summaryrefslogtreecommitdiff
path: root/kernel/bpf/memalloc.c
blob: 550f02e2cb1327e1d082a895f3be16f22bf06f57 (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
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
// SPDX-License-Identifier: GPL-2.0-only
/* Copyright (c) 2022 Meta Platforms, Inc. and affiliates. */
#include <linux/mm.h>
#include <linux/llist.h>
#include <linux/bpf.h>
#include <linux/irq_work.h>
#include <linux/bpf_mem_alloc.h>
#include <linux/memcontrol.h>
#include <asm/local.h>

/* Any context (including NMI) BPF specific memory allocator.
 *
 * Tracing BPF programs can attach to kprobe and fentry. Hence they
 * run in unknown context where calling plain kmalloc() might not be safe.
 *
 * Front-end kmalloc() with per-cpu per-bucket cache of free elements.
 * Refill this cache asynchronously from irq_work.
 *
 * CPU_0 buckets
 * 16 32 64 96 128 196 256 512 1024 2048 4096
 * ...
 * CPU_N buckets
 * 16 32 64 96 128 196 256 512 1024 2048 4096
 *
 * The buckets are prefilled at the start.
 * BPF programs always run with migration disabled.
 * It's safe to allocate from cache of the current cpu with irqs disabled.
 * Free-ing is always done into bucket of the current cpu as well.
 * irq_work trims extra free elements from buckets with kfree
 * and refills them with kmalloc, so global kmalloc logic takes care
 * of freeing objects allocated by one cpu and freed on another.
 *
 * Every allocated objected is padded with extra 8 bytes that contains
 * struct llist_node.
 */
#define LLIST_NODE_SZ sizeof(struct llist_node)

/* similar to kmalloc, but sizeof == 8 bucket is gone */
static u8 size_index[24] __ro_after_init = {
	3,	/* 8 */
	3,	/* 16 */
	4,	/* 24 */
	4,	/* 32 */
	5,	/* 40 */
	5,	/* 48 */
	5,	/* 56 */
	5,	/* 64 */
	1,	/* 72 */
	1,	/* 80 */
	1,	/* 88 */
	1,	/* 96 */
	6,	/* 104 */
	6,	/* 112 */
	6,	/* 120 */
	6,	/* 128 */
	2,	/* 136 */
	2,	/* 144 */
	2,	/* 152 */
	2,	/* 160 */
	2,	/* 168 */
	2,	/* 176 */
	2,	/* 184 */
	2	/* 192 */
};

static int bpf_mem_cache_idx(size_t size)
{
	if (!size || size > 4096)
		return -1;

	if (size <= 192)
		return size_index[(size - 1) / 8] - 1;

	return fls(size - 1) - 2;
}

#define NUM_CACHES 11

struct bpf_mem_cache {
	/* per-cpu list of free objects of size 'unit_size'.
	 * All accesses are done with interrupts disabled and 'active' counter
	 * protection with __llist_add() and __llist_del_first().
	 */
	struct llist_head free_llist;
	local_t active;

	/* Operations on the free_list from unit_alloc/unit_free/bpf_mem_refill
	 * are sequenced by per-cpu 'active' counter. But unit_free() cannot
	 * fail. When 'active' is busy the unit_free() will add an object to
	 * free_llist_extra.
	 */
	struct llist_head free_llist_extra;

	struct irq_work refill_work;
	struct obj_cgroup *objcg;
	int unit_size;
	/* count of objects in free_llist */
	int free_cnt;
	int low_watermark, high_watermark, batch;
	int percpu_size;
	bool draining;
	struct bpf_mem_cache *tgt;

	/* list of objects to be freed after RCU GP */
	struct llist_head free_by_rcu;
	struct llist_node *free_by_rcu_tail;
	struct llist_head waiting_for_gp;
	struct llist_node *waiting_for_gp_tail;
	struct rcu_head rcu;
	atomic_t call_rcu_in_progress;
	struct llist_head free_llist_extra_rcu;

	/* list of objects to be freed after RCU tasks trace GP */
	struct llist_head free_by_rcu_ttrace;
	struct llist_head waiting_for_gp_ttrace;
	struct rcu_head rcu_ttrace;
	atomic_t call_rcu_ttrace_in_progress;
};

struct bpf_mem_caches {
	struct bpf_mem_cache cache[NUM_CACHES];
};

static const u16 sizes[NUM_CACHES] = {96, 192, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096};

static struct llist_node notrace *__llist_del_first(struct llist_head *head)
{
	struct llist_node *entry, *next;

	entry = head->first;
	if (!entry)
		return NULL;
	next = entry->next;
	head->first = next;
	return entry;
}

static void *__alloc(struct bpf_mem_cache *c, int node, gfp_t flags)
{
	if (c->percpu_size) {
		void **obj = kmalloc_node(c->percpu_size, flags, node);
		void *pptr = __alloc_percpu_gfp(c->unit_size, 8, flags);

		if (!obj || !pptr) {
			free_percpu(pptr);
			kfree(obj);
			return NULL;
		}
		obj[1] = pptr;
		return obj;
	}

	return kmalloc_node(c->unit_size, flags | __GFP_ZERO, node);
}

static struct mem_cgroup *get_memcg(const struct bpf_mem_cache *c)
{
#ifdef CONFIG_MEMCG_KMEM
	if (c->objcg)
		return get_mem_cgroup_from_objcg(c->objcg);
#endif

#ifdef CONFIG_MEMCG
	return root_mem_cgroup;
#else
	return NULL;
#endif
}

static void inc_active(struct bpf_mem_cache *c, unsigned long *flags)
{
	if (IS_ENABLED(CONFIG_PREEMPT_RT))
		/* In RT irq_work runs in per-cpu kthread, so disable
		 * interrupts to avoid preemption and interrupts and
		 * reduce the chance of bpf prog executing on this cpu
		 * when active counter is busy.
		 */
		local_irq_save(*flags);
	/* alloc_bulk runs from irq_work which will not preempt a bpf
	 * program that does unit_alloc/unit_free since IRQs are
	 * disabled there. There is no race to increment 'active'
	 * counter. It protects free_llist from corruption in case NMI
	 * bpf prog preempted this loop.
	 */
	WARN_ON_ONCE(local_inc_return(&c->active) != 1);
}

static void dec_active(struct bpf_mem_cache *c, unsigned long *flags)
{
	local_dec(&c->active);
	if (IS_ENABLED(CONFIG_PREEMPT_RT))
		local_irq_restore(*flags);
}

static void add_obj_to_free_list(struct bpf_mem_cache *c, void *obj)
{
	unsigned long flags;

	inc_active(c, &flags);
	__llist_add(obj, &c->free_llist);
	c->free_cnt++;
	dec_active(c, &flags);
}

/* Mostly runs from irq_work except __init phase. */
static void alloc_bulk(struct bpf_mem_cache *c, int cnt, int node, bool atomic)
{
	struct mem_cgroup *memcg = NULL, *old_memcg;
	gfp_t gfp;
	void *obj;
	int i;

	gfp = __GFP_NOWARN | __GFP_ACCOUNT;
	gfp |= atomic ? GFP_NOWAIT : GFP_KERNEL;

	for (i = 0; i < cnt; i++) {
		/*
		 * For every 'c' llist_del_first(&c->free_by_rcu_ttrace); is
		 * done only by one CPU == current CPU. Other CPUs might
		 * llist_add() and llist_del_all() in parallel.
		 */
		obj = llist_del_first(&c->free_by_rcu_ttrace);
		if (!obj)
			break;
		add_obj_to_free_list(c, obj);
	}
	if (i >= cnt)
		return;

	for (; i < cnt; i++) {
		obj = llist_del_first(&c->waiting_for_gp_ttrace);
		if (!obj)
			break;
		add_obj_to_free_list(c, obj);
	}
	if (i >= cnt)
		return;

	memcg = get_memcg(c);
	old_memcg = set_active_memcg(memcg);
	for (; i < cnt; i++) {
		/* Allocate, but don't deplete atomic reserves that typical
		 * GFP_ATOMIC would do. irq_work runs on this cpu and kmalloc
		 * will allocate from the current numa node which is what we
		 * want here.
		 */
		obj = __alloc(c, node, gfp);
		if (!obj)
			break;
		add_obj_to_free_list(c, obj);
	}
	set_active_memcg(old_memcg);
	mem_cgroup_put(memcg);
}

static void free_one(void *obj, bool percpu)
{
	if (percpu) {
		free_percpu(((void **)obj)[1]);
		kfree(obj);
		return;
	}

	kfree(obj);
}

static int free_all(struct llist_node *llnode, bool percpu)
{
	struct llist_node *pos, *t;
	int cnt = 0;

	llist_for_each_safe(pos, t, llnode) {
		free_one(pos, percpu);
		cnt++;
	}
	return cnt;
}

static void __free_rcu(struct rcu_head *head)
{
	struct bpf_mem_cache *c = container_of(head, struct bpf_mem_cache, rcu_ttrace);

	free_all(llist_del_all(&c->waiting_for_gp_ttrace), !!c->percpu_size);
	atomic_set(&c->call_rcu_ttrace_in_progress, 0);
}

static void __free_rcu_tasks_trace(struct rcu_head *head)
{
	/* If RCU Tasks Trace grace period implies RCU grace period,
	 * there is no need to invoke call_rcu().
	 */
	if (rcu_trace_implies_rcu_gp())
		__free_rcu(head);
	else
		call_rcu(head, __free_rcu);
}

static void enque_to_free(struct bpf_mem_cache *c, void *obj)
{
	struct llist_node *llnode = obj;

	/* bpf_mem_cache is a per-cpu object. Freeing happens in irq_work.
	 * Nothing races to add to free_by_rcu_ttrace list.
	 */
	llist_add(llnode, &c->free_by_rcu_ttrace);
}

static void do_call_rcu_ttrace(struct bpf_mem_cache *c)
{
	struct llist_node *llnode, *t;

	if (atomic_xchg(&c->call_rcu_ttrace_in_progress, 1)) {
		if (unlikely(READ_ONCE(c->draining))) {
			llnode = llist_del_all(&c->free_by_rcu_ttrace);
			free_all(llnode, !!c->percpu_size);
		}
		return;
	}

	WARN_ON_ONCE(!llist_empty(&c->waiting_for_gp_ttrace));
	llist_for_each_safe(llnode, t, llist_del_all(&c->free_by_rcu_ttrace))
		llist_add(llnode, &c->waiting_for_gp_ttrace);

	if (unlikely(READ_ONCE(c->draining))) {
		__free_rcu(&c->rcu_ttrace);
		return;
	}

	/* Use call_rcu_tasks_trace() to wait for sleepable progs to finish.
	 * If RCU Tasks Trace grace period implies RCU grace period, free
	 * these elements directly, else use call_rcu() to wait for normal
	 * progs to finish and finally do free_one() on each element.
	 */
	call_rcu_tasks_trace(&c->rcu_ttrace, __free_rcu_tasks_trace);
}

static void free_bulk(struct bpf_mem_cache *c)
{
	struct bpf_mem_cache *tgt = c->tgt;
	struct llist_node *llnode, *t;
	unsigned long flags;
	int cnt;

	WARN_ON_ONCE(tgt->unit_size != c->unit_size);
	WARN_ON_ONCE(tgt->percpu_size != c->percpu_size);

	do {
		inc_active(c, &flags);
		llnode = __llist_del_first(&c->free_llist);
		if (llnode)
			cnt = --c->free_cnt;
		else
			cnt = 0;
		dec_active(c, &flags);
		if (llnode)
			enque_to_free(tgt, llnode);
	} while (cnt > (c->high_watermark + c->low_watermark) / 2);

	/* and drain free_llist_extra */
	llist_for_each_safe(llnode, t, llist_del_all(&c->free_llist_extra))
		enque_to_free(tgt, llnode);
	do_call_rcu_ttrace(tgt);
}

static void __free_by_rcu(struct rcu_head *head)
{
	struct bpf_mem_cache *c = container_of(head, struct bpf_mem_cache, rcu);
	struct bpf_mem_cache *tgt = c->tgt;
	struct llist_node *llnode;

	WARN_ON_ONCE(tgt->unit_size != c->unit_size);
	WARN_ON_ONCE(tgt->percpu_size != c->percpu_size);

	llnode = llist_del_all(&c->waiting_for_gp);
	if (!llnode)
		goto out;

	llist_add_batch(llnode, c->waiting_for_gp_tail, &tgt->free_by_rcu_ttrace);

	/* Objects went through regular RCU GP. Send them to RCU tasks trace */
	do_call_rcu_ttrace(tgt);
out:
	atomic_set(&c->call_rcu_in_progress, 0);
}

static void check_free_by_rcu(struct bpf_mem_cache *c)
{
	struct llist_node *llnode, *t;
	unsigned long flags;

	/* drain free_llist_extra_rcu */
	if (unlikely(!llist_empty(&c->free_llist_extra_rcu))) {
		inc_active(c, &flags);
		llist_for_each_safe(llnode, t, llist_del_all(&c->free_llist_extra_rcu))
			if (__llist_add(llnode, &c->free_by_rcu))
				c->free_by_rcu_tail = llnode;
		dec_active(c, &flags);
	}

	if (llist_empty(&c->free_by_rcu))
		return;

	if (atomic_xchg(&c->call_rcu_in_progress, 1)) {
		/*
		 * Instead of kmalloc-ing new rcu_head and triggering 10k
		 * call_rcu() to hit rcutree.qhimark and force RCU to notice
		 * the overload just ask RCU to hurry up. There could be many
		 * objects in free_by_rcu list.
		 * This hint reduces memory consumption for an artificial
		 * benchmark from 2 Gbyte to 150 Mbyte.
		 */
		rcu_request_urgent_qs_task(current);
		return;
	}

	WARN_ON_ONCE(!llist_empty(&c->waiting_for_gp));

	inc_active(c, &flags);
	WRITE_ONCE(c->waiting_for_gp.first, __llist_del_all(&c->free_by_rcu));
	c->waiting_for_gp_tail = c->free_by_rcu_tail;
	dec_active(c, &flags);

	if (unlikely(READ_ONCE(c->draining))) {
		free_all(llist_del_all(&c->waiting_for_gp), !!c->percpu_size);
		atomic_set(&c->call_rcu_in_progress, 0);
	} else {
		call_rcu_hurry(&c->rcu, __free_by_rcu);
	}
}

static void bpf_mem_refill(struct irq_work *work)
{
	struct bpf_mem_cache *c = container_of(work, struct bpf_mem_cache, refill_work);
	int cnt;

	/* Racy access to free_cnt. It doesn't need to be 100% accurate */
	cnt = c->free_cnt;
	if (cnt < c->low_watermark)
		/* irq_work runs on this cpu and kmalloc will allocate
		 * from the current numa node which is what we want here.
		 */
		alloc_bulk(c, c->batch, NUMA_NO_NODE, true);
	else if (cnt > c->high_watermark)
		free_bulk(c);

	check_free_by_rcu(c);
}

static void notrace irq_work_raise(struct bpf_mem_cache *c)
{
	irq_work_queue(&c->refill_work);
}

/* For typical bpf map case that uses bpf_mem_cache_alloc and single bucket
 * the freelist cache will be elem_size * 64 (or less) on each cpu.
 *
 * For bpf programs that don't have statically known allocation sizes and
 * assuming (low_mark + high_mark) / 2 as an average number of elements per
 * bucket and all buckets are used the total amount of memory in freelists
 * on each cpu will be:
 * 64*16 + 64*32 + 64*64 + 64*96 + 64*128 + 64*196 + 64*256 + 32*512 + 16*1024 + 8*2048 + 4*4096
 * == ~ 116 Kbyte using below heuristic.
 * Initialized, but unused bpf allocator (not bpf map specific one) will
 * consume ~ 11 Kbyte per cpu.
 * Typical case will be between 11K and 116K closer to 11K.
 * bpf progs can and should share bpf_mem_cache when possible.
 *
 * Percpu allocation is typically rare. To avoid potential unnecessary large
 * memory consumption, set low_mark = 1 and high_mark = 3, resulting in c->batch = 1.
 */
static void init_refill_work(struct bpf_mem_cache *c)
{
	init_irq_work(&c->refill_work, bpf_mem_refill);
	if (c->percpu_size) {
		c->low_watermark = 1;
		c->high_watermark = 3;
	} else if (c->unit_size <= 256) {
		c->low_watermark = 32;
		c->high_watermark = 96;
	} else {
		/* When page_size == 4k, order-0 cache will have low_mark == 2
		 * and high_mark == 6 with batch alloc of 3 individual pages at
		 * a time.
		 * 8k allocs and above low == 1, high == 3, batch == 1.
		 */
		c->low_watermark = max(32 * 256 / c->unit_size, 1);
		c->high_watermark = max(96 * 256 / c->unit_size, 3);
	}
	c->batch = max((c->high_watermark - c->low_watermark) / 4 * 3, 1);
}

static void prefill_mem_cache(struct bpf_mem_cache *c, int cpu)
{
	int cnt = 1;

	/* To avoid consuming memory, for non-percpu allocation, assume that
	 * 1st run of bpf prog won't be doing more than 4 map_update_elem from
	 * irq disabled region if unit size is less than or equal to 256.
	 * For all other cases, let us just do one allocation.
	 */
	if (!c->percpu_size && c->unit_size <= 256)
		cnt = 4;
	alloc_bulk(c, cnt, cpu_to_node(cpu), false);
}

/* When size != 0 bpf_mem_cache for each cpu.
 * This is typical bpf hash map use case when all elements have equal size.
 *
 * When size == 0 allocate 11 bpf_mem_cache-s for each cpu, then rely on
 * kmalloc/kfree. Max allocation size is 4096 in this case.
 * This is bpf_dynptr and bpf_kptr use case.
 */
int bpf_mem_alloc_init(struct bpf_mem_alloc *ma, int size, bool percpu)
{
	struct bpf_mem_caches *cc, __percpu *pcc;
	struct bpf_mem_cache *c, __percpu *pc;
	struct obj_cgroup *objcg = NULL;
	int cpu, i, unit_size, percpu_size = 0;

	if (percpu && size == 0)
		return -EINVAL;

	/* room for llist_node and per-cpu pointer */
	if (percpu)
		percpu_size = LLIST_NODE_SZ + sizeof(void *);
	ma->percpu = percpu;

	if (size) {
		pc = __alloc_percpu_gfp(sizeof(*pc), 8, GFP_KERNEL);
		if (!pc)
			return -ENOMEM;

		if (!percpu)
			size += LLIST_NODE_SZ; /* room for llist_node */
		unit_size = size;

#ifdef CONFIG_MEMCG_KMEM
		if (memcg_bpf_enabled())
			objcg = get_obj_cgroup_from_current();
#endif
		ma->objcg = objcg;

		for_each_possible_cpu(cpu) {
			c = per_cpu_ptr(pc, cpu);
			c->unit_size = unit_size;
			c->objcg = objcg;
			c->percpu_size = percpu_size;
			c->tgt = c;
			init_refill_work(c);
			prefill_mem_cache(c, cpu);
		}
		ma->cache = pc;
		return 0;
	}

	pcc = __alloc_percpu_gfp(sizeof(*cc), 8, GFP_KERNEL);
	if (!pcc)
		return -ENOMEM;
#ifdef CONFIG_MEMCG_KMEM
	objcg = get_obj_cgroup_from_current();
#endif
	ma->objcg = objcg;
	for_each_possible_cpu(cpu) {
		cc = per_cpu_ptr(pcc, cpu);
		for (i = 0; i < NUM_CACHES; i++) {
			c = &cc->cache[i];
			c->unit_size = sizes[i];
			c->objcg = objcg;
			c->percpu_size = percpu_size;
			c->tgt = c;

			init_refill_work(c);
			prefill_mem_cache(c, cpu);
		}
	}

	ma->caches = pcc;
	return 0;
}

int bpf_mem_alloc_percpu_init(struct bpf_mem_alloc *ma, struct obj_cgroup *objcg)
{
	struct bpf_mem_caches __percpu *pcc;

	pcc = __alloc_percpu_gfp(sizeof(struct bpf_mem_caches), 8, GFP_KERNEL);
	if (!pcc)
		return -ENOMEM;

	ma->caches = pcc;
	ma->objcg = objcg;
	ma->percpu = true;
	return 0;
}

int bpf_mem_alloc_percpu_unit_init(struct bpf_mem_alloc *ma, int size)
{
	struct bpf_mem_caches *cc, __percpu *pcc;
	int cpu, i, unit_size, percpu_size;
	struct obj_cgroup *objcg;
	struct bpf_mem_cache *c;

	i = bpf_mem_cache_idx(size);
	if (i < 0)
		return -EINVAL;

	/* room for llist_node and per-cpu pointer */
	percpu_size = LLIST_NODE_SZ + sizeof(void *);

	unit_size = sizes[i];
	objcg = ma->objcg;
	pcc = ma->caches;

	for_each_possible_cpu(cpu) {
		cc = per_cpu_ptr(pcc, cpu);
		c = &cc->cache[i];
		if (c->unit_size)
			break;

		c->unit_size = unit_size;
		c->objcg = objcg;
		c->percpu_size = percpu_size;
		c->tgt = c;

		init_refill_work(c);
		prefill_mem_cache(c, cpu);
	}

	return 0;
}

static void drain_mem_cache(struct bpf_mem_cache *c)
{
	bool percpu = !!c->percpu_size;

	/* No progs are using this bpf_mem_cache, but htab_map_free() called
	 * bpf_mem_cache_free() for all remaining elements and they can be in
	 * free_by_rcu_ttrace or in waiting_for_gp_ttrace lists, so drain those lists now.
	 *
	 * Except for waiting_for_gp_ttrace list, there are no concurrent operations
	 * on these lists, so it is safe to use __llist_del_all().
	 */
	free_all(llist_del_all(&c->free_by_rcu_ttrace), percpu);
	free_all(llist_del_all(&c->waiting_for_gp_ttrace), percpu);
	free_all(__llist_del_all(&c->free_llist), percpu);
	free_all(__llist_del_all(&c->free_llist_extra), percpu);
	free_all(__llist_del_all(&c->free_by_rcu), percpu);
	free_all(__llist_del_all(&c->free_llist_extra_rcu), percpu);
	free_all(llist_del_all(&c->waiting_for_gp), percpu);
}

static void check_mem_cache(struct bpf_mem_cache *c)
{
	WARN_ON_ONCE(!llist_empty(&c->free_by_rcu_ttrace));
	WARN_ON_ONCE(!llist_empty(&c->waiting_for_gp_ttrace));
	WARN_ON_ONCE(!llist_empty(&c->free_llist));
	WARN_ON_ONCE(!llist_empty(&c->free_llist_extra));
	WARN_ON_ONCE(!llist_empty(&c->free_by_rcu));
	WARN_ON_ONCE(!llist_empty(&c->free_llist_extra_rcu));
	WARN_ON_ONCE(!llist_empty(&c->waiting_for_gp));
}

static void check_leaked_objs(struct bpf_mem_alloc *ma)
{
	struct bpf_mem_caches *cc;
	struct bpf_mem_cache *c;
	int cpu, i;

	if (ma->cache) {
		for_each_possible_cpu(cpu) {
			c = per_cpu_ptr(ma->cache, cpu);
			check_mem_cache(c);
		}
	}
	if (ma->caches) {
		for_each_possible_cpu(cpu) {
			cc = per_cpu_ptr(ma->caches, cpu);
			for (i = 0; i < NUM_CACHES; i++) {
				c = &cc->cache[i];
				check_mem_cache(c);
			}
		}
	}
}

static void free_mem_alloc_no_barrier(struct bpf_mem_alloc *ma)
{
	check_leaked_objs(ma);
	free_percpu(ma->cache);
	free_percpu(ma->caches);
	ma->cache = NULL;
	ma->caches = NULL;
}

static void free_mem_alloc(struct bpf_mem_alloc *ma)
{
	/* waiting_for_gp[_ttrace] lists were drained, but RCU callbacks
	 * might still execute. Wait for them.
	 *
	 * rcu_barrier_tasks_trace() doesn't imply synchronize_rcu_tasks_trace(),
	 * but rcu_barrier_tasks_trace() and rcu_barrier() below are only used
	 * to wait for the pending __free_rcu_tasks_trace() and __free_rcu(),
	 * so if call_rcu(head, __free_rcu) is skipped due to
	 * rcu_trace_implies_rcu_gp(), it will be OK to skip rcu_barrier() by
	 * using rcu_trace_implies_rcu_gp() as well.
	 */
	rcu_barrier(); /* wait for __free_by_rcu */
	rcu_barrier_tasks_trace(); /* wait for __free_rcu */
	if (!rcu_trace_implies_rcu_gp())
		rcu_barrier();
	free_mem_alloc_no_barrier(ma);
}

static void free_mem_alloc_deferred(struct work_struct *work)
{
	struct bpf_mem_alloc *ma = container_of(work, struct bpf_mem_alloc, work);

	free_mem_alloc(ma);
	kfree(ma);
}

static void destroy_mem_alloc(struct bpf_mem_alloc *ma, int rcu_in_progress)
{
	struct bpf_mem_alloc *copy;

	if (!rcu_in_progress) {
		/* Fast path. No callbacks are pending, hence no need to do
		 * rcu_barrier-s.
		 */
		free_mem_alloc_no_barrier(ma);
		return;
	}

	copy = kmemdup(ma, sizeof(*ma), GFP_KERNEL);
	if (!copy) {
		/* Slow path with inline barrier-s */
		free_mem_alloc(ma);
		return;
	}

	/* Defer barriers into worker to let the rest of map memory to be freed */
	memset(ma, 0, sizeof(*ma));
	INIT_WORK(&copy->work, free_mem_alloc_deferred);
	queue_work(system_unbound_wq, &copy->work);
}

void bpf_mem_alloc_destroy(struct bpf_mem_alloc *ma)
{
	struct bpf_mem_caches *cc;
	struct bpf_mem_cache *c;
	int cpu, i, rcu_in_progress;

	if (ma->cache) {
		rcu_in_progress = 0;
		for_each_possible_cpu(cpu) {
			c = per_cpu_ptr(ma->cache, cpu);
			WRITE_ONCE(c->draining, true);
			irq_work_sync(&c->refill_work);
			drain_mem_cache(c);
			rcu_in_progress += atomic_read(&c->call_rcu_ttrace_in_progress);
			rcu_in_progress += atomic_read(&c->call_rcu_in_progress);
		}
		if (ma->objcg)
			obj_cgroup_put(ma->objcg);
		destroy_mem_alloc(ma, rcu_in_progress);
	}
	if (ma->caches) {
		rcu_in_progress = 0;
		for_each_possible_cpu(cpu) {
			cc = per_cpu_ptr(ma->caches, cpu);
			for (i = 0; i < NUM_CACHES; i++) {
				c = &cc->cache[i];
				WRITE_ONCE(c->draining, true);
				irq_work_sync(&c->refill_work);
				drain_mem_cache(c);
				rcu_in_progress += atomic_read(&c->call_rcu_ttrace_in_progress);
				rcu_in_progress += atomic_read(&c->call_rcu_in_progress);
			}
		}
		if (ma->objcg)
			obj_cgroup_put(ma->objcg);
		destroy_mem_alloc(ma, rcu_in_progress);
	}
}

/* notrace is necessary here and in other functions to make sure
 * bpf programs cannot attach to them and cause llist corruptions.
 */
static void notrace *unit_alloc(struct bpf_mem_cache *c)
{
	struct llist_node *llnode = NULL;
	unsigned long flags;
	int cnt = 0;

	/* Disable irqs to prevent the following race for majority of prog types:
	 * prog_A
	 *   bpf_mem_alloc
	 *      preemption or irq -> prog_B
	 *        bpf_mem_alloc
	 *
	 * but prog_B could be a perf_event NMI prog.
	 * Use per-cpu 'active' counter to order free_list access between
	 * unit_alloc/unit_free/bpf_mem_refill.
	 */
	local_irq_save(flags);
	if (local_inc_return(&c->active) == 1) {
		llnode = __llist_del_first(&c->free_llist);
		if (llnode) {
			cnt = --c->free_cnt;
			*(struct bpf_mem_cache **)llnode = c;
		}
	}
	local_dec(&c->active);

	WARN_ON(cnt < 0);

	if (cnt < c->low_watermark)
		irq_work_raise(c);
	/* Enable IRQ after the enqueue of irq work completes, so irq work
	 * will run after IRQ is enabled and free_llist may be refilled by
	 * irq work before other task preempts current task.
	 */
	local_irq_restore(flags);

	return llnode;
}

/* Though 'ptr' object could have been allocated on a different cpu
 * add it to the free_llist of the current cpu.
 * Let kfree() logic deal with it when it's later called from irq_work.
 */
static void notrace unit_free(struct bpf_mem_cache *c, void *ptr)
{
	struct llist_node *llnode = ptr - LLIST_NODE_SZ;
	unsigned long flags;
	int cnt = 0;

	BUILD_BUG_ON(LLIST_NODE_SZ > 8);

	/*
	 * Remember bpf_mem_cache that allocated this object.
	 * The hint is not accurate.
	 */
	c->tgt = *(struct bpf_mem_cache **)llnode;

	local_irq_save(flags);
	if (local_inc_return(&c->active) == 1) {
		__llist_add(llnode, &c->free_llist);
		cnt = ++c->free_cnt;
	} else {
		/* unit_free() cannot fail. Therefore add an object to atomic
		 * llist. free_bulk() will drain it. Though free_llist_extra is
		 * a per-cpu list we have to use atomic llist_add here, since
		 * it also can be interrupted by bpf nmi prog that does another
		 * unit_free() into the same free_llist_extra.
		 */
		llist_add(llnode, &c->free_llist_extra);
	}
	local_dec(&c->active);

	if (cnt > c->high_watermark)
		/* free few objects from current cpu into global kmalloc pool */
		irq_work_raise(c);
	/* Enable IRQ after irq_work_raise() completes, otherwise when current
	 * task is preempted by task which does unit_alloc(), unit_alloc() may
	 * return NULL unexpectedly because irq work is already pending but can
	 * not been triggered and free_llist can not be refilled timely.
	 */
	local_irq_restore(flags);
}

static void notrace unit_free_rcu(struct bpf_mem_cache *c, void *ptr)
{
	struct llist_node *llnode = ptr - LLIST_NODE_SZ;
	unsigned long flags;

	c->tgt = *(struct bpf_mem_cache **)llnode;

	local_irq_save(flags);
	if (local_inc_return(&c->active) == 1) {
		if (__llist_add(llnode, &c->free_by_rcu))
			c->free_by_rcu_tail = llnode;
	} else {
		llist_add(llnode, &c->free_llist_extra_rcu);
	}
	local_dec(&c->active);

	if (!atomic_read(&c->call_rcu_in_progress))
		irq_work_raise(c);
	local_irq_restore(flags);
}

/* Called from BPF program or from sys_bpf syscall.
 * In both cases migration is disabled.
 */
void notrace *bpf_mem_alloc(struct bpf_mem_alloc *ma, size_t size)
{
	int idx;
	void *ret;

	if (!size)
		return NULL;

	if (!ma->percpu)
		size += LLIST_NODE_SZ;
	idx = bpf_mem_cache_idx(size);
	if (idx < 0)
		return NULL;

	ret = unit_alloc(this_cpu_ptr(ma->caches)->cache + idx);
	return !ret ? NULL : ret + LLIST_NODE_SZ;
}

void notrace bpf_mem_free(struct bpf_mem_alloc *ma, void *ptr)
{
	struct bpf_mem_cache *c;
	int idx;

	if (!ptr)
		return;

	c = *(void **)(ptr - LLIST_NODE_SZ);
	idx = bpf_mem_cache_idx(c->unit_size);
	if (WARN_ON_ONCE(idx < 0))
		return;

	unit_free(this_cpu_ptr(ma->caches)->cache + idx, ptr);
}

void notrace bpf_mem_free_rcu(struct bpf_mem_alloc *ma, void *ptr)
{
	struct bpf_mem_cache *c;
	int idx;

	if (!ptr)
		return;

	c = *(void **)(ptr - LLIST_NODE_SZ);
	idx = bpf_mem_cache_idx(c->unit_size);
	if (WARN_ON_ONCE(idx < 0))
		return;

	unit_free_rcu(this_cpu_ptr(ma->caches)->cache + idx, ptr);
}

void notrace *bpf_mem_cache_alloc(struct bpf_mem_alloc *ma)
{
	void *ret;

	ret = unit_alloc(this_cpu_ptr(ma->cache));
	return !ret ? NULL : ret + LLIST_NODE_SZ;
}

void notrace bpf_mem_cache_free(struct bpf_mem_alloc *ma, void *ptr)
{
	if (!ptr)
		return;

	unit_free(this_cpu_ptr(ma->cache), ptr);
}

void notrace bpf_mem_cache_free_rcu(struct bpf_mem_alloc *ma, void *ptr)
{
	if (!ptr)
		return;

	unit_free_rcu(this_cpu_ptr(ma->cache), ptr);
}

/* Directly does a kfree() without putting 'ptr' back to the free_llist
 * for reuse and without waiting for a rcu_tasks_trace gp.
 * The caller must first go through the rcu_tasks_trace gp for 'ptr'
 * before calling bpf_mem_cache_raw_free().
 * It could be used when the rcu_tasks_trace callback does not have
 * a hold on the original bpf_mem_alloc object that allocated the
 * 'ptr'. This should only be used in the uncommon code path.
 * Otherwise, the bpf_mem_alloc's free_llist cannot be refilled
 * and may affect performance.
 */
void bpf_mem_cache_raw_free(void *ptr)
{
	if (!ptr)
		return;

	kfree(ptr - LLIST_NODE_SZ);
}

/* When flags == GFP_KERNEL, it signals that the caller will not cause
 * deadlock when using kmalloc. bpf_mem_cache_alloc_flags() will use
 * kmalloc if the free_llist is empty.
 */
void notrace *bpf_mem_cache_alloc_flags(struct bpf_mem_alloc *ma, gfp_t flags)
{
	struct bpf_mem_cache *c;
	void *ret;

	c = this_cpu_ptr(ma->cache);

	ret = unit_alloc(c);
	if (!ret && flags == GFP_KERNEL) {
		struct mem_cgroup *memcg, *old_memcg;

		memcg = get_memcg(c);
		old_memcg = set_active_memcg(memcg);
		ret = __alloc(c, NUMA_NO_NODE, GFP_KERNEL | __GFP_NOWARN | __GFP_ACCOUNT);
		if (ret)
			*(struct bpf_mem_cache **)ret = c;
		set_active_memcg(old_memcg);
		mem_cgroup_put(memcg);
	}

	return !ret ? NULL : ret + LLIST_NODE_SZ;
}