bridge_driver.c 184.2 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2016 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
 */

#include <config.h>

#include <sys/types.h>
#include <sys/poll.h>
#include <stdarg.h>
#include <unistd.h>
#include <sys/utsname.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <signal.h>
#include <pwd.h>
#include <sys/wait.h>
#include <sys/ioctl.h>
35
#include <net/if.h>
36 37 38
#if HAVE_SYS_SYSCTL_H
# include <sys/sysctl.h>
#endif
39

40
#include "virerror.h"
41
#include "datatypes.h"
42
#include "bridge_driver.h"
43
#include "bridge_driver_platform.h"
44
#include "device_conf.h"
45
#include "driver.h"
46
#include "virbuffer.h"
47
#include "virpidfile.h"
48
#include "vircommand.h"
49
#include "viralloc.h"
50
#include "viruuid.h"
51
#include "viriptables.h"
52
#include "virlog.h"
53
#include "virdnsmasq.h"
54
#include "configmake.h"
55
#include "virnetlink.h"
56
#include "virnetdev.h"
57
#include "virnetdevip.h"
58
#include "virnetdevbridge.h"
59
#include "virnetdevopenvswitch.h"
60
#include "virnetdevtap.h"
61
#include "virnetdevvportprofile.h"
62
#include "virpci.h"
63
#include "virdbus.h"
64
#include "virfile.h"
65
#include "virstring.h"
66
#include "viraccessapicheck.h"
67
#include "network_event.h"
68
#include "virhook.h"
69
#include "virjson.h"
70

71
#define VIR_FROM_THIS VIR_FROM_NETWORK
72
#define MAX_BRIDGE_ID 256
73

74 75 76 77 78 79 80
/**
 * 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)

81 82
#define SYSCTL_PATH "/proc/sys"

83 84
VIR_LOG_INIT("network.bridge_driver");

85
static virNetworkDriverStatePtr network_driver;
86

87

88 89 90 91 92 93 94 95
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;
}
96

97 98 99

static void
networkDriverLock(virNetworkDriverStatePtr driver)
100
{
101
    virMutexLock(&driver->lock);
102
}
103 104 105 106


static void
networkDriverUnlock(virNetworkDriverStatePtr driver)
107
{
108
    virMutexUnlock(&driver->lock);
109 110
}

111

112 113 114 115
static dnsmasqCapsPtr
networkGetDnsmasqCaps(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr ret;
116
    networkDriverLock(driver);
117
    ret = virObjectRef(driver->dnsmasqCaps);
118
    networkDriverUnlock(driver);
119 120 121
    return ret;
}

122

123 124 125 126 127 128 129 130
static int
networkDnsmasqCapsRefresh(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr caps;

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

131
    networkDriverLock(driver);
132 133
    virObjectUnref(driver->dnsmasqCaps);
    driver->dnsmasqCaps = caps;
134
    networkDriverUnlock(driver);
135 136 137
    return 0;
}

138

139 140 141 142 143
static int
networkStateCleanup(void);

static int
networkStartNetwork(virNetworkDriverStatePtr driver,
144
                    virNetworkObjPtr obj);
145 146 147

static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
148
                       virNetworkObjPtr obj);
149 150 151

static int
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
152
                           virNetworkObjPtr obj);
153 154 155

static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
156
                              virNetworkObjPtr obj);
157

158
static int
159
networkStartNetworkExternal(virNetworkObjPtr obj);
160

161
static int
162
networkShutdownNetworkExternal(virNetworkObjPtr obj);
163

164
static void
165
networkReloadFirewallRules(virNetworkDriverStatePtr driver, bool startup);
166

167 168
static void
networkRefreshDaemons(virNetworkDriverStatePtr driver);
169

170
static int
171
networkPlugBandwidth(virNetworkObjPtr obj,
172
                     virDomainNetDefPtr iface);
173

174
static int
175
networkUnplugBandwidth(virNetworkObjPtr obj,
176
                       virDomainNetDefPtr iface);
177

178
static void
179
networkNetworkObjTaint(virNetworkObjPtr obj,
180
                       virNetworkTaintFlags taint);
181

182

183 184 185
static virNetworkObjPtr
networkObjFromNetwork(virNetworkPtr net)
{
186
    virNetworkDriverStatePtr driver = networkGetDriver();
187
    virNetworkObjPtr obj;
188 189
    char uuidstr[VIR_UUID_STRING_BUFLEN];

190 191
    obj = virNetworkObjFindByUUID(driver->networks, net->uuid);
    if (!obj) {
192 193 194 195 196 197
        virUUIDFormat(net->uuid, uuidstr);
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching uuid '%s' (%s)"),
                       uuidstr, net->name);
    }

198
    return obj;
199 200
}

201

202
static int
203
networkRunHook(virNetworkObjPtr obj,
204
               virDomainDefPtr dom,
205
               virDomainNetDefPtr iface,
206 207 208
               int op,
               int sub_op)
{
209
    virNetworkDefPtr def;
210 211 212 213 214 215
    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)) {
216 217
        if (!obj) {
            VIR_DEBUG("Not running hook as @obj is NULL");
218 219 220
            ret = 0;
            goto cleanup;
        }
221
        def = virNetworkObjGetDef(obj);
222

223 224
        virBufferAddLit(&buf, "<hookData>\n");
        virBufferAdjustIndent(&buf, 2);
225
        if (iface && virDomainNetDefFormat(&buf, iface, NULL, 0) < 0)
226
            goto cleanup;
227
        if (virNetworkDefFormatBuf(&buf, def, 0) < 0)
228
            goto cleanup;
229
        if (dom && virDomainDefFormatInternal(dom, NULL, 0, &buf, NULL) < 0)
230 231 232 233 234
            goto cleanup;

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

235
        if (virBufferCheckError(&buf) < 0)
236 237
            goto cleanup;

238
        xml = virBufferContentAndReset(&buf);
239
        hookret = virHookCall(VIR_HOOK_DRIVER_NETWORK, def->name,
240 241 242 243 244 245 246
                              op, sub_op, NULL, xml, NULL);

        /*
         * If the script raised an error, pass it to the callee.
         */
        if (hookret < 0)
            goto cleanup;
247

248
        networkNetworkObjTaint(obj, VIR_NETWORK_TAINT_HOOK);
249 250 251
    }

    ret = 0;
252
 cleanup:
253 254 255 256 257 258 259
    virBufferFreeAndReset(&buf);
    VIR_FREE(xml);
    VIR_FREE(net_xml);
    VIR_FREE(dom_xml);
    return ret;
}

260

261
static char *
262 263
networkDnsmasqLeaseFileNameDefault(virNetworkDriverStatePtr driver,
                                   const char *netname)
264 265 266
{
    char *leasefile;

267
    ignore_value(virAsprintf(&leasefile, "%s/%s.leases",
268
                             driver->dnsmasqStateDir, netname));
269 270 271
    return leasefile;
}

272

273
static char *
274 275
networkDnsmasqLeaseFileNameCustom(virNetworkDriverStatePtr driver,
                                  const char *bridge)
276 277 278 279
{
    char *leasefile;

    ignore_value(virAsprintf(&leasefile, "%s/%s.status",
280
                             driver->dnsmasqStateDir, bridge));
281 282 283
    return leasefile;
}

284

285
static char *
286 287
networkDnsmasqConfigFileName(virNetworkDriverStatePtr driver,
                             const char *netname)
288 289 290
{
    char *conffile;

291
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
292
                             driver->dnsmasqStateDir, netname));
293 294 295
    return conffile;
}

296

297 298 299 300 301 302
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

303
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
304 305 306
    return pidfilebase;
}

307

308
static char *
309 310
networkRadvdConfigFileName(virNetworkDriverStatePtr driver,
                           const char *netname)
311 312 313
{
    char *configfile;

314
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
315
                             driver->radvdStateDir, netname));
316 317
    return configfile;
}
318

319

320 321
/* do needed cleanup steps and remove the network from the list */
static int
322
networkRemoveInactive(virNetworkDriverStatePtr driver,
323
                      virNetworkObjPtr obj)
324 325
{
    char *leasefile = NULL;
326
    char *customleasefile = NULL;
327
    char *radvdconfigfile = NULL;
328
    char *configfile = NULL;
329
    char *radvdpidbase = NULL;
330
    char *statusfile = NULL;
M
Michal Privoznik 已提交
331
    char *macMapFile = NULL;
332
    dnsmasqContext *dctx = NULL;
333
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(obj);
334 335 336 337

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
338
    if (!(dctx = dnsmasqContextNew(def->name,
339
                                   driver->dnsmasqStateDir))) {
340
        goto cleanup;
341
    }
342

343
    if (!(leasefile = networkDnsmasqLeaseFileNameDefault(driver, def->name)))
344 345
        goto cleanup;

346
    if (!(customleasefile = networkDnsmasqLeaseFileNameCustom(driver, def->bridge)))
347 348
        goto cleanup;

349
    if (!(radvdconfigfile = networkRadvdConfigFileName(driver, def->name)))
350
        goto cleanup;
351 352

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
353
        goto cleanup;
354

355
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
356
        goto cleanup;
357

358
    if (!(statusfile = virNetworkConfigFile(driver->stateDir, def->name)))
359
        goto cleanup;
360

361
    if (!(macMapFile = virMacMapFileName(driver->dnsmasqStateDir, def->bridge)))
M
Michal Privoznik 已提交
362 363
        goto cleanup;

364 365 366
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
367
    unlink(customleasefile);
368
    unlink(configfile);
369

M
Michal Privoznik 已提交
370 371 372
    /* MAC map manager */
    unlink(macMapFile);

373 374
    /* radvd */
    unlink(radvdconfigfile);
375
    virPidFileDelete(driver->pidDir, radvdpidbase);
376

377 378 379
    /* remove status file */
    unlink(statusfile);

380
    /* remove the network definition */
381
    virNetworkObjRemoveInactive(driver->networks, obj);
382 383 384

    ret = 0;

385
 cleanup:
386
    VIR_FREE(leasefile);
387
    VIR_FREE(configfile);
388
    VIR_FREE(customleasefile);
389 390
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
391
    VIR_FREE(statusfile);
M
Michal Privoznik 已提交
392
    VIR_FREE(macMapFile);
393 394 395 396
    dnsmasqContextFree(dctx);
    return ret;
}

397

398 399 400
static char *
networkBridgeDummyNicName(const char *brname)
{
401
    static const char dummyNicSuffix[] = "-nic";
402 403
    char *nicname;

404 405 406 407 408 409 410
    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.
         */
411 412 413 414 415
        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));
416
    } else {
417
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
418
    }
419 420 421
    return nicname;
}

422

423 424
static int
networkUpdateState(virNetworkObjPtr obj,
425
                   void *opaque)
426
{
427
    virNetworkDefPtr def;
428
    virNetworkDriverStatePtr driver = opaque;
429
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
430
    virMacMapPtr macmap;
431
    char *macMapFile = NULL;
432
    int ret = -1;
433

434
    virObjectLock(obj);
435
    if (!virNetworkObjIsActive(obj)) {
436 437
        ret = 0;
        goto cleanup;
438
    }
439
    def = virNetworkObjGetDef(obj);
440

441
    switch ((virNetworkForwardType) def->forward.type) {
442 443 444
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
445
    case VIR_NETWORK_FORWARD_OPEN:
446
        /* If bridge doesn't exist, then mark it inactive */
447
        if (!(def->bridge && virNetDevExists(def->bridge) == 1))
448
            virNetworkObjSetActive(obj, false);
449

450
        if (!(macMapFile = virMacMapFileName(driver->dnsmasqStateDir,
451
                                             def->bridge)))
452 453
            goto cleanup;

454
        if (!(macmap = virMacMapNew(macMapFile)))
455 456
            goto cleanup;

457 458
        virNetworkObjSetMacMap(obj, macmap);

459
        break;
460

461
    case VIR_NETWORK_FORWARD_BRIDGE:
462 463
        if (def->bridge) {
            if (virNetDevExists(def->bridge) != 1)
464
                virNetworkObjSetActive(obj, false);
465
            break;
466
        }
467 468 469 470 471 472 473 474 475
        /* 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;
476

477 478 479
    case VIR_NETWORK_FORWARD_HOSTDEV:
        /* so far no extra checks */
        break;
480 481 482 483 484

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        goto cleanup;
485
    }
486

487
    /* Try and read dnsmasq/radvd pids of active networks */
488
    if (virNetworkObjIsActive(obj) && def->ips && (def->nips > 0)) {
489 490
        pid_t radvdPid;
        pid_t dnsmasqPid;
491
        char *radvdpidbase;
492

493
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
494
                                           def->name,
495
                                           &dnsmasqPid,
496
                                           dnsmasqCapsGetBinaryPath(dnsmasq_caps)));
497
        virNetworkObjSetDnsmasqPid(obj, dnsmasqPid);
M
Michal Privoznik 已提交
498

499
        radvdpidbase = networkRadvdPidfileBasename(def->name);
500 501
        if (!radvdpidbase)
            goto cleanup;
502

503 504
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           radvdpidbase,
505 506
                                           &radvdPid, RADVD));
        virNetworkObjSetRadvdPid(obj, radvdPid);
507
        VIR_FREE(radvdpidbase);
508
    }
509

510 511
    ret = 0;
 cleanup:
512
    virObjectUnlock(obj);
513
    virObjectUnref(dnsmasq_caps);
514
    VIR_FREE(macMapFile);
515 516
    return ret;
}
517

518

519
static int
520
networkAutostartConfig(virNetworkObjPtr obj,
521
                       void *opaque)
522
{
523
    virNetworkDriverStatePtr driver = opaque;
524
    int ret = -1;
525

526
    virObjectLock(obj);
527
    if (virNetworkObjIsAutostart(obj) &&
528 529
        !virNetworkObjIsActive(obj) &&
        networkStartNetwork(driver, obj) < 0)
530 531 532 533
        goto cleanup;

    ret = 0;
 cleanup:
534
    virObjectUnlock(obj);
535
    return ret;
536 537
}

538

539 540 541
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
542 543
                             DBusMessage *message,
                             void *user_data)
544
{
545 546
    virNetworkDriverStatePtr driver = user_data;

547 548 549 550 551 552
    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.");
553
        networkReloadFirewallRules(driver, false);
554 555 556 557 558 559
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

560

561
/**
562
 * networkStateInitialize:
563
 *
J
Ján Tomko 已提交
564
 * Initialization function for the QEMU daemon
565 566
 */
static int
567 568 569
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
570
{
571 572 573
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
574 575 576
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
577

578
    if (VIR_ALLOC(network_driver) < 0)
579
        goto error;
580

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

586 587
    network_driver->privileged = privileged;

588 589 590 591
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
592
    if (privileged) {
593
        if (VIR_STRDUP(network_driver->networkConfigDir,
594
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
595
            VIR_STRDUP(network_driver->networkAutostartDir,
596
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
597
            VIR_STRDUP(network_driver->stateDir,
598
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
599
            VIR_STRDUP(network_driver->pidDir,
600
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
601
            VIR_STRDUP(network_driver->dnsmasqStateDir,
602
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
603
            VIR_STRDUP(network_driver->radvdStateDir,
604 605
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
606
    } else {
607 608 609
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
610
            goto error;
611

612
        if ((virAsprintf(&network_driver->networkConfigDir,
613
                         "%s/qemu/networks", configdir) < 0) ||
614
            (virAsprintf(&network_driver->networkAutostartDir,
615
                         "%s/qemu/networks/autostart", configdir) < 0) ||
616
            (virAsprintf(&network_driver->stateDir,
617
                         "%s/network/lib", rundir) < 0) ||
618
            (virAsprintf(&network_driver->pidDir,
619
                         "%s/network/run", rundir) < 0) ||
620
            (virAsprintf(&network_driver->dnsmasqStateDir,
621
                         "%s/dnsmasq/lib", rundir) < 0) ||
622
            (virAsprintf(&network_driver->radvdStateDir,
623
                         "%s/radvd/lib", rundir) < 0)) {
624
            goto error;
625
        }
626 627
    }

628
    if (virFileMakePath(network_driver->stateDir) < 0) {
629 630
        virReportSystemError(errno,
                             _("cannot create directory %s"),
631
                             network_driver->stateDir);
632 633 634
        goto error;
    }

635
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
636
    network_driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
637

638
    if (!(network_driver->networks = virNetworkObjListNew()))
639 640
        goto error;

641 642
    if (virNetworkObjLoadAllState(network_driver->networks,
                                  network_driver->stateDir) < 0)
643 644
        goto error;

645 646 647
    if (virNetworkObjLoadAllConfigs(network_driver->networks,
                                    network_driver->networkConfigDir,
                                    network_driver->networkAutostartDir) < 0)
648 649
        goto error;

650 651 652 653
    /* 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). */
654
    virNetworkObjListForEach(network_driver->networks,
655
                             networkUpdateState,
656 657
                             network_driver);
    virNetworkObjListPrune(network_driver->networks,
658 659
                           VIR_CONNECT_LIST_NETWORKS_INACTIVE |
                           VIR_CONNECT_LIST_NETWORKS_TRANSIENT);
660
    networkReloadFirewallRules(network_driver, true);
661
    networkRefreshDaemons(network_driver);
662

663
    network_driver->networkEventState = virObjectEventStateNew();
664

665 666 667
#ifdef HAVE_FIREWALLD
    if (!(sysbus = virDBusGetSystemBus())) {
        VIR_WARN("DBus not available, disabling firewalld support "
668
                 "in bridge_network_driver: %s", virGetLastErrorMessage());
669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685
    } 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,
686
                                   network_driver, NULL);
687 688 689
    }
#endif

690
    ret = 0;
691
 cleanup:
692 693 694
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
695

696
 error:
697
    networkStateCleanup();
698
    goto cleanup;
699 700
}

701

702 703 704 705 706 707 708 709
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
710
    if (!network_driver)
711 712
        return;

713
    virNetworkObjListForEach(network_driver->networks,
714
                             networkAutostartConfig,
715
                             network_driver);
716 717
}

718

719
/**
720
 * networkStateReload:
721
 *
J
Ján Tomko 已提交
722
 * Function to restart the QEMU daemon, it will recheck the configuration
723 724 725
 * files and update its state and the networking
 */
static int
726 727
networkStateReload(void)
{
728
    if (!network_driver)
729 730
        return 0;

731 732 733 734 735
    virNetworkObjLoadAllState(network_driver->networks,
                              network_driver->stateDir);
    virNetworkObjLoadAllConfigs(network_driver->networks,
                                network_driver->networkConfigDir,
                                network_driver->networkAutostartDir);
736
    networkReloadFirewallRules(network_driver, false);
737 738
    networkRefreshDaemons(network_driver);
    virNetworkObjListForEach(network_driver->networks,
739
                             networkAutostartConfig,
740
                             network_driver);
741 742 743 744 745
    return 0;
}


/**
746
 * networkStateCleanup:
747
 *
J
Ján Tomko 已提交
748
 * Shutdown the QEMU daemon, it will stop all active domains and networks
749 750
 */
static int
751 752
networkStateCleanup(void)
{
753
    if (!network_driver)
754 755
        return -1;

756
    virObjectUnref(network_driver->networkEventState);
757

758
    /* free inactive networks */
759
    virObjectUnref(network_driver->networks);
760

761 762 763 764 765 766
    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);
767

768
    virObjectUnref(network_driver->dnsmasqCaps);
769

770
    virMutexDestroy(&network_driver->lock);
771

772
    VIR_FREE(network_driver);
773 774 775 776 777

    return 0;
}


778 779 780 781 782 783 784 785
static virDrvOpenStatus
networkConnectOpen(virConnectPtr conn,
                   virConnectAuthPtr auth ATTRIBUTE_UNUSED,
                   virConfPtr conf ATTRIBUTE_UNUSED,
                   unsigned int flags)
{
    virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);

786 787 788 789 790 791 792 793 794 795 796
    if (network_driver == NULL) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("network state driver is not active"));
        return VIR_DRV_OPEN_ERROR;
    }

    if (network_driver->privileged) {
        if (STRNEQ(conn->uri->path, "/system")) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unexpected network URI path '%s', try network:///system"),
                           conn->uri->path);
797 798
            return VIR_DRV_OPEN_ERROR;
        }
799 800 801 802 803 804
    } else {
        if (STRNEQ(conn->uri->path, "/session")) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unexpected network URI path '%s', try network:///session"),
                           conn->uri->path);
            return VIR_DRV_OPEN_ERROR;
805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839
        }
    }

    if (virConnectOpenEnsureACL(conn) < 0)
        return VIR_DRV_OPEN_ERROR;

    return VIR_DRV_OPEN_SUCCESS;
}

static int networkConnectClose(virConnectPtr conn ATTRIBUTE_UNUSED)
{
    return 0;
}


static int networkConnectIsSecure(virConnectPtr conn ATTRIBUTE_UNUSED)
{
    /* Trivially secure, since always inside the daemon */
    return 1;
}


static int networkConnectIsEncrypted(virConnectPtr conn ATTRIBUTE_UNUSED)
{
    /* Not encrypted, but remote driver takes care of that */
    return 0;
}


static int networkConnectIsAlive(virConnectPtr conn ATTRIBUTE_UNUSED)
{
    return 1;
}


840 841 842 843 844
/* networkKillDaemon:
 *
 * kill the specified pid/name, and wait a bit to make sure it's dead.
 */
static int
845 846 847
networkKillDaemon(pid_t pid,
                  const char *daemonName,
                  const char *networkName)
848
{
849 850
    size_t i;
    int ret = -1;
851 852 853 854 855 856 857
    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.
     */
858
    for (i = 0; i < 25; i++) {
859
        int signum = 0;
860
        if (i == 0) {
861
            signum = SIGTERM;
862
        } else if (i == 15) {
863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897
            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);
898
 cleanup:
899 900 901
    return ret;
}

902

J
Ján Tomko 已提交
903 904 905
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
906
static int
G
Gene Czarcinski 已提交
907
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
908
                                 virNetworkIPDefPtr ipdef)
909
{
910
    size_t i;
G
Gene Czarcinski 已提交
911
    bool ipv6 = false;
912

G
Gene Czarcinski 已提交
913 914
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
915 916
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
917
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
918 919
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
920
                return -1;
921
    }
922

G
Gene Czarcinski 已提交
923 924 925
    return 0;
}

926

G
Gene Czarcinski 已提交
927 928 929 930
static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
931
    size_t i, j;
G
Gene Czarcinski 已提交
932

933 934
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
935
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
936
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
937
                for (j = 0; j < host->nnames; j++)
938 939
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
940 941
            }
        }
942 943
    }

944
    return 0;
945 946 947
}


948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984
static int
networkDnsmasqConfLocalPTRs(virBufferPtr buf,
                            virNetworkDefPtr def)
{
    virNetworkIPDefPtr ip;
    size_t i;
    char *ptr = NULL;
    int rc;

    for (i = 0; i < def->nips; i++) {
        ip = def->ips + i;

        if (ip->localPTR != VIR_TRISTATE_BOOL_YES)
            continue;

        if ((rc = virSocketAddrPTRDomain(&ip->address,
                                         virNetworkIPDefPrefix(ip),
                                         &ptr)) < 0) {
            if (rc == -2) {
                int family = VIR_SOCKET_ADDR_FAMILY(&ip->address);
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("PTR domain for %s network with prefix %u "
                                 "cannot be automatically created"),
                               (family == AF_INET) ? "IPv4" : "IPv6",
                               virNetworkIPDefPrefix(ip));
            }
            return -1;
        }

        virBufferAsprintf(buf, "local=/%s/\n", ptr);
        VIR_FREE(ptr);
    }

    return 0;
}


985
int
986
networkDnsmasqConfContents(virNetworkObjPtr obj,
987 988 989 990
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
991
{
992
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
993
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
994
    int r, ret = -1;
995
    int nbleases = 0;
996
    size_t i;
997
    virNetworkDNSDefPtr dns = &def->dns;
998
    bool wantDNS = dns->enable != VIR_TRISTATE_BOOL_NO;
999
    virNetworkIPDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
G
Gene Czarcinski 已提交
1000
    bool ipv6SLAAC;
1001
    char *saddr = NULL, *eaddr = NULL;
1002

1003 1004
    *configstr = NULL;

1005
    /*
1006 1007 1008
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
1009
     *
1010 1011
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
1012
     */
1013 1014 1015 1016 1017 1018

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

1019
    /* create dnsmasq config file appropriate for this network */
1020 1021

    /* Don't forget to update networkxml2conftest :-) */
1022
    virBufferAsprintf(&configbuf,
1023 1024 1025 1026 1027 1028 1029
                      "##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"
1030
                      "strict-order\n",
1031
                      def->name);
1032

1033 1034 1035 1036 1037 1038
    /* if dns is disabled, set its listening port to 0, which
     * tells dnsmasq to not listen
     */
    if (!wantDNS)
        virBufferAddLit(&configbuf, "port=0\n");

1039
    if (wantDNS && def->dns.forwarders) {
1040 1041 1042 1043 1044 1045 1046 1047 1048
        /* addNoResolv should be set to true if there are any entries
         * that specify an IP address for requests, but no domain
         * qualifier (implying that all requests otherwise "unclaimed"
         * should be sent to that address). if it is still false when
         * we've looked at all entries, it means we still need the
         * host's resolv.conf for some cases.
         */
        bool addNoResolv = false;

1049 1050
        for (i = 0; i < def->dns.nfwds; i++) {
            virNetworkDNSForwarderPtr fwd = &def->dns.forwarders[i];
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060

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

                if (!addr)
                    goto cleanup;
                virBufferAsprintf(&configbuf, "%s\n", addr);
1061
                VIR_FREE(addr);
1062 1063
                if (!fwd->domain)
                    addNoResolv = true;
1064 1065 1066 1067
            } else {
                /* "don't forward requests for this domain" */
                virBufferAddLit(&configbuf, "#\n");
            }
1068
        }
1069 1070
        if (addNoResolv)
            virBufferAddLit(&configbuf, "no-resolv\n");
1071 1072
    }

1073 1074
    if (def->domain) {
        if (def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
1075 1076
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
1077
                              def->domain);
1078
        }
1079
        virBufferAsprintf(&configbuf,
1080 1081
                          "domain=%s\n"
                          "expand-hosts\n",
1082
                          def->domain);
1083
    }
1084

1085
    if (wantDNS &&
1086
        networkDnsmasqConfLocalPTRs(&configbuf, def) < 0)
1087 1088
        goto cleanup;

1089
    if (wantDNS && def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
1090 1091 1092 1093
        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)
1094
         */
1095
        virBufferAddLit(&configbuf, "local=//\n");
1096
    }
1097

1098
    if (pidfile)
1099
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
1100

1101
    /* dnsmasq will *always* listen on localhost unless told otherwise */
P
Pavel Timofeev 已提交
1102
#ifdef __linux__
1103
    virBufferAddLit(&configbuf, "except-interface=lo\n");
P
Pavel Timofeev 已提交
1104 1105 1106 1107
#else
    /* BSD family OSes and Solaris call loopback interface as lo0 */
    virBufferAddLit(&configbuf, "except-interface=lo0\n");
#endif
1108

1109 1110 1111 1112 1113 1114 1115 1116
    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.
         */
1117
        virBufferAsprintf(&configbuf,
1118 1119
                          "bind-dynamic\n"
                          "interface=%s\n",
1120
                          def->bridge);
1121
    } else {
1122
        virBufferAddLit(&configbuf, "bind-interfaces\n");
1123 1124 1125 1126
        /*
         * --interface does not actually work with dnsmasq < 2.47,
         * due to DAD for ipv6 addresses on the interface.
         *
1127
         * virCommandAddArgList(cmd, "--interface", def->bridge, NULL);
1128 1129 1130
         *
         * So listen on all defined IPv[46] addresses
         */
1131
        for (i = 0;
1132
             (tmpipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
1133
             i++) {
1134 1135 1136 1137
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
1138

1139
            /* also part of CVE 2012-3411 - if the host's version of
1140
             * dnsmasq doesn't have bind-dynamic, only allow listening on
1141 1142
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
1143 1144
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
1145 1146 1147 1148
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
1149
                                 "The version of dnsmasq on this host (%d.%d) "
1150 1151 1152 1153
                                 "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 "
1154 1155 1156 1157 1158 1159
                                 "(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);
1160
                VIR_FREE(ipaddr);
1161 1162
                goto cleanup;
            }
1163
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1164 1165 1166
            VIR_FREE(ipaddr);
        }
    }
1167

1168 1169
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1170
     * guaranteed to not work, and set no-resolv so that no dns
1171 1172 1173
     * 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).
1174 1175 1176
     * IPv6 RA always contains an implicit default route
     * via the sender's link-local address. The only thing we can do
     * is set the lifetime of this route to 0, i.e. disable it.
1177
     */
1178
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1179
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1180
                        "no-resolv\n");
1181 1182 1183 1184
        if (dnsmasqCapsGet(caps, DNSMASQ_CAPS_RA_PARAM)) {
            /* interface=* (any), interval=0 (default), lifetime=0 (seconds) */
            virBufferAddLit(&configbuf, "ra-param=*,0,0\n");
        }
1185
    }
1186

1187 1188 1189 1190 1191
    if (wantDNS) {
        for (i = 0; i < dns->ntxts; i++) {
            virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
                              dns->txts[i].name,
                              dns->txts[i].value);
1192
        }
1193

1194 1195 1196 1197 1198 1199 1200 1201
        for (i = 0; i < dns->nsrvs; i++) {
            /* service/protocol are required, and should have been validated
             * by the parser.
             */
            if (!dns->srvs[i].service) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Missing required 'service' "
                                 "attribute in SRV record of network '%s'"),
1202
                               def->name);
1203 1204 1205 1206 1207 1208
                goto cleanup;
            }
            if (!dns->srvs[i].protocol) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Missing required 'service' "
                                 "attribute in SRV record of network '%s'"),
1209
                               def->name);
1210 1211 1212 1213 1214 1215 1216
                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);
1217

1218 1219 1220 1221 1222 1223 1224
            /* domain is optional - it defaults to the domain of this network */
            if (dns->srvs[i].domain)
                virBufferAsprintf(&configbuf, ".%s", dns->srvs[i].domain);

            /* If target is empty or ".", that means "the service is
             * decidedly not available at this domain" (RFC2782). In that
             * case, any port, priority, or weight is irrelevant.
1225
             */
1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245
            if (dns->srvs[i].target && STRNEQ(dns->srvs[i].target, ".")) {

                virBufferAsprintf(&configbuf, ",%s", dns->srvs[i].target);
                /* port, priority, and weight are optional, but are
                 * identified by their position in the line. If an item is
                 * unspecified, but something later in the line *is*
                 * specified, we need to give the default value for the
                 * unspecified item. (According to the dnsmasq manpage,
                 * the default for port is 1).
                 */
                if (dns->srvs[i].port ||
                    dns->srvs[i].priority || dns->srvs[i].weight)
                    virBufferAsprintf(&configbuf, ",%d",
                                      dns->srvs[i].port ? dns->srvs[i].port : 1);
                if (dns->srvs[i].priority || dns->srvs[i].weight)
                    virBufferAsprintf(&configbuf, ",%d", dns->srvs[i].priority);
                if (dns->srvs[i].weight)
                    virBufferAsprintf(&configbuf, ",%d", dns->srvs[i].weight);
            }
            virBufferAddLit(&configbuf, "\n");
1246
        }
1247 1248
    }

G
Gene Czarcinski 已提交
1249
    /* Find the first dhcp for both IPv4 and IPv6 */
1250
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
1251
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
1252
         i++) {
G
Gene Czarcinski 已提交
1253 1254 1255 1256
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1257 1258
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269
                    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,
1270 1271 1272 1273 1274 1275 1276 1277 1278
                                   _("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
1279 1280 1281 1282
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1283 1284
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297
                    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 "
1298 1299 1300 1301
                 "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
1302 1303 1304 1305 1306
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1307 1308
        int prefix;

1309
        prefix = virNetworkIPDefPrefix(ipdef);
1310 1311 1312
        if (prefix < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
1313
                           def->bridge);
1314 1315
            goto cleanup;
        }
1316
        for (r = 0; r < ipdef->nranges; r++) {
1317 1318
            int thisRange;

1319 1320
            if (!(saddr = virSocketAddrFormat(&ipdef->ranges[r].start)) ||
                !(eaddr = virSocketAddrFormat(&ipdef->ranges[r].end)))
1321
                goto cleanup;
1322

1323
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s",
1324
                              saddr, eaddr);
1325
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1326
                virBufferAsprintf(&configbuf, ",%d", prefix);
1327 1328
            virBufferAddLit(&configbuf, "\n");

1329
            VIR_FREE(saddr);
1330
            VIR_FREE(eaddr);
1331
            thisRange = virSocketAddrGetRange(&ipdef->ranges[r].start,
1332 1333
                                              &ipdef->ranges[r].end,
                                              &ipdef->address,
1334
                                              virNetworkIPDefPrefix(ipdef));
1335 1336 1337
            if (thisRange < 0)
                goto cleanup;
            nbleases += thisRange;
1338
        }
1339

1340
        /*
1341 1342 1343 1344
         * 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)
1345 1346
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1347
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1348 1349
            if (!bridgeaddr)
                goto cleanup;
1350 1351 1352
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static",
                              bridgeaddr);
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1353
                virBufferAsprintf(&configbuf, ",%d", prefix);
1354
            virBufferAddLit(&configbuf, "\n");
1355 1356
            VIR_FREE(bridgeaddr);
        }
1357

G
Gene Czarcinski 已提交
1358 1359
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1360

G
Gene Czarcinski 已提交
1361 1362
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
1363
            if (ipdef->nranges || ipdef->nhosts) {
1364
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1365 1366
                virBufferAddLit(&configbuf, "dhcp-authoritative\n");
            }
1367

G
Gene Czarcinski 已提交
1368
            if (ipdef->tftproot) {
1369 1370
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1371
            }
1372

G
Gene Czarcinski 已提交
1373 1374 1375
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1376

1377
                    if (!bootserver)
G
Gene Czarcinski 已提交
1378
                        goto cleanup;
1379
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1380
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1381 1382
                    VIR_FREE(bootserver);
                } else {
1383
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1384 1385 1386 1387 1388
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1389

1390
    if (nbleases > 0)
1391
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1392

G
Gene Czarcinski 已提交
1393 1394
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1395
        goto cleanup;
G
Gene Czarcinski 已提交
1396 1397 1398 1399 1400 1401

    /* 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)
1402 1403
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1404

1405 1406
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1407
     */
1408 1409 1410 1411
    if (wantDNS) {
        virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
                          dctx->addnhostsfile->path);
    }
G
Gene Czarcinski 已提交
1412 1413 1414

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1415
        if (ipv6def) {
1416
            virBufferAddLit(&configbuf, "enable-ra\n");
1417
        } else {
1418
            for (i = 0;
1419
                 (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i));
1420
                 i++) {
G
Gene Czarcinski 已提交
1421 1422 1423 1424
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1425 1426
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1427 1428
                    VIR_FREE(bridgeaddr);
                }
1429
            }
1430
        }
1431 1432
    }

1433 1434 1435
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1436
    ret = 0;
G
Gene Czarcinski 已提交
1437

1438
 cleanup:
1439 1440
    VIR_FREE(saddr);
    VIR_FREE(eaddr);
1441
    virBufferFreeAndReset(&configbuf);
1442
    return ret;
1443 1444
}

1445

1446
/* build the dnsmasq command line */
1447 1448
static int ATTRIBUTE_NONNULL(3)
networkBuildDhcpDaemonCommandLine(virNetworkDriverStatePtr driver,
1449
                                  virNetworkObjPtr obj,
1450
                                  virCommandPtr *cmdout,
1451 1452
                                  char *pidfile,
                                  dnsmasqContext *dctx)
1453
{
1454
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1455
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1456
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1457
    int ret = -1;
1458 1459
    char *configfile = NULL;
    char *configstr = NULL;
1460
    char *leaseshelper_path = NULL;
1461

1462
    virNetworkObjSetDnsmasqPid(obj, -1);
1463

1464
    if (networkDnsmasqConfContents(obj, pidfile, &configstr,
1465
                                   dctx, dnsmasq_caps) < 0)
1466 1467 1468 1469 1470
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1471
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
1472 1473 1474 1475 1476
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1477 1478
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1479 1480 1481
        goto cleanup;
    }

1482 1483
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1484
                                                  abs_topbuilddir "/src",
1485 1486 1487
                                                  LIBEXECDIR)))
        goto cleanup;

1488
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(dnsmasq_caps));
1489
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1490 1491
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1492
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1493
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", def->bridge);
1494

1495
    *cmdout = cmd;
1496
    ret = 0;
1497
 cleanup:
1498
    virObjectUnref(dnsmasq_caps);
1499 1500
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1501
    VIR_FREE(leaseshelper_path);
1502 1503 1504
    return ret;
}

1505

1506
static int
1507
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
1508
                       virNetworkObjPtr obj)
1509
{
1510
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1511 1512 1513
    virNetworkIPDefPtr ipdef;
    size_t i;
    bool needDnsmasq = false;
1514 1515
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
1516
    pid_t dnsmasqPid;
1517
    int ret = -1;
1518
    dnsmasqContext *dctx = NULL;
1519

1520
    /* see if there are any IP addresses that need a dhcp server */
1521
    i = 0;
1522
    while ((ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i))) {
1523
        i++;
1524 1525 1526 1527
        if (ipdef->nranges || ipdef->nhosts)
            needDnsmasq = true;
    }

1528 1529 1530 1531 1532 1533
    if (i == 0) {
        /* no IP addresses at all, so we don't need to run */
        ret = 0;
        goto cleanup;
    }

1534
    if (!needDnsmasq && def->dns.enable == VIR_TRISTATE_BOOL_NO) {
1535
        /* no DHCP services needed, and user disabled DNS service */
1536 1537 1538 1539
        ret = 0;
        goto cleanup;
    }

1540
    if (virFileMakePath(driver->pidDir) < 0) {
1541
        virReportSystemError(errno,
1542
                             _("cannot create directory %s"),
1543
                             driver->pidDir);
1544
        goto cleanup;
1545 1546
    }

1547
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, def->name)))
1548
        goto cleanup;
1549

1550
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1551
        virReportSystemError(errno,
1552
                             _("cannot create directory %s"),
1553
                             driver->dnsmasqStateDir);
1554 1555 1556
        goto cleanup;
    }

1557
    dctx = dnsmasqContextNew(def->name, driver->dnsmasqStateDir);
1558 1559 1560
    if (dctx == NULL)
        goto cleanup;

1561
    if (networkDnsmasqCapsRefresh(driver) < 0)
1562
        goto cleanup;
1563

1564
    ret = networkBuildDhcpDaemonCommandLine(driver, obj, &cmd, pidfile, dctx);
1565 1566 1567 1568 1569
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1570
        goto cleanup;
1571

G
Guido Günther 已提交
1572
    ret = virCommandRun(cmd, NULL);
1573
    if (ret < 0)
1574 1575 1576
        goto cleanup;

    /*
1577 1578 1579 1580 1581
     * 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
1582 1583
     */

1584
    ret = virPidFileRead(driver->pidDir, def->name, &dnsmasqPid);
1585
    if (ret < 0)
1586
        goto cleanup;
1587
    virNetworkObjSetDnsmasqPid(obj, dnsmasqPid);
1588

1589
    ret = 0;
1590
 cleanup:
1591
    VIR_FREE(pidfile);
1592
    virCommandFree(cmd);
1593
    dnsmasqContextFree(dctx);
1594 1595 1596
    return ret;
}

1597

1598 1599
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1600 1601
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1602 1603 1604
 *
 *  Returns 0 on success, -1 on failure.
 */
1605
static int
1606
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
1607
                         virNetworkObjPtr obj)
1608
{
1609
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1610 1611
    int ret = -1;
    size_t i;
1612
    pid_t dnsmasqPid;
1613
    virNetworkIPDefPtr ipdef, ipv4def, ipv6def;
1614
    dnsmasqContext *dctx = NULL;
1615

G
Gene Czarcinski 已提交
1616
    /* if no IP addresses specified, nothing to do */
1617
    if (!virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1618 1619
        return 0;

1620
    /* if there's no running dnsmasq, just start it */
1621 1622
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid <= 0 || (kill(dnsmasqPid, 0) < 0))
1623
        return networkStartDhcpDaemon(driver, obj);
1624

1625 1626
    VIR_INFO("Refreshing dnsmasq for network %s", def->bridge);
    if (!(dctx = dnsmasqContextNew(def->name,
1627
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1628
        goto cleanup;
1629
    }
G
Gene Czarcinski 已提交
1630 1631 1632 1633 1634 1635

    /* 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;
1636
    for (i = 0;
1637
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
1638
         i++) {
G
Gene Czarcinski 已提交
1639 1640
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1641 1642
    }

G
Gene Czarcinski 已提交
1643
    ipv6def = NULL;
1644
    for (i = 0;
1645
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i));
1646
         i++) {
G
Gene Czarcinski 已提交
1647 1648
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1649 1650
    }

G
Gene Czarcinski 已提交
1651
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1652
        goto cleanup;
G
Gene Czarcinski 已提交
1653 1654

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1655
        goto cleanup;
1656

1657
    if (networkBuildDnsmasqHostsList(dctx, &def->dns) < 0)
J
Ján Tomko 已提交
1658
        goto cleanup;
1659 1660

    if ((ret = dnsmasqSave(dctx)) < 0)
1661
        goto cleanup;
1662

1663 1664
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    ret = kill(dnsmasqPid, SIGHUP);
1665
 cleanup:
1666 1667 1668 1669
    dnsmasqContextFree(dctx);
    return ret;
}

1670

1671 1672 1673 1674 1675 1676 1677 1678
/* 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
1679
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
1680
                         virNetworkObjPtr obj)
1681
{
1682
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1683 1684
    pid_t dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);

1685
    /* if there is a running dnsmasq, kill it */
1686
    if (dnsmasqPid > 0) {
1687
        networkKillDaemon(dnsmasqPid, "dnsmasq", def->name);
1688
        virNetworkObjSetDnsmasqPid(obj, -1);
1689
    }
1690
    /* now start dnsmasq if it should be started */
1691
    return networkStartDhcpDaemon(driver, obj);
1692 1693
}

1694

G
Gene Czarcinski 已提交
1695 1696 1697 1698 1699 1700
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";

1701
static int
1702
networkRadvdConfContents(virNetworkObjPtr obj,
1703
                         char **configstr)
1704
{
1705
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
E
Eric Blake 已提交
1706
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1707 1708
    int ret = -1;
    size_t i;
1709
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
1710
    bool v6present = false, dhcp6 = false;
1711 1712

    *configstr = NULL;
1713

G
Gene Czarcinski 已提交
1714
    /* Check if DHCPv6 is needed */
1715
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i)); i++) {
G
Gene Czarcinski 已提交
1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728
        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;
    }

1729 1730 1731
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1732
    virBufferAsprintf(&configbuf, "interface %s\n"
1733 1734
                      "{\n"
                      "  AdvSendAdvert on;\n"
1735
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1736 1737
                      "  AdvManagedFlag %s;\n"
                      "%s",
1738
                      def->bridge,
G
Gene Czarcinski 已提交
1739 1740
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1741 1742

    /* add a section for each IPv6 address in the config */
1743
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i)); i++) {
1744 1745 1746
        int prefix;
        char *netaddr;

1747
        prefix = virNetworkIPDefPrefix(ipdef);
1748
        if (prefix < 0) {
1749 1750
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
1751
                           def->bridge);
1752 1753
            goto cleanup;
        }
1754
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1755
            goto cleanup;
1756
        virBufferAsprintf(&configbuf,
1757
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1758 1759 1760
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1761 1762 1763
        VIR_FREE(netaddr);
    }

1764
    virBufferAddLit(&configbuf, "};\n");
1765

1766
    if (virBufferCheckError(&configbuf) < 0)
1767
        goto cleanup;
1768

1769 1770
    *configstr = virBufferContentAndReset(&configbuf);

1771
    ret = 0;
1772
 cleanup:
1773 1774 1775 1776
    virBufferFreeAndReset(&configbuf);
    return ret;
}

1777

1778
/* write file and return its name (which must be freed by caller) */
1779
static int
1780
networkRadvdConfWrite(virNetworkDriverStatePtr driver,
1781
                      virNetworkObjPtr obj,
1782
                      char **configFile)
1783
{
1784
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1785 1786 1787 1788 1789 1790 1791 1792 1793
    int ret = -1;
    char *configStr = NULL;
    char *myConfigFile = NULL;

    if (!configFile)
        configFile = &myConfigFile;

    *configFile = NULL;

1794
    if (networkRadvdConfContents(obj, &configStr) < 0)
1795 1796 1797 1798
        goto cleanup;

    if (!configStr) {
        ret = 0;
1799 1800 1801 1802
        goto cleanup;
    }

    /* construct the filename */
1803
    if (!(*configFile = networkRadvdConfigFileName(driver, def->name)))
1804 1805
        goto cleanup;
    /* write the file */
1806
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1807 1808
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1809 1810 1811 1812 1813
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1814
 cleanup:
1815 1816 1817 1818 1819
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

1820

1821
static int
1822
networkStartRadvd(virNetworkDriverStatePtr driver,
1823
                  virNetworkObjPtr obj)
1824
{
1825
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1826
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1827
    pid_t radvdPid;
1828 1829 1830 1831 1832 1833
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

1834
    virNetworkObjSetRadvdPid(obj, -1);
1835

G
Gene Czarcinski 已提交
1836
    /* Is dnsmasq handling RA? */
1837
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
G
Gene Czarcinski 已提交
1838 1839 1840 1841
        ret = 0;
        goto cleanup;
    }

1842
    if (!virNetworkDefGetIPByIndex(def, AF_INET6, 0)) {
1843 1844 1845 1846 1847 1848 1849 1850 1851 1852
        /* 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);
1853 1854 1855
        goto cleanup;
    }

1856
    if (virFileMakePath(driver->pidDir) < 0) {
1857 1858
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1859
                             driver->pidDir);
1860 1861
        goto cleanup;
    }
1862
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1863 1864
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1865
                             driver->radvdStateDir);
1866 1867 1868 1869
        goto cleanup;
    }

    /* construct pidfile name */
1870
    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
1871
        goto cleanup;
1872
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1873 1874
        goto cleanup;

1875
    if (networkRadvdConfWrite(driver, obj, &configfile) < 0)
1876 1877
        goto cleanup;

1878 1879 1880 1881
    /* 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
1882
     * virPidFileRead() below will fail if we use them).
1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897
     * 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;

1898
    if (virPidFileRead(driver->pidDir, radvdpidbase, &radvdPid) < 0)
1899
        goto cleanup;
1900
    virNetworkObjSetRadvdPid(obj, radvdPid);
1901 1902

    ret = 0;
1903
 cleanup:
1904
    virObjectUnref(dnsmasq_caps);
1905 1906 1907 1908 1909 1910 1911
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1912

1913
static int
1914
networkRefreshRadvd(virNetworkDriverStatePtr driver,
1915
                    virNetworkObjPtr obj)
1916
{
1917
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1918
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
G
Gene Czarcinski 已提交
1919
    char *radvdpidbase;
1920
    pid_t radvdPid;
G
Gene Czarcinski 已提交
1921 1922

    /* Is dnsmasq handling RA? */
1923 1924
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
        virObjectUnref(dnsmasq_caps);
1925 1926
        radvdPid = virNetworkObjGetRadvdPid(obj);
        if (radvdPid <= 0)
G
Gene Czarcinski 已提交
1927 1928
            return 0;
        /* radvd should not be running but in case it is */
1929 1930
        if ((networkKillDaemon(radvdPid, "radvd", def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(def->name))
G
Gene Czarcinski 已提交
1931
             != NULL)) {
1932
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1933 1934
            VIR_FREE(radvdpidbase);
        }
1935
        virNetworkObjSetRadvdPid(obj, -1);
G
Gene Czarcinski 已提交
1936 1937
        return 0;
    }
1938
    virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1939

1940
    /* if there's no running radvd, just start it */
1941 1942
    radvdPid = virNetworkObjGetRadvdPid(obj);
    if (radvdPid <= 0 || (kill(radvdPid, 0) < 0))
1943
        return networkStartRadvd(driver, obj);
1944

1945
    if (!virNetworkDefGetIPByIndex(def, AF_INET6, 0)) {
1946 1947 1948 1949
        /* no IPv6 addresses, so we don't need to run radvd */
        return 0;
    }

1950
    if (networkRadvdConfWrite(driver, obj, NULL) < 0)
1951 1952
        return -1;

1953
    return kill(radvdPid, SIGHUP);
1954 1955
}

1956

1957 1958
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1959
static int
1960
networkRestartRadvd(virNetworkObjPtr obj)
1961
{
1962
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1963
    char *radvdpidbase;
1964
    pid_t radvdPid = virNeworkObjGetRadvdPid(obj);
1965 1966

    /* if there is a running radvd, kill it */
1967
    if (radvdPid > 0) {
1968 1969 1970 1971
        /* 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).
         */
1972 1973
        if ((networkKillDaemon(radvdPid, "radvd", def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(def->name))
1974
             != NULL)) {
1975
            virPidFileDelete(driver->pidDir, radvdpidbase);
1976 1977
            VIR_FREE(radvdpidbase);
        }
1978
        virNetworkObjSetRadvdPid(obj, -1);
1979 1980
    }
    /* now start radvd if it should be started */
1981
    return networkStartRadvd(obj);
1982 1983 1984
}
#endif /* #if 0 */

1985

1986
static int
1987
networkRefreshDaemonsHelper(virNetworkObjPtr obj,
1988
                            void *opaque)
1989
{
1990
    virNetworkDefPtr def;
1991
    virNetworkDriverStatePtr driver = opaque;
1992

1993
    virObjectLock(obj);
1994
    def = virNetworkObjGetDef(obj);
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
    if (virNetworkObjIsActive(obj)) {
        switch ((virNetworkForwardType) def->forward.type) {
        case VIR_NETWORK_FORWARD_NONE:
        case VIR_NETWORK_FORWARD_NAT:
        case VIR_NETWORK_FORWARD_ROUTE:
        case VIR_NETWORK_FORWARD_OPEN:
            /* Only the three L3 network types that are configured by
             * libvirt will have a dnsmasq or radvd daemon associated
             * with them.  Here we send a SIGHUP to an existing
             * dnsmasq and/or radvd, or restart them if they've
             * disappeared.
             */
            networkRefreshDhcpDaemon(driver, obj);
            networkRefreshRadvd(driver, obj);
            break;

        case VIR_NETWORK_FORWARD_BRIDGE:
        case VIR_NETWORK_FORWARD_PRIVATE:
        case VIR_NETWORK_FORWARD_VEPA:
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
        case VIR_NETWORK_FORWARD_HOSTDEV:
            break;

        case VIR_NETWORK_FORWARD_LAST:
        default:
            virReportEnumRangeError(virNetworkForwardType, def->forward.type);
            goto cleanup;
        }
2023
    }
2024 2025

 cleanup:
2026
    virObjectUnlock(obj);
2027 2028 2029
    return 0;
}

2030

2031 2032 2033 2034
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
2035
networkRefreshDaemons(virNetworkDriverStatePtr driver)
2036 2037
{
    VIR_INFO("Refreshing network daemons");
2038 2039
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
2040
                             driver);
2041
}
2042

2043

2044
static int
2045
networkReloadFirewallRulesHelper(virNetworkObjPtr obj,
2046 2047
                                 void *opaque ATTRIBUTE_UNUSED)
{
2048
    virNetworkDefPtr def;
2049

2050
    virObjectLock(obj);
2051
    def = virNetworkObjGetDef(obj);
2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077
    if (virNetworkObjIsActive(obj)) {
        switch ((virNetworkForwardType) def->forward.type) {
        case VIR_NETWORK_FORWARD_NONE:
        case VIR_NETWORK_FORWARD_NAT:
        case VIR_NETWORK_FORWARD_ROUTE:
            /* Only three of the L3 network types that are configured by
             * libvirt need to have iptables rules reloaded. The 4th L3
             * network type, forward='open', doesn't need this because it
             * has no iptables rules.
             */
            networkRemoveFirewallRules(def);
            ignore_value(networkAddFirewallRules(def));
            break;

        case VIR_NETWORK_FORWARD_OPEN:
        case VIR_NETWORK_FORWARD_BRIDGE:
        case VIR_NETWORK_FORWARD_PRIVATE:
        case VIR_NETWORK_FORWARD_VEPA:
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
        case VIR_NETWORK_FORWARD_HOSTDEV:
            break;

        case VIR_NETWORK_FORWARD_LAST:
        default:
            virReportEnumRangeError(virNetworkForwardType, def->forward.type);
            goto cleanup;
2078 2079
        }
    }
2080 2081

 cleanup:
2082
    virObjectUnlock(obj);
2083
    return 0;
2084 2085
}

2086

2087
static void
2088
networkReloadFirewallRules(virNetworkDriverStatePtr driver, bool startup)
2089
{
2090
    VIR_INFO("Reloading iptables rules");
2091 2092
    if (networkPreReloadFirewallRules(startup) < 0)
        return;
2093 2094 2095
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
2096
    networkPostReloadFirewallRules(startup);
2097 2098
}

2099

2100
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
2101
static int
2102 2103
networkEnableIPForwarding(bool enableIPv4,
                          bool enableIPv6)
2104
{
2105
    int ret = 0;
2106 2107 2108 2109
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
2110
                           &enabled, sizeof(enabled));
2111 2112
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
2113
                           &enabled, sizeof(enabled));
2114
#else
2115
    if (enableIPv4)
2116
        ret = virFileWriteStr(SYSCTL_PATH "/net/ipv4/ip_forward", "1\n", 0);
2117
    if (enableIPv6 && ret == 0)
2118 2119
        ret = virFileWriteStr(SYSCTL_PATH "/net/ipv6/conf/all/forwarding", "1\n", 0);

2120
#endif
2121
    return ret;
2122 2123
}

2124

2125
static int
2126
networkSetIPv6Sysctls(virNetworkObjPtr obj)
2127
{
2128
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2129 2130
    char *field = NULL;
    int ret = -1;
2131
    bool enableIPv6 = !!virNetworkDefGetIPByIndex(def, AF_INET6, 0);
2132

2133 2134 2135 2136 2137
    /* 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",
2138
                    def->bridge) < 0)
2139
       goto cleanup;
2140

2141 2142
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
2143
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
2144
                      def->bridge);
2145 2146 2147
        ret = 0;
        goto cleanup;
    }
2148

2149 2150 2151
    if (virFileWriteStr(field, enableIPv6 ? "0" : "1", 0) < 0) {
        virReportSystemError(errno,
                             _("cannot write to %s to enable/disable IPv6 "
2152
                               "on bridge %s"), field, def->bridge);
2153
        goto cleanup;
2154
    }
2155
    VIR_FREE(field);
2156

2157 2158
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
2159 2160 2161 2162 2163 2164
     */

    /* 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",
2165
                    def->bridge) < 0)
2166 2167
        goto cleanup;

2168
    if (virFileWriteStr(field, "0", 0) < 0) {
2169
        virReportSystemError(errno,
2170 2171 2172 2173 2174
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

2175 2176 2177 2178
    /* 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",
2179
                    def->bridge) < 0)
2180 2181
        goto cleanup;

2182
    if (virFileWriteStr(field, "0", 0) < 0) {
2183
        virReportSystemError(errno,
2184
                             _("cannot disable %s"), field);
2185 2186 2187 2188
        goto cleanup;
    }

    ret = 0;
2189
 cleanup:
2190 2191 2192 2193
    VIR_FREE(field);
    return ret;
}

2194

2195
/* add an IP address to a bridge */
2196
static int
2197
networkAddAddrToBridge(virNetworkObjPtr obj,
2198
                       virNetworkIPDefPtr ipdef)
2199
{
2200
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2201
    int prefix = virNetworkIPDefPrefix(ipdef);
2202 2203

    if (prefix < 0) {
2204 2205
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
2206
                       def->bridge);
2207 2208 2209
        return -1;
    }

2210
    if (virNetDevIPAddrAdd(def->bridge, &ipdef->address, NULL, prefix) < 0)
2211 2212 2213 2214 2215
        return -1;

    return 0;
}

2216 2217

static int
2218
networkStartHandleMACTableManagerMode(virNetworkObjPtr obj,
2219 2220
                                      const char *macTapIfName)
{
2221 2222
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
    const char *brname = def->bridge;
2223 2224

    if (brname &&
2225
        def->macTableManager == VIR_NETWORK_BRIDGE_MAC_TABLE_MANAGER_LIBVIRT) {
2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238
        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;
}


2239 2240
/* add an IP (static) route to a bridge */
static int
2241
networkAddRouteToBridge(virNetworkObjPtr obj,
2242
                        virNetDevIPRoutePtr routedef)
2243
{
2244
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2245 2246 2247 2248
    int prefix = virNetDevIPRouteGetPrefix(routedef);
    unsigned int metric = virNetDevIPRouteGetMetric(routedef);
    virSocketAddrPtr addr = virNetDevIPRouteGetAddress(routedef);
    virSocketAddrPtr gateway = virNetDevIPRouteGetGateway(routedef);
2249 2250 2251 2252 2253

    if (prefix < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' has an invalid netmask "
                         "or IP address in route definition"),
2254
                       def->name);
2255 2256 2257
        return -1;
    }

2258
    if (virNetDevIPRouteAdd(def->bridge, addr, prefix, gateway, metric) < 0)
2259
        return -1;
2260

2261 2262 2263
    return 0;
}

2264
static int
2265
networkWaitDadFinish(virNetworkObjPtr obj)
2266
{
2267
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2268
    virNetworkIPDefPtr ipdef;
2269 2270 2271 2272
    virSocketAddrPtr *addrs = NULL, addr = NULL;
    size_t naddrs = 0;
    int ret = -1;

2273
    VIR_DEBUG("Begin waiting for IPv6 DAD on network %s", def->name);
2274

2275
    while ((ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, naddrs))) {
2276 2277 2278 2279 2280
        addr = &ipdef->address;
        if (VIR_APPEND_ELEMENT_COPY(addrs, naddrs, addr) < 0)
            goto cleanup;
    }

2281
    ret = (naddrs == 0) ? 0 : virNetDevIPWaitDadFinish(addrs, naddrs);
2282 2283 2284 2285

 cleanup:
    VIR_FREE(addrs);
    VIR_DEBUG("Finished waiting for IPv6 DAD on network %s with status %d",
2286
              def->name, ret);
2287 2288 2289
    return ret;
}

2290

2291
static int
2292
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
2293
                           virNetworkObjPtr obj)
2294
{
2295
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2296
    size_t i;
2297
    bool v4present = false, v6present = false;
2298
    virErrorPtr save_err = NULL;
2299
    virNetworkIPDefPtr ipdef;
2300
    virNetDevIPRoutePtr routedef;
2301
    char *macTapIfName = NULL;
2302
    virMacMapPtr macmap;
M
Michal Privoznik 已提交
2303
    char *macMapFile = NULL;
2304
    int tapfd = -1;
2305
    pid_t dnsmasqPid;
2306

2307
    /* Check to see if any network IP collides with an existing route */
2308
    if (networkCheckRouteCollision(def) < 0)
2309 2310
        return -1;

2311
    /* Create and configure the bridge device */
2312
    if (!def->bridge) {
2313 2314 2315 2316 2317 2318 2319 2320 2321 2322
        /* 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"),
2323
                       def->name);
2324 2325
        return -1;
    }
2326
    if (virNetDevBridgeCreate(def->bridge) < 0)
2327 2328
        return -1;

2329
    if (def->mac_specified) {
2330 2331 2332 2333 2334 2335
        /* 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.
         */
2336
        macTapIfName = networkBridgeDummyNicName(def->bridge);
2337
        if (!macTapIfName)
2338
            goto err0;
2339
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2340 2341
        if (virNetDevTapCreateInBridgePort(def->bridge,
                                           &macTapIfName, &def->mac,
2342
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2343
                                           NULL, def->mtu, NULL,
2344 2345 2346
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2347 2348 2349 2350 2351
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2352
    if (!(macMapFile = virMacMapFileName(driver->dnsmasqStateDir,
2353
                                         def->bridge)) ||
2354
        !(macmap = virMacMapNew(macMapFile)))
M
Michal Privoznik 已提交
2355 2356
        goto err1;

2357
    virNetworkObjSetMacMap(obj, macmap);
2358
    macmap = NULL;
2359

2360
    /* Set bridge options */
2361 2362 2363 2364

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2365
    if (virNetDevBridgeSetSTPDelay(def->bridge, def->delay * 1000) < 0)
2366
        goto err1;
2367

2368
    if (virNetDevBridgeSetSTP(def->bridge, def->stp ? true : false) < 0)
2369
        goto err1;
2370

2371 2372 2373
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
2374
    if (networkSetIPv6Sysctls(obj) < 0)
2375
        goto err1;
2376

2377
    /* Add "once per network" rules */
2378 2379
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN &&
        networkAddFirewallRules(def) < 0)
2380 2381
        goto err1;

2382
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i)); i++) {
2383
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2384
            v4present = true;
2385
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2386
            v6present = true;
2387

2388
        /* Add the IP address/netmask to the bridge */
2389
        if (networkAddAddrToBridge(obj, ipdef) < 0)
2390
            goto err2;
2391 2392
    }

2393
    if (networkStartHandleMACTableManagerMode(obj, macTapIfName) < 0)
2394 2395
        goto err2;

2396
    /* Bring up the bridge interface */
2397
    if (virNetDevSetOnline(def->bridge, 1) < 0)
2398
        goto err2;
2399

2400
    for (i = 0; i < def->nroutes; i++) {
2401 2402
        virSocketAddrPtr gateway = NULL;

2403
        routedef = def->routes[i];
2404
        gateway = virNetDevIPRouteGetGateway(routedef);
2405

2406 2407 2408
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2409
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2410
            if (networkAddRouteToBridge(obj, routedef) < 0) {
2411 2412 2413 2414 2415 2416
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2417
    /* If forward.type != NONE, turn on global IP forwarding */
2418
    if (def->forward.type != VIR_NETWORK_FORWARD_NONE) {
2419
        if (v6present && !virNetDevIPCheckIPv6Forwarding())
2420 2421 2422 2423 2424 2425 2426
            goto err3; /* Precise error message already provided */

        if (networkEnableIPForwarding(v4present, v6present) < 0) {
            virReportSystemError(errno, "%s",
                                 _("failed to enable IP forwarding"));
            goto err3;
        }
2427 2428
    }

2429

2430
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2431
    if ((v4present || v6present) &&
2432
        networkStartDhcpDaemon(driver, obj) < 0)
2433
        goto err3;
2434

2435
    /* start radvd if there are any ipv6 addresses */
2436
    if (v6present && networkStartRadvd(driver, obj) < 0)
2437 2438
        goto err4;

2439 2440 2441
    /* dnsmasq does not wait for DAD to complete before daemonizing,
     * so we need to wait for it ourselves.
     */
2442
    if (v6present && networkWaitDadFinish(obj) < 0)
2443 2444 2445 2446
        goto err4;

    /* DAD has finished, dnsmasq is now bound to the
     * bridge's IPv6 address, so we can set the dummy tun down.
2447 2448 2449 2450 2451 2452 2453
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
            goto err4;
        VIR_FORCE_CLOSE(tapfd);
    }

2454
    if (virNetDevBandwidthSet(def->bridge, def->bandwidth, true, true) < 0)
2455 2456
        goto err5;

2457
    VIR_FREE(macTapIfName);
M
Michal Privoznik 已提交
2458
    VIR_FREE(macMapFile);
2459 2460 2461

    return 0;

2462
 err5:
2463 2464
    if (def->bandwidth)
       virNetDevBandwidthClear(def->bridge);
2465

2466 2467 2468 2469
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2470 2471 2472 2473
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid > 0) {
        kill(dnsmasqPid, SIGTERM);
        virNetworkObjSetDnsmasqPid(obj, -1);
2474 2475
    }

2476 2477 2478
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2479
    ignore_value(virNetDevSetOnline(def->bridge, 0));
2480

2481 2482 2483
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2484 2485
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(def);
2486 2487

 err1:
2488 2489 2490
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2491
    if (macTapIfName) {
2492
        VIR_FORCE_CLOSE(tapfd);
2493
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2494 2495
        VIR_FREE(macTapIfName);
    }
2496
    virNetworkObjUnrefMacMap(obj);
M
Michal Privoznik 已提交
2497
    VIR_FREE(macMapFile);
2498 2499

 err0:
2500 2501
    if (!save_err)
        save_err = virSaveLastError();
2502
    ignore_value(virNetDevBridgeDelete(def->bridge));
2503

2504 2505 2506 2507
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2508
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2509 2510 2511
    return -1;
}

2512

2513 2514
static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
2515
                              virNetworkObjPtr obj)
2516
{
2517
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2518 2519 2520
    pid_t radvdPid;
    pid_t dnsmasqPid;

2521 2522
    if (def->bandwidth)
        virNetDevBandwidthClear(def->bridge);
2523

2524
    virNetworkObjUnrefMacMap(obj);
M
Michal Privoznik 已提交
2525

2526 2527
    radvdPid = virNetworkObjGetRadvdPid(obj);
    if (radvdPid > 0) {
2528 2529
        char *radvdpidbase;

2530
        kill(radvdPid, SIGTERM);
2531
        /* attempt to delete the pidfile we created */
2532
        if ((radvdpidbase = networkRadvdPidfileBasename(def->name))) {
2533
            virPidFileDelete(driver->pidDir, radvdpidbase);
2534 2535 2536 2537
            VIR_FREE(radvdpidbase);
        }
    }

2538 2539 2540
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid > 0)
        kill(dnsmasqPid, SIGTERM);
2541

2542 2543
    if (def->mac_specified) {
        char *macTapIfName = networkBridgeDummyNicName(def->bridge);
2544
        if (macTapIfName) {
2545
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2546 2547 2548 2549
            VIR_FREE(macTapIfName);
        }
    }

2550
    ignore_value(virNetDevSetOnline(def->bridge, 0));
2551

2552 2553
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(def);
2554

2555
    ignore_value(virNetDevBridgeDelete(def->bridge));
2556

2557
    /* See if its still alive and really really kill it */
2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid > 0 &&
        (kill(dnsmasqPid, 0) == 0))
        kill(dnsmasqPid, SIGKILL);
    virNetworkObjSetDnsmasqPid(obj, -1);

    radvdPid = virNetworkObjGetRadvdPid(obj);
    if (radvdPid > 0 &&
        (kill(radvdPid, 0) == 0))
        kill(radvdPid, SIGKILL);
    virNetworkObjSetRadvdPid(obj, -1);
2569

2570 2571 2572
    return 0;
}

2573

2574
static int
2575
networkStartNetworkBridge(virNetworkObjPtr obj)
2576 2577 2578 2579 2580
{
    /* 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.
     */
2581
    return networkStartHandleMACTableManagerMode(obj, NULL);
2582 2583
}

2584

2585
static int
2586
networkShutdownNetworkBridge(virNetworkObjPtr obj ATTRIBUTE_UNUSED)
2587 2588 2589 2590 2591 2592 2593 2594 2595
{
    /* 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;
}


2596 2597 2598 2599 2600 2601 2602 2603 2604
/* 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;
2605
    unsigned int maxVirtFns = 0;
2606 2607 2608 2609 2610 2611
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2612 2613 2614
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2615 2616
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev, &vfNames,
                                      &virtFns, &numVirtFns, &maxVirtFns)) < 0) {
2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631
        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];

2632
        switch ((virNetworkForwardType) netdef->forward.type) {
2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662
        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:
2663
        case VIR_NETWORK_FORWARD_OPEN:
2664 2665
            /* by definition these will never be encountered here */
            break;
2666 2667 2668 2669 2670

        case VIR_NETWORK_FORWARD_LAST:
        default:
            virReportEnumRangeError(virNetworkForwardType, netdef->forward.type);
            goto cleanup;
2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707
        }
    }

    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;
}


2708
static int
2709
networkStartNetworkExternal(virNetworkObjPtr obj)
2710 2711
{
    /* put anything here that needs to be done each time a network of
2712
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2713 2714 2715
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2716
    return networkCreateInterfacePool(virNetworkObjGetDef(obj));
2717 2718
}

2719 2720

static int
2721
networkShutdownNetworkExternal(virNetworkObjPtr obj ATTRIBUTE_UNUSED)
2722 2723
{
    /* put anything here that needs to be done each time a network of
2724
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2725 2726 2727 2728 2729 2730
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

2731

2732
static int
2733
networkStartNetwork(virNetworkDriverStatePtr driver,
2734
                    virNetworkObjPtr obj)
2735
{
2736
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2737 2738
    int ret = -1;

2739
    VIR_DEBUG("driver=%p, network=%p", driver, obj);
2740

2741
    if (virNetworkObjIsActive(obj)) {
2742 2743
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2744
        return ret;
2745 2746
    }

2747 2748 2749
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2750
    if (virNetworkObjSetDefTransient(obj, true) < 0)
2751
        goto cleanup;
2752

2753 2754
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2755
    if (networkRunHook(obj, NULL, NULL,
2756 2757 2758 2759
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2760
    switch ((virNetworkForwardType) def->forward.type) {
2761 2762 2763 2764

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2765
    case VIR_NETWORK_FORWARD_OPEN:
2766
        if (networkStartNetworkVirtual(driver, obj) < 0)
2767
            goto cleanup;
2768 2769 2770
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2771
        if (def->bridge) {
2772
            if (networkStartNetworkBridge(obj) < 0)
2773 2774 2775 2776 2777 2778 2779
                goto cleanup;
            break;
        }
        /* intentionally fall through to the macvtap/direct case for
         * VIR_NETWORK_FORWARD_BRIDGE with no bridge device defined
         * (since that is macvtap bridge mode).
         */
2780 2781
        ATTRIBUTE_FALLTHROUGH;

2782 2783 2784
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2785
    case VIR_NETWORK_FORWARD_HOSTDEV:
2786
        if (networkStartNetworkExternal(obj) < 0)
2787
            goto cleanup;
2788
        break;
2789 2790 2791 2792 2793

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        goto cleanup;
2794 2795
    }

2796
    /* finally we can call the 'started' hook script if any */
2797
    if (networkRunHook(obj, NULL, NULL,
2798 2799 2800 2801
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2802 2803 2804
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2805
    VIR_DEBUG("Writing network status to disk");
2806
    if (virNetworkObjSaveStatus(driver->stateDir, obj) < 0)
2807
        goto cleanup;
2808

2809
    virNetworkObjSetActive(obj, true);
2810
    VIR_INFO("Network '%s' started up", def->name);
2811
    ret = 0;
2812

2813
 cleanup:
2814
    if (ret < 0) {
2815
        virNetworkObjUnsetDefTransient(obj);
2816 2817
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2818
        networkShutdownNetwork(driver, obj);
2819 2820 2821 2822 2823 2824 2825
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2826

2827 2828
static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
2829
                       virNetworkObjPtr obj)
2830
{
2831
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2832 2833 2834
    int ret = 0;
    char *stateFile;

2835
    VIR_INFO("Shutting down network '%s'", def->name);
2836

2837
    if (!virNetworkObjIsActive(obj))
2838 2839
        return 0;

2840
    stateFile = virNetworkConfigFile(driver->stateDir, def->name);
2841 2842 2843 2844 2845 2846
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2847
    switch ((virNetworkForwardType) def->forward.type) {
2848 2849 2850 2851

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2852
    case VIR_NETWORK_FORWARD_OPEN:
2853
        ret = networkShutdownNetworkVirtual(driver, obj);
2854 2855 2856
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2857
        if (def->bridge) {
2858
            ret = networkShutdownNetworkBridge(obj);
2859 2860 2861 2862 2863 2864
            break;
        }
        /* intentionally fall through to the macvtap/direct case for
         * VIR_NETWORK_FORWARD_BRIDGE with no bridge device defined
         * (since that is macvtap bridge mode).
         */
2865 2866
        ATTRIBUTE_FALLTHROUGH;

2867 2868 2869
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2870
    case VIR_NETWORK_FORWARD_HOSTDEV:
2871
        ret = networkShutdownNetworkExternal(obj);
2872
        break;
2873 2874 2875 2876 2877

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        return -1;
2878 2879
    }

2880
    /* now that we know it's stopped call the hook if present */
2881
    networkRunHook(obj, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2882 2883
                   VIR_HOOK_SUBOP_END);

2884
    virNetworkObjSetActive(obj, false);
2885
    virNetworkObjUnsetDefTransient(obj);
2886
    return ret;
2887 2888 2889
}


2890 2891 2892
static virNetworkPtr
networkLookupByUUID(virConnectPtr conn,
                    const unsigned char *uuid)
2893
{
2894
    virNetworkDriverStatePtr driver = networkGetDriver();
2895
    virNetworkObjPtr obj;
2896
    virNetworkDefPtr def;
2897
    virNetworkPtr net = NULL;
2898

2899 2900
    obj = virNetworkObjFindByUUID(driver->networks, uuid);
    if (!obj) {
2901 2902
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2903
        virReportError(VIR_ERR_NO_NETWORK,
2904 2905
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2906
        goto cleanup;
2907
    }
2908
    def = virNetworkObjGetDef(obj);
2909

2910
    if (virNetworkLookupByUUIDEnsureACL(conn, def) < 0)
2911 2912
        goto cleanup;

2913
    net = virGetNetwork(conn, def->name, def->uuid);
2914

2915
 cleanup:
2916 2917
    virNetworkObjEndAPI(&obj);
    return net;
2918 2919
}

2920 2921 2922 2923

static virNetworkPtr
networkLookupByName(virConnectPtr conn,
                    const char *name)
2924
{
2925
    virNetworkDriverStatePtr driver = networkGetDriver();
2926
    virNetworkObjPtr obj;
2927
    virNetworkDefPtr def;
2928
    virNetworkPtr net = NULL;
2929

2930 2931
    obj = virNetworkObjFindByName(driver->networks, name);
    if (!obj) {
2932 2933
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2934
        goto cleanup;
2935
    }
2936
    def = virNetworkObjGetDef(obj);
2937

2938
    if (virNetworkLookupByNameEnsureACL(conn, def) < 0)
2939 2940
        goto cleanup;

2941
    net = virGetNetwork(conn, def->name, def->uuid);
2942

2943
 cleanup:
2944 2945
    virNetworkObjEndAPI(&obj);
    return net;
2946 2947
}

2948 2949 2950

static int
networkConnectNumOfNetworks(virConnectPtr conn)
2951
{
2952
    virNetworkDriverStatePtr driver = networkGetDriver();
2953
    int nactive;
2954

2955 2956 2957
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2958 2959 2960 2961
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
2962

2963 2964 2965
    return nactive;
}

2966 2967 2968 2969

static int
networkConnectListNetworks(virConnectPtr conn,
                           char **const names,
2970
                           int maxnames)
2971 2972
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2973
    int got = 0;
2974

2975 2976 2977
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2978
    got = virNetworkObjListGetNames(driver->networks,
2979
                                    true, names, maxnames,
2980 2981
                                    virConnectListNetworksCheckACL,
                                    conn);
2982

2983 2984 2985
    return got;
}

2986 2987 2988

static int
networkConnectNumOfDefinedNetworks(virConnectPtr conn)
2989
{
2990
    virNetworkDriverStatePtr driver = networkGetDriver();
2991
    int ninactive = 0;
2992

2993 2994 2995
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2996 2997 2998 2999
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
3000

3001 3002 3003
    return ninactive;
}

3004 3005 3006 3007

static int
networkConnectListDefinedNetworks(virConnectPtr conn,
                                  char **const names,
3008
                                  int maxnames)
3009 3010
{
    virNetworkDriverStatePtr driver = networkGetDriver();
3011
    int got = 0;
3012

3013 3014 3015
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

3016
    got = virNetworkObjListGetNames(driver->networks,
3017
                                    false, names, maxnames,
3018 3019
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
3020 3021 3022
    return got;
}

3023

3024
static int
3025 3026 3027
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
3028
{
3029
    virNetworkDriverStatePtr driver = networkGetDriver();
3030 3031 3032 3033
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

3034 3035 3036
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

3037
    ret = virNetworkObjListExport(conn, driver->networks, nets,
3038 3039
                                  virConnectListAllNetworksCheckACL,
                                  flags);
3040

3041
 cleanup:
3042 3043
    return ret;
}
3044

3045

3046 3047 3048 3049 3050 3051 3052 3053
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
3054
    virNetworkDriverStatePtr driver = networkGetDriver();
3055 3056 3057 3058 3059 3060
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
3061
                                       net, eventID, callback,
3062 3063 3064
                                       opaque, freecb, &ret) < 0)
        ret = -1;

3065
 cleanup:
3066 3067 3068
    return ret;
}

3069

3070 3071 3072 3073
static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
3074
    virNetworkDriverStatePtr driver = networkGetDriver();
3075 3076 3077 3078 3079
    int ret = -1;

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

3080 3081
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
3082
                                        callbackID, true) < 0)
3083 3084 3085
        goto cleanup;

    ret = 0;
3086

3087
 cleanup:
3088 3089 3090
    return ret;
}

3091 3092 3093

static int
networkIsActive(virNetworkPtr net)
3094 3095 3096 3097
{
    virNetworkObjPtr obj;
    int ret = -1;

3098 3099
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3100

3101
    if (virNetworkIsActiveEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
3102 3103
        goto cleanup;

3104 3105
    ret = virNetworkObjIsActive(obj);

3106
 cleanup:
3107
    virNetworkObjEndAPI(&obj);
3108 3109 3110
    return ret;
}

3111 3112 3113

static int
networkIsPersistent(virNetworkPtr net)
3114 3115 3116 3117
{
    virNetworkObjPtr obj;
    int ret = -1;

3118 3119
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3120

3121
    if (virNetworkIsPersistentEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
3122 3123
        goto cleanup;

3124
    ret = virNetworkObjIsPersistent(obj);
3125

3126
 cleanup:
3127
    virNetworkObjEndAPI(&obj);
3128 3129 3130 3131
    return ret;
}


3132 3133
/*
 * networkFindUnusedBridgeName() - try to find a bridge name that is
3134 3135 3136
 * 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.
3137 3138 3139 3140 3141 3142 3143
 */
static int
networkFindUnusedBridgeName(virNetworkObjListPtr nets,
                            virNetworkDefPtr def)
{
    int ret = -1, id = 0;
    char *newname = NULL;
3144 3145 3146 3147 3148
    const char *templ = "virbr%d";
    const char *p;

    if (def->bridge &&
        (p = strchr(def->bridge, '%')) == strrchr(def->bridge, '%') &&
3149
        p && p[1] == 'd')
3150
        templ = def->bridge;
3151 3152 3153 3154

    do {
        if (virAsprintf(&newname, templ, id) < 0)
            goto cleanup;
3155 3156 3157 3158 3159
        /* 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).
         */
3160
        if (!(virNetworkObjBridgeInUse(nets, newname, def->name) ||
3161
              virNetDevExists(newname) == 1)) {
3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194
            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")) {
3195
        if (virNetworkObjBridgeInUse(nets, def->bridge, def->name)) {
3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212
            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;
}


3213
static int
3214
networkValidate(virNetworkDriverStatePtr driver,
3215
                virNetworkDefPtr def)
3216
{
3217
    size_t i, j;
3218 3219
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
3220
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
3221
    bool ipv4def = false, ipv6def = false;
3222
    bool bandwidthAllowed = true;
3223
    bool usesInterface = false, usesAddress = false;
3224

3225 3226 3227
    if (virXMLCheckIllegalChars("name", def->name, "\n") < 0)
        return -1;

3228 3229 3230
    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
3231 3232 3233 3234 3235
    switch ((virNetworkForwardType) def->forward.type) {
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
    case VIR_NETWORK_FORWARD_OPEN:
3236 3237 3238 3239
        /* 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)
3240 3241 3242
            return -1;

        virNetworkSetBridgeMacAddr(def);
3243 3244 3245 3246 3247 3248 3249
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
    case VIR_NETWORK_FORWARD_HOSTDEV:
3250
        /* They are also the only types that currently support setting
3251 3252
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
3253
         */
3254 3255 3256 3257 3258 3259 3260 3261
        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;
        }
3262
        if (virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0)) {
3263 3264 3265 3266
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3267
                           virNetworkForwardTypeToString(def->forward.type));
3268 3269
            return -1;
        }
3270
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
3271 3272 3273 3274
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3275
                           virNetworkForwardTypeToString(def->forward.type));
3276 3277 3278 3279 3280 3281 3282
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3283
                           virNetworkForwardTypeToString(def->forward.type));
3284 3285
            return -1;
        }
3286 3287 3288 3289 3290 3291 3292 3293
        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;
        }
3294
        bandwidthAllowed = false;
3295 3296 3297 3298 3299 3300
        break;

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        return -1;
3301 3302
    }

3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315
    /* 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++) {
3316 3317 3318
        virNetworkForwardIfDefPtr iface = &def->forward.ifs[i];
        char *sysfs_path = NULL;

3319
        switch ((virNetworkForwardHostdevDeviceType)iface->type) {
3320 3321
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV:
            usesInterface = true;
3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332

            if (def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("hostdev network '%s' lists '%s' "
                                 "in the device pool, but hostdev "
                                 "networks require all devices to "
                                 "be listed by PCI address, not "
                                 "network device name"),
                               def->name, iface->device.dev);
                return -1;
            }
3333
            break;
3334 3335

        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI: {
3336
            usesAddress = true;
3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361

            if (def->forward.type != VIR_NETWORK_FORWARD_HOSTDEV) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has forward mode '%s' "
                                 " but lists a device by PCI address "
                                 "in the device pool. This is only "
                                 "supported for networks with forward "
                                 "mode 'hostdev'"),
                               def->name,
                               virNetworkForwardTypeToString(def->forward.type));
                return -1;
            }

            if (virPCIDeviceAddressGetSysfsFile(&iface->device.pci, &sysfs_path) < 0)
                return -1;

            if (!virPCIIsVirtualFunction(sysfs_path)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("device '%s' in network '%s' is not "
                                 "an SR-IOV Virtual Function"),
                               sysfs_path, def->name);
                VIR_FREE(sysfs_path);
                return -1;
            }
            VIR_FREE(sysfs_path);
3362
            break;
3363 3364
        }

3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377
        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
3378 3379 3380
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
3381
    for (i = 0;
3382
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
3383
         i++) {
G
Gene Czarcinski 已提交
3384 3385 3386 3387 3388
        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 -- "
3389 3390
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407
                    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;
                }
3408 3409 3410
            }
        }
    }
3411 3412 3413 3414 3415 3416

    /* 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.
     */

3417 3418 3419
    vlanAllowed = (def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV ||
                   def->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH ||
                   (def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
J
Ján Tomko 已提交
3420 3421
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
3422
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH));
3423 3424

    vlanUsed = def->vlan.nTags > 0;
3425 3426
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3427 3428 3429 3430 3431
            /* 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.
             */
3432
            if (def->portGroups[i].virtPortProfile) {
3433
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3434
                    def->portGroups[i].virtPortProfile->virtPortType
3435 3436 3437 3438 3439 3440 3441
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3442
        }
3443
        if (def->portGroups[i].isDefault) {
3444 3445 3446 3447 3448
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3449
                               def->name, defaultPortGroup->name,
3450
                               def->portGroups[i].name);
3451
                return -1;
3452
            }
3453
            defaultPortGroup = &def->portGroups[i];
3454
        }
3455 3456 3457 3458 3459 3460 3461 3462 3463
        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;
            }
        }
3464 3465 3466 3467 3468 3469 3470 3471
        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;
        }
3472
    }
3473 3474 3475 3476 3477 3478 3479
    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.
         */
3480 3481 3482 3483 3484 3485
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499

    if (def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
        for (i = 0; i < def->nPortGroups; i++) {
            if (def->portGroups[i].bandwidth) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("unsupported <bandwidth> element "
                                 "in <portgroup name='%s'> of "
                                 "network '%s' with forward mode='%s'"),
                               def->portGroups[i].name, def->name,
                               virNetworkForwardTypeToString(def->forward.type));
                return -1;
            }
        }
    }
3500 3501 3502
    return 0;
}

3503 3504 3505 3506

static virNetworkPtr
networkCreateXML(virConnectPtr conn,
                 const char *xml)
3507
{
3508
    virNetworkDriverStatePtr driver = networkGetDriver();
3509
    virNetworkDefPtr newDef;
3510
    virNetworkObjPtr obj = NULL;
3511
    virNetworkDefPtr def;
3512
    virNetworkPtr net = NULL;
3513
    virObjectEventPtr event = NULL;
3514

3515
    if (!(newDef = virNetworkDefParseString(xml)))
3516
        goto cleanup;
3517

3518
    if (virNetworkCreateXMLEnsureACL(conn, newDef) < 0)
3519 3520
        goto cleanup;

3521
    if (networkValidate(driver, newDef) < 0)
J
Ján Tomko 已提交
3522
        goto cleanup;
3523

3524 3525 3526
    /* 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.
3527
     */
3528
    if (!(obj = virNetworkObjAssignDef(driver->networks, newDef,
3529 3530
                                       VIR_NETWORK_OBJ_LIST_ADD_LIVE |
                                       VIR_NETWORK_OBJ_LIST_ADD_CHECK_LIVE)))
3531
        goto cleanup;
3532 3533
    newDef = NULL;
    def = virNetworkObjGetDef(obj);
3534

3535 3536
    if (networkStartNetwork(driver, obj) < 0) {
        virNetworkObjRemoveInactive(driver->networks, obj);
3537
        goto cleanup;
3538 3539
    }

3540 3541
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3542 3543
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3544

3545 3546
    VIR_INFO("Creating network '%s'", def->name);
    net = virGetNetwork(conn, def->name, def->uuid);
3547

3548
 cleanup:
3549
    virNetworkDefFree(newDef);
3550
    virObjectEventStateQueue(driver->networkEventState, event);
3551 3552
    virNetworkObjEndAPI(&obj);
    return net;
3553 3554
}

3555 3556 3557 3558

static virNetworkPtr
networkDefineXML(virConnectPtr conn,
                 const char *xml)
3559
{
3560
    virNetworkDriverStatePtr driver = networkGetDriver();
3561
    virNetworkDefPtr def = NULL;
3562
    bool freeDef = true;
3563 3564
    virNetworkObjPtr obj = NULL;
    virNetworkPtr net = NULL;
3565
    virObjectEventPtr event = NULL;
3566

3567
    if (!(def = virNetworkDefParseString(xml)))
3568
        goto cleanup;
3569

3570 3571 3572
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3573
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3574
        goto cleanup;
3575

3576
    if (!(obj = virNetworkObjAssignDef(driver->networks, def, 0)))
J
Ján Tomko 已提交
3577
        goto cleanup;
3578

3579
    /* def was assigned to network object */
3580
    freeDef = false;
3581 3582

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3583 3584
        if (!virNetworkObjIsActive(obj)) {
            virNetworkObjRemoveInactive(driver->networks, obj);
3585 3586
            goto cleanup;
        }
3587 3588 3589 3590
        /* if network was active already, just undo new persistent
         * definition by making it transient.
         * XXX - this isn't necessarily the correct thing to do.
         */
3591
        virNetworkObjUpdateAssignDef(obj, NULL, false);
3592 3593 3594
        goto cleanup;
    }

3595
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3596 3597
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3598

3599
    VIR_INFO("Defining network '%s'", def->name);
3600
    net = virGetNetwork(conn, def->name, def->uuid);
3601

3602
 cleanup:
3603
    virObjectEventStateQueue(driver->networkEventState, event);
3604
    if (freeDef)
J
Ján Tomko 已提交
3605
        virNetworkDefFree(def);
3606 3607
    virNetworkObjEndAPI(&obj);
    return net;
3608 3609
}

3610

3611
static int
3612 3613
networkUndefine(virNetworkPtr net)
{
3614
    virNetworkDriverStatePtr driver = networkGetDriver();
3615
    virNetworkObjPtr obj;
3616
    virNetworkDefPtr def;
3617
    int ret = -1;
3618
    bool active = false;
3619
    virObjectEventPtr event = NULL;
3620

3621
    if (!(obj = networkObjFromNetwork(net)))
3622
        goto cleanup;
3623
    def = virNetworkObjGetDef(obj);
3624

3625
    if (virNetworkUndefineEnsureACL(net->conn, def) < 0)
3626 3627
        goto cleanup;

3628
    if (virNetworkObjIsActive(obj))
3629
        active = true;
3630

3631
    if (!virNetworkObjIsPersistent(obj)) {
3632 3633 3634 3635 3636
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                       _("can't undefine transient network"));
        goto cleanup;
    }

3637
    /* remove autostart link */
3638 3639
    if (virNetworkObjDeleteConfig(driver->networkConfigDir,
                                  driver->networkAutostartDir,
3640
                                  obj) < 0)
3641
        goto cleanup;
3642

3643 3644
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3645 3646
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3647

3648
    VIR_INFO("Undefining network '%s'", def->name);
3649
    if (!active) {
3650
        if (networkRemoveInactive(driver, obj) < 0)
3651
            goto cleanup;
3652 3653 3654 3655 3656
    } else {

        /* if the network still exists, it was active, and we need to make
         * it transient (by deleting the persistent def)
         */
3657
        virNetworkObjUpdateAssignDef(obj, NULL, false);
3658 3659
    }

3660
    ret = 0;
3661

3662
 cleanup:
3663
    virObjectEventStateQueue(driver->networkEventState, event);
3664
    virNetworkObjEndAPI(&obj);
3665
    return ret;
3666 3667
}

3668

3669 3670 3671 3672 3673 3674 3675 3676
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3677
    virNetworkDriverStatePtr driver = networkGetDriver();
3678
    virNetworkObjPtr obj = NULL;
3679
    virNetworkDefPtr def;
3680 3681
    int isActive, ret = -1;
    size_t i;
3682
    virNetworkIPDefPtr ipdef;
3683
    bool oldDhcpActive = false;
3684
    bool needFirewallRefresh = false;
3685

3686 3687 3688 3689 3690

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3691
    if (!(obj = networkObjFromNetwork(net)))
3692
        goto cleanup;
3693
    def = virNetworkObjGetDef(obj);
3694

3695
    if (virNetworkUpdateEnsureACL(net->conn, def, flags) < 0)
3696 3697
        goto cleanup;

3698
    /* see if we are listening for dhcp pre-modification */
3699
    for (i = 0;
3700
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
3701
         i++) {
3702 3703 3704 3705 3706 3707
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3708 3709
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3710
     */
3711
    isActive = virNetworkObjIsActive(obj);
3712 3713
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3714 3715 3716 3717 3718 3719 3720
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3721 3722 3723 3724
    if (isActive && (flags & VIR_NETWORK_UPDATE_AFFECT_LIVE)) {
        /* Take care of anything that must be done before updating the
         * live NetworkDef.
         */
3725 3726 3727 3728
        switch ((virNetworkForwardType) def->forward.type) {
        case VIR_NETWORK_FORWARD_NONE:
        case VIR_NETWORK_FORWARD_NAT:
        case VIR_NETWORK_FORWARD_ROUTE:
3729 3730 3731 3732 3733 3734 3735 3736 3737 3738
            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).
                 */
3739 3740
                networkRemoveFirewallRules(def);
                needFirewallRefresh = true;
3741 3742 3743 3744
                break;
            default:
                break;
            }
3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758
            break;

        case VIR_NETWORK_FORWARD_OPEN:
        case VIR_NETWORK_FORWARD_BRIDGE:
        case VIR_NETWORK_FORWARD_PRIVATE:
        case VIR_NETWORK_FORWARD_VEPA:
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
        case VIR_NETWORK_FORWARD_HOSTDEV:
            break;

        case VIR_NETWORK_FORWARD_LAST:
        default:
            virReportEnumRangeError(virNetworkForwardType, def->forward.type);
            goto cleanup;
3759 3760 3761
        }
    }

3762
    /* update the network config in memory/on disk */
3763
    if (virNetworkObjUpdate(obj, command, section, parentIndex, xml, flags) < 0) {
3764
        if (needFirewallRefresh)
3765
            ignore_value(networkAddFirewallRules(def));
3766 3767 3768
        goto cleanup;
    }

3769 3770 3771 3772
    /* @def is replaced */
    def = virNetworkObjGetDef(obj);

    if (needFirewallRefresh && networkAddFirewallRules(def) < 0)
3773 3774 3775 3776 3777
        goto cleanup;

    if (flags & VIR_NETWORK_UPDATE_AFFECT_CONFIG) {
        /* save updated persistent config to disk */
        if (virNetworkSaveConfig(driver->networkConfigDir,
3778
                                 virNetworkObjGetPersistentDef(obj)) < 0) {
3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791
            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 ||
3792 3793 3794 3795 3796 3797 3798 3799
            section == VIR_NETWORK_SECTION_IP_DHCP_RANGE ||
            section == VIR_NETWORK_SECTION_DNS_TXT ||
            section == VIR_NETWORK_SECTION_DNS_SRV) {
            /* these sections all change things on the dnsmasq
             * commandline (i.e. in the .conf file), so we need to
             * kill and restart dnsmasq, because dnsmasq sets its uid
             * to "nobody" after it starts, and is unable to re-read
             * the conf file (owned by root, mode 600)
3800
             */
3801
            if (networkRestartDhcpDaemon(driver, obj) < 0)
3802 3803
                goto cleanup;

3804 3805 3806 3807 3808 3809 3810 3811
        } 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;

3812
            for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
3813
                 i++) {
3814 3815 3816 3817 3818 3819 3820
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3821 3822
                 networkRestartDhcpDaemon(driver, obj) < 0) ||
                networkRefreshDhcpDaemon(driver, obj) < 0) {
3823 3824 3825
                goto cleanup;
            }

3826 3827 3828 3829
        } else if (section == VIR_NETWORK_SECTION_DNS_HOST) {
            /* this section only changes data in an external file
             * (not the .conf file) so we can just update the config
             * files and send SIGHUP to dnsmasq.
3830
             */
3831
            if (networkRefreshDhcpDaemon(driver, obj) < 0)
3832 3833 3834 3835 3836 3837 3838 3839
                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.
             */
3840
            if (networkRefreshRadvd(driver, obj) < 0)
3841 3842 3843 3844
                goto cleanup;
        }

        /* save current network state to disk */
3845
        if ((ret = virNetworkObjSaveStatus(driver->stateDir, obj)) < 0)
3846 3847
            goto cleanup;
    }
3848 3849

    /* call the 'updated' network hook script */
3850
    if (networkRunHook(obj, NULL, NULL, VIR_HOOK_NETWORK_OP_UPDATED,
3851 3852 3853
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

3854
    ret = 0;
3855
 cleanup:
3856
    virNetworkObjEndAPI(&obj);
3857 3858 3859
    return ret;
}

3860 3861 3862

static int
networkCreate(virNetworkPtr net)
3863
{
3864
    virNetworkDriverStatePtr driver = networkGetDriver();
3865
    virNetworkObjPtr obj;
3866
    virNetworkDefPtr def;
3867
    int ret = -1;
3868
    virObjectEventPtr event = NULL;
3869

3870
    if (!(obj = networkObjFromNetwork(net)))
3871
        goto cleanup;
3872
    def = virNetworkObjGetDef(obj);
3873

3874
    if (virNetworkCreateEnsureACL(net->conn, def) < 0)
3875 3876
        goto cleanup;

3877
    if ((ret = networkStartNetwork(driver, obj)) < 0)
3878
        goto cleanup;
3879

3880 3881
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3882 3883
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3884

3885
 cleanup:
3886
    virObjectEventStateQueue(driver->networkEventState, event);
3887
    virNetworkObjEndAPI(&obj);
3888
    return ret;
3889 3890
}

3891 3892 3893

static int
networkDestroy(virNetworkPtr net)
3894
{
3895
    virNetworkDriverStatePtr driver = networkGetDriver();
3896
    virNetworkObjPtr obj;
3897
    virNetworkDefPtr def;
3898
    int ret = -1;
3899
    virObjectEventPtr event = NULL;
3900

3901
    if (!(obj = networkObjFromNetwork(net)))
3902
        goto cleanup;
3903
    def = virNetworkObjGetDef(obj);
3904

3905
    if (virNetworkDestroyEnsureACL(net->conn, def) < 0)
3906 3907
        goto cleanup;

3908
    if (!virNetworkObjIsActive(obj)) {
3909
        virReportError(VIR_ERR_OPERATION_INVALID,
3910
                       _("network '%s' is not active"),
3911
                       def->name);
3912 3913 3914
        goto cleanup;
    }

3915
    if ((ret = networkShutdownNetwork(driver, obj)) < 0)
3916
        goto cleanup;
3917 3918
    /* @def replaced in virNetworkObjUnsetDefTransient*/
    def = virNetworkObjGetDef(obj);
3919

3920 3921
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3922 3923
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3924

3925 3926
    if (!virNetworkObjIsPersistent(obj) &&
        networkRemoveInactive(driver, obj) < 0) {
3927 3928
        ret = -1;
        goto cleanup;
3929
    }
3930

3931
 cleanup:
3932
    virObjectEventStateQueue(driver->networkEventState, event);
3933
    virNetworkObjEndAPI(&obj);
3934 3935 3936
    return ret;
}

3937 3938 3939 3940

static char *
networkGetXMLDesc(virNetworkPtr net,
                  unsigned int flags)
3941
{
3942
    virNetworkObjPtr obj;
3943
    virNetworkDefPtr curDef;
3944
    virNetworkDefPtr def;
3945
    virNetworkDefPtr newDef;
3946
    char *ret = NULL;
3947

3948
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3949

3950
    if (!(obj = networkObjFromNetwork(net)))
3951
        return ret;
3952 3953
    def = virNetworkObjGetDef(obj);
    newDef = virNetworkObjGetNewDef(obj);
3954

3955
    if (virNetworkGetXMLDescEnsureACL(net->conn, def) < 0)
3956 3957
        goto cleanup;

3958 3959
    if ((flags & VIR_NETWORK_XML_INACTIVE) && newDef)
        curDef = newDef;
3960
    else
3961
        curDef = def;
3962

3963
    ret = virNetworkDefFormat(curDef, flags);
3964

3965
 cleanup:
3966
    virNetworkObjEndAPI(&obj);
3967
    return ret;
3968 3969
}

3970 3971 3972 3973

static char *
networkGetBridgeName(virNetworkPtr net)
{
3974
    virNetworkObjPtr obj;
3975
    virNetworkDefPtr def;
3976 3977
    char *bridge = NULL;

3978
    if (!(obj = networkObjFromNetwork(net)))
3979
        return bridge;
3980
    def = virNetworkObjGetDef(obj);
3981

3982
    if (virNetworkGetBridgeNameEnsureACL(net->conn, def) < 0)
3983 3984
        goto cleanup;

3985
    if (!(def->bridge)) {
3986 3987
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
3988
                       def->name);
3989 3990 3991
        goto cleanup;
    }

3992
    ignore_value(VIR_STRDUP(bridge, def->bridge));
3993

3994
 cleanup:
3995
    virNetworkObjEndAPI(&obj);
3996 3997 3998
    return bridge;
}

3999 4000 4001 4002

static int
networkGetAutostart(virNetworkPtr net,
                    int *autostart)
4003
{
4004
    virNetworkObjPtr obj;
4005
    int ret = -1;
4006

4007
    if (!(obj = networkObjFromNetwork(net)))
4008
        return ret;
4009

4010
    if (virNetworkGetAutostartEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
4011 4012
        goto cleanup;

4013
    *autostart = virNetworkObjIsAutostart(obj) ? 1 : 0;
4014
    ret = 0;
4015

4016
 cleanup:
4017
    virNetworkObjEndAPI(&obj);
4018
    return ret;
4019 4020
}

4021 4022 4023 4024

static int
networkSetAutostart(virNetworkPtr net,
                    int autostart)
4025
{
4026
    virNetworkDriverStatePtr driver = networkGetDriver();
4027
    virNetworkObjPtr obj;
4028
    virNetworkDefPtr def;
4029
    char *configFile = NULL, *autostartLink = NULL;
4030 4031
    bool new_autostart;
    bool cur_autostart;
4032
    int ret = -1;
4033

4034
    if (!(obj = networkObjFromNetwork(net)))
4035
        goto cleanup;
4036
    def = virNetworkObjGetDef(obj);
4037

4038
    if (virNetworkSetAutostartEnsureACL(net->conn, def) < 0)
4039 4040
        goto cleanup;

4041
    if (!virNetworkObjIsPersistent(obj)) {
4042 4043
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
4044 4045 4046
        goto cleanup;
    }

4047 4048 4049
    new_autostart = (autostart != 0);
    cur_autostart = virNetworkObjIsAutostart(obj);
    if (cur_autostart != new_autostart) {
4050 4051
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir,
                                               def->name)) == NULL)
4052
            goto cleanup;
4053 4054
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir,
                                                  def->name)) == NULL)
4055 4056
            goto cleanup;

4057
        if (new_autostart) {
4058
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
4059
                virReportSystemError(errno,
4060 4061
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
4062 4063
                goto cleanup;
            }
4064

4065
            if (symlink(configFile, autostartLink) < 0) {
4066
                virReportSystemError(errno,
4067
                                     _("Failed to create symlink '%s' to '%s'"),
4068
                                     autostartLink, configFile);
4069 4070 4071
                goto cleanup;
            }
        } else {
4072
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
4073
                virReportSystemError(errno,
4074
                                     _("Failed to delete symlink '%s'"),
4075
                                     autostartLink);
4076 4077
                goto cleanup;
            }
4078 4079
        }

4080
        virNetworkObjSetAutostart(obj, new_autostart);
4081
    }
4082

4083
    ret = 0;
4084

4085
 cleanup:
4086 4087
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
4088
    virNetworkObjEndAPI(&obj);
4089
    return ret;
4090 4091
}

4092

4093
static int
4094
networkGetDHCPLeases(virNetworkPtr net,
4095 4096 4097
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
4098
{
4099
    virNetworkDriverStatePtr driver = networkGetDriver();
4100 4101 4102
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
4103
    size_t size = 0;
4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114
    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;
4115
    virNetworkIPDefPtr ipdef_tmp = NULL;
4116 4117
    virNetworkDHCPLeasePtr lease = NULL;
    virNetworkDHCPLeasePtr *leases_ret = NULL;
4118
    virNetworkObjPtr obj;
4119
    virNetworkDefPtr def;
4120
    virMacAddr mac_addr;
4121 4122 4123

    virCheckFlags(0, -1);

4124 4125 4126 4127 4128 4129
    /* only to check if the MAC is valid */
    if (mac && virMacAddrParse(mac, &mac_addr) < 0) {
        virReportError(VIR_ERR_INVALID_MAC, "%s", mac);
        return -1;
    }

4130
    if (!(obj = networkObjFromNetwork(net)))
4131
        return -1;
4132
    def = virNetworkObjGetDef(obj);
4133

4134
    if (virNetworkGetDHCPLeasesEnsureACL(net->conn, def) < 0)
4135
        goto cleanup;
4136 4137

    /* Retrieve custom leases file location */
4138
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(driver, def->bridge);
4139 4140

    /* Read entire contents */
4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154
    if ((custom_lease_file_len = virFileReadAllQuiet(custom_lease_file,
                                                     VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX,
                                                     &lease_entries)) < 0) {
        /* Not all networks are guaranteed to have leases file.
         * Only those which run dnsmasq. Therefore, if we failed
         * to read the leases file, don't report error. Return 0
         * leases instead. */
        if (errno == ENOENT) {
            rv = 0;
        } else {
            virReportSystemError(errno,
                                 _("Unable to read leases file: %s"),
                                 custom_lease_file);
        }
4155 4156 4157 4158 4159 4160 4161 4162 4163 4164
        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;
        }

4165
        if (!virJSONValueIsArray(leases_array)) {
4166
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
4167
                           _("Malformed lease_entries array"));
4168 4169
            goto error;
        }
4170
        size = virJSONValueArraySize(leases_array);
4171 4172
    }

4173
    currtime = (long long)time(NULL);
4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189

    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;
        }

4190
        if (mac && virMacAddrCompare(mac, mac_tmp))
4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221
            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 */
4222 4223
            for (j = 0; j < def->nips; j++) {
                ipdef_tmp = &def->ips[j];
4224 4225 4226 4227 4228 4229 4230 4231

                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)) {
4232
                    lease->prefix = virSocketAddrGetIPPrefix(&ipdef_tmp->address,
4233 4234 4235 4236 4237 4238 4239 4240
                                                             &ipdef_tmp->netmask,
                                                             ipdef_tmp->prefix);
                    break;
                }
            }

            if ((VIR_STRDUP(lease->mac, mac_tmp) < 0) ||
                (VIR_STRDUP(lease->ipaddr, ip_tmp) < 0) ||
4241
                (VIR_STRDUP(lease->iface, def->bridge) < 0))
4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273
                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);
4274
    VIR_FREE(lease_entries);
4275 4276
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
4277

4278
    virNetworkObjEndAPI(&obj);
4279

4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290
    return rv;

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

4291

4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326
/* A unified function to log network connections and disconnections */

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

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

4327

4328 4329 4330 4331 4332 4333 4334 4335 4336
/* 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:
4337
 * @dom: domain definition that @iface belongs to
4338 4339 4340 4341 4342 4343 4344 4345 4346 4347
 * @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.
 */
4348
static int
4349 4350
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
4351
{
4352
    virNetworkDriverStatePtr driver = networkGetDriver();
4353
    virDomainNetType actualType = iface->type;
4354
    virNetworkObjPtr obj = NULL;
4355
    virNetworkDefPtr netdef = NULL;
4356
    virNetDevBandwidthPtr bandwidth = NULL;
4357 4358 4359
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
4360
    virNetworkForwardIfDefPtr dev = NULL;
4361
    size_t i;
4362 4363 4364
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
4365
        goto validate;
4366 4367 4368 4369

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

4370 4371
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
4372 4373 4374
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4375
        goto error;
4376
    }
4377
    netdef = virNetworkObjGetDef(obj);
4378

4379
    if (!virNetworkObjIsActive(obj)) {
4380 4381 4382 4383 4384 4385
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

4386 4387 4388
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

4389 4390 4391
    /* 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
4392
     */
4393 4394 4395 4396 4397 4398
    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.
     */
4399 4400 4401 4402 4403 4404

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

4405 4406
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
4407
        goto error;
4408

4409 4410 4411 4412 4413 4414 4415 4416
    /* 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;

4417 4418
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
4419

4420 4421 4422 4423 4424 4425 4426 4427 4428 4429
    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;

4430 4431 4432 4433 4434
    switch ((virNetworkForwardType) netdef->forward.type) {
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
    case VIR_NETWORK_FORWARD_OPEN:
4435
        /* for these forward types, the actual net type really *is*
4436
         * NETWORK; we just keep the info from the portgroup in
4437
         * iface->data.network.actual
J
Ján Tomko 已提交
4438
         */
4439
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
4440

4441
        /* we also store the bridge device and macTableManager settings
4442 4443 4444 4445 4446 4447 4448
         * 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;
4449 4450
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4451

4452
        if (networkPlugBandwidth(obj, iface) < 0)
4453
            goto error;
4454
        break;
4455

4456
    case VIR_NETWORK_FORWARD_HOSTDEV: {
4457
        virDomainHostdevSubsysPCIBackendType backend;
4458

4459
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
4460
        if (networkCreateInterfacePool(netdef) < 0)
4461 4462 4463
            goto error;

        /* pick first dev with 0 connections */
4464 4465 4466
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480
                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;
4481
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
4482
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
4483
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
4484

E
Eric Blake 已提交
4485
        switch (netdef->forward.driverName) {
4486
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
4487
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
4488 4489
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
4490
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
4491 4492
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
4493
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504
            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;

4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528
        /* 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;
            }
        }
4529 4530
        break;
    }
4531

4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543
    case VIR_NETWORK_FORWARD_BRIDGE:
        if (netdef->bridge) {
            /* <forward type='bridge'/> <bridge name='xxx'/>
             * is VIR_DOMAIN_NET_TYPE_BRIDGE
             */

            iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
            if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                           netdef->bridge) < 0)
                goto error;
            iface->data.network.actual->data.bridge.macTableManager
               = netdef->macTableManager;
4544

4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577
            /* 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) {
                /* 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);
                    goto error;
                }
            }
            break;
        }

        /* intentionally fall through to the direct case for
         * VIR_NETWORK_FORWARD_BRIDGE with no bridge device defined
         */
        ATTRIBUTE_FALLTHROUGH;

    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
4578 4579 4580 4581 4582
        /* <forward type='bridge|private|vepa|passthrough'> are all
         * VIR_DOMAIN_NET_TYPE_DIRECT.
         */

        /* Set type=direct and appropriate <source mode='xxx'/> */
4583
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4584 4585 4586 4587 4588 4589

        /* NO need to check the value returned from virNetDevMacVLanModeTypeFromString
         * it must be valid for these forward type(bridge|private|vepa|passthrough)
         */
        iface->data.network.actual->data.direct.mode =
            virNetDevMacVLanModeTypeFromString(virNetworkForwardTypeToString(netdef->forward.type));
4590

4591 4592 4593 4594 4595 4596 4597 4598
        /* 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) {
4599
            goto error;
4600
        }
4601
        virtport = iface->data.network.actual->virtPortProfile;
4602
        if (virtport) {
4603 4604 4605 4606 4607 4608 4609 4610
            /* 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);
4611
                goto error;
4612 4613
            }
        }
4614

4615 4616 4617
        /* 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).
         */
4618
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4619 4620 4621 4622
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4623
            goto error;
4624 4625 4626
        } else {
            /* pick an interface from the pool */

4627
            if (networkCreateInterfacePool(netdef) < 0)
4628 4629
                goto error;

4630 4631 4632 4633 4634
            /* 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.
4635
             */
4636 4637
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4638 4639 4640
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4641

4642
                /* pick first dev with 0 connections */
4643 4644 4645
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4646 4647 4648 4649 4650
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4651
                dev = &netdef->forward.ifs[0];
4652 4653 4654
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4655 4656 4657 4658
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4659 4660 4661 4662
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4663
                goto error;
4664
            }
4665 4666
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4667
                goto error;
4668
        }
4669 4670 4671 4672 4673 4674
        break;

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, netdef->forward.type);
        goto error;
4675 4676
    }

4677 4678
    if (virNetworkObjMacMgrAdd(obj, driver->dnsmasqStateDir,
                               dom->name, &iface->mac) < 0)
M
Michal Privoznik 已提交
4679 4680
        goto error;

4681
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4682
        goto error;
4683

4684
 validate:
4685 4686 4687 4688 4689
    /* 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.
     */

4690
    if (virDomainNetGetActualVlan(iface)) {
4691 4692 4693 4694
        /* vlan configuration via libvirt is only supported for PCI
         * Passthrough SR-IOV devices (hostdev or macvtap passthru
         * mode) and openvswitch bridges. Otherwise log an error and
         * fail
4695 4696
         */
        if (!(actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV ||
4697 4698 4699
              (actualType == VIR_DOMAIN_NET_TYPE_DIRECT &&
               virDomainNetGetActualDirectMode(iface)
               == VIR_NETDEV_MACVLAN_MODE_PASSTHRU) ||
4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718
              (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;
        }
    }
4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729
    if (virDomainNetGetActualBandwidth(iface)) {
        /* bandwidth configuration via libvirt is not supported for
         * hostdev network devices
         */
        if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("bandwidth settings are not supported "
                             "for hostdev interfaces"));
            goto error;
        }
    }
4730 4731 4732

    if (netdef) {
        netdef->connections++;
4733
        if (dev)
4734 4735
            dev->connections++;
        /* finally we can call the 'plugged' hook script if any */
4736
        if (networkRunHook(obj, dom, iface,
4737 4738 4739
                           VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
                           VIR_HOOK_SUBOP_BEGIN) < 0) {
            /* adjust for failure */
4740
            netdef->connections--;
4741 4742 4743 4744
            if (dev)
                dev->connections--;
            goto error;
        }
4745
        networkLogAllocation(netdef, actualType, dev, iface, true);
4746 4747
    }

4748
    ret = 0;
4749

4750
 cleanup:
4751
    virNetworkObjEndAPI(&obj);
4752 4753
    return ret;

4754
 error:
4755
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4756 4757 4758
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4759
    goto cleanup;
4760 4761
}

4762

4763
/* networkNotifyActualDevice:
4764
 * @dom: domain definition that @iface belongs to
4765 4766 4767 4768 4769
 * @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
4770 4771
 * order, or re-attach the interface's tap device to the network's
 * bridge.
4772
 *
4773
 * No return value (but does log any failures)
4774
 */
4775
static void
4776 4777
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4778
{
4779
    virNetworkDriverStatePtr driver = networkGetDriver();
4780
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4781
    virNetworkObjPtr obj;
4782
    virNetworkDefPtr netdef;
4783
    virNetworkForwardIfDefPtr dev = NULL;
4784
    size_t i;
4785
    char *master = NULL;
4786 4787

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
4788
        return;
4789

4790 4791
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
4792 4793 4794
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4795 4796
        goto error;
    }
4797
    netdef = virNetworkObjGetDef(obj);
4798

4799
    if (!virNetworkObjIsActive(obj)) {
4800 4801 4802 4803 4804 4805
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816
    /* 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;

4817 4818
    /* see if we're connected to the correct bridge */
    if (netdef->bridge) {
4819 4820
        bool useOVS = false;

4821 4822 4823
        if (virNetDevGetMaster(iface->ifname, &master) < 0)
            goto error;

4824 4825 4826 4827 4828 4829 4830 4831
        /* IFLA_MASTER for a tap on an OVS switch is always "ovs-system" */
        if (STREQ_NULLABLE(master, "ovs-system")) {
            useOVS = true;
            VIR_FREE(master);
            if (virNetDevOpenvswitchInterfaceGetMaster(iface->ifname, &master) < 0)
                goto error;
        }

4832 4833
        if (STRNEQ_NULLABLE(netdef->bridge, master)) {
            /* disconnect from current (incorrect) bridge */
4834 4835 4836 4837 4838 4839 4840
            if (master) {
                VIR_INFO("Removing %s from %s", iface->ifname, master);
                if (useOVS)
                    ignore_value(virNetDevOpenvswitchRemovePort(master, iface->ifname));
                else
                    ignore_value(virNetDevBridgeRemovePort(master, iface->ifname));
            }
4841 4842 4843 4844 4845 4846

            /* attach/reattach to correct bridge.
             * NB: we can't notify the guest of any MTU change anyway,
             * so there is no point in trying to learn the actualMTU
             * (final arg to virNetDevTapAttachBridge())
             */
4847
            VIR_INFO("Attaching %s to %s", iface->ifname, netdef->bridge);
4848 4849 4850 4851 4852 4853 4854 4855 4856 4857
            if (virNetDevTapAttachBridge(iface->ifname, netdef->bridge,
                                         &iface->mac, dom->uuid,
                                         virDomainNetGetActualVirtPortProfile(iface),
                                         virDomainNetGetActualVlan(iface),
                                         iface->mtu, NULL) < 0) {
                goto error;
            }
        }
    }

4858
    if (!iface->data.network.actual ||
4859 4860
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4861 4862
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4863 4864
    }

4865
    if (networkCreateInterfacePool(netdef) < 0)
4866
        goto error;
4867

4868
    if (netdef->forward.nifs == 0) {
4869
        virReportError(VIR_ERR_INTERNAL_ERROR,
4870 4871
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4872
                       netdef->name);
4873
        goto error;
4874
    }
4875

4876 4877
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4878

4879 4880 4881 4882 4883 4884 4885 4886 4887
        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 */
4888 4889
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4890
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4891 4892
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4893 4894 4895 4896 4897
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4898
            virReportError(VIR_ERR_INTERNAL_ERROR,
4899 4900
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4901
                           netdef->name, actualDev);
4902
            goto error;
4903 4904
        }

4905
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4906 4907
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4908
         */
4909
        if ((dev->connections > 0) &&
4910 4911
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4912 4913
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4914
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4915
            virReportError(VIR_ERR_INTERNAL_ERROR,
4916 4917
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4918
                           netdef->name, actualDev);
4919
            goto error;
4920
        }
4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932
    }  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 */
4933 4934
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4935
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4936
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4937 4938
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4939 4940 4941 4942 4943 4944 4945 4946 4947
                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,
4948 4949 4950 4951
                           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
4952
            goto error;
4953 4954 4955 4956 4957 4958 4959
        }

        /* 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) &&
4960
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4961 4962 4963 4964 4965 4966 4967 4968 4969
            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;
        }
4970 4971
    }

4972
 success:
4973
    netdef->connections++;
4974 4975
    if (dev)
        dev->connections++;
4976
    /* finally we can call the 'plugged' hook script if any */
4977
    if (networkRunHook(obj, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
4978 4979 4980 4981 4982 4983 4984
                       VIR_HOOK_SUBOP_BEGIN) < 0) {
        /* adjust for failure */
        if (dev)
            dev->connections--;
        netdef->connections--;
        goto error;
    }
4985
    networkLogAllocation(netdef, actualType, dev, iface, true);
4986

4987
 cleanup:
4988
    virNetworkObjEndAPI(&obj);
4989
    VIR_FREE(master);
4990
    return;
4991

4992
 error:
4993
    goto cleanup;
4994 4995 4996 4997
}


/* networkReleaseActualDevice:
4998
 * @dom: domain definition that @iface belongs to
4999 5000 5001 5002 5003 5004 5005 5006 5007
 * @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.
 */
5008
static int
5009 5010
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
5011
{
5012
    virNetworkDriverStatePtr driver = networkGetDriver();
5013
    virDomainNetType actualType = virDomainNetGetActualType(iface);
5014
    virNetworkObjPtr obj;
5015
    virNetworkDefPtr netdef;
5016
    virNetworkForwardIfDefPtr dev = NULL;
5017 5018
    size_t i;
    int ret = -1;
5019 5020

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
5021
        return 0;
5022

5023 5024
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
5025 5026 5027
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
5028 5029
        goto error;
    }
5030
    netdef = virNetworkObjGetDef(obj);
5031

5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050
    switch ((virNetworkForwardType) netdef->forward.type) {
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
    case VIR_NETWORK_FORWARD_OPEN:
        if (iface->data.network.actual && networkUnplugBandwidth(obj, iface) < 0)
            goto error;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
    case VIR_NETWORK_FORWARD_HOSTDEV:
        break;

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, netdef->forward.type);
5051
        goto error;
5052
    }
5053

5054 5055 5056
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
5057 5058
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
5059 5060
    }

5061
    if (netdef->forward.nifs == 0) {
5062
        virReportError(VIR_ERR_INTERNAL_ERROR,
5063
                       _("network '%s' uses a direct/hostdev mode, but "
5064 5065
                         "has no forward dev and no interface pool"),
                       netdef->name);
5066
        goto error;
5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078
    }

    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;
        }
5079

5080 5081
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
5082
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
5083 5084
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
5085 5086 5087
                break;
            }
        }
5088

5089
        if (!dev) {
5090
            virReportError(VIR_ERR_INTERNAL_ERROR,
5091 5092
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
5093
                           netdef->name, actualDev);
5094
            goto error;
5095
        }
5096 5097 5098 5099 5100 5101 5102 5103 5104 5105
    } 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;
        }

5106 5107
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
5108
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
5109
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
5110 5111
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
5112 5113 5114 5115 5116 5117 5118 5119 5120
                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,
5121 5122 5123 5124
                           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
5125
            goto error;
5126
        }
J
Ján Tomko 已提交
5127
    }
5128

5129
 success:
5130
    virNetworkObjMacMgrDel(obj, driver->dnsmasqStateDir, dom->name, &iface->mac);
M
Michal Privoznik 已提交
5131

5132
    if (iface->data.network.actual) {
5133
        netdef->connections--;
5134 5135
        if (dev)
            dev->connections--;
5136
        /* finally we can call the 'unplugged' hook script if any */
5137
        networkRunHook(obj, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
5138
                       VIR_HOOK_SUBOP_BEGIN);
5139
        networkLogAllocation(netdef, actualType, dev, iface, false);
5140
    }
5141
    ret = 0;
5142
 cleanup:
5143
    virNetworkObjEndAPI(&obj);
5144 5145 5146 5147
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
5148
    return ret;
5149

5150
 error:
5151
    goto cleanup;
5152
}
5153

5154

5155 5156 5157
/**
 * networkCheckBandwidth:
 * @net: network QoS
5158
 * @ifaceBand: interface QoS (may be NULL if no QoS)
5159
 * @oldBandwidth: new interface QoS (may be NULL if no QoS)
5160
 * @ifaceMac: interface MAC (used in error messages for identification)
5161 5162
 * @new_rate: new rate for non guaranteed class
 *
5163 5164 5165 5166 5167 5168 5169 5170
 * Function checks if @ifaceBand can be satisfied on @net. However, sometimes it
 * may happen that the interface that @ifaceBand corresponds to is already
 * plugged into the @net and the bandwidth is to be updated. In that case we
 * need to check if new bandwidth can be satisfied. If that's the case
 * @ifaceBand should point to new bandwidth settings and @oldBandwidth to
 * current ones. If you want to suppress this functionality just pass
 * @oldBandwidth == NULL.
 *
5171 5172 5173 5174 5175
 * 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
5176
networkCheckBandwidth(virNetworkObjPtr obj,
5177
                      virNetDevBandwidthPtr ifaceBand,
5178
                      virNetDevBandwidthPtr oldBandwidth,
5179
                      virMacAddr ifaceMac,
5180 5181 5182
                      unsigned long long *new_rate)
{
    int ret = -1;
5183 5184
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
    virNetDevBandwidthPtr netBand = def->bandwidth;
5185
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5186 5187 5188
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

5189
    virMacAddrFormat(&ifaceMac, ifmac);
5190 5191 5192 5193 5194 5195

    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"),
5196
                       ifmac, def->name);
5197 5198 5199
        return -1;
    }

5200 5201
    if (((!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor) &&
         (!oldBandwidth || !oldBandwidth->in || !oldBandwidth->in->floor)) ||
5202 5203
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
5204
        return 1;
5205
    }
5206 5207

    tmp_new_rate = netBand->in->average;
5208 5209 5210 5211
    if (oldBandwidth && oldBandwidth->in)
        tmp_floor_sum -= oldBandwidth->in->floor;
    if (ifaceBand && ifaceBand->in)
        tmp_floor_sum += ifaceBand->in->floor;
5212 5213 5214 5215 5216 5217 5218 5219 5220

    /* 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,
5221 5222
                           def->bridge,
                           def->name);
5223 5224 5225 5226 5227 5228 5229 5230 5231
            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,
5232 5233
                       def->bridge,
                       def->name);
5234 5235 5236
        goto cleanup;
    }

5237 5238
    if (new_rate)
        *new_rate = tmp_new_rate;
5239 5240
    ret = 0;

5241
 cleanup:
5242 5243 5244
    return ret;
}

5245

5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256
/**
 * 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
5257
networkNextClassID(virNetworkObjPtr obj)
5258
{
5259
    ssize_t ret = 0;
5260
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5261

5262 5263
    if ((ret = virBitmapNextClearBit(classIdMap, -1)) < 0)
        ret = virBitmapSize(classIdMap);
5264

5265
    if (virBitmapSetBitExpand(classIdMap, ret) < 0)
5266 5267 5268 5269 5270
        return -1;

    return ret;
}

5271

5272
static int
5273
networkPlugBandwidthImpl(virNetworkObjPtr obj,
5274 5275 5276
                         virDomainNetDefPtr iface,
                         virNetDevBandwidthPtr ifaceBand,
                         unsigned long long new_rate)
5277
{
5278
    virNetworkDriverStatePtr driver = networkGetDriver();
5279
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
5280
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5281
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5282
    ssize_t class_id = 0;
5283 5284
    int plug_ret;
    int ret = -1;
5285 5286

    /* generate new class_id */
5287
    if ((class_id = networkNextClassID(obj)) < 0) {
5288 5289 5290 5291 5292
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Could not generate next class ID"));
        goto cleanup;
    }

5293
    plug_ret = virNetDevBandwidthPlug(def->bridge, def->bandwidth,
5294
                                      &iface->mac, ifaceBand, class_id);
5295
    if (plug_ret < 0) {
5296
        ignore_value(virNetDevBandwidthUnplug(def->bridge, class_id));
5297 5298 5299 5300 5301 5302
        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 */
5303 5304
    tmp_floor_sum += ifaceBand->in->floor;
    virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5305
    /* update status file */
5306
    if (virNetworkObjSaveStatus(driver->stateDir, obj) < 0) {
5307
        ignore_value(virBitmapClearBit(classIdMap, class_id));
5308 5309
        tmp_floor_sum -= ifaceBand->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5310
        iface->data.network.actual->class_id = 0;
5311
        ignore_value(virNetDevBandwidthUnplug(def->bridge, class_id));
5312 5313
        goto cleanup;
    }
5314
    /* update rate for non guaranteed NICs */
5315
    new_rate -= tmp_floor_sum;
5316 5317
    if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                     def->bandwidth, new_rate) < 0)
5318
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
5319
                 def->bridge);
5320 5321

    ret = 0;
5322 5323 5324 5325 5326 5327
 cleanup:
    return ret;
}


static int
5328
networkPlugBandwidth(virNetworkObjPtr obj,
5329 5330 5331 5332 5333 5334 5335 5336
                     virDomainNetDefPtr iface)
{
    int ret = -1;
    int plug_ret;
    unsigned long long new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);

5337
    if ((plug_ret = networkCheckBandwidth(obj, ifaceBand, NULL,
5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357
                                          iface->mac, &new_rate)) < 0) {
        /* helper reported error */
        goto cleanup;
    }

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

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

5358
    if (networkPlugBandwidthImpl(obj, iface, ifaceBand, new_rate) < 0)
5359 5360 5361
        goto cleanup;

    ret = 0;
5362

5363
 cleanup:
5364 5365 5366
    return ret;
}

5367

5368
static int
5369
networkUnplugBandwidth(virNetworkObjPtr obj,
5370 5371
                       virDomainNetDefPtr iface)
{
5372
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
5373
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5374
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5375
    virNetworkDriverStatePtr driver = networkGetDriver();
5376 5377
    int ret = 0;
    unsigned long long new_rate;
5378
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
5379 5380 5381

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
5382
        if (!def->bandwidth || !def->bandwidth->in) {
5383
            VIR_WARN("Network %s has no bandwidth but unplug requested",
5384
                     def->name);
5385 5386
            goto cleanup;
        }
5387
        /* we must remove class from bridge */
5388
        new_rate = def->bandwidth->in->average;
5389

5390 5391
        if (def->bandwidth->in->peak > 0)
            new_rate = def->bandwidth->in->peak;
5392

5393
        ret = virNetDevBandwidthUnplug(def->bridge,
5394 5395 5396 5397
                                       iface->data.network.actual->class_id);
        if (ret < 0)
            goto cleanup;
        /* update sum of 'floor'-s of attached NICs */
5398 5399 5400
        tmp_floor_sum -= ifaceBand->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);

5401
        /* return class ID */
5402
        ignore_value(virBitmapClearBit(classIdMap,
5403 5404
                                       iface->data.network.actual->class_id));
        /* update status file */
5405
        if (virNetworkObjSaveStatus(driver->stateDir, obj) < 0) {
5406 5407
            tmp_floor_sum += ifaceBand->in->floor;
            virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5408
            ignore_value(virBitmapSetBit(classIdMap,
5409 5410 5411
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
5412
        /* update rate for non guaranteed NICs */
5413
        new_rate -= tmp_floor_sum;
5414 5415
        if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                         def->bandwidth, new_rate) < 0)
5416
            VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
5417
                     def->bridge);
5418 5419 5420 5421
        /* no class is associated any longer */
        iface->data.network.actual->class_id = 0;
    }

5422
 cleanup:
5423 5424
    return ret;
}
5425

5426

5427
static void
5428
networkNetworkObjTaint(virNetworkObjPtr obj,
5429
                       virNetworkTaintFlags taint)
5430
{
5431 5432
    virNetworkDefPtr def = virNetworkObjGetDef(obj);

5433
    if (virNetworkObjTaint(obj, taint)) {
5434
        char uuidstr[VIR_UUID_STRING_BUFLEN];
5435
        virUUIDFormat(def->uuid, uuidstr);
5436 5437

        VIR_WARN("Network name='%s' uuid=%s is tainted: %s",
5438
                 def->name, uuidstr, virNetworkTaintTypeToString(taint));
5439 5440
    }
}
5441 5442 5443 5444 5445 5446


static bool
networkBandwidthGenericChecks(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
5447
    virNetDevBandwidthPtr ifaceBand;
5448 5449 5450 5451 5452 5453 5454 5455
    unsigned long long old_floor, new_floor;

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

5456
    ifaceBand = virDomainNetGetActualBandwidth(iface);
5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467
    old_floor = new_floor = 0;

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

    return new_floor != old_floor;
}


5468
static bool
5469 5470 5471 5472
networkBandwidthChangeAllowed(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
5473
    virNetworkObjPtr obj = NULL;
5474 5475 5476 5477 5478 5479
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    bool ret = false;

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

5480 5481
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
5482 5483 5484 5485 5486 5487
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return false;
    }

5488
    if (networkCheckBandwidth(obj, newBandwidth, ifaceBand, iface->mac, NULL) < 0)
5489 5490 5491 5492 5493
        goto cleanup;

    ret = true;

 cleanup:
5494
    virNetworkObjEndAPI(&obj);
5495 5496
    return ret;
}
5497 5498


5499
static int
5500 5501 5502 5503
networkBandwidthUpdate(virDomainNetDefPtr iface,
                       virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
5504
    virNetworkObjPtr obj = NULL;
5505
    virNetworkDefPtr def;
5506
    unsigned long long tmp_floor_sum;
5507 5508 5509 5510 5511 5512 5513 5514
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    unsigned long long new_rate = 0;
    int plug_ret;
    int ret = -1;

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

5515 5516
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
5517 5518 5519 5520 5521
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return ret;
    }
5522
    def = virNetworkObjGetDef(obj);
5523

5524
    if ((plug_ret = networkCheckBandwidth(obj, newBandwidth, ifaceBand,
5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537
                                          iface->mac, &new_rate)) < 0) {
        /* helper reported error */
        goto cleanup;
    }

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

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

5538
    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
5539
        newBandwidth->in && newBandwidth->in->floor) {
5540 5541
        /* Either we just need to update @floor .. */

5542
        if (virNetDevBandwidthUpdateRate(def->bridge,
5543
                                         iface->data.network.actual->class_id,
5544
                                         def->bandwidth,
5545 5546 5547
                                         newBandwidth->in->floor) < 0)
            goto cleanup;

5548 5549 5550 5551 5552
        tmp_floor_sum = virNetworkObjGetFloorSum(obj);
        tmp_floor_sum -= ifaceBand->in->floor;
        tmp_floor_sum += newBandwidth->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);
        new_rate -= tmp_floor_sum;
5553

5554 5555
        if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                         def->bandwidth, new_rate) < 0 ||
5556
            virNetworkObjSaveStatus(driver->stateDir, obj) < 0) {
5557
            /* Ouch, rollback */
5558 5559 5560
            tmp_floor_sum -= newBandwidth->in->floor;
            tmp_floor_sum += ifaceBand->in->floor;
            virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5561

5562
            ignore_value(virNetDevBandwidthUpdateRate(def->bridge,
5563
                                                      iface->data.network.actual->class_id,
5564
                                                      def->bandwidth,
5565 5566 5567 5568 5569 5570
                                                      ifaceBand->in->floor));
            goto cleanup;
        }
    } else if (newBandwidth->in && newBandwidth->in->floor) {
        /* .. or we need to plug in new .. */

5571
        if (networkPlugBandwidthImpl(obj, iface, newBandwidth, new_rate) < 0)
5572 5573 5574 5575
            goto cleanup;
    } else {
        /* .. or unplug old. */

5576
        if (networkUnplugBandwidth(obj, iface) < 0)
5577 5578 5579 5580 5581
            goto cleanup;
    }

    ret = 0;
 cleanup:
5582
    virNetworkObjEndAPI(&obj);
5583 5584
    return ret;
}
5585 5586 5587 5588 5589 5590 5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612


static virNetworkDriver networkDriver = {
    .name = "bridge",
    .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 */
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
    .networkUndefine = networkUndefine, /* 0.2.0 */
    .networkUpdate = networkUpdate, /* 0.10.2 */
    .networkCreate = networkCreate, /* 0.2.0 */
    .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 */
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
};

5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624

static virHypervisorDriver networkHypervisorDriver = {
    .name = "network",
    .connectOpen = networkConnectOpen, /* 4.1.0 */
    .connectClose = networkConnectClose, /* 4.1.0 */
    .connectIsEncrypted = networkConnectIsEncrypted, /* 4.1.0 */
    .connectIsSecure = networkConnectIsSecure, /* 4.1.0 */
    .connectIsAlive = networkConnectIsAlive, /* 4.1.0 */
};


static virConnectDriver networkConnectDriver = {
5625
    .localOnly = true,
5626
    .uriSchemes = (const char *[]){ "network", NULL },
5627 5628 5629 5630 5631
    .hypervisorDriver = &networkHypervisorDriver,
    .networkDriver = &networkDriver,
};


5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642
static virStateDriver networkStateDriver = {
    .name = "bridge",
    .stateInitialize  = networkStateInitialize,
    .stateAutoStart  = networkStateAutoStart,
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
};

int
networkRegister(void)
{
5643 5644
    if (virRegisterConnectDriver(&networkConnectDriver, false) < 0)
        return -1;
5645 5646 5647 5648
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
        return -1;
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
5649 5650 5651 5652

    virDomainNetSetDeviceImpl(
        networkAllocateActualDevice,
        networkNotifyActualDevice,
5653 5654
        networkReleaseActualDevice,
        networkBandwidthChangeAllowed,
5655
        networkBandwidthUpdate);
5656

5657 5658
    return 0;
}