sch_generic.h 21.7 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>
13
#include <linux/workqueue.h>
L
Linus Torvalds 已提交
14
#include <net/gen_stats.h>
15
#include <net/rtnetlink.h>
L
Linus Torvalds 已提交
16 17 18 19 20 21

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

E
Eric Dumazet 已提交
22
struct qdisc_rate_table {
L
Linus Torvalds 已提交
23 24 25 26 27 28
	struct tc_ratespec rate;
	u32		data[256];
	struct qdisc_rate_table *next;
	int		refcnt;
};

E
Eric Dumazet 已提交
29
enum qdisc_state_t {
30
	__QDISC_STATE_SCHED,
31
	__QDISC_STATE_DEACTIVATED,
32 33
};

34
struct qdisc_size_table {
E
Eric Dumazet 已提交
35
	struct rcu_head		rcu;
36 37 38 39 40 41
	struct list_head	list;
	struct tc_sizespec	szopts;
	int			refcnt;
	u16			data[];
};

42 43 44 45 46 47 48 49
/* 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
50
struct Qdisc {
51 52 53 54
	int 			(*enqueue)(struct sk_buff *skb,
					   struct Qdisc *sch,
					   struct sk_buff **to_free);
	struct sk_buff *	(*dequeue)(struct Qdisc *sch);
55
	unsigned int		flags;
56
#define TCQ_F_BUILTIN		1
57 58 59
#define TCQ_F_INGRESS		2
#define TCQ_F_CAN_BYPASS	4
#define TCQ_F_MQROOT		8
60 61 62 63 64 65 66
#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.
				      */
67
#define TCQ_F_WARN_NONWC	(1 << 16)
68
#define TCQ_F_CPUSTATS		0x20 /* run using percpu statistics */
69 70 71
#define TCQ_F_NOPARENT		0x40 /* root of its hierarchy :
				      * qdisc_tree_decrease_qlen() should stop.
				      */
72
#define TCQ_F_INVISIBLE		0x80 /* invisible by default in dump */
73
	u32			limit;
74
	const struct Qdisc_ops	*ops;
E
Eric Dumazet 已提交
75
	struct qdisc_size_table	__rcu *stab;
76
	struct hlist_node       hash;
L
Linus Torvalds 已提交
77 78
	u32			handle;
	u32			parent;
79

80 81
	struct netdev_queue	*dev_queue;

82
	struct net_rate_estimator __rcu *rate_est;
83 84 85
	struct gnet_stats_basic_cpu __percpu *cpu_bstats;
	struct gnet_stats_queue	__percpu *cpu_qstats;

86 87 88
	/*
	 * For performance sake on SMP, we put highly modified fields at the end
	 */
89
	struct sk_buff		*gso_skb ____cacheline_aligned_in_smp;
90
	struct qdisc_skb_head	q;
91
	struct gnet_stats_basic_packed bstats;
92
	seqcount_t		running;
93
	struct gnet_stats_queue	qstats;
94 95 96
	unsigned long		state;
	struct Qdisc            *next_sched;
	struct sk_buff		*skb_bad_txq;
97
	int			padded;
98
	refcount_t		refcnt;
99 100

	spinlock_t		busylock ____cacheline_aligned_in_smp;
L
Linus Torvalds 已提交
101 102
};

103 104 105 106 107 108 109
static inline void qdisc_refcount_inc(struct Qdisc *qdisc)
{
	if (qdisc->flags & TCQ_F_BUILTIN)
		return;
	refcount_inc(&qdisc->refcnt);
}

110
static inline bool qdisc_is_running(const struct Qdisc *qdisc)
111
{
112
	return (raw_read_seqcount(&qdisc->running) & 1) ? true : false;
113 114 115 116
}

static inline bool qdisc_run_begin(struct Qdisc *qdisc)
{
117 118
	if (qdisc_is_running(qdisc))
		return false;
119 120 121 122 123
	/* 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_);
124
	return true;
125 126 127 128
}

static inline void qdisc_run_end(struct Qdisc *qdisc)
{
129
	write_seqcount_end(&qdisc->running);
130 131
}

132 133 134 135 136 137 138 139 140 141 142 143 144 145 146
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
147
struct Qdisc_class_ops {
L
Linus Torvalds 已提交
148
	/* Child qdisc manipulation */
149
	struct netdev_queue *	(*select_queue)(struct Qdisc *, struct tcmsg *);
L
Linus Torvalds 已提交
150 151 152
	int			(*graft)(struct Qdisc *, unsigned long cl,
					struct Qdisc *, struct Qdisc **);
	struct Qdisc *		(*leaf)(struct Qdisc *, unsigned long cl);
153
	void			(*qlen_notify)(struct Qdisc *, unsigned long);
L
Linus Torvalds 已提交
154 155

	/* Class manipulation routines */
156
	unsigned long		(*find)(struct Qdisc *, u32 classid);
L
Linus Torvalds 已提交
157
	int			(*change)(struct Qdisc *, u32, u32,
158
					struct nlattr **, unsigned long *);
L
Linus Torvalds 已提交
159 160 161 162
	int			(*delete)(struct Qdisc *, unsigned long);
	void			(*walk)(struct Qdisc *, struct qdisc_walker * arg);

	/* Filter manipulation */
163
	struct tcf_block *	(*tcf_block)(struct Qdisc *, unsigned long);
L
Linus Torvalds 已提交
164 165 166 167 168 169 170 171 172 173 174
	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
175
struct Qdisc_ops {
L
Linus Torvalds 已提交
176
	struct Qdisc_ops	*next;
177
	const struct Qdisc_class_ops	*cl_ops;
L
Linus Torvalds 已提交
178 179 180
	char			id[IFNAMSIZ];
	int			priv_size;

181 182 183
	int 			(*enqueue)(struct sk_buff *skb,
					   struct Qdisc *sch,
					   struct sk_buff **to_free);
L
Linus Torvalds 已提交
184
	struct sk_buff *	(*dequeue)(struct Qdisc *);
185
	struct sk_buff *	(*peek)(struct Qdisc *);
L
Linus Torvalds 已提交
186

187
	int			(*init)(struct Qdisc *, struct nlattr *arg);
L
Linus Torvalds 已提交
188 189
	void			(*reset)(struct Qdisc *);
	void			(*destroy)(struct Qdisc *);
190
	int			(*change)(struct Qdisc *, struct nlattr *arg);
191
	void			(*attach)(struct Qdisc *);
L
Linus Torvalds 已提交
192 193 194 195 196 197 198 199

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

	struct module		*owner;
};


E
Eric Dumazet 已提交
200
struct tcf_result {
201 202 203 204 205 206 207
	union {
		struct {
			unsigned long	class;
			u32		classid;
		};
		const struct tcf_proto *goto_tp;
	};
L
Linus Torvalds 已提交
208 209
};

E
Eric Dumazet 已提交
210
struct tcf_proto_ops {
211
	struct list_head	head;
L
Linus Torvalds 已提交
212 213
	char			kind[IFNAMSIZ];

214 215 216
	int			(*classify)(struct sk_buff *,
					    const struct tcf_proto *,
					    struct tcf_result *);
L
Linus Torvalds 已提交
217
	int			(*init)(struct tcf_proto*);
218
	void			(*destroy)(struct tcf_proto*);
L
Linus Torvalds 已提交
219

220
	void*			(*get)(struct tcf_proto*, u32 handle);
221
	int			(*change)(struct net *net, struct sk_buff *,
222
					struct tcf_proto*, unsigned long,
223
					u32 handle, struct nlattr **,
224 225
					void **, bool);
	int			(*delete)(struct tcf_proto*, void *, bool*);
L
Linus Torvalds 已提交
226
	void			(*walk)(struct tcf_proto*, struct tcf_walker *arg);
227
	void			(*bind_class)(void *, u32, unsigned long);
L
Linus Torvalds 已提交
228 229

	/* rtnetlink specific */
230
	int			(*dump)(struct net*, struct tcf_proto*, void *,
L
Linus Torvalds 已提交
231 232 233 234 235
					struct sk_buff *skb, struct tcmsg*);

	struct module		*owner;
};

E
Eric Dumazet 已提交
236
struct tcf_proto {
L
Linus Torvalds 已提交
237
	/* Fast access part */
J
John Fastabend 已提交
238 239
	struct tcf_proto __rcu	*next;
	void __rcu		*root;
240 241 242
	int			(*classify)(struct sk_buff *,
					    const struct tcf_proto *,
					    struct tcf_result *);
A
Al Viro 已提交
243
	__be16			protocol;
L
Linus Torvalds 已提交
244 245 246 247 248 249

	/* All the rest */
	u32			prio;
	u32			classid;
	struct Qdisc		*q;
	void			*data;
250
	const struct tcf_proto_ops	*ops;
251
	struct tcf_chain	*chain;
J
John Fastabend 已提交
252
	struct rcu_head		rcu;
L
Linus Torvalds 已提交
253 254
};

255 256
struct qdisc_skb_cb {
	unsigned int		pkt_len;
257
	u16			slave_dev_queue_mapping;
258
	u16			tc_classid;
259 260
#define QDISC_CB_PRIV_LEN 20
	unsigned char		data[QDISC_CB_PRIV_LEN];
261 262
};

263 264
typedef void tcf_chain_head_change_t(struct tcf_proto *tp_head, void *priv);

265 266
struct tcf_chain {
	struct tcf_proto __rcu *filter_chain;
267 268
	tcf_chain_head_change_t *chain_head_change;
	void *chain_head_change_priv;
269 270 271 272
	struct list_head list;
	struct tcf_block *block;
	u32 index; /* chain index */
	unsigned int refcnt;
273 274
};

275
struct tcf_block {
276
	struct list_head chain_list;
277
	struct net *net;
278
	struct Qdisc *q;
279
	struct list_head cb_list;
280
	struct work_struct work;
281 282
};

283 284 285
static inline void qdisc_cb_private_validate(const struct sk_buff *skb, int sz)
{
	struct qdisc_skb_cb *qcb;
286 287

	BUILD_BUG_ON(sizeof(skb->cb) < offsetof(struct qdisc_skb_cb, data) + sz);
288 289 290
	BUILD_BUG_ON(sizeof(qcb->data) < sz);
}

291
static inline int qdisc_qlen(const struct Qdisc *q)
292 293 294 295
{
	return q->q.qlen;
}

296
static inline struct qdisc_skb_cb *qdisc_skb_cb(const struct sk_buff *skb)
297 298 299 300
{
	return (struct qdisc_skb_cb *)skb->cb;
}

301 302 303 304 305
static inline spinlock_t *qdisc_lock(struct Qdisc *qdisc)
{
	return &qdisc->q.lock;
}

306
static inline struct Qdisc *qdisc_root(const struct Qdisc *qdisc)
307
{
308 309 310
	struct Qdisc *q = rcu_dereference_rtnl(qdisc->dev_queue->qdisc);

	return q;
311 312
}

313
static inline struct Qdisc *qdisc_root_sleeping(const struct Qdisc *qdisc)
314 315 316 317
{
	return qdisc->dev_queue->qdisc_sleeping;
}

318 319 320 321 322 323 324 325 326 327 328
/* 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.
 */
329
static inline spinlock_t *qdisc_root_lock(const struct Qdisc *qdisc)
330 331 332
{
	struct Qdisc *root = qdisc_root(qdisc);

333
	ASSERT_RTNL();
334
	return qdisc_lock(root);
335 336
}

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

	ASSERT_RTNL();
	return qdisc_lock(root);
}

345 346 347 348 349 350 351 352
static inline seqcount_t *qdisc_root_sleeping_running(const struct Qdisc *qdisc)
{
	struct Qdisc *root = qdisc_root_sleeping(qdisc);

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

353
static inline struct net_device *qdisc_dev(const struct Qdisc *qdisc)
354 355 356
{
	return qdisc->dev_queue->dev;
}
L
Linus Torvalds 已提交
357

358
static inline void sch_tree_lock(const struct Qdisc *q)
359
{
J
Jarek Poplawski 已提交
360
	spin_lock_bh(qdisc_root_sleeping_lock(q));
361 362
}

363
static inline void sch_tree_unlock(const struct Qdisc *q)
364
{
J
Jarek Poplawski 已提交
365
	spin_unlock_bh(qdisc_root_sleeping_lock(q));
366 367
}

368 369
extern struct Qdisc noop_qdisc;
extern struct Qdisc_ops noop_qdisc_ops;
370 371
extern struct Qdisc_ops pfifo_fast_ops;
extern struct Qdisc_ops mq_qdisc_ops;
P
Phil Sutter 已提交
372
extern struct Qdisc_ops noqueue_qdisc_ops;
373
extern const struct Qdisc_ops *default_qdisc_ops;
374 375 376 377 378 379
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;
}
380

E
Eric Dumazet 已提交
381
struct Qdisc_class_common {
382 383 384 385
	u32			classid;
	struct hlist_node	hnode;
};

E
Eric Dumazet 已提交
386
struct Qdisc_class_hash {
387 388 389 390 391 392 393 394 395 396 397 398 399 400
	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 *
401
qdisc_class_find(const struct Qdisc_class_hash *hash, u32 id)
402 403 404 405
{
	struct Qdisc_class_common *cl;
	unsigned int h;

406 407 408
	if (!id)
		return NULL;

409
	h = qdisc_class_hash(id, hash->hashmask);
410
	hlist_for_each_entry(cl, &hash->hash[h], hnode) {
411 412 413 414 415 416
		if (cl->classid == id)
			return cl;
	}
	return NULL;
}

417 418 419 420 421 422 423
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;
}

424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440
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);
441 442
void qdisc_tree_reduce_backlog(struct Qdisc *qdisc, unsigned int n,
			       unsigned int len);
443
struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
444
			  const struct Qdisc_ops *ops);
445
struct Qdisc *qdisc_create_dflt(struct netdev_queue *dev_queue,
446
				const struct Qdisc_ops *ops, u32 parentid);
447 448
void __qdisc_calculate_pkt_len(struct sk_buff *skb,
			       const struct qdisc_size_table *stab);
449
int skb_do_redirect(struct sk_buff *);
L
Linus Torvalds 已提交
450

451 452 453
static inline void skb_reset_tc(struct sk_buff *skb)
{
#ifdef CONFIG_NET_CLS_ACT
454
	skb->tc_redirected = 0;
455 456 457
#endif
}

458 459 460
static inline bool skb_at_tc_ingress(const struct sk_buff *skb)
{
#ifdef CONFIG_NET_CLS_ACT
461
	return skb->tc_at_ingress;
462 463 464 465 466
#else
	return false;
#endif
}

467 468 469 470 471 472 473 474 475 476 477
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;
}

478 479
/* 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)
480
{
481 482
	struct Qdisc *qdisc;

483
	for (; i < dev->num_tx_queues; i++) {
484
		qdisc = rtnl_dereference(netdev_get_tx_queue(dev, i)->qdisc);
485 486 487 488 489 490
		if (qdisc) {
			spin_lock_bh(qdisc_lock(qdisc));
			qdisc_reset(qdisc);
			spin_unlock_bh(qdisc_lock(qdisc));
		}
	}
491 492 493 494
}

static inline void qdisc_reset_all_tx(struct net_device *dev)
{
495
	qdisc_reset_all_tx_gt(dev, 0);
496 497
}

498 499 500
/* Are all TX queues of the device empty?  */
static inline bool qdisc_all_tx_empty(const struct net_device *dev)
{
501
	unsigned int i;
502 503

	rcu_read_lock();
504 505
	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
506
		const struct Qdisc *q = rcu_dereference(txq->qdisc);
507

508 509
		if (q->q.qlen) {
			rcu_read_unlock();
510
			return false;
511
		}
512
	}
513
	rcu_read_unlock();
514
	return true;
515 516
}

517
/* Are any of the TX qdiscs changing?  */
518
static inline bool qdisc_tx_changing(const struct net_device *dev)
519
{
520
	unsigned int i;
521

522 523
	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
524
		if (rcu_access_pointer(txq->qdisc) != txq->qdisc_sleeping)
525 526 527
			return true;
	}
	return false;
528 529
}

530
/* Is the device using the noop qdisc on all queues?  */
531 532
static inline bool qdisc_tx_is_noop(const struct net_device *dev)
{
533
	unsigned int i;
534

535 536
	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
537
		if (rcu_access_pointer(txq->qdisc) != &noop_qdisc)
538 539 540
			return false;
	}
	return true;
541 542
}

543
static inline unsigned int qdisc_pkt_len(const struct sk_buff *skb)
544
{
545
	return qdisc_skb_cb(skb)->pkt_len;
546 547
}

548
/* additional qdisc xmit flags (NET_XMIT_MASK in linux/netdevice.h) */
549 550
enum net_xmit_qdisc_t {
	__NET_XMIT_STOLEN = 0x00010000,
551
	__NET_XMIT_BYPASS = 0x00020000,
552 553
};

554
#ifdef CONFIG_NET_CLS_ACT
555 556 557 558 559
#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
560 561
static inline void qdisc_calculate_pkt_len(struct sk_buff *skb,
					   const struct Qdisc *sch)
562
{
563
#ifdef CONFIG_NET_SCHED
E
Eric Dumazet 已提交
564 565 566 567
	struct qdisc_size_table *stab = rcu_dereference_bh(sch->stab);

	if (stab)
		__qdisc_calculate_pkt_len(skb, stab);
568
#endif
E
Eric Dumazet 已提交
569 570
}

571 572
static inline int qdisc_enqueue(struct sk_buff *skb, struct Qdisc *sch,
				struct sk_buff **to_free)
E
Eric Dumazet 已提交
573 574
{
	qdisc_calculate_pkt_len(skb, sch);
575
	return sch->enqueue(skb, sch, to_free);
576 577
}

578 579 580 581
static inline bool qdisc_is_percpu_stats(const struct Qdisc *q)
{
	return q->flags & TCQ_F_CPUSTATS;
}
582

583 584 585 586 587 588 589
static inline void _bstats_update(struct gnet_stats_basic_packed *bstats,
				  __u64 bytes, __u32 packets)
{
	bstats->bytes += bytes;
	bstats->packets += packets;
}

590 591 592
static inline void bstats_update(struct gnet_stats_basic_packed *bstats,
				 const struct sk_buff *skb)
{
593 594 595 596 597 598 599 600 601 602 603
	_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);
604 605
}

606 607
static inline void bstats_cpu_update(struct gnet_stats_basic_cpu *bstats,
				     const struct sk_buff *skb)
608 609 610 611 612 613
{
	u64_stats_update_begin(&bstats->syncp);
	bstats_update(&bstats->bstats, skb);
	u64_stats_update_end(&bstats->syncp);
}

614 615 616 617 618 619
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);
}

620 621
static inline void qdisc_bstats_update(struct Qdisc *sch,
				       const struct sk_buff *skb)
622
{
623
	bstats_update(&sch->bstats, skb);
624 625
}

626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642
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;
}

643
static inline void qstats_drop_inc(struct gnet_stats_queue *qstats)
644
{
645
	qstats->drops++;
646 647
}

648
static inline void qstats_overlimit_inc(struct gnet_stats_queue *qstats)
649
{
650 651
	qstats->overlimits++;
}
652

653 654 655 656 657 658 659
static inline void qdisc_qstats_drop(struct Qdisc *sch)
{
	qstats_drop_inc(&sch->qstats);
}

static inline void qdisc_qstats_cpu_drop(struct Qdisc *sch)
{
660
	this_cpu_inc(sch->cpu_qstats->drops);
661 662
}

663 664 665 666 667
static inline void qdisc_qstats_overlimit(struct Qdisc *sch)
{
	sch->qstats.overlimits++;
}

668 669 670 671 672 673 674
static inline void qdisc_skb_head_init(struct qdisc_skb_head *qh)
{
	qh->head = NULL;
	qh->tail = NULL;
	qh->qlen = 0;
}

675
static inline int __qdisc_enqueue_tail(struct sk_buff *skb, struct Qdisc *sch,
676
				       struct qdisc_skb_head *qh)
677
{
678 679 680 681 682 683 684 685 686 687 688
	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++;
689
	qdisc_qstats_backlog_inc(sch, skb);
690 691 692 693 694 695 696 697 698

	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);
}

699
static inline struct sk_buff *__qdisc_dequeue_head(struct qdisc_skb_head *qh)
700
{
701 702 703 704 705 706 707 708 709
	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;
	}
710

711 712 713 714 715 716 717
	return skb;
}

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

718
	if (likely(skb != NULL)) {
719
		qdisc_qstats_backlog_dec(sch, skb);
720 721
		qdisc_bstats_update(sch, skb);
	}
722 723 724 725

	return skb;
}

726 727 728 729 730 731 732 733 734
/* 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;
}

735
static inline unsigned int __qdisc_queue_drop_head(struct Qdisc *sch,
736
						   struct qdisc_skb_head *qh,
737
						   struct sk_buff **to_free)
738
{
739
	struct sk_buff *skb = __qdisc_dequeue_head(qh);
740 741 742

	if (likely(skb != NULL)) {
		unsigned int len = qdisc_pkt_len(skb);
743

744
		qdisc_qstats_backlog_dec(sch, skb);
745
		__qdisc_drop(skb, to_free);
746 747 748 749 750 751
		return len;
	}

	return 0;
}

752 753
static inline unsigned int qdisc_queue_drop_head(struct Qdisc *sch,
						 struct sk_buff **to_free)
754
{
755
	return __qdisc_queue_drop_head(sch, &sch->q, to_free);
756 757
}

758 759
static inline struct sk_buff *qdisc_peek_head(struct Qdisc *sch)
{
760 761 762
	const struct qdisc_skb_head *qh = &sch->q;

	return qh->head;
763 764
}

765 766 767 768
/* 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 */
769
	if (!sch->gso_skb) {
770
		sch->gso_skb = sch->dequeue(sch);
771
		if (sch->gso_skb) {
772
			/* it's still part of the queue */
773
			qdisc_qstats_backlog_inc(sch, sch->gso_skb);
774
			sch->q.qlen++;
775
		}
776
	}
777 778 779 780 781 782 783 784 785

	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;

786
	if (skb) {
787
		sch->gso_skb = NULL;
788
		qdisc_qstats_backlog_dec(sch, skb);
789 790
		sch->q.qlen--;
	} else {
791
		skb = sch->dequeue(sch);
792
	}
793 794 795 796

	return skb;
}

797
static inline void __qdisc_reset_queue(struct qdisc_skb_head *qh)
798 799 800 801 802
{
	/*
	 * We do not know the backlog in bytes of this list, it
	 * is up to the caller to correct it
	 */
803 804 805 806 807 808 809
	ASSERT_RTNL();
	if (qh->qlen) {
		rtnl_kfree_skbs(qh->head, qh->tail);

		qh->head = NULL;
		qh->tail = NULL;
		qh->qlen = 0;
810
	}
811 812 813 814
}

static inline void qdisc_reset_queue(struct Qdisc *sch)
{
815
	__qdisc_reset_queue(&sch->q);
816 817 818
	sch->qstats.backlog = 0;
}

819 820 821 822 823 824 825 826 827
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) {
828 829 830
		unsigned int qlen = old->q.qlen;
		unsigned int backlog = old->qstats.backlog;

831
		qdisc_reset(old);
832
		qdisc_tree_reduce_backlog(old, qlen, backlog);
833 834 835 836 837 838
	}
	sch_tree_unlock(sch);

	return old;
}

839 840 841 842 843 844
static inline void rtnl_qdisc_drop(struct sk_buff *skb, struct Qdisc *sch)
{
	rtnl_kfree_skbs(skb, skb);
	qdisc_qstats_drop(sch);
}

845 846 847

static inline int qdisc_drop(struct sk_buff *skb, struct Qdisc *sch,
			     struct sk_buff **to_free)
848
{
849
	__qdisc_drop(skb, to_free);
850
	qdisc_qstats_drop(sch);
851 852 853 854

	return NET_XMIT_DROP;
}

855 856 857 858 859
/* 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)
{
860 861 862
	int slot = pktlen + rtab->rate.cell_align + rtab->rate.overhead;
	if (slot < 0)
		slot = 0;
863 864
	slot >>= rtab->rate.cell_log;
	if (slot > 255)
E
Eric Dumazet 已提交
865
		return rtab->data[255]*(slot >> 8) + rtab->data[slot & 0xFF];
866 867 868
	return rtab->data[slot];
}

869
struct psched_ratecfg {
870
	u64	rate_bytes_ps; /* bytes per second */
871 872
	u32	mult;
	u16	overhead;
873
	u8	linklayer;
874
	u8	shift;
875 876 877 878 879
};

static inline u64 psched_l2t_ns(const struct psched_ratecfg *r,
				unsigned int len)
{
880 881 882 883 884 885
	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;
886 887
}

888
void psched_ratecfg_precompute(struct psched_ratecfg *r,
889 890
			       const struct tc_ratespec *conf,
			       u64 rate64);
891

892 893
static inline void psched_ratecfg_getrate(struct tc_ratespec *res,
					  const struct psched_ratecfg *r)
894
{
895
	memset(res, 0, sizeof(*res));
896 897 898 899 900 901 902

	/* 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);

903
	res->overhead = r->overhead;
904
	res->linklayer = (r->linklayer & TC_LINKLAYER_MASK);
905 906
}

L
Linus Torvalds 已提交
907
#endif