bridge_driver.c 134.2 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2012 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

#include <sys/types.h>
#include <sys/poll.h>
#include <limits.h>
#include <string.h>
#include <stdio.h>
#include <stdarg.h>
#include <stdlib.h>
#include <unistd.h>
#include <errno.h>
#include <sys/utsname.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <signal.h>
#include <paths.h>
#include <pwd.h>
#include <stdio.h>
#include <sys/wait.h>
#include <sys/ioctl.h>
44
#include <net/if.h>
45

46
#include "virterror_internal.h"
47
#include "datatypes.h"
48
#include "bridge_driver.h"
49
#include "network_conf.h"
50
#include "device_conf.h"
51 52
#include "driver.h"
#include "buf.h"
53
#include "virpidfile.h"
54
#include "util.h"
55
#include "command.h"
56 57 58
#include "memory.h"
#include "uuid.h"
#include "iptables.h"
59
#include "logging.h"
60
#include "dnsmasq.h"
61
#include "configmake.h"
62
#include "virnetdev.h"
63
#include "pci.h"
64 65
#include "virnetdevbridge.h"
#include "virnetdevtap.h"
66
#include "virnetdevvportprofile.h"
67
#include "virdbus.h"
68

69 70
#define NETWORK_PID_DIR LOCALSTATEDIR "/run/libvirt/network"
#define NETWORK_STATE_DIR LOCALSTATEDIR "/lib/libvirt/network"
71

72
#define DNSMASQ_STATE_DIR LOCALSTATEDIR "/lib/libvirt/dnsmasq"
73
#define RADVD_STATE_DIR LOCALSTATEDIR "/lib/libvirt/radvd"
74

75 76
#define VIR_FROM_THIS VIR_FROM_NETWORK

77 78
/* Main driver state */
struct network_driver {
79
    virMutex lock;
80

81
    virNetworkObjList networks;
82 83 84 85 86 87 88

    iptablesContext *iptables;
    char *networkConfigDir;
    char *networkAutostartDir;
    char *logDir;
};

89 90 91

static void networkDriverLock(struct network_driver *driver)
{
92
    virMutexLock(&driver->lock);
93 94 95
}
static void networkDriverUnlock(struct network_driver *driver)
{
96
    virMutexUnlock(&driver->lock);
97 98
}

99 100
static int networkShutdown(void);

101 102 103 104 105 106 107
static int networkStartNetwork(struct network_driver *driver,
                               virNetworkObjPtr network);

static int networkShutdownNetwork(struct network_driver *driver,
                                  virNetworkObjPtr network);

static int networkStartNetworkVirtual(struct network_driver *driver,
108
                                     virNetworkObjPtr network);
109

110 111 112 113 114 115 116
static int networkShutdownNetworkVirtual(struct network_driver *driver,
                                        virNetworkObjPtr network);

static int networkStartNetworkExternal(struct network_driver *driver,
                                     virNetworkObjPtr network);

static int networkShutdownNetworkExternal(struct network_driver *driver,
117
                                        virNetworkObjPtr network);
118

119
static void networkReloadIptablesRules(struct network_driver *driver);
120
static void networkRefreshDaemons(struct network_driver *driver);
121

122 123
static struct network_driver *driverState = NULL;

124
static char *
125
networkDnsmasqLeaseFileNameDefault(const char *netname)
126 127 128
{
    char *leasefile;

129 130
    ignore_value(virAsprintf(&leasefile, DNSMASQ_STATE_DIR "/%s.leases",
                             netname));
131 132 133
    return leasefile;
}

134 135 136
networkDnsmasqLeaseFileNameFunc networkDnsmasqLeaseFileName =
    networkDnsmasqLeaseFileNameDefault;

137 138 139 140 141 142
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

143
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
144 145 146 147 148 149 150 151
    return pidfilebase;
}

static char *
networkRadvdConfigFileName(const char *netname)
{
    char *configfile;

152 153
    ignore_value(virAsprintf(&configfile, RADVD_STATE_DIR "/%s-radvd.conf",
                             netname));
154 155
    return configfile;
}
156

157 158 159
static char *
networkBridgeDummyNicName(const char *brname)
{
160
    static const char dummyNicSuffix[] = "-nic";
161 162
    char *nicname;

163 164 165 166 167 168 169
    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.
         */
170 171 172 173 174
        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));
175
    } else {
176
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
177
    }
178 179 180
    return nicname;
}

181 182 183 184 185 186 187 188 189 190 191
static void
networkFindActiveConfigs(struct network_driver *driver) {
    unsigned int i;

    for (i = 0 ; i < driver->networks.count ; i++) {
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkDefPtr tmp;
        char *config;

        virNetworkObjLock(obj);

192
        if ((config = virNetworkConfigFile(NETWORK_STATE_DIR,
193 194 195 196 197 198 199 200 201 202 203 204
                                           obj->def->name)) == NULL) {
            virNetworkObjUnlock(obj);
            continue;
        }

        if (access(config, R_OK) < 0) {
            VIR_FREE(config);
            virNetworkObjUnlock(obj);
            continue;
        }

        /* Try and load the live config */
205
        tmp = virNetworkDefParseFile(config);
206 207 208 209 210 211 212 213
        VIR_FREE(config);
        if (tmp) {
            obj->newDef = obj->def;
            obj->def = tmp;
        }

        /* If bridge exists, then mark it active */
        if (obj->def->bridge &&
H
Hu Tao 已提交
214
            virNetDevExists(obj->def->bridge) == 1) {
215 216
            obj->active = 1;

217 218
            /* Try and read dnsmasq/radvd pids if any */
            if (obj->def->ips && (obj->def->nips > 0)) {
219 220 221 222
                char *radvdpidbase;

                ignore_value(virPidFileReadIfAlive(NETWORK_PID_DIR, obj->def->name,
                                                   &obj->dnsmasqPid, DNSMASQ));
223

224
                if (!(radvdpidbase = networkRadvdPidfileBasename(obj->def->name))) {
225
                    virReportOOMError();
226 227
                    goto cleanup;
                }
228 229
                ignore_value(virPidFileReadIfAlive(NETWORK_PID_DIR, radvdpidbase,
                                                   &obj->radvdPid, RADVD));
230
                VIR_FREE(radvdpidbase);
231 232 233
            }
        }

234
    cleanup:
235 236 237 238 239
        virNetworkObjUnlock(obj);
    }
}


240 241 242
static void
networkAutostartConfigs(struct network_driver *driver) {
    unsigned int i;
243

244
    for (i = 0 ; i < driver->networks.count ; i++) {
245
        virNetworkObjLock(driver->networks.objs[i]);
246
        if (driver->networks.objs[i]->autostart &&
247 248
            !virNetworkObjIsActive(driver->networks.objs[i])) {
            if (networkStartNetwork(driver, driver->networks.objs[i]) < 0) {
249
            /* failed to start but already logged */
250
            }
251
        }
252
        virNetworkObjUnlock(driver->networks.objs[i]);
253 254 255
    }
}

256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
                             DBusMessage *message, void *user_data) {
    struct network_driver *_driverState = user_data;

    if (dbus_message_is_signal(message, DBUS_INTERFACE_DBUS,
                               "NameOwnerChanged") ||
        dbus_message_is_signal(message, "org.fedoraproject.FirewallD1",
                               "Reloaded"))
    {
        VIR_DEBUG("Reload in bridge_driver because of firewalld.");
        networkReloadIptablesRules(_driverState);
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

275 276 277 278 279 280
/**
 * networkStartup:
 *
 * Initialization function for the QEmu daemon
 */
static int
281
networkStartup(int privileged) {
282
    char *base = NULL;
283 284 285
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
286 287

    if (VIR_ALLOC(driverState) < 0)
288
        goto error;
289

290 291 292 293
    if (virMutexInit(&driverState->lock) < 0) {
        VIR_FREE(driverState);
        goto error;
    }
294 295
    networkDriverLock(driverState);

296
    if (privileged) {
297
        if (virAsprintf(&driverState->logDir,
298
                        "%s/log/libvirt/qemu", LOCALSTATEDIR) == -1)
299 300
            goto out_of_memory;

301
        if ((base = strdup (SYSCONFDIR "/libvirt")) == NULL)
302 303
            goto out_of_memory;
    } else {
304
        char *userdir = virGetUserCacheDirectory();
305 306 307

        if (!userdir)
            goto error;
308

309
        if (virAsprintf(&driverState->logDir,
310
                        "%s/qemu/log", userdir) == -1) {
311
            VIR_FREE(userdir);
312
            goto out_of_memory;
313
        }
314
        VIR_FREE(userdir);
315

316
        userdir = virGetUserConfigDirectory();
317
        if (virAsprintf(&base, "%s", userdir) == -1) {
318
            VIR_FREE(userdir);
319 320
            goto out_of_memory;
        }
321
        VIR_FREE(userdir);
322 323 324 325 326
    }

    /* Configuration paths are either ~/.libvirt/qemu/... (session) or
     * /etc/libvirt/qemu/... (system).
     */
327
    if (virAsprintf(&driverState->networkConfigDir, "%s/qemu/networks", base) == -1)
328 329
        goto out_of_memory;

330 331
    if (virAsprintf(&driverState->networkAutostartDir, "%s/qemu/networks/autostart",
                    base) == -1)
332 333 334 335
        goto out_of_memory;

    VIR_FREE(base);

336
    if (!(driverState->iptables = iptablesContextNew())) {
337
        goto out_of_memory;
338 339 340
    }


341
    if (virNetworkLoadAllConfigs(&driverState->networks,
342
                                 driverState->networkConfigDir,
343 344 345
                                 driverState->networkAutostartDir) < 0)
        goto error;

346
    networkFindActiveConfigs(driverState);
347
    networkReloadIptablesRules(driverState);
348
    networkRefreshDaemons(driverState);
349 350
    networkAutostartConfigs(driverState);

351 352
    networkDriverUnlock(driverState);

353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378
#ifdef HAVE_FIREWALLD
    if (!(sysbus = virDBusGetSystemBus())) {
        virErrorPtr err = virGetLastError();
        VIR_WARN("DBus not available, disabling firewalld support "
                 "in bridge_driver: %s", err->message);
    } 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,
                                   driverState, NULL);
    }
#endif

379 380
    return 0;

381
out_of_memory:
382
    virReportOOMError();
383 384

error:
385 386 387
    if (driverState)
        networkDriverUnlock(driverState);

388
    VIR_FREE(base);
389
    networkShutdown();
390 391 392 393 394 395 396 397 398 399 400
    return -1;
}

/**
 * networkReload:
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
networkReload(void) {
401 402 403
    if (!driverState)
        return 0;

404
    networkDriverLock(driverState);
405
    virNetworkLoadAllConfigs(&driverState->networks,
406 407
                             driverState->networkConfigDir,
                             driverState->networkAutostartDir);
408
    networkReloadIptablesRules(driverState);
409
    networkRefreshDaemons(driverState);
410
    networkAutostartConfigs(driverState);
411
    networkDriverUnlock(driverState);
412 413 414 415 416 417 418 419 420 421 422 423 424
    return 0;
}

/**
 * networkActive:
 *
 * Checks if the QEmu daemon is active, i.e. has an active domain or
 * an active network
 *
 * Returns 1 if active, 0 otherwise
 */
static int
networkActive(void) {
425
    unsigned int i;
426
    int active = 0;
427

428 429 430
    if (!driverState)
        return 0;

431
    networkDriverLock(driverState);
432 433
    for (i = 0 ; i < driverState->networks.count ; i++) {
        virNetworkObjPtr net = driverState->networks.objs[i];
434
        virNetworkObjLock(net);
D
Daniel P. Berrange 已提交
435
        if (virNetworkObjIsActive(net))
436
            active = 1;
437
        virNetworkObjUnlock(net);
438
    }
439
    networkDriverUnlock(driverState);
440
    return active;
441 442 443 444 445 446 447 448 449 450 451 452
}

/**
 * networkShutdown:
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
networkShutdown(void) {
    if (!driverState)
        return -1;

453 454
    networkDriverLock(driverState);

455
    /* free inactive networks */
456
    virNetworkObjListFree(&driverState->networks);
457 458 459 460 461 462 463 464

    VIR_FREE(driverState->logDir);
    VIR_FREE(driverState->networkConfigDir);
    VIR_FREE(driverState->networkAutostartDir);

    if (driverState->iptables)
        iptablesContextFree(driverState->iptables);

465
    networkDriverUnlock(driverState);
466
    virMutexDestroy(&driverState->lock);
467

468 469 470 471 472 473
    VIR_FREE(driverState);

    return 0;
}


474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532
/* networkKillDaemon:
 *
 * kill the specified pid/name, and wait a bit to make sure it's dead.
 */
static int
networkKillDaemon(pid_t pid, const char *daemonName, const char *networkName)
{
    int ii, ret = -1;
    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.
     */
    for (ii = 0; ii < 25; ii++) {
        int signum = 0;
        if (ii == 0)
            signum = SIGTERM;
        else if (ii == 15) {
            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);
cleanup:
    return ret;
}

533 534 535 536
static int
networkBuildDnsmasqHostsfile(dnsmasqContext *dctx,
                             virNetworkIpDefPtr ipdef,
                             virNetworkDNSDefPtr dnsdef)
537
{
538
    unsigned int i, j;
539

540 541
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
542
        if ((host->mac) && VIR_SOCKET_ADDR_VALID(&host->ip))
543 544
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip, host->name) < 0)
                return -1;
545
    }
546

547 548 549
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
            virNetworkDNSHostsDefPtr host = &(dnsdef->hosts[i]);
550
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
551
                for (j = 0; j < host->nnames; j++)
552 553
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
554 555
            }
        }
556 557
    }

558
    return 0;
559 560 561
}


562
static int
563
networkBuildDnsmasqArgv(virNetworkObjPtr network,
564
                        virNetworkIpDefPtr ipdef,
565
                        const char *pidfile,
566 567 568
                        virCommandPtr cmd,
                        dnsmasqContext *dctx)
{
569
    int r, ret = -1;
570
    int nbleases = 0;
571
    int ii;
572 573 574 575
    char *record = NULL;
    char *recordPort = NULL;
    char *recordWeight = NULL;
    char *recordPriority = NULL;
576
    virNetworkIpDefPtr tmpipdef;
577 578

    /*
579
     * NB, be careful about syntax for dnsmasq options in long format.
580 581 582 583 584 585 586 587 588 589 590 591 592
     *
     * If the flag has a mandatory argument, it can be given using
     * either syntax:
     *
     *     --foo bar
     *     --foo=bar
     *
     * If the flag has a optional argument, it *must* be given using
     * the syntax:
     *
     *     --foo=bar
     *
     * It is hard to determine whether a flag is optional or not,
593 594
     * without reading the dnsmasq source :-( The manpage is not
     * very explicit on this.
595
     */
596 597 598 599 600

    /*
     * Needed to ensure dnsmasq uses same algorithm for processing
     * multiple namedriver entries in /etc/resolv.conf as GLibC.
     */
601
    virCommandAddArgList(cmd, "--strict-order", "--bind-interfaces", NULL);
602

603
    if (network->def->domain)
604 605 606 607
        virCommandAddArgPair(cmd, "--domain", network->def->domain);
    /* need to specify local even if no domain specified */
    virCommandAddArgFormat(cmd, "--local=/%s/",
                           network->def->domain ? network->def->domain : "");
608
    virCommandAddArg(cmd, "--domain-needed");
609

610 611
    if (pidfile)
        virCommandAddArgPair(cmd, "--pid-file", pidfile);
612

613
    /* *no* conf file */
614
    virCommandAddArg(cmd, "--conf-file=");
615

616 617 618
    virCommandAddArgList(cmd,
                         "--except-interface", "lo",
                         NULL);
619

620 621
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
622 623 624 625
     * guaranteed to not work, and set --no-resolv so that no dns
     * 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).
626
     */
627 628 629 630
    if (network->def->forwardType == VIR_NETWORK_FORWARD_NONE) {
        virCommandAddArgList(cmd, "--dhcp-option=3",
                             "--no-resolv", NULL);
    }
631

632 633 634 635 636
    if (network->def->dns != NULL) {
        virNetworkDNSDefPtr dns = network->def->dns;
        int i;

        for (i = 0; i < dns->ntxtrecords; i++) {
E
Eric Blake 已提交
637
            virCommandAddArgFormat(cmd, "--txt-record=%s,%s",
638 639
                                   dns->txtrecords[i].name,
                                   dns->txtrecords[i].value);
640
        }
641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676

        for (i = 0; i < dns->nsrvrecords; i++) {
            if (dns->srvrecords[i].service && dns->srvrecords[i].protocol) {
                if (dns->srvrecords[i].port) {
                    if (virAsprintf(&recordPort, "%d", dns->srvrecords[i].port) < 0) {
                        virReportOOMError();
                        goto cleanup;
                    }
                }
                if (dns->srvrecords[i].priority) {
                    if (virAsprintf(&recordPriority, "%d", dns->srvrecords[i].priority) < 0) {
                        virReportOOMError();
                        goto cleanup;
                    }
                }
                if (dns->srvrecords[i].weight) {
                    if (virAsprintf(&recordWeight, "%d", dns->srvrecords[i].weight) < 0) {
                        virReportOOMError();
                        goto cleanup;
                    }
                }

                if (virAsprintf(&record, "%s.%s.%s,%s,%s,%s,%s",
                                dns->srvrecords[i].service,
                                dns->srvrecords[i].protocol,
                                dns->srvrecords[i].domain   ? dns->srvrecords[i].domain : "",
                                dns->srvrecords[i].target   ? dns->srvrecords[i].target : "",
                                recordPort                  ? recordPort                : "",
                                recordPriority              ? recordPriority            : "",
                                recordWeight                ? recordWeight              : "") < 0) {
                    virReportOOMError();
                    goto cleanup;
                }

                virCommandAddArgPair(cmd, "--srv-host", record);
                VIR_FREE(record);
677 678 679
                VIR_FREE(recordPort);
                VIR_FREE(recordWeight);
                VIR_FREE(recordPriority);
680 681
            }
        }
682 683
    }

684 685 686 687 688 689 690 691 692 693 694
    /*
     * --interface does not actually work with dnsmasq < 2.47,
     * due to DAD for ipv6 addresses on the interface.
     *
     * virCommandAddArgList(cmd, "--interface", ipdef->bridge, NULL);
     *
     * So listen on all defined IPv[46] addresses
     */
    for (ii = 0;
         (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, ii));
         ii++) {
695
        char *ipaddr = virSocketAddrFormat(&tmpipdef->address);
696 697 698 699 700 701
        if (!ipaddr)
            goto cleanup;
        virCommandAddArgList(cmd, "--listen-address", ipaddr, NULL);
        VIR_FREE(ipaddr);
    }

702
    if (ipdef) {
703
        for (r = 0 ; r < ipdef->nranges ; r++) {
704
            char *saddr = virSocketAddrFormat(&ipdef->ranges[r].start);
705 706
            if (!saddr)
                goto cleanup;
707
            char *eaddr = virSocketAddrFormat(&ipdef->ranges[r].end);
708 709 710 711 712 713
            if (!eaddr) {
                VIR_FREE(saddr);
                goto cleanup;
            }
            virCommandAddArg(cmd, "--dhcp-range");
            virCommandAddArgFormat(cmd, "%s,%s", saddr, eaddr);
714
            VIR_FREE(saddr);
715
            VIR_FREE(eaddr);
716 717
            nbleases += virSocketAddrGetRange(&ipdef->ranges[r].start,
                                              &ipdef->ranges[r].end);
718
        }
719

720 721 722 723 724 725
        /*
         * 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.
         */
        if (!ipdef->nranges && ipdef->nhosts) {
726
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
727 728 729 730 731 732
            if (!bridgeaddr)
                goto cleanup;
            virCommandAddArg(cmd, "--dhcp-range");
            virCommandAddArgFormat(cmd, "%s,static", bridgeaddr);
            VIR_FREE(bridgeaddr);
        }
733

734
        if (ipdef->nranges > 0) {
735 736 737 738 739
            char *leasefile = networkDnsmasqLeaseFileName(network->def->name);
            if (!leasefile)
                goto cleanup;
            virCommandAddArgFormat(cmd, "--dhcp-leasefile=%s", leasefile);
            VIR_FREE(leasefile);
740 741
            virCommandAddArgFormat(cmd, "--dhcp-lease-max=%d", nbleases);
        }
742

743 744
        if (ipdef->nranges || ipdef->nhosts)
            virCommandAddArg(cmd, "--dhcp-no-override");
745

746 747 748 749
        /* add domain to any non-qualified hostnames in /etc/hosts or addn-hosts */
        if (network->def->domain)
           virCommandAddArg(cmd, "--expand-hosts");

750 751 752 753 754 755 756 757 758
        if (networkBuildDnsmasqHostsfile(dctx, ipdef, network->def->dns) < 0)
            goto cleanup;

        if (dctx->hostsfile->nhosts)
            virCommandAddArgPair(cmd, "--dhcp-hostsfile",
                                 dctx->hostsfile->path);
        if (dctx->addnhostsfile->nhosts)
            virCommandAddArgPair(cmd, "--addn-hosts",
                                 dctx->addnhostsfile->path);
759

760 761 762 763 764 765 766
        if (ipdef->tftproot) {
            virCommandAddArgList(cmd, "--enable-tftp",
                                 "--tftp-root", ipdef->tftproot,
                                 NULL);
        }
        if (ipdef->bootfile) {
            virCommandAddArg(cmd, "--dhcp-boot");
767 768
            if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
769

770 771 772 773 774 775 776 777
                if (!bootserver)
                    goto cleanup;
                virCommandAddArgFormat(cmd, "%s%s%s",
                                       ipdef->bootfile, ",,", bootserver);
                VIR_FREE(bootserver);
            } else {
                virCommandAddArg(cmd, ipdef->bootfile);
            }
778
        }
779 780
    }

781 782
    ret = 0;
cleanup:
783 784 785 786
    VIR_FREE(record);
    VIR_FREE(recordPort);
    VIR_FREE(recordWeight);
    VIR_FREE(recordPriority);
787
    return ret;
788 789
}

790 791
int
networkBuildDhcpDaemonCommandLine(virNetworkObjPtr network, virCommandPtr *cmdout,
792
                                  char *pidfile, dnsmasqContext *dctx)
793
{
794
    virCommandPtr cmd = NULL;
795
    int ret = -1, ii;
796
    virNetworkIpDefPtr ipdef;
797 798

    network->dnsmasqPid = -1;
799

800 801 802 803 804 805 806
    /* Look for first IPv4 address that has dhcp defined. */
    /* We support dhcp config on 1 IPv4 interface only. */
    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, ii));
         ii++) {
        if (ipdef->nranges || ipdef->nhosts)
            break;
807
    }
808
    /* If no IPv4 addresses had dhcp info, pick the first (if there were any). */
809
    if (!ipdef)
810 811 812 813 814 815 816 817
        ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, 0);

    /* If there are no IP addresses at all (v4 or v6), return now, since
     * there won't be any address for dnsmasq to listen on anyway.
     * If there are any addresses, even if no dhcp ranges or static entries,
     * we should continue and run dnsmasq, just for the DNS capabilities.
     */
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0))
818
        return 0;
819

820
    cmd = virCommandNew(DNSMASQ);
821
    if (networkBuildDnsmasqArgv(network, ipdef, pidfile, cmd, dctx) < 0) {
822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839
        goto cleanup;
    }

    if (cmdout)
        *cmdout = cmd;
    ret = 0;
cleanup:
    if (ret < 0)
        virCommandFree(cmd);
    return ret;
}

static int
networkStartDhcpDaemon(virNetworkObjPtr network)
{
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
840
    dnsmasqContext *dctx = NULL;
841

842 843 844 845 846 847
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0)) {
        /* no IPv6 addresses, so we don't need to run radvd */
        ret = 0;
        goto cleanup;
    }

848 849
    if (virFileMakePath(NETWORK_PID_DIR) < 0) {
        virReportSystemError(errno,
850 851
                             _("cannot create directory %s"),
                             NETWORK_PID_DIR);
852
        goto cleanup;
853
    }
854 855
    if (virFileMakePath(NETWORK_STATE_DIR) < 0) {
        virReportSystemError(errno,
856 857
                             _("cannot create directory %s"),
                             NETWORK_STATE_DIR);
858
        goto cleanup;
859 860
    }

861
    if (!(pidfile = virPidFileBuildPath(NETWORK_PID_DIR, network->def->name))) {
862
        virReportOOMError();
863
        goto cleanup;
864 865
    }

866 867
    if (virFileMakePath(DNSMASQ_STATE_DIR) < 0) {
        virReportSystemError(errno,
868 869 870 871 872
                             _("cannot create directory %s"),
                             DNSMASQ_STATE_DIR);
        goto cleanup;
    }

873 874 875 876 877 878 879 880 881 882
    dctx = dnsmasqContextNew(network->def->name, DNSMASQ_STATE_DIR);
    if (dctx == NULL)
        goto cleanup;

    ret = networkBuildDhcpDaemonCommandLine(network, &cmd, pidfile, dctx);
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
883
        goto cleanup;
884

G
Guido Günther 已提交
885 886
    ret = virCommandRun(cmd, NULL);
    if (ret < 0) {
887
        goto cleanup;
G
Guido Günther 已提交
888
    }
889 890

    /*
891 892 893 894 895
     * 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
896 897
     */

898
    ret = virPidFileRead(NETWORK_PID_DIR, network->def->name,
899 900
                         &network->dnsmasqPid);
    if (ret < 0)
901
        goto cleanup;
902

903 904 905
    ret = 0;
cleanup:
    VIR_FREE(pidfile);
906
    virCommandFree(cmd);
907
    dnsmasqContextFree(dctx);
908 909 910
    return ret;
}

911 912 913 914 915 916
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
 *  them.
 *
 *  Returns 0 on success, -1 on failure.
 */
917
static int
918
networkRefreshDhcpDaemon(virNetworkObjPtr network)
919
{
920
    int ret = -1, ii;
921
    virNetworkIpDefPtr ipdef;
922
    dnsmasqContext *dctx = NULL;
923

924 925 926
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
        return networkStartDhcpDaemon(network);
927

928 929 930 931 932 933 934
    /* Look for first IPv4 address that has dhcp defined. */
    /* We support dhcp config on 1 IPv4 interface only. */
    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, ii));
         ii++) {
        if (ipdef->nranges || ipdef->nhosts)
            break;
935
    }
936 937 938
    /* If no IPv4 addresses had dhcp info, pick the first (if there were any). */
    if (!ipdef)
        ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, 0);
939

940 941 942
    if (!ipdef) {
        /* no <ip> elements, so nothing to do */
        return 0;
943 944
    }

945
    if (!(dctx = dnsmasqContextNew(network->def->name, DNSMASQ_STATE_DIR)))
946
        goto cleanup;
947 948 949 950 951

    if (networkBuildDnsmasqHostsfile(dctx, ipdef, network->def->dns) < 0)
       goto cleanup;

    if ((ret = dnsmasqSave(dctx)) < 0)
952
        goto cleanup;
953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974

    ret = kill(network->dnsmasqPid, SIGHUP);
cleanup:
    dnsmasqContextFree(dctx);
    return ret;
}

/* networkRestartDhcpDaemon:
 *
 * kill and restart dnsmasq, in order to update any config that is on
 * the dnsmasq commandline (and any placed in separate config files).
 *
 *  Returns 0 on success, -1 on failure.
 */
static int
networkRestartDhcpDaemon(virNetworkObjPtr network)
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
975
    }
976 977 978 979 980 981 982 983 984 985 986 987 988
    /* now start dnsmasq if it should be started */
    return networkStartDhcpDaemon(network);
}

static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;;
    int ret = -1, ii;
    virNetworkIpDefPtr ipdef;
    bool v6present = false;

    *configstr = NULL;
989 990

    /* create radvd config file appropriate for this network */
991
    virBufferAsprintf(&configbuf, "interface %s\n"
992 993 994 995 996 997
                      "{\n"
                      "  AdvSendAdvert on;\n"
                      "  AdvManagedFlag off;\n"
                      "  AdvOtherConfigFlag off;\n"
                      "\n",
                      network->def->bridge);
998 999

    /* add a section for each IPv6 address in the config */
1000 1001 1002 1003 1004 1005
    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, ii));
         ii++) {
        int prefix;
        char *netaddr;

1006
        v6present = true;
1007 1008
        prefix = virNetworkIpDefPrefix(ipdef);
        if (prefix < 0) {
1009 1010 1011
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1012 1013
            goto cleanup;
        }
1014
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1015
            goto cleanup;
1016
        virBufferAsprintf(&configbuf,
1017 1018 1019 1020 1021 1022 1023 1024 1025 1026
                          "  prefix %s/%d\n"
                          "  {\n"
                          "    AdvOnLink on;\n"
                          "    AdvAutonomous on;\n"
                          "    AdvRouterAddr off;\n"
                          "  };\n",
                          netaddr, prefix);
        VIR_FREE(netaddr);
    }

1027 1028 1029
    /* only create the string if we found at least one IPv6 address */
    if (v6present) {
        virBufferAddLit(&configbuf, "};\n");
1030

1031 1032 1033 1034 1035 1036 1037 1038
        if (virBufferError(&configbuf)) {
            virReportOOMError();
            goto cleanup;
        }
        if (!(*configstr = virBufferContentAndReset(&configbuf))) {
            virReportOOMError();
            goto cleanup;
        }
1039
    }
1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064

    ret = 0;
cleanup:
    virBufferFreeAndReset(&configbuf);
    return ret;
}

/* write file and return it's name (which must be freed by caller) */
static int
networkRadvdConfWrite(virNetworkObjPtr network, char **configFile)
{
    int ret = -1;
    char *configStr = NULL;
    char *myConfigFile = NULL;

    if (!configFile)
        configFile = &myConfigFile;

    *configFile = NULL;

    if (networkRadvdConfContents(network, &configStr) < 0)
        goto cleanup;

    if (!configStr) {
        ret = 0;
1065 1066 1067 1068
        goto cleanup;
    }

    /* construct the filename */
1069
    if (!(*configFile = networkRadvdConfigFileName(network->def->name))) {
1070 1071 1072 1073
        virReportOOMError();
        goto cleanup;
    }
    /* write the file */
1074
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1075 1076
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109
                             *configFile);
        goto cleanup;
    }

    ret = 0;
cleanup:
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
networkStartRadvd(virNetworkObjPtr network)
{
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

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

    if (!virFileIsExecutable(RADVD)) {
        virReportSystemError(errno,
                             _("Cannot find %s - "
                               "Possibly the package isn't installed"),
                             RADVD);
1110 1111 1112
        goto cleanup;
    }

1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138
    if (virFileMakePath(NETWORK_PID_DIR) < 0) {
        virReportSystemError(errno,
                             _("cannot create directory %s"),
                             NETWORK_PID_DIR);
        goto cleanup;
    }
    if (virFileMakePath(RADVD_STATE_DIR) < 0) {
        virReportSystemError(errno,
                             _("cannot create directory %s"),
                             RADVD_STATE_DIR);
        goto cleanup;
    }

    /* construct pidfile name */
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
        virReportOOMError();
        goto cleanup;
    }
    if (!(pidfile = virPidFileBuildPath(NETWORK_PID_DIR, radvdpidbase))) {
        virReportOOMError();
        goto cleanup;
    }

    if (networkRadvdConfWrite(network, &configfile) < 0)
        goto cleanup;

1139 1140 1141 1142
    /* 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
1143
     * virPidFileRead() below will fail if we use them).
1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158
     * 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;

1159
    if (virPidFileRead(NETWORK_PID_DIR, radvdpidbase, &network->radvdPid) < 0)
1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170
        goto cleanup;

    ret = 0;
cleanup:
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188
static int
networkRefreshRadvd(virNetworkObjPtr network)
{
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
        return networkStartRadvd(network);

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

    if (networkRadvdConfWrite(network, NULL) < 0)
        return -1;

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

1189 1190
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215
static int
networkRestartRadvd(virNetworkObjPtr network)
{
    char *radvdpidbase;

    /* if there is a running radvd, kill it */
    if (network->radvdPid > 0) {
        /* essentially ignore errors from the following two functions,
         * since there's really no better recovery to be done than to
         * just push ahead (and that may be exactly what's needed).
         */
        if ((networkKillDaemon(network->dnsmasqPid, "radvd",
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
            virPidFileDelete(NETWORK_PID_DIR, radvdpidbase);
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
networkRefreshDaemons(struct network_driver *driver)
{
    unsigned int i;

    VIR_INFO("Refreshing network daemons");

    for (i = 0 ; i < driver->networks.count ; i++) {
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
            ((network->def->forwardType == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forwardType == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forwardType == VIR_NETWORK_FORWARD_ROUTE))) {
            /* Only the three L3 network types that are configured by
             * libvirt will have a dnsmasq or radvd daemon associated
             * with them.  Here we send a SIGHUP to an existing
             * dnsmasq and/or radvd, or restart them if they've
             * disappeared.
             */
            networkRefreshDhcpDaemon(network);
            networkRefreshRadvd(network);
        }
        virNetworkObjUnlock(network);
    }
}

1247
static int
1248
networkAddMasqueradingIptablesRules(struct network_driver *driver,
1249 1250
                                    virNetworkObjPtr network,
                                    virNetworkIpDefPtr ipdef)
1251 1252
{
    int prefix = virNetworkIpDefPrefix(ipdef);
1253
    const char *forwardIf = virNetworkDefForwardIf(network->def, 0);
1254 1255

    if (prefix < 0) {
1256 1257 1258
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Invalid prefix or netmask for '%s'"),
                       network->def->bridge);
1259 1260
        goto masqerr1;
    }
1261

1262
    /* allow forwarding packets from the bridge interface */
1263
    if (iptablesAddForwardAllowOut(driver->iptables,
1264
                                   &ipdef->address,
1265
                                   prefix,
1266
                                   network->def->bridge,
1267
                                   forwardIf) < 0) {
1268 1269 1270
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow forwarding from '%s'"),
                       network->def->bridge);
1271 1272 1273
        goto masqerr1;
    }

1274 1275 1276
    /* allow forwarding packets to the bridge interface if they are
     * part of an existing connection
     */
1277
    if (iptablesAddForwardAllowRelatedIn(driver->iptables,
1278
                                         &ipdef->address,
1279
                                         prefix,
1280
                                         network->def->bridge,
1281
                                         forwardIf) < 0) {
1282 1283 1284
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow forwarding to '%s'"),
                       network->def->bridge);
1285 1286 1287
        goto masqerr2;
    }

1288 1289 1290 1291 1292
    /*
     * Enable masquerading.
     *
     * We need to end up with 3 rules in the table in this order
     *
E
Eric Blake 已提交
1293 1294
     *  1. protocol=tcp with sport mapping restriction
     *  2. protocol=udp with sport mapping restriction
1295 1296 1297
     *  3. generic any protocol
     *
     * The sport mappings are required, because default IPtables
E
Eric Blake 已提交
1298
     * MASQUERADE maintain port numbers unchanged where possible.
1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311
     *
     * NFS can be configured to only "trust" port numbers < 1023.
     *
     * Guests using NAT thus need to be prevented from having port
     * numbers < 1023, otherwise they can bypass the NFS "security"
     * check on the source port number.
     *
     * Since we use '--insert' to add rules to the header of the
     * chain, we actually need to add them in the reverse of the
     * order just mentioned !
     */

    /* First the generic masquerade rule for other protocols */
1312
    if (iptablesAddForwardMasquerade(driver->iptables,
1313
                                     &ipdef->address,
1314
                                     prefix,
1315
                                     forwardIf,
1316
                                     NULL) < 0) {
1317 1318 1319 1320 1321
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       forwardIf ?
                       _("failed to add iptables rule to enable masquerading to %s") :
                       _("failed to add iptables rule to enable masquerading"),
                       forwardIf);
1322 1323 1324
        goto masqerr3;
    }

1325
    /* UDP with a source port restriction */
1326
    if (iptablesAddForwardMasquerade(driver->iptables,
1327
                                     &ipdef->address,
1328
                                     prefix,
1329
                                     forwardIf,
1330
                                     "udp") < 0) {
1331 1332 1333 1334 1335
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       forwardIf ?
                       _("failed to add iptables rule to enable UDP masquerading to %s") :
                       _("failed to add iptables rule to enable UDP masquerading"),
                       forwardIf);
1336 1337 1338 1339
        goto masqerr4;
    }

    /* TCP with a source port restriction */
1340
    if (iptablesAddForwardMasquerade(driver->iptables,
1341
                                     &ipdef->address,
1342
                                     prefix,
1343
                                     forwardIf,
1344
                                     "tcp") < 0) {
1345 1346 1347 1348 1349
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       forwardIf ?
                       _("failed to add iptables rule to enable TCP masquerading to %s") :
                       _("failed to add iptables rule to enable TCP masquerading"),
                       forwardIf);
1350 1351 1352
        goto masqerr5;
    }

1353
    return 0;
1354

1355 1356
 masqerr5:
    iptablesRemoveForwardMasquerade(driver->iptables,
1357
                                    &ipdef->address,
1358
                                    prefix,
1359
                                    forwardIf,
1360 1361 1362
                                    "udp");
 masqerr4:
    iptablesRemoveForwardMasquerade(driver->iptables,
1363
                                    &ipdef->address,
1364
                                    prefix,
1365
                                    forwardIf,
1366
                                    NULL);
1367 1368
 masqerr3:
    iptablesRemoveForwardAllowRelatedIn(driver->iptables,
1369
                                        &ipdef->address,
1370
                                        prefix,
1371
                                        network->def->bridge,
1372
                                        forwardIf);
1373 1374
 masqerr2:
    iptablesRemoveForwardAllowOut(driver->iptables,
1375
                                  &ipdef->address,
1376
                                  prefix,
1377
                                  network->def->bridge,
1378
                                  forwardIf);
1379
 masqerr1:
1380
    return -1;
1381 1382
}

1383 1384 1385 1386 1387 1388
static void
networkRemoveMasqueradingIptablesRules(struct network_driver *driver,
                                       virNetworkObjPtr network,
                                       virNetworkIpDefPtr ipdef)
{
    int prefix = virNetworkIpDefPrefix(ipdef);
1389
    const char *forwardIf = virNetworkDefForwardIf(network->def, 0);
1390 1391 1392 1393 1394

    if (prefix >= 0) {
        iptablesRemoveForwardMasquerade(driver->iptables,
                                        &ipdef->address,
                                        prefix,
1395
                                        forwardIf,
1396 1397 1398 1399
                                        "tcp");
        iptablesRemoveForwardMasquerade(driver->iptables,
                                        &ipdef->address,
                                        prefix,
1400
                                        forwardIf,
1401 1402 1403 1404
                                        "udp");
        iptablesRemoveForwardMasquerade(driver->iptables,
                                        &ipdef->address,
                                        prefix,
1405
                                        forwardIf,
1406 1407 1408 1409 1410 1411
                                        NULL);

        iptablesRemoveForwardAllowRelatedIn(driver->iptables,
                                            &ipdef->address,
                                            prefix,
                                            network->def->bridge,
1412
                                            forwardIf);
1413 1414 1415 1416
        iptablesRemoveForwardAllowOut(driver->iptables,
                                      &ipdef->address,
                                      prefix,
                                      network->def->bridge,
1417
                                      forwardIf);
1418 1419 1420
    }
}

1421
static int
1422
networkAddRoutingIptablesRules(struct network_driver *driver,
1423
                               virNetworkObjPtr network,
1424 1425
                               virNetworkIpDefPtr ipdef)
{
1426
    int prefix = virNetworkIpDefPrefix(ipdef);
1427
    const char *forwardIf = virNetworkDefForwardIf(network->def, 0);
1428 1429

    if (prefix < 0) {
1430 1431 1432
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Invalid prefix or netmask for '%s'"),
                       network->def->bridge);
1433 1434
        goto routeerr1;
    }
1435

1436
    /* allow routing packets from the bridge interface */
1437
    if (iptablesAddForwardAllowOut(driver->iptables,
1438
                                   &ipdef->address,
1439
                                   prefix,
1440
                                   network->def->bridge,
1441
                                   forwardIf) < 0) {
1442 1443 1444
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow routing from '%s'"),
                       network->def->bridge);
1445 1446 1447 1448
        goto routeerr1;
    }

    /* allow routing packets to the bridge interface */
1449
    if (iptablesAddForwardAllowIn(driver->iptables,
1450
                                  &ipdef->address,
1451
                                  prefix,
1452
                                  network->def->bridge,
1453
                                  forwardIf) < 0) {
1454 1455 1456
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow routing to '%s'"),
                       network->def->bridge);
1457 1458 1459
        goto routeerr2;
    }

1460
    return 0;
1461

1462
routeerr2:
1463
    iptablesRemoveForwardAllowOut(driver->iptables,
1464
                                  &ipdef->address,
1465
                                  prefix,
1466
                                  network->def->bridge,
1467
                                  forwardIf);
1468
routeerr1:
1469
    return -1;
1470 1471
}

1472 1473 1474 1475 1476 1477
static void
networkRemoveRoutingIptablesRules(struct network_driver *driver,
                                  virNetworkObjPtr network,
                                  virNetworkIpDefPtr ipdef)
{
    int prefix = virNetworkIpDefPrefix(ipdef);
1478
    const char *forwardIf = virNetworkDefForwardIf(network->def, 0);
1479 1480 1481 1482 1483 1484

    if (prefix >= 0) {
        iptablesRemoveForwardAllowIn(driver->iptables,
                                     &ipdef->address,
                                     prefix,
                                     network->def->bridge,
1485
                                     forwardIf);
1486 1487 1488 1489 1490

        iptablesRemoveForwardAllowOut(driver->iptables,
                                      &ipdef->address,
                                      prefix,
                                      network->def->bridge,
1491
                                      forwardIf);
1492 1493 1494
    }
}

1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507
/* Add all once/network rules required for IPv6 (if any IPv6 addresses are defined) */
static int
networkAddGeneralIp6tablesRules(struct network_driver *driver,
                               virNetworkObjPtr network)
{

    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0))
        return 0;

    /* Catch all rules to block forwarding to/from bridges */

    if (iptablesAddForwardRejectOut(driver->iptables, AF_INET6,
                                    network->def->bridge) < 0) {
1508 1509 1510
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to block outbound traffic from '%s'"),
                       network->def->bridge);
1511 1512 1513 1514 1515
        goto err1;
    }

    if (iptablesAddForwardRejectIn(driver->iptables, AF_INET6,
                                   network->def->bridge) < 0) {
1516 1517 1518
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to block inbound traffic to '%s'"),
                       network->def->bridge);
1519 1520 1521 1522 1523 1524
        goto err2;
    }

    /* Allow traffic between guests on the same bridge */
    if (iptablesAddForwardAllowCross(driver->iptables, AF_INET6,
                                     network->def->bridge) < 0) {
1525 1526 1527
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to allow cross bridge traffic on '%s'"),
                       network->def->bridge);
1528 1529 1530
        goto err3;
    }

1531 1532 1533
    /* allow DNS over IPv6 */
    if (iptablesAddTcpInput(driver->iptables, AF_INET6,
                            network->def->bridge, 53) < 0) {
1534 1535 1536
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to allow DNS requests from '%s'"),
                       network->def->bridge);
1537 1538 1539 1540 1541
        goto err4;
    }

    if (iptablesAddUdpInput(driver->iptables, AF_INET6,
                            network->def->bridge, 53) < 0) {
1542 1543 1544
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to allow DNS requests from '%s'"),
                       network->def->bridge);
1545 1546 1547
        goto err5;
    }

1548 1549 1550
    return 0;

    /* unwind in reverse order from the point of failure */
1551 1552 1553 1554
err5:
    iptablesRemoveTcpInput(driver->iptables, AF_INET6, network->def->bridge, 53);
err4:
    iptablesRemoveForwardAllowCross(driver->iptables, AF_INET6, network->def->bridge);
1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574
err3:
    iptablesRemoveForwardRejectIn(driver->iptables, AF_INET6, network->def->bridge);
err2:
    iptablesRemoveForwardRejectOut(driver->iptables, AF_INET6, network->def->bridge);
err1:
    return -1;
}

static void
networkRemoveGeneralIp6tablesRules(struct network_driver *driver,
                                  virNetworkObjPtr network)
{
    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0))
        return;

    iptablesRemoveForwardAllowCross(driver->iptables, AF_INET6, network->def->bridge);
    iptablesRemoveForwardRejectIn(driver->iptables, AF_INET6, network->def->bridge);
    iptablesRemoveForwardRejectOut(driver->iptables, AF_INET6, network->def->bridge);
}

1575
static int
1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589
networkAddGeneralIptablesRules(struct network_driver *driver,
                               virNetworkObjPtr network)
{
    int ii;
    virNetworkIpDefPtr ipv4def;

    /* First look for first IPv4 address that has dhcp or tftpboot defined. */
    /* We support dhcp config on 1 IPv4 interface only. */
    for (ii = 0;
         (ipv4def = virNetworkDefGetIpByIndex(network->def, AF_INET, ii));
         ii++) {
        if (ipv4def->nranges || ipv4def->nhosts || ipv4def->tftproot)
            break;
    }
1590 1591

    /* allow DHCP requests through to dnsmasq */
1592

1593 1594
    if (iptablesAddTcpInput(driver->iptables, AF_INET,
                            network->def->bridge, 67) < 0) {
1595 1596 1597
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow DHCP requests from '%s'"),
                       network->def->bridge);
1598 1599 1600
        goto err1;
    }

1601 1602
    if (iptablesAddUdpInput(driver->iptables, AF_INET,
                            network->def->bridge, 67) < 0) {
1603 1604 1605
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow DHCP requests from '%s'"),
                       network->def->bridge);
1606 1607 1608
        goto err2;
    }

1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619
    /* If we are doing local DHCP service on this network, attempt to
     * add a rule that will fixup the checksum of DHCP response
     * packets back to the guests (but report failure without
     * aborting, since not all iptables implementations support it).
     */

    if (ipv4def && (ipv4def->nranges || ipv4def->nhosts) &&
        (iptablesAddOutputFixUdpChecksum(driver->iptables,
                                         network->def->bridge, 68) < 0)) {
        VIR_WARN("Could not add rule to fixup DHCP response checksums "
                 "on network '%s'.", network->def->name);
1620
        VIR_WARN("May need to update iptables package & kernel to support CHECKSUM rule.");
1621 1622
    }

1623
    /* allow DNS requests through to dnsmasq */
1624 1625
    if (iptablesAddTcpInput(driver->iptables, AF_INET,
                            network->def->bridge, 53) < 0) {
1626 1627 1628
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow DNS requests from '%s'"),
                       network->def->bridge);
1629 1630 1631
        goto err3;
    }

1632 1633
    if (iptablesAddUdpInput(driver->iptables, AF_INET,
                            network->def->bridge, 53) < 0) {
1634 1635 1636
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow DNS requests from '%s'"),
                       network->def->bridge);
1637 1638 1639
        goto err4;
    }

1640 1641
    /* allow TFTP requests through to dnsmasq if necessary */
    if (ipv4def && ipv4def->tftproot &&
1642 1643
        iptablesAddUdpInput(driver->iptables, AF_INET,
                            network->def->bridge, 69) < 0) {
1644 1645 1646
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow TFTP requests from '%s'"),
                       network->def->bridge);
1647
        goto err5;
1648 1649
    }

1650 1651
    /* Catch all rules to block forwarding to/from bridges */

1652 1653
    if (iptablesAddForwardRejectOut(driver->iptables, AF_INET,
                                    network->def->bridge) < 0) {
1654 1655 1656
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to block outbound traffic from '%s'"),
                       network->def->bridge);
1657
        goto err6;
1658 1659
    }

1660 1661
    if (iptablesAddForwardRejectIn(driver->iptables, AF_INET,
                                   network->def->bridge) < 0) {
1662 1663 1664
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to block inbound traffic to '%s'"),
                       network->def->bridge);
1665
        goto err7;
1666 1667 1668
    }

    /* Allow traffic between guests on the same bridge */
1669 1670
    if (iptablesAddForwardAllowCross(driver->iptables, AF_INET,
                                     network->def->bridge) < 0) {
1671 1672 1673
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow cross bridge traffic on '%s'"),
                       network->def->bridge);
1674
        goto err8;
1675 1676
    }

1677 1678 1679 1680 1681
    /* add IPv6 general rules, if needed */
    if (networkAddGeneralIp6tablesRules(driver, network) < 0) {
        goto err9;
    }

1682
    return 0;
1683

1684
    /* unwind in reverse order from the point of failure */
1685 1686
err9:
    iptablesRemoveForwardAllowCross(driver->iptables, AF_INET, network->def->bridge);
1687
err8:
1688
    iptablesRemoveForwardRejectIn(driver->iptables, AF_INET, network->def->bridge);
1689
err7:
1690
    iptablesRemoveForwardRejectOut(driver->iptables, AF_INET, network->def->bridge);
1691 1692
err6:
    if (ipv4def && ipv4def->tftproot) {
1693
        iptablesRemoveUdpInput(driver->iptables, AF_INET, network->def->bridge, 69);
1694
    }
1695
err5:
1696
    iptablesRemoveUdpInput(driver->iptables, AF_INET, network->def->bridge, 53);
1697
err4:
1698
    iptablesRemoveTcpInput(driver->iptables, AF_INET, network->def->bridge, 53);
1699
err3:
1700
    iptablesRemoveUdpInput(driver->iptables, AF_INET, network->def->bridge, 67);
1701
err2:
1702
    iptablesRemoveTcpInput(driver->iptables, AF_INET, network->def->bridge, 67);
1703
err1:
1704
    return -1;
1705 1706 1707
}

static void
1708 1709 1710 1711 1712
networkRemoveGeneralIptablesRules(struct network_driver *driver,
                                  virNetworkObjPtr network)
{
    int ii;
    virNetworkIpDefPtr ipv4def;
1713

1714 1715
    networkRemoveGeneralIp6tablesRules(driver, network);

1716 1717 1718 1719 1720
    for (ii = 0;
         (ipv4def = virNetworkDefGetIpByIndex(network->def, AF_INET, ii));
         ii++) {
        if (ipv4def->nranges || ipv4def->nhosts || ipv4def->tftproot)
            break;
1721
    }
1722

1723 1724 1725
    iptablesRemoveForwardAllowCross(driver->iptables, AF_INET, network->def->bridge);
    iptablesRemoveForwardRejectIn(driver->iptables, AF_INET, network->def->bridge);
    iptablesRemoveForwardRejectOut(driver->iptables, AF_INET, network->def->bridge);
1726
    if (ipv4def && ipv4def->tftproot) {
1727
        iptablesRemoveUdpInput(driver->iptables, AF_INET, network->def->bridge, 69);
1728
    }
1729 1730
    iptablesRemoveUdpInput(driver->iptables, AF_INET, network->def->bridge, 53);
    iptablesRemoveTcpInput(driver->iptables, AF_INET, network->def->bridge, 53);
1731 1732 1733 1734
    if (ipv4def && (ipv4def->nranges || ipv4def->nhosts)) {
        iptablesRemoveOutputFixUdpChecksum(driver->iptables,
                                           network->def->bridge, 68);
    }
1735 1736
    iptablesRemoveUdpInput(driver->iptables, AF_INET, network->def->bridge, 67);
    iptablesRemoveTcpInput(driver->iptables, AF_INET, network->def->bridge, 67);
1737 1738
}

1739 1740 1741 1742 1743
static int
networkAddIpSpecificIptablesRules(struct network_driver *driver,
                                  virNetworkObjPtr network,
                                  virNetworkIpDefPtr ipdef)
{
1744 1745 1746
    /* NB: in the case of IPv6, routing rules are added when the
     * forward mode is NAT. This is because IPv6 has no NAT.
     */
1747

1748
    if (network->def->forwardType == VIR_NETWORK_FORWARD_NAT) {
1749
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
1750
            return networkAddMasqueradingIptablesRules(driver, network, ipdef);
1751
        else if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1752 1753 1754 1755
            return networkAddRoutingIptablesRules(driver, network, ipdef);
    } else if (network->def->forwardType == VIR_NETWORK_FORWARD_ROUTE) {
        return networkAddRoutingIptablesRules(driver, network, ipdef);
    }
1756 1757 1758 1759 1760 1761 1762 1763
    return 0;
}

static void
networkRemoveIpSpecificIptablesRules(struct network_driver *driver,
                                     virNetworkObjPtr network,
                                     virNetworkIpDefPtr ipdef)
{
1764
    if (network->def->forwardType == VIR_NETWORK_FORWARD_NAT) {
1765
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
1766
            networkRemoveMasqueradingIptablesRules(driver, network, ipdef);
1767
        else if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1768 1769
            networkRemoveRoutingIptablesRules(driver, network, ipdef);
    } else if (network->def->forwardType == VIR_NETWORK_FORWARD_ROUTE) {
1770
        networkRemoveRoutingIptablesRules(driver, network, ipdef);
1771
    }
1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824
}

/* Add all rules for all ip addresses (and general rules) on a network */
static int
networkAddIptablesRules(struct network_driver *driver,
                        virNetworkObjPtr network)
{
    int ii;
    virNetworkIpDefPtr ipdef;

    /* Add "once per network" rules */
    if (networkAddGeneralIptablesRules(driver, network) < 0)
        return -1;

    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, ii));
         ii++) {
        /* Add address-specific iptables rules */
        if (networkAddIpSpecificIptablesRules(driver, network, ipdef) < 0) {
            goto err;
        }
    }
    return 0;

err:
    /* The final failed call to networkAddIpSpecificIptablesRules will
     * have removed any rules it created, but we need to remove those
     * added for previous IP addresses.
     */
    while ((--ii >= 0) &&
           (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, ii))) {
        networkRemoveIpSpecificIptablesRules(driver, network, ipdef);
    }
    networkRemoveGeneralIptablesRules(driver, network);
    return -1;
}

/* Remove all rules for all ip addresses (and general rules) on a network */
static void
networkRemoveIptablesRules(struct network_driver *driver,
                           virNetworkObjPtr network)
{
    int ii;
    virNetworkIpDefPtr ipdef;

    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, ii));
         ii++) {
        networkRemoveIpSpecificIptablesRules(driver, network, ipdef);
    }
    networkRemoveGeneralIptablesRules(driver, network);
}

1825 1826 1827 1828 1829
static void
networkReloadIptablesRules(struct network_driver *driver)
{
    unsigned int i;

1830
    VIR_INFO("Reloading iptables rules");
1831 1832

    for (i = 0 ; i < driver->networks.count ; i++) {
1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
            ((network->def->forwardType == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forwardType == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forwardType == VIR_NETWORK_FORWARD_ROUTE))) {
            /* Only the three L3 network types that are configured by libvirt
             * need to have iptables rules reloaded.
             */
            networkRemoveIptablesRules(driver, network);
            if (networkAddIptablesRules(driver, network) < 0) {
1845 1846
                /* failed to add but already logged */
            }
1847
        }
1848
        virNetworkObjUnlock(network);
1849 1850 1851
    }
}

1852
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
1853
static int
1854
networkEnableIpForwarding(bool enableIPv4, bool enableIPv6)
1855
{
1856 1857 1858 1859 1860 1861
    int ret = 0;
    if (enableIPv4)
        ret = virFileWriteStr("/proc/sys/net/ipv4/ip_forward", "1\n", 0);
    if (enableIPv6 && ret == 0)
        ret = virFileWriteStr("/proc/sys/net/ipv6/conf/all/forwarding", "1\n", 0);
    return ret;
1862 1863
}

1864 1865
#define SYSCTL_PATH "/proc/sys"

1866 1867
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
1868 1869 1870 1871
{
    char *field = NULL;
    int ret = -1;

1872 1873 1874 1875 1876 1877 1878 1879 1880
    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0)) {
        /* Only set disable_ipv6 if there are no ipv6 addresses defined for
         * the network.
         */
        if (virAsprintf(&field, SYSCTL_PATH "/net/ipv6/conf/%s/disable_ipv6",
                        network->def->bridge) < 0) {
            virReportOOMError();
            goto cleanup;
        }
1881

1882 1883 1884 1885 1886 1887
        if (access(field, W_OK) < 0 && errno == ENOENT) {
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
            ret = 0;
            goto cleanup;
        }
1888

1889 1890 1891 1892 1893 1894 1895
        if (virFileWriteStr(field, "1", 0) < 0) {
            virReportSystemError(errno,
                                 _("cannot write to %s to disable IPv6 on bridge %s"),
                                 field, network->def->bridge);
            goto cleanup;
        }
        VIR_FREE(field);
1896 1897
    }

1898 1899 1900 1901 1902 1903 1904 1905 1906
    /* The rest of the ipv6 sysctl tunables should always be set,
     * whether or not we're using ipv6 on this bridge.
     */

    /* 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",
                    network->def->bridge) < 0) {
1907
        virReportOOMError();
1908 1909 1910
        goto cleanup;
    }

1911
    if (virFileWriteStr(field, "0", 0) < 0) {
1912
        virReportSystemError(errno,
1913 1914 1915 1916 1917
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

1918 1919 1920 1921 1922
    /* 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",
                    network->def->bridge) < 0) {
1923
        virReportOOMError();
1924 1925 1926
        goto cleanup;
    }

1927
    if (virFileWriteStr(field, "0", 0) < 0) {
1928
        virReportSystemError(errno,
1929
                             _("cannot disable %s"), field);
1930 1931 1932 1933 1934 1935 1936 1937 1938
        goto cleanup;
    }

    ret = 0;
cleanup:
    VIR_FREE(field);
    return ret;
}

1939 1940 1941 1942 1943 1944
#define PROC_NET_ROUTE "/proc/net/route"

/* XXX: This function can be a lot more exhaustive, there are certainly
 *      other scenarios where we can ruin host network connectivity.
 * XXX: Using a proper library is preferred over parsing /proc
 */
1945 1946
static int
networkCheckRouteCollision(virNetworkObjPtr network)
1947
{
1948
    int ret = 0, len;
1949 1950 1951 1952 1953
    char *cur, *buf = NULL;
    enum {MAX_ROUTE_SIZE = 1024*64};

    /* Read whole routing table into memory */
    if ((len = virFileReadAll(PROC_NET_ROUTE, MAX_ROUTE_SIZE, &buf)) < 0)
1954
        goto out;
1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972

    /* Dropping the last character shouldn't hurt */
    if (len > 0)
        buf[len-1] = '\0';

    VIR_DEBUG("%s output:\n%s", PROC_NET_ROUTE, buf);

    if (!STRPREFIX (buf, "Iface"))
        goto out;

    /* First line is just headings, skip it */
    cur = strchr(buf, '\n');
    if (cur)
        cur++;

    while (cur) {
        char iface[17], dest[128], mask[128];
        unsigned int addr_val, mask_val;
1973 1974
        virNetworkIpDefPtr ipdef;
        int num, ii;
1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

        /* NUL-terminate the line, so sscanf doesn't go beyond a newline.  */
        char *nl = strchr(cur, '\n');
        if (nl) {
            *nl++ = '\0';
        }

        num = sscanf(cur, "%16s %127s %*s %*s %*s %*s %*s %127s",
                     iface, dest, mask);
        cur = nl;

        if (num != 3) {
            VIR_DEBUG("Failed to parse %s", PROC_NET_ROUTE);
            continue;
        }

        if (virStrToLong_ui(dest, NULL, 16, &addr_val) < 0) {
            VIR_DEBUG("Failed to convert network address %s to uint", dest);
            continue;
        }

        if (virStrToLong_ui(mask, NULL, 16, &mask_val) < 0) {
            VIR_DEBUG("Failed to convert network mask %s to uint", mask);
            continue;
        }

        addr_val &= mask_val;

2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
        for (ii = 0;
             (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, ii));
             ii++) {

            unsigned int net_dest;
            virSocketAddr netmask;

            if (virNetworkIpDefNetmask(ipdef, &netmask) < 0) {
                VIR_WARN("Failed to get netmask of '%s'",
                         network->def->bridge);
                continue;
            }

            net_dest = (ipdef->address.data.inet4.sin_addr.s_addr &
                        netmask.data.inet4.sin_addr.s_addr);

            if ((net_dest == addr_val) &&
                (netmask.data.inet4.sin_addr.s_addr == mask_val)) {
2021 2022 2023
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Network is already in use by interface %s"),
                               iface);
2024 2025 2026
                ret = -1;
                goto out;
            }
2027 2028 2029 2030 2031 2032 2033 2034
        }
    }

out:
    VIR_FREE(buf);
    return ret;
}

2035
static int
D
Daniel P. Berrange 已提交
2036
networkAddAddrToBridge(virNetworkObjPtr network,
2037
                       virNetworkIpDefPtr ipdef)
2038
{
2039 2040 2041
    int prefix = virNetworkIpDefPrefix(ipdef);

    if (prefix < 0) {
2042 2043 2044
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
2045 2046 2047
        return -1;
    }

2048 2049
    if (virNetDevSetIPv4Address(network->def->bridge,
                                &ipdef->address, prefix) < 0)
2050 2051 2052 2053 2054 2055
        return -1;

    return 0;
}

static int
2056
networkStartNetworkVirtual(struct network_driver *driver,
2057 2058
                          virNetworkObjPtr network)
{
2059
    int ii;
2060
    bool v4present = false, v6present = false;
2061 2062
    virErrorPtr save_err = NULL;
    virNetworkIpDefPtr ipdef;
2063
    char *macTapIfName = NULL;
2064

2065 2066
    /* Check to see if any network IP collides with an existing route */
    if (networkCheckRouteCollision(network) < 0)
2067 2068
        return -1;

2069
    /* Create and configure the bridge device */
2070
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
2071 2072
        return -1;

2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084
    if (network->def->mac_specified) {
        /* To set a mac for the bridge, we need to define a dummy tap
         * device, set its mac, then attach it to the bridge. As long
         * as its mac address is lower than any other interface that
         * gets attached, the bridge will always maintain this mac
         * address.
         */
        macTapIfName = networkBridgeDummyNicName(network->def->bridge);
        if (!macTapIfName) {
            virReportOOMError();
            goto err0;
        }
2085
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
2086
                                           &macTapIfName, &network->def->mac,
2087
                                           NULL, NULL, NULL, NULL,
2088
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE) < 0) {
2089 2090 2091 2092 2093
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2094
    /* Set bridge options */
2095 2096 2097 2098

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2099
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2100
                                   network->def->delay * 1000) < 0)
2101
        goto err1;
2102

2103
    if (virNetDevBridgeSetSTP(network->def->bridge,
2104
                              network->def->stp ? true : false) < 0)
2105
        goto err1;
2106

2107 2108 2109 2110
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2111
        goto err1;
2112

2113 2114 2115 2116 2117 2118 2119
    /* Add "once per network" rules */
    if (networkAddIptablesRules(driver, network) < 0)
        goto err1;

    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, ii));
         ii++) {
2120
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2121
            v4present = true;
2122
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2123
            v6present = true;
2124

2125
        /* Add the IP address/netmask to the bridge */
D
Daniel P. Berrange 已提交
2126
        if (networkAddAddrToBridge(network, ipdef) < 0) {
2127
            goto err2;
2128
        }
2129 2130
    }

2131
    /* Bring up the bridge interface */
2132
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2133
        goto err2;
2134

2135
    /* If forwardType != NONE, turn on global IP forwarding */
2136
    if (network->def->forwardType != VIR_NETWORK_FORWARD_NONE &&
2137
        networkEnableIpForwarding(v4present, v6present) < 0) {
2138
        virReportSystemError(errno, "%s",
2139
                             _("failed to enable IP forwarding"));
2140
        goto err3;
2141 2142
    }

2143

2144 2145
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
    if ((v4present || v6present) && networkStartDhcpDaemon(network) < 0)
2146
        goto err3;
2147

2148 2149 2150 2151
    /* start radvd if there are any ipv6 addresses */
    if (v6present && networkStartRadvd(network) < 0)
        goto err4;

2152
    if (virNetDevBandwidthSet(network->def->bridge, network->def->bandwidth) < 0) {
2153 2154 2155
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("cannot set bandwidth limits on %s"),
                       network->def->bridge);
2156 2157 2158
        goto err5;
    }

2159
    VIR_FREE(macTapIfName);
2160 2161 2162

    return 0;

2163
 err5:
2164
    virNetDevBandwidthClear(network->def->bridge);
2165

2166 2167 2168 2169
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2170 2171 2172 2173 2174
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2175 2176 2177
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2178
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2179

2180 2181 2182 2183 2184 2185
 err2:
    if (!save_err)
        save_err = virSaveLastError();
    networkRemoveIptablesRules(driver, network);

 err1:
2186 2187 2188
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2189 2190 2191 2192
    if (macTapIfName) {
        ignore_value(virNetDevTapDelete(macTapIfName));
        VIR_FREE(macTapIfName);
    }
2193 2194

 err0:
2195 2196
    if (!save_err)
        save_err = virSaveLastError();
2197
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2198

2199 2200 2201 2202
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2203 2204 2205
    return -1;
}

2206
static int networkShutdownNetworkVirtual(struct network_driver *driver,
2207 2208
                                        virNetworkObjPtr network)
{
2209
    virNetDevBandwidthClear(network->def->bridge);
2210

2211 2212 2213 2214 2215 2216 2217 2218
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
        if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
            virReportOOMError();
        } else {
2219
            virPidFileDelete(NETWORK_PID_DIR, radvdpidbase);
2220 2221 2222 2223
            VIR_FREE(radvdpidbase);
        }
    }

2224 2225 2226
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2227
    if (network->def->mac_specified) {
2228
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2229 2230 2231
        if (!macTapIfName) {
            virReportOOMError();
        } else {
2232
            ignore_value(virNetDevTapDelete(macTapIfName));
2233 2234 2235 2236
            VIR_FREE(macTapIfName);
        }
    }

2237
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2238

2239 2240
    networkRemoveIptablesRules(driver, network);

2241
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2242

2243
    /* See if its still alive and really really kill it */
2244
    if (network->dnsmasqPid > 0 &&
2245
        (kill(network->dnsmasqPid, 0) == 0))
2246 2247
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2248 2249 2250 2251 2252 2253

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

2254 2255 2256 2257 2258 2259 2260 2261
    return 0;
}

static int
networkStartNetworkExternal(struct network_driver *driver ATTRIBUTE_UNUSED,
                            virNetworkObjPtr network ATTRIBUTE_UNUSED)
{
    /* put anything here that needs to be done each time a network of
2262
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2263 2264 2265 2266 2267 2268 2269 2270 2271 2272
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int networkShutdownNetworkExternal(struct network_driver *driver ATTRIBUTE_UNUSED,
                                        virNetworkObjPtr network ATTRIBUTE_UNUSED)
{
    /* put anything here that needs to be done each time a network of
2273
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
networkStartNetwork(struct network_driver *driver,
                    virNetworkObjPtr network)
{
    int ret = 0;

    if (virNetworkObjIsActive(network)) {
2287 2288
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2289 2290 2291
        return -1;
    }

2292 2293 2294
    if (virNetworkObjSetDefTransient(network, true) < 0)
        return -1;

2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306
    switch (network->def->forwardType) {

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
        ret = networkStartNetworkVirtual(driver, network);
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2307
    case VIR_NETWORK_FORWARD_HOSTDEV:
2308 2309 2310 2311 2312 2313 2314 2315 2316 2317
        ret = networkStartNetworkExternal(driver, network);
        break;
    }

    if (ret < 0)
        return ret;

    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2318
    if ((ret = virNetworkSaveStatus(NETWORK_STATE_DIR, network)) < 0) {
2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366
        goto error;
    }

    VIR_INFO("Starting up network '%s'", network->def->name);
    network->active = 1;

error:
    if (ret < 0) {
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
        networkShutdownNetwork(driver, network);
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

static int networkShutdownNetwork(struct network_driver *driver,
                                        virNetworkObjPtr network)
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

    stateFile = virNetworkConfigFile(NETWORK_STATE_DIR, network->def->name);
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

    switch (network->def->forwardType) {

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
        ret = networkShutdownNetworkVirtual(driver, network);
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2367
    case VIR_NETWORK_FORWARD_HOSTDEV:
2368 2369 2370 2371
        ret = networkShutdownNetworkExternal(driver, network);
        break;
    }

2372 2373 2374 2375 2376 2377 2378 2379
    network->active = 0;

    if (network->newDef) {
        virNetworkDefFree(network->def);
        network->def = network->newDef;
        network->newDef = NULL;
    }

2380
    return ret;
2381 2382 2383
}


2384 2385 2386 2387 2388
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
                                         const unsigned char *uuid) {
    struct network_driver *driver = conn->networkPrivateData;
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2389

2390
    networkDriverLock(driver);
2391
    network = virNetworkFindByUUID(&driver->networks, uuid);
2392
    networkDriverUnlock(driver);
2393
    if (!network) {
2394 2395
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
2396
        goto cleanup;
2397 2398
    }

2399 2400 2401
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

cleanup:
2402 2403
    if (network)
        virNetworkObjUnlock(network);
2404
    return ret;
2405 2406
}

2407 2408 2409 2410 2411 2412
static virNetworkPtr networkLookupByName(virConnectPtr conn,
                                         const char *name) {
    struct network_driver *driver = conn->networkPrivateData;
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2413
    networkDriverLock(driver);
2414
    network = virNetworkFindByName(&driver->networks, name);
2415
    networkDriverUnlock(driver);
2416
    if (!network) {
2417 2418
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2419
        goto cleanup;
2420 2421
    }

2422 2423 2424
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

cleanup:
2425 2426
    if (network)
        virNetworkObjUnlock(network);
2427
    return ret;
2428 2429 2430 2431
}

static virDrvOpenStatus networkOpenNetwork(virConnectPtr conn,
                                           virConnectAuthPtr auth ATTRIBUTE_UNUSED,
2432 2433 2434 2435
                                           unsigned int flags)
{
    virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);

2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448
    if (!driverState)
        return VIR_DRV_OPEN_DECLINED;

    conn->networkPrivateData = driverState;
    return VIR_DRV_OPEN_SUCCESS;
}

static int networkCloseNetwork(virConnectPtr conn) {
    conn->networkPrivateData = NULL;
    return 0;
}

static int networkNumNetworks(virConnectPtr conn) {
2449
    int nactive = 0, i;
2450
    struct network_driver *driver = conn->networkPrivateData;
2451

2452 2453 2454
    networkDriverLock(driver);
    for (i = 0 ; i < driver->networks.count ; i++) {
        virNetworkObjLock(driver->networks.objs[i]);
D
Daniel P. Berrange 已提交
2455
        if (virNetworkObjIsActive(driver->networks.objs[i]))
2456
            nactive++;
2457 2458 2459
        virNetworkObjUnlock(driver->networks.objs[i]);
    }
    networkDriverUnlock(driver);
2460

2461 2462 2463 2464
    return nactive;
}

static int networkListNetworks(virConnectPtr conn, char **const names, int nnames) {
2465
    struct network_driver *driver = conn->networkPrivateData;
2466
    int got = 0, i;
2467

2468
    networkDriverLock(driver);
2469
    for (i = 0 ; i < driver->networks.count && got < nnames ; i++) {
2470
        virNetworkObjLock(driver->networks.objs[i]);
D
Daniel P. Berrange 已提交
2471
        if (virNetworkObjIsActive(driver->networks.objs[i])) {
2472
            if (!(names[got] = strdup(driver->networks.objs[i]->def->name))) {
2473
                virNetworkObjUnlock(driver->networks.objs[i]);
2474
                virReportOOMError();
2475 2476 2477 2478
                goto cleanup;
            }
            got++;
        }
2479
        virNetworkObjUnlock(driver->networks.objs[i]);
2480
    }
2481 2482
    networkDriverUnlock(driver);

2483 2484 2485
    return got;

 cleanup:
2486
    networkDriverUnlock(driver);
2487 2488 2489 2490 2491 2492
    for (i = 0 ; i < got ; i++)
        VIR_FREE(names[i]);
    return -1;
}

static int networkNumDefinedNetworks(virConnectPtr conn) {
2493
    int ninactive = 0, i;
2494
    struct network_driver *driver = conn->networkPrivateData;
2495

2496 2497 2498
    networkDriverLock(driver);
    for (i = 0 ; i < driver->networks.count ; i++) {
        virNetworkObjLock(driver->networks.objs[i]);
D
Daniel P. Berrange 已提交
2499
        if (!virNetworkObjIsActive(driver->networks.objs[i]))
2500
            ninactive++;
2501 2502 2503
        virNetworkObjUnlock(driver->networks.objs[i]);
    }
    networkDriverUnlock(driver);
2504

2505 2506 2507 2508
    return ninactive;
}

static int networkListDefinedNetworks(virConnectPtr conn, char **const names, int nnames) {
2509
    struct network_driver *driver = conn->networkPrivateData;
2510
    int got = 0, i;
2511

2512
    networkDriverLock(driver);
2513
    for (i = 0 ; i < driver->networks.count && got < nnames ; i++) {
2514
        virNetworkObjLock(driver->networks.objs[i]);
D
Daniel P. Berrange 已提交
2515
        if (!virNetworkObjIsActive(driver->networks.objs[i])) {
2516
            if (!(names[got] = strdup(driver->networks.objs[i]->def->name))) {
2517
                virNetworkObjUnlock(driver->networks.objs[i]);
2518
                virReportOOMError();
2519 2520 2521 2522
                goto cleanup;
            }
            got++;
        }
2523
        virNetworkObjUnlock(driver->networks.objs[i]);
2524
    }
2525
    networkDriverUnlock(driver);
2526 2527 2528
    return got;

 cleanup:
2529
    networkDriverUnlock(driver);
2530 2531 2532 2533 2534
    for (i = 0 ; i < got ; i++)
        VIR_FREE(names[i]);
    return -1;
}

2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550
static int
networkListAllNetworks(virConnectPtr conn,
                       virNetworkPtr **nets,
                       unsigned int flags)
{
    struct network_driver *driver = conn->networkPrivateData;
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

    networkDriverLock(driver);
    ret = virNetworkList(conn, driver->networks, nets, flags);
    networkDriverUnlock(driver);

    return ret;
}
2551 2552 2553

static int networkIsActive(virNetworkPtr net)
{
2554
    struct network_driver *driver = net->conn->networkPrivateData;
2555 2556 2557 2558 2559 2560 2561
    virNetworkObjPtr obj;
    int ret = -1;

    networkDriverLock(driver);
    obj = virNetworkFindByUUID(&driver->networks, net->uuid);
    networkDriverUnlock(driver);
    if (!obj) {
2562
        virReportError(VIR_ERR_NO_NETWORK, NULL);
2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574
        goto cleanup;
    }
    ret = virNetworkObjIsActive(obj);

cleanup:
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}

static int networkIsPersistent(virNetworkPtr net)
{
2575
    struct network_driver *driver = net->conn->networkPrivateData;
2576 2577 2578 2579 2580 2581 2582
    virNetworkObjPtr obj;
    int ret = -1;

    networkDriverLock(driver);
    obj = virNetworkFindByUUID(&driver->networks, net->uuid);
    networkDriverUnlock(driver);
    if (!obj) {
2583
        virReportError(VIR_ERR_NO_NETWORK, NULL);
2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594
        goto cleanup;
    }
    ret = obj->persistent;

cleanup:
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}


2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630
static int
networkValidate(virNetworkDefPtr def)
{
    int ii;
    bool vlanUsed, vlanAllowed;

    /* 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.
     */

    vlanAllowed = (def->forwardType == VIR_NETWORK_FORWARD_BRIDGE &&
                   def->virtPortProfile &&
                   def->virtPortProfile->virtPortType == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH);

    vlanUsed = def->vlan.nTags > 0;
    for (ii = 0; ii < def->nPortGroups && !(vlanUsed && vlanAllowed); ii++) {
        if (def->portGroups[ii].vlan.nTags > 0)
            vlanUsed = true;
        if (def->forwardType == VIR_NETWORK_FORWARD_BRIDGE &&
            def->portGroups[ii].virtPortProfile &&
            (def->portGroups[ii].virtPortProfile->virtPortType
             == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH)) {
            vlanAllowed = true;
        }
    }
    if (vlanUsed && !vlanAllowed) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
    return 0;
}

2631
static virNetworkPtr networkCreate(virConnectPtr conn, const char *xml) {
2632
    struct network_driver *driver = conn->networkPrivateData;
2633
    virNetworkDefPtr def;
2634
    virNetworkObjPtr network = NULL;
2635
    virNetworkPtr ret = NULL;
2636

2637 2638
    networkDriverLock(driver);

2639
    if (!(def = virNetworkDefParseString(xml)))
2640
        goto cleanup;
2641

2642 2643 2644
    if (virNetworkObjIsDuplicate(&driver->networks, def, 1) < 0)
        goto cleanup;

2645 2646 2647 2648 2649 2650
    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
    if (def->forwardType == VIR_NETWORK_FORWARD_NONE ||
        def->forwardType == VIR_NETWORK_FORWARD_NAT ||
        def->forwardType == VIR_NETWORK_FORWARD_ROUTE) {
2651

2652 2653 2654 2655 2656
        if (virNetworkSetBridgeName(&driver->networks, def, 1))
            goto cleanup;

        virNetworkSetBridgeMacAddr(def);
    }
2657

2658 2659 2660
    if (networkValidate(def) < 0)
       goto cleanup;

2661 2662 2663 2664
    /* NB: "live" is false because this transient network hasn't yet
     * been started
     */
    if (!(network = virNetworkAssignDef(&driver->networks, def, false)))
2665 2666
        goto cleanup;
    def = NULL;
2667

2668
    if (networkStartNetwork(driver, network) < 0) {
2669 2670
        virNetworkRemoveInactive(&driver->networks,
                                 network);
2671
        network = NULL;
2672
        goto cleanup;
2673 2674
    }

2675
    VIR_INFO("Creating network '%s'", network->def->name);
2676 2677 2678 2679
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

cleanup:
    virNetworkDefFree(def);
2680 2681 2682
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
2683
    return ret;
2684 2685 2686
}

static virNetworkPtr networkDefine(virConnectPtr conn, const char *xml) {
2687
    struct network_driver *driver = conn->networkPrivateData;
2688
    virNetworkIpDefPtr ipdef, ipv4def = NULL;
2689
    virNetworkDefPtr def;
2690
    bool freeDef = true;
2691
    virNetworkObjPtr network = NULL;
2692
    virNetworkPtr ret = NULL;
2693
    int ii;
2694
    dnsmasqContext* dctx = NULL;
2695

2696 2697
    networkDriverLock(driver);

2698
    if (!(def = virNetworkDefParseString(xml)))
2699
        goto cleanup;
2700

E
Eric Blake 已提交
2701
    if (virNetworkObjIsDuplicate(&driver->networks, def, 0) < 0)
2702 2703
        goto cleanup;

2704 2705 2706 2707 2708 2709
    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
    if (def->forwardType == VIR_NETWORK_FORWARD_NONE ||
        def->forwardType == VIR_NETWORK_FORWARD_NAT ||
        def->forwardType == VIR_NETWORK_FORWARD_ROUTE) {
2710

2711 2712 2713 2714 2715
        if (virNetworkSetBridgeName(&driver->networks, def, 1))
            goto cleanup;

        virNetworkSetBridgeMacAddr(def);
    }
2716

2717
    /* We only support dhcp on one IPv4 address per defined network */
2718
    for (ii = 0;
2719
         (ipdef = virNetworkDefGetIpByIndex(def, AF_UNSPEC, ii));
2720
         ii++) {
2721
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
2722 2723
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
2724 2725 2726 2727
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                                   _("Multiple dhcp sections found. "
                                     "dhcp is supported only for a "
                                     "single IPv4 address on each network"));
2728 2729 2730 2731 2732 2733 2734
                    goto cleanup;
                } else {
                    ipv4def = ipdef;
                }
            }
        }
    }
2735

2736 2737 2738
    if (networkValidate(def) < 0)
       goto cleanup;

2739
    if (!(network = virNetworkAssignDef(&driver->networks, def, false)))
2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750
        goto cleanup;
    freeDef = false;

    network->persistent = 1;

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
        virNetworkRemoveInactive(&driver->networks, network);
        network = NULL;
        goto cleanup;
    }

2751
    if (ipv4def) {
2752
        dctx = dnsmasqContextNew(def->name, DNSMASQ_STATE_DIR);
2753
        if (dctx == NULL ||
2754
            networkBuildDnsmasqHostsfile(dctx, ipv4def, def->dns) < 0 ||
2755
            dnsmasqSave(dctx) < 0)
2756 2757 2758
            goto cleanup;
    }

2759 2760
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
2761 2762

cleanup:
2763 2764
    if (freeDef)
       virNetworkDefFree(def);
2765
    dnsmasqContextFree(dctx);
2766 2767 2768
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
2769
    return ret;
2770 2771 2772
}

static int networkUndefine(virNetworkPtr net) {
2773
    struct network_driver *driver = net->conn->networkPrivateData;
2774
    virNetworkObjPtr network;
2775 2776
    virNetworkIpDefPtr ipdef;
    bool dhcp_present = false, v6present = false;
2777
    int ret = -1, ii;
2778

2779 2780
    networkDriverLock(driver);

2781
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
2782
    if (!network) {
2783 2784
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
2785
        goto cleanup;
2786 2787
    }

D
Daniel P. Berrange 已提交
2788
    if (virNetworkObjIsActive(network)) {
2789 2790
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is still active"));
2791
        goto cleanup;
2792 2793
    }

2794
    if (virNetworkDeleteConfig(driver->networkConfigDir,
2795 2796
                               driver->networkAutostartDir,
                               network) < 0)
2797
        goto cleanup;
2798

2799 2800
    /* we only support dhcp on one IPv4 address per defined network */
    for (ii = 0;
2801
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, ii));
2802
         ii++) {
2803
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
2804 2805
            if (ipdef->nranges || ipdef->nhosts)
                dhcp_present = true;
2806
        } else if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6)) {
2807 2808
            v6present = true;
        }
2809
    }
2810 2811

    if (dhcp_present) {
2812
        char *leasefile;
2813 2814 2815 2816 2817 2818
        dnsmasqContext *dctx = dnsmasqContextNew(network->def->name, DNSMASQ_STATE_DIR);
        if (dctx == NULL)
            goto cleanup;

        dnsmasqDelete(dctx);
        dnsmasqContextFree(dctx);
2819 2820 2821 2822 2823 2824

        leasefile = networkDnsmasqLeaseFileName(network->def->name);
        if (!leasefile)
            goto cleanup;
        unlink(leasefile);
        VIR_FREE(leasefile);
2825 2826
    }

2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842
    if (v6present) {
        char *configfile = networkRadvdConfigFileName(network->def->name);

        if (!configfile) {
            virReportOOMError();
            goto cleanup;
        }
        unlink(configfile);
        VIR_FREE(configfile);

        char *radvdpidbase = networkRadvdPidfileBasename(network->def->name);

        if (!(radvdpidbase)) {
            virReportOOMError();
            goto cleanup;
        }
2843
        virPidFileDelete(NETWORK_PID_DIR, radvdpidbase);
2844 2845 2846 2847
        VIR_FREE(radvdpidbase);

    }

2848
    VIR_INFO("Undefining network '%s'", network->def->name);
2849 2850
    virNetworkRemoveInactive(&driver->networks,
                             network);
2851
    network = NULL;
2852
    ret = 0;
2853

2854
cleanup:
2855 2856 2857
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
2858
    return ret;
2859 2860
}

2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network = NULL;
    int isActive, ret = -1;

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

    networkDriverLock(driver);

    network = virNetworkFindByUUID(&driver->networks, net->uuid);
    if (!network) {
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
        goto cleanup;
    }

    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
    */
    isActive = virNetworkObjIsActive(network);
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE
                   | VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

    /* update the network config in memory/on disk */
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0)
        goto cleanup;

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

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

        } else if (section == VIR_NETWORK_SECTION_IP_DHCP_HOST ||
                   section == VIR_NETWORK_SECTION_DNS_HOST ||
                   section == VIR_NETWORK_SECTION_DNS_TXT ||
                   section == VIR_NETWORK_SECTION_DNS_SRV) {
            /* these sections only change things in config files, so we
             * can just update the config files and send SIGHUP to
             * dnsmasq.
             */
            if (networkRefreshDhcpDaemon(network) < 0)
                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.
             */
            if (networkRefreshRadvd(network) < 0)
                goto cleanup;
        }

2948 2949 2950 2951 2952 2953
        if ((section == VIR_NETWORK_SECTION_IP ||
             section == VIR_NETWORK_SECTION_FORWARD ||
             section == VIR_NETWORK_SECTION_FORWARD_INTERFACE) &&
           (network->def->forwardType == VIR_NETWORK_FORWARD_NONE ||
            network->def->forwardType == VIR_NETWORK_FORWARD_NAT ||
            network->def->forwardType == VIR_NETWORK_FORWARD_ROUTE)) {
2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972
            /* these could affect the iptables rules */
            networkRemoveIptablesRules(driver, network);
            if (networkAddIptablesRules(driver, network) < 0)
                goto cleanup;

        }

        /* save current network state to disk */
        if ((ret = virNetworkSaveStatus(NETWORK_STATE_DIR, network)) < 0)
            goto cleanup;
    }
    ret = 0;
cleanup:
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
    return ret;
}

2973
static int networkStart(virNetworkPtr net) {
2974 2975 2976
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
    int ret = -1;
2977

2978
    networkDriverLock(driver);
2979
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
2980

2981
    if (!network) {
2982 2983
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
2984
        goto cleanup;
2985 2986
    }

2987
    ret = networkStartNetwork(driver, network);
2988 2989

cleanup:
2990 2991
    if (network)
        virNetworkObjUnlock(network);
2992
    networkDriverUnlock(driver);
2993
    return ret;
2994 2995 2996
}

static int networkDestroy(virNetworkPtr net) {
2997 2998 2999
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
    int ret = -1;
3000

3001
    networkDriverLock(driver);
3002
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3003

3004
    if (!network) {
3005 3006
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3007
        goto cleanup;
3008 3009
    }

D
Daniel P. Berrange 已提交
3010
    if (!virNetworkObjIsActive(network)) {
3011 3012
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is not active"));
3013 3014 3015
        goto cleanup;
    }

3016
    ret = networkShutdownNetwork(driver, network);
3017
    if (!network->persistent) {
3018 3019 3020 3021
        virNetworkRemoveInactive(&driver->networks,
                                 network);
        network = NULL;
    }
3022

3023
cleanup:
3024 3025
    if (network)
        virNetworkObjUnlock(network);
3026
    networkDriverUnlock(driver);
3027 3028 3029
    return ret;
}

3030
static char *networkGetXMLDesc(virNetworkPtr net,
3031
                               unsigned int flags)
3032
{
3033 3034
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
3035
    virNetworkDefPtr def;
3036
    char *ret = NULL;
3037

3038
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3039

3040
    networkDriverLock(driver);
3041
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3042 3043
    networkDriverUnlock(driver);

3044
    if (!network) {
3045 3046
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3047
        goto cleanup;
3048 3049
    }

3050 3051 3052 3053 3054 3055
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3056 3057

cleanup:
3058 3059
    if (network)
        virNetworkObjUnlock(network);
3060
    return ret;
3061 3062 3063
}

static char *networkGetBridgeName(virNetworkPtr net) {
3064 3065 3066 3067
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
    char *bridge = NULL;

3068
    networkDriverLock(driver);
3069
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3070 3071
    networkDriverUnlock(driver);

3072
    if (!network) {
3073 3074
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching id"));
3075
        goto cleanup;
3076 3077
    }

3078
    if (!(network->def->bridge)) {
3079 3080 3081
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3082 3083 3084
        goto cleanup;
    }

3085
    bridge = strdup(network->def->bridge);
3086
    if (!bridge)
3087
        virReportOOMError();
3088 3089

cleanup:
3090 3091
    if (network)
        virNetworkObjUnlock(network);
3092 3093 3094 3095 3096
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
                             int *autostart) {
3097 3098 3099
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
    int ret = -1;
3100

3101
    networkDriverLock(driver);
3102
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3103
    networkDriverUnlock(driver);
3104
    if (!network) {
3105 3106
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3107
        goto cleanup;
3108 3109 3110
    }

    *autostart = network->autostart;
3111
    ret = 0;
3112

3113
cleanup:
3114 3115
    if (network)
        virNetworkObjUnlock(network);
3116
    return ret;
3117 3118 3119
}

static int networkSetAutostart(virNetworkPtr net,
3120
                               int autostart) {
3121 3122
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
3123
    char *configFile = NULL, *autostartLink = NULL;
3124
    int ret = -1;
3125

3126
    networkDriverLock(driver);
3127
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3128

3129
    if (!network) {
3130 3131
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3132
        goto cleanup;
3133 3134
    }

3135
    if (!network->persistent) {
3136 3137
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3138 3139 3140
        goto cleanup;
    }

3141 3142
    autostart = (autostart != 0);

3143
    if (network->autostart != autostart) {
3144
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3145
            goto cleanup;
3146
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3147 3148
            goto cleanup;

3149
        if (autostart) {
3150
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3151
                virReportSystemError(errno,
3152 3153
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3154 3155
                goto cleanup;
            }
3156

3157
            if (symlink(configFile, autostartLink) < 0) {
3158
                virReportSystemError(errno,
3159
                                     _("Failed to create symlink '%s' to '%s'"),
3160
                                     autostartLink, configFile);
3161 3162 3163
                goto cleanup;
            }
        } else {
3164
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3165
                virReportSystemError(errno,
3166
                                     _("Failed to delete symlink '%s'"),
3167
                                     autostartLink);
3168 3169
                goto cleanup;
            }
3170 3171
        }

3172
        network->autostart = autostart;
3173
    }
3174
    ret = 0;
3175

3176
cleanup:
3177 3178
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3179 3180
    if (network)
        virNetworkObjUnlock(network);
3181
    networkDriverUnlock(driver);
3182
    return ret;
3183 3184 3185 3186 3187
}


static virNetworkDriver networkDriver = {
    "Network",
3188 3189 3190 3191 3192 3193
    .open = networkOpenNetwork, /* 0.2.0 */
    .close = networkCloseNetwork, /* 0.2.0 */
    .numOfNetworks = networkNumNetworks, /* 0.2.0 */
    .listNetworks = networkListNetworks, /* 0.2.0 */
    .numOfDefinedNetworks = networkNumDefinedNetworks, /* 0.2.0 */
    .listDefinedNetworks = networkListDefinedNetworks, /* 0.2.0 */
3194
    .listAllNetworks = networkListAllNetworks, /* 0.10.2 */
3195 3196 3197 3198 3199
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
    .networkCreateXML = networkCreate, /* 0.2.0 */
    .networkDefineXML = networkDefine, /* 0.2.0 */
    .networkUndefine = networkUndefine, /* 0.2.0 */
3200
    .networkUpdate = networkUpdate, /* 0.10.2 */
3201 3202 3203 3204 3205 3206 3207 3208
    .networkCreate = networkStart, /* 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 */
3209 3210 3211
};

static virStateDriver networkStateDriver = {
3212
    "Network",
3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223
    networkStartup,
    networkShutdown,
    networkReload,
    networkActive,
};

int networkRegister(void) {
    virRegisterNetworkDriver(&networkDriver);
    virRegisterStateDriver(&networkStateDriver);
    return 0;
}
3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234

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

/* 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.
 */

3235 3236 3237 3238 3239 3240 3241 3242 3243
/* 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) {
    unsigned int num_virt_fns = 0;
    char **vfname = NULL;
3244
    struct pci_config_address **virt_fns;
3245 3246 3247
    int ret = -1, ii = 0;

    if ((virNetDevGetVirtualFunctions(netdef->forwardPfs->dev,
3248
                                      &vfname, &virt_fns, &num_virt_fns)) < 0) {
3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Could not get Virtual functions on %s"),
                       netdef->forwardPfs->dev);
        goto finish;
    }

    if (num_virt_fns == 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("No Vf's present on SRIOV PF %s"),
                       netdef->forwardPfs->dev);
       goto finish;
    }

    if ((VIR_ALLOC_N(netdef->forwardIfs, num_virt_fns)) < 0) {
        virReportOOMError();
        goto finish;
    }

    netdef->nForwardIfs = num_virt_fns;

    for (ii = 0; ii < netdef->nForwardIfs; ii++) {
3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283
        if ((netdef->forwardType == VIR_NETWORK_FORWARD_BRIDGE) ||
            (netdef->forwardType == VIR_NETWORK_FORWARD_PRIVATE) ||
            (netdef->forwardType == VIR_NETWORK_FORWARD_VEPA) ||
            (netdef->forwardType == VIR_NETWORK_FORWARD_PASSTHROUGH)) {
            netdef->forwardIfs[ii].type = VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV;
            if(vfname[ii]) {
                netdef->forwardIfs[ii].device.dev = strdup(vfname[ii]);
                if (!netdef->forwardIfs[ii].device.dev) {
                    virReportOOMError();
                    goto finish;
                }
            }
            else {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
3284
                               _("Direct mode types require interface names"));
3285 3286
                goto finish;
            }
3287
        }
3288 3289 3290 3291 3292 3293 3294 3295
        else if (netdef->forwardType == VIR_NETWORK_FORWARD_HOSTDEV) {
            /* VF's are always PCI devices */
            netdef->forwardIfs[ii].type = VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI;
            netdef->forwardIfs[ii].device.pci.domain = virt_fns[ii]->domain;
            netdef->forwardIfs[ii].device.pci.bus = virt_fns[ii]->bus;
            netdef->forwardIfs[ii].device.pci.slot = virt_fns[ii]->slot;
            netdef->forwardIfs[ii].device.pci.function = virt_fns[ii]->function;
        }
3296 3297 3298 3299
    }

    ret = 0;
finish:
3300
    for (ii = 0; ii < num_virt_fns; ii++) {
3301
        VIR_FREE(vfname[ii]);
3302 3303
        VIR_FREE(virt_fns[ii]);
    }
3304
    VIR_FREE(vfname);
3305
    VIR_FREE(virt_fns);
3306 3307 3308
    return ret;
}

3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323
/* networkAllocateActualDevice:
 * @iface: the original NetDef from the domain
 *
 * Looks up the network reference by iface, allocates a physical
 * device from that network (if appropriate), and returns with the
 * virDomainActualNetDef filled in accordingly. If there are no
 * changes to be made in the netdef, then just leave the actualdef
 * empty.
 *
 * Returns 0 on success, -1 on failure.
 */
int
networkAllocateActualDevice(virDomainNetDefPtr iface)
{
    struct network_driver *driver = driverState;
3324 3325 3326 3327 3328 3329
    enum virDomainNetType actualType = iface->type;
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3330
    virNetworkForwardIfDefPtr dev = NULL;
3331
    int ii;
3332 3333
    int ret = -1;

3334 3335 3336 3337
    /* it's handy to have this initialized if we skip directly to validate */
    if (iface->vlan.nTags > 0)
        vlan = &iface->vlan;

3338
    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3339
        goto validate;
3340 3341 3342 3343 3344 3345 3346 3347

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

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
3348 3349 3350
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3351
        goto error;
3352 3353
    }
    netdef = network->def;
3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368

    /* 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.
    */
    portgroup = virPortGroupFindByName(netdef, iface->data.network.portgroup);

    /* If there is already interface-specific bandwidth, just use that
     * (already in NetDef). Otherwise, if there is bandwidth info in
     * the portgroup, fill that into the ActualDef.
     */
    if (portgroup && !iface->bandwidth) {
        if (!iface->data.network.actual
            && (VIR_ALLOC(iface->data.network.actual) < 0)) {
            virReportOOMError();
3369
            goto error;
3370 3371
        }

3372 3373
        if (virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                   portgroup->bandwidth) < 0)
3374
            goto error;
3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387
    }

    if ((netdef->forwardType == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forwardType == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forwardType == VIR_NETWORK_FORWARD_ROUTE)) {
        /* for these forward types, the actual net type really *is*
         *NETWORK; we just keep the info from the portgroup in
         * iface->data.network.actual
        */
        if (iface->data.network.actual)
            iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
    } else if ((netdef->forwardType == VIR_NETWORK_FORWARD_BRIDGE) &&
               netdef->bridge) {
3388 3389 3390 3391 3392

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

3393 3394
        if (!iface->data.network.actual
            && (VIR_ALLOC(iface->data.network.actual) < 0)) {
3395
            virReportOOMError();
3396
            goto error;
3397 3398
        }

3399
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
3400 3401 3402
        iface->data.network.actual->data.bridge.brname = strdup(netdef->bridge);
        if (!iface->data.network.actual->data.bridge.brname) {
            virReportOOMError();
3403
            goto error;
3404 3405
        }

3406 3407 3408 3409 3410 3411 3412 3413
        /* merge virtualports from interface, network, and portgroup to
         * arrive at actual virtualport to use
         */
        if (virNetDevVPortProfileMerge3(&iface->data.network.actual->virtPortProfile,
                                        iface->virtPortProfile,
                                        netdef->virtPortProfile,
                                        portgroup
                                        ? portgroup->virtPortProfile : NULL) < 0) {
3414
            goto error;
3415 3416 3417 3418 3419 3420 3421 3422 3423 3424
        }
        virtport = iface->data.network.actual->virtPortProfile;
        if (virtport) {
            /* only type='openvswitch' is allowed for bridges */
            if (virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("<virtualport type='%s'> not supported for network "
                                 "'%s' which uses a bridge device"),
                               virNetDevVPortTypeToString(virtport->virtPortType),
                               netdef->name);
3425
                goto error;
3426 3427 3428
            }
        }

3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489
    } else if (netdef->forwardType == VIR_NETWORK_FORWARD_HOSTDEV) {

        if (!iface->data.network.actual
            && (VIR_ALLOC(iface->data.network.actual) < 0)) {
            virReportOOMError();
            goto error;
        }

        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
        if (netdef->nForwardPfs > 0 && netdef->nForwardIfs <= 0 &&
            networkCreateInterfacePool(netdef) < 0) {
            goto error;
        }

        /* pick first dev with 0 connections */
        for (ii = 0; ii < netdef->nForwardIfs; ii++) {
            if (netdef->forwardIfs[ii].connections == 0) {
                dev = &netdef->forwardIfs[ii];
                break;
            }
        }
        if (!dev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' requires exclusive access "
                             "to interfaces, but none are available"),
                           netdef->name);
            goto error;
        }
        iface->data.network.actual->data.hostdev.def.parent.type = VIR_DOMAIN_DEVICE_NET;
        iface->data.network.actual->data.hostdev.def.parent.data.net = iface;
        iface->data.network.actual->data.hostdev.def.info = &iface->info;
        iface->data.network.actual->data.hostdev.def.mode = VIR_DOMAIN_HOSTDEV_MODE_SUBSYS;
        iface->data.network.actual->data.hostdev.def.managed = netdef->managed;
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci = dev->device.pci;

        /* merge virtualports from interface, network, and portgroup to
         * arrive at actual virtualport to use
         */
        if (virNetDevVPortProfileMerge3(&iface->data.network.actual->virtPortProfile,
                                        iface->virtPortProfile,
                                        netdef->virtPortProfile,
                                        portgroup
                                        ? portgroup->virtPortProfile : NULL) < 0) {
            goto error;
        }
        virtport = iface->data.network.actual->virtPortProfile;
        if (virtport) {
            /* make sure type is supported for hostdev connections */
            if (virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBG &&
                virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBH) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("<virtualport type='%s'> not supported for network "
                                 "'%s' which uses an SR-IOV Virtual Function "
                                 "via PCI passthrough"),
                               virNetDevVPortTypeToString(virtport->virtPortType),
                               netdef->name);
                goto error;
            }
        }

3490 3491 3492 3493 3494 3495 3496 3497 3498
    } else if ((netdef->forwardType == VIR_NETWORK_FORWARD_BRIDGE) ||
               (netdef->forwardType == VIR_NETWORK_FORWARD_PRIVATE) ||
               (netdef->forwardType == VIR_NETWORK_FORWARD_VEPA) ||
               (netdef->forwardType == VIR_NETWORK_FORWARD_PASSTHROUGH)) {

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

3499 3500
        if (!iface->data.network.actual
            && (VIR_ALLOC(iface->data.network.actual) < 0)) {
3501
            virReportOOMError();
3502
            goto error;
3503 3504 3505
        }

        /* Set type=direct and appropriate <source mode='xxx'/> */
3506
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
3507 3508
        switch (netdef->forwardType) {
        case VIR_NETWORK_FORWARD_BRIDGE:
3509
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
3510 3511
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
3512
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
3513 3514
            break;
        case VIR_NETWORK_FORWARD_VEPA:
3515
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
3516 3517
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
3518
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
3519 3520 3521
            break;
        }

3522 3523 3524 3525 3526 3527 3528 3529
        /* merge virtualports from interface, network, and portgroup to
         * arrive at actual virtualport to use
         */
        if (virNetDevVPortProfileMerge3(&iface->data.network.actual->virtPortProfile,
                                        iface->virtPortProfile,
                                        netdef->virtPortProfile,
                                        portgroup
                                        ? portgroup->virtPortProfile : NULL) < 0) {
3530
            goto error;
3531
        }
3532
        virtport = iface->data.network.actual->virtPortProfile;
3533
        if (virtport) {
3534 3535 3536 3537 3538 3539 3540 3541
            /* make sure type is supported for macvtap connections */
            if (virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBG &&
                virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBH) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("<virtualport type='%s'> not supported for network "
                                 "'%s' which uses a macvtap device"),
                               virNetDevVPortTypeToString(virtport->virtPortType),
                               netdef->name);
3542
                goto error;
3543 3544
            }
        }
3545

3546 3547 3548
        /* 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).
         */
3549
        if ((netdef->nForwardIfs <= 0) && (netdef->nForwardPfs <= 0)) {
3550 3551 3552 3553
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
3554
            goto error;
3555 3556 3557
        } else {
            /* pick an interface from the pool */

3558 3559 3560 3561 3562
            if (netdef->nForwardPfs > 0 && netdef->nForwardIfs == 0 &&
                networkCreateInterfacePool(netdef) < 0) {
                goto error;
            }

3563 3564 3565 3566 3567
            /* 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.
3568
             */
3569 3570 3571 3572 3573
            if ((netdef->forwardType == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forwardType == VIR_NETWORK_FORWARD_PRIVATE) &&
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
3574

3575
                /* pick first dev with 0 connections */
3576
                for (ii = 0; ii < netdef->nForwardIfs; ii++) {
3577
                    if (netdef->forwardIfs[ii].connections == 0) {
3578 3579 3580 3581 3582 3583 3584 3585
                        dev = &netdef->forwardIfs[ii];
                        break;
                    }
                }
            } else {
                /* pick least used dev */
                dev = &netdef->forwardIfs[0];
                for (ii = 1; ii < netdef->nForwardIfs; ii++) {
3586
                    if (netdef->forwardIfs[ii].connections < dev->connections)
3587 3588 3589 3590 3591
                        dev = &netdef->forwardIfs[ii];
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
3592 3593 3594 3595
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
3596
                goto error;
3597
            }
3598
            iface->data.network.actual->data.direct.linkdev = strdup(dev->device.dev);
3599 3600
            if (!iface->data.network.actual->data.direct.linkdev) {
                virReportOOMError();
3601
                goto error;
3602 3603 3604 3605
            }
        }
    }

3606
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
3607
        goto error;
3608

3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651
    /* copy appropriate vlan info to actualNet */
    if (iface->vlan.nTags > 0)
        vlan = &iface->vlan;
    else if (portgroup && portgroup->vlan.nTags > 0)
        vlan = &portgroup->vlan;
    else if (netdef && netdef->vlan.nTags > 0)
        vlan = &netdef->vlan;

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

validate:
    /* 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.
     */

    if (vlan) {
        /* vlan configuration via libvirt is only supported for
         * PCI Passthrough SR-IOV devices and openvswitch bridges.
         * otherwise log an error and fail
         */
        if (!(actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV ||
              (actualType == VIR_DOMAIN_NET_TYPE_BRIDGE &&
               virtport && virtport->virtPortType
               == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH))) {
            if (netdef) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("an interface connecting to network '%s' "
                                 "is requesting a vlan tag, but that is not "
                                 "supported for this type of network"),
                               netdef->name);
            } else {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("an interface of type '%s' "
                                 "is requesting a vlan tag, but that is not "
                                 "supported for this type of connection"),
                               virDomainNetTypeToString(iface->type));
            }
            goto error;
        }
    }

3652 3653
    if (dev) {
        /* we are now assured of success, so mark the allocation */
3654
        dev->connections++;
3655 3656 3657 3658 3659 3660 3661 3662 3663
        if (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV) {
            VIR_DEBUG("Using physical device %s, %d connections",
                      dev->device.dev, dev->connections);
        } else {
            VIR_DEBUG("Using physical device %04x:%02x:%02x.%x, connections %d",
                      dev->device.pci.domain, dev->device.pci.bus,
                      dev->device.pci.slot, dev->device.pci.function,
                      dev->connections);
        }
3664
    }
3665

3666 3667 3668 3669 3670
    if (netdef) {
        netdef->connections++;
        VIR_DEBUG("Using network %s, %d connections",
                  netdef->name, netdef->connections);
    }
3671
    ret = 0;
3672

3673 3674 3675
cleanup:
    if (network)
        virNetworkObjUnlock(network);
3676 3677 3678 3679
    return ret;

error:
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
3680 3681 3682
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
3683
    goto cleanup;
3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699
}

/* networkNotifyActualDevice:
 * @iface:  the domain's NetDef with an "actual" device already filled in.
 *
 * Called to notify the network driver when libvirtd is restarted and
 * finds an already running domain. If appropriate it will force an
 * allocation of the actual->direct.linkdev to get everything back in
 * order.
 *
 * Returns 0 on success, -1 on failure.
 */
int
networkNotifyActualDevice(virDomainNetDefPtr iface)
{
    struct network_driver *driver = driverState;
3700
    enum virDomainNetType actualType = virDomainNetGetActualType(iface);
3701 3702
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
3703 3704
    virNetworkForwardIfDefPtr dev = NULL;
    int ii, ret = -1;
3705 3706 3707 3708 3709 3710 3711 3712

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
       return 0;

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
3713 3714 3715
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3716 3717 3718 3719 3720
        goto error;
    }
    netdef = network->def;

    if (!iface->data.network.actual ||
3721 3722
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
3723 3724
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
3725 3726
    }

3727 3728
    if (netdef->nForwardPfs > 0 && netdef->nForwardIfs == 0 &&
        networkCreateInterfacePool(netdef) < 0) {
3729
        goto error;
3730
    }
3731
    if (netdef->nForwardIfs == 0) {
3732
        virReportError(VIR_ERR_INTERNAL_ERROR,
3733 3734
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
3735
                       netdef->name);
3736
        goto error;
3737
    }
3738

3739 3740
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
3741

3742 3743 3744 3745 3746 3747 3748 3749 3750
        actualDev = virDomainNetGetActualDirectDev(iface);
        if (!actualDev) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("the interface uses a direct mode, "
                             "but has no source dev"));
            goto error;
        }

        /* find the matching interface and increment its connections */
3751
        for (ii = 0; ii < netdef->nForwardIfs; ii++) {
3752 3753 3754
            if (netdef->forwardIfs[ii].type
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
                STREQ(actualDev, netdef->forwardIfs[ii].device.dev)) {
3755 3756 3757 3758 3759 3760
                dev = &netdef->forwardIfs[ii];
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
3761
            virReportError(VIR_ERR_INTERNAL_ERROR,
3762 3763
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
3764
                           netdef->name, actualDev);
3765
            goto error;
3766 3767
        }

3768
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
3769 3770
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
3771
         */
3772
        if ((dev->connections > 0) &&
3773 3774
            ((netdef->forwardType == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forwardType == VIR_NETWORK_FORWARD_PRIVATE) &&
3775 3776
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
3777
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
3778
            virReportError(VIR_ERR_INTERNAL_ERROR,
3779 3780
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
3781
                           netdef->name, actualDev);
3782
            goto error;
3783
        }
3784

3785
        /* we are now assured of success, so mark the allocation */
3786
        dev->connections++;
3787
        VIR_DEBUG("Using physical device %s, connections %d",
3788
                  dev->device.dev, dev->connections);
3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845

    }  else /* if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) */ {
        virDomainHostdevDefPtr hostdev;

        hostdev = virDomainNetGetActualHostdev(iface);
        if (!hostdev) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("the interface uses a hostdev mode, "
                             "but has no hostdev"));
            goto error;
        }

        /* find the matching interface and increment its connections */
        for (ii = 0; ii < netdef->nForwardIfs; ii++) {
            if (netdef->forwardIfs[ii].type
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
                (virDevicePCIAddressEqual(hostdev->source.subsys.u.pci,
                                          netdef->forwardIfs[ii].device.pci) == 0)) {
                dev = &netdef->forwardIfs[ii];
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have "
                             "PCI device %04x:%02x:%02x.%x in use by domain"),
                           netdef->name,
                           hostdev->source.subsys.u.pci.domain,
                           hostdev->source.subsys.u.pci.bus,
                           hostdev->source.subsys.u.pci.slot,
                           hostdev->source.subsys.u.pci.function);
                goto error;
        }

        /* 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) &&
            netdef->forwardType == VIR_NETWORK_FORWARD_HOSTDEV) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' claims the PCI device at "
                             "domain=%d bus=%d slot=%d function=%d "
                             "is already in use by a different domain"),
                           netdef->name,
                           dev->device.pci.domain, dev->device.pci.bus,
                           dev->device.pci.slot, dev->device.pci.function);
            goto error;
        }

        /* we are now assured of success, so mark the allocation */
        dev->connections++;
        VIR_DEBUG("Using physical device %04x:%02x:%02x.%x, connections %d",
                  dev->device.pci.domain, dev->device.pci.bus,
                  dev->device.pci.slot, dev->device.pci.function,
                  dev->connections);
3846 3847
    }

3848
success:
3849 3850 3851
    netdef->connections++;
    VIR_DEBUG("Using network %s, %d connections",
              netdef->name, netdef->connections);
3852 3853 3854 3855 3856
    ret = 0;
cleanup:
    if (network)
        virNetworkObjUnlock(network);
    return ret;
3857 3858 3859

error:
    goto cleanup;
3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876
}


/* networkReleaseActualDevice:
 * @iface:  a domain's NetDef (interface definition)
 *
 * Given a domain <interface> element that previously had its <actual>
 * element filled in (and possibly a physical device allocated to it),
 * free up the physical device for use by someone else, and free the
 * virDomainActualNetDef.
 *
 * Returns 0 on success, -1 on failure.
 */
int
networkReleaseActualDevice(virDomainNetDefPtr iface)
{
    struct network_driver *driver = driverState;
3877
    enum virDomainNetType actualType = virDomainNetGetActualType(iface);
3878
    virNetworkObjPtr network;
3879
    virNetworkDefPtr netdef;
3880 3881
    virNetworkForwardIfDefPtr dev = NULL;
    int ii, ret = -1;
3882 3883 3884 3885 3886 3887 3888 3889

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
       return 0;

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
3890 3891 3892
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3893 3894 3895 3896
        goto error;
    }
    netdef = network->def;

3897 3898 3899
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
3900 3901
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
3902 3903 3904
    }

    if (netdef->nForwardIfs == 0) {
3905
        virReportError(VIR_ERR_INTERNAL_ERROR,
3906
                       _("network '%s' uses a direct/hostdev mode, but "
3907 3908
                         "has no forward dev and no interface pool"),
                       netdef->name);
3909
        goto error;
3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921
    }

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

        actualDev = virDomainNetGetActualDirectDev(iface);
        if (!actualDev) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("the interface uses a direct mode, "
                             "but has no source dev"));
            goto error;
        }
3922 3923

        for (ii = 0; ii < netdef->nForwardIfs; ii++) {
3924 3925 3926
            if (netdef->forwardIfs[ii].type
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
                STREQ(actualDev, netdef->forwardIfs[ii].device.dev)) {
3927 3928 3929 3930
                dev = &netdef->forwardIfs[ii];
                break;
            }
        }
3931

3932
        if (!dev) {
3933
            virReportError(VIR_ERR_INTERNAL_ERROR,
3934 3935
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
3936
                           netdef->name, actualDev);
3937
            goto error;
3938 3939
        }

3940
        dev->connections--;
3941
        VIR_DEBUG("Releasing physical device %s, connections %d",
3942
                  dev->device.dev, dev->connections);
3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981

    } else /* if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) */ {
        virDomainHostdevDefPtr hostdev;

        hostdev = virDomainNetGetActualHostdev(iface);
        if (!hostdev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("the interface uses a hostdev mode, but has no hostdev"));
            goto error;
        }

        for (ii = 0; ii < netdef->nForwardIfs; ii++) {
            if (netdef->forwardIfs[ii].type
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
                (virDevicePCIAddressEqual(hostdev->source.subsys.u.pci,
                                          netdef->forwardIfs[ii].device.pci) == 0)) {
                dev = &netdef->forwardIfs[ii];
                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,
                           hostdev->source.subsys.u.pci.domain,
                           hostdev->source.subsys.u.pci.bus,
                           hostdev->source.subsys.u.pci.slot,
                           hostdev->source.subsys.u.pci.function);
                goto error;
        }

        dev->connections--;
        VIR_DEBUG("Releasing physical device %04x:%02x:%02x.%x, connections %d",
                  dev->device.pci.domain, dev->device.pci.bus,
                  dev->device.pci.slot, dev->device.pci.function,
                  dev->connections);
   }
3982

3983
success:
3984 3985 3986
    netdef->connections--;
    VIR_DEBUG("Releasing network %s, %d connections",
              netdef->name, netdef->connections);
3987 3988 3989 3990
    ret = 0;
cleanup:
    if (network)
        virNetworkObjUnlock(network);
3991 3992 3993 3994
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
3995
    return ret;
3996 3997 3998

error:
    goto cleanup;
3999
}
4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024

/*
 * networkGetNetworkAddress:
 * @netname: the name of a network
 * @netaddr: string representation of IP address for that network.
 *
 * Attempt to return an IP (v4) address associated with the named
 * network. If a libvirt virtual network, that will be provided in the
 * configuration. For host bridge and direct (macvtap) networks, we
 * must do an ioctl to learn the address.
 *
 * Note: This function returns the 1st IPv4 address it finds. It might
 * be useful if it was more flexible, but the current use (getting a
 * listen address for qemu's vnc/spice graphics server) can only use a
 * single address anyway.
 *
 * Returns 0 on success, and puts a string (which must be free'd by
 * the caller) into *netaddr. Returns -1 on failure or -2 if
 * completely unsupported.
 */
int
networkGetNetworkAddress(const char *netname, char **netaddr)
{
    int ret = -1;
    struct network_driver *driver = driverState;
4025
    virNetworkObjPtr network;
4026 4027 4028 4029
    virNetworkDefPtr netdef;
    virNetworkIpDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4030
    char *dev_name = NULL;
4031 4032 4033 4034 4035 4036

    *netaddr = NULL;
    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, netname);
    networkDriverUnlock(driver);
    if (!network) {
4037 4038 4039
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4040
        goto error;
4041 4042 4043 4044 4045 4046 4047 4048 4049 4050
    }
    netdef = network->def;

    switch (netdef->forwardType) {
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
        /* if there's an ipv4def, get it's address */
        ipdef = virNetworkDefGetIpByIndex(netdef, AF_INET, 0);
        if (!ipdef) {
4051 4052 4053
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have an IPv4 address"),
                           netdef->name);
4054 4055 4056 4057 4058 4059
            break;
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4060
        if ((dev_name = netdef->bridge))
4061 4062 4063 4064 4065 4066 4067 4068 4069
            break;
        /*
         * fall through if netdef->bridge wasn't set, since this is
         * also a direct-mode interface.
         */
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
        if ((netdef->nForwardIfs > 0) && netdef->forwardIfs)
4070
            dev_name = netdef->forwardIfs[0].device.dev;
4071

4072
        if (!dev_name) {
4073 4074 4075
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4076 4077 4078 4079
        }
        break;
    }

4080
    if (dev_name) {
4081
        if (virNetDevGetIPv4Address(dev_name, &addr) < 0)
4082
            goto error;
4083
        addrptr = &addr;
4084 4085
    }

4086 4087 4088
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
        goto error;
4089 4090
    }

4091
    ret = 0;
4092 4093 4094 4095
cleanup:
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4096 4097 4098

error:
    goto cleanup;
4099
}