bridge_driver.c 159.8 KB
Newer Older
G
Gene Czarcinski 已提交
1

2
/*
3
 * bridge_driver.c: core driver methods for managing network
4
 *
5
 * Copyright (C) 2006-2013 Red Hat, Inc.
6 7 8 9 10 11 12 13 14 15 16 17 18
 * 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
19
 * License along with this library.  If not, see
O
Osier Yang 已提交
20
 * <http://www.gnu.org/licenses/>.
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 <sys/wait.h>
#include <sys/ioctl.h>
44
#include <net/if.h>
45

46
#include "virerror.h"
47
#include "datatypes.h"
48
#include "bridge_driver.h"
49
#include "network_conf.h"
50
#include "device_conf.h"
51
#include "driver.h"
52
#include "virbuffer.h"
53
#include "virpidfile.h"
54
#include "virutil.h"
55
#include "vircommand.h"
56
#include "viralloc.h"
57
#include "viruuid.h"
58
#include "viriptables.h"
59
#include "virlog.h"
60
#include "virdnsmasq.h"
61
#include "configmake.h"
62
#include "virnetdev.h"
63
#include "virpci.h"
64 65
#include "virnetdevbridge.h"
#include "virnetdevtap.h"
66
#include "virnetdevvportprofile.h"
67
#include "virdbus.h"
68
#include "virfile.h"
69

70 71
#define NETWORK_PID_DIR LOCALSTATEDIR "/run/libvirt/network"
#define NETWORK_STATE_DIR LOCALSTATEDIR "/lib/libvirt/network"
72

73
#define DNSMASQ_STATE_DIR LOCALSTATEDIR "/lib/libvirt/dnsmasq"
74
#define RADVD_STATE_DIR LOCALSTATEDIR "/lib/libvirt/radvd"
75

76 77
#define VIR_FROM_THIS VIR_FROM_NETWORK

78 79
/* Main driver state */
struct network_driver {
80
    virMutex lock;
81

82
    virNetworkObjList networks;
83 84 85 86 87

    iptablesContext *iptables;
    char *networkConfigDir;
    char *networkAutostartDir;
    char *logDir;
88
    dnsmasqCapsPtr dnsmasqCaps;
89 90
};

91 92 93

static void networkDriverLock(struct network_driver *driver)
{
94
    virMutexLock(&driver->lock);
95 96 97
}
static void networkDriverUnlock(struct network_driver *driver)
{
98
    virMutexUnlock(&driver->lock);
99 100
}

101
static int networkStateCleanup(void);
102

103 104 105 106 107 108 109
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,
110
                                     virNetworkObjPtr network);
111

112 113 114 115 116 117 118
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,
119
                                        virNetworkObjPtr network);
120

121
static void networkReloadIptablesRules(struct network_driver *driver);
122
static void networkRefreshDaemons(struct network_driver *driver);
123

124 125 126 127 128
static int networkPlugBandwidth(virNetworkObjPtr net,
                                virDomainNetDefPtr iface);
static int networkUnplugBandwidth(virNetworkObjPtr net,
                                  virDomainNetDefPtr iface);

129 130
static struct network_driver *driverState = NULL;

131
static char *
132
networkDnsmasqLeaseFileNameDefault(const char *netname)
133 134 135
{
    char *leasefile;

136 137
    ignore_value(virAsprintf(&leasefile, DNSMASQ_STATE_DIR "/%s.leases",
                             netname));
138 139 140
    return leasefile;
}

141 142 143
networkDnsmasqLeaseFileNameFunc networkDnsmasqLeaseFileName =
    networkDnsmasqLeaseFileNameDefault;

144 145 146 147 148 149 150 151 152 153
static char *
networkDnsmasqConfigFileName(const char *netname)
{
    char *conffile;

    ignore_value(virAsprintf(&conffile, DNSMASQ_STATE_DIR "/%s.conf",
                             netname));
    return conffile;
}

154 155 156 157 158 159
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

160
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
161 162 163 164 165 166 167 168
    return pidfilebase;
}

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

169 170
    ignore_value(virAsprintf(&configfile, RADVD_STATE_DIR "/%s-radvd.conf",
                             netname));
171 172
    return configfile;
}
173

174 175 176 177 178 179 180
/* do needed cleanup steps and remove the network from the list */
static int
networkRemoveInactive(struct network_driver *driver,
                      virNetworkObjPtr net)
{
    char *leasefile = NULL;
    char *radvdconfigfile = NULL;
181
    char *configfile = NULL;
182
    char *radvdpidbase = NULL;
183
    char *statusfile = NULL;
184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201
    dnsmasqContext *dctx = NULL;
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(net);

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
    if (!(dctx = dnsmasqContextNew(def->name, DNSMASQ_STATE_DIR)))
        goto cleanup;

    if (!(leasefile = networkDnsmasqLeaseFileName(def->name)))
        goto cleanup;

    if (!(radvdconfigfile = networkRadvdConfigFileName(def->name)))
        goto no_memory;

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
        goto no_memory;

202 203 204
    if (!(configfile = networkDnsmasqConfigFileName(def->name)))
        goto no_memory;

205 206 207
    if (!(statusfile = virNetworkConfigFile(NETWORK_STATE_DIR, def->name)))
        goto no_memory;

208 209 210
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
211
    unlink(configfile);
212 213 214 215 216

    /* radvd */
    unlink(radvdconfigfile);
    virPidFileDelete(NETWORK_PID_DIR, radvdpidbase);

217 218 219
    /* remove status file */
    unlink(statusfile);

220 221 222 223 224 225 226
    /* remove the network definition */
    virNetworkRemoveInactive(&driver->networks, net);

    ret = 0;

cleanup:
    VIR_FREE(leasefile);
227
    VIR_FREE(configfile);
228 229
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
230
    VIR_FREE(statusfile);
231 232 233 234 235 236 237 238
    dnsmasqContextFree(dctx);
    return ret;

no_memory:
    virReportOOMError();
    goto cleanup;
}

239 240 241
static char *
networkBridgeDummyNicName(const char *brname)
{
242
    static const char dummyNicSuffix[] = "-nic";
243 244
    char *nicname;

245 246 247 248 249 250 251
    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.
         */
252 253 254 255 256
        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));
257
    } else {
258
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
259
    }
260 261 262
    return nicname;
}

263
static void
264 265
networkFindActiveConfigs(struct network_driver *driver)
{
266 267 268 269 270 271 272 273 274
    unsigned int i;

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

        virNetworkObjLock(obj);

        /* If bridge exists, then mark it active */
        if (obj->def->bridge &&
H
Hu Tao 已提交
275
            virNetDevExists(obj->def->bridge) == 1) {
276 277
            obj->active = 1;

278 279
            /* Try and read dnsmasq/radvd pids if any */
            if (obj->def->ips && (obj->def->nips > 0)) {
280 281 282
                char *radvdpidbase;

                ignore_value(virPidFileReadIfAlive(NETWORK_PID_DIR, obj->def->name,
283 284
                                                   &obj->dnsmasqPid,
                                                   dnsmasqCapsGetBinaryPath(driver->dnsmasqCaps)));
285

286
                if (!(radvdpidbase = networkRadvdPidfileBasename(obj->def->name))) {
287
                    virReportOOMError();
288 289
                    goto cleanup;
                }
290 291
                ignore_value(virPidFileReadIfAlive(NETWORK_PID_DIR, radvdpidbase,
                                                   &obj->radvdPid, RADVD));
292
                VIR_FREE(radvdpidbase);
293 294 295
            }
        }

296
    cleanup:
297 298
        virNetworkObjUnlock(obj);
    }
299 300 301 302 303 304 305 306 307 308 309 310 311 312 313

    /* remove inactive transient networks */
    i = 0;
    while (i < driver->networks.count) {
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);

        if (!obj->persistent && !obj->active) {
            networkRemoveInactive(driver, obj);
            continue;
        }

        virNetworkObjUnlock(obj);
        i++;
    }
314 315 316
}


317 318 319
static void
networkAutostartConfigs(struct network_driver *driver) {
    unsigned int i;
320

321
    for (i = 0 ; i < driver->networks.count ; i++) {
322
        virNetworkObjLock(driver->networks.objs[i]);
323
        if (driver->networks.objs[i]->autostart &&
324 325
            !virNetworkObjIsActive(driver->networks.objs[i])) {
            if (networkStartNetwork(driver, driver->networks.objs[i]) < 0) {
326
            /* failed to start but already logged */
327
            }
328
        }
329
        virNetworkObjUnlock(driver->networks.objs[i]);
330 331 332
    }
}

333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351
#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

352
/**
353
 * networkStateInitialize:
354 355 356 357
 *
 * Initialization function for the QEmu daemon
 */
static int
358 359 360
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
361
{
362
    char *base = NULL;
363 364 365
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
366 367

    if (VIR_ALLOC(driverState) < 0)
368
        goto error;
369

370 371 372 373
    if (virMutexInit(&driverState->lock) < 0) {
        VIR_FREE(driverState);
        goto error;
    }
374 375
    networkDriverLock(driverState);

376
    if (privileged) {
377
        if (virAsprintf(&driverState->logDir,
378
                        "%s/log/libvirt/qemu", LOCALSTATEDIR) == -1)
379 380
            goto out_of_memory;

381
        if ((base = strdup(SYSCONFDIR "/libvirt")) == NULL)
382 383
            goto out_of_memory;
    } else {
384
        char *userdir = virGetUserCacheDirectory();
385 386 387

        if (!userdir)
            goto error;
388

389
        if (virAsprintf(&driverState->logDir,
390
                        "%s/qemu/log", userdir) == -1) {
391
            VIR_FREE(userdir);
392
            goto out_of_memory;
393
        }
394
        VIR_FREE(userdir);
395

396
        userdir = virGetUserConfigDirectory();
397
        if (virAsprintf(&base, "%s", userdir) == -1) {
398
            VIR_FREE(userdir);
399 400
            goto out_of_memory;
        }
401
        VIR_FREE(userdir);
402 403 404 405 406
    }

    /* Configuration paths are either ~/.libvirt/qemu/... (session) or
     * /etc/libvirt/qemu/... (system).
     */
407
    if (virAsprintf(&driverState->networkConfigDir, "%s/qemu/networks", base) == -1)
408 409
        goto out_of_memory;

410 411
    if (virAsprintf(&driverState->networkAutostartDir, "%s/qemu/networks/autostart",
                    base) == -1)
412 413 414 415
        goto out_of_memory;

    VIR_FREE(base);

416
    if (!(driverState->iptables = iptablesContextNew())) {
417
        goto out_of_memory;
418 419
    }

420 421
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
    driverState->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
422

423 424 425 426
    if (virNetworkLoadAllState(&driverState->networks,
                               NETWORK_STATE_DIR) < 0)
        goto error;

427
    if (virNetworkLoadAllConfigs(&driverState->networks,
428
                                 driverState->networkConfigDir,
429 430 431
                                 driverState->networkAutostartDir) < 0)
        goto error;

432
    networkFindActiveConfigs(driverState);
433
    networkReloadIptablesRules(driverState);
434
    networkRefreshDaemons(driverState);
435 436
    networkAutostartConfigs(driverState);

437 438
    networkDriverUnlock(driverState);

439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464
#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

465 466
    return 0;

467
out_of_memory:
468
    virReportOOMError();
469 470

error:
471 472 473
    if (driverState)
        networkDriverUnlock(driverState);

474
    VIR_FREE(base);
475
    networkStateCleanup();
476 477 478 479
    return -1;
}

/**
480
 * networkStateReload:
481 482 483 484 485
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
486
networkStateReload(void) {
487 488 489
    if (!driverState)
        return 0;

490
    networkDriverLock(driverState);
491 492
    virNetworkLoadAllState(&driverState->networks,
                           NETWORK_STATE_DIR);
493
    virNetworkLoadAllConfigs(&driverState->networks,
494 495
                             driverState->networkConfigDir,
                             driverState->networkAutostartDir);
496
    networkReloadIptablesRules(driverState);
497
    networkRefreshDaemons(driverState);
498
    networkAutostartConfigs(driverState);
499
    networkDriverUnlock(driverState);
500 501 502 503 504
    return 0;
}


/**
505
 * networkStateCleanup:
506 507 508 509
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
510
networkStateCleanup(void) {
511 512 513
    if (!driverState)
        return -1;

514 515
    networkDriverLock(driverState);

516
    /* free inactive networks */
517
    virNetworkObjListFree(&driverState->networks);
518 519 520 521 522 523 524 525

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

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

526 527
    virObjectUnref(driverState->dnsmasqCaps);

528
    networkDriverUnlock(driverState);
529
    virMutexDestroy(&driverState->lock);
530

531 532 533 534 535 536
    VIR_FREE(driverState);

    return 0;
}


537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595
/* 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;
}

G
Gene Czarcinski 已提交
596 597 598 599
    /* the following does not build a file, it builds a list
     * which is later saved into a file
     */

600
static int
G
Gene Czarcinski 已提交
601 602
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
                                 virNetworkIpDefPtr ipdef)
603
{
G
Gene Czarcinski 已提交
604 605
    unsigned int i;
    bool ipv6 = false;
606

G
Gene Czarcinski 已提交
607 608
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
609 610
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
611
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
612 613
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
614
                return -1;
615
    }
616

G
Gene Czarcinski 已提交
617 618 619 620 621 622 623 624 625
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
    unsigned int i, j;

626 627
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
628
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
629
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
630
                for (j = 0; j < host->nnames; j++)
631 632
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
633 634
            }
        }
635 636
    }

637
    return 0;
638 639 640
}


641 642
int
networkDnsmasqConfContents(virNetworkObjPtr network,
643 644 645 646
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
647
{
648
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
649
    int r, ret = -1;
650
    int nbleases = 0;
651
    int ii;
652 653 654 655
    char *record = NULL;
    char *recordPort = NULL;
    char *recordWeight = NULL;
    char *recordPriority = NULL;
656
    virNetworkDNSDefPtr dns = &network->def->dns;
G
Gene Czarcinski 已提交
657 658
    virNetworkIpDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
    bool ipv6SLAAC;
659

660 661
    *configstr = NULL;

662
    /*
663 664 665
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
666
     *
667 668
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
669
     */
670 671 672 673 674 675

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

676 677
    /* create dnsmasq config file appropriate for this network */
    virBufferAsprintf(&configbuf,
678 679 680 681 682 683 684 685 686 687 688 689
                      "##WARNING:  THIS IS AN AUTO-GENERATED FILE. "
                      "CHANGES TO IT ARE LIKELY TO BE\n"
                      "##OVERWRITTEN AND LOST.  Changes to this "
                      "configuration should be made using:\n"
                      "##    virsh net-edit %s\n"
                      "## or other application using the libvirt API.\n"
                      "##\n## dnsmasq conf file created by libvirt\n"
                      "strict-order\n"
                      "domain-needed\n",
                      network->def->name);

    if (network->def->domain) {
690
        virBufferAsprintf(&configbuf,
691 692 693 694 695
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
    /* need to specify local even if no domain specified */
696
    virBufferAsprintf(&configbuf,
697 698
                      "local=/%s/\n",
                      network->def->domain ? network->def->domain : "");
699

700
    if (pidfile)
701
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
702

703 704 705
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

706 707 708 709 710 711 712 713
    if (dnsmasqCapsGet(caps, DNSMASQ_CAPS_BIND_DYNAMIC)) {
        /* using --bind-dynamic with only --interface (no
         * --listen-address) prevents dnsmasq from responding to dns
         * queries that arrive on some interface other than our bridge
         * interface (in other words, requests originating somewhere
         * other than one of the virtual guests connected directly to
         * this network). This was added in response to CVE 2012-3411.
         */
714
        virBufferAsprintf(&configbuf,
715 716 717
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
718
    } else {
719
        virBufferAddLit(&configbuf, "bind-interfaces\n");
720 721 722 723 724 725 726 727 728 729 730 731 732 733 734
        /*
         * --interface does not actually work with dnsmasq < 2.47,
         * due to DAD for ipv6 addresses on the interface.
         *
         * virCommandAddArgList(cmd, "--interface", network->def->bridge, NULL);
         *
         * So listen on all defined IPv[46] addresses
         */
        for (ii = 0;
             (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, ii));
             ii++) {
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
735

736
            /* also part of CVE 2012-3411 - if the host's version of
737
             * dnsmasq doesn't have bind-dynamic, only allow listening on
738 739
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
740 741
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
742 743 744 745
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
746
                                 "The version of dnsmasq on this host (%d.%d) "
747 748 749 750
                                 "doesn't support the bind-dynamic option or "
                                 "use SO_BINDTODEVICE on listening sockets, "
                                 "one of which is required for safe operation "
                                 "on a publicly routable subnet "
751 752 753 754 755 756
                                 "(see CVE-2012-3411). You must either "
                                 "upgrade dnsmasq, or use a private/local "
                                 "subnet range for this network "
                                 "(as described in RFC1918/RFC3484/RFC4193)."),
                               ipaddr, (int)version / 1000000,
                               (int)(version % 1000000) / 1000);
757 758
                goto cleanup;
            }
759
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
760 761 762
            VIR_FREE(ipaddr);
        }
    }
763

764 765
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
766
     * guaranteed to not work, and set no-resolv so that no dns
767 768 769
     * 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).
770
     */
771
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
772
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
773
                        "no-resolv\n");
774
    }
775

776
    for (ii = 0; ii < dns->ntxts; ii++) {
777 778 779
        virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
                          dns->txts[ii].name,
                          dns->txts[ii].value);
780
    }
781

782 783 784 785 786 787
    for (ii = 0; ii < dns->nsrvs; ii++) {
        if (dns->srvs[ii].service && dns->srvs[ii].protocol) {
            if (dns->srvs[ii].port) {
                if (virAsprintf(&recordPort, "%d", dns->srvs[ii].port) < 0) {
                    virReportOOMError();
                    goto cleanup;
788
                }
789 790 791 792 793
            }
            if (dns->srvs[ii].priority) {
                if (virAsprintf(&recordPriority, "%d", dns->srvs[ii].priority) < 0) {
                    virReportOOMError();
                    goto cleanup;
794
                }
795 796 797
            }
            if (dns->srvs[ii].weight) {
                if (virAsprintf(&recordWeight, "%d", dns->srvs[ii].weight) < 0) {
798 799 800
                    virReportOOMError();
                    goto cleanup;
                }
801
            }
802

803 804 805
            if (virAsprintf(&record, "%s.%s.%s,%s,%s,%s,%s",
                            dns->srvs[ii].service,
                            dns->srvs[ii].protocol,
806 807 808 809 810
                            dns->srvs[ii].domain ? dns->srvs[ii].domain : "",
                            dns->srvs[ii].target ? dns->srvs[ii].target : "",
                            recordPort           ? recordPort           : "",
                            recordPriority       ? recordPriority       : "",
                            recordWeight         ? recordWeight         : "") < 0) {
811 812
                virReportOOMError();
                goto cleanup;
813
            }
814

815
            virBufferAsprintf(&configbuf, "srv-host=%s\n", record);
816 817 818 819
            VIR_FREE(record);
            VIR_FREE(recordPort);
            VIR_FREE(recordWeight);
            VIR_FREE(recordPriority);
820
        }
821 822
    }

G
Gene Czarcinski 已提交
823 824 825 826 827 828 829 830
    /* Find the first dhcp for both IPv4 and IPv6 */
    for (ii = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, ii));
         ii++) {
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
831 832
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
833 834 835 836 837 838 839 840 841 842 843
                    goto cleanup;
                } else {
                    ipv4def = ipdef;
                }
            }
        }
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (!DNSMASQ_DHCPv6_SUPPORT(caps)) {
                    unsigned long version = dnsmasqCapsGetVersion(caps);
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
844 845 846 847 848 849 850 851 852
                                   _("The version of dnsmasq on this host "
                                     "(%d.%d) doesn't adequately support "
                                     "IPv6 dhcp range or dhcp host "
                                     "specification. Version %d.%d or later "
                                     "is required."),
                                   (int)version / 1000000,
                                   (int)(version % 1000000) / 1000,
                                   DNSMASQ_DHCPv6_MAJOR_REQD,
                                   DNSMASQ_DHCPv6_MINOR_REQD);
G
Gene Czarcinski 已提交
853 854 855 856
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
857 858
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
859 860 861 862 863 864 865 866 867 868 869 870 871
                    goto cleanup;
                } else {
                    ipv6def = ipdef;
                }
            } else {
                ipv6SLAAC = true;
            }
        }
    }

    if (ipv6def && ipv6SLAAC) {
        VIR_WARN("For IPv6, when DHCP is specified for one address, then "
                 "state-full Router Advertising will occur.  The additional "
872 873 874 875
                 "IPv6 addresses specified require manually configured guest "
                 "network to work properly since both state-full (DHCP) "
                 "and state-less (SLAAC) addressing are not supported "
                 "on the same network interface.");
G
Gene Czarcinski 已提交
876 877 878 879 880
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
881
        for (r = 0 ; r < ipdef->nranges ; r++) {
882
            char *saddr = virSocketAddrFormat(&ipdef->ranges[r].start);
883 884
            if (!saddr)
                goto cleanup;
885
            char *eaddr = virSocketAddrFormat(&ipdef->ranges[r].end);
886 887 888 889
            if (!eaddr) {
                VIR_FREE(saddr);
                goto cleanup;
            }
890
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s\n",
891
                              saddr, eaddr);
892
            VIR_FREE(saddr);
893
            VIR_FREE(eaddr);
894 895
            nbleases += virSocketAddrGetRange(&ipdef->ranges[r].start,
                                              &ipdef->ranges[r].end);
896
        }
897

898
        /*
899 900 901 902
         * For static-only DHCP, i.e. with no range but at least one
         * host element, we have to add a special --dhcp-range option
         * to enable the service in dnsmasq. (this is for dhcp-hosts=
         * support)
903 904
         */
        if (!ipdef->nranges && ipdef->nhosts) {
905
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
906 907
            if (!bridgeaddr)
                goto cleanup;
908
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static\n", bridgeaddr);
909 910
            VIR_FREE(bridgeaddr);
        }
911

G
Gene Czarcinski 已提交
912 913
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
914

G
Gene Czarcinski 已提交
915 916 917
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts)
918
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
919

G
Gene Czarcinski 已提交
920
            if (ipdef->tftproot) {
921 922
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
923
            }
924

G
Gene Czarcinski 已提交
925 926 927
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
928

G
Gene Czarcinski 已提交
929 930 931 932
                    if (!bootserver) {
                        virReportOOMError();
                        goto cleanup;
                    }
933
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
934
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
935 936
                    VIR_FREE(bootserver);
                } else {
937
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
938 939 940 941 942
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
943

G
Gene Czarcinski 已提交
944 945 946 947 948
    if (nbleases > 0) {
        char *leasefile = networkDnsmasqLeaseFileName(network->def->name);
        if (!leasefile) {
            virReportOOMError();
            goto cleanup;
949
        }
950
        virBufferAsprintf(&configbuf, "dhcp-leasefile=%s\n", leasefile);
G
Gene Czarcinski 已提交
951
        VIR_FREE(leasefile);
952
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
G
Gene Czarcinski 已提交
953
    }
954

G
Gene Czarcinski 已提交
955 956
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
957
        goto cleanup;
G
Gene Czarcinski 已提交
958 959 960 961 962 963

    /* Even if there are currently no static hosts, if we're
     * listening for DHCP, we should write a 0-length hosts
     * file to allow for runtime additions.
     */
    if (ipv4def || ipv6def)
964 965
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
966

967 968
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
969
     */
970
    virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
971
                      dctx->addnhostsfile->path);
G
Gene Czarcinski 已提交
972 973 974 975

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
        if (ipv6def)
976
            virBufferAddLit(&configbuf, "enable-ra\n");
G
Gene Czarcinski 已提交
977 978
        else {
            for (ii = 0;
979 980
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, ii));
                 ii++) {
G
Gene Czarcinski 已提交
981 982 983 984
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
985 986
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
987 988
                    VIR_FREE(bridgeaddr);
                }
989
            }
990
        }
991 992
    }

993 994 995
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

996
    ret = 0;
G
Gene Czarcinski 已提交
997

998
cleanup:
999
    virBufferFreeAndReset(&configbuf);
1000 1001 1002 1003
    VIR_FREE(record);
    VIR_FREE(recordPort);
    VIR_FREE(recordWeight);
    VIR_FREE(recordPriority);
1004
    return ret;
1005 1006
}

1007
/* build the dnsmasq command line */
1008 1009 1010
static int ATTRIBUTE_NONNULL(2)
networkBuildDhcpDaemonCommandLine(virNetworkObjPtr network,
                                  virCommandPtr *cmdout,
1011 1012
                                  char *pidfile, dnsmasqContext *dctx,
                                  dnsmasqCapsPtr caps)
1013
{
1014
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1015
    int ret = -1;
1016 1017
    char *configfile = NULL;
    char *configstr = NULL;
1018 1019

    network->dnsmasqPid = -1;
1020

1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036
    if (networkDnsmasqConfContents(network, pidfile, &configstr, dctx, caps) < 0)
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
    if (!(configfile = networkDnsmasqConfigFileName(network->def->name))) {
        virReportOOMError();
        goto cleanup;
    }

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
                         _("couldn't write dnsmasq config file '%s'"),
                         configfile);
1037 1038 1039
        goto cleanup;
    }

1040 1041
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(caps));
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1042
    *cmdout = cmd;
1043 1044 1045 1046 1047 1048
    ret = 0;
cleanup:
    return ret;
}

static int
1049 1050
networkStartDhcpDaemon(struct network_driver *driver,
                       virNetworkObjPtr network)
1051 1052 1053 1054
{
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1055
    dnsmasqContext *dctx = NULL;
1056

1057
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0)) {
G
Gene Czarcinski 已提交
1058
        /* no IP addresses, so we don't need to run */
1059 1060 1061 1062
        ret = 0;
        goto cleanup;
    }

1063 1064
    if (virFileMakePath(NETWORK_PID_DIR) < 0) {
        virReportSystemError(errno,
1065 1066
                             _("cannot create directory %s"),
                             NETWORK_PID_DIR);
1067
        goto cleanup;
1068
    }
1069 1070
    if (virFileMakePath(NETWORK_STATE_DIR) < 0) {
        virReportSystemError(errno,
1071 1072
                             _("cannot create directory %s"),
                             NETWORK_STATE_DIR);
1073
        goto cleanup;
1074 1075
    }

1076
    if (!(pidfile = virPidFileBuildPath(NETWORK_PID_DIR, network->def->name))) {
1077
        virReportOOMError();
1078
        goto cleanup;
1079 1080
    }

1081 1082
    if (virFileMakePath(DNSMASQ_STATE_DIR) < 0) {
        virReportSystemError(errno,
1083 1084 1085 1086 1087
                             _("cannot create directory %s"),
                             DNSMASQ_STATE_DIR);
        goto cleanup;
    }

1088 1089 1090 1091
    dctx = dnsmasqContextNew(network->def->name, DNSMASQ_STATE_DIR);
    if (dctx == NULL)
        goto cleanup;

1092 1093 1094 1095
    dnsmasqCapsRefresh(&driver->dnsmasqCaps, false);

    ret = networkBuildDhcpDaemonCommandLine(network, &cmd, pidfile,
                                            dctx, driver->dnsmasqCaps);
1096 1097 1098 1099 1100
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1101
        goto cleanup;
1102

G
Guido Günther 已提交
1103 1104
    ret = virCommandRun(cmd, NULL);
    if (ret < 0) {
1105
        goto cleanup;
G
Guido Günther 已提交
1106
    }
1107 1108

    /*
1109 1110 1111 1112 1113
     * 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
1114 1115
     */

1116
    ret = virPidFileRead(NETWORK_PID_DIR, network->def->name,
1117 1118
                         &network->dnsmasqPid);
    if (ret < 0)
1119
        goto cleanup;
1120

1121 1122 1123
    ret = 0;
cleanup:
    VIR_FREE(pidfile);
1124
    virCommandFree(cmd);
1125
    dnsmasqContextFree(dctx);
1126 1127 1128
    return ret;
}

1129 1130
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1131 1132
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1133 1134 1135
 *
 *  Returns 0 on success, -1 on failure.
 */
1136
static int
1137 1138
networkRefreshDhcpDaemon(struct network_driver *driver,
                         virNetworkObjPtr network)
1139
{
1140
    int ret = -1, ii;
G
Gene Czarcinski 已提交
1141
    virNetworkIpDefPtr ipdef, ipv4def, ipv6def;
1142
    dnsmasqContext *dctx = NULL;
1143

G
Gene Czarcinski 已提交
1144
    /* if no IP addresses specified, nothing to do */
1145
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1146 1147
        return 0;

1148 1149
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1150
        return networkStartDhcpDaemon(driver, network);
1151

G
Gene Czarcinski 已提交
1152 1153 1154 1155 1156 1157 1158 1159 1160
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
    if (!(dctx = dnsmasqContextNew(network->def->name, DNSMASQ_STATE_DIR)))
        goto cleanup;

    /* Look for first IPv4 address that has dhcp defined.
     * We only support dhcp-host config on one IPv4 subnetwork
     * and on one IPv6 subnetwork.
     */
    ipv4def = NULL;
1161 1162 1163
    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, ii));
         ii++) {
G
Gene Czarcinski 已提交
1164 1165
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1166 1167
    }

G
Gene Czarcinski 已提交
1168 1169 1170 1171 1172 1173
    ipv6def = NULL;
    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, ii));
         ii++) {
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1174 1175
    }

G
Gene Czarcinski 已提交
1176 1177 1178 1179 1180
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
           goto cleanup;

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
           goto cleanup;
1181

G
Gene Czarcinski 已提交
1182
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
1183 1184 1185
       goto cleanup;

    if ((ret = dnsmasqSave(dctx)) < 0)
1186
        goto cleanup;
1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201

    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
1202 1203
networkRestartDhcpDaemon(struct network_driver *driver,
                         virNetworkObjPtr network)
1204 1205 1206 1207 1208 1209
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1210
    }
1211
    /* now start dnsmasq if it should be started */
1212
    return networkStartDhcpDaemon(driver, network);
1213 1214
}

G
Gene Czarcinski 已提交
1215 1216 1217 1218 1219 1220
static char radvd1[] = "  AdvOtherConfigFlag off;\n\n";
static char radvd2[] = "    AdvAutonomous off;\n";
static char radvd3[] = "    AdvOnLink on;\n"
                       "    AdvAutonomous on;\n"
                       "    AdvRouterAddr off;\n";

1221 1222 1223
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1224
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1225 1226
    int ret = -1, ii;
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
1227
    bool v6present = false, dhcp6 = false;
1228 1229

    *configstr = NULL;
1230

G
Gene Czarcinski 已提交
1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247
    /* Check if DHCPv6 is needed */
    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, ii));
         ii++) {
        v6present = true;
        if (ipdef->nranges || ipdef->nhosts) {
            dhcp6 = true;
            break;
        }
    }

    /* If there are no IPv6 addresses, then we are done */
    if (!v6present) {
        ret = 0;
        goto cleanup;
    }

1248 1249 1250
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1251
    virBufferAsprintf(&configbuf, "interface %s\n"
1252 1253
                      "{\n"
                      "  AdvSendAdvert on;\n"
1254
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1255 1256 1257 1258 1259
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1260 1261

    /* add a section for each IPv6 address in the config */
1262 1263 1264 1265 1266 1267 1268 1269
    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, ii));
         ii++) {
        int prefix;
        char *netaddr;

        prefix = virNetworkIpDefPrefix(ipdef);
        if (prefix < 0) {
1270 1271 1272
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1273 1274
            goto cleanup;
        }
1275
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1276
            goto cleanup;
1277
        virBufferAsprintf(&configbuf,
1278
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1279 1280 1281
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1282 1283 1284
        VIR_FREE(netaddr);
    }

1285 1286 1287
    /* only create the string if we found at least one IPv6 address */
    if (v6present) {
        virBufferAddLit(&configbuf, "};\n");
1288

1289 1290 1291 1292 1293 1294 1295 1296
        if (virBufferError(&configbuf)) {
            virReportOOMError();
            goto cleanup;
        }
        if (!(*configstr = virBufferContentAndReset(&configbuf))) {
            virReportOOMError();
            goto cleanup;
        }
1297
    }
1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322

    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;
1323 1324 1325 1326
        goto cleanup;
    }

    /* construct the filename */
1327
    if (!(*configFile = networkRadvdConfigFileName(network->def->name))) {
1328 1329 1330 1331
        virReportOOMError();
        goto cleanup;
    }
    /* write the file */
1332
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1333 1334
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346
                             *configFile);
        goto cleanup;
    }

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

static int
G
Gene Czarcinski 已提交
1347 1348
networkStartRadvd(struct network_driver *driver ATTRIBUTE_UNUSED,
                        virNetworkObjPtr network)
1349 1350 1351 1352 1353 1354 1355 1356 1357
{
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1358
    /* Is dnsmasq handling RA? */
1359
   if (DNSMASQ_RA_SUPPORT(driver->dnsmasqCaps)) {
G
Gene Czarcinski 已提交
1360 1361 1362 1363
        ret = 0;
        goto cleanup;
    }

1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374
    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);
1375 1376 1377
        goto cleanup;
    }

1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403
    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;

1404 1405 1406 1407
    /* 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
1408
     * virPidFileRead() below will fail if we use them).
1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423
     * 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;

1424
    if (virPidFileRead(NETWORK_PID_DIR, radvdpidbase, &network->radvdPid) < 0)
1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435
        goto cleanup;

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

1436
static int
1437 1438
networkRefreshRadvd(struct network_driver *driver ATTRIBUTE_UNUSED,
                    virNetworkObjPtr network)
1439
{
G
Gene Czarcinski 已提交
1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457
    char *radvdpidbase;

    /* Is dnsmasq handling RA? */
    if (DNSMASQ_RA_SUPPORT(driver->dnsmasqCaps)) {
        if (network->radvdPid <= 0)
            return 0;
        /* radvd should not be running but in case it is */
        if ((networkKillDaemon(network->radvdPid, "radvd",
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
            virPidFileDelete(NETWORK_PID_DIR, radvdpidbase);
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }

1458 1459
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
G
Gene Czarcinski 已提交
1460
        return networkStartRadvd(driver, network);
1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472

    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);
}

1473 1474
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1475
static int
1476 1477
networkRestartRadvd(struct network_driver *driver,
                    virNetworkObjPtr network)
1478 1479 1480 1481 1482 1483 1484 1485 1486
{
    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).
         */
G
Gene Czarcinski 已提交
1487
        if ((networkKillDaemon(network->radvdPid, "radvd",
1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500
                               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 */

1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515
/* 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) &&
1516 1517 1518
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
1519 1520 1521 1522 1523 1524
            /* 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.
             */
1525 1526
            networkRefreshDhcpDaemon(driver, network);
            networkRefreshRadvd(driver, network);
1527 1528 1529 1530 1531
        }
        virNetworkObjUnlock(network);
    }
}

1532
static int
1533
networkAddMasqueradingIptablesRules(struct network_driver *driver,
1534 1535
                                    virNetworkObjPtr network,
                                    virNetworkIpDefPtr ipdef)
1536 1537
{
    int prefix = virNetworkIpDefPrefix(ipdef);
1538
    const char *forwardIf = virNetworkDefForwardIf(network->def, 0);
1539 1540

    if (prefix < 0) {
1541 1542 1543
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Invalid prefix or netmask for '%s'"),
                       network->def->bridge);
1544 1545
        goto masqerr1;
    }
1546

1547
    /* allow forwarding packets from the bridge interface */
1548
    if (iptablesAddForwardAllowOut(driver->iptables,
1549
                                   &ipdef->address,
1550
                                   prefix,
1551
                                   network->def->bridge,
1552
                                   forwardIf) < 0) {
1553 1554 1555
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow forwarding from '%s'"),
                       network->def->bridge);
1556 1557 1558
        goto masqerr1;
    }

1559 1560 1561
    /* allow forwarding packets to the bridge interface if they are
     * part of an existing connection
     */
1562
    if (iptablesAddForwardAllowRelatedIn(driver->iptables,
1563
                                         &ipdef->address,
1564
                                         prefix,
1565
                                         network->def->bridge,
1566
                                         forwardIf) < 0) {
1567 1568 1569
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow forwarding to '%s'"),
                       network->def->bridge);
1570 1571 1572
        goto masqerr2;
    }

1573 1574 1575 1576 1577
    /*
     * Enable masquerading.
     *
     * We need to end up with 3 rules in the table in this order
     *
E
Eric Blake 已提交
1578 1579
     *  1. protocol=tcp with sport mapping restriction
     *  2. protocol=udp with sport mapping restriction
1580 1581 1582
     *  3. generic any protocol
     *
     * The sport mappings are required, because default IPtables
E
Eric Blake 已提交
1583
     * MASQUERADE maintain port numbers unchanged where possible.
1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596
     *
     * 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 */
1597
    if (iptablesAddForwardMasquerade(driver->iptables,
1598
                                     &ipdef->address,
1599
                                     prefix,
1600
                                     forwardIf,
1601 1602
                                     &network->def->forward.addr,
                                     &network->def->forward.port,
1603
                                     NULL) < 0) {
1604 1605 1606 1607 1608
        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);
1609 1610 1611
        goto masqerr3;
    }

1612
    /* UDP with a source port restriction */
1613
    if (iptablesAddForwardMasquerade(driver->iptables,
1614
                                     &ipdef->address,
1615
                                     prefix,
1616
                                     forwardIf,
1617 1618
                                     &network->def->forward.addr,
                                     &network->def->forward.port,
1619
                                     "udp") < 0) {
1620 1621 1622 1623 1624
        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);
1625 1626 1627 1628
        goto masqerr4;
    }

    /* TCP with a source port restriction */
1629
    if (iptablesAddForwardMasquerade(driver->iptables,
1630
                                     &ipdef->address,
1631
                                     prefix,
1632
                                     forwardIf,
1633 1634
                                     &network->def->forward.addr,
                                     &network->def->forward.port,
1635
                                     "tcp") < 0) {
1636 1637 1638 1639 1640
        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);
1641 1642 1643
        goto masqerr5;
    }

1644
    return 0;
1645

1646 1647
 masqerr5:
    iptablesRemoveForwardMasquerade(driver->iptables,
1648
                                    &ipdef->address,
1649
                                    prefix,
1650
                                    forwardIf,
1651 1652
                                    &network->def->forward.addr,
                                    &network->def->forward.port,
1653 1654 1655
                                    "udp");
 masqerr4:
    iptablesRemoveForwardMasquerade(driver->iptables,
1656
                                    &ipdef->address,
1657
                                    prefix,
1658
                                    forwardIf,
1659 1660
                                    &network->def->forward.addr,
                                    &network->def->forward.port,
1661
                                    NULL);
1662 1663
 masqerr3:
    iptablesRemoveForwardAllowRelatedIn(driver->iptables,
1664
                                        &ipdef->address,
1665
                                        prefix,
1666
                                        network->def->bridge,
1667
                                        forwardIf);
1668 1669
 masqerr2:
    iptablesRemoveForwardAllowOut(driver->iptables,
1670
                                  &ipdef->address,
1671
                                  prefix,
1672
                                  network->def->bridge,
1673
                                  forwardIf);
1674
 masqerr1:
1675
    return -1;
1676 1677
}

1678 1679 1680 1681 1682 1683
static void
networkRemoveMasqueradingIptablesRules(struct network_driver *driver,
                                       virNetworkObjPtr network,
                                       virNetworkIpDefPtr ipdef)
{
    int prefix = virNetworkIpDefPrefix(ipdef);
1684
    const char *forwardIf = virNetworkDefForwardIf(network->def, 0);
1685 1686 1687 1688 1689

    if (prefix >= 0) {
        iptablesRemoveForwardMasquerade(driver->iptables,
                                        &ipdef->address,
                                        prefix,
1690
                                        forwardIf,
1691 1692
                                        &network->def->forward.addr,
                                        &network->def->forward.port,
1693 1694 1695 1696
                                        "tcp");
        iptablesRemoveForwardMasquerade(driver->iptables,
                                        &ipdef->address,
                                        prefix,
1697
                                        forwardIf,
1698 1699
                                        &network->def->forward.addr,
                                        &network->def->forward.port,
1700 1701 1702 1703
                                        "udp");
        iptablesRemoveForwardMasquerade(driver->iptables,
                                        &ipdef->address,
                                        prefix,
1704
                                        forwardIf,
1705 1706
                                        &network->def->forward.addr,
                                        &network->def->forward.port,
1707 1708 1709 1710 1711 1712
                                        NULL);

        iptablesRemoveForwardAllowRelatedIn(driver->iptables,
                                            &ipdef->address,
                                            prefix,
                                            network->def->bridge,
1713
                                            forwardIf);
1714 1715 1716 1717
        iptablesRemoveForwardAllowOut(driver->iptables,
                                      &ipdef->address,
                                      prefix,
                                      network->def->bridge,
1718
                                      forwardIf);
1719 1720 1721
    }
}

1722
static int
1723
networkAddRoutingIptablesRules(struct network_driver *driver,
1724
                               virNetworkObjPtr network,
1725 1726
                               virNetworkIpDefPtr ipdef)
{
1727
    int prefix = virNetworkIpDefPrefix(ipdef);
1728
    const char *forwardIf = virNetworkDefForwardIf(network->def, 0);
1729 1730

    if (prefix < 0) {
1731 1732 1733
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Invalid prefix or netmask for '%s'"),
                       network->def->bridge);
1734 1735
        goto routeerr1;
    }
1736

1737
    /* allow routing packets from the bridge interface */
1738
    if (iptablesAddForwardAllowOut(driver->iptables,
1739
                                   &ipdef->address,
1740
                                   prefix,
1741
                                   network->def->bridge,
1742
                                   forwardIf) < 0) {
1743 1744 1745
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow routing from '%s'"),
                       network->def->bridge);
1746 1747 1748 1749
        goto routeerr1;
    }

    /* allow routing packets to the bridge interface */
1750
    if (iptablesAddForwardAllowIn(driver->iptables,
1751
                                  &ipdef->address,
1752
                                  prefix,
1753
                                  network->def->bridge,
1754
                                  forwardIf) < 0) {
1755 1756 1757
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow routing to '%s'"),
                       network->def->bridge);
1758 1759 1760
        goto routeerr2;
    }

1761
    return 0;
1762

1763
routeerr2:
1764
    iptablesRemoveForwardAllowOut(driver->iptables,
1765
                                  &ipdef->address,
1766
                                  prefix,
1767
                                  network->def->bridge,
1768
                                  forwardIf);
1769
routeerr1:
1770
    return -1;
1771 1772
}

1773 1774 1775 1776 1777 1778
static void
networkRemoveRoutingIptablesRules(struct network_driver *driver,
                                  virNetworkObjPtr network,
                                  virNetworkIpDefPtr ipdef)
{
    int prefix = virNetworkIpDefPrefix(ipdef);
1779
    const char *forwardIf = virNetworkDefForwardIf(network->def, 0);
1780 1781 1782 1783 1784 1785

    if (prefix >= 0) {
        iptablesRemoveForwardAllowIn(driver->iptables,
                                     &ipdef->address,
                                     prefix,
                                     network->def->bridge,
1786
                                     forwardIf);
1787 1788 1789 1790 1791

        iptablesRemoveForwardAllowOut(driver->iptables,
                                      &ipdef->address,
                                      prefix,
                                      network->def->bridge,
1792
                                      forwardIf);
1793 1794 1795
    }
}

1796 1797
/* Add all once/network rules required for IPv6.
 * If no IPv6 addresses are defined and <network ipv6='yes'> is
G
Gene Czarcinski 已提交
1798 1799
 * specified, then allow IPv6 commuinications between virtual systems.
 * If any IPv6 addresses are defined, then add the rules for regular operation.
1800
 */
1801 1802 1803 1804 1805
static int
networkAddGeneralIp6tablesRules(struct network_driver *driver,
                               virNetworkObjPtr network)
{

1806 1807
    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0) &&
        !network->def->ipv6nogw) {
1808
        return 0;
1809
    }
1810 1811 1812 1813 1814

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

    if (iptablesAddForwardRejectOut(driver->iptables, AF_INET6,
                                    network->def->bridge) < 0) {
1815 1816 1817
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to block outbound traffic from '%s'"),
                       network->def->bridge);
1818 1819 1820 1821 1822
        goto err1;
    }

    if (iptablesAddForwardRejectIn(driver->iptables, AF_INET6,
                                   network->def->bridge) < 0) {
1823 1824 1825
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to block inbound traffic to '%s'"),
                       network->def->bridge);
1826 1827 1828 1829 1830 1831
        goto err2;
    }

    /* Allow traffic between guests on the same bridge */
    if (iptablesAddForwardAllowCross(driver->iptables, AF_INET6,
                                     network->def->bridge) < 0) {
1832 1833 1834
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to allow cross bridge traffic on '%s'"),
                       network->def->bridge);
1835 1836 1837
        goto err3;
    }

1838 1839 1840 1841
    /* if no IPv6 addresses are defined, we are done. */
    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0))
        return 0;

1842 1843 1844
    /* allow DNS over IPv6 */
    if (iptablesAddTcpInput(driver->iptables, AF_INET6,
                            network->def->bridge, 53) < 0) {
1845 1846 1847
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to allow DNS requests from '%s'"),
                       network->def->bridge);
1848 1849 1850 1851 1852
        goto err4;
    }

    if (iptablesAddUdpInput(driver->iptables, AF_INET6,
                            network->def->bridge, 53) < 0) {
1853 1854 1855
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to allow DNS requests from '%s'"),
                       network->def->bridge);
1856 1857 1858
        goto err5;
    }

G
Gene Czarcinski 已提交
1859 1860 1861 1862 1863 1864 1865 1866
    if (iptablesAddUdpInput(driver->iptables, AF_INET6,
                            network->def->bridge, 547) < 0) {
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add ip6tables rule to allow DHCP6 requests from '%s'"),
                       network->def->bridge);
        goto err6;
    }

1867 1868 1869
    return 0;

    /* unwind in reverse order from the point of failure */
G
Gene Czarcinski 已提交
1870 1871
err6:
    iptablesRemoveUdpInput(driver->iptables, AF_INET6, network->def->bridge, 53);
1872 1873 1874 1875
err5:
    iptablesRemoveTcpInput(driver->iptables, AF_INET6, network->def->bridge, 53);
err4:
    iptablesRemoveForwardAllowCross(driver->iptables, AF_INET6, network->def->bridge);
1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887
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)
{
1888
    if (!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0) &&
G
Gene Czarcinski 已提交
1889
        !network->def->ipv6nogw) {
1890
        return;
G
Gene Czarcinski 已提交
1891
    }
1892
    if (virNetworkDefGetIpByIndex(network->def, AF_INET6, 0)) {
G
Gene Czarcinski 已提交
1893
        iptablesRemoveUdpInput(driver->iptables, AF_INET6, network->def->bridge, 547);
1894 1895 1896
        iptablesRemoveUdpInput(driver->iptables, AF_INET6, network->def->bridge, 53);
        iptablesRemoveTcpInput(driver->iptables, AF_INET6, network->def->bridge, 53);
    }
1897

1898 1899 1900
    /* the following rules are there if no IPv6 address has been defined
     * but network->def->ipv6nogw == true
     */
1901 1902 1903 1904 1905
    iptablesRemoveForwardAllowCross(driver->iptables, AF_INET6, network->def->bridge);
    iptablesRemoveForwardRejectIn(driver->iptables, AF_INET6, network->def->bridge);
    iptablesRemoveForwardRejectOut(driver->iptables, AF_INET6, network->def->bridge);
}

1906
static int
1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920
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;
    }
1921 1922

    /* allow DHCP requests through to dnsmasq */
1923

1924 1925
    if (iptablesAddTcpInput(driver->iptables, AF_INET,
                            network->def->bridge, 67) < 0) {
1926 1927 1928
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow DHCP requests from '%s'"),
                       network->def->bridge);
1929 1930 1931
        goto err1;
    }

1932 1933
    if (iptablesAddUdpInput(driver->iptables, AF_INET,
                            network->def->bridge, 67) < 0) {
1934 1935 1936
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow DHCP requests from '%s'"),
                       network->def->bridge);
1937 1938 1939
        goto err2;
    }

1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950
    /* 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);
1951
        VIR_WARN("May need to update iptables package & kernel to support CHECKSUM rule.");
1952 1953
    }

1954
    /* allow DNS requests through to dnsmasq */
1955 1956
    if (iptablesAddTcpInput(driver->iptables, AF_INET,
                            network->def->bridge, 53) < 0) {
1957 1958 1959
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow DNS requests from '%s'"),
                       network->def->bridge);
1960 1961 1962
        goto err3;
    }

1963 1964
    if (iptablesAddUdpInput(driver->iptables, AF_INET,
                            network->def->bridge, 53) < 0) {
1965 1966 1967
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow DNS requests from '%s'"),
                       network->def->bridge);
1968 1969 1970
        goto err4;
    }

1971 1972
    /* allow TFTP requests through to dnsmasq if necessary */
    if (ipv4def && ipv4def->tftproot &&
1973 1974
        iptablesAddUdpInput(driver->iptables, AF_INET,
                            network->def->bridge, 69) < 0) {
1975 1976 1977
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow TFTP requests from '%s'"),
                       network->def->bridge);
1978
        goto err5;
1979 1980
    }

1981 1982
    /* Catch all rules to block forwarding to/from bridges */

1983 1984
    if (iptablesAddForwardRejectOut(driver->iptables, AF_INET,
                                    network->def->bridge) < 0) {
1985 1986 1987
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to block outbound traffic from '%s'"),
                       network->def->bridge);
1988
        goto err6;
1989 1990
    }

1991 1992
    if (iptablesAddForwardRejectIn(driver->iptables, AF_INET,
                                   network->def->bridge) < 0) {
1993 1994 1995
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to block inbound traffic to '%s'"),
                       network->def->bridge);
1996
        goto err7;
1997 1998 1999
    }

    /* Allow traffic between guests on the same bridge */
2000 2001
    if (iptablesAddForwardAllowCross(driver->iptables, AF_INET,
                                     network->def->bridge) < 0) {
2002 2003 2004
        virReportError(VIR_ERR_SYSTEM_ERROR,
                       _("failed to add iptables rule to allow cross bridge traffic on '%s'"),
                       network->def->bridge);
2005
        goto err8;
2006 2007
    }

2008 2009 2010 2011 2012
    /* add IPv6 general rules, if needed */
    if (networkAddGeneralIp6tablesRules(driver, network) < 0) {
        goto err9;
    }

2013
    return 0;
2014

2015
    /* unwind in reverse order from the point of failure */
2016 2017
err9:
    iptablesRemoveForwardAllowCross(driver->iptables, AF_INET, network->def->bridge);
2018
err8:
2019
    iptablesRemoveForwardRejectIn(driver->iptables, AF_INET, network->def->bridge);
2020
err7:
2021
    iptablesRemoveForwardRejectOut(driver->iptables, AF_INET, network->def->bridge);
2022 2023
err6:
    if (ipv4def && ipv4def->tftproot) {
2024
        iptablesRemoveUdpInput(driver->iptables, AF_INET, network->def->bridge, 69);
2025
    }
2026
err5:
2027
    iptablesRemoveUdpInput(driver->iptables, AF_INET, network->def->bridge, 53);
2028
err4:
2029
    iptablesRemoveTcpInput(driver->iptables, AF_INET, network->def->bridge, 53);
2030
err3:
2031
    iptablesRemoveUdpInput(driver->iptables, AF_INET, network->def->bridge, 67);
2032
err2:
2033
    iptablesRemoveTcpInput(driver->iptables, AF_INET, network->def->bridge, 67);
2034
err1:
2035
    return -1;
2036 2037 2038
}

static void
2039 2040 2041 2042 2043
networkRemoveGeneralIptablesRules(struct network_driver *driver,
                                  virNetworkObjPtr network)
{
    int ii;
    virNetworkIpDefPtr ipv4def;
2044

2045 2046
    networkRemoveGeneralIp6tablesRules(driver, network);

2047 2048 2049 2050 2051
    for (ii = 0;
         (ipv4def = virNetworkDefGetIpByIndex(network->def, AF_INET, ii));
         ii++) {
        if (ipv4def->nranges || ipv4def->nhosts || ipv4def->tftproot)
            break;
2052
    }
2053

2054 2055 2056
    iptablesRemoveForwardAllowCross(driver->iptables, AF_INET, network->def->bridge);
    iptablesRemoveForwardRejectIn(driver->iptables, AF_INET, network->def->bridge);
    iptablesRemoveForwardRejectOut(driver->iptables, AF_INET, network->def->bridge);
2057
    if (ipv4def && ipv4def->tftproot) {
2058
        iptablesRemoveUdpInput(driver->iptables, AF_INET, network->def->bridge, 69);
2059
    }
2060 2061
    iptablesRemoveUdpInput(driver->iptables, AF_INET, network->def->bridge, 53);
    iptablesRemoveTcpInput(driver->iptables, AF_INET, network->def->bridge, 53);
2062 2063 2064 2065
    if (ipv4def && (ipv4def->nranges || ipv4def->nhosts)) {
        iptablesRemoveOutputFixUdpChecksum(driver->iptables,
                                           network->def->bridge, 68);
    }
2066 2067
    iptablesRemoveUdpInput(driver->iptables, AF_INET, network->def->bridge, 67);
    iptablesRemoveTcpInput(driver->iptables, AF_INET, network->def->bridge, 67);
2068 2069
}

2070 2071 2072 2073 2074
static int
networkAddIpSpecificIptablesRules(struct network_driver *driver,
                                  virNetworkObjPtr network,
                                  virNetworkIpDefPtr ipdef)
{
2075 2076 2077
    /* NB: in the case of IPv6, routing rules are added when the
     * forward mode is NAT. This is because IPv6 has no NAT.
     */
2078

2079
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) {
2080
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2081
            return networkAddMasqueradingIptablesRules(driver, network, ipdef);
2082
        else if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2083
            return networkAddRoutingIptablesRules(driver, network, ipdef);
2084
    } else if (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2085 2086
        return networkAddRoutingIptablesRules(driver, network, ipdef);
    }
2087 2088 2089 2090 2091 2092 2093 2094
    return 0;
}

static void
networkRemoveIpSpecificIptablesRules(struct network_driver *driver,
                                     virNetworkObjPtr network,
                                     virNetworkIpDefPtr ipdef)
{
2095
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) {
2096
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2097
            networkRemoveMasqueradingIptablesRules(driver, network, ipdef);
2098
        else if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2099
            networkRemoveRoutingIptablesRules(driver, network, ipdef);
2100
    } else if (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2101
        networkRemoveRoutingIptablesRules(driver, network, ipdef);
2102
    }
2103 2104 2105 2106 2107 2108 2109 2110 2111
}

/* 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;
2112
    virErrorPtr orig_error;
2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128

    /* 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:
2129 2130 2131
    /* store the previous error message before attempting removal of rules */
    orig_error = virSaveLastError();

2132 2133 2134 2135 2136 2137 2138 2139 2140
    /* 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);
2141 2142 2143

    /* return the original error */
    virSetError(orig_error);
2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162
    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);
}

2163 2164 2165 2166 2167
static void
networkReloadIptablesRules(struct network_driver *driver)
{
    unsigned int i;

2168
    VIR_INFO("Reloading iptables rules");
2169 2170

    for (i = 0 ; i < driver->networks.count ; i++) {
2171 2172 2173 2174
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
2175 2176 2177
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
2178 2179 2180 2181 2182
            /* 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) {
2183 2184
                /* failed to add but already logged */
            }
2185
        }
2186
        virNetworkObjUnlock(network);
2187 2188 2189
    }
}

2190
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
2191
static int
2192
networkEnableIpForwarding(bool enableIPv4, bool enableIPv6)
2193
{
2194 2195 2196 2197 2198 2199
    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;
2200 2201
}

2202 2203
#define SYSCTL_PATH "/proc/sys"

2204 2205
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
2206 2207 2208 2209
{
    char *field = NULL;
    int ret = -1;

2210 2211 2212 2213 2214 2215 2216 2217 2218
    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;
        }
2219

2220 2221 2222 2223 2224 2225
        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;
        }
2226

2227 2228 2229 2230 2231 2232 2233
        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);
2234 2235
    }

2236 2237 2238 2239 2240 2241 2242 2243 2244
    /* 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) {
2245
        virReportOOMError();
2246 2247 2248
        goto cleanup;
    }

2249
    if (virFileWriteStr(field, "0", 0) < 0) {
2250
        virReportSystemError(errno,
2251 2252 2253 2254 2255
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

2256 2257 2258 2259 2260
    /* 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) {
2261
        virReportOOMError();
2262 2263 2264
        goto cleanup;
    }

2265
    if (virFileWriteStr(field, "0", 0) < 0) {
2266
        virReportSystemError(errno,
2267
                             _("cannot disable %s"), field);
2268 2269 2270 2271 2272 2273 2274 2275 2276
        goto cleanup;
    }

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

2277 2278 2279 2280 2281 2282
#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
 */
2283 2284
static int
networkCheckRouteCollision(virNetworkObjPtr network)
2285
{
2286
    int ret = 0, len;
2287 2288 2289 2290 2291
    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)
2292
        goto out;
2293 2294 2295 2296 2297 2298 2299

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

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

2300
    if (!STRPREFIX(buf, "Iface"))
2301 2302 2303 2304 2305 2306 2307 2308 2309 2310
        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;
2311 2312
        virNetworkIpDefPtr ipdef;
        int num, ii;
2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340

        /* 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;

2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358
        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)) {
2359 2360 2361
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Network is already in use by interface %s"),
                               iface);
2362 2363 2364
                ret = -1;
                goto out;
            }
2365 2366 2367 2368 2369 2370 2371 2372
        }
    }

out:
    VIR_FREE(buf);
    return ret;
}

2373
static int
D
Daniel P. Berrange 已提交
2374
networkAddAddrToBridge(virNetworkObjPtr network,
2375
                       virNetworkIpDefPtr ipdef)
2376
{
2377 2378 2379
    int prefix = virNetworkIpDefPrefix(ipdef);

    if (prefix < 0) {
2380 2381 2382
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
2383 2384 2385
        return -1;
    }

2386 2387
    if (virNetDevSetIPv4Address(network->def->bridge,
                                &ipdef->address, prefix) < 0)
2388 2389 2390 2391 2392 2393
        return -1;

    return 0;
}

static int
2394
networkStartNetworkVirtual(struct network_driver *driver,
2395 2396
                          virNetworkObjPtr network)
{
2397
    int ii;
2398
    bool v4present = false, v6present = false;
2399 2400
    virErrorPtr save_err = NULL;
    virNetworkIpDefPtr ipdef;
2401
    char *macTapIfName = NULL;
2402
    int tapfd = -1;
2403

2404 2405
    /* Check to see if any network IP collides with an existing route */
    if (networkCheckRouteCollision(network) < 0)
2406 2407
        return -1;

2408
    /* Create and configure the bridge device */
2409
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
2410 2411
        return -1;

2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423
    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;
        }
2424
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2425
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
2426
                                           &macTapIfName, &network->def->mac,
2427 2428 2429 2430
                                           NULL, &tapfd, NULL, NULL,
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2431 2432 2433 2434 2435
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2436
    /* Set bridge options */
2437 2438 2439 2440

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2441
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2442
                                   network->def->delay * 1000) < 0)
2443
        goto err1;
2444

2445
    if (virNetDevBridgeSetSTP(network->def->bridge,
2446
                              network->def->stp ? true : false) < 0)
2447
        goto err1;
2448

2449 2450 2451 2452
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2453
        goto err1;
2454

2455 2456 2457 2458 2459 2460 2461
    /* Add "once per network" rules */
    if (networkAddIptablesRules(driver, network) < 0)
        goto err1;

    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, ii));
         ii++) {
2462
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2463
            v4present = true;
2464
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2465
            v6present = true;
2466

2467
        /* Add the IP address/netmask to the bridge */
D
Daniel P. Berrange 已提交
2468
        if (networkAddAddrToBridge(network, ipdef) < 0) {
2469
            goto err2;
2470
        }
2471 2472
    }

2473
    /* Bring up the bridge interface */
2474
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2475
        goto err2;
2476

2477 2478
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2479
        networkEnableIpForwarding(v4present, v6present) < 0) {
2480
        virReportSystemError(errno, "%s",
2481
                             _("failed to enable IP forwarding"));
2482
        goto err3;
2483 2484
    }

2485

2486
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2487 2488
    if ((v4present || v6present) &&
        networkStartDhcpDaemon(driver, network) < 0)
2489
        goto err3;
2490

2491
    /* start radvd if there are any ipv6 addresses */
G
Gene Czarcinski 已提交
2492
    if (v6present && networkStartRadvd(driver, network) < 0)
2493 2494
        goto err4;

2495 2496 2497 2498 2499 2500 2501 2502 2503
    /* DAD has happened (dnsmasq waits for it), dnsmasq is now bound to the
     * bridge's IPv6 address, so we can now set the dummy tun down.
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
            goto err4;
        VIR_FORCE_CLOSE(tapfd);
    }

2504 2505
    if (virNetDevBandwidthSet(network->def->bridge,
                              network->def->bandwidth, true) < 0) {
2506 2507 2508
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("cannot set bandwidth limits on %s"),
                       network->def->bridge);
2509 2510 2511
        goto err5;
    }

2512
    VIR_FREE(macTapIfName);
2513 2514 2515

    return 0;

2516
 err5:
2517
    virNetDevBandwidthClear(network->def->bridge);
2518

2519 2520 2521 2522
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2523 2524 2525 2526 2527
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2528 2529 2530
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2531
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2532

2533 2534 2535 2536 2537 2538
 err2:
    if (!save_err)
        save_err = virSaveLastError();
    networkRemoveIptablesRules(driver, network);

 err1:
2539 2540 2541
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2542
    if (macTapIfName) {
2543
        VIR_FORCE_CLOSE(tapfd);
H
Hu Tao 已提交
2544 2545 2546
        ignore_value(virNetDevTapDelete(macTapIfName));
        VIR_FREE(macTapIfName);
    }
2547 2548

 err0:
2549 2550
    if (!save_err)
        save_err = virSaveLastError();
2551
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2552

2553 2554 2555 2556
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2557
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2558 2559 2560
    return -1;
}

2561
static int networkShutdownNetworkVirtual(struct network_driver *driver,
2562 2563
                                        virNetworkObjPtr network)
{
2564
    virNetDevBandwidthClear(network->def->bridge);
2565

2566 2567 2568 2569 2570 2571 2572 2573
    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 {
2574
            virPidFileDelete(NETWORK_PID_DIR, radvdpidbase);
2575 2576 2577 2578
            VIR_FREE(radvdpidbase);
        }
    }

2579 2580 2581
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2582
    if (network->def->mac_specified) {
2583
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2584 2585 2586
        if (!macTapIfName) {
            virReportOOMError();
        } else {
2587
            ignore_value(virNetDevTapDelete(macTapIfName));
2588 2589 2590 2591
            VIR_FREE(macTapIfName);
        }
    }

2592
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2593

2594 2595
    networkRemoveIptablesRules(driver, network);

2596
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2597

2598
    /* See if its still alive and really really kill it */
2599
    if (network->dnsmasqPid > 0 &&
2600
        (kill(network->dnsmasqPid, 0) == 0))
2601 2602
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2603 2604 2605 2606 2607 2608

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

2609 2610 2611 2612 2613 2614 2615 2616
    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
2617
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2618 2619 2620 2621 2622 2623 2624 2625 2626 2627
     * 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
2628
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641
     * 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)) {
2642 2643
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2644 2645 2646
        return -1;
    }

2647 2648 2649
    if (virNetworkObjSetDefTransient(network, true) < 0)
        return -1;

2650
    switch (network->def->forward.type) {
2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661

    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:
2662
    case VIR_NETWORK_FORWARD_HOSTDEV:
2663 2664 2665 2666
        ret = networkStartNetworkExternal(driver, network);
        break;
    }

2667 2668
    if (ret < 0) {
        virNetworkObjUnsetDefTransient(network);
2669
        return ret;
2670
    }
2671 2672 2673 2674

    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2675
    if ((ret = virNetworkSaveStatus(NETWORK_STATE_DIR, network)) < 0) {
2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711
        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);

2712
    switch (network->def->forward.type) {
2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723

    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:
2724
    case VIR_NETWORK_FORWARD_HOSTDEV:
2725 2726 2727 2728
        ret = networkShutdownNetworkExternal(driver, network);
        break;
    }

2729
    network->active = 0;
2730
    virNetworkObjUnsetDefTransient(network);
2731
    return ret;
2732 2733 2734
}


2735 2736 2737 2738 2739
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
                                         const unsigned char *uuid) {
    struct network_driver *driver = conn->networkPrivateData;
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2740

2741
    networkDriverLock(driver);
2742
    network = virNetworkFindByUUID(&driver->networks, uuid);
2743
    networkDriverUnlock(driver);
2744
    if (!network) {
2745 2746
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
2747
        goto cleanup;
2748 2749
    }

2750 2751 2752
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

cleanup:
2753 2754
    if (network)
        virNetworkObjUnlock(network);
2755
    return ret;
2756 2757
}

2758 2759 2760 2761 2762 2763
static virNetworkPtr networkLookupByName(virConnectPtr conn,
                                         const char *name) {
    struct network_driver *driver = conn->networkPrivateData;
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2764
    networkDriverLock(driver);
2765
    network = virNetworkFindByName(&driver->networks, name);
2766
    networkDriverUnlock(driver);
2767
    if (!network) {
2768 2769
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2770
        goto cleanup;
2771 2772
    }

2773 2774 2775
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

cleanup:
2776 2777
    if (network)
        virNetworkObjUnlock(network);
2778
    return ret;
2779 2780
}

2781 2782 2783
static virDrvOpenStatus networkOpen(virConnectPtr conn,
                                    virConnectAuthPtr auth ATTRIBUTE_UNUSED,
                                    unsigned int flags)
2784 2785 2786
{
    virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);

2787 2788 2789 2790 2791 2792 2793
    if (!driverState)
        return VIR_DRV_OPEN_DECLINED;

    conn->networkPrivateData = driverState;
    return VIR_DRV_OPEN_SUCCESS;
}

2794
static int networkClose(virConnectPtr conn) {
2795 2796 2797 2798
    conn->networkPrivateData = NULL;
    return 0;
}

2799
static int networkConnectNumOfNetworks(virConnectPtr conn) {
2800
    int nactive = 0, i;
2801
    struct network_driver *driver = conn->networkPrivateData;
2802

2803 2804 2805
    networkDriverLock(driver);
    for (i = 0 ; i < driver->networks.count ; i++) {
        virNetworkObjLock(driver->networks.objs[i]);
D
Daniel P. Berrange 已提交
2806
        if (virNetworkObjIsActive(driver->networks.objs[i]))
2807
            nactive++;
2808 2809 2810
        virNetworkObjUnlock(driver->networks.objs[i]);
    }
    networkDriverUnlock(driver);
2811

2812 2813 2814
    return nactive;
}

2815
static int networkConnectListNetworks(virConnectPtr conn, char **const names, int nnames) {
2816
    struct network_driver *driver = conn->networkPrivateData;
2817
    int got = 0, i;
2818

2819
    networkDriverLock(driver);
2820
    for (i = 0 ; i < driver->networks.count && got < nnames ; i++) {
2821
        virNetworkObjLock(driver->networks.objs[i]);
D
Daniel P. Berrange 已提交
2822
        if (virNetworkObjIsActive(driver->networks.objs[i])) {
2823
            if (!(names[got] = strdup(driver->networks.objs[i]->def->name))) {
2824
                virNetworkObjUnlock(driver->networks.objs[i]);
2825
                virReportOOMError();
2826 2827 2828 2829
                goto cleanup;
            }
            got++;
        }
2830
        virNetworkObjUnlock(driver->networks.objs[i]);
2831
    }
2832 2833
    networkDriverUnlock(driver);

2834 2835 2836
    return got;

 cleanup:
2837
    networkDriverUnlock(driver);
2838 2839 2840 2841 2842
    for (i = 0 ; i < got ; i++)
        VIR_FREE(names[i]);
    return -1;
}

2843
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn) {
2844
    int ninactive = 0, i;
2845
    struct network_driver *driver = conn->networkPrivateData;
2846

2847 2848 2849
    networkDriverLock(driver);
    for (i = 0 ; i < driver->networks.count ; i++) {
        virNetworkObjLock(driver->networks.objs[i]);
D
Daniel P. Berrange 已提交
2850
        if (!virNetworkObjIsActive(driver->networks.objs[i]))
2851
            ninactive++;
2852 2853 2854
        virNetworkObjUnlock(driver->networks.objs[i]);
    }
    networkDriverUnlock(driver);
2855

2856 2857 2858
    return ninactive;
}

2859
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames) {
2860
    struct network_driver *driver = conn->networkPrivateData;
2861
    int got = 0, i;
2862

2863
    networkDriverLock(driver);
2864
    for (i = 0 ; i < driver->networks.count && got < nnames ; i++) {
2865
        virNetworkObjLock(driver->networks.objs[i]);
D
Daniel P. Berrange 已提交
2866
        if (!virNetworkObjIsActive(driver->networks.objs[i])) {
2867
            if (!(names[got] = strdup(driver->networks.objs[i]->def->name))) {
2868
                virNetworkObjUnlock(driver->networks.objs[i]);
2869
                virReportOOMError();
2870 2871 2872 2873
                goto cleanup;
            }
            got++;
        }
2874
        virNetworkObjUnlock(driver->networks.objs[i]);
2875
    }
2876
    networkDriverUnlock(driver);
2877 2878 2879
    return got;

 cleanup:
2880
    networkDriverUnlock(driver);
2881 2882 2883 2884 2885
    for (i = 0 ; i < got ; i++)
        VIR_FREE(names[i]);
    return -1;
}

2886
static int
2887 2888 2889
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901
{
    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;
}
2902 2903 2904

static int networkIsActive(virNetworkPtr net)
{
2905
    struct network_driver *driver = net->conn->networkPrivateData;
2906 2907 2908 2909 2910 2911 2912
    virNetworkObjPtr obj;
    int ret = -1;

    networkDriverLock(driver);
    obj = virNetworkFindByUUID(&driver->networks, net->uuid);
    networkDriverUnlock(driver);
    if (!obj) {
2913
        virReportError(VIR_ERR_NO_NETWORK, NULL);
2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925
        goto cleanup;
    }
    ret = virNetworkObjIsActive(obj);

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

static int networkIsPersistent(virNetworkPtr net)
{
2926
    struct network_driver *driver = net->conn->networkPrivateData;
2927 2928 2929 2930 2931 2932 2933
    virNetworkObjPtr obj;
    int ret = -1;

    networkDriverLock(driver);
    obj = virNetworkFindByUUID(&driver->networks, net->uuid);
    networkDriverUnlock(driver);
    if (!obj) {
2934
        virReportError(VIR_ERR_NO_NETWORK, NULL);
2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945
        goto cleanup;
    }
    ret = obj->persistent;

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


2946
static int
2947 2948 2949
networkValidate(struct network_driver *driver,
                virNetworkDefPtr def,
                bool check_active)
2950 2951
{
    int ii;
2952 2953
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2954
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
2955
    bool ipv4def = false, ipv6def = false;
2956 2957 2958 2959 2960 2961 2962 2963

    /* check for duplicate networks */
    if (virNetworkObjIsDuplicate(&driver->networks, def, check_active) < 0)
        return -1;

    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
2964 2965 2966
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2967 2968 2969 2970 2971

        if (virNetworkSetBridgeName(&driver->networks, def, 1))
            return -1;

        virNetworkSetBridgeMacAddr(def);
2972 2973 2974 2975 2976 2977 2978 2979 2980
    } else {
        /* They are also the only types that currently support setting
         * an IP address for the host-side device (bridge)
         */
        if (virNetworkDefGetIpByIndex(def, AF_UNSPEC, 0)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2981
                           virNetworkForwardTypeToString(def->forward.type));
2982 2983
            return -1;
        }
2984
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
2985 2986 2987 2988
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2989
                           virNetworkForwardTypeToString(def->forward.type));
2990 2991 2992 2993 2994 2995 2996
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2997
                           virNetworkForwardTypeToString(def->forward.type));
2998 2999
            return -1;
        }
3000 3001
    }

G
Gene Czarcinski 已提交
3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(def, AF_UNSPEC, ii));
         ii++) {
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("Multiple IPv4 dhcp sections found -- "
3013 3014
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031
                    return -1;
                } else {
                    ipv4def = true;
                }
            }
        }
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("Multiple IPv6 dhcp sections found -- "
                                 "dhcp is supported only for a "
                                 "single IPv6 address on each network"));
                    return -1;
                } else {
                    ipv6def = true;
                }
3032 3033 3034
            }
        }
    }
3035 3036 3037 3038 3039 3040

    /* 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.
     */

3041
    vlanAllowed = (def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
3042 3043 3044 3045
                   def->virtPortProfile &&
                   def->virtPortProfile->virtPortType == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH);

    vlanUsed = def->vlan.nTags > 0;
3046 3047 3048 3049 3050 3051 3052 3053
    for (ii = 0; ii < def->nPortGroups; ii++) {
        if (vlanUsed || def->portGroups[ii].vlan.nTags > 0) {
            /* anyone using this portgroup will get a vlan tag. Verify
             * that they will also be using an openvswitch connection,
             * as that is the only type of network that currently
             * supports a vlan tag.
             */
            if (def->portGroups[ii].virtPortProfile) {
3054
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3055 3056 3057 3058 3059 3060 3061 3062
                    def->portGroups[ii].virtPortProfile->virtPortType
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3063
        }
3064 3065 3066 3067 3068 3069
        if (def->portGroups[ii].isDefault) {
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3070
                               def->name, defaultPortGroup->name,
3071
                               def->portGroups[ii].name);
3072
                return -1;
3073
            }
3074
            defaultPortGroup = &def->portGroups[ii];
3075
        }
3076
    }
3077 3078 3079 3080 3081 3082 3083
    if (badVlanUse ||
        (vlanUsed && !vlanAllowed && !defaultPortGroup)) {
        /* NB: if defaultPortGroup is set, we don't directly look at
         * vlanUsed && !vlanAllowed, because the network will never be
         * used without having a portgroup added in, so all necessary
         * checks were done in the loop above.
         */
3084 3085 3086 3087 3088 3089 3090 3091 3092
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
    return 0;
}

3093
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml) {
3094
    struct network_driver *driver = conn->networkPrivateData;
3095
    virNetworkDefPtr def;
3096
    virNetworkObjPtr network = NULL;
3097
    virNetworkPtr ret = NULL;
3098

3099 3100
    networkDriverLock(driver);

3101
    if (!(def = virNetworkDefParseString(xml)))
3102
        goto cleanup;
3103

3104
    if (networkValidate(driver, def, true) < 0)
3105 3106
       goto cleanup;

3107 3108 3109 3110
    /* NB: "live" is false because this transient network hasn't yet
     * been started
     */
    if (!(network = virNetworkAssignDef(&driver->networks, def, false)))
3111 3112
        goto cleanup;
    def = NULL;
3113

3114
    if (networkStartNetwork(driver, network) < 0) {
3115 3116
        virNetworkRemoveInactive(&driver->networks,
                                 network);
3117
        network = NULL;
3118
        goto cleanup;
3119 3120
    }

3121
    VIR_INFO("Creating network '%s'", network->def->name);
3122 3123 3124 3125
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

cleanup:
    virNetworkDefFree(def);
3126 3127 3128
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
3129
    return ret;
3130 3131
}

3132
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml) {
3133
    struct network_driver *driver = conn->networkPrivateData;
3134
    virNetworkDefPtr def = NULL;
3135
    bool freeDef = true;
3136
    virNetworkObjPtr network = NULL;
3137
    virNetworkPtr ret = NULL;
3138

3139 3140
    networkDriverLock(driver);

3141
    if (!(def = virNetworkDefParseString(xml)))
3142
        goto cleanup;
3143

3144
    if (networkValidate(driver, def, false) < 0)
3145 3146
       goto cleanup;

3147 3148 3149 3150 3151 3152 3153 3154
    if ((network = virNetworkFindByName(&driver->networks, def->name))) {
        network->persistent = 1;
        if (virNetworkObjAssignDef(network, def, false) < 0)
            goto cleanup;
    } else {
        if (!(network = virNetworkAssignDef(&driver->networks, def, false)))
            goto cleanup;
    }
3155

3156 3157 3158
    /* define makes the network persistent - always */
    network->persistent = 1;

3159 3160
    /* def was asigned */
    freeDef = false;
3161 3162

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3163 3164 3165 3166 3167 3168 3169 3170
        if (!virNetworkObjIsActive(network)) {
            virNetworkRemoveInactive(&driver->networks, network);
            network = NULL;
            goto cleanup;
        }
        network->persistent = 0;
        virNetworkDefFree(network->newDef);
        network->newDef = NULL;
3171 3172 3173
        goto cleanup;
    }

3174 3175
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
3176 3177

cleanup:
3178 3179
    if (freeDef)
       virNetworkDefFree(def);
3180 3181 3182
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
3183
    return ret;
3184 3185
}

3186 3187
static int
networkUndefine(virNetworkPtr net) {
3188
    struct network_driver *driver = net->conn->networkPrivateData;
3189
    virNetworkObjPtr network;
3190
    int ret = -1;
3191
    bool active = false;
3192

3193 3194
    networkDriverLock(driver);

3195
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3196
    if (!network) {
3197 3198
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3199
        goto cleanup;
3200 3201
    }

3202 3203
    if (virNetworkObjIsActive(network))
        active = true;
3204

3205
    if (virNetworkDeleteConfig(driver->networkConfigDir,
3206 3207
                               driver->networkAutostartDir,
                               network) < 0)
3208
        goto cleanup;
3209

3210 3211
    /* make the network transient */
    network->persistent = 0;
3212
    network->autostart = 0;
3213 3214 3215
    virNetworkDefFree(network->newDef);
    network->newDef = NULL;

3216
    VIR_INFO("Undefining network '%s'", network->def->name);
3217 3218 3219 3220 3221
    if (!active) {
        if (networkRemoveInactive(driver, network) < 0) {
            network = NULL;
            goto cleanup;
        }
3222
        network = NULL;
3223 3224
    }

3225
    ret = 0;
3226

3227
cleanup:
3228 3229 3230
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
3231
    return ret;
3232 3233
}

3234 3235 3236 3237 3238 3239 3240 3241 3242 3243
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;
3244 3245 3246 3247
    int isActive, ret = -1, ii;
    virNetworkIpDefPtr ipdef;
    bool oldDhcpActive = false;

3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261

    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;
    }

3262 3263 3264 3265 3266 3267 3268 3269 3270 3271
    /* see if we are listening for dhcp pre-modification */
    for (ii = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, ii));
         ii++) {
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3272 3273 3274 3275
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
    */
    isActive = virNetworkObjIsActive(network);
3276 3277
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309
        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.
             */
3310
            if (networkRestartDhcpDaemon(driver, network) < 0)
3311 3312
                goto cleanup;

3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330
        } else if (section == VIR_NETWORK_SECTION_IP_DHCP_HOST) {
            /* if we previously weren't listening for dhcp and now we
             * are (or vice-versa) then we need to do a restart,
             * otherwise we just need to do a refresh (redo the config
             * files and send SIGHUP)
             */
            bool newDhcpActive = false;

            for (ii = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, ii));
                 ii++) {
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3331 3332
                 networkRestartDhcpDaemon(driver, network) < 0) ||
                networkRefreshDhcpDaemon(driver, network) < 0) {
3333 3334 3335 3336
                goto cleanup;
            }

        } else if (section == VIR_NETWORK_SECTION_DNS_HOST ||
3337 3338 3339 3340 3341 3342
                   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.
             */
3343
            if (networkRefreshDhcpDaemon(driver, network) < 0)
3344 3345 3346 3347 3348 3349 3350 3351
                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.
             */
3352
            if (networkRefreshRadvd(driver, network) < 0)
3353 3354 3355
                goto cleanup;
        }

3356 3357 3358
        if ((section == VIR_NETWORK_SECTION_IP ||
             section == VIR_NETWORK_SECTION_FORWARD ||
             section == VIR_NETWORK_SECTION_FORWARD_INTERFACE) &&
3359 3360 3361
           (network->def->forward.type == VIR_NETWORK_FORWARD_NONE ||
            network->def->forward.type == VIR_NETWORK_FORWARD_NAT ||
            network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380
            /* 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;
}

3381
static int networkCreate(virNetworkPtr net) {
3382 3383 3384
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
    int ret = -1;
3385

3386
    networkDriverLock(driver);
3387
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3388

3389
    if (!network) {
3390 3391
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3392
        goto cleanup;
3393 3394
    }

3395
    ret = networkStartNetwork(driver, network);
3396 3397

cleanup:
3398 3399
    if (network)
        virNetworkObjUnlock(network);
3400
    networkDriverUnlock(driver);
3401
    return ret;
3402 3403 3404
}

static int networkDestroy(virNetworkPtr net) {
3405 3406 3407
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
    int ret = -1;
3408

3409
    networkDriverLock(driver);
3410
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3411

3412
    if (!network) {
3413 3414
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3415
        goto cleanup;
3416 3417
    }

D
Daniel P. Berrange 已提交
3418
    if (!virNetworkObjIsActive(network)) {
3419 3420
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is not active"));
3421 3422 3423
        goto cleanup;
    }

3424 3425 3426
    if ((ret = networkShutdownNetwork(driver, network)) < 0)
        goto cleanup;

3427
    if (!network->persistent) {
3428 3429 3430 3431 3432
        if (networkRemoveInactive(driver, network) < 0) {
            network = NULL;
            ret = -1;
            goto cleanup;
        }
3433 3434
        network = NULL;
    }
3435

3436
cleanup:
3437 3438
    if (network)
        virNetworkObjUnlock(network);
3439
    networkDriverUnlock(driver);
3440 3441 3442
    return ret;
}

3443
static char *networkGetXMLDesc(virNetworkPtr net,
3444
                               unsigned int flags)
3445
{
3446 3447
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
3448
    virNetworkDefPtr def;
3449
    char *ret = NULL;
3450

3451
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3452

3453
    networkDriverLock(driver);
3454
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3455 3456
    networkDriverUnlock(driver);

3457
    if (!network) {
3458 3459
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3460
        goto cleanup;
3461 3462
    }

3463 3464 3465 3466 3467 3468
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3469 3470

cleanup:
3471 3472
    if (network)
        virNetworkObjUnlock(network);
3473
    return ret;
3474 3475 3476
}

static char *networkGetBridgeName(virNetworkPtr net) {
3477 3478 3479 3480
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
    char *bridge = NULL;

3481
    networkDriverLock(driver);
3482
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3483 3484
    networkDriverUnlock(driver);

3485
    if (!network) {
3486 3487
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching id"));
3488
        goto cleanup;
3489 3490
    }

3491
    if (!(network->def->bridge)) {
3492 3493 3494
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3495 3496 3497
        goto cleanup;
    }

3498
    bridge = strdup(network->def->bridge);
3499
    if (!bridge)
3500
        virReportOOMError();
3501 3502

cleanup:
3503 3504
    if (network)
        virNetworkObjUnlock(network);
3505 3506 3507 3508 3509
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
                             int *autostart) {
3510 3511 3512
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
    int ret = -1;
3513

3514
    networkDriverLock(driver);
3515
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3516
    networkDriverUnlock(driver);
3517
    if (!network) {
3518 3519
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3520
        goto cleanup;
3521 3522 3523
    }

    *autostart = network->autostart;
3524
    ret = 0;
3525

3526
cleanup:
3527 3528
    if (network)
        virNetworkObjUnlock(network);
3529
    return ret;
3530 3531 3532
}

static int networkSetAutostart(virNetworkPtr net,
3533
                               int autostart) {
3534 3535
    struct network_driver *driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
3536
    char *configFile = NULL, *autostartLink = NULL;
3537
    int ret = -1;
3538

3539
    networkDriverLock(driver);
3540
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3541

3542
    if (!network) {
3543 3544
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3545
        goto cleanup;
3546 3547
    }

3548
    if (!network->persistent) {
3549 3550
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3551 3552 3553
        goto cleanup;
    }

3554 3555
    autostart = (autostart != 0);

3556
    if (network->autostart != autostart) {
3557
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3558
            goto cleanup;
3559
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3560 3561
            goto cleanup;

3562
        if (autostart) {
3563
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3564
                virReportSystemError(errno,
3565 3566
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3567 3568
                goto cleanup;
            }
3569

3570
            if (symlink(configFile, autostartLink) < 0) {
3571
                virReportSystemError(errno,
3572
                                     _("Failed to create symlink '%s' to '%s'"),
3573
                                     autostartLink, configFile);
3574 3575 3576
                goto cleanup;
            }
        } else {
3577
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3578
                virReportSystemError(errno,
3579
                                     _("Failed to delete symlink '%s'"),
3580
                                     autostartLink);
3581 3582
                goto cleanup;
            }
3583 3584
        }

3585
        network->autostart = autostart;
3586
    }
3587
    ret = 0;
3588

3589
cleanup:
3590 3591
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3592 3593
    if (network)
        virNetworkObjUnlock(network);
3594
    networkDriverUnlock(driver);
3595
    return ret;
3596 3597 3598 3599 3600
}


static virNetworkDriver networkDriver = {
    "Network",
3601 3602 3603 3604 3605 3606 3607
    .networkOpen = networkOpen, /* 0.2.0 */
    .networkClose = networkClose, /* 0.2.0 */
    .connectNumOfNetworks = networkConnectNumOfNetworks, /* 0.2.0 */
    .connectListNetworks = networkConnectListNetworks, /* 0.2.0 */
    .connectNumOfDefinedNetworks = networkConnectNumOfDefinedNetworks, /* 0.2.0 */
    .connectListDefinedNetworks = networkConnectListDefinedNetworks, /* 0.2.0 */
    .connectListAllNetworks = networkConnectListAllNetworks, /* 0.10.2 */
3608 3609
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3610 3611
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3612
    .networkUndefine = networkUndefine, /* 0.2.0 */
3613
    .networkUpdate = networkUpdate, /* 0.10.2 */
3614
    .networkCreate = networkCreate, /* 0.2.0 */
3615 3616 3617 3618 3619 3620 3621
    .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 */
3622 3623 3624
};

static virStateDriver networkStateDriver = {
3625
    .name = "Network",
3626 3627 3628
    .stateInitialize  = networkStateInitialize,
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3629 3630 3631 3632 3633 3634 3635
};

int networkRegister(void) {
    virRegisterNetworkDriver(&networkDriver);
    virRegisterStateDriver(&networkStateDriver);
    return 0;
}
3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646

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

/* 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.
 */

3647 3648 3649 3650 3651 3652 3653 3654 3655
/* 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;
3656
    virPCIDeviceAddressPtr *virt_fns;
3657 3658
    int ret = -1, ii = 0;

3659
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev,
3660
                                      &vfname, &virt_fns, &num_virt_fns)) < 0) {
3661 3662
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Could not get Virtual functions on %s"),
3663
                       netdef->forward.pfs->dev);
3664 3665 3666 3667 3668 3669
        goto finish;
    }

    if (num_virt_fns == 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("No Vf's present on SRIOV PF %s"),
3670
                       netdef->forward.pfs->dev);
3671 3672 3673
       goto finish;
    }

3674
    if ((VIR_ALLOC_N(netdef->forward.ifs, num_virt_fns)) < 0) {
3675 3676 3677 3678
        virReportOOMError();
        goto finish;
    }

3679
    netdef->forward.nifs = num_virt_fns;
3680

3681 3682 3683 3684 3685 3686
    for (ii = 0; ii < netdef->forward.nifs; ii++) {
        if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) ||
            (netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) ||
            (netdef->forward.type == VIR_NETWORK_FORWARD_VEPA) ||
            (netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH)) {
            netdef->forward.ifs[ii].type = VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV;
3687
            if (vfname[ii]) {
3688 3689
                netdef->forward.ifs[ii].device.dev = strdup(vfname[ii]);
                if (!netdef->forward.ifs[ii].device.dev) {
3690 3691 3692 3693 3694 3695
                    virReportOOMError();
                    goto finish;
                }
            }
            else {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
3696
                               _("Direct mode types require interface names"));
3697 3698
                goto finish;
            }
3699
        }
3700
        else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3701
            /* VF's are always PCI devices */
3702 3703 3704 3705 3706
            netdef->forward.ifs[ii].type = VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI;
            netdef->forward.ifs[ii].device.pci.domain = virt_fns[ii]->domain;
            netdef->forward.ifs[ii].device.pci.bus = virt_fns[ii]->bus;
            netdef->forward.ifs[ii].device.pci.slot = virt_fns[ii]->slot;
            netdef->forward.ifs[ii].device.pci.function = virt_fns[ii]->function;
3707
        }
3708 3709 3710 3711
    }

    ret = 0;
finish:
3712
    for (ii = 0; ii < num_virt_fns; ii++) {
3713
        VIR_FREE(vfname[ii]);
3714 3715
        VIR_FREE(virt_fns[ii]);
    }
3716
    VIR_FREE(vfname);
3717
    VIR_FREE(virt_fns);
3718 3719 3720
    return ret;
}

3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735
/* 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;
3736 3737 3738
    enum virDomainNetType actualType = iface->type;
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
3739
    virNetDevBandwidthPtr bandwidth = NULL;
3740 3741 3742
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3743
    virNetworkForwardIfDefPtr dev = NULL;
3744
    int ii;
3745 3746 3747
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3748
        goto validate;
3749 3750 3751 3752 3753 3754 3755 3756

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

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
3757 3758 3759
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3760
        goto error;
3761 3762
    }
    netdef = network->def;
3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773

    /* 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.
     */
3774 3775 3776 3777 3778 3779 3780

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

    if (bandwidth) {
3781 3782 3783
        if (!iface->data.network.actual
            && (VIR_ALLOC(iface->data.network.actual) < 0)) {
            virReportOOMError();
3784
            goto error;
3785 3786
        }

3787
        if (virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
3788
                                   bandwidth) < 0)
3789
            goto error;
3790 3791
    }

3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809
    /* copy appropriate vlan info to actualNet */
    if (iface->vlan.nTags > 0)
        vlan = &iface->vlan;
    else if (portgroup && portgroup->vlan.nTags > 0)
        vlan = &portgroup->vlan;
    else if (netdef->vlan.nTags > 0)
        vlan = &netdef->vlan;

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

3810 3811 3812
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
3813 3814 3815 3816 3817 3818
        /* 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;
3819 3820 3821 3822

        if (networkPlugBandwidth(network, iface) < 0)
            goto error;

3823
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
3824
               netdef->bridge) {
3825 3826 3827 3828 3829

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

3830 3831
        if (!iface->data.network.actual
            && (VIR_ALLOC(iface->data.network.actual) < 0)) {
3832
            virReportOOMError();
3833
            goto error;
3834 3835
        }

3836
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
3837 3838 3839
        iface->data.network.actual->data.bridge.brname = strdup(netdef->bridge);
        if (!iface->data.network.actual->data.bridge.brname) {
            virReportOOMError();
3840
            goto error;
3841 3842
        }

3843 3844 3845 3846 3847 3848 3849 3850
        /* 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) {
3851
            goto error;
3852 3853 3854 3855 3856 3857 3858 3859 3860 3861
        }
        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);
3862
                goto error;
3863 3864 3865
            }
        }

3866
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3867

3868 3869
        virDomainHostdevSubsysPciBackendType backend;

3870 3871 3872 3873 3874 3875 3876
        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;
3877
        if (netdef->forward.npfs > 0 && netdef->forward.nifs <= 0 &&
3878 3879 3880 3881 3882
            networkCreateInterfacePool(netdef) < 0) {
            goto error;
        }

        /* pick first dev with 0 connections */
3883 3884 3885
        for (ii = 0; ii < netdef->forward.nifs; ii++) {
            if (netdef->forward.ifs[ii].connections == 0) {
                dev = &netdef->forward.ifs[ii];
3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899
                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;
3900
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
3901
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
3902
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
3903

3904 3905 3906
        switch (netdef->forward.driverName)
        {
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
3907
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
3908 3909
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
3910
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
3911 3912
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
3913
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924
            break;
        default:
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unrecognized driver name value %d "
                             " in network '%s'"),
                           netdef->forward.driverName, netdef->name);
            goto error;
        }
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.backend
            = backend;

3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949
        /* 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;
            }
        }

3950 3951 3952 3953
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) ||
               (netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) ||
               (netdef->forward.type == VIR_NETWORK_FORWARD_VEPA) ||
               (netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH)) {
3954 3955 3956 3957 3958

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

3959 3960
        if (!iface->data.network.actual
            && (VIR_ALLOC(iface->data.network.actual) < 0)) {
3961
            virReportOOMError();
3962
            goto error;
3963 3964 3965
        }

        /* Set type=direct and appropriate <source mode='xxx'/> */
3966
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
3967
        switch (netdef->forward.type) {
3968
        case VIR_NETWORK_FORWARD_BRIDGE:
3969
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
3970 3971
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
3972
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
3973 3974
            break;
        case VIR_NETWORK_FORWARD_VEPA:
3975
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
3976 3977
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
3978
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
3979 3980 3981
            break;
        }

3982 3983 3984 3985 3986 3987 3988 3989
        /* 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) {
3990
            goto error;
3991
        }
3992
        virtport = iface->data.network.actual->virtPortProfile;
3993
        if (virtport) {
3994 3995 3996 3997 3998 3999 4000 4001
            /* 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);
4002
                goto error;
4003 4004
            }
        }
4005

4006 4007 4008
        /* 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).
         */
4009
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4010 4011 4012 4013
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4014
            goto error;
4015 4016 4017
        } else {
            /* pick an interface from the pool */

4018
            if (netdef->forward.npfs > 0 && netdef->forward.nifs == 0 &&
4019 4020 4021 4022
                networkCreateInterfacePool(netdef) < 0) {
                goto error;
            }

4023 4024 4025 4026 4027
            /* 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.
4028
             */
4029 4030
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4031 4032 4033
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4034

4035
                /* pick first dev with 0 connections */
4036 4037 4038
                for (ii = 0; ii < netdef->forward.nifs; ii++) {
                    if (netdef->forward.ifs[ii].connections == 0) {
                        dev = &netdef->forward.ifs[ii];
4039 4040 4041 4042 4043
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4044 4045 4046 4047
                dev = &netdef->forward.ifs[0];
                for (ii = 1; ii < netdef->forward.nifs; ii++) {
                    if (netdef->forward.ifs[ii].connections < dev->connections)
                        dev = &netdef->forward.ifs[ii];
4048 4049 4050 4051
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4052 4053 4054 4055
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4056
                goto error;
4057
            }
4058
            iface->data.network.actual->data.direct.linkdev = strdup(dev->device.dev);
4059 4060
            if (!iface->data.network.actual->data.direct.linkdev) {
                virReportOOMError();
4061
                goto error;
4062 4063 4064 4065
            }
        }
    }

4066
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4067
        goto error;
4068

4069 4070 4071 4072 4073 4074
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.
     */

4075
    if (virDomainNetGetActualVlan(iface)) {
4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100
        /* 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;
        }
    }

4101 4102
    if (dev) {
        /* we are now assured of success, so mark the allocation */
4103
        dev->connections++;
4104 4105 4106 4107 4108 4109 4110 4111 4112
        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);
        }
4113
    }
4114

4115 4116 4117 4118 4119
    if (netdef) {
        netdef->connections++;
        VIR_DEBUG("Using network %s, %d connections",
                  netdef->name, netdef->connections);
    }
4120
    ret = 0;
4121

4122 4123 4124
cleanup:
    if (network)
        virNetworkObjUnlock(network);
4125 4126 4127 4128
    return ret;

error:
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4129 4130 4131
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4132
    goto cleanup;
4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148
}

/* 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;
4149
    enum virDomainNetType actualType = virDomainNetGetActualType(iface);
4150 4151
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4152 4153
    virNetworkForwardIfDefPtr dev = NULL;
    int ii, ret = -1;
4154 4155 4156 4157 4158 4159 4160 4161

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

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
4162 4163 4164
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4165 4166 4167 4168 4169
        goto error;
    }
    netdef = network->def;

    if (!iface->data.network.actual ||
4170 4171
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4172 4173
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4174 4175
    }

4176
    if (netdef->forward.npfs > 0 && netdef->forward.nifs == 0 &&
4177
        networkCreateInterfacePool(netdef) < 0) {
4178
        goto error;
4179
    }
4180
    if (netdef->forward.nifs == 0) {
4181
        virReportError(VIR_ERR_INTERNAL_ERROR,
4182 4183
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4184
                       netdef->name);
4185
        goto error;
4186
    }
4187

4188 4189
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4190

4191 4192 4193 4194 4195 4196 4197 4198 4199
        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 */
4200 4201
        for (ii = 0; ii < netdef->forward.nifs; ii++) {
            if (netdef->forward.ifs[ii].type
4202
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4203 4204
                STREQ(actualDev, netdef->forward.ifs[ii].device.dev)) {
                dev = &netdef->forward.ifs[ii];
4205 4206 4207 4208 4209
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4210
            virReportError(VIR_ERR_INTERNAL_ERROR,
4211 4212
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4213
                           netdef->name, actualDev);
4214
            goto error;
4215 4216
        }

4217
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4218 4219
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4220
         */
4221
        if ((dev->connections > 0) &&
4222 4223
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4224 4225
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4226
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4227
            virReportError(VIR_ERR_INTERNAL_ERROR,
4228 4229
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4230
                           netdef->name, actualDev);
4231
            goto error;
4232
        }
4233

4234
        /* we are now assured of success, so mark the allocation */
4235
        dev->connections++;
4236
        VIR_DEBUG("Using physical device %s, connections %d",
4237
                  dev->device.dev, dev->connections);
4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250

    }  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 */
4251 4252
        for (ii = 0; ii < netdef->forward.nifs; ii++) {
            if (netdef->forward.ifs[ii].type
4253
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4254
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4255 4256
                                         &netdef->forward.ifs[ii].device.pci)) {
                dev = &netdef->forward.ifs[ii];
4257 4258 4259 4260 4261 4262 4263 4264 4265
                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,
4266 4267 4268 4269
                           hostdev->source.subsys.u.pci.addr.domain,
                           hostdev->source.subsys.u.pci.addr.bus,
                           hostdev->source.subsys.u.pci.addr.slot,
                           hostdev->source.subsys.u.pci.addr.function);
4270 4271 4272 4273 4274 4275 4276 4277
                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) &&
4278
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294
            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);
4295 4296
    }

4297
success:
4298 4299 4300
    netdef->connections++;
    VIR_DEBUG("Using network %s, %d connections",
              netdef->name, netdef->connections);
4301 4302 4303 4304 4305
    ret = 0;
cleanup:
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4306 4307 4308

error:
    goto cleanup;
4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325
}


/* 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;
4326
    enum virDomainNetType actualType = virDomainNetGetActualType(iface);
4327
    virNetworkObjPtr network;
4328
    virNetworkDefPtr netdef;
4329 4330
    virNetworkForwardIfDefPtr dev = NULL;
    int ii, ret = -1;
4331 4332 4333 4334 4335 4336 4337 4338

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

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
4339 4340 4341
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4342 4343 4344 4345
        goto error;
    }
    netdef = network->def;

4346 4347 4348 4349 4350 4351
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) &&
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4352 4353 4354
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4355 4356
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4357 4358
    }

4359
    if (netdef->forward.nifs == 0) {
4360
        virReportError(VIR_ERR_INTERNAL_ERROR,
4361
                       _("network '%s' uses a direct/hostdev mode, but "
4362 4363
                         "has no forward dev and no interface pool"),
                       netdef->name);
4364
        goto error;
4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376
    }

    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;
        }
4377

4378 4379
        for (ii = 0; ii < netdef->forward.nifs; ii++) {
            if (netdef->forward.ifs[ii].type
4380
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4381 4382
                STREQ(actualDev, netdef->forward.ifs[ii].device.dev)) {
                dev = &netdef->forward.ifs[ii];
4383 4384 4385
                break;
            }
        }
4386

4387
        if (!dev) {
4388
            virReportError(VIR_ERR_INTERNAL_ERROR,
4389 4390
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4391
                           netdef->name, actualDev);
4392
            goto error;
4393 4394
        }

4395
        dev->connections--;
4396
        VIR_DEBUG("Releasing physical device %s, connections %d",
4397
                  dev->device.dev, dev->connections);
4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408

    } 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;
        }

4409 4410
        for (ii = 0; ii < netdef->forward.nifs; ii++) {
            if (netdef->forward.ifs[ii].type
4411
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4412 4413
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
                                         &netdef->forward.ifs[ii].device.pci)) {
4414
                dev = &netdef->forward.ifs[ii];
4415 4416 4417 4418 4419 4420 4421 4422 4423
                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,
4424 4425 4426 4427
                           hostdev->source.subsys.u.pci.addr.domain,
                           hostdev->source.subsys.u.pci.addr.bus,
                           hostdev->source.subsys.u.pci.addr.slot,
                           hostdev->source.subsys.u.pci.addr.function);
4428 4429 4430 4431 4432 4433 4434 4435 4436
                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);
   }
4437

4438
success:
4439 4440 4441
    netdef->connections--;
    VIR_DEBUG("Releasing network %s, %d connections",
              netdef->name, netdef->connections);
4442 4443 4444 4445
    ret = 0;
cleanup:
    if (network)
        virNetworkObjUnlock(network);
4446 4447 4448 4449
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4450
    return ret;
4451 4452 4453

error:
    goto cleanup;
4454
}
4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479

/*
 * 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;
4480
    virNetworkObjPtr network;
4481 4482 4483 4484
    virNetworkDefPtr netdef;
    virNetworkIpDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4485
    char *dev_name = NULL;
4486 4487 4488 4489 4490 4491

    *netaddr = NULL;
    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, netname);
    networkDriverUnlock(driver);
    if (!network) {
4492 4493 4494
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4495
        goto error;
4496 4497 4498
    }
    netdef = network->def;

4499
    switch (netdef->forward.type) {
4500 4501 4502 4503 4504 4505
    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) {
4506 4507 4508
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have an IPv4 address"),
                           netdef->name);
4509 4510 4511 4512 4513 4514
            break;
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4515
        if ((dev_name = netdef->bridge))
4516 4517 4518 4519 4520 4521 4522 4523
            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:
4524 4525
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4526

4527
        if (!dev_name) {
4528 4529 4530
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4531 4532 4533 4534
        }
        break;
    }

4535
    if (dev_name) {
4536
        if (virNetDevGetIPv4Address(dev_name, &addr) < 0)
4537
            goto error;
4538
        addrptr = &addr;
4539 4540
    }

4541 4542 4543
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
        goto error;
4544 4545
    }

4546
    ret = 0;
4547 4548 4549 4550
cleanup:
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4551 4552 4553

error:
    goto cleanup;
4554
}
4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572

/**
 * networkCheckBandwidth:
 * @net: network QoS
 * @iface: interface QoS
 * @new_rate: new rate for non guaranteed class
 *
 * Returns: -1 if plugging would overcommit network QoS
 *           0 if plugging is safe (@new_rate updated)
 *           1 if no QoS is set (@new_rate untouched)
 */
static int
networkCheckBandwidth(virNetworkObjPtr net,
                      virDomainNetDefPtr iface,
                      unsigned long long *new_rate)
{
    int ret = -1;
    virNetDevBandwidthPtr netBand = net->def->bandwidth;
4573
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4574 4575 4576 4577
    unsigned long long tmp_floor_sum = net->floor_sum;
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588
    virMacAddrFormat(&iface->mac, ifmac);

    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
        !(netBand && netBand->in)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("Invalid use of 'floor' on interface with MAC "
                         "address %s - network '%s' has no inbound QoS set"),
                       ifmac, net->def->name);
        return -1;
    }

4589
    if (!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor ||
4590 4591
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4592
        return 1;
4593
    }
4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662

    tmp_new_rate = netBand->in->average;
    tmp_floor_sum += ifaceBand->in->floor;

    /* check against peak */
    if (netBand->in->peak) {
        tmp_new_rate = netBand->in->peak;
        if (tmp_floor_sum > netBand->in->peak) {
            virReportError(VIR_ERR_OPERATION_INVALID,
                           _("Cannot plug '%s' interface into '%s' because it "
                             "would overcommit 'peak' on network '%s'"),
                           ifmac,
                           net->def->bridge,
                           net->def->name);
            goto cleanup;
        }
    } else if (tmp_floor_sum > netBand->in->average) {
        /* tmp_floor_sum can be between 'average' and 'peak' iff 'peak' is set.
         * Otherwise, tmp_floor_sum must be below 'average'. */
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("Cannot plug '%s' interface into '%s' because it "
                         "would overcommit 'average' on network '%s'"),
                       ifmac,
                       net->def->bridge,
                       net->def->name);
        goto cleanup;
    }

    *new_rate = tmp_new_rate;
    ret = 0;

cleanup:
    return ret;
}

/**
 * networkNextClassID:
 * @net: network object
 *
 * Find next free class ID. @net is supposed
 * to be locked already. If there is a free ID,
 * it is marked as used and returned.
 *
 * Returns next free class ID or -1 if none is available.
 */
static ssize_t
networkNextClassID(virNetworkObjPtr net)
{
    size_t ret = 0;
    bool is_set = false;

    while (virBitmapGetBit(net->class_id, ret, &is_set) == 0 && is_set)
        ret++;

    if (is_set || virBitmapSetBit(net->class_id, ret) < 0)
        return -1;

    return ret;
}

static int
networkPlugBandwidth(virNetworkObjPtr net,
                     virDomainNetDefPtr iface)
{
    int ret = -1;
    int plug_ret;
    unsigned long long new_rate = 0;
    ssize_t class_id = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];
4663
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691

    if ((plug_ret = networkCheckBandwidth(net, iface, &new_rate)) < 0) {
        /* helper reported error */
        goto cleanup;
    }

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

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

    /* generate new class_id */
    if ((class_id = networkNextClassID(net)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Could not generate next class ID"));
        goto cleanup;
    }

4692 4693
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
4694 4695 4696 4697 4698 4699 4700 4701
    if (plug_ret < 0) {
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }

    /* QoS was set, generate new class ID */
    iface->data.network.actual->class_id = class_id;
    /* update sum of 'floor'-s of attached NICs */
4702
    net->floor_sum += ifaceBand->in->floor;
4703 4704 4705
    /* update status file */
    if (virNetworkSaveStatus(NETWORK_STATE_DIR, net) < 0) {
        ignore_value(virBitmapClearBit(net->class_id, class_id));
4706
        net->floor_sum -= ifaceBand->in->floor;
4707 4708 4709 4710
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729
    /* update rate for non guaranteed NICs */
    new_rate -= net->floor_sum;
    if (virNetDevBandwidthUpdateRate(net->def->bridge, "1:2",
                                     net->def->bandwidth, new_rate) < 0)
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                 net->def->bridge);

    ret = 0;

cleanup:
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
    int ret = 0;
    unsigned long long new_rate;
4730
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
        /* we must remove class from bridge */
        new_rate = net->def->bandwidth->in->average;

        if (net->def->bandwidth->in->peak > 0)
            new_rate = net->def->bandwidth->in->peak;

        ret = virNetDevBandwidthUnplug(net->def->bridge,
                                       iface->data.network.actual->class_id);
        if (ret < 0)
            goto cleanup;
        /* update sum of 'floor'-s of attached NICs */
4745
        net->floor_sum -= ifaceBand->in->floor;
4746 4747 4748 4749 4750
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
        if (virNetworkSaveStatus(NETWORK_STATE_DIR, net) < 0) {
4751
            net->floor_sum += ifaceBand->in->floor;
4752 4753 4754 4755
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768
        /* update rate for non guaranteed NICs */
        new_rate -= net->floor_sum;
        if (virNetDevBandwidthUpdateRate(net->def->bridge, "1:2",
                                         net->def->bandwidth, new_rate) < 0)
            VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                     net->def->bridge);
        /* no class is associated any longer */
        iface->data.network.actual->class_id = 0;
    }

cleanup:
    return ret;
}