xref: /linux/mm/list_lru.c (revision 50a0844bf8c4d38be540e423672ef9408d029252)
1 // SPDX-License-Identifier: GPL-2.0-only
2 /*
3  * Copyright (c) 2013 Red Hat, Inc. and Parallels Inc. All rights reserved.
4  * Authors: David Chinner and Glauber Costa
5  *
6  * Generic LRU infrastructure
7  */
8 #include <linux/kernel.h>
9 #include <linux/module.h>
10 #include <linux/mm.h>
11 #include <linux/list_lru.h>
12 #include <linux/slab.h>
13 #include <linux/mutex.h>
14 #include <linux/memcontrol.h>
15 #include "slab.h"
16 #include "internal.h"
17 
18 #ifdef CONFIG_MEMCG_KMEM
19 static LIST_HEAD(memcg_list_lrus);
20 static DEFINE_MUTEX(list_lrus_mutex);
21 
22 static inline bool list_lru_memcg_aware(struct list_lru *lru)
23 {
24 	return lru->memcg_aware;
25 }
26 
27 static void list_lru_register(struct list_lru *lru)
28 {
29 	if (!list_lru_memcg_aware(lru))
30 		return;
31 
32 	mutex_lock(&list_lrus_mutex);
33 	list_add(&lru->list, &memcg_list_lrus);
34 	mutex_unlock(&list_lrus_mutex);
35 }
36 
37 static void list_lru_unregister(struct list_lru *lru)
38 {
39 	if (!list_lru_memcg_aware(lru))
40 		return;
41 
42 	mutex_lock(&list_lrus_mutex);
43 	list_del(&lru->list);
44 	mutex_unlock(&list_lrus_mutex);
45 }
46 
47 static int lru_shrinker_id(struct list_lru *lru)
48 {
49 	return lru->shrinker_id;
50 }
51 
52 static inline struct list_lru_one *
53 list_lru_from_memcg_idx(struct list_lru *lru, int nid, int idx)
54 {
55 	if (list_lru_memcg_aware(lru) && idx >= 0) {
56 		struct list_lru_memcg *mlru = xa_load(&lru->xa, idx);
57 
58 		return mlru ? &mlru->node[nid] : NULL;
59 	}
60 	return &lru->node[nid].lru;
61 }
62 #else
63 static void list_lru_register(struct list_lru *lru)
64 {
65 }
66 
67 static void list_lru_unregister(struct list_lru *lru)
68 {
69 }
70 
71 static int lru_shrinker_id(struct list_lru *lru)
72 {
73 	return -1;
74 }
75 
76 static inline bool list_lru_memcg_aware(struct list_lru *lru)
77 {
78 	return false;
79 }
80 
81 static inline struct list_lru_one *
82 list_lru_from_memcg_idx(struct list_lru *lru, int nid, int idx)
83 {
84 	return &lru->node[nid].lru;
85 }
86 #endif /* CONFIG_MEMCG_KMEM */
87 
88 bool list_lru_add(struct list_lru *lru, struct list_head *item, int nid,
89 		    struct mem_cgroup *memcg)
90 {
91 	struct list_lru_node *nlru = &lru->node[nid];
92 	struct list_lru_one *l;
93 
94 	spin_lock(&nlru->lock);
95 	if (list_empty(item)) {
96 		l = list_lru_from_memcg_idx(lru, nid, memcg_kmem_id(memcg));
97 		list_add_tail(item, &l->list);
98 		/* Set shrinker bit if the first element was added */
99 		if (!l->nr_items++)
100 			set_shrinker_bit(memcg, nid, lru_shrinker_id(lru));
101 		nlru->nr_items++;
102 		spin_unlock(&nlru->lock);
103 		return true;
104 	}
105 	spin_unlock(&nlru->lock);
106 	return false;
107 }
108 EXPORT_SYMBOL_GPL(list_lru_add);
109 
110 bool list_lru_add_obj(struct list_lru *lru, struct list_head *item)
111 {
112 	int nid = page_to_nid(virt_to_page(item));
113 	struct mem_cgroup *memcg = list_lru_memcg_aware(lru) ?
114 		mem_cgroup_from_slab_obj(item) : NULL;
115 
116 	return list_lru_add(lru, item, nid, memcg);
117 }
118 EXPORT_SYMBOL_GPL(list_lru_add_obj);
119 
120 bool list_lru_del(struct list_lru *lru, struct list_head *item, int nid,
121 		    struct mem_cgroup *memcg)
122 {
123 	struct list_lru_node *nlru = &lru->node[nid];
124 	struct list_lru_one *l;
125 
126 	spin_lock(&nlru->lock);
127 	if (!list_empty(item)) {
128 		l = list_lru_from_memcg_idx(lru, nid, memcg_kmem_id(memcg));
129 		list_del_init(item);
130 		l->nr_items--;
131 		nlru->nr_items--;
132 		spin_unlock(&nlru->lock);
133 		return true;
134 	}
135 	spin_unlock(&nlru->lock);
136 	return false;
137 }
138 EXPORT_SYMBOL_GPL(list_lru_del);
139 
140 bool list_lru_del_obj(struct list_lru *lru, struct list_head *item)
141 {
142 	int nid = page_to_nid(virt_to_page(item));
143 	struct mem_cgroup *memcg = list_lru_memcg_aware(lru) ?
144 		mem_cgroup_from_slab_obj(item) : NULL;
145 
146 	return list_lru_del(lru, item, nid, memcg);
147 }
148 EXPORT_SYMBOL_GPL(list_lru_del_obj);
149 
150 void list_lru_isolate(struct list_lru_one *list, struct list_head *item)
151 {
152 	list_del_init(item);
153 	list->nr_items--;
154 }
155 EXPORT_SYMBOL_GPL(list_lru_isolate);
156 
157 void list_lru_isolate_move(struct list_lru_one *list, struct list_head *item,
158 			   struct list_head *head)
159 {
160 	list_move(item, head);
161 	list->nr_items--;
162 }
163 EXPORT_SYMBOL_GPL(list_lru_isolate_move);
164 
165 unsigned long list_lru_count_one(struct list_lru *lru,
166 				 int nid, struct mem_cgroup *memcg)
167 {
168 	struct list_lru_one *l;
169 	long count;
170 
171 	rcu_read_lock();
172 	l = list_lru_from_memcg_idx(lru, nid, memcg_kmem_id(memcg));
173 	count = l ? READ_ONCE(l->nr_items) : 0;
174 	rcu_read_unlock();
175 
176 	if (unlikely(count < 0))
177 		count = 0;
178 
179 	return count;
180 }
181 EXPORT_SYMBOL_GPL(list_lru_count_one);
182 
183 unsigned long list_lru_count_node(struct list_lru *lru, int nid)
184 {
185 	struct list_lru_node *nlru;
186 
187 	nlru = &lru->node[nid];
188 	return nlru->nr_items;
189 }
190 EXPORT_SYMBOL_GPL(list_lru_count_node);
191 
192 static unsigned long
193 __list_lru_walk_one(struct list_lru *lru, int nid, int memcg_idx,
194 		    list_lru_walk_cb isolate, void *cb_arg,
195 		    unsigned long *nr_to_walk)
196 {
197 	struct list_lru_node *nlru = &lru->node[nid];
198 	struct list_lru_one *l;
199 	struct list_head *item, *n;
200 	unsigned long isolated = 0;
201 
202 restart:
203 	l = list_lru_from_memcg_idx(lru, nid, memcg_idx);
204 	if (!l)
205 		goto out;
206 
207 	list_for_each_safe(item, n, &l->list) {
208 		enum lru_status ret;
209 
210 		/*
211 		 * decrement nr_to_walk first so that we don't livelock if we
212 		 * get stuck on large numbers of LRU_RETRY items
213 		 */
214 		if (!*nr_to_walk)
215 			break;
216 		--*nr_to_walk;
217 
218 		ret = isolate(item, l, &nlru->lock, cb_arg);
219 		switch (ret) {
220 		case LRU_REMOVED_RETRY:
221 			assert_spin_locked(&nlru->lock);
222 			fallthrough;
223 		case LRU_REMOVED:
224 			isolated++;
225 			nlru->nr_items--;
226 			/*
227 			 * If the lru lock has been dropped, our list
228 			 * traversal is now invalid and so we have to
229 			 * restart from scratch.
230 			 */
231 			if (ret == LRU_REMOVED_RETRY)
232 				goto restart;
233 			break;
234 		case LRU_ROTATE:
235 			list_move_tail(item, &l->list);
236 			break;
237 		case LRU_SKIP:
238 			break;
239 		case LRU_RETRY:
240 			/*
241 			 * The lru lock has been dropped, our list traversal is
242 			 * now invalid and so we have to restart from scratch.
243 			 */
244 			assert_spin_locked(&nlru->lock);
245 			goto restart;
246 		case LRU_STOP:
247 			assert_spin_locked(&nlru->lock);
248 			goto out;
249 		default:
250 			BUG();
251 		}
252 	}
253 out:
254 	return isolated;
255 }
256 
257 unsigned long
258 list_lru_walk_one(struct list_lru *lru, int nid, struct mem_cgroup *memcg,
259 		  list_lru_walk_cb isolate, void *cb_arg,
260 		  unsigned long *nr_to_walk)
261 {
262 	struct list_lru_node *nlru = &lru->node[nid];
263 	unsigned long ret;
264 
265 	spin_lock(&nlru->lock);
266 	ret = __list_lru_walk_one(lru, nid, memcg_kmem_id(memcg), isolate,
267 				  cb_arg, nr_to_walk);
268 	spin_unlock(&nlru->lock);
269 	return ret;
270 }
271 EXPORT_SYMBOL_GPL(list_lru_walk_one);
272 
273 unsigned long
274 list_lru_walk_one_irq(struct list_lru *lru, int nid, struct mem_cgroup *memcg,
275 		      list_lru_walk_cb isolate, void *cb_arg,
276 		      unsigned long *nr_to_walk)
277 {
278 	struct list_lru_node *nlru = &lru->node[nid];
279 	unsigned long ret;
280 
281 	spin_lock_irq(&nlru->lock);
282 	ret = __list_lru_walk_one(lru, nid, memcg_kmem_id(memcg), isolate,
283 				  cb_arg, nr_to_walk);
284 	spin_unlock_irq(&nlru->lock);
285 	return ret;
286 }
287 
288 unsigned long list_lru_walk_node(struct list_lru *lru, int nid,
289 				 list_lru_walk_cb isolate, void *cb_arg,
290 				 unsigned long *nr_to_walk)
291 {
292 	long isolated = 0;
293 
294 	isolated += list_lru_walk_one(lru, nid, NULL, isolate, cb_arg,
295 				      nr_to_walk);
296 
297 #ifdef CONFIG_MEMCG_KMEM
298 	if (*nr_to_walk > 0 && list_lru_memcg_aware(lru)) {
299 		struct list_lru_memcg *mlru;
300 		unsigned long index;
301 
302 		xa_for_each(&lru->xa, index, mlru) {
303 			struct list_lru_node *nlru = &lru->node[nid];
304 
305 			spin_lock(&nlru->lock);
306 			isolated += __list_lru_walk_one(lru, nid, index,
307 							isolate, cb_arg,
308 							nr_to_walk);
309 			spin_unlock(&nlru->lock);
310 
311 			if (*nr_to_walk <= 0)
312 				break;
313 		}
314 	}
315 #endif
316 
317 	return isolated;
318 }
319 EXPORT_SYMBOL_GPL(list_lru_walk_node);
320 
321 static void init_one_lru(struct list_lru_one *l)
322 {
323 	INIT_LIST_HEAD(&l->list);
324 	l->nr_items = 0;
325 }
326 
327 #ifdef CONFIG_MEMCG_KMEM
328 static struct list_lru_memcg *memcg_init_list_lru_one(gfp_t gfp)
329 {
330 	int nid;
331 	struct list_lru_memcg *mlru;
332 
333 	mlru = kmalloc(struct_size(mlru, node, nr_node_ids), gfp);
334 	if (!mlru)
335 		return NULL;
336 
337 	for_each_node(nid)
338 		init_one_lru(&mlru->node[nid]);
339 
340 	return mlru;
341 }
342 
343 static void memcg_list_lru_free(struct list_lru *lru, int src_idx)
344 {
345 	struct list_lru_memcg *mlru = xa_erase_irq(&lru->xa, src_idx);
346 
347 	/*
348 	 * The __list_lru_walk_one() can walk the list of this node.
349 	 * We need kvfree_rcu() here. And the walking of the list
350 	 * is under lru->node[nid]->lock, which can serve as a RCU
351 	 * read-side critical section.
352 	 */
353 	if (mlru)
354 		kvfree_rcu(mlru, rcu);
355 }
356 
357 static inline void memcg_init_list_lru(struct list_lru *lru, bool memcg_aware)
358 {
359 	if (memcg_aware)
360 		xa_init_flags(&lru->xa, XA_FLAGS_LOCK_IRQ);
361 	lru->memcg_aware = memcg_aware;
362 }
363 
364 static void memcg_destroy_list_lru(struct list_lru *lru)
365 {
366 	XA_STATE(xas, &lru->xa, 0);
367 	struct list_lru_memcg *mlru;
368 
369 	if (!list_lru_memcg_aware(lru))
370 		return;
371 
372 	xas_lock_irq(&xas);
373 	xas_for_each(&xas, mlru, ULONG_MAX) {
374 		kfree(mlru);
375 		xas_store(&xas, NULL);
376 	}
377 	xas_unlock_irq(&xas);
378 }
379 
380 static void memcg_reparent_list_lru_node(struct list_lru *lru, int nid,
381 					 int src_idx, struct mem_cgroup *dst_memcg)
382 {
383 	struct list_lru_node *nlru = &lru->node[nid];
384 	int dst_idx = dst_memcg->kmemcg_id;
385 	struct list_lru_one *src, *dst;
386 
387 	/*
388 	 * Since list_lru_{add,del} may be called under an IRQ-safe lock,
389 	 * we have to use IRQ-safe primitives here to avoid deadlock.
390 	 */
391 	spin_lock_irq(&nlru->lock);
392 
393 	src = list_lru_from_memcg_idx(lru, nid, src_idx);
394 	if (!src)
395 		goto out;
396 	dst = list_lru_from_memcg_idx(lru, nid, dst_idx);
397 
398 	list_splice_init(&src->list, &dst->list);
399 
400 	if (src->nr_items) {
401 		dst->nr_items += src->nr_items;
402 		set_shrinker_bit(dst_memcg, nid, lru_shrinker_id(lru));
403 		src->nr_items = 0;
404 	}
405 out:
406 	spin_unlock_irq(&nlru->lock);
407 }
408 
409 static void memcg_reparent_list_lru(struct list_lru *lru,
410 				    int src_idx, struct mem_cgroup *dst_memcg)
411 {
412 	int i;
413 
414 	for_each_node(i)
415 		memcg_reparent_list_lru_node(lru, i, src_idx, dst_memcg);
416 
417 	memcg_list_lru_free(lru, src_idx);
418 }
419 
420 void memcg_reparent_list_lrus(struct mem_cgroup *memcg, struct mem_cgroup *parent)
421 {
422 	struct cgroup_subsys_state *css;
423 	struct list_lru *lru;
424 	int src_idx = memcg->kmemcg_id;
425 
426 	/*
427 	 * Change kmemcg_id of this cgroup and all its descendants to the
428 	 * parent's id, and then move all entries from this cgroup's list_lrus
429 	 * to ones of the parent.
430 	 *
431 	 * After we have finished, all list_lrus corresponding to this cgroup
432 	 * are guaranteed to remain empty. So we can safely free this cgroup's
433 	 * list lrus in memcg_list_lru_free().
434 	 *
435 	 * Changing ->kmemcg_id to the parent can prevent memcg_list_lru_alloc()
436 	 * from allocating list lrus for this cgroup after memcg_list_lru_free()
437 	 * call.
438 	 */
439 	rcu_read_lock();
440 	css_for_each_descendant_pre(css, &memcg->css) {
441 		struct mem_cgroup *child;
442 
443 		child = mem_cgroup_from_css(css);
444 		WRITE_ONCE(child->kmemcg_id, parent->kmemcg_id);
445 	}
446 	rcu_read_unlock();
447 
448 	mutex_lock(&list_lrus_mutex);
449 	list_for_each_entry(lru, &memcg_list_lrus, list)
450 		memcg_reparent_list_lru(lru, src_idx, parent);
451 	mutex_unlock(&list_lrus_mutex);
452 }
453 
454 static inline bool memcg_list_lru_allocated(struct mem_cgroup *memcg,
455 					    struct list_lru *lru)
456 {
457 	int idx = memcg->kmemcg_id;
458 
459 	return idx < 0 || xa_load(&lru->xa, idx);
460 }
461 
462 int memcg_list_lru_alloc(struct mem_cgroup *memcg, struct list_lru *lru,
463 			 gfp_t gfp)
464 {
465 	int i;
466 	unsigned long flags;
467 	struct list_lru_memcg_table {
468 		struct list_lru_memcg *mlru;
469 		struct mem_cgroup *memcg;
470 	} *table;
471 	XA_STATE(xas, &lru->xa, 0);
472 
473 	if (!list_lru_memcg_aware(lru) || memcg_list_lru_allocated(memcg, lru))
474 		return 0;
475 
476 	gfp &= GFP_RECLAIM_MASK;
477 	table = kmalloc_array(memcg->css.cgroup->level, sizeof(*table), gfp);
478 	if (!table)
479 		return -ENOMEM;
480 
481 	/*
482 	 * Because the list_lru can be reparented to the parent cgroup's
483 	 * list_lru, we should make sure that this cgroup and all its
484 	 * ancestors have allocated list_lru_memcg.
485 	 */
486 	for (i = 0; memcg; memcg = parent_mem_cgroup(memcg), i++) {
487 		if (memcg_list_lru_allocated(memcg, lru))
488 			break;
489 
490 		table[i].memcg = memcg;
491 		table[i].mlru = memcg_init_list_lru_one(gfp);
492 		if (!table[i].mlru) {
493 			while (i--)
494 				kfree(table[i].mlru);
495 			kfree(table);
496 			return -ENOMEM;
497 		}
498 	}
499 
500 	xas_lock_irqsave(&xas, flags);
501 	while (i--) {
502 		int index = READ_ONCE(table[i].memcg->kmemcg_id);
503 		struct list_lru_memcg *mlru = table[i].mlru;
504 
505 		xas_set(&xas, index);
506 retry:
507 		if (unlikely(index < 0 || xas_error(&xas) || xas_load(&xas))) {
508 			kfree(mlru);
509 		} else {
510 			xas_store(&xas, mlru);
511 			if (xas_error(&xas) == -ENOMEM) {
512 				xas_unlock_irqrestore(&xas, flags);
513 				if (xas_nomem(&xas, gfp))
514 					xas_set_err(&xas, 0);
515 				xas_lock_irqsave(&xas, flags);
516 				/*
517 				 * The xas lock has been released, this memcg
518 				 * can be reparented before us. So reload
519 				 * memcg id. More details see the comments
520 				 * in memcg_reparent_list_lrus().
521 				 */
522 				index = READ_ONCE(table[i].memcg->kmemcg_id);
523 				if (index < 0)
524 					xas_set_err(&xas, 0);
525 				else if (!xas_error(&xas) && index != xas.xa_index)
526 					xas_set(&xas, index);
527 				goto retry;
528 			}
529 		}
530 	}
531 	/* xas_nomem() is used to free memory instead of memory allocation. */
532 	if (xas.xa_alloc)
533 		xas_nomem(&xas, gfp);
534 	xas_unlock_irqrestore(&xas, flags);
535 	kfree(table);
536 
537 	return xas_error(&xas);
538 }
539 #else
540 static inline void memcg_init_list_lru(struct list_lru *lru, bool memcg_aware)
541 {
542 }
543 
544 static void memcg_destroy_list_lru(struct list_lru *lru)
545 {
546 }
547 #endif /* CONFIG_MEMCG_KMEM */
548 
549 int __list_lru_init(struct list_lru *lru, bool memcg_aware,
550 		    struct lock_class_key *key, struct shrinker *shrinker)
551 {
552 	int i;
553 
554 #ifdef CONFIG_MEMCG_KMEM
555 	if (shrinker)
556 		lru->shrinker_id = shrinker->id;
557 	else
558 		lru->shrinker_id = -1;
559 
560 	if (mem_cgroup_kmem_disabled())
561 		memcg_aware = false;
562 #endif
563 
564 	lru->node = kcalloc(nr_node_ids, sizeof(*lru->node), GFP_KERNEL);
565 	if (!lru->node)
566 		return -ENOMEM;
567 
568 	for_each_node(i) {
569 		spin_lock_init(&lru->node[i].lock);
570 		if (key)
571 			lockdep_set_class(&lru->node[i].lock, key);
572 		init_one_lru(&lru->node[i].lru);
573 	}
574 
575 	memcg_init_list_lru(lru, memcg_aware);
576 	list_lru_register(lru);
577 
578 	return 0;
579 }
580 EXPORT_SYMBOL_GPL(__list_lru_init);
581 
582 void list_lru_destroy(struct list_lru *lru)
583 {
584 	/* Already destroyed or not yet initialized? */
585 	if (!lru->node)
586 		return;
587 
588 	list_lru_unregister(lru);
589 
590 	memcg_destroy_list_lru(lru);
591 	kfree(lru->node);
592 	lru->node = NULL;
593 
594 #ifdef CONFIG_MEMCG_KMEM
595 	lru->shrinker_id = -1;
596 #endif
597 }
598 EXPORT_SYMBOL_GPL(list_lru_destroy);
599