bridge_driver.c 163.4 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2015 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

#include <sys/types.h>
#include <sys/poll.h>
#include <limits.h>
#include <string.h>
#include <stdio.h>
#include <stdarg.h>
#include <stdlib.h>
#include <unistd.h>
#include <errno.h>
#include <sys/utsname.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <signal.h>
#include <paths.h>
#include <pwd.h>
#include <sys/wait.h>
#include <sys/ioctl.h>
43
#include <net/if.h>
44
#include <dirent.h>
45 46 47
#if HAVE_SYS_SYSCTL_H
# include <sys/sysctl.h>
#endif
48

49
#include "virerror.h"
50
#include "datatypes.h"
51
#include "bridge_driver.h"
52
#include "bridge_driver_platform.h"
53
#include "network_conf.h"
54
#include "device_conf.h"
55
#include "driver.h"
56
#include "virbuffer.h"
57
#include "virpidfile.h"
58
#include "vircommand.h"
59
#include "viralloc.h"
60
#include "viruuid.h"
61
#include "viriptables.h"
62
#include "virlog.h"
63
#include "virdnsmasq.h"
64
#include "configmake.h"
65
#include "virnetdev.h"
66
#include "virpci.h"
67 68
#include "virnetdevbridge.h"
#include "virnetdevtap.h"
69
#include "virnetdevvportprofile.h"
70
#include "virdbus.h"
71
#include "virfile.h"
72
#include "virstring.h"
73
#include "viraccessapicheck.h"
74
#include "network_event.h"
75
#include "virhook.h"
76
#include "virjson.h"
77

78
#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);
203 204
        if (iface && virDomainNetDefFormat(&buf, iface, 0) < 0)
            goto cleanup;
205 206 207 208 209 210 211 212
        if (virNetworkDefFormatBuf(&buf, network->def, 0) < 0)
            goto cleanup;
        if (dom && virDomainDefFormatInternal(dom, 0, &buf) < 0)
            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

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

408 409
    case VIR_NETWORK_FORWARD_BRIDGE:
        if (!(obj->def->bridge && virNetDevExists(obj->def->bridge) == 1)) {
410 411
                obj->active = 0;
            break;
412
        }
413 414 415 416 417 418 419 420 421
        /* 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;
422

423 424 425
    case VIR_NETWORK_FORWARD_HOSTDEV:
        /* so far no extra checks */
        break;
426
    }
427

428 429 430
    /* Try and read dnsmasq/radvd pids of active networks */
    if (obj->active && obj->def->ips && (obj->def->nips > 0)) {
        char *radvdpidbase;
431

432 433 434
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           obj->def->name,
                                           &obj->dnsmasqPid,
435
                                           dnsmasqCapsGetBinaryPath(dnsmasq_caps)));
436 437 438
        radvdpidbase = networkRadvdPidfileBasename(obj->def->name);
        if (!radvdpidbase)
            goto cleanup;
439

440 441 442 443
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           radvdpidbase,
                                           &obj->radvdPid, RADVD));
        VIR_FREE(radvdpidbase);
444
    }
445

446 447
    ret = 0;
 cleanup:
448
    virObjectUnlock(obj);
449
    virObjectUnref(dnsmasq_caps);
450 451
    return ret;
}
452

453

454 455
static int
networkAutostartConfig(virNetworkObjPtr net,
456
                       void *opaque)
457
{
458
    virNetworkDriverStatePtr driver = opaque;
459
    int ret = -1;
460

461
    virObjectLock(net);
462 463
    if (net->autostart &&
        !virNetworkObjIsActive(net) &&
464
        networkStartNetwork(driver, net) < 0)
465 466 467 468
        goto cleanup;

    ret = 0;
 cleanup:
469
    virObjectUnlock(net);
470
    return ret;
471 472
}

473 474 475
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
476
                             DBusMessage *message, void *user_data)
477
{
478 479
    virNetworkDriverStatePtr driver = user_data;

480 481 482 483 484 485
    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.");
486
        networkReloadFirewallRules(driver);
487 488 489 490 491 492
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

493
static int
494
networkMigrateStateFiles(virNetworkDriverStatePtr driver)
495 496 497 498 499 500 501 502 503 504 505 506 507
{
    /* 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;
508
    int direrr;
509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527
    struct dirent *entry;
    char *oldPath = NULL, *newPath = NULL;
    char *contents = NULL;

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

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

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

528
    while ((direrr = virDirRead(dir, &entry, oldStateDir)) > 0) {
529 530 531
        if (entry->d_type != DT_UNKNOWN &&
            entry->d_type != DT_REG)
            continue;
532

533
        if (STREQ(entry->d_name, ".") ||
534 535 536 537 538
            STREQ(entry->d_name, ".."))
            continue;

        if (virAsprintf(&oldPath, "%s/%s",
                        oldStateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
539
            goto cleanup;
540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556

        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;
            }
        }

557
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
J
Ján Tomko 已提交
558
            goto cleanup;
559 560 561

        if (virAsprintf(&newPath, "%s/%s",
                        driver->stateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
562
            goto cleanup;
563 564 565 566 567 568 569 570 571 572 573 574
        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);
    }
575
    if (direrr < 0)
J
Ján Tomko 已提交
576
        goto cleanup;
577 578 579 580 581 582 583 584 585 586

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

587
/**
588
 * networkStateInitialize:
589 590 591 592
 *
 * Initialization function for the QEmu daemon
 */
static int
593 594 595
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
596
{
597 598 599
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
600 601 602
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
603

604
    if (VIR_ALLOC(network_driver) < 0)
605
        goto error;
606

607 608
    if (virMutexInit(&network_driver->lock) < 0) {
        VIR_FREE(network_driver);
609 610
        goto error;
    }
611

612 613 614 615
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
616
    if (privileged) {
617
        if (VIR_STRDUP(network_driver->networkConfigDir,
618
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
619
            VIR_STRDUP(network_driver->networkAutostartDir,
620
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
621
            VIR_STRDUP(network_driver->stateDir,
622
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
623
            VIR_STRDUP(network_driver->pidDir,
624
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
625
            VIR_STRDUP(network_driver->dnsmasqStateDir,
626
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
627
            VIR_STRDUP(network_driver->radvdStateDir,
628 629
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
630 631 632 633 634

        /* migration from old to new location is only applicable for
         * privileged mode - unprivileged mode directories haven't
         * changed location.
         */
635
        if (networkMigrateStateFiles(network_driver) < 0)
636
            goto error;
637
    } else {
638 639 640
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
641
            goto error;
642

643
        if ((virAsprintf(&network_driver->networkConfigDir,
644
                         "%s/qemu/networks", configdir) < 0) ||
645
            (virAsprintf(&network_driver->networkAutostartDir,
646
                         "%s/qemu/networks/autostart", configdir) < 0) ||
647
            (virAsprintf(&network_driver->stateDir,
648
                         "%s/network/lib", rundir) < 0) ||
649
            (virAsprintf(&network_driver->pidDir,
650
                         "%s/network/run", rundir) < 0) ||
651
            (virAsprintf(&network_driver->dnsmasqStateDir,
652
                         "%s/dnsmasq/lib", rundir) < 0) ||
653
            (virAsprintf(&network_driver->radvdStateDir,
654
                         "%s/radvd/lib", rundir) < 0)) {
655
            goto error;
656
        }
657 658
    }

659
    if (virFileMakePath(network_driver->stateDir) < 0) {
660 661
        virReportSystemError(errno,
                             _("cannot create directory %s"),
662
                             network_driver->stateDir);
663 664 665
        goto error;
    }

666
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
667
    network_driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
668

669
    if (!(network_driver->networks = virNetworkObjListNew()))
670 671
        goto error;

672 673
    if (virNetworkLoadAllState(network_driver->networks,
                               network_driver->stateDir) < 0)
674 675
        goto error;

676 677 678
    if (virNetworkLoadAllConfigs(network_driver->networks,
                                 network_driver->networkConfigDir,
                                 network_driver->networkAutostartDir) < 0)
679 680
        goto error;

681 682 683 684
    /* 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). */
685
    virNetworkObjListForEach(network_driver->networks,
686
                             networkUpdateState,
687 688
                             network_driver);
    virNetworkObjListPrune(network_driver->networks,
689 690
                           VIR_CONNECT_LIST_NETWORKS_INACTIVE |
                           VIR_CONNECT_LIST_NETWORKS_TRANSIENT);
691 692
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
693

694
    network_driver->networkEventState = virObjectEventStateNew();
695

696 697 698 699
#ifdef HAVE_FIREWALLD
    if (!(sysbus = virDBusGetSystemBus())) {
        virErrorPtr err = virGetLastError();
        VIR_WARN("DBus not available, disabling firewalld support "
700
                 "in bridge_network_driver: %s", err->message);
701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
    } 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,
718
                                   network_driver, NULL);
719 720 721
    }
#endif

722
    ret = 0;
723
 cleanup:
724 725 726
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
727

728
 error:
729 730
    if (network_driver)
        networkDriverUnlock(network_driver);
731
    networkStateCleanup();
732
    goto cleanup;
733 734
}

735 736 737 738 739 740 741 742
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
743
    if (!network_driver)
744 745
        return;

746
    virNetworkObjListForEach(network_driver->networks,
747
                             networkAutostartConfig,
748
                             network_driver);
749 750
}

751
/**
752
 * networkStateReload:
753 754 755 756 757
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
758 759
networkStateReload(void)
{
760
    if (!network_driver)
761 762
        return 0;

763 764 765 766 767 768 769 770
    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,
771 772
                             networkAutostartConfig,
                             NULL);
773 774 775 776 777
    return 0;
}


/**
778
 * networkStateCleanup:
779 780 781 782
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
783 784
networkStateCleanup(void)
{
785
    if (!network_driver)
786 787
        return -1;

788
    virObjectEventStateFree(network_driver->networkEventState);
789

790
    /* free inactive networks */
791
    virObjectUnref(network_driver->networks);
792

793 794 795 796 797 798
    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);
799

800
    virObjectUnref(network_driver->dnsmasqCaps);
801

802
    virMutexDestroy(&network_driver->lock);
803

804
    VIR_FREE(network_driver);
805 806 807 808 809

    return 0;
}


810 811 812 813 814 815 816
/* 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)
{
817 818
    size_t i;
    int ret = -1;
819 820 821 822 823 824 825
    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.
     */
826
    for (i = 0; i < 25; i++) {
827
        int signum = 0;
828
        if (i == 0) {
829
            signum = SIGTERM;
830
        } else if (i == 15) {
831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865
            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);
866
 cleanup:
867 868 869
    return ret;
}

J
Ján Tomko 已提交
870 871 872
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
G
Gene Czarcinski 已提交
873

874
static int
G
Gene Czarcinski 已提交
875 876
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
                                 virNetworkIpDefPtr ipdef)
877
{
878
    size_t i;
G
Gene Czarcinski 已提交
879
    bool ipv6 = false;
880

G
Gene Czarcinski 已提交
881 882
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
883 884
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
885
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
886 887
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
888
                return -1;
889
    }
890

G
Gene Czarcinski 已提交
891 892 893 894 895 896 897
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
898
    size_t i, j;
G
Gene Czarcinski 已提交
899

900 901
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
902
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
903
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
904
                for (j = 0; j < host->nnames; j++)
905 906
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
907 908
            }
        }
909 910
    }

911
    return 0;
912 913 914
}


915 916
int
networkDnsmasqConfContents(virNetworkObjPtr network,
917 918 919 920
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
921
{
922
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
923
    int r, ret = -1;
924
    int nbleases = 0;
925
    size_t i;
926
    virNetworkDNSDefPtr dns = &network->def->dns;
G
Gene Czarcinski 已提交
927 928
    virNetworkIpDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
    bool ipv6SLAAC;
929
    char *saddr = NULL, *eaddr = NULL;
930

931 932
    *configstr = NULL;

933
    /*
934 935 936
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
937
     *
938 939
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
940
     */
941 942 943 944 945 946

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

947 948
    /* create dnsmasq config file appropriate for this network */
    virBufferAsprintf(&configbuf,
949 950 951 952 953 954 955
                      "##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"
956
                      "strict-order\n",
957 958
                      network->def->name);

959 960 961 962
    if (network->def->dns.forwarders) {
        virBufferAddLit(&configbuf, "no-resolv\n");
        for (i = 0; i < network->def->dns.nfwds; i++) {
            virBufferAsprintf(&configbuf, "server=%s\n",
J
Ján Tomko 已提交
963
                              network->def->dns.forwarders[i]);
964 965 966
        }
    }

967
    if (network->def->domain) {
968 969 970 971 972
        if (network->def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
                              network->def->domain);
        }
973
        virBufferAsprintf(&configbuf,
974 975 976 977
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
978

J
Ján Tomko 已提交
979
    if (network->def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
980 981 982 983
        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)
984
         */
985
        virBufferAddLit(&configbuf, "local=//\n");
986
    }
987

988
    if (pidfile)
989
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
990

991 992 993
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

994 995 996 997 998 999 1000 1001
    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.
         */
1002
        virBufferAsprintf(&configbuf,
1003 1004 1005
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
1006
    } else {
1007
        virBufferAddLit(&configbuf, "bind-interfaces\n");
1008 1009 1010 1011 1012 1013 1014 1015
        /*
         * --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
         */
1016 1017 1018
        for (i = 0;
             (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
             i++) {
1019 1020 1021 1022
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
1023

1024
            /* also part of CVE 2012-3411 - if the host's version of
1025
             * dnsmasq doesn't have bind-dynamic, only allow listening on
1026 1027
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
1028 1029
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
1030 1031 1032 1033
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
1034
                                 "The version of dnsmasq on this host (%d.%d) "
1035 1036 1037 1038
                                 "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 "
1039 1040 1041 1042 1043 1044
                                 "(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);
1045
                VIR_FREE(ipaddr);
1046 1047
                goto cleanup;
            }
1048
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1049 1050 1051
            VIR_FREE(ipaddr);
        }
    }
1052

1053 1054
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1055
     * guaranteed to not work, and set no-resolv so that no dns
1056 1057 1058
     * 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).
1059
     */
1060
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1061
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1062
                        "no-resolv\n");
1063
    }
1064

1065
    for (i = 0; i < dns->ntxts; i++) {
1066
        virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
1067 1068
                          dns->txts[i].name,
                          dns->txts[i].value);
1069
    }
1070

1071
    for (i = 0; i < dns->nsrvs; i++) {
1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093
        /* 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);
1094

1095 1096 1097
        /* domain is optional - it defaults to the domain of this network */
        if (dns->srvs[i].domain)
            virBufferAsprintf(&configbuf, ".%s", dns->srvs[i].domain);
1098

1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120
        /* If target is empty or ".", that means "the service is
         * decidedly not available at this domain" (RFC2782). In that
         * case, any port, priority, or weight is irrelevant.
         */
        if (dns->srvs[i].target && STRNEQ(dns->srvs[i].target, ".")) {

            virBufferAsprintf(&configbuf, ",%s", dns->srvs[i].target);
            /* port, priority, and weight are optional, but are
             * identified by their position in the line. If an item is
             * unspecified, but something later in the line *is*
             * specified, we need to give the default value for the
             * unspecified item. (According to the dnsmasq manpage,
             * the default for port is 1).
             */
            if (dns->srvs[i].port ||
                dns->srvs[i].priority || dns->srvs[i].weight)
                virBufferAsprintf(&configbuf, ",%d",
                                  dns->srvs[i].port ? dns->srvs[i].port : 1);
            if (dns->srvs[i].priority || dns->srvs[i].weight)
                virBufferAsprintf(&configbuf, ",%d", dns->srvs[i].priority);
            if (dns->srvs[i].weight)
                virBufferAsprintf(&configbuf, ",%d", dns->srvs[i].weight);
1121
        }
1122
        virBufferAddLit(&configbuf, "\n");
1123 1124
    }

G
Gene Czarcinski 已提交
1125
    /* Find the first dhcp for both IPv4 and IPv6 */
1126 1127 1128
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
1129 1130 1131 1132
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1133 1134
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145
                    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,
1146 1147 1148 1149 1150 1151 1152 1153 1154
                                   _("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
1155 1156 1157 1158
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1159 1160
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173
                    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 "
1174 1175 1176 1177
                 "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
1178 1179 1180 1181 1182
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1183
        for (r = 0; r < ipdef->nranges; r++) {
1184 1185
            if (!(saddr = virSocketAddrFormat(&ipdef->ranges[r].start)) ||
                !(eaddr = virSocketAddrFormat(&ipdef->ranges[r].end)))
1186
                goto cleanup;
1187

1188
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s\n",
1189
                              saddr, eaddr);
1190
            VIR_FREE(saddr);
1191
            VIR_FREE(eaddr);
1192
            nbleases += virSocketAddrGetRange(&ipdef->ranges[r].start,
1193 1194 1195
                                              &ipdef->ranges[r].end,
                                              &ipdef->address,
                                              virNetworkIpDefPrefix(ipdef));
1196
        }
1197

1198
        /*
1199 1200 1201 1202
         * 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)
1203 1204
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1205
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1206 1207
            if (!bridgeaddr)
                goto cleanup;
1208
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static\n", bridgeaddr);
1209 1210
            VIR_FREE(bridgeaddr);
        }
1211

G
Gene Czarcinski 已提交
1212 1213
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1214

G
Gene Czarcinski 已提交
1215 1216 1217
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts)
1218
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1219

G
Gene Czarcinski 已提交
1220
            if (ipdef->tftproot) {
1221 1222
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1223
            }
1224

G
Gene Czarcinski 已提交
1225 1226 1227
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1228

1229
                    if (!bootserver)
G
Gene Czarcinski 已提交
1230
                        goto cleanup;
1231
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1232
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1233 1234
                    VIR_FREE(bootserver);
                } else {
1235
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1236 1237 1238 1239 1240
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1241

1242
    if (nbleases > 0)
1243
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1244

G
Gene Czarcinski 已提交
1245 1246
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1247
        goto cleanup;
G
Gene Czarcinski 已提交
1248 1249 1250 1251 1252 1253

    /* 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)
1254 1255
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1256

1257 1258
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1259
     */
1260
    virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
1261
                      dctx->addnhostsfile->path);
G
Gene Czarcinski 已提交
1262 1263 1264

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1265
        if (ipv6def) {
1266
            virBufferAddLit(&configbuf, "enable-ra\n");
1267
        } else {
1268 1269 1270
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
                 i++) {
G
Gene Czarcinski 已提交
1271 1272 1273 1274
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1275 1276
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1277 1278
                    VIR_FREE(bridgeaddr);
                }
1279
            }
1280
        }
1281 1282
    }

1283 1284 1285
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1286
    ret = 0;
G
Gene Czarcinski 已提交
1287

1288
 cleanup:
1289 1290
    VIR_FREE(saddr);
    VIR_FREE(eaddr);
1291
    virBufferFreeAndReset(&configbuf);
1292
    return ret;
1293 1294
}

1295
/* build the dnsmasq command line */
1296 1297 1298
static int ATTRIBUTE_NONNULL(3)
networkBuildDhcpDaemonCommandLine(virNetworkDriverStatePtr driver,
                                  virNetworkObjPtr network,
1299
                                  virCommandPtr *cmdout,
1300 1301
                                  char *pidfile,
                                  dnsmasqContext *dctx)
1302
{
1303
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1304
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1305
    int ret = -1;
1306 1307
    char *configfile = NULL;
    char *configstr = NULL;
1308
    char *leaseshelper_path = NULL;
1309 1310

    network->dnsmasqPid = -1;
1311

1312
    if (networkDnsmasqConfContents(network, pidfile, &configstr,
1313
                                   dctx, dnsmasq_caps) < 0)
1314 1315 1316 1317 1318
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1319
    if (!(configfile = networkDnsmasqConfigFileName(driver, network->def->name)))
1320 1321 1322 1323 1324
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1325 1326
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1327 1328 1329
        goto cleanup;
    }

1330 1331
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1332
                                                  abs_topbuilddir "/src",
1333 1334 1335
                                                  LIBEXECDIR)))
        goto cleanup;

1336
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(dnsmasq_caps));
1337
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1338 1339
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1340
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1341
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", network->def->bridge);
1342

1343
    *cmdout = cmd;
1344
    ret = 0;
1345
 cleanup:
1346
    virObjectUnref(dnsmasq_caps);
1347 1348
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1349
    VIR_FREE(leaseshelper_path);
1350 1351 1352 1353
    return ret;
}

static int
1354 1355
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
1356 1357 1358 1359
{
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1360
    dnsmasqContext *dctx = NULL;
1361

1362
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0)) {
G
Gene Czarcinski 已提交
1363
        /* no IP addresses, so we don't need to run */
1364 1365 1366 1367
        ret = 0;
        goto cleanup;
    }

1368
    if (virFileMakePath(driver->pidDir) < 0) {
1369
        virReportSystemError(errno,
1370
                             _("cannot create directory %s"),
1371
                             driver->pidDir);
1372
        goto cleanup;
1373 1374
    }

1375
    if (!(pidfile = virPidFileBuildPath(driver->pidDir,
1376
                                        network->def->name)))
1377
        goto cleanup;
1378

1379
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1380
        virReportSystemError(errno,
1381
                             _("cannot create directory %s"),
1382
                             driver->dnsmasqStateDir);
1383 1384 1385
        goto cleanup;
    }

1386
    dctx = dnsmasqContextNew(network->def->name, driver->dnsmasqStateDir);
1387 1388 1389
    if (dctx == NULL)
        goto cleanup;

1390
    if (networkDnsmasqCapsRefresh(driver) < 0)
1391
        goto cleanup;
1392

1393
    ret = networkBuildDhcpDaemonCommandLine(driver, network, &cmd, pidfile, dctx);
1394 1395 1396 1397 1398
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1399
        goto cleanup;
1400

G
Guido Günther 已提交
1401
    ret = virCommandRun(cmd, NULL);
1402
    if (ret < 0)
1403 1404 1405
        goto cleanup;

    /*
1406 1407 1408 1409 1410
     * 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
1411 1412
     */

1413
    ret = virPidFileRead(driver->pidDir, network->def->name,
1414 1415
                         &network->dnsmasqPid);
    if (ret < 0)
1416
        goto cleanup;
1417

1418
    ret = 0;
1419
 cleanup:
1420
    VIR_FREE(pidfile);
1421
    virCommandFree(cmd);
1422
    dnsmasqContextFree(dctx);
1423 1424 1425
    return ret;
}

1426 1427
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1428 1429
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1430 1431 1432
 *
 *  Returns 0 on success, -1 on failure.
 */
1433
static int
1434 1435
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1436
{
1437 1438
    int ret = -1;
    size_t i;
G
Gene Czarcinski 已提交
1439
    virNetworkIpDefPtr ipdef, ipv4def, ipv6def;
1440
    dnsmasqContext *dctx = NULL;
1441

G
Gene Czarcinski 已提交
1442
    /* if no IP addresses specified, nothing to do */
1443
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1444 1445
        return 0;

1446 1447
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1448
        return networkStartDhcpDaemon(driver, network);
1449

G
Gene Czarcinski 已提交
1450
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1451
    if (!(dctx = dnsmasqContextNew(network->def->name,
1452
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1453
        goto cleanup;
1454
    }
G
Gene Czarcinski 已提交
1455 1456 1457 1458 1459 1460

    /* 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;
1461 1462 1463
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
G
Gene Czarcinski 已提交
1464 1465
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1466 1467
    }

G
Gene Czarcinski 已提交
1468
    ipv6def = NULL;
1469 1470 1471
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1472 1473
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1474 1475
    }

G
Gene Czarcinski 已提交
1476
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1477
        goto cleanup;
G
Gene Czarcinski 已提交
1478 1479

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1480
        goto cleanup;
1481

G
Gene Czarcinski 已提交
1482
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
J
Ján Tomko 已提交
1483
        goto cleanup;
1484 1485

    if ((ret = dnsmasqSave(dctx)) < 0)
1486
        goto cleanup;
1487 1488

    ret = kill(network->dnsmasqPid, SIGHUP);
1489
 cleanup:
1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501
    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
1502 1503
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1504 1505 1506 1507 1508 1509
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1510
    }
1511
    /* now start dnsmasq if it should be started */
1512
    return networkStartDhcpDaemon(driver, network);
1513 1514
}

G
Gene Czarcinski 已提交
1515 1516 1517 1518 1519 1520
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";

1521 1522 1523
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1524
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1525 1526
    int ret = -1;
    size_t i;
1527
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
1528
    bool v6present = false, dhcp6 = false;
1529 1530

    *configstr = NULL;
1531

G
Gene Czarcinski 已提交
1532
    /* Check if DHCPv6 is needed */
1533 1534 1535
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548
        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;
    }

1549 1550 1551
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1552
    virBufferAsprintf(&configbuf, "interface %s\n"
1553 1554
                      "{\n"
                      "  AdvSendAdvert on;\n"
1555
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1556 1557 1558 1559 1560
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1561 1562

    /* add a section for each IPv6 address in the config */
1563 1564 1565
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
1566 1567 1568 1569 1570
        int prefix;
        char *netaddr;

        prefix = virNetworkIpDefPrefix(ipdef);
        if (prefix < 0) {
1571 1572 1573
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1574 1575
            goto cleanup;
        }
1576
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1577
            goto cleanup;
1578
        virBufferAsprintf(&configbuf,
1579
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1580 1581 1582
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1583 1584 1585
        VIR_FREE(netaddr);
    }

1586
    virBufferAddLit(&configbuf, "};\n");
1587

1588
    if (virBufferCheckError(&configbuf) < 0)
1589
        goto cleanup;
1590

1591 1592
    *configstr = virBufferContentAndReset(&configbuf);

1593
    ret = 0;
1594
 cleanup:
1595 1596 1597 1598 1599 1600
    virBufferFreeAndReset(&configbuf);
    return ret;
}

/* write file and return it's name (which must be freed by caller) */
static int
1601 1602 1603
networkRadvdConfWrite(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr network,
                      char **configFile)
1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618
{
    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;
1619 1620 1621 1622
        goto cleanup;
    }

    /* construct the filename */
1623
    if (!(*configFile = networkRadvdConfigFileName(driver, network->def->name)))
1624 1625
        goto cleanup;
    /* write the file */
1626
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1627 1628
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1629 1630 1631 1632 1633
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1634
 cleanup:
1635 1636 1637 1638 1639 1640
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1641 1642
networkStartRadvd(virNetworkDriverStatePtr driver,
                  virNetworkObjPtr network)
1643
{
1644
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1645 1646 1647 1648 1649 1650 1651 1652
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1653
    /* Is dnsmasq handling RA? */
1654
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
G
Gene Czarcinski 已提交
1655 1656 1657 1658
        ret = 0;
        goto cleanup;
    }

1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669
    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0)) {
        /* no IPv6 addresses, so we don't need to run radvd */
        ret = 0;
        goto cleanup;
    }

    if (!virFileIsExecutable(RADVD)) {
        virReportSystemError(errno,
                             _("Cannot find %s - "
                               "Possibly the package isn't installed"),
                             RADVD);
1670 1671 1672
        goto cleanup;
    }

1673
    if (virFileMakePath(driver->pidDir) < 0) {
1674 1675
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1676
                             driver->pidDir);
1677 1678
        goto cleanup;
    }
1679
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1680 1681
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1682
                             driver->radvdStateDir);
1683 1684 1685 1686
        goto cleanup;
    }

    /* construct pidfile name */
1687
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1688
        goto cleanup;
1689
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1690 1691
        goto cleanup;

1692
    if (networkRadvdConfWrite(driver, network, &configfile) < 0)
1693 1694
        goto cleanup;

1695 1696 1697 1698
    /* 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
1699
     * virPidFileRead() below will fail if we use them).
1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714
     * 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;

1715
    if (virPidFileRead(driver->pidDir, radvdpidbase, &network->radvdPid) < 0)
1716 1717 1718
        goto cleanup;

    ret = 0;
1719
 cleanup:
1720
    virObjectUnref(dnsmasq_caps);
1721 1722 1723 1724 1725 1726 1727
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1728
static int
1729 1730
networkRefreshRadvd(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
1731
{
1732
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
G
Gene Czarcinski 已提交
1733 1734 1735
    char *radvdpidbase;

    /* Is dnsmasq handling RA? */
1736 1737
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
        virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1738 1739 1740 1741 1742 1743 1744
        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)) {
1745
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1746 1747 1748 1749 1750
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }
1751
    virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1752

1753 1754
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
1755
        return networkStartRadvd(driver, network);
1756 1757 1758 1759 1760 1761

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

1762
    if (networkRadvdConfWrite(driver, network, NULL) < 0)
1763 1764 1765 1766 1767
        return -1;

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

1768 1769
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1770
static int
1771
networkRestartRadvd(virNetworkObjPtr network)
1772 1773 1774 1775 1776 1777 1778 1779 1780
{
    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
1781
        if ((networkKillDaemon(network->radvdPid, "radvd",
1782 1783 1784
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1785
            virPidFileDelete(driver->pidDir, radvdpidbase);
1786 1787 1788 1789 1790 1791 1792 1793 1794
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1795 1796
static int
networkRefreshDaemonsHelper(virNetworkObjPtr net,
1797
                            void *opaque)
1798
{
1799
    virNetworkDriverStatePtr driver = opaque;
1800

1801
    virObjectLock(net);
1802 1803 1804 1805 1806 1807 1808 1809 1810 1811
    if (virNetworkObjIsActive(net) &&
        ((net->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
        /* Only the three L3 network types that are configured by
         * libvirt will have a dnsmasq or radvd daemon associated
         * with them.  Here we send a SIGHUP to an existing
         * dnsmasq and/or radvd, or restart them if they've
         * disappeared.
         */
1812 1813
        networkRefreshDhcpDaemon(driver, net);
        networkRefreshRadvd(driver, net);
1814
    }
1815
    virObjectUnlock(net);
1816 1817 1818
    return 0;
}

1819 1820 1821 1822
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
1823
networkRefreshDaemons(virNetworkDriverStatePtr driver)
1824 1825
{
    VIR_INFO("Refreshing network daemons");
1826 1827
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
1828
                             driver);
1829
}
1830

1831 1832 1833 1834 1835
static int
networkReloadFirewallRulesHelper(virNetworkObjPtr net,
                                 void *opaque ATTRIBUTE_UNUSED)
{

1836
    virObjectLock(net);
1837 1838 1839 1840 1841 1842 1843 1844 1845 1846
    if (virNetworkObjIsActive(net) &&
        ((net->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
        /* Only the three L3 network types that are configured by libvirt
         * need to have iptables rules reloaded.
         */
        networkRemoveFirewallRules(net->def);
        if (networkAddFirewallRules(net->def) < 0) {
            /* failed to add but already logged */
1847 1848
        }
    }
1849
    virObjectUnlock(net);
1850
    return 0;
1851 1852
}

1853
static void
1854
networkReloadFirewallRules(virNetworkDriverStatePtr driver)
1855
{
1856
    VIR_INFO("Reloading iptables rules");
1857 1858 1859
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
1860 1861
}

1862
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
1863
static int
1864
networkEnableIpForwarding(bool enableIPv4, bool enableIPv6)
1865
{
1866
    int ret = 0;
1867 1868 1869 1870
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
1871
                           &enabled, sizeof(enabled));
1872 1873
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
1874
                           &enabled, sizeof(enabled));
1875
#else
1876 1877 1878 1879
    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);
1880
#endif
1881
    return ret;
1882 1883
}

1884 1885
#define SYSCTL_PATH "/proc/sys"

1886 1887
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
1888 1889 1890
{
    char *field = NULL;
    int ret = -1;
1891
    bool enableIPv6 =  !!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0);
1892

1893 1894 1895 1896 1897 1898 1899
    /* 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;
1900

1901 1902
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
1903 1904
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
1905 1906 1907
        ret = 0;
        goto cleanup;
    }
1908

1909 1910 1911 1912 1913
    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;
1914
    }
1915
    VIR_FREE(field);
1916

1917 1918
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
1919 1920 1921 1922 1923 1924
     */

    /* 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",
1925
                    network->def->bridge) < 0)
1926 1927
        goto cleanup;

1928
    if (virFileWriteStr(field, "0", 0) < 0) {
1929
        virReportSystemError(errno,
1930 1931 1932 1933 1934
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

1935 1936 1937 1938
    /* 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",
1939
                    network->def->bridge) < 0)
1940 1941
        goto cleanup;

1942
    if (virFileWriteStr(field, "0", 0) < 0) {
1943
        virReportSystemError(errno,
1944
                             _("cannot disable %s"), field);
1945 1946 1947 1948
        goto cleanup;
    }

    ret = 0;
1949
 cleanup:
1950 1951 1952 1953
    VIR_FREE(field);
    return ret;
}

1954
/* add an IP address to a bridge */
1955
static int
D
Daniel P. Berrange 已提交
1956
networkAddAddrToBridge(virNetworkObjPtr network,
1957
                       virNetworkIpDefPtr ipdef)
1958
{
1959 1960 1961
    int prefix = virNetworkIpDefPrefix(ipdef);

    if (prefix < 0) {
1962 1963 1964
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
1965 1966 1967
        return -1;
    }

1968 1969
    if (virNetDevSetIPAddress(network->def->bridge,
                              &ipdef->address, prefix) < 0)
1970 1971 1972 1973 1974
        return -1;

    return 0;
}

1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

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;
}


1998 1999 2000 2001 2002
/* add an IP (static) route to a bridge */
static int
networkAddRouteToBridge(virNetworkObjPtr network,
                        virNetworkRouteDefPtr routedef)
{
2003 2004 2005 2006
    int prefix = virNetworkRouteDefGetPrefix(routedef);
    unsigned int metric = virNetworkRouteDefGetMetric(routedef);
    virSocketAddrPtr addr = virNetworkRouteDefGetAddress(routedef);
    virSocketAddrPtr gateway = virNetworkRouteDefGetGateway(routedef);
2007 2008 2009 2010 2011 2012 2013 2014 2015

    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;
    }

2016 2017
    if (virNetDevAddRoute(network->def->bridge, addr,
                          prefix, gateway, metric) < 0) {
2018 2019 2020 2021 2022
        return -1;
    }
    return 0;
}

2023
static int
2024 2025
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
                           virNetworkObjPtr network)
2026
{
2027
    size_t i;
2028
    bool v4present = false, v6present = false;
2029 2030
    virErrorPtr save_err = NULL;
    virNetworkIpDefPtr ipdef;
2031
    virNetworkRouteDefPtr routedef;
2032
    char *macTapIfName = NULL;
2033
    int tapfd = -1;
2034

2035
    /* Check to see if any network IP collides with an existing route */
2036
    if (networkCheckRouteCollision(network->def) < 0)
2037 2038
        return -1;

2039
    /* Create and configure the bridge device */
2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053
    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;
    }
2054
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
2055 2056
        return -1;

2057 2058 2059 2060 2061 2062 2063 2064
    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);
2065
        if (!macTapIfName)
2066
            goto err0;
2067
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2068
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
2069
                                           &macTapIfName, &network->def->mac,
2070
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2071 2072 2073
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2074 2075 2076 2077 2078
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2079
    /* Set bridge options */
2080 2081 2082 2083

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2084
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2085
                                   network->def->delay * 1000) < 0)
2086
        goto err1;
2087

2088
    if (virNetDevBridgeSetSTP(network->def->bridge,
2089
                              network->def->stp ? true : false) < 0)
2090
        goto err1;
2091

2092 2093 2094 2095
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2096
        goto err1;
2097

2098
    /* Add "once per network" rules */
2099
    if (networkAddFirewallRules(network->def) < 0)
2100 2101
        goto err1;

2102 2103 2104
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
2105
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2106
            v4present = true;
2107
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2108
            v6present = true;
2109

2110
        /* Add the IP address/netmask to the bridge */
2111
        if (networkAddAddrToBridge(network, ipdef) < 0)
2112
            goto err2;
2113 2114
    }

2115 2116 2117
    if (networkStartHandleMACTableManagerMode(network, macTapIfName) < 0)
        goto err2;

2118
    /* Bring up the bridge interface */
2119
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2120
        goto err2;
2121

2122
    for (i = 0; i < network->def->nroutes; i++) {
2123 2124 2125 2126 2127
        virSocketAddrPtr gateway = NULL;

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

2128 2129 2130
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2131
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2132 2133 2134 2135 2136 2137 2138
            if (networkAddRouteToBridge(network, routedef) < 0) {
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2139 2140
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2141
        networkEnableIpForwarding(v4present, v6present) < 0) {
2142
        virReportSystemError(errno, "%s",
2143
                             _("failed to enable IP forwarding"));
2144
        goto err3;
2145 2146
    }

2147

2148
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2149
    if ((v4present || v6present) &&
2150
        networkStartDhcpDaemon(driver, network) < 0)
2151
        goto err3;
2152

2153
    /* start radvd if there are any ipv6 addresses */
2154
    if (v6present && networkStartRadvd(driver, network) < 0)
2155 2156
        goto err4;

2157 2158 2159 2160 2161 2162 2163 2164 2165
    /* DAD has happened (dnsmasq waits for it), dnsmasq is now bound to the
     * bridge's IPv6 address, so we can now set the dummy tun down.
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
            goto err4;
        VIR_FORCE_CLOSE(tapfd);
    }

2166
    if (virNetDevBandwidthSet(network->def->bridge,
2167
                              network->def->bandwidth, true) < 0)
2168 2169
        goto err5;

2170
    VIR_FREE(macTapIfName);
2171 2172 2173

    return 0;

2174
 err5:
2175 2176
    if (network->def->bandwidth)
       virNetDevBandwidthClear(network->def->bridge);
2177

2178 2179 2180 2181
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2182 2183 2184 2185 2186
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2187 2188 2189
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2190
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2191

2192 2193 2194
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2195
    networkRemoveFirewallRules(network->def);
2196 2197

 err1:
2198 2199 2200
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2201
    if (macTapIfName) {
2202
        VIR_FORCE_CLOSE(tapfd);
2203
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2204 2205
        VIR_FREE(macTapIfName);
    }
2206 2207

 err0:
2208 2209
    if (!save_err)
        save_err = virSaveLastError();
2210
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2211

2212 2213 2214 2215
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2216
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2217 2218 2219
    return -1;
}

2220 2221 2222
static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
                              virNetworkObjPtr network)
2223
{
2224 2225
    if (network->def->bandwidth)
        virNetDevBandwidthClear(network->def->bridge);
2226

2227 2228 2229 2230 2231
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2232
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2233
            virPidFileDelete(driver->pidDir, radvdpidbase);
2234 2235 2236 2237
            VIR_FREE(radvdpidbase);
        }
    }

2238 2239 2240
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2241
    if (network->def->mac_specified) {
2242
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2243
        if (macTapIfName) {
2244
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2245 2246 2247 2248
            VIR_FREE(macTapIfName);
        }
    }

2249
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2250

2251
    networkRemoveFirewallRules(network->def);
2252

2253
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2254

2255
    /* See if its still alive and really really kill it */
2256
    if (network->dnsmasqPid > 0 &&
2257
        (kill(network->dnsmasqPid, 0) == 0))
2258 2259
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2260 2261 2262 2263 2264 2265

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

2266 2267 2268
    return 0;
}

2269

2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290
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;
}


2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305
/* networkCreateInterfacePool:
 * @netdef: the original NetDef from the network
 *
 * Creates an implicit interface pool of VF's when a PF dev is given
 */
static int
networkCreateInterfacePool(virNetworkDefPtr netdef)
{
    size_t numVirtFns = 0;
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2306 2307 2308
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev,
                                      &vfNames, &virtFns, &numVirtFns)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Could not get Virtual functions on %s"),
                       netdef->forward.pfs->dev);
        goto cleanup;
    }

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

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

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

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

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

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

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

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


2397
static int
2398
networkStartNetworkExternal(virNetworkObjPtr network)
2399 2400
{
    /* put anything here that needs to be done each time a network of
2401
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2402 2403 2404
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2405
    return networkCreateInterfacePool(network->def);
2406 2407
}

2408
static int networkShutdownNetworkExternal(virNetworkObjPtr network ATTRIBUTE_UNUSED)
2409 2410
{
    /* put anything here that needs to be done each time a network of
2411
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2412 2413 2414 2415 2416 2417 2418
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2419 2420
networkStartNetwork(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
2421
{
2422 2423 2424
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2425 2426

    if (virNetworkObjIsActive(network)) {
2427 2428
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2429
        return ret;
2430 2431
    }

2432 2433 2434
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2435
    if (virNetworkObjSetDefTransient(network, true) < 0)
2436
        goto cleanup;
2437

2438 2439
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2440
    if (networkRunHook(network, NULL, NULL,
2441 2442 2443 2444
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

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

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2450
        if (networkStartNetworkVirtual(driver, network) < 0)
2451
            goto cleanup;
2452 2453 2454
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2455 2456 2457 2458
       if (networkStartNetworkBridge(network) < 0)
          goto cleanup;
       break;

2459 2460 2461
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2462
    case VIR_NETWORK_FORWARD_HOSTDEV:
2463
        if (networkStartNetworkExternal(network) < 0)
2464
            goto cleanup;
2465 2466 2467
        break;
    }

2468
    /* finally we can call the 'started' hook script if any */
2469
    if (networkRunHook(network, NULL, NULL,
2470 2471 2472 2473
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2474 2475 2476
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2477
    VIR_DEBUG("Writing network status to disk");
2478
    if (virNetworkSaveStatus(driver->stateDir, network) < 0)
2479
        goto cleanup;
2480 2481

    network->active = 1;
2482 2483
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2484

2485
 cleanup:
2486
    if (ret < 0) {
2487
        virNetworkObjUnsetDefTransient(network);
2488 2489
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2490
        networkShutdownNetwork(driver, network);
2491 2492 2493 2494 2495 2496 2497
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2498 2499 2500
static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
2501 2502 2503 2504 2505 2506 2507 2508 2509
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

2510
    stateFile = virNetworkConfigFile(driver->stateDir,
2511
                                     network->def->name);
2512 2513 2514 2515 2516 2517
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2518
    switch (network->def->forward.type) {
2519 2520 2521 2522

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2523
        ret = networkShutdownNetworkVirtual(driver, network);
2524 2525 2526
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2527 2528 2529
        ret = networkShutdownNetworkBridge(network);
        break;

2530 2531 2532
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2533
    case VIR_NETWORK_FORWARD_HOSTDEV:
2534
        ret = networkShutdownNetworkExternal(network);
2535 2536 2537
        break;
    }

2538
    /* now that we know it's stopped call the hook if present */
2539
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2540 2541
                   VIR_HOOK_SUBOP_END);

2542
    network->active = 0;
2543
    virNetworkObjUnsetDefTransient(network);
2544
    return ret;
2545 2546 2547
}


2548
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2549 2550
                                         const unsigned char *uuid)
{
2551
    virNetworkDriverStatePtr driver = networkGetDriver();
2552 2553
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2554

2555
    network = virNetworkObjFindByUUID(driver->networks, uuid);
2556
    if (!network) {
2557 2558
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2559
        virReportError(VIR_ERR_NO_NETWORK,
2560 2561
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2562
        goto cleanup;
2563 2564
    }

2565 2566 2567
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2568 2569
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2570
 cleanup:
2571
    virNetworkObjEndAPI(&network);
2572
    return ret;
2573 2574
}

2575
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2576 2577
                                         const char *name)
{
2578
    virNetworkDriverStatePtr driver = networkGetDriver();
2579 2580 2581
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2582
    network = virNetworkObjFindByName(driver->networks, name);
2583
    if (!network) {
2584 2585
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2586
        goto cleanup;
2587 2588
    }

2589 2590 2591
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2592 2593
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2594
 cleanup:
2595
    virNetworkObjEndAPI(&network);
2596
    return ret;
2597 2598
}

2599 2600
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2601
    virNetworkDriverStatePtr driver = networkGetDriver();
2602
    int nactive;
2603

2604 2605 2606
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2607 2608 2609 2610
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
2611

2612 2613 2614
    return nactive;
}

2615 2616 2617 2618 2619
static int networkConnectListNetworks(virConnectPtr conn,
                                      char **const names,
                                      int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2620
    int got = 0;
2621

2622 2623 2624
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2625 2626 2627 2628
    got = virNetworkObjListGetNames(driver->networks,
                                    true, names, nnames,
                                    virConnectListNetworksCheckACL,
                                    conn);
2629

2630 2631 2632
    return got;
}

2633 2634
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2635
    virNetworkDriverStatePtr driver = networkGetDriver();
2636
    int ninactive = 0;
2637

2638 2639 2640
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2641 2642 2643 2644
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
2645

2646 2647 2648
    return ninactive;
}

2649 2650 2651
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2652
    int got = 0;
2653

2654 2655 2656
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2657 2658 2659 2660
    got = virNetworkObjListGetNames(driver->networks,
                                    false, names, nnames,
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
2661 2662 2663
    return got;
}

2664
static int
2665 2666 2667
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2668
{
2669
    virNetworkDriverStatePtr driver = networkGetDriver();
2670 2671 2672 2673
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2674 2675 2676
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2677
    ret = virNetworkObjListExport(conn, driver->networks, nets,
2678 2679
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2680

2681
 cleanup:
2682 2683
    return ret;
}
2684

2685 2686 2687 2688 2689 2690 2691 2692
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
2693
    virNetworkDriverStatePtr driver = networkGetDriver();
2694 2695 2696 2697 2698 2699
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
2700
                                       net, eventID, callback,
2701 2702 2703
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2704
 cleanup:
2705 2706 2707 2708 2709 2710 2711
    return ret;
}

static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
2712
    virNetworkDriverStatePtr driver = networkGetDriver();
2713 2714 2715 2716 2717
    int ret = -1;

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

2718 2719 2720 2721 2722 2723
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
2724

2725
 cleanup:
2726 2727 2728
    return ret;
}

2729 2730 2731 2732 2733
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2734 2735
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2736 2737 2738 2739

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

2740 2741
    ret = virNetworkObjIsActive(obj);

2742
 cleanup:
2743
    virNetworkObjEndAPI(&obj);
2744 2745 2746 2747 2748 2749 2750 2751
    return ret;
}

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

2752 2753
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2754 2755 2756 2757

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

2758 2759
    ret = obj->persistent;

2760
 cleanup:
2761
    virNetworkObjEndAPI(&obj);
2762 2763 2764 2765
    return ret;
}


2766 2767
/*
 * networkFindUnusedBridgeName() - try to find a bridge name that is
2768 2769 2770
 * 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.
2771 2772 2773 2774 2775 2776 2777 2778
 */
static int
networkFindUnusedBridgeName(virNetworkObjListPtr nets,
                            virNetworkDefPtr def)
{

    int ret = -1, id = 0;
    char *newname = NULL;
2779 2780 2781 2782 2783
    const char *templ = "virbr%d";
    const char *p;

    if (def->bridge &&
        (p = strchr(def->bridge, '%')) == strrchr(def->bridge, '%') &&
2784
        p && p[1] == 'd')
2785
        templ = def->bridge;
2786 2787 2788 2789

    do {
        if (virAsprintf(&newname, templ, id) < 0)
            goto cleanup;
2790 2791 2792 2793 2794 2795 2796
        /* 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)) {
2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848
            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;
}


2849
static int
2850
networkValidate(virNetworkDriverStatePtr driver,
2851
                virNetworkDefPtr def)
2852
{
2853
    size_t i, j;
2854 2855
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2856
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
2857
    bool ipv4def = false, ipv6def = false;
2858
    bool bandwidthAllowed = true;
2859
    bool usesInterface = false, usesAddress = false;
2860 2861 2862 2863

    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
2864 2865 2866
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2867

2868 2869 2870 2871
        /* 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)
2872 2873 2874
            return -1;

        virNetworkSetBridgeMacAddr(def);
2875 2876
    } else {
        /* They are also the only types that currently support setting
2877 2878
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
2879
         */
2880 2881 2882 2883 2884 2885 2886 2887
        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;
        }
2888 2889 2890 2891 2892
        if (virNetworkDefGetIpByIndex(def, AF_UNSPEC, 0)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2893
                           virNetworkForwardTypeToString(def->forward.type));
2894 2895
            return -1;
        }
2896
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
2897 2898 2899 2900
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2901
                           virNetworkForwardTypeToString(def->forward.type));
2902 2903 2904 2905 2906 2907 2908
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2909
                           virNetworkForwardTypeToString(def->forward.type));
2910 2911
            return -1;
        }
2912 2913 2914 2915 2916 2917 2918 2919
        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;
        }
2920
        bandwidthAllowed = false;
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
    /* 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
2957 2958 2959
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
2960 2961 2962
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
2963 2964 2965 2966 2967
        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 -- "
2968 2969
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986
                    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;
                }
2987 2988 2989
            }
        }
    }
2990 2991 2992 2993 2994 2995

    /* 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.
     */

2996
    vlanAllowed = ((def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
J
Ján Tomko 已提交
2997 2998
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
2999 3000
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) ||
                   def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV);
3001 3002

    vlanUsed = def->vlan.nTags > 0;
3003 3004
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3005 3006 3007 3008 3009
            /* 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.
             */
3010
            if (def->portGroups[i].virtPortProfile) {
3011
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3012
                    def->portGroups[i].virtPortProfile->virtPortType
3013 3014 3015 3016 3017 3018 3019
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3020
        }
3021
        if (def->portGroups[i].isDefault) {
3022 3023 3024 3025 3026
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3027
                               def->name, defaultPortGroup->name,
3028
                               def->portGroups[i].name);
3029
                return -1;
3030
            }
3031
            defaultPortGroup = &def->portGroups[i];
3032
        }
3033 3034 3035 3036 3037 3038 3039 3040 3041
        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;
            }
        }
3042 3043 3044 3045 3046 3047 3048 3049
        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;
        }
3050
    }
3051 3052 3053 3054 3055 3056 3057
    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.
         */
3058 3059 3060 3061 3062 3063 3064 3065 3066
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
    return 0;
}

3067 3068
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
3069
    virNetworkDriverStatePtr driver = networkGetDriver();
3070
    virNetworkDefPtr def;
3071
    virNetworkObjPtr network = NULL;
3072
    virNetworkPtr ret = NULL;
3073
    virObjectEventPtr event = NULL;
3074

3075
    if (!(def = virNetworkDefParseString(xml)))
3076
        goto cleanup;
3077

3078 3079 3080
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3081
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3082
        goto cleanup;
3083

3084 3085 3086
    /* 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.
3087
     */
3088 3089 3090
    if (!(network = virNetworkAssignDef(driver->networks, def,
                                        VIR_NETWORK_OBJ_LIST_ADD_LIVE |
                                        VIR_NETWORK_OBJ_LIST_ADD_CHECK_LIVE)))
3091 3092
        goto cleanup;
    def = NULL;
3093

3094
    if (networkStartNetwork(driver, network) < 0) {
3095
        virNetworkRemoveInactive(driver->networks,
3096
                                 network);
3097
        goto cleanup;
3098 3099
    }

3100 3101
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3102 3103
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3104

3105
    VIR_INFO("Creating network '%s'", network->def->name);
3106 3107
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

3108
 cleanup:
3109
    virNetworkDefFree(def);
3110 3111
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3112
    virNetworkObjEndAPI(&network);
3113
    return ret;
3114 3115
}

3116 3117
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
3118
    virNetworkDriverStatePtr driver = networkGetDriver();
3119
    virNetworkDefPtr def = NULL;
3120
    bool freeDef = true;
3121
    virNetworkObjPtr network = NULL;
3122
    virNetworkPtr ret = NULL;
3123
    virObjectEventPtr event = NULL;
3124

3125
    if (!(def = virNetworkDefParseString(xml)))
3126
        goto cleanup;
3127

3128 3129 3130
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3131
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3132
        goto cleanup;
3133

3134
    if (!(network = virNetworkAssignDef(driver->networks, def, 0)))
J
Ján Tomko 已提交
3135
        goto cleanup;
3136

3137
    /* def was assigned to network object */
3138
    freeDef = false;
3139 3140

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3141
        if (!virNetworkObjIsActive(network)) {
3142
            virNetworkRemoveInactive(driver->networks, network);
3143 3144
            goto cleanup;
        }
3145 3146 3147 3148 3149
        /* 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);
3150 3151 3152
        goto cleanup;
    }

3153
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3154 3155
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3156

3157 3158
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
3159

3160
 cleanup:
3161 3162
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3163
    if (freeDef)
J
Ján Tomko 已提交
3164
        virNetworkDefFree(def);
3165
    virNetworkObjEndAPI(&network);
3166
    return ret;
3167 3168
}

3169
static int
3170 3171
networkUndefine(virNetworkPtr net)
{
3172
    virNetworkDriverStatePtr driver = networkGetDriver();
3173
    virNetworkObjPtr network;
3174
    int ret = -1;
3175
    bool active = false;
3176
    virObjectEventPtr event = NULL;
3177

3178
    if (!(network = networkObjFromNetwork(net)))
3179
        goto cleanup;
3180

3181 3182 3183
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3184 3185
    if (virNetworkObjIsActive(network))
        active = true;
3186

3187
    /* remove autostart link */
3188
    if (virNetworkDeleteConfig(driver->networkConfigDir,
3189 3190
                               driver->networkAutostartDir,
                               network) < 0)
3191
        goto cleanup;
3192

3193 3194
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3195 3196
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3197

3198
    VIR_INFO("Undefining network '%s'", network->def->name);
3199
    if (!active) {
3200
        if (networkRemoveInactive(driver, network) < 0)
3201
            goto cleanup;
3202 3203 3204 3205 3206 3207
    } 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);
3208 3209
    }

3210
    ret = 0;
3211

3212
 cleanup:
3213 3214
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3215
    virNetworkObjEndAPI(&network);
3216
    return ret;
3217 3218
}

3219 3220 3221 3222 3223 3224 3225 3226
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3227
    virNetworkDriverStatePtr driver = networkGetDriver();
3228
    virNetworkObjPtr network = NULL;
3229 3230
    int isActive, ret = -1;
    size_t i;
3231 3232
    virNetworkIpDefPtr ipdef;
    bool oldDhcpActive = false;
3233
    bool needFirewallRefresh = false;
3234

3235 3236 3237 3238 3239

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3240
    if (!(network = networkObjFromNetwork(net)))
3241 3242
        goto cleanup;

3243 3244 3245
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3246
    /* see if we are listening for dhcp pre-modification */
3247 3248 3249
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
3250 3251 3252 3253 3254 3255
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3256 3257
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3258
     */
3259
    isActive = virNetworkObjIsActive(network);
3260 3261
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3262 3263 3264 3265 3266 3267 3268
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285
    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).
                 */
3286
                networkRemoveFirewallRules(network->def);
3287 3288 3289 3290 3291 3292 3293 3294
                needFirewallRefresh = true;
                break;
            default:
                break;
            }
        }
    }

3295
    /* update the network config in memory/on disk */
3296 3297
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3298
            ignore_value(networkAddFirewallRules(network->def));
3299 3300 3301
        goto cleanup;
    }

3302
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325
        goto cleanup;

    if (flags & VIR_NETWORK_UPDATE_AFFECT_CONFIG) {
        /* save updated persistent config to disk */
        if (virNetworkSaveConfig(driver->networkConfigDir,
                                 virNetworkObjGetPersistentDef(network)) < 0) {
            goto cleanup;
        }
    }

    if (isActive && (flags & VIR_NETWORK_UPDATE_AFFECT_LIVE)) {
        /* rewrite dnsmasq host files, restart dnsmasq, update iptables
         * rules, etc, according to which section was modified. Note that
         * some sections require multiple actions, so a single switch
         * statement is inadequate.
         */
        if (section == VIR_NETWORK_SECTION_BRIDGE ||
            section == VIR_NETWORK_SECTION_DOMAIN ||
            section == VIR_NETWORK_SECTION_IP ||
            section == VIR_NETWORK_SECTION_IP_DHCP_RANGE) {
            /* these sections all change things on the dnsmasq commandline,
             * so we need to kill and restart dnsmasq.
             */
3326
            if (networkRestartDhcpDaemon(driver, network) < 0)
3327 3328
                goto cleanup;

3329 3330 3331 3332 3333 3334 3335 3336
        } 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;

3337 3338 3339
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
                 i++) {
3340 3341 3342 3343 3344 3345 3346
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3347 3348
                 networkRestartDhcpDaemon(driver, network) < 0) ||
                networkRefreshDhcpDaemon(driver, network) < 0) {
3349 3350 3351 3352
                goto cleanup;
            }

        } else if (section == VIR_NETWORK_SECTION_DNS_HOST ||
3353 3354 3355 3356 3357 3358
                   section == VIR_NETWORK_SECTION_DNS_TXT ||
                   section == VIR_NETWORK_SECTION_DNS_SRV) {
            /* these sections only change things in config files, so we
             * can just update the config files and send SIGHUP to
             * dnsmasq.
             */
3359
            if (networkRefreshDhcpDaemon(driver, network) < 0)
3360 3361 3362 3363 3364 3365 3366 3367
                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.
             */
3368
            if (networkRefreshRadvd(driver, network) < 0)
3369 3370 3371 3372
                goto cleanup;
        }

        /* save current network state to disk */
3373
        if ((ret = virNetworkSaveStatus(driver->stateDir,
3374
                                        network)) < 0) {
3375
            goto cleanup;
3376
        }
3377 3378
    }
    ret = 0;
3379
 cleanup:
3380
    virNetworkObjEndAPI(&network);
3381 3382 3383
    return ret;
}

3384 3385
static int networkCreate(virNetworkPtr net)
{
3386
    virNetworkDriverStatePtr driver = networkGetDriver();
3387 3388
    virNetworkObjPtr network;
    int ret = -1;
3389
    virObjectEventPtr event = NULL;
3390

3391
    if (!(network = networkObjFromNetwork(net)))
3392
        goto cleanup;
3393

3394 3395 3396
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3397
    if ((ret = networkStartNetwork(driver, network)) < 0)
3398
        goto cleanup;
3399

3400 3401
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3402 3403
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3404

3405
 cleanup:
3406 3407
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3408
    virNetworkObjEndAPI(&network);
3409
    return ret;
3410 3411
}

3412 3413
static int networkDestroy(virNetworkPtr net)
{
3414
    virNetworkDriverStatePtr driver = networkGetDriver();
3415 3416
    virNetworkObjPtr network;
    int ret = -1;
3417
    virObjectEventPtr event = NULL;
3418

3419
    if (!(network = networkObjFromNetwork(net)))
3420
        goto cleanup;
3421

3422 3423 3424
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3425
    if (!virNetworkObjIsActive(network)) {
3426
        virReportError(VIR_ERR_OPERATION_INVALID,
3427 3428
                       _("network '%s' is not active"),
                       network->def->name);
3429 3430 3431
        goto cleanup;
    }

3432
    if ((ret = networkShutdownNetwork(driver, network)) < 0)
3433 3434
        goto cleanup;

3435 3436
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3437 3438
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3439

3440
    if (!network->persistent &&
3441
        networkRemoveInactive(driver, network) < 0) {
3442 3443
        ret = -1;
        goto cleanup;
3444
    }
3445

3446
 cleanup:
3447 3448
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3449
    virNetworkObjEndAPI(&network);
3450 3451 3452
    return ret;
}

3453
static char *networkGetXMLDesc(virNetworkPtr net,
3454
                               unsigned int flags)
3455
{
3456
    virNetworkObjPtr network;
3457
    virNetworkDefPtr def;
3458
    char *ret = NULL;
3459

3460
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3461

3462 3463
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3464

3465 3466 3467
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3468 3469 3470 3471 3472 3473
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3474

3475
 cleanup:
3476
    virNetworkObjEndAPI(&network);
3477
    return ret;
3478 3479 3480
}

static char *networkGetBridgeName(virNetworkPtr net) {
3481 3482 3483
    virNetworkObjPtr network;
    char *bridge = NULL;

3484 3485
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3486

3487 3488 3489
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3490
    if (!(network->def->bridge)) {
3491 3492 3493
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3494 3495 3496
        goto cleanup;
    }

3497
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3498

3499
 cleanup:
3500
    virNetworkObjEndAPI(&network);
3501 3502 3503 3504
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
J
Ján Tomko 已提交
3505
                               int *autostart)
3506
{
3507 3508
    virNetworkObjPtr network;
    int ret = -1;
3509

3510 3511
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3512

3513 3514 3515
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3516
    *autostart = network->autostart;
3517
    ret = 0;
3518

3519
 cleanup:
3520
    virNetworkObjEndAPI(&network);
3521
    return ret;
3522 3523 3524
}

static int networkSetAutostart(virNetworkPtr net,
3525 3526
                               int autostart)
{
3527
    virNetworkDriverStatePtr driver = networkGetDriver();
3528
    virNetworkObjPtr network;
3529
    char *configFile = NULL, *autostartLink = NULL;
3530
    int ret = -1;
3531

3532

3533
    if (!(network = networkObjFromNetwork(net)))
3534
        goto cleanup;
3535

3536 3537 3538
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3539
    if (!network->persistent) {
3540 3541
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3542 3543 3544
        goto cleanup;
    }

3545 3546
    autostart = (autostart != 0);

3547
    if (network->autostart != autostart) {
3548
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3549
            goto cleanup;
3550
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3551 3552
            goto cleanup;

3553
        if (autostart) {
3554
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3555
                virReportSystemError(errno,
3556 3557
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3558 3559
                goto cleanup;
            }
3560

3561
            if (symlink(configFile, autostartLink) < 0) {
3562
                virReportSystemError(errno,
3563
                                     _("Failed to create symlink '%s' to '%s'"),
3564
                                     autostartLink, configFile);
3565 3566 3567
                goto cleanup;
            }
        } else {
3568
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3569
                virReportSystemError(errno,
3570
                                     _("Failed to delete symlink '%s'"),
3571
                                     autostartLink);
3572 3573
                goto cleanup;
            }
3574 3575
        }

3576
        network->autostart = autostart;
3577
    }
3578
    ret = 0;
3579

3580
 cleanup:
3581 3582
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3583
    virNetworkObjEndAPI(&network);
3584
    return ret;
3585 3586
}

3587
static int
3588 3589 3590 3591
networkGetDHCPLeases(virNetworkPtr network,
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
3592
{
3593
    virNetworkDriverStatePtr driver = networkGetDriver();
3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
    int size = 0;
    int custom_lease_file_len = 0;
    bool need_results = !!leases;
    long long currtime = 0;
    long long expirytime_tmp = -1;
    bool ipv6 = false;
    char *lease_entries = NULL;
    char *custom_lease_file = NULL;
    const char *ip_tmp = NULL;
    const char *mac_tmp = NULL;
    virJSONValuePtr lease_tmp = NULL;
    virJSONValuePtr leases_array = NULL;
    virNetworkIpDefPtr ipdef_tmp = NULL;
    virNetworkDHCPLeasePtr lease = NULL;
    virNetworkDHCPLeasePtr *leases_ret = NULL;
3612 3613 3614 3615 3616 3617 3618 3619 3620
    virNetworkObjPtr obj;

    virCheckFlags(0, -1);

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

    if (virNetworkGetDHCPLeasesEnsureACL(network->conn, obj->def) < 0)
        goto cleanup;
3621 3622

    /* Retrieve custom leases file location */
3623
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(driver, obj->def->bridge);
3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666

    /* 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;
        }

3667
        if (mac && virMacAddrCompare(mac, mac_tmp))
3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717
            continue;

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

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

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

            lease->expirytime = expirytime_tmp;

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

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

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

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

            if ((VIR_STRDUP(lease->mac, mac_tmp) < 0) ||
                (VIR_STRDUP(lease->ipaddr, ip_tmp) < 0) ||
3718
                (VIR_STRDUP(lease->iface, obj->def->bridge) < 0))
3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750
                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);
3751
    VIR_FREE(lease_entries);
3752 3753
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
3754

3755
    virNetworkObjEndAPI(&obj);
3756

3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767
    return rv;

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

3768 3769

static virNetworkDriver networkDriver = {
3770
    .name = "bridge",
3771 3772 3773 3774 3775
    .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 */
3776 3777
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
3778 3779
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3780 3781
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3782
    .networkUndefine = networkUndefine, /* 0.2.0 */
3783
    .networkUpdate = networkUpdate, /* 0.10.2 */
3784
    .networkCreate = networkCreate, /* 0.2.0 */
3785 3786 3787 3788 3789 3790 3791
    .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 */
3792
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
3793 3794 3795
};

static virStateDriver networkStateDriver = {
3796
    .name = "bridge",
3797
    .stateInitialize  = networkStateInitialize,
3798
    .stateAutoStart  = networkStateAutoStart,
3799 3800
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3801 3802
};

3803 3804
int networkRegister(void)
{
3805
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
3806
        return -1;
3807 3808
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
3809 3810
    return 0;
}
3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822

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

/* 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:
3823
 * @dom: domain definition that @iface belongs to
3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834
 * @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
3835 3836
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
3837
{
3838
    virNetworkDriverStatePtr driver = networkGetDriver();
3839
    virDomainNetType actualType = iface->type;
3840 3841
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
3842
    virNetDevBandwidthPtr bandwidth = NULL;
3843 3844 3845
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3846
    virNetworkForwardIfDefPtr dev = NULL;
3847
    size_t i;
3848 3849 3850
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3851
        goto validate;
3852 3853 3854 3855

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

3856
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
3857
    if (!network) {
3858 3859 3860
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3861
        goto error;
3862 3863
    }
    netdef = network->def;
3864

3865 3866 3867 3868 3869 3870 3871
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

3872 3873 3874
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

3875 3876 3877
    /* 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
3878
     */
3879 3880 3881 3882 3883 3884
    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.
     */
3885 3886 3887 3888 3889 3890

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

3891 3892
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
3893
        goto error;
3894

3895 3896 3897 3898 3899 3900 3901 3902
    /* 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;

3903 3904
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
3905

3906 3907 3908 3909 3910 3911 3912 3913 3914 3915
    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;

3916 3917 3918
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
3919
        /* for these forward types, the actual net type really *is*
3920
         * NETWORK; we just keep the info from the portgroup in
3921
         * iface->data.network.actual
J
Ján Tomko 已提交
3922
         */
3923
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
3924

3925
        /* we also store the bridge device and macTableManager settings
3926 3927 3928 3929 3930 3931 3932
         * 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;
3933 3934
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
3935

3936 3937 3938
        if (networkPlugBandwidth(network, iface) < 0)
            goto error;

3939
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
3940
               netdef->bridge) {
3941 3942 3943 3944 3945

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

3946
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
3947 3948
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
3949
            goto error;
3950 3951
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
3952

3953 3954 3955 3956 3957 3958 3959 3960
        /* 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) {
3961
            goto error;
3962 3963 3964 3965 3966 3967 3968 3969 3970 3971
        }
        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);
3972
                goto error;
3973 3974 3975
            }
        }

3976
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3977

3978
        virDomainHostdevSubsysPCIBackendType backend;
3979

3980
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
3981
        if (networkCreateInterfacePool(netdef) < 0)
3982 3983 3984
            goto error;

        /* pick first dev with 0 connections */
3985 3986 3987
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001
                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;
4002
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
4003
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
4004
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
4005

E
Eric Blake 已提交
4006
        switch (netdef->forward.driverName) {
4007
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
4008
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
4009 4010
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
4011
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
4012 4013
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
4014
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025
            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;

4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050
        /* 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;
            }
        }

4051 4052 4053 4054
    } 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)) {
4055 4056 4057 4058 4059 4060

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
4061
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4062
        switch (netdef->forward.type) {
4063
        case VIR_NETWORK_FORWARD_BRIDGE:
4064
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
4065 4066
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
4067
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
4068 4069
            break;
        case VIR_NETWORK_FORWARD_VEPA:
4070
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
4071 4072
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
4073
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
4074 4075 4076
            break;
        }

4077 4078 4079 4080 4081 4082 4083 4084
        /* 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) {
4085
            goto error;
4086
        }
4087
        virtport = iface->data.network.actual->virtPortProfile;
4088
        if (virtport) {
4089 4090 4091 4092 4093 4094 4095 4096
            /* 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);
4097
                goto error;
4098 4099
            }
        }
4100

4101 4102 4103
        /* 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).
         */
4104
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4105 4106 4107 4108
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4109
            goto error;
4110 4111 4112
        } else {
            /* pick an interface from the pool */

4113
            if (networkCreateInterfacePool(netdef) < 0)
4114 4115
                goto error;

4116 4117 4118 4119 4120
            /* 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.
4121
             */
4122 4123
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4124 4125 4126
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4127

4128
                /* pick first dev with 0 connections */
4129 4130 4131
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4132 4133 4134 4135 4136
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4137
                dev = &netdef->forward.ifs[0];
4138 4139 4140
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4141 4142 4143 4144
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4145 4146 4147 4148
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4149
                goto error;
4150
            }
4151 4152
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4153
                goto error;
4154 4155 4156
        }
    }

4157
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4158
        goto error;
4159

4160
 validate:
4161 4162 4163 4164 4165
    /* 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.
     */

4166
    if (virDomainNetGetActualVlan(iface)) {
4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195
        /* vlan configuration via libvirt is only supported for
         * PCI Passthrough SR-IOV devices and openvswitch bridges.
         * otherwise log an error and fail
         */
        if (!(actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV ||
              (actualType == VIR_DOMAIN_NET_TYPE_BRIDGE &&
               virtport && virtport->virtPortType
               == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH))) {
            if (netdef) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("an interface connecting to network '%s' "
                                 "is requesting a vlan tag, but that is not "
                                 "supported for this type of network"),
                               netdef->name);
            } else {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("an interface of type '%s' "
                                 "is requesting a vlan tag, but that is not "
                                 "supported for this type of connection"),
                               virDomainNetTypeToString(iface->type));
            }
            goto error;
        }
    }

    if (netdef) {
        netdef->connections++;
        VIR_DEBUG("Using network %s, %d connections",
                  netdef->name, netdef->connections);
4196

4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215
        if (dev) {
            /* mark the allocation */
            dev->connections++;
            if (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV) {
                VIR_DEBUG("Using physical device %s, %d connections",
                          dev->device.dev, dev->connections);
            } else {
                VIR_DEBUG("Using physical device %04x:%02x:%02x.%x, connections %d",
                          dev->device.pci.domain, dev->device.pci.bus,
                          dev->device.pci.slot, dev->device.pci.function,
                          dev->connections);
            }
        }

        /* finally we can call the 'plugged' hook script if any */
        if (networkRunHook(network, dom, iface,
                           VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
                           VIR_HOOK_SUBOP_BEGIN) < 0) {
            /* adjust for failure */
4216
            netdef->connections--;
4217 4218 4219 4220
            if (dev)
                dev->connections--;
            goto error;
        }
4221 4222
    }

4223
    ret = 0;
4224

4225
 cleanup:
4226
    virNetworkObjEndAPI(&network);
4227 4228
    return ret;

4229
 error:
4230
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4231 4232 4233
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4234
    goto cleanup;
4235 4236 4237
}

/* networkNotifyActualDevice:
4238
 * @dom: domain definition that @iface belongs to
4239 4240 4241 4242 4243 4244 4245 4246 4247 4248
 * @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
4249 4250
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4251
{
4252
    virNetworkDriverStatePtr driver = networkGetDriver();
4253
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4254 4255
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4256
    virNetworkForwardIfDefPtr dev = NULL;
4257 4258
    size_t i;
    int ret = -1;
4259 4260

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4261
        return 0;
4262

4263
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4264
    if (!network) {
4265 4266 4267
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4268 4269 4270 4271
        goto error;
    }
    netdef = network->def;

4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282
    /* 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;

4283
    if (!iface->data.network.actual ||
4284 4285
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4286 4287
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4288 4289
    }

4290
    if (networkCreateInterfacePool(netdef) < 0)
4291
        goto error;
4292

4293
    if (netdef->forward.nifs == 0) {
4294
        virReportError(VIR_ERR_INTERNAL_ERROR,
4295 4296
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4297
                       netdef->name);
4298
        goto error;
4299
    }
4300

4301 4302
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4303

4304 4305 4306 4307 4308 4309 4310 4311 4312
        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 */
4313 4314
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4315
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4316 4317
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4318 4319 4320 4321 4322
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4323
            virReportError(VIR_ERR_INTERNAL_ERROR,
4324 4325
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4326
                           netdef->name, actualDev);
4327
            goto error;
4328 4329
        }

4330
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4331 4332
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4333
         */
4334
        if ((dev->connections > 0) &&
4335 4336
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4337 4338
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4339
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4340
            virReportError(VIR_ERR_INTERNAL_ERROR,
4341 4342
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4343
                           netdef->name, actualDev);
4344
            goto error;
4345
        }
4346

4347
        /* we are now assured of success, so mark the allocation */
4348
        dev->connections++;
4349
        VIR_DEBUG("Using physical device %s, connections %d",
4350
                  dev->device.dev, dev->connections);
4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363

    }  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 */
4364 4365
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4366
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4367
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4368 4369
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4370 4371 4372 4373 4374 4375 4376 4377 4378
                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,
4379 4380 4381 4382
                           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
4383
            goto error;
4384 4385 4386 4387 4388 4389 4390
        }

        /* 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) &&
4391
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' claims the PCI device at "
                             "domain=%d bus=%d slot=%d function=%d "
                             "is already in use by a different domain"),
                           netdef->name,
                           dev->device.pci.domain, dev->device.pci.bus,
                           dev->device.pci.slot, dev->device.pci.function);
            goto error;
        }

        /* we are now assured of success, so mark the allocation */
        dev->connections++;
        VIR_DEBUG("Using physical device %04x:%02x:%02x.%x, connections %d",
                  dev->device.pci.domain, dev->device.pci.bus,
                  dev->device.pci.slot, dev->device.pci.function,
                  dev->connections);
4408 4409
    }

4410
 success:
4411 4412 4413
    netdef->connections++;
    VIR_DEBUG("Using network %s, %d connections",
              netdef->name, netdef->connections);
4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424

    /* 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;
    }

4425
    ret = 0;
4426
 cleanup:
4427
    virNetworkObjEndAPI(&network);
4428
    return ret;
4429

4430
 error:
4431
    goto cleanup;
4432 4433 4434 4435
}


/* networkReleaseActualDevice:
4436
 * @dom: domain definition that @iface belongs to
4437 4438 4439 4440 4441 4442 4443 4444 4445 4446
 * @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
4447 4448
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4449
{
4450
    virNetworkDriverStatePtr driver = networkGetDriver();
4451
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4452
    virNetworkObjPtr network;
4453
    virNetworkDefPtr netdef;
4454
    virNetworkForwardIfDefPtr dev = NULL;
4455 4456
    size_t i;
    int ret = -1;
4457 4458

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4459
        return 0;
4460

4461
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4462
    if (!network) {
4463 4464 4465
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4466 4467 4468 4469
        goto error;
    }
    netdef = network->def;

4470 4471
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4472 4473 4474 4475 4476
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) &&
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4477 4478 4479
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4480 4481
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4482 4483
    }

4484
    if (netdef->forward.nifs == 0) {
4485
        virReportError(VIR_ERR_INTERNAL_ERROR,
4486
                       _("network '%s' uses a direct/hostdev mode, but "
4487 4488
                         "has no forward dev and no interface pool"),
                       netdef->name);
4489
        goto error;
4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501
    }

    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;
        }
4502

4503 4504
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4505
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4506 4507
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4508 4509 4510
                break;
            }
        }
4511

4512
        if (!dev) {
4513
            virReportError(VIR_ERR_INTERNAL_ERROR,
4514 4515
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4516
                           netdef->name, actualDev);
4517
            goto error;
4518 4519
        }

4520
        dev->connections--;
4521
        VIR_DEBUG("Releasing physical device %s, connections %d",
4522
                  dev->device.dev, dev->connections);
4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533

    } 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;
        }

4534 4535
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4536
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4537
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4538 4539
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4540 4541 4542 4543 4544 4545 4546 4547 4548
                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,
4549 4550 4551 4552
                           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
4553
            goto error;
4554 4555 4556 4557 4558 4559 4560
        }

        dev->connections--;
        VIR_DEBUG("Releasing physical device %04x:%02x:%02x.%x, connections %d",
                  dev->device.pci.domain, dev->device.pci.bus,
                  dev->device.pci.slot, dev->device.pci.function,
                  dev->connections);
J
Ján Tomko 已提交
4561
    }
4562

4563
 success:
4564
    if (iface->data.network.actual) {
4565
        netdef->connections--;
4566 4567
        VIR_DEBUG("Releasing network %s, %d connections",
                  netdef->name, netdef->connections);
4568

4569 4570 4571 4572
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
    }
4573
    ret = 0;
4574
 cleanup:
4575
    virNetworkObjEndAPI(&network);
4576 4577 4578 4579
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4580
    return ret;
4581

4582
 error:
4583
    goto cleanup;
4584
}
4585 4586 4587 4588 4589 4590

/*
 * networkGetNetworkAddress:
 * @netname: the name of a network
 * @netaddr: string representation of IP address for that network.
 *
4591
 * Attempt to return an IP address associated with the named
4592 4593 4594 4595
 * 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.
 *
4596
 * Note: This function returns the first IP address it finds. It might
4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607
 * 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)
{
4608
    virNetworkDriverStatePtr driver = networkGetDriver();
4609
    int ret = -1;
4610
    virNetworkObjPtr network;
4611 4612 4613 4614
    virNetworkDefPtr netdef;
    virNetworkIpDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4615
    char *dev_name = NULL;
4616 4617

    *netaddr = NULL;
4618
    network = virNetworkObjFindByName(driver->networks, netname);
4619
    if (!network) {
4620 4621 4622
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4623
        goto cleanup;
4624 4625 4626
    }
    netdef = network->def;

4627
    switch (netdef->forward.type) {
4628 4629 4630
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
4631
        ipdef = virNetworkDefGetIpByIndex(netdef, AF_UNSPEC, 0);
4632
        if (!ipdef) {
4633
            virReportError(VIR_ERR_INTERNAL_ERROR,
4634
                           _("network '%s' doesn't have an IP address"),
4635
                           netdef->name);
4636
            goto cleanup;
4637 4638 4639 4640 4641
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4642
        if ((dev_name = netdef->bridge))
4643 4644 4645 4646 4647 4648 4649 4650
            break;
        /*
         * fall through if netdef->bridge wasn't set, since this is
         * also a direct-mode interface.
         */
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
4651 4652
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4653

4654
        if (!dev_name) {
4655 4656 4657
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4658
            goto cleanup;
4659 4660 4661 4662
        }
        break;
    }

4663
    if (dev_name) {
4664
        if (virNetDevGetIPAddress(dev_name, &addr) < 0)
4665
            goto cleanup;
4666
        addrptr = &addr;
4667 4668
    }

4669 4670
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
4671
        goto cleanup;
4672 4673
    }

4674
    ret = 0;
4675
 cleanup:
4676
    virNetworkObjEndAPI(&network);
4677 4678
    return ret;
}
4679 4680 4681 4682

/**
 * networkCheckBandwidth:
 * @net: network QoS
4683 4684
 * @ifaceBand: interface QoS (may be NULL if no QoS)
 * @ifaceMac: interface MAC (used in error messages for identification)
4685 4686 4687 4688 4689 4690 4691 4692
 * @new_rate: new rate for non guaranteed class
 *
 * Returns: -1 if plugging would overcommit network QoS
 *           0 if plugging is safe (@new_rate updated)
 *           1 if no QoS is set (@new_rate untouched)
 */
static int
networkCheckBandwidth(virNetworkObjPtr net,
4693 4694
                      virNetDevBandwidthPtr ifaceBand,
                      virMacAddr ifaceMac,
4695 4696 4697 4698 4699 4700 4701 4702
                      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];

4703
    virMacAddrFormat(&ifaceMac, ifmac);
4704 4705 4706 4707 4708 4709 4710 4711 4712 4713

    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;
    }

4714
    if (!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor ||
4715 4716
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4717
        return 1;
4718
    }
4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749

    tmp_new_rate = netBand->in->average;
    tmp_floor_sum += ifaceBand->in->floor;

    /* check against peak */
    if (netBand->in->peak) {
        tmp_new_rate = netBand->in->peak;
        if (tmp_floor_sum > netBand->in->peak) {
            virReportError(VIR_ERR_OPERATION_INVALID,
                           _("Cannot plug '%s' interface into '%s' because it "
                             "would overcommit 'peak' on network '%s'"),
                           ifmac,
                           net->def->bridge,
                           net->def->name);
            goto cleanup;
        }
    } else if (tmp_floor_sum > netBand->in->average) {
        /* tmp_floor_sum can be between 'average' and 'peak' iff 'peak' is set.
         * Otherwise, tmp_floor_sum must be below 'average'. */
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("Cannot plug '%s' interface into '%s' because it "
                         "would overcommit 'average' on network '%s'"),
                       ifmac,
                       net->def->bridge,
                       net->def->name);
        goto cleanup;
    }

    *new_rate = tmp_new_rate;
    ret = 0;

4750
 cleanup:
4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766
    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)
{
4767
    ssize_t ret = 0;
4768

4769
    ret = virBitmapNextClearBit(net->class_id, -1);
4770

4771
    if (ret < 0 || virBitmapSetBit(net->class_id, ret) < 0)
4772 4773 4774 4775 4776 4777 4778 4779 4780
        return -1;

    return ret;
}

static int
networkPlugBandwidth(virNetworkObjPtr net,
                     virDomainNetDefPtr iface)
{
4781
    virNetworkDriverStatePtr driver = networkGetDriver();
4782 4783 4784 4785 4786
    int ret = -1;
    int plug_ret;
    unsigned long long new_rate = 0;
    ssize_t class_id = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];
4787
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4788

4789 4790
    if ((plug_ret = networkCheckBandwidth(net, ifaceBand,
                                          iface->mac, &new_rate)) < 0) {
4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816
        /* helper reported error */
        goto cleanup;
    }

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

    virMacAddrFormat(&iface->mac, ifmac);
    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK ||
        !iface->data.network.actual) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Cannot set bandwidth on interface '%s' of type %d"),
                       ifmac, iface->type);
        goto cleanup;
    }

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

4817 4818
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
4819 4820 4821 4822 4823 4824 4825 4826
    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 */
4827
    net->floor_sum += ifaceBand->in->floor;
4828
    /* update status file */
4829
    if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
4830
        ignore_value(virBitmapClearBit(net->class_id, class_id));
4831
        net->floor_sum -= ifaceBand->in->floor;
4832 4833 4834 4835
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
4836 4837 4838 4839 4840 4841 4842 4843 4844
    /* update rate for non guaranteed NICs */
    new_rate -= net->floor_sum;
    if (virNetDevBandwidthUpdateRate(net->def->bridge, "1:2",
                                     net->def->bandwidth, new_rate) < 0)
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                 net->def->bridge);

    ret = 0;

4845
 cleanup:
4846 4847 4848 4849 4850 4851 4852
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
4853
    virNetworkDriverStatePtr driver = networkGetDriver();
4854 4855
    int ret = 0;
    unsigned long long new_rate;
4856
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4857 4858 4859

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
4860 4861 4862 4863 4864
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875
        /* 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 */
4876
        net->floor_sum -= ifaceBand->in->floor;
4877 4878 4879 4880
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
4881
        if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
4882
            net->floor_sum += ifaceBand->in->floor;
4883 4884 4885 4886
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
4887 4888 4889 4890 4891 4892 4893 4894 4895 4896
        /* update rate for non guaranteed NICs */
        new_rate -= net->floor_sum;
        if (virNetDevBandwidthUpdateRate(net->def->bridge, "1:2",
                                         net->def->bandwidth, new_rate) < 0)
            VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                     net->def->bridge);
        /* no class is associated any longer */
        iface->data.network.actual->class_id = 0;
    }

4897
 cleanup:
4898 4899
    return ret;
}
4900 4901 4902

static void
networkNetworkObjTaint(virNetworkObjPtr net,
4903
                       virNetworkTaintFlags taint)
4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914
{
    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));
    }
}