352 		block = xfs_btree_get_block(cur, level, &bp);
2353 #ifdef DEBUG
2354 		error = xfs_btree_check_block(cur, block, level, bp);
2355 		if (error)
2356 			return error;
2357 #endif
2358 		ptr = cur->bc_levels[level].ptr;
2359 		kp = xfs_btree_key_addr(cur, ptr, block);
2360 		xfs_btree_copy_keys(cur, kp, &key, 1);
2361 		xfs_btree_log_keys(cur, bp, ptr, ptr);
2362 	}
2363 
2364 	return 0;
2365 }
2366 
2367 /*
2368  * Update the record referred to by cur to the value in the
2369  * given record. This either works (return 0) or gets an
2370  * EFSCORRUPTED error.
2371  */
2372 int
xfs_btree_update(struct xfs_btree_cur * cur,union xfs_btree_rec * rec)2373 xfs_btree_update(
2374 	struct xfs_btree_cur	*cur,
2375 	union xfs_btree_rec	*rec)
2376 {
2377 	struct xfs_btree_block	*block;
2378 	struct xfs_buf		*bp;
2379 	int			error;
2380 	int			ptr;
2381 	union xfs_btree_rec	*rp;
2382 
2383 	/* Pick up the current block. */
2384 	block = xfs_btree_get_block(cur, 0, &bp);
2385 
2386 #ifdef DEBUG
2387 	error = xfs_btree_check_block(cur, block, 0, bp);
2388 	if (error)
2389 		goto error0;
2390 #endif
2391 	/* Get the address of the rec to be updated. */
2392 	ptr = cur->bc_levels[0].ptr;
2393 	rp = xfs_btree_rec_addr(cur, ptr, block);
2394 
2395 	/* Fill in the new contents and log them. */
2396 	xfs_btree_copy_recs(cur, rp, rec, 1);
2397 	xfs_btree_log_recs(cur, bp, ptr, ptr);
2398 
2399 	/* Pass new key value up to our parent. */
2400 	if (xfs_btree_needs_key_update(cur, ptr)) {
2401 		error = xfs_btree_update_keys(cur, 0);
2402 		if (error)
2403 			goto error0;
2404 	}
2405 
2406 	return 0;
2407 
2408 error0:
2409 	return error;
2410 }
2411 
2412 /*
2413  * Move 1 record left from cur/level if possible.
2414  * Update cur to reflect the new path.
2415  */
2416 STATIC int					/* error */
xfs_btree_lshift(struct xfs_btree_cur * cur,int level,int * stat)2417 xfs_btree_lshift(
2418 	struct xfs_btree_cur	*cur,
2419 	int			level,
2420 	int			*stat)		/* success/failure */
2421 {
2422 	struct xfs_buf		*lbp;		/* left buffer pointer */
2423 	struct xfs_btree_block	*left;		/* left btree block */
2424 	int			lrecs;		/* left record count */
2425 	struct xfs_buf		*rbp;		/* right buffer pointer */
2426 	struct xfs_btree_block	*right;		/* right btree block */
2427 	struct xfs_btree_cur	*tcur;		/* temporary btree cursor */
2428 	int			rrecs;		/* right record count */
2429 	union xfs_btree_ptr	lptr;		/* left btree pointer */
2430 	union xfs_btree_key	*rkp = NULL;	/* right btree key */
2431 	union xfs_btree_ptr	*rpp = NULL;	/* right address pointer */
2432 	union xfs_btree_rec	*rrp = NULL;	/* right record pointer */
2433 	int			error;		/* error return value */
2434 	int			i;
2435 
2436 	if (xfs_btree_at_iroot(cur, level))
2437 		goto out0;
2438 
2439 	/* Set up variables for this block as "right". */
2440 	right = xfs_btree_get_block(cur, level, &rbp);
2441 
2442 #ifdef DEBUG
2443 	error = xfs_btree_check_block(cur, right, level, rbp);
2444 	if (error)
2445 		goto error0;
2446 #endif
2447 
2448 	/* If we've got no left sibling then we can't shift an entry left. */
2449 	xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
2450 	if (xfs_btree_ptr_is_null(cur, &lptr))
2451 		goto out0;
2452 
2453 	/*
2454 	 * If the cursor entry is the one that would be moved, don't
2455 	 * do it... it's too complicated.
2456 	 */
2457 	if (cur->bc_levels[level].ptr <= 1)
2458 		goto out0;
2459 
2460 	/* Set up the left neighbor as "left". */
2461 	error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
2462 	if (error)
2463 		goto error0;
2464 
2465 	/* If it's full, it can't take another entry. */
2466 	lrecs = xfs_btree_get_numrecs(left);
2467 	if (lrecs == cur->bc_ops->get_maxrecs(cur, level))
2468 		goto out0;
2469 
2470 	rrecs = xfs_btree_get_numrecs(right);
2471 
2472 	/*
2473 	 * We add one entry to the left side and remove one for the right side.
2474 	 * Account for it here, the changes will be updated on disk and logged
2475 	 * later.
2476 	 */
2477 	lrecs++;
2478 	rrecs--;
2479 
2480 	XFS_BTREE_STATS_INC(cur, lshift);
2481 	XFS_BTREE_STATS_ADD(cur, moves, 1);
2482 
2483 	/*
2484 	 * If non-leaf, copy a key and a ptr to the left block.
2485 	 * Log the changes to the left block.
2486 	 */
2487 	if (level > 0) {
2488 		/* It's a non-leaf.  Move keys and pointers. */
2489 		union xfs_btree_key	*lkp;	/* left btree key */
2490 		union xfs_btree_ptr	*lpp;	/* left address pointer */
2491 
2492 		lkp = xfs_btree_key_addr(cur, lrecs, left);
2493 		rkp = xfs_btree_key_addr(cur, 1, right);
2494 
2495 		lpp = xfs_btree_ptr_addr(cur, lrecs, left);
2496 		rpp = xfs_btree_ptr_addr(cur, 1, right);
2497 
2498 		error = xfs_btree_debug_check_ptr(cur, rpp, 0, level);
2499 		if (error)
2500 			goto error0;
2501 
2502 		xfs_btree_copy_keys(cur, lkp, rkp, 1);
2503 		xfs_btree_copy_ptrs(cur, lpp, rpp, 1);
2504 
2505 		xfs_btree_log_keys(cur, lbp, lrecs, lrecs);
2506 		xfs_btree_log_ptrs(cur, lbp, lrecs, lrecs);
2507 
2508 		ASSERT(cur->bc_ops->keys_inorder(cur,
2509 			xfs_btree_key_addr(cur, lrecs - 1, left), lkp));
2510 	} else {
2511 		/* It's a leaf.  Move records.  */
2512 		union xfs_btree_rec	*lrp;	/* left record pointer */
2513 
2514 		lrp = xfs_btree_rec_addr(cur, lrecs, left);
2515 		rrp = xfs_btree_rec_addr(cur, 1, right);
2516 
2517 		xfs_btree_copy_recs(cur, lrp, rrp, 1);
2518 		xfs_btree_log_recs(cur, lbp, lrecs, lrecs);
2519 
2520 		ASSERT(cur->bc_ops->recs_inorder(cur,
2521 			xfs_btree_rec_addr(cur, lrecs - 1, left), lrp));
2522 	}
2523 
2524 	xfs_btree_set_numrecs(left, lrecs);
2525 	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
2526 
2527 	xfs_btree_set_numrecs(right, rrecs);
2528 	xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
2529 
2530 	/*
2531 	 * Slide the contents of right down one entry.
2532 	 */
2533 	XFS_BTREE_STATS_ADD(cur, moves, rrecs - 1);
2534 	if (level > 0) {
2535 		/* It's a nonleaf. operate on keys and ptrs */
2536 		for (i = 0; i < rrecs; i++) {
2537 			error = xfs_btree_debug_check_ptr(cur, rpp, i + 1, level);
2538 			if (error)
2539 				goto error0;
2540 		}
2541 
2542 		xfs_btree_shift_keys(cur,
2543 				xfs_btree_key_addr(cur, 2, right),
2544 				-1, rrecs);
2545 		xfs_btree_shift_ptrs(cur,
2546 				xfs_btree_ptr_addr(cur, 2, right),
2547 				-1, rrecs);
2548 
2549 		xfs_btree_log_keys(cur, rbp, 1, rrecs);
2550 		xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
2551 	} else {
2552 		/* It's a leaf. operate on records */
2553 		xfs_btree_shift_recs(cur,
2554 			xfs_btree_rec_addr(cur, 2, right),
2555 			-1, rrecs);
2556 		xfs_btree_log_recs(cur, rbp, 1, rrecs);
2557 	}
2558 
2559 	/*
2560 	 * Using a temporary cursor, update the parent key values of the
2561 	 * block on the left.
2562 	 */
2563 	if (cur->bc_ops->geom_flags & XFS_BTGEO_OVERLAPPING) {
2564 		error = xfs_btree_dup_cursor(cur, &tcur);
2565 		if (error)
2566 			goto error0;
2567 		i = xfs_btree_firstrec(tcur, level);
2568 		if (XFS_IS_CORRUPT(tcur->bc_mp, i != 1)) {
2569 			xfs_btree_mark_sick(cur);
2570 			error = -EFSCORRUPTED;
2571 			goto error0;
2572 		}
2573 
2574 		error = xfs_btree_decrement(tcur, level, &i);
2575 		if (error)
2576 			goto error1;
2577 
2578 		/* Update the parent high keys of the left block, if needed. */
2579 		error = xfs_btree_update_keys(tcur, level);
2580 		if (error)
2581 			goto error1;
2582 
2583 		xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
2584 	}
2585 
2586 	/* Update the parent keys of the right block. */
2587 	error = xfs_btree_update_keys(cur, level);
2588 	if (error)
2589 		goto error0;
2590 
2591 	/* Slide the cursor value left one. */
2592 	cur->bc_levels[level].ptr--;
2593 
2594 	*stat = 1;
2595 	return 0;
2596 
2597 out0:
2598 	*stat = 0;
2599 	return 0;
2600 
2601 error0:
2602 	return error;
2603 
2604 error1:
2605 	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
2606 	return error;
2607 }
2608 
2609 /*
2610  * Move 1 record right from cur/level if possible.
2611  * Update cur to reflect the new path.
2612  */
2613 STATIC int					/* error */
xfs_btree_rshift(struct xfs_btree_cur * cur,int level,int * stat)2614 xfs_btree_rshift(
2615 	struct xfs_btree_cur	*cur,
2616 	int			level,
2617 	int			*stat)		/* success/failure */
2618 {
2619 	struct xfs_buf		*lbp;		/* left buffer pointer */
2620 	struct xfs_btree_block	*left;		/* left btree block */
2621 	struct xfs_buf		*rbp;		/* right buffer pointer */
2622 	struct xfs_btree_block	*right;		/* right btree block */
2623 	struct xfs_btree_cur	*tcur;		/* temporary btree cursor */
2624 	union xfs_btree_ptr	rptr;		/* right block pointer */
2625 	union xfs_btree_key	*rkp;		/* right btree key */
2626 	int			rrecs;		/* right record count */
2627 	int			lrecs;		/* left record count */
2628 	int			error;		/* error return value */
2629 	int			i;		/* loop counter */
2630 
2631 	if (xfs_btree_at_iroot(cur, level))
2632 		goto out0;
2633 
2634 	/* Set up variables for this block as "left". */
2635 	left = xfs_btree_get_block(cur, level, &lbp);
2636 
2637 #ifdef DEBUG
2638 	error = xfs_btree_check_block(cur, left, level, lbp);
2639 	if (error)
2640 		goto error0;
2641 #endif
2642 
2643 	/* If we've got no right sibling then we can't shift an entry right. */
2644 	xfs_btree_get_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
2645 	if (xfs_btree_ptr_is_null(cur, &rptr))
2646 		goto out0;
2647 
2648 	/*
2649 	 * If the cursor entry is the one that would be moved, don't
2650 	 * do it... it's too complicated.
2651 	 */
2652 	lrecs = xfs_btree_get_numrecs(left);
2653 	if (cur->bc_levels[level].ptr >= lrecs)
2654 		goto out0;
2655 
2656 	/* Set up the right neighbor as "right". */
2657 	error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
2658 	if (error)
2659 		goto error0;
2660 
2661 	/* If it's full, it can't take another entry. */
2662 	rrecs = xfs_btree_get_numrecs(right);
2663 	if (rrecs == cur->bc_ops->get_maxrecs(cur, level))
2664 		goto out0;
2665 
2666 	XFS_BTREE_STATS_INC(cur, rshift);
2667 	XFS_BTREE_STATS_ADD(cur, moves, rrecs);
2668 
2669 	/*
2670 	 * Make a hole at the start of the right neighbor block, then
2671 	 * copy the last left block entry to the hole.
2672 	 */
2673 	if (level > 0) {
2674 		/* It's a nonleaf. make a hole in the keys and ptrs */
2675 		union xfs_btree_key	*lkp;
2676 		union xfs_btree_ptr	*lpp;
2677 		union xfs_btree_ptr	*rpp;
2678 
2679 		lkp = xfs_btree_key_addr(cur, lrecs, left);
2680 		lpp = xfs_btree_ptr_addr(cur, lrecs, left);
2681 		rkp = xfs_btree_key_addr(cur, 1, right);
2682 		rpp = xfs_btree_ptr_addr(cur, 1, right);
2683 
2684 		for (i = rrecs - 1; i >= 0; i--) {
2685 			error = xfs_btree_debug_check_ptr(cur, rpp, i, level);
2686 			if (error)
2687 				goto error0;
2688 		}
2689 
2690 		xfs_btree_shift_keys(cur, rkp, 1, rrecs);
2691 		xfs_btree_shift_ptrs(cur, rpp, 1, rrecs);
2692 
2693 		error = xfs_btree_debug_check_ptr(cur, lpp, 0, level);
2694 		if (error)
2695 			goto error0;
2696 
2697 		/* Now put the new data in, and log it. */
2698 		xfs_btree_copy_keys(cur, rkp, lkp, 1);
2699 		xfs_btree_copy_ptrs(cur, rpp, lpp, 1);
2700 
2701 		xfs_btree_log_keys(cur, rbp, 1, rrecs + 1);
2702 		xfs_btree_log_ptrs(cur, rbp, 1, rrecs + 1);
2703 
2704 		ASSERT(cur->bc_ops->keys_inorder(cur, rkp,
2705 			xfs_btree_key_addr(cur, 2, right)));
2706 	} else {
2707 		/* It's a leaf. make a hole in the records */
2708 		union xfs_btree_rec	*lrp;
2709 		union xfs_btree_rec	*rrp;
2710 
2711 		lrp = xfs_btree_rec_addr(cur, lrecs, left);
2712 		rrp = xfs_btree_rec_addr(cur, 1, right);
2713 
2714 		xfs_btree_shift_recs(cur, rrp, 1, rrecs);
2715 
2716 		/* Now put the new data in, and log it. */
2717 		xfs_btree_copy_recs(cur, rrp, lrp, 1);
2718 		xfs_btree_log_recs(cur, rbp, 1, rrecs + 1);
2719 	}
2720 
2721 	/*
2722 	 * Decrement and log left's numrecs, bump and log right's numrecs.
2723 	 */
2724 	xfs_btree_set_numrecs(left, --lrecs);
2725 	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
2726 
2727 	xfs_btree_set_numrecs(right, ++rrecs);
2728 	xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
2729 
2730 	/*
2731 	 * Using a temporary cursor, update the parent key values of the
2732 	 * block on the right.
2733 	 */
2734 	error = xfs_btree_dup_cursor(cur, &tcur);
2735 	if (error)
2736 		goto error0;
2737 	i = xfs_btree_lastrec(tcur, level);
2738 	if (XFS_IS_CORRUPT(tcur->bc_mp, i != 1)) {
2739 		xfs_btree_mark_sick(cur);
2740 		error = -EFSCORRUPTED;
2741 		goto error0;
2742 	}
2743 
2744 	error = xfs_btree_increment(tcur, level, &i);
2745 	if (error)
2746 		goto error1;
2747 
2748 	/* Update the parent high keys of the left block, if needed. */
2749 	if (cur->bc_ops->geom_flags & XFS_BTGEO_OVERLAPPING) {
2750 		error = xfs_btree_update_keys(cur, level);
2751 		if (error)
2752 			goto error1;
2753 	}
2754 
2755 	/* Update the parent keys of the right block. */
2756 	error = xfs_btree_update_keys(tcur, level);
2757 	if (error)
2758 		goto error1;
2759 
2760 	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
2761 
2762 	*stat = 1;
2763 	return 0;
2764 
2765 out0:
2766 	*stat = 0;
2767 	return 0;
2768 
2769 error0:
2770 	return error;
2771 
2772 error1:
2773 	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
2774 	return error;
2775 }
2776 
2777 static inline int
xfs_btree_alloc_block(struct xfs_btree_cur * cur,const union xfs_btree_ptr * hint_block,union xfs_btree_ptr * new_block,int * stat)2778 xfs_btree_alloc_block(
2779 	struct xfs_btree_cur		*cur,
2780 	const union xfs_btree_ptr	*hint_block,
2781 	union xfs_btree_ptr		*new_block,
2782 	int				*stat)
2783 {
2784 	int				error;
2785 
2786 	/*
2787 	 * Don't allow block allocation for a staging cursor, because staging
2788 	 * cursors do not support regular btree modifications.
2789 	 *
2790 	 * Bulk loading uses a separate callback to obtain new blocks from a
2791 	 * preallocated list, which prevents ENOSPC failures during loading.
2792 	 */
2793 	if (unlikely(cur->bc_flags & XFS_BTREE_STAGING)) {
2794 		ASSERT(0);
2795 		return -EFSCORRUPTED;
2796 	}
2797 
2798 	error = cur->bc_ops->alloc_block(cur, hint_block, new_block, stat);
2799 	trace_xfs_btree_alloc_block(cur, new_block, *stat, error);
2800 	return error;
2801 }
2802 
2803 /*
2804  * Split cur/level block in half.
2805  * Return new block number and the key to its first
2806  * record (to be inserted into parent).
2807  */
2808 STATIC int					/* error */
__xfs_btree_split(struct xfs_btree_cur * cur,int level,union xfs_btree_ptr * ptrp,union xfs_btree_key * key,struct xfs_btree_cur ** curp,int * stat)2809 __xfs_btree_split(
2810 	struct xfs_btree_cur	*cur,
2811 	int			level,
2812 	union xfs_btree_ptr	*ptrp,
2813 	union xfs_btree_key	*key,
2814 	struct xfs_btree_cur	**curp,
2815 	int			*stat)		/* success/failure */
2816 {
2817 	union xfs_btree_ptr	lptr;		/* left sibling block ptr */
2818 	struct xfs_buf		*lbp;		/* left buffer pointer */
2819 	struct xfs_btree_block	*left;		/* left btree block */
2820 	union xfs_btree_ptr	rptr;		/* right sibling block ptr */
2821 	struct xfs_buf		*rbp;		/* right buffer pointer */
2822 	struct xfs_btree_block	*right;		/* right btree block */
2823 	union xfs_btree_ptr	rrptr;		/* right-right sibling ptr */
2824 	struct xfs_buf		*rrbp;		/* right-right buffer pointer */
2825 	struct xfs_btree_block	*rrblock;	/* right-right btree block */
2826 	int			lrecs;
2827 	int			rrecs;
2828 	int			src_index;
2829 	int			error;		/* error return value */
2830 	int			i;
2831 
2832 	XFS_BTREE_STATS_INC(cur, split);
2833 
2834 	/* Set up left block (current one). */
2835 	left = xfs_btree_get_block(cur, level, &lbp);
2836 
2837 #ifdef DEBUG
2838 	error = xfs_btree_check_block(cur, left, level, lbp);
2839 	if (error)
2840 		goto error0;
2841 #endif
2842 
2843 	xfs_btree_buf_to_ptr(cur, lbp, &lptr);
2844 
2845 	/* Allocate the new block. If we can't do it, we're toast. Give up. */
2846 	error = xfs_btree_alloc_block(cur, &lptr, &rptr, stat);
2847 	if (error)
2848 		goto error0;
2849 	if (*stat == 0)
2850 		goto out0;
2851 	XFS_BTREE_STATS_INC(cur, alloc);
2852 
2853 	/* Set up the new block as "right". */
2854 	error = xfs_btree_get_buf_block(cur, &rptr, &right, &rbp);
2855 	if (error)
2856 		goto error0;
2857 
2858 	/* Fill in the btree header for the new right block. */
2859 	xfs_btree_init_block_cur(cur, rbp, xfs_btree_get_level(left), 0);
2860 
2861 	/*
2862 	 * Split the entries between the old and the new block evenly.
2863 	 * Make sure that if there's an odd number of entries now, that
2864 	 * each new block will have the same number of entries.
2865 	 */
2866 	lrecs = xfs_btree_get_numrecs(left);
2867 	rrecs = lrecs / 2;
2868 	if ((lrecs & 1) && cur->bc_levels[level].ptr <= rrecs + 1)
2869 		rrecs++;
2870 	src_index = (lrecs - rrecs + 1);
2871 
2872 	XFS_BTREE_STATS_ADD(cur, moves, rrecs);
2873 
2874 	/* Adjust numrecs for the later get_*_keys() calls. */
2875 	lrecs -= rrecs;
2876 	xfs_btree_set_numrecs(left, lrecs);
2877 	xfs_btree_set_numrecs(right, xfs_btree_get_numrecs(right) + rrecs);
2878 
2879 	/*
2880 	 * Copy btree block entries from the left block over to the
2881 	 * new block, the right. Update the right block and log the
2882 	 * changes.
2883 	 */
2884 	if (level > 0) {
2885 		/* It's a non-leaf.  Move keys and pointers. */
2886 		union xfs_btree_key	*lkp;	/* left btree key */
2887 		union xfs_btree_ptr	*lpp;	/* left address pointer */
2888 		union xfs_btree_key	*rkp;	/* right btree key */
2889 		union xfs_btree_ptr	*rpp;	/* right address pointer */
2890 
2891 		lkp = xfs_btree_key_addr(cur, src_index, left);
2892 		lpp = xfs_btree_ptr_addr(cur, src_index, left);
2893 		rkp = xfs_btree_key_addr(cur, 1, right);
2894 		rpp = xfs_btree_ptr_addr(cur, 1, right);
2895 
2896 		for (i = src_index; i < rrecs; i++) {
2897 			error = xfs_btree_debug_check_ptr(cur, lpp, i, level);
2898 			if (error)
2899 				goto error0;
2900 		}
2901 
2902 		/* Copy the keys & pointers to the new block. */
2903 		xfs_btree_copy_keys(cur, rkp, lkp, rrecs);
2904 		xfs_btree_copy_ptrs(cur, rpp, lpp, rrecs);
2905 
2906 		xfs_btree_log_keys(cur, rbp, 1, rrecs);
2907 		xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
2908 
2909 		/* Stash the keys of the new block for later insertion. */
2910 		xfs_btree_get_node_keys(cur, right, key);
2911 	} else {
2912 		/* It's a leaf.  Move records.  */
2913 		union xfs_btree_rec	*lrp;	/* left record pointer */
2914 		union xfs_btree_rec	*rrp;	/* right record pointer */
2915 
2916 		lrp = xfs_btree_rec_addr(cur, src_index, left);
2917 		rrp = xfs_btree_rec_addr(cur, 1, right);
2918 
2919 		/* Copy records to the new block. */
2920 		xfs_btree_copy_recs(cur, rrp, lrp, rrecs);
2921 		xfs_btree_log_recs(cur, rbp, 1, rrecs);
2922 
2923 		/* Stash the keys of the new block for later insertion. */
2924 		xfs_btree_get_leaf_keys(cur, right, key);
2925 	}
2926 
2927 	/*
2928 	 * Find the left block number by looking in the buffer.
2929 	 * Adjust sibling pointers.
2930 	 */
2931 	xfs_btree_get_sibling(cur, left, &rrptr, XFS_BB_RIGHTSIB);
2932 	xfs_btree_set_sibling(cur, right, &rrptr, XFS_BB_RIGHTSIB);
2933 	xfs_btree_set_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
2934 	xfs_btree_set_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
2935 
2936 	xfs_btree_log_block(cur, rbp, XFS_BB_ALL_BITS);
2937 	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
2938 
2939 	/*
2940 	 * If there's a block to the new block's right, make that block
2941 	 * point back to right instead of to left.
2942 	 */
2943 	if (!xfs_btree_ptr_is_null(cur, &rrptr)) {
2944 		error = xfs_btree_read_buf_block(cur, &rrptr,
2945 							0, &rrblock, &rrbp);
2946 		if (error)
2947 			goto error0;
2948 		xfs_btree_set_sibling(cur, rrblock, &rptr, XFS_BB_LEFTSIB);
2949 		xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
2950 	}
2951 
2952 	/* Update the parent high keys of the left block, if needed. */
2953 	if (cur->bc_ops->geom_flags & XFS_BTGEO_OVERLAPPING) {
2954 		error = xfs_btree_update_keys(cur, level);
2955 		if (error)
2956 			goto error0;
2957 	}
2958 
2959 	/*
2960 	 * If the cursor is really in the right block, move it there.
2961 	 * If it's just pointing past the last entry in left, then we'll
2962 	 * insert there, so don't change anything in that case.
2963 	 */
2964 	if (cur->bc_levels[level].ptr > lrecs + 1) {
2965 		xfs_btree_setbuf(cur, level, rbp);
2966 		cur->bc_levels[level].ptr -= lrecs;
2967 	}
2968 	/*
2969 	 * If there are more levels, we'll need another cursor which refers
2970 	 * the right block, no matter where this cursor was.
2971 	 */
2972 	if (level + 1 < cur->bc_nlevels) {
2973 		error = xfs_btree_dup_cursor(cur, curp);
2974 		if (error)
2975 			goto error0;
2976 		(*curp)->bc_levels[level + 1].ptr++;
2977 	}
2978 	*ptrp = rptr;
2979 	*stat = 1;
2980 	return 0;
2981 out0:
2982 	*stat = 0;
2983 	return 0;
2984 
2985 error0:
2986 	return error;
2987 }
2988 
2989 #ifdef __KERNEL__
2990 struct xfs_btree_split_args {
2991 	struct xfs_btree_cur	*cur;
2992 	int			level;
2993 	union xfs_btree_ptr	*ptrp;
2994 	union xfs_btree_key	*key;
2995 	struct xfs_btree_cur	**curp;
2996 	int			*stat;		/* success/failure */
2997 	int			result;
2998 	bool			kswapd;	/* allocation in kswapd context */
2999 	struct completion	*done;
3000 	struct work_struct	work;
3001 };
3002 
3003 /*
3004  * Stack switching interfaces for allocation
3005  */
3006 static void
xfs_btree_split_worker(struct work_struct * work)3007 xfs_btree_split_worker(
3008 	struct work_struct	*work)
3009 {
3010 	struct xfs_btree_split_args	*args = container_of(work,
3011 						struct xfs_btree_split_args, work);
3012 	unsigned long		pflags;
3013 	unsigned long		new_pflags = 0;
3014 
3015 	/*
3016 	 * we are in a transaction context here, but may also be doing work
3017 	 * in kswapd context, and hence we may need to inherit that state
3018 	 * temporarily to ensure that we don't block waiting for memory reclaim
3019 	 * in any way.
3020 	 */
3021 	if (args->kswapd)
3022 		new_pflags |= PF_MEMALLOC | PF_KSWAPD;
3023 
3024 	current_set_flags_nested(&pflags, new_pflags);
3025 	xfs_trans_set_context(args->cur->bc_tp);
3026 
3027 	args->result = __xfs_btree_split(args->cur, args->level, args->ptrp,
3028 					 args->key, args->curp, args->stat);
3029 
3030 	xfs_trans_clear_context(args->cur->bc_tp);
3031 	current_restore_flags_nested(&pflags, new_pflags);
3032 
3033 	/*
3034 	 * Do not access args after complete() has run here. We don't own args
3035 	 * and the owner may run and free args before we return here.
3036 	 */
3037 	complete(args->done);
3038 
3039 }
3040 
3041 /*
3042  * BMBT split requests often come in with little stack to work on so we push
3043  * them off to a worker thread so there is lots of stack to use. For the other
3044  * btree types, just call directly to avoid the context switch overhead here.
3045  *
3046  * Care must be taken here - the work queue rescuer thread introduces potential
3047  * AGF <> worker queue deadlocks if the BMBT block allocation has to lock new
3048  * AGFs to allocate blocks. A task being run by the rescuer could attempt to
3049  * lock an AGF that is already locked by a task queued to run by the rescuer,
3050  * resulting in an ABBA deadlock as the rescuer cannot run the lock holder to
3051  * release it until the current thread it is running gains the lock.
3052  *
3053  * To avoid this issue, we only ever queue BMBT splits that don't have an AGF
3054  * already locked to allocate from. The only place that doesn't hold an AGF
3055  * locked is unwritten extent conversion at IO completion, but that has already
3056  * been offloaded to a worker thread and hence has no stack consumption issues
3057  * we have to worry about.
3058  */
3059 STATIC int					/* error */
xfs_btree_split(struct xfs_btree_cur * cur,int level,union xfs_btree_ptr * ptrp,union xfs_btree_key * key,struct xfs_btree_cur ** curp,int * stat)3060 xfs_btree_split(
3061 	struct xfs_btree_cur	*cur,
3062 	int			level,
3063 	union xfs_btree_ptr	*ptrp,
3064 	union xfs_btree_key	*key,
3065 	struct xfs_btree_cur	**curp,
3066 	int			*stat)		/* success/failure */
3067 {
3068 	struct xfs_btree_split_args	args;
3069 	DECLARE_COMPLETION_ONSTACK(done);
3070 
3071 	if (!xfs_btree_is_bmap(cur->bc_ops) ||
3072 	    cur->bc_tp->t_highest_agno == NULLAGNUMBER)
3073 		return __xfs_btree_split(cur, level, ptrp, key, curp, stat);
3074 
3075 	args.cur = cur;
3076 	args.level = level;
3077 	args.ptrp = ptrp;
3078 	args.key = key;
3079 	args.curp = curp;
3080 	args.stat = stat;
3081 	args.done = &done;
3082 	args.kswapd = current_is_kswapd();
3083 	INIT_WORK_ONSTACK(&args.work, xfs_btree_split_worker);
3084 	queue_work(xfs_alloc_wq, &args.work);
3085 	wait_for_completion(&done);
3086 	destroy_work_on_stack(&args.work);
3087 	return args.result;
3088 }
3089 #else
3090 #define xfs_btree_split	__xfs_btree_split
3091 #endif /* __KERNEL__ */
3092 
3093 /*
3094  * Copy the old inode root contents into a real block and make the
3095  * broot point to it.
3096  */
3097 int						/* error */
xfs_btree_new_iroot(struct xfs_btree_cur * cur,int * logflags,int * stat)3098 xfs_btree_new_iroot(
3099 	struct xfs_btree_cur	*cur,		/* btree cursor */
3100 	int			*logflags,	/* logging flags for inode */
3101 	int			*stat)		/* return status - 0 fail */
3102 {
3103 	struct xfs_buf		*cbp;		/* buffer for cblock */
3104 	struct xfs_btree_block	*block;		/* btree block */
3105 	struct xfs_btree_block	*cblock;	/* child btree block */
3106 	union xfs_btree_key	*ckp;		/* child key pointer */
3107 	union xfs_btree_ptr	*cpp;		/* child ptr pointer */
3108 	union xfs_btree_key	*kp;		/* pointer to btree key */
3109 	union xfs_btree_ptr	*pp;		/* pointer to block addr */
3110 	union xfs_btree_ptr	nptr;		/* new block addr */
3111 	int			level;		/* btree level */
3112 	int			error;		/* error return code */
3113 	int			i;		/* loop counter */
3114 
3115 	XFS_BTREE_STATS_INC(cur, newroot);
3116 
3117 	ASSERT(cur->bc_ops->type == XFS_BTREE_TYPE_INODE);
3118 
3119 	level = cur->bc_nlevels - 1;
3120 
3121 	block = xfs_btree_get_iroot(cur);
3122 	pp = xfs_btree_ptr_addr(cur, 1, block);
3123 
3124 	/* Allocate the new block. If we can't do it, we're toast. Give up. */
3125 	error = xfs_btree_alloc_block(cur, pp, &nptr, stat);
3126 	if (error)
3127 		goto error0;
3128 	if (*stat == 0)
3129 		return 0;
3130 
3131 	XFS_BTREE_STATS_INC(cur, alloc);
3132 
3133 	/* Copy the root into a real block. */
3134 	error = xfs_btree_get_buf_block(cur, &nptr, &cblock, &cbp);
3135 	if (error)
3136 		goto error0;
3137 
3138 	/*
3139 	 * we can't just memcpy() the root in for CRC enabled btree blocks.
3140 	 * In that case have to also ensure the blkno remains correct
3141 	 */
3142 	memcpy(cblock, block, xfs_btree_block_len(cur));
3143 	if (xfs_has_crc(cur->bc_mp)) {
3144 		__be64 bno = cpu_to_be64(xfs_buf_daddr(cbp));
3145 		if (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN)
3146 			cblock->bb_u.l.bb_blkno = bno;
3147 		else
3148 			cblock->bb_u.s.bb_blkno = bno;
3149 	}
3150 
3151 	be16_add_cpu(&block->bb_level, 1);
3152 	xfs_btree_set_numrecs(block, 1);
3153 	cur->bc_nlevels++;
3154 	ASSERT(cur->bc_nlevels <= cur->bc_maxlevels);
3155 	cur->bc_levels[level + 1].ptr = 1;
3156 
3157 	kp = xfs_btree_key_addr(cur, 1, block);
3158 	ckp = xfs_btree_key_addr(cur, 1, cblock);
3159 	xfs_btree_copy_keys(cur, ckp, kp, xfs_btree_get_numrecs(cblock));
3160 
3161 	cpp = xfs_btree_ptr_addr(cur, 1, cblock);
3162 	for (i = 0; i < be16_to_cpu(cblock->bb_numrecs); i++) {
3163 		error = xfs_btree_debug_check_ptr(cur, pp, i, level);
3164 		if (error)
3165 			goto error0;
3166 	}
3167 
3168 	xfs_btree_copy_ptrs(cur, cpp, pp, xfs_btree_get_numrecs(cblock));
3169 
3170 	error = xfs_btree_debug_check_ptr(cur, &nptr, 0, level);
3171 	if (error)
3172 		goto error0;
3173 
3174 	xfs_btree_copy_ptrs(cur, pp, &nptr, 1);
3175 
3176 	xfs_iroot_realloc(cur->bc_ino.ip,
3177 			  1 - xfs_btree_get_numrecs(cblock),
3178 			  cur->bc_ino.whichfork);
3179 
3180 	xfs_btree_setbuf(cur, level, cbp);
3181 
3182 	/*
3183 	 * Do all this logging at the end so that
3184 	 * the root is at the right level.
3185 	 */
3186 	xfs_btree_log_block(cur, cbp, XFS_BB_ALL_BITS);
3187 	xfs_btree_log_keys(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
3188 	xfs_btree_log_ptrs(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
3189 
3190 	*logflags |=
3191 		XFS_ILOG_CORE | xfs_ilog_fbroot(cur->bc_ino.whichfork);
3192 	*stat = 1;
3193 	return 0;
3194 error0:
3195 	return error;
3196 }
3197 
3198 static void
xfs_btree_set_root(struct xfs_btree_cur * cur,const union xfs_btree_ptr * ptr,int inc)3199 xfs_btree_set_root(
3200 	struct xfs_btree_cur		*cur,
3201 	const union xfs_btree_ptr	*ptr,
3202 	int				inc)
3203 {
3204 	if (cur->bc_flags & XFS_BTREE_STAGING) {
3205 		/* Update the btree root information for a per-AG fake root. */
3206 		cur->bc_ag.afake->af_root = be32_to_cpu(ptr->s);
3207 		cur->bc_ag.afake->af_levels += inc;
3208 	} else {
3209 		cur->bc_ops->set_root(cur, ptr, inc);
3210 	}
3211 }
3212 
3213 /*
3214  * Allocate a new root block, fill it in.
3215  */
3216 STATIC int				/* error */
xfs_btree_new_root(struct xfs_btree_cur * cur,int * stat)3217 xfs_btree_new_root(
3218 	struct xfs_btree_cur	*cur,	/* btree cursor */
3219 	int			*stat)	/* success/failure */
3220 {
3221 	struct xfs_btree_block	*block;	/* one half of the old root block */
3222 	struct xfs_buf		*bp;	/* buffer containing block */
3223 	int			error;	/* error return value */
3224 	struct xfs_buf		*lbp;	/* left buffer pointer */
3225 	struct xfs_btree_block	*left;	/* left btree block */
3226 	struct xfs_buf		*nbp;	/* new (root) buffer */
3227 	struct xfs_btree_block	*new;	/* new (root) btree block */
3228 	int			nptr;	/* new value for key index, 1 or 2 */
3229 	struct xfs_buf		*rbp;	/* right buffer pointer */
3230 	struct xfs_btree_block	*right;	/* right btree block */
3231 	union xfs_btree_ptr	rptr;
3232 	union xfs_btree_ptr	lptr;
3233 
3234 	XFS_BTREE_STATS_INC(cur, newroot);
3235 
3236 	/* initialise our start point from the cursor */
3237 	xfs_btree_init_ptr_from_cur(cur, &rptr);
3238 
3239 	/* Allocate the new block. If we can't do it, we're toast. Give up. */
3240 	error = xfs_btree_alloc_block(cur, &rptr, &lptr, stat);
3241 	if (error)
3242 		goto error0;
3243 	if (*stat == 0)
3244 		goto out0;
3245 	XFS_BTREE_STATS_INC(cur, alloc);
3246 
3247 	/* Set up the new block. */
3248 	error = xfs_btree_get_buf_block(cur, &lptr, &new, &nbp);
3249 	if (error)
3250 		goto error0;
3251 
3252 	/* Set the root in the holding structure  increasing the level by 1. */
3253 	xfs_btree_set_root(cur, &lptr, 1);
3254 
3255 	/*
3256 	 * At the previous root level there are now two blocks: the old root,
3257 	 * and the new block generated when it was split.  We don't know which
3258 	 * one the cursor is pointing at, so we set up variables "left" and
3259 	 * "right" for each case.
3260 	 */
3261 	block = xfs_btree_get_block(cur, cur->bc_nlevels - 1, &bp);
3262 
3263 #ifdef DEBUG
3264 	error = xfs_btree_check_block(cur, block, cur->bc_nlevels - 1, bp);
3265 	if (error)
3266 		goto error0;
3267 #endif
3268 
3269 	xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
3270 	if (!xfs_btree_ptr_is_null(cur, &rptr)) {
3271 		/* Our block is left, pick up the right block. */
3272 		lbp = bp;
3273 		xfs_btree_buf_to_ptr(cur, lbp, &lptr);
3274 		left = block;
3275 		error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
3276 		if (error)
3277 			goto error0;
3278 		bp = rbp;
3279 		nptr = 1;
3280 	} else {
3281 		/* Our block is right, pick up the left block. */
3282 		rbp = bp;
3283 		xfs_btree_buf_to_ptr(cur, rbp, &rptr);
3284 		right = block;
3285 		xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
3286 		error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
3287 		if (error)
3288 			goto error0;
3289 		bp = lbp;
3290 		nptr = 2;
3291 	}
3292 
3293 	/* Fill in the new block's btree header and log it. */
3294 	xfs_btree_init_block_cur(cur, nbp, cur->bc_nlevels, 2);
3295 	xfs_btree_log_block(cur, nbp, XFS_BB_ALL_BITS);
3296 	ASSERT(!xfs_btree_ptr_is_null(cur, &lptr) &&
3297 			!xfs_btree_ptr_is_null(cur, &rptr));
3298 
3299 	/* Fill in the key data in the new root. */
3300 	if (xfs_btree_get_level(left) > 0) {
3301 		/*
3302 		 * Get the keys for the left block's keys and put them directly
3303 		 * in the parent block.  Do the same for the right block.
3304 		 */
3305 		xfs_btree_get_node_keys(cur, left,
3306 				xfs_btree_key_addr(cur, 1, new));
3307 		xfs_btree_get_node_keys(cur, right,
3308 				xfs_btree_key_addr(cur, 2, new));
3309 	} else {
3310 		/*
3311 		 * Get the keys for the left block's records and put them
3312 		 * directly in the parent block.  Do the same for the right
3313 		 * block.
3314 		 */
3315 		xfs_btree_get_leaf_keys(cur, left,
3316 			xfs_btree_key_addr(cur, 1, new));
3317 		xfs_btree_get_leaf_keys(cur, right,
3318 			xfs_btree_key_addr(cur, 2, new));
3319 	}
3320 	xfs_btree_log_keys(cur, nbp, 1, 2);
3321 
3322 	/* Fill in the pointer data in the new root. */
3323 	xfs_btree_copy_ptrs(cur,
3324 		xfs_btree_ptr_addr(cur, 1, new), &lptr, 1);
3325 	xfs_btree_copy_ptrs(cur,
3326 		xfs_btree_ptr_addr(cur, 2, new), &rptr, 1);
3327 	xfs_btree_log_ptrs(cur, nbp, 1, 2);
3328 
3329 	/* Fix up the cursor. */
3330 	xfs_btree_setbuf(cur, cur->bc_nlevels, nbp);
3331 	cur->bc_levels[cur->bc_nlevels].ptr = nptr;
3332 	cur->bc_nlevels++;
3333 	ASSERT(cur->bc_nlevels <= cur->bc_maxlevels);
3334 	*stat = 1;
3335 	return 0;
3336 error0:
3337 	return error;
3338 out0:
3339 	*stat = 0;
3340 	return 0;
3341 }
3342 
3343 STATIC int
xfs_btree_make_block_unfull(struct xfs_btree_cur * cur,int level,int numrecs,int * oindex,int * index,union xfs_btree_ptr * nptr,struct xfs_btree_cur ** ncur,union xfs_btree_key * key,int * stat)3344 xfs_btree_make_block_unfull(
3345 	struct xfs_btree_cur	*cur,	/* btree cursor */
3346 	int			level,	/* btree level */
3347 	int			numrecs,/* # of recs in block */
3348 	int			*oindex,/* old tree index */
3349 	int			*index,	/* new tree index */
3350 	union xfs_btree_ptr	*nptr,	/* new btree ptr */
3351 	struct xfs_btree_cur	**ncur,	/* new btree cursor */
3352 	union xfs_btree_key	*key,	/* key of new block */
3353 	int			*stat)
3354 {
3355 	int			error = 0;
3356 
3357 	if (xfs_btree_at_iroot(cur, level)) {
3358 		struct xfs_inode *ip = cur->bc_ino.ip;
3359 
3360 		if (numrecs < cur->bc_ops->get_dmaxrecs(cur, level)) {
3361 			/* A root block that can be made bigger. */
3362 			xfs_iroot_realloc(ip, 1, cur->bc_ino.whichfork);
3363 			*stat = 1;
3364 		} else {
3365 			/* A root block that needs replacing */
3366 			int	logflags = 0;
3367 
3368 			error = xfs_btree_new_iroot(cur, &logflags, stat);
3369 			if (error || *stat == 0)
3370 				return error;
3371 
3372 			xfs_trans_log_inode(cur->bc_tp, ip, logflags);
3373 		}
3374 
3375 		return 0;
3376 	}
3377 
3378 	/* First, try shifting an entry to the right neighbor. */
3379 	error = xfs_btree_rshift(cur, level, stat);
3380 	if (error || *stat)
3381 		return error;
3382 
3383 	/* Next, try shifting an entry to the left neighbor. */
3384 	error = xfs_btree_lshift(cur, level, stat);
3385 	if (error)
3386 		return error;
3387 
3388 	if (*stat) {
3389 		*oindex = *index = cur->bc_levels[level].ptr;
3390 		return 0;
3391 	}
3392 
3393 	/*
3394 	 * Next, try splitting the current block in half.
3395 	 *
3396 	 * If this works we have to re-set our variables because we
3397 	 * could be in a different block now.
3398 	 */
3399 	error = xfs_btree_split(cur, level, nptr, key, ncur, stat);
3400 	if (error || *stat == 0)
3401 		return error;
3402 
3403 
3404 	*index = cur->bc_levels[level].ptr;
3405 	return 0;
3406 }
3407 
3408 /*
3409  * Insert one record/level.  Return information to the caller
3410  * allowing the next level up to proceed if necessary.
3411  */
3412 STATIC int
xfs_btree_insrec(struct xfs_btree_cur * cur,int level,union xfs_btree_ptr * ptrp,union xfs_btree_rec * rec,union xfs_btree_key * key,struct xfs_btree_cur ** curp,int * stat)3413 xfs_btree_insrec(
3414 	struct xfs_btree_cur	*cur,	/* btree cursor */
3415 	int			level,	/* level to insert record at */
3416 	union xfs_btree_ptr	*ptrp,	/* i/o: block number inserted */
3417 	union xfs_btree_rec	*rec,	/* record to insert */
3418 	union xfs_btree_key	*key,	/* i/o: block key for ptrp */
3419 	struct xfs_btree_cur	**curp,	/* output: new cursor replacing cur */
3420 	int			*stat)	/* success/failure */
3421 {
3422 	struct xfs_btree_block	*block;	/* btree block */
3423 	struct xfs_buf		*bp;	/* buffer for block */
3424 	union xfs_btree_ptr	nptr;	/* new block ptr */
3425 	struct xfs_btree_cur	*ncur = NULL;	/* new btree cursor */
3426 	union xfs_btree_key	nkey;	/* new block key */
3427 	union xfs_btree_key	*lkey;
3428 	int			optr;	/* old key/record index */
3429 	int			ptr;	/* key/record index */
3430 	int			numrecs;/* number of records */
3431 	int			error;	/* error return value */
3432 	int			i;
3433 	xfs_daddr_t		old_bn;
3434 
3435 	ncur = NULL;
3436 	lkey = &nkey;
3437 
3438 	/*
3439 	 * If we have an external root pointer, and we've made it to the
3440 	 * root level, allocate a new root block and we're done.
3441 	 */
3442 	if (cur->bc_ops->type != XFS_BTREE_TYPE_INODE &&
3443 	    level >= cur->bc_nlevels) {
3444 		error = xfs_btree_new_root(cur, stat);
3445 		xfs_btree_set_ptr_null(cur, ptrp);
3446 
3447 		return error;
3448 	}
3449 
3450 	/* If we're off the left edge, return failure. */
3451 	ptr = cur->bc_levels[level].ptr;
3452 	if (ptr == 0) {
3453 		*stat = 0;
3454 		return 0;
3455 	}
3456 
3457 	optr = ptr;
3458 
3459 	XFS_BTREE_STATS_INC(cur, insrec);
3460 
3461 	/* Get pointers to the btree buffer and block. */
3462 	block = xfs_btree_get_block(cur, level, &bp);
3463 	old_bn = bp ? xfs_buf_daddr(bp) : XFS_BUF_DADDR_NULL;
3464 	numrecs = xfs_btree_get_numrecs(block);
3465 
3466 #ifdef DEBUG
3467 	error = xfs_btree_check_block(cur, block, level, bp);
3468 	if (error)
3469 		goto error0;
3470 
3471 	/* Check that the new entry is being inserted in the right place. */
3472 	if (ptr <= numrecs) {
3473 		if (level == 0) {
3474 			ASSERT(cur->bc_ops->recs_inorder(cur, rec,
3475 				xfs_btree_rec_addr(cur, ptr, block)));
3476 		} else {
3477 			ASSERT(cur->bc_ops->keys_inorder(cur, key,
3478 				xfs_btree_key_addr(cur, ptr, block)));
3479 		}
3480 	}
3481 #endif
3482 
3483 	/*
3484 	 * If the block is full, we can't insert the new entry until we
3485 	 * make the block un-full.
3486 	 */
3487 	xfs_btree_set_ptr_null(cur, &nptr);
3488 	if (numrecs == cur->bc_ops->get_maxrecs(cur, level)) {
3489 		error = xfs_btree_make_block_unfull(cur, level, numrecs,
3490 					&optr, &ptr, &nptr, &ncur, lkey, stat);
3491 		if (error || *stat == 0)
3492 			goto error0;
3493 	}
3494 
3495 	/*
3496 	 * The current block may have changed if the block was
3497 	 * previously full and we have just made space in it.
3498 	 */
3499 	block = xfs_btree_get_block(cur, level, &bp);
3500 	numrecs = xfs_btree_get_numrecs(block);
3501 
3502 #ifdef DEBUG
3503 	error = xfs_btree_check_block(cur, block, level, bp);
3504 	if (error)
3505 		goto error0;
3506 #endif
3507 
3508 	/*
3509 	 * At this point we know there's room for our new entry in the block
3510 	 * we're pointing at.
3511 	 */
3512 	XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr + 1);
3513 
3514 	if (level > 0) {
3515 		/* It's a nonleaf. make a hole in the keys and ptrs */
3516 		union xfs_btree_key	*kp;
3517 		union xfs_btree_ptr	*pp;
3518 
3519 		kp = xfs_btree_key_addr(cur, ptr, block);
3520 		pp = xfs_btree_ptr_addr(cur, ptr, block);
3521 
3522 		for (i = numrecs - ptr; i >= 0; i--) {
3523 			error = xfs_btree_debug_check_ptr(cur, pp, i, level);
3524 			if (error)
3525 				goto error0;
3526 		}
3527 
3528 		xfs_btree_shift_keys(cur, kp, 1, numrecs - ptr + 1);
3529 		xfs_btree_shift_ptrs(cur, pp, 1, numrecs - ptr + 1);
3530 
3531 		error = xfs_btree_debug_check_ptr(cur, ptrp, 0, level);
3532 		if (error)
3533 			goto error0;
3534 
3535 		/* Now put the new data in, bump numrecs and log it. */
3536 		xfs_btree_copy_keys(cur, kp, key, 1);
3537 		xfs_btree_copy_ptrs(cur, pp, ptrp, 1);
3538 		numrecs++;
3539 		xfs_btree_set_numrecs(block, numrecs);
3540 		xfs_btree_log_ptrs(cur, bp, ptr, numrecs);
3541 		xfs_btree_log_keys(cur, bp, ptr, numrecs);
3542 #ifdef DEBUG
3543 		if (ptr < numrecs) {
3544 			ASSERT(cur->bc_ops->keys_inorder(cur, kp,
3545 				xfs_btree_key_addr(cur, ptr + 1, block)));
3546 		}
3547 #endif
3548 	} else {
3549 		/* It's a leaf. make a hole in the records */
3550 		union xfs_btree_rec             *rp;
3551 
3552 		rp = xfs_btree_rec_addr(cur, ptr, block);
3553 
3554 		xfs_btree_shift_recs(cur, rp, 1, numrecs - ptr + 1);
3555 
3556 		/* Now put the new data in, bump numrecs and log it. */
3557 		xfs_btree_copy_recs(cur, rp, rec, 1);
3558 		xfs_btree_set_numrecs(block, ++numrecs);
3559 		xfs_btree_log_recs(cur, bp, ptr, numrecs);
3560 #ifdef DEBUG
3561 		if (ptr < numrecs) {
3562 			ASSERT(cur->bc_ops->recs_inorder(cur, rp,
3563 				xfs_btree_rec_addr(cur, ptr + 1, block)));
3564 		}
3565 #endif
3566 	}
3567 
3568 	/* Log the new number of records in the btree header. */
3569 	xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
3570 
3571 	/*
3572 	 * If we just inserted into a new tree block, we have to
3573 	 * recalculate nkey here because nkey is out of date.
3574 	 *
3575 	 * Otherwise we're just updating an existing block (having shoved
3576 	 * some records into the new tree block), so use the regular key
3577 	 * update mechanism.
3578 	 */
3579 	if (bp && xfs_buf_daddr(bp) != old_bn) {
3580 		xfs_btree_get_keys(cur, block, lkey);
3581 	} else if (xfs_btree_needs_key_update(cur, optr)) {
3582 		error = xfs_btree_update_keys(cur, level);
3583 		if (error)
3584 			goto error0;
3585 	}
3586 
3587 	/*
3588 	 * Return the new block number, if any.
3589 	 * If there is one, give back a record value and a cursor too.
3590 	 */
3591 	*ptrp = nptr;
3592 	if (!xfs_btree_ptr_is_null(cur, &nptr)) {
3593 		xfs_btree_copy_keys(cur, key, lkey, 1);
3594 		*curp = ncur;
3595 	}
3596 
3597 	*stat = 1;
3598 	return 0;
3599 
3600 error0:
3601 	if (ncur)
3602 		xfs_btree_del_cursor(ncur, error);
3603 	return error;
3604 }
3605 
3606 /*
3607  * Insert the record at the point referenced by cur.
3608  *
3609  * A multi-level split of the tree on insert will invalidate the original
3610  * cursor.  All callers of this function should assume that the cursor is
3611  * no longer valid and revalidate it.
3612  */
3613 int
xfs_btree_insert(struct xfs_btree_cur * cur,int * stat)3614 xfs_btree_insert(
3615 	struct xfs_btree_cur	*cur,
3616 	int			*stat)
3617 {
3618 	int			error;	/* error return value */
3619 	int			i;	/* result value, 0 for failure */
3620 	int			level;	/* current level number in btree */
3621 	union xfs_btree_ptr	nptr;	/* new block number (split result) */
3622 	struct xfs_btree_cur	*ncur;	/* new cursor (split result) */
3623 	struct xfs_btree_cur	*pcur;	/* previous level's cursor */
3624 	union xfs_btree_key	bkey;	/* key of block to insert */
3625 	union xfs_btree_key	*key;
3626 	union xfs_btree_rec	rec;	/* record to insert */
3627 
3628 	level = 0;
3629 	ncur = NULL;
3630 	pcur = cur;
3631 	key = &bkey;
3632 
3633 	xfs_btree_set_ptr_null(cur, &nptr);
3634 
3635 	/* Make a key out of the record data to be inserted, and save it. */
3636 	cur->bc_ops->init_rec_from_cur(cur, &rec);
3637 	cur->bc_ops->init_key_from_rec(key, &rec);
3638 
3639 	/*
3640 	 * Loop going up the tree, starting at the leaf level.
3641 	 * Stop when we don't get a split block, that must mean that
3642 	 * the insert is finished with this level.
3643 	 */
3644 	do {
3645 		/*
3646 		 * Insert nrec/nptr into this level of the tree.
3647 		 * Note if we fail, nptr will be null.
3648 		 */
3649 		error = xfs_btree_insrec(pcur, level, &nptr, &rec, key,
3650 				&ncur, &i);
3651 		if (error) {
3652 			if (pcur != cur)
3653 				xfs_btree_del_cursor(pcur, XFS_BTREE_ERROR);
3654 			goto error0;
3655 		}
3656 
3657 		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
3658 			xfs_btree_mark_sick(cur);
3659 			error = -EFSCORRUPTED;
3660 			goto error0;
3661 		}
3662 		level++;
3663 
3664 		/*
3665 		 * See if the cursor we just used is trash.
3666 		 * Can't trash the caller's cursor, but otherwise we should
3667 		 * if ncur is a new cursor or we're about to be done.
3668 		 */
3669 		if (pcur != cur &&
3670 		    (ncur || xfs_btree_ptr_is_null(cur, &nptr))) {
3671 			/* Save the state from the cursor before we trash it */
3672 			if (cur->bc_ops->update_cursor &&
3673 			    !(cur->bc_flags & XFS_BTREE_STAGING))
3674 				cur->bc_ops->update_cursor(pcur, cur);
3675 			cur->bc_nlevels = pcur->bc_nlevels;
3676 			xfs_btree_del_cursor(pcur, XFS_BTREE_NOERROR);
3677 		}
3678 		/* If we got a new cursor, switch to it. */
3679 		if (ncur) {
3680 			pcur = ncur;
3681 			ncur = NULL;
3682 		}
3683 	} while (!xfs_btree_ptr_is_null(cur, &nptr));
3684 
3685 	*stat = i;
3686 	return 0;
3687 error0:
3688 	return error;
3689 }
3690 
3691 /*
3692  * Try to merge a non-leaf block back into the inode root.
3693  *
3694  * Note: the killroot names comes from the fact that we're effectively
3695  * killing the old root block.  But because we can't just delete the
3696  * inode we have to copy the single block it was pointing to into the
3697  * inode.
3698  */
3699 STATIC int
xfs_btree_kill_iroot(struct xfs_btree_cur * cur)3700 xfs_btree_kill_iroot(
3701 	struct xfs_btree_cur	*cur)
3702 {
3703 	int			whichfork = cur->bc_ino.whichfork;
3704 	struct xfs_inode	*ip = cur->bc_ino.ip;
3705 	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
3706 	struct xfs_btree_block	*block;
3707 	struct xfs_btree_block	*cblock;
3708 	union xfs_btree_key	*kp;
3709 	union xfs_btree_key	*ckp;
3710 	union xfs_btree_ptr	*pp;
3711 	union xfs_btree_ptr	*cpp;
3712 	struct xfs_buf		*cbp;
3713 	int			level;
3714 	int			index;
3715 	int			numrecs;
3716 	int			error;
3717 #ifdef DEBUG
3718 	union xfs_btree_ptr	ptr;
3719 #endif
3720 	int			i;
3721 
3722 	ASSERT(cur->bc_ops->type == XFS_BTREE_TYPE_INODE);
3723 	ASSERT(cur->bc_nlevels > 1);
3724 
3725 	/*
3726 	 * Don't deal with the root block needs to be a leaf case.
3727 	 * We're just going to turn the thing back into extents anyway.
3728 	 */
3729 	level = cur->bc_nlevels - 1;
3730 	if (level == 1)
3731 		goto out0;
3732 
3733 	/*
3734 	 * Give up if the root has multiple children.
3735 	 */
3736 	block = xfs_btree_get_iroot(cur);
3737 	if (xfs_btree_get_numrecs(block) != 1)
3738 		goto out0;
3739 
3740 	cblock = xfs_btree_get_block(cur, level - 1, &cbp);
3741 	numrecs = xfs_btree_get_numrecs(cblock);
3742 
3743 	/*
3744 	 * Only do this if the next level will fit.
3745 	 * Then the data must be copied up to the inode,
3746 	 * instead of freeing the root you free the next level.
3747 	 */
3748 	if (numrecs > cur->bc_ops->get_dmaxrecs(cur, level))
3749 		goto out0;
3750 
3751 	XFS_BTREE_STATS_INC(cur, killroot);
3752 
3753 #ifdef DEBUG
3754 	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
3755 	ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
3756 	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
3757 	ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
3758 #endif
3759 
3760 	index = numrecs - cur->bc_ops->get_maxrecs(cur, level);
3761 	if (index) {
3762 		xfs_iroot_realloc(cur->bc_ino.ip, index,
3763 				  cur->bc_ino.whichfork);
3764 		block = ifp->if_broot;
3765 	}
3766 
3767 	be16_add_cpu(&block->bb_numrecs, index);
3768 	ASSERT(block->bb_numrecs == cblock->bb_numrecs);
3769 
3770 	kp = xfs_btree_key_addr(cur, 1, block);
3771 	ckp = xfs_btree_key_addr(cur, 1, cblock);
3772 	xfs_btree_copy_keys(cur, kp, ckp, numrecs);
3773 
3774 	pp = xfs_btree_ptr_addr(cur, 1, block);
3775 	cpp = xfs_btree_ptr_addr(cur, 1, cblock);
3776 
3777 	for (i = 0; i < numrecs; i++) {
3778 		error = xfs_btree_debug_check_ptr(cur, cpp, i, level - 1);
3779 		if (error)
3780 			return error;
3781 	}
3782 
3783 	xfs_btree_copy_ptrs(cur, pp, cpp, numrecs);
3784 
3785 	error = xfs_btree_free_block(cur, cbp);
3786 	if (error)
3787 		return error;
3788 
3789 	cur->bc_levels[level - 1].bp = NULL;
3790 	be16_add_cpu(&block->bb_level, -1);
3791 	xfs_trans_log_inode(cur->bc_tp, ip,
3792 		XFS_ILOG_CORE | xfs_ilog_fbroot(cur->bc_ino.whichfork));
3793 	cur->bc_nlevels--;
3794 out0:
3795 	return 0;
3796 }
3797 
3798 /*
3799  * Kill the current root node, and replace it with it's only child node.
3800  */
3801 STATIC int
xfs_btree_kill_root(struct xfs_btree_cur * cur,struct xfs_buf * bp,int level,union xfs_btree_ptr * newroot)3802 xfs_btree_kill_root(
3803 	struct xfs_btree_cur	*cur,
3804 	struct xfs_buf		*bp,
3805 	int			level,
3806 	union xfs_btree_ptr	*newroot)
3807 {
3808 	int			error;
3809 
3810 	XFS_BTREE_STATS_INC(cur, killroot);
3811 
3812 	/*
3813 	 * Update the root pointer, decreasing the level by 1 and then
3814 	 * free the old root.
3815 	 */
3816 	xfs_btree_set_root(cur, newroot, -1);
3817 
3818 	error = xfs_btree_free_block(cur, bp);
3819 	if (error)
3820 		return error;
3821 
3822 	cur->bc_levels[level].bp = NULL;
3823 	cur->bc_levels[level].ra = 0;
3824 	cur->bc_nlevels--;
3825 
3826 	return 0;
3827 }
3828 
3829 STATIC int
xfs_btree_dec_cursor(struct xfs_btree_cur * cur,int level,int * stat)3830 xfs_btree_dec_cursor(
3831 	struct xfs_btree_cur	*cur,
3832 	int			level,
3833 	int			*stat)
3834 {
3835 	int			error;
3836 	int			i;
3837 
3838 	if (level > 0) {
3839 		error = xfs_btree_decrement(cur, level, &i);
3840 		if (error)
3841 			return error;
3842 	}
3843 
3844 	*stat = 1;
3845 	return 0;
3846 }
3847 
3848 /*
3849  * Single level of the btree record deletion routine.
3850  * Delete record pointed to by cur/level.
3851  * Remove the record from its block then rebalance the tree.
3852  * Return 0 for error, 1 for done, 2 to go on to the next level.
3853  */
3854 STATIC int					/* error */
xfs_btree_delrec(struct xfs_btree_cur * cur,int level,int * stat)3855 xfs_btree_delrec(
3856 	struct xfs_btree_cur	*cur,		/* btree cursor */
3857 	int			level,		/* level removing record from */
3858 	int			*stat)		/* fail/done/go-on */
3859 {
3860 	struct xfs_btree_block	*block;		/* btree block */
3861 	union xfs_btree_ptr	cptr;		/* current block ptr */
3862 	struct xfs_buf		*bp;		/* buffer for block */
3863 	int			error;		/* error return value */
3864 	int			i;		/* loop counter */
3865 	union xfs_btree_ptr	lptr;		/* left sibling block ptr */
3866 	struct xfs_buf		*lbp;		/* left buffer pointer */
3867 	struct xfs_btree_block	*left;		/* left btree block */
3868 	int			lrecs = 0;	/* left record count */
3869 	int			ptr;		/* key/record index */
3870 	union xfs_btree_ptr	rptr;		/* right sibling block ptr */
3871 	struct xfs_buf		*rbp;		/* right buffer pointer */
3872 	struct xfs_btree_block	*right;		/* right btree block */
3873 	struct xfs_btree_block	*rrblock;	/* right-right btree block */
3874 	struct xfs_buf		*rrbp;		/* right-right buffer pointer */
3875 	int			rrecs = 0;	/* right record count */
3876 	struct xfs_btree_cur	*tcur;		/* temporary btree cursor */
3877 	int			numrecs;	/* temporary numrec count */
3878 
3879 	tcur = NULL;
3880 
3881 	/* Get the index of the entry being deleted, check for nothing there. */
3882 	ptr = cur->bc_levels[level].ptr;
3883 	if (ptr == 0) {
3884 		*stat = 0;
3885 		return 0;
3886 	}
3887 
3888 	/* Get the buffer & block containing the record or key/ptr. */
3889 	block = xfs_btree_get_block(cur, level, &bp);
3890 	numrecs = xfs_btree_get_numrecs(block);
3891 
3892 #ifdef DEBUG
3893 	error = xfs_btree_check_block(cur, block, level, bp);
3894 	if (error)
3895 		goto error0;
3896 #endif
3897 
3898 	/* Fail if we're off the end of the block. */
3899 	if (ptr > numrecs) {
3900 		*stat = 0;
3901 		return 0;
3902 	}
3903 
3904 	XFS_BTREE_STATS_INC(cur, delrec);
3905 	XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr);
3906 
3907 	/* Excise the entries being deleted. */
3908 	if (level > 0) {
3909 		/* It's a nonleaf. operate on keys and ptrs */
3910 		union xfs_btree_key	*lkp;
3911 		union xfs_btree_ptr	*lpp;
3912 
3913 		lkp = xfs_btree_key_addr(cur, ptr + 1, block);
3914 		lpp = xfs_btree_ptr_addr(cur, ptr + 1, block);
3915 
3916 		for (i = 0; i < numrecs - ptr; i++) {
3917 			error = xfs_btree_debug_check_ptr(cur, lpp, i, level);
3918 			if (error)
3919 				goto error0;
3920 		}
3921 
3922 		if (ptr < numrecs) {
3923 			xfs_btree_shift_keys(cur, lkp, -1, numrecs - ptr);
3924 			xfs_btree_shift_ptrs(cur, lpp, -1, numrecs - ptr);
3925 			xfs_btree_log_keys(cur, bp, ptr, numrecs - 1);
3926 			xfs_btree_log_ptrs(cur, bp, ptr, numrecs - 1);
3927 		}
3928 	} else {
3929 		/* It's a leaf. operate on records */
3930 		if (ptr < numrecs) {
3931 			xfs_btree_shift_recs(cur,
3932 				xfs_btree_rec_addr(cur, ptr + 1, block),
3933 				-1, numrecs - ptr);
3934 			xfs_btree_log_recs(cur, bp, ptr, numrecs - 1);
3935 		}
3936 	}
3937 
3938 	/*
3939 	 * Decrement and log the number of entries in the block.
3940 	 */
3941 	xfs_btree_set_numrecs(block, --numrecs);
3942 	xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
3943 
3944 	/*
3945 	 * We're at the root level.  First, shrink the root block in-memory.
3946 	 * Try to get rid of the next level down.  If we can't then there's
3947 	 * nothing left to do.
3948 	 */
3949 	if (xfs_btree_at_iroot(cur, level)) {
3950 		xfs_iroot_realloc(cur->bc_ino.ip, -1, cur->bc_ino.whichfork);
3951 
3952 		error = xfs_btree_kill_iroot(cur);
3953 		if (error)
3954 			goto error0;
3955 
3956 		error = xfs_btree_dec_cursor(cur, level, stat);
3957 		if (error)
3958 			goto error0;
3959 		*stat = 1;
3960 		return 0;
3961 	}
3962 
3963 	/*
3964 	 * If this is the root level, and there's only one entry left, and it's
3965 	 * NOT the leaf level, then we can get rid of this level.
3966 	 */
3967 	if (level == cur->bc_nlevels - 1) {
3968 		if (numrecs == 1 && level > 0) {
3969 			union xfs_btree_ptr	*pp;
3970 			/*
3971 			 * pp is still set to the first pointer in the block.
3972 			 * Make it the new root of the btree.
3973 			 */
3974 			pp = xfs_btree_ptr_addr(cur, 1, block);
3975 			error = xfs_btree_kill_root(cur, bp, level, pp);
3976 			if (error)
3977 				goto error0;
3978 		} else if (level > 0) {
3979 			error = xfs_btree_dec_cursor(cur, level, stat);
3980 			if (error)
3981 				goto error0;
3982 		}
3983 		*stat = 1;
3984 		return 0;
3985 	}
3986 
3987 	/*
3988 	 * If we deleted the leftmost entry in the block, update the
3989 	 * key values above us in the tree.
3990 	 */
3991 	if (xfs_btree_needs_key_update(cur, ptr)) {
3992 		error = xfs_btree_update_keys(cur, level);
3993 		if (error)
3994 			goto error0;
3995 	}
3996 
3997 	/*
3998 	 * If the number of records remaining in the block is at least
3999 	 * the minimum, we're done.
4000 	 */
4001 	if (numrecs >= cur->bc_ops->get_minrecs(cur, level)) {
4002 		error = xfs_btree_dec_cursor(cur, level, stat);
4003 		if (error)
4004 			goto error0;
4005 		return 0;
4006 	}
4007 
4008 	/*
4009 	 * Otherwise, we have to move some records around to keep the
4010 	 * tree balanced.  Look at the left and right sibling blocks to
4011 	 * see if we can re-balance by moving only one record.
4012 	 */
4013 	xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
4014 	xfs_btree_get_sibling(cur, block, &lptr, XFS_BB_LEFTSIB);
4015 
4016 	if (cur->bc_ops->type == XFS_BTREE_TYPE_INODE) {
4017 		/*
4018 		 * One child of root, need to get a chance to copy its contents
4019 		 * into the root and delete it. Can't go up to next level,
4020 		 * there's nothing to delete there.
4021 		 */
4022 		if (xfs_btree_ptr_is_null(cur, &rptr) &&
4023 		    xfs_btree_ptr_is_null(cur, &lptr) &&
4024 		    level == cur->bc_nlevels - 2) {
4025 			error = xfs_btree_kill_iroot(cur);
4026 			if (!error)
4027 				error = xfs_btree_dec_cursor(cur, level, stat);
4028 			if (error)
4029 				goto error0;
4030 			return 0;
4031 		}
4032 	}
4033 
4034 	ASSERT(!xfs_btree_ptr_is_null(cur, &rptr) ||
4035 	       !xfs_btree_ptr_is_null(cur, &lptr));
4036 
4037 	/*
4038 	 * Duplicate the cursor so our btree manipulations here won't
4039 	 * disrupt the next level up.
4040 	 */
4041 	error = xfs_btree_dup_cursor(cur, &tcur);
4042 	if (error)
4043 		goto error0;
4044 
4045 	/*
4046 	 * If there's a right sibling, see if it's ok to shift an entry
4047 	 * out of it.
4048 	 */
4049 	if (!xfs_btree_ptr_is_null(cur, &rptr)) {
4050 		/*
4051 		 * Move the temp cursor to the last entry in the next block.
4052 		 * Actually any entry but the first would suffice.
4053 		 */
4054 		i = xfs_btree_lastrec(tcur, level);
4055 		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
4056 			xfs_btree_mark_sick(cur);
4057 			error = -EFSCORRUPTED;
4058 			goto error0;
4059 		}
4060 
4061 		error = xfs_btree_increment(tcur, level, &i);
4062 		if (error)
4063 			goto error0;
4064 		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
4065 			xfs_btree_mark_sick(cur);
4066 			error = -EFSCORRUPTED;
4067 			goto error0;
4068 		}
4069 
4070 		i = xfs_btree_lastrec(tcur, level);
4071 		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
4072 			xfs_btree_mark_sick(cur);
4073 			error = -EFSCORRUPTED;
4074 			goto error0;
4075 		}
4076 
4077 		/* Grab a pointer to the block. */
4078 		right = xfs_btree_get_block(tcur, level, &rbp);
4079 #ifdef DEBUG
4080 		error = xfs_btree_check_block(tcur, right, level, rbp);
4081 		if (error)
4082 			goto error0;
4083 #endif
4084 		/* Grab the current block number, for future use. */
4085 		xfs_btree_get_sibling(tcur, right, &cptr, XFS_BB_LEFTSIB);
4086 
4087 		/*
4088 		 * If right block is full enough so that removing one entry
4089 		 * won't make it too empty, and left-shifting an entry out
4090 		 * of right to us works, we're done.
4091 		 */
4092 		if (xfs_btree_get_numrecs(right) - 1 >=
4093 		    cur->bc_ops->get_minrecs(tcur, level)) {
4094 			error = xfs_btree_lshift(tcur, level, &i);
4095 			if (error)
4096 				goto error0;
4097 			if (i) {
4098 				ASSERT(xfs_btree_get_numrecs(block) >=
4099 				       cur->bc_ops->get_minrecs(tcur, level));
4100 
4101 				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
4102 				tcur = NULL;
4103 
4104 				error = xfs_btree_dec_cursor(cur, level, stat);
4105 				if (error)
4106 					goto error0;
4107 				return 0;
4108 			}
4109 		}
4110 
4111 		/*
4112 		 * Otherwise, grab the number of records in right for
4113 		 * future reference, and fix up the temp cursor to point
4114 		 * to our block again (last record).
4115 		 */
4116 		rrecs = xfs_btree_get_numrecs(right);
4117 		if (!xfs_btree_ptr_is_null(cur, &lptr)) {
4118 			i = xfs_btree_firstrec(tcur, level);
4119 			if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
4120 				xfs_btree_mark_sick(cur);
4121 				error = -EFSCORRUPTED;
4122 				goto error0;
4123 			}
4124 
4125 			error = xfs_btree_decrement(tcur, level, &i);
4126 			if (error)
4127 				goto error0;
4128 			if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
4129 				xfs_btree_mark_sick(cur);
4130 				error = -EFSCORRUPTED;
4131 				goto error0;
4132 			}
4133 		}
4134 	}
4135 
4136 	/*
4137 	 * If there's a left sibling, see if it's ok to shift an entry
4138 	 * out of it.
4139 	 */
4140 	if (!xfs_btree_ptr_is_null(cur, &lptr)) {
4141 		/*
4142 		 * Move the temp cursor to the first entry in the
4143 		 * previous block.
4144 		 */
4145 		i = xfs_btree_firstrec(tcur, level);
4146 		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
4147 			xfs_btree_mark_sick(cur);
4148 			error = -EFSCORRUPTED;
4149 			goto error0;
4150 		}
4151 
4152 		error = xfs_btree_decrement(tcur, level, &i);
4153 		if (error)
4154 			goto error0;
4155 		i = xfs_btree_firstrec(tcur, level);
4156 		if (XFS_IS_CORRUPT(cur->bc_mp, i != 1)) {
4157 			xfs_btree_mark_sick(cur);
4158 			error = -EFSCORRUPTED;
4159 			goto error0;
4160 		}
4161 
4162 		/* Grab a pointer to the block. */
4163 		left = xfs_btree_get_block(tcur, level, &lbp);
4164 #ifdef DEBUG
4165 		error = xfs_btree_check_block(cur, left, level, lbp);
4166 		if (error)
4167 			goto error0;
4168 #endif
4169 		/* Grab the current block number, for future use. */
4170 		xfs_btree_get_sibling(tcur, left, &cptr, XFS_BB_RIGHTSIB);
4171 
4172 		/*
4173 		 * If left block is full enough so that removing one entry
4174 		 * won't make it too empty, and right-shifting an entry out
4175 		 * of left to us works, we're done.
4176 		 */
4177 		if (xfs_btree_get_numrecs(left) - 1 >=
4178 		    cur->bc_ops->get_minrecs(tcur, level)) {
4179 			error = xfs_btree_rshift(tcur, level, &i);
4180 			if (error)
4181 				goto error0;
4182 			if (i) {
4183 				ASSERT(xfs_btree_get_numrecs(block) >=
4184 				       cur->bc_ops->get_minrecs(tcur, level));
4185 				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
4186 				tcur = NULL;
4187 				if (level == 0)
4188 					cur->bc_levels[0].ptr++;
4189 
4190 				*stat = 1;
4191 				return 0;
4192 			}
4193 		}
4194 
4195 		/*
4196 		 * Otherwise, grab the number of records in right for
4197 		 * future reference.
4198 		 */
4199 		lrecs = xfs_btree_get_numrecs(left);
4200 	}
4201 
4202 	/* Delete the temp cursor, we're done with it. */
4203 	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
4204 	tcur = NULL;
4205 
4206 	/* If here, we need to do a join to keep the tree balanced. */
4207 	ASSERT(!xfs_btree_ptr_is_null(cur, &cptr));
4208 
4209 	if (!xfs_btree_ptr_is_null(cur, &lptr) &&
4210 	    lrecs + xfs_btree_get_numrecs(block) <=
4211 			cur->bc_ops->get_maxrecs(cur, level)) {
4212 		/*
4213 		 * Set "right" to be the starting block,
4214 		 * "left" to be the left neighbor.
4215 		 */
4216 		rptr = cptr;
4217 		right = block;
4218 		rbp = bp;
4219 		error = xfs_btree_read_buf_block(cur, &lptr, 0, &left, &lbp);
4220 		if (error)
4221 			goto error0;
4222 
4223 	/*
4224 	 * If that won't work, see if we can join with the right neighbor block.
4225 	 */
4226 	} else if (!xfs_btree_ptr_is_null(cur, &rptr) &&
4227 		   rrecs + xfs_btree_get_numrecs(block) <=
4228 			cur->bc_ops->get_maxrecs(cur, level)) {
4229 		/*
4230 		 * Set "left" to be the starting block,
4231 		 * "right" to be the right neighbor.
4232 		 */
4233 		lptr = cptr;
4234 		left = block;
4235 		lbp = bp;
4236 		error = xfs_btree_read_buf_block(cur, &rptr, 0, &right, &rbp);
4237 		if (error)
4238 			goto error0;
4239 
4240 	/*
4241 	 * Otherwise, we can't fix the imbalance.
4242 	 * Just return.  This is probably a logic error, but it's not fatal.
4243 	 */
4244 	} else {
4245 		error = xfs_btree_dec_cursor(cur, level, stat);
4246 		if (error)
4247 			goto error0;
4248 		return 0;
4249 	}
4250 
4251 	rrecs = xfs_btree_get_numrecs(right);
4252 	lrecs = xfs_btree_get_numrecs(left);
4253 
4254 	/*
4255 	 * We're now going to join "left" and "right" by moving all the stuff
4256 	 * in "right" to "left" and deleting "right".
4257 	 */
4258 	XFS_BTREE_STATS_ADD(cur, moves, rrecs);
4259 	if (level > 0) {
4260 		/* It's a non-leaf.  Move keys and pointers. */
4261 		union xfs_btree_key	*lkp;	/* left btree key */
4262 		union xfs_btree_ptr	*lpp;	/* left address pointer */
4263 		union xfs_btree_key	*rkp;	/* right btree key */
4264 		union xfs_btree_ptr	*rpp;	/* right address pointer */
4265 
4266 		lkp = xfs_btree_key_addr(cur, lrecs + 1, left);
4267 		lpp = xfs_btree_ptr_addr(cur, lrecs + 1, left);
4268 		rkp = xfs_btree_key_addr(cur, 1, right);
4269 		rpp = xfs_btree_ptr_addr(cur, 1, right);
4270 
4271 		for (i = 1; i < rrecs; i++) {
4272 			error = xfs_btree_debug_check_ptr(cur, rpp, i, level);
4273 			if (error)
4274 				goto error0;
4275 		}
4276 
4277 		xfs_btree_copy_keys(cur, lkp, rkp, rrecs);
4278 		xfs_btree_copy_ptrs(cur, lpp, rpp, rrecs);
4279 
4280 		xfs_btree_log_keys(cur, lbp, lrecs + 1, lrecs + rrecs);
4281 		xfs_btree_log_ptrs(cur, lbp, lrecs + 1, lrecs + rrecs);
4282 	} else {
4283 		/* It's a leaf.  Move records.  */
4284 		union xfs_btree_rec	*lrp;	/* left record pointer */
4285 		union xfs_btree_rec	*rrp;	/* right record pointer */
4286 
4287 		lrp = xfs_btree_rec_addr(cur, lrecs + 1, left);
4288 		rrp = xfs_btree_rec_addr(cur, 1, right);
4289 
4290 		xfs_btree_copy_recs(cur, lrp, rrp, rrecs);
4291 		xfs_btree_log_recs(cur, lbp, lrecs + 1, lrecs + rrecs);
4292 	}
4293 
4294 	XFS_BTREE_STATS_INC(cur, join);
4295 
4296 	/*
4297 	 * Fix up the number of records and right block pointer in the
4298 	 * surviving block, and log it.
4299 	 */
4300 	xfs_btree_set_numrecs(left, lrecs + rrecs);
4301 	xfs_btree_get_sibling(cur, right, &cptr, XFS_BB_RIGHTSIB);
4302 	xfs_btree_set_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
4303 	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
4304 
4305 	/* If there is a right sibling, point it to the remaining block. */
4306 	xfs_btree_get_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
4307 	if (!xfs_btree_ptr_is_null(cur, &cptr)) {
4308 		error = xfs_btree_read_buf_block(cur, &cptr, 0, &rrblock, &rrbp);
4309 		if (error)
4310 			goto error0;
4311 		xfs_btree_set_sibling(cur, rrblock, &lptr, XFS_BB_LEFTSIB);
4312 		xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
4313 	}
4314 
4315 	/* Free the deleted block. */
4316 	error = xfs_btree_free_block(cur, rbp);
4317 	if (error)
4318 		goto error0;
4319 
4320 	/*
4321 	 * If we joined with the left neighbor, set the buffer in the
4322 	 * cursor to the left block, and fix up the index.
4323 	 */
4324 	if (bp != lbp) {
4325 		cur->bc_levels[level].bp = lbp;
4326 		cur->bc_levels[level].ptr += lrecs;
4327 		cur->bc_levels[level].ra = 0;
4328 	}
4329 	/*
4330 	 * If we joined with the right neighbor and there's a level above
4331 	 * us, increment the cursor at that level.
4332 	 */
4333 	else if (cur->bc_ops->type == XFS_BTREE_TYPE_INODE ||
4334 		 level + 1 < cur->bc_nlevels) {
4335 		error = xfs_btree_increment(cur, level + 1, &i);
4336 		if (error)
4337 			goto error0;
4338 	}
4339 
4340 	/*
4341 	 * Readjust the ptr at this level if it's not a leaf, since it's
4342 	 * still pointing at the deletion point, which makes the cursor
4343 	 * inconsistent.  If this makes the ptr 0, the caller fixes it up.
4344 	 * We can't use decrement because it would change the next level up.
4345 	 */
4346 	if (level > 0)
4347 		cur->bc_levels[level].ptr--;
4348 
4349 	/*
4350 	 * We combined blocks, so we have to update the parent keys if the
4351 	 * btree supports overlapped intervals.  However,
4352 	 * bc_levels[level + 1].ptr points to the old block so that the caller
4353 	 * knows which record to delete.  Therefore, the caller must be savvy
4354 	 * enough to call updkeys for us if we return stat == 2.  The other
4355 	 * exit points from this function don't require deletions further up
4356 	 * the tree, so they can call updkeys directly.
4357 	 */
4358 
4359 	/* Return value means the next level up has something to do. */
4360 	*stat = 2;
4361 	return 0;
4362 
4363 error0:
4364 	if (tcur)
4365 		xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
4366 	return error;
4367 }
4368 
4369 /*
4370  * Delete the record pointed to by cur.
4371  * The cursor refers to the place where the record was (could be inserted)
4372  * when the operation returns.
4373  */
4374 int					/* error */
xfs_btree_delete(struct xfs_btree_cur * cur,int * stat)4375 xfs_btree_delete(
4376 	struct xfs_btree_cur	*cur,
4377 	int			*stat)	/* success/failure */
4378 {
4379 	int			error;	/* error return value */
4380 	int			level;
4381 	int			i;
4382 	bool			joined = false;
4383 
4384 	/*
4385 	 * Go up the tree, starting at leaf level.
4386 	 *
4387 	 * If 2 is returned then a join was done; go to the next level.
4388 	 * Otherwise we are done.
4389 	 */
4390 	for (level = 0, i = 2; i == 2; level++) {
4391 		error = xfs_btree_delrec(cur, level, &i);
4392 		if (error)
4393 			goto error0;
4394 		if (i == 2)
4395 			joined = true;
4396 	}
4397 
4398 	/*
4399 	 * If we combined blocks as part of deleting the record, delrec won't
4400 	 * have updated the parent high keys so we have to do that here.
4401 	 */
4402 	if (joined && (cur->bc_ops->geom_flags & XFS_BTGEO_OVERLAPPING)) {
4403 		error = xfs_btree_updkeys_force(cur, 0);
4404 		if (error)
4405 			goto error0;
4406 	}
4407 
4408 	if (i == 0) {
4409 		for (level = 1; level < cur->bc_nlevels; level++) {
4410 			if (cur->bc_levels[level].ptr == 0) {
4411 				error = xfs_btree_decrement(cur, level, &i);
4412 				if (error)
4413 					goto error0;
4414 				break;
4415 			}
4416 		}
4417 	}
4418 
4419 	*stat = i;
4420 	return 0;
4421 error0:
4422 	return error;
4423 }
4424 
4425 /*
4426  * Get the data from the pointed-to record.
4427  */
4428 int					/* error */
xfs_btree_get_rec(struct xfs_btree_cur * cur,union xfs_btree_rec ** recp,int * stat)4429 xfs_btree_get_rec(
4430 	struct xfs_btree_cur	*cur,	/* btree cursor */
4431 	union xfs_btree_rec	**recp,	/* output: btree record */
4432 	int			*stat)	/* output: success/failure */
4433 {
4434 	struct xfs_btree_block	*block;	/* btree block */
4435 	struct xfs_buf		*bp;	/* buffer pointer */
4436 	int			ptr;	/* record number */
4437 #ifdef DEBUG
4438 	int			error;	/* error return value */
4439 #endif
4440 
4441 	ptr = cur->bc_levels[0].ptr;
4442 	block = xfs_btree_get_block(cur, 0, &bp);
4443 
4444 #ifdef DEBUG
4445 	error = xfs_btree_check_block(cur, block, 0, bp);
4446 	if (error)
4447 		return error;
4448 #endif
4449 
4450 	/*
4451 	 * Off the right end or left end, return failure.
4452 	 */
4453 	if (ptr > xfs_btree_get_numrecs(block) || ptr <= 0) {
4454 		*stat = 0;
4455 		return 0;
4456 	}
4457 
4458 	/*
4459 	 * Point to the record and extract its data.
4460 	 */
4461 	*recp = xfs_btree_rec_addr(cur, ptr, block);
4462 	*stat = 1;
4463 	return 0;
4464 }
4465 
4466 /* Visit a block in a btree. */
4467 STATIC int
xfs_btree_visit_block(struct xfs_btree_cur * cur,int level,xfs_btree_visit_blocks_fn fn,void * data)4468 xfs_btree_visit_block(
4469 	struct xfs_btree_cur		*cur,
4470 	int				level,
4471 	xfs_btree_visit_blocks_fn	fn,
4472 	void				*data)
4473 {
4474 	struct xfs_btree_block		*block;
4475 	struct xfs_buf			*bp;
4476 	union xfs_btree_ptr		rptr, bufptr;
4477 	int				error;
4478 
4479 	/* do right sibling readahead */
4480 	xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
4481 	block = xfs_btree_get_block(cur, level, &bp);
4482 
4483 	/* process the block */
4484 	error = fn(cur, level, data);
4485 	if (error)
4486 		return error;
4487 
4488 	/* now read rh sibling block for next iteration */
4489 	xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
4490 	if (xfs_btree_ptr_is_null(cur, &rptr))
4491 		return -ENOENT;
4492 
4493 	/*
4494 	 * We only visit blocks once in this walk, so we have to avoid the
4495 	 * internal xfs_btree_lookup_get_block() optimisation where it will
4496 	 * return the same block without checking if the right sibling points
4497 	 * back to us and creates a cyclic reference in the btree.
4498 	 */
4499 	xfs_btree_buf_to_ptr(cur, bp, &bufptr);
4500 	if (xfs_btree_ptrs_equal(cur, &rptr, &bufptr)) {
4501 		xfs_btree_mark_sick(cur);
4502 		return -EFSCORRUPTED;
4503 	}
4504 
4505 	return xfs_btree_lookup_get_block(cur, level, &rptr, &block);
4506 }
4507 
4508 
4509 /* Visit every block in a btree. */
4510 int
xfs_btree_visit_blocks(struct xfs_btree_cur * cur,xfs_btree_visit_blocks_fn fn,unsigned int flags,void * data)4511 xfs_btree_visit_blocks(
4512 	struct xfs_btree_cur		*cur,
4513 	xfs_btree_visit_blocks_fn	fn,
4514 	unsigned int			flags,
4515 	void				*data)
4516 {
4517 	union xfs_btree_ptr		lptr;
4518 	int				level;
4519 	struct xfs_btree_block		*block = NULL;
4520 	int				error = 0;
4521 
4522 	xfs_btree_init_ptr_from_cur(cur, &lptr);
4523 
4524 	/* for each level */
4525 	for (level = cur->bc_nlevels - 1; level >= 0; level--) {
4526 		/* grab the left hand block */
4527 		error = xfs_btree_lookup_get_block(cur, level, &lptr, &block);
4528 		if (error)
4529 			return error;
4530 
4531 		/* readahead the left most block for the next level down */
4532 		if (level > 0) {
4533 			union xfs_btree_ptr     *ptr;
4534 
4535 			ptr = xfs_btree_ptr_addr(cur, 1, block);
4536 			xfs_btree_readahead_ptr(cur, ptr, 1);
4537 
4538 			/* save for the next iteration of the loop */
4539 			xfs_btree_copy_ptrs(cur, &lptr, ptr, 1);
4540 
4541 			if (!(flags & XFS_BTREE_VISIT_LEAVES))
4542 				continue;
4543 		} else if (!(flags & XFS_BTREE_VISIT_RECORDS)) {
4544 			continue;
4545 		}
4546 
4547 		/* for each buffer in the level */
4548 		do {
4549 			error = xfs_btree_visit_block(cur, level, fn, data);
4550 		} while (!error);
4551 
4552 		if (error != -ENOENT)
4553 			return error;
4554 	}
4555 
4556 	return 0;
4557 }
4558 
4559 /*
4560  * Change the owner of a btree.
4561  *
4562  * The mechanism we use here is ordered buffer logging. Because we don't know
4563  * how many buffers were are going to need to modify, we don't really want to
4564  * have to make transaction reservations for the worst case of every buffer in a
4565  * full size btree as that may be more space that we can fit in the log....
4566  *
4567  * We do the btree walk in the most optimal manner possible - we have sibling
4568  * pointers so we can just walk all the blocks on each level from left to right
4569  * in a single pass, and then move to the next level and do the same. We can
4570  * also do readahead on the sibling pointers to get IO moving more quickly,
4571  * though for slow disks this is unlikely to make much difference to performance
4572  * as the amount of CPU work we have to do before moving to the next block is
4573  * relatively small.
4574  *
4575  * For each btree block that we load, modify the owner appropriately, set the
4576  * buffer as an ordered buffer and log it appropriately. We need to ensure that
4577  * we mark the region we change dirty so that if the buffer is relogged in
4578  * a subsequent transaction the changes we make here as an ordered buffer are
4579  * correctly relogged in that transaction.  If we are in recovery context, then
4580  * just queue the modified buffer as delayed write buffer so the transaction
4581  * recovery completion writes the changes to disk.
4582  */
4583 struct xfs_btree_block_change_owner_info {
4584 	uint64_t		new_owner;
4585 	struct list_head	*buffer_list;
4586 };
4587 
4588 static int
xfs_btree_block_change_owner(struct xfs_btree_cur * cur,int level,void * data)4589 xfs_btree_block_change_owner(
4590 	struct xfs_btree_cur	*cur,
4591 	int			level,
4592 	void			*data)
4593 {
4594 	struct xfs_btree_block_change_owner_info	*bbcoi = data;
4595 	struct xfs_btree_block	*block;
4596 	struct xfs_buf		*bp;
4597 
4598 	/* modify the owner */
4599 	block = xfs_btree_get_block(cur, level, &bp);
4600 	if (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN) {
4601 		if (block->bb_u.l.bb_owner == cpu_to_be64(bbcoi->new_owner))
4602 			return 0;
4603 		block->bb_u.l.bb_owner = cpu_to_be64(bbcoi->new_owner);
4604 	} else {
4605 		if (block->bb_u.s.bb_owner == cpu_to_be32(bbcoi->new_owner))
4606 			return 0;
4607 		block->bb_u.s.bb_owner = cpu_to_be32(bbcoi->new_owner);
4608 	}
4609 
4610 	/*
4611 	 * If the block is a root block hosted in an inode, we might not have a
4612 	 * buffer pointer here and we shouldn't attempt to log the change as the
4613 	 * information is already held in the inode and discarded when the root
4614 	 * block is formatted into the on-disk inode fork. We still change it,
4615 	 * though, so everything is consistent in memory.
4616 	 */
4617 	if (!bp) {
4618 		ASSERT(cur->bc_ops->type == XFS_BTREE_TYPE_INODE);
4619 		ASSERT(level == cur->bc_nlevels - 1);
4620 		return 0;
4621 	}
4622 
4623 	if (cur->bc_tp) {
4624 		if (!xfs_trans_ordered_buf(cur->bc_tp, bp)) {
4625 			xfs_btree_log_block(cur, bp, XFS_BB_OWNER);
4626 			return -EAGAIN;
4627 		}
4628 	} else {
4629 		xfs_buf_delwri_queue(bp, bbcoi->buffer_list);
4630 	}
4631 
4632 	return 0;
4633 }
4634 
4635 int
xfs_btree_change_owner(struct xfs_btree_cur * cur,uint64_t new_owner,struct list_head * buffer_list)4636 xfs_btree_change_owner(
4637 	struct xfs_btree_cur	*cur,
4638 	uint64_t		new_owner,
4639 	struct list_head	*buffer_list)
4640 {
4641 	struct xfs_btree_block_change_owner_info	bbcoi;
4642 
4643 	bbcoi.new_owner = new_owner;
4644 	bbcoi.buffer_list = buffer_list;
4645 
4646 	return xfs_btree_visit_blocks(cur, xfs_btree_block_change_owner,
4647 			XFS_BTREE_VISIT_ALL, &bbcoi);
4648 }
4649 
4650 /* Verify the v5 fields of a long-format btree block. */
4651 xfs_failaddr_t
xfs_btree_fsblock_v5hdr_verify(struct xfs_buf * bp,uint64_t owner)4652 xfs_btree_fsblock_v5hdr_verify(
4653 	struct xfs_buf		*bp,
4654 	uint64_t		owner)
4655 {
4656 	struct xfs_mount	*mp = bp->b_mount;
4657 	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
4658 
4659 	if (!xfs_has_crc(mp))
4660 		return __this_address;
4661 	if (!uuid_equal(&block->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid))
4662 		return __this_address;
4663 	if (block->bb_u.l.bb_blkno != cpu_to_be64(xfs_buf_daddr(bp)))
4664 		return __this_address;
4665 	if (owner != XFS_RMAP_OWN_UNKNOWN &&
4666 	    be64_to_cpu(block->bb_u.l.bb_owner) != owner)
4667 		return __this_address;
4668 	return NULL;
4669 }
4670 
4671 /* Verify a long-format btree block. */
4672 xfs_failaddr_t
xfs_btree_fsblock_verify(struct xfs_buf * bp,unsigned int max_recs)4673 xfs_btree_fsblock_verify(
4674 	struct xfs_buf		*bp,
4675 	unsigned int		max_recs)
4676 {
4677 	struct xfs_mount	*mp = bp->b_mount;
4678 	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
4679 	xfs_fsblock_t		fsb;
4680 	xfs_failaddr_t		fa;
4681 
4682 	ASSERT(!xfs_buftarg_is_mem(bp->b_target));
4683 
4684 	/* numrecs verification */
4685 	if (be16_to_cpu(block->bb_numrecs) > max_recs)
4686 		return __this_address;
4687 
4688 	/* sibling pointer verification */
4689 	fsb = XFS_DADDR_TO_FSB(mp, xfs_buf_daddr(bp));
4690 	fa = xfs_btree_check_fsblock_siblings(mp, fsb,
4691 			block->bb_u.l.bb_leftsib);
4692 	if (!fa)
4693 		fa = xfs_btree_check_fsblock_siblings(mp, fsb,
4694 				block->bb_u.l.bb_rightsib);
4695 	return fa;
4696 }
4697 
4698 /* Verify an in-memory btree block. */
4699 xfs_failaddr_t
xfs_btree_memblock_verify(struct xfs_buf * bp,unsigned int max_recs)4700 xfs_btree_memblock_verify(
4701 	struct xfs_buf		*bp,
4702 	unsigned int		max_recs)
4703 {
4704 	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
4705 	struct xfs_buftarg	*btp = bp->b_target;
4706 	xfs_failaddr_t		fa;
4707 	xfbno_t			bno;
4708 
4709 	ASSERT(xfs_buftarg_is_mem(bp->b_target));
4710 
4711 	/* numrecs verification */
4712 	if (be16_to_cpu(block->bb_numrecs) > max_recs)
4713 		return __this_address;
4714 
4715 	/* sibling pointer verification */
4716 	bno = xfs_daddr_to_xfbno(xfs_buf_daddr(bp));
4717 	fa = xfs_btree_check_memblock_siblings(btp, bno,
4718 			block->bb_u.l.bb_leftsib);
4719 	if (fa)
4720 		return fa;
4721 	fa = xfs_btree_check_memblock_siblings(btp, bno,
4722 			block->bb_u.l.bb_rightsib);
4723 	if (fa)
4724 		return fa;
4725 
4726 	return NULL;
4727 }
4728 /**
4729  * xfs_btree_agblock_v5hdr_verify() -- verify the v5 fields of a short-format
4730  *				      btree block
4731  *
4732  * @bp: buffer containing the btree block
4733  */
4734 xfs_failaddr_t
xfs_btree_agblock_v5hdr_verify(struct xfs_buf * bp)4735 xfs_btree_agblock_v5hdr_verify(
4736 	struct xfs_buf		*bp)
4737 {
4738 	struct xfs_mount	*mp = bp->b_mount;
4739 	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
4740 	struct xfs_perag	*pag = bp->b_pag;
4741 
4742 	if (!xfs_has_crc(mp))
4743 		return __this_address;
4744 	if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid))
4745 		return __this_address;
4746 	if (block->bb_u.s.bb_blkno != cpu_to_be64(xfs_buf_daddr(bp)))
4747 		return __this_address;
4748 	if (pag && be32_to_cpu(block->bb_u.s.bb_owner) != pag->pag_agno)
4749 		return __this_address;
4750 	return NULL;
4751 }
4752 
4753 /**
4754  * xfs_btree_agblock_verify() -- verify a short-format btree block
4755  *
4756  * @bp: buffer containing the btree block
4757  * @max_recs: maximum records allowed in this btree node
4758  */
4759 xfs_failaddr_t
xfs_btree_agblock_verify(struct xfs_buf * bp,unsigned int max_recs)4760 xfs_btree_agblock_verify(
4761 	struct xfs_buf		*bp,
4762 	unsigned int		max_recs)
4763 {
4764 	struct xfs_mount	*mp = bp->b_mount;
4765 	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
4766 	xfs_agblock_t		agbno;
4767 	xfs_failaddr_t		fa;
4768 
4769 	ASSERT(!xfs_buftarg_is_mem(bp->b_target));
4770 
4771 	/* numrecs verification */
4772 	if (be16_to_cpu(block->bb_numrecs) > max_recs)
4773 		return __this_address;
4774 
4775 	/* sibling pointer verification */
4776 	agbno = xfs_daddr_to_agbno(mp, xfs_buf_daddr(bp));
4777 	fa = xfs_btree_check_agblock_siblings(bp->b_pag, agbno,
4778 			block->bb_u.s.bb_leftsib);
4779 	if (!fa)
4780 		fa = xfs_btree_check_agblock_siblings(bp->b_pag, agbno,
4781 				block->bb_u.s.bb_rightsib);
4782 	return fa;
4783 }
4784 
4785 /*
4786  * For the given limits on leaf and keyptr records per block, calculate the
4787  * height of the tree needed to index the number of leaf records.
4788  */
4789 unsigned int
xfs_btree_compute_maxlevels(const unsigned int * limits,unsigned long long records)4790 xfs_btree_compute_maxlevels(
4791 	const unsigned int	*limits,
4792 	unsigned long long	records)
4793 {
4794 	unsigned long long	level_blocks = howmany_64(records, limits[0]);
4795 	unsigned int		height = 1;
4796 
4797 	while (level_blocks > 1) {
4798 		level_blocks = howmany_64(level_blocks, limits[1]);
4799 		height++;
4800 	}
4801 
4802 	return height;
4803 }
4804 
4805 /*
4806  * For the given limits on leaf and keyptr records per block, calculate the
4807  * number of blocks needed to index the given number of leaf records.
4808  */
4809 unsigned long long
xfs_btree_calc_size(const unsigned int * limits,unsigned long long records)4810 xfs_btree_calc_size(
4811 	const unsigned int	*limits,
4812 	unsigned long long	records)
4813 {
4814 	unsigned long long	level_blocks = howmany_64(records, limits[0]);
4815 	unsigned long long	blocks = level_blocks;
4816 
4817 	while (level_blocks > 1) {
4818 		level_blocks = howmany_64(level_blocks, limits[1]);
4819 		blocks += level_blocks;
4820 	}
4821 
4822 	return blocks;
4823 }
4824 
4825 /*
4826  * Given a number of available blocks for the btree to consume with records and
4827  * pointers, calculate the height of the tree needed to index all the records
4828  * that space can hold based on the number of pointers each interior node
4829  * holds.
4830  *
4831  * We start by assuming a single level tree consumes a single block, then track
4832  * the number of blocks each node level consumes until we no longer have space
4833  * to store the next node level. At this point, we are indexing all the leaf
4834  * blocks in the space, and there's no more free space to split the tree any
4835  * further. That's our maximum btree height.
4836  */
4837 unsigned int
xfs_btree_space_to_height(const unsigned int * limits,unsigned long long leaf_blocks)4838 xfs_btree_space_to_height(
4839 	const unsigned int	*limits,
4840 	unsigned long long	leaf_blocks)
4841 {
4842 	/*
4843 	 * The root btree block can have fewer than minrecs pointers in it
4844 	 * because the tree might not be big enough to require that amount of
4845 	 * fanout. Hence it has a minimum size of 2 pointers, not limits[1].
4846 	 */
4847 	unsigned long long	node_blocks = 2;
4848 	unsigned long long	blocks_left = leaf_blocks - 1;
4849 	unsigned int		height = 1;
4850 
4851 	if (leaf_blocks < 1)
4852 		return 0;
4853 
4854 	while (node_blocks < blocks_left) {
4855 		blocks_left -= node_blocks;
4856 		node_blocks *= limits[1];
4857 		height++;
4858 	}
4859 
4860 	return height;
4861 }
4862 
4863 /*
4864  * Query a regular btree for all records overlapping a given interval.
4865  * Start with a LE lookup of the key of low_rec and return all records
4866  * until we find a record with a key greater than the key of high_rec.
4867  */
4868 STATIC int
xfs_btree_simple_query_range(struct xfs_btree_cur * cur,const union xfs_btree_key * low_key,const union xfs_btree_key * high_key,xfs_btree_query_range_fn fn,void * priv)4869 xfs_btree_simple_query_range(
4870 	struct xfs_btree_cur		*cur,
4871 	const union xfs_btree_key	*low_key,
4872 	const union xfs_btree_key	*high_key,
4873 	xfs_btree_query_range_fn	fn,
4874 	void				*priv)
4875 {
4876 	union xfs_btree_rec		*recp;
4877 	union xfs_btree_key		rec_key;
4878 	int				stat;
4879 	bool				firstrec = true;
4880 	int				error;
4881 
4882 	ASSERT(cur->bc_ops->init_high_key_from_rec);
4883 	ASSERT(cur->bc_ops->diff_two_keys);
4884 
4885 	/*
4886 	 * Find the leftmost record.  The btree cursor must be set
4887 	 * to the low record used to generate low_key.
4888 	 */
4889 	stat = 0;
4890 	error = xfs_btree_lookup(cur, XFS_LOOKUP_LE, &stat);
4891 	if (error)
4892 		goto out;
4893 
4894 	/* Nothing?  See if there's anything to the right. */
4895 	if (!stat) {
4896 		error = xfs_btree_increment(cur, 0, &stat);
4897 		if (error)
4898 			goto out;
4899 	}
4900 
4901 	while (stat) {
4902 		/* Find the record. */
4903 		error = xfs_btree_get_rec(cur, &recp, &stat);
4904 		if (error || !stat)
4905 			break;
4906 
4907 		/* Skip if low_key > high_key(rec). */
4908 		if (firstrec) {
4909 			cur->bc_ops->init_high_key_from_rec(&rec_key, recp);
4910 			firstrec = false;
4911 			if (xfs_btree_keycmp_gt(cur, low_key, &rec_key))
4912 				goto advloop;
4913 		}
4914 
4915 		/* Stop if low_key(rec) > high_key. */
4916 		cur->bc_ops->init_key_from_rec(&rec_key, recp);
4917 		if (xfs_btree_keycmp_gt(cur, &rec_key, high_key))
4918 			break;
4919 
4920 		/* Callback */
4921 		error = fn(cur, recp, priv);
4922 		if (error)
4923 			break;
4924 
4925 advloop:
4926 		/* Move on to the next record. */
4927 		error = xfs_btree_increment(cur, 0, &stat);
4928 		if (error)
4929 			break;
4930 	}
4931 
4932 out:
4933 	return error;
4934 }
4935 
4936 /*
4937  * Query an overlapped interval btree for all records overlapping a given
4938  * interval.  This function roughly follows the algorithm given in
4939  * "Interval Trees" of _Introduction to Algorithms_, which is section
4940  * 14.3 in the 2nd and 3rd editions.
4941  *
4942  * First, generate keys for the low and high records passed in.
4943  *
4944  * For any leaf node, generate the high and low keys for the record.
4945  * If the record keys overlap with the query low/high keys, pass the
4946  * record to the function iterator.
4947  *
4948  * For any internal node, compare the low and high keys of each
4949  * pointer against the query low/high keys.  If there's an overlap,
4950  * follow the pointer.
4951  *
4952  * As an optimization, we stop scanning a block when we find a low key
4953  * that is greater than the query's high key.
4954  */
4955 STATIC int
xfs_btree_overlapped_query_range(struct xfs_btree_cur * cur,const union xfs_btree_key * low_key,const union xfs_btree_key * high_key,xfs_btree_query_range_fn fn,void * priv)4956 xfs_btree_overlapped_query_range(
4957 	struct xfs_btree_cur		*cur,
4958 	const union xfs_btree_key	*low_key,
4959 	const union xfs_btree_key	*high_key,
4960 	xfs_btree_query_range_fn	fn,
4961 	void				*priv)
4962 {
4963 	union xfs_btree_ptr		ptr;
4964 	union xfs_btree_ptr		*pp;
4965 	union xfs_btree_key		rec_key;
4966 	union xfs_btree_key		rec_hkey;
4967 	union xfs_btree_key		*lkp;
4968 	union xfs_btree_key		*hkp;
4969 	union xfs_btree_rec		*recp;
4970 	struct xfs_btree_block		*block;
4971 	int				level;
4972 	struct xfs_buf			*bp;
4973 	int				i;
4974 	int				error;
4975 
4976 	/* Load the root of the btree. */
4977 	level = cur->bc_nlevels - 1;
4978 	xfs_btree_init_ptr_from_cur(cur, &ptr);
4979 	error = xfs_btree_lookup_get_block(cur, level, &ptr, &block);
4980 	if (error)
4981 		return error;
4982 	xfs_btree_get_block(cur, level, &bp);
4983 	trace_xfs_btree_overlapped_query_range(cur, level, bp);
4984 #ifdef DEBUG
4985 	error = xfs_btree_check_block(cur, block, level, bp);
4986 	if (error)
4987 		goto out;
4988 #endif
4989 	cur->bc_levels[level].ptr = 1;
4990 
4991 	while (level < cur->bc_nlevels) {
4992 		block = xfs_btree_get_block(cur, level, &bp);
4993 
4994 		/* End of node, pop back towards the root. */
4995 		if (cur->bc_levels[level].ptr >
4996 					be16_to_cpu(block->bb_numrecs)) {
4997 pop_up:
4998 			if (level < cur->bc_nlevels - 1)
4999 				cur->bc_levels[level + 1].ptr++;
5000 			level++;
5001 			continue;
5002 		}
5003 
5004 		if (level == 0) {
5005 			/* Handle a leaf node. */
5006 			recp = xfs_btree_rec_addr(cur, cur->bc_levels[0].ptr,
5007 					block);
5008 
5009 			cur->bc_ops->init_high_key_from_rec(&rec_hkey, recp);
5010 			cur->bc_ops->init_key_from_rec(&rec_key, recp);
5011 
5012 			/*
5013 			 * If (query's high key < record's low key), then there
5014 			 * are no more interesting records in this block.  Pop
5015 			 * up to the leaf level to find more record blocks.
5016 			 *
5017 			 * If (record's high key >= query's low key) and
5018 			 *    (query's high key >= record's low key), then
5019 			 * this record overlaps the query range; callback.
5020 			 */
5021 			if (xfs_btree_keycmp_lt(cur, high_key, &rec_key))
5022 				goto pop_up;
5023 			if (xfs_btree_keycmp_ge(cur, &rec_hkey, low_key)) {
5024 				error = fn(cur, recp, priv);
5025 				if (error)
5026 					break;
5027 			}
5028 			cur->bc_levels[level].ptr++;
5029 			continue;
5030 		}
5031 
5032 		/* Handle an internal node. */
5033 		lkp = xfs_btree_key_addr(cur, cur->bc_levels[level].ptr, block);
5034 		hkp = xfs_btree_high_key_addr(cur, cur->bc_levels[level].ptr,
5035 				block);
5036 		pp = xfs_btree_ptr_addr(cur, cur->bc_levels[level].ptr, block);
5037 
5038 		/*
5039 		 * If (query's high key < pointer's low key), then there are no
5040 		 * more interesting keys in this block.  Pop up one leaf level
5041 		 * to continue looking for records.
5042 		 *
5043 		 * If (pointer's high key >= query's low key) and
5044 		 *    (query's high key >= pointer's low key), then
5045 		 * this record overlaps the query range; follow pointer.
5046 		 */
5047 		if (xfs_btree_keycmp_lt(cur, high_key, lkp))
5048 			goto pop_up;
5049 		if (xfs_btree_keycmp_ge(cur, hkp, low_key)) {
5050 			level--;
5051 			error = xfs_btree_lookup_get_block(cur, level, pp,
5052 					&block);
5053 			if (error)
5054 				goto out;
5055 			xfs_btree_get_block(cur, level, &bp);
5056 			trace_xfs_btree_overlapped_query_range(cur, level, bp);
5057 #ifdef DEBUG
5058 			error = xfs_btree_check_block(cur, block, level, bp);
5059 			if (error)
5060 				goto out;
5061 #endif
5062 			cur->bc_levels[level].ptr = 1;
5063 			continue;
5064 		}
5065 		cur->bc_levels[level].ptr++;
5066 	}
5067 
5068 out:
5069 	/*
5070 	 * If we don't end this function with the cursor pointing at a record
5071 	 * block, a subsequent non-error cursor deletion will not release
5072 	 * node-level buffers, causing a buffer leak.  This is quite possible
5073 	 * with a zero-results range query, so release the buffers if we
5074 	 * failed to return any results.
5075 	 */
5076 	if (cur->bc_levels[0].bp == NULL) {
5077 		for (i = 0; i < cur->bc_nlevels; i++) {
5078 			if (cur->bc_levels[i].bp) {
5079 				xfs_trans_brelse(cur->bc_tp,
5080 						cur->bc_levels[i].bp);
5081 				cur->bc_levels[i].bp = NULL;
5082 				cur->bc_levels[i].ptr = 0;
5083 				cur->bc_levels[i].ra = 0;
5084 			}
5085 		}
5086 	}
5087 
5088 	return error;
5089 }
5090 
5091 static inline void
xfs_btree_key_from_irec(struct xfs_btree_cur * cur,union xfs_btree_key * key,const union xfs_btree_irec * irec)5092 xfs_btree_key_from_irec(
5093 	struct xfs_btree_cur		*cur,
5094 	union xfs_btree_key		*key,
5095 	const union xfs_btree_irec	*irec)
5096 {
5097 	union xfs_btree_rec		rec;
5098 
5099 	cur->bc_rec = *irec;
5100 	cur->bc_ops->init_rec_from_cur(cur, &rec);
5101 	cur->bc_ops->init_key_from_rec(key, &rec);
5102 }
5103 
5104 /*
5105  * Query a btree for all records overlapping a given interval of keys.  The
5106  * supplied function will be called with each record found; return one of the
5107  * XFS_BTREE_QUERY_RANGE_{CONTINUE,ABORT} values or the usual negative error
5108  * code.  This function returns -ECANCELED, zero, or a negative error code.
5109  */
5110 int
xfs_btree_query_range(struct xfs_btree_cur * cur,const union xfs_btree_irec * low_rec,const union xfs_btree_irec * high_rec,xfs_btree_query_range_fn fn,void * priv)5111 xfs_btree_query_range(
5112 	struct xfs_btree_cur		*cur,
5113 	const union xfs_btree_irec	*low_rec,
5114 	const union xfs_btree_irec	*high_rec,
5115 	xfs_btree_query_range_fn	fn,
5116 	void				*priv)
5117 {
5118 	union xfs_btree_key		low_key;
5119 	union xfs_btree_key		high_key;
5120 
5121 	/* Find the keys of both ends of the interval. */
5122 	xfs_btree_key_from_irec(cur, &high_key, high_rec);
5123 	xfs_btree_key_from_irec(cur, &low_key, low_rec);
5124 
5125 	/* Enforce low key <= high key. */
5126 	if (!xfs_btree_keycmp_le(cur, &low_key, &high_key))
5127 		return -EINVAL;
5128 
5129 	if (!(cur->bc_ops->geom_flags & XFS_BTGEO_OVERLAPPING))
5130 		return xfs_btree_simple_query_range(cur, &low_key,
5131 				&high_key, fn, priv);
5132 	return xfs_btree_overlapped_query_range(cur, &low_key, &high_key,
5133 			fn, priv);
5134 }
5135 
5136 /* Query a btree for all records. */
5137 int
xfs_btree_query_all(struct xfs_btree_cur * cur,xfs_btree_query_range_fn fn,void * priv)5138 xfs_btree_query_all(
5139 	struct xfs_btree_cur		*cur,
5140 	xfs_btree_query_range_fn	fn,
5141 	void				*priv)
5142 {
5143 	union xfs_btree_key		low_key;
5144 	union xfs_btree_key		high_key;
5145 
5146 	memset(&cur->bc_rec, 0, sizeof(cur->bc_rec));
5147 	memset(&low_key, 0, sizeof(low_key));
5148 	memset(&high_key, 0xFF, sizeof(high_key));
5149 
5150 	return xfs_btree_simple_query_range(cur, &low_key, &high_key, fn, priv);
5151 }
5152 
5153 static int
xfs_btree_count_blocks_helper(struct xfs_btree_cur * cur,int level,void * data)5154 xfs_btree_count_blocks_helper(
5155 	struct xfs_btree_cur	*cur,
5156 	int			level,
5157 	void			*data)
5158 {
5159 	xfs_extlen_t		*blocks = data;
5160 	(*blocks)++;
5161 
5162 	return 0;
5163 }
5164 
5165 /* Count the blocks in a btree and return the result in *blocks. */
5166 int
xfs_btree_count_blocks(struct xfs_btree_cur * cur,xfs_extlen_t * blocks)5167 xfs_btree_count_blocks(
5168 	struct xfs_btree_cur	*cur,
5169 	xfs_extlen_t		*blocks)
5170 {
5171 	*blocks = 0;
5172 	return xfs_btree_visit_blocks(cur, xfs_btree_count_blocks_helper,
5173 			XFS_BTREE_VISIT_ALL, blocks);
5174 }
5175 
5176 /* Compare two btree pointers. */
5177 int64_t
xfs_btree_diff_two_ptrs(struct xfs_btree_cur * cur,const union xfs_btree_ptr * a,const union xfs_btree_ptr * b)5178 xfs_btree_diff_two_ptrs(
5179 	struct xfs_btree_cur		*cur,
5180 	const union xfs_btree_ptr	*a,
5181 	const union xfs_btree_ptr	*b)
5182 {
5183 	if (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN)
5184 		return (int64_t)be64_to_cpu(a->l) - be64_to_cpu(b->l);
5185 	return (int64_t)be32_to_cpu(a->s) - be32_to_cpu(b->s);
5186 }
5187 
5188 struct xfs_btree_has_records {
5189 	/* Keys for the start and end of the range we want to know about. */
5190 	union xfs_btree_key		start_key;
5191 	union xfs_btree_key		end_key;
5192 
5193 	/* Mask for key comparisons, if desired. */
5194 	const union xfs_btree_key	*key_mask;
5195 
5196 	/* Highest record key we've seen so far. */
5197 	union xfs_btree_key		high_key;
5198 
5199 	enum xbtree_recpacking		outcome;
5200 };
5201 
5202 STATIC int
xfs_btree_has_records_helper(struct xfs_btree_cur * cur,const union xfs_btree_rec * rec,void * priv)5203 xfs_btree_has_records_helper(
5204 	struct xfs_btree_cur		*cur,
5205 	const union xfs_btree_rec	*rec,
5206 	void				*priv)
5207 {
5208 	union xfs_btree_key		rec_key;
5209 	union xfs_btree_key		rec_high_key;
5210 	struct xfs_btree_has_records	*info = priv;
5211 	enum xbtree_key_contig		key_contig;
5212 
5213 	cur->bc_ops->init_key_from_rec(&rec_key, rec);
5214 
5215 	if (info->outcome == XBTREE_RECPACKING_EMPTY) {
5216 		info->outcome = XBTREE_RECPACKING_SPARSE;
5217 
5218 		/*
5219 		 * If the first record we find does not overlap the start key,
5220 		 * then there is a hole at the start of the search range.
5221 		 * Classify this as sparse and stop immediately.
5222 		 */
5223 		if (xfs_btree_masked_keycmp_lt(cur, &info->start_key, &rec_key,
5224 					info->key_mask))
5225 			return -ECANCELED;
5226 	} else {
5227 		/*
5228 		 * If a subsequent record does not overlap with the any record
5229 		 * we've seen so far, there is a hole in the middle of the
5230 		 * search range.  Classify this as sparse and stop.
5231 		 * If the keys overlap and this btree does not allow overlap,
5232 		 * signal corruption.
5233 		 */
5234 		key_contig = cur->bc_ops->keys_contiguous(cur, &info->high_key,
5235 					&rec_key, info->key_mask);
5236 		if (key_contig == XBTREE_KEY_OVERLAP &&
5237 				!(cur->bc_ops->geom_flags & XFS_BTGEO_OVERLAPPING))
5238 			return -EFSCORRUPTED;
5239 		if (key_contig == XBTREE_KEY_GAP)
5240 			return -ECANCELED;
5241 	}
5242 
5243 	/*
5244 	 * If high_key(rec) is larger than any other high key we've seen,
5245 	 * remember it for later.
5246 	 */
5247 	cur->bc_ops->init_high_key_from_rec(&rec_high_key, rec);
5248 	if (xfs_btree_masked_keycmp_gt(cur, &rec_high_key, &info->high_key,
5249 				info->key_mask))
5250 		info->high_key = rec_high_key; /* struct copy */
5251 
5252 	return 0;
5253 }
5254 
5255 /*
5256  * Scan part of the keyspace of a btree and tell us if that keyspace does not
5257  * map to any records; is fully mapped to records; or is partially mapped to
5258  * records.  This is the btree record equivalent to determining if a file is
5259  * sparse.
5260  *
5261  * For most btree types, the record scan should use all available btree key
5262  * fields to compare the keys encountered.  These callers should pass NULL for
5263  * @mask.  However, some callers (e.g.  scanning physical space in the rmapbt)
5264  * want to ignore some part of the btree record keyspace when performing the
5265  * comparison.  These callers should pass in a union xfs_btree_key object with
5266  * the fields that *should* be a part of the comparison set to any nonzero
5267  * value, and the rest zeroed.
5268  */
5269 int
xfs_btree_has_records(struct xfs_btree_cur * cur,const union xfs_btree_irec * low,const union xfs_btree_irec * high,const union xfs_btree_key * mask,enum xbtree_recpacking * outcome)5270 xfs_btree_has_records(
5271 	struct xfs_btree_cur		*cur,
5272 	const union xfs_btree_irec	*low,
5273 	const union xfs_btree_irec	*high,
5274 	const union xfs_btree_key	*mask,
5275 	enum xbtree_recpacking		*outcome)
5276 {
5277 	struct xfs_btree_has_records	info = {
5278 		.outcome		= XBTREE_RECPACKING_EMPTY,
5279 		.key_mask		= mask,
5280 	};
5281 	int				error;
5282 
5283 	/* Not all btrees support this operation. */
5284 	if (!cur->bc_ops->keys_contiguous) {
5285 		ASSERT(0);
5286 		return -EOPNOTSUPP;
5287 	}
5288 
5289 	xfs_btree_key_from_irec(cur, &info.start_key, low);
5290 	xfs_btree_key_from_irec(cur, &info.end_key, high);
5291 
5292 	error = xfs_btree_query_range(cur, low, high,
5293 			xfs_btree_has_records_helper, &info);
5294 	if (error == -ECANCELED)
5295 		goto out;
5296 	if (error)
5297 		return error;
5298 
5299 	if (info.outcome == XBTREE_RECPACKING_EMPTY)
5300 		goto out;
5301 
5302 	/*
5303 	 * If the largest high_key(rec) we saw during the walk is greater than
5304 	 * the end of the search range, classify this as full.  Otherwise,
5305 	 * there is a hole at the end of the search range.
5306 	 */
5307 	if (xfs_btree_masked_keycmp_ge(cur, &info.high_key, &info.end_key,
5308 				mask))
5309 		info.outcome = XBTREE_RECPACKING_FULL;
5310 
5311 out:
5312 	*outcome = info.outcome;
5313 	return 0;
5314 }
5315 
5316 /* Are there more records in this btree? */
5317 bool
xfs_btree_has_more_records(struct xfs_btree_cur * cur)5318 xfs_btree_has_more_records(
5319 	struct xfs_btree_cur	*cur)
5320 {
5321 	struct xfs_btree_block	*block;
5322 	struct xfs_buf		*bp;
5323 
5324 	block = xfs_btree_get_block(cur, 0, &bp);
5325 
5326 	/* There are still records in this block. */
5327 	if (cur->bc_levels[0].ptr < xfs_btree_get_numrecs(block))
5328 		return true;
5329 
5330 	/* There are more record blocks. */
5331 	if (cur->bc_ops->ptr_len == XFS_BTREE_LONG_PTR_LEN)
5332 		return block->bb_u.l.bb_rightsib != cpu_to_be64(NULLFSBLOCK);
5333 	else
5334 		return block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK);
5335 }
5336 
5337 /* Set up all the btree cursor caches. */
5338 int __init
xfs_btree_init_cur_caches(void)5339 xfs_btree_init_cur_caches(void)
5340 {
5341 	int		error;
5342 
5343 	error = xfs_allocbt_init_cur_cache();
5344 	if (error)
5345 		return error;
5346 	error = xfs_inobt_init_cur_cache();
5347 	if (error)
5348 		goto err;
5349 	error = xfs_bmbt_init_cur_cache();
5350 	if (error)
5351 		goto err;
5352 	error = xfs_rmapbt_init_cur_cache();
5353 	if (error)
5354 		goto err;
5355 	error = xfs_refcountbt_init_cur_cache();
5356 	if (error)
5357 		goto err;
5358 
5359 	return 0;
5360 err:
5361 	xfs_btree_destroy_cur_caches();
5362 	return error;
5363 }
5364 
5365 /* Destroy all the btree cursor caches, if they've been allocated. */
5366 void
xfs_btree_destroy_cur_caches(void)5367 xfs_btree_destroy_cur_caches(void)
5368 {
5369 	xfs_allocbt_destroy_cur_cache();
5370 	xfs_inobt_destroy_cur_cache();
5371 	xfs_bmbt_destroy_cur_cache();
5372 	xfs_rmapbt_destroy_cur_cache();
5373 	xfs_refcountbt_destroy_cur_cache();
5374 }
5375 
5376 /* Move the btree cursor before the first record. */
5377 int
xfs_btree_goto_left_edge(struct xfs_btree_cur * cur)5378 xfs_btree_goto_left_edge(
5379 	struct xfs_btree_cur	*cur)
5380 {
5381 	int			stat = 0;
5382 	int			error;
5383 
5384 	memset(&cur->bc_rec, 0, sizeof(cur->bc_rec));
5385 	error = xfs_btree_lookup(cur, XFS_LOOKUP_LE, &stat);
5386 	if (error)
5387 		return error;
5388 	if (!stat)
5389 		return 0;
5390 
5391 	error = xfs_btree_decrement(cur, 0, &stat);
5392 	if (error)
5393 		return error;
5394 	if (stat != 0) {
5395 		ASSERT(0);
5396 		xfs_btree_mark_sick(cur);
5397 		return -EFSCORRUPTED;
5398 	}
5399 
5400 	return 0;
5401 }
5402