1 /* 2 * net/sched/sch_netem.c Network emulator 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. 8 * 9 * Many of the algorithms and ideas for this came from 10 * NIST Net which is not copyrighted. 11 * 12 * Authors: Stephen Hemminger <shemminger@osdl.org> 13 * Catalin(ux aka Dino) BOIE <catab at umbrella dot ro> 14 */ 15 16 #include <linux/mm.h> 17 #include <linux/module.h> 18 #include <linux/slab.h> 19 #include <linux/types.h> 20 #include <linux/kernel.h> 21 #include <linux/errno.h> 22 #include <linux/skbuff.h> 23 #include <linux/vmalloc.h> 24 #include <linux/rtnetlink.h> 25 #include <linux/reciprocal_div.h> 26 #include <linux/rbtree.h> 27 28 #include <net/netlink.h> 29 #include <net/pkt_sched.h> 30 #include <net/inet_ecn.h> 31 32 #define VERSION "1.3" 33 34 /* Network Emulation Queuing algorithm. 35 ==================================== 36 37 Sources: [1] Mark Carson, Darrin Santay, "NIST Net - A Linux-based 38 Network Emulation Tool 39 [2] Luigi Rizzo, DummyNet for FreeBSD 40 41 ---------------------------------------------------------------- 42 43 This started out as a simple way to delay outgoing packets to 44 test TCP but has grown to include most of the functionality 45 of a full blown network emulator like NISTnet. It can delay 46 packets and add random jitter (and correlation). The random 47 distribution can be loaded from a table as well to provide 48 normal, Pareto, or experimental curves. Packet loss, 49 duplication, and reordering can also be emulated. 50 51 This qdisc does not do classification that can be handled in 52 layering other disciplines. It does not need to do bandwidth 53 control either since that can be handled by using token 54 bucket or other rate control. 55 56 Correlated Loss Generator models 57 58 Added generation of correlated loss according to the 59 "Gilbert-Elliot" model, a 4-state markov model. 60 61 References: 62 [1] NetemCLG Home http://netgroup.uniroma2.it/NetemCLG 63 [2] S. Salsano, F. Ludovici, A. Ordine, "Definition of a general 64 and intuitive loss model for packet networks and its implementation 65 in the Netem module in the Linux kernel", available in [1] 66 67 Authors: Stefano Salsano <stefano.salsano at uniroma2.it 68 Fabio Ludovici <fabio.ludovici at yahoo.it> 69 */ 70 71 struct disttable { 72 u32 size; 73 s16 table[0]; 74 }; 75 76 struct netem_sched_data { 77 /* internal t(ime)fifo qdisc uses t_root and sch->limit */ 78 struct rb_root t_root; 79 80 /* optional qdisc for classful handling (NULL at netem init) */ 81 struct Qdisc *qdisc; 82 83 struct qdisc_watchdog watchdog; 84 85 s64 latency; 86 s64 jitter; 87 88 u32 loss; 89 u32 ecn; 90 u32 limit; 91 u32 counter; 92 u32 gap; 93 u32 duplicate; 94 u32 reorder; 95 u32 corrupt; 96 u64 rate; 97 s32 packet_overhead; 98 u32 cell_size; 99 struct reciprocal_value cell_size_reciprocal; 100 s32 cell_overhead; 101 102 struct crndstate { 103 u32 last; 104 u32 rho; 105 } delay_cor, loss_cor, dup_cor, reorder_cor, corrupt_cor; 106 107 struct disttable *delay_dist; 108 109 enum { 110 CLG_RANDOM, 111 CLG_4_STATES, 112 CLG_GILB_ELL, 113 } loss_model; 114 115 enum { 116 TX_IN_GAP_PERIOD = 1, 117 TX_IN_BURST_PERIOD, 118 LOST_IN_GAP_PERIOD, 119 LOST_IN_BURST_PERIOD, 120 } _4_state_model; 121 122 enum { 123 GOOD_STATE = 1, 124 BAD_STATE, 125 } GE_state_model; 126 127 /* Correlated Loss Generation models */ 128 struct clgstate { 129 /* state of the Markov chain */ 130 u8 state; 131 132 /* 4-states and Gilbert-Elliot models */ 133 u32 a1; /* p13 for 4-states or p for GE */ 134 u32 a2; /* p31 for 4-states or r for GE */ 135 u32 a3; /* p32 for 4-states or h for GE */ 136 u32 a4; /* p14 for 4-states or 1-k for GE */ 137 u32 a5; /* p23 used only in 4-states */ 138 } clg; 139 140 struct tc_netem_slot slot_config; 141 struct slotstate { 142 u64 slot_next; 143 s32 packets_left; 144 s32 bytes_left; 145 } slot; 146 147 struct disttable *slot_dist; 148 }; 149 150 /* Time stamp put into socket buffer control block 151 * Only valid when skbs are in our internal t(ime)fifo queue. 152 * 153 * As skb->rbnode uses same storage than skb->next, skb->prev and skb->tstamp, 154 * and skb->next & skb->prev are scratch space for a qdisc, 155 * we save skb->tstamp value in skb->cb[] before destroying it. 156 */ 157 struct netem_skb_cb { 158 u64 time_to_send; 159 }; 160 161 static inline struct netem_skb_cb *netem_skb_cb(struct sk_buff *skb) 162 { 163 /* we assume we can use skb next/prev/tstamp as storage for rb_node */ 164 qdisc_cb_private_validate(skb, sizeof(struct netem_skb_cb)); 165 return (struct netem_skb_cb *)qdisc_skb_cb(skb)->data; 166 } 167 168 /* init_crandom - initialize correlated random number generator 169 * Use entropy source for initial seed. 170 */ 171 static void init_crandom(struct crndstate *state, unsigned long rho) 172 { 173 state->rho = rho; 174 state->last = prandom_u32(); 175 } 176 177 /* get_crandom - correlated random number generator 178 * Next number depends on last value. 179 * rho is scaled to avoid floating point. 180 */ 181 static u32 get_crandom(struct crndstate *state) 182 { 183 u64 value, rho; 184 unsigned long answer; 185 186 if (!state || state->rho == 0) /* no correlation */ 187 return prandom_u32(); 188 189 value = prandom_u32(); 190 rho = (u64)state->rho + 1; 191 answer = (value * ((1ull<<32) - rho) + state->last * rho) >> 32; 192 state->last = answer; 193 return answer; 194 } 195 196 /* loss_4state - 4-state model loss generator 197 * Generates losses according to the 4-state Markov chain adopted in 198 * the GI (General and Intuitive) loss model. 199 */ 200 static bool loss_4state(struct netem_sched_data *q) 201 { 202 struct clgstate *clg = &q->clg; 203 u32 rnd = prandom_u32(); 204 205 /* 206 * Makes a comparison between rnd and the transition 207 * probabilities outgoing from the current state, then decides the 208 * next state and if the next packet has to be transmitted or lost. 209 * The four states correspond to: 210 * TX_IN_GAP_PERIOD => successfully transmitted packets within a gap period 211 * LOST_IN_BURST_PERIOD => isolated losses within a gap period 212 * LOST_IN_GAP_PERIOD => lost packets within a burst period 213 * TX_IN_GAP_PERIOD => successfully transmitted packets within a burst period 214 */ 215 switch (clg->state) { 216 case TX_IN_GAP_PERIOD: 217 if (rnd < clg->a4) { 218 clg->state = LOST_IN_BURST_PERIOD; 219 return true; 220 } else if (clg->a4 < rnd && rnd < clg->a1 + clg->a4) { 221 clg->state = LOST_IN_GAP_PERIOD; 222 return true; 223 } else if (clg->a1 + clg->a4 < rnd) { 224 clg->state = TX_IN_GAP_PERIOD; 225 } 226 227 break; 228 case TX_IN_BURST_PERIOD: 229 if (rnd < clg->a5) { 230 clg->state = LOST_IN_GAP_PERIOD; 231 return true; 232 } else { 233 clg->state = TX_IN_BURST_PERIOD; 234 } 235 236 break; 237 case LOST_IN_GAP_PERIOD: 238 if (rnd < clg->a3) 239 clg->state = TX_IN_BURST_PERIOD; 240 else if (clg->a3 < rnd && rnd < clg->a2 + clg->a3) { 241 clg->state = TX_IN_GAP_PERIOD; 242 } else if (clg->a2 + clg->a3 < rnd) { 243 clg->state = LOST_IN_GAP_PERIOD; 244 return true; 245 } 246 break; 247 case LOST_IN_BURST_PERIOD: 248 clg->state = TX_IN_GAP_PERIOD; 249 break; 250 } 251 252 return false; 253 } 254 255 /* loss_gilb_ell - Gilbert-Elliot model loss generator 256 * Generates losses according to the Gilbert-Elliot loss model or 257 * its special cases (Gilbert or Simple Gilbert) 258 * 259 * Makes a comparison between random number and the transition 260 * probabilities outgoing from the current state, then decides the 261 * next state. A second random number is extracted and the comparison 262 * with the loss probability of the current state decides if the next 263 * packet will be transmitted or lost. 264 */ 265 static bool loss_gilb_ell(struct netem_sched_data *q) 266 { 267 struct clgstate *clg = &q->clg; 268 269 switch (clg->state) { 270 case GOOD_STATE: 271 if (prandom_u32() < clg->a1) 272 clg->state = BAD_STATE; 273 if (prandom_u32() < clg->a4) 274 return true; 275 break; 276 case BAD_STATE: 277 if (prandom_u32() < clg->a2) 278 clg->state = GOOD_STATE; 279 if (prandom_u32() > clg->a3) 280 return true; 281 } 282 283 return false; 284 } 285 286 static bool loss_event(struct netem_sched_data *q) 287 { 288 switch (q->loss_model) { 289 case CLG_RANDOM: 290 /* Random packet drop 0 => none, ~0 => all */ 291 return q->loss && q->loss >= get_crandom(&q->loss_cor); 292 293 case CLG_4_STATES: 294 /* 4state loss model algorithm (used also for GI model) 295 * Extracts a value from the markov 4 state loss generator, 296 * if it is 1 drops a packet and if needed writes the event in 297 * the kernel logs 298 */ 299 return loss_4state(q); 300 301 case CLG_GILB_ELL: 302 /* Gilbert-Elliot loss model algorithm 303 * Extracts a value from the Gilbert-Elliot loss generator, 304 * if it is 1 drops a packet and if needed writes the event in 305 * the kernel logs 306 */ 307 return loss_gilb_ell(q); 308 } 309 310 return false; /* not reached */ 311 } 312 313 314 /* tabledist - return a pseudo-randomly distributed value with mean mu and 315 * std deviation sigma. Uses table lookup to approximate the desired 316 * distribution, and a uniformly-distributed pseudo-random source. 317 */ 318 static s64 tabledist(s64 mu, s32 sigma, 319 struct crndstate *state, 320 const struct disttable *dist) 321 { 322 s64 x; 323 long t; 324 u32 rnd; 325 326 if (sigma == 0) 327 return mu; 328 329 rnd = get_crandom(state); 330 331 /* default uniform distribution */ 332 if (dist == NULL) 333 return ((rnd % (2 * sigma)) + mu) - sigma; 334 335 t = dist->table[rnd % dist->size]; 336 x = (sigma % NETEM_DIST_SCALE) * t; 337 if (x >= 0) 338 x += NETEM_DIST_SCALE/2; 339 else 340 x -= NETEM_DIST_SCALE/2; 341 342 return x / NETEM_DIST_SCALE + (sigma / NETEM_DIST_SCALE) * t + mu; 343 } 344 345 static u64 packet_time_ns(u64 len, const struct netem_sched_data *q) 346 { 347 len += q->packet_overhead; 348 349 if (q->cell_size) { 350 u32 cells = reciprocal_divide(len, q->cell_size_reciprocal); 351 352 if (len > cells * q->cell_size) /* extra cell needed for remainder */ 353 cells++; 354 len = cells * (q->cell_size + q->cell_overhead); 355 } 356 357 return div64_u64(len * NSEC_PER_SEC, q->rate); 358 } 359 360 static void tfifo_reset(struct Qdisc *sch) 361 { 362 struct netem_sched_data *q = qdisc_priv(sch); 363 struct rb_node *p = rb_first(&q->t_root); 364 365 while (p) { 366 struct sk_buff *skb = rb_to_skb(p); 367 368 p = rb_next(p); 369 rb_erase(&skb->rbnode, &q->t_root); 370 rtnl_kfree_skbs(skb, skb); 371 } 372 } 373 374 static void tfifo_enqueue(struct sk_buff *nskb, struct Qdisc *sch) 375 { 376 struct netem_sched_data *q = qdisc_priv(sch); 377 u64 tnext = netem_skb_cb(nskb)->time_to_send; 378 struct rb_node **p = &q->t_root.rb_node, *parent = NULL; 379 380 while (*p) { 381 struct sk_buff *skb; 382 383 parent = *p; 384 skb = rb_to_skb(parent); 385 if (tnext >= netem_skb_cb(skb)->time_to_send) 386 p = &parent->rb_right; 387 else 388 p = &parent->rb_left; 389 } 390 rb_link_node(&nskb->rbnode, parent, p); 391 rb_insert_color(&nskb->rbnode, &q->t_root); 392 sch->q.qlen++; 393 } 394 395 /* netem can't properly corrupt a megapacket (like we get from GSO), so instead 396 * when we statistically choose to corrupt one, we instead segment it, returning 397 * the first packet to be corrupted, and re-enqueue the remaining frames 398 */ 399 static struct sk_buff *netem_segment(struct sk_buff *skb, struct Qdisc *sch, 400 struct sk_buff **to_free) 401 { 402 struct sk_buff *segs; 403 netdev_features_t features = netif_skb_features(skb); 404 405 segs = skb_gso_segment(skb, features & ~NETIF_F_GSO_MASK); 406 407 if (IS_ERR_OR_NULL(segs)) { 408 qdisc_drop(skb, sch, to_free); 409 return NULL; 410 } 411 consume_skb(skb); 412 return segs; 413 } 414 415 static void netem_enqueue_skb_head(struct qdisc_skb_head *qh, struct sk_buff *skb) 416 { 417 skb->next = qh->head; 418 419 if (!qh->head) 420 qh->tail = skb; 421 qh->head = skb; 422 qh->qlen++; 423 } 424 425 /* 426 * Insert one skb into qdisc. 427 * Note: parent depends on return value to account for queue length. 428 * NET_XMIT_DROP: queue length didn't change. 429 * NET_XMIT_SUCCESS: one skb was queued. 430 */ 431 static int netem_enqueue(struct sk_buff *skb, struct Qdisc *sch, 432 struct sk_buff **to_free) 433 { 434 struct netem_sched_data *q = qdisc_priv(sch); 435 /* We don't fill cb now as skb_unshare() may invalidate it */ 436 struct netem_skb_cb *cb; 437 struct sk_buff *skb2; 438 struct sk_buff *segs = NULL; 439 unsigned int len = 0, last_len, prev_len = qdisc_pkt_len(skb); 440 int nb = 0; 441 int count = 1; 442 int rc = NET_XMIT_SUCCESS; 443 444 /* Random duplication */ 445 if (q->duplicate && q->duplicate >= get_crandom(&q->dup_cor)) 446 ++count; 447 448 /* Drop packet? */ 449 if (loss_event(q)) { 450 if (q->ecn && INET_ECN_set_ce(skb)) 451 qdisc_qstats_drop(sch); /* mark packet */ 452 else 453 --count; 454 } 455 if (count == 0) { 456 qdisc_qstats_drop(sch); 457 __qdisc_drop(skb, to_free); 458 return NET_XMIT_SUCCESS | __NET_XMIT_BYPASS; 459 } 460 461 /* If a delay is expected, orphan the skb. (orphaning usually takes 462 * place at TX completion time, so _before_ the link transit delay) 463 */ 464 if (q->latency || q->jitter || q->rate) 465 skb_orphan_partial(skb); 466 467 /* 468 * If we need to duplicate packet, then re-insert at top of the 469 * qdisc tree, since parent queuer expects that only one 470 * skb will be queued. 471 */ 472 if (count > 1 && (skb2 = skb_clone(skb, GFP_ATOMIC)) != NULL) { 473 struct Qdisc *rootq = qdisc_root(sch); 474 u32 dupsave = q->duplicate; /* prevent duplicating a dup... */ 475 476 q->duplicate = 0; 477 rootq->enqueue(skb2, rootq, to_free); 478 q->duplicate = dupsave; 479 } 480 481 /* 482 * Randomized packet corruption. 483 * Make copy if needed since we are modifying 484 * If packet is going to be hardware checksummed, then 485 * do it now in software before we mangle it. 486 */ 487 if (q->corrupt && q->corrupt >= get_crandom(&q->corrupt_cor)) { 488 if (skb_is_gso(skb)) { 489 segs = netem_segment(skb, sch, to_free); 490 if (!segs) 491 return NET_XMIT_DROP; 492 } else { 493 segs = skb; 494 } 495 496 skb = segs; 497 segs = segs->next; 498 499 skb = skb_unshare(skb, GFP_ATOMIC); 500 if (unlikely(!skb)) { 501 qdisc_qstats_drop(sch); 502 goto finish_segs; 503 } 504 if (skb->ip_summed == CHECKSUM_PARTIAL && 505 skb_checksum_help(skb)) { 506 qdisc_drop(skb, sch, to_free); 507 goto finish_segs; 508 } 509 510 skb->data[prandom_u32() % skb_headlen(skb)] ^= 511 1<<(prandom_u32() % 8); 512 } 513 514 if (unlikely(sch->q.qlen >= sch->limit)) 515 return qdisc_drop_all(skb, sch, to_free); 516 517 qdisc_qstats_backlog_inc(sch, skb); 518 519 cb = netem_skb_cb(skb); 520 if (q->gap == 0 || /* not doing reordering */ 521 q->counter < q->gap - 1 || /* inside last reordering gap */ 522 q->reorder < get_crandom(&q->reorder_cor)) { 523 u64 now; 524 s64 delay; 525 526 delay = tabledist(q->latency, q->jitter, 527 &q->delay_cor, q->delay_dist); 528 529 now = ktime_get_ns(); 530 531 if (q->rate) { 532 struct netem_skb_cb *last = NULL; 533 534 if (sch->q.tail) 535 last = netem_skb_cb(sch->q.tail); 536 if (q->t_root.rb_node) { 537 struct sk_buff *t_skb; 538 struct netem_skb_cb *t_last; 539 540 t_skb = skb_rb_last(&q->t_root); 541 t_last = netem_skb_cb(t_skb); 542 if (!last || 543 t_last->time_to_send > last->time_to_send) { 544 last = t_last; 545 } 546 } 547 548 if (last) { 549 /* 550 * Last packet in queue is reference point (now), 551 * calculate this time bonus and subtract 552 * from delay. 553 */ 554 delay -= last->time_to_send - now; 555 delay = max_t(s64, 0, delay); 556 now = last->time_to_send; 557 } 558 559 delay += packet_time_ns(qdisc_pkt_len(skb), q); 560 } 561 562 cb->time_to_send = now + delay; 563 ++q->counter; 564 tfifo_enqueue(skb, sch); 565 } else { 566 /* 567 * Do re-ordering by putting one out of N packets at the front 568 * of the queue. 569 */ 570 cb->time_to_send = ktime_get_ns(); 571 q->counter = 0; 572 573 netem_enqueue_skb_head(&sch->q, skb); 574 sch->qstats.requeues++; 575 } 576 577 finish_segs: 578 if (segs) { 579 while (segs) { 580 skb2 = segs->next; 581 segs->next = NULL; 582 qdisc_skb_cb(segs)->pkt_len = segs->len; 583 last_len = segs->len; 584 rc = qdisc_enqueue(segs, sch, to_free); 585 if (rc != NET_XMIT_SUCCESS) { 586 if (net_xmit_drop_count(rc)) 587 qdisc_qstats_drop(sch); 588 } else { 589 nb++; 590 len += last_len; 591 } 592 segs = skb2; 593 } 594 sch->q.qlen += nb; 595 if (nb > 1) 596 qdisc_tree_reduce_backlog(sch, 1 - nb, prev_len - len); 597 } 598 return NET_XMIT_SUCCESS; 599 } 600 601 /* Delay the next round with a new future slot with a 602 * correct number of bytes and packets. 603 */ 604 605 static void get_slot_next(struct netem_sched_data *q, u64 now) 606 { 607 s64 next_delay; 608 609 if (!q->slot_dist) 610 next_delay = q->slot_config.min_delay + 611 (prandom_u32() * 612 (q->slot_config.max_delay - 613 q->slot_config.min_delay) >> 32); 614 else 615 next_delay = tabledist(q->slot_config.dist_delay, 616 (s32)(q->slot_config.dist_jitter), 617 NULL, q->slot_dist); 618 619 q->slot.slot_next = now + next_delay; 620 q->slot.packets_left = q->slot_config.max_packets; 621 q->slot.bytes_left = q->slot_config.max_bytes; 622 } 623 624 static struct sk_buff *netem_dequeue(struct Qdisc *sch) 625 { 626 struct netem_sched_data *q = qdisc_priv(sch); 627 struct sk_buff *skb; 628 struct rb_node *p; 629 630 tfifo_dequeue: 631 skb = __qdisc_dequeue_head(&sch->q); 632 if (skb) { 633 qdisc_qstats_backlog_dec(sch, skb); 634 deliver: 635 qdisc_bstats_update(sch, skb); 636 return skb; 637 } 638 p = rb_first(&q->t_root); 639 if (p) { 640 u64 time_to_send; 641 u64 now = ktime_get_ns(); 642 643 skb = rb_to_skb(p); 644 645 /* if more time remaining? */ 646 time_to_send = netem_skb_cb(skb)->time_to_send; 647 if (q->slot.slot_next && q->slot.slot_next < time_to_send) 648 get_slot_next(q, now); 649 650 if (time_to_send <= now && q->slot.slot_next <= now) { 651 rb_erase(p, &q->t_root); 652 sch->q.qlen--; 653 qdisc_qstats_backlog_dec(sch, skb); 654 skb->next = NULL; 655 skb->prev = NULL; 656 /* skb->dev shares skb->rbnode area, 657 * we need to restore its value. 658 */ 659 skb->dev = qdisc_dev(sch); 660 661 #ifdef CONFIG_NET_CLS_ACT 662 /* 663 * If it's at ingress let's pretend the delay is 664 * from the network (tstamp will be updated). 665 */ 666 if (skb->tc_redirected && skb->tc_from_ingress) 667 skb->tstamp = 0; 668 #endif 669 670 if (q->slot.slot_next) { 671 q->slot.packets_left--; 672 q->slot.bytes_left -= qdisc_pkt_len(skb); 673 if (q->slot.packets_left <= 0 || 674 q->slot.bytes_left <= 0) 675 get_slot_next(q, now); 676 } 677 678 if (q->qdisc) { 679 unsigned int pkt_len = qdisc_pkt_len(skb); 680 struct sk_buff *to_free = NULL; 681 int err; 682 683 err = qdisc_enqueue(skb, q->qdisc, &to_free); 684 kfree_skb_list(to_free); 685 if (err != NET_XMIT_SUCCESS && 686 net_xmit_drop_count(err)) { 687 qdisc_qstats_drop(sch); 688 qdisc_tree_reduce_backlog(sch, 1, 689 pkt_len); 690 } 691 goto tfifo_dequeue; 692 } 693 goto deliver; 694 } 695 696 if (q->qdisc) { 697 skb = q->qdisc->ops->dequeue(q->qdisc); 698 if (skb) 699 goto deliver; 700 } 701 702 qdisc_watchdog_schedule_ns(&q->watchdog, 703 max(time_to_send, 704 q->slot.slot_next)); 705 } 706 707 if (q->qdisc) { 708 skb = q->qdisc->ops->dequeue(q->qdisc); 709 if (skb) 710 goto deliver; 711 } 712 return NULL; 713 } 714 715 static void netem_reset(struct Qdisc *sch) 716 { 717 struct netem_sched_data *q = qdisc_priv(sch); 718 719 qdisc_reset_queue(sch); 720 tfifo_reset(sch); 721 if (q->qdisc) 722 qdisc_reset(q->qdisc); 723 qdisc_watchdog_cancel(&q->watchdog); 724 } 725 726 static void dist_free(struct disttable *d) 727 { 728 kvfree(d); 729 } 730 731 /* 732 * Distribution data is a variable size payload containing 733 * signed 16 bit values. 734 */ 735 736 static int get_dist_table(struct Qdisc *sch, struct disttable **tbl, 737 const struct nlattr *attr) 738 { 739 size_t n = nla_len(attr)/sizeof(__s16); 740 const __s16 *data = nla_data(attr); 741 spinlock_t *root_lock; 742 struct disttable *d; 743 int i; 744 745 if (n > NETEM_DIST_MAX) 746 return -EINVAL; 747 748 d = kvmalloc(sizeof(struct disttable) + n * sizeof(s16), GFP_KERNEL); 749 if (!d) 750 return -ENOMEM; 751 752 d->size = n; 753 for (i = 0; i < n; i++) 754 d->table[i] = data[i]; 755 756 root_lock = qdisc_root_sleeping_lock(sch); 757 758 spin_lock_bh(root_lock); 759 swap(*tbl, d); 760 spin_unlock_bh(root_lock); 761 762 dist_free(d); 763 return 0; 764 } 765 766 static void get_slot(struct netem_sched_data *q, const struct nlattr *attr) 767 { 768 const struct tc_netem_slot *c = nla_data(attr); 769 770 q->slot_config = *c; 771 if (q->slot_config.max_packets == 0) 772 q->slot_config.max_packets = INT_MAX; 773 if (q->slot_config.max_bytes == 0) 774 q->slot_config.max_bytes = INT_MAX; 775 q->slot.packets_left = q->slot_config.max_packets; 776 q->slot.bytes_left = q->slot_config.max_bytes; 777 if (q->slot_config.min_delay | q->slot_config.max_delay | 778 q->slot_config.dist_jitter) 779 q->slot.slot_next = ktime_get_ns(); 780 else 781 q->slot.slot_next = 0; 782 } 783 784 static void get_correlation(struct netem_sched_data *q, const struct nlattr *attr) 785 { 786 const struct tc_netem_corr *c = nla_data(attr); 787 788 init_crandom(&q->delay_cor, c->delay_corr); 789 init_crandom(&q->loss_cor, c->loss_corr); 790 init_crandom(&q->dup_cor, c->dup_corr); 791 } 792 793 static void get_reorder(struct netem_sched_data *q, const struct nlattr *attr) 794 { 795 const struct tc_netem_reorder *r = nla_data(attr); 796 797 q->reorder = r->probability; 798 init_crandom(&q->reorder_cor, r->correlation); 799 } 800 801 static void get_corrupt(struct netem_sched_data *q, const struct nlattr *attr) 802 { 803 const struct tc_netem_corrupt *r = nla_data(attr); 804 805 q->corrupt = r->probability; 806 init_crandom(&q->corrupt_cor, r->correlation); 807 } 808 809 static void get_rate(struct netem_sched_data *q, const struct nlattr *attr) 810 { 811 const struct tc_netem_rate *r = nla_data(attr); 812 813 q->rate = r->rate; 814 q->packet_overhead = r->packet_overhead; 815 q->cell_size = r->cell_size; 816 q->cell_overhead = r->cell_overhead; 817 if (q->cell_size) 818 q->cell_size_reciprocal = reciprocal_value(q->cell_size); 819 else 820 q->cell_size_reciprocal = (struct reciprocal_value) { 0 }; 821 } 822 823 static int get_loss_clg(struct netem_sched_data *q, const struct nlattr *attr) 824 { 825 const struct nlattr *la; 826 int rem; 827 828 nla_for_each_nested(la, attr, rem) { 829 u16 type = nla_type(la); 830 831 switch (type) { 832 case NETEM_LOSS_GI: { 833 const struct tc_netem_gimodel *gi = nla_data(la); 834 835 if (nla_len(la) < sizeof(struct tc_netem_gimodel)) { 836 pr_info("netem: incorrect gi model size\n"); 837 return -EINVAL; 838 } 839 840 q->loss_model = CLG_4_STATES; 841 842 q->clg.state = TX_IN_GAP_PERIOD; 843 q->clg.a1 = gi->p13; 844 q->clg.a2 = gi->p31; 845 q->clg.a3 = gi->p32; 846 q->clg.a4 = gi->p14; 847 q->clg.a5 = gi->p23; 848 break; 849 } 850 851 case NETEM_LOSS_GE: { 852 const struct tc_netem_gemodel *ge = nla_data(la); 853 854 if (nla_len(la) < sizeof(struct tc_netem_gemodel)) { 855 pr_info("netem: incorrect ge model size\n"); 856 return -EINVAL; 857 } 858 859 q->loss_model = CLG_GILB_ELL; 860 q->clg.state = GOOD_STATE; 861 q->clg.a1 = ge->p; 862 q->clg.a2 = ge->r; 863 q->clg.a3 = ge->h; 864 q->clg.a4 = ge->k1; 865 break; 866 } 867 868 default: 869 pr_info("netem: unknown loss type %u\n", type); 870 return -EINVAL; 871 } 872 } 873 874 return 0; 875 } 876 877 static const struct nla_policy netem_policy[TCA_NETEM_MAX + 1] = { 878 [TCA_NETEM_CORR] = { .len = sizeof(struct tc_netem_corr) }, 879 [TCA_NETEM_REORDER] = { .len = sizeof(struct tc_netem_reorder) }, 880 [TCA_NETEM_CORRUPT] = { .len = sizeof(struct tc_netem_corrupt) }, 881 [TCA_NETEM_RATE] = { .len = sizeof(struct tc_netem_rate) }, 882 [TCA_NETEM_LOSS] = { .type = NLA_NESTED }, 883 [TCA_NETEM_ECN] = { .type = NLA_U32 }, 884 [TCA_NETEM_RATE64] = { .type = NLA_U64 }, 885 [TCA_NETEM_LATENCY64] = { .type = NLA_S64 }, 886 [TCA_NETEM_JITTER64] = { .type = NLA_S64 }, 887 [TCA_NETEM_SLOT] = { .len = sizeof(struct tc_netem_slot) }, 888 }; 889 890 static int parse_attr(struct nlattr *tb[], int maxtype, struct nlattr *nla, 891 const struct nla_policy *policy, int len) 892 { 893 int nested_len = nla_len(nla) - NLA_ALIGN(len); 894 895 if (nested_len < 0) { 896 pr_info("netem: invalid attributes len %d\n", nested_len); 897 return -EINVAL; 898 } 899 900 if (nested_len >= nla_attr_size(0)) 901 return nla_parse(tb, maxtype, nla_data(nla) + NLA_ALIGN(len), 902 nested_len, policy, NULL); 903 904 memset(tb, 0, sizeof(struct nlattr *) * (maxtype + 1)); 905 return 0; 906 } 907 908 /* Parse netlink message to set options */ 909 static int netem_change(struct Qdisc *sch, struct nlattr *opt, 910 struct netlink_ext_ack *extack) 911 { 912 struct netem_sched_data *q = qdisc_priv(sch); 913 struct nlattr *tb[TCA_NETEM_MAX + 1]; 914 struct tc_netem_qopt *qopt; 915 struct clgstate old_clg; 916 int old_loss_model = CLG_RANDOM; 917 int ret; 918 919 if (opt == NULL) 920 return -EINVAL; 921 922 qopt = nla_data(opt); 923 ret = parse_attr(tb, TCA_NETEM_MAX, opt, netem_policy, sizeof(*qopt)); 924 if (ret < 0) 925 return ret; 926 927 /* backup q->clg and q->loss_model */ 928 old_clg = q->clg; 929 old_loss_model = q->loss_model; 930 931 if (tb[TCA_NETEM_LOSS]) { 932 ret = get_loss_clg(q, tb[TCA_NETEM_LOSS]); 933 if (ret) { 934 q->loss_model = old_loss_model; 935 return ret; 936 } 937 } else { 938 q->loss_model = CLG_RANDOM; 939 } 940 941 if (tb[TCA_NETEM_DELAY_DIST]) { 942 ret = get_dist_table(sch, &q->delay_dist, 943 tb[TCA_NETEM_DELAY_DIST]); 944 if (ret) 945 goto get_table_failure; 946 } 947 948 if (tb[TCA_NETEM_SLOT_DIST]) { 949 ret = get_dist_table(sch, &q->slot_dist, 950 tb[TCA_NETEM_SLOT_DIST]); 951 if (ret) 952 goto get_table_failure; 953 } 954 955 sch->limit = qopt->limit; 956 957 q->latency = PSCHED_TICKS2NS(qopt->latency); 958 q->jitter = PSCHED_TICKS2NS(qopt->jitter); 959 q->limit = qopt->limit; 960 q->gap = qopt->gap; 961 q->counter = 0; 962 q->loss = qopt->loss; 963 q->duplicate = qopt->duplicate; 964 965 /* for compatibility with earlier versions. 966 * if gap is set, need to assume 100% probability 967 */ 968 if (q->gap) 969 q->reorder = ~0; 970 971 if (tb[TCA_NETEM_CORR]) 972 get_correlation(q, tb[TCA_NETEM_CORR]); 973 974 if (tb[TCA_NETEM_REORDER]) 975 get_reorder(q, tb[TCA_NETEM_REORDER]); 976 977 if (tb[TCA_NETEM_CORRUPT]) 978 get_corrupt(q, tb[TCA_NETEM_CORRUPT]); 979 980 if (tb[TCA_NETEM_RATE]) 981 get_rate(q, tb[TCA_NETEM_RATE]); 982 983 if (tb[TCA_NETEM_RATE64]) 984 q->rate = max_t(u64, q->rate, 985 nla_get_u64(tb[TCA_NETEM_RATE64])); 986 987 if (tb[TCA_NETEM_LATENCY64]) 988 q->latency = nla_get_s64(tb[TCA_NETEM_LATENCY64]); 989 990 if (tb[TCA_NETEM_JITTER64]) 991 q->jitter = nla_get_s64(tb[TCA_NETEM_JITTER64]); 992 993 if (tb[TCA_NETEM_ECN]) 994 q->ecn = nla_get_u32(tb[TCA_NETEM_ECN]); 995 996 if (tb[TCA_NETEM_SLOT]) 997 get_slot(q, tb[TCA_NETEM_SLOT]); 998 999 return ret; 1000 1001 get_table_failure: 1002 /* recover clg and loss_model, in case of 1003 * q->clg and q->loss_model were modified 1004 * in get_loss_clg() 1005 */ 1006 q->clg = old_clg; 1007 q->loss_model = old_loss_model; 1008 return ret; 1009 } 1010 1011 static int netem_init(struct Qdisc *sch, struct nlattr *opt, 1012 struct netlink_ext_ack *extack) 1013 { 1014 struct netem_sched_data *q = qdisc_priv(sch); 1015 int ret; 1016 1017 qdisc_watchdog_init(&q->watchdog, sch); 1018 1019 if (!opt) 1020 return -EINVAL; 1021 1022 q->loss_model = CLG_RANDOM; 1023 ret = netem_change(sch, opt, extack); 1024 if (ret) 1025 pr_info("netem: change failed\n"); 1026 return ret; 1027 } 1028 1029 static void netem_destroy(struct Qdisc *sch) 1030 { 1031 struct netem_sched_data *q = qdisc_priv(sch); 1032 1033 qdisc_watchdog_cancel(&q->watchdog); 1034 if (q->qdisc) 1035 qdisc_destroy(q->qdisc); 1036 dist_free(q->delay_dist); 1037 dist_free(q->slot_dist); 1038 } 1039 1040 static int dump_loss_model(const struct netem_sched_data *q, 1041 struct sk_buff *skb) 1042 { 1043 struct nlattr *nest; 1044 1045 nest = nla_nest_start(skb, TCA_NETEM_LOSS); 1046 if (nest == NULL) 1047 goto nla_put_failure; 1048 1049 switch (q->loss_model) { 1050 case CLG_RANDOM: 1051 /* legacy loss model */ 1052 nla_nest_cancel(skb, nest); 1053 return 0; /* no data */ 1054 1055 case CLG_4_STATES: { 1056 struct tc_netem_gimodel gi = { 1057 .p13 = q->clg.a1, 1058 .p31 = q->clg.a2, 1059 .p32 = q->clg.a3, 1060 .p14 = q->clg.a4, 1061 .p23 = q->clg.a5, 1062 }; 1063 1064 if (nla_put(skb, NETEM_LOSS_GI, sizeof(gi), &gi)) 1065 goto nla_put_failure; 1066 break; 1067 } 1068 case CLG_GILB_ELL: { 1069 struct tc_netem_gemodel ge = { 1070 .p = q->clg.a1, 1071 .r = q->clg.a2, 1072 .h = q->clg.a3, 1073 .k1 = q->clg.a4, 1074 }; 1075 1076 if (nla_put(skb, NETEM_LOSS_GE, sizeof(ge), &ge)) 1077 goto nla_put_failure; 1078 break; 1079 } 1080 } 1081 1082 nla_nest_end(skb, nest); 1083 return 0; 1084 1085 nla_put_failure: 1086 nla_nest_cancel(skb, nest); 1087 return -1; 1088 } 1089 1090 static int netem_dump(struct Qdisc *sch, struct sk_buff *skb) 1091 { 1092 const struct netem_sched_data *q = qdisc_priv(sch); 1093 struct nlattr *nla = (struct nlattr *) skb_tail_pointer(skb); 1094 struct tc_netem_qopt qopt; 1095 struct tc_netem_corr cor; 1096 struct tc_netem_reorder reorder; 1097 struct tc_netem_corrupt corrupt; 1098 struct tc_netem_rate rate; 1099 struct tc_netem_slot slot; 1100 1101 qopt.latency = min_t(psched_tdiff_t, PSCHED_NS2TICKS(q->latency), 1102 UINT_MAX); 1103 qopt.jitter = min_t(psched_tdiff_t, PSCHED_NS2TICKS(q->jitter), 1104 UINT_MAX); 1105 qopt.limit = q->limit; 1106 qopt.loss = q->loss; 1107 qopt.gap = q->gap; 1108 qopt.duplicate = q->duplicate; 1109 if (nla_put(skb, TCA_OPTIONS, sizeof(qopt), &qopt)) 1110 goto nla_put_failure; 1111 1112 if (nla_put(skb, TCA_NETEM_LATENCY64, sizeof(q->latency), &q->latency)) 1113 goto nla_put_failure; 1114 1115 if (nla_put(skb, TCA_NETEM_JITTER64, sizeof(q->jitter), &q->jitter)) 1116 goto nla_put_failure; 1117 1118 cor.delay_corr = q->delay_cor.rho; 1119 cor.loss_corr = q->loss_cor.rho; 1120 cor.dup_corr = q->dup_cor.rho; 1121 if (nla_put(skb, TCA_NETEM_CORR, sizeof(cor), &cor)) 1122 goto nla_put_failure; 1123 1124 reorder.probability = q->reorder; 1125 reorder.correlation = q->reorder_cor.rho; 1126 if (nla_put(skb, TCA_NETEM_REORDER, sizeof(reorder), &reorder)) 1127 goto nla_put_failure; 1128 1129 corrupt.probability = q->corrupt; 1130 corrupt.correlation = q->corrupt_cor.rho; 1131 if (nla_put(skb, TCA_NETEM_CORRUPT, sizeof(corrupt), &corrupt)) 1132 goto nla_put_failure; 1133 1134 if (q->rate >= (1ULL << 32)) { 1135 if (nla_put_u64_64bit(skb, TCA_NETEM_RATE64, q->rate, 1136 TCA_NETEM_PAD)) 1137 goto nla_put_failure; 1138 rate.rate = ~0U; 1139 } else { 1140 rate.rate = q->rate; 1141 } 1142 rate.packet_overhead = q->packet_overhead; 1143 rate.cell_size = q->cell_size; 1144 rate.cell_overhead = q->cell_overhead; 1145 if (nla_put(skb, TCA_NETEM_RATE, sizeof(rate), &rate)) 1146 goto nla_put_failure; 1147 1148 if (q->ecn && nla_put_u32(skb, TCA_NETEM_ECN, q->ecn)) 1149 goto nla_put_failure; 1150 1151 if (dump_loss_model(q, skb) != 0) 1152 goto nla_put_failure; 1153 1154 if (q->slot_config.min_delay | q->slot_config.max_delay | 1155 q->slot_config.dist_jitter) { 1156 slot = q->slot_config; 1157 if (slot.max_packets == INT_MAX) 1158 slot.max_packets = 0; 1159 if (slot.max_bytes == INT_MAX) 1160 slot.max_bytes = 0; 1161 if (nla_put(skb, TCA_NETEM_SLOT, sizeof(slot), &slot)) 1162 goto nla_put_failure; 1163 } 1164 1165 return nla_nest_end(skb, nla); 1166 1167 nla_put_failure: 1168 nlmsg_trim(skb, nla); 1169 return -1; 1170 } 1171 1172 static int netem_dump_class(struct Qdisc *sch, unsigned long cl, 1173 struct sk_buff *skb, struct tcmsg *tcm) 1174 { 1175 struct netem_sched_data *q = qdisc_priv(sch); 1176 1177 if (cl != 1 || !q->qdisc) /* only one class */ 1178 return -ENOENT; 1179 1180 tcm->tcm_handle |= TC_H_MIN(1); 1181 tcm->tcm_info = q->qdisc->handle; 1182 1183 return 0; 1184 } 1185 1186 static int netem_graft(struct Qdisc *sch, unsigned long arg, struct Qdisc *new, 1187 struct Qdisc **old, struct netlink_ext_ack *extack) 1188 { 1189 struct netem_sched_data *q = qdisc_priv(sch); 1190 1191 *old = qdisc_replace(sch, new, &q->qdisc); 1192 return 0; 1193 } 1194 1195 static struct Qdisc *netem_leaf(struct Qdisc *sch, unsigned long arg) 1196 { 1197 struct netem_sched_data *q = qdisc_priv(sch); 1198 return q->qdisc; 1199 } 1200 1201 static unsigned long netem_find(struct Qdisc *sch, u32 classid) 1202 { 1203 return 1; 1204 } 1205 1206 static void netem_walk(struct Qdisc *sch, struct qdisc_walker *walker) 1207 { 1208 if (!walker->stop) { 1209 if (walker->count >= walker->skip) 1210 if (walker->fn(sch, 1, walker) < 0) { 1211 walker->stop = 1; 1212 return; 1213 } 1214 walker->count++; 1215 } 1216 } 1217 1218 static const struct Qdisc_class_ops netem_class_ops = { 1219 .graft = netem_graft, 1220 .leaf = netem_leaf, 1221 .find = netem_find, 1222 .walk = netem_walk, 1223 .dump = netem_dump_class, 1224 }; 1225 1226 static struct Qdisc_ops netem_qdisc_ops __read_mostly = { 1227 .id = "netem", 1228 .cl_ops = &netem_class_ops, 1229 .priv_size = sizeof(struct netem_sched_data), 1230 .enqueue = netem_enqueue, 1231 .dequeue = netem_dequeue, 1232 .peek = qdisc_peek_dequeued, 1233 .init = netem_init, 1234 .reset = netem_reset, 1235 .destroy = netem_destroy, 1236 .change = netem_change, 1237 .dump = netem_dump, 1238 .owner = THIS_MODULE, 1239 }; 1240 1241 1242 static int __init netem_module_init(void) 1243 { 1244 pr_info("netem: version " VERSION "\n"); 1245 return register_qdisc(&netem_qdisc_ops); 1246 } 1247 static void __exit netem_module_exit(void) 1248 { 1249 unregister_qdisc(&netem_qdisc_ops); 1250 } 1251 module_init(netem_module_init) 1252 module_exit(netem_module_exit) 1253 MODULE_LICENSE("GPL"); 1254