bridge_driver.c 184.3 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 1415 1416
    /* Configure DHCP to tell clients about the MTU. */
    if (def->mtu > 0)
        virBufferAsprintf(&configbuf, "dhcp-option=option:mtu,%d\n", def->mtu);

G
Gene Czarcinski 已提交
1417 1418
    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1419
        if (ipv6def) {
1420
            virBufferAddLit(&configbuf, "enable-ra\n");
1421
        } else {
1422
            for (i = 0;
1423
                 (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i));
1424
                 i++) {
G
Gene Czarcinski 已提交
1425 1426 1427 1428
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1429 1430
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1431 1432
                    VIR_FREE(bridgeaddr);
                }
1433
            }
1434
        }
1435 1436
    }

1437 1438 1439
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1440
    ret = 0;
G
Gene Czarcinski 已提交
1441

1442
 cleanup:
1443 1444
    VIR_FREE(saddr);
    VIR_FREE(eaddr);
1445
    virBufferFreeAndReset(&configbuf);
1446
    return ret;
1447 1448
}

1449

1450
/* build the dnsmasq command line */
1451 1452
static int ATTRIBUTE_NONNULL(3)
networkBuildDhcpDaemonCommandLine(virNetworkDriverStatePtr driver,
1453
                                  virNetworkObjPtr obj,
1454
                                  virCommandPtr *cmdout,
1455 1456
                                  char *pidfile,
                                  dnsmasqContext *dctx)
1457
{
1458
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1459
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1460
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1461
    int ret = -1;
1462 1463
    char *configfile = NULL;
    char *configstr = NULL;
1464
    char *leaseshelper_path = NULL;
1465

1466
    virNetworkObjSetDnsmasqPid(obj, -1);
1467

1468
    if (networkDnsmasqConfContents(obj, pidfile, &configstr,
1469
                                   dctx, dnsmasq_caps) < 0)
1470 1471 1472 1473 1474
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1475
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
1476 1477 1478 1479 1480
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1481 1482
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1483 1484 1485
        goto cleanup;
    }

1486 1487
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1488
                                                  abs_topbuilddir "/src",
1489 1490 1491
                                                  LIBEXECDIR)))
        goto cleanup;

1492
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(dnsmasq_caps));
1493
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1494 1495
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1496
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1497
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", def->bridge);
1498

1499
    *cmdout = cmd;
1500
    ret = 0;
1501
 cleanup:
1502
    virObjectUnref(dnsmasq_caps);
1503 1504
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1505
    VIR_FREE(leaseshelper_path);
1506 1507 1508
    return ret;
}

1509

1510
static int
1511
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
1512
                       virNetworkObjPtr obj)
1513
{
1514
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1515 1516 1517
    virNetworkIPDefPtr ipdef;
    size_t i;
    bool needDnsmasq = false;
1518 1519
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
1520
    pid_t dnsmasqPid;
1521
    int ret = -1;
1522
    dnsmasqContext *dctx = NULL;
1523

1524
    /* see if there are any IP addresses that need a dhcp server */
1525
    i = 0;
1526
    while ((ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i))) {
1527
        i++;
1528 1529 1530 1531
        if (ipdef->nranges || ipdef->nhosts)
            needDnsmasq = true;
    }

1532 1533 1534 1535 1536 1537
    if (i == 0) {
        /* no IP addresses at all, so we don't need to run */
        ret = 0;
        goto cleanup;
    }

1538
    if (!needDnsmasq && def->dns.enable == VIR_TRISTATE_BOOL_NO) {
1539
        /* no DHCP services needed, and user disabled DNS service */
1540 1541 1542 1543
        ret = 0;
        goto cleanup;
    }

1544
    if (virFileMakePath(driver->pidDir) < 0) {
1545
        virReportSystemError(errno,
1546
                             _("cannot create directory %s"),
1547
                             driver->pidDir);
1548
        goto cleanup;
1549 1550
    }

1551
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, def->name)))
1552
        goto cleanup;
1553

1554
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1555
        virReportSystemError(errno,
1556
                             _("cannot create directory %s"),
1557
                             driver->dnsmasqStateDir);
1558 1559 1560
        goto cleanup;
    }

1561
    dctx = dnsmasqContextNew(def->name, driver->dnsmasqStateDir);
1562 1563 1564
    if (dctx == NULL)
        goto cleanup;

1565
    if (networkDnsmasqCapsRefresh(driver) < 0)
1566
        goto cleanup;
1567

1568
    ret = networkBuildDhcpDaemonCommandLine(driver, obj, &cmd, pidfile, dctx);
1569 1570 1571 1572 1573
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1574
        goto cleanup;
1575

G
Guido Günther 已提交
1576
    ret = virCommandRun(cmd, NULL);
1577
    if (ret < 0)
1578 1579 1580
        goto cleanup;

    /*
1581 1582 1583 1584 1585
     * 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
1586 1587
     */

1588
    ret = virPidFileRead(driver->pidDir, def->name, &dnsmasqPid);
1589
    if (ret < 0)
1590
        goto cleanup;
1591
    virNetworkObjSetDnsmasqPid(obj, dnsmasqPid);
1592

1593
    ret = 0;
1594
 cleanup:
1595
    VIR_FREE(pidfile);
1596
    virCommandFree(cmd);
1597
    dnsmasqContextFree(dctx);
1598 1599 1600
    return ret;
}

1601

1602 1603
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1604 1605
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1606 1607 1608
 *
 *  Returns 0 on success, -1 on failure.
 */
1609
static int
1610
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
1611
                         virNetworkObjPtr obj)
1612
{
1613
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1614 1615
    int ret = -1;
    size_t i;
1616
    pid_t dnsmasqPid;
1617
    virNetworkIPDefPtr ipdef, ipv4def, ipv6def;
1618
    dnsmasqContext *dctx = NULL;
1619

G
Gene Czarcinski 已提交
1620
    /* if no IP addresses specified, nothing to do */
1621
    if (!virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1622 1623
        return 0;

1624
    /* if there's no running dnsmasq, just start it */
1625 1626
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid <= 0 || (kill(dnsmasqPid, 0) < 0))
1627
        return networkStartDhcpDaemon(driver, obj);
1628

1629 1630
    VIR_INFO("Refreshing dnsmasq for network %s", def->bridge);
    if (!(dctx = dnsmasqContextNew(def->name,
1631
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1632
        goto cleanup;
1633
    }
G
Gene Czarcinski 已提交
1634 1635 1636 1637 1638 1639

    /* 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;
1640
    for (i = 0;
1641
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
1642
         i++) {
G
Gene Czarcinski 已提交
1643 1644
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1645 1646
    }

G
Gene Czarcinski 已提交
1647
    ipv6def = NULL;
1648
    for (i = 0;
1649
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i));
1650
         i++) {
G
Gene Czarcinski 已提交
1651 1652
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1653 1654
    }

G
Gene Czarcinski 已提交
1655
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1656
        goto cleanup;
G
Gene Czarcinski 已提交
1657 1658

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1659
        goto cleanup;
1660

1661
    if (networkBuildDnsmasqHostsList(dctx, &def->dns) < 0)
J
Ján Tomko 已提交
1662
        goto cleanup;
1663 1664

    if ((ret = dnsmasqSave(dctx)) < 0)
1665
        goto cleanup;
1666

1667 1668
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    ret = kill(dnsmasqPid, SIGHUP);
1669
 cleanup:
1670 1671 1672 1673
    dnsmasqContextFree(dctx);
    return ret;
}

1674

1675 1676 1677 1678 1679 1680 1681 1682
/* 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
1683
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
1684
                         virNetworkObjPtr obj)
1685
{
1686
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1687 1688
    pid_t dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);

1689
    /* if there is a running dnsmasq, kill it */
1690
    if (dnsmasqPid > 0) {
1691
        networkKillDaemon(dnsmasqPid, "dnsmasq", def->name);
1692
        virNetworkObjSetDnsmasqPid(obj, -1);
1693
    }
1694
    /* now start dnsmasq if it should be started */
1695
    return networkStartDhcpDaemon(driver, obj);
1696 1697
}

1698

G
Gene Czarcinski 已提交
1699 1700 1701 1702 1703 1704
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";

1705
static int
1706
networkRadvdConfContents(virNetworkObjPtr obj,
1707
                         char **configstr)
1708
{
1709
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
E
Eric Blake 已提交
1710
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1711 1712
    int ret = -1;
    size_t i;
1713
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
1714
    bool v6present = false, dhcp6 = false;
1715 1716

    *configstr = NULL;
1717

G
Gene Czarcinski 已提交
1718
    /* Check if DHCPv6 is needed */
1719
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i)); i++) {
G
Gene Czarcinski 已提交
1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732
        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;
    }

1733 1734 1735
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1736
    virBufferAsprintf(&configbuf, "interface %s\n"
1737 1738
                      "{\n"
                      "  AdvSendAdvert on;\n"
1739
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1740 1741
                      "  AdvManagedFlag %s;\n"
                      "%s",
1742
                      def->bridge,
G
Gene Czarcinski 已提交
1743 1744
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1745 1746

    /* add a section for each IPv6 address in the config */
1747
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i)); i++) {
1748 1749 1750
        int prefix;
        char *netaddr;

1751
        prefix = virNetworkIPDefPrefix(ipdef);
1752
        if (prefix < 0) {
1753 1754
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
1755
                           def->bridge);
1756 1757
            goto cleanup;
        }
1758
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1759
            goto cleanup;
1760
        virBufferAsprintf(&configbuf,
1761
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1762 1763 1764
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1765 1766 1767
        VIR_FREE(netaddr);
    }

1768
    virBufferAddLit(&configbuf, "};\n");
1769

1770
    if (virBufferCheckError(&configbuf) < 0)
1771
        goto cleanup;
1772

1773 1774
    *configstr = virBufferContentAndReset(&configbuf);

1775
    ret = 0;
1776
 cleanup:
1777 1778 1779 1780
    virBufferFreeAndReset(&configbuf);
    return ret;
}

1781

1782
/* write file and return its name (which must be freed by caller) */
1783
static int
1784
networkRadvdConfWrite(virNetworkDriverStatePtr driver,
1785
                      virNetworkObjPtr obj,
1786
                      char **configFile)
1787
{
1788
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1789 1790 1791 1792 1793 1794 1795 1796 1797
    int ret = -1;
    char *configStr = NULL;
    char *myConfigFile = NULL;

    if (!configFile)
        configFile = &myConfigFile;

    *configFile = NULL;

1798
    if (networkRadvdConfContents(obj, &configStr) < 0)
1799 1800 1801 1802
        goto cleanup;

    if (!configStr) {
        ret = 0;
1803 1804 1805 1806
        goto cleanup;
    }

    /* construct the filename */
1807
    if (!(*configFile = networkRadvdConfigFileName(driver, def->name)))
1808 1809
        goto cleanup;
    /* write the file */
1810
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1811 1812
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1813 1814 1815 1816 1817
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1818
 cleanup:
1819 1820 1821 1822 1823
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

1824

1825
static int
1826
networkStartRadvd(virNetworkDriverStatePtr driver,
1827
                  virNetworkObjPtr obj)
1828
{
1829
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1830
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1831
    pid_t radvdPid;
1832 1833 1834 1835 1836 1837
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

1838
    virNetworkObjSetRadvdPid(obj, -1);
1839

G
Gene Czarcinski 已提交
1840
    /* Is dnsmasq handling RA? */
1841
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
G
Gene Czarcinski 已提交
1842 1843 1844 1845
        ret = 0;
        goto cleanup;
    }

1846
    if (!virNetworkDefGetIPByIndex(def, AF_INET6, 0)) {
1847 1848 1849 1850 1851 1852 1853 1854 1855 1856
        /* 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);
1857 1858 1859
        goto cleanup;
    }

1860
    if (virFileMakePath(driver->pidDir) < 0) {
1861 1862
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1863
                             driver->pidDir);
1864 1865
        goto cleanup;
    }
1866
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1867 1868
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1869
                             driver->radvdStateDir);
1870 1871 1872 1873
        goto cleanup;
    }

    /* construct pidfile name */
1874
    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
1875
        goto cleanup;
1876
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1877 1878
        goto cleanup;

1879
    if (networkRadvdConfWrite(driver, obj, &configfile) < 0)
1880 1881
        goto cleanup;

1882 1883 1884 1885
    /* 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
1886
     * virPidFileRead() below will fail if we use them).
1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901
     * 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;

1902
    if (virPidFileRead(driver->pidDir, radvdpidbase, &radvdPid) < 0)
1903
        goto cleanup;
1904
    virNetworkObjSetRadvdPid(obj, radvdPid);
1905 1906

    ret = 0;
1907
 cleanup:
1908
    virObjectUnref(dnsmasq_caps);
1909 1910 1911 1912 1913 1914 1915
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1916

1917
static int
1918
networkRefreshRadvd(virNetworkDriverStatePtr driver,
1919
                    virNetworkObjPtr obj)
1920
{
1921
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1922
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
G
Gene Czarcinski 已提交
1923
    char *radvdpidbase;
1924
    pid_t radvdPid;
G
Gene Czarcinski 已提交
1925 1926

    /* Is dnsmasq handling RA? */
1927 1928
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
        virObjectUnref(dnsmasq_caps);
1929 1930
        radvdPid = virNetworkObjGetRadvdPid(obj);
        if (radvdPid <= 0)
G
Gene Czarcinski 已提交
1931 1932
            return 0;
        /* radvd should not be running but in case it is */
1933 1934
        if ((networkKillDaemon(radvdPid, "radvd", def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(def->name))
G
Gene Czarcinski 已提交
1935
             != NULL)) {
1936
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1937 1938
            VIR_FREE(radvdpidbase);
        }
1939
        virNetworkObjSetRadvdPid(obj, -1);
G
Gene Czarcinski 已提交
1940 1941
        return 0;
    }
1942
    virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1943

1944
    /* if there's no running radvd, just start it */
1945 1946
    radvdPid = virNetworkObjGetRadvdPid(obj);
    if (radvdPid <= 0 || (kill(radvdPid, 0) < 0))
1947
        return networkStartRadvd(driver, obj);
1948

1949
    if (!virNetworkDefGetIPByIndex(def, AF_INET6, 0)) {
1950 1951 1952 1953
        /* no IPv6 addresses, so we don't need to run radvd */
        return 0;
    }

1954
    if (networkRadvdConfWrite(driver, obj, NULL) < 0)
1955 1956
        return -1;

1957
    return kill(radvdPid, SIGHUP);
1958 1959
}

1960

1961 1962
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1963
static int
1964
networkRestartRadvd(virNetworkObjPtr obj)
1965
{
1966
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1967
    char *radvdpidbase;
1968
    pid_t radvdPid = virNeworkObjGetRadvdPid(obj);
1969 1970

    /* if there is a running radvd, kill it */
1971
    if (radvdPid > 0) {
1972 1973 1974 1975
        /* 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).
         */
1976 1977
        if ((networkKillDaemon(radvdPid, "radvd", def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(def->name))
1978
             != NULL)) {
1979
            virPidFileDelete(driver->pidDir, radvdpidbase);
1980 1981
            VIR_FREE(radvdpidbase);
        }
1982
        virNetworkObjSetRadvdPid(obj, -1);
1983 1984
    }
    /* now start radvd if it should be started */
1985
    return networkStartRadvd(obj);
1986 1987 1988
}
#endif /* #if 0 */

1989

1990
static int
1991
networkRefreshDaemonsHelper(virNetworkObjPtr obj,
1992
                            void *opaque)
1993
{
1994
    virNetworkDefPtr def;
1995
    virNetworkDriverStatePtr driver = opaque;
1996

1997
    virObjectLock(obj);
1998
    def = virNetworkObjGetDef(obj);
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026
    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;
        }
2027
    }
2028 2029

 cleanup:
2030
    virObjectUnlock(obj);
2031 2032 2033
    return 0;
}

2034

2035 2036 2037 2038
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
2039
networkRefreshDaemons(virNetworkDriverStatePtr driver)
2040 2041
{
    VIR_INFO("Refreshing network daemons");
2042 2043
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
2044
                             driver);
2045
}
2046

2047

2048
static int
2049
networkReloadFirewallRulesHelper(virNetworkObjPtr obj,
2050 2051
                                 void *opaque ATTRIBUTE_UNUSED)
{
2052
    virNetworkDefPtr def;
2053

2054
    virObjectLock(obj);
2055
    def = virNetworkObjGetDef(obj);
2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081
    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;
2082 2083
        }
    }
2084 2085

 cleanup:
2086
    virObjectUnlock(obj);
2087
    return 0;
2088 2089
}

2090

2091
static void
2092
networkReloadFirewallRules(virNetworkDriverStatePtr driver, bool startup)
2093
{
2094
    VIR_INFO("Reloading iptables rules");
2095 2096
    if (networkPreReloadFirewallRules(startup) < 0)
        return;
2097 2098 2099
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
2100
    networkPostReloadFirewallRules(startup);
2101 2102
}

2103

2104
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
2105
static int
2106 2107
networkEnableIPForwarding(bool enableIPv4,
                          bool enableIPv6)
2108
{
2109
    int ret = 0;
2110 2111 2112 2113
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
2114
                           &enabled, sizeof(enabled));
2115 2116
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
2117
                           &enabled, sizeof(enabled));
2118
#else
2119
    if (enableIPv4)
2120
        ret = virFileWriteStr(SYSCTL_PATH "/net/ipv4/ip_forward", "1\n", 0);
2121
    if (enableIPv6 && ret == 0)
2122 2123
        ret = virFileWriteStr(SYSCTL_PATH "/net/ipv6/conf/all/forwarding", "1\n", 0);

2124
#endif
2125
    return ret;
2126 2127
}

2128

2129
static int
2130
networkSetIPv6Sysctls(virNetworkObjPtr obj)
2131
{
2132
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2133 2134
    char *field = NULL;
    int ret = -1;
2135
    bool enableIPv6 = !!virNetworkDefGetIPByIndex(def, AF_INET6, 0);
2136

2137 2138 2139 2140 2141
    /* 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",
2142
                    def->bridge) < 0)
2143
       goto cleanup;
2144

2145 2146
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
2147
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
2148
                      def->bridge);
2149 2150 2151
        ret = 0;
        goto cleanup;
    }
2152

2153 2154 2155
    if (virFileWriteStr(field, enableIPv6 ? "0" : "1", 0) < 0) {
        virReportSystemError(errno,
                             _("cannot write to %s to enable/disable IPv6 "
2156
                               "on bridge %s"), field, def->bridge);
2157
        goto cleanup;
2158
    }
2159
    VIR_FREE(field);
2160

2161 2162
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
2163 2164 2165 2166 2167 2168
     */

    /* 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",
2169
                    def->bridge) < 0)
2170 2171
        goto cleanup;

2172
    if (virFileWriteStr(field, "0", 0) < 0) {
2173
        virReportSystemError(errno,
2174 2175 2176 2177 2178
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

2179 2180 2181 2182
    /* 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",
2183
                    def->bridge) < 0)
2184 2185
        goto cleanup;

2186
    if (virFileWriteStr(field, "0", 0) < 0) {
2187
        virReportSystemError(errno,
2188
                             _("cannot disable %s"), field);
2189 2190 2191 2192
        goto cleanup;
    }

    ret = 0;
2193
 cleanup:
2194 2195 2196 2197
    VIR_FREE(field);
    return ret;
}

2198

2199
/* add an IP address to a bridge */
2200
static int
2201
networkAddAddrToBridge(virNetworkObjPtr obj,
2202
                       virNetworkIPDefPtr ipdef)
2203
{
2204
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2205
    int prefix = virNetworkIPDefPrefix(ipdef);
2206 2207

    if (prefix < 0) {
2208 2209
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
2210
                       def->bridge);
2211 2212 2213
        return -1;
    }

2214
    if (virNetDevIPAddrAdd(def->bridge, &ipdef->address, NULL, prefix) < 0)
2215 2216 2217 2218 2219
        return -1;

    return 0;
}

2220 2221

static int
2222
networkStartHandleMACTableManagerMode(virNetworkObjPtr obj,
2223 2224
                                      const char *macTapIfName)
{
2225 2226
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
    const char *brname = def->bridge;
2227 2228

    if (brname &&
2229
        def->macTableManager == VIR_NETWORK_BRIDGE_MAC_TABLE_MANAGER_LIBVIRT) {
2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242
        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;
}


2243 2244
/* add an IP (static) route to a bridge */
static int
2245
networkAddRouteToBridge(virNetworkObjPtr obj,
2246
                        virNetDevIPRoutePtr routedef)
2247
{
2248
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2249 2250 2251 2252
    int prefix = virNetDevIPRouteGetPrefix(routedef);
    unsigned int metric = virNetDevIPRouteGetMetric(routedef);
    virSocketAddrPtr addr = virNetDevIPRouteGetAddress(routedef);
    virSocketAddrPtr gateway = virNetDevIPRouteGetGateway(routedef);
2253 2254 2255 2256 2257

    if (prefix < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' has an invalid netmask "
                         "or IP address in route definition"),
2258
                       def->name);
2259 2260 2261
        return -1;
    }

2262
    if (virNetDevIPRouteAdd(def->bridge, addr, prefix, gateway, metric) < 0)
2263
        return -1;
2264

2265 2266 2267
    return 0;
}

2268
static int
2269
networkWaitDadFinish(virNetworkObjPtr obj)
2270
{
2271
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2272
    virNetworkIPDefPtr ipdef;
2273 2274 2275 2276
    virSocketAddrPtr *addrs = NULL, addr = NULL;
    size_t naddrs = 0;
    int ret = -1;

2277
    VIR_DEBUG("Begin waiting for IPv6 DAD on network %s", def->name);
2278

2279
    while ((ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, naddrs))) {
2280 2281 2282 2283 2284
        addr = &ipdef->address;
        if (VIR_APPEND_ELEMENT_COPY(addrs, naddrs, addr) < 0)
            goto cleanup;
    }

2285
    ret = (naddrs == 0) ? 0 : virNetDevIPWaitDadFinish(addrs, naddrs);
2286 2287 2288 2289

 cleanup:
    VIR_FREE(addrs);
    VIR_DEBUG("Finished waiting for IPv6 DAD on network %s with status %d",
2290
              def->name, ret);
2291 2292 2293
    return ret;
}

2294

2295
static int
2296
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
2297
                           virNetworkObjPtr obj)
2298
{
2299
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2300
    size_t i;
2301
    bool v4present = false, v6present = false;
2302
    virErrorPtr save_err = NULL;
2303
    virNetworkIPDefPtr ipdef;
2304
    virNetDevIPRoutePtr routedef;
2305
    char *macTapIfName = NULL;
2306
    virMacMapPtr macmap;
M
Michal Privoznik 已提交
2307
    char *macMapFile = NULL;
2308
    int tapfd = -1;
2309
    pid_t dnsmasqPid;
2310

2311
    /* Check to see if any network IP collides with an existing route */
2312
    if (networkCheckRouteCollision(def) < 0)
2313 2314
        return -1;

2315
    /* Create and configure the bridge device */
2316
    if (!def->bridge) {
2317 2318 2319 2320 2321 2322 2323 2324 2325 2326
        /* 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"),
2327
                       def->name);
2328 2329
        return -1;
    }
2330
    if (virNetDevBridgeCreate(def->bridge) < 0)
2331 2332
        return -1;

2333
    if (def->mac_specified) {
2334 2335 2336 2337 2338 2339
        /* 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.
         */
2340
        macTapIfName = networkBridgeDummyNicName(def->bridge);
2341
        if (!macTapIfName)
2342
            goto err0;
2343
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2344 2345
        if (virNetDevTapCreateInBridgePort(def->bridge,
                                           &macTapIfName, &def->mac,
2346
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2347
                                           NULL, def->mtu, NULL,
2348 2349 2350
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2351 2352 2353 2354 2355
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2356
    if (!(macMapFile = virMacMapFileName(driver->dnsmasqStateDir,
2357
                                         def->bridge)) ||
2358
        !(macmap = virMacMapNew(macMapFile)))
M
Michal Privoznik 已提交
2359 2360
        goto err1;

2361
    virNetworkObjSetMacMap(obj, macmap);
2362
    macmap = NULL;
2363

2364
    /* Set bridge options */
2365 2366 2367 2368

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2369
    if (virNetDevBridgeSetSTPDelay(def->bridge, def->delay * 1000) < 0)
2370
        goto err1;
2371

2372
    if (virNetDevBridgeSetSTP(def->bridge, def->stp ? true : false) < 0)
2373
        goto err1;
2374

2375 2376 2377
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
2378
    if (networkSetIPv6Sysctls(obj) < 0)
2379
        goto err1;
2380

2381
    /* Add "once per network" rules */
2382 2383
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN &&
        networkAddFirewallRules(def) < 0)
2384 2385
        goto err1;

2386
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i)); i++) {
2387
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2388
            v4present = true;
2389
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2390
            v6present = true;
2391

2392
        /* Add the IP address/netmask to the bridge */
2393
        if (networkAddAddrToBridge(obj, ipdef) < 0)
2394
            goto err2;
2395 2396
    }

2397
    if (networkStartHandleMACTableManagerMode(obj, macTapIfName) < 0)
2398 2399
        goto err2;

2400
    /* Bring up the bridge interface */
2401
    if (virNetDevSetOnline(def->bridge, 1) < 0)
2402
        goto err2;
2403

2404
    for (i = 0; i < def->nroutes; i++) {
2405 2406
        virSocketAddrPtr gateway = NULL;

2407
        routedef = def->routes[i];
2408
        gateway = virNetDevIPRouteGetGateway(routedef);
2409

2410 2411 2412
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2413
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2414
            if (networkAddRouteToBridge(obj, routedef) < 0) {
2415 2416 2417 2418 2419 2420
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2421
    /* If forward.type != NONE, turn on global IP forwarding */
2422
    if (def->forward.type != VIR_NETWORK_FORWARD_NONE) {
2423
        if (v6present && !virNetDevIPCheckIPv6Forwarding())
2424 2425 2426 2427 2428 2429 2430
            goto err3; /* Precise error message already provided */

        if (networkEnableIPForwarding(v4present, v6present) < 0) {
            virReportSystemError(errno, "%s",
                                 _("failed to enable IP forwarding"));
            goto err3;
        }
2431 2432
    }

2433

2434
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2435
    if ((v4present || v6present) &&
2436
        networkStartDhcpDaemon(driver, obj) < 0)
2437
        goto err3;
2438

2439
    /* start radvd if there are any ipv6 addresses */
2440
    if (v6present && networkStartRadvd(driver, obj) < 0)
2441 2442
        goto err4;

2443 2444 2445
    /* dnsmasq does not wait for DAD to complete before daemonizing,
     * so we need to wait for it ourselves.
     */
2446
    if (v6present && networkWaitDadFinish(obj) < 0)
2447 2448 2449 2450
        goto err4;

    /* DAD has finished, dnsmasq is now bound to the
     * bridge's IPv6 address, so we can set the dummy tun down.
2451 2452 2453 2454 2455 2456 2457
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
            goto err4;
        VIR_FORCE_CLOSE(tapfd);
    }

2458
    if (virNetDevBandwidthSet(def->bridge, def->bandwidth, true, true) < 0)
2459 2460
        goto err5;

2461
    VIR_FREE(macTapIfName);
M
Michal Privoznik 已提交
2462
    VIR_FREE(macMapFile);
2463 2464 2465

    return 0;

2466
 err5:
2467 2468
    if (def->bandwidth)
       virNetDevBandwidthClear(def->bridge);
2469

2470 2471 2472 2473
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2474 2475 2476 2477
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid > 0) {
        kill(dnsmasqPid, SIGTERM);
        virNetworkObjSetDnsmasqPid(obj, -1);
2478 2479
    }

2480 2481 2482
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2483
    ignore_value(virNetDevSetOnline(def->bridge, 0));
2484

2485 2486 2487
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2488 2489
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(def);
2490 2491

 err1:
2492 2493 2494
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2495
    if (macTapIfName) {
2496
        VIR_FORCE_CLOSE(tapfd);
2497
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2498 2499
        VIR_FREE(macTapIfName);
    }
2500
    virNetworkObjUnrefMacMap(obj);
M
Michal Privoznik 已提交
2501
    VIR_FREE(macMapFile);
2502 2503

 err0:
2504 2505
    if (!save_err)
        save_err = virSaveLastError();
2506
    ignore_value(virNetDevBridgeDelete(def->bridge));
2507

2508 2509 2510 2511
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2512
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2513 2514 2515
    return -1;
}

2516

2517 2518
static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
2519
                              virNetworkObjPtr obj)
2520
{
2521
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2522 2523 2524
    pid_t radvdPid;
    pid_t dnsmasqPid;

2525 2526
    if (def->bandwidth)
        virNetDevBandwidthClear(def->bridge);
2527

2528
    virNetworkObjUnrefMacMap(obj);
M
Michal Privoznik 已提交
2529

2530 2531
    radvdPid = virNetworkObjGetRadvdPid(obj);
    if (radvdPid > 0) {
2532 2533
        char *radvdpidbase;

2534
        kill(radvdPid, SIGTERM);
2535
        /* attempt to delete the pidfile we created */
2536
        if ((radvdpidbase = networkRadvdPidfileBasename(def->name))) {
2537
            virPidFileDelete(driver->pidDir, radvdpidbase);
2538 2539 2540 2541
            VIR_FREE(radvdpidbase);
        }
    }

2542 2543 2544
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid > 0)
        kill(dnsmasqPid, SIGTERM);
2545

2546 2547
    if (def->mac_specified) {
        char *macTapIfName = networkBridgeDummyNicName(def->bridge);
2548
        if (macTapIfName) {
2549
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2550 2551 2552 2553
            VIR_FREE(macTapIfName);
        }
    }

2554
    ignore_value(virNetDevSetOnline(def->bridge, 0));
2555

2556 2557
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(def);
2558

2559
    ignore_value(virNetDevBridgeDelete(def->bridge));
2560

2561
    /* See if its still alive and really really kill it */
2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572
    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);
2573

2574 2575 2576
    return 0;
}

2577

2578
static int
2579
networkStartNetworkBridge(virNetworkObjPtr obj)
2580 2581 2582 2583 2584
{
    /* 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.
     */
2585
    return networkStartHandleMACTableManagerMode(obj, NULL);
2586 2587
}

2588

2589
static int
2590
networkShutdownNetworkBridge(virNetworkObjPtr obj ATTRIBUTE_UNUSED)
2591 2592 2593 2594 2595 2596 2597 2598 2599
{
    /* 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;
}


2600 2601 2602 2603 2604 2605 2606 2607 2608
/* 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;
2609
    unsigned int maxVirtFns = 0;
2610 2611 2612 2613 2614 2615
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2616 2617 2618
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2619 2620
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev, &vfNames,
                                      &virtFns, &numVirtFns, &maxVirtFns)) < 0) {
2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635
        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];

2636
        switch ((virNetworkForwardType) netdef->forward.type) {
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 2663 2664 2665 2666
        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:
2667
        case VIR_NETWORK_FORWARD_OPEN:
2668 2669
            /* by definition these will never be encountered here */
            break;
2670 2671 2672 2673 2674

        case VIR_NETWORK_FORWARD_LAST:
        default:
            virReportEnumRangeError(virNetworkForwardType, netdef->forward.type);
            goto cleanup;
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 2708 2709 2710 2711
        }
    }

    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;
}


2712
static int
2713
networkStartNetworkExternal(virNetworkObjPtr obj)
2714 2715
{
    /* put anything here that needs to be done each time a network of
2716
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2717 2718 2719
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2720
    return networkCreateInterfacePool(virNetworkObjGetDef(obj));
2721 2722
}

2723 2724

static int
2725
networkShutdownNetworkExternal(virNetworkObjPtr obj ATTRIBUTE_UNUSED)
2726 2727
{
    /* put anything here that needs to be done each time a network of
2728
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2729 2730 2731 2732 2733 2734
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

2735

2736
static int
2737
networkStartNetwork(virNetworkDriverStatePtr driver,
2738
                    virNetworkObjPtr obj)
2739
{
2740
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2741 2742
    int ret = -1;

2743
    VIR_DEBUG("driver=%p, network=%p", driver, obj);
2744

2745
    if (virNetworkObjIsActive(obj)) {
2746 2747
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2748
        return ret;
2749 2750
    }

2751 2752 2753
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2754
    if (virNetworkObjSetDefTransient(obj, true) < 0)
2755
        goto cleanup;
2756

2757 2758
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2759
    if (networkRunHook(obj, NULL, NULL,
2760 2761 2762 2763
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2764
    switch ((virNetworkForwardType) def->forward.type) {
2765 2766 2767 2768

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2769
    case VIR_NETWORK_FORWARD_OPEN:
2770
        if (networkStartNetworkVirtual(driver, obj) < 0)
2771
            goto cleanup;
2772 2773 2774
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2775
        if (def->bridge) {
2776
            if (networkStartNetworkBridge(obj) < 0)
2777 2778 2779 2780 2781 2782 2783
                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).
         */
2784 2785
        ATTRIBUTE_FALLTHROUGH;

2786 2787 2788
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2789
    case VIR_NETWORK_FORWARD_HOSTDEV:
2790
        if (networkStartNetworkExternal(obj) < 0)
2791
            goto cleanup;
2792
        break;
2793 2794 2795 2796 2797

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        goto cleanup;
2798 2799
    }

2800
    /* finally we can call the 'started' hook script if any */
2801
    if (networkRunHook(obj, NULL, NULL,
2802 2803 2804 2805
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2806 2807 2808
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2809
    VIR_DEBUG("Writing network status to disk");
2810
    if (virNetworkObjSaveStatus(driver->stateDir, obj) < 0)
2811
        goto cleanup;
2812

2813
    virNetworkObjSetActive(obj, true);
2814
    VIR_INFO("Network '%s' started up", def->name);
2815
    ret = 0;
2816

2817
 cleanup:
2818
    if (ret < 0) {
2819
        virNetworkObjUnsetDefTransient(obj);
2820 2821
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2822
        networkShutdownNetwork(driver, obj);
2823 2824 2825 2826 2827 2828 2829
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2830

2831 2832
static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
2833
                       virNetworkObjPtr obj)
2834
{
2835
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2836 2837 2838
    int ret = 0;
    char *stateFile;

2839
    VIR_INFO("Shutting down network '%s'", def->name);
2840

2841
    if (!virNetworkObjIsActive(obj))
2842 2843
        return 0;

2844
    stateFile = virNetworkConfigFile(driver->stateDir, def->name);
2845 2846 2847 2848 2849 2850
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2851
    switch ((virNetworkForwardType) def->forward.type) {
2852 2853 2854 2855

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2856
    case VIR_NETWORK_FORWARD_OPEN:
2857
        ret = networkShutdownNetworkVirtual(driver, obj);
2858 2859 2860
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2861
        if (def->bridge) {
2862
            ret = networkShutdownNetworkBridge(obj);
2863 2864 2865 2866 2867 2868
            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).
         */
2869 2870
        ATTRIBUTE_FALLTHROUGH;

2871 2872 2873
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2874
    case VIR_NETWORK_FORWARD_HOSTDEV:
2875
        ret = networkShutdownNetworkExternal(obj);
2876
        break;
2877 2878 2879 2880 2881

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        return -1;
2882 2883
    }

2884
    /* now that we know it's stopped call the hook if present */
2885
    networkRunHook(obj, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2886 2887
                   VIR_HOOK_SUBOP_END);

2888
    virNetworkObjSetActive(obj, false);
2889
    virNetworkObjUnsetDefTransient(obj);
2890
    return ret;
2891 2892 2893
}


2894 2895 2896
static virNetworkPtr
networkLookupByUUID(virConnectPtr conn,
                    const unsigned char *uuid)
2897
{
2898
    virNetworkDriverStatePtr driver = networkGetDriver();
2899
    virNetworkObjPtr obj;
2900
    virNetworkDefPtr def;
2901
    virNetworkPtr net = NULL;
2902

2903 2904
    obj = virNetworkObjFindByUUID(driver->networks, uuid);
    if (!obj) {
2905 2906
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2907
        virReportError(VIR_ERR_NO_NETWORK,
2908 2909
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2910
        goto cleanup;
2911
    }
2912
    def = virNetworkObjGetDef(obj);
2913

2914
    if (virNetworkLookupByUUIDEnsureACL(conn, def) < 0)
2915 2916
        goto cleanup;

2917
    net = virGetNetwork(conn, def->name, def->uuid);
2918

2919
 cleanup:
2920 2921
    virNetworkObjEndAPI(&obj);
    return net;
2922 2923
}

2924 2925 2926 2927

static virNetworkPtr
networkLookupByName(virConnectPtr conn,
                    const char *name)
2928
{
2929
    virNetworkDriverStatePtr driver = networkGetDriver();
2930
    virNetworkObjPtr obj;
2931
    virNetworkDefPtr def;
2932
    virNetworkPtr net = NULL;
2933

2934 2935
    obj = virNetworkObjFindByName(driver->networks, name);
    if (!obj) {
2936 2937
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2938
        goto cleanup;
2939
    }
2940
    def = virNetworkObjGetDef(obj);
2941

2942
    if (virNetworkLookupByNameEnsureACL(conn, def) < 0)
2943 2944
        goto cleanup;

2945
    net = virGetNetwork(conn, def->name, def->uuid);
2946

2947
 cleanup:
2948 2949
    virNetworkObjEndAPI(&obj);
    return net;
2950 2951
}

2952 2953 2954

static int
networkConnectNumOfNetworks(virConnectPtr conn)
2955
{
2956
    virNetworkDriverStatePtr driver = networkGetDriver();
2957
    int nactive;
2958

2959 2960 2961
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2962 2963 2964 2965
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
2966

2967 2968 2969
    return nactive;
}

2970 2971 2972 2973

static int
networkConnectListNetworks(virConnectPtr conn,
                           char **const names,
2974
                           int maxnames)
2975 2976
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2977
    int got = 0;
2978

2979 2980 2981
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2982
    got = virNetworkObjListGetNames(driver->networks,
2983
                                    true, names, maxnames,
2984 2985
                                    virConnectListNetworksCheckACL,
                                    conn);
2986

2987 2988 2989
    return got;
}

2990 2991 2992

static int
networkConnectNumOfDefinedNetworks(virConnectPtr conn)
2993
{
2994
    virNetworkDriverStatePtr driver = networkGetDriver();
2995
    int ninactive = 0;
2996

2997 2998 2999
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

3000 3001 3002 3003
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
3004

3005 3006 3007
    return ninactive;
}

3008 3009 3010 3011

static int
networkConnectListDefinedNetworks(virConnectPtr conn,
                                  char **const names,
3012
                                  int maxnames)
3013 3014
{
    virNetworkDriverStatePtr driver = networkGetDriver();
3015
    int got = 0;
3016

3017 3018 3019
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

3020
    got = virNetworkObjListGetNames(driver->networks,
3021
                                    false, names, maxnames,
3022 3023
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
3024 3025 3026
    return got;
}

3027

3028
static int
3029 3030 3031
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
3032
{
3033
    virNetworkDriverStatePtr driver = networkGetDriver();
3034 3035 3036 3037
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

3038 3039 3040
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

3041
    ret = virNetworkObjListExport(conn, driver->networks, nets,
3042 3043
                                  virConnectListAllNetworksCheckACL,
                                  flags);
3044

3045
 cleanup:
3046 3047
    return ret;
}
3048

3049

3050 3051 3052 3053 3054 3055 3056 3057
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
3058
    virNetworkDriverStatePtr driver = networkGetDriver();
3059 3060 3061 3062 3063 3064
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
3065
                                       net, eventID, callback,
3066 3067 3068
                                       opaque, freecb, &ret) < 0)
        ret = -1;

3069
 cleanup:
3070 3071 3072
    return ret;
}

3073

3074 3075 3076 3077
static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
3078
    virNetworkDriverStatePtr driver = networkGetDriver();
3079 3080 3081 3082 3083
    int ret = -1;

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

3084 3085
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
3086
                                        callbackID, true) < 0)
3087 3088 3089
        goto cleanup;

    ret = 0;
3090

3091
 cleanup:
3092 3093 3094
    return ret;
}

3095 3096 3097

static int
networkIsActive(virNetworkPtr net)
3098 3099 3100 3101
{
    virNetworkObjPtr obj;
    int ret = -1;

3102 3103
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3104

3105
    if (virNetworkIsActiveEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
3106 3107
        goto cleanup;

3108 3109
    ret = virNetworkObjIsActive(obj);

3110
 cleanup:
3111
    virNetworkObjEndAPI(&obj);
3112 3113 3114
    return ret;
}

3115 3116 3117

static int
networkIsPersistent(virNetworkPtr net)
3118 3119 3120 3121
{
    virNetworkObjPtr obj;
    int ret = -1;

3122 3123
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3124

3125
    if (virNetworkIsPersistentEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
3126 3127
        goto cleanup;

3128
    ret = virNetworkObjIsPersistent(obj);
3129

3130
 cleanup:
3131
    virNetworkObjEndAPI(&obj);
3132 3133 3134 3135
    return ret;
}


3136 3137
/*
 * networkFindUnusedBridgeName() - try to find a bridge name that is
3138 3139 3140
 * 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.
3141 3142 3143 3144 3145 3146 3147
 */
static int
networkFindUnusedBridgeName(virNetworkObjListPtr nets,
                            virNetworkDefPtr def)
{
    int ret = -1, id = 0;
    char *newname = NULL;
3148 3149 3150 3151 3152
    const char *templ = "virbr%d";
    const char *p;

    if (def->bridge &&
        (p = strchr(def->bridge, '%')) == strrchr(def->bridge, '%') &&
3153
        p && p[1] == 'd')
3154
        templ = def->bridge;
3155 3156 3157 3158

    do {
        if (virAsprintf(&newname, templ, id) < 0)
            goto cleanup;
3159 3160 3161 3162 3163
        /* 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).
         */
3164
        if (!(virNetworkObjBridgeInUse(nets, newname, def->name) ||
3165
              virNetDevExists(newname) == 1)) {
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 3195 3196 3197 3198
            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")) {
3199
        if (virNetworkObjBridgeInUse(nets, def->bridge, def->name)) {
3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216
            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;
}


3217
static int
3218
networkValidate(virNetworkDriverStatePtr driver,
3219
                virNetworkDefPtr def)
3220
{
3221
    size_t i, j;
3222 3223
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
3224
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
3225
    bool ipv4def = false, ipv6def = false;
3226
    bool bandwidthAllowed = true;
3227
    bool usesInterface = false, usesAddress = false;
3228

3229 3230 3231
    if (virXMLCheckIllegalChars("name", def->name, "\n") < 0)
        return -1;

3232 3233 3234
    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
3235 3236 3237 3238 3239
    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:
3240 3241 3242 3243
        /* 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)
3244 3245 3246
            return -1;

        virNetworkSetBridgeMacAddr(def);
3247 3248 3249 3250 3251 3252 3253
        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:
3254
        /* They are also the only types that currently support setting
3255 3256
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
3257
         */
3258 3259 3260 3261 3262 3263 3264 3265
        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;
        }
3266
        if (virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0)) {
3267 3268 3269 3270
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3271
                           virNetworkForwardTypeToString(def->forward.type));
3272 3273
            return -1;
        }
3274
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
3275 3276 3277 3278
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3279
                           virNetworkForwardTypeToString(def->forward.type));
3280 3281 3282 3283 3284 3285 3286
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3287
                           virNetworkForwardTypeToString(def->forward.type));
3288 3289
            return -1;
        }
3290 3291 3292 3293 3294 3295 3296 3297
        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;
        }
3298
        bandwidthAllowed = false;
3299 3300 3301 3302 3303 3304
        break;

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        return -1;
3305 3306
    }

3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319
    /* 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++) {
3320 3321 3322
        virNetworkForwardIfDefPtr iface = &def->forward.ifs[i];
        char *sysfs_path = NULL;

3323
        switch ((virNetworkForwardHostdevDeviceType)iface->type) {
3324 3325
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV:
            usesInterface = true;
3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336

            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;
            }
3337
            break;
3338 3339

        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI: {
3340
            usesAddress = true;
3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365

            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);
3366
            break;
3367 3368
        }

3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381
        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
3382 3383 3384
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
3385
    for (i = 0;
3386
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
3387
         i++) {
G
Gene Czarcinski 已提交
3388 3389 3390 3391 3392
        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 -- "
3393 3394
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411
                    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;
                }
3412 3413 3414
            }
        }
    }
3415 3416 3417 3418 3419 3420

    /* 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.
     */

3421 3422 3423
    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
3424 3425
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
3426
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH));
3427 3428

    vlanUsed = def->vlan.nTags > 0;
3429 3430
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3431 3432 3433 3434 3435
            /* 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.
             */
3436
            if (def->portGroups[i].virtPortProfile) {
3437
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3438
                    def->portGroups[i].virtPortProfile->virtPortType
3439 3440 3441 3442 3443 3444 3445
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3446
        }
3447
        if (def->portGroups[i].isDefault) {
3448 3449 3450 3451 3452
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3453
                               def->name, defaultPortGroup->name,
3454
                               def->portGroups[i].name);
3455
                return -1;
3456
            }
3457
            defaultPortGroup = &def->portGroups[i];
3458
        }
3459 3460 3461 3462 3463 3464 3465 3466 3467
        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;
            }
        }
3468 3469 3470 3471 3472 3473 3474 3475
        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;
        }
3476
    }
3477 3478 3479 3480 3481 3482 3483
    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.
         */
3484 3485 3486 3487 3488 3489
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503

    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;
            }
        }
    }
3504 3505 3506
    return 0;
}

3507 3508 3509 3510

static virNetworkPtr
networkCreateXML(virConnectPtr conn,
                 const char *xml)
3511
{
3512
    virNetworkDriverStatePtr driver = networkGetDriver();
3513
    virNetworkDefPtr newDef;
3514
    virNetworkObjPtr obj = NULL;
3515
    virNetworkDefPtr def;
3516
    virNetworkPtr net = NULL;
3517
    virObjectEventPtr event = NULL;
3518

3519
    if (!(newDef = virNetworkDefParseString(xml)))
3520
        goto cleanup;
3521

3522
    if (virNetworkCreateXMLEnsureACL(conn, newDef) < 0)
3523 3524
        goto cleanup;

3525
    if (networkValidate(driver, newDef) < 0)
J
Ján Tomko 已提交
3526
        goto cleanup;
3527

3528 3529 3530
    /* 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.
3531
     */
3532
    if (!(obj = virNetworkObjAssignDef(driver->networks, newDef,
3533 3534
                                       VIR_NETWORK_OBJ_LIST_ADD_LIVE |
                                       VIR_NETWORK_OBJ_LIST_ADD_CHECK_LIVE)))
3535
        goto cleanup;
3536 3537
    newDef = NULL;
    def = virNetworkObjGetDef(obj);
3538

3539 3540
    if (networkStartNetwork(driver, obj) < 0) {
        virNetworkObjRemoveInactive(driver->networks, obj);
3541
        goto cleanup;
3542 3543
    }

3544 3545
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3546 3547
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3548

3549 3550
    VIR_INFO("Creating network '%s'", def->name);
    net = virGetNetwork(conn, def->name, def->uuid);
3551

3552
 cleanup:
3553
    virNetworkDefFree(newDef);
3554
    virObjectEventStateQueue(driver->networkEventState, event);
3555 3556
    virNetworkObjEndAPI(&obj);
    return net;
3557 3558
}

3559 3560 3561 3562

static virNetworkPtr
networkDefineXML(virConnectPtr conn,
                 const char *xml)
3563
{
3564
    virNetworkDriverStatePtr driver = networkGetDriver();
3565
    virNetworkDefPtr def = NULL;
3566
    bool freeDef = true;
3567 3568
    virNetworkObjPtr obj = NULL;
    virNetworkPtr net = NULL;
3569
    virObjectEventPtr event = NULL;
3570

3571
    if (!(def = virNetworkDefParseString(xml)))
3572
        goto cleanup;
3573

3574 3575 3576
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3577
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3578
        goto cleanup;
3579

3580
    if (!(obj = virNetworkObjAssignDef(driver->networks, def, 0)))
J
Ján Tomko 已提交
3581
        goto cleanup;
3582

3583
    /* def was assigned to network object */
3584
    freeDef = false;
3585 3586

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3587 3588
        if (!virNetworkObjIsActive(obj)) {
            virNetworkObjRemoveInactive(driver->networks, obj);
3589 3590
            goto cleanup;
        }
3591 3592 3593 3594
        /* if network was active already, just undo new persistent
         * definition by making it transient.
         * XXX - this isn't necessarily the correct thing to do.
         */
3595
        virNetworkObjUpdateAssignDef(obj, NULL, false);
3596 3597 3598
        goto cleanup;
    }

3599
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3600 3601
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3602

3603
    VIR_INFO("Defining network '%s'", def->name);
3604
    net = virGetNetwork(conn, def->name, def->uuid);
3605

3606
 cleanup:
3607
    virObjectEventStateQueue(driver->networkEventState, event);
3608
    if (freeDef)
J
Ján Tomko 已提交
3609
        virNetworkDefFree(def);
3610 3611
    virNetworkObjEndAPI(&obj);
    return net;
3612 3613
}

3614

3615
static int
3616 3617
networkUndefine(virNetworkPtr net)
{
3618
    virNetworkDriverStatePtr driver = networkGetDriver();
3619
    virNetworkObjPtr obj;
3620
    virNetworkDefPtr def;
3621
    int ret = -1;
3622
    bool active = false;
3623
    virObjectEventPtr event = NULL;
3624

3625
    if (!(obj = networkObjFromNetwork(net)))
3626
        goto cleanup;
3627
    def = virNetworkObjGetDef(obj);
3628

3629
    if (virNetworkUndefineEnsureACL(net->conn, def) < 0)
3630 3631
        goto cleanup;

3632
    if (virNetworkObjIsActive(obj))
3633
        active = true;
3634

3635
    if (!virNetworkObjIsPersistent(obj)) {
3636 3637 3638 3639 3640
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                       _("can't undefine transient network"));
        goto cleanup;
    }

3641
    /* remove autostart link */
3642 3643
    if (virNetworkObjDeleteConfig(driver->networkConfigDir,
                                  driver->networkAutostartDir,
3644
                                  obj) < 0)
3645
        goto cleanup;
3646

3647 3648
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3649 3650
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3651

3652
    VIR_INFO("Undefining network '%s'", def->name);
3653
    if (!active) {
3654
        if (networkRemoveInactive(driver, obj) < 0)
3655
            goto cleanup;
3656 3657 3658 3659 3660
    } else {

        /* if the network still exists, it was active, and we need to make
         * it transient (by deleting the persistent def)
         */
3661
        virNetworkObjUpdateAssignDef(obj, NULL, false);
3662 3663
    }

3664
    ret = 0;
3665

3666
 cleanup:
3667
    virObjectEventStateQueue(driver->networkEventState, event);
3668
    virNetworkObjEndAPI(&obj);
3669
    return ret;
3670 3671
}

3672

3673 3674 3675 3676 3677 3678 3679 3680
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3681
    virNetworkDriverStatePtr driver = networkGetDriver();
3682
    virNetworkObjPtr obj = NULL;
3683
    virNetworkDefPtr def;
3684 3685
    int isActive, ret = -1;
    size_t i;
3686
    virNetworkIPDefPtr ipdef;
3687
    bool oldDhcpActive = false;
3688
    bool needFirewallRefresh = false;
3689

3690 3691 3692 3693 3694

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3695
    if (!(obj = networkObjFromNetwork(net)))
3696
        goto cleanup;
3697
    def = virNetworkObjGetDef(obj);
3698

3699
    if (virNetworkUpdateEnsureACL(net->conn, def, flags) < 0)
3700 3701
        goto cleanup;

3702
    /* see if we are listening for dhcp pre-modification */
3703
    for (i = 0;
3704
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
3705
         i++) {
3706 3707 3708 3709 3710 3711
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3712 3713
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3714
     */
3715
    isActive = virNetworkObjIsActive(obj);
3716 3717
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3718 3719 3720 3721 3722 3723 3724
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3725 3726 3727 3728
    if (isActive && (flags & VIR_NETWORK_UPDATE_AFFECT_LIVE)) {
        /* Take care of anything that must be done before updating the
         * live NetworkDef.
         */
3729 3730 3731 3732
        switch ((virNetworkForwardType) def->forward.type) {
        case VIR_NETWORK_FORWARD_NONE:
        case VIR_NETWORK_FORWARD_NAT:
        case VIR_NETWORK_FORWARD_ROUTE:
3733 3734 3735 3736 3737 3738 3739 3740 3741 3742
            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).
                 */
3743 3744
                networkRemoveFirewallRules(def);
                needFirewallRefresh = true;
3745 3746 3747 3748
                break;
            default:
                break;
            }
3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762
            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;
3763 3764 3765
        }
    }

3766
    /* update the network config in memory/on disk */
3767
    if (virNetworkObjUpdate(obj, command, section, parentIndex, xml, flags) < 0) {
3768
        if (needFirewallRefresh)
3769
            ignore_value(networkAddFirewallRules(def));
3770 3771 3772
        goto cleanup;
    }

3773 3774 3775 3776
    /* @def is replaced */
    def = virNetworkObjGetDef(obj);

    if (needFirewallRefresh && networkAddFirewallRules(def) < 0)
3777 3778 3779 3780 3781
        goto cleanup;

    if (flags & VIR_NETWORK_UPDATE_AFFECT_CONFIG) {
        /* save updated persistent config to disk */
        if (virNetworkSaveConfig(driver->networkConfigDir,
3782
                                 virNetworkObjGetPersistentDef(obj)) < 0) {
3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795
            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 ||
3796 3797 3798 3799 3800 3801 3802 3803
            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)
3804
             */
3805
            if (networkRestartDhcpDaemon(driver, obj) < 0)
3806 3807
                goto cleanup;

3808 3809 3810 3811 3812 3813 3814 3815
        } 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;

3816
            for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
3817
                 i++) {
3818 3819 3820 3821 3822 3823 3824
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3825 3826
                 networkRestartDhcpDaemon(driver, obj) < 0) ||
                networkRefreshDhcpDaemon(driver, obj) < 0) {
3827 3828 3829
                goto cleanup;
            }

3830 3831 3832 3833
        } 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.
3834
             */
3835
            if (networkRefreshDhcpDaemon(driver, obj) < 0)
3836 3837 3838 3839 3840 3841 3842 3843
                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.
             */
3844
            if (networkRefreshRadvd(driver, obj) < 0)
3845 3846 3847 3848
                goto cleanup;
        }

        /* save current network state to disk */
3849
        if ((ret = virNetworkObjSaveStatus(driver->stateDir, obj)) < 0)
3850 3851
            goto cleanup;
    }
3852 3853

    /* call the 'updated' network hook script */
3854
    if (networkRunHook(obj, NULL, NULL, VIR_HOOK_NETWORK_OP_UPDATED,
3855 3856 3857
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

3858
    ret = 0;
3859
 cleanup:
3860
    virNetworkObjEndAPI(&obj);
3861 3862 3863
    return ret;
}

3864 3865 3866

static int
networkCreate(virNetworkPtr net)
3867
{
3868
    virNetworkDriverStatePtr driver = networkGetDriver();
3869
    virNetworkObjPtr obj;
3870
    virNetworkDefPtr def;
3871
    int ret = -1;
3872
    virObjectEventPtr event = NULL;
3873

3874
    if (!(obj = networkObjFromNetwork(net)))
3875
        goto cleanup;
3876
    def = virNetworkObjGetDef(obj);
3877

3878
    if (virNetworkCreateEnsureACL(net->conn, def) < 0)
3879 3880
        goto cleanup;

3881
    if ((ret = networkStartNetwork(driver, obj)) < 0)
3882
        goto cleanup;
3883

3884 3885
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3886 3887
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3888

3889
 cleanup:
3890
    virObjectEventStateQueue(driver->networkEventState, event);
3891
    virNetworkObjEndAPI(&obj);
3892
    return ret;
3893 3894
}

3895 3896 3897

static int
networkDestroy(virNetworkPtr net)
3898
{
3899
    virNetworkDriverStatePtr driver = networkGetDriver();
3900
    virNetworkObjPtr obj;
3901
    virNetworkDefPtr def;
3902
    int ret = -1;
3903
    virObjectEventPtr event = NULL;
3904

3905
    if (!(obj = networkObjFromNetwork(net)))
3906
        goto cleanup;
3907
    def = virNetworkObjGetDef(obj);
3908

3909
    if (virNetworkDestroyEnsureACL(net->conn, def) < 0)
3910 3911
        goto cleanup;

3912
    if (!virNetworkObjIsActive(obj)) {
3913
        virReportError(VIR_ERR_OPERATION_INVALID,
3914
                       _("network '%s' is not active"),
3915
                       def->name);
3916 3917 3918
        goto cleanup;
    }

3919
    if ((ret = networkShutdownNetwork(driver, obj)) < 0)
3920
        goto cleanup;
3921 3922
    /* @def replaced in virNetworkObjUnsetDefTransient*/
    def = virNetworkObjGetDef(obj);
3923

3924 3925
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3926 3927
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3928

3929 3930
    if (!virNetworkObjIsPersistent(obj) &&
        networkRemoveInactive(driver, obj) < 0) {
3931 3932
        ret = -1;
        goto cleanup;
3933
    }
3934

3935
 cleanup:
3936
    virObjectEventStateQueue(driver->networkEventState, event);
3937
    virNetworkObjEndAPI(&obj);
3938 3939 3940
    return ret;
}

3941 3942 3943 3944

static char *
networkGetXMLDesc(virNetworkPtr net,
                  unsigned int flags)
3945
{
3946
    virNetworkObjPtr obj;
3947
    virNetworkDefPtr curDef;
3948
    virNetworkDefPtr def;
3949
    virNetworkDefPtr newDef;
3950
    char *ret = NULL;
3951

3952
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3953

3954
    if (!(obj = networkObjFromNetwork(net)))
3955
        return ret;
3956 3957
    def = virNetworkObjGetDef(obj);
    newDef = virNetworkObjGetNewDef(obj);
3958

3959
    if (virNetworkGetXMLDescEnsureACL(net->conn, def) < 0)
3960 3961
        goto cleanup;

3962 3963
    if ((flags & VIR_NETWORK_XML_INACTIVE) && newDef)
        curDef = newDef;
3964
    else
3965
        curDef = def;
3966

3967
    ret = virNetworkDefFormat(curDef, flags);
3968

3969
 cleanup:
3970
    virNetworkObjEndAPI(&obj);
3971
    return ret;
3972 3973
}

3974 3975 3976 3977

static char *
networkGetBridgeName(virNetworkPtr net)
{
3978
    virNetworkObjPtr obj;
3979
    virNetworkDefPtr def;
3980 3981
    char *bridge = NULL;

3982
    if (!(obj = networkObjFromNetwork(net)))
3983
        return bridge;
3984
    def = virNetworkObjGetDef(obj);
3985

3986
    if (virNetworkGetBridgeNameEnsureACL(net->conn, def) < 0)
3987 3988
        goto cleanup;

3989
    if (!(def->bridge)) {
3990 3991
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
3992
                       def->name);
3993 3994 3995
        goto cleanup;
    }

3996
    ignore_value(VIR_STRDUP(bridge, def->bridge));
3997

3998
 cleanup:
3999
    virNetworkObjEndAPI(&obj);
4000 4001 4002
    return bridge;
}

4003 4004 4005 4006

static int
networkGetAutostart(virNetworkPtr net,
                    int *autostart)
4007
{
4008
    virNetworkObjPtr obj;
4009
    int ret = -1;
4010

4011
    if (!(obj = networkObjFromNetwork(net)))
4012
        return ret;
4013

4014
    if (virNetworkGetAutostartEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
4015 4016
        goto cleanup;

4017
    *autostart = virNetworkObjIsAutostart(obj) ? 1 : 0;
4018
    ret = 0;
4019

4020
 cleanup:
4021
    virNetworkObjEndAPI(&obj);
4022
    return ret;
4023 4024
}

4025 4026 4027 4028

static int
networkSetAutostart(virNetworkPtr net,
                    int autostart)
4029
{
4030
    virNetworkDriverStatePtr driver = networkGetDriver();
4031
    virNetworkObjPtr obj;
4032
    virNetworkDefPtr def;
4033
    char *configFile = NULL, *autostartLink = NULL;
4034 4035
    bool new_autostart;
    bool cur_autostart;
4036
    int ret = -1;
4037

4038
    if (!(obj = networkObjFromNetwork(net)))
4039
        goto cleanup;
4040
    def = virNetworkObjGetDef(obj);
4041

4042
    if (virNetworkSetAutostartEnsureACL(net->conn, def) < 0)
4043 4044
        goto cleanup;

4045
    if (!virNetworkObjIsPersistent(obj)) {
4046 4047
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
4048 4049 4050
        goto cleanup;
    }

4051 4052 4053
    new_autostart = (autostart != 0);
    cur_autostart = virNetworkObjIsAutostart(obj);
    if (cur_autostart != new_autostart) {
4054 4055
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir,
                                               def->name)) == NULL)
4056
            goto cleanup;
4057 4058
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir,
                                                  def->name)) == NULL)
4059 4060
            goto cleanup;

4061
        if (new_autostart) {
4062
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
4063
                virReportSystemError(errno,
4064 4065
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
4066 4067
                goto cleanup;
            }
4068

4069
            if (symlink(configFile, autostartLink) < 0) {
4070
                virReportSystemError(errno,
4071
                                     _("Failed to create symlink '%s' to '%s'"),
4072
                                     autostartLink, configFile);
4073 4074 4075
                goto cleanup;
            }
        } else {
4076
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
4077
                virReportSystemError(errno,
4078
                                     _("Failed to delete symlink '%s'"),
4079
                                     autostartLink);
4080 4081
                goto cleanup;
            }
4082 4083
        }

4084
        virNetworkObjSetAutostart(obj, new_autostart);
4085
    }
4086

4087
    ret = 0;
4088

4089
 cleanup:
4090 4091
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
4092
    virNetworkObjEndAPI(&obj);
4093
    return ret;
4094 4095
}

4096

4097
static int
4098
networkGetDHCPLeases(virNetworkPtr net,
4099 4100 4101
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
4102
{
4103
    virNetworkDriverStatePtr driver = networkGetDriver();
4104 4105 4106
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
4107
    size_t size = 0;
4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118
    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;
4119
    virNetworkIPDefPtr ipdef_tmp = NULL;
4120 4121
    virNetworkDHCPLeasePtr lease = NULL;
    virNetworkDHCPLeasePtr *leases_ret = NULL;
4122
    virNetworkObjPtr obj;
4123
    virNetworkDefPtr def;
4124
    virMacAddr mac_addr;
4125 4126 4127

    virCheckFlags(0, -1);

4128 4129 4130 4131 4132 4133
    /* only to check if the MAC is valid */
    if (mac && virMacAddrParse(mac, &mac_addr) < 0) {
        virReportError(VIR_ERR_INVALID_MAC, "%s", mac);
        return -1;
    }

4134
    if (!(obj = networkObjFromNetwork(net)))
4135
        return -1;
4136
    def = virNetworkObjGetDef(obj);
4137

4138
    if (virNetworkGetDHCPLeasesEnsureACL(net->conn, def) < 0)
4139
        goto cleanup;
4140 4141

    /* Retrieve custom leases file location */
4142
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(driver, def->bridge);
4143 4144

    /* Read entire contents */
4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158
    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);
        }
4159 4160 4161 4162 4163 4164 4165 4166 4167 4168
        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;
        }

4169
        if (!virJSONValueIsArray(leases_array)) {
4170
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
4171
                           _("Malformed lease_entries array"));
4172 4173
            goto error;
        }
4174
        size = virJSONValueArraySize(leases_array);
4175 4176
    }

4177
    currtime = (long long)time(NULL);
4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193

    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;
        }

4194
        if (mac && virMacAddrCompare(mac, mac_tmp))
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 4222 4223 4224 4225
            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 */
4226 4227
            for (j = 0; j < def->nips; j++) {
                ipdef_tmp = &def->ips[j];
4228 4229 4230 4231 4232 4233 4234 4235

                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)) {
4236
                    lease->prefix = virSocketAddrGetIPPrefix(&ipdef_tmp->address,
4237 4238 4239 4240 4241 4242 4243 4244
                                                             &ipdef_tmp->netmask,
                                                             ipdef_tmp->prefix);
                    break;
                }
            }

            if ((VIR_STRDUP(lease->mac, mac_tmp) < 0) ||
                (VIR_STRDUP(lease->ipaddr, ip_tmp) < 0) ||
4245
                (VIR_STRDUP(lease->iface, def->bridge) < 0))
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 4274 4275 4276 4277
                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);
4278
    VIR_FREE(lease_entries);
4279 4280
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
4281

4282
    virNetworkObjEndAPI(&obj);
4283

4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294
    return rv;

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

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 4327 4328 4329 4330
/* 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);
        }
    }
}

4331

4332 4333 4334 4335 4336 4337 4338 4339 4340
/* 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:
4341
 * @dom: domain definition that @iface belongs to
4342 4343 4344 4345 4346 4347 4348 4349 4350 4351
 * @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.
 */
4352
static int
4353 4354
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
4355
{
4356
    virNetworkDriverStatePtr driver = networkGetDriver();
4357
    virDomainNetType actualType = iface->type;
4358
    virNetworkObjPtr obj = NULL;
4359
    virNetworkDefPtr netdef = NULL;
4360
    virNetDevBandwidthPtr bandwidth = NULL;
4361 4362 4363
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
4364
    virNetworkForwardIfDefPtr dev = NULL;
4365
    size_t i;
4366 4367 4368
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
4369
        goto validate;
4370 4371 4372 4373

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

4374 4375
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
4376 4377 4378
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4379
        goto error;
4380
    }
4381
    netdef = virNetworkObjGetDef(obj);
4382

4383
    if (!virNetworkObjIsActive(obj)) {
4384 4385 4386 4387 4388 4389
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

4390 4391 4392
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

4393 4394 4395
    /* 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
4396
     */
4397 4398 4399 4400 4401 4402
    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.
     */
4403 4404 4405 4406 4407 4408

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

4409 4410
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
4411
        goto error;
4412

4413 4414 4415 4416 4417 4418 4419 4420
    /* 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;

4421 4422
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
4423

4424 4425 4426 4427 4428 4429 4430 4431 4432 4433
    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;

4434 4435 4436 4437 4438
    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:
4439
        /* for these forward types, the actual net type really *is*
4440
         * NETWORK; we just keep the info from the portgroup in
4441
         * iface->data.network.actual
J
Ján Tomko 已提交
4442
         */
4443
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
4444

4445
        /* we also store the bridge device and macTableManager settings
4446 4447 4448 4449 4450 4451 4452
         * 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;
4453 4454
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4455

4456
        if (networkPlugBandwidth(obj, iface) < 0)
4457
            goto error;
4458
        break;
4459

4460
    case VIR_NETWORK_FORWARD_HOSTDEV: {
4461
        virDomainHostdevSubsysPCIBackendType backend;
4462

4463
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
4464
        if (networkCreateInterfacePool(netdef) < 0)
4465 4466 4467
            goto error;

        /* pick first dev with 0 connections */
4468 4469 4470
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484
                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;
4485
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
4486
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
4487
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
4488

E
Eric Blake 已提交
4489
        switch (netdef->forward.driverName) {
4490
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
4491
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
4492 4493
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
4494
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
4495 4496
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
4497
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508
            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;

4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532
        /* 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;
            }
        }
4533 4534
        break;
    }
4535

4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547
    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;
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 4578 4579 4580 4581
            /* 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:
4582 4583 4584 4585 4586
        /* <forward type='bridge|private|vepa|passthrough'> are all
         * VIR_DOMAIN_NET_TYPE_DIRECT.
         */

        /* Set type=direct and appropriate <source mode='xxx'/> */
4587
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4588 4589 4590 4591 4592 4593

        /* 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));
4594

4595 4596 4597 4598 4599 4600 4601 4602
        /* 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) {
4603
            goto error;
4604
        }
4605
        virtport = iface->data.network.actual->virtPortProfile;
4606
        if (virtport) {
4607 4608 4609 4610 4611 4612 4613 4614
            /* 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);
4615
                goto error;
4616 4617
            }
        }
4618

4619 4620 4621
        /* 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).
         */
4622
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4623 4624 4625 4626
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4627
            goto error;
4628 4629 4630
        } else {
            /* pick an interface from the pool */

4631
            if (networkCreateInterfacePool(netdef) < 0)
4632 4633
                goto error;

4634 4635 4636 4637 4638
            /* 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.
4639
             */
4640 4641
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4642 4643 4644
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4645

4646
                /* pick first dev with 0 connections */
4647 4648 4649
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4650 4651 4652 4653 4654
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4655
                dev = &netdef->forward.ifs[0];
4656 4657 4658
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4659 4660 4661 4662
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4663 4664 4665 4666
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4667
                goto error;
4668
            }
4669 4670
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4671
                goto error;
4672
        }
4673 4674 4675 4676 4677 4678
        break;

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, netdef->forward.type);
        goto error;
4679 4680
    }

4681 4682
    if (virNetworkObjMacMgrAdd(obj, driver->dnsmasqStateDir,
                               dom->name, &iface->mac) < 0)
M
Michal Privoznik 已提交
4683 4684
        goto error;

4685
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4686
        goto error;
4687

4688
 validate:
4689 4690 4691 4692 4693
    /* 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.
     */

4694
    if (virDomainNetGetActualVlan(iface)) {
4695 4696 4697 4698
        /* 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
4699 4700
         */
        if (!(actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV ||
4701 4702 4703
              (actualType == VIR_DOMAIN_NET_TYPE_DIRECT &&
               virDomainNetGetActualDirectMode(iface)
               == VIR_NETDEV_MACVLAN_MODE_PASSTHRU) ||
4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722
              (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;
        }
    }
4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733
    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;
        }
    }
4734 4735 4736

    if (netdef) {
        netdef->connections++;
4737
        if (dev)
4738 4739
            dev->connections++;
        /* finally we can call the 'plugged' hook script if any */
4740
        if (networkRunHook(obj, dom, iface,
4741 4742 4743
                           VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
                           VIR_HOOK_SUBOP_BEGIN) < 0) {
            /* adjust for failure */
4744
            netdef->connections--;
4745 4746 4747 4748
            if (dev)
                dev->connections--;
            goto error;
        }
4749
        networkLogAllocation(netdef, actualType, dev, iface, true);
4750 4751
    }

4752
    ret = 0;
4753

4754
 cleanup:
4755
    virNetworkObjEndAPI(&obj);
4756 4757
    return ret;

4758
 error:
4759
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4760 4761 4762
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4763
    goto cleanup;
4764 4765
}

4766

4767
/* networkNotifyActualDevice:
4768
 * @dom: domain definition that @iface belongs to
4769 4770 4771 4772 4773
 * @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
4774 4775
 * order, or re-attach the interface's tap device to the network's
 * bridge.
4776
 *
4777
 * No return value (but does log any failures)
4778
 */
4779
static void
4780 4781
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4782
{
4783
    virNetworkDriverStatePtr driver = networkGetDriver();
4784
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4785
    virNetworkObjPtr obj;
4786
    virNetworkDefPtr netdef;
4787
    virNetworkForwardIfDefPtr dev = NULL;
4788
    size_t i;
4789
    char *master = NULL;
4790 4791

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
4792
        return;
4793

4794 4795
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
4796 4797 4798
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4799 4800
        goto error;
    }
4801
    netdef = virNetworkObjGetDef(obj);
4802

4803
    if (!virNetworkObjIsActive(obj)) {
4804 4805 4806 4807 4808 4809
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820
    /* 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;

4821 4822
    /* see if we're connected to the correct bridge */
    if (netdef->bridge) {
4823 4824
        bool useOVS = false;

4825 4826 4827
        if (virNetDevGetMaster(iface->ifname, &master) < 0)
            goto error;

4828 4829 4830 4831 4832 4833 4834 4835
        /* 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;
        }

4836 4837
        if (STRNEQ_NULLABLE(netdef->bridge, master)) {
            /* disconnect from current (incorrect) bridge */
4838 4839 4840 4841 4842 4843 4844
            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));
            }
4845 4846 4847 4848 4849 4850

            /* 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())
             */
4851
            VIR_INFO("Attaching %s to %s", iface->ifname, netdef->bridge);
4852 4853 4854 4855 4856 4857 4858 4859 4860 4861
            if (virNetDevTapAttachBridge(iface->ifname, netdef->bridge,
                                         &iface->mac, dom->uuid,
                                         virDomainNetGetActualVirtPortProfile(iface),
                                         virDomainNetGetActualVlan(iface),
                                         iface->mtu, NULL) < 0) {
                goto error;
            }
        }
    }

4862
    if (!iface->data.network.actual ||
4863 4864
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4865 4866
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4867 4868
    }

4869
    if (networkCreateInterfacePool(netdef) < 0)
4870
        goto error;
4871

4872
    if (netdef->forward.nifs == 0) {
4873
        virReportError(VIR_ERR_INTERNAL_ERROR,
4874 4875
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4876
                       netdef->name);
4877
        goto error;
4878
    }
4879

4880 4881
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4882

4883 4884 4885 4886 4887 4888 4889 4890 4891
        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 */
4892 4893
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4894
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4895 4896
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4897 4898 4899 4900 4901
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4902
            virReportError(VIR_ERR_INTERNAL_ERROR,
4903 4904
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4905
                           netdef->name, actualDev);
4906
            goto error;
4907 4908
        }

4909
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4910 4911
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4912
         */
4913
        if ((dev->connections > 0) &&
4914 4915
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4916 4917
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4918
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4919
            virReportError(VIR_ERR_INTERNAL_ERROR,
4920 4921
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4922
                           netdef->name, actualDev);
4923
            goto error;
4924
        }
4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936
    }  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 */
4937 4938
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4939
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4940
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4941 4942
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4943 4944 4945 4946 4947 4948 4949 4950 4951
                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,
4952 4953 4954 4955
                           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
4956
            goto error;
4957 4958 4959 4960 4961 4962 4963
        }

        /* 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) &&
4964
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4965 4966 4967 4968 4969 4970 4971 4972 4973
            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;
        }
4974 4975
    }

4976
 success:
4977
    netdef->connections++;
4978 4979
    if (dev)
        dev->connections++;
4980
    /* finally we can call the 'plugged' hook script if any */
4981
    if (networkRunHook(obj, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
4982 4983 4984 4985 4986 4987 4988
                       VIR_HOOK_SUBOP_BEGIN) < 0) {
        /* adjust for failure */
        if (dev)
            dev->connections--;
        netdef->connections--;
        goto error;
    }
4989
    networkLogAllocation(netdef, actualType, dev, iface, true);
4990

4991
 cleanup:
4992
    virNetworkObjEndAPI(&obj);
4993
    VIR_FREE(master);
4994
    return;
4995

4996
 error:
4997
    goto cleanup;
4998 4999 5000 5001
}


/* networkReleaseActualDevice:
5002
 * @dom: domain definition that @iface belongs to
5003 5004 5005 5006 5007 5008 5009 5010 5011
 * @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.
 */
5012
static int
5013 5014
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
5015
{
5016
    virNetworkDriverStatePtr driver = networkGetDriver();
5017
    virDomainNetType actualType = virDomainNetGetActualType(iface);
5018
    virNetworkObjPtr obj;
5019
    virNetworkDefPtr netdef;
5020
    virNetworkForwardIfDefPtr dev = NULL;
5021 5022
    size_t i;
    int ret = -1;
5023 5024

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
5025
        return 0;
5026

5027 5028
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
5029 5030 5031
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
5032 5033
        goto error;
    }
5034
    netdef = virNetworkObjGetDef(obj);
5035

5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054
    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);
5055
        goto error;
5056
    }
5057

5058 5059 5060
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
5061 5062
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
5063 5064
    }

5065
    if (netdef->forward.nifs == 0) {
5066
        virReportError(VIR_ERR_INTERNAL_ERROR,
5067
                       _("network '%s' uses a direct/hostdev mode, but "
5068 5069
                         "has no forward dev and no interface pool"),
                       netdef->name);
5070
        goto error;
5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082
    }

    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;
        }
5083

5084 5085
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
5086
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
5087 5088
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
5089 5090 5091
                break;
            }
        }
5092

5093
        if (!dev) {
5094
            virReportError(VIR_ERR_INTERNAL_ERROR,
5095 5096
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
5097
                           netdef->name, actualDev);
5098
            goto error;
5099
        }
5100 5101 5102 5103 5104 5105 5106 5107 5108 5109
    } 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;
        }

5110 5111
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
5112
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
5113
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
5114 5115
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
5116 5117 5118 5119 5120 5121 5122 5123 5124
                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,
5125 5126 5127 5128
                           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
5129
            goto error;
5130
        }
J
Ján Tomko 已提交
5131
    }
5132

5133
 success:
5134
    virNetworkObjMacMgrDel(obj, driver->dnsmasqStateDir, dom->name, &iface->mac);
M
Michal Privoznik 已提交
5135

5136
    if (iface->data.network.actual) {
5137
        netdef->connections--;
5138 5139
        if (dev)
            dev->connections--;
5140
        /* finally we can call the 'unplugged' hook script if any */
5141
        networkRunHook(obj, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
5142
                       VIR_HOOK_SUBOP_BEGIN);
5143
        networkLogAllocation(netdef, actualType, dev, iface, false);
5144
    }
5145
    ret = 0;
5146
 cleanup:
5147
    virNetworkObjEndAPI(&obj);
5148 5149 5150 5151
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
5152
    return ret;
5153

5154
 error:
5155
    goto cleanup;
5156
}
5157

5158

5159 5160 5161
/**
 * networkCheckBandwidth:
 * @net: network QoS
5162
 * @ifaceBand: interface QoS (may be NULL if no QoS)
5163
 * @oldBandwidth: new interface QoS (may be NULL if no QoS)
5164
 * @ifaceMac: interface MAC (used in error messages for identification)
5165 5166
 * @new_rate: new rate for non guaranteed class
 *
5167 5168 5169 5170 5171 5172 5173 5174
 * 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.
 *
5175 5176 5177 5178 5179
 * 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
5180
networkCheckBandwidth(virNetworkObjPtr obj,
5181
                      virNetDevBandwidthPtr ifaceBand,
5182
                      virNetDevBandwidthPtr oldBandwidth,
5183
                      virMacAddr ifaceMac,
5184 5185 5186
                      unsigned long long *new_rate)
{
    int ret = -1;
5187 5188
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
    virNetDevBandwidthPtr netBand = def->bandwidth;
5189
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5190 5191 5192
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

5193
    virMacAddrFormat(&ifaceMac, ifmac);
5194 5195 5196 5197 5198 5199

    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"),
5200
                       ifmac, def->name);
5201 5202 5203
        return -1;
    }

5204 5205
    if (((!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor) &&
         (!oldBandwidth || !oldBandwidth->in || !oldBandwidth->in->floor)) ||
5206 5207
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
5208
        return 1;
5209
    }
5210 5211

    tmp_new_rate = netBand->in->average;
5212 5213 5214 5215
    if (oldBandwidth && oldBandwidth->in)
        tmp_floor_sum -= oldBandwidth->in->floor;
    if (ifaceBand && ifaceBand->in)
        tmp_floor_sum += ifaceBand->in->floor;
5216 5217 5218 5219 5220 5221 5222 5223 5224

    /* 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,
5225 5226
                           def->bridge,
                           def->name);
5227 5228 5229 5230 5231 5232 5233 5234 5235
            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,
5236 5237
                       def->bridge,
                       def->name);
5238 5239 5240
        goto cleanup;
    }

5241 5242
    if (new_rate)
        *new_rate = tmp_new_rate;
5243 5244
    ret = 0;

5245
 cleanup:
5246 5247 5248
    return ret;
}

5249

5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260
/**
 * 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
5261
networkNextClassID(virNetworkObjPtr obj)
5262
{
5263
    ssize_t ret = 0;
5264
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5265

5266 5267
    if ((ret = virBitmapNextClearBit(classIdMap, -1)) < 0)
        ret = virBitmapSize(classIdMap);
5268

5269
    if (virBitmapSetBitExpand(classIdMap, ret) < 0)
5270 5271 5272 5273 5274
        return -1;

    return ret;
}

5275

5276
static int
5277
networkPlugBandwidthImpl(virNetworkObjPtr obj,
5278 5279 5280
                         virDomainNetDefPtr iface,
                         virNetDevBandwidthPtr ifaceBand,
                         unsigned long long new_rate)
5281
{
5282
    virNetworkDriverStatePtr driver = networkGetDriver();
5283
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
5284
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5285
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5286
    ssize_t class_id = 0;
5287 5288
    int plug_ret;
    int ret = -1;
5289 5290

    /* generate new class_id */
5291
    if ((class_id = networkNextClassID(obj)) < 0) {
5292 5293 5294 5295 5296
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Could not generate next class ID"));
        goto cleanup;
    }

5297
    plug_ret = virNetDevBandwidthPlug(def->bridge, def->bandwidth,
5298
                                      &iface->mac, ifaceBand, class_id);
5299
    if (plug_ret < 0) {
5300
        ignore_value(virNetDevBandwidthUnplug(def->bridge, class_id));
5301 5302 5303 5304 5305 5306
        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 */
5307 5308
    tmp_floor_sum += ifaceBand->in->floor;
    virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5309
    /* update status file */
5310
    if (virNetworkObjSaveStatus(driver->stateDir, obj) < 0) {
5311
        ignore_value(virBitmapClearBit(classIdMap, class_id));
5312 5313
        tmp_floor_sum -= ifaceBand->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5314
        iface->data.network.actual->class_id = 0;
5315
        ignore_value(virNetDevBandwidthUnplug(def->bridge, class_id));
5316 5317
        goto cleanup;
    }
5318
    /* update rate for non guaranteed NICs */
5319
    new_rate -= tmp_floor_sum;
5320 5321
    if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                     def->bandwidth, new_rate) < 0)
5322
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
5323
                 def->bridge);
5324 5325

    ret = 0;
5326 5327 5328 5329 5330 5331
 cleanup:
    return ret;
}


static int
5332
networkPlugBandwidth(virNetworkObjPtr obj,
5333 5334 5335 5336 5337 5338 5339 5340
                     virDomainNetDefPtr iface)
{
    int ret = -1;
    int plug_ret;
    unsigned long long new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);

5341
    if ((plug_ret = networkCheckBandwidth(obj, ifaceBand, NULL,
5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361
                                          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;
    }

5362
    if (networkPlugBandwidthImpl(obj, iface, ifaceBand, new_rate) < 0)
5363 5364 5365
        goto cleanup;

    ret = 0;
5366

5367
 cleanup:
5368 5369 5370
    return ret;
}

5371

5372
static int
5373
networkUnplugBandwidth(virNetworkObjPtr obj,
5374 5375
                       virDomainNetDefPtr iface)
{
5376
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
5377
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5378
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5379
    virNetworkDriverStatePtr driver = networkGetDriver();
5380 5381
    int ret = 0;
    unsigned long long new_rate;
5382
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
5383 5384 5385

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
5386
        if (!def->bandwidth || !def->bandwidth->in) {
5387
            VIR_WARN("Network %s has no bandwidth but unplug requested",
5388
                     def->name);
5389 5390
            goto cleanup;
        }
5391
        /* we must remove class from bridge */
5392
        new_rate = def->bandwidth->in->average;
5393

5394 5395
        if (def->bandwidth->in->peak > 0)
            new_rate = def->bandwidth->in->peak;
5396

5397
        ret = virNetDevBandwidthUnplug(def->bridge,
5398 5399 5400 5401
                                       iface->data.network.actual->class_id);
        if (ret < 0)
            goto cleanup;
        /* update sum of 'floor'-s of attached NICs */
5402 5403 5404
        tmp_floor_sum -= ifaceBand->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);

5405
        /* return class ID */
5406
        ignore_value(virBitmapClearBit(classIdMap,
5407 5408
                                       iface->data.network.actual->class_id));
        /* update status file */
5409
        if (virNetworkObjSaveStatus(driver->stateDir, obj) < 0) {
5410 5411
            tmp_floor_sum += ifaceBand->in->floor;
            virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5412
            ignore_value(virBitmapSetBit(classIdMap,
5413 5414 5415
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
5416
        /* update rate for non guaranteed NICs */
5417
        new_rate -= tmp_floor_sum;
5418 5419
        if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                         def->bandwidth, new_rate) < 0)
5420
            VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
5421
                     def->bridge);
5422 5423 5424 5425
        /* no class is associated any longer */
        iface->data.network.actual->class_id = 0;
    }

5426
 cleanup:
5427 5428
    return ret;
}
5429

5430

5431
static void
5432
networkNetworkObjTaint(virNetworkObjPtr obj,
5433
                       virNetworkTaintFlags taint)
5434
{
5435 5436
    virNetworkDefPtr def = virNetworkObjGetDef(obj);

5437
    if (virNetworkObjTaint(obj, taint)) {
5438
        char uuidstr[VIR_UUID_STRING_BUFLEN];
5439
        virUUIDFormat(def->uuid, uuidstr);
5440 5441

        VIR_WARN("Network name='%s' uuid=%s is tainted: %s",
5442
                 def->name, uuidstr, virNetworkTaintTypeToString(taint));
5443 5444
    }
}
5445 5446 5447 5448 5449 5450


static bool
networkBandwidthGenericChecks(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
5451
    virNetDevBandwidthPtr ifaceBand;
5452 5453 5454 5455 5456 5457 5458 5459
    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;
    }

5460
    ifaceBand = virDomainNetGetActualBandwidth(iface);
5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471
    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;
}


5472
static bool
5473 5474 5475 5476
networkBandwidthChangeAllowed(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
5477
    virNetworkObjPtr obj = NULL;
5478 5479 5480 5481 5482 5483
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    bool ret = false;

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

5484 5485
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
5486 5487 5488 5489 5490 5491
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return false;
    }

5492
    if (networkCheckBandwidth(obj, newBandwidth, ifaceBand, iface->mac, NULL) < 0)
5493 5494 5495 5496 5497
        goto cleanup;

    ret = true;

 cleanup:
5498
    virNetworkObjEndAPI(&obj);
5499 5500
    return ret;
}
5501 5502


5503
static int
5504 5505 5506 5507
networkBandwidthUpdate(virDomainNetDefPtr iface,
                       virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
5508
    virNetworkObjPtr obj = NULL;
5509
    virNetworkDefPtr def;
5510
    unsigned long long tmp_floor_sum;
5511 5512 5513 5514 5515 5516 5517 5518
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    unsigned long long new_rate = 0;
    int plug_ret;
    int ret = -1;

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

5519 5520
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
5521 5522 5523 5524 5525
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return ret;
    }
5526
    def = virNetworkObjGetDef(obj);
5527

5528
    if ((plug_ret = networkCheckBandwidth(obj, newBandwidth, ifaceBand,
5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541
                                          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: */

5542
    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
5543
        newBandwidth->in && newBandwidth->in->floor) {
5544 5545
        /* Either we just need to update @floor .. */

5546
        if (virNetDevBandwidthUpdateRate(def->bridge,
5547
                                         iface->data.network.actual->class_id,
5548
                                         def->bandwidth,
5549 5550 5551
                                         newBandwidth->in->floor) < 0)
            goto cleanup;

5552 5553 5554 5555 5556
        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;
5557

5558 5559
        if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                         def->bandwidth, new_rate) < 0 ||
5560
            virNetworkObjSaveStatus(driver->stateDir, obj) < 0) {
5561
            /* Ouch, rollback */
5562 5563 5564
            tmp_floor_sum -= newBandwidth->in->floor;
            tmp_floor_sum += ifaceBand->in->floor;
            virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5565

5566
            ignore_value(virNetDevBandwidthUpdateRate(def->bridge,
5567
                                                      iface->data.network.actual->class_id,
5568
                                                      def->bandwidth,
5569 5570 5571 5572 5573 5574
                                                      ifaceBand->in->floor));
            goto cleanup;
        }
    } else if (newBandwidth->in && newBandwidth->in->floor) {
        /* .. or we need to plug in new .. */

5575
        if (networkPlugBandwidthImpl(obj, iface, newBandwidth, new_rate) < 0)
5576 5577 5578 5579
            goto cleanup;
    } else {
        /* .. or unplug old. */

5580
        if (networkUnplugBandwidth(obj, iface) < 0)
5581 5582 5583 5584 5585
            goto cleanup;
    }

    ret = 0;
 cleanup:
5586
    virNetworkObjEndAPI(&obj);
5587 5588
    return ret;
}
5589 5590 5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616


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 */
};

5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627 5628

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 = {
5629
    .localOnly = true,
5630
    .uriSchemes = (const char *[]){ "network", NULL },
5631 5632 5633 5634 5635
    .hypervisorDriver = &networkHypervisorDriver,
    .networkDriver = &networkDriver,
};


5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646
static virStateDriver networkStateDriver = {
    .name = "bridge",
    .stateInitialize  = networkStateInitialize,
    .stateAutoStart  = networkStateAutoStart,
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
};

int
networkRegister(void)
{
5647 5648
    if (virRegisterConnectDriver(&networkConnectDriver, false) < 0)
        return -1;
5649 5650 5651 5652
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
        return -1;
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
5653 5654 5655 5656

    virDomainNetSetDeviceImpl(
        networkAllocateActualDevice,
        networkNotifyActualDevice,
5657 5658
        networkReleaseActualDevice,
        networkBandwidthChangeAllowed,
5659
        networkBandwidthUpdate);
5660

5661 5662
    return 0;
}