bridge_driver.c 157.7 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
static int
networkUpdateState(virNetworkObjPtr obj,
                   void *opaque ATTRIBUTE_UNUSED)
348
{
349
    int ret = -1;
350

351 352 353 354 355
    virNetworkObjLock(obj);
    if (!virNetworkObjIsActive(obj)) {
        virNetworkObjUnlock(obj);
        return 0;
    }
356

357 358 359 360 361 362 363 364
    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;
365

366 367
    case VIR_NETWORK_FORWARD_BRIDGE:
        if (!(obj->def->bridge && virNetDevExists(obj->def->bridge) == 1)) {
368 369
                obj->active = 0;
            break;
370
        }
371 372 373 374 375 376 377 378 379
        /* 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;
380

381 382 383
    case VIR_NETWORK_FORWARD_HOSTDEV:
        /* so far no extra checks */
        break;
384
    }
385

386 387 388
    /* Try and read dnsmasq/radvd pids of active networks */
    if (obj->active && obj->def->ips && (obj->def->nips > 0)) {
        char *radvdpidbase;
389

390 391 392 393 394 395 396
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           obj->def->name,
                                           &obj->dnsmasqPid,
                                           dnsmasqCapsGetBinaryPath(driver->dnsmasqCaps)));
        radvdpidbase = networkRadvdPidfileBasename(obj->def->name);
        if (!radvdpidbase)
            goto cleanup;
397

398 399 400 401
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           radvdpidbase,
                                           &obj->radvdPid, RADVD));
        VIR_FREE(radvdpidbase);
402
    }
403

404 405 406 407 408
    ret = 0;
 cleanup:
    virNetworkObjUnlock(obj);
    return ret;
}
409

410 411 412
static int
networkAutostartConfig(virNetworkObjPtr net,
                       void *opaque ATTRIBUTE_UNUSED)
413
{
414
    int ret = -1;
415

416 417 418 419 420 421 422 423 424 425
    virNetworkObjLock(net);
    if (net->autostart &&
        !virNetworkObjIsActive(net) &&
        networkStartNetwork(net) < 0)
        goto cleanup;

    ret = 0;
 cleanup:
    virNetworkObjUnlock(net);
    return ret;
426 427
}

428 429 430
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
431
                             DBusMessage *message, void *user_data ATTRIBUTE_UNUSED)
432
{
433 434 435 436 437 438
    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.");
439
        networkReloadFirewallRules();
440 441 442 443 444 445
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

446
static int
447
networkMigrateStateFiles(void)
448 449 450 451 452 453 454 455 456 457 458 459 460
{
    /* 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;
461
    int direrr;
462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480
    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;
    }

481
    while ((direrr = virDirRead(dir, &entry, oldStateDir)) > 0) {
482 483 484
        if (entry->d_type != DT_UNKNOWN &&
            entry->d_type != DT_REG)
            continue;
485

486
        if (STREQ(entry->d_name, ".") ||
487 488 489 490 491
            STREQ(entry->d_name, ".."))
            continue;

        if (virAsprintf(&oldPath, "%s/%s",
                        oldStateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
492
            goto cleanup;
493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509

        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;
            }
        }

510
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
J
Ján Tomko 已提交
511
            goto cleanup;
512 513 514

        if (virAsprintf(&newPath, "%s/%s",
                        driver->stateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
515
            goto cleanup;
516 517 518 519 520 521 522 523 524 525 526 527
        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);
    }
528
    if (direrr < 0)
J
Ján Tomko 已提交
529
        goto cleanup;
530 531 532 533 534 535 536 537 538 539

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

540
/**
541
 * networkStateInitialize:
542 543 544 545
 *
 * Initialization function for the QEmu daemon
 */
static int
546 547 548
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
549
{
550 551 552
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
553 554 555
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
556

557
    if (VIR_ALLOC(driver) < 0)
558
        goto error;
559

560 561
    if (virMutexInit(&driver->lock) < 0) {
        VIR_FREE(driver);
562 563
        goto error;
    }
564
    networkDriverLock();
565

566 567 568 569
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
570
    if (privileged) {
571
        if (VIR_STRDUP(driver->networkConfigDir,
572
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
573
            VIR_STRDUP(driver->networkAutostartDir,
574
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
575
            VIR_STRDUP(driver->stateDir,
576
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
577
            VIR_STRDUP(driver->pidDir,
578
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
579
            VIR_STRDUP(driver->dnsmasqStateDir,
580
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
581
            VIR_STRDUP(driver->radvdStateDir,
582 583
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
584 585 586 587 588

        /* migration from old to new location is only applicable for
         * privileged mode - unprivileged mode directories haven't
         * changed location.
         */
589
        if (networkMigrateStateFiles() < 0)
590
            goto error;
591
    } else {
592 593 594
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
595
            goto error;
596

597
        if ((virAsprintf(&driver->networkConfigDir,
598
                         "%s/qemu/networks", configdir) < 0) ||
599
            (virAsprintf(&driver->networkAutostartDir,
600
                         "%s/qemu/networks/autostart", configdir) < 0) ||
601
            (virAsprintf(&driver->stateDir,
602
                         "%s/network/lib", rundir) < 0) ||
603
            (virAsprintf(&driver->pidDir,
604
                         "%s/network/run", rundir) < 0) ||
605
            (virAsprintf(&driver->dnsmasqStateDir,
606
                         "%s/dnsmasq/lib", rundir) < 0) ||
607
            (virAsprintf(&driver->radvdStateDir,
608
                         "%s/radvd/lib", rundir) < 0)) {
609
            goto error;
610
        }
611 612
    }

613
    if (virFileMakePath(driver->stateDir) < 0) {
614 615
        virReportSystemError(errno,
                             _("cannot create directory %s"),
616
                             driver->stateDir);
617 618 619
        goto error;
    }

620
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
621
    driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
622

623
    if (!(driver->networks = virNetworkObjListNew()))
624 625 626
        goto error;

    if (virNetworkLoadAllState(driver->networks,
627
                               driver->stateDir) < 0)
628 629
        goto error;

630
    if (virNetworkLoadAllConfigs(driver->networks,
631 632
                                 driver->networkConfigDir,
                                 driver->networkAutostartDir) < 0)
633 634
        goto error;

635 636 637 638 639 640 641 642 643 644 645

    /* 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). */
    virNetworkObjListForEach(driver->networks,
                             networkUpdateState,
                             NULL);
    virNetworkObjListPrune(driver->networks,
                           VIR_CONNECT_LIST_NETWORKS_INACTIVE |
                           VIR_CONNECT_LIST_NETWORKS_TRANSIENT);
646 647
    networkReloadFirewallRules();
    networkRefreshDaemons();
648

649
    driver->networkEventState = virObjectEventStateNew();
650

651
    networkDriverUnlock();
652

653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674
#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,
675
                                   NULL, NULL);
676 677 678
    }
#endif

679
    ret = 0;
680
 cleanup:
681 682 683
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
684

685
 error:
686 687
    if (driver)
        networkDriverUnlock();
688
    networkStateCleanup();
689
    goto cleanup;
690 691
}

692 693 694 695 696 697 698 699
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
700
    if (!driver)
701 702
        return;

703
    networkDriverLock();
704 705 706
    virNetworkObjListForEach(driver->networks,
                             networkAutostartConfig,
                             NULL);
707
    networkDriverUnlock();
708 709
}

710
/**
711
 * networkStateReload:
712 713 714 715 716
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
717 718
networkStateReload(void)
{
719
    if (!driver)
720 721
        return 0;

722
    networkDriverLock();
723
    virNetworkLoadAllState(driver->networks,
724
                           driver->stateDir);
725
    virNetworkLoadAllConfigs(driver->networks,
726 727 728 729
                             driver->networkConfigDir,
                             driver->networkAutostartDir);
    networkReloadFirewallRules();
    networkRefreshDaemons();
730 731 732
    virNetworkObjListForEach(driver->networks,
                             networkAutostartConfig,
                             NULL);
733
    networkDriverUnlock();
734 735 736 737 738
    return 0;
}


/**
739
 * networkStateCleanup:
740 741 742 743
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
744 745
networkStateCleanup(void)
{
746
    if (!driver)
747 748
        return -1;

749
    networkDriverLock();
750

751
    virObjectEventStateFree(driver->networkEventState);
752

753
    /* free inactive networks */
754
    virObjectUnref(driver->networks);
755

756 757 758 759 760 761
    VIR_FREE(driver->networkConfigDir);
    VIR_FREE(driver->networkAutostartDir);
    VIR_FREE(driver->stateDir);
    VIR_FREE(driver->pidDir);
    VIR_FREE(driver->dnsmasqStateDir);
    VIR_FREE(driver->radvdStateDir);
762

763
    virObjectUnref(driver->dnsmasqCaps);
764

765 766
    networkDriverUnlock();
    virMutexDestroy(&driver->lock);
767

768
    VIR_FREE(driver);
769 770 771 772 773

    return 0;
}


774 775 776 777 778 779 780
/* 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)
{
781 782
    size_t i;
    int ret = -1;
783 784 785 786 787 788 789
    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.
     */
790
    for (i = 0; i < 25; i++) {
791
        int signum = 0;
792
        if (i == 0) {
793
            signum = SIGTERM;
794
        } else if (i == 15) {
795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829
            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);
830
 cleanup:
831 832 833
    return ret;
}

J
Ján Tomko 已提交
834 835 836
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
G
Gene Czarcinski 已提交
837

838
static int
G
Gene Czarcinski 已提交
839 840
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
                                 virNetworkIpDefPtr ipdef)
841
{
842
    size_t i;
G
Gene Czarcinski 已提交
843
    bool ipv6 = false;
844

G
Gene Czarcinski 已提交
845 846
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
847 848
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
849
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
850 851
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
852
                return -1;
853
    }
854

G
Gene Czarcinski 已提交
855 856 857 858 859 860 861
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
862
    size_t i, j;
G
Gene Czarcinski 已提交
863

864 865
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
866
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
867
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
868
                for (j = 0; j < host->nnames; j++)
869 870
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
871 872
            }
        }
873 874
    }

875
    return 0;
876 877 878
}


879 880
int
networkDnsmasqConfContents(virNetworkObjPtr network,
881 882 883 884
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
885
{
886
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
887
    int r, ret = -1;
888
    int nbleases = 0;
889
    size_t i;
890
    virNetworkDNSDefPtr dns = &network->def->dns;
G
Gene Czarcinski 已提交
891 892
    virNetworkIpDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
    bool ipv6SLAAC;
893

894 895
    *configstr = NULL;

896
    /*
897 898 899
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
900
     *
901 902
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
903
     */
904 905 906 907 908 909

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

910 911
    /* create dnsmasq config file appropriate for this network */
    virBufferAsprintf(&configbuf,
912 913 914 915 916 917 918
                      "##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"
919
                      "strict-order\n",
920 921
                      network->def->name);

922 923 924 925
    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
926
                              network->def->dns.forwarders[i]);
927 928 929
        }
    }

930
    if (network->def->domain) {
931 932 933 934 935
        if (network->def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
                              network->def->domain);
        }
936
        virBufferAsprintf(&configbuf,
937 938 939 940
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
941

J
Ján Tomko 已提交
942
    if (network->def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
943 944 945 946
        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)
947
         */
948
        virBufferAddLit(&configbuf, "local=//\n");
949
    }
950

951
    if (pidfile)
952
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
953

954 955 956
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

957 958 959 960 961 962 963 964
    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.
         */
965
        virBufferAsprintf(&configbuf,
966 967 968
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
969
    } else {
970
        virBufferAddLit(&configbuf, "bind-interfaces\n");
971 972 973 974 975 976 977 978
        /*
         * --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
         */
979 980 981
        for (i = 0;
             (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
             i++) {
982 983 984 985
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
986

987
            /* also part of CVE 2012-3411 - if the host's version of
988
             * dnsmasq doesn't have bind-dynamic, only allow listening on
989 990
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
991 992
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
993 994 995 996
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
997
                                 "The version of dnsmasq on this host (%d.%d) "
998 999 1000 1001
                                 "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 "
1002 1003 1004 1005 1006 1007
                                 "(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);
1008
                VIR_FREE(ipaddr);
1009 1010
                goto cleanup;
            }
1011
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1012 1013 1014
            VIR_FREE(ipaddr);
        }
    }
1015

1016 1017
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1018
     * guaranteed to not work, and set no-resolv so that no dns
1019 1020 1021
     * 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).
1022
     */
1023
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1024
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1025
                        "no-resolv\n");
1026
    }
1027

1028
    for (i = 0; i < dns->ntxts; i++) {
1029
        virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
1030 1031
                          dns->txts[i].name,
                          dns->txts[i].value);
1032
    }
1033

1034
    for (i = 0; i < dns->nsrvs; i++) {
1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056
        /* 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);
1057

1058 1059 1060
        /* domain is optional - it defaults to the domain of this network */
        if (dns->srvs[i].domain)
            virBufferAsprintf(&configbuf, ".%s", dns->srvs[i].domain);
1061

1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083
        /* 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);
1084
        }
1085
        virBufferAddLit(&configbuf, "\n");
1086 1087
    }

G
Gene Czarcinski 已提交
1088
    /* Find the first dhcp for both IPv4 and IPv6 */
1089 1090 1091
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
1092 1093 1094 1095
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1096 1097
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108
                    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,
1109 1110 1111 1112 1113 1114 1115 1116 1117
                                   _("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
1118 1119 1120 1121
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1122 1123
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136
                    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 "
1137 1138 1139 1140
                 "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
1141 1142 1143 1144 1145
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1146
        for (r = 0; r < ipdef->nranges; r++) {
1147
            char *saddr = virSocketAddrFormat(&ipdef->ranges[r].start);
1148 1149
            if (!saddr)
                goto cleanup;
1150
            char *eaddr = virSocketAddrFormat(&ipdef->ranges[r].end);
1151 1152 1153 1154
            if (!eaddr) {
                VIR_FREE(saddr);
                goto cleanup;
            }
1155
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s\n",
1156
                              saddr, eaddr);
1157
            VIR_FREE(saddr);
1158
            VIR_FREE(eaddr);
1159 1160
            nbleases += virSocketAddrGetRange(&ipdef->ranges[r].start,
                                              &ipdef->ranges[r].end);
1161
        }
1162

1163
        /*
1164 1165 1166 1167
         * 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)
1168 1169
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1170
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1171 1172
            if (!bridgeaddr)
                goto cleanup;
1173
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static\n", bridgeaddr);
1174 1175
            VIR_FREE(bridgeaddr);
        }
1176

G
Gene Czarcinski 已提交
1177 1178
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1179

G
Gene Czarcinski 已提交
1180 1181 1182
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts)
1183
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1184

G
Gene Czarcinski 已提交
1185
            if (ipdef->tftproot) {
1186 1187
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1188
            }
1189

G
Gene Czarcinski 已提交
1190 1191 1192
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1193

1194
                    if (!bootserver)
G
Gene Czarcinski 已提交
1195
                        goto cleanup;
1196
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1197
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1198 1199
                    VIR_FREE(bootserver);
                } else {
1200
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1201 1202 1203 1204 1205
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1206

1207
    if (nbleases > 0)
1208
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1209

G
Gene Czarcinski 已提交
1210 1211
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1212
        goto cleanup;
G
Gene Czarcinski 已提交
1213 1214 1215 1216 1217 1218

    /* 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)
1219 1220
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1221

1222 1223
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1224
     */
1225
    virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
1226
                      dctx->addnhostsfile->path);
G
Gene Czarcinski 已提交
1227 1228 1229

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1230
        if (ipv6def) {
1231
            virBufferAddLit(&configbuf, "enable-ra\n");
1232
        } else {
1233 1234 1235
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
                 i++) {
G
Gene Czarcinski 已提交
1236 1237 1238 1239
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1240 1241
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1242 1243
                    VIR_FREE(bridgeaddr);
                }
1244
            }
1245
        }
1246 1247
    }

1248 1249 1250
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1251
    ret = 0;
G
Gene Czarcinski 已提交
1252

1253
 cleanup:
1254
    virBufferFreeAndReset(&configbuf);
1255
    return ret;
1256 1257
}

1258
/* build the dnsmasq command line */
1259 1260 1261
static int ATTRIBUTE_NONNULL(2)
networkBuildDhcpDaemonCommandLine(virNetworkObjPtr network,
                                  virCommandPtr *cmdout,
1262 1263
                                  char *pidfile, dnsmasqContext *dctx,
                                  dnsmasqCapsPtr caps)
1264
{
1265
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1266
    int ret = -1;
1267 1268
    char *configfile = NULL;
    char *configstr = NULL;
1269
    char *leaseshelper_path = NULL;
1270 1271

    network->dnsmasqPid = -1;
1272

1273 1274 1275 1276 1277 1278
    if (networkDnsmasqConfContents(network, pidfile, &configstr, dctx, caps) < 0)
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1279
    if (!(configfile = networkDnsmasqConfigFileName(network->def->name)))
1280 1281 1282 1283 1284
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1285 1286
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1287 1288 1289
        goto cleanup;
    }

1290 1291
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1292
                                                  abs_topbuilddir "/src",
1293 1294 1295
                                                  LIBEXECDIR)))
        goto cleanup;

1296 1297
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(caps));
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1298 1299
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1300
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1301
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", network->def->bridge);
1302

1303
    *cmdout = cmd;
1304
    ret = 0;
1305
 cleanup:
1306 1307
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1308
    VIR_FREE(leaseshelper_path);
1309 1310 1311 1312
    return ret;
}

static int
1313
networkStartDhcpDaemon(virNetworkObjPtr network)
1314 1315 1316 1317
{
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1318
    dnsmasqContext *dctx = NULL;
1319

1320
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0)) {
G
Gene Czarcinski 已提交
1321
        /* no IP addresses, so we don't need to run */
1322 1323 1324 1325
        ret = 0;
        goto cleanup;
    }

1326
    if (virFileMakePath(driver->pidDir) < 0) {
1327
        virReportSystemError(errno,
1328
                             _("cannot create directory %s"),
1329
                             driver->pidDir);
1330
        goto cleanup;
1331 1332
    }

1333
    if (!(pidfile = virPidFileBuildPath(driver->pidDir,
1334
                                        network->def->name)))
1335
        goto cleanup;
1336

1337
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1338
        virReportSystemError(errno,
1339
                             _("cannot create directory %s"),
1340
                             driver->dnsmasqStateDir);
1341 1342 1343
        goto cleanup;
    }

1344
    dctx = dnsmasqContextNew(network->def->name, driver->dnsmasqStateDir);
1345 1346 1347
    if (dctx == NULL)
        goto cleanup;

1348 1349
    if (dnsmasqCapsRefresh(&driver->dnsmasqCaps, NULL) < 0)
        goto cleanup;
1350 1351 1352

    ret = networkBuildDhcpDaemonCommandLine(network, &cmd, pidfile,
                                            dctx, driver->dnsmasqCaps);
1353 1354 1355 1356 1357
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1358
        goto cleanup;
1359

G
Guido Günther 已提交
1360
    ret = virCommandRun(cmd, NULL);
1361
    if (ret < 0)
1362 1363 1364
        goto cleanup;

    /*
1365 1366 1367 1368 1369
     * 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
1370 1371
     */

1372
    ret = virPidFileRead(driver->pidDir, network->def->name,
1373 1374
                         &network->dnsmasqPid);
    if (ret < 0)
1375
        goto cleanup;
1376

1377
    ret = 0;
1378
 cleanup:
1379
    VIR_FREE(pidfile);
1380
    virCommandFree(cmd);
1381
    dnsmasqContextFree(dctx);
1382 1383 1384
    return ret;
}

1385 1386
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1387 1388
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1389 1390 1391
 *
 *  Returns 0 on success, -1 on failure.
 */
1392
static int
1393
networkRefreshDhcpDaemon(virNetworkObjPtr network)
1394
{
1395 1396
    int ret = -1;
    size_t i;
G
Gene Czarcinski 已提交
1397
    virNetworkIpDefPtr ipdef, ipv4def, ipv6def;
1398
    dnsmasqContext *dctx = NULL;
1399

G
Gene Czarcinski 已提交
1400
    /* if no IP addresses specified, nothing to do */
1401
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1402 1403
        return 0;

1404 1405
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1406
        return networkStartDhcpDaemon(network);
1407

G
Gene Czarcinski 已提交
1408
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1409
    if (!(dctx = dnsmasqContextNew(network->def->name,
1410
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1411
        goto cleanup;
1412
    }
G
Gene Czarcinski 已提交
1413 1414 1415 1416 1417 1418

    /* 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;
1419 1420 1421
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
G
Gene Czarcinski 已提交
1422 1423
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1424 1425
    }

G
Gene Czarcinski 已提交
1426
    ipv6def = NULL;
1427 1428 1429
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1430 1431
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1432 1433
    }

G
Gene Czarcinski 已提交
1434
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1435
        goto cleanup;
G
Gene Czarcinski 已提交
1436 1437

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1438
        goto cleanup;
1439

G
Gene Czarcinski 已提交
1440
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
J
Ján Tomko 已提交
1441
        goto cleanup;
1442 1443

    if ((ret = dnsmasqSave(dctx)) < 0)
1444
        goto cleanup;
1445 1446

    ret = kill(network->dnsmasqPid, SIGHUP);
1447
 cleanup:
1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459
    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
1460
networkRestartDhcpDaemon(virNetworkObjPtr network)
1461 1462 1463 1464 1465 1466
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1467
    }
1468
    /* now start dnsmasq if it should be started */
1469
    return networkStartDhcpDaemon(network);
1470 1471
}

G
Gene Czarcinski 已提交
1472 1473 1474 1475 1476 1477
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";

1478 1479 1480
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1481
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1482 1483
    int ret = -1;
    size_t i;
1484
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
1485
    bool v6present = false, dhcp6 = false;
1486 1487

    *configstr = NULL;
1488

G
Gene Czarcinski 已提交
1489
    /* Check if DHCPv6 is needed */
1490 1491 1492
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505
        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;
    }

1506 1507 1508
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1509
    virBufferAsprintf(&configbuf, "interface %s\n"
1510 1511
                      "{\n"
                      "  AdvSendAdvert on;\n"
1512
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1513 1514 1515 1516 1517
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1518 1519

    /* add a section for each IPv6 address in the config */
1520 1521 1522
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
1523 1524 1525 1526 1527
        int prefix;
        char *netaddr;

        prefix = virNetworkIpDefPrefix(ipdef);
        if (prefix < 0) {
1528 1529 1530
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1531 1532
            goto cleanup;
        }
1533
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1534
            goto cleanup;
1535
        virBufferAsprintf(&configbuf,
1536
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1537 1538 1539
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1540 1541 1542
        VIR_FREE(netaddr);
    }

1543
    virBufferAddLit(&configbuf, "};\n");
1544

1545
    if (virBufferCheckError(&configbuf) < 0)
1546
        goto cleanup;
1547

1548 1549
    *configstr = virBufferContentAndReset(&configbuf);

1550
    ret = 0;
1551
 cleanup:
1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573
    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;
1574 1575 1576 1577
        goto cleanup;
    }

    /* construct the filename */
1578
    if (!(*configFile = networkRadvdConfigFileName(network->def->name)))
1579 1580
        goto cleanup;
    /* write the file */
1581
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1582 1583
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1584 1585 1586 1587 1588
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1589
 cleanup:
1590 1591 1592 1593 1594 1595
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1596
networkStartRadvd(virNetworkObjPtr network)
1597 1598 1599 1600 1601 1602 1603 1604 1605
{
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1606
    /* Is dnsmasq handling RA? */
J
Ján Tomko 已提交
1607
    if (DNSMASQ_RA_SUPPORT(driver->dnsmasqCaps)) {
G
Gene Czarcinski 已提交
1608 1609 1610 1611
        ret = 0;
        goto cleanup;
    }

1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622
    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);
1623 1624 1625
        goto cleanup;
    }

1626
    if (virFileMakePath(driver->pidDir) < 0) {
1627 1628
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1629
                             driver->pidDir);
1630 1631
        goto cleanup;
    }
1632
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1633 1634
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1635
                             driver->radvdStateDir);
1636 1637 1638 1639
        goto cleanup;
    }

    /* construct pidfile name */
1640
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1641
        goto cleanup;
1642
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1643 1644 1645 1646 1647
        goto cleanup;

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

1648 1649 1650 1651
    /* 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
1652
     * virPidFileRead() below will fail if we use them).
1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667
     * 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;

1668
    if (virPidFileRead(driver->pidDir, radvdpidbase, &network->radvdPid) < 0)
1669 1670 1671
        goto cleanup;

    ret = 0;
1672
 cleanup:
1673 1674 1675 1676 1677 1678 1679
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1680
static int
1681
networkRefreshRadvd(virNetworkObjPtr network)
1682
{
G
Gene Czarcinski 已提交
1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693
    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)) {
1694
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1695 1696 1697 1698 1699 1700
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }

1701 1702
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
1703
        return networkStartRadvd(network);
1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715

    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);
}

1716 1717
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1718
static int
1719
networkRestartRadvd(virNetworkObjPtr network)
1720 1721 1722 1723 1724 1725 1726 1727 1728
{
    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
1729
        if ((networkKillDaemon(network->radvdPid, "radvd",
1730 1731 1732
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1733
            virPidFileDelete(driver->pidDir, radvdpidbase);
1734 1735 1736 1737 1738 1739 1740 1741 1742
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765
static int
networkRefreshDaemonsHelper(virNetworkObjPtr net,
                            void *opaque ATTRIBUTE_UNUSED)
{

    virNetworkObjLock(net);
    if (virNetworkObjIsActive(net) &&
        ((net->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
        /* 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.
         */
        networkRefreshDhcpDaemon(net);
        networkRefreshRadvd(net);
    }
    virNetworkObjUnlock(net);
    return 0;
}

1766 1767 1768 1769
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
1770
networkRefreshDaemons(void)
1771 1772
{
    VIR_INFO("Refreshing network daemons");
1773 1774 1775 1776
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
                             NULL);
}
1777

1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793
static int
networkReloadFirewallRulesHelper(virNetworkObjPtr net,
                                 void *opaque ATTRIBUTE_UNUSED)
{

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

1800
static void
1801
networkReloadFirewallRules(void)
1802
{
1803
    VIR_INFO("Reloading iptables rules");
1804 1805 1806
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
1807 1808
}

1809
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
1810
static int
1811
networkEnableIpForwarding(bool enableIPv4, bool enableIPv6)
1812
{
1813
    int ret = 0;
1814 1815 1816 1817
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
1818
                           &enabled, sizeof(enabled));
1819 1820
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
1821
                           &enabled, sizeof(enabled));
1822
#else
1823 1824 1825 1826
    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);
1827
#endif
1828
    return ret;
1829 1830
}

1831 1832
#define SYSCTL_PATH "/proc/sys"

1833 1834
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
1835 1836 1837
{
    char *field = NULL;
    int ret = -1;
1838
    bool enableIPv6 =  !!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0);
1839

1840 1841 1842 1843 1844 1845 1846
    /* 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;
1847

1848 1849
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
1850 1851
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
1852 1853 1854
        ret = 0;
        goto cleanup;
    }
1855

1856 1857 1858 1859 1860
    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;
1861
    }
1862
    VIR_FREE(field);
1863

1864 1865
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
1866 1867 1868 1869 1870 1871
     */

    /* 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",
1872
                    network->def->bridge) < 0)
1873 1874
        goto cleanup;

1875
    if (virFileWriteStr(field, "0", 0) < 0) {
1876
        virReportSystemError(errno,
1877 1878 1879 1880 1881
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

1882 1883 1884 1885
    /* 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",
1886
                    network->def->bridge) < 0)
1887 1888
        goto cleanup;

1889
    if (virFileWriteStr(field, "0", 0) < 0) {
1890
        virReportSystemError(errno,
1891
                             _("cannot disable %s"), field);
1892 1893 1894 1895
        goto cleanup;
    }

    ret = 0;
1896
 cleanup:
1897 1898 1899 1900
    VIR_FREE(field);
    return ret;
}

1901
/* add an IP address to a bridge */
1902
static int
D
Daniel P. Berrange 已提交
1903
networkAddAddrToBridge(virNetworkObjPtr network,
1904
                       virNetworkIpDefPtr ipdef)
1905
{
1906 1907 1908
    int prefix = virNetworkIpDefPrefix(ipdef);

    if (prefix < 0) {
1909 1910 1911
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
1912 1913 1914
        return -1;
    }

1915 1916
    if (virNetDevSetIPAddress(network->def->bridge,
                              &ipdef->address, prefix) < 0)
1917 1918 1919 1920 1921
        return -1;

    return 0;
}

1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944

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;
}


1945 1946 1947 1948 1949
/* add an IP (static) route to a bridge */
static int
networkAddRouteToBridge(virNetworkObjPtr network,
                        virNetworkRouteDefPtr routedef)
{
1950 1951 1952 1953
    int prefix = virNetworkRouteDefGetPrefix(routedef);
    unsigned int metric = virNetworkRouteDefGetMetric(routedef);
    virSocketAddrPtr addr = virNetworkRouteDefGetAddress(routedef);
    virSocketAddrPtr gateway = virNetworkRouteDefGetGateway(routedef);
1954 1955 1956 1957 1958 1959 1960 1961 1962

    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;
    }

1963 1964
    if (virNetDevAddRoute(network->def->bridge, addr,
                          prefix, gateway, metric) < 0) {
1965 1966 1967 1968 1969
        return -1;
    }
    return 0;
}

1970
static int
1971
networkStartNetworkVirtual(virNetworkObjPtr network)
1972
{
1973
    size_t i;
1974
    bool v4present = false, v6present = false;
1975 1976
    virErrorPtr save_err = NULL;
    virNetworkIpDefPtr ipdef;
1977
    virNetworkRouteDefPtr routedef;
1978
    char *macTapIfName = NULL;
1979
    int tapfd = -1;
1980

1981
    /* Check to see if any network IP collides with an existing route */
1982
    if (networkCheckRouteCollision(network->def) < 0)
1983 1984
        return -1;

1985
    /* Create and configure the bridge device */
1986
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
1987 1988
        return -1;

1989 1990 1991 1992 1993 1994 1995 1996
    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);
1997
        if (!macTapIfName)
1998
            goto err0;
1999
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2000
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
2001
                                           &macTapIfName, &network->def->mac,
2002
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2003 2004 2005
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2006 2007 2008 2009 2010
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2011
    /* Set bridge options */
2012 2013 2014 2015

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2016
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2017
                                   network->def->delay * 1000) < 0)
2018
        goto err1;
2019

2020
    if (virNetDevBridgeSetSTP(network->def->bridge,
2021
                              network->def->stp ? true : false) < 0)
2022
        goto err1;
2023

2024 2025 2026 2027
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2028
        goto err1;
2029

2030
    /* Add "once per network" rules */
2031
    if (networkAddFirewallRules(network->def) < 0)
2032 2033
        goto err1;

2034 2035 2036
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
2037
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2038
            v4present = true;
2039
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2040
            v6present = true;
2041

2042
        /* Add the IP address/netmask to the bridge */
2043
        if (networkAddAddrToBridge(network, ipdef) < 0)
2044
            goto err2;
2045 2046
    }

2047 2048 2049
    if (networkStartHandleMACTableManagerMode(network, macTapIfName) < 0)
        goto err2;

2050
    /* Bring up the bridge interface */
2051
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2052
        goto err2;
2053

2054
    for (i = 0; i < network->def->nroutes; i++) {
2055 2056 2057 2058 2059
        virSocketAddrPtr gateway = NULL;

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

2060 2061 2062
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2063
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2064 2065 2066 2067 2068 2069 2070
            if (networkAddRouteToBridge(network, routedef) < 0) {
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2071 2072
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2073
        networkEnableIpForwarding(v4present, v6present) < 0) {
2074
        virReportSystemError(errno, "%s",
2075
                             _("failed to enable IP forwarding"));
2076
        goto err3;
2077 2078
    }

2079

2080
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2081
    if ((v4present || v6present) &&
2082
        networkStartDhcpDaemon(network) < 0)
2083
        goto err3;
2084

2085
    /* start radvd if there are any ipv6 addresses */
2086
    if (v6present && networkStartRadvd(network) < 0)
2087 2088
        goto err4;

2089 2090 2091 2092 2093 2094 2095 2096 2097
    /* 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);
    }

2098
    if (virNetDevBandwidthSet(network->def->bridge,
2099
                              network->def->bandwidth, true) < 0)
2100 2101
        goto err5;

2102
    VIR_FREE(macTapIfName);
2103 2104 2105

    return 0;

2106
 err5:
2107 2108
    if (network->def->bandwidth)
       virNetDevBandwidthClear(network->def->bridge);
2109

2110 2111 2112 2113
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2114 2115 2116 2117 2118
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2119 2120 2121
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2122
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2123

2124 2125 2126
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2127
    networkRemoveFirewallRules(network->def);
2128 2129

 err1:
2130 2131 2132
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2133
    if (macTapIfName) {
2134
        VIR_FORCE_CLOSE(tapfd);
2135
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2136 2137
        VIR_FREE(macTapIfName);
    }
2138 2139

 err0:
2140 2141
    if (!save_err)
        save_err = virSaveLastError();
2142
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2143

2144 2145 2146 2147
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2148
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2149 2150 2151
    return -1;
}

2152
static int networkShutdownNetworkVirtual(virNetworkObjPtr network)
2153
{
2154 2155
    if (network->def->bandwidth)
        virNetDevBandwidthClear(network->def->bridge);
2156

2157 2158 2159 2160 2161
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2162
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2163
            virPidFileDelete(driver->pidDir, radvdpidbase);
2164 2165 2166 2167
            VIR_FREE(radvdpidbase);
        }
    }

2168 2169 2170
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2171
    if (network->def->mac_specified) {
2172
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2173
        if (macTapIfName) {
2174
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2175 2176 2177 2178
            VIR_FREE(macTapIfName);
        }
    }

2179
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2180

2181
    networkRemoveFirewallRules(network->def);
2182

2183
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2184

2185
    /* See if its still alive and really really kill it */
2186
    if (network->dnsmasqPid > 0 &&
2187
        (kill(network->dnsmasqPid, 0) == 0))
2188 2189
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2190 2191 2192 2193 2194 2195

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

2196 2197 2198
    return 0;
}

2199

2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220
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;
}


2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235
/* 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;

2236 2237 2238
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

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 2324 2325 2326
    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;
}


2327
static int
2328
networkStartNetworkExternal(virNetworkObjPtr network)
2329 2330
{
    /* put anything here that needs to be done each time a network of
2331
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2332 2333 2334
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2335
    return networkCreateInterfacePool(network->def);
2336 2337
}

2338
static int networkShutdownNetworkExternal(virNetworkObjPtr network ATTRIBUTE_UNUSED)
2339 2340
{
    /* put anything here that needs to be done each time a network of
2341
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2342 2343 2344 2345 2346 2347 2348
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2349
networkStartNetwork(virNetworkObjPtr network)
2350
{
2351 2352 2353
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2354 2355

    if (virNetworkObjIsActive(network)) {
2356 2357
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2358
        return ret;
2359 2360
    }

2361 2362 2363
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2364
    if (virNetworkObjSetDefTransient(network, true) < 0)
2365
        goto cleanup;
2366

2367 2368
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2369
    if (networkRunHook(network, NULL, NULL,
2370 2371 2372 2373
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2374
    switch (network->def->forward.type) {
2375 2376 2377 2378

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2379
        if (networkStartNetworkVirtual(network) < 0)
2380
            goto cleanup;
2381 2382 2383
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2384 2385 2386 2387
       if (networkStartNetworkBridge(network) < 0)
          goto cleanup;
       break;

2388 2389 2390
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2391
    case VIR_NETWORK_FORWARD_HOSTDEV:
2392
        if (networkStartNetworkExternal(network) < 0)
2393
            goto cleanup;
2394 2395 2396
        break;
    }

2397
    /* finally we can call the 'started' hook script if any */
2398
    if (networkRunHook(network, NULL, NULL,
2399 2400 2401 2402
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2403 2404 2405
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2406
    VIR_DEBUG("Writing network status to disk");
2407
    if (virNetworkSaveStatus(driver->stateDir, network) < 0)
2408
        goto cleanup;
2409 2410

    network->active = 1;
2411 2412
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2413

2414
 cleanup:
2415
    if (ret < 0) {
2416
        virNetworkObjUnsetDefTransient(network);
2417 2418
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2419
        networkShutdownNetwork(network);
2420 2421 2422 2423 2424 2425 2426
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2427
static int networkShutdownNetwork(virNetworkObjPtr network)
2428 2429 2430 2431 2432 2433 2434 2435 2436
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

2437
    stateFile = virNetworkConfigFile(driver->stateDir,
2438
                                     network->def->name);
2439 2440 2441 2442 2443 2444
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2445
    switch (network->def->forward.type) {
2446 2447 2448 2449

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2450
        ret = networkShutdownNetworkVirtual(network);
2451 2452 2453
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2454 2455 2456
        ret = networkShutdownNetworkBridge(network);
        break;

2457 2458 2459
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2460
    case VIR_NETWORK_FORWARD_HOSTDEV:
2461
        ret = networkShutdownNetworkExternal(network);
2462 2463 2464
        break;
    }

2465
    /* now that we know it's stopped call the hook if present */
2466
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2467 2468
                   VIR_HOOK_SUBOP_END);

2469
    network->active = 0;
2470
    virNetworkObjUnsetDefTransient(network);
2471
    return ret;
2472 2473 2474
}


2475
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2476 2477
                                         const unsigned char *uuid)
{
2478 2479
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2480

2481
    networkDriverLock();
2482
    network = virNetworkObjFindByUUID(driver->networks, uuid);
2483
    networkDriverUnlock();
2484
    if (!network) {
2485 2486
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2487
        virReportError(VIR_ERR_NO_NETWORK,
2488 2489
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2490
        goto cleanup;
2491 2492
    }

2493 2494 2495
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2496 2497
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2498
 cleanup:
2499 2500
    if (network)
        virNetworkObjUnlock(network);
2501
    return ret;
2502 2503
}

2504
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2505 2506
                                         const char *name)
{
2507 2508 2509
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2510
    networkDriverLock();
2511
    network = virNetworkObjFindByName(driver->networks, name);
2512
    networkDriverUnlock();
2513
    if (!network) {
2514 2515
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2516
        goto cleanup;
2517 2518
    }

2519 2520 2521
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2522 2523
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2524
 cleanup:
2525 2526
    if (network)
        virNetworkObjUnlock(network);
2527
    return ret;
2528 2529
}

2530 2531
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2532
    int nactive;
2533

2534 2535 2536
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2537
    networkDriverLock();
2538 2539 2540 2541
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
2542
    networkDriverUnlock();
2543

2544 2545 2546
    return nactive;
}

2547
static int networkConnectListNetworks(virConnectPtr conn, char **const names, int nnames) {
2548
    int got = 0;
2549

2550 2551 2552
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2553
    networkDriverLock();
2554 2555 2556 2557
    got = virNetworkObjListGetNames(driver->networks,
                                    true, names, nnames,
                                    virConnectListNetworksCheckACL,
                                    conn);
2558
    networkDriverUnlock();
2559

2560 2561 2562
    return got;
}

2563 2564
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2565
    int ninactive = 0;
2566

2567 2568 2569
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2570
    networkDriverLock();
2571 2572 2573 2574
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
2575
    networkDriverUnlock();
2576

2577 2578 2579
    return ninactive;
}

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

2583 2584 2585
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2586
    networkDriverLock();
2587 2588 2589 2590
    got = virNetworkObjListGetNames(driver->networks,
                                    false, names, nnames,
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
2591
    networkDriverUnlock();
2592 2593 2594
    return got;
}

2595
static int
2596 2597 2598
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2599 2600 2601 2602 2603
{
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2604 2605 2606
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2607
    networkDriverLock();
2608
    ret = virNetworkObjListExport(conn, driver->networks, nets,
2609 2610
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2611
    networkDriverUnlock();
2612

2613
 cleanup:
2614 2615
    return ret;
}
2616

2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630
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,
2631
                                       net, eventID, callback,
2632 2633 2634
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2635
 cleanup:
2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647
    return ret;
}

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

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

2648 2649 2650 2651 2652 2653
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
2654

2655
 cleanup:
2656 2657 2658
    return ret;
}

2659 2660 2661 2662 2663
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2664 2665
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2666 2667 2668 2669

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

2670 2671
    ret = virNetworkObjIsActive(obj);

2672
 cleanup:
2673 2674 2675 2676 2677 2678 2679 2680 2681 2682
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}

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

2683 2684
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2685 2686 2687 2688

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

2689 2690
    ret = obj->persistent;

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


2698
static int
2699
networkValidate(virNetworkDefPtr def,
2700
                bool check_active)
2701
{
2702
    size_t i, j;
2703 2704
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2705
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
2706
    bool ipv4def = false, ipv6def = false;
2707
    bool bandwidthAllowed = true;
2708
    bool usesInterface = false, usesAddress = false;
2709 2710

    /* check for duplicate networks */
2711
    if (virNetworkObjIsDuplicate(driver->networks, def, check_active) < 0)
2712 2713 2714 2715 2716
        return -1;

    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
2717 2718 2719
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2720

2721
        if (virNetworkSetBridgeName(driver->networks, def, 1))
2722 2723 2724
            return -1;

        virNetworkSetBridgeMacAddr(def);
2725 2726
    } else {
        /* They are also the only types that currently support setting
2727 2728
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
2729
         */
2730 2731 2732 2733 2734 2735 2736 2737
        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;
        }
2738 2739 2740 2741 2742
        if (virNetworkDefGetIpByIndex(def, AF_UNSPEC, 0)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2743
                           virNetworkForwardTypeToString(def->forward.type));
2744 2745
            return -1;
        }
2746
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
2747 2748 2749 2750
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2751
                           virNetworkForwardTypeToString(def->forward.type));
2752 2753 2754 2755 2756 2757 2758
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2759
                           virNetworkForwardTypeToString(def->forward.type));
2760 2761
            return -1;
        }
2762 2763 2764 2765 2766 2767 2768 2769
        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;
        }
2770
        bandwidthAllowed = false;
2771 2772
    }

2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806
    /* 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
2807 2808 2809
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
2810 2811 2812
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
2813 2814 2815 2816 2817
        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 -- "
2818 2819
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836
                    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;
                }
2837 2838 2839
            }
        }
    }
2840 2841 2842 2843 2844 2845

    /* 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.
     */

2846
    vlanAllowed = ((def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
J
Ján Tomko 已提交
2847 2848
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
2849 2850
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) ||
                   def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV);
2851 2852

    vlanUsed = def->vlan.nTags > 0;
2853 2854
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
2855 2856 2857 2858 2859
            /* 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.
             */
2860
            if (def->portGroups[i].virtPortProfile) {
2861
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
2862
                    def->portGroups[i].virtPortProfile->virtPortType
2863 2864 2865 2866 2867 2868 2869
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
2870
        }
2871
        if (def->portGroups[i].isDefault) {
2872 2873 2874 2875 2876
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
2877
                               def->name, defaultPortGroup->name,
2878
                               def->portGroups[i].name);
2879
                return -1;
2880
            }
2881
            defaultPortGroup = &def->portGroups[i];
2882
        }
2883 2884 2885 2886 2887 2888 2889 2890 2891
        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;
            }
        }
2892 2893 2894 2895 2896 2897 2898 2899
        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;
        }
2900
    }
2901 2902 2903 2904 2905 2906 2907
    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.
         */
2908 2909 2910 2911 2912 2913 2914 2915 2916
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
    return 0;
}

2917 2918
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
2919
    virNetworkDefPtr def;
2920
    virNetworkObjPtr network = NULL;
2921
    virNetworkPtr ret = NULL;
2922
    virObjectEventPtr event = NULL;
2923

2924
    networkDriverLock();
2925

2926
    if (!(def = virNetworkDefParseString(xml)))
2927
        goto cleanup;
2928

2929 2930 2931
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

2932
    if (networkValidate(def, true) < 0)
J
Ján Tomko 已提交
2933
        goto cleanup;
2934

2935 2936 2937
    /* 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.
2938
     */
2939
    if (!(network = virNetworkAssignDef(driver->networks, def, true)))
2940 2941
        goto cleanup;
    def = NULL;
2942

2943
    if (networkStartNetwork(network) < 0) {
2944
        virNetworkRemoveInactive(driver->networks,
2945
                                 network);
2946
        network = NULL;
2947
        goto cleanup;
2948 2949
    }

2950 2951
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
2952 2953
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
2954

2955
    VIR_INFO("Creating network '%s'", network->def->name);
2956 2957
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2958
 cleanup:
2959
    virNetworkDefFree(def);
2960 2961
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
2962 2963
    if (network)
        virNetworkObjUnlock(network);
2964
    networkDriverUnlock();
2965
    return ret;
2966 2967
}

2968 2969
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
2970
    virNetworkDefPtr def = NULL;
2971
    bool freeDef = true;
2972
    virNetworkObjPtr network = NULL;
2973
    virNetworkPtr ret = NULL;
2974
    virObjectEventPtr event = NULL;
2975

2976
    networkDriverLock();
2977

2978
    if (!(def = virNetworkDefParseString(xml)))
2979
        goto cleanup;
2980

2981 2982 2983
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

2984
    if (networkValidate(def, false) < 0)
J
Ján Tomko 已提交
2985
        goto cleanup;
2986

2987
    if (!(network = virNetworkAssignDef(driver->networks, def, false)))
J
Ján Tomko 已提交
2988
        goto cleanup;
2989

2990
    /* def was assigned to network object */
2991
    freeDef = false;
2992 2993

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
2994
        if (!virNetworkObjIsActive(network)) {
2995
            virNetworkRemoveInactive(driver->networks, network);
2996 2997 2998
            network = NULL;
            goto cleanup;
        }
2999 3000 3001 3002 3003
        /* 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);
3004 3005 3006
        goto cleanup;
    }

3007
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3008 3009
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3010

3011 3012
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
3013

3014
 cleanup:
3015 3016
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3017
    if (freeDef)
J
Ján Tomko 已提交
3018
        virNetworkDefFree(def);
3019 3020
    if (network)
        virNetworkObjUnlock(network);
3021
    networkDriverUnlock();
3022
    return ret;
3023 3024
}

3025
static int
3026 3027
networkUndefine(virNetworkPtr net)
{
3028
    virNetworkObjPtr network;
3029
    int ret = -1;
3030
    bool active = false;
3031
    virObjectEventPtr event = NULL;
3032

3033
    networkDriverLock();
3034

3035
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
3036
    if (!network) {
3037 3038
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3039
        goto cleanup;
3040 3041
    }

3042 3043 3044
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3045 3046
    if (virNetworkObjIsActive(network))
        active = true;
3047

3048
    /* remove autostart link */
3049
    if (virNetworkDeleteConfig(driver->networkConfigDir,
3050 3051
                               driver->networkAutostartDir,
                               network) < 0)
3052
        goto cleanup;
3053

3054 3055
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3056 3057
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3058

3059
    VIR_INFO("Undefining network '%s'", network->def->name);
3060
    if (!active) {
3061
        if (networkRemoveInactive(network) < 0) {
3062 3063 3064
            network = NULL;
            goto cleanup;
        }
3065
        network = NULL;
3066 3067 3068 3069 3070 3071
    } 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);
3072 3073
    }

3074
    ret = 0;
3075

3076
 cleanup:
3077 3078
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3079 3080
    if (network)
        virNetworkObjUnlock(network);
3081
    networkDriverUnlock();
3082
    return ret;
3083 3084
}

3085 3086 3087 3088 3089 3090 3091 3092 3093
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
    virNetworkObjPtr network = NULL;
3094 3095
    int isActive, ret = -1;
    size_t i;
3096 3097
    virNetworkIpDefPtr ipdef;
    bool oldDhcpActive = false;
3098
    bool needFirewallRefresh = false;
3099

3100 3101 3102 3103 3104

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3105
    networkDriverLock();
3106

3107
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
3108 3109 3110 3111 3112 3113
    if (!network) {
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
        goto cleanup;
    }

3114 3115 3116
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3117
    /* see if we are listening for dhcp pre-modification */
3118 3119 3120
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
3121 3122 3123 3124 3125 3126
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3127 3128
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3129
     */
3130
    isActive = virNetworkObjIsActive(network);
3131 3132
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3133 3134 3135 3136 3137 3138 3139
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156
    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).
                 */
3157
                networkRemoveFirewallRules(network->def);
3158 3159 3160 3161 3162 3163 3164 3165
                needFirewallRefresh = true;
                break;
            default:
                break;
            }
        }
    }

3166
    /* update the network config in memory/on disk */
3167 3168
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3169
            ignore_value(networkAddFirewallRules(network->def));
3170 3171 3172
        goto cleanup;
    }

3173
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196
        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.
             */
3197
            if (networkRestartDhcpDaemon(network) < 0)
3198 3199
                goto cleanup;

3200 3201 3202 3203 3204 3205 3206 3207
        } 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;

3208 3209 3210
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
                 i++) {
3211 3212 3213 3214 3215 3216 3217
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3218 3219
                 networkRestartDhcpDaemon(network) < 0) ||
                networkRefreshDhcpDaemon(network) < 0) {
3220 3221 3222 3223
                goto cleanup;
            }

        } else if (section == VIR_NETWORK_SECTION_DNS_HOST ||
3224 3225 3226 3227 3228 3229
                   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.
             */
3230
            if (networkRefreshDhcpDaemon(network) < 0)
3231 3232 3233 3234 3235 3236 3237 3238
                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.
             */
3239
            if (networkRefreshRadvd(network) < 0)
3240 3241 3242 3243
                goto cleanup;
        }

        /* save current network state to disk */
3244
        if ((ret = virNetworkSaveStatus(driver->stateDir,
3245
                                        network)) < 0) {
3246
            goto cleanup;
3247
        }
3248 3249
    }
    ret = 0;
3250
 cleanup:
3251 3252
    if (network)
        virNetworkObjUnlock(network);
3253
    networkDriverUnlock();
3254 3255 3256
    return ret;
}

3257 3258
static int networkCreate(virNetworkPtr net)
{
3259 3260
    virNetworkObjPtr network;
    int ret = -1;
3261
    virObjectEventPtr event = NULL;
3262

3263
    networkDriverLock();
3264
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
3265

3266
    if (!network) {
3267 3268
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3269
        goto cleanup;
3270 3271
    }

3272 3273 3274
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3275
    if ((ret = networkStartNetwork(network)) < 0)
3276
        goto cleanup;
3277

3278 3279
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3280 3281
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3282

3283
 cleanup:
3284 3285
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3286 3287
    if (network)
        virNetworkObjUnlock(network);
3288
    networkDriverUnlock();
3289
    return ret;
3290 3291
}

3292 3293
static int networkDestroy(virNetworkPtr net)
{
3294 3295
    virNetworkObjPtr network;
    int ret = -1;
3296
    virObjectEventPtr event = NULL;
3297

3298
    networkDriverLock();
3299
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
3300

3301
    if (!network) {
3302 3303
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3304
        goto cleanup;
3305 3306
    }

3307 3308 3309
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3310
    if (!virNetworkObjIsActive(network)) {
3311
        virReportError(VIR_ERR_OPERATION_INVALID,
3312 3313
                       _("network '%s' is not active"),
                       network->def->name);
3314 3315 3316
        goto cleanup;
    }

3317
    if ((ret = networkShutdownNetwork(network)) < 0)
3318 3319
        goto cleanup;

3320 3321
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3322 3323
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3324

3325
    if (!network->persistent) {
3326
        if (networkRemoveInactive(network) < 0) {
3327 3328 3329 3330
            network = NULL;
            ret = -1;
            goto cleanup;
        }
3331 3332
        network = NULL;
    }
3333

3334
 cleanup:
3335 3336
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3337 3338
    if (network)
        virNetworkObjUnlock(network);
3339
    networkDriverUnlock();
3340 3341 3342
    return ret;
}

3343
static char *networkGetXMLDesc(virNetworkPtr net,
3344
                               unsigned int flags)
3345
{
3346
    virNetworkObjPtr network;
3347
    virNetworkDefPtr def;
3348
    char *ret = NULL;
3349

3350
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3351

3352 3353
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3354

3355 3356 3357
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3358 3359 3360 3361 3362 3363
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3364

3365
 cleanup:
3366 3367
    if (network)
        virNetworkObjUnlock(network);
3368
    return ret;
3369 3370 3371
}

static char *networkGetBridgeName(virNetworkPtr net) {
3372 3373 3374
    virNetworkObjPtr network;
    char *bridge = NULL;

3375 3376
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3377

3378 3379 3380
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3381
    if (!(network->def->bridge)) {
3382 3383 3384
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3385 3386 3387
        goto cleanup;
    }

3388
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3389

3390
 cleanup:
3391 3392
    if (network)
        virNetworkObjUnlock(network);
3393 3394 3395 3396
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
J
Ján Tomko 已提交
3397
                               int *autostart)
3398
{
3399 3400
    virNetworkObjPtr network;
    int ret = -1;
3401

3402 3403
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3404

3405 3406 3407
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3408
    *autostart = network->autostart;
3409
    ret = 0;
3410

3411
 cleanup:
3412 3413
    if (network)
        virNetworkObjUnlock(network);
3414
    return ret;
3415 3416 3417
}

static int networkSetAutostart(virNetworkPtr net,
3418 3419
                               int autostart)
{
3420
    virNetworkObjPtr network;
3421
    char *configFile = NULL, *autostartLink = NULL;
3422
    int ret = -1;
3423

3424
    networkDriverLock();
3425
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
3426

3427
    if (!network) {
3428 3429
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3430
        goto cleanup;
3431 3432
    }

3433 3434 3435
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3436
    if (!network->persistent) {
3437 3438
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3439 3440 3441
        goto cleanup;
    }

3442 3443
    autostart = (autostart != 0);

3444
    if (network->autostart != autostart) {
3445
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3446
            goto cleanup;
3447
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3448 3449
            goto cleanup;

3450
        if (autostart) {
3451
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3452
                virReportSystemError(errno,
3453 3454
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3455 3456
                goto cleanup;
            }
3457

3458
            if (symlink(configFile, autostartLink) < 0) {
3459
                virReportSystemError(errno,
3460
                                     _("Failed to create symlink '%s' to '%s'"),
3461
                                     autostartLink, configFile);
3462 3463 3464
                goto cleanup;
            }
        } else {
3465
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3466
                virReportSystemError(errno,
3467
                                     _("Failed to delete symlink '%s'"),
3468
                                     autostartLink);
3469 3470
                goto cleanup;
            }
3471 3472
        }

3473
        network->autostart = autostart;
3474
    }
3475
    ret = 0;
3476

3477
 cleanup:
3478 3479
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3480 3481
    if (network)
        virNetworkObjUnlock(network);
3482
    networkDriverUnlock();
3483
    return ret;
3484 3485
}

3486
static int
3487 3488 3489 3490
networkGetDHCPLeases(virNetworkPtr network,
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509
{
    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;
3510 3511 3512 3513 3514 3515 3516 3517 3518
    virNetworkObjPtr obj;

    virCheckFlags(0, -1);

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

    if (virNetworkGetDHCPLeasesEnsureACL(network->conn, obj->def) < 0)
        goto cleanup;
3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564

    /* 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;
        }

3565
        if (mac && virMacAddrCompare(mac, mac_tmp))
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 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615
            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) ||
3616
                (VIR_STRDUP(lease->iface, obj->def->bridge) < 0))
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
                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);
3649
    VIR_FREE(lease_entries);
3650 3651
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
3652 3653 3654 3655

    if (obj)
        virNetworkObjUnlock(obj);

3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666
    return rv;

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

3667 3668

static virNetworkDriver networkDriver = {
3669
    .name = "bridge",
3670 3671 3672 3673 3674
    .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 */
3675 3676
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
3677 3678
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3679 3680
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3681
    .networkUndefine = networkUndefine, /* 0.2.0 */
3682
    .networkUpdate = networkUpdate, /* 0.10.2 */
3683
    .networkCreate = networkCreate, /* 0.2.0 */
3684 3685 3686 3687 3688 3689 3690
    .networkDestroy = networkDestroy, /* 0.2.0 */
    .networkGetXMLDesc = networkGetXMLDesc, /* 0.2.0 */
    .networkGetBridgeName = networkGetBridgeName, /* 0.2.0 */
    .networkGetAutostart = networkGetAutostart, /* 0.2.1 */
    .networkSetAutostart = networkSetAutostart, /* 0.2.1 */
    .networkIsActive = networkIsActive, /* 0.7.3 */
    .networkIsPersistent = networkIsPersistent, /* 0.7.3 */
3691
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
3692 3693 3694
};

static virStateDriver networkStateDriver = {
3695
    .name = "bridge",
3696
    .stateInitialize  = networkStateInitialize,
3697
    .stateAutoStart  = networkStateAutoStart,
3698 3699
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3700 3701
};

3702 3703
int networkRegister(void)
{
3704
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
3705
        return -1;
3706 3707
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
3708 3709
    return 0;
}
3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721

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

/* 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:
3722
 * @dom: domain definition that @iface belongs to
3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733
 * @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
3734 3735
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
3736
{
3737
    virDomainNetType actualType = iface->type;
3738 3739
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
3740
    virNetDevBandwidthPtr bandwidth = NULL;
3741 3742 3743
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3744
    virNetworkForwardIfDefPtr dev = NULL;
3745
    size_t i;
3746 3747 3748
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3749
        goto validate;
3750 3751 3752 3753

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

3754
    networkDriverLock();
3755
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
3756
    networkDriverUnlock();
3757
    if (!network) {
3758 3759 3760
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3761
        goto error;
3762 3763
    }
    netdef = network->def;
3764

3765 3766 3767 3768 3769 3770 3771
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

3772 3773 3774
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

3775 3776 3777
    /* 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
3778
     */
3779 3780 3781 3782 3783 3784
    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.
     */
3785 3786 3787 3788 3789 3790

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

3791 3792
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
3793
        goto error;
3794

3795 3796 3797 3798 3799 3800 3801 3802
    /* 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;

3803 3804
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
3805

3806 3807 3808 3809 3810 3811 3812 3813 3814 3815
    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;

3816 3817 3818
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
3819 3820 3821
        /* 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
3822
         */
3823
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
3824

3825
        /* we also store the bridge device and macTableManager settings
3826 3827 3828 3829 3830 3831 3832
         * 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;
3833 3834
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
3835

3836 3837 3838
        if (networkPlugBandwidth(network, iface) < 0)
            goto error;

3839
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
3840
               netdef->bridge) {
3841 3842 3843 3844 3845

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

3846
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
3847 3848
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
3849
            goto error;
3850 3851
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
3852

3853 3854 3855 3856 3857 3858 3859 3860
        /* 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) {
3861
            goto error;
3862 3863 3864 3865 3866 3867 3868 3869 3870 3871
        }
        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);
3872
                goto error;
3873 3874 3875
            }
        }

3876
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3877

3878
        virDomainHostdevSubsysPCIBackendType backend;
3879

3880
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
3881
        if (networkCreateInterfacePool(netdef) < 0)
3882 3883 3884
            goto error;

        /* pick first dev with 0 connections */
3885 3886 3887
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901
                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;
3902
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
3903
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
3904
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
3905

E
Eric Blake 已提交
3906
        switch (netdef->forward.driverName) {
3907
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
3908
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
3909 3910
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
3911
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
3912 3913
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
3914
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925
            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;

3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950
        /* 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;
            }
        }

3951 3952 3953 3954
    } 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)) {
3955 3956 3957 3958 3959 3960

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
3961
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
3962
        switch (netdef->forward.type) {
3963
        case VIR_NETWORK_FORWARD_BRIDGE:
3964
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
3965 3966
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
3967
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
3968 3969
            break;
        case VIR_NETWORK_FORWARD_VEPA:
3970
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
3971 3972
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
3973
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
3974 3975 3976
            break;
        }

3977 3978 3979 3980 3981 3982 3983 3984
        /* 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) {
3985
            goto error;
3986
        }
3987
        virtport = iface->data.network.actual->virtPortProfile;
3988
        if (virtport) {
3989 3990 3991 3992 3993 3994 3995 3996
            /* 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);
3997
                goto error;
3998 3999
            }
        }
4000

4001 4002 4003
        /* 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).
         */
4004
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4005 4006 4007 4008
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4009
            goto error;
4010 4011 4012
        } else {
            /* pick an interface from the pool */

4013
            if (networkCreateInterfacePool(netdef) < 0)
4014 4015
                goto error;

4016 4017 4018 4019 4020
            /* 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.
4021
             */
4022 4023
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4024 4025 4026
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4027

4028
                /* pick first dev with 0 connections */
4029 4030 4031
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4032 4033 4034 4035 4036
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4037
                dev = &netdef->forward.ifs[0];
4038 4039 4040
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4041 4042 4043 4044
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4045 4046 4047 4048
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4049
                goto error;
4050
            }
4051 4052
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4053
                goto error;
4054 4055 4056
        }
    }

4057
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4058
        goto error;
4059

4060
 validate:
4061 4062 4063 4064 4065
    /* 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.
     */

4066
    if (virDomainNetGetActualVlan(iface)) {
4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095
        /* 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);
4096

4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115
        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 */
4116
            netdef->connections--;
4117 4118 4119 4120
            if (dev)
                dev->connections--;
            goto error;
        }
4121 4122
    }

4123
    ret = 0;
4124

4125
 cleanup:
4126 4127
    if (network)
        virNetworkObjUnlock(network);
4128 4129
    return ret;

4130
 error:
4131
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4132 4133 4134
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4135
    goto cleanup;
4136 4137 4138
}

/* networkNotifyActualDevice:
4139
 * @dom: domain definition that @iface belongs to
4140 4141 4142 4143 4144 4145 4146 4147 4148 4149
 * @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
4150 4151
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4152
{
4153
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4154 4155
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4156
    virNetworkForwardIfDefPtr dev = NULL;
4157 4158
    size_t i;
    int ret = -1;
4159 4160

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4161
        return 0;
4162

4163
    networkDriverLock();
4164
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4165
    networkDriverUnlock();
4166
    if (!network) {
4167 4168 4169
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4170 4171 4172 4173
        goto error;
    }
    netdef = network->def;

4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184
    /* 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;

4185
    if (!iface->data.network.actual ||
4186 4187
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4188 4189
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4190 4191
    }

4192
    if (networkCreateInterfacePool(netdef) < 0)
4193
        goto error;
4194

4195
    if (netdef->forward.nifs == 0) {
4196
        virReportError(VIR_ERR_INTERNAL_ERROR,
4197 4198
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4199
                       netdef->name);
4200
        goto error;
4201
    }
4202

4203 4204
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4205

4206 4207 4208 4209 4210 4211 4212 4213 4214
        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 */
4215 4216
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4217
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4218 4219
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4220 4221 4222 4223 4224
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4225
            virReportError(VIR_ERR_INTERNAL_ERROR,
4226 4227
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4228
                           netdef->name, actualDev);
4229
            goto error;
4230 4231
        }

4232
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4233 4234
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4235
         */
4236
        if ((dev->connections > 0) &&
4237 4238
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4239 4240
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4241
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4242
            virReportError(VIR_ERR_INTERNAL_ERROR,
4243 4244
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4245
                           netdef->name, actualDev);
4246
            goto error;
4247
        }
4248

4249
        /* we are now assured of success, so mark the allocation */
4250
        dev->connections++;
4251
        VIR_DEBUG("Using physical device %s, connections %d",
4252
                  dev->device.dev, dev->connections);
4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265

    }  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 */
4266 4267
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4268
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4269
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4270 4271
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4272 4273 4274 4275 4276 4277 4278 4279 4280
                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,
4281 4282 4283 4284
                           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
4285
            goto error;
4286 4287 4288 4289 4290 4291 4292
        }

        /* 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) &&
4293
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309
            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);
4310 4311
    }

4312
 success:
4313 4314 4315
    netdef->connections++;
    VIR_DEBUG("Using network %s, %d connections",
              netdef->name, netdef->connections);
4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326

    /* 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;
    }

4327
    ret = 0;
4328
 cleanup:
4329 4330 4331
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4332

4333
 error:
4334
    goto cleanup;
4335 4336 4337 4338
}


/* networkReleaseActualDevice:
4339
 * @dom: domain definition that @iface belongs to
4340 4341 4342 4343 4344 4345 4346 4347 4348 4349
 * @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
4350 4351
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4352
{
4353
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4354
    virNetworkObjPtr network;
4355
    virNetworkDefPtr netdef;
4356
    virNetworkForwardIfDefPtr dev = NULL;
4357 4358
    size_t i;
    int ret = -1;
4359 4360

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4361
        return 0;
4362

4363
    networkDriverLock();
4364
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4365
    networkDriverUnlock();
4366
    if (!network) {
4367 4368 4369
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4370 4371 4372 4373
        goto error;
    }
    netdef = network->def;

4374 4375
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4376 4377 4378 4379 4380
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) &&
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4381 4382 4383
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4384 4385
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4386 4387
    }

4388
    if (netdef->forward.nifs == 0) {
4389
        virReportError(VIR_ERR_INTERNAL_ERROR,
4390
                       _("network '%s' uses a direct/hostdev mode, but "
4391 4392
                         "has no forward dev and no interface pool"),
                       netdef->name);
4393
        goto error;
4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405
    }

    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;
        }
4406

4407 4408
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4409
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4410 4411
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4412 4413 4414
                break;
            }
        }
4415

4416
        if (!dev) {
4417
            virReportError(VIR_ERR_INTERNAL_ERROR,
4418 4419
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4420
                           netdef->name, actualDev);
4421
            goto error;
4422 4423
        }

4424
        dev->connections--;
4425
        VIR_DEBUG("Releasing physical device %s, connections %d",
4426
                  dev->device.dev, dev->connections);
4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437

    } 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;
        }

4438 4439
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4440
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4441
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4442 4443
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4444 4445 4446 4447 4448 4449 4450 4451 4452
                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,
4453 4454 4455 4456
                           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
4457
            goto error;
4458 4459 4460 4461 4462 4463 4464
        }

        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
4465
    }
4466

4467
 success:
4468
    if (iface->data.network.actual) {
4469
        netdef->connections--;
4470 4471
        VIR_DEBUG("Releasing network %s, %d connections",
                  netdef->name, netdef->connections);
4472

4473 4474 4475 4476
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
    }
4477
    ret = 0;
4478
 cleanup:
4479 4480
    if (network)
        virNetworkObjUnlock(network);
4481 4482 4483 4484
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4485
    return ret;
4486

4487
 error:
4488
    goto cleanup;
4489
}
4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513

/*
 * 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;
4514
    virNetworkObjPtr network;
4515 4516 4517 4518
    virNetworkDefPtr netdef;
    virNetworkIpDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4519
    char *dev_name = NULL;
4520 4521

    *netaddr = NULL;
4522
    networkDriverLock();
4523
    network = virNetworkObjFindByName(driver->networks, netname);
4524
    networkDriverUnlock();
4525
    if (!network) {
4526 4527 4528
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4529
        goto cleanup;
4530 4531 4532
    }
    netdef = network->def;

4533
    switch (netdef->forward.type) {
4534 4535 4536 4537 4538 4539
    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) {
4540 4541 4542
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have an IPv4 address"),
                           netdef->name);
4543
            goto cleanup;
4544 4545 4546 4547 4548
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4549
        if ((dev_name = netdef->bridge))
4550 4551 4552 4553 4554 4555 4556 4557
            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:
4558 4559
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4560

4561
        if (!dev_name) {
4562 4563 4564
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4565
            goto cleanup;
4566 4567 4568 4569
        }
        break;
    }

4570
    if (dev_name) {
4571
        if (virNetDevGetIPv4Address(dev_name, &addr) < 0)
4572
            goto cleanup;
4573
        addrptr = &addr;
4574 4575
    }

4576 4577
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
4578
        goto cleanup;
4579 4580
    }

4581
    ret = 0;
4582
 cleanup:
4583 4584 4585 4586
    if (network)
        virNetworkObjUnlock(network);
    return ret;
}
4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604

/**
 * 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;
4605
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4606 4607 4608 4609
    unsigned long long tmp_floor_sum = net->floor_sum;
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620
    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;
    }

4621
    if (!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor ||
4622 4623
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4624
        return 1;
4625
    }
4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656

    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;

4657
 cleanup:
4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673
    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)
{
4674
    ssize_t ret = 0;
4675

4676
    ret = virBitmapNextClearBit(net->class_id, -1);
4677

4678
    if (ret < 0 || virBitmapSetBit(net->class_id, ret) < 0)
4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692
        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];
4693
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721

    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;
    }

4722 4723
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
4724 4725 4726 4727 4728 4729 4730 4731
    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 */
4732
    net->floor_sum += ifaceBand->in->floor;
4733
    /* update status file */
4734
    if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
4735
        ignore_value(virBitmapClearBit(net->class_id, class_id));
4736
        net->floor_sum -= ifaceBand->in->floor;
4737 4738 4739 4740
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
4741 4742 4743 4744 4745 4746 4747 4748 4749
    /* 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;

4750
 cleanup:
4751 4752 4753 4754 4755 4756 4757 4758 4759
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
    int ret = 0;
    unsigned long long new_rate;
4760
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4761 4762 4763

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
4764 4765 4766 4767 4768
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779
        /* 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 */
4780
        net->floor_sum -= ifaceBand->in->floor;
4781 4782 4783 4784
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
4785
        if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
4786
            net->floor_sum += ifaceBand->in->floor;
4787 4788 4789 4790
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
4791 4792 4793 4794 4795 4796 4797 4798 4799 4800
        /* 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;
    }

4801
 cleanup:
4802 4803
    return ret;
}
4804 4805 4806

static void
networkNetworkObjTaint(virNetworkObjPtr net,
4807
                       virNetworkTaintFlags taint)
4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818
{
    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));
    }
}