summaryrefslogtreecommitdiffstats
path: root/mm/frontswap.c
blob: 2183a56c7874ba3e61930a95d5d87bde08ef9bd6 (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
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
// SPDX-License-Identifier: GPL-2.0-only
/*
 * Frontswap frontend
 *
 * This code provides the generic "frontend" layer to call a matching
 * "backend" driver implementation of frontswap.  See
 * Documentation/vm/frontswap.rst for more information.
 *
 * Copyright (C) 2009-2012 Oracle Corp.  All rights reserved.
 * Author: Dan Magenheimer
 */

#include <linux/mman.h>
#include <linux/swap.h>
#include <linux/swapops.h>
#include <linux/security.h>
#include <linux/module.h>
#include <linux/debugfs.h>
#include <linux/frontswap.h>
#include <linux/swapfile.h>

DEFINE_STATIC_KEY_FALSE(frontswap_enabled_key);

/*
 * frontswap_ops are added by frontswap_register_ops, and provide the
 * frontswap "backend" implementation functions.  Multiple implementations
 * may be registered, but implementations can never deregister.  This
 * is a simple singly-linked list of all registered implementations.
 */
static struct frontswap_ops *frontswap_ops __read_mostly;

#define for_each_frontswap_ops(ops)		\
	for ((ops) = frontswap_ops; (ops); (ops) = (ops)->next)

/*
 * If enabled, frontswap_store will return failure even on success.  As
 * a result, the swap subsystem will always write the page to swap, in
 * effect converting frontswap into a writethrough cache.  In this mode,
 * there is no direct reduction in swap writes, but a frontswap backend
 * can unilaterally "reclaim" any pages in use with no data loss, thus
 * providing increases control over maximum memory usage due to frontswap.
 */
static bool frontswap_writethrough_enabled __read_mostly;

/*
 * If enabled, the underlying tmem implementation is capable of doing
 * exclusive gets, so frontswap_load, on a successful tmem_get must
 * mark the page as no longer in frontswap AND mark it dirty.
 */
static bool frontswap_tmem_exclusive_gets_enabled __read_mostly;

#ifdef CONFIG_DEBUG_FS
/*
 * Counters available via /sys/kernel/debug/frontswap (if debugfs is
 * properly configured).  These are for information only so are not protected
 * against increment races.
 */
static u64 frontswap_loads;
static u64 frontswap_succ_stores;
static u64 frontswap_failed_stores;
static u64 frontswap_invalidates;

static inline void inc_frontswap_loads(void) {
	data_race(frontswap_loads++);
}
static inline void inc_frontswap_succ_stores(void) {
	data_race(frontswap_succ_stores++);
}
static inline void inc_frontswap_failed_stores(void) {
	data_race(frontswap_failed_stores++);
}
static inline void inc_frontswap_invalidates(void) {
	data_race(frontswap_invalidates++);
}
#else
static inline void inc_frontswap_loads(void) { }
static inline void inc_frontswap_succ_stores(void) { }
static inline void inc_frontswap_failed_stores(void) { }
static inline void inc_frontswap_invalidates(void) { }
#endif

/*
 * Due to the asynchronous nature of the backends loading potentially
 * _after_ the swap system has been activated, we have chokepoints
 * on all frontswap functions to not call the backend until the backend
 * has registered.
 *
 * This would not guards us against the user deciding to call swapoff right as
 * we are calling the backend to initialize (so swapon is in action).
 * Fortunately for us, the swapon_mutex has been taken by the callee so we are
 * OK. The other scenario where calls to frontswap_store (called via
 * swap_writepage) is racing with frontswap_invalidate_area (called via
 * swapoff) is again guarded by the swap subsystem.
 *
 * While no backend is registered all calls to frontswap_[store|load|
 * invalidate_area|invalidate_page] are ignored or fail.
 *
 * The time between the backend being registered and the swap file system
 * calling the backend (via the frontswap_* functions) is indeterminate as
 * frontswap_ops is not atomic_t (or a value guarded by a spinlock).
 * That is OK as we are comfortable missing some of these calls to the newly
 * registered backend.
 *
 * Obviously the opposite (unloading the backend) must be done after all
 * the frontswap_[store|load|invalidate_area|invalidate_page] start
 * ignoring or failing the requests.  However, there is currently no way
 * to unload a backend once it is registered.
 */

/*
 * Register operations for frontswap
 */
void frontswap_register_ops(struct frontswap_ops *ops)
{
	DECLARE_BITMAP(a, MAX_SWAPFILES);
	DECLARE_BITMAP(b, MAX_SWAPFILES);
	struct swap_info_struct *si;
	unsigned int i;

	bitmap_zero(a, MAX_SWAPFILES);
	bitmap_zero(b, MAX_SWAPFILES);

	spin_lock(&swap_lock);
	plist_for_each_entry(si, &swap_active_head, list) {
		if (!WARN_ON(!si->frontswap_map))
			set_bit(si->type, a);
	}
	spin_unlock(&swap_lock);

	/* the new ops needs to know the currently active swap devices */
	for_each_set_bit(i, a, MAX_SWAPFILES)
		ops->init(i);

	/*
	 * Setting frontswap_ops must happen after the ops->init() calls
	 * above; cmpxchg implies smp_mb() which will ensure the init is
	 * complete at this point.
	 */
	do {
		ops->next = frontswap_ops;
	} while (cmpxchg(&frontswap_ops, ops->next, ops) != ops->next);

	static_branch_inc(&frontswap_enabled_key);

	spin_lock(&swap_lock);
	plist_for_each_entry(si, &swap_active_head, list) {
		if (si->frontswap_map)
			set_bit(si->type, b);
	}
	spin_unlock(&swap_lock);

	/*
	 * On the very unlikely chance that a swap device was added or
	 * removed between setting the "a" list bits and the ops init
	 * calls, we re-check and do init or invalidate for any changed
	 * bits.
	 */
	if (unlikely(!bitmap_equal(a, b, MAX_SWAPFILES))) {
		for (i = 0; i < MAX_SWAPFILES; i++) {
			if (!test_bit(i, a) && test_bit(i, b))
				ops->init(i);
			else if (test_bit(i, a) && !test_bit(i, b))
				ops->invalidate_area(i);
		}
	}
}
EXPORT_SYMBOL(frontswap_register_ops);

/*
 * Enable/disable frontswap writethrough (see above).
 */
void frontswap_writethrough(bool enable)
{
	frontswap_writethrough_enabled = enable;
}
EXPORT_SYMBOL(frontswap_writethrough);

/*
 * Enable/disable frontswap exclusive gets (see above).
 */
void frontswap_tmem_exclusive_gets(bool enable)
{
	frontswap_tmem_exclusive_gets_enabled = enable;
}
EXPORT_SYMBOL(frontswap_tmem_exclusive_gets);

/*
 * Called when a swap device is swapon'd.
 */
void __frontswap_init(unsigned type, unsigned long *map)
{
	struct swap_info_struct *sis = swap_info[type];
	struct frontswap_ops *ops;

	VM_BUG_ON(sis == NULL);

	/*
	 * p->frontswap is a bitmap that we MUST have to figure out which page
	 * has gone in frontswap. Without it there is no point of continuing.
	 */
	if (WARN_ON(!map))
		return;
	/*
	 * Irregardless of whether the frontswap backend has been loaded
	 * before this function or it will be later, we _MUST_ have the
	 * p->frontswap set to something valid to work properly.
	 */
	frontswap_map_set(sis, map);

	for_each_frontswap_ops(ops)
		ops->init(type);
}
EXPORT_SYMBOL(__frontswap_init);

bool __frontswap_test(struct swap_info_struct *sis,
				pgoff_t offset)
{
	if (sis->frontswap_map)
		return test_bit(offset, sis->frontswap_map);
	return false;
}
EXPORT_SYMBOL(__frontswap_test);

static inline void __frontswap_set(struct swap_info_struct *sis,
				   pgoff_t offset)
{
	set_bit(offset, sis->frontswap_map);
	atomic_inc(&sis->frontswap_pages);
}

static inline void __frontswap_clear(struct swap_info_struct *sis,
				     pgoff_t offset)
{
	clear_bit(offset, sis->frontswap_map);
	atomic_dec(&sis->frontswap_pages);
}

/*
 * "Store" data from a page to frontswap and associate it with the page's
 * swaptype and offset.  Page must be locked and in the swap cache.
 * If frontswap already contains a page with matching swaptype and
 * offset, the frontswap implementation may either overwrite the data and
 * return success or invalidate the page from frontswap and return failure.
 */
int __frontswap_store(struct page *page)
{
	int ret = -1;
	swp_entry_t entry = { .val = page_private(page), };
	int type = swp_type(entry);
	struct swap_info_struct *sis = swap_info[type];
	pgoff_t offset = swp_offset(entry);
	struct frontswap_ops *ops;

	VM_BUG_ON(!frontswap_ops);
	VM_BUG_ON(!PageLocked(page));
	VM_BUG_ON(sis == NULL);

	/*
	 * If a dup, we must remove the old page first; we can't leave the
	 * old page no matter if the store of the new page succeeds or fails,
	 * and we can't rely on the new page replacing the old page as we may
	 * not store to the same implementation that contains the old page.
	 */
	if (__frontswap_test(sis, offset)) {
		__frontswap_clear(sis, offset);
		for_each_frontswap_ops(ops)
			ops->invalidate_page(type, offset);
	}

	/* Try to store in each implementation, until one succeeds. */
	for_each_frontswap_ops(ops) {
		ret = ops->store(type, offset, page);
		if (!ret) /* successful store */
			break;
	}
	if (ret == 0) {
		__frontswap_set(sis, offset);
		inc_frontswap_succ_stores();
	} else {
		inc_frontswap_failed_stores();
	}
	if (frontswap_writethrough_enabled)
		/* report failure so swap also writes to swap device */
		ret = -1;
	return ret;
}
EXPORT_SYMBOL(__frontswap_store);

/*
 * "Get" data from frontswap associated with swaptype and offset that were
 * specified when the data was put to frontswap and use it to fill the
 * specified page with data. Page must be locked and in the swap cache.
 */
int __frontswap_load(struct page *page)
{
	int ret = -1;
	swp_entry_t entry = { .val = page_private(page), };
	int type = swp_type(entry);
	struct swap_info_struct *sis = swap_info[type];
	pgoff_t offset = swp_offset(entry);
	struct frontswap_ops *ops;

	VM_BUG_ON(!frontswap_ops);
	VM_BUG_ON(!PageLocked(page));
	VM_BUG_ON(sis == NULL);

	if (!__frontswap_test(sis, offset))
		return -1;

	/* Try loading from each implementation, until one succeeds. */
	for_each_frontswap_ops(ops) {
		ret = ops->load(type, offset, page