dn_sched_fq_codel.c revision 301772
1/* 2 * FQ_Codel - The FlowQueue-Codel scheduler/AQM 3 * 4 * $FreeBSD: stable/10/sys/netpfil/ipfw/dn_sched_fq_codel.c 301772 2016-06-10 00:00:25Z truckman $ 5 * 6 * Copyright (C) 2016 Centre for Advanced Internet Architectures, 7 * Swinburne University of Technology, Melbourne, Australia. 8 * Portions of this code were made possible in part by a gift from 9 * The Comcast Innovation Fund. 10 * Implemented by Rasool Al-Saadi <ralsaadi@swin.edu.au> 11 * 12 * Redistribution and use in source and binary forms, with or without 13 * modification, are permitted provided that the following conditions 14 * are met: 15 * 1. Redistributions of source code must retain the above copyright 16 * notice, this list of conditions and the following disclaimer. 17 * 2. Redistributions in binary form must reproduce the above copyright 18 * notice, this list of conditions and the following disclaimer in the 19 * documentation and/or other materials provided with the distribution. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 */ 33 34#ifdef _KERNEL 35#include <sys/malloc.h> 36#include <sys/socket.h> 37//#include <sys/socketvar.h> 38#include <sys/kernel.h> 39#include <sys/mbuf.h> 40#include <sys/module.h> 41#include <net/if.h> /* IFNAMSIZ */ 42#include <netinet/in.h> 43#include <netinet/ip_var.h> /* ipfw_rule_ref */ 44#include <netinet/ip_fw.h> /* flow_id */ 45#include <netinet/ip_dummynet.h> 46 47#include <sys/proc.h> 48#include <sys/rwlock.h> 49 50#include <netpfil/ipfw/ip_fw_private.h> 51#include <sys/sysctl.h> 52#include <netinet/ip.h> 53#include <netinet/ip6.h> 54#include <netinet/ip_icmp.h> 55#include <netinet/tcp.h> 56#include <netinet/udp.h> 57#include <sys/queue.h> 58#include <sys/hash.h> 59 60#include <netpfil/ipfw/dn_heap.h> 61#include <netpfil/ipfw/ip_dn_private.h> 62 63#include <netpfil/ipfw/dn_aqm.h> 64#include <netpfil/ipfw/dn_aqm_codel.h> 65#include <netpfil/ipfw/dn_sched.h> 66#include <netpfil/ipfw/dn_sched_fq_codel.h> 67#include <netpfil/ipfw/dn_sched_fq_codel_helper.h> 68 69#else 70#include <dn_test.h> 71#endif 72 73/* NOTE: In fq_codel module, we reimplements CoDel AQM functions 74 * because fq_codel use different flows (sub-queues) structure and 75 * dn_queue includes many variables not needed by a flow (sub-queue 76 * )i.e. avoid extra overhead (88 bytes vs 208 bytes). 77 * Also, CoDel functions manages stats of sub-queues as well as the main queue. 78 */ 79 80#define DN_SCHED_FQ_CODEL 6 81 82static struct dn_alg fq_codel_desc; 83 84/* fq_codel default parameters including codel */ 85struct dn_sch_fq_codel_parms 86fq_codel_sysctl = {{5000 * AQM_TIME_1US, 100000 * AQM_TIME_1US, 87 CODEL_ECN_ENABLED}, 1024, 10240, 1514}; 88 89static int 90fqcodel_sysctl_interval_handler(SYSCTL_HANDLER_ARGS) 91{ 92 int error; 93 long value; 94 95 value = fq_codel_sysctl.ccfg.interval; 96 value /= AQM_TIME_1US; 97 error = sysctl_handle_long(oidp, &value, 0, req); 98 if (error != 0 || req->newptr == NULL) 99 return (error); 100 if (value < 1 || value > 100 * AQM_TIME_1S) 101 return (EINVAL); 102 fq_codel_sysctl.ccfg.interval = value * AQM_TIME_1US ; 103 104 return (0); 105} 106 107static int 108fqcodel_sysctl_target_handler(SYSCTL_HANDLER_ARGS) 109{ 110 int error; 111 long value; 112 113 value = fq_codel_sysctl.ccfg.target; 114 value /= AQM_TIME_1US; 115 error = sysctl_handle_long(oidp, &value, 0, req); 116 if (error != 0 || req->newptr == NULL) 117 return (error); 118 if (value < 1 || value > 5 * AQM_TIME_1S) 119 return (EINVAL); 120 fq_codel_sysctl.ccfg.target = value * AQM_TIME_1US ; 121 122 return (0); 123} 124 125 126SYSBEGIN(f4) 127 128SYSCTL_DECL(_net_inet); 129SYSCTL_DECL(_net_inet_ip); 130SYSCTL_DECL(_net_inet_ip_dummynet); 131static SYSCTL_NODE(_net_inet_ip_dummynet, OID_AUTO, fqcodel, 132 CTLFLAG_RW, 0, "FQ_CODEL"); 133 134#ifdef SYSCTL_NODE 135 136SYSCTL_PROC(_net_inet_ip_dummynet_fqcodel, OID_AUTO, target, 137 CTLTYPE_LONG | CTLFLAG_RW, NULL, 0, fqcodel_sysctl_target_handler, "L", 138 "FQ_CoDel target in microsecond"); 139SYSCTL_PROC(_net_inet_ip_dummynet_fqcodel, OID_AUTO, interval, 140 CTLTYPE_LONG | CTLFLAG_RW, NULL, 0, fqcodel_sysctl_interval_handler, "L", 141 "FQ_CoDel interval in microsecond"); 142 143SYSCTL_UINT(_net_inet_ip_dummynet_fqcodel, OID_AUTO, quantum, 144 CTLFLAG_RW, &fq_codel_sysctl.quantum, 1514, "FQ_CoDel quantum"); 145SYSCTL_UINT(_net_inet_ip_dummynet_fqcodel, OID_AUTO, flows, 146 CTLFLAG_RW, &fq_codel_sysctl.flows_cnt, 1024, 147 "Number of queues for FQ_CoDel"); 148SYSCTL_UINT(_net_inet_ip_dummynet_fqcodel, OID_AUTO, limit, 149 CTLFLAG_RW, &fq_codel_sysctl.limit, 10240, "FQ_CoDel queues size limit"); 150#endif 151 152/* Drop a packet form the head of codel queue */ 153static void 154codel_drop_head(struct fq_codel_flow *q, struct fq_codel_si *si) 155{ 156 struct mbuf *m = q->mq.head; 157 158 if (m == NULL) 159 return; 160 q->mq.head = m->m_nextpkt; 161 162 fq_update_stats(q, si, -m->m_pkthdr.len, 1); 163 164 if (si->main_q.ni.length == 0) /* queue is now idle */ 165 si->main_q.q_time = dn_cfg.curr_time; 166 167 FREE_PKT(m); 168} 169 170/* Enqueue a packet 'm' to a queue 'q' and add timestamp to that packet. 171 * Return 1 when unable to add timestamp, otherwise return 0 172 */ 173static int 174codel_enqueue(struct fq_codel_flow *q, struct mbuf *m, struct fq_codel_si *si) 175{ 176 uint64_t len; 177 178 len = m->m_pkthdr.len; 179 /* finding maximum packet size */ 180 if (len > q->cst.maxpkt_size) 181 q->cst.maxpkt_size = len; 182 183 /* Add timestamp to mbuf as MTAG */ 184 struct m_tag *mtag; 185 mtag = m_tag_locate(m, MTAG_ABI_COMPAT, DN_AQM_MTAG_TS, NULL); 186 if (mtag == NULL) 187 mtag = m_tag_alloc(MTAG_ABI_COMPAT, DN_AQM_MTAG_TS, sizeof(aqm_time_t), 188 M_NOWAIT); 189 if (mtag == NULL) { 190 m_freem(m); 191 goto drop; 192 } 193 *(aqm_time_t *)(mtag + 1) = AQM_UNOW; 194 m_tag_prepend(m, mtag); 195 196 mq_append(&q->mq, m); 197 fq_update_stats(q, si, len, 0); 198 return 0; 199 200drop: 201 fq_update_stats(q, si, len, 1); 202 m_freem(m); 203 return 1; 204} 205 206/* 207 * Classify a packet to queue number using Jenkins hash function. 208 * Return: queue number 209 * the input of the hash are protocol no, perturbation, src IP, dst IP, 210 * src port, dst port, 211 */ 212static inline int 213fq_codel_classify_flow(struct mbuf *m, uint16_t fcount, struct fq_codel_si *si) 214{ 215 struct ip *ip; 216 struct tcphdr *th; 217 struct udphdr *uh; 218 uint8_t tuple[41]; 219 uint16_t hash=0; 220 221//#ifdef INET6 222 struct ip6_hdr *ip6; 223 int isip6; 224 isip6 = (mtod(m, struct ip *)->ip_v == 6) ? 1 : 0; 225 226 if(isip6) { 227 ip6 = mtod(m, struct ip6_hdr *); 228 *((uint8_t *) &tuple[0]) = ip6->ip6_nxt; 229 *((uint32_t *) &tuple[1]) = si->perturbation; 230 memcpy(&tuple[5], ip6->ip6_src.s6_addr, 16); 231 memcpy(&tuple[21], ip6->ip6_dst.s6_addr, 16); 232 233 switch (ip6->ip6_nxt) { 234 case IPPROTO_TCP: 235 th = (struct tcphdr *)(ip6 + 1); 236 *((uint16_t *) &tuple[37]) = th->th_dport; 237 *((uint16_t *) &tuple[39]) = th->th_sport; 238 break; 239 240 case IPPROTO_UDP: 241 uh = (struct udphdr *)(ip6 + 1); 242 *((uint16_t *) &tuple[37]) = uh->uh_dport; 243 *((uint16_t *) &tuple[39]) = uh->uh_sport; 244 break; 245 default: 246 memset(&tuple[37], 0, 4); 247 248 } 249 250 hash = jenkins_hash(tuple, 41, HASHINIT) % fcount; 251 return hash; 252 } 253//#endif 254 255 /* IPv4 */ 256 ip = mtod(m, struct ip *); 257 *((uint8_t *) &tuple[0]) = ip->ip_p; 258 *((uint32_t *) &tuple[1]) = si->perturbation; 259 *((uint32_t *) &tuple[5]) = ip->ip_src.s_addr; 260 *((uint32_t *) &tuple[9]) = ip->ip_dst.s_addr; 261 262 switch (ip->ip_p) { 263 case IPPROTO_TCP: 264 th = (struct tcphdr *)(ip + 1); 265 *((uint16_t *) &tuple[13]) = th->th_dport; 266 *((uint16_t *) &tuple[15]) = th->th_sport; 267 break; 268 269 case IPPROTO_UDP: 270 uh = (struct udphdr *)(ip + 1); 271 *((uint16_t *) &tuple[13]) = uh->uh_dport; 272 *((uint16_t *) &tuple[15]) = uh->uh_sport; 273 break; 274 default: 275 memset(&tuple[13], 0, 4); 276 277 } 278 hash = jenkins_hash(tuple, 17, HASHINIT) % fcount; 279 280 return hash; 281} 282 283/* 284 * Enqueue a packet into an appropriate queue according to 285 * FQ_CODEL algorithm. 286 */ 287static int 288fq_codel_enqueue(struct dn_sch_inst *_si, struct dn_queue *_q, 289 struct mbuf *m) 290{ 291 struct fq_codel_si *si; 292 struct fq_codel_schk *schk; 293 struct dn_sch_fq_codel_parms *param; 294 struct dn_queue *mainq; 295 int idx, drop, i, maxidx; 296 297 mainq = (struct dn_queue *)(_si + 1); 298 si = (struct fq_codel_si *)_si; 299 schk = (struct fq_codel_schk *)(si->_si.sched+1); 300 param = &schk->cfg; 301 302 /* classify a packet to queue number*/ 303 idx = fq_codel_classify_flow(m, param->flows_cnt, si); 304 /* enqueue packet into appropriate queue using CoDel AQM. 305 * Note: 'codel_enqueue' function returns 1 only when it unable to 306 * add timestamp to packet (no limit check)*/ 307 drop = codel_enqueue(&si->flows[idx], m, si); 308 309 /* codel unable to timestamp a packet */ 310 if (drop) 311 return 1; 312 313 /* If the flow (sub-queue) is not active ,then add it to the tail of 314 * new flows list, initialize and activate it. 315 */ 316 if (!si->flows[idx].active ) { 317 STAILQ_INSERT_TAIL(&si->newflows, &si->flows[idx], flowchain); 318 si->flows[idx].deficit = param->quantum; 319 si->flows[idx].cst.dropping = false; 320 si->flows[idx].cst.first_above_time = 0; 321 si->flows[idx].active = 1; 322 //D("activate %d",idx); 323 } 324 325 /* check the limit for all queues and remove a packet from the 326 * largest one 327 */ 328 if (mainq->ni.length > schk->cfg.limit) { D("over limit"); 329 /* find first active flow */ 330 for (maxidx = 0; maxidx < schk->cfg.flows_cnt; maxidx++) 331 if (si->flows[maxidx].active) 332 break; 333 if (maxidx < schk->cfg.flows_cnt) { 334 /* find the largest sub- queue */ 335 for (i = maxidx + 1; i < schk->cfg.flows_cnt; i++) 336 if (si->flows[i].active && si->flows[i].stats.length > 337 si->flows[maxidx].stats.length) 338 maxidx = i; 339 codel_drop_head(&si->flows[maxidx], si); 340 D("maxidx = %d",maxidx); 341 drop = 1; 342 } 343 } 344 345 return drop; 346} 347 348/* 349 * Dequeue a packet from an appropriate queue according to 350 * FQ_CODEL algorithm. 351 */ 352static struct mbuf * 353fq_codel_dequeue(struct dn_sch_inst *_si) 354{ 355 struct fq_codel_si *si; 356 struct fq_codel_schk *schk; 357 struct dn_sch_fq_codel_parms *param; 358 struct fq_codel_flow *f; 359 struct mbuf *mbuf; 360 struct fq_codel_list *fq_codel_flowlist; 361 362 si = (struct fq_codel_si *)_si; 363 schk = (struct fq_codel_schk *)(si->_si.sched+1); 364 param = &schk->cfg; 365 366 do { 367 /* select a list to start with */ 368 if (STAILQ_EMPTY(&si->newflows)) 369 fq_codel_flowlist = &si->oldflows; 370 else 371 fq_codel_flowlist = &si->newflows; 372 373 /* Both new and old queue lists are empty, return NULL */ 374 if (STAILQ_EMPTY(fq_codel_flowlist)) 375 return NULL; 376 377 f = STAILQ_FIRST(fq_codel_flowlist); 378 while (f != NULL) { 379 /* if there is no flow(sub-queue) deficit, increase deficit 380 * by quantum, move the flow to the tail of old flows list 381 * and try another flow. 382 * Otherwise, the flow will be used for dequeue. 383 */ 384 if (f->deficit < 0) { 385 f->deficit += param->quantum; 386 STAILQ_REMOVE_HEAD(fq_codel_flowlist, flowchain); 387 STAILQ_INSERT_TAIL(&si->oldflows, f, flowchain); 388 } else 389 break; 390 391 f = STAILQ_FIRST(fq_codel_flowlist); 392 } 393 394 /* the new flows list is empty, try old flows list */ 395 if (STAILQ_EMPTY(fq_codel_flowlist)) 396 continue; 397 398 /* Dequeue a packet from the selected flow */ 399 mbuf = fqc_codel_dequeue(f, si); 400 401 /* Codel did not return a packet */ 402 if (!mbuf) { 403 /* If the selected flow belongs to new flows list, then move 404 * it to the tail of old flows list. Otherwise, deactivate it and 405 * remove it from the old list and 406 */ 407 if (fq_codel_flowlist == &si->newflows) { 408 STAILQ_REMOVE_HEAD(fq_codel_flowlist, flowchain); 409 STAILQ_INSERT_TAIL(&si->oldflows, f, flowchain); 410 } else { 411 f->active = 0; 412 STAILQ_REMOVE_HEAD(fq_codel_flowlist, flowchain); 413 } 414 /* start again */ 415 continue; 416 } 417 418 /* we have a packet to return, 419 * update flow deficit and return the packet*/ 420 f->deficit -= mbuf->m_pkthdr.len; 421 return mbuf; 422 423 } while (1); 424 425 /* unreachable point */ 426 return NULL; 427} 428 429/* 430 * Initialize fq_codel scheduler instance. 431 * also, allocate memory for flows array. 432 */ 433static int 434fq_codel_new_sched(struct dn_sch_inst *_si) 435{ 436 struct fq_codel_si *si; 437 struct dn_queue *q; 438 struct fq_codel_schk *schk; 439 int i; 440 441 si = (struct fq_codel_si *)_si; 442 schk = (struct fq_codel_schk *)(_si->sched+1); 443 444 if(si->flows) { 445 D("si already configured!"); 446 return 0; 447 } 448 449 /* init the main queue */ 450 q = &si->main_q; 451 set_oid(&q->ni.oid, DN_QUEUE, sizeof(*q)); 452 q->_si = _si; 453 q->fs = _si->sched->fs; 454 455 /* allocate memory for flows array */ 456 si->flows = malloc(schk->cfg.flows_cnt * sizeof(struct fq_codel_flow), 457 M_DUMMYNET, M_NOWAIT | M_ZERO); 458 if (si->flows == NULL) { 459 D("cannot allocate memory for fq_codel configuration parameters"); 460 return ENOMEM ; 461 } 462 463 /* init perturbation for this si */ 464 si->perturbation = random(); 465 466 /* init the old and new flows lists */ 467 STAILQ_INIT(&si->newflows); 468 STAILQ_INIT(&si->oldflows); 469 470 /* init the flows (sub-queues) */ 471 for (i = 0; i < schk->cfg.flows_cnt; i++) { 472 /* init codel */ 473 si->flows[i].cst.maxpkt_size = 500; 474 } 475 476 fq_codel_desc.ref_count++; 477 return 0; 478} 479 480/* 481 * Free fq_codel scheduler instance. 482 */ 483static int 484fq_codel_free_sched(struct dn_sch_inst *_si) 485{ 486 struct fq_codel_si *si = (struct fq_codel_si *)_si ; 487 488 /* free the flows array */ 489 free(si->flows , M_DUMMYNET); 490 si->flows = NULL; 491 fq_codel_desc.ref_count--; 492 493 return 0; 494} 495 496/* 497 * Configure fq_codel scheduler. 498 * the configurations for the scheduler is passed from userland. 499 */ 500static int 501fq_codel_config(struct dn_schk *_schk) 502{ 503 struct fq_codel_schk *schk; 504 struct dn_extra_parms *ep; 505 struct dn_sch_fq_codel_parms *fqc_cfg; 506 507 schk = (struct fq_codel_schk *)(_schk+1); 508 ep = (struct dn_extra_parms *) _schk->cfg; 509 510 /* par array contains fq_codel configuration as follow 511 * Codel: 0- target,1- interval, 2- flags 512 * FQ_CODEL: 3- quantum, 4- limit, 5- flows 513 */ 514 if (ep && ep->oid.len ==sizeof(*ep) && 515 ep->oid.subtype == DN_SCH_PARAMS) { 516 517 fqc_cfg = &schk->cfg; 518 if (ep->par[0] < 0) 519 fqc_cfg->ccfg.target = fq_codel_sysctl.ccfg.target; 520 else 521 fqc_cfg->ccfg.target = ep->par[0] * AQM_TIME_1US; 522 523 if (ep->par[1] < 0) 524 fqc_cfg->ccfg.interval = fq_codel_sysctl.ccfg.interval; 525 else 526 fqc_cfg->ccfg.interval = ep->par[1] * AQM_TIME_1US; 527 528 if (ep->par[2] < 0) 529 fqc_cfg->ccfg.flags = 0; 530 else 531 fqc_cfg->ccfg.flags = ep->par[2]; 532 533 /* FQ configurations */ 534 if (ep->par[3] < 0) 535 fqc_cfg->quantum = fq_codel_sysctl.quantum; 536 else 537 fqc_cfg->quantum = ep->par[3]; 538 539 if (ep->par[4] < 0) 540 fqc_cfg->limit = fq_codel_sysctl.limit; 541 else 542 fqc_cfg->limit = ep->par[4]; 543 544 if (ep->par[5] < 0) 545 fqc_cfg->flows_cnt = fq_codel_sysctl.flows_cnt; 546 else 547 fqc_cfg->flows_cnt = ep->par[5]; 548 549 /* Bound the configurations */ 550 fqc_cfg->ccfg.target = BOUND_VAR(fqc_cfg->ccfg.target, 1 , 551 5 * AQM_TIME_1S); ; 552 fqc_cfg->ccfg.interval = BOUND_VAR(fqc_cfg->ccfg.interval, 1, 553 100 * AQM_TIME_1S); 554 555 fqc_cfg->quantum = BOUND_VAR(fqc_cfg->quantum,1, 9000); 556 fqc_cfg->limit= BOUND_VAR(fqc_cfg->limit,1,20480); 557 fqc_cfg->flows_cnt= BOUND_VAR(fqc_cfg->flows_cnt,1,65536); 558 } 559 else 560 return 1; 561 562 return 0; 563} 564 565/* 566 * Return fq_codel scheduler configurations 567 * the configurations for the scheduler is passed to userland. 568 */ 569static int 570fq_codel_getconfig (struct dn_schk *_schk, struct dn_extra_parms *ep) { 571 572 struct fq_codel_schk *schk = (struct fq_codel_schk *)(_schk+1); 573 struct dn_sch_fq_codel_parms *fqc_cfg; 574 575 fqc_cfg = &schk->cfg; 576 577 strcpy(ep->name, fq_codel_desc.name); 578 ep->par[0] = fqc_cfg->ccfg.target / AQM_TIME_1US; 579 ep->par[1] = fqc_cfg->ccfg.interval / AQM_TIME_1US; 580 ep->par[2] = fqc_cfg->ccfg.flags; 581 582 ep->par[3] = fqc_cfg->quantum; 583 ep->par[4] = fqc_cfg->limit; 584 ep->par[5] = fqc_cfg->flows_cnt; 585 586 return 0; 587} 588 589/* 590 * fq_codel scheduler descriptor 591 * contains the type of the scheduler, the name, the size of extra 592 * data structures, and function pointers. 593 */ 594static struct dn_alg fq_codel_desc = { 595 _SI( .type = ) DN_SCHED_FQ_CODEL, 596 _SI( .name = ) "FQ_CODEL", 597 _SI( .flags = ) 0, 598 599 _SI( .schk_datalen = ) sizeof(struct fq_codel_schk), 600 _SI( .si_datalen = ) sizeof(struct fq_codel_si) - sizeof(struct dn_sch_inst), 601 _SI( .q_datalen = ) 0, 602 603 _SI( .enqueue = ) fq_codel_enqueue, 604 _SI( .dequeue = ) fq_codel_dequeue, 605 _SI( .config = ) fq_codel_config, /* new sched i.e. sched X config ...*/ 606 _SI( .destroy = ) NULL, /*sched x delete */ 607 _SI( .new_sched = ) fq_codel_new_sched, /* new schd instance */ 608 _SI( .free_sched = ) fq_codel_free_sched, /* delete schd instance */ 609 _SI( .new_fsk = ) NULL, 610 _SI( .free_fsk = ) NULL, 611 _SI( .new_queue = ) NULL, 612 _SI( .free_queue = ) NULL, 613 _SI( .getconfig = ) fq_codel_getconfig, 614 _SI( .ref_count = ) 0 615}; 616 617DECLARE_DNSCHED_MODULE(dn_fq_codel, &fq_codel_desc); 618