1/*
2 * Copyright (C) 2012 Red Hat, Inc.
3 *
4 * This file is released under the GPL.
5 */
6#ifndef _LINUX_DM_ARRAY_H
7#define _LINUX_DM_ARRAY_H
8
9#include "dm-btree.h"
10
11/*----------------------------------------------------------------*/
12
13/*
14 * The dm-array is a persistent version of an array.  It packs the data
15 * more efficiently than a btree which will result in less disk space use,
16 * and a performance boost.  The element get and set operations are still
17 * O(ln(n)), but with a much smaller constant.
18 *
19 * The value type structure is reused from the btree type to support proper
20 * reference counting of values.
21 *
22 * The arrays implicitly know their length, and bounds are checked for
23 * lookups and updated.  It doesn't store this in an accessible place
24 * because it would waste a whole metadata block.  Make sure you store the
25 * size along with the array root in your encompassing data.
26 *
27 * Array entries are indexed via an unsigned integer starting from zero.
28 * Arrays are not sparse; if you resize an array to have 'n' entries then
29 * 'n - 1' will be the last valid index.
30 *
31 * Typical use:
32 *
33 * a) initialise a dm_array_info structure.  This describes the array
34 *    values and ties it into a specific transaction manager.  It holds no
35 *    instance data; the same info can be used for many similar arrays if
36 *    you wish.
37 *
38 * b) Get yourself a root.  The root is the index of a block of data on the
39 *    disk that holds a particular instance of an array.  You may have a
40 *    pre existing root in your metadata that you wish to use, or you may
41 *    want to create a brand new, empty array with dm_array_empty().
42 *
43 * Like the other data structures in this library, dm_array objects are
44 * immutable between transactions.  Update functions will return you the
45 * root for a _new_ array.  If you've incremented the old root, via
46 * dm_tm_inc(), before calling the update function you may continue to use
47 * it in parallel with the new root.
48 *
49 * c) resize an array with dm_array_resize().
50 *
51 * d) Get a value from the array with dm_array_get_value().
52 *
53 * e) Set a value in the array with dm_array_set_value().
54 *
55 * f) Walk an array of values in index order with dm_array_walk().  More
56 *    efficient than making many calls to dm_array_get_value().
57 *
58 * g) Destroy the array with dm_array_del().  This tells the transaction
59 *    manager that you're no longer using this data structure so it can
60 *    recycle it's blocks.  (dm_array_dec() would be a better name for it,
61 *    but del is in keeping with dm_btree_del()).
62 */
63
64/*
65 * Describes an array.  Don't initialise this structure yourself, use the
66 * init function below.
67 */
68struct dm_array_info {
69	struct dm_transaction_manager *tm;
70	struct dm_btree_value_type value_type;
71	struct dm_btree_info btree_info;
72};
73
74/*
75 * Sets up a dm_array_info structure.  You don't need to do anything with
76 * this structure when you finish using it.
77 *
78 * info - the structure being filled in.
79 * tm   - the transaction manager that should supervise this structure.
80 * vt   - describes the leaf values.
81 */
82void dm_array_info_init(struct dm_array_info *info,
83			struct dm_transaction_manager *tm,
84			struct dm_btree_value_type *vt);
85
86/*
87 * Create an empty, zero length array.
88 *
89 * info - describes the array
90 * root - on success this will be filled out with the root block
91 */
92int dm_array_empty(struct dm_array_info *info, dm_block_t *root);
93
94/*
95 * Resizes the array.
96 *
97 * info - describes the array
98 * root - the root block of the array on disk
99 * old_size - the caller is responsible for remembering the size of
100 *            the array
101 * new_size - can be bigger or smaller than old_size
102 * value - if we're growing the array the new entries will have this value
103 * new_root - on success, points to the new root block
104 *
105 * If growing the inc function for 'value' will be called the appropriate
106 * number of times.  So if the caller is holding a reference they may want
107 * to drop it.
108 */
109int dm_array_resize(struct dm_array_info *info, dm_block_t root,
110		    uint32_t old_size, uint32_t new_size,
111		    const void *value, dm_block_t *new_root)
112	__dm_written_to_disk(value);
113
114/*
115 * Creates a new array populated with values provided by a callback
116 * function.  This is more efficient than creating an empty array,
117 * resizing, and then setting values since that process incurs a lot of
118 * copying.
119 *
120 * Assumes 32bit values for now since it's only used by the cache hint
121 * array.
122 *
123 * info - describes the array
124 * root - the root block of the array on disk
125 * size - the number of entries in the array
126 * fn - the callback
127 * context - passed to the callback
128 */
129typedef int (*value_fn)(uint32_t index, void *value_le, void *context);
130int dm_array_new(struct dm_array_info *info, dm_block_t *root,
131		 uint32_t size, value_fn fn, void *context);
132
133/*
134 * Frees a whole array.  The value_type's decrement operation will be called
135 * for all values in the array
136 */
137int dm_array_del(struct dm_array_info *info, dm_block_t root);
138
139/*
140 * Lookup a value in the array
141 *
142 * info - describes the array
143 * root - root block of the array
144 * index - array index
145 * value - the value to be read.  Will be in on-disk format of course.
146 *
147 * -ENODATA will be returned if the index is out of bounds.
148 */
149int dm_array_get_value(struct dm_array_info *info, dm_block_t root,
150		       uint32_t index, void *value);
151
152/*
153 * Set an entry in the array.
154 *
155 * info - describes the array
156 * root - root block of the array
157 * index - array index
158 * value - value to be written to disk.  Make sure you confirm the value is
159 *         in on-disk format with__dm_bless_for_disk() before calling.
160 * new_root - the new root block
161 *
162 * The old value being overwritten will be decremented, the new value
163 * incremented.
164 *
165 * -ENODATA will be returned if the index is out of bounds.
166 */
167int dm_array_set_value(struct dm_array_info *info, dm_block_t root,
168		       uint32_t index, const void *value, dm_block_t *new_root)
169	__dm_written_to_disk(value);
170
171/*
172 * Walk through all the entries in an array.
173 *
174 * info - describes the array
175 * root - root block of the array
176 * fn - called back for every element
177 * context - passed to the callback
178 */
179int dm_array_walk(struct dm_array_info *info, dm_block_t root,
180		  int (*fn)(void *context, uint64_t key, void *leaf),
181		  void *context);
182
183/*----------------------------------------------------------------*/
184
185/*
186 * Cursor api.
187 *
188 * This lets you iterate through all the entries in an array efficiently
189 * (it will preload metadata).
190 *
191 * I'm using a cursor, rather than a walk function with a callback because
192 * the cache target needs to iterate both the mapping and hint arrays in
193 * unison.
194 */
195struct dm_array_cursor {
196	struct dm_array_info *info;
197	struct dm_btree_cursor cursor;
198
199	struct dm_block *block;
200	struct array_block *ab;
201	unsigned index;
202};
203
204int dm_array_cursor_begin(struct dm_array_info *info,
205			  dm_block_t root, struct dm_array_cursor *c);
206void dm_array_cursor_end(struct dm_array_cursor *c);
207
208uint32_t dm_array_cursor_index(struct dm_array_cursor *c);
209int dm_array_cursor_next(struct dm_array_cursor *c);
210int dm_array_cursor_skip(struct dm_array_cursor *c, uint32_t count);
211
212/*
213 * value_le is only valid while the cursor points at the current value.
214 */
215void dm_array_cursor_get_value(struct dm_array_cursor *c, void **value_le);
216
217/*----------------------------------------------------------------*/
218
219#endif	/* _LINUX_DM_ARRAY_H */
220