1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
4 * Copyright (c) 2013 Red Hat, Inc.
5 * All Rights Reserved.
6 */
7 #include "xfs.h"
8 #include "xfs_fs.h"
9 #include "xfs_shared.h"
10 #include "xfs_format.h"
11 #include "xfs_log_format.h"
12 #include "xfs_trans_resv.h"
13 #include "xfs_bit.h"
14 #include "xfs_mount.h"
15 #include "xfs_inode.h"
16 #include "xfs_dir2.h"
17 #include "xfs_dir2_priv.h"
18 #include "xfs_trans.h"
19 #include "xfs_bmap.h"
20 #include "xfs_attr_leaf.h"
21 #include "xfs_error.h"
22 #include "xfs_trace.h"
23 #include "xfs_buf_item.h"
24 #include "xfs_log.h"
25 #include "xfs_errortag.h"
26 #include "xfs_health.h"
27
28 /*
29 * xfs_da_btree.c
30 *
31 * Routines to implement directories as Btrees of hashed names.
32 */
33
34 /*========================================================================
35 * Function prototypes for the kernel.
36 *========================================================================*/
37
38 /*
39 * Routines used for growing the Btree.
40 */
41 STATIC int xfs_da3_root_split(xfs_da_state_t *state,
42 xfs_da_state_blk_t *existing_root,
43 xfs_da_state_blk_t *new_child);
44 STATIC int xfs_da3_node_split(xfs_da_state_t *state,
45 xfs_da_state_blk_t *existing_blk,
46 xfs_da_state_blk_t *split_blk,
47 xfs_da_state_blk_t *blk_to_add,
48 int treelevel,
49 int *result);
50 STATIC void xfs_da3_node_rebalance(xfs_da_state_t *state,
51 xfs_da_state_blk_t *node_blk_1,
52 xfs_da_state_blk_t *node_blk_2);
53 STATIC void xfs_da3_node_add(xfs_da_state_t *state,
54 xfs_da_state_blk_t *old_node_blk,
55 xfs_da_state_blk_t *new_node_blk);
56
57 /*
58 * Routines used for shrinking the Btree.
59 */
60 STATIC int xfs_da3_root_join(xfs_da_state_t *state,
61 xfs_da_state_blk_t *root_blk);
62 STATIC int xfs_da3_node_toosmall(xfs_da_state_t *state, int *retval);
63 STATIC void xfs_da3_node_remove(xfs_da_state_t *state,
64 xfs_da_state_blk_t *drop_blk);
65 STATIC void xfs_da3_node_unbalance(xfs_da_state_t *state,
66 xfs_da_state_blk_t *src_node_blk,
67 xfs_da_state_blk_t *dst_node_blk);
68
69 /*
70 * Utility routines.
71 */
72 STATIC int xfs_da3_blk_unlink(xfs_da_state_t *state,
73 xfs_da_state_blk_t *drop_blk,
74 xfs_da_state_blk_t *save_blk);
75
76
77 struct kmem_cache *xfs_da_state_cache; /* anchor for dir/attr state */
78
79 /*
80 * Allocate a dir-state structure.
81 * We don't put them on the stack since they're large.
82 */
83 struct xfs_da_state *
84 xfs_da_state_alloc(
85 struct xfs_da_args *args)
86 {
87 struct xfs_da_state *state;
88
89 state = kmem_cache_zalloc(xfs_da_state_cache,
90 GFP_KERNEL | __GFP_NOLOCKDEP | __GFP_NOFAIL);
91 state->args = args;
92 state->mp = args->dp->i_mount;
93 return state;
94 }
95
96 /*
97 * Kill the altpath contents of a da-state structure.
98 */
99 STATIC void
100 xfs_da_state_kill_altpath(xfs_da_state_t *state)
101 {
102 int i;
103
104 for (i = 0; i < state->altpath.active; i++)
105 state->altpath.blk[i].bp = NULL;
106 state->altpath.active = 0;
107 }
108
109 /*
110 * Free a da-state structure.
111 */
112 void
113 xfs_da_state_free(xfs_da_state_t *state)
114 {
115 xfs_da_state_kill_altpath(state);
116 #ifdef DEBUG
117 memset((char *)state, 0, sizeof(*state));
118 #endif /* DEBUG */
119 kmem_cache_free(xfs_da_state_cache, state);
120 }
121
122 void
123 xfs_da_state_reset(
124 struct xfs_da_state *state,
125 struct xfs_da_args *args)
126 {
127 xfs_da_state_kill_altpath(state);
128 memset(state, 0, sizeof(struct xfs_da_state));
129 state->args = args;
130 state->mp = state->args->dp->i_mount;
131 }
132
133 static inline int xfs_dabuf_nfsb(struct xfs_mount *mp, int whichfork)
134 {
135 if (whichfork == XFS_DATA_FORK)
136 return mp->m_dir_geo->fsbcount;
137 return mp->m_attr_geo->fsbcount;
138 }
139
140 void
141 xfs_da3_node_hdr_from_disk(
142 struct xfs_mount *mp,
143 struct xfs_da3_icnode_hdr *to,
144 struct xfs_da_intnode *from)
145 {
146 if (xfs_has_crc(mp)) {
147 struct xfs_da3_intnode *from3 = (struct xfs_da3_intnode *)from;
148
149 to->forw = be32_to_cpu(from3->hdr.info.hdr.forw);
150 to->back = be32_to_cpu(from3->hdr.info.hdr.back);
151 to->magic = be16_to_cpu(from3->hdr.info.hdr.magic);
152 to->count = be16_to_cpu(from3->hdr.__count);
153 to->level = be16_to_cpu(from3->hdr.__level);
154 to->btree = from3->__btree;
155 ASSERT(to->magic == XFS_DA3_NODE_MAGIC);
156 } else {
157 to->forw = be32_to_cpu(from->hdr.info.forw);
158 to->back = be32_to_cpu(from->hdr.info.back);
159 to->magic = be16_to_cpu(from->hdr.info.magic);
160 to->count = be16_to_cpu(from->hdr.__count);
161 to->level = be16_to_cpu(from->hdr.__level);
162 to->btree = from->__btree;
163 ASSERT(to->magic == XFS_DA_NODE_MAGIC);
164 }
165 }
166
167 void
168 xfs_da3_node_hdr_to_disk(
169 struct xfs_mount *mp,
170 struct xfs_da_intnode *to,
171 struct xfs_da3_icnode_hdr *from)
172 {
173 if (xfs_has_crc(mp)) {
174 struct xfs_da3_intnode *to3 = (struct xfs_da3_intnode *)to;
175
176 ASSERT(from->magic == XFS_DA3_NODE_MAGIC);
177 to3->hdr.info.hdr.forw = cpu_to_be32(from->forw);
178 to3->hdr.info.hdr.back = cpu_to_be32(from->back);
179 to3->hdr.info.hdr.magic = cpu_to_be16(from->magic);
180 to3->hdr.__count = cpu_to_be16(from->count);
181 to3->hdr.__level = cpu_to_be16(from->level);
182 } else {
183 ASSERT(from->magic == XFS_DA_NODE_MAGIC);
184 to->hdr.info.forw = cpu_to_be32(from->forw);
185 to->hdr.info.back = cpu_to_be32(from->back);
186 to->hdr.info.magic = cpu_to_be16(from->magic);
187 to->hdr.__count = cpu_to_be16(from->count);
188 to->hdr.__level = cpu_to_be16(from->level);
189 }
190 }
191
192 /*
193 * Verify an xfs_da3_blkinfo structure. Note that the da3 fields are only
194 * accessible on v5 filesystems. This header format is common across da node,
195 * attr leaf and dir leaf blocks.
196 */
197 xfs_failaddr_t
198 xfs_da3_blkinfo_verify(
199 struct xfs_buf *bp,
200 struct xfs_da3_blkinfo *hdr3)
201 {
202 struct xfs_mount *mp = bp->b_mount;
203 struct xfs_da_blkinfo *hdr = &hdr3->hdr;
204
205 if (!xfs_verify_magic16(bp, hdr->magic))
206 return __this_address;
207
208 if (xfs_has_crc(mp)) {
209 if (!uuid_equal(&hdr3->uuid, &mp->m_sb.sb_meta_uuid))
210 return __this_address;
211 if (be64_to_cpu(hdr3->blkno) != xfs_buf_daddr(bp))
212 return __this_address;
213 if (!xfs_log_check_lsn(mp, be64_to_cpu(hdr3->lsn)))
214 return __this_address;
215 }
216
217 return NULL;
218 }
219
220 static xfs_failaddr_t
221 xfs_da3_node_verify(
222 struct xfs_buf *bp)
223 {
224 struct xfs_mount *mp = bp->b_mount;
225 struct xfs_da_intnode *hdr = bp->b_addr;
226 struct xfs_da3_icnode_hdr ichdr;
227 xfs_failaddr_t fa;
228
229 xfs_da3_node_hdr_from_disk(mp, &ichdr, hdr);
230
231 fa = xfs_da3_blkinfo_verify(bp, bp->b_addr);
232 if (fa)
233 return fa;
234
235 if (ichdr.level == 0)
236 return __this_address;
237 if (ichdr.level > XFS_DA_NODE_MAXDEPTH)
238 return __this_address;
239 if (ichdr.count == 0)
240 return __this_address;
241
242 /*
243 * we don't know if the node is for and attribute or directory tree,
244 * so only fail if the count is outside both bounds
245 */
246 if (ichdr.count > mp->m_dir_geo->node_ents &&
247 ichdr.count > mp->m_attr_geo->node_ents)
248 return __this_address;
249
250 /* XXX: hash order check? */
251
252 return NULL;
253 }
254
255 xfs_failaddr_t
256 xfs_da3_node_header_check(
257 struct xfs_buf *bp,
258 xfs_ino_t owner)
259 {
260 struct xfs_mount *mp = bp->b_mount;
261
262 if (xfs_has_crc(mp)) {
263 struct xfs_da3_blkinfo *hdr3 = bp->b_addr;
264
265 if (hdr3->hdr.magic != cpu_to_be16(XFS_DA3_NODE_MAGIC))
266 return __this_address;
267
268 if (be64_to_cpu(hdr3->owner) != owner)
269 return __this_address;
270 }
271
272 return NULL;
273 }
274
275 xfs_failaddr_t
276 xfs_da3_header_check(
277 struct xfs_buf *bp,
278 xfs_ino_t owner)
279 {
280 struct xfs_mount *mp = bp->b_mount;
281 struct xfs_da_blkinfo *hdr = bp->b_addr;
282
283 if (!xfs_has_crc(mp))
284 return NULL;
285
286 switch (hdr->magic) {
287 case cpu_to_be16(XFS_ATTR3_LEAF_MAGIC):
288 return xfs_attr3_leaf_header_check(bp, owner);
289 case cpu_to_be16(XFS_DA3_NODE_MAGIC):
290 return xfs_da3_node_header_check(bp, owner);
291 case cpu_to_be16(XFS_DIR3_LEAF1_MAGIC):
292 case cpu_to_be16(XFS_DIR3_LEAFN_MAGIC):
293 return xfs_dir3_leaf_header_check(bp, owner);
294 }
295
296 ASSERT(0);
297 return NULL;
298 }
299
300 static void
301 xfs_da3_node_write_verify(
302 struct xfs_buf *bp)
303 {
304 struct xfs_mount *mp = bp->b_mount;
305 struct xfs_buf_log_item *bip = bp->b_log_item;
306 struct xfs_da3_node_hdr *hdr3 = bp->b_addr;
307 xfs_failaddr_t fa;
308
309 fa = xfs_da3_node_verify(bp);
310 if (fa) {
311 xfs_verifier_error(bp, -EFSCORRUPTED, fa);
312 return;
313 }
314
315 if (!xfs_has_crc(mp))
316 return;
317
318 if (bip)
319 hdr3->info.lsn = cpu_to_be64(bip->bli_item.li_lsn);
320
321 xfs_buf_update_cksum(bp, XFS_DA3_NODE_CRC_OFF);
322 }
323
324 /*
325 * leaf/node format detection on trees is sketchy, so a node read can be done on
326 * leaf level blocks when detection identifies the tree as a node format tree
327 * incorrectly. In this case, we need to swap the verifier to match the correct
328 * format of the block being read.
329 */
330 static void
331 xfs_da3_node_read_verify(
332 struct xfs_buf *bp)
333 {
334 struct xfs_da_blkinfo *info = bp->b_addr;
335 xfs_failaddr_t fa;
336
337 switch (be16_to_cpu(info->magic)) {
338 case XFS_DA3_NODE_MAGIC:
339 if (!xfs_buf_verify_cksum(bp, XFS_DA3_NODE_CRC_OFF)) {
340 xfs_verifier_error(bp, -EFSBADCRC,
341 __this_address);
342 break;
343 }
344 fallthrough;
345 case XFS_DA_NODE_MAGIC:
346 fa = xfs_da3_node_verify(bp);
347 if (fa)
348 xfs_verifier_error(bp, -EFSCORRUPTED, fa);
349 return;
350 case XFS_ATTR_LEAF_MAGIC:
351 case XFS_ATTR3_LEAF_MAGIC:
352 bp->b_ops = &xfs_attr3_leaf_buf_ops;
353 bp->b_ops->verify_read(bp);
354 return;
355 case XFS_DIR2_LEAFN_MAGIC:
356 case XFS_DIR3_LEAFN_MAGIC:
357 bp->b_ops = &xfs_dir3_leafn_buf_ops;
358 bp->b_ops->verify_read(bp);
359 return;
360 default:
361 xfs_verifier_error(bp, -EFSCORRUPTED, __this_address);
362 break;
363 }
364 }
365
366 /* Verify the structure of a da3 block. */
367 static xfs_failaddr_t
368 xfs_da3_node_verify_struct(
369 struct xfs_buf *bp)
370 {
371 struct xfs_da_blkinfo *info = bp->b_addr;
372
373 switch (be16_to_cpu(info->magic)) {
374 case XFS_DA3_NODE_MAGIC:
375 case XFS_DA_NODE_MAGIC:
376 return xfs_da3_node_verify(bp);
377 case XFS_ATTR_LEAF_MAGIC:
378 case XFS_ATTR3_LEAF_MAGIC:
379 bp->b_ops = &xfs_attr3_leaf_buf_ops;
380 return bp->b_ops->verify_struct(bp);
381 case XFS_DIR2_LEAFN_MAGIC:
382 case XFS_DIR3_LEAFN_MAGIC:
383 bp->b_ops = &xfs_dir3_leafn_buf_ops;
384 return bp->b_ops->verify_struct(bp);
385 default:
386 return __this_address;
387 }
388 }
389
390 const struct xfs_buf_ops xfs_da3_node_buf_ops = {
391 .name = "xfs_da3_node",
392 .magic16 = { cpu_to_be16(XFS_DA_NODE_MAGIC),
393 cpu_to_be16(XFS_DA3_NODE_MAGIC) },
394 .verify_read = xfs_da3_node_read_verify,
395 .verify_write = xfs_da3_node_write_verify,
396 .verify_struct = xfs_da3_node_verify_struct,
397 };
398
399 static int
400 xfs_da3_node_set_type(
401 struct xfs_trans *tp,
402 struct xfs_inode *dp,
403 int whichfork,
404 struct xfs_buf *bp)
405 {
406 struct xfs_da_blkinfo *info = bp->b_addr;
407
408 switch (be16_to_cpu(info->magic)) {
409 case XFS_DA_NODE_MAGIC:
410 case XFS_DA3_NODE_MAGIC:
411 xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DA_NODE_BUF);
412 return 0;
413 case XFS_ATTR_LEAF_MAGIC:
414 case XFS_ATTR3_LEAF_MAGIC:
415 xfs_trans_buf_set_type(tp, bp, XFS_BLFT_ATTR_LEAF_BUF);
416 return 0;
417 case XFS_DIR2_LEAFN_MAGIC:
418 case XFS_DIR3_LEAFN_MAGIC:
419 xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_LEAFN_BUF);
420 return 0;
421 default:
422 XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, tp->t_mountp,
423 info, sizeof(*info));
424 xfs_trans_brelse(tp, bp);
425 xfs_dirattr_mark_sick(dp, whichfork);
426 return -EFSCORRUPTED;
427 }
428 }
429
430 int
431 xfs_da3_node_read(
432 struct xfs_trans *tp,
433 struct xfs_inode *dp,
434 xfs_dablk_t bno,
435 struct xfs_buf **bpp,
436 int whichfork)
437 {
438 int error;
439
440 error = xfs_da_read_buf(tp, dp, bno, 0, bpp, whichfork,
441 &xfs_da3_node_buf_ops);
442 if (error || !*bpp || !tp)
443 return error;
444 return xfs_da3_node_set_type(tp, dp, whichfork, *bpp);
445 }
446
447 int
448 xfs_da3_node_read_mapped(
449 struct xfs_trans *tp,
450 struct xfs_inode *dp,
451 xfs_daddr_t mappedbno,
452 struct xfs_buf **bpp,
453 int whichfork)
454 {
455 struct xfs_mount *mp = dp->i_mount;
456 int error;
457
458 error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, mappedbno,
459 XFS_FSB_TO_BB(mp, xfs_dabuf_nfsb(mp, whichfork)), 0,
460 bpp, &xfs_da3_node_buf_ops);
461 if (xfs_metadata_is_sick(error))
462 xfs_dirattr_mark_sick(dp, whichfork);
463 if (error || !*bpp)
464 return error;
465
466 if (whichfork == XFS_ATTR_FORK)
467 xfs_buf_set_ref(*bpp, XFS_ATTR_BTREE_REF);
468 else
469 xfs_buf_set_ref(*bpp, XFS_DIR_BTREE_REF);
470
471 if (!tp)
472 return 0;
473 return xfs_da3_node_set_type(tp, dp, whichfork, *bpp);
474 }
475
476 /*
477 * Copy src directory/attr leaf/node buffer to the dst.
478 * For v5 file systems make sure the right blkno is stamped in.
479 */
480 void
481 xfs_da_buf_copy(
482 struct xfs_buf *dst,
483 struct xfs_buf *src,
484 size_t size)
485 {
486 struct xfs_da3_blkinfo *da3 = dst->b_addr;
487
488 memcpy(dst->b_addr, src->b_addr, size);
489 dst->b_ops = src->b_ops;
490 xfs_trans_buf_copy_type(dst, src);
491 if (xfs_has_crc(dst->b_mount))
492 da3->blkno = cpu_to_be64(xfs_buf_daddr(dst));
493 }
494
495 /*========================================================================
496 * Routines used for growing the Btree.
497 *========================================================================*/
498
499 /*
500 * Create the initial contents of an intermediate node.
501 */
502 int
503 xfs_da3_node_create(
504 struct xfs_da_args *args,
505 xfs_dablk_t blkno,
506 int level,
507 struct xfs_buf **bpp,
508 int whichfork)
509 {
510 struct xfs_da_intnode *node;
511 struct xfs_trans *tp = args->trans;
512 struct xfs_mount *mp = tp->t_mountp;
513 struct xfs_da3_icnode_hdr ichdr = {0};
514 struct xfs_buf *bp;
515 int error;
516 struct xfs_inode *dp = args->dp;
517
518 trace_xfs_da_node_create(args);
519 ASSERT(level <= XFS_DA_NODE_MAXDEPTH);
520
521 error = xfs_da_get_buf(tp, dp, blkno, &bp, whichfork);
522 if (error)
523 return error;
524 bp->b_ops = &xfs_da3_node_buf_ops;
525 xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DA_NODE_BUF);
526 node = bp->b_addr;
527
528 if (xfs_has_crc(mp)) {
529 struct xfs_da3_node_hdr *hdr3 = bp->b_addr;
530
531 memset(hdr3, 0, sizeof(struct xfs_da3_node_hdr));
532 ichdr.magic = XFS_DA3_NODE_MAGIC;
533 hdr3->info.blkno = cpu_to_be64(xfs_buf_daddr(bp));
534 hdr3->info.owner = cpu_to_be64(args->owner);
535 uuid_copy(&hdr3->info.uuid, &mp->m_sb.sb_meta_uuid);
536 } else {
537 ichdr.magic = XFS_DA_NODE_MAGIC;
538 }
539 ichdr.level = level;
540
541 xfs_da3_node_hdr_to_disk(dp->i_mount, node, &ichdr);
542 xfs_trans_log_buf(tp, bp,
543 XFS_DA_LOGRANGE(node, &node->hdr, args->geo->node_hdr_size));
544
545 *bpp = bp;
546 return 0;
547 }
548
549 /*
550 * Split a leaf node, rebalance, then possibly split
551 * intermediate nodes, rebalance, etc.
552 */
553 int /* error */
554 xfs_da3_split(
555 struct xfs_da_state *state)
556 {
557 struct xfs_da_state_blk *oldblk;
558 struct xfs_da_state_blk *newblk;
559 struct xfs_da_state_blk *addblk;
560 struct xfs_da_intnode *node;
561 int max;
562 int action = 0;
563 int error;
564 int i;
565
566 trace_xfs_da_split(state->args);
567
568 if (XFS_TEST_ERROR(false, state->mp, XFS_ERRTAG_DA_LEAF_SPLIT))
569 return -EIO;
570
571 /*
572 * Walk back up the tree splitting/inserting/adjusting as necessary.
573 * If we need to insert and there isn't room, split the node, then
574 * decide which fragment to insert the new block from below into.
575 * Note that we may split the root this way, but we need more fixup.
576 */
577 max = state->path.active - 1;
578 ASSERT((max >= 0) && (max < XFS_DA_NODE_MAXDEPTH));
579 ASSERT(state->path.blk[max].magic == XFS_ATTR_LEAF_MAGIC ||
580 state->path.blk[max].magic == XFS_DIR2_LEAFN_MAGIC);
581
582 addblk = &state->path.blk[max]; /* initial dummy value */
583 for (i = max; (i >= 0) && addblk; state->path.active--, i--) {
584 oldblk = &state->path.blk[i];
585 newblk = &state->altpath.blk[i];
586
587 /*
588 * If a leaf node then
589 * Allocate a new leaf node, then rebalance across them.
590 * else if an intermediate node then
591 * We split on the last layer, must we split the node?
592 */
593 switch (oldblk->magic) {
594 case XFS_ATTR_LEAF_MAGIC:
595 error = xfs_attr3_leaf_split(state, oldblk, newblk);
596 if ((error != 0) && (error != -ENOSPC)) {
597 return error; /* GROT: attr is inconsistent */
598 }
599 if (!error) {
600 addblk = newblk;
601 break;
602 }
603 /*
604 * Entry wouldn't fit, split the leaf again. The new
605 * extrablk will be consumed by xfs_da3_node_split if
606 * the node is split.
607 */
608 state->extravalid = 1;
609 if (state->inleaf) {
610 state->extraafter = 0; /* before newblk */
611 trace_xfs_attr_leaf_split_before(state->args);
612 error = xfs_attr3_leaf_split(state, oldblk,
613 &state->extrablk);
614 } else {
615 state->extraafter = 1; /* after newblk */
616 trace_xfs_attr_leaf_split_after(state->args);
617 error = xfs_attr3_leaf_split(state, newblk,
618 &state->extrablk);
619 }
620 if (error)
621 return error; /* GROT: attr inconsistent */
622 addblk = newblk;
623 break;
624 case XFS_DIR2_LEAFN_MAGIC:
625 error = xfs_dir2_leafn_split(state, oldblk, newblk);
626 if (error)
627 return error;
628 addblk = newblk;
629 break;
630 case XFS_DA_NODE_MAGIC:
631 error = xfs_da3_node_split(state, oldblk, newblk, addblk,
632 max - i, &action);
633 addblk->bp = NULL;
634 if (error)
635 return error; /* GROT: dir is inconsistent */
636 /*
637 * Record the newly split block for the next time thru?
638 */
639 if (action)
640 addblk = newblk;
641 else
642 addblk = NULL;
643 break;
644 }
645
646 /*
647 * Update the btree to show the new hashval for this child.
648 */
649 xfs_da3_fixhashpath(state, &state->path);
650 }
651 if (!addblk)
652 return 0;
653
654 /*
655 * xfs_da3_node_split() should have consumed any extra blocks we added
656 * during a double leaf split in the attr fork. This is guaranteed as
657 * we can't be here if the attr fork only has a single leaf block.
658 */
659 ASSERT(state->extravalid == 0 ||
660 state->path.blk[max].magic == XFS_DIR2_LEAFN_MAGIC);
661
662 /*
663 * Split the root node.
664 */
665 ASSERT(state->path.active == 0);
666 oldblk = &state->path.blk[0];
667 error = xfs_da3_root_split(state, oldblk, addblk);
668 if (error)
669 goto out;
670
671 /*
672 * Update pointers to the node which used to be block 0 and just got
673 * bumped because of the addition of a new root node. Note that the
674 * original block 0 could be at any position in the list of blocks in
675 * the tree.
676 *
677 * Note: the magic numbers and sibling pointers are in the same physical
678 * place for both v2 and v3 headers (by design). Hence it doesn't matter
679 * which version of the xfs_da_intnode structure we use here as the
680 * result will be the same using either structure.
681 */
682 node = oldblk->bp->b_addr;
683 if (node->hdr.info.forw) {
684 if (be32_to_cpu(node->hdr.info.forw) != addblk->blkno) {
685 xfs_buf_mark_corrupt(oldblk->bp);
686 xfs_da_mark_sick(state->args);
687 error = -EFSCORRUPTED;
688 goto out;
689 }
690 node = addblk->bp->b_addr;
691 node->hdr.info.back = cpu_to_be32(oldblk->blkno);
692 xfs_trans_log_buf(state->args->trans, addblk->bp,
693 XFS_DA_LOGRANGE(node, &node->hdr.info,
694 sizeof(node->hdr.info)));
695 }
696 node = oldblk->bp->b_addr;
697 if (node->hdr.info.back) {
698 if (be32_to_cpu(node->hdr.info.back) != addblk->blkno) {
699 xfs_buf_mark_corrupt(oldblk->bp);
700 xfs_da_mark_sick(state->args);
701 error = -EFSCORRUPTED;
702 goto out;
703 }
704 node = addblk->bp->b_addr;
705 node->hdr.info.forw = cpu_to_be32(oldblk->blkno);
706 xfs_trans_log_buf(state->args->trans, addblk->bp,
707 XFS_DA_LOGRANGE(node, &node->hdr.info,
708 sizeof(node->hdr.info)));
709 }
710 out:
711 addblk->bp = NULL;
712 return error;
713 }
714
715 /*
716 * Split the root. We have to create a new root and point to the two
717 * parts (the split old root) that we just created. Copy block zero to
718 * the EOF, extending the inode in process.
719 */
720 STATIC int /* error */
721 xfs_da3_root_split(
722 struct xfs_da_state *state,
723 struct xfs_da_state_blk *blk1,
724 struct xfs_da_state_blk *blk2)
725 {
726 struct xfs_da_intnode *node;
727 struct xfs_da_intnode *oldroot;
728 struct xfs_da_node_entry *btree;
729 struct xfs_da3_icnode_hdr nodehdr;
730 struct xfs_da_args *args;
731 struct xfs_buf *bp;
732 struct xfs_inode *dp;
733 struct xfs_trans *tp;
734 struct xfs_dir2_leaf *leaf;
735 xfs_dablk_t blkno;
736 int level;
737 int error;
738 int size;
739
740 trace_xfs_da_root_split(state->args);
741
742 /*
743 * Copy the existing (incorrect) block from the root node position
744 * to a free space somewhere.
745 */
746 args = state->args;
747 error = xfs_da_grow_inode(args, &blkno);
748 if (error)
749 return error;
750
751 dp = args->dp;
752 tp = args->trans;
753 error = xfs_da_get_buf(tp, dp, blkno, &bp, args->whichfork);
754 if (error)
755 return error;
756 node = bp->b_addr;
757 oldroot = blk1->bp->b_addr;
758 if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
759 oldroot->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC)) {
760 struct xfs_da3_icnode_hdr icnodehdr;
761
762 xfs_da3_node_hdr_from_disk(dp->i_mount, &icnodehdr, oldroot);
763 btree = icnodehdr.btree;
764 size = (int)((char *)&btree[icnodehdr.count] - (char *)oldroot);
765 level = icnodehdr.level;
766 } else {
767 struct xfs_dir3_icleaf_hdr leafhdr;
768
769 leaf = (xfs_dir2_leaf_t *)oldroot;
770 xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &leafhdr, leaf);
771
772 ASSERT(leafhdr.magic == XFS_DIR2_LEAFN_MAGIC ||
773 leafhdr.magic == XFS_DIR3_LEAFN_MAGIC);
774 size = (int)((char *)&leafhdr.ents[leafhdr.count] -
775 (char *)leaf);
776 level = 0;
777 }
778
779 /*
780 * Copy old root to new buffer and log it.
781 */
782 xfs_da_buf_copy(bp, blk1->bp, size);
783 xfs_trans_log_buf(tp, bp, 0, size - 1);
784
785 /*
786 * Update blk1 to point to new buffer.
787 */
788 blk1->bp = bp;
789 blk1->blkno = blkno;
790
791 /*
792 * Set up the new root node.
793 */
794 error = xfs_da3_node_create(args,
795 (args->whichfork == XFS_DATA_FORK) ? args->geo->leafblk : 0,
796 level + 1, &bp, args->whichfork);
797 if (error)
798 return error;
799
800 node = bp->b_addr;
801 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
802 btree = nodehdr.btree;
803 btree[0].hashval = cpu_to_be32(blk1->hashval);
804 btree[0].before = cpu_to_be32(blk1->blkno);
805 btree[1].hashval = cpu_to_be32(blk2->hashval);
806 btree[1].before = cpu_to_be32(blk2->blkno);
807 nodehdr.count = 2;
808 xfs_da3_node_hdr_to_disk(dp->i_mount, node, &nodehdr);
809
810 #ifdef DEBUG
811 if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
812 oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC)) {
813 ASSERT(blk1->blkno >= args->geo->leafblk &&
814 blk1->blkno < args->geo->freeblk);
815 ASSERT(blk2->blkno >= args->geo->leafblk &&
816 blk2->blkno < args->geo->freeblk);
817 }
818 #endif
819
820 /* Header is already logged by xfs_da_node_create */
821 xfs_trans_log_buf(tp, bp,
822 XFS_DA_LOGRANGE(node, btree, sizeof(xfs_da_node_entry_t) * 2));
823
824 return 0;
825 }
826
827 /*
828 * Split the node, rebalance, then add the new entry.
829 */
830 STATIC int /* error */
831 xfs_da3_node_split(
832 struct xfs_da_state *state,
833 struct xfs_da_state_blk *oldblk,
834 struct xfs_da_state_blk *newblk,
835 struct xfs_da_state_blk *addblk,
836 int treelevel,
837 int *result)
838 {
839 struct xfs_da_intnode *node;
840 struct xfs_da3_icnode_hdr nodehdr;
841 xfs_dablk_t blkno;
842 int newcount;
843 int error;
844 int useextra;
845 struct xfs_inode *dp = state->args->dp;
846
847 trace_xfs_da_node_split(state->args);
848
849 node = oldblk->bp->b_addr;
850 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
851
852 /*
853 * With V2 dirs the extra block is data or freespace.
854 */
855 useextra = state->extravalid && state->args->whichfork == XFS_ATTR_FORK;
856 newcount = 1 + useextra;
857 /*
858 * Do we have to split the node?
859 */
860 if (nodehdr.count + newcount > state->args->geo->node_ents) {
861 /*
862 * Allocate a new node, add to the doubly linked chain of
863 * nodes, then move some of our excess entries into it.
864 */
865 error = xfs_da_grow_inode(state->args, &blkno);
866 if (error)
867 return error; /* GROT: dir is inconsistent */
868
869 error = xfs_da3_node_create(state->args, blkno, treelevel,
870 &newblk->bp, state->args->whichfork);
871 if (error)
872 return error; /* GROT: dir is inconsistent */
873 newblk->blkno = blkno;
874 newblk->magic = XFS_DA_NODE_MAGIC;
875 xfs_da3_node_rebalance(state, oldblk, newblk);
876 error = xfs_da3_blk_link(state, oldblk, newblk);
877 if (error)
878 return error;
879 *result = 1;
880 } else {
881 *result = 0;
882 }
883
884 /*
885 * Insert the new entry(s) into the correct block
886 * (updating last hashval in the process).
887 *
888 * xfs_da3_node_add() inserts BEFORE the given index,
889 * and as a result of using node_lookup_int() we always
890 * point to a valid entry (not after one), but a split
891 * operation always results in a new block whose hashvals
892 * FOLLOW the current block.
893 *
894 * If we had double-split op below us, then add the extra block too.
895 */
896 node = oldblk->bp->b_addr;
897 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
898 if (oldblk->index <= nodehdr.count) {
899 oldblk->index++;
900 xfs_da3_node_add(state, oldblk, addblk);
901 if (useextra) {
902 if (state->extraafter)
903 oldblk->index++;
904 xfs_da3_node_add(state, oldblk, &state->extrablk);
905 state->extravalid = 0;
906 }
907 } else {
908 newblk->index++;
909 xfs_da3_node_add(state, newblk, addblk);
910 if (useextra) {
911 if (state->extraafter)
912 newblk->index++;
913 xfs_da3_node_add(state, newblk, &state->extrablk);
914 state->extravalid = 0;
915 }
916 }
917
918 return 0;
919 }
920
921 /*
922 * Balance the btree elements between two intermediate nodes,
923 * usually one full and one empty.
924 *
925 * NOTE: if blk2 is empty, then it will get the upper half of blk1.
926 */
927 STATIC void
928 xfs_da3_node_rebalance(
929 struct xfs_da_state *state,
930 struct xfs_da_state_blk *blk1,
931 struct xfs_da_state_blk *blk2)
932 {
933 struct xfs_da_intnode *node1;
934 struct xfs_da_intnode *node2;
935 struct xfs_da_node_entry *btree1;
936 struct xfs_da_node_entry *btree2;
937 struct xfs_da_node_entry *btree_s;
938 struct xfs_da_node_entry *btree_d;
939 struct xfs_da3_icnode_hdr nodehdr1;
940 struct xfs_da3_icnode_hdr nodehdr2;
941 struct xfs_trans *tp;
942 int count;
943 int tmp;
944 int swap = 0;
945 struct xfs_inode *dp = state->args->dp;
946
947 trace_xfs_da_node_rebalance(state->args);
948
949 node1 = blk1->bp->b_addr;
950 node2 = blk2->bp->b_addr;
951 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr1, node1);
952 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr2, node2);
953 btree1 = nodehdr1.btree;
954 btree2 = nodehdr2.btree;
955
956 /*
957 * Figure out how many entries need to move, and in which direction.
958 * Swap the nodes around if that makes it simpler.
959 */
960 if (nodehdr1.count > 0 && nodehdr2.count > 0 &&
961 ((be32_to_cpu(btree2[0].hashval) < be32_to_cpu(btree1[0].hashval)) ||
962 (be32_to_cpu(btree2[nodehdr2.count - 1].hashval) <
963 be32_to_cpu(btree1[nodehdr1.count - 1].hashval)))) {
964 swap(node1, node2);
965 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr1, node1);
966 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr2, node2);
967 btree1 = nodehdr1.btree;
968 btree2 = nodehdr2.btree;
969 swap = 1;
970 }
971
972 count = (nodehdr1.count - nodehdr2.count) / 2;
973 if (count == 0)
974 return;
975 tp = state->args->trans;
976 /*
977 * Two cases: high-to-low and low-to-high.
978 */
979 if (count > 0) {
980 /*
981 * Move elements in node2 up to make a hole.
982 */
983 tmp = nodehdr2.count;
984 if (tmp > 0) {
985 tmp *= (uint)sizeof(xfs_da_node_entry_t);
986 btree_s = &btree2[0];
987 btree_d = &btree2[count];
988 memmove(btree_d, btree_s, tmp);
989 }
990
991 /*
992 * Move the req'd B-tree elements from high in node1 to
993 * low in node2.
994 */
995 nodehdr2.count += count;
996 tmp = count * (uint)sizeof(xfs_da_node_entry_t);
997 btree_s = &btree1[nodehdr1.count - count];
998 btree_d = &btree2[0];
999 memcpy(btree_d, btree_s, tmp);
1000 nodehdr1.count -= count;
1001 } else {
1002 /*
1003 * Move the req'd B-tree elements from low in node2 to
1004 * high in node1.
1005 */
1006 count = -count;
1007 tmp = count * (uint)sizeof(xfs_da_node_entry_t);
1008 btree_s = &btree2[0];
1009 btree_d = &btree1[nodehdr1.count];
1010 memcpy(btree_d, btree_s, tmp);
1011 nodehdr1.count += count;
1012
1013 xfs_trans_log_buf(tp, blk1->bp,
1014 XFS_DA_LOGRANGE(node1, btree_d, tmp));
1015
1016 /*
1017 * Move elements in node2 down to fill the hole.
1018 */
1019 tmp = nodehdr2.count - count;
1020 tmp *= (uint)sizeof(xfs_da_node_entry_t);
1021 btree_s = &btree2[count];
1022 btree_d = &btree2[0];
1023 memmove(btree_d, btree_s, tmp);
1024 nodehdr2.count -= count;
1025 }
1026
1027 /*
1028 * Log header of node 1 and all current bits of node 2.
1029 */
1030 xfs_da3_node_hdr_to_disk(dp->i_mount, node1, &nodehdr1);
1031 xfs_trans_log_buf(tp, blk1->bp,
1032 XFS_DA_LOGRANGE(node1, &node1->hdr,
1033 state->args->geo->node_hdr_size));
1034
1035 xfs_da3_node_hdr_to_disk(dp->i_mount, node2, &nodehdr2);
1036 xfs_trans_log_buf(tp, blk2->bp,
1037 XFS_DA_LOGRANGE(node2, &node2->hdr,
1038 state->args->geo->node_hdr_size +
1039 (sizeof(btree2[0]) * nodehdr2.count)));
1040
1041 /*
1042 * Record the last hashval from each block for upward propagation.
1043 * (note: don't use the swapped node pointers)
1044 */
1045 if (swap) {
1046 node1 = blk1->bp->b_addr;
1047 node2 = blk2->bp->b_addr;
1048 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr1, node1);
1049 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr2, node2);
1050 btree1 = nodehdr1.btree;
1051 btree2 = nodehdr2.btree;
1052 }
1053 blk1->hashval = be32_to_cpu(btree1[nodehdr1.count - 1].hashval);
1054 blk2->hashval = be32_to_cpu(btree2[nodehdr2.count - 1].hashval);
1055
1056 /*
1057 * Adjust the expected index for insertion.
1058 */
1059 if (blk1->index >= nodehdr1.count) {
1060 blk2->index = blk1->index - nodehdr1.count;
1061 blk1->index = nodehdr1.count + 1; /* make it invalid */
1062 }
1063 }
1064
1065 /*
1066 * Add a new entry to an intermediate node.
1067 */
1068 STATIC void
1069 xfs_da3_node_add(
1070 struct xfs_da_state *state,
1071 struct xfs_da_state_blk *oldblk,
1072 struct xfs_da_state_blk *newblk)
1073 {
1074 struct xfs_da_intnode *node;
1075 struct xfs_da3_icnode_hdr nodehdr;
1076 struct xfs_da_node_entry *btree;
1077 int tmp;
1078 struct xfs_inode *dp = state->args->dp;
1079
1080 trace_xfs_da_node_add(state->args);
1081
1082 node = oldblk->bp->b_addr;
1083 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
1084 btree = nodehdr.btree;
1085
1086 ASSERT(oldblk->index >= 0 && oldblk->index <= nodehdr.count);
1087 ASSERT(newblk->blkno != 0);
1088 if (state->args->whichfork == XFS_DATA_FORK)
1089 ASSERT(newblk->blkno >= state->args->geo->leafblk &&
1090 newblk->blkno < state->args->geo->freeblk);
1091
1092 /*
1093 * We may need to make some room before we insert the new node.
1094 */
1095 tmp = 0;
1096 if (oldblk->index < nodehdr.count) {
1097 tmp = (nodehdr.count - oldblk->index) * (uint)sizeof(*btree);
1098 memmove(&btree[oldblk->index + 1], &btree[oldblk->index], tmp);
1099 }
1100 btree[oldblk->index].hashval = cpu_to_be32(newblk->hashval);
1101 btree[oldblk->index].before = cpu_to_be32(newblk->blkno);
1102 xfs_trans_log_buf(state->args->trans, oldblk->bp,
1103 XFS_DA_LOGRANGE(node, &btree[oldblk->index],
1104 tmp + sizeof(*btree)));
1105
1106 nodehdr.count += 1;
1107 xfs_da3_node_hdr_to_disk(dp->i_mount, node, &nodehdr);
1108 xfs_trans_log_buf(state->args->trans, oldblk->bp,
1109 XFS_DA_LOGRANGE(node, &node->hdr,
1110 state->args->geo->node_hdr_size));
1111
1112 /*
1113 * Copy the last hash value from the oldblk to propagate upwards.
1114 */
1115 oldblk->hashval = be32_to_cpu(btree[nodehdr.count - 1].hashval);
1116 }
1117
1118 /*========================================================================
1119 * Routines used for shrinking the Btree.
1120 *========================================================================*/
1121
1122 /*
1123 * Deallocate an empty leaf node, remove it from its parent,
1124 * possibly deallocating that block, etc...
1125 */
1126 int
1127 xfs_da3_join(
1128 struct xfs_da_state *state)
1129 {
1130 struct xfs_da_state_blk *drop_blk;
1131 struct xfs_da_state_blk *save_blk;
1132 int action = 0;
1133 int error;
1134
1135 trace_xfs_da_join(state->args);
1136
1137 drop_blk = &state->path.blk[ state->path.active-1 ];
1138 save_blk = &state->altpath.blk[ state->path.active-1 ];
1139 ASSERT(state->path.blk[0].magic == XFS_DA_NODE_MAGIC);
1140 ASSERT(drop_blk->magic == XFS_ATTR_LEAF_MAGIC ||
1141 drop_blk->magic == XFS_DIR2_LEAFN_MAGIC);
1142
1143 /*
1144 * Walk back up the tree joining/deallocating as necessary.
1145 * When we stop dropping blocks, break out.
1146 */
1147 for ( ; state->path.active >= 2; drop_blk--, save_blk--,
1148 state->path.active--) {
1149 /*
1150 * See if we can combine the block with a neighbor.
1151 * (action == 0) => no options, just leave
1152 * (action == 1) => coalesce, then unlink
1153 * (action == 2) => block empty, unlink it
1154 */
1155 switch (drop_blk->magic) {
1156 case XFS_ATTR_LEAF_MAGIC:
1157 error = xfs_attr3_leaf_toosmall(state, &action);
1158 if (error)
1159 return error;
1160 if (action == 0)
1161 return 0;
1162 xfs_attr3_leaf_unbalance(state, drop_blk, save_blk);
1163 break;
1164 case XFS_DIR2_LEAFN_MAGIC:
1165 error = xfs_dir2_leafn_toosmall(state, &action);
1166 if (error)
1167 return error;
1168 if (action == 0)
1169 return 0;
1170 xfs_dir2_leafn_unbalance(state, drop_blk, save_blk);
1171 break;
1172 case XFS_DA_NODE_MAGIC:
1173 /*
1174 * Remove the offending node, fixup hashvals,
1175 * check for a toosmall neighbor.
1176 */
1177 xfs_da3_node_remove(state, drop_blk);
1178 xfs_da3_fixhashpath(state, &state->path);
1179 error = xfs_da3_node_toosmall(state, &action);
1180 if (error)
1181 return error;
1182 if (action == 0)
1183 return 0;
1184 xfs_da3_node_unbalance(state, drop_blk, save_blk);
1185 break;
1186 }
1187 xfs_da3_fixhashpath(state, &state->altpath);
1188 error = xfs_da3_blk_unlink(state, drop_blk, save_blk);
1189 xfs_da_state_kill_altpath(state);
1190 if (error)
1191 return error;
1192 error = xfs_da_shrink_inode(state->args, drop_blk->blkno,
1193 drop_blk->bp);
1194 drop_blk->bp = NULL;
1195 if (error)
1196 return error;
1197 }
1198 /*
1199 * We joined all the way to the top. If it turns out that
1200 * we only have one entry in the root, make the child block
1201 * the new root.
1202 */
1203 xfs_da3_node_remove(state, drop_blk);
1204 xfs_da3_fixhashpath(state, &state->path);
1205 error = xfs_da3_root_join(state, &state->path.blk[0]);
1206 return error;
1207 }
1208
1209 #ifdef DEBUG
1210 static void
1211 xfs_da_blkinfo_onlychild_validate(struct xfs_da_blkinfo *blkinfo, __u16 level)
1212 {
1213 __be16 magic = blkinfo->magic;
1214
1215 if (level == 1) {
1216 ASSERT(magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
1217 magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) ||
1218 magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC) ||
1219 magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC));
1220 } else {
1221 ASSERT(magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
1222 magic == cpu_to_be16(XFS_DA3_NODE_MAGIC));
1223 }
1224 ASSERT(!blkinfo->forw);
1225 ASSERT(!blkinfo->back);
1226 }
1227 #else /* !DEBUG */
1228 #define xfs_da_blkinfo_onlychild_validate(blkinfo, level)
1229 #endif /* !DEBUG */
1230
1231 /*
1232 * We have only one entry in the root. Copy the only remaining child of
1233 * the old root to block 0 as the new root node.
1234 */
1235 STATIC int
1236 xfs_da3_root_join(
1237 struct xfs_da_state *state,
1238 struct xfs_da_state_blk *root_blk)
1239 {
1240 struct xfs_da_intnode *oldroot;
1241 struct xfs_da_args *args;
1242 xfs_dablk_t child;
1243 struct xfs_buf *bp;
1244 struct xfs_da3_icnode_hdr oldroothdr;
1245 int error;
1246 struct xfs_inode *dp = state->args->dp;
1247 xfs_failaddr_t fa;
1248
1249 trace_xfs_da_root_join(state->args);
1250
1251 ASSERT(root_blk->magic == XFS_DA_NODE_MAGIC);
1252
1253 args = state->args;
1254 oldroot = root_blk->bp->b_addr;
1255 xfs_da3_node_hdr_from_disk(dp->i_mount, &oldroothdr, oldroot);
1256 ASSERT(oldroothdr.forw == 0);
1257 ASSERT(oldroothdr.back == 0);
1258
1259 /*
1260 * If the root has more than one child, then don't do anything.
1261 */
1262 if (oldroothdr.count > 1)
1263 return 0;
1264
1265 /*
1266 * Read in the (only) child block, then copy those bytes into
1267 * the root block's buffer and free the original child block.
1268 */
1269 child = be32_to_cpu(oldroothdr.btree[0].before);
1270 ASSERT(child != 0);
1271 error = xfs_da3_node_read(args->trans, dp, child, &bp, args->whichfork);
1272 if (error)
1273 return error;
1274 fa = xfs_da3_header_check(bp, args->owner);
1275 if (fa) {
1276 __xfs_buf_mark_corrupt(bp, fa);
1277 xfs_trans_brelse(args->trans, bp);
1278 xfs_da_mark_sick(args);
1279 return -EFSCORRUPTED;
1280 }
1281 xfs_da_blkinfo_onlychild_validate(bp->b_addr, oldroothdr.level);
1282
1283 /*
1284 * Copy child to root buffer and log it.
1285 */
1286 xfs_da_buf_copy(root_blk->bp, bp, args->geo->blksize);
1287 xfs_trans_log_buf(args->trans, root_blk->bp, 0,
1288 args->geo->blksize - 1);
1289 /*
1290 * Now we can drop the child buffer.
1291 */
1292 error = xfs_da_shrink_inode(args, child, bp);
1293 return error;
1294 }
1295
1296 /*
1297 * Check a node block and its neighbors to see if the block should be
1298 * collapsed into one or the other neighbor. Always keep the block
1299 * with the smaller block number.
1300 * If the current block is over 50% full, don't try to join it, return 0.
1301 * If the block is empty, fill in the state structure and return 2.
1302 * If it can be collapsed, fill in the state structure and return 1.
1303 * If nothing can be done, return 0.
1304 */
1305 STATIC int
1306 xfs_da3_node_toosmall(
1307 struct xfs_da_state *state,
1308 int *action)
1309 {
1310 struct xfs_da_intnode *node;
1311 struct xfs_da_state_blk *blk;
1312 struct xfs_da_blkinfo *info;
1313 xfs_dablk_t blkno;
1314 struct xfs_buf *bp;
1315 xfs_failaddr_t fa;
1316 struct xfs_da3_icnode_hdr nodehdr;
1317 int count;
1318 int forward;
1319 int error;
1320 int retval;
1321 int i;
1322 struct xfs_inode *dp = state->args->dp;
1323
1324 trace_xfs_da_node_toosmall(state->args);
1325
1326 /*
1327 * Check for the degenerate case of the block being over 50% full.
1328 * If so, it's not worth even looking to see if we might be able
1329 * to coalesce with a sibling.
1330 */
1331 blk = &state->path.blk[ state->path.active-1 ];
1332 info = blk->bp->b_addr;
1333 node = (xfs_da_intnode_t *)info;
1334 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
1335 if (nodehdr.count > (state->args->geo->node_ents >> 1)) {
1336 *action = 0; /* blk over 50%, don't try to join */
1337 return 0; /* blk over 50%, don't try to join */
1338 }
1339
1340 /*
1341 * Check for the degenerate case of the block being empty.
1342 * If the block is empty, we'll simply delete it, no need to
1343 * coalesce it with a sibling block. We choose (arbitrarily)
1344 * to merge with the forward block unless it is NULL.
1345 */
1346 if (nodehdr.count == 0) {
1347 /*
1348 * Make altpath point to the block we want to keep and
1349 * path point to the block we want to drop (this one).
1350 */
1351 forward = (info->forw != 0);
1352 memcpy(&state->altpath, &state->path, sizeof(state->path));
1353 error = xfs_da3_path_shift(state, &state->altpath, forward,
1354 0, &retval);
1355 if (error)
1356 return error;
1357 if (retval) {
1358 *action = 0;
1359 } else {
1360 *action = 2;
1361 }
1362 return 0;
1363 }
1364
1365 /*
1366 * Examine each sibling block to see if we can coalesce with
1367 * at least 25% free space to spare. We need to figure out
1368 * whether to merge with the forward or the backward block.
1369 * We prefer coalescing with the lower numbered sibling so as
1370 * to shrink a directory over time.
1371 */
1372 count = state->args->geo->node_ents;
1373 count -= state->args->geo->node_ents >> 2;
1374 count -= nodehdr.count;
1375
1376 /* start with smaller blk num */
1377 forward = nodehdr.forw < nodehdr.back;
1378 for (i = 0; i < 2; forward = !forward, i++) {
1379 struct xfs_da3_icnode_hdr thdr;
1380 if (forward)
1381 blkno = nodehdr.forw;
1382 else
1383 blkno = nodehdr.back;
1384 if (blkno == 0)
1385 continue;
1386 error = xfs_da3_node_read(state->args->trans, dp, blkno, &bp,
1387 state->args->whichfork);
1388 if (error)
1389 return error;
1390 fa = xfs_da3_node_header_check(bp, state->args->owner);
1391 if (fa) {
1392 __xfs_buf_mark_corrupt(bp, fa);
1393 xfs_trans_brelse(state->args->trans, bp);
1394 xfs_da_mark_sick(state->args);
1395 return -EFSCORRUPTED;
1396 }
1397
1398 node = bp->b_addr;
1399 xfs_da3_node_hdr_from_disk(dp->i_mount, &thdr, node);
1400 xfs_trans_brelse(state->args->trans, bp);
1401
1402 if (count - thdr.count >= 0)
1403 break; /* fits with at least 25% to spare */
1404 }
1405 if (i >= 2) {
1406 *action = 0;
1407 return 0;
1408 }
1409
1410 /*
1411 * Make altpath point to the block we want to keep (the lower
1412 * numbered block) and path point to the block we want to drop.
1413 */
1414 memcpy(&state->altpath, &state->path, sizeof(state->path));
1415 if (blkno < blk->blkno) {
1416 error = xfs_da3_path_shift(state, &state->altpath, forward,
1417 0, &retval);
1418 } else {
1419 error = xfs_da3_path_shift(state, &state->path, forward,
1420 0, &retval);
1421 }
1422 if (error)
1423 return error;
1424 if (retval) {
1425 *action = 0;
1426 return 0;
1427 }
1428 *action = 1;
1429 return 0;
1430 }
1431
1432 /*
1433 * Pick up the last hashvalue from an intermediate node.
1434 */
1435 STATIC uint
1436 xfs_da3_node_lasthash(
1437 struct xfs_inode *dp,
1438 struct xfs_buf *bp,
1439 int *count)
1440 {
1441 struct xfs_da3_icnode_hdr nodehdr;
1442
1443 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, bp->b_addr);
1444 if (count)
1445 *count = nodehdr.count;
1446 if (!nodehdr.count)
1447 return 0;
1448 return be32_to_cpu(nodehdr.btree[nodehdr.count - 1].hashval);
1449 }
1450
1451 /*
1452 * Walk back up the tree adjusting hash values as necessary,
1453 * when we stop making changes, return.
1454 */
1455 void
1456 xfs_da3_fixhashpath(
1457 struct xfs_da_state *state,
1458 struct xfs_da_state_path *path)
1459 {
1460 struct xfs_da_state_blk *blk;
1461 struct xfs_da_intnode *node;
1462 struct xfs_da_node_entry *btree;
1463 xfs_dahash_t lasthash=0;
1464 int level;
1465 int count;
1466 struct xfs_inode *dp = state->args->dp;
1467
1468 trace_xfs_da_fixhashpath(state->args);
1469
1470 level = path->active-1;
1471 blk = &path->blk[ level ];
1472 switch (blk->magic) {
1473 case XFS_ATTR_LEAF_MAGIC:
1474 lasthash = xfs_attr_leaf_lasthash(blk->bp, &count);
1475 if (count == 0)
1476 return;
1477 break;
1478 case XFS_DIR2_LEAFN_MAGIC:
1479 lasthash = xfs_dir2_leaf_lasthash(dp, blk->bp, &count);
1480 if (count == 0)
1481 return;
1482 break;
1483 case XFS_DA_NODE_MAGIC:
1484 lasthash = xfs_da3_node_lasthash(dp, blk->bp, &count);
1485 if (count == 0)
1486 return;
1487 break;
1488 }
1489 for (blk--, level--; level >= 0; blk--, level--) {
1490 struct xfs_da3_icnode_hdr nodehdr;
1491
1492 node = blk->bp->b_addr;
1493 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
1494 btree = nodehdr.btree;
1495 if (be32_to_cpu(btree[blk->index].hashval) == lasthash)
1496 break;
1497 blk->hashval = lasthash;
1498 btree[blk->index].hashval = cpu_to_be32(lasthash);
1499 xfs_trans_log_buf(state->args->trans, blk->bp,
1500 XFS_DA_LOGRANGE(node, &btree[blk->index],
1501 sizeof(*btree)));
1502
1503 lasthash = be32_to_cpu(btree[nodehdr.count - 1].hashval);
1504 }
1505 }
1506
1507 /*
1508 * Remove an entry from an intermediate node.
1509 */
1510 STATIC void
1511 xfs_da3_node_remove(
1512 struct xfs_da_state *state,
1513 struct xfs_da_state_blk *drop_blk)
1514 {
1515 struct xfs_da_intnode *node;
1516 struct xfs_da3_icnode_hdr nodehdr;
1517 struct xfs_da_node_entry *btree;
1518 int index;
1519 int tmp;
1520 struct xfs_inode *dp = state->args->dp;
1521
1522 trace_xfs_da_node_remove(state->args);
1523
1524 node = drop_blk->bp->b_addr;
1525 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
1526 ASSERT(drop_blk->index < nodehdr.count);
1527 ASSERT(drop_blk->index >= 0);
1528
1529 /*
1530 * Copy over the offending entry, or just zero it out.
1531 */
1532 index = drop_blk->index;
1533 btree = nodehdr.btree;
1534 if (index < nodehdr.count - 1) {
1535 tmp = nodehdr.count - index - 1;
1536 tmp *= (uint)sizeof(xfs_da_node_entry_t);
1537 memmove(&btree[index], &btree[index + 1], tmp);
1538 xfs_trans_log_buf(state->args->trans, drop_blk->bp,
1539 XFS_DA_LOGRANGE(node, &btree[index], tmp));
1540 index = nodehdr.count - 1;
1541 }
1542 memset(&btree[index], 0, sizeof(xfs_da_node_entry_t));
1543 xfs_trans_log_buf(state->args->trans, drop_blk->bp,
1544 XFS_DA_LOGRANGE(node, &btree[index], sizeof(btree[index])));
1545 nodehdr.count -= 1;
1546 xfs_da3_node_hdr_to_disk(dp->i_mount, node, &nodehdr);
1547 xfs_trans_log_buf(state->args->trans, drop_blk->bp,
1548 XFS_DA_LOGRANGE(node, &node->hdr, state->args->geo->node_hdr_size));
1549
1550 /*
1551 * Copy the last hash value from the block to propagate upwards.
1552 */
1553 drop_blk->hashval = be32_to_cpu(btree[index - 1].hashval);
1554 }
1555
1556 /*
1557 * Unbalance the elements between two intermediate nodes,
1558 * move all Btree elements from one node into another.
1559 */
1560 STATIC void
1561 xfs_da3_node_unbalance(
1562 struct xfs_da_state *state,
1563 struct xfs_da_state_blk *drop_blk,
1564 struct xfs_da_state_blk *save_blk)
1565 {
1566 struct xfs_da_intnode *drop_node;
1567 struct xfs_da_intnode *save_node;
1568 struct xfs_da_node_entry *drop_btree;
1569 struct xfs_da_node_entry *save_btree;
1570 struct xfs_da3_icnode_hdr drop_hdr;
1571 struct xfs_da3_icnode_hdr save_hdr;
1572 struct xfs_trans *tp;
1573 int sindex;
1574 int tmp;
1575 struct xfs_inode *dp = state->args->dp;
1576
1577 trace_xfs_da_node_unbalance(state->args);
1578
1579 drop_node = drop_blk->bp->b_addr;
1580 save_node = save_blk->bp->b_addr;
1581 xfs_da3_node_hdr_from_disk(dp->i_mount, &drop_hdr, drop_node);
1582 xfs_da3_node_hdr_from_disk(dp->i_mount, &save_hdr, save_node);
1583 drop_btree = drop_hdr.btree;
1584 save_btree = save_hdr.btree;
1585 tp = state->args->trans;
1586
1587 /*
1588 * If the dying block has lower hashvals, then move all the
1589 * elements in the remaining block up to make a hole.
1590 */
1591 if ((be32_to_cpu(drop_btree[0].hashval) <
1592 be32_to_cpu(save_btree[0].hashval)) ||
1593 (be32_to_cpu(drop_btree[drop_hdr.count - 1].hashval) <
1594 be32_to_cpu(save_btree[save_hdr.count - 1].hashval))) {
1595 /* XXX: check this - is memmove dst correct? */
1596 tmp = save_hdr.count * sizeof(xfs_da_node_entry_t);
1597 memmove(&save_btree[drop_hdr.count], &save_btree[0], tmp);
1598
1599 sindex = 0;
1600 xfs_trans_log_buf(tp, save_blk->bp,
1601 XFS_DA_LOGRANGE(save_node, &save_btree[0],
1602 (save_hdr.count + drop_hdr.count) *
1603 sizeof(xfs_da_node_entry_t)));
1604 } else {
1605 sindex = save_hdr.count;
1606 xfs_trans_log_buf(tp, save_blk->bp,
1607 XFS_DA_LOGRANGE(save_node, &save_btree[sindex],
1608 drop_hdr.count * sizeof(xfs_da_node_entry_t)));
1609 }
1610
1611 /*
1612 * Move all the B-tree elements from drop_blk to save_blk.
1613 */
1614 tmp = drop_hdr.count * (uint)sizeof(xfs_da_node_entry_t);
1615 memcpy(&save_btree[sindex], &drop_btree[0], tmp);
1616 save_hdr.count += drop_hdr.count;
1617
1618 xfs_da3_node_hdr_to_disk(dp->i_mount, save_node, &save_hdr);
1619 xfs_trans_log_buf(tp, save_blk->bp,
1620 XFS_DA_LOGRANGE(save_node, &save_node->hdr,
1621 state->args->geo->node_hdr_size));
1622
1623 /*
1624 * Save the last hashval in the remaining block for upward propagation.
1625 */
1626 save_blk->hashval = be32_to_cpu(save_btree[save_hdr.count - 1].hashval);
1627 }
1628
1629 /*========================================================================
1630 * Routines used for finding things in the Btree.
1631 *========================================================================*/
1632
1633 /*
1634 * Walk down the Btree looking for a particular filename, filling
1635 * in the state structure as we go.
1636 *
1637 * We will set the state structure to point to each of the elements
1638 * in each of the nodes where either the hashval is or should be.
1639 *
1640 * We support duplicate hashval's so for each entry in the current
1641 * node that could contain the desired hashval, descend. This is a
1642 * pruned depth-first tree search.
1643 */
1644 int /* error */
1645 xfs_da3_node_lookup_int(
1646 struct xfs_da_state *state,
1647 int *result)
1648 {
1649 struct xfs_da_state_blk *blk;
1650 struct xfs_da_blkinfo *curr;
1651 struct xfs_da_intnode *node;
1652 struct xfs_da_node_entry *btree;
1653 struct xfs_da3_icnode_hdr nodehdr;
1654 struct xfs_da_args *args;
1655 xfs_failaddr_t fa;
1656 xfs_dablk_t blkno;
1657 xfs_dahash_t hashval;
1658 xfs_dahash_t btreehashval;
1659 int probe;
1660 int span;
1661 int max;
1662 int error;
1663 int retval;
1664 unsigned int expected_level = 0;
1665 uint16_t magic;
1666 struct xfs_inode *dp = state->args->dp;
1667
1668 args = state->args;
1669
1670 /*
1671 * Descend thru the B-tree searching each level for the right
1672 * node to use, until the right hashval is found.
1673 */
1674 blkno = args->geo->leafblk;
1675 for (blk = &state->path.blk[0], state->path.active = 1;
1676 state->path.active <= XFS_DA_NODE_MAXDEPTH;
1677 blk++, state->path.active++) {
1678 /*
1679 * Read the next node down in the tree.
1680 */
1681 blk->blkno = blkno;
1682 error = xfs_da3_node_read(args->trans, args->dp, blkno,
1683 &blk->bp, args->whichfork);
1684 if (error) {
1685 blk->blkno = 0;
1686 state->path.active--;
1687 return error;
1688 }
1689 curr = blk->bp->b_addr;
1690 magic = be16_to_cpu(curr->magic);
1691
1692 if (magic == XFS_ATTR_LEAF_MAGIC ||
1693 magic == XFS_ATTR3_LEAF_MAGIC) {
1694 fa = xfs_attr3_leaf_header_check(blk->bp, args->owner);
1695 if (fa) {
1696 __xfs_buf_mark_corrupt(blk->bp, fa);
1697 xfs_da_mark_sick(args);
1698 return -EFSCORRUPTED;
1699 }
1700 blk->magic = XFS_ATTR_LEAF_MAGIC;
1701 blk->hashval = xfs_attr_leaf_lasthash(blk->bp, NULL);
1702 break;
1703 }
1704
1705 if (magic == XFS_DIR2_LEAFN_MAGIC ||
1706 magic == XFS_DIR3_LEAFN_MAGIC) {
1707 fa = xfs_dir3_leaf_header_check(blk->bp, args->owner);
1708 if (fa) {
1709 __xfs_buf_mark_corrupt(blk->bp, fa);
1710 xfs_da_mark_sick(args);
1711 return -EFSCORRUPTED;
1712 }
1713 blk->magic = XFS_DIR2_LEAFN_MAGIC;
1714 blk->hashval = xfs_dir2_leaf_lasthash(args->dp,
1715 blk->bp, NULL);
1716 break;
1717 }
1718
1719 if (magic != XFS_DA_NODE_MAGIC && magic != XFS_DA3_NODE_MAGIC) {
1720 xfs_buf_mark_corrupt(blk->bp);
1721 xfs_da_mark_sick(args);
1722 return -EFSCORRUPTED;
1723 }
1724
1725 fa = xfs_da3_node_header_check(blk->bp, args->owner);
1726 if (fa) {
1727 __xfs_buf_mark_corrupt(blk->bp, fa);
1728 xfs_da_mark_sick(args);
1729 return -EFSCORRUPTED;
1730 }
1731
1732 blk->magic = XFS_DA_NODE_MAGIC;
1733
1734 /*
1735 * Search an intermediate node for a match.
1736 */
1737 node = blk->bp->b_addr;
1738 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr, node);
1739 btree = nodehdr.btree;
1740
1741 /* Tree taller than we can handle; bail out! */
1742 if (nodehdr.level >= XFS_DA_NODE_MAXDEPTH) {
1743 xfs_buf_mark_corrupt(blk->bp);
1744 xfs_da_mark_sick(args);
1745 return -EFSCORRUPTED;
1746 }
1747
1748 /* Check the level from the root. */
1749 if (blkno == args->geo->leafblk)
1750 expected_level = nodehdr.level - 1;
1751 else if (expected_level != nodehdr.level) {
1752 xfs_buf_mark_corrupt(blk->bp);
1753 xfs_da_mark_sick(args);
1754 return -EFSCORRUPTED;
1755 } else
1756 expected_level--;
1757
1758 max = nodehdr.count;
1759 blk->hashval = be32_to_cpu(btree[max - 1].hashval);
1760
1761 /*
1762 * Binary search. (note: small blocks will skip loop)
1763 */
1764 probe = span = max / 2;
1765 hashval = args->hashval;
1766 while (span > 4) {
1767 span /= 2;
1768 btreehashval = be32_to_cpu(btree[probe].hashval);
1769 if (btreehashval < hashval)
1770 probe += span;
1771 else if (btreehashval > hashval)
1772 probe -= span;
1773 else
1774 break;
1775 }
1776 ASSERT((probe >= 0) && (probe < max));
1777 ASSERT((span <= 4) ||
1778 (be32_to_cpu(btree[probe].hashval) == hashval));
1779
1780 /*
1781 * Since we may have duplicate hashval's, find the first
1782 * matching hashval in the node.
1783 */
1784 while (probe > 0 &&
1785 be32_to_cpu(btree[probe].hashval) >= hashval) {
1786 probe--;
1787 }
1788 while (probe < max &&
1789 be32_to_cpu(btree[probe].hashval) < hashval) {
1790 probe++;
1791 }
1792
1793 /*
1794 * Pick the right block to descend on.
1795 */
1796 if (probe == max) {
1797 blk->index = max - 1;
1798 blkno = be32_to_cpu(btree[max - 1].before);
1799 } else {
1800 blk->index = probe;
1801 blkno = be32_to_cpu(btree[probe].before);
1802 }
1803
1804 /* We can't point back to the root. */
1805 if (XFS_IS_CORRUPT(dp->i_mount, blkno == args->geo->leafblk)) {
1806 xfs_da_mark_sick(args);
1807 return -EFSCORRUPTED;
1808 }
1809 }
1810
1811 if (XFS_IS_CORRUPT(dp->i_mount, expected_level != 0)) {
1812 xfs_da_mark_sick(args);
1813 return -EFSCORRUPTED;
1814 }
1815
1816 /*
1817 * A leaf block that ends in the hashval that we are interested in
1818 * (final hashval == search hashval) means that the next block may
1819 * contain more entries with the same hashval, shift upward to the
1820 * next leaf and keep searching.
1821 */
1822 for (;;) {
1823 if (blk->magic == XFS_DIR2_LEAFN_MAGIC) {
1824 retval = xfs_dir2_leafn_lookup_int(blk->bp, args,
1825 &blk->index, state);
1826 } else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
1827 retval = xfs_attr3_leaf_lookup_int(blk->bp, args);
1828 blk->index = args->index;
1829 args->blkno = blk->blkno;
1830 } else {
1831 ASSERT(0);
1832 xfs_da_mark_sick(args);
1833 return -EFSCORRUPTED;
1834 }
1835 if (((retval == -ENOENT) || (retval == -ENOATTR)) &&
1836 (blk->hashval == args->hashval)) {
1837 error = xfs_da3_path_shift(state, &state->path, 1, 1,
1838 &retval);
1839 if (error)
1840 return error;
1841 if (retval == 0) {
1842 continue;
1843 } else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
1844 /* path_shift() gives ENOENT */
1845 retval = -ENOATTR;
1846 }
1847 }
1848 break;
1849 }
1850 *result = retval;
1851 return 0;
1852 }
1853
1854 /*========================================================================
1855 * Utility routines.
1856 *========================================================================*/
1857
1858 /*
1859 * Compare two intermediate nodes for "order".
1860 */
1861 STATIC int
1862 xfs_da3_node_order(
1863 struct xfs_inode *dp,
1864 struct xfs_buf *node1_bp,
1865 struct xfs_buf *node2_bp)
1866 {
1867 struct xfs_da_intnode *node1;
1868 struct xfs_da_intnode *node2;
1869 struct xfs_da_node_entry *btree1;
1870 struct xfs_da_node_entry *btree2;
1871 struct xfs_da3_icnode_hdr node1hdr;
1872 struct xfs_da3_icnode_hdr node2hdr;
1873
1874 node1 = node1_bp->b_addr;
1875 node2 = node2_bp->b_addr;
1876 xfs_da3_node_hdr_from_disk(dp->i_mount, &node1hdr, node1);
1877 xfs_da3_node_hdr_from_disk(dp->i_mount, &node2hdr, node2);
1878 btree1 = node1hdr.btree;
1879 btree2 = node2hdr.btree;
1880
1881 if (node1hdr.count > 0 && node2hdr.count > 0 &&
1882 ((be32_to_cpu(btree2[0].hashval) < be32_to_cpu(btree1[0].hashval)) ||
1883 (be32_to_cpu(btree2[node2hdr.count - 1].hashval) <
1884 be32_to_cpu(btree1[node1hdr.count - 1].hashval)))) {
1885 return 1;
1886 }
1887 return 0;
1888 }
1889
1890 /*
1891 * Link a new block into a doubly linked list of blocks (of whatever type).
1892 */
1893 int /* error */
1894 xfs_da3_blk_link(
1895 struct xfs_da_state *state,
1896 struct xfs_da_state_blk *old_blk,
1897 struct xfs_da_state_blk *new_blk)
1898 {
1899 struct xfs_da_blkinfo *old_info;
1900 struct xfs_da_blkinfo *new_info;
1901 struct xfs_da_blkinfo *tmp_info;
1902 struct xfs_da_args *args;
1903 struct xfs_buf *bp;
1904 xfs_failaddr_t fa;
1905 int before = 0;
1906 int error;
1907 struct xfs_inode *dp = state->args->dp;
1908
1909 /*
1910 * Set up environment.
1911 */
1912 args = state->args;
1913 ASSERT(args != NULL);
1914 old_info = old_blk->bp->b_addr;
1915 new_info = new_blk->bp->b_addr;
1916 ASSERT(old_blk->magic == XFS_DA_NODE_MAGIC ||
1917 old_blk->magic == XFS_DIR2_LEAFN_MAGIC ||
1918 old_blk->magic == XFS_ATTR_LEAF_MAGIC);
1919
1920 switch (old_blk->magic) {
1921 case XFS_ATTR_LEAF_MAGIC:
1922 before = xfs_attr_leaf_order(old_blk->bp, new_blk->bp);
1923 break;
1924 case XFS_DIR2_LEAFN_MAGIC:
1925 before = xfs_dir2_leafn_order(dp, old_blk->bp, new_blk->bp);
1926 break;
1927 case XFS_DA_NODE_MAGIC:
1928 before = xfs_da3_node_order(dp, old_blk->bp, new_blk->bp);
1929 break;
1930 }
1931
1932 /*
1933 * Link blocks in appropriate order.
1934 */
1935 if (before) {
1936 /*
1937 * Link new block in before existing block.
1938 */
1939 trace_xfs_da_link_before(args);
1940 new_info->forw = cpu_to_be32(old_blk->blkno);
1941 new_info->back = old_info->back;
1942 if (old_info->back) {
1943 error = xfs_da3_node_read(args->trans, dp,
1944 be32_to_cpu(old_info->back),
1945 &bp, args->whichfork);
1946 if (error)
1947 return error;
1948 fa = xfs_da3_header_check(bp, args->owner);
1949 if (fa) {
1950 __xfs_buf_mark_corrupt(bp, fa);
1951 xfs_trans_brelse(args->trans, bp);
1952 xfs_da_mark_sick(args);
1953 return -EFSCORRUPTED;
1954 }
1955 ASSERT(bp != NULL);
1956 tmp_info = bp->b_addr;
1957 ASSERT(tmp_info->magic == old_info->magic);
1958 ASSERT(be32_to_cpu(tmp_info->forw) == old_blk->blkno);
1959 tmp_info->forw = cpu_to_be32(new_blk->blkno);
1960 xfs_trans_log_buf(args->trans, bp, 0, sizeof(*tmp_info)-1);
1961 }
1962 old_info->back = cpu_to_be32(new_blk->blkno);
1963 } else {
1964 /*
1965 * Link new block in after existing block.
1966 */
1967 trace_xfs_da_link_after(args);
1968 new_info->forw = old_info->forw;
1969 new_info->back = cpu_to_be32(old_blk->blkno);
1970 if (old_info->forw) {
1971 error = xfs_da3_node_read(args->trans, dp,
1972 be32_to_cpu(old_info->forw),
1973 &bp, args->whichfork);
1974 if (error)
1975 return error;
1976 fa = xfs_da3_header_check(bp, args->owner);
1977 if (fa) {
1978 __xfs_buf_mark_corrupt(bp, fa);
1979 xfs_trans_brelse(args->trans, bp);
1980 xfs_da_mark_sick(args);
1981 return -EFSCORRUPTED;
1982 }
1983 ASSERT(bp != NULL);
1984 tmp_info = bp->b_addr;
1985 ASSERT(tmp_info->magic == old_info->magic);
1986 ASSERT(be32_to_cpu(tmp_info->back) == old_blk->blkno);
1987 tmp_info->back = cpu_to_be32(new_blk->blkno);
1988 xfs_trans_log_buf(args->trans, bp, 0, sizeof(*tmp_info)-1);
1989 }
1990 old_info->forw = cpu_to_be32(new_blk->blkno);
1991 }
1992
1993 xfs_trans_log_buf(args->trans, old_blk->bp, 0, sizeof(*tmp_info) - 1);
1994 xfs_trans_log_buf(args->trans, new_blk->bp, 0, sizeof(*tmp_info) - 1);
1995 return 0;
1996 }
1997
1998 /*
1999 * Unlink a block from a doubly linked list of blocks.
2000 */
2001 STATIC int /* error */
2002 xfs_da3_blk_unlink(
2003 struct xfs_da_state *state,
2004 struct xfs_da_state_blk *drop_blk,
2005 struct xfs_da_state_blk *save_blk)
2006 {
2007 struct xfs_da_blkinfo *drop_info;
2008 struct xfs_da_blkinfo *save_info;
2009 struct xfs_da_blkinfo *tmp_info;
2010 struct xfs_da_args *args;
2011 struct xfs_buf *bp;
2012 xfs_failaddr_t fa;
2013 int error;
2014
2015 /*
2016 * Set up environment.
2017 */
2018 args = state->args;
2019 ASSERT(args != NULL);
2020 save_info = save_blk->bp->b_addr;
2021 drop_info = drop_blk->bp->b_addr;
2022 ASSERT(save_blk->magic == XFS_DA_NODE_MAGIC ||
2023 save_blk->magic == XFS_DIR2_LEAFN_MAGIC ||
2024 save_blk->magic == XFS_ATTR_LEAF_MAGIC);
2025 ASSERT(save_blk->magic == drop_blk->magic);
2026 ASSERT((be32_to_cpu(save_info->forw) == drop_blk->blkno) ||
2027 (be32_to_cpu(save_info->back) == drop_blk->blkno));
2028 ASSERT((be32_to_cpu(drop_info->forw) == save_blk->blkno) ||
2029 (be32_to_cpu(drop_info->back) == save_blk->blkno));
2030
2031 /*
2032 * Unlink the leaf block from the doubly linked chain of leaves.
2033 */
2034 if (be32_to_cpu(save_info->back) == drop_blk->blkno) {
2035 trace_xfs_da_unlink_back(args);
2036 save_info->back = drop_info->back;
2037 if (drop_info->back) {
2038 error = xfs_da3_node_read(args->trans, args->dp,
2039 be32_to_cpu(drop_info->back),
2040 &bp, args->whichfork);
2041 if (error)
2042 return error;
2043 fa = xfs_da3_header_check(bp, args->owner);
2044 if (fa) {
2045 __xfs_buf_mark_corrupt(bp, fa);
2046 xfs_trans_brelse(args->trans, bp);
2047 xfs_da_mark_sick(args);
2048 return -EFSCORRUPTED;
2049 }
2050 ASSERT(bp != NULL);
2051 tmp_info = bp->b_addr;
2052 ASSERT(tmp_info->magic == save_info->magic);
2053 ASSERT(be32_to_cpu(tmp_info->forw) == drop_blk->blkno);
2054 tmp_info->forw = cpu_to_be32(save_blk->blkno);
2055 xfs_trans_log_buf(args->trans, bp, 0,
2056 sizeof(*tmp_info) - 1);
2057 }
2058 } else {
2059 trace_xfs_da_unlink_forward(args);
2060 save_info->forw = drop_info->forw;
2061 if (drop_info->forw) {
2062 error = xfs_da3_node_read(args->trans, args->dp,
2063 be32_to_cpu(drop_info->forw),
2064 &bp, args->whichfork);
2065 if (error)
2066 return error;
2067 fa = xfs_da3_header_check(bp, args->owner);
2068 if (fa) {
2069 __xfs_buf_mark_corrupt(bp, fa);
2070 xfs_trans_brelse(args->trans, bp);
2071 xfs_da_mark_sick(args);
2072 return -EFSCORRUPTED;
2073 }
2074 ASSERT(bp != NULL);
2075 tmp_info = bp->b_addr;
2076 ASSERT(tmp_info->magic == save_info->magic);
2077 ASSERT(be32_to_cpu(tmp_info->back) == drop_blk->blkno);
2078 tmp_info->back = cpu_to_be32(save_blk->blkno);
2079 xfs_trans_log_buf(args->trans, bp, 0,
2080 sizeof(*tmp_info) - 1);
2081 }
2082 }
2083
2084 xfs_trans_log_buf(args->trans, save_blk->bp, 0, sizeof(*save_info) - 1);
2085 return 0;
2086 }
2087
2088 /*
2089 * Move a path "forward" or "!forward" one block at the current level.
2090 *
2091 * This routine will adjust a "path" to point to the next block
2092 * "forward" (higher hashvalues) or "!forward" (lower hashvals) in the
2093 * Btree, including updating pointers to the intermediate nodes between
2094 * the new bottom and the root.
2095 */
2096 int /* error */
2097 xfs_da3_path_shift(
2098 struct xfs_da_state *state,
2099 struct xfs_da_state_path *path,
2100 int forward,
2101 int release,
2102 int *result)
2103 {
2104 struct xfs_da_state_blk *blk;
2105 struct xfs_da_blkinfo *info;
2106 struct xfs_da_args *args;
2107 struct xfs_da_node_entry *btree;
2108 struct xfs_da3_icnode_hdr nodehdr;
2109 struct xfs_buf *bp;
2110 xfs_failaddr_t fa;
2111 xfs_dablk_t blkno = 0;
2112 int level;
2113 int error;
2114 struct xfs_inode *dp = state->args->dp;
2115
2116 trace_xfs_da_path_shift(state->args);
2117
2118 /*
2119 * Roll up the Btree looking for the first block where our
2120 * current index is not at the edge of the block. Note that
2121 * we skip the bottom layer because we want the sibling block.
2122 */
2123 args = state->args;
2124 ASSERT(args != NULL);
2125 ASSERT(path != NULL);
2126 ASSERT((path->active > 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
2127 level = (path->active-1) - 1; /* skip bottom layer in path */
2128 for (; level >= 0; level--) {
2129 blk = &path->blk[level];
2130 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr,
2131 blk->bp->b_addr);
2132
2133 if (forward && (blk->index < nodehdr.count - 1)) {
2134 blk->index++;
2135 blkno = be32_to_cpu(nodehdr.btree[blk->index].before);
2136 break;
2137 } else if (!forward && (blk->index > 0)) {
2138 blk->index--;
2139 blkno = be32_to_cpu(nodehdr.btree[blk->index].before);
2140 break;
2141 }
2142 }
2143 if (level < 0) {
2144 *result = -ENOENT; /* we're out of our tree */
2145 ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
2146 return 0;
2147 }
2148
2149 /*
2150 * Roll down the edge of the subtree until we reach the
2151 * same depth we were at originally.
2152 */
2153 for (blk++, level++; level < path->active; blk++, level++) {
2154 /*
2155 * Read the next child block into a local buffer.
2156 */
2157 error = xfs_da3_node_read(args->trans, dp, blkno, &bp,
2158 args->whichfork);
2159 if (error)
2160 return error;
2161
2162 /*
2163 * Release the old block (if it's dirty, the trans doesn't
2164 * actually let go) and swap the local buffer into the path
2165 * structure. This ensures failure of the above read doesn't set
2166 * a NULL buffer in an active slot in the path.
2167 */
2168 if (release)
2169 xfs_trans_brelse(args->trans, blk->bp);
2170 blk->blkno = blkno;
2171 blk->bp = bp;
2172
2173 info = blk->bp->b_addr;
2174 ASSERT(info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
2175 info->magic == cpu_to_be16(XFS_DA3_NODE_MAGIC) ||
2176 info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
2177 info->magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) ||
2178 info->magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC) ||
2179 info->magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC));
2180
2181
2182 /*
2183 * Note: we flatten the magic number to a single type so we
2184 * don't have to compare against crc/non-crc types elsewhere.
2185 */
2186 switch (be16_to_cpu(info->magic)) {
2187 case XFS_DA_NODE_MAGIC:
2188 case XFS_DA3_NODE_MAGIC:
2189 fa = xfs_da3_node_header_check(blk->bp, args->owner);
2190 if (fa) {
2191 __xfs_buf_mark_corrupt(blk->bp, fa);
2192 xfs_da_mark_sick(args);
2193 return -EFSCORRUPTED;
2194 }
2195 blk->magic = XFS_DA_NODE_MAGIC;
2196 xfs_da3_node_hdr_from_disk(dp->i_mount, &nodehdr,
2197 bp->b_addr);
2198 btree = nodehdr.btree;
2199 blk->hashval = be32_to_cpu(btree[nodehdr.count - 1].hashval);
2200 if (forward)
2201 blk->index = 0;
2202 else
2203 blk->index = nodehdr.count - 1;
2204 blkno = be32_to_cpu(btree[blk->index].before);
2205 break;
2206 case XFS_ATTR_LEAF_MAGIC:
2207 case XFS_ATTR3_LEAF_MAGIC:
2208 fa = xfs_attr3_leaf_header_check(blk->bp, args->owner);
2209 if (fa) {
2210 __xfs_buf_mark_corrupt(blk->bp, fa);
2211 xfs_da_mark_sick(args);
2212 return -EFSCORRUPTED;
2213 }
2214 blk->magic = XFS_ATTR_LEAF_MAGIC;
2215 ASSERT(level == path->active-1);
2216 blk->index = 0;
2217 blk->hashval = xfs_attr_leaf_lasthash(blk->bp, NULL);
2218 break;
2219 case XFS_DIR2_LEAFN_MAGIC:
2220 case XFS_DIR3_LEAFN_MAGIC:
2221 fa = xfs_dir3_leaf_header_check(blk->bp, args->owner);
2222 if (fa) {
2223 __xfs_buf_mark_corrupt(blk->bp, fa);
2224 xfs_da_mark_sick(args);
2225 return -EFSCORRUPTED;
2226 }
2227 blk->magic = XFS_DIR2_LEAFN_MAGIC;
2228 ASSERT(level == path->active-1);
2229 blk->index = 0;
2230 blk->hashval = xfs_dir2_leaf_lasthash(args->dp,
2231 blk->bp, NULL);
2232 break;
2233 default:
2234 ASSERT(0);
2235 break;
2236 }
2237 }
2238 *result = 0;
2239 return 0;
2240 }
2241
2242
2243 /*========================================================================
2244 * Utility routines.
2245 *========================================================================*/
2246
2247 /*
2248 * Implement a simple hash on a character string.
2249 * Rotate the hash value by 7 bits, then XOR each character in.
2250 * This is implemented with some source-level loop unrolling.
2251 */
2252 xfs_dahash_t
2253 xfs_da_hashname(const uint8_t *name, int namelen)
2254 {
2255 xfs_dahash_t hash;
2256
2257 /*
2258 * Do four characters at a time as long as we can.
2259 */
2260 for (hash = 0; namelen >= 4; namelen -= 4, name += 4)
2261 hash = (name[0] << 21) ^ (name[1] << 14) ^ (name[2] << 7) ^
2262 (name[3] << 0) ^ rol32(hash, 7 * 4);
2263
2264 /*
2265 * Now do the rest of the characters.
2266 */
2267 switch (namelen) {
2268 case 3:
2269 return (name[0] << 14) ^ (name[1] << 7) ^ (name[2] << 0) ^
2270 rol32(hash, 7 * 3);
2271 case 2:
2272 return (name[0] << 7) ^ (name[1] << 0) ^ rol32(hash, 7 * 2);
2273 case 1:
2274 return (name[0] << 0) ^ rol32(hash, 7 * 1);
2275 default: /* case 0: */
2276 return hash;
2277 }
2278 }
2279
2280 enum xfs_dacmp
2281 xfs_da_compname(
2282 struct xfs_da_args *args,
2283 const unsigned char *name,
2284 int len)
2285 {
2286 return (args->namelen == len && memcmp(args->name, name, len) == 0) ?
2287 XFS_CMP_EXACT : XFS_CMP_DIFFERENT;
2288 }
2289
2290 int
2291 xfs_da_grow_inode_int(
2292 struct xfs_da_args *args,
2293 xfs_fileoff_t *bno,
2294 int count)
2295 {
2296 struct xfs_trans *tp = args->trans;
2297 struct xfs_inode *dp = args->dp;
2298 int w = args->whichfork;
2299 xfs_rfsblock_t nblks = dp->i_nblocks;
2300 struct xfs_bmbt_irec map, *mapp = ↦
2301 int nmap, error, got, i, mapi = 1;
2302
2303 /*
2304 * Find a spot in the file space to put the new block.
2305 */
2306 error = xfs_bmap_first_unused(tp, dp, count, bno, w);
2307 if (error)
2308 return error;
2309
2310 /*
2311 * Try mapping it in one filesystem block.
2312 */
2313 nmap = 1;
2314 error = xfs_bmapi_write(tp, dp, *bno, count,
2315 xfs_bmapi_aflag(w)|XFS_BMAPI_METADATA|XFS_BMAPI_CONTIG,
2316 args->total, &map, &nmap);
2317 if (error == -ENOSPC && count > 1) {
2318 xfs_fileoff_t b;
2319 int c;
2320
2321 /*
2322 * If we didn't get it and the block might work if fragmented,
2323 * try without the CONTIG flag. Loop until we get it all.
2324 */
2325 mapp = kmalloc(sizeof(*mapp) * count,
2326 GFP_KERNEL | __GFP_NOFAIL);
2327 for (b = *bno, mapi = 0; b < *bno + count; ) {
2328 c = (int)(*bno + count - b);
2329 nmap = min(XFS_BMAP_MAX_NMAP, c);
2330 error = xfs_bmapi_write(tp, dp, b, c,
2331 xfs_bmapi_aflag(w)|XFS_BMAPI_METADATA,
2332 args->total, &mapp[mapi], &nmap);
2333 if (error)
2334 goto out_free_map;
2335 mapi += nmap;
2336 b = mapp[mapi - 1].br_startoff +
2337 mapp[mapi - 1].br_blockcount;
2338 }
2339 }
2340 if (error)
2341 goto out_free_map;
2342
2343 /*
2344 * Count the blocks we got, make sure it matches the total.
2345 */
2346 for (i = 0, got = 0; i < mapi; i++)
2347 got += mapp[i].br_blockcount;
2348 if (got != count || mapp[0].br_startoff != *bno ||
2349 mapp[mapi - 1].br_startoff + mapp[mapi - 1].br_blockcount !=
2350 *bno + count) {
2351 error = -ENOSPC;
2352 goto out_free_map;
2353 }
2354
2355 /* account for newly allocated blocks in reserved blocks total */
2356 args->total -= dp->i_nblocks - nblks;
2357
2358 out_free_map:
2359 if (mapp != &map)
2360 kfree(mapp);
2361 return error;
2362 }
2363
2364 /*
2365 * Add a block to the btree ahead of the file.
2366 * Return the new block number to the caller.
2367 */
2368 int
2369 xfs_da_grow_inode(
2370 struct xfs_da_args *args,
2371 xfs_dablk_t *new_blkno)
2372 {
2373 xfs_fileoff_t bno;
2374 int error;
2375
2376 trace_xfs_da_grow_inode(args);
2377
2378 bno = args->geo->leafblk;
2379 error = xfs_da_grow_inode_int(args, &bno, args->geo->fsbcount);
2380 if (!error)
2381 *new_blkno = (xfs_dablk_t)bno;
2382 return error;
2383 }
2384
2385 /*
2386 * Ick. We need to always be able to remove a btree block, even
2387 * if there's no space reservation because the filesystem is full.
2388 * This is called if xfs_bunmapi on a btree block fails due to ENOSPC.
2389 * It swaps the target block with the last block in the file. The
2390 * last block in the file can always be removed since it can't cause
2391 * a bmap btree split to do that.
2392 */
2393 STATIC int
2394 xfs_da3_swap_lastblock(
2395 struct xfs_da_args *args,
2396 xfs_dablk_t *dead_blknop,
2397 struct xfs_buf **dead_bufp)
2398 {
2399 struct xfs_da_blkinfo *dead_info;
2400 struct xfs_da_blkinfo *sib_info;
2401 struct xfs_da_intnode *par_node;
2402 struct xfs_da_intnode *dead_node;
2403 struct xfs_dir2_leaf *dead_leaf2;
2404 struct xfs_da_node_entry *btree;
2405 struct xfs_da3_icnode_hdr par_hdr;
2406 struct xfs_inode *dp;
2407 struct xfs_trans *tp;
2408 struct xfs_mount *mp;
2409 struct xfs_buf *dead_buf;
2410 struct xfs_buf *last_buf;
2411 struct xfs_buf *sib_buf;
2412 struct xfs_buf *par_buf;
2413 xfs_failaddr_t fa;
2414 xfs_dahash_t dead_hash;
2415 xfs_fileoff_t lastoff;
2416 xfs_dablk_t dead_blkno;
2417 xfs_dablk_t last_blkno;
2418 xfs_dablk_t sib_blkno;
2419 xfs_dablk_t par_blkno;
2420 int error;
2421 int w;
2422 int entno;
2423 int level;
2424 int dead_level;
2425
2426 trace_xfs_da_swap_lastblock(args);
2427
2428 dead_buf = *dead_bufp;
2429 dead_blkno = *dead_blknop;
2430 tp = args->trans;
2431 dp = args->dp;
2432 w = args->whichfork;
2433 ASSERT(w == XFS_DATA_FORK);
2434 mp = dp->i_mount;
2435 lastoff = args->geo->freeblk;
2436 error = xfs_bmap_last_before(tp, dp, &lastoff, w);
2437 if (error)
2438 return error;
2439 if (XFS_IS_CORRUPT(mp, lastoff == 0)) {
2440 xfs_da_mark_sick(args);
2441 return -EFSCORRUPTED;
2442 }
2443 /*
2444 * Read the last block in the btree space.
2445 */
2446 last_blkno = (xfs_dablk_t)lastoff - args->geo->fsbcount;
2447 error = xfs_da3_node_read(tp, dp, last_blkno, &last_buf, w);
2448 if (error)
2449 return error;
2450 fa = xfs_da3_header_check(last_buf, args->owner);
2451 if (fa) {
2452 __xfs_buf_mark_corrupt(last_buf, fa);
2453 xfs_trans_brelse(tp, last_buf);
2454 xfs_da_mark_sick(args);
2455 return -EFSCORRUPTED;
2456 }
2457
2458 /*
2459 * Copy the last block into the dead buffer and log it.
2460 */
2461 xfs_da_buf_copy(dead_buf, last_buf, args->geo->blksize);
2462 xfs_trans_log_buf(tp, dead_buf, 0, args->geo->blksize - 1);
2463 dead_info = dead_buf->b_addr;
2464
2465 /*
2466 * Get values from the moved block.
2467 */
2468 if (dead_info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
2469 dead_info->magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC)) {
2470 struct xfs_dir3_icleaf_hdr leafhdr;
2471 struct xfs_dir2_leaf_entry *ents;
2472
2473 dead_leaf2 = (xfs_dir2_leaf_t *)dead_info;
2474 xfs_dir2_leaf_hdr_from_disk(dp->i_mount, &leafhdr,
2475 dead_leaf2);
2476 ents = leafhdr.ents;
2477 dead_level = 0;
2478 dead_hash = be32_to_cpu(ents[leafhdr.count - 1].hashval);
2479 } else {
2480 struct xfs_da3_icnode_hdr deadhdr;
2481
2482 dead_node = (xfs_da_intnode_t *)dead_info;
2483 xfs_da3_node_hdr_from_disk(dp->i_mount, &deadhdr, dead_node);
2484 btree = deadhdr.btree;
2485 dead_level = deadhdr.level;
2486 dead_hash = be32_to_cpu(btree[deadhdr.count - 1].hashval);
2487 }
2488 sib_buf = par_buf = NULL;
2489 /*
2490 * If the moved block has a left sibling, fix up the pointers.
2491 */
2492 if ((sib_blkno = be32_to_cpu(dead_info->back))) {
2493 error = xfs_da3_node_read(tp, dp, sib_blkno, &sib_buf, w);
2494 if (error)
2495 goto done;
2496 fa = xfs_da3_header_check(sib_buf, args->owner);
2497 if (fa) {
2498 __xfs_buf_mark_corrupt(sib_buf, fa);
2499 xfs_da_mark_sick(args);
2500 error = -EFSCORRUPTED;
2501 goto done;
2502 }
2503 sib_info = sib_buf->b_addr;
2504 if (XFS_IS_CORRUPT(mp,
2505 be32_to_cpu(sib_info->forw) != last_blkno ||
2506 sib_info->magic != dead_info->magic)) {
2507 xfs_da_mark_sick(args);
2508 error = -EFSCORRUPTED;
2509 goto done;
2510 }
2511 sib_info->forw = cpu_to_be32(dead_blkno);
2512 xfs_trans_log_buf(tp, sib_buf,
2513 XFS_DA_LOGRANGE(sib_info, &sib_info->forw,
2514 sizeof(sib_info->forw)));
2515 sib_buf = NULL;
2516 }
2517 /*
2518 * If the moved block has a right sibling, fix up the pointers.
2519 */
2520 if ((sib_blkno = be32_to_cpu(dead_info->forw))) {
2521 error = xfs_da3_node_read(tp, dp, sib_blkno, &sib_buf, w);
2522 if (error)
2523 goto done;
2524 fa = xfs_da3_header_check(sib_buf, args->owner);
2525 if (fa) {
2526 __xfs_buf_mark_corrupt(sib_buf, fa);
2527 xfs_da_mark_sick(args);
2528 error = -EFSCORRUPTED;
2529 goto done;
2530 }
2531 sib_info = sib_buf->b_addr;
2532 if (XFS_IS_CORRUPT(mp,
2533 be32_to_cpu(sib_info->back) != last_blkno ||
2534 sib_info->magic != dead_info->magic)) {
2535 xfs_da_mark_sick(args);
2536 error = -EFSCORRUPTED;
2537 goto done;
2538 }
2539 sib_info->back = cpu_to_be32(dead_blkno);
2540 xfs_trans_log_buf(tp, sib_buf,
2541 XFS_DA_LOGRANGE(sib_info, &sib_info->back,
2542 sizeof(sib_info->back)));
2543 sib_buf = NULL;
2544 }
2545 par_blkno = args->geo->leafblk;
2546 level = -1;
2547 /*
2548 * Walk down the tree looking for the parent of the moved block.
2549 */
2550 for (;;) {
2551 error = xfs_da3_node_read(tp, dp, par_blkno, &par_buf, w);
2552 if (error)
2553 goto done;
2554 fa = xfs_da3_node_header_check(par_buf, args->owner);
2555 if (fa) {
2556 __xfs_buf_mark_corrupt(par_buf, fa);
2557 xfs_da_mark_sick(args);
2558 error = -EFSCORRUPTED;
2559 goto done;
2560 }
2561 par_node = par_buf->b_addr;
2562 xfs_da3_node_hdr_from_disk(dp->i_mount, &par_hdr, par_node);
2563 if (XFS_IS_CORRUPT(mp,
2564 level >= 0 && level != par_hdr.level + 1)) {
2565 xfs_da_mark_sick(args);
2566 error = -EFSCORRUPTED;
2567 goto done;
2568 }
2569 level = par_hdr.level;
2570 btree = par_hdr.btree;
2571 for (entno = 0;
2572 entno < par_hdr.count &&
2573 be32_to_cpu(btree[entno].hashval) < dead_hash;
2574 entno++)
2575 continue;
2576 if (XFS_IS_CORRUPT(mp, entno == par_hdr.count)) {
2577 xfs_da_mark_sick(args);
2578 error = -EFSCORRUPTED;
2579 goto done;
2580 }
2581 par_blkno = be32_to_cpu(btree[entno].before);
2582 if (level == dead_level + 1)
2583 break;
2584 xfs_trans_brelse(tp, par_buf);
2585 par_buf = NULL;
2586 }
2587 /*
2588 * We're in the right parent block.
2589 * Look for the right entry.
2590 */
2591 for (;;) {
2592 for (;
2593 entno < par_hdr.count &&
2594 be32_to_cpu(btree[entno].before) != last_blkno;
2595 entno++)
2596 continue;
2597 if (entno < par_hdr.count)
2598 break;
2599 par_blkno = par_hdr.forw;
2600 xfs_trans_brelse(tp, par_buf);
2601 par_buf = NULL;
2602 if (XFS_IS_CORRUPT(mp, par_blkno == 0)) {
2603 xfs_da_mark_sick(args);
2604 error = -EFSCORRUPTED;
2605 goto done;
2606 }
2607 error = xfs_da3_node_read(tp, dp, par_blkno, &par_buf, w);
2608 if (error)
2609 goto done;
2610 fa = xfs_da3_node_header_check(par_buf, args->owner);
2611 if (fa) {
2612 __xfs_buf_mark_corrupt(par_buf, fa);
2613 xfs_da_mark_sick(args);
2614 error = -EFSCORRUPTED;
2615 goto done;
2616 }
2617 par_node = par_buf->b_addr;
2618 xfs_da3_node_hdr_from_disk(dp->i_mount, &par_hdr, par_node);
2619 if (XFS_IS_CORRUPT(mp, par_hdr.level != level)) {
2620 xfs_da_mark_sick(args);
2621 error = -EFSCORRUPTED;
2622 goto done;
2623 }
2624 btree = par_hdr.btree;
2625 entno = 0;
2626 }
2627 /*
2628 * Update the parent entry pointing to the moved block.
2629 */
2630 btree[entno].before = cpu_to_be32(dead_blkno);
2631 xfs_trans_log_buf(tp, par_buf,
2632 XFS_DA_LOGRANGE(par_node, &btree[entno].before,
2633 sizeof(btree[entno].before)));
2634 *dead_blknop = last_blkno;
2635 *dead_bufp = last_buf;
2636 return 0;
2637 done:
2638 if (par_buf)
2639 xfs_trans_brelse(tp, par_buf);
2640 if (sib_buf)
2641 xfs_trans_brelse(tp, sib_buf);
2642 xfs_trans_brelse(tp, last_buf);
2643 return error;
2644 }
2645
2646 /*
2647 * Remove a btree block from a directory or attribute.
2648 */
2649 int
2650 xfs_da_shrink_inode(
2651 struct xfs_da_args *args,
2652 xfs_dablk_t dead_blkno,
2653 struct xfs_buf *dead_buf)
2654 {
2655 struct xfs_inode *dp;
2656 int done, error, w, count;
2657 struct xfs_trans *tp;
2658
2659 trace_xfs_da_shrink_inode(args);
2660
2661 dp = args->dp;
2662 w = args->whichfork;
2663 tp = args->trans;
2664 count = args->geo->fsbcount;
2665 for (;;) {
2666 /*
2667 * Remove extents. If we get ENOSPC for a dir we have to move
2668 * the last block to the place we want to kill.
2669 */
2670 error = xfs_bunmapi(tp, dp, dead_blkno, count,
2671 xfs_bmapi_aflag(w), 0, &done);
2672 if (error == -ENOSPC) {
2673 if (w != XFS_DATA_FORK)
2674 break;
2675 error = xfs_da3_swap_lastblock(args, &dead_blkno,
2676 &dead_buf);
2677 if (error)
2678 break;
2679 } else {
2680 break;
2681 }
2682 }
2683 xfs_trans_binval(tp, dead_buf);
2684 return error;
2685 }
2686
2687 static int
2688 xfs_dabuf_map(
2689 struct xfs_inode *dp,
2690 xfs_dablk_t bno,
2691 unsigned int flags,
2692 int whichfork,
2693 struct xfs_buf_map **mapp,
2694 int *nmaps)
2695 {
2696 struct xfs_mount *mp = dp->i_mount;
2697 int nfsb = xfs_dabuf_nfsb(mp, whichfork);
2698 struct xfs_bmbt_irec irec, *irecs = &irec;
2699 struct xfs_buf_map *map = *mapp;
2700 xfs_fileoff_t off = bno;
2701 int error = 0, nirecs, i;
2702
2703 if (nfsb > 1)
2704 irecs = kzalloc(sizeof(irec) * nfsb,
2705 GFP_KERNEL | __GFP_NOLOCKDEP | __GFP_NOFAIL);
2706
2707 nirecs = nfsb;
2708 error = xfs_bmapi_read(dp, bno, nfsb, irecs, &nirecs,
2709 xfs_bmapi_aflag(whichfork));
2710 if (error)
2711 goto out_free_irecs;
2712
2713 /*
2714 * Use the caller provided map for the single map case, else allocate a
2715 * larger one that needs to be free by the caller.
2716 */
2717 if (nirecs > 1) {
2718 map = kzalloc(nirecs * sizeof(struct xfs_buf_map),
2719 GFP_KERNEL | __GFP_NOLOCKDEP | __GFP_NOFAIL);
2720 if (!map) {
2721 error = -ENOMEM;
2722 goto out_free_irecs;
2723 }
2724 *mapp = map;
2725 }
2726
2727 for (i = 0; i < nirecs; i++) {
2728 if (irecs[i].br_startblock == HOLESTARTBLOCK ||
2729 irecs[i].br_startblock == DELAYSTARTBLOCK)
2730 goto invalid_mapping;
2731 if (off != irecs[i].br_startoff)
2732 goto invalid_mapping;
2733
2734 map[i].bm_bn = XFS_FSB_TO_DADDR(mp, irecs[i].br_startblock);
2735 map[i].bm_len = XFS_FSB_TO_BB(mp, irecs[i].br_blockcount);
2736 off += irecs[i].br_blockcount;
2737 }
2738
2739 if (off != bno + nfsb)
2740 goto invalid_mapping;
2741
2742 *nmaps = nirecs;
2743 out_free_irecs:
2744 if (irecs != &irec)
2745 kfree(irecs);
2746 return error;
2747
2748 invalid_mapping:
2749 /* Caller ok with no mapping. */
2750 if (XFS_IS_CORRUPT(mp, !(flags & XFS_DABUF_MAP_HOLE_OK))) {
2751 xfs_dirattr_mark_sick(dp, whichfork);
2752 error = -EFSCORRUPTED;
2753 if (xfs_error_level >= XFS_ERRLEVEL_LOW) {
2754 xfs_alert(mp, "%s: bno %u inode %llu",
2755 __func__, bno, dp->i_ino);
2756
2757 for (i = 0; i < nirecs; i++) {
2758 xfs_alert(mp,
2759 "[%02d] br_startoff %lld br_startblock %lld br_blockcount %lld br_state %d",
2760 i, irecs[i].br_startoff,
2761 irecs[i].br_startblock,
2762 irecs[i].br_blockcount,
2763 irecs[i].br_state);
2764 }
2765 }
2766 } else {
2767 *nmaps = 0;
2768 }
2769 goto out_free_irecs;
2770 }
2771
2772 /*
2773 * Get a buffer for the dir/attr block.
2774 */
2775 int
2776 xfs_da_get_buf(
2777 struct xfs_trans *tp,
2778 struct xfs_inode *dp,
2779 xfs_dablk_t bno,
2780 struct xfs_buf **bpp,
2781 int whichfork)
2782 {
2783 struct xfs_mount *mp = dp->i_mount;
2784 struct xfs_buf *bp;
2785 struct xfs_buf_map map, *mapp = ↦
2786 int nmap = 1;
2787 int error;
2788
2789 *bpp = NULL;
2790 error = xfs_dabuf_map(dp, bno, 0, whichfork, &mapp, &nmap);
2791 if (error || nmap == 0)
2792 goto out_free;
2793
2794 error = xfs_trans_get_buf_map(tp, mp->m_ddev_targp, mapp, nmap, 0, &bp);
2795 if (error)
2796 goto out_free;
2797
2798 *bpp = bp;
2799
2800 out_free:
2801 if (mapp != &map)
2802 kfree(mapp);
2803
2804 return error;
2805 }
2806
2807 /*
2808 * Get a buffer for the dir/attr block, fill in the contents.
2809 */
2810 int
2811 xfs_da_read_buf(
2812 struct xfs_trans *tp,
2813 struct xfs_inode *dp,
2814 xfs_dablk_t bno,
2815 unsigned int flags,
2816 struct xfs_buf **bpp,
2817 int whichfork,
2818 const struct xfs_buf_ops *ops)
2819 {
2820 struct xfs_mount *mp = dp->i_mount;
2821 struct xfs_buf *bp;
2822 struct xfs_buf_map map, *mapp = ↦
2823 int nmap = 1;
2824 int error;
2825
2826 *bpp = NULL;
2827 error = xfs_dabuf_map(dp, bno, flags, whichfork, &mapp, &nmap);
2828 if (error || !nmap)
2829 goto out_free;
2830
2831 error = xfs_trans_read_buf_map(mp, tp, mp->m_ddev_targp, mapp, nmap, 0,
2832 &bp, ops);
2833 if (xfs_metadata_is_sick(error))
2834 xfs_dirattr_mark_sick(dp, whichfork);
2835 if (error)
2836 goto out_free;
2837
2838 if (whichfork == XFS_ATTR_FORK)
2839 xfs_buf_set_ref(bp, XFS_ATTR_BTREE_REF);
2840 else
2841 xfs_buf_set_ref(bp, XFS_DIR_BTREE_REF);
2842 *bpp = bp;
2843 out_free:
2844 if (mapp != &map)
2845 kfree(mapp);
2846
2847 return error;
2848 }
2849
2850 /*
2851 * Readahead the dir/attr block.
2852 */
2853 int
2854 xfs_da_reada_buf(
2855 struct xfs_inode *dp,
2856 xfs_dablk_t bno,
2857 unsigned int flags,
2858 int whichfork,
2859 const struct xfs_buf_ops *ops)
2860 {
2861 struct xfs_buf_map map;
2862 struct xfs_buf_map *mapp;
2863 int nmap;
2864 int error;
2865
2866 mapp = ↦
2867 nmap = 1;
2868 error = xfs_dabuf_map(dp, bno, flags, whichfork, &mapp, &nmap);
2869 if (error || !nmap)
2870 goto out_free;
2871
2872 xfs_buf_readahead_map(dp->i_mount->m_ddev_targp, mapp, nmap, ops);
2873
2874 out_free:
2875 if (mapp != &map)
2876 kfree(mapp);
2877
2878 return error;
2879 }
2880
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.