bridge_driver.c 155.0 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
#include "virjson.h"
77

78 79
#define VIR_FROM_THIS VIR_FROM_NETWORK

80 81 82 83 84 85 86
/**
 * VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX:
 *
 * Macro providing the upper limit on the size of leases file
 */
#define VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX (32 * 1024 * 1024)

87 88
VIR_LOG_INIT("network.bridge_driver");

89 90 91 92
static virNetworkDriverStatePtr driver;


static void networkDriverLock(void)
93
{
94
    virMutexLock(&driver->lock);
95
}
96
static void networkDriverUnlock(void)
97
{
98
    virMutexUnlock(&driver->lock);
99 100
}

101
static int networkStateCleanup(void);
102

103
static int networkStartNetwork(virNetworkObjPtr network);
104

105
static int networkShutdownNetwork(virNetworkObjPtr network);
106

107
static int networkStartNetworkVirtual(virNetworkObjPtr network);
108

109
static int networkShutdownNetworkVirtual(virNetworkObjPtr network);
110

111
static int networkStartNetworkExternal(virNetworkObjPtr network);
112

113
static int networkShutdownNetworkExternal(virNetworkObjPtr network);
114

115 116
static void networkReloadFirewallRules(void);
static void networkRefreshDaemons(void);
117

118 119 120 121 122
static int networkPlugBandwidth(virNetworkObjPtr net,
                                virDomainNetDefPtr iface);
static int networkUnplugBandwidth(virNetworkObjPtr net,
                                  virDomainNetDefPtr iface);

123
static void networkNetworkObjTaint(virNetworkObjPtr net,
124
                                   virNetworkTaintFlags taint);
125

126 127 128 129 130 131
static virNetworkObjPtr
networkObjFromNetwork(virNetworkPtr net)
{
    virNetworkObjPtr network;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

132
    networkDriverLock();
133
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
134
    networkDriverUnlock();
135 136 137 138 139 140 141 142 143 144 145

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

    return network;
}

146 147 148
static int
networkRunHook(virNetworkObjPtr network,
               virDomainDefPtr dom,
149
               virDomainNetDefPtr iface,
150 151 152 153 154 155 156 157 158
               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)) {
159 160 161 162 163 164
        if (!network) {
            VIR_DEBUG("Not running hook as @network is NULL");
            ret = 0;
            goto cleanup;
        }

165 166
        virBufferAddLit(&buf, "<hookData>\n");
        virBufferAdjustIndent(&buf, 2);
167 168
        if (iface && virDomainNetDefFormat(&buf, iface, 0) < 0)
            goto cleanup;
169 170 171 172 173 174 175 176
        if (virNetworkDefFormatBuf(&buf, network->def, 0) < 0)
            goto cleanup;
        if (dom && virDomainDefFormatInternal(dom, 0, &buf) < 0)
            goto cleanup;

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

177
        if (virBufferCheckError(&buf) < 0)
178 179
            goto cleanup;

180
        xml = virBufferContentAndReset(&buf);
181 182 183 184 185 186 187 188
        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;
189 190

        networkNetworkObjTaint(network, VIR_NETWORK_TAINT_HOOK);
191 192 193
    }

    ret = 0;
194
 cleanup:
195 196 197 198 199 200 201
    virBufferFreeAndReset(&buf);
    VIR_FREE(xml);
    VIR_FREE(net_xml);
    VIR_FREE(dom_xml);
    return ret;
}

202
static char *
203
networkDnsmasqLeaseFileNameDefault(const char *netname)
204 205 206
{
    char *leasefile;

207
    ignore_value(virAsprintf(&leasefile, "%s/%s.leases",
208
                             driver->dnsmasqStateDir, netname));
209 210 211
    return leasefile;
}

212 213 214
networkDnsmasqLeaseFileNameFunc networkDnsmasqLeaseFileName =
    networkDnsmasqLeaseFileNameDefault;

215 216 217 218 219 220
static char *
networkDnsmasqLeaseFileNameCustom(const char *bridge)
{
    char *leasefile;

    ignore_value(virAsprintf(&leasefile, "%s/%s.status",
221
                             driver->dnsmasqStateDir, bridge));
222 223 224
    return leasefile;
}

225 226 227 228 229
static char *
networkDnsmasqConfigFileName(const char *netname)
{
    char *conffile;

230
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
231
                             driver->dnsmasqStateDir, netname));
232 233 234
    return conffile;
}

235 236 237 238 239 240
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

241
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
242 243 244 245 246 247 248 249
    return pidfilebase;
}

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

250
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
251
                             driver->radvdStateDir, netname));
252 253
    return configfile;
}
254

255 256
/* do needed cleanup steps and remove the network from the list */
static int
257
networkRemoveInactive(virNetworkObjPtr net)
258 259
{
    char *leasefile = NULL;
260
    char *customleasefile = NULL;
261
    char *radvdconfigfile = NULL;
262
    char *configfile = NULL;
263
    char *radvdpidbase = NULL;
264
    char *statusfile = NULL;
265 266 267 268 269 270
    dnsmasqContext *dctx = NULL;
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(net);

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
271
    if (!(dctx = dnsmasqContextNew(def->name,
272
                                   driver->dnsmasqStateDir))) {
273
        goto cleanup;
274
    }
275 276 277 278

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

279 280 281
    if (!(customleasefile = networkDnsmasqLeaseFileNameCustom(def->bridge)))
        goto cleanup;

282
    if (!(radvdconfigfile = networkRadvdConfigFileName(def->name)))
283
        goto cleanup;
284 285

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
286
        goto cleanup;
287

288
    if (!(configfile = networkDnsmasqConfigFileName(def->name)))
289
        goto cleanup;
290

291
    if (!(statusfile
292
          = virNetworkConfigFile(driver->stateDir, def->name)))
293
        goto cleanup;
294

295 296 297
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
298
    unlink(customleasefile);
299
    unlink(configfile);
300 301 302

    /* radvd */
    unlink(radvdconfigfile);
303
    virPidFileDelete(driver->pidDir, radvdpidbase);
304

305 306 307
    /* remove status file */
    unlink(statusfile);

308 309 310 311 312
    /* remove the network definition */
    virNetworkRemoveInactive(&driver->networks, net);

    ret = 0;

313
 cleanup:
314
    VIR_FREE(leasefile);
315
    VIR_FREE(configfile);
316
    VIR_FREE(customleasefile);
317 318
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
319
    VIR_FREE(statusfile);
320 321 322 323
    dnsmasqContextFree(dctx);
    return ret;
}

324 325 326
static char *
networkBridgeDummyNicName(const char *brname)
{
327
    static const char dummyNicSuffix[] = "-nic";
328 329
    char *nicname;

330 331 332 333 334 335 336
    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.
         */
337 338 339 340 341
        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));
342
    } else {
343
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
344
    }
345 346 347
    return nicname;
}

348 349 350
/* 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). */
351
static void
352
networkUpdateAllState(void)
353
{
354
    size_t i;
355

356
    for (i = 0; i < driver->networks.count; i++) {
357 358
        virNetworkObjPtr obj = driver->networks.objs[i];

359
        if (!obj->active)
J
Ján Tomko 已提交
360
            continue;
361

362 363
        virNetworkObjLock(obj);

364 365 366 367 368 369 370 371
        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;
372

373 374 375 376 377 378 379 380 381 382 383 384 385 386 387
        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;
388

389 390 391 392
        case VIR_NETWORK_FORWARD_HOSTDEV:
            /* so far no extra checks */
            break;
        }
393

394 395 396 397
        /* Try and read dnsmasq/radvd pids of active networks */
        if (obj->active && obj->def->ips && (obj->def->nips > 0)) {
            char *radvdpidbase;

398
            ignore_value(virPidFileReadIfAlive(driver->pidDir,
399 400 401 402 403 404
                                               obj->def->name,
                                               &obj->dnsmasqPid,
                                               dnsmasqCapsGetBinaryPath(driver->dnsmasqCaps)));
            radvdpidbase = networkRadvdPidfileBasename(obj->def->name);
            if (!radvdpidbase)
                break;
405
            ignore_value(virPidFileReadIfAlive(driver->pidDir,
406 407 408
                                               radvdpidbase,
                                               &obj->radvdPid, RADVD));
            VIR_FREE(radvdpidbase);
409 410 411 412
        }

        virNetworkObjUnlock(obj);
    }
413 414 415 416 417 418 419 420

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

        if (!obj->persistent && !obj->active) {
421
            networkRemoveInactive(obj);
422 423 424 425 426 427
            continue;
        }

        virNetworkObjUnlock(obj);
        i++;
    }
428 429 430
}


431
static void
432
networkAutostartConfigs(void)
433
{
434
    size_t i;
435

436
    for (i = 0; i < driver->networks.count; i++) {
437
        virNetworkObjLock(driver->networks.objs[i]);
438
        if (driver->networks.objs[i]->autostart &&
439
            !virNetworkObjIsActive(driver->networks.objs[i])) {
440
            if (networkStartNetwork(driver->networks.objs[i]) < 0) {
J
Ján Tomko 已提交
441
                /* failed to start but already logged */
442
            }
443
        }
444
        virNetworkObjUnlock(driver->networks.objs[i]);
445 446 447
    }
}

448 449 450
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
451
                             DBusMessage *message, void *user_data ATTRIBUTE_UNUSED)
452
{
453 454 455 456 457 458
    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.");
459
        networkReloadFirewallRules();
460 461 462 463 464 465
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

466
static int
467
networkMigrateStateFiles(void)
468 469 470 471 472 473 474 475 476 477 478 479 480
{
    /* 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;
481
    int direrr;
482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500
    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;
    }

501
    while ((direrr = virDirRead(dir, &entry, oldStateDir)) > 0) {
502 503 504
        if (entry->d_type != DT_UNKNOWN &&
            entry->d_type != DT_REG)
            continue;
505

506
        if (STREQ(entry->d_name, ".") ||
507 508 509 510 511
            STREQ(entry->d_name, ".."))
            continue;

        if (virAsprintf(&oldPath, "%s/%s",
                        oldStateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
512
            goto cleanup;
513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529

        if (entry->d_type == DT_UNKNOWN) {
            struct stat st;

            if (lstat(oldPath, &st) < 0) {
                virReportSystemError(errno,
                                     _("failed to stat network status file '%s'"),
                                     oldPath);
                goto cleanup;
            }

            if (!S_ISREG(st.st_mode)) {
                VIR_FREE(oldPath);
                continue;
            }
        }

530
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
J
Ján Tomko 已提交
531
            goto cleanup;
532 533 534

        if (virAsprintf(&newPath, "%s/%s",
                        driver->stateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
535
            goto cleanup;
536 537 538 539 540 541 542 543 544 545 546 547
        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);
    }
548
    if (direrr < 0)
J
Ján Tomko 已提交
549
        goto cleanup;
550 551 552 553 554 555 556 557 558 559

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

560
/**
561
 * networkStateInitialize:
562 563 564 565
 *
 * Initialization function for the QEmu daemon
 */
static int
566 567 568
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
569
{
570 571 572
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
573 574 575
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
576

577
    if (VIR_ALLOC(driver) < 0)
578
        goto error;
579

580 581
    if (virMutexInit(&driver->lock) < 0) {
        VIR_FREE(driver);
582 583
        goto error;
    }
584
    networkDriverLock();
585

586 587 588 589
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
590
    if (privileged) {
591
        if (VIR_STRDUP(driver->networkConfigDir,
592
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
593
            VIR_STRDUP(driver->networkAutostartDir,
594
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
595
            VIR_STRDUP(driver->stateDir,
596
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
597
            VIR_STRDUP(driver->pidDir,
598
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
599
            VIR_STRDUP(driver->dnsmasqStateDir,
600
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
601
            VIR_STRDUP(driver->radvdStateDir,
602 603
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
604 605 606 607 608

        /* migration from old to new location is only applicable for
         * privileged mode - unprivileged mode directories haven't
         * changed location.
         */
609
        if (networkMigrateStateFiles() < 0)
610
            goto error;
611
    } else {
612 613 614
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
615
            goto error;
616

617
        if ((virAsprintf(&driver->networkConfigDir,
618
                         "%s/qemu/networks", configdir) < 0) ||
619
            (virAsprintf(&driver->networkAutostartDir,
620
                         "%s/qemu/networks/autostart", configdir) < 0) ||
621
            (virAsprintf(&driver->stateDir,
622
                         "%s/network/lib", rundir) < 0) ||
623
            (virAsprintf(&driver->pidDir,
624
                         "%s/network/run", rundir) < 0) ||
625
            (virAsprintf(&driver->dnsmasqStateDir,
626
                         "%s/dnsmasq/lib", rundir) < 0) ||
627
            (virAsprintf(&driver->radvdStateDir,
628
                         "%s/radvd/lib", rundir) < 0)) {
629
            goto error;
630
        }
631 632
    }

633
    if (virFileMakePath(driver->stateDir) < 0) {
634 635
        virReportSystemError(errno,
                             _("cannot create directory %s"),
636
                             driver->stateDir);
637 638 639
        goto error;
    }

640
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
641
    driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
642

643 644
    if (virNetworkLoadAllState(&driver->networks,
                               driver->stateDir) < 0)
645 646
        goto error;

647 648 649
    if (virNetworkLoadAllConfigs(&driver->networks,
                                 driver->networkConfigDir,
                                 driver->networkAutostartDir) < 0)
650 651
        goto error;

652 653 654
    networkUpdateAllState();
    networkReloadFirewallRules();
    networkRefreshDaemons();
655

656
    driver->networkEventState = virObjectEventStateNew();
657

658
    networkDriverUnlock();
659

660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681
#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,
682
                                   NULL, NULL);
683 684 685
    }
#endif

686
    ret = 0;
687
 cleanup:
688 689 690
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
691

692
 error:
693 694
    if (driver)
        networkDriverUnlock();
695
    networkStateCleanup();
696
    goto cleanup;
697 698
}

699 700 701 702 703 704 705 706
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
707
    if (!driver)
708 709
        return;

710 711 712
    networkDriverLock();
    networkAutostartConfigs();
    networkDriverUnlock();
713 714
}

715
/**
716
 * networkStateReload:
717 718 719 720 721
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
722 723
networkStateReload(void)
{
724
    if (!driver)
725 726
        return 0;

727 728 729 730 731 732 733 734 735 736
    networkDriverLock();
    virNetworkLoadAllState(&driver->networks,
                           driver->stateDir);
    virNetworkLoadAllConfigs(&driver->networks,
                             driver->networkConfigDir,
                             driver->networkAutostartDir);
    networkReloadFirewallRules();
    networkRefreshDaemons();
    networkAutostartConfigs();
    networkDriverUnlock();
737 738 739 740 741
    return 0;
}


/**
742
 * networkStateCleanup:
743 744 745 746
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
747 748
networkStateCleanup(void)
{
749
    if (!driver)
750 751
        return -1;

752
    networkDriverLock();
753

754
    virObjectEventStateFree(driver->networkEventState);
755

756
    /* free inactive networks */
757
    virNetworkObjListFree(&driver->networks);
758

759 760 761 762 763 764
    VIR_FREE(driver->networkConfigDir);
    VIR_FREE(driver->networkAutostartDir);
    VIR_FREE(driver->stateDir);
    VIR_FREE(driver->pidDir);
    VIR_FREE(driver->dnsmasqStateDir);
    VIR_FREE(driver->radvdStateDir);
765

766
    virObjectUnref(driver->dnsmasqCaps);
767

768 769
    networkDriverUnlock();
    virMutexDestroy(&driver->lock);
770

771
    VIR_FREE(driver);
772 773 774 775 776

    return 0;
}


777 778 779 780 781 782 783
/* 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)
{
784 785
    size_t i;
    int ret = -1;
786 787 788 789 790 791 792
    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.
     */
793
    for (i = 0; i < 25; i++) {
794
        int signum = 0;
795
        if (i == 0) {
796
            signum = SIGTERM;
797
        } else if (i == 15) {
798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832
            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);
833
 cleanup:
834 835 836
    return ret;
}

J
Ján Tomko 已提交
837 838 839
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
G
Gene Czarcinski 已提交
840

841
static int
G
Gene Czarcinski 已提交
842 843
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
                                 virNetworkIpDefPtr ipdef)
844
{
845
    size_t i;
G
Gene Czarcinski 已提交
846
    bool ipv6 = false;
847

G
Gene Czarcinski 已提交
848 849
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
850 851
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
852
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
853 854
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
855
                return -1;
856
    }
857

G
Gene Czarcinski 已提交
858 859 860 861 862 863 864
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
865
    size_t i, j;
G
Gene Czarcinski 已提交
866

867 868
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
869
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
870
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
871
                for (j = 0; j < host->nnames; j++)
872 873
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
874 875
            }
        }
876 877
    }

878
    return 0;
879 880 881
}


882 883
int
networkDnsmasqConfContents(virNetworkObjPtr network,
884 885 886 887
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
888
{
889
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
890
    int r, ret = -1;
891
    int nbleases = 0;
892
    size_t i;
893
    virNetworkDNSDefPtr dns = &network->def->dns;
G
Gene Czarcinski 已提交
894 895
    virNetworkIpDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
    bool ipv6SLAAC;
896

897 898
    *configstr = NULL;

899
    /*
900 901 902
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
903
     *
904 905
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
906
     */
907 908 909 910 911 912

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

913 914
    /* create dnsmasq config file appropriate for this network */
    virBufferAsprintf(&configbuf,
915 916 917 918 919 920 921
                      "##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"
922
                      "strict-order\n",
923 924
                      network->def->name);

925 926 927 928
    if (network->def->dns.forwarders) {
        virBufferAddLit(&configbuf, "no-resolv\n");
        for (i = 0; i < network->def->dns.nfwds; i++) {
            virBufferAsprintf(&configbuf, "server=%s\n",
J
Ján Tomko 已提交
929
                              network->def->dns.forwarders[i]);
930 931 932
        }
    }

933
    if (network->def->domain) {
934
        virBufferAsprintf(&configbuf,
935 936 937 938
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
939

J
Ján Tomko 已提交
940
    if (network->def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
941 942 943 944
        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)
945
         */
946
        virBufferAddLit(&configbuf, "local=//\n");
947
    }
948

949
    if (pidfile)
950
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
951

952 953 954
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

955 956 957 958 959 960 961 962
    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.
         */
963
        virBufferAsprintf(&configbuf,
964 965 966
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
967
    } else {
968
        virBufferAddLit(&configbuf, "bind-interfaces\n");
969 970 971 972 973 974 975 976
        /*
         * --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
         */
977 978 979
        for (i = 0;
             (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
             i++) {
980 981 982 983
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
984

985
            /* also part of CVE 2012-3411 - if the host's version of
986
             * dnsmasq doesn't have bind-dynamic, only allow listening on
987 988
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
989 990
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
991 992 993 994
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
995
                                 "The version of dnsmasq on this host (%d.%d) "
996 997 998 999
                                 "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 "
1000 1001 1002 1003 1004 1005
                                 "(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);
1006
                VIR_FREE(ipaddr);
1007 1008
                goto cleanup;
            }
1009
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1010 1011 1012
            VIR_FREE(ipaddr);
        }
    }
1013

1014 1015
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1016
     * guaranteed to not work, and set no-resolv so that no dns
1017 1018 1019
     * 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).
1020
     */
1021
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1022
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1023
                        "no-resolv\n");
1024
    }
1025

1026
    for (i = 0; i < dns->ntxts; i++) {
1027
        virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
1028 1029
                          dns->txts[i].name,
                          dns->txts[i].value);
1030
    }
1031

1032
    for (i = 0; i < dns->nsrvs; i++) {
1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
        /* 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);
1055

1056 1057 1058
        /* domain is optional - it defaults to the domain of this network */
        if (dns->srvs[i].domain)
            virBufferAsprintf(&configbuf, ".%s", dns->srvs[i].domain);
1059

1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081
        /* 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);
1082
        }
1083
        virBufferAddLit(&configbuf, "\n");
1084 1085
    }

G
Gene Czarcinski 已提交
1086
    /* Find the first dhcp for both IPv4 and IPv6 */
1087 1088 1089
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
1090 1091 1092 1093
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1094 1095
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106
                    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,
1107 1108 1109 1110 1111 1112 1113 1114 1115
                                   _("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
1116 1117 1118 1119
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1120 1121
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134
                    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 "
1135 1136 1137 1138
                 "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
1139 1140 1141 1142 1143
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1144
        for (r = 0; r < ipdef->nranges; r++) {
1145
            char *saddr = virSocketAddrFormat(&ipdef->ranges[r].start);
1146 1147
            if (!saddr)
                goto cleanup;
1148
            char *eaddr = virSocketAddrFormat(&ipdef->ranges[r].end);
1149 1150 1151 1152
            if (!eaddr) {
                VIR_FREE(saddr);
                goto cleanup;
            }
1153
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s\n",
1154
                              saddr, eaddr);
1155
            VIR_FREE(saddr);
1156
            VIR_FREE(eaddr);
1157 1158
            nbleases += virSocketAddrGetRange(&ipdef->ranges[r].start,
                                              &ipdef->ranges[r].end);
1159
        }
1160

1161
        /*
1162 1163 1164 1165
         * 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)
1166 1167
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1168
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1169 1170
            if (!bridgeaddr)
                goto cleanup;
1171
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static\n", bridgeaddr);
1172 1173
            VIR_FREE(bridgeaddr);
        }
1174

G
Gene Czarcinski 已提交
1175 1176
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1177

G
Gene Czarcinski 已提交
1178 1179 1180
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts)
1181
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1182

G
Gene Czarcinski 已提交
1183
            if (ipdef->tftproot) {
1184 1185
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1186
            }
1187

G
Gene Czarcinski 已提交
1188 1189 1190
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1191

1192
                    if (!bootserver)
G
Gene Czarcinski 已提交
1193
                        goto cleanup;
1194
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1195
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1196 1197
                    VIR_FREE(bootserver);
                } else {
1198
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1199 1200 1201 1202 1203
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1204

G
Gene Czarcinski 已提交
1205 1206
    if (nbleases > 0) {
        char *leasefile = networkDnsmasqLeaseFileName(network->def->name);
1207
        if (!leasefile)
G
Gene Czarcinski 已提交
1208
            goto cleanup;
1209
        virBufferAsprintf(&configbuf, "dhcp-leasefile=%s\n", leasefile);
G
Gene Czarcinski 已提交
1210
        VIR_FREE(leasefile);
1211
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
G
Gene Czarcinski 已提交
1212
    }
1213

G
Gene Czarcinski 已提交
1214 1215
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1216
        goto cleanup;
G
Gene Czarcinski 已提交
1217 1218 1219 1220 1221 1222

    /* 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)
1223 1224
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1225

1226 1227
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1228
     */
1229
    virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
1230
                      dctx->addnhostsfile->path);
G
Gene Czarcinski 已提交
1231 1232 1233

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1234
        if (ipv6def) {
1235
            virBufferAddLit(&configbuf, "enable-ra\n");
1236
        } else {
1237 1238 1239
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
                 i++) {
G
Gene Czarcinski 已提交
1240 1241 1242 1243
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1244 1245
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1246 1247
                    VIR_FREE(bridgeaddr);
                }
1248
            }
1249
        }
1250 1251
    }

1252 1253 1254
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1255
    ret = 0;
G
Gene Czarcinski 已提交
1256

1257
 cleanup:
1258
    virBufferFreeAndReset(&configbuf);
1259
    return ret;
1260 1261
}

1262
/* build the dnsmasq command line */
1263 1264 1265
static int ATTRIBUTE_NONNULL(2)
networkBuildDhcpDaemonCommandLine(virNetworkObjPtr network,
                                  virCommandPtr *cmdout,
1266 1267
                                  char *pidfile, dnsmasqContext *dctx,
                                  dnsmasqCapsPtr caps)
1268
{
1269
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1270
    int ret = -1;
1271 1272
    char *configfile = NULL;
    char *configstr = NULL;
1273
    char *leaseshelper_path = NULL;
1274 1275

    network->dnsmasqPid = -1;
1276

1277 1278 1279 1280 1281 1282
    if (networkDnsmasqConfContents(network, pidfile, &configstr, dctx, caps) < 0)
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1283
    if (!(configfile = networkDnsmasqConfigFileName(network->def->name)))
1284 1285 1286 1287 1288
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1289 1290
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1291 1292 1293
        goto cleanup;
    }

1294 1295 1296 1297 1298 1299
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
                                                  "src",
                                                  LIBEXECDIR)))
        goto cleanup;

1300 1301
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(caps));
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1302 1303
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1304
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1305
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", network->def->bridge);
1306

1307
    *cmdout = cmd;
1308
    ret = 0;
1309
 cleanup:
1310 1311
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1312
    VIR_FREE(leaseshelper_path);
1313 1314 1315 1316
    return ret;
}

static int
1317
networkStartDhcpDaemon(virNetworkObjPtr network)
1318 1319 1320 1321
{
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1322
    dnsmasqContext *dctx = NULL;
1323

1324
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0)) {
G
Gene Czarcinski 已提交
1325
        /* no IP addresses, so we don't need to run */
1326 1327 1328 1329
        ret = 0;
        goto cleanup;
    }

1330
    if (virFileMakePath(driver->pidDir) < 0) {
1331
        virReportSystemError(errno,
1332
                             _("cannot create directory %s"),
1333
                             driver->pidDir);
1334
        goto cleanup;
1335 1336
    }

1337
    if (!(pidfile = virPidFileBuildPath(driver->pidDir,
1338
                                        network->def->name)))
1339
        goto cleanup;
1340

1341
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1342
        virReportSystemError(errno,
1343
                             _("cannot create directory %s"),
1344
                             driver->dnsmasqStateDir);
1345 1346 1347
        goto cleanup;
    }

1348
    dctx = dnsmasqContextNew(network->def->name, driver->dnsmasqStateDir);
1349 1350 1351
    if (dctx == NULL)
        goto cleanup;

1352 1353
    if (dnsmasqCapsRefresh(&driver->dnsmasqCaps, NULL) < 0)
        goto cleanup;
1354 1355 1356

    ret = networkBuildDhcpDaemonCommandLine(network, &cmd, pidfile,
                                            dctx, driver->dnsmasqCaps);
1357 1358 1359 1360 1361
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1362
        goto cleanup;
1363

G
Guido Günther 已提交
1364
    ret = virCommandRun(cmd, NULL);
1365
    if (ret < 0)
1366 1367 1368
        goto cleanup;

    /*
1369 1370 1371 1372 1373
     * 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
1374 1375
     */

1376
    ret = virPidFileRead(driver->pidDir, network->def->name,
1377 1378
                         &network->dnsmasqPid);
    if (ret < 0)
1379
        goto cleanup;
1380

1381
    ret = 0;
1382
 cleanup:
1383
    VIR_FREE(pidfile);
1384
    virCommandFree(cmd);
1385
    dnsmasqContextFree(dctx);
1386 1387 1388
    return ret;
}

1389 1390
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1391 1392
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1393 1394 1395
 *
 *  Returns 0 on success, -1 on failure.
 */
1396
static int
1397
networkRefreshDhcpDaemon(virNetworkObjPtr network)
1398
{
1399 1400
    int ret = -1;
    size_t i;
G
Gene Czarcinski 已提交
1401
    virNetworkIpDefPtr ipdef, ipv4def, ipv6def;
1402
    dnsmasqContext *dctx = NULL;
1403

G
Gene Czarcinski 已提交
1404
    /* if no IP addresses specified, nothing to do */
1405
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1406 1407
        return 0;

1408 1409
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1410
        return networkStartDhcpDaemon(network);
1411

G
Gene Czarcinski 已提交
1412
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1413
    if (!(dctx = dnsmasqContextNew(network->def->name,
1414
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1415
        goto cleanup;
1416
    }
G
Gene Czarcinski 已提交
1417 1418 1419 1420 1421 1422

    /* 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;
1423 1424 1425
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
G
Gene Czarcinski 已提交
1426 1427
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1428 1429
    }

G
Gene Czarcinski 已提交
1430
    ipv6def = NULL;
1431 1432 1433
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1434 1435
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1436 1437
    }

G
Gene Czarcinski 已提交
1438
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1439
        goto cleanup;
G
Gene Czarcinski 已提交
1440 1441

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1442
        goto cleanup;
1443

G
Gene Czarcinski 已提交
1444
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
J
Ján Tomko 已提交
1445
        goto cleanup;
1446 1447

    if ((ret = dnsmasqSave(dctx)) < 0)
1448
        goto cleanup;
1449 1450

    ret = kill(network->dnsmasqPid, SIGHUP);
1451
 cleanup:
1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463
    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
1464
networkRestartDhcpDaemon(virNetworkObjPtr network)
1465 1466 1467 1468 1469 1470
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1471
    }
1472
    /* now start dnsmasq if it should be started */
1473
    return networkStartDhcpDaemon(network);
1474 1475
}

G
Gene Czarcinski 已提交
1476 1477 1478 1479 1480 1481
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";

1482 1483 1484
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1485
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1486 1487
    int ret = -1;
    size_t i;
1488
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
1489
    bool v6present = false, dhcp6 = false;
1490 1491

    *configstr = NULL;
1492

G
Gene Czarcinski 已提交
1493
    /* Check if DHCPv6 is needed */
1494 1495 1496
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509
        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;
    }

1510 1511 1512
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1513
    virBufferAsprintf(&configbuf, "interface %s\n"
1514 1515
                      "{\n"
                      "  AdvSendAdvert on;\n"
1516
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1517 1518 1519 1520 1521
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1522 1523

    /* add a section for each IPv6 address in the config */
1524 1525 1526
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
1527 1528 1529 1530 1531
        int prefix;
        char *netaddr;

        prefix = virNetworkIpDefPrefix(ipdef);
        if (prefix < 0) {
1532 1533 1534
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1535 1536
            goto cleanup;
        }
1537
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1538
            goto cleanup;
1539
        virBufferAsprintf(&configbuf,
1540
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1541 1542 1543
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1544 1545 1546
        VIR_FREE(netaddr);
    }

1547
    virBufferAddLit(&configbuf, "};\n");
1548

1549
    if (virBufferCheckError(&configbuf) < 0)
1550
        goto cleanup;
1551

1552 1553
    *configstr = virBufferContentAndReset(&configbuf);

1554
    ret = 0;
1555
 cleanup:
1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577
    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;
1578 1579 1580 1581
        goto cleanup;
    }

    /* construct the filename */
1582
    if (!(*configFile = networkRadvdConfigFileName(network->def->name)))
1583 1584
        goto cleanup;
    /* write the file */
1585
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1586 1587
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1588 1589 1590 1591 1592
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1593
 cleanup:
1594 1595 1596 1597 1598 1599
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1600
networkStartRadvd(virNetworkObjPtr network)
1601 1602 1603 1604 1605 1606 1607 1608 1609
{
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1610
    /* Is dnsmasq handling RA? */
J
Ján Tomko 已提交
1611
    if (DNSMASQ_RA_SUPPORT(driver->dnsmasqCaps)) {
G
Gene Czarcinski 已提交
1612 1613 1614 1615
        ret = 0;
        goto cleanup;
    }

1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626
    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);
1627 1628 1629
        goto cleanup;
    }

1630
    if (virFileMakePath(driver->pidDir) < 0) {
1631 1632
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1633
                             driver->pidDir);
1634 1635
        goto cleanup;
    }
1636
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1637 1638
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1639
                             driver->radvdStateDir);
1640 1641 1642 1643
        goto cleanup;
    }

    /* construct pidfile name */
1644
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1645
        goto cleanup;
1646
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1647 1648 1649 1650 1651
        goto cleanup;

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

1652 1653 1654 1655
    /* 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
1656
     * virPidFileRead() below will fail if we use them).
1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671
     * 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;

1672
    if (virPidFileRead(driver->pidDir, radvdpidbase, &network->radvdPid) < 0)
1673 1674 1675
        goto cleanup;

    ret = 0;
1676
 cleanup:
1677 1678 1679 1680 1681 1682 1683
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1684
static int
1685
networkRefreshRadvd(virNetworkObjPtr network)
1686
{
G
Gene Czarcinski 已提交
1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697
    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)) {
1698
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1699 1700 1701 1702 1703 1704
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }

1705 1706
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
1707
        return networkStartRadvd(network);
1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719

    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);
}

1720 1721
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1722
static int
1723
networkRestartRadvd(virNetworkObjPtr network)
1724 1725 1726 1727 1728 1729 1730 1731 1732
{
    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
1733
        if ((networkKillDaemon(network->radvdPid, "radvd",
1734 1735 1736
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1737
            virPidFileDelete(driver->pidDir, radvdpidbase);
1738 1739 1740 1741 1742 1743 1744 1745 1746
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1747 1748 1749 1750
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
1751
networkRefreshDaemons(void)
1752
{
1753
    size_t i;
1754 1755 1756

    VIR_INFO("Refreshing network daemons");

1757
    for (i = 0; i < driver->networks.count; i++) {
1758 1759 1760 1761
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
1762 1763 1764
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
1765 1766 1767 1768 1769 1770
            /* 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.
             */
1771 1772
            networkRefreshDhcpDaemon(network);
            networkRefreshRadvd(network);
1773 1774 1775 1776 1777
        }
        virNetworkObjUnlock(network);
    }
}

1778
static void
1779
networkReloadFirewallRules(void)
1780
{
1781
    size_t i;
1782

1783
    VIR_INFO("Reloading iptables rules");
1784

1785
    for (i = 0; i < driver->networks.count; i++) {
1786 1787 1788 1789
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
1790 1791 1792
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
1793 1794 1795
            /* Only the three L3 network types that are configured by libvirt
             * need to have iptables rules reloaded.
             */
1796 1797
            networkRemoveFirewallRules(network->def);
            if (networkAddFirewallRules(network->def) < 0) {
1798 1799
                /* failed to add but already logged */
            }
1800
        }
1801
        virNetworkObjUnlock(network);
1802 1803 1804
    }
}

1805
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
1806
static int
1807
networkEnableIpForwarding(bool enableIPv4, bool enableIPv6)
1808
{
1809
    int ret = 0;
1810 1811 1812 1813
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
1814
                           &enabled, sizeof(enabled));
1815 1816
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
1817
                           &enabled, sizeof(enabled));
1818
#else
1819 1820 1821 1822
    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);
1823
#endif
1824
    return ret;
1825 1826
}

1827 1828
#define SYSCTL_PATH "/proc/sys"

1829 1830
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
1831 1832 1833
{
    char *field = NULL;
    int ret = -1;
1834
    bool enableIPv6 =  !!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0);
1835

1836 1837 1838 1839 1840 1841 1842
    /* set disable_ipv6 if there are no ipv6 addresses defined for the
     * network. But also unset it if there *are* ipv6 addresses, as we
     * can't be sure of its default value.
     */
    if (virAsprintf(&field, SYSCTL_PATH "/net/ipv6/conf/%s/disable_ipv6",
                    network->def->bridge) < 0)
       goto cleanup;
1843

1844 1845
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
1846 1847
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
1848 1849 1850
        ret = 0;
        goto cleanup;
    }
1851

1852 1853 1854 1855 1856
    if (virFileWriteStr(field, enableIPv6 ? "0" : "1", 0) < 0) {
        virReportSystemError(errno,
                             _("cannot write to %s to enable/disable IPv6 "
                               "on bridge %s"), field, network->def->bridge);
        goto cleanup;
1857
    }
1858
    VIR_FREE(field);
1859

1860 1861
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
1862 1863 1864 1865 1866 1867
     */

    /* 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",
1868
                    network->def->bridge) < 0)
1869 1870
        goto cleanup;

1871
    if (virFileWriteStr(field, "0", 0) < 0) {
1872
        virReportSystemError(errno,
1873 1874 1875 1876 1877
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

1878 1879 1880 1881
    /* 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",
1882
                    network->def->bridge) < 0)
1883 1884
        goto cleanup;

1885
    if (virFileWriteStr(field, "0", 0) < 0) {
1886
        virReportSystemError(errno,
1887
                             _("cannot disable %s"), field);
1888 1889 1890 1891
        goto cleanup;
    }

    ret = 0;
1892
 cleanup:
1893 1894 1895 1896
    VIR_FREE(field);
    return ret;
}

1897
/* add an IP address to a bridge */
1898
static int
D
Daniel P. Berrange 已提交
1899
networkAddAddrToBridge(virNetworkObjPtr network,
1900
                       virNetworkIpDefPtr ipdef)
1901
{
1902 1903 1904
    int prefix = virNetworkIpDefPrefix(ipdef);

    if (prefix < 0) {
1905 1906 1907
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
1908 1909 1910
        return -1;
    }

1911 1912
    if (virNetDevSetIPv4Address(network->def->bridge,
                                &ipdef->address, prefix) < 0)
1913 1914 1915 1916 1917
        return -1;

    return 0;
}

1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930
/* 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,
1931
                                    (VIR_SOCKET_ADDR_IS_FAMILY(addr, AF_INET)
1932 1933 1934 1935 1936 1937 1938
                                     ? "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) &&
J
Ján Tomko 已提交
1939
                  virSocketAddrEqual(mask, &zero)))
1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966
            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;
}

1967
static int
1968
networkStartNetworkVirtual(virNetworkObjPtr network)
1969
{
1970
    size_t i;
1971
    bool v4present = false, v6present = false;
1972 1973
    virErrorPtr save_err = NULL;
    virNetworkIpDefPtr ipdef;
1974
    virNetworkRouteDefPtr routedef;
1975
    char *macTapIfName = NULL;
1976
    int tapfd = -1;
1977

1978
    /* Check to see if any network IP collides with an existing route */
1979
    if (networkCheckRouteCollision(network->def) < 0)
1980 1981
        return -1;

1982
    /* Create and configure the bridge device */
1983
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
1984 1985
        return -1;

1986 1987 1988 1989 1990 1991 1992 1993
    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);
1994
        if (!macTapIfName)
1995
            goto err0;
1996
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
1997
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
1998
                                           &macTapIfName, &network->def->mac,
1999
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2000 2001 2002
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2003 2004 2005 2006 2007
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2008
    /* Set bridge options */
2009 2010 2011 2012

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2013
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2014
                                   network->def->delay * 1000) < 0)
2015
        goto err1;
2016

2017
    if (virNetDevBridgeSetSTP(network->def->bridge,
2018
                              network->def->stp ? true : false) < 0)
2019
        goto err1;
2020

2021 2022 2023 2024
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2025
        goto err1;
2026

2027
    /* Add "once per network" rules */
2028
    if (networkAddFirewallRules(network->def) < 0)
2029 2030
        goto err1;

2031 2032 2033
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
2034
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2035
            v4present = true;
2036
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2037
            v6present = true;
2038

2039
        /* Add the IP address/netmask to the bridge */
2040
        if (networkAddAddrToBridge(network, ipdef) < 0)
2041
            goto err2;
2042 2043
    }

2044
    /* Bring up the bridge interface */
2045
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2046
        goto err2;
2047

2048 2049
    for (i = 0; i < network->def->nroutes; i++) {
        routedef = &network->def->routes[i];
2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060
        /* 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 */
            }
        }
    }

2061 2062
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2063
        networkEnableIpForwarding(v4present, v6present) < 0) {
2064
        virReportSystemError(errno, "%s",
2065
                             _("failed to enable IP forwarding"));
2066
        goto err3;
2067 2068
    }

2069

2070
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2071
    if ((v4present || v6present) &&
2072
        networkStartDhcpDaemon(network) < 0)
2073
        goto err3;
2074

2075
    /* start radvd if there are any ipv6 addresses */
2076
    if (v6present && networkStartRadvd(network) < 0)
2077 2078
        goto err4;

2079 2080 2081 2082 2083 2084 2085 2086 2087
    /* 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);
    }

2088
    if (virNetDevBandwidthSet(network->def->bridge,
2089
                              network->def->bandwidth, true) < 0)
2090 2091
        goto err5;

2092
    VIR_FREE(macTapIfName);
2093 2094 2095

    return 0;

2096
 err5:
2097
    virNetDevBandwidthClear(network->def->bridge);
2098

2099 2100 2101 2102
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2103 2104 2105 2106 2107
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2108 2109 2110
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2111
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2112

2113 2114 2115
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2116
    networkRemoveFirewallRules(network->def);
2117 2118

 err1:
2119 2120 2121
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2122
    if (macTapIfName) {
2123
        VIR_FORCE_CLOSE(tapfd);
2124
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2125 2126
        VIR_FREE(macTapIfName);
    }
2127 2128

 err0:
2129 2130
    if (!save_err)
        save_err = virSaveLastError();
2131
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2132

2133 2134 2135 2136
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2137
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2138 2139 2140
    return -1;
}

2141
static int networkShutdownNetworkVirtual(virNetworkObjPtr network)
2142
{
2143
    virNetDevBandwidthClear(network->def->bridge);
2144

2145 2146 2147 2148 2149
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2150
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2151
            virPidFileDelete(driver->pidDir, radvdpidbase);
2152 2153 2154 2155
            VIR_FREE(radvdpidbase);
        }
    }

2156 2157 2158
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2159
    if (network->def->mac_specified) {
2160
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2161
        if (macTapIfName) {
2162
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2163 2164 2165 2166
            VIR_FREE(macTapIfName);
        }
    }

2167
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2168

2169
    networkRemoveFirewallRules(network->def);
2170

2171
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2172

2173
    /* See if its still alive and really really kill it */
2174
    if (network->dnsmasqPid > 0 &&
2175
        (kill(network->dnsmasqPid, 0) == 0))
2176 2177
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2178 2179 2180 2181 2182 2183

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

2184 2185 2186
    return 0;
}

2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202

/* networkCreateInterfacePool:
 * @netdef: the original NetDef from the network
 *
 * Creates an implicit interface pool of VF's when a PF dev is given
 */
static int
networkCreateInterfacePool(virNetworkDefPtr netdef)
{
    size_t numVirtFns = 0;
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2203 2204 2205
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev,
                                      &vfNames, &virtFns, &numVirtFns)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Could not get Virtual functions on %s"),
                       netdef->forward.pfs->dev);
        goto cleanup;
    }

    if (VIR_ALLOC_N(netdef->forward.ifs, numVirtFns) < 0)
        goto cleanup;

    for (i = 0; i < numVirtFns; i++) {
        virPCIDeviceAddressPtr thisVirtFn = virtFns[i];
        const char *thisName = vfNames[i];
        virNetworkForwardIfDefPtr thisIf
            = &netdef->forward.ifs[netdef->forward.nifs];

        switch (netdef->forward.type) {
        case VIR_NETWORK_FORWARD_BRIDGE:
        case VIR_NETWORK_FORWARD_PRIVATE:
        case VIR_NETWORK_FORWARD_VEPA:
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
            if (thisName) {
                if (VIR_STRDUP(thisIf->device.dev, thisName) < 0)
                    goto cleanup;
                thisIf->type = VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV;
                netdef->forward.nifs++;
            } else {
                VIR_WARN("VF %zu of SRIOV PF %s couldn't be added to the "
                         "interface pool because it isn't bound "
                         "to a network driver - possibly in use elsewhere",
                         i, netdef->forward.pfs->dev);
            }
            break;

        case VIR_NETWORK_FORWARD_HOSTDEV:
            /* VF's are always PCI devices */
            thisIf->type = VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI;
            thisIf->device.pci.domain = thisVirtFn->domain;
            thisIf->device.pci.bus = thisVirtFn->bus;
            thisIf->device.pci.slot = thisVirtFn->slot;
            thisIf->device.pci.function = thisVirtFn->function;
            netdef->forward.nifs++;
            break;

        case VIR_NETWORK_FORWARD_NONE:
        case VIR_NETWORK_FORWARD_NAT:
        case VIR_NETWORK_FORWARD_ROUTE:
        case VIR_NETWORK_FORWARD_LAST:
            /* by definition these will never be encountered here */
            break;
        }
    }

    if (netdef->forward.nifs == 0) {
        /* If we don't get at least one interface in the pool, declare
         * failure
         */
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("No usable Vf's present on SRIOV PF %s"),
                       netdef->forward.pfs->dev);
        goto cleanup;
    }

    ret = 0;
 cleanup:
    if (ret < 0) {
        /* free all the entries made before error */
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV)
                VIR_FREE(netdef->forward.ifs[i].device.dev);
        }
        netdef->forward.nifs = 0;
    }
    if (netdef->forward.nifs == 0)
        VIR_FREE(netdef->forward.ifs);

    for (i = 0; i < numVirtFns; i++) {
        VIR_FREE(vfNames[i]);
        VIR_FREE(virtFns[i]);
    }
    VIR_FREE(vfNames);
    VIR_FREE(virtFns);
    return ret;
}


2294
static int
2295
networkStartNetworkExternal(virNetworkObjPtr network)
2296 2297
{
    /* put anything here that needs to be done each time a network of
2298
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2299 2300 2301
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2302
    return networkCreateInterfacePool(network->def);
2303 2304
}

2305
static int networkShutdownNetworkExternal(virNetworkObjPtr network ATTRIBUTE_UNUSED)
2306 2307
{
    /* put anything here that needs to be done each time a network of
2308
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2309 2310 2311 2312 2313 2314 2315
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2316
networkStartNetwork(virNetworkObjPtr network)
2317
{
2318 2319 2320
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2321 2322

    if (virNetworkObjIsActive(network)) {
2323 2324
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2325
        return ret;
2326 2327
    }

2328 2329 2330
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2331
    if (virNetworkObjSetDefTransient(network, true) < 0)
2332
        goto cleanup;
2333

2334 2335
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2336
    if (networkRunHook(network, NULL, NULL,
2337 2338 2339 2340
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2341
    switch (network->def->forward.type) {
2342 2343 2344 2345

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2346
        if (networkStartNetworkVirtual(network) < 0)
2347
            goto cleanup;
2348 2349 2350 2351 2352 2353
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2354
    case VIR_NETWORK_FORWARD_HOSTDEV:
2355
        if (networkStartNetworkExternal(network) < 0)
2356
            goto cleanup;
2357 2358 2359
        break;
    }

2360
    /* finally we can call the 'started' hook script if any */
2361
    if (networkRunHook(network, NULL, NULL,
2362 2363 2364 2365
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2366 2367 2368
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2369
    VIR_DEBUG("Writing network status to disk");
2370
    if (virNetworkSaveStatus(driver->stateDir, network) < 0)
2371
        goto cleanup;
2372 2373

    network->active = 1;
2374 2375
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2376

2377
 cleanup:
2378
    if (ret < 0) {
2379
        virNetworkObjUnsetDefTransient(network);
2380 2381
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2382
        networkShutdownNetwork(network);
2383 2384 2385 2386 2387 2388 2389
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2390
static int networkShutdownNetwork(virNetworkObjPtr network)
2391 2392 2393 2394 2395 2396 2397 2398 2399
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

2400
    stateFile = virNetworkConfigFile(driver->stateDir,
2401
                                     network->def->name);
2402 2403 2404 2405 2406 2407
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2408
    switch (network->def->forward.type) {
2409 2410 2411 2412

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2413
        ret = networkShutdownNetworkVirtual(network);
2414 2415 2416 2417 2418 2419
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2420
    case VIR_NETWORK_FORWARD_HOSTDEV:
2421
        ret = networkShutdownNetworkExternal(network);
2422 2423 2424
        break;
    }

2425
    /* now that we know it's stopped call the hook if present */
2426
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2427 2428
                   VIR_HOOK_SUBOP_END);

2429
    network->active = 0;
2430
    virNetworkObjUnsetDefTransient(network);
2431
    return ret;
2432 2433 2434
}


2435
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2436 2437
                                         const unsigned char *uuid)
{
2438 2439
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2440

2441
    networkDriverLock();
2442
    network = virNetworkFindByUUID(&driver->networks, uuid);
2443
    networkDriverUnlock();
2444
    if (!network) {
2445 2446
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
2447
        goto cleanup;
2448 2449
    }

2450 2451 2452
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2453 2454
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2455
 cleanup:
2456 2457
    if (network)
        virNetworkObjUnlock(network);
2458
    return ret;
2459 2460
}

2461
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2462 2463
                                         const char *name)
{
2464 2465 2466
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2467
    networkDriverLock();
2468
    network = virNetworkFindByName(&driver->networks, name);
2469
    networkDriverUnlock();
2470
    if (!network) {
2471 2472
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2473
        goto cleanup;
2474 2475
    }

2476 2477 2478
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2479 2480
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2481
 cleanup:
2482 2483
    if (network)
        virNetworkObjUnlock(network);
2484
    return ret;
2485 2486
}

2487
static virDrvOpenStatus networkOpen(virConnectPtr conn ATTRIBUTE_UNUSED,
2488 2489
                                    virConnectAuthPtr auth ATTRIBUTE_UNUSED,
                                    unsigned int flags)
2490 2491 2492
{
    virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);

2493
    if (!driver)
2494 2495 2496 2497 2498
        return VIR_DRV_OPEN_DECLINED;

    return VIR_DRV_OPEN_SUCCESS;
}

2499
static int networkClose(virConnectPtr conn ATTRIBUTE_UNUSED)
2500
{
2501 2502 2503
    return 0;
}

2504 2505
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2506 2507
    int nactive = 0;
    size_t i;
2508

2509 2510 2511
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2512
    networkDriverLock();
2513
    for (i = 0; i < driver->networks.count; i++) {
2514 2515 2516 2517
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectNumOfNetworksCheckACL(conn, obj->def) &&
            virNetworkObjIsActive(obj))
2518
            nactive++;
2519
        virNetworkObjUnlock(obj);
2520
    }
2521
    networkDriverUnlock();
2522

2523 2524 2525
    return nactive;
}

2526
static int networkConnectListNetworks(virConnectPtr conn, char **const names, int nnames) {
2527 2528
    int got = 0;
    size_t i;
2529

2530 2531 2532
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2533
    networkDriverLock();
2534
    for (i = 0; i < driver->networks.count && got < nnames; i++) {
2535 2536 2537 2538 2539 2540
        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);
2541 2542 2543 2544
                goto cleanup;
            }
            got++;
        }
2545
        virNetworkObjUnlock(obj);
2546
    }
2547
    networkDriverUnlock();
2548

2549 2550 2551
    return got;

 cleanup:
2552
    networkDriverUnlock();
2553
    for (i = 0; i < got; i++)
2554 2555 2556 2557
        VIR_FREE(names[i]);
    return -1;
}

2558 2559
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2560 2561
    int ninactive = 0;
    size_t i;
2562

2563 2564 2565
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2566
    networkDriverLock();
2567
    for (i = 0; i < driver->networks.count; i++) {
2568 2569 2570 2571
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectNumOfDefinedNetworksCheckACL(conn, obj->def) &&
            !virNetworkObjIsActive(obj))
2572
            ninactive++;
2573
        virNetworkObjUnlock(obj);
2574
    }
2575
    networkDriverUnlock();
2576

2577 2578 2579
    return ninactive;
}

2580
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames) {
2581 2582
    int got = 0;
    size_t i;
2583

2584 2585 2586
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2587
    networkDriverLock();
2588
    for (i = 0; i < driver->networks.count && got < nnames; i++) {
2589 2590 2591 2592 2593 2594
        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);
2595 2596 2597 2598
                goto cleanup;
            }
            got++;
        }
2599
        virNetworkObjUnlock(obj);
2600
    }
2601
    networkDriverUnlock();
2602 2603 2604
    return got;

 cleanup:
2605
    networkDriverUnlock();
2606
    for (i = 0; i < got; i++)
2607 2608 2609 2610
        VIR_FREE(names[i]);
    return -1;
}

2611
static int
2612 2613 2614
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2615 2616 2617 2618 2619
{
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2620 2621 2622
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2623
    networkDriverLock();
2624 2625 2626
    ret = virNetworkObjListExport(conn, driver->networks, nets,
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2627
    networkDriverUnlock();
2628

2629
 cleanup:
2630 2631
    return ret;
}
2632

2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
2647
                                       net, eventID, callback,
2648 2649 2650
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2651
 cleanup:
2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663
    return ret;
}

static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
    int ret = -1;

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

2664 2665 2666 2667 2668 2669
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
2670

2671
 cleanup:
2672 2673 2674
    return ret;
}

2675 2676 2677 2678 2679
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2680 2681
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2682 2683 2684 2685

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

2686 2687
    ret = virNetworkObjIsActive(obj);

2688
 cleanup:
2689 2690 2691 2692 2693 2694 2695 2696 2697 2698
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}

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

2699 2700
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2701 2702 2703 2704

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

2705 2706
    ret = obj->persistent;

2707
 cleanup:
2708 2709 2710 2711 2712 2713
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}


2714
static int
2715
networkValidate(virNetworkDefPtr def,
2716
                bool check_active)
2717
{
2718
    size_t i;
2719 2720
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2721
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
2722
    bool ipv4def = false, ipv6def = false;
2723 2724 2725 2726 2727 2728 2729 2730

    /* 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
     */
2731 2732 2733
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2734 2735 2736 2737 2738

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

        virNetworkSetBridgeMacAddr(def);
2739 2740
    } else {
        /* They are also the only types that currently support setting
2741 2742
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
2743
         */
2744 2745 2746 2747 2748 2749 2750 2751
        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;
        }
2752 2753 2754 2755 2756
        if (virNetworkDefGetIpByIndex(def, AF_UNSPEC, 0)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2757
                           virNetworkForwardTypeToString(def->forward.type));
2758 2759
            return -1;
        }
2760
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
2761 2762 2763 2764
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2765
                           virNetworkForwardTypeToString(def->forward.type));
2766 2767 2768 2769 2770 2771 2772
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2773
                           virNetworkForwardTypeToString(def->forward.type));
2774 2775
            return -1;
        }
2776 2777 2778 2779 2780 2781 2782 2783
        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;
        }
2784 2785
    }

G
Gene Czarcinski 已提交
2786 2787 2788
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
2789 2790 2791
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
2792 2793 2794 2795 2796
        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 -- "
2797 2798
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815
                    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;
                }
2816 2817 2818
            }
        }
    }
2819 2820 2821 2822 2823 2824

    /* 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.
     */

2825
    vlanAllowed = ((def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
J
Ján Tomko 已提交
2826 2827
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
2828 2829
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) ||
                   def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV);
2830 2831

    vlanUsed = def->vlan.nTags > 0;
2832 2833
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
2834 2835 2836 2837 2838
            /* 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.
             */
2839
            if (def->portGroups[i].virtPortProfile) {
2840
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
2841
                    def->portGroups[i].virtPortProfile->virtPortType
2842 2843 2844 2845 2846 2847 2848
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
2849
        }
2850
        if (def->portGroups[i].isDefault) {
2851 2852 2853 2854 2855
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
2856
                               def->name, defaultPortGroup->name,
2857
                               def->portGroups[i].name);
2858
                return -1;
2859
            }
2860
            defaultPortGroup = &def->portGroups[i];
2861
        }
2862
    }
2863 2864 2865 2866 2867 2868 2869
    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.
         */
2870 2871 2872 2873 2874 2875 2876 2877 2878
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
    return 0;
}

2879 2880
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
2881
    virNetworkDefPtr def;
2882
    virNetworkObjPtr network = NULL;
2883
    virNetworkPtr ret = NULL;
2884
    virObjectEventPtr event = NULL;
2885

2886
    networkDriverLock();
2887

2888
    if (!(def = virNetworkDefParseString(xml)))
2889
        goto cleanup;
2890

2891 2892 2893
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

2894
    if (networkValidate(def, true) < 0)
J
Ján Tomko 已提交
2895
        goto cleanup;
2896

2897 2898 2899
    /* 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.
2900
     */
2901
    if (!(network = virNetworkAssignDef(&driver->networks, def, true)))
2902 2903
        goto cleanup;
    def = NULL;
2904

2905
    if (networkStartNetwork(network) < 0) {
2906 2907
        virNetworkRemoveInactive(&driver->networks,
                                 network);
2908
        network = NULL;
2909
        goto cleanup;
2910 2911
    }

2912 2913
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
2914 2915
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
2916

2917
    VIR_INFO("Creating network '%s'", network->def->name);
2918 2919
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2920
 cleanup:
2921
    virNetworkDefFree(def);
2922 2923
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
2924 2925
    if (network)
        virNetworkObjUnlock(network);
2926
    networkDriverUnlock();
2927
    return ret;
2928 2929
}

2930 2931
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
2932
    virNetworkDefPtr def = NULL;
2933
    bool freeDef = true;
2934
    virNetworkObjPtr network = NULL;
2935
    virNetworkPtr ret = NULL;
2936
    virObjectEventPtr event = NULL;
2937

2938
    networkDriverLock();
2939

2940
    if (!(def = virNetworkDefParseString(xml)))
2941
        goto cleanup;
2942

2943 2944 2945
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

2946
    if (networkValidate(def, false) < 0)
J
Ján Tomko 已提交
2947
        goto cleanup;
2948

2949
    if (!(network = virNetworkAssignDef(&driver->networks, def, false)))
J
Ján Tomko 已提交
2950
        goto cleanup;
2951

2952
    /* def was assigned to network object */
2953
    freeDef = false;
2954 2955

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
2956 2957 2958 2959 2960
        if (!virNetworkObjIsActive(network)) {
            virNetworkRemoveInactive(&driver->networks, network);
            network = NULL;
            goto cleanup;
        }
2961 2962 2963 2964 2965
        /* 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);
2966 2967 2968
        goto cleanup;
    }

2969
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
2970 2971
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
2972

2973 2974
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
2975

2976
 cleanup:
2977 2978
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
2979
    if (freeDef)
J
Ján Tomko 已提交
2980
        virNetworkDefFree(def);
2981 2982
    if (network)
        virNetworkObjUnlock(network);
2983
    networkDriverUnlock();
2984
    return ret;
2985 2986
}

2987
static int
2988 2989
networkUndefine(virNetworkPtr net)
{
2990
    virNetworkObjPtr network;
2991
    int ret = -1;
2992
    bool active = false;
2993
    virObjectEventPtr event = NULL;
2994

2995
    networkDriverLock();
2996

2997
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
2998
    if (!network) {
2999 3000
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3001
        goto cleanup;
3002 3003
    }

3004 3005 3006
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3007 3008
    if (virNetworkObjIsActive(network))
        active = true;
3009

3010
    /* remove autostart link */
3011
    if (virNetworkDeleteConfig(driver->networkConfigDir,
3012 3013
                               driver->networkAutostartDir,
                               network) < 0)
3014
        goto cleanup;
3015
    network->autostart = 0;
3016

3017 3018
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3019 3020
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3021

3022
    VIR_INFO("Undefining network '%s'", network->def->name);
3023
    if (!active) {
3024
        if (networkRemoveInactive(network) < 0) {
3025 3026 3027
            network = NULL;
            goto cleanup;
        }
3028
        network = NULL;
3029 3030 3031 3032 3033 3034
    } 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);
3035 3036
    }

3037
    ret = 0;
3038

3039
 cleanup:
3040 3041
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3042 3043
    if (network)
        virNetworkObjUnlock(network);
3044
    networkDriverUnlock();
3045
    return ret;
3046 3047
}

3048 3049 3050 3051 3052 3053 3054 3055 3056
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
    virNetworkObjPtr network = NULL;
3057 3058
    int isActive, ret = -1;
    size_t i;
3059 3060
    virNetworkIpDefPtr ipdef;
    bool oldDhcpActive = false;
3061
    bool needFirewallRefresh = false;
3062

3063 3064 3065 3066 3067

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3068
    networkDriverLock();
3069 3070 3071 3072 3073 3074 3075 3076

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

3077 3078 3079
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3080
    /* see if we are listening for dhcp pre-modification */
3081 3082 3083
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
3084 3085 3086 3087 3088 3089
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3090 3091
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3092
     */
3093
    isActive = virNetworkObjIsActive(network);
3094 3095
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3096 3097 3098 3099 3100 3101 3102
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119
    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).
                 */
3120
                networkRemoveFirewallRules(network->def);
3121 3122 3123 3124 3125 3126 3127 3128
                needFirewallRefresh = true;
                break;
            default:
                break;
            }
        }
    }

3129
    /* update the network config in memory/on disk */
3130 3131
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3132
            ignore_value(networkAddFirewallRules(network->def));
3133 3134 3135
        goto cleanup;
    }

3136
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159
        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.
             */
3160
            if (networkRestartDhcpDaemon(network) < 0)
3161 3162
                goto cleanup;

3163 3164 3165 3166 3167 3168 3169 3170
        } 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;

3171 3172 3173
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
                 i++) {
3174 3175 3176 3177 3178 3179 3180
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3181 3182
                 networkRestartDhcpDaemon(network) < 0) ||
                networkRefreshDhcpDaemon(network) < 0) {
3183 3184 3185 3186
                goto cleanup;
            }

        } else if (section == VIR_NETWORK_SECTION_DNS_HOST ||
3187 3188 3189 3190 3191 3192
                   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.
             */
3193
            if (networkRefreshDhcpDaemon(network) < 0)
3194 3195 3196 3197 3198 3199 3200 3201
                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.
             */
3202
            if (networkRefreshRadvd(network) < 0)
3203 3204 3205 3206
                goto cleanup;
        }

        /* save current network state to disk */
3207
        if ((ret = virNetworkSaveStatus(driver->stateDir,
3208
                                        network)) < 0) {
3209
            goto cleanup;
3210
        }
3211 3212
    }
    ret = 0;
3213
 cleanup:
3214 3215
    if (network)
        virNetworkObjUnlock(network);
3216
    networkDriverUnlock();
3217 3218 3219
    return ret;
}

3220 3221
static int networkCreate(virNetworkPtr net)
{
3222 3223
    virNetworkObjPtr network;
    int ret = -1;
3224
    virObjectEventPtr event = NULL;
3225

3226
    networkDriverLock();
3227
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3228

3229
    if (!network) {
3230 3231
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3232
        goto cleanup;
3233 3234
    }

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

3238
    if ((ret = networkStartNetwork(network)) < 0)
3239
        goto cleanup;
3240

3241 3242
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3243 3244
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3245

3246
 cleanup:
3247 3248
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3249 3250
    if (network)
        virNetworkObjUnlock(network);
3251
    networkDriverUnlock();
3252
    return ret;
3253 3254
}

3255 3256
static int networkDestroy(virNetworkPtr net)
{
3257 3258
    virNetworkObjPtr network;
    int ret = -1;
3259
    virObjectEventPtr event = NULL;
3260

3261
    networkDriverLock();
3262
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3263

3264
    if (!network) {
3265 3266
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3267
        goto cleanup;
3268 3269
    }

3270 3271 3272
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3273
    if (!virNetworkObjIsActive(network)) {
3274
        virReportError(VIR_ERR_OPERATION_INVALID,
3275 3276
                       _("network '%s' is not active"),
                       network->def->name);
3277 3278 3279
        goto cleanup;
    }

3280
    if ((ret = networkShutdownNetwork(network)) < 0)
3281 3282
        goto cleanup;

3283 3284
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3285 3286
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3287

3288
    if (!network->persistent) {
3289
        if (networkRemoveInactive(network) < 0) {
3290 3291 3292 3293
            network = NULL;
            ret = -1;
            goto cleanup;
        }
3294 3295
        network = NULL;
    }
3296

3297
 cleanup:
3298 3299
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3300 3301
    if (network)
        virNetworkObjUnlock(network);
3302
    networkDriverUnlock();
3303 3304 3305
    return ret;
}

3306
static char *networkGetXMLDesc(virNetworkPtr net,
3307
                               unsigned int flags)
3308
{
3309
    virNetworkObjPtr network;
3310
    virNetworkDefPtr def;
3311
    char *ret = NULL;
3312

3313
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3314

3315 3316
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3317

3318 3319 3320
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3321 3322 3323 3324 3325 3326
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3327

3328
 cleanup:
3329 3330
    if (network)
        virNetworkObjUnlock(network);
3331
    return ret;
3332 3333 3334
}

static char *networkGetBridgeName(virNetworkPtr net) {
3335 3336 3337
    virNetworkObjPtr network;
    char *bridge = NULL;

3338 3339
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3340

3341 3342 3343
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3344
    if (!(network->def->bridge)) {
3345 3346 3347
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3348 3349 3350
        goto cleanup;
    }

3351
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3352

3353
 cleanup:
3354 3355
    if (network)
        virNetworkObjUnlock(network);
3356 3357 3358 3359
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
J
Ján Tomko 已提交
3360
                               int *autostart)
3361
{
3362 3363
    virNetworkObjPtr network;
    int ret = -1;
3364

3365 3366
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3367

3368 3369 3370
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3371
    *autostart = network->autostart;
3372
    ret = 0;
3373

3374
 cleanup:
3375 3376
    if (network)
        virNetworkObjUnlock(network);
3377
    return ret;
3378 3379 3380
}

static int networkSetAutostart(virNetworkPtr net,
3381 3382
                               int autostart)
{
3383
    virNetworkObjPtr network;
3384
    char *configFile = NULL, *autostartLink = NULL;
3385
    int ret = -1;
3386

3387
    networkDriverLock();
3388
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3389

3390
    if (!network) {
3391 3392
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3393
        goto cleanup;
3394 3395
    }

3396 3397 3398
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3399
    if (!network->persistent) {
3400 3401
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3402 3403 3404
        goto cleanup;
    }

3405 3406
    autostart = (autostart != 0);

3407
    if (network->autostart != autostart) {
3408
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3409
            goto cleanup;
3410
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3411 3412
            goto cleanup;

3413
        if (autostart) {
3414
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3415
                virReportSystemError(errno,
3416 3417
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3418 3419
                goto cleanup;
            }
3420

3421
            if (symlink(configFile, autostartLink) < 0) {
3422
                virReportSystemError(errno,
3423
                                     _("Failed to create symlink '%s' to '%s'"),
3424
                                     autostartLink, configFile);
3425 3426 3427
                goto cleanup;
            }
        } else {
3428
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3429
                virReportSystemError(errno,
3430
                                     _("Failed to delete symlink '%s'"),
3431
                                     autostartLink);
3432 3433
                goto cleanup;
            }
3434 3435
        }

3436
        network->autostart = autostart;
3437
    }
3438
    ret = 0;
3439

3440
 cleanup:
3441 3442
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3443 3444
    if (network)
        virNetworkObjUnlock(network);
3445
    networkDriverUnlock();
3446
    return ret;
3447 3448
}

3449
static int
3450 3451 3452 3453
networkGetDHCPLeases(virNetworkPtr network,
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472
{
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
    int size = 0;
    int custom_lease_file_len = 0;
    bool need_results = !!leases;
    long long currtime = 0;
    long long expirytime_tmp = -1;
    bool ipv6 = false;
    char *lease_entries = NULL;
    char *custom_lease_file = NULL;
    const char *ip_tmp = NULL;
    const char *mac_tmp = NULL;
    virJSONValuePtr lease_tmp = NULL;
    virJSONValuePtr leases_array = NULL;
    virNetworkIpDefPtr ipdef_tmp = NULL;
    virNetworkDHCPLeasePtr lease = NULL;
    virNetworkDHCPLeasePtr *leases_ret = NULL;
3473 3474 3475 3476 3477 3478 3479 3480 3481
    virNetworkObjPtr obj;

    virCheckFlags(0, -1);

    if (!(obj = networkObjFromNetwork(network)))
        return -1;

    if (virNetworkGetDHCPLeasesEnsureACL(network->conn, obj->def) < 0)
        goto cleanup;
3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527

    /* Retrieve custom leases file location */
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(obj->def->bridge);

    /* Read entire contents */
    if ((custom_lease_file_len = virFileReadAll(custom_lease_file,
                                                VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX,
                                                &lease_entries)) < 0) {
        /* Even though src/network/leaseshelper.c guarantees the existence of
         * leases file (even if no leases are present), and the control reaches
         * here, instead of reporting error, return 0 leases */
        rv = 0;
        goto error;
    }

    if (custom_lease_file_len) {
        if (!(leases_array = virJSONValueFromString(lease_entries))) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("invalid json in file: %s"), custom_lease_file);
            goto error;
        }

        if ((size = virJSONValueArraySize(leases_array)) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("couldn't fetch array of leases"));
            goto error;
        }
    }

    currtime = (long long) time(NULL);

    for (i = 0; i < size; i++) {
        if (!(lease_tmp = virJSONValueArrayGet(leases_array, i))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("failed to parse json"));
            goto error;
        }

        if (!(mac_tmp = virJSONValueObjectGetString(lease_tmp, "mac-address"))) {
            /* leaseshelper program guarantees that lease will be stored only if
             * mac-address is known otherwise not */
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("found lease without mac-address"));
            goto error;
        }

3528
        if (mac && virMacAddrCompare(mac, mac_tmp))
3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578
            continue;

        if (virJSONValueObjectGetNumberLong(lease_tmp, "expiry-time", &expirytime_tmp) < 0) {
            /* A lease cannot be present without expiry-time */
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("found lease without expiry-time"));
            goto error;
        }

        /* Do not report expired lease */
        if (expirytime_tmp < currtime)
            continue;

        if (need_results) {
            if (VIR_ALLOC(lease) < 0)
                goto error;

            lease->expirytime = expirytime_tmp;

            if (!(ip_tmp = virJSONValueObjectGetString(lease_tmp, "ip-address"))) {
                /* A lease without ip-address makes no sense */
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("found lease without ip-address"));
                goto error;
            }

            /* Unlike IPv4, IPv6 uses ':' instead of '.' as separator */
            ipv6 = strchr(ip_tmp, ':') ? true : false;
            lease->type = ipv6 ? VIR_IP_ADDR_TYPE_IPV6 : VIR_IP_ADDR_TYPE_IPV4;

            /* Obtain prefix */
            for (j = 0; j < obj->def->nips; j++) {
                ipdef_tmp = &obj->def->ips[j];

                if (ipv6 && VIR_SOCKET_ADDR_IS_FAMILY(&ipdef_tmp->address,
                                                      AF_INET6)) {
                    lease->prefix = ipdef_tmp->prefix;
                    break;
                }
                if (!ipv6 && VIR_SOCKET_ADDR_IS_FAMILY(&ipdef_tmp->address,
                                                      AF_INET)) {
                    lease->prefix = virSocketAddrGetIpPrefix(&ipdef_tmp->address,
                                                             &ipdef_tmp->netmask,
                                                             ipdef_tmp->prefix);
                    break;
                }
            }

            if ((VIR_STRDUP(lease->mac, mac_tmp) < 0) ||
                (VIR_STRDUP(lease->ipaddr, ip_tmp) < 0) ||
3579
                (VIR_STRDUP(lease->iface, obj->def->bridge) < 0))
3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617
                goto error;

            /* Fields that can be NULL */
            if ((VIR_STRDUP(lease->iaid,
                            virJSONValueObjectGetString(lease_tmp, "iaid")) < 0) ||
                (VIR_STRDUP(lease->clientid,
                            virJSONValueObjectGetString(lease_tmp, "client-id")) < 0) ||
                (VIR_STRDUP(lease->hostname,
                            virJSONValueObjectGetString(lease_tmp, "hostname")) < 0))
                goto error;

            if (VIR_INSERT_ELEMENT(leases_ret, nleases, nleases, lease) < 0)
                goto error;

        } else {
            nleases++;
        }

        VIR_FREE(lease);
    }

    if (need_results && mac && !leases_ret) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("no lease with matching MAC address: %s"), mac);
        goto error;
    }

    if (leases_ret) {
        /* NULL terminated array */
        ignore_value(VIR_REALLOC_N(leases_ret, nleases + 1));
        *leases = leases_ret;
        leases_ret = NULL;
    }

    rv = nleases;

 cleanup:
    VIR_FREE(lease);
3618
    VIR_FREE(lease_entries);
3619 3620
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
3621 3622 3623 3624

    if (obj)
        virNetworkObjUnlock(obj);

3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635
    return rv;

 error:
    if (leases_ret) {
        for (i = 0; i < nleases; i++)
            virNetworkDHCPLeaseFree(leases_ret[i]);
        VIR_FREE(leases_ret);
    }
    goto cleanup;
}

3636 3637 3638

static virNetworkDriver networkDriver = {
    "Network",
3639 3640 3641 3642 3643 3644 3645
    .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 */
3646 3647
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
3648 3649
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3650 3651
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3652
    .networkUndefine = networkUndefine, /* 0.2.0 */
3653
    .networkUpdate = networkUpdate, /* 0.10.2 */
3654
    .networkCreate = networkCreate, /* 0.2.0 */
3655 3656 3657 3658 3659 3660 3661
    .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 */
3662
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
3663 3664 3665
};

static virStateDriver networkStateDriver = {
3666
    .name = "Network",
3667
    .stateInitialize  = networkStateInitialize,
3668
    .stateAutoStart  = networkStateAutoStart,
3669 3670
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3671 3672
};

3673 3674
int networkRegister(void)
{
3675 3676
    if (virRegisterNetworkDriver(&networkDriver) < 0)
        return -1;
3677 3678
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
3679 3680
    return 0;
}
3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692

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

/* 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.
 */

/* networkAllocateActualDevice:
3693
 * @dom: domain definition that @iface belongs to
3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704
 * @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
3705 3706
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
3707
{
3708
    virDomainNetType actualType = iface->type;
3709 3710
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
3711
    virNetDevBandwidthPtr bandwidth = NULL;
3712 3713 3714
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3715
    virNetworkForwardIfDefPtr dev = NULL;
3716
    size_t i;
3717 3718 3719
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3720
        goto validate;
3721 3722 3723 3724

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

3725
    networkDriverLock();
3726
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
3727
    networkDriverUnlock();
3728
    if (!network) {
3729 3730 3731
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3732
        goto error;
3733 3734
    }
    netdef = network->def;
3735

3736 3737 3738 3739 3740 3741 3742
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

3743 3744 3745
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

3746 3747 3748
    /* 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.
J
Ján Tomko 已提交
3749
     */
3750 3751 3752 3753 3754 3755
    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.
     */
3756 3757 3758 3759 3760 3761

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

3762 3763
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
3764
        goto error;
3765

3766 3767 3768 3769 3770 3771 3772 3773
    /* 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;

3774 3775
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
3776

3777 3778 3779 3780 3781 3782 3783 3784 3785 3786
    if (iface->trustGuestRxFilters)
       iface->data.network.actual->trustGuestRxFilters
          = iface->trustGuestRxFilters;
    else if (portgroup && portgroup->trustGuestRxFilters)
       iface->data.network.actual->trustGuestRxFilters
          = portgroup->trustGuestRxFilters;
    else if (netdef->trustGuestRxFilters)
       iface->data.network.actual->trustGuestRxFilters
          = netdef->trustGuestRxFilters;

3787 3788 3789
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
3790 3791 3792
        /* for these forward types, the actual net type really *is*
         *NETWORK; we just keep the info from the portgroup in
         * iface->data.network.actual
J
Ján Tomko 已提交
3793
         */
3794
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
3795 3796 3797 3798

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

3799
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
3800
               netdef->bridge) {
3801 3802 3803 3804 3805

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

3806
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
3807 3808
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
3809
            goto error;
3810

3811 3812 3813 3814 3815 3816 3817 3818
        /* 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) {
3819
            goto error;
3820 3821 3822 3823 3824 3825 3826 3827 3828 3829
        }
        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);
3830
                goto error;
3831 3832 3833
            }
        }

3834
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3835

3836
        virDomainHostdevSubsysPCIBackendType backend;
3837

3838
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
3839
        if (networkCreateInterfacePool(netdef) < 0)
3840 3841 3842
            goto error;

        /* pick first dev with 0 connections */
3843 3844 3845
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859
                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;
3860
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
3861
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
3862
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
3863

E
Eric Blake 已提交
3864
        switch (netdef->forward.driverName) {
3865
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
3866
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
3867 3868
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
3869
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
3870 3871
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
3872
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883
            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;

3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908
        /* 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;
            }
        }

3909 3910 3911 3912
    } 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)) {
3913 3914 3915 3916 3917 3918

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
3919
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
3920
        switch (netdef->forward.type) {
3921
        case VIR_NETWORK_FORWARD_BRIDGE:
3922
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
3923 3924
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
3925
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
3926 3927
            break;
        case VIR_NETWORK_FORWARD_VEPA:
3928
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
3929 3930
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
3931
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
3932 3933 3934
            break;
        }

3935 3936 3937 3938 3939 3940 3941 3942
        /* 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) {
3943
            goto error;
3944
        }
3945
        virtport = iface->data.network.actual->virtPortProfile;
3946
        if (virtport) {
3947 3948 3949 3950 3951 3952 3953 3954
            /* 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);
3955
                goto error;
3956 3957
            }
        }
3958

3959 3960 3961
        /* 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).
         */
3962
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
3963 3964 3965 3966
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
3967
            goto error;
3968 3969 3970
        } else {
            /* pick an interface from the pool */

3971
            if (networkCreateInterfacePool(netdef) < 0)
3972 3973
                goto error;

3974 3975 3976 3977 3978
            /* 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.
3979
             */
3980 3981
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
3982 3983 3984
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
3985

3986
                /* pick first dev with 0 connections */
3987 3988 3989
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
3990 3991 3992 3993 3994
                        break;
                    }
                }
            } else {
                /* pick least used dev */
3995
                dev = &netdef->forward.ifs[0];
3996 3997 3998
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
3999 4000 4001 4002
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4003 4004 4005 4006
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4007
                goto error;
4008
            }
4009 4010
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4011
                goto error;
4012 4013 4014
        }
    }

4015
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4016
        goto error;
4017

4018
 validate:
4019 4020 4021 4022 4023
    /* 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.
     */

4024
    if (virDomainNetGetActualVlan(iface)) {
4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053
        /* 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);
4054

4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073
        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 */
4074
            netdef->connections--;
4075 4076 4077 4078
            if (dev)
                dev->connections--;
            goto error;
        }
4079 4080
    }

4081
    ret = 0;
4082

4083
 cleanup:
4084 4085
    if (network)
        virNetworkObjUnlock(network);
4086 4087
    return ret;

4088
 error:
4089
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4090 4091 4092
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4093
    goto cleanup;
4094 4095 4096
}

/* networkNotifyActualDevice:
4097
 * @dom: domain definition that @iface belongs to
4098 4099 4100 4101 4102 4103 4104 4105 4106 4107
 * @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
4108 4109
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4110
{
4111
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4112 4113
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4114
    virNetworkForwardIfDefPtr dev = NULL;
4115 4116
    size_t i;
    int ret = -1;
4117 4118

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4119
        return 0;
4120

4121
    networkDriverLock();
4122
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
4123
    networkDriverUnlock();
4124
    if (!network) {
4125 4126 4127
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4128 4129 4130 4131 4132
        goto error;
    }
    netdef = network->def;

    if (!iface->data.network.actual ||
4133 4134
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4135 4136
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4137 4138
    }

4139
    if (networkCreateInterfacePool(netdef) < 0)
4140
        goto error;
4141

4142
    if (netdef->forward.nifs == 0) {
4143
        virReportError(VIR_ERR_INTERNAL_ERROR,
4144 4145
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4146
                       netdef->name);
4147
        goto error;
4148
    }
4149

4150 4151
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4152

4153 4154 4155 4156 4157 4158 4159 4160 4161
        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 */
4162 4163
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4164
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4165 4166
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4167 4168 4169 4170 4171
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4172
            virReportError(VIR_ERR_INTERNAL_ERROR,
4173 4174
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4175
                           netdef->name, actualDev);
4176
            goto error;
4177 4178
        }

4179
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4180 4181
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4182
         */
4183
        if ((dev->connections > 0) &&
4184 4185
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4186 4187
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4188
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4189
            virReportError(VIR_ERR_INTERNAL_ERROR,
4190 4191
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4192
                           netdef->name, actualDev);
4193
            goto error;
4194
        }
4195

4196
        /* we are now assured of success, so mark the allocation */
4197
        dev->connections++;
4198
        VIR_DEBUG("Using physical device %s, connections %d",
4199
                  dev->device.dev, dev->connections);
4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212

    }  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 */
4213 4214
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4215
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4216
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4217 4218
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4219 4220 4221 4222 4223 4224 4225 4226 4227
                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,
4228 4229 4230 4231
                           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);
J
Ján Tomko 已提交
4232
            goto error;
4233 4234 4235 4236 4237 4238 4239
        }

        /* 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) &&
4240
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256
            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);
4257 4258
    }

4259
 success:
4260 4261 4262
    netdef->connections++;
    VIR_DEBUG("Using network %s, %d connections",
              netdef->name, netdef->connections);
4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273

    /* 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;
    }

4274
    ret = 0;
4275
 cleanup:
4276 4277 4278
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4279

4280
 error:
4281
    goto cleanup;
4282 4283 4284 4285
}


/* networkReleaseActualDevice:
4286
 * @dom: domain definition that @iface belongs to
4287 4288 4289 4290 4291 4292 4293 4294 4295 4296
 * @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
4297 4298
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4299
{
4300
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4301
    virNetworkObjPtr network;
4302
    virNetworkDefPtr netdef;
4303
    virNetworkForwardIfDefPtr dev = NULL;
4304 4305
    size_t i;
    int ret = -1;
4306 4307

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4308
        return 0;
4309

4310
    networkDriverLock();
4311
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
4312
    networkDriverUnlock();
4313
    if (!network) {
4314 4315 4316
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4317 4318 4319 4320
        goto error;
    }
    netdef = network->def;

4321 4322
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4323 4324 4325 4326 4327
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) &&
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4328 4329 4330
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4331 4332
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4333 4334
    }

4335
    if (netdef->forward.nifs == 0) {
4336
        virReportError(VIR_ERR_INTERNAL_ERROR,
4337
                       _("network '%s' uses a direct/hostdev mode, but "
4338 4339
                         "has no forward dev and no interface pool"),
                       netdef->name);
4340
        goto error;
4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352
    }

    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;
        }
4353

4354 4355
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4356
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4357 4358
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4359 4360 4361
                break;
            }
        }
4362

4363
        if (!dev) {
4364
            virReportError(VIR_ERR_INTERNAL_ERROR,
4365 4366
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4367
                           netdef->name, actualDev);
4368
            goto error;
4369 4370
        }

4371
        dev->connections--;
4372
        VIR_DEBUG("Releasing physical device %s, connections %d",
4373
                  dev->device.dev, dev->connections);
4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384

    } 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;
        }

4385 4386
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4387
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4388
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4389 4390
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4391 4392 4393 4394 4395 4396 4397 4398 4399
                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,
4400 4401 4402 4403
                           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);
J
Ján Tomko 已提交
4404
            goto error;
4405 4406 4407 4408 4409 4410 4411
        }

        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);
J
Ján Tomko 已提交
4412
    }
4413

4414
 success:
4415
    if (iface->data.network.actual) {
4416
        netdef->connections--;
4417 4418
        VIR_DEBUG("Releasing network %s, %d connections",
                  netdef->name, netdef->connections);
4419

4420 4421 4422 4423
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
    }
4424
    ret = 0;
4425
 cleanup:
4426 4427
    if (network)
        virNetworkObjUnlock(network);
4428 4429 4430 4431
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4432
    return ret;
4433

4434
 error:
4435
    goto cleanup;
4436
}
4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460

/*
 * 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;
4461
    virNetworkObjPtr network;
4462 4463 4464 4465
    virNetworkDefPtr netdef;
    virNetworkIpDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4466
    char *dev_name = NULL;
4467 4468

    *netaddr = NULL;
4469
    networkDriverLock();
4470
    network = virNetworkFindByName(&driver->networks, netname);
4471
    networkDriverUnlock();
4472
    if (!network) {
4473 4474 4475
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4476
        goto error;
4477 4478 4479
    }
    netdef = network->def;

4480
    switch (netdef->forward.type) {
4481 4482 4483 4484 4485 4486
    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) {
4487 4488 4489
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have an IPv4 address"),
                           netdef->name);
4490 4491 4492 4493 4494 4495
            break;
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4496
        if ((dev_name = netdef->bridge))
4497 4498 4499 4500 4501 4502 4503 4504
            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:
4505 4506
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4507

4508
        if (!dev_name) {
4509 4510 4511
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4512 4513 4514 4515
        }
        break;
    }

4516
    if (dev_name) {
4517
        if (virNetDevGetIPv4Address(dev_name, &addr) < 0)
4518
            goto error;
4519
        addrptr = &addr;
4520 4521
    }

4522 4523 4524
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
        goto error;
4525 4526
    }

4527
    ret = 0;
4528
 cleanup:
4529 4530 4531
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4532

4533
 error:
4534
    goto cleanup;
4535
}
4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553

/**
 * 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;
4554
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4555 4556 4557 4558
    unsigned long long tmp_floor_sum = net->floor_sum;
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569
    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;
    }

4570
    if (!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor ||
4571 4572
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4573
        return 1;
4574
    }
4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605

    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;

4606
 cleanup:
4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643
    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];
4644
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672

    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;
    }

4673 4674
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
4675 4676 4677 4678 4679 4680 4681 4682
    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 */
4683
    net->floor_sum += ifaceBand->in->floor;
4684
    /* update status file */
4685
    if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
4686
        ignore_value(virBitmapClearBit(net->class_id, class_id));
4687
        net->floor_sum -= ifaceBand->in->floor;
4688 4689 4690 4691
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
4692 4693 4694 4695 4696 4697 4698 4699 4700
    /* 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;

4701
 cleanup:
4702 4703 4704 4705 4706 4707 4708 4709 4710
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
    int ret = 0;
    unsigned long long new_rate;
4711
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4712 4713 4714

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
4715 4716 4717 4718 4719
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730
        /* 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 */
4731
        net->floor_sum -= ifaceBand->in->floor;
4732 4733 4734 4735
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
4736
        if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
4737
            net->floor_sum += ifaceBand->in->floor;
4738 4739 4740 4741
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
4742 4743 4744 4745 4746 4747 4748 4749 4750 4751
        /* 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;
    }

4752
 cleanup:
4753 4754
    return ret;
}
4755 4756 4757

static void
networkNetworkObjTaint(virNetworkObjPtr net,
4758
                       virNetworkTaintFlags taint)
4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769
{
    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));
    }
}