?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Linux v2.6.16-rc1
[linux-2.6/next.git] / fs / xfs / xfs_da_btree.c
blob473671fa5c13e1cff88cd4ad075ceecbf1a70ec6
1 /*
2 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
3 * All Rights Reserved.
4 *
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as
7 * published by the Free Software Foundation.
8 *
9 * This program is distributed in the hope that it would be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
13 *
14 * You should have received a copy of the GNU General Public License
15 * along with this program; if not, write the Free Software Foundation,
16 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
17 */
18 #include "xfs.h"
19 #include "xfs_fs.h"
20 #include "xfs_types.h"
21 #include "xfs_bit.h"
22 #include "xfs_log.h"
23 #include "xfs_inum.h"
24 #include "xfs_trans.h"
25 #include "xfs_sb.h"
26 #include "xfs_ag.h"
27 #include "xfs_dir.h"
28 #include "xfs_dir2.h"
29 #include "xfs_dmapi.h"
30 #include "xfs_mount.h"
31 #include "xfs_da_btree.h"
32 #include "xfs_bmap_btree.h"
33 #include "xfs_alloc_btree.h"
34 #include "xfs_ialloc_btree.h"
35 #include "xfs_dir_sf.h"
36 #include "xfs_dir2_sf.h"
37 #include "xfs_attr_sf.h"
38 #include "xfs_dinode.h"
39 #include "xfs_inode.h"
40 #include "xfs_inode_item.h"
41 #include "xfs_alloc.h"
42 #include "xfs_btree.h"
43 #include "xfs_bmap.h"
44 #include "xfs_attr.h"
45 #include "xfs_attr_leaf.h"
46 #include "xfs_dir_leaf.h"
47 #include "xfs_dir2_data.h"
48 #include "xfs_dir2_leaf.h"
49 #include "xfs_dir2_block.h"
50 #include "xfs_dir2_node.h"
51 #include "xfs_error.h"
52
53 /*
54 * xfs_da_btree.c
55 *
56 * Routines to implement directories as Btrees of hashed names.
57 */
58
59 /*========================================================================
60 * Function prototypes for the kernel.
61 *========================================================================*/
62
63 /*
64 * Routines used for growing the Btree.
65 */
66 STATIC int xfs_da_root_split(xfs_da_state_t *state,
67 xfs_da_state_blk_t *existing_root,
68 xfs_da_state_blk_t *new_child);
69 STATIC int xfs_da_node_split(xfs_da_state_t *state,
70 xfs_da_state_blk_t *existing_blk,
71 xfs_da_state_blk_t *split_blk,
72 xfs_da_state_blk_t *blk_to_add,
73 int treelevel,
74 int *result);
75 STATIC void xfs_da_node_rebalance(xfs_da_state_t *state,
76 xfs_da_state_blk_t *node_blk_1,
77 xfs_da_state_blk_t *node_blk_2);
78 STATIC void xfs_da_node_add(xfs_da_state_t *state,
79 xfs_da_state_blk_t *old_node_blk,
80 xfs_da_state_blk_t *new_node_blk);
81
82 /*
83 * Routines used for shrinking the Btree.
84 */
85 STATIC int xfs_da_root_join(xfs_da_state_t *state,
86 xfs_da_state_blk_t *root_blk);
87 STATIC int xfs_da_node_toosmall(xfs_da_state_t *state, int *retval);
88 STATIC void xfs_da_node_remove(xfs_da_state_t *state,
89 xfs_da_state_blk_t *drop_blk);
90 STATIC void xfs_da_node_unbalance(xfs_da_state_t *state,
91 xfs_da_state_blk_t *src_node_blk,
92 xfs_da_state_blk_t *dst_node_blk);
93
94 /*
95 * Utility routines.
96 */
97 STATIC uint xfs_da_node_lasthash(xfs_dabuf_t *bp, int *count);
98 STATIC int xfs_da_node_order(xfs_dabuf_t *node1_bp, xfs_dabuf_t *node2_bp);
99 STATIC xfs_dabuf_t *xfs_da_buf_make(int nbuf, xfs_buf_t **bps, inst_t *ra);
100 STATIC int xfs_da_blk_unlink(xfs_da_state_t *state,
101 xfs_da_state_blk_t *drop_blk,
102 xfs_da_state_blk_t *save_blk);
103 STATIC void xfs_da_state_kill_altpath(xfs_da_state_t *state);
104
105 /*========================================================================
106 * Routines used for growing the Btree.
107 *========================================================================*/
108
109 /*
110 * Create the initial contents of an intermediate node.
111 */
112 int
113 xfs_da_node_create(xfs_da_args_t *args, xfs_dablk_t blkno, int level,
114 xfs_dabuf_t **bpp, int whichfork)
115 {
116 xfs_da_intnode_t *node;
117 xfs_dabuf_t *bp;
118 int error;
119 xfs_trans_t *tp;
120
121 tp = args->trans;
122 error = xfs_da_get_buf(tp, args->dp, blkno, -1, &bp, whichfork);
123 if (error)
124 return(error);
125 ASSERT(bp != NULL);
126 node = bp->data;
127 node->hdr.info.forw = 0;
128 node->hdr.info.back = 0;
129 INT_SET(node->hdr.info.magic, ARCH_CONVERT, XFS_DA_NODE_MAGIC);
130 node->hdr.info.pad = 0;
131 node->hdr.count = 0;
132 INT_SET(node->hdr.level, ARCH_CONVERT, level);
133
134 xfs_da_log_buf(tp, bp,
135 XFS_DA_LOGRANGE(node, &node->hdr, sizeof(node->hdr)));
136
137 *bpp = bp;
138 return(0);
139 }
140
141 /*
142 * Split a leaf node, rebalance, then possibly split
143 * intermediate nodes, rebalance, etc.
144 */
145 int /* error */
146 xfs_da_split(xfs_da_state_t *state)
147 {
148 xfs_da_state_blk_t *oldblk, *newblk, *addblk;
149 xfs_da_intnode_t *node;
150 xfs_dabuf_t *bp;
151 int max, action, error, i;
152
153 /*
154 * Walk back up the tree splitting/inserting/adjusting as necessary.
155 * If we need to insert and there isn't room, split the node, then
156 * decide which fragment to insert the new block from below into.
157 * Note that we may split the root this way, but we need more fixup.
158 */
159 max = state->path.active - 1;
160 ASSERT((max >= 0) && (max < XFS_DA_NODE_MAXDEPTH));
161 ASSERT(state->path.blk[max].magic == XFS_ATTR_LEAF_MAGIC ||
162 state->path.blk[max].magic == XFS_DIRX_LEAF_MAGIC(state->mp));
163
164 addblk = &state->path.blk[max]; /* initial dummy value */
165 for (i = max; (i >= 0) && addblk; state->path.active--, i--) {
166 oldblk = &state->path.blk[i];
167 newblk = &state->altpath.blk[i];
168
169 /*
170 * If a leaf node then
171 * Allocate a new leaf node, then rebalance across them.
172 * else if an intermediate node then
173 * We split on the last layer, must we split the node?
174 */
175 switch (oldblk->magic) {
176 case XFS_ATTR_LEAF_MAGIC:
177 error = xfs_attr_leaf_split(state, oldblk, newblk);
178 if ((error != 0) && (error != ENOSPC)) {
179 return(error); /* GROT: attr is inconsistent */
180 }
181 if (!error) {
182 addblk = newblk;
183 break;
184 }
185 /*
186 * Entry wouldn't fit, split the leaf again.
187 */
188 state->extravalid = 1;
189 if (state->inleaf) {
190 state->extraafter = 0; /* before newblk */
191 error = xfs_attr_leaf_split(state, oldblk,
192 &state->extrablk);
193 } else {
194 state->extraafter = 1; /* after newblk */
195 error = xfs_attr_leaf_split(state, newblk,
196 &state->extrablk);
197 }
198 if (error)
199 return(error); /* GROT: attr inconsistent */
200 addblk = newblk;
201 break;
202 case XFS_DIR_LEAF_MAGIC:
203 ASSERT(XFS_DIR_IS_V1(state->mp));
204 error = xfs_dir_leaf_split(state, oldblk, newblk);
205 if ((error != 0) && (error != ENOSPC)) {
206 return(error); /* GROT: dir is inconsistent */
207 }
208 if (!error) {
209 addblk = newblk;
210 break;
211 }
212 /*
213 * Entry wouldn't fit, split the leaf again.
214 */
215 state->extravalid = 1;
216 if (state->inleaf) {
217 state->extraafter = 0; /* before newblk */
218 error = xfs_dir_leaf_split(state, oldblk,
219 &state->extrablk);
220 if (error)
221 return(error); /* GROT: dir incon. */
222 addblk = newblk;
223 } else {
224 state->extraafter = 1; /* after newblk */
225 error = xfs_dir_leaf_split(state, newblk,
226 &state->extrablk);
227 if (error)
228 return(error); /* GROT: dir incon. */
229 addblk = newblk;
230 }
231 break;
232 case XFS_DIR2_LEAFN_MAGIC:
233 ASSERT(XFS_DIR_IS_V2(state->mp));
234 error = xfs_dir2_leafn_split(state, oldblk, newblk);
235 if (error)
236 return error;
237 addblk = newblk;
238 break;
239 case XFS_DA_NODE_MAGIC:
240 error = xfs_da_node_split(state, oldblk, newblk, addblk,
241 max - i, &action);
242 xfs_da_buf_done(addblk->bp);
243 addblk->bp = NULL;
244 if (error)
245 return(error); /* GROT: dir is inconsistent */
246 /*
247 * Record the newly split block for the next time thru?
248 */
249 if (action)
250 addblk = newblk;
251 else
252 addblk = NULL;
253 break;
254 }
255
256 /*
257 * Update the btree to show the new hashval for this child.
258 */
259 xfs_da_fixhashpath(state, &state->path);
260 /*
261 * If we won't need this block again, it's getting dropped
262 * from the active path by the loop control, so we need
263 * to mark it done now.
264 */
265 if (i > 0 || !addblk)
266 xfs_da_buf_done(oldblk->bp);
267 }
268 if (!addblk)
269 return(0);
270
271 /*
272 * Split the root node.
273 */
274 ASSERT(state->path.active == 0);
275 oldblk = &state->path.blk[0];
276 error = xfs_da_root_split(state, oldblk, addblk);
277 if (error) {
278 xfs_da_buf_done(oldblk->bp);
279 xfs_da_buf_done(addblk->bp);
280 addblk->bp = NULL;
281 return(error); /* GROT: dir is inconsistent */
282 }
283
284 /*
285 * Update pointers to the node which used to be block 0 and
286 * just got bumped because of the addition of a new root node.
287 * There might be three blocks involved if a double split occurred,
288 * and the original block 0 could be at any position in the list.
289 */
290
291 node = oldblk->bp->data;
292 if (node->hdr.info.forw) {
293 if (INT_GET(node->hdr.info.forw, ARCH_CONVERT) == addblk->blkno) {
294 bp = addblk->bp;
295 } else {
296 ASSERT(state->extravalid);
297 bp = state->extrablk.bp;
298 }
299 node = bp->data;
300 INT_SET(node->hdr.info.back, ARCH_CONVERT, oldblk->blkno);
301 xfs_da_log_buf(state->args->trans, bp,
302 XFS_DA_LOGRANGE(node, &node->hdr.info,
303 sizeof(node->hdr.info)));
304 }
305 node = oldblk->bp->data;
306 if (INT_GET(node->hdr.info.back, ARCH_CONVERT)) {
307 if (INT_GET(node->hdr.info.back, ARCH_CONVERT) == addblk->blkno) {
308 bp = addblk->bp;
309 } else {
310 ASSERT(state->extravalid);
311 bp = state->extrablk.bp;
312 }
313 node = bp->data;
314 INT_SET(node->hdr.info.forw, ARCH_CONVERT, oldblk->blkno);
315 xfs_da_log_buf(state->args->trans, bp,
316 XFS_DA_LOGRANGE(node, &node->hdr.info,
317 sizeof(node->hdr.info)));
318 }
319 xfs_da_buf_done(oldblk->bp);
320 xfs_da_buf_done(addblk->bp);
321 addblk->bp = NULL;
322 return(0);
323 }
324
325 /*
326 * Split the root. We have to create a new root and point to the two
327 * parts (the split old root) that we just created. Copy block zero to
328 * the EOF, extending the inode in process.
329 */
330 STATIC int /* error */
331 xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
332 xfs_da_state_blk_t *blk2)
333 {
334 xfs_da_intnode_t *node, *oldroot;
335 xfs_da_args_t *args;
336 xfs_dablk_t blkno;
337 xfs_dabuf_t *bp;
338 int error, size;
339 xfs_inode_t *dp;
340 xfs_trans_t *tp;
341 xfs_mount_t *mp;
342 xfs_dir2_leaf_t *leaf;
343
344 /*
345 * Copy the existing (incorrect) block from the root node position
346 * to a free space somewhere.
347 */
348 args = state->args;
349 ASSERT(args != NULL);
350 error = xfs_da_grow_inode(args, &blkno);
351 if (error)
352 return(error);
353 dp = args->dp;
354 tp = args->trans;
355 mp = state->mp;
356 error = xfs_da_get_buf(tp, dp, blkno, -1, &bp, args->whichfork);
357 if (error)
358 return(error);
359 ASSERT(bp != NULL);
360 node = bp->data;
361 oldroot = blk1->bp->data;
362 if (INT_GET(oldroot->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC) {
363 size = (int)((char *)&oldroot->btree[INT_GET(oldroot->hdr.count, ARCH_CONVERT)] -
364 (char *)oldroot);
365 } else {
366 ASSERT(XFS_DIR_IS_V2(mp));
367 ASSERT(INT_GET(oldroot->hdr.info.magic, ARCH_CONVERT) == XFS_DIR2_LEAFN_MAGIC);
368 leaf = (xfs_dir2_leaf_t *)oldroot;
369 size = (int)((char *)&leaf->ents[INT_GET(leaf->hdr.count, ARCH_CONVERT)] -
370 (char *)leaf);
371 }
372 memcpy(node, oldroot, size);
373 xfs_da_log_buf(tp, bp, 0, size - 1);
374 xfs_da_buf_done(blk1->bp);
375 blk1->bp = bp;
376 blk1->blkno = blkno;
377
378 /*
379 * Set up the new root node.
380 */
381 error = xfs_da_node_create(args,
382 args->whichfork == XFS_DATA_FORK &&
383 XFS_DIR_IS_V2(mp) ? mp->m_dirleafblk : 0,
384 INT_GET(node->hdr.level, ARCH_CONVERT) + 1, &bp, args->whichfork);
385 if (error)
386 return(error);
387 node = bp->data;
388 INT_SET(node->btree[0].hashval, ARCH_CONVERT, blk1->hashval);
389 INT_SET(node->btree[0].before, ARCH_CONVERT, blk1->blkno);
390 INT_SET(node->btree[1].hashval, ARCH_CONVERT, blk2->hashval);
391 INT_SET(node->btree[1].before, ARCH_CONVERT, blk2->blkno);
392 INT_SET(node->hdr.count, ARCH_CONVERT, 2);
393
394 #ifdef DEBUG
395 if (INT_GET(oldroot->hdr.info.magic, ARCH_CONVERT) == XFS_DIR2_LEAFN_MAGIC) {
396 ASSERT(blk1->blkno >= mp->m_dirleafblk &&
397 blk1->blkno < mp->m_dirfreeblk);
398 ASSERT(blk2->blkno >= mp->m_dirleafblk &&
399 blk2->blkno < mp->m_dirfreeblk);
400 }
401 #endif
402
403 /* Header is already logged by xfs_da_node_create */
404 xfs_da_log_buf(tp, bp,
405 XFS_DA_LOGRANGE(node, node->btree,
406 sizeof(xfs_da_node_entry_t) * 2));
407 xfs_da_buf_done(bp);
408
409 return(0);
410 }
411
412 /*
413 * Split the node, rebalance, then add the new entry.
414 */
415 STATIC int /* error */
416 xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
417 xfs_da_state_blk_t *newblk,
418 xfs_da_state_blk_t *addblk,
419 int treelevel, int *result)
420 {
421 xfs_da_intnode_t *node;
422 xfs_dablk_t blkno;
423 int newcount, error;
424 int useextra;
425
426 node = oldblk->bp->data;
427 ASSERT(INT_GET(node->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
428
429 /*
430 * With V2 the extra block is data or freespace.
431 */
432 useextra = state->extravalid && (XFS_DIR_IS_V1(state->mp) ||
433 state->args->whichfork == XFS_ATTR_FORK);
434 newcount = 1 + useextra;
435 /*
436 * Do we have to split the node?
437 */
438 if ((INT_GET(node->hdr.count, ARCH_CONVERT) + newcount) > state->node_ents) {
439 /*
440 * Allocate a new node, add to the doubly linked chain of
441 * nodes, then move some of our excess entries into it.
442 */
443 error = xfs_da_grow_inode(state->args, &blkno);
444 if (error)
445 return(error); /* GROT: dir is inconsistent */
446
447 error = xfs_da_node_create(state->args, blkno, treelevel,
448 &newblk->bp, state->args->whichfork);
449 if (error)
450 return(error); /* GROT: dir is inconsistent */
451 newblk->blkno = blkno;
452 newblk->magic = XFS_DA_NODE_MAGIC;
453 xfs_da_node_rebalance(state, oldblk, newblk);
454 error = xfs_da_blk_link(state, oldblk, newblk);
455 if (error)
456 return(error);
457 *result = 1;
458 } else {
459 *result = 0;
460 }
461
462 /*
463 * Insert the new entry(s) into the correct block
464 * (updating last hashval in the process).
465 *
466 * xfs_da_node_add() inserts BEFORE the given index,
467 * and as a result of using node_lookup_int() we always
468 * point to a valid entry (not after one), but a split
469 * operation always results in a new block whose hashvals
470 * FOLLOW the current block.
471 *
472 * If we had double-split op below us, then add the extra block too.
473 */
474 node = oldblk->bp->data;
475 if (oldblk->index <= INT_GET(node->hdr.count, ARCH_CONVERT)) {
476 oldblk->index++;
477 xfs_da_node_add(state, oldblk, addblk);
478 if (useextra) {
479 if (state->extraafter)
480 oldblk->index++;
481 xfs_da_node_add(state, oldblk, &state->extrablk);
482 state->extravalid = 0;
483 }
484 } else {
485 newblk->index++;
486 xfs_da_node_add(state, newblk, addblk);
487 if (useextra) {
488 if (state->extraafter)
489 newblk->index++;
490 xfs_da_node_add(state, newblk, &state->extrablk);
491 state->extravalid = 0;
492 }
493 }
494
495 return(0);
496 }
497
498 /*
499 * Balance the btree elements between two intermediate nodes,
500 * usually one full and one empty.
501 *
502 * NOTE: if blk2 is empty, then it will get the upper half of blk1.
503 */
504 STATIC void
505 xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
506 xfs_da_state_blk_t *blk2)
507 {
508 xfs_da_intnode_t *node1, *node2, *tmpnode;
509 xfs_da_node_entry_t *btree_s, *btree_d;
510 int count, tmp;
511 xfs_trans_t *tp;
512
513 node1 = blk1->bp->data;
514 node2 = blk2->bp->data;
515 /*
516 * Figure out how many entries need to move, and in which direction.
517 * Swap the nodes around if that makes it simpler.
518 */
519 if ((INT_GET(node1->hdr.count, ARCH_CONVERT) > 0) && (INT_GET(node2->hdr.count, ARCH_CONVERT) > 0) &&
520 ((INT_GET(node2->btree[ 0 ].hashval, ARCH_CONVERT) < INT_GET(node1->btree[ 0 ].hashval, ARCH_CONVERT)) ||
521 (INT_GET(node2->btree[ INT_GET(node2->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT) <
522 INT_GET(node1->btree[ INT_GET(node1->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT)))) {
523 tmpnode = node1;
524 node1 = node2;
525 node2 = tmpnode;
526 }
527 ASSERT(INT_GET(node1->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
528 ASSERT(INT_GET(node2->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
529 count = (INT_GET(node1->hdr.count, ARCH_CONVERT) - INT_GET(node2->hdr.count, ARCH_CONVERT)) / 2;
530 if (count == 0)
531 return;
532 tp = state->args->trans;
533 /*
534 * Two cases: high-to-low and low-to-high.
535 */
536 if (count > 0) {
537 /*
538 * Move elements in node2 up to make a hole.
539 */
540 if ((tmp = INT_GET(node2->hdr.count, ARCH_CONVERT)) > 0) {
541 tmp *= (uint)sizeof(xfs_da_node_entry_t);
542 btree_s = &node2->btree[0];
543 btree_d = &node2->btree[count];
544 memmove(btree_d, btree_s, tmp);
545 }
546
547 /*
548 * Move the req'd B-tree elements from high in node1 to
549 * low in node2.
550 */
551 INT_MOD(node2->hdr.count, ARCH_CONVERT, count);
552 tmp = count * (uint)sizeof(xfs_da_node_entry_t);
553 btree_s = &node1->btree[INT_GET(node1->hdr.count, ARCH_CONVERT) - count];
554 btree_d = &node2->btree[0];
555 memcpy(btree_d, btree_s, tmp);
556 INT_MOD(node1->hdr.count, ARCH_CONVERT, -(count));
557
558 } else {
559 /*
560 * Move the req'd B-tree elements from low in node2 to
561 * high in node1.
562 */
563 count = -count;
564 tmp = count * (uint)sizeof(xfs_da_node_entry_t);
565 btree_s = &node2->btree[0];
566 btree_d = &node1->btree[INT_GET(node1->hdr.count, ARCH_CONVERT)];
567 memcpy(btree_d, btree_s, tmp);
568 INT_MOD(node1->hdr.count, ARCH_CONVERT, count);
569 xfs_da_log_buf(tp, blk1->bp,
570 XFS_DA_LOGRANGE(node1, btree_d, tmp));
571
572 /*
573 * Move elements in node2 down to fill the hole.
574 */
575 tmp = INT_GET(node2->hdr.count, ARCH_CONVERT) - count;
576 tmp *= (uint)sizeof(xfs_da_node_entry_t);
577 btree_s = &node2->btree[count];
578 btree_d = &node2->btree[0];
579 memmove(btree_d, btree_s, tmp);
580 INT_MOD(node2->hdr.count, ARCH_CONVERT, -(count));
581 }
582
583 /*
584 * Log header of node 1 and all current bits of node 2.
585 */
586 xfs_da_log_buf(tp, blk1->bp,
587 XFS_DA_LOGRANGE(node1, &node1->hdr, sizeof(node1->hdr)));
588 xfs_da_log_buf(tp, blk2->bp,
589 XFS_DA_LOGRANGE(node2, &node2->hdr,
590 sizeof(node2->hdr) +
591 sizeof(node2->btree[0]) * INT_GET(node2->hdr.count, ARCH_CONVERT)));
592
593 /*
594 * Record the last hashval from each block for upward propagation.
595 * (note: don't use the swapped node pointers)
596 */
597 node1 = blk1->bp->data;
598 node2 = blk2->bp->data;
599 blk1->hashval = INT_GET(node1->btree[ INT_GET(node1->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT);
600 blk2->hashval = INT_GET(node2->btree[ INT_GET(node2->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT);
601
602 /*
603 * Adjust the expected index for insertion.
604 */
605 if (blk1->index >= INT_GET(node1->hdr.count, ARCH_CONVERT)) {
606 blk2->index = blk1->index - INT_GET(node1->hdr.count, ARCH_CONVERT);
607 blk1->index = INT_GET(node1->hdr.count, ARCH_CONVERT) + 1; /* make it invalid */
608 }
609 }
610
611 /*
612 * Add a new entry to an intermediate node.
613 */
614 STATIC void
615 xfs_da_node_add(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
616 xfs_da_state_blk_t *newblk)
617 {
618 xfs_da_intnode_t *node;
619 xfs_da_node_entry_t *btree;
620 int tmp;
621 xfs_mount_t *mp;
622
623 node = oldblk->bp->data;
624 mp = state->mp;
625 ASSERT(INT_GET(node->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
626 ASSERT((oldblk->index >= 0) && (oldblk->index <= INT_GET(node->hdr.count, ARCH_CONVERT)));
627 ASSERT(newblk->blkno != 0);
628 if (state->args->whichfork == XFS_DATA_FORK && XFS_DIR_IS_V2(mp))
629 ASSERT(newblk->blkno >= mp->m_dirleafblk &&
630 newblk->blkno < mp->m_dirfreeblk);
631
632 /*
633 * We may need to make some room before we insert the new node.
634 */
635 tmp = 0;
636 btree = &node->btree[ oldblk->index ];
637 if (oldblk->index < INT_GET(node->hdr.count, ARCH_CONVERT)) {
638 tmp = (INT_GET(node->hdr.count, ARCH_CONVERT) - oldblk->index) * (uint)sizeof(*btree);
639 memmove(btree + 1, btree, tmp);
640 }
641 INT_SET(btree->hashval, ARCH_CONVERT, newblk->hashval);
642 INT_SET(btree->before, ARCH_CONVERT, newblk->blkno);
643 xfs_da_log_buf(state->args->trans, oldblk->bp,
644 XFS_DA_LOGRANGE(node, btree, tmp + sizeof(*btree)));
645 INT_MOD(node->hdr.count, ARCH_CONVERT, +1);
646 xfs_da_log_buf(state->args->trans, oldblk->bp,
647 XFS_DA_LOGRANGE(node, &node->hdr, sizeof(node->hdr)));
648
649 /*
650 * Copy the last hash value from the oldblk to propagate upwards.
651 */
652 oldblk->hashval = INT_GET(node->btree[ INT_GET(node->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT);
653 }
654
655 /*========================================================================
656 * Routines used for shrinking the Btree.
657 *========================================================================*/
658
659 /*
660 * Deallocate an empty leaf node, remove it from its parent,
661 * possibly deallocating that block, etc...
662 */
663 int
664 xfs_da_join(xfs_da_state_t *state)
665 {
666 xfs_da_state_blk_t *drop_blk, *save_blk;
667 int action, error;
668
669 action = 0;
670 drop_blk = &state->path.blk[ state->path.active-1 ];
671 save_blk = &state->altpath.blk[ state->path.active-1 ];
672 ASSERT(state->path.blk[0].magic == XFS_DA_NODE_MAGIC);
673 ASSERT(drop_blk->magic == XFS_ATTR_LEAF_MAGIC ||
674 drop_blk->magic == XFS_DIRX_LEAF_MAGIC(state->mp));
675
676 /*
677 * Walk back up the tree joining/deallocating as necessary.
678 * When we stop dropping blocks, break out.
679 */
680 for ( ; state->path.active >= 2; drop_blk--, save_blk--,
681 state->path.active--) {
682 /*
683 * See if we can combine the block with a neighbor.
684 * (action == 0) => no options, just leave
685 * (action == 1) => coalesce, then unlink
686 * (action == 2) => block empty, unlink it
687 */
688 switch (drop_blk->magic) {
689 case XFS_ATTR_LEAF_MAGIC:
690 error = xfs_attr_leaf_toosmall(state, &action);
691 if (error)
692 return(error);
693 if (action == 0)
694 return(0);
695 xfs_attr_leaf_unbalance(state, drop_blk, save_blk);
696 break;
697 case XFS_DIR_LEAF_MAGIC:
698 ASSERT(XFS_DIR_IS_V1(state->mp));
699 error = xfs_dir_leaf_toosmall(state, &action);
700 if (error)
701 return(error);
702 if (action == 0)
703 return(0);
704 xfs_dir_leaf_unbalance(state, drop_blk, save_blk);
705 break;
706 case XFS_DIR2_LEAFN_MAGIC:
707 ASSERT(XFS_DIR_IS_V2(state->mp));
708 error = xfs_dir2_leafn_toosmall(state, &action);
709 if (error)
710 return error;
711 if (action == 0)
712 return 0;
713 xfs_dir2_leafn_unbalance(state, drop_blk, save_blk);
714 break;
715 case XFS_DA_NODE_MAGIC:
716 /*
717 * Remove the offending node, fixup hashvals,
718 * check for a toosmall neighbor.
719 */
720 xfs_da_node_remove(state, drop_blk);
721 xfs_da_fixhashpath(state, &state->path);
722 error = xfs_da_node_toosmall(state, &action);
723 if (error)
724 return(error);
725 if (action == 0)
726 return 0;
727 xfs_da_node_unbalance(state, drop_blk, save_blk);
728 break;
729 }
730 xfs_da_fixhashpath(state, &state->altpath);
731 error = xfs_da_blk_unlink(state, drop_blk, save_blk);
732 xfs_da_state_kill_altpath(state);
733 if (error)
734 return(error);
735 error = xfs_da_shrink_inode(state->args, drop_blk->blkno,
736 drop_blk->bp);
737 drop_blk->bp = NULL;
738 if (error)
739 return(error);
740 }
741 /*
742 * We joined all the way to the top. If it turns out that
743 * we only have one entry in the root, make the child block
744 * the new root.
745 */
746 xfs_da_node_remove(state, drop_blk);
747 xfs_da_fixhashpath(state, &state->path);
748 error = xfs_da_root_join(state, &state->path.blk[0]);
749 return(error);
750 }
751
752 /*
753 * We have only one entry in the root. Copy the only remaining child of
754 * the old root to block 0 as the new root node.
755 */
756 STATIC int
757 xfs_da_root_join(xfs_da_state_t *state, xfs_da_state_blk_t *root_blk)
758 {
759 xfs_da_intnode_t *oldroot;
760 /* REFERENCED */
761 xfs_da_blkinfo_t *blkinfo;
762 xfs_da_args_t *args;
763 xfs_dablk_t child;
764 xfs_dabuf_t *bp;
765 int error;
766
767 args = state->args;
768 ASSERT(args != NULL);
769 ASSERT(root_blk->magic == XFS_DA_NODE_MAGIC);
770 oldroot = root_blk->bp->data;
771 ASSERT(INT_GET(oldroot->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
772 ASSERT(!oldroot->hdr.info.forw);
773 ASSERT(!oldroot->hdr.info.back);
774
775 /*
776 * If the root has more than one child, then don't do anything.
777 */
778 if (INT_GET(oldroot->hdr.count, ARCH_CONVERT) > 1)
779 return(0);
780
781 /*
782 * Read in the (only) child block, then copy those bytes into
783 * the root block's buffer and free the original child block.
784 */
785 child = INT_GET(oldroot->btree[ 0 ].before, ARCH_CONVERT);
786 ASSERT(child != 0);
787 error = xfs_da_read_buf(args->trans, args->dp, child, -1, &bp,
788 args->whichfork);
789 if (error)
790 return(error);
791 ASSERT(bp != NULL);
792 blkinfo = bp->data;
793 if (INT_GET(oldroot->hdr.level, ARCH_CONVERT) == 1) {
794 ASSERT(INT_GET(blkinfo->magic, ARCH_CONVERT) == XFS_DIRX_LEAF_MAGIC(state->mp) ||
795 INT_GET(blkinfo->magic, ARCH_CONVERT) == XFS_ATTR_LEAF_MAGIC);
796 } else {
797 ASSERT(INT_GET(blkinfo->magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
798 }
799 ASSERT(!blkinfo->forw);
800 ASSERT(!blkinfo->back);
801 memcpy(root_blk->bp->data, bp->data, state->blocksize);
802 xfs_da_log_buf(args->trans, root_blk->bp, 0, state->blocksize - 1);
803 error = xfs_da_shrink_inode(args, child, bp);
804 return(error);
805 }
806
807 /*
808 * Check a node block and its neighbors to see if the block should be
809 * collapsed into one or the other neighbor. Always keep the block
810 * with the smaller block number.
811 * If the current block is over 50% full, don't try to join it, return 0.
812 * If the block is empty, fill in the state structure and return 2.
813 * If it can be collapsed, fill in the state structure and return 1.
814 * If nothing can be done, return 0.
815 */
816 STATIC int
817 xfs_da_node_toosmall(xfs_da_state_t *state, int *action)
818 {
819 xfs_da_intnode_t *node;
820 xfs_da_state_blk_t *blk;
821 xfs_da_blkinfo_t *info;
822 int count, forward, error, retval, i;
823 xfs_dablk_t blkno;
824 xfs_dabuf_t *bp;
825
826 /*
827 * Check for the degenerate case of the block being over 50% full.
828 * If so, it's not worth even looking to see if we might be able
829 * to coalesce with a sibling.
830 */
831 blk = &state->path.blk[ state->path.active-1 ];
832 info = blk->bp->data;
833 ASSERT(INT_GET(info->magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
834 node = (xfs_da_intnode_t *)info;
835 count = INT_GET(node->hdr.count, ARCH_CONVERT);
836 if (count > (state->node_ents >> 1)) {
837 *action = 0; /* blk over 50%, don't try to join */
838 return(0); /* blk over 50%, don't try to join */
839 }
840
841 /*
842 * Check for the degenerate case of the block being empty.
843 * If the block is empty, we'll simply delete it, no need to
844 * coalesce it with a sibling block. We choose (aribtrarily)
845 * to merge with the forward block unless it is NULL.
846 */
847 if (count == 0) {
848 /*
849 * Make altpath point to the block we want to keep and
850 * path point to the block we want to drop (this one).
851 */
852 forward = info->forw;
853 memcpy(&state->altpath, &state->path, sizeof(state->path));
854 error = xfs_da_path_shift(state, &state->altpath, forward,
855 0, &retval);
856 if (error)
857 return(error);
858 if (retval) {
859 *action = 0;
860 } else {
861 *action = 2;
862 }
863 return(0);
864 }
865
866 /*
867 * Examine each sibling block to see if we can coalesce with
868 * at least 25% free space to spare. We need to figure out
869 * whether to merge with the forward or the backward block.
870 * We prefer coalescing with the lower numbered sibling so as
871 * to shrink a directory over time.
872 */
873 /* start with smaller blk num */
874 forward = (INT_GET(info->forw, ARCH_CONVERT)
875 < INT_GET(info->back, ARCH_CONVERT));
876 for (i = 0; i < 2; forward = !forward, i++) {
877 if (forward)
878 blkno = INT_GET(info->forw, ARCH_CONVERT);
879 else
880 blkno = INT_GET(info->back, ARCH_CONVERT);
881 if (blkno == 0)
882 continue;
883 error = xfs_da_read_buf(state->args->trans, state->args->dp,
884 blkno, -1, &bp, state->args->whichfork);
885 if (error)
886 return(error);
887 ASSERT(bp != NULL);
888
889 node = (xfs_da_intnode_t *)info;
890 count = state->node_ents;
891 count -= state->node_ents >> 2;
892 count -= INT_GET(node->hdr.count, ARCH_CONVERT);
893 node = bp->data;
894 ASSERT(INT_GET(node->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
895 count -= INT_GET(node->hdr.count, ARCH_CONVERT);
896 xfs_da_brelse(state->args->trans, bp);
897 if (count >= 0)
898 break; /* fits with at least 25% to spare */
899 }
900 if (i >= 2) {
901 *action = 0;
902 return(0);
903 }
904
905 /*
906 * Make altpath point to the block we want to keep (the lower
907 * numbered block) and path point to the block we want to drop.
908 */
909 memcpy(&state->altpath, &state->path, sizeof(state->path));
910 if (blkno < blk->blkno) {
911 error = xfs_da_path_shift(state, &state->altpath, forward,
912 0, &retval);
913 if (error) {
914 return(error);
915 }
916 if (retval) {
917 *action = 0;
918 return(0);
919 }
920 } else {
921 error = xfs_da_path_shift(state, &state->path, forward,
922 0, &retval);
923 if (error) {
924 return(error);
925 }
926 if (retval) {
927 *action = 0;
928 return(0);
929 }
930 }
931 *action = 1;
932 return(0);
933 }
934
935 /*
936 * Walk back up the tree adjusting hash values as necessary,
937 * when we stop making changes, return.
938 */
939 void
940 xfs_da_fixhashpath(xfs_da_state_t *state, xfs_da_state_path_t *path)
941 {
942 xfs_da_state_blk_t *blk;
943 xfs_da_intnode_t *node;
944 xfs_da_node_entry_t *btree;
945 xfs_dahash_t lasthash=0;
946 int level, count;
947
948 level = path->active-1;
949 blk = &path->blk[ level ];
950 switch (blk->magic) {
951 case XFS_ATTR_LEAF_MAGIC:
952 lasthash = xfs_attr_leaf_lasthash(blk->bp, &count);
953 if (count == 0)
954 return;
955 break;
956 case XFS_DIR_LEAF_MAGIC:
957 ASSERT(XFS_DIR_IS_V1(state->mp));
958 lasthash = xfs_dir_leaf_lasthash(blk->bp, &count);
959 if (count == 0)
960 return;
961 break;
962 case XFS_DIR2_LEAFN_MAGIC:
963 ASSERT(XFS_DIR_IS_V2(state->mp));
964 lasthash = xfs_dir2_leafn_lasthash(blk->bp, &count);
965 if (count == 0)
966 return;
967 break;
968 case XFS_DA_NODE_MAGIC:
969 lasthash = xfs_da_node_lasthash(blk->bp, &count);
970 if (count == 0)
971 return;
972 break;
973 }
974 for (blk--, level--; level >= 0; blk--, level--) {
975 node = blk->bp->data;
976 ASSERT(INT_GET(node->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
977 btree = &node->btree[ blk->index ];
978 if (INT_GET(btree->hashval, ARCH_CONVERT) == lasthash)
979 break;
980 blk->hashval = lasthash;
981 INT_SET(btree->hashval, ARCH_CONVERT, lasthash);
982 xfs_da_log_buf(state->args->trans, blk->bp,
983 XFS_DA_LOGRANGE(node, btree, sizeof(*btree)));
984
985 lasthash = INT_GET(node->btree[ INT_GET(node->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT);
986 }
987 }
988
989 /*
990 * Remove an entry from an intermediate node.
991 */
992 STATIC void
993 xfs_da_node_remove(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk)
994 {
995 xfs_da_intnode_t *node;
996 xfs_da_node_entry_t *btree;
997 int tmp;
998
999 node = drop_blk->bp->data;
1000 ASSERT(drop_blk->index < INT_GET(node->hdr.count, ARCH_CONVERT));
1001 ASSERT(drop_blk->index >= 0);
1002
1003 /*
1004 * Copy over the offending entry, or just zero it out.
1005 */
1006 btree = &node->btree[drop_blk->index];
1007 if (drop_blk->index < (INT_GET(node->hdr.count, ARCH_CONVERT)-1)) {
1008 tmp = INT_GET(node->hdr.count, ARCH_CONVERT) - drop_blk->index - 1;
1009 tmp *= (uint)sizeof(xfs_da_node_entry_t);
1010 memmove(btree, btree + 1, tmp);
1011 xfs_da_log_buf(state->args->trans, drop_blk->bp,
1012 XFS_DA_LOGRANGE(node, btree, tmp));
1013 btree = &node->btree[ INT_GET(node->hdr.count, ARCH_CONVERT)-1 ];
1014 }
1015 memset((char *)btree, 0, sizeof(xfs_da_node_entry_t));
1016 xfs_da_log_buf(state->args->trans, drop_blk->bp,
1017 XFS_DA_LOGRANGE(node, btree, sizeof(*btree)));
1018 INT_MOD(node->hdr.count, ARCH_CONVERT, -1);
1019 xfs_da_log_buf(state->args->trans, drop_blk->bp,
1020 XFS_DA_LOGRANGE(node, &node->hdr, sizeof(node->hdr)));
1021
1022 /*
1023 * Copy the last hash value from the block to propagate upwards.
1024 */
1025 btree--;
1026 drop_blk->hashval = INT_GET(btree->hashval, ARCH_CONVERT);
1027 }
1028
1029 /*
1030 * Unbalance the btree elements between two intermediate nodes,
1031 * move all Btree elements from one node into another.
1032 */
1033 STATIC void
1034 xfs_da_node_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
1035 xfs_da_state_blk_t *save_blk)
1036 {
1037 xfs_da_intnode_t *drop_node, *save_node;
1038 xfs_da_node_entry_t *btree;
1039 int tmp;
1040 xfs_trans_t *tp;
1041
1042 drop_node = drop_blk->bp->data;
1043 save_node = save_blk->bp->data;
1044 ASSERT(INT_GET(drop_node->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
1045 ASSERT(INT_GET(save_node->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
1046 tp = state->args->trans;
1047
1048 /*
1049 * If the dying block has lower hashvals, then move all the
1050 * elements in the remaining block up to make a hole.
1051 */
1052 if ((INT_GET(drop_node->btree[ 0 ].hashval, ARCH_CONVERT) < INT_GET(save_node->btree[ 0 ].hashval, ARCH_CONVERT)) ||
1053 (INT_GET(drop_node->btree[ INT_GET(drop_node->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT) <
1054 INT_GET(save_node->btree[ INT_GET(save_node->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT)))
1055 {
1056 btree = &save_node->btree[ INT_GET(drop_node->hdr.count, ARCH_CONVERT) ];
1057 tmp = INT_GET(save_node->hdr.count, ARCH_CONVERT) * (uint)sizeof(xfs_da_node_entry_t);
1058 memmove(btree, &save_node->btree[0], tmp);
1059 btree = &save_node->btree[0];
1060 xfs_da_log_buf(tp, save_blk->bp,
1061 XFS_DA_LOGRANGE(save_node, btree,
1062 (INT_GET(save_node->hdr.count, ARCH_CONVERT) + INT_GET(drop_node->hdr.count, ARCH_CONVERT)) *
1063 sizeof(xfs_da_node_entry_t)));
1064 } else {
1065 btree = &save_node->btree[ INT_GET(save_node->hdr.count, ARCH_CONVERT) ];
1066 xfs_da_log_buf(tp, save_blk->bp,
1067 XFS_DA_LOGRANGE(save_node, btree,
1068 INT_GET(drop_node->hdr.count, ARCH_CONVERT) *
1069 sizeof(xfs_da_node_entry_t)));
1070 }
1071
1072 /*
1073 * Move all the B-tree elements from drop_blk to save_blk.
1074 */
1075 tmp = INT_GET(drop_node->hdr.count, ARCH_CONVERT) * (uint)sizeof(xfs_da_node_entry_t);
1076 memcpy(btree, &drop_node->btree[0], tmp);
1077 INT_MOD(save_node->hdr.count, ARCH_CONVERT, INT_GET(drop_node->hdr.count, ARCH_CONVERT));
1078
1079 xfs_da_log_buf(tp, save_blk->bp,
1080 XFS_DA_LOGRANGE(save_node, &save_node->hdr,
1081 sizeof(save_node->hdr)));
1082
1083 /*
1084 * Save the last hashval in the remaining block for upward propagation.
1085 */
1086 save_blk->hashval = INT_GET(save_node->btree[ INT_GET(save_node->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT);
1087 }
1088
1089 /*========================================================================
1090 * Routines used for finding things in the Btree.
1091 *========================================================================*/
1092
1093 /*
1094 * Walk down the Btree looking for a particular filename, filling
1095 * in the state structure as we go.
1096 *
1097 * We will set the state structure to point to each of the elements
1098 * in each of the nodes where either the hashval is or should be.
1099 *
1100 * We support duplicate hashval's so for each entry in the current
1101 * node that could contain the desired hashval, descend. This is a
1102 * pruned depth-first tree search.
1103 */
1104 int /* error */
1105 xfs_da_node_lookup_int(xfs_da_state_t *state, int *result)
1106 {
1107 xfs_da_state_blk_t *blk;
1108 xfs_da_blkinfo_t *curr;
1109 xfs_da_intnode_t *node;
1110 xfs_da_node_entry_t *btree;
1111 xfs_dablk_t blkno;
1112 int probe, span, max, error, retval;
1113 xfs_dahash_t hashval;
1114 xfs_da_args_t *args;
1115
1116 args = state->args;
1117
1118 /*
1119 * Descend thru the B-tree searching each level for the right
1120 * node to use, until the right hashval is found.
1121 */
1122 if (args->whichfork == XFS_DATA_FORK && XFS_DIR_IS_V2(state->mp))
1123 blkno = state->mp->m_dirleafblk;
1124 else
1125 blkno = 0;
1126 for (blk = &state->path.blk[0], state->path.active = 1;
1127 state->path.active <= XFS_DA_NODE_MAXDEPTH;
1128 blk++, state->path.active++) {
1129 /*
1130 * Read the next node down in the tree.
1131 */
1132 blk->blkno = blkno;
1133 error = xfs_da_read_buf(args->trans, args->dp, blkno,
1134 -1, &blk->bp, args->whichfork);
1135 if (error) {
1136 blk->blkno = 0;
1137 state->path.active--;
1138 return(error);
1139 }
1140 curr = blk->bp->data;
1141 ASSERT(INT_GET(curr->magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC ||
1142 INT_GET(curr->magic, ARCH_CONVERT) == XFS_DIRX_LEAF_MAGIC(state->mp) ||
1143 INT_GET(curr->magic, ARCH_CONVERT) == XFS_ATTR_LEAF_MAGIC);
1144
1145 /*
1146 * Search an intermediate node for a match.
1147 */
1148 blk->magic = INT_GET(curr->magic, ARCH_CONVERT);
1149 if (INT_GET(curr->magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC) {
1150 node = blk->bp->data;
1151 blk->hashval = INT_GET(node->btree[ INT_GET(node->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT);
1152
1153 /*
1154 * Binary search. (note: small blocks will skip loop)
1155 */
1156 max = INT_GET(node->hdr.count, ARCH_CONVERT);
1157 probe = span = max / 2;
1158 hashval = args->hashval;
1159 for (btree = &node->btree[probe]; span > 4;
1160 btree = &node->btree[probe]) {
1161 span /= 2;
1162 if (INT_GET(btree->hashval, ARCH_CONVERT) < hashval)
1163 probe += span;
1164 else if (INT_GET(btree->hashval, ARCH_CONVERT) > hashval)
1165 probe -= span;
1166 else
1167 break;
1168 }
1169 ASSERT((probe >= 0) && (probe < max));
1170 ASSERT((span <= 4) || (INT_GET(btree->hashval, ARCH_CONVERT) == hashval));
1171
1172 /*
1173 * Since we may have duplicate hashval's, find the first
1174 * matching hashval in the node.
1175 */
1176 while ((probe > 0) && (INT_GET(btree->hashval, ARCH_CONVERT) >= hashval)) {
1177 btree--;
1178 probe--;
1179 }
1180 while ((probe < max) && (INT_GET(btree->hashval, ARCH_CONVERT) < hashval)) {
1181 btree++;
1182 probe++;
1183 }
1184
1185 /*
1186 * Pick the right block to descend on.
1187 */
1188 if (probe == max) {
1189 blk->index = max-1;
1190 blkno = INT_GET(node->btree[ max-1 ].before, ARCH_CONVERT);
1191 } else {
1192 blk->index = probe;
1193 blkno = INT_GET(btree->before, ARCH_CONVERT);
1194 }
1195 }
1196 else if (INT_GET(curr->magic, ARCH_CONVERT) == XFS_ATTR_LEAF_MAGIC) {
1197 blk->hashval = xfs_attr_leaf_lasthash(blk->bp, NULL);
1198 break;
1199 }
1200 else if (INT_GET(curr->magic, ARCH_CONVERT) == XFS_DIR_LEAF_MAGIC) {
1201 blk->hashval = xfs_dir_leaf_lasthash(blk->bp, NULL);
1202 break;
1203 }
1204 else if (INT_GET(curr->magic, ARCH_CONVERT) == XFS_DIR2_LEAFN_MAGIC) {
1205 blk->hashval = xfs_dir2_leafn_lasthash(blk->bp, NULL);
1206 break;
1207 }
1208 }
1209
1210 /*
1211 * A leaf block that ends in the hashval that we are interested in
1212 * (final hashval == search hashval) means that the next block may
1213 * contain more entries with the same hashval, shift upward to the
1214 * next leaf and keep searching.
1215 */
1216 for (;;) {
1217 if (blk->magic == XFS_DIR_LEAF_MAGIC) {
1218 ASSERT(XFS_DIR_IS_V1(state->mp));
1219 retval = xfs_dir_leaf_lookup_int(blk->bp, args,
1220 &blk->index);
1221 } else if (blk->magic == XFS_DIR2_LEAFN_MAGIC) {
1222 ASSERT(XFS_DIR_IS_V2(state->mp));
1223 retval = xfs_dir2_leafn_lookup_int(blk->bp, args,
1224 &blk->index, state);
1225 }
1226 else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
1227 retval = xfs_attr_leaf_lookup_int(blk->bp, args);
1228 blk->index = args->index;
1229 args->blkno = blk->blkno;
1230 }
1231 if (((retval == ENOENT) || (retval == ENOATTR)) &&
1232 (blk->hashval == args->hashval)) {
1233 error = xfs_da_path_shift(state, &state->path, 1, 1,
1234 &retval);
1235 if (error)
1236 return(error);
1237 if (retval == 0) {
1238 continue;
1239 }
1240 else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
1241 /* path_shift() gives ENOENT */
1242 retval = XFS_ERROR(ENOATTR);
1243 }
1244 }
1245 break;
1246 }
1247 *result = retval;
1248 return(0);
1249 }
1250
1251 /*========================================================================
1252 * Utility routines.
1253 *========================================================================*/
1254
1255 /*
1256 * Link a new block into a doubly linked list of blocks (of whatever type).
1257 */
1258 int /* error */
1259 xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
1260 xfs_da_state_blk_t *new_blk)
1261 {
1262 xfs_da_blkinfo_t *old_info, *new_info, *tmp_info;
1263 xfs_da_args_t *args;
1264 int before=0, error;
1265 xfs_dabuf_t *bp;
1266
1267 /*
1268 * Set up environment.
1269 */
1270 args = state->args;
1271 ASSERT(args != NULL);
1272 old_info = old_blk->bp->data;
1273 new_info = new_blk->bp->data;
1274 ASSERT(old_blk->magic == XFS_DA_NODE_MAGIC ||
1275 old_blk->magic == XFS_DIRX_LEAF_MAGIC(state->mp) ||
1276 old_blk->magic == XFS_ATTR_LEAF_MAGIC);
1277 ASSERT(old_blk->magic == INT_GET(old_info->magic, ARCH_CONVERT));
1278 ASSERT(new_blk->magic == INT_GET(new_info->magic, ARCH_CONVERT));
1279 ASSERT(old_blk->magic == new_blk->magic);
1280
1281 switch (old_blk->magic) {
1282 case XFS_ATTR_LEAF_MAGIC:
1283 before = xfs_attr_leaf_order(old_blk->bp, new_blk->bp);
1284 break;
1285 case XFS_DIR_LEAF_MAGIC:
1286 ASSERT(XFS_DIR_IS_V1(state->mp));
1287 before = xfs_dir_leaf_order(old_blk->bp, new_blk->bp);
1288 break;
1289 case XFS_DIR2_LEAFN_MAGIC:
1290 ASSERT(XFS_DIR_IS_V2(state->mp));
1291 before = xfs_dir2_leafn_order(old_blk->bp, new_blk->bp);
1292 break;
1293 case XFS_DA_NODE_MAGIC:
1294 before = xfs_da_node_order(old_blk->bp, new_blk->bp);
1295 break;
1296 }
1297
1298 /*
1299 * Link blocks in appropriate order.
1300 */
1301 if (before) {
1302 /*
1303 * Link new block in before existing block.
1304 */
1305 INT_SET(new_info->forw, ARCH_CONVERT, old_blk->blkno);
1306 new_info->back = old_info->back; /* INT_: direct copy */
1307 if (INT_GET(old_info->back, ARCH_CONVERT)) {
1308 error = xfs_da_read_buf(args->trans, args->dp,
1309 INT_GET(old_info->back,
1310 ARCH_CONVERT), -1, &bp,
1311 args->whichfork);
1312 if (error)
1313 return(error);
1314 ASSERT(bp != NULL);
1315 tmp_info = bp->data;
1316 ASSERT(INT_GET(tmp_info->magic, ARCH_CONVERT) == INT_GET(old_info->magic, ARCH_CONVERT));
1317 ASSERT(INT_GET(tmp_info->forw, ARCH_CONVERT) == old_blk->blkno);
1318 INT_SET(tmp_info->forw, ARCH_CONVERT, new_blk->blkno);
1319 xfs_da_log_buf(args->trans, bp, 0, sizeof(*tmp_info)-1);
1320 xfs_da_buf_done(bp);
1321 }
1322 INT_SET(old_info->back, ARCH_CONVERT, new_blk->blkno);
1323 } else {
1324 /*
1325 * Link new block in after existing block.
1326 */
1327 new_info->forw = old_info->forw; /* INT_: direct copy */
1328 INT_SET(new_info->back, ARCH_CONVERT, old_blk->blkno);
1329 if (INT_GET(old_info->forw, ARCH_CONVERT)) {
1330 error = xfs_da_read_buf(args->trans, args->dp,
1331 INT_GET(old_info->forw, ARCH_CONVERT), -1, &bp,
1332 args->whichfork);
1333 if (error)
1334 return(error);
1335 ASSERT(bp != NULL);
1336 tmp_info = bp->data;
1337 ASSERT(INT_GET(tmp_info->magic, ARCH_CONVERT)
1338 == INT_GET(old_info->magic, ARCH_CONVERT));
1339 ASSERT(INT_GET(tmp_info->back, ARCH_CONVERT)
1340 == old_blk->blkno);
1341 INT_SET(tmp_info->back, ARCH_CONVERT, new_blk->blkno);
1342 xfs_da_log_buf(args->trans, bp, 0, sizeof(*tmp_info)-1);
1343 xfs_da_buf_done(bp);
1344 }
1345 INT_SET(old_info->forw, ARCH_CONVERT, new_blk->blkno);
1346 }
1347
1348 xfs_da_log_buf(args->trans, old_blk->bp, 0, sizeof(*tmp_info) - 1);
1349 xfs_da_log_buf(args->trans, new_blk->bp, 0, sizeof(*tmp_info) - 1);
1350 return(0);
1351 }
1352
1353 /*
1354 * Compare two intermediate nodes for "order".
1355 */
1356 STATIC int
1357 xfs_da_node_order(xfs_dabuf_t *node1_bp, xfs_dabuf_t *node2_bp)
1358 {
1359 xfs_da_intnode_t *node1, *node2;
1360
1361 node1 = node1_bp->data;
1362 node2 = node2_bp->data;
1363 ASSERT((INT_GET(node1->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC) &&
1364 (INT_GET(node2->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC));
1365 if ((INT_GET(node1->hdr.count, ARCH_CONVERT) > 0) && (INT_GET(node2->hdr.count, ARCH_CONVERT) > 0) &&
1366 ((INT_GET(node2->btree[ 0 ].hashval, ARCH_CONVERT) <
1367 INT_GET(node1->btree[ 0 ].hashval, ARCH_CONVERT)) ||
1368 (INT_GET(node2->btree[ INT_GET(node2->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT) <
1369 INT_GET(node1->btree[ INT_GET(node1->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT)))) {
1370 return(1);
1371 }
1372 return(0);
1373 }
1374
1375 /*
1376 * Pick up the last hashvalue from an intermediate node.
1377 */
1378 STATIC uint
1379 xfs_da_node_lasthash(xfs_dabuf_t *bp, int *count)
1380 {
1381 xfs_da_intnode_t *node;
1382
1383 node = bp->data;
1384 ASSERT(INT_GET(node->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
1385 if (count)
1386 *count = INT_GET(node->hdr.count, ARCH_CONVERT);
1387 if (!node->hdr.count)
1388 return(0);
1389 return(INT_GET(node->btree[ INT_GET(node->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT));
1390 }
1391
1392 /*
1393 * Unlink a block from a doubly linked list of blocks.
1394 */
1395 STATIC int /* error */
1396 xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
1397 xfs_da_state_blk_t *save_blk)
1398 {
1399 xfs_da_blkinfo_t *drop_info, *save_info, *tmp_info;
1400 xfs_da_args_t *args;
1401 xfs_dabuf_t *bp;
1402 int error;
1403
1404 /*
1405 * Set up environment.
1406 */
1407 args = state->args;
1408 ASSERT(args != NULL);
1409 save_info = save_blk->bp->data;
1410 drop_info = drop_blk->bp->data;
1411 ASSERT(save_blk->magic == XFS_DA_NODE_MAGIC ||
1412 save_blk->magic == XFS_DIRX_LEAF_MAGIC(state->mp) ||
1413 save_blk->magic == XFS_ATTR_LEAF_MAGIC);
1414 ASSERT(save_blk->magic == INT_GET(save_info->magic, ARCH_CONVERT));
1415 ASSERT(drop_blk->magic == INT_GET(drop_info->magic, ARCH_CONVERT));
1416 ASSERT(save_blk->magic == drop_blk->magic);
1417 ASSERT((INT_GET(save_info->forw, ARCH_CONVERT) == drop_blk->blkno) ||
1418 (INT_GET(save_info->back, ARCH_CONVERT) == drop_blk->blkno));
1419 ASSERT((INT_GET(drop_info->forw, ARCH_CONVERT) == save_blk->blkno) ||
1420 (INT_GET(drop_info->back, ARCH_CONVERT) == save_blk->blkno));
1421
1422 /*
1423 * Unlink the leaf block from the doubly linked chain of leaves.
1424 */
1425 if (INT_GET(save_info->back, ARCH_CONVERT) == drop_blk->blkno) {
1426 save_info->back = drop_info->back; /* INT_: direct copy */
1427 if (INT_GET(drop_info->back, ARCH_CONVERT)) {
1428 error = xfs_da_read_buf(args->trans, args->dp,
1429 INT_GET(drop_info->back,
1430 ARCH_CONVERT), -1, &bp,
1431 args->whichfork);
1432 if (error)
1433 return(error);
1434 ASSERT(bp != NULL);
1435 tmp_info = bp->data;
1436 ASSERT(INT_GET(tmp_info->magic, ARCH_CONVERT) == INT_GET(save_info->magic, ARCH_CONVERT));
1437 ASSERT(INT_GET(tmp_info->forw, ARCH_CONVERT) == drop_blk->blkno);
1438 INT_SET(tmp_info->forw, ARCH_CONVERT, save_blk->blkno);
1439 xfs_da_log_buf(args->trans, bp, 0,
1440 sizeof(*tmp_info) - 1);
1441 xfs_da_buf_done(bp);
1442 }
1443 } else {
1444 save_info->forw = drop_info->forw; /* INT_: direct copy */
1445 if (INT_GET(drop_info->forw, ARCH_CONVERT)) {
1446 error = xfs_da_read_buf(args->trans, args->dp,
1447 INT_GET(drop_info->forw, ARCH_CONVERT), -1, &bp,
1448 args->whichfork);
1449 if (error)
1450 return(error);
1451 ASSERT(bp != NULL);
1452 tmp_info = bp->data;
1453 ASSERT(INT_GET(tmp_info->magic, ARCH_CONVERT)
1454 == INT_GET(save_info->magic, ARCH_CONVERT));
1455 ASSERT(INT_GET(tmp_info->back, ARCH_CONVERT)
1456 == drop_blk->blkno);
1457 INT_SET(tmp_info->back, ARCH_CONVERT, save_blk->blkno);
1458 xfs_da_log_buf(args->trans, bp, 0,
1459 sizeof(*tmp_info) - 1);
1460 xfs_da_buf_done(bp);
1461 }
1462 }
1463
1464 xfs_da_log_buf(args->trans, save_blk->bp, 0, sizeof(*save_info) - 1);
1465 return(0);
1466 }
1467
1468 /*
1469 * Move a path "forward" or "!forward" one block at the current level.
1470 *
1471 * This routine will adjust a "path" to point to the next block
1472 * "forward" (higher hashvalues) or "!forward" (lower hashvals) in the
1473 * Btree, including updating pointers to the intermediate nodes between
1474 * the new bottom and the root.
1475 */
1476 int /* error */
1477 xfs_da_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path,
1478 int forward, int release, int *result)
1479 {
1480 xfs_da_state_blk_t *blk;
1481 xfs_da_blkinfo_t *info;
1482 xfs_da_intnode_t *node;
1483 xfs_da_args_t *args;
1484 xfs_dablk_t blkno=0;
1485 int level, error;
1486
1487 /*
1488 * Roll up the Btree looking for the first block where our
1489 * current index is not at the edge of the block. Note that
1490 * we skip the bottom layer because we want the sibling block.
1491 */
1492 args = state->args;
1493 ASSERT(args != NULL);
1494 ASSERT(path != NULL);
1495 ASSERT((path->active > 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
1496 level = (path->active-1) - 1; /* skip bottom layer in path */
1497 for (blk = &path->blk[level]; level >= 0; blk--, level--) {
1498 ASSERT(blk->bp != NULL);
1499 node = blk->bp->data;
1500 ASSERT(INT_GET(node->hdr.info.magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
1501 if (forward && (blk->index < INT_GET(node->hdr.count, ARCH_CONVERT)-1)) {
1502 blk->index++;
1503 blkno = INT_GET(node->btree[ blk->index ].before, ARCH_CONVERT);
1504 break;
1505 } else if (!forward && (blk->index > 0)) {
1506 blk->index--;
1507 blkno = INT_GET(node->btree[ blk->index ].before, ARCH_CONVERT);
1508 break;
1509 }
1510 }
1511 if (level < 0) {
1512 *result = XFS_ERROR(ENOENT); /* we're out of our tree */
1513 ASSERT(args->oknoent);
1514 return(0);
1515 }
1516
1517 /*
1518 * Roll down the edge of the subtree until we reach the
1519 * same depth we were at originally.
1520 */
1521 for (blk++, level++; level < path->active; blk++, level++) {
1522 /*
1523 * Release the old block.
1524 * (if it's dirty, trans won't actually let go)
1525 */
1526 if (release)
1527 xfs_da_brelse(args->trans, blk->bp);
1528
1529 /*
1530 * Read the next child block.
1531 */
1532 blk->blkno = blkno;
1533 error = xfs_da_read_buf(args->trans, args->dp, blkno, -1,
1534 &blk->bp, args->whichfork);
1535 if (error)
1536 return(error);
1537 ASSERT(blk->bp != NULL);
1538 info = blk->bp->data;
1539 ASSERT(INT_GET(info->magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC ||
1540 INT_GET(info->magic, ARCH_CONVERT) == XFS_DIRX_LEAF_MAGIC(state->mp) ||
1541 INT_GET(info->magic, ARCH_CONVERT) == XFS_ATTR_LEAF_MAGIC);
1542 blk->magic = INT_GET(info->magic, ARCH_CONVERT);
1543 if (INT_GET(info->magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC) {
1544 node = (xfs_da_intnode_t *)info;
1545 blk->hashval = INT_GET(node->btree[ INT_GET(node->hdr.count, ARCH_CONVERT)-1 ].hashval, ARCH_CONVERT);
1546 if (forward)
1547 blk->index = 0;
1548 else
1549 blk->index = INT_GET(node->hdr.count, ARCH_CONVERT)-1;
1550 blkno = INT_GET(node->btree[ blk->index ].before, ARCH_CONVERT);
1551 } else {
1552 ASSERT(level == path->active-1);
1553 blk->index = 0;
1554 switch(blk->magic) {
1555 case XFS_ATTR_LEAF_MAGIC:
1556 blk->hashval = xfs_attr_leaf_lasthash(blk->bp,
1557 NULL);
1558 break;
1559 case XFS_DIR_LEAF_MAGIC:
1560 ASSERT(XFS_DIR_IS_V1(state->mp));
1561 blk->hashval = xfs_dir_leaf_lasthash(blk->bp,
1562 NULL);
1563 break;
1564 case XFS_DIR2_LEAFN_MAGIC:
1565 ASSERT(XFS_DIR_IS_V2(state->mp));
1566 blk->hashval = xfs_dir2_leafn_lasthash(blk->bp,
1567 NULL);
1568 break;
1569 default:
1570 ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC ||
1571 blk->magic ==
1572 XFS_DIRX_LEAF_MAGIC(state->mp));
1573 break;
1574 }
1575 }
1576 }
1577 *result = 0;
1578 return(0);
1579 }
1580
1581
1582 /*========================================================================
1583 * Utility routines.
1584 *========================================================================*/
1585
1586 /*
1587 * Implement a simple hash on a character string.
1588 * Rotate the hash value by 7 bits, then XOR each character in.
1589 * This is implemented with some source-level loop unrolling.
1590 */
1591 xfs_dahash_t
1592 xfs_da_hashname(const uchar_t *name, int namelen)
1593 {
1594 xfs_dahash_t hash;
1595
1596 /*
1597 * Do four characters at a time as long as we can.
1598 */
1599 for (hash = 0; namelen >= 4; namelen -= 4, name += 4)
1600 hash = (name[0] << 21) ^ (name[1] << 14) ^ (name[2] << 7) ^
1601 (name[3] << 0) ^ rol32(hash, 7 * 4);
1602
1603 /*
1604 * Now do the rest of the characters.
1605 */
1606 switch (namelen) {
1607 case 3:
1608 return (name[0] << 14) ^ (name[1] << 7) ^ (name[2] << 0) ^
1609 rol32(hash, 7 * 3);
1610 case 2:
1611 return (name[0] << 7) ^ (name[1] << 0) ^ rol32(hash, 7 * 2);
1612 case 1:
1613 return (name[0] << 0) ^ rol32(hash, 7 * 1);
1614 default: /* case 0: */
1615 return hash;
1616 }
1617 }
1618
1619 /*
1620 * Add a block to the btree ahead of the file.
1621 * Return the new block number to the caller.
1622 */
1623 int
1624 xfs_da_grow_inode(xfs_da_args_t *args, xfs_dablk_t *new_blkno)
1625 {
1626 xfs_fileoff_t bno, b;
1627 xfs_bmbt_irec_t map;
1628 xfs_bmbt_irec_t *mapp;
1629 xfs_inode_t *dp;
1630 int nmap, error, w, count, c, got, i, mapi;
1631 xfs_fsize_t size;
1632 xfs_trans_t *tp;
1633 xfs_mount_t *mp;
1634
1635 dp = args->dp;
1636 mp = dp->i_mount;
1637 w = args->whichfork;
1638 tp = args->trans;
1639 /*
1640 * For new directories adjust the file offset and block count.
1641 */
1642 if (w == XFS_DATA_FORK && XFS_DIR_IS_V2(mp)) {
1643 bno = mp->m_dirleafblk;
1644 count = mp->m_dirblkfsbs;
1645 } else {
1646 bno = 0;
1647 count = 1;
1648 }
1649 /*
1650 * Find a spot in the file space to put the new block.
1651 */
1652 if ((error = xfs_bmap_first_unused(tp, dp, count, &bno, w))) {
1653 return error;
1654 }
1655 if (w == XFS_DATA_FORK && XFS_DIR_IS_V2(mp))
1656 ASSERT(bno >= mp->m_dirleafblk && bno < mp->m_dirfreeblk);
1657 /*
1658 * Try mapping it in one filesystem block.
1659 */
1660 nmap = 1;
1661 ASSERT(args->firstblock != NULL);
1662 if ((error = xfs_bmapi(tp, dp, bno, count,
1663 XFS_BMAPI_AFLAG(w)|XFS_BMAPI_WRITE|XFS_BMAPI_METADATA|
1664 XFS_BMAPI_CONTIG,
1665 args->firstblock, args->total, &map, &nmap,
1666 args->flist))) {
1667 return error;
1668 }
1669 ASSERT(nmap <= 1);
1670 if (nmap == 1) {
1671 mapp = &map;
1672 mapi = 1;
1673 }
1674 /*
1675 * If we didn't get it and the block might work if fragmented,
1676 * try without the CONTIG flag. Loop until we get it all.
1677 */
1678 else if (nmap == 0 && count > 1) {
1679 mapp = kmem_alloc(sizeof(*mapp) * count, KM_SLEEP);
1680 for (b = bno, mapi = 0; b < bno + count; ) {
1681 nmap = MIN(XFS_BMAP_MAX_NMAP, count);
1682 c = (int)(bno + count - b);
1683 if ((error = xfs_bmapi(tp, dp, b, c,
1684 XFS_BMAPI_AFLAG(w)|XFS_BMAPI_WRITE|
1685 XFS_BMAPI_METADATA,
1686 args->firstblock, args->total,
1687 &mapp[mapi], &nmap, args->flist))) {
1688 kmem_free(mapp, sizeof(*mapp) * count);
1689 return error;
1690 }
1691 if (nmap < 1)
1692 break;
1693 mapi += nmap;
1694 b = mapp[mapi - 1].br_startoff +
1695 mapp[mapi - 1].br_blockcount;
1696 }
1697 } else {
1698 mapi = 0;
1699 mapp = NULL;
1700 }
1701 /*
1702 * Count the blocks we got, make sure it matches the total.
1703 */
1704 for (i = 0, got = 0; i < mapi; i++)
1705 got += mapp[i].br_blockcount;
1706 if (got != count || mapp[0].br_startoff != bno ||
1707 mapp[mapi - 1].br_startoff + mapp[mapi - 1].br_blockcount !=
1708 bno + count) {
1709 if (mapp != &map)
1710 kmem_free(mapp, sizeof(*mapp) * count);
1711 return XFS_ERROR(ENOSPC);
1712 }
1713 if (mapp != &map)
1714 kmem_free(mapp, sizeof(*mapp) * count);
1715 *new_blkno = (xfs_dablk_t)bno;
1716 /*
1717 * For version 1 directories, adjust the file size if it changed.
1718 */
1719 if (w == XFS_DATA_FORK && XFS_DIR_IS_V1(mp)) {
1720 ASSERT(mapi == 1);
1721 if ((error = xfs_bmap_last_offset(tp, dp, &bno, w)))
1722 return error;
1723 size = XFS_FSB_TO_B(mp, bno);
1724 if (size != dp->i_d.di_size) {
1725 dp->i_d.di_size = size;
1726 xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
1727 }
1728 }
1729 return 0;
1730 }
1731
1732 /*
1733 * Ick. We need to always be able to remove a btree block, even
1734 * if there's no space reservation because the filesystem is full.
1735 * This is called if xfs_bunmapi on a btree block fails due to ENOSPC.
1736 * It swaps the target block with the last block in the file. The
1737 * last block in the file can always be removed since it can't cause
1738 * a bmap btree split to do that.
1739 */
1740 STATIC int
1741 xfs_da_swap_lastblock(xfs_da_args_t *args, xfs_dablk_t *dead_blknop,
1742 xfs_dabuf_t **dead_bufp)
1743 {
1744 xfs_dablk_t dead_blkno, last_blkno, sib_blkno, par_blkno;
1745 xfs_dabuf_t *dead_buf, *last_buf, *sib_buf, *par_buf;
1746 xfs_fileoff_t lastoff;
1747 xfs_inode_t *ip;
1748 xfs_trans_t *tp;
1749 xfs_mount_t *mp;
1750 int error, w, entno, level, dead_level;
1751 xfs_da_blkinfo_t *dead_info, *sib_info;
1752 xfs_da_intnode_t *par_node, *dead_node;
1753 xfs_dir_leafblock_t *dead_leaf;
1754 xfs_dir2_leaf_t *dead_leaf2;
1755 xfs_dahash_t dead_hash;
1756
1757 dead_buf = *dead_bufp;
1758 dead_blkno = *dead_blknop;
1759 tp = args->trans;
1760 ip = args->dp;
1761 w = args->whichfork;
1762 ASSERT(w == XFS_DATA_FORK);
1763 mp = ip->i_mount;
1764 if (XFS_DIR_IS_V2(mp)) {
1765 lastoff = mp->m_dirfreeblk;
1766 error = xfs_bmap_last_before(tp, ip, &lastoff, w);
1767 } else
1768 error = xfs_bmap_last_offset(tp, ip, &lastoff, w);
1769 if (error)
1770 return error;
1771 if (unlikely(lastoff == 0)) {
1772 XFS_ERROR_REPORT("xfs_da_swap_lastblock(1)", XFS_ERRLEVEL_LOW,
1773 mp);
1774 return XFS_ERROR(EFSCORRUPTED);
1775 }
1776 /*
1777 * Read the last block in the btree space.
1778 */
1779 last_blkno = (xfs_dablk_t)lastoff - mp->m_dirblkfsbs;
1780 if ((error = xfs_da_read_buf(tp, ip, last_blkno, -1, &last_buf, w)))
1781 return error;
1782 /*
1783 * Copy the last block into the dead buffer and log it.
1784 */
1785 memcpy(dead_buf->data, last_buf->data, mp->m_dirblksize);
1786 xfs_da_log_buf(tp, dead_buf, 0, mp->m_dirblksize - 1);
1787 dead_info = dead_buf->data;
1788 /*
1789 * Get values from the moved block.
1790 */
1791 if (INT_GET(dead_info->magic, ARCH_CONVERT) == XFS_DIR_LEAF_MAGIC) {
1792 ASSERT(XFS_DIR_IS_V1(mp));
1793 dead_leaf = (xfs_dir_leafblock_t *)dead_info;
1794 dead_level = 0;
1795 dead_hash =
1796 INT_GET(dead_leaf->entries[INT_GET(dead_leaf->hdr.count, ARCH_CONVERT) - 1].hashval, ARCH_CONVERT);
1797 } else if (INT_GET(dead_info->magic, ARCH_CONVERT) == XFS_DIR2_LEAFN_MAGIC) {
1798 ASSERT(XFS_DIR_IS_V2(mp));
1799 dead_leaf2 = (xfs_dir2_leaf_t *)dead_info;
1800 dead_level = 0;
1801 dead_hash = INT_GET(dead_leaf2->ents[INT_GET(dead_leaf2->hdr.count, ARCH_CONVERT) - 1].hashval, ARCH_CONVERT);
1802 } else {
1803 ASSERT(INT_GET(dead_info->magic, ARCH_CONVERT) == XFS_DA_NODE_MAGIC);
1804 dead_node = (xfs_da_intnode_t *)dead_info;
1805 dead_level = INT_GET(dead_node->hdr.level, ARCH_CONVERT);
1806 dead_hash = INT_GET(dead_node->btree[INT_GET(dead_node->hdr.count, ARCH_CONVERT) - 1].hashval, ARCH_CONVERT);
1807 }
1808 sib_buf = par_buf = NULL;
1809 /*
1810 * If the moved block has a left sibling, fix up the pointers.
1811 */
1812 if ((sib_blkno = INT_GET(dead_info->back, ARCH_CONVERT))) {
1813 if ((error = xfs_da_read_buf(tp, ip, sib_blkno, -1, &sib_buf, w)))
1814 goto done;
1815 sib_info = sib_buf->data;
1816 if (unlikely(
1817 INT_GET(sib_info->forw, ARCH_CONVERT) != last_blkno ||
1818 INT_GET(sib_info->magic, ARCH_CONVERT) != INT_GET(dead_info->magic, ARCH_CONVERT))) {
1819 XFS_ERROR_REPORT("xfs_da_swap_lastblock(2)",
1820 XFS_ERRLEVEL_LOW, mp);
1821 error = XFS_ERROR(EFSCORRUPTED);
1822 goto done;
1823 }
1824 INT_SET(sib_info->forw, ARCH_CONVERT, dead_blkno);
1825 xfs_da_log_buf(tp, sib_buf,
1826 XFS_DA_LOGRANGE(sib_info, &sib_info->forw,
1827 sizeof(sib_info->forw)));
1828 xfs_da_buf_done(sib_buf);
1829 sib_buf = NULL;
1830 }
1831 /*
1832 * If the moved block has a right sibling, fix up the pointers.
1833 */
1834 if ((sib_blkno = INT_GET(dead_info->forw, ARCH_CONVERT))) {
1835 if ((error = xfs_da_read_buf(tp, ip, sib_blkno, -1, &sib_buf, w)))
1836 goto done;
1837 sib_info = sib_buf->data;
1838 if (unlikely(
1839 INT_GET(sib_info->back, ARCH_CONVERT) != last_blkno
1840 || INT_GET(sib_info->magic, ARCH_CONVERT)
1841 != INT_GET(dead_info->magic, ARCH_CONVERT))) {
1842 XFS_ERROR_REPORT("xfs_da_swap_lastblock(3)",
1843 XFS_ERRLEVEL_LOW, mp);
1844 error = XFS_ERROR(EFSCORRUPTED);
1845 goto done;
1846 }
1847 INT_SET(sib_info->back, ARCH_CONVERT, dead_blkno);
1848 xfs_da_log_buf(tp, sib_buf,
1849 XFS_DA_LOGRANGE(sib_info, &sib_info->back,
1850 sizeof(sib_info->back)));
1851 xfs_da_buf_done(sib_buf);
1852 sib_buf = NULL;
1853 }
1854 par_blkno = XFS_DIR_IS_V1(mp) ? 0 : mp->m_dirleafblk;
1855 level = -1;
1856 /*
1857 * Walk down the tree looking for the parent of the moved block.
1858 */
1859 for (;;) {
1860 if ((error = xfs_da_read_buf(tp, ip, par_blkno, -1, &par_buf, w)))
1861 goto done;
1862 par_node = par_buf->data;
1863 if (unlikely(
1864 INT_GET(par_node->hdr.info.magic, ARCH_CONVERT) != XFS_DA_NODE_MAGIC ||
1865 (level >= 0 && level != INT_GET(par_node->hdr.level, ARCH_CONVERT) + 1))) {
1866 XFS_ERROR_REPORT("xfs_da_swap_lastblock(4)",
1867 XFS_ERRLEVEL_LOW, mp);
1868 error = XFS_ERROR(EFSCORRUPTED);
1869 goto done;
1870 }
1871 level = INT_GET(par_node->hdr.level, ARCH_CONVERT);
1872 for (entno = 0;
1873 entno < INT_GET(par_node->hdr.count, ARCH_CONVERT) &&
1874 INT_GET(par_node->btree[entno].hashval, ARCH_CONVERT) < dead_hash;
1875 entno++)
1876 continue;
1877 if (unlikely(entno == INT_GET(par_node->hdr.count, ARCH_CONVERT))) {
1878 XFS_ERROR_REPORT("xfs_da_swap_lastblock(5)",
1879 XFS_ERRLEVEL_LOW, mp);
1880 error = XFS_ERROR(EFSCORRUPTED);
1881 goto done;
1882 }
1883 par_blkno = INT_GET(par_node->btree[entno].before, ARCH_CONVERT);
1884 if (level == dead_level + 1)
1885 break;
1886 xfs_da_brelse(tp, par_buf);
1887 par_buf = NULL;
1888 }
1889 /*
1890 * We're in the right parent block.
1891 * Look for the right entry.
1892 */
1893 for (;;) {
1894 for (;
1895 entno < INT_GET(par_node->hdr.count, ARCH_CONVERT) &&
1896 INT_GET(par_node->btree[entno].before, ARCH_CONVERT) != last_blkno;
1897 entno++)
1898 continue;
1899 if (entno < INT_GET(par_node->hdr.count, ARCH_CONVERT))
1900 break;
1901 par_blkno = INT_GET(par_node->hdr.info.forw, ARCH_CONVERT);
1902 xfs_da_brelse(tp, par_buf);
1903 par_buf = NULL;
1904 if (unlikely(par_blkno == 0)) {
1905 XFS_ERROR_REPORT("xfs_da_swap_lastblock(6)",
1906 XFS_ERRLEVEL_LOW, mp);
1907 error = XFS_ERROR(EFSCORRUPTED);
1908 goto done;
1909 }
1910 if ((error = xfs_da_read_buf(tp, ip, par_blkno, -1, &par_buf, w)))
1911 goto done;
1912 par_node = par_buf->data;
1913 if (unlikely(
1914 INT_GET(par_node->hdr.level, ARCH_CONVERT) != level ||
1915 INT_GET(par_node->hdr.info.magic, ARCH_CONVERT) != XFS_DA_NODE_MAGIC)) {
1916 XFS_ERROR_REPORT("xfs_da_swap_lastblock(7)",
1917 XFS_ERRLEVEL_LOW, mp);
1918 error = XFS_ERROR(EFSCORRUPTED);
1919 goto done;
1920 }
1921 entno = 0;
1922 }
1923 /*
1924 * Update the parent entry pointing to the moved block.
1925 */
1926 INT_SET(par_node->btree[entno].before, ARCH_CONVERT, dead_blkno);
1927 xfs_da_log_buf(tp, par_buf,
1928 XFS_DA_LOGRANGE(par_node, &par_node->btree[entno].before,
1929 sizeof(par_node->btree[entno].before)));
1930 xfs_da_buf_done(par_buf);
1931 xfs_da_buf_done(dead_buf);
1932 *dead_blknop = last_blkno;
1933 *dead_bufp = last_buf;
1934 return 0;
1935 done:
1936 if (par_buf)
1937 xfs_da_brelse(tp, par_buf);
1938 if (sib_buf)
1939 xfs_da_brelse(tp, sib_buf);
1940 xfs_da_brelse(tp, last_buf);
1941 return error;
1942 }
1943
1944 /*
1945 * Remove a btree block from a directory or attribute.
1946 */
1947 int
1948 xfs_da_shrink_inode(xfs_da_args_t *args, xfs_dablk_t dead_blkno,
1949 xfs_dabuf_t *dead_buf)
1950 {
1951 xfs_inode_t *dp;
1952 int done, error, w, count;
1953 xfs_fileoff_t bno;
1954 xfs_fsize_t size;
1955 xfs_trans_t *tp;
1956 xfs_mount_t *mp;
1957
1958 dp = args->dp;
1959 w = args->whichfork;
1960 tp = args->trans;
1961 mp = dp->i_mount;
1962 if (w == XFS_DATA_FORK && XFS_DIR_IS_V2(mp))
1963 count = mp->m_dirblkfsbs;
1964 else
1965 count = 1;
1966 for (;;) {
1967 /*
1968 * Remove extents. If we get ENOSPC for a dir we have to move
1969 * the last block to the place we want to kill.
1970 */
1971 if ((error = xfs_bunmapi(tp, dp, dead_blkno, count,
1972 XFS_BMAPI_AFLAG(w)|XFS_BMAPI_METADATA,
1973 0, args->firstblock, args->flist,
1974 &done)) == ENOSPC) {
1975 if (w != XFS_DATA_FORK)
1976 goto done;
1977 if ((error = xfs_da_swap_lastblock(args, &dead_blkno,
1978 &dead_buf)))
1979 goto done;
1980 } else if (error)
1981 goto done;
1982 else
1983 break;
1984 }
1985 ASSERT(done);
1986 xfs_da_binval(tp, dead_buf);
1987 /*
1988 * Adjust the directory size for version 1.
1989 */
1990 if (w == XFS_DATA_FORK && XFS_DIR_IS_V1(mp)) {
1991 if ((error = xfs_bmap_last_offset(tp, dp, &bno, w)))
1992 return error;
1993 size = XFS_FSB_TO_B(dp->i_mount, bno);
1994 if (size != dp->i_d.di_size) {
1995 dp->i_d.di_size = size;
1996 xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
1997 }
1998 }
1999 return 0;
2000 done:
2001 xfs_da_binval(tp, dead_buf);
2002 return error;
2003 }
2004
2005 /*
2006 * See if the mapping(s) for this btree block are valid, i.e.
2007 * don't contain holes, are logically contiguous, and cover the whole range.
2008 */
2009 STATIC int
2010 xfs_da_map_covers_blocks(
2011 int nmap,
2012 xfs_bmbt_irec_t *mapp,
2013 xfs_dablk_t bno,
2014 int count)
2015 {
2016 int i;
2017 xfs_fileoff_t off;
2018
2019 for (i = 0, off = bno; i < nmap; i++) {
2020 if (mapp[i].br_startblock == HOLESTARTBLOCK ||
2021 mapp[i].br_startblock == DELAYSTARTBLOCK) {
2022 return 0;
2023 }
2024 if (off != mapp[i].br_startoff) {
2025 return 0;
2026 }
2027 off += mapp[i].br_blockcount;
2028 }
2029 return off == bno + count;
2030 }
2031
2032 /*
2033 * Make a dabuf.
2034 * Used for get_buf, read_buf, read_bufr, and reada_buf.
2035 */
2036 STATIC int
2037 xfs_da_do_buf(
2038 xfs_trans_t *trans,
2039 xfs_inode_t *dp,
2040 xfs_dablk_t bno,
2041 xfs_daddr_t *mappedbnop,
2042 xfs_dabuf_t **bpp,
2043 int whichfork,
2044 int caller,
2045 inst_t *ra)
2046 {
2047 xfs_buf_t *bp = NULL;
2048 xfs_buf_t **bplist;
2049 int error=0;
2050 int i;
2051 xfs_bmbt_irec_t map;
2052 xfs_bmbt_irec_t *mapp;
2053 xfs_daddr_t mappedbno;
2054 xfs_mount_t *mp;
2055 int nbplist=0;
2056 int nfsb;
2057 int nmap;
2058 xfs_dabuf_t *rbp;
2059
2060 mp = dp->i_mount;
2061 if (whichfork == XFS_DATA_FORK && XFS_DIR_IS_V2(mp))
2062 nfsb = mp->m_dirblkfsbs;
2063 else
2064 nfsb = 1;
2065 mappedbno = *mappedbnop;
2066 /*
2067 * Caller doesn't have a mapping. -2 means don't complain
2068 * if we land in a hole.
2069 */
2070 if (mappedbno == -1 || mappedbno == -2) {
2071 /*
2072 * Optimize the one-block case.
2073 */
2074 if (nfsb == 1) {
2075 xfs_fsblock_t fsb;
2076
2077 if ((error =
2078 xfs_bmapi_single(trans, dp, whichfork, &fsb,
2079 (xfs_fileoff_t)bno))) {
2080 return error;
2081 }
2082 mapp = &map;
2083 if (fsb == NULLFSBLOCK) {
2084 nmap = 0;
2085 } else {
2086 map.br_startblock = fsb;
2087 map.br_startoff = (xfs_fileoff_t)bno;
2088 map.br_blockcount = 1;
2089 nmap = 1;
2090 }
2091 } else {
2092 mapp = kmem_alloc(sizeof(*mapp) * nfsb, KM_SLEEP);
2093 nmap = nfsb;
2094 if ((error = xfs_bmapi(trans, dp, (xfs_fileoff_t)bno,
2095 nfsb,
2096 XFS_BMAPI_METADATA |
2097 XFS_BMAPI_AFLAG(whichfork),
2098 NULL, 0, mapp, &nmap, NULL)))
2099 goto exit0;
2100 }
2101 } else {
2102 map.br_startblock = XFS_DADDR_TO_FSB(mp, mappedbno);
2103 map.br_startoff = (xfs_fileoff_t)bno;
2104 map.br_blockcount = nfsb;
2105 mapp = &map;
2106 nmap = 1;
2107 }
2108 if (!xfs_da_map_covers_blocks(nmap, mapp, bno, nfsb)) {
2109 error = mappedbno == -2 ? 0 : XFS_ERROR(EFSCORRUPTED);
2110 if (unlikely(error == EFSCORRUPTED)) {
2111 if (xfs_error_level >= XFS_ERRLEVEL_LOW) {
2112 int i;
2113 cmn_err(CE_ALERT, "xfs_da_do_buf: bno %lld\n",
2114 (long long)bno);
2115 cmn_err(CE_ALERT, "dir: inode %lld\n",
2116 (long long)dp->i_ino);
2117 for (i = 0; i < nmap; i++) {
2118 cmn_err(CE_ALERT,
2119 "[%02d] br_startoff %lld br_startblock %lld br_blockcount %lld br_state %d\n",
2120 i,
2121 (long long)mapp[i].br_startoff,
2122 (long long)mapp[i].br_startblock,
2123 (long long)mapp[i].br_blockcount,
2124 mapp[i].br_state);
2125 }
2126 }
2127 XFS_ERROR_REPORT("xfs_da_do_buf(1)",
2128 XFS_ERRLEVEL_LOW, mp);
2129 }
2130 goto exit0;
2131 }
2132 if (caller != 3 && nmap > 1) {
2133 bplist = kmem_alloc(sizeof(*bplist) * nmap, KM_SLEEP);
2134 nbplist = 0;
2135 } else
2136 bplist = NULL;
2137 /*
2138 * Turn the mapping(s) into buffer(s).
2139 */
2140 for (i = 0; i < nmap; i++) {
2141 int nmapped;
2142
2143 mappedbno = XFS_FSB_TO_DADDR(mp, mapp[i].br_startblock);
2144 if (i == 0)
2145 *mappedbnop = mappedbno;
2146 nmapped = (int)XFS_FSB_TO_BB(mp, mapp[i].br_blockcount);
2147 switch (caller) {
2148 case 0:
2149 bp = xfs_trans_get_buf(trans, mp->m_ddev_targp,
2150 mappedbno, nmapped, 0);
2151 error = bp ? XFS_BUF_GETERROR(bp) : XFS_ERROR(EIO);
2152 break;
2153 case 1:
2154 case 2:
2155 bp = NULL;
2156 error = xfs_trans_read_buf(mp, trans, mp->m_ddev_targp,
2157 mappedbno, nmapped, 0, &bp);
2158 break;
2159 case 3:
2160 xfs_baread(mp->m_ddev_targp, mappedbno, nmapped);
2161 error = 0;
2162 bp = NULL;
2163 break;
2164 }
2165 if (error) {
2166 if (bp)
2167 xfs_trans_brelse(trans, bp);
2168 goto exit1;
2169 }
2170 if (!bp)
2171 continue;
2172 if (caller == 1) {
2173 if (whichfork == XFS_ATTR_FORK) {
2174 XFS_BUF_SET_VTYPE_REF(bp, B_FS_ATTR_BTREE,
2175 XFS_ATTR_BTREE_REF);
2176 } else {
2177 XFS_BUF_SET_VTYPE_REF(bp, B_FS_DIR_BTREE,
2178 XFS_DIR_BTREE_REF);
2179 }
2180 }
2181 if (bplist) {
2182 bplist[nbplist++] = bp;
2183 }
2184 }
2185 /*
2186 * Build a dabuf structure.
2187 */
2188 if (bplist) {
2189 rbp = xfs_da_buf_make(nbplist, bplist, ra);
2190 } else if (bp)
2191 rbp = xfs_da_buf_make(1, &bp, ra);
2192 else
2193 rbp = NULL;
2194 /*
2195 * For read_buf, check the magic number.
2196 */
2197 if (caller == 1) {
2198 xfs_dir2_data_t *data;
2199 xfs_dir2_free_t *free;
2200 xfs_da_blkinfo_t *info;
2201 uint magic, magic1;
2202
2203 info = rbp->data;
2204 data = rbp->data;
2205 free = rbp->data;
2206 magic = INT_GET(info->magic, ARCH_CONVERT);
2207 magic1 = INT_GET(data->hdr.magic, ARCH_CONVERT);
2208 if (unlikely(
2209 XFS_TEST_ERROR((magic != XFS_DA_NODE_MAGIC) &&
2210 (magic != XFS_DIR_LEAF_MAGIC) &&
2211 (magic != XFS_ATTR_LEAF_MAGIC) &&
2212 (magic != XFS_DIR2_LEAF1_MAGIC) &&
2213 (magic != XFS_DIR2_LEAFN_MAGIC) &&
2214 (magic1 != XFS_DIR2_BLOCK_MAGIC) &&
2215 (magic1 != XFS_DIR2_DATA_MAGIC) &&
2216 (INT_GET(free->hdr.magic, ARCH_CONVERT) != XFS_DIR2_FREE_MAGIC),
2217 mp, XFS_ERRTAG_DA_READ_BUF,
2218 XFS_RANDOM_DA_READ_BUF))) {
2219 xfs_buftrace("DA READ ERROR", rbp->bps[0]);
2220 XFS_CORRUPTION_ERROR("xfs_da_do_buf(2)",
2221 XFS_ERRLEVEL_LOW, mp, info);
2222 error = XFS_ERROR(EFSCORRUPTED);
2223 xfs_da_brelse(trans, rbp);
2224 nbplist = 0;
2225 goto exit1;
2226 }
2227 }
2228 if (bplist) {
2229 kmem_free(bplist, sizeof(*bplist) * nmap);
2230 }
2231 if (mapp != &map) {
2232 kmem_free(mapp, sizeof(*mapp) * nfsb);
2233 }
2234 if (bpp)
2235 *bpp = rbp;
2236 return 0;
2237 exit1:
2238 if (bplist) {
2239 for (i = 0; i < nbplist; i++)
2240 xfs_trans_brelse(trans, bplist[i]);
2241 kmem_free(bplist, sizeof(*bplist) * nmap);
2242 }
2243 exit0:
2244 if (mapp != &map)
2245 kmem_free(mapp, sizeof(*mapp) * nfsb);
2246 if (bpp)
2247 *bpp = NULL;
2248 return error;
2249 }
2250
2251 /*
2252 * Get a buffer for the dir/attr block.
2253 */
2254 int
2255 xfs_da_get_buf(
2256 xfs_trans_t *trans,
2257 xfs_inode_t *dp,
2258 xfs_dablk_t bno,
2259 xfs_daddr_t mappedbno,
2260 xfs_dabuf_t **bpp,
2261 int whichfork)
2262 {
2263 return xfs_da_do_buf(trans, dp, bno, &mappedbno, bpp, whichfork, 0,
2264 (inst_t *)__return_address);
2265 }
2266
2267 /*
2268 * Get a buffer for the dir/attr block, fill in the contents.
2269 */
2270 int
2271 xfs_da_read_buf(
2272 xfs_trans_t *trans,
2273 xfs_inode_t *dp,
2274 xfs_dablk_t bno,
2275 xfs_daddr_t mappedbno,
2276 xfs_dabuf_t **bpp,
2277 int whichfork)
2278 {
2279 return xfs_da_do_buf(trans, dp, bno, &mappedbno, bpp, whichfork, 1,
2280 (inst_t *)__return_address);
2281 }
2282
2283 /*
2284 * Readahead the dir/attr block.
2285 */
2286 xfs_daddr_t
2287 xfs_da_reada_buf(
2288 xfs_trans_t *trans,
2289 xfs_inode_t *dp,
2290 xfs_dablk_t bno,
2291 int whichfork)
2292 {
2293 xfs_daddr_t rval;
2294
2295 rval = -1;
2296 if (xfs_da_do_buf(trans, dp, bno, &rval, NULL, whichfork, 3,
2297 (inst_t *)__return_address))
2298 return -1;
2299 else
2300 return rval;
2301 }
2302
2303 /*
2304 * Calculate the number of bits needed to hold i different values.
2305 */
2306 uint
2307 xfs_da_log2_roundup(uint i)
2308 {
2309 uint rval;
2310
2311 for (rval = 0; rval < NBBY * sizeof(i); rval++) {
2312 if ((1 << rval) >= i)
2313 break;
2314 }
2315 return(rval);
2316 }
2317
2318 kmem_zone_t *xfs_da_state_zone; /* anchor for state struct zone */
2319 kmem_zone_t *xfs_dabuf_zone; /* dabuf zone */
2320
2321 /*
2322 * Allocate a dir-state structure.
2323 * We don't put them on the stack since they're large.
2324 */
2325 xfs_da_state_t *
2326 xfs_da_state_alloc(void)
2327 {
2328 return kmem_zone_zalloc(xfs_da_state_zone, KM_SLEEP);
2329 }
2330
2331 /*
2332 * Kill the altpath contents of a da-state structure.
2333 */
2334 STATIC void
2335 xfs_da_state_kill_altpath(xfs_da_state_t *state)
2336 {
2337 int i;
2338
2339 for (i = 0; i < state->altpath.active; i++) {
2340 if (state->altpath.blk[i].bp) {
2341 if (state->altpath.blk[i].bp != state->path.blk[i].bp)
2342 xfs_da_buf_done(state->altpath.blk[i].bp);
2343 state->altpath.blk[i].bp = NULL;
2344 }
2345 }
2346 state->altpath.active = 0;
2347 }
2348
2349 /*
2350 * Free a da-state structure.
2351 */
2352 void
2353 xfs_da_state_free(xfs_da_state_t *state)
2354 {
2355 int i;
2356
2357 xfs_da_state_kill_altpath(state);
2358 for (i = 0; i < state->path.active; i++) {
2359 if (state->path.blk[i].bp)
2360 xfs_da_buf_done(state->path.blk[i].bp);
2361 }
2362 if (state->extravalid && state->extrablk.bp)
2363 xfs_da_buf_done(state->extrablk.bp);
2364 #ifdef DEBUG
2365 memset((char *)state, 0, sizeof(*state));
2366 #endif /* DEBUG */
2367 kmem_zone_free(xfs_da_state_zone, state);
2368 }
2369
2370 #ifdef XFS_DABUF_DEBUG
2371 xfs_dabuf_t *xfs_dabuf_global_list;
2372 lock_t xfs_dabuf_global_lock;
2373 #endif
2374
2375 /*
2376 * Create a dabuf.
2377 */
2378 /* ARGSUSED */
2379 STATIC xfs_dabuf_t *
2380 xfs_da_buf_make(int nbuf, xfs_buf_t **bps, inst_t *ra)
2381 {
2382 xfs_buf_t *bp;
2383 xfs_dabuf_t *dabuf;
2384 int i;
2385 int off;
2386
2387 if (nbuf == 1)
2388 dabuf = kmem_zone_alloc(xfs_dabuf_zone, KM_SLEEP);
2389 else
2390 dabuf = kmem_alloc(XFS_DA_BUF_SIZE(nbuf), KM_SLEEP);
2391 dabuf->dirty = 0;
2392 #ifdef XFS_DABUF_DEBUG
2393 dabuf->ra = ra;
2394 dabuf->target = XFS_BUF_TARGET(bps[0]);
2395 dabuf->blkno = XFS_BUF_ADDR(bps[0]);
2396 #endif
2397 if (nbuf == 1) {
2398 dabuf->nbuf = 1;
2399 bp = bps[0];
2400 dabuf->bbcount = (short)BTOBB(XFS_BUF_COUNT(bp));
2401 dabuf->data = XFS_BUF_PTR(bp);
2402 dabuf->bps[0] = bp;
2403 } else {
2404 dabuf->nbuf = nbuf;
2405 for (i = 0, dabuf->bbcount = 0; i < nbuf; i++) {
2406 dabuf->bps[i] = bp = bps[i];
2407 dabuf->bbcount += BTOBB(XFS_BUF_COUNT(bp));
2408 }
2409 dabuf->data = kmem_alloc(BBTOB(dabuf->bbcount), KM_SLEEP);
2410 for (i = off = 0; i < nbuf; i++, off += XFS_BUF_COUNT(bp)) {
2411 bp = bps[i];
2412 memcpy((char *)dabuf->data + off, XFS_BUF_PTR(bp),
2413 XFS_BUF_COUNT(bp));
2414 }
2415 }
2416 #ifdef XFS_DABUF_DEBUG
2417 {
2418 SPLDECL(s);
2419 xfs_dabuf_t *p;
2420
2421 s = mutex_spinlock(&xfs_dabuf_global_lock);
2422 for (p = xfs_dabuf_global_list; p; p = p->next) {
2423 ASSERT(p->blkno != dabuf->blkno ||
2424 p->target != dabuf->target);
2425 }
2426 dabuf->prev = NULL;
2427 if (xfs_dabuf_global_list)
2428 xfs_dabuf_global_list->prev = dabuf;
2429 dabuf->next = xfs_dabuf_global_list;
2430 xfs_dabuf_global_list = dabuf;
2431 mutex_spinunlock(&xfs_dabuf_global_lock, s);
2432 }
2433 #endif
2434 return dabuf;
2435 }
2436
2437 /*
2438 * Un-dirty a dabuf.
2439 */
2440 STATIC void
2441 xfs_da_buf_clean(xfs_dabuf_t *dabuf)
2442 {
2443 xfs_buf_t *bp;
2444 int i;
2445 int off;
2446
2447 if (dabuf->dirty) {
2448 ASSERT(dabuf->nbuf > 1);
2449 dabuf->dirty = 0;
2450 for (i = off = 0; i < dabuf->nbuf;
2451 i++, off += XFS_BUF_COUNT(bp)) {
2452 bp = dabuf->bps[i];
2453 memcpy(XFS_BUF_PTR(bp), (char *)dabuf->data + off,
2454 XFS_BUF_COUNT(bp));
2455 }
2456 }
2457 }
2458
2459 /*
2460 * Release a dabuf.
2461 */
2462 void
2463 xfs_da_buf_done(xfs_dabuf_t *dabuf)
2464 {
2465 ASSERT(dabuf);
2466 ASSERT(dabuf->nbuf && dabuf->data && dabuf->bbcount && dabuf->bps[0]);
2467 if (dabuf->dirty)
2468 xfs_da_buf_clean(dabuf);
2469 if (dabuf->nbuf > 1)
2470 kmem_free(dabuf->data, BBTOB(dabuf->bbcount));
2471 #ifdef XFS_DABUF_DEBUG
2472 {
2473 SPLDECL(s);
2474
2475 s = mutex_spinlock(&xfs_dabuf_global_lock);
2476 if (dabuf->prev)
2477 dabuf->prev->next = dabuf->next;
2478 else
2479 xfs_dabuf_global_list = dabuf->next;
2480 if (dabuf->next)
2481 dabuf->next->prev = dabuf->prev;
2482 mutex_spinunlock(&xfs_dabuf_global_lock, s);
2483 }
2484 memset(dabuf, 0, XFS_DA_BUF_SIZE(dabuf->nbuf));
2485 #endif
2486 if (dabuf->nbuf == 1)
2487 kmem_zone_free(xfs_dabuf_zone, dabuf);
2488 else
2489 kmem_free(dabuf, XFS_DA_BUF_SIZE(dabuf->nbuf));
2490 }
2491
2492 /*
2493 * Log transaction from a dabuf.
2494 */
2495 void
2496 xfs_da_log_buf(xfs_trans_t *tp, xfs_dabuf_t *dabuf, uint first, uint last)
2497 {
2498 xfs_buf_t *bp;
2499 uint f;
2500 int i;
2501 uint l;
2502 int off;
2503
2504 ASSERT(dabuf->nbuf && dabuf->data && dabuf->bbcount && dabuf->bps[0]);
2505 if (dabuf->nbuf == 1) {
2506 ASSERT(dabuf->data == (void *)XFS_BUF_PTR(dabuf->bps[0]));
2507 xfs_trans_log_buf(tp, dabuf->bps[0], first, last);
2508 return;
2509 }
2510 dabuf->dirty = 1;
2511 ASSERT(first <= last);
2512 for (i = off = 0; i < dabuf->nbuf; i++, off += XFS_BUF_COUNT(bp)) {
2513 bp = dabuf->bps[i];
2514 f = off;
2515 l = f + XFS_BUF_COUNT(bp) - 1;
2516 if (f < first)
2517 f = first;
2518 if (l > last)
2519 l = last;
2520 if (f <= l)
2521 xfs_trans_log_buf(tp, bp, f - off, l - off);
2522 /*
2523 * B_DONE is set by xfs_trans_log buf.
2524 * If we don't set it on a new buffer (get not read)
2525 * then if we don't put anything in the buffer it won't
2526 * be set, and at commit it it released into the cache,
2527 * and then a read will fail.
2528 */
2529 else if (!(XFS_BUF_ISDONE(bp)))
2530 XFS_BUF_DONE(bp);
2531 }
2532 ASSERT(last < off);
2533 }
2534
2535 /*
2536 * Release dabuf from a transaction.
2537 * Have to free up the dabuf before the buffers are released,
2538 * since the synchronization on the dabuf is really the lock on the buffer.
2539 */
2540 void
2541 xfs_da_brelse(xfs_trans_t *tp, xfs_dabuf_t *dabuf)
2542 {
2543 xfs_buf_t *bp;
2544 xfs_buf_t **bplist;
2545 int i;
2546 int nbuf;
2547
2548 ASSERT(dabuf->nbuf && dabuf->data && dabuf->bbcount && dabuf->bps[0]);
2549 if ((nbuf = dabuf->nbuf) == 1) {
2550 bplist = &bp;
2551 bp = dabuf->bps[0];
2552 } else {
2553 bplist = kmem_alloc(nbuf * sizeof(*bplist), KM_SLEEP);
2554 memcpy(bplist, dabuf->bps, nbuf * sizeof(*bplist));
2555 }
2556 xfs_da_buf_done(dabuf);
2557 for (i = 0; i < nbuf; i++)
2558 xfs_trans_brelse(tp, bplist[i]);
2559 if (bplist != &bp)
2560 kmem_free(bplist, nbuf * sizeof(*bplist));
2561 }
2562
2563 /*
2564 * Invalidate dabuf from a transaction.
2565 */
2566 void
2567 xfs_da_binval(xfs_trans_t *tp, xfs_dabuf_t *dabuf)
2568 {
2569 xfs_buf_t *bp;
2570 xfs_buf_t **bplist;
2571 int i;
2572 int nbuf;
2573
2574 ASSERT(dabuf->nbuf && dabuf->data && dabuf->bbcount && dabuf->bps[0]);
2575 if ((nbuf = dabuf->nbuf) == 1) {
2576 bplist = &bp;
2577 bp = dabuf->bps[0];
2578 } else {
2579 bplist = kmem_alloc(nbuf * sizeof(*bplist), KM_SLEEP);
2580 memcpy(bplist, dabuf->bps, nbuf * sizeof(*bplist));
2581 }
2582 xfs_da_buf_done(dabuf);
2583 for (i = 0; i < nbuf; i++)
2584 xfs_trans_binval(tp, bplist[i]);
2585 if (bplist != &bp)
2586 kmem_free(bplist, nbuf * sizeof(*bplist));
2587 }
2588
2589 /*
2590 * Get the first daddr from a dabuf.
2591 */
2592 xfs_daddr_t
2593 xfs_da_blkno(xfs_dabuf_t *dabuf)
2594 {
2595 ASSERT(dabuf->nbuf);
2596 ASSERT(dabuf->data);
2597 return XFS_BUF_ADDR(dabuf->bps[0]);
2598 }
linux-next