net/sched/sch_gred.c

   1 /*
   2  * net/sched/sch_gred.c Generic Random Early Detection queue.
   3  *
   4  *
   5  *              This program is free software; you can redistribute it and/or
   6  *              modify it under the terms of the GNU General Public License
   7  *              as published by the Free Software Foundation; either version
   8  *              2 of the License, or (at your option) any later version.
   9  *
  10  * Authors:    J Hadi Salim (hadi@cyberus.ca) 1998-2002
  11  *
  12  *             991129: -  Bug fix with grio mode
  13  *                     - a better sing. AvgQ mode with Grio(WRED)
  14  *                     - A finer grained VQ dequeue based on sugestion
  15  *                       from Ren Liu
  16  *                     - More error checks
  17  *
  18  *
  19  *
  20  *  For all the glorious comments look at Alexey's sch_red.c
  21  */
  22
  23 #include <linux/config.h>
  24 #include <linux/module.h>
  25 #include <asm/uaccess.h>
  26 #include <asm/system.h>
  27 #include <linux/bitops.h>
  28 #include <linux/types.h>
  29 #include <linux/kernel.h>
  30 #include <linux/sched.h>
  31 #include <linux/string.h>
  32 #include <linux/mm.h>
  33 #include <linux/socket.h>
  34 #include <linux/sockios.h>
  35 #include <linux/in.h>
  36 #include <linux/errno.h>
  37 #include <linux/interrupt.h>
  38 #include <linux/if_ether.h>
  39 #include <linux/inet.h>
  40 #include <linux/netdevice.h>
  41 #include <linux/etherdevice.h>
  42 #include <linux/notifier.h>
  43 #include <net/ip.h>
  44 #include <net/route.h>
  45 #include <linux/skbuff.h>
  46 #include <net/sock.h>
  47 #include <net/pkt_sched.h>
  48 #include <net/red.h>
  49
  50 #if 1 /* control */
  51 #define DPRINTK(format,args...) printk(KERN_DEBUG format,##args)
  52 #else
  53 #define DPRINTK(format,args...)
  54 #endif
  55
  56 #if 0 /* data */
  57 #define D2PRINTK(format,args...) printk(KERN_DEBUG format,##args)
  58 #else
  59 #define D2PRINTK(format,args...)
  60 #endif
  61
  62 #define GRED_DEF_PRIO (MAX_DPs / 2)
  63
  64 struct gred_sched_data;
  65 struct gred_sched;
  66
  67 struct gred_sched_data
  68 {
  69         u32             limit;          /* HARD maximal queue length    */
  70         u32             DP;             /* the drop pramaters */
  71         u32             bytesin;        /* bytes seen on virtualQ so far*/
  72         u32             packetsin;      /* packets seen on virtualQ so far*/
  73         u32             backlog;        /* bytes on the virtualQ */
  74         u8              prio;        /* the prio of this vq */
  75
  76         struct red_parms parms;
  77         struct red_stats stats;
  78 };
  79
  80 enum {
  81         GRED_WRED_MODE = 1,
  82         GRED_RIO_MODE,
  83 };
  84
  85 struct gred_sched
  86 {
  87         struct gred_sched_data *tab[MAX_DPs];
  88         unsigned long   flags;
  89         u32             DPs;
  90         u32             def;
  91         u8              initd;
  92 };
  93
  94 static inline int gred_wred_mode(struct gred_sched *table)
  95 {
  96         return test_bit(GRED_WRED_MODE, &table->flags);
  97 }
  98
  99 static inline void gred_enable_wred_mode(struct gred_sched *table)
 100 {
 101         __set_bit(GRED_WRED_MODE, &table->flags);
 102 }
 103
 104 static inline void gred_disable_wred_mode(struct gred_sched *table)
 105 {
 106         __clear_bit(GRED_WRED_MODE, &table->flags);
 107 }
 108
 109 static inline int gred_rio_mode(struct gred_sched *table)
 110 {
 111         return test_bit(GRED_RIO_MODE, &table->flags);
 112 }
 113
 114 static inline void gred_enable_rio_mode(struct gred_sched *table)
 115 {
 116         __set_bit(GRED_RIO_MODE, &table->flags);
 117 }
 118
 119 static inline void gred_disable_rio_mode(struct gred_sched *table)
 120 {
 121         __clear_bit(GRED_RIO_MODE, &table->flags);
 122 }
 123
 124 static inline int gred_wred_mode_check(struct Qdisc *sch)
 125 {
 126         struct gred_sched *table = qdisc_priv(sch);
 127         int i;
 128
 129         /* Really ugly O(n^2) but shouldn't be necessary too frequent. */
 130         for (i = 0; i < table->DPs; i++) {
 131                 struct gred_sched_data *q = table->tab[i];
 132                 int n;
 133
 134                 if (q == NULL)
 135                         continue;
 136
 137                 for (n = 0; n < table->DPs; n++)
 138                         if (table->tab[n] && table->tab[n] != q &&
 139                             table->tab[n]->prio == q->prio)
 140                                 return 1;
 141         }
 142
 143         return 0;
 144 }
 145
 146 static inline unsigned int gred_backlog(struct gred_sched *table,
 147                                         struct gred_sched_data *q,
 148                                         struct Qdisc *sch)
 149 {
 150         if (gred_wred_mode(table))
 151                 return sch->qstats.backlog;
 152         else
 153                 return q->backlog;
 154 }
 155
 156 static int
 157 gred_enqueue(struct sk_buff *skb, struct Qdisc* sch)
 158 {
 159         struct gred_sched_data *q=NULL;
 160         struct gred_sched *t= qdisc_priv(sch);
 161         unsigned long qavg = 0;
 162         int i=0;
 163
 164         if (!t->initd && skb_queue_len(&sch->q) < (sch->dev->tx_queue_len ? : 1)) {
 165                 D2PRINTK("NO GRED Queues setup yet! Enqueued anyway\n");
 166                 goto do_enqueue;
 167         }
 168
 169
 170         if ( ((skb->tc_index&0xf) > (t->DPs -1)) || !(q=t->tab[skb->tc_index&0xf])) {
 171                 printk("GRED: setting to default (%d)\n ",t->def);
 172                 if (!(q=t->tab[t->def])) {
 173                         DPRINTK("GRED: setting to default FAILED! dropping!! "
 174                             "(%d)\n ", t->def);
 175                         goto drop;
 176                 }
 177                 /* fix tc_index? --could be controvesial but needed for
 178                    requeueing */
 179                 skb->tc_index=(skb->tc_index&0xfffffff0) | t->def;
 180         }
 181
 182         D2PRINTK("gred_enqueue virtualQ 0x%x classid %x backlog %d "
 183             "general backlog %d\n",skb->tc_index&0xf,sch->handle,q->backlog,
 184             sch->qstats.backlog);
 185         /* sum up all the qaves of prios <= to ours to get the new qave*/
 186         if (!gred_wred_mode(t) && gred_rio_mode(t)) {
 187                 for (i=0;i<t->DPs;i++) {
 188                         if ((!t->tab[i]) || (i==q->DP))
 189                                 continue;
 190
 191                         if (t->tab[i]->prio < q->prio &&
 192                             !red_is_idling(&t->tab[i]->parms))
 193                                 qavg +=t->tab[i]->parms.qavg;
 194                 }
 195
 196         }
 197
 198         q->packetsin++;
 199         q->bytesin+=skb->len;
 200
 201         if (gred_wred_mode(t)) {
 202                 qavg = 0;
 203                 q->parms.qavg = t->tab[t->def]->parms.qavg;
 204                 q->parms.qidlestart = t->tab[t->def]->parms.qidlestart;
 205         }
 206
 207         q->parms.qavg = red_calc_qavg(&q->parms, gred_backlog(t, q, sch));
 208
 209         if (red_is_idling(&q->parms))
 210                 red_end_of_idle_period(&q->parms);
 211
 212         if (gred_wred_mode(t))
 213                 t->tab[t->def]->parms.qavg = q->parms.qavg;
 214
 215         switch (red_action(&q->parms, q->parms.qavg + qavg)) {
 216                 case RED_DONT_MARK:
 217                         break;
 218
 219                 case RED_PROB_MARK:
 220                         sch->qstats.overlimits++;
 221                         q->stats.prob_drop++;
 222                         goto drop;
 223
 224                 case RED_HARD_MARK:
 225                         sch->qstats.overlimits++;
 226                         q->stats.forced_drop++;
 227                         goto drop;
 228         }
 229
 230         if (q->backlog + skb->len <= q->limit) {
 231                 q->backlog += skb->len;
 232 do_enqueue:
 233                 __skb_queue_tail(&sch->q, skb);
 234                 sch->qstats.backlog += skb->len;
 235                 sch->bstats.bytes += skb->len;
 236                 sch->bstats.packets++;
 237                 return 0;
 238         }
 239
 240         q->stats.pdrop++;
 241 drop:
 242         kfree_skb(skb);
 243         sch->qstats.drops++;
 244         return NET_XMIT_DROP;
 245 }
 246
 247 static int
 248 gred_requeue(struct sk_buff *skb, struct Qdisc* sch)
 249 {
 250         struct gred_sched_data *q;
 251         struct gred_sched *t= qdisc_priv(sch);
 252         q= t->tab[(skb->tc_index&0xf)];
 253 /* error checking here -- probably unnecessary */
 254
 255         if (red_is_idling(&q->parms))
 256                 red_end_of_idle_period(&q->parms);
 257
 258         __skb_queue_head(&sch->q, skb);
 259         sch->qstats.backlog += skb->len;
 260         sch->qstats.requeues++;
 261         q->backlog += skb->len;
 262         return 0;
 263 }
 264
 265 static struct sk_buff *
 266 gred_dequeue(struct Qdisc* sch)
 267 {
 268         struct sk_buff *skb;
 269         struct gred_sched_data *q;
 270         struct gred_sched *t= qdisc_priv(sch);
 271
 272         skb = __skb_dequeue(&sch->q);
 273         if (skb) {
 274                 sch->qstats.backlog -= skb->len;
 275                 q= t->tab[(skb->tc_index&0xf)];
 276                 if (q) {
 277                         q->backlog -= skb->len;
 278                         if (!q->backlog && !gred_wred_mode(t))
 279                                 red_start_of_idle_period(&q->parms);
 280                 } else {
 281                         D2PRINTK("gred_dequeue: skb has bad tcindex %x\n",skb->tc_index&0xf);
 282                 }
 283                 return skb;
 284         }
 285
 286         if (gred_wred_mode(t)) {
 287                         q= t->tab[t->def];
 288                         if (!q)
 289                                 D2PRINTK("no default VQ set: Results will be "
 290                                        "screwed up\n");
 291                         else
 292                                 red_start_of_idle_period(&q->parms);
 293         }
 294
 295         return NULL;
 296 }
 297
 298 static unsigned int gred_drop(struct Qdisc* sch)
 299 {
 300         struct sk_buff *skb;
 301
 302         struct gred_sched_data *q;
 303         struct gred_sched *t= qdisc_priv(sch);
 304
 305         skb = __skb_dequeue_tail(&sch->q);
 306         if (skb) {
 307                 unsigned int len = skb->len;
 308                 sch->qstats.backlog -= len;
 309                 sch->qstats.drops++;
 310                 q= t->tab[(skb->tc_index&0xf)];
 311                 if (q) {
 312                         q->backlog -= len;
 313                         q->stats.other++;
 314                         if (!q->backlog && !gred_wred_mode(t))
 315                                 red_start_of_idle_period(&q->parms);
 316                 } else {
 317                         D2PRINTK("gred_dequeue: skb has bad tcindex %x\n",skb->tc_index&0xf);
 318                 }
 319
 320                 kfree_skb(skb);
 321                 return len;
 322         }
 323
 324         q=t->tab[t->def];
 325         if (!q) {
 326                 D2PRINTK("no default VQ set: Results might be screwed up\n");
 327                 return 0;
 328         }
 329
 330         red_start_of_idle_period(&q->parms);
 331         return 0;
 332
 333 }
 334
 335 static void gred_reset(struct Qdisc* sch)
 336 {
 337         int i;
 338         struct gred_sched_data *q;
 339         struct gred_sched *t= qdisc_priv(sch);
 340
 341         __skb_queue_purge(&sch->q);
 342
 343         sch->qstats.backlog = 0;
 344
 345         for (i=0;i<t->DPs;i++) {
 346                 q= t->tab[i];
 347                 if (!q)
 348                         continue;
 349                 red_restart(&q->parms);
 350                 q->backlog = 0;
 351         }
 352 }
 353
 354 static inline void gred_destroy_vq(struct gred_sched_data *q)
 355 {
 356         kfree(q);
 357 }
 358
 359 static inline int gred_change_table_def(struct Qdisc *sch, struct rtattr *dps)
 360 {
 361         struct gred_sched *table = qdisc_priv(sch);
 362         struct tc_gred_sopt *sopt;
 363         int i;
 364
 365         if (dps == NULL || RTA_PAYLOAD(dps) < sizeof(*sopt))
 366                 return -EINVAL;
 367
 368         sopt = RTA_DATA(dps);
 369
 370         if (sopt->DPs > MAX_DPs || sopt->DPs == 0 || sopt->def_DP >= sopt->DPs)
 371                 return -EINVAL;
 372
 373         sch_tree_lock(sch);
 374         table->DPs = sopt->DPs;
 375         table->def = sopt->def_DP;
 376
 377         /*
 378          * Every entry point to GRED is synchronized with the above code
 379          * and the DP is checked against DPs, i.e. shadowed VQs can no
 380          * longer be found so we can unlock right here.
 381          */
 382         sch_tree_unlock(sch);
 383
 384         if (sopt->grio) {
 385                 gred_enable_rio_mode(table);
 386                 gred_disable_wred_mode(table);
 387                 if (gred_wred_mode_check(sch))
 388                         gred_enable_wred_mode(table);
 389         } else {
 390                 gred_disable_rio_mode(table);
 391                 gred_disable_wred_mode(table);
 392         }
 393
 394         for (i = table->DPs; i < MAX_DPs; i++) {
 395                 if (table->tab[i]) {
 396                         printk(KERN_WARNING "GRED: Warning: Destroying "
 397                                "shadowed VQ 0x%x\n", i);
 398                         gred_destroy_vq(table->tab[i]);
 399                         table->tab[i] = NULL;
 400                 }
 401         }
 402
 403         table->initd = 0;
 404
 405         return 0;
 406 }
 407
 408 static inline int gred_change_vq(struct Qdisc *sch, int dp,
 409                                  struct tc_gred_qopt *ctl, int prio, u8 *stab)
 410 {
 411         struct gred_sched *table = qdisc_priv(sch);
 412         struct gred_sched_data *q;
 413
 414         if (table->tab[dp] == NULL) {
 415                 table->tab[dp] = kmalloc(sizeof(*q), GFP_KERNEL);
 416                 if (table->tab[dp] == NULL)
 417                         return -ENOMEM;
 418                 memset(table->tab[dp], 0, sizeof(*q));
 419         }
 420
 421         q = table->tab[dp];
 422         q->DP = dp;
 423         q->prio = prio;
 424         q->limit = ctl->limit;
 425
 426         if (q->backlog == 0)
 427                 red_end_of_idle_period(&q->parms);
 428
 429         red_set_parms(&q->parms,
 430                       ctl->qth_min, ctl->qth_max, ctl->Wlog, ctl->Plog,
 431                       ctl->Scell_log, stab);
 432
 433         return 0;
 434 }
 435
 436 static int gred_change(struct Qdisc *sch, struct rtattr *opt)
 437 {
 438         struct gred_sched *table = qdisc_priv(sch);
 439         struct tc_gred_qopt *ctl;
 440         struct rtattr *tb[TCA_GRED_MAX];
 441         int err = -EINVAL, prio = GRED_DEF_PRIO;
 442         u8 *stab;
 443
 444         if (opt == NULL || rtattr_parse_nested(tb, TCA_GRED_MAX, opt))
 445                 return -EINVAL;
 446
 447         if (tb[TCA_GRED_PARMS-1] == NULL && tb[TCA_GRED_STAB-1] == NULL)
 448                 return gred_change_table_def(sch, opt);
 449
 450         if (tb[TCA_GRED_PARMS-1] == NULL ||
 451             RTA_PAYLOAD(tb[TCA_GRED_PARMS-1]) < sizeof(*ctl) ||
 452             tb[TCA_GRED_STAB-1] == NULL ||
 453             RTA_PAYLOAD(tb[TCA_GRED_STAB-1]) < 256)
 454                 return -EINVAL;
 455
 456         ctl = RTA_DATA(tb[TCA_GRED_PARMS-1]);
 457         stab = RTA_DATA(tb[TCA_GRED_STAB-1]);
 458
 459         if (ctl->DP >= table->DPs)
 460                 goto errout;
 461
 462         if (gred_rio_mode(table)) {
 463                 if (ctl->prio == 0) {
 464                         int def_prio = GRED_DEF_PRIO;
 465
 466                         if (table->tab[table->def])
 467                                 def_prio = table->tab[table->def]->prio;
 468
 469                         printk(KERN_DEBUG "GRED: DP %u does not have a prio "
 470                                "setting default to %d\n", ctl->DP, def_prio);
 471
 472                         prio = def_prio;
 473                 } else
 474                         prio = ctl->prio;
 475         }
 476
 477         sch_tree_lock(sch);
 478
 479         err = gred_change_vq(sch, ctl->DP, ctl, prio, stab);
 480         if (err < 0)
 481                 goto errout_locked;
 482
 483         if (table->tab[table->def] == NULL) {
 484                 if (gred_rio_mode(table))
 485                         prio = table->tab[ctl->DP]->prio;
 486
 487                 err = gred_change_vq(sch, table->def, ctl, prio, stab);
 488                 if (err < 0)
 489                         goto errout_locked;
 490         }
 491
 492         table->initd = 1;
 493
 494         if (gred_rio_mode(table)) {
 495                 gred_disable_wred_mode(table);
 496                 if (gred_wred_mode_check(sch))
 497                         gred_enable_wred_mode(table);
 498         }
 499
 500         err = 0;
 501
 502 errout_locked:
 503         sch_tree_unlock(sch);
 504 errout:
 505         return err;
 506 }
 507
 508 static int gred_init(struct Qdisc *sch, struct rtattr *opt)
 509 {
 510         struct rtattr *tb[TCA_GRED_MAX];
 511
 512         if (opt == NULL || rtattr_parse_nested(tb, TCA_GRED_MAX, opt))
 513                 return -EINVAL;
 514
 515         if (tb[TCA_GRED_PARMS-1] || tb[TCA_GRED_STAB-1])
 516                 return -EINVAL;
 517
 518         return gred_change_table_def(sch, tb[TCA_GRED_DPS-1]);
 519 }
 520
 521 static int gred_dump(struct Qdisc *sch, struct sk_buff *skb)
 522 {
 523         struct gred_sched *table = qdisc_priv(sch);
 524         struct rtattr *parms, *opts = NULL;
 525         int i;
 526         struct tc_gred_sopt sopt = {
 527                 .DPs    = table->DPs,
 528                 .def_DP = table->def,
 529                 .grio   = gred_rio_mode(table),
 530         };
 531
 532         opts = RTA_NEST(skb, TCA_OPTIONS);
 533         RTA_PUT(skb, TCA_GRED_DPS, sizeof(sopt), &sopt);
 534         parms = RTA_NEST(skb, TCA_GRED_PARMS);
 535
 536         for (i = 0; i < MAX_DPs; i++) {
 537                 struct gred_sched_data *q = table->tab[i];
 538                 struct tc_gred_qopt opt;
 539
 540                 memset(&opt, 0, sizeof(opt));
 541
 542                 if (!q) {
 543                         /* hack -- fix at some point with proper message
 544                            This is how we indicate to tc that there is no VQ
 545                            at this DP */
 546
 547                         opt.DP = MAX_DPs + i;
 548                         goto append_opt;
 549                 }
 550
 551                 opt.limit       = q->limit;
 552                 opt.DP          = q->DP;
 553                 opt.backlog     = q->backlog;
 554                 opt.prio        = q->prio;
 555                 opt.qth_min     = q->parms.qth_min >> q->parms.Wlog;
 556                 opt.qth_max     = q->parms.qth_max >> q->parms.Wlog;
 557                 opt.Wlog        = q->parms.Wlog;
 558                 opt.Plog        = q->parms.Plog;
 559                 opt.Scell_log   = q->parms.Scell_log;
 560                 opt.other       = q->stats.other;
 561                 opt.early       = q->stats.prob_drop;
 562                 opt.forced      = q->stats.forced_drop;
 563                 opt.pdrop       = q->stats.pdrop;
 564                 opt.packets     = q->packetsin;
 565                 opt.bytesin     = q->bytesin;
 566
 567                 if (gred_wred_mode(table)) {
 568                         q->parms.qidlestart =
 569                                 table->tab[table->def]->parms.qidlestart;
 570                         q->parms.qavg = table->tab[table->def]->parms.qavg;
 571                 }
 572
 573                 opt.qave = red_calc_qavg(&q->parms, q->parms.qavg);
 574
 575 append_opt:
 576                 RTA_APPEND(skb, sizeof(opt), &opt);
 577         }
 578
 579         RTA_NEST_END(skb, parms);
 580
 581         return RTA_NEST_END(skb, opts);
 582
 583 rtattr_failure:
 584         return RTA_NEST_CANCEL(skb, opts);
 585 }
 586
 587 static void gred_destroy(struct Qdisc *sch)
 588 {
 589         struct gred_sched *table = qdisc_priv(sch);
 590         int i;
 591
 592         for (i = 0;i < table->DPs; i++) {
 593                 if (table->tab[i])
 594                         gred_destroy_vq(table->tab[i]);
 595         }
 596 }
 597
 598 static struct Qdisc_ops gred_qdisc_ops = {
 599         .next           =       NULL,
 600         .cl_ops         =       NULL,
 601         .id             =       "gred",
 602         .priv_size      =       sizeof(struct gred_sched),
 603         .enqueue        =       gred_enqueue,
 604         .dequeue        =       gred_dequeue,
 605         .requeue        =       gred_requeue,
 606         .drop           =       gred_drop,
 607         .init           =       gred_init,
 608         .reset          =       gred_reset,
 609         .destroy        =       gred_destroy,
 610         .change         =       gred_change,
 611         .dump           =       gred_dump,
 612         .owner          =       THIS_MODULE,
 613 };
 614
 615 static int __init gred_module_init(void)
 616 {
 617         return register_qdisc(&gred_qdisc_ops);
 618 }
 619 static void __exit gred_module_exit(void)
 620 {
 621         unregister_qdisc(&gred_qdisc_ops);
 622 }
 623 module_init(gred_module_init)
 624 module_exit(gred_module_exit)
 625 MODULE_LICENSE("GPL");