sch_api.c 39.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
/*
 * net/sched/sch_api.c	Packet scheduler API.
 *
 *		This program is free software; you can redistribute it and/or
 *		modify it under the terms of the GNU General Public License
 *		as published by the Free Software Foundation; either version
 *		2 of the License, or (at your option) any later version.
 *
 * Authors:	Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
 *
 * Fixes:
 *
 * Rani Assaf <rani@magic.metawire.com> :980802: JIFFIES and CPU clock sources are repaired.
 * Eduardo J. Blanco <ejbs@netlabs.com.uy> :990222: kmod support
 * Jamal Hadi Salim <hadi@nortelnetworks.com>: 990601: ingress support
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/skbuff.h>
#include <linux/init.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/kmod.h>
#include <linux/list.h>
29
#include <linux/hrtimer.h>
30
#include <linux/lockdep.h>
L
Linus Torvalds 已提交
31

32
#include <net/net_namespace.h>
33
#include <net/sock.h>
34
#include <net/netlink.h>
L
Linus Torvalds 已提交
35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99
#include <net/pkt_sched.h>

static int qdisc_notify(struct sk_buff *oskb, struct nlmsghdr *n, u32 clid,
			struct Qdisc *old, struct Qdisc *new);
static int tclass_notify(struct sk_buff *oskb, struct nlmsghdr *n,
			 struct Qdisc *q, unsigned long cl, int event);

/*

   Short review.
   -------------

   This file consists of two interrelated parts:

   1. queueing disciplines manager frontend.
   2. traffic classes manager frontend.

   Generally, queueing discipline ("qdisc") is a black box,
   which is able to enqueue packets and to dequeue them (when
   device is ready to send something) in order and at times
   determined by algorithm hidden in it.

   qdisc's are divided to two categories:
   - "queues", which have no internal structure visible from outside.
   - "schedulers", which split all the packets to "traffic classes",
     using "packet classifiers" (look at cls_api.c)

   In turn, classes may have child qdiscs (as rule, queues)
   attached to them etc. etc. etc.

   The goal of the routines in this file is to translate
   information supplied by user in the form of handles
   to more intelligible for kernel form, to make some sanity
   checks and part of work, which is common to all qdiscs
   and to provide rtnetlink notifications.

   All real intelligent work is done inside qdisc modules.



   Every discipline has two major routines: enqueue and dequeue.

   ---dequeue

   dequeue usually returns a skb to send. It is allowed to return NULL,
   but it does not mean that queue is empty, it just means that
   discipline does not want to send anything this time.
   Queue is really empty if q->q.qlen == 0.
   For complicated disciplines with multiple queues q->q is not
   real packet queue, but however q->q.qlen must be valid.

   ---enqueue

   enqueue returns 0, if packet was enqueued successfully.
   If packet (this one or another one) was dropped, it returns
   not zero error code.
   NET_XMIT_DROP 	- this packet dropped
     Expected action: do not backoff, but wait until queue will clear.
   NET_XMIT_CN	 	- probably this packet enqueued, but another one dropped.
     Expected action: backoff or ignore
   NET_XMIT_POLICED	- dropped by police.
     Expected action: backoff or error to real-time apps.

   Auxiliary routines:

100 101 102 103
   ---peek

   like dequeue but without removing a packet from the queue

L
Linus Torvalds 已提交
104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148
   ---reset

   returns qdisc to initial state: purge all buffers, clear all
   timers, counters (except for statistics) etc.

   ---init

   initializes newly created qdisc.

   ---destroy

   destroys resources allocated by init and during lifetime of qdisc.

   ---change

   changes qdisc parameters.
 */

/* Protects list of registered TC modules. It is pure SMP lock. */
static DEFINE_RWLOCK(qdisc_mod_lock);


/************************************************
 *	Queueing disciplines manipulation.	*
 ************************************************/


/* The list of all installed queueing disciplines. */

static struct Qdisc_ops *qdisc_base;

/* Register/uregister queueing discipline */

int register_qdisc(struct Qdisc_ops *qops)
{
	struct Qdisc_ops *q, **qp;
	int rc = -EEXIST;

	write_lock(&qdisc_mod_lock);
	for (qp = &qdisc_base; (q = *qp) != NULL; qp = &q->next)
		if (!strcmp(qops->id, q->id))
			goto out;

	if (qops->enqueue == NULL)
		qops->enqueue = noop_qdisc_ops.enqueue;
149 150 151 152 153 154 155 156
	if (qops->peek == NULL) {
		if (qops->dequeue == NULL) {
			qops->peek = noop_qdisc_ops.peek;
		} else {
			rc = -EINVAL;
			goto out;
		}
	}
L
Linus Torvalds 已提交
157 158 159 160 161 162 163 164 165 166
	if (qops->dequeue == NULL)
		qops->dequeue = noop_qdisc_ops.dequeue;

	qops->next = NULL;
	*qp = qops;
	rc = 0;
out:
	write_unlock(&qdisc_mod_lock);
	return rc;
}
167
EXPORT_SYMBOL(register_qdisc);
L
Linus Torvalds 已提交
168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185

int unregister_qdisc(struct Qdisc_ops *qops)
{
	struct Qdisc_ops *q, **qp;
	int err = -ENOENT;

	write_lock(&qdisc_mod_lock);
	for (qp = &qdisc_base; (q=*qp)!=NULL; qp = &q->next)
		if (q == qops)
			break;
	if (q) {
		*qp = q->next;
		q->next = NULL;
		err = 0;
	}
	write_unlock(&qdisc_mod_lock);
	return err;
}
186
EXPORT_SYMBOL(unregister_qdisc);
L
Linus Torvalds 已提交
187 188 189 190 191

/* We know handle. Find qdisc among all qdisc's attached to device
   (root qdisc, all its children, children of children etc.)
 */

H
Hannes Eder 已提交
192
static struct Qdisc *qdisc_match_from_root(struct Qdisc *root, u32 handle)
193 194 195 196 197 198 199 200 201 202 203 204 205 206
{
	struct Qdisc *q;

	if (!(root->flags & TCQ_F_BUILTIN) &&
	    root->handle == handle)
		return root;

	list_for_each_entry(q, &root->list, list) {
		if (q->handle == handle)
			return q;
	}
	return NULL;
}

207 208
static void qdisc_list_add(struct Qdisc *q)
{
209
	if ((q->parent != TC_H_ROOT) && !(q->flags & TCQ_F_INGRESS))
210 211 212 213 214
		list_add_tail(&q->list, &qdisc_root_sleeping(q)->list);
}

void qdisc_list_del(struct Qdisc *q)
{
215
	if ((q->parent != TC_H_ROOT) && !(q->flags & TCQ_F_INGRESS))
216 217 218 219
		list_del(&q->list);
}
EXPORT_SYMBOL(qdisc_list_del);

220
struct Qdisc *qdisc_lookup(struct net_device *dev, u32 handle)
L
Linus Torvalds 已提交
221
{
222
	unsigned int i;
223 224
	struct Qdisc *q;

225 226
	for (i = 0; i < dev->num_tx_queues; i++) {
		struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
227
		struct Qdisc *txq_root = txq->qdisc_sleeping;
L
Linus Torvalds 已提交
228

229 230
		q = qdisc_match_from_root(txq_root, handle);
		if (q)
231
			goto out;
L
Linus Torvalds 已提交
232
	}
233 234

	q = qdisc_match_from_root(dev->rx_queue.qdisc_sleeping, handle);
235
out:
236
	return q;
L
Linus Torvalds 已提交
237 238 239 240 241 242
}

static struct Qdisc *qdisc_leaf(struct Qdisc *p, u32 classid)
{
	unsigned long cl;
	struct Qdisc *leaf;
243
	const struct Qdisc_class_ops *cops = p->ops->cl_ops;
L
Linus Torvalds 已提交
244 245 246 247 248 249 250 251 252 253 254 255 256 257

	if (cops == NULL)
		return NULL;
	cl = cops->get(p, classid);

	if (cl == 0)
		return NULL;
	leaf = cops->leaf(p, cl);
	cops->put(p, cl);
	return leaf;
}

/* Find queueing discipline by name */

258
static struct Qdisc_ops *qdisc_lookup_ops(struct nlattr *kind)
L
Linus Torvalds 已提交
259 260 261 262 263 264
{
	struct Qdisc_ops *q = NULL;

	if (kind) {
		read_lock(&qdisc_mod_lock);
		for (q = qdisc_base; q; q = q->next) {
265
			if (nla_strcmp(kind, q->id) == 0) {
L
Linus Torvalds 已提交
266 267 268 269 270 271 272 273 274 275 276 277
				if (!try_module_get(q->owner))
					q = NULL;
				break;
			}
		}
		read_unlock(&qdisc_mod_lock);
	}
	return q;
}

static struct qdisc_rate_table *qdisc_rtab_list;

278
struct qdisc_rate_table *qdisc_get_rtab(struct tc_ratespec *r, struct nlattr *tab)
L
Linus Torvalds 已提交
279 280 281 282 283 284 285 286 287 288
{
	struct qdisc_rate_table *rtab;

	for (rtab = qdisc_rtab_list; rtab; rtab = rtab->next) {
		if (memcmp(&rtab->rate, r, sizeof(struct tc_ratespec)) == 0) {
			rtab->refcnt++;
			return rtab;
		}
	}

289 290
	if (tab == NULL || r->rate == 0 || r->cell_log == 0 ||
	    nla_len(tab) != TC_RTAB_SIZE)
L
Linus Torvalds 已提交
291 292 293 294 295 296
		return NULL;

	rtab = kmalloc(sizeof(*rtab), GFP_KERNEL);
	if (rtab) {
		rtab->rate = *r;
		rtab->refcnt = 1;
297
		memcpy(rtab->data, nla_data(tab), 1024);
L
Linus Torvalds 已提交
298 299 300 301 302
		rtab->next = qdisc_rtab_list;
		qdisc_rtab_list = rtab;
	}
	return rtab;
}
303
EXPORT_SYMBOL(qdisc_get_rtab);
L
Linus Torvalds 已提交
304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319

void qdisc_put_rtab(struct qdisc_rate_table *tab)
{
	struct qdisc_rate_table *rtab, **rtabp;

	if (!tab || --tab->refcnt)
		return;

	for (rtabp = &qdisc_rtab_list; (rtab=*rtabp) != NULL; rtabp = &rtab->next) {
		if (rtab == tab) {
			*rtabp = rtab->next;
			kfree(rtab);
			return;
		}
	}
}
320
EXPORT_SYMBOL(qdisc_put_rtab);
L
Linus Torvalds 已提交
321

322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356
static LIST_HEAD(qdisc_stab_list);
static DEFINE_SPINLOCK(qdisc_stab_lock);

static const struct nla_policy stab_policy[TCA_STAB_MAX + 1] = {
	[TCA_STAB_BASE]	= { .len = sizeof(struct tc_sizespec) },
	[TCA_STAB_DATA] = { .type = NLA_BINARY },
};

static struct qdisc_size_table *qdisc_get_stab(struct nlattr *opt)
{
	struct nlattr *tb[TCA_STAB_MAX + 1];
	struct qdisc_size_table *stab;
	struct tc_sizespec *s;
	unsigned int tsize = 0;
	u16 *tab = NULL;
	int err;

	err = nla_parse_nested(tb, TCA_STAB_MAX, opt, stab_policy);
	if (err < 0)
		return ERR_PTR(err);
	if (!tb[TCA_STAB_BASE])
		return ERR_PTR(-EINVAL);

	s = nla_data(tb[TCA_STAB_BASE]);

	if (s->tsize > 0) {
		if (!tb[TCA_STAB_DATA])
			return ERR_PTR(-EINVAL);
		tab = nla_data(tb[TCA_STAB_DATA]);
		tsize = nla_len(tb[TCA_STAB_DATA]) / sizeof(u16);
	}

	if (!s || tsize != s->tsize || (!tab && tsize > 0))
		return ERR_PTR(-EINVAL);

357
	spin_lock(&qdisc_stab_lock);
358 359 360 361 362 363 364

	list_for_each_entry(stab, &qdisc_stab_list, list) {
		if (memcmp(&stab->szopts, s, sizeof(*s)))
			continue;
		if (tsize > 0 && memcmp(stab->data, tab, tsize * sizeof(u16)))
			continue;
		stab->refcnt++;
365
		spin_unlock(&qdisc_stab_lock);
366 367 368
		return stab;
	}

369
	spin_unlock(&qdisc_stab_lock);
370 371 372 373 374 375 376 377 378 379

	stab = kmalloc(sizeof(*stab) + tsize * sizeof(u16), GFP_KERNEL);
	if (!stab)
		return ERR_PTR(-ENOMEM);

	stab->refcnt = 1;
	stab->szopts = *s;
	if (tsize > 0)
		memcpy(stab->data, tab, tsize * sizeof(u16));

380
	spin_lock(&qdisc_stab_lock);
381
	list_add_tail(&stab->list, &qdisc_stab_list);
382
	spin_unlock(&qdisc_stab_lock);
383 384 385 386 387 388 389 390 391

	return stab;
}

void qdisc_put_stab(struct qdisc_size_table *tab)
{
	if (!tab)
		return;

392
	spin_lock(&qdisc_stab_lock);
393 394 395 396 397 398

	if (--tab->refcnt == 0) {
		list_del(&tab->list);
		kfree(tab);
	}

399
	spin_unlock(&qdisc_stab_lock);
400 401 402 403 404 405 406 407
}
EXPORT_SYMBOL(qdisc_put_stab);

static int qdisc_dump_stab(struct sk_buff *skb, struct qdisc_size_table *stab)
{
	struct nlattr *nest;

	nest = nla_nest_start(skb, TCA_STAB);
408 409
	if (nest == NULL)
		goto nla_put_failure;
410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446
	NLA_PUT(skb, TCA_STAB_BASE, sizeof(stab->szopts), &stab->szopts);
	nla_nest_end(skb, nest);

	return skb->len;

nla_put_failure:
	return -1;
}

void qdisc_calculate_pkt_len(struct sk_buff *skb, struct qdisc_size_table *stab)
{
	int pkt_len, slot;

	pkt_len = skb->len + stab->szopts.overhead;
	if (unlikely(!stab->szopts.tsize))
		goto out;

	slot = pkt_len + stab->szopts.cell_align;
	if (unlikely(slot < 0))
		slot = 0;

	slot >>= stab->szopts.cell_log;
	if (likely(slot < stab->szopts.tsize))
		pkt_len = stab->data[slot];
	else
		pkt_len = stab->data[stab->szopts.tsize - 1] *
				(slot / stab->szopts.tsize) +
				stab->data[slot % stab->szopts.tsize];

	pkt_len <<= stab->szopts.size_log;
out:
	if (unlikely(pkt_len < 1))
		pkt_len = 1;
	qdisc_skb_cb(skb)->pkt_len = pkt_len;
}
EXPORT_SYMBOL(qdisc_calculate_pkt_len);

447 448 449 450 451 452
static enum hrtimer_restart qdisc_watchdog(struct hrtimer *timer)
{
	struct qdisc_watchdog *wd = container_of(timer, struct qdisc_watchdog,
						 timer);

	wd->qdisc->flags &= ~TCQ_F_THROTTLED;
453
	__netif_schedule(qdisc_root(wd->qdisc));
454

455 456 457 458 459 460 461 462 463 464 465 466 467 468 469
	return HRTIMER_NORESTART;
}

void qdisc_watchdog_init(struct qdisc_watchdog *wd, struct Qdisc *qdisc)
{
	hrtimer_init(&wd->timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
	wd->timer.function = qdisc_watchdog;
	wd->qdisc = qdisc;
}
EXPORT_SYMBOL(qdisc_watchdog_init);

void qdisc_watchdog_schedule(struct qdisc_watchdog *wd, psched_time_t expires)
{
	ktime_t time;

470 471 472 473
	if (test_bit(__QDISC_STATE_DEACTIVATED,
		     &qdisc_root_sleeping(wd->qdisc)->state))
		return;

474 475 476 477 478 479 480 481 482 483 484 485 486
	wd->qdisc->flags |= TCQ_F_THROTTLED;
	time = ktime_set(0, 0);
	time = ktime_add_ns(time, PSCHED_US2NS(expires));
	hrtimer_start(&wd->timer, time, HRTIMER_MODE_ABS);
}
EXPORT_SYMBOL(qdisc_watchdog_schedule);

void qdisc_watchdog_cancel(struct qdisc_watchdog *wd)
{
	hrtimer_cancel(&wd->timer);
	wd->qdisc->flags &= ~TCQ_F_THROTTLED;
}
EXPORT_SYMBOL(qdisc_watchdog_cancel);
L
Linus Torvalds 已提交
487

488
static struct hlist_head *qdisc_class_hash_alloc(unsigned int n)
489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591
{
	unsigned int size = n * sizeof(struct hlist_head), i;
	struct hlist_head *h;

	if (size <= PAGE_SIZE)
		h = kmalloc(size, GFP_KERNEL);
	else
		h = (struct hlist_head *)
			__get_free_pages(GFP_KERNEL, get_order(size));

	if (h != NULL) {
		for (i = 0; i < n; i++)
			INIT_HLIST_HEAD(&h[i]);
	}
	return h;
}

static void qdisc_class_hash_free(struct hlist_head *h, unsigned int n)
{
	unsigned int size = n * sizeof(struct hlist_head);

	if (size <= PAGE_SIZE)
		kfree(h);
	else
		free_pages((unsigned long)h, get_order(size));
}

void qdisc_class_hash_grow(struct Qdisc *sch, struct Qdisc_class_hash *clhash)
{
	struct Qdisc_class_common *cl;
	struct hlist_node *n, *next;
	struct hlist_head *nhash, *ohash;
	unsigned int nsize, nmask, osize;
	unsigned int i, h;

	/* Rehash when load factor exceeds 0.75 */
	if (clhash->hashelems * 4 <= clhash->hashsize * 3)
		return;
	nsize = clhash->hashsize * 2;
	nmask = nsize - 1;
	nhash = qdisc_class_hash_alloc(nsize);
	if (nhash == NULL)
		return;

	ohash = clhash->hash;
	osize = clhash->hashsize;

	sch_tree_lock(sch);
	for (i = 0; i < osize; i++) {
		hlist_for_each_entry_safe(cl, n, next, &ohash[i], hnode) {
			h = qdisc_class_hash(cl->classid, nmask);
			hlist_add_head(&cl->hnode, &nhash[h]);
		}
	}
	clhash->hash     = nhash;
	clhash->hashsize = nsize;
	clhash->hashmask = nmask;
	sch_tree_unlock(sch);

	qdisc_class_hash_free(ohash, osize);
}
EXPORT_SYMBOL(qdisc_class_hash_grow);

int qdisc_class_hash_init(struct Qdisc_class_hash *clhash)
{
	unsigned int size = 4;

	clhash->hash = qdisc_class_hash_alloc(size);
	if (clhash->hash == NULL)
		return -ENOMEM;
	clhash->hashsize  = size;
	clhash->hashmask  = size - 1;
	clhash->hashelems = 0;
	return 0;
}
EXPORT_SYMBOL(qdisc_class_hash_init);

void qdisc_class_hash_destroy(struct Qdisc_class_hash *clhash)
{
	qdisc_class_hash_free(clhash->hash, clhash->hashsize);
}
EXPORT_SYMBOL(qdisc_class_hash_destroy);

void qdisc_class_hash_insert(struct Qdisc_class_hash *clhash,
			     struct Qdisc_class_common *cl)
{
	unsigned int h;

	INIT_HLIST_NODE(&cl->hnode);
	h = qdisc_class_hash(cl->classid, clhash->hashmask);
	hlist_add_head(&cl->hnode, &clhash->hash[h]);
	clhash->hashelems++;
}
EXPORT_SYMBOL(qdisc_class_hash_insert);

void qdisc_class_hash_remove(struct Qdisc_class_hash *clhash,
			     struct Qdisc_class_common *cl)
{
	hlist_del(&cl->hnode);
	clhash->hashelems--;
}
EXPORT_SYMBOL(qdisc_class_hash_remove);

L
Linus Torvalds 已提交
592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607
/* Allocate an unique handle from space managed by kernel */

static u32 qdisc_alloc_handle(struct net_device *dev)
{
	int i = 0x10000;
	static u32 autohandle = TC_H_MAKE(0x80000000U, 0);

	do {
		autohandle += TC_H_MAKE(0x10000U, 0);
		if (autohandle == TC_H_MAKE(TC_H_ROOT, 0))
			autohandle = TC_H_MAKE(0x80000000U, 0);
	} while	(qdisc_lookup(dev, autohandle) && --i > 0);

	return i>0 ? autohandle : 0;
}

608
/* Attach toplevel qdisc to device queue. */
L
Linus Torvalds 已提交
609

610 611
static struct Qdisc *dev_graft_qdisc(struct netdev_queue *dev_queue,
				     struct Qdisc *qdisc)
L
Linus Torvalds 已提交
612
{
613
	struct Qdisc *oqdisc = dev_queue->qdisc_sleeping;
614 615
	spinlock_t *root_lock;

616
	root_lock = qdisc_lock(oqdisc);
617 618
	spin_lock_bh(root_lock);

619 620 621
	/* Prune old scheduler */
	if (oqdisc && atomic_read(&oqdisc->refcnt) <= 1)
		qdisc_reset(oqdisc);
L
Linus Torvalds 已提交
622

623 624 625 626
	/* ... and graft new one */
	if (qdisc == NULL)
		qdisc = &noop_qdisc;
	dev_queue->qdisc_sleeping = qdisc;
627
	rcu_assign_pointer(dev_queue->qdisc, &noop_qdisc);
L
Linus Torvalds 已提交
628

629
	spin_unlock_bh(root_lock);
L
Linus Torvalds 已提交
630 631 632 633

	return oqdisc;
}

634 635
void qdisc_tree_decrease_qlen(struct Qdisc *sch, unsigned int n)
{
636
	const struct Qdisc_class_ops *cops;
637 638 639 640 641 642
	unsigned long cl;
	u32 parentid;

	if (n == 0)
		return;
	while ((parentid = sch->parent)) {
643 644 645
		if (TC_H_MAJ(parentid) == TC_H_MAJ(TC_H_INGRESS))
			return;

646
		sch = qdisc_lookup(qdisc_dev(sch), TC_H_MAJ(parentid));
647 648 649 650
		if (sch == NULL) {
			WARN_ON(parentid != TC_H_ROOT);
			return;
		}
651 652 653 654 655 656 657 658 659 660
		cops = sch->ops->cl_ops;
		if (cops->qlen_notify) {
			cl = cops->get(sch, parentid);
			cops->qlen_notify(sch, cl);
			cops->put(sch, cl);
		}
		sch->q.qlen -= n;
	}
}
EXPORT_SYMBOL(qdisc_tree_decrease_qlen);
L
Linus Torvalds 已提交
661

662 663 664 665 666
static void notify_and_destroy(struct sk_buff *skb, struct nlmsghdr *n, u32 clid,
			       struct Qdisc *old, struct Qdisc *new)
{
	if (new || old)
		qdisc_notify(skb, n, clid, old, new);
L
Linus Torvalds 已提交
667

668
	if (old)
669 670 671 672 673 674 675 676 677 678
		qdisc_destroy(old);
}

/* Graft qdisc "new" to class "classid" of qdisc "parent" or
 * to device "dev".
 *
 * When appropriate send a netlink notification using 'skb'
 * and "n".
 *
 * On success, destroy old qdisc.
L
Linus Torvalds 已提交
679 680 681
 */

static int qdisc_graft(struct net_device *dev, struct Qdisc *parent,
682 683
		       struct sk_buff *skb, struct nlmsghdr *n, u32 classid,
		       struct Qdisc *new, struct Qdisc *old)
L
Linus Torvalds 已提交
684
{
685
	struct Qdisc *q = old;
L
Linus Torvalds 已提交
686 687
	int err = 0;

688
	if (parent == NULL) {
689 690 691 692
		unsigned int i, num_q, ingress;

		ingress = 0;
		num_q = dev->num_tx_queues;
693 694
		if ((q && q->flags & TCQ_F_INGRESS) ||
		    (new && new->flags & TCQ_F_INGRESS)) {
695 696 697 698 699 700 701 702 703 704 705 706 707
			num_q = 1;
			ingress = 1;
		}

		if (dev->flags & IFF_UP)
			dev_deactivate(dev);

		for (i = 0; i < num_q; i++) {
			struct netdev_queue *dev_queue = &dev->rx_queue;

			if (!ingress)
				dev_queue = netdev_get_tx_queue(dev, i);

708 709 710 711
			old = dev_graft_qdisc(dev_queue, new);
			if (new && i > 0)
				atomic_inc(&new->refcnt);

712
			notify_and_destroy(skb, n, classid, old, new);
L
Linus Torvalds 已提交
713
		}
714 715 716

		if (dev->flags & IFF_UP)
			dev_activate(dev);
L
Linus Torvalds 已提交
717
	} else {
718
		const struct Qdisc_class_ops *cops = parent->ops->cl_ops;
L
Linus Torvalds 已提交
719 720 721 722 723 724

		err = -EINVAL;

		if (cops) {
			unsigned long cl = cops->get(parent, classid);
			if (cl) {
725
				err = cops->graft(parent, cl, new, &old);
L
Linus Torvalds 已提交
726 727 728
				cops->put(parent, cl);
			}
		}
729 730
		if (!err)
			notify_and_destroy(skb, n, classid, old, new);
L
Linus Torvalds 已提交
731 732 733 734
	}
	return err;
}

735 736 737 738
/* lockdep annotation is needed for ingress; egress gets it only for name */
static struct lock_class_key qdisc_tx_lock;
static struct lock_class_key qdisc_rx_lock;

L
Linus Torvalds 已提交
739 740 741 742 743 744 745
/*
   Allocate and initialize new qdisc.

   Parameters are passed via opt.
 */

static struct Qdisc *
746 747
qdisc_create(struct net_device *dev, struct netdev_queue *dev_queue,
	     u32 parent, u32 handle, struct nlattr **tca, int *errp)
L
Linus Torvalds 已提交
748 749
{
	int err;
750
	struct nlattr *kind = tca[TCA_KIND];
L
Linus Torvalds 已提交
751 752
	struct Qdisc *sch;
	struct Qdisc_ops *ops;
753
	struct qdisc_size_table *stab;
L
Linus Torvalds 已提交
754 755

	ops = qdisc_lookup_ops(kind);
756
#ifdef CONFIG_MODULES
L
Linus Torvalds 已提交
757 758
	if (ops == NULL && kind != NULL) {
		char name[IFNAMSIZ];
759
		if (nla_strlcpy(name, kind, IFNAMSIZ) < IFNAMSIZ) {
L
Linus Torvalds 已提交
760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783
			/* We dropped the RTNL semaphore in order to
			 * perform the module load.  So, even if we
			 * succeeded in loading the module we have to
			 * tell the caller to replay the request.  We
			 * indicate this using -EAGAIN.
			 * We replay the request because the device may
			 * go away in the mean time.
			 */
			rtnl_unlock();
			request_module("sch_%s", name);
			rtnl_lock();
			ops = qdisc_lookup_ops(kind);
			if (ops != NULL) {
				/* We will try again qdisc_lookup_ops,
				 * so don't keep a reference.
				 */
				module_put(ops->owner);
				err = -EAGAIN;
				goto err_out;
			}
		}
	}
#endif

784
	err = -ENOENT;
L
Linus Torvalds 已提交
785 786 787
	if (ops == NULL)
		goto err_out;

788
	sch = qdisc_alloc(dev_queue, ops);
789 790
	if (IS_ERR(sch)) {
		err = PTR_ERR(sch);
L
Linus Torvalds 已提交
791
		goto err_out2;
792
	}
L
Linus Torvalds 已提交
793

794 795
	sch->parent = parent;

796
	if (handle == TC_H_INGRESS) {
L
Linus Torvalds 已提交
797
		sch->flags |= TCQ_F_INGRESS;
798
		handle = TC_H_MAKE(TC_H_INGRESS, 0);
799
		lockdep_set_class(qdisc_lock(sch), &qdisc_rx_lock);
800 801 802 803 804 805 806
	} else {
		if (handle == 0) {
			handle = qdisc_alloc_handle(dev);
			err = -ENOMEM;
			if (handle == 0)
				goto err_out3;
		}
807
		lockdep_set_class(qdisc_lock(sch), &qdisc_tx_lock);
L
Linus Torvalds 已提交
808 809
	}

810
	sch->handle = handle;
L
Linus Torvalds 已提交
811

812
	if (!ops->init || (err = ops->init(sch, tca[TCA_OPTIONS])) == 0) {
813 814 815 816 817 818 819 820
		if (tca[TCA_STAB]) {
			stab = qdisc_get_stab(tca[TCA_STAB]);
			if (IS_ERR(stab)) {
				err = PTR_ERR(stab);
				goto err_out3;
			}
			sch->stab = stab;
		}
821
		if (tca[TCA_RATE]) {
822 823 824 825 826 827 828 829
			spinlock_t *root_lock;

			if ((sch->parent != TC_H_ROOT) &&
			    !(sch->flags & TCQ_F_INGRESS))
				root_lock = qdisc_root_sleeping_lock(sch);
			else
				root_lock = qdisc_lock(sch);

830
			err = gen_new_estimator(&sch->bstats, &sch->rate_est,
831
						root_lock, tca[TCA_RATE]);
832 833 834 835 836 837 838 839 840 841 842
			if (err) {
				/*
				 * Any broken qdiscs that would require
				 * a ops->reset() here? The qdisc was never
				 * in action so it shouldn't be necessary.
				 */
				if (ops->destroy)
					ops->destroy(sch);
				goto err_out3;
			}
		}
843 844

		qdisc_list_add(sch);
L
Linus Torvalds 已提交
845 846 847 848

		return sch;
	}
err_out3:
849
	qdisc_put_stab(sch->stab);
L
Linus Torvalds 已提交
850
	dev_put(dev);
851
	kfree((char *) sch - sch->padded);
L
Linus Torvalds 已提交
852 853 854 855 856 857 858
err_out2:
	module_put(ops->owner);
err_out:
	*errp = err;
	return NULL;
}

859
static int qdisc_change(struct Qdisc *sch, struct nlattr **tca)
L
Linus Torvalds 已提交
860
{
861 862
	struct qdisc_size_table *stab = NULL;
	int err = 0;
L
Linus Torvalds 已提交
863

864
	if (tca[TCA_OPTIONS]) {
L
Linus Torvalds 已提交
865 866
		if (sch->ops->change == NULL)
			return -EINVAL;
867
		err = sch->ops->change(sch, tca[TCA_OPTIONS]);
L
Linus Torvalds 已提交
868 869 870
		if (err)
			return err;
	}
871 872 873 874 875 876 877 878 879 880

	if (tca[TCA_STAB]) {
		stab = qdisc_get_stab(tca[TCA_STAB]);
		if (IS_ERR(stab))
			return PTR_ERR(stab);
	}

	qdisc_put_stab(sch->stab);
	sch->stab = stab;

881
	if (tca[TCA_RATE])
882 883
		/* NB: ignores errors from replace_estimator
		   because change can't be undone. */
L
Linus Torvalds 已提交
884
		gen_replace_estimator(&sch->bstats, &sch->rate_est,
885 886 887
					    qdisc_root_sleeping_lock(sch),
					    tca[TCA_RATE]);

L
Linus Torvalds 已提交
888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918
	return 0;
}

struct check_loop_arg
{
	struct qdisc_walker 	w;
	struct Qdisc		*p;
	int			depth;
};

static int check_loop_fn(struct Qdisc *q, unsigned long cl, struct qdisc_walker *w);

static int check_loop(struct Qdisc *q, struct Qdisc *p, int depth)
{
	struct check_loop_arg	arg;

	if (q->ops->cl_ops == NULL)
		return 0;

	arg.w.stop = arg.w.skip = arg.w.count = 0;
	arg.w.fn = check_loop_fn;
	arg.depth = depth;
	arg.p = p;
	q->ops->cl_ops->walk(q, &arg.w);
	return arg.w.stop ? -ELOOP : 0;
}

static int
check_loop_fn(struct Qdisc *q, unsigned long cl, struct qdisc_walker *w)
{
	struct Qdisc *leaf;
919
	const struct Qdisc_class_ops *cops = q->ops->cl_ops;
L
Linus Torvalds 已提交
920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936
	struct check_loop_arg *arg = (struct check_loop_arg *)w;

	leaf = cops->leaf(q, cl);
	if (leaf) {
		if (leaf == arg->p || arg->depth > 7)
			return -ELOOP;
		return check_loop(leaf, arg->p, arg->depth + 1);
	}
	return 0;
}

/*
 * Delete/get qdisc.
 */

static int tc_get_qdisc(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
{
937
	struct net *net = sock_net(skb->sk);
L
Linus Torvalds 已提交
938
	struct tcmsg *tcm = NLMSG_DATA(n);
939
	struct nlattr *tca[TCA_MAX + 1];
L
Linus Torvalds 已提交
940 941 942 943 944 945
	struct net_device *dev;
	u32 clid = tcm->tcm_parent;
	struct Qdisc *q = NULL;
	struct Qdisc *p = NULL;
	int err;

946 947 948
	if (net != &init_net)
		return -EINVAL;

949
	if ((dev = __dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
L
Linus Torvalds 已提交
950 951
		return -ENODEV;

952 953 954 955
	err = nlmsg_parse(n, sizeof(*tcm), tca, TCA_MAX, NULL);
	if (err < 0)
		return err;

L
Linus Torvalds 已提交
956 957 958 959 960 961 962
	if (clid) {
		if (clid != TC_H_ROOT) {
			if (TC_H_MAJ(clid) != TC_H_MAJ(TC_H_INGRESS)) {
				if ((p = qdisc_lookup(dev, TC_H_MAJ(clid))) == NULL)
					return -ENOENT;
				q = qdisc_leaf(p, clid);
			} else { /* ingress */
963
				q = dev->rx_queue.qdisc_sleeping;
964
			}
L
Linus Torvalds 已提交
965
		} else {
966 967
			struct netdev_queue *dev_queue;
			dev_queue = netdev_get_tx_queue(dev, 0);
968
			q = dev_queue->qdisc_sleeping;
L
Linus Torvalds 已提交
969 970 971 972 973 974 975 976 977 978 979
		}
		if (!q)
			return -ENOENT;

		if (tcm->tcm_handle && q->handle != tcm->tcm_handle)
			return -EINVAL;
	} else {
		if ((q = qdisc_lookup(dev, tcm->tcm_handle)) == NULL)
			return -ENOENT;
	}

980
	if (tca[TCA_KIND] && nla_strcmp(tca[TCA_KIND], q->ops->id))
L
Linus Torvalds 已提交
981 982 983 984 985 986 987
		return -EINVAL;

	if (n->nlmsg_type == RTM_DELQDISC) {
		if (!clid)
			return -EINVAL;
		if (q->handle == 0)
			return -ENOENT;
988
		if ((err = qdisc_graft(dev, p, skb, n, clid, NULL, q)) != 0)
L
Linus Torvalds 已提交
989 990 991 992 993 994 995 996 997 998 999 1000 1001
			return err;
	} else {
		qdisc_notify(skb, n, clid, NULL, q);
	}
	return 0;
}

/*
   Create/change qdisc.
 */

static int tc_modify_qdisc(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
{
1002
	struct net *net = sock_net(skb->sk);
L
Linus Torvalds 已提交
1003
	struct tcmsg *tcm;
1004
	struct nlattr *tca[TCA_MAX + 1];
L
Linus Torvalds 已提交
1005 1006 1007 1008 1009
	struct net_device *dev;
	u32 clid;
	struct Qdisc *q, *p;
	int err;

1010 1011 1012
	if (net != &init_net)
		return -EINVAL;

L
Linus Torvalds 已提交
1013 1014 1015 1016 1017 1018
replay:
	/* Reinit, just in case something touches this. */
	tcm = NLMSG_DATA(n);
	clid = tcm->tcm_parent;
	q = p = NULL;

1019
	if ((dev = __dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
L
Linus Torvalds 已提交
1020 1021
		return -ENODEV;

1022 1023 1024 1025
	err = nlmsg_parse(n, sizeof(*tcm), tca, TCA_MAX, NULL);
	if (err < 0)
		return err;

L
Linus Torvalds 已提交
1026 1027 1028 1029 1030 1031 1032
	if (clid) {
		if (clid != TC_H_ROOT) {
			if (clid != TC_H_INGRESS) {
				if ((p = qdisc_lookup(dev, TC_H_MAJ(clid))) == NULL)
					return -ENOENT;
				q = qdisc_leaf(p, clid);
			} else { /*ingress */
1033
				q = dev->rx_queue.qdisc_sleeping;
L
Linus Torvalds 已提交
1034 1035
			}
		} else {
1036 1037
			struct netdev_queue *dev_queue;
			dev_queue = netdev_get_tx_queue(dev, 0);
1038
			q = dev_queue->qdisc_sleeping;
L
Linus Torvalds 已提交
1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
		}

		/* It may be default qdisc, ignore it */
		if (q && q->handle == 0)
			q = NULL;

		if (!q || !tcm->tcm_handle || q->handle != tcm->tcm_handle) {
			if (tcm->tcm_handle) {
				if (q && !(n->nlmsg_flags&NLM_F_REPLACE))
					return -EEXIST;
				if (TC_H_MIN(tcm->tcm_handle))
					return -EINVAL;
				if ((q = qdisc_lookup(dev, tcm->tcm_handle)) == NULL)
					goto create_n_graft;
				if (n->nlmsg_flags&NLM_F_EXCL)
					return -EEXIST;
1055
				if (tca[TCA_KIND] && nla_strcmp(tca[TCA_KIND], q->ops->id))
L
Linus Torvalds 已提交
1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087
					return -EINVAL;
				if (q == p ||
				    (p && check_loop(q, p, 0)))
					return -ELOOP;
				atomic_inc(&q->refcnt);
				goto graft;
			} else {
				if (q == NULL)
					goto create_n_graft;

				/* This magic test requires explanation.
				 *
				 *   We know, that some child q is already
				 *   attached to this parent and have choice:
				 *   either to change it or to create/graft new one.
				 *
				 *   1. We are allowed to create/graft only
				 *   if CREATE and REPLACE flags are set.
				 *
				 *   2. If EXCL is set, requestor wanted to say,
				 *   that qdisc tcm_handle is not expected
				 *   to exist, so that we choose create/graft too.
				 *
				 *   3. The last case is when no flags are set.
				 *   Alas, it is sort of hole in API, we
				 *   cannot decide what to do unambiguously.
				 *   For now we select create/graft, if
				 *   user gave KIND, which does not match existing.
				 */
				if ((n->nlmsg_flags&NLM_F_CREATE) &&
				    (n->nlmsg_flags&NLM_F_REPLACE) &&
				    ((n->nlmsg_flags&NLM_F_EXCL) ||
1088 1089
				     (tca[TCA_KIND] &&
				      nla_strcmp(tca[TCA_KIND], q->ops->id))))
L
Linus Torvalds 已提交
1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103
					goto create_n_graft;
			}
		}
	} else {
		if (!tcm->tcm_handle)
			return -EINVAL;
		q = qdisc_lookup(dev, tcm->tcm_handle);
	}

	/* Change qdisc parameters */
	if (q == NULL)
		return -ENOENT;
	if (n->nlmsg_flags&NLM_F_EXCL)
		return -EEXIST;
1104
	if (tca[TCA_KIND] && nla_strcmp(tca[TCA_KIND], q->ops->id))
L
Linus Torvalds 已提交
1105 1106 1107 1108 1109 1110 1111 1112 1113 1114
		return -EINVAL;
	err = qdisc_change(q, tca);
	if (err == 0)
		qdisc_notify(skb, n, clid, NULL, q);
	return err;

create_n_graft:
	if (!(n->nlmsg_flags&NLM_F_CREATE))
		return -ENOENT;
	if (clid == TC_H_INGRESS)
1115 1116
		q = qdisc_create(dev, &dev->rx_queue,
				 tcm->tcm_parent, tcm->tcm_parent,
1117
				 tca, &err);
1118
	else
1119
		q = qdisc_create(dev, netdev_get_tx_queue(dev, 0),
1120
				 tcm->tcm_parent, tcm->tcm_handle,
1121
				 tca, &err);
L
Linus Torvalds 已提交
1122 1123 1124 1125 1126 1127 1128
	if (q == NULL) {
		if (err == -EAGAIN)
			goto replay;
		return err;
	}

graft:
1129 1130 1131 1132 1133
	err = qdisc_graft(dev, p, skb, n, clid, q, NULL);
	if (err) {
		if (q)
			qdisc_destroy(q);
		return err;
L
Linus Torvalds 已提交
1134
	}
1135

L
Linus Torvalds 已提交
1136 1137 1138 1139
	return 0;
}

static int tc_fill_qdisc(struct sk_buff *skb, struct Qdisc *q, u32 clid,
J
Jamal Hadi Salim 已提交
1140
			 u32 pid, u32 seq, u16 flags, int event)
L
Linus Torvalds 已提交
1141 1142 1143
{
	struct tcmsg *tcm;
	struct nlmsghdr  *nlh;
1144
	unsigned char *b = skb_tail_pointer(skb);
L
Linus Torvalds 已提交
1145 1146
	struct gnet_dump d;

J
Jamal Hadi Salim 已提交
1147
	nlh = NLMSG_NEW(skb, pid, seq, event, sizeof(*tcm), flags);
L
Linus Torvalds 已提交
1148 1149
	tcm = NLMSG_DATA(nlh);
	tcm->tcm_family = AF_UNSPEC;
1150 1151
	tcm->tcm__pad1 = 0;
	tcm->tcm__pad2 = 0;
1152
	tcm->tcm_ifindex = qdisc_dev(q)->ifindex;
L
Linus Torvalds 已提交
1153 1154 1155
	tcm->tcm_parent = clid;
	tcm->tcm_handle = q->handle;
	tcm->tcm_info = atomic_read(&q->refcnt);
1156
	NLA_PUT_STRING(skb, TCA_KIND, q->ops->id);
L
Linus Torvalds 已提交
1157
	if (q->ops->dump && q->ops->dump(q, skb) < 0)
1158
		goto nla_put_failure;
L
Linus Torvalds 已提交
1159 1160
	q->qstats.qlen = q->q.qlen;

1161 1162 1163
	if (q->stab && qdisc_dump_stab(skb, q->stab) < 0)
		goto nla_put_failure;

1164 1165
	if (gnet_stats_start_copy_compat(skb, TCA_STATS2, TCA_STATS, TCA_XSTATS,
					 qdisc_root_sleeping_lock(q), &d) < 0)
1166
		goto nla_put_failure;
L
Linus Torvalds 已提交
1167 1168

	if (q->ops->dump_stats && q->ops->dump_stats(q, &d) < 0)
1169
		goto nla_put_failure;
L
Linus Torvalds 已提交
1170 1171 1172 1173

	if (gnet_stats_copy_basic(&d, &q->bstats) < 0 ||
	    gnet_stats_copy_rate_est(&d, &q->rate_est) < 0 ||
	    gnet_stats_copy_queue(&d, &q->qstats) < 0)
1174
		goto nla_put_failure;
1175

L
Linus Torvalds 已提交
1176
	if (gnet_stats_finish_copy(&d) < 0)
1177
		goto nla_put_failure;
1178

1179
	nlh->nlmsg_len = skb_tail_pointer(skb) - b;
L
Linus Torvalds 已提交
1180 1181 1182
	return skb->len;

nlmsg_failure:
1183
nla_put_failure:
1184
	nlmsg_trim(skb, b);
L
Linus Torvalds 已提交
1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207
	return -1;
}

static int qdisc_notify(struct sk_buff *oskb, struct nlmsghdr *n,
			u32 clid, struct Qdisc *old, struct Qdisc *new)
{
	struct sk_buff *skb;
	u32 pid = oskb ? NETLINK_CB(oskb).pid : 0;

	skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
	if (!skb)
		return -ENOBUFS;

	if (old && old->handle) {
		if (tc_fill_qdisc(skb, old, clid, pid, n->nlmsg_seq, 0, RTM_DELQDISC) < 0)
			goto err_out;
	}
	if (new) {
		if (tc_fill_qdisc(skb, new, clid, pid, n->nlmsg_seq, old ? NLM_F_REPLACE : 0, RTM_NEWQDISC) < 0)
			goto err_out;
	}

	if (skb->len)
1208
		return rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
L
Linus Torvalds 已提交
1209 1210 1211 1212 1213 1214

err_out:
	kfree_skb(skb);
	return -EINVAL;
}

1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259
static bool tc_qdisc_dump_ignore(struct Qdisc *q)
{
	return (q->flags & TCQ_F_BUILTIN) ? true : false;
}

static int tc_dump_qdisc_root(struct Qdisc *root, struct sk_buff *skb,
			      struct netlink_callback *cb,
			      int *q_idx_p, int s_q_idx)
{
	int ret = 0, q_idx = *q_idx_p;
	struct Qdisc *q;

	if (!root)
		return 0;

	q = root;
	if (q_idx < s_q_idx) {
		q_idx++;
	} else {
		if (!tc_qdisc_dump_ignore(q) &&
		    tc_fill_qdisc(skb, q, q->parent, NETLINK_CB(cb->skb).pid,
				  cb->nlh->nlmsg_seq, NLM_F_MULTI, RTM_NEWQDISC) <= 0)
			goto done;
		q_idx++;
	}
	list_for_each_entry(q, &root->list, list) {
		if (q_idx < s_q_idx) {
			q_idx++;
			continue;
		}
		if (!tc_qdisc_dump_ignore(q) && 
		    tc_fill_qdisc(skb, q, q->parent, NETLINK_CB(cb->skb).pid,
				  cb->nlh->nlmsg_seq, NLM_F_MULTI, RTM_NEWQDISC) <= 0)
			goto done;
		q_idx++;
	}

out:
	*q_idx_p = q_idx;
	return ret;
done:
	ret = -1;
	goto out;
}

L
Linus Torvalds 已提交
1260 1261
static int tc_dump_qdisc(struct sk_buff *skb, struct netlink_callback *cb)
{
1262
	struct net *net = sock_net(skb->sk);
L
Linus Torvalds 已提交
1263 1264 1265 1266
	int idx, q_idx;
	int s_idx, s_q_idx;
	struct net_device *dev;

1267 1268 1269
	if (net != &init_net)
		return 0;

L
Linus Torvalds 已提交
1270 1271 1272
	s_idx = cb->args[0];
	s_q_idx = q_idx = cb->args[1];
	read_lock(&dev_base_lock);
1273
	idx = 0;
1274
	for_each_netdev(&init_net, dev) {
1275 1276
		struct netdev_queue *dev_queue;

L
Linus Torvalds 已提交
1277
		if (idx < s_idx)
1278
			goto cont;
L
Linus Torvalds 已提交
1279 1280 1281
		if (idx > s_idx)
			s_q_idx = 0;
		q_idx = 0;
1282 1283

		dev_queue = netdev_get_tx_queue(dev, 0);
1284
		if (tc_dump_qdisc_root(dev_queue->qdisc_sleeping, skb, cb, &q_idx, s_q_idx) < 0)
1285 1286 1287
			goto done;

		dev_queue = &dev->rx_queue;
1288
		if (tc_dump_qdisc_root(dev_queue->qdisc_sleeping, skb, cb, &q_idx, s_q_idx) < 0)
1289 1290
			goto done;

1291 1292
cont:
		idx++;
L
Linus Torvalds 已提交
1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313
	}

done:
	read_unlock(&dev_base_lock);

	cb->args[0] = idx;
	cb->args[1] = q_idx;

	return skb->len;
}



/************************************************
 *	Traffic classes manipulation.		*
 ************************************************/



static int tc_ctl_tclass(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
{
1314
	struct net *net = sock_net(skb->sk);
1315
	struct netdev_queue *dev_queue;
L
Linus Torvalds 已提交
1316
	struct tcmsg *tcm = NLMSG_DATA(n);
1317
	struct nlattr *tca[TCA_MAX + 1];
L
Linus Torvalds 已提交
1318 1319
	struct net_device *dev;
	struct Qdisc *q = NULL;
1320
	const struct Qdisc_class_ops *cops;
L
Linus Torvalds 已提交
1321 1322 1323 1324 1325 1326 1327
	unsigned long cl = 0;
	unsigned long new_cl;
	u32 pid = tcm->tcm_parent;
	u32 clid = tcm->tcm_handle;
	u32 qid = TC_H_MAJ(clid);
	int err;

1328 1329 1330
	if (net != &init_net)
		return -EINVAL;

1331
	if ((dev = __dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
L
Linus Torvalds 已提交
1332 1333
		return -ENODEV;

1334 1335 1336 1337
	err = nlmsg_parse(n, sizeof(*tcm), tca, TCA_MAX, NULL);
	if (err < 0)
		return err;

L
Linus Torvalds 已提交
1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352
	/*
	   parent == TC_H_UNSPEC - unspecified parent.
	   parent == TC_H_ROOT   - class is root, which has no parent.
	   parent == X:0	 - parent is root class.
	   parent == X:Y	 - parent is a node in hierarchy.
	   parent == 0:Y	 - parent is X:Y, where X:0 is qdisc.

	   handle == 0:0	 - generate handle from kernel pool.
	   handle == 0:Y	 - class is X:Y, where X:0 is qdisc.
	   handle == X:Y	 - clear.
	   handle == X:0	 - root class.
	 */

	/* Step 1. Determine qdisc handle X:0 */

1353
	dev_queue = netdev_get_tx_queue(dev, 0);
L
Linus Torvalds 已提交
1354 1355 1356 1357 1358 1359 1360 1361 1362 1363
	if (pid != TC_H_ROOT) {
		u32 qid1 = TC_H_MAJ(pid);

		if (qid && qid1) {
			/* If both majors are known, they must be identical. */
			if (qid != qid1)
				return -EINVAL;
		} else if (qid1) {
			qid = qid1;
		} else if (qid == 0)
1364
			qid = dev_queue->qdisc_sleeping->handle;
L
Linus Torvalds 已提交
1365 1366 1367 1368 1369 1370 1371 1372 1373 1374

		/* Now qid is genuine qdisc handle consistent
		   both with parent and child.

		   TC_H_MAJ(pid) still may be unspecified, complete it now.
		 */
		if (pid)
			pid = TC_H_MAKE(qid, pid);
	} else {
		if (qid == 0)
1375
			qid = dev_queue->qdisc_sleeping->handle;
L
Linus Torvalds 已提交
1376 1377 1378
	}

	/* OK. Locate qdisc */
1379
	if ((q = qdisc_lookup(dev, qid)) == NULL)
L
Linus Torvalds 已提交
1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402
		return -ENOENT;

	/* An check that it supports classes */
	cops = q->ops->cl_ops;
	if (cops == NULL)
		return -EINVAL;

	/* Now try to get class */
	if (clid == 0) {
		if (pid == TC_H_ROOT)
			clid = qid;
	} else
		clid = TC_H_MAKE(qid, clid);

	if (clid)
		cl = cops->get(q, clid);

	if (cl == 0) {
		err = -ENOENT;
		if (n->nlmsg_type != RTM_NEWTCLASS || !(n->nlmsg_flags&NLM_F_CREATE))
			goto out;
	} else {
		switch (n->nlmsg_type) {
1403
		case RTM_NEWTCLASS:
L
Linus Torvalds 已提交
1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436
			err = -EEXIST;
			if (n->nlmsg_flags&NLM_F_EXCL)
				goto out;
			break;
		case RTM_DELTCLASS:
			err = cops->delete(q, cl);
			if (err == 0)
				tclass_notify(skb, n, q, cl, RTM_DELTCLASS);
			goto out;
		case RTM_GETTCLASS:
			err = tclass_notify(skb, n, q, cl, RTM_NEWTCLASS);
			goto out;
		default:
			err = -EINVAL;
			goto out;
		}
	}

	new_cl = cl;
	err = cops->change(q, clid, pid, tca, &new_cl);
	if (err == 0)
		tclass_notify(skb, n, q, new_cl, RTM_NEWTCLASS);

out:
	if (cl)
		cops->put(q, cl);

	return err;
}


static int tc_fill_tclass(struct sk_buff *skb, struct Qdisc *q,
			  unsigned long cl,
J
Jamal Hadi Salim 已提交
1437
			  u32 pid, u32 seq, u16 flags, int event)
L
Linus Torvalds 已提交
1438 1439 1440
{
	struct tcmsg *tcm;
	struct nlmsghdr  *nlh;
1441
	unsigned char *b = skb_tail_pointer(skb);
L
Linus Torvalds 已提交
1442
	struct gnet_dump d;
1443
	const struct Qdisc_class_ops *cl_ops = q->ops->cl_ops;
L
Linus Torvalds 已提交
1444

J
Jamal Hadi Salim 已提交
1445
	nlh = NLMSG_NEW(skb, pid, seq, event, sizeof(*tcm), flags);
L
Linus Torvalds 已提交
1446 1447
	tcm = NLMSG_DATA(nlh);
	tcm->tcm_family = AF_UNSPEC;
1448
	tcm->tcm_ifindex = qdisc_dev(q)->ifindex;
L
Linus Torvalds 已提交
1449 1450 1451
	tcm->tcm_parent = q->handle;
	tcm->tcm_handle = q->handle;
	tcm->tcm_info = 0;
1452
	NLA_PUT_STRING(skb, TCA_KIND, q->ops->id);
L
Linus Torvalds 已提交
1453
	if (cl_ops->dump && cl_ops->dump(q, cl, skb, tcm) < 0)
1454
		goto nla_put_failure;
L
Linus Torvalds 已提交
1455

1456 1457
	if (gnet_stats_start_copy_compat(skb, TCA_STATS2, TCA_STATS, TCA_XSTATS,
					 qdisc_root_sleeping_lock(q), &d) < 0)
1458
		goto nla_put_failure;
L
Linus Torvalds 已提交
1459 1460

	if (cl_ops->dump_stats && cl_ops->dump_stats(q, cl, &d) < 0)
1461
		goto nla_put_failure;
L
Linus Torvalds 已提交
1462 1463

	if (gnet_stats_finish_copy(&d) < 0)
1464
		goto nla_put_failure;
L
Linus Torvalds 已提交
1465

1466
	nlh->nlmsg_len = skb_tail_pointer(skb) - b;
L
Linus Torvalds 已提交
1467 1468 1469
	return skb->len;

nlmsg_failure:
1470
nla_put_failure:
1471
	nlmsg_trim(skb, b);
L
Linus Torvalds 已提交
1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489
	return -1;
}

static int tclass_notify(struct sk_buff *oskb, struct nlmsghdr *n,
			  struct Qdisc *q, unsigned long cl, int event)
{
	struct sk_buff *skb;
	u32 pid = oskb ? NETLINK_CB(oskb).pid : 0;

	skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
	if (!skb)
		return -ENOBUFS;

	if (tc_fill_tclass(skb, q, cl, pid, n->nlmsg_seq, 0, event) < 0) {
		kfree_skb(skb);
		return -EINVAL;
	}

1490
	return rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
L
Linus Torvalds 已提交
1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507
}

struct qdisc_dump_args
{
	struct qdisc_walker w;
	struct sk_buff *skb;
	struct netlink_callback *cb;
};

static int qdisc_class_dump(struct Qdisc *q, unsigned long cl, struct qdisc_walker *arg)
{
	struct qdisc_dump_args *a = (struct qdisc_dump_args *)arg;

	return tc_fill_tclass(a->skb, q, cl, NETLINK_CB(a->cb->skb).pid,
			      a->cb->nlh->nlmsg_seq, NLM_F_MULTI, RTM_NEWTCLASS);
}

1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556
static int tc_dump_tclass_qdisc(struct Qdisc *q, struct sk_buff *skb,
				struct tcmsg *tcm, struct netlink_callback *cb,
				int *t_p, int s_t)
{
	struct qdisc_dump_args arg;

	if (tc_qdisc_dump_ignore(q) ||
	    *t_p < s_t || !q->ops->cl_ops ||
	    (tcm->tcm_parent &&
	     TC_H_MAJ(tcm->tcm_parent) != q->handle)) {
		(*t_p)++;
		return 0;
	}
	if (*t_p > s_t)
		memset(&cb->args[1], 0, sizeof(cb->args)-sizeof(cb->args[0]));
	arg.w.fn = qdisc_class_dump;
	arg.skb = skb;
	arg.cb = cb;
	arg.w.stop  = 0;
	arg.w.skip = cb->args[1];
	arg.w.count = 0;
	q->ops->cl_ops->walk(q, &arg.w);
	cb->args[1] = arg.w.count;
	if (arg.w.stop)
		return -1;
	(*t_p)++;
	return 0;
}

static int tc_dump_tclass_root(struct Qdisc *root, struct sk_buff *skb,
			       struct tcmsg *tcm, struct netlink_callback *cb,
			       int *t_p, int s_t)
{
	struct Qdisc *q;

	if (!root)
		return 0;

	if (tc_dump_tclass_qdisc(root, skb, tcm, cb, t_p, s_t) < 0)
		return -1;

	list_for_each_entry(q, &root->list, list) {
		if (tc_dump_tclass_qdisc(q, skb, tcm, cb, t_p, s_t) < 0)
			return -1;
	}

	return 0;
}

L
Linus Torvalds 已提交
1557 1558
static int tc_dump_tclass(struct sk_buff *skb, struct netlink_callback *cb)
{
1559
	struct tcmsg *tcm = (struct tcmsg*)NLMSG_DATA(cb->nlh);
1560
	struct net *net = sock_net(skb->sk);
1561
	struct netdev_queue *dev_queue;
L
Linus Torvalds 已提交
1562
	struct net_device *dev;
1563
	int t, s_t;
L
Linus Torvalds 已提交
1564

1565 1566 1567
	if (net != &init_net)
		return 0;

L
Linus Torvalds 已提交
1568 1569
	if (cb->nlh->nlmsg_len < NLMSG_LENGTH(sizeof(*tcm)))
		return 0;
1570
	if ((dev = dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
L
Linus Torvalds 已提交
1571 1572 1573 1574 1575
		return 0;

	s_t = cb->args[0];
	t = 0;

1576
	dev_queue = netdev_get_tx_queue(dev, 0);
1577
	if (tc_dump_tclass_root(dev_queue->qdisc_sleeping, skb, tcm, cb, &t, s_t) < 0)
1578 1579 1580
		goto done;

	dev_queue = &dev->rx_queue;
1581
	if (tc_dump_tclass_root(dev_queue->qdisc_sleeping, skb, tcm, cb, &t, s_t) < 0)
1582
		goto done;
L
Linus Torvalds 已提交
1583

1584
done:
L
Linus Torvalds 已提交
1585 1586 1587 1588 1589 1590 1591 1592 1593 1594
	cb->args[0] = t;

	dev_put(dev);
	return skb->len;
}

/* Main classifier routine: scans classifier chain attached
   to this qdisc, (optionally) tests for protocol and asks
   specific classifiers.
 */
1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615
int tc_classify_compat(struct sk_buff *skb, struct tcf_proto *tp,
		       struct tcf_result *res)
{
	__be16 protocol = skb->protocol;
	int err = 0;

	for (; tp; tp = tp->next) {
		if ((tp->protocol == protocol ||
		     tp->protocol == htons(ETH_P_ALL)) &&
		    (err = tp->classify(skb, tp, res)) >= 0) {
#ifdef CONFIG_NET_CLS_ACT
			if (err != TC_ACT_RECLASSIFY && skb->tc_verd)
				skb->tc_verd = SET_TC_VERD(skb->tc_verd, 0);
#endif
			return err;
		}
	}
	return -1;
}
EXPORT_SYMBOL(tc_classify_compat);

L
Linus Torvalds 已提交
1616
int tc_classify(struct sk_buff *skb, struct tcf_proto *tp,
1617
		struct tcf_result *res)
L
Linus Torvalds 已提交
1618 1619
{
	int err = 0;
1620
	__be16 protocol;
L
Linus Torvalds 已提交
1621 1622 1623 1624 1625 1626
#ifdef CONFIG_NET_CLS_ACT
	struct tcf_proto *otp = tp;
reclassify:
#endif
	protocol = skb->protocol;

1627
	err = tc_classify_compat(skb, tp, res);
L
Linus Torvalds 已提交
1628
#ifdef CONFIG_NET_CLS_ACT
1629 1630 1631 1632 1633 1634 1635 1636 1637
	if (err == TC_ACT_RECLASSIFY) {
		u32 verd = G_TC_VERD(skb->tc_verd);
		tp = otp;

		if (verd++ >= MAX_REC_LOOP) {
			printk("rule prio %u protocol %02x reclassify loop, "
			       "packet dropped\n",
			       tp->prio&0xffff, ntohs(tp->protocol));
			return TC_ACT_SHOT;
L
Linus Torvalds 已提交
1638
		}
1639 1640
		skb->tc_verd = SET_TC_VERD(skb->tc_verd, verd);
		goto reclassify;
L
Linus Torvalds 已提交
1641
	}
1642 1643
#endif
	return err;
L
Linus Torvalds 已提交
1644
}
1645
EXPORT_SYMBOL(tc_classify);
L
Linus Torvalds 已提交
1646

1647 1648 1649 1650 1651 1652 1653
void tcf_destroy(struct tcf_proto *tp)
{
	tp->ops->destroy(tp);
	module_put(tp->ops->owner);
	kfree(tp);
}

1654
void tcf_destroy_chain(struct tcf_proto **fl)
1655 1656 1657
{
	struct tcf_proto *tp;

1658 1659
	while ((tp = *fl) != NULL) {
		*fl = tp->next;
1660 1661 1662 1663 1664
		tcf_destroy(tp);
	}
}
EXPORT_SYMBOL(tcf_destroy_chain);

L
Linus Torvalds 已提交
1665 1666 1667
#ifdef CONFIG_PROC_FS
static int psched_show(struct seq_file *seq, void *v)
{
1668 1669 1670
	struct timespec ts;

	hrtimer_get_res(CLOCK_MONOTONIC, &ts);
L
Linus Torvalds 已提交
1671
	seq_printf(seq, "%08x %08x %08x %08x\n",
1672
		   (u32)NSEC_PER_USEC, (u32)PSCHED_US2NS(1),
P
Patrick McHardy 已提交
1673
		   1000000,
1674
		   (u32)NSEC_PER_SEC/(u32)ktime_to_ns(timespec_to_ktime(ts)));
L
Linus Torvalds 已提交
1675 1676 1677 1678 1679 1680 1681 1682 1683

	return 0;
}

static int psched_open(struct inode *inode, struct file *file)
{
	return single_open(file, psched_show, PDE(inode)->data);
}

1684
static const struct file_operations psched_fops = {
L
Linus Torvalds 已提交
1685 1686 1687 1688 1689
	.owner = THIS_MODULE,
	.open = psched_open,
	.read  = seq_read,
	.llseek = seq_lseek,
	.release = single_release,
1690
};
L
Linus Torvalds 已提交
1691 1692 1693 1694 1695 1696
#endif

static int __init pktsched_init(void)
{
	register_qdisc(&pfifo_qdisc_ops);
	register_qdisc(&bfifo_qdisc_ops);
1697
	proc_net_fops_create(&init_net, "psched", 0, &psched_fops);
L
Linus Torvalds 已提交
1698

1699 1700 1701 1702 1703 1704 1705
	rtnl_register(PF_UNSPEC, RTM_NEWQDISC, tc_modify_qdisc, NULL);
	rtnl_register(PF_UNSPEC, RTM_DELQDISC, tc_get_qdisc, NULL);
	rtnl_register(PF_UNSPEC, RTM_GETQDISC, tc_get_qdisc, tc_dump_qdisc);
	rtnl_register(PF_UNSPEC, RTM_NEWTCLASS, tc_ctl_tclass, NULL);
	rtnl_register(PF_UNSPEC, RTM_DELTCLASS, tc_ctl_tclass, NULL);
	rtnl_register(PF_UNSPEC, RTM_GETTCLASS, tc_ctl_tclass, tc_dump_tclass);

L
Linus Torvalds 已提交
1706 1707 1708 1709
	return 0;
}

subsys_initcall(pktsched_init);