2 * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README
6 * Written by Anatoly P. Pinchuk pap@namesys.botik.ru
7 * Programm System Institute
8 * Pereslavl-Zalessky Russia
12 * This file contains functions dealing with S+tree
26 * decrement_counters_in_path
28 * pathrelse_and_restore
32 * search_for_position_by_key
34 * prepare_for_direct_item
35 * prepare_for_direntry_item
36 * prepare_for_delete_or_cut
37 * calc_deleted_bytes_number
40 * reiserfs_delete_item
41 * reiserfs_delete_solid_item
42 * reiserfs_delete_object
43 * maybe_indirect_to_direct
44 * indirect_to_direct_roll_back
45 * reiserfs_cut_from_item
47 * reiserfs_do_truncate
48 * reiserfs_paste_into_item
49 * reiserfs_insert_item
52 #include <linux/config.h>
53 #include <linux/time.h>
54 #include <linux/string.h>
55 #include <linux/pagemap.h>
56 #include <linux/reiserfs_fs.h>
57 #include <linux/smp_lock.h>
58 #include <linux/buffer_head.h>
59 #include <linux/quotaops.h>
60 #include <linux/vs_dlimit.h>
62 /* Does the buffer contain a disk block which is in the tree. */
63 inline int B_IS_IN_TREE(const struct buffer_head *p_s_bh)
66 RFALSE(B_LEVEL(p_s_bh) > MAX_HEIGHT,
67 "PAP-1010: block (%b) has too big level (%z)", p_s_bh, p_s_bh);
69 return (B_LEVEL(p_s_bh) != FREE_LEVEL);
73 // to gets item head in le form
75 inline void copy_item_head(struct item_head *p_v_to,
76 const struct item_head *p_v_from)
78 memcpy(p_v_to, p_v_from, IH_SIZE);
81 /* k1 is pointer to on-disk structure which is stored in little-endian
82 form. k2 is pointer to cpu variable. For key of items of the same
83 object this returns 0.
84 Returns: -1 if key1 < key2
87 inline int comp_short_keys(const struct reiserfs_key *le_key,
88 const struct cpu_key *cpu_key)
91 n = le32_to_cpu(le_key->k_dir_id);
92 if (n < cpu_key->on_disk_key.k_dir_id)
94 if (n > cpu_key->on_disk_key.k_dir_id)
96 n = le32_to_cpu(le_key->k_objectid);
97 if (n < cpu_key->on_disk_key.k_objectid)
99 if (n > cpu_key->on_disk_key.k_objectid)
104 /* k1 is pointer to on-disk structure which is stored in little-endian
105 form. k2 is pointer to cpu variable.
106 Compare keys using all 4 key fields.
107 Returns: -1 if key1 < key2 0
108 if key1 = key2 1 if key1 > key2 */
109 static inline int comp_keys(const struct reiserfs_key *le_key,
110 const struct cpu_key *cpu_key)
114 retval = comp_short_keys(le_key, cpu_key);
117 if (le_key_k_offset(le_key_version(le_key), le_key) <
118 cpu_key_k_offset(cpu_key))
120 if (le_key_k_offset(le_key_version(le_key), le_key) >
121 cpu_key_k_offset(cpu_key))
124 if (cpu_key->key_length == 3)
127 /* this part is needed only when tail conversion is in progress */
128 if (le_key_k_type(le_key_version(le_key), le_key) <
129 cpu_key_k_type(cpu_key))
132 if (le_key_k_type(le_key_version(le_key), le_key) >
133 cpu_key_k_type(cpu_key))
139 inline int comp_short_le_keys(const struct reiserfs_key *key1,
140 const struct reiserfs_key *key2)
142 __u32 *p_s_1_u32, *p_s_2_u32;
143 int n_key_length = REISERFS_SHORT_KEY_LEN;
145 p_s_1_u32 = (__u32 *) key1;
146 p_s_2_u32 = (__u32 *) key2;
147 for (; n_key_length--; ++p_s_1_u32, ++p_s_2_u32) {
148 if (le32_to_cpu(*p_s_1_u32) < le32_to_cpu(*p_s_2_u32))
150 if (le32_to_cpu(*p_s_1_u32) > le32_to_cpu(*p_s_2_u32))
156 inline void le_key2cpu_key(struct cpu_key *to, const struct reiserfs_key *from)
159 to->on_disk_key.k_dir_id = le32_to_cpu(from->k_dir_id);
160 to->on_disk_key.k_objectid = le32_to_cpu(from->k_objectid);
162 // find out version of the key
163 version = le_key_version(from);
164 to->version = version;
165 to->on_disk_key.k_offset = le_key_k_offset(version, from);
166 to->on_disk_key.k_type = le_key_k_type(version, from);
169 // this does not say which one is bigger, it only returns 1 if keys
170 // are not equal, 0 otherwise
171 inline int comp_le_keys(const struct reiserfs_key *k1,
172 const struct reiserfs_key *k2)
174 return memcmp(k1, k2, sizeof(struct reiserfs_key));
177 /**************************************************************************
178 * Binary search toolkit function *
179 * Search for an item in the array by the item key *
180 * Returns: 1 if found, 0 if not found; *
181 * *p_n_pos = number of the searched element if found, else the *
182 * number of the first element that is larger than p_v_key. *
183 **************************************************************************/
184 /* For those not familiar with binary search: n_lbound is the leftmost item that it
185 could be, n_rbound the rightmost item that it could be. We examine the item
186 halfway between n_lbound and n_rbound, and that tells us either that we can increase
187 n_lbound, or decrease n_rbound, or that we have found it, or if n_lbound <= n_rbound that
188 there are no possible items, and we have not found it. With each examination we
189 cut the number of possible items it could be by one more than half rounded down,
191 static inline int bin_search(const void *p_v_key, /* Key to search for. */
192 const void *p_v_base, /* First item in the array. */
193 int p_n_num, /* Number of items in the array. */
194 int p_n_width, /* Item size in the array.
195 searched. Lest the reader be
196 confused, note that this is crafted
197 as a general function, and when it
198 is applied specifically to the array
199 of item headers in a node, p_n_width
200 is actually the item header size not
202 int *p_n_pos /* Number of the searched for element. */
205 int n_rbound, n_lbound, n_j;
207 for (n_j = ((n_rbound = p_n_num - 1) + (n_lbound = 0)) / 2;
208 n_lbound <= n_rbound; n_j = (n_rbound + n_lbound) / 2)
210 ((struct reiserfs_key *)((char *)p_v_base +
212 (struct cpu_key *)p_v_key)) {
221 return ITEM_FOUND; /* Key found in the array. */
224 /* bin_search did not find given key, it returns position of key,
225 that is minimal and greater than the given one. */
227 return ITEM_NOT_FOUND;
230 #ifdef CONFIG_REISERFS_CHECK
231 extern struct tree_balance *cur_tb;
234 /* Minimal possible key. It is never in the tree. */
235 const struct reiserfs_key MIN_KEY = { 0, 0, {{0, 0},} };
237 /* Maximal possible key. It is never in the tree. */
238 static const struct reiserfs_key MAX_KEY = {
239 __constant_cpu_to_le32(0xffffffff),
240 __constant_cpu_to_le32(0xffffffff),
241 {{__constant_cpu_to_le32(0xffffffff),
242 __constant_cpu_to_le32(0xffffffff)},}
245 /* Get delimiting key of the buffer by looking for it in the buffers in the path, starting from the bottom
246 of the path, and going upwards. We must check the path's validity at each step. If the key is not in
247 the path, there is no delimiting key in the tree (buffer is first or last buffer in tree), and in this
248 case we return a special key, either MIN_KEY or MAX_KEY. */
249 static inline const struct reiserfs_key *get_lkey(const struct path
251 const struct super_block
254 int n_position, n_path_offset = p_s_chk_path->path_length;
255 struct buffer_head *p_s_parent;
257 RFALSE(n_path_offset < FIRST_PATH_ELEMENT_OFFSET,
258 "PAP-5010: invalid offset in the path");
260 /* While not higher in path than first element. */
261 while (n_path_offset-- > FIRST_PATH_ELEMENT_OFFSET) {
263 RFALSE(!buffer_uptodate
264 (PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset)),
265 "PAP-5020: parent is not uptodate");
267 /* Parent at the path is not in the tree now. */
270 PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset)))
272 /* Check whether position in the parent is correct. */
274 PATH_OFFSET_POSITION(p_s_chk_path,
276 B_NR_ITEMS(p_s_parent))
278 /* Check whether parent at the path really points to the child. */
279 if (B_N_CHILD_NUM(p_s_parent, n_position) !=
280 PATH_OFFSET_PBUFFER(p_s_chk_path,
281 n_path_offset + 1)->b_blocknr)
283 /* Return delimiting key if position in the parent is not equal to zero. */
285 return B_N_PDELIM_KEY(p_s_parent, n_position - 1);
287 /* Return MIN_KEY if we are in the root of the buffer tree. */
288 if (PATH_OFFSET_PBUFFER(p_s_chk_path, FIRST_PATH_ELEMENT_OFFSET)->
289 b_blocknr == SB_ROOT_BLOCK(p_s_sb))
294 /* Get delimiting key of the buffer at the path and its right neighbor. */
295 inline const struct reiserfs_key *get_rkey(const struct path *p_s_chk_path,
296 const struct super_block *p_s_sb)
298 int n_position, n_path_offset = p_s_chk_path->path_length;
299 struct buffer_head *p_s_parent;
301 RFALSE(n_path_offset < FIRST_PATH_ELEMENT_OFFSET,
302 "PAP-5030: invalid offset in the path");
304 while (n_path_offset-- > FIRST_PATH_ELEMENT_OFFSET) {
306 RFALSE(!buffer_uptodate
307 (PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset)),
308 "PAP-5040: parent is not uptodate");
310 /* Parent at the path is not in the tree now. */
313 PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset)))
315 /* Check whether position in the parent is correct. */
317 PATH_OFFSET_POSITION(p_s_chk_path,
319 B_NR_ITEMS(p_s_parent))
321 /* Check whether parent at the path really points to the child. */
322 if (B_N_CHILD_NUM(p_s_parent, n_position) !=
323 PATH_OFFSET_PBUFFER(p_s_chk_path,
324 n_path_offset + 1)->b_blocknr)
326 /* Return delimiting key if position in the parent is not the last one. */
327 if (n_position != B_NR_ITEMS(p_s_parent))
328 return B_N_PDELIM_KEY(p_s_parent, n_position);
330 /* Return MAX_KEY if we are in the root of the buffer tree. */
331 if (PATH_OFFSET_PBUFFER(p_s_chk_path, FIRST_PATH_ELEMENT_OFFSET)->
332 b_blocknr == SB_ROOT_BLOCK(p_s_sb))
337 /* Check whether a key is contained in the tree rooted from a buffer at a path. */
338 /* This works by looking at the left and right delimiting keys for the buffer in the last path_element in
339 the path. These delimiting keys are stored at least one level above that buffer in the tree. If the
340 buffer is the first or last node in the tree order then one of the delimiting keys may be absent, and in
341 this case get_lkey and get_rkey return a special key which is MIN_KEY or MAX_KEY. */
342 static inline int key_in_buffer(struct path *p_s_chk_path, /* Path which should be checked. */
343 const struct cpu_key *p_s_key, /* Key which should be checked. */
344 struct super_block *p_s_sb /* Super block pointer. */
348 RFALSE(!p_s_key || p_s_chk_path->path_length < FIRST_PATH_ELEMENT_OFFSET
349 || p_s_chk_path->path_length > MAX_HEIGHT,
350 "PAP-5050: pointer to the key(%p) is NULL or invalid path length(%d)",
351 p_s_key, p_s_chk_path->path_length);
352 RFALSE(!PATH_PLAST_BUFFER(p_s_chk_path)->b_bdev,
353 "PAP-5060: device must not be NODEV");
355 if (comp_keys(get_lkey(p_s_chk_path, p_s_sb), p_s_key) == 1)
356 /* left delimiting key is bigger, that the key we look for */
358 // if ( comp_keys(p_s_key, get_rkey(p_s_chk_path, p_s_sb)) != -1 )
359 if (comp_keys(get_rkey(p_s_chk_path, p_s_sb), p_s_key) != 1)
360 /* p_s_key must be less than right delimitiing key */
365 inline void decrement_bcount(struct buffer_head *p_s_bh)
368 if (atomic_read(&(p_s_bh->b_count))) {
373 "PAP-5070: decrement_bcount: trying to free free buffer %b",
378 /* Decrement b_count field of the all buffers in the path. */
379 void decrement_counters_in_path(struct path *p_s_search_path)
381 int n_path_offset = p_s_search_path->path_length;
383 RFALSE(n_path_offset < ILLEGAL_PATH_ELEMENT_OFFSET ||
384 n_path_offset > EXTENDED_MAX_HEIGHT - 1,
385 "PAP-5080: invalid path offset of %d", n_path_offset);
387 while (n_path_offset > ILLEGAL_PATH_ELEMENT_OFFSET) {
388 struct buffer_head *bh;
390 bh = PATH_OFFSET_PBUFFER(p_s_search_path, n_path_offset--);
391 decrement_bcount(bh);
393 p_s_search_path->path_length = ILLEGAL_PATH_ELEMENT_OFFSET;
396 int reiserfs_check_path(struct path *p)
398 RFALSE(p->path_length != ILLEGAL_PATH_ELEMENT_OFFSET,
399 "path not properly relsed");
403 /* Release all buffers in the path. Restore dirty bits clean
404 ** when preparing the buffer for the log
406 ** only called from fix_nodes()
408 void pathrelse_and_restore(struct super_block *s, struct path *p_s_search_path)
410 int n_path_offset = p_s_search_path->path_length;
412 RFALSE(n_path_offset < ILLEGAL_PATH_ELEMENT_OFFSET,
413 "clm-4000: invalid path offset");
415 while (n_path_offset > ILLEGAL_PATH_ELEMENT_OFFSET) {
416 reiserfs_restore_prepared_buffer(s,
420 brelse(PATH_OFFSET_PBUFFER(p_s_search_path, n_path_offset--));
422 p_s_search_path->path_length = ILLEGAL_PATH_ELEMENT_OFFSET;
425 /* Release all buffers in the path. */
426 void pathrelse(struct path *p_s_search_path)
428 int n_path_offset = p_s_search_path->path_length;
430 RFALSE(n_path_offset < ILLEGAL_PATH_ELEMENT_OFFSET,
431 "PAP-5090: invalid path offset");
433 while (n_path_offset > ILLEGAL_PATH_ELEMENT_OFFSET)
434 brelse(PATH_OFFSET_PBUFFER(p_s_search_path, n_path_offset--));
436 p_s_search_path->path_length = ILLEGAL_PATH_ELEMENT_OFFSET;
439 static int is_leaf(char *buf, int blocksize, struct buffer_head *bh)
441 struct block_head *blkh;
442 struct item_head *ih;
448 blkh = (struct block_head *)buf;
449 if (blkh_level(blkh) != DISK_LEAF_NODE_LEVEL) {
450 reiserfs_warning(NULL,
451 "is_leaf: this should be caught earlier");
455 nr = blkh_nr_item(blkh);
456 if (nr < 1 || nr > ((blocksize - BLKH_SIZE) / (IH_SIZE + MIN_ITEM_LEN))) {
457 /* item number is too big or too small */
458 reiserfs_warning(NULL, "is_leaf: nr_item seems wrong: %z", bh);
461 ih = (struct item_head *)(buf + BLKH_SIZE) + nr - 1;
462 used_space = BLKH_SIZE + IH_SIZE * nr + (blocksize - ih_location(ih));
463 if (used_space != blocksize - blkh_free_space(blkh)) {
464 /* free space does not match to calculated amount of use space */
465 reiserfs_warning(NULL, "is_leaf: free space seems wrong: %z",
469 // FIXME: it is_leaf will hit performance too much - we may have
472 /* check tables of item heads */
473 ih = (struct item_head *)(buf + BLKH_SIZE);
474 prev_location = blocksize;
475 for (i = 0; i < nr; i++, ih++) {
476 if (le_ih_k_type(ih) == TYPE_ANY) {
477 reiserfs_warning(NULL,
478 "is_leaf: wrong item type for item %h",
482 if (ih_location(ih) >= blocksize
483 || ih_location(ih) < IH_SIZE * nr) {
484 reiserfs_warning(NULL,
485 "is_leaf: item location seems wrong: %h",
489 if (ih_item_len(ih) < 1
490 || ih_item_len(ih) > MAX_ITEM_LEN(blocksize)) {
491 reiserfs_warning(NULL,
492 "is_leaf: item length seems wrong: %h",
496 if (prev_location - ih_location(ih) != ih_item_len(ih)) {
497 reiserfs_warning(NULL,
498 "is_leaf: item location seems wrong (second one): %h",
502 prev_location = ih_location(ih);
505 // one may imagine much more checks
509 /* returns 1 if buf looks like an internal node, 0 otherwise */
510 static int is_internal(char *buf, int blocksize, struct buffer_head *bh)
512 struct block_head *blkh;
516 blkh = (struct block_head *)buf;
517 nr = blkh_level(blkh);
518 if (nr <= DISK_LEAF_NODE_LEVEL || nr > MAX_HEIGHT) {
519 /* this level is not possible for internal nodes */
520 reiserfs_warning(NULL,
521 "is_internal: this should be caught earlier");
525 nr = blkh_nr_item(blkh);
526 if (nr > (blocksize - BLKH_SIZE - DC_SIZE) / (KEY_SIZE + DC_SIZE)) {
527 /* for internal which is not root we might check min number of keys */
528 reiserfs_warning(NULL,
529 "is_internal: number of key seems wrong: %z",
534 used_space = BLKH_SIZE + KEY_SIZE * nr + DC_SIZE * (nr + 1);
535 if (used_space != blocksize - blkh_free_space(blkh)) {
536 reiserfs_warning(NULL,
537 "is_internal: free space seems wrong: %z", bh);
540 // one may imagine much more checks
544 // make sure that bh contains formatted node of reiserfs tree of
546 static int is_tree_node(struct buffer_head *bh, int level)
548 if (B_LEVEL(bh) != level) {
549 reiserfs_warning(NULL,
550 "is_tree_node: node level %d does not match to the expected one %d",
554 if (level == DISK_LEAF_NODE_LEVEL)
555 return is_leaf(bh->b_data, bh->b_size, bh);
557 return is_internal(bh->b_data, bh->b_size, bh);
560 #define SEARCH_BY_KEY_READA 16
562 /* The function is NOT SCHEDULE-SAFE! */
563 static void search_by_key_reada(struct super_block *s,
564 struct buffer_head **bh,
565 unsigned long *b, int num)
569 for (i = 0; i < num; i++) {
570 bh[i] = sb_getblk(s, b[i]);
572 for (j = 0; j < i; j++) {
574 * note, this needs attention if we are getting rid of the BKL
575 * you have to make sure the prepared bit isn't set on this buffer
577 if (!buffer_uptodate(bh[j]))
578 ll_rw_block(READA, 1, bh + j);
583 /**************************************************************************
584 * Algorithm SearchByKey *
585 * look for item in the Disk S+Tree by its key *
586 * Input: p_s_sb - super block *
587 * p_s_key - pointer to the key to search *
588 * Output: ITEM_FOUND, ITEM_NOT_FOUND or IO_ERROR *
589 * p_s_search_path - path from the root to the needed leaf *
590 **************************************************************************/
592 /* This function fills up the path from the root to the leaf as it
593 descends the tree looking for the key. It uses reiserfs_bread to
594 try to find buffers in the cache given their block number. If it
595 does not find them in the cache it reads them from disk. For each
596 node search_by_key finds using reiserfs_bread it then uses
597 bin_search to look through that node. bin_search will find the
598 position of the block_number of the next node if it is looking
599 through an internal node. If it is looking through a leaf node
600 bin_search will find the position of the item which has key either
601 equal to given key, or which is the maximal key less than the given
602 key. search_by_key returns a path that must be checked for the
603 correctness of the top of the path but need not be checked for the
604 correctness of the bottom of the path */
605 /* The function is NOT SCHEDULE-SAFE! */
606 int search_by_key(struct super_block *p_s_sb, const struct cpu_key *p_s_key, /* Key to search. */
607 struct path *p_s_search_path, /* This structure was
608 allocated and initialized
610 function. It is filled up
612 int n_stop_level /* How far down the tree to search. To
613 stop at leaf level - set to
614 DISK_LEAF_NODE_LEVEL */
619 struct buffer_head *p_s_bh;
620 struct path_element *p_s_last_element;
621 int n_node_level, n_retval;
622 int right_neighbor_of_leaf_node;
624 struct buffer_head *reada_bh[SEARCH_BY_KEY_READA];
625 unsigned long reada_blocks[SEARCH_BY_KEY_READA];
628 #ifdef CONFIG_REISERFS_CHECK
629 int n_repeat_counter = 0;
632 PROC_INFO_INC(p_s_sb, search_by_key);
634 /* As we add each node to a path we increase its count. This means that
635 we must be careful to release all nodes in a path before we either
636 discard the path struct or re-use the path struct, as we do here. */
638 decrement_counters_in_path(p_s_search_path);
640 right_neighbor_of_leaf_node = 0;
642 /* With each iteration of this loop we search through the items in the
643 current node, and calculate the next current node(next path element)
644 for the next iteration of this loop.. */
645 n_block_number = SB_ROOT_BLOCK(p_s_sb);
649 #ifdef CONFIG_REISERFS_CHECK
650 if (!(++n_repeat_counter % 50000))
651 reiserfs_warning(p_s_sb, "PAP-5100: search_by_key: %s:"
652 "there were %d iterations of while loop "
653 "looking for key %K",
654 current->comm, n_repeat_counter,
658 /* prep path to have another element added to it. */
660 PATH_OFFSET_PELEMENT(p_s_search_path,
661 ++p_s_search_path->path_length);
662 fs_gen = get_generation(p_s_sb);
664 /* Read the next tree node, and set the last element in the path to
665 have a pointer to it. */
666 if ((p_s_bh = p_s_last_element->pe_buffer =
667 sb_getblk(p_s_sb, n_block_number))) {
668 if (!buffer_uptodate(p_s_bh) && reada_count > 1) {
669 search_by_key_reada(p_s_sb, reada_bh,
670 reada_blocks, reada_count);
672 ll_rw_block(READ, 1, &p_s_bh);
673 wait_on_buffer(p_s_bh);
674 if (!buffer_uptodate(p_s_bh))
678 p_s_search_path->path_length--;
679 pathrelse(p_s_search_path);
683 if (expected_level == -1)
684 expected_level = SB_TREE_HEIGHT(p_s_sb);
687 /* It is possible that schedule occurred. We must check whether the key
688 to search is still in the tree rooted from the current buffer. If
689 not then repeat search from the root. */
690 if (fs_changed(fs_gen, p_s_sb) &&
691 (!B_IS_IN_TREE(p_s_bh) ||
692 B_LEVEL(p_s_bh) != expected_level ||
693 !key_in_buffer(p_s_search_path, p_s_key, p_s_sb))) {
694 PROC_INFO_INC(p_s_sb, search_by_key_fs_changed);
695 PROC_INFO_INC(p_s_sb, search_by_key_restarted);
696 PROC_INFO_INC(p_s_sb,
697 sbk_restarted[expected_level - 1]);
698 decrement_counters_in_path(p_s_search_path);
700 /* Get the root block number so that we can repeat the search
701 starting from the root. */
702 n_block_number = SB_ROOT_BLOCK(p_s_sb);
704 right_neighbor_of_leaf_node = 0;
706 /* repeat search from the root */
710 /* only check that the key is in the buffer if p_s_key is not
711 equal to the MAX_KEY. Latter case is only possible in
712 "finish_unfinished()" processing during mount. */
713 RFALSE(comp_keys(&MAX_KEY, p_s_key) &&
714 !key_in_buffer(p_s_search_path, p_s_key, p_s_sb),
715 "PAP-5130: key is not in the buffer");
716 #ifdef CONFIG_REISERFS_CHECK
718 print_cur_tb("5140");
719 reiserfs_panic(p_s_sb,
720 "PAP-5140: search_by_key: schedule occurred in do_balance!");
724 // make sure, that the node contents look like a node of
726 if (!is_tree_node(p_s_bh, expected_level)) {
727 reiserfs_warning(p_s_sb, "vs-5150: search_by_key: "
728 "invalid format found in block %ld. Fsck?",
730 pathrelse(p_s_search_path);
734 /* ok, we have acquired next formatted node in the tree */
735 n_node_level = B_LEVEL(p_s_bh);
737 PROC_INFO_BH_STAT(p_s_sb, p_s_bh, n_node_level - 1);
739 RFALSE(n_node_level < n_stop_level,
740 "vs-5152: tree level (%d) is less than stop level (%d)",
741 n_node_level, n_stop_level);
743 n_retval = bin_search(p_s_key, B_N_PITEM_HEAD(p_s_bh, 0),
746 DISK_LEAF_NODE_LEVEL) ? IH_SIZE :
748 &(p_s_last_element->pe_position));
749 if (n_node_level == n_stop_level) {
753 /* we are not in the stop level */
754 if (n_retval == ITEM_FOUND)
755 /* item has been found, so we choose the pointer which is to the right of the found one */
756 p_s_last_element->pe_position++;
758 /* if item was not found we choose the position which is to
759 the left of the found item. This requires no code,
760 bin_search did it already. */
762 /* So we have chosen a position in the current node which is
763 an internal node. Now we calculate child block number by
764 position in the node. */
766 B_N_CHILD_NUM(p_s_bh, p_s_last_element->pe_position);
768 /* if we are going to read leaf nodes, try for read ahead as well */
769 if ((p_s_search_path->reada & PATH_READA) &&
770 n_node_level == DISK_LEAF_NODE_LEVEL + 1) {
771 int pos = p_s_last_element->pe_position;
772 int limit = B_NR_ITEMS(p_s_bh);
773 struct reiserfs_key *le_key;
775 if (p_s_search_path->reada & PATH_READA_BACK)
777 while (reada_count < SEARCH_BY_KEY_READA) {
780 reada_blocks[reada_count++] =
781 B_N_CHILD_NUM(p_s_bh, pos);
782 if (p_s_search_path->reada & PATH_READA_BACK)
788 * check to make sure we're in the same object
790 le_key = B_N_PDELIM_KEY(p_s_bh, pos);
791 if (le32_to_cpu(le_key->k_objectid) !=
792 p_s_key->on_disk_key.k_objectid) {
800 /* Form the path to an item and position in this item which contains
801 file byte defined by p_s_key. If there is no such item
802 corresponding to the key, we point the path to the item with
803 maximal key less than p_s_key, and *p_n_pos_in_item is set to one
804 past the last entry/byte in the item. If searching for entry in a
805 directory item, and it is not found, *p_n_pos_in_item is set to one
806 entry more than the entry with maximal key which is less than the
809 Note that if there is no entry in this same node which is one more,
810 then we point to an imaginary entry. for direct items, the
811 position is in units of bytes, for indirect items the position is
812 in units of blocknr entries, for directory items the position is in
813 units of directory entries. */
815 /* The function is NOT SCHEDULE-SAFE! */
816 int search_for_position_by_key(struct super_block *p_s_sb, /* Pointer to the super block. */
817 const struct cpu_key *p_cpu_key, /* Key to search (cpu variable) */
818 struct path *p_s_search_path /* Filled up by this function. */
821 struct item_head *p_le_ih; /* pointer to on-disk structure */
823 loff_t item_offset, offset;
824 struct reiserfs_dir_entry de;
827 /* If searching for directory entry. */
828 if (is_direntry_cpu_key(p_cpu_key))
829 return search_by_entry_key(p_s_sb, p_cpu_key, p_s_search_path,
832 /* If not searching for directory entry. */
834 /* If item is found. */
835 retval = search_item(p_s_sb, p_cpu_key, p_s_search_path);
836 if (retval == IO_ERROR)
838 if (retval == ITEM_FOUND) {
842 (PATH_PLAST_BUFFER(p_s_search_path),
843 PATH_LAST_POSITION(p_s_search_path))),
844 "PAP-5165: item length equals zero");
846 pos_in_item(p_s_search_path) = 0;
847 return POSITION_FOUND;
850 RFALSE(!PATH_LAST_POSITION(p_s_search_path),
851 "PAP-5170: position equals zero");
853 /* Item is not found. Set path to the previous item. */
855 B_N_PITEM_HEAD(PATH_PLAST_BUFFER(p_s_search_path),
856 --PATH_LAST_POSITION(p_s_search_path));
857 n_blk_size = p_s_sb->s_blocksize;
859 if (comp_short_keys(&(p_le_ih->ih_key), p_cpu_key)) {
860 return FILE_NOT_FOUND;
862 // FIXME: quite ugly this far
864 item_offset = le_ih_k_offset(p_le_ih);
865 offset = cpu_key_k_offset(p_cpu_key);
867 /* Needed byte is contained in the item pointed to by the path. */
868 if (item_offset <= offset &&
869 item_offset + op_bytes_number(p_le_ih, n_blk_size) > offset) {
870 pos_in_item(p_s_search_path) = offset - item_offset;
871 if (is_indirect_le_ih(p_le_ih)) {
872 pos_in_item(p_s_search_path) /= n_blk_size;
874 return POSITION_FOUND;
877 /* Needed byte is not contained in the item pointed to by the
878 path. Set pos_in_item out of the item. */
879 if (is_indirect_le_ih(p_le_ih))
880 pos_in_item(p_s_search_path) =
881 ih_item_len(p_le_ih) / UNFM_P_SIZE;
883 pos_in_item(p_s_search_path) = ih_item_len(p_le_ih);
885 return POSITION_NOT_FOUND;
888 /* Compare given item and item pointed to by the path. */
889 int comp_items(const struct item_head *stored_ih, const struct path *p_s_path)
891 struct buffer_head *p_s_bh;
892 struct item_head *ih;
894 /* Last buffer at the path is not in the tree. */
895 if (!B_IS_IN_TREE(p_s_bh = PATH_PLAST_BUFFER(p_s_path)))
898 /* Last path position is invalid. */
899 if (PATH_LAST_POSITION(p_s_path) >= B_NR_ITEMS(p_s_bh))
902 /* we need only to know, whether it is the same item */
903 ih = get_ih(p_s_path);
904 return memcmp(stored_ih, ih, IH_SIZE);
907 /* unformatted nodes are not logged anymore, ever. This is safe
910 #define held_by_others(bh) (atomic_read(&(bh)->b_count) > 1)
912 // block can not be forgotten as it is in I/O or held by someone
913 #define block_in_use(bh) (buffer_locked(bh) || (held_by_others(bh)))
915 // prepare for delete or cut of direct item
916 static inline int prepare_for_direct_item(struct path *path,
917 struct item_head *le_ih,
919 loff_t new_file_length, int *cut_size)
923 if (new_file_length == max_reiserfs_offset(inode)) {
924 /* item has to be deleted */
925 *cut_size = -(IH_SIZE + ih_item_len(le_ih));
928 // new file gets truncated
929 if (get_inode_item_key_version(inode) == KEY_FORMAT_3_6) {
931 round_len = ROUND_UP(new_file_length);
932 /* this was n_new_file_length < le_ih ... */
933 if (round_len < le_ih_k_offset(le_ih)) {
934 *cut_size = -(IH_SIZE + ih_item_len(le_ih));
935 return M_DELETE; /* Delete this item. */
937 /* Calculate first position and size for cutting from item. */
938 pos_in_item(path) = round_len - (le_ih_k_offset(le_ih) - 1);
939 *cut_size = -(ih_item_len(le_ih) - pos_in_item(path));
941 return M_CUT; /* Cut from this item. */
944 // old file: items may have any length
946 if (new_file_length < le_ih_k_offset(le_ih)) {
947 *cut_size = -(IH_SIZE + ih_item_len(le_ih));
948 return M_DELETE; /* Delete this item. */
950 /* Calculate first position and size for cutting from item. */
951 *cut_size = -(ih_item_len(le_ih) -
953 new_file_length + 1 - le_ih_k_offset(le_ih)));
954 return M_CUT; /* Cut from this item. */
957 static inline int prepare_for_direntry_item(struct path *path,
958 struct item_head *le_ih,
960 loff_t new_file_length,
963 if (le_ih_k_offset(le_ih) == DOT_OFFSET &&
964 new_file_length == max_reiserfs_offset(inode)) {
965 RFALSE(ih_entry_count(le_ih) != 2,
966 "PAP-5220: incorrect empty directory item (%h)", le_ih);
967 *cut_size = -(IH_SIZE + ih_item_len(le_ih));
968 return M_DELETE; /* Delete the directory item containing "." and ".." entry. */
971 if (ih_entry_count(le_ih) == 1) {
972 /* Delete the directory item such as there is one record only
974 *cut_size = -(IH_SIZE + ih_item_len(le_ih));
978 /* Cut one record from the directory item. */
981 entry_length(get_last_bh(path), le_ih, pos_in_item(path)));
985 /* If the path points to a directory or direct item, calculate mode and the size cut, for balance.
986 If the path points to an indirect item, remove some number of its unformatted nodes.
987 In case of file truncate calculate whether this item must be deleted/truncated or last
988 unformatted node of this item will be converted to a direct item.
989 This function returns a determination of what balance mode the calling function should employ. */
990 static char prepare_for_delete_or_cut(struct reiserfs_transaction_handle *th, struct inode *inode, struct path *p_s_path, const struct cpu_key *p_s_item_key, int *p_n_removed, /* Number of unformatted nodes which were removed
991 from end of the file. */
992 int *p_n_cut_size, unsigned long long n_new_file_length /* MAX_KEY_OFFSET in case of delete. */
995 struct super_block *p_s_sb = inode->i_sb;
996 struct item_head *p_le_ih = PATH_PITEM_HEAD(p_s_path);
997 struct buffer_head *p_s_bh = PATH_PLAST_BUFFER(p_s_path);
999 BUG_ON(!th->t_trans_id);
1001 /* Stat_data item. */
1002 if (is_statdata_le_ih(p_le_ih)) {
1004 RFALSE(n_new_file_length != max_reiserfs_offset(inode),
1005 "PAP-5210: mode must be M_DELETE");
1007 *p_n_cut_size = -(IH_SIZE + ih_item_len(p_le_ih));
1011 /* Directory item. */
1012 if (is_direntry_le_ih(p_le_ih))
1013 return prepare_for_direntry_item(p_s_path, p_le_ih, inode,
1018 if (is_direct_le_ih(p_le_ih))
1019 return prepare_for_direct_item(p_s_path, p_le_ih, inode,
1020 n_new_file_length, p_n_cut_size);
1022 /* Case of an indirect item. */
1024 int n_unfm_number, /* Number of the item unformatted nodes. */
1025 n_counter, n_blk_size;
1026 __le32 *p_n_unfm_pointer; /* Pointer to the unformatted node number. */
1028 struct item_head s_ih; /* Item header. */
1029 char c_mode; /* Returned mode of the balance. */
1032 n_blk_size = p_s_sb->s_blocksize;
1034 /* Search for the needed object indirect item until there are no unformatted nodes to be removed. */
1037 p_s_bh = PATH_PLAST_BUFFER(p_s_path);
1038 /* Copy indirect item header to a temp variable. */
1039 copy_item_head(&s_ih, PATH_PITEM_HEAD(p_s_path));
1040 /* Calculate number of unformatted nodes in this item. */
1041 n_unfm_number = I_UNFM_NUM(&s_ih);
1043 RFALSE(!is_indirect_le_ih(&s_ih) || !n_unfm_number ||
1044 pos_in_item(p_s_path) + 1 != n_unfm_number,
1045 "PAP-5240: invalid item %h "
1046 "n_unfm_number = %d *p_n_pos_in_item = %d",
1047 &s_ih, n_unfm_number, pos_in_item(p_s_path));
1049 /* Calculate balance mode and position in the item to remove unformatted nodes. */
1050 if (n_new_file_length == max_reiserfs_offset(inode)) { /* Case of delete. */
1051 pos_in_item(p_s_path) = 0;
1052 *p_n_cut_size = -(IH_SIZE + ih_item_len(&s_ih));
1054 } else { /* Case of truncate. */
1055 if (n_new_file_length < le_ih_k_offset(&s_ih)) {
1056 pos_in_item(p_s_path) = 0;
1058 -(IH_SIZE + ih_item_len(&s_ih));
1059 c_mode = M_DELETE; /* Delete this item. */
1061 /* indirect item must be truncated starting from *p_n_pos_in_item-th position */
1062 pos_in_item(p_s_path) =
1063 (n_new_file_length + n_blk_size -
1064 le_ih_k_offset(&s_ih)) >> p_s_sb->
1067 RFALSE(pos_in_item(p_s_path) >
1069 "PAP-5250: invalid position in the item");
1071 /* Either convert last unformatted node of indirect item to direct item or increase
1073 if (pos_in_item(p_s_path) ==
1075 *p_n_cut_size = 0; /* Nothing to cut. */
1076 return M_CONVERT; /* Maybe convert last unformatted node to the direct item. */
1078 /* Calculate size to cut. */
1080 -(ih_item_len(&s_ih) -
1081 pos_in_item(p_s_path) *
1084 c_mode = M_CUT; /* Cut from this indirect item. */
1088 RFALSE(n_unfm_number <= pos_in_item(p_s_path),
1089 "PAP-5260: invalid position in the indirect item");
1091 /* pointers to be cut */
1092 n_unfm_number -= pos_in_item(p_s_path);
1093 /* Set pointer to the last unformatted node pointer that is to be cut. */
1095 (__le32 *) B_I_PITEM(p_s_bh,
1096 &s_ih) + I_UNFM_NUM(&s_ih) -
1099 /* We go through the unformatted nodes pointers of the indirect
1100 item and look for the unformatted nodes in the cache. If we
1101 found some of them we free it, zero corresponding indirect item
1102 entry and log buffer containing that indirect item. For this we
1103 need to prepare last path element for logging. If some
1104 unformatted node has b_count > 1 we must not free this
1105 unformatted node since it is in use. */
1106 reiserfs_prepare_for_journal(p_s_sb, p_s_bh, 1);
1107 // note: path could be changed, first line in for loop takes care
1110 for (n_counter = *p_n_removed;
1111 n_counter < n_unfm_number;
1112 n_counter++, p_n_unfm_pointer--) {
1115 if (item_moved(&s_ih, p_s_path)) {
1119 RFALSE(p_n_unfm_pointer <
1120 (__le32 *) B_I_PITEM(p_s_bh, &s_ih)
1121 || p_n_unfm_pointer >
1122 (__le32 *) B_I_PITEM(p_s_bh,
1124 I_UNFM_NUM(&s_ih) - 1,
1125 "vs-5265: pointer out of range");
1127 /* Hole, nothing to remove. */
1128 if (!get_block_num(p_n_unfm_pointer, 0)) {
1135 tmp = get_block_num(p_n_unfm_pointer, 0);
1136 put_block_num(p_n_unfm_pointer, 0, 0);
1137 journal_mark_dirty(th, p_s_sb, p_s_bh);
1138 reiserfs_free_block(th, inode, tmp, 1);
1139 if (item_moved(&s_ih, p_s_path)) {
1145 /* a trick. If the buffer has been logged, this
1146 ** will do nothing. If we've broken the loop without
1147 ** logging it, it will restore the buffer
1150 reiserfs_restore_prepared_buffer(p_s_sb, p_s_bh);
1152 /* This loop can be optimized. */
1153 } while ((*p_n_removed < n_unfm_number || need_research) &&
1154 search_for_position_by_key(p_s_sb, p_s_item_key,
1158 RFALSE(*p_n_removed < n_unfm_number,
1159 "PAP-5310: indirect item is not found");
1160 RFALSE(item_moved(&s_ih, p_s_path),
1161 "after while, comp failed, retry");
1163 if (c_mode == M_CUT)
1164 pos_in_item(p_s_path) *= UNFM_P_SIZE;
1169 /* Calculate number of bytes which will be deleted or cut during balance */
1170 static int calc_deleted_bytes_number(struct tree_balance *p_s_tb, char c_mode)
1173 struct item_head *p_le_ih = PATH_PITEM_HEAD(p_s_tb->tb_path);
1175 if (is_statdata_le_ih(p_le_ih))
1180 M_DELETE) ? ih_item_len(p_le_ih) : -p_s_tb->insert_size[0];
1181 if (is_direntry_le_ih(p_le_ih)) {
1182 // return EMPTY_DIR_SIZE; /* We delete emty directoris only. */
1183 // we can't use EMPTY_DIR_SIZE, as old format dirs have a different
1184 // empty size. ick. FIXME, is this right?
1189 if (is_indirect_le_ih(p_le_ih))
1190 n_del_size = (n_del_size / UNFM_P_SIZE) * (PATH_PLAST_BUFFER(p_s_tb->tb_path)->b_size); // - get_ih_free_space (p_le_ih);
1194 static void init_tb_struct(struct reiserfs_transaction_handle *th,
1195 struct tree_balance *p_s_tb,
1196 struct super_block *p_s_sb,
1197 struct path *p_s_path, int n_size)
1200 BUG_ON(!th->t_trans_id);
1202 memset(p_s_tb, '\0', sizeof(struct tree_balance));
1203 p_s_tb->transaction_handle = th;
1204 p_s_tb->tb_sb = p_s_sb;
1205 p_s_tb->tb_path = p_s_path;
1206 PATH_OFFSET_PBUFFER(p_s_path, ILLEGAL_PATH_ELEMENT_OFFSET) = NULL;
1207 PATH_OFFSET_POSITION(p_s_path, ILLEGAL_PATH_ELEMENT_OFFSET) = 0;
1208 p_s_tb->insert_size[0] = n_size;
1211 void padd_item(char *item, int total_length, int length)
1215 for (i = total_length; i > length;)
1219 #ifdef REISERQUOTA_DEBUG
1220 char key2type(struct reiserfs_key *ih)
1222 if (is_direntry_le_key(2, ih))
1224 if (is_direct_le_key(2, ih))
1226 if (is_indirect_le_key(2, ih))
1228 if (is_statdata_le_key(2, ih))
1233 char head2type(struct item_head *ih)
1235 if (is_direntry_le_ih(ih))
1237 if (is_direct_le_ih(ih))
1239 if (is_indirect_le_ih(ih))
1241 if (is_statdata_le_ih(ih))
1247 /* Delete object item. */
1248 int reiserfs_delete_item(struct reiserfs_transaction_handle *th, struct path *p_s_path, /* Path to the deleted item. */
1249 const struct cpu_key *p_s_item_key, /* Key to search for the deleted item. */
1250 struct inode *p_s_inode, /* inode is here just to update i_blocks and quotas */
1251 struct buffer_head *p_s_un_bh)
1252 { /* NULL or unformatted node pointer. */
1253 struct super_block *p_s_sb = p_s_inode->i_sb;
1254 struct tree_balance s_del_balance;
1255 struct item_head s_ih;
1256 struct item_head *q_ih;
1257 int quota_cut_bytes;
1258 int n_ret_value, n_del_size, n_removed;
1260 #ifdef CONFIG_REISERFS_CHECK
1265 BUG_ON(!th->t_trans_id);
1267 init_tb_struct(th, &s_del_balance, p_s_sb, p_s_path,
1268 0 /*size is unknown */ );
1273 #ifdef CONFIG_REISERFS_CHECK
1277 prepare_for_delete_or_cut(th, p_s_inode, p_s_path,
1278 p_s_item_key, &n_removed,
1280 max_reiserfs_offset(p_s_inode));
1282 RFALSE(c_mode != M_DELETE, "PAP-5320: mode must be M_DELETE");
1284 copy_item_head(&s_ih, PATH_PITEM_HEAD(p_s_path));
1285 s_del_balance.insert_size[0] = n_del_size;
1287 n_ret_value = fix_nodes(M_DELETE, &s_del_balance, NULL, NULL);
1288 if (n_ret_value != REPEAT_SEARCH)
1291 PROC_INFO_INC(p_s_sb, delete_item_restarted);
1293 // file system changed, repeat search
1295 search_for_position_by_key(p_s_sb, p_s_item_key, p_s_path);
1296 if (n_ret_value == IO_ERROR)
1298 if (n_ret_value == FILE_NOT_FOUND) {
1299 reiserfs_warning(p_s_sb,
1300 "vs-5340: reiserfs_delete_item: "
1301 "no items of the file %K found",
1307 if (n_ret_value != CARRY_ON) {
1308 unfix_nodes(&s_del_balance);
1311 // reiserfs_delete_item returns item length when success
1312 n_ret_value = calc_deleted_bytes_number(&s_del_balance, M_DELETE);
1313 q_ih = get_ih(p_s_path);
1314 quota_cut_bytes = ih_item_len(q_ih);
1316 /* hack so the quota code doesn't have to guess if the file
1317 ** has a tail. On tail insert, we allocate quota for 1 unformatted node.
1318 ** We test the offset because the tail might have been
1319 ** split into multiple items, and we only want to decrement for
1320 ** the unfm node once
1322 if (!S_ISLNK(p_s_inode->i_mode) && is_direct_le_ih(q_ih)) {
1323 if ((le_ih_k_offset(q_ih) & (p_s_sb->s_blocksize - 1)) == 1) {
1324 quota_cut_bytes = p_s_sb->s_blocksize + UNFM_P_SIZE;
1326 quota_cut_bytes = 0;
1334 /* We are in direct2indirect conversion, so move tail contents
1335 to the unformatted node */
1336 /* note, we do the copy before preparing the buffer because we
1337 ** don't care about the contents of the unformatted node yet.
1338 ** the only thing we really care about is the direct item's data
1339 ** is in the unformatted node.
1341 ** Otherwise, we would have to call reiserfs_prepare_for_journal on
1342 ** the unformatted node, which might schedule, meaning we'd have to
1343 ** loop all the way back up to the start of the while loop.
1345 ** The unformatted node must be dirtied later on. We can't be
1346 ** sure here if the entire tail has been deleted yet.
1348 ** p_s_un_bh is from the page cache (all unformatted nodes are
1349 ** from the page cache) and might be a highmem page. So, we
1350 ** can't use p_s_un_bh->b_data.
1354 data = kmap_atomic(p_s_un_bh->b_page, KM_USER0);
1355 off = ((le_ih_k_offset(&s_ih) - 1) & (PAGE_CACHE_SIZE - 1));
1357 B_I_PITEM(PATH_PLAST_BUFFER(p_s_path), &s_ih),
1359 kunmap_atomic(data, KM_USER0);
1361 /* Perform balancing after all resources have been collected at once. */
1362 do_balance(&s_del_balance, NULL, NULL, M_DELETE);
1364 #ifdef REISERQUOTA_DEBUG
1365 reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE,
1366 "reiserquota delete_item(): freeing %u, id=%u type=%c",
1367 quota_cut_bytes, p_s_inode->i_uid, head2type(&s_ih));
1369 DLIMIT_FREE_SPACE(p_s_inode, quota_cut_bytes);
1370 DQUOT_FREE_SPACE_NODIRTY(p_s_inode, quota_cut_bytes);
1372 /* Return deleted body length */
1376 /* Summary Of Mechanisms For Handling Collisions Between Processes:
1378 deletion of the body of the object is performed by iput(), with the
1379 result that if multiple processes are operating on a file, the
1380 deletion of the body of the file is deferred until the last process
1381 that has an open inode performs its iput().
1383 writes and truncates are protected from collisions by use of
1386 creates, linking, and mknod are protected from collisions with other
1387 processes by making the reiserfs_add_entry() the last step in the
1388 creation, and then rolling back all changes if there was a collision.
1392 /* this deletes item which never gets split */
1393 void reiserfs_delete_solid_item(struct reiserfs_transaction_handle *th,
1394 struct inode *inode, struct reiserfs_key *key)
1396 struct tree_balance tb;
1397 INITIALIZE_PATH(path);
1400 struct cpu_key cpu_key;
1402 int quota_cut_bytes = 0;
1404 BUG_ON(!th->t_trans_id);
1406 le_key2cpu_key(&cpu_key, key);
1409 retval = search_item(th->t_super, &cpu_key, &path);
1410 if (retval == IO_ERROR) {
1411 reiserfs_warning(th->t_super,
1412 "vs-5350: reiserfs_delete_solid_item: "
1413 "i/o failure occurred trying to delete %K",
1417 if (retval != ITEM_FOUND) {
1419 // No need for a warning, if there is just no free space to insert '..' item into the newly-created subdir
1421 ((unsigned long long)
1422 GET_HASH_VALUE(le_key_k_offset
1423 (le_key_version(key), key)) == 0
1424 && (unsigned long long)
1425 GET_GENERATION_NUMBER(le_key_k_offset
1426 (le_key_version(key),
1428 reiserfs_warning(th->t_super,
1429 "vs-5355: reiserfs_delete_solid_item: %k not found",
1435 item_len = ih_item_len(PATH_PITEM_HEAD(&path));
1436 init_tb_struct(th, &tb, th->t_super, &path,
1437 -(IH_SIZE + item_len));
1439 quota_cut_bytes = ih_item_len(PATH_PITEM_HEAD(&path));
1441 retval = fix_nodes(M_DELETE, &tb, NULL, NULL);
1442 if (retval == REPEAT_SEARCH) {
1443 PROC_INFO_INC(th->t_super, delete_solid_item_restarted);
1447 if (retval == CARRY_ON) {
1448 do_balance(&tb, NULL, NULL, M_DELETE);
1449 if (inode) { /* Should we count quota for item? (we don't count quotas for save-links) */
1450 #ifdef REISERQUOTA_DEBUG
1451 reiserfs_debug(th->t_super, REISERFS_DEBUG_CODE,
1452 "reiserquota delete_solid_item(): freeing %u id=%u type=%c",
1453 quota_cut_bytes, inode->i_uid,
1456 DLIMIT_FREE_SPACE(inode, quota_cut_bytes);
1457 DQUOT_FREE_SPACE_NODIRTY(inode,
1462 // IO_ERROR, NO_DISK_SPACE, etc
1463 reiserfs_warning(th->t_super,
1464 "vs-5360: reiserfs_delete_solid_item: "
1465 "could not delete %K due to fix_nodes failure",
1471 reiserfs_check_path(&path);
1474 int reiserfs_delete_object(struct reiserfs_transaction_handle *th,
1475 struct inode *inode)
1479 BUG_ON(!th->t_trans_id);
1481 /* for directory this deletes item containing "." and ".." */
1483 reiserfs_do_truncate(th, inode, NULL, 0 /*no timestamp updates */ );
1487 #if defined( USE_INODE_GENERATION_COUNTER )
1488 if (!old_format_only(th->t_super)) {
1489 __le32 *inode_generation;
1492 &REISERFS_SB(th->t_super)->s_rs->s_inode_generation;
1494 cpu_to_le32(le32_to_cpu(*inode_generation) + 1);
1496 /* USE_INODE_GENERATION_COUNTER */
1498 reiserfs_delete_solid_item(th, inode, INODE_PKEY(inode));
1503 static void unmap_buffers(struct page *page, loff_t pos)
1505 struct buffer_head *bh;
1506 struct buffer_head *head;
1507 struct buffer_head *next;
1508 unsigned long tail_index;
1509 unsigned long cur_index;
1512 if (page_has_buffers(page)) {
1513 tail_index = pos & (PAGE_CACHE_SIZE - 1);
1515 head = page_buffers(page);
1518 next = bh->b_this_page;
1520 /* we want to unmap the buffers that contain the tail, and
1521 ** all the buffers after it (since the tail must be at the
1522 ** end of the file). We don't want to unmap file data
1523 ** before the tail, since it might be dirty and waiting to
1526 cur_index += bh->b_size;
1527 if (cur_index > tail_index) {
1528 reiserfs_unmap_buffer(bh);
1531 } while (bh != head);
1532 if (PAGE_SIZE == bh->b_size) {
1533 clear_page_dirty(page);
1539 static int maybe_indirect_to_direct(struct reiserfs_transaction_handle *th,
1540 struct inode *p_s_inode,
1542 struct path *p_s_path,
1543 const struct cpu_key *p_s_item_key,
1544 loff_t n_new_file_size, char *p_c_mode)
1546 struct super_block *p_s_sb = p_s_inode->i_sb;
1547 int n_block_size = p_s_sb->s_blocksize;
1549 BUG_ON(!th->t_trans_id);
1551 if (n_new_file_size != p_s_inode->i_size)
1554 /* the page being sent in could be NULL if there was an i/o error
1555 ** reading in the last block. The user will hit problems trying to
1556 ** read the file, but for now we just skip the indirect2direct
1558 if (atomic_read(&p_s_inode->i_count) > 1 ||
1559 !tail_has_to_be_packed(p_s_inode) ||
1560 !page || (REISERFS_I(p_s_inode)->i_flags & i_nopack_mask)) {
1561 // leave tail in an unformatted node
1562 *p_c_mode = M_SKIP_BALANCING;
1564 n_block_size - (n_new_file_size & (n_block_size - 1));
1565 pathrelse(p_s_path);
1568 /* Permorm the conversion to a direct_item. */
1569 /*return indirect_to_direct (p_s_inode, p_s_path, p_s_item_key, n_new_file_size, p_c_mode); */
1570 return indirect2direct(th, p_s_inode, page, p_s_path, p_s_item_key,
1571 n_new_file_size, p_c_mode);
1574 /* we did indirect_to_direct conversion. And we have inserted direct
1575 item successesfully, but there were no disk space to cut unfm
1576 pointer being converted. Therefore we have to delete inserted
1578 static void indirect_to_direct_roll_back(struct reiserfs_transaction_handle *th,
1579 struct inode *inode, struct path *path)
1581 struct cpu_key tail_key;
1584 BUG_ON(!th->t_trans_id);
1586 make_cpu_key(&tail_key, inode, inode->i_size + 1, TYPE_DIRECT, 4); // !!!!
1587 tail_key.key_length = 4;
1590 (cpu_key_k_offset(&tail_key) & (inode->i_sb->s_blocksize - 1)) - 1;
1592 /* look for the last byte of the tail */
1593 if (search_for_position_by_key(inode->i_sb, &tail_key, path) ==
1595 reiserfs_panic(inode->i_sb,
1596 "vs-5615: indirect_to_direct_roll_back: found invalid item");
1597 RFALSE(path->pos_in_item !=
1598 ih_item_len(PATH_PITEM_HEAD(path)) - 1,
1599 "vs-5616: appended bytes found");
1600 PATH_LAST_POSITION(path)--;
1603 reiserfs_delete_item(th, path, &tail_key, inode,
1604 NULL /*unbh not needed */ );
1606 || removed > tail_len,
1607 "vs-5617: there was tail %d bytes, removed item length %d bytes",
1609 tail_len -= removed;
1610 set_cpu_key_k_offset(&tail_key,
1611 cpu_key_k_offset(&tail_key) - removed);
1613 reiserfs_warning(inode->i_sb,
1614 "indirect_to_direct_roll_back: indirect_to_direct conversion has been rolled back due to lack of disk space");
1615 //mark_file_without_tail (inode);
1616 mark_inode_dirty(inode);
1619 /* (Truncate or cut entry) or delete object item. Returns < 0 on failure */
1620 int reiserfs_cut_from_item(struct reiserfs_transaction_handle *th,
1621 struct path *p_s_path,
1622 struct cpu_key *p_s_item_key,
1623 struct inode *p_s_inode,
1624 struct page *page, loff_t n_new_file_size)
1626 struct super_block *p_s_sb = p_s_inode->i_sb;
1627 /* Every function which is going to call do_balance must first
1628 create a tree_balance structure. Then it must fill up this
1629 structure by using the init_tb_struct and fix_nodes functions.
1630 After that we can make tree balancing. */
1631 struct tree_balance s_cut_balance;
1632 struct item_head *p_le_ih;
1633 int n_cut_size = 0, /* Amount to be cut. */
1634 n_ret_value = CARRY_ON, n_removed = 0, /* Number of the removed unformatted nodes. */
1635 n_is_inode_locked = 0;
1636 char c_mode; /* Mode of the balance. */
1638 int quota_cut_bytes;
1639 loff_t tail_pos = 0;
1641 BUG_ON(!th->t_trans_id);
1643 init_tb_struct(th, &s_cut_balance, p_s_inode->i_sb, p_s_path,
1646 /* Repeat this loop until we either cut the item without needing
1647 to balance, or we fix_nodes without schedule occurring */
1649 /* Determine the balance mode, position of the first byte to
1650 be cut, and size to be cut. In case of the indirect item
1651 free unformatted nodes which are pointed to by the cut
1655 prepare_for_delete_or_cut(th, p_s_inode, p_s_path,
1656 p_s_item_key, &n_removed,
1657 &n_cut_size, n_new_file_size);
1658 if (c_mode == M_CONVERT) {
1659 /* convert last unformatted node to direct item or leave
1660 tail in the unformatted node */
1661 RFALSE(n_ret_value != CARRY_ON,
1662 "PAP-5570: can not convert twice");
1665 maybe_indirect_to_direct(th, p_s_inode, page,
1666 p_s_path, p_s_item_key,
1667 n_new_file_size, &c_mode);
1668 if (c_mode == M_SKIP_BALANCING)
1669 /* tail has been left in the unformatted node */
1672 n_is_inode_locked = 1;
1674 /* removing of last unformatted node will change value we
1675 have to return to truncate. Save it */
1676 retval2 = n_ret_value;
1677 /*retval2 = p_s_sb->s_blocksize - (n_new_file_size & (p_s_sb->s_blocksize - 1)); */
1679 /* So, we have performed the first part of the conversion:
1680 inserting the new direct item. Now we are removing the
1681 last unformatted node pointer. Set key to search for
1683 set_cpu_key_k_type(p_s_item_key, TYPE_INDIRECT);
1684 p_s_item_key->key_length = 4;
1686 (n_new_file_size & (p_s_sb->s_blocksize - 1));
1687 tail_pos = n_new_file_size;
1688 set_cpu_key_k_offset(p_s_item_key, n_new_file_size + 1);
1689 if (search_for_position_by_key
1690 (p_s_sb, p_s_item_key,
1691 p_s_path) == POSITION_NOT_FOUND) {
1692 print_block(PATH_PLAST_BUFFER(p_s_path), 3,
1693 PATH_LAST_POSITION(p_s_path) - 1,
1694 PATH_LAST_POSITION(p_s_path) + 1);
1695 reiserfs_panic(p_s_sb,
1696 "PAP-5580: reiserfs_cut_from_item: item to convert does not exist (%K)",
1701 if (n_cut_size == 0) {
1702 pathrelse(p_s_path);
1706 s_cut_balance.insert_size[0] = n_cut_size;
1708 n_ret_value = fix_nodes(c_mode, &s_cut_balance, NULL, NULL);
1709 if (n_ret_value != REPEAT_SEARCH)
1712 PROC_INFO_INC(p_s_sb, cut_from_item_restarted);
1715 search_for_position_by_key(p_s_sb, p_s_item_key, p_s_path);
1716 if (n_ret_value == POSITION_FOUND)
1719 reiserfs_warning(p_s_sb,
1720 "PAP-5610: reiserfs_cut_from_item: item %K not found",
1722 unfix_nodes(&s_cut_balance);
1723 return (n_ret_value == IO_ERROR) ? -EIO : -ENOENT;
1726 // check fix_nodes results (IO_ERROR or NO_DISK_SPACE)
1727 if (n_ret_value != CARRY_ON) {
1728 if (n_is_inode_locked) {
1729 // FIXME: this seems to be not needed: we are always able
1731 indirect_to_direct_roll_back(th, p_s_inode, p_s_path);
1733 if (n_ret_value == NO_DISK_SPACE)
1734 reiserfs_warning(p_s_sb, "NO_DISK_SPACE");
1735 unfix_nodes(&s_cut_balance);
1739 /* go ahead and perform balancing */
1741 RFALSE(c_mode == M_PASTE || c_mode == M_INSERT, "invalid mode");
1743 /* Calculate number of bytes that need to be cut from the item. */
1746 M_DELETE) ? ih_item_len(get_ih(p_s_path)) : -s_cut_balance.
1749 n_ret_value = calc_deleted_bytes_number(&s_cut_balance, c_mode);
1751 n_ret_value = retval2;
1753 /* For direct items, we only change the quota when deleting the last
1756 p_le_ih = PATH_PITEM_HEAD(s_cut_balance.tb_path);
1757 if (!S_ISLNK(p_s_inode->i_mode) && is_direct_le_ih(p_le_ih)) {
1758 if (c_mode == M_DELETE &&
1759 (le_ih_k_offset(p_le_ih) & (p_s_sb->s_blocksize - 1)) ==
1761 // FIXME: this is to keep 3.5 happy
1762 REISERFS_I(p_s_inode)->i_first_direct_byte = U32_MAX;
1763 quota_cut_bytes = p_s_sb->s_blocksize + UNFM_P_SIZE;
1765 quota_cut_bytes = 0;
1768 #ifdef CONFIG_REISERFS_CHECK
1769 if (n_is_inode_locked) {
1770 struct item_head *le_ih =
1771 PATH_PITEM_HEAD(s_cut_balance.tb_path);
1772 /* we are going to complete indirect2direct conversion. Make
1773 sure, that we exactly remove last unformatted node pointer
1775 if (!is_indirect_le_ih(le_ih))
1776 reiserfs_panic(p_s_sb,
1777 "vs-5652: reiserfs_cut_from_item: "
1778 "item must be indirect %h", le_ih);
1780 if (c_mode == M_DELETE && ih_item_len(le_ih) != UNFM_P_SIZE)
1781 reiserfs_panic(p_s_sb,
1782 "vs-5653: reiserfs_cut_from_item: "
1783 "completing indirect2direct conversion indirect item %h "
1784 "being deleted must be of 4 byte long",
1788 && s_cut_balance.insert_size[0] != -UNFM_P_SIZE) {
1789 reiserfs_panic(p_s_sb,
1790 "vs-5654: reiserfs_cut_from_item: "
1791 "can not complete indirect2direct conversion of %h (CUT, insert_size==%d)",
1792 le_ih, s_cut_balance.insert_size[0]);
1794 /* it would be useful to make sure, that right neighboring
1795 item is direct item of this file */
1799 do_balance(&s_cut_balance, NULL, NULL, c_mode);
1800 if (n_is_inode_locked) {
1801 /* we've done an indirect->direct conversion. when the data block
1802 ** was freed, it was removed from the list of blocks that must
1803 ** be flushed before the transaction commits, make sure to
1804 ** unmap and invalidate it
1806 unmap_buffers(page, tail_pos);
1807 REISERFS_I(p_s_inode)->i_flags &= ~i_pack_on_close_mask;
1809 #ifdef REISERQUOTA_DEBUG
1810 reiserfs_debug(p_s_inode->i_sb, REISERFS_DEBUG_CODE,
1811 "reiserquota cut_from_item(): freeing %u id=%u type=%c",
1812 quota_cut_bytes, p_s_inode->i_uid, '?');
1814 DLIMIT_FREE_SPACE(p_s_inode, quota_cut_bytes);
1815 DQUOT_FREE_SPACE_NODIRTY(p_s_inode, quota_cut_bytes);
1819 static void truncate_directory(struct reiserfs_transaction_handle *th,
1820 struct inode *inode)
1822 BUG_ON(!th->t_trans_id);
1824 reiserfs_warning(inode->i_sb,
1825 "vs-5655: truncate_directory: link count != 0");
1827 set_le_key_k_offset(KEY_FORMAT_3_5, INODE_PKEY(inode), DOT_OFFSET);
1828 set_le_key_k_type(KEY_FORMAT_3_5, INODE_PKEY(inode), TYPE_DIRENTRY);
1829 reiserfs_delete_solid_item(th, inode, INODE_PKEY(inode));
1830 reiserfs_update_sd(th, inode);
1831 set_le_key_k_offset(KEY_FORMAT_3_5, INODE_PKEY(inode), SD_OFFSET);
1832 set_le_key_k_type(KEY_FORMAT_3_5, INODE_PKEY(inode), TYPE_STAT_DATA);
1835 /* Truncate file to the new size. Note, this must be called with a transaction
1837 int reiserfs_do_truncate(struct reiserfs_transaction_handle *th, struct inode *p_s_inode, /* ->i_size contains new
1839 struct page *page, /* up to date for last block */
1840 int update_timestamps /* when it is called by
1841 file_release to convert
1842 the tail - no timestamps
1843 should be updated */
1846 INITIALIZE_PATH(s_search_path); /* Path to the current object item. */
1847 struct item_head *p_le_ih; /* Pointer to an item header. */
1848 struct cpu_key s_item_key; /* Key to search for a previous file item. */
1849 loff_t n_file_size, /* Old file size. */
1850 n_new_file_size; /* New file size. */
1851 int n_deleted; /* Number of deleted or truncated bytes. */
1855 BUG_ON(!th->t_trans_id);
1857 (S_ISREG(p_s_inode->i_mode) || S_ISDIR(p_s_inode->i_mode)
1858 || S_ISLNK(p_s_inode->i_mode)))
1861 if (S_ISDIR(p_s_inode->i_mode)) {
1862 // deletion of directory - no need to update timestamps
1863 truncate_directory(th, p_s_inode);
1867 /* Get new file size. */
1868 n_new_file_size = p_s_inode->i_size;
1870 // FIXME: note, that key type is unimportant here
1871 make_cpu_key(&s_item_key, p_s_inode, max_reiserfs_offset(p_s_inode),
1875 search_for_position_by_key(p_s_inode->i_sb, &s_item_key,
1877 if (retval == IO_ERROR) {
1878 reiserfs_warning(p_s_inode->i_sb,
1879 "vs-5657: reiserfs_do_truncate: "
1880 "i/o failure occurred trying to truncate %K",
1885 if (retval == POSITION_FOUND || retval == FILE_NOT_FOUND) {
1886 reiserfs_warning(p_s_inode->i_sb,
1887 "PAP-5660: reiserfs_do_truncate: "
1888 "wrong result %d of search for %K", retval,
1895 s_search_path.pos_in_item--;
1897 /* Get real file size (total length of all file items) */
1898 p_le_ih = PATH_PITEM_HEAD(&s_search_path);
1899 if (is_statdata_le_ih(p_le_ih))
1902 loff_t offset = le_ih_k_offset(p_le_ih);
1904 op_bytes_number(p_le_ih, p_s_inode->i_sb->s_blocksize);
1906 /* this may mismatch with real file size: if last direct item
1907 had no padding zeros and last unformatted node had no free
1908 space, this file would have this file size */
1909 n_file_size = offset + bytes - 1;
1912 * are we doing a full truncate or delete, if so
1913 * kick in the reada code
1915 if (n_new_file_size == 0)
1916 s_search_path.reada = PATH_READA | PATH_READA_BACK;
1918 if (n_file_size == 0 || n_file_size < n_new_file_size) {
1919 goto update_and_out;
1922 /* Update key to search for the last file item. */
1923 set_cpu_key_k_offset(&s_item_key, n_file_size);
1926 /* Cut or delete file item. */
1928 reiserfs_cut_from_item(th, &s_search_path, &s_item_key,
1929 p_s_inode, page, n_new_file_size);
1930 if (n_deleted < 0) {
1931 reiserfs_warning(p_s_inode->i_sb,
1932 "vs-5665: reiserfs_do_truncate: reiserfs_cut_from_item failed");
1933 reiserfs_check_path(&s_search_path);
1937 RFALSE(n_deleted > n_file_size,
1938 "PAP-5670: reiserfs_cut_from_item: too many bytes deleted: deleted %d, file_size %lu, item_key %K",
1939 n_deleted, n_file_size, &s_item_key);
1941 /* Change key to search the last file item. */
1942 n_file_size -= n_deleted;
1944 set_cpu_key_k_offset(&s_item_key, n_file_size);
1946 /* While there are bytes to truncate and previous file item is presented in the tree. */
1949 ** This loop could take a really long time, and could log
1950 ** many more blocks than a transaction can hold. So, we do a polite
1951 ** journal end here, and if the transaction needs ending, we make
1952 ** sure the file is consistent before ending the current trans
1953 ** and starting a new one
1955 if (journal_transaction_should_end(th, th->t_blocks_allocated)) {
1956 int orig_len_alloc = th->t_blocks_allocated;
1957 decrement_counters_in_path(&s_search_path);
1959 if (update_timestamps) {
1960 p_s_inode->i_mtime = p_s_inode->i_ctime =
1963 reiserfs_update_sd(th, p_s_inode);
1965 err = journal_end(th, p_s_inode->i_sb, orig_len_alloc);
1968 err = journal_begin(th, p_s_inode->i_sb,
1969 JOURNAL_PER_BALANCE_CNT * 6);
1972 reiserfs_update_inode_transaction(p_s_inode);
1974 } while (n_file_size > ROUND_UP(n_new_file_size) &&
1975 search_for_position_by_key(p_s_inode->i_sb, &s_item_key,
1976 &s_search_path) == POSITION_FOUND);
1978 RFALSE(n_file_size > ROUND_UP(n_new_file_size),
1979 "PAP-5680: truncate did not finish: new_file_size %Ld, current %Ld, oid %d",
1980 n_new_file_size, n_file_size, s_item_key.on_disk_key.k_objectid);
1983 if (update_timestamps) {
1984 // this is truncate, not file closing
1985 p_s_inode->i_mtime = p_s_inode->i_ctime = CURRENT_TIME_SEC;
1987 reiserfs_update_sd(th, p_s_inode);
1990 pathrelse(&s_search_path);
1994 #ifdef CONFIG_REISERFS_CHECK
1995 // this makes sure, that we __append__, not overwrite or add holes
1996 static void check_research_for_paste(struct path *path,
1997 const struct cpu_key *p_s_key)
1999 struct item_head *found_ih = get_ih(path);
2001 if (is_direct_le_ih(found_ih)) {
2002 if (le_ih_k_offset(found_ih) +
2003 op_bytes_number(found_ih,
2004 get_last_bh(path)->b_size) !=
2005 cpu_key_k_offset(p_s_key)
2006 || op_bytes_number(found_ih,
2007 get_last_bh(path)->b_size) !=
2009 reiserfs_panic(NULL,
2010 "PAP-5720: check_research_for_paste: "
2011 "found direct item %h or position (%d) does not match to key %K",
2012 found_ih, pos_in_item(path), p_s_key);
2014 if (is_indirect_le_ih(found_ih)) {
2015 if (le_ih_k_offset(found_ih) +
2016 op_bytes_number(found_ih,
2017 get_last_bh(path)->b_size) !=
2018 cpu_key_k_offset(p_s_key)
2019 || I_UNFM_NUM(found_ih) != pos_in_item(path)
2020 || get_ih_free_space(found_ih) != 0)
2021 reiserfs_panic(NULL,
2022 "PAP-5730: check_research_for_paste: "
2023 "found indirect item (%h) or position (%d) does not match to key (%K)",
2024 found_ih, pos_in_item(path), p_s_key);
2027 #endif /* config reiserfs check */
2029 /* Paste bytes to the existing item. Returns bytes number pasted into the item. */
2030 int reiserfs_paste_into_item(struct reiserfs_transaction_handle *th, struct path *p_s_search_path, /* Path to the pasted item. */
2031 const struct cpu_key *p_s_key, /* Key to search for the needed item. */
2032 struct inode *inode, /* Inode item belongs to */
2033 const char *p_c_body, /* Pointer to the bytes to paste. */
2035 { /* Size of pasted bytes. */
2036 struct tree_balance s_paste_balance;
2040 BUG_ON(!th->t_trans_id);
2042 fs_gen = get_generation(inode->i_sb);
2044 #ifdef REISERQUOTA_DEBUG
2045 reiserfs_debug(inode->i_sb, REISERFS_DEBUG_CODE,
2046 "reiserquota paste_into_item(): allocating %u id=%u type=%c",
2047 n_pasted_size, inode->i_uid,
2048 key2type(&(p_s_key->on_disk_key)));
2051 if (DQUOT_ALLOC_SPACE_NODIRTY(inode, n_pasted_size)) {
2052 pathrelse(p_s_search_path);
2055 if (DLIMIT_ALLOC_SPACE(inode, n_pasted_size)) {
2056 DQUOT_FREE_SPACE_NODIRTY(inode, n_pasted_size);
2057 pathrelse(p_s_search_path);
2060 init_tb_struct(th, &s_paste_balance, th->t_super, p_s_search_path,
2062 #ifdef DISPLACE_NEW_PACKING_LOCALITIES
2063 s_paste_balance.key = p_s_key->on_disk_key;
2066 /* DQUOT_* can schedule, must check before the fix_nodes */
2067 if (fs_changed(fs_gen, inode->i_sb)) {
2072 fix_nodes(M_PASTE, &s_paste_balance, NULL,
2073 p_c_body)) == REPEAT_SEARCH) {
2075 /* file system changed while we were in the fix_nodes */
2076 PROC_INFO_INC(th->t_super, paste_into_item_restarted);
2078 search_for_position_by_key(th->t_super, p_s_key,
2080 if (retval == IO_ERROR) {
2084 if (retval == POSITION_FOUND) {
2085 reiserfs_warning(inode->i_sb,
2086 "PAP-5710: reiserfs_paste_into_item: entry or pasted byte (%K) exists",
2091 #ifdef CONFIG_REISERFS_CHECK
2092 check_research_for_paste(p_s_search_path, p_s_key);
2096 /* Perform balancing after all resources are collected by fix_nodes, and
2097 accessing them will not risk triggering schedule. */
2098 if (retval == CARRY_ON) {
2099 do_balance(&s_paste_balance, NULL /*ih */ , p_c_body, M_PASTE);
2102 retval = (retval == NO_DISK_SPACE) ? -ENOSPC : -EIO;
2104 /* this also releases the path */
2105 unfix_nodes(&s_paste_balance);
2106 #ifdef REISERQUOTA_DEBUG
2107 reiserfs_debug(inode->i_sb, REISERFS_DEBUG_CODE,
2108 "reiserquota paste_into_item(): freeing %u id=%u type=%c",
2109 n_pasted_size, inode->i_uid,
2110 key2type(&(p_s_key->on_disk_key)));
2112 DLIMIT_FREE_SPACE(inode, n_pasted_size);
2113 DQUOT_FREE_SPACE_NODIRTY(inode, n_pasted_size);
2117 /* Insert new item into the buffer at the path. */
2118 int reiserfs_insert_item(struct reiserfs_transaction_handle *th, struct path *p_s_path, /* Path to the inserteded item. */
2119 const struct cpu_key *key, struct item_head *p_s_ih, /* Pointer to the item header to insert. */
2120 struct inode *inode, const char *p_c_body)
2121 { /* Pointer to the bytes to insert. */
2122 struct tree_balance s_ins_balance;
2125 int quota_bytes = 0;
2127 BUG_ON(!th->t_trans_id);
2129 if (inode) { /* Do we count quotas for item? */
2130 fs_gen = get_generation(inode->i_sb);
2131 quota_bytes = ih_item_len(p_s_ih);
2133 /* hack so the quota code doesn't have to guess if the file has
2134 ** a tail, links are always tails, so there's no guessing needed
2136 if (!S_ISLNK(inode->i_mode) && is_direct_le_ih(p_s_ih)) {
2137 quota_bytes = inode->i_sb->s_blocksize + UNFM_P_SIZE;
2139 #ifdef REISERQUOTA_DEBUG
2140 reiserfs_debug(inode->i_sb, REISERFS_DEBUG_CODE,
2141 "reiserquota insert_item(): allocating %u id=%u type=%c",
2142 quota_bytes, inode->i_uid, head2type(p_s_ih));
2144 /* We can't dirty inode here. It would be immediately written but
2145 * appropriate stat item isn't inserted yet... */
2146 if (DQUOT_ALLOC_SPACE_NODIRTY(inode, quota_bytes)) {
2147 pathrelse(p_s_path);
2150 if (DLIMIT_ALLOC_SPACE(inode, quota_bytes)) {
2151 DQUOT_FREE_SPACE_NODIRTY(inode, quota_bytes);
2152 pathrelse(p_s_path);
2156 init_tb_struct(th, &s_ins_balance, th->t_super, p_s_path,
2157 IH_SIZE + ih_item_len(p_s_ih));
2158 #ifdef DISPLACE_NEW_PACKING_LOCALITIES
2159 s_ins_balance.key = key->on_disk_key;
2161 /* DQUOT_* can schedule, must check to be sure calling fix_nodes is safe */
2162 if (inode && fs_changed(fs_gen, inode->i_sb)) {
2167 fix_nodes(M_INSERT, &s_ins_balance, p_s_ih,
2168 p_c_body)) == REPEAT_SEARCH) {
2170 /* file system changed while we were in the fix_nodes */
2171 PROC_INFO_INC(th->t_super, insert_item_restarted);
2172 retval = search_item(th->t_super, key, p_s_path);
2173 if (retval == IO_ERROR) {
2177 if (retval == ITEM_FOUND) {
2178 reiserfs_warning(th->t_super,
2179 "PAP-5760: reiserfs_insert_item: "
2180 "key %K already exists in the tree",
2187 /* make balancing after all resources will be collected at a time */
2188 if (retval == CARRY_ON) {
2189 do_balance(&s_ins_balance, p_s_ih, p_c_body, M_INSERT);
2193 retval = (retval == NO_DISK_SPACE) ? -ENOSPC : -EIO;
2195 /* also releases the path */
2196 unfix_nodes(&s_ins_balance);
2197 #ifdef REISERQUOTA_DEBUG
2198 reiserfs_debug(th->t_super, REISERFS_DEBUG_CODE,
2199 "reiserquota insert_item(): freeing %u id=%u type=%c",
2200 quota_bytes, inode->i_uid, head2type(p_s_ih));
2203 DLIMIT_FREE_SPACE(inode, quota_bytes);
2204 DQUOT_FREE_SPACE_NODIRTY(inode, quota_bytes);