net/sched/sch_choke.c

   1 /*
   2  * net/sched/sch_choke.c        CHOKE scheduler
   3  *
   4  * Copyright (c) 2011 Stephen Hemminger <shemminger@vyatta.com>
   5  * Copyright (c) 2011 Eric Dumazet <eric.dumazet@gmail.com>
   6  *
   7  * This program is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU General Public License
   9  * version 2 as published by the Free Software Foundation.
  10  *
  11  */
  12
  13 #include <linux/module.h>
  14 #include <linux/types.h>
  15 #include <linux/kernel.h>
  16 #include <linux/skbuff.h>
  17 #include <linux/vmalloc.h>
  18 #include <net/pkt_sched.h>
  19 #include <net/inet_ecn.h>
  20 #include <net/red.h>
  21 #include <net/flow_keys.h>
  22
  23 /*
  24    CHOKe stateless AQM for fair bandwidth allocation
  25    =================================================
  26
  27    CHOKe (CHOose and Keep for responsive flows, CHOose and Kill for
  28    unresponsive flows) is a variant of RED that penalizes misbehaving flows but
  29    maintains no flow state. The difference from RED is an additional step
  30    during the enqueuing process. If average queue size is over the
  31    low threshold (qmin), a packet is chosen at random from the queue.
  32    If both the new and chosen packet are from the same flow, both
  33    are dropped. Unlike RED, CHOKe is not really a "classful" qdisc because it
  34    needs to access packets in queue randomly. It has a minimal class
  35    interface to allow overriding the builtin flow classifier with
  36    filters.
  37
  38    Source:
  39    R. Pan, B. Prabhakar, and K. Psounis, "CHOKe, A Stateless
  40    Active Queue Management Scheme for Approximating Fair Bandwidth Allocation",
  41    IEEE INFOCOM, 2000.
  42
  43    A. Tang, J. Wang, S. Low, "Understanding CHOKe: Throughput and Spatial
  44    Characteristics", IEEE/ACM Transactions on Networking, 2004
  45
  46  */
  47
  48 /* Upper bound on size of sk_buff table (packets) */
  49 #define CHOKE_MAX_QUEUE (128*1024 - 1)
  50
  51 struct choke_sched_data {
  52 /* Parameters */
  53         u32              limit;
  54         unsigned char    flags;
  55
  56         struct red_parms parms;
  57
  58 /* Variables */
  59         struct red_vars  vars;
  60         struct tcf_proto *filter_list;
  61         struct {
  62                 u32     prob_drop;      /* Early probability drops */
  63                 u32     prob_mark;      /* Early probability marks */
  64                 u32     forced_drop;    /* Forced drops, qavg > max_thresh */
  65                 u32     forced_mark;    /* Forced marks, qavg > max_thresh */
  66                 u32     pdrop;          /* Drops due to queue limits */
  67                 u32     other;          /* Drops due to drop() calls */
  68                 u32     matched;        /* Drops to flow match */
  69         } stats;
  70
  71         unsigned int     head;
  72         unsigned int     tail;
  73
  74         unsigned int     tab_mask; /* size - 1 */
  75
  76         struct sk_buff **tab;
  77 };
  78
  79 /* number of elements in queue including holes */
  80 static unsigned int choke_len(const struct choke_sched_data *q)
  81 {
  82         return (q->tail - q->head) & q->tab_mask;
  83 }
  84
  85 /* Is ECN parameter configured */
  86 static int use_ecn(const struct choke_sched_data *q)
  87 {
  88         return q->flags & TC_RED_ECN;
  89 }
  90
  91 /* Should packets over max just be dropped (versus marked) */
  92 static int use_harddrop(const struct choke_sched_data *q)
  93 {
  94         return q->flags & TC_RED_HARDDROP;
  95 }
  96
  97 /* Move head pointer forward to skip over holes */
  98 static void choke_zap_head_holes(struct choke_sched_data *q)
  99 {
 100         do {
 101                 q->head = (q->head + 1) & q->tab_mask;
 102                 if (q->head == q->tail)
 103                         break;
 104         } while (q->tab[q->head] == NULL);
 105 }
 106
 107 /* Move tail pointer backwards to reuse holes */
 108 static void choke_zap_tail_holes(struct choke_sched_data *q)
 109 {
 110         do {
 111                 q->tail = (q->tail - 1) & q->tab_mask;
 112                 if (q->head == q->tail)
 113                         break;
 114         } while (q->tab[q->tail] == NULL);
 115 }
 116
 117 /* Drop packet from queue array by creating a "hole" */
 118 static void choke_drop_by_idx(struct Qdisc *sch, unsigned int idx)
 119 {
 120         struct choke_sched_data *q = qdisc_priv(sch);
 121         struct sk_buff *skb = q->tab[idx];
 122
 123         q->tab[idx] = NULL;
 124
 125         if (idx == q->head)
 126                 choke_zap_head_holes(q);
 127         if (idx == q->tail)
 128                 choke_zap_tail_holes(q);
 129
 130         sch->qstats.backlog -= qdisc_pkt_len(skb);
 131         qdisc_drop(skb, sch);
 132         qdisc_tree_decrease_qlen(sch, 1);
 133         --sch->q.qlen;
 134 }
 135
 136 struct choke_skb_cb {
 137         u16                     classid;
 138         u8                      keys_valid;
 139         struct flow_keys        keys;
 140 };
 141
 142 static inline struct choke_skb_cb *choke_skb_cb(const struct sk_buff *skb)
 143 {
 144         qdisc_cb_private_validate(skb, sizeof(struct choke_skb_cb));
 145         return (struct choke_skb_cb *)qdisc_skb_cb(skb)->data;
 146 }
 147
 148 static inline void choke_set_classid(struct sk_buff *skb, u16 classid)
 149 {
 150         choke_skb_cb(skb)->classid = classid;
 151 }
 152
 153 static u16 choke_get_classid(const struct sk_buff *skb)
 154 {
 155         return choke_skb_cb(skb)->classid;
 156 }
 157
 158 /*
 159  * Compare flow of two packets
 160  *  Returns true only if source and destination address and port match.
 161  *          false for special cases
 162  */
 163 static bool choke_match_flow(struct sk_buff *skb1,
 164                              struct sk_buff *skb2)
 165 {
 166         if (skb1->protocol != skb2->protocol)
 167                 return false;
 168
 169         if (!choke_skb_cb(skb1)->keys_valid) {
 170                 choke_skb_cb(skb1)->keys_valid = 1;
 171                 skb_flow_dissect(skb1, &choke_skb_cb(skb1)->keys);
 172         }
 173
 174         if (!choke_skb_cb(skb2)->keys_valid) {
 175                 choke_skb_cb(skb2)->keys_valid = 1;
 176                 skb_flow_dissect(skb2, &choke_skb_cb(skb2)->keys);
 177         }
 178
 179         return !memcmp(&choke_skb_cb(skb1)->keys,
 180                        &choke_skb_cb(skb2)->keys,
 181                        sizeof(struct flow_keys));
 182 }
 183
 184 /*
 185  * Classify flow using either:
 186  *  1. pre-existing classification result in skb
 187  *  2. fast internal classification
 188  *  3. use TC filter based classification
 189  */
 190 static bool choke_classify(struct sk_buff *skb,
 191                            struct Qdisc *sch, int *qerr)
 192
 193 {
 194         struct choke_sched_data *q = qdisc_priv(sch);
 195         struct tcf_result res;
 196         int result;
 197
 198         result = tc_classify(skb, q->filter_list, &res);
 199         if (result >= 0) {
 200 #ifdef CONFIG_NET_CLS_ACT
 201                 switch (result) {
 202                 case TC_ACT_STOLEN:
 203                 case TC_ACT_QUEUED:
 204                         *qerr = NET_XMIT_SUCCESS | __NET_XMIT_STOLEN;
 205                 case TC_ACT_SHOT:
 206                         return false;
 207                 }
 208 #endif
 209                 choke_set_classid(skb, TC_H_MIN(res.classid));
 210                 return true;
 211         }
 212
 213         return false;
 214 }
 215
 216 /*
 217  * Select a packet at random from queue
 218  * HACK: since queue can have holes from previous deletion; retry several
 219  *   times to find a random skb but then just give up and return the head
 220  * Will return NULL if queue is empty (q->head == q->tail)
 221  */
 222 static struct sk_buff *choke_peek_random(const struct choke_sched_data *q,
 223                                          unsigned int *pidx)
 224 {
 225         struct sk_buff *skb;
 226         int retrys = 3;
 227
 228         do {
 229                 *pidx = (q->head + prandom_u32_max(choke_len(q))) & q->tab_mask;
 230                 skb = q->tab[*pidx];
 231                 if (skb)
 232                         return skb;
 233         } while (--retrys > 0);
 234
 235         return q->tab[*pidx = q->head];
 236 }
 237
 238 /*
 239  * Compare new packet with random packet in queue
 240  * returns true if matched and sets *pidx
 241  */
 242 static bool choke_match_random(const struct choke_sched_data *q,
 243                                struct sk_buff *nskb,
 244                                unsigned int *pidx)
 245 {
 246         struct sk_buff *oskb;
 247
 248         if (q->head == q->tail)
 249                 return false;
 250
 251         oskb = choke_peek_random(q, pidx);
 252         if (q->filter_list)
 253                 return choke_get_classid(nskb) == choke_get_classid(oskb);
 254
 255         return choke_match_flow(oskb, nskb);
 256 }
 257
 258 static int choke_enqueue(struct sk_buff *skb, struct Qdisc *sch)
 259 {
 260         struct choke_sched_data *q = qdisc_priv(sch);
 261         const struct red_parms *p = &q->parms;
 262         int ret = NET_XMIT_SUCCESS | __NET_XMIT_BYPASS;
 263
 264         if (q->filter_list) {
 265                 /* If using external classifiers, get result and record it. */
 266                 if (!choke_classify(skb, sch, &ret))
 267                         goto other_drop;        /* Packet was eaten by filter */
 268         }
 269
 270         choke_skb_cb(skb)->keys_valid = 0;
 271         /* Compute average queue usage (see RED) */
 272         q->vars.qavg = red_calc_qavg(p, &q->vars, sch->q.qlen);
 273         if (red_is_idling(&q->vars))
 274                 red_end_of_idle_period(&q->vars);
 275
 276         /* Is queue small? */
 277         if (q->vars.qavg <= p->qth_min)
 278                 q->vars.qcount = -1;
 279         else {
 280                 unsigned int idx;
 281
 282                 /* Draw a packet at random from queue and compare flow */
 283                 if (choke_match_random(q, skb, &idx)) {
 284                         q->stats.matched++;
 285                         choke_drop_by_idx(sch, idx);
 286                         goto congestion_drop;
 287                 }
 288
 289                 /* Queue is large, always mark/drop */
 290                 if (q->vars.qavg > p->qth_max) {
 291                         q->vars.qcount = -1;
 292
 293                         sch->qstats.overlimits++;
 294                         if (use_harddrop(q) || !use_ecn(q) ||
 295                             !INET_ECN_set_ce(skb)) {
 296                                 q->stats.forced_drop++;
 297                                 goto congestion_drop;
 298                         }
 299
 300                         q->stats.forced_mark++;
 301                 } else if (++q->vars.qcount) {
 302                         if (red_mark_probability(p, &q->vars, q->vars.qavg)) {
 303                                 q->vars.qcount = 0;
 304                                 q->vars.qR = red_random(p);
 305
 306                                 sch->qstats.overlimits++;
 307                                 if (!use_ecn(q) || !INET_ECN_set_ce(skb)) {
 308                                         q->stats.prob_drop++;
 309                                         goto congestion_drop;
 310                                 }
 311
 312                                 q->stats.prob_mark++;
 313                         }
 314                 } else
 315                         q->vars.qR = red_random(p);
 316         }
 317
 318         /* Admit new packet */
 319         if (sch->q.qlen < q->limit) {
 320                 q->tab[q->tail] = skb;
 321                 q->tail = (q->tail + 1) & q->tab_mask;
 322                 ++sch->q.qlen;
 323                 sch->qstats.backlog += qdisc_pkt_len(skb);
 324                 return NET_XMIT_SUCCESS;
 325         }
 326
 327         q->stats.pdrop++;
 328         return qdisc_drop(skb, sch);
 329
 330 congestion_drop:
 331         qdisc_drop(skb, sch);
 332         return NET_XMIT_CN;
 333
 334 other_drop:
 335         if (ret & __NET_XMIT_BYPASS)
 336                 sch->qstats.drops++;
 337         kfree_skb(skb);
 338         return ret;
 339 }
 340
 341 static struct sk_buff *choke_dequeue(struct Qdisc *sch)
 342 {
 343         struct choke_sched_data *q = qdisc_priv(sch);
 344         struct sk_buff *skb;
 345
 346         if (q->head == q->tail) {
 347                 if (!red_is_idling(&q->vars))
 348                         red_start_of_idle_period(&q->vars);
 349                 return NULL;
 350         }
 351
 352         skb = q->tab[q->head];
 353         q->tab[q->head] = NULL;
 354         choke_zap_head_holes(q);
 355         --sch->q.qlen;
 356         sch->qstats.backlog -= qdisc_pkt_len(skb);
 357         qdisc_bstats_update(sch, skb);
 358
 359         return skb;
 360 }
 361
 362 static unsigned int choke_drop(struct Qdisc *sch)
 363 {
 364         struct choke_sched_data *q = qdisc_priv(sch);
 365         unsigned int len;
 366
 367         len = qdisc_queue_drop(sch);
 368         if (len > 0)
 369                 q->stats.other++;
 370         else {
 371                 if (!red_is_idling(&q->vars))
 372                         red_start_of_idle_period(&q->vars);
 373         }
 374
 375         return len;
 376 }
 377
 378 static void choke_reset(struct Qdisc *sch)
 379 {
 380         struct choke_sched_data *q = qdisc_priv(sch);
 381
 382         red_restart(&q->vars);
 383 }
 384
 385 static const struct nla_policy choke_policy[TCA_CHOKE_MAX + 1] = {
 386         [TCA_CHOKE_PARMS]       = { .len = sizeof(struct tc_red_qopt) },
 387         [TCA_CHOKE_STAB]        = { .len = RED_STAB_SIZE },
 388         [TCA_CHOKE_MAX_P]       = { .type = NLA_U32 },
 389 };
 390
 391
 392 static void choke_free(void *addr)
 393 {
 394         kvfree(addr);
 395 }
 396
 397 static int choke_change(struct Qdisc *sch, struct nlattr *opt)
 398 {
 399         struct choke_sched_data *q = qdisc_priv(sch);
 400         struct nlattr *tb[TCA_CHOKE_MAX + 1];
 401         const struct tc_red_qopt *ctl;
 402         int err;
 403         struct sk_buff **old = NULL;
 404         unsigned int mask;
 405         u32 max_P;
 406
 407         if (opt == NULL)
 408                 return -EINVAL;
 409
 410         err = nla_parse_nested(tb, TCA_CHOKE_MAX, opt, choke_policy);
 411         if (err < 0)
 412                 return err;
 413
 414         if (tb[TCA_CHOKE_PARMS] == NULL ||
 415             tb[TCA_CHOKE_STAB] == NULL)
 416                 return -EINVAL;
 417
 418         max_P = tb[TCA_CHOKE_MAX_P] ? nla_get_u32(tb[TCA_CHOKE_MAX_P]) : 0;
 419
 420         ctl = nla_data(tb[TCA_CHOKE_PARMS]);
 421
 422         if (ctl->limit > CHOKE_MAX_QUEUE)
 423                 return -EINVAL;
 424
 425         mask = roundup_pow_of_two(ctl->limit + 1) - 1;
 426         if (mask != q->tab_mask) {
 427                 struct sk_buff **ntab;
 428
 429                 ntab = kcalloc(mask + 1, sizeof(struct sk_buff *),
 430                                GFP_KERNEL | __GFP_NOWARN);
 431                 if (!ntab)
 432                         ntab = vzalloc((mask + 1) * sizeof(struct sk_buff *));
 433                 if (!ntab)
 434                         return -ENOMEM;
 435
 436                 sch_tree_lock(sch);
 437                 old = q->tab;
 438                 if (old) {
 439                         unsigned int oqlen = sch->q.qlen, tail = 0;
 440
 441                         while (q->head != q->tail) {
 442                                 struct sk_buff *skb = q->tab[q->head];
 443
 444                                 q->head = (q->head + 1) & q->tab_mask;
 445                                 if (!skb)
 446                                         continue;
 447                                 if (tail < mask) {
 448                                         ntab[tail++] = skb;
 449                                         continue;
 450                                 }
 451                                 sch->qstats.backlog -= qdisc_pkt_len(skb);
 452                                 --sch->q.qlen;
 453                                 qdisc_drop(skb, sch);
 454                         }
 455                         qdisc_tree_decrease_qlen(sch, oqlen - sch->q.qlen);
 456                         q->head = 0;
 457                         q->tail = tail;
 458                 }
 459
 460                 q->tab_mask = mask;
 461                 q->tab = ntab;
 462         } else
 463                 sch_tree_lock(sch);
 464
 465         q->flags = ctl->flags;
 466         q->limit = ctl->limit;
 467
 468         red_set_parms(&q->parms, ctl->qth_min, ctl->qth_max, ctl->Wlog,
 469                       ctl->Plog, ctl->Scell_log,
 470                       nla_data(tb[TCA_CHOKE_STAB]),
 471                       max_P);
 472         red_set_vars(&q->vars);
 473
 474         if (q->head == q->tail)
 475                 red_end_of_idle_period(&q->vars);
 476
 477         sch_tree_unlock(sch);
 478         choke_free(old);
 479         return 0;
 480 }
 481
 482 static int choke_init(struct Qdisc *sch, struct nlattr *opt)
 483 {
 484         return choke_change(sch, opt);
 485 }
 486
 487 static int choke_dump(struct Qdisc *sch, struct sk_buff *skb)
 488 {
 489         struct choke_sched_data *q = qdisc_priv(sch);
 490         struct nlattr *opts = NULL;
 491         struct tc_red_qopt opt = {
 492                 .limit          = q->limit,
 493                 .flags          = q->flags,
 494                 .qth_min        = q->parms.qth_min >> q->parms.Wlog,
 495                 .qth_max        = q->parms.qth_max >> q->parms.Wlog,
 496                 .Wlog           = q->parms.Wlog,
 497                 .Plog           = q->parms.Plog,
 498                 .Scell_log      = q->parms.Scell_log,
 499         };
 500
 501         opts = nla_nest_start(skb, TCA_OPTIONS);
 502         if (opts == NULL)
 503                 goto nla_put_failure;
 504
 505         if (nla_put(skb, TCA_CHOKE_PARMS, sizeof(opt), &opt) ||
 506             nla_put_u32(skb, TCA_CHOKE_MAX_P, q->parms.max_P))
 507                 goto nla_put_failure;
 508         return nla_nest_end(skb, opts);
 509
 510 nla_put_failure:
 511         nla_nest_cancel(skb, opts);
 512         return -EMSGSIZE;
 513 }
 514
 515 static int choke_dump_stats(struct Qdisc *sch, struct gnet_dump *d)
 516 {
 517         struct choke_sched_data *q = qdisc_priv(sch);
 518         struct tc_choke_xstats st = {
 519                 .early  = q->stats.prob_drop + q->stats.forced_drop,
 520                 .marked = q->stats.prob_mark + q->stats.forced_mark,
 521                 .pdrop  = q->stats.pdrop,
 522                 .other  = q->stats.other,
 523                 .matched = q->stats.matched,
 524         };
 525
 526         return gnet_stats_copy_app(d, &st, sizeof(st));
 527 }
 528
 529 static void choke_destroy(struct Qdisc *sch)
 530 {
 531         struct choke_sched_data *q = qdisc_priv(sch);
 532
 533         tcf_destroy_chain(&q->filter_list);
 534         choke_free(q->tab);
 535 }
 536
 537 static struct Qdisc *choke_leaf(struct Qdisc *sch, unsigned long arg)
 538 {
 539         return NULL;
 540 }
 541
 542 static unsigned long choke_get(struct Qdisc *sch, u32 classid)
 543 {
 544         return 0;
 545 }
 546
 547 static void choke_put(struct Qdisc *q, unsigned long cl)
 548 {
 549 }
 550
 551 static unsigned long choke_bind(struct Qdisc *sch, unsigned long parent,
 552                                 u32 classid)
 553 {
 554         return 0;
 555 }
 556
 557 static struct tcf_proto **choke_find_tcf(struct Qdisc *sch, unsigned long cl)
 558 {
 559         struct choke_sched_data *q = qdisc_priv(sch);
 560
 561         if (cl)
 562                 return NULL;
 563         return &q->filter_list;
 564 }
 565
 566 static int choke_dump_class(struct Qdisc *sch, unsigned long cl,
 567                           struct sk_buff *skb, struct tcmsg *tcm)
 568 {
 569         tcm->tcm_handle |= TC_H_MIN(cl);
 570         return 0;
 571 }
 572
 573 static void choke_walk(struct Qdisc *sch, struct qdisc_walker *arg)
 574 {
 575         if (!arg->stop) {
 576                 if (arg->fn(sch, 1, arg) < 0) {
 577                         arg->stop = 1;
 578                         return;
 579                 }
 580                 arg->count++;
 581         }
 582 }
 583
 584 static const struct Qdisc_class_ops choke_class_ops = {
 585         .leaf           =       choke_leaf,
 586         .get            =       choke_get,
 587         .put            =       choke_put,
 588         .tcf_chain      =       choke_find_tcf,
 589         .bind_tcf       =       choke_bind,
 590         .unbind_tcf     =       choke_put,
 591         .dump           =       choke_dump_class,
 592         .walk           =       choke_walk,
 593 };
 594
 595 static struct sk_buff *choke_peek_head(struct Qdisc *sch)
 596 {
 597         struct choke_sched_data *q = qdisc_priv(sch);
 598
 599         return (q->head != q->tail) ? q->tab[q->head] : NULL;
 600 }
 601
 602 static struct Qdisc_ops choke_qdisc_ops __read_mostly = {
 603         .id             =       "choke",
 604         .priv_size      =       sizeof(struct choke_sched_data),
 605
 606         .enqueue        =       choke_enqueue,
 607         .dequeue        =       choke_dequeue,
 608         .peek           =       choke_peek_head,
 609         .drop           =       choke_drop,
 610         .init           =       choke_init,
 611         .destroy        =       choke_destroy,
 612         .reset          =       choke_reset,
 613         .change         =       choke_change,
 614         .dump           =       choke_dump,
 615         .dump_stats     =       choke_dump_stats,
 616         .owner          =       THIS_MODULE,
 617 };
 618
 619 static int __init choke_module_init(void)
 620 {
 621         return register_qdisc(&choke_qdisc_ops);
 622 }
 623
 624 static void __exit choke_module_exit(void)
 625 {
 626         unregister_qdisc(&choke_qdisc_ops);
 627 }
 628
 629 module_init(choke_module_init)
 630 module_exit(choke_module_exit)
 631
 632 MODULE_LICENSE("GPL");