summaryrefslogtreecommitdiff
path: root/fs/erofs/utils.c
blob: 518bdd69c82325c70fee1ec437fda5fb7b3ef010 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
// SPDX-License-Identifier: GPL-2.0-only
/*
 * Copyright (C) 2018 HUAWEI, Inc.
 *             https://www.huawei.com/
 */
#include "internal.h"

struct page *erofs_allocpage(struct page **pagepool, gfp_t gfp)
{
	struct page *page = *pagepool;

	if (page) {
		DBG_BUGON(page_ref_count(page) != 1);
		*pagepool = (struct page *)page_private(page);
	} else {
		page = alloc_page(gfp);
	}
	return page;
}

void erofs_release_pages(struct page **pagepool)
{
	while (*pagepool) {
		struct page *page = *pagepool;

		*pagepool = (struct page *)page_private(page);
		put_page(page);
	}
}

#ifdef CONFIG_EROFS_FS_ZIP
/* global shrink count (for all mounted EROFS instances) */
static atomic_long_t erofs_global_shrink_cnt;

static bool erofs_workgroup_get(struct erofs_workgroup *grp)
{
	if (lockref_get_not_zero(&grp->lockref))
		return true;

	spin_lock(&grp->lockref.lock);
	if (__lockref_is_dead(&grp->lockref)) {
		spin_unlock(&grp->lockref.lock);
		return false;
	}

	if (!grp->lockref.count++)
		atomic_long_dec(&erofs_global_shrink_cnt);
	spin_unlock(&grp->lockref.lock);
	return true;
}

struct erofs_workgroup *erofs_find_workgroup(struct super_block *sb,
					     pgoff_t index)
{
	struct erofs_sb_info *sbi = EROFS_SB(sb);
	struct erofs_workgroup *grp;

repeat:
	rcu_read_lock();
	grp = xa_load(&sbi->managed_pslots, index);
	if (grp) {
		if (!erofs_workgroup_get(grp)) {
			/* prefer to relax rcu read side */
			rcu_read_unlock();
			goto repeat;
		}

		DBG_BUGON(index != grp->index);
	}
	rcu_read_unlock();
	return grp;
}

struct erofs_workgroup *erofs_insert_workgroup(struct super_block *sb,
					       struct erofs_workgroup *grp)
{
	struct erofs_sb_info *const sbi = EROFS_SB(sb);
	struct erofs_workgroup *pre;

	DBG_BUGON(grp->lockref.count < 1);
repeat:
	xa_lock(&sbi->managed_pslots);
	pre = __xa_cmpxchg(&sbi->managed_pslots, grp->index,
			   NULL, grp, GFP_KERNEL);
	if (pre) {
		if (xa_is_err(pre)) {
			pre = ERR_PTR(xa_err(pre));
		} else if (!erofs_workgroup_get(pre)) {
			/* try to legitimize the current in-tree one */
			xa_unlock(&sbi->managed_pslots);
			cond_resched();
			goto repeat;
		}
		grp = pre;
	}
	xa_unlock(&sbi->managed_pslots);
	return grp;
}

static void  __erofs_workgroup_free(struct erofs_workgroup *grp)
{
	atomic_long_dec(&erofs_global_shrink_cnt);
	erofs_workgroup_free_rcu(grp);
}

void erofs_workgroup_put(struct erofs_workgroup *grp)
{
	if (lockref_put_or_lock(&grp->lockref))
		return;

	DBG_BUGON(__lockref_is_dead(&grp->lockref));
	if (grp->lockref.count == 1)
		atomic_long_inc(&erofs_global_shrink_cnt);
	--grp->lockref.count;
	spin_unlock(&grp->lockref.lock);
}

static bool erofs_try_to_release_workgroup(struct erofs_sb_info *sbi,
					   struct erofs_workgroup *grp)
{
	int free = false;

	spin_lock(&grp->lockref.lock);
	if (grp->lockref.count)
		goto out;

	/*
	 * Note that all cached pages should be detached before deleted from
	 * the XArray. Otherwise some cached pages could be still attached to
	 * the orphan old workgroup when the new one is available in the tree.
	 */
	if (erofs_try_to_free_all_cached_folios(sbi, grp))
		goto out;

	/*
	 * It's impossible to fail after the workgroup is freezed,
	 * however in order to avoid some race conditions, add a
	 * DBG_BUGON to observe this in advance.
	 */
	DBG_BUGON(__xa_erase(&sbi->managed_pslots, grp->index) != grp);

	lockref_mark_dead(&grp->lockref);
	free = true;
out:
	spin_unlock(&grp->lockref.lock);
	if (free)
		__erofs_workgroup_free(grp);
	return free;
}

static unsigned long erofs_shrink_workstation(struct erofs_sb_info *sbi,
					      unsigned long nr_shrink)
{
	struct erofs_workgroup *grp;
	unsigned int freed = 0;
	unsigned long index;

	xa_lock(&sbi->managed_pslots);
	xa_for_each(&sbi->managed_pslots, index, grp) {
		/* try to shrink each valid workgroup */
		if (!erofs_try_to_release_workgroup(sbi, grp))
			continue;
		xa_unlock(&sbi->managed_pslots);

		++freed;
		if (!--nr_shrink)
			return freed;
		xa_lock(&sbi->managed_pslots);
	}
	xa_unlock(&sbi->managed_pslots);
	return freed;
}

/* protected by 'erofs_sb_list_lock' */
static unsigned int shrinker_run_no;

/* protects the mounted 'erofs_sb_list' */
static DEFINE_SPINLOCK(erofs_sb_list_lock);
static LIST_HEAD(erofs_sb_list);

void erofs_shrinker_register(struct super_block *sb)
{
	struct erofs_sb_info *sbi = EROFS_SB(sb);

	mutex_init(&sbi->umount_mutex);

	spin_lock(&erofs_sb_list_lock);
	list_add(&sbi->list, &erofs_sb_list);
	spin_unlock(&erofs_sb_list_lock);
}

void erofs_shrinker_unregister(struct super_block *sb)
{
	struct erofs_sb_info *const sbi = EROFS_SB(sb);

	mutex_lock(&sbi->umount_mutex);
	/* clean up all remaining workgroups in memory */
	erofs_shrink_workstation(sbi, ~0UL);

	spin_lock(&erofs_sb_list_lock);
	list_del(&sbi->list);
	spin_unlock(&erofs_sb_list_lock);
	mutex_unlock(&sbi->umount_mutex);
}

static unsigned long erofs_shrink_count(struct shrinker *shrink,
					struct shrink_control *sc)
{
	return atomic_long_read(&erofs_global_shrink_cnt);
}

static unsigned long erofs_shrink_scan(struct shrinker *shrink,
				       struct shrink_control *sc)
{
	struct erofs_sb_info *sbi;
	struct list_head *p;

	unsigned long nr = sc->nr_to_scan;
	unsigned int run_no;
	unsigned long freed = 0;

	spin_lock(&erofs_sb_list_lock);
	do {
		run_no = ++shrinker_run_no;
	} while (run_no == 0);

	/* Iterate over all mounted superblocks and try to shrink them */
	p = erofs_sb_list.next;
	while (p != &erofs_sb_list) {
		sbi = list_entry(p, struct erofs_sb_info, list);

		/*
		 * We move the ones we do to the end of the list, so we stop
		 * when we see one we have already done.
		 */
		if (sbi->shrinker_run_no == run_no)
			break;

		if (!mutex_trylock(&sbi->umount_mutex)) {
			p = p->next;
			continue;
		}

		spin_unlock(&erofs_sb_list_lock);
		sbi->shrinker_run_no = run_no;

		freed += erofs_shrink_workstation(sbi, nr - freed);

		spin_lock(&erofs_sb_list_lock);
		/* Get the next list element before we move this one */
		p = p->next;

		/*
		 * Move this one to the end of the list to provide some
		 * fairness.
		 */
		list_move_tail(&sbi->list, &erofs_sb_list);
		mutex_unlock(&sbi->umount_mutex);

		if (freed >= nr)
			break;
	}
	spin_unlock(&erofs_sb_list_lock);
	return freed;
}

static struct shrinker *erofs_shrinker_info;

int __init erofs_init_shrinker(void)
{
	erofs_shrinker_info = shrinker_alloc(0, "erofs-shrinker");
	if (!erofs_shrinker_info)
		return -ENOMEM;

	erofs_shrinker_info->count_objects = erofs_shrink_count;
	erofs_shrinker_info->scan_objects = erofs_shrink_scan;

	shrinker_register(erofs_shrinker_info);

	return 0;
}

void erofs_exit_shrinker(void)
{
	shrinker_free(erofs_shrinker_info);
}
#endif	/* !CONFIG_EROFS_FS_ZIP */