virnetdevbandwidth.c 24.6 KB
Newer Older
1
/*
2
 * Copyright (C) 2009-2015 Red Hat, Inc.
3 4 5 6 7 8 9 10 11 12 13 14
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
15
 * License along with this library.  If not, see
O
Osier Yang 已提交
16
 * <http://www.gnu.org/licenses/>.
17 18 19
 */

#include <config.h>
20
#include <unistd.h>
21 22

#include "virnetdevbandwidth.h"
23
#include "vircommand.h"
24
#include "viralloc.h"
25
#include "virerror.h"
26
#include "virstring.h"
27
#include "virutil.h"
28 29 30 31 32 33 34 35 36 37 38 39 40 41

#define VIR_FROM_THIS VIR_FROM_NONE

void
virNetDevBandwidthFree(virNetDevBandwidthPtr def)
{
    if (!def)
        return;

    VIR_FREE(def->in);
    VIR_FREE(def->out);
    VIR_FREE(def);
}

42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64
static void
virNetDevBandwidthCmdAddOptimalQuantum(virCommandPtr cmd,
                                       const virNetDevBandwidthRate *rate)
{
    const unsigned long long mtu = 1500;
    unsigned long long r2q;

    /* When two or more classes compete for unused bandwidth they are each
     * given some number of bytes before serving other competing class. This
     * number is called quantum. It's advised in HTB docs that the number
     * should be equal to MTU. The class quantum is computed from its rate
     * divided by global r2q parameter. However, if rate is too small the
     * default value will not suffice and thus we must provide our own value.
     * */

    r2q = rate->average * 1024 / 8 / mtu;
    if (!r2q)
        r2q = 1;

    virCommandAddArg(cmd, "quantum");
    virCommandAddArgFormat(cmd, "%llu", r2q);
}

65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116
/**
 * virNetDevBandwidthManipulateFilter:
 * @ifname: interface to operate on
 * @ifmac_ptr: MAC of the interface to create filter over
 * @id: filter ID
 * @class_id: where to place traffic
 * @remove_old: whether to remove the filter
 * @create_new: whether to create the filter
 *
 * TC filters are as crucial for traffic shaping as QDiscs. While
 * QDiscs act like black boxes deciding which packets should be
 * held up and which should be sent immediately, it's the filter
 * that places a packet into the box. So, we may end up
 * constructing a set of filters on a single device (e.g. a
 * bridge) and filter the traffic into QDiscs based on the
 * originating vNET device.
 *
 * Long story short, @ifname is the interface where the filter
 * should be created. The @ifmac_ptr is the MAC address for which
 * the filter should be created (usually different to the MAC
 * address of @ifname). Then, like everything - even filters have
 * an @id which should be unique (per @ifname). And @class_id
 * tells into which QDisc should filter place the traffic.
 *
 * This function can be used for both, removing stale filter
 * (@remove_old set to true) and creating new one (@create_new
 * set to true). Both at once for the same price!
 *
 * Returns: 0 on success,
 *         -1 otherwise (with error reported).
 */
static int ATTRIBUTE_NONNULL(1)
virNetDevBandwidthManipulateFilter(const char *ifname,
                                   const virMacAddr *ifmac_ptr,
                                   unsigned int id,
                                   const char *class_id,
                                   bool remove_old,
                                   bool create_new)
{
    int ret = -1;
    char *filter_id = NULL;
    virCommandPtr cmd = NULL;
    unsigned char ifmac[VIR_MAC_BUFLEN];
    char *mac[2] = {NULL, NULL};

    if (!(remove_old || create_new)) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("filter creation API error"));
        goto cleanup;
    }

    /* u32 filters must have 800:: prefix. Don't ask. */
117
    filter_id = g_strdup_printf("800::%u", id);
118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133

    if (remove_old) {
        int cmd_ret = 0;

        cmd = virCommandNew(TC);
        virCommandAddArgList(cmd, "filter", "del", "dev", ifname,
                             "prio", "2", "handle",  filter_id, "u32", NULL);

        if (virCommandRun(cmd, &cmd_ret) < 0)
            goto cleanup;

    }

    if (create_new) {
        virMacAddrGetRaw(ifmac_ptr, ifmac);

134 135 136
        mac[0] = g_strdup_printf("0x%02x%02x%02x%02x", ifmac[2],
                                 ifmac[3], ifmac[4], ifmac[5]);
        mac[1] = g_strdup_printf("0x%02x%02x", ifmac[0], ifmac[1]);
137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164

        virCommandFree(cmd);
        cmd = virCommandNew(TC);
        /* Okay, this not nice. But since libvirt does not necessarily track
         * interface IP address(es), and tc fw filter simply refuse to use
         * ebtables marks, we need to use u32 selector to match MAC address.
         * If libvirt will ever know something, remove this FIXME
         */
        virCommandAddArgList(cmd, "filter", "add", "dev", ifname, "protocol", "ip",
                             "prio", "2", "handle", filter_id, "u32",
                             "match", "u16", "0x0800", "0xffff", "at", "-2",
                             "match", "u32", mac[0], "0xffffffff", "at", "-12",
                             "match", "u16", mac[1], "0xffff", "at", "-14",
                             "flowid", class_id, NULL);

        if (virCommandRun(cmd, NULL) < 0)
            goto cleanup;
    }

    ret = 0;
 cleanup:
    VIR_FREE(mac[1]);
    VIR_FREE(mac[0]);
    VIR_FREE(filter_id);
    virCommandFree(cmd);
    return ret;
}

165 166 167 168 169

/**
 * virNetDevBandwidthSet:
 * @ifname: on which interface
 * @bandwidth: rates to set (may be NULL)
170
 * @hierarchical_class: whether to create hierarchical class
171
 * @swapped: true if IN/OUT should be set contrariwise
172 173 174 175
 *
 * This function enables QoS on specified interface
 * and set given traffic limits for both, incoming
 * and outgoing traffic. Any previous setting get
176 177 178
 * overwritten. If @hierarchical_class is TRUE, create
 * hierarchical class. It is used to guarantee minimal
 * throughput ('floor' attribute in NIC).
179
 *
180 181 182 183 184 185 186
 * If @swapped is set, the IN part of @bandwidth is set on
 * @ifname's TX, and vice versa. If it is not set, IN is set on
 * RX and OUT on TX. This is because for some types of interfaces
 * domain and the host live on the same side of the interface (so
 * domain's RX/TX is host's RX/TX), and for some it's swapped
 * (domain's RX/TX is hosts's TX/RX).
 *
187 188 189 190
 * Return 0 on success, -1 otherwise.
 */
int
virNetDevBandwidthSet(const char *ifname,
191
                      const virNetDevBandwidth *bandwidth,
192 193
                      bool hierarchical_class,
                      bool swapped)
194 195
{
    int ret = -1;
196
    virNetDevBandwidthRatePtr rx = NULL, tx = NULL; /* From domain POV */
197 198 199 200 201 202 203 204 205 206 207
    virCommandPtr cmd = NULL;
    char *average = NULL;
    char *peak = NULL;
    char *burst = NULL;

    if (!bandwidth) {
        /* nothing to be enabled */
        ret = 0;
        goto cleanup;
    }

208 209 210 211 212 213 214
    if (geteuid() != 0) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED, "%s",
                       _("Network bandwidth tuning is not available"
                         " in session mode"));
        return -1;
    }

215 216 217 218 219 220 221
    if (!ifname) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED, "%s",
                       _("Unable to set bandwidth for interface because "
                         "device name is unknown"));
        return -1;
    }

222 223 224 225 226 227 228 229
    if (swapped) {
        rx = bandwidth->out;
        tx = bandwidth->in;
    } else {
        rx = bandwidth->in;
        tx = bandwidth->out;
    }

230
    virNetDevBandwidthClear(ifname);
231

232
    if (tx && tx->average) {
233 234 235 236 237
        average = g_strdup_printf("%llukbps", tx->average);
        if (tx->peak)
            peak = g_strdup_printf("%llukbps", tx->peak);
        if (tx->burst)
            burst = g_strdup_printf("%llukb", tx->burst);
238 239 240

        cmd = virCommandNew(TC);
        virCommandAddArgList(cmd, "qdisc", "add", "dev", ifname, "root",
241 242
                             "handle", "1:", "htb", "default",
                             hierarchical_class ? "2" : "1", NULL);
243 244 245
        if (virCommandRun(cmd, NULL) < 0)
            goto cleanup;

246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273
        /* If we are creating a hierarchical class, all non guaranteed traffic
         * goes to the 1:2 class which will adjust 'rate' dynamically as NICs
         * with guaranteed throughput are plugged and unplugged. Class 1:1
         * exists so we don't exceed the maximum limit for the network. For each
         * NIC with guaranteed throughput a separate classid will be created.
         * NB '1:' is just a shorter notation of '1:0'.
         *
         * To get a picture how this works:
         *
         * +-----+     +---------+     +-----------+      +-----------+     +-----+
         * |     |     |  qdisc  |     | class 1:1 |      | class 1:2 |     |     |
         * | NIC |     | def 1:2 |     |   rate    |      |   rate    |     | sfq |
         * |     | --> |         | --> |   peak    | -+-> |   peak    | --> |     |
         * +-----+     +---------+     +-----------+  |   +-----------+     +-----+
         *                                            |
         *                                            |   +-----------+     +-----+
         *                                            |   | class 1:3 |     |     |
         *                                            |   |   rate    |     | sfq |
         *                                            +-> |   peak    | --> |     |
         *                                            |   +-----------+     +-----+
         *                                           ...
         *                                            |   +-----------+     +-----+
         *                                            |   | class 1:n |     |     |
         *                                            |   |   rate    |     | sfq |
         *                                            +-> |   peak    | --> |     |
         *                                                +-----------+     +-----+
         *
         * After the routing decision, when is it clear a packet is to be sent
274
         * via a particular NIC, it is sent to the root qdisc (queuing
275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314
         * discipline). In this case HTB (Hierarchical Token Bucket). It has
         * only one direct child class (with id 1:1) which shapes the overall
         * rate that is sent through the NIC.  This class has at least one child
         * (1:2) which is meant for all non-privileged (non guaranteed) traffic
         * from all domains. Then, for each interface with guaranteed
         * throughput, a separate class (1:n) is created. Imagine a class is a
         * box. Whenever a packet ends up in a class it is stored in this box
         * until the kernel sends it, then it is removed from box. Packets are
         * placed into boxes based on rules (filters) - e.g. depending on
         * destination IP/MAC address. If there is no rule to be applied, the
         * root qdisc has a default where such packets go (1:2 in this case).
         * Packets come in over and over again and boxes get filled more and
         * more. Imagine that kernel sends packets just once a second. So it
         * starts to traverse through this tree. It starts with the root qdisc
         * and through 1:1 it gets to 1:2. It sends packets up to 1:2's 'rate'.
         * Then it moves to 1:3 and again sends packets up to 1:3's 'rate'.  The
         * whole process is repeated until 1:n is processed. So now we have
         * ensured each class its guaranteed bandwidth. If the sum of sent data
         * doesn't exceed the 'rate' in 1:1 class, we can go further and send
         * more packets. The rest of available bandwidth is distributed to the
         * 1:2,1:3...1:n classes by ratio of their 'rate'. As soon as the root
         * 'rate' limit is reached or there are no more packets to send, we stop
         * sending and wait another second. Each class has an SFQ qdisc which
         * shuffles packets in boxes stochastically, so one sender cannot
         * starve others.
         *
         * Therefore, whenever we want to plug in a new guaranteed interface, we
         * need to create a new class and adjust the 'rate' of the 1:2 class.
         * When unplugging we do the exact opposite - remove the associated
         * class, and adjust the 'rate'.
         *
         * This description is rather long, but it is still a good idea to read
         * it before you dig into the code.
         */
        if (hierarchical_class) {
            virCommandFree(cmd);
            cmd = virCommandNew(TC);
            virCommandAddArgList(cmd, "class", "add", "dev", ifname, "parent",
                                 "1:", "classid", "1:1", "htb", "rate", average,
                                 "ceil", peak ? peak : average, NULL);
315
            virNetDevBandwidthCmdAddOptimalQuantum(cmd, tx);
316 317 318
            if (virCommandRun(cmd, NULL) < 0)
                goto cleanup;
        }
319 320
        virCommandFree(cmd);
        cmd = virCommandNew(TC);
E
Eric Blake 已提交
321
        virCommandAddArgList(cmd, "class", "add", "dev", ifname, "parent",
322 323 324
                             hierarchical_class ? "1:1" : "1:", "classid",
                             hierarchical_class ? "1:2" : "1:1", "htb",
                             "rate", average, NULL);
325 326 327 328 329 330

        if (peak)
            virCommandAddArgList(cmd, "ceil", peak, NULL);
        if (burst)
            virCommandAddArgList(cmd, "burst", burst, NULL);

331
        virNetDevBandwidthCmdAddOptimalQuantum(cmd, tx);
332 333 334
        if (virCommandRun(cmd, NULL) < 0)
            goto cleanup;

335 336 337
        virCommandFree(cmd);
        cmd = virCommandNew(TC);
        virCommandAddArgList(cmd, "qdisc", "add", "dev", ifname, "parent",
338 339
                             hierarchical_class ? "1:2" : "1:1",
                             "handle", "2:", "sfq", "perturb",
340 341 342 343 344
                             "10", NULL);

        if (virCommandRun(cmd, NULL) < 0)
            goto cleanup;

345 346
        virCommandFree(cmd);
        cmd = virCommandNew(TC);
E
Eric Blake 已提交
347
        virCommandAddArgList(cmd, "filter", "add", "dev", ifname, "parent",
348 349
                             "1:0", "protocol", "all", "prio", "1", "handle",
                             "1", "fw", "flowid", "1", NULL);
350 351 352 353 354 355 356 357 358

        if (virCommandRun(cmd, NULL) < 0)
            goto cleanup;

        VIR_FREE(average);
        VIR_FREE(peak);
        VIR_FREE(burst);
    }

359
    if (rx) {
360 361
        average = g_strdup_printf("%llukbps", rx->average);
        burst = g_strdup_printf("%llukb", rx->burst ? rx->burst : rx->average);
362 363 364 365 366 367 368 369 370 371 372

        virCommandFree(cmd);
        cmd = virCommandNew(TC);
            virCommandAddArgList(cmd, "qdisc", "add", "dev", ifname,
                                 "ingress", NULL);

        if (virCommandRun(cmd, NULL) < 0)
            goto cleanup;

        virCommandFree(cmd);
        cmd = virCommandNew(TC);
373
        /* Set filter to match all ingress traffic */
374
        virCommandAddArgList(cmd, "filter", "add", "dev", ifname, "parent",
375 376
                             "ffff:", "protocol", "all", "u32", "match", "u32",
                             "0", "0", "police", "rate", average,
377
                             "burst", burst, "mtu", "64kb", "drop", "flowid",
378 379 380 381 382 383 384 385
                             ":1", NULL);

        if (virCommandRun(cmd, NULL) < 0)
            goto cleanup;
    }

    ret = 0;

386
 cleanup:
387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407
    virCommandFree(cmd);
    VIR_FREE(average);
    VIR_FREE(peak);
    VIR_FREE(burst);
    return ret;
}

/**
 * virNetDevBandwidthClear:
 * @ifname: on which interface
 *
 * This function tries to disable QoS on specified interface
 * by deleting root and ingress qdisc. However, this may fail
 * if we try to remove the default one.
 *
 * Return 0 on success, -1 otherwise.
 */
int
virNetDevBandwidthClear(const char *ifname)
{
    int ret = 0;
408
    int dummy; /* for ignoring the exit status */
409 410
    virCommandPtr cmd = NULL;

411 412 413
    if (!ifname)
       return 0;

414 415 416
    cmd = virCommandNew(TC);
    virCommandAddArgList(cmd, "qdisc", "del", "dev", ifname, "root", NULL);

417
    if (virCommandRun(cmd, &dummy) < 0)
418 419 420 421 422 423 424
        ret = -1;

    virCommandFree(cmd);

    cmd = virCommandNew(TC);
    virCommandAddArgList(cmd, "qdisc",  "del", "dev", ifname, "ingress", NULL);

425
    if (virCommandRun(cmd, &dummy) < 0)
426
        ret = -1;
427

428 429 430 431 432 433 434 435 436 437 438 439 440 441 442
    virCommandFree(cmd);

    return ret;
}

/*
 * virNetDevBandwidthCopy:
 * @dest: destination
 * @src:  source (may be NULL)
 *
 * Returns -1 on OOM error (which gets reported),
 * 0 otherwise.
 */
int
virNetDevBandwidthCopy(virNetDevBandwidthPtr *dest,
443
                       const virNetDevBandwidth *src)
444 445 446 447 448 449 450 451 452
{
    int ret = -1;

    *dest = NULL;
    if (!src) {
        /* nothing to be copied */
        return 0;
    }

453
    if (VIR_ALLOC(*dest) < 0)
454 455 456
        goto cleanup;

    if (src->in) {
457
        if (VIR_ALLOC((*dest)->in) < 0)
458 459 460 461 462 463 464 465 466 467 468 469 470 471
            goto cleanup;
        memcpy((*dest)->in, src->in, sizeof(*src->in));
    }

    if (src->out) {
        if (VIR_ALLOC((*dest)->out) < 0) {
            VIR_FREE((*dest)->in);
            goto cleanup;
        }
        memcpy((*dest)->out, src->out, sizeof(*src->out));
    }

    ret = 0;

472
 cleanup:
473 474 475 476 477 478 479 480
    if (ret < 0) {
        virNetDevBandwidthFree(*dest);
        *dest = NULL;
    }
    return ret;
}

bool
481 482
virNetDevBandwidthEqual(const virNetDevBandwidth *a,
                        const virNetDevBandwidth *b)
483
{
M
Michal Privoznik 已提交
484 485
    if (!a && !b)
        return true;
486

M
Michal Privoznik 已提交
487 488
    if (!a || !b)
        return false;
489

M
Michal Privoznik 已提交
490
    /* in */
491 492 493 494 495 496
    if (a->in) {
        if (!b->in)
            return false;

        if (a->in->average != b->in->average ||
            a->in->peak != b->in->peak ||
497
            a->in->floor != b->in->floor ||
498 499 500
            a->in->burst != b->in->burst)
            return false;
    } else if (b->in) {
M
Michal Privoznik 已提交
501
        return false;
502
    }
503

J
Ján Tomko 已提交
504
    /* out */
505 506 507 508 509 510
    if (a->out) {
        if (!b->out)
            return false;

        if (a->out->average != b->out->average ||
            a->out->peak != b->out->peak ||
511
            a->out->floor != b->out->floor ||
512 513 514
            a->out->burst != b->out->burst)
            return false;
    } else if (b->out) {
M
Michal Privoznik 已提交
515
        return false;
516
    }
517

M
Michal Privoznik 已提交
518
    return true;
519
}
520 521 522 523 524

/*
 * virNetDevBandwidthPlug:
 * @brname: name of the bridge
 * @net_bandwidth: QoS settings on @brname
525
 * @ifmac_ptr: MAC of interface
526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543
 * @bandwidth: QoS settings for interface
 * @id: unique ID (MUST be greater than 2)
 *
 * Set bridge part of interface QoS settings, e.g. guaranteed
 * bandwidth.  @id is an unique ID (among @brname) from which
 * other identifiers for class, qdisc and filter are derived.
 * However, two classes were already set up (by
 * virNetDevBandwidthSet). That's why this @id MUST be greater
 * than 2. You may want to keep passed @id, as it is used later
 * by virNetDevBandwidthUnplug.
 *
 * Returns:
 * 0 if QoS set successfully
 * -1 otherwise.
 */
int
virNetDevBandwidthPlug(const char *brname,
                       virNetDevBandwidthPtr net_bandwidth,
544
                       const virMacAddr *ifmac_ptr,
545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570
                       virNetDevBandwidthPtr bandwidth,
                       unsigned int id)
{
    int ret = -1;
    virCommandPtr cmd = NULL;
    char *class_id = NULL;
    char *qdisc_id = NULL;
    char *floor = NULL;
    char *ceil = NULL;
    char ifmacStr[VIR_MAC_STRING_BUFLEN];

    if (id <= 2) {
        virReportError(VIR_ERR_INTERNAL_ERROR, _("Invalid class ID %d"), id);
        return -1;
    }

    virMacAddrFormat(ifmac_ptr, ifmacStr);

    if (!net_bandwidth || !net_bandwidth->in) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Bridge '%s' has no QoS set, therefore "
                         "unable to set 'floor' on '%s'"),
                       brname, ifmacStr);
        return -1;
    }

571 572 573 574 575 576
    class_id = g_strdup_printf("1:%x", id);
    qdisc_id = g_strdup_printf("%x:", id);
    floor = g_strdup_printf("%llukbps", bandwidth->in->floor);
    ceil = g_strdup_printf("%llukbps", net_bandwidth->in->peak ?
                           net_bandwidth->in->peak :
                           net_bandwidth->in->average);
577 578 579 580 581

    cmd = virCommandNew(TC);
    virCommandAddArgList(cmd, "class", "add", "dev", brname, "parent", "1:1",
                         "classid", class_id, "htb", "rate", floor,
                         "ceil", ceil, NULL);
582
    virNetDevBandwidthCmdAddOptimalQuantum(cmd, bandwidth->in);
583 584 585 586 587 588 589 590 591 592 593 594 595

    if (virCommandRun(cmd, NULL) < 0)
        goto cleanup;

    virCommandFree(cmd);
    cmd = virCommandNew(TC);
    virCommandAddArgList(cmd, "qdisc", "add", "dev", brname, "parent",
                         class_id, "handle", qdisc_id, "sfq", "perturb",
                         "10", NULL);

    if (virCommandRun(cmd, NULL) < 0)
        goto cleanup;

596 597
    if (virNetDevBandwidthManipulateFilter(brname, ifmac_ptr, id,
                                           class_id, false, true) < 0)
598 599 600 601
        goto cleanup;

    ret = 0;

602
 cleanup:
603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634
    VIR_FREE(ceil);
    VIR_FREE(floor);
    VIR_FREE(qdisc_id);
    VIR_FREE(class_id);
    virCommandFree(cmd);
    return ret;
}

/*
 * virNetDevBandwidthUnplug:
 * @brname: from which bridge are we unplugging
 * @id: unique identifier (MUST be greater than 2)
 *
 * Remove QoS settings from bridge.
 *
 * Returns 0 on success, -1 otherwise.
 */
int
virNetDevBandwidthUnplug(const char *brname,
                         unsigned int id)
{
    int ret = -1;
    int cmd_ret = 0;
    virCommandPtr cmd = NULL;
    char *class_id = NULL;
    char *qdisc_id = NULL;

    if (id <= 2) {
        virReportError(VIR_ERR_INTERNAL_ERROR, _("Invalid class ID %d"), id);
        return -1;
    }

635 636
    class_id = g_strdup_printf("1:%x", id);
    qdisc_id = g_strdup_printf("%x:", id);
637 638 639 640 641 642 643 644 645 646

    cmd = virCommandNew(TC);
    virCommandAddArgList(cmd, "qdisc", "del", "dev", brname,
                         "handle", qdisc_id, NULL);

    /* Don't threat tc errors as fatal, but
     * try to remove as much as possible */
    if (virCommandRun(cmd, &cmd_ret) < 0)
        goto cleanup;

647 648
    if (virNetDevBandwidthManipulateFilter(brname, NULL, id,
                                           NULL, true, false) < 0)
649 650
        goto cleanup;

651
    virCommandFree(cmd);
652 653 654 655 656 657 658 659 660
    cmd = virCommandNew(TC);
    virCommandAddArgList(cmd, "class", "del", "dev", brname,
                         "classid", class_id, NULL);

    if (virCommandRun(cmd, &cmd_ret) < 0)
        goto cleanup;

    ret = 0;

661
 cleanup:
662 663 664 665 666
    VIR_FREE(qdisc_id);
    VIR_FREE(class_id);
    virCommandFree(cmd);
    return ret;
}
667 668 669 670

/**
 * virNetDevBandwidthUpdateRate:
 * @ifname: interface name
671 672
 * @id: unique identifier
 * @bandwidth: used to derive 'ceil' of class with @id
673 674 675 676 677 678 679 680 681 682 683
 * @new_rate: new rate
 *
 * This function updates the 'rate' attribute of HTB class.
 * It can be used whenever a new interface is plugged to a
 * bridge to adjust average throughput of non guaranteed
 * NICs.
 *
 * Returns 0 on success, -1 otherwise.
 */
int
virNetDevBandwidthUpdateRate(const char *ifname,
684
                             unsigned int id,
685 686 687 688 689
                             virNetDevBandwidthPtr bandwidth,
                             unsigned long long new_rate)
{
    int ret = -1;
    virCommandPtr cmd = NULL;
690
    char *class_id = NULL;
691 692 693
    char *rate = NULL;
    char *ceil = NULL;

694 695 696 697 698
    class_id = g_strdup_printf("1:%x", id);
    rate = g_strdup_printf("%llukbps", new_rate);
    ceil = g_strdup_printf("%llukbps", bandwidth->in->peak ?
                           bandwidth->in->peak :
                           bandwidth->in->average);
699 700 701 702 703

    cmd = virCommandNew(TC);
    virCommandAddArgList(cmd, "class", "change", "dev", ifname,
                         "classid", class_id, "htb", "rate", rate,
                         "ceil", ceil, NULL);
704
    virNetDevBandwidthCmdAddOptimalQuantum(cmd, bandwidth->in);
705 706 707 708 709 710

    if (virCommandRun(cmd, NULL) < 0)
        goto cleanup;

    ret = 0;

711
 cleanup:
712
    virCommandFree(cmd);
713
    VIR_FREE(class_id);
714 715 716 717
    VIR_FREE(rate);
    VIR_FREE(ceil);
    return ret;
}
718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743

/**
 * virNetDevBandwidthUpdateFilter:
 * @ifname: interface to operate on
 * @ifmac_ptr: new MAC to update the filter with
 * @id: filter ID
 *
 * Sometimes the host environment is so dynamic, that even a
 * guest's MAC addresses change on the fly. When that happens we
 * must update our QoS hierarchy so that the guest's traffic is
 * placed into the correct QDiscs.  This function updates the
 * filter for the interface @ifname with the unique identifier
 * @id so that it uses the new MAC address of the guest interface
 * @ifmac_ptr.
 *
 * Returns: 0 on success,
 *         -1 on failure (with error reported).
 */
int
virNetDevBandwidthUpdateFilter(const char *ifname,
                               const virMacAddr *ifmac_ptr,
                               unsigned int id)
{
    int ret = -1;
    char *class_id = NULL;

744
    class_id = g_strdup_printf("1:%x", id);
745 746 747 748 749 750 751 752 753 754

    if (virNetDevBandwidthManipulateFilter(ifname, ifmac_ptr, id,
                                           class_id, true, true) < 0)
        goto cleanup;

    ret = 0;
 cleanup:
    VIR_FREE(class_id);
    return ret;
}