bridge_driver.c 178.2 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2016 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
 *
 * 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 <pwd.h>
#include <sys/wait.h>
#include <sys/ioctl.h>
42
#include <net/if.h>
43
#include <dirent.h>
44 45 46
#if HAVE_SYS_SYSCTL_H
# include <sys/sysctl.h>
#endif
47

48
#include "virerror.h"
49
#include "datatypes.h"
50
#include "bridge_driver.h"
51
#include "bridge_driver_platform.h"
52
#include "network_conf.h"
53
#include "device_conf.h"
54
#include "driver.h"
55
#include "virbuffer.h"
56
#include "virpidfile.h"
57
#include "vircommand.h"
58
#include "viralloc.h"
59
#include "viruuid.h"
60
#include "viriptables.h"
61
#include "virlog.h"
62
#include "virdnsmasq.h"
63
#include "configmake.h"
64
#include "virnetdev.h"
65
#include "virnetdevip.h"
66 67
#include "virnetdevbridge.h"
#include "virnetdevtap.h"
68
#include "virnetdevvportprofile.h"
69
#include "virpci.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
#define VIR_FROM_THIS VIR_FROM_NETWORK
79
#define MAX_BRIDGE_ID 256
80

81 82 83 84 85 86 87
/**
 * 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)

88 89
VIR_LOG_INIT("network.bridge_driver");

90
static virNetworkDriverStatePtr network_driver;
91

92 93 94 95 96 97 98 99
static virNetworkDriverStatePtr
networkGetDriver(void)
{
    /* Maybe one day we can store @network_driver in the
     * connection object, but until then, it's just a global
     * variable which is returned. */
    return network_driver;
}
100

101
static void networkDriverLock(virNetworkDriverStatePtr driver)
102
{
103
    virMutexLock(&driver->lock);
104
}
105
static void networkDriverUnlock(virNetworkDriverStatePtr driver)
106
{
107
    virMutexUnlock(&driver->lock);
108 109
}

110 111 112 113
static dnsmasqCapsPtr
networkGetDnsmasqCaps(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr ret;
114
    networkDriverLock(driver);
115
    ret = virObjectRef(driver->dnsmasqCaps);
116
    networkDriverUnlock(driver);
117 118 119 120 121 122 123 124 125 126 127
    return ret;
}

static int
networkDnsmasqCapsRefresh(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr caps;

    if (!(caps = dnsmasqCapsNewFromBinary(DNSMASQ)))
        return -1;

128
    networkDriverLock(driver);
129 130
    virObjectUnref(driver->dnsmasqCaps);
    driver->dnsmasqCaps = caps;
131
    networkDriverUnlock(driver);
132 133 134
    return 0;
}

135
static int networkStateCleanup(void);
136

137 138
static int networkStartNetwork(virNetworkDriverStatePtr driver,
                               virNetworkObjPtr network);
139

140 141
static int networkShutdownNetwork(virNetworkDriverStatePtr driver,
                                  virNetworkObjPtr network);
142

143 144
static int networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
                                      virNetworkObjPtr network);
145

146 147
static int networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
                                         virNetworkObjPtr network);
148

149
static int networkStartNetworkExternal(virNetworkObjPtr network);
150

151
static int networkShutdownNetworkExternal(virNetworkObjPtr network);
152

153 154
static void networkReloadFirewallRules(virNetworkDriverStatePtr driver);
static void networkRefreshDaemons(virNetworkDriverStatePtr driver);
155

156 157 158 159 160
static int networkPlugBandwidth(virNetworkObjPtr net,
                                virDomainNetDefPtr iface);
static int networkUnplugBandwidth(virNetworkObjPtr net,
                                  virDomainNetDefPtr iface);

161
static void networkNetworkObjTaint(virNetworkObjPtr net,
162
                                   virNetworkTaintFlags taint);
163

164 165 166
static virNetworkObjPtr
networkObjFromNetwork(virNetworkPtr net)
{
167
    virNetworkDriverStatePtr driver = networkGetDriver();
168 169 170
    virNetworkObjPtr network;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

171
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
172 173 174 175 176 177 178 179 180 181
    if (!network) {
        virUUIDFormat(net->uuid, uuidstr);
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching uuid '%s' (%s)"),
                       uuidstr, net->name);
    }

    return network;
}

182 183 184
static int
networkRunHook(virNetworkObjPtr network,
               virDomainDefPtr dom,
185
               virDomainNetDefPtr iface,
186 187 188 189 190 191 192 193 194
               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)) {
195 196 197 198 199 200
        if (!network) {
            VIR_DEBUG("Not running hook as @network is NULL");
            ret = 0;
            goto cleanup;
        }

201 202
        virBufferAddLit(&buf, "<hookData>\n");
        virBufferAdjustIndent(&buf, 2);
203
        if (iface && virDomainNetDefFormat(&buf, iface, NULL, 0) < 0)
204
            goto cleanup;
205 206
        if (virNetworkDefFormatBuf(&buf, network->def, 0) < 0)
            goto cleanup;
207
        if (dom && virDomainDefFormatInternal(dom, NULL, 0, &buf) < 0)
208 209 210 211 212
            goto cleanup;

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

213
        if (virBufferCheckError(&buf) < 0)
214 215
            goto cleanup;

216
        xml = virBufferContentAndReset(&buf);
217 218 219 220 221 222 223 224
        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;
225 226

        networkNetworkObjTaint(network, VIR_NETWORK_TAINT_HOOK);
227 228 229
    }

    ret = 0;
230
 cleanup:
231 232 233 234 235 236 237
    virBufferFreeAndReset(&buf);
    VIR_FREE(xml);
    VIR_FREE(net_xml);
    VIR_FREE(dom_xml);
    return ret;
}

238
static char *
239 240
networkDnsmasqLeaseFileNameDefault(virNetworkDriverStatePtr driver,
                                   const char *netname)
241 242 243
{
    char *leasefile;

244
    ignore_value(virAsprintf(&leasefile, "%s/%s.leases",
245
                             driver->dnsmasqStateDir, netname));
246 247 248
    return leasefile;
}

249
static char *
250 251
networkDnsmasqLeaseFileNameCustom(virNetworkDriverStatePtr driver,
                                  const char *bridge)
252 253 254 255
{
    char *leasefile;

    ignore_value(virAsprintf(&leasefile, "%s/%s.status",
256
                             driver->dnsmasqStateDir, bridge));
257 258 259
    return leasefile;
}

260
static char *
261 262
networkDnsmasqConfigFileName(virNetworkDriverStatePtr driver,
                             const char *netname)
263 264 265
{
    char *conffile;

266
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
267
                             driver->dnsmasqStateDir, netname));
268 269 270
    return conffile;
}

271 272 273 274 275 276
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

277
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
278 279 280 281
    return pidfilebase;
}

static char *
282 283
networkRadvdConfigFileName(virNetworkDriverStatePtr driver,
                           const char *netname)
284 285 286
{
    char *configfile;

287
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
288
                             driver->radvdStateDir, netname));
289 290
    return configfile;
}
291

292 293
/* do needed cleanup steps and remove the network from the list */
static int
294 295
networkRemoveInactive(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr net)
296 297
{
    char *leasefile = NULL;
298
    char *customleasefile = NULL;
299
    char *radvdconfigfile = NULL;
300
    char *configfile = NULL;
301
    char *radvdpidbase = NULL;
302
    char *statusfile = NULL;
303 304 305 306 307 308
    dnsmasqContext *dctx = NULL;
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(net);

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
309
    if (!(dctx = dnsmasqContextNew(def->name,
310
                                   driver->dnsmasqStateDir))) {
311
        goto cleanup;
312
    }
313

314
    if (!(leasefile = networkDnsmasqLeaseFileNameDefault(driver, def->name)))
315 316
        goto cleanup;

317
    if (!(customleasefile = networkDnsmasqLeaseFileNameCustom(driver, def->bridge)))
318 319
        goto cleanup;

320
    if (!(radvdconfigfile = networkRadvdConfigFileName(driver, def->name)))
321
        goto cleanup;
322 323

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
324
        goto cleanup;
325

326
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
327
        goto cleanup;
328

329
    if (!(statusfile = virNetworkConfigFile(driver->stateDir, def->name)))
330
        goto cleanup;
331

332 333 334
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
335
    unlink(customleasefile);
336
    unlink(configfile);
337 338 339

    /* radvd */
    unlink(radvdconfigfile);
340
    virPidFileDelete(driver->pidDir, radvdpidbase);
341

342 343 344
    /* remove status file */
    unlink(statusfile);

345
    /* remove the network definition */
346
    virNetworkRemoveInactive(driver->networks, net);
347 348 349

    ret = 0;

350
 cleanup:
351
    VIR_FREE(leasefile);
352
    VIR_FREE(configfile);
353
    VIR_FREE(customleasefile);
354 355
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
356
    VIR_FREE(statusfile);
357 358 359 360
    dnsmasqContextFree(dctx);
    return ret;
}

361 362 363
static char *
networkBridgeDummyNicName(const char *brname)
{
364
    static const char dummyNicSuffix[] = "-nic";
365 366
    char *nicname;

367 368 369 370 371 372 373
    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.
         */
374 375 376 377 378
        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));
379
    } else {
380
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
381
    }
382 383 384
    return nicname;
}

385 386
static int
networkUpdateState(virNetworkObjPtr obj,
387
                   void *opaque)
388
{
389
    virNetworkDriverStatePtr driver = opaque;
390
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
391
    int ret = -1;
392

393
    virObjectLock(obj);
394
    if (!virNetworkObjIsActive(obj)) {
395 396
        ret = 0;
        goto cleanup;
397
    }
398

399 400 401 402
    switch (obj->def->forward.type) {
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
403
    case VIR_NETWORK_FORWARD_OPEN:
404 405 406 407
        /* If bridge doesn't exist, then mark it inactive */
        if (!(obj->def->bridge && virNetDevExists(obj->def->bridge) == 1))
            obj->active = 0;
        break;
408

409
    case VIR_NETWORK_FORWARD_BRIDGE:
410 411
        if (obj->def->bridge) {
            if (virNetDevExists(obj->def->bridge) != 1)
412 413
                obj->active = 0;
            break;
414
        }
415 416 417 418 419 420 421 422 423
        /* 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;
424

425 426 427
    case VIR_NETWORK_FORWARD_HOSTDEV:
        /* so far no extra checks */
        break;
428
    }
429

430 431 432
    /* Try and read dnsmasq/radvd pids of active networks */
    if (obj->active && obj->def->ips && (obj->def->nips > 0)) {
        char *radvdpidbase;
433

434 435 436
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           obj->def->name,
                                           &obj->dnsmasqPid,
437
                                           dnsmasqCapsGetBinaryPath(dnsmasq_caps)));
438 439 440
        radvdpidbase = networkRadvdPidfileBasename(obj->def->name);
        if (!radvdpidbase)
            goto cleanup;
441

442 443 444 445
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           radvdpidbase,
                                           &obj->radvdPid, RADVD));
        VIR_FREE(radvdpidbase);
446
    }
447

448 449
    ret = 0;
 cleanup:
450
    virObjectUnlock(obj);
451
    virObjectUnref(dnsmasq_caps);
452 453
    return ret;
}
454

455

456 457
static int
networkAutostartConfig(virNetworkObjPtr net,
458
                       void *opaque)
459
{
460
    virNetworkDriverStatePtr driver = opaque;
461
    int ret = -1;
462

463
    virObjectLock(net);
464 465
    if (net->autostart &&
        !virNetworkObjIsActive(net) &&
466
        networkStartNetwork(driver, net) < 0)
467 468 469 470
        goto cleanup;

    ret = 0;
 cleanup:
471
    virObjectUnlock(net);
472
    return ret;
473 474
}

475 476 477
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
478
                             DBusMessage *message, void *user_data)
479
{
480 481
    virNetworkDriverStatePtr driver = user_data;

482 483 484 485 486 487
    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.");
488
        networkReloadFirewallRules(driver);
489 490 491 492 493 494
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

495
static int
496
networkMigrateStateFiles(virNetworkDriverStatePtr driver)
497 498 499 500 501 502 503 504 505 506 507 508 509
{
    /* 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;
510
    int direrr;
511 512 513
    struct dirent *entry;
    char *oldPath = NULL, *newPath = NULL;
    char *contents = NULL;
J
Ján Tomko 已提交
514
    int rc;
515

J
Ján Tomko 已提交
516 517
    if ((rc = virDirOpenIfExists(&dir, oldStateDir)) <= 0)
        return rc;
518 519 520 521 522 523 524

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

525
    while ((direrr = virDirRead(dir, &entry, oldStateDir)) > 0) {
526 527 528
        if (entry->d_type != DT_UNKNOWN &&
            entry->d_type != DT_REG)
            continue;
529 530 531

        if (virAsprintf(&oldPath, "%s/%s",
                        oldStateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
532
            goto cleanup;
533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549

        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;
            }
        }

550
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
J
Ján Tomko 已提交
551
            goto cleanup;
552 553 554

        if (virAsprintf(&newPath, "%s/%s",
                        driver->stateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
555
            goto cleanup;
556 557 558 559 560 561 562 563 564 565 566 567
        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);
    }
568
    if (direrr < 0)
J
Ján Tomko 已提交
569
        goto cleanup;
570 571 572

    ret = 0;
 cleanup:
J
Ján Tomko 已提交
573
    VIR_DIR_CLOSE(dir);
574 575 576 577 578 579
    VIR_FREE(oldPath);
    VIR_FREE(newPath);
    VIR_FREE(contents);
    return ret;
}

580
/**
581
 * networkStateInitialize:
582 583 584 585
 *
 * Initialization function for the QEmu daemon
 */
static int
586 587 588
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
589
{
590 591 592
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
593 594 595
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
596

597
    if (VIR_ALLOC(network_driver) < 0)
598
        goto error;
599

600 601
    if (virMutexInit(&network_driver->lock) < 0) {
        VIR_FREE(network_driver);
602 603
        goto error;
    }
604

605 606 607 608
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
609
    if (privileged) {
610
        if (VIR_STRDUP(network_driver->networkConfigDir,
611
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
612
            VIR_STRDUP(network_driver->networkAutostartDir,
613
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
614
            VIR_STRDUP(network_driver->stateDir,
615
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
616
            VIR_STRDUP(network_driver->pidDir,
617
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
618
            VIR_STRDUP(network_driver->dnsmasqStateDir,
619
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
620
            VIR_STRDUP(network_driver->radvdStateDir,
621 622
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
623 624 625 626 627

        /* migration from old to new location is only applicable for
         * privileged mode - unprivileged mode directories haven't
         * changed location.
         */
628
        if (networkMigrateStateFiles(network_driver) < 0)
629
            goto error;
630
    } else {
631 632 633
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
634
            goto error;
635

636
        if ((virAsprintf(&network_driver->networkConfigDir,
637
                         "%s/qemu/networks", configdir) < 0) ||
638
            (virAsprintf(&network_driver->networkAutostartDir,
639
                         "%s/qemu/networks/autostart", configdir) < 0) ||
640
            (virAsprintf(&network_driver->stateDir,
641
                         "%s/network/lib", rundir) < 0) ||
642
            (virAsprintf(&network_driver->pidDir,
643
                         "%s/network/run", rundir) < 0) ||
644
            (virAsprintf(&network_driver->dnsmasqStateDir,
645
                         "%s/dnsmasq/lib", rundir) < 0) ||
646
            (virAsprintf(&network_driver->radvdStateDir,
647
                         "%s/radvd/lib", rundir) < 0)) {
648
            goto error;
649
        }
650 651
    }

652
    if (virFileMakePath(network_driver->stateDir) < 0) {
653 654
        virReportSystemError(errno,
                             _("cannot create directory %s"),
655
                             network_driver->stateDir);
656 657 658
        goto error;
    }

659
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
660
    network_driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
661

662
    if (!(network_driver->networks = virNetworkObjListNew()))
663 664
        goto error;

665 666
    if (virNetworkLoadAllState(network_driver->networks,
                               network_driver->stateDir) < 0)
667 668
        goto error;

669 670 671
    if (virNetworkLoadAllConfigs(network_driver->networks,
                                 network_driver->networkConfigDir,
                                 network_driver->networkAutostartDir) < 0)
672 673
        goto error;

674 675 676 677
    /* 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). */
678
    virNetworkObjListForEach(network_driver->networks,
679
                             networkUpdateState,
680 681
                             network_driver);
    virNetworkObjListPrune(network_driver->networks,
682 683
                           VIR_CONNECT_LIST_NETWORKS_INACTIVE |
                           VIR_CONNECT_LIST_NETWORKS_TRANSIENT);
684 685
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
686

687
    network_driver->networkEventState = virObjectEventStateNew();
688

689 690 691
#ifdef HAVE_FIREWALLD
    if (!(sysbus = virDBusGetSystemBus())) {
        VIR_WARN("DBus not available, disabling firewalld support "
692
                 "in bridge_network_driver: %s", virGetLastErrorMessage());
693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709
    } 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,
710
                                   network_driver, NULL);
711 712 713
    }
#endif

714
    ret = 0;
715
 cleanup:
716 717 718
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
719

720
 error:
721 722
    if (network_driver)
        networkDriverUnlock(network_driver);
723
    networkStateCleanup();
724
    goto cleanup;
725 726
}

727 728 729 730 731 732 733 734
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
735
    if (!network_driver)
736 737
        return;

738
    virNetworkObjListForEach(network_driver->networks,
739
                             networkAutostartConfig,
740
                             network_driver);
741 742
}

743
/**
744
 * networkStateReload:
745 746 747 748 749
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
750 751
networkStateReload(void)
{
752
    if (!network_driver)
753 754
        return 0;

755 756 757 758 759 760 761 762
    virNetworkLoadAllState(network_driver->networks,
                           network_driver->stateDir);
    virNetworkLoadAllConfigs(network_driver->networks,
                             network_driver->networkConfigDir,
                             network_driver->networkAutostartDir);
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
    virNetworkObjListForEach(network_driver->networks,
763
                             networkAutostartConfig,
764
                             network_driver);
765 766 767 768 769
    return 0;
}


/**
770
 * networkStateCleanup:
771 772 773 774
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
775 776
networkStateCleanup(void)
{
777
    if (!network_driver)
778 779
        return -1;

780
    virObjectUnref(network_driver->networkEventState);
781

782
    /* free inactive networks */
783
    virObjectUnref(network_driver->networks);
784

785 786 787 788 789 790
    VIR_FREE(network_driver->networkConfigDir);
    VIR_FREE(network_driver->networkAutostartDir);
    VIR_FREE(network_driver->stateDir);
    VIR_FREE(network_driver->pidDir);
    VIR_FREE(network_driver->dnsmasqStateDir);
    VIR_FREE(network_driver->radvdStateDir);
791

792
    virObjectUnref(network_driver->dnsmasqCaps);
793

794
    virMutexDestroy(&network_driver->lock);
795

796
    VIR_FREE(network_driver);
797 798 799 800 801

    return 0;
}


802 803 804 805 806 807 808
/* 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)
{
809 810
    size_t i;
    int ret = -1;
811 812 813 814 815 816 817
    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.
     */
818
    for (i = 0; i < 25; i++) {
819
        int signum = 0;
820
        if (i == 0) {
821
            signum = SIGTERM;
822
        } else if (i == 15) {
823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857
            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);
858
 cleanup:
859 860 861
    return ret;
}

J
Ján Tomko 已提交
862 863 864
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
G
Gene Czarcinski 已提交
865

866
static int
G
Gene Czarcinski 已提交
867
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
868
                                 virNetworkIPDefPtr ipdef)
869
{
870
    size_t i;
G
Gene Czarcinski 已提交
871
    bool ipv6 = false;
872

G
Gene Czarcinski 已提交
873 874
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
875 876
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
877
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
878 879
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
880
                return -1;
881
    }
882

G
Gene Czarcinski 已提交
883 884 885 886 887 888 889
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
890
    size_t i, j;
G
Gene Czarcinski 已提交
891

892 893
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
894
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
895
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
896
                for (j = 0; j < host->nnames; j++)
897 898
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
899 900
            }
        }
901 902
    }

903
    return 0;
904 905 906
}


907 908
int
networkDnsmasqConfContents(virNetworkObjPtr network,
909 910 911 912
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
913
{
914
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
915
    int r, ret = -1;
916
    int nbleases = 0;
917
    size_t i;
918
    virNetworkDNSDefPtr dns = &network->def->dns;
919
    bool wantDNS = dns->enable != VIR_TRISTATE_BOOL_NO;
920
    virNetworkIPDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
G
Gene Czarcinski 已提交
921
    bool ipv6SLAAC;
922
    char *saddr = NULL, *eaddr = NULL;
923

924 925
    *configstr = NULL;

926
    /*
927 928 929
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
930
     *
931 932
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
933
     */
934 935 936 937 938 939

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

940 941
    /* create dnsmasq config file appropriate for this network */
    virBufferAsprintf(&configbuf,
942 943 944 945 946 947 948
                      "##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"
949
                      "strict-order\n",
950 951
                      network->def->name);

952 953 954 955 956 957 958
    /* if dns is disabled, set its listening port to 0, which
     * tells dnsmasq to not listen
     */
    if (!wantDNS)
        virBufferAddLit(&configbuf, "port=0\n");

    if (wantDNS && network->def->dns.forwarders) {
959 960
        virBufferAddLit(&configbuf, "no-resolv\n");
        for (i = 0; i < network->def->dns.nfwds; i++) {
961 962 963 964 965 966 967 968 969 970 971
            virNetworkDNSForwarderPtr fwd = &network->def->dns.forwarders[i];

            virBufferAddLit(&configbuf, "server=");
            if (fwd->domain)
                virBufferAsprintf(&configbuf, "/%s/", fwd->domain);
            if (VIR_SOCKET_ADDR_VALID(&fwd->addr)) {
                char *addr = virSocketAddrFormat(&fwd->addr);

                if (!addr)
                    goto cleanup;
                virBufferAsprintf(&configbuf, "%s\n", addr);
972
                VIR_FREE(addr);
973 974 975 976
            } else {
                /* "don't forward requests for this domain" */
                virBufferAddLit(&configbuf, "#\n");
            }
977 978 979
        }
    }

980
    if (network->def->domain) {
981 982 983 984 985
        if (network->def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
                              network->def->domain);
        }
986
        virBufferAsprintf(&configbuf,
987 988 989 990
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
991

992
    if (wantDNS && network->def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
993 994 995 996
        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)
997
         */
998
        virBufferAddLit(&configbuf, "local=//\n");
999
    }
1000

1001
    if (pidfile)
1002
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
1003

1004 1005 1006
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

1007 1008 1009 1010 1011 1012 1013 1014
    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.
         */
1015
        virBufferAsprintf(&configbuf,
1016 1017 1018
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
1019
    } else {
1020
        virBufferAddLit(&configbuf, "bind-interfaces\n");
1021 1022 1023 1024 1025 1026 1027 1028
        /*
         * --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
         */
1029
        for (i = 0;
1030
             (tmpipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i));
1031
             i++) {
1032 1033 1034 1035
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
1036

1037
            /* also part of CVE 2012-3411 - if the host's version of
1038
             * dnsmasq doesn't have bind-dynamic, only allow listening on
1039 1040
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
1041 1042
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
1043 1044 1045 1046
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
1047
                                 "The version of dnsmasq on this host (%d.%d) "
1048 1049 1050 1051
                                 "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 "
1052 1053 1054 1055 1056 1057
                                 "(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);
1058
                VIR_FREE(ipaddr);
1059 1060
                goto cleanup;
            }
1061
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1062 1063 1064
            VIR_FREE(ipaddr);
        }
    }
1065

1066 1067
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1068
     * guaranteed to not work, and set no-resolv so that no dns
1069 1070 1071
     * 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).
1072 1073 1074
     * IPv6 RA always contains an implicit default route
     * via the sender's link-local address. The only thing we can do
     * is set the lifetime of this route to 0, i.e. disable it.
1075
     */
1076
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1077
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1078
                        "no-resolv\n");
1079 1080 1081 1082
        if (dnsmasqCapsGet(caps, DNSMASQ_CAPS_RA_PARAM)) {
            /* interface=* (any), interval=0 (default), lifetime=0 (seconds) */
            virBufferAddLit(&configbuf, "ra-param=*,0,0\n");
        }
1083
    }
1084

1085 1086 1087 1088 1089
    if (wantDNS) {
        for (i = 0; i < dns->ntxts; i++) {
            virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
                              dns->txts[i].name,
                              dns->txts[i].value);
1090
        }
1091

1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114
        for (i = 0; i < dns->nsrvs; i++) {
            /* 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);
1115

1116 1117 1118 1119 1120 1121 1122
            /* domain is optional - it defaults to the domain of this network */
            if (dns->srvs[i].domain)
                virBufferAsprintf(&configbuf, ".%s", dns->srvs[i].domain);

            /* 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.
1123
             */
1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143
            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);
            }
            virBufferAddLit(&configbuf, "\n");
1144
        }
1145 1146
    }

G
Gene Czarcinski 已提交
1147
    /* Find the first dhcp for both IPv4 and IPv6 */
1148
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
1149
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i));
1150
         i++) {
G
Gene Czarcinski 已提交
1151 1152 1153 1154
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1155 1156
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167
                    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,
1168 1169 1170 1171 1172 1173 1174 1175 1176
                                   _("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
1177 1178 1179 1180
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1181 1182
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195
                    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 "
1196 1197 1198 1199
                 "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
1200 1201 1202 1203 1204
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1205 1206
        int prefix;

1207
        prefix = virNetworkIPDefPrefix(ipdef);
1208 1209 1210 1211 1212 1213
        if (prefix < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
            goto cleanup;
        }
1214
        for (r = 0; r < ipdef->nranges; r++) {
1215 1216
            int thisRange;

1217 1218
            if (!(saddr = virSocketAddrFormat(&ipdef->ranges[r].start)) ||
                !(eaddr = virSocketAddrFormat(&ipdef->ranges[r].end)))
1219
                goto cleanup;
1220

1221
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s",
1222
                              saddr, eaddr);
1223
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1224
                virBufferAsprintf(&configbuf, ",%d", prefix);
1225 1226
            virBufferAddLit(&configbuf, "\n");

1227
            VIR_FREE(saddr);
1228
            VIR_FREE(eaddr);
1229
            thisRange = virSocketAddrGetRange(&ipdef->ranges[r].start,
1230 1231
                                              &ipdef->ranges[r].end,
                                              &ipdef->address,
1232
                                              virNetworkIPDefPrefix(ipdef));
1233 1234 1235
            if (thisRange < 0)
                goto cleanup;
            nbleases += thisRange;
1236
        }
1237

1238
        /*
1239 1240 1241 1242
         * 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)
1243 1244
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1245
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1246 1247
            if (!bridgeaddr)
                goto cleanup;
1248 1249 1250
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static",
                              bridgeaddr);
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1251
                virBufferAsprintf(&configbuf, ",%d", prefix);
1252
            virBufferAddLit(&configbuf, "\n");
1253 1254
            VIR_FREE(bridgeaddr);
        }
1255

G
Gene Czarcinski 已提交
1256 1257
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1258

G
Gene Czarcinski 已提交
1259 1260
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
1261
            if (ipdef->nranges || ipdef->nhosts) {
1262
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1263 1264
                virBufferAddLit(&configbuf, "dhcp-authoritative\n");
            }
1265

G
Gene Czarcinski 已提交
1266
            if (ipdef->tftproot) {
1267 1268
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1269
            }
1270

G
Gene Czarcinski 已提交
1271 1272 1273
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1274

1275
                    if (!bootserver)
G
Gene Czarcinski 已提交
1276
                        goto cleanup;
1277
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1278
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1279 1280
                    VIR_FREE(bootserver);
                } else {
1281
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1282 1283 1284 1285 1286
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1287

1288
    if (nbleases > 0)
1289
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1290

G
Gene Czarcinski 已提交
1291 1292
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1293
        goto cleanup;
G
Gene Czarcinski 已提交
1294 1295 1296 1297 1298 1299

    /* 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)
1300 1301
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1302

1303 1304
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1305
     */
1306 1307 1308 1309
    if (wantDNS) {
        virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
                          dctx->addnhostsfile->path);
    }
G
Gene Czarcinski 已提交
1310 1311 1312

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1313
        if (ipv6def) {
1314
            virBufferAddLit(&configbuf, "enable-ra\n");
1315
        } else {
1316
            for (i = 0;
1317
                 (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1318
                 i++) {
G
Gene Czarcinski 已提交
1319 1320 1321 1322
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1323 1324
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1325 1326
                    VIR_FREE(bridgeaddr);
                }
1327
            }
1328
        }
1329 1330
    }

1331 1332 1333
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1334
    ret = 0;
G
Gene Czarcinski 已提交
1335

1336
 cleanup:
1337 1338
    VIR_FREE(saddr);
    VIR_FREE(eaddr);
1339
    virBufferFreeAndReset(&configbuf);
1340
    return ret;
1341 1342
}

1343
/* build the dnsmasq command line */
1344 1345 1346
static int ATTRIBUTE_NONNULL(3)
networkBuildDhcpDaemonCommandLine(virNetworkDriverStatePtr driver,
                                  virNetworkObjPtr network,
1347
                                  virCommandPtr *cmdout,
1348 1349
                                  char *pidfile,
                                  dnsmasqContext *dctx)
1350
{
1351
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1352
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1353
    int ret = -1;
1354 1355
    char *configfile = NULL;
    char *configstr = NULL;
1356
    char *leaseshelper_path = NULL;
1357 1358

    network->dnsmasqPid = -1;
1359

1360
    if (networkDnsmasqConfContents(network, pidfile, &configstr,
1361
                                   dctx, dnsmasq_caps) < 0)
1362 1363 1364 1365 1366
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1367
    if (!(configfile = networkDnsmasqConfigFileName(driver, network->def->name)))
1368 1369 1370 1371 1372
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1373 1374
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1375 1376 1377
        goto cleanup;
    }

1378 1379
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1380
                                                  abs_topbuilddir "/src",
1381 1382 1383
                                                  LIBEXECDIR)))
        goto cleanup;

1384
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(dnsmasq_caps));
1385
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1386 1387
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1388
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1389
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", network->def->bridge);
1390

1391
    *cmdout = cmd;
1392
    ret = 0;
1393
 cleanup:
1394
    virObjectUnref(dnsmasq_caps);
1395 1396
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1397
    VIR_FREE(leaseshelper_path);
1398 1399 1400 1401
    return ret;
}

static int
1402 1403
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
1404
{
1405 1406 1407
    virNetworkIPDefPtr ipdef;
    size_t i;
    bool needDnsmasq = false;
1408 1409 1410
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1411
    dnsmasqContext *dctx = NULL;
1412

1413
    if (!(ipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, 0))) {
G
Gene Czarcinski 已提交
1414
        /* no IP addresses, so we don't need to run */
1415 1416 1417 1418
        ret = 0;
        goto cleanup;
    }

1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430
    /* see if there are any IP addresses that need a dhcp server */
    for (i = 0; ipdef && !needDnsmasq;
         ipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i + 1)) {
        if (ipdef->nranges || ipdef->nhosts)
            needDnsmasq = true;
    }

    if (!needDnsmasq && network->def->dns.enable == VIR_TRISTATE_BOOL_NO) {
        ret = 0;
        goto cleanup;
    }

1431
    if (virFileMakePath(driver->pidDir) < 0) {
1432
        virReportSystemError(errno,
1433
                             _("cannot create directory %s"),
1434
                             driver->pidDir);
1435
        goto cleanup;
1436 1437
    }

1438
    if (!(pidfile = virPidFileBuildPath(driver->pidDir,
1439
                                        network->def->name)))
1440
        goto cleanup;
1441

1442
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1443
        virReportSystemError(errno,
1444
                             _("cannot create directory %s"),
1445
                             driver->dnsmasqStateDir);
1446 1447 1448
        goto cleanup;
    }

1449
    dctx = dnsmasqContextNew(network->def->name, driver->dnsmasqStateDir);
1450 1451 1452
    if (dctx == NULL)
        goto cleanup;

1453
    if (networkDnsmasqCapsRefresh(driver) < 0)
1454
        goto cleanup;
1455

1456
    ret = networkBuildDhcpDaemonCommandLine(driver, network, &cmd, pidfile, dctx);
1457 1458 1459 1460 1461
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1462
        goto cleanup;
1463

G
Guido Günther 已提交
1464
    ret = virCommandRun(cmd, NULL);
1465
    if (ret < 0)
1466 1467 1468
        goto cleanup;

    /*
1469 1470 1471 1472 1473
     * 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
1474 1475
     */

1476
    ret = virPidFileRead(driver->pidDir, network->def->name,
1477 1478
                         &network->dnsmasqPid);
    if (ret < 0)
1479
        goto cleanup;
1480

1481
    ret = 0;
1482
 cleanup:
1483
    VIR_FREE(pidfile);
1484
    virCommandFree(cmd);
1485
    dnsmasqContextFree(dctx);
1486 1487 1488
    return ret;
}

1489 1490
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1491 1492
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1493 1494 1495
 *
 *  Returns 0 on success, -1 on failure.
 */
1496
static int
1497 1498
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1499
{
1500 1501
    int ret = -1;
    size_t i;
1502
    virNetworkIPDefPtr ipdef, ipv4def, ipv6def;
1503
    dnsmasqContext *dctx = NULL;
1504

G
Gene Czarcinski 已提交
1505
    /* if no IP addresses specified, nothing to do */
1506
    if (!virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1507 1508
        return 0;

1509 1510
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1511
        return networkStartDhcpDaemon(driver, network);
1512

G
Gene Czarcinski 已提交
1513
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1514
    if (!(dctx = dnsmasqContextNew(network->def->name,
1515
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1516
        goto cleanup;
1517
    }
G
Gene Czarcinski 已提交
1518 1519 1520 1521 1522 1523

    /* 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;
1524
    for (i = 0;
1525
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET, i));
1526
         i++) {
G
Gene Czarcinski 已提交
1527 1528
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1529 1530
    }

G
Gene Czarcinski 已提交
1531
    ipv6def = NULL;
1532
    for (i = 0;
1533
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1534
         i++) {
G
Gene Czarcinski 已提交
1535 1536
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1537 1538
    }

G
Gene Czarcinski 已提交
1539
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1540
        goto cleanup;
G
Gene Czarcinski 已提交
1541 1542

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1543
        goto cleanup;
1544

G
Gene Czarcinski 已提交
1545
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
J
Ján Tomko 已提交
1546
        goto cleanup;
1547 1548

    if ((ret = dnsmasqSave(dctx)) < 0)
1549
        goto cleanup;
1550 1551

    ret = kill(network->dnsmasqPid, SIGHUP);
1552
 cleanup:
1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564
    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
1565 1566
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1567 1568 1569 1570 1571 1572
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1573
    }
1574
    /* now start dnsmasq if it should be started */
1575
    return networkStartDhcpDaemon(driver, network);
1576 1577
}

G
Gene Czarcinski 已提交
1578 1579 1580 1581 1582 1583
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";

1584 1585 1586
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1587
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1588 1589
    int ret = -1;
    size_t i;
1590
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
1591
    bool v6present = false, dhcp6 = false;
1592 1593

    *configstr = NULL;
1594

G
Gene Czarcinski 已提交
1595
    /* Check if DHCPv6 is needed */
1596
    for (i = 0;
1597
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1598
         i++) {
G
Gene Czarcinski 已提交
1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611
        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;
    }

1612 1613 1614
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1615
    virBufferAsprintf(&configbuf, "interface %s\n"
1616 1617
                      "{\n"
                      "  AdvSendAdvert on;\n"
1618
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1619 1620 1621 1622 1623
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1624 1625

    /* add a section for each IPv6 address in the config */
1626
    for (i = 0;
1627
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1628
         i++) {
1629 1630 1631
        int prefix;
        char *netaddr;

1632
        prefix = virNetworkIPDefPrefix(ipdef);
1633
        if (prefix < 0) {
1634 1635 1636
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1637 1638
            goto cleanup;
        }
1639
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1640
            goto cleanup;
1641
        virBufferAsprintf(&configbuf,
1642
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1643 1644 1645
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1646 1647 1648
        VIR_FREE(netaddr);
    }

1649
    virBufferAddLit(&configbuf, "};\n");
1650

1651
    if (virBufferCheckError(&configbuf) < 0)
1652
        goto cleanup;
1653

1654 1655
    *configstr = virBufferContentAndReset(&configbuf);

1656
    ret = 0;
1657
 cleanup:
1658 1659 1660 1661
    virBufferFreeAndReset(&configbuf);
    return ret;
}

1662
/* write file and return its name (which must be freed by caller) */
1663
static int
1664 1665 1666
networkRadvdConfWrite(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr network,
                      char **configFile)
1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681
{
    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;
1682 1683 1684 1685
        goto cleanup;
    }

    /* construct the filename */
1686
    if (!(*configFile = networkRadvdConfigFileName(driver, network->def->name)))
1687 1688
        goto cleanup;
    /* write the file */
1689
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1690 1691
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1692 1693 1694 1695 1696
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1697
 cleanup:
1698 1699 1700 1701 1702 1703
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1704 1705
networkStartRadvd(virNetworkDriverStatePtr driver,
                  virNetworkObjPtr network)
1706
{
1707
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1708 1709 1710 1711 1712 1713 1714 1715
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1716
    /* Is dnsmasq handling RA? */
1717
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
G
Gene Czarcinski 已提交
1718 1719 1720 1721
        ret = 0;
        goto cleanup;
    }

1722
    if (!virNetworkDefGetIPByIndex(network->def, AF_INET6, 0)) {
1723 1724 1725 1726 1727 1728 1729 1730 1731 1732
        /* 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);
1733 1734 1735
        goto cleanup;
    }

1736
    if (virFileMakePath(driver->pidDir) < 0) {
1737 1738
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1739
                             driver->pidDir);
1740 1741
        goto cleanup;
    }
1742
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1743 1744
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1745
                             driver->radvdStateDir);
1746 1747 1748 1749
        goto cleanup;
    }

    /* construct pidfile name */
1750
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1751
        goto cleanup;
1752
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1753 1754
        goto cleanup;

1755
    if (networkRadvdConfWrite(driver, network, &configfile) < 0)
1756 1757
        goto cleanup;

1758 1759 1760 1761
    /* 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
1762
     * virPidFileRead() below will fail if we use them).
1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777
     * 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;

1778
    if (virPidFileRead(driver->pidDir, radvdpidbase, &network->radvdPid) < 0)
1779 1780 1781
        goto cleanup;

    ret = 0;
1782
 cleanup:
1783
    virObjectUnref(dnsmasq_caps);
1784 1785 1786 1787 1788 1789 1790
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1791
static int
1792 1793
networkRefreshRadvd(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
1794
{
1795
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
G
Gene Czarcinski 已提交
1796 1797 1798
    char *radvdpidbase;

    /* Is dnsmasq handling RA? */
1799 1800
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
        virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1801 1802 1803 1804 1805 1806 1807
        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)) {
1808
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1809 1810 1811 1812 1813
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }
1814
    virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1815

1816 1817
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
1818
        return networkStartRadvd(driver, network);
1819

1820
    if (!virNetworkDefGetIPByIndex(network->def, AF_INET6, 0)) {
1821 1822 1823 1824
        /* no IPv6 addresses, so we don't need to run radvd */
        return 0;
    }

1825
    if (networkRadvdConfWrite(driver, network, NULL) < 0)
1826 1827 1828 1829 1830
        return -1;

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

1831 1832
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1833
static int
1834
networkRestartRadvd(virNetworkObjPtr network)
1835 1836 1837 1838 1839 1840 1841 1842 1843
{
    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
1844
        if ((networkKillDaemon(network->radvdPid, "radvd",
1845 1846 1847
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1848
            virPidFileDelete(driver->pidDir, radvdpidbase);
1849 1850 1851 1852 1853 1854 1855 1856 1857
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1858 1859
static int
networkRefreshDaemonsHelper(virNetworkObjPtr net,
1860
                            void *opaque)
1861
{
1862
    virNetworkDriverStatePtr driver = opaque;
1863

1864
    virObjectLock(net);
1865 1866 1867
    if (virNetworkObjIsActive(net) &&
        ((net->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
1868 1869
         (net->def->forward.type == VIR_NETWORK_FORWARD_ROUTE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_OPEN))) {
1870 1871 1872 1873 1874 1875
        /* 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.
         */
1876 1877
        networkRefreshDhcpDaemon(driver, net);
        networkRefreshRadvd(driver, net);
1878
    }
1879
    virObjectUnlock(net);
1880 1881 1882
    return 0;
}

1883 1884 1885 1886
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
1887
networkRefreshDaemons(virNetworkDriverStatePtr driver)
1888 1889
{
    VIR_INFO("Refreshing network daemons");
1890 1891
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
1892
                             driver);
1893
}
1894

1895 1896 1897 1898 1899
static int
networkReloadFirewallRulesHelper(virNetworkObjPtr net,
                                 void *opaque ATTRIBUTE_UNUSED)
{

1900
    virObjectLock(net);
1901 1902 1903 1904
    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))) {
1905 1906 1907 1908
        /* Only three of the L3 network types that are configured by
         * libvirt need to have iptables rules reloaded. The 4th L3
         * network type, forward='open', doesn't need this because it
         * has no iptables rules.
1909 1910 1911 1912
         */
        networkRemoveFirewallRules(net->def);
        if (networkAddFirewallRules(net->def) < 0) {
            /* failed to add but already logged */
1913 1914
        }
    }
1915
    virObjectUnlock(net);
1916
    return 0;
1917 1918
}

1919
static void
1920
networkReloadFirewallRules(virNetworkDriverStatePtr driver)
1921
{
1922
    VIR_INFO("Reloading iptables rules");
1923 1924 1925
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
1926 1927
}

1928
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
1929
static int
1930
networkEnableIPForwarding(bool enableIPv4, bool enableIPv6)
1931
{
1932
    int ret = 0;
1933 1934 1935 1936
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
1937
                           &enabled, sizeof(enabled));
1938 1939
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
1940
                           &enabled, sizeof(enabled));
1941
#else
1942 1943 1944 1945
    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);
1946
#endif
1947
    return ret;
1948 1949
}

1950 1951
#define SYSCTL_PATH "/proc/sys"

1952 1953
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
1954 1955 1956
{
    char *field = NULL;
    int ret = -1;
1957
    bool enableIPv6 =  !!virNetworkDefGetIPByIndex(network->def, AF_INET6, 0);
1958

1959 1960 1961 1962 1963 1964 1965
    /* 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;
1966

1967 1968
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
1969 1970
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
1971 1972 1973
        ret = 0;
        goto cleanup;
    }
1974

1975 1976 1977 1978 1979
    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;
1980
    }
1981
    VIR_FREE(field);
1982

1983 1984
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
1985 1986 1987 1988 1989 1990
     */

    /* 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",
1991
                    network->def->bridge) < 0)
1992 1993
        goto cleanup;

1994
    if (virFileWriteStr(field, "0", 0) < 0) {
1995
        virReportSystemError(errno,
1996 1997 1998 1999 2000
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

2001 2002 2003 2004
    /* 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",
2005
                    network->def->bridge) < 0)
2006 2007
        goto cleanup;

2008
    if (virFileWriteStr(field, "0", 0) < 0) {
2009
        virReportSystemError(errno,
2010
                             _("cannot disable %s"), field);
2011 2012 2013 2014
        goto cleanup;
    }

    ret = 0;
2015
 cleanup:
2016 2017 2018 2019
    VIR_FREE(field);
    return ret;
}

2020
/* add an IP address to a bridge */
2021
static int
D
Daniel P. Berrange 已提交
2022
networkAddAddrToBridge(virNetworkObjPtr network,
2023
                       virNetworkIPDefPtr ipdef)
2024
{
2025
    int prefix = virNetworkIPDefPrefix(ipdef);
2026 2027

    if (prefix < 0) {
2028 2029 2030
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
2031 2032 2033
        return -1;
    }

2034 2035
    if (virNetDevIPAddrAdd(network->def->bridge,
                           &ipdef->address, NULL, prefix) < 0)
2036 2037 2038 2039 2040
        return -1;

    return 0;
}

2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063

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;
}


2064 2065 2066
/* add an IP (static) route to a bridge */
static int
networkAddRouteToBridge(virNetworkObjPtr network,
2067
                        virNetDevIPRoutePtr routedef)
2068
{
2069 2070 2071 2072
    int prefix = virNetDevIPRouteGetPrefix(routedef);
    unsigned int metric = virNetDevIPRouteGetMetric(routedef);
    virSocketAddrPtr addr = virNetDevIPRouteGetAddress(routedef);
    virSocketAddrPtr gateway = virNetDevIPRouteGetGateway(routedef);
2073 2074 2075 2076 2077 2078 2079 2080 2081

    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;
    }

2082 2083
    if (virNetDevIPRouteAdd(network->def->bridge, addr,
                            prefix, gateway, metric) < 0) {
2084 2085 2086 2087 2088
        return -1;
    }
    return 0;
}

2089 2090 2091
static int
networkWaitDadFinish(virNetworkObjPtr network)
{
2092
    virNetworkIPDefPtr ipdef;
2093 2094 2095 2096 2097 2098
    virSocketAddrPtr *addrs = NULL, addr = NULL;
    size_t naddrs = 0;
    int ret = -1;

    VIR_DEBUG("Begin waiting for IPv6 DAD on network %s", network->def->name);

2099
    while ((ipdef = virNetworkDefGetIPByIndex(network->def,
2100 2101 2102 2103 2104 2105
                                              AF_INET6, naddrs))) {
        addr = &ipdef->address;
        if (VIR_APPEND_ELEMENT_COPY(addrs, naddrs, addr) < 0)
            goto cleanup;
    }

2106
    ret = (naddrs == 0) ? 0 : virNetDevIPWaitDadFinish(addrs, naddrs);
2107 2108 2109 2110 2111 2112 2113 2114

 cleanup:
    VIR_FREE(addrs);
    VIR_DEBUG("Finished waiting for IPv6 DAD on network %s with status %d",
              network->def->name, ret);
    return ret;
}

2115
static int
2116 2117
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
                           virNetworkObjPtr network)
2118
{
2119
    size_t i;
2120
    bool v4present = false, v6present = false;
2121
    virErrorPtr save_err = NULL;
2122
    virNetworkIPDefPtr ipdef;
2123
    virNetDevIPRoutePtr routedef;
2124
    char *macTapIfName = NULL;
2125
    int tapfd = -1;
2126

2127
    /* Check to see if any network IP collides with an existing route */
2128
    if (networkCheckRouteCollision(network->def) < 0)
2129 2130
        return -1;

2131
    /* Create and configure the bridge device */
2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145
    if (!network->def->bridge) {
        /* bridge name can only be empty if the config files were
         * edited directly. Otherwise networkValidate() (called after
         * parsing the XML from networkCreateXML() and
         * networkDefine()) guarantees we will have a valid bridge
         * name before this point. Since hand editing of the config
         * files is explicitly prohibited we can, with clear
         * conscience, log an error and fail at this point.
         */
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' has no bridge name defined"),
                       network->def->name);
        return -1;
    }
2146
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
2147 2148
        return -1;

2149 2150 2151 2152 2153 2154 2155 2156
    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);
2157
        if (!macTapIfName)
2158
            goto err0;
2159
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2160
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
2161
                                           &macTapIfName, &network->def->mac,
2162
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2163 2164 2165
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2166 2167 2168 2169 2170
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2171
    /* Set bridge options */
2172 2173 2174 2175

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2176
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2177
                                   network->def->delay * 1000) < 0)
2178
        goto err1;
2179

2180
    if (virNetDevBridgeSetSTP(network->def->bridge,
2181
                              network->def->stp ? true : false) < 0)
2182
        goto err1;
2183

2184 2185 2186 2187
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2188
        goto err1;
2189

2190
    /* Add "once per network" rules */
2191 2192
    if (network->def->forward.type != VIR_NETWORK_FORWARD_OPEN &&
        networkAddFirewallRules(network->def) < 0)
2193 2194
        goto err1;

2195
    for (i = 0;
2196
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i));
2197
         i++) {
2198
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2199
            v4present = true;
2200
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2201
            v6present = true;
2202

2203
        /* Add the IP address/netmask to the bridge */
2204
        if (networkAddAddrToBridge(network, ipdef) < 0)
2205
            goto err2;
2206 2207
    }

2208 2209 2210
    if (networkStartHandleMACTableManagerMode(network, macTapIfName) < 0)
        goto err2;

2211
    /* Bring up the bridge interface */
2212
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2213
        goto err2;
2214

2215
    for (i = 0; i < network->def->nroutes; i++) {
2216 2217 2218
        virSocketAddrPtr gateway = NULL;

        routedef = network->def->routes[i];
2219
        gateway = virNetDevIPRouteGetGateway(routedef);
2220

2221 2222 2223
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2224
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2225 2226 2227 2228 2229 2230 2231
            if (networkAddRouteToBridge(network, routedef) < 0) {
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2232 2233
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2234
        networkEnableIPForwarding(v4present, v6present) < 0) {
2235
        virReportSystemError(errno, "%s",
2236
                             _("failed to enable IP forwarding"));
2237
        goto err3;
2238 2239
    }

2240

2241
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2242
    if ((v4present || v6present) &&
2243
        networkStartDhcpDaemon(driver, network) < 0)
2244
        goto err3;
2245

2246
    /* start radvd if there are any ipv6 addresses */
2247
    if (v6present && networkStartRadvd(driver, network) < 0)
2248 2249
        goto err4;

2250 2251 2252 2253 2254 2255 2256 2257
    /* dnsmasq does not wait for DAD to complete before daemonizing,
     * so we need to wait for it ourselves.
     */
    if (v6present && networkWaitDadFinish(network) < 0)
        goto err4;

    /* DAD has finished, dnsmasq is now bound to the
     * bridge's IPv6 address, so we can set the dummy tun down.
2258 2259 2260 2261 2262 2263 2264
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
            goto err4;
        VIR_FORCE_CLOSE(tapfd);
    }

2265
    if (virNetDevBandwidthSet(network->def->bridge,
2266
                              network->def->bandwidth, true) < 0)
2267 2268
        goto err5;

2269
    VIR_FREE(macTapIfName);
2270 2271 2272

    return 0;

2273
 err5:
2274 2275
    if (network->def->bandwidth)
       virNetDevBandwidthClear(network->def->bridge);
2276

2277 2278 2279 2280
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2281 2282 2283 2284 2285
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2286 2287 2288
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2289
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2290

2291 2292 2293
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2294 2295
    if (network->def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(network->def);
2296 2297

 err1:
2298 2299 2300
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2301
    if (macTapIfName) {
2302
        VIR_FORCE_CLOSE(tapfd);
2303
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2304 2305
        VIR_FREE(macTapIfName);
    }
2306 2307

 err0:
2308 2309
    if (!save_err)
        save_err = virSaveLastError();
2310
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2311

2312 2313 2314 2315
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2316
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2317 2318 2319
    return -1;
}

2320 2321 2322
static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
                              virNetworkObjPtr network)
2323
{
2324 2325
    if (network->def->bandwidth)
        virNetDevBandwidthClear(network->def->bridge);
2326

2327 2328 2329 2330 2331
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2332
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2333
            virPidFileDelete(driver->pidDir, radvdpidbase);
2334 2335 2336 2337
            VIR_FREE(radvdpidbase);
        }
    }

2338 2339 2340
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2341
    if (network->def->mac_specified) {
2342
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2343
        if (macTapIfName) {
2344
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2345 2346 2347 2348
            VIR_FREE(macTapIfName);
        }
    }

2349
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2350

2351 2352
    if (network->def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(network->def);
2353

2354
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2355

2356
    /* See if its still alive and really really kill it */
2357
    if (network->dnsmasqPid > 0 &&
2358
        (kill(network->dnsmasqPid, 0) == 0))
2359 2360
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2361 2362 2363 2364 2365 2366

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

2367 2368 2369
    return 0;
}

2370

2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391
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;
}


2392 2393 2394 2395 2396 2397 2398 2399 2400
/* 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;
2401
    unsigned int maxVirtFns = 0;
2402 2403 2404 2405 2406 2407
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2408 2409 2410
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2411 2412
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev, &vfNames,
                                      &virtFns, &numVirtFns, &maxVirtFns)) < 0) {
2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458
        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:
2459
        case VIR_NETWORK_FORWARD_OPEN:
2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499
        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;
}


2500
static int
2501
networkStartNetworkExternal(virNetworkObjPtr network)
2502 2503
{
    /* put anything here that needs to be done each time a network of
2504
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2505 2506 2507
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2508
    return networkCreateInterfacePool(network->def);
2509 2510
}

2511
static int networkShutdownNetworkExternal(virNetworkObjPtr network ATTRIBUTE_UNUSED)
2512 2513
{
    /* put anything here that needs to be done each time a network of
2514
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2515 2516 2517 2518 2519 2520 2521
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2522 2523
networkStartNetwork(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
2524
{
2525 2526 2527
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2528 2529

    if (virNetworkObjIsActive(network)) {
2530 2531
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2532
        return ret;
2533 2534
    }

2535 2536 2537
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2538
    if (virNetworkObjSetDefTransient(network, true) < 0)
2539
        goto cleanup;
2540

2541 2542
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2543
    if (networkRunHook(network, NULL, NULL,
2544 2545 2546 2547
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2548
    switch (network->def->forward.type) {
2549 2550 2551 2552

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2553
    case VIR_NETWORK_FORWARD_OPEN:
2554
        if (networkStartNetworkVirtual(driver, network) < 0)
2555
            goto cleanup;
2556 2557 2558
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2559 2560 2561 2562 2563 2564 2565 2566 2567
        if (network->def->bridge) {
            if (networkStartNetworkBridge(network) < 0)
                goto cleanup;
            break;
        }
        /* intentionally fall through to the macvtap/direct case for
         * VIR_NETWORK_FORWARD_BRIDGE with no bridge device defined
         * (since that is macvtap bridge mode).
         */
2568 2569 2570
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2571
    case VIR_NETWORK_FORWARD_HOSTDEV:
2572
        if (networkStartNetworkExternal(network) < 0)
2573
            goto cleanup;
2574 2575 2576
        break;
    }

2577
    /* finally we can call the 'started' hook script if any */
2578
    if (networkRunHook(network, NULL, NULL,
2579 2580 2581 2582
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2583 2584 2585
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2586
    VIR_DEBUG("Writing network status to disk");
2587
    if (virNetworkSaveStatus(driver->stateDir, network) < 0)
2588
        goto cleanup;
2589 2590

    network->active = 1;
2591 2592
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2593

2594
 cleanup:
2595
    if (ret < 0) {
2596
        virNetworkObjUnsetDefTransient(network);
2597 2598
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2599
        networkShutdownNetwork(driver, network);
2600 2601 2602 2603 2604 2605 2606
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2607 2608 2609
static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
2610 2611 2612 2613 2614 2615 2616 2617 2618
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

2619
    stateFile = virNetworkConfigFile(driver->stateDir,
2620
                                     network->def->name);
2621 2622 2623 2624 2625 2626
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2627
    switch (network->def->forward.type) {
2628 2629 2630 2631

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2632
    case VIR_NETWORK_FORWARD_OPEN:
2633
        ret = networkShutdownNetworkVirtual(driver, network);
2634 2635 2636
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2637 2638 2639 2640 2641 2642 2643 2644
        if (network->def->bridge) {
            ret = networkShutdownNetworkBridge(network);
            break;
        }
        /* intentionally fall through to the macvtap/direct case for
         * VIR_NETWORK_FORWARD_BRIDGE with no bridge device defined
         * (since that is macvtap bridge mode).
         */
2645 2646 2647
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2648
    case VIR_NETWORK_FORWARD_HOSTDEV:
2649
        ret = networkShutdownNetworkExternal(network);
2650 2651 2652
        break;
    }

2653
    /* now that we know it's stopped call the hook if present */
2654
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2655 2656
                   VIR_HOOK_SUBOP_END);

2657
    network->active = 0;
2658
    virNetworkObjUnsetDefTransient(network);
2659
    return ret;
2660 2661 2662
}


2663
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2664 2665
                                         const unsigned char *uuid)
{
2666
    virNetworkDriverStatePtr driver = networkGetDriver();
2667 2668
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2669

2670
    network = virNetworkObjFindByUUID(driver->networks, uuid);
2671
    if (!network) {
2672 2673
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2674
        virReportError(VIR_ERR_NO_NETWORK,
2675 2676
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2677
        goto cleanup;
2678 2679
    }

2680 2681 2682
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2683 2684
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2685
 cleanup:
2686
    virNetworkObjEndAPI(&network);
2687
    return ret;
2688 2689
}

2690
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2691 2692
                                         const char *name)
{
2693
    virNetworkDriverStatePtr driver = networkGetDriver();
2694 2695 2696
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2697
    network = virNetworkObjFindByName(driver->networks, name);
2698
    if (!network) {
2699 2700
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2701
        goto cleanup;
2702 2703
    }

2704 2705 2706
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2707 2708
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2709
 cleanup:
2710
    virNetworkObjEndAPI(&network);
2711
    return ret;
2712 2713
}

2714 2715
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2716
    virNetworkDriverStatePtr driver = networkGetDriver();
2717
    int nactive;
2718

2719 2720 2721
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2722 2723 2724 2725
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
2726

2727 2728 2729
    return nactive;
}

2730 2731 2732 2733 2734
static int networkConnectListNetworks(virConnectPtr conn,
                                      char **const names,
                                      int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2735
    int got = 0;
2736

2737 2738 2739
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2740 2741 2742 2743
    got = virNetworkObjListGetNames(driver->networks,
                                    true, names, nnames,
                                    virConnectListNetworksCheckACL,
                                    conn);
2744

2745 2746 2747
    return got;
}

2748 2749
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2750
    virNetworkDriverStatePtr driver = networkGetDriver();
2751
    int ninactive = 0;
2752

2753 2754 2755
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2756 2757 2758 2759
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
2760

2761 2762 2763
    return ninactive;
}

2764 2765 2766
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2767
    int got = 0;
2768

2769 2770 2771
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2772 2773 2774 2775
    got = virNetworkObjListGetNames(driver->networks,
                                    false, names, nnames,
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
2776 2777 2778
    return got;
}

2779
static int
2780 2781 2782
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2783
{
2784
    virNetworkDriverStatePtr driver = networkGetDriver();
2785 2786 2787 2788
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2789 2790 2791
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2792
    ret = virNetworkObjListExport(conn, driver->networks, nets,
2793 2794
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2795

2796
 cleanup:
2797 2798
    return ret;
}
2799

2800 2801 2802 2803 2804 2805 2806 2807
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
2808
    virNetworkDriverStatePtr driver = networkGetDriver();
2809 2810 2811 2812 2813 2814
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
2815
                                       net, eventID, callback,
2816 2817 2818
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2819
 cleanup:
2820 2821 2822 2823 2824 2825 2826
    return ret;
}

static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
2827
    virNetworkDriverStatePtr driver = networkGetDriver();
2828 2829 2830 2831 2832
    int ret = -1;

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

2833 2834 2835 2836 2837 2838
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
2839

2840
 cleanup:
2841 2842 2843
    return ret;
}

2844 2845 2846 2847 2848
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2849 2850
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2851 2852 2853 2854

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

2855 2856
    ret = virNetworkObjIsActive(obj);

2857
 cleanup:
2858
    virNetworkObjEndAPI(&obj);
2859 2860 2861 2862 2863 2864 2865 2866
    return ret;
}

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

2867 2868
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2869 2870 2871 2872

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

2873 2874
    ret = obj->persistent;

2875
 cleanup:
2876
    virNetworkObjEndAPI(&obj);
2877 2878 2879 2880
    return ret;
}


2881 2882
/*
 * networkFindUnusedBridgeName() - try to find a bridge name that is
2883 2884 2885
 * unused by the currently configured libvirt networks, as well as by
 * the host system itself (possibly created by someone/something other
 * than libvirt). Set this network's name to that new name.
2886 2887 2888 2889 2890 2891 2892 2893
 */
static int
networkFindUnusedBridgeName(virNetworkObjListPtr nets,
                            virNetworkDefPtr def)
{

    int ret = -1, id = 0;
    char *newname = NULL;
2894 2895 2896 2897 2898
    const char *templ = "virbr%d";
    const char *p;

    if (def->bridge &&
        (p = strchr(def->bridge, '%')) == strrchr(def->bridge, '%') &&
2899
        p && p[1] == 'd')
2900
        templ = def->bridge;
2901 2902 2903 2904

    do {
        if (virAsprintf(&newname, templ, id) < 0)
            goto cleanup;
2905 2906 2907 2908 2909 2910 2911
        /* check if this name is used in another libvirt network or
         * there is an existing device with that name. ignore errors
         * from virNetDevExists(), just in case it isn't implemented
         * on this platform (probably impossible).
         */
        if (!(virNetworkBridgeInUse(nets, newname, def->name) ||
              virNetDevExists(newname) == 1)) {
2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963
            VIR_FREE(def->bridge); /*could contain template */
            def->bridge = newname;
            ret = 0;
            goto cleanup;
        }
        VIR_FREE(newname);
    } while (++id <= MAX_BRIDGE_ID);

    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("Bridge generation exceeded max id %d"),
                   MAX_BRIDGE_ID);
    ret = 0;
 cleanup:
    if (ret < 0)
        VIR_FREE(newname);
    return ret;
}



/*
 * networkValidateBridgeName() - if no bridge name is set, or if the
 * bridge name contains a %d (indicating that this is a template for
 * the actual name) try to set an appropriate bridge name.  If a
 * bridge name *is* set, make sure it doesn't conflict with any other
 * network's bridge name.
 */
static int
networkBridgeNameValidate(virNetworkObjListPtr nets,
                          virNetworkDefPtr def)
{
    int ret = -1;

    if (def->bridge && !strstr(def->bridge, "%d")) {
        if (virNetworkBridgeInUse(nets, def->bridge, def->name)) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge name '%s' already in use."),
                           def->bridge);
            goto cleanup;
        }
    } else {
        /* Allocate a bridge name */
        if (networkFindUnusedBridgeName(nets, def) < 0)
            goto cleanup;
    }

    ret = 0;
 cleanup:
    return ret;
}


2964
static int
2965
networkValidate(virNetworkDriverStatePtr driver,
2966
                virNetworkDefPtr def)
2967
{
2968
    size_t i, j;
2969 2970
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2971
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
2972
    bool ipv4def = false, ipv6def = false;
2973
    bool bandwidthAllowed = true;
2974
    bool usesInterface = false, usesAddress = false;
2975

2976 2977 2978
    if (virXMLCheckIllegalChars("name", def->name, "\n") < 0)
        return -1;

2979 2980 2981
    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
2982 2983
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
2984 2985
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE ||
        def->forward.type == VIR_NETWORK_FORWARD_OPEN) {
2986

2987 2988 2989 2990
        /* if no bridge name was given in the config, find a name
         * unused by any other libvirt networks and assign it.
         */
        if (networkBridgeNameValidate(driver->networks, def) < 0)
2991 2992 2993
            return -1;

        virNetworkSetBridgeMacAddr(def);
2994 2995
    } else {
        /* They are also the only types that currently support setting
2996 2997
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
2998
         */
2999 3000 3001 3002 3003 3004 3005 3006
        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;
        }
3007
        if (virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0)) {
3008 3009 3010 3011
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3012
                           virNetworkForwardTypeToString(def->forward.type));
3013 3014
            return -1;
        }
3015
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
3016 3017 3018 3019
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3020
                           virNetworkForwardTypeToString(def->forward.type));
3021 3022 3023 3024 3025 3026 3027
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3028
                           virNetworkForwardTypeToString(def->forward.type));
3029 3030
            return -1;
        }
3031 3032 3033 3034 3035 3036 3037 3038
        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;
        }
3039
        bandwidthAllowed = false;
3040 3041
    }

3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075
    /* 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
3076 3077 3078
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
3079
    for (i = 0;
3080
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
3081
         i++) {
G
Gene Czarcinski 已提交
3082 3083 3084 3085 3086
        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 -- "
3087 3088
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105
                    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;
                }
3106 3107 3108
            }
        }
    }
3109 3110 3111 3112 3113 3114

    /* 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.
     */

3115 3116 3117
    vlanAllowed = (def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV ||
                   def->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH ||
                   (def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
J
Ján Tomko 已提交
3118 3119
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
3120
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH));
3121 3122

    vlanUsed = def->vlan.nTags > 0;
3123 3124
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3125 3126 3127 3128 3129
            /* 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.
             */
3130
            if (def->portGroups[i].virtPortProfile) {
3131
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3132
                    def->portGroups[i].virtPortProfile->virtPortType
3133 3134 3135 3136 3137 3138 3139
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3140
        }
3141
        if (def->portGroups[i].isDefault) {
3142 3143 3144 3145 3146
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3147
                               def->name, defaultPortGroup->name,
3148
                               def->portGroups[i].name);
3149
                return -1;
3150
            }
3151
            defaultPortGroup = &def->portGroups[i];
3152
        }
3153 3154 3155 3156 3157 3158 3159 3160 3161
        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;
            }
        }
3162 3163 3164 3165 3166 3167 3168 3169
        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;
        }
3170
    }
3171 3172 3173 3174 3175 3176 3177
    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.
         */
3178 3179 3180 3181 3182 3183
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197

    if (def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
        for (i = 0; i < def->nPortGroups; i++) {
            if (def->portGroups[i].bandwidth) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("unsupported <bandwidth> element "
                                 "in <portgroup name='%s'> of "
                                 "network '%s' with forward mode='%s'"),
                               def->portGroups[i].name, def->name,
                               virNetworkForwardTypeToString(def->forward.type));
                return -1;
            }
        }
    }
3198 3199 3200
    return 0;
}

3201 3202
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
3203
    virNetworkDriverStatePtr driver = networkGetDriver();
3204
    virNetworkDefPtr def;
3205
    virNetworkObjPtr network = NULL;
3206
    virNetworkPtr ret = NULL;
3207
    virObjectEventPtr event = NULL;
3208

3209
    if (!(def = virNetworkDefParseString(xml)))
3210
        goto cleanup;
3211

3212 3213 3214
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3215
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3216
        goto cleanup;
3217

3218 3219 3220
    /* 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.
3221
     */
3222 3223 3224
    if (!(network = virNetworkAssignDef(driver->networks, def,
                                        VIR_NETWORK_OBJ_LIST_ADD_LIVE |
                                        VIR_NETWORK_OBJ_LIST_ADD_CHECK_LIVE)))
3225 3226
        goto cleanup;
    def = NULL;
3227

3228
    if (networkStartNetwork(driver, network) < 0) {
3229
        virNetworkRemoveInactive(driver->networks,
3230
                                 network);
3231
        goto cleanup;
3232 3233
    }

3234 3235
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3236 3237
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3238

3239
    VIR_INFO("Creating network '%s'", network->def->name);
3240 3241
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

3242
 cleanup:
3243
    virNetworkDefFree(def);
3244 3245
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3246
    virNetworkObjEndAPI(&network);
3247
    return ret;
3248 3249
}

3250 3251
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
3252
    virNetworkDriverStatePtr driver = networkGetDriver();
3253
    virNetworkDefPtr def = NULL;
3254
    bool freeDef = true;
3255
    virNetworkObjPtr network = NULL;
3256
    virNetworkPtr ret = NULL;
3257
    virObjectEventPtr event = NULL;
3258

3259
    if (!(def = virNetworkDefParseString(xml)))
3260
        goto cleanup;
3261

3262 3263 3264
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3265
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3266
        goto cleanup;
3267

3268
    if (!(network = virNetworkAssignDef(driver->networks, def, 0)))
J
Ján Tomko 已提交
3269
        goto cleanup;
3270

3271
    /* def was assigned to network object */
3272
    freeDef = false;
3273 3274

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3275
        if (!virNetworkObjIsActive(network)) {
3276
            virNetworkRemoveInactive(driver->networks, network);
3277 3278
            goto cleanup;
        }
3279 3280 3281 3282 3283
        /* 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);
3284 3285 3286
        goto cleanup;
    }

3287
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3288 3289
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3290

3291 3292
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
3293

3294
 cleanup:
3295 3296
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3297
    if (freeDef)
J
Ján Tomko 已提交
3298
        virNetworkDefFree(def);
3299
    virNetworkObjEndAPI(&network);
3300
    return ret;
3301 3302
}

3303
static int
3304 3305
networkUndefine(virNetworkPtr net)
{
3306
    virNetworkDriverStatePtr driver = networkGetDriver();
3307
    virNetworkObjPtr network;
3308
    int ret = -1;
3309
    bool active = false;
3310
    virObjectEventPtr event = NULL;
3311

3312
    if (!(network = networkObjFromNetwork(net)))
3313
        goto cleanup;
3314

3315 3316 3317
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3318 3319
    if (virNetworkObjIsActive(network))
        active = true;
3320

3321 3322 3323 3324 3325 3326
    if (!network->persistent) {
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                       _("can't undefine transient network"));
        goto cleanup;
    }

3327
    /* remove autostart link */
3328
    if (virNetworkDeleteConfig(driver->networkConfigDir,
3329 3330
                               driver->networkAutostartDir,
                               network) < 0)
3331
        goto cleanup;
3332

3333 3334
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3335 3336
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3337

3338
    VIR_INFO("Undefining network '%s'", network->def->name);
3339
    if (!active) {
3340
        if (networkRemoveInactive(driver, network) < 0)
3341
            goto cleanup;
3342 3343 3344 3345 3346 3347
    } 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);
3348 3349
    }

3350
    ret = 0;
3351

3352
 cleanup:
3353 3354
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3355
    virNetworkObjEndAPI(&network);
3356
    return ret;
3357 3358
}

3359 3360 3361 3362 3363 3364 3365 3366
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3367
    virNetworkDriverStatePtr driver = networkGetDriver();
3368
    virNetworkObjPtr network = NULL;
3369 3370
    int isActive, ret = -1;
    size_t i;
3371
    virNetworkIPDefPtr ipdef;
3372
    bool oldDhcpActive = false;
3373
    bool needFirewallRefresh = false;
3374

3375 3376 3377 3378 3379

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3380
    if (!(network = networkObjFromNetwork(net)))
3381 3382
        goto cleanup;

3383 3384 3385
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3386
    /* see if we are listening for dhcp pre-modification */
3387
    for (i = 0;
3388
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET, i));
3389
         i++) {
3390 3391 3392 3393 3394 3395
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3396 3397
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3398
     */
3399
    isActive = virNetworkObjIsActive(network);
3400 3401
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3402 3403 3404 3405 3406 3407 3408
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425
    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).
                 */
3426 3427 3428 3429
                if (network->def->forward.type != VIR_NETWORK_FORWARD_OPEN) {
                    networkRemoveFirewallRules(network->def);
                    needFirewallRefresh = true;
                }
3430 3431 3432 3433 3434 3435 3436
                break;
            default:
                break;
            }
        }
    }

3437
    /* update the network config in memory/on disk */
3438 3439
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3440
            ignore_value(networkAddFirewallRules(network->def));
3441 3442 3443
        goto cleanup;
    }

3444
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463
        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 ||
3464 3465 3466 3467 3468 3469 3470 3471
            section == VIR_NETWORK_SECTION_IP_DHCP_RANGE ||
            section == VIR_NETWORK_SECTION_DNS_TXT ||
            section == VIR_NETWORK_SECTION_DNS_SRV) {
            /* these sections all change things on the dnsmasq
             * commandline (i.e. in the .conf file), so we need to
             * kill and restart dnsmasq, because dnsmasq sets its uid
             * to "nobody" after it starts, and is unable to re-read
             * the conf file (owned by root, mode 600)
3472
             */
3473
            if (networkRestartDhcpDaemon(driver, network) < 0)
3474 3475
                goto cleanup;

3476 3477 3478 3479 3480 3481 3482 3483
        } 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;

3484
            for (i = 0;
3485
                 (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET, i));
3486
                 i++) {
3487 3488 3489 3490 3491 3492 3493
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3494 3495
                 networkRestartDhcpDaemon(driver, network) < 0) ||
                networkRefreshDhcpDaemon(driver, network) < 0) {
3496 3497 3498
                goto cleanup;
            }

3499 3500 3501 3502
        } else if (section == VIR_NETWORK_SECTION_DNS_HOST) {
            /* this section only changes data in an external file
             * (not the .conf file) so we can just update the config
             * files and send SIGHUP to dnsmasq.
3503
             */
3504
            if (networkRefreshDhcpDaemon(driver, network) < 0)
3505 3506 3507 3508 3509 3510 3511 3512
                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.
             */
3513
            if (networkRefreshRadvd(driver, network) < 0)
3514 3515 3516 3517
                goto cleanup;
        }

        /* save current network state to disk */
3518
        if ((ret = virNetworkSaveStatus(driver->stateDir,
3519
                                        network)) < 0) {
3520
            goto cleanup;
3521
        }
3522
    }
3523 3524 3525 3526 3527 3528

    /* call the 'updated' network hook script */
    if (networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_UPDATED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

3529
    ret = 0;
3530
 cleanup:
3531
    virNetworkObjEndAPI(&network);
3532 3533 3534
    return ret;
}

3535 3536
static int networkCreate(virNetworkPtr net)
{
3537
    virNetworkDriverStatePtr driver = networkGetDriver();
3538 3539
    virNetworkObjPtr network;
    int ret = -1;
3540
    virObjectEventPtr event = NULL;
3541

3542
    if (!(network = networkObjFromNetwork(net)))
3543
        goto cleanup;
3544

3545 3546 3547
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3548
    if ((ret = networkStartNetwork(driver, network)) < 0)
3549
        goto cleanup;
3550

3551 3552
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3553 3554
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3555

3556
 cleanup:
3557 3558
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3559
    virNetworkObjEndAPI(&network);
3560
    return ret;
3561 3562
}

3563 3564
static int networkDestroy(virNetworkPtr net)
{
3565
    virNetworkDriverStatePtr driver = networkGetDriver();
3566 3567
    virNetworkObjPtr network;
    int ret = -1;
3568
    virObjectEventPtr event = NULL;
3569

3570
    if (!(network = networkObjFromNetwork(net)))
3571
        goto cleanup;
3572

3573 3574 3575
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3576
    if (!virNetworkObjIsActive(network)) {
3577
        virReportError(VIR_ERR_OPERATION_INVALID,
3578 3579
                       _("network '%s' is not active"),
                       network->def->name);
3580 3581 3582
        goto cleanup;
    }

3583
    if ((ret = networkShutdownNetwork(driver, network)) < 0)
3584 3585
        goto cleanup;

3586 3587
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3588 3589
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3590

3591
    if (!network->persistent &&
3592
        networkRemoveInactive(driver, network) < 0) {
3593 3594
        ret = -1;
        goto cleanup;
3595
    }
3596

3597
 cleanup:
3598 3599
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3600
    virNetworkObjEndAPI(&network);
3601 3602 3603
    return ret;
}

3604
static char *networkGetXMLDesc(virNetworkPtr net,
3605
                               unsigned int flags)
3606
{
3607
    virNetworkObjPtr network;
3608
    virNetworkDefPtr def;
3609
    char *ret = NULL;
3610

3611
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3612

3613 3614
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3615

3616 3617 3618
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3619 3620 3621 3622 3623 3624
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3625

3626
 cleanup:
3627
    virNetworkObjEndAPI(&network);
3628
    return ret;
3629 3630 3631
}

static char *networkGetBridgeName(virNetworkPtr net) {
3632 3633 3634
    virNetworkObjPtr network;
    char *bridge = NULL;

3635 3636
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3637

3638 3639 3640
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3641
    if (!(network->def->bridge)) {
3642 3643 3644
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3645 3646 3647
        goto cleanup;
    }

3648
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3649

3650
 cleanup:
3651
    virNetworkObjEndAPI(&network);
3652 3653 3654 3655
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
J
Ján Tomko 已提交
3656
                               int *autostart)
3657
{
3658 3659
    virNetworkObjPtr network;
    int ret = -1;
3660

3661 3662
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3663

3664 3665 3666
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3667
    *autostart = network->autostart;
3668
    ret = 0;
3669

3670
 cleanup:
3671
    virNetworkObjEndAPI(&network);
3672
    return ret;
3673 3674 3675
}

static int networkSetAutostart(virNetworkPtr net,
3676 3677
                               int autostart)
{
3678
    virNetworkDriverStatePtr driver = networkGetDriver();
3679
    virNetworkObjPtr network;
3680
    char *configFile = NULL, *autostartLink = NULL;
3681
    int ret = -1;
3682

3683

3684
    if (!(network = networkObjFromNetwork(net)))
3685
        goto cleanup;
3686

3687 3688 3689
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3690
    if (!network->persistent) {
3691 3692
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3693 3694 3695
        goto cleanup;
    }

3696 3697
    autostart = (autostart != 0);

3698
    if (network->autostart != autostart) {
3699
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3700
            goto cleanup;
3701
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3702 3703
            goto cleanup;

3704
        if (autostart) {
3705
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3706
                virReportSystemError(errno,
3707 3708
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3709 3710
                goto cleanup;
            }
3711

3712
            if (symlink(configFile, autostartLink) < 0) {
3713
                virReportSystemError(errno,
3714
                                     _("Failed to create symlink '%s' to '%s'"),
3715
                                     autostartLink, configFile);
3716 3717 3718
                goto cleanup;
            }
        } else {
3719
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3720
                virReportSystemError(errno,
3721
                                     _("Failed to delete symlink '%s'"),
3722
                                     autostartLink);
3723 3724
                goto cleanup;
            }
3725 3726
        }

3727
        network->autostart = autostart;
3728
    }
3729
    ret = 0;
3730

3731
 cleanup:
3732 3733
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3734
    virNetworkObjEndAPI(&network);
3735
    return ret;
3736 3737
}

3738
static int
3739 3740 3741 3742
networkGetDHCPLeases(virNetworkPtr network,
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
3743
{
3744
    virNetworkDriverStatePtr driver = networkGetDriver();
3745 3746 3747
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
3748
    ssize_t size = 0;
3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759
    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;
3760
    virNetworkIPDefPtr ipdef_tmp = NULL;
3761 3762
    virNetworkDHCPLeasePtr lease = NULL;
    virNetworkDHCPLeasePtr *leases_ret = NULL;
3763
    virNetworkObjPtr obj;
3764
    virMacAddr mac_addr;
3765 3766 3767

    virCheckFlags(0, -1);

3768 3769 3770 3771 3772 3773
    /* only to check if the MAC is valid */
    if (mac && virMacAddrParse(mac, &mac_addr) < 0) {
        virReportError(VIR_ERR_INVALID_MAC, "%s", mac);
        return -1;
    }

3774 3775 3776 3777 3778
    if (!(obj = networkObjFromNetwork(network)))
        return -1;

    if (virNetworkGetDHCPLeasesEnsureACL(network->conn, obj->def) < 0)
        goto cleanup;
3779 3780

    /* Retrieve custom leases file location */
3781
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(driver, obj->def->bridge);
3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824

    /* 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;
        }

3825
        if (mac && virMacAddrCompare(mac, mac_tmp))
3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866
            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)) {
3867
                    lease->prefix = virSocketAddrGetIPPrefix(&ipdef_tmp->address,
3868 3869 3870 3871 3872 3873 3874 3875
                                                             &ipdef_tmp->netmask,
                                                             ipdef_tmp->prefix);
                    break;
                }
            }

            if ((VIR_STRDUP(lease->mac, mac_tmp) < 0) ||
                (VIR_STRDUP(lease->ipaddr, ip_tmp) < 0) ||
3876
                (VIR_STRDUP(lease->iface, obj->def->bridge) < 0))
3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908
                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);
3909
    VIR_FREE(lease_entries);
3910 3911
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
3912

3913
    virNetworkObjEndAPI(&obj);
3914

3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925
    return rv;

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

3926 3927

static virNetworkDriver networkDriver = {
3928
    .name = "bridge",
3929 3930 3931 3932 3933
    .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 */
3934 3935
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
3936 3937
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3938 3939
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3940
    .networkUndefine = networkUndefine, /* 0.2.0 */
3941
    .networkUpdate = networkUpdate, /* 0.10.2 */
3942
    .networkCreate = networkCreate, /* 0.2.0 */
3943 3944 3945 3946 3947 3948 3949
    .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 */
3950
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
3951 3952 3953
};

static virStateDriver networkStateDriver = {
3954
    .name = "bridge",
3955
    .stateInitialize  = networkStateInitialize,
3956
    .stateAutoStart  = networkStateAutoStart,
3957 3958
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3959 3960
};

3961 3962
int networkRegister(void)
{
3963
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
3964
        return -1;
3965 3966
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
3967 3968
    return 0;
}
3969 3970 3971

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

3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006
/* A unified function to log network connections and disconnections */

static void
networkLogAllocation(virNetworkDefPtr netdef,
                     virDomainNetType actualType,
                     virNetworkForwardIfDefPtr dev,
                     virDomainNetDefPtr iface,
                     bool inUse)
{
    char macStr[VIR_MAC_STRING_BUFLEN];
    const char *verb = inUse ? "using" : "releasing";

    if (!dev) {
        VIR_INFO("MAC %s %s network %s (%d connections)",
                 virMacAddrFormat(&iface->mac, macStr), verb,
                 netdef->name, netdef->connections);
    } else {
        if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) {
            VIR_INFO("MAC %s %s network %s (%d connections) "
                     "physical device %04x:%02x:%02x.%x (%d connections)",
                     virMacAddrFormat(&iface->mac, macStr), verb,
                     netdef->name, netdef->connections,
                     dev->device.pci.domain, dev->device.pci.bus,
                     dev->device.pci.slot, dev->device.pci.function,
                     dev->connections);
        } else {
            VIR_INFO("MAC %s %s network %s (%d connections) "
                     "physical device %s (%d connections)",
                     virMacAddrFormat(&iface->mac, macStr), verb,
                     netdef->name, netdef->connections,
                     dev->device.dev, dev->connections);
        }
    }
}

4007 4008 4009 4010 4011 4012 4013 4014 4015
/* 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:
4016
 * @dom: domain definition that @iface belongs to
4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027
 * @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
4028 4029
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
4030
{
4031
    virNetworkDriverStatePtr driver = networkGetDriver();
4032
    virDomainNetType actualType = iface->type;
4033 4034
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
4035
    virNetDevBandwidthPtr bandwidth = NULL;
4036 4037 4038
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
4039
    virNetworkForwardIfDefPtr dev = NULL;
4040
    size_t i;
4041 4042 4043
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
4044
        goto validate;
4045 4046 4047 4048

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

4049
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4050
    if (!network) {
4051 4052 4053
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4054
        goto error;
4055 4056
    }
    netdef = network->def;
4057

4058 4059 4060 4061 4062 4063 4064
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

4065 4066 4067
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

4068 4069 4070
    /* 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
4071
     */
4072 4073 4074 4075 4076 4077
    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.
     */
4078 4079 4080 4081 4082 4083

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

4084 4085
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
4086
        goto error;
4087

4088 4089 4090 4091 4092 4093 4094 4095
    /* 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;

4096 4097
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
4098

4099 4100 4101 4102 4103 4104 4105 4106 4107 4108
    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;

4109 4110
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
4111 4112
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_OPEN)) {
4113
        /* for these forward types, the actual net type really *is*
4114
         * NETWORK; we just keep the info from the portgroup in
4115
         * iface->data.network.actual
J
Ján Tomko 已提交
4116
         */
4117
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
4118

4119
        /* we also store the bridge device and macTableManager settings
4120 4121 4122 4123 4124 4125 4126
         * 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;
4127 4128
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4129

4130 4131 4132
        if (networkPlugBandwidth(network, iface) < 0)
            goto error;

4133
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
4134
               netdef->bridge) {
4135 4136 4137 4138 4139

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

4140
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
4141 4142
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
4143
            goto error;
4144 4145
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4146

4147 4148 4149 4150 4151 4152 4153 4154
        /* 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) {
4155
            goto error;
4156 4157 4158 4159 4160 4161 4162 4163 4164 4165
        }
        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);
4166
                goto error;
4167 4168 4169
            }
        }

4170
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4171

4172
        virDomainHostdevSubsysPCIBackendType backend;
4173

4174
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
4175
        if (networkCreateInterfacePool(netdef) < 0)
4176 4177 4178
            goto error;

        /* pick first dev with 0 connections */
4179 4180 4181
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195
                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;
4196
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
4197
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
4198
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
4199

E
Eric Blake 已提交
4200
        switch (netdef->forward.driverName) {
4201
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
4202
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
4203 4204
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
4205
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
4206 4207
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
4208
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219
            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;

4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244
        /* 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;
            }
        }

4245 4246 4247 4248
    } 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)) {
4249 4250 4251 4252 4253 4254

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
4255
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4256
        switch (netdef->forward.type) {
4257
        case VIR_NETWORK_FORWARD_BRIDGE:
4258
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
4259 4260
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
4261
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
4262 4263
            break;
        case VIR_NETWORK_FORWARD_VEPA:
4264
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
4265 4266
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
4267
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
4268 4269 4270
            break;
        }

4271 4272 4273 4274 4275 4276 4277 4278
        /* 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) {
4279
            goto error;
4280
        }
4281
        virtport = iface->data.network.actual->virtPortProfile;
4282
        if (virtport) {
4283 4284 4285 4286 4287 4288 4289 4290
            /* 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);
4291
                goto error;
4292 4293
            }
        }
4294

4295 4296 4297
        /* 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).
         */
4298
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4299 4300 4301 4302
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4303
            goto error;
4304 4305 4306
        } else {
            /* pick an interface from the pool */

4307
            if (networkCreateInterfacePool(netdef) < 0)
4308 4309
                goto error;

4310 4311 4312 4313 4314
            /* 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.
4315
             */
4316 4317
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4318 4319 4320
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4321

4322
                /* pick first dev with 0 connections */
4323 4324 4325
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4326 4327 4328 4329 4330
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4331
                dev = &netdef->forward.ifs[0];
4332 4333 4334
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4335 4336 4337 4338
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4339 4340 4341 4342
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4343
                goto error;
4344
            }
4345 4346
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4347
                goto error;
4348 4349 4350
        }
    }

4351
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4352
        goto error;
4353

4354
 validate:
4355 4356 4357 4358 4359
    /* 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.
     */

4360
    if (virDomainNetGetActualVlan(iface)) {
4361 4362 4363 4364
        /* vlan configuration via libvirt is only supported for PCI
         * Passthrough SR-IOV devices (hostdev or macvtap passthru
         * mode) and openvswitch bridges. Otherwise log an error and
         * fail
4365 4366
         */
        if (!(actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV ||
4367 4368 4369
              (actualType == VIR_DOMAIN_NET_TYPE_DIRECT &&
               virDomainNetGetActualDirectMode(iface)
               == VIR_NETDEV_MACVLAN_MODE_PASSTHRU) ||
4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388
              (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;
        }
    }
4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399
    if (virDomainNetGetActualBandwidth(iface)) {
        /* bandwidth configuration via libvirt is not supported for
         * hostdev network devices
         */
        if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("bandwidth settings are not supported "
                             "for hostdev interfaces"));
            goto error;
        }
    }
4400 4401 4402

    if (netdef) {
        netdef->connections++;
4403
        if (dev)
4404 4405 4406 4407 4408 4409
            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 */
4410
            netdef->connections--;
4411 4412 4413 4414
            if (dev)
                dev->connections--;
            goto error;
        }
4415
        networkLogAllocation(netdef, actualType, dev, iface, true);
4416 4417
    }

4418
    ret = 0;
4419

4420
 cleanup:
4421
    virNetworkObjEndAPI(&network);
4422 4423
    return ret;

4424
 error:
4425
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4426 4427 4428
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4429
    goto cleanup;
4430 4431 4432
}

/* networkNotifyActualDevice:
4433
 * @dom: domain definition that @iface belongs to
4434 4435 4436 4437 4438 4439 4440 4441 4442 4443
 * @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
4444 4445
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4446
{
4447
    virNetworkDriverStatePtr driver = networkGetDriver();
4448
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4449 4450
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4451
    virNetworkForwardIfDefPtr dev = NULL;
4452 4453
    size_t i;
    int ret = -1;
4454 4455

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4456
        return 0;
4457

4458
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4459
    if (!network) {
4460 4461 4462
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4463 4464 4465 4466
        goto error;
    }
    netdef = network->def;

4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477
    /* 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;

4478
    if (!iface->data.network.actual ||
4479 4480
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4481 4482
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4483 4484
    }

4485
    if (networkCreateInterfacePool(netdef) < 0)
4486
        goto error;
4487

4488
    if (netdef->forward.nifs == 0) {
4489
        virReportError(VIR_ERR_INTERNAL_ERROR,
4490 4491
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4492
                       netdef->name);
4493
        goto error;
4494
    }
4495

4496 4497
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4498

4499 4500 4501 4502 4503 4504 4505 4506 4507
        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 */
4508 4509
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4510
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4511 4512
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4513 4514 4515 4516 4517
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4518
            virReportError(VIR_ERR_INTERNAL_ERROR,
4519 4520
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4521
                           netdef->name, actualDev);
4522
            goto error;
4523 4524
        }

4525
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4526 4527
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4528
         */
4529
        if ((dev->connections > 0) &&
4530 4531
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4532 4533
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4534
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4535
            virReportError(VIR_ERR_INTERNAL_ERROR,
4536 4537
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4538
                           netdef->name, actualDev);
4539
            goto error;
4540
        }
4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552
    }  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 */
4553 4554
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4555
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4556
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4557 4558
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4559 4560 4561 4562 4563 4564 4565 4566 4567
                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,
4568 4569 4570 4571
                           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
4572
            goto error;
4573 4574 4575 4576 4577 4578 4579
        }

        /* 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) &&
4580
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4581 4582 4583 4584 4585 4586 4587 4588 4589
            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;
        }
4590 4591
    }

4592
 success:
4593
    netdef->connections++;
4594 4595
    if (dev)
        dev->connections++;
4596 4597 4598 4599 4600 4601 4602 4603 4604
    /* 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;
    }
4605
    networkLogAllocation(netdef, actualType, dev, iface, true);
4606

4607
    ret = 0;
4608
 cleanup:
4609
    virNetworkObjEndAPI(&network);
4610
    return ret;
4611

4612
 error:
4613
    goto cleanup;
4614 4615 4616
}


4617

4618
/* networkReleaseActualDevice:
4619
 * @dom: domain definition that @iface belongs to
4620 4621 4622 4623 4624 4625 4626 4627 4628 4629
 * @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
4630 4631
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4632
{
4633
    virNetworkDriverStatePtr driver = networkGetDriver();
4634
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4635
    virNetworkObjPtr network;
4636
    virNetworkDefPtr netdef;
4637
    virNetworkForwardIfDefPtr dev = NULL;
4638 4639
    size_t i;
    int ret = -1;
4640 4641

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4642
        return 0;
4643

4644
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4645
    if (!network) {
4646 4647 4648
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4649 4650 4651 4652
        goto error;
    }
    netdef = network->def;

4653 4654
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4655
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
4656 4657
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE ||
         netdef->forward.type == VIR_NETWORK_FORWARD_OPEN) &&
4658 4659 4660
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4661 4662 4663
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4664 4665
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4666 4667
    }

4668
    if (netdef->forward.nifs == 0) {
4669
        virReportError(VIR_ERR_INTERNAL_ERROR,
4670
                       _("network '%s' uses a direct/hostdev mode, but "
4671 4672
                         "has no forward dev and no interface pool"),
                       netdef->name);
4673
        goto error;
4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685
    }

    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;
        }
4686

4687 4688
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4689
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4690 4691
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4692 4693 4694
                break;
            }
        }
4695

4696
        if (!dev) {
4697
            virReportError(VIR_ERR_INTERNAL_ERROR,
4698 4699
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4700
                           netdef->name, actualDev);
4701
            goto error;
4702
        }
4703 4704 4705 4706 4707 4708 4709 4710 4711 4712
    } 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;
        }

4713 4714
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4715
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4716
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4717 4718
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4719 4720 4721 4722 4723 4724 4725 4726 4727
                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,
4728 4729 4730 4731
                           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
4732
            goto error;
4733
        }
J
Ján Tomko 已提交
4734
    }
4735

4736
 success:
4737
    if (iface->data.network.actual) {
4738
        netdef->connections--;
4739 4740
        if (dev)
            dev->connections--;
4741 4742 4743
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
4744
        networkLogAllocation(netdef, actualType, dev, iface, false);
4745
    }
4746
    ret = 0;
4747
 cleanup:
4748
    virNetworkObjEndAPI(&network);
4749 4750 4751 4752
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4753
    return ret;
4754

4755
 error:
4756
    goto cleanup;
4757
}
4758 4759 4760 4761 4762 4763

/*
 * networkGetNetworkAddress:
 * @netname: the name of a network
 * @netaddr: string representation of IP address for that network.
 *
4764
 * Attempt to return an IP address associated with the named
4765 4766 4767 4768
 * 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.
 *
4769
 * Note: This function returns the first IP address it finds. It might
4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780
 * 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)
{
4781
    virNetworkDriverStatePtr driver = networkGetDriver();
4782
    int ret = -1;
4783
    virNetworkObjPtr network;
4784
    virNetworkDefPtr netdef;
4785
    virNetworkIPDefPtr ipdef;
4786 4787
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4788
    char *dev_name = NULL;
4789 4790

    *netaddr = NULL;
4791
    network = virNetworkObjFindByName(driver->networks, netname);
4792
    if (!network) {
4793 4794 4795
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4796
        goto cleanup;
4797 4798 4799
    }
    netdef = network->def;

4800
    switch (netdef->forward.type) {
4801 4802 4803
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
4804
    case VIR_NETWORK_FORWARD_OPEN:
4805
        ipdef = virNetworkDefGetIPByIndex(netdef, AF_UNSPEC, 0);
4806
        if (!ipdef) {
4807
            virReportError(VIR_ERR_INTERNAL_ERROR,
4808
                           _("network '%s' doesn't have an IP address"),
4809
                           netdef->name);
4810
            goto cleanup;
4811 4812 4813 4814 4815
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4816
        if ((dev_name = netdef->bridge))
4817 4818
            break;
        /*
4819 4820
         * fall through if netdef->bridge wasn't set, since that is
         * macvtap bridge mode network.
4821 4822 4823 4824
         */
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
4825 4826
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4827

4828
        if (!dev_name) {
4829 4830 4831
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4832
            goto cleanup;
4833 4834 4835 4836
        }
        break;
    }

4837
    if (dev_name) {
4838
        if (virNetDevIPAddrGet(dev_name, &addr) < 0)
4839
            goto cleanup;
4840
        addrptr = &addr;
4841 4842
    }

4843 4844
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
4845
        goto cleanup;
4846 4847
    }

4848
    ret = 0;
4849
 cleanup:
4850
    virNetworkObjEndAPI(&network);
4851 4852
    return ret;
}
4853

4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887
/* networkGetActualType:
 * @dom: domain definition that @iface belongs to
 * @iface: the original NetDef from the domain
 *
 * Looks up the network reference by iface, and returns the actual
 * type of the connection without allocating any resources.
 *
 * Returns 0 on success, -1 on failure.
 */
int
networkGetActualType(virDomainNetDefPtr iface)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
    int ret = -1;

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

    if (iface->data.network.actual)
        return iface->data.network.actual->type;

    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!network) {
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return -1;
    }
    netdef = network->def;

    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
4888 4889
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_OPEN)) {
4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926
        /* for these forward types, the actual net type really *is*
         * NETWORK; we just keep the info from the portgroup in
         * iface->data.network.actual
         */
        ret = VIR_DOMAIN_NET_TYPE_NETWORK;

    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
               netdef->bridge) {

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

        ret = VIR_DOMAIN_NET_TYPE_BRIDGE;

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

        ret = VIR_DOMAIN_NET_TYPE_HOSTDEV;

    } 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)) {

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

        ret = VIR_DOMAIN_NET_TYPE_DIRECT;

    }

    virNetworkObjEndAPI(&network);
    return ret;
}


4927 4928 4929
/**
 * networkCheckBandwidth:
 * @net: network QoS
4930
 * @ifaceBand: interface QoS (may be NULL if no QoS)
4931
 * @oldBandwidth: new interface QoS (may be NULL if no QoS)
4932
 * @ifaceMac: interface MAC (used in error messages for identification)
4933 4934
 * @new_rate: new rate for non guaranteed class
 *
4935 4936 4937 4938 4939 4940 4941 4942
 * Function checks if @ifaceBand can be satisfied on @net. However, sometimes it
 * may happen that the interface that @ifaceBand corresponds to is already
 * plugged into the @net and the bandwidth is to be updated. In that case we
 * need to check if new bandwidth can be satisfied. If that's the case
 * @ifaceBand should point to new bandwidth settings and @oldBandwidth to
 * current ones. If you want to suppress this functionality just pass
 * @oldBandwidth == NULL.
 *
4943 4944 4945 4946 4947 4948
 * 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,
4949
                      virNetDevBandwidthPtr ifaceBand,
4950
                      virNetDevBandwidthPtr oldBandwidth,
4951
                      virMacAddr ifaceMac,
4952 4953 4954 4955 4956 4957 4958 4959
                      unsigned long long *new_rate)
{
    int ret = -1;
    virNetDevBandwidthPtr netBand = net->def->bandwidth;
    unsigned long long tmp_floor_sum = net->floor_sum;
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

4960
    virMacAddrFormat(&ifaceMac, ifmac);
4961 4962 4963 4964 4965 4966 4967 4968 4969 4970

    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;
    }

4971 4972
    if (((!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor) &&
         (!oldBandwidth || !oldBandwidth->in || !oldBandwidth->in->floor)) ||
4973 4974
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4975
        return 1;
4976
    }
4977 4978

    tmp_new_rate = netBand->in->average;
4979 4980 4981 4982
    if (oldBandwidth && oldBandwidth->in)
        tmp_floor_sum -= oldBandwidth->in->floor;
    if (ifaceBand && ifaceBand->in)
        tmp_floor_sum += ifaceBand->in->floor;
4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007

    /* 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;
    }

5008 5009
    if (new_rate)
        *new_rate = tmp_new_rate;
5010 5011
    ret = 0;

5012
 cleanup:
5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028
    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)
{
5029
    ssize_t ret = 0;
5030

5031
    ret = virBitmapNextClearBit(net->class_id, -1);
5032

5033
    if (ret < 0 || virBitmapSetBit(net->class_id, ret) < 0)
5034 5035 5036 5037 5038
        return -1;

    return ret;
}

5039

5040
static int
5041 5042 5043 5044
networkPlugBandwidthImpl(virNetworkObjPtr net,
                         virDomainNetDefPtr iface,
                         virNetDevBandwidthPtr ifaceBand,
                         unsigned long long new_rate)
5045
{
5046
    virNetworkDriverStatePtr driver = networkGetDriver();
5047
    ssize_t class_id = 0;
5048 5049
    int plug_ret;
    int ret = -1;
5050 5051 5052 5053 5054 5055 5056 5057

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

5058 5059
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
5060 5061 5062 5063 5064 5065 5066 5067
    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 */
5068
    net->floor_sum += ifaceBand->in->floor;
5069
    /* update status file */
5070
    if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
5071
        ignore_value(virBitmapClearBit(net->class_id, class_id));
5072
        net->floor_sum -= ifaceBand->in->floor;
5073 5074 5075 5076
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
5077 5078
    /* update rate for non guaranteed NICs */
    new_rate -= net->floor_sum;
5079
    if (virNetDevBandwidthUpdateRate(net->def->bridge, 2,
5080 5081 5082 5083 5084
                                     net->def->bandwidth, new_rate) < 0)
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                 net->def->bridge);

    ret = 0;
5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124
 cleanup:
    return ret;
}


static int
networkPlugBandwidth(virNetworkObjPtr net,
                     virDomainNetDefPtr iface)
{
    int ret = -1;
    int plug_ret;
    unsigned long long new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);

    if ((plug_ret = networkCheckBandwidth(net, ifaceBand, NULL,
                                          iface->mac, &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;
    }

    if (networkPlugBandwidthImpl(net, iface, ifaceBand, new_rate) < 0)
        goto cleanup;

    ret = 0;
5125

5126
 cleanup:
5127 5128 5129 5130 5131 5132 5133
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
5134
    virNetworkDriverStatePtr driver = networkGetDriver();
5135 5136
    int ret = 0;
    unsigned long long new_rate;
5137
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
5138 5139 5140

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
5141 5142 5143 5144 5145
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156
        /* 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 */
5157
        net->floor_sum -= ifaceBand->in->floor;
5158 5159 5160 5161
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
5162
        if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
5163
            net->floor_sum += ifaceBand->in->floor;
5164 5165 5166 5167
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
5168 5169
        /* update rate for non guaranteed NICs */
        new_rate -= net->floor_sum;
5170
        if (virNetDevBandwidthUpdateRate(net->def->bridge, 2,
5171 5172 5173 5174 5175 5176 5177
                                         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;
    }

5178
 cleanup:
5179 5180
    return ret;
}
5181 5182 5183

static void
networkNetworkObjTaint(virNetworkObjPtr net,
5184
                       virNetworkTaintFlags taint)
5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195
{
    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));
    }
}
5196 5197 5198 5199 5200 5201


static bool
networkBandwidthGenericChecks(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
5202
    virNetDevBandwidthPtr ifaceBand;
5203 5204 5205 5206 5207 5208 5209 5210
    unsigned long long old_floor, new_floor;

    if (virDomainNetGetActualType(iface) != VIR_DOMAIN_NET_TYPE_NETWORK) {
        /* This is not an interface that's plugged into a network.
         * We don't care. Thus from our POV bandwidth change is allowed. */
        return false;
    }

5211
    ifaceBand = virDomainNetGetActualBandwidth(iface);
5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251
    old_floor = new_floor = 0;

    if (ifaceBand && ifaceBand->in)
        old_floor = ifaceBand->in->floor;
    if (newBandwidth && newBandwidth->in)
        new_floor = newBandwidth->in->floor;

    return new_floor != old_floor;
}


bool
networkBandwidthChangeAllowed(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
    virNetworkObjPtr network = NULL;
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    bool ret = false;

    if (!networkBandwidthGenericChecks(iface, newBandwidth))
        return true;

    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!network) {
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return false;
    }

    if (networkCheckBandwidth(network, newBandwidth, ifaceBand, iface->mac, NULL) < 0)
        goto cleanup;

    ret = true;

 cleanup:
    virNetworkObjEndAPI(&network);
    return ret;
}
5252 5253 5254 5255 5256 5257 5258 5259 5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289


int
networkBandwidthUpdate(virDomainNetDefPtr iface,
                       virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
    virNetworkObjPtr network = NULL;
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    unsigned long long new_rate = 0;
    int plug_ret;
    int ret = -1;

    if (!networkBandwidthGenericChecks(iface, newBandwidth))
        return 0;

    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!network) {
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return ret;
    }

    if ((plug_ret = networkCheckBandwidth(network, newBandwidth, ifaceBand,
                                          iface->mac, &new_rate)) < 0) {
        /* helper reported error */
        goto cleanup;
    }

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

    /* Okay, there are three possible scenarios: */

5290 5291
    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
        newBandwidth && newBandwidth->in && newBandwidth->in->floor) {
5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333
        /* Either we just need to update @floor .. */

        if (virNetDevBandwidthUpdateRate(network->def->bridge,
                                         iface->data.network.actual->class_id,
                                         network->def->bandwidth,
                                         newBandwidth->in->floor) < 0)
            goto cleanup;

        network->floor_sum -= ifaceBand->in->floor;
        network->floor_sum += newBandwidth->in->floor;
        new_rate -= network->floor_sum;

        if (virNetDevBandwidthUpdateRate(network->def->bridge, 2,
                                         network->def->bandwidth, new_rate) < 0 ||
            virNetworkSaveStatus(driver->stateDir, network) < 0) {
            /* Ouch, rollback */
            network->floor_sum -= newBandwidth->in->floor;
            network->floor_sum += ifaceBand->in->floor;

            ignore_value(virNetDevBandwidthUpdateRate(network->def->bridge,
                                                      iface->data.network.actual->class_id,
                                                      network->def->bandwidth,
                                                      ifaceBand->in->floor));
            goto cleanup;
        }
    } else if (newBandwidth->in && newBandwidth->in->floor) {
        /* .. or we need to plug in new .. */

        if (networkPlugBandwidthImpl(network, iface, newBandwidth, new_rate) < 0)
            goto cleanup;
    } else {
        /* .. or unplug old. */

        if (networkUnplugBandwidth(network, iface) < 0)
            goto cleanup;
    }

    ret = 0;
 cleanup:
    virNetworkObjEndAPI(&network);
    return ret;
}