1 /* 2 * net/sched/sch_sfq.c Stochastic Fairness Queueing discipline. 3 * 4 * This program is free software; you can redistribute it and/or 5 * modify it under the terms of the GNU General Public License 6 * as published by the Free Software Foundation; either version 7 * 2 of the License, or (at your option) any later version. 8 * 9 * Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru> 10 */ 11 12 #include <linux/module.h> 13 #include <linux/types.h> 14 #include <linux/kernel.h> 15 #include <linux/jiffies.h> 16 #include <linux/string.h> 17 #include <linux/in.h> 18 #include <linux/errno.h> 19 #include <linux/init.h> 20 #include <linux/skbuff.h> 21 #include <linux/jhash.h> 22 #include <linux/slab.h> 23 #include <linux/vmalloc.h> 24 #include <net/netlink.h> 25 #include <net/pkt_sched.h> 26 #include <net/flow_keys.h> 27 #include <net/red.h> 28 29 30 /* Stochastic Fairness Queuing algorithm. 31 ======================================= 32 33 Source: 34 Paul E. McKenney "Stochastic Fairness Queuing", 35 IEEE INFOCOMM'90 Proceedings, San Francisco, 1990. 36 37 Paul E. McKenney "Stochastic Fairness Queuing", 38 "Interworking: Research and Experience", v.2, 1991, p.113-131. 39 40 41 See also: 42 M. Shreedhar and George Varghese "Efficient Fair 43 Queuing using Deficit Round Robin", Proc. SIGCOMM 95. 44 45 46 This is not the thing that is usually called (W)FQ nowadays. 47 It does not use any timestamp mechanism, but instead 48 processes queues in round-robin order. 49 50 ADVANTAGE: 51 52 - It is very cheap. Both CPU and memory requirements are minimal. 53 54 DRAWBACKS: 55 56 - "Stochastic" -> It is not 100% fair. 57 When hash collisions occur, several flows are considered as one. 58 59 - "Round-robin" -> It introduces larger delays than virtual clock 60 based schemes, and should not be used for isolating interactive 61 traffic from non-interactive. It means, that this scheduler 62 should be used as leaf of CBQ or P3, which put interactive traffic 63 to higher priority band. 64 65 We still need true WFQ for top level CSZ, but using WFQ 66 for the best effort traffic is absolutely pointless: 67 SFQ is superior for this purpose. 68 69 IMPLEMENTATION: 70 This implementation limits : 71 - maximal queue length per flow to 127 packets. 72 - max mtu to 2^18-1; 73 - max 65408 flows, 74 - number of hash buckets to 65536. 75 76 It is easy to increase these values, but not in flight. */ 77 78 #define SFQ_MAX_DEPTH 127 /* max number of packets per flow */ 79 #define SFQ_DEFAULT_FLOWS 128 80 #define SFQ_MAX_FLOWS (0x10000 - SFQ_MAX_DEPTH - 1) /* max number of flows */ 81 #define SFQ_EMPTY_SLOT 0xffff 82 #define SFQ_DEFAULT_HASH_DIVISOR 1024 83 84 /* We use 16 bits to store allot, and want to handle packets up to 64K 85 * Scale allot by 8 (1<<3) so that no overflow occurs. 86 */ 87 #define SFQ_ALLOT_SHIFT 3 88 #define SFQ_ALLOT_SIZE(X) DIV_ROUND_UP(X, 1 << SFQ_ALLOT_SHIFT) 89 90 /* This type should contain at least SFQ_MAX_DEPTH + 1 + SFQ_MAX_FLOWS values */ 91 typedef u16 sfq_index; 92 93 /* 94 * We dont use pointers to save space. 95 * Small indexes [0 ... SFQ_MAX_FLOWS - 1] are 'pointers' to slots[] array 96 * while following values [SFQ_MAX_FLOWS ... SFQ_MAX_FLOWS + SFQ_MAX_DEPTH] 97 * are 'pointers' to dep[] array 98 */ 99 struct sfq_head { 100 sfq_index next; 101 sfq_index prev; 102 }; 103 104 struct sfq_slot { 105 struct sk_buff *skblist_next; 106 struct sk_buff *skblist_prev; 107 sfq_index qlen; /* number of skbs in skblist */ 108 sfq_index next; /* next slot in sfq RR chain */ 109 struct sfq_head dep; /* anchor in dep[] chains */ 110 unsigned short hash; /* hash value (index in ht[]) */ 111 short allot; /* credit for this slot */ 112 113 unsigned int backlog; 114 struct red_vars vars; 115 }; 116 117 struct sfq_sched_data { 118 /* frequently used fields */ 119 int limit; /* limit of total number of packets in this qdisc */ 120 unsigned int divisor; /* number of slots in hash table */ 121 u8 headdrop; 122 u8 maxdepth; /* limit of packets per flow */ 123 124 u32 perturbation; 125 u8 cur_depth; /* depth of longest slot */ 126 u8 flags; 127 unsigned short scaled_quantum; /* SFQ_ALLOT_SIZE(quantum) */ 128 struct tcf_proto __rcu *filter_list; 129 sfq_index *ht; /* Hash table ('divisor' slots) */ 130 struct sfq_slot *slots; /* Flows table ('maxflows' entries) */ 131 132 struct red_parms *red_parms; 133 struct tc_sfqred_stats stats; 134 struct sfq_slot *tail; /* current slot in round */ 135 136 struct sfq_head dep[SFQ_MAX_DEPTH + 1]; 137 /* Linked lists of slots, indexed by depth 138 * dep[0] : list of unused flows 139 * dep[1] : list of flows with 1 packet 140 * dep[X] : list of flows with X packets 141 */ 142 143 unsigned int maxflows; /* number of flows in flows array */ 144 int perturb_period; 145 unsigned int quantum; /* Allotment per round: MUST BE >= MTU */ 146 struct timer_list perturb_timer; 147 }; 148 149 /* 150 * sfq_head are either in a sfq_slot or in dep[] array 151 */ 152 static inline struct sfq_head *sfq_dep_head(struct sfq_sched_data *q, sfq_index val) 153 { 154 if (val < SFQ_MAX_FLOWS) 155 return &q->slots[val].dep; 156 return &q->dep[val - SFQ_MAX_FLOWS]; 157 } 158 159 /* 160 * In order to be able to quickly rehash our queue when timer changes 161 * q->perturbation, we store flow_keys in skb->cb[] 162 */ 163 struct sfq_skb_cb { 164 struct flow_keys keys; 165 }; 166 167 static inline struct sfq_skb_cb *sfq_skb_cb(const struct sk_buff *skb) 168 { 169 qdisc_cb_private_validate(skb, sizeof(struct sfq_skb_cb)); 170 return (struct sfq_skb_cb *)qdisc_skb_cb(skb)->data; 171 } 172 173 static unsigned int sfq_hash(const struct sfq_sched_data *q, 174 const struct sk_buff *skb) 175 { 176 const struct flow_keys *keys = &sfq_skb_cb(skb)->keys; 177 unsigned int hash; 178 179 hash = jhash_3words((__force u32)keys->dst, 180 (__force u32)keys->src ^ keys->ip_proto, 181 (__force u32)keys->ports, q->perturbation); 182 return hash & (q->divisor - 1); 183 } 184 185 static unsigned int sfq_classify(struct sk_buff *skb, struct Qdisc *sch, 186 int *qerr) 187 { 188 struct sfq_sched_data *q = qdisc_priv(sch); 189 struct tcf_result res; 190 struct tcf_proto *fl; 191 int result; 192 193 if (TC_H_MAJ(skb->priority) == sch->handle && 194 TC_H_MIN(skb->priority) > 0 && 195 TC_H_MIN(skb->priority) <= q->divisor) 196 return TC_H_MIN(skb->priority); 197 198 fl = rcu_dereference_bh(q->filter_list); 199 if (!fl) { 200 skb_flow_dissect(skb, &sfq_skb_cb(skb)->keys); 201 return sfq_hash(q, skb) + 1; 202 } 203 204 *qerr = NET_XMIT_SUCCESS | __NET_XMIT_BYPASS; 205 result = tc_classify(skb, fl, &res); 206 if (result >= 0) { 207 #ifdef CONFIG_NET_CLS_ACT 208 switch (result) { 209 case TC_ACT_STOLEN: 210 case TC_ACT_QUEUED: 211 *qerr = NET_XMIT_SUCCESS | __NET_XMIT_STOLEN; 212 case TC_ACT_SHOT: 213 return 0; 214 } 215 #endif 216 if (TC_H_MIN(res.classid) <= q->divisor) 217 return TC_H_MIN(res.classid); 218 } 219 return 0; 220 } 221 222 /* 223 * x : slot number [0 .. SFQ_MAX_FLOWS - 1] 224 */ 225 static inline void sfq_link(struct sfq_sched_data *q, sfq_index x) 226 { 227 sfq_index p, n; 228 struct sfq_slot *slot = &q->slots[x]; 229 int qlen = slot->qlen; 230 231 p = qlen + SFQ_MAX_FLOWS; 232 n = q->dep[qlen].next; 233 234 slot->dep.next = n; 235 slot->dep.prev = p; 236 237 q->dep[qlen].next = x; /* sfq_dep_head(q, p)->next = x */ 238 sfq_dep_head(q, n)->prev = x; 239 } 240 241 #define sfq_unlink(q, x, n, p) \ 242 do { \ 243 n = q->slots[x].dep.next; \ 244 p = q->slots[x].dep.prev; \ 245 sfq_dep_head(q, p)->next = n; \ 246 sfq_dep_head(q, n)->prev = p; \ 247 } while (0) 248 249 250 static inline void sfq_dec(struct sfq_sched_data *q, sfq_index x) 251 { 252 sfq_index p, n; 253 int d; 254 255 sfq_unlink(q, x, n, p); 256 257 d = q->slots[x].qlen--; 258 if (n == p && q->cur_depth == d) 259 q->cur_depth--; 260 sfq_link(q, x); 261 } 262 263 static inline void sfq_inc(struct sfq_sched_data *q, sfq_index x) 264 { 265 sfq_index p, n; 266 int d; 267 268 sfq_unlink(q, x, n, p); 269 270 d = ++q->slots[x].qlen; 271 if (q->cur_depth < d) 272 q->cur_depth = d; 273 sfq_link(q, x); 274 } 275 276 /* helper functions : might be changed when/if skb use a standard list_head */ 277 278 /* remove one skb from tail of slot queue */ 279 static inline struct sk_buff *slot_dequeue_tail(struct sfq_slot *slot) 280 { 281 struct sk_buff *skb = slot->skblist_prev; 282 283 slot->skblist_prev = skb->prev; 284 skb->prev->next = (struct sk_buff *)slot; 285 skb->next = skb->prev = NULL; 286 return skb; 287 } 288 289 /* remove one skb from head of slot queue */ 290 static inline struct sk_buff *slot_dequeue_head(struct sfq_slot *slot) 291 { 292 struct sk_buff *skb = slot->skblist_next; 293 294 slot->skblist_next = skb->next; 295 skb->next->prev = (struct sk_buff *)slot; 296 skb->next = skb->prev = NULL; 297 return skb; 298 } 299 300 static inline void slot_queue_init(struct sfq_slot *slot) 301 { 302 memset(slot, 0, sizeof(*slot)); 303 slot->skblist_prev = slot->skblist_next = (struct sk_buff *)slot; 304 } 305 306 /* add skb to slot queue (tail add) */ 307 static inline void slot_queue_add(struct sfq_slot *slot, struct sk_buff *skb) 308 { 309 skb->prev = slot->skblist_prev; 310 skb->next = (struct sk_buff *)slot; 311 slot->skblist_prev->next = skb; 312 slot->skblist_prev = skb; 313 } 314 315 static unsigned int sfq_drop(struct Qdisc *sch) 316 { 317 struct sfq_sched_data *q = qdisc_priv(sch); 318 sfq_index x, d = q->cur_depth; 319 struct sk_buff *skb; 320 unsigned int len; 321 struct sfq_slot *slot; 322 323 /* Queue is full! Find the longest slot and drop tail packet from it */ 324 if (d > 1) { 325 x = q->dep[d].next; 326 slot = &q->slots[x]; 327 drop: 328 skb = q->headdrop ? slot_dequeue_head(slot) : slot_dequeue_tail(slot); 329 len = qdisc_pkt_len(skb); 330 slot->backlog -= len; 331 sfq_dec(q, x); 332 kfree_skb(skb); 333 sch->q.qlen--; 334 qdisc_qstats_drop(sch); 335 qdisc_qstats_backlog_dec(sch, skb); 336 return len; 337 } 338 339 if (d == 1) { 340 /* It is difficult to believe, but ALL THE SLOTS HAVE LENGTH 1. */ 341 x = q->tail->next; 342 slot = &q->slots[x]; 343 q->tail->next = slot->next; 344 q->ht[slot->hash] = SFQ_EMPTY_SLOT; 345 goto drop; 346 } 347 348 return 0; 349 } 350 351 /* Is ECN parameter configured */ 352 static int sfq_prob_mark(const struct sfq_sched_data *q) 353 { 354 return q->flags & TC_RED_ECN; 355 } 356 357 /* Should packets over max threshold just be marked */ 358 static int sfq_hard_mark(const struct sfq_sched_data *q) 359 { 360 return (q->flags & (TC_RED_ECN | TC_RED_HARDDROP)) == TC_RED_ECN; 361 } 362 363 static int sfq_headdrop(const struct sfq_sched_data *q) 364 { 365 return q->headdrop; 366 } 367 368 static int 369 sfq_enqueue(struct sk_buff *skb, struct Qdisc *sch) 370 { 371 struct sfq_sched_data *q = qdisc_priv(sch); 372 unsigned int hash; 373 sfq_index x, qlen; 374 struct sfq_slot *slot; 375 int uninitialized_var(ret); 376 struct sk_buff *head; 377 int delta; 378 379 hash = sfq_classify(skb, sch, &ret); 380 if (hash == 0) { 381 if (ret & __NET_XMIT_BYPASS) 382 qdisc_qstats_drop(sch); 383 kfree_skb(skb); 384 return ret; 385 } 386 hash--; 387 388 x = q->ht[hash]; 389 slot = &q->slots[x]; 390 if (x == SFQ_EMPTY_SLOT) { 391 x = q->dep[0].next; /* get a free slot */ 392 if (x >= SFQ_MAX_FLOWS) 393 return qdisc_drop(skb, sch); 394 q->ht[hash] = x; 395 slot = &q->slots[x]; 396 slot->hash = hash; 397 slot->backlog = 0; /* should already be 0 anyway... */ 398 red_set_vars(&slot->vars); 399 goto enqueue; 400 } 401 if (q->red_parms) { 402 slot->vars.qavg = red_calc_qavg_no_idle_time(q->red_parms, 403 &slot->vars, 404 slot->backlog); 405 switch (red_action(q->red_parms, 406 &slot->vars, 407 slot->vars.qavg)) { 408 case RED_DONT_MARK: 409 break; 410 411 case RED_PROB_MARK: 412 qdisc_qstats_overlimit(sch); 413 if (sfq_prob_mark(q)) { 414 /* We know we have at least one packet in queue */ 415 if (sfq_headdrop(q) && 416 INET_ECN_set_ce(slot->skblist_next)) { 417 q->stats.prob_mark_head++; 418 break; 419 } 420 if (INET_ECN_set_ce(skb)) { 421 q->stats.prob_mark++; 422 break; 423 } 424 } 425 q->stats.prob_drop++; 426 goto congestion_drop; 427 428 case RED_HARD_MARK: 429 qdisc_qstats_overlimit(sch); 430 if (sfq_hard_mark(q)) { 431 /* We know we have at least one packet in queue */ 432 if (sfq_headdrop(q) && 433 INET_ECN_set_ce(slot->skblist_next)) { 434 q->stats.forced_mark_head++; 435 break; 436 } 437 if (INET_ECN_set_ce(skb)) { 438 q->stats.forced_mark++; 439 break; 440 } 441 } 442 q->stats.forced_drop++; 443 goto congestion_drop; 444 } 445 } 446 447 if (slot->qlen >= q->maxdepth) { 448 congestion_drop: 449 if (!sfq_headdrop(q)) 450 return qdisc_drop(skb, sch); 451 452 /* We know we have at least one packet in queue */ 453 head = slot_dequeue_head(slot); 454 delta = qdisc_pkt_len(head) - qdisc_pkt_len(skb); 455 sch->qstats.backlog -= delta; 456 slot->backlog -= delta; 457 qdisc_drop(head, sch); 458 459 slot_queue_add(slot, skb); 460 return NET_XMIT_CN; 461 } 462 463 enqueue: 464 qdisc_qstats_backlog_inc(sch, skb); 465 slot->backlog += qdisc_pkt_len(skb); 466 slot_queue_add(slot, skb); 467 sfq_inc(q, x); 468 if (slot->qlen == 1) { /* The flow is new */ 469 if (q->tail == NULL) { /* It is the first flow */ 470 slot->next = x; 471 } else { 472 slot->next = q->tail->next; 473 q->tail->next = x; 474 } 475 /* We put this flow at the end of our flow list. 476 * This might sound unfair for a new flow to wait after old ones, 477 * but we could endup servicing new flows only, and freeze old ones. 478 */ 479 q->tail = slot; 480 /* We could use a bigger initial quantum for new flows */ 481 slot->allot = q->scaled_quantum; 482 } 483 if (++sch->q.qlen <= q->limit) 484 return NET_XMIT_SUCCESS; 485 486 qlen = slot->qlen; 487 sfq_drop(sch); 488 /* Return Congestion Notification only if we dropped a packet 489 * from this flow. 490 */ 491 if (qlen != slot->qlen) 492 return NET_XMIT_CN; 493 494 /* As we dropped a packet, better let upper stack know this */ 495 qdisc_tree_decrease_qlen(sch, 1); 496 return NET_XMIT_SUCCESS; 497 } 498 499 static struct sk_buff * 500 sfq_dequeue(struct Qdisc *sch) 501 { 502 struct sfq_sched_data *q = qdisc_priv(sch); 503 struct sk_buff *skb; 504 sfq_index a, next_a; 505 struct sfq_slot *slot; 506 507 /* No active slots */ 508 if (q->tail == NULL) 509 return NULL; 510 511 next_slot: 512 a = q->tail->next; 513 slot = &q->slots[a]; 514 if (slot->allot <= 0) { 515 q->tail = slot; 516 slot->allot += q->scaled_quantum; 517 goto next_slot; 518 } 519 skb = slot_dequeue_head(slot); 520 sfq_dec(q, a); 521 qdisc_bstats_update(sch, skb); 522 sch->q.qlen--; 523 qdisc_qstats_backlog_dec(sch, skb); 524 slot->backlog -= qdisc_pkt_len(skb); 525 /* Is the slot empty? */ 526 if (slot->qlen == 0) { 527 q->ht[slot->hash] = SFQ_EMPTY_SLOT; 528 next_a = slot->next; 529 if (a == next_a) { 530 q->tail = NULL; /* no more active slots */ 531 return skb; 532 } 533 q->tail->next = next_a; 534 } else { 535 slot->allot -= SFQ_ALLOT_SIZE(qdisc_pkt_len(skb)); 536 } 537 return skb; 538 } 539 540 static void 541 sfq_reset(struct Qdisc *sch) 542 { 543 struct sk_buff *skb; 544 545 while ((skb = sfq_dequeue(sch)) != NULL) 546 kfree_skb(skb); 547 } 548 549 /* 550 * When q->perturbation is changed, we rehash all queued skbs 551 * to avoid OOO (Out Of Order) effects. 552 * We dont use sfq_dequeue()/sfq_enqueue() because we dont want to change 553 * counters. 554 */ 555 static void sfq_rehash(struct Qdisc *sch) 556 { 557 struct sfq_sched_data *q = qdisc_priv(sch); 558 struct sk_buff *skb; 559 int i; 560 struct sfq_slot *slot; 561 struct sk_buff_head list; 562 int dropped = 0; 563 564 __skb_queue_head_init(&list); 565 566 for (i = 0; i < q->maxflows; i++) { 567 slot = &q->slots[i]; 568 if (!slot->qlen) 569 continue; 570 while (slot->qlen) { 571 skb = slot_dequeue_head(slot); 572 sfq_dec(q, i); 573 __skb_queue_tail(&list, skb); 574 } 575 slot->backlog = 0; 576 red_set_vars(&slot->vars); 577 q->ht[slot->hash] = SFQ_EMPTY_SLOT; 578 } 579 q->tail = NULL; 580 581 while ((skb = __skb_dequeue(&list)) != NULL) { 582 unsigned int hash = sfq_hash(q, skb); 583 sfq_index x = q->ht[hash]; 584 585 slot = &q->slots[x]; 586 if (x == SFQ_EMPTY_SLOT) { 587 x = q->dep[0].next; /* get a free slot */ 588 if (x >= SFQ_MAX_FLOWS) { 589 drop: 590 qdisc_qstats_backlog_dec(sch, skb); 591 kfree_skb(skb); 592 dropped++; 593 continue; 594 } 595 q->ht[hash] = x; 596 slot = &q->slots[x]; 597 slot->hash = hash; 598 } 599 if (slot->qlen >= q->maxdepth) 600 goto drop; 601 slot_queue_add(slot, skb); 602 if (q->red_parms) 603 slot->vars.qavg = red_calc_qavg(q->red_parms, 604 &slot->vars, 605 slot->backlog); 606 slot->backlog += qdisc_pkt_len(skb); 607 sfq_inc(q, x); 608 if (slot->qlen == 1) { /* The flow is new */ 609 if (q->tail == NULL) { /* It is the first flow */ 610 slot->next = x; 611 } else { 612 slot->next = q->tail->next; 613 q->tail->next = x; 614 } 615 q->tail = slot; 616 slot->allot = q->scaled_quantum; 617 } 618 } 619 sch->q.qlen -= dropped; 620 qdisc_tree_decrease_qlen(sch, dropped); 621 } 622 623 static void sfq_perturbation(unsigned long arg) 624 { 625 struct Qdisc *sch = (struct Qdisc *)arg; 626 struct sfq_sched_data *q = qdisc_priv(sch); 627 spinlock_t *root_lock = qdisc_lock(qdisc_root_sleeping(sch)); 628 629 spin_lock(root_lock); 630 q->perturbation = prandom_u32(); 631 if (!q->filter_list && q->tail) 632 sfq_rehash(sch); 633 spin_unlock(root_lock); 634 635 if (q->perturb_period) 636 mod_timer(&q->perturb_timer, jiffies + q->perturb_period); 637 } 638 639 static int sfq_change(struct Qdisc *sch, struct nlattr *opt) 640 { 641 struct sfq_sched_data *q = qdisc_priv(sch); 642 struct tc_sfq_qopt *ctl = nla_data(opt); 643 struct tc_sfq_qopt_v1 *ctl_v1 = NULL; 644 unsigned int qlen; 645 struct red_parms *p = NULL; 646 647 if (opt->nla_len < nla_attr_size(sizeof(*ctl))) 648 return -EINVAL; 649 if (opt->nla_len >= nla_attr_size(sizeof(*ctl_v1))) 650 ctl_v1 = nla_data(opt); 651 if (ctl->divisor && 652 (!is_power_of_2(ctl->divisor) || ctl->divisor > 65536)) 653 return -EINVAL; 654 if (ctl_v1 && ctl_v1->qth_min) { 655 p = kmalloc(sizeof(*p), GFP_KERNEL); 656 if (!p) 657 return -ENOMEM; 658 } 659 sch_tree_lock(sch); 660 if (ctl->quantum) { 661 q->quantum = ctl->quantum; 662 q->scaled_quantum = SFQ_ALLOT_SIZE(q->quantum); 663 } 664 q->perturb_period = ctl->perturb_period * HZ; 665 if (ctl->flows) 666 q->maxflows = min_t(u32, ctl->flows, SFQ_MAX_FLOWS); 667 if (ctl->divisor) { 668 q->divisor = ctl->divisor; 669 q->maxflows = min_t(u32, q->maxflows, q->divisor); 670 } 671 if (ctl_v1) { 672 if (ctl_v1->depth) 673 q->maxdepth = min_t(u32, ctl_v1->depth, SFQ_MAX_DEPTH); 674 if (p) { 675 swap(q->red_parms, p); 676 red_set_parms(q->red_parms, 677 ctl_v1->qth_min, ctl_v1->qth_max, 678 ctl_v1->Wlog, 679 ctl_v1->Plog, ctl_v1->Scell_log, 680 NULL, 681 ctl_v1->max_P); 682 } 683 q->flags = ctl_v1->flags; 684 q->headdrop = ctl_v1->headdrop; 685 } 686 if (ctl->limit) { 687 q->limit = min_t(u32, ctl->limit, q->maxdepth * q->maxflows); 688 q->maxflows = min_t(u32, q->maxflows, q->limit); 689 } 690 691 qlen = sch->q.qlen; 692 while (sch->q.qlen > q->limit) 693 sfq_drop(sch); 694 qdisc_tree_decrease_qlen(sch, qlen - sch->q.qlen); 695 696 del_timer(&q->perturb_timer); 697 if (q->perturb_period) { 698 mod_timer(&q->perturb_timer, jiffies + q->perturb_period); 699 q->perturbation = prandom_u32(); 700 } 701 sch_tree_unlock(sch); 702 kfree(p); 703 return 0; 704 } 705 706 static void *sfq_alloc(size_t sz) 707 { 708 void *ptr = kmalloc(sz, GFP_KERNEL | __GFP_NOWARN); 709 710 if (!ptr) 711 ptr = vmalloc(sz); 712 return ptr; 713 } 714 715 static void sfq_free(void *addr) 716 { 717 kvfree(addr); 718 } 719 720 static void sfq_destroy(struct Qdisc *sch) 721 { 722 struct sfq_sched_data *q = qdisc_priv(sch); 723 724 tcf_destroy_chain(&q->filter_list); 725 q->perturb_period = 0; 726 del_timer_sync(&q->perturb_timer); 727 sfq_free(q->ht); 728 sfq_free(q->slots); 729 kfree(q->red_parms); 730 } 731 732 static int sfq_init(struct Qdisc *sch, struct nlattr *opt) 733 { 734 struct sfq_sched_data *q = qdisc_priv(sch); 735 int i; 736 737 q->perturb_timer.function = sfq_perturbation; 738 q->perturb_timer.data = (unsigned long)sch; 739 init_timer_deferrable(&q->perturb_timer); 740 741 for (i = 0; i < SFQ_MAX_DEPTH + 1; i++) { 742 q->dep[i].next = i + SFQ_MAX_FLOWS; 743 q->dep[i].prev = i + SFQ_MAX_FLOWS; 744 } 745 746 q->limit = SFQ_MAX_DEPTH; 747 q->maxdepth = SFQ_MAX_DEPTH; 748 q->cur_depth = 0; 749 q->tail = NULL; 750 q->divisor = SFQ_DEFAULT_HASH_DIVISOR; 751 q->maxflows = SFQ_DEFAULT_FLOWS; 752 q->quantum = psched_mtu(qdisc_dev(sch)); 753 q->scaled_quantum = SFQ_ALLOT_SIZE(q->quantum); 754 q->perturb_period = 0; 755 q->perturbation = prandom_u32(); 756 757 if (opt) { 758 int err = sfq_change(sch, opt); 759 if (err) 760 return err; 761 } 762 763 q->ht = sfq_alloc(sizeof(q->ht[0]) * q->divisor); 764 q->slots = sfq_alloc(sizeof(q->slots[0]) * q->maxflows); 765 if (!q->ht || !q->slots) { 766 sfq_destroy(sch); 767 return -ENOMEM; 768 } 769 for (i = 0; i < q->divisor; i++) 770 q->ht[i] = SFQ_EMPTY_SLOT; 771 772 for (i = 0; i < q->maxflows; i++) { 773 slot_queue_init(&q->slots[i]); 774 sfq_link(q, i); 775 } 776 if (q->limit >= 1) 777 sch->flags |= TCQ_F_CAN_BYPASS; 778 else 779 sch->flags &= ~TCQ_F_CAN_BYPASS; 780 return 0; 781 } 782 783 static int sfq_dump(struct Qdisc *sch, struct sk_buff *skb) 784 { 785 struct sfq_sched_data *q = qdisc_priv(sch); 786 unsigned char *b = skb_tail_pointer(skb); 787 struct tc_sfq_qopt_v1 opt; 788 struct red_parms *p = q->red_parms; 789 790 memset(&opt, 0, sizeof(opt)); 791 opt.v0.quantum = q->quantum; 792 opt.v0.perturb_period = q->perturb_period / HZ; 793 opt.v0.limit = q->limit; 794 opt.v0.divisor = q->divisor; 795 opt.v0.flows = q->maxflows; 796 opt.depth = q->maxdepth; 797 opt.headdrop = q->headdrop; 798 799 if (p) { 800 opt.qth_min = p->qth_min >> p->Wlog; 801 opt.qth_max = p->qth_max >> p->Wlog; 802 opt.Wlog = p->Wlog; 803 opt.Plog = p->Plog; 804 opt.Scell_log = p->Scell_log; 805 opt.max_P = p->max_P; 806 } 807 memcpy(&opt.stats, &q->stats, sizeof(opt.stats)); 808 opt.flags = q->flags; 809 810 if (nla_put(skb, TCA_OPTIONS, sizeof(opt), &opt)) 811 goto nla_put_failure; 812 813 return skb->len; 814 815 nla_put_failure: 816 nlmsg_trim(skb, b); 817 return -1; 818 } 819 820 static struct Qdisc *sfq_leaf(struct Qdisc *sch, unsigned long arg) 821 { 822 return NULL; 823 } 824 825 static unsigned long sfq_get(struct Qdisc *sch, u32 classid) 826 { 827 return 0; 828 } 829 830 static unsigned long sfq_bind(struct Qdisc *sch, unsigned long parent, 831 u32 classid) 832 { 833 /* we cannot bypass queue discipline anymore */ 834 sch->flags &= ~TCQ_F_CAN_BYPASS; 835 return 0; 836 } 837 838 static void sfq_put(struct Qdisc *q, unsigned long cl) 839 { 840 } 841 842 static struct tcf_proto __rcu **sfq_find_tcf(struct Qdisc *sch, 843 unsigned long cl) 844 { 845 struct sfq_sched_data *q = qdisc_priv(sch); 846 847 if (cl) 848 return NULL; 849 return &q->filter_list; 850 } 851 852 static int sfq_dump_class(struct Qdisc *sch, unsigned long cl, 853 struct sk_buff *skb, struct tcmsg *tcm) 854 { 855 tcm->tcm_handle |= TC_H_MIN(cl); 856 return 0; 857 } 858 859 static int sfq_dump_class_stats(struct Qdisc *sch, unsigned long cl, 860 struct gnet_dump *d) 861 { 862 struct sfq_sched_data *q = qdisc_priv(sch); 863 sfq_index idx = q->ht[cl - 1]; 864 struct gnet_stats_queue qs = { 0 }; 865 struct tc_sfq_xstats xstats = { 0 }; 866 867 if (idx != SFQ_EMPTY_SLOT) { 868 const struct sfq_slot *slot = &q->slots[idx]; 869 870 xstats.allot = slot->allot << SFQ_ALLOT_SHIFT; 871 qs.qlen = slot->qlen; 872 qs.backlog = slot->backlog; 873 } 874 if (gnet_stats_copy_queue(d, NULL, &qs, qs.qlen) < 0) 875 return -1; 876 return gnet_stats_copy_app(d, &xstats, sizeof(xstats)); 877 } 878 879 static void sfq_walk(struct Qdisc *sch, struct qdisc_walker *arg) 880 { 881 struct sfq_sched_data *q = qdisc_priv(sch); 882 unsigned int i; 883 884 if (arg->stop) 885 return; 886 887 for (i = 0; i < q->divisor; i++) { 888 if (q->ht[i] == SFQ_EMPTY_SLOT || 889 arg->count < arg->skip) { 890 arg->count++; 891 continue; 892 } 893 if (arg->fn(sch, i + 1, arg) < 0) { 894 arg->stop = 1; 895 break; 896 } 897 arg->count++; 898 } 899 } 900 901 static const struct Qdisc_class_ops sfq_class_ops = { 902 .leaf = sfq_leaf, 903 .get = sfq_get, 904 .put = sfq_put, 905 .tcf_chain = sfq_find_tcf, 906 .bind_tcf = sfq_bind, 907 .unbind_tcf = sfq_put, 908 .dump = sfq_dump_class, 909 .dump_stats = sfq_dump_class_stats, 910 .walk = sfq_walk, 911 }; 912 913 static struct Qdisc_ops sfq_qdisc_ops __read_mostly = { 914 .cl_ops = &sfq_class_ops, 915 .id = "sfq", 916 .priv_size = sizeof(struct sfq_sched_data), 917 .enqueue = sfq_enqueue, 918 .dequeue = sfq_dequeue, 919 .peek = qdisc_peek_dequeued, 920 .drop = sfq_drop, 921 .init = sfq_init, 922 .reset = sfq_reset, 923 .destroy = sfq_destroy, 924 .change = NULL, 925 .dump = sfq_dump, 926 .owner = THIS_MODULE, 927 }; 928 929 static int __init sfq_module_init(void) 930 { 931 return register_qdisc(&sfq_qdisc_ops); 932 } 933 static void __exit sfq_module_exit(void) 934 { 935 unregister_qdisc(&sfq_qdisc_ops); 936 } 937 module_init(sfq_module_init) 938 module_exit(sfq_module_exit) 939 MODULE_LICENSE("GPL"); 940