sch_generic.h 22.6 KB
Newer Older
1
/* SPDX-License-Identifier: GPL-2.0 */
L
Linus Torvalds 已提交
2 3 4 5 6 7 8 9
#ifndef __NET_SCHED_GENERIC_H
#define __NET_SCHED_GENERIC_H

#include <linux/netdevice.h>
#include <linux/types.h>
#include <linux/rcupdate.h>
#include <linux/pkt_sched.h>
#include <linux/pkt_cls.h>
10
#include <linux/percpu.h>
11
#include <linux/dynamic_queue_limits.h>
12
#include <linux/list.h>
13
#include <linux/refcount.h>
14
#include <linux/workqueue.h>
L
Linus Torvalds 已提交
15
#include <net/gen_stats.h>
16
#include <net/rtnetlink.h>
L
Linus Torvalds 已提交
17 18 19 20 21 22

struct Qdisc_ops;
struct qdisc_walker;
struct tcf_walker;
struct module;

E
Eric Dumazet 已提交
23
struct qdisc_rate_table {
L
Linus Torvalds 已提交
24 25 26 27 28 29
	struct tc_ratespec rate;
	u32		data[256];
	struct qdisc_rate_table *next;
	int		refcnt;
};

E
Eric Dumazet 已提交
30
enum qdisc_state_t {
31
	__QDISC_STATE_SCHED,
32
	__QDISC_STATE_DEACTIVATED,
33 34
};

35
struct qdisc_size_table {
E
Eric Dumazet 已提交
36
	struct rcu_head		rcu;
37 38 39 40 41 42
	struct list_head	list;
	struct tc_sizespec	szopts;
	int			refcnt;
	u16			data[];
};

43 44 45 46 47 48 49 50
/* similar to sk_buff_head, but skb->prev pointer is undefined. */
struct qdisc_skb_head {
	struct sk_buff	*head;
	struct sk_buff	*tail;
	__u32		qlen;
	spinlock_t	lock;
};

E
Eric Dumazet 已提交
51
struct Qdisc {
52 53 54 55
	int 			(*enqueue)(struct sk_buff *skb,
					   struct Qdisc *sch,
					   struct sk_buff **to_free);
	struct sk_buff *	(*dequeue)(struct Qdisc *sch);
56
	unsigned int		flags;
57
#define TCQ_F_BUILTIN		1
58 59 60
#define TCQ_F_INGRESS		2
#define TCQ_F_CAN_BYPASS	4
#define TCQ_F_MQROOT		8
61 62 63 64 65 66 67
#define TCQ_F_ONETXQUEUE	0x10 /* dequeue_skb() can assume all skbs are for
				      * q->dev_queue : It can test
				      * netif_xmit_frozen_or_stopped() before
				      * dequeueing next packet.
				      * Its true for MQ/MQPRIO slaves, or non
				      * multiqueue device.
				      */
68
#define TCQ_F_WARN_NONWC	(1 << 16)
69
#define TCQ_F_CPUSTATS		0x20 /* run using percpu statistics */
70 71 72
#define TCQ_F_NOPARENT		0x40 /* root of its hierarchy :
				      * qdisc_tree_decrease_qlen() should stop.
				      */
73
#define TCQ_F_INVISIBLE		0x80 /* invisible by default in dump */
74
	u32			limit;
75
	const struct Qdisc_ops	*ops;
E
Eric Dumazet 已提交
76
	struct qdisc_size_table	__rcu *stab;
77
	struct hlist_node       hash;
L
Linus Torvalds 已提交
78 79
	u32			handle;
	u32			parent;
80

81 82
	struct netdev_queue	*dev_queue;

83
	struct net_rate_estimator __rcu *rate_est;
84 85 86
	struct gnet_stats_basic_cpu __percpu *cpu_bstats;
	struct gnet_stats_queue	__percpu *cpu_qstats;

87 88 89
	/*
	 * For performance sake on SMP, we put highly modified fields at the end
	 */
90
	struct sk_buff		*gso_skb ____cacheline_aligned_in_smp;
91
	struct qdisc_skb_head	q;
92
	struct gnet_stats_basic_packed bstats;
93
	seqcount_t		running;
94
	struct gnet_stats_queue	qstats;
95 96 97
	unsigned long		state;
	struct Qdisc            *next_sched;
	struct sk_buff		*skb_bad_txq;
98
	int			padded;
99
	refcount_t		refcnt;
100 101

	spinlock_t		busylock ____cacheline_aligned_in_smp;
L
Linus Torvalds 已提交
102 103
};

104 105 106 107 108 109 110
static inline void qdisc_refcount_inc(struct Qdisc *qdisc)
{
	if (qdisc->flags & TCQ_F_BUILTIN)
		return;
	refcount_inc(&qdisc->refcnt);
}

111
static inline bool qdisc_is_running(const struct Qdisc *qdisc)
112
{
113
	return (raw_read_seqcount(&qdisc->running) & 1) ? true : false;
114 115 116 117
}

static inline bool qdisc_run_begin(struct Qdisc *qdisc)
{
118 119
	if (qdisc_is_running(qdisc))
		return false;
120 121 122 123 124
	/* Variant of write_seqcount_begin() telling lockdep a trylock
	 * was attempted.
	 */
	raw_write_seqcount_begin(&qdisc->running);
	seqcount_acquire(&qdisc->running.dep_map, 0, 1, _RET_IP_);
125
	return true;
126 127 128 129
}

static inline void qdisc_run_end(struct Qdisc *qdisc)
{
130
	write_seqcount_end(&qdisc->running);
131 132
}

133 134 135 136 137 138 139 140 141 142 143 144 145 146 147
static inline bool qdisc_may_bulk(const struct Qdisc *qdisc)
{
	return qdisc->flags & TCQ_F_ONETXQUEUE;
}

static inline int qdisc_avail_bulklimit(const struct netdev_queue *txq)
{
#ifdef CONFIG_BQL
	/* Non-BQL migrated drivers will return 0, too. */
	return dql_avail(&txq->dql);
#else
	return 0;
#endif
}

E
Eric Dumazet 已提交
148
struct Qdisc_class_ops {
L
Linus Torvalds 已提交
149
	/* Child qdisc manipulation */
150
	struct netdev_queue *	(*select_queue)(struct Qdisc *, struct tcmsg *);
L
Linus Torvalds 已提交
151 152 153
	int			(*graft)(struct Qdisc *, unsigned long cl,
					struct Qdisc *, struct Qdisc **);
	struct Qdisc *		(*leaf)(struct Qdisc *, unsigned long cl);
154
	void			(*qlen_notify)(struct Qdisc *, unsigned long);
L
Linus Torvalds 已提交
155 156

	/* Class manipulation routines */
157
	unsigned long		(*find)(struct Qdisc *, u32 classid);
L
Linus Torvalds 已提交
158
	int			(*change)(struct Qdisc *, u32, u32,
159
					struct nlattr **, unsigned long *);
L
Linus Torvalds 已提交
160 161 162 163
	int			(*delete)(struct Qdisc *, unsigned long);
	void			(*walk)(struct Qdisc *, struct qdisc_walker * arg);

	/* Filter manipulation */
164
	struct tcf_block *	(*tcf_block)(struct Qdisc *, unsigned long);
L
Linus Torvalds 已提交
165 166 167 168 169 170 171 172 173 174 175
	unsigned long		(*bind_tcf)(struct Qdisc *, unsigned long,
					u32 classid);
	void			(*unbind_tcf)(struct Qdisc *, unsigned long);

	/* rtnetlink specific */
	int			(*dump)(struct Qdisc *, unsigned long,
					struct sk_buff *skb, struct tcmsg*);
	int			(*dump_stats)(struct Qdisc *, unsigned long,
					struct gnet_dump *);
};

E
Eric Dumazet 已提交
176
struct Qdisc_ops {
L
Linus Torvalds 已提交
177
	struct Qdisc_ops	*next;
178
	const struct Qdisc_class_ops	*cl_ops;
L
Linus Torvalds 已提交
179 180 181
	char			id[IFNAMSIZ];
	int			priv_size;

182 183 184
	int 			(*enqueue)(struct sk_buff *skb,
					   struct Qdisc *sch,
					   struct sk_buff **to_free);
L
Linus Torvalds 已提交
185
	struct sk_buff *	(*dequeue)(struct Qdisc *);
186
	struct sk_buff *	(*peek)(struct Qdisc *);
L
Linus Torvalds 已提交
187

188
	int			(*init)(struct Qdisc *, struct nlattr *arg);
L
Linus Torvalds 已提交
189 190
	void			(*reset)(struct Qdisc *);
	void			(*destroy)(struct Qdisc *);
191
	int			(*change)(struct Qdisc *, struct nlattr *arg);
192
	void			(*attach)(struct Qdisc *);
L
Linus Torvalds 已提交
193 194 195 196 197 198 199 200

	int			(*dump)(struct Qdisc *, struct sk_buff *);
	int			(*dump_stats)(struct Qdisc *, struct gnet_dump *);

	struct module		*owner;
};


E
Eric Dumazet 已提交
201
struct tcf_result {
202 203 204 205 206 207 208
	union {
		struct {
			unsigned long	class;
			u32		classid;
		};
		const struct tcf_proto *goto_tp;
	};
L
Linus Torvalds 已提交
209 210
};

E
Eric Dumazet 已提交
211
struct tcf_proto_ops {
212
	struct list_head	head;
L
Linus Torvalds 已提交
213 214
	char			kind[IFNAMSIZ];

215 216 217
	int			(*classify)(struct sk_buff *,
					    const struct tcf_proto *,
					    struct tcf_result *);
L
Linus Torvalds 已提交
218
	int			(*init)(struct tcf_proto*);
219
	void			(*destroy)(struct tcf_proto*);
L
Linus Torvalds 已提交
220

221
	void*			(*get)(struct tcf_proto*, u32 handle);
222
	int			(*change)(struct net *net, struct sk_buff *,
223
					struct tcf_proto*, unsigned long,
224
					u32 handle, struct nlattr **,
225 226
					void **, bool);
	int			(*delete)(struct tcf_proto*, void *, bool*);
L
Linus Torvalds 已提交
227
	void			(*walk)(struct tcf_proto*, struct tcf_walker *arg);
228
	void			(*bind_class)(void *, u32, unsigned long);
L
Linus Torvalds 已提交
229 230

	/* rtnetlink specific */
231
	int			(*dump)(struct net*, struct tcf_proto*, void *,
L
Linus Torvalds 已提交
232 233 234 235 236
					struct sk_buff *skb, struct tcmsg*);

	struct module		*owner;
};

E
Eric Dumazet 已提交
237
struct tcf_proto {
L
Linus Torvalds 已提交
238
	/* Fast access part */
J
John Fastabend 已提交
239 240
	struct tcf_proto __rcu	*next;
	void __rcu		*root;
241 242 243
	int			(*classify)(struct sk_buff *,
					    const struct tcf_proto *,
					    struct tcf_result *);
A
Al Viro 已提交
244
	__be16			protocol;
L
Linus Torvalds 已提交
245 246 247 248 249 250

	/* All the rest */
	u32			prio;
	u32			classid;
	struct Qdisc		*q;
	void			*data;
251
	const struct tcf_proto_ops	*ops;
252
	struct tcf_chain	*chain;
J
John Fastabend 已提交
253
	struct rcu_head		rcu;
L
Linus Torvalds 已提交
254 255
};

256 257
struct qdisc_skb_cb {
	unsigned int		pkt_len;
258
	u16			slave_dev_queue_mapping;
259
	u16			tc_classid;
260 261
#define QDISC_CB_PRIV_LEN 20
	unsigned char		data[QDISC_CB_PRIV_LEN];
262 263
};

264 265
typedef void tcf_chain_head_change_t(struct tcf_proto *tp_head, void *priv);

266 267
struct tcf_chain {
	struct tcf_proto __rcu *filter_chain;
268 269
	tcf_chain_head_change_t *chain_head_change;
	void *chain_head_change_priv;
270 271 272 273
	struct list_head list;
	struct tcf_block *block;
	u32 index; /* chain index */
	unsigned int refcnt;
274 275
};

276
struct tcf_block {
277
	struct list_head chain_list;
278
	struct net *net;
279
	struct Qdisc *q;
280
	struct list_head cb_list;
281
	struct work_struct work;
282 283
};

284 285 286
static inline void qdisc_cb_private_validate(const struct sk_buff *skb, int sz)
{
	struct qdisc_skb_cb *qcb;
287 288

	BUILD_BUG_ON(sizeof(skb->cb) < offsetof(struct qdisc_skb_cb, data) + sz);
289 290 291
	BUILD_BUG_ON(sizeof(qcb->data) < sz);
}

292
static inline int qdisc_qlen(const struct Qdisc *q)
293 294 295 296
{
	return q->q.qlen;
}

297
static inline struct qdisc_skb_cb *qdisc_skb_cb(const struct sk_buff *skb)
298 299 300 301
{
	return (struct qdisc_skb_cb *)skb->cb;
}

302 303 304 305 306
static inline spinlock_t *qdisc_lock(struct Qdisc *qdisc)
{
	return &qdisc->q.lock;
}

307
static inline struct Qdisc *qdisc_root(const struct Qdisc *qdisc)
308
{
309 310 311
	struct Qdisc *q = rcu_dereference_rtnl(qdisc->dev_queue->qdisc);

	return q;
312 313
}

314
static inline struct Qdisc *qdisc_root_sleeping(const struct Qdisc *qdisc)
315 316 317 318
{
	return qdisc->dev_queue->qdisc_sleeping;
}

319 320 321 322 323 324 325 326 327 328 329
/* The qdisc root lock is a mechanism by which to top level
 * of a qdisc tree can be locked from any qdisc node in the
 * forest.  This allows changing the configuration of some
 * aspect of the qdisc tree while blocking out asynchronous
 * qdisc access in the packet processing paths.
 *
 * It is only legal to do this when the root will not change
 * on us.  Otherwise we'll potentially lock the wrong qdisc
 * root.  This is enforced by holding the RTNL semaphore, which
 * all users of this lock accessor must do.
 */
330
static inline spinlock_t *qdisc_root_lock(const struct Qdisc *qdisc)
331 332 333
{
	struct Qdisc *root = qdisc_root(qdisc);

334
	ASSERT_RTNL();
335
	return qdisc_lock(root);
336 337
}

338
static inline spinlock_t *qdisc_root_sleeping_lock(const struct Qdisc *qdisc)
339 340 341 342 343 344 345
{
	struct Qdisc *root = qdisc_root_sleeping(qdisc);

	ASSERT_RTNL();
	return qdisc_lock(root);
}

346 347 348 349 350 351 352 353
static inline seqcount_t *qdisc_root_sleeping_running(const struct Qdisc *qdisc)
{
	struct Qdisc *root = qdisc_root_sleeping(qdisc);

	ASSERT_RTNL();
	return &root->running;
}

354
static inline struct net_device *qdisc_dev(const struct Qdisc *qdisc)
355 356 357
{
	return qdisc->dev_queue->dev;
}
L
Linus Torvalds 已提交
358

359
static inline void sch_tree_lock(const struct Qdisc *q)
360
{
J
Jarek Poplawski 已提交
361
	spin_lock_bh(qdisc_root_sleeping_lock(q));
362 363
}

364
static inline void sch_tree_unlock(const struct Qdisc *q)
365
{
J
Jarek Poplawski 已提交
366
	spin_unlock_bh(qdisc_root_sleeping_lock(q));
367 368
}

369 370
extern struct Qdisc noop_qdisc;
extern struct Qdisc_ops noop_qdisc_ops;
371 372
extern struct Qdisc_ops pfifo_fast_ops;
extern struct Qdisc_ops mq_qdisc_ops;
P
Phil Sutter 已提交
373
extern struct Qdisc_ops noqueue_qdisc_ops;
374
extern const struct Qdisc_ops *default_qdisc_ops;
375 376 377 378 379 380
static inline const struct Qdisc_ops *
get_default_qdisc_ops(const struct net_device *dev, int ntx)
{
	return ntx < dev->real_num_tx_queues ?
			default_qdisc_ops : &pfifo_fast_ops;
}
381

E
Eric Dumazet 已提交
382
struct Qdisc_class_common {
383 384 385 386
	u32			classid;
	struct hlist_node	hnode;
};

E
Eric Dumazet 已提交
387
struct Qdisc_class_hash {
388 389 390 391 392 393 394 395 396 397 398 399 400 401
	struct hlist_head	*hash;
	unsigned int		hashsize;
	unsigned int		hashmask;
	unsigned int		hashelems;
};

static inline unsigned int qdisc_class_hash(u32 id, u32 mask)
{
	id ^= id >> 8;
	id ^= id >> 4;
	return id & mask;
}

static inline struct Qdisc_class_common *
402
qdisc_class_find(const struct Qdisc_class_hash *hash, u32 id)
403 404 405 406
{
	struct Qdisc_class_common *cl;
	unsigned int h;

407 408 409
	if (!id)
		return NULL;

410
	h = qdisc_class_hash(id, hash->hashmask);
411
	hlist_for_each_entry(cl, &hash->hash[h], hnode) {
412 413 414 415 416 417
		if (cl->classid == id)
			return cl;
	}
	return NULL;
}

418 419 420 421 422 423 424
static inline int tc_classid_to_hwtc(struct net_device *dev, u32 classid)
{
	u32 hwtc = TC_H_MIN(classid) - TC_H_MIN_PRIORITY;

	return (hwtc < netdev_get_num_tc(dev)) ? hwtc : -EINVAL;
}

425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441
int qdisc_class_hash_init(struct Qdisc_class_hash *);
void qdisc_class_hash_insert(struct Qdisc_class_hash *,
			     struct Qdisc_class_common *);
void qdisc_class_hash_remove(struct Qdisc_class_hash *,
			     struct Qdisc_class_common *);
void qdisc_class_hash_grow(struct Qdisc *, struct Qdisc_class_hash *);
void qdisc_class_hash_destroy(struct Qdisc_class_hash *);

void dev_init_scheduler(struct net_device *dev);
void dev_shutdown(struct net_device *dev);
void dev_activate(struct net_device *dev);
void dev_deactivate(struct net_device *dev);
void dev_deactivate_many(struct list_head *head);
struct Qdisc *dev_graft_qdisc(struct netdev_queue *dev_queue,
			      struct Qdisc *qdisc);
void qdisc_reset(struct Qdisc *qdisc);
void qdisc_destroy(struct Qdisc *qdisc);
442 443
void qdisc_tree_reduce_backlog(struct Qdisc *qdisc, unsigned int n,
			       unsigned int len);
444
struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
445
			  const struct Qdisc_ops *ops);
446
struct Qdisc *qdisc_create_dflt(struct netdev_queue *dev_queue,
447
				const struct Qdisc_ops *ops, u32 parentid);
448 449
void __qdisc_calculate_pkt_len(struct sk_buff *skb,
			       const struct qdisc_size_table *stab);
450
int skb_do_redirect(struct sk_buff *);
L
Linus Torvalds 已提交
451

452 453 454
static inline void skb_reset_tc(struct sk_buff *skb)
{
#ifdef CONFIG_NET_CLS_ACT
455
	skb->tc_redirected = 0;
456 457 458
#endif
}

459 460 461
static inline bool skb_at_tc_ingress(const struct sk_buff *skb)
{
#ifdef CONFIG_NET_CLS_ACT
462
	return skb->tc_at_ingress;
463 464 465 466 467
#else
	return false;
#endif
}

468 469 470 471 472 473 474 475 476 477 478
static inline bool skb_skip_tc_classify(struct sk_buff *skb)
{
#ifdef CONFIG_NET_CLS_ACT
	if (skb->tc_skip_classify) {
		skb->tc_skip_classify = 0;
		return true;
	}
#endif
	return false;
}

479 480
/* Reset all TX qdiscs greater then index of a device.  */
static inline void qdisc_reset_all_tx_gt(struct net_device *dev, unsigned int i)
481
{
482 483
	struct Qdisc *qdisc;

484
	for (; i < dev->num_tx_queues; i++) {
485
		qdisc = rtnl_dereference(netdev_get_tx_queue(dev, i)->qdisc);
486 487 488 489 490 491
		if (qdisc) {
			spin_lock_bh(qdisc_lock(qdisc));
			qdisc_reset(qdisc);
			spin_unlock_bh(qdisc_lock(qdisc));
		}
	}
492 493 494 495
}

static inline void qdisc_reset_all_tx(struct net_device *dev)
{
496
	qdisc_reset_all_tx_gt(dev, 0);
497 498
}

499 500 501
/* Are all TX queues of the device empty?  */
static inline bool qdisc_all_tx_empty(const struct net_device *dev)
{
502
	unsigned int i;
503 504

	rcu_read_lock();
505 506
	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
507
		const struct Qdisc *q = rcu_dereference(txq->qdisc);
508

509 510
		if (q->q.qlen) {
			rcu_read_unlock();
511
			return false;
512
		}
513
	}
514
	rcu_read_unlock();
515
	return true;
516 517
}

518
/* Are any of the TX qdiscs changing?  */
519
static inline bool qdisc_tx_changing(const struct net_device *dev)
520
{
521
	unsigned int i;
522

523 524
	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
525
		if (rcu_access_pointer(txq->qdisc) != txq->qdisc_sleeping)
526 527 528
			return true;
	}
	return false;
529 530
}

531
/* Is the device using the noop qdisc on all queues?  */
532 533
static inline bool qdisc_tx_is_noop(const struct net_device *dev)
{
534
	unsigned int i;
535

536 537
	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
538
		if (rcu_access_pointer(txq->qdisc) != &noop_qdisc)
539 540 541
			return false;
	}
	return true;
542 543
}

544
static inline unsigned int qdisc_pkt_len(const struct sk_buff *skb)
545
{
546
	return qdisc_skb_cb(skb)->pkt_len;
547 548
}

549
/* additional qdisc xmit flags (NET_XMIT_MASK in linux/netdevice.h) */
550 551
enum net_xmit_qdisc_t {
	__NET_XMIT_STOLEN = 0x00010000,
552
	__NET_XMIT_BYPASS = 0x00020000,
553 554
};

555
#ifdef CONFIG_NET_CLS_ACT
556 557 558 559 560
#define net_xmit_drop_count(e)	((e) & __NET_XMIT_STOLEN ? 0 : 1)
#else
#define net_xmit_drop_count(e)	(1)
#endif

E
Eric Dumazet 已提交
561 562
static inline void qdisc_calculate_pkt_len(struct sk_buff *skb,
					   const struct Qdisc *sch)
563
{
564
#ifdef CONFIG_NET_SCHED
E
Eric Dumazet 已提交
565 566 567 568
	struct qdisc_size_table *stab = rcu_dereference_bh(sch->stab);

	if (stab)
		__qdisc_calculate_pkt_len(skb, stab);
569
#endif
E
Eric Dumazet 已提交
570 571
}

572 573
static inline int qdisc_enqueue(struct sk_buff *skb, struct Qdisc *sch,
				struct sk_buff **to_free)
E
Eric Dumazet 已提交
574 575
{
	qdisc_calculate_pkt_len(skb, sch);
576
	return sch->enqueue(skb, sch, to_free);
577 578
}

579 580 581 582
static inline bool qdisc_is_percpu_stats(const struct Qdisc *q)
{
	return q->flags & TCQ_F_CPUSTATS;
}
583

584 585 586 587 588 589 590
static inline void _bstats_update(struct gnet_stats_basic_packed *bstats,
				  __u64 bytes, __u32 packets)
{
	bstats->bytes += bytes;
	bstats->packets += packets;
}

591 592 593
static inline void bstats_update(struct gnet_stats_basic_packed *bstats,
				 const struct sk_buff *skb)
{
594 595 596 597 598 599 600 601 602 603 604
	_bstats_update(bstats,
		       qdisc_pkt_len(skb),
		       skb_is_gso(skb) ? skb_shinfo(skb)->gso_segs : 1);
}

static inline void _bstats_cpu_update(struct gnet_stats_basic_cpu *bstats,
				      __u64 bytes, __u32 packets)
{
	u64_stats_update_begin(&bstats->syncp);
	_bstats_update(&bstats->bstats, bytes, packets);
	u64_stats_update_end(&bstats->syncp);
605 606
}

607 608
static inline void bstats_cpu_update(struct gnet_stats_basic_cpu *bstats,
				     const struct sk_buff *skb)
609 610 611 612 613 614
{
	u64_stats_update_begin(&bstats->syncp);
	bstats_update(&bstats->bstats, skb);
	u64_stats_update_end(&bstats->syncp);
}

615 616 617 618 619 620
static inline void qdisc_bstats_cpu_update(struct Qdisc *sch,
					   const struct sk_buff *skb)
{
	bstats_cpu_update(this_cpu_ptr(sch->cpu_bstats), skb);
}

621 622
static inline void qdisc_bstats_update(struct Qdisc *sch,
				       const struct sk_buff *skb)
623
{
624
	bstats_update(&sch->bstats, skb);
625 626
}

627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643
static inline void qdisc_qstats_backlog_dec(struct Qdisc *sch,
					    const struct sk_buff *skb)
{
	sch->qstats.backlog -= qdisc_pkt_len(skb);
}

static inline void qdisc_qstats_backlog_inc(struct Qdisc *sch,
					    const struct sk_buff *skb)
{
	sch->qstats.backlog += qdisc_pkt_len(skb);
}

static inline void __qdisc_qstats_drop(struct Qdisc *sch, int count)
{
	sch->qstats.drops += count;
}

644
static inline void qstats_drop_inc(struct gnet_stats_queue *qstats)
645
{
646
	qstats->drops++;
647 648
}

649
static inline void qstats_overlimit_inc(struct gnet_stats_queue *qstats)
650
{
651 652
	qstats->overlimits++;
}
653

654 655 656 657 658 659 660
static inline void qdisc_qstats_drop(struct Qdisc *sch)
{
	qstats_drop_inc(&sch->qstats);
}

static inline void qdisc_qstats_cpu_drop(struct Qdisc *sch)
{
661
	this_cpu_inc(sch->cpu_qstats->drops);
662 663
}

664 665 666 667 668
static inline void qdisc_qstats_overlimit(struct Qdisc *sch)
{
	sch->qstats.overlimits++;
}

669 670 671 672 673 674 675
static inline void qdisc_skb_head_init(struct qdisc_skb_head *qh)
{
	qh->head = NULL;
	qh->tail = NULL;
	qh->qlen = 0;
}

676
static inline int __qdisc_enqueue_tail(struct sk_buff *skb, struct Qdisc *sch,
677
				       struct qdisc_skb_head *qh)
678
{
679 680 681 682 683 684 685 686 687 688 689
	struct sk_buff *last = qh->tail;

	if (last) {
		skb->next = NULL;
		last->next = skb;
		qh->tail = skb;
	} else {
		qh->tail = skb;
		qh->head = skb;
	}
	qh->qlen++;
690
	qdisc_qstats_backlog_inc(sch, skb);
691 692 693 694 695 696 697 698 699

	return NET_XMIT_SUCCESS;
}

static inline int qdisc_enqueue_tail(struct sk_buff *skb, struct Qdisc *sch)
{
	return __qdisc_enqueue_tail(skb, sch, &sch->q);
}

700
static inline struct sk_buff *__qdisc_dequeue_head(struct qdisc_skb_head *qh)
701
{
702 703 704 705 706 707 708 709 710
	struct sk_buff *skb = qh->head;

	if (likely(skb != NULL)) {
		qh->head = skb->next;
		qh->qlen--;
		if (qh->head == NULL)
			qh->tail = NULL;
		skb->next = NULL;
	}
711

712 713 714 715 716 717 718
	return skb;
}

static inline struct sk_buff *qdisc_dequeue_head(struct Qdisc *sch)
{
	struct sk_buff *skb = __qdisc_dequeue_head(&sch->q);

719
	if (likely(skb != NULL)) {
720
		qdisc_qstats_backlog_dec(sch, skb);
721 722
		qdisc_bstats_update(sch, skb);
	}
723 724 725 726

	return skb;
}

727 728 729 730 731 732 733 734 735
/* Instead of calling kfree_skb() while root qdisc lock is held,
 * queue the skb for future freeing at end of __dev_xmit_skb()
 */
static inline void __qdisc_drop(struct sk_buff *skb, struct sk_buff **to_free)
{
	skb->next = *to_free;
	*to_free = skb;
}

736
static inline unsigned int __qdisc_queue_drop_head(struct Qdisc *sch,
737
						   struct qdisc_skb_head *qh,
738
						   struct sk_buff **to_free)
739
{
740
	struct sk_buff *skb = __qdisc_dequeue_head(qh);
741 742 743

	if (likely(skb != NULL)) {
		unsigned int len = qdisc_pkt_len(skb);
744

745
		qdisc_qstats_backlog_dec(sch, skb);
746
		__qdisc_drop(skb, to_free);
747 748 749 750 751 752
		return len;
	}

	return 0;
}

753 754
static inline unsigned int qdisc_queue_drop_head(struct Qdisc *sch,
						 struct sk_buff **to_free)
755
{
756
	return __qdisc_queue_drop_head(sch, &sch->q, to_free);
757 758
}

759 760
static inline struct sk_buff *qdisc_peek_head(struct Qdisc *sch)
{
761 762 763
	const struct qdisc_skb_head *qh = &sch->q;

	return qh->head;
764 765
}

766 767 768 769
/* generic pseudo peek method for non-work-conserving qdisc */
static inline struct sk_buff *qdisc_peek_dequeued(struct Qdisc *sch)
{
	/* we can reuse ->gso_skb because peek isn't called for root qdiscs */
770
	if (!sch->gso_skb) {
771
		sch->gso_skb = sch->dequeue(sch);
772
		if (sch->gso_skb) {
773
			/* it's still part of the queue */
774
			qdisc_qstats_backlog_inc(sch, sch->gso_skb);
775
			sch->q.qlen++;
776
		}
777
	}
778 779 780 781 782 783 784 785 786

	return sch->gso_skb;
}

/* use instead of qdisc->dequeue() for all qdiscs queried with ->peek() */
static inline struct sk_buff *qdisc_dequeue_peeked(struct Qdisc *sch)
{
	struct sk_buff *skb = sch->gso_skb;

787
	if (skb) {
788
		sch->gso_skb = NULL;
789
		qdisc_qstats_backlog_dec(sch, skb);
790 791
		sch->q.qlen--;
	} else {
792
		skb = sch->dequeue(sch);
793
	}
794 795 796 797

	return skb;
}

798
static inline void __qdisc_reset_queue(struct qdisc_skb_head *qh)
799 800 801 802 803
{
	/*
	 * We do not know the backlog in bytes of this list, it
	 * is up to the caller to correct it
	 */
804 805 806 807 808 809 810
	ASSERT_RTNL();
	if (qh->qlen) {
		rtnl_kfree_skbs(qh->head, qh->tail);

		qh->head = NULL;
		qh->tail = NULL;
		qh->qlen = 0;
811
	}
812 813 814 815
}

static inline void qdisc_reset_queue(struct Qdisc *sch)
{
816
	__qdisc_reset_queue(&sch->q);
817 818 819
	sch->qstats.backlog = 0;
}

820 821 822 823 824 825 826 827 828
static inline struct Qdisc *qdisc_replace(struct Qdisc *sch, struct Qdisc *new,
					  struct Qdisc **pold)
{
	struct Qdisc *old;

	sch_tree_lock(sch);
	old = *pold;
	*pold = new;
	if (old != NULL) {
829 830 831
		unsigned int qlen = old->q.qlen;
		unsigned int backlog = old->qstats.backlog;

832
		qdisc_reset(old);
833
		qdisc_tree_reduce_backlog(old, qlen, backlog);
834 835 836 837 838 839
	}
	sch_tree_unlock(sch);

	return old;
}

840 841 842 843 844 845
static inline void rtnl_qdisc_drop(struct sk_buff *skb, struct Qdisc *sch)
{
	rtnl_kfree_skbs(skb, skb);
	qdisc_qstats_drop(sch);
}

846 847 848

static inline int qdisc_drop(struct sk_buff *skb, struct Qdisc *sch,
			     struct sk_buff **to_free)
849
{
850
	__qdisc_drop(skb, to_free);
851
	qdisc_qstats_drop(sch);
852 853 854 855

	return NET_XMIT_DROP;
}

856 857 858 859 860
/* Length to Time (L2T) lookup in a qdisc_rate_table, to determine how
   long it will take to send a packet given its size.
 */
static inline u32 qdisc_l2t(struct qdisc_rate_table* rtab, unsigned int pktlen)
{
861 862 863
	int slot = pktlen + rtab->rate.cell_align + rtab->rate.overhead;
	if (slot < 0)
		slot = 0;
864 865
	slot >>= rtab->rate.cell_log;
	if (slot > 255)
E
Eric Dumazet 已提交
866
		return rtab->data[255]*(slot >> 8) + rtab->data[slot & 0xFF];
867 868 869
	return rtab->data[slot];
}

870
struct psched_ratecfg {
871
	u64	rate_bytes_ps; /* bytes per second */
872 873
	u32	mult;
	u16	overhead;
874
	u8	linklayer;
875
	u8	shift;
876 877 878 879 880
};

static inline u64 psched_l2t_ns(const struct psched_ratecfg *r,
				unsigned int len)
{
881 882 883 884 885 886
	len += r->overhead;

	if (unlikely(r->linklayer == TC_LINKLAYER_ATM))
		return ((u64)(DIV_ROUND_UP(len,48)*53) * r->mult) >> r->shift;

	return ((u64)len * r->mult) >> r->shift;
887 888
}

889
void psched_ratecfg_precompute(struct psched_ratecfg *r,
890 891
			       const struct tc_ratespec *conf,
			       u64 rate64);
892

893 894
static inline void psched_ratecfg_getrate(struct tc_ratespec *res,
					  const struct psched_ratecfg *r)
895
{
896
	memset(res, 0, sizeof(*res));
897 898 899 900 901 902 903

	/* legacy struct tc_ratespec has a 32bit @rate field
	 * Qdisc using 64bit rate should add new attributes
	 * in order to maintain compatibility.
	 */
	res->rate = min_t(u64, r->rate_bytes_ps, ~0U);

904
	res->overhead = r->overhead;
905
	res->linklayer = (r->linklayer & TC_LINKLAYER_MASK);
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
/* Mini Qdisc serves for specific needs of ingress/clsact Qdisc.
 * The fast path only needs to access filter list and to update stats
 */
struct mini_Qdisc {
	struct tcf_proto *filter_list;
	struct gnet_stats_basic_cpu __percpu *cpu_bstats;
	struct gnet_stats_queue	__percpu *cpu_qstats;
	struct rcu_head rcu;
};

static inline void mini_qdisc_bstats_cpu_update(struct mini_Qdisc *miniq,
						const struct sk_buff *skb)
{
	bstats_cpu_update(this_cpu_ptr(miniq->cpu_bstats), skb);
}

static inline void mini_qdisc_qstats_cpu_drop(struct mini_Qdisc *miniq)
{
	this_cpu_inc(miniq->cpu_qstats->drops);
}

struct mini_Qdisc_pair {
	struct mini_Qdisc miniq1;
	struct mini_Qdisc miniq2;
	struct mini_Qdisc __rcu **p_miniq;
};

void mini_qdisc_pair_swap(struct mini_Qdisc_pair *miniqp,
			  struct tcf_proto *tp_head);
void mini_qdisc_pair_init(struct mini_Qdisc_pair *miniqp, struct Qdisc *qdisc,
			  struct mini_Qdisc __rcu **p_miniq);

L
Linus Torvalds 已提交
940
#endif