bridge_driver.c 190.6 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
#ifdef HAVE_SYSCTLBYNAME
37 38
# 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
#include "virnetworkportdef.h"
71

72 73
#include <libxml/xpathInternals.h>

74
#define VIR_FROM_THIS VIR_FROM_NETWORK
75
#define MAX_BRIDGE_ID 256
76

77 78 79 80 81 82 83
/**
 * 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)

84 85
#define SYSCTL_PATH "/proc/sys"

86 87
VIR_LOG_INIT("network.bridge_driver");

88 89
#define DNSMASQ_NAMESPACE_HREF "http://libvirt.org/schemas/network/dnsmasq/1.0"

90
static virNetworkDriverStatePtr network_driver;
91

92

93 94 95 96 97 98 99 100
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;
}
101

102 103 104

static void
networkDriverLock(virNetworkDriverStatePtr driver)
105
{
106
    virMutexLock(&driver->lock);
107
}
108 109 110 111


static void
networkDriverUnlock(virNetworkDriverStatePtr driver)
112
{
113
    virMutexUnlock(&driver->lock);
114 115
}

116

117 118 119 120
static dnsmasqCapsPtr
networkGetDnsmasqCaps(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr ret;
121
    networkDriverLock(driver);
122
    ret = virObjectRef(driver->dnsmasqCaps);
123
    networkDriverUnlock(driver);
124 125 126
    return ret;
}

127

128 129 130 131 132 133 134 135
static int
networkDnsmasqCapsRefresh(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr caps;

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

136
    networkDriverLock(driver);
137 138
    virObjectUnref(driver->dnsmasqCaps);
    driver->dnsmasqCaps = caps;
139
    networkDriverUnlock(driver);
140 141 142
    return 0;
}

143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250

static void
networkDnsmasqDefNamespaceFree(void *nsdata)
{
    networkDnsmasqXmlNsDefPtr def = nsdata;
    if (!def)
        return;

    virStringListFreeCount(def->options, def->noptions);

    VIR_FREE(def);
}


static int
networkDnsmasqDefNamespaceParseOptions(networkDnsmasqXmlNsDefPtr nsdef,
                                       xmlXPathContextPtr ctxt)
{
    VIR_AUTOFREE(xmlNodePtr *) nodes = NULL;
    ssize_t nnodes;
    size_t i;

    if ((nnodes = virXPathNodeSet("./dnsmasq:options/dnsmasq:option",
                                  ctxt, &nodes)) < 0)
        return -1;

    if (nnodes == 0)
        return 0;

    if (VIR_ALLOC_N(nsdef->options, nnodes) < 0)
        return -1;

    for (i = 0; i < nnodes; i++) {
        if (!(nsdef->options[nsdef->noptions++] = virXMLPropString(nodes[i], "value"))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("No dnsmasq options value specified"));
            return -1;
        }
    }

    return 0;
}


static int
networkDnsmasqDefNamespaceParse(xmlXPathContextPtr ctxt,
                                void **data)
{
    networkDnsmasqXmlNsDefPtr nsdata = NULL;
    int ret = -1;

    if (xmlXPathRegisterNs(ctxt, BAD_CAST "dnsmasq",
                           BAD_CAST DNSMASQ_NAMESPACE_HREF) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to register xml namespace '%s'"),
                       DNSMASQ_NAMESPACE_HREF);
        return -1;
    }

    if (VIR_ALLOC(nsdata) < 0)
        return -1;

    if (networkDnsmasqDefNamespaceParseOptions(nsdata, ctxt))
        goto cleanup;

    if (nsdata->noptions > 0)
        VIR_STEAL_PTR(*data, nsdata);

    ret = 0;

 cleanup:
    networkDnsmasqDefNamespaceFree(nsdata);
    return ret;
}


static int
networkDnsmasqDefNamespaceFormatXML(virBufferPtr buf,
                                    void *nsdata)
{
    networkDnsmasqXmlNsDefPtr def = nsdata;
    size_t i;

    if (!def->noptions)
        return 0;

    virBufferAddLit(buf, "<dnsmasq:options>\n");
    virBufferAdjustIndent(buf, 2);

    for (i = 0; i < def->noptions; i++) {
        virBufferEscapeString(buf, "<dnsmasq:option value='%s'/>\n",
                              def->options[i]);
    }

    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</dnsmasq:options>\n");

    return 0;
}


static const char *
networkDnsmasqDefNamespaceHref(void)
{
    return "xmlns:dnsmasq='" DNSMASQ_NAMESPACE_HREF "'";
}


251
virXMLNamespace networkDnsmasqXMLNamespace = {
252 253 254
    .parse = networkDnsmasqDefNamespaceParse,
    .free = networkDnsmasqDefNamespaceFree,
    .format = networkDnsmasqDefNamespaceFormatXML,
255
    .prefix = "dnsmasq",
256 257 258 259 260
    .href = networkDnsmasqDefNamespaceHref,
};


virNetworkXMLOptionPtr
C
Cole Robinson 已提交
261 262
networkDnsmasqCreateXMLConf(void)
{
263
    return virNetworkXMLOptionNew(&networkDnsmasqXMLNamespace);
C
Cole Robinson 已提交
264 265
}

266

267 268 269 270 271
static int
networkStateCleanup(void);

static int
networkStartNetwork(virNetworkDriverStatePtr driver,
272
                    virNetworkObjPtr obj);
273 274 275

static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
276
                       virNetworkObjPtr obj);
277 278 279

static int
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
280
                           virNetworkObjPtr obj);
281 282 283

static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
284
                              virNetworkObjPtr obj);
285

286
static int
287
networkStartNetworkExternal(virNetworkObjPtr obj);
288

289
static int
290
networkShutdownNetworkExternal(virNetworkObjPtr obj);
291

292
static void
293
networkReloadFirewallRules(virNetworkDriverStatePtr driver, bool startup);
294

295 296
static void
networkRefreshDaemons(virNetworkDriverStatePtr driver);
297

298
static int
299
networkPlugBandwidth(virNetworkObjPtr obj,
300 301 302
                     virMacAddrPtr mac,
                     virNetDevBandwidthPtr ifaceBand,
                     unsigned int *class_id);
303

304
static int
305
networkUnplugBandwidth(virNetworkObjPtr obj,
306 307
                       virNetDevBandwidthPtr ifaceBand,
                       unsigned int *class_id);
308

309
static void
310
networkNetworkObjTaint(virNetworkObjPtr obj,
311
                       virNetworkTaintFlags taint);
312

313

314 315 316
static virNetworkObjPtr
networkObjFromNetwork(virNetworkPtr net)
{
317
    virNetworkDriverStatePtr driver = networkGetDriver();
318
    virNetworkObjPtr obj;
319 320
    char uuidstr[VIR_UUID_STRING_BUFLEN];

321 322
    obj = virNetworkObjFindByUUID(driver->networks, net->uuid);
    if (!obj) {
323 324 325 326 327 328
        virUUIDFormat(net->uuid, uuidstr);
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching uuid '%s' (%s)"),
                       uuidstr, net->name);
    }

329
    return obj;
330 331
}

332

333
static int
334
networkRunHook(virNetworkObjPtr obj,
335
               virNetworkPortDefPtr port,
336 337 338
               int op,
               int sub_op)
{
339
    virNetworkDefPtr def;
340
    virBuffer buf = VIR_BUFFER_INITIALIZER;
341
    char *xml = NULL;
342 343 344 345
    int hookret;
    int ret = -1;

    if (virHookPresent(VIR_HOOK_DRIVER_NETWORK)) {
346 347
        if (!obj) {
            VIR_DEBUG("Not running hook as @obj is NULL");
348 349 350
            ret = 0;
            goto cleanup;
        }
351
        def = virNetworkObjGetDef(obj);
352

353 354
        virBufferAddLit(&buf, "<hookData>\n");
        virBufferAdjustIndent(&buf, 2);
355
        if (virNetworkDefFormatBuf(&buf, def, network_driver->xmlopt, 0) < 0)
356
            goto cleanup;
357
        if (port && virNetworkPortDefFormatBuf(&buf, port) < 0)
358 359 360 361 362
            goto cleanup;

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

363
        if (virBufferCheckError(&buf) < 0)
364 365
            goto cleanup;

366
        xml = virBufferContentAndReset(&buf);
367
        hookret = virHookCall(VIR_HOOK_DRIVER_NETWORK, def->name,
368 369 370 371 372 373 374
                              op, sub_op, NULL, xml, NULL);

        /*
         * If the script raised an error, pass it to the callee.
         */
        if (hookret < 0)
            goto cleanup;
375

376
        networkNetworkObjTaint(obj, VIR_NETWORK_TAINT_HOOK);
377 378 379
    }

    ret = 0;
380
 cleanup:
381 382 383 384 385
    virBufferFreeAndReset(&buf);
    VIR_FREE(xml);
    return ret;
}

386

387
static char *
388 389
networkDnsmasqLeaseFileNameDefault(virNetworkDriverStatePtr driver,
                                   const char *netname)
390 391 392
{
    char *leasefile;

393
    ignore_value(virAsprintf(&leasefile, "%s/%s.leases",
394
                             driver->dnsmasqStateDir, netname));
395 396 397
    return leasefile;
}

398

399
static char *
400 401
networkDnsmasqLeaseFileNameCustom(virNetworkDriverStatePtr driver,
                                  const char *bridge)
402 403 404 405
{
    char *leasefile;

    ignore_value(virAsprintf(&leasefile, "%s/%s.status",
406
                             driver->dnsmasqStateDir, bridge));
407 408 409
    return leasefile;
}

410

411
static char *
412 413
networkDnsmasqConfigFileName(virNetworkDriverStatePtr driver,
                             const char *netname)
414 415 416
{
    char *conffile;

417
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
418
                             driver->dnsmasqStateDir, netname));
419 420 421
    return conffile;
}

422

423 424 425 426 427 428
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

429
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
430 431 432
    return pidfilebase;
}

433

434
static char *
435 436
networkRadvdConfigFileName(virNetworkDriverStatePtr driver,
                           const char *netname)
437 438 439
{
    char *configfile;

440
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
441
                             driver->radvdStateDir, netname));
442 443
    return configfile;
}
444

445

446 447
/* do needed cleanup steps and remove the network from the list */
static int
448
networkRemoveInactive(virNetworkDriverStatePtr driver,
449
                      virNetworkObjPtr obj)
450 451
{
    char *leasefile = NULL;
452
    char *customleasefile = NULL;
453
    char *radvdconfigfile = NULL;
454
    char *configfile = NULL;
455
    char *radvdpidbase = NULL;
456
    char *statusfile = NULL;
M
Michal Privoznik 已提交
457
    char *macMapFile = NULL;
458
    dnsmasqContext *dctx = NULL;
459
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(obj);
460 461 462 463

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
464
    if (!(dctx = dnsmasqContextNew(def->name,
465
                                   driver->dnsmasqStateDir))) {
466
        goto cleanup;
467
    }
468

469
    if (!(leasefile = networkDnsmasqLeaseFileNameDefault(driver, def->name)))
470 471
        goto cleanup;

472
    if (!(customleasefile = networkDnsmasqLeaseFileNameCustom(driver, def->bridge)))
473 474
        goto cleanup;

475
    if (!(radvdconfigfile = networkRadvdConfigFileName(driver, def->name)))
476
        goto cleanup;
477 478

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
479
        goto cleanup;
480

481
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
482
        goto cleanup;
483

484
    if (!(statusfile = virNetworkConfigFile(driver->stateDir, def->name)))
485
        goto cleanup;
486

487
    if (!(macMapFile = virMacMapFileName(driver->dnsmasqStateDir, def->bridge)))
M
Michal Privoznik 已提交
488 489
        goto cleanup;

490 491 492
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
493
    unlink(customleasefile);
494
    unlink(configfile);
495

M
Michal Privoznik 已提交
496 497 498
    /* MAC map manager */
    unlink(macMapFile);

499 500
    /* radvd */
    unlink(radvdconfigfile);
501
    virPidFileDelete(driver->pidDir, radvdpidbase);
502

503 504 505
    /* remove status file */
    unlink(statusfile);

506
    /* remove the network definition */
507
    virNetworkObjRemoveInactive(driver->networks, obj);
508 509 510

    ret = 0;

511
 cleanup:
512
    VIR_FREE(leasefile);
513
    VIR_FREE(configfile);
514
    VIR_FREE(customleasefile);
515 516
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
517
    VIR_FREE(statusfile);
M
Michal Privoznik 已提交
518
    VIR_FREE(macMapFile);
519 520 521 522
    dnsmasqContextFree(dctx);
    return ret;
}

523

524 525 526
static char *
networkBridgeDummyNicName(const char *brname)
{
527
    static const char dummyNicSuffix[] = "-nic";
528 529
    char *nicname;

530 531 532 533 534 535 536
    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.
         */
537 538 539 540 541
        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));
542
    } else {
543
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
544
    }
545 546 547
    return nicname;
}

548

549 550
static int
networkUpdateState(virNetworkObjPtr obj,
551
                   void *opaque)
552
{
553
    virNetworkDefPtr def;
554
    virNetworkDriverStatePtr driver = opaque;
555
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
556
    virMacMapPtr macmap;
557
    char *macMapFile = NULL;
558
    int ret = -1;
559

560
    virObjectLock(obj);
561
    if (!virNetworkObjIsActive(obj)) {
562 563
        ret = 0;
        goto cleanup;
564
    }
565
    def = virNetworkObjGetDef(obj);
566

567
    switch ((virNetworkForwardType) def->forward.type) {
568 569 570
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
571
    case VIR_NETWORK_FORWARD_OPEN:
572
        /* If bridge doesn't exist, then mark it inactive */
573
        if (!(def->bridge && virNetDevExists(def->bridge) == 1))
574
            virNetworkObjSetActive(obj, false);
575

576
        if (!(macMapFile = virMacMapFileName(driver->dnsmasqStateDir,
577
                                             def->bridge)))
578 579
            goto cleanup;

580
        if (!(macmap = virMacMapNew(macMapFile)))
581 582
            goto cleanup;

583 584
        virNetworkObjSetMacMap(obj, macmap);

585
        break;
586

587
    case VIR_NETWORK_FORWARD_BRIDGE:
588 589
        if (def->bridge) {
            if (virNetDevExists(def->bridge) != 1)
590
                virNetworkObjSetActive(obj, false);
591
            break;
592
        }
593 594 595 596 597 598 599 600 601
        /* 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;
602

603 604 605
    case VIR_NETWORK_FORWARD_HOSTDEV:
        /* so far no extra checks */
        break;
606 607 608 609 610

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        goto cleanup;
611
    }
612

613
    /* Try and read dnsmasq/radvd pids of active networks */
614
    if (virNetworkObjIsActive(obj) && def->ips && (def->nips > 0)) {
615 616
        pid_t radvdPid;
        pid_t dnsmasqPid;
617
        char *radvdpidbase;
618

619
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
620
                                           def->name,
621
                                           &dnsmasqPid,
622
                                           dnsmasqCapsGetBinaryPath(dnsmasq_caps)));
623
        virNetworkObjSetDnsmasqPid(obj, dnsmasqPid);
M
Michal Privoznik 已提交
624

625
        radvdpidbase = networkRadvdPidfileBasename(def->name);
626 627
        if (!radvdpidbase)
            goto cleanup;
628

629 630
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           radvdpidbase,
631 632
                                           &radvdPid, RADVD));
        virNetworkObjSetRadvdPid(obj, radvdPid);
633
        VIR_FREE(radvdpidbase);
634
    }
635

636 637
    ret = 0;
 cleanup:
638
    virObjectUnlock(obj);
639
    virObjectUnref(dnsmasq_caps);
640
    VIR_FREE(macMapFile);
641 642
    return ret;
}
643

644

645
static int
646
networkAutostartConfig(virNetworkObjPtr obj,
647
                       void *opaque)
648
{
649
    virNetworkDriverStatePtr driver = opaque;
650
    int ret = -1;
651

652
    virObjectLock(obj);
653
    if (virNetworkObjIsAutostart(obj) &&
654 655
        !virNetworkObjIsActive(obj) &&
        networkStartNetwork(driver, obj) < 0)
656 657 658 659
        goto cleanup;

    ret = 0;
 cleanup:
660
    virObjectUnlock(obj);
661
    return ret;
662 663
}

664

665
#ifdef WITH_FIREWALLD
666 667
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
668 669
                             DBusMessage *message,
                             void *user_data)
670
{
671
    virNetworkDriverStatePtr driver = user_data;
672
    bool reload = false;
673

674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697
    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) {
698
        VIR_DEBUG("Reload in bridge_driver because of firewalld.");
699
        networkReloadFirewallRules(driver, false);
700 701 702 703 704 705
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

706

707
/**
708
 * networkStateInitialize:
709
 *
J
Ján Tomko 已提交
710
 * Initialization function for the QEMU daemon
711 712
 */
static int
713 714 715
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
716
{
717
    int ret = VIR_DRV_STATE_INIT_ERROR;
718 719
    char *configdir = NULL;
    char *rundir = NULL;
720
#ifdef WITH_FIREWALLD
721 722
    DBusConnection *sysbus = NULL;
#endif
723

724
    if (VIR_ALLOC(network_driver) < 0)
725
        goto error;
726

727
    network_driver->lockFD = -1;
728 729
    if (virMutexInit(&network_driver->lock) < 0) {
        VIR_FREE(network_driver);
730 731
        goto error;
    }
732

733 734
    network_driver->privileged = privileged;

C
Cole Robinson 已提交
735 736 737
    if (!(network_driver->xmlopt = networkDnsmasqCreateXMLConf()))
        goto error;

738 739 740 741
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
742
    if (privileged) {
743
        if (VIR_STRDUP(network_driver->networkConfigDir,
744
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
745
            VIR_STRDUP(network_driver->networkAutostartDir,
746
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
747
            VIR_STRDUP(network_driver->stateDir,
748
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
749
            VIR_STRDUP(network_driver->pidDir,
750
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
751
            VIR_STRDUP(network_driver->dnsmasqStateDir,
752
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
753
            VIR_STRDUP(network_driver->radvdStateDir,
754 755
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
756
    } else {
757 758 759
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
760
            goto error;
761

762
        if ((virAsprintf(&network_driver->networkConfigDir,
763
                         "%s/qemu/networks", configdir) < 0) ||
764
            (virAsprintf(&network_driver->networkAutostartDir,
765
                         "%s/qemu/networks/autostart", configdir) < 0) ||
766
            (virAsprintf(&network_driver->stateDir,
767
                         "%s/network/lib", rundir) < 0) ||
768
            (virAsprintf(&network_driver->pidDir,
769
                         "%s/network/run", rundir) < 0) ||
770
            (virAsprintf(&network_driver->dnsmasqStateDir,
771
                         "%s/dnsmasq/lib", rundir) < 0) ||
772
            (virAsprintf(&network_driver->radvdStateDir,
773
                         "%s/radvd/lib", rundir) < 0)) {
774
            goto error;
775
        }
776 777
    }

778
    if (virFileMakePath(network_driver->stateDir) < 0) {
779 780
        virReportSystemError(errno,
                             _("cannot create directory %s"),
781
                             network_driver->stateDir);
782 783 784
        goto error;
    }

785 786 787 788 789
    if ((network_driver->lockFD =
         virPidFileAcquire(network_driver->stateDir, "driver",
                           true, getpid())) < 0)
        goto error;

790
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
791
    network_driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
792

793
    if (!(network_driver->networks = virNetworkObjListNew()))
794 795
        goto error;

796
    if (virNetworkObjLoadAllState(network_driver->networks,
797 798
                                  network_driver->stateDir,
                                  network_driver->xmlopt) < 0)
799 800
        goto error;

801 802
    if (virNetworkObjLoadAllConfigs(network_driver->networks,
                                    network_driver->networkConfigDir,
803 804
                                    network_driver->networkAutostartDir,
                                    network_driver->xmlopt) < 0)
805 806
        goto error;

807 808 809 810
    /* 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). */
811
    virNetworkObjListForEach(network_driver->networks,
812
                             networkUpdateState,
813 814
                             network_driver);
    virNetworkObjListPrune(network_driver->networks,
815 816
                           VIR_CONNECT_LIST_NETWORKS_INACTIVE |
                           VIR_CONNECT_LIST_NETWORKS_TRANSIENT);
817
    networkReloadFirewallRules(network_driver, true);
818
    networkRefreshDaemons(network_driver);
819

820 821 822 823
    virNetworkObjListForEach(network_driver->networks,
                             networkAutostartConfig,
                             network_driver);

824
    network_driver->networkEventState = virObjectEventStateNew();
825

826
#ifdef WITH_FIREWALLD
827 828
    if (!(sysbus = virDBusGetSystemBus())) {
        VIR_WARN("DBus not available, disabling firewalld support "
829
                 "in bridge_network_driver: %s", virGetLastErrorMessage());
830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846
    } 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,
847
                                   network_driver, NULL);
848 849 850
    }
#endif

851
    ret = VIR_DRV_STATE_INIT_COMPLETE;
852
 cleanup:
853 854 855
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
856

857
 error:
858
    networkStateCleanup();
859
    goto cleanup;
860 861
}

862

863
/**
864
 * networkStateReload:
865
 *
J
Ján Tomko 已提交
866
 * Function to restart the QEMU daemon, it will recheck the configuration
867 868 869
 * files and update its state and the networking
 */
static int
870 871
networkStateReload(void)
{
872
    if (!network_driver)
873 874
        return 0;

875
    virNetworkObjLoadAllState(network_driver->networks,
876 877
                              network_driver->stateDir,
                              network_driver->xmlopt);
878 879
    virNetworkObjLoadAllConfigs(network_driver->networks,
                                network_driver->networkConfigDir,
880 881
                                network_driver->networkAutostartDir,
                                network_driver->xmlopt);
882
    networkReloadFirewallRules(network_driver, false);
883 884
    networkRefreshDaemons(network_driver);
    virNetworkObjListForEach(network_driver->networks,
885
                             networkAutostartConfig,
886
                             network_driver);
887 888 889 890 891
    return 0;
}


/**
892
 * networkStateCleanup:
893
 *
J
Ján Tomko 已提交
894
 * Shutdown the QEMU daemon, it will stop all active domains and networks
895 896
 */
static int
897 898
networkStateCleanup(void)
{
899
    if (!network_driver)
900 901
        return -1;

902
    virObjectUnref(network_driver->networkEventState);
C
Cole Robinson 已提交
903
    virObjectUnref(network_driver->xmlopt);
904

905
    /* free inactive networks */
906
    virObjectUnref(network_driver->networks);
907

908 909 910 911
    if (network_driver->lockFD != -1)
        virPidFileRelease(network_driver->stateDir, "driver",
                          network_driver->lockFD);

912 913 914 915 916 917
    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);
918

919
    virObjectUnref(network_driver->dnsmasqCaps);
920

921
    virMutexDestroy(&network_driver->lock);
922

923
    VIR_FREE(network_driver);
924 925 926 927 928

    return 0;
}


929 930 931 932 933 934 935 936
static virDrvOpenStatus
networkConnectOpen(virConnectPtr conn,
                   virConnectAuthPtr auth ATTRIBUTE_UNUSED,
                   virConfPtr conf ATTRIBUTE_UNUSED,
                   unsigned int flags)
{
    virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);

937 938 939 940 941 942 943 944 945 946 947
    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);
948 949
            return VIR_DRV_OPEN_ERROR;
        }
950 951 952 953 954 955
    } 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;
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
        }
    }

    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;
}


991 992 993 994 995
/* networkKillDaemon:
 *
 * kill the specified pid/name, and wait a bit to make sure it's dead.
 */
static int
996 997 998
networkKillDaemon(pid_t pid,
                  const char *daemonName,
                  const char *networkName)
999
{
1000 1001
    size_t i;
    int ret = -1;
1002 1003 1004 1005 1006 1007 1008
    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.
     */
1009
    for (i = 0; i < 25; i++) {
1010
        int signum = 0;
1011
        if (i == 0) {
1012
            signum = SIGTERM;
1013
        } else if (i == 15) {
1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048
            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);
1049
 cleanup:
1050 1051 1052
    return ret;
}

1053

J
Ján Tomko 已提交
1054 1055 1056
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
1057
static int
G
Gene Czarcinski 已提交
1058
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
1059
                                 virNetworkIPDefPtr ipdef)
1060
{
1061
    size_t i;
G
Gene Czarcinski 已提交
1062
    bool ipv6 = false;
1063

G
Gene Czarcinski 已提交
1064 1065
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
1066 1067
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
1068
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
1069 1070
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
1071
                return -1;
1072
    }
1073

G
Gene Czarcinski 已提交
1074 1075 1076
    return 0;
}

1077

G
Gene Czarcinski 已提交
1078 1079 1080 1081
static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
1082
    size_t i, j;
G
Gene Czarcinski 已提交
1083

1084 1085
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
1086
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
1087
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
1088
                for (j = 0; j < host->nnames; j++)
1089 1090
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
1091 1092
            }
        }
1093 1094
    }

1095
    return 0;
1096 1097 1098
}


1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135
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;
}


1136
int
1137
networkDnsmasqConfContents(virNetworkObjPtr obj,
1138 1139 1140 1141
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
1142
{
1143
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1144
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1145
    int r, ret = -1;
1146
    int nbleases = 0;
1147
    size_t i;
1148
    virNetworkDNSDefPtr dns = &def->dns;
1149
    bool wantDNS = dns->enable != VIR_TRISTATE_BOOL_NO;
1150
    virNetworkIPDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
G
Gene Czarcinski 已提交
1151
    bool ipv6SLAAC;
1152
    char *saddr = NULL, *eaddr = NULL;
1153

1154 1155
    *configstr = NULL;

1156
    /*
1157 1158 1159
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
1160
     *
1161 1162
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
1163
     */
1164 1165 1166 1167 1168 1169

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

1170
    /* create dnsmasq config file appropriate for this network */
1171 1172

    /* Don't forget to update networkxml2conftest :-) */
1173
    virBufferAsprintf(&configbuf,
1174 1175 1176 1177 1178 1179 1180
                      "##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"
1181
                      "strict-order\n",
1182
                      def->name);
1183

1184 1185 1186 1187 1188 1189
    /* if dns is disabled, set its listening port to 0, which
     * tells dnsmasq to not listen
     */
    if (!wantDNS)
        virBufferAddLit(&configbuf, "port=0\n");

1190
    if (wantDNS && def->dns.forwarders) {
1191 1192 1193 1194 1195 1196 1197 1198 1199
        /* 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;

1200 1201
        for (i = 0; i < def->dns.nfwds; i++) {
            virNetworkDNSForwarderPtr fwd = &def->dns.forwarders[i];
1202 1203 1204 1205 1206 1207 1208 1209 1210 1211

            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);
1212
                VIR_FREE(addr);
1213 1214
                if (!fwd->domain)
                    addNoResolv = true;
1215 1216 1217 1218
            } else {
                /* "don't forward requests for this domain" */
                virBufferAddLit(&configbuf, "#\n");
            }
1219
        }
1220 1221
        if (addNoResolv)
            virBufferAddLit(&configbuf, "no-resolv\n");
1222 1223
    }

1224 1225
    if (def->domain) {
        if (def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
1226 1227
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
1228
                              def->domain);
1229
        }
1230
        virBufferAsprintf(&configbuf,
1231 1232
                          "domain=%s\n"
                          "expand-hosts\n",
1233
                          def->domain);
1234
    }
1235

1236
    if (wantDNS &&
1237
        networkDnsmasqConfLocalPTRs(&configbuf, def) < 0)
1238 1239
        goto cleanup;

1240
    if (wantDNS && def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
1241 1242 1243 1244
        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)
1245
         */
1246
        virBufferAddLit(&configbuf, "local=//\n");
1247
    }
1248

1249
    if (pidfile)
1250
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
1251

1252
    /* dnsmasq will *always* listen on localhost unless told otherwise */
P
Pavel Timofeev 已提交
1253
#ifdef __linux__
1254
    virBufferAddLit(&configbuf, "except-interface=lo\n");
P
Pavel Timofeev 已提交
1255 1256 1257 1258
#else
    /* BSD family OSes and Solaris call loopback interface as lo0 */
    virBufferAddLit(&configbuf, "except-interface=lo0\n");
#endif
1259

1260 1261 1262 1263 1264 1265 1266 1267
    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.
         */
1268
        virBufferAsprintf(&configbuf,
1269 1270
                          "bind-dynamic\n"
                          "interface=%s\n",
1271
                          def->bridge);
1272
    } else {
1273
        virBufferAddLit(&configbuf, "bind-interfaces\n");
1274 1275 1276 1277
        /*
         * --interface does not actually work with dnsmasq < 2.47,
         * due to DAD for ipv6 addresses on the interface.
         *
1278
         * virCommandAddArgList(cmd, "--interface", def->bridge, NULL);
1279 1280 1281
         *
         * So listen on all defined IPv[46] addresses
         */
1282
        for (i = 0;
1283
             (tmpipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
1284
             i++) {
1285 1286 1287 1288
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
1289

1290
            /* also part of CVE 2012-3411 - if the host's version of
1291
             * dnsmasq doesn't have bind-dynamic, only allow listening on
1292 1293
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
1294 1295
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
1296 1297 1298 1299
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
1300
                                 "The version of dnsmasq on this host (%d.%d) "
1301 1302 1303 1304
                                 "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 "
1305 1306 1307 1308 1309 1310
                                 "(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);
1311
                VIR_FREE(ipaddr);
1312 1313
                goto cleanup;
            }
1314
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1315 1316 1317
            VIR_FREE(ipaddr);
        }
    }
1318

1319 1320
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1321
     * guaranteed to not work, and set no-resolv so that no dns
1322 1323 1324
     * 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).
1325 1326 1327
     * 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.
1328
     */
1329
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1330
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1331
                        "no-resolv\n");
1332 1333 1334 1335
        if (dnsmasqCapsGet(caps, DNSMASQ_CAPS_RA_PARAM)) {
            /* interface=* (any), interval=0 (default), lifetime=0 (seconds) */
            virBufferAddLit(&configbuf, "ra-param=*,0,0\n");
        }
1336
    }
1337

1338 1339 1340 1341 1342
    if (wantDNS) {
        for (i = 0; i < dns->ntxts; i++) {
            virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
                              dns->txts[i].name,
                              dns->txts[i].value);
1343
        }
1344

1345 1346 1347 1348 1349 1350 1351 1352
        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'"),
1353
                               def->name);
1354 1355 1356 1357 1358 1359
                goto cleanup;
            }
            if (!dns->srvs[i].protocol) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Missing required 'service' "
                                 "attribute in SRV record of network '%s'"),
1360
                               def->name);
1361 1362 1363 1364 1365 1366 1367
                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);
1368

1369 1370 1371 1372 1373 1374 1375
            /* 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.
1376
             */
1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396
            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");
1397
        }
1398 1399
    }

G
Gene Czarcinski 已提交
1400
    /* Find the first dhcp for both IPv4 and IPv6 */
1401
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
1402
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
1403
         i++) {
G
Gene Czarcinski 已提交
1404 1405 1406 1407
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1408 1409
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420
                    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,
1421 1422 1423 1424 1425 1426 1427 1428 1429
                                   _("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
1430 1431 1432 1433
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1434 1435
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448
                    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 "
1449 1450 1451 1452
                 "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
1453 1454 1455 1456 1457
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1458 1459
        int prefix;

1460
        prefix = virNetworkIPDefPrefix(ipdef);
1461 1462 1463
        if (prefix < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
1464
                           def->bridge);
1465 1466
            goto cleanup;
        }
1467
        for (r = 0; r < ipdef->nranges; r++) {
1468 1469
            int thisRange;

1470 1471
            if (!(saddr = virSocketAddrFormat(&ipdef->ranges[r].start)) ||
                !(eaddr = virSocketAddrFormat(&ipdef->ranges[r].end)))
1472
                goto cleanup;
1473

1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494
            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);
            }
1495

1496
            VIR_FREE(saddr);
1497
            VIR_FREE(eaddr);
1498
            thisRange = virSocketAddrGetRange(&ipdef->ranges[r].start,
1499 1500
                                              &ipdef->ranges[r].end,
                                              &ipdef->address,
1501
                                              virNetworkIPDefPrefix(ipdef));
1502 1503 1504
            if (thisRange < 0)
                goto cleanup;
            nbleases += thisRange;
1505
        }
1506

1507
        /*
1508 1509 1510 1511
         * 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)
1512 1513
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1514
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1515 1516
            if (!bridgeaddr)
                goto cleanup;
1517 1518 1519
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static",
                              bridgeaddr);
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1520
                virBufferAsprintf(&configbuf, ",%d", prefix);
1521
            virBufferAddLit(&configbuf, "\n");
1522 1523
            VIR_FREE(bridgeaddr);
        }
1524

G
Gene Czarcinski 已提交
1525 1526
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1527

G
Gene Czarcinski 已提交
1528 1529
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
1530
            if (ipdef->nranges || ipdef->nhosts) {
1531
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1532 1533
                virBufferAddLit(&configbuf, "dhcp-authoritative\n");
            }
1534

G
Gene Czarcinski 已提交
1535
            if (ipdef->tftproot) {
1536 1537
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1538
            }
1539

G
Gene Czarcinski 已提交
1540 1541 1542
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1543

1544
                    if (!bootserver)
G
Gene Czarcinski 已提交
1545
                        goto cleanup;
1546
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1547
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1548 1549
                    VIR_FREE(bootserver);
                } else {
1550
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1551 1552 1553 1554 1555
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1556

1557
    if (nbleases > 0)
1558
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1559

G
Gene Czarcinski 已提交
1560 1561
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1562
        goto cleanup;
G
Gene Czarcinski 已提交
1563 1564 1565 1566 1567 1568

    /* 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)
1569 1570
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1571

1572 1573
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1574
     */
1575 1576 1577 1578
    if (wantDNS) {
        virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
                          dctx->addnhostsfile->path);
    }
G
Gene Czarcinski 已提交
1579

1580 1581 1582 1583
    /* 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
1584 1585
    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1586
        if (ipv6def) {
1587
            virBufferAddLit(&configbuf, "enable-ra\n");
1588
        } else {
1589
            for (i = 0;
1590
                 (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i));
1591
                 i++) {
G
Gene Czarcinski 已提交
1592 1593 1594 1595
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1596 1597
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1598 1599
                    VIR_FREE(bridgeaddr);
                }
1600
            }
1601
        }
1602 1603
    }

1604 1605 1606 1607 1608 1609
    if (def->namespaceData) {
        networkDnsmasqXmlNsDefPtr dnsmasqxmlns = def->namespaceData;
        for (i = 0; i < dnsmasqxmlns->noptions; i++)
            virBufferAsprintf(&configbuf, "%s\n", dnsmasqxmlns->options[i]);
    }

1610 1611 1612
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1613
    ret = 0;
G
Gene Czarcinski 已提交
1614

1615
 cleanup:
1616 1617
    VIR_FREE(saddr);
    VIR_FREE(eaddr);
1618
    virBufferFreeAndReset(&configbuf);
1619
    return ret;
1620 1621
}

1622

1623
/* build the dnsmasq command line */
1624 1625
static int ATTRIBUTE_NONNULL(3)
networkBuildDhcpDaemonCommandLine(virNetworkDriverStatePtr driver,
1626
                                  virNetworkObjPtr obj,
1627
                                  virCommandPtr *cmdout,
1628 1629
                                  char *pidfile,
                                  dnsmasqContext *dctx)
1630
{
1631
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1632
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1633
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1634
    int ret = -1;
1635 1636
    char *configfile = NULL;
    char *configstr = NULL;
1637
    char *leaseshelper_path = NULL;
1638

1639
    virNetworkObjSetDnsmasqPid(obj, -1);
1640

1641
    if (networkDnsmasqConfContents(obj, pidfile, &configstr,
1642
                                   dctx, dnsmasq_caps) < 0)
1643 1644 1645 1646 1647
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1648
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
1649 1650 1651 1652 1653
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1654 1655
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1656 1657 1658
        goto cleanup;
    }

1659 1660
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1661
                                                  abs_top_builddir "/src",
1662 1663 1664
                                                  LIBEXECDIR)))
        goto cleanup;

1665
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(dnsmasq_caps));
1666
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1667 1668
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1669
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1670
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", def->bridge);
1671

1672
    *cmdout = cmd;
1673
    ret = 0;
1674
 cleanup:
1675
    virObjectUnref(dnsmasq_caps);
1676 1677
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1678
    VIR_FREE(leaseshelper_path);
1679 1680 1681
    return ret;
}

1682

1683
static int
1684
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
1685
                       virNetworkObjPtr obj)
1686
{
1687
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1688 1689 1690
    virNetworkIPDefPtr ipdef;
    size_t i;
    bool needDnsmasq = false;
1691 1692
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
1693
    pid_t dnsmasqPid;
1694
    int ret = -1;
1695
    dnsmasqContext *dctx = NULL;
1696

1697
    /* see if there are any IP addresses that need a dhcp server */
1698
    i = 0;
1699
    while ((ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i))) {
1700
        i++;
1701 1702 1703 1704
        if (ipdef->nranges || ipdef->nhosts)
            needDnsmasq = true;
    }

1705 1706 1707 1708 1709 1710
    if (i == 0) {
        /* no IP addresses at all, so we don't need to run */
        ret = 0;
        goto cleanup;
    }

1711
    if (!needDnsmasq && def->dns.enable == VIR_TRISTATE_BOOL_NO) {
1712
        /* no DHCP services needed, and user disabled DNS service */
1713 1714 1715 1716
        ret = 0;
        goto cleanup;
    }

1717
    if (virFileMakePath(driver->pidDir) < 0) {
1718
        virReportSystemError(errno,
1719
                             _("cannot create directory %s"),
1720
                             driver->pidDir);
1721
        goto cleanup;
1722 1723
    }

1724
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, def->name)))
1725
        goto cleanup;
1726

1727
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1728
        virReportSystemError(errno,
1729
                             _("cannot create directory %s"),
1730
                             driver->dnsmasqStateDir);
1731 1732 1733
        goto cleanup;
    }

1734
    dctx = dnsmasqContextNew(def->name, driver->dnsmasqStateDir);
1735 1736 1737
    if (dctx == NULL)
        goto cleanup;

1738
    if (networkDnsmasqCapsRefresh(driver) < 0)
1739
        goto cleanup;
1740

1741
    ret = networkBuildDhcpDaemonCommandLine(driver, obj, &cmd, pidfile, dctx);
1742 1743 1744 1745 1746
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1747
        goto cleanup;
1748

G
Guido Günther 已提交
1749
    ret = virCommandRun(cmd, NULL);
1750
    if (ret < 0)
1751 1752 1753
        goto cleanup;

    /*
1754 1755 1756 1757 1758
     * 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
1759 1760
     */

1761
    ret = virPidFileRead(driver->pidDir, def->name, &dnsmasqPid);
1762
    if (ret < 0)
1763
        goto cleanup;
1764
    virNetworkObjSetDnsmasqPid(obj, dnsmasqPid);
1765

1766
    ret = 0;
1767
 cleanup:
1768
    VIR_FREE(pidfile);
1769
    virCommandFree(cmd);
1770
    dnsmasqContextFree(dctx);
1771 1772 1773
    return ret;
}

1774

1775 1776
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1777 1778
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1779 1780 1781
 *
 *  Returns 0 on success, -1 on failure.
 */
1782
static int
1783
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
1784
                         virNetworkObjPtr obj)
1785
{
1786
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1787 1788
    int ret = -1;
    size_t i;
1789
    pid_t dnsmasqPid;
1790
    virNetworkIPDefPtr ipdef, ipv4def, ipv6def;
1791
    dnsmasqContext *dctx = NULL;
1792

G
Gene Czarcinski 已提交
1793
    /* if no IP addresses specified, nothing to do */
1794
    if (!virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1795 1796
        return 0;

1797
    /* if there's no running dnsmasq, just start it */
1798 1799
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid <= 0 || (kill(dnsmasqPid, 0) < 0))
1800
        return networkStartDhcpDaemon(driver, obj);
1801

1802 1803
    VIR_INFO("Refreshing dnsmasq for network %s", def->bridge);
    if (!(dctx = dnsmasqContextNew(def->name,
1804
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1805
        goto cleanup;
1806
    }
G
Gene Czarcinski 已提交
1807 1808 1809 1810 1811 1812

    /* 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;
1813
    for (i = 0;
1814
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
1815
         i++) {
G
Gene Czarcinski 已提交
1816 1817
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1818 1819
    }

G
Gene Czarcinski 已提交
1820
    ipv6def = NULL;
1821
    for (i = 0;
1822
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i));
1823
         i++) {
G
Gene Czarcinski 已提交
1824 1825
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1826 1827
    }

G
Gene Czarcinski 已提交
1828
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1829
        goto cleanup;
G
Gene Czarcinski 已提交
1830 1831

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1832
        goto cleanup;
1833

1834
    if (networkBuildDnsmasqHostsList(dctx, &def->dns) < 0)
J
Ján Tomko 已提交
1835
        goto cleanup;
1836 1837

    if ((ret = dnsmasqSave(dctx)) < 0)
1838
        goto cleanup;
1839

1840 1841
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    ret = kill(dnsmasqPid, SIGHUP);
1842
 cleanup:
1843 1844 1845 1846
    dnsmasqContextFree(dctx);
    return ret;
}

1847

1848 1849 1850 1851 1852 1853 1854 1855
/* 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
1856
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
1857
                         virNetworkObjPtr obj)
1858
{
1859
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1860 1861
    pid_t dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);

1862
    /* if there is a running dnsmasq, kill it */
1863
    if (dnsmasqPid > 0) {
1864
        networkKillDaemon(dnsmasqPid, "dnsmasq", def->name);
1865
        virNetworkObjSetDnsmasqPid(obj, -1);
1866
    }
1867
    /* now start dnsmasq if it should be started */
1868
    return networkStartDhcpDaemon(driver, obj);
1869 1870
}

1871

G
Gene Czarcinski 已提交
1872 1873 1874 1875 1876 1877
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";

1878
static int
1879
networkRadvdConfContents(virNetworkObjPtr obj,
1880
                         char **configstr)
1881
{
1882
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
E
Eric Blake 已提交
1883
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1884 1885
    int ret = -1;
    size_t i;
1886
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
1887
    bool v6present = false, dhcp6 = false;
1888 1889

    *configstr = NULL;
1890

G
Gene Czarcinski 已提交
1891
    /* Check if DHCPv6 is needed */
1892
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i)); i++) {
G
Gene Czarcinski 已提交
1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905
        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;
    }

1906 1907 1908
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1909
    virBufferAsprintf(&configbuf, "interface %s\n"
1910 1911
                      "{\n"
                      "  AdvSendAdvert on;\n"
1912
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1913 1914
                      "  AdvManagedFlag %s;\n"
                      "%s",
1915
                      def->bridge,
G
Gene Czarcinski 已提交
1916 1917
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1918 1919

    /* add a section for each IPv6 address in the config */
1920
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i)); i++) {
1921 1922 1923
        int prefix;
        char *netaddr;

1924
        prefix = virNetworkIPDefPrefix(ipdef);
1925
        if (prefix < 0) {
1926 1927
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
1928
                           def->bridge);
1929 1930
            goto cleanup;
        }
1931
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1932
            goto cleanup;
1933
        virBufferAsprintf(&configbuf,
1934
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1935 1936 1937
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1938 1939 1940
        VIR_FREE(netaddr);
    }

1941
    virBufferAddLit(&configbuf, "};\n");
1942

1943
    if (virBufferCheckError(&configbuf) < 0)
1944
        goto cleanup;
1945

1946 1947
    *configstr = virBufferContentAndReset(&configbuf);

1948
    ret = 0;
1949
 cleanup:
1950 1951 1952 1953
    virBufferFreeAndReset(&configbuf);
    return ret;
}

1954

1955
/* write file and return its name (which must be freed by caller) */
1956
static int
1957
networkRadvdConfWrite(virNetworkDriverStatePtr driver,
1958
                      virNetworkObjPtr obj,
1959
                      char **configFile)
1960
{
1961
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1962 1963 1964 1965 1966 1967 1968 1969 1970
    int ret = -1;
    char *configStr = NULL;
    char *myConfigFile = NULL;

    if (!configFile)
        configFile = &myConfigFile;

    *configFile = NULL;

1971
    if (networkRadvdConfContents(obj, &configStr) < 0)
1972 1973 1974 1975
        goto cleanup;

    if (!configStr) {
        ret = 0;
1976 1977 1978 1979
        goto cleanup;
    }

    /* construct the filename */
1980
    if (!(*configFile = networkRadvdConfigFileName(driver, def->name)))
1981 1982
        goto cleanup;
    /* write the file */
1983
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1984 1985
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1986 1987 1988 1989 1990
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1991
 cleanup:
1992 1993 1994 1995 1996
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

1997

1998
static int
1999
networkStartRadvd(virNetworkDriverStatePtr driver,
2000
                  virNetworkObjPtr obj)
2001
{
2002
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2003
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
2004
    pid_t radvdPid;
2005 2006 2007 2008 2009 2010
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

2011
    virNetworkObjSetRadvdPid(obj, -1);
2012

G
Gene Czarcinski 已提交
2013
    /* Is dnsmasq handling RA? */
2014
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
G
Gene Czarcinski 已提交
2015 2016 2017 2018
        ret = 0;
        goto cleanup;
    }

2019
    if (!virNetworkDefGetIPByIndex(def, AF_INET6, 0)) {
2020 2021 2022 2023 2024 2025 2026 2027 2028 2029
        /* 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);
2030 2031 2032
        goto cleanup;
    }

2033
    if (virFileMakePath(driver->pidDir) < 0) {
2034 2035
        virReportSystemError(errno,
                             _("cannot create directory %s"),
2036
                             driver->pidDir);
2037 2038
        goto cleanup;
    }
2039
    if (virFileMakePath(driver->radvdStateDir) < 0) {
2040 2041
        virReportSystemError(errno,
                             _("cannot create directory %s"),
2042
                             driver->radvdStateDir);
2043 2044 2045 2046
        goto cleanup;
    }

    /* construct pidfile name */
2047
    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
2048
        goto cleanup;
2049
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
2050 2051
        goto cleanup;

2052
    if (networkRadvdConfWrite(driver, obj, &configfile) < 0)
2053 2054
        goto cleanup;

2055 2056 2057 2058
    /* 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
2059
     * virPidFileRead() below will fail if we use them).
2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074
     * 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;

2075
    if (virPidFileRead(driver->pidDir, radvdpidbase, &radvdPid) < 0)
2076
        goto cleanup;
2077
    virNetworkObjSetRadvdPid(obj, radvdPid);
2078 2079

    ret = 0;
2080
 cleanup:
2081
    virObjectUnref(dnsmasq_caps);
2082 2083 2084 2085 2086 2087 2088
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

2089

2090
static int
2091
networkRefreshRadvd(virNetworkDriverStatePtr driver,
2092
                    virNetworkObjPtr obj)
2093
{
2094
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2095
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
G
Gene Czarcinski 已提交
2096
    char *radvdpidbase;
2097
    pid_t radvdPid;
G
Gene Czarcinski 已提交
2098 2099

    /* Is dnsmasq handling RA? */
2100 2101
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
        virObjectUnref(dnsmasq_caps);
2102 2103
        radvdPid = virNetworkObjGetRadvdPid(obj);
        if (radvdPid <= 0)
G
Gene Czarcinski 已提交
2104 2105
            return 0;
        /* radvd should not be running but in case it is */
2106 2107
        if ((networkKillDaemon(radvdPid, "radvd", def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(def->name))
G
Gene Czarcinski 已提交
2108
             != NULL)) {
2109
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
2110 2111
            VIR_FREE(radvdpidbase);
        }
2112
        virNetworkObjSetRadvdPid(obj, -1);
G
Gene Czarcinski 已提交
2113 2114
        return 0;
    }
2115
    virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
2116

2117
    /* if there's no running radvd, just start it */
2118 2119
    radvdPid = virNetworkObjGetRadvdPid(obj);
    if (radvdPid <= 0 || (kill(radvdPid, 0) < 0))
2120
        return networkStartRadvd(driver, obj);
2121

2122
    if (!virNetworkDefGetIPByIndex(def, AF_INET6, 0)) {
2123 2124 2125 2126
        /* no IPv6 addresses, so we don't need to run radvd */
        return 0;
    }

2127
    if (networkRadvdConfWrite(driver, obj, NULL) < 0)
2128 2129
        return -1;

2130
    return kill(radvdPid, SIGHUP);
2131 2132
}

2133

2134 2135
#if 0
/* currently unused, so it causes a build error unless we #if it out */
2136
static int
2137
networkRestartRadvd(virNetworkObjPtr obj)
2138
{
2139
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2140
    char *radvdpidbase;
2141
    pid_t radvdPid = virNeworkObjGetRadvdPid(obj);
2142 2143

    /* if there is a running radvd, kill it */
2144
    if (radvdPid > 0) {
2145 2146 2147 2148
        /* 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).
         */
2149 2150
        if ((networkKillDaemon(radvdPid, "radvd", def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(def->name))
2151
             != NULL)) {
2152
            virPidFileDelete(driver->pidDir, radvdpidbase);
2153 2154
            VIR_FREE(radvdpidbase);
        }
2155
        virNetworkObjSetRadvdPid(obj, -1);
2156 2157
    }
    /* now start radvd if it should be started */
2158
    return networkStartRadvd(obj);
2159 2160 2161
}
#endif /* #if 0 */

2162

2163
static int
2164
networkRefreshDaemonsHelper(virNetworkObjPtr obj,
2165
                            void *opaque)
2166
{
2167
    virNetworkDefPtr def;
2168
    virNetworkDriverStatePtr driver = opaque;
2169

2170
    virObjectLock(obj);
2171
    def = virNetworkObjGetDef(obj);
2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199
    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;
        }
2200
    }
2201 2202

 cleanup:
2203
    virObjectUnlock(obj);
2204 2205 2206
    return 0;
}

2207

2208 2209 2210 2211
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
2212
networkRefreshDaemons(virNetworkDriverStatePtr driver)
2213 2214
{
    VIR_INFO("Refreshing network daemons");
2215 2216
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
2217
                             driver);
2218
}
2219

2220

2221
static int
2222
networkReloadFirewallRulesHelper(virNetworkObjPtr obj,
2223 2224
                                 void *opaque ATTRIBUTE_UNUSED)
{
2225
    virNetworkDefPtr def;
2226

2227
    virObjectLock(obj);
2228
    def = virNetworkObjGetDef(obj);
2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254
    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;
2255 2256
        }
    }
2257 2258

 cleanup:
2259
    virObjectUnlock(obj);
2260
    return 0;
2261 2262
}

2263

2264
static void
2265
networkReloadFirewallRules(virNetworkDriverStatePtr driver, bool startup)
2266
{
2267
    VIR_INFO("Reloading iptables rules");
2268 2269 2270 2271
    /* 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;
2272
    networkPreReloadFirewallRules(driver, startup);
2273 2274 2275
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
2276
    networkPostReloadFirewallRules(startup);
2277 2278
}

2279

2280
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
2281
static int
2282 2283
networkEnableIPForwarding(bool enableIPv4,
                          bool enableIPv6)
2284
{
2285
    int ret = 0;
2286 2287 2288 2289
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
2290
                           &enabled, sizeof(enabled));
2291 2292
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
2293
                           &enabled, sizeof(enabled));
2294
#else
2295
    if (enableIPv4)
2296
        ret = virFileWriteStr(SYSCTL_PATH "/net/ipv4/ip_forward", "1\n", 0);
2297
    if (enableIPv6 && ret == 0)
2298 2299
        ret = virFileWriteStr(SYSCTL_PATH "/net/ipv6/conf/all/forwarding", "1\n", 0);

2300
#endif
2301
    return ret;
2302 2303
}

2304

2305
static int
2306
networkSetIPv6Sysctls(virNetworkObjPtr obj)
2307
{
2308
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2309 2310
    char *field = NULL;
    int ret = -1;
2311
    bool enableIPv6 = !!virNetworkDefGetIPByIndex(def, AF_INET6, 0);
2312

2313 2314 2315 2316 2317
    /* 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",
2318
                    def->bridge) < 0)
2319
       goto cleanup;
2320

2321 2322
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
2323
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
2324
                      def->bridge);
2325 2326 2327
        ret = 0;
        goto cleanup;
    }
2328

2329 2330 2331
    if (virFileWriteStr(field, enableIPv6 ? "0" : "1", 0) < 0) {
        virReportSystemError(errno,
                             _("cannot write to %s to enable/disable IPv6 "
2332
                               "on bridge %s"), field, def->bridge);
2333
        goto cleanup;
2334
    }
2335
    VIR_FREE(field);
2336

2337 2338
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
2339 2340 2341 2342 2343 2344
     */

    /* 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",
2345
                    def->bridge) < 0)
2346 2347
        goto cleanup;

2348
    if (virFileWriteStr(field, "0", 0) < 0) {
2349
        virReportSystemError(errno,
2350 2351 2352 2353 2354
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

2355 2356 2357 2358
    /* 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",
2359
                    def->bridge) < 0)
2360 2361
        goto cleanup;

2362
    if (virFileWriteStr(field, "0", 0) < 0) {
2363
        virReportSystemError(errno,
2364
                             _("cannot disable %s"), field);
2365 2366 2367 2368
        goto cleanup;
    }

    ret = 0;
2369
 cleanup:
2370 2371 2372 2373
    VIR_FREE(field);
    return ret;
}

2374

2375
/* add an IP address to a bridge */
2376
static int
2377
networkAddAddrToBridge(virNetworkObjPtr obj,
2378
                       virNetworkIPDefPtr ipdef)
2379
{
2380
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2381
    int prefix = virNetworkIPDefPrefix(ipdef);
2382 2383

    if (prefix < 0) {
2384 2385
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
2386
                       def->bridge);
2387 2388 2389
        return -1;
    }

2390
    if (virNetDevIPAddrAdd(def->bridge, &ipdef->address, NULL, prefix) < 0)
2391 2392 2393 2394 2395
        return -1;

    return 0;
}

2396 2397

static int
2398
networkStartHandleMACTableManagerMode(virNetworkObjPtr obj,
2399 2400
                                      const char *macTapIfName)
{
2401 2402
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
    const char *brname = def->bridge;
2403 2404

    if (brname &&
2405
        def->macTableManager == VIR_NETWORK_BRIDGE_MAC_TABLE_MANAGER_LIBVIRT) {
2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418
        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;
}


2419 2420
/* add an IP (static) route to a bridge */
static int
2421
networkAddRouteToBridge(virNetworkObjPtr obj,
2422
                        virNetDevIPRoutePtr routedef)
2423
{
2424
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2425 2426 2427 2428
    int prefix = virNetDevIPRouteGetPrefix(routedef);
    unsigned int metric = virNetDevIPRouteGetMetric(routedef);
    virSocketAddrPtr addr = virNetDevIPRouteGetAddress(routedef);
    virSocketAddrPtr gateway = virNetDevIPRouteGetGateway(routedef);
2429 2430 2431 2432 2433

    if (prefix < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' has an invalid netmask "
                         "or IP address in route definition"),
2434
                       def->name);
2435 2436 2437
        return -1;
    }

2438
    if (virNetDevIPRouteAdd(def->bridge, addr, prefix, gateway, metric) < 0)
2439
        return -1;
2440

2441 2442 2443
    return 0;
}

2444
static int
2445
networkWaitDadFinish(virNetworkObjPtr obj)
2446
{
2447
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2448
    virNetworkIPDefPtr ipdef;
2449 2450 2451 2452
    virSocketAddrPtr *addrs = NULL, addr = NULL;
    size_t naddrs = 0;
    int ret = -1;

2453
    VIR_DEBUG("Begin waiting for IPv6 DAD on network %s", def->name);
2454

2455
    while ((ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, naddrs))) {
2456 2457 2458 2459 2460
        addr = &ipdef->address;
        if (VIR_APPEND_ELEMENT_COPY(addrs, naddrs, addr) < 0)
            goto cleanup;
    }

2461
    ret = (naddrs == 0) ? 0 : virNetDevIPWaitDadFinish(addrs, naddrs);
2462 2463 2464 2465

 cleanup:
    VIR_FREE(addrs);
    VIR_DEBUG("Finished waiting for IPv6 DAD on network %s with status %d",
2466
              def->name, ret);
2467 2468 2469
    return ret;
}

2470

2471
static int
2472
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
2473
                           virNetworkObjPtr obj)
2474
{
2475
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2476
    size_t i;
2477
    bool v4present = false, v6present = false;
2478
    virErrorPtr save_err = NULL;
2479
    virNetworkIPDefPtr ipdef;
2480
    virNetDevIPRoutePtr routedef;
2481
    char *macTapIfName = NULL;
2482
    virMacMapPtr macmap;
M
Michal Privoznik 已提交
2483
    char *macMapFile = NULL;
2484
    int tapfd = -1;
2485
    bool dnsmasqStarted = false;
2486
    bool devOnline = false;
2487
    bool firewalRulesAdded = false;
2488

2489
    /* Check to see if any network IP collides with an existing route */
2490
    if (networkCheckRouteCollision(def) < 0)
2491 2492
        return -1;

2493
    /* Create and configure the bridge device */
2494
    if (!def->bridge) {
2495 2496 2497 2498 2499 2500 2501 2502 2503 2504
        /* 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"),
2505
                       def->name);
2506 2507
        return -1;
    }
2508
    if (virNetDevBridgeCreate(def->bridge) < 0)
2509 2510
        return -1;

2511
    if (def->mac_specified) {
2512 2513 2514 2515 2516 2517
        /* 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.
         */
2518
        macTapIfName = networkBridgeDummyNicName(def->bridge);
2519
        if (!macTapIfName)
2520
            goto error;
2521
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2522 2523
        if (virNetDevTapCreateInBridgePort(def->bridge,
                                           &macTapIfName, &def->mac,
2524
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2525
                                           NULL, def->mtu, NULL,
2526 2527 2528
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2529
            VIR_FREE(macTapIfName);
2530
            goto error;
2531 2532 2533
        }
    }

2534
    if (!(macMapFile = virMacMapFileName(driver->dnsmasqStateDir,
2535
                                         def->bridge)) ||
2536
        !(macmap = virMacMapNew(macMapFile)))
2537
        goto error;
M
Michal Privoznik 已提交
2538

2539
    virNetworkObjSetMacMap(obj, macmap);
2540
    macmap = NULL;
2541

2542
    /* Set bridge options */
2543 2544 2545 2546

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2547
    if (virNetDevBridgeSetSTPDelay(def->bridge, def->delay * 1000) < 0)
2548
        goto error;
2549

2550
    if (virNetDevBridgeSetSTP(def->bridge, def->stp ? true : false) < 0)
2551
        goto error;
2552

2553 2554 2555
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
2556
    if (networkSetIPv6Sysctls(obj) < 0)
2557
        goto error;
2558

2559
    /* Add "once per network" rules */
2560 2561
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN &&
        networkAddFirewallRules(def) < 0)
2562
        goto error;
2563

2564 2565
    firewalRulesAdded = true;

2566
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i)); i++) {
2567
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2568
            v4present = true;
2569
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2570
            v6present = true;
2571

2572
        /* Add the IP address/netmask to the bridge */
2573
        if (networkAddAddrToBridge(obj, ipdef) < 0)
2574
            goto error;
2575 2576
    }

2577
    if (networkStartHandleMACTableManagerMode(obj, macTapIfName) < 0)
2578
        goto error;
2579

2580
    /* Bring up the bridge interface */
2581
    if (virNetDevSetOnline(def->bridge, 1) < 0)
2582
        goto error;
2583

2584 2585
    devOnline = true;

2586
    for (i = 0; i < def->nroutes; i++) {
2587 2588
        virSocketAddrPtr gateway = NULL;

2589
        routedef = def->routes[i];
2590
        gateway = virNetDevIPRouteGetGateway(routedef);
2591

2592 2593 2594
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2595
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2596
            if (networkAddRouteToBridge(obj, routedef) < 0) {
2597 2598 2599 2600 2601 2602
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2603
    /* If forward.type != NONE, turn on global IP forwarding */
2604
    if (def->forward.type != VIR_NETWORK_FORWARD_NONE) {
2605
        if (v6present && !virNetDevIPCheckIPv6Forwarding())
2606
            goto error; /* Precise error message already provided */
2607 2608 2609 2610

        if (networkEnableIPForwarding(v4present, v6present) < 0) {
            virReportSystemError(errno, "%s",
                                 _("failed to enable IP forwarding"));
2611
            goto error;
2612
        }
2613 2614
    }

2615

2616
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2617
    if ((v4present || v6present) &&
2618
        networkStartDhcpDaemon(driver, obj) < 0)
2619
        goto error;
2620

2621 2622
    dnsmasqStarted = true;

2623
    /* start radvd if there are any ipv6 addresses */
2624
    if (v6present && networkStartRadvd(driver, obj) < 0)
2625
        goto error;
2626

2627 2628 2629
    /* dnsmasq does not wait for DAD to complete before daemonizing,
     * so we need to wait for it ourselves.
     */
2630
    if (v6present && networkWaitDadFinish(obj) < 0)
2631
        goto error;
2632 2633 2634

    /* DAD has finished, dnsmasq is now bound to the
     * bridge's IPv6 address, so we can set the dummy tun down.
2635 2636 2637
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
2638
            goto error;
2639 2640 2641
        VIR_FORCE_CLOSE(tapfd);
    }

2642
    if (virNetDevBandwidthSet(def->bridge, def->bandwidth, true, true) < 0)
2643
        goto error;
2644

2645
    VIR_FREE(macTapIfName);
M
Michal Privoznik 已提交
2646
    VIR_FREE(macMapFile);
2647 2648 2649

    return 0;

2650
 error:
2651
    virErrorPreserveLast(&save_err);
2652 2653
    if (def->bandwidth)
       virNetDevBandwidthClear(def->bridge);
2654

2655 2656
    if (dnsmasqStarted) {
        pid_t dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
2657 2658
        kill(dnsmasqPid, SIGTERM);
        virNetworkObjSetDnsmasqPid(obj, -1);
2659 2660
    }

2661 2662
    if (devOnline)
        ignore_value(virNetDevSetOnline(def->bridge, 0));
2663

2664 2665
    if (firewalRulesAdded &&
        def->forward.type != VIR_NETWORK_FORWARD_OPEN)
2666
        networkRemoveFirewallRules(def);
2667

H
Hu Tao 已提交
2668
    if (macTapIfName) {
2669
        VIR_FORCE_CLOSE(tapfd);
2670
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2671 2672
        VIR_FREE(macTapIfName);
    }
2673
    virNetworkObjUnrefMacMap(obj);
M
Michal Privoznik 已提交
2674
    VIR_FREE(macMapFile);
2675

2676
    ignore_value(virNetDevBridgeDelete(def->bridge));
2677

2678
    virErrorRestore(&save_err);
2679
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2680 2681 2682
    return -1;
}

2683

2684 2685
static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
2686
                              virNetworkObjPtr obj)
2687
{
2688
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2689 2690 2691
    pid_t radvdPid;
    pid_t dnsmasqPid;

2692 2693
    if (def->bandwidth)
        virNetDevBandwidthClear(def->bridge);
2694

2695
    virNetworkObjUnrefMacMap(obj);
M
Michal Privoznik 已提交
2696

2697 2698
    radvdPid = virNetworkObjGetRadvdPid(obj);
    if (radvdPid > 0) {
2699 2700
        char *radvdpidbase;

2701
        kill(radvdPid, SIGTERM);
2702
        /* attempt to delete the pidfile we created */
2703
        if ((radvdpidbase = networkRadvdPidfileBasename(def->name))) {
2704
            virPidFileDelete(driver->pidDir, radvdpidbase);
2705 2706 2707 2708
            VIR_FREE(radvdpidbase);
        }
    }

2709 2710 2711
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid > 0)
        kill(dnsmasqPid, SIGTERM);
2712

2713 2714
    if (def->mac_specified) {
        char *macTapIfName = networkBridgeDummyNicName(def->bridge);
2715
        if (macTapIfName) {
2716
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2717 2718 2719 2720
            VIR_FREE(macTapIfName);
        }
    }

2721
    ignore_value(virNetDevSetOnline(def->bridge, 0));
2722

2723 2724
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(def);
2725

2726
    ignore_value(virNetDevBridgeDelete(def->bridge));
2727

2728
    /* See if its still alive and really really kill it */
2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739
    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);
2740

2741 2742 2743
    return 0;
}

2744

2745
static int
2746
networkStartNetworkBridge(virNetworkObjPtr obj)
2747 2748 2749 2750 2751
{
    /* 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.
     */
2752
    return networkStartHandleMACTableManagerMode(obj, NULL);
2753 2754
}

2755

2756
static int
2757
networkShutdownNetworkBridge(virNetworkObjPtr obj ATTRIBUTE_UNUSED)
2758 2759 2760 2761 2762 2763 2764 2765 2766
{
    /* 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;
}


2767 2768 2769 2770 2771 2772 2773 2774 2775
/* 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;
2776
    unsigned int maxVirtFns = 0;
2777 2778 2779 2780 2781 2782
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2783 2784 2785
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2786 2787
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev, &vfNames,
                                      &virtFns, &numVirtFns, &maxVirtFns)) < 0) {
2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802
        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];

2803
        switch ((virNetworkForwardType) netdef->forward.type) {
2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833
        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:
2834
        case VIR_NETWORK_FORWARD_OPEN:
2835 2836
            /* by definition these will never be encountered here */
            break;
2837 2838 2839 2840 2841

        case VIR_NETWORK_FORWARD_LAST:
        default:
            virReportEnumRangeError(virNetworkForwardType, netdef->forward.type);
            goto cleanup;
2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878
        }
    }

    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;
}


2879
static int
2880
networkStartNetworkExternal(virNetworkObjPtr obj)
2881 2882
{
    /* put anything here that needs to be done each time a network of
2883
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2884 2885 2886
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2887
    return networkCreateInterfacePool(virNetworkObjGetDef(obj));
2888 2889
}

2890 2891

static int
2892
networkShutdownNetworkExternal(virNetworkObjPtr obj ATTRIBUTE_UNUSED)
2893 2894
{
    /* put anything here that needs to be done each time a network of
2895
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2896 2897 2898 2899 2900 2901
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

2902

2903
static int
2904
networkStartNetwork(virNetworkDriverStatePtr driver,
2905
                    virNetworkObjPtr obj)
2906
{
2907
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2908 2909
    int ret = -1;

2910
    VIR_DEBUG("driver=%p, network=%p", driver, obj);
2911

2912
    if (virNetworkObjIsActive(obj)) {
2913 2914
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2915
        return ret;
2916 2917
    }

2918 2919
    VIR_DEBUG("Beginning network startup process");

2920 2921
    virNetworkObjDeleteAllPorts(obj, driver->stateDir);

2922
    VIR_DEBUG("Setting current network def as transient");
2923
    if (virNetworkObjSetDefTransient(obj, true, network_driver->xmlopt) < 0)
2924
        goto cleanup;
2925

2926 2927
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2928
    if (networkRunHook(obj, NULL,
2929 2930 2931 2932
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2933
    switch ((virNetworkForwardType) def->forward.type) {
2934 2935 2936 2937

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2938
    case VIR_NETWORK_FORWARD_OPEN:
2939
        if (networkStartNetworkVirtual(driver, obj) < 0)
2940
            goto cleanup;
2941 2942 2943
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2944
        if (def->bridge) {
2945
            if (networkStartNetworkBridge(obj) < 0)
2946 2947 2948 2949 2950 2951 2952
                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).
         */
2953 2954
        ATTRIBUTE_FALLTHROUGH;

2955 2956 2957
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2958
    case VIR_NETWORK_FORWARD_HOSTDEV:
2959
        if (networkStartNetworkExternal(obj) < 0)
2960
            goto cleanup;
2961
        break;
2962 2963 2964 2965 2966

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        goto cleanup;
2967 2968
    }

2969 2970
    virNetworkObjSetFloorSum(obj, 0);

2971
    /* finally we can call the 'started' hook script if any */
2972
    if (networkRunHook(obj, NULL,
2973 2974 2975 2976
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2977 2978 2979
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2980
    VIR_DEBUG("Writing network status to disk");
2981 2982
    if (virNetworkObjSaveStatus(driver->stateDir,
                                obj, network_driver->xmlopt) < 0)
2983
        goto cleanup;
2984

2985
    virNetworkObjSetActive(obj, true);
2986
    VIR_INFO("Network '%s' started up", def->name);
2987
    ret = 0;
2988

2989
 cleanup:
2990
    if (ret < 0) {
2991 2992 2993
        virErrorPtr save_err;

        virErrorPreserveLast(&save_err);
2994 2995
        virNetworkObjUnsetDefTransient(obj);
        networkShutdownNetwork(driver, obj);
2996
        virErrorRestore(&save_err);
2997 2998 2999 3000
    }
    return ret;
}

3001

3002 3003
static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
3004
                       virNetworkObjPtr obj)
3005
{
3006
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
3007 3008 3009
    int ret = 0;
    char *stateFile;

3010
    VIR_INFO("Shutting down network '%s'", def->name);
3011

3012
    if (!virNetworkObjIsActive(obj))
3013 3014
        return 0;

3015
    stateFile = virNetworkConfigFile(driver->stateDir, def->name);
3016 3017 3018 3019 3020 3021
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

3022
    switch ((virNetworkForwardType) def->forward.type) {
3023 3024 3025 3026

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
3027
    case VIR_NETWORK_FORWARD_OPEN:
3028
        ret = networkShutdownNetworkVirtual(driver, obj);
3029 3030 3031
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
3032
        if (def->bridge) {
3033
            ret = networkShutdownNetworkBridge(obj);
3034 3035 3036 3037 3038 3039
            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).
         */
3040 3041
        ATTRIBUTE_FALLTHROUGH;

3042 3043 3044
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
3045
    case VIR_NETWORK_FORWARD_HOSTDEV:
3046
        ret = networkShutdownNetworkExternal(obj);
3047
        break;
3048 3049 3050 3051 3052

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        return -1;
3053 3054
    }

3055
    /* now that we know it's stopped call the hook if present */
3056
    networkRunHook(obj, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
3057 3058
                   VIR_HOOK_SUBOP_END);

3059
    virNetworkObjSetActive(obj, false);
3060
    virNetworkObjUnsetDefTransient(obj);
3061
    return ret;
3062 3063 3064
}


3065 3066 3067
static virNetworkPtr
networkLookupByUUID(virConnectPtr conn,
                    const unsigned char *uuid)
3068
{
3069
    virNetworkDriverStatePtr driver = networkGetDriver();
3070
    virNetworkObjPtr obj;
3071
    virNetworkDefPtr def;
3072
    virNetworkPtr net = NULL;
3073

3074 3075
    obj = virNetworkObjFindByUUID(driver->networks, uuid);
    if (!obj) {
3076 3077
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
3078
        virReportError(VIR_ERR_NO_NETWORK,
3079 3080
                       _("no network with matching uuid '%s'"),
                       uuidstr);
3081
        goto cleanup;
3082
    }
3083
    def = virNetworkObjGetDef(obj);
3084

3085
    if (virNetworkLookupByUUIDEnsureACL(conn, def) < 0)
3086 3087
        goto cleanup;

3088
    net = virGetNetwork(conn, def->name, def->uuid);
3089

3090
 cleanup:
3091 3092
    virNetworkObjEndAPI(&obj);
    return net;
3093 3094
}

3095 3096 3097 3098

static virNetworkPtr
networkLookupByName(virConnectPtr conn,
                    const char *name)
3099
{
3100
    virNetworkDriverStatePtr driver = networkGetDriver();
3101
    virNetworkObjPtr obj;
3102
    virNetworkDefPtr def;
3103
    virNetworkPtr net = NULL;
3104

3105 3106
    obj = virNetworkObjFindByName(driver->networks, name);
    if (!obj) {
3107 3108
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
3109
        goto cleanup;
3110
    }
3111
    def = virNetworkObjGetDef(obj);
3112

3113
    if (virNetworkLookupByNameEnsureACL(conn, def) < 0)
3114 3115
        goto cleanup;

3116
    net = virGetNetwork(conn, def->name, def->uuid);
3117

3118
 cleanup:
3119 3120
    virNetworkObjEndAPI(&obj);
    return net;
3121 3122
}

3123 3124 3125

static int
networkConnectNumOfNetworks(virConnectPtr conn)
3126
{
3127
    virNetworkDriverStatePtr driver = networkGetDriver();
3128
    int nactive;
3129

3130 3131 3132
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

3133 3134 3135 3136
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
3137

3138 3139 3140
    return nactive;
}

3141 3142 3143 3144

static int
networkConnectListNetworks(virConnectPtr conn,
                           char **const names,
3145
                           int maxnames)
3146 3147
{
    virNetworkDriverStatePtr driver = networkGetDriver();
3148
    int got = 0;
3149

3150 3151 3152
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

3153
    got = virNetworkObjListGetNames(driver->networks,
3154
                                    true, names, maxnames,
3155 3156
                                    virConnectListNetworksCheckACL,
                                    conn);
3157

3158 3159 3160
    return got;
}

3161 3162 3163

static int
networkConnectNumOfDefinedNetworks(virConnectPtr conn)
3164
{
3165
    virNetworkDriverStatePtr driver = networkGetDriver();
3166
    int ninactive = 0;
3167

3168 3169 3170
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

3171 3172 3173 3174
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
3175

3176 3177 3178
    return ninactive;
}

3179 3180 3181 3182

static int
networkConnectListDefinedNetworks(virConnectPtr conn,
                                  char **const names,
3183
                                  int maxnames)
3184 3185
{
    virNetworkDriverStatePtr driver = networkGetDriver();
3186
    int got = 0;
3187

3188 3189 3190
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

3191
    got = virNetworkObjListGetNames(driver->networks,
3192
                                    false, names, maxnames,
3193 3194
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
3195 3196 3197
    return got;
}

3198

3199
static int
3200 3201 3202
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
3203
{
3204
    virNetworkDriverStatePtr driver = networkGetDriver();
3205 3206 3207 3208
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

3209 3210 3211
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

3212
    ret = virNetworkObjListExport(conn, driver->networks, nets,
3213 3214
                                  virConnectListAllNetworksCheckACL,
                                  flags);
3215

3216
 cleanup:
3217 3218
    return ret;
}
3219

3220

3221 3222 3223 3224 3225 3226 3227 3228
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
3229
    virNetworkDriverStatePtr driver = networkGetDriver();
3230 3231 3232 3233 3234 3235
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
3236
                                       net, eventID, callback,
3237 3238 3239
                                       opaque, freecb, &ret) < 0)
        ret = -1;

3240
 cleanup:
3241 3242 3243
    return ret;
}

3244

3245 3246 3247 3248
static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
3249
    virNetworkDriverStatePtr driver = networkGetDriver();
3250 3251 3252 3253 3254
    int ret = -1;

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

3255 3256
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
3257
                                        callbackID, true) < 0)
3258 3259 3260
        goto cleanup;

    ret = 0;
3261

3262
 cleanup:
3263 3264 3265
    return ret;
}

3266 3267 3268

static int
networkIsActive(virNetworkPtr net)
3269 3270 3271 3272
{
    virNetworkObjPtr obj;
    int ret = -1;

3273 3274
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3275

3276
    if (virNetworkIsActiveEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
3277 3278
        goto cleanup;

3279 3280
    ret = virNetworkObjIsActive(obj);

3281
 cleanup:
3282
    virNetworkObjEndAPI(&obj);
3283 3284 3285
    return ret;
}

3286 3287 3288

static int
networkIsPersistent(virNetworkPtr net)
3289 3290 3291 3292
{
    virNetworkObjPtr obj;
    int ret = -1;

3293 3294
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3295

3296
    if (virNetworkIsPersistentEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
3297 3298
        goto cleanup;

3299
    ret = virNetworkObjIsPersistent(obj);
3300

3301
 cleanup:
3302
    virNetworkObjEndAPI(&obj);
3303 3304 3305 3306
    return ret;
}


3307 3308
/*
 * networkFindUnusedBridgeName() - try to find a bridge name that is
3309 3310 3311
 * 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.
3312 3313 3314 3315 3316 3317 3318
 */
static int
networkFindUnusedBridgeName(virNetworkObjListPtr nets,
                            virNetworkDefPtr def)
{
    int ret = -1, id = 0;
    char *newname = NULL;
3319 3320 3321 3322 3323
    const char *templ = "virbr%d";
    const char *p;

    if (def->bridge &&
        (p = strchr(def->bridge, '%')) == strrchr(def->bridge, '%') &&
3324
        p && p[1] == 'd')
3325
        templ = def->bridge;
3326 3327 3328 3329

    do {
        if (virAsprintf(&newname, templ, id) < 0)
            goto cleanup;
3330 3331 3332 3333 3334
        /* 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).
         */
3335
        if (!(virNetworkObjBridgeInUse(nets, newname, def->name) ||
3336
              virNetDevExists(newname) == 1)) {
3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369
            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")) {
3370
        if (virNetworkObjBridgeInUse(nets, def->bridge, def->name)) {
3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387
            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;
}


3388
static int
3389
networkValidate(virNetworkDriverStatePtr driver,
3390
                virNetworkDefPtr def)
3391
{
3392
    size_t i, j;
3393 3394
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
3395
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
3396
    bool ipv4def = false, ipv6def = false;
3397
    bool bandwidthAllowed = false;
3398
    bool usesInterface = false, usesAddress = false;
3399

3400 3401 3402
    if (virXMLCheckIllegalChars("name", def->name, "\n") < 0)
        return -1;

3403 3404 3405
    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
3406 3407 3408 3409 3410
    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:
3411 3412 3413 3414
        /* 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)
3415 3416 3417
            return -1;

        virNetworkSetBridgeMacAddr(def);
3418
        bandwidthAllowed = true;
3419 3420 3421
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
3422 3423 3424 3425 3426
        if (def->bridge != NULL)
            bandwidthAllowed = true;

        ATTRIBUTE_FALLTHROUGH;

3427 3428 3429 3430
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
    case VIR_NETWORK_FORWARD_HOSTDEV:
3431
        /* They are also the only types that currently support setting
3432 3433
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
3434
         */
3435 3436 3437 3438 3439 3440 3441 3442
        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;
        }
3443
        if (virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0)) {
3444 3445 3446 3447
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3448
                           virNetworkForwardTypeToString(def->forward.type));
3449 3450
            return -1;
        }
3451
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
3452 3453 3454 3455
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3456
                           virNetworkForwardTypeToString(def->forward.type));
3457 3458 3459 3460 3461 3462 3463
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3464
                           virNetworkForwardTypeToString(def->forward.type));
3465 3466
            return -1;
        }
3467 3468 3469 3470 3471 3472
        break;

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, def->forward.type);
        return -1;
3473 3474
    }

3475 3476 3477 3478 3479 3480 3481 3482 3483 3484
    if (def->bandwidth &&
        !bandwidthAllowed) {
        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;
    }

3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497
    /* 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++) {
3498 3499 3500
        virNetworkForwardIfDefPtr iface = &def->forward.ifs[i];
        char *sysfs_path = NULL;

3501
        switch ((virNetworkForwardHostdevDeviceType)iface->type) {
3502 3503
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV:
            usesInterface = true;
3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514

            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;
            }
3515
            break;
3516 3517

        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI: {
3518
            usesAddress = true;
3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543

            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);
3544
            break;
3545 3546
        }

3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559
        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
3560 3561 3562
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
3563
    for (i = 0;
3564
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
3565
         i++) {
G
Gene Czarcinski 已提交
3566 3567 3568 3569 3570
        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 -- "
3571 3572
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589
                    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;
                }
3590 3591 3592
            }
        }
    }
3593 3594 3595 3596 3597 3598

    /* 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.
     */

3599 3600 3601
    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
3602 3603
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
3604
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH));
3605 3606

    vlanUsed = def->vlan.nTags > 0;
3607 3608
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3609 3610 3611 3612 3613
            /* 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.
             */
3614
            if (def->portGroups[i].virtPortProfile) {
3615
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3616
                    def->portGroups[i].virtPortProfile->virtPortType
3617 3618 3619 3620 3621 3622 3623
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3624
        }
3625
        if (def->portGroups[i].isDefault) {
3626 3627 3628 3629 3630
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3631
                               def->name, defaultPortGroup->name,
3632
                               def->portGroups[i].name);
3633
                return -1;
3634
            }
3635
            defaultPortGroup = &def->portGroups[i];
3636
        }
3637 3638 3639 3640 3641 3642 3643 3644 3645
        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;
            }
        }
3646 3647 3648 3649 3650 3651 3652 3653
        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;
        }
3654
    }
3655 3656 3657 3658 3659 3660 3661
    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.
         */
3662 3663 3664 3665 3666 3667
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681

    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;
            }
        }
    }
3682 3683 3684
    return 0;
}

3685 3686 3687 3688

static virNetworkPtr
networkCreateXML(virConnectPtr conn,
                 const char *xml)
3689
{
3690
    virNetworkDriverStatePtr driver = networkGetDriver();
3691
    virNetworkDefPtr newDef;
3692
    virNetworkObjPtr obj = NULL;
3693
    virNetworkDefPtr def;
3694
    virNetworkPtr net = NULL;
3695
    virObjectEventPtr event = NULL;
3696

3697
    if (!(newDef = virNetworkDefParseString(xml, network_driver->xmlopt)))
3698
        goto cleanup;
3699

3700
    if (virNetworkCreateXMLEnsureACL(conn, newDef) < 0)
3701 3702
        goto cleanup;

3703
    if (networkValidate(driver, newDef) < 0)
J
Ján Tomko 已提交
3704
        goto cleanup;
3705

3706 3707 3708
    /* 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.
3709
     */
3710
    if (!(obj = virNetworkObjAssignDef(driver->networks, newDef,
3711 3712
                                       VIR_NETWORK_OBJ_LIST_ADD_LIVE |
                                       VIR_NETWORK_OBJ_LIST_ADD_CHECK_LIVE)))
3713
        goto cleanup;
3714 3715
    newDef = NULL;
    def = virNetworkObjGetDef(obj);
3716

3717 3718
    if (networkStartNetwork(driver, obj) < 0) {
        virNetworkObjRemoveInactive(driver->networks, obj);
3719
        goto cleanup;
3720 3721
    }

3722 3723
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3724 3725
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3726

3727 3728
    VIR_INFO("Creating network '%s'", def->name);
    net = virGetNetwork(conn, def->name, def->uuid);
3729

3730
 cleanup:
3731
    virNetworkDefFree(newDef);
3732
    virObjectEventStateQueue(driver->networkEventState, event);
3733 3734
    virNetworkObjEndAPI(&obj);
    return net;
3735 3736
}

3737 3738 3739 3740

static virNetworkPtr
networkDefineXML(virConnectPtr conn,
                 const char *xml)
3741
{
3742
    virNetworkDriverStatePtr driver = networkGetDriver();
3743
    virNetworkDefPtr def = NULL;
3744
    bool freeDef = true;
3745 3746
    virNetworkObjPtr obj = NULL;
    virNetworkPtr net = NULL;
3747
    virObjectEventPtr event = NULL;
3748

3749
    if (!(def = virNetworkDefParseString(xml, network_driver->xmlopt)))
3750
        goto cleanup;
3751

3752 3753 3754
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3755
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3756
        goto cleanup;
3757

3758
    if (!(obj = virNetworkObjAssignDef(driver->networks, def, 0)))
J
Ján Tomko 已提交
3759
        goto cleanup;
3760

3761
    /* def was assigned to network object */
3762
    freeDef = false;
3763

3764 3765
    if (virNetworkSaveConfig(driver->networkConfigDir,
                             def, network_driver->xmlopt) < 0) {
3766 3767
        if (!virNetworkObjIsActive(obj)) {
            virNetworkObjRemoveInactive(driver->networks, obj);
3768 3769
            goto cleanup;
        }
3770 3771 3772 3773
        /* if network was active already, just undo new persistent
         * definition by making it transient.
         * XXX - this isn't necessarily the correct thing to do.
         */
3774
        virNetworkObjUpdateAssignDef(obj, NULL, false);
3775 3776 3777
        goto cleanup;
    }

3778
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3779 3780
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3781

3782
    VIR_INFO("Defining network '%s'", def->name);
3783
    net = virGetNetwork(conn, def->name, def->uuid);
3784

3785
 cleanup:
3786
    virObjectEventStateQueue(driver->networkEventState, event);
3787
    if (freeDef)
J
Ján Tomko 已提交
3788
        virNetworkDefFree(def);
3789 3790
    virNetworkObjEndAPI(&obj);
    return net;
3791 3792
}

3793

3794
static int
3795 3796
networkUndefine(virNetworkPtr net)
{
3797
    virNetworkDriverStatePtr driver = networkGetDriver();
3798
    virNetworkObjPtr obj;
3799
    virNetworkDefPtr def;
3800
    int ret = -1;
3801
    bool active = false;
3802
    virObjectEventPtr event = NULL;
3803

3804
    if (!(obj = networkObjFromNetwork(net)))
3805
        goto cleanup;
3806
    def = virNetworkObjGetDef(obj);
3807

3808
    if (virNetworkUndefineEnsureACL(net->conn, def) < 0)
3809 3810
        goto cleanup;

3811
    if (virNetworkObjIsActive(obj))
3812
        active = true;
3813

3814
    if (!virNetworkObjIsPersistent(obj)) {
3815 3816 3817 3818 3819
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                       _("can't undefine transient network"));
        goto cleanup;
    }

3820
    /* remove autostart link */
3821 3822
    if (virNetworkObjDeleteConfig(driver->networkConfigDir,
                                  driver->networkAutostartDir,
3823
                                  obj) < 0)
3824
        goto cleanup;
3825

3826 3827
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3828 3829
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3830

3831
    VIR_INFO("Undefining network '%s'", def->name);
3832
    if (!active) {
3833
        if (networkRemoveInactive(driver, obj) < 0)
3834
            goto cleanup;
3835 3836 3837 3838 3839
    } else {

        /* if the network still exists, it was active, and we need to make
         * it transient (by deleting the persistent def)
         */
3840
        virNetworkObjUpdateAssignDef(obj, NULL, false);
3841 3842
    }

3843
    ret = 0;
3844

3845
 cleanup:
3846
    virObjectEventStateQueue(driver->networkEventState, event);
3847
    virNetworkObjEndAPI(&obj);
3848
    return ret;
3849 3850
}

3851

3852 3853 3854 3855 3856 3857 3858 3859
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3860
    virNetworkDriverStatePtr driver = networkGetDriver();
3861
    virNetworkObjPtr obj = NULL;
3862
    virNetworkDefPtr def;
3863 3864
    int isActive, ret = -1;
    size_t i;
3865
    virNetworkIPDefPtr ipdef;
3866
    bool oldDhcpActive = false;
3867
    bool needFirewallRefresh = false;
3868

3869 3870 3871 3872 3873

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

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

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

3881
    /* see if we are listening for dhcp pre-modification */
3882
    for (i = 0;
3883
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
3884
         i++) {
3885 3886 3887 3888 3889 3890
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3891 3892
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3893
     */
3894
    isActive = virNetworkObjIsActive(obj);
3895 3896
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3897 3898 3899 3900 3901 3902 3903
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3904 3905 3906 3907
    if (isActive && (flags & VIR_NETWORK_UPDATE_AFFECT_LIVE)) {
        /* Take care of anything that must be done before updating the
         * live NetworkDef.
         */
3908 3909 3910 3911
        switch ((virNetworkForwardType) def->forward.type) {
        case VIR_NETWORK_FORWARD_NONE:
        case VIR_NETWORK_FORWARD_NAT:
        case VIR_NETWORK_FORWARD_ROUTE:
3912 3913 3914 3915 3916 3917 3918 3919 3920 3921
            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).
                 */
3922 3923
                networkRemoveFirewallRules(def);
                needFirewallRefresh = true;
3924 3925 3926 3927
                break;
            default:
                break;
            }
3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941
            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;
3942 3943 3944
        }
    }

3945
    /* update the network config in memory/on disk */
3946 3947 3948
    if (virNetworkObjUpdate(obj, command, section,
                            parentIndex, xml,
                            network_driver->xmlopt, flags) < 0) {
3949
        if (needFirewallRefresh)
3950
            ignore_value(networkAddFirewallRules(def));
3951 3952 3953
        goto cleanup;
    }

3954 3955 3956 3957
    /* @def is replaced */
    def = virNetworkObjGetDef(obj);

    if (needFirewallRefresh && networkAddFirewallRules(def) < 0)
3958 3959 3960 3961 3962
        goto cleanup;

    if (flags & VIR_NETWORK_UPDATE_AFFECT_CONFIG) {
        /* save updated persistent config to disk */
        if (virNetworkSaveConfig(driver->networkConfigDir,
3963 3964
                                 virNetworkObjGetPersistentDef(obj),
                                 network_driver->xmlopt) < 0) {
3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977
            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 ||
3978 3979 3980 3981 3982 3983 3984 3985
            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)
3986
             */
3987
            if (networkRestartDhcpDaemon(driver, obj) < 0)
3988 3989
                goto cleanup;

3990 3991 3992 3993 3994 3995 3996 3997
        } 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;

3998
            for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
3999
                 i++) {
4000 4001 4002 4003 4004 4005 4006
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
4007 4008
                 networkRestartDhcpDaemon(driver, obj) < 0) ||
                networkRefreshDhcpDaemon(driver, obj) < 0) {
4009 4010 4011
                goto cleanup;
            }

4012 4013 4014 4015
        } 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.
4016
             */
4017
            if (networkRefreshDhcpDaemon(driver, obj) < 0)
4018 4019 4020 4021 4022 4023 4024 4025
                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.
             */
4026
            if (networkRefreshRadvd(driver, obj) < 0)
4027 4028 4029 4030
                goto cleanup;
        }

        /* save current network state to disk */
4031 4032
        if ((ret = virNetworkObjSaveStatus(driver->stateDir,
                                           obj, network_driver->xmlopt)) < 0)
4033 4034
            goto cleanup;
    }
4035 4036

    /* call the 'updated' network hook script */
4037
    if (networkRunHook(obj, NULL, VIR_HOOK_NETWORK_OP_UPDATED,
4038 4039 4040
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

4041
    ret = 0;
4042
 cleanup:
4043
    virNetworkObjEndAPI(&obj);
4044 4045 4046
    return ret;
}

4047 4048 4049

static int
networkCreate(virNetworkPtr net)
4050
{
4051
    virNetworkDriverStatePtr driver = networkGetDriver();
4052
    virNetworkObjPtr obj;
4053
    virNetworkDefPtr def;
4054
    int ret = -1;
4055
    virObjectEventPtr event = NULL;
4056

4057
    if (!(obj = networkObjFromNetwork(net)))
4058
        goto cleanup;
4059
    def = virNetworkObjGetDef(obj);
4060

4061
    if (virNetworkCreateEnsureACL(net->conn, def) < 0)
4062 4063
        goto cleanup;

4064
    if ((ret = networkStartNetwork(driver, obj)) < 0)
4065
        goto cleanup;
4066

4067 4068
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
4069 4070
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
4071

4072
 cleanup:
4073
    virObjectEventStateQueue(driver->networkEventState, event);
4074
    virNetworkObjEndAPI(&obj);
4075
    return ret;
4076 4077
}

4078 4079 4080

static int
networkDestroy(virNetworkPtr net)
4081
{
4082
    virNetworkDriverStatePtr driver = networkGetDriver();
4083
    virNetworkObjPtr obj;
4084
    virNetworkDefPtr def;
4085
    int ret = -1;
4086
    virObjectEventPtr event = NULL;
4087

4088
    if (!(obj = networkObjFromNetwork(net)))
4089
        goto cleanup;
4090
    def = virNetworkObjGetDef(obj);
4091

4092
    if (virNetworkDestroyEnsureACL(net->conn, def) < 0)
4093 4094
        goto cleanup;

4095
    if (!virNetworkObjIsActive(obj)) {
4096
        virReportError(VIR_ERR_OPERATION_INVALID,
4097
                       _("network '%s' is not active"),
4098
                       def->name);
4099 4100 4101
        goto cleanup;
    }

4102
    if ((ret = networkShutdownNetwork(driver, obj)) < 0)
4103
        goto cleanup;
4104 4105 4106

    virNetworkObjDeleteAllPorts(obj, driver->stateDir);

4107 4108
    /* @def replaced in virNetworkObjUnsetDefTransient*/
    def = virNetworkObjGetDef(obj);
4109

4110 4111
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
4112 4113
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
4114

4115 4116
    if (!virNetworkObjIsPersistent(obj) &&
        networkRemoveInactive(driver, obj) < 0) {
4117 4118
        ret = -1;
        goto cleanup;
4119
    }
4120

4121
 cleanup:
4122
    virObjectEventStateQueue(driver->networkEventState, event);
4123
    virNetworkObjEndAPI(&obj);
4124 4125 4126
    return ret;
}

4127 4128 4129 4130

static char *
networkGetXMLDesc(virNetworkPtr net,
                  unsigned int flags)
4131
{
4132
    virNetworkObjPtr obj;
4133
    virNetworkDefPtr curDef;
4134
    virNetworkDefPtr def;
4135
    virNetworkDefPtr newDef;
4136
    char *ret = NULL;
4137

4138
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
4139

4140
    if (!(obj = networkObjFromNetwork(net)))
4141
        return ret;
4142 4143
    def = virNetworkObjGetDef(obj);
    newDef = virNetworkObjGetNewDef(obj);
4144

4145
    if (virNetworkGetXMLDescEnsureACL(net->conn, def) < 0)
4146 4147
        goto cleanup;

4148 4149
    if ((flags & VIR_NETWORK_XML_INACTIVE) && newDef)
        curDef = newDef;
4150
    else
4151
        curDef = def;
4152

4153
    ret = virNetworkDefFormat(curDef, network_driver->xmlopt, flags);
4154

4155
 cleanup:
4156
    virNetworkObjEndAPI(&obj);
4157
    return ret;
4158 4159
}

4160 4161 4162 4163

static char *
networkGetBridgeName(virNetworkPtr net)
{
4164
    virNetworkObjPtr obj;
4165
    virNetworkDefPtr def;
4166 4167
    char *bridge = NULL;

4168
    if (!(obj = networkObjFromNetwork(net)))
4169
        return bridge;
4170
    def = virNetworkObjGetDef(obj);
4171

4172
    if (virNetworkGetBridgeNameEnsureACL(net->conn, def) < 0)
4173 4174
        goto cleanup;

4175
    if (!(def->bridge)) {
4176 4177
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
4178
                       def->name);
4179 4180 4181
        goto cleanup;
    }

4182
    ignore_value(VIR_STRDUP(bridge, def->bridge));
4183

4184
 cleanup:
4185
    virNetworkObjEndAPI(&obj);
4186 4187 4188
    return bridge;
}

4189 4190 4191 4192

static int
networkGetAutostart(virNetworkPtr net,
                    int *autostart)
4193
{
4194
    virNetworkObjPtr obj;
4195
    int ret = -1;
4196

4197
    if (!(obj = networkObjFromNetwork(net)))
4198
        return ret;
4199

4200
    if (virNetworkGetAutostartEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
4201 4202
        goto cleanup;

4203
    *autostart = virNetworkObjIsAutostart(obj) ? 1 : 0;
4204
    ret = 0;
4205

4206
 cleanup:
4207
    virNetworkObjEndAPI(&obj);
4208
    return ret;
4209 4210
}

4211 4212 4213 4214

static int
networkSetAutostart(virNetworkPtr net,
                    int autostart)
4215
{
4216
    virNetworkDriverStatePtr driver = networkGetDriver();
4217
    virNetworkObjPtr obj;
4218
    virNetworkDefPtr def;
4219
    char *configFile = NULL, *autostartLink = NULL;
4220 4221
    bool new_autostart;
    bool cur_autostart;
4222
    int ret = -1;
4223

4224
    if (!(obj = networkObjFromNetwork(net)))
4225
        goto cleanup;
4226
    def = virNetworkObjGetDef(obj);
4227

4228
    if (virNetworkSetAutostartEnsureACL(net->conn, def) < 0)
4229 4230
        goto cleanup;

4231
    if (!virNetworkObjIsPersistent(obj)) {
4232 4233
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
4234 4235 4236
        goto cleanup;
    }

4237 4238 4239
    new_autostart = (autostart != 0);
    cur_autostart = virNetworkObjIsAutostart(obj);
    if (cur_autostart != new_autostart) {
4240 4241
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir,
                                               def->name)) == NULL)
4242
            goto cleanup;
4243 4244
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir,
                                                  def->name)) == NULL)
4245 4246
            goto cleanup;

4247
        if (new_autostart) {
4248
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
4249
                virReportSystemError(errno,
4250 4251
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
4252 4253
                goto cleanup;
            }
4254

4255
            if (symlink(configFile, autostartLink) < 0) {
4256
                virReportSystemError(errno,
4257
                                     _("Failed to create symlink '%s' to '%s'"),
4258
                                     autostartLink, configFile);
4259 4260 4261
                goto cleanup;
            }
        } else {
4262
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
4263
                virReportSystemError(errno,
4264
                                     _("Failed to delete symlink '%s'"),
4265
                                     autostartLink);
4266 4267
                goto cleanup;
            }
4268 4269
        }

4270
        virNetworkObjSetAutostart(obj, new_autostart);
4271
    }
4272

4273
    ret = 0;
4274

4275
 cleanup:
4276 4277
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
4278
    virNetworkObjEndAPI(&obj);
4279
    return ret;
4280 4281
}

4282

4283
static int
4284
networkGetDHCPLeases(virNetworkPtr net,
4285 4286 4287
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
4288
{
4289
    virNetworkDriverStatePtr driver = networkGetDriver();
4290 4291 4292
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
4293
    size_t size = 0;
4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304
    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;
4305
    virNetworkIPDefPtr ipdef_tmp = NULL;
4306 4307
    virNetworkDHCPLeasePtr lease = NULL;
    virNetworkDHCPLeasePtr *leases_ret = NULL;
4308
    virNetworkObjPtr obj;
4309
    virNetworkDefPtr def;
4310
    virMacAddr mac_addr;
4311 4312 4313

    virCheckFlags(0, -1);

4314 4315 4316 4317 4318 4319
    /* only to check if the MAC is valid */
    if (mac && virMacAddrParse(mac, &mac_addr) < 0) {
        virReportError(VIR_ERR_INVALID_MAC, "%s", mac);
        return -1;
    }

4320
    if (!(obj = networkObjFromNetwork(net)))
4321
        return -1;
4322
    def = virNetworkObjGetDef(obj);
4323

4324
    if (virNetworkGetDHCPLeasesEnsureACL(net->conn, def) < 0)
4325
        goto cleanup;
4326 4327

    /* Retrieve custom leases file location */
4328
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(driver, def->bridge);
4329 4330

    /* Read entire contents */
4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344
    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);
        }
4345 4346 4347 4348 4349 4350 4351 4352 4353 4354
        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;
        }

4355
        if (!virJSONValueIsArray(leases_array)) {
4356
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
4357
                           _("Malformed lease_entries array"));
4358 4359
            goto error;
        }
4360
        size = virJSONValueArraySize(leases_array);
4361 4362
    }

4363
    currtime = (long long)time(NULL);
4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379

    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;
        }

4380
        if (mac && virMacAddrCompare(mac, mac_tmp))
4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411
            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 */
4412 4413
            for (j = 0; j < def->nips; j++) {
                ipdef_tmp = &def->ips[j];
4414 4415 4416 4417 4418 4419 4420 4421

                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)) {
4422
                    lease->prefix = virSocketAddrGetIPPrefix(&ipdef_tmp->address,
4423 4424 4425 4426 4427 4428 4429 4430
                                                             &ipdef_tmp->netmask,
                                                             ipdef_tmp->prefix);
                    break;
                }
            }

            if ((VIR_STRDUP(lease->mac, mac_tmp) < 0) ||
                (VIR_STRDUP(lease->ipaddr, ip_tmp) < 0) ||
4431
                (VIR_STRDUP(lease->iface, def->bridge) < 0))
4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463
                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);
4464
    VIR_FREE(lease_entries);
4465 4466
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
4467

4468
    virNetworkObjEndAPI(&obj);
4469

4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480
    return rv;

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

4481

4482 4483 4484 4485 4486
/* A unified function to log network connections and disconnections */

static void
networkLogAllocation(virNetworkDefPtr netdef,
                     virNetworkForwardIfDefPtr dev,
4487
                     virMacAddrPtr mac,
4488 4489 4490 4491 4492
                     bool inUse)
{
    char macStr[VIR_MAC_STRING_BUFLEN];
    const char *verb = inUse ? "using" : "releasing";

4493
    virMacAddrFormat(mac, macStr);
4494 4495
    if (!dev) {
        VIR_INFO("MAC %s %s network %s (%d connections)",
4496
                 macStr, verb, netdef->name, netdef->connections);
4497
    } else {
4498
        if (dev->type == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI) {
4499 4500
            VIR_INFO("MAC %s %s network %s (%d connections) "
                     "physical device %04x:%02x:%02x.%x (%d connections)",
4501
                     macStr, verb, netdef->name, netdef->connections,
4502 4503 4504 4505 4506 4507
                     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)",
4508
                     macStr, verb, netdef->name, netdef->connections,
4509 4510 4511 4512 4513
                     dev->device.dev, dev->connections);
        }
    }
}

4514

4515 4516 4517 4518 4519 4520 4521 4522
/* 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.
 */

4523 4524 4525
/* networkAllocatePort:
 * @obj: the network to allocate from
 * @port: the port definition to allocate
4526
 *
4527
 * Looks up the network reference by port, allocates a physical
4528
 * device from that network (if appropriate), and returns with the
4529
 * port configuration filled in accordingly.
4530 4531 4532
 *
 * Returns 0 on success, -1 on failure.
 */
4533
static int
4534 4535
networkAllocatePort(virNetworkObjPtr obj,
                    virNetworkPortDefPtr port)
4536
{
4537
    virNetworkDriverStatePtr driver = networkGetDriver();
4538 4539
    virNetworkDefPtr netdef = NULL;
    virPortGroupDefPtr portgroup = NULL;
4540
    virNetworkForwardIfDefPtr dev = NULL;
4541
    size_t i;
4542
    int ret = -1;
4543
    virNetDevVPortProfilePtr portprofile = NULL;
4544

4545
    netdef = virNetworkObjGetDef(obj);
4546
    VIR_DEBUG("Allocating port from net %s", netdef->name);
4547

4548
    if (!virNetworkObjIsActive(obj)) {
4549 4550 4551
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
4552
        goto cleanup;
4553 4554
    }

4555
    VIR_DEBUG("Interface port group %s", port->group);
4556 4557 4558
    /* 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
4559
     */
4560
    portgroup = virPortGroupFindByName(netdef, port->group);
4561

4562 4563 4564 4565
    if (!port->bandwidth) {
        if (portgroup && portgroup->bandwidth &&
            virNetDevBandwidthCopy(&port->bandwidth,
                                   portgroup->bandwidth) < 0)
4566
            goto cleanup;
4567
    }
4568

4569 4570 4571 4572 4573 4574
    if (port->vlan.nTags == 0) {
        virNetDevVlanPtr vlan = NULL;
        if (portgroup && portgroup->vlan.nTags > 0)
            vlan = &portgroup->vlan;
        else if (netdef->vlan.nTags > 0)
            vlan = &netdef->vlan;
4575

4576
        if (vlan && virNetDevVlanCopy(&port->vlan, vlan) < 0)
4577
            goto cleanup;
4578
    }
4579

4580 4581 4582 4583 4584 4585
    if (!port->trustGuestRxFilters) {
        if (portgroup && portgroup->trustGuestRxFilters)
            port->trustGuestRxFilters = portgroup->trustGuestRxFilters;
        else if (netdef->trustGuestRxFilters)
            port->trustGuestRxFilters = netdef->trustGuestRxFilters;
    }
4586

4587 4588 4589
    /* merge virtualports from interface, network, and portgroup to
     * arrive at actual virtualport to use
     */
4590 4591
    if (virNetDevVPortProfileMerge3(&portprofile,
                                    port->virtPortProfile,
4592 4593 4594
                                    netdef->virtPortProfile,
                                    portgroup
                                    ? portgroup->virtPortProfile : NULL) < 0) {
4595
                goto cleanup;
4596 4597 4598 4599
    }
    if (portprofile) {
        VIR_FREE(port->virtPortProfile);
        port->virtPortProfile = portprofile;
4600 4601
    }

4602
    VIR_DEBUG("Processing forward type %d", netdef->forward.type);
4603 4604 4605 4606 4607
    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:
4608 4609 4610 4611
        /* for these forward types, the actual net type really *is*
         * NETWORK; we just keep the info from the portgroup in
         * iface->data.network.actual
         */
4612
        port->plugtype = VIR_NETWORK_PORT_PLUG_TYPE_NETWORK;
4613

4614
        if (VIR_STRDUP(port->plug.bridge.brname, netdef->bridge) < 0)
4615
            goto cleanup;
4616
        port->plug.bridge.macTableManager = netdef->macTableManager;
4617

4618
        if (port->virtPortProfile) {
4619 4620 4621
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("<virtualport type='%s'> not supported for network "
                             "'%s' which uses IP forwarding"),
4622
                           virNetDevVPortTypeToString(port->virtPortProfile->virtPortType),
4623
                           netdef->name);
4624
            goto cleanup;
4625 4626
        }

4627
        if (networkPlugBandwidth(obj, &port->mac, port->bandwidth, &port->class_id) < 0)
4628
            goto cleanup;
4629
        break;
4630

4631
    case VIR_NETWORK_FORWARD_HOSTDEV: {
4632
        port->plugtype = VIR_NETWORK_PORT_PLUG_TYPE_HOSTDEV_PCI;
4633

4634
        if (networkCreateInterfacePool(netdef) < 0)
4635
            goto cleanup;
4636 4637

        /* pick first dev with 0 connections */
4638 4639 4640
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
4641 4642 4643 4644 4645 4646 4647 4648
                break;
            }
        }
        if (!dev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' requires exclusive access "
                             "to interfaces, but none are available"),
                           netdef->name);
4649
            goto cleanup;
4650
        }
4651 4652 4653
        port->plug.hostdevpci.addr = dev->device.pci;
        port->plug.hostdevpci.driver = netdef->forward.driverName;
        port->plug.hostdevpci.managed = netdef->forward.managed;
4654

4655
        if (port->virtPortProfile) {
4656
            /* make sure type is supported for hostdev connections */
4657 4658
            if (port->virtPortProfile->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBG &&
                port->virtPortProfile->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBH) {
4659 4660 4661 4662
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("<virtualport type='%s'> not supported for network "
                                 "'%s' which uses an SR-IOV Virtual Function "
                                 "via PCI passthrough"),
4663
                               virNetDevVPortTypeToString(port->virtPortProfile->virtPortType),
4664
                               netdef->name);
4665
                goto cleanup;
4666 4667
            }
        }
4668 4669
        break;
    }
4670

4671 4672 4673 4674 4675 4676
    case VIR_NETWORK_FORWARD_BRIDGE:
        if (netdef->bridge) {
            /* <forward type='bridge'/> <bridge name='xxx'/>
             * is VIR_DOMAIN_NET_TYPE_BRIDGE
             */

4677 4678
            port->plugtype = VIR_NETWORK_PORT_PLUG_TYPE_BRIDGE;
            if (VIR_STRDUP(port->plug.bridge.brname, netdef->bridge) < 0)
4679
                goto cleanup;
4680
            port->plug.bridge.macTableManager = netdef->macTableManager;
4681

4682
            if (port->virtPortProfile) {
4683
                /* only type='openvswitch' is allowed for bridges */
4684
                if (port->virtPortProfile->virtPortType != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
4685 4686 4687
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                                   _("<virtualport type='%s'> not supported for network "
                                     "'%s' which uses a bridge device"),
4688
                                   virNetDevVPortTypeToString(port->virtPortProfile->virtPortType),
4689
                                   netdef->name);
4690
                    goto cleanup;
4691 4692
                }
            }
4693

4694
            if (networkPlugBandwidth(obj, &port->mac, port->bandwidth, &port->class_id) < 0)
4695
                goto cleanup;
4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706
            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:
4707 4708 4709 4710 4711
        /* <forward type='bridge|private|vepa|passthrough'> are all
         * VIR_DOMAIN_NET_TYPE_DIRECT.
         */

        /* Set type=direct and appropriate <source mode='xxx'/> */
4712
        port->plugtype = VIR_NETWORK_PORT_PLUG_TYPE_DIRECT;
4713 4714 4715 4716

        /* NO need to check the value returned from virNetDevMacVLanModeTypeFromString
         * it must be valid for these forward type(bridge|private|vepa|passthrough)
         */
4717
        port->plug.direct.mode =
4718
            virNetDevMacVLanModeTypeFromString(virNetworkForwardTypeToString(netdef->forward.type));
4719

4720
        if (port->virtPortProfile) {
4721
            /* make sure type is supported for macvtap connections */
4722 4723
            if (port->virtPortProfile->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBG &&
                port->virtPortProfile->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBH) {
4724 4725 4726
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("<virtualport type='%s'> not supported for network "
                                 "'%s' which uses a macvtap device"),
4727
                               virNetDevVPortTypeToString(port->virtPortProfile->virtPortType),
4728
                               netdef->name);
4729
                goto cleanup;
4730 4731
            }
        }
4732

4733 4734 4735
        /* 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).
         */
4736
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4737 4738 4739 4740
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4741
            goto cleanup;
4742 4743 4744
        } else {
            /* pick an interface from the pool */

4745
            if (networkCreateInterfacePool(netdef) < 0)
4746
                goto cleanup;
4747

4748 4749 4750 4751 4752
            /* 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.
4753
             */
4754 4755
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4756 4757
                 port->virtPortProfile &&
                 (port->virtPortProfile->virtPortType
4758
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4759

4760
                /* pick first dev with 0 connections */
4761 4762 4763
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4764 4765 4766 4767 4768
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4769
                dev = &netdef->forward.ifs[0];
4770 4771 4772
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4773 4774 4775 4776
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4777 4778 4779 4780
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4781
                goto cleanup;
4782
            }
4783
            if (VIR_STRDUP(port->plug.direct.linkdev,
4784
                           dev->device.dev) < 0)
4785
                goto cleanup;
4786
        }
4787 4788 4789 4790 4791
        break;

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, netdef->forward.type);
4792
        goto cleanup;
4793 4794
    }

4795
    if (virNetworkObjMacMgrAdd(obj, driver->dnsmasqStateDir,
4796 4797
                               port->ownername, &port->mac) < 0)
        goto cleanup;
M
Michal Privoznik 已提交
4798

4799
    if (virNetDevVPortProfileCheckComplete(port->virtPortProfile, true) < 0)
4800
        goto cleanup;
4801

4802 4803 4804 4805
    /* 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.
     */
4806
    VIR_DEBUG("Sanity check port config");
4807

4808
    if (port->vlan.nTags) {
4809 4810 4811 4812
        /* 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
4813
         */
4814 4815 4816
        if (!(port->plugtype == VIR_NETWORK_PORT_PLUG_TYPE_HOSTDEV_PCI ||
              (port->plugtype == VIR_NETWORK_PORT_PLUG_TYPE_DIRECT &&
               port->plug.direct.mode == VIR_NETDEV_MACVLAN_MODE_PASSTHRU) ||
4817
              (port->plugtype == VIR_NETWORK_PORT_PLUG_TYPE_BRIDGE &&
4818 4819
               port->virtPortProfile &&
               port->virtPortProfile->virtPortType == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH))) {
4820 4821 4822 4823 4824
            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);
4825
            goto cleanup;
4826 4827
        }
    }
4828 4829 4830 4831 4832 4833 4834 4835

    /* bandwidth configuration via libvirt is not supported for
     * hostdev network devices
     */
    if (port->bandwidth && port->plugtype == VIR_NETWORK_PORT_PLUG_TYPE_HOSTDEV_PCI) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("bandwidth settings are not supported "
                         "for hostdev interfaces"));
4836
        goto cleanup;
4837
    }
4838

4839 4840 4841 4842
    netdef->connections++;
    if (dev)
        dev->connections++;
    /* finally we can call the 'plugged' hook script if any */
4843 4844
    if (networkRunHook(obj, port,
                       VIR_HOOK_NETWORK_OP_PORT_CREATED,
4845 4846 4847
                       VIR_HOOK_SUBOP_BEGIN) < 0) {
        /* adjust for failure */
        netdef->connections--;
4848
        if (dev)
4849
            dev->connections--;
4850
        goto cleanup;
4851
    }
4852
    networkLogAllocation(netdef, dev, &port->mac, true);
4853 4854

    VIR_DEBUG("Port allocated");
4855

4856
    ret = 0;
4857
 cleanup:
4858
    return ret;
4859
}
4860

4861

4862 4863 4864
/* networkNotifyPort:
 * @obj: the network to notify
 * @port: the port definition to notify
4865 4866 4867 4868
 *
 * 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
4869
 * order.
4870
 */
4871
static int
4872 4873
networkNotifyPort(virNetworkObjPtr obj,
                  virNetworkPortDefPtr port)
4874 4875
{
    virNetworkDefPtr netdef;
4876
    virNetworkForwardIfDefPtr dev = NULL;
4877
    size_t i;
4878
    int ret = -1;
4879

4880
    netdef = virNetworkObjGetDef(obj);
4881

4882
    if (!virNetworkObjIsActive(obj)) {
4883 4884 4885
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
4886
        goto cleanup;
4887
    }
4888

4889 4890 4891 4892
    switch (port->plugtype) {
    case VIR_NETWORK_PORT_PLUG_TYPE_NONE:
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Unexpectedly got a network port without a plug"));
4893
        goto cleanup;
4894

4895 4896 4897 4898
    case VIR_NETWORK_PORT_PLUG_TYPE_NETWORK:
    case VIR_NETWORK_PORT_PLUG_TYPE_BRIDGE:
        /* see if we're connected to the correct bridge */
        if (!netdef->bridge) {
4899
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
4900
                           _("Unexpectedly got a network port without a network bridge"));
4901
            goto cleanup;
4902
        }
4903 4904 4905 4906
        break;

    case VIR_NETWORK_PORT_PLUG_TYPE_DIRECT:
        if (networkCreateInterfacePool(netdef) < 0)
4907
            goto cleanup;
4908 4909

        /* find the matching interface and increment its connections */
4910 4911
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4912
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4913 4914
                STREQ(port->plug.direct.linkdev,
                      netdef->forward.ifs[i].device.dev)) {
4915
                dev = &netdef->forward.ifs[i];
4916 4917 4918 4919 4920
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4921
            virReportError(VIR_ERR_INTERNAL_ERROR,
4922
                           _("network '%s' doesn't have dev='%s' "
4923 4924 4925
                             "in use by network port '%s'"),
                           netdef->name, port->plug.direct.linkdev,
                           port->uuid);
4926
            goto cleanup;
4927 4928
        }

4929
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4930 4931
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4932
         */
4933
        if ((dev->connections > 0) &&
4934 4935
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4936 4937
              port->virtPortProfile &&
              (port->virtPortProfile->virtPortType == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4938
            virReportError(VIR_ERR_INTERNAL_ERROR,
4939
                           _("network '%s' claims dev='%s' is already in "
4940 4941
                             "use by a different port"),
                           netdef->name, port->plug.direct.linkdev);
4942
            goto cleanup;
4943
        }
4944
        break;
4945

4946 4947 4948
    case VIR_NETWORK_PORT_PLUG_TYPE_HOSTDEV_PCI:

        if (networkCreateInterfacePool(netdef) < 0)
4949
            goto cleanup;
4950 4951

        /* find the matching interface and increment its connections */
4952 4953
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4954
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4955
                virPCIDeviceAddressEqual(&port->plug.hostdevpci.addr,
4956 4957
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
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 "
4965
                             "PCI device %04x:%02x:%02x.%x in use by network port"),
4966
                           netdef->name,
4967 4968 4969 4970
                           port->plug.hostdevpci.addr.domain,
                           port->plug.hostdevpci.addr.bus,
                           port->plug.hostdevpci.addr.slot,
                           port->plug.hostdevpci.addr.function);
4971
            goto cleanup;
4972 4973 4974 4975 4976 4977 4978
        }

        /* 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) &&
4979
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4980 4981 4982
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' claims the PCI device at "
                             "domain=%d bus=%d slot=%d function=%d "
4983
                             "is already in use by a different network port"),
4984 4985 4986
                           netdef->name,
                           dev->device.pci.domain, dev->device.pci.bus,
                           dev->device.pci.slot, dev->device.pci.function);
4987
            goto cleanup;
4988
        }
4989 4990 4991 4992 4993 4994

        break;

    case VIR_NETWORK_PORT_PLUG_TYPE_LAST:
    default:
        virReportEnumRangeError(virNetworkPortPlugType, port->plugtype);
4995
        goto cleanup;
4996 4997
    }

4998
    netdef->connections++;
4999 5000
    if (dev)
        dev->connections++;
5001
    /* finally we can call the 'plugged' hook script if any */
5002
    if (networkRunHook(obj, port, VIR_HOOK_NETWORK_OP_PORT_CREATED,
5003 5004 5005 5006 5007
                       VIR_HOOK_SUBOP_BEGIN) < 0) {
        /* adjust for failure */
        if (dev)
            dev->connections--;
        netdef->connections--;
5008
        goto cleanup;
5009
    }
5010 5011
    networkLogAllocation(netdef, dev, &port->mac, true);

5012
    ret = 0;
5013 5014 5015 5016 5017
 cleanup:
    return ret;
}


5018 5019 5020
/* networkReleasePort:
 * @obj: the network to release from
 * @port: the port definition to release
5021 5022 5023 5024 5025 5026 5027 5028
 *
 * 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.
 */
5029
static int
5030 5031
networkReleasePort(virNetworkObjPtr obj,
                   virNetworkPortDefPtr port)
5032
{
5033
    virNetworkDriverStatePtr driver = networkGetDriver();
5034
    virNetworkDefPtr netdef;
5035
    virNetworkForwardIfDefPtr dev = NULL;
5036 5037
    size_t i;
    int ret = -1;
5038

5039
    netdef = virNetworkObjGetDef(obj);
5040

5041 5042 5043
    switch ((virNetworkPortPlugType)port->plugtype) {
    case VIR_NETWORK_PORT_PLUG_TYPE_NONE:
        VIR_DEBUG("Releasing network device with no plug type");
5044 5045
        break;

5046 5047 5048 5049 5050
    case VIR_NETWORK_PORT_PLUG_TYPE_NETWORK:
    case VIR_NETWORK_PORT_PLUG_TYPE_BRIDGE:
        if (networkUnplugBandwidth(obj, port->bandwidth,
                                   &port->class_id) < 0)
            goto cleanup;
5051 5052
        break;

5053 5054 5055 5056 5057 5058 5059
    case VIR_NETWORK_PORT_PLUG_TYPE_DIRECT:
        if (netdef->forward.nifs == 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
            goto cleanup;
5060
        }
5061

5062 5063
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
5064
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
5065
                STREQ(port->plug.direct.linkdev, netdef->forward.ifs[i].device.dev)) {
5066
                dev = &netdef->forward.ifs[i];
5067 5068 5069
                break;
            }
        }
5070

5071
        if (!dev) {
5072
            virReportError(VIR_ERR_INTERNAL_ERROR,
5073 5074
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
5075 5076
                           netdef->name, port->plug.direct.linkdev);
            goto cleanup;
5077
        }
5078
        break;
5079

5080 5081
    case VIR_NETWORK_PORT_PLUG_TYPE_HOSTDEV_PCI:
        if (netdef->forward.nifs == 0) {
5082
            virReportError(VIR_ERR_INTERNAL_ERROR,
5083 5084 5085 5086
                           _("network '%s' uses a hostdev mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
            goto cleanup;
5087 5088
        }

5089 5090
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
5091
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
5092
                virPCIDeviceAddressEqual(&port->plug.hostdevpci.addr,
5093 5094
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
5095 5096 5097 5098 5099 5100 5101 5102 5103
                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,
5104 5105 5106 5107 5108
                           port->plug.hostdevpci.addr.domain,
                           port->plug.hostdevpci.addr.bus,
                           port->plug.hostdevpci.addr.slot,
                           port->plug.hostdevpci.addr.function);
            goto cleanup;
5109
        }
5110 5111 5112 5113 5114 5115
        break;

    case VIR_NETWORK_PORT_PLUG_TYPE_LAST:
    default:
        virReportEnumRangeError(virNetworkPortPlugType, port->plugtype);
        goto cleanup;
J
Ján Tomko 已提交
5116
    }
5117

5118
    virNetworkObjMacMgrDel(obj, driver->dnsmasqStateDir, port->ownername, &port->mac);
5119 5120 5121 5122 5123

    netdef->connections--;
    if (dev)
        dev->connections--;
    /* finally we can call the 'unplugged' hook script if any */
5124
    networkRunHook(obj, port, VIR_HOOK_NETWORK_OP_PORT_DELETED,
5125 5126
                   VIR_HOOK_SUBOP_BEGIN);
    networkLogAllocation(netdef, dev, &port->mac, false);
M
Michal Privoznik 已提交
5127

5128 5129 5130 5131 5132 5133
    ret = 0;
 cleanup:
    return ret;
}


5134 5135 5136
/**
 * networkCheckBandwidth:
 * @net: network QoS
5137
 * @ifaceBand: interface QoS (may be NULL if no QoS)
5138
 * @oldBandwidth: new interface QoS (may be NULL if no QoS)
5139
 * @ifaceMac: interface MAC (used in error messages for identification)
5140 5141
 * @new_rate: new rate for non guaranteed class
 *
5142 5143 5144 5145 5146 5147 5148 5149
 * 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.
 *
5150 5151 5152 5153 5154
 * 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
5155
networkCheckBandwidth(virNetworkObjPtr obj,
5156
                      virNetDevBandwidthPtr ifaceBand,
5157
                      virNetDevBandwidthPtr oldBandwidth,
5158
                      virMacAddrPtr ifaceMac,
5159 5160 5161
                      unsigned long long *new_rate)
{
    int ret = -1;
5162 5163
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
    virNetDevBandwidthPtr netBand = def->bandwidth;
5164
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5165 5166 5167
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

5168
    virMacAddrFormat(ifaceMac, ifmac);
5169 5170 5171 5172 5173 5174

    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"),
5175
                       ifmac, def->name);
5176 5177 5178
        return -1;
    }

5179 5180
    if (((!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor) &&
         (!oldBandwidth || !oldBandwidth->in || !oldBandwidth->in->floor)) ||
5181 5182
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
5183
        return 1;
5184
    }
5185 5186

    tmp_new_rate = netBand->in->average;
5187 5188 5189 5190
    if (oldBandwidth && oldBandwidth->in)
        tmp_floor_sum -= oldBandwidth->in->floor;
    if (ifaceBand && ifaceBand->in)
        tmp_floor_sum += ifaceBand->in->floor;
5191 5192 5193 5194 5195 5196

    /* 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,
5197 5198 5199
                           _("Cannot plug '%s' interface into '%s' because "
                             "new combined inbound floor=%llu would overcommit "
                             "peak=%llu on network '%s'"),
5200
                           ifmac,
5201
                           def->bridge,
5202 5203
                           tmp_floor_sum,
                           netBand->in->peak,
5204
                           def->name);
5205 5206 5207 5208 5209 5210
            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,
5211 5212 5213
                       _("Cannot plug '%s' interface into '%s' because "
                         "new combined inbound floor=%llu would overcommit "
                         "average=%llu on network '%s'"),
5214
                       ifmac,
5215
                       def->bridge,
5216 5217
                       tmp_floor_sum,
                       netBand->in->average,
5218
                       def->name);
5219 5220 5221
        goto cleanup;
    }

5222 5223
    if (new_rate)
        *new_rate = tmp_new_rate;
5224 5225
    ret = 0;

5226
 cleanup:
5227 5228 5229
    return ret;
}

5230

5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241
/**
 * 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
5242
networkNextClassID(virNetworkObjPtr obj)
5243
{
5244
    ssize_t ret = 0;
5245
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5246

5247 5248
    if ((ret = virBitmapNextClearBit(classIdMap, -1)) < 0)
        ret = virBitmapSize(classIdMap);
5249

5250
    if (virBitmapSetBitExpand(classIdMap, ret) < 0)
5251 5252 5253 5254 5255
        return -1;

    return ret;
}

5256

5257
static int
5258
networkPlugBandwidthImpl(virNetworkObjPtr obj,
5259
                         virMacAddrPtr mac,
5260
                         virNetDevBandwidthPtr ifaceBand,
5261
                         unsigned int *class_id,
5262
                         unsigned long long new_rate)
5263
{
5264
    virNetworkDriverStatePtr driver = networkGetDriver();
5265
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
5266
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5267
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5268
    ssize_t next_id = 0;
5269 5270
    int plug_ret;
    int ret = -1;
5271 5272

    /* generate new class_id */
5273
    if ((next_id = networkNextClassID(obj)) < 0) {
5274 5275 5276 5277 5278
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Could not generate next class ID"));
        goto cleanup;
    }

5279
    plug_ret = virNetDevBandwidthPlug(def->bridge, def->bandwidth,
5280
                                      mac, ifaceBand, next_id);
5281
    if (plug_ret < 0) {
5282
        ignore_value(virNetDevBandwidthUnplug(def->bridge, next_id));
5283 5284 5285 5286
        goto cleanup;
    }

    /* QoS was set, generate new class ID */
5287
    *class_id = next_id;
5288
    /* update sum of 'floor'-s of attached NICs */
5289 5290
    tmp_floor_sum += ifaceBand->in->floor;
    virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5291
    /* update status file */
5292
    if (virNetworkObjSaveStatus(driver->stateDir, obj, network_driver->xmlopt) < 0) {
5293
        ignore_value(virBitmapClearBit(classIdMap, next_id));
5294 5295
        tmp_floor_sum -= ifaceBand->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5296 5297
        *class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(def->bridge, next_id));
5298 5299
        goto cleanup;
    }
5300
    /* update rate for non guaranteed NICs */
5301
    new_rate -= tmp_floor_sum;
5302 5303
    if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                     def->bandwidth, new_rate) < 0)
5304
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
5305
                 def->bridge);
5306 5307

    ret = 0;
5308 5309 5310 5311 5312 5313
 cleanup:
    return ret;
}


static int
5314
networkPlugBandwidth(virNetworkObjPtr obj,
5315 5316 5317
                     virMacAddrPtr mac,
                     virNetDevBandwidthPtr ifaceBand,
                     unsigned int *class_id)
5318 5319 5320 5321 5322 5323
{
    int ret = -1;
    int plug_ret;
    unsigned long long new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

5324
    if ((plug_ret = networkCheckBandwidth(obj, ifaceBand, NULL,
5325
                                          mac, &new_rate)) < 0) {
5326 5327 5328 5329 5330 5331 5332 5333 5334 5335
        /* helper reported error */
        goto cleanup;
    }

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

5336
    virMacAddrFormat(mac, ifmac);
5337

5338
    if (networkPlugBandwidthImpl(obj, mac, ifaceBand, class_id, new_rate) < 0)
5339 5340 5341
        goto cleanup;

    ret = 0;
5342

5343
 cleanup:
5344 5345 5346
    return ret;
}

5347

5348
static int
5349
networkUnplugBandwidth(virNetworkObjPtr obj,
5350 5351
                       virNetDevBandwidthPtr ifaceBand,
                       unsigned int *class_id)
5352
{
5353
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
5354
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5355
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5356
    virNetworkDriverStatePtr driver = networkGetDriver();
5357 5358 5359
    int ret = 0;
    unsigned long long new_rate;

5360
    if (class_id && *class_id) {
5361
        if (!def->bandwidth || !def->bandwidth->in) {
5362
            VIR_WARN("Network %s has no bandwidth but unplug requested",
5363
                     def->name);
5364 5365
            goto cleanup;
        }
5366
        /* we must remove class from bridge */
5367
        new_rate = def->bandwidth->in->average;
5368

5369 5370
        if (def->bandwidth->in->peak > 0)
            new_rate = def->bandwidth->in->peak;
5371

5372
        ret = virNetDevBandwidthUnplug(def->bridge, *class_id);
5373 5374 5375
        if (ret < 0)
            goto cleanup;
        /* update sum of 'floor'-s of attached NICs */
5376 5377 5378
        tmp_floor_sum -= ifaceBand->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);

5379
        /* return class ID */
5380
        ignore_value(virBitmapClearBit(classIdMap, *class_id));
5381
        /* update status file */
5382 5383
        if (virNetworkObjSaveStatus(driver->stateDir,
                                    obj, network_driver->xmlopt) < 0) {
5384 5385
            tmp_floor_sum += ifaceBand->in->floor;
            virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5386
            ignore_value(virBitmapSetBit(classIdMap, *class_id));
5387 5388
            goto cleanup;
        }
5389
        /* update rate for non guaranteed NICs */
5390
        new_rate -= tmp_floor_sum;
5391 5392
        if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                         def->bandwidth, new_rate) < 0)
5393
            VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
5394
                     def->bridge);
5395
        /* no class is associated any longer */
5396
        *class_id = 0;
5397 5398
    }

5399
 cleanup:
5400 5401
    return ret;
}
5402

5403

5404
static void
5405
networkNetworkObjTaint(virNetworkObjPtr obj,
5406
                       virNetworkTaintFlags taint)
5407
{
5408 5409
    virNetworkDefPtr def = virNetworkObjGetDef(obj);

5410
    if (virNetworkObjTaint(obj, taint)) {
5411
        char uuidstr[VIR_UUID_STRING_BUFLEN];
5412
        virUUIDFormat(def->uuid, uuidstr);
5413 5414

        VIR_WARN("Network name='%s' uuid=%s is tainted: %s",
5415
                 def->name, uuidstr, virNetworkTaintTypeToString(taint));
5416 5417
    }
}
5418 5419


5420
static int
5421 5422 5423 5424 5425
networkUpdatePortBandwidth(virNetworkObjPtr obj,
                           virMacAddrPtr mac,
                           unsigned int *class_id,
                           virNetDevBandwidthPtr oldBandwidth,
                           virNetDevBandwidthPtr newBandwidth)
5426 5427
{
    virNetworkDriverStatePtr driver = networkGetDriver();
5428
    virNetworkDefPtr def;
5429
    unsigned long long tmp_floor_sum;
5430
    unsigned long long new_rate = 0;
5431
    unsigned long long old_floor, new_floor;
5432
    int plug_ret;
5433 5434 5435

    old_floor = new_floor = 0;

5436 5437
    if (oldBandwidth && oldBandwidth->in)
        old_floor = oldBandwidth->in->floor;
5438 5439 5440 5441
    if (newBandwidth && newBandwidth->in)
        new_floor = newBandwidth->in->floor;

    if (new_floor == old_floor)
5442 5443
        return 0;

5444
    def = virNetworkObjGetDef(obj);
5445

5446 5447
    if ((plug_ret = networkCheckBandwidth(obj, newBandwidth, oldBandwidth,
                                          mac, &new_rate)) < 0) {
5448
        /* helper reported error */
5449
        return -1;
5450 5451 5452 5453
    }

    if (plug_ret > 0) {
        /* no QoS needs to be set; claim success */
5454
        return 0;
5455 5456 5457 5458
    }

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

5459
    if (oldBandwidth && oldBandwidth->in && oldBandwidth->in->floor &&
5460
        newBandwidth->in && newBandwidth->in->floor) {
5461 5462
        /* Either we just need to update @floor .. */

5463
        if (virNetDevBandwidthUpdateRate(def->bridge,
5464
                                         *class_id,
5465
                                         def->bandwidth,
5466
                                         newBandwidth->in->floor) < 0)
5467
            return -1;
5468

5469
        tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5470
        tmp_floor_sum -= oldBandwidth->in->floor;
5471 5472 5473
        tmp_floor_sum += newBandwidth->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);
        new_rate -= tmp_floor_sum;
5474

5475 5476
        if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                         def->bandwidth, new_rate) < 0 ||
5477 5478
            virNetworkObjSaveStatus(driver->stateDir,
                                    obj, network_driver->xmlopt) < 0) {
5479
            /* Ouch, rollback */
5480
            tmp_floor_sum -= newBandwidth->in->floor;
5481
            tmp_floor_sum += oldBandwidth->in->floor;
5482
            virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5483

5484
            ignore_value(virNetDevBandwidthUpdateRate(def->bridge,
5485
                                                      *class_id,
5486
                                                      def->bandwidth,
5487 5488
                                                      oldBandwidth->in->floor));
            return -1;
5489 5490 5491 5492
        }
    } else if (newBandwidth->in && newBandwidth->in->floor) {
        /* .. or we need to plug in new .. */

5493 5494
        if (networkPlugBandwidthImpl(obj, mac, newBandwidth,
                                     class_id,
5495
                                     new_rate) < 0)
5496
            return -1;
5497 5498 5499
    } else {
        /* .. or unplug old. */

5500 5501
        if (networkUnplugBandwidth(obj, oldBandwidth, class_id) < 0)
            return -1;
5502 5503
    }

5504 5505 5506 5507
    return 0;
}


5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594
static virNetworkPortPtr
networkPortLookupByUUID(virNetworkPtr net,
                        const unsigned char *uuid)
{
    virNetworkObjPtr obj;
    virNetworkDefPtr def;
    virNetworkPortDefPtr portdef = NULL;
    virNetworkPortPtr ret = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    virUUIDFormat(uuid, uuidstr);

    if (!(obj = networkObjFromNetwork(net)))
        return ret;

    def = virNetworkObjGetDef(obj);

    if (!(portdef = virNetworkObjLookupPort(obj, uuid)))
        goto cleanup;

    if (virNetworkPortLookupByUUIDEnsureACL(net->conn, def, portdef) < 0)
        goto cleanup;

    if (!virNetworkObjIsActive(obj)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       def->name);
        goto cleanup;
    }

    ret = virGetNetworkPort(net, uuid);

 cleanup:
    virNetworkObjEndAPI(&obj);
    return ret;
}


static virNetworkPortPtr
networkPortCreateXML(virNetworkPtr net,
                     const char *xmldesc,
                     unsigned int flags)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
    virNetworkObjPtr obj;
    virNetworkDefPtr def;
    virNetworkPortDefPtr portdef = NULL;
    virNetworkPortPtr ret = NULL;
    int rc;

    virCheckFlags(VIR_NETWORK_PORT_CREATE_RECLAIM, NULL);

    if (!(obj = networkObjFromNetwork(net)))
        return ret;

    def = virNetworkObjGetDef(obj);

    if (!(portdef = virNetworkPortDefParseString(xmldesc)))
        goto cleanup;

    if (virNetworkPortCreateXMLEnsureACL(net->conn, def, portdef) < 0)
        goto cleanup;

    if (!virNetworkObjIsActive(obj)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       def->name);
        goto cleanup;
    }

    if (portdef->plugtype == VIR_NETWORK_PORT_PLUG_TYPE_NONE) {
        if (flags & VIR_NETWORK_PORT_CREATE_RECLAIM) {
            virReportError(VIR_ERR_INVALID_ARG, "%s",
                           _("Port reclaim requested but plug type is none"));
            goto cleanup;
        }
    } else {
        if (!(flags & VIR_NETWORK_PORT_CREATE_RECLAIM)) {
            virReportError(VIR_ERR_INVALID_ARG, "%s",
                           _("Port reclaim not requested but plug type is not none"));
            goto cleanup;
        }
    }

    if (flags & VIR_NETWORK_PORT_CREATE_RECLAIM)
        rc = networkNotifyPort(obj, portdef);
    else
        rc = networkAllocatePort(obj, portdef);
5595 5596 5597 5598
    if (rc < 0)
        goto cleanup;

    if (virNetworkObjAddPort(obj, portdef, driver->stateDir) < 0) {
5599
        virErrorPtr save_err;
5600

5601
        virErrorPreserveLast(&save_err);
5602
        ignore_value(networkReleasePort(obj, portdef));
5603
        virNetworkPortDefFree(portdef);
5604 5605
        virErrorRestore(&save_err);

5606 5607 5608 5609 5610 5611 5612 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 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656 5657 5658 5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796 5797 5798 5799 5800 5801 5802 5803 5804 5805 5806 5807 5808 5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828 5829 5830 5831 5832 5833 5834 5835 5836 5837 5838 5839 5840 5841 5842 5843 5844 5845 5846 5847 5848 5849 5850 5851 5852 5853 5854 5855 5856 5857 5858 5859 5860 5861 5862 5863 5864 5865 5866 5867 5868 5869 5870 5871 5872 5873 5874 5875 5876 5877 5878 5879 5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895
        goto cleanup;
    }

    ret = virGetNetworkPort(net, portdef->uuid);
 cleanup:
    virNetworkObjEndAPI(&obj);
    return ret;
}


static char *
networkPortGetXMLDesc(virNetworkPortPtr port,
                      unsigned int flags)
{
    virNetworkObjPtr obj;
    virNetworkDefPtr def;
    virNetworkPortDefPtr portdef = NULL;
    char *ret = NULL;

    virCheckFlags(0, NULL);

    if (!(obj = networkObjFromNetwork(port->net)))
        return ret;

    def = virNetworkObjGetDef(obj);

    if (!(portdef = virNetworkObjLookupPort(obj, port->uuid)))
        goto cleanup;

    if (virNetworkPortGetXMLDescEnsureACL(port->net->conn, def, portdef) < 0)
        goto cleanup;

    if (!virNetworkObjIsActive(obj)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       def->name);
        goto cleanup;
    }

   if (!(ret = virNetworkPortDefFormat(portdef)))
       goto cleanup;

 cleanup:
    virNetworkObjEndAPI(&obj);
    return ret;
}


static int
networkPortDelete(virNetworkPortPtr port,
                  unsigned int flags)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
    virNetworkObjPtr obj;
    virNetworkDefPtr def;
    virNetworkPortDefPtr portdef;
    int ret = -1;

    virCheckFlags(0, -1);

    if (!(obj = networkObjFromNetwork(port->net)))
        return ret;

    def = virNetworkObjGetDef(obj);

    if (!(portdef = virNetworkObjLookupPort(obj, port->uuid)))
        goto cleanup;

    if (virNetworkPortDeleteEnsureACL(port->net->conn, def, portdef) < 0)
        goto cleanup;

    if (!virNetworkObjIsActive(obj)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       def->name);
        goto cleanup;
    }

    if (networkReleasePort(obj, portdef) < 0)
        goto cleanup;

    virNetworkObjDeletePort(obj, port->uuid, driver->stateDir);

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


static int
networkPortSetParameters(virNetworkPortPtr port,
                         virTypedParameterPtr params,
                         int nparams,
                         unsigned int flags)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
    virNetworkObjPtr obj;
    virNetworkDefPtr def;
    virNetworkPortDefPtr portdef;
    virNetDevBandwidthPtr bandwidth = NULL;
    char *dir = NULL;
    int ret = -1;
    size_t i;

    virCheckFlags(0, -1);

    if (!(obj = networkObjFromNetwork(port->net)))
        return ret;

    def = virNetworkObjGetDef(obj);

    if (!(portdef = virNetworkObjLookupPort(obj, port->uuid)))
        goto cleanup;

    if (virNetworkPortSetParametersEnsureACL(port->net->conn, def, portdef) < 0)
        goto cleanup;

    if (!virNetworkObjIsActive(obj)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       def->name);
        goto cleanup;
    }

    if (!(dir = virNetworkObjGetPortStatusDir(obj, driver->stateDir)))
        goto cleanup;

    if ((VIR_ALLOC(bandwidth) < 0) ||
        (VIR_ALLOC(bandwidth->in) < 0) ||
        (VIR_ALLOC(bandwidth->out) < 0))
        goto cleanup;

    for (i = 0; i < nparams; i++) {
        virTypedParameterPtr param = &params[i];

        if (STREQ(param->field, VIR_NETWORK_PORT_BANDWIDTH_IN_AVERAGE)) {
            bandwidth->in->average = param->value.ui;
        } else if (STREQ(param->field, VIR_NETWORK_PORT_BANDWIDTH_IN_PEAK)) {
            bandwidth->in->peak = param->value.ui;
        } else if (STREQ(param->field, VIR_NETWORK_PORT_BANDWIDTH_IN_BURST)) {
            bandwidth->in->burst = param->value.ui;
        } else if (STREQ(param->field, VIR_NETWORK_PORT_BANDWIDTH_IN_FLOOR)) {
            bandwidth->in->floor = param->value.ui;
        } else if (STREQ(param->field, VIR_NETWORK_PORT_BANDWIDTH_OUT_AVERAGE)) {
            bandwidth->out->average = param->value.ui;
        } else if (STREQ(param->field, VIR_NETWORK_PORT_BANDWIDTH_OUT_PEAK)) {
            bandwidth->out->peak = param->value.ui;
        } else if (STREQ(param->field, VIR_NETWORK_PORT_BANDWIDTH_OUT_BURST)) {
            bandwidth->out->burst = param->value.ui;
        }
    }

    /* average or floor are mandatory, peak and burst are optional.
     * So if no average or floor is given, we free inbound/outbound
     * here which causes inbound/outbound to not be set. */
    if (!bandwidth->in->average && !bandwidth->in->floor)
        VIR_FREE(bandwidth->in);
    if (!bandwidth->out->average)
        VIR_FREE(bandwidth->out);

    if (networkUpdatePortBandwidth(obj,
                                   &portdef->mac,
                                   &portdef->class_id,
                                   portdef->bandwidth,
                                   bandwidth) < 0)
        goto cleanup;

    virNetDevBandwidthFree(portdef->bandwidth);
    portdef->bandwidth = bandwidth;
    bandwidth = NULL;

    if (virNetworkPortDefSaveStatus(portdef, dir) < 0)
        goto cleanup;

    ret = 0;
 cleanup:
    virNetDevBandwidthFree(bandwidth);
    virNetworkObjEndAPI(&obj);
    VIR_FREE(dir);
    return ret;
}


static int
networkPortGetParameters(virNetworkPortPtr port,
                         virTypedParameterPtr *params,
                         int *nparams,
                         unsigned int flags)
{
    virNetworkObjPtr obj;
    virNetworkDefPtr def;
    virNetworkPortDefPtr portdef;
    int maxparams = 0;
    int ret = -1;

    virCheckFlags(0, -1);

    *params = NULL;
    *nparams = 0;

    if (!(obj = networkObjFromNetwork(port->net)))
        return ret;

    def = virNetworkObjGetDef(obj);

    if (!(portdef = virNetworkObjLookupPort(obj, port->uuid)))
        goto cleanup;

    if (virNetworkPortGetParametersEnsureACL(port->net->conn, def, portdef) < 0)
        goto cleanup;

    if (!virNetworkObjIsActive(obj)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       def->name);
        goto cleanup;
    }

    if (portdef->bandwidth) {
        if ((portdef->bandwidth->in != NULL) &&
            (virTypedParamsAddUInt(params, nparams, &maxparams,
                                   VIR_NETWORK_PORT_BANDWIDTH_IN_AVERAGE,
                                   portdef->bandwidth->in->average) < 0 ||
             virTypedParamsAddUInt(params, nparams, &maxparams,
                                   VIR_NETWORK_PORT_BANDWIDTH_IN_PEAK,
                                   portdef->bandwidth->in->peak) < 0 ||
             virTypedParamsAddUInt(params, nparams, &maxparams,
                                   VIR_NETWORK_PORT_BANDWIDTH_IN_FLOOR,
                                   portdef->bandwidth->in->floor) < 0 ||
             virTypedParamsAddUInt(params, nparams, &maxparams,
                                   VIR_NETWORK_PORT_BANDWIDTH_IN_BURST,
                                   portdef->bandwidth->in->burst) < 0))
            goto cleanup;

        if ((portdef->bandwidth->out != NULL) &&
            (virTypedParamsAddUInt(params, nparams, &maxparams,
                                   VIR_NETWORK_PORT_BANDWIDTH_OUT_AVERAGE,
                                   portdef->bandwidth->out->average) < 0 ||
             virTypedParamsAddUInt(params, nparams, &maxparams,
                                   VIR_NETWORK_PORT_BANDWIDTH_OUT_PEAK,
                                   portdef->bandwidth->out->peak) < 0 ||
             virTypedParamsAddUInt(params, nparams, &maxparams,
                                   VIR_NETWORK_PORT_BANDWIDTH_OUT_BURST,
                                   portdef->bandwidth->out->burst) < 0))
            goto cleanup;
    }

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


static int
networkListAllPorts(virNetworkPtr net,
                    virNetworkPortPtr **ports,
                    unsigned int flags)
{
    virNetworkObjPtr obj;
    virNetworkDefPtr def;
    int ret = -1;

    virCheckFlags(0, -1);

    if (!(obj = networkObjFromNetwork(net)))
        return ret;

    def = virNetworkObjGetDef(obj);

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

    if (!virNetworkObjIsActive(obj)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       def->name);
        goto cleanup;
    }

    ret = virNetworkObjPortListExport(net, obj, ports,
                                      virNetworkListAllPortsCheckACL);

 cleanup:
    virNetworkObjEndAPI(&obj);
    return ret;
}


5896 5897 5898 5899 5900 5901 5902 5903 5904 5905 5906 5907 5908 5909 5910 5911 5912 5913 5914 5915 5916 5917 5918 5919
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 */
5920 5921 5922 5923 5924 5925 5926
    .networkPortLookupByUUID = networkPortLookupByUUID, /* 5.5.0 */
    .networkPortCreateXML = networkPortCreateXML, /* 5.5.0 */
    .networkPortGetXMLDesc = networkPortGetXMLDesc, /* 5.5.0 */
    .networkPortDelete = networkPortDelete, /* 5.5.0 */
    .networkListAllPorts = networkListAllPorts, /* 5.5.0 */
    .networkPortGetParameters = networkPortGetParameters, /* 5.5.0 */
    .networkPortSetParameters = networkPortSetParameters, /* 5.5.0 */
5927 5928
};

5929 5930 5931 5932 5933 5934 5935 5936 5937 5938 5939 5940

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 = {
5941
    .localOnly = true,
5942
    .uriSchemes = (const char *[]){ "network", NULL },
5943 5944 5945 5946 5947
    .hypervisorDriver = &networkHypervisorDriver,
    .networkDriver = &networkDriver,
};


5948 5949 5950 5951 5952 5953 5954 5955 5956 5957
static virStateDriver networkStateDriver = {
    .name = "bridge",
    .stateInitialize  = networkStateInitialize,
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
};

int
networkRegister(void)
{
5958 5959
    if (virRegisterConnectDriver(&networkConnectDriver, false) < 0)
        return -1;
5960 5961 5962 5963 5964 5965
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
        return -1;
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
    return 0;
}