bridge_driver.c 158.2 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2014 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
 *
 * 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>
43
#include <net/if.h>
44
#include <dirent.h>
45 46 47
#if HAVE_SYS_SYSCTL_H
# include <sys/sysctl.h>
#endif
48

49
#include "virerror.h"
50
#include "datatypes.h"
51
#include "bridge_driver.h"
52
#include "bridge_driver_platform.h"
53
#include "network_conf.h"
54
#include "device_conf.h"
55
#include "driver.h"
56
#include "virbuffer.h"
57
#include "virpidfile.h"
58
#include "vircommand.h"
59
#include "viralloc.h"
60
#include "viruuid.h"
61
#include "viriptables.h"
62
#include "virlog.h"
63
#include "virdnsmasq.h"
64
#include "configmake.h"
65
#include "virnetdev.h"
66
#include "virpci.h"
67 68
#include "virnetdevbridge.h"
#include "virnetdevtap.h"
69
#include "virnetdevvportprofile.h"
70
#include "virdbus.h"
71
#include "virfile.h"
72
#include "virstring.h"
73
#include "viraccessapicheck.h"
74
#include "network_event.h"
75
#include "virhook.h"
76
#include "virjson.h"
77

78 79
#define VIR_FROM_THIS VIR_FROM_NETWORK

80 81 82 83 84 85 86
/**
 * VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX:
 *
 * Macro providing the upper limit on the size of leases file
 */
#define VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX (32 * 1024 * 1024)

87 88
VIR_LOG_INIT("network.bridge_driver");

89
static void networkDriverLock(virNetworkDriverStatePtr driver)
90
{
91
    virMutexLock(&driver->lock);
92
}
93
static void networkDriverUnlock(virNetworkDriverStatePtr driver)
94
{
95
    virMutexUnlock(&driver->lock);
96 97
}

98
static int networkStateCleanup(void);
99

100
static int networkStartNetwork(virNetworkDriverStatePtr driver,
101 102
                               virNetworkObjPtr network);

103
static int networkShutdownNetwork(virNetworkDriverStatePtr driver,
104 105
                                  virNetworkObjPtr network);

106
static int networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
107
                                     virNetworkObjPtr network);
108

109
static int networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
R
Roman Bogorodskiy 已提交
110
                                         virNetworkObjPtr network);
111

112
static int networkStartNetworkExternal(virNetworkDriverStatePtr driver,
113 114
                                     virNetworkObjPtr network);

115
static int networkShutdownNetworkExternal(virNetworkDriverStatePtr driver,
116
                                        virNetworkObjPtr network);
117

118
static void networkReloadFirewallRules(virNetworkDriverStatePtr driver);
119
static void networkRefreshDaemons(virNetworkDriverStatePtr driver);
120

121 122 123 124 125
static int networkPlugBandwidth(virNetworkObjPtr net,
                                virDomainNetDefPtr iface);
static int networkUnplugBandwidth(virNetworkObjPtr net,
                                  virDomainNetDefPtr iface);

126
static void networkNetworkObjTaint(virNetworkObjPtr net,
127
                                   virNetworkTaintFlags taint);
128

129
static virNetworkDriverStatePtr driverState = NULL;
130

131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151
static virNetworkObjPtr
networkObjFromNetwork(virNetworkPtr net)
{
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
    virNetworkObjPtr network;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

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

    if (!network) {
        virUUIDFormat(net->uuid, uuidstr);
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching uuid '%s' (%s)"),
                       uuidstr, net->name);
    }

    return network;
}

152 153 154
static int
networkRunHook(virNetworkObjPtr network,
               virDomainDefPtr dom,
155
               virDomainNetDefPtr iface,
156 157 158 159 160 161 162 163 164
               int op,
               int sub_op)
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *xml = NULL, *net_xml = NULL, *dom_xml = NULL;
    int hookret;
    int ret = -1;

    if (virHookPresent(VIR_HOOK_DRIVER_NETWORK)) {
165 166 167 168 169 170
        if (!network) {
            VIR_DEBUG("Not running hook as @network is NULL");
            ret = 0;
            goto cleanup;
        }

171 172
        virBufferAddLit(&buf, "<hookData>\n");
        virBufferAdjustIndent(&buf, 2);
173 174
        if (iface && virDomainNetDefFormat(&buf, iface, 0) < 0)
            goto cleanup;
175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194
        if (virNetworkDefFormatBuf(&buf, network->def, 0) < 0)
            goto cleanup;
        if (dom && virDomainDefFormatInternal(dom, 0, &buf) < 0)
            goto cleanup;

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

        if (virBufferError(&buf) ||
            !(xml = virBufferContentAndReset(&buf)))
            goto cleanup;

        hookret = virHookCall(VIR_HOOK_DRIVER_NETWORK, network->def->name,
                              op, sub_op, NULL, xml, NULL);

        /*
         * If the script raised an error, pass it to the callee.
         */
        if (hookret < 0)
            goto cleanup;
195 196

        networkNetworkObjTaint(network, VIR_NETWORK_TAINT_HOOK);
197 198 199
    }

    ret = 0;
200
 cleanup:
201 202 203 204 205 206 207
    virBufferFreeAndReset(&buf);
    VIR_FREE(xml);
    VIR_FREE(net_xml);
    VIR_FREE(dom_xml);
    return ret;
}

208
static char *
209
networkDnsmasqLeaseFileNameDefault(const char *netname)
210 211 212
{
    char *leasefile;

213 214
    ignore_value(virAsprintf(&leasefile, "%s/%s.leases",
                             driverState->dnsmasqStateDir, netname));
215 216 217
    return leasefile;
}

218 219 220
networkDnsmasqLeaseFileNameFunc networkDnsmasqLeaseFileName =
    networkDnsmasqLeaseFileNameDefault;

221 222 223 224 225 226 227 228 229 230
static char *
networkDnsmasqLeaseFileNameCustom(const char *bridge)
{
    char *leasefile;

    ignore_value(virAsprintf(&leasefile, "%s/%s.status",
                             driverState->dnsmasqStateDir, bridge));
    return leasefile;
}

231 232 233 234 235
static char *
networkDnsmasqConfigFileName(const char *netname)
{
    char *conffile;

236 237
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
                             driverState->dnsmasqStateDir, netname));
238 239 240
    return conffile;
}

241 242 243 244 245 246
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

247
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
248 249 250 251 252 253 254 255
    return pidfilebase;
}

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

256 257
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
                             driverState->radvdStateDir, netname));
258 259
    return configfile;
}
260

261 262
/* do needed cleanup steps and remove the network from the list */
static int
263
networkRemoveInactive(virNetworkDriverStatePtr driver,
264 265 266
                      virNetworkObjPtr net)
{
    char *leasefile = NULL;
267
    char *customleasefile = NULL;
268
    char *radvdconfigfile = NULL;
269
    char *configfile = NULL;
270
    char *radvdpidbase = NULL;
271
    char *statusfile = NULL;
272 273 274 275 276 277
    dnsmasqContext *dctx = NULL;
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(net);

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
278 279
    if (!(dctx = dnsmasqContextNew(def->name,
                                   driverState->dnsmasqStateDir))) {
280
        goto cleanup;
281
    }
282 283 284 285

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

286 287 288
    if (!(customleasefile = networkDnsmasqLeaseFileNameCustom(def->bridge)))
        goto cleanup;

289
    if (!(radvdconfigfile = networkRadvdConfigFileName(def->name)))
290
        goto cleanup;
291 292

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
293
        goto cleanup;
294

295
    if (!(configfile = networkDnsmasqConfigFileName(def->name)))
296
        goto cleanup;
297

298 299
    if (!(statusfile
          = virNetworkConfigFile(driverState->stateDir, def->name)))
300
        goto cleanup;
301

302 303 304
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
305
    unlink(customleasefile);
306
    unlink(configfile);
307 308 309

    /* radvd */
    unlink(radvdconfigfile);
310
    virPidFileDelete(driverState->pidDir, radvdpidbase);
311

312 313 314
    /* remove status file */
    unlink(statusfile);

315 316 317 318 319
    /* remove the network definition */
    virNetworkRemoveInactive(&driver->networks, net);

    ret = 0;

320
 cleanup:
321
    VIR_FREE(leasefile);
322
    VIR_FREE(configfile);
323
    VIR_FREE(customleasefile);
324 325
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
326
    VIR_FREE(statusfile);
327 328 329 330
    dnsmasqContextFree(dctx);
    return ret;
}

331 332 333
static char *
networkBridgeDummyNicName(const char *brname)
{
334
    static const char dummyNicSuffix[] = "-nic";
335 336
    char *nicname;

337 338 339 340 341 342 343
    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.
         */
344 345 346 347 348
        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));
349
    } else {
350
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
351
    }
352 353 354
    return nicname;
}

355 356 357
/* Update the internal status of all allegedly active networks
 * according to external conditions on the host (i.e. anything that
 * isn't stored directly in each network's state file). */
358
static void
359
networkUpdateAllState(virNetworkDriverStatePtr driver)
360
{
361
    size_t i;
362

363
    for (i = 0; i < driver->networks.count; i++) {
364 365
        virNetworkObjPtr obj = driver->networks.objs[i];

366 367 368
        if (!obj->active)
           continue;

369 370
        virNetworkObjLock(obj);

371 372 373 374 375 376 377 378
        switch (obj->def->forward.type) {
        case VIR_NETWORK_FORWARD_NONE:
        case VIR_NETWORK_FORWARD_NAT:
        case VIR_NETWORK_FORWARD_ROUTE:
            /* If bridge doesn't exist, then mark it inactive */
            if (!(obj->def->bridge && virNetDevExists(obj->def->bridge) == 1))
                obj->active = 0;
            break;
379

380 381 382 383 384 385 386 387 388 389 390 391 392 393 394
        case VIR_NETWORK_FORWARD_BRIDGE:
            if (obj->def->bridge) {
                if (virNetDevExists(obj->def->bridge) != 1)
                    obj->active = 0;
                break;
            }
            /* intentionally drop through to common case for all
             * macvtap networks (forward='bridge' with no bridge
             * device defined is macvtap using its 'bridge' mode)
             */
        case VIR_NETWORK_FORWARD_PRIVATE:
        case VIR_NETWORK_FORWARD_VEPA:
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
            /* so far no extra checks */
            break;
395

396 397 398 399
        case VIR_NETWORK_FORWARD_HOSTDEV:
            /* so far no extra checks */
            break;
        }
400

401 402 403 404 405 406 407 408 409 410 411 412 413 414 415
        /* Try and read dnsmasq/radvd pids of active networks */
        if (obj->active && obj->def->ips && (obj->def->nips > 0)) {
            char *radvdpidbase;

            ignore_value(virPidFileReadIfAlive(driverState->pidDir,
                                               obj->def->name,
                                               &obj->dnsmasqPid,
                                               dnsmasqCapsGetBinaryPath(driver->dnsmasqCaps)));
            radvdpidbase = networkRadvdPidfileBasename(obj->def->name);
            if (!radvdpidbase)
                break;
            ignore_value(virPidFileReadIfAlive(driverState->pidDir,
                                               radvdpidbase,
                                               &obj->radvdPid, RADVD));
            VIR_FREE(radvdpidbase);
416 417 418 419
        }

        virNetworkObjUnlock(obj);
    }
420 421 422 423 424 425 426 427 428 429 430 431 432 433 434

    /* 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++;
    }
435 436 437
}


438
static void
439 440
networkAutostartConfigs(virNetworkDriverStatePtr driver)
{
441
    size_t i;
442

443
    for (i = 0; i < driver->networks.count; i++) {
444
        virNetworkObjLock(driver->networks.objs[i]);
445
        if (driver->networks.objs[i]->autostart &&
446 447
            !virNetworkObjIsActive(driver->networks.objs[i])) {
            if (networkStartNetwork(driver, driver->networks.objs[i]) < 0) {
448
            /* failed to start but already logged */
449
            }
450
        }
451
        virNetworkObjUnlock(driver->networks.objs[i]);
452 453 454
    }
}

455 456 457
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
458 459
                             DBusMessage *message, void *user_data)
{
460
    virNetworkDriverStatePtr _driverState = user_data;
461 462 463 464 465 466 467

    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.");
468
        networkReloadFirewallRules(_driverState);
469 470 471 472 473 474
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

475 476 477 478 479 480 481 482 483 484 485 486 487 488 489
static int
networkMigrateStateFiles(virNetworkDriverStatePtr driver)
{
    /* Due to a change in location of network state xml beginning in
     * libvirt 1.2.4 (from /var/lib/libvirt/network to
     * /var/run/libvirt/network), we must check for state files in two
     * locations. Anything found in the old location must be written
     * to the new location, then erased from the old location. (Note
     * that we read/write the file rather than calling rename()
     * because the old and new state directories are likely in
     * different filesystems).
     */
    int ret = -1;
    const char *oldStateDir = LOCALSTATEDIR "/lib/libvirt/network";
    DIR *dir;
490
    int direrr;
491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509
    struct dirent *entry;
    char *oldPath = NULL, *newPath = NULL;
    char *contents = NULL;

    if (!(dir = opendir(oldStateDir))) {
        if (errno == ENOENT)
            return 0;

        virReportSystemError(errno, _("failed to open directory '%s'"),
                             oldStateDir);
        return -1;
    }

    if (virFileMakePath(driver->stateDir) < 0) {
        virReportSystemError(errno, _("cannot create directory %s"),
                             driver->stateDir);
        goto cleanup;
    }

510
    while ((direrr = virDirRead(dir, &entry, oldStateDir)) > 0) {
511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537

        if (entry->d_type != DT_REG ||
            STREQ(entry->d_name, ".") ||
            STREQ(entry->d_name, ".."))
            continue;

        if (virAsprintf(&oldPath, "%s/%s",
                        oldStateDir, entry->d_name) < 0)
           goto cleanup;
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
           goto cleanup;

        if (virAsprintf(&newPath, "%s/%s",
                        driver->stateDir, entry->d_name) < 0)
           goto cleanup;
        if (virFileWriteStr(newPath, contents, S_IRUSR | S_IWUSR) < 0) {
            virReportSystemError(errno,
                                 _("failed to write network status file '%s'"),
                                 newPath);
            goto cleanup;
        }

        unlink(oldPath);
        VIR_FREE(oldPath);
        VIR_FREE(newPath);
        VIR_FREE(contents);
    }
538 539
    if (direrr < 0)
       goto cleanup;
540 541 542 543 544 545 546 547 548 549

    ret = 0;
 cleanup:
    closedir(dir);
    VIR_FREE(oldPath);
    VIR_FREE(newPath);
    VIR_FREE(contents);
    return ret;
}

550
/**
551
 * networkStateInitialize:
552 553 554 555
 *
 * Initialization function for the QEmu daemon
 */
static int
556 557 558
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
559
{
560 561 562
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
563 564 565
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
566 567

    if (VIR_ALLOC(driverState) < 0)
568
        goto error;
569

570 571 572 573
    if (virMutexInit(&driverState->lock) < 0) {
        VIR_FREE(driverState);
        goto error;
    }
574 575
    networkDriverLock(driverState);

576 577 578 579
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
580
    if (privileged) {
581 582 583 584 585
        if (VIR_STRDUP(driverState->networkConfigDir,
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
            VIR_STRDUP(driverState->networkAutostartDir,
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
            VIR_STRDUP(driverState->stateDir,
586
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
587 588 589 590 591 592 593
            VIR_STRDUP(driverState->pidDir,
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
            VIR_STRDUP(driverState->dnsmasqStateDir,
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
            VIR_STRDUP(driverState->radvdStateDir,
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
594 595 596 597 598 599 600

        /* migration from old to new location is only applicable for
         * privileged mode - unprivileged mode directories haven't
         * changed location.
         */
        if (networkMigrateStateFiles(driverState) < 0)
            goto error;
601
    } else {
602 603 604
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
605
            goto error;
606

607 608 609 610 611 612 613 614 615 616 617 618
        if ((virAsprintf(&driverState->networkConfigDir,
                         "%s/qemu/networks", configdir) < 0) ||
            (virAsprintf(&driverState->networkAutostartDir,
                         "%s/qemu/networks/autostart", configdir) < 0) ||
            (virAsprintf(&driverState->stateDir,
                         "%s/network/lib", rundir) < 0) ||
            (virAsprintf(&driverState->pidDir,
                         "%s/network/run", rundir) < 0) ||
            (virAsprintf(&driverState->dnsmasqStateDir,
                         "%s/dnsmasq/lib", rundir) < 0) ||
            (virAsprintf(&driverState->radvdStateDir,
                         "%s/radvd/lib", rundir) < 0)) {
619
            goto error;
620
        }
621 622
    }

623 624 625 626 627 628 629
    if (virFileMakePath(driverState->stateDir) < 0) {
        virReportSystemError(errno,
                             _("cannot create directory %s"),
                             driverState->stateDir);
        goto error;
    }

630 631
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
    driverState->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
632

633
    if (virNetworkLoadAllState(&driverState->networks,
634
                               driverState->stateDir) < 0)
635 636
        goto error;

637
    if (virNetworkLoadAllConfigs(&driverState->networks,
638
                                 driverState->networkConfigDir,
639 640 641
                                 driverState->networkAutostartDir) < 0)
        goto error;

642
    networkUpdateAllState(driverState);
643
    networkReloadFirewallRules(driverState);
644
    networkRefreshDaemons(driverState);
645

646 647
    driverState->networkEventState = virObjectEventStateNew();

648 649
    networkDriverUnlock(driverState);

650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675
#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

676
    ret = 0;
677
 cleanup:
678 679 680
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
681

682
 error:
683 684
    if (driverState)
        networkDriverUnlock(driverState);
685
    networkStateCleanup();
686
    goto cleanup;
687 688
}

689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
    if (!driverState)
        return;

    networkDriverLock(driverState);
    networkAutostartConfigs(driverState);
    networkDriverUnlock(driverState);
}

705
/**
706
 * networkStateReload:
707 708 709 710 711
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
712 713
networkStateReload(void)
{
714 715 716
    if (!driverState)
        return 0;

717
    networkDriverLock(driverState);
718
    virNetworkLoadAllState(&driverState->networks,
719
                           driverState->stateDir);
720
    virNetworkLoadAllConfigs(&driverState->networks,
721 722
                             driverState->networkConfigDir,
                             driverState->networkAutostartDir);
723
    networkReloadFirewallRules(driverState);
724
    networkRefreshDaemons(driverState);
725
    networkAutostartConfigs(driverState);
726
    networkDriverUnlock(driverState);
727 728 729 730 731
    return 0;
}


/**
732
 * networkStateCleanup:
733 734 735 736
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
737 738
networkStateCleanup(void)
{
739 740 741
    if (!driverState)
        return -1;

742 743
    networkDriverLock(driverState);

744 745
    virObjectEventStateFree(driverState->networkEventState);

746
    /* free inactive networks */
747
    virNetworkObjListFree(&driverState->networks);
748 749 750

    VIR_FREE(driverState->networkConfigDir);
    VIR_FREE(driverState->networkAutostartDir);
751 752 753 754
    VIR_FREE(driverState->stateDir);
    VIR_FREE(driverState->pidDir);
    VIR_FREE(driverState->dnsmasqStateDir);
    VIR_FREE(driverState->radvdStateDir);
755

756 757
    virObjectUnref(driverState->dnsmasqCaps);

758
    networkDriverUnlock(driverState);
759
    virMutexDestroy(&driverState->lock);
760

761 762 763 764 765 766
    VIR_FREE(driverState);

    return 0;
}


767 768 769 770 771 772 773
/* 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)
{
774 775
    size_t i;
    int ret = -1;
776 777 778 779 780 781 782
    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.
     */
783
    for (i = 0; i < 25; i++) {
784
        int signum = 0;
785
        if (i == 0)
786
            signum = SIGTERM;
787
        else if (i == 15) {
788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822
            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);
823
 cleanup:
824 825 826
    return ret;
}

G
Gene Czarcinski 已提交
827 828 829 830
    /* the following does not build a file, it builds a list
     * which is later saved into a file
     */

831
static int
G
Gene Czarcinski 已提交
832 833
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
                                 virNetworkIpDefPtr ipdef)
834
{
835
    size_t i;
G
Gene Czarcinski 已提交
836
    bool ipv6 = false;
837

G
Gene Czarcinski 已提交
838 839
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
840 841
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
842
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
843 844
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
845
                return -1;
846
    }
847

G
Gene Czarcinski 已提交
848 849 850 851 852 853 854
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
855
    size_t i, j;
G
Gene Czarcinski 已提交
856

857 858
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
859
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
860
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
861
                for (j = 0; j < host->nnames; j++)
862 863
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
864 865
            }
        }
866 867
    }

868
    return 0;
869 870 871
}


872 873
int
networkDnsmasqConfContents(virNetworkObjPtr network,
874 875 876 877
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
878
{
879
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
880
    int r, ret = -1;
881
    int nbleases = 0;
882
    size_t i;
883
    virNetworkDNSDefPtr dns = &network->def->dns;
G
Gene Czarcinski 已提交
884 885
    virNetworkIpDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
    bool ipv6SLAAC;
886

887 888
    *configstr = NULL;

889
    /*
890 891 892
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
893
     *
894 895
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
896
     */
897 898 899 900 901 902

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

903 904
    /* create dnsmasq config file appropriate for this network */
    virBufferAsprintf(&configbuf,
905 906 907 908 909 910 911
                      "##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"
912
                      "strict-order\n",
913 914
                      network->def->name);

915 916 917 918 919 920 921 922
    if (network->def->dns.forwarders) {
        virBufferAddLit(&configbuf, "no-resolv\n");
        for (i = 0; i < network->def->dns.nfwds; i++) {
            virBufferAsprintf(&configbuf, "server=%s\n",
                               network->def->dns.forwarders[i]);
        }
    }

923
    if (network->def->domain) {
924
        virBufferAsprintf(&configbuf,
925 926 927 928
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
929

930 931
    if (network->def->dns.forwardPlainNames
        == VIR_NETWORK_DNS_FORWARD_PLAIN_NAMES_NO) {
932 933 934 935
        virBufferAddLit(&configbuf, "domain-needed\n");
        /* need to specify local=// whether or not a domain is
         * specified, unless the config says we should forward "plain"
         * names (i.e. not fully qualified, no '.' characters)
936
         */
937
        virBufferAddLit(&configbuf, "local=//\n");
938
    }
939

940
    if (pidfile)
941
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
942

943 944 945
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

946 947 948 949 950 951 952 953
    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.
         */
954
        virBufferAsprintf(&configbuf,
955 956 957
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
958
    } else {
959
        virBufferAddLit(&configbuf, "bind-interfaces\n");
960 961 962 963 964 965 966 967
        /*
         * --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
         */
968 969 970
        for (i = 0;
             (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
             i++) {
971 972 973 974
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
975

976
            /* also part of CVE 2012-3411 - if the host's version of
977
             * dnsmasq doesn't have bind-dynamic, only allow listening on
978 979
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
980 981
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
982 983 984 985
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
986
                                 "The version of dnsmasq on this host (%d.%d) "
987 988 989 990
                                 "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 "
991 992 993 994 995 996
                                 "(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);
997 998
                goto cleanup;
            }
999
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1000 1001 1002
            VIR_FREE(ipaddr);
        }
    }
1003

1004 1005
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1006
     * guaranteed to not work, and set no-resolv so that no dns
1007 1008 1009
     * 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).
1010
     */
1011
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1012
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1013
                        "no-resolv\n");
1014
    }
1015

1016
    for (i = 0; i < dns->ntxts; i++) {
1017
        virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
1018 1019
                          dns->txts[i].name,
                          dns->txts[i].value);
1020
    }
1021

1022
    for (i = 0; i < dns->nsrvs; i++) {
1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044
        /* service/protocol are required, and should have been validated
         * by the parser.
         */
        if (!dns->srvs[i].service) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Missing required 'service' "
                             "attribute in SRV record of network '%s'"),
                           network->def->name);
            goto cleanup;
        }
        if (!dns->srvs[i].protocol) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Missing required 'service' "
                             "attribute in SRV record of network '%s'"),
                           network->def->name);
            goto cleanup;
        }
        /* RFC2782 requires that service and protocol be preceded by
         * an underscore.
         */
        virBufferAsprintf(&configbuf, "srv-host=_%s._%s",
                          dns->srvs[i].service, dns->srvs[i].protocol);
1045

1046 1047 1048
        /* domain is optional - it defaults to the domain of this network */
        if (dns->srvs[i].domain)
            virBufferAsprintf(&configbuf, ".%s", dns->srvs[i].domain);
1049

1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071
        /* If target is empty or ".", that means "the service is
         * decidedly not available at this domain" (RFC2782). In that
         * case, any port, priority, or weight is irrelevant.
         */
        if (dns->srvs[i].target && STRNEQ(dns->srvs[i].target, ".")) {

            virBufferAsprintf(&configbuf, ",%s", dns->srvs[i].target);
            /* port, priority, and weight are optional, but are
             * identified by their position in the line. If an item is
             * unspecified, but something later in the line *is*
             * specified, we need to give the default value for the
             * unspecified item. (According to the dnsmasq manpage,
             * the default for port is 1).
             */
            if (dns->srvs[i].port ||
                dns->srvs[i].priority || dns->srvs[i].weight)
                virBufferAsprintf(&configbuf, ",%d",
                                  dns->srvs[i].port ? dns->srvs[i].port : 1);
            if (dns->srvs[i].priority || dns->srvs[i].weight)
                virBufferAsprintf(&configbuf, ",%d", dns->srvs[i].priority);
            if (dns->srvs[i].weight)
                virBufferAsprintf(&configbuf, ",%d", dns->srvs[i].weight);
1072
        }
1073
        virBufferAddLit(&configbuf, "\n");
1074 1075
    }

G
Gene Czarcinski 已提交
1076
    /* Find the first dhcp for both IPv4 and IPv6 */
1077 1078 1079
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
1080 1081 1082 1083
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1084 1085
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096
                    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,
1097 1098 1099 1100 1101 1102 1103 1104 1105
                                   _("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
1106 1107 1108 1109
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1110 1111
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
                    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 "
1125 1126 1127 1128
                 "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
1129 1130 1131 1132 1133
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1134
        for (r = 0; r < ipdef->nranges; r++) {
1135
            char *saddr = virSocketAddrFormat(&ipdef->ranges[r].start);
1136 1137
            if (!saddr)
                goto cleanup;
1138
            char *eaddr = virSocketAddrFormat(&ipdef->ranges[r].end);
1139 1140 1141 1142
            if (!eaddr) {
                VIR_FREE(saddr);
                goto cleanup;
            }
1143
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s\n",
1144
                              saddr, eaddr);
1145
            VIR_FREE(saddr);
1146
            VIR_FREE(eaddr);
1147 1148
            nbleases += virSocketAddrGetRange(&ipdef->ranges[r].start,
                                              &ipdef->ranges[r].end);
1149
        }
1150

1151
        /*
1152 1153 1154 1155
         * 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)
1156 1157
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1158
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1159 1160
            if (!bridgeaddr)
                goto cleanup;
1161
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static\n", bridgeaddr);
1162 1163
            VIR_FREE(bridgeaddr);
        }
1164

G
Gene Czarcinski 已提交
1165 1166
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1167

G
Gene Czarcinski 已提交
1168 1169 1170
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts)
1171
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1172

G
Gene Czarcinski 已提交
1173
            if (ipdef->tftproot) {
1174 1175
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1176
            }
1177

G
Gene Czarcinski 已提交
1178 1179 1180
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1181

1182
                    if (!bootserver)
G
Gene Czarcinski 已提交
1183
                        goto cleanup;
1184
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1185
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1186 1187
                    VIR_FREE(bootserver);
                } else {
1188
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1189 1190 1191 1192 1193
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1194

G
Gene Czarcinski 已提交
1195 1196
    if (nbleases > 0) {
        char *leasefile = networkDnsmasqLeaseFileName(network->def->name);
1197
        if (!leasefile)
G
Gene Czarcinski 已提交
1198
            goto cleanup;
1199
        virBufferAsprintf(&configbuf, "dhcp-leasefile=%s\n", leasefile);
G
Gene Czarcinski 已提交
1200
        VIR_FREE(leasefile);
1201
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
G
Gene Czarcinski 已提交
1202
    }
1203

G
Gene Czarcinski 已提交
1204 1205
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1206
        goto cleanup;
G
Gene Czarcinski 已提交
1207 1208 1209 1210 1211 1212

    /* 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)
1213 1214
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1215

1216 1217
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1218
     */
1219
    virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
1220
                      dctx->addnhostsfile->path);
G
Gene Czarcinski 已提交
1221 1222 1223 1224

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
        if (ipv6def)
1225
            virBufferAddLit(&configbuf, "enable-ra\n");
G
Gene Czarcinski 已提交
1226
        else {
1227 1228 1229
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
                 i++) {
G
Gene Czarcinski 已提交
1230 1231 1232 1233
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1234 1235
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1236 1237
                    VIR_FREE(bridgeaddr);
                }
1238
            }
1239
        }
1240 1241
    }

1242 1243 1244
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1245
    ret = 0;
G
Gene Czarcinski 已提交
1246

1247
 cleanup:
1248
    virBufferFreeAndReset(&configbuf);
1249
    return ret;
1250 1251
}

1252
/* build the dnsmasq command line */
1253 1254 1255
static int ATTRIBUTE_NONNULL(2)
networkBuildDhcpDaemonCommandLine(virNetworkObjPtr network,
                                  virCommandPtr *cmdout,
1256 1257
                                  char *pidfile, dnsmasqContext *dctx,
                                  dnsmasqCapsPtr caps)
1258
{
1259
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1260
    int ret = -1;
1261 1262
    char *configfile = NULL;
    char *configstr = NULL;
1263
    char *leaseshelper_path = NULL;
1264 1265

    network->dnsmasqPid = -1;
1266

1267 1268 1269 1270 1271 1272
    if (networkDnsmasqConfContents(network, pidfile, &configstr, dctx, caps) < 0)
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1273
    if (!(configfile = networkDnsmasqConfigFileName(network->def->name)))
1274 1275 1276 1277 1278 1279 1280
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
                         _("couldn't write dnsmasq config file '%s'"),
                         configfile);
1281 1282 1283
        goto cleanup;
    }

1284 1285 1286 1287 1288 1289
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
                                                  "src",
                                                  LIBEXECDIR)))
        goto cleanup;

1290 1291
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(caps));
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1292 1293
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);

1294
    *cmdout = cmd;
1295
    ret = 0;
1296
 cleanup:
1297 1298
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1299
    VIR_FREE(leaseshelper_path);
1300 1301 1302 1303
    return ret;
}

static int
1304
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
1305
                       virNetworkObjPtr network)
1306 1307 1308 1309
{
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1310
    dnsmasqContext *dctx = NULL;
1311

1312
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0)) {
G
Gene Czarcinski 已提交
1313
        /* no IP addresses, so we don't need to run */
1314 1315 1316 1317
        ret = 0;
        goto cleanup;
    }

1318
    if (virFileMakePath(driverState->pidDir) < 0) {
1319
        virReportSystemError(errno,
1320
                             _("cannot create directory %s"),
1321
                             driverState->pidDir);
1322
        goto cleanup;
1323 1324
    }

1325
    if (!(pidfile = virPidFileBuildPath(driverState->pidDir,
1326
                                        network->def->name)))
1327
        goto cleanup;
1328

1329
    if (virFileMakePath(driverState->dnsmasqStateDir) < 0) {
1330
        virReportSystemError(errno,
1331
                             _("cannot create directory %s"),
1332
                             driverState->dnsmasqStateDir);
1333 1334 1335
        goto cleanup;
    }

1336
    dctx = dnsmasqContextNew(network->def->name, driverState->dnsmasqStateDir);
1337 1338 1339
    if (dctx == NULL)
        goto cleanup;

1340 1341
    if (dnsmasqCapsRefresh(&driver->dnsmasqCaps, NULL) < 0)
        goto cleanup;
1342 1343 1344

    ret = networkBuildDhcpDaemonCommandLine(network, &cmd, pidfile,
                                            dctx, driver->dnsmasqCaps);
1345 1346 1347 1348 1349
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1350
        goto cleanup;
1351

G
Guido Günther 已提交
1352 1353
    ret = virCommandRun(cmd, NULL);
    if (ret < 0) {
1354
        goto cleanup;
G
Guido Günther 已提交
1355
    }
1356 1357

    /*
1358 1359 1360 1361 1362
     * 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
1363 1364
     */

1365
    ret = virPidFileRead(driverState->pidDir, network->def->name,
1366 1367
                         &network->dnsmasqPid);
    if (ret < 0)
1368
        goto cleanup;
1369

1370
    ret = 0;
1371
 cleanup:
1372
    VIR_FREE(pidfile);
1373
    virCommandFree(cmd);
1374
    dnsmasqContextFree(dctx);
1375 1376 1377
    return ret;
}

1378 1379
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1380 1381
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1382 1383 1384
 *
 *  Returns 0 on success, -1 on failure.
 */
1385
static int
1386
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
1387
                         virNetworkObjPtr network)
1388
{
1389 1390
    int ret = -1;
    size_t i;
G
Gene Czarcinski 已提交
1391
    virNetworkIpDefPtr ipdef, ipv4def, ipv6def;
1392
    dnsmasqContext *dctx = NULL;
1393

G
Gene Czarcinski 已提交
1394
    /* if no IP addresses specified, nothing to do */
1395
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1396 1397
        return 0;

1398 1399
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1400
        return networkStartDhcpDaemon(driver, network);
1401

G
Gene Czarcinski 已提交
1402
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1403 1404
    if (!(dctx = dnsmasqContextNew(network->def->name,
                                   driverState->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1405
        goto cleanup;
1406
    }
G
Gene Czarcinski 已提交
1407 1408 1409 1410 1411 1412

    /* 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;
1413 1414 1415
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
G
Gene Czarcinski 已提交
1416 1417
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1418 1419
    }

G
Gene Czarcinski 已提交
1420
    ipv6def = NULL;
1421 1422 1423
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1424 1425
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1426 1427
    }

G
Gene Czarcinski 已提交
1428 1429 1430 1431 1432
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
           goto cleanup;

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
           goto cleanup;
1433

G
Gene Czarcinski 已提交
1434
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
1435 1436 1437
       goto cleanup;

    if ((ret = dnsmasqSave(dctx)) < 0)
1438
        goto cleanup;
1439 1440

    ret = kill(network->dnsmasqPid, SIGHUP);
1441
 cleanup:
1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453
    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
1454
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
1455
                         virNetworkObjPtr network)
1456 1457 1458 1459 1460 1461
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1462
    }
1463
    /* now start dnsmasq if it should be started */
1464
    return networkStartDhcpDaemon(driver, network);
1465 1466
}

G
Gene Czarcinski 已提交
1467 1468 1469 1470 1471 1472
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";

1473 1474 1475
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1476
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1477 1478
    int ret = -1;
    size_t i;
1479
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
1480
    bool v6present = false, dhcp6 = false;
1481 1482

    *configstr = NULL;
1483

G
Gene Czarcinski 已提交
1484
    /* Check if DHCPv6 is needed */
1485 1486 1487
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500
        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;
    }

1501 1502 1503
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1504
    virBufferAsprintf(&configbuf, "interface %s\n"
1505 1506
                      "{\n"
                      "  AdvSendAdvert on;\n"
1507
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1508 1509 1510 1511 1512
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1513 1514

    /* add a section for each IPv6 address in the config */
1515 1516 1517
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
1518 1519 1520 1521 1522
        int prefix;
        char *netaddr;

        prefix = virNetworkIpDefPrefix(ipdef);
        if (prefix < 0) {
1523 1524 1525
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1526 1527
            goto cleanup;
        }
1528
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1529
            goto cleanup;
1530
        virBufferAsprintf(&configbuf,
1531
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1532 1533 1534
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1535 1536 1537
        VIR_FREE(netaddr);
    }

1538 1539 1540
    /* only create the string if we found at least one IPv6 address */
    if (v6present) {
        virBufferAddLit(&configbuf, "};\n");
1541

1542 1543 1544 1545 1546 1547 1548 1549
        if (virBufferError(&configbuf)) {
            virReportOOMError();
            goto cleanup;
        }
        if (!(*configstr = virBufferContentAndReset(&configbuf))) {
            virReportOOMError();
            goto cleanup;
        }
1550
    }
1551 1552

    ret = 0;
1553
 cleanup:
1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575
    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;
1576 1577 1578 1579
        goto cleanup;
    }

    /* construct the filename */
1580
    if (!(*configFile = networkRadvdConfigFileName(network->def->name)))
1581 1582
        goto cleanup;
    /* write the file */
1583
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1584 1585
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1586 1587 1588 1589 1590
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1591
 cleanup:
1592 1593 1594 1595 1596 1597
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1598
networkStartRadvd(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
G
Gene Czarcinski 已提交
1599
                        virNetworkObjPtr network)
1600 1601 1602 1603 1604 1605 1606 1607 1608
{
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1609
    /* Is dnsmasq handling RA? */
1610
   if (DNSMASQ_RA_SUPPORT(driver->dnsmasqCaps)) {
G
Gene Czarcinski 已提交
1611 1612 1613 1614
        ret = 0;
        goto cleanup;
    }

1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625
    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);
1626 1627 1628
        goto cleanup;
    }

1629
    if (virFileMakePath(driverState->pidDir) < 0) {
1630 1631
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1632
                             driverState->pidDir);
1633 1634
        goto cleanup;
    }
1635
    if (virFileMakePath(driverState->radvdStateDir) < 0) {
1636 1637
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1638
                             driverState->radvdStateDir);
1639 1640 1641 1642
        goto cleanup;
    }

    /* construct pidfile name */
1643
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1644
        goto cleanup;
1645
    if (!(pidfile = virPidFileBuildPath(driverState->pidDir, radvdpidbase)))
1646 1647 1648 1649 1650
        goto cleanup;

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

1651 1652 1653 1654
    /* 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
1655
     * virPidFileRead() below will fail if we use them).
1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670
     * 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;

1671
    if (virPidFileRead(driverState->pidDir, radvdpidbase, &network->radvdPid) < 0)
1672 1673 1674
        goto cleanup;

    ret = 0;
1675
 cleanup:
1676 1677 1678 1679 1680 1681 1682
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1683
static int
1684
networkRefreshRadvd(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
1685
                    virNetworkObjPtr network)
1686
{
G
Gene Czarcinski 已提交
1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697
    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)) {
1698
            virPidFileDelete(driverState->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1699 1700 1701 1702 1703 1704
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }

1705 1706
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
G
Gene Czarcinski 已提交
1707
        return networkStartRadvd(driver, network);
1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719

    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);
}

1720 1721
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1722
static int
1723
networkRestartRadvd(virNetworkDriverStatePtr driver,
1724
                    virNetworkObjPtr network)
1725 1726 1727 1728 1729 1730 1731 1732 1733
{
    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
1734
        if ((networkKillDaemon(network->radvdPid, "radvd",
1735 1736 1737
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1738
            virPidFileDelete(driverState->pidDir, radvdpidbase);
1739 1740 1741 1742 1743 1744 1745 1746 1747
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1748 1749 1750 1751
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
1752
networkRefreshDaemons(virNetworkDriverStatePtr driver)
1753
{
1754
    size_t i;
1755 1756 1757

    VIR_INFO("Refreshing network daemons");

1758
    for (i = 0; i < driver->networks.count; i++) {
1759 1760 1761 1762
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
1763 1764 1765
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
1766 1767 1768 1769 1770 1771
            /* 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.
             */
1772 1773
            networkRefreshDhcpDaemon(driver, network);
            networkRefreshRadvd(driver, network);
1774 1775 1776 1777 1778
        }
        virNetworkObjUnlock(network);
    }
}

1779
static void
1780
networkReloadFirewallRules(virNetworkDriverStatePtr driver)
1781
{
1782
    size_t i;
1783

1784
    VIR_INFO("Reloading iptables rules");
1785

1786
    for (i = 0; i < driver->networks.count; i++) {
1787 1788 1789 1790
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
1791 1792 1793
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
1794 1795 1796
            /* Only the three L3 network types that are configured by libvirt
             * need to have iptables rules reloaded.
             */
1797 1798
            networkRemoveFirewallRules(network->def);
            if (networkAddFirewallRules(network->def) < 0) {
1799 1800
                /* failed to add but already logged */
            }
1801
        }
1802
        virNetworkObjUnlock(network);
1803 1804 1805
    }
}

1806
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
1807
static int
1808
networkEnableIpForwarding(bool enableIPv4, bool enableIPv6)
1809
{
1810
    int ret = 0;
1811 1812 1813 1814 1815 1816 1817 1818 1819
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
                            &enabled, sizeof(enabled));
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
                            &enabled, sizeof(enabled));
#else
1820 1821 1822 1823
    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);
1824
#endif
1825
    return ret;
1826 1827
}

1828 1829
#define SYSCTL_PATH "/proc/sys"

1830 1831
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
1832 1833 1834 1835
{
    char *field = NULL;
    int ret = -1;

1836 1837 1838 1839 1840
    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",
1841
                        network->def->bridge) < 0)
1842
            goto cleanup;
1843

1844 1845 1846 1847 1848 1849
        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;
        }
1850

1851 1852 1853 1854 1855 1856 1857
        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);
1858 1859
    }

1860 1861 1862 1863 1864 1865 1866 1867
    /* 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",
1868
                    network->def->bridge) < 0)
1869 1870
        goto cleanup;

1871
    if (virFileWriteStr(field, "0", 0) < 0) {
1872
        virReportSystemError(errno,
1873 1874 1875 1876 1877
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

1878 1879 1880 1881
    /* 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",
1882
                    network->def->bridge) < 0)
1883 1884
        goto cleanup;

1885
    if (virFileWriteStr(field, "0", 0) < 0) {
1886
        virReportSystemError(errno,
1887
                             _("cannot disable %s"), field);
1888 1889 1890 1891
        goto cleanup;
    }

    ret = 0;
1892
 cleanup:
1893 1894 1895 1896
    VIR_FREE(field);
    return ret;
}

1897
/* add an IP address to a bridge */
1898
static int
D
Daniel P. Berrange 已提交
1899
networkAddAddrToBridge(virNetworkObjPtr network,
1900
                       virNetworkIpDefPtr ipdef)
1901
{
1902 1903 1904
    int prefix = virNetworkIpDefPrefix(ipdef);

    if (prefix < 0) {
1905 1906 1907
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
1908 1909 1910
        return -1;
    }

1911 1912
    if (virNetDevSetIPv4Address(network->def->bridge,
                                &ipdef->address, prefix) < 0)
1913 1914 1915 1916 1917
        return -1;

    return 0;
}

1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930
/* add an IP (static) route to a bridge */
static int
networkAddRouteToBridge(virNetworkObjPtr network,
                        virNetworkRouteDefPtr routedef)
{
    int prefix = 0;
    unsigned int metric;
    virSocketAddrPtr addr = &routedef->address;
    virSocketAddrPtr mask = &routedef->netmask;
    virSocketAddr zero;

    /* this creates an all-0 address of the appropriate family */
    ignore_value(virSocketAddrParse(&zero,
1931
                                    (VIR_SOCKET_ADDR_IS_FAMILY(addr, AF_INET)
1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966
                                     ? "0.0.0.0" : "::"),
                                    VIR_SOCKET_ADDR_FAMILY(addr)));

    if (virSocketAddrEqual(addr, &zero)) {
        if (routedef->has_prefix && routedef->prefix == 0)
            prefix = 0;
        else if ((VIR_SOCKET_ADDR_IS_FAMILY(mask, AF_INET) &&
                virSocketAddrEqual(mask, &zero)))
            prefix = 0;
        else
            prefix = virSocketAddrGetIpPrefix(addr, mask, routedef->prefix);
    } else {
        prefix = virSocketAddrGetIpPrefix(addr, mask, routedef->prefix);
    }

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

    if (routedef->has_metric && routedef->metric > 0)
        metric = routedef->metric;
    else
        metric = 1;

    if (virNetDevAddRoute(network->def->bridge, &routedef->address,
                          prefix, &routedef->gateway, metric) < 0) {
        return -1;
    }
    return 0;
}

1967
static int
1968
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
1969 1970
                          virNetworkObjPtr network)
{
1971
    size_t i;
1972
    bool v4present = false, v6present = false;
1973 1974
    virErrorPtr save_err = NULL;
    virNetworkIpDefPtr ipdef;
1975
    virNetworkRouteDefPtr routedef;
1976
    char *macTapIfName = NULL;
1977
    int tapfd = -1;
1978

1979
    /* Check to see if any network IP collides with an existing route */
1980
    if (networkCheckRouteCollision(network->def) < 0)
1981 1982
        return -1;

1983
    /* Create and configure the bridge device */
1984
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
1985 1986
        return -1;

1987 1988 1989 1990 1991 1992 1993 1994
    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);
1995
        if (!macTapIfName)
1996
            goto err0;
1997
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
1998
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
1999
                                           &macTapIfName, &network->def->mac,
2000
                                           NULL, &tapfd, 1, NULL, NULL,
2001 2002 2003
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2004 2005 2006 2007 2008
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2009
    /* Set bridge options */
2010 2011 2012 2013

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2014
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2015
                                   network->def->delay * 1000) < 0)
2016
        goto err1;
2017

2018
    if (virNetDevBridgeSetSTP(network->def->bridge,
2019
                              network->def->stp ? true : false) < 0)
2020
        goto err1;
2021

2022 2023 2024 2025
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2026
        goto err1;
2027

2028
    /* Add "once per network" rules */
2029
    if (networkAddFirewallRules(network->def) < 0)
2030 2031
        goto err1;

2032 2033 2034
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
2035
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2036
            v4present = true;
2037
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2038
            v6present = true;
2039

2040
        /* Add the IP address/netmask to the bridge */
D
Daniel P. Berrange 已提交
2041
        if (networkAddAddrToBridge(network, ipdef) < 0) {
2042
            goto err2;
2043
        }
2044 2045
    }

2046
    /* Bring up the bridge interface */
2047
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2048
        goto err2;
2049

2050 2051
    for (i = 0; i < network->def->nroutes; i++) {
        routedef = &network->def->routes[i];
2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
        if (VIR_SOCKET_ADDR_VALID(&routedef->gateway)) {
            if (networkAddRouteToBridge(network, routedef) < 0) {
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2063 2064
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2065
        networkEnableIpForwarding(v4present, v6present) < 0) {
2066
        virReportSystemError(errno, "%s",
2067
                             _("failed to enable IP forwarding"));
2068
        goto err3;
2069 2070
    }

2071

2072
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2073 2074
    if ((v4present || v6present) &&
        networkStartDhcpDaemon(driver, network) < 0)
2075
        goto err3;
2076

2077
    /* start radvd if there are any ipv6 addresses */
G
Gene Czarcinski 已提交
2078
    if (v6present && networkStartRadvd(driver, network) < 0)
2079 2080
        goto err4;

2081 2082 2083 2084 2085 2086 2087 2088 2089
    /* 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);
    }

2090 2091
    if (virNetDevBandwidthSet(network->def->bridge,
                              network->def->bandwidth, true) < 0) {
2092 2093 2094
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("cannot set bandwidth limits on %s"),
                       network->def->bridge);
2095 2096 2097
        goto err5;
    }

2098
    VIR_FREE(macTapIfName);
2099 2100 2101

    return 0;

2102
 err5:
2103
    virNetDevBandwidthClear(network->def->bridge);
2104

2105 2106 2107 2108
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2109 2110 2111 2112 2113
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2114 2115 2116
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2117
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2118

2119 2120 2121
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2122
    networkRemoveFirewallRules(network->def);
2123 2124

 err1:
2125 2126 2127
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2128
    if (macTapIfName) {
2129
        VIR_FORCE_CLOSE(tapfd);
H
Hu Tao 已提交
2130 2131 2132
        ignore_value(virNetDevTapDelete(macTapIfName));
        VIR_FREE(macTapIfName);
    }
2133 2134

 err0:
2135 2136
    if (!save_err)
        save_err = virSaveLastError();
2137
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2138

2139 2140 2141 2142
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2143
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2144 2145 2146
    return -1;
}

2147
static int networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
R
Roman Bogorodskiy 已提交
2148
                                         virNetworkObjPtr network)
2149
{
2150
    virNetDevBandwidthClear(network->def->bridge);
2151

2152 2153 2154 2155 2156
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2157
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2158
            virPidFileDelete(driverState->pidDir, radvdpidbase);
2159 2160 2161 2162
            VIR_FREE(radvdpidbase);
        }
    }

2163 2164 2165
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2166
    if (network->def->mac_specified) {
2167
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2168
        if (macTapIfName) {
2169
            ignore_value(virNetDevTapDelete(macTapIfName));
2170 2171 2172 2173
            VIR_FREE(macTapIfName);
        }
    }

2174
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2175

2176
    networkRemoveFirewallRules(network->def);
2177

2178
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2179

2180
    /* See if its still alive and really really kill it */
2181
    if (network->dnsmasqPid > 0 &&
2182
        (kill(network->dnsmasqPid, 0) == 0))
2183 2184
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2185 2186 2187 2188 2189 2190

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

2191 2192 2193 2194
    return 0;
}

static int
2195
networkStartNetworkExternal(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
2196 2197 2198
                            virNetworkObjPtr network ATTRIBUTE_UNUSED)
{
    /* put anything here that needs to be done each time a network of
2199
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2200 2201 2202 2203 2204 2205
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

2206
static int networkShutdownNetworkExternal(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
2207 2208 2209
                                        virNetworkObjPtr network ATTRIBUTE_UNUSED)
{
    /* put anything here that needs to be done each time a network of
2210
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2211 2212 2213 2214 2215 2216 2217
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2218
networkStartNetwork(virNetworkDriverStatePtr driver,
2219 2220
                    virNetworkObjPtr network)
{
2221 2222 2223
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2224 2225

    if (virNetworkObjIsActive(network)) {
2226 2227
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2228
        return ret;
2229 2230
    }

2231 2232 2233
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2234
    if (virNetworkObjSetDefTransient(network, true) < 0)
2235
        goto cleanup;
2236

2237 2238
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2239
    if (networkRunHook(network, NULL, NULL,
2240 2241 2242 2243
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2244
    switch (network->def->forward.type) {
2245 2246 2247 2248

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2249 2250
        if (networkStartNetworkVirtual(driver, network) < 0)
            goto cleanup;
2251 2252 2253 2254 2255 2256
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2257
    case VIR_NETWORK_FORWARD_HOSTDEV:
2258 2259
        if (networkStartNetworkExternal(driver, network) < 0)
            goto cleanup;
2260 2261 2262
        break;
    }

2263
    /* finally we can call the 'started' hook script if any */
2264
    if (networkRunHook(network, NULL, NULL,
2265 2266 2267 2268
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2269 2270 2271
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2272 2273 2274
    VIR_DEBUG("Writing network status to disk");
    if (virNetworkSaveStatus(driverState->stateDir, network) < 0)
        goto cleanup;
2275 2276

    network->active = 1;
2277 2278
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2279

2280
 cleanup:
2281
    if (ret < 0) {
2282
        virNetworkObjUnsetDefTransient(network);
2283 2284 2285 2286 2287 2288 2289 2290 2291 2292
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
        networkShutdownNetwork(driver, network);
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2293
static int networkShutdownNetwork(virNetworkDriverStatePtr driver,
2294 2295 2296 2297 2298 2299 2300 2301 2302 2303
                                        virNetworkObjPtr network)
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

2304 2305
    stateFile = virNetworkConfigFile(driverState->stateDir,
                                     network->def->name);
2306 2307 2308 2309 2310 2311
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2312
    switch (network->def->forward.type) {
2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323

    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:
2324
    case VIR_NETWORK_FORWARD_HOSTDEV:
2325 2326 2327 2328
        ret = networkShutdownNetworkExternal(driver, network);
        break;
    }

2329
    /* now that we know it's stopped call the hook if present */
2330
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2331 2332
                   VIR_HOOK_SUBOP_END);

2333
    network->active = 0;
2334
    virNetworkObjUnsetDefTransient(network);
2335
    return ret;
2336 2337 2338
}


2339
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2340 2341
                                         const unsigned char *uuid)
{
2342
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2343 2344
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2345

2346
    networkDriverLock(driver);
2347
    network = virNetworkFindByUUID(&driver->networks, uuid);
2348
    networkDriverUnlock(driver);
2349
    if (!network) {
2350 2351
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
2352
        goto cleanup;
2353 2354
    }

2355 2356 2357
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2358 2359
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2360
 cleanup:
2361 2362
    if (network)
        virNetworkObjUnlock(network);
2363
    return ret;
2364 2365
}

2366
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2367 2368
                                         const char *name)
{
2369
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2370 2371 2372
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2373
    networkDriverLock(driver);
2374
    network = virNetworkFindByName(&driver->networks, name);
2375
    networkDriverUnlock(driver);
2376
    if (!network) {
2377 2378
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2379
        goto cleanup;
2380 2381
    }

2382 2383 2384
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2385 2386
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2387
 cleanup:
2388 2389
    if (network)
        virNetworkObjUnlock(network);
2390
    return ret;
2391 2392
}

2393 2394 2395
static virDrvOpenStatus networkOpen(virConnectPtr conn,
                                    virConnectAuthPtr auth ATTRIBUTE_UNUSED,
                                    unsigned int flags)
2396 2397 2398
{
    virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);

2399 2400 2401 2402 2403 2404 2405
    if (!driverState)
        return VIR_DRV_OPEN_DECLINED;

    conn->networkPrivateData = driverState;
    return VIR_DRV_OPEN_SUCCESS;
}

2406 2407
static int networkClose(virConnectPtr conn)
{
2408 2409 2410 2411
    conn->networkPrivateData = NULL;
    return 0;
}

2412 2413
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2414 2415
    int nactive = 0;
    size_t i;
2416
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2417

2418 2419 2420
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2421
    networkDriverLock(driver);
2422
    for (i = 0; i < driver->networks.count; i++) {
2423 2424 2425 2426
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectNumOfNetworksCheckACL(conn, obj->def) &&
            virNetworkObjIsActive(obj))
2427
            nactive++;
2428
        virNetworkObjUnlock(obj);
2429 2430
    }
    networkDriverUnlock(driver);
2431

2432 2433 2434
    return nactive;
}

2435
static int networkConnectListNetworks(virConnectPtr conn, char **const names, int nnames) {
2436
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2437 2438
    int got = 0;
    size_t i;
2439

2440 2441 2442
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2443
    networkDriverLock(driver);
2444
    for (i = 0; i < driver->networks.count && got < nnames; i++) {
2445 2446 2447 2448 2449 2450
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectListNetworksCheckACL(conn, obj->def) &&
            virNetworkObjIsActive(obj)) {
            if (VIR_STRDUP(names[got], obj->def->name) < 0) {
                virNetworkObjUnlock(obj);
2451 2452 2453 2454
                goto cleanup;
            }
            got++;
        }
2455
        virNetworkObjUnlock(obj);
2456
    }
2457 2458
    networkDriverUnlock(driver);

2459 2460 2461
    return got;

 cleanup:
2462
    networkDriverUnlock(driver);
2463
    for (i = 0; i < got; i++)
2464 2465 2466 2467
        VIR_FREE(names[i]);
    return -1;
}

2468 2469
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2470 2471
    int ninactive = 0;
    size_t i;
2472
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2473

2474 2475 2476
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2477
    networkDriverLock(driver);
2478
    for (i = 0; i < driver->networks.count; i++) {
2479 2480 2481 2482
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectNumOfDefinedNetworksCheckACL(conn, obj->def) &&
            !virNetworkObjIsActive(obj))
2483
            ninactive++;
2484
        virNetworkObjUnlock(obj);
2485 2486
    }
    networkDriverUnlock(driver);
2487

2488 2489 2490
    return ninactive;
}

2491
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames) {
2492
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2493 2494
    int got = 0;
    size_t i;
2495

2496 2497 2498
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2499
    networkDriverLock(driver);
2500
    for (i = 0; i < driver->networks.count && got < nnames; i++) {
2501 2502 2503 2504 2505 2506
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectListDefinedNetworksCheckACL(conn, obj->def) &&
            !virNetworkObjIsActive(obj)) {
            if (VIR_STRDUP(names[got], obj->def->name) < 0) {
                virNetworkObjUnlock(obj);
2507 2508 2509 2510
                goto cleanup;
            }
            got++;
        }
2511
        virNetworkObjUnlock(obj);
2512
    }
2513
    networkDriverUnlock(driver);
2514 2515 2516
    return got;

 cleanup:
2517
    networkDriverUnlock(driver);
2518
    for (i = 0; i < got; i++)
2519 2520 2521 2522
        VIR_FREE(names[i]);
    return -1;
}

2523
static int
2524 2525 2526
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2527
{
2528
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2529 2530 2531 2532
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2533 2534 2535
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2536
    networkDriverLock(driver);
2537 2538 2539
    ret = virNetworkObjListExport(conn, driver->networks, nets,
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2540 2541
    networkDriverUnlock(driver);

2542
 cleanup:
2543 2544
    return ret;
}
2545

2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
2561
                                       net, eventID, callback,
2562 2563 2564
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2565
 cleanup:
2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578
    return ret;
}

static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
    int ret = -1;

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

2579 2580 2581 2582 2583 2584
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
2585

2586
 cleanup:
2587 2588 2589
    return ret;
}

2590 2591 2592 2593 2594
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2595 2596
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2597 2598 2599 2600

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

2601 2602
    ret = virNetworkObjIsActive(obj);

2603
 cleanup:
2604 2605 2606 2607 2608 2609 2610 2611 2612 2613
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}

static int networkIsPersistent(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2614 2615
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2616 2617 2618 2619

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

2620 2621
    ret = obj->persistent;

2622
 cleanup:
2623 2624 2625 2626 2627 2628
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}


2629
static int
2630
networkValidate(virNetworkDriverStatePtr driver,
2631 2632
                virNetworkDefPtr def,
                bool check_active)
2633
{
2634
    size_t i;
2635 2636
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2637
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
2638
    bool ipv4def = false, ipv6def = false;
2639 2640 2641 2642 2643 2644 2645 2646

    /* 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
     */
2647 2648 2649
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2650 2651 2652 2653 2654

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

        virNetworkSetBridgeMacAddr(def);
2655 2656
    } else {
        /* They are also the only types that currently support setting
2657 2658
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
2659
         */
2660 2661 2662 2663 2664 2665 2666 2667
        if (def->mac_specified) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <mac> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
                           virNetworkForwardTypeToString(def->forward.type));
            return -1;
        }
2668 2669 2670 2671 2672
        if (virNetworkDefGetIpByIndex(def, AF_UNSPEC, 0)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2673
                           virNetworkForwardTypeToString(def->forward.type));
2674 2675
            return -1;
        }
2676
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
2677 2678 2679 2680
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2681
                           virNetworkForwardTypeToString(def->forward.type));
2682 2683 2684 2685 2686 2687 2688
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2689
                           virNetworkForwardTypeToString(def->forward.type));
2690 2691
            return -1;
        }
2692 2693 2694 2695 2696 2697 2698 2699
        if (def->bandwidth) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported network-wide <bandwidth> element "
                             "in network %s with forward mode='%s'"),
                           def->name,
                           virNetworkForwardTypeToString(def->forward.type));
            return -1;
        }
2700 2701
    }

G
Gene Czarcinski 已提交
2702 2703 2704
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
2705 2706 2707
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
2708 2709 2710 2711 2712
        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 -- "
2713 2714
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731
                    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;
                }
2732 2733 2734
            }
        }
    }
2735 2736 2737 2738 2739 2740

    /* 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.
     */

2741
    vlanAllowed = ((def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
2742
                   def->virtPortProfile &&
2743 2744 2745
                   def->virtPortProfile->virtPortType
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) ||
                   def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV);
2746 2747

    vlanUsed = def->vlan.nTags > 0;
2748 2749
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
2750 2751 2752 2753 2754
            /* 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.
             */
2755
            if (def->portGroups[i].virtPortProfile) {
2756
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
2757
                    def->portGroups[i].virtPortProfile->virtPortType
2758 2759 2760 2761 2762 2763 2764
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
2765
        }
2766
        if (def->portGroups[i].isDefault) {
2767 2768 2769 2770 2771
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
2772
                               def->name, defaultPortGroup->name,
2773
                               def->portGroups[i].name);
2774
                return -1;
2775
            }
2776
            defaultPortGroup = &def->portGroups[i];
2777
        }
2778
    }
2779 2780 2781 2782 2783 2784 2785
    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.
         */
2786 2787 2788 2789 2790 2791 2792 2793 2794
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
    return 0;
}

2795 2796
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
2797
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2798
    virNetworkDefPtr def;
2799
    virNetworkObjPtr network = NULL;
2800
    virNetworkPtr ret = NULL;
2801
    virObjectEventPtr event = NULL;
2802

2803 2804
    networkDriverLock(driver);

2805
    if (!(def = virNetworkDefParseString(xml)))
2806
        goto cleanup;
2807

2808 2809 2810
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

2811
    if (networkValidate(driver, def, true) < 0)
2812 2813
       goto cleanup;

2814 2815 2816
    /* NB: even though this transient network hasn't yet been started,
     * we assign the def with live = true in anticipation that it will
     * be started momentarily.
2817
     */
2818
    if (!(network = virNetworkAssignDef(&driver->networks, def, true)))
2819 2820
        goto cleanup;
    def = NULL;
2821

2822
    if (networkStartNetwork(driver, network) < 0) {
2823 2824
        virNetworkRemoveInactive(&driver->networks,
                                 network);
2825
        network = NULL;
2826
        goto cleanup;
2827 2828
    }

2829 2830
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
2831 2832
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
2833

2834
    VIR_INFO("Creating network '%s'", network->def->name);
2835 2836
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2837
 cleanup:
2838
    virNetworkDefFree(def);
2839 2840
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
2841 2842 2843
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
2844
    return ret;
2845 2846
}

2847 2848
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
2849
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2850
    virNetworkDefPtr def = NULL;
2851
    bool freeDef = true;
2852
    virNetworkObjPtr network = NULL;
2853
    virNetworkPtr ret = NULL;
2854
    virObjectEventPtr event = NULL;
2855

2856 2857
    networkDriverLock(driver);

2858
    if (!(def = virNetworkDefParseString(xml)))
2859
        goto cleanup;
2860

2861 2862 2863
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

2864
    if (networkValidate(driver, def, false) < 0)
2865 2866
       goto cleanup;

2867 2868
    if (!(network = virNetworkAssignDef(&driver->networks, def, false)))
       goto cleanup;
2869

2870
    /* def was assigned to network object */
2871
    freeDef = false;
2872 2873

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
2874 2875 2876 2877 2878
        if (!virNetworkObjIsActive(network)) {
            virNetworkRemoveInactive(&driver->networks, network);
            network = NULL;
            goto cleanup;
        }
2879 2880 2881 2882 2883
        /* if network was active already, just undo new persistent
         * definition by making it transient.
         * XXX - this isn't necessarily the correct thing to do.
         */
        virNetworkObjAssignDef(network, NULL, false);
2884 2885 2886
        goto cleanup;
    }

2887
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
2888 2889
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
2890

2891 2892
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
2893

2894
 cleanup:
2895 2896
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
2897 2898
    if (freeDef)
       virNetworkDefFree(def);
2899 2900 2901
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
2902
    return ret;
2903 2904
}

2905
static int
2906 2907
networkUndefine(virNetworkPtr net)
{
2908
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
2909
    virNetworkObjPtr network;
2910
    int ret = -1;
2911
    bool active = false;
2912
    virObjectEventPtr event = NULL;
2913

2914 2915
    networkDriverLock(driver);

2916
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
2917
    if (!network) {
2918 2919
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
2920
        goto cleanup;
2921 2922
    }

2923 2924 2925
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

2926 2927
    if (virNetworkObjIsActive(network))
        active = true;
2928

2929
    /* remove autostart link */
2930
    if (virNetworkDeleteConfig(driver->networkConfigDir,
2931 2932
                               driver->networkAutostartDir,
                               network) < 0)
2933
        goto cleanup;
2934
    network->autostart = 0;
2935

2936 2937
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
2938 2939
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
2940

2941
    VIR_INFO("Undefining network '%s'", network->def->name);
2942 2943 2944 2945 2946
    if (!active) {
        if (networkRemoveInactive(driver, network) < 0) {
            network = NULL;
            goto cleanup;
        }
2947
        network = NULL;
2948 2949 2950 2951 2952 2953
    } else {

        /* if the network still exists, it was active, and we need to make
         * it transient (by deleting the persistent def)
         */
        virNetworkObjAssignDef(network, NULL, false);
2954 2955
    }

2956
    ret = 0;
2957

2958
 cleanup:
2959 2960
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
2961 2962 2963
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
2964
    return ret;
2965 2966
}

2967 2968 2969 2970 2971 2972 2973 2974
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
2975
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
2976
    virNetworkObjPtr network = NULL;
2977 2978
    int isActive, ret = -1;
    size_t i;
2979 2980
    virNetworkIpDefPtr ipdef;
    bool oldDhcpActive = false;
2981
    bool needFirewallRefresh = false;
2982

2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996

    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;
    }

2997 2998 2999
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3000
    /* see if we are listening for dhcp pre-modification */
3001 3002 3003
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
3004 3005 3006 3007 3008 3009
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3010 3011 3012 3013
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
    */
    isActive = virNetworkObjIsActive(network);
3014 3015
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3016 3017 3018 3019 3020 3021 3022
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039
    if (isActive && (flags & VIR_NETWORK_UPDATE_AFFECT_LIVE)) {
        /* Take care of anything that must be done before updating the
         * live NetworkDef.
         */
        if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE ||
            network->def->forward.type == VIR_NETWORK_FORWARD_NAT ||
            network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
            switch (section) {
            case VIR_NETWORK_SECTION_FORWARD:
            case VIR_NETWORK_SECTION_FORWARD_INTERFACE:
            case VIR_NETWORK_SECTION_IP:
            case VIR_NETWORK_SECTION_IP_DHCP_RANGE:
            case VIR_NETWORK_SECTION_IP_DHCP_HOST:
                /* these could affect the firewall rules, so remove the
                 * old rules (and remember to load new ones after the
                 * update).
                 */
3040
                networkRemoveFirewallRules(network->def);
3041 3042 3043 3044 3045 3046 3047 3048
                needFirewallRefresh = true;
                break;
            default:
                break;
            }
        }
    }

3049
    /* update the network config in memory/on disk */
3050 3051
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3052
            ignore_value(networkAddFirewallRules(network->def));
3053 3054 3055
        goto cleanup;
    }

3056
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079
        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.
             */
3080
            if (networkRestartDhcpDaemon(driver, network) < 0)
3081 3082
                goto cleanup;

3083 3084 3085 3086 3087 3088 3089 3090
        } 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;

3091 3092 3093
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
                 i++) {
3094 3095 3096 3097 3098 3099 3100
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3101 3102
                 networkRestartDhcpDaemon(driver, network) < 0) ||
                networkRefreshDhcpDaemon(driver, network) < 0) {
3103 3104 3105 3106
                goto cleanup;
            }

        } else if (section == VIR_NETWORK_SECTION_DNS_HOST ||
3107 3108 3109 3110 3111 3112
                   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.
             */
3113
            if (networkRefreshDhcpDaemon(driver, network) < 0)
3114 3115 3116 3117 3118 3119 3120 3121
                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.
             */
3122
            if (networkRefreshRadvd(driver, network) < 0)
3123 3124 3125 3126
                goto cleanup;
        }

        /* save current network state to disk */
3127 3128
        if ((ret = virNetworkSaveStatus(driverState->stateDir,
                                        network)) < 0) {
3129
            goto cleanup;
3130
        }
3131 3132
    }
    ret = 0;
3133
 cleanup:
3134 3135 3136 3137 3138 3139
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
    return ret;
}

3140 3141
static int networkCreate(virNetworkPtr net)
{
3142
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3143 3144
    virNetworkObjPtr network;
    int ret = -1;
3145
    virObjectEventPtr event = NULL;
3146

3147
    networkDriverLock(driver);
3148
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3149

3150
    if (!network) {
3151 3152
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3153
        goto cleanup;
3154 3155
    }

3156 3157 3158
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3159
    ret = networkStartNetwork(driver, network);
3160

3161 3162
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3163 3164
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3165

3166
 cleanup:
3167 3168
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3169 3170
    if (network)
        virNetworkObjUnlock(network);
3171
    networkDriverUnlock(driver);
3172
    return ret;
3173 3174
}

3175 3176
static int networkDestroy(virNetworkPtr net)
{
3177
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3178 3179
    virNetworkObjPtr network;
    int ret = -1;
3180
    virObjectEventPtr event = NULL;
3181

3182
    networkDriverLock(driver);
3183
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3184

3185
    if (!network) {
3186 3187
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3188
        goto cleanup;
3189 3190
    }

3191 3192 3193
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3194
    if (!virNetworkObjIsActive(network)) {
3195
        virReportError(VIR_ERR_OPERATION_INVALID,
3196 3197
                       _("network '%s' is not active"),
                       network->def->name);
3198 3199 3200
        goto cleanup;
    }

3201 3202 3203
    if ((ret = networkShutdownNetwork(driver, network)) < 0)
        goto cleanup;

3204 3205
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3206 3207
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3208

3209
    if (!network->persistent) {
3210 3211 3212 3213 3214
        if (networkRemoveInactive(driver, network) < 0) {
            network = NULL;
            ret = -1;
            goto cleanup;
        }
3215 3216
        network = NULL;
    }
3217

3218
 cleanup:
3219 3220
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3221 3222
    if (network)
        virNetworkObjUnlock(network);
3223
    networkDriverUnlock(driver);
3224 3225 3226
    return ret;
}

3227
static char *networkGetXMLDesc(virNetworkPtr net,
3228
                               unsigned int flags)
3229
{
3230
    virNetworkObjPtr network;
3231
    virNetworkDefPtr def;
3232
    char *ret = NULL;
3233

3234
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3235

3236 3237
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3238

3239 3240 3241
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3242 3243 3244 3245 3246 3247
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3248

3249
 cleanup:
3250 3251
    if (network)
        virNetworkObjUnlock(network);
3252
    return ret;
3253 3254 3255
}

static char *networkGetBridgeName(virNetworkPtr net) {
3256 3257 3258
    virNetworkObjPtr network;
    char *bridge = NULL;

3259 3260
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3261

3262 3263 3264
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3265
    if (!(network->def->bridge)) {
3266 3267 3268
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3269 3270 3271
        goto cleanup;
    }

3272
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3273

3274
 cleanup:
3275 3276
    if (network)
        virNetworkObjUnlock(network);
3277 3278 3279 3280
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
3281 3282
                             int *autostart)
{
3283 3284
    virNetworkObjPtr network;
    int ret = -1;
3285

3286 3287
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3288

3289 3290 3291
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3292
    *autostart = network->autostart;
3293
    ret = 0;
3294

3295
 cleanup:
3296 3297
    if (network)
        virNetworkObjUnlock(network);
3298
    return ret;
3299 3300 3301
}

static int networkSetAutostart(virNetworkPtr net,
3302 3303
                               int autostart)
{
3304
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3305
    virNetworkObjPtr network;
3306
    char *configFile = NULL, *autostartLink = NULL;
3307
    int ret = -1;
3308

3309
    networkDriverLock(driver);
3310
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3311

3312
    if (!network) {
3313 3314
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3315
        goto cleanup;
3316 3317
    }

3318 3319 3320
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3321
    if (!network->persistent) {
3322 3323
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3324 3325 3326
        goto cleanup;
    }

3327 3328
    autostart = (autostart != 0);

3329
    if (network->autostart != autostart) {
3330
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3331
            goto cleanup;
3332
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3333 3334
            goto cleanup;

3335
        if (autostart) {
3336
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3337
                virReportSystemError(errno,
3338 3339
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3340 3341
                goto cleanup;
            }
3342

3343
            if (symlink(configFile, autostartLink) < 0) {
3344
                virReportSystemError(errno,
3345
                                     _("Failed to create symlink '%s' to '%s'"),
3346
                                     autostartLink, configFile);
3347 3348 3349
                goto cleanup;
            }
        } else {
3350
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3351
                virReportSystemError(errno,
3352
                                     _("Failed to delete symlink '%s'"),
3353
                                     autostartLink);
3354 3355
                goto cleanup;
            }
3356 3357
        }

3358
        network->autostart = autostart;
3359
    }
3360
    ret = 0;
3361

3362
 cleanup:
3363 3364
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3365 3366
    if (network)
        virNetworkObjUnlock(network);
3367
    networkDriverUnlock(driver);
3368
    return ret;
3369 3370
}

3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492
static int
networkGetDHCPLeasesHelper(virNetworkObjPtr obj,
                           const char *mac,
                           virNetworkDHCPLeasePtr **leases)
{
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
    int size = 0;
    int custom_lease_file_len = 0;
    bool need_results = !!leases;
    long long currtime = 0;
    long long expirytime_tmp = -1;
    bool ipv6 = false;
    char *lease_entries = NULL;
    char *custom_lease_file = NULL;
    const char *ip_tmp = NULL;
    const char *mac_tmp = NULL;
    virJSONValuePtr lease_tmp = NULL;
    virJSONValuePtr leases_array = NULL;
    virNetworkIpDefPtr ipdef_tmp = NULL;
    virNetworkDHCPLeasePtr lease = NULL;
    virNetworkDHCPLeasePtr *leases_ret = NULL;

    /* Retrieve custom leases file location */
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(obj->def->bridge);

    /* Read entire contents */
    if ((custom_lease_file_len = virFileReadAll(custom_lease_file,
                                                VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX,
                                                &lease_entries)) < 0) {
        /* Even though src/network/leaseshelper.c guarantees the existence of
         * leases file (even if no leases are present), and the control reaches
         * here, instead of reporting error, return 0 leases */
        rv = 0;
        goto error;
    }

    if (custom_lease_file_len) {
        if (!(leases_array = virJSONValueFromString(lease_entries))) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("invalid json in file: %s"), custom_lease_file);
            goto error;
        }

        if ((size = virJSONValueArraySize(leases_array)) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("couldn't fetch array of leases"));
            goto error;
        }
    }

    currtime = (long long) time(NULL);

    for (i = 0; i < size; i++) {
        if (!(lease_tmp = virJSONValueArrayGet(leases_array, i))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("failed to parse json"));
            goto error;
        }

        if (!(mac_tmp = virJSONValueObjectGetString(lease_tmp, "mac-address"))) {
            /* leaseshelper program guarantees that lease will be stored only if
             * mac-address is known otherwise not */
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("found lease without mac-address"));
            goto error;
        }

        if (mac && virMacAddrCompare(mac, mac_tmp)) {
            virJSONValueFree(lease_tmp);
            continue;
        }

        if (virJSONValueObjectGetNumberLong(lease_tmp, "expiry-time", &expirytime_tmp) < 0) {
            /* A lease cannot be present without expiry-time */
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("found lease without expiry-time"));
            goto error;
        }

        /* Do not report expired lease */
        if (expirytime_tmp < currtime)
            continue;

        if (need_results) {
            if (VIR_ALLOC(lease) < 0)
                goto error;

            lease->expirytime = expirytime_tmp;

            if (!(ip_tmp = virJSONValueObjectGetString(lease_tmp, "ip-address"))) {
                /* A lease without ip-address makes no sense */
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("found lease without ip-address"));
                goto error;
            }

            /* Unlike IPv4, IPv6 uses ':' instead of '.' as separator */
            ipv6 = strchr(ip_tmp, ':') ? true : false;
            lease->type = ipv6 ? VIR_IP_ADDR_TYPE_IPV6 : VIR_IP_ADDR_TYPE_IPV4;

            /* Obtain prefix */
            for (j = 0; j < obj->def->nips; j++) {
                ipdef_tmp = &obj->def->ips[j];

                if (ipv6 && VIR_SOCKET_ADDR_IS_FAMILY(&ipdef_tmp->address,
                                                      AF_INET6)) {
                    lease->prefix = ipdef_tmp->prefix;
                    break;
                }
                if (!ipv6 && VIR_SOCKET_ADDR_IS_FAMILY(&ipdef_tmp->address,
                                                      AF_INET)) {
                    lease->prefix = virSocketAddrGetIpPrefix(&ipdef_tmp->address,
                                                             &ipdef_tmp->netmask,
                                                             ipdef_tmp->prefix);
                    break;
                }
            }

            if ((VIR_STRDUP(lease->mac, mac_tmp) < 0) ||
                (VIR_STRDUP(lease->ipaddr, ip_tmp) < 0) ||
3493
                (VIR_STRDUP(lease->iface, obj->def->bridge) < 0))
3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592
                goto error;

            /* Fields that can be NULL */
            if ((VIR_STRDUP(lease->iaid,
                            virJSONValueObjectGetString(lease_tmp, "iaid")) < 0) ||
                (VIR_STRDUP(lease->clientid,
                            virJSONValueObjectGetString(lease_tmp, "client-id")) < 0) ||
                (VIR_STRDUP(lease->hostname,
                            virJSONValueObjectGetString(lease_tmp, "hostname")) < 0))
                goto error;

            if (VIR_INSERT_ELEMENT(leases_ret, nleases, nleases, lease) < 0)
                goto error;

        } else {
            nleases++;
        }

        VIR_FREE(lease);
    }

    if (need_results && mac && !leases_ret) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("no lease with matching MAC address: %s"), mac);
        goto error;
    }

    if (leases_ret) {
        /* NULL terminated array */
        ignore_value(VIR_REALLOC_N(leases_ret, nleases + 1));
        *leases = leases_ret;
        leases_ret = NULL;
    }

    rv = nleases;

 cleanup:
    VIR_FREE(lease);
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
    return rv;

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

static int
networkGetDHCPLeases(virNetworkPtr network,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
{
    int rv = -1;
    virNetworkObjPtr obj;

    virCheckFlags(0, -1);

    if (!(obj = networkObjFromNetwork(network)))
        return rv;

    if (virNetworkGetDHCPLeasesEnsureACL(network->conn, obj->def) < 0)
        goto cleanup;

    rv = networkGetDHCPLeasesHelper(obj, NULL, leases);

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

static int
networkGetDHCPLeasesForMAC(virNetworkPtr network,
                           const char *mac,
                           virNetworkDHCPLeasePtr **leases,
                           unsigned int flags)
{
    int rv = -1;
    virNetworkObjPtr obj;

    virCheckFlags(0, -1);

    if (!(obj = networkObjFromNetwork(network)))
        return rv;

    if (virNetworkGetDHCPLeasesForMACEnsureACL(network->conn, obj->def) < 0)
        goto cleanup;

    rv = networkGetDHCPLeasesHelper(obj, mac, leases);

 cleanup:
    if (obj)
        virNetworkObjUnlock(obj);
    return rv;
}
3593 3594 3595

static virNetworkDriver networkDriver = {
    "Network",
3596 3597 3598 3599 3600 3601 3602
    .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 */
3603 3604
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
3605 3606
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3607 3608
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3609
    .networkUndefine = networkUndefine, /* 0.2.0 */
3610
    .networkUpdate = networkUpdate, /* 0.10.2 */
3611
    .networkCreate = networkCreate, /* 0.2.0 */
3612 3613 3614 3615 3616 3617 3618
    .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 */
3619 3620
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
    .networkGetDHCPLeasesForMAC = networkGetDHCPLeasesForMAC, /* 1.2.6 */
3621 3622 3623
};

static virStateDriver networkStateDriver = {
3624
    .name = "Network",
3625
    .stateInitialize  = networkStateInitialize,
3626
    .stateAutoStart  = networkStateAutoStart,
3627 3628
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3629 3630
};

3631 3632
int networkRegister(void)
{
3633 3634
    if (virRegisterNetworkDriver(&networkDriver) < 0)
        return -1;
3635 3636
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
3637 3638
    return 0;
}
3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649

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

/* 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.
 */

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
3656 3657
networkCreateInterfacePool(virNetworkDefPtr netdef)
{
3658
    size_t num_virt_fns = 0;
3659
    char **vfname = NULL;
3660
    virPCIDeviceAddressPtr *virt_fns;
3661 3662
    int ret = -1;
    size_t i;
3663

3664
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev,
3665
                                      &vfname, &virt_fns, &num_virt_fns)) < 0) {
3666 3667
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Could not get Virtual functions on %s"),
3668
                       netdef->forward.pfs->dev);
3669 3670 3671 3672 3673 3674
        goto finish;
    }

    if (num_virt_fns == 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("No Vf's present on SRIOV PF %s"),
3675
                       netdef->forward.pfs->dev);
3676 3677 3678
       goto finish;
    }

3679
    if (VIR_ALLOC_N(netdef->forward.ifs, num_virt_fns) < 0)
3680 3681
        goto finish;

3682
    netdef->forward.nifs = num_virt_fns;
3683

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

    ret = 0;
3710
 finish:
3711 3712 3713
    for (i = 0; i < num_virt_fns; i++) {
        VIR_FREE(vfname[i]);
        VIR_FREE(virt_fns[i]);
3714
    }
3715
    VIR_FREE(vfname);
3716
    VIR_FREE(virt_fns);
3717 3718 3719
    return ret;
}

3720
/* networkAllocateActualDevice:
3721
 * @dom: domain definition that @iface belongs to
3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732
 * @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
3733 3734
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
3735
{
3736
    virNetworkDriverStatePtr driver = driverState;
3737
    virDomainNetType actualType = iface->type;
3738 3739
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
3740
    virNetDevBandwidthPtr bandwidth = NULL;
3741 3742 3743
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3744
    virNetworkForwardIfDefPtr dev = NULL;
3745
    size_t i;
3746 3747 3748
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3749
        goto validate;
3750 3751 3752 3753 3754 3755 3756 3757

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

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

3765 3766 3767 3768 3769 3770 3771
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

3772 3773 3774
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

3775 3776 3777 3778 3779 3780 3781 3782 3783 3784
    /* 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.
     */
3785 3786 3787 3788 3789 3790

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

3791 3792
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
3793
        goto error;
3794

3795 3796 3797 3798 3799 3800 3801 3802
    /* 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;

3803 3804
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
3805

3806 3807 3808
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
3809 3810 3811 3812
        /* for these forward types, the actual net type really *is*
         *NETWORK; we just keep the info from the portgroup in
         * iface->data.network.actual
        */
3813
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
3814 3815 3816 3817

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

3818
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
3819
               netdef->bridge) {
3820 3821 3822 3823 3824

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

3825
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
3826 3827
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
3828
            goto error;
3829

3830 3831 3832 3833 3834 3835 3836 3837
        /* 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) {
3838
            goto error;
3839 3840 3841 3842 3843 3844 3845 3846 3847 3848
        }
        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);
3849
                goto error;
3850 3851 3852
            }
        }

3853
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3854

3855
        virDomainHostdevSubsysPCIBackendType backend;
3856

3857
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
3858
        if (netdef->forward.npfs > 0 && netdef->forward.nifs <= 0 &&
3859 3860 3861 3862 3863
            networkCreateInterfacePool(netdef) < 0) {
            goto error;
        }

        /* pick first dev with 0 connections */
3864 3865 3866
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880
                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;
3881
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
3882
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
3883
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
3884

3885 3886 3887
        switch (netdef->forward.driverName)
        {
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
3888
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
3889 3890
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
3891
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
3892 3893
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
3894
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905
            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;

3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930
        /* 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;
            }
        }

3931 3932 3933 3934
    } 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)) {
3935 3936 3937 3938 3939 3940

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
3941
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
3942
        switch (netdef->forward.type) {
3943
        case VIR_NETWORK_FORWARD_BRIDGE:
3944
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
3945 3946
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
3947
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
3948 3949
            break;
        case VIR_NETWORK_FORWARD_VEPA:
3950
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
3951 3952
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
3953
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
3954 3955 3956
            break;
        }

3957 3958 3959 3960 3961 3962 3963 3964
        /* 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) {
3965
            goto error;
3966
        }
3967
        virtport = iface->data.network.actual->virtPortProfile;
3968
        if (virtport) {
3969 3970 3971 3972 3973 3974 3975 3976
            /* 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);
3977
                goto error;
3978 3979
            }
        }
3980

3981 3982 3983
        /* 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).
         */
3984
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
3985 3986 3987 3988
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
3989
            goto error;
3990 3991 3992
        } else {
            /* pick an interface from the pool */

3993
            if (netdef->forward.npfs > 0 && netdef->forward.nifs == 0 &&
3994 3995 3996 3997
                networkCreateInterfacePool(netdef) < 0) {
                goto error;
            }

3998 3999 4000 4001 4002
            /* 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.
4003
             */
4004 4005
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4006 4007 4008
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4009

4010
                /* pick first dev with 0 connections */
4011 4012 4013
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4014 4015 4016 4017 4018
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4019
                dev = &netdef->forward.ifs[0];
4020 4021 4022
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4023 4024 4025 4026
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4027 4028 4029 4030
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4031
                goto error;
4032
            }
4033 4034
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4035
                goto error;
4036 4037 4038
        }
    }

4039
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4040
        goto error;
4041

4042
 validate:
4043 4044 4045 4046 4047
    /* 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.
     */

4048
    if (virDomainNetGetActualVlan(iface)) {
4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077
        /* 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;
        }
    }

    if (netdef) {
        netdef->connections++;
        VIR_DEBUG("Using network %s, %d connections",
                  netdef->name, netdef->connections);
4078

4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097
        if (dev) {
            /* mark the allocation */
            dev->connections++;
            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);
            }
        }

        /* finally we can call the 'plugged' hook script if any */
        if (networkRunHook(network, dom, iface,
                           VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
                           VIR_HOOK_SUBOP_BEGIN) < 0) {
            /* adjust for failure */
4098
            netdef->connections--;
4099 4100 4101 4102
            if (dev)
                dev->connections--;
            goto error;
        }
4103 4104
    }

4105
    ret = 0;
4106

4107
 cleanup:
4108 4109
    if (network)
        virNetworkObjUnlock(network);
4110 4111
    return ret;

4112
 error:
4113
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4114 4115 4116
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4117
    goto cleanup;
4118 4119 4120
}

/* networkNotifyActualDevice:
4121
 * @dom: domain definition that @iface belongs to
4122 4123 4124 4125 4126 4127 4128 4129 4130 4131
 * @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
4132 4133
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4134
{
4135
    virNetworkDriverStatePtr driver = driverState;
4136
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4137 4138
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4139
    virNetworkForwardIfDefPtr dev = NULL;
4140 4141
    size_t i;
    int ret = -1;
4142 4143 4144 4145 4146 4147 4148 4149

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

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
4150 4151 4152
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4153 4154 4155 4156 4157
        goto error;
    }
    netdef = network->def;

    if (!iface->data.network.actual ||
4158 4159
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4160 4161
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4162 4163
    }

4164
    if (netdef->forward.npfs > 0 && netdef->forward.nifs == 0 &&
4165
        networkCreateInterfacePool(netdef) < 0) {
4166
        goto error;
4167
    }
4168
    if (netdef->forward.nifs == 0) {
4169
        virReportError(VIR_ERR_INTERNAL_ERROR,
4170 4171
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4172
                       netdef->name);
4173
        goto error;
4174
    }
4175

4176 4177
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4178

4179 4180 4181 4182 4183 4184 4185 4186 4187
        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 */
4188 4189
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4190
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4191 4192
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4193 4194 4195 4196 4197
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4198
            virReportError(VIR_ERR_INTERNAL_ERROR,
4199 4200
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4201
                           netdef->name, actualDev);
4202
            goto error;
4203 4204
        }

4205
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4206 4207
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4208
         */
4209
        if ((dev->connections > 0) &&
4210 4211
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4212 4213
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4214
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4215
            virReportError(VIR_ERR_INTERNAL_ERROR,
4216 4217
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4218
                           netdef->name, actualDev);
4219
            goto error;
4220
        }
4221

4222
        /* we are now assured of success, so mark the allocation */
4223
        dev->connections++;
4224
        VIR_DEBUG("Using physical device %s, connections %d",
4225
                  dev->device.dev, dev->connections);
4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238

    }  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 */
4239 4240
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4241
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4242
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4243 4244
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4245 4246 4247 4248 4249 4250 4251 4252 4253
                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,
4254 4255 4256 4257
                           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);
4258 4259 4260 4261 4262 4263 4264 4265
                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) &&
4266
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282
            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);
4283 4284
    }

4285
 success:
4286 4287 4288
    netdef->connections++;
    VIR_DEBUG("Using network %s, %d connections",
              netdef->name, netdef->connections);
4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299

    /* finally we can call the 'plugged' hook script if any */
    if (networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
                       VIR_HOOK_SUBOP_BEGIN) < 0) {
        /* adjust for failure */
        if (dev)
            dev->connections--;
        netdef->connections--;
        goto error;
    }

4300
    ret = 0;
4301
 cleanup:
4302 4303 4304
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4305

4306
 error:
4307
    goto cleanup;
4308 4309 4310 4311
}


/* networkReleaseActualDevice:
4312
 * @dom: domain definition that @iface belongs to
4313 4314 4315 4316 4317 4318 4319 4320 4321 4322
 * @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
4323 4324
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4325
{
4326
    virNetworkDriverStatePtr driver = driverState;
4327
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4328
    virNetworkObjPtr network;
4329
    virNetworkDefPtr netdef;
4330
    virNetworkForwardIfDefPtr dev = NULL;
4331 4332
    size_t i;
    int ret = -1;
4333 4334 4335 4336 4337 4338 4339 4340

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

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
4341 4342 4343
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4344 4345 4346 4347
        goto error;
    }
    netdef = network->def;

4348 4349
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4350 4351 4352 4353 4354
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) &&
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4355 4356 4357
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4358 4359
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4360 4361
    }

4362
    if (netdef->forward.nifs == 0) {
4363
        virReportError(VIR_ERR_INTERNAL_ERROR,
4364
                       _("network '%s' uses a direct/hostdev mode, but "
4365 4366
                         "has no forward dev and no interface pool"),
                       netdef->name);
4367
        goto error;
4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379
    }

    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;
        }
4380

4381 4382
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4383
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4384 4385
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4386 4387 4388
                break;
            }
        }
4389

4390
        if (!dev) {
4391
            virReportError(VIR_ERR_INTERNAL_ERROR,
4392 4393
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4394
                           netdef->name, actualDev);
4395
            goto error;
4396 4397
        }

4398
        dev->connections--;
4399
        VIR_DEBUG("Releasing physical device %s, connections %d",
4400
                  dev->device.dev, dev->connections);
4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411

    } 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;
        }

4412 4413
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4414
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4415
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4416 4417
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4418 4419 4420 4421 4422 4423 4424 4425 4426
                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,
4427 4428 4429 4430
                           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);
4431 4432 4433 4434 4435 4436 4437 4438 4439
                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);
   }
4440

4441
 success:
4442
    if (iface->data.network.actual) {
4443
        netdef->connections--;
4444 4445
        VIR_DEBUG("Releasing network %s, %d connections",
                  netdef->name, netdef->connections);
4446

4447 4448 4449 4450
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
    }
4451
    ret = 0;
4452
 cleanup:
4453 4454
    if (network)
        virNetworkObjUnlock(network);
4455 4456 4457 4458
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4459
    return ret;
4460

4461
 error:
4462
    goto cleanup;
4463
}
4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487

/*
 * 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;
4488
    virNetworkDriverStatePtr driver = driverState;
4489
    virNetworkObjPtr network;
4490 4491 4492 4493
    virNetworkDefPtr netdef;
    virNetworkIpDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4494
    char *dev_name = NULL;
4495 4496 4497 4498 4499 4500

    *netaddr = NULL;
    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, netname);
    networkDriverUnlock(driver);
    if (!network) {
4501 4502 4503
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4504
        goto error;
4505 4506 4507
    }
    netdef = network->def;

4508
    switch (netdef->forward.type) {
4509 4510 4511 4512 4513 4514
    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) {
4515 4516 4517
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have an IPv4 address"),
                           netdef->name);
4518 4519 4520 4521 4522 4523
            break;
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4524
        if ((dev_name = netdef->bridge))
4525 4526 4527 4528 4529 4530 4531 4532
            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:
4533 4534
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4535

4536
        if (!dev_name) {
4537 4538 4539
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4540 4541 4542 4543
        }
        break;
    }

4544
    if (dev_name) {
4545
        if (virNetDevGetIPv4Address(dev_name, &addr) < 0)
4546
            goto error;
4547
        addrptr = &addr;
4548 4549
    }

4550 4551 4552
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
        goto error;
4553 4554
    }

4555
    ret = 0;
4556
 cleanup:
4557 4558 4559
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4560

4561
 error:
4562
    goto cleanup;
4563
}
4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581

/**
 * 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;
4582
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4583 4584 4585 4586
    unsigned long long tmp_floor_sum = net->floor_sum;
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597
    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;
    }

4598
    if (!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor ||
4599 4600
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4601
        return 1;
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

    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;

4634
 cleanup:
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 4663 4664 4665 4666 4667 4668 4669 4670 4671
    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];
4672
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700

    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;
    }

4701 4702
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
4703 4704 4705 4706 4707 4708 4709 4710
    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 */
4711
    net->floor_sum += ifaceBand->in->floor;
4712
    /* update status file */
4713
    if (virNetworkSaveStatus(driverState->stateDir, net) < 0) {
4714
        ignore_value(virBitmapClearBit(net->class_id, class_id));
4715
        net->floor_sum -= ifaceBand->in->floor;
4716 4717 4718 4719
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
4720 4721 4722 4723 4724 4725 4726 4727 4728
    /* 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;

4729
 cleanup:
4730 4731 4732 4733 4734 4735 4736 4737 4738
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
    int ret = 0;
    unsigned long long new_rate;
4739
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4740 4741 4742

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
4743 4744 4745 4746 4747
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758
        /* 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 */
4759
        net->floor_sum -= ifaceBand->in->floor;
4760 4761 4762 4763
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
4764
        if (virNetworkSaveStatus(driverState->stateDir, net) < 0) {
4765
            net->floor_sum += ifaceBand->in->floor;
4766 4767 4768 4769
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
4770 4771 4772 4773 4774 4775 4776 4777 4778 4779
        /* 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;
    }

4780
 cleanup:
4781 4782
    return ret;
}
4783 4784 4785

static void
networkNetworkObjTaint(virNetworkObjPtr net,
4786
                       virNetworkTaintFlags taint)
4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797
{
    if (virNetworkObjTaint(net, taint)) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(net->def->uuid, uuidstr);

        VIR_WARN("Network name='%s' uuid=%s is tainted: %s",
                 net->def->name,
                 uuidstr,
                 virNetworkTaintTypeToString(taint));
    }
}