net/sched/sch_red.c

   1 /*
   2  * net/sched/sch_red.c  Random Early Detection queue.
   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  * Changes:
  12  * J Hadi Salim 980914: computation fixes
  13  * Alexey Makarenko <makar@phoenix.kharkov.ua> 990814: qave on idle link was calculated incorrectly.
  14  * J Hadi Salim 980816:  ECN support
  15  */
  16
  17 #include <linux/module.h>
  18 #include <linux/types.h>
  19 #include <linux/kernel.h>
  20 #include <linux/skbuff.h>
  21 #include <net/pkt_sched.h>
  22 #include <net/inet_ecn.h>
  23 #include <net/red.h>
  24
  25
  26 /*      Parameters, settable by user:
  27         -----------------------------
  28
  29         limit           - bytes (must be > qth_max + burst)
  30
  31         Hard limit on queue length, should be chosen >qth_max
  32         to allow packet bursts. This parameter does not
  33         affect the algorithms behaviour and can be chosen
  34         arbitrarily high (well, less than ram size)
  35         Really, this limit will never be reached
  36         if RED works correctly.
  37  */
  38
  39 struct red_sched_data
  40 {
  41         u32                     limit;          /* HARD maximal queue length */
  42         unsigned char           flags;
  43         struct red_parms        parms;
  44         struct red_stats        stats;
  45         struct Qdisc            *qdisc;
  46 };
  47
  48 static inline int red_use_ecn(struct red_sched_data *q)
  49 {
  50         return q->flags & TC_RED_ECN;
  51 }
  52
  53 static inline int red_use_harddrop(struct red_sched_data *q)
  54 {
  55         return q->flags & TC_RED_HARDDROP;
  56 }
  57
  58 static int red_enqueue(struct sk_buff *skb, struct Qdisc* sch)
  59 {
  60         struct red_sched_data *q = qdisc_priv(sch);
  61         struct Qdisc *child = q->qdisc;
  62         int ret;
  63
  64         q->parms.qavg = red_calc_qavg(&q->parms, child->qstats.backlog);
  65
  66         if (red_is_idling(&q->parms))
  67                 red_end_of_idle_period(&q->parms);
  68
  69         switch (red_action(&q->parms, q->parms.qavg)) {
  70                 case RED_DONT_MARK:
  71                         break;
  72
  73                 case RED_PROB_MARK:
  74                         sch->qstats.overlimits++;
  75                         if (!red_use_ecn(q) || !INET_ECN_set_ce(skb)) {
  76                                 q->stats.prob_drop++;
  77                                 goto congestion_drop;
  78                         }
  79
  80                         q->stats.prob_mark++;
  81                         break;
  82
  83                 case RED_HARD_MARK:
  84                         sch->qstats.overlimits++;
  85                         if (red_use_harddrop(q) || !red_use_ecn(q) ||
  86                             !INET_ECN_set_ce(skb)) {
  87                                 q->stats.forced_drop++;
  88                                 goto congestion_drop;
  89                         }
  90
  91                         q->stats.forced_mark++;
  92                         break;
  93         }
  94
  95         ret = qdisc_enqueue(skb, child);
  96         if (likely(ret == NET_XMIT_SUCCESS)) {
  97                 sch->bstats.bytes += qdisc_pkt_len(skb);
  98                 sch->bstats.packets++;
  99                 sch->q.qlen++;
 100         } else if (net_xmit_drop_count(ret)) {
 101                 q->stats.pdrop++;
 102                 sch->qstats.drops++;
 103         }
 104         return ret;
 105
 106 congestion_drop:
 107         qdisc_drop(skb, sch);
 108         return NET_XMIT_CN;
 109 }
 110
 111 static struct sk_buff * red_dequeue(struct Qdisc* sch)
 112 {
 113         struct sk_buff *skb;
 114         struct red_sched_data *q = qdisc_priv(sch);
 115         struct Qdisc *child = q->qdisc;
 116
 117         skb = child->dequeue(child);
 118         if (skb)
 119                 sch->q.qlen--;
 120         else if (!red_is_idling(&q->parms))
 121                 red_start_of_idle_period(&q->parms);
 122
 123         return skb;
 124 }
 125
 126 static struct sk_buff * red_peek(struct Qdisc* sch)
 127 {
 128         struct red_sched_data *q = qdisc_priv(sch);
 129         struct Qdisc *child = q->qdisc;
 130
 131         return child->ops->peek(child);
 132 }
 133
 134 static unsigned int red_drop(struct Qdisc* sch)
 135 {
 136         struct red_sched_data *q = qdisc_priv(sch);
 137         struct Qdisc *child = q->qdisc;
 138         unsigned int len;
 139
 140         if (child->ops->drop && (len = child->ops->drop(child)) > 0) {
 141                 q->stats.other++;
 142                 sch->qstats.drops++;
 143                 sch->q.qlen--;
 144                 return len;
 145         }
 146
 147         if (!red_is_idling(&q->parms))
 148                 red_start_of_idle_period(&q->parms);
 149
 150         return 0;
 151 }
 152
 153 static void red_reset(struct Qdisc* sch)
 154 {
 155         struct red_sched_data *q = qdisc_priv(sch);
 156
 157         qdisc_reset(q->qdisc);
 158         sch->q.qlen = 0;
 159         red_restart(&q->parms);
 160 }
 161
 162 static void red_destroy(struct Qdisc *sch)
 163 {
 164         struct red_sched_data *q = qdisc_priv(sch);
 165         qdisc_destroy(q->qdisc);
 166 }
 167
 168 static const struct nla_policy red_policy[TCA_RED_MAX + 1] = {
 169         [TCA_RED_PARMS] = { .len = sizeof(struct tc_red_qopt) },
 170         [TCA_RED_STAB]  = { .len = RED_STAB_SIZE },
 171 };
 172
 173 static int red_change(struct Qdisc *sch, struct nlattr *opt)
 174 {
 175         struct red_sched_data *q = qdisc_priv(sch);
 176         struct nlattr *tb[TCA_RED_MAX + 1];
 177         struct tc_red_qopt *ctl;
 178         struct Qdisc *child = NULL;
 179         int err;
 180
 181         if (opt == NULL)
 182                 return -EINVAL;
 183
 184         err = nla_parse_nested(tb, TCA_RED_MAX, opt, red_policy);
 185         if (err < 0)
 186                 return err;
 187
 188         if (tb[TCA_RED_PARMS] == NULL ||
 189             tb[TCA_RED_STAB] == NULL)
 190                 return -EINVAL;
 191
 192         ctl = nla_data(tb[TCA_RED_PARMS]);
 193
 194         if (ctl->limit > 0) {
 195                 child = fifo_create_dflt(sch, &bfifo_qdisc_ops, ctl->limit);
 196                 if (IS_ERR(child))
 197                         return PTR_ERR(child);
 198         }
 199
 200         sch_tree_lock(sch);
 201         q->flags = ctl->flags;
 202         q->limit = ctl->limit;
 203         if (child) {
 204                 qdisc_tree_decrease_qlen(q->qdisc, q->qdisc->q.qlen);
 205                 qdisc_destroy(q->qdisc);
 206                 q->qdisc = child;
 207         }
 208
 209         red_set_parms(&q->parms, ctl->qth_min, ctl->qth_max, ctl->Wlog,
 210                                  ctl->Plog, ctl->Scell_log,
 211                                  nla_data(tb[TCA_RED_STAB]));
 212
 213         if (skb_queue_empty(&sch->q))
 214                 red_end_of_idle_period(&q->parms);
 215
 216         sch_tree_unlock(sch);
 217         return 0;
 218 }
 219
 220 static int red_init(struct Qdisc* sch, struct nlattr *opt)
 221 {
 222         struct red_sched_data *q = qdisc_priv(sch);
 223
 224         q->qdisc = &noop_qdisc;
 225         return red_change(sch, opt);
 226 }
 227
 228 static int red_dump(struct Qdisc *sch, struct sk_buff *skb)
 229 {
 230         struct red_sched_data *q = qdisc_priv(sch);
 231         struct nlattr *opts = NULL;
 232         struct tc_red_qopt opt = {
 233                 .limit          = q->limit,
 234                 .flags          = q->flags,
 235                 .qth_min        = q->parms.qth_min >> q->parms.Wlog,
 236                 .qth_max        = q->parms.qth_max >> q->parms.Wlog,
 237                 .Wlog           = q->parms.Wlog,
 238                 .Plog           = q->parms.Plog,
 239                 .Scell_log      = q->parms.Scell_log,
 240         };
 241
 242         opts = nla_nest_start(skb, TCA_OPTIONS);
 243         if (opts == NULL)
 244                 goto nla_put_failure;
 245         NLA_PUT(skb, TCA_RED_PARMS, sizeof(opt), &opt);
 246         return nla_nest_end(skb, opts);
 247
 248 nla_put_failure:
 249         nla_nest_cancel(skb, opts);
 250         return -EMSGSIZE;
 251 }
 252
 253 static int red_dump_stats(struct Qdisc *sch, struct gnet_dump *d)
 254 {
 255         struct red_sched_data *q = qdisc_priv(sch);
 256         struct tc_red_xstats st = {
 257                 .early  = q->stats.prob_drop + q->stats.forced_drop,
 258                 .pdrop  = q->stats.pdrop,
 259                 .other  = q->stats.other,
 260                 .marked = q->stats.prob_mark + q->stats.forced_mark,
 261         };
 262
 263         return gnet_stats_copy_app(d, &st, sizeof(st));
 264 }
 265
 266 static int red_dump_class(struct Qdisc *sch, unsigned long cl,
 267                           struct sk_buff *skb, struct tcmsg *tcm)
 268 {
 269         struct red_sched_data *q = qdisc_priv(sch);
 270
 271         tcm->tcm_handle |= TC_H_MIN(1);
 272         tcm->tcm_info = q->qdisc->handle;
 273         return 0;
 274 }
 275
 276 static int red_graft(struct Qdisc *sch, unsigned long arg, struct Qdisc *new,
 277                      struct Qdisc **old)
 278 {
 279         struct red_sched_data *q = qdisc_priv(sch);
 280
 281         if (new == NULL)
 282                 new = &noop_qdisc;
 283
 284         sch_tree_lock(sch);
 285         *old = q->qdisc;
 286         q->qdisc = new;
 287         qdisc_tree_decrease_qlen(*old, (*old)->q.qlen);
 288         qdisc_reset(*old);
 289         sch_tree_unlock(sch);
 290         return 0;
 291 }
 292
 293 static struct Qdisc *red_leaf(struct Qdisc *sch, unsigned long arg)
 294 {
 295         struct red_sched_data *q = qdisc_priv(sch);
 296         return q->qdisc;
 297 }
 298
 299 static unsigned long red_get(struct Qdisc *sch, u32 classid)
 300 {
 301         return 1;
 302 }
 303
 304 static void red_put(struct Qdisc *sch, unsigned long arg)
 305 {
 306 }
 307
 308 static void red_walk(struct Qdisc *sch, struct qdisc_walker *walker)
 309 {
 310         if (!walker->stop) {
 311                 if (walker->count >= walker->skip)
 312                         if (walker->fn(sch, 1, walker) < 0) {
 313                                 walker->stop = 1;
 314                                 return;
 315                         }
 316                 walker->count++;
 317         }
 318 }
 319
 320 static const struct Qdisc_class_ops red_class_ops = {
 321         .graft          =       red_graft,
 322         .leaf           =       red_leaf,
 323         .get            =       red_get,
 324         .put            =       red_put,
 325         .walk           =       red_walk,
 326         .dump           =       red_dump_class,
 327 };
 328
 329 static struct Qdisc_ops red_qdisc_ops __read_mostly = {
 330         .id             =       "red",
 331         .priv_size      =       sizeof(struct red_sched_data),
 332         .cl_ops         =       &red_class_ops,
 333         .enqueue        =       red_enqueue,
 334         .dequeue        =       red_dequeue,
 335         .peek           =       red_peek,
 336         .drop           =       red_drop,
 337         .init           =       red_init,
 338         .reset          =       red_reset,
 339         .destroy        =       red_destroy,
 340         .change         =       red_change,
 341         .dump           =       red_dump,
 342         .dump_stats     =       red_dump_stats,
 343         .owner          =       THIS_MODULE,
 344 };
 345
 346 static int __init red_module_init(void)
 347 {
 348         return register_qdisc(&red_qdisc_ops);
 349 }
 350
 351 static void __exit red_module_exit(void)
 352 {
 353         unregister_qdisc(&red_qdisc_ops);
 354 }
 355
 356 module_init(red_module_init)
 357 module_exit(red_module_exit)
 358
 359 MODULE_LICENSE("GPL");