aboutsummaryrefslogtreecommitdiffstats
path: root/net/sched/sch_fq.c
blob: d5185c44e9a5f521ca99243b6e9b53ec05b84d49 (plain) (blame)
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
/*
 * net/sched/sch_fq.c Fair Queue Packet Scheduler (per flow pacing)
 *
 *  Copyright (C) 2013-2015 Eric Dumazet <edumazet@google.com>
 *
 *	This program is free software; you can redistribute it and/or
 *	modify it under the terms of the GNU General Public License
 *	as published by the Free Software Foundation; either version
 *	2 of the License, or (at your option) any later version.
 *
 *  Meant to be mostly used for locally generated traffic :
 *  Fast classification depends on skb->sk being set before reaching us.
 *  If not, (router workload), we use rxhash as fallback, with 32 bits wide hash.
 *  All packets belonging to a socket are considered as a 'flow'.
 *
 *  Flows are dynamically allocated and stored in a hash table of RB trees
 *  They are also part of one Round Robin 'queues' (new or old flows)
 *
 *  Burst avoidance (aka pacing) capability :
 *
 *  Transport (eg TCP) can set in sk->sk_pacing_rate a rate, enqueue a
 *  bunch of packets, and this packet scheduler adds delay between
 *  packets to respect rate limitation.
 *
 *  enqueue() :
 *   - lookup one RB tree (out of 1024 or more) to find the flow.
 *     If non existent flow, create it, add it to the tree.
 *     Add skb to the per flow list of skb (fifo).
 *   - Use a special fifo for high prio packets
 *
 *  dequeue() : serves flows in Round Robin
 *  Note : When a flow becomes empty, we do not immediately remove it from
 *  rb trees, for performance reasons (its expected to send additional packets,
 *  or SLAB cache will reuse socket for another flow)
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/jiffies.h>
#include <linux/string.h>
#include <linux/in.h>
#include <linux/errno.h>
#include <linux/init.h>
#include <linux/skbuff.h>
#include <linux/slab.h>
#include <linux/rbtree.h>
#include <linux/hash.h>
#include <linux/prefetch.h>
#include <linux/vmalloc.h>
#include <net/netlink.h>
#include <net/pkt_sched.h>
#include <net/sock.h>
#include <net/tcp_states.h>
#include <net/tcp.h>

/*
 * Per flow structure, dynamically allocated
 */
struct fq_flow {
	struct sk_buff	*head;		/* list of skbs for this flow : first skb */
	union {
		struct sk_buff *tail;	/* last skb in the list */
		unsigned long  age;	/* jiffies when flow was emptied, for gc */
	};
	struct rb_node	fq_node;	/* anchor in fq_root[] trees */
	struct sock	*sk;
	int		qlen;		/* number of packets in flow queue */
	int		credit;
	u32		socket_hash;	/* sk_hash */
	struct fq_flow *next;		/* next pointer in RR lists, or &detached */

	struct rb_node  rate_node;	/* anchor in q->delayed tree */
	u64		time_next_packet;
};

struct fq_flow_head {
	struct fq_flow *first;
	struct fq_flow *last;
};

struct fq_sched_data {
	struct fq_flow_head new_flows;

	struct fq_flow_head old_flows;

	struct rb_root	delayed;	/* for rate limited flows */
	u64		time_next_delayed_flow;
	unsigned long	unthrottle_latency_ns;

	struct fq_flow	internal;	/* for non classified or high prio packets */
	u32		quantum;
	u32		initial_quantum;
	u32		flow_refill_delay;
	u32		flow_max_rate;	/* optional max rate per flow */
	u32		flow_plimit;	/* max packets per flow */
	u32		orphan_mask;	/* mask for orphaned skb */
	u32		low_rate_threshold;
	struct rb_root	*fq_root;
	u8		rate_enable;
	u8		fq_trees_log;

	u32		flows;
	u32		inactive_flows;
	u32		throttled_flows;

	u64		stat_gc_flows;
	u64		stat_internal_packets;
	u64		stat_tcp_retrans;
	u64		stat_throttled;
	u64		stat_flows_plimit;
	u64		stat_pkts_too_long;
	u64		stat_allocation_errors;
	struct qdisc_watchdog watchdog;
};

/* special value to mark a detached flow (not on old/new list) */
static struct fq_flow detached, throttled;

static void fq_flow_set_detached(struct fq_flow *f)
{
	f->next = &detached;
	f->age = jiffies;
}

static bool fq_flow_is_detached(const struct fq_flow *f)
{
	return f->next == &detached;
}

static bool fq_flow_is_throttled(const struct fq_flow *f)
{
	return f->next == &throttled;
}

static void fq_flow_add_tail(struct fq_flow_head *head, struct fq_flow *flow)
{
	if (head->first)
		head->last->next = flow;
	else
		head->first = flow;
	head->last = flow;
	flow->next = NULL;
}

static void fq_flow_unset_throttled(struct fq_sched_data *q, struct fq_flow *f)
{
	rb_erase(&f->rate_node, &q->delayed);
	q->throttled_flows--;
	fq_flow_add_tail(&q->old_flows, f);
}

static void fq_flow_set_throttled(struct fq_sched_data *q, struct fq_flow *f)
{
	struct rb_node **p = &q->delayed.rb_node, *parent = NULL;

	while (*p) {
		struct fq_flow *aux;

		parent = *p;
		aux = rb_entry(parent, struct fq_flow, rate_node);
		if (f->time_next_packet >= aux->time_next_packet)
			p = &parent->rb_right;
		else
			p = &parent->rb_left;
	}
	rb_link_node(&f->rate_node, parent, p);
	rb_insert_color(&f->rate_node, &q->delayed);
	q->throttled_flows++;
	q->stat_throttled++;

	f->next = &throttled;
	if (q->time_next_delayed_flow > f->time_next_packet)
		q->time_next_delayed_flow = f->time_next_packet;
}


static struct kmem_cache *fq_flow_cachep __read_mostly;


/* limit number of collected flows per round */
#define FQ_GC_MAX 8
#define FQ_GC_AGE (3*HZ)

static bool fq_gc_candidate(const struct fq_flow *f)
{
	return fq_flow_is_detached(f) &&
	       time_after(jiffies, f->age + FQ_GC_AGE);
}

static void fq_gc(struct fq_sched_data *q,
		  struct rb_root *root,
		  struct sock *sk)
{
	struct fq_flow *f, *tofree[FQ_GC_MAX];
	struct rb_node **p, *parent;
	int fcnt = 0;

	p = &root->rb_node;
	parent = NULL;
	while (*p) {
		parent = *p;

		f = rb_entry(parent, struct fq_flow, fq_node);
		if (f->sk == sk)
			break;

		if (fq_gc_candidate(f)) {
			tofree[fcnt++] = f;
			if (fcnt == FQ_GC_MAX)
				break;
		}

		if (f->sk > sk)
			p = &parent->rb_right;
		else
			p = &parent->rb_left;
	}

	q->flows -= fcnt;
	q->inactive_flows -= fcnt;
	q->stat_gc_flows += fcnt;
	while (fcnt) {
		struct fq_flow *f = tofree[--fcnt];

		rb_erase(&f->fq_node, root);
		kmem_cache_free(fq_flow_cachep, f);
	}
}

static struct fq_flow *fq_classify(struct sk_buff *skb, struct fq_sched_data *q)
{
	struct rb_node **p, *parent;
	struct sock *sk = skb->sk;
	struct rb_root *root;
	struct fq_flow *f;

	/* warning: no starvation prevention... */
	if (unlikely((skb->priority & TC_PRIO_MAX) == TC_PRIO_CONTROL))
		return &q->internal;

	/* SYNACK messages are attached to a TCP_NEW_SYN_RECV request socket
	 * or a listener (SYNCOOKIE mode)
	 * 1) request sockets are not full blown,
	 *    they do not contain sk_pacing_rate
	 * 2) They are not part of a 'flow' yet
	 * 3) We do not want to rate limit them (eg SYNFLOOD attack),
	 *    especially if the listener set SO_MAX_PACING_RATE
	 * 4) We pretend they are orphaned
	 */
	if (!sk || sk_listener(sk)) {
		unsigned long hash = skb_get_hash(skb) & q->orphan_mask;

		/* By forcing low order bit to 1, we make sure to not
		 * collide with a local flow (socket pointers are word aligned)
		 */
		sk = (struct sock *)((hash << 1) | 1UL);
		skb_orphan(skb);
	}

	root = &q->fq_root[hash_ptr(sk, q->fq_trees_log)];

	if (q->flows >= (2U << q->fq_trees_log) &&
	    q->inactive_flows > q->flows/2)
		fq_gc(q, root, sk);

	p = &root->rb_node;
	parent = NULL;
	while (*p) {
		parent = *p;

		f = rb_entry(parent, struct fq_flow, fq_node);
		if (f->sk == sk) {
			/* socket might have been reallocated, so check
			 * if its sk_hash is the same.
			 * It not, we need to refill credit with
			 * initial quantum
			 */
			if (unlikely(skb->sk &&
				     f->socket_hash != sk->sk_hash)) {
				f->credit = q->initial_quantum;
				f->socket_hash = sk->sk_hash;
				if (fq_flow_is_throttled(f))
					fq_flow_unset_throttled(q, f);
				f->time_next_packet = 0ULL;
			}
			return f;
		}
		if (f->sk > sk)
			p = &parent->rb_right;
		else
			p = &parent->rb_left;
	}

	f = kmem_cache_zalloc(fq_flow_cachep, GFP_ATOMIC | __GFP_NOWARN);
	if (unlikely(!f)) {
		q->stat_allocation_errors++;
		return &q->internal;
	}
	fq_flow_set_detached(f);
	f->sk = sk;
	if (skb->sk)
		f->socket_hash = sk->sk_hash;
	f->credit = q->initial_quantum;

	rb_link_node(&f->fq_node, parent, p);
	rb_insert_color(&f->fq_node, root);

	q->flows++;
	q->inactive_flows++;
	return f;
}


/* remove one skb from head of flow queue */
static struct sk_buff *fq_dequeue_head(struct Qdisc *sch, struct fq_flow *flow)
{
	struct sk_buff *skb = flow->head;

	if (skb) {
		flow->head = skb->next;
		skb_mark_not_on_list(skb);
		flow->qlen--;
		qdisc_qstats_backlog_dec(sch, skb);
		sch->q.qlen--;
	}
	return skb;
}

/* We might add in the future detection of retransmits
 * For the time being, just return false
 */
static bool skb_is_retransmit(struct sk_buff *skb)
{
	return false;
}

/* add skb to flow queue
 * flow queue is a linked list, kind of FIFO, except for TCP retransmits
 * We special case tcp retransmits to be transmitted before other packets.
 * We rely on fact that TCP retransmits are unlikely, so we do not waste
 * a separate queue or a pointer.
 * head->  [retrans pkt 1]
 *         [retrans pkt 2]
 *         [ normal pkt 1]
 *         [ normal pkt 2]
 *         [ normal pkt 3]
 * tail->  [ normal pkt 4]
 */
static void flow_queue_add(struct fq_flow *flow, struct sk_buff *skb)
{
	struct sk_buff *prev, *head = flow->head;

	skb->next = NULL;
	if (!head) {
		flow->head = skb;
		flow->tail = skb;
		return;
	}
	if (likely(!skb_is_retransmit(skb))) {
		flow->tail->next = skb;
		flow->tail = skb;
		return;
	}

	/* This skb is a tcp retransmit,
	 * find the last retrans packet in the queue
	 */
	prev = NULL;
	while (skb_is_retransmit(head)) {
		prev = head;
		head = head->next;
		if (!head)
			break;
	}
	if (!prev) { /* no rtx packet in queue, become the new head */
		skb->next = flow->head;
		flow->head = skb;
	} else {
		if (prev == flow->tail)
			flow->tail = skb;
		else
			skb->next = prev->next;
		prev->next = skb;
	}
}

static int fq_enqueue(struct sk_buff *skb, struct Qdisc *sch,
		      struct sk_buff **to_free)
{
	struct fq_sched_data *q = qdisc_priv(sch);
	struct fq_flow *f;

	if (unlikely(sch->q.qlen >= sch->limit))
		return qdisc_drop(skb, sch, to_free);

	f = fq_classify(skb, q);
	if (unlikely(f->qlen >= q->flow_plimit && f != &q->internal)) {
		q->stat_flows_plimit++;
		return qdisc_drop(skb, sch, to_free);
	}

	f->qlen++;
	if (skb_is_retransmit(skb))
		q->stat_tcp_retrans++;
	qdisc_qstats_backlog_inc(sch, skb);
	if (fq_flow_is_detached(f)) {
		struct sock *sk = skb->sk;

		fq_flow_add_tail(&q->new_flows, f);
		if (time_after(jiffies, f->age + q->flow_refill_delay))
			f->credit = max_t(u32, f->credit, q->quantum);
		if (sk && q->rate_enable) {
			if (unlikely(smp_load_acquire(&sk->sk_pacing_status) !=
				     SK_PACING_FQ))
				smp_store_release(&sk->sk_pacing_status,
						  SK_PACING_FQ);
		}
		q->inactive_flows--;
	}

	/* Note: this overwrites f->age */
	flow_queue_add(f, skb);

	if (unlikely(f == &q->internal)) {
		q->stat_internal_packets++;
	}
	sch->q.qlen++;

	return NET_XMIT_SUCCESS;
}

static void fq_check_throttled(struct fq_sched_data *q, u64 now)
{
	unsigned long sample;
	struct rb_node *p;

	if (q->time_next_delayed_flow > now)
		return;

	/* Update unthrottle latency EWMA.
	 * This is cheap and can help diagnosing timer/latency problems.
	 */
	sample = (unsigned long)(now - q->time_next_delayed_flow);
	q->unthrottle_latency_ns -= q->unthrottle_latency_ns >> 3;
	q->unthrottle_latency_ns += sample >> 3;

	q->time_next_delayed_flow = ~0ULL;
	while ((p = rb_first(&q->delayed)) != NULL) {
		struct fq_flow *f = rb_entry(p, struct fq_flow, rate_node);

		if (f->time_next_packet > now) {
			q->time_next_delayed_flow = f->time_next_packet;
			break;
		}
		fq_flow_unset_throttled(q, f);
	}
}

static struct sk_buff *fq_dequeue(struct Qdisc *sch)
{
	struct fq_sched_data *q = qdisc_priv(sch);
	u64 now = ktime_get_tai_ns();
	struct fq_flow_head *head;
	struct sk_buff *skb;
	struct fq_flow *f;
	u32 rate, plen;

	skb = fq_dequeue_head(sch, &q->internal);
	if (skb)
		goto out;
	fq_check_throttled(q, now);
begin:
	head = &q->new_flows;
	if (!head->first) {
		head = &q->old_flows;
		if (!head->first) {
			if (q->time_next_delayed_flow != ~0ULL)
				qdisc_watchdog_schedule_ns(&q->watchdog,
							   q->time_next_delayed_flow);
			return NULL;
		}
	}
	f = head->first;

	if (f->credit <= 0) {
		f->credit += q->quantum;
		head->first = f->next;
		fq_flow_add_tail(&q->old_flows, f);
		goto begin;
	}

	skb = f->head;
	if (unlikely(skb && now < f->time_next_packet &&
		     !skb_is_tcp_pure_ack(skb))) {
		head->first = f->next;
		fq_flow_set_throttled(q, f);
		goto begin;
	}

	skb = fq_dequeue_head(sch, f);
	if (!skb) {
		head->first = f->next;
		/* force a pass through old_flows to prevent starvation */
		if ((head == &q->new_flows) && q->old_flows.first) {
			fq_flow_add_tail(&q->old_flows, f);
		} else {
			fq_flow_set_detached(f);
			q->inactive_flows++;
		}
		goto begin;
	}
	prefetch(&skb->end);
	f->credit -= qdisc_pkt_len(skb);

	if (!q->rate_enable)
		goto out;

	/* Do not pace locally generated ack packets */
	if (skb_is_tcp_pure_ack(skb))
		goto out;

	rate = q->flow_max_rate;
	if (skb->sk)
		rate = min(skb->sk->sk_pacing_rate, rate);

	if (rate <= q->low_rate_threshold) {
		f->credit = 0;
		plen = qdisc_pkt_len(skb);
	} else {
		plen = max(qdisc_pkt_len(skb), q->quantum);
		if (f->credit > 0)
			goto out;
	}
	if (rate != ~0U) {
		u64 len = (u64)plen * NSEC_PER_SEC;

		if (likely(rate))
			do_div(len, rate);
		/* Since socket rate can change later,
		 * clamp the delay to 1 second.
		 * Really, providers of too big packets should be fixed !
		 */
		if (unlikely(len > NSEC_PER_SEC)) {
			len = NSEC_PER_SEC;
			q->stat_pkts_too_long++;
		}
		/* Account for schedule/timers drifts.
		 * f->time_next_packet was set when prior packet was sent,
		 * and current time (@now) can be too late by tens of us.
		 */
		if (f->time_next_packet)
			len -= min(len/2, now - f->time_next_packet);
		f->time_next_packet = now + len;
	}
out:
	qdisc_bstats_update(sch, skb);
	return skb;
}

static void fq_flow_purge(struct fq_flow *flow)
{
	rtnl_kfree_skbs(flow->head, flow->tail);
	flow->head = NULL;
	flow->qlen = 0;
}

static void fq_reset(struct Qdisc *sch)
{
	struct fq_sched_data *q = qdisc_priv(sch);
	struct rb_root *root;
	struct rb_node *p;
	struct fq_flow *f;
	unsigned int idx;

	sch->q.qlen = 0;
	sch->qstats.backlog = 0;

	fq_flow_purge(&q->internal);

	if (!q->fq_root)
		return;

	for (idx = 0; idx < (1U << q->fq_trees_log); idx++) {
		root = &q->fq_root[idx];
		while ((p = rb_first(root)) != NULL) {
			f = rb_entry(p, struct fq_flow, fq_node);
			rb_erase(p, root);

			fq_flow_purge(f);

			kmem_cache_free(fq_flow_cachep, f);
		}
	}
	q->new_flows.first	= NULL;
	q->old_flows.first	= NULL;
	q->delayed		= RB_ROOT;
	q->flows		= 0;
	q->inactive_flows	= 0;
	q->throttled_flows	= 0;
}

static void fq_rehash(struct fq_sched_data *q,
		      struct rb_root *old_array, u32 old_log,
		      struct rb_root *new_array, u32 new_log)
{
	struct rb_node *op, **np, *parent;
	struct rb_root *oroot, *nroot;
	struct fq_flow *of, *nf;
	int fcnt = 0;
	u32 idx;

	for (idx = 0; idx < (1U << old_log); idx++) {
		oroot = &old_array[idx];
		while ((op = rb_first(oroot)) != NULL) {
			rb_erase(op, oroot);
			of = rb_entry(op, struct fq_flow, fq_node);
			if (fq_gc_candidate(of)) {
				fcnt++;
				kmem_cache_free(fq_flow_cachep, of);
				continue;
			}
			nroot = &new_array[hash_ptr(of->sk, new_log)];

			np = &nroot->rb_node;
			parent = NULL;
			while (*np) {
				parent = *np;

				nf = rb_entry(parent, struct fq_flow, fq_node);
				BUG_ON(nf->sk == of->sk);

				if (nf->sk > of->sk)
					np = &parent->rb_right;
				else
					np = &parent->rb_left;
			}

			rb_link_node(&of->fq_node, parent, np);
			rb_insert_color(&of->fq_node, nroot);
		}
	}
	q->flows -= fcnt;
	q->inactive_flows -= fcnt;
	q->stat_gc_flows += fcnt;
}

static void fq_free(void *addr)
{
	kvfree(addr);
}

static int fq_resize(struct Qdisc *sch, u32 log)
{
	struct fq_sched_data *q = qdisc_priv(sch);
	struct rb_root *array;
	void *old_fq_root;
	u32 idx;

	if (q->fq_root && log == q->fq_trees_log)
		return 0;

	/* If XPS was setup, we can allocate memory on right NUMA node */
	array = kvmalloc_node(sizeof(struct rb_root) << log, GFP_KERNEL | __GFP_RETRY_MAYFAIL,
			      netdev_queue_numa_node_read(sch->dev_queue));
	if (!array)
		return -ENOMEM;

	for (idx = 0; idx < (1U << log); idx++)
		array[idx] = RB_ROOT;

	sch_tree_lock(sch);

	old_fq_root = q->fq_root;
	if (old_fq_root)
		fq_rehash(q, old_fq_root, q->fq_trees_log, array, log);

	q->fq_root = array;
	q->fq_trees_log = log;

	sch_tree_unlock(sch);

	fq_free(old_fq_root);

	return 0;
}

static const struct nla_policy fq_policy[TCA_FQ_MAX + 1] = {
	[TCA_FQ_PLIMIT]			= { .type = NLA_U32 },
	[TCA_FQ_FLOW_PLIMIT]		= { .type = NLA_U32 },
	[TCA_FQ_QUANTUM]		= { .type = NLA_U32 },
	[TCA_FQ_INITIAL_QUANTUM]	= { .type = NLA_U32 },
	[TCA_FQ_RATE_ENABLE]		= { .type = NLA_U32 },
	[TCA_FQ_FLOW_DEFAULT_RATE]	= { .type = NLA_U32 },
	[TCA_FQ_FLOW_MAX_RATE]		= { .type = NLA_U32 },
	[TCA_FQ_BUCKETS_LOG]		= { .type = NLA_U32 },
	[TCA_FQ_FLOW_REFILL_DELAY]	= { .type = NLA_U32 },
	[TCA_FQ_LOW_RATE_THRESHOLD]	= { .type = NLA_U32 },
};

static int fq_change(struct Qdisc *sch, struct nlattr *opt,
		     struct netlink_ext_ack *extack)
{
	struct fq_sched_data *q = qdisc_priv(sch);
	struct nlattr *tb[TCA_FQ_MAX + 1];
	int err, drop_count = 0;
	unsigned drop_len = 0;
	u32 fq_log;

	if (!opt)
		return -EINVAL;

	err = nla_parse_nested(tb, TCA_FQ_MAX, opt, fq_policy, NULL);
	if (err < 0)
		return err;

	sch_tree_lock(sch);

	fq_log = q->fq_trees_log;

	if (tb[TCA_FQ_BUCKETS_LOG]) {
		u32 nval = nla_get_u32(tb[TCA_FQ_BUCKETS_LOG]);

		if (nval >= 1 && nval <= ilog2(256*1024))
			fq_log = nval;
		else
			err = -EINVAL;
	}
	if (tb[TCA_FQ_PLIMIT])
		sch->limit = nla_get_u32(tb[TCA_FQ_PLIMIT]);

	if (tb[TCA_FQ_FLOW_PLIMIT])
		q->flow_plimit = nla_get_u32(tb[TCA_FQ_FLOW_PLIMIT]);

	if (tb[TCA_FQ_QUANTUM]) {
		u32 quantum = nla_get_u32(tb[TCA_FQ_QUANTUM]);

		if (quantum > 0)
			q->quantum = quantum;
		else
			err = -EINVAL;
	}

	if (tb[TCA_FQ_INITIAL_QUANTUM])
		q->initial_quantum = nla_get_u32(tb[TCA_FQ_INITIAL_QUANTUM]);

	if (tb[TCA_FQ_FLOW_DEFAULT_RATE])
		pr_warn_ratelimited("sch_fq: defrate %u ignored.\n",
				    nla_get_u32(tb[TCA_FQ_FLOW_DEFAULT_RATE]));

	if (tb[TCA_FQ_FLOW_MAX_RATE])
		q->flow_max_rate = nla_get_u32(tb[TCA_FQ_FLOW_MAX_RATE]);

	if (tb[TCA_FQ_LOW_RATE_THRESHOLD])
		q->low_rate_threshold =
			nla_get_u32(tb[TCA_FQ_LOW_RATE_THRESHOLD]);

	if (tb[TCA_FQ_RATE_ENABLE]) {
		u32 enable = nla_get_u32(tb[TCA_FQ_RATE_ENABLE]);

		if (enable <= 1)
			q->rate_enable = enable;
		else
			err = -EINVAL;
	}

	if (tb[TCA_FQ_FLOW_REFILL_DELAY]) {
		u32 usecs_delay = nla_get_u32(tb[TCA_FQ_FLOW_REFILL_DELAY]) ;

		q->flow_refill_delay = usecs_to_jiffies(usecs_delay);
	}

	if (tb[TCA_FQ_ORPHAN_MASK])
		q->orphan_mask = nla_get_u32(tb[TCA_FQ_ORPHAN_MASK]);

	if (!err) {
		sch_tree_unlock(sch);
		err = fq_resize(sch, fq_log);
		sch_tree_lock(sch);
	}
	while (sch->q.qlen > sch->limit) {
		struct sk_buff *skb = fq_dequeue(sch);

		if (!skb)
			break;
		drop_len += qdisc_pkt_len(skb);
		rtnl_kfree_skbs(skb, skb);
		drop_count++;
	}
	qdisc_tree_reduce_backlog(sch, drop_count, drop_len);

	sch_tree_unlock(sch);
	return err;
}

static void fq_destroy(struct Qdisc *sch)
{
	struct fq_sched_data *q = qdisc_priv(sch);

	fq_reset(sch);
	fq_free(q->fq_root);
	qdisc_watchdog_cancel(&q->watchdog);
}

static int fq_init(struct Qdisc *sch, struct nlattr *opt,
		   struct netlink_ext_ack *extack)
{
	struct fq_sched_data *q = qdisc_priv(sch);
	int err;

	sch->limit		= 10000;
	q->flow_plimit		= 100;
	q->quantum		= 2 * psched_mtu(qdisc_dev(sch));
	q->initial_quantum	= 10 * psched_mtu(qdisc_dev(sch));
	q->flow_refill_delay	= msecs_to_jiffies(40);
	q->flow_max_rate	= ~0U;
	q->time_next_delayed_flow = ~0ULL;
	q->rate_enable		= 1;
	q->new_flows.first	= NULL;
	q->old_flows.first	= NULL;
	q->delayed		= RB_ROOT;
	q->fq_root		= NULL;
	q->fq_trees_log		= ilog2(1024);
	q->orphan_mask		= 1024 - 1;
	q->low_rate_threshold	= 550000 / 8;
	qdisc_watchdog_init_clockid(&q->watchdog, sch, CLOCK_TAI);

	if (opt)
		err = fq_change(sch, opt, extack);
	else
		err = fq_resize(sch, q->fq_trees_log);

	return err;
}

static int fq_dump(struct Qdisc *sch, struct sk_buff *skb)
{
	struct fq_sched_data *q = qdisc_priv(sch);
	struct nlattr *opts;

	opts = nla_nest_start(skb, TCA_OPTIONS);
	if (opts == NULL)
		goto nla_put_failure;

	/* TCA_FQ_FLOW_DEFAULT_RATE is not used anymore */

	if (nla_put_u32(skb, TCA_FQ_PLIMIT, sch->limit) ||
	    nla_put_u32(skb, TCA_FQ_FLOW_PLIMIT, q->flow_plimit) ||
	    nla_put_u32(skb, TCA_FQ_QUANTUM, q->quantum) ||
	    nla_put_u32(skb, TCA_FQ_INITIAL_QUANTUM, q->initial_quantum) ||
	    nla_put_u32(skb, TCA_FQ_RATE_ENABLE, q->rate_enable) ||
	    nla_put_u32(skb, TCA_FQ_FLOW_MAX_RATE, q->flow_max_rate) ||
	    nla_put_u32(skb, TCA_FQ_FLOW_REFILL_DELAY,
			jiffies_to_usecs(q->flow_refill_delay)) ||
	    nla_put_u32(skb, TCA_FQ_ORPHAN_MASK, q->orphan_mask) ||
	    nla_put_u32(skb, TCA_FQ_LOW_RATE_THRESHOLD,
			q->low_rate_threshold) ||
	    nla_put_u32(skb, TCA_FQ_BUCKETS_LOG, q->fq_trees_log))
		goto nla_put_failure;

	return nla_nest_end(skb, opts);

nla_put_failure:
	return -1;
}

static int fq_dump_stats(struct Qdisc *sch, struct gnet_dump *d)
{
	struct fq_sched_data *q = qdisc_priv(sch);
	struct tc_fq_qd_stats st;

	sch_tree_lock(sch);

	st.gc_flows		  = q->stat_gc_flows;
	st.highprio_packets	  = q->stat_internal_packets;
	st.tcp_retrans		  = q->stat_tcp_retrans;
	st.throttled		  = q->stat_throttled;
	st.flows_plimit		  = q->stat_flows_plimit;
	st.pkts_too_long	  = q->stat_pkts_too_long;
	st.allocation_errors	  = q->stat_allocation_errors;
	st.time_next_delayed_flow = q->time_next_delayed_flow - ktime_get_tai_ns();
	st.flows		  = q->flows;
	st.inactive_flows	  = q->inactive_flows;
	st.throttled_flows	  = q->throttled_flows;
	st.unthrottle_latency_ns  = min_t(unsigned long,
					  q->unthrottle_latency_ns, ~0U);
	sch_tree_unlock(sch);

	return gnet_stats_copy_app(d, &st, sizeof(st));
}

static struct Qdisc_ops fq_qdisc_ops __read_mostly = {
	.id		=	"fq",
	.priv_size	=	sizeof(struct fq_sched_data),

	.enqueue	=	fq_enqueue,
	.dequeue	=	fq_dequeue,
	.peek		=	qdisc_peek_dequeued,
	.init		=	fq_init,
	.reset		=	fq_reset,
	.destroy	=	fq_destroy,
	.change		=	fq_change,
	.dump		=	fq_dump,
	.dump_stats	=	fq_dump_stats,
	.owner		=	THIS_MODULE,
};

static int __init fq_module_init(void)
{
	int ret;

	fq_flow_cachep = kmem_cache_create("fq_flow_cache",
					   sizeof(struct fq_flow),
					   0, 0, NULL);
	if (!fq_flow_cachep)
		return -ENOMEM;

	ret = register_qdisc(&fq_qdisc_ops);
	if (ret)
		kmem_cache_destroy(fq_flow_cachep);
	return ret;
}

static void __exit fq_module_exit(void)
{
	unregister_qdisc(&fq_qdisc_ops);
	kmem_cache_destroy(fq_flow_cachep);
}

module_init(fq_module_init)
module_exit(fq_module_exit)
MODULE_AUTHOR("Eric Dumazet");
MODULE_LICENSE("GPL");