bridge_driver.c 161.0 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2013 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * 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
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

#include <sys/types.h>
#include <sys/poll.h>
#include <limits.h>
#include <string.h>
#include <stdio.h>
#include <stdarg.h>
#include <stdlib.h>
#include <unistd.h>
#include <errno.h>
#include <sys/utsname.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <signal.h>
#include <paths.h>
#include <pwd.h>
#include <sys/wait.h>
#include <sys/ioctl.h>
43
#include <net/if.h>
44

45
#include "virerror.h"
46
#include "datatypes.h"
47
#include "bridge_driver.h"
48
#include "network_conf.h"
49
#include "device_conf.h"
50
#include "driver.h"
51
#include "virbuffer.h"
52
#include "virpidfile.h"
53
#include "vircommand.h"
54
#include "viralloc.h"
55
#include "viruuid.h"
56
#include "viriptables.h"
57
#include "virlog.h"
58
#include "virdnsmasq.h"
59
#include "configmake.h"
60
#include "virnetdev.h"
61
#include "virpci.h"
62 63
#include "virnetdevbridge.h"
#include "virnetdevtap.h"
64
#include "virnetdevvportprofile.h"
65
#include "virdbus.h"
66
#include "virfile.h"
67
#include "virstring.h"
68
#include "viraccessapicheck.h"
69

70 71
#define VIR_FROM_THIS VIR_FROM_NETWORK

72
/* Main driver state */
73
struct _virNetworkDriverState {
74
    virMutex lock;
75

76
    virNetworkObjList networks;
77 78 79

    char *networkConfigDir;
    char *networkAutostartDir;
80 81 82 83
    char *stateDir;
    char *pidDir;
    char *dnsmasqStateDir;
    char *radvdStateDir;
84
    dnsmasqCapsPtr dnsmasqCaps;
85 86
};

87 88
typedef struct _virNetworkDriverState virNetworkDriverState;
typedef virNetworkDriverState *virNetworkDriverStatePtr;
89

90
static void networkDriverLock(virNetworkDriverStatePtr driver)
91
{
92
    virMutexLock(&driver->lock);
93
}
94
static void networkDriverUnlock(virNetworkDriverStatePtr driver)
95
{
96
    virMutexUnlock(&driver->lock);
97 98
}

99
static int networkStateCleanup(void);
100

101
static int networkStartNetwork(virNetworkDriverStatePtr driver,
102 103
                               virNetworkObjPtr network);

104
static int networkShutdownNetwork(virNetworkDriverStatePtr driver,
105 106
                                  virNetworkObjPtr network);

107
static int networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
108
                                     virNetworkObjPtr network);
109

110
static int networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
R
Roman Bogorodskiy 已提交
111
                                         virNetworkObjPtr network);
112

113
static int networkStartNetworkExternal(virNetworkDriverStatePtr driver,
114 115
                                     virNetworkObjPtr network);

116
static int networkShutdownNetworkExternal(virNetworkDriverStatePtr driver,
117
                                        virNetworkObjPtr network);
118

119 120
static void networkReloadIptablesRules(virNetworkDriverStatePtr driver);
static void networkRefreshDaemons(virNetworkDriverStatePtr driver);
121

122 123 124 125 126
static int networkPlugBandwidth(virNetworkObjPtr net,
                                virDomainNetDefPtr iface);
static int networkUnplugBandwidth(virNetworkObjPtr net,
                                  virDomainNetDefPtr iface);

127
static virNetworkDriverStatePtr driverState = NULL;
128

129
static char *
130
networkDnsmasqLeaseFileNameDefault(const char *netname)
131 132 133
{
    char *leasefile;

134 135
    ignore_value(virAsprintf(&leasefile, "%s/%s.leases",
                             driverState->dnsmasqStateDir, netname));
136 137 138
    return leasefile;
}

139 140 141
networkDnsmasqLeaseFileNameFunc networkDnsmasqLeaseFileName =
    networkDnsmasqLeaseFileNameDefault;

142 143 144 145 146
static char *
networkDnsmasqConfigFileName(const char *netname)
{
    char *conffile;

147 148
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
                             driverState->dnsmasqStateDir, netname));
149 150 151
    return conffile;
}

152 153 154 155 156 157
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

158
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
159 160 161 162 163 164 165 166
    return pidfilebase;
}

static char *
networkRadvdConfigFileName(const char *netname)
{
    char *configfile;

167 168
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
                             driverState->radvdStateDir, netname));
169 170
    return configfile;
}
171

172 173
/* do needed cleanup steps and remove the network from the list */
static int
174
networkRemoveInactive(virNetworkDriverStatePtr driver,
175 176 177 178
                      virNetworkObjPtr net)
{
    char *leasefile = NULL;
    char *radvdconfigfile = NULL;
179
    char *configfile = NULL;
180
    char *radvdpidbase = NULL;
181
    char *statusfile = NULL;
182 183 184 185 186 187
    dnsmasqContext *dctx = NULL;
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(net);

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
188 189
    if (!(dctx = dnsmasqContextNew(def->name,
                                   driverState->dnsmasqStateDir))) {
190
        goto cleanup;
191
    }
192 193 194 195 196

    if (!(leasefile = networkDnsmasqLeaseFileName(def->name)))
        goto cleanup;

    if (!(radvdconfigfile = networkRadvdConfigFileName(def->name)))
197
        goto cleanup;
198 199

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
200
        goto cleanup;
201

202
    if (!(configfile = networkDnsmasqConfigFileName(def->name)))
203
        goto cleanup;
204

205 206
    if (!(statusfile
          = virNetworkConfigFile(driverState->stateDir, def->name)))
207
        goto cleanup;
208

209 210 211
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
212
    unlink(configfile);
213 214 215

    /* radvd */
    unlink(radvdconfigfile);
216
    virPidFileDelete(driverState->pidDir, radvdpidbase);
217

218 219 220
    /* remove status file */
    unlink(statusfile);

221 222 223 224 225 226 227
    /* remove the network definition */
    virNetworkRemoveInactive(&driver->networks, net);

    ret = 0;

cleanup:
    VIR_FREE(leasefile);
228
    VIR_FREE(configfile);
229 230
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
231
    VIR_FREE(statusfile);
232 233 234 235
    dnsmasqContextFree(dctx);
    return ret;
}

236 237 238
static char *
networkBridgeDummyNicName(const char *brname)
{
239
    static const char dummyNicSuffix[] = "-nic";
240 241
    char *nicname;

242 243 244 245 246 247 248
    if (strlen(brname) + sizeof(dummyNicSuffix) > IFNAMSIZ) {
        /* because the length of an ifname is limited to IFNAMSIZ-1
         * (usually 15), and we're adding 4 more characters, we must
         * truncate the original name to 11 to fit. In order to catch
         * a possible numeric ending (eg virbr0, virbr1, etc), we grab
         * the first 8 and last 3 characters of the string.
         */
249 250 251 252 253
        ignore_value(virAsprintf(&nicname, "%.*s%s%s",
                                 /* space for last 3 chars + "-nic" + NULL */
                                 (int)(IFNAMSIZ - (3 + sizeof(dummyNicSuffix))),
                                 brname, brname + strlen(brname) - 3,
                                 dummyNicSuffix));
254
    } else {
255
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
256
    }
257 258 259
    return nicname;
}

260
static void
261
networkFindActiveConfigs(virNetworkDriverStatePtr driver)
262
{
263
    size_t i;
264

265
    for (i = 0; i < driver->networks.count; i++) {
266 267 268 269 270 271
        virNetworkObjPtr obj = driver->networks.objs[i];

        virNetworkObjLock(obj);

        /* If bridge exists, then mark it active */
        if (obj->def->bridge &&
H
Hu Tao 已提交
272
            virNetDevExists(obj->def->bridge) == 1) {
273 274
            obj->active = 1;

275 276
            /* Try and read dnsmasq/radvd pids if any */
            if (obj->def->ips && (obj->def->nips > 0)) {
277 278
                char *radvdpidbase;

279
                ignore_value(virPidFileReadIfAlive(driverState->pidDir, obj->def->name,
280 281
                                                   &obj->dnsmasqPid,
                                                   dnsmasqCapsGetBinaryPath(driver->dnsmasqCaps)));
282

283
                if (!(radvdpidbase = networkRadvdPidfileBasename(obj->def->name)))
284
                    goto cleanup;
285
                ignore_value(virPidFileReadIfAlive(driverState->pidDir, radvdpidbase,
286
                                                   &obj->radvdPid, RADVD));
287
                VIR_FREE(radvdpidbase);
288 289 290
            }
        }

291
    cleanup:
292 293
        virNetworkObjUnlock(obj);
    }
294 295 296 297 298 299 300 301 302 303 304 305 306 307 308

    /* remove inactive transient networks */
    i = 0;
    while (i < driver->networks.count) {
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);

        if (!obj->persistent && !obj->active) {
            networkRemoveInactive(driver, obj);
            continue;
        }

        virNetworkObjUnlock(obj);
        i++;
    }
309 310 311
}


312
static void
313
networkAutostartConfigs(virNetworkDriverStatePtr driver) {
314
    size_t i;
315

316
    for (i = 0; i < driver->networks.count; i++) {
317
        virNetworkObjLock(driver->networks.objs[i]);
318
        if (driver->networks.objs[i]->autostart &&
319 320
            !virNetworkObjIsActive(driver->networks.objs[i])) {
            if (networkStartNetwork(driver, driver->networks.objs[i]) < 0) {
321
            /* failed to start but already logged */
322
            }
323
        }
324
        virNetworkObjUnlock(driver->networks.objs[i]);
325 326 327
    }
}

328 329 330 331
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
                             DBusMessage *message, void *user_data) {
332
    virNetworkDriverStatePtr _driverState = user_data;
333 334 335 336 337 338 339 340 341 342 343 344 345 346

    if (dbus_message_is_signal(message, DBUS_INTERFACE_DBUS,
                               "NameOwnerChanged") ||
        dbus_message_is_signal(message, "org.fedoraproject.FirewallD1",
                               "Reloaded"))
    {
        VIR_DEBUG("Reload in bridge_driver because of firewalld.");
        networkReloadIptablesRules(_driverState);
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

347
/**
348
 * networkStateInitialize:
349 350 351 352
 *
 * Initialization function for the QEmu daemon
 */
static int
353 354 355
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
356
{
357 358 359
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
360 361 362
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
363 364

    if (VIR_ALLOC(driverState) < 0)
365
        goto error;
366

367 368 369 370
    if (virMutexInit(&driverState->lock) < 0) {
        VIR_FREE(driverState);
        goto error;
    }
371 372
    networkDriverLock(driverState);

373 374 375 376 377 378 379 380 381
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     *
     * NB: The qemu driver puts its domain state in /var/run, and I
     * think the network driver should have used /var/run too (instead
     * of /var/lib), but it's been this way for a long time, and we
     * probably shouldn't change it now.
     */
382
    if (privileged) {
383 384 385 386 387 388 389 390 391 392 393 394 395
        if (VIR_STRDUP(driverState->networkConfigDir,
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
            VIR_STRDUP(driverState->networkAutostartDir,
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
            VIR_STRDUP(driverState->stateDir,
                       LOCALSTATEDIR "/lib/libvirt/network") < 0 ||
            VIR_STRDUP(driverState->pidDir,
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
            VIR_STRDUP(driverState->dnsmasqStateDir,
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
            VIR_STRDUP(driverState->radvdStateDir,
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
396
    } else {
397 398 399
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
400
            goto error;
401

402 403 404 405 406 407 408 409 410 411 412 413
        if ((virAsprintf(&driverState->networkConfigDir,
                         "%s/qemu/networks", configdir) < 0) ||
            (virAsprintf(&driverState->networkAutostartDir,
                         "%s/qemu/networks/autostart", configdir) < 0) ||
            (virAsprintf(&driverState->stateDir,
                         "%s/network/lib", rundir) < 0) ||
            (virAsprintf(&driverState->pidDir,
                         "%s/network/run", rundir) < 0) ||
            (virAsprintf(&driverState->dnsmasqStateDir,
                         "%s/dnsmasq/lib", rundir) < 0) ||
            (virAsprintf(&driverState->radvdStateDir,
                         "%s/radvd/lib", rundir) < 0)) {
414
            goto error;
415
        }
416 417
    }

418 419
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
    driverState->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
420

421
    if (virNetworkLoadAllState(&driverState->networks,
422
                               driverState->stateDir) < 0)
423 424
        goto error;

425
    if (virNetworkLoadAllConfigs(&driverState->networks,
426
                                 driverState->networkConfigDir,
427 428 429
                                 driverState->networkAutostartDir) < 0)
        goto error;

430
    networkFindActiveConfigs(driverState);
431
    networkReloadIptablesRules(driverState);
432
    networkRefreshDaemons(driverState);
433 434
    networkAutostartConfigs(driverState);

435 436
    networkDriverUnlock(driverState);

437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462
#ifdef HAVE_FIREWALLD
    if (!(sysbus = virDBusGetSystemBus())) {
        virErrorPtr err = virGetLastError();
        VIR_WARN("DBus not available, disabling firewalld support "
                 "in bridge_driver: %s", err->message);
    } else {
        /* add matches for
         * NameOwnerChanged on org.freedesktop.DBus for firewalld start/stop
         * Reloaded on org.fedoraproject.FirewallD1 for firewalld reload
         */
        dbus_bus_add_match(sysbus,
                           "type='signal'"
                           ",interface='"DBUS_INTERFACE_DBUS"'"
                           ",member='NameOwnerChanged'"
                           ",arg0='org.fedoraproject.FirewallD1'",
                           NULL);
        dbus_bus_add_match(sysbus,
                           "type='signal'"
                           ",interface='org.fedoraproject.FirewallD1'"
                           ",member='Reloaded'",
                           NULL);
        dbus_connection_add_filter(sysbus, firewalld_dbus_filter_bridge,
                                   driverState, NULL);
    }
#endif

463 464 465 466 467
    ret = 0;
cleanup:
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
468

469
error:
470 471
    if (driverState)
        networkDriverUnlock(driverState);
472
    networkStateCleanup();
473
    goto cleanup;
474 475 476
}

/**
477
 * networkStateReload:
478 479 480 481 482
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
483
networkStateReload(void) {
484 485 486
    if (!driverState)
        return 0;

487
    networkDriverLock(driverState);
488
    virNetworkLoadAllState(&driverState->networks,
489
                           driverState->stateDir);
490
    virNetworkLoadAllConfigs(&driverState->networks,
491 492
                             driverState->networkConfigDir,
                             driverState->networkAutostartDir);
493
    networkReloadIptablesRules(driverState);
494
    networkRefreshDaemons(driverState);
495
    networkAutostartConfigs(driverState);
496
    networkDriverUnlock(driverState);
497 498 499 500 501
    return 0;
}


/**
502
 * networkStateCleanup:
503 504 505 506
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
507
networkStateCleanup(void) {
508 509 510
    if (!driverState)
        return -1;

511 512
    networkDriverLock(driverState);

513
    /* free inactive networks */
514
    virNetworkObjListFree(&driverState->networks);
515 516 517

    VIR_FREE(driverState->networkConfigDir);
    VIR_FREE(driverState->networkAutostartDir);
518 519 520 521
    VIR_FREE(driverState->stateDir);
    VIR_FREE(driverState->pidDir);
    VIR_FREE(driverState->dnsmasqStateDir);
    VIR_FREE(driverState->radvdStateDir);
522

523 524
    virObjectUnref(driverState->dnsmasqCaps);

525
    networkDriverUnlock(driverState);
526
    virMutexDestroy(&driverState->lock);
527

528 529 530 531 532 533
    VIR_FREE(driverState);

    return 0;
}


534 535 536 537 538 539 540
/* networkKillDaemon:
 *
 * kill the specified pid/name, and wait a bit to make sure it's dead.
 */
static int
networkKillDaemon(pid_t pid, const char *daemonName, const char *networkName)
{
541 542
    size_t i;
    int ret = -1;
543 544 545 546 547 548 549
    const char *signame = "TERM";

    /* send SIGTERM, then wait up to 3 seconds for the process to
     * disappear, send SIGKILL, then wait for up to another 2
     * seconds. If that fails, log a warning and continue, hoping
     * for the best.
     */
550
    for (i = 0; i < 25; i++) {
551
        int signum = 0;
552
        if (i == 0)
553
            signum = SIGTERM;
554
        else if (i == 15) {
555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593
            signum = SIGKILL;
            signame = "KILL";
        }
        if (kill(pid, signum) < 0) {
            if (errno == ESRCH) {
                ret = 0;
            } else {
                char ebuf[1024];
                VIR_WARN("Failed to terminate %s process %d "
                         "for network '%s' with SIG%s: %s",
                         daemonName, pid, networkName, signame,
                         virStrerror(errno, ebuf, sizeof(ebuf)));
            }
            goto cleanup;
        }
        /* NB: since networks have no reference count like
         * domains, there is no safe way to unlock the network
         * object temporarily, and so we can't follow the
         * procedure used by the qemu driver of 1) unlock driver
         * 2) sleep, 3) add ref to object 4) unlock object, 5)
         * re-lock driver, 6) re-lock object. We may need to add
         * that functionality eventually, but for now this
         * function is rarely used and, at worst, leaving the
         * network driver locked during this loop of sleeps will
         * have the effect of holding up any other thread trying
         * to make modifications to a network for up to 5 seconds;
         * since modifications to networks are much less common
         * than modifications to domains, this seems a reasonable
         * tradeoff in exchange for less code disruption.
         */
        usleep(20 * 1000);
    }
    VIR_WARN("Timed out waiting after SIG%s to %s process %d "
             "(network '%s')",
             signame, daemonName, pid, networkName);
cleanup:
    return ret;
}

G
Gene Czarcinski 已提交
594 595 596 597
    /* the following does not build a file, it builds a list
     * which is later saved into a file
     */

598
static int
G
Gene Czarcinski 已提交
599 600
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
                                 virNetworkIpDefPtr ipdef)
601
{
602
    size_t i;
G
Gene Czarcinski 已提交
603
    bool ipv6 = false;
604

G
Gene Czarcinski 已提交
605 606
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
607 608
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
609
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
610 611
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
612
                return -1;
613
    }
614

G
Gene Czarcinski 已提交
615 616 617 618 619 620 621
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
622
    size_t i, j;
G
Gene Czarcinski 已提交
623

624 625
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
626
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
627
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
628
                for (j = 0; j < host->nnames; j++)
629 630
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
631 632
            }
        }
633 634
    }

635
    return 0;
636 637 638
}


639 640
int
networkDnsmasqConfContents(virNetworkObjPtr network,
641 642 643 644
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
645
{
646
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
647
    int r, ret = -1;
648
    int nbleases = 0;
649
    size_t i;
650 651 652 653
    char *record = NULL;
    char *recordPort = NULL;
    char *recordWeight = NULL;
    char *recordPriority = NULL;
654
    virNetworkDNSDefPtr dns = &network->def->dns;
G
Gene Czarcinski 已提交
655 656
    virNetworkIpDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
    bool ipv6SLAAC;
657

658 659
    *configstr = NULL;

660
    /*
661 662 663
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
664
     *
665 666
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
667
     */
668 669 670 671 672 673

    /*
     * Needed to ensure dnsmasq uses same algorithm for processing
     * multiple namedriver entries in /etc/resolv.conf as GLibC.
     */

674 675
    /* create dnsmasq config file appropriate for this network */
    virBufferAsprintf(&configbuf,
676 677 678 679 680 681 682 683 684 685 686 687
                      "##WARNING:  THIS IS AN AUTO-GENERATED FILE. "
                      "CHANGES TO IT ARE LIKELY TO BE\n"
                      "##OVERWRITTEN AND LOST.  Changes to this "
                      "configuration should be made using:\n"
                      "##    virsh net-edit %s\n"
                      "## or other application using the libvirt API.\n"
                      "##\n## dnsmasq conf file created by libvirt\n"
                      "strict-order\n"
                      "domain-needed\n",
                      network->def->name);

    if (network->def->domain) {
688
        virBufferAsprintf(&configbuf,
689 690 691 692 693
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
    /* need to specify local even if no domain specified */
694
    virBufferAsprintf(&configbuf,
695 696
                      "local=/%s/\n",
                      network->def->domain ? network->def->domain : "");
697

698
    if (pidfile)
699
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
700

701 702 703
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

704 705 706 707 708 709 710 711
    if (dnsmasqCapsGet(caps, DNSMASQ_CAPS_BIND_DYNAMIC)) {
        /* using --bind-dynamic with only --interface (no
         * --listen-address) prevents dnsmasq from responding to dns
         * queries that arrive on some interface other than our bridge
         * interface (in other words, requests originating somewhere
         * other than one of the virtual guests connected directly to
         * this network). This was added in response to CVE 2012-3411.
         */
712
        virBufferAsprintf(&configbuf,
713 714 715
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
716
    } else {
717
        virBufferAddLit(&configbuf, "bind-interfaces\n");
718 719 720 721 722 723 724 725
        /*
         * --interface does not actually work with dnsmasq < 2.47,
         * due to DAD for ipv6 addresses on the interface.
         *
         * virCommandAddArgList(cmd, "--interface", network->def->bridge, NULL);
         *
         * So listen on all defined IPv[46] addresses
         */
726 727 728
        for (i = 0;
             (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
             i++) {
729 730 731 732
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
733

734
            /* also part of CVE 2012-3411 - if the host's version of
735
             * dnsmasq doesn't have bind-dynamic, only allow listening on
736 737
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
738 739
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
740 741 742 743
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
744
                                 "The version of dnsmasq on this host (%d.%d) "
745 746 747 748
                                 "doesn't support the bind-dynamic option or "
                                 "use SO_BINDTODEVICE on listening sockets, "
                                 "one of which is required for safe operation "
                                 "on a publicly routable subnet "
749 750 751 752 753 754
                                 "(see CVE-2012-3411). You must either "
                                 "upgrade dnsmasq, or use a private/local "
                                 "subnet range for this network "
                                 "(as described in RFC1918/RFC3484/RFC4193)."),
                               ipaddr, (int)version / 1000000,
                               (int)(version % 1000000) / 1000);
755 756
                goto cleanup;
            }
757
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
758 759 760
            VIR_FREE(ipaddr);
        }
    }
761

762 763
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
764
     * guaranteed to not work, and set no-resolv so that no dns
765 766 767
     * requests are forwarded on to the dns server listed in the
     * host's /etc/resolv.conf (since this could be used as a channel
     * to build a connection to the outside).
768
     */
769
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
770
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
771
                        "no-resolv\n");
772
    }
773

774
    for (i = 0; i < dns->ntxts; i++) {
775
        virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
776 777
                          dns->txts[i].name,
                          dns->txts[i].value);
778
    }
779

780 781 782 783
    for (i = 0; i < dns->nsrvs; i++) {
        if (dns->srvs[i].service && dns->srvs[i].protocol) {
            if (dns->srvs[i].port &&
                virAsprintf(&recordPort, "%d", dns->srvs[i].port) < 0)
784
                goto cleanup;
785 786
            if (dns->srvs[i].priority &&
                virAsprintf(&recordPriority, "%d", dns->srvs[i].priority) < 0)
787
                goto cleanup;
788 789
            if (dns->srvs[i].weight &&
                virAsprintf(&recordWeight, "%d", dns->srvs[i].weight) < 0)
790
                goto cleanup;
791

792
            if (virAsprintf(&record, "%s.%s.%s,%s,%s,%s,%s",
793 794 795 796
                            dns->srvs[i].service,
                            dns->srvs[i].protocol,
                            dns->srvs[i].domain ? dns->srvs[i].domain : "",
                            dns->srvs[i].target ? dns->srvs[i].target : "",
797 798
                            recordPort           ? recordPort           : "",
                            recordPriority       ? recordPriority       : "",
799
                            recordWeight         ? recordWeight         : "") < 0)
800 801
                goto cleanup;

802
            virBufferAsprintf(&configbuf, "srv-host=%s\n", record);
803 804 805 806
            VIR_FREE(record);
            VIR_FREE(recordPort);
            VIR_FREE(recordWeight);
            VIR_FREE(recordPriority);
807
        }
808 809
    }

G
Gene Czarcinski 已提交
810
    /* Find the first dhcp for both IPv4 and IPv6 */
811 812 813
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
814 815 816 817
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
818 819
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
820 821 822 823 824 825 826 827 828 829 830
                    goto cleanup;
                } else {
                    ipv4def = ipdef;
                }
            }
        }
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (!DNSMASQ_DHCPv6_SUPPORT(caps)) {
                    unsigned long version = dnsmasqCapsGetVersion(caps);
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
831 832 833 834 835 836 837 838 839
                                   _("The version of dnsmasq on this host "
                                     "(%d.%d) doesn't adequately support "
                                     "IPv6 dhcp range or dhcp host "
                                     "specification. Version %d.%d or later "
                                     "is required."),
                                   (int)version / 1000000,
                                   (int)(version % 1000000) / 1000,
                                   DNSMASQ_DHCPv6_MAJOR_REQD,
                                   DNSMASQ_DHCPv6_MINOR_REQD);
G
Gene Czarcinski 已提交
840 841 842 843
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
844 845
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
846 847 848 849 850 851 852 853 854 855 856 857 858
                    goto cleanup;
                } else {
                    ipv6def = ipdef;
                }
            } else {
                ipv6SLAAC = true;
            }
        }
    }

    if (ipv6def && ipv6SLAAC) {
        VIR_WARN("For IPv6, when DHCP is specified for one address, then "
                 "state-full Router Advertising will occur.  The additional "
859 860 861 862
                 "IPv6 addresses specified require manually configured guest "
                 "network to work properly since both state-full (DHCP) "
                 "and state-less (SLAAC) addressing are not supported "
                 "on the same network interface.");
G
Gene Czarcinski 已提交
863 864 865 866 867
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
868
        for (r = 0; r < ipdef->nranges; r++) {
869
            char *saddr = virSocketAddrFormat(&ipdef->ranges[r].start);
870 871
            if (!saddr)
                goto cleanup;
872
            char *eaddr = virSocketAddrFormat(&ipdef->ranges[r].end);
873 874 875 876
            if (!eaddr) {
                VIR_FREE(saddr);
                goto cleanup;
            }
877
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s\n",
878
                              saddr, eaddr);
879
            VIR_FREE(saddr);
880
            VIR_FREE(eaddr);
881 882
            nbleases += virSocketAddrGetRange(&ipdef->ranges[r].start,
                                              &ipdef->ranges[r].end);
883
        }
884

885
        /*
886 887 888 889
         * For static-only DHCP, i.e. with no range but at least one
         * host element, we have to add a special --dhcp-range option
         * to enable the service in dnsmasq. (this is for dhcp-hosts=
         * support)
890 891
         */
        if (!ipdef->nranges && ipdef->nhosts) {
892
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
893 894
            if (!bridgeaddr)
                goto cleanup;
895
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static\n", bridgeaddr);
896 897
            VIR_FREE(bridgeaddr);
        }
898

G
Gene Czarcinski 已提交
899 900
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
901

G
Gene Czarcinski 已提交
902 903 904
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts)
905
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
906

G
Gene Czarcinski 已提交
907
            if (ipdef->tftproot) {
908 909
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
910
            }
911

G
Gene Czarcinski 已提交
912 913 914
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
915

916
                    if (!bootserver)
G
Gene Czarcinski 已提交
917
                        goto cleanup;
918
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
919
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
920 921
                    VIR_FREE(bootserver);
                } else {
922
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
923 924 925 926 927
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
928

G
Gene Czarcinski 已提交
929 930
    if (nbleases > 0) {
        char *leasefile = networkDnsmasqLeaseFileName(network->def->name);
931
        if (!leasefile)
G
Gene Czarcinski 已提交
932
            goto cleanup;
933
        virBufferAsprintf(&configbuf, "dhcp-leasefile=%s\n", leasefile);
G
Gene Czarcinski 已提交
934
        VIR_FREE(leasefile);
935
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
G
Gene Czarcinski 已提交
936
    }
937

G
Gene Czarcinski 已提交
938 939
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
940
        goto cleanup;
G
Gene Czarcinski 已提交
941 942 943 944 945 946

    /* Even if there are currently no static hosts, if we're
     * listening for DHCP, we should write a 0-length hosts
     * file to allow for runtime additions.
     */
    if (ipv4def || ipv6def)
947 948
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
949

950 951
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
952
     */
953
    virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
954
                      dctx->addnhostsfile->path);
G
Gene Czarcinski 已提交
955 956 957 958

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
        if (ipv6def)
959
            virBufferAddLit(&configbuf, "enable-ra\n");
G
Gene Czarcinski 已提交
960
        else {
961 962 963
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
                 i++) {
G
Gene Czarcinski 已提交
964 965 966 967
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
968 969
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
970 971
                    VIR_FREE(bridgeaddr);
                }
972
            }
973
        }
974 975
    }

976 977 978
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

979
    ret = 0;
G
Gene Czarcinski 已提交
980

981
cleanup:
982
    virBufferFreeAndReset(&configbuf);
983 984 985 986
    VIR_FREE(record);
    VIR_FREE(recordPort);
    VIR_FREE(recordWeight);
    VIR_FREE(recordPriority);
987
    return ret;
988 989
}

990
/* build the dnsmasq command line */
991 992 993
static int ATTRIBUTE_NONNULL(2)
networkBuildDhcpDaemonCommandLine(virNetworkObjPtr network,
                                  virCommandPtr *cmdout,
994 995
                                  char *pidfile, dnsmasqContext *dctx,
                                  dnsmasqCapsPtr caps)
996
{
997
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
998
    int ret = -1;
999 1000
    char *configfile = NULL;
    char *configstr = NULL;
1001 1002

    network->dnsmasqPid = -1;
1003

1004 1005 1006 1007 1008 1009
    if (networkDnsmasqConfContents(network, pidfile, &configstr, dctx, caps) < 0)
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1010
    if (!(configfile = networkDnsmasqConfigFileName(network->def->name)))
1011 1012 1013 1014 1015 1016 1017
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
                         _("couldn't write dnsmasq config file '%s'"),
                         configfile);
1018 1019 1020
        goto cleanup;
    }

1021 1022
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(caps));
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1023
    *cmdout = cmd;
1024 1025 1026 1027 1028 1029
    ret = 0;
cleanup:
    return ret;
}

static int
1030
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
1031
                       virNetworkObjPtr network)
1032 1033 1034 1035
{
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1036
    dnsmasqContext *dctx = NULL;
1037

1038
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0)) {
G
Gene Czarcinski 已提交
1039
        /* no IP addresses, so we don't need to run */
1040 1041 1042 1043
        ret = 0;
        goto cleanup;
    }

1044
    if (virFileMakePath(driverState->pidDir) < 0) {
1045
        virReportSystemError(errno,
1046
                             _("cannot create directory %s"),
1047
                             driverState->pidDir);
1048
        goto cleanup;
1049
    }
1050
    if (virFileMakePath(driverState->stateDir) < 0) {
1051
        virReportSystemError(errno,
1052
                             _("cannot create directory %s"),
1053
                             driverState->stateDir);
1054
        goto cleanup;
1055 1056
    }

1057
    if (!(pidfile = virPidFileBuildPath(driverState->pidDir,
1058
                                        network->def->name)))
1059
        goto cleanup;
1060

1061
    if (virFileMakePath(driverState->dnsmasqStateDir) < 0) {
1062
        virReportSystemError(errno,
1063
                             _("cannot create directory %s"),
1064
                             driverState->dnsmasqStateDir);
1065 1066 1067
        goto cleanup;
    }

1068
    dctx = dnsmasqContextNew(network->def->name, driverState->dnsmasqStateDir);
1069 1070 1071
    if (dctx == NULL)
        goto cleanup;

1072 1073 1074 1075
    dnsmasqCapsRefresh(&driver->dnsmasqCaps, false);

    ret = networkBuildDhcpDaemonCommandLine(network, &cmd, pidfile,
                                            dctx, driver->dnsmasqCaps);
1076 1077 1078 1079 1080
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1081
        goto cleanup;
1082

G
Guido Günther 已提交
1083 1084
    ret = virCommandRun(cmd, NULL);
    if (ret < 0) {
1085
        goto cleanup;
G
Guido Günther 已提交
1086
    }
1087 1088

    /*
1089 1090 1091 1092 1093
     * There really is no race here - when dnsmasq daemonizes, its
     * leader process stays around until its child has actually
     * written its pidfile. So by time virCommandRun exits it has
     * waitpid'd and guaranteed the proess has started and written a
     * pid
1094 1095
     */

1096
    ret = virPidFileRead(driverState->pidDir, network->def->name,
1097 1098
                         &network->dnsmasqPid);
    if (ret < 0)
1099
        goto cleanup;
1100

1101 1102 1103
    ret = 0;
cleanup:
    VIR_FREE(pidfile);
1104
    virCommandFree(cmd);
1105
    dnsmasqContextFree(dctx);
1106 1107 1108
    return ret;
}

1109 1110
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1111 1112
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1113 1114 1115
 *
 *  Returns 0 on success, -1 on failure.
 */
1116
static int
1117
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
1118
                         virNetworkObjPtr network)
1119
{
1120 1121
    int ret = -1;
    size_t i;
G
Gene Czarcinski 已提交
1122
    virNetworkIpDefPtr ipdef, ipv4def, ipv6def;
1123
    dnsmasqContext *dctx = NULL;
1124

G
Gene Czarcinski 已提交
1125
    /* if no IP addresses specified, nothing to do */
1126
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1127 1128
        return 0;

1129 1130
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1131
        return networkStartDhcpDaemon(driver, network);
1132

G
Gene Czarcinski 已提交
1133
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1134 1135
    if (!(dctx = dnsmasqContextNew(network->def->name,
                                   driverState->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1136
        goto cleanup;
1137
    }
G
Gene Czarcinski 已提交
1138 1139 1140 1141 1142 1143

    /* Look for first IPv4 address that has dhcp defined.
     * We only support dhcp-host config on one IPv4 subnetwork
     * and on one IPv6 subnetwork.
     */
    ipv4def = NULL;
1144 1145 1146
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
G
Gene Czarcinski 已提交
1147 1148
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1149 1150
    }

G
Gene Czarcinski 已提交
1151
    ipv6def = NULL;
1152 1153 1154
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1155 1156
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1157 1158
    }

G
Gene Czarcinski 已提交
1159 1160 1161 1162 1163
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
           goto cleanup;

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
           goto cleanup;
1164

G
Gene Czarcinski 已提交
1165
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
1166 1167 1168
       goto cleanup;

    if ((ret = dnsmasqSave(dctx)) < 0)
1169
        goto cleanup;
1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184

    ret = kill(network->dnsmasqPid, SIGHUP);
cleanup:
    dnsmasqContextFree(dctx);
    return ret;
}

/* networkRestartDhcpDaemon:
 *
 * kill and restart dnsmasq, in order to update any config that is on
 * the dnsmasq commandline (and any placed in separate config files).
 *
 *  Returns 0 on success, -1 on failure.
 */
static int
1185
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
1186
                         virNetworkObjPtr network)
1187 1188 1189 1190 1191 1192
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1193
    }
1194
    /* now start dnsmasq if it should be started */
1195
    return networkStartDhcpDaemon(driver, network);
1196 1197
}

G
Gene Czarcinski 已提交
1198 1199 1200 1201 1202 1203
static char radvd1[] = "  AdvOtherConfigFlag off;\n\n";
static char radvd2[] = "    AdvAutonomous off;\n";
static char radvd3[] = "    AdvOnLink on;\n"
                       "    AdvAutonomous on;\n"
                       "    AdvRouterAddr off;\n";

1204 1205 1206
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1207
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1208 1209
    int ret = -1;
    size_t i;
1210
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
1211
    bool v6present = false, dhcp6 = false;
1212 1213

    *configstr = NULL;
1214

G
Gene Czarcinski 已提交
1215
    /* Check if DHCPv6 is needed */
1216 1217 1218
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231
        v6present = true;
        if (ipdef->nranges || ipdef->nhosts) {
            dhcp6 = true;
            break;
        }
    }

    /* If there are no IPv6 addresses, then we are done */
    if (!v6present) {
        ret = 0;
        goto cleanup;
    }

1232 1233 1234
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1235
    virBufferAsprintf(&configbuf, "interface %s\n"
1236 1237
                      "{\n"
                      "  AdvSendAdvert on;\n"
1238
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1239 1240 1241 1242 1243
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1244 1245

    /* add a section for each IPv6 address in the config */
1246 1247 1248
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
1249 1250 1251 1252 1253
        int prefix;
        char *netaddr;

        prefix = virNetworkIpDefPrefix(ipdef);
        if (prefix < 0) {
1254 1255 1256
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1257 1258
            goto cleanup;
        }
1259
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1260
            goto cleanup;
1261
        virBufferAsprintf(&configbuf,
1262
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1263 1264 1265
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1266 1267 1268
        VIR_FREE(netaddr);
    }

1269 1270 1271
    /* only create the string if we found at least one IPv6 address */
    if (v6present) {
        virBufferAddLit(&configbuf, "};\n");
1272

1273 1274 1275 1276 1277 1278 1279 1280
        if (virBufferError(&configbuf)) {
            virReportOOMError();
            goto cleanup;
        }
        if (!(*configstr = virBufferContentAndReset(&configbuf))) {
            virReportOOMError();
            goto cleanup;
        }
1281
    }
1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306

    ret = 0;
cleanup:
    virBufferFreeAndReset(&configbuf);
    return ret;
}

/* write file and return it's name (which must be freed by caller) */
static int
networkRadvdConfWrite(virNetworkObjPtr network, char **configFile)
{
    int ret = -1;
    char *configStr = NULL;
    char *myConfigFile = NULL;

    if (!configFile)
        configFile = &myConfigFile;

    *configFile = NULL;

    if (networkRadvdConfContents(network, &configStr) < 0)
        goto cleanup;

    if (!configStr) {
        ret = 0;
1307 1308 1309 1310
        goto cleanup;
    }

    /* construct the filename */
1311
    if (!(*configFile = networkRadvdConfigFileName(network->def->name)))
1312 1313
        goto cleanup;
    /* write the file */
1314
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1315 1316
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328
                             *configFile);
        goto cleanup;
    }

    ret = 0;
cleanup:
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1329
networkStartRadvd(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
G
Gene Czarcinski 已提交
1330
                        virNetworkObjPtr network)
1331 1332 1333 1334 1335 1336 1337 1338 1339
{
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1340
    /* Is dnsmasq handling RA? */
1341
   if (DNSMASQ_RA_SUPPORT(driver->dnsmasqCaps)) {
G
Gene Czarcinski 已提交
1342 1343 1344 1345
        ret = 0;
        goto cleanup;
    }

1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356
    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0)) {
        /* no IPv6 addresses, so we don't need to run radvd */
        ret = 0;
        goto cleanup;
    }

    if (!virFileIsExecutable(RADVD)) {
        virReportSystemError(errno,
                             _("Cannot find %s - "
                               "Possibly the package isn't installed"),
                             RADVD);
1357 1358 1359
        goto cleanup;
    }

1360
    if (virFileMakePath(driverState->pidDir) < 0) {
1361 1362
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1363
                             driverState->pidDir);
1364 1365
        goto cleanup;
    }
1366
    if (virFileMakePath(driverState->radvdStateDir) < 0) {
1367 1368
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1369
                             driverState->radvdStateDir);
1370 1371 1372 1373
        goto cleanup;
    }

    /* construct pidfile name */
1374
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1375
        goto cleanup;
1376
    if (!(pidfile = virPidFileBuildPath(driverState->pidDir, radvdpidbase)))
1377 1378 1379 1380 1381
        goto cleanup;

    if (networkRadvdConfWrite(network, &configfile) < 0)
        goto cleanup;

1382 1383 1384 1385
    /* prevent radvd from daemonizing itself with "--debug 1", and use
     * a dummy pidfile name - virCommand will create the pidfile we
     * want to use (this is necessary because radvd's internal
     * daemonization and pidfile creation causes a race, and the
1386
     * virPidFileRead() below will fail if we use them).
1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401
     * Unfortunately, it isn't possible to tell radvd to not create
     * its own pidfile, so we just let it do so, with a slightly
     * different name. Unused, but harmless.
     */
    cmd = virCommandNewArgList(RADVD, "--debug", "1",
                               "--config", configfile,
                               "--pidfile", NULL);
    virCommandAddArgFormat(cmd, "%s-bin", pidfile);

    virCommandSetPidFile(cmd, pidfile);
    virCommandDaemonize(cmd);

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

1402
    if (virPidFileRead(driverState->pidDir, radvdpidbase, &network->radvdPid) < 0)
1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413
        goto cleanup;

    ret = 0;
cleanup:
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1414
static int
1415
networkRefreshRadvd(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
1416
                    virNetworkObjPtr network)
1417
{
G
Gene Czarcinski 已提交
1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428
    char *radvdpidbase;

    /* Is dnsmasq handling RA? */
    if (DNSMASQ_RA_SUPPORT(driver->dnsmasqCaps)) {
        if (network->radvdPid <= 0)
            return 0;
        /* radvd should not be running but in case it is */
        if ((networkKillDaemon(network->radvdPid, "radvd",
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1429
            virPidFileDelete(driverState->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1430 1431 1432 1433 1434 1435
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }

1436 1437
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
G
Gene Czarcinski 已提交
1438
        return networkStartRadvd(driver, network);
1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450

    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0)) {
        /* no IPv6 addresses, so we don't need to run radvd */
        return 0;
    }

    if (networkRadvdConfWrite(network, NULL) < 0)
        return -1;

    return kill(network->radvdPid, SIGHUP);
}

1451 1452
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1453
static int
1454
networkRestartRadvd(virNetworkDriverStatePtr driver,
1455
                    virNetworkObjPtr network)
1456 1457 1458 1459 1460 1461 1462 1463 1464
{
    char *radvdpidbase;

    /* if there is a running radvd, kill it */
    if (network->radvdPid > 0) {
        /* essentially ignore errors from the following two functions,
         * since there's really no better recovery to be done than to
         * just push ahead (and that may be exactly what's needed).
         */
G
Gene Czarcinski 已提交
1465
        if ((networkKillDaemon(network->radvdPid, "radvd",
1466 1467 1468
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1469
            virPidFileDelete(driverState->pidDir, radvdpidbase);
1470 1471 1472 1473 1474 1475 1476 1477 1478
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1479 1480 1481 1482
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
1483
networkRefreshDaemons(virNetworkDriverStatePtr driver)
1484
{
1485
    size_t i;
1486 1487 1488

    VIR_INFO("Refreshing network daemons");

1489
    for (i = 0; i < driver->networks.count; i++) {
1490 1491 1492 1493
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
1494 1495 1496
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
1497 1498 1499 1500 1501 1502
            /* Only the three L3 network types that are configured by
             * libvirt will have a dnsmasq or radvd daemon associated
             * with them.  Here we send a SIGHUP to an existing
             * dnsmasq and/or radvd, or restart them if they've
             * disappeared.
             */
1503 1504
            networkRefreshDhcpDaemon(driver, network);
            networkRefreshRadvd(driver, network);
1505 1506 1507 1508 1509
        }
        virNetworkObjUnlock(network);
    }
}

1510
static int
R
Roman Bogorodskiy 已提交
1511
networkAddMasqueradingIptablesRules(virNetworkObjPtr network,
1512
                                    virNetworkIpDefPtr ipdef)
1513 1514
{
    int prefix = virNetworkIpDefPrefix(ipdef);
1515
    const char *forwardIf = virNetworkDefForwardIf(network->def, 0);
1516 1517

    if (prefix < 0) {
1518 1519 1520
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Invalid prefix or netmask for '%s'"),
                       network->def->bridge);
1521 1522
        goto masqerr1;
    }
1523

1524
    /* allow forwarding packets from the bridge interface */
R
Roman Bogorodskiy 已提交
1525
    if (iptablesAddForwardAllowOut(&ipdef->address,
1526
                                   prefix,
1527
                                   network->def->bridge,
1528
                                   forwardIf) < 0) {
1529 1530 1531
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow forwarding from '%s'"),
                       network->def->bridge);
1532 1533 1534
        goto masqerr1;
    }

1535 1536 1537
    /* allow forwarding packets to the bridge interface if they are
     * part of an existing connection
     */
R
Roman Bogorodskiy 已提交
1538
    if (iptablesAddForwardAllowRelatedIn(&ipdef->address,
1539
                                         prefix,
1540
                                         network->def->bridge,
1541
                                         forwardIf) < 0) {
1542 1543 1544
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow forwarding to '%s'"),
                       network->def->bridge);
1545 1546 1547
        goto masqerr2;
    }

1548 1549 1550 1551 1552
    /*
     * Enable masquerading.
     *
     * We need to end up with 3 rules in the table in this order
     *
E
Eric Blake 已提交
1553 1554
     *  1. protocol=tcp with sport mapping restriction
     *  2. protocol=udp with sport mapping restriction
1555 1556 1557
     *  3. generic any protocol
     *
     * The sport mappings are required, because default IPtables
E
Eric Blake 已提交
1558
     * MASQUERADE maintain port numbers unchanged where possible.
1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571
     *
     * NFS can be configured to only "trust" port numbers < 1023.
     *
     * Guests using NAT thus need to be prevented from having port
     * numbers < 1023, otherwise they can bypass the NFS "security"
     * check on the source port number.
     *
     * Since we use '--insert' to add rules to the header of the
     * chain, we actually need to add them in the reverse of the
     * order just mentioned !
     */

    /* First the generic masquerade rule for other protocols */
R
Roman Bogorodskiy 已提交
1572
    if (iptablesAddForwardMasquerade(&ipdef->address,
1573
                                     prefix,
1574
                                     forwardIf,
1575 1576
                                     &network->def->forward.addr,
                                     &network->def->forward.port,
1577
                                     NULL) < 0) {
1578 1579 1580 1581 1582 1583 1584
        if (forwardIf)
            virReportError(VIR_ERR_SYSTEM_ERROR,
                           _("failed to add iptables rule to enable masquerading to %s"),
                           forwardIf);
        else
            virReportError(VIR_ERR_SYSTEM_ERROR, "%s",
                           _("failed to add iptables rule to enable masquerading"));
1585 1586 1587
        goto masqerr3;
    }

1588
    /* UDP with a source port restriction */
R
Roman Bogorodskiy 已提交
1589
    if (iptablesAddForwardMasquerade(&ipdef->address,
1590
                                     prefix,
1591
                                     forwardIf,
1592 1593
                                     &network->def->forward.addr,
                                     &network->def->forward.port,
1594
                                     "udp") < 0) {
1595 1596 1597 1598 1599 1600 1601
        if (forwardIf)
            virReportError(VIR_ERR_SYSTEM_ERROR,
                           _("failed to add iptables rule to enable UDP masquerading to %s"),
                           forwardIf);
        else
            virReportError(VIR_ERR_SYSTEM_ERROR, "%s",
                           _("failed to add iptables rule to enable UDP masquerading"));
1602 1603 1604 1605
        goto masqerr4;
    }

    /* TCP with a source port restriction */
R
Roman Bogorodskiy 已提交
1606
    if (iptablesAddForwardMasquerade(&ipdef->address,
1607
                                     prefix,
1608
                                     forwardIf,
1609 1610
                                     &network->def->forward.addr,
                                     &network->def->forward.port,
1611
                                     "tcp") < 0) {
1612 1613 1614 1615 1616 1617 1618
        if (forwardIf)
            virReportError(VIR_ERR_SYSTEM_ERROR,
                           _("failed to add iptables rule to enable TCP masquerading to %s"),
                           forwardIf);
        else
            virReportError(VIR_ERR_SYSTEM_ERROR, "%s",
                           _("failed to add iptables rule to enable TCP masquerading"));
1619 1620 1621
        goto masqerr5;
    }

1622
    return 0;
1623

1624
 masqerr5:
R
Roman Bogorodskiy 已提交
1625
    iptablesRemoveForwardMasquerade(&ipdef->address,
1626
                                    prefix,
1627
                                    forwardIf,
1628 1629
                                    &network->def->forward.addr,
                                    &network->def->forward.port,
1630 1631
                                    "udp");
 masqerr4:
R
Roman Bogorodskiy 已提交
1632
    iptablesRemoveForwardMasquerade(&ipdef->address,
1633
                                    prefix,
1634
                                    forwardIf,
1635 1636
                                    &network->def->forward.addr,
                                    &network->def->forward.port,
1637
                                    NULL);
1638
 masqerr3:
R
Roman Bogorodskiy 已提交
1639
    iptablesRemoveForwardAllowRelatedIn(&ipdef->address,
1640
                                        prefix,
1641
                                        network->def->bridge,
1642
                                        forwardIf);
1643
 masqerr2:
R
Roman Bogorodskiy 已提交
1644
    iptablesRemoveForwardAllowOut(&ipdef->address,
1645
                                  prefix,
1646
                                  network->def->bridge,
1647
                                  forwardIf);
1648
 masqerr1:
1649
    return -1;
1650 1651
}

1652
static void
R
Roman Bogorodskiy 已提交
1653
networkRemoveMasqueradingIptablesRules(virNetworkObjPtr network,
1654 1655 1656
                                       virNetworkIpDefPtr ipdef)
{
    int prefix = virNetworkIpDefPrefix(ipdef);
1657
    const char *forwardIf = virNetworkDefForwardIf(network->def, 0);
1658 1659

    if (prefix >= 0) {
R
Roman Bogorodskiy 已提交
1660
        iptablesRemoveForwardMasquerade(&ipdef->address,
1661
                                        prefix,
1662
                                        forwardIf,
1663 1664
                                        &network->def->forward.addr,
                                        &network->def->forward.port,
1665
                                        "tcp");
R
Roman Bogorodskiy 已提交
1666
        iptablesRemoveForwardMasquerade(&ipdef->address,
1667
                                        prefix,
1668
                                        forwardIf,
1669 1670
                                        &network->def->forward.addr,
                                        &network->def->forward.port,
1671
                                        "udp");
R
Roman Bogorodskiy 已提交
1672
        iptablesRemoveForwardMasquerade(&ipdef->address,
1673
                                        prefix,
1674
                                        forwardIf,
1675 1676
                                        &network->def->forward.addr,
                                        &network->def->forward.port,
1677 1678
                                        NULL);

R
Roman Bogorodskiy 已提交
1679
        iptablesRemoveForwardAllowRelatedIn(&ipdef->address,
1680 1681
                                            prefix,
                                            network->def->bridge,
1682
                                            forwardIf);
R
Roman Bogorodskiy 已提交
1683
        iptablesRemoveForwardAllowOut(&ipdef->address,
1684 1685
                                      prefix,
                                      network->def->bridge,
1686
                                      forwardIf);
1687 1688 1689
    }
}

1690
static int
R
Roman Bogorodskiy 已提交
1691
networkAddRoutingIptablesRules(virNetworkObjPtr network,
1692 1693
                               virNetworkIpDefPtr ipdef)
{
1694
    int prefix = virNetworkIpDefPrefix(ipdef);
1695
    const char *forwardIf = virNetworkDefForwardIf(network->def, 0);
1696 1697

    if (prefix < 0) {
1698 1699 1700
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Invalid prefix or netmask for '%s'"),
                       network->def->bridge);
1701 1702
        goto routeerr1;
    }
1703

1704
    /* allow routing packets from the bridge interface */
R
Roman Bogorodskiy 已提交
1705
    if (iptablesAddForwardAllowOut(&ipdef->address,
1706
                                   prefix,
1707
                                   network->def->bridge,
1708
                                   forwardIf) < 0) {
1709 1710 1711
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow routing from '%s'"),
                       network->def->bridge);
1712 1713 1714 1715
        goto routeerr1;
    }

    /* allow routing packets to the bridge interface */
R
Roman Bogorodskiy 已提交
1716
    if (iptablesAddForwardAllowIn(&ipdef->address,
1717
                                  prefix,
1718
                                  network->def->bridge,
1719
                                  forwardIf) < 0) {
1720 1721 1722
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow routing to '%s'"),
                       network->def->bridge);
1723 1724 1725
        goto routeerr2;
    }

1726
    return 0;
1727

1728
routeerr2:
R
Roman Bogorodskiy 已提交
1729
    iptablesRemoveForwardAllowOut(&ipdef->address,
1730
                                  prefix,
1731
                                  network->def->bridge,
1732
                                  forwardIf);
1733
routeerr1:
1734
    return -1;
1735 1736
}

1737
static void
R
Roman Bogorodskiy 已提交
1738
networkRemoveRoutingIptablesRules(virNetworkObjPtr network,
1739 1740 1741
                                  virNetworkIpDefPtr ipdef)
{
    int prefix = virNetworkIpDefPrefix(ipdef);
1742
    const char *forwardIf = virNetworkDefForwardIf(network->def, 0);
1743 1744

    if (prefix >= 0) {
R
Roman Bogorodskiy 已提交
1745
        iptablesRemoveForwardAllowIn(&ipdef->address,
1746 1747
                                     prefix,
                                     network->def->bridge,
1748
                                     forwardIf);
1749

R
Roman Bogorodskiy 已提交
1750
        iptablesRemoveForwardAllowOut(&ipdef->address,
1751 1752
                                      prefix,
                                      network->def->bridge,
1753
                                      forwardIf);
1754 1755 1756
    }
}

1757 1758
/* Add all once/network rules required for IPv6.
 * If no IPv6 addresses are defined and <network ipv6='yes'> is
G
Gene Czarcinski 已提交
1759 1760
 * specified, then allow IPv6 commuinications between virtual systems.
 * If any IPv6 addresses are defined, then add the rules for regular operation.
1761
 */
1762
static int
R
Roman Bogorodskiy 已提交
1763
networkAddGeneralIp6tablesRules(virNetworkObjPtr network)
1764 1765
{

1766 1767
    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0) &&
        !network->def->ipv6nogw) {
1768
        return 0;
1769
    }
1770 1771 1772

    /* Catch all rules to block forwarding to/from bridges */

R
Roman Bogorodskiy 已提交
1773
    if (iptablesAddForwardRejectOut(AF_INET6, network->def->bridge) < 0) {
1774 1775 1776
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to block outbound traffic from '%s'"),
                       network->def->bridge);
1777 1778 1779
        goto err1;
    }

R
Roman Bogorodskiy 已提交
1780
    if (iptablesAddForwardRejectIn(AF_INET6, network->def->bridge) < 0) {
1781 1782 1783
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to block inbound traffic to '%s'"),
                       network->def->bridge);
1784 1785 1786 1787
        goto err2;
    }

    /* Allow traffic between guests on the same bridge */
R
Roman Bogorodskiy 已提交
1788
    if (iptablesAddForwardAllowCross(AF_INET6, network->def->bridge) < 0) {
1789 1790 1791
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to allow cross bridge traffic on '%s'"),
                       network->def->bridge);
1792 1793 1794
        goto err3;
    }

1795 1796 1797 1798
    /* if no IPv6 addresses are defined, we are done. */
    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0))
        return 0;

1799
    /* allow DNS over IPv6 */
R
Roman Bogorodskiy 已提交
1800
    if (iptablesAddTcpInput(AF_INET6, network->def->bridge, 53) < 0) {
1801 1802 1803
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to allow DNS requests from '%s'"),
                       network->def->bridge);
1804 1805 1806
        goto err4;
    }

R
Roman Bogorodskiy 已提交
1807
    if (iptablesAddUdpInput(AF_INET6, network->def->bridge, 53) < 0) {
1808 1809 1810
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to allow DNS requests from '%s'"),
                       network->def->bridge);
1811 1812 1813
        goto err5;
    }

R
Roman Bogorodskiy 已提交
1814
    if (iptablesAddUdpInput(AF_INET6, network->def->bridge, 547) < 0) {
G
Gene Czarcinski 已提交
1815 1816 1817 1818 1819 1820
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to allow DHCP6 requests from '%s'"),
                       network->def->bridge);
        goto err6;
    }

1821 1822 1823
    return 0;

    /* unwind in reverse order from the point of failure */
G
Gene Czarcinski 已提交
1824
err6:
R
Roman Bogorodskiy 已提交
1825
    iptablesRemoveUdpInput(AF_INET6, network->def->bridge, 53);
1826
err5:
R
Roman Bogorodskiy 已提交
1827
    iptablesRemoveTcpInput(AF_INET6, network->def->bridge, 53);
1828
err4:
R
Roman Bogorodskiy 已提交
1829
    iptablesRemoveForwardAllowCross(AF_INET6, network->def->bridge);
1830
err3:
R
Roman Bogorodskiy 已提交
1831
    iptablesRemoveForwardRejectIn(AF_INET6, network->def->bridge);
1832
err2:
R
Roman Bogorodskiy 已提交
1833
    iptablesRemoveForwardRejectOut(AF_INET6, network->def->bridge);
1834 1835 1836 1837 1838
err1:
    return -1;
}

static void
R
Roman Bogorodskiy 已提交
1839
networkRemoveGeneralIp6tablesRules(virNetworkObjPtr network)
1840
{
1841
    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0) &&
G
Gene Czarcinski 已提交
1842
        !network->def->ipv6nogw) {
1843
        return;
G
Gene Czarcinski 已提交
1844
    }
1845
    if (virNetworkDefGetIpByIndex(network->def, AF_INET6, 0)) {
R
Roman Bogorodskiy 已提交
1846 1847 1848
        iptablesRemoveUdpInput(AF_INET6, network->def->bridge, 547);
        iptablesRemoveUdpInput(AF_INET6, network->def->bridge, 53);
        iptablesRemoveTcpInput(AF_INET6, network->def->bridge, 53);
1849
    }
1850

1851 1852 1853
    /* the following rules are there if no IPv6 address has been defined
     * but network->def->ipv6nogw == true
     */
R
Roman Bogorodskiy 已提交
1854 1855 1856
    iptablesRemoveForwardAllowCross(AF_INET6, network->def->bridge);
    iptablesRemoveForwardRejectIn(AF_INET6, network->def->bridge);
    iptablesRemoveForwardRejectOut(AF_INET6, network->def->bridge);
1857 1858
}

1859
static int
R
Roman Bogorodskiy 已提交
1860
networkAddGeneralIptablesRules(virNetworkObjPtr network)
1861
{
1862
    size_t i;
1863 1864 1865 1866
    virNetworkIpDefPtr ipv4def;

    /* First look for first IPv4 address that has dhcp or tftpboot defined. */
    /* We support dhcp config on 1 IPv4 interface only. */
1867 1868 1869
    for (i = 0;
         (ipv4def = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
1870 1871 1872
        if (ipv4def->nranges || ipv4def->nhosts || ipv4def->tftproot)
            break;
    }
1873 1874

    /* allow DHCP requests through to dnsmasq */
1875

R
Roman Bogorodskiy 已提交
1876
    if (iptablesAddTcpInput(AF_INET, network->def->bridge, 67) < 0) {
1877 1878 1879
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow DHCP requests from '%s'"),
                       network->def->bridge);
1880 1881 1882
        goto err1;
    }

R
Roman Bogorodskiy 已提交
1883
    if (iptablesAddUdpInput(AF_INET, network->def->bridge, 67) < 0) {
1884 1885 1886
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow DHCP requests from '%s'"),
                       network->def->bridge);
1887 1888 1889
        goto err2;
    }

1890 1891 1892 1893 1894 1895 1896
    /* If we are doing local DHCP service on this network, attempt to
     * add a rule that will fixup the checksum of DHCP response
     * packets back to the guests (but report failure without
     * aborting, since not all iptables implementations support it).
     */

    if (ipv4def && (ipv4def->nranges || ipv4def->nhosts) &&
R
Roman Bogorodskiy 已提交
1897
        (iptablesAddOutputFixUdpChecksum(network->def->bridge, 68) < 0)) {
1898 1899
        VIR_WARN("Could not add rule to fixup DHCP response checksums "
                 "on network '%s'.", network->def->name);
1900
        VIR_WARN("May need to update iptables package & kernel to support CHECKSUM rule.");
1901 1902
    }

1903
    /* allow DNS requests through to dnsmasq */
R
Roman Bogorodskiy 已提交
1904
    if (iptablesAddTcpInput(AF_INET, network->def->bridge, 53) < 0) {
1905 1906 1907
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow DNS requests from '%s'"),
                       network->def->bridge);
1908 1909 1910
        goto err3;
    }

R
Roman Bogorodskiy 已提交
1911
    if (iptablesAddUdpInput(AF_INET, network->def->bridge, 53) < 0) {
1912 1913 1914
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow DNS requests from '%s'"),
                       network->def->bridge);
1915 1916 1917
        goto err4;
    }

1918 1919
    /* allow TFTP requests through to dnsmasq if necessary */
    if (ipv4def && ipv4def->tftproot &&
R
Roman Bogorodskiy 已提交
1920
        iptablesAddUdpInput(AF_INET, network->def->bridge, 69) < 0) {
1921 1922 1923
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow TFTP requests from '%s'"),
                       network->def->bridge);
1924
        goto err5;
1925 1926
    }

1927 1928
    /* Catch all rules to block forwarding to/from bridges */

R
Roman Bogorodskiy 已提交
1929
    if (iptablesAddForwardRejectOut(AF_INET, network->def->bridge) < 0) {
1930 1931 1932
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to block outbound traffic from '%s'"),
                       network->def->bridge);
1933
        goto err6;
1934 1935
    }

R
Roman Bogorodskiy 已提交
1936
    if (iptablesAddForwardRejectIn(AF_INET, network->def->bridge) < 0) {
1937 1938 1939
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to block inbound traffic to '%s'"),
                       network->def->bridge);
1940
        goto err7;
1941 1942 1943
    }

    /* Allow traffic between guests on the same bridge */
R
Roman Bogorodskiy 已提交
1944
    if (iptablesAddForwardAllowCross(AF_INET, network->def->bridge) < 0) {
1945 1946 1947
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow cross bridge traffic on '%s'"),
                       network->def->bridge);
1948
        goto err8;
1949 1950
    }

1951
    /* add IPv6 general rules, if needed */
R
Roman Bogorodskiy 已提交
1952
    if (networkAddGeneralIp6tablesRules(network) < 0) {
1953 1954 1955
        goto err9;
    }

1956
    return 0;
1957

1958
    /* unwind in reverse order from the point of failure */
1959
err9:
R
Roman Bogorodskiy 已提交
1960
    iptablesRemoveForwardAllowCross(AF_INET, network->def->bridge);
1961
err8:
R
Roman Bogorodskiy 已提交
1962
    iptablesRemoveForwardRejectIn(AF_INET, network->def->bridge);
1963
err7:
R
Roman Bogorodskiy 已提交
1964
    iptablesRemoveForwardRejectOut(AF_INET, network->def->bridge);
1965 1966
err6:
    if (ipv4def && ipv4def->tftproot) {
R
Roman Bogorodskiy 已提交
1967
        iptablesRemoveUdpInput(AF_INET, network->def->bridge, 69);
1968
    }
1969
err5:
R
Roman Bogorodskiy 已提交
1970
    iptablesRemoveUdpInput(AF_INET, network->def->bridge, 53);
1971
err4:
R
Roman Bogorodskiy 已提交
1972
    iptablesRemoveTcpInput(AF_INET, network->def->bridge, 53);
1973
err3:
R
Roman Bogorodskiy 已提交
1974
    iptablesRemoveUdpInput(AF_INET, network->def->bridge, 67);
1975
err2:
R
Roman Bogorodskiy 已提交
1976
    iptablesRemoveTcpInput(AF_INET, network->def->bridge, 67);
1977
err1:
1978
    return -1;
1979 1980 1981
}

static void
R
Roman Bogorodskiy 已提交
1982
networkRemoveGeneralIptablesRules(virNetworkObjPtr network)
1983
{
1984
    size_t i;
1985
    virNetworkIpDefPtr ipv4def;
1986

R
Roman Bogorodskiy 已提交
1987
    networkRemoveGeneralIp6tablesRules(network);
1988

1989 1990 1991
    for (i = 0;
         (ipv4def = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
1992 1993
        if (ipv4def->nranges || ipv4def->nhosts || ipv4def->tftproot)
            break;
1994
    }
1995

R
Roman Bogorodskiy 已提交
1996 1997 1998
    iptablesRemoveForwardAllowCross(AF_INET, network->def->bridge);
    iptablesRemoveForwardRejectIn(AF_INET, network->def->bridge);
    iptablesRemoveForwardRejectOut(AF_INET, network->def->bridge);
1999
    if (ipv4def && ipv4def->tftproot) {
R
Roman Bogorodskiy 已提交
2000
        iptablesRemoveUdpInput(AF_INET, network->def->bridge, 69);
2001
    }
R
Roman Bogorodskiy 已提交
2002 2003
    iptablesRemoveUdpInput(AF_INET, network->def->bridge, 53);
    iptablesRemoveTcpInput(AF_INET, network->def->bridge, 53);
2004
    if (ipv4def && (ipv4def->nranges || ipv4def->nhosts)) {
R
Roman Bogorodskiy 已提交
2005
        iptablesRemoveOutputFixUdpChecksum(network->def->bridge, 68);
2006
    }
R
Roman Bogorodskiy 已提交
2007 2008
    iptablesRemoveUdpInput(AF_INET, network->def->bridge, 67);
    iptablesRemoveTcpInput(AF_INET, network->def->bridge, 67);
2009 2010
}

2011
static int
R
Roman Bogorodskiy 已提交
2012
networkAddIpSpecificIptablesRules(virNetworkObjPtr network,
2013 2014
                                  virNetworkIpDefPtr ipdef)
{
2015 2016 2017
    /* NB: in the case of IPv6, routing rules are added when the
     * forward mode is NAT. This is because IPv6 has no NAT.
     */
2018

2019
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) {
2020
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
R
Roman Bogorodskiy 已提交
2021
            return networkAddMasqueradingIptablesRules(network, ipdef);
2022
        else if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
R
Roman Bogorodskiy 已提交
2023
            return networkAddRoutingIptablesRules(network, ipdef);
2024
    } else if (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
R
Roman Bogorodskiy 已提交
2025
        return networkAddRoutingIptablesRules(network, ipdef);
2026
    }
2027 2028 2029 2030
    return 0;
}

static void
R
Roman Bogorodskiy 已提交
2031
networkRemoveIpSpecificIptablesRules(virNetworkObjPtr network,
2032 2033
                                     virNetworkIpDefPtr ipdef)
{
2034
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) {
2035
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
R
Roman Bogorodskiy 已提交
2036
            networkRemoveMasqueradingIptablesRules(network, ipdef);
2037
        else if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
R
Roman Bogorodskiy 已提交
2038
            networkRemoveRoutingIptablesRules(network, ipdef);
2039
    } else if (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
R
Roman Bogorodskiy 已提交
2040
        networkRemoveRoutingIptablesRules(network, ipdef);
2041
    }
2042 2043 2044 2045
}

/* Add all rules for all ip addresses (and general rules) on a network */
static int
R
Roman Bogorodskiy 已提交
2046
networkAddIptablesRules(virNetworkObjPtr network)
2047
{
2048
    size_t i, j;
2049
    virNetworkIpDefPtr ipdef;
2050
    virErrorPtr orig_error;
2051 2052

    /* Add "once per network" rules */
R
Roman Bogorodskiy 已提交
2053
    if (networkAddGeneralIptablesRules(network) < 0)
2054 2055
        return -1;

2056 2057 2058
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
2059
        /* Add address-specific iptables rules */
R
Roman Bogorodskiy 已提交
2060
        if (networkAddIpSpecificIptablesRules(network, ipdef) < 0) {
2061 2062 2063 2064 2065 2066
            goto err;
        }
    }
    return 0;

err:
2067 2068 2069
    /* store the previous error message before attempting removal of rules */
    orig_error = virSaveLastError();

2070 2071 2072 2073
    /* The final failed call to networkAddIpSpecificIptablesRules will
     * have removed any rules it created, but we need to remove those
     * added for previous IP addresses.
     */
2074 2075 2076
    for (j = 0; j < i; j++) {
        if ((ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, j)))
            networkRemoveIpSpecificIptablesRules(network, ipdef);
2077
    }
R
Roman Bogorodskiy 已提交
2078
    networkRemoveGeneralIptablesRules(network);
2079 2080 2081

    /* return the original error */
    virSetError(orig_error);
2082
    virFreeError(orig_error);
2083 2084 2085 2086 2087
    return -1;
}

/* Remove all rules for all ip addresses (and general rules) on a network */
static void
R
Roman Bogorodskiy 已提交
2088
networkRemoveIptablesRules(virNetworkObjPtr network)
2089
{
2090
    size_t i;
2091 2092
    virNetworkIpDefPtr ipdef;

2093 2094 2095
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
R
Roman Bogorodskiy 已提交
2096
        networkRemoveIpSpecificIptablesRules(network, ipdef);
2097
    }
R
Roman Bogorodskiy 已提交
2098
    networkRemoveGeneralIptablesRules(network);
2099 2100
}

2101
static void
2102
networkReloadIptablesRules(virNetworkDriverStatePtr driver)
2103
{
2104
    size_t i;
2105

2106
    VIR_INFO("Reloading iptables rules");
2107

2108
    for (i = 0; i < driver->networks.count; i++) {
2109 2110 2111 2112
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
2113 2114 2115
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
2116 2117 2118
            /* Only the three L3 network types that are configured by libvirt
             * need to have iptables rules reloaded.
             */
R
Roman Bogorodskiy 已提交
2119 2120
            networkRemoveIptablesRules(network);
            if (networkAddIptablesRules(network) < 0) {
2121 2122
                /* failed to add but already logged */
            }
2123
        }
2124
        virNetworkObjUnlock(network);
2125 2126 2127
    }
}

2128
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
2129
static int
2130
networkEnableIpForwarding(bool enableIPv4, bool enableIPv6)
2131
{
2132 2133 2134 2135 2136 2137
    int ret = 0;
    if (enableIPv4)
        ret = virFileWriteStr("/proc/sys/net/ipv4/ip_forward", "1\n", 0);
    if (enableIPv6 && ret == 0)
        ret = virFileWriteStr("/proc/sys/net/ipv6/conf/all/forwarding", "1\n", 0);
    return ret;
2138 2139
}

2140 2141
#define SYSCTL_PATH "/proc/sys"

2142 2143
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
2144 2145 2146 2147
{
    char *field = NULL;
    int ret = -1;

2148 2149 2150 2151 2152
    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0)) {
        /* Only set disable_ipv6 if there are no ipv6 addresses defined for
         * the network.
         */
        if (virAsprintf(&field, SYSCTL_PATH "/net/ipv6/conf/%s/disable_ipv6",
2153
                        network->def->bridge) < 0)
2154
            goto cleanup;
2155

2156 2157 2158 2159 2160 2161
        if (access(field, W_OK) < 0 && errno == ENOENT) {
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
            ret = 0;
            goto cleanup;
        }
2162

2163 2164 2165 2166 2167 2168 2169
        if (virFileWriteStr(field, "1", 0) < 0) {
            virReportSystemError(errno,
                                 _("cannot write to %s to disable IPv6 on bridge %s"),
                                 field, network->def->bridge);
            goto cleanup;
        }
        VIR_FREE(field);
2170 2171
    }

2172 2173 2174 2175 2176 2177 2178 2179
    /* The rest of the ipv6 sysctl tunables should always be set,
     * whether or not we're using ipv6 on this bridge.
     */

    /* Prevent guests from hijacking the host network by sending out
     * their own router advertisements.
     */
    if (virAsprintf(&field, SYSCTL_PATH "/net/ipv6/conf/%s/accept_ra",
2180
                    network->def->bridge) < 0)
2181 2182
        goto cleanup;

2183
    if (virFileWriteStr(field, "0", 0) < 0) {
2184
        virReportSystemError(errno,
2185 2186 2187 2188 2189
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

2190 2191 2192 2193
    /* All interfaces used as a gateway (which is what this is, by
     * definition), must always have autoconf=0.
     */
    if (virAsprintf(&field, SYSCTL_PATH "/net/ipv6/conf/%s/autoconf",
2194
                    network->def->bridge) < 0)
2195 2196
        goto cleanup;

2197
    if (virFileWriteStr(field, "0", 0) < 0) {
2198
        virReportSystemError(errno,
2199
                             _("cannot disable %s"), field);
2200 2201 2202 2203 2204 2205 2206 2207 2208
        goto cleanup;
    }

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

2209 2210 2211 2212 2213 2214
#define PROC_NET_ROUTE "/proc/net/route"

/* XXX: This function can be a lot more exhaustive, there are certainly
 *      other scenarios where we can ruin host network connectivity.
 * XXX: Using a proper library is preferred over parsing /proc
 */
2215 2216
static int
networkCheckRouteCollision(virNetworkObjPtr network)
2217
{
2218
    int ret = 0, len;
2219
    char *cur, *buf = NULL;
2220 2221
    /* allow for up to 100000 routes (each line is 128 bytes) */
    enum {MAX_ROUTE_SIZE = 128*100000};
2222 2223 2224

    /* Read whole routing table into memory */
    if ((len = virFileReadAll(PROC_NET_ROUTE, MAX_ROUTE_SIZE, &buf)) < 0)
2225
        goto out;
2226 2227 2228 2229 2230 2231 2232

    /* Dropping the last character shouldn't hurt */
    if (len > 0)
        buf[len-1] = '\0';

    VIR_DEBUG("%s output:\n%s", PROC_NET_ROUTE, buf);

2233
    if (!STRPREFIX(buf, "Iface"))
2234 2235 2236 2237 2238 2239 2240 2241 2242 2243
        goto out;

    /* First line is just headings, skip it */
    cur = strchr(buf, '\n');
    if (cur)
        cur++;

    while (cur) {
        char iface[17], dest[128], mask[128];
        unsigned int addr_val, mask_val;
2244
        virNetworkIpDefPtr ipdef;
2245 2246
        int num;
        size_t i;
2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274

        /* NUL-terminate the line, so sscanf doesn't go beyond a newline.  */
        char *nl = strchr(cur, '\n');
        if (nl) {
            *nl++ = '\0';
        }

        num = sscanf(cur, "%16s %127s %*s %*s %*s %*s %*s %127s",
                     iface, dest, mask);
        cur = nl;

        if (num != 3) {
            VIR_DEBUG("Failed to parse %s", PROC_NET_ROUTE);
            continue;
        }

        if (virStrToLong_ui(dest, NULL, 16, &addr_val) < 0) {
            VIR_DEBUG("Failed to convert network address %s to uint", dest);
            continue;
        }

        if (virStrToLong_ui(mask, NULL, 16, &mask_val) < 0) {
            VIR_DEBUG("Failed to convert network mask %s to uint", mask);
            continue;
        }

        addr_val &= mask_val;

2275 2276 2277
        for (i = 0;
             (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
             i++) {
2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292

            unsigned int net_dest;
            virSocketAddr netmask;

            if (virNetworkIpDefNetmask(ipdef, &netmask) < 0) {
                VIR_WARN("Failed to get netmask of '%s'",
                         network->def->bridge);
                continue;
            }

            net_dest = (ipdef->address.data.inet4.sin_addr.s_addr &
                        netmask.data.inet4.sin_addr.s_addr);

            if ((net_dest == addr_val) &&
                (netmask.data.inet4.sin_addr.s_addr == mask_val)) {
2293 2294 2295
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Network is already in use by interface %s"),
                               iface);
2296 2297 2298
                ret = -1;
                goto out;
            }
2299 2300 2301 2302 2303 2304 2305 2306
        }
    }

out:
    VIR_FREE(buf);
    return ret;
}

2307
/* add an IP address to a bridge */
2308
static int
D
Daniel P. Berrange 已提交
2309
networkAddAddrToBridge(virNetworkObjPtr network,
2310
                       virNetworkIpDefPtr ipdef)
2311
{
2312 2313 2314
    int prefix = virNetworkIpDefPrefix(ipdef);

    if (prefix < 0) {
2315 2316 2317
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
2318 2319 2320
        return -1;
    }

2321 2322
    if (virNetDevSetIPv4Address(network->def->bridge,
                                &ipdef->address, prefix) < 0)
2323 2324 2325 2326 2327
        return -1;

    return 0;
}

2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376
/* add an IP (static) route to a bridge */
static int
networkAddRouteToBridge(virNetworkObjPtr network,
                        virNetworkRouteDefPtr routedef)
{
    int prefix = 0;
    unsigned int metric;
    virSocketAddrPtr addr = &routedef->address;
    virSocketAddrPtr mask = &routedef->netmask;
    virSocketAddr zero;

    /* this creates an all-0 address of the appropriate family */
    ignore_value(virSocketAddrParse(&zero,
                                    (VIR_SOCKET_ADDR_IS_FAMILY(addr,AF_INET)
                                     ? "0.0.0.0" : "::"),
                                    VIR_SOCKET_ADDR_FAMILY(addr)));

    if (virSocketAddrEqual(addr, &zero)) {
        if (routedef->has_prefix && routedef->prefix == 0)
            prefix = 0;
        else if ((VIR_SOCKET_ADDR_IS_FAMILY(mask, AF_INET) &&
                virSocketAddrEqual(mask, &zero)))
            prefix = 0;
        else
            prefix = virSocketAddrGetIpPrefix(addr, mask, routedef->prefix);
    } else {
        prefix = virSocketAddrGetIpPrefix(addr, mask, routedef->prefix);
    }

    if (prefix < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' has an invalid netmask "
                         "or IP address in route definition"),
                       network->def->name);
        return -1;
    }

    if (routedef->has_metric && routedef->metric > 0)
        metric = routedef->metric;
    else
        metric = 1;

    if (virNetDevAddRoute(network->def->bridge, &routedef->address,
                          prefix, &routedef->gateway, metric) < 0) {
        return -1;
    }
    return 0;
}

2377
static int
2378
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
2379 2380
                          virNetworkObjPtr network)
{
2381
    size_t i;
2382
    bool v4present = false, v6present = false;
2383 2384
    virErrorPtr save_err = NULL;
    virNetworkIpDefPtr ipdef;
2385
    virNetworkRouteDefPtr routedef;
2386
    char *macTapIfName = NULL;
2387
    int tapfd = -1;
2388

2389 2390
    /* Check to see if any network IP collides with an existing route */
    if (networkCheckRouteCollision(network) < 0)
2391 2392
        return -1;

2393
    /* Create and configure the bridge device */
2394
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
2395 2396
        return -1;

2397 2398 2399 2400 2401 2402 2403 2404
    if (network->def->mac_specified) {
        /* To set a mac for the bridge, we need to define a dummy tap
         * device, set its mac, then attach it to the bridge. As long
         * as its mac address is lower than any other interface that
         * gets attached, the bridge will always maintain this mac
         * address.
         */
        macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2405
        if (!macTapIfName)
2406
            goto err0;
2407
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2408
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
2409
                                           &macTapIfName, &network->def->mac,
2410
                                           NULL, &tapfd, 1, NULL, NULL,
2411 2412 2413
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2414 2415 2416 2417 2418
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2419
    /* Set bridge options */
2420 2421 2422 2423

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2424
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2425
                                   network->def->delay * 1000) < 0)
2426
        goto err1;
2427

2428
    if (virNetDevBridgeSetSTP(network->def->bridge,
2429
                              network->def->stp ? true : false) < 0)
2430
        goto err1;
2431

2432 2433 2434 2435
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2436
        goto err1;
2437

2438
    /* Add "once per network" rules */
R
Roman Bogorodskiy 已提交
2439
    if (networkAddIptablesRules(network) < 0)
2440 2441
        goto err1;

2442 2443 2444
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
2445
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2446
            v4present = true;
2447
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2448
            v6present = true;
2449

2450
        /* Add the IP address/netmask to the bridge */
D
Daniel P. Berrange 已提交
2451
        if (networkAddAddrToBridge(network, ipdef) < 0) {
2452
            goto err2;
2453
        }
2454 2455
    }

2456
    /* Bring up the bridge interface */
2457
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2458
        goto err2;
2459

2460 2461
    for (i = 0; i < network->def->nroutes; i++) {
        routedef = &network->def->routes[i];
2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
        if (VIR_SOCKET_ADDR_VALID(&routedef->gateway)) {
            if (networkAddRouteToBridge(network, routedef) < 0) {
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2473 2474
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2475
        networkEnableIpForwarding(v4present, v6present) < 0) {
2476
        virReportSystemError(errno, "%s",
2477
                             _("failed to enable IP forwarding"));
2478
        goto err3;
2479 2480
    }

2481

2482
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2483 2484
    if ((v4present || v6present) &&
        networkStartDhcpDaemon(driver, network) < 0)
2485
        goto err3;
2486

2487
    /* start radvd if there are any ipv6 addresses */
G
Gene Czarcinski 已提交
2488
    if (v6present && networkStartRadvd(driver, network) < 0)
2489 2490
        goto err4;

2491 2492 2493 2494 2495 2496 2497 2498 2499
    /* DAD has happened (dnsmasq waits for it), dnsmasq is now bound to the
     * bridge's IPv6 address, so we can now set the dummy tun down.
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
            goto err4;
        VIR_FORCE_CLOSE(tapfd);
    }

2500 2501
    if (virNetDevBandwidthSet(network->def->bridge,
                              network->def->bandwidth, true) < 0) {
2502 2503 2504
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("cannot set bandwidth limits on %s"),
                       network->def->bridge);
2505 2506 2507
        goto err5;
    }

2508
    VIR_FREE(macTapIfName);
2509 2510 2511

    return 0;

2512
 err5:
2513
    virNetDevBandwidthClear(network->def->bridge);
2514

2515 2516 2517 2518
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2519 2520 2521 2522 2523
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2524 2525 2526
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2527
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2528

2529 2530 2531
 err2:
    if (!save_err)
        save_err = virSaveLastError();
R
Roman Bogorodskiy 已提交
2532
    networkRemoveIptablesRules(network);
2533 2534

 err1:
2535 2536 2537
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2538
    if (macTapIfName) {
2539
        VIR_FORCE_CLOSE(tapfd);
H
Hu Tao 已提交
2540 2541 2542
        ignore_value(virNetDevTapDelete(macTapIfName));
        VIR_FREE(macTapIfName);
    }
2543 2544

 err0:
2545 2546
    if (!save_err)
        save_err = virSaveLastError();
2547
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2548

2549 2550 2551 2552
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2553
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2554 2555 2556
    return -1;
}

2557
static int networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
R
Roman Bogorodskiy 已提交
2558
                                         virNetworkObjPtr network)
2559
{
2560
    virNetDevBandwidthClear(network->def->bridge);
2561

2562 2563 2564 2565 2566
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2567
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2568
            virPidFileDelete(driverState->pidDir, radvdpidbase);
2569 2570 2571 2572
            VIR_FREE(radvdpidbase);
        }
    }

2573 2574 2575
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2576
    if (network->def->mac_specified) {
2577
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2578
        if (macTapIfName) {
2579
            ignore_value(virNetDevTapDelete(macTapIfName));
2580 2581 2582 2583
            VIR_FREE(macTapIfName);
        }
    }

2584
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2585

R
Roman Bogorodskiy 已提交
2586
    networkRemoveIptablesRules(network);
2587

2588
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2589

2590
    /* See if its still alive and really really kill it */
2591
    if (network->dnsmasqPid > 0 &&
2592
        (kill(network->dnsmasqPid, 0) == 0))
2593 2594
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2595 2596 2597 2598 2599 2600

    if (network->radvdPid > 0 &&
        (kill(network->radvdPid, 0) == 0))
        kill(network->radvdPid, SIGKILL);
    network->radvdPid = -1;

2601 2602 2603 2604
    return 0;
}

static int
2605
networkStartNetworkExternal(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
2606 2607 2608
                            virNetworkObjPtr network ATTRIBUTE_UNUSED)
{
    /* put anything here that needs to be done each time a network of
2609
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2610 2611 2612 2613 2614 2615
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

2616
static int networkShutdownNetworkExternal(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
2617 2618 2619
                                        virNetworkObjPtr network ATTRIBUTE_UNUSED)
{
    /* put anything here that needs to be done each time a network of
2620
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2621 2622 2623 2624 2625 2626 2627
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2628
networkStartNetwork(virNetworkDriverStatePtr driver,
2629 2630 2631 2632 2633
                    virNetworkObjPtr network)
{
    int ret = 0;

    if (virNetworkObjIsActive(network)) {
2634 2635
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2636 2637 2638
        return -1;
    }

2639 2640 2641
    if (virNetworkObjSetDefTransient(network, true) < 0)
        return -1;

2642
    switch (network->def->forward.type) {
2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
        ret = networkStartNetworkVirtual(driver, network);
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2654
    case VIR_NETWORK_FORWARD_HOSTDEV:
2655 2656 2657 2658
        ret = networkStartNetworkExternal(driver, network);
        break;
    }

2659 2660
    if (ret < 0) {
        virNetworkObjUnsetDefTransient(network);
2661
        return ret;
2662
    }
2663 2664 2665 2666

    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2667 2668
    if ((ret = virNetworkSaveStatus(driverState->stateDir,
                                    network)) < 0) {
2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686
        goto error;
    }

    VIR_INFO("Starting up network '%s'", network->def->name);
    network->active = 1;

error:
    if (ret < 0) {
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
        networkShutdownNetwork(driver, network);
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2687
static int networkShutdownNetwork(virNetworkDriverStatePtr driver,
2688 2689 2690 2691 2692 2693 2694 2695 2696 2697
                                        virNetworkObjPtr network)
{
    int ret = 0;
    char *stateFile;

    VIR_INFO("Shutting down network '%s'", network->def->name);

    if (!virNetworkObjIsActive(network))
        return 0;

2698 2699
    stateFile = virNetworkConfigFile(driverState->stateDir,
                                     network->def->name);
2700 2701 2702 2703 2704 2705
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2706
    switch (network->def->forward.type) {
2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
        ret = networkShutdownNetworkVirtual(driver, network);
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2718
    case VIR_NETWORK_FORWARD_HOSTDEV:
2719 2720 2721 2722
        ret = networkShutdownNetworkExternal(driver, network);
        break;
    }

2723
    network->active = 0;
2724
    virNetworkObjUnsetDefTransient(network);
2725
    return ret;
2726 2727 2728
}


2729 2730
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
                                         const unsigned char *uuid) {
2731
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2732 2733
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2734

2735
    networkDriverLock(driver);
2736
    network = virNetworkFindByUUID(&driver->networks, uuid);
2737
    networkDriverUnlock(driver);
2738
    if (!network) {
2739 2740
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
2741
        goto cleanup;
2742 2743
    }

2744 2745 2746
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2747 2748 2749
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

cleanup:
2750 2751
    if (network)
        virNetworkObjUnlock(network);
2752
    return ret;
2753 2754
}

2755 2756
static virNetworkPtr networkLookupByName(virConnectPtr conn,
                                         const char *name) {
2757
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2758 2759 2760
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2761
    networkDriverLock(driver);
2762
    network = virNetworkFindByName(&driver->networks, name);
2763
    networkDriverUnlock(driver);
2764
    if (!network) {
2765 2766
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2767
        goto cleanup;
2768 2769
    }

2770 2771 2772
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2773 2774 2775
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

cleanup:
2776 2777
    if (network)
        virNetworkObjUnlock(network);
2778
    return ret;
2779 2780
}

2781 2782 2783
static virDrvOpenStatus networkOpen(virConnectPtr conn,
                                    virConnectAuthPtr auth ATTRIBUTE_UNUSED,
                                    unsigned int flags)
2784 2785 2786
{
    virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);

2787 2788 2789 2790 2791 2792 2793
    if (!driverState)
        return VIR_DRV_OPEN_DECLINED;

    conn->networkPrivateData = driverState;
    return VIR_DRV_OPEN_SUCCESS;
}

2794
static int networkClose(virConnectPtr conn) {
2795 2796 2797 2798
    conn->networkPrivateData = NULL;
    return 0;
}

2799
static int networkConnectNumOfNetworks(virConnectPtr conn) {
2800 2801
    int nactive = 0;
    size_t i;
2802
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2803

2804 2805 2806
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2807
    networkDriverLock(driver);
2808
    for (i = 0; i < driver->networks.count; i++) {
2809 2810 2811 2812
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectNumOfNetworksCheckACL(conn, obj->def) &&
            virNetworkObjIsActive(obj))
2813
            nactive++;
2814
        virNetworkObjUnlock(obj);
2815 2816
    }
    networkDriverUnlock(driver);
2817

2818 2819 2820
    return nactive;
}

2821
static int networkConnectListNetworks(virConnectPtr conn, char **const names, int nnames) {
2822
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2823 2824
    int got = 0;
    size_t i;
2825

2826 2827 2828
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2829
    networkDriverLock(driver);
2830
    for (i = 0; i < driver->networks.count && got < nnames; i++) {
2831 2832 2833 2834 2835 2836
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectListNetworksCheckACL(conn, obj->def) &&
            virNetworkObjIsActive(obj)) {
            if (VIR_STRDUP(names[got], obj->def->name) < 0) {
                virNetworkObjUnlock(obj);
2837 2838 2839 2840
                goto cleanup;
            }
            got++;
        }
2841
        virNetworkObjUnlock(obj);
2842
    }
2843 2844
    networkDriverUnlock(driver);

2845 2846 2847
    return got;

 cleanup:
2848
    networkDriverUnlock(driver);
2849
    for (i = 0; i < got; i++)
2850 2851 2852 2853
        VIR_FREE(names[i]);
    return -1;
}

2854
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn) {
2855 2856
    int ninactive = 0;
    size_t i;
2857
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2858

2859 2860 2861
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2862
    networkDriverLock(driver);
2863
    for (i = 0; i < driver->networks.count; i++) {
2864 2865 2866 2867
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectNumOfDefinedNetworksCheckACL(conn, obj->def) &&
            !virNetworkObjIsActive(obj))
2868
            ninactive++;
2869
        virNetworkObjUnlock(obj);
2870 2871
    }
    networkDriverUnlock(driver);
2872

2873 2874 2875
    return ninactive;
}

2876
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames) {
2877
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2878 2879
    int got = 0;
    size_t i;
2880

2881 2882 2883
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2884
    networkDriverLock(driver);
2885
    for (i = 0; i < driver->networks.count && got < nnames; i++) {
2886 2887 2888 2889 2890 2891
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectListDefinedNetworksCheckACL(conn, obj->def) &&
            !virNetworkObjIsActive(obj)) {
            if (VIR_STRDUP(names[got], obj->def->name) < 0) {
                virNetworkObjUnlock(obj);
2892 2893 2894 2895
                goto cleanup;
            }
            got++;
        }
2896
        virNetworkObjUnlock(obj);
2897
    }
2898
    networkDriverUnlock(driver);
2899 2900 2901
    return got;

 cleanup:
2902
    networkDriverUnlock(driver);
2903
    for (i = 0; i < got; i++)
2904 2905 2906 2907
        VIR_FREE(names[i]);
    return -1;
}

2908
static int
2909 2910 2911
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2912
{
2913
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2914 2915 2916 2917
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2918 2919 2920
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2921
    networkDriverLock(driver);
2922 2923 2924
    ret = virNetworkObjListExport(conn, driver->networks, nets,
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2925 2926
    networkDriverUnlock(driver);

2927
cleanup:
2928 2929
    return ret;
}
2930 2931 2932

static int networkIsActive(virNetworkPtr net)
{
2933
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
2934 2935 2936 2937 2938 2939 2940
    virNetworkObjPtr obj;
    int ret = -1;

    networkDriverLock(driver);
    obj = virNetworkFindByUUID(&driver->networks, net->uuid);
    networkDriverUnlock(driver);
    if (!obj) {
2941
        virReportError(VIR_ERR_NO_NETWORK, NULL);
2942 2943
        goto cleanup;
    }
2944 2945 2946 2947

    if (virNetworkIsActiveEnsureACL(net->conn, obj->def) < 0)
        goto cleanup;

2948 2949 2950 2951 2952 2953 2954 2955 2956 2957
    ret = virNetworkObjIsActive(obj);

cleanup:
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}

static int networkIsPersistent(virNetworkPtr net)
{
2958
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
2959 2960 2961 2962 2963 2964 2965
    virNetworkObjPtr obj;
    int ret = -1;

    networkDriverLock(driver);
    obj = virNetworkFindByUUID(&driver->networks, net->uuid);
    networkDriverUnlock(driver);
    if (!obj) {
2966
        virReportError(VIR_ERR_NO_NETWORK, NULL);
2967 2968
        goto cleanup;
    }
2969 2970 2971 2972

    if (virNetworkIsPersistentEnsureACL(net->conn, obj->def) < 0)
        goto cleanup;

2973 2974 2975 2976 2977 2978 2979 2980 2981
    ret = obj->persistent;

cleanup:
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}


2982
static int
2983
networkValidate(virNetworkDriverStatePtr driver,
2984 2985
                virNetworkDefPtr def,
                bool check_active)
2986
{
2987
    size_t i;
2988 2989
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2990
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
2991
    bool ipv4def = false, ipv6def = false;
2992 2993 2994 2995 2996 2997 2998 2999

    /* check for duplicate networks */
    if (virNetworkObjIsDuplicate(&driver->networks, def, check_active) < 0)
        return -1;

    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
3000 3001 3002
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
3003 3004 3005 3006 3007

        if (virNetworkSetBridgeName(&driver->networks, def, 1))
            return -1;

        virNetworkSetBridgeMacAddr(def);
3008 3009 3010 3011 3012 3013 3014 3015 3016
    } else {
        /* They are also the only types that currently support setting
         * an IP address for the host-side device (bridge)
         */
        if (virNetworkDefGetIpByIndex(def, AF_UNSPEC, 0)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3017
                           virNetworkForwardTypeToString(def->forward.type));
3018 3019
            return -1;
        }
3020
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
3021 3022 3023 3024
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3025
                           virNetworkForwardTypeToString(def->forward.type));
3026 3027 3028 3029 3030 3031 3032
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3033
                           virNetworkForwardTypeToString(def->forward.type));
3034 3035
            return -1;
        }
3036 3037
    }

G
Gene Czarcinski 已提交
3038 3039 3040
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
3041 3042 3043
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
3044 3045 3046 3047 3048
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("Multiple IPv4 dhcp sections found -- "
3049 3050
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067
                    return -1;
                } else {
                    ipv4def = true;
                }
            }
        }
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("Multiple IPv6 dhcp sections found -- "
                                 "dhcp is supported only for a "
                                 "single IPv6 address on each network"));
                    return -1;
                } else {
                    ipv6def = true;
                }
3068 3069 3070
            }
        }
    }
3071 3072 3073 3074 3075 3076

    /* The only type of networks that currently support transparent
     * vlan configuration are those using hostdev sr-iov devices from
     * a pool, and those using an Open vSwitch bridge.
     */

3077
    vlanAllowed = ((def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
3078
                   def->virtPortProfile &&
3079 3080 3081
                   def->virtPortProfile->virtPortType
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) ||
                   def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV);
3082 3083

    vlanUsed = def->vlan.nTags > 0;
3084 3085
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3086 3087 3088 3089 3090
            /* anyone using this portgroup will get a vlan tag. Verify
             * that they will also be using an openvswitch connection,
             * as that is the only type of network that currently
             * supports a vlan tag.
             */
3091
            if (def->portGroups[i].virtPortProfile) {
3092
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3093
                    def->portGroups[i].virtPortProfile->virtPortType
3094 3095 3096 3097 3098 3099 3100
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3101
        }
3102
        if (def->portGroups[i].isDefault) {
3103 3104 3105 3106 3107
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3108
                               def->name, defaultPortGroup->name,
3109
                               def->portGroups[i].name);
3110
                return -1;
3111
            }
3112
            defaultPortGroup = &def->portGroups[i];
3113
        }
3114
    }
3115 3116 3117 3118 3119 3120 3121
    if (badVlanUse ||
        (vlanUsed && !vlanAllowed && !defaultPortGroup)) {
        /* NB: if defaultPortGroup is set, we don't directly look at
         * vlanUsed && !vlanAllowed, because the network will never be
         * used without having a portgroup added in, so all necessary
         * checks were done in the loop above.
         */
3122 3123 3124 3125 3126 3127 3128 3129 3130
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
    return 0;
}

3131
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml) {
3132
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
3133
    virNetworkDefPtr def;
3134
    virNetworkObjPtr network = NULL;
3135
    virNetworkPtr ret = NULL;
3136

3137 3138
    networkDriverLock(driver);

3139
    if (!(def = virNetworkDefParseString(xml)))
3140
        goto cleanup;
3141

3142 3143 3144
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3145
    if (networkValidate(driver, def, true) < 0)
3146 3147
       goto cleanup;

3148 3149 3150 3151
    /* NB: "live" is false because this transient network hasn't yet
     * been started
     */
    if (!(network = virNetworkAssignDef(&driver->networks, def, false)))
3152 3153
        goto cleanup;
    def = NULL;
3154

3155
    if (networkStartNetwork(driver, network) < 0) {
3156 3157
        virNetworkRemoveInactive(&driver->networks,
                                 network);
3158
        network = NULL;
3159
        goto cleanup;
3160 3161
    }

3162
    VIR_INFO("Creating network '%s'", network->def->name);
3163 3164 3165 3166
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

cleanup:
    virNetworkDefFree(def);
3167 3168 3169
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
3170
    return ret;
3171 3172
}

3173
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml) {
3174
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
3175
    virNetworkDefPtr def = NULL;
3176
    bool freeDef = true;
3177
    virNetworkObjPtr network = NULL;
3178
    virNetworkPtr ret = NULL;
3179

3180 3181
    networkDriverLock(driver);

3182
    if (!(def = virNetworkDefParseString(xml)))
3183
        goto cleanup;
3184

3185 3186 3187
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3188
    if (networkValidate(driver, def, false) < 0)
3189 3190
       goto cleanup;

3191 3192 3193 3194 3195 3196 3197 3198
    if ((network = virNetworkFindByName(&driver->networks, def->name))) {
        network->persistent = 1;
        if (virNetworkObjAssignDef(network, def, false) < 0)
            goto cleanup;
    } else {
        if (!(network = virNetworkAssignDef(&driver->networks, def, false)))
            goto cleanup;
    }
3199

3200 3201 3202
    /* define makes the network persistent - always */
    network->persistent = 1;

3203 3204
    /* def was asigned */
    freeDef = false;
3205 3206

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3207 3208 3209 3210 3211 3212 3213 3214
        if (!virNetworkObjIsActive(network)) {
            virNetworkRemoveInactive(&driver->networks, network);
            network = NULL;
            goto cleanup;
        }
        network->persistent = 0;
        virNetworkDefFree(network->newDef);
        network->newDef = NULL;
3215 3216 3217
        goto cleanup;
    }

3218 3219
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
3220 3221

cleanup:
3222 3223
    if (freeDef)
       virNetworkDefFree(def);
3224 3225 3226
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
3227
    return ret;
3228 3229
}

3230 3231
static int
networkUndefine(virNetworkPtr net) {
3232
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3233
    virNetworkObjPtr network;
3234
    int ret = -1;
3235
    bool active = false;
3236

3237 3238
    networkDriverLock(driver);

3239
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3240
    if (!network) {
3241 3242
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3243
        goto cleanup;
3244 3245
    }

3246 3247 3248
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3249 3250
    if (virNetworkObjIsActive(network))
        active = true;
3251

3252
    if (virNetworkDeleteConfig(driver->networkConfigDir,
3253 3254
                               driver->networkAutostartDir,
                               network) < 0)
3255
        goto cleanup;
3256

3257 3258
    /* make the network transient */
    network->persistent = 0;
3259
    network->autostart = 0;
3260 3261 3262
    virNetworkDefFree(network->newDef);
    network->newDef = NULL;

3263
    VIR_INFO("Undefining network '%s'", network->def->name);
3264 3265 3266 3267 3268
    if (!active) {
        if (networkRemoveInactive(driver, network) < 0) {
            network = NULL;
            goto cleanup;
        }
3269
        network = NULL;
3270 3271
    }

3272
    ret = 0;
3273

3274
cleanup:
3275 3276 3277
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
3278
    return ret;
3279 3280
}

3281 3282 3283 3284 3285 3286 3287 3288
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3289
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3290
    virNetworkObjPtr network = NULL;
3291 3292
    int isActive, ret = -1;
    size_t i;
3293 3294 3295
    virNetworkIpDefPtr ipdef;
    bool oldDhcpActive = false;

3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

    networkDriverLock(driver);

    network = virNetworkFindByUUID(&driver->networks, net->uuid);
    if (!network) {
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
        goto cleanup;
    }

3310 3311 3312
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3313
    /* see if we are listening for dhcp pre-modification */
3314 3315 3316
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
3317 3318 3319 3320 3321 3322
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3323 3324 3325 3326
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
    */
    isActive = virNetworkObjIsActive(network);
3327 3328
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

    /* update the network config in memory/on disk */
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0)
        goto cleanup;

    if (flags & VIR_NETWORK_UPDATE_AFFECT_CONFIG) {
        /* save updated persistent config to disk */
        if (virNetworkSaveConfig(driver->networkConfigDir,
                                 virNetworkObjGetPersistentDef(network)) < 0) {
            goto cleanup;
        }
    }

    if (isActive && (flags & VIR_NETWORK_UPDATE_AFFECT_LIVE)) {
        /* rewrite dnsmasq host files, restart dnsmasq, update iptables
         * rules, etc, according to which section was modified. Note that
         * some sections require multiple actions, so a single switch
         * statement is inadequate.
         */
        if (section == VIR_NETWORK_SECTION_BRIDGE ||
            section == VIR_NETWORK_SECTION_DOMAIN ||
            section == VIR_NETWORK_SECTION_IP ||
            section == VIR_NETWORK_SECTION_IP_DHCP_RANGE) {
            /* these sections all change things on the dnsmasq commandline,
             * so we need to kill and restart dnsmasq.
             */
3361
            if (networkRestartDhcpDaemon(driver, network) < 0)
3362 3363
                goto cleanup;

3364 3365 3366 3367 3368 3369 3370 3371
        } else if (section == VIR_NETWORK_SECTION_IP_DHCP_HOST) {
            /* if we previously weren't listening for dhcp and now we
             * are (or vice-versa) then we need to do a restart,
             * otherwise we just need to do a refresh (redo the config
             * files and send SIGHUP)
             */
            bool newDhcpActive = false;

3372 3373 3374
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
                 i++) {
3375 3376 3377 3378 3379 3380 3381
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3382 3383
                 networkRestartDhcpDaemon(driver, network) < 0) ||
                networkRefreshDhcpDaemon(driver, network) < 0) {
3384 3385 3386 3387
                goto cleanup;
            }

        } else if (section == VIR_NETWORK_SECTION_DNS_HOST ||
3388 3389 3390 3391 3392 3393
                   section == VIR_NETWORK_SECTION_DNS_TXT ||
                   section == VIR_NETWORK_SECTION_DNS_SRV) {
            /* these sections only change things in config files, so we
             * can just update the config files and send SIGHUP to
             * dnsmasq.
             */
3394
            if (networkRefreshDhcpDaemon(driver, network) < 0)
3395 3396 3397 3398 3399 3400 3401 3402
                goto cleanup;

        }

        if (section == VIR_NETWORK_SECTION_IP) {
            /* only a change in IP addresses will affect radvd, and all of radvd's
             * config is stored in the conf file which will be re-read with a SIGHUP.
             */
3403
            if (networkRefreshRadvd(driver, network) < 0)
3404 3405 3406
                goto cleanup;
        }

3407 3408 3409
        if ((section == VIR_NETWORK_SECTION_IP ||
             section == VIR_NETWORK_SECTION_FORWARD ||
             section == VIR_NETWORK_SECTION_FORWARD_INTERFACE) &&
3410 3411 3412
           (network->def->forward.type == VIR_NETWORK_FORWARD_NONE ||
            network->def->forward.type == VIR_NETWORK_FORWARD_NAT ||
            network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
3413
            /* these could affect the iptables rules */
R
Roman Bogorodskiy 已提交
3414 3415
            networkRemoveIptablesRules(network);
            if (networkAddIptablesRules(network) < 0)
3416 3417 3418 3419 3420
                goto cleanup;

        }

        /* save current network state to disk */
3421 3422
        if ((ret = virNetworkSaveStatus(driverState->stateDir,
                                        network)) < 0) {
3423
            goto cleanup;
3424
        }
3425 3426 3427 3428 3429 3430 3431 3432 3433
    }
    ret = 0;
cleanup:
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
    return ret;
}

3434
static int networkCreate(virNetworkPtr net) {
3435
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3436 3437
    virNetworkObjPtr network;
    int ret = -1;
3438

3439
    networkDriverLock(driver);
3440
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3441

3442
    if (!network) {
3443 3444
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3445
        goto cleanup;
3446 3447
    }

3448 3449 3450
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3451
    ret = networkStartNetwork(driver, network);
3452 3453

cleanup:
3454 3455
    if (network)
        virNetworkObjUnlock(network);
3456
    networkDriverUnlock(driver);
3457
    return ret;
3458 3459 3460
}

static int networkDestroy(virNetworkPtr net) {
3461
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3462 3463
    virNetworkObjPtr network;
    int ret = -1;
3464

3465
    networkDriverLock(driver);
3466
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3467

3468
    if (!network) {
3469 3470
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3471
        goto cleanup;
3472 3473
    }

3474 3475 3476
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3477
    if (!virNetworkObjIsActive(network)) {
3478 3479
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is not active"));
3480 3481 3482
        goto cleanup;
    }

3483 3484 3485
    if ((ret = networkShutdownNetwork(driver, network)) < 0)
        goto cleanup;

3486
    if (!network->persistent) {
3487 3488 3489 3490 3491
        if (networkRemoveInactive(driver, network) < 0) {
            network = NULL;
            ret = -1;
            goto cleanup;
        }
3492 3493
        network = NULL;
    }
3494

3495
cleanup:
3496 3497
    if (network)
        virNetworkObjUnlock(network);
3498
    networkDriverUnlock(driver);
3499 3500 3501
    return ret;
}

3502
static char *networkGetXMLDesc(virNetworkPtr net,
3503
                               unsigned int flags)
3504
{
3505
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3506
    virNetworkObjPtr network;
3507
    virNetworkDefPtr def;
3508
    char *ret = NULL;
3509

3510
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3511

3512
    networkDriverLock(driver);
3513
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3514 3515
    networkDriverUnlock(driver);

3516
    if (!network) {
3517 3518
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3519
        goto cleanup;
3520 3521
    }

3522 3523 3524
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3525 3526 3527 3528 3529 3530
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3531 3532

cleanup:
3533 3534
    if (network)
        virNetworkObjUnlock(network);
3535
    return ret;
3536 3537 3538
}

static char *networkGetBridgeName(virNetworkPtr net) {
3539
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3540 3541 3542
    virNetworkObjPtr network;
    char *bridge = NULL;

3543
    networkDriverLock(driver);
3544
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3545 3546
    networkDriverUnlock(driver);

3547
    if (!network) {
3548 3549
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching id"));
3550
        goto cleanup;
3551 3552
    }

3553 3554 3555
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3556
    if (!(network->def->bridge)) {
3557 3558 3559
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3560 3561 3562
        goto cleanup;
    }

3563
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3564 3565

cleanup:
3566 3567
    if (network)
        virNetworkObjUnlock(network);
3568 3569 3570 3571 3572
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
                             int *autostart) {
3573
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3574 3575
    virNetworkObjPtr network;
    int ret = -1;
3576

3577
    networkDriverLock(driver);
3578
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3579
    networkDriverUnlock(driver);
3580
    if (!network) {
3581 3582
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3583
        goto cleanup;
3584 3585
    }

3586 3587 3588
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3589
    *autostart = network->autostart;
3590
    ret = 0;
3591

3592
cleanup:
3593 3594
    if (network)
        virNetworkObjUnlock(network);
3595
    return ret;
3596 3597 3598
}

static int networkSetAutostart(virNetworkPtr net,
3599
                               int autostart) {
3600
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3601
    virNetworkObjPtr network;
3602
    char *configFile = NULL, *autostartLink = NULL;
3603
    int ret = -1;
3604

3605
    networkDriverLock(driver);
3606
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3607

3608
    if (!network) {
3609 3610
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3611
        goto cleanup;
3612 3613
    }

3614 3615 3616
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3617
    if (!network->persistent) {
3618 3619
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3620 3621 3622
        goto cleanup;
    }

3623 3624
    autostart = (autostart != 0);

3625
    if (network->autostart != autostart) {
3626
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3627
            goto cleanup;
3628
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3629 3630
            goto cleanup;

3631
        if (autostart) {
3632
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3633
                virReportSystemError(errno,
3634 3635
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3636 3637
                goto cleanup;
            }
3638

3639
            if (symlink(configFile, autostartLink) < 0) {
3640
                virReportSystemError(errno,
3641
                                     _("Failed to create symlink '%s' to '%s'"),
3642
                                     autostartLink, configFile);
3643 3644 3645
                goto cleanup;
            }
        } else {
3646
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3647
                virReportSystemError(errno,
3648
                                     _("Failed to delete symlink '%s'"),
3649
                                     autostartLink);
3650 3651
                goto cleanup;
            }
3652 3653
        }

3654
        network->autostart = autostart;
3655
    }
3656
    ret = 0;
3657

3658
cleanup:
3659 3660
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3661 3662
    if (network)
        virNetworkObjUnlock(network);
3663
    networkDriverUnlock(driver);
3664
    return ret;
3665 3666 3667 3668 3669
}


static virNetworkDriver networkDriver = {
    "Network",
3670 3671 3672 3673 3674 3675 3676
    .networkOpen = networkOpen, /* 0.2.0 */
    .networkClose = networkClose, /* 0.2.0 */
    .connectNumOfNetworks = networkConnectNumOfNetworks, /* 0.2.0 */
    .connectListNetworks = networkConnectListNetworks, /* 0.2.0 */
    .connectNumOfDefinedNetworks = networkConnectNumOfDefinedNetworks, /* 0.2.0 */
    .connectListDefinedNetworks = networkConnectListDefinedNetworks, /* 0.2.0 */
    .connectListAllNetworks = networkConnectListAllNetworks, /* 0.10.2 */
3677 3678
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3679 3680
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3681
    .networkUndefine = networkUndefine, /* 0.2.0 */
3682
    .networkUpdate = networkUpdate, /* 0.10.2 */
3683
    .networkCreate = networkCreate, /* 0.2.0 */
3684 3685 3686 3687 3688 3689 3690
    .networkDestroy = networkDestroy, /* 0.2.0 */
    .networkGetXMLDesc = networkGetXMLDesc, /* 0.2.0 */
    .networkGetBridgeName = networkGetBridgeName, /* 0.2.0 */
    .networkGetAutostart = networkGetAutostart, /* 0.2.1 */
    .networkSetAutostart = networkSetAutostart, /* 0.2.1 */
    .networkIsActive = networkIsActive, /* 0.7.3 */
    .networkIsPersistent = networkIsPersistent, /* 0.7.3 */
3691 3692 3693
};

static virStateDriver networkStateDriver = {
3694
    .name = "Network",
3695 3696 3697
    .stateInitialize  = networkStateInitialize,
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3698 3699 3700 3701 3702 3703 3704
};

int networkRegister(void) {
    virRegisterNetworkDriver(&networkDriver);
    virRegisterStateDriver(&networkStateDriver);
    return 0;
}
3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715

/********************************************************/

/* Private API to deal with logical switch capabilities.
 * These functions are exported so that other parts of libvirt can
 * call them, but are not part of the public API and not in the
 * driver's function table. If we ever have more than one network
 * driver, we will need to present these functions via a second
 * "backend" function table.
 */

3716 3717 3718 3719 3720 3721 3722 3723 3724
/* networkCreateInterfacePool:
 * @netdef: the original NetDef from the network
 *
 * Creates an implicit interface pool of VF's when a PF dev is given
 */
static int
networkCreateInterfacePool(virNetworkDefPtr netdef) {
    unsigned int num_virt_fns = 0;
    char **vfname = NULL;
3725
    virPCIDeviceAddressPtr *virt_fns;
3726 3727
    int ret = -1;
    size_t i;
3728

3729
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev,
3730
                                      &vfname, &virt_fns, &num_virt_fns)) < 0) {
3731 3732
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Could not get Virtual functions on %s"),
3733
                       netdef->forward.pfs->dev);
3734 3735 3736 3737 3738 3739
        goto finish;
    }

    if (num_virt_fns == 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("No Vf's present on SRIOV PF %s"),
3740
                       netdef->forward.pfs->dev);
3741 3742 3743
       goto finish;
    }

3744
    if (VIR_ALLOC_N(netdef->forward.ifs, num_virt_fns) < 0)
3745 3746
        goto finish;

3747
    netdef->forward.nifs = num_virt_fns;
3748

3749
    for (i = 0; i < netdef->forward.nifs; i++) {
3750 3751 3752 3753
        if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) ||
            (netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) ||
            (netdef->forward.type == VIR_NETWORK_FORWARD_VEPA) ||
            (netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH)) {
3754 3755 3756
            netdef->forward.ifs[i].type = VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV;
            if (vfname[i]) {
                if (VIR_STRDUP(netdef->forward.ifs[i].device.dev, vfname[i]) < 0)
3757
                    goto finish;
3758
            } else {
3759
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
3760
                               _("Direct mode types require interface names"));
3761 3762
                goto finish;
            }
3763
        }
3764
        else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3765
            /* VF's are always PCI devices */
3766 3767 3768 3769 3770
            netdef->forward.ifs[i].type = VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI;
            netdef->forward.ifs[i].device.pci.domain = virt_fns[i]->domain;
            netdef->forward.ifs[i].device.pci.bus = virt_fns[i]->bus;
            netdef->forward.ifs[i].device.pci.slot = virt_fns[i]->slot;
            netdef->forward.ifs[i].device.pci.function = virt_fns[i]->function;
3771
        }
3772 3773 3774 3775
    }

    ret = 0;
finish:
3776 3777 3778
    for (i = 0; i < num_virt_fns; i++) {
        VIR_FREE(vfname[i]);
        VIR_FREE(virt_fns[i]);
3779
    }
3780
    VIR_FREE(vfname);
3781
    VIR_FREE(virt_fns);
3782 3783 3784
    return ret;
}

3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798
/* networkAllocateActualDevice:
 * @iface: the original NetDef from the domain
 *
 * Looks up the network reference by iface, allocates a physical
 * device from that network (if appropriate), and returns with the
 * virDomainActualNetDef filled in accordingly. If there are no
 * changes to be made in the netdef, then just leave the actualdef
 * empty.
 *
 * Returns 0 on success, -1 on failure.
 */
int
networkAllocateActualDevice(virDomainNetDefPtr iface)
{
3799
    virNetworkDriverStatePtr driver = driverState;
3800 3801 3802
    enum virDomainNetType actualType = iface->type;
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
3803
    virNetDevBandwidthPtr bandwidth = NULL;
3804 3805 3806
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3807
    virNetworkForwardIfDefPtr dev = NULL;
3808
    size_t i;
3809 3810 3811
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3812
        goto validate;
3813 3814 3815 3816 3817 3818 3819 3820

    virDomainActualNetDefFree(iface->data.network.actual);
    iface->data.network.actual = NULL;

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
3821 3822 3823
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3824
        goto error;
3825 3826
    }
    netdef = network->def;
3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837

    /* portgroup can be present for any type of network, in particular
     * for bandwidth information, so we need to check for that and
     * fill it in appropriately for all forward types.
    */
    portgroup = virPortGroupFindByName(netdef, iface->data.network.portgroup);

    /* If there is already interface-specific bandwidth, just use that
     * (already in NetDef). Otherwise, if there is bandwidth info in
     * the portgroup, fill that into the ActualDef.
     */
3838 3839 3840 3841 3842 3843 3844

    if (iface->bandwidth)
        bandwidth = iface->bandwidth;
    else if (portgroup && portgroup->bandwidth)
        bandwidth = portgroup->bandwidth;

    if (bandwidth) {
3845 3846
        if (!iface->data.network.actual &&
            VIR_ALLOC(iface->data.network.actual) < 0)
3847
            goto error;
3848
        if (virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
3849
                                   bandwidth) < 0)
3850
            goto error;
3851 3852
    }

3853 3854 3855 3856 3857 3858 3859 3860 3861
    /* copy appropriate vlan info to actualNet */
    if (iface->vlan.nTags > 0)
        vlan = &iface->vlan;
    else if (portgroup && portgroup->vlan.nTags > 0)
        vlan = &portgroup->vlan;
    else if (netdef->vlan.nTags > 0)
        vlan = &netdef->vlan;

    if (vlan) {
3862 3863
        if (!iface->data.network.actual &&
            VIR_ALLOC(iface->data.network.actual) < 0)
3864 3865 3866 3867 3868
            goto error;
        if (virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
           goto error;
    }

3869 3870 3871
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
3872 3873 3874 3875 3876 3877
        /* for these forward types, the actual net type really *is*
         *NETWORK; we just keep the info from the portgroup in
         * iface->data.network.actual
        */
        if (iface->data.network.actual)
            iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
3878 3879 3880 3881

        if (networkPlugBandwidth(network, iface) < 0)
            goto error;

3882
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
3883
               netdef->bridge) {
3884 3885 3886 3887 3888

        /* <forward type='bridge'/> <bridge name='xxx'/>
         * is VIR_DOMAIN_NET_TYPE_BRIDGE
         */

3889 3890
        if (!iface->data.network.actual &&
            VIR_ALLOC(iface->data.network.actual) < 0)
3891
            goto error;
3892

3893
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
3894 3895
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
3896
            goto error;
3897

3898 3899 3900 3901 3902 3903 3904 3905
        /* merge virtualports from interface, network, and portgroup to
         * arrive at actual virtualport to use
         */
        if (virNetDevVPortProfileMerge3(&iface->data.network.actual->virtPortProfile,
                                        iface->virtPortProfile,
                                        netdef->virtPortProfile,
                                        portgroup
                                        ? portgroup->virtPortProfile : NULL) < 0) {
3906
            goto error;
3907 3908 3909 3910 3911 3912 3913 3914 3915 3916
        }
        virtport = iface->data.network.actual->virtPortProfile;
        if (virtport) {
            /* only type='openvswitch' is allowed for bridges */
            if (virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("<virtualport type='%s'> not supported for network "
                                 "'%s' which uses a bridge device"),
                               virNetDevVPortTypeToString(virtport->virtPortType),
                               netdef->name);
3917
                goto error;
3918 3919 3920
            }
        }

3921
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3922

3923 3924
        virDomainHostdevSubsysPciBackendType backend;

3925 3926
        if (!iface->data.network.actual &&
            VIR_ALLOC(iface->data.network.actual) < 0)
3927 3928 3929
            goto error;

        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
3930
        if (netdef->forward.npfs > 0 && netdef->forward.nifs <= 0 &&
3931 3932 3933 3934 3935
            networkCreateInterfacePool(netdef) < 0) {
            goto error;
        }

        /* pick first dev with 0 connections */
3936 3937 3938
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952
                break;
            }
        }
        if (!dev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' requires exclusive access "
                             "to interfaces, but none are available"),
                           netdef->name);
            goto error;
        }
        iface->data.network.actual->data.hostdev.def.parent.type = VIR_DOMAIN_DEVICE_NET;
        iface->data.network.actual->data.hostdev.def.parent.data.net = iface;
        iface->data.network.actual->data.hostdev.def.info = &iface->info;
        iface->data.network.actual->data.hostdev.def.mode = VIR_DOMAIN_HOSTDEV_MODE_SUBSYS;
3953
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
3954
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
3955
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
3956

3957 3958 3959
        switch (netdef->forward.driverName)
        {
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
3960
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
3961 3962
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
3963
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
3964 3965
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
3966
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977
            break;
        default:
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unrecognized driver name value %d "
                             " in network '%s'"),
                           netdef->forward.driverName, netdef->name);
            goto error;
        }
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.backend
            = backend;

3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002
        /* merge virtualports from interface, network, and portgroup to
         * arrive at actual virtualport to use
         */
        if (virNetDevVPortProfileMerge3(&iface->data.network.actual->virtPortProfile,
                                        iface->virtPortProfile,
                                        netdef->virtPortProfile,
                                        portgroup
                                        ? portgroup->virtPortProfile : NULL) < 0) {
            goto error;
        }
        virtport = iface->data.network.actual->virtPortProfile;
        if (virtport) {
            /* make sure type is supported for hostdev connections */
            if (virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBG &&
                virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBH) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("<virtualport type='%s'> not supported for network "
                                 "'%s' which uses an SR-IOV Virtual Function "
                                 "via PCI passthrough"),
                               virNetDevVPortTypeToString(virtport->virtPortType),
                               netdef->name);
                goto error;
            }
        }

4003 4004 4005 4006
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) ||
               (netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) ||
               (netdef->forward.type == VIR_NETWORK_FORWARD_VEPA) ||
               (netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH)) {
4007 4008 4009 4010 4011

        /* <forward type='bridge|private|vepa|passthrough'> are all
         * VIR_DOMAIN_NET_TYPE_DIRECT.
         */

4012 4013
        if (!iface->data.network.actual &&
            VIR_ALLOC(iface->data.network.actual) < 0)
4014
            goto error;
4015 4016

        /* Set type=direct and appropriate <source mode='xxx'/> */
4017
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4018
        switch (netdef->forward.type) {
4019
        case VIR_NETWORK_FORWARD_BRIDGE:
4020
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
4021 4022
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
4023
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
4024 4025
            break;
        case VIR_NETWORK_FORWARD_VEPA:
4026
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
4027 4028
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
4029
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
4030 4031 4032
            break;
        }

4033 4034 4035 4036 4037 4038 4039 4040
        /* merge virtualports from interface, network, and portgroup to
         * arrive at actual virtualport to use
         */
        if (virNetDevVPortProfileMerge3(&iface->data.network.actual->virtPortProfile,
                                        iface->virtPortProfile,
                                        netdef->virtPortProfile,
                                        portgroup
                                        ? portgroup->virtPortProfile : NULL) < 0) {
4041
            goto error;
4042
        }
4043
        virtport = iface->data.network.actual->virtPortProfile;
4044
        if (virtport) {
4045 4046 4047 4048 4049 4050 4051 4052
            /* make sure type is supported for macvtap connections */
            if (virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBG &&
                virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBH) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("<virtualport type='%s'> not supported for network "
                                 "'%s' which uses a macvtap device"),
                               virNetDevVPortTypeToString(virtport->virtPortType),
                               netdef->name);
4053
                goto error;
4054 4055
            }
        }
4056

4057 4058 4059
        /* If there is only a single device, just return it (caller will detect
         * any error if exclusive use is required but could not be acquired).
         */
4060
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4061 4062 4063 4064
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4065
            goto error;
4066 4067 4068
        } else {
            /* pick an interface from the pool */

4069
            if (netdef->forward.npfs > 0 && netdef->forward.nifs == 0 &&
4070 4071 4072 4073
                networkCreateInterfacePool(netdef) < 0) {
                goto error;
            }

4074 4075 4076 4077 4078
            /* PASSTHROUGH mode, and PRIVATE Mode + 802.1Qbh both
             * require exclusive access to a device, so current
             * connections count must be 0.  Other modes can share, so
             * just search for the one with the lowest number of
             * connections.
4079
             */
4080 4081
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4082 4083 4084
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4085

4086
                /* pick first dev with 0 connections */
4087 4088 4089
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4090 4091 4092 4093 4094
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4095
                dev = &netdef->forward.ifs[0];
4096 4097 4098
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4099 4100 4101 4102
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4103 4104 4105 4106
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4107
                goto error;
4108
            }
4109 4110
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4111
                goto error;
4112 4113 4114
        }
    }

4115
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4116
        goto error;
4117

4118 4119 4120 4121 4122 4123
validate:
    /* make sure that everything now specified for the device is
     * actually supported on this type of network. NB: network,
     * netdev, and iface->data.network.actual may all be NULL.
     */

4124
    if (virDomainNetGetActualVlan(iface)) {
4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149
        /* vlan configuration via libvirt is only supported for
         * PCI Passthrough SR-IOV devices and openvswitch bridges.
         * otherwise log an error and fail
         */
        if (!(actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV ||
              (actualType == VIR_DOMAIN_NET_TYPE_BRIDGE &&
               virtport && virtport->virtPortType
               == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH))) {
            if (netdef) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("an interface connecting to network '%s' "
                                 "is requesting a vlan tag, but that is not "
                                 "supported for this type of network"),
                               netdef->name);
            } else {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("an interface of type '%s' "
                                 "is requesting a vlan tag, but that is not "
                                 "supported for this type of connection"),
                               virDomainNetTypeToString(iface->type));
            }
            goto error;
        }
    }

4150 4151
    if (dev) {
        /* we are now assured of success, so mark the allocation */
4152
        dev->connections++;
4153 4154 4155 4156 4157 4158 4159 4160 4161
        if (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV) {
            VIR_DEBUG("Using physical device %s, %d connections",
                      dev->device.dev, dev->connections);
        } else {
            VIR_DEBUG("Using physical device %04x:%02x:%02x.%x, connections %d",
                      dev->device.pci.domain, dev->device.pci.bus,
                      dev->device.pci.slot, dev->device.pci.function,
                      dev->connections);
        }
4162
    }
4163

4164 4165 4166 4167 4168
    if (netdef) {
        netdef->connections++;
        VIR_DEBUG("Using network %s, %d connections",
                  netdef->name, netdef->connections);
    }
4169
    ret = 0;
4170

4171 4172 4173
cleanup:
    if (network)
        virNetworkObjUnlock(network);
4174 4175 4176 4177
    return ret;

error:
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4178 4179 4180
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4181
    goto cleanup;
4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196
}

/* networkNotifyActualDevice:
 * @iface:  the domain's NetDef with an "actual" device already filled in.
 *
 * Called to notify the network driver when libvirtd is restarted and
 * finds an already running domain. If appropriate it will force an
 * allocation of the actual->direct.linkdev to get everything back in
 * order.
 *
 * Returns 0 on success, -1 on failure.
 */
int
networkNotifyActualDevice(virDomainNetDefPtr iface)
{
4197
    virNetworkDriverStatePtr driver = driverState;
4198
    enum virDomainNetType actualType = virDomainNetGetActualType(iface);
4199 4200
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4201
    virNetworkForwardIfDefPtr dev = NULL;
4202 4203
    size_t i;
    int ret = -1;
4204 4205 4206 4207 4208 4209 4210 4211

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
       return 0;

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
4212 4213 4214
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4215 4216 4217 4218 4219
        goto error;
    }
    netdef = network->def;

    if (!iface->data.network.actual ||
4220 4221
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4222 4223
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4224 4225
    }

4226
    if (netdef->forward.npfs > 0 && netdef->forward.nifs == 0 &&
4227
        networkCreateInterfacePool(netdef) < 0) {
4228
        goto error;
4229
    }
4230
    if (netdef->forward.nifs == 0) {
4231
        virReportError(VIR_ERR_INTERNAL_ERROR,
4232 4233
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4234
                       netdef->name);
4235
        goto error;
4236
    }
4237

4238 4239
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4240

4241 4242 4243 4244 4245 4246 4247 4248 4249
        actualDev = virDomainNetGetActualDirectDev(iface);
        if (!actualDev) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("the interface uses a direct mode, "
                             "but has no source dev"));
            goto error;
        }

        /* find the matching interface and increment its connections */
4250 4251
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4252
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4253 4254
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4255 4256 4257 4258 4259
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4260
            virReportError(VIR_ERR_INTERNAL_ERROR,
4261 4262
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4263
                           netdef->name, actualDev);
4264
            goto error;
4265 4266
        }

4267
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4268 4269
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4270
         */
4271
        if ((dev->connections > 0) &&
4272 4273
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4274 4275
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4276
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4277
            virReportError(VIR_ERR_INTERNAL_ERROR,
4278 4279
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4280
                           netdef->name, actualDev);
4281
            goto error;
4282
        }
4283

4284
        /* we are now assured of success, so mark the allocation */
4285
        dev->connections++;
4286
        VIR_DEBUG("Using physical device %s, connections %d",
4287
                  dev->device.dev, dev->connections);
4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300

    }  else /* if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) */ {
        virDomainHostdevDefPtr hostdev;

        hostdev = virDomainNetGetActualHostdev(iface);
        if (!hostdev) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("the interface uses a hostdev mode, "
                             "but has no hostdev"));
            goto error;
        }

        /* find the matching interface and increment its connections */
4301 4302
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4303
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4304
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4305 4306
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4307 4308 4309 4310 4311 4312 4313 4314 4315
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have "
                             "PCI device %04x:%02x:%02x.%x in use by domain"),
                           netdef->name,
4316 4317 4318 4319
                           hostdev->source.subsys.u.pci.addr.domain,
                           hostdev->source.subsys.u.pci.addr.bus,
                           hostdev->source.subsys.u.pci.addr.slot,
                           hostdev->source.subsys.u.pci.addr.function);
4320 4321 4322 4323 4324 4325 4326 4327
                goto error;
        }

        /* PASSTHROUGH mode, PRIVATE Mode + 802.1Qbh, and hostdev (PCI
         * passthrough) all require exclusive access to a device, so
         * current connections count must be 0 in those cases.
         */
        if ((dev->connections > 0) &&
4328
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' claims the PCI device at "
                             "domain=%d bus=%d slot=%d function=%d "
                             "is already in use by a different domain"),
                           netdef->name,
                           dev->device.pci.domain, dev->device.pci.bus,
                           dev->device.pci.slot, dev->device.pci.function);
            goto error;
        }

        /* we are now assured of success, so mark the allocation */
        dev->connections++;
        VIR_DEBUG("Using physical device %04x:%02x:%02x.%x, connections %d",
                  dev->device.pci.domain, dev->device.pci.bus,
                  dev->device.pci.slot, dev->device.pci.function,
                  dev->connections);
4345 4346
    }

4347
success:
4348 4349 4350
    netdef->connections++;
    VIR_DEBUG("Using network %s, %d connections",
              netdef->name, netdef->connections);
4351 4352 4353 4354 4355
    ret = 0;
cleanup:
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4356 4357 4358

error:
    goto cleanup;
4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374
}


/* networkReleaseActualDevice:
 * @iface:  a domain's NetDef (interface definition)
 *
 * Given a domain <interface> element that previously had its <actual>
 * element filled in (and possibly a physical device allocated to it),
 * free up the physical device for use by someone else, and free the
 * virDomainActualNetDef.
 *
 * Returns 0 on success, -1 on failure.
 */
int
networkReleaseActualDevice(virDomainNetDefPtr iface)
{
4375
    virNetworkDriverStatePtr driver = driverState;
4376
    enum virDomainNetType actualType = virDomainNetGetActualType(iface);
4377
    virNetworkObjPtr network;
4378
    virNetworkDefPtr netdef;
4379
    virNetworkForwardIfDefPtr dev = NULL;
4380 4381
    size_t i;
    int ret = -1;
4382 4383 4384 4385 4386 4387 4388 4389

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
       return 0;

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
4390 4391 4392
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4393 4394 4395 4396
        goto error;
    }
    netdef = network->def;

4397 4398 4399 4400 4401 4402
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) &&
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4403 4404 4405
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4406 4407
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4408 4409
    }

4410
    if (netdef->forward.nifs == 0) {
4411
        virReportError(VIR_ERR_INTERNAL_ERROR,
4412
                       _("network '%s' uses a direct/hostdev mode, but "
4413 4414
                         "has no forward dev and no interface pool"),
                       netdef->name);
4415
        goto error;
4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427
    }

    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;

        actualDev = virDomainNetGetActualDirectDev(iface);
        if (!actualDev) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("the interface uses a direct mode, "
                             "but has no source dev"));
            goto error;
        }
4428

4429 4430
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4431
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4432 4433
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4434 4435 4436
                break;
            }
        }
4437

4438
        if (!dev) {
4439
            virReportError(VIR_ERR_INTERNAL_ERROR,
4440 4441
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4442
                           netdef->name, actualDev);
4443
            goto error;
4444 4445
        }

4446
        dev->connections--;
4447
        VIR_DEBUG("Releasing physical device %s, connections %d",
4448
                  dev->device.dev, dev->connections);
4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459

    } else /* if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) */ {
        virDomainHostdevDefPtr hostdev;

        hostdev = virDomainNetGetActualHostdev(iface);
        if (!hostdev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("the interface uses a hostdev mode, but has no hostdev"));
            goto error;
        }

4460 4461
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4462
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4463
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4464 4465
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4466 4467 4468 4469 4470 4471 4472 4473 4474
                break;
            }
        }

        if (!dev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have "
                             "PCI device %04x:%02x:%02x.%x in use by domain"),
                           netdef->name,
4475 4476 4477 4478
                           hostdev->source.subsys.u.pci.addr.domain,
                           hostdev->source.subsys.u.pci.addr.bus,
                           hostdev->source.subsys.u.pci.addr.slot,
                           hostdev->source.subsys.u.pci.addr.function);
4479 4480 4481 4482 4483 4484 4485 4486 4487
                goto error;
        }

        dev->connections--;
        VIR_DEBUG("Releasing physical device %04x:%02x:%02x.%x, connections %d",
                  dev->device.pci.domain, dev->device.pci.bus,
                  dev->device.pci.slot, dev->device.pci.function,
                  dev->connections);
   }
4488

4489
success:
4490 4491 4492
    netdef->connections--;
    VIR_DEBUG("Releasing network %s, %d connections",
              netdef->name, netdef->connections);
4493 4494 4495 4496
    ret = 0;
cleanup:
    if (network)
        virNetworkObjUnlock(network);
4497 4498 4499 4500
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4501
    return ret;
4502 4503 4504

error:
    goto cleanup;
4505
}
4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529

/*
 * networkGetNetworkAddress:
 * @netname: the name of a network
 * @netaddr: string representation of IP address for that network.
 *
 * Attempt to return an IP (v4) address associated with the named
 * network. If a libvirt virtual network, that will be provided in the
 * configuration. For host bridge and direct (macvtap) networks, we
 * must do an ioctl to learn the address.
 *
 * Note: This function returns the 1st IPv4 address it finds. It might
 * be useful if it was more flexible, but the current use (getting a
 * listen address for qemu's vnc/spice graphics server) can only use a
 * single address anyway.
 *
 * Returns 0 on success, and puts a string (which must be free'd by
 * the caller) into *netaddr. Returns -1 on failure or -2 if
 * completely unsupported.
 */
int
networkGetNetworkAddress(const char *netname, char **netaddr)
{
    int ret = -1;
4530
    virNetworkDriverStatePtr driver = driverState;
4531
    virNetworkObjPtr network;
4532 4533 4534 4535
    virNetworkDefPtr netdef;
    virNetworkIpDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4536
    char *dev_name = NULL;
4537 4538 4539 4540 4541 4542

    *netaddr = NULL;
    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, netname);
    networkDriverUnlock(driver);
    if (!network) {
4543 4544 4545
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4546
        goto error;
4547 4548 4549
    }
    netdef = network->def;

4550
    switch (netdef->forward.type) {
4551 4552 4553 4554 4555 4556
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
        /* if there's an ipv4def, get it's address */
        ipdef = virNetworkDefGetIpByIndex(netdef, AF_INET, 0);
        if (!ipdef) {
4557 4558 4559
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have an IPv4 address"),
                           netdef->name);
4560 4561 4562 4563 4564 4565
            break;
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4566
        if ((dev_name = netdef->bridge))
4567 4568 4569 4570 4571 4572 4573 4574
            break;
        /*
         * fall through if netdef->bridge wasn't set, since this is
         * also a direct-mode interface.
         */
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
4575 4576
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4577

4578
        if (!dev_name) {
4579 4580 4581
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4582 4583 4584 4585
        }
        break;
    }

4586
    if (dev_name) {
4587
        if (virNetDevGetIPv4Address(dev_name, &addr) < 0)
4588
            goto error;
4589
        addrptr = &addr;
4590 4591
    }

4592 4593 4594
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
        goto error;
4595 4596
    }

4597
    ret = 0;
4598 4599 4600 4601
cleanup:
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4602 4603 4604

error:
    goto cleanup;
4605
}
4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623

/**
 * networkCheckBandwidth:
 * @net: network QoS
 * @iface: interface QoS
 * @new_rate: new rate for non guaranteed class
 *
 * Returns: -1 if plugging would overcommit network QoS
 *           0 if plugging is safe (@new_rate updated)
 *           1 if no QoS is set (@new_rate untouched)
 */
static int
networkCheckBandwidth(virNetworkObjPtr net,
                      virDomainNetDefPtr iface,
                      unsigned long long *new_rate)
{
    int ret = -1;
    virNetDevBandwidthPtr netBand = net->def->bandwidth;
4624
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4625 4626 4627 4628
    unsigned long long tmp_floor_sum = net->floor_sum;
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639
    virMacAddrFormat(&iface->mac, ifmac);

    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
        !(netBand && netBand->in)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("Invalid use of 'floor' on interface with MAC "
                         "address %s - network '%s' has no inbound QoS set"),
                       ifmac, net->def->name);
        return -1;
    }

4640
    if (!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor ||
4641 4642
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4643
        return 1;
4644
    }
4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713

    tmp_new_rate = netBand->in->average;
    tmp_floor_sum += ifaceBand->in->floor;

    /* check against peak */
    if (netBand->in->peak) {
        tmp_new_rate = netBand->in->peak;
        if (tmp_floor_sum > netBand->in->peak) {
            virReportError(VIR_ERR_OPERATION_INVALID,
                           _("Cannot plug '%s' interface into '%s' because it "
                             "would overcommit 'peak' on network '%s'"),
                           ifmac,
                           net->def->bridge,
                           net->def->name);
            goto cleanup;
        }
    } else if (tmp_floor_sum > netBand->in->average) {
        /* tmp_floor_sum can be between 'average' and 'peak' iff 'peak' is set.
         * Otherwise, tmp_floor_sum must be below 'average'. */
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("Cannot plug '%s' interface into '%s' because it "
                         "would overcommit 'average' on network '%s'"),
                       ifmac,
                       net->def->bridge,
                       net->def->name);
        goto cleanup;
    }

    *new_rate = tmp_new_rate;
    ret = 0;

cleanup:
    return ret;
}

/**
 * networkNextClassID:
 * @net: network object
 *
 * Find next free class ID. @net is supposed
 * to be locked already. If there is a free ID,
 * it is marked as used and returned.
 *
 * Returns next free class ID or -1 if none is available.
 */
static ssize_t
networkNextClassID(virNetworkObjPtr net)
{
    size_t ret = 0;
    bool is_set = false;

    while (virBitmapGetBit(net->class_id, ret, &is_set) == 0 && is_set)
        ret++;

    if (is_set || virBitmapSetBit(net->class_id, ret) < 0)
        return -1;

    return ret;
}

static int
networkPlugBandwidth(virNetworkObjPtr net,
                     virDomainNetDefPtr iface)
{
    int ret = -1;
    int plug_ret;
    unsigned long long new_rate = 0;
    ssize_t class_id = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];
4714
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742

    if ((plug_ret = networkCheckBandwidth(net, iface, &new_rate)) < 0) {
        /* helper reported error */
        goto cleanup;
    }

    if (plug_ret > 0) {
        /* no QoS needs to be set; claim success */
        ret = 0;
        goto cleanup;
    }

    virMacAddrFormat(&iface->mac, ifmac);
    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK ||
        !iface->data.network.actual) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Cannot set bandwidth on interface '%s' of type %d"),
                       ifmac, iface->type);
        goto cleanup;
    }

    /* generate new class_id */
    if ((class_id = networkNextClassID(net)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Could not generate next class ID"));
        goto cleanup;
    }

4743 4744
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
4745 4746 4747 4748 4749 4750 4751 4752
    if (plug_ret < 0) {
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }

    /* QoS was set, generate new class ID */
    iface->data.network.actual->class_id = class_id;
    /* update sum of 'floor'-s of attached NICs */
4753
    net->floor_sum += ifaceBand->in->floor;
4754
    /* update status file */
4755
    if (virNetworkSaveStatus(driverState->stateDir, net) < 0) {
4756
        ignore_value(virBitmapClearBit(net->class_id, class_id));
4757
        net->floor_sum -= ifaceBand->in->floor;
4758 4759 4760 4761
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780
    /* update rate for non guaranteed NICs */
    new_rate -= net->floor_sum;
    if (virNetDevBandwidthUpdateRate(net->def->bridge, "1:2",
                                     net->def->bandwidth, new_rate) < 0)
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                 net->def->bridge);

    ret = 0;

cleanup:
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
    int ret = 0;
    unsigned long long new_rate;
4781
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4782 4783 4784

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
4785 4786 4787 4788 4789
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800
        /* we must remove class from bridge */
        new_rate = net->def->bandwidth->in->average;

        if (net->def->bandwidth->in->peak > 0)
            new_rate = net->def->bandwidth->in->peak;

        ret = virNetDevBandwidthUnplug(net->def->bridge,
                                       iface->data.network.actual->class_id);
        if (ret < 0)
            goto cleanup;
        /* update sum of 'floor'-s of attached NICs */
4801
        net->floor_sum -= ifaceBand->in->floor;
4802 4803 4804 4805
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
4806
        if (virNetworkSaveStatus(driverState->stateDir, net) < 0) {
4807
            net->floor_sum += ifaceBand->in->floor;
4808 4809 4810 4811
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824
        /* update rate for non guaranteed NICs */
        new_rate -= net->floor_sum;
        if (virNetDevBandwidthUpdateRate(net->def->bridge, "1:2",
                                         net->def->bandwidth, new_rate) < 0)
            VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                     net->def->bridge);
        /* no class is associated any longer */
        iface->data.network.actual->class_id = 0;
    }

cleanup:
    return ret;
}