sch_generic.h 21.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8
#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>
9
#include <linux/percpu.h>
10
#include <linux/dynamic_queue_limits.h>
11
#include <linux/list.h>
12
#include <linux/refcount.h>
L
Linus Torvalds 已提交
13
#include <net/gen_stats.h>
14
#include <net/rtnetlink.h>
L
Linus Torvalds 已提交
15 16 17 18 19 20

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

E
Eric Dumazet 已提交
21
struct qdisc_rate_table {
L
Linus Torvalds 已提交
22 23 24 25 26 27
	struct tc_ratespec rate;
	u32		data[256];
	struct qdisc_rate_table *next;
	int		refcnt;
};

E
Eric Dumazet 已提交
28
enum qdisc_state_t {
29
	__QDISC_STATE_SCHED,
30
	__QDISC_STATE_DEACTIVATED,
31 32
};

33
struct qdisc_size_table {
E
Eric Dumazet 已提交
34
	struct rcu_head		rcu;
35 36 37 38 39 40
	struct list_head	list;
	struct tc_sizespec	szopts;
	int			refcnt;
	u16			data[];
};

41 42 43 44 45 46 47 48
/* 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
49
struct Qdisc {
50 51 52 53
	int 			(*enqueue)(struct sk_buff *skb,
					   struct Qdisc *sch,
					   struct sk_buff **to_free);
	struct sk_buff *	(*dequeue)(struct Qdisc *sch);
54
	unsigned int		flags;
55
#define TCQ_F_BUILTIN		1
56 57 58
#define TCQ_F_INGRESS		2
#define TCQ_F_CAN_BYPASS	4
#define TCQ_F_MQROOT		8
59 60 61 62 63 64 65
#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.
				      */
66
#define TCQ_F_WARN_NONWC	(1 << 16)
67
#define TCQ_F_CPUSTATS		0x20 /* run using percpu statistics */
68 69 70
#define TCQ_F_NOPARENT		0x40 /* root of its hierarchy :
				      * qdisc_tree_decrease_qlen() should stop.
				      */
71
#define TCQ_F_INVISIBLE		0x80 /* invisible by default in dump */
72
	u32			limit;
73
	const struct Qdisc_ops	*ops;
E
Eric Dumazet 已提交
74
	struct qdisc_size_table	__rcu *stab;
75
	struct hlist_node       hash;
L
Linus Torvalds 已提交
76 77
	u32			handle;
	u32			parent;
78

79 80
	struct netdev_queue	*dev_queue;

81
	struct net_rate_estimator __rcu *rate_est;
82 83 84
	struct gnet_stats_basic_cpu __percpu *cpu_bstats;
	struct gnet_stats_queue	__percpu *cpu_qstats;

85 86 87
	/*
	 * For performance sake on SMP, we put highly modified fields at the end
	 */
88
	struct sk_buff		*gso_skb ____cacheline_aligned_in_smp;
89
	struct qdisc_skb_head	q;
90
	struct gnet_stats_basic_packed bstats;
91
	seqcount_t		running;
92
	struct gnet_stats_queue	qstats;
93 94 95
	unsigned long		state;
	struct Qdisc            *next_sched;
	struct sk_buff		*skb_bad_txq;
96
	int			padded;
97
	refcount_t		refcnt;
98 99

	spinlock_t		busylock ____cacheline_aligned_in_smp;
L
Linus Torvalds 已提交
100 101
};

102 103 104 105 106 107 108
static inline void qdisc_refcount_inc(struct Qdisc *qdisc)
{
	if (qdisc->flags & TCQ_F_BUILTIN)
		return;
	refcount_inc(&qdisc->refcnt);
}

109
static inline bool qdisc_is_running(const struct Qdisc *qdisc)
110
{
111
	return (raw_read_seqcount(&qdisc->running) & 1) ? true : false;
112 113 114 115
}

static inline bool qdisc_run_begin(struct Qdisc *qdisc)
{
116 117
	if (qdisc_is_running(qdisc))
		return false;
118 119 120 121 122
	/* 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_);
123
	return true;
124 125 126 127
}

static inline void qdisc_run_end(struct Qdisc *qdisc)
{
128
	write_seqcount_end(&qdisc->running);
129 130
}

131 132 133 134 135 136 137 138 139 140 141 142 143 144 145
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
146
struct Qdisc_class_ops {
L
Linus Torvalds 已提交
147
	/* Child qdisc manipulation */
148
	struct netdev_queue *	(*select_queue)(struct Qdisc *, struct tcmsg *);
L
Linus Torvalds 已提交
149 150 151
	int			(*graft)(struct Qdisc *, unsigned long cl,
					struct Qdisc *, struct Qdisc **);
	struct Qdisc *		(*leaf)(struct Qdisc *, unsigned long cl);
152
	void			(*qlen_notify)(struct Qdisc *, unsigned long);
L
Linus Torvalds 已提交
153 154

	/* Class manipulation routines */
155
	unsigned long		(*find)(struct Qdisc *, u32 classid);
L
Linus Torvalds 已提交
156
	int			(*change)(struct Qdisc *, u32, u32,
157
					struct nlattr **, unsigned long *);
L
Linus Torvalds 已提交
158 159 160 161
	int			(*delete)(struct Qdisc *, unsigned long);
	void			(*walk)(struct Qdisc *, struct qdisc_walker * arg);

	/* Filter manipulation */
162
	struct tcf_block *	(*tcf_block)(struct Qdisc *, unsigned long);
L
Linus Torvalds 已提交
163 164 165 166 167 168 169 170 171 172 173
	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
174
struct Qdisc_ops {
L
Linus Torvalds 已提交
175
	struct Qdisc_ops	*next;
176
	const struct Qdisc_class_ops	*cl_ops;
L
Linus Torvalds 已提交
177 178 179
	char			id[IFNAMSIZ];
	int			priv_size;

180 181 182
	int 			(*enqueue)(struct sk_buff *skb,
					   struct Qdisc *sch,
					   struct sk_buff **to_free);
L
Linus Torvalds 已提交
183
	struct sk_buff *	(*dequeue)(struct Qdisc *);
184
	struct sk_buff *	(*peek)(struct Qdisc *);
L
Linus Torvalds 已提交
185

186
	int			(*init)(struct Qdisc *, struct nlattr *arg);
L
Linus Torvalds 已提交
187 188
	void			(*reset)(struct Qdisc *);
	void			(*destroy)(struct Qdisc *);
189
	int			(*change)(struct Qdisc *, struct nlattr *arg);
190
	void			(*attach)(struct Qdisc *);
L
Linus Torvalds 已提交
191 192 193 194 195 196 197 198

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

	struct module		*owner;
};


E
Eric Dumazet 已提交
199
struct tcf_result {
200 201 202 203 204 205 206
	union {
		struct {
			unsigned long	class;
			u32		classid;
		};
		const struct tcf_proto *goto_tp;
	};
L
Linus Torvalds 已提交
207 208
};

E
Eric Dumazet 已提交
209
struct tcf_proto_ops {
210
	struct list_head	head;
L
Linus Torvalds 已提交
211 212
	char			kind[IFNAMSIZ];

213 214 215
	int			(*classify)(struct sk_buff *,
					    const struct tcf_proto *,
					    struct tcf_result *);
L
Linus Torvalds 已提交
216
	int			(*init)(struct tcf_proto*);
217
	void			(*destroy)(struct tcf_proto*);
L
Linus Torvalds 已提交
218

219
	void*			(*get)(struct tcf_proto*, u32 handle);
220
	int			(*change)(struct net *net, struct sk_buff *,
221
					struct tcf_proto*, unsigned long,
222
					u32 handle, struct nlattr **,
223 224
					void **, bool);
	int			(*delete)(struct tcf_proto*, void *, bool*);
L
Linus Torvalds 已提交
225
	void			(*walk)(struct tcf_proto*, struct tcf_walker *arg);
226
	void			(*bind_class)(void *, u32, unsigned long);
L
Linus Torvalds 已提交
227 228

	/* rtnetlink specific */
229
	int			(*dump)(struct net*, struct tcf_proto*, void *,
L
Linus Torvalds 已提交
230 231 232 233 234
					struct sk_buff *skb, struct tcmsg*);

	struct module		*owner;
};

E
Eric Dumazet 已提交
235
struct tcf_proto {
L
Linus Torvalds 已提交
236
	/* Fast access part */
J
John Fastabend 已提交
237 238
	struct tcf_proto __rcu	*next;
	void __rcu		*root;
239 240 241
	int			(*classify)(struct sk_buff *,
					    const struct tcf_proto *,
					    struct tcf_result *);
A
Al Viro 已提交
242
	__be16			protocol;
L
Linus Torvalds 已提交
243 244 245 246 247 248

	/* All the rest */
	u32			prio;
	u32			classid;
	struct Qdisc		*q;
	void			*data;
249
	const struct tcf_proto_ops	*ops;
250
	struct tcf_chain	*chain;
J
John Fastabend 已提交
251
	struct rcu_head		rcu;
L
Linus Torvalds 已提交
252 253
};

254 255
struct qdisc_skb_cb {
	unsigned int		pkt_len;
256
	u16			slave_dev_queue_mapping;
257
	u16			tc_classid;
258 259
#define QDISC_CB_PRIV_LEN 20
	unsigned char		data[QDISC_CB_PRIV_LEN];
260 261
};

262 263
struct tcf_chain {
	struct tcf_proto __rcu *filter_chain;
264
	struct tcf_proto __rcu **p_filter_chain;
265 266 267 268
	struct list_head list;
	struct tcf_block *block;
	u32 index; /* chain index */
	unsigned int refcnt;
269 270
};

271
struct tcf_block {
272
	struct list_head chain_list;
273 274
};

275 276 277
static inline void qdisc_cb_private_validate(const struct sk_buff *skb, int sz)
{
	struct qdisc_skb_cb *qcb;
278 279

	BUILD_BUG_ON(sizeof(skb->cb) < offsetof(struct qdisc_skb_cb, data) + sz);
280 281 282
	BUILD_BUG_ON(sizeof(qcb->data) < sz);
}

283
static inline int qdisc_qlen(const struct Qdisc *q)
284 285 286 287
{
	return q->q.qlen;
}

288
static inline struct qdisc_skb_cb *qdisc_skb_cb(const struct sk_buff *skb)
289 290 291 292
{
	return (struct qdisc_skb_cb *)skb->cb;
}

293 294 295 296 297
static inline spinlock_t *qdisc_lock(struct Qdisc *qdisc)
{
	return &qdisc->q.lock;
}

298
static inline struct Qdisc *qdisc_root(const struct Qdisc *qdisc)
299
{
300 301 302
	struct Qdisc *q = rcu_dereference_rtnl(qdisc->dev_queue->qdisc);

	return q;
303 304
}

305
static inline struct Qdisc *qdisc_root_sleeping(const struct Qdisc *qdisc)
306 307 308 309
{
	return qdisc->dev_queue->qdisc_sleeping;
}

310 311 312 313 314 315 316 317 318 319 320
/* 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.
 */
321
static inline spinlock_t *qdisc_root_lock(const struct Qdisc *qdisc)
322 323 324
{
	struct Qdisc *root = qdisc_root(qdisc);

325
	ASSERT_RTNL();
326
	return qdisc_lock(root);
327 328
}

329
static inline spinlock_t *qdisc_root_sleeping_lock(const struct Qdisc *qdisc)
330 331 332 333 334 335 336
{
	struct Qdisc *root = qdisc_root_sleeping(qdisc);

	ASSERT_RTNL();
	return qdisc_lock(root);
}

337 338 339 340 341 342 343 344
static inline seqcount_t *qdisc_root_sleeping_running(const struct Qdisc *qdisc)
{
	struct Qdisc *root = qdisc_root_sleeping(qdisc);

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

345
static inline struct net_device *qdisc_dev(const struct Qdisc *qdisc)
346 347 348
{
	return qdisc->dev_queue->dev;
}
L
Linus Torvalds 已提交
349

350
static inline void sch_tree_lock(const struct Qdisc *q)
351
{
J
Jarek Poplawski 已提交
352
	spin_lock_bh(qdisc_root_sleeping_lock(q));
353 354
}

355
static inline void sch_tree_unlock(const struct Qdisc *q)
356
{
J
Jarek Poplawski 已提交
357
	spin_unlock_bh(qdisc_root_sleeping_lock(q));
358 359 360 361
}

#define tcf_tree_lock(tp)	sch_tree_lock((tp)->q)
#define tcf_tree_unlock(tp)	sch_tree_unlock((tp)->q)
L
Linus Torvalds 已提交
362

363 364
extern struct Qdisc noop_qdisc;
extern struct Qdisc_ops noop_qdisc_ops;
365 366
extern struct Qdisc_ops pfifo_fast_ops;
extern struct Qdisc_ops mq_qdisc_ops;
P
Phil Sutter 已提交
367
extern struct Qdisc_ops noqueue_qdisc_ops;
368
extern const struct Qdisc_ops *default_qdisc_ops;
369 370 371 372 373 374
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;
}
375

E
Eric Dumazet 已提交
376
struct Qdisc_class_common {
377 378 379 380
	u32			classid;
	struct hlist_node	hnode;
};

E
Eric Dumazet 已提交
381
struct Qdisc_class_hash {
382 383 384 385 386 387 388 389 390 391 392 393 394 395
	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 *
396
qdisc_class_find(const struct Qdisc_class_hash *hash, u32 id)
397 398 399 400
{
	struct Qdisc_class_common *cl;
	unsigned int h;

401 402 403
	if (!id)
		return NULL;

404
	h = qdisc_class_hash(id, hash->hashmask);
405
	hlist_for_each_entry(cl, &hash->hash[h], hnode) {
406 407 408 409 410 411
		if (cl->classid == id)
			return cl;
	}
	return NULL;
}

412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428
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);
429 430
void qdisc_tree_reduce_backlog(struct Qdisc *qdisc, unsigned int n,
			       unsigned int len);
431
struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
432
			  const struct Qdisc_ops *ops);
433
struct Qdisc *qdisc_create_dflt(struct netdev_queue *dev_queue,
434
				const struct Qdisc_ops *ops, u32 parentid);
435 436
void __qdisc_calculate_pkt_len(struct sk_buff *skb,
			       const struct qdisc_size_table *stab);
437
int skb_do_redirect(struct sk_buff *);
L
Linus Torvalds 已提交
438

439 440 441
static inline void skb_reset_tc(struct sk_buff *skb)
{
#ifdef CONFIG_NET_CLS_ACT
442
	skb->tc_redirected = 0;
443 444 445
#endif
}

446 447 448
static inline bool skb_at_tc_ingress(const struct sk_buff *skb)
{
#ifdef CONFIG_NET_CLS_ACT
449
	return skb->tc_at_ingress;
450 451 452 453 454
#else
	return false;
#endif
}

455 456 457 458 459 460 461 462 463 464 465
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;
}

466 467
/* 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)
468
{
469 470
	struct Qdisc *qdisc;

471
	for (; i < dev->num_tx_queues; i++) {
472
		qdisc = rtnl_dereference(netdev_get_tx_queue(dev, i)->qdisc);
473 474 475 476 477 478
		if (qdisc) {
			spin_lock_bh(qdisc_lock(qdisc));
			qdisc_reset(qdisc);
			spin_unlock_bh(qdisc_lock(qdisc));
		}
	}
479 480 481 482
}

static inline void qdisc_reset_all_tx(struct net_device *dev)
{
483
	qdisc_reset_all_tx_gt(dev, 0);
484 485
}

486 487 488
/* Are all TX queues of the device empty?  */
static inline bool qdisc_all_tx_empty(const struct net_device *dev)
{
489
	unsigned int i;
490 491

	rcu_read_lock();
492 493
	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
494
		const struct Qdisc *q = rcu_dereference(txq->qdisc);
495

496 497
		if (q->q.qlen) {
			rcu_read_unlock();
498
			return false;
499
		}
500
	}
501
	rcu_read_unlock();
502
	return true;
503 504
}

505
/* Are any of the TX qdiscs changing?  */
506
static inline bool qdisc_tx_changing(const struct net_device *dev)
507
{
508
	unsigned int i;
509

510 511
	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
512
		if (rcu_access_pointer(txq->qdisc) != txq->qdisc_sleeping)
513 514 515
			return true;
	}
	return false;
516 517
}

518
/* Is the device using the noop qdisc on all queues?  */
519 520
static inline bool qdisc_tx_is_noop(const struct net_device *dev)
{
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) != &noop_qdisc)
526 527 528
			return false;
	}
	return true;
529 530
}

531
static inline unsigned int qdisc_pkt_len(const struct sk_buff *skb)
532
{
533
	return qdisc_skb_cb(skb)->pkt_len;
534 535
}

536
/* additional qdisc xmit flags (NET_XMIT_MASK in linux/netdevice.h) */
537 538
enum net_xmit_qdisc_t {
	__NET_XMIT_STOLEN = 0x00010000,
539
	__NET_XMIT_BYPASS = 0x00020000,
540 541
};

542
#ifdef CONFIG_NET_CLS_ACT
543 544 545 546 547
#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
548 549
static inline void qdisc_calculate_pkt_len(struct sk_buff *skb,
					   const struct Qdisc *sch)
550
{
551
#ifdef CONFIG_NET_SCHED
E
Eric Dumazet 已提交
552 553 554 555
	struct qdisc_size_table *stab = rcu_dereference_bh(sch->stab);

	if (stab)
		__qdisc_calculate_pkt_len(skb, stab);
556
#endif
E
Eric Dumazet 已提交
557 558
}

559 560
static inline int qdisc_enqueue(struct sk_buff *skb, struct Qdisc *sch,
				struct sk_buff **to_free)
E
Eric Dumazet 已提交
561 562
{
	qdisc_calculate_pkt_len(skb, sch);
563
	return sch->enqueue(skb, sch, to_free);
564 565
}

566 567 568 569
static inline bool qdisc_is_percpu_stats(const struct Qdisc *q)
{
	return q->flags & TCQ_F_CPUSTATS;
}
570

571 572 573 574 575 576 577
static inline void _bstats_update(struct gnet_stats_basic_packed *bstats,
				  __u64 bytes, __u32 packets)
{
	bstats->bytes += bytes;
	bstats->packets += packets;
}

578 579 580
static inline void bstats_update(struct gnet_stats_basic_packed *bstats,
				 const struct sk_buff *skb)
{
581 582 583 584 585 586 587 588 589 590 591
	_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);
592 593
}

594 595
static inline void bstats_cpu_update(struct gnet_stats_basic_cpu *bstats,
				     const struct sk_buff *skb)
596 597 598 599 600 601
{
	u64_stats_update_begin(&bstats->syncp);
	bstats_update(&bstats->bstats, skb);
	u64_stats_update_end(&bstats->syncp);
}

602 603 604 605 606 607
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);
}

608 609
static inline void qdisc_bstats_update(struct Qdisc *sch,
				       const struct sk_buff *skb)
610
{
611
	bstats_update(&sch->bstats, skb);
612 613
}

614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630
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;
}

631
static inline void qstats_drop_inc(struct gnet_stats_queue *qstats)
632
{
633
	qstats->drops++;
634 635
}

636
static inline void qstats_overlimit_inc(struct gnet_stats_queue *qstats)
637
{
638 639
	qstats->overlimits++;
}
640

641 642 643 644 645 646 647
static inline void qdisc_qstats_drop(struct Qdisc *sch)
{
	qstats_drop_inc(&sch->qstats);
}

static inline void qdisc_qstats_cpu_drop(struct Qdisc *sch)
{
648
	this_cpu_inc(sch->cpu_qstats->drops);
649 650
}

651 652 653 654 655
static inline void qdisc_qstats_overlimit(struct Qdisc *sch)
{
	sch->qstats.overlimits++;
}

656 657 658 659 660 661 662
static inline void qdisc_skb_head_init(struct qdisc_skb_head *qh)
{
	qh->head = NULL;
	qh->tail = NULL;
	qh->qlen = 0;
}

663
static inline int __qdisc_enqueue_tail(struct sk_buff *skb, struct Qdisc *sch,
664
				       struct qdisc_skb_head *qh)
665
{
666 667 668 669 670 671 672 673 674 675 676
	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++;
677
	qdisc_qstats_backlog_inc(sch, skb);
678 679 680 681 682 683 684 685 686

	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);
}

687
static inline struct sk_buff *__qdisc_dequeue_head(struct qdisc_skb_head *qh)
688
{
689 690 691 692 693 694 695 696 697
	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;
	}
698

699 700 701 702 703 704 705
	return skb;
}

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

706
	if (likely(skb != NULL)) {
707
		qdisc_qstats_backlog_dec(sch, skb);
708 709
		qdisc_bstats_update(sch, skb);
	}
710 711 712 713

	return skb;
}

714 715 716 717 718 719 720 721 722
/* 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;
}

723
static inline unsigned int __qdisc_queue_drop_head(struct Qdisc *sch,
724
						   struct qdisc_skb_head *qh,
725
						   struct sk_buff **to_free)
726
{
727
	struct sk_buff *skb = __qdisc_dequeue_head(qh);
728 729 730

	if (likely(skb != NULL)) {
		unsigned int len = qdisc_pkt_len(skb);
731

732
		qdisc_qstats_backlog_dec(sch, skb);
733
		__qdisc_drop(skb, to_free);
734 735 736 737 738 739
		return len;
	}

	return 0;
}

740 741
static inline unsigned int qdisc_queue_drop_head(struct Qdisc *sch,
						 struct sk_buff **to_free)
742
{
743
	return __qdisc_queue_drop_head(sch, &sch->q, to_free);
744 745
}

746 747
static inline struct sk_buff *qdisc_peek_head(struct Qdisc *sch)
{
748 749 750
	const struct qdisc_skb_head *qh = &sch->q;

	return qh->head;
751 752
}

753 754 755 756
/* 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 */
757
	if (!sch->gso_skb) {
758
		sch->gso_skb = sch->dequeue(sch);
759
		if (sch->gso_skb) {
760
			/* it's still part of the queue */
761
			qdisc_qstats_backlog_inc(sch, sch->gso_skb);
762
			sch->q.qlen++;
763
		}
764
	}
765 766 767 768 769 770 771 772 773

	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;

774
	if (skb) {
775
		sch->gso_skb = NULL;
776
		qdisc_qstats_backlog_dec(sch, skb);
777 778
		sch->q.qlen--;
	} else {
779
		skb = sch->dequeue(sch);
780
	}
781 782 783 784

	return skb;
}

785
static inline void __qdisc_reset_queue(struct qdisc_skb_head *qh)
786 787 788 789 790
{
	/*
	 * We do not know the backlog in bytes of this list, it
	 * is up to the caller to correct it
	 */
791 792 793 794 795 796 797
	ASSERT_RTNL();
	if (qh->qlen) {
		rtnl_kfree_skbs(qh->head, qh->tail);

		qh->head = NULL;
		qh->tail = NULL;
		qh->qlen = 0;
798
	}
799 800 801 802
}

static inline void qdisc_reset_queue(struct Qdisc *sch)
{
803
	__qdisc_reset_queue(&sch->q);
804 805 806
	sch->qstats.backlog = 0;
}

807 808 809 810 811 812 813 814 815
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) {
816 817 818
		unsigned int qlen = old->q.qlen;
		unsigned int backlog = old->qstats.backlog;

819
		qdisc_reset(old);
820
		qdisc_tree_reduce_backlog(old, qlen, backlog);
821 822 823 824 825 826
	}
	sch_tree_unlock(sch);

	return old;
}

827 828 829 830 831 832
static inline void rtnl_qdisc_drop(struct sk_buff *skb, struct Qdisc *sch)
{
	rtnl_kfree_skbs(skb, skb);
	qdisc_qstats_drop(sch);
}

833 834 835

static inline int qdisc_drop(struct sk_buff *skb, struct Qdisc *sch,
			     struct sk_buff **to_free)
836
{
837
	__qdisc_drop(skb, to_free);
838
	qdisc_qstats_drop(sch);
839 840 841 842

	return NET_XMIT_DROP;
}

843 844 845 846 847
/* 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)
{
848 849 850
	int slot = pktlen + rtab->rate.cell_align + rtab->rate.overhead;
	if (slot < 0)
		slot = 0;
851 852
	slot >>= rtab->rate.cell_log;
	if (slot > 255)
E
Eric Dumazet 已提交
853
		return rtab->data[255]*(slot >> 8) + rtab->data[slot & 0xFF];
854 855 856
	return rtab->data[slot];
}

857
struct psched_ratecfg {
858
	u64	rate_bytes_ps; /* bytes per second */
859 860
	u32	mult;
	u16	overhead;
861
	u8	linklayer;
862
	u8	shift;
863 864 865 866 867
};

static inline u64 psched_l2t_ns(const struct psched_ratecfg *r,
				unsigned int len)
{
868 869 870 871 872 873
	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;
874 875
}

876
void psched_ratecfg_precompute(struct psched_ratecfg *r,
877 878
			       const struct tc_ratespec *conf,
			       u64 rate64);
879

880 881
static inline void psched_ratecfg_getrate(struct tc_ratespec *res,
					  const struct psched_ratecfg *r)
882
{
883
	memset(res, 0, sizeof(*res));
884 885 886 887 888 889 890

	/* 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);

891
	res->overhead = r->overhead;
892
	res->linklayer = (r->linklayer & TC_LINKLAYER_MASK);
893 894
}

L
Linus Torvalds 已提交
895
#endif