sch_generic.h 20.0 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>
L
Linus Torvalds 已提交
11
#include <net/gen_stats.h>
12
#include <net/rtnetlink.h>
L
Linus Torvalds 已提交
13 14 15 16 17 18

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

E
Eric Dumazet 已提交
19
struct qdisc_rate_table {
L
Linus Torvalds 已提交
20 21 22 23 24 25
	struct tc_ratespec rate;
	u32		data[256];
	struct qdisc_rate_table *next;
	int		refcnt;
};

E
Eric Dumazet 已提交
26
enum qdisc_state_t {
27
	__QDISC_STATE_SCHED,
28
	__QDISC_STATE_DEACTIVATED,
29 30
};

31
struct qdisc_size_table {
E
Eric Dumazet 已提交
32
	struct rcu_head		rcu;
33 34 35 36 37 38
	struct list_head	list;
	struct tc_sizespec	szopts;
	int			refcnt;
	u16			data[];
};

E
Eric Dumazet 已提交
39
struct Qdisc {
40 41 42 43
	int 			(*enqueue)(struct sk_buff *skb,
					   struct Qdisc *sch,
					   struct sk_buff **to_free);
	struct sk_buff *	(*dequeue)(struct Qdisc *sch);
44
	unsigned int		flags;
45
#define TCQ_F_BUILTIN		1
46 47 48
#define TCQ_F_INGRESS		2
#define TCQ_F_CAN_BYPASS	4
#define TCQ_F_MQROOT		8
49 50 51 52 53 54 55
#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.
				      */
56
#define TCQ_F_WARN_NONWC	(1 << 16)
57
#define TCQ_F_CPUSTATS		0x20 /* run using percpu statistics */
58 59 60
#define TCQ_F_NOPARENT		0x40 /* root of its hierarchy :
				      * qdisc_tree_decrease_qlen() should stop.
				      */
61
	u32			limit;
62
	const struct Qdisc_ops	*ops;
E
Eric Dumazet 已提交
63
	struct qdisc_size_table	__rcu *stab;
64
	struct hlist_node       hash;
L
Linus Torvalds 已提交
65 66
	u32			handle;
	u32			parent;
67 68
	void			*u32_node;

69 70
	struct netdev_queue	*dev_queue;

71
	struct gnet_stats_rate_est64	rate_est;
72 73 74
	struct gnet_stats_basic_cpu __percpu *cpu_bstats;
	struct gnet_stats_queue	__percpu *cpu_qstats;

75 76 77
	/*
	 * For performance sake on SMP, we put highly modified fields at the end
	 */
78
	struct sk_buff		*gso_skb ____cacheline_aligned_in_smp;
79
	struct sk_buff_head	q;
80
	struct gnet_stats_basic_packed bstats;
81
	seqcount_t		running;
82
	struct gnet_stats_queue	qstats;
83 84 85
	unsigned long		state;
	struct Qdisc            *next_sched;
	struct sk_buff		*skb_bad_txq;
86
	struct rcu_head		rcu_head;
87 88 89 90
	int			padded;
	atomic_t		refcnt;

	spinlock_t		busylock ____cacheline_aligned_in_smp;
L
Linus Torvalds 已提交
91 92
};

93
static inline bool qdisc_is_running(const struct Qdisc *qdisc)
94
{
95
	return (raw_read_seqcount(&qdisc->running) & 1) ? true : false;
96 97 98 99
}

static inline bool qdisc_run_begin(struct Qdisc *qdisc)
{
100 101
	if (qdisc_is_running(qdisc))
		return false;
102 103 104 105 106
	/* 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_);
107
	return true;
108 109 110 111
}

static inline void qdisc_run_end(struct Qdisc *qdisc)
{
112
	write_seqcount_end(&qdisc->running);
113 114
}

115 116 117 118 119 120 121 122 123 124 125 126 127 128 129
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
130
struct Qdisc_class_ops {
L
Linus Torvalds 已提交
131
	/* Child qdisc manipulation */
132
	struct netdev_queue *	(*select_queue)(struct Qdisc *, struct tcmsg *);
L
Linus Torvalds 已提交
133 134 135
	int			(*graft)(struct Qdisc *, unsigned long cl,
					struct Qdisc *, struct Qdisc **);
	struct Qdisc *		(*leaf)(struct Qdisc *, unsigned long cl);
136
	void			(*qlen_notify)(struct Qdisc *, unsigned long);
L
Linus Torvalds 已提交
137 138 139 140 141

	/* Class manipulation routines */
	unsigned long		(*get)(struct Qdisc *, u32 classid);
	void			(*put)(struct Qdisc *, unsigned long);
	int			(*change)(struct Qdisc *, u32, u32,
142
					struct nlattr **, unsigned long *);
L
Linus Torvalds 已提交
143 144 145 146
	int			(*delete)(struct Qdisc *, unsigned long);
	void			(*walk)(struct Qdisc *, struct qdisc_walker * arg);

	/* Filter manipulation */
J
John Fastabend 已提交
147
	struct tcf_proto __rcu ** (*tcf_chain)(struct Qdisc *, unsigned long);
148
	bool			(*tcf_cl_offload)(u32 classid);
L
Linus Torvalds 已提交
149 150 151 152 153 154 155 156 157 158 159
	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
160
struct Qdisc_ops {
L
Linus Torvalds 已提交
161
	struct Qdisc_ops	*next;
162
	const struct Qdisc_class_ops	*cl_ops;
L
Linus Torvalds 已提交
163 164 165
	char			id[IFNAMSIZ];
	int			priv_size;

166 167 168
	int 			(*enqueue)(struct sk_buff *skb,
					   struct Qdisc *sch,
					   struct sk_buff **to_free);
L
Linus Torvalds 已提交
169
	struct sk_buff *	(*dequeue)(struct Qdisc *);
170
	struct sk_buff *	(*peek)(struct Qdisc *);
L
Linus Torvalds 已提交
171

172
	int			(*init)(struct Qdisc *, struct nlattr *arg);
L
Linus Torvalds 已提交
173 174
	void			(*reset)(struct Qdisc *);
	void			(*destroy)(struct Qdisc *);
175
	int			(*change)(struct Qdisc *, struct nlattr *arg);
176
	void			(*attach)(struct Qdisc *);
L
Linus Torvalds 已提交
177 178 179 180 181 182 183 184

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

	struct module		*owner;
};


E
Eric Dumazet 已提交
185
struct tcf_result {
L
Linus Torvalds 已提交
186 187 188 189
	unsigned long	class;
	u32		classid;
};

E
Eric Dumazet 已提交
190
struct tcf_proto_ops {
191
	struct list_head	head;
L
Linus Torvalds 已提交
192 193
	char			kind[IFNAMSIZ];

194 195 196
	int			(*classify)(struct sk_buff *,
					    const struct tcf_proto *,
					    struct tcf_result *);
L
Linus Torvalds 已提交
197
	int			(*init)(struct tcf_proto*);
198
	bool			(*destroy)(struct tcf_proto*, bool);
L
Linus Torvalds 已提交
199 200

	unsigned long		(*get)(struct tcf_proto*, u32 handle);
201
	int			(*change)(struct net *net, struct sk_buff *,
202
					struct tcf_proto*, unsigned long,
203
					u32 handle, struct nlattr **,
204
					unsigned long *, bool);
L
Linus Torvalds 已提交
205 206 207 208
	int			(*delete)(struct tcf_proto*, unsigned long);
	void			(*walk)(struct tcf_proto*, struct tcf_walker *arg);

	/* rtnetlink specific */
209
	int			(*dump)(struct net*, struct tcf_proto*, unsigned long,
L
Linus Torvalds 已提交
210 211 212 213 214
					struct sk_buff *skb, struct tcmsg*);

	struct module		*owner;
};

E
Eric Dumazet 已提交
215
struct tcf_proto {
L
Linus Torvalds 已提交
216
	/* Fast access part */
J
John Fastabend 已提交
217 218
	struct tcf_proto __rcu	*next;
	void __rcu		*root;
219 220 221
	int			(*classify)(struct sk_buff *,
					    const struct tcf_proto *,
					    struct tcf_result *);
A
Al Viro 已提交
222
	__be16			protocol;
L
Linus Torvalds 已提交
223 224 225 226 227 228

	/* All the rest */
	u32			prio;
	u32			classid;
	struct Qdisc		*q;
	void			*data;
229
	const struct tcf_proto_ops	*ops;
J
John Fastabend 已提交
230
	struct rcu_head		rcu;
L
Linus Torvalds 已提交
231 232
};

233 234
struct qdisc_skb_cb {
	unsigned int		pkt_len;
235
	u16			slave_dev_queue_mapping;
236
	u16			tc_classid;
237 238
#define QDISC_CB_PRIV_LEN 20
	unsigned char		data[QDISC_CB_PRIV_LEN];
239 240
};

241 242 243
static inline void qdisc_cb_private_validate(const struct sk_buff *skb, int sz)
{
	struct qdisc_skb_cb *qcb;
244 245

	BUILD_BUG_ON(sizeof(skb->cb) < offsetof(struct qdisc_skb_cb, data) + sz);
246 247 248
	BUILD_BUG_ON(sizeof(qcb->data) < sz);
}

249
static inline int qdisc_qlen(const struct Qdisc *q)
250 251 252 253
{
	return q->q.qlen;
}

254
static inline struct qdisc_skb_cb *qdisc_skb_cb(const struct sk_buff *skb)
255 256 257 258
{
	return (struct qdisc_skb_cb *)skb->cb;
}

259 260 261 262 263
static inline spinlock_t *qdisc_lock(struct Qdisc *qdisc)
{
	return &qdisc->q.lock;
}

264
static inline struct Qdisc *qdisc_root(const struct Qdisc *qdisc)
265
{
266 267 268
	struct Qdisc *q = rcu_dereference_rtnl(qdisc->dev_queue->qdisc);

	return q;
269 270
}

271
static inline struct Qdisc *qdisc_root_sleeping(const struct Qdisc *qdisc)
272 273 274 275
{
	return qdisc->dev_queue->qdisc_sleeping;
}

276 277 278 279 280 281 282 283 284 285 286
/* 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.
 */
287
static inline spinlock_t *qdisc_root_lock(const struct Qdisc *qdisc)
288 289 290
{
	struct Qdisc *root = qdisc_root(qdisc);

291
	ASSERT_RTNL();
292
	return qdisc_lock(root);
293 294
}

295
static inline spinlock_t *qdisc_root_sleeping_lock(const struct Qdisc *qdisc)
296 297 298 299 300 301 302
{
	struct Qdisc *root = qdisc_root_sleeping(qdisc);

	ASSERT_RTNL();
	return qdisc_lock(root);
}

303 304 305 306 307 308 309 310
static inline seqcount_t *qdisc_root_sleeping_running(const struct Qdisc *qdisc)
{
	struct Qdisc *root = qdisc_root_sleeping(qdisc);

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

311
static inline struct net_device *qdisc_dev(const struct Qdisc *qdisc)
312 313 314
{
	return qdisc->dev_queue->dev;
}
L
Linus Torvalds 已提交
315

316
static inline void sch_tree_lock(const struct Qdisc *q)
317
{
J
Jarek Poplawski 已提交
318
	spin_lock_bh(qdisc_root_sleeping_lock(q));
319 320
}

321
static inline void sch_tree_unlock(const struct Qdisc *q)
322
{
J
Jarek Poplawski 已提交
323
	spin_unlock_bh(qdisc_root_sleeping_lock(q));
324 325 326 327
}

#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
328

329 330
extern struct Qdisc noop_qdisc;
extern struct Qdisc_ops noop_qdisc_ops;
331 332
extern struct Qdisc_ops pfifo_fast_ops;
extern struct Qdisc_ops mq_qdisc_ops;
P
Phil Sutter 已提交
333
extern struct Qdisc_ops noqueue_qdisc_ops;
334
extern const struct Qdisc_ops *default_qdisc_ops;
335 336 337 338 339 340
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;
}
341

E
Eric Dumazet 已提交
342
struct Qdisc_class_common {
343 344 345 346
	u32			classid;
	struct hlist_node	hnode;
};

E
Eric Dumazet 已提交
347
struct Qdisc_class_hash {
348 349 350 351 352 353 354 355 356 357 358 359 360 361
	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 *
362
qdisc_class_find(const struct Qdisc_class_hash *hash, u32 id)
363 364 365 366 367
{
	struct Qdisc_class_common *cl;
	unsigned int h;

	h = qdisc_class_hash(id, hash->hashmask);
368
	hlist_for_each_entry(cl, &hash->hash[h], hnode) {
369 370 371 372 373 374
		if (cl->classid == id)
			return cl;
	}
	return NULL;
}

375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391
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);
392 393
void qdisc_tree_reduce_backlog(struct Qdisc *qdisc, unsigned int n,
			       unsigned int len);
394
struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
395
			  const struct Qdisc_ops *ops);
396
struct Qdisc *qdisc_create_dflt(struct netdev_queue *dev_queue,
397
				const struct Qdisc_ops *ops, u32 parentid);
398 399
void __qdisc_calculate_pkt_len(struct sk_buff *skb,
			       const struct qdisc_size_table *stab);
400
bool tcf_destroy(struct tcf_proto *tp, bool force);
J
John Fastabend 已提交
401
void tcf_destroy_chain(struct tcf_proto __rcu **fl);
402
int skb_do_redirect(struct sk_buff *);
L
Linus Torvalds 已提交
403

404 405 406 407 408 409 410 411 412
static inline bool skb_at_tc_ingress(const struct sk_buff *skb)
{
#ifdef CONFIG_NET_CLS_ACT
	return G_TC_AT(skb->tc_verd) & AT_INGRESS;
#else
	return false;
#endif
}

413 414
/* 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)
415
{
416 417
	struct Qdisc *qdisc;

418
	for (; i < dev->num_tx_queues; i++) {
419
		qdisc = rtnl_dereference(netdev_get_tx_queue(dev, i)->qdisc);
420 421 422 423 424 425
		if (qdisc) {
			spin_lock_bh(qdisc_lock(qdisc));
			qdisc_reset(qdisc);
			spin_unlock_bh(qdisc_lock(qdisc));
		}
	}
426 427 428 429
}

static inline void qdisc_reset_all_tx(struct net_device *dev)
{
430
	qdisc_reset_all_tx_gt(dev, 0);
431 432
}

433 434 435
/* Are all TX queues of the device empty?  */
static inline bool qdisc_all_tx_empty(const struct net_device *dev)
{
436
	unsigned int i;
437 438

	rcu_read_lock();
439 440
	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
441
		const struct Qdisc *q = rcu_dereference(txq->qdisc);
442

443 444
		if (q->q.qlen) {
			rcu_read_unlock();
445
			return false;
446
		}
447
	}
448
	rcu_read_unlock();
449
	return true;
450 451
}

452
/* Are any of the TX qdiscs changing?  */
453
static inline bool qdisc_tx_changing(const struct net_device *dev)
454
{
455
	unsigned int i;
456

457 458
	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
459
		if (rcu_access_pointer(txq->qdisc) != txq->qdisc_sleeping)
460 461 462
			return true;
	}
	return false;
463 464
}

465
/* Is the device using the noop qdisc on all queues?  */
466 467
static inline bool qdisc_tx_is_noop(const struct net_device *dev)
{
468
	unsigned int i;
469

470 471
	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
472
		if (rcu_access_pointer(txq->qdisc) != &noop_qdisc)
473 474 475
			return false;
	}
	return true;
476 477
}

478
static inline unsigned int qdisc_pkt_len(const struct sk_buff *skb)
479
{
480
	return qdisc_skb_cb(skb)->pkt_len;
481 482
}

483
/* additional qdisc xmit flags (NET_XMIT_MASK in linux/netdevice.h) */
484 485
enum net_xmit_qdisc_t {
	__NET_XMIT_STOLEN = 0x00010000,
486
	__NET_XMIT_BYPASS = 0x00020000,
487 488
};

489
#ifdef CONFIG_NET_CLS_ACT
490 491 492 493 494
#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
495 496
static inline void qdisc_calculate_pkt_len(struct sk_buff *skb,
					   const struct Qdisc *sch)
497
{
498
#ifdef CONFIG_NET_SCHED
E
Eric Dumazet 已提交
499 500 501 502
	struct qdisc_size_table *stab = rcu_dereference_bh(sch->stab);

	if (stab)
		__qdisc_calculate_pkt_len(skb, stab);
503
#endif
E
Eric Dumazet 已提交
504 505
}

506 507
static inline int qdisc_enqueue(struct sk_buff *skb, struct Qdisc *sch,
				struct sk_buff **to_free)
E
Eric Dumazet 已提交
508 509
{
	qdisc_calculate_pkt_len(skb, sch);
510
	return sch->enqueue(skb, sch, to_free);
511 512
}

513 514 515 516
static inline bool qdisc_is_percpu_stats(const struct Qdisc *q)
{
	return q->flags & TCQ_F_CPUSTATS;
}
517

518 519 520 521 522 523 524
static inline void _bstats_update(struct gnet_stats_basic_packed *bstats,
				  __u64 bytes, __u32 packets)
{
	bstats->bytes += bytes;
	bstats->packets += packets;
}

525 526 527
static inline void bstats_update(struct gnet_stats_basic_packed *bstats,
				 const struct sk_buff *skb)
{
528 529 530 531 532 533 534 535 536 537 538
	_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);
539 540
}

541 542
static inline void bstats_cpu_update(struct gnet_stats_basic_cpu *bstats,
				     const struct sk_buff *skb)
543 544 545 546 547 548
{
	u64_stats_update_begin(&bstats->syncp);
	bstats_update(&bstats->bstats, skb);
	u64_stats_update_end(&bstats->syncp);
}

549 550 551 552 553 554
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);
}

555 556
static inline void qdisc_bstats_update(struct Qdisc *sch,
				       const struct sk_buff *skb)
557
{
558
	bstats_update(&sch->bstats, skb);
559 560
}

561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577
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;
}

578
static inline void qstats_drop_inc(struct gnet_stats_queue *qstats)
579
{
580
	qstats->drops++;
581 582
}

583
static inline void qstats_overlimit_inc(struct gnet_stats_queue *qstats)
584
{
585 586
	qstats->overlimits++;
}
587

588 589 590 591 592 593 594
static inline void qdisc_qstats_drop(struct Qdisc *sch)
{
	qstats_drop_inc(&sch->qstats);
}

static inline void qdisc_qstats_cpu_drop(struct Qdisc *sch)
{
595
	this_cpu_inc(sch->cpu_qstats->drops);
596 597
}

598 599 600 601 602
static inline void qdisc_qstats_overlimit(struct Qdisc *sch)
{
	sch->qstats.overlimits++;
}

603 604 605 606
static inline int __qdisc_enqueue_tail(struct sk_buff *skb, struct Qdisc *sch,
				       struct sk_buff_head *list)
{
	__skb_queue_tail(list, skb);
607
	qdisc_qstats_backlog_inc(sch, skb);
608 609 610 611 612 613 614 615 616

	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);
}

617
static inline struct sk_buff *__qdisc_dequeue_head(struct sk_buff_head *list)
618 619 620
{
	struct sk_buff *skb = __skb_dequeue(list);

621 622 623 624 625 626 627
	return skb;
}

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

628
	if (likely(skb != NULL)) {
629
		qdisc_qstats_backlog_dec(sch, skb);
630 631
		qdisc_bstats_update(sch, skb);
	}
632 633 634 635

	return skb;
}

636 637 638 639 640 641 642 643 644
/* 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;
}

645
static inline unsigned int __qdisc_queue_drop_head(struct Qdisc *sch,
646 647
						   struct sk_buff_head *list,
						   struct sk_buff **to_free)
648
{
649
	struct sk_buff *skb = __skb_dequeue(list);
650 651 652

	if (likely(skb != NULL)) {
		unsigned int len = qdisc_pkt_len(skb);
653

654
		qdisc_qstats_backlog_dec(sch, skb);
655
		__qdisc_drop(skb, to_free);
656 657 658 659 660 661
		return len;
	}

	return 0;
}

662 663
static inline unsigned int qdisc_queue_drop_head(struct Qdisc *sch,
						 struct sk_buff **to_free)
664
{
665
	return __qdisc_queue_drop_head(sch, &sch->q, to_free);
666 667
}

668 669 670 671 672
static inline struct sk_buff *qdisc_peek_head(struct Qdisc *sch)
{
	return skb_peek(&sch->q);
}

673 674 675 676
/* 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 */
677
	if (!sch->gso_skb) {
678
		sch->gso_skb = sch->dequeue(sch);
679
		if (sch->gso_skb) {
680
			/* it's still part of the queue */
681
			qdisc_qstats_backlog_inc(sch, sch->gso_skb);
682
			sch->q.qlen++;
683
		}
684
	}
685 686 687 688 689 690 691 692 693

	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;

694
	if (skb) {
695
		sch->gso_skb = NULL;
696
		qdisc_qstats_backlog_dec(sch, skb);
697 698
		sch->q.qlen--;
	} else {
699
		skb = sch->dequeue(sch);
700
	}
701 702 703 704

	return skb;
}

705
static inline void __qdisc_reset_queue(struct sk_buff_head *list)
706 707 708 709 710
{
	/*
	 * We do not know the backlog in bytes of this list, it
	 * is up to the caller to correct it
	 */
711 712 713 714
	if (!skb_queue_empty(list)) {
		rtnl_kfree_skbs(list->next, list->prev);
		__skb_queue_head_init(list);
	}
715 716 717 718
}

static inline void qdisc_reset_queue(struct Qdisc *sch)
{
719
	__qdisc_reset_queue(&sch->q);
720 721 722
	sch->qstats.backlog = 0;
}

723 724 725 726 727 728 729 730 731
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) {
732
		qdisc_tree_reduce_backlog(old, old->q.qlen, old->qstats.backlog);
733 734 735 736 737 738 739
		qdisc_reset(old);
	}
	sch_tree_unlock(sch);

	return old;
}

740 741 742 743 744 745
static inline void rtnl_qdisc_drop(struct sk_buff *skb, struct Qdisc *sch)
{
	rtnl_kfree_skbs(skb, skb);
	qdisc_qstats_drop(sch);
}

746 747 748

static inline int qdisc_drop(struct sk_buff *skb, struct Qdisc *sch,
			     struct sk_buff **to_free)
749
{
750
	__qdisc_drop(skb, to_free);
751
	qdisc_qstats_drop(sch);
752 753 754 755

	return NET_XMIT_DROP;
}

756 757 758 759 760
/* 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)
{
761 762 763
	int slot = pktlen + rtab->rate.cell_align + rtab->rate.overhead;
	if (slot < 0)
		slot = 0;
764 765
	slot >>= rtab->rate.cell_log;
	if (slot > 255)
E
Eric Dumazet 已提交
766
		return rtab->data[255]*(slot >> 8) + rtab->data[slot & 0xFF];
767 768 769
	return rtab->data[slot];
}

770
struct psched_ratecfg {
771
	u64	rate_bytes_ps; /* bytes per second */
772 773
	u32	mult;
	u16	overhead;
774
	u8	linklayer;
775
	u8	shift;
776 777 778 779 780
};

static inline u64 psched_l2t_ns(const struct psched_ratecfg *r,
				unsigned int len)
{
781 782 783 784 785 786
	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;
787 788
}

789
void psched_ratecfg_precompute(struct psched_ratecfg *r,
790 791
			       const struct tc_ratespec *conf,
			       u64 rate64);
792

793 794
static inline void psched_ratecfg_getrate(struct tc_ratespec *res,
					  const struct psched_ratecfg *r)
795
{
796
	memset(res, 0, sizeof(*res));
797 798 799 800 801 802 803

	/* 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);

804
	res->overhead = r->overhead;
805
	res->linklayer = (r->linklayer & TC_LINKLAYER_MASK);
806 807
}

L
Linus Torvalds 已提交
808
#endif