sch_fq_codel.c 15.4 KB
Newer Older
E
Eric Dumazet 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 29 30 31 32 33 34 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197
/*
 * Fair Queue CoDel discipline
 *
 *	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.
 *
 *  Copyright (C) 2012 Eric Dumazet <edumazet@google.com>
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/jiffies.h>
#include <linux/string.h>
#include <linux/in.h>
#include <linux/errno.h>
#include <linux/init.h>
#include <linux/skbuff.h>
#include <linux/jhash.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>
#include <net/netlink.h>
#include <net/pkt_sched.h>
#include <net/flow_keys.h>
#include <net/codel.h>

/*	Fair Queue CoDel.
 *
 * Principles :
 * Packets are classified (internal classifier or external) on flows.
 * This is a Stochastic model (as we use a hash, several flows
 *			       might be hashed on same slot)
 * Each flow has a CoDel managed queue.
 * Flows are linked onto two (Round Robin) lists,
 * so that new flows have priority on old ones.
 *
 * For a given flow, packets are not reordered (CoDel uses a FIFO)
 * head drops only.
 * ECN capability is on by default.
 * Low memory footprint (64 bytes per flow)
 */

struct fq_codel_flow {
	struct sk_buff	  *head;
	struct sk_buff	  *tail;
	struct list_head  flowchain;
	int		  deficit;
	u32		  dropped; /* number of drops (or ECN marks) on this flow */
	struct codel_vars cvars;
}; /* please try to keep this structure <= 64 bytes */

struct fq_codel_sched_data {
	struct tcf_proto *filter_list;	/* optional external classifier */
	struct fq_codel_flow *flows;	/* Flows table [flows_cnt] */
	u32		*backlogs;	/* backlog table [flows_cnt] */
	u32		flows_cnt;	/* number of flows */
	u32		perturbation;	/* hash perturbation */
	u32		quantum;	/* psched_mtu(qdisc_dev(sch)); */
	struct codel_params cparams;
	struct codel_stats cstats;
	u32		drop_overlimit;
	u32		new_flow_count;

	struct list_head new_flows;	/* list of new flows */
	struct list_head old_flows;	/* list of old flows */
};

static unsigned int fq_codel_hash(const struct fq_codel_sched_data *q,
				  const struct sk_buff *skb)
{
	struct flow_keys keys;
	unsigned int hash;

	skb_flow_dissect(skb, &keys);
	hash = jhash_3words((__force u32)keys.dst,
			    (__force u32)keys.src ^ keys.ip_proto,
			    (__force u32)keys.ports, q->perturbation);
	return ((u64)hash * q->flows_cnt) >> 32;
}

static unsigned int fq_codel_classify(struct sk_buff *skb, struct Qdisc *sch,
				      int *qerr)
{
	struct fq_codel_sched_data *q = qdisc_priv(sch);
	struct tcf_result res;
	int result;

	if (TC_H_MAJ(skb->priority) == sch->handle &&
	    TC_H_MIN(skb->priority) > 0 &&
	    TC_H_MIN(skb->priority) <= q->flows_cnt)
		return TC_H_MIN(skb->priority);

	if (!q->filter_list)
		return fq_codel_hash(q, skb) + 1;

	*qerr = NET_XMIT_SUCCESS | __NET_XMIT_BYPASS;
	result = tc_classify(skb, q->filter_list, &res);
	if (result >= 0) {
#ifdef CONFIG_NET_CLS_ACT
		switch (result) {
		case TC_ACT_STOLEN:
		case TC_ACT_QUEUED:
			*qerr = NET_XMIT_SUCCESS | __NET_XMIT_STOLEN;
		case TC_ACT_SHOT:
			return 0;
		}
#endif
		if (TC_H_MIN(res.classid) <= q->flows_cnt)
			return TC_H_MIN(res.classid);
	}
	return 0;
}

/* helper functions : might be changed when/if skb use a standard list_head */

/* remove one skb from head of slot queue */
static inline struct sk_buff *dequeue_head(struct fq_codel_flow *flow)
{
	struct sk_buff *skb = flow->head;

	flow->head = skb->next;
	skb->next = NULL;
	return skb;
}

/* add skb to flow queue (tail add) */
static inline void flow_queue_add(struct fq_codel_flow *flow,
				  struct sk_buff *skb)
{
	if (flow->head == NULL)
		flow->head = skb;
	else
		flow->tail->next = skb;
	flow->tail = skb;
	skb->next = NULL;
}

static unsigned int fq_codel_drop(struct Qdisc *sch)
{
	struct fq_codel_sched_data *q = qdisc_priv(sch);
	struct sk_buff *skb;
	unsigned int maxbacklog = 0, idx = 0, i, len;
	struct fq_codel_flow *flow;

	/* Queue is full! Find the fat flow and drop packet from it.
	 * This might sound expensive, but with 1024 flows, we scan
	 * 4KB of memory, and we dont need to handle a complex tree
	 * in fast path (packet queue/enqueue) with many cache misses.
	 */
	for (i = 0; i < q->flows_cnt; i++) {
		if (q->backlogs[i] > maxbacklog) {
			maxbacklog = q->backlogs[i];
			idx = i;
		}
	}
	flow = &q->flows[idx];
	skb = dequeue_head(flow);
	len = qdisc_pkt_len(skb);
	q->backlogs[idx] -= len;
	kfree_skb(skb);
	sch->q.qlen--;
	sch->qstats.drops++;
	sch->qstats.backlog -= len;
	flow->dropped++;
	return idx;
}

static int fq_codel_enqueue(struct sk_buff *skb, struct Qdisc *sch)
{
	struct fq_codel_sched_data *q = qdisc_priv(sch);
	unsigned int idx;
	struct fq_codel_flow *flow;
	int uninitialized_var(ret);

	idx = fq_codel_classify(skb, sch, &ret);
	if (idx == 0) {
		if (ret & __NET_XMIT_BYPASS)
			sch->qstats.drops++;
		kfree_skb(skb);
		return ret;
	}
	idx--;

	codel_set_enqueue_time(skb);
	flow = &q->flows[idx];
	flow_queue_add(flow, skb);
	q->backlogs[idx] += qdisc_pkt_len(skb);
	sch->qstats.backlog += qdisc_pkt_len(skb);

	if (list_empty(&flow->flowchain)) {
		list_add_tail(&flow->flowchain, &q->new_flows);
		q->new_flow_count++;
		flow->deficit = q->quantum;
		flow->dropped = 0;
	}
198
	if (++sch->q.qlen <= sch->limit)
E
Eric Dumazet 已提交
199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218
		return NET_XMIT_SUCCESS;

	q->drop_overlimit++;
	/* Return Congestion Notification only if we dropped a packet
	 * from this flow.
	 */
	if (fq_codel_drop(sch) == idx)
		return NET_XMIT_CN;

	/* As we dropped a packet, better let upper stack know this */
	qdisc_tree_decrease_qlen(sch, 1);
	return NET_XMIT_SUCCESS;
}

/* This is the specific function called from codel_dequeue()
 * to dequeue a packet from queue. Note: backlog is handled in
 * codel, we dont need to reduce it here.
 */
static struct sk_buff *dequeue(struct codel_vars *vars, struct Qdisc *sch)
{
219
	struct fq_codel_sched_data *q = qdisc_priv(sch);
E
Eric Dumazet 已提交
220 221 222 223 224 225
	struct fq_codel_flow *flow;
	struct sk_buff *skb = NULL;

	flow = container_of(vars, struct fq_codel_flow, cvars);
	if (flow->head) {
		skb = dequeue_head(flow);
226
		q->backlogs[flow - q->flows] -= qdisc_pkt_len(skb);
E
Eric Dumazet 已提交
227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258
		sch->q.qlen--;
	}
	return skb;
}

static struct sk_buff *fq_codel_dequeue(struct Qdisc *sch)
{
	struct fq_codel_sched_data *q = qdisc_priv(sch);
	struct sk_buff *skb;
	struct fq_codel_flow *flow;
	struct list_head *head;
	u32 prev_drop_count, prev_ecn_mark;

begin:
	head = &q->new_flows;
	if (list_empty(head)) {
		head = &q->old_flows;
		if (list_empty(head))
			return NULL;
	}
	flow = list_first_entry(head, struct fq_codel_flow, flowchain);

	if (flow->deficit <= 0) {
		flow->deficit += q->quantum;
		list_move_tail(&flow->flowchain, &q->old_flows);
		goto begin;
	}

	prev_drop_count = q->cstats.drop_count;
	prev_ecn_mark = q->cstats.ecn_mark;

	skb = codel_dequeue(sch, &q->cparams, &flow->cvars, &q->cstats,
259
			    dequeue);
E
Eric Dumazet 已提交
260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367

	flow->dropped += q->cstats.drop_count - prev_drop_count;
	flow->dropped += q->cstats.ecn_mark - prev_ecn_mark;

	if (!skb) {
		/* force a pass through old_flows to prevent starvation */
		if ((head == &q->new_flows) && !list_empty(&q->old_flows))
			list_move_tail(&flow->flowchain, &q->old_flows);
		else
			list_del_init(&flow->flowchain);
		goto begin;
	}
	qdisc_bstats_update(sch, skb);
	flow->deficit -= qdisc_pkt_len(skb);
	/* We cant call qdisc_tree_decrease_qlen() if our qlen is 0,
	 * or HTB crashes. Defer it for next round.
	 */
	if (q->cstats.drop_count && sch->q.qlen) {
		qdisc_tree_decrease_qlen(sch, q->cstats.drop_count);
		q->cstats.drop_count = 0;
	}
	return skb;
}

static void fq_codel_reset(struct Qdisc *sch)
{
	struct sk_buff *skb;

	while ((skb = fq_codel_dequeue(sch)) != NULL)
		kfree_skb(skb);
}

static const struct nla_policy fq_codel_policy[TCA_FQ_CODEL_MAX + 1] = {
	[TCA_FQ_CODEL_TARGET]	= { .type = NLA_U32 },
	[TCA_FQ_CODEL_LIMIT]	= { .type = NLA_U32 },
	[TCA_FQ_CODEL_INTERVAL]	= { .type = NLA_U32 },
	[TCA_FQ_CODEL_ECN]	= { .type = NLA_U32 },
	[TCA_FQ_CODEL_FLOWS]	= { .type = NLA_U32 },
	[TCA_FQ_CODEL_QUANTUM]	= { .type = NLA_U32 },
};

static int fq_codel_change(struct Qdisc *sch, struct nlattr *opt)
{
	struct fq_codel_sched_data *q = qdisc_priv(sch);
	struct nlattr *tb[TCA_FQ_CODEL_MAX + 1];
	int err;

	if (!opt)
		return -EINVAL;

	err = nla_parse_nested(tb, TCA_FQ_CODEL_MAX, opt, fq_codel_policy);
	if (err < 0)
		return err;
	if (tb[TCA_FQ_CODEL_FLOWS]) {
		if (q->flows)
			return -EINVAL;
		q->flows_cnt = nla_get_u32(tb[TCA_FQ_CODEL_FLOWS]);
		if (!q->flows_cnt ||
		    q->flows_cnt > 65536)
			return -EINVAL;
	}
	sch_tree_lock(sch);

	if (tb[TCA_FQ_CODEL_TARGET]) {
		u64 target = nla_get_u32(tb[TCA_FQ_CODEL_TARGET]);

		q->cparams.target = (target * NSEC_PER_USEC) >> CODEL_SHIFT;
	}

	if (tb[TCA_FQ_CODEL_INTERVAL]) {
		u64 interval = nla_get_u32(tb[TCA_FQ_CODEL_INTERVAL]);

		q->cparams.interval = (interval * NSEC_PER_USEC) >> CODEL_SHIFT;
	}

	if (tb[TCA_FQ_CODEL_LIMIT])
		sch->limit = nla_get_u32(tb[TCA_FQ_CODEL_LIMIT]);

	if (tb[TCA_FQ_CODEL_ECN])
		q->cparams.ecn = !!nla_get_u32(tb[TCA_FQ_CODEL_ECN]);

	if (tb[TCA_FQ_CODEL_QUANTUM])
		q->quantum = max(256U, nla_get_u32(tb[TCA_FQ_CODEL_QUANTUM]));

	while (sch->q.qlen > sch->limit) {
		struct sk_buff *skb = fq_codel_dequeue(sch);

		kfree_skb(skb);
		q->cstats.drop_count++;
	}
	qdisc_tree_decrease_qlen(sch, q->cstats.drop_count);
	q->cstats.drop_count = 0;

	sch_tree_unlock(sch);
	return 0;
}

static void *fq_codel_zalloc(size_t sz)
{
	void *ptr = kzalloc(sz, GFP_KERNEL | __GFP_NOWARN);

	if (!ptr)
		ptr = vzalloc(sz);
	return ptr;
}

static void fq_codel_free(void *addr)
{
W
WANG Cong 已提交
368
	kvfree(addr);
E
Eric Dumazet 已提交
369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387
}

static void fq_codel_destroy(struct Qdisc *sch)
{
	struct fq_codel_sched_data *q = qdisc_priv(sch);

	tcf_destroy_chain(&q->filter_list);
	fq_codel_free(q->backlogs);
	fq_codel_free(q->flows);
}

static int fq_codel_init(struct Qdisc *sch, struct nlattr *opt)
{
	struct fq_codel_sched_data *q = qdisc_priv(sch);
	int i;

	sch->limit = 10*1024;
	q->flows_cnt = 1024;
	q->quantum = psched_mtu(qdisc_dev(sch));
388
	q->perturbation = prandom_u32();
E
Eric Dumazet 已提交
389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414
	INIT_LIST_HEAD(&q->new_flows);
	INIT_LIST_HEAD(&q->old_flows);
	codel_params_init(&q->cparams);
	codel_stats_init(&q->cstats);
	q->cparams.ecn = true;

	if (opt) {
		int err = fq_codel_change(sch, opt);
		if (err)
			return err;
	}

	if (!q->flows) {
		q->flows = fq_codel_zalloc(q->flows_cnt *
					   sizeof(struct fq_codel_flow));
		if (!q->flows)
			return -ENOMEM;
		q->backlogs = fq_codel_zalloc(q->flows_cnt * sizeof(u32));
		if (!q->backlogs) {
			fq_codel_free(q->flows);
			return -ENOMEM;
		}
		for (i = 0; i < q->flows_cnt; i++) {
			struct fq_codel_flow *flow = q->flows + i;

			INIT_LIST_HEAD(&flow->flowchain);
E
Eric Dumazet 已提交
415
			codel_vars_init(&flow->cvars);
E
Eric Dumazet 已提交
416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447
		}
	}
	if (sch->limit >= 1)
		sch->flags |= TCQ_F_CAN_BYPASS;
	else
		sch->flags &= ~TCQ_F_CAN_BYPASS;
	return 0;
}

static int fq_codel_dump(struct Qdisc *sch, struct sk_buff *skb)
{
	struct fq_codel_sched_data *q = qdisc_priv(sch);
	struct nlattr *opts;

	opts = nla_nest_start(skb, TCA_OPTIONS);
	if (opts == NULL)
		goto nla_put_failure;

	if (nla_put_u32(skb, TCA_FQ_CODEL_TARGET,
			codel_time_to_us(q->cparams.target)) ||
	    nla_put_u32(skb, TCA_FQ_CODEL_LIMIT,
			sch->limit) ||
	    nla_put_u32(skb, TCA_FQ_CODEL_INTERVAL,
			codel_time_to_us(q->cparams.interval)) ||
	    nla_put_u32(skb, TCA_FQ_CODEL_ECN,
			q->cparams.ecn) ||
	    nla_put_u32(skb, TCA_FQ_CODEL_QUANTUM,
			q->quantum) ||
	    nla_put_u32(skb, TCA_FQ_CODEL_FLOWS,
			q->flows_cnt))
		goto nla_put_failure;

448
	return nla_nest_end(skb, opts);
E
Eric Dumazet 已提交
449 450 451 452 453 454 455 456 457 458 459 460 461

nla_put_failure:
	return -1;
}

static int fq_codel_dump_stats(struct Qdisc *sch, struct gnet_dump *d)
{
	struct fq_codel_sched_data *q = qdisc_priv(sch);
	struct tc_fq_codel_xstats st = {
		.type				= TCA_FQ_CODEL_XSTATS_QDISC,
	};
	struct list_head *pos;

S
Sasha Levin 已提交
462 463 464 465 466
	st.qdisc_stats.maxpacket = q->cstats.maxpacket;
	st.qdisc_stats.drop_overlimit = q->drop_overlimit;
	st.qdisc_stats.ecn_mark = q->cstats.ecn_mark;
	st.qdisc_stats.new_flow_count = q->new_flow_count;

E
Eric Dumazet 已提交
467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620
	list_for_each(pos, &q->new_flows)
		st.qdisc_stats.new_flows_len++;

	list_for_each(pos, &q->old_flows)
		st.qdisc_stats.old_flows_len++;

	return gnet_stats_copy_app(d, &st, sizeof(st));
}

static struct Qdisc *fq_codel_leaf(struct Qdisc *sch, unsigned long arg)
{
	return NULL;
}

static unsigned long fq_codel_get(struct Qdisc *sch, u32 classid)
{
	return 0;
}

static unsigned long fq_codel_bind(struct Qdisc *sch, unsigned long parent,
			      u32 classid)
{
	/* we cannot bypass queue discipline anymore */
	sch->flags &= ~TCQ_F_CAN_BYPASS;
	return 0;
}

static void fq_codel_put(struct Qdisc *q, unsigned long cl)
{
}

static struct tcf_proto **fq_codel_find_tcf(struct Qdisc *sch, unsigned long cl)
{
	struct fq_codel_sched_data *q = qdisc_priv(sch);

	if (cl)
		return NULL;
	return &q->filter_list;
}

static int fq_codel_dump_class(struct Qdisc *sch, unsigned long cl,
			  struct sk_buff *skb, struct tcmsg *tcm)
{
	tcm->tcm_handle |= TC_H_MIN(cl);
	return 0;
}

static int fq_codel_dump_class_stats(struct Qdisc *sch, unsigned long cl,
				     struct gnet_dump *d)
{
	struct fq_codel_sched_data *q = qdisc_priv(sch);
	u32 idx = cl - 1;
	struct gnet_stats_queue qs = { 0 };
	struct tc_fq_codel_xstats xstats;

	if (idx < q->flows_cnt) {
		const struct fq_codel_flow *flow = &q->flows[idx];
		const struct sk_buff *skb = flow->head;

		memset(&xstats, 0, sizeof(xstats));
		xstats.type = TCA_FQ_CODEL_XSTATS_CLASS;
		xstats.class_stats.deficit = flow->deficit;
		xstats.class_stats.ldelay =
			codel_time_to_us(flow->cvars.ldelay);
		xstats.class_stats.count = flow->cvars.count;
		xstats.class_stats.lastcount = flow->cvars.lastcount;
		xstats.class_stats.dropping = flow->cvars.dropping;
		if (flow->cvars.dropping) {
			codel_tdiff_t delta = flow->cvars.drop_next -
					      codel_get_time();

			xstats.class_stats.drop_next = (delta >= 0) ?
				codel_time_to_us(delta) :
				-codel_time_to_us(-delta);
		}
		while (skb) {
			qs.qlen++;
			skb = skb->next;
		}
		qs.backlog = q->backlogs[idx];
		qs.drops = flow->dropped;
	}
	if (gnet_stats_copy_queue(d, &qs) < 0)
		return -1;
	if (idx < q->flows_cnt)
		return gnet_stats_copy_app(d, &xstats, sizeof(xstats));
	return 0;
}

static void fq_codel_walk(struct Qdisc *sch, struct qdisc_walker *arg)
{
	struct fq_codel_sched_data *q = qdisc_priv(sch);
	unsigned int i;

	if (arg->stop)
		return;

	for (i = 0; i < q->flows_cnt; i++) {
		if (list_empty(&q->flows[i].flowchain) ||
		    arg->count < arg->skip) {
			arg->count++;
			continue;
		}
		if (arg->fn(sch, i + 1, arg) < 0) {
			arg->stop = 1;
			break;
		}
		arg->count++;
	}
}

static const struct Qdisc_class_ops fq_codel_class_ops = {
	.leaf		=	fq_codel_leaf,
	.get		=	fq_codel_get,
	.put		=	fq_codel_put,
	.tcf_chain	=	fq_codel_find_tcf,
	.bind_tcf	=	fq_codel_bind,
	.unbind_tcf	=	fq_codel_put,
	.dump		=	fq_codel_dump_class,
	.dump_stats	=	fq_codel_dump_class_stats,
	.walk		=	fq_codel_walk,
};

static struct Qdisc_ops fq_codel_qdisc_ops __read_mostly = {
	.cl_ops		=	&fq_codel_class_ops,
	.id		=	"fq_codel",
	.priv_size	=	sizeof(struct fq_codel_sched_data),
	.enqueue	=	fq_codel_enqueue,
	.dequeue	=	fq_codel_dequeue,
	.peek		=	qdisc_peek_dequeued,
	.drop		=	fq_codel_drop,
	.init		=	fq_codel_init,
	.reset		=	fq_codel_reset,
	.destroy	=	fq_codel_destroy,
	.change		=	fq_codel_change,
	.dump		=	fq_codel_dump,
	.dump_stats =	fq_codel_dump_stats,
	.owner		=	THIS_MODULE,
};

static int __init fq_codel_module_init(void)
{
	return register_qdisc(&fq_codel_qdisc_ops);
}

static void __exit fq_codel_module_exit(void)
{
	unregister_qdisc(&fq_codel_qdisc_ops);
}

module_init(fq_codel_module_init)
module_exit(fq_codel_module_exit)
MODULE_AUTHOR("Eric Dumazet");
MODULE_LICENSE("GPL");