1 /* 2 * net/sched/sch_tbf.c Token Bucket Filter 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 * Dmitry Torokhov <dtor@mail.ru> - allow attaching inner qdiscs - 11 * original idea by Martin Devera 12 * 13 */ 14 15 #include <linux/module.h> 16 #include <linux/types.h> 17 #include <linux/kernel.h> 18 #include <linux/string.h> 19 #include <linux/errno.h> 20 #include <linux/skbuff.h> 21 #include <net/netlink.h> 22 #include <net/sch_generic.h> 23 #include <net/pkt_sched.h> 24 25 26 /* Simple Token Bucket Filter. 27 ======================================= 28 29 SOURCE. 30 ------- 31 32 None. 33 34 Description. 35 ------------ 36 37 A data flow obeys TBF with rate R and depth B, if for any 38 time interval t_i...t_f the number of transmitted bits 39 does not exceed B + R*(t_f-t_i). 40 41 Packetized version of this definition: 42 The sequence of packets of sizes s_i served at moments t_i 43 obeys TBF, if for any i<=k: 44 45 s_i+....+s_k <= B + R*(t_k - t_i) 46 47 Algorithm. 48 ---------- 49 50 Let N(t_i) be B/R initially and N(t) grow continuously with time as: 51 52 N(t+delta) = min{B/R, N(t) + delta} 53 54 If the first packet in queue has length S, it may be 55 transmitted only at the time t_* when S/R <= N(t_*), 56 and in this case N(t) jumps: 57 58 N(t_* + 0) = N(t_* - 0) - S/R. 59 60 61 62 Actually, QoS requires two TBF to be applied to a data stream. 63 One of them controls steady state burst size, another 64 one with rate P (peak rate) and depth M (equal to link MTU) 65 limits bursts at a smaller time scale. 66 67 It is easy to see that P>R, and B>M. If P is infinity, this double 68 TBF is equivalent to a single one. 69 70 When TBF works in reshaping mode, latency is estimated as: 71 72 lat = max ((L-B)/R, (L-M)/P) 73 74 75 NOTES. 76 ------ 77 78 If TBF throttles, it starts a watchdog timer, which will wake it up 79 when it is ready to transmit. 80 Note that the minimal timer resolution is 1/HZ. 81 If no new packets arrive during this period, 82 or if the device is not awaken by EOI for some previous packet, 83 TBF can stop its activity for 1/HZ. 84 85 86 This means, that with depth B, the maximal rate is 87 88 R_crit = B*HZ 89 90 F.e. for 10Mbit ethernet and HZ=100 the minimal allowed B is ~10Kbytes. 91 92 Note that the peak rate TBF is much more tough: with MTU 1500 93 P_crit = 150Kbytes/sec. So, if you need greater peak 94 rates, use alpha with HZ=1000 :-) 95 96 With classful TBF, limit is just kept for backwards compatibility. 97 It is passed to the default bfifo qdisc - if the inner qdisc is 98 changed the limit is not effective anymore. 99 */ 100 101 struct tbf_sched_data { 102 /* Parameters */ 103 u32 limit; /* Maximal length of backlog: bytes */ 104 u32 max_size; 105 s64 buffer; /* Token bucket depth/rate: MUST BE >= MTU/B */ 106 s64 mtu; 107 struct psched_ratecfg rate; 108 struct psched_ratecfg peak; 109 110 /* Variables */ 111 s64 tokens; /* Current number of B tokens */ 112 s64 ptokens; /* Current number of P tokens */ 113 s64 t_c; /* Time check-point */ 114 struct Qdisc *qdisc; /* Inner qdisc, default - bfifo queue */ 115 struct qdisc_watchdog watchdog; /* Watchdog timer */ 116 }; 117 118 119 /* Time to Length, convert time in ns to length in bytes 120 * to determinate how many bytes can be sent in given time. 121 */ 122 static u64 psched_ns_t2l(const struct psched_ratecfg *r, 123 u64 time_in_ns) 124 { 125 /* The formula is : 126 * len = (time_in_ns * r->rate_bytes_ps) / NSEC_PER_SEC 127 */ 128 u64 len = time_in_ns * r->rate_bytes_ps; 129 130 do_div(len, NSEC_PER_SEC); 131 132 if (unlikely(r->linklayer == TC_LINKLAYER_ATM)) { 133 do_div(len, 53); 134 len = len * 48; 135 } 136 137 if (len > r->overhead) 138 len -= r->overhead; 139 else 140 len = 0; 141 142 return len; 143 } 144 145 /* 146 * Return length of individual segments of a gso packet, 147 * including all headers (MAC, IP, TCP/UDP) 148 */ 149 static unsigned int skb_gso_mac_seglen(const struct sk_buff *skb) 150 { 151 unsigned int hdr_len = skb_transport_header(skb) - skb_mac_header(skb); 152 return hdr_len + skb_gso_transport_seglen(skb); 153 } 154 155 /* GSO packet is too big, segment it so that tbf can transmit 156 * each segment in time 157 */ 158 static int tbf_segment(struct sk_buff *skb, struct Qdisc *sch) 159 { 160 struct tbf_sched_data *q = qdisc_priv(sch); 161 struct sk_buff *segs, *nskb; 162 netdev_features_t features = netif_skb_features(skb); 163 unsigned int len = 0, prev_len = qdisc_pkt_len(skb); 164 int ret, nb; 165 166 segs = skb_gso_segment(skb, features & ~NETIF_F_GSO_MASK); 167 168 if (IS_ERR_OR_NULL(segs)) 169 return qdisc_reshape_fail(skb, sch); 170 171 nb = 0; 172 while (segs) { 173 nskb = segs->next; 174 segs->next = NULL; 175 qdisc_skb_cb(segs)->pkt_len = segs->len; 176 len += segs->len; 177 ret = qdisc_enqueue(segs, q->qdisc); 178 if (ret != NET_XMIT_SUCCESS) { 179 if (net_xmit_drop_count(ret)) 180 qdisc_qstats_drop(sch); 181 } else { 182 nb++; 183 } 184 segs = nskb; 185 } 186 sch->q.qlen += nb; 187 if (nb > 1) 188 qdisc_tree_reduce_backlog(sch, 1 - nb, prev_len - len); 189 consume_skb(skb); 190 return nb > 0 ? NET_XMIT_SUCCESS : NET_XMIT_DROP; 191 } 192 193 static int tbf_enqueue(struct sk_buff *skb, struct Qdisc *sch) 194 { 195 struct tbf_sched_data *q = qdisc_priv(sch); 196 int ret; 197 198 if (qdisc_pkt_len(skb) > q->max_size) { 199 if (skb_is_gso(skb) && skb_gso_mac_seglen(skb) <= q->max_size) 200 return tbf_segment(skb, sch); 201 return qdisc_reshape_fail(skb, sch); 202 } 203 ret = qdisc_enqueue(skb, q->qdisc); 204 if (ret != NET_XMIT_SUCCESS) { 205 if (net_xmit_drop_count(ret)) 206 qdisc_qstats_drop(sch); 207 return ret; 208 } 209 210 sch->q.qlen++; 211 return NET_XMIT_SUCCESS; 212 } 213 214 static unsigned int tbf_drop(struct Qdisc *sch) 215 { 216 struct tbf_sched_data *q = qdisc_priv(sch); 217 unsigned int len = 0; 218 219 if (q->qdisc->ops->drop && (len = q->qdisc->ops->drop(q->qdisc)) != 0) { 220 sch->q.qlen--; 221 qdisc_qstats_drop(sch); 222 } 223 return len; 224 } 225 226 static bool tbf_peak_present(const struct tbf_sched_data *q) 227 { 228 return q->peak.rate_bytes_ps; 229 } 230 231 static struct sk_buff *tbf_dequeue(struct Qdisc *sch) 232 { 233 struct tbf_sched_data *q = qdisc_priv(sch); 234 struct sk_buff *skb; 235 236 skb = q->qdisc->ops->peek(q->qdisc); 237 238 if (skb) { 239 s64 now; 240 s64 toks; 241 s64 ptoks = 0; 242 unsigned int len = qdisc_pkt_len(skb); 243 244 now = ktime_get_ns(); 245 toks = min_t(s64, now - q->t_c, q->buffer); 246 247 if (tbf_peak_present(q)) { 248 ptoks = toks + q->ptokens; 249 if (ptoks > q->mtu) 250 ptoks = q->mtu; 251 ptoks -= (s64) psched_l2t_ns(&q->peak, len); 252 } 253 toks += q->tokens; 254 if (toks > q->buffer) 255 toks = q->buffer; 256 toks -= (s64) psched_l2t_ns(&q->rate, len); 257 258 if ((toks|ptoks) >= 0) { 259 skb = qdisc_dequeue_peeked(q->qdisc); 260 if (unlikely(!skb)) 261 return NULL; 262 263 q->t_c = now; 264 q->tokens = toks; 265 q->ptokens = ptoks; 266 sch->q.qlen--; 267 qdisc_unthrottled(sch); 268 qdisc_bstats_update(sch, skb); 269 return skb; 270 } 271 272 qdisc_watchdog_schedule_ns(&q->watchdog, 273 now + max_t(long, -toks, -ptoks), 274 true); 275 276 /* Maybe we have a shorter packet in the queue, 277 which can be sent now. It sounds cool, 278 but, however, this is wrong in principle. 279 We MUST NOT reorder packets under these circumstances. 280 281 Really, if we split the flow into independent 282 subflows, it would be a very good solution. 283 This is the main idea of all FQ algorithms 284 (cf. CSZ, HPFQ, HFSC) 285 */ 286 287 qdisc_qstats_overlimit(sch); 288 } 289 return NULL; 290 } 291 292 static void tbf_reset(struct Qdisc *sch) 293 { 294 struct tbf_sched_data *q = qdisc_priv(sch); 295 296 qdisc_reset(q->qdisc); 297 sch->q.qlen = 0; 298 q->t_c = ktime_get_ns(); 299 q->tokens = q->buffer; 300 q->ptokens = q->mtu; 301 qdisc_watchdog_cancel(&q->watchdog); 302 } 303 304 static const struct nla_policy tbf_policy[TCA_TBF_MAX + 1] = { 305 [TCA_TBF_PARMS] = { .len = sizeof(struct tc_tbf_qopt) }, 306 [TCA_TBF_RTAB] = { .type = NLA_BINARY, .len = TC_RTAB_SIZE }, 307 [TCA_TBF_PTAB] = { .type = NLA_BINARY, .len = TC_RTAB_SIZE }, 308 [TCA_TBF_RATE64] = { .type = NLA_U64 }, 309 [TCA_TBF_PRATE64] = { .type = NLA_U64 }, 310 [TCA_TBF_BURST] = { .type = NLA_U32 }, 311 [TCA_TBF_PBURST] = { .type = NLA_U32 }, 312 }; 313 314 static int tbf_change(struct Qdisc *sch, struct nlattr *opt) 315 { 316 int err; 317 struct tbf_sched_data *q = qdisc_priv(sch); 318 struct nlattr *tb[TCA_TBF_MAX + 1]; 319 struct tc_tbf_qopt *qopt; 320 struct Qdisc *child = NULL; 321 struct psched_ratecfg rate; 322 struct psched_ratecfg peak; 323 u64 max_size; 324 s64 buffer, mtu; 325 u64 rate64 = 0, prate64 = 0; 326 327 err = nla_parse_nested(tb, TCA_TBF_MAX, opt, tbf_policy); 328 if (err < 0) 329 return err; 330 331 err = -EINVAL; 332 if (tb[TCA_TBF_PARMS] == NULL) 333 goto done; 334 335 qopt = nla_data(tb[TCA_TBF_PARMS]); 336 if (qopt->rate.linklayer == TC_LINKLAYER_UNAWARE) 337 qdisc_put_rtab(qdisc_get_rtab(&qopt->rate, 338 tb[TCA_TBF_RTAB])); 339 340 if (qopt->peakrate.linklayer == TC_LINKLAYER_UNAWARE) 341 qdisc_put_rtab(qdisc_get_rtab(&qopt->peakrate, 342 tb[TCA_TBF_PTAB])); 343 344 buffer = min_t(u64, PSCHED_TICKS2NS(qopt->buffer), ~0U); 345 mtu = min_t(u64, PSCHED_TICKS2NS(qopt->mtu), ~0U); 346 347 if (tb[TCA_TBF_RATE64]) 348 rate64 = nla_get_u64(tb[TCA_TBF_RATE64]); 349 psched_ratecfg_precompute(&rate, &qopt->rate, rate64); 350 351 if (tb[TCA_TBF_BURST]) { 352 max_size = nla_get_u32(tb[TCA_TBF_BURST]); 353 buffer = psched_l2t_ns(&rate, max_size); 354 } else { 355 max_size = min_t(u64, psched_ns_t2l(&rate, buffer), ~0U); 356 } 357 358 if (qopt->peakrate.rate) { 359 if (tb[TCA_TBF_PRATE64]) 360 prate64 = nla_get_u64(tb[TCA_TBF_PRATE64]); 361 psched_ratecfg_precompute(&peak, &qopt->peakrate, prate64); 362 if (peak.rate_bytes_ps <= rate.rate_bytes_ps) { 363 pr_warn_ratelimited("sch_tbf: peakrate %llu is lower than or equals to rate %llu !\n", 364 peak.rate_bytes_ps, rate.rate_bytes_ps); 365 err = -EINVAL; 366 goto done; 367 } 368 369 if (tb[TCA_TBF_PBURST]) { 370 u32 pburst = nla_get_u32(tb[TCA_TBF_PBURST]); 371 max_size = min_t(u32, max_size, pburst); 372 mtu = psched_l2t_ns(&peak, pburst); 373 } else { 374 max_size = min_t(u64, max_size, psched_ns_t2l(&peak, mtu)); 375 } 376 } else { 377 memset(&peak, 0, sizeof(peak)); 378 } 379 380 if (max_size < psched_mtu(qdisc_dev(sch))) 381 pr_warn_ratelimited("sch_tbf: burst %llu is lower than device %s mtu (%u) !\n", 382 max_size, qdisc_dev(sch)->name, 383 psched_mtu(qdisc_dev(sch))); 384 385 if (!max_size) { 386 err = -EINVAL; 387 goto done; 388 } 389 390 if (q->qdisc != &noop_qdisc) { 391 err = fifo_set_limit(q->qdisc, qopt->limit); 392 if (err) 393 goto done; 394 } else if (qopt->limit > 0) { 395 child = fifo_create_dflt(sch, &bfifo_qdisc_ops, qopt->limit); 396 if (IS_ERR(child)) { 397 err = PTR_ERR(child); 398 goto done; 399 } 400 } 401 402 sch_tree_lock(sch); 403 if (child) { 404 qdisc_tree_reduce_backlog(q->qdisc, q->qdisc->q.qlen, 405 q->qdisc->qstats.backlog); 406 qdisc_destroy(q->qdisc); 407 q->qdisc = child; 408 } 409 q->limit = qopt->limit; 410 if (tb[TCA_TBF_PBURST]) 411 q->mtu = mtu; 412 else 413 q->mtu = PSCHED_TICKS2NS(qopt->mtu); 414 q->max_size = max_size; 415 if (tb[TCA_TBF_BURST]) 416 q->buffer = buffer; 417 else 418 q->buffer = PSCHED_TICKS2NS(qopt->buffer); 419 q->tokens = q->buffer; 420 q->ptokens = q->mtu; 421 422 memcpy(&q->rate, &rate, sizeof(struct psched_ratecfg)); 423 memcpy(&q->peak, &peak, sizeof(struct psched_ratecfg)); 424 425 sch_tree_unlock(sch); 426 err = 0; 427 done: 428 return err; 429 } 430 431 static int tbf_init(struct Qdisc *sch, struct nlattr *opt) 432 { 433 struct tbf_sched_data *q = qdisc_priv(sch); 434 435 if (opt == NULL) 436 return -EINVAL; 437 438 q->t_c = ktime_get_ns(); 439 qdisc_watchdog_init(&q->watchdog, sch); 440 q->qdisc = &noop_qdisc; 441 442 return tbf_change(sch, opt); 443 } 444 445 static void tbf_destroy(struct Qdisc *sch) 446 { 447 struct tbf_sched_data *q = qdisc_priv(sch); 448 449 qdisc_watchdog_cancel(&q->watchdog); 450 qdisc_destroy(q->qdisc); 451 } 452 453 static int tbf_dump(struct Qdisc *sch, struct sk_buff *skb) 454 { 455 struct tbf_sched_data *q = qdisc_priv(sch); 456 struct nlattr *nest; 457 struct tc_tbf_qopt opt; 458 459 sch->qstats.backlog = q->qdisc->qstats.backlog; 460 nest = nla_nest_start(skb, TCA_OPTIONS); 461 if (nest == NULL) 462 goto nla_put_failure; 463 464 opt.limit = q->limit; 465 psched_ratecfg_getrate(&opt.rate, &q->rate); 466 if (tbf_peak_present(q)) 467 psched_ratecfg_getrate(&opt.peakrate, &q->peak); 468 else 469 memset(&opt.peakrate, 0, sizeof(opt.peakrate)); 470 opt.mtu = PSCHED_NS2TICKS(q->mtu); 471 opt.buffer = PSCHED_NS2TICKS(q->buffer); 472 if (nla_put(skb, TCA_TBF_PARMS, sizeof(opt), &opt)) 473 goto nla_put_failure; 474 if (q->rate.rate_bytes_ps >= (1ULL << 32) && 475 nla_put_u64_64bit(skb, TCA_TBF_RATE64, q->rate.rate_bytes_ps, 476 TCA_TBF_PAD)) 477 goto nla_put_failure; 478 if (tbf_peak_present(q) && 479 q->peak.rate_bytes_ps >= (1ULL << 32) && 480 nla_put_u64_64bit(skb, TCA_TBF_PRATE64, q->peak.rate_bytes_ps, 481 TCA_TBF_PAD)) 482 goto nla_put_failure; 483 484 return nla_nest_end(skb, nest); 485 486 nla_put_failure: 487 nla_nest_cancel(skb, nest); 488 return -1; 489 } 490 491 static int tbf_dump_class(struct Qdisc *sch, unsigned long cl, 492 struct sk_buff *skb, struct tcmsg *tcm) 493 { 494 struct tbf_sched_data *q = qdisc_priv(sch); 495 496 tcm->tcm_handle |= TC_H_MIN(1); 497 tcm->tcm_info = q->qdisc->handle; 498 499 return 0; 500 } 501 502 static int tbf_graft(struct Qdisc *sch, unsigned long arg, struct Qdisc *new, 503 struct Qdisc **old) 504 { 505 struct tbf_sched_data *q = qdisc_priv(sch); 506 507 if (new == NULL) 508 new = &noop_qdisc; 509 510 *old = qdisc_replace(sch, new, &q->qdisc); 511 return 0; 512 } 513 514 static struct Qdisc *tbf_leaf(struct Qdisc *sch, unsigned long arg) 515 { 516 struct tbf_sched_data *q = qdisc_priv(sch); 517 return q->qdisc; 518 } 519 520 static unsigned long tbf_get(struct Qdisc *sch, u32 classid) 521 { 522 return 1; 523 } 524 525 static void tbf_put(struct Qdisc *sch, unsigned long arg) 526 { 527 } 528 529 static void tbf_walk(struct Qdisc *sch, struct qdisc_walker *walker) 530 { 531 if (!walker->stop) { 532 if (walker->count >= walker->skip) 533 if (walker->fn(sch, 1, walker) < 0) { 534 walker->stop = 1; 535 return; 536 } 537 walker->count++; 538 } 539 } 540 541 static const struct Qdisc_class_ops tbf_class_ops = { 542 .graft = tbf_graft, 543 .leaf = tbf_leaf, 544 .get = tbf_get, 545 .put = tbf_put, 546 .walk = tbf_walk, 547 .dump = tbf_dump_class, 548 }; 549 550 static struct Qdisc_ops tbf_qdisc_ops __read_mostly = { 551 .next = NULL, 552 .cl_ops = &tbf_class_ops, 553 .id = "tbf", 554 .priv_size = sizeof(struct tbf_sched_data), 555 .enqueue = tbf_enqueue, 556 .dequeue = tbf_dequeue, 557 .peek = qdisc_peek_dequeued, 558 .drop = tbf_drop, 559 .init = tbf_init, 560 .reset = tbf_reset, 561 .destroy = tbf_destroy, 562 .change = tbf_change, 563 .dump = tbf_dump, 564 .owner = THIS_MODULE, 565 }; 566 567 static int __init tbf_module_init(void) 568 { 569 return register_qdisc(&tbf_qdisc_ops); 570 } 571 572 static void __exit tbf_module_exit(void) 573 { 574 unregister_qdisc(&tbf_qdisc_ops); 575 } 576 module_init(tbf_module_init) 577 module_exit(tbf_module_exit) 578 MODULE_LICENSE("GPL"); 579