• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  * Copyright (C) 2011 Red Hat, Inc.
3  *
4  * This file is released under the GPL.
5  */
6 
7 #include "dm-btree-internal.h"
8 #include "dm-space-map.h"
9 #include "dm-transaction-manager.h"
10 
11 #include <linux/export.h>
12 #include <linux/device-mapper.h>
13 
14 #define DM_MSG_PREFIX "btree"
15 
16 /*----------------------------------------------------------------
17  * Array manipulation
18  *--------------------------------------------------------------*/
memcpy_disk(void * dest,const void * src,size_t len)19 static void memcpy_disk(void *dest, const void *src, size_t len)
20 	__dm_written_to_disk(src)
21 {
22 	memcpy(dest, src, len);
23 	__dm_unbless_for_disk(src);
24 }
25 
array_insert(void * base,size_t elt_size,unsigned nr_elts,unsigned index,void * elt)26 static void array_insert(void *base, size_t elt_size, unsigned nr_elts,
27 			 unsigned index, void *elt)
28 	__dm_written_to_disk(elt)
29 {
30 	if (index < nr_elts)
31 		memmove(base + (elt_size * (index + 1)),
32 			base + (elt_size * index),
33 			(nr_elts - index) * elt_size);
34 
35 	memcpy_disk(base + (elt_size * index), elt, elt_size);
36 }
37 
38 /*----------------------------------------------------------------*/
39 
40 /* makes the assumption that no two keys are the same. */
bsearch(struct btree_node * n,uint64_t key,int want_hi)41 static int bsearch(struct btree_node *n, uint64_t key, int want_hi)
42 {
43 	int lo = -1, hi = le32_to_cpu(n->header.nr_entries);
44 
45 	while (hi - lo > 1) {
46 		int mid = lo + ((hi - lo) / 2);
47 		uint64_t mid_key = le64_to_cpu(n->keys[mid]);
48 
49 		if (mid_key == key)
50 			return mid;
51 
52 		if (mid_key < key)
53 			lo = mid;
54 		else
55 			hi = mid;
56 	}
57 
58 	return want_hi ? hi : lo;
59 }
60 
lower_bound(struct btree_node * n,uint64_t key)61 int lower_bound(struct btree_node *n, uint64_t key)
62 {
63 	return bsearch(n, key, 0);
64 }
65 
inc_children(struct dm_transaction_manager * tm,struct btree_node * n,struct dm_btree_value_type * vt)66 void inc_children(struct dm_transaction_manager *tm, struct btree_node *n,
67 		  struct dm_btree_value_type *vt)
68 {
69 	unsigned i;
70 	uint32_t nr_entries = le32_to_cpu(n->header.nr_entries);
71 
72 	if (le32_to_cpu(n->header.flags) & INTERNAL_NODE)
73 		for (i = 0; i < nr_entries; i++)
74 			dm_tm_inc(tm, value64(n, i));
75 	else if (vt->inc)
76 		for (i = 0; i < nr_entries; i++)
77 			vt->inc(vt->context, value_ptr(n, i));
78 }
79 
insert_at(size_t value_size,struct btree_node * node,unsigned index,uint64_t key,void * value)80 static int insert_at(size_t value_size, struct btree_node *node, unsigned index,
81 		      uint64_t key, void *value)
82 		      __dm_written_to_disk(value)
83 {
84 	uint32_t nr_entries = le32_to_cpu(node->header.nr_entries);
85 	__le64 key_le = cpu_to_le64(key);
86 
87 	if (index > nr_entries ||
88 	    index >= le32_to_cpu(node->header.max_entries)) {
89 		DMERR("too many entries in btree node for insert");
90 		__dm_unbless_for_disk(value);
91 		return -ENOMEM;
92 	}
93 
94 	__dm_bless_for_disk(&key_le);
95 
96 	array_insert(node->keys, sizeof(*node->keys), nr_entries, index, &key_le);
97 	array_insert(value_base(node), value_size, nr_entries, index, value);
98 	node->header.nr_entries = cpu_to_le32(nr_entries + 1);
99 
100 	return 0;
101 }
102 
103 /*----------------------------------------------------------------*/
104 
105 /*
106  * We want 3n entries (for some n).  This works more nicely for repeated
107  * insert remove loops than (2n + 1).
108  */
calc_max_entries(size_t value_size,size_t block_size)109 static uint32_t calc_max_entries(size_t value_size, size_t block_size)
110 {
111 	uint32_t total, n;
112 	size_t elt_size = sizeof(uint64_t) + value_size; /* key + value */
113 
114 	block_size -= sizeof(struct node_header);
115 	total = block_size / elt_size;
116 	n = total / 3;		/* rounds down */
117 
118 	return 3 * n;
119 }
120 
dm_btree_empty(struct dm_btree_info * info,dm_block_t * root)121 int dm_btree_empty(struct dm_btree_info *info, dm_block_t *root)
122 {
123 	int r;
124 	struct dm_block *b;
125 	struct btree_node *n;
126 	size_t block_size;
127 	uint32_t max_entries;
128 
129 	r = new_block(info, &b);
130 	if (r < 0)
131 		return r;
132 
133 	block_size = dm_bm_block_size(dm_tm_get_bm(info->tm));
134 	max_entries = calc_max_entries(info->value_type.size, block_size);
135 
136 	n = dm_block_data(b);
137 	memset(n, 0, block_size);
138 	n->header.flags = cpu_to_le32(LEAF_NODE);
139 	n->header.nr_entries = cpu_to_le32(0);
140 	n->header.max_entries = cpu_to_le32(max_entries);
141 	n->header.value_size = cpu_to_le32(info->value_type.size);
142 
143 	*root = dm_block_location(b);
144 	return unlock_block(info, b);
145 }
146 EXPORT_SYMBOL_GPL(dm_btree_empty);
147 
148 /*----------------------------------------------------------------*/
149 
150 /*
151  * Deletion uses a recursive algorithm, since we have limited stack space
152  * we explicitly manage our own stack on the heap.
153  */
154 #define MAX_SPINE_DEPTH 64
155 struct frame {
156 	struct dm_block *b;
157 	struct btree_node *n;
158 	unsigned level;
159 	unsigned nr_children;
160 	unsigned current_child;
161 };
162 
163 struct del_stack {
164 	struct dm_transaction_manager *tm;
165 	int top;
166 	struct frame spine[MAX_SPINE_DEPTH];
167 };
168 
top_frame(struct del_stack * s,struct frame ** f)169 static int top_frame(struct del_stack *s, struct frame **f)
170 {
171 	if (s->top < 0) {
172 		DMERR("btree deletion stack empty");
173 		return -EINVAL;
174 	}
175 
176 	*f = s->spine + s->top;
177 
178 	return 0;
179 }
180 
unprocessed_frames(struct del_stack * s)181 static int unprocessed_frames(struct del_stack *s)
182 {
183 	return s->top >= 0;
184 }
185 
push_frame(struct del_stack * s,dm_block_t b,unsigned level)186 static int push_frame(struct del_stack *s, dm_block_t b, unsigned level)
187 {
188 	int r;
189 	uint32_t ref_count;
190 
191 	if (s->top >= MAX_SPINE_DEPTH - 1) {
192 		DMERR("btree deletion stack out of memory");
193 		return -ENOMEM;
194 	}
195 
196 	r = dm_tm_ref(s->tm, b, &ref_count);
197 	if (r)
198 		return r;
199 
200 	if (ref_count > 1)
201 		/*
202 		 * This is a shared node, so we can just decrement it's
203 		 * reference counter and leave the children.
204 		 */
205 		dm_tm_dec(s->tm, b);
206 
207 	else {
208 		struct frame *f = s->spine + ++s->top;
209 
210 		r = dm_tm_read_lock(s->tm, b, &btree_node_validator, &f->b);
211 		if (r) {
212 			s->top--;
213 			return r;
214 		}
215 
216 		f->n = dm_block_data(f->b);
217 		f->level = level;
218 		f->nr_children = le32_to_cpu(f->n->header.nr_entries);
219 		f->current_child = 0;
220 	}
221 
222 	return 0;
223 }
224 
pop_frame(struct del_stack * s)225 static void pop_frame(struct del_stack *s)
226 {
227 	struct frame *f = s->spine + s->top--;
228 
229 	dm_tm_dec(s->tm, dm_block_location(f->b));
230 	dm_tm_unlock(s->tm, f->b);
231 }
232 
is_internal_level(struct dm_btree_info * info,struct frame * f)233 static bool is_internal_level(struct dm_btree_info *info, struct frame *f)
234 {
235 	return f->level < (info->levels - 1);
236 }
237 
dm_btree_del(struct dm_btree_info * info,dm_block_t root)238 int dm_btree_del(struct dm_btree_info *info, dm_block_t root)
239 {
240 	int r;
241 	struct del_stack *s;
242 
243 	s = kmalloc(sizeof(*s), GFP_KERNEL);
244 	if (!s)
245 		return -ENOMEM;
246 	s->tm = info->tm;
247 	s->top = -1;
248 
249 	r = push_frame(s, root, 0);
250 	if (r)
251 		goto out;
252 
253 	while (unprocessed_frames(s)) {
254 		uint32_t flags;
255 		struct frame *f;
256 		dm_block_t b;
257 
258 		r = top_frame(s, &f);
259 		if (r)
260 			goto out;
261 
262 		if (f->current_child >= f->nr_children) {
263 			pop_frame(s);
264 			continue;
265 		}
266 
267 		flags = le32_to_cpu(f->n->header.flags);
268 		if (flags & INTERNAL_NODE) {
269 			b = value64(f->n, f->current_child);
270 			f->current_child++;
271 			r = push_frame(s, b, f->level);
272 			if (r)
273 				goto out;
274 
275 		} else if (is_internal_level(info, f)) {
276 			b = value64(f->n, f->current_child);
277 			f->current_child++;
278 			r = push_frame(s, b, f->level + 1);
279 			if (r)
280 				goto out;
281 
282 		} else {
283 			if (info->value_type.dec) {
284 				unsigned i;
285 
286 				for (i = 0; i < f->nr_children; i++)
287 					info->value_type.dec(info->value_type.context,
288 							     value_ptr(f->n, i));
289 			}
290 			f->current_child = f->nr_children;
291 		}
292 	}
293 
294 out:
295 	kfree(s);
296 	return r;
297 }
298 EXPORT_SYMBOL_GPL(dm_btree_del);
299 
300 /*----------------------------------------------------------------*/
301 
btree_lookup_raw(struct ro_spine * s,dm_block_t block,uint64_t key,int (* search_fn)(struct btree_node *,uint64_t),uint64_t * result_key,void * v,size_t value_size)302 static int btree_lookup_raw(struct ro_spine *s, dm_block_t block, uint64_t key,
303 			    int (*search_fn)(struct btree_node *, uint64_t),
304 			    uint64_t *result_key, void *v, size_t value_size)
305 {
306 	int i, r;
307 	uint32_t flags, nr_entries;
308 
309 	do {
310 		r = ro_step(s, block);
311 		if (r < 0)
312 			return r;
313 
314 		i = search_fn(ro_node(s), key);
315 
316 		flags = le32_to_cpu(ro_node(s)->header.flags);
317 		nr_entries = le32_to_cpu(ro_node(s)->header.nr_entries);
318 		if (i < 0 || i >= nr_entries)
319 			return -ENODATA;
320 
321 		if (flags & INTERNAL_NODE)
322 			block = value64(ro_node(s), i);
323 
324 	} while (!(flags & LEAF_NODE));
325 
326 	*result_key = le64_to_cpu(ro_node(s)->keys[i]);
327 	memcpy(v, value_ptr(ro_node(s), i), value_size);
328 
329 	return 0;
330 }
331 
dm_btree_lookup(struct dm_btree_info * info,dm_block_t root,uint64_t * keys,void * value_le)332 int dm_btree_lookup(struct dm_btree_info *info, dm_block_t root,
333 		    uint64_t *keys, void *value_le)
334 {
335 	unsigned level, last_level = info->levels - 1;
336 	int r = -ENODATA;
337 	uint64_t rkey;
338 	__le64 internal_value_le;
339 	struct ro_spine spine;
340 
341 	init_ro_spine(&spine, info);
342 	for (level = 0; level < info->levels; level++) {
343 		size_t size;
344 		void *value_p;
345 
346 		if (level == last_level) {
347 			value_p = value_le;
348 			size = info->value_type.size;
349 
350 		} else {
351 			value_p = &internal_value_le;
352 			size = sizeof(uint64_t);
353 		}
354 
355 		r = btree_lookup_raw(&spine, root, keys[level],
356 				     lower_bound, &rkey,
357 				     value_p, size);
358 
359 		if (!r) {
360 			if (rkey != keys[level]) {
361 				exit_ro_spine(&spine);
362 				return -ENODATA;
363 			}
364 		} else {
365 			exit_ro_spine(&spine);
366 			return r;
367 		}
368 
369 		root = le64_to_cpu(internal_value_le);
370 	}
371 	exit_ro_spine(&spine);
372 
373 	return r;
374 }
375 EXPORT_SYMBOL_GPL(dm_btree_lookup);
376 
377 /*
378  * Splits a node by creating a sibling node and shifting half the nodes
379  * contents across.  Assumes there is a parent node, and it has room for
380  * another child.
381  *
382  * Before:
383  *	  +--------+
384  *	  | Parent |
385  *	  +--------+
386  *	     |
387  *	     v
388  *	+----------+
389  *	| A ++++++ |
390  *	+----------+
391  *
392  *
393  * After:
394  *		+--------+
395  *		| Parent |
396  *		+--------+
397  *		  |	|
398  *		  v	+------+
399  *	    +---------+	       |
400  *	    | A* +++  |	       v
401  *	    +---------+	  +-------+
402  *			  | B +++ |
403  *			  +-------+
404  *
405  * Where A* is a shadow of A.
406  */
btree_split_sibling(struct shadow_spine * s,dm_block_t root,unsigned parent_index,uint64_t key)407 static int btree_split_sibling(struct shadow_spine *s, dm_block_t root,
408 			       unsigned parent_index, uint64_t key)
409 {
410 	int r;
411 	size_t size;
412 	unsigned nr_left, nr_right;
413 	struct dm_block *left, *right, *parent;
414 	struct btree_node *ln, *rn, *pn;
415 	__le64 location;
416 
417 	left = shadow_current(s);
418 
419 	r = new_block(s->info, &right);
420 	if (r < 0)
421 		return r;
422 
423 	ln = dm_block_data(left);
424 	rn = dm_block_data(right);
425 
426 	nr_left = le32_to_cpu(ln->header.nr_entries) / 2;
427 	nr_right = le32_to_cpu(ln->header.nr_entries) - nr_left;
428 
429 	ln->header.nr_entries = cpu_to_le32(nr_left);
430 
431 	rn->header.flags = ln->header.flags;
432 	rn->header.nr_entries = cpu_to_le32(nr_right);
433 	rn->header.max_entries = ln->header.max_entries;
434 	rn->header.value_size = ln->header.value_size;
435 	memcpy(rn->keys, ln->keys + nr_left, nr_right * sizeof(rn->keys[0]));
436 
437 	size = le32_to_cpu(ln->header.flags) & INTERNAL_NODE ?
438 		sizeof(uint64_t) : s->info->value_type.size;
439 	memcpy(value_ptr(rn, 0), value_ptr(ln, nr_left),
440 	       size * nr_right);
441 
442 	/*
443 	 * Patch up the parent
444 	 */
445 	parent = shadow_parent(s);
446 
447 	pn = dm_block_data(parent);
448 	location = cpu_to_le64(dm_block_location(left));
449 	__dm_bless_for_disk(&location);
450 	memcpy_disk(value_ptr(pn, parent_index),
451 		    &location, sizeof(__le64));
452 
453 	location = cpu_to_le64(dm_block_location(right));
454 	__dm_bless_for_disk(&location);
455 
456 	r = insert_at(sizeof(__le64), pn, parent_index + 1,
457 		      le64_to_cpu(rn->keys[0]), &location);
458 	if (r)
459 		return r;
460 
461 	if (key < le64_to_cpu(rn->keys[0])) {
462 		unlock_block(s->info, right);
463 		s->nodes[1] = left;
464 	} else {
465 		unlock_block(s->info, left);
466 		s->nodes[1] = right;
467 	}
468 
469 	return 0;
470 }
471 
472 /*
473  * Splits a node by creating two new children beneath the given node.
474  *
475  * Before:
476  *	  +----------+
477  *	  | A ++++++ |
478  *	  +----------+
479  *
480  *
481  * After:
482  *	+------------+
483  *	| A (shadow) |
484  *	+------------+
485  *	    |	|
486  *   +------+	+----+
487  *   |		     |
488  *   v		     v
489  * +-------+	 +-------+
490  * | B +++ |	 | C +++ |
491  * +-------+	 +-------+
492  */
btree_split_beneath(struct shadow_spine * s,uint64_t key)493 static int btree_split_beneath(struct shadow_spine *s, uint64_t key)
494 {
495 	int r;
496 	size_t size;
497 	unsigned nr_left, nr_right;
498 	struct dm_block *left, *right, *new_parent;
499 	struct btree_node *pn, *ln, *rn;
500 	__le64 val;
501 
502 	new_parent = shadow_current(s);
503 
504 	r = new_block(s->info, &left);
505 	if (r < 0)
506 		return r;
507 
508 	r = new_block(s->info, &right);
509 	if (r < 0) {
510 		/* FIXME: put left */
511 		return r;
512 	}
513 
514 	pn = dm_block_data(new_parent);
515 	ln = dm_block_data(left);
516 	rn = dm_block_data(right);
517 
518 	nr_left = le32_to_cpu(pn->header.nr_entries) / 2;
519 	nr_right = le32_to_cpu(pn->header.nr_entries) - nr_left;
520 
521 	ln->header.flags = pn->header.flags;
522 	ln->header.nr_entries = cpu_to_le32(nr_left);
523 	ln->header.max_entries = pn->header.max_entries;
524 	ln->header.value_size = pn->header.value_size;
525 
526 	rn->header.flags = pn->header.flags;
527 	rn->header.nr_entries = cpu_to_le32(nr_right);
528 	rn->header.max_entries = pn->header.max_entries;
529 	rn->header.value_size = pn->header.value_size;
530 
531 	memcpy(ln->keys, pn->keys, nr_left * sizeof(pn->keys[0]));
532 	memcpy(rn->keys, pn->keys + nr_left, nr_right * sizeof(pn->keys[0]));
533 
534 	size = le32_to_cpu(pn->header.flags) & INTERNAL_NODE ?
535 		sizeof(__le64) : s->info->value_type.size;
536 	memcpy(value_ptr(ln, 0), value_ptr(pn, 0), nr_left * size);
537 	memcpy(value_ptr(rn, 0), value_ptr(pn, nr_left),
538 	       nr_right * size);
539 
540 	/* new_parent should just point to l and r now */
541 	pn->header.flags = cpu_to_le32(INTERNAL_NODE);
542 	pn->header.nr_entries = cpu_to_le32(2);
543 	pn->header.max_entries = cpu_to_le32(
544 		calc_max_entries(sizeof(__le64),
545 				 dm_bm_block_size(
546 					 dm_tm_get_bm(s->info->tm))));
547 	pn->header.value_size = cpu_to_le32(sizeof(__le64));
548 
549 	val = cpu_to_le64(dm_block_location(left));
550 	__dm_bless_for_disk(&val);
551 	pn->keys[0] = ln->keys[0];
552 	memcpy_disk(value_ptr(pn, 0), &val, sizeof(__le64));
553 
554 	val = cpu_to_le64(dm_block_location(right));
555 	__dm_bless_for_disk(&val);
556 	pn->keys[1] = rn->keys[0];
557 	memcpy_disk(value_ptr(pn, 1), &val, sizeof(__le64));
558 
559 	/*
560 	 * rejig the spine.  This is ugly, since it knows too
561 	 * much about the spine
562 	 */
563 	if (s->nodes[0] != new_parent) {
564 		unlock_block(s->info, s->nodes[0]);
565 		s->nodes[0] = new_parent;
566 	}
567 	if (key < le64_to_cpu(rn->keys[0])) {
568 		unlock_block(s->info, right);
569 		s->nodes[1] = left;
570 	} else {
571 		unlock_block(s->info, left);
572 		s->nodes[1] = right;
573 	}
574 	s->count = 2;
575 
576 	return 0;
577 }
578 
btree_insert_raw(struct shadow_spine * s,dm_block_t root,struct dm_btree_value_type * vt,uint64_t key,unsigned * index)579 static int btree_insert_raw(struct shadow_spine *s, dm_block_t root,
580 			    struct dm_btree_value_type *vt,
581 			    uint64_t key, unsigned *index)
582 {
583 	int r, i = *index, top = 1;
584 	struct btree_node *node;
585 
586 	for (;;) {
587 		r = shadow_step(s, root, vt);
588 		if (r < 0)
589 			return r;
590 
591 		node = dm_block_data(shadow_current(s));
592 
593 		/*
594 		 * We have to patch up the parent node, ugly, but I don't
595 		 * see a way to do this automatically as part of the spine
596 		 * op.
597 		 */
598 		if (shadow_has_parent(s) && i >= 0) { /* FIXME: second clause unness. */
599 			__le64 location = cpu_to_le64(dm_block_location(shadow_current(s)));
600 
601 			__dm_bless_for_disk(&location);
602 			memcpy_disk(value_ptr(dm_block_data(shadow_parent(s)), i),
603 				    &location, sizeof(__le64));
604 		}
605 
606 		node = dm_block_data(shadow_current(s));
607 
608 		if (node->header.nr_entries == node->header.max_entries) {
609 			if (top)
610 				r = btree_split_beneath(s, key);
611 			else
612 				r = btree_split_sibling(s, root, i, key);
613 
614 			if (r < 0)
615 				return r;
616 		}
617 
618 		node = dm_block_data(shadow_current(s));
619 
620 		i = lower_bound(node, key);
621 
622 		if (le32_to_cpu(node->header.flags) & LEAF_NODE)
623 			break;
624 
625 		if (i < 0) {
626 			/* change the bounds on the lowest key */
627 			node->keys[0] = cpu_to_le64(key);
628 			i = 0;
629 		}
630 
631 		root = value64(node, i);
632 		top = 0;
633 	}
634 
635 	if (i < 0 || le64_to_cpu(node->keys[i]) != key)
636 		i++;
637 
638 	*index = i;
639 	return 0;
640 }
641 
insert(struct dm_btree_info * info,dm_block_t root,uint64_t * keys,void * value,dm_block_t * new_root,int * inserted)642 static int insert(struct dm_btree_info *info, dm_block_t root,
643 		  uint64_t *keys, void *value, dm_block_t *new_root,
644 		  int *inserted)
645 		  __dm_written_to_disk(value)
646 {
647 	int r, need_insert;
648 	unsigned level, index = -1, last_level = info->levels - 1;
649 	dm_block_t block = root;
650 	struct shadow_spine spine;
651 	struct btree_node *n;
652 	struct dm_btree_value_type le64_type;
653 
654 	le64_type.context = NULL;
655 	le64_type.size = sizeof(__le64);
656 	le64_type.inc = NULL;
657 	le64_type.dec = NULL;
658 	le64_type.equal = NULL;
659 
660 	init_shadow_spine(&spine, info);
661 
662 	for (level = 0; level < (info->levels - 1); level++) {
663 		r = btree_insert_raw(&spine, block, &le64_type, keys[level], &index);
664 		if (r < 0)
665 			goto bad;
666 
667 		n = dm_block_data(shadow_current(&spine));
668 		need_insert = ((index >= le32_to_cpu(n->header.nr_entries)) ||
669 			       (le64_to_cpu(n->keys[index]) != keys[level]));
670 
671 		if (need_insert) {
672 			dm_block_t new_tree;
673 			__le64 new_le;
674 
675 			r = dm_btree_empty(info, &new_tree);
676 			if (r < 0)
677 				goto bad;
678 
679 			new_le = cpu_to_le64(new_tree);
680 			__dm_bless_for_disk(&new_le);
681 
682 			r = insert_at(sizeof(uint64_t), n, index,
683 				      keys[level], &new_le);
684 			if (r)
685 				goto bad;
686 		}
687 
688 		if (level < last_level)
689 			block = value64(n, index);
690 	}
691 
692 	r = btree_insert_raw(&spine, block, &info->value_type,
693 			     keys[level], &index);
694 	if (r < 0)
695 		goto bad;
696 
697 	n = dm_block_data(shadow_current(&spine));
698 	need_insert = ((index >= le32_to_cpu(n->header.nr_entries)) ||
699 		       (le64_to_cpu(n->keys[index]) != keys[level]));
700 
701 	if (need_insert) {
702 		if (inserted)
703 			*inserted = 1;
704 
705 		r = insert_at(info->value_type.size, n, index,
706 			      keys[level], value);
707 		if (r)
708 			goto bad_unblessed;
709 	} else {
710 		if (inserted)
711 			*inserted = 0;
712 
713 		if (info->value_type.dec &&
714 		    (!info->value_type.equal ||
715 		     !info->value_type.equal(
716 			     info->value_type.context,
717 			     value_ptr(n, index),
718 			     value))) {
719 			info->value_type.dec(info->value_type.context,
720 					     value_ptr(n, index));
721 		}
722 		memcpy_disk(value_ptr(n, index),
723 			    value, info->value_type.size);
724 	}
725 
726 	*new_root = shadow_root(&spine);
727 	exit_shadow_spine(&spine);
728 
729 	return 0;
730 
731 bad:
732 	__dm_unbless_for_disk(value);
733 bad_unblessed:
734 	exit_shadow_spine(&spine);
735 	return r;
736 }
737 
dm_btree_insert(struct dm_btree_info * info,dm_block_t root,uint64_t * keys,void * value,dm_block_t * new_root)738 int dm_btree_insert(struct dm_btree_info *info, dm_block_t root,
739 		    uint64_t *keys, void *value, dm_block_t *new_root)
740 		    __dm_written_to_disk(value)
741 {
742 	return insert(info, root, keys, value, new_root, NULL);
743 }
744 EXPORT_SYMBOL_GPL(dm_btree_insert);
745 
dm_btree_insert_notify(struct dm_btree_info * info,dm_block_t root,uint64_t * keys,void * value,dm_block_t * new_root,int * inserted)746 int dm_btree_insert_notify(struct dm_btree_info *info, dm_block_t root,
747 			   uint64_t *keys, void *value, dm_block_t *new_root,
748 			   int *inserted)
749 			   __dm_written_to_disk(value)
750 {
751 	return insert(info, root, keys, value, new_root, inserted);
752 }
753 EXPORT_SYMBOL_GPL(dm_btree_insert_notify);
754 
755 /*----------------------------------------------------------------*/
756 
find_highest_key(struct ro_spine * s,dm_block_t block,uint64_t * result_key,dm_block_t * next_block)757 static int find_highest_key(struct ro_spine *s, dm_block_t block,
758 			    uint64_t *result_key, dm_block_t *next_block)
759 {
760 	int i, r;
761 	uint32_t flags;
762 
763 	do {
764 		r = ro_step(s, block);
765 		if (r < 0)
766 			return r;
767 
768 		flags = le32_to_cpu(ro_node(s)->header.flags);
769 		i = le32_to_cpu(ro_node(s)->header.nr_entries);
770 		if (!i)
771 			return -ENODATA;
772 		else
773 			i--;
774 
775 		*result_key = le64_to_cpu(ro_node(s)->keys[i]);
776 		if (next_block || flags & INTERNAL_NODE)
777 			block = value64(ro_node(s), i);
778 
779 	} while (flags & INTERNAL_NODE);
780 
781 	if (next_block)
782 		*next_block = block;
783 	return 0;
784 }
785 
dm_btree_find_highest_key(struct dm_btree_info * info,dm_block_t root,uint64_t * result_keys)786 int dm_btree_find_highest_key(struct dm_btree_info *info, dm_block_t root,
787 			      uint64_t *result_keys)
788 {
789 	int r = 0, count = 0, level;
790 	struct ro_spine spine;
791 
792 	init_ro_spine(&spine, info);
793 	for (level = 0; level < info->levels; level++) {
794 		r = find_highest_key(&spine, root, result_keys + level,
795 				     level == info->levels - 1 ? NULL : &root);
796 		if (r == -ENODATA) {
797 			r = 0;
798 			break;
799 
800 		} else if (r)
801 			break;
802 
803 		count++;
804 	}
805 	exit_ro_spine(&spine);
806 
807 	return r ? r : count;
808 }
809 EXPORT_SYMBOL_GPL(dm_btree_find_highest_key);
810 
811 /*
812  * FIXME: We shouldn't use a recursive algorithm when we have limited stack
813  * space.  Also this only works for single level trees.
814  */
walk_node(struct ro_spine * s,dm_block_t block,int (* fn)(void * context,uint64_t * keys,void * leaf),void * context)815 static int walk_node(struct ro_spine *s, dm_block_t block,
816 		     int (*fn)(void *context, uint64_t *keys, void *leaf),
817 		     void *context)
818 {
819 	int r;
820 	unsigned i, nr;
821 	struct btree_node *n;
822 	uint64_t keys;
823 
824 	r = ro_step(s, block);
825 	n = ro_node(s);
826 
827 	nr = le32_to_cpu(n->header.nr_entries);
828 	for (i = 0; i < nr; i++) {
829 		if (le32_to_cpu(n->header.flags) & INTERNAL_NODE) {
830 			r = walk_node(s, value64(n, i), fn, context);
831 			if (r)
832 				goto out;
833 		} else {
834 			keys = le64_to_cpu(*key_ptr(n, i));
835 			r = fn(context, &keys, value_ptr(n, i));
836 			if (r)
837 				goto out;
838 		}
839 	}
840 
841 out:
842 	ro_pop(s);
843 	return r;
844 }
845 
dm_btree_walk(struct dm_btree_info * info,dm_block_t root,int (* fn)(void * context,uint64_t * keys,void * leaf),void * context)846 int dm_btree_walk(struct dm_btree_info *info, dm_block_t root,
847 		  int (*fn)(void *context, uint64_t *keys, void *leaf),
848 		  void *context)
849 {
850 	int r;
851 	struct ro_spine spine;
852 
853 	BUG_ON(info->levels > 1);
854 
855 	init_ro_spine(&spine, info);
856 	r = walk_node(&spine, root, fn, context);
857 	exit_ro_spine(&spine);
858 
859 	return r;
860 }
861 EXPORT_SYMBOL_GPL(dm_btree_walk);
862