bridge_driver.c 185.2 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2016 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
 */

#include <config.h>

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

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

71
#define VIR_FROM_THIS VIR_FROM_NETWORK
72
#define MAX_BRIDGE_ID 256
73

74 75 76 77 78 79 80
/**
 * VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX:
 *
 * Macro providing the upper limit on the size of leases file
 */
#define VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX (32 * 1024 * 1024)

81 82
#define SYSCTL_PATH "/proc/sys"

83 84
VIR_LOG_INIT("network.bridge_driver");

85
static virNetworkDriverStatePtr network_driver;
86

87

88 89 90 91 92 93 94 95
static virNetworkDriverStatePtr
networkGetDriver(void)
{
    /* Maybe one day we can store @network_driver in the
     * connection object, but until then, it's just a global
     * variable which is returned. */
    return network_driver;
}
96

97 98 99

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


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

111

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

122

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

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

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

138

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

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

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

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

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

158
static int
159
networkStartNetworkExternal(virNetworkObjPtr obj);
160

161
static int
162
networkShutdownNetworkExternal(virNetworkObjPtr obj);
163

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

167 168
static void
networkRefreshDaemons(virNetworkDriverStatePtr driver);
169

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

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

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

182

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

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

198
    return obj;
199 200
}

201

202
static int
203
networkRunHook(virNetworkObjPtr obj,
204
               virDomainDefPtr dom,
205
               virDomainNetDefPtr iface,
206 207 208
               int op,
               int sub_op)
{
209
    virNetworkDefPtr def;
210 211 212 213 214 215
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *xml = NULL, *net_xml = NULL, *dom_xml = NULL;
    int hookret;
    int ret = -1;

    if (virHookPresent(VIR_HOOK_DRIVER_NETWORK)) {
216 217
        if (!obj) {
            VIR_DEBUG("Not running hook as @obj is NULL");
218 219 220
            ret = 0;
            goto cleanup;
        }
221
        def = virNetworkObjGetDef(obj);
222

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

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

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

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

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

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

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

260

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

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

272

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

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

284

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

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

296

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

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

307

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

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

319

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

    int ret = -1;

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

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

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

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

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

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

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

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

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

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

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

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

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

    ret = 0;

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

397

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

404 405 406 407 408 409 410
    if (strlen(brname) + sizeof(dummyNicSuffix) > IFNAMSIZ) {
        /* because the length of an ifname is limited to IFNAMSIZ-1
         * (usually 15), and we're adding 4 more characters, we must
         * truncate the original name to 11 to fit. In order to catch
         * a possible numeric ending (eg virbr0, virbr1, etc), we grab
         * the first 8 and last 3 characters of the string.
         */
411 412 413 414 415
        ignore_value(virAsprintf(&nicname, "%.*s%s%s",
                                 /* space for last 3 chars + "-nic" + NULL */
                                 (int)(IFNAMSIZ - (3 + sizeof(dummyNicSuffix))),
                                 brname, brname + strlen(brname) - 3,
                                 dummyNicSuffix));
416
    } else {
417
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
418
    }
419 420 421
    return nicname;
}

422

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

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

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

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

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

457 458
        virNetworkObjSetMacMap(obj, macmap);

459
        break;
460

461
    case VIR_NETWORK_FORWARD_BRIDGE:
462 463
        if (def->bridge) {
            if (virNetDevExists(def->bridge) != 1)
464
                virNetworkObjSetActive(obj, false);
465
            break;
466
        }
467 468 469 470 471 472 473 474 475
        /* intentionally drop through to common case for all
         * macvtap networks (forward='bridge' with no bridge
         * device defined is macvtap using its 'bridge' mode)
         */
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
        /* so far no extra checks */
        break;
476

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

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

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

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

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

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

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

518

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

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

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

538

539
#ifdef WITH_FIREWALLD
540 541
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
542 543
                             DBusMessage *message,
                             void *user_data)
544
{
545
    virNetworkDriverStatePtr driver = user_data;
546
    bool reload = false;
547

548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571
    if (dbus_message_is_signal(message,
                               "org.fedoraproject.FirewallD1", "Reloaded")) {
        reload = true;

    } else if (dbus_message_is_signal(message,
                                      DBUS_INTERFACE_DBUS, "NameOwnerChanged")) {

        VIR_AUTOFREE(char *) name = NULL;
        VIR_AUTOFREE(char *) old_owner = NULL;
        VIR_AUTOFREE(char *) new_owner = NULL;

        if (virDBusMessageDecode(message, "sss", &name, &old_owner, &new_owner) < 0) {
            VIR_WARN("Failed to decode DBus NameOwnerChanged message");
            return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
        }
        /*
         * if new_owner is empty, firewalld is shutting down. If it is
         * non-empty, then it is starting
         */
        if (new_owner && *new_owner)
            reload = true;
    }

    if (reload) {
572
        VIR_DEBUG("Reload in bridge_driver because of firewalld.");
573
        networkReloadFirewallRules(driver, false);
574 575 576 577 578 579
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

580

581
/**
582
 * networkStateInitialize:
583
 *
J
Ján Tomko 已提交
584
 * Initialization function for the QEMU daemon
585 586
 */
static int
587 588 589
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
590
{
591 592 593
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
594
#ifdef WITH_FIREWALLD
595 596
    DBusConnection *sysbus = NULL;
#endif
597

598
    if (VIR_ALLOC(network_driver) < 0)
599
        goto error;
600

601 602
    if (virMutexInit(&network_driver->lock) < 0) {
        VIR_FREE(network_driver);
603 604
        goto error;
    }
605

606 607
    network_driver->privileged = privileged;

608 609 610 611
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
612
    if (privileged) {
613
        if (VIR_STRDUP(network_driver->networkConfigDir,
614
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
615
            VIR_STRDUP(network_driver->networkAutostartDir,
616
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
617
            VIR_STRDUP(network_driver->stateDir,
618
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
619
            VIR_STRDUP(network_driver->pidDir,
620
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
621
            VIR_STRDUP(network_driver->dnsmasqStateDir,
622
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
623
            VIR_STRDUP(network_driver->radvdStateDir,
624 625
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
626
    } else {
627 628 629
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
630
            goto error;
631

632
        if ((virAsprintf(&network_driver->networkConfigDir,
633
                         "%s/qemu/networks", configdir) < 0) ||
634
            (virAsprintf(&network_driver->networkAutostartDir,
635
                         "%s/qemu/networks/autostart", configdir) < 0) ||
636
            (virAsprintf(&network_driver->stateDir,
637
                         "%s/network/lib", rundir) < 0) ||
638
            (virAsprintf(&network_driver->pidDir,
639
                         "%s/network/run", rundir) < 0) ||
640
            (virAsprintf(&network_driver->dnsmasqStateDir,
641
                         "%s/dnsmasq/lib", rundir) < 0) ||
642
            (virAsprintf(&network_driver->radvdStateDir,
643
                         "%s/radvd/lib", rundir) < 0)) {
644
            goto error;
645
        }
646 647
    }

648
    if (virFileMakePath(network_driver->stateDir) < 0) {
649 650
        virReportSystemError(errno,
                             _("cannot create directory %s"),
651
                             network_driver->stateDir);
652 653 654
        goto error;
    }

655
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
656
    network_driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
657

658
    if (!(network_driver->networks = virNetworkObjListNew()))
659 660
        goto error;

661 662
    if (virNetworkObjLoadAllState(network_driver->networks,
                                  network_driver->stateDir) < 0)
663 664
        goto error;

665 666 667
    if (virNetworkObjLoadAllConfigs(network_driver->networks,
                                    network_driver->networkConfigDir,
                                    network_driver->networkAutostartDir) < 0)
668 669
        goto error;

670 671 672 673
    /* 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). */
674
    virNetworkObjListForEach(network_driver->networks,
675
                             networkUpdateState,
676 677
                             network_driver);
    virNetworkObjListPrune(network_driver->networks,
678 679
                           VIR_CONNECT_LIST_NETWORKS_INACTIVE |
                           VIR_CONNECT_LIST_NETWORKS_TRANSIENT);
680
    networkReloadFirewallRules(network_driver, true);
681
    networkRefreshDaemons(network_driver);
682

683 684 685 686
    virNetworkObjListForEach(network_driver->networks,
                             networkAutostartConfig,
                             network_driver);

687
    network_driver->networkEventState = virObjectEventStateNew();
688

689
#ifdef WITH_FIREWALLD
690 691
    if (!(sysbus = virDBusGetSystemBus())) {
        VIR_WARN("DBus not available, disabling firewalld support "
692
                 "in bridge_network_driver: %s", virGetLastErrorMessage());
693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709
    } else {
        /* add matches for
         * NameOwnerChanged on org.freedesktop.DBus for firewalld start/stop
         * Reloaded on org.fedoraproject.FirewallD1 for firewalld reload
         */
        dbus_bus_add_match(sysbus,
                           "type='signal'"
                           ",interface='"DBUS_INTERFACE_DBUS"'"
                           ",member='NameOwnerChanged'"
                           ",arg0='org.fedoraproject.FirewallD1'",
                           NULL);
        dbus_bus_add_match(sysbus,
                           "type='signal'"
                           ",interface='org.fedoraproject.FirewallD1'"
                           ",member='Reloaded'",
                           NULL);
        dbus_connection_add_filter(sysbus, firewalld_dbus_filter_bridge,
710
                                   network_driver, NULL);
711 712 713
    }
#endif

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

720
 error:
721
    networkStateCleanup();
722
    goto cleanup;
723 724
}

725

726
/**
727
 * networkStateReload:
728
 *
J
Ján Tomko 已提交
729
 * Function to restart the QEMU daemon, it will recheck the configuration
730 731 732
 * files and update its state and the networking
 */
static int
733 734
networkStateReload(void)
{
735
    if (!network_driver)
736 737
        return 0;

738 739 740 741 742
    virNetworkObjLoadAllState(network_driver->networks,
                              network_driver->stateDir);
    virNetworkObjLoadAllConfigs(network_driver->networks,
                                network_driver->networkConfigDir,
                                network_driver->networkAutostartDir);
743
    networkReloadFirewallRules(network_driver, false);
744 745
    networkRefreshDaemons(network_driver);
    virNetworkObjListForEach(network_driver->networks,
746
                             networkAutostartConfig,
747
                             network_driver);
748 749 750 751 752
    return 0;
}


/**
753
 * networkStateCleanup:
754
 *
J
Ján Tomko 已提交
755
 * Shutdown the QEMU daemon, it will stop all active domains and networks
756 757
 */
static int
758 759
networkStateCleanup(void)
{
760
    if (!network_driver)
761 762
        return -1;

763
    virObjectUnref(network_driver->networkEventState);
764

765
    /* free inactive networks */
766
    virObjectUnref(network_driver->networks);
767

768 769 770 771 772 773
    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);
774

775
    virObjectUnref(network_driver->dnsmasqCaps);
776

777
    virMutexDestroy(&network_driver->lock);
778

779
    VIR_FREE(network_driver);
780 781 782 783 784

    return 0;
}


785 786 787 788 789 790 791 792
static virDrvOpenStatus
networkConnectOpen(virConnectPtr conn,
                   virConnectAuthPtr auth ATTRIBUTE_UNUSED,
                   virConfPtr conf ATTRIBUTE_UNUSED,
                   unsigned int flags)
{
    virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);

793 794 795 796 797 798 799 800 801 802 803
    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);
804 805
            return VIR_DRV_OPEN_ERROR;
        }
806 807 808 809 810 811
    } 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;
812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846
        }
    }

    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;
}


847 848 849 850 851
/* networkKillDaemon:
 *
 * kill the specified pid/name, and wait a bit to make sure it's dead.
 */
static int
852 853 854
networkKillDaemon(pid_t pid,
                  const char *daemonName,
                  const char *networkName)
855
{
856 857
    size_t i;
    int ret = -1;
858 859 860 861 862 863 864
    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.
     */
865
    for (i = 0; i < 25; i++) {
866
        int signum = 0;
867
        if (i == 0) {
868
            signum = SIGTERM;
869
        } else if (i == 15) {
870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904
            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);
905
 cleanup:
906 907 908
    return ret;
}

909

J
Ján Tomko 已提交
910 911 912
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
913
static int
G
Gene Czarcinski 已提交
914
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
915
                                 virNetworkIPDefPtr ipdef)
916
{
917
    size_t i;
G
Gene Czarcinski 已提交
918
    bool ipv6 = false;
919

G
Gene Czarcinski 已提交
920 921
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
922 923
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
924
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
925 926
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
927
                return -1;
928
    }
929

G
Gene Czarcinski 已提交
930 931 932
    return 0;
}

933

G
Gene Czarcinski 已提交
934 935 936 937
static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
938
    size_t i, j;
G
Gene Czarcinski 已提交
939

940 941
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
942
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
943
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
944
                for (j = 0; j < host->nnames; j++)
945 946
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
947 948
            }
        }
949 950
    }

951
    return 0;
952 953 954
}


955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991
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;
}


992
int
993
networkDnsmasqConfContents(virNetworkObjPtr obj,
994 995 996 997
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
998
{
999
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1000
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1001
    int r, ret = -1;
1002
    int nbleases = 0;
1003
    size_t i;
1004
    virNetworkDNSDefPtr dns = &def->dns;
1005
    bool wantDNS = dns->enable != VIR_TRISTATE_BOOL_NO;
1006
    virNetworkIPDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
G
Gene Czarcinski 已提交
1007
    bool ipv6SLAAC;
1008
    char *saddr = NULL, *eaddr = NULL;
1009

1010 1011
    *configstr = NULL;

1012
    /*
1013 1014 1015
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
1016
     *
1017 1018
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
1019
     */
1020 1021 1022 1023 1024 1025

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

1026
    /* create dnsmasq config file appropriate for this network */
1027 1028

    /* Don't forget to update networkxml2conftest :-) */
1029
    virBufferAsprintf(&configbuf,
1030 1031 1032 1033 1034 1035 1036
                      "##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"
1037
                      "strict-order\n",
1038
                      def->name);
1039

1040 1041 1042 1043 1044 1045
    /* if dns is disabled, set its listening port to 0, which
     * tells dnsmasq to not listen
     */
    if (!wantDNS)
        virBufferAddLit(&configbuf, "port=0\n");

1046
    if (wantDNS && def->dns.forwarders) {
1047 1048 1049 1050 1051 1052 1053 1054 1055
        /* 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;

1056 1057
        for (i = 0; i < def->dns.nfwds; i++) {
            virNetworkDNSForwarderPtr fwd = &def->dns.forwarders[i];
1058 1059 1060 1061 1062 1063 1064 1065 1066 1067

            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);
1068
                VIR_FREE(addr);
1069 1070
                if (!fwd->domain)
                    addNoResolv = true;
1071 1072 1073 1074
            } else {
                /* "don't forward requests for this domain" */
                virBufferAddLit(&configbuf, "#\n");
            }
1075
        }
1076 1077
        if (addNoResolv)
            virBufferAddLit(&configbuf, "no-resolv\n");
1078 1079
    }

1080 1081
    if (def->domain) {
        if (def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
1082 1083
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
1084
                              def->domain);
1085
        }
1086
        virBufferAsprintf(&configbuf,
1087 1088
                          "domain=%s\n"
                          "expand-hosts\n",
1089
                          def->domain);
1090
    }
1091

1092
    if (wantDNS &&
1093
        networkDnsmasqConfLocalPTRs(&configbuf, def) < 0)
1094 1095
        goto cleanup;

1096
    if (wantDNS && def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
1097 1098 1099 1100
        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)
1101
         */
1102
        virBufferAddLit(&configbuf, "local=//\n");
1103
    }
1104

1105
    if (pidfile)
1106
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
1107

1108
    /* dnsmasq will *always* listen on localhost unless told otherwise */
P
Pavel Timofeev 已提交
1109
#ifdef __linux__
1110
    virBufferAddLit(&configbuf, "except-interface=lo\n");
P
Pavel Timofeev 已提交
1111 1112 1113 1114
#else
    /* BSD family OSes and Solaris call loopback interface as lo0 */
    virBufferAddLit(&configbuf, "except-interface=lo0\n");
#endif
1115

1116 1117 1118 1119 1120 1121 1122 1123
    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.
         */
1124
        virBufferAsprintf(&configbuf,
1125 1126
                          "bind-dynamic\n"
                          "interface=%s\n",
1127
                          def->bridge);
1128
    } else {
1129
        virBufferAddLit(&configbuf, "bind-interfaces\n");
1130 1131 1132 1133
        /*
         * --interface does not actually work with dnsmasq < 2.47,
         * due to DAD for ipv6 addresses on the interface.
         *
1134
         * virCommandAddArgList(cmd, "--interface", def->bridge, NULL);
1135 1136 1137
         *
         * So listen on all defined IPv[46] addresses
         */
1138
        for (i = 0;
1139
             (tmpipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
1140
             i++) {
1141 1142 1143 1144
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
1145

1146
            /* also part of CVE 2012-3411 - if the host's version of
1147
             * dnsmasq doesn't have bind-dynamic, only allow listening on
1148 1149
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
1150 1151
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
1152 1153 1154 1155
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
1156
                                 "The version of dnsmasq on this host (%d.%d) "
1157 1158 1159 1160
                                 "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 "
1161 1162 1163 1164 1165 1166
                                 "(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);
1167
                VIR_FREE(ipaddr);
1168 1169
                goto cleanup;
            }
1170
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1171 1172 1173
            VIR_FREE(ipaddr);
        }
    }
1174

1175 1176
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1177
     * guaranteed to not work, and set no-resolv so that no dns
1178 1179 1180
     * 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).
1181 1182 1183
     * 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.
1184
     */
1185
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1186
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1187
                        "no-resolv\n");
1188 1189 1190 1191
        if (dnsmasqCapsGet(caps, DNSMASQ_CAPS_RA_PARAM)) {
            /* interface=* (any), interval=0 (default), lifetime=0 (seconds) */
            virBufferAddLit(&configbuf, "ra-param=*,0,0\n");
        }
1192
    }
1193

1194 1195 1196 1197 1198
    if (wantDNS) {
        for (i = 0; i < dns->ntxts; i++) {
            virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
                              dns->txts[i].name,
                              dns->txts[i].value);
1199
        }
1200

1201 1202 1203 1204 1205 1206 1207 1208
        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'"),
1209
                               def->name);
1210 1211 1212 1213 1214 1215
                goto cleanup;
            }
            if (!dns->srvs[i].protocol) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Missing required 'service' "
                                 "attribute in SRV record of network '%s'"),
1216
                               def->name);
1217 1218 1219 1220 1221 1222 1223
                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);
1224

1225 1226 1227 1228 1229 1230 1231
            /* 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.
1232
             */
1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252
            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");
1253
        }
1254 1255
    }

G
Gene Czarcinski 已提交
1256
    /* Find the first dhcp for both IPv4 and IPv6 */
1257
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
1258
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
1259
         i++) {
G
Gene Czarcinski 已提交
1260 1261 1262 1263
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1264 1265
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276
                    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,
1277 1278 1279 1280 1281 1282 1283 1284 1285
                                   _("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
1286 1287 1288 1289
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1290 1291
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304
                    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 "
1305 1306 1307 1308
                 "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
1309 1310 1311 1312 1313
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1314 1315
        int prefix;

1316
        prefix = virNetworkIPDefPrefix(ipdef);
1317 1318 1319
        if (prefix < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
1320
                           def->bridge);
1321 1322
            goto cleanup;
        }
1323
        for (r = 0; r < ipdef->nranges; r++) {
1324 1325
            int thisRange;

1326 1327
            if (!(saddr = virSocketAddrFormat(&ipdef->ranges[r].start)) ||
                !(eaddr = virSocketAddrFormat(&ipdef->ranges[r].end)))
1328
                goto cleanup;
1329

1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6)) {
               virBufferAsprintf(&configbuf, "dhcp-range=%s,%s,%d\n",
                                 saddr, eaddr, prefix);
            } else {
                /* IPv4 - dnsmasq requires a netmask rather than prefix */
                virSocketAddr netmask;
                VIR_AUTOFREE(char *) netmaskStr = NULL;

                if (virSocketAddrPrefixToNetmask(prefix, &netmask, AF_INET) < 0) {
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("Failed to translate bridge '%s' "
                                     "prefix %d to netmask"),
                                   def->bridge, prefix);
                    goto cleanup;
                }

                if (!(netmaskStr = virSocketAddrFormat(&netmask)))
                    goto cleanup;
                virBufferAsprintf(&configbuf, "dhcp-range=%s,%s,%s\n",
                                  saddr, eaddr, netmaskStr);
            }
1351

1352
            VIR_FREE(saddr);
1353
            VIR_FREE(eaddr);
1354
            thisRange = virSocketAddrGetRange(&ipdef->ranges[r].start,
1355 1356
                                              &ipdef->ranges[r].end,
                                              &ipdef->address,
1357
                                              virNetworkIPDefPrefix(ipdef));
1358 1359 1360
            if (thisRange < 0)
                goto cleanup;
            nbleases += thisRange;
1361
        }
1362

1363
        /*
1364 1365 1366 1367
         * 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)
1368 1369
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1370
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1371 1372
            if (!bridgeaddr)
                goto cleanup;
1373 1374 1375
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static",
                              bridgeaddr);
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1376
                virBufferAsprintf(&configbuf, ",%d", prefix);
1377
            virBufferAddLit(&configbuf, "\n");
1378 1379
            VIR_FREE(bridgeaddr);
        }
1380

G
Gene Czarcinski 已提交
1381 1382
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1383

G
Gene Czarcinski 已提交
1384 1385
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
1386
            if (ipdef->nranges || ipdef->nhosts) {
1387
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1388 1389
                virBufferAddLit(&configbuf, "dhcp-authoritative\n");
            }
1390

G
Gene Czarcinski 已提交
1391
            if (ipdef->tftproot) {
1392 1393
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1394
            }
1395

G
Gene Czarcinski 已提交
1396 1397 1398
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1399

1400
                    if (!bootserver)
G
Gene Czarcinski 已提交
1401
                        goto cleanup;
1402
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1403
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1404 1405
                    VIR_FREE(bootserver);
                } else {
1406
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1407 1408 1409 1410 1411
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1412

1413
    if (nbleases > 0)
1414
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1415

G
Gene Czarcinski 已提交
1416 1417
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1418
        goto cleanup;
G
Gene Czarcinski 已提交
1419 1420 1421 1422 1423 1424

    /* 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)
1425 1426
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1427

1428 1429
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1430
     */
1431 1432 1433 1434
    if (wantDNS) {
        virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
                          dctx->addnhostsfile->path);
    }
G
Gene Czarcinski 已提交
1435

1436 1437 1438 1439
    /* 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
1440 1441
    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1442
        if (ipv6def) {
1443
            virBufferAddLit(&configbuf, "enable-ra\n");
1444
        } else {
1445
            for (i = 0;
1446
                 (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i));
1447
                 i++) {
G
Gene Czarcinski 已提交
1448 1449 1450 1451
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1452 1453
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1454 1455
                    VIR_FREE(bridgeaddr);
                }
1456
            }
1457
        }
1458 1459
    }

1460 1461 1462
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1463
    ret = 0;
G
Gene Czarcinski 已提交
1464

1465
 cleanup:
1466 1467
    VIR_FREE(saddr);
    VIR_FREE(eaddr);
1468
    virBufferFreeAndReset(&configbuf);
1469
    return ret;
1470 1471
}

1472

1473
/* build the dnsmasq command line */
1474 1475
static int ATTRIBUTE_NONNULL(3)
networkBuildDhcpDaemonCommandLine(virNetworkDriverStatePtr driver,
1476
                                  virNetworkObjPtr obj,
1477
                                  virCommandPtr *cmdout,
1478 1479
                                  char *pidfile,
                                  dnsmasqContext *dctx)
1480
{
1481
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1482
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1483
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1484
    int ret = -1;
1485 1486
    char *configfile = NULL;
    char *configstr = NULL;
1487
    char *leaseshelper_path = NULL;
1488

1489
    virNetworkObjSetDnsmasqPid(obj, -1);
1490

1491
    if (networkDnsmasqConfContents(obj, pidfile, &configstr,
1492
                                   dctx, dnsmasq_caps) < 0)
1493 1494 1495 1496 1497
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1498
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
1499 1500 1501 1502 1503
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1504 1505
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1506 1507 1508
        goto cleanup;
    }

1509 1510
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1511
                                                  abs_top_builddir "/src",
1512 1513 1514
                                                  LIBEXECDIR)))
        goto cleanup;

1515
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(dnsmasq_caps));
1516
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1517 1518
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1519
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1520
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", def->bridge);
1521

1522
    *cmdout = cmd;
1523
    ret = 0;
1524
 cleanup:
1525
    virObjectUnref(dnsmasq_caps);
1526 1527
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1528
    VIR_FREE(leaseshelper_path);
1529 1530 1531
    return ret;
}

1532

1533
static int
1534
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
1535
                       virNetworkObjPtr obj)
1536
{
1537
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1538 1539 1540
    virNetworkIPDefPtr ipdef;
    size_t i;
    bool needDnsmasq = false;
1541 1542
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
1543
    pid_t dnsmasqPid;
1544
    int ret = -1;
1545
    dnsmasqContext *dctx = NULL;
1546

1547
    /* see if there are any IP addresses that need a dhcp server */
1548
    i = 0;
1549
    while ((ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i))) {
1550
        i++;
1551 1552 1553 1554
        if (ipdef->nranges || ipdef->nhosts)
            needDnsmasq = true;
    }

1555 1556 1557 1558 1559 1560
    if (i == 0) {
        /* no IP addresses at all, so we don't need to run */
        ret = 0;
        goto cleanup;
    }

1561
    if (!needDnsmasq && def->dns.enable == VIR_TRISTATE_BOOL_NO) {
1562
        /* no DHCP services needed, and user disabled DNS service */
1563 1564 1565 1566
        ret = 0;
        goto cleanup;
    }

1567
    if (virFileMakePath(driver->pidDir) < 0) {
1568
        virReportSystemError(errno,
1569
                             _("cannot create directory %s"),
1570
                             driver->pidDir);
1571
        goto cleanup;
1572 1573
    }

1574
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, def->name)))
1575
        goto cleanup;
1576

1577
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1578
        virReportSystemError(errno,
1579
                             _("cannot create directory %s"),
1580
                             driver->dnsmasqStateDir);
1581 1582 1583
        goto cleanup;
    }

1584
    dctx = dnsmasqContextNew(def->name, driver->dnsmasqStateDir);
1585 1586 1587
    if (dctx == NULL)
        goto cleanup;

1588
    if (networkDnsmasqCapsRefresh(driver) < 0)
1589
        goto cleanup;
1590

1591
    ret = networkBuildDhcpDaemonCommandLine(driver, obj, &cmd, pidfile, dctx);
1592 1593 1594 1595 1596
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1597
        goto cleanup;
1598

G
Guido Günther 已提交
1599
    ret = virCommandRun(cmd, NULL);
1600
    if (ret < 0)
1601 1602 1603
        goto cleanup;

    /*
1604 1605 1606 1607 1608
     * 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
1609 1610
     */

1611
    ret = virPidFileRead(driver->pidDir, def->name, &dnsmasqPid);
1612
    if (ret < 0)
1613
        goto cleanup;
1614
    virNetworkObjSetDnsmasqPid(obj, dnsmasqPid);
1615

1616
    ret = 0;
1617
 cleanup:
1618
    VIR_FREE(pidfile);
1619
    virCommandFree(cmd);
1620
    dnsmasqContextFree(dctx);
1621 1622 1623
    return ret;
}

1624

1625 1626
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1627 1628
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1629 1630 1631
 *
 *  Returns 0 on success, -1 on failure.
 */
1632
static int
1633
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
1634
                         virNetworkObjPtr obj)
1635
{
1636
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1637 1638
    int ret = -1;
    size_t i;
1639
    pid_t dnsmasqPid;
1640
    virNetworkIPDefPtr ipdef, ipv4def, ipv6def;
1641
    dnsmasqContext *dctx = NULL;
1642

G
Gene Czarcinski 已提交
1643
    /* if no IP addresses specified, nothing to do */
1644
    if (!virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1645 1646
        return 0;

1647
    /* if there's no running dnsmasq, just start it */
1648 1649
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid <= 0 || (kill(dnsmasqPid, 0) < 0))
1650
        return networkStartDhcpDaemon(driver, obj);
1651

1652 1653
    VIR_INFO("Refreshing dnsmasq for network %s", def->bridge);
    if (!(dctx = dnsmasqContextNew(def->name,
1654
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1655
        goto cleanup;
1656
    }
G
Gene Czarcinski 已提交
1657 1658 1659 1660 1661 1662

    /* 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;
1663
    for (i = 0;
1664
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
1665
         i++) {
G
Gene Czarcinski 已提交
1666 1667
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1668 1669
    }

G
Gene Czarcinski 已提交
1670
    ipv6def = NULL;
1671
    for (i = 0;
1672
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i));
1673
         i++) {
G
Gene Czarcinski 已提交
1674 1675
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1676 1677
    }

G
Gene Czarcinski 已提交
1678
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1679
        goto cleanup;
G
Gene Czarcinski 已提交
1680 1681

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1682
        goto cleanup;
1683

1684
    if (networkBuildDnsmasqHostsList(dctx, &def->dns) < 0)
J
Ján Tomko 已提交
1685
        goto cleanup;
1686 1687

    if ((ret = dnsmasqSave(dctx)) < 0)
1688
        goto cleanup;
1689

1690 1691
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    ret = kill(dnsmasqPid, SIGHUP);
1692
 cleanup:
1693 1694 1695 1696
    dnsmasqContextFree(dctx);
    return ret;
}

1697

1698 1699 1700 1701 1702 1703 1704 1705
/* 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
1706
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
1707
                         virNetworkObjPtr obj)
1708
{
1709
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1710 1711
    pid_t dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);

1712
    /* if there is a running dnsmasq, kill it */
1713
    if (dnsmasqPid > 0) {
1714
        networkKillDaemon(dnsmasqPid, "dnsmasq", def->name);
1715
        virNetworkObjSetDnsmasqPid(obj, -1);
1716
    }
1717
    /* now start dnsmasq if it should be started */
1718
    return networkStartDhcpDaemon(driver, obj);
1719 1720
}

1721

G
Gene Czarcinski 已提交
1722 1723 1724 1725 1726 1727
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";

1728
static int
1729
networkRadvdConfContents(virNetworkObjPtr obj,
1730
                         char **configstr)
1731
{
1732
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
E
Eric Blake 已提交
1733
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1734 1735
    int ret = -1;
    size_t i;
1736
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
1737
    bool v6present = false, dhcp6 = false;
1738 1739

    *configstr = NULL;
1740

G
Gene Czarcinski 已提交
1741
    /* Check if DHCPv6 is needed */
1742
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i)); i++) {
G
Gene Czarcinski 已提交
1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755
        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;
    }

1756 1757 1758
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1759
    virBufferAsprintf(&configbuf, "interface %s\n"
1760 1761
                      "{\n"
                      "  AdvSendAdvert on;\n"
1762
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1763 1764
                      "  AdvManagedFlag %s;\n"
                      "%s",
1765
                      def->bridge,
G
Gene Czarcinski 已提交
1766 1767
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1768 1769

    /* add a section for each IPv6 address in the config */
1770
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i)); i++) {
1771 1772 1773
        int prefix;
        char *netaddr;

1774
        prefix = virNetworkIPDefPrefix(ipdef);
1775
        if (prefix < 0) {
1776 1777
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
1778
                           def->bridge);
1779 1780
            goto cleanup;
        }
1781
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1782
            goto cleanup;
1783
        virBufferAsprintf(&configbuf,
1784
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1785 1786 1787
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1788 1789 1790
        VIR_FREE(netaddr);
    }

1791
    virBufferAddLit(&configbuf, "};\n");
1792

1793
    if (virBufferCheckError(&configbuf) < 0)
1794
        goto cleanup;
1795

1796 1797
    *configstr = virBufferContentAndReset(&configbuf);

1798
    ret = 0;
1799
 cleanup:
1800 1801 1802 1803
    virBufferFreeAndReset(&configbuf);
    return ret;
}

1804

1805
/* write file and return its name (which must be freed by caller) */
1806
static int
1807
networkRadvdConfWrite(virNetworkDriverStatePtr driver,
1808
                      virNetworkObjPtr obj,
1809
                      char **configFile)
1810
{
1811
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1812 1813 1814 1815 1816 1817 1818 1819 1820
    int ret = -1;
    char *configStr = NULL;
    char *myConfigFile = NULL;

    if (!configFile)
        configFile = &myConfigFile;

    *configFile = NULL;

1821
    if (networkRadvdConfContents(obj, &configStr) < 0)
1822 1823 1824 1825
        goto cleanup;

    if (!configStr) {
        ret = 0;
1826 1827 1828 1829
        goto cleanup;
    }

    /* construct the filename */
1830
    if (!(*configFile = networkRadvdConfigFileName(driver, def->name)))
1831 1832
        goto cleanup;
    /* write the file */
1833
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1834 1835
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1836 1837 1838 1839 1840
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1841
 cleanup:
1842 1843 1844 1845 1846
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

1847

1848
static int
1849
networkStartRadvd(virNetworkDriverStatePtr driver,
1850
                  virNetworkObjPtr obj)
1851
{
1852
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1853
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1854
    pid_t radvdPid;
1855 1856 1857 1858 1859 1860
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

1861
    virNetworkObjSetRadvdPid(obj, -1);
1862

G
Gene Czarcinski 已提交
1863
    /* Is dnsmasq handling RA? */
1864
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
G
Gene Czarcinski 已提交
1865 1866 1867 1868
        ret = 0;
        goto cleanup;
    }

1869
    if (!virNetworkDefGetIPByIndex(def, AF_INET6, 0)) {
1870 1871 1872 1873 1874 1875 1876 1877 1878 1879
        /* 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);
1880 1881 1882
        goto cleanup;
    }

1883
    if (virFileMakePath(driver->pidDir) < 0) {
1884 1885
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1886
                             driver->pidDir);
1887 1888
        goto cleanup;
    }
1889
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1890 1891
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1892
                             driver->radvdStateDir);
1893 1894 1895 1896
        goto cleanup;
    }

    /* construct pidfile name */
1897
    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
1898
        goto cleanup;
1899
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1900 1901
        goto cleanup;

1902
    if (networkRadvdConfWrite(driver, obj, &configfile) < 0)
1903 1904
        goto cleanup;

1905 1906 1907 1908
    /* 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
1909
     * virPidFileRead() below will fail if we use them).
1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924
     * 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;

1925
    if (virPidFileRead(driver->pidDir, radvdpidbase, &radvdPid) < 0)
1926
        goto cleanup;
1927
    virNetworkObjSetRadvdPid(obj, radvdPid);
1928 1929

    ret = 0;
1930
 cleanup:
1931
    virObjectUnref(dnsmasq_caps);
1932 1933 1934 1935 1936 1937 1938
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1939

1940
static int
1941
networkRefreshRadvd(virNetworkDriverStatePtr driver,
1942
                    virNetworkObjPtr obj)
1943
{
1944
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1945
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
G
Gene Czarcinski 已提交
1946
    char *radvdpidbase;
1947
    pid_t radvdPid;
G
Gene Czarcinski 已提交
1948 1949

    /* Is dnsmasq handling RA? */
1950 1951
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
        virObjectUnref(dnsmasq_caps);
1952 1953
        radvdPid = virNetworkObjGetRadvdPid(obj);
        if (radvdPid <= 0)
G
Gene Czarcinski 已提交
1954 1955
            return 0;
        /* radvd should not be running but in case it is */
1956 1957
        if ((networkKillDaemon(radvdPid, "radvd", def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(def->name))
G
Gene Czarcinski 已提交
1958
             != NULL)) {
1959
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1960 1961
            VIR_FREE(radvdpidbase);
        }
1962
        virNetworkObjSetRadvdPid(obj, -1);
G
Gene Czarcinski 已提交
1963 1964
        return 0;
    }
1965
    virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1966

1967
    /* if there's no running radvd, just start it */
1968 1969
    radvdPid = virNetworkObjGetRadvdPid(obj);
    if (radvdPid <= 0 || (kill(radvdPid, 0) < 0))
1970
        return networkStartRadvd(driver, obj);
1971

1972
    if (!virNetworkDefGetIPByIndex(def, AF_INET6, 0)) {
1973 1974 1975 1976
        /* no IPv6 addresses, so we don't need to run radvd */
        return 0;
    }

1977
    if (networkRadvdConfWrite(driver, obj, NULL) < 0)
1978 1979
        return -1;

1980
    return kill(radvdPid, SIGHUP);
1981 1982
}

1983

1984 1985
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1986
static int
1987
networkRestartRadvd(virNetworkObjPtr obj)
1988
{
1989
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1990
    char *radvdpidbase;
1991
    pid_t radvdPid = virNeworkObjGetRadvdPid(obj);
1992 1993

    /* if there is a running radvd, kill it */
1994
    if (radvdPid > 0) {
1995 1996 1997 1998
        /* 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).
         */
1999 2000
        if ((networkKillDaemon(radvdPid, "radvd", def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(def->name))
2001
             != NULL)) {
2002
            virPidFileDelete(driver->pidDir, radvdpidbase);
2003 2004
            VIR_FREE(radvdpidbase);
        }
2005
        virNetworkObjSetRadvdPid(obj, -1);
2006 2007
    }
    /* now start radvd if it should be started */
2008
    return networkStartRadvd(obj);
2009 2010 2011
}
#endif /* #if 0 */

2012

2013
static int
2014
networkRefreshDaemonsHelper(virNetworkObjPtr obj,
2015
                            void *opaque)
2016
{
2017
    virNetworkDefPtr def;
2018
    virNetworkDriverStatePtr driver = opaque;
2019

2020
    virObjectLock(obj);
2021
    def = virNetworkObjGetDef(obj);
2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049
    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;
        }
2050
    }
2051 2052

 cleanup:
2053
    virObjectUnlock(obj);
2054 2055 2056
    return 0;
}

2057

2058 2059 2060 2061
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
2062
networkRefreshDaemons(virNetworkDriverStatePtr driver)
2063 2064
{
    VIR_INFO("Refreshing network daemons");
2065 2066
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
2067
                             driver);
2068
}
2069

2070

2071
static int
2072
networkReloadFirewallRulesHelper(virNetworkObjPtr obj,
2073 2074
                                 void *opaque ATTRIBUTE_UNUSED)
{
2075
    virNetworkDefPtr def;
2076

2077
    virObjectLock(obj);
2078
    def = virNetworkObjGetDef(obj);
2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104
    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;
2105 2106
        }
    }
2107 2108

 cleanup:
2109
    virObjectUnlock(obj);
2110
    return 0;
2111 2112
}

2113

2114
static void
2115
networkReloadFirewallRules(virNetworkDriverStatePtr driver, bool startup)
2116
{
2117
    VIR_INFO("Reloading iptables rules");
2118 2119 2120 2121
    /* Ideally we'd not even register the driver when unprivilegd
     * but until we untangle the virt driver that's not viable */
    if (!driver->privileged)
        return;
2122
    networkPreReloadFirewallRules(startup);
2123 2124 2125
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
2126
    networkPostReloadFirewallRules(startup);
2127 2128
}

2129

2130
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
2131
static int
2132 2133
networkEnableIPForwarding(bool enableIPv4,
                          bool enableIPv6)
2134
{
2135
    int ret = 0;
2136 2137 2138 2139
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
2140
                           &enabled, sizeof(enabled));
2141 2142
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
2143
                           &enabled, sizeof(enabled));
2144
#else
2145
    if (enableIPv4)
2146
        ret = virFileWriteStr(SYSCTL_PATH "/net/ipv4/ip_forward", "1\n", 0);
2147
    if (enableIPv6 && ret == 0)
2148 2149
        ret = virFileWriteStr(SYSCTL_PATH "/net/ipv6/conf/all/forwarding", "1\n", 0);

2150
#endif
2151
    return ret;
2152 2153
}

2154

2155
static int
2156
networkSetIPv6Sysctls(virNetworkObjPtr obj)
2157
{
2158
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2159 2160
    char *field = NULL;
    int ret = -1;
2161
    bool enableIPv6 = !!virNetworkDefGetIPByIndex(def, AF_INET6, 0);
2162

2163 2164 2165 2166 2167
    /* 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",
2168
                    def->bridge) < 0)
2169
       goto cleanup;
2170

2171 2172
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
2173
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
2174
                      def->bridge);
2175 2176 2177
        ret = 0;
        goto cleanup;
    }
2178

2179 2180 2181
    if (virFileWriteStr(field, enableIPv6 ? "0" : "1", 0) < 0) {
        virReportSystemError(errno,
                             _("cannot write to %s to enable/disable IPv6 "
2182
                               "on bridge %s"), field, def->bridge);
2183
        goto cleanup;
2184
    }
2185
    VIR_FREE(field);
2186

2187 2188
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
2189 2190 2191 2192 2193 2194
     */

    /* 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",
2195
                    def->bridge) < 0)
2196 2197
        goto cleanup;

2198
    if (virFileWriteStr(field, "0", 0) < 0) {
2199
        virReportSystemError(errno,
2200 2201 2202 2203 2204
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

2205 2206 2207 2208
    /* 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",
2209
                    def->bridge) < 0)
2210 2211
        goto cleanup;

2212
    if (virFileWriteStr(field, "0", 0) < 0) {
2213
        virReportSystemError(errno,
2214
                             _("cannot disable %s"), field);
2215 2216 2217 2218
        goto cleanup;
    }

    ret = 0;
2219
 cleanup:
2220 2221 2222 2223
    VIR_FREE(field);
    return ret;
}

2224

2225
/* add an IP address to a bridge */
2226
static int
2227
networkAddAddrToBridge(virNetworkObjPtr obj,
2228
                       virNetworkIPDefPtr ipdef)
2229
{
2230
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2231
    int prefix = virNetworkIPDefPrefix(ipdef);
2232 2233

    if (prefix < 0) {
2234 2235
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
2236
                       def->bridge);
2237 2238 2239
        return -1;
    }

2240
    if (virNetDevIPAddrAdd(def->bridge, &ipdef->address, NULL, prefix) < 0)
2241 2242 2243 2244 2245
        return -1;

    return 0;
}

2246 2247

static int
2248
networkStartHandleMACTableManagerMode(virNetworkObjPtr obj,
2249 2250
                                      const char *macTapIfName)
{
2251 2252
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
    const char *brname = def->bridge;
2253 2254

    if (brname &&
2255
        def->macTableManager == VIR_NETWORK_BRIDGE_MAC_TABLE_MANAGER_LIBVIRT) {
2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268
        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;
}


2269 2270
/* add an IP (static) route to a bridge */
static int
2271
networkAddRouteToBridge(virNetworkObjPtr obj,
2272
                        virNetDevIPRoutePtr routedef)
2273
{
2274
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2275 2276 2277 2278
    int prefix = virNetDevIPRouteGetPrefix(routedef);
    unsigned int metric = virNetDevIPRouteGetMetric(routedef);
    virSocketAddrPtr addr = virNetDevIPRouteGetAddress(routedef);
    virSocketAddrPtr gateway = virNetDevIPRouteGetGateway(routedef);
2279 2280 2281 2282 2283

    if (prefix < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' has an invalid netmask "
                         "or IP address in route definition"),
2284
                       def->name);
2285 2286 2287
        return -1;
    }

2288
    if (virNetDevIPRouteAdd(def->bridge, addr, prefix, gateway, metric) < 0)
2289
        return -1;
2290

2291 2292 2293
    return 0;
}

2294
static int
2295
networkWaitDadFinish(virNetworkObjPtr obj)
2296
{
2297
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2298
    virNetworkIPDefPtr ipdef;
2299 2300 2301 2302
    virSocketAddrPtr *addrs = NULL, addr = NULL;
    size_t naddrs = 0;
    int ret = -1;

2303
    VIR_DEBUG("Begin waiting for IPv6 DAD on network %s", def->name);
2304

2305
    while ((ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, naddrs))) {
2306 2307 2308 2309 2310
        addr = &ipdef->address;
        if (VIR_APPEND_ELEMENT_COPY(addrs, naddrs, addr) < 0)
            goto cleanup;
    }

2311
    ret = (naddrs == 0) ? 0 : virNetDevIPWaitDadFinish(addrs, naddrs);
2312 2313 2314 2315

 cleanup:
    VIR_FREE(addrs);
    VIR_DEBUG("Finished waiting for IPv6 DAD on network %s with status %d",
2316
              def->name, ret);
2317 2318 2319
    return ret;
}

2320

2321
static int
2322
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
2323
                           virNetworkObjPtr obj)
2324
{
2325
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2326
    size_t i;
2327
    bool v4present = false, v6present = false;
2328
    virErrorPtr save_err = NULL;
2329
    virNetworkIPDefPtr ipdef;
2330
    virNetDevIPRoutePtr routedef;
2331
    char *macTapIfName = NULL;
2332
    virMacMapPtr macmap;
M
Michal Privoznik 已提交
2333
    char *macMapFile = NULL;
2334
    int tapfd = -1;
2335
    pid_t dnsmasqPid;
2336

2337
    /* Check to see if any network IP collides with an existing route */
2338
    if (networkCheckRouteCollision(def) < 0)
2339 2340
        return -1;

2341
    /* Create and configure the bridge device */
2342
    if (!def->bridge) {
2343 2344 2345 2346 2347 2348 2349 2350 2351 2352
        /* 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"),
2353
                       def->name);
2354 2355
        return -1;
    }
2356
    if (virNetDevBridgeCreate(def->bridge) < 0)
2357 2358
        return -1;

2359
    if (def->mac_specified) {
2360 2361 2362 2363 2364 2365
        /* 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.
         */
2366
        macTapIfName = networkBridgeDummyNicName(def->bridge);
2367
        if (!macTapIfName)
2368
            goto err0;
2369
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2370 2371
        if (virNetDevTapCreateInBridgePort(def->bridge,
                                           &macTapIfName, &def->mac,
2372
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2373
                                           NULL, def->mtu, NULL,
2374 2375 2376
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2377 2378 2379 2380 2381
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2382
    if (!(macMapFile = virMacMapFileName(driver->dnsmasqStateDir,
2383
                                         def->bridge)) ||
2384
        !(macmap = virMacMapNew(macMapFile)))
M
Michal Privoznik 已提交
2385 2386
        goto err1;

2387
    virNetworkObjSetMacMap(obj, macmap);
2388
    macmap = NULL;
2389

2390
    /* Set bridge options */
2391 2392 2393 2394

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2395
    if (virNetDevBridgeSetSTPDelay(def->bridge, def->delay * 1000) < 0)
2396
        goto err1;
2397

2398
    if (virNetDevBridgeSetSTP(def->bridge, def->stp ? true : false) < 0)
2399
        goto err1;
2400

2401 2402 2403
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
2404
    if (networkSetIPv6Sysctls(obj) < 0)
2405
        goto err1;
2406

2407
    /* Add "once per network" rules */
2408 2409
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN &&
        networkAddFirewallRules(def) < 0)
2410 2411
        goto err1;

2412
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i)); i++) {
2413
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2414
            v4present = true;
2415
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2416
            v6present = true;
2417

2418
        /* Add the IP address/netmask to the bridge */
2419
        if (networkAddAddrToBridge(obj, ipdef) < 0)
2420
            goto err2;
2421 2422
    }

2423
    if (networkStartHandleMACTableManagerMode(obj, macTapIfName) < 0)
2424 2425
        goto err2;

2426
    /* Bring up the bridge interface */
2427
    if (virNetDevSetOnline(def->bridge, 1) < 0)
2428
        goto err2;
2429

2430
    for (i = 0; i < def->nroutes; i++) {
2431 2432
        virSocketAddrPtr gateway = NULL;

2433
        routedef = def->routes[i];
2434
        gateway = virNetDevIPRouteGetGateway(routedef);
2435

2436 2437 2438
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2439
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2440
            if (networkAddRouteToBridge(obj, routedef) < 0) {
2441 2442 2443 2444 2445 2446
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2447
    /* If forward.type != NONE, turn on global IP forwarding */
2448
    if (def->forward.type != VIR_NETWORK_FORWARD_NONE) {
2449
        if (v6present && !virNetDevIPCheckIPv6Forwarding())
2450 2451 2452 2453 2454 2455 2456
            goto err3; /* Precise error message already provided */

        if (networkEnableIPForwarding(v4present, v6present) < 0) {
            virReportSystemError(errno, "%s",
                                 _("failed to enable IP forwarding"));
            goto err3;
        }
2457 2458
    }

2459

2460
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2461
    if ((v4present || v6present) &&
2462
        networkStartDhcpDaemon(driver, obj) < 0)
2463
        goto err3;
2464

2465
    /* start radvd if there are any ipv6 addresses */
2466
    if (v6present && networkStartRadvd(driver, obj) < 0)
2467 2468
        goto err4;

2469 2470 2471
    /* dnsmasq does not wait for DAD to complete before daemonizing,
     * so we need to wait for it ourselves.
     */
2472
    if (v6present && networkWaitDadFinish(obj) < 0)
2473 2474 2475 2476
        goto err4;

    /* DAD has finished, dnsmasq is now bound to the
     * bridge's IPv6 address, so we can set the dummy tun down.
2477 2478 2479 2480 2481 2482 2483
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
            goto err4;
        VIR_FORCE_CLOSE(tapfd);
    }

2484
    if (virNetDevBandwidthSet(def->bridge, def->bandwidth, true, true) < 0)
2485 2486
        goto err5;

2487
    VIR_FREE(macTapIfName);
M
Michal Privoznik 已提交
2488
    VIR_FREE(macMapFile);
2489 2490 2491

    return 0;

2492
 err5:
2493 2494
    if (def->bandwidth)
       virNetDevBandwidthClear(def->bridge);
2495

2496 2497 2498 2499
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2500 2501 2502 2503
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid > 0) {
        kill(dnsmasqPid, SIGTERM);
        virNetworkObjSetDnsmasqPid(obj, -1);
2504 2505
    }

2506 2507 2508
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2509
    ignore_value(virNetDevSetOnline(def->bridge, 0));
2510

2511 2512 2513
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2514 2515
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(def);
2516 2517

 err1:
2518 2519 2520
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2521
    if (macTapIfName) {
2522
        VIR_FORCE_CLOSE(tapfd);
2523
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2524 2525
        VIR_FREE(macTapIfName);
    }
2526
    virNetworkObjUnrefMacMap(obj);
M
Michal Privoznik 已提交
2527
    VIR_FREE(macMapFile);
2528 2529

 err0:
2530 2531
    if (!save_err)
        save_err = virSaveLastError();
2532
    ignore_value(virNetDevBridgeDelete(def->bridge));
2533

2534 2535 2536 2537
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2538
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2539 2540 2541
    return -1;
}

2542

2543 2544
static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
2545
                              virNetworkObjPtr obj)
2546
{
2547
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2548 2549 2550
    pid_t radvdPid;
    pid_t dnsmasqPid;

2551 2552
    if (def->bandwidth)
        virNetDevBandwidthClear(def->bridge);
2553

2554
    virNetworkObjUnrefMacMap(obj);
M
Michal Privoznik 已提交
2555

2556 2557
    radvdPid = virNetworkObjGetRadvdPid(obj);
    if (radvdPid > 0) {
2558 2559
        char *radvdpidbase;

2560
        kill(radvdPid, SIGTERM);
2561
        /* attempt to delete the pidfile we created */
2562
        if ((radvdpidbase = networkRadvdPidfileBasename(def->name))) {
2563
            virPidFileDelete(driver->pidDir, radvdpidbase);
2564 2565 2566 2567
            VIR_FREE(radvdpidbase);
        }
    }

2568 2569 2570
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid > 0)
        kill(dnsmasqPid, SIGTERM);
2571

2572 2573
    if (def->mac_specified) {
        char *macTapIfName = networkBridgeDummyNicName(def->bridge);
2574
        if (macTapIfName) {
2575
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2576 2577 2578 2579
            VIR_FREE(macTapIfName);
        }
    }

2580
    ignore_value(virNetDevSetOnline(def->bridge, 0));
2581

2582 2583
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(def);
2584

2585
    ignore_value(virNetDevBridgeDelete(def->bridge));
2586

2587
    /* See if its still alive and really really kill it */
2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598
    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);
2599

2600 2601 2602
    return 0;
}

2603

2604
static int
2605
networkStartNetworkBridge(virNetworkObjPtr obj)
2606 2607 2608 2609 2610
{
    /* 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.
     */
2611
    return networkStartHandleMACTableManagerMode(obj, NULL);
2612 2613
}

2614

2615
static int
2616
networkShutdownNetworkBridge(virNetworkObjPtr obj ATTRIBUTE_UNUSED)
2617 2618 2619 2620 2621 2622 2623 2624 2625
{
    /* 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;
}


2626 2627 2628 2629 2630 2631 2632 2633 2634
/* 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;
2635
    unsigned int maxVirtFns = 0;
2636 2637 2638 2639 2640 2641
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2642 2643 2644
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2645 2646
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev, &vfNames,
                                      &virtFns, &numVirtFns, &maxVirtFns)) < 0) {
2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661
        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];

2662
        switch ((virNetworkForwardType) netdef->forward.type) {
2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692
        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:
2693
        case VIR_NETWORK_FORWARD_OPEN:
2694 2695
            /* by definition these will never be encountered here */
            break;
2696 2697 2698 2699 2700

        case VIR_NETWORK_FORWARD_LAST:
        default:
            virReportEnumRangeError(virNetworkForwardType, netdef->forward.type);
            goto cleanup;
2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737
        }
    }

    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;
}


2738
static int
2739
networkStartNetworkExternal(virNetworkObjPtr obj)
2740 2741
{
    /* put anything here that needs to be done each time a network of
2742
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2743 2744 2745
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2746
    return networkCreateInterfacePool(virNetworkObjGetDef(obj));
2747 2748
}

2749 2750

static int
2751
networkShutdownNetworkExternal(virNetworkObjPtr obj ATTRIBUTE_UNUSED)
2752 2753
{
    /* put anything here that needs to be done each time a network of
2754
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2755 2756 2757 2758 2759 2760
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

2761

2762
static int
2763
networkStartNetwork(virNetworkDriverStatePtr driver,
2764
                    virNetworkObjPtr obj)
2765
{
2766
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2767 2768
    int ret = -1;

2769
    VIR_DEBUG("driver=%p, network=%p", driver, obj);
2770

2771
    if (virNetworkObjIsActive(obj)) {
2772 2773
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2774
        return ret;
2775 2776
    }

2777 2778 2779
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2780
    if (virNetworkObjSetDefTransient(obj, true) < 0)
2781
        goto cleanup;
2782

2783 2784
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2785
    if (networkRunHook(obj, NULL, NULL,
2786 2787 2788 2789
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2790
    switch ((virNetworkForwardType) def->forward.type) {
2791 2792 2793 2794

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2795
    case VIR_NETWORK_FORWARD_OPEN:
2796
        if (networkStartNetworkVirtual(driver, obj) < 0)
2797
            goto cleanup;
2798 2799 2800
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2801
        if (def->bridge) {
2802
            if (networkStartNetworkBridge(obj) < 0)
2803 2804 2805 2806 2807 2808 2809
                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).
         */
2810 2811
        ATTRIBUTE_FALLTHROUGH;

2812 2813 2814
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2815
    case VIR_NETWORK_FORWARD_HOSTDEV:
2816
        if (networkStartNetworkExternal(obj) < 0)
2817
            goto cleanup;
2818
        break;
2819 2820 2821 2822 2823

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        goto cleanup;
2824 2825
    }

2826
    /* finally we can call the 'started' hook script if any */
2827
    if (networkRunHook(obj, NULL, NULL,
2828 2829 2830 2831
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2832 2833 2834
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2835
    VIR_DEBUG("Writing network status to disk");
2836
    if (virNetworkObjSaveStatus(driver->stateDir, obj) < 0)
2837
        goto cleanup;
2838

2839
    virNetworkObjSetActive(obj, true);
2840
    VIR_INFO("Network '%s' started up", def->name);
2841
    ret = 0;
2842

2843
 cleanup:
2844
    if (ret < 0) {
2845
        virNetworkObjUnsetDefTransient(obj);
2846 2847
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2848
        networkShutdownNetwork(driver, obj);
2849 2850 2851 2852 2853 2854 2855
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2856

2857 2858
static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
2859
                       virNetworkObjPtr obj)
2860
{
2861
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2862 2863 2864
    int ret = 0;
    char *stateFile;

2865
    VIR_INFO("Shutting down network '%s'", def->name);
2866

2867
    if (!virNetworkObjIsActive(obj))
2868 2869
        return 0;

2870
    stateFile = virNetworkConfigFile(driver->stateDir, def->name);
2871 2872 2873 2874 2875 2876
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2877
    switch ((virNetworkForwardType) def->forward.type) {
2878 2879 2880 2881

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2882
    case VIR_NETWORK_FORWARD_OPEN:
2883
        ret = networkShutdownNetworkVirtual(driver, obj);
2884 2885 2886
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2887
        if (def->bridge) {
2888
            ret = networkShutdownNetworkBridge(obj);
2889 2890 2891 2892 2893 2894
            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).
         */
2895 2896
        ATTRIBUTE_FALLTHROUGH;

2897 2898 2899
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2900
    case VIR_NETWORK_FORWARD_HOSTDEV:
2901
        ret = networkShutdownNetworkExternal(obj);
2902
        break;
2903 2904 2905 2906 2907

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        return -1;
2908 2909
    }

2910
    /* now that we know it's stopped call the hook if present */
2911
    networkRunHook(obj, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2912 2913
                   VIR_HOOK_SUBOP_END);

2914
    virNetworkObjSetActive(obj, false);
2915
    virNetworkObjUnsetDefTransient(obj);
2916
    return ret;
2917 2918 2919
}


2920 2921 2922
static virNetworkPtr
networkLookupByUUID(virConnectPtr conn,
                    const unsigned char *uuid)
2923
{
2924
    virNetworkDriverStatePtr driver = networkGetDriver();
2925
    virNetworkObjPtr obj;
2926
    virNetworkDefPtr def;
2927
    virNetworkPtr net = NULL;
2928

2929 2930
    obj = virNetworkObjFindByUUID(driver->networks, uuid);
    if (!obj) {
2931 2932
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2933
        virReportError(VIR_ERR_NO_NETWORK,
2934 2935
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2936
        goto cleanup;
2937
    }
2938
    def = virNetworkObjGetDef(obj);
2939

2940
    if (virNetworkLookupByUUIDEnsureACL(conn, def) < 0)
2941 2942
        goto cleanup;

2943
    net = virGetNetwork(conn, def->name, def->uuid);
2944

2945
 cleanup:
2946 2947
    virNetworkObjEndAPI(&obj);
    return net;
2948 2949
}

2950 2951 2952 2953

static virNetworkPtr
networkLookupByName(virConnectPtr conn,
                    const char *name)
2954
{
2955
    virNetworkDriverStatePtr driver = networkGetDriver();
2956
    virNetworkObjPtr obj;
2957
    virNetworkDefPtr def;
2958
    virNetworkPtr net = NULL;
2959

2960 2961
    obj = virNetworkObjFindByName(driver->networks, name);
    if (!obj) {
2962 2963
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2964
        goto cleanup;
2965
    }
2966
    def = virNetworkObjGetDef(obj);
2967

2968
    if (virNetworkLookupByNameEnsureACL(conn, def) < 0)
2969 2970
        goto cleanup;

2971
    net = virGetNetwork(conn, def->name, def->uuid);
2972

2973
 cleanup:
2974 2975
    virNetworkObjEndAPI(&obj);
    return net;
2976 2977
}

2978 2979 2980

static int
networkConnectNumOfNetworks(virConnectPtr conn)
2981
{
2982
    virNetworkDriverStatePtr driver = networkGetDriver();
2983
    int nactive;
2984

2985 2986 2987
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2988 2989 2990 2991
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
2992

2993 2994 2995
    return nactive;
}

2996 2997 2998 2999

static int
networkConnectListNetworks(virConnectPtr conn,
                           char **const names,
3000
                           int maxnames)
3001 3002
{
    virNetworkDriverStatePtr driver = networkGetDriver();
3003
    int got = 0;
3004

3005 3006 3007
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

3008
    got = virNetworkObjListGetNames(driver->networks,
3009
                                    true, names, maxnames,
3010 3011
                                    virConnectListNetworksCheckACL,
                                    conn);
3012

3013 3014 3015
    return got;
}

3016 3017 3018

static int
networkConnectNumOfDefinedNetworks(virConnectPtr conn)
3019
{
3020
    virNetworkDriverStatePtr driver = networkGetDriver();
3021
    int ninactive = 0;
3022

3023 3024 3025
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

3026 3027 3028 3029
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
3030

3031 3032 3033
    return ninactive;
}

3034 3035 3036 3037

static int
networkConnectListDefinedNetworks(virConnectPtr conn,
                                  char **const names,
3038
                                  int maxnames)
3039 3040
{
    virNetworkDriverStatePtr driver = networkGetDriver();
3041
    int got = 0;
3042

3043 3044 3045
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

3046
    got = virNetworkObjListGetNames(driver->networks,
3047
                                    false, names, maxnames,
3048 3049
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
3050 3051 3052
    return got;
}

3053

3054
static int
3055 3056 3057
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
3058
{
3059
    virNetworkDriverStatePtr driver = networkGetDriver();
3060 3061 3062 3063
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

3064 3065 3066
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

3067
    ret = virNetworkObjListExport(conn, driver->networks, nets,
3068 3069
                                  virConnectListAllNetworksCheckACL,
                                  flags);
3070

3071
 cleanup:
3072 3073
    return ret;
}
3074

3075

3076 3077 3078 3079 3080 3081 3082 3083
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
3084
    virNetworkDriverStatePtr driver = networkGetDriver();
3085 3086 3087 3088 3089 3090
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
3091
                                       net, eventID, callback,
3092 3093 3094
                                       opaque, freecb, &ret) < 0)
        ret = -1;

3095
 cleanup:
3096 3097 3098
    return ret;
}

3099

3100 3101 3102 3103
static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
3104
    virNetworkDriverStatePtr driver = networkGetDriver();
3105 3106 3107 3108 3109
    int ret = -1;

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

3110 3111
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
3112
                                        callbackID, true) < 0)
3113 3114 3115
        goto cleanup;

    ret = 0;
3116

3117
 cleanup:
3118 3119 3120
    return ret;
}

3121 3122 3123

static int
networkIsActive(virNetworkPtr net)
3124 3125 3126 3127
{
    virNetworkObjPtr obj;
    int ret = -1;

3128 3129
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3130

3131
    if (virNetworkIsActiveEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
3132 3133
        goto cleanup;

3134 3135
    ret = virNetworkObjIsActive(obj);

3136
 cleanup:
3137
    virNetworkObjEndAPI(&obj);
3138 3139 3140
    return ret;
}

3141 3142 3143

static int
networkIsPersistent(virNetworkPtr net)
3144 3145 3146 3147
{
    virNetworkObjPtr obj;
    int ret = -1;

3148 3149
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3150

3151
    if (virNetworkIsPersistentEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
3152 3153
        goto cleanup;

3154
    ret = virNetworkObjIsPersistent(obj);
3155

3156
 cleanup:
3157
    virNetworkObjEndAPI(&obj);
3158 3159 3160 3161
    return ret;
}


3162 3163
/*
 * networkFindUnusedBridgeName() - try to find a bridge name that is
3164 3165 3166
 * 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.
3167 3168 3169 3170 3171 3172 3173
 */
static int
networkFindUnusedBridgeName(virNetworkObjListPtr nets,
                            virNetworkDefPtr def)
{
    int ret = -1, id = 0;
    char *newname = NULL;
3174 3175 3176 3177 3178
    const char *templ = "virbr%d";
    const char *p;

    if (def->bridge &&
        (p = strchr(def->bridge, '%')) == strrchr(def->bridge, '%') &&
3179
        p && p[1] == 'd')
3180
        templ = def->bridge;
3181 3182 3183 3184

    do {
        if (virAsprintf(&newname, templ, id) < 0)
            goto cleanup;
3185 3186 3187 3188 3189
        /* 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).
         */
3190
        if (!(virNetworkObjBridgeInUse(nets, newname, def->name) ||
3191
              virNetDevExists(newname) == 1)) {
3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224
            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")) {
3225
        if (virNetworkObjBridgeInUse(nets, def->bridge, def->name)) {
3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242
            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;
}


3243
static int
3244
networkValidate(virNetworkDriverStatePtr driver,
3245
                virNetworkDefPtr def)
3246
{
3247
    size_t i, j;
3248 3249
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
3250
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
3251
    bool ipv4def = false, ipv6def = false;
3252
    bool bandwidthAllowed = true;
3253
    bool usesInterface = false, usesAddress = false;
3254

3255 3256 3257
    if (virXMLCheckIllegalChars("name", def->name, "\n") < 0)
        return -1;

3258 3259 3260
    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
3261 3262 3263 3264 3265
    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:
3266 3267 3268 3269
        /* 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)
3270 3271 3272
            return -1;

        virNetworkSetBridgeMacAddr(def);
3273 3274 3275 3276 3277 3278 3279
        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:
3280
        /* They are also the only types that currently support setting
3281 3282
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
3283
         */
3284 3285 3286 3287 3288 3289 3290 3291
        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;
        }
3292
        if (virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0)) {
3293 3294 3295 3296
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3297
                           virNetworkForwardTypeToString(def->forward.type));
3298 3299
            return -1;
        }
3300
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
3301 3302 3303 3304
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3305
                           virNetworkForwardTypeToString(def->forward.type));
3306 3307 3308 3309 3310 3311 3312
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3313
                           virNetworkForwardTypeToString(def->forward.type));
3314 3315
            return -1;
        }
3316 3317 3318 3319 3320 3321 3322 3323
        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;
        }
3324
        bandwidthAllowed = false;
3325 3326 3327 3328 3329 3330
        break;

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        return -1;
3331 3332
    }

3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345
    /* 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++) {
3346 3347 3348
        virNetworkForwardIfDefPtr iface = &def->forward.ifs[i];
        char *sysfs_path = NULL;

3349
        switch ((virNetworkForwardHostdevDeviceType)iface->type) {
3350 3351
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV:
            usesInterface = true;
3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362

            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;
            }
3363
            break;
3364 3365

        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI: {
3366
            usesAddress = true;
3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391

            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);
3392
            break;
3393 3394
        }

3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407
        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
3408 3409 3410
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
3411
    for (i = 0;
3412
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
3413
         i++) {
G
Gene Czarcinski 已提交
3414 3415 3416 3417 3418
        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 -- "
3419 3420
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437
                    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;
                }
3438 3439 3440
            }
        }
    }
3441 3442 3443 3444 3445 3446

    /* 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.
     */

3447 3448 3449
    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
3450 3451
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
3452
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH));
3453 3454

    vlanUsed = def->vlan.nTags > 0;
3455 3456
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3457 3458 3459 3460 3461
            /* 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.
             */
3462
            if (def->portGroups[i].virtPortProfile) {
3463
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3464
                    def->portGroups[i].virtPortProfile->virtPortType
3465 3466 3467 3468 3469 3470 3471
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3472
        }
3473
        if (def->portGroups[i].isDefault) {
3474 3475 3476 3477 3478
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3479
                               def->name, defaultPortGroup->name,
3480
                               def->portGroups[i].name);
3481
                return -1;
3482
            }
3483
            defaultPortGroup = &def->portGroups[i];
3484
        }
3485 3486 3487 3488 3489 3490 3491 3492 3493
        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;
            }
        }
3494 3495 3496 3497 3498 3499 3500 3501
        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;
        }
3502
    }
3503 3504 3505 3506 3507 3508 3509
    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.
         */
3510 3511 3512 3513 3514 3515
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529

    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;
            }
        }
    }
3530 3531 3532
    return 0;
}

3533 3534 3535 3536

static virNetworkPtr
networkCreateXML(virConnectPtr conn,
                 const char *xml)
3537
{
3538
    virNetworkDriverStatePtr driver = networkGetDriver();
3539
    virNetworkDefPtr newDef;
3540
    virNetworkObjPtr obj = NULL;
3541
    virNetworkDefPtr def;
3542
    virNetworkPtr net = NULL;
3543
    virObjectEventPtr event = NULL;
3544

3545
    if (!(newDef = virNetworkDefParseString(xml)))
3546
        goto cleanup;
3547

3548
    if (virNetworkCreateXMLEnsureACL(conn, newDef) < 0)
3549 3550
        goto cleanup;

3551
    if (networkValidate(driver, newDef) < 0)
J
Ján Tomko 已提交
3552
        goto cleanup;
3553

3554 3555 3556
    /* 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.
3557
     */
3558
    if (!(obj = virNetworkObjAssignDef(driver->networks, newDef,
3559 3560
                                       VIR_NETWORK_OBJ_LIST_ADD_LIVE |
                                       VIR_NETWORK_OBJ_LIST_ADD_CHECK_LIVE)))
3561
        goto cleanup;
3562 3563
    newDef = NULL;
    def = virNetworkObjGetDef(obj);
3564

3565 3566
    if (networkStartNetwork(driver, obj) < 0) {
        virNetworkObjRemoveInactive(driver->networks, obj);
3567
        goto cleanup;
3568 3569
    }

3570 3571
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3572 3573
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3574

3575 3576
    VIR_INFO("Creating network '%s'", def->name);
    net = virGetNetwork(conn, def->name, def->uuid);
3577

3578
 cleanup:
3579
    virNetworkDefFree(newDef);
3580
    virObjectEventStateQueue(driver->networkEventState, event);
3581 3582
    virNetworkObjEndAPI(&obj);
    return net;
3583 3584
}

3585 3586 3587 3588

static virNetworkPtr
networkDefineXML(virConnectPtr conn,
                 const char *xml)
3589
{
3590
    virNetworkDriverStatePtr driver = networkGetDriver();
3591
    virNetworkDefPtr def = NULL;
3592
    bool freeDef = true;
3593 3594
    virNetworkObjPtr obj = NULL;
    virNetworkPtr net = NULL;
3595
    virObjectEventPtr event = NULL;
3596

3597
    if (!(def = virNetworkDefParseString(xml)))
3598
        goto cleanup;
3599

3600 3601 3602
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3603
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3604
        goto cleanup;
3605

3606
    if (!(obj = virNetworkObjAssignDef(driver->networks, def, 0)))
J
Ján Tomko 已提交
3607
        goto cleanup;
3608

3609
    /* def was assigned to network object */
3610
    freeDef = false;
3611 3612

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3613 3614
        if (!virNetworkObjIsActive(obj)) {
            virNetworkObjRemoveInactive(driver->networks, obj);
3615 3616
            goto cleanup;
        }
3617 3618 3619 3620
        /* if network was active already, just undo new persistent
         * definition by making it transient.
         * XXX - this isn't necessarily the correct thing to do.
         */
3621
        virNetworkObjUpdateAssignDef(obj, NULL, false);
3622 3623 3624
        goto cleanup;
    }

3625
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3626 3627
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3628

3629
    VIR_INFO("Defining network '%s'", def->name);
3630
    net = virGetNetwork(conn, def->name, def->uuid);
3631

3632
 cleanup:
3633
    virObjectEventStateQueue(driver->networkEventState, event);
3634
    if (freeDef)
J
Ján Tomko 已提交
3635
        virNetworkDefFree(def);
3636 3637
    virNetworkObjEndAPI(&obj);
    return net;
3638 3639
}

3640

3641
static int
3642 3643
networkUndefine(virNetworkPtr net)
{
3644
    virNetworkDriverStatePtr driver = networkGetDriver();
3645
    virNetworkObjPtr obj;
3646
    virNetworkDefPtr def;
3647
    int ret = -1;
3648
    bool active = false;
3649
    virObjectEventPtr event = NULL;
3650

3651
    if (!(obj = networkObjFromNetwork(net)))
3652
        goto cleanup;
3653
    def = virNetworkObjGetDef(obj);
3654

3655
    if (virNetworkUndefineEnsureACL(net->conn, def) < 0)
3656 3657
        goto cleanup;

3658
    if (virNetworkObjIsActive(obj))
3659
        active = true;
3660

3661
    if (!virNetworkObjIsPersistent(obj)) {
3662 3663 3664 3665 3666
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                       _("can't undefine transient network"));
        goto cleanup;
    }

3667
    /* remove autostart link */
3668 3669
    if (virNetworkObjDeleteConfig(driver->networkConfigDir,
                                  driver->networkAutostartDir,
3670
                                  obj) < 0)
3671
        goto cleanup;
3672

3673 3674
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3675 3676
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3677

3678
    VIR_INFO("Undefining network '%s'", def->name);
3679
    if (!active) {
3680
        if (networkRemoveInactive(driver, obj) < 0)
3681
            goto cleanup;
3682 3683 3684 3685 3686
    } else {

        /* if the network still exists, it was active, and we need to make
         * it transient (by deleting the persistent def)
         */
3687
        virNetworkObjUpdateAssignDef(obj, NULL, false);
3688 3689
    }

3690
    ret = 0;
3691

3692
 cleanup:
3693
    virObjectEventStateQueue(driver->networkEventState, event);
3694
    virNetworkObjEndAPI(&obj);
3695
    return ret;
3696 3697
}

3698

3699 3700 3701 3702 3703 3704 3705 3706
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3707
    virNetworkDriverStatePtr driver = networkGetDriver();
3708
    virNetworkObjPtr obj = NULL;
3709
    virNetworkDefPtr def;
3710 3711
    int isActive, ret = -1;
    size_t i;
3712
    virNetworkIPDefPtr ipdef;
3713
    bool oldDhcpActive = false;
3714
    bool needFirewallRefresh = false;
3715

3716 3717 3718 3719 3720

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3721
    if (!(obj = networkObjFromNetwork(net)))
3722
        goto cleanup;
3723
    def = virNetworkObjGetDef(obj);
3724

3725
    if (virNetworkUpdateEnsureACL(net->conn, def, flags) < 0)
3726 3727
        goto cleanup;

3728
    /* see if we are listening for dhcp pre-modification */
3729
    for (i = 0;
3730
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
3731
         i++) {
3732 3733 3734 3735 3736 3737
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3738 3739
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3740
     */
3741
    isActive = virNetworkObjIsActive(obj);
3742 3743
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3744 3745 3746 3747 3748 3749 3750
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3751 3752 3753 3754
    if (isActive && (flags & VIR_NETWORK_UPDATE_AFFECT_LIVE)) {
        /* Take care of anything that must be done before updating the
         * live NetworkDef.
         */
3755 3756 3757 3758
        switch ((virNetworkForwardType) def->forward.type) {
        case VIR_NETWORK_FORWARD_NONE:
        case VIR_NETWORK_FORWARD_NAT:
        case VIR_NETWORK_FORWARD_ROUTE:
3759 3760 3761 3762 3763 3764 3765 3766 3767 3768
            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).
                 */
3769 3770
                networkRemoveFirewallRules(def);
                needFirewallRefresh = true;
3771 3772 3773 3774
                break;
            default:
                break;
            }
3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788
            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;
3789 3790 3791
        }
    }

3792
    /* update the network config in memory/on disk */
3793
    if (virNetworkObjUpdate(obj, command, section, parentIndex, xml, flags) < 0) {
3794
        if (needFirewallRefresh)
3795
            ignore_value(networkAddFirewallRules(def));
3796 3797 3798
        goto cleanup;
    }

3799 3800 3801 3802
    /* @def is replaced */
    def = virNetworkObjGetDef(obj);

    if (needFirewallRefresh && networkAddFirewallRules(def) < 0)
3803 3804 3805 3806 3807
        goto cleanup;

    if (flags & VIR_NETWORK_UPDATE_AFFECT_CONFIG) {
        /* save updated persistent config to disk */
        if (virNetworkSaveConfig(driver->networkConfigDir,
3808
                                 virNetworkObjGetPersistentDef(obj)) < 0) {
3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821
            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 ||
3822 3823 3824 3825 3826 3827 3828 3829
            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)
3830
             */
3831
            if (networkRestartDhcpDaemon(driver, obj) < 0)
3832 3833
                goto cleanup;

3834 3835 3836 3837 3838 3839 3840 3841
        } 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;

3842
            for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
3843
                 i++) {
3844 3845 3846 3847 3848 3849 3850
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3851 3852
                 networkRestartDhcpDaemon(driver, obj) < 0) ||
                networkRefreshDhcpDaemon(driver, obj) < 0) {
3853 3854 3855
                goto cleanup;
            }

3856 3857 3858 3859
        } 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.
3860
             */
3861
            if (networkRefreshDhcpDaemon(driver, obj) < 0)
3862 3863 3864 3865 3866 3867 3868 3869
                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.
             */
3870
            if (networkRefreshRadvd(driver, obj) < 0)
3871 3872 3873 3874
                goto cleanup;
        }

        /* save current network state to disk */
3875
        if ((ret = virNetworkObjSaveStatus(driver->stateDir, obj)) < 0)
3876 3877
            goto cleanup;
    }
3878 3879

    /* call the 'updated' network hook script */
3880
    if (networkRunHook(obj, NULL, NULL, VIR_HOOK_NETWORK_OP_UPDATED,
3881 3882 3883
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

3884
    ret = 0;
3885
 cleanup:
3886
    virNetworkObjEndAPI(&obj);
3887 3888 3889
    return ret;
}

3890 3891 3892

static int
networkCreate(virNetworkPtr net)
3893
{
3894
    virNetworkDriverStatePtr driver = networkGetDriver();
3895
    virNetworkObjPtr obj;
3896
    virNetworkDefPtr def;
3897
    int ret = -1;
3898
    virObjectEventPtr event = NULL;
3899

3900
    if (!(obj = networkObjFromNetwork(net)))
3901
        goto cleanup;
3902
    def = virNetworkObjGetDef(obj);
3903

3904
    if (virNetworkCreateEnsureACL(net->conn, def) < 0)
3905 3906
        goto cleanup;

3907
    if ((ret = networkStartNetwork(driver, obj)) < 0)
3908
        goto cleanup;
3909

3910 3911
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3912 3913
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3914

3915
 cleanup:
3916
    virObjectEventStateQueue(driver->networkEventState, event);
3917
    virNetworkObjEndAPI(&obj);
3918
    return ret;
3919 3920
}

3921 3922 3923

static int
networkDestroy(virNetworkPtr net)
3924
{
3925
    virNetworkDriverStatePtr driver = networkGetDriver();
3926
    virNetworkObjPtr obj;
3927
    virNetworkDefPtr def;
3928
    int ret = -1;
3929
    virObjectEventPtr event = NULL;
3930

3931
    if (!(obj = networkObjFromNetwork(net)))
3932
        goto cleanup;
3933
    def = virNetworkObjGetDef(obj);
3934

3935
    if (virNetworkDestroyEnsureACL(net->conn, def) < 0)
3936 3937
        goto cleanup;

3938
    if (!virNetworkObjIsActive(obj)) {
3939
        virReportError(VIR_ERR_OPERATION_INVALID,
3940
                       _("network '%s' is not active"),
3941
                       def->name);
3942 3943 3944
        goto cleanup;
    }

3945
    if ((ret = networkShutdownNetwork(driver, obj)) < 0)
3946
        goto cleanup;
3947 3948
    /* @def replaced in virNetworkObjUnsetDefTransient*/
    def = virNetworkObjGetDef(obj);
3949

3950 3951
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3952 3953
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3954

3955 3956
    if (!virNetworkObjIsPersistent(obj) &&
        networkRemoveInactive(driver, obj) < 0) {
3957 3958
        ret = -1;
        goto cleanup;
3959
    }
3960

3961
 cleanup:
3962
    virObjectEventStateQueue(driver->networkEventState, event);
3963
    virNetworkObjEndAPI(&obj);
3964 3965 3966
    return ret;
}

3967 3968 3969 3970

static char *
networkGetXMLDesc(virNetworkPtr net,
                  unsigned int flags)
3971
{
3972
    virNetworkObjPtr obj;
3973
    virNetworkDefPtr curDef;
3974
    virNetworkDefPtr def;
3975
    virNetworkDefPtr newDef;
3976
    char *ret = NULL;
3977

3978
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3979

3980
    if (!(obj = networkObjFromNetwork(net)))
3981
        return ret;
3982 3983
    def = virNetworkObjGetDef(obj);
    newDef = virNetworkObjGetNewDef(obj);
3984

3985
    if (virNetworkGetXMLDescEnsureACL(net->conn, def) < 0)
3986 3987
        goto cleanup;

3988 3989
    if ((flags & VIR_NETWORK_XML_INACTIVE) && newDef)
        curDef = newDef;
3990
    else
3991
        curDef = def;
3992

3993
    ret = virNetworkDefFormat(curDef, flags);
3994

3995
 cleanup:
3996
    virNetworkObjEndAPI(&obj);
3997
    return ret;
3998 3999
}

4000 4001 4002 4003

static char *
networkGetBridgeName(virNetworkPtr net)
{
4004
    virNetworkObjPtr obj;
4005
    virNetworkDefPtr def;
4006 4007
    char *bridge = NULL;

4008
    if (!(obj = networkObjFromNetwork(net)))
4009
        return bridge;
4010
    def = virNetworkObjGetDef(obj);
4011

4012
    if (virNetworkGetBridgeNameEnsureACL(net->conn, def) < 0)
4013 4014
        goto cleanup;

4015
    if (!(def->bridge)) {
4016 4017
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
4018
                       def->name);
4019 4020 4021
        goto cleanup;
    }

4022
    ignore_value(VIR_STRDUP(bridge, def->bridge));
4023

4024
 cleanup:
4025
    virNetworkObjEndAPI(&obj);
4026 4027 4028
    return bridge;
}

4029 4030 4031 4032

static int
networkGetAutostart(virNetworkPtr net,
                    int *autostart)
4033
{
4034
    virNetworkObjPtr obj;
4035
    int ret = -1;
4036

4037
    if (!(obj = networkObjFromNetwork(net)))
4038
        return ret;
4039

4040
    if (virNetworkGetAutostartEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
4041 4042
        goto cleanup;

4043
    *autostart = virNetworkObjIsAutostart(obj) ? 1 : 0;
4044
    ret = 0;
4045

4046
 cleanup:
4047
    virNetworkObjEndAPI(&obj);
4048
    return ret;
4049 4050
}

4051 4052 4053 4054

static int
networkSetAutostart(virNetworkPtr net,
                    int autostart)
4055
{
4056
    virNetworkDriverStatePtr driver = networkGetDriver();
4057
    virNetworkObjPtr obj;
4058
    virNetworkDefPtr def;
4059
    char *configFile = NULL, *autostartLink = NULL;
4060 4061
    bool new_autostart;
    bool cur_autostart;
4062
    int ret = -1;
4063

4064
    if (!(obj = networkObjFromNetwork(net)))
4065
        goto cleanup;
4066
    def = virNetworkObjGetDef(obj);
4067

4068
    if (virNetworkSetAutostartEnsureACL(net->conn, def) < 0)
4069 4070
        goto cleanup;

4071
    if (!virNetworkObjIsPersistent(obj)) {
4072 4073
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
4074 4075 4076
        goto cleanup;
    }

4077 4078 4079
    new_autostart = (autostart != 0);
    cur_autostart = virNetworkObjIsAutostart(obj);
    if (cur_autostart != new_autostart) {
4080 4081
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir,
                                               def->name)) == NULL)
4082
            goto cleanup;
4083 4084
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir,
                                                  def->name)) == NULL)
4085 4086
            goto cleanup;

4087
        if (new_autostart) {
4088
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
4089
                virReportSystemError(errno,
4090 4091
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
4092 4093
                goto cleanup;
            }
4094

4095
            if (symlink(configFile, autostartLink) < 0) {
4096
                virReportSystemError(errno,
4097
                                     _("Failed to create symlink '%s' to '%s'"),
4098
                                     autostartLink, configFile);
4099 4100 4101
                goto cleanup;
            }
        } else {
4102
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
4103
                virReportSystemError(errno,
4104
                                     _("Failed to delete symlink '%s'"),
4105
                                     autostartLink);
4106 4107
                goto cleanup;
            }
4108 4109
        }

4110
        virNetworkObjSetAutostart(obj, new_autostart);
4111
    }
4112

4113
    ret = 0;
4114

4115
 cleanup:
4116 4117
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
4118
    virNetworkObjEndAPI(&obj);
4119
    return ret;
4120 4121
}

4122

4123
static int
4124
networkGetDHCPLeases(virNetworkPtr net,
4125 4126 4127
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
4128
{
4129
    virNetworkDriverStatePtr driver = networkGetDriver();
4130 4131 4132
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
4133
    size_t size = 0;
4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144
    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;
4145
    virNetworkIPDefPtr ipdef_tmp = NULL;
4146 4147
    virNetworkDHCPLeasePtr lease = NULL;
    virNetworkDHCPLeasePtr *leases_ret = NULL;
4148
    virNetworkObjPtr obj;
4149
    virNetworkDefPtr def;
4150
    virMacAddr mac_addr;
4151 4152 4153

    virCheckFlags(0, -1);

4154 4155 4156 4157 4158 4159
    /* only to check if the MAC is valid */
    if (mac && virMacAddrParse(mac, &mac_addr) < 0) {
        virReportError(VIR_ERR_INVALID_MAC, "%s", mac);
        return -1;
    }

4160
    if (!(obj = networkObjFromNetwork(net)))
4161
        return -1;
4162
    def = virNetworkObjGetDef(obj);
4163

4164
    if (virNetworkGetDHCPLeasesEnsureACL(net->conn, def) < 0)
4165
        goto cleanup;
4166 4167

    /* Retrieve custom leases file location */
4168
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(driver, def->bridge);
4169 4170

    /* Read entire contents */
4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184
    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);
        }
4185 4186 4187 4188 4189 4190 4191 4192 4193 4194
        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;
        }

4195
        if (!virJSONValueIsArray(leases_array)) {
4196
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
4197
                           _("Malformed lease_entries array"));
4198 4199
            goto error;
        }
4200
        size = virJSONValueArraySize(leases_array);
4201 4202
    }

4203
    currtime = (long long)time(NULL);
4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219

    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;
        }

4220
        if (mac && virMacAddrCompare(mac, mac_tmp))
4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251
            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 */
4252 4253
            for (j = 0; j < def->nips; j++) {
                ipdef_tmp = &def->ips[j];
4254 4255 4256 4257 4258 4259 4260 4261

                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)) {
4262
                    lease->prefix = virSocketAddrGetIPPrefix(&ipdef_tmp->address,
4263 4264 4265 4266 4267 4268 4269 4270
                                                             &ipdef_tmp->netmask,
                                                             ipdef_tmp->prefix);
                    break;
                }
            }

            if ((VIR_STRDUP(lease->mac, mac_tmp) < 0) ||
                (VIR_STRDUP(lease->ipaddr, ip_tmp) < 0) ||
4271
                (VIR_STRDUP(lease->iface, def->bridge) < 0))
4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303
                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);
4304
    VIR_FREE(lease_entries);
4305 4306
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
4307

4308
    virNetworkObjEndAPI(&obj);
4309

4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320
    return rv;

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

4321

4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356
/* 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);
        }
    }
}

4357

4358 4359 4360 4361 4362 4363 4364 4365 4366
/* 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:
4367
 * @dom: domain definition that @iface belongs to
4368 4369 4370 4371 4372 4373 4374 4375 4376 4377
 * @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.
 */
4378
static int
4379 4380
networkAllocateActualDevice(virNetworkPtr net,
                            virDomainDefPtr dom,
4381
                            virDomainNetDefPtr iface)
4382
{
4383
    virNetworkDriverStatePtr driver = networkGetDriver();
4384
    virDomainNetType actualType = iface->type;
4385
    virNetworkObjPtr obj = NULL;
4386
    virNetworkDefPtr netdef = NULL;
4387
    virNetDevBandwidthPtr bandwidth = NULL;
4388 4389 4390
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
4391
    virNetworkForwardIfDefPtr dev = NULL;
4392
    size_t i;
4393 4394
    int ret = -1;

4395 4396 4397 4398 4399 4400 4401 4402
    obj = virNetworkObjFindByName(driver->networks, net->name);
    if (!obj) {
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       net->name);
        goto error;
    }

4403 4404 4405 4406 4407
    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Expected an interface for a virtual network"));
        goto error;
    }
4408 4409 4410 4411

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

4412
    netdef = virNetworkObjGetDef(obj);
4413

4414
    if (!virNetworkObjIsActive(obj)) {
4415 4416 4417 4418 4419 4420
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

4421 4422 4423
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

4424 4425 4426
    /* 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
4427
     */
4428 4429 4430 4431 4432 4433
    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.
     */
4434 4435 4436 4437 4438 4439

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

4440 4441
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
4442
        goto error;
4443

4444 4445 4446 4447 4448 4449 4450 4451
    /* 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;

4452 4453
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
4454

4455 4456 4457 4458 4459 4460 4461 4462 4463 4464
    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;

4465 4466 4467 4468 4469
    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:
4470
        /* for these forward types, the actual net type really *is*
4471
         * NETWORK; we just keep the info from the portgroup in
4472
         * iface->data.network.actual
J
Ján Tomko 已提交
4473
         */
4474
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
4475

4476
        /* we also store the bridge device and macTableManager settings
4477 4478 4479 4480 4481 4482 4483
         * 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;
4484 4485
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4486

4487
        if (networkPlugBandwidth(obj, iface) < 0)
4488
            goto error;
4489
        break;
4490

4491
    case VIR_NETWORK_FORWARD_HOSTDEV: {
4492
        virDomainHostdevSubsysPCIBackendType backend;
4493

4494
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
4495
        if (networkCreateInterfacePool(netdef) < 0)
4496 4497 4498
            goto error;

        /* pick first dev with 0 connections */
4499 4500 4501
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
4502 4503 4504 4505 4506 4507 4508 4509 4510 4511
                break;
            }
        }
        if (!dev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' requires exclusive access "
                             "to interfaces, but none are available"),
                           netdef->name);
            goto error;
        }
4512
        iface->data.network.actual->data.hostdev.def.parentnet = iface;
4513 4514
        iface->data.network.actual->data.hostdev.def.info = &iface->info;
        iface->data.network.actual->data.hostdev.def.mode = VIR_DOMAIN_HOSTDEV_MODE_SUBSYS;
4515
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
4516
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
4517
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
4518

E
Eric Blake 已提交
4519
        switch (netdef->forward.driverName) {
4520
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
4521
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
4522 4523
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
4524
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
4525 4526
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
4527
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538
            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;

4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562
        /* 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;
            }
        }
4563 4564
        break;
    }
4565

4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577
    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;
4578

4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611
            /* 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:
4612 4613 4614 4615 4616
        /* <forward type='bridge|private|vepa|passthrough'> are all
         * VIR_DOMAIN_NET_TYPE_DIRECT.
         */

        /* Set type=direct and appropriate <source mode='xxx'/> */
4617
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4618 4619 4620 4621 4622 4623

        /* 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));
4624

4625 4626 4627 4628 4629 4630 4631 4632
        /* 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) {
4633
            goto error;
4634
        }
4635
        virtport = iface->data.network.actual->virtPortProfile;
4636
        if (virtport) {
4637 4638 4639 4640 4641 4642 4643 4644
            /* 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);
4645
                goto error;
4646 4647
            }
        }
4648

4649 4650 4651
        /* 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).
         */
4652
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4653 4654 4655 4656
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4657
            goto error;
4658 4659 4660
        } else {
            /* pick an interface from the pool */

4661
            if (networkCreateInterfacePool(netdef) < 0)
4662 4663
                goto error;

4664 4665 4666 4667 4668
            /* 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.
4669
             */
4670 4671
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4672 4673 4674
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4675

4676
                /* pick first dev with 0 connections */
4677 4678 4679
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4680 4681 4682 4683 4684
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4685
                dev = &netdef->forward.ifs[0];
4686 4687 4688
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4689 4690 4691 4692
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4693 4694 4695 4696
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4697
                goto error;
4698
            }
4699 4700
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4701
                goto error;
4702
        }
4703 4704 4705 4706 4707 4708
        break;

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, netdef->forward.type);
        goto error;
4709 4710
    }

4711 4712
    if (virNetworkObjMacMgrAdd(obj, driver->dnsmasqStateDir,
                               dom->name, &iface->mac) < 0)
M
Michal Privoznik 已提交
4713 4714
        goto error;

4715
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4716
        goto error;
4717

4718 4719 4720 4721 4722
    /* 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.
     */

4723
    if (virDomainNetGetActualVlan(iface)) {
4724 4725 4726 4727
        /* 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
4728 4729
         */
        if (!(actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV ||
4730 4731 4732
              (actualType == VIR_DOMAIN_NET_TYPE_DIRECT &&
               virDomainNetGetActualDirectMode(iface)
               == VIR_NETDEV_MACVLAN_MODE_PASSTHRU) ||
4733 4734 4735
              (actualType == VIR_DOMAIN_NET_TYPE_BRIDGE &&
               virtport && virtport->virtPortType
               == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH))) {
4736 4737 4738 4739 4740
            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);
4741 4742 4743
            goto error;
        }
    }
4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754
    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;
        }
    }
4755

4756 4757 4758 4759 4760 4761 4762 4763 4764
    netdef->connections++;
    if (dev)
        dev->connections++;
    /* finally we can call the 'plugged' hook script if any */
    if (networkRunHook(obj, dom, iface,
                       VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
                       VIR_HOOK_SUBOP_BEGIN) < 0) {
        /* adjust for failure */
        netdef->connections--;
4765
        if (dev)
4766 4767
            dev->connections--;
        goto error;
4768
    }
4769
    networkLogAllocation(netdef, actualType, dev, iface, true);
4770

4771
    ret = 0;
4772

4773
 cleanup:
4774
    virNetworkObjEndAPI(&obj);
4775 4776
    return ret;

4777
 error:
4778
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4779 4780 4781
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4782
    goto cleanup;
4783 4784
}

4785

4786
/* networkNotifyActualDevice:
4787
 * @dom: domain definition that @iface belongs to
4788 4789 4790 4791 4792
 * @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
4793 4794
 * order, or re-attach the interface's tap device to the network's
 * bridge.
4795
 *
4796
 * No return value (but does log any failures)
4797
 */
4798
static void
4799 4800
networkNotifyActualDevice(virNetworkPtr net,
                          virDomainDefPtr dom,
4801
                          virDomainNetDefPtr iface)
4802
{
4803
    virNetworkDriverStatePtr driver = networkGetDriver();
4804
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4805
    virNetworkObjPtr obj;
4806
    virNetworkDefPtr netdef;
4807
    virNetworkForwardIfDefPtr dev = NULL;
4808
    size_t i;
4809
    char *master = NULL;
4810

4811 4812 4813 4814 4815 4816 4817 4818
    obj = virNetworkObjFindByName(driver->networks, net->name);
    if (!obj) {
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       net->name);
        goto error;
    }

4819 4820 4821 4822 4823
    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Expected an interface for a virtual network"));
        goto error;
    }
4824

4825
    netdef = virNetworkObjGetDef(obj);
4826

4827
    if (!virNetworkObjIsActive(obj)) {
4828 4829 4830 4831 4832 4833
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

4834 4835
    /* see if we're connected to the correct bridge */
    if (netdef->bridge) {
4836 4837
        bool useOVS = false;

4838 4839 4840
        if (virNetDevGetMaster(iface->ifname, &master) < 0)
            goto error;

4841 4842 4843 4844 4845 4846 4847 4848
        /* 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;
        }

4849 4850
        if (STRNEQ_NULLABLE(netdef->bridge, master)) {
            /* disconnect from current (incorrect) bridge */
4851 4852 4853 4854 4855 4856 4857
            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));
            }
4858 4859 4860 4861 4862 4863

            /* 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())
             */
4864
            VIR_INFO("Attaching %s to %s", iface->ifname, netdef->bridge);
4865 4866 4867 4868 4869 4870 4871 4872 4873 4874
            if (virNetDevTapAttachBridge(iface->ifname, netdef->bridge,
                                         &iface->mac, dom->uuid,
                                         virDomainNetGetActualVirtPortProfile(iface),
                                         virDomainNetGetActualVlan(iface),
                                         iface->mtu, NULL) < 0) {
                goto error;
            }
        }
    }

4875
    if (!iface->data.network.actual ||
4876 4877
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4878 4879
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4880 4881
    }

4882
    if (networkCreateInterfacePool(netdef) < 0)
4883
        goto error;
4884

4885
    if (netdef->forward.nifs == 0) {
4886
        virReportError(VIR_ERR_INTERNAL_ERROR,
4887 4888
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4889
                       netdef->name);
4890
        goto error;
4891
    }
4892

4893 4894
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4895

4896 4897 4898 4899 4900 4901 4902 4903 4904
        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 */
4905 4906
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4907
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4908 4909
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4910 4911 4912 4913 4914
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4915
            virReportError(VIR_ERR_INTERNAL_ERROR,
4916 4917
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4918
                           netdef->name, actualDev);
4919
            goto error;
4920 4921
        }

4922
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4923 4924
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4925
         */
4926
        if ((dev->connections > 0) &&
4927 4928
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4929 4930
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4931
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4932
            virReportError(VIR_ERR_INTERNAL_ERROR,
4933 4934
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4935
                           netdef->name, actualDev);
4936
            goto error;
4937
        }
4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949
    }  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 */
4950 4951
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4952
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4953
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4954 4955
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4956 4957 4958 4959 4960 4961 4962 4963 4964
                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,
4965 4966 4967 4968
                           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
4969
            goto error;
4970 4971 4972 4973 4974 4975 4976
        }

        /* 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) &&
4977
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4978 4979 4980 4981 4982 4983 4984 4985 4986
            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;
        }
4987 4988
    }

4989
 success:
4990
    netdef->connections++;
4991 4992
    if (dev)
        dev->connections++;
4993
    /* finally we can call the 'plugged' hook script if any */
4994
    if (networkRunHook(obj, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
4995 4996 4997 4998 4999 5000 5001
                       VIR_HOOK_SUBOP_BEGIN) < 0) {
        /* adjust for failure */
        if (dev)
            dev->connections--;
        netdef->connections--;
        goto error;
    }
5002
    networkLogAllocation(netdef, actualType, dev, iface, true);
5003

5004
 cleanup:
5005
    virNetworkObjEndAPI(&obj);
5006
    VIR_FREE(master);
5007
    return;
5008

5009
 error:
5010
    goto cleanup;
5011 5012 5013 5014
}


/* networkReleaseActualDevice:
5015
 * @dom: domain definition that @iface belongs to
5016 5017 5018 5019 5020 5021 5022 5023 5024
 * @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.
 */
5025
static int
5026 5027
networkReleaseActualDevice(virNetworkPtr net,
                           virDomainDefPtr dom,
5028
                           virDomainNetDefPtr iface)
5029
{
5030
    virNetworkDriverStatePtr driver = networkGetDriver();
5031
    virDomainNetType actualType = virDomainNetGetActualType(iface);
5032
    virNetworkObjPtr obj;
5033
    virNetworkDefPtr netdef;
5034
    virNetworkForwardIfDefPtr dev = NULL;
5035 5036
    size_t i;
    int ret = -1;
5037

5038 5039 5040 5041 5042 5043 5044 5045
    obj = virNetworkObjFindByName(driver->networks, net->name);
    if (!obj) {
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       net->name);
        goto error;
    }

5046 5047 5048 5049 5050
    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Expected an interface for a virtual network"));
        goto error;
    }
5051

5052
    netdef = virNetworkObjGetDef(obj);
5053

5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072
    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);
5073
        goto error;
5074
    }
5075

5076 5077 5078
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
5079 5080
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
5081 5082
    }

5083
    if (netdef->forward.nifs == 0) {
5084
        virReportError(VIR_ERR_INTERNAL_ERROR,
5085
                       _("network '%s' uses a direct/hostdev mode, but "
5086 5087
                         "has no forward dev and no interface pool"),
                       netdef->name);
5088
        goto error;
5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100
    }

    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;
        }
5101

5102 5103
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
5104
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
5105 5106
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
5107 5108 5109
                break;
            }
        }
5110

5111
        if (!dev) {
5112
            virReportError(VIR_ERR_INTERNAL_ERROR,
5113 5114
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
5115
                           netdef->name, actualDev);
5116
            goto error;
5117
        }
5118 5119 5120 5121 5122 5123 5124 5125 5126 5127
    } 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;
        }

5128 5129
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
5130
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
5131
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
5132 5133
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
5134 5135 5136 5137 5138 5139 5140 5141 5142
                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,
5143 5144 5145 5146
                           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
5147
            goto error;
5148
        }
J
Ján Tomko 已提交
5149
    }
5150

5151
 success:
5152
    virNetworkObjMacMgrDel(obj, driver->dnsmasqStateDir, dom->name, &iface->mac);
M
Michal Privoznik 已提交
5153

5154
    if (iface->data.network.actual) {
5155
        netdef->connections--;
5156 5157
        if (dev)
            dev->connections--;
5158
        /* finally we can call the 'unplugged' hook script if any */
5159
        networkRunHook(obj, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
5160
                       VIR_HOOK_SUBOP_BEGIN);
5161
        networkLogAllocation(netdef, actualType, dev, iface, false);
5162
    }
5163
    ret = 0;
5164
 cleanup:
5165
    virNetworkObjEndAPI(&obj);
5166 5167 5168 5169
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
5170
    return ret;
5171

5172
 error:
5173
    goto cleanup;
5174
}
5175

5176

5177 5178 5179
/**
 * networkCheckBandwidth:
 * @net: network QoS
5180
 * @ifaceBand: interface QoS (may be NULL if no QoS)
5181
 * @oldBandwidth: new interface QoS (may be NULL if no QoS)
5182
 * @ifaceMac: interface MAC (used in error messages for identification)
5183 5184
 * @new_rate: new rate for non guaranteed class
 *
5185 5186 5187 5188 5189 5190 5191 5192
 * 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.
 *
5193 5194 5195 5196 5197
 * 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
5198
networkCheckBandwidth(virNetworkObjPtr obj,
5199
                      virNetDevBandwidthPtr ifaceBand,
5200
                      virNetDevBandwidthPtr oldBandwidth,
5201
                      virMacAddr ifaceMac,
5202 5203 5204
                      unsigned long long *new_rate)
{
    int ret = -1;
5205 5206
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
    virNetDevBandwidthPtr netBand = def->bandwidth;
5207
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5208 5209 5210
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

5211
    virMacAddrFormat(&ifaceMac, ifmac);
5212 5213 5214 5215 5216 5217

    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"),
5218
                       ifmac, def->name);
5219 5220 5221
        return -1;
    }

5222 5223
    if (((!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor) &&
         (!oldBandwidth || !oldBandwidth->in || !oldBandwidth->in->floor)) ||
5224 5225
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
5226
        return 1;
5227
    }
5228 5229

    tmp_new_rate = netBand->in->average;
5230 5231 5232 5233
    if (oldBandwidth && oldBandwidth->in)
        tmp_floor_sum -= oldBandwidth->in->floor;
    if (ifaceBand && ifaceBand->in)
        tmp_floor_sum += ifaceBand->in->floor;
5234 5235 5236 5237 5238 5239 5240 5241 5242

    /* 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,
5243 5244
                           def->bridge,
                           def->name);
5245 5246 5247 5248 5249 5250 5251 5252 5253
            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,
5254 5255
                       def->bridge,
                       def->name);
5256 5257 5258
        goto cleanup;
    }

5259 5260
    if (new_rate)
        *new_rate = tmp_new_rate;
5261 5262
    ret = 0;

5263
 cleanup:
5264 5265 5266
    return ret;
}

5267

5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278
/**
 * 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
5279
networkNextClassID(virNetworkObjPtr obj)
5280
{
5281
    ssize_t ret = 0;
5282
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5283

5284 5285
    if ((ret = virBitmapNextClearBit(classIdMap, -1)) < 0)
        ret = virBitmapSize(classIdMap);
5286

5287
    if (virBitmapSetBitExpand(classIdMap, ret) < 0)
5288 5289 5290 5291 5292
        return -1;

    return ret;
}

5293

5294
static int
5295
networkPlugBandwidthImpl(virNetworkObjPtr obj,
5296 5297 5298
                         virDomainNetDefPtr iface,
                         virNetDevBandwidthPtr ifaceBand,
                         unsigned long long new_rate)
5299
{
5300
    virNetworkDriverStatePtr driver = networkGetDriver();
5301
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
5302
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5303
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5304
    ssize_t class_id = 0;
5305 5306
    int plug_ret;
    int ret = -1;
5307 5308

    /* generate new class_id */
5309
    if ((class_id = networkNextClassID(obj)) < 0) {
5310 5311 5312 5313 5314
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Could not generate next class ID"));
        goto cleanup;
    }

5315
    plug_ret = virNetDevBandwidthPlug(def->bridge, def->bandwidth,
5316
                                      &iface->mac, ifaceBand, class_id);
5317
    if (plug_ret < 0) {
5318
        ignore_value(virNetDevBandwidthUnplug(def->bridge, class_id));
5319 5320 5321 5322 5323 5324
        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 */
5325 5326
    tmp_floor_sum += ifaceBand->in->floor;
    virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5327
    /* update status file */
5328
    if (virNetworkObjSaveStatus(driver->stateDir, obj) < 0) {
5329
        ignore_value(virBitmapClearBit(classIdMap, class_id));
5330 5331
        tmp_floor_sum -= ifaceBand->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5332
        iface->data.network.actual->class_id = 0;
5333
        ignore_value(virNetDevBandwidthUnplug(def->bridge, class_id));
5334 5335
        goto cleanup;
    }
5336
    /* update rate for non guaranteed NICs */
5337
    new_rate -= tmp_floor_sum;
5338 5339
    if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                     def->bandwidth, new_rate) < 0)
5340
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
5341
                 def->bridge);
5342 5343

    ret = 0;
5344 5345 5346 5347 5348 5349
 cleanup:
    return ret;
}


static int
5350
networkPlugBandwidth(virNetworkObjPtr obj,
5351 5352 5353 5354 5355 5356 5357 5358
                     virDomainNetDefPtr iface)
{
    int ret = -1;
    int plug_ret;
    unsigned long long new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);

5359
    if ((plug_ret = networkCheckBandwidth(obj, ifaceBand, NULL,
5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379
                                          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;
    }

5380
    if (networkPlugBandwidthImpl(obj, iface, ifaceBand, new_rate) < 0)
5381 5382 5383
        goto cleanup;

    ret = 0;
5384

5385
 cleanup:
5386 5387 5388
    return ret;
}

5389

5390
static int
5391
networkUnplugBandwidth(virNetworkObjPtr obj,
5392 5393
                       virDomainNetDefPtr iface)
{
5394
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
5395
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5396
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5397
    virNetworkDriverStatePtr driver = networkGetDriver();
5398 5399
    int ret = 0;
    unsigned long long new_rate;
5400
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
5401 5402 5403

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
5404
        if (!def->bandwidth || !def->bandwidth->in) {
5405
            VIR_WARN("Network %s has no bandwidth but unplug requested",
5406
                     def->name);
5407 5408
            goto cleanup;
        }
5409
        /* we must remove class from bridge */
5410
        new_rate = def->bandwidth->in->average;
5411

5412 5413
        if (def->bandwidth->in->peak > 0)
            new_rate = def->bandwidth->in->peak;
5414

5415
        ret = virNetDevBandwidthUnplug(def->bridge,
5416 5417 5418 5419
                                       iface->data.network.actual->class_id);
        if (ret < 0)
            goto cleanup;
        /* update sum of 'floor'-s of attached NICs */
5420 5421 5422
        tmp_floor_sum -= ifaceBand->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);

5423
        /* return class ID */
5424
        ignore_value(virBitmapClearBit(classIdMap,
5425 5426
                                       iface->data.network.actual->class_id));
        /* update status file */
5427
        if (virNetworkObjSaveStatus(driver->stateDir, obj) < 0) {
5428 5429
            tmp_floor_sum += ifaceBand->in->floor;
            virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5430
            ignore_value(virBitmapSetBit(classIdMap,
5431 5432 5433
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
5434
        /* update rate for non guaranteed NICs */
5435
        new_rate -= tmp_floor_sum;
5436 5437
        if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                         def->bandwidth, new_rate) < 0)
5438
            VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
5439
                     def->bridge);
5440 5441 5442 5443
        /* no class is associated any longer */
        iface->data.network.actual->class_id = 0;
    }

5444
 cleanup:
5445 5446
    return ret;
}
5447

5448

5449
static void
5450
networkNetworkObjTaint(virNetworkObjPtr obj,
5451
                       virNetworkTaintFlags taint)
5452
{
5453 5454
    virNetworkDefPtr def = virNetworkObjGetDef(obj);

5455
    if (virNetworkObjTaint(obj, taint)) {
5456
        char uuidstr[VIR_UUID_STRING_BUFLEN];
5457
        virUUIDFormat(def->uuid, uuidstr);
5458 5459

        VIR_WARN("Network name='%s' uuid=%s is tainted: %s",
5460
                 def->name, uuidstr, virNetworkTaintTypeToString(taint));
5461 5462
    }
}
5463 5464 5465 5466 5467 5468


static bool
networkBandwidthGenericChecks(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
5469
    virNetDevBandwidthPtr ifaceBand;
5470 5471 5472 5473 5474 5475 5476 5477
    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;
    }

5478
    ifaceBand = virDomainNetGetActualBandwidth(iface);
5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489
    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;
}


5490
static bool
5491 5492 5493 5494
networkBandwidthChangeAllowed(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
5495
    virNetworkObjPtr obj = NULL;
5496 5497 5498 5499 5500 5501
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    bool ret = false;

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

5502 5503
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
5504 5505 5506 5507 5508 5509
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return false;
    }

5510
    if (networkCheckBandwidth(obj, newBandwidth, ifaceBand, iface->mac, NULL) < 0)
5511 5512 5513 5514 5515
        goto cleanup;

    ret = true;

 cleanup:
5516
    virNetworkObjEndAPI(&obj);
5517 5518
    return ret;
}
5519 5520


5521
static int
5522 5523 5524 5525
networkBandwidthUpdate(virDomainNetDefPtr iface,
                       virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
5526
    virNetworkObjPtr obj = NULL;
5527
    virNetworkDefPtr def;
5528
    unsigned long long tmp_floor_sum;
5529 5530 5531 5532 5533
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    unsigned long long new_rate = 0;
    int plug_ret;
    int ret = -1;

5534 5535 5536 5537 5538 5539
    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Expected an interface for a virtual network"));
        return -1;
    }

5540 5541 5542
    if (!networkBandwidthGenericChecks(iface, newBandwidth))
        return 0;

5543 5544
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
5545 5546 5547 5548 5549
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return ret;
    }
5550
    def = virNetworkObjGetDef(obj);
5551

5552
    if ((plug_ret = networkCheckBandwidth(obj, newBandwidth, ifaceBand,
5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565
                                          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: */

5566
    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
5567
        newBandwidth->in && newBandwidth->in->floor) {
5568 5569
        /* Either we just need to update @floor .. */

5570
        if (virNetDevBandwidthUpdateRate(def->bridge,
5571
                                         iface->data.network.actual->class_id,
5572
                                         def->bandwidth,
5573 5574 5575
                                         newBandwidth->in->floor) < 0)
            goto cleanup;

5576 5577 5578 5579 5580
        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;
5581

5582 5583
        if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                         def->bandwidth, new_rate) < 0 ||
5584
            virNetworkObjSaveStatus(driver->stateDir, obj) < 0) {
5585
            /* Ouch, rollback */
5586 5587 5588
            tmp_floor_sum -= newBandwidth->in->floor;
            tmp_floor_sum += ifaceBand->in->floor;
            virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5589

5590
            ignore_value(virNetDevBandwidthUpdateRate(def->bridge,
5591
                                                      iface->data.network.actual->class_id,
5592
                                                      def->bandwidth,
5593 5594 5595 5596 5597 5598
                                                      ifaceBand->in->floor));
            goto cleanup;
        }
    } else if (newBandwidth->in && newBandwidth->in->floor) {
        /* .. or we need to plug in new .. */

5599
        if (networkPlugBandwidthImpl(obj, iface, newBandwidth, new_rate) < 0)
5600 5601 5602 5603
            goto cleanup;
    } else {
        /* .. or unplug old. */

5604
        if (networkUnplugBandwidth(obj, iface) < 0)
5605 5606 5607 5608 5609
            goto cleanup;
    }

    ret = 0;
 cleanup:
5610
    virNetworkObjEndAPI(&obj);
5611 5612
    return ret;
}
5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640


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 */
};

5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652

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 = {
5653
    .localOnly = true,
5654
    .uriSchemes = (const char *[]){ "network", NULL },
5655 5656 5657 5658 5659
    .hypervisorDriver = &networkHypervisorDriver,
    .networkDriver = &networkDriver,
};


5660 5661 5662 5663 5664 5665 5666 5667 5668 5669
static virStateDriver networkStateDriver = {
    .name = "bridge",
    .stateInitialize  = networkStateInitialize,
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
};

int
networkRegister(void)
{
5670 5671
    if (virRegisterConnectDriver(&networkConnectDriver, false) < 0)
        return -1;
5672 5673 5674 5675
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
        return -1;
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
5676 5677 5678 5679

    virDomainNetSetDeviceImpl(
        networkAllocateActualDevice,
        networkNotifyActualDevice,
5680 5681
        networkReleaseActualDevice,
        networkBandwidthChangeAllowed,
5682
        networkBandwidthUpdate);
5683

5684 5685
    return 0;
}