bridge_driver.c 158.7 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2015 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 = virNetworkObjFindByUUID(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;
}

212 213 214 215 216 217
static char *
networkDnsmasqLeaseFileNameCustom(const char *bridge)
{
    char *leasefile;

    ignore_value(virAsprintf(&leasefile, "%s/%s.status",
218
                             driver->dnsmasqStateDir, bridge));
219 220 221
    return leasefile;
}

222 223 224 225 226
static char *
networkDnsmasqConfigFileName(const char *netname)
{
    char *conffile;

227
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
228
                             driver->dnsmasqStateDir, netname));
229 230 231
    return conffile;
}

232 233 234 235 236 237
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

238
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
239 240 241 242 243 244 245 246
    return pidfilebase;
}

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

247
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
248
                             driver->radvdStateDir, netname));
249 250
    return configfile;
}
251

252 253
/* do needed cleanup steps and remove the network from the list */
static int
254
networkRemoveInactive(virNetworkObjPtr net)
255 256
{
    char *leasefile = NULL;
257
    char *customleasefile = NULL;
258
    char *radvdconfigfile = NULL;
259
    char *configfile = NULL;
260
    char *radvdpidbase = NULL;
261
    char *statusfile = NULL;
262 263 264 265 266 267
    dnsmasqContext *dctx = NULL;
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(net);

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
268
    if (!(dctx = dnsmasqContextNew(def->name,
269
                                   driver->dnsmasqStateDir))) {
270
        goto cleanup;
271
    }
272

273
    if (!(leasefile = networkDnsmasqLeaseFileNameDefault(def->name)))
274 275
        goto cleanup;

276 277 278
    if (!(customleasefile = networkDnsmasqLeaseFileNameCustom(def->bridge)))
        goto cleanup;

279
    if (!(radvdconfigfile = networkRadvdConfigFileName(def->name)))
280
        goto cleanup;
281 282

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
283
        goto cleanup;
284

285
    if (!(configfile = networkDnsmasqConfigFileName(def->name)))
286
        goto cleanup;
287

288
    if (!(statusfile
289
          = virNetworkConfigFile(driver->stateDir, def->name)))
290
        goto cleanup;
291

292 293 294
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
295
    unlink(customleasefile);
296
    unlink(configfile);
297 298 299

    /* radvd */
    unlink(radvdconfigfile);
300
    virPidFileDelete(driver->pidDir, radvdpidbase);
301

302 303 304
    /* remove status file */
    unlink(statusfile);

305
    /* remove the network definition */
306
    virNetworkRemoveInactive(driver->networks, net);
307 308 309

    ret = 0;

310
 cleanup:
311
    VIR_FREE(leasefile);
312
    VIR_FREE(configfile);
313
    VIR_FREE(customleasefile);
314 315
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
316
    VIR_FREE(statusfile);
317 318 319 320
    dnsmasqContextFree(dctx);
    return ret;
}

321 322 323
static char *
networkBridgeDummyNicName(const char *brname)
{
324
    static const char dummyNicSuffix[] = "-nic";
325 326
    char *nicname;

327 328 329 330 331 332 333
    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.
         */
334 335 336 337 338
        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));
339
    } else {
340
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
341
    }
342 343 344
    return nicname;
}

345 346 347
/* 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). */
348
static void
349
networkUpdateAllState(void)
350
{
351
    size_t i;
352

353 354
    for (i = 0; i < driver->networks->count; i++) {
        virNetworkObjPtr obj = driver->networks->objs[i];
355 356

        virNetworkObjLock(obj);
357 358 359 360
        if (!virNetworkObjIsActive(obj)) {
            virNetworkObjUnlock(obj);
            continue;
        }
361

362 363 364 365 366 367 368 369
        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;
370

371 372 373 374 375 376 377 378 379 380 381 382 383 384 385
        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;
386

387 388 389 390
        case VIR_NETWORK_FORWARD_HOSTDEV:
            /* so far no extra checks */
            break;
        }
391

392 393 394 395
        /* Try and read dnsmasq/radvd pids of active networks */
        if (obj->active && obj->def->ips && (obj->def->nips > 0)) {
            char *radvdpidbase;

396
            ignore_value(virPidFileReadIfAlive(driver->pidDir,
397 398 399 400 401 402
                                               obj->def->name,
                                               &obj->dnsmasqPid,
                                               dnsmasqCapsGetBinaryPath(driver->dnsmasqCaps)));
            radvdpidbase = networkRadvdPidfileBasename(obj->def->name);
            if (!radvdpidbase)
                break;
403
            ignore_value(virPidFileReadIfAlive(driver->pidDir,
404 405 406
                                               radvdpidbase,
                                               &obj->radvdPid, RADVD));
            VIR_FREE(radvdpidbase);
407 408 409 410
        }

        virNetworkObjUnlock(obj);
    }
411 412 413

    /* remove inactive transient networks */
    i = 0;
414 415
    while (i < driver->networks->count) {
        virNetworkObjPtr obj = driver->networks->objs[i];
416 417 418
        virNetworkObjLock(obj);

        if (!obj->persistent && !obj->active) {
419
            networkRemoveInactive(obj);
420 421 422 423 424 425
            continue;
        }

        virNetworkObjUnlock(obj);
        i++;
    }
426 427 428
}


429
static void
430
networkAutostartConfigs(void)
431
{
432
    size_t i;
433

434 435 436 437 438
    for (i = 0; i < driver->networks->count; i++) {
        virNetworkObjLock(driver->networks->objs[i]);
        if (driver->networks->objs[i]->autostart &&
            !virNetworkObjIsActive(driver->networks->objs[i])) {
            if (networkStartNetwork(driver->networks->objs[i]) < 0) {
J
Ján Tomko 已提交
439
                /* failed to start but already logged */
440
            }
441
        }
442
        virNetworkObjUnlock(driver->networks->objs[i]);
443 444 445
    }
}

446 447 448
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
449
                             DBusMessage *message, void *user_data ATTRIBUTE_UNUSED)
450
{
451 452 453 454 455 456
    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.");
457
        networkReloadFirewallRules();
458 459 460 461 462 463
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

464
static int
465
networkMigrateStateFiles(void)
466 467 468 469 470 471 472 473 474 475 476 477 478
{
    /* 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;
479
    int direrr;
480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498
    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;
    }

499
    while ((direrr = virDirRead(dir, &entry, oldStateDir)) > 0) {
500 501 502
        if (entry->d_type != DT_UNKNOWN &&
            entry->d_type != DT_REG)
            continue;
503

504
        if (STREQ(entry->d_name, ".") ||
505 506 507 508 509
            STREQ(entry->d_name, ".."))
            continue;

        if (virAsprintf(&oldPath, "%s/%s",
                        oldStateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
510
            goto cleanup;
511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527

        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;
            }
        }

528
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
J
Ján Tomko 已提交
529
            goto cleanup;
530 531 532

        if (virAsprintf(&newPath, "%s/%s",
                        driver->stateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
533
            goto cleanup;
534 535 536 537 538 539 540 541 542 543 544 545
        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);
    }
546
    if (direrr < 0)
J
Ján Tomko 已提交
547
        goto cleanup;
548 549 550 551 552 553 554 555 556 557

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

558
/**
559
 * networkStateInitialize:
560 561 562 563
 *
 * Initialization function for the QEmu daemon
 */
static int
564 565 566
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
567
{
568 569 570
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
571 572 573
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
574

575
    if (VIR_ALLOC(driver) < 0)
576
        goto error;
577

578 579
    if (virMutexInit(&driver->lock) < 0) {
        VIR_FREE(driver);
580 581
        goto error;
    }
582
    networkDriverLock();
583

584 585 586 587
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
588
    if (privileged) {
589
        if (VIR_STRDUP(driver->networkConfigDir,
590
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
591
            VIR_STRDUP(driver->networkAutostartDir,
592
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
593
            VIR_STRDUP(driver->stateDir,
594
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
595
            VIR_STRDUP(driver->pidDir,
596
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
597
            VIR_STRDUP(driver->dnsmasqStateDir,
598
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
599
            VIR_STRDUP(driver->radvdStateDir,
600 601
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
602 603 604 605 606

        /* migration from old to new location is only applicable for
         * privileged mode - unprivileged mode directories haven't
         * changed location.
         */
607
        if (networkMigrateStateFiles() < 0)
608
            goto error;
609
    } else {
610 611 612
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
613
            goto error;
614

615
        if ((virAsprintf(&driver->networkConfigDir,
616
                         "%s/qemu/networks", configdir) < 0) ||
617
            (virAsprintf(&driver->networkAutostartDir,
618
                         "%s/qemu/networks/autostart", configdir) < 0) ||
619
            (virAsprintf(&driver->stateDir,
620
                         "%s/network/lib", rundir) < 0) ||
621
            (virAsprintf(&driver->pidDir,
622
                         "%s/network/run", rundir) < 0) ||
623
            (virAsprintf(&driver->dnsmasqStateDir,
624
                         "%s/dnsmasq/lib", rundir) < 0) ||
625
            (virAsprintf(&driver->radvdStateDir,
626
                         "%s/radvd/lib", rundir) < 0)) {
627
            goto error;
628
        }
629 630
    }

631
    if (virFileMakePath(driver->stateDir) < 0) {
632 633
        virReportSystemError(errno,
                             _("cannot create directory %s"),
634
                             driver->stateDir);
635 636 637
        goto error;
    }

638
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
639
    driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
640

641 642 643 644
    if (VIR_ALLOC(driver->networks) < 0)
        goto error;

    if (virNetworkLoadAllState(driver->networks,
645
                               driver->stateDir) < 0)
646 647
        goto error;

648
    if (virNetworkLoadAllConfigs(driver->networks,
649 650
                                 driver->networkConfigDir,
                                 driver->networkAutostartDir) < 0)
651 652
        goto error;

653 654 655
    networkUpdateAllState();
    networkReloadFirewallRules();
    networkRefreshDaemons();
656

657
    driver->networkEventState = virObjectEventStateNew();
658

659
    networkDriverUnlock();
660

661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682
#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,
683
                                   NULL, NULL);
684 685 686
    }
#endif

687
    ret = 0;
688
 cleanup:
689 690 691
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
692

693
 error:
694 695
    if (driver)
        networkDriverUnlock();
696
    networkStateCleanup();
697
    goto cleanup;
698 699
}

700 701 702 703 704 705 706 707
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
708
    if (!driver)
709 710
        return;

711 712 713
    networkDriverLock();
    networkAutostartConfigs();
    networkDriverUnlock();
714 715
}

716
/**
717
 * networkStateReload:
718 719 720 721 722
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
723 724
networkStateReload(void)
{
725
    if (!driver)
726 727
        return 0;

728
    networkDriverLock();
729
    virNetworkLoadAllState(driver->networks,
730
                           driver->stateDir);
731
    virNetworkLoadAllConfigs(driver->networks,
732 733 734 735 736 737
                             driver->networkConfigDir,
                             driver->networkAutostartDir);
    networkReloadFirewallRules();
    networkRefreshDaemons();
    networkAutostartConfigs();
    networkDriverUnlock();
738 739 740 741 742
    return 0;
}


/**
743
 * networkStateCleanup:
744 745 746 747
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
748 749
networkStateCleanup(void)
{
750
    if (!driver)
751 752
        return -1;

753
    networkDriverLock();
754

755
    virObjectEventStateFree(driver->networkEventState);
756

757
    /* free inactive networks */
758 759
    virNetworkObjListFree(driver->networks);
    VIR_FREE(driver->networks);
760

761 762 763 764 765 766
    VIR_FREE(driver->networkConfigDir);
    VIR_FREE(driver->networkAutostartDir);
    VIR_FREE(driver->stateDir);
    VIR_FREE(driver->pidDir);
    VIR_FREE(driver->dnsmasqStateDir);
    VIR_FREE(driver->radvdStateDir);
767

768
    virObjectUnref(driver->dnsmasqCaps);
769

770 771
    networkDriverUnlock();
    virMutexDestroy(&driver->lock);
772

773
    VIR_FREE(driver);
774 775 776 777 778

    return 0;
}


779 780 781 782 783 784 785
/* 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)
{
786 787
    size_t i;
    int ret = -1;
788 789 790 791 792 793 794
    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.
     */
795
    for (i = 0; i < 25; i++) {
796
        int signum = 0;
797
        if (i == 0) {
798
            signum = SIGTERM;
799
        } else if (i == 15) {
800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834
            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);
835
 cleanup:
836 837 838
    return ret;
}

J
Ján Tomko 已提交
839 840 841
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
G
Gene Czarcinski 已提交
842

843
static int
G
Gene Czarcinski 已提交
844 845
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
                                 virNetworkIpDefPtr ipdef)
846
{
847
    size_t i;
G
Gene Czarcinski 已提交
848
    bool ipv6 = false;
849

G
Gene Czarcinski 已提交
850 851
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
852 853
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
854
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
855 856
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
857
                return -1;
858
    }
859

G
Gene Czarcinski 已提交
860 861 862 863 864 865 866
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
867
    size_t i, j;
G
Gene Czarcinski 已提交
868

869 870
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
871
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
872
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
873
                for (j = 0; j < host->nnames; j++)
874 875
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
876 877
            }
        }
878 879
    }

880
    return 0;
881 882 883
}


884 885
int
networkDnsmasqConfContents(virNetworkObjPtr network,
886 887 888 889
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
890
{
891
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
892
    int r, ret = -1;
893
    int nbleases = 0;
894
    size_t i;
895
    virNetworkDNSDefPtr dns = &network->def->dns;
G
Gene Czarcinski 已提交
896 897
    virNetworkIpDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
    bool ipv6SLAAC;
898

899 900
    *configstr = NULL;

901
    /*
902 903 904
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
905
     *
906 907
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
908
     */
909 910 911 912 913 914

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

915 916
    /* create dnsmasq config file appropriate for this network */
    virBufferAsprintf(&configbuf,
917 918 919 920 921 922 923
                      "##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"
924
                      "strict-order\n",
925 926
                      network->def->name);

927 928 929 930
    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
931
                              network->def->dns.forwarders[i]);
932 933 934
        }
    }

935
    if (network->def->domain) {
936 937 938 939 940
        if (network->def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
                              network->def->domain);
        }
941
        virBufferAsprintf(&configbuf,
942 943 944 945
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
946

J
Ján Tomko 已提交
947
    if (network->def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
948 949 950 951
        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)
952
         */
953
        virBufferAddLit(&configbuf, "local=//\n");
954
    }
955

956
    if (pidfile)
957
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
958

959 960 961
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

962 963 964 965 966 967 968 969
    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.
         */
970
        virBufferAsprintf(&configbuf,
971 972 973
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
974
    } else {
975
        virBufferAddLit(&configbuf, "bind-interfaces\n");
976 977 978 979 980 981 982 983
        /*
         * --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
         */
984 985 986
        for (i = 0;
             (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
             i++) {
987 988 989 990
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
991

992
            /* also part of CVE 2012-3411 - if the host's version of
993
             * dnsmasq doesn't have bind-dynamic, only allow listening on
994 995
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
996 997
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
998 999 1000 1001
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
1002
                                 "The version of dnsmasq on this host (%d.%d) "
1003 1004 1005 1006
                                 "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 "
1007 1008 1009 1010 1011 1012
                                 "(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);
1013
                VIR_FREE(ipaddr);
1014 1015
                goto cleanup;
            }
1016
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1017 1018 1019
            VIR_FREE(ipaddr);
        }
    }
1020

1021 1022
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1023
     * guaranteed to not work, and set no-resolv so that no dns
1024 1025 1026
     * 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).
1027
     */
1028
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1029
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1030
                        "no-resolv\n");
1031
    }
1032

1033
    for (i = 0; i < dns->ntxts; i++) {
1034
        virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
1035 1036
                          dns->txts[i].name,
                          dns->txts[i].value);
1037
    }
1038

1039
    for (i = 0; i < dns->nsrvs; i++) {
1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061
        /* 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);
1062

1063 1064 1065
        /* domain is optional - it defaults to the domain of this network */
        if (dns->srvs[i].domain)
            virBufferAsprintf(&configbuf, ".%s", dns->srvs[i].domain);
1066

1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088
        /* 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);
1089
        }
1090
        virBufferAddLit(&configbuf, "\n");
1091 1092
    }

G
Gene Czarcinski 已提交
1093
    /* Find the first dhcp for both IPv4 and IPv6 */
1094 1095 1096
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
1097 1098 1099 1100
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1101 1102
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113
                    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,
1114 1115 1116 1117 1118 1119 1120 1121 1122
                                   _("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
1123 1124 1125 1126
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1127 1128
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141
                    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 "
1142 1143 1144 1145
                 "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
1146 1147 1148 1149 1150
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1151
        for (r = 0; r < ipdef->nranges; r++) {
1152
            char *saddr = virSocketAddrFormat(&ipdef->ranges[r].start);
1153 1154
            if (!saddr)
                goto cleanup;
1155
            char *eaddr = virSocketAddrFormat(&ipdef->ranges[r].end);
1156 1157 1158 1159
            if (!eaddr) {
                VIR_FREE(saddr);
                goto cleanup;
            }
1160
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s\n",
1161
                              saddr, eaddr);
1162
            VIR_FREE(saddr);
1163
            VIR_FREE(eaddr);
1164 1165
            nbleases += virSocketAddrGetRange(&ipdef->ranges[r].start,
                                              &ipdef->ranges[r].end);
1166
        }
1167

1168
        /*
1169 1170 1171 1172
         * 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)
1173 1174
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1175
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1176 1177
            if (!bridgeaddr)
                goto cleanup;
1178
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static\n", bridgeaddr);
1179 1180
            VIR_FREE(bridgeaddr);
        }
1181

G
Gene Czarcinski 已提交
1182 1183
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1184

G
Gene Czarcinski 已提交
1185 1186 1187
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts)
1188
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1189

G
Gene Czarcinski 已提交
1190
            if (ipdef->tftproot) {
1191 1192
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1193
            }
1194

G
Gene Czarcinski 已提交
1195 1196 1197
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1198

1199
                    if (!bootserver)
G
Gene Czarcinski 已提交
1200
                        goto cleanup;
1201
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1202
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1203 1204
                    VIR_FREE(bootserver);
                } else {
1205
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1206 1207 1208 1209 1210
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1211

1212
    if (nbleases > 0)
1213
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1214

G
Gene Czarcinski 已提交
1215 1216
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1217
        goto cleanup;
G
Gene Czarcinski 已提交
1218 1219 1220 1221 1222 1223

    /* 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)
1224 1225
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1226

1227 1228
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1229
     */
1230
    virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
1231
                      dctx->addnhostsfile->path);
G
Gene Czarcinski 已提交
1232 1233 1234

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1235
        if (ipv6def) {
1236
            virBufferAddLit(&configbuf, "enable-ra\n");
1237
        } else {
1238 1239 1240
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
                 i++) {
G
Gene Czarcinski 已提交
1241 1242 1243 1244
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1245 1246
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1247 1248
                    VIR_FREE(bridgeaddr);
                }
1249
            }
1250
        }
1251 1252
    }

1253 1254 1255
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1256
    ret = 0;
G
Gene Czarcinski 已提交
1257

1258
 cleanup:
1259
    virBufferFreeAndReset(&configbuf);
1260
    return ret;
1261 1262
}

1263
/* build the dnsmasq command line */
1264 1265 1266
static int ATTRIBUTE_NONNULL(2)
networkBuildDhcpDaemonCommandLine(virNetworkObjPtr network,
                                  virCommandPtr *cmdout,
1267 1268
                                  char *pidfile, dnsmasqContext *dctx,
                                  dnsmasqCapsPtr caps)
1269
{
1270
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1271
    int ret = -1;
1272 1273
    char *configfile = NULL;
    char *configstr = NULL;
1274
    char *leaseshelper_path = NULL;
1275 1276

    network->dnsmasqPid = -1;
1277

1278 1279 1280 1281 1282 1283
    if (networkDnsmasqConfContents(network, pidfile, &configstr, dctx, caps) < 0)
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1284
    if (!(configfile = networkDnsmasqConfigFileName(network->def->name)))
1285 1286 1287 1288 1289
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1290 1291
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1292 1293 1294
        goto cleanup;
    }

1295 1296
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1297
                                                  abs_topbuilddir "/src",
1298 1299 1300
                                                  LIBEXECDIR)))
        goto cleanup;

1301 1302
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(caps));
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1303 1304
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1305
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1306
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", network->def->bridge);
1307

1308
    *cmdout = cmd;
1309
    ret = 0;
1310
 cleanup:
1311 1312
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1313
    VIR_FREE(leaseshelper_path);
1314 1315 1316 1317
    return ret;
}

static int
1318
networkStartDhcpDaemon(virNetworkObjPtr network)
1319 1320 1321 1322
{
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1323
    dnsmasqContext *dctx = NULL;
1324

1325
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0)) {
G
Gene Czarcinski 已提交
1326
        /* no IP addresses, so we don't need to run */
1327 1328 1329 1330
        ret = 0;
        goto cleanup;
    }

1331
    if (virFileMakePath(driver->pidDir) < 0) {
1332
        virReportSystemError(errno,
1333
                             _("cannot create directory %s"),
1334
                             driver->pidDir);
1335
        goto cleanup;
1336 1337
    }

1338
    if (!(pidfile = virPidFileBuildPath(driver->pidDir,
1339
                                        network->def->name)))
1340
        goto cleanup;
1341

1342
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1343
        virReportSystemError(errno,
1344
                             _("cannot create directory %s"),
1345
                             driver->dnsmasqStateDir);
1346 1347 1348
        goto cleanup;
    }

1349
    dctx = dnsmasqContextNew(network->def->name, driver->dnsmasqStateDir);
1350 1351 1352
    if (dctx == NULL)
        goto cleanup;

1353 1354
    if (dnsmasqCapsRefresh(&driver->dnsmasqCaps, NULL) < 0)
        goto cleanup;
1355 1356 1357

    ret = networkBuildDhcpDaemonCommandLine(network, &cmd, pidfile,
                                            dctx, driver->dnsmasqCaps);
1358 1359 1360 1361 1362
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1363
        goto cleanup;
1364

G
Guido Günther 已提交
1365
    ret = virCommandRun(cmd, NULL);
1366
    if (ret < 0)
1367 1368 1369
        goto cleanup;

    /*
1370 1371 1372 1373 1374
     * 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
1375 1376
     */

1377
    ret = virPidFileRead(driver->pidDir, network->def->name,
1378 1379
                         &network->dnsmasqPid);
    if (ret < 0)
1380
        goto cleanup;
1381

1382
    ret = 0;
1383
 cleanup:
1384
    VIR_FREE(pidfile);
1385
    virCommandFree(cmd);
1386
    dnsmasqContextFree(dctx);
1387 1388 1389
    return ret;
}

1390 1391
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1392 1393
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1394 1395 1396
 *
 *  Returns 0 on success, -1 on failure.
 */
1397
static int
1398
networkRefreshDhcpDaemon(virNetworkObjPtr network)
1399
{
1400 1401
    int ret = -1;
    size_t i;
G
Gene Czarcinski 已提交
1402
    virNetworkIpDefPtr ipdef, ipv4def, ipv6def;
1403
    dnsmasqContext *dctx = NULL;
1404

G
Gene Czarcinski 已提交
1405
    /* if no IP addresses specified, nothing to do */
1406
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1407 1408
        return 0;

1409 1410
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1411
        return networkStartDhcpDaemon(network);
1412

G
Gene Czarcinski 已提交
1413
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1414
    if (!(dctx = dnsmasqContextNew(network->def->name,
1415
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1416
        goto cleanup;
1417
    }
G
Gene Czarcinski 已提交
1418 1419 1420 1421 1422 1423

    /* 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;
1424 1425 1426
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
G
Gene Czarcinski 已提交
1427 1428
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1429 1430
    }

G
Gene Czarcinski 已提交
1431
    ipv6def = NULL;
1432 1433 1434
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1435 1436
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1437 1438
    }

G
Gene Czarcinski 已提交
1439
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1440
        goto cleanup;
G
Gene Czarcinski 已提交
1441 1442

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1443
        goto cleanup;
1444

G
Gene Czarcinski 已提交
1445
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
J
Ján Tomko 已提交
1446
        goto cleanup;
1447 1448

    if ((ret = dnsmasqSave(dctx)) < 0)
1449
        goto cleanup;
1450 1451

    ret = kill(network->dnsmasqPid, SIGHUP);
1452
 cleanup:
1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464
    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
1465
networkRestartDhcpDaemon(virNetworkObjPtr network)
1466 1467 1468 1469 1470 1471
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1472
    }
1473
    /* now start dnsmasq if it should be started */
1474
    return networkStartDhcpDaemon(network);
1475 1476
}

G
Gene Czarcinski 已提交
1477 1478 1479 1480 1481 1482
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";

1483 1484 1485
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1486
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1487 1488
    int ret = -1;
    size_t i;
1489
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
1490
    bool v6present = false, dhcp6 = false;
1491 1492

    *configstr = NULL;
1493

G
Gene Czarcinski 已提交
1494
    /* Check if DHCPv6 is needed */
1495 1496 1497
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510
        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;
    }

1511 1512 1513
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1514
    virBufferAsprintf(&configbuf, "interface %s\n"
1515 1516
                      "{\n"
                      "  AdvSendAdvert on;\n"
1517
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1518 1519 1520 1521 1522
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1523 1524

    /* add a section for each IPv6 address in the config */
1525 1526 1527
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
1528 1529 1530 1531 1532
        int prefix;
        char *netaddr;

        prefix = virNetworkIpDefPrefix(ipdef);
        if (prefix < 0) {
1533 1534 1535
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1536 1537
            goto cleanup;
        }
1538
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1539
            goto cleanup;
1540
        virBufferAsprintf(&configbuf,
1541
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1542 1543 1544
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1545 1546 1547
        VIR_FREE(netaddr);
    }

1548
    virBufferAddLit(&configbuf, "};\n");
1549

1550
    if (virBufferCheckError(&configbuf) < 0)
1551
        goto cleanup;
1552

1553 1554
    *configstr = virBufferContentAndReset(&configbuf);

1555
    ret = 0;
1556
 cleanup:
1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578
    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;
1579 1580 1581 1582
        goto cleanup;
    }

    /* construct the filename */
1583
    if (!(*configFile = networkRadvdConfigFileName(network->def->name)))
1584 1585
        goto cleanup;
    /* write the file */
1586
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1587 1588
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1589 1590 1591 1592 1593
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1594
 cleanup:
1595 1596 1597 1598 1599 1600
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1601
networkStartRadvd(virNetworkObjPtr network)
1602 1603 1604 1605 1606 1607 1608 1609 1610
{
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1611
    /* Is dnsmasq handling RA? */
J
Ján Tomko 已提交
1612
    if (DNSMASQ_RA_SUPPORT(driver->dnsmasqCaps)) {
G
Gene Czarcinski 已提交
1613 1614 1615 1616
        ret = 0;
        goto cleanup;
    }

1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627
    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);
1628 1629 1630
        goto cleanup;
    }

1631
    if (virFileMakePath(driver->pidDir) < 0) {
1632 1633
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1634
                             driver->pidDir);
1635 1636
        goto cleanup;
    }
1637
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1638 1639
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1640
                             driver->radvdStateDir);
1641 1642 1643 1644
        goto cleanup;
    }

    /* construct pidfile name */
1645
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1646
        goto cleanup;
1647
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1648 1649 1650 1651 1652
        goto cleanup;

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

1653 1654 1655 1656
    /* 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
1657
     * virPidFileRead() below will fail if we use them).
1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672
     * 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;

1673
    if (virPidFileRead(driver->pidDir, radvdpidbase, &network->radvdPid) < 0)
1674 1675 1676
        goto cleanup;

    ret = 0;
1677
 cleanup:
1678 1679 1680 1681 1682 1683 1684
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1685
static int
1686
networkRefreshRadvd(virNetworkObjPtr network)
1687
{
G
Gene Czarcinski 已提交
1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698
    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)) {
1699
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1700 1701 1702 1703 1704 1705
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }

1706 1707
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
1708
        return networkStartRadvd(network);
1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720

    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);
}

1721 1722
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1723
static int
1724
networkRestartRadvd(virNetworkObjPtr network)
1725 1726 1727 1728 1729 1730 1731 1732 1733
{
    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
1734
        if ((networkKillDaemon(network->radvdPid, "radvd",
1735 1736 1737
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1738
            virPidFileDelete(driver->pidDir, radvdpidbase);
1739 1740 1741 1742 1743 1744 1745 1746 1747
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1748 1749 1750 1751
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
1752
networkRefreshDaemons(void)
1753
{
1754
    size_t i;
1755 1756 1757

    VIR_INFO("Refreshing network daemons");

1758 1759
    for (i = 0; i < driver->networks->count; i++) {
        virNetworkObjPtr network = driver->networks->objs[i];
1760 1761 1762

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
1763 1764 1765
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
1766 1767 1768 1769 1770 1771
            /* 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.
             */
1772 1773
            networkRefreshDhcpDaemon(network);
            networkRefreshRadvd(network);
1774 1775 1776 1777 1778
        }
        virNetworkObjUnlock(network);
    }
}

1779
static void
1780
networkReloadFirewallRules(void)
1781
{
1782
    size_t i;
1783

1784
    VIR_INFO("Reloading iptables rules");
1785

1786 1787
    for (i = 0; i < driver->networks->count; i++) {
        virNetworkObjPtr network = driver->networks->objs[i];
1788 1789 1790

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
1791 1792 1793
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
1794 1795 1796
            /* Only the three L3 network types that are configured by libvirt
             * need to have iptables rules reloaded.
             */
1797 1798
            networkRemoveFirewallRules(network->def);
            if (networkAddFirewallRules(network->def) < 0) {
1799 1800
                /* failed to add but already logged */
            }
1801
        }
1802
        virNetworkObjUnlock(network);
1803 1804 1805
    }
}

1806
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
1807
static int
1808
networkEnableIpForwarding(bool enableIPv4, bool enableIPv6)
1809
{
1810
    int ret = 0;
1811 1812 1813 1814
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
1815
                           &enabled, sizeof(enabled));
1816 1817
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
1818
                           &enabled, sizeof(enabled));
1819
#else
1820 1821 1822 1823
    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);
1824
#endif
1825
    return ret;
1826 1827
}

1828 1829
#define SYSCTL_PATH "/proc/sys"

1830 1831
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
1832 1833 1834
{
    char *field = NULL;
    int ret = -1;
1835
    bool enableIPv6 =  !!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0);
1836

1837 1838 1839 1840 1841 1842 1843
    /* 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;
1844

1845 1846
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
1847 1848
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
1849 1850 1851
        ret = 0;
        goto cleanup;
    }
1852

1853 1854 1855 1856 1857
    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;
1858
    }
1859
    VIR_FREE(field);
1860

1861 1862
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
1863 1864 1865 1866 1867 1868
     */

    /* 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",
1869
                    network->def->bridge) < 0)
1870 1871
        goto cleanup;

1872
    if (virFileWriteStr(field, "0", 0) < 0) {
1873
        virReportSystemError(errno,
1874 1875 1876 1877 1878
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

1879 1880 1881 1882
    /* 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",
1883
                    network->def->bridge) < 0)
1884 1885
        goto cleanup;

1886
    if (virFileWriteStr(field, "0", 0) < 0) {
1887
        virReportSystemError(errno,
1888
                             _("cannot disable %s"), field);
1889 1890 1891 1892
        goto cleanup;
    }

    ret = 0;
1893
 cleanup:
1894 1895 1896 1897
    VIR_FREE(field);
    return ret;
}

1898
/* add an IP address to a bridge */
1899
static int
D
Daniel P. Berrange 已提交
1900
networkAddAddrToBridge(virNetworkObjPtr network,
1901
                       virNetworkIpDefPtr ipdef)
1902
{
1903 1904 1905
    int prefix = virNetworkIpDefPrefix(ipdef);

    if (prefix < 0) {
1906 1907 1908
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
1909 1910 1911
        return -1;
    }

1912 1913
    if (virNetDevSetIPAddress(network->def->bridge,
                              &ipdef->address, prefix) < 0)
1914 1915 1916 1917 1918
        return -1;

    return 0;
}

1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941

static int
networkStartHandleMACTableManagerMode(virNetworkObjPtr network,
                                      const char *macTapIfName)
{
    const char *brname = network->def->bridge;

    if (brname &&
        network->def->macTableManager
        == VIR_NETWORK_BRIDGE_MAC_TABLE_MANAGER_LIBVIRT) {
        if (virNetDevBridgeSetVlanFiltering(brname, true) < 0)
            return -1;
        if (macTapIfName) {
            if (virNetDevBridgePortSetLearning(brname, macTapIfName, false) < 0)
                return -1;
            if (virNetDevBridgePortSetUnicastFlood(brname, macTapIfName, false) < 0)
                return -1;
        }
    }
    return 0;
}


1942 1943 1944 1945 1946
/* add an IP (static) route to a bridge */
static int
networkAddRouteToBridge(virNetworkObjPtr network,
                        virNetworkRouteDefPtr routedef)
{
1947 1948 1949 1950
    int prefix = virNetworkRouteDefGetPrefix(routedef);
    unsigned int metric = virNetworkRouteDefGetMetric(routedef);
    virSocketAddrPtr addr = virNetworkRouteDefGetAddress(routedef);
    virSocketAddrPtr gateway = virNetworkRouteDefGetGateway(routedef);
1951 1952 1953 1954 1955 1956 1957 1958 1959

    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;
    }

1960 1961
    if (virNetDevAddRoute(network->def->bridge, addr,
                          prefix, gateway, metric) < 0) {
1962 1963 1964 1965 1966
        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 2045 2046
    if (networkStartHandleMACTableManagerMode(network, macTapIfName) < 0)
        goto err2;

2047
    /* Bring up the bridge interface */
2048
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2049
        goto err2;
2050

2051
    for (i = 0; i < network->def->nroutes; i++) {
2052 2053 2054 2055 2056
        virSocketAddrPtr gateway = NULL;

        routedef = network->def->routes[i];
        gateway = virNetworkRouteDefGetGateway(routedef);

2057 2058 2059
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2060
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2061 2062 2063 2064 2065 2066 2067
            if (networkAddRouteToBridge(network, routedef) < 0) {
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2068 2069
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2070
        networkEnableIpForwarding(v4present, v6present) < 0) {
2071
        virReportSystemError(errno, "%s",
2072
                             _("failed to enable IP forwarding"));
2073
        goto err3;
2074 2075
    }

2076

2077
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2078
    if ((v4present || v6present) &&
2079
        networkStartDhcpDaemon(network) < 0)
2080
        goto err3;
2081

2082
    /* start radvd if there are any ipv6 addresses */
2083
    if (v6present && networkStartRadvd(network) < 0)
2084 2085
        goto err4;

2086 2087 2088 2089 2090 2091 2092 2093 2094
    /* 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);
    }

2095
    if (virNetDevBandwidthSet(network->def->bridge,
2096
                              network->def->bandwidth, true) < 0)
2097 2098
        goto err5;

2099
    VIR_FREE(macTapIfName);
2100 2101 2102

    return 0;

2103
 err5:
2104 2105
    if (network->def->bandwidth)
       virNetDevBandwidthClear(network->def->bridge);
2106

2107 2108 2109 2110
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2111 2112 2113 2114 2115
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2116 2117 2118
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2119
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2120

2121 2122 2123
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2124
    networkRemoveFirewallRules(network->def);
2125 2126

 err1:
2127 2128 2129
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2130
    if (macTapIfName) {
2131
        VIR_FORCE_CLOSE(tapfd);
2132
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2133 2134
        VIR_FREE(macTapIfName);
    }
2135 2136

 err0:
2137 2138
    if (!save_err)
        save_err = virSaveLastError();
2139
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2140

2141 2142 2143 2144
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2145
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2146 2147 2148
    return -1;
}

2149
static int networkShutdownNetworkVirtual(virNetworkObjPtr network)
2150
{
2151 2152
    if (network->def->bandwidth)
        virNetDevBandwidthClear(network->def->bridge);
2153

2154 2155 2156 2157 2158
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2159
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2160
            virPidFileDelete(driver->pidDir, radvdpidbase);
2161 2162 2163 2164
            VIR_FREE(radvdpidbase);
        }
    }

2165 2166 2167
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2168
    if (network->def->mac_specified) {
2169
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2170
        if (macTapIfName) {
2171
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2172 2173 2174 2175
            VIR_FREE(macTapIfName);
        }
    }

2176
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2177

2178
    networkRemoveFirewallRules(network->def);
2179

2180
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2181

2182
    /* See if its still alive and really really kill it */
2183
    if (network->dnsmasqPid > 0 &&
2184
        (kill(network->dnsmasqPid, 0) == 0))
2185 2186
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2187 2188 2189 2190 2191 2192

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

2193 2194 2195
    return 0;
}

2196

2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217
static int
networkStartNetworkBridge(virNetworkObjPtr network)
{
    /* put anything here that needs to be done each time a network of
     * type BRIDGE, is started. On failure, undo anything you've done,
     * and return -1. On success return 0.
     */
    return networkStartHandleMACTableManagerMode(network, NULL);
}

static int
networkShutdownNetworkBridge(virNetworkObjPtr network ATTRIBUTE_UNUSED)
{
    /* put anything here that needs to be done each time a network of
     * type BRIDGE is shutdown. On failure, undo anything you've done,
     * and return -1. On success return 0.
     */
    return 0;
}


2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232
/* 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;

2233 2234 2235
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

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 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323
    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;
}


2324
static int
2325
networkStartNetworkExternal(virNetworkObjPtr network)
2326 2327
{
    /* put anything here that needs to be done each time a network of
2328
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2329 2330 2331
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2332
    return networkCreateInterfacePool(network->def);
2333 2334
}

2335
static int networkShutdownNetworkExternal(virNetworkObjPtr network ATTRIBUTE_UNUSED)
2336 2337
{
    /* put anything here that needs to be done each time a network of
2338
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2339 2340 2341 2342 2343 2344 2345
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2346
networkStartNetwork(virNetworkObjPtr network)
2347
{
2348 2349 2350
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2351 2352

    if (virNetworkObjIsActive(network)) {
2353 2354
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2355
        return ret;
2356 2357
    }

2358 2359 2360
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2361
    if (virNetworkObjSetDefTransient(network, true) < 0)
2362
        goto cleanup;
2363

2364 2365
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2366
    if (networkRunHook(network, NULL, NULL,
2367 2368 2369 2370
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2371
    switch (network->def->forward.type) {
2372 2373 2374 2375

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2376
        if (networkStartNetworkVirtual(network) < 0)
2377
            goto cleanup;
2378 2379 2380
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2381 2382 2383 2384
       if (networkStartNetworkBridge(network) < 0)
          goto cleanup;
       break;

2385 2386 2387
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2388
    case VIR_NETWORK_FORWARD_HOSTDEV:
2389
        if (networkStartNetworkExternal(network) < 0)
2390
            goto cleanup;
2391 2392 2393
        break;
    }

2394
    /* finally we can call the 'started' hook script if any */
2395
    if (networkRunHook(network, NULL, NULL,
2396 2397 2398 2399
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2400 2401 2402
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2403
    VIR_DEBUG("Writing network status to disk");
2404
    if (virNetworkSaveStatus(driver->stateDir, network) < 0)
2405
        goto cleanup;
2406 2407

    network->active = 1;
2408 2409
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2410

2411
 cleanup:
2412
    if (ret < 0) {
2413
        virNetworkObjUnsetDefTransient(network);
2414 2415
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2416
        networkShutdownNetwork(network);
2417 2418 2419 2420 2421 2422 2423
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2424
static int networkShutdownNetwork(virNetworkObjPtr network)
2425 2426 2427 2428 2429 2430 2431 2432 2433
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

2434
    stateFile = virNetworkConfigFile(driver->stateDir,
2435
                                     network->def->name);
2436 2437 2438 2439 2440 2441
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2442
    switch (network->def->forward.type) {
2443 2444 2445 2446

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2447
        ret = networkShutdownNetworkVirtual(network);
2448 2449 2450
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2451 2452 2453
        ret = networkShutdownNetworkBridge(network);
        break;

2454 2455 2456
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2457
    case VIR_NETWORK_FORWARD_HOSTDEV:
2458
        ret = networkShutdownNetworkExternal(network);
2459 2460 2461
        break;
    }

2462
    /* now that we know it's stopped call the hook if present */
2463
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2464 2465
                   VIR_HOOK_SUBOP_END);

2466
    network->active = 0;
2467
    virNetworkObjUnsetDefTransient(network);
2468
    return ret;
2469 2470 2471
}


2472
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2473 2474
                                         const unsigned char *uuid)
{
2475 2476
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2477

2478
    networkDriverLock();
2479
    network = virNetworkObjFindByUUID(driver->networks, uuid);
2480
    networkDriverUnlock();
2481
    if (!network) {
2482 2483
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2484
        virReportError(VIR_ERR_NO_NETWORK,
2485 2486
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2487
        goto cleanup;
2488 2489
    }

2490 2491 2492
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2493 2494
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2495
 cleanup:
2496 2497
    if (network)
        virNetworkObjUnlock(network);
2498
    return ret;
2499 2500
}

2501
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2502 2503
                                         const char *name)
{
2504 2505 2506
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2507
    networkDriverLock();
2508
    network = virNetworkFindByName(driver->networks, name);
2509
    networkDriverUnlock();
2510
    if (!network) {
2511 2512
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2513
        goto cleanup;
2514 2515
    }

2516 2517 2518
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2519 2520
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2521
 cleanup:
2522 2523
    if (network)
        virNetworkObjUnlock(network);
2524
    return ret;
2525 2526
}

2527 2528
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2529 2530
    int nactive = 0;
    size_t i;
2531

2532 2533 2534
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2535
    networkDriverLock();
2536 2537
    for (i = 0; i < driver->networks->count; i++) {
        virNetworkObjPtr obj = driver->networks->objs[i];
2538 2539 2540
        virNetworkObjLock(obj);
        if (virConnectNumOfNetworksCheckACL(conn, obj->def) &&
            virNetworkObjIsActive(obj))
2541
            nactive++;
2542
        virNetworkObjUnlock(obj);
2543
    }
2544
    networkDriverUnlock();
2545

2546 2547 2548
    return nactive;
}

2549
static int networkConnectListNetworks(virConnectPtr conn, char **const names, int nnames) {
2550 2551
    int got = 0;
    size_t i;
2552

2553 2554 2555
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2556
    networkDriverLock();
2557 2558
    for (i = 0; i < driver->networks->count && got < nnames; i++) {
        virNetworkObjPtr obj = driver->networks->objs[i];
2559 2560 2561 2562 2563
        virNetworkObjLock(obj);
        if (virConnectListNetworksCheckACL(conn, obj->def) &&
            virNetworkObjIsActive(obj)) {
            if (VIR_STRDUP(names[got], obj->def->name) < 0) {
                virNetworkObjUnlock(obj);
2564 2565 2566 2567
                goto cleanup;
            }
            got++;
        }
2568
        virNetworkObjUnlock(obj);
2569
    }
2570
    networkDriverUnlock();
2571

2572 2573 2574
    return got;

 cleanup:
2575
    networkDriverUnlock();
2576
    for (i = 0; i < got; i++)
2577 2578 2579 2580
        VIR_FREE(names[i]);
    return -1;
}

2581 2582
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2583 2584
    int ninactive = 0;
    size_t i;
2585

2586 2587 2588
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2589
    networkDriverLock();
2590 2591
    for (i = 0; i < driver->networks->count; i++) {
        virNetworkObjPtr obj = driver->networks->objs[i];
2592 2593 2594
        virNetworkObjLock(obj);
        if (virConnectNumOfDefinedNetworksCheckACL(conn, obj->def) &&
            !virNetworkObjIsActive(obj))
2595
            ninactive++;
2596
        virNetworkObjUnlock(obj);
2597
    }
2598
    networkDriverUnlock();
2599

2600 2601 2602
    return ninactive;
}

2603
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames) {
2604 2605
    int got = 0;
    size_t i;
2606

2607 2608 2609
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2610
    networkDriverLock();
2611 2612
    for (i = 0; i < driver->networks->count && got < nnames; i++) {
        virNetworkObjPtr obj = driver->networks->objs[i];
2613 2614 2615 2616 2617
        virNetworkObjLock(obj);
        if (virConnectListDefinedNetworksCheckACL(conn, obj->def) &&
            !virNetworkObjIsActive(obj)) {
            if (VIR_STRDUP(names[got], obj->def->name) < 0) {
                virNetworkObjUnlock(obj);
2618 2619 2620 2621
                goto cleanup;
            }
            got++;
        }
2622
        virNetworkObjUnlock(obj);
2623
    }
2624
    networkDriverUnlock();
2625 2626 2627
    return got;

 cleanup:
2628
    networkDriverUnlock();
2629
    for (i = 0; i < got; i++)
2630 2631 2632 2633
        VIR_FREE(names[i]);
    return -1;
}

2634
static int
2635 2636 2637
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2638 2639 2640 2641 2642
{
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2643 2644 2645
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2646
    networkDriverLock();
2647
    ret = virNetworkObjListExport(conn, driver->networks, nets,
2648 2649
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2650
    networkDriverUnlock();
2651

2652
 cleanup:
2653 2654
    return ret;
}
2655

2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669
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,
2670
                                       net, eventID, callback,
2671 2672 2673
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2674
 cleanup:
2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686
    return ret;
}

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

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

2687 2688 2689 2690 2691 2692
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
2693

2694
 cleanup:
2695 2696 2697
    return ret;
}

2698 2699 2700 2701 2702
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2703 2704
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2705 2706 2707 2708

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

2709 2710
    ret = virNetworkObjIsActive(obj);

2711
 cleanup:
2712 2713 2714 2715 2716 2717 2718 2719 2720 2721
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}

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

2722 2723
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2724 2725 2726 2727

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

2728 2729
    ret = obj->persistent;

2730
 cleanup:
2731 2732 2733 2734 2735 2736
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}


2737
static int
2738
networkValidate(virNetworkDefPtr def,
2739
                bool check_active)
2740
{
2741
    size_t i, j;
2742 2743
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2744
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
2745
    bool ipv4def = false, ipv6def = false;
2746
    bool bandwidthAllowed = true;
2747
    bool usesInterface = false, usesAddress = false;
2748 2749

    /* check for duplicate networks */
2750
    if (virNetworkObjIsDuplicate(driver->networks, def, check_active) < 0)
2751 2752 2753 2754 2755
        return -1;

    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
2756 2757 2758
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2759

2760
        if (virNetworkSetBridgeName(driver->networks, def, 1))
2761 2762 2763
            return -1;

        virNetworkSetBridgeMacAddr(def);
2764 2765
    } else {
        /* They are also the only types that currently support setting
2766 2767
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
2768
         */
2769 2770 2771 2772 2773 2774 2775 2776
        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;
        }
2777 2778 2779 2780 2781
        if (virNetworkDefGetIpByIndex(def, AF_UNSPEC, 0)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2782
                           virNetworkForwardTypeToString(def->forward.type));
2783 2784
            return -1;
        }
2785
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
2786 2787 2788 2789
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2790
                           virNetworkForwardTypeToString(def->forward.type));
2791 2792 2793 2794 2795 2796 2797
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2798
                           virNetworkForwardTypeToString(def->forward.type));
2799 2800
            return -1;
        }
2801 2802 2803 2804 2805 2806 2807 2808
        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;
        }
2809
        bandwidthAllowed = false;
2810 2811
    }

2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845
    /* we support configs with a single PF defined:
     *   <pf dev='eth0'/>
     * or with a list of netdev names:
     *   <interface dev='eth9'/>
     * OR a list of PCI addresses
     *   <address type='pci' domain='0' bus='4' slot='0' function='1'/>
     * but not any combination of those.
     *
     * Since <interface> and <address> are for some strange reason
     * stored in the same array, we need to cycle through it and check
     * the type of each.
     */
    for (i = 0; i < def->forward.nifs; i++) {
        switch ((virNetworkForwardHostdevDeviceType)
                def->forward.ifs[i].type) {
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV:
            usesInterface = true;
            break;
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI:
            usesAddress = true;
            break;
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NONE:
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_LAST:
            break;
        }
    }
    if ((def->forward.npfs > 0) + usesInterface + usesAddress > 1) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<address>, <interface>, and <pf> elements of "
                         "<forward> in network %s are mutually exclusive"),
                       def->name);
        return -1;
    }

G
Gene Czarcinski 已提交
2846 2847 2848
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
2849 2850 2851
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
2852 2853 2854 2855 2856
        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 -- "
2857 2858
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875
                    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;
                }
2876 2877 2878
            }
        }
    }
2879 2880 2881 2882 2883 2884

    /* 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.
     */

2885
    vlanAllowed = ((def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
J
Ján Tomko 已提交
2886 2887
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
2888 2889
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) ||
                   def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV);
2890 2891

    vlanUsed = def->vlan.nTags > 0;
2892 2893
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
2894 2895 2896 2897 2898
            /* 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.
             */
2899
            if (def->portGroups[i].virtPortProfile) {
2900
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
2901
                    def->portGroups[i].virtPortProfile->virtPortType
2902 2903 2904 2905 2906 2907 2908
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
2909
        }
2910
        if (def->portGroups[i].isDefault) {
2911 2912 2913 2914 2915
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
2916
                               def->name, defaultPortGroup->name,
2917
                               def->portGroups[i].name);
2918
                return -1;
2919
            }
2920
            defaultPortGroup = &def->portGroups[i];
2921
        }
2922 2923 2924 2925 2926 2927 2928 2929 2930
        for (j = i + 1; j < def->nPortGroups; j++) {
            if (STREQ(def->portGroups[i].name, def->portGroups[j].name)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("multiple <portgroup> elements with the "
                                 "same name (%s) in network '%s'"),
                               def->portGroups[i].name, def->name);
                return -1;
            }
        }
2931 2932 2933 2934 2935 2936 2937 2938
        if (def->portGroups[i].bandwidth && !bandwidthAllowed) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <bandwidth> element in network '%s' "
                             "in portgroup '%s' with forward mode='%s'"),
                           def->name, def->portGroups[i].name,
                           virNetworkForwardTypeToString(def->forward.type));
            return -1;
        }
2939
    }
2940 2941 2942 2943 2944 2945 2946
    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.
         */
2947 2948 2949 2950 2951 2952 2953 2954 2955
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
    return 0;
}

2956 2957
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
2958
    virNetworkDefPtr def;
2959
    virNetworkObjPtr network = NULL;
2960
    virNetworkPtr ret = NULL;
2961
    virObjectEventPtr event = NULL;
2962

2963
    networkDriverLock();
2964

2965
    if (!(def = virNetworkDefParseString(xml)))
2966
        goto cleanup;
2967

2968 2969 2970
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

2971
    if (networkValidate(def, true) < 0)
J
Ján Tomko 已提交
2972
        goto cleanup;
2973

2974 2975 2976
    /* 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.
2977
     */
2978
    if (!(network = virNetworkAssignDef(driver->networks, def, true)))
2979 2980
        goto cleanup;
    def = NULL;
2981

2982
    if (networkStartNetwork(network) < 0) {
2983
        virNetworkRemoveInactive(driver->networks,
2984
                                 network);
2985
        network = NULL;
2986
        goto cleanup;
2987 2988
    }

2989 2990
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
2991 2992
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
2993

2994
    VIR_INFO("Creating network '%s'", network->def->name);
2995 2996
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2997
 cleanup:
2998
    virNetworkDefFree(def);
2999 3000
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3001 3002
    if (network)
        virNetworkObjUnlock(network);
3003
    networkDriverUnlock();
3004
    return ret;
3005 3006
}

3007 3008
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
3009
    virNetworkDefPtr def = NULL;
3010
    bool freeDef = true;
3011
    virNetworkObjPtr network = NULL;
3012
    virNetworkPtr ret = NULL;
3013
    virObjectEventPtr event = NULL;
3014

3015
    networkDriverLock();
3016

3017
    if (!(def = virNetworkDefParseString(xml)))
3018
        goto cleanup;
3019

3020 3021 3022
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3023
    if (networkValidate(def, false) < 0)
J
Ján Tomko 已提交
3024
        goto cleanup;
3025

3026
    if (!(network = virNetworkAssignDef(driver->networks, def, false)))
J
Ján Tomko 已提交
3027
        goto cleanup;
3028

3029
    /* def was assigned to network object */
3030
    freeDef = false;
3031 3032

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3033
        if (!virNetworkObjIsActive(network)) {
3034
            virNetworkRemoveInactive(driver->networks, network);
3035 3036 3037
            network = NULL;
            goto cleanup;
        }
3038 3039 3040 3041 3042
        /* 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);
3043 3044 3045
        goto cleanup;
    }

3046
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3047 3048
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3049

3050 3051
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
3052

3053
 cleanup:
3054 3055
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3056
    if (freeDef)
J
Ján Tomko 已提交
3057
        virNetworkDefFree(def);
3058 3059
    if (network)
        virNetworkObjUnlock(network);
3060
    networkDriverUnlock();
3061
    return ret;
3062 3063
}

3064
static int
3065 3066
networkUndefine(virNetworkPtr net)
{
3067
    virNetworkObjPtr network;
3068
    int ret = -1;
3069
    bool active = false;
3070
    virObjectEventPtr event = NULL;
3071

3072
    networkDriverLock();
3073

3074
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
3075
    if (!network) {
3076 3077
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3078
        goto cleanup;
3079 3080
    }

3081 3082 3083
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3084 3085
    if (virNetworkObjIsActive(network))
        active = true;
3086

3087
    /* remove autostart link */
3088
    if (virNetworkDeleteConfig(driver->networkConfigDir,
3089 3090
                               driver->networkAutostartDir,
                               network) < 0)
3091
        goto cleanup;
3092
    network->autostart = 0;
3093

3094 3095
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3096 3097
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3098

3099
    VIR_INFO("Undefining network '%s'", network->def->name);
3100
    if (!active) {
3101
        if (networkRemoveInactive(network) < 0) {
3102 3103 3104
            network = NULL;
            goto cleanup;
        }
3105
        network = NULL;
3106 3107 3108 3109 3110 3111
    } 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);
3112 3113
    }

3114
    ret = 0;
3115

3116
 cleanup:
3117 3118
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3119 3120
    if (network)
        virNetworkObjUnlock(network);
3121
    networkDriverUnlock();
3122
    return ret;
3123 3124
}

3125 3126 3127 3128 3129 3130 3131 3132 3133
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
    virNetworkObjPtr network = NULL;
3134 3135
    int isActive, ret = -1;
    size_t i;
3136 3137
    virNetworkIpDefPtr ipdef;
    bool oldDhcpActive = false;
3138
    bool needFirewallRefresh = false;
3139

3140 3141 3142 3143 3144

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3145
    networkDriverLock();
3146

3147
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
3148 3149 3150 3151 3152 3153
    if (!network) {
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
        goto cleanup;
    }

3154 3155 3156
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3157
    /* see if we are listening for dhcp pre-modification */
3158 3159 3160
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
3161 3162 3163 3164 3165 3166
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3167 3168
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3169
     */
3170
    isActive = virNetworkObjIsActive(network);
3171 3172
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3173 3174 3175 3176 3177 3178 3179
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196
    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).
                 */
3197
                networkRemoveFirewallRules(network->def);
3198 3199 3200 3201 3202 3203 3204 3205
                needFirewallRefresh = true;
                break;
            default:
                break;
            }
        }
    }

3206
    /* update the network config in memory/on disk */
3207 3208
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3209
            ignore_value(networkAddFirewallRules(network->def));
3210 3211 3212
        goto cleanup;
    }

3213
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236
        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.
             */
3237
            if (networkRestartDhcpDaemon(network) < 0)
3238 3239
                goto cleanup;

3240 3241 3242 3243 3244 3245 3246 3247
        } 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;

3248 3249 3250
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
                 i++) {
3251 3252 3253 3254 3255 3256 3257
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3258 3259
                 networkRestartDhcpDaemon(network) < 0) ||
                networkRefreshDhcpDaemon(network) < 0) {
3260 3261 3262 3263
                goto cleanup;
            }

        } else if (section == VIR_NETWORK_SECTION_DNS_HOST ||
3264 3265 3266 3267 3268 3269
                   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.
             */
3270
            if (networkRefreshDhcpDaemon(network) < 0)
3271 3272 3273 3274 3275 3276 3277 3278
                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.
             */
3279
            if (networkRefreshRadvd(network) < 0)
3280 3281 3282 3283
                goto cleanup;
        }

        /* save current network state to disk */
3284
        if ((ret = virNetworkSaveStatus(driver->stateDir,
3285
                                        network)) < 0) {
3286
            goto cleanup;
3287
        }
3288 3289
    }
    ret = 0;
3290
 cleanup:
3291 3292
    if (network)
        virNetworkObjUnlock(network);
3293
    networkDriverUnlock();
3294 3295 3296
    return ret;
}

3297 3298
static int networkCreate(virNetworkPtr net)
{
3299 3300
    virNetworkObjPtr network;
    int ret = -1;
3301
    virObjectEventPtr event = NULL;
3302

3303
    networkDriverLock();
3304
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
3305

3306
    if (!network) {
3307 3308
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3309
        goto cleanup;
3310 3311
    }

3312 3313 3314
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3315
    if ((ret = networkStartNetwork(network)) < 0)
3316
        goto cleanup;
3317

3318 3319
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3320 3321
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3322

3323
 cleanup:
3324 3325
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3326 3327
    if (network)
        virNetworkObjUnlock(network);
3328
    networkDriverUnlock();
3329
    return ret;
3330 3331
}

3332 3333
static int networkDestroy(virNetworkPtr net)
{
3334 3335
    virNetworkObjPtr network;
    int ret = -1;
3336
    virObjectEventPtr event = NULL;
3337

3338
    networkDriverLock();
3339
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
3340

3341
    if (!network) {
3342 3343
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3344
        goto cleanup;
3345 3346
    }

3347 3348 3349
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3350
    if (!virNetworkObjIsActive(network)) {
3351
        virReportError(VIR_ERR_OPERATION_INVALID,
3352 3353
                       _("network '%s' is not active"),
                       network->def->name);
3354 3355 3356
        goto cleanup;
    }

3357
    if ((ret = networkShutdownNetwork(network)) < 0)
3358 3359
        goto cleanup;

3360 3361
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3362 3363
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3364

3365
    if (!network->persistent) {
3366
        if (networkRemoveInactive(network) < 0) {
3367 3368 3369 3370
            network = NULL;
            ret = -1;
            goto cleanup;
        }
3371 3372
        network = NULL;
    }
3373

3374
 cleanup:
3375 3376
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3377 3378
    if (network)
        virNetworkObjUnlock(network);
3379
    networkDriverUnlock();
3380 3381 3382
    return ret;
}

3383
static char *networkGetXMLDesc(virNetworkPtr net,
3384
                               unsigned int flags)
3385
{
3386
    virNetworkObjPtr network;
3387
    virNetworkDefPtr def;
3388
    char *ret = NULL;
3389

3390
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3391

3392 3393
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3394

3395 3396 3397
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3398 3399 3400 3401 3402 3403
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3404

3405
 cleanup:
3406 3407
    if (network)
        virNetworkObjUnlock(network);
3408
    return ret;
3409 3410 3411
}

static char *networkGetBridgeName(virNetworkPtr net) {
3412 3413 3414
    virNetworkObjPtr network;
    char *bridge = NULL;

3415 3416
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3417

3418 3419 3420
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3421
    if (!(network->def->bridge)) {
3422 3423 3424
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3425 3426 3427
        goto cleanup;
    }

3428
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3429

3430
 cleanup:
3431 3432
    if (network)
        virNetworkObjUnlock(network);
3433 3434 3435 3436
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
J
Ján Tomko 已提交
3437
                               int *autostart)
3438
{
3439 3440
    virNetworkObjPtr network;
    int ret = -1;
3441

3442 3443
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3444

3445 3446 3447
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3448
    *autostart = network->autostart;
3449
    ret = 0;
3450

3451
 cleanup:
3452 3453
    if (network)
        virNetworkObjUnlock(network);
3454
    return ret;
3455 3456 3457
}

static int networkSetAutostart(virNetworkPtr net,
3458 3459
                               int autostart)
{
3460
    virNetworkObjPtr network;
3461
    char *configFile = NULL, *autostartLink = NULL;
3462
    int ret = -1;
3463

3464
    networkDriverLock();
3465
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
3466

3467
    if (!network) {
3468 3469
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3470
        goto cleanup;
3471 3472
    }

3473 3474 3475
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3476
    if (!network->persistent) {
3477 3478
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3479 3480 3481
        goto cleanup;
    }

3482 3483
    autostart = (autostart != 0);

3484
    if (network->autostart != autostart) {
3485
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3486
            goto cleanup;
3487
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3488 3489
            goto cleanup;

3490
        if (autostart) {
3491
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3492
                virReportSystemError(errno,
3493 3494
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3495 3496
                goto cleanup;
            }
3497

3498
            if (symlink(configFile, autostartLink) < 0) {
3499
                virReportSystemError(errno,
3500
                                     _("Failed to create symlink '%s' to '%s'"),
3501
                                     autostartLink, configFile);
3502 3503 3504
                goto cleanup;
            }
        } else {
3505
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3506
                virReportSystemError(errno,
3507
                                     _("Failed to delete symlink '%s'"),
3508
                                     autostartLink);
3509 3510
                goto cleanup;
            }
3511 3512
        }

3513
        network->autostart = autostart;
3514
    }
3515
    ret = 0;
3516

3517
 cleanup:
3518 3519
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3520 3521
    if (network)
        virNetworkObjUnlock(network);
3522
    networkDriverUnlock();
3523
    return ret;
3524 3525
}

3526
static int
3527 3528 3529 3530
networkGetDHCPLeases(virNetworkPtr network,
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549
{
    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;
3550 3551 3552 3553 3554 3555 3556 3557 3558
    virNetworkObjPtr obj;

    virCheckFlags(0, -1);

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

    if (virNetworkGetDHCPLeasesEnsureACL(network->conn, obj->def) < 0)
        goto cleanup;
3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 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

    /* 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;
        }

3605
        if (mac && virMacAddrCompare(mac, mac_tmp))
3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655
            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) ||
3656
                (VIR_STRDUP(lease->iface, obj->def->bridge) < 0))
3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688
                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 (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);
3689
    VIR_FREE(lease_entries);
3690 3691
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
3692 3693 3694 3695

    if (obj)
        virNetworkObjUnlock(obj);

3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706
    return rv;

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

3707 3708

static virNetworkDriver networkDriver = {
3709
    .name = "bridge",
3710 3711 3712 3713 3714
    .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 */
3715 3716
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
3717 3718
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3719 3720
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3721
    .networkUndefine = networkUndefine, /* 0.2.0 */
3722
    .networkUpdate = networkUpdate, /* 0.10.2 */
3723
    .networkCreate = networkCreate, /* 0.2.0 */
3724 3725 3726 3727 3728 3729 3730
    .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 */
3731
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
3732 3733 3734
};

static virStateDriver networkStateDriver = {
3735
    .name = "bridge",
3736
    .stateInitialize  = networkStateInitialize,
3737
    .stateAutoStart  = networkStateAutoStart,
3738 3739
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3740 3741
};

3742 3743
int networkRegister(void)
{
3744
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
3745
        return -1;
3746 3747
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
3748 3749
    return 0;
}
3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761

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

/* 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:
3762
 * @dom: domain definition that @iface belongs to
3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773
 * @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
3774 3775
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
3776
{
3777
    virDomainNetType actualType = iface->type;
3778 3779
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
3780
    virNetDevBandwidthPtr bandwidth = NULL;
3781 3782 3783
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3784
    virNetworkForwardIfDefPtr dev = NULL;
3785
    size_t i;
3786 3787 3788
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3789
        goto validate;
3790 3791 3792 3793

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

3794
    networkDriverLock();
3795
    network = virNetworkFindByName(driver->networks, iface->data.network.name);
3796
    networkDriverUnlock();
3797
    if (!network) {
3798 3799 3800
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3801
        goto error;
3802 3803
    }
    netdef = network->def;
3804

3805 3806 3807 3808 3809 3810 3811
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

3812 3813 3814
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

3815 3816 3817
    /* 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
3818
     */
3819 3820 3821 3822 3823 3824
    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.
     */
3825 3826 3827 3828 3829 3830

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

3831 3832
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
3833
        goto error;
3834

3835 3836 3837 3838 3839 3840 3841 3842
    /* 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;

3843 3844
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
3845

3846 3847 3848 3849 3850 3851 3852 3853 3854 3855
    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;

3856 3857 3858
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
3859 3860 3861
        /* 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
3862
         */
3863
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
3864

3865
        /* we also store the bridge device and macTableManager settings
3866 3867 3868 3869 3870 3871 3872
         * in iface->data.network.actual->data.bridge for later use
         * after the domain's tap device is created (to attach to the
         * bridge and set flood/learning mode on the tap device)
         */
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
            goto error;
3873 3874
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
3875

3876 3877 3878
        if (networkPlugBandwidth(network, iface) < 0)
            goto error;

3879
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
3880
               netdef->bridge) {
3881 3882 3883 3884 3885

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

3886
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
3887 3888
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
3889
            goto error;
3890 3891
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
3892

3893 3894 3895 3896 3897 3898 3899 3900
        /* 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) {
3901
            goto error;
3902 3903 3904 3905 3906 3907 3908 3909 3910 3911
        }
        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);
3912
                goto error;
3913 3914 3915
            }
        }

3916
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3917

3918
        virDomainHostdevSubsysPCIBackendType backend;
3919

3920
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
3921
        if (networkCreateInterfacePool(netdef) < 0)
3922 3923 3924
            goto error;

        /* pick first dev with 0 connections */
3925 3926 3927
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941
                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;
3942
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
3943
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
3944
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
3945

E
Eric Blake 已提交
3946
        switch (netdef->forward.driverName) {
3947
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
3948
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
3949 3950
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
3951
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
3952 3953
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
3954
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965
            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;

3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990
        /* 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;
            }
        }

3991 3992 3993 3994
    } 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)) {
3995 3996 3997 3998 3999 4000

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
4001
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4002
        switch (netdef->forward.type) {
4003
        case VIR_NETWORK_FORWARD_BRIDGE:
4004
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
4005 4006
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
4007
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
4008 4009
            break;
        case VIR_NETWORK_FORWARD_VEPA:
4010
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
4011 4012
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
4013
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
4014 4015 4016
            break;
        }

4017 4018 4019 4020 4021 4022 4023 4024
        /* 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) {
4025
            goto error;
4026
        }
4027
        virtport = iface->data.network.actual->virtPortProfile;
4028
        if (virtport) {
4029 4030 4031 4032 4033 4034 4035 4036
            /* 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);
4037
                goto error;
4038 4039
            }
        }
4040

4041 4042 4043
        /* 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).
         */
4044
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4045 4046 4047 4048
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4049
            goto error;
4050 4051 4052
        } else {
            /* pick an interface from the pool */

4053
            if (networkCreateInterfacePool(netdef) < 0)
4054 4055
                goto error;

4056 4057 4058 4059 4060
            /* 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.
4061
             */
4062 4063
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4064 4065 4066
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4067

4068
                /* pick first dev with 0 connections */
4069 4070 4071
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4072 4073 4074 4075 4076
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4077
                dev = &netdef->forward.ifs[0];
4078 4079 4080
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4081 4082 4083 4084
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4085 4086 4087 4088
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4089
                goto error;
4090
            }
4091 4092
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4093
                goto error;
4094 4095 4096
        }
    }

4097
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4098
        goto error;
4099

4100
 validate:
4101 4102 4103 4104 4105
    /* 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.
     */

4106
    if (virDomainNetGetActualVlan(iface)) {
4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135
        /* 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);
4136

4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155
        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 */
4156
            netdef->connections--;
4157 4158 4159 4160
            if (dev)
                dev->connections--;
            goto error;
        }
4161 4162
    }

4163
    ret = 0;
4164

4165
 cleanup:
4166 4167
    if (network)
        virNetworkObjUnlock(network);
4168 4169
    return ret;

4170
 error:
4171
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4172 4173 4174
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4175
    goto cleanup;
4176 4177 4178
}

/* networkNotifyActualDevice:
4179
 * @dom: domain definition that @iface belongs to
4180 4181 4182 4183 4184 4185 4186 4187 4188 4189
 * @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
4190 4191
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4192
{
4193
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4194 4195
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4196
    virNetworkForwardIfDefPtr dev = NULL;
4197 4198
    size_t i;
    int ret = -1;
4199 4200

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4201
        return 0;
4202

4203
    networkDriverLock();
4204
    network = virNetworkFindByName(driver->networks, iface->data.network.name);
4205
    networkDriverUnlock();
4206
    if (!network) {
4207 4208 4209
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4210 4211 4212 4213
        goto error;
    }
    netdef = network->def;

4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224
    /* if we're restarting libvirtd after an upgrade from a version
     * that didn't save bridge name in actualNetDef for
     * actualType==network, we need to copy it in so that it will be
     * available in all cases
     */
    if (actualType == VIR_DOMAIN_NET_TYPE_NETWORK &&
        !iface->data.network.actual->data.bridge.brname &&
        (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                    netdef->bridge) < 0))
            goto error;

4225
    if (!iface->data.network.actual ||
4226 4227
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4228 4229
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4230 4231
    }

4232
    if (networkCreateInterfacePool(netdef) < 0)
4233
        goto error;
4234

4235
    if (netdef->forward.nifs == 0) {
4236
        virReportError(VIR_ERR_INTERNAL_ERROR,
4237 4238
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4239
                       netdef->name);
4240
        goto error;
4241
    }
4242

4243 4244
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4245

4246 4247 4248 4249 4250 4251 4252 4253 4254
        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 */
4255 4256
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4257
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4258 4259
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4260 4261 4262 4263 4264
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4265
            virReportError(VIR_ERR_INTERNAL_ERROR,
4266 4267
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4268
                           netdef->name, actualDev);
4269
            goto error;
4270 4271
        }

4272
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4273 4274
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4275
         */
4276
        if ((dev->connections > 0) &&
4277 4278
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4279 4280
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4281
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4282
            virReportError(VIR_ERR_INTERNAL_ERROR,
4283 4284
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4285
                           netdef->name, actualDev);
4286
            goto error;
4287
        }
4288

4289
        /* we are now assured of success, so mark the allocation */
4290
        dev->connections++;
4291
        VIR_DEBUG("Using physical device %s, connections %d",
4292
                  dev->device.dev, dev->connections);
4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305

    }  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 */
4306 4307
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4308
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4309
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4310 4311
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4312 4313 4314 4315 4316 4317 4318 4319 4320
                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,
4321 4322 4323 4324
                           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
4325
            goto error;
4326 4327 4328 4329 4330 4331 4332
        }

        /* 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) &&
4333
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349
            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);
4350 4351
    }

4352
 success:
4353 4354 4355
    netdef->connections++;
    VIR_DEBUG("Using network %s, %d connections",
              netdef->name, netdef->connections);
4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366

    /* 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;
    }

4367
    ret = 0;
4368
 cleanup:
4369 4370 4371
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4372

4373
 error:
4374
    goto cleanup;
4375 4376 4377 4378
}


/* networkReleaseActualDevice:
4379
 * @dom: domain definition that @iface belongs to
4380 4381 4382 4383 4384 4385 4386 4387 4388 4389
 * @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
4390 4391
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4392
{
4393
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4394
    virNetworkObjPtr network;
4395
    virNetworkDefPtr netdef;
4396
    virNetworkForwardIfDefPtr dev = NULL;
4397 4398
    size_t i;
    int ret = -1;
4399 4400

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4401
        return 0;
4402

4403
    networkDriverLock();
4404
    network = virNetworkFindByName(driver->networks, iface->data.network.name);
4405
    networkDriverUnlock();
4406
    if (!network) {
4407 4408 4409
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4410 4411 4412 4413
        goto error;
    }
    netdef = network->def;

4414 4415
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4416 4417 4418 4419 4420
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) &&
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4421 4422 4423
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4424 4425
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4426 4427
    }

4428
    if (netdef->forward.nifs == 0) {
4429
        virReportError(VIR_ERR_INTERNAL_ERROR,
4430
                       _("network '%s' uses a direct/hostdev mode, but "
4431 4432
                         "has no forward dev and no interface pool"),
                       netdef->name);
4433
        goto error;
4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445
    }

    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;
        }
4446

4447 4448
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4449
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4450 4451
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4452 4453 4454
                break;
            }
        }
4455

4456
        if (!dev) {
4457
            virReportError(VIR_ERR_INTERNAL_ERROR,
4458 4459
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4460
                           netdef->name, actualDev);
4461
            goto error;
4462 4463
        }

4464
        dev->connections--;
4465
        VIR_DEBUG("Releasing physical device %s, connections %d",
4466
                  dev->device.dev, dev->connections);
4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477

    } 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;
        }

4478 4479
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4480
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4481
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4482 4483
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4484 4485 4486 4487 4488 4489 4490 4491 4492
                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,
4493 4494 4495 4496
                           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
4497
            goto error;
4498 4499 4500 4501 4502 4503 4504
        }

        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
4505
    }
4506

4507
 success:
4508
    if (iface->data.network.actual) {
4509
        netdef->connections--;
4510 4511
        VIR_DEBUG("Releasing network %s, %d connections",
                  netdef->name, netdef->connections);
4512

4513 4514 4515 4516
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
    }
4517
    ret = 0;
4518
 cleanup:
4519 4520
    if (network)
        virNetworkObjUnlock(network);
4521 4522 4523 4524
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4525
    return ret;
4526

4527
 error:
4528
    goto cleanup;
4529
}
4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553

/*
 * 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;
4554
    virNetworkObjPtr network;
4555 4556 4557 4558
    virNetworkDefPtr netdef;
    virNetworkIpDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4559
    char *dev_name = NULL;
4560 4561

    *netaddr = NULL;
4562
    networkDriverLock();
4563
    network = virNetworkFindByName(driver->networks, netname);
4564
    networkDriverUnlock();
4565
    if (!network) {
4566 4567 4568
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4569
        goto cleanup;
4570 4571 4572
    }
    netdef = network->def;

4573
    switch (netdef->forward.type) {
4574 4575 4576 4577 4578 4579
    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) {
4580 4581 4582
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have an IPv4 address"),
                           netdef->name);
4583
            goto cleanup;
4584 4585 4586 4587 4588
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4589
        if ((dev_name = netdef->bridge))
4590 4591 4592 4593 4594 4595 4596 4597
            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:
4598 4599
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4600

4601
        if (!dev_name) {
4602 4603 4604
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4605
            goto cleanup;
4606 4607 4608 4609
        }
        break;
    }

4610
    if (dev_name) {
4611
        if (virNetDevGetIPv4Address(dev_name, &addr) < 0)
4612
            goto cleanup;
4613
        addrptr = &addr;
4614 4615
    }

4616 4617
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
4618
        goto cleanup;
4619 4620
    }

4621
    ret = 0;
4622
 cleanup:
4623 4624 4625 4626
    if (network)
        virNetworkObjUnlock(network);
    return ret;
}
4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644

/**
 * 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;
4645
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4646 4647 4648 4649
    unsigned long long tmp_floor_sum = net->floor_sum;
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660
    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;
    }

4661
    if (!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor ||
4662 4663
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4664
        return 1;
4665
    }
4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696

    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;

4697
 cleanup:
4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734
    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];
4735
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763

    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;
    }

4764 4765
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
4766 4767 4768 4769 4770 4771 4772 4773
    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 */
4774
    net->floor_sum += ifaceBand->in->floor;
4775
    /* update status file */
4776
    if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
4777
        ignore_value(virBitmapClearBit(net->class_id, class_id));
4778
        net->floor_sum -= ifaceBand->in->floor;
4779 4780 4781 4782
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
4783 4784 4785 4786 4787 4788 4789 4790 4791
    /* 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;

4792
 cleanup:
4793 4794 4795 4796 4797 4798 4799 4800 4801
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
    int ret = 0;
    unsigned long long new_rate;
4802
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4803 4804 4805

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
4806 4807 4808 4809 4810
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821
        /* 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 */
4822
        net->floor_sum -= ifaceBand->in->floor;
4823 4824 4825 4826
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
4827
        if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
4828
            net->floor_sum += ifaceBand->in->floor;
4829 4830 4831 4832
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
4833 4834 4835 4836 4837 4838 4839 4840 4841 4842
        /* 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;
    }

4843
 cleanup:
4844 4845
    return ret;
}
4846 4847 4848

static void
networkNetworkObjTaint(virNetworkObjPtr net,
4849
                       virNetworkTaintFlags taint)
4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860
{
    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));
    }
}