2 * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README
4 /* Reiserfs block (de)allocator, bitmap-based. */
6 #include <linux/time.h>
8 #include <linux/errno.h>
9 #include <linux/buffer_head.h>
10 #include <linux/kernel.h>
11 #include <linux/pagemap.h>
12 #include <linux/vmalloc.h>
13 #include <linux/quotaops.h>
14 #include <linux/seq_file.h>
16 #define PREALLOCATION_SIZE 9
18 /* different reiserfs block allocator options */
20 #define SB_ALLOC_OPTS(s) (REISERFS_SB(s)->s_alloc_options.bits)
22 #define _ALLOC_concentrating_formatted_nodes 0
23 #define _ALLOC_displacing_large_files 1
24 #define _ALLOC_displacing_new_packing_localities 2
25 #define _ALLOC_old_hashed_relocation 3
26 #define _ALLOC_new_hashed_relocation 4
27 #define _ALLOC_skip_busy 5
28 #define _ALLOC_displace_based_on_dirid 6
29 #define _ALLOC_hashed_formatted_nodes 7
30 #define _ALLOC_old_way 8
31 #define _ALLOC_hundredth_slices 9
32 #define _ALLOC_dirid_groups 10
33 #define _ALLOC_oid_groups 11
34 #define _ALLOC_packing_groups 12
36 #define concentrating_formatted_nodes(s) test_bit(_ALLOC_concentrating_formatted_nodes, &SB_ALLOC_OPTS(s))
37 #define displacing_large_files(s) test_bit(_ALLOC_displacing_large_files, &SB_ALLOC_OPTS(s))
38 #define displacing_new_packing_localities(s) test_bit(_ALLOC_displacing_new_packing_localities, &SB_ALLOC_OPTS(s))
40 #define SET_OPTION(optname) \
42 reiserfs_info(s, "block allocator option \"%s\" is set", #optname); \
43 set_bit(_ALLOC_ ## optname , &SB_ALLOC_OPTS(s)); \
45 #define TEST_OPTION(optname, s) \
46 test_bit(_ALLOC_ ## optname , &SB_ALLOC_OPTS(s))
48 static inline void get_bit_address(struct super_block *s,
50 unsigned int *bmap_nr,
54 * It is in the bitmap block number equal to the block
55 * number divided by the number of bits in a block.
57 *bmap_nr = block >> (s->s_blocksize_bits + 3);
58 /* Within that bitmap block it is located at bit offset *offset. */
59 *offset = block & ((s->s_blocksize << 3) - 1);
62 int is_reusable(struct super_block *s, b_blocknr_t block, int bit_value)
64 unsigned int bmap, offset;
65 unsigned int bmap_count = reiserfs_bmap_count(s);
67 if (block == 0 || block >= SB_BLOCK_COUNT(s)) {
68 reiserfs_error(s, "vs-4010",
69 "block number is out of range %lu (%u)",
70 block, SB_BLOCK_COUNT(s));
74 get_bit_address(s, block, &bmap, &offset);
77 * Old format filesystem? Unlikely, but the bitmaps are all
78 * up front so we need to account for it.
80 if (unlikely(test_bit(REISERFS_OLD_FORMAT,
81 &(REISERFS_SB(s)->s_properties)))) {
82 b_blocknr_t bmap1 = REISERFS_SB(s)->s_sbh->b_blocknr + 1;
84 block <= bmap1 + bmap_count) {
85 reiserfs_error(s, "vs-4019", "bitmap block %lu(%u) "
86 "can't be freed or reused",
92 reiserfs_error(s, "vs-4020", "bitmap block %lu(%u) "
93 "can't be freed or reused",
99 if (bmap >= bmap_count) {
100 reiserfs_error(s, "vs-4030", "bitmap for requested block "
101 "is out of range: block=%lu, bitmap_nr=%u",
106 if (bit_value == 0 && block == SB_ROOT_BLOCK(s)) {
107 reiserfs_error(s, "vs-4050", "this is root block (%u), "
108 "it must be busy", SB_ROOT_BLOCK(s));
116 * Searches in journal structures for a given block number (bmap, off).
117 * If block is found in reiserfs journal it suggests next free block
120 static inline int is_block_in_journal(struct super_block *s, unsigned int bmap,
125 if (reiserfs_in_journal(s, bmap, off, 1, &tmp)) {
126 if (tmp) { /* hint supplied */
128 PROC_INFO_INC(s, scan_bitmap.in_journal_hint);
130 (*next) = off + 1; /* inc offset to avoid looping. */
131 PROC_INFO_INC(s, scan_bitmap.in_journal_nohint);
133 PROC_INFO_INC(s, scan_bitmap.retry);
140 * Searches for a window of zero bits with given minimum and maximum
141 * lengths in one bitmap block
143 static int scan_bitmap_block(struct reiserfs_transaction_handle *th,
144 unsigned int bmap_n, int *beg, int boundary,
145 int min, int max, int unfm)
147 struct super_block *s = th->t_super;
148 struct reiserfs_bitmap_info *bi = &SB_AP_BITMAP(s)[bmap_n];
149 struct buffer_head *bh;
153 BUG_ON(!th->t_trans_id);
155 RFALSE(bmap_n >= reiserfs_bmap_count(s), "Bitmap %u is out of "
156 "range (0..%u)", bmap_n, reiserfs_bmap_count(s) - 1);
157 PROC_INFO_INC(s, scan_bitmap.bmap);
160 reiserfs_error(s, "jdm-4055", "NULL bitmap info pointer "
161 "for bitmap %d", bmap_n);
165 bh = reiserfs_read_bitmap_block(s, bmap_n);
171 if (bi->free_count < min) {
173 return 0; /* No free blocks in this bitmap */
176 /* search for a first zero bit -- beginning of a window */
177 *beg = reiserfs_find_next_zero_le_bit
178 ((unsigned long *)(bh->b_data), boundary, *beg);
181 * search for a zero bit fails or the rest of bitmap block
182 * cannot contain a zero window of minimum size
184 if (*beg + min > boundary) {
189 if (unfm && is_block_in_journal(s, bmap_n, *beg, beg))
191 /* first zero bit found; we check next bits */
192 for (end = *beg + 1;; end++) {
193 if (end >= *beg + max || end >= boundary
194 || reiserfs_test_le_bit(end, bh->b_data)) {
200 * finding the other end of zero bit window requires
201 * looking into journal structures (in case of
202 * searching for free blocks for unformatted nodes)
204 if (unfm && is_block_in_journal(s, bmap_n, end, &next))
209 * now (*beg) points to beginning of zero bits window,
210 * (end) points to one bit after the window end
213 /* found window of proper size */
214 if (end - *beg >= min) {
216 reiserfs_prepare_for_journal(s, bh, 1);
218 * try to set all blocks used checking are
221 for (i = *beg; i < end; i++) {
222 /* Don't check in journal again. */
223 if (reiserfs_test_and_set_le_bit
226 * bit was set by another process while
227 * we slept in prepare_for_journal()
229 PROC_INFO_INC(s, scan_bitmap.stolen);
232 * we can continue with smaller set
233 * of allocated blocks, if length of
234 * this set is more or equal to `min'
236 if (i >= *beg + min) {
242 * otherwise we clear all bit
246 reiserfs_clear_le_bit
248 reiserfs_restore_prepared_buffer(s, bh);
252 * Search again in current block
258 bi->free_count -= (end - *beg);
259 journal_mark_dirty(th, bh);
262 /* free block count calculation */
263 reiserfs_prepare_for_journal(s, SB_BUFFER_WITH_SB(s),
265 PUT_SB_FREE_BLOCKS(s, SB_FREE_BLOCKS(s) - (end - *beg));
266 journal_mark_dirty(th, SB_BUFFER_WITH_SB(s));
275 static int bmap_hash_id(struct super_block *s, u32 id)
277 char *hash_in = NULL;
284 hash_in = (char *)(&id);
285 hash = keyed_hash(hash_in, 4);
286 bm = hash % reiserfs_bmap_count(s);
290 /* this can only be true when SB_BMAP_NR = 1 */
291 if (bm >= reiserfs_bmap_count(s))
297 * hashes the id and then returns > 0 if the block group for the
298 * corresponding hash is full
300 static inline int block_group_used(struct super_block *s, u32 id)
302 int bm = bmap_hash_id(s, id);
303 struct reiserfs_bitmap_info *info = &SB_AP_BITMAP(s)[bm];
306 * If we don't have cached information on this bitmap block, we're
307 * going to have to load it later anyway. Loading it here allows us
308 * to make a better decision. This favors long-term performance gain
309 * with a better on-disk layout vs. a short term gain of skipping the
310 * read and potentially having a bad placement.
312 if (info->free_count == UINT_MAX) {
313 struct buffer_head *bh = reiserfs_read_bitmap_block(s, bm);
317 if (info->free_count > ((s->s_blocksize << 3) * 60 / 100)) {
324 * the packing is returned in disk byte order
326 __le32 reiserfs_choose_packing(struct inode * dir)
329 if (TEST_OPTION(packing_groups, dir->i_sb)) {
330 u32 parent_dir = le32_to_cpu(INODE_PKEY(dir)->k_dir_id);
332 * some versions of reiserfsck expect packing locality 1 to be
335 if (parent_dir == 1 || block_group_used(dir->i_sb, parent_dir))
336 packing = INODE_PKEY(dir)->k_objectid;
338 packing = INODE_PKEY(dir)->k_dir_id;
340 packing = INODE_PKEY(dir)->k_objectid;
345 * Tries to find contiguous zero bit window (given size) in given region of
346 * bitmap and place new blocks there. Returns number of allocated blocks.
348 static int scan_bitmap(struct reiserfs_transaction_handle *th,
349 b_blocknr_t * start, b_blocknr_t finish,
350 int min, int max, int unfm, sector_t file_block)
352 int nr_allocated = 0;
353 struct super_block *s = th->t_super;
354 unsigned int bm, off;
355 unsigned int end_bm, end_off;
356 unsigned int off_max = s->s_blocksize << 3;
358 BUG_ON(!th->t_trans_id);
360 PROC_INFO_INC(s, scan_bitmap.call);
362 /* No point in looking for more free blocks */
363 if (SB_FREE_BLOCKS(s) <= 0)
366 get_bit_address(s, *start, &bm, &off);
367 get_bit_address(s, finish, &end_bm, &end_off);
368 if (bm > reiserfs_bmap_count(s))
370 if (end_bm > reiserfs_bmap_count(s))
371 end_bm = reiserfs_bmap_count(s);
374 * When the bitmap is more than 10% free, anyone can allocate.
375 * When it's less than 10% free, only files that already use the
376 * bitmap are allowed. Once we pass 80% full, this restriction
379 * We do this so that files that grow later still have space close to
380 * their original allocation. This improves locality, and presumably
381 * performance as a result.
383 * This is only an allocation policy and does not make up for getting a
384 * bad hint. Decent hinting must be implemented for this to work well.
386 if (TEST_OPTION(skip_busy, s)
387 && SB_FREE_BLOCKS(s) > SB_BLOCK_COUNT(s) / 20) {
388 for (; bm < end_bm; bm++, off = 0) {
389 if ((off && (!unfm || (file_block != 0)))
390 || SB_AP_BITMAP(s)[bm].free_count >
391 (s->s_blocksize << 3) / 10)
393 scan_bitmap_block(th, bm, &off, off_max,
398 /* we know from above that start is a reasonable number */
399 get_bit_address(s, *start, &bm, &off);
402 for (; bm < end_bm; bm++, off = 0) {
404 scan_bitmap_block(th, bm, &off, off_max, min, max, unfm);
410 scan_bitmap_block(th, bm, &off, end_off + 1, min, max, unfm);
413 *start = bm * off_max + off;
418 static void _reiserfs_free_block(struct reiserfs_transaction_handle *th,
419 struct inode *inode, b_blocknr_t block,
422 struct super_block *s = th->t_super;
423 struct reiserfs_super_block *rs;
424 struct buffer_head *sbh, *bmbh;
425 struct reiserfs_bitmap_info *apbi;
426 unsigned int nr, offset;
428 BUG_ON(!th->t_trans_id);
430 PROC_INFO_INC(s, free_block);
432 rs = SB_DISK_SUPER_BLOCK(s);
433 sbh = SB_BUFFER_WITH_SB(s);
434 apbi = SB_AP_BITMAP(s);
436 get_bit_address(s, block, &nr, &offset);
438 if (nr >= reiserfs_bmap_count(s)) {
439 reiserfs_error(s, "vs-4075", "block %lu is out of range",
444 bmbh = reiserfs_read_bitmap_block(s, nr);
448 reiserfs_prepare_for_journal(s, bmbh, 1);
450 /* clear bit for the given block in bit map */
451 if (!reiserfs_test_and_clear_le_bit(offset, bmbh->b_data)) {
452 reiserfs_error(s, "vs-4080",
453 "block %lu: bit already cleared", block);
455 apbi[nr].free_count++;
456 journal_mark_dirty(th, bmbh);
459 reiserfs_prepare_for_journal(s, sbh, 1);
460 /* update super block */
461 set_sb_free_blocks(rs, sb_free_blocks(rs) + 1);
463 journal_mark_dirty(th, sbh);
464 if (for_unformatted) {
465 int depth = reiserfs_write_unlock_nested(s);
466 dquot_free_block_nodirty(inode, 1);
467 reiserfs_write_lock_nested(s, depth);
471 void reiserfs_free_block(struct reiserfs_transaction_handle *th,
472 struct inode *inode, b_blocknr_t block,
475 struct super_block *s = th->t_super;
476 BUG_ON(!th->t_trans_id);
478 RFALSE(!s, "vs-4061: trying to free block on nonexistent device");
479 if (!is_reusable(s, block, 1))
482 if (block > sb_block_count(REISERFS_SB(s)->s_rs)) {
483 reiserfs_error(th->t_super, "bitmap-4072",
484 "Trying to free block outside file system "
485 "boundaries (%lu > %lu)",
486 block, sb_block_count(REISERFS_SB(s)->s_rs));
489 /* mark it before we clear it, just in case */
490 journal_mark_freed(th, s, block);
491 _reiserfs_free_block(th, inode, block, for_unformatted);
494 /* preallocated blocks don't need to be run through journal_mark_freed */
495 static void reiserfs_free_prealloc_block(struct reiserfs_transaction_handle *th,
496 struct inode *inode, b_blocknr_t block)
498 BUG_ON(!th->t_trans_id);
500 "vs-4060: trying to free block on nonexistent device");
501 if (!is_reusable(th->t_super, block, 1))
503 _reiserfs_free_block(th, inode, block, 1);
506 static void __discard_prealloc(struct reiserfs_transaction_handle *th,
507 struct reiserfs_inode_info *ei)
509 unsigned long save = ei->i_prealloc_block;
511 struct inode *inode = &ei->vfs_inode;
512 BUG_ON(!th->t_trans_id);
513 #ifdef CONFIG_REISERFS_CHECK
514 if (ei->i_prealloc_count < 0)
515 reiserfs_error(th->t_super, "zam-4001",
516 "inode has negative prealloc blocks count.");
518 while (ei->i_prealloc_count > 0) {
519 reiserfs_free_prealloc_block(th, inode, ei->i_prealloc_block);
520 ei->i_prealloc_block++;
521 ei->i_prealloc_count--;
525 reiserfs_update_sd(th, inode);
526 ei->i_prealloc_block = save;
527 list_del_init(&(ei->i_prealloc_list));
530 /* FIXME: It should be inline function */
531 void reiserfs_discard_prealloc(struct reiserfs_transaction_handle *th,
534 struct reiserfs_inode_info *ei = REISERFS_I(inode);
535 BUG_ON(!th->t_trans_id);
536 if (ei->i_prealloc_count)
537 __discard_prealloc(th, ei);
540 void reiserfs_discard_all_prealloc(struct reiserfs_transaction_handle *th)
542 struct list_head *plist = &SB_JOURNAL(th->t_super)->j_prealloc_list;
544 BUG_ON(!th->t_trans_id);
546 while (!list_empty(plist)) {
547 struct reiserfs_inode_info *ei;
548 ei = list_entry(plist->next, struct reiserfs_inode_info,
550 #ifdef CONFIG_REISERFS_CHECK
551 if (!ei->i_prealloc_count) {
552 reiserfs_error(th->t_super, "zam-4001",
553 "inode is in prealloc list but has "
554 "no preallocated blocks.");
557 __discard_prealloc(th, ei);
561 void reiserfs_init_alloc_options(struct super_block *s)
563 set_bit(_ALLOC_skip_busy, &SB_ALLOC_OPTS(s));
564 set_bit(_ALLOC_dirid_groups, &SB_ALLOC_OPTS(s));
565 set_bit(_ALLOC_packing_groups, &SB_ALLOC_OPTS(s));
568 /* block allocator related options are parsed here */
569 int reiserfs_parse_alloc_options(struct super_block *s, char *options)
571 char *this_char, *value;
573 /* clear default settings */
574 REISERFS_SB(s)->s_alloc_options.bits = 0;
576 while ((this_char = strsep(&options, ":")) != NULL) {
577 if ((value = strchr(this_char, '=')) != NULL)
580 if (!strcmp(this_char, "concentrating_formatted_nodes")) {
582 SET_OPTION(concentrating_formatted_nodes);
584 && *value) ? simple_strtoul(value, &value,
586 if (temp <= 0 || temp > 100) {
587 REISERFS_SB(s)->s_alloc_options.border = 10;
589 REISERFS_SB(s)->s_alloc_options.border =
594 if (!strcmp(this_char, "displacing_large_files")) {
595 SET_OPTION(displacing_large_files);
596 REISERFS_SB(s)->s_alloc_options.large_file_size =
598 && *value) ? simple_strtoul(value, &value, 0) : 16;
601 if (!strcmp(this_char, "displacing_new_packing_localities")) {
602 SET_OPTION(displacing_new_packing_localities);
606 if (!strcmp(this_char, "old_hashed_relocation")) {
607 SET_OPTION(old_hashed_relocation);
611 if (!strcmp(this_char, "new_hashed_relocation")) {
612 SET_OPTION(new_hashed_relocation);
616 if (!strcmp(this_char, "dirid_groups")) {
617 SET_OPTION(dirid_groups);
620 if (!strcmp(this_char, "oid_groups")) {
621 SET_OPTION(oid_groups);
624 if (!strcmp(this_char, "packing_groups")) {
625 SET_OPTION(packing_groups);
628 if (!strcmp(this_char, "hashed_formatted_nodes")) {
629 SET_OPTION(hashed_formatted_nodes);
633 if (!strcmp(this_char, "skip_busy")) {
634 SET_OPTION(skip_busy);
638 if (!strcmp(this_char, "hundredth_slices")) {
639 SET_OPTION(hundredth_slices);
643 if (!strcmp(this_char, "old_way")) {
648 if (!strcmp(this_char, "displace_based_on_dirid")) {
649 SET_OPTION(displace_based_on_dirid);
653 if (!strcmp(this_char, "preallocmin")) {
654 REISERFS_SB(s)->s_alloc_options.preallocmin =
656 && *value) ? simple_strtoul(value, &value, 0) : 4;
660 if (!strcmp(this_char, "preallocsize")) {
661 REISERFS_SB(s)->s_alloc_options.preallocsize =
663 && *value) ? simple_strtoul(value, &value,
669 reiserfs_warning(s, "zam-4001", "unknown option - %s",
674 reiserfs_info(s, "allocator options = [%08x]\n", SB_ALLOC_OPTS(s));
678 static void print_sep(struct seq_file *seq, int *first)
686 void show_alloc_options(struct seq_file *seq, struct super_block *s)
690 if (SB_ALLOC_OPTS(s) == ((1 << _ALLOC_skip_busy) |
691 (1 << _ALLOC_dirid_groups) | (1 << _ALLOC_packing_groups)))
694 seq_puts(seq, ",alloc=");
696 if (TEST_OPTION(concentrating_formatted_nodes, s)) {
697 print_sep(seq, &first);
698 if (REISERFS_SB(s)->s_alloc_options.border != 10) {
699 seq_printf(seq, "concentrating_formatted_nodes=%d",
700 100 / REISERFS_SB(s)->s_alloc_options.border);
702 seq_puts(seq, "concentrating_formatted_nodes");
704 if (TEST_OPTION(displacing_large_files, s)) {
705 print_sep(seq, &first);
706 if (REISERFS_SB(s)->s_alloc_options.large_file_size != 16) {
707 seq_printf(seq, "displacing_large_files=%lu",
708 REISERFS_SB(s)->s_alloc_options.large_file_size);
710 seq_puts(seq, "displacing_large_files");
712 if (TEST_OPTION(displacing_new_packing_localities, s)) {
713 print_sep(seq, &first);
714 seq_puts(seq, "displacing_new_packing_localities");
716 if (TEST_OPTION(old_hashed_relocation, s)) {
717 print_sep(seq, &first);
718 seq_puts(seq, "old_hashed_relocation");
720 if (TEST_OPTION(new_hashed_relocation, s)) {
721 print_sep(seq, &first);
722 seq_puts(seq, "new_hashed_relocation");
724 if (TEST_OPTION(dirid_groups, s)) {
725 print_sep(seq, &first);
726 seq_puts(seq, "dirid_groups");
728 if (TEST_OPTION(oid_groups, s)) {
729 print_sep(seq, &first);
730 seq_puts(seq, "oid_groups");
732 if (TEST_OPTION(packing_groups, s)) {
733 print_sep(seq, &first);
734 seq_puts(seq, "packing_groups");
736 if (TEST_OPTION(hashed_formatted_nodes, s)) {
737 print_sep(seq, &first);
738 seq_puts(seq, "hashed_formatted_nodes");
740 if (TEST_OPTION(skip_busy, s)) {
741 print_sep(seq, &first);
742 seq_puts(seq, "skip_busy");
744 if (TEST_OPTION(hundredth_slices, s)) {
745 print_sep(seq, &first);
746 seq_puts(seq, "hundredth_slices");
748 if (TEST_OPTION(old_way, s)) {
749 print_sep(seq, &first);
750 seq_puts(seq, "old_way");
752 if (TEST_OPTION(displace_based_on_dirid, s)) {
753 print_sep(seq, &first);
754 seq_puts(seq, "displace_based_on_dirid");
756 if (REISERFS_SB(s)->s_alloc_options.preallocmin != 0) {
757 print_sep(seq, &first);
758 seq_printf(seq, "preallocmin=%d",
759 REISERFS_SB(s)->s_alloc_options.preallocmin);
761 if (REISERFS_SB(s)->s_alloc_options.preallocsize != 17) {
762 print_sep(seq, &first);
763 seq_printf(seq, "preallocsize=%d",
764 REISERFS_SB(s)->s_alloc_options.preallocsize);
768 static inline void new_hashed_relocation(reiserfs_blocknr_hint_t * hint)
771 if (hint->formatted_node) {
772 hash_in = (char *)&hint->key.k_dir_id;
775 /*hint->search_start = hint->beg;*/
776 hash_in = (char *)&hint->key.k_dir_id;
778 if (TEST_OPTION(displace_based_on_dirid, hint->th->t_super))
779 hash_in = (char *)(&INODE_PKEY(hint->inode)->k_dir_id);
782 (char *)(&INODE_PKEY(hint->inode)->k_objectid);
786 hint->beg + keyed_hash(hash_in, 4) % (hint->end - hint->beg);
790 * Relocation based on dirid, hashing them into a given bitmap block
791 * files. Formatted nodes are unaffected, a separate policy covers them
793 static void dirid_groups(reiserfs_blocknr_hint_t * hint)
798 struct super_block *sb = hint->th->t_super;
800 dirid = le32_to_cpu(INODE_PKEY(hint->inode)->k_dir_id);
801 else if (hint->formatted_node)
802 dirid = hint->key.k_dir_id;
805 bm = bmap_hash_id(sb, dirid);
806 hash = bm * (sb->s_blocksize << 3);
807 /* give a portion of the block group to metadata */
809 hash += sb->s_blocksize / 2;
810 hint->search_start = hash;
815 * Relocation based on oid, hashing them into a given bitmap block
816 * files. Formatted nodes are unaffected, a separate policy covers them
818 static void oid_groups(reiserfs_blocknr_hint_t * hint)
826 dirid = le32_to_cpu(INODE_PKEY(hint->inode)->k_dir_id);
829 * keep the root dir and it's first set of subdirs close to
830 * the start of the disk
833 hash = (hint->inode->i_sb->s_blocksize << 3);
835 oid = le32_to_cpu(INODE_PKEY(hint->inode)->k_objectid);
836 bm = bmap_hash_id(hint->inode->i_sb, oid);
837 hash = bm * (hint->inode->i_sb->s_blocksize << 3);
839 hint->search_start = hash;
844 * returns 1 if it finds an indirect item and gets valid hint info
845 * from it, otherwise 0
847 static int get_left_neighbor(reiserfs_blocknr_hint_t * hint)
849 struct treepath *path;
850 struct buffer_head *bh;
851 struct item_head *ih;
857 * reiserfs code can call this function w/o pointer to path
858 * structure supplied; then we rely on supplied search_start
864 bh = get_last_bh(path);
865 RFALSE(!bh, "green-4002: Illegal path specified to get_left_neighbor");
866 ih = tp_item_head(path);
867 pos_in_item = path->pos_in_item;
868 item = tp_item_body(path);
870 hint->search_start = bh->b_blocknr;
873 * for indirect item: go to left and look for the first non-hole entry
874 * in the indirect item
876 if (!hint->formatted_node && is_indirect_le_ih(ih)) {
877 if (pos_in_item == I_UNFM_NUM(ih))
879 while (pos_in_item >= 0) {
880 int t = get_block_num(item, pos_in_item);
882 hint->search_start = t;
890 /* does result value fit into specified region? */
895 * should be, if formatted node, then try to put on first part of the device
896 * specified as number of percent with mount option device, else try to put
897 * on last of device. This is not to say it is good code to do so,
898 * but the effect should be measured.
900 static inline void set_border_in_hint(struct super_block *s,
901 reiserfs_blocknr_hint_t * hint)
904 SB_BLOCK_COUNT(s) / REISERFS_SB(s)->s_alloc_options.border;
906 if (hint->formatted_node)
907 hint->end = border - 1;
912 static inline void displace_large_file(reiserfs_blocknr_hint_t * hint)
914 if (TEST_OPTION(displace_based_on_dirid, hint->th->t_super))
917 keyed_hash((char *)(&INODE_PKEY(hint->inode)->k_dir_id),
918 4) % (hint->end - hint->beg);
922 keyed_hash((char *)(&INODE_PKEY(hint->inode)->k_objectid),
923 4) % (hint->end - hint->beg);
926 static inline void hash_formatted_node(reiserfs_blocknr_hint_t * hint)
931 hash_in = (char *)&hint->key.k_dir_id;
932 else if (TEST_OPTION(displace_based_on_dirid, hint->th->t_super))
933 hash_in = (char *)(&INODE_PKEY(hint->inode)->k_dir_id);
935 hash_in = (char *)(&INODE_PKEY(hint->inode)->k_objectid);
938 hint->beg + keyed_hash(hash_in, 4) % (hint->end - hint->beg);
942 this_blocknr_allocation_would_make_it_a_large_file(reiserfs_blocknr_hint_t *
945 return hint->block ==
946 REISERFS_SB(hint->th->t_super)->s_alloc_options.large_file_size;
949 #ifdef DISPLACE_NEW_PACKING_LOCALITIES
950 static inline void displace_new_packing_locality(reiserfs_blocknr_hint_t * hint)
952 struct in_core_key *key = &hint->key;
954 hint->th->displace_new_blocks = 0;
956 hint->beg + keyed_hash((char *)(&key->k_objectid),
957 4) % (hint->end - hint->beg);
961 static inline int old_hashed_relocation(reiserfs_blocknr_hint_t * hint)
966 if (hint->formatted_node || hint->inode == NULL) {
970 hash_in = le32_to_cpu((INODE_PKEY(hint->inode))->k_dir_id);
972 hint->beg + (u32) keyed_hash(((char *)(&hash_in)),
973 4) % (hint->end - hint->beg - 1);
974 if (border > hint->search_start)
975 hint->search_start = border;
980 static inline int old_way(reiserfs_blocknr_hint_t * hint)
984 if (hint->formatted_node || hint->inode == NULL) {
990 le32_to_cpu(INODE_PKEY(hint->inode)->k_dir_id) % (hint->end -
992 if (border > hint->search_start)
993 hint->search_start = border;
998 static inline void hundredth_slices(reiserfs_blocknr_hint_t * hint)
1000 struct in_core_key *key = &hint->key;
1001 b_blocknr_t slice_start;
1004 (keyed_hash((char *)(&key->k_dir_id), 4) % 100) * (hint->end / 100);
1005 if (slice_start > hint->search_start
1006 || slice_start + (hint->end / 100) <= hint->search_start) {
1007 hint->search_start = slice_start;
1011 static void determine_search_start(reiserfs_blocknr_hint_t * hint,
1014 struct super_block *s = hint->th->t_super;
1018 hint->end = SB_BLOCK_COUNT(s) - 1;
1020 /* This is former border algorithm. Now with tunable border offset */
1021 if (concentrating_formatted_nodes(s))
1022 set_border_in_hint(s, hint);
1024 #ifdef DISPLACE_NEW_PACKING_LOCALITIES
1026 * whenever we create a new directory, we displace it. At first
1027 * we will hash for location, later we might look for a moderately
1028 * empty place for it
1030 if (displacing_new_packing_localities(s)
1031 && hint->th->displace_new_blocks) {
1032 displace_new_packing_locality(hint);
1035 * we do not continue determine_search_start,
1036 * if new packing locality is being displaced
1043 * all persons should feel encouraged to add more special cases
1044 * here and test them
1047 if (displacing_large_files(s) && !hint->formatted_node
1048 && this_blocknr_allocation_would_make_it_a_large_file(hint)) {
1049 displace_large_file(hint);
1054 * if none of our special cases is relevant, use the left
1055 * neighbor in the tree order of the new node we are allocating for
1057 if (hint->formatted_node && TEST_OPTION(hashed_formatted_nodes, s)) {
1058 hash_formatted_node(hint);
1062 unfm_hint = get_left_neighbor(hint);
1065 * Mimic old block allocator behaviour, that is if VFS allowed for
1066 * preallocation, new blocks are displaced based on directory ID.
1067 * Also, if suggested search_start is less than last preallocated
1068 * block, we start searching from it, assuming that HDD dataflow
1069 * is faster in forward direction
1071 if (TEST_OPTION(old_way, s)) {
1072 if (!hint->formatted_node) {
1073 if (!reiserfs_hashed_relocation(s))
1075 else if (!reiserfs_no_unhashed_relocation(s))
1076 old_hashed_relocation(hint);
1079 && hint->search_start <
1080 REISERFS_I(hint->inode)->i_prealloc_block)
1081 hint->search_start =
1082 REISERFS_I(hint->inode)->i_prealloc_block;
1087 /* This is an approach proposed by Hans */
1088 if (TEST_OPTION(hundredth_slices, s)
1089 && !(displacing_large_files(s) && !hint->formatted_node)) {
1090 hundredth_slices(hint);
1094 /* old_hashed_relocation only works on unformatted */
1095 if (!unfm_hint && !hint->formatted_node &&
1096 TEST_OPTION(old_hashed_relocation, s)) {
1097 old_hashed_relocation(hint);
1100 /* new_hashed_relocation works with both formatted/unformatted nodes */
1101 if ((!unfm_hint || hint->formatted_node) &&
1102 TEST_OPTION(new_hashed_relocation, s)) {
1103 new_hashed_relocation(hint);
1106 /* dirid grouping works only on unformatted nodes */
1107 if (!unfm_hint && !hint->formatted_node && TEST_OPTION(dirid_groups, s)) {
1110 #ifdef DISPLACE_NEW_PACKING_LOCALITIES
1111 if (hint->formatted_node && TEST_OPTION(dirid_groups, s)) {
1116 /* oid grouping works only on unformatted nodes */
1117 if (!unfm_hint && !hint->formatted_node && TEST_OPTION(oid_groups, s)) {
1123 static int determine_prealloc_size(reiserfs_blocknr_hint_t * hint)
1125 /* make minimum size a mount option and benchmark both ways */
1126 /* we preallocate blocks only for regular files, specific size */
1127 /* benchmark preallocating always and see what happens */
1129 hint->prealloc_size = 0;
1131 if (!hint->formatted_node && hint->preallocate) {
1132 if (S_ISREG(hint->inode->i_mode)
1133 && hint->inode->i_size >=
1134 REISERFS_SB(hint->th->t_super)->s_alloc_options.
1135 preallocmin * hint->inode->i_sb->s_blocksize)
1136 hint->prealloc_size =
1137 REISERFS_SB(hint->th->t_super)->s_alloc_options.
1143 static inline int allocate_without_wrapping_disk(reiserfs_blocknr_hint_t * hint,
1144 b_blocknr_t * new_blocknrs,
1146 b_blocknr_t finish, int min,
1150 int rest = amount_needed;
1153 while (rest > 0 && start <= finish) {
1154 nr_allocated = scan_bitmap(hint->th, &start, finish, min,
1155 rest + prealloc_size,
1156 !hint->formatted_node, hint->block);
1158 if (nr_allocated == 0) /* no new blocks allocated, return */
1161 /* fill free_blocknrs array first */
1162 while (rest > 0 && nr_allocated > 0) {
1163 *new_blocknrs++ = start++;
1168 /* do we have something to fill prealloc. array also ? */
1169 if (nr_allocated > 0) {
1171 * it means prealloc_size was greater that 0 and
1172 * we do preallocation
1174 list_add(&REISERFS_I(hint->inode)->i_prealloc_list,
1175 &SB_JOURNAL(hint->th->t_super)->
1177 REISERFS_I(hint->inode)->i_prealloc_block = start;
1178 REISERFS_I(hint->inode)->i_prealloc_count =
1184 return (amount_needed - rest);
1187 static inline int blocknrs_and_prealloc_arrays_from_search_start
1188 (reiserfs_blocknr_hint_t * hint, b_blocknr_t * new_blocknrs,
1189 int amount_needed) {
1190 struct super_block *s = hint->th->t_super;
1191 b_blocknr_t start = hint->search_start;
1192 b_blocknr_t finish = SB_BLOCK_COUNT(s) - 1;
1194 int nr_allocated = 0;
1197 determine_prealloc_size(hint);
1198 if (!hint->formatted_node) {
1200 #ifdef REISERQUOTA_DEBUG
1201 reiserfs_debug(s, REISERFS_DEBUG_CODE,
1202 "reiserquota: allocating %d blocks id=%u",
1203 amount_needed, hint->inode->i_uid);
1205 depth = reiserfs_write_unlock_nested(s);
1207 dquot_alloc_block_nodirty(hint->inode, amount_needed);
1208 if (quota_ret) { /* Quota exceeded? */
1209 reiserfs_write_lock_nested(s, depth);
1210 return QUOTA_EXCEEDED;
1212 if (hint->preallocate && hint->prealloc_size) {
1213 #ifdef REISERQUOTA_DEBUG
1214 reiserfs_debug(s, REISERFS_DEBUG_CODE,
1215 "reiserquota: allocating (prealloc) %d blocks id=%u",
1216 hint->prealloc_size, hint->inode->i_uid);
1218 quota_ret = dquot_prealloc_block_nodirty(hint->inode,
1219 hint->prealloc_size);
1221 hint->preallocate = hint->prealloc_size = 0;
1223 /* for unformatted nodes, force large allocations */
1224 reiserfs_write_lock_nested(s, depth);
1229 case 0: /* Search from hint->search_start to end of disk */
1230 start = hint->search_start;
1231 finish = SB_BLOCK_COUNT(s) - 1;
1233 case 1: /* Search from hint->beg to hint->search_start */
1235 finish = hint->search_start;
1237 case 2: /* Last chance: Search from 0 to hint->beg */
1242 /* We've tried searching everywhere, not enough space */
1243 /* Free the blocks */
1244 if (!hint->formatted_node) {
1245 #ifdef REISERQUOTA_DEBUG
1246 reiserfs_debug(s, REISERFS_DEBUG_CODE,
1247 "reiserquota: freeing (nospace) %d blocks id=%u",
1249 hint->prealloc_size -
1251 hint->inode->i_uid);
1253 /* Free not allocated blocks */
1254 depth = reiserfs_write_unlock_nested(s);
1255 dquot_free_block_nodirty(hint->inode,
1256 amount_needed + hint->prealloc_size -
1258 reiserfs_write_lock_nested(s, depth);
1260 while (nr_allocated--)
1261 reiserfs_free_block(hint->th, hint->inode,
1262 new_blocknrs[nr_allocated],
1263 !hint->formatted_node);
1265 return NO_DISK_SPACE;
1267 } while ((nr_allocated += allocate_without_wrapping_disk(hint,
1277 if (!hint->formatted_node &&
1278 amount_needed + hint->prealloc_size >
1279 nr_allocated + REISERFS_I(hint->inode)->i_prealloc_count) {
1280 /* Some of preallocation blocks were not allocated */
1281 #ifdef REISERQUOTA_DEBUG
1282 reiserfs_debug(s, REISERFS_DEBUG_CODE,
1283 "reiserquota: freeing (failed prealloc) %d blocks id=%u",
1284 amount_needed + hint->prealloc_size -
1286 REISERFS_I(hint->inode)->i_prealloc_count,
1287 hint->inode->i_uid);
1290 depth = reiserfs_write_unlock_nested(s);
1291 dquot_free_block_nodirty(hint->inode, amount_needed +
1292 hint->prealloc_size - nr_allocated -
1293 REISERFS_I(hint->inode)->
1295 reiserfs_write_lock_nested(s, depth);
1301 /* grab new blocknrs from preallocated list */
1302 /* return amount still needed after using them */
1303 static int use_preallocated_list_if_available(reiserfs_blocknr_hint_t * hint,
1304 b_blocknr_t * new_blocknrs,
1307 struct inode *inode = hint->inode;
1309 if (REISERFS_I(inode)->i_prealloc_count > 0) {
1310 while (amount_needed) {
1312 *new_blocknrs++ = REISERFS_I(inode)->i_prealloc_block++;
1313 REISERFS_I(inode)->i_prealloc_count--;
1317 if (REISERFS_I(inode)->i_prealloc_count <= 0) {
1318 list_del(&REISERFS_I(inode)->i_prealloc_list);
1323 /* return amount still needed after using preallocated blocks */
1324 return amount_needed;
1327 int reiserfs_allocate_blocknrs(reiserfs_blocknr_hint_t *hint,
1328 b_blocknr_t *new_blocknrs,
1330 /* Amount of blocks we have already reserved */
1333 int initial_amount_needed = amount_needed;
1335 struct super_block *s = hint->th->t_super;
1337 /* Check if there is enough space, taking into account reserved space */
1338 if (SB_FREE_BLOCKS(s) - REISERFS_SB(s)->reserved_blocks <
1339 amount_needed - reserved_by_us)
1340 return NO_DISK_SPACE;
1341 /* should this be if !hint->inode && hint->preallocate? */
1342 /* do you mean hint->formatted_node can be removed ? - Zam */
1344 * hint->formatted_node cannot be removed because we try to access
1345 * inode information here, and there is often no inode associated with
1346 * metadata allocations - green
1349 if (!hint->formatted_node && hint->preallocate) {
1350 amount_needed = use_preallocated_list_if_available
1351 (hint, new_blocknrs, amount_needed);
1354 * We have all the block numbers we need from the
1357 if (amount_needed == 0)
1359 new_blocknrs += (initial_amount_needed - amount_needed);
1362 /* find search start and save it in hint structure */
1363 determine_search_start(hint, amount_needed);
1364 if (hint->search_start >= SB_BLOCK_COUNT(s))
1365 hint->search_start = SB_BLOCK_COUNT(s) - 1;
1367 /* allocation itself; fill new_blocknrs and preallocation arrays */
1368 ret = blocknrs_and_prealloc_arrays_from_search_start
1369 (hint, new_blocknrs, amount_needed);
1372 * We used prealloc. list to fill (partially) new_blocknrs array.
1373 * If final allocation fails we need to return blocks back to
1374 * prealloc. list or just free them. -- Zam (I chose second
1377 if (ret != CARRY_ON) {
1378 while (amount_needed++ < initial_amount_needed) {
1379 reiserfs_free_block(hint->th, hint->inode,
1380 *(--new_blocknrs), 1);
1386 void reiserfs_cache_bitmap_metadata(struct super_block *sb,
1387 struct buffer_head *bh,
1388 struct reiserfs_bitmap_info *info)
1390 unsigned long *cur = (unsigned long *)(bh->b_data + bh->b_size);
1392 /* The first bit must ALWAYS be 1 */
1393 if (!reiserfs_test_le_bit(0, (unsigned long *)bh->b_data))
1394 reiserfs_error(sb, "reiserfs-2025", "bitmap block %lu is "
1395 "corrupted: first bit must be 1", bh->b_blocknr);
1397 info->free_count = 0;
1399 while (--cur >= (unsigned long *)bh->b_data) {
1400 /* 0 and ~0 are special, we can optimize for them */
1402 info->free_count += BITS_PER_LONG;
1403 else if (*cur != ~0L) /* A mix, investigate */
1404 info->free_count += BITS_PER_LONG - hweight_long(*cur);
1408 struct buffer_head *reiserfs_read_bitmap_block(struct super_block *sb,
1409 unsigned int bitmap)
1411 b_blocknr_t block = (sb->s_blocksize << 3) * bitmap;
1412 struct reiserfs_bitmap_info *info = SB_AP_BITMAP(sb) + bitmap;
1413 struct buffer_head *bh;
1416 * Way old format filesystems had the bitmaps packed up front.
1417 * I doubt there are any of these left, but just in case...
1419 if (unlikely(test_bit(REISERFS_OLD_FORMAT,
1420 &(REISERFS_SB(sb)->s_properties))))
1421 block = REISERFS_SB(sb)->s_sbh->b_blocknr + 1 + bitmap;
1422 else if (bitmap == 0)
1423 block = (REISERFS_DISK_OFFSET_IN_BYTES >> sb->s_blocksize_bits) + 1;
1425 bh = sb_bread(sb, block);
1427 reiserfs_warning(sb, "sh-2029: %s: bitmap block (#%u) "
1428 "reading failed", __func__, block);
1430 if (buffer_locked(bh)) {
1432 PROC_INFO_INC(sb, scan_bitmap.wait);
1433 depth = reiserfs_write_unlock_nested(sb);
1434 __wait_on_buffer(bh);
1435 reiserfs_write_lock_nested(sb, depth);
1437 BUG_ON(!buffer_uptodate(bh));
1438 BUG_ON(atomic_read(&bh->b_count) == 0);
1440 if (info->free_count == UINT_MAX)
1441 reiserfs_cache_bitmap_metadata(sb, bh, info);
1447 int reiserfs_init_bitmap_cache(struct super_block *sb)
1449 struct reiserfs_bitmap_info *bitmap;
1450 unsigned int bmap_nr = reiserfs_bmap_count(sb);
1452 bitmap = vmalloc(sizeof(*bitmap) * bmap_nr);
1456 memset(bitmap, 0xff, sizeof(*bitmap) * bmap_nr);
1458 SB_AP_BITMAP(sb) = bitmap;
1463 void reiserfs_free_bitmap_cache(struct super_block *sb)
1465 if (SB_AP_BITMAP(sb)) {
1466 vfree(SB_AP_BITMAP(sb));
1467 SB_AP_BITMAP(sb) = NULL;