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 struct sk_buff **to_free) 160 { 161 struct tbf_sched_data *q = qdisc_priv(sch); 162 struct sk_buff *segs, *nskb; 163 netdev_features_t features = netif_skb_features(skb); 164 unsigned int len = 0, prev_len = qdisc_pkt_len(skb); 165 int ret, nb; 166 167 segs = skb_gso_segment(skb, features & ~NETIF_F_GSO_MASK); 168 169 if (IS_ERR_OR_NULL(segs)) 170 return qdisc_drop(skb, sch, to_free); 171 172 nb = 0; 173 while (segs) { 174 nskb = segs->next; 175 segs->next = NULL; 176 qdisc_skb_cb(segs)->pkt_len = segs->len; 177 len += segs->len; 178 ret = qdisc_enqueue(segs, q->qdisc, to_free); 179 if (ret != NET_XMIT_SUCCESS) { 180 if (net_xmit_drop_count(ret)) 181 qdisc_qstats_drop(sch); 182 } else { 183 nb++; 184 } 185 segs = nskb; 186 } 187 sch->q.qlen += nb; 188 if (nb > 1) 189 qdisc_tree_reduce_backlog(sch, 1 - nb, prev_len - len); 190 consume_skb(skb); 191 return nb > 0 ? NET_XMIT_SUCCESS : NET_XMIT_DROP; 192 } 193 194 static int tbf_enqueue(struct sk_buff *skb, struct Qdisc *sch, 195 struct sk_buff **to_free) 196 { 197 struct tbf_sched_data *q = qdisc_priv(sch); 198 int ret; 199 200 if (qdisc_pkt_len(skb) > q->max_size) { 201 if (skb_is_gso(skb) && skb_gso_mac_seglen(skb) <= q->max_size) 202 return tbf_segment(skb, sch, to_free); 203 return qdisc_drop(skb, sch, to_free); 204 } 205 ret = qdisc_enqueue(skb, q->qdisc, to_free); 206 if (ret != NET_XMIT_SUCCESS) { 207 if (net_xmit_drop_count(ret)) 208 qdisc_qstats_drop(sch); 209 return ret; 210 } 211 212 qdisc_qstats_backlog_inc(sch, skb); 213 sch->q.qlen++; 214 return NET_XMIT_SUCCESS; 215 } 216 217 static bool tbf_peak_present(const struct tbf_sched_data *q) 218 { 219 return q->peak.rate_bytes_ps; 220 } 221 222 static struct sk_buff *tbf_dequeue(struct Qdisc *sch) 223 { 224 struct tbf_sched_data *q = qdisc_priv(sch); 225 struct sk_buff *skb; 226 227 skb = q->qdisc->ops->peek(q->qdisc); 228 229 if (skb) { 230 s64 now; 231 s64 toks; 232 s64 ptoks = 0; 233 unsigned int len = qdisc_pkt_len(skb); 234 235 now = ktime_get_ns(); 236 toks = min_t(s64, now - q->t_c, q->buffer); 237 238 if (tbf_peak_present(q)) { 239 ptoks = toks + q->ptokens; 240 if (ptoks > q->mtu) 241 ptoks = q->mtu; 242 ptoks -= (s64) psched_l2t_ns(&q->peak, len); 243 } 244 toks += q->tokens; 245 if (toks > q->buffer) 246 toks = q->buffer; 247 toks -= (s64) psched_l2t_ns(&q->rate, len); 248 249 if ((toks|ptoks) >= 0) { 250 skb = qdisc_dequeue_peeked(q->qdisc); 251 if (unlikely(!skb)) 252 return NULL; 253 254 q->t_c = now; 255 q->tokens = toks; 256 q->ptokens = ptoks; 257 qdisc_qstats_backlog_dec(sch, skb); 258 sch->q.qlen--; 259 qdisc_bstats_update(sch, skb); 260 return skb; 261 } 262 263 qdisc_watchdog_schedule_ns(&q->watchdog, 264 now + max_t(long, -toks, -ptoks)); 265 266 /* Maybe we have a shorter packet in the queue, 267 which can be sent now. It sounds cool, 268 but, however, this is wrong in principle. 269 We MUST NOT reorder packets under these circumstances. 270 271 Really, if we split the flow into independent 272 subflows, it would be a very good solution. 273 This is the main idea of all FQ algorithms 274 (cf. CSZ, HPFQ, HFSC) 275 */ 276 277 qdisc_qstats_overlimit(sch); 278 } 279 return NULL; 280 } 281 282 static void tbf_reset(struct Qdisc *sch) 283 { 284 struct tbf_sched_data *q = qdisc_priv(sch); 285 286 qdisc_reset(q->qdisc); 287 sch->qstats.backlog = 0; 288 sch->q.qlen = 0; 289 q->t_c = ktime_get_ns(); 290 q->tokens = q->buffer; 291 q->ptokens = q->mtu; 292 qdisc_watchdog_cancel(&q->watchdog); 293 } 294 295 static const struct nla_policy tbf_policy[TCA_TBF_MAX + 1] = { 296 [TCA_TBF_PARMS] = { .len = sizeof(struct tc_tbf_qopt) }, 297 [TCA_TBF_RTAB] = { .type = NLA_BINARY, .len = TC_RTAB_SIZE }, 298 [TCA_TBF_PTAB] = { .type = NLA_BINARY, .len = TC_RTAB_SIZE }, 299 [TCA_TBF_RATE64] = { .type = NLA_U64 }, 300 [TCA_TBF_PRATE64] = { .type = NLA_U64 }, 301 [TCA_TBF_BURST] = { .type = NLA_U32 }, 302 [TCA_TBF_PBURST] = { .type = NLA_U32 }, 303 }; 304 305 static int tbf_change(struct Qdisc *sch, struct nlattr *opt) 306 { 307 int err; 308 struct tbf_sched_data *q = qdisc_priv(sch); 309 struct nlattr *tb[TCA_TBF_MAX + 1]; 310 struct tc_tbf_qopt *qopt; 311 struct Qdisc *child = NULL; 312 struct psched_ratecfg rate; 313 struct psched_ratecfg peak; 314 u64 max_size; 315 s64 buffer, mtu; 316 u64 rate64 = 0, prate64 = 0; 317 318 err = nla_parse_nested(tb, TCA_TBF_MAX, opt, tbf_policy, NULL); 319 if (err < 0) 320 return err; 321 322 err = -EINVAL; 323 if (tb[TCA_TBF_PARMS] == NULL) 324 goto done; 325 326 qopt = nla_data(tb[TCA_TBF_PARMS]); 327 if (qopt->rate.linklayer == TC_LINKLAYER_UNAWARE) 328 qdisc_put_rtab(qdisc_get_rtab(&qopt->rate, 329 tb[TCA_TBF_RTAB])); 330 331 if (qopt->peakrate.linklayer == TC_LINKLAYER_UNAWARE) 332 qdisc_put_rtab(qdisc_get_rtab(&qopt->peakrate, 333 tb[TCA_TBF_PTAB])); 334 335 buffer = min_t(u64, PSCHED_TICKS2NS(qopt->buffer), ~0U); 336 mtu = min_t(u64, PSCHED_TICKS2NS(qopt->mtu), ~0U); 337 338 if (tb[TCA_TBF_RATE64]) 339 rate64 = nla_get_u64(tb[TCA_TBF_RATE64]); 340 psched_ratecfg_precompute(&rate, &qopt->rate, rate64); 341 342 if (tb[TCA_TBF_BURST]) { 343 max_size = nla_get_u32(tb[TCA_TBF_BURST]); 344 buffer = psched_l2t_ns(&rate, max_size); 345 } else { 346 max_size = min_t(u64, psched_ns_t2l(&rate, buffer), ~0U); 347 } 348 349 if (qopt->peakrate.rate) { 350 if (tb[TCA_TBF_PRATE64]) 351 prate64 = nla_get_u64(tb[TCA_TBF_PRATE64]); 352 psched_ratecfg_precompute(&peak, &qopt->peakrate, prate64); 353 if (peak.rate_bytes_ps <= rate.rate_bytes_ps) { 354 pr_warn_ratelimited("sch_tbf: peakrate %llu is lower than or equals to rate %llu !\n", 355 peak.rate_bytes_ps, rate.rate_bytes_ps); 356 err = -EINVAL; 357 goto done; 358 } 359 360 if (tb[TCA_TBF_PBURST]) { 361 u32 pburst = nla_get_u32(tb[TCA_TBF_PBURST]); 362 max_size = min_t(u32, max_size, pburst); 363 mtu = psched_l2t_ns(&peak, pburst); 364 } else { 365 max_size = min_t(u64, max_size, psched_ns_t2l(&peak, mtu)); 366 } 367 } else { 368 memset(&peak, 0, sizeof(peak)); 369 } 370 371 if (max_size < psched_mtu(qdisc_dev(sch))) 372 pr_warn_ratelimited("sch_tbf: burst %llu is lower than device %s mtu (%u) !\n", 373 max_size, qdisc_dev(sch)->name, 374 psched_mtu(qdisc_dev(sch))); 375 376 if (!max_size) { 377 err = -EINVAL; 378 goto done; 379 } 380 381 if (q->qdisc != &noop_qdisc) { 382 err = fifo_set_limit(q->qdisc, qopt->limit); 383 if (err) 384 goto done; 385 } else if (qopt->limit > 0) { 386 child = fifo_create_dflt(sch, &bfifo_qdisc_ops, qopt->limit); 387 if (IS_ERR(child)) { 388 err = PTR_ERR(child); 389 goto done; 390 } 391 } 392 393 sch_tree_lock(sch); 394 if (child) { 395 qdisc_tree_reduce_backlog(q->qdisc, q->qdisc->q.qlen, 396 q->qdisc->qstats.backlog); 397 qdisc_destroy(q->qdisc); 398 q->qdisc = child; 399 if (child != &noop_qdisc) 400 qdisc_hash_add(child, true); 401 } 402 q->limit = qopt->limit; 403 if (tb[TCA_TBF_PBURST]) 404 q->mtu = mtu; 405 else 406 q->mtu = PSCHED_TICKS2NS(qopt->mtu); 407 q->max_size = max_size; 408 if (tb[TCA_TBF_BURST]) 409 q->buffer = buffer; 410 else 411 q->buffer = PSCHED_TICKS2NS(qopt->buffer); 412 q->tokens = q->buffer; 413 q->ptokens = q->mtu; 414 415 memcpy(&q->rate, &rate, sizeof(struct psched_ratecfg)); 416 memcpy(&q->peak, &peak, sizeof(struct psched_ratecfg)); 417 418 sch_tree_unlock(sch); 419 err = 0; 420 done: 421 return err; 422 } 423 424 static int tbf_init(struct Qdisc *sch, struct nlattr *opt) 425 { 426 struct tbf_sched_data *q = qdisc_priv(sch); 427 428 qdisc_watchdog_init(&q->watchdog, sch); 429 q->qdisc = &noop_qdisc; 430 431 if (opt == NULL) 432 return -EINVAL; 433 434 q->t_c = ktime_get_ns(); 435 436 return tbf_change(sch, opt); 437 } 438 439 static void tbf_destroy(struct Qdisc *sch) 440 { 441 struct tbf_sched_data *q = qdisc_priv(sch); 442 443 qdisc_watchdog_cancel(&q->watchdog); 444 qdisc_destroy(q->qdisc); 445 } 446 447 static int tbf_dump(struct Qdisc *sch, struct sk_buff *skb) 448 { 449 struct tbf_sched_data *q = qdisc_priv(sch); 450 struct nlattr *nest; 451 struct tc_tbf_qopt opt; 452 453 sch->qstats.backlog = q->qdisc->qstats.backlog; 454 nest = nla_nest_start(skb, TCA_OPTIONS); 455 if (nest == NULL) 456 goto nla_put_failure; 457 458 opt.limit = q->limit; 459 psched_ratecfg_getrate(&opt.rate, &q->rate); 460 if (tbf_peak_present(q)) 461 psched_ratecfg_getrate(&opt.peakrate, &q->peak); 462 else 463 memset(&opt.peakrate, 0, sizeof(opt.peakrate)); 464 opt.mtu = PSCHED_NS2TICKS(q->mtu); 465 opt.buffer = PSCHED_NS2TICKS(q->buffer); 466 if (nla_put(skb, TCA_TBF_PARMS, sizeof(opt), &opt)) 467 goto nla_put_failure; 468 if (q->rate.rate_bytes_ps >= (1ULL << 32) && 469 nla_put_u64_64bit(skb, TCA_TBF_RATE64, q->rate.rate_bytes_ps, 470 TCA_TBF_PAD)) 471 goto nla_put_failure; 472 if (tbf_peak_present(q) && 473 q->peak.rate_bytes_ps >= (1ULL << 32) && 474 nla_put_u64_64bit(skb, TCA_TBF_PRATE64, q->peak.rate_bytes_ps, 475 TCA_TBF_PAD)) 476 goto nla_put_failure; 477 478 return nla_nest_end(skb, nest); 479 480 nla_put_failure: 481 nla_nest_cancel(skb, nest); 482 return -1; 483 } 484 485 static int tbf_dump_class(struct Qdisc *sch, unsigned long cl, 486 struct sk_buff *skb, struct tcmsg *tcm) 487 { 488 struct tbf_sched_data *q = qdisc_priv(sch); 489 490 tcm->tcm_handle |= TC_H_MIN(1); 491 tcm->tcm_info = q->qdisc->handle; 492 493 return 0; 494 } 495 496 static int tbf_graft(struct Qdisc *sch, unsigned long arg, struct Qdisc *new, 497 struct Qdisc **old) 498 { 499 struct tbf_sched_data *q = qdisc_priv(sch); 500 501 if (new == NULL) 502 new = &noop_qdisc; 503 504 *old = qdisc_replace(sch, new, &q->qdisc); 505 return 0; 506 } 507 508 static struct Qdisc *tbf_leaf(struct Qdisc *sch, unsigned long arg) 509 { 510 struct tbf_sched_data *q = qdisc_priv(sch); 511 return q->qdisc; 512 } 513 514 static unsigned long tbf_find(struct Qdisc *sch, u32 classid) 515 { 516 return 1; 517 } 518 519 static void tbf_walk(struct Qdisc *sch, struct qdisc_walker *walker) 520 { 521 if (!walker->stop) { 522 if (walker->count >= walker->skip) 523 if (walker->fn(sch, 1, walker) < 0) { 524 walker->stop = 1; 525 return; 526 } 527 walker->count++; 528 } 529 } 530 531 static const struct Qdisc_class_ops tbf_class_ops = { 532 .graft = tbf_graft, 533 .leaf = tbf_leaf, 534 .find = tbf_find, 535 .walk = tbf_walk, 536 .dump = tbf_dump_class, 537 }; 538 539 static struct Qdisc_ops tbf_qdisc_ops __read_mostly = { 540 .next = NULL, 541 .cl_ops = &tbf_class_ops, 542 .id = "tbf", 543 .priv_size = sizeof(struct tbf_sched_data), 544 .enqueue = tbf_enqueue, 545 .dequeue = tbf_dequeue, 546 .peek = qdisc_peek_dequeued, 547 .init = tbf_init, 548 .reset = tbf_reset, 549 .destroy = tbf_destroy, 550 .change = tbf_change, 551 .dump = tbf_dump, 552 .owner = THIS_MODULE, 553 }; 554 555 static int __init tbf_module_init(void) 556 { 557 return register_qdisc(&tbf_qdisc_ops); 558 } 559 560 static void __exit tbf_module_exit(void) 561 { 562 unregister_qdisc(&tbf_qdisc_ops); 563 } 564 module_init(tbf_module_init) 565 module_exit(tbf_module_exit) 566 MODULE_LICENSE("GPL"); 567