net/sched/sch_fq_codel.c

   1 // SPDX-License-Identifier: GPL-2.0-or-later
   2 /*
   3  * Fair Queue CoDel discipline
   4  *
   5  *  Copyright (C) 2012,2015 Eric Dumazet <edumazet@google.com>
   6  */
   7
   8 #include <linux/module.h>
   9 #include <linux/types.h>
  10 #include <linux/kernel.h>
  11 #include <linux/jiffies.h>
  12 #include <linux/string.h>
  13 #include <linux/in.h>
  14 #include <linux/errno.h>
  15 #include <linux/init.h>
  16 #include <linux/skbuff.h>
  17 #include <linux/slab.h>
  18 #include <linux/vmalloc.h>
  19 #include <net/netlink.h>
  20 #include <net/pkt_sched.h>
  21 #include <net/pkt_cls.h>
  22 #include <net/codel.h>
  23 #include <net/codel_impl.h>
  24 #include <net/codel_qdisc.h>
  25
  26 /*      Fair Queue CoDel.
  27  *
  28  * Principles :
  29  * Packets are classified (internal classifier or external) on flows.
  30  * This is a Stochastic model (as we use a hash, several flows
  31  *                             might be hashed on same slot)
  32  * Each flow has a CoDel managed queue.
  33  * Flows are linked onto two (Round Robin) lists,
  34  * so that new flows have priority on old ones.
  35  *
  36  * For a given flow, packets are not reordered (CoDel uses a FIFO)
  37  * head drops only.
  38  * ECN capability is on by default.
  39  * Low memory footprint (64 bytes per flow)
  40  */
  41
  42 struct fq_codel_flow {
  43         struct sk_buff    *head;
  44         struct sk_buff    *tail;
  45         struct list_head  flowchain;
  46         int               deficit;
  47         struct codel_vars cvars;
  48 }; /* please try to keep this structure <= 64 bytes */
  49
  50 struct fq_codel_sched_data {
  51         struct tcf_proto __rcu *filter_list; /* optional external classifier */
  52         struct tcf_block *block;
  53         struct fq_codel_flow *flows;    /* Flows table [flows_cnt] */
  54         u32             *backlogs;      /* backlog table [flows_cnt] */
  55         u32             flows_cnt;      /* number of flows */
  56         u32             quantum;        /* psched_mtu(qdisc_dev(sch)); */
  57         u32             drop_batch_size;
  58         u32             memory_limit;
  59         struct codel_params cparams;
  60         struct codel_stats cstats;
  61         u32             memory_usage;
  62         u32             drop_overmemory;
  63         u32             drop_overlimit;
  64         u32             new_flow_count;
  65
  66         struct list_head new_flows;     /* list of new flows */
  67         struct list_head old_flows;     /* list of old flows */
  68 };
  69
  70 static unsigned int fq_codel_hash(const struct fq_codel_sched_data *q,
  71                                   struct sk_buff *skb)
  72 {
  73         return reciprocal_scale(skb_get_hash(skb), q->flows_cnt);
  74 }
  75
  76 static unsigned int fq_codel_classify(struct sk_buff *skb, struct Qdisc *sch,
  77                                       int *qerr)
  78 {
  79         struct fq_codel_sched_data *q = qdisc_priv(sch);
  80         struct tcf_proto *filter;
  81         struct tcf_result res;
  82         int result;
  83
  84         if (TC_H_MAJ(skb->priority) == sch->handle &&
  85             TC_H_MIN(skb->priority) > 0 &&
  86             TC_H_MIN(skb->priority) <= q->flows_cnt)
  87                 return TC_H_MIN(skb->priority);
  88
  89         filter = rcu_dereference_bh(q->filter_list);
  90         if (!filter)
  91                 return fq_codel_hash(q, skb) + 1;
  92
  93         *qerr = NET_XMIT_SUCCESS | __NET_XMIT_BYPASS;
  94         result = tcf_classify(skb, NULL, filter, &res, false);
  95         if (result >= 0) {
  96 #ifdef CONFIG_NET_CLS_ACT
  97                 switch (result) {
  98                 case TC_ACT_STOLEN:
  99                 case TC_ACT_QUEUED:
 100                 case TC_ACT_TRAP:
 101                         *qerr = NET_XMIT_SUCCESS | __NET_XMIT_STOLEN;
 102                         fallthrough;
 103                 case TC_ACT_SHOT:
 104                         return 0;
 105                 }
 106 #endif
 107                 if (TC_H_MIN(res.classid) <= q->flows_cnt)
 108                         return TC_H_MIN(res.classid);
 109         }
 110         return 0;
 111 }
 112
 113 /* helper functions : might be changed when/if skb use a standard list_head */
 114
 115 /* remove one skb from head of slot queue */
 116 static inline struct sk_buff *dequeue_head(struct fq_codel_flow *flow)
 117 {
 118         struct sk_buff *skb = flow->head;
 119
 120         flow->head = skb->next;
 121         skb_mark_not_on_list(skb);
 122         return skb;
 123 }
 124
 125 /* add skb to flow queue (tail add) */
 126 static inline void flow_queue_add(struct fq_codel_flow *flow,
 127                                   struct sk_buff *skb)
 128 {
 129         if (flow->head == NULL)
 130                 flow->head = skb;
 131         else
 132                 flow->tail->next = skb;
 133         flow->tail = skb;
 134         skb->next = NULL;
 135 }
 136
 137 static unsigned int fq_codel_drop(struct Qdisc *sch, unsigned int max_packets,
 138                                   struct sk_buff **to_free)
 139 {
 140         struct fq_codel_sched_data *q = qdisc_priv(sch);
 141         struct sk_buff *skb;
 142         unsigned int maxbacklog = 0, idx = 0, i, len;
 143         struct fq_codel_flow *flow;
 144         unsigned int threshold;
 145         unsigned int mem = 0;
 146
 147         /* Queue is full! Find the fat flow and drop packet(s) from it.
 148          * This might sound expensive, but with 1024 flows, we scan
 149          * 4KB of memory, and we dont need to handle a complex tree
 150          * in fast path (packet queue/enqueue) with many cache misses.
 151          * In stress mode, we'll try to drop 64 packets from the flow,
 152          * amortizing this linear lookup to one cache line per drop.
 153          */
 154         for (i = 0; i < q->flows_cnt; i++) {
 155                 if (q->backlogs[i] > maxbacklog) {
 156                         maxbacklog = q->backlogs[i];
 157                         idx = i;
 158                 }
 159         }
 160
 161         /* Our goal is to drop half of this fat flow backlog */
 162         threshold = maxbacklog >> 1;
 163
 164         flow = &q->flows[idx];
 165         len = 0;
 166         i = 0;
 167         do {
 168                 skb = dequeue_head(flow);
 169                 len += qdisc_pkt_len(skb);
 170                 mem += get_codel_cb(skb)->mem_usage;
 171                 __qdisc_drop(skb, to_free);
 172         } while (++i < max_packets && len < threshold);
 173
 174         /* Tell codel to increase its signal strength also */
 175         flow->cvars.count += i;
 176         q->backlogs[idx] -= len;
 177         q->memory_usage -= mem;
 178         sch->qstats.drops += i;
 179         sch->qstats.backlog -= len;
 180         sch->q.qlen -= i;
 181         return idx;
 182 }
 183
 184 static int fq_codel_enqueue(struct sk_buff *skb, struct Qdisc *sch,
 185                             struct sk_buff **to_free)
 186 {
 187         struct fq_codel_sched_data *q = qdisc_priv(sch);
 188         unsigned int idx, prev_backlog, prev_qlen;
 189         struct fq_codel_flow *flow;
 190         int ret;
 191         unsigned int pkt_len;
 192         bool memory_limited;
 193
 194         idx = fq_codel_classify(skb, sch, &ret);
 195         if (idx == 0) {
 196                 if (ret & __NET_XMIT_BYPASS)
 197                         qdisc_qstats_drop(sch);
 198                 __qdisc_drop(skb, to_free);
 199                 return ret;
 200         }
 201         idx--;
 202
 203         codel_set_enqueue_time(skb);
 204         flow = &q->flows[idx];
 205         flow_queue_add(flow, skb);
 206         q->backlogs[idx] += qdisc_pkt_len(skb);
 207         qdisc_qstats_backlog_inc(sch, skb);
 208
 209         if (list_empty(&flow->flowchain)) {
 210                 list_add_tail(&flow->flowchain, &q->new_flows);
 211                 q->new_flow_count++;
 212                 flow->deficit = q->quantum;
 213         }
 214         get_codel_cb(skb)->mem_usage = skb->truesize;
 215         q->memory_usage += get_codel_cb(skb)->mem_usage;
 216         memory_limited = q->memory_usage > q->memory_limit;
 217         if (++sch->q.qlen <= sch->limit && !memory_limited)
 218                 return NET_XMIT_SUCCESS;
 219
 220         prev_backlog = sch->qstats.backlog;
 221         prev_qlen = sch->q.qlen;
 222
 223         /* save this packet length as it might be dropped by fq_codel_drop() */
 224         pkt_len = qdisc_pkt_len(skb);
 225         /* fq_codel_drop() is quite expensive, as it performs a linear search
 226          * in q->backlogs[] to find a fat flow.
 227          * So instead of dropping a single packet, drop half of its backlog
 228          * with a 64 packets limit to not add a too big cpu spike here.
 229          */
 230         ret = fq_codel_drop(sch, q->drop_batch_size, to_free);
 231
 232         prev_qlen -= sch->q.qlen;
 233         prev_backlog -= sch->qstats.backlog;
 234         q->drop_overlimit += prev_qlen;
 235         if (memory_limited)
 236                 q->drop_overmemory += prev_qlen;
 237
 238         /* As we dropped packet(s), better let upper stack know this.
 239          * If we dropped a packet for this flow, return NET_XMIT_CN,
 240          * but in this case, our parents wont increase their backlogs.
 241          */
 242         if (ret == idx) {
 243                 qdisc_tree_reduce_backlog(sch, prev_qlen - 1,
 244                                           prev_backlog - pkt_len);
 245                 return NET_XMIT_CN;
 246         }
 247         qdisc_tree_reduce_backlog(sch, prev_qlen, prev_backlog);
 248         return NET_XMIT_SUCCESS;
 249 }
 250
 251 /* This is the specific function called from codel_dequeue()
 252  * to dequeue a packet from queue. Note: backlog is handled in
 253  * codel, we dont need to reduce it here.
 254  */
 255 static struct sk_buff *dequeue_func(struct codel_vars *vars, void *ctx)
 256 {
 257         struct Qdisc *sch = ctx;
 258         struct fq_codel_sched_data *q = qdisc_priv(sch);
 259         struct fq_codel_flow *flow;
 260         struct sk_buff *skb = NULL;
 261
 262         flow = container_of(vars, struct fq_codel_flow, cvars);
 263         if (flow->head) {
 264                 skb = dequeue_head(flow);
 265                 q->backlogs[flow - q->flows] -= qdisc_pkt_len(skb);
 266                 q->memory_usage -= get_codel_cb(skb)->mem_usage;
 267                 sch->q.qlen--;
 268                 sch->qstats.backlog -= qdisc_pkt_len(skb);
 269         }
 270         return skb;
 271 }
 272
 273 static void drop_func(struct sk_buff *skb, void *ctx)
 274 {
 275         struct Qdisc *sch = ctx;
 276
 277         kfree_skb(skb);
 278         qdisc_qstats_drop(sch);
 279 }
 280
 281 static struct sk_buff *fq_codel_dequeue(struct Qdisc *sch)
 282 {
 283         struct fq_codel_sched_data *q = qdisc_priv(sch);
 284         struct sk_buff *skb;
 285         struct fq_codel_flow *flow;
 286         struct list_head *head;
 287
 288 begin:
 289         head = &q->new_flows;
 290         if (list_empty(head)) {
 291                 head = &q->old_flows;
 292                 if (list_empty(head))
 293                         return NULL;
 294         }
 295         flow = list_first_entry(head, struct fq_codel_flow, flowchain);
 296
 297         if (flow->deficit <= 0) {
 298                 flow->deficit += q->quantum;
 299                 list_move_tail(&flow->flowchain, &q->old_flows);
 300                 goto begin;
 301         }
 302
 303         skb = codel_dequeue(sch, &sch->qstats.backlog, &q->cparams,
 304                             &flow->cvars, &q->cstats, qdisc_pkt_len,
 305                             codel_get_enqueue_time, drop_func, dequeue_func);
 306
 307         if (!skb) {
 308                 /* force a pass through old_flows to prevent starvation */
 309                 if ((head == &q->new_flows) && !list_empty(&q->old_flows))
 310                         list_move_tail(&flow->flowchain, &q->old_flows);
 311                 else
 312                         list_del_init(&flow->flowchain);
 313                 goto begin;
 314         }
 315         qdisc_bstats_update(sch, skb);
 316         flow->deficit -= qdisc_pkt_len(skb);
 317         /* We cant call qdisc_tree_reduce_backlog() if our qlen is 0,
 318          * or HTB crashes. Defer it for next round.
 319          */
 320         if (q->cstats.drop_count && sch->q.qlen) {
 321                 qdisc_tree_reduce_backlog(sch, q->cstats.drop_count,
 322                                           q->cstats.drop_len);
 323                 q->cstats.drop_count = 0;
 324                 q->cstats.drop_len = 0;
 325         }
 326         return skb;
 327 }
 328
 329 static void fq_codel_flow_purge(struct fq_codel_flow *flow)
 330 {
 331         rtnl_kfree_skbs(flow->head, flow->tail);
 332         flow->head = NULL;
 333 }
 334
 335 static void fq_codel_reset(struct Qdisc *sch)
 336 {
 337         struct fq_codel_sched_data *q = qdisc_priv(sch);
 338         int i;
 339
 340         INIT_LIST_HEAD(&q->new_flows);
 341         INIT_LIST_HEAD(&q->old_flows);
 342         for (i = 0; i < q->flows_cnt; i++) {
 343                 struct fq_codel_flow *flow = q->flows + i;
 344
 345                 fq_codel_flow_purge(flow);
 346                 INIT_LIST_HEAD(&flow->flowchain);
 347                 codel_vars_init(&flow->cvars);
 348         }
 349         memset(q->backlogs, 0, q->flows_cnt * sizeof(u32));
 350         sch->q.qlen = 0;
 351         sch->qstats.backlog = 0;
 352         q->memory_usage = 0;
 353 }
 354
 355 static const struct nla_policy fq_codel_policy[TCA_FQ_CODEL_MAX + 1] = {
 356         [TCA_FQ_CODEL_TARGET]   = { .type = NLA_U32 },
 357         [TCA_FQ_CODEL_LIMIT]    = { .type = NLA_U32 },
 358         [TCA_FQ_CODEL_INTERVAL] = { .type = NLA_U32 },
 359         [TCA_FQ_CODEL_ECN]      = { .type = NLA_U32 },
 360         [TCA_FQ_CODEL_FLOWS]    = { .type = NLA_U32 },
 361         [TCA_FQ_CODEL_QUANTUM]  = { .type = NLA_U32 },
 362         [TCA_FQ_CODEL_CE_THRESHOLD] = { .type = NLA_U32 },
 363         [TCA_FQ_CODEL_DROP_BATCH_SIZE] = { .type = NLA_U32 },
 364         [TCA_FQ_CODEL_MEMORY_LIMIT] = { .type = NLA_U32 },
 365 };
 366
 367 static int fq_codel_change(struct Qdisc *sch, struct nlattr *opt,
 368                            struct netlink_ext_ack *extack)
 369 {
 370         struct fq_codel_sched_data *q = qdisc_priv(sch);
 371         struct nlattr *tb[TCA_FQ_CODEL_MAX + 1];
 372         int err;
 373
 374         if (!opt)
 375                 return -EINVAL;
 376
 377         err = nla_parse_nested_deprecated(tb, TCA_FQ_CODEL_MAX, opt,
 378                                           fq_codel_policy, NULL);
 379         if (err < 0)
 380                 return err;
 381         if (tb[TCA_FQ_CODEL_FLOWS]) {
 382                 if (q->flows)
 383                         return -EINVAL;
 384                 q->flows_cnt = nla_get_u32(tb[TCA_FQ_CODEL_FLOWS]);
 385                 if (!q->flows_cnt ||
 386                     q->flows_cnt > 65536)
 387                         return -EINVAL;
 388         }
 389         sch_tree_lock(sch);
 390
 391         if (tb[TCA_FQ_CODEL_TARGET]) {
 392                 u64 target = nla_get_u32(tb[TCA_FQ_CODEL_TARGET]);
 393
 394                 q->cparams.target = (target * NSEC_PER_USEC) >> CODEL_SHIFT;
 395         }
 396
 397         if (tb[TCA_FQ_CODEL_CE_THRESHOLD]) {
 398                 u64 val = nla_get_u32(tb[TCA_FQ_CODEL_CE_THRESHOLD]);
 399
 400                 q->cparams.ce_threshold = (val * NSEC_PER_USEC) >> CODEL_SHIFT;
 401         }
 402
 403         if (tb[TCA_FQ_CODEL_INTERVAL]) {
 404                 u64 interval = nla_get_u32(tb[TCA_FQ_CODEL_INTERVAL]);
 405
 406                 q->cparams.interval = (interval * NSEC_PER_USEC) >> CODEL_SHIFT;
 407         }
 408
 409         if (tb[TCA_FQ_CODEL_LIMIT])
 410                 sch->limit = nla_get_u32(tb[TCA_FQ_CODEL_LIMIT]);
 411
 412         if (tb[TCA_FQ_CODEL_ECN])
 413                 q->cparams.ecn = !!nla_get_u32(tb[TCA_FQ_CODEL_ECN]);
 414
 415         if (tb[TCA_FQ_CODEL_QUANTUM])
 416                 q->quantum = max(256U, nla_get_u32(tb[TCA_FQ_CODEL_QUANTUM]));
 417
 418         if (tb[TCA_FQ_CODEL_DROP_BATCH_SIZE])
 419                 q->drop_batch_size = max(1U, nla_get_u32(tb[TCA_FQ_CODEL_DROP_BATCH_SIZE]));
 420
 421         if (tb[TCA_FQ_CODEL_MEMORY_LIMIT])
 422                 q->memory_limit = min(1U << 31, nla_get_u32(tb[TCA_FQ_CODEL_MEMORY_LIMIT]));
 423
 424         while (sch->q.qlen > sch->limit ||
 425                q->memory_usage > q->memory_limit) {
 426                 struct sk_buff *skb = fq_codel_dequeue(sch);
 427
 428                 q->cstats.drop_len += qdisc_pkt_len(skb);
 429                 rtnl_kfree_skbs(skb, skb);
 430                 q->cstats.drop_count++;
 431         }
 432         qdisc_tree_reduce_backlog(sch, q->cstats.drop_count, q->cstats.drop_len);
 433         q->cstats.drop_count = 0;
 434         q->cstats.drop_len = 0;
 435
 436         sch_tree_unlock(sch);
 437         return 0;
 438 }
 439
 440 static void fq_codel_destroy(struct Qdisc *sch)
 441 {
 442         struct fq_codel_sched_data *q = qdisc_priv(sch);
 443
 444         tcf_block_put(q->block);
 445         kvfree(q->backlogs);
 446         kvfree(q->flows);
 447 }
 448
 449 static int fq_codel_init(struct Qdisc *sch, struct nlattr *opt,
 450                          struct netlink_ext_ack *extack)
 451 {
 452         struct fq_codel_sched_data *q = qdisc_priv(sch);
 453         int i;
 454         int err;
 455
 456         sch->limit = 10*1024;
 457         q->flows_cnt = 1024;
 458         q->memory_limit = 32 << 20; /* 32 MBytes */
 459         q->drop_batch_size = 64;
 460         q->quantum = psched_mtu(qdisc_dev(sch));
 461         INIT_LIST_HEAD(&q->new_flows);
 462         INIT_LIST_HEAD(&q->old_flows);
 463         codel_params_init(&q->cparams);
 464         codel_stats_init(&q->cstats);
 465         q->cparams.ecn = true;
 466         q->cparams.mtu = psched_mtu(qdisc_dev(sch));
 467
 468         if (opt) {
 469                 err = fq_codel_change(sch, opt, extack);
 470                 if (err)
 471                         goto init_failure;
 472         }
 473
 474         err = tcf_block_get(&q->block, &q->filter_list, sch, extack);
 475         if (err)
 476                 goto init_failure;
 477
 478         if (!q->flows) {
 479                 q->flows = kvcalloc(q->flows_cnt,
 480                                     sizeof(struct fq_codel_flow),
 481                                     GFP_KERNEL);
 482                 if (!q->flows) {
 483                         err = -ENOMEM;
 484                         goto init_failure;
 485                 }
 486                 q->backlogs = kvcalloc(q->flows_cnt, sizeof(u32), GFP_KERNEL);
 487                 if (!q->backlogs) {
 488                         err = -ENOMEM;
 489                         goto alloc_failure;
 490                 }
 491                 for (i = 0; i < q->flows_cnt; i++) {
 492                         struct fq_codel_flow *flow = q->flows + i;
 493
 494                         INIT_LIST_HEAD(&flow->flowchain);
 495                         codel_vars_init(&flow->cvars);
 496                 }
 497         }
 498         if (sch->limit >= 1)
 499                 sch->flags |= TCQ_F_CAN_BYPASS;
 500         else
 501                 sch->flags &= ~TCQ_F_CAN_BYPASS;
 502         return 0;
 503
 504 alloc_failure:
 505         kvfree(q->flows);
 506         q->flows = NULL;
 507 init_failure:
 508         q->flows_cnt = 0;
 509         return err;
 510 }
 511
 512 static int fq_codel_dump(struct Qdisc *sch, struct sk_buff *skb)
 513 {
 514         struct fq_codel_sched_data *q = qdisc_priv(sch);
 515         struct nlattr *opts;
 516
 517         opts = nla_nest_start_noflag(skb, TCA_OPTIONS);
 518         if (opts == NULL)
 519                 goto nla_put_failure;
 520
 521         if (nla_put_u32(skb, TCA_FQ_CODEL_TARGET,
 522                         codel_time_to_us(q->cparams.target)) ||
 523             nla_put_u32(skb, TCA_FQ_CODEL_LIMIT,
 524                         sch->limit) ||
 525             nla_put_u32(skb, TCA_FQ_CODEL_INTERVAL,
 526                         codel_time_to_us(q->cparams.interval)) ||
 527             nla_put_u32(skb, TCA_FQ_CODEL_ECN,
 528                         q->cparams.ecn) ||
 529             nla_put_u32(skb, TCA_FQ_CODEL_QUANTUM,
 530                         q->quantum) ||
 531             nla_put_u32(skb, TCA_FQ_CODEL_DROP_BATCH_SIZE,
 532                         q->drop_batch_size) ||
 533             nla_put_u32(skb, TCA_FQ_CODEL_MEMORY_LIMIT,
 534                         q->memory_limit) ||
 535             nla_put_u32(skb, TCA_FQ_CODEL_FLOWS,
 536                         q->flows_cnt))
 537                 goto nla_put_failure;
 538
 539         if (q->cparams.ce_threshold != CODEL_DISABLED_THRESHOLD &&
 540             nla_put_u32(skb, TCA_FQ_CODEL_CE_THRESHOLD,
 541                         codel_time_to_us(q->cparams.ce_threshold)))
 542                 goto nla_put_failure;
 543
 544         return nla_nest_end(skb, opts);
 545
 546 nla_put_failure:
 547         return -1;
 548 }
 549
 550 static int fq_codel_dump_stats(struct Qdisc *sch, struct gnet_dump *d)
 551 {
 552         struct fq_codel_sched_data *q = qdisc_priv(sch);
 553         struct tc_fq_codel_xstats st = {
 554                 .type                           = TCA_FQ_CODEL_XSTATS_QDISC,
 555         };
 556         struct list_head *pos;
 557
 558         st.qdisc_stats.maxpacket = q->cstats.maxpacket;
 559         st.qdisc_stats.drop_overlimit = q->drop_overlimit;
 560         st.qdisc_stats.ecn_mark = q->cstats.ecn_mark;
 561         st.qdisc_stats.new_flow_count = q->new_flow_count;
 562         st.qdisc_stats.ce_mark = q->cstats.ce_mark;
 563         st.qdisc_stats.memory_usage  = q->memory_usage;
 564         st.qdisc_stats.drop_overmemory = q->drop_overmemory;
 565
 566         sch_tree_lock(sch);
 567         list_for_each(pos, &q->new_flows)
 568                 st.qdisc_stats.new_flows_len++;
 569
 570         list_for_each(pos, &q->old_flows)
 571                 st.qdisc_stats.old_flows_len++;
 572         sch_tree_unlock(sch);
 573
 574         return gnet_stats_copy_app(d, &st, sizeof(st));
 575 }
 576
 577 static struct Qdisc *fq_codel_leaf(struct Qdisc *sch, unsigned long arg)
 578 {
 579         return NULL;
 580 }
 581
 582 static unsigned long fq_codel_find(struct Qdisc *sch, u32 classid)
 583 {
 584         return 0;
 585 }
 586
 587 static unsigned long fq_codel_bind(struct Qdisc *sch, unsigned long parent,
 588                               u32 classid)
 589 {
 590         return 0;
 591 }
 592
 593 static void fq_codel_unbind(struct Qdisc *q, unsigned long cl)
 594 {
 595 }
 596
 597 static struct tcf_block *fq_codel_tcf_block(struct Qdisc *sch, unsigned long cl,
 598                                             struct netlink_ext_ack *extack)
 599 {
 600         struct fq_codel_sched_data *q = qdisc_priv(sch);
 601
 602         if (cl)
 603                 return NULL;
 604         return q->block;
 605 }
 606
 607 static int fq_codel_dump_class(struct Qdisc *sch, unsigned long cl,
 608                           struct sk_buff *skb, struct tcmsg *tcm)
 609 {
 610         tcm->tcm_handle |= TC_H_MIN(cl);
 611         return 0;
 612 }
 613
 614 static int fq_codel_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 615                                      struct gnet_dump *d)
 616 {
 617         struct fq_codel_sched_data *q = qdisc_priv(sch);
 618         u32 idx = cl - 1;
 619         struct gnet_stats_queue qs = { 0 };
 620         struct tc_fq_codel_xstats xstats;
 621
 622         if (idx < q->flows_cnt) {
 623                 const struct fq_codel_flow *flow = &q->flows[idx];
 624                 const struct sk_buff *skb;
 625
 626                 memset(&xstats, 0, sizeof(xstats));
 627                 xstats.type = TCA_FQ_CODEL_XSTATS_CLASS;
 628                 xstats.class_stats.deficit = flow->deficit;
 629                 xstats.class_stats.ldelay =
 630                         codel_time_to_us(flow->cvars.ldelay);
 631                 xstats.class_stats.count = flow->cvars.count;
 632                 xstats.class_stats.lastcount = flow->cvars.lastcount;
 633                 xstats.class_stats.dropping = flow->cvars.dropping;
 634                 if (flow->cvars.dropping) {
 635                         codel_tdiff_t delta = flow->cvars.drop_next -
 636                                               codel_get_time();
 637
 638                         xstats.class_stats.drop_next = (delta >= 0) ?
 639                                 codel_time_to_us(delta) :
 640                                 -codel_time_to_us(-delta);
 641                 }
 642                 if (flow->head) {
 643                         sch_tree_lock(sch);
 644                         skb = flow->head;
 645                         while (skb) {
 646                                 qs.qlen++;
 647                                 skb = skb->next;
 648                         }
 649                         sch_tree_unlock(sch);
 650                 }
 651                 qs.backlog = q->backlogs[idx];
 652                 qs.drops = 0;
 653         }
 654         if (gnet_stats_copy_queue(d, NULL, &qs, qs.qlen) < 0)
 655                 return -1;
 656         if (idx < q->flows_cnt)
 657                 return gnet_stats_copy_app(d, &xstats, sizeof(xstats));
 658         return 0;
 659 }
 660
 661 static void fq_codel_walk(struct Qdisc *sch, struct qdisc_walker *arg)
 662 {
 663         struct fq_codel_sched_data *q = qdisc_priv(sch);
 664         unsigned int i;
 665
 666         if (arg->stop)
 667                 return;
 668
 669         for (i = 0; i < q->flows_cnt; i++) {
 670                 if (list_empty(&q->flows[i].flowchain) ||
 671                     arg->count < arg->skip) {
 672                         arg->count++;
 673                         continue;
 674                 }
 675                 if (arg->fn(sch, i + 1, arg) < 0) {
 676                         arg->stop = 1;
 677                         break;
 678                 }
 679                 arg->count++;
 680         }
 681 }
 682
 683 static const struct Qdisc_class_ops fq_codel_class_ops = {
 684         .leaf           =       fq_codel_leaf,
 685         .find           =       fq_codel_find,
 686         .tcf_block      =       fq_codel_tcf_block,
 687         .bind_tcf       =       fq_codel_bind,
 688         .unbind_tcf     =       fq_codel_unbind,
 689         .dump           =       fq_codel_dump_class,
 690         .dump_stats     =       fq_codel_dump_class_stats,
 691         .walk           =       fq_codel_walk,
 692 };
 693
 694 static struct Qdisc_ops fq_codel_qdisc_ops __read_mostly = {
 695         .cl_ops         =       &fq_codel_class_ops,
 696         .id             =       "fq_codel",
 697         .priv_size      =       sizeof(struct fq_codel_sched_data),
 698         .enqueue        =       fq_codel_enqueue,
 699         .dequeue        =       fq_codel_dequeue,
 700         .peek           =       qdisc_peek_dequeued,
 701         .init           =       fq_codel_init,
 702         .reset          =       fq_codel_reset,
 703         .destroy        =       fq_codel_destroy,
 704         .change         =       fq_codel_change,
 705         .dump           =       fq_codel_dump,
 706         .dump_stats =   fq_codel_dump_stats,
 707         .owner          =       THIS_MODULE,
 708 };
 709
 710 static int __init fq_codel_module_init(void)
 711 {
 712         return register_qdisc(&fq_codel_qdisc_ops);
 713 }
 714
 715 static void __exit fq_codel_module_exit(void)
 716 {
 717         unregister_qdisc(&fq_codel_qdisc_ops);
 718 }
 719
 720 module_init(fq_codel_module_init)
 721 module_exit(fq_codel_module_exit)
 722 MODULE_AUTHOR("Eric Dumazet");
 723 MODULE_LICENSE("GPL");
 724 MODULE_DESCRIPTION("Fair Queue CoDel discipline");