bridge_driver.c 150.2 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2014 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
#include <dirent.h>
45 46 47
#if HAVE_SYS_SYSCTL_H
# include <sys/sysctl.h>
#endif
48

49
#include "virerror.h"
50
#include "datatypes.h"
51
#include "bridge_driver.h"
52
#include "bridge_driver_platform.h"
53
#include "network_conf.h"
54
#include "device_conf.h"
55
#include "driver.h"
56
#include "virbuffer.h"
57
#include "virpidfile.h"
58
#include "vircommand.h"
59
#include "viralloc.h"
60
#include "viruuid.h"
61
#include "viriptables.h"
62
#include "virlog.h"
63
#include "virdnsmasq.h"
64
#include "configmake.h"
65
#include "virnetdev.h"
66
#include "virpci.h"
67 68
#include "virnetdevbridge.h"
#include "virnetdevtap.h"
69
#include "virnetdevvportprofile.h"
70
#include "virdbus.h"
71
#include "virfile.h"
72
#include "virstring.h"
73
#include "viraccessapicheck.h"
74
#include "network_event.h"
75
#include "virhook.h"
76

77 78
#define VIR_FROM_THIS VIR_FROM_NETWORK

79 80
VIR_LOG_INIT("network.bridge_driver");

81
static void networkDriverLock(virNetworkDriverStatePtr driver)
82
{
83
    virMutexLock(&driver->lock);
84
}
85
static void networkDriverUnlock(virNetworkDriverStatePtr driver)
86
{
87
    virMutexUnlock(&driver->lock);
88 89
}

90
static int networkStateCleanup(void);
91

92
static int networkStartNetwork(virNetworkDriverStatePtr driver,
93 94
                               virNetworkObjPtr network);

95
static int networkShutdownNetwork(virNetworkDriverStatePtr driver,
96 97
                                  virNetworkObjPtr network);

98
static int networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
99
                                     virNetworkObjPtr network);
100

101
static int networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
R
Roman Bogorodskiy 已提交
102
                                         virNetworkObjPtr network);
103

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

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

110
static void networkReloadFirewallRules(virNetworkDriverStatePtr driver);
111
static void networkRefreshDaemons(virNetworkDriverStatePtr driver);
112

113 114 115 116 117
static int networkPlugBandwidth(virNetworkObjPtr net,
                                virDomainNetDefPtr iface);
static int networkUnplugBandwidth(virNetworkObjPtr net,
                                  virDomainNetDefPtr iface);

118 119 120
static void networkNetworkObjTaint(virNetworkObjPtr net,
                                   enum virNetworkTaintFlags taint);

121
static virNetworkDriverStatePtr driverState = NULL;
122

123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143
static virNetworkObjPtr
networkObjFromNetwork(virNetworkPtr net)
{
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

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

    if (!network) {
        virUUIDFormat(net->uuid, uuidstr);
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching uuid '%s' (%s)"),
                       uuidstr, net->name);
    }

    return network;
}

144 145 146
static int
networkRunHook(virNetworkObjPtr network,
               virDomainDefPtr dom,
147
               virDomainNetDefPtr iface,
148 149 150 151 152 153 154 155 156
               int op,
               int sub_op)
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *xml = NULL, *net_xml = NULL, *dom_xml = NULL;
    int hookret;
    int ret = -1;

    if (virHookPresent(VIR_HOOK_DRIVER_NETWORK)) {
157 158 159 160 161 162
        if (!network) {
            VIR_DEBUG("Not running hook as @network is NULL");
            ret = 0;
            goto cleanup;
        }

163 164
        virBufferAddLit(&buf, "<hookData>\n");
        virBufferAdjustIndent(&buf, 2);
165 166
        if (iface && virDomainNetDefFormat(&buf, iface, 0) < 0)
            goto cleanup;
167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186
        if (virNetworkDefFormatBuf(&buf, network->def, 0) < 0)
            goto cleanup;
        if (dom && virDomainDefFormatInternal(dom, 0, &buf) < 0)
            goto cleanup;

        virBufferAdjustIndent(&buf, -2);
        virBufferAddLit(&buf, "</hookData>");

        if (virBufferError(&buf) ||
            !(xml = virBufferContentAndReset(&buf)))
            goto cleanup;

        hookret = virHookCall(VIR_HOOK_DRIVER_NETWORK, network->def->name,
                              op, sub_op, NULL, xml, NULL);

        /*
         * If the script raised an error, pass it to the callee.
         */
        if (hookret < 0)
            goto cleanup;
187 188

        networkNetworkObjTaint(network, VIR_NETWORK_TAINT_HOOK);
189 190 191
    }

    ret = 0;
192
 cleanup:
193 194 195 196 197 198 199
    virBufferFreeAndReset(&buf);
    VIR_FREE(xml);
    VIR_FREE(net_xml);
    VIR_FREE(dom_xml);
    return ret;
}

200
static char *
201
networkDnsmasqLeaseFileNameDefault(const char *netname)
202 203 204
{
    char *leasefile;

205 206
    ignore_value(virAsprintf(&leasefile, "%s/%s.leases",
                             driverState->dnsmasqStateDir, netname));
207 208 209
    return leasefile;
}

210 211 212
networkDnsmasqLeaseFileNameFunc networkDnsmasqLeaseFileName =
    networkDnsmasqLeaseFileNameDefault;

213 214 215 216 217
static char *
networkDnsmasqConfigFileName(const char *netname)
{
    char *conffile;

218 219
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
                             driverState->dnsmasqStateDir, netname));
220 221 222
    return conffile;
}

223 224 225 226 227 228
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

229
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
230 231 232 233 234 235 236 237
    return pidfilebase;
}

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

238 239
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
                             driverState->radvdStateDir, netname));
240 241
    return configfile;
}
242

243 244
/* do needed cleanup steps and remove the network from the list */
static int
245
networkRemoveInactive(virNetworkDriverStatePtr driver,
246 247 248 249
                      virNetworkObjPtr net)
{
    char *leasefile = NULL;
    char *radvdconfigfile = NULL;
250
    char *configfile = NULL;
251
    char *radvdpidbase = NULL;
252
    char *statusfile = NULL;
253 254 255 256 257 258
    dnsmasqContext *dctx = NULL;
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(net);

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
259 260
    if (!(dctx = dnsmasqContextNew(def->name,
                                   driverState->dnsmasqStateDir))) {
261
        goto cleanup;
262
    }
263 264 265 266 267

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

    if (!(radvdconfigfile = networkRadvdConfigFileName(def->name)))
268
        goto cleanup;
269 270

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
271
        goto cleanup;
272

273
    if (!(configfile = networkDnsmasqConfigFileName(def->name)))
274
        goto cleanup;
275

276 277
    if (!(statusfile
          = virNetworkConfigFile(driverState->stateDir, def->name)))
278
        goto cleanup;
279

280 281 282
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
283
    unlink(configfile);
284 285 286

    /* radvd */
    unlink(radvdconfigfile);
287
    virPidFileDelete(driverState->pidDir, radvdpidbase);
288

289 290 291
    /* remove status file */
    unlink(statusfile);

292 293 294 295 296
    /* remove the network definition */
    virNetworkRemoveInactive(&driver->networks, net);

    ret = 0;

297
 cleanup:
298
    VIR_FREE(leasefile);
299
    VIR_FREE(configfile);
300 301
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
302
    VIR_FREE(statusfile);
303 304 305 306
    dnsmasqContextFree(dctx);
    return ret;
}

307 308 309
static char *
networkBridgeDummyNicName(const char *brname)
{
310
    static const char dummyNicSuffix[] = "-nic";
311 312
    char *nicname;

313 314 315 316 317 318 319
    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.
         */
320 321 322 323 324
        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));
325
    } else {
326
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
327
    }
328 329 330
    return nicname;
}

331 332 333
/* Update the internal status of all allegedly active networks
 * according to external conditions on the host (i.e. anything that
 * isn't stored directly in each network's state file). */
334
static void
335
networkUpdateAllState(virNetworkDriverStatePtr driver)
336
{
337
    size_t i;
338

339
    for (i = 0; i < driver->networks.count; i++) {
340 341
        virNetworkObjPtr obj = driver->networks.objs[i];

342 343 344
        if (!obj->active)
           continue;

345 346
        virNetworkObjLock(obj);

347 348 349 350 351 352 353 354
        switch (obj->def->forward.type) {
        case VIR_NETWORK_FORWARD_NONE:
        case VIR_NETWORK_FORWARD_NAT:
        case VIR_NETWORK_FORWARD_ROUTE:
            /* If bridge doesn't exist, then mark it inactive */
            if (!(obj->def->bridge && virNetDevExists(obj->def->bridge) == 1))
                obj->active = 0;
            break;
355

356 357 358 359 360 361 362 363 364 365 366 367 368 369 370
        case VIR_NETWORK_FORWARD_BRIDGE:
            if (obj->def->bridge) {
                if (virNetDevExists(obj->def->bridge) != 1)
                    obj->active = 0;
                break;
            }
            /* intentionally drop through to common case for all
             * macvtap networks (forward='bridge' with no bridge
             * device defined is macvtap using its 'bridge' mode)
             */
        case VIR_NETWORK_FORWARD_PRIVATE:
        case VIR_NETWORK_FORWARD_VEPA:
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
            /* so far no extra checks */
            break;
371

372 373 374 375
        case VIR_NETWORK_FORWARD_HOSTDEV:
            /* so far no extra checks */
            break;
        }
376

377 378 379 380 381 382 383 384 385 386 387 388 389 390 391
        /* Try and read dnsmasq/radvd pids of active networks */
        if (obj->active && obj->def->ips && (obj->def->nips > 0)) {
            char *radvdpidbase;

            ignore_value(virPidFileReadIfAlive(driverState->pidDir,
                                               obj->def->name,
                                               &obj->dnsmasqPid,
                                               dnsmasqCapsGetBinaryPath(driver->dnsmasqCaps)));
            radvdpidbase = networkRadvdPidfileBasename(obj->def->name);
            if (!radvdpidbase)
                break;
            ignore_value(virPidFileReadIfAlive(driverState->pidDir,
                                               radvdpidbase,
                                               &obj->radvdPid, RADVD));
            VIR_FREE(radvdpidbase);
392 393 394 395
        }

        virNetworkObjUnlock(obj);
    }
396 397 398 399 400 401 402 403 404 405 406 407 408 409 410

    /* 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++;
    }
411 412 413
}


414
static void
415 416
networkAutostartConfigs(virNetworkDriverStatePtr driver)
{
417
    size_t i;
418

419
    for (i = 0; i < driver->networks.count; i++) {
420
        virNetworkObjLock(driver->networks.objs[i]);
421
        if (driver->networks.objs[i]->autostart &&
422 423
            !virNetworkObjIsActive(driver->networks.objs[i])) {
            if (networkStartNetwork(driver, driver->networks.objs[i]) < 0) {
424
            /* failed to start but already logged */
425
            }
426
        }
427
        virNetworkObjUnlock(driver->networks.objs[i]);
428 429 430
    }
}

431 432 433
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
434 435
                             DBusMessage *message, void *user_data)
{
436
    virNetworkDriverStatePtr _driverState = user_data;
437 438 439 440 441 442 443

    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.");
444
        networkReloadFirewallRules(_driverState);
445 446 447 448 449 450
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532
static int
networkMigrateStateFiles(virNetworkDriverStatePtr driver)
{
    /* Due to a change in location of network state xml beginning in
     * libvirt 1.2.4 (from /var/lib/libvirt/network to
     * /var/run/libvirt/network), we must check for state files in two
     * locations. Anything found in the old location must be written
     * to the new location, then erased from the old location. (Note
     * that we read/write the file rather than calling rename()
     * because the old and new state directories are likely in
     * different filesystems).
     */
    int ret = -1;
    const char *oldStateDir = LOCALSTATEDIR "/lib/libvirt/network";
    DIR *dir;
    struct dirent *entry;
    char *oldPath = NULL, *newPath = NULL;
    char *contents = NULL;

    if (!(dir = opendir(oldStateDir))) {
        if (errno == ENOENT)
            return 0;

        virReportSystemError(errno, _("failed to open directory '%s'"),
                             oldStateDir);
        return -1;
    }

    if (virFileMakePath(driver->stateDir) < 0) {
        virReportSystemError(errno, _("cannot create directory %s"),
                             driver->stateDir);
        goto cleanup;
    }

    for (;;) {
        errno = 0;
        entry = readdir(dir);
        if (!entry) {
            if (errno) {
                virReportSystemError(errno, _("failed to read directory '%s'"),
                                     oldStateDir);
                goto cleanup;
            }
            break;
        }

        if (entry->d_type != DT_REG ||
            STREQ(entry->d_name, ".") ||
            STREQ(entry->d_name, ".."))
            continue;

        if (virAsprintf(&oldPath, "%s/%s",
                        oldStateDir, entry->d_name) < 0)
           goto cleanup;
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
           goto cleanup;

        if (virAsprintf(&newPath, "%s/%s",
                        driver->stateDir, entry->d_name) < 0)
           goto cleanup;
        if (virFileWriteStr(newPath, contents, S_IRUSR | S_IWUSR) < 0) {
            virReportSystemError(errno,
                                 _("failed to write network status file '%s'"),
                                 newPath);
            goto cleanup;
        }

        unlink(oldPath);
        VIR_FREE(oldPath);
        VIR_FREE(newPath);
        VIR_FREE(contents);
    }

    ret = 0;
 cleanup:
    closedir(dir);
    VIR_FREE(oldPath);
    VIR_FREE(newPath);
    VIR_FREE(contents);
    return ret;
}

533
/**
534
 * networkStateInitialize:
535 536 537 538
 *
 * Initialization function for the QEmu daemon
 */
static int
539 540 541
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
542
{
543 544 545
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
546 547 548
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
549 550

    if (VIR_ALLOC(driverState) < 0)
551
        goto error;
552

553 554 555 556
    if (virMutexInit(&driverState->lock) < 0) {
        VIR_FREE(driverState);
        goto error;
    }
557 558
    networkDriverLock(driverState);

559 560 561 562
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
563
    if (privileged) {
564 565 566 567 568
        if (VIR_STRDUP(driverState->networkConfigDir,
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
            VIR_STRDUP(driverState->networkAutostartDir,
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
            VIR_STRDUP(driverState->stateDir,
569
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
570 571 572 573 574 575 576
            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;
577 578 579 580 581 582 583

        /* migration from old to new location is only applicable for
         * privileged mode - unprivileged mode directories haven't
         * changed location.
         */
        if (networkMigrateStateFiles(driverState) < 0)
            goto error;
584
    } else {
585 586 587
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
588
            goto error;
589

590 591 592 593 594 595 596 597 598 599 600 601
        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)) {
602
            goto error;
603
        }
604 605
    }

606 607 608 609 610 611 612
    if (virFileMakePath(driverState->stateDir) < 0) {
        virReportSystemError(errno,
                             _("cannot create directory %s"),
                             driverState->stateDir);
        goto error;
    }

613 614
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
    driverState->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
615

616
    if (virNetworkLoadAllState(&driverState->networks,
617
                               driverState->stateDir) < 0)
618 619
        goto error;

620
    if (virNetworkLoadAllConfigs(&driverState->networks,
621
                                 driverState->networkConfigDir,
622 623 624
                                 driverState->networkAutostartDir) < 0)
        goto error;

625
    networkUpdateAllState(driverState);
626
    networkReloadFirewallRules(driverState);
627
    networkRefreshDaemons(driverState);
628

629 630
    driverState->networkEventState = virObjectEventStateNew();

631 632
    networkDriverUnlock(driverState);

633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658
#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

659
    ret = 0;
660
 cleanup:
661 662 663
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
664

665
 error:
666 667
    if (driverState)
        networkDriverUnlock(driverState);
668
    networkStateCleanup();
669
    goto cleanup;
670 671
}

672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
    if (!driverState)
        return;

    networkDriverLock(driverState);
    networkAutostartConfigs(driverState);
    networkDriverUnlock(driverState);
}

688
/**
689
 * networkStateReload:
690 691 692 693 694
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
695 696
networkStateReload(void)
{
697 698 699
    if (!driverState)
        return 0;

700
    networkDriverLock(driverState);
701
    virNetworkLoadAllState(&driverState->networks,
702
                           driverState->stateDir);
703
    virNetworkLoadAllConfigs(&driverState->networks,
704 705
                             driverState->networkConfigDir,
                             driverState->networkAutostartDir);
706
    networkReloadFirewallRules(driverState);
707
    networkRefreshDaemons(driverState);
708
    networkAutostartConfigs(driverState);
709
    networkDriverUnlock(driverState);
710 711 712 713 714
    return 0;
}


/**
715
 * networkStateCleanup:
716 717 718 719
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
720 721
networkStateCleanup(void)
{
722 723 724
    if (!driverState)
        return -1;

725 726
    networkDriverLock(driverState);

727 728
    virObjectEventStateFree(driverState->networkEventState);

729
    /* free inactive networks */
730
    virNetworkObjListFree(&driverState->networks);
731 732 733

    VIR_FREE(driverState->networkConfigDir);
    VIR_FREE(driverState->networkAutostartDir);
734 735 736 737
    VIR_FREE(driverState->stateDir);
    VIR_FREE(driverState->pidDir);
    VIR_FREE(driverState->dnsmasqStateDir);
    VIR_FREE(driverState->radvdStateDir);
738

739 740
    virObjectUnref(driverState->dnsmasqCaps);

741
    networkDriverUnlock(driverState);
742
    virMutexDestroy(&driverState->lock);
743

744 745 746 747 748 749
    VIR_FREE(driverState);

    return 0;
}


750 751 752 753 754 755 756
/* 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)
{
757 758
    size_t i;
    int ret = -1;
759 760 761 762 763 764 765
    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.
     */
766
    for (i = 0; i < 25; i++) {
767
        int signum = 0;
768
        if (i == 0)
769
            signum = SIGTERM;
770
        else if (i == 15) {
771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805
            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);
806
 cleanup:
807 808 809
    return ret;
}

G
Gene Czarcinski 已提交
810 811 812 813
    /* the following does not build a file, it builds a list
     * which is later saved into a file
     */

814
static int
G
Gene Czarcinski 已提交
815 816
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
                                 virNetworkIpDefPtr ipdef)
817
{
818
    size_t i;
G
Gene Czarcinski 已提交
819
    bool ipv6 = false;
820

G
Gene Czarcinski 已提交
821 822
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
823 824
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
825
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
826 827
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
828
                return -1;
829
    }
830

G
Gene Czarcinski 已提交
831 832 833 834 835 836 837
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
838
    size_t i, j;
G
Gene Czarcinski 已提交
839

840 841
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
842
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
843
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
844
                for (j = 0; j < host->nnames; j++)
845 846
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
847 848
            }
        }
849 850
    }

851
    return 0;
852 853 854
}


855 856
int
networkDnsmasqConfContents(virNetworkObjPtr network,
857 858 859 860
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
861
{
862
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
863
    int r, ret = -1;
864
    int nbleases = 0;
865
    size_t i;
866
    virNetworkDNSDefPtr dns = &network->def->dns;
G
Gene Czarcinski 已提交
867 868
    virNetworkIpDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
    bool ipv6SLAAC;
869

870 871
    *configstr = NULL;

872
    /*
873 874 875
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
876
     *
877 878
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
879
     */
880 881 882 883 884 885

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

886 887
    /* create dnsmasq config file appropriate for this network */
    virBufferAsprintf(&configbuf,
888 889 890 891 892 893 894
                      "##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"
895
                      "strict-order\n",
896 897
                      network->def->name);

898 899 900 901 902 903 904 905
    if (network->def->dns.forwarders) {
        virBufferAddLit(&configbuf, "no-resolv\n");
        for (i = 0; i < network->def->dns.nfwds; i++) {
            virBufferAsprintf(&configbuf, "server=%s\n",
                               network->def->dns.forwarders[i]);
        }
    }

906
    if (network->def->domain) {
907
        virBufferAsprintf(&configbuf,
908 909 910 911
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
912

913 914
    if (network->def->dns.forwardPlainNames
        == VIR_NETWORK_DNS_FORWARD_PLAIN_NAMES_NO) {
915 916 917 918
        virBufferAddLit(&configbuf, "domain-needed\n");
        /* need to specify local=// whether or not a domain is
         * specified, unless the config says we should forward "plain"
         * names (i.e. not fully qualified, no '.' characters)
919
         */
920
        virBufferAddLit(&configbuf, "local=//\n");
921
    }
922

923
    if (pidfile)
924
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
925

926 927 928
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

929 930 931 932 933 934 935 936
    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.
         */
937
        virBufferAsprintf(&configbuf,
938 939 940
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
941
    } else {
942
        virBufferAddLit(&configbuf, "bind-interfaces\n");
943 944 945 946 947 948 949 950
        /*
         * --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
         */
951 952 953
        for (i = 0;
             (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
             i++) {
954 955 956 957
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
958

959
            /* also part of CVE 2012-3411 - if the host's version of
960
             * dnsmasq doesn't have bind-dynamic, only allow listening on
961 962
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
963 964
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
965 966 967 968
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
969
                                 "The version of dnsmasq on this host (%d.%d) "
970 971 972 973
                                 "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 "
974 975 976 977 978 979
                                 "(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);
980 981
                goto cleanup;
            }
982
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
983 984 985
            VIR_FREE(ipaddr);
        }
    }
986

987 988
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
989
     * guaranteed to not work, and set no-resolv so that no dns
990 991 992
     * 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).
993
     */
994
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
995
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
996
                        "no-resolv\n");
997
    }
998

999
    for (i = 0; i < dns->ntxts; i++) {
1000
        virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
1001 1002
                          dns->txts[i].name,
                          dns->txts[i].value);
1003
    }
1004

1005
    for (i = 0; i < dns->nsrvs; i++) {
1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027
        /* service/protocol are required, and should have been validated
         * by the parser.
         */
        if (!dns->srvs[i].service) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Missing required 'service' "
                             "attribute in SRV record of network '%s'"),
                           network->def->name);
            goto cleanup;
        }
        if (!dns->srvs[i].protocol) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Missing required 'service' "
                             "attribute in SRV record of network '%s'"),
                           network->def->name);
            goto cleanup;
        }
        /* RFC2782 requires that service and protocol be preceded by
         * an underscore.
         */
        virBufferAsprintf(&configbuf, "srv-host=_%s._%s",
                          dns->srvs[i].service, dns->srvs[i].protocol);
1028

1029 1030 1031
        /* domain is optional - it defaults to the domain of this network */
        if (dns->srvs[i].domain)
            virBufferAsprintf(&configbuf, ".%s", dns->srvs[i].domain);
1032

1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
        /* If target is empty or ".", that means "the service is
         * decidedly not available at this domain" (RFC2782). In that
         * case, any port, priority, or weight is irrelevant.
         */
        if (dns->srvs[i].target && STRNEQ(dns->srvs[i].target, ".")) {

            virBufferAsprintf(&configbuf, ",%s", dns->srvs[i].target);
            /* port, priority, and weight are optional, but are
             * identified by their position in the line. If an item is
             * unspecified, but something later in the line *is*
             * specified, we need to give the default value for the
             * unspecified item. (According to the dnsmasq manpage,
             * the default for port is 1).
             */
            if (dns->srvs[i].port ||
                dns->srvs[i].priority || dns->srvs[i].weight)
                virBufferAsprintf(&configbuf, ",%d",
                                  dns->srvs[i].port ? dns->srvs[i].port : 1);
            if (dns->srvs[i].priority || dns->srvs[i].weight)
                virBufferAsprintf(&configbuf, ",%d", dns->srvs[i].priority);
            if (dns->srvs[i].weight)
                virBufferAsprintf(&configbuf, ",%d", dns->srvs[i].weight);
1055
        }
1056
        virBufferAddLit(&configbuf, "\n");
1057 1058
    }

G
Gene Czarcinski 已提交
1059
    /* Find the first dhcp for both IPv4 and IPv6 */
1060 1061 1062
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
1063 1064 1065 1066
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1067 1068
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079
                    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,
1080 1081 1082 1083 1084 1085 1086 1087 1088
                                   _("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
1089 1090 1091 1092
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1093 1094
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107
                    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 "
1108 1109 1110 1111
                 "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
1112 1113 1114 1115 1116
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1117
        for (r = 0; r < ipdef->nranges; r++) {
1118
            char *saddr = virSocketAddrFormat(&ipdef->ranges[r].start);
1119 1120
            if (!saddr)
                goto cleanup;
1121
            char *eaddr = virSocketAddrFormat(&ipdef->ranges[r].end);
1122 1123 1124 1125
            if (!eaddr) {
                VIR_FREE(saddr);
                goto cleanup;
            }
1126
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s\n",
1127
                              saddr, eaddr);
1128
            VIR_FREE(saddr);
1129
            VIR_FREE(eaddr);
1130 1131
            nbleases += virSocketAddrGetRange(&ipdef->ranges[r].start,
                                              &ipdef->ranges[r].end);
1132
        }
1133

1134
        /*
1135 1136 1137 1138
         * 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)
1139 1140
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1141
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1142 1143
            if (!bridgeaddr)
                goto cleanup;
1144
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static\n", bridgeaddr);
1145 1146
            VIR_FREE(bridgeaddr);
        }
1147

G
Gene Czarcinski 已提交
1148 1149
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1150

G
Gene Czarcinski 已提交
1151 1152 1153
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts)
1154
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1155

G
Gene Czarcinski 已提交
1156
            if (ipdef->tftproot) {
1157 1158
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1159
            }
1160

G
Gene Czarcinski 已提交
1161 1162 1163
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1164

1165
                    if (!bootserver)
G
Gene Czarcinski 已提交
1166
                        goto cleanup;
1167
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1168
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1169 1170
                    VIR_FREE(bootserver);
                } else {
1171
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1172 1173 1174 1175 1176
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1177

G
Gene Czarcinski 已提交
1178 1179
    if (nbleases > 0) {
        char *leasefile = networkDnsmasqLeaseFileName(network->def->name);
1180
        if (!leasefile)
G
Gene Czarcinski 已提交
1181
            goto cleanup;
1182
        virBufferAsprintf(&configbuf, "dhcp-leasefile=%s\n", leasefile);
G
Gene Czarcinski 已提交
1183
        VIR_FREE(leasefile);
1184
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
G
Gene Czarcinski 已提交
1185
    }
1186

G
Gene Czarcinski 已提交
1187 1188
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1189
        goto cleanup;
G
Gene Czarcinski 已提交
1190 1191 1192 1193 1194 1195

    /* 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)
1196 1197
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1198

1199 1200
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1201
     */
1202
    virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
1203
                      dctx->addnhostsfile->path);
G
Gene Czarcinski 已提交
1204 1205 1206 1207

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
        if (ipv6def)
1208
            virBufferAddLit(&configbuf, "enable-ra\n");
G
Gene Czarcinski 已提交
1209
        else {
1210 1211 1212
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
                 i++) {
G
Gene Czarcinski 已提交
1213 1214 1215 1216
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1217 1218
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1219 1220
                    VIR_FREE(bridgeaddr);
                }
1221
            }
1222
        }
1223 1224
    }

1225 1226 1227
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1228
    ret = 0;
G
Gene Czarcinski 已提交
1229

1230
 cleanup:
1231
    virBufferFreeAndReset(&configbuf);
1232
    return ret;
1233 1234
}

1235
/* build the dnsmasq command line */
1236 1237 1238
static int ATTRIBUTE_NONNULL(2)
networkBuildDhcpDaemonCommandLine(virNetworkObjPtr network,
                                  virCommandPtr *cmdout,
1239 1240
                                  char *pidfile, dnsmasqContext *dctx,
                                  dnsmasqCapsPtr caps)
1241
{
1242
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1243
    int ret = -1;
1244 1245
    char *configfile = NULL;
    char *configstr = NULL;
1246 1247

    network->dnsmasqPid = -1;
1248

1249 1250 1251 1252 1253 1254
    if (networkDnsmasqConfContents(network, pidfile, &configstr, dctx, caps) < 0)
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1255
    if (!(configfile = networkDnsmasqConfigFileName(network->def->name)))
1256 1257 1258 1259 1260 1261 1262
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
                         _("couldn't write dnsmasq config file '%s'"),
                         configfile);
1263 1264 1265
        goto cleanup;
    }

1266 1267
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(caps));
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1268
    *cmdout = cmd;
1269
    ret = 0;
1270
 cleanup:
1271 1272
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1273 1274 1275 1276
    return ret;
}

static int
1277
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
1278
                       virNetworkObjPtr network)
1279 1280 1281 1282
{
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1283
    dnsmasqContext *dctx = NULL;
1284

1285
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0)) {
G
Gene Czarcinski 已提交
1286
        /* no IP addresses, so we don't need to run */
1287 1288 1289 1290
        ret = 0;
        goto cleanup;
    }

1291
    if (virFileMakePath(driverState->pidDir) < 0) {
1292
        virReportSystemError(errno,
1293
                             _("cannot create directory %s"),
1294
                             driverState->pidDir);
1295
        goto cleanup;
1296 1297
    }

1298
    if (!(pidfile = virPidFileBuildPath(driverState->pidDir,
1299
                                        network->def->name)))
1300
        goto cleanup;
1301

1302
    if (virFileMakePath(driverState->dnsmasqStateDir) < 0) {
1303
        virReportSystemError(errno,
1304
                             _("cannot create directory %s"),
1305
                             driverState->dnsmasqStateDir);
1306 1307 1308
        goto cleanup;
    }

1309
    dctx = dnsmasqContextNew(network->def->name, driverState->dnsmasqStateDir);
1310 1311 1312
    if (dctx == NULL)
        goto cleanup;

1313 1314
    if (dnsmasqCapsRefresh(&driver->dnsmasqCaps, NULL) < 0)
        goto cleanup;
1315 1316 1317

    ret = networkBuildDhcpDaemonCommandLine(network, &cmd, pidfile,
                                            dctx, driver->dnsmasqCaps);
1318 1319 1320 1321 1322
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1323
        goto cleanup;
1324

G
Guido Günther 已提交
1325 1326
    ret = virCommandRun(cmd, NULL);
    if (ret < 0) {
1327
        goto cleanup;
G
Guido Günther 已提交
1328
    }
1329 1330

    /*
1331 1332 1333 1334 1335
     * 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
1336 1337
     */

1338
    ret = virPidFileRead(driverState->pidDir, network->def->name,
1339 1340
                         &network->dnsmasqPid);
    if (ret < 0)
1341
        goto cleanup;
1342

1343
    ret = 0;
1344
 cleanup:
1345
    VIR_FREE(pidfile);
1346
    virCommandFree(cmd);
1347
    dnsmasqContextFree(dctx);
1348 1349 1350
    return ret;
}

1351 1352
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1353 1354
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1355 1356 1357
 *
 *  Returns 0 on success, -1 on failure.
 */
1358
static int
1359
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
1360
                         virNetworkObjPtr network)
1361
{
1362 1363
    int ret = -1;
    size_t i;
G
Gene Czarcinski 已提交
1364
    virNetworkIpDefPtr ipdef, ipv4def, ipv6def;
1365
    dnsmasqContext *dctx = NULL;
1366

G
Gene Czarcinski 已提交
1367
    /* if no IP addresses specified, nothing to do */
1368
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1369 1370
        return 0;

1371 1372
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1373
        return networkStartDhcpDaemon(driver, network);
1374

G
Gene Czarcinski 已提交
1375
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1376 1377
    if (!(dctx = dnsmasqContextNew(network->def->name,
                                   driverState->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1378
        goto cleanup;
1379
    }
G
Gene Czarcinski 已提交
1380 1381 1382 1383 1384 1385

    /* 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;
1386 1387 1388
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
G
Gene Czarcinski 已提交
1389 1390
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1391 1392
    }

G
Gene Czarcinski 已提交
1393
    ipv6def = NULL;
1394 1395 1396
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1397 1398
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1399 1400
    }

G
Gene Czarcinski 已提交
1401 1402 1403 1404 1405
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
           goto cleanup;

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
           goto cleanup;
1406

G
Gene Czarcinski 已提交
1407
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
1408 1409 1410
       goto cleanup;

    if ((ret = dnsmasqSave(dctx)) < 0)
1411
        goto cleanup;
1412 1413

    ret = kill(network->dnsmasqPid, SIGHUP);
1414
 cleanup:
1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426
    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
1427
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
1428
                         virNetworkObjPtr network)
1429 1430 1431 1432 1433 1434
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1435
    }
1436
    /* now start dnsmasq if it should be started */
1437
    return networkStartDhcpDaemon(driver, network);
1438 1439
}

G
Gene Czarcinski 已提交
1440 1441 1442 1443 1444 1445
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";

1446 1447 1448
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1449
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1450 1451
    int ret = -1;
    size_t i;
1452
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
1453
    bool v6present = false, dhcp6 = false;
1454 1455

    *configstr = NULL;
1456

G
Gene Czarcinski 已提交
1457
    /* Check if DHCPv6 is needed */
1458 1459 1460
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473
        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;
    }

1474 1475 1476
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1477
    virBufferAsprintf(&configbuf, "interface %s\n"
1478 1479
                      "{\n"
                      "  AdvSendAdvert on;\n"
1480
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1481 1482 1483 1484 1485
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1486 1487

    /* add a section for each IPv6 address in the config */
1488 1489 1490
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
1491 1492 1493 1494 1495
        int prefix;
        char *netaddr;

        prefix = virNetworkIpDefPrefix(ipdef);
        if (prefix < 0) {
1496 1497 1498
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1499 1500
            goto cleanup;
        }
1501
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1502
            goto cleanup;
1503
        virBufferAsprintf(&configbuf,
1504
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1505 1506 1507
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1508 1509 1510
        VIR_FREE(netaddr);
    }

1511 1512 1513
    /* only create the string if we found at least one IPv6 address */
    if (v6present) {
        virBufferAddLit(&configbuf, "};\n");
1514

1515 1516 1517 1518 1519 1520 1521 1522
        if (virBufferError(&configbuf)) {
            virReportOOMError();
            goto cleanup;
        }
        if (!(*configstr = virBufferContentAndReset(&configbuf))) {
            virReportOOMError();
            goto cleanup;
        }
1523
    }
1524 1525

    ret = 0;
1526
 cleanup:
1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548
    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;
1549 1550 1551 1552
        goto cleanup;
    }

    /* construct the filename */
1553
    if (!(*configFile = networkRadvdConfigFileName(network->def->name)))
1554 1555
        goto cleanup;
    /* write the file */
1556
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1557 1558
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1559 1560 1561 1562 1563
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1564
 cleanup:
1565 1566 1567 1568 1569 1570
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1571
networkStartRadvd(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
G
Gene Czarcinski 已提交
1572
                        virNetworkObjPtr network)
1573 1574 1575 1576 1577 1578 1579 1580 1581
{
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1582
    /* Is dnsmasq handling RA? */
1583
   if (DNSMASQ_RA_SUPPORT(driver->dnsmasqCaps)) {
G
Gene Czarcinski 已提交
1584 1585 1586 1587
        ret = 0;
        goto cleanup;
    }

1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598
    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);
1599 1600 1601
        goto cleanup;
    }

1602
    if (virFileMakePath(driverState->pidDir) < 0) {
1603 1604
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1605
                             driverState->pidDir);
1606 1607
        goto cleanup;
    }
1608
    if (virFileMakePath(driverState->radvdStateDir) < 0) {
1609 1610
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1611
                             driverState->radvdStateDir);
1612 1613 1614 1615
        goto cleanup;
    }

    /* construct pidfile name */
1616
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1617
        goto cleanup;
1618
    if (!(pidfile = virPidFileBuildPath(driverState->pidDir, radvdpidbase)))
1619 1620 1621 1622 1623
        goto cleanup;

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

1624 1625 1626 1627
    /* 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
1628
     * virPidFileRead() below will fail if we use them).
1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643
     * 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;

1644
    if (virPidFileRead(driverState->pidDir, radvdpidbase, &network->radvdPid) < 0)
1645 1646 1647
        goto cleanup;

    ret = 0;
1648
 cleanup:
1649 1650 1651 1652 1653 1654 1655
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1656
static int
1657
networkRefreshRadvd(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
1658
                    virNetworkObjPtr network)
1659
{
G
Gene Czarcinski 已提交
1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670
    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)) {
1671
            virPidFileDelete(driverState->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1672 1673 1674 1675 1676 1677
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }

1678 1679
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
G
Gene Czarcinski 已提交
1680
        return networkStartRadvd(driver, network);
1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692

    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);
}

1693 1694
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1695
static int
1696
networkRestartRadvd(virNetworkDriverStatePtr driver,
1697
                    virNetworkObjPtr network)
1698 1699 1700 1701 1702 1703 1704 1705 1706
{
    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
1707
        if ((networkKillDaemon(network->radvdPid, "radvd",
1708 1709 1710
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1711
            virPidFileDelete(driverState->pidDir, radvdpidbase);
1712 1713 1714 1715 1716 1717 1718 1719 1720
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1721 1722 1723 1724
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
1725
networkRefreshDaemons(virNetworkDriverStatePtr driver)
1726
{
1727
    size_t i;
1728 1729 1730

    VIR_INFO("Refreshing network daemons");

1731
    for (i = 0; i < driver->networks.count; i++) {
1732 1733 1734 1735
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
1736 1737 1738
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
1739 1740 1741 1742 1743 1744
            /* 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.
             */
1745 1746
            networkRefreshDhcpDaemon(driver, network);
            networkRefreshRadvd(driver, network);
1747 1748 1749 1750 1751
        }
        virNetworkObjUnlock(network);
    }
}

1752
static void
1753
networkReloadFirewallRules(virNetworkDriverStatePtr driver)
1754
{
1755
    size_t i;
1756

1757
    VIR_INFO("Reloading iptables rules");
1758

1759
    for (i = 0; i < driver->networks.count; i++) {
1760 1761 1762 1763
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
1764 1765 1766
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
1767 1768 1769
            /* Only the three L3 network types that are configured by libvirt
             * need to have iptables rules reloaded.
             */
1770 1771
            networkRemoveFirewallRules(network->def);
            if (networkAddFirewallRules(network->def) < 0) {
1772 1773
                /* failed to add but already logged */
            }
1774
        }
1775
        virNetworkObjUnlock(network);
1776 1777 1778
    }
}

1779
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
1780
static int
1781
networkEnableIpForwarding(bool enableIPv4, bool enableIPv6)
1782
{
1783
    int ret = 0;
1784 1785 1786 1787 1788 1789 1790 1791 1792
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
                            &enabled, sizeof(enabled));
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
                            &enabled, sizeof(enabled));
#else
1793 1794 1795 1796
    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);
1797
#endif
1798
    return ret;
1799 1800
}

1801 1802
#define SYSCTL_PATH "/proc/sys"

1803 1804
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
1805 1806 1807 1808
{
    char *field = NULL;
    int ret = -1;

1809 1810 1811 1812 1813
    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",
1814
                        network->def->bridge) < 0)
1815
            goto cleanup;
1816

1817 1818 1819 1820 1821 1822
        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;
        }
1823

1824 1825 1826 1827 1828 1829 1830
        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);
1831 1832
    }

1833 1834 1835 1836 1837 1838 1839 1840
    /* 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",
1841
                    network->def->bridge) < 0)
1842 1843
        goto cleanup;

1844
    if (virFileWriteStr(field, "0", 0) < 0) {
1845
        virReportSystemError(errno,
1846 1847 1848 1849 1850
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

1851 1852 1853 1854
    /* 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",
1855
                    network->def->bridge) < 0)
1856 1857
        goto cleanup;

1858
    if (virFileWriteStr(field, "0", 0) < 0) {
1859
        virReportSystemError(errno,
1860
                             _("cannot disable %s"), field);
1861 1862 1863 1864
        goto cleanup;
    }

    ret = 0;
1865
 cleanup:
1866 1867 1868 1869
    VIR_FREE(field);
    return ret;
}

1870
/* add an IP address to a bridge */
1871
static int
D
Daniel P. Berrange 已提交
1872
networkAddAddrToBridge(virNetworkObjPtr network,
1873
                       virNetworkIpDefPtr ipdef)
1874
{
1875 1876 1877
    int prefix = virNetworkIpDefPrefix(ipdef);

    if (prefix < 0) {
1878 1879 1880
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
1881 1882 1883
        return -1;
    }

1884 1885
    if (virNetDevSetIPv4Address(network->def->bridge,
                                &ipdef->address, prefix) < 0)
1886 1887 1888 1889 1890
        return -1;

    return 0;
}

1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903
/* 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,
1904
                                    (VIR_SOCKET_ADDR_IS_FAMILY(addr, AF_INET)
1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939
                                     ? "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;
}

1940
static int
1941
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
1942 1943
                          virNetworkObjPtr network)
{
1944
    size_t i;
1945
    bool v4present = false, v6present = false;
1946 1947
    virErrorPtr save_err = NULL;
    virNetworkIpDefPtr ipdef;
1948
    virNetworkRouteDefPtr routedef;
1949
    char *macTapIfName = NULL;
1950
    int tapfd = -1;
1951

1952
    /* Check to see if any network IP collides with an existing route */
1953
    if (networkCheckRouteCollision(network->def) < 0)
1954 1955
        return -1;

1956
    /* Create and configure the bridge device */
1957
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
1958 1959
        return -1;

1960 1961 1962 1963 1964 1965 1966 1967
    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);
1968
        if (!macTapIfName)
1969
            goto err0;
1970
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
1971
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
1972
                                           &macTapIfName, &network->def->mac,
1973
                                           NULL, &tapfd, 1, NULL, NULL,
1974 1975 1976
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
1977 1978 1979 1980 1981
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

1982
    /* Set bridge options */
1983 1984 1985 1986

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
1987
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
1988
                                   network->def->delay * 1000) < 0)
1989
        goto err1;
1990

1991
    if (virNetDevBridgeSetSTP(network->def->bridge,
1992
                              network->def->stp ? true : false) < 0)
1993
        goto err1;
1994

1995 1996 1997 1998
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
1999
        goto err1;
2000

2001
    /* Add "once per network" rules */
2002
    if (networkAddFirewallRules(network->def) < 0)
2003 2004
        goto err1;

2005 2006 2007
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
2008
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2009
            v4present = true;
2010
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2011
            v6present = true;
2012

2013
        /* Add the IP address/netmask to the bridge */
D
Daniel P. Berrange 已提交
2014
        if (networkAddAddrToBridge(network, ipdef) < 0) {
2015
            goto err2;
2016
        }
2017 2018
    }

2019
    /* Bring up the bridge interface */
2020
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2021
        goto err2;
2022

2023 2024
    for (i = 0; i < network->def->nroutes; i++) {
        routedef = &network->def->routes[i];
2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035
        /* 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 */
            }
        }
    }

2036 2037
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2038
        networkEnableIpForwarding(v4present, v6present) < 0) {
2039
        virReportSystemError(errno, "%s",
2040
                             _("failed to enable IP forwarding"));
2041
        goto err3;
2042 2043
    }

2044

2045
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2046 2047
    if ((v4present || v6present) &&
        networkStartDhcpDaemon(driver, network) < 0)
2048
        goto err3;
2049

2050
    /* start radvd if there are any ipv6 addresses */
G
Gene Czarcinski 已提交
2051
    if (v6present && networkStartRadvd(driver, network) < 0)
2052 2053
        goto err4;

2054 2055 2056 2057 2058 2059 2060 2061 2062
    /* 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);
    }

2063 2064
    if (virNetDevBandwidthSet(network->def->bridge,
                              network->def->bandwidth, true) < 0) {
2065 2066 2067
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("cannot set bandwidth limits on %s"),
                       network->def->bridge);
2068 2069 2070
        goto err5;
    }

2071
    VIR_FREE(macTapIfName);
2072 2073 2074

    return 0;

2075
 err5:
2076
    virNetDevBandwidthClear(network->def->bridge);
2077

2078 2079 2080 2081
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2082 2083 2084 2085 2086
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2087 2088 2089
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2090
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2091

2092 2093 2094
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2095
    networkRemoveFirewallRules(network->def);
2096 2097

 err1:
2098 2099 2100
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2101
    if (macTapIfName) {
2102
        VIR_FORCE_CLOSE(tapfd);
H
Hu Tao 已提交
2103 2104 2105
        ignore_value(virNetDevTapDelete(macTapIfName));
        VIR_FREE(macTapIfName);
    }
2106 2107

 err0:
2108 2109
    if (!save_err)
        save_err = virSaveLastError();
2110
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2111

2112 2113 2114 2115
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2116
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2117 2118 2119
    return -1;
}

2120
static int networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
R
Roman Bogorodskiy 已提交
2121
                                         virNetworkObjPtr network)
2122
{
2123
    virNetDevBandwidthClear(network->def->bridge);
2124

2125 2126 2127 2128 2129
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2130
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2131
            virPidFileDelete(driverState->pidDir, radvdpidbase);
2132 2133 2134 2135
            VIR_FREE(radvdpidbase);
        }
    }

2136 2137 2138
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2139
    if (network->def->mac_specified) {
2140
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2141
        if (macTapIfName) {
2142
            ignore_value(virNetDevTapDelete(macTapIfName));
2143 2144 2145 2146
            VIR_FREE(macTapIfName);
        }
    }

2147
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2148

2149
    networkRemoveFirewallRules(network->def);
2150

2151
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2152

2153
    /* See if its still alive and really really kill it */
2154
    if (network->dnsmasqPid > 0 &&
2155
        (kill(network->dnsmasqPid, 0) == 0))
2156 2157
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2158 2159 2160 2161 2162 2163

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

2164 2165 2166 2167
    return 0;
}

static int
2168
networkStartNetworkExternal(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
2169 2170 2171
                            virNetworkObjPtr network ATTRIBUTE_UNUSED)
{
    /* put anything here that needs to be done each time a network of
2172
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2173 2174 2175 2176 2177 2178
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

2179
static int networkShutdownNetworkExternal(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
2180 2181 2182
                                        virNetworkObjPtr network ATTRIBUTE_UNUSED)
{
    /* put anything here that needs to be done each time a network of
2183
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2184 2185 2186 2187 2188 2189 2190
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2191
networkStartNetwork(virNetworkDriverStatePtr driver,
2192 2193
                    virNetworkObjPtr network)
{
2194 2195 2196
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2197 2198

    if (virNetworkObjIsActive(network)) {
2199 2200
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2201
        return ret;
2202 2203
    }

2204 2205 2206
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2207
    if (virNetworkObjSetDefTransient(network, true) < 0)
2208
        goto cleanup;
2209

2210 2211
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2212
    if (networkRunHook(network, NULL, NULL,
2213 2214 2215 2216
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2217
    switch (network->def->forward.type) {
2218 2219 2220 2221

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2222 2223
        if (networkStartNetworkVirtual(driver, network) < 0)
            goto cleanup;
2224 2225 2226 2227 2228 2229
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2230
    case VIR_NETWORK_FORWARD_HOSTDEV:
2231 2232
        if (networkStartNetworkExternal(driver, network) < 0)
            goto cleanup;
2233 2234 2235
        break;
    }

2236
    /* finally we can call the 'started' hook script if any */
2237
    if (networkRunHook(network, NULL, NULL,
2238 2239 2240 2241
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2242 2243 2244
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2245 2246 2247
    VIR_DEBUG("Writing network status to disk");
    if (virNetworkSaveStatus(driverState->stateDir, network) < 0)
        goto cleanup;
2248 2249

    network->active = 1;
2250 2251
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2252

2253
 cleanup:
2254
    if (ret < 0) {
2255
        virNetworkObjUnsetDefTransient(network);
2256 2257 2258 2259 2260 2261 2262 2263 2264 2265
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
        networkShutdownNetwork(driver, network);
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2266
static int networkShutdownNetwork(virNetworkDriverStatePtr driver,
2267 2268 2269 2270 2271 2272 2273 2274 2275 2276
                                        virNetworkObjPtr network)
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

2277 2278
    stateFile = virNetworkConfigFile(driverState->stateDir,
                                     network->def->name);
2279 2280 2281 2282 2283 2284
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2285
    switch (network->def->forward.type) {
2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296

    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:
2297
    case VIR_NETWORK_FORWARD_HOSTDEV:
2298 2299 2300 2301
        ret = networkShutdownNetworkExternal(driver, network);
        break;
    }

2302
    /* now that we know it's stopped call the hook if present */
2303
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2304 2305
                   VIR_HOOK_SUBOP_END);

2306
    network->active = 0;
2307
    virNetworkObjUnsetDefTransient(network);
2308
    return ret;
2309 2310 2311
}


2312
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2313 2314
                                         const unsigned char *uuid)
{
2315
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2316 2317
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2318

2319
    networkDriverLock(driver);
2320
    network = virNetworkFindByUUID(&driver->networks, uuid);
2321
    networkDriverUnlock(driver);
2322
    if (!network) {
2323 2324
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
2325
        goto cleanup;
2326 2327
    }

2328 2329 2330
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2331 2332
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2333
 cleanup:
2334 2335
    if (network)
        virNetworkObjUnlock(network);
2336
    return ret;
2337 2338
}

2339
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2340 2341
                                         const char *name)
{
2342
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2343 2344 2345
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2346
    networkDriverLock(driver);
2347
    network = virNetworkFindByName(&driver->networks, name);
2348
    networkDriverUnlock(driver);
2349
    if (!network) {
2350 2351
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2352
        goto cleanup;
2353 2354
    }

2355 2356 2357
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2358 2359
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2360
 cleanup:
2361 2362
    if (network)
        virNetworkObjUnlock(network);
2363
    return ret;
2364 2365
}

2366 2367 2368
static virDrvOpenStatus networkOpen(virConnectPtr conn,
                                    virConnectAuthPtr auth ATTRIBUTE_UNUSED,
                                    unsigned int flags)
2369 2370 2371
{
    virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);

2372 2373 2374 2375 2376 2377 2378
    if (!driverState)
        return VIR_DRV_OPEN_DECLINED;

    conn->networkPrivateData = driverState;
    return VIR_DRV_OPEN_SUCCESS;
}

2379 2380
static int networkClose(virConnectPtr conn)
{
2381 2382 2383 2384
    conn->networkPrivateData = NULL;
    return 0;
}

2385 2386
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2387 2388
    int nactive = 0;
    size_t i;
2389
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2390

2391 2392 2393
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2394
    networkDriverLock(driver);
2395
    for (i = 0; i < driver->networks.count; i++) {
2396 2397 2398 2399
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectNumOfNetworksCheckACL(conn, obj->def) &&
            virNetworkObjIsActive(obj))
2400
            nactive++;
2401
        virNetworkObjUnlock(obj);
2402 2403
    }
    networkDriverUnlock(driver);
2404

2405 2406 2407
    return nactive;
}

2408
static int networkConnectListNetworks(virConnectPtr conn, char **const names, int nnames) {
2409
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2410 2411
    int got = 0;
    size_t i;
2412

2413 2414 2415
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2416
    networkDriverLock(driver);
2417
    for (i = 0; i < driver->networks.count && got < nnames; i++) {
2418 2419 2420 2421 2422 2423
        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);
2424 2425 2426 2427
                goto cleanup;
            }
            got++;
        }
2428
        virNetworkObjUnlock(obj);
2429
    }
2430 2431
    networkDriverUnlock(driver);

2432 2433 2434
    return got;

 cleanup:
2435
    networkDriverUnlock(driver);
2436
    for (i = 0; i < got; i++)
2437 2438 2439 2440
        VIR_FREE(names[i]);
    return -1;
}

2441 2442
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2443 2444
    int ninactive = 0;
    size_t i;
2445
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2446

2447 2448 2449
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2450
    networkDriverLock(driver);
2451
    for (i = 0; i < driver->networks.count; i++) {
2452 2453 2454 2455
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectNumOfDefinedNetworksCheckACL(conn, obj->def) &&
            !virNetworkObjIsActive(obj))
2456
            ninactive++;
2457
        virNetworkObjUnlock(obj);
2458 2459
    }
    networkDriverUnlock(driver);
2460

2461 2462 2463
    return ninactive;
}

2464
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames) {
2465
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2466 2467
    int got = 0;
    size_t i;
2468

2469 2470 2471
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2472
    networkDriverLock(driver);
2473
    for (i = 0; i < driver->networks.count && got < nnames; i++) {
2474 2475 2476 2477 2478 2479
        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);
2480 2481 2482 2483
                goto cleanup;
            }
            got++;
        }
2484
        virNetworkObjUnlock(obj);
2485
    }
2486
    networkDriverUnlock(driver);
2487 2488 2489
    return got;

 cleanup:
2490
    networkDriverUnlock(driver);
2491
    for (i = 0; i < got; i++)
2492 2493 2494 2495
        VIR_FREE(names[i]);
    return -1;
}

2496
static int
2497 2498 2499
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2500
{
2501
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2502 2503 2504 2505
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2506 2507 2508
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2509
    networkDriverLock(driver);
2510 2511 2512
    ret = virNetworkObjListExport(conn, driver->networks, nets,
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2513 2514
    networkDriverUnlock(driver);

2515
 cleanup:
2516 2517
    return ret;
}
2518

2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
    int ret = -1;

    if (virConnectNetworkEventRegisterAnyEnsureACL(conn) < 0)
        goto cleanup;

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
2534
                                       net, eventID, callback,
2535 2536 2537
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2538
 cleanup:
2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551
    return ret;
}

static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
    int ret = -1;

    if (virConnectNetworkEventDeregisterAnyEnsureACL(conn) < 0)
        goto cleanup;

2552 2553 2554 2555 2556 2557
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
2558

2559
 cleanup:
2560 2561 2562
    return ret;
}

2563 2564 2565 2566 2567
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2568 2569
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2570 2571 2572 2573

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

2574 2575
    ret = virNetworkObjIsActive(obj);

2576
 cleanup:
2577 2578 2579 2580 2581 2582 2583 2584 2585 2586
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}

static int networkIsPersistent(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2587 2588
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2589 2590 2591 2592

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

2593 2594
    ret = obj->persistent;

2595
 cleanup:
2596 2597 2598 2599 2600 2601
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}


2602
static int
2603
networkValidate(virNetworkDriverStatePtr driver,
2604 2605
                virNetworkDefPtr def,
                bool check_active)
2606
{
2607
    size_t i;
2608 2609
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2610
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
2611
    bool ipv4def = false, ipv6def = false;
2612 2613 2614 2615 2616 2617 2618 2619

    /* 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
     */
2620 2621 2622
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2623 2624 2625 2626 2627

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

        virNetworkSetBridgeMacAddr(def);
2628 2629
    } else {
        /* They are also the only types that currently support setting
2630 2631
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
2632
         */
2633 2634 2635 2636 2637 2638 2639 2640
        if (def->mac_specified) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <mac> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
                           virNetworkForwardTypeToString(def->forward.type));
            return -1;
        }
2641 2642 2643 2644 2645
        if (virNetworkDefGetIpByIndex(def, AF_UNSPEC, 0)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2646
                           virNetworkForwardTypeToString(def->forward.type));
2647 2648
            return -1;
        }
2649
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
2650 2651 2652 2653
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2654
                           virNetworkForwardTypeToString(def->forward.type));
2655 2656 2657 2658 2659 2660 2661
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2662
                           virNetworkForwardTypeToString(def->forward.type));
2663 2664
            return -1;
        }
2665 2666 2667 2668 2669 2670 2671 2672
        if (def->bandwidth) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported network-wide <bandwidth> element "
                             "in network %s with forward mode='%s'"),
                           def->name,
                           virNetworkForwardTypeToString(def->forward.type));
            return -1;
        }
2673 2674
    }

G
Gene Czarcinski 已提交
2675 2676 2677
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
2678 2679 2680
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
2681 2682 2683 2684 2685
        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 -- "
2686 2687
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704
                    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;
                }
2705 2706 2707
            }
        }
    }
2708 2709 2710 2711 2712 2713

    /* 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.
     */

2714
    vlanAllowed = ((def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
2715
                   def->virtPortProfile &&
2716 2717 2718
                   def->virtPortProfile->virtPortType
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) ||
                   def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV);
2719 2720

    vlanUsed = def->vlan.nTags > 0;
2721 2722
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
2723 2724 2725 2726 2727
            /* 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.
             */
2728
            if (def->portGroups[i].virtPortProfile) {
2729
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
2730
                    def->portGroups[i].virtPortProfile->virtPortType
2731 2732 2733 2734 2735 2736 2737
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
2738
        }
2739
        if (def->portGroups[i].isDefault) {
2740 2741 2742 2743 2744
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
2745
                               def->name, defaultPortGroup->name,
2746
                               def->portGroups[i].name);
2747
                return -1;
2748
            }
2749
            defaultPortGroup = &def->portGroups[i];
2750
        }
2751
    }
2752 2753 2754 2755 2756 2757 2758
    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.
         */
2759 2760 2761 2762 2763 2764 2765 2766 2767
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
    return 0;
}

2768 2769
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
2770
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2771
    virNetworkDefPtr def;
2772
    virNetworkObjPtr network = NULL;
2773
    virNetworkPtr ret = NULL;
2774
    virObjectEventPtr event = NULL;
2775

2776 2777
    networkDriverLock(driver);

2778
    if (!(def = virNetworkDefParseString(xml)))
2779
        goto cleanup;
2780

2781 2782 2783
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

2784
    if (networkValidate(driver, def, true) < 0)
2785 2786
       goto cleanup;

2787 2788 2789
    /* NB: even though this transient network hasn't yet been started,
     * we assign the def with live = true in anticipation that it will
     * be started momentarily.
2790
     */
2791
    if (!(network = virNetworkAssignDef(&driver->networks, def, true)))
2792 2793
        goto cleanup;
    def = NULL;
2794

2795
    if (networkStartNetwork(driver, network) < 0) {
2796 2797
        virNetworkRemoveInactive(&driver->networks,
                                 network);
2798
        network = NULL;
2799
        goto cleanup;
2800 2801
    }

2802 2803
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
2804 2805
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
2806

2807
    VIR_INFO("Creating network '%s'", network->def->name);
2808 2809
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2810
 cleanup:
2811
    virNetworkDefFree(def);
2812 2813
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
2814 2815 2816
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
2817
    return ret;
2818 2819
}

2820 2821
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
2822
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2823
    virNetworkDefPtr def = NULL;
2824
    bool freeDef = true;
2825
    virNetworkObjPtr network = NULL;
2826
    virNetworkPtr ret = NULL;
2827
    virObjectEventPtr event = NULL;
2828

2829 2830
    networkDriverLock(driver);

2831
    if (!(def = virNetworkDefParseString(xml)))
2832
        goto cleanup;
2833

2834 2835 2836
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

2837
    if (networkValidate(driver, def, false) < 0)
2838 2839
       goto cleanup;

2840 2841
    if (!(network = virNetworkAssignDef(&driver->networks, def, false)))
       goto cleanup;
2842

2843
    /* def was assigned to network object */
2844
    freeDef = false;
2845 2846

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
2847 2848 2849 2850 2851
        if (!virNetworkObjIsActive(network)) {
            virNetworkRemoveInactive(&driver->networks, network);
            network = NULL;
            goto cleanup;
        }
2852 2853 2854 2855 2856
        /* if network was active already, just undo new persistent
         * definition by making it transient.
         * XXX - this isn't necessarily the correct thing to do.
         */
        virNetworkObjAssignDef(network, NULL, false);
2857 2858 2859
        goto cleanup;
    }

2860
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
2861 2862
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
2863

2864 2865
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
2866

2867
 cleanup:
2868 2869
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
2870 2871
    if (freeDef)
       virNetworkDefFree(def);
2872 2873 2874
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
2875
    return ret;
2876 2877
}

2878
static int
2879 2880
networkUndefine(virNetworkPtr net)
{
2881
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
2882
    virNetworkObjPtr network;
2883
    int ret = -1;
2884
    bool active = false;
2885
    virObjectEventPtr event = NULL;
2886

2887 2888
    networkDriverLock(driver);

2889
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
2890
    if (!network) {
2891 2892
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
2893
        goto cleanup;
2894 2895
    }

2896 2897 2898
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

2899 2900
    if (virNetworkObjIsActive(network))
        active = true;
2901

2902
    /* remove autostart link */
2903
    if (virNetworkDeleteConfig(driver->networkConfigDir,
2904 2905
                               driver->networkAutostartDir,
                               network) < 0)
2906
        goto cleanup;
2907
    network->autostart = 0;
2908

2909 2910
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
2911 2912
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
2913

2914
    VIR_INFO("Undefining network '%s'", network->def->name);
2915 2916 2917 2918 2919
    if (!active) {
        if (networkRemoveInactive(driver, network) < 0) {
            network = NULL;
            goto cleanup;
        }
2920
        network = NULL;
2921 2922 2923 2924 2925 2926
    } else {

        /* if the network still exists, it was active, and we need to make
         * it transient (by deleting the persistent def)
         */
        virNetworkObjAssignDef(network, NULL, false);
2927 2928
    }

2929
    ret = 0;
2930

2931
 cleanup:
2932 2933
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
2934 2935 2936
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
2937
    return ret;
2938 2939
}

2940 2941 2942 2943 2944 2945 2946 2947
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
2948
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
2949
    virNetworkObjPtr network = NULL;
2950 2951
    int isActive, ret = -1;
    size_t i;
2952 2953
    virNetworkIpDefPtr ipdef;
    bool oldDhcpActive = false;
2954
    bool needFirewallRefresh = false;
2955

2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969

    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;
    }

2970 2971 2972
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

2973
    /* see if we are listening for dhcp pre-modification */
2974 2975 2976
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
2977 2978 2979 2980 2981 2982
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

2983 2984 2985 2986
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
    */
    isActive = virNetworkObjIsActive(network);
2987 2988
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
2989 2990 2991 2992 2993 2994 2995
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012
    if (isActive && (flags & VIR_NETWORK_UPDATE_AFFECT_LIVE)) {
        /* Take care of anything that must be done before updating the
         * live NetworkDef.
         */
        if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE ||
            network->def->forward.type == VIR_NETWORK_FORWARD_NAT ||
            network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
            switch (section) {
            case VIR_NETWORK_SECTION_FORWARD:
            case VIR_NETWORK_SECTION_FORWARD_INTERFACE:
            case VIR_NETWORK_SECTION_IP:
            case VIR_NETWORK_SECTION_IP_DHCP_RANGE:
            case VIR_NETWORK_SECTION_IP_DHCP_HOST:
                /* these could affect the firewall rules, so remove the
                 * old rules (and remember to load new ones after the
                 * update).
                 */
3013
                networkRemoveFirewallRules(network->def);
3014 3015 3016 3017 3018 3019 3020 3021
                needFirewallRefresh = true;
                break;
            default:
                break;
            }
        }
    }

3022
    /* update the network config in memory/on disk */
3023 3024
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3025
            ignore_value(networkAddFirewallRules(network->def));
3026 3027 3028
        goto cleanup;
    }

3029
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052
        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.
             */
3053
            if (networkRestartDhcpDaemon(driver, network) < 0)
3054 3055
                goto cleanup;

3056 3057 3058 3059 3060 3061 3062 3063
        } 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;

3064 3065 3066
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
                 i++) {
3067 3068 3069 3070 3071 3072 3073
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3074 3075
                 networkRestartDhcpDaemon(driver, network) < 0) ||
                networkRefreshDhcpDaemon(driver, network) < 0) {
3076 3077 3078 3079
                goto cleanup;
            }

        } else if (section == VIR_NETWORK_SECTION_DNS_HOST ||
3080 3081 3082 3083 3084 3085
                   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.
             */
3086
            if (networkRefreshDhcpDaemon(driver, network) < 0)
3087 3088 3089 3090 3091 3092 3093 3094
                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.
             */
3095
            if (networkRefreshRadvd(driver, network) < 0)
3096 3097 3098 3099
                goto cleanup;
        }

        /* save current network state to disk */
3100 3101
        if ((ret = virNetworkSaveStatus(driverState->stateDir,
                                        network)) < 0) {
3102
            goto cleanup;
3103
        }
3104 3105
    }
    ret = 0;
3106
 cleanup:
3107 3108 3109 3110 3111 3112
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
    return ret;
}

3113 3114
static int networkCreate(virNetworkPtr net)
{
3115
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3116 3117
    virNetworkObjPtr network;
    int ret = -1;
3118
    virObjectEventPtr event = NULL;
3119

3120
    networkDriverLock(driver);
3121
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3122

3123
    if (!network) {
3124 3125
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3126
        goto cleanup;
3127 3128
    }

3129 3130 3131
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3132
    ret = networkStartNetwork(driver, network);
3133

3134 3135
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3136 3137
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3138

3139
 cleanup:
3140 3141
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3142 3143
    if (network)
        virNetworkObjUnlock(network);
3144
    networkDriverUnlock(driver);
3145
    return ret;
3146 3147
}

3148 3149
static int networkDestroy(virNetworkPtr net)
{
3150
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3151 3152
    virNetworkObjPtr network;
    int ret = -1;
3153
    virObjectEventPtr event = NULL;
3154

3155
    networkDriverLock(driver);
3156
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3157

3158
    if (!network) {
3159 3160
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3161
        goto cleanup;
3162 3163
    }

3164 3165 3166
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3167
    if (!virNetworkObjIsActive(network)) {
3168
        virReportError(VIR_ERR_OPERATION_INVALID,
3169 3170
                       _("network '%s' is not active"),
                       network->def->name);
3171 3172 3173
        goto cleanup;
    }

3174 3175 3176
    if ((ret = networkShutdownNetwork(driver, network)) < 0)
        goto cleanup;

3177 3178
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3179 3180
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3181

3182
    if (!network->persistent) {
3183 3184 3185 3186 3187
        if (networkRemoveInactive(driver, network) < 0) {
            network = NULL;
            ret = -1;
            goto cleanup;
        }
3188 3189
        network = NULL;
    }
3190

3191
 cleanup:
3192 3193
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3194 3195
    if (network)
        virNetworkObjUnlock(network);
3196
    networkDriverUnlock(driver);
3197 3198 3199
    return ret;
}

3200
static char *networkGetXMLDesc(virNetworkPtr net,
3201
                               unsigned int flags)
3202
{
3203
    virNetworkObjPtr network;
3204
    virNetworkDefPtr def;
3205
    char *ret = NULL;
3206

3207
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3208

3209 3210
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3211

3212 3213 3214
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3215 3216 3217 3218 3219 3220
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3221

3222
 cleanup:
3223 3224
    if (network)
        virNetworkObjUnlock(network);
3225
    return ret;
3226 3227 3228
}

static char *networkGetBridgeName(virNetworkPtr net) {
3229 3230 3231
    virNetworkObjPtr network;
    char *bridge = NULL;

3232 3233
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3234

3235 3236 3237
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3238
    if (!(network->def->bridge)) {
3239 3240 3241
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3242 3243 3244
        goto cleanup;
    }

3245
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3246

3247
 cleanup:
3248 3249
    if (network)
        virNetworkObjUnlock(network);
3250 3251 3252 3253
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
3254 3255
                             int *autostart)
{
3256 3257
    virNetworkObjPtr network;
    int ret = -1;
3258

3259 3260
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3261

3262 3263 3264
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3265
    *autostart = network->autostart;
3266
    ret = 0;
3267

3268
 cleanup:
3269 3270
    if (network)
        virNetworkObjUnlock(network);
3271
    return ret;
3272 3273 3274
}

static int networkSetAutostart(virNetworkPtr net,
3275 3276
                               int autostart)
{
3277
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3278
    virNetworkObjPtr network;
3279
    char *configFile = NULL, *autostartLink = NULL;
3280
    int ret = -1;
3281

3282
    networkDriverLock(driver);
3283
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3284

3285
    if (!network) {
3286 3287
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3288
        goto cleanup;
3289 3290
    }

3291 3292 3293
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3294
    if (!network->persistent) {
3295 3296
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3297 3298 3299
        goto cleanup;
    }

3300 3301
    autostart = (autostart != 0);

3302
    if (network->autostart != autostart) {
3303
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3304
            goto cleanup;
3305
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3306 3307
            goto cleanup;

3308
        if (autostart) {
3309
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3310
                virReportSystemError(errno,
3311 3312
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3313 3314
                goto cleanup;
            }
3315

3316
            if (symlink(configFile, autostartLink) < 0) {
3317
                virReportSystemError(errno,
3318
                                     _("Failed to create symlink '%s' to '%s'"),
3319
                                     autostartLink, configFile);
3320 3321 3322
                goto cleanup;
            }
        } else {
3323
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3324
                virReportSystemError(errno,
3325
                                     _("Failed to delete symlink '%s'"),
3326
                                     autostartLink);
3327 3328
                goto cleanup;
            }
3329 3330
        }

3331
        network->autostart = autostart;
3332
    }
3333
    ret = 0;
3334

3335
 cleanup:
3336 3337
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3338 3339
    if (network)
        virNetworkObjUnlock(network);
3340
    networkDriverUnlock(driver);
3341
    return ret;
3342 3343 3344 3345 3346
}


static virNetworkDriver networkDriver = {
    "Network",
3347 3348 3349 3350 3351 3352 3353
    .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 */
3354 3355
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
3356 3357
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3358 3359
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3360
    .networkUndefine = networkUndefine, /* 0.2.0 */
3361
    .networkUpdate = networkUpdate, /* 0.10.2 */
3362
    .networkCreate = networkCreate, /* 0.2.0 */
3363 3364 3365 3366 3367 3368 3369
    .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 */
3370 3371 3372
};

static virStateDriver networkStateDriver = {
3373
    .name = "Network",
3374
    .stateInitialize  = networkStateInitialize,
3375
    .stateAutoStart  = networkStateAutoStart,
3376 3377
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3378 3379
};

3380 3381
int networkRegister(void)
{
3382 3383
    if (virRegisterNetworkDriver(&networkDriver) < 0)
        return -1;
3384 3385
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
3386 3387
    return 0;
}
3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398

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

/* 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.
 */

3399 3400 3401 3402 3403 3404
/* networkCreateInterfacePool:
 * @netdef: the original NetDef from the network
 *
 * Creates an implicit interface pool of VF's when a PF dev is given
 */
static int
3405 3406
networkCreateInterfacePool(virNetworkDefPtr netdef)
{
3407
    size_t num_virt_fns = 0;
3408
    char **vfname = NULL;
3409
    virPCIDeviceAddressPtr *virt_fns;
3410 3411
    int ret = -1;
    size_t i;
3412

3413
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev,
3414
                                      &vfname, &virt_fns, &num_virt_fns)) < 0) {
3415 3416
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Could not get Virtual functions on %s"),
3417
                       netdef->forward.pfs->dev);
3418 3419 3420 3421 3422 3423
        goto finish;
    }

    if (num_virt_fns == 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("No Vf's present on SRIOV PF %s"),
3424
                       netdef->forward.pfs->dev);
3425 3426 3427
       goto finish;
    }

3428
    if (VIR_ALLOC_N(netdef->forward.ifs, num_virt_fns) < 0)
3429 3430
        goto finish;

3431
    netdef->forward.nifs = num_virt_fns;
3432

3433
    for (i = 0; i < netdef->forward.nifs; i++) {
3434 3435 3436 3437
        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)) {
3438 3439 3440
            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)
3441
                    goto finish;
3442
            } else {
3443
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
3444
                               _("Direct mode types require interface names"));
3445 3446
                goto finish;
            }
3447
        }
3448
        else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3449
            /* VF's are always PCI devices */
3450 3451 3452 3453 3454
            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;
3455
        }
3456 3457 3458
    }

    ret = 0;
3459
 finish:
3460 3461 3462
    for (i = 0; i < num_virt_fns; i++) {
        VIR_FREE(vfname[i]);
        VIR_FREE(virt_fns[i]);
3463
    }
3464
    VIR_FREE(vfname);
3465
    VIR_FREE(virt_fns);
3466 3467 3468
    return ret;
}

3469
/* networkAllocateActualDevice:
3470
 * @dom: domain definition that @iface belongs to
3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481
 * @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
3482 3483
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
3484
{
3485
    virNetworkDriverStatePtr driver = driverState;
3486 3487 3488
    enum virDomainNetType actualType = iface->type;
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
3489
    virNetDevBandwidthPtr bandwidth = NULL;
3490 3491 3492
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3493
    virNetworkForwardIfDefPtr dev = NULL;
3494
    size_t i;
3495 3496 3497
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3498
        goto validate;
3499 3500 3501 3502 3503 3504 3505 3506

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

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
3507 3508 3509
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3510
        goto error;
3511 3512
    }
    netdef = network->def;
3513

3514 3515 3516 3517 3518 3519 3520
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

3521 3522 3523
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

3524 3525 3526 3527 3528 3529 3530 3531 3532 3533
    /* 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.
     */
3534 3535 3536 3537 3538 3539

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

3540 3541
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
3542
        goto error;
3543

3544 3545 3546 3547 3548 3549 3550 3551
    /* 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;

3552 3553
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
3554

3555 3556 3557
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
3558 3559 3560 3561
        /* for these forward types, the actual net type really *is*
         *NETWORK; we just keep the info from the portgroup in
         * iface->data.network.actual
        */
3562
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
3563 3564 3565 3566

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

3567
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
3568
               netdef->bridge) {
3569 3570 3571 3572 3573

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

3574
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
3575 3576
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
3577
            goto error;
3578

3579 3580 3581 3582 3583 3584 3585 3586
        /* 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) {
3587
            goto error;
3588 3589 3590 3591 3592 3593 3594 3595 3596 3597
        }
        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);
3598
                goto error;
3599 3600 3601
            }
        }

3602
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3603

3604
        virDomainHostdevSubsysPCIBackendType backend;
3605

3606
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
3607
        if (netdef->forward.npfs > 0 && netdef->forward.nifs <= 0 &&
3608 3609 3610 3611 3612
            networkCreateInterfacePool(netdef) < 0) {
            goto error;
        }

        /* pick first dev with 0 connections */
3613 3614 3615
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629
                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;
3630
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
3631
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
3632
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
3633

3634 3635 3636
        switch (netdef->forward.driverName)
        {
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
3637
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
3638 3639
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
3640
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
3641 3642
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
3643
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654
            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;

3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679
        /* 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;
            }
        }

3680 3681 3682 3683
    } 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)) {
3684 3685 3686 3687 3688 3689

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
3690
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
3691
        switch (netdef->forward.type) {
3692
        case VIR_NETWORK_FORWARD_BRIDGE:
3693
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
3694 3695
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
3696
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
3697 3698
            break;
        case VIR_NETWORK_FORWARD_VEPA:
3699
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
3700 3701
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
3702
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
3703 3704 3705
            break;
        }

3706 3707 3708 3709 3710 3711 3712 3713
        /* 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) {
3714
            goto error;
3715
        }
3716
        virtport = iface->data.network.actual->virtPortProfile;
3717
        if (virtport) {
3718 3719 3720 3721 3722 3723 3724 3725
            /* 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);
3726
                goto error;
3727 3728
            }
        }
3729

3730 3731 3732
        /* 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).
         */
3733
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
3734 3735 3736 3737
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
3738
            goto error;
3739 3740 3741
        } else {
            /* pick an interface from the pool */

3742
            if (netdef->forward.npfs > 0 && netdef->forward.nifs == 0 &&
3743 3744 3745 3746
                networkCreateInterfacePool(netdef) < 0) {
                goto error;
            }

3747 3748 3749 3750 3751
            /* 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.
3752
             */
3753 3754
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
3755 3756 3757
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
3758

3759
                /* pick first dev with 0 connections */
3760 3761 3762
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
3763 3764 3765 3766 3767
                        break;
                    }
                }
            } else {
                /* pick least used dev */
3768
                dev = &netdef->forward.ifs[0];
3769 3770 3771
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
3772 3773 3774 3775
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
3776 3777 3778 3779
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
3780
                goto error;
3781
            }
3782 3783
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
3784
                goto error;
3785 3786 3787
        }
    }

3788
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
3789
        goto error;
3790

3791
 validate:
3792 3793 3794 3795 3796
    /* 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.
     */

3797
    if (virDomainNetGetActualVlan(iface)) {
3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826
        /* 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;
        }
    }

    if (netdef) {
        netdef->connections++;
        VIR_DEBUG("Using network %s, %d connections",
                  netdef->name, netdef->connections);
3827

3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846
        if (dev) {
            /* mark the allocation */
            dev->connections++;
            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);
            }
        }

        /* finally we can call the 'plugged' hook script if any */
        if (networkRunHook(network, dom, iface,
                           VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
                           VIR_HOOK_SUBOP_BEGIN) < 0) {
            /* adjust for failure */
3847
            netdef->connections--;
3848 3849 3850 3851
            if (dev)
                dev->connections--;
            goto error;
        }
3852 3853
    }

3854
    ret = 0;
3855

3856
 cleanup:
3857 3858
    if (network)
        virNetworkObjUnlock(network);
3859 3860
    return ret;

3861
 error:
3862
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
3863 3864 3865
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
3866
    goto cleanup;
3867 3868 3869
}

/* networkNotifyActualDevice:
3870
 * @dom: domain definition that @iface belongs to
3871 3872 3873 3874 3875 3876 3877 3878 3879 3880
 * @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
3881 3882
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
3883
{
3884
    virNetworkDriverStatePtr driver = driverState;
3885
    enum virDomainNetType actualType = virDomainNetGetActualType(iface);
3886 3887
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
3888
    virNetworkForwardIfDefPtr dev = NULL;
3889 3890
    size_t i;
    int ret = -1;
3891 3892 3893 3894 3895 3896 3897 3898

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

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
3899 3900 3901
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3902 3903 3904 3905 3906
        goto error;
    }
    netdef = network->def;

    if (!iface->data.network.actual ||
3907 3908
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
3909 3910
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
3911 3912
    }

3913
    if (netdef->forward.npfs > 0 && netdef->forward.nifs == 0 &&
3914
        networkCreateInterfacePool(netdef) < 0) {
3915
        goto error;
3916
    }
3917
    if (netdef->forward.nifs == 0) {
3918
        virReportError(VIR_ERR_INTERNAL_ERROR,
3919 3920
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
3921
                       netdef->name);
3922
        goto error;
3923
    }
3924

3925 3926
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
3927

3928 3929 3930 3931 3932 3933 3934 3935 3936
        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 */
3937 3938
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
3939
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
3940 3941
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
3942 3943 3944 3945 3946
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
3947
            virReportError(VIR_ERR_INTERNAL_ERROR,
3948 3949
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
3950
                           netdef->name, actualDev);
3951
            goto error;
3952 3953
        }

3954
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
3955 3956
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
3957
         */
3958
        if ((dev->connections > 0) &&
3959 3960
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
3961 3962
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
3963
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
3964
            virReportError(VIR_ERR_INTERNAL_ERROR,
3965 3966
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
3967
                           netdef->name, actualDev);
3968
            goto error;
3969
        }
3970

3971
        /* we are now assured of success, so mark the allocation */
3972
        dev->connections++;
3973
        VIR_DEBUG("Using physical device %s, connections %d",
3974
                  dev->device.dev, dev->connections);
3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987

    }  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 */
3988 3989
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
3990
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
3991
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
3992 3993
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
3994 3995 3996 3997 3998 3999 4000 4001 4002
                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,
4003 4004 4005 4006
                           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);
4007 4008 4009 4010 4011 4012 4013 4014
                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) &&
4015
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031
            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);
4032 4033
    }

4034
 success:
4035 4036 4037
    netdef->connections++;
    VIR_DEBUG("Using network %s, %d connections",
              netdef->name, netdef->connections);
4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048

    /* finally we can call the 'plugged' hook script if any */
    if (networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
                       VIR_HOOK_SUBOP_BEGIN) < 0) {
        /* adjust for failure */
        if (dev)
            dev->connections--;
        netdef->connections--;
        goto error;
    }

4049
    ret = 0;
4050
 cleanup:
4051 4052 4053
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4054

4055
 error:
4056
    goto cleanup;
4057 4058 4059 4060
}


/* networkReleaseActualDevice:
4061
 * @dom: domain definition that @iface belongs to
4062 4063 4064 4065 4066 4067 4068 4069 4070 4071
 * @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
4072 4073
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4074
{
4075
    virNetworkDriverStatePtr driver = driverState;
4076
    enum virDomainNetType actualType = virDomainNetGetActualType(iface);
4077
    virNetworkObjPtr network;
4078
    virNetworkDefPtr netdef;
4079
    virNetworkForwardIfDefPtr dev = NULL;
4080 4081
    size_t i;
    int ret = -1;
4082 4083 4084 4085 4086 4087 4088 4089

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

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
4090 4091 4092
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4093 4094 4095 4096
        goto error;
    }
    netdef = network->def;

4097 4098
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4099 4100 4101 4102 4103
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) &&
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4104 4105 4106
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4107 4108
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4109 4110
    }

4111
    if (netdef->forward.nifs == 0) {
4112
        virReportError(VIR_ERR_INTERNAL_ERROR,
4113
                       _("network '%s' uses a direct/hostdev mode, but "
4114 4115
                         "has no forward dev and no interface pool"),
                       netdef->name);
4116
        goto error;
4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128
    }

    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;
        }
4129

4130 4131
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4132
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4133 4134
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4135 4136 4137
                break;
            }
        }
4138

4139
        if (!dev) {
4140
            virReportError(VIR_ERR_INTERNAL_ERROR,
4141 4142
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4143
                           netdef->name, actualDev);
4144
            goto error;
4145 4146
        }

4147
        dev->connections--;
4148
        VIR_DEBUG("Releasing physical device %s, connections %d",
4149
                  dev->device.dev, dev->connections);
4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160

    } 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;
        }

4161 4162
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4163
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4164
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4165 4166
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4167 4168 4169 4170 4171 4172 4173 4174 4175
                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,
4176 4177 4178 4179
                           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);
4180 4181 4182 4183 4184 4185 4186 4187 4188
                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);
   }
4189

4190
 success:
4191
    if (iface->data.network.actual) {
4192
        netdef->connections--;
4193 4194
        VIR_DEBUG("Releasing network %s, %d connections",
                  netdef->name, netdef->connections);
4195

4196 4197 4198 4199
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
    }
4200
    ret = 0;
4201
 cleanup:
4202 4203
    if (network)
        virNetworkObjUnlock(network);
4204 4205 4206 4207
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4208
    return ret;
4209

4210
 error:
4211
    goto cleanup;
4212
}
4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236

/*
 * 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;
4237
    virNetworkDriverStatePtr driver = driverState;
4238
    virNetworkObjPtr network;
4239 4240 4241 4242
    virNetworkDefPtr netdef;
    virNetworkIpDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4243
    char *dev_name = NULL;
4244 4245 4246 4247 4248 4249

    *netaddr = NULL;
    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, netname);
    networkDriverUnlock(driver);
    if (!network) {
4250 4251 4252
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4253
        goto error;
4254 4255 4256
    }
    netdef = network->def;

4257
    switch (netdef->forward.type) {
4258 4259 4260 4261 4262 4263
    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) {
4264 4265 4266
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have an IPv4 address"),
                           netdef->name);
4267 4268 4269 4270 4271 4272
            break;
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4273
        if ((dev_name = netdef->bridge))
4274 4275 4276 4277 4278 4279 4280 4281
            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:
4282 4283
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4284

4285
        if (!dev_name) {
4286 4287 4288
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4289 4290 4291 4292
        }
        break;
    }

4293
    if (dev_name) {
4294
        if (virNetDevGetIPv4Address(dev_name, &addr) < 0)
4295
            goto error;
4296
        addrptr = &addr;
4297 4298
    }

4299 4300 4301
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
        goto error;
4302 4303
    }

4304
    ret = 0;
4305
 cleanup:
4306 4307 4308
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4309

4310
 error:
4311
    goto cleanup;
4312
}
4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330

/**
 * 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;
4331
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4332 4333 4334 4335
    unsigned long long tmp_floor_sum = net->floor_sum;
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346
    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;
    }

4347
    if (!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor ||
4348 4349
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4350
        return 1;
4351
    }
4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382

    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;

4383
 cleanup:
4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420
    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];
4421
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449

    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;
    }

4450 4451
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
4452 4453 4454 4455 4456 4457 4458 4459
    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 */
4460
    net->floor_sum += ifaceBand->in->floor;
4461
    /* update status file */
4462
    if (virNetworkSaveStatus(driverState->stateDir, net) < 0) {
4463
        ignore_value(virBitmapClearBit(net->class_id, class_id));
4464
        net->floor_sum -= ifaceBand->in->floor;
4465 4466 4467 4468
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
4469 4470 4471 4472 4473 4474 4475 4476 4477
    /* 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;

4478
 cleanup:
4479 4480 4481 4482 4483 4484 4485 4486 4487
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
    int ret = 0;
    unsigned long long new_rate;
4488
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4489 4490 4491

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
4492 4493 4494 4495 4496
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507
        /* 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 */
4508
        net->floor_sum -= ifaceBand->in->floor;
4509 4510 4511 4512
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
4513
        if (virNetworkSaveStatus(driverState->stateDir, net) < 0) {
4514
            net->floor_sum += ifaceBand->in->floor;
4515 4516 4517 4518
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
4519 4520 4521 4522 4523 4524 4525 4526 4527 4528
        /* 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;
    }

4529
 cleanup:
4530 4531
    return ret;
}
4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546

static void
networkNetworkObjTaint(virNetworkObjPtr net,
                       enum virNetworkTaintFlags taint)
{
    if (virNetworkObjTaint(net, taint)) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(net->def->uuid, uuidstr);

        VIR_WARN("Network name='%s' uuid=%s is tainted: %s",
                 net->def->name,
                 uuidstr,
                 virNetworkTaintTypeToString(taint));
    }
}