1/*
2 * Copyright (C) 2012 Red Hat. All rights reserved.
3 *
4 * writeback cache policy supporting flushing out dirty cache blocks.
5 *
6 * This file is released under the GPL.
7 */
8
9#include "dm-cache-policy.h"
10#include "dm.h"
11
12#include <linux/hash.h>
13#include <linux/module.h>
14#include <linux/slab.h>
15#include <linux/vmalloc.h>
16
17/*----------------------------------------------------------------*/
18
19#define DM_MSG_PREFIX "cache cleaner"
20
21/* Cache entry struct. */
22struct wb_cache_entry {
23	struct list_head list;
24	struct hlist_node hlist;
25
26	dm_oblock_t oblock;
27	dm_cblock_t cblock;
28	bool dirty:1;
29	bool pending:1;
30};
31
32struct hash {
33	struct hlist_head *table;
34	dm_block_t hash_bits;
35	unsigned nr_buckets;
36};
37
38struct policy {
39	struct dm_cache_policy policy;
40	spinlock_t lock;
41
42	struct list_head free;
43	struct list_head clean;
44	struct list_head clean_pending;
45	struct list_head dirty;
46
47	/*
48	 * We know exactly how many cblocks will be needed,
49	 * so we can allocate them up front.
50	 */
51	dm_cblock_t cache_size, nr_cblocks_allocated;
52	struct wb_cache_entry *cblocks;
53	struct hash chash;
54};
55
56/*----------------------------------------------------------------------------*/
57
58/*
59 * Low-level functions.
60 */
61static unsigned next_power(unsigned n, unsigned min)
62{
63	return roundup_pow_of_two(max(n, min));
64}
65
66static struct policy *to_policy(struct dm_cache_policy *p)
67{
68	return container_of(p, struct policy, policy);
69}
70
71static struct list_head *list_pop(struct list_head *q)
72{
73	struct list_head *r = q->next;
74
75	list_del(r);
76
77	return r;
78}
79
80/*----------------------------------------------------------------------------*/
81
82/* Allocate/free various resources. */
83static int alloc_hash(struct hash *hash, unsigned elts)
84{
85	hash->nr_buckets = next_power(elts >> 4, 16);
86	hash->hash_bits = ffs(hash->nr_buckets) - 1;
87	hash->table = vzalloc(sizeof(*hash->table) * hash->nr_buckets);
88
89	return hash->table ? 0 : -ENOMEM;
90}
91
92static void free_hash(struct hash *hash)
93{
94	vfree(hash->table);
95}
96
97static int alloc_cache_blocks_with_hash(struct policy *p, dm_cblock_t cache_size)
98{
99	int r = -ENOMEM;
100
101	p->cblocks = vzalloc(sizeof(*p->cblocks) * from_cblock(cache_size));
102	if (p->cblocks) {
103		unsigned u = from_cblock(cache_size);
104
105		while (u--)
106			list_add(&p->cblocks[u].list, &p->free);
107
108		p->nr_cblocks_allocated = 0;
109
110		/* Cache entries hash. */
111		r = alloc_hash(&p->chash, from_cblock(cache_size));
112		if (r)
113			vfree(p->cblocks);
114	}
115
116	return r;
117}
118
119static void free_cache_blocks_and_hash(struct policy *p)
120{
121	free_hash(&p->chash);
122	vfree(p->cblocks);
123}
124
125static struct wb_cache_entry *alloc_cache_entry(struct policy *p)
126{
127	struct wb_cache_entry *e;
128
129	BUG_ON(from_cblock(p->nr_cblocks_allocated) >= from_cblock(p->cache_size));
130
131	e = list_entry(list_pop(&p->free), struct wb_cache_entry, list);
132	p->nr_cblocks_allocated = to_cblock(from_cblock(p->nr_cblocks_allocated) + 1);
133
134	return e;
135}
136
137/*----------------------------------------------------------------------------*/
138
139/* Hash functions (lookup, insert, remove). */
140static struct wb_cache_entry *lookup_cache_entry(struct policy *p, dm_oblock_t oblock)
141{
142	struct hash *hash = &p->chash;
143	unsigned h = hash_64(from_oblock(oblock), hash->hash_bits);
144	struct wb_cache_entry *cur;
145	struct hlist_head *bucket = &hash->table[h];
146
147	hlist_for_each_entry(cur, bucket, hlist) {
148		if (cur->oblock == oblock) {
149			/* Move upfront bucket for faster access. */
150			hlist_del(&cur->hlist);
151			hlist_add_head(&cur->hlist, bucket);
152			return cur;
153		}
154	}
155
156	return NULL;
157}
158
159static void insert_cache_hash_entry(struct policy *p, struct wb_cache_entry *e)
160{
161	unsigned h = hash_64(from_oblock(e->oblock), p->chash.hash_bits);
162
163	hlist_add_head(&e->hlist, &p->chash.table[h]);
164}
165
166static void remove_cache_hash_entry(struct wb_cache_entry *e)
167{
168	hlist_del(&e->hlist);
169}
170
171/* Public interface (see dm-cache-policy.h */
172static int wb_map(struct dm_cache_policy *pe, dm_oblock_t oblock,
173		  bool can_block, bool can_migrate, bool discarded_oblock,
174		  struct bio *bio, struct policy_result *result)
175{
176	struct policy *p = to_policy(pe);
177	struct wb_cache_entry *e;
178	unsigned long flags;
179
180	result->op = POLICY_MISS;
181
182	if (can_block)
183		spin_lock_irqsave(&p->lock, flags);
184
185	else if (!spin_trylock_irqsave(&p->lock, flags))
186		return -EWOULDBLOCK;
187
188	e = lookup_cache_entry(p, oblock);
189	if (e) {
190		result->op = POLICY_HIT;
191		result->cblock = e->cblock;
192
193	}
194
195	spin_unlock_irqrestore(&p->lock, flags);
196
197	return 0;
198}
199
200static int wb_lookup(struct dm_cache_policy *pe, dm_oblock_t oblock, dm_cblock_t *cblock)
201{
202	int r;
203	struct policy *p = to_policy(pe);
204	struct wb_cache_entry *e;
205	unsigned long flags;
206
207	if (!spin_trylock_irqsave(&p->lock, flags))
208		return -EWOULDBLOCK;
209
210	e = lookup_cache_entry(p, oblock);
211	if (e) {
212		*cblock = e->cblock;
213		r = 0;
214
215	} else
216		r = -ENOENT;
217
218	spin_unlock_irqrestore(&p->lock, flags);
219
220	return r;
221}
222
223static void __set_clear_dirty(struct dm_cache_policy *pe, dm_oblock_t oblock, bool set)
224{
225	struct policy *p = to_policy(pe);
226	struct wb_cache_entry *e;
227
228	e = lookup_cache_entry(p, oblock);
229	BUG_ON(!e);
230
231	if (set) {
232		if (!e->dirty) {
233			e->dirty = true;
234			list_move(&e->list, &p->dirty);
235		}
236
237	} else {
238		if (e->dirty) {
239			e->pending = false;
240			e->dirty = false;
241			list_move(&e->list, &p->clean);
242		}
243	}
244}
245
246static void wb_set_dirty(struct dm_cache_policy *pe, dm_oblock_t oblock)
247{
248	struct policy *p = to_policy(pe);
249	unsigned long flags;
250
251	spin_lock_irqsave(&p->lock, flags);
252	__set_clear_dirty(pe, oblock, true);
253	spin_unlock_irqrestore(&p->lock, flags);
254}
255
256static void wb_clear_dirty(struct dm_cache_policy *pe, dm_oblock_t oblock)
257{
258	struct policy *p = to_policy(pe);
259	unsigned long flags;
260
261	spin_lock_irqsave(&p->lock, flags);
262	__set_clear_dirty(pe, oblock, false);
263	spin_unlock_irqrestore(&p->lock, flags);
264}
265
266static void add_cache_entry(struct policy *p, struct wb_cache_entry *e)
267{
268	insert_cache_hash_entry(p, e);
269	if (e->dirty)
270		list_add(&e->list, &p->dirty);
271	else
272		list_add(&e->list, &p->clean);
273}
274
275static int wb_load_mapping(struct dm_cache_policy *pe,
276			   dm_oblock_t oblock, dm_cblock_t cblock,
277			   uint32_t hint, bool hint_valid)
278{
279	int r;
280	struct policy *p = to_policy(pe);
281	struct wb_cache_entry *e = alloc_cache_entry(p);
282
283	if (e) {
284		e->cblock = cblock;
285		e->oblock = oblock;
286		e->dirty = false; /* blocks default to clean */
287		add_cache_entry(p, e);
288		r = 0;
289
290	} else
291		r = -ENOMEM;
292
293	return r;
294}
295
296static void wb_destroy(struct dm_cache_policy *pe)
297{
298	struct policy *p = to_policy(pe);
299
300	free_cache_blocks_and_hash(p);
301	kfree(p);
302}
303
304static struct wb_cache_entry *__wb_force_remove_mapping(struct policy *p, dm_oblock_t oblock)
305{
306	struct wb_cache_entry *r = lookup_cache_entry(p, oblock);
307
308	BUG_ON(!r);
309
310	remove_cache_hash_entry(r);
311	list_del(&r->list);
312
313	return r;
314}
315
316static void wb_remove_mapping(struct dm_cache_policy *pe, dm_oblock_t oblock)
317{
318	struct policy *p = to_policy(pe);
319	struct wb_cache_entry *e;
320	unsigned long flags;
321
322	spin_lock_irqsave(&p->lock, flags);
323	e = __wb_force_remove_mapping(p, oblock);
324	list_add_tail(&e->list, &p->free);
325	BUG_ON(!from_cblock(p->nr_cblocks_allocated));
326	p->nr_cblocks_allocated = to_cblock(from_cblock(p->nr_cblocks_allocated) - 1);
327	spin_unlock_irqrestore(&p->lock, flags);
328}
329
330static void wb_force_mapping(struct dm_cache_policy *pe,
331				dm_oblock_t current_oblock, dm_oblock_t oblock)
332{
333	struct policy *p = to_policy(pe);
334	struct wb_cache_entry *e;
335	unsigned long flags;
336
337	spin_lock_irqsave(&p->lock, flags);
338	e = __wb_force_remove_mapping(p, current_oblock);
339	e->oblock = oblock;
340	add_cache_entry(p, e);
341	spin_unlock_irqrestore(&p->lock, flags);
342}
343
344static struct wb_cache_entry *get_next_dirty_entry(struct policy *p)
345{
346	struct list_head *l;
347	struct wb_cache_entry *r;
348
349	if (list_empty(&p->dirty))
350		return NULL;
351
352	l = list_pop(&p->dirty);
353	r = container_of(l, struct wb_cache_entry, list);
354	list_add(l, &p->clean_pending);
355
356	return r;
357}
358
359static int wb_writeback_work(struct dm_cache_policy *pe,
360			     dm_oblock_t *oblock,
361			     dm_cblock_t *cblock)
362{
363	int r = -ENOENT;
364	struct policy *p = to_policy(pe);
365	struct wb_cache_entry *e;
366	unsigned long flags;
367
368	spin_lock_irqsave(&p->lock, flags);
369
370	e = get_next_dirty_entry(p);
371	if (e) {
372		*oblock = e->oblock;
373		*cblock = e->cblock;
374		r = 0;
375	}
376
377	spin_unlock_irqrestore(&p->lock, flags);
378
379	return r;
380}
381
382static dm_cblock_t wb_residency(struct dm_cache_policy *pe)
383{
384	return to_policy(pe)->nr_cblocks_allocated;
385}
386
387/* Init the policy plugin interface function pointers. */
388static void init_policy_functions(struct policy *p)
389{
390	p->policy.destroy = wb_destroy;
391	p->policy.map = wb_map;
392	p->policy.lookup = wb_lookup;
393	p->policy.set_dirty = wb_set_dirty;
394	p->policy.clear_dirty = wb_clear_dirty;
395	p->policy.load_mapping = wb_load_mapping;
396	p->policy.walk_mappings = NULL;
397	p->policy.remove_mapping = wb_remove_mapping;
398	p->policy.writeback_work = wb_writeback_work;
399	p->policy.force_mapping = wb_force_mapping;
400	p->policy.residency = wb_residency;
401	p->policy.tick = NULL;
402}
403
404static struct dm_cache_policy *wb_create(dm_cblock_t cache_size,
405					 sector_t origin_size,
406					 sector_t cache_block_size)
407{
408	int r;
409	struct policy *p = kzalloc(sizeof(*p), GFP_KERNEL);
410
411	if (!p)
412		return NULL;
413
414	init_policy_functions(p);
415	INIT_LIST_HEAD(&p->free);
416	INIT_LIST_HEAD(&p->clean);
417	INIT_LIST_HEAD(&p->clean_pending);
418	INIT_LIST_HEAD(&p->dirty);
419
420	p->cache_size = cache_size;
421	spin_lock_init(&p->lock);
422
423	/* Allocate cache entry structs and add them to free list. */
424	r = alloc_cache_blocks_with_hash(p, cache_size);
425	if (!r)
426		return &p->policy;
427
428	kfree(p);
429
430	return NULL;
431}
432/*----------------------------------------------------------------------------*/
433
434static struct dm_cache_policy_type wb_policy_type = {
435	.name = "cleaner",
436	.version = {1, 0, 0},
437	.hint_size = 0,
438	.owner = THIS_MODULE,
439	.create = wb_create
440};
441
442static int __init wb_init(void)
443{
444	int r = dm_cache_policy_register(&wb_policy_type);
445
446	if (r < 0)
447		DMERR("register failed %d", r);
448	else
449		DMINFO("version %u.%u.%u loaded",
450		       wb_policy_type.version[0],
451		       wb_policy_type.version[1],
452		       wb_policy_type.version[2]);
453
454	return r;
455}
456
457static void __exit wb_exit(void)
458{
459	dm_cache_policy_unregister(&wb_policy_type);
460}
461
462module_init(wb_init);
463module_exit(wb_exit);
464
465MODULE_AUTHOR("Heinz Mauelshagen <dm-devel@redhat.com>");
466MODULE_LICENSE("GPL");
467MODULE_DESCRIPTION("cleaner cache policy");
468