bridge_driver.c 157.8 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,
J
Ján Tomko 已提交
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,
J
Ján Tomko 已提交
113
                                       virNetworkObjPtr network);
114

115
static int networkShutdownNetworkExternal(virNetworkDriverStatePtr driver,
J
Ján Tomko 已提交
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
        if (virNetworkDefFormatBuf(&buf, network->def, 0) < 0)
            goto cleanup;
        if (dom && virDomainDefFormatInternal(dom, 0, &buf) < 0)
            goto cleanup;

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

183
        if (virBufferCheckError(&buf) < 0)
184 185
            goto cleanup;

186
        xml = virBufferContentAndReset(&buf);
187 188 189 190 191 192 193 194
        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
        if (!obj->active)
J
Ján Tomko 已提交
367
            continue;
368

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) {
J
Ján Tomko 已提交
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

        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)
J
Ján Tomko 已提交
519
            goto cleanup;
520
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
J
Ján Tomko 已提交
521
            goto cleanup;
522 523 524

        if (virAsprintf(&newPath, "%s/%s",
                        driver->stateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
525
            goto cleanup;
526 527 528 529 530 531 532 533 534 535 536 537
        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
    if (direrr < 0)
J
Ján Tomko 已提交
539
        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;
}

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

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

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

J
Ján Tomko 已提交
930
    if (network->def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
931 932 933 934
        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)
935
         */
936
        virBufferAddLit(&configbuf, "local=//\n");
937
    }
938

939
    if (pidfile)
940
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
941

942 943 944
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

945 946 947 948 949 950 951 952
    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.
         */
953
        virBufferAsprintf(&configbuf,
954 955 956
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
957
    } else {
958
        virBufferAddLit(&configbuf, "bind-interfaces\n");
959 960 961 962 963 964 965 966
        /*
         * --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
         */
967 968 969
        for (i = 0;
             (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
             i++) {
970 971 972 973
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
974

975
            /* also part of CVE 2012-3411 - if the host's version of
976
             * dnsmasq doesn't have bind-dynamic, only allow listening on
977 978
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
979 980
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
981 982 983 984
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
985
                                 "The version of dnsmasq on this host (%d.%d) "
986 987 988 989
                                 "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 "
990 991 992 993 994 995
                                 "(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);
996
                VIR_FREE(ipaddr);
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

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1224
        if (ipv6def) {
1225
            virBufferAddLit(&configbuf, "enable-ra\n");
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
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1279 1280
                             _("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
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1429
        goto cleanup;
G
Gene Czarcinski 已提交
1430 1431

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1432
        goto cleanup;
1433

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

    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
    virBufferAddLit(&configbuf, "};\n");
1539

1540
    if (virBufferCheckError(&configbuf) < 0)
1541
        goto cleanup;
1542

1543 1544
    *configstr = virBufferContentAndReset(&configbuf);

1545
    ret = 0;
1546
 cleanup:
1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568
    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;
1569 1570 1571 1572
        goto cleanup;
    }

    /* construct the filename */
1573
    if (!(*configFile = networkRadvdConfigFileName(network->def->name)))
1574 1575
        goto cleanup;
    /* write the file */
1576
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1577 1578
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1579 1580 1581 1582 1583
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1584
 cleanup:
1585 1586 1587 1588 1589 1590
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1591
networkStartRadvd(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
J
Ján Tomko 已提交
1592
                  virNetworkObjPtr network)
1593 1594 1595 1596 1597 1598 1599 1600 1601
{
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1602
    /* Is dnsmasq handling RA? */
J
Ján Tomko 已提交
1603
    if (DNSMASQ_RA_SUPPORT(driver->dnsmasqCaps)) {
G
Gene Czarcinski 已提交
1604 1605 1606 1607
        ret = 0;
        goto cleanup;
    }

1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618
    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);
1619 1620 1621
        goto cleanup;
    }

1622
    if (virFileMakePath(driverState->pidDir) < 0) {
1623 1624
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1625
                             driverState->pidDir);
1626 1627
        goto cleanup;
    }
1628
    if (virFileMakePath(driverState->radvdStateDir) < 0) {
1629 1630
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1631
                             driverState->radvdStateDir);
1632 1633 1634 1635
        goto cleanup;
    }

    /* construct pidfile name */
1636
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1637
        goto cleanup;
1638
    if (!(pidfile = virPidFileBuildPath(driverState->pidDir, radvdpidbase)))
1639 1640 1641 1642 1643
        goto cleanup;

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

1644 1645 1646 1647
    /* 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
1648
     * virPidFileRead() below will fail if we use them).
1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663
     * 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;

1664
    if (virPidFileRead(driverState->pidDir, radvdpidbase, &network->radvdPid) < 0)
1665 1666 1667
        goto cleanup;

    ret = 0;
1668
 cleanup:
1669 1670 1671 1672 1673 1674 1675
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1676
static int
1677
networkRefreshRadvd(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
1678
                    virNetworkObjPtr network)
1679
{
G
Gene Czarcinski 已提交
1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690
    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)) {
1691
            virPidFileDelete(driverState->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1692 1693 1694 1695 1696 1697
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }

1698 1699
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
G
Gene Czarcinski 已提交
1700
        return networkStartRadvd(driver, network);
1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712

    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);
}

1713 1714
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1715
static int
1716
networkRestartRadvd(virNetworkDriverStatePtr driver,
1717
                    virNetworkObjPtr network)
1718 1719 1720 1721 1722 1723 1724 1725 1726
{
    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
1727
        if ((networkKillDaemon(network->radvdPid, "radvd",
1728 1729 1730
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1731
            virPidFileDelete(driverState->pidDir, radvdpidbase);
1732 1733 1734 1735 1736 1737 1738 1739 1740
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1741 1742 1743 1744
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
1745
networkRefreshDaemons(virNetworkDriverStatePtr driver)
1746
{
1747
    size_t i;
1748 1749 1750

    VIR_INFO("Refreshing network daemons");

1751
    for (i = 0; i < driver->networks.count; i++) {
1752 1753 1754 1755
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
1756 1757 1758
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
1759 1760 1761 1762 1763 1764
            /* 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.
             */
1765 1766
            networkRefreshDhcpDaemon(driver, network);
            networkRefreshRadvd(driver, network);
1767 1768 1769 1770 1771
        }
        virNetworkObjUnlock(network);
    }
}

1772
static void
1773
networkReloadFirewallRules(virNetworkDriverStatePtr driver)
1774
{
1775
    size_t i;
1776

1777
    VIR_INFO("Reloading iptables rules");
1778

1779
    for (i = 0; i < driver->networks.count; i++) {
1780 1781 1782 1783
        virNetworkObjPtr network = driver->networks.objs[i];

        virNetworkObjLock(network);
        if (virNetworkObjIsActive(network) &&
1784 1785 1786
            ((network->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
             (network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
1787 1788 1789
            /* Only the three L3 network types that are configured by libvirt
             * need to have iptables rules reloaded.
             */
1790 1791
            networkRemoveFirewallRules(network->def);
            if (networkAddFirewallRules(network->def) < 0) {
1792 1793
                /* failed to add but already logged */
            }
1794
        }
1795
        virNetworkObjUnlock(network);
1796 1797 1798
    }
}

1799
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
1800
static int
1801
networkEnableIpForwarding(bool enableIPv4, bool enableIPv6)
1802
{
1803
    int ret = 0;
1804 1805 1806 1807
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
1808
                           &enabled, sizeof(enabled));
1809 1810
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
1811
                           &enabled, sizeof(enabled));
1812
#else
1813 1814 1815 1816
    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);
1817
#endif
1818
    return ret;
1819 1820
}

1821 1822
#define SYSCTL_PATH "/proc/sys"

1823 1824
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
1825 1826 1827
{
    char *field = NULL;
    int ret = -1;
1828
    bool enableIPv6 =  !!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0);
1829

1830 1831 1832 1833 1834 1835 1836
    /* set disable_ipv6 if there are no ipv6 addresses defined for the
     * network. But also unset it if there *are* ipv6 addresses, as we
     * can't be sure of its default value.
     */
    if (virAsprintf(&field, SYSCTL_PATH "/net/ipv6/conf/%s/disable_ipv6",
                    network->def->bridge) < 0)
       goto cleanup;
1837

1838 1839
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
1840 1841
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
1842 1843 1844
        ret = 0;
        goto cleanup;
    }
1845

1846 1847 1848 1849 1850
    if (virFileWriteStr(field, enableIPv6 ? "0" : "1", 0) < 0) {
        virReportSystemError(errno,
                             _("cannot write to %s to enable/disable IPv6 "
                               "on bridge %s"), field, network->def->bridge);
        goto cleanup;
1851
    }
1852
    VIR_FREE(field);
1853

1854 1855
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
1856 1857 1858 1859 1860 1861
     */

    /* 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",
1862
                    network->def->bridge) < 0)
1863 1864
        goto cleanup;

1865
    if (virFileWriteStr(field, "0", 0) < 0) {
1866
        virReportSystemError(errno,
1867 1868 1869 1870 1871
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

1872 1873 1874 1875
    /* 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",
1876
                    network->def->bridge) < 0)
1877 1878
        goto cleanup;

1879
    if (virFileWriteStr(field, "0", 0) < 0) {
1880
        virReportSystemError(errno,
1881
                             _("cannot disable %s"), field);
1882 1883 1884 1885
        goto cleanup;
    }

    ret = 0;
1886
 cleanup:
1887 1888 1889 1890
    VIR_FREE(field);
    return ret;
}

1891
/* add an IP address to a bridge */
1892
static int
D
Daniel P. Berrange 已提交
1893
networkAddAddrToBridge(virNetworkObjPtr network,
1894
                       virNetworkIpDefPtr ipdef)
1895
{
1896 1897 1898
    int prefix = virNetworkIpDefPrefix(ipdef);

    if (prefix < 0) {
1899 1900 1901
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
1902 1903 1904
        return -1;
    }

1905 1906
    if (virNetDevSetIPv4Address(network->def->bridge,
                                &ipdef->address, prefix) < 0)
1907 1908 1909 1910 1911
        return -1;

    return 0;
}

1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924
/* 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,
1925
                                    (VIR_SOCKET_ADDR_IS_FAMILY(addr, AF_INET)
1926 1927 1928 1929 1930 1931 1932
                                     ? "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) &&
J
Ján Tomko 已提交
1933
                  virSocketAddrEqual(mask, &zero)))
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
            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;
}

1961
static int
1962
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
J
Ján Tomko 已提交
1963
                           virNetworkObjPtr network)
1964
{
1965
    size_t i;
1966
    bool v4present = false, v6present = false;
1967 1968
    virErrorPtr save_err = NULL;
    virNetworkIpDefPtr ipdef;
1969
    virNetworkRouteDefPtr routedef;
1970
    char *macTapIfName = NULL;
1971
    int tapfd = -1;
1972

1973
    /* Check to see if any network IP collides with an existing route */
1974
    if (networkCheckRouteCollision(network->def) < 0)
1975 1976
        return -1;

1977
    /* Create and configure the bridge device */
1978
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
1979 1980
        return -1;

1981 1982 1983 1984 1985 1986 1987 1988
    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);
1989
        if (!macTapIfName)
1990
            goto err0;
1991
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
1992
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
1993
                                           &macTapIfName, &network->def->mac,
1994
                                           NULL, &tapfd, 1, NULL, NULL,
1995 1996 1997
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
1998 1999 2000 2001 2002
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2003
    /* Set bridge options */
2004 2005 2006 2007

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2008
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2009
                                   network->def->delay * 1000) < 0)
2010
        goto err1;
2011

2012
    if (virNetDevBridgeSetSTP(network->def->bridge,
2013
                              network->def->stp ? true : false) < 0)
2014
        goto err1;
2015

2016 2017 2018 2019
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2020
        goto err1;
2021

2022
    /* Add "once per network" rules */
2023
    if (networkAddFirewallRules(network->def) < 0)
2024 2025
        goto err1;

2026 2027 2028
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
2029
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2030
            v4present = true;
2031
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2032
            v6present = true;
2033

2034
        /* Add the IP address/netmask to the bridge */
D
Daniel P. Berrange 已提交
2035
        if (networkAddAddrToBridge(network, ipdef) < 0) {
2036
            goto err2;
2037
        }
2038 2039
    }

2040
    /* Bring up the bridge interface */
2041
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2042
        goto err2;
2043

2044 2045
    for (i = 0; i < network->def->nroutes; i++) {
        routedef = &network->def->routes[i];
2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056
        /* 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 */
            }
        }
    }

2057 2058
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2059
        networkEnableIpForwarding(v4present, v6present) < 0) {
2060
        virReportSystemError(errno, "%s",
2061
                             _("failed to enable IP forwarding"));
2062
        goto err3;
2063 2064
    }

2065

2066
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2067 2068
    if ((v4present || v6present) &&
        networkStartDhcpDaemon(driver, network) < 0)
2069
        goto err3;
2070

2071
    /* start radvd if there are any ipv6 addresses */
G
Gene Czarcinski 已提交
2072
    if (v6present && networkStartRadvd(driver, network) < 0)
2073 2074
        goto err4;

2075 2076 2077 2078 2079 2080 2081 2082 2083
    /* 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);
    }

2084
    if (virNetDevBandwidthSet(network->def->bridge,
2085
                              network->def->bandwidth, true) < 0)
2086 2087
        goto err5;

2088
    VIR_FREE(macTapIfName);
2089 2090 2091

    return 0;

2092
 err5:
2093
    virNetDevBandwidthClear(network->def->bridge);
2094

2095 2096 2097 2098
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2099 2100 2101 2102 2103
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2104 2105 2106
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2107
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2108

2109 2110 2111
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2112
    networkRemoveFirewallRules(network->def);
2113 2114

 err1:
2115 2116 2117
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2118
    if (macTapIfName) {
2119
        VIR_FORCE_CLOSE(tapfd);
H
Hu Tao 已提交
2120 2121 2122
        ignore_value(virNetDevTapDelete(macTapIfName));
        VIR_FREE(macTapIfName);
    }
2123 2124

 err0:
2125 2126
    if (!save_err)
        save_err = virSaveLastError();
2127
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2128

2129 2130 2131 2132
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2133
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2134 2135 2136
    return -1;
}

2137
static int networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
R
Roman Bogorodskiy 已提交
2138
                                         virNetworkObjPtr network)
2139
{
2140
    virNetDevBandwidthClear(network->def->bridge);
2141

2142 2143 2144 2145 2146
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2147
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2148
            virPidFileDelete(driverState->pidDir, radvdpidbase);
2149 2150 2151 2152
            VIR_FREE(radvdpidbase);
        }
    }

2153 2154 2155
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2156
    if (network->def->mac_specified) {
2157
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2158
        if (macTapIfName) {
2159
            ignore_value(virNetDevTapDelete(macTapIfName));
2160 2161 2162 2163
            VIR_FREE(macTapIfName);
        }
    }

2164
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2165

2166
    networkRemoveFirewallRules(network->def);
2167

2168
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2169

2170
    /* See if its still alive and really really kill it */
2171
    if (network->dnsmasqPid > 0 &&
2172
        (kill(network->dnsmasqPid, 0) == 0))
2173 2174
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2175 2176 2177 2178 2179 2180

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

2181 2182 2183
    return 0;
}

2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199

/* networkCreateInterfacePool:
 * @netdef: the original NetDef from the network
 *
 * Creates an implicit interface pool of VF's when a PF dev is given
 */
static int
networkCreateInterfacePool(virNetworkDefPtr netdef)
{
    size_t numVirtFns = 0;
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2200 2201 2202
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev,
                                      &vfNames, &virtFns, &numVirtFns)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Could not get Virtual functions on %s"),
                       netdef->forward.pfs->dev);
        goto cleanup;
    }

    if (VIR_ALLOC_N(netdef->forward.ifs, numVirtFns) < 0)
        goto cleanup;

    for (i = 0; i < numVirtFns; i++) {
        virPCIDeviceAddressPtr thisVirtFn = virtFns[i];
        const char *thisName = vfNames[i];
        virNetworkForwardIfDefPtr thisIf
            = &netdef->forward.ifs[netdef->forward.nifs];

        switch (netdef->forward.type) {
        case VIR_NETWORK_FORWARD_BRIDGE:
        case VIR_NETWORK_FORWARD_PRIVATE:
        case VIR_NETWORK_FORWARD_VEPA:
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
            if (thisName) {
                if (VIR_STRDUP(thisIf->device.dev, thisName) < 0)
                    goto cleanup;
                thisIf->type = VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV;
                netdef->forward.nifs++;
            } else {
                VIR_WARN("VF %zu of SRIOV PF %s couldn't be added to the "
                         "interface pool because it isn't bound "
                         "to a network driver - possibly in use elsewhere",
                         i, netdef->forward.pfs->dev);
            }
            break;

        case VIR_NETWORK_FORWARD_HOSTDEV:
            /* VF's are always PCI devices */
            thisIf->type = VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI;
            thisIf->device.pci.domain = thisVirtFn->domain;
            thisIf->device.pci.bus = thisVirtFn->bus;
            thisIf->device.pci.slot = thisVirtFn->slot;
            thisIf->device.pci.function = thisVirtFn->function;
            netdef->forward.nifs++;
            break;

        case VIR_NETWORK_FORWARD_NONE:
        case VIR_NETWORK_FORWARD_NAT:
        case VIR_NETWORK_FORWARD_ROUTE:
        case VIR_NETWORK_FORWARD_LAST:
            /* by definition these will never be encountered here */
            break;
        }
    }

    if (netdef->forward.nifs == 0) {
        /* If we don't get at least one interface in the pool, declare
         * failure
         */
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("No usable Vf's present on SRIOV PF %s"),
                       netdef->forward.pfs->dev);
        goto cleanup;
    }

    ret = 0;
 cleanup:
    if (ret < 0) {
        /* free all the entries made before error */
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV)
                VIR_FREE(netdef->forward.ifs[i].device.dev);
        }
        netdef->forward.nifs = 0;
    }
    if (netdef->forward.nifs == 0)
        VIR_FREE(netdef->forward.ifs);

    for (i = 0; i < numVirtFns; i++) {
        VIR_FREE(vfNames[i]);
        VIR_FREE(virtFns[i]);
    }
    VIR_FREE(vfNames);
    VIR_FREE(virtFns);
    return ret;
}


2291
static int
2292
networkStartNetworkExternal(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
2293
                            virNetworkObjPtr network)
2294 2295
{
    /* put anything here that needs to be done each time a network of
2296
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2297 2298 2299
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2300
    return networkCreateInterfacePool(network->def);
2301 2302
}

2303
static int networkShutdownNetworkExternal(virNetworkDriverStatePtr driver ATTRIBUTE_UNUSED,
J
Ján Tomko 已提交
2304
                                          virNetworkObjPtr network ATTRIBUTE_UNUSED)
2305 2306
{
    /* put anything here that needs to be done each time a network of
2307
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2308 2309 2310 2311 2312 2313 2314
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2315
networkStartNetwork(virNetworkDriverStatePtr driver,
2316 2317
                    virNetworkObjPtr network)
{
2318 2319 2320
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2321 2322

    if (virNetworkObjIsActive(network)) {
2323 2324
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2325
        return ret;
2326 2327
    }

2328 2329 2330
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2331
    if (virNetworkObjSetDefTransient(network, true) < 0)
2332
        goto cleanup;
2333

2334 2335
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2336
    if (networkRunHook(network, NULL, NULL,
2337 2338 2339 2340
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2341
    switch (network->def->forward.type) {
2342 2343 2344 2345

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2346 2347
        if (networkStartNetworkVirtual(driver, network) < 0)
            goto cleanup;
2348 2349 2350 2351 2352 2353
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2354
    case VIR_NETWORK_FORWARD_HOSTDEV:
2355 2356
        if (networkStartNetworkExternal(driver, network) < 0)
            goto cleanup;
2357 2358 2359
        break;
    }

2360
    /* finally we can call the 'started' hook script if any */
2361
    if (networkRunHook(network, NULL, NULL,
2362 2363 2364 2365
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2366 2367 2368
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2369 2370 2371
    VIR_DEBUG("Writing network status to disk");
    if (virNetworkSaveStatus(driverState->stateDir, network) < 0)
        goto cleanup;
2372 2373

    network->active = 1;
2374 2375
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2376

2377
 cleanup:
2378
    if (ret < 0) {
2379
        virNetworkObjUnsetDefTransient(network);
2380 2381 2382 2383 2384 2385 2386 2387 2388 2389
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
        networkShutdownNetwork(driver, network);
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2390
static int networkShutdownNetwork(virNetworkDriverStatePtr driver,
J
Ján Tomko 已提交
2391
                                  virNetworkObjPtr network)
2392 2393 2394 2395 2396 2397 2398 2399 2400
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

2401 2402
    stateFile = virNetworkConfigFile(driverState->stateDir,
                                     network->def->name);
2403 2404 2405 2406 2407 2408
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2409
    switch (network->def->forward.type) {
2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420

    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:
2421
    case VIR_NETWORK_FORWARD_HOSTDEV:
2422 2423 2424 2425
        ret = networkShutdownNetworkExternal(driver, network);
        break;
    }

2426
    /* now that we know it's stopped call the hook if present */
2427
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2428 2429
                   VIR_HOOK_SUBOP_END);

2430
    network->active = 0;
2431
    virNetworkObjUnsetDefTransient(network);
2432
    return ret;
2433 2434 2435
}


2436
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2437 2438
                                         const unsigned char *uuid)
{
2439
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2440 2441
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2442

2443
    networkDriverLock(driver);
2444
    network = virNetworkFindByUUID(&driver->networks, uuid);
2445
    networkDriverUnlock(driver);
2446
    if (!network) {
2447 2448
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
2449
        goto cleanup;
2450 2451
    }

2452 2453 2454
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2455 2456
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2457
 cleanup:
2458 2459
    if (network)
        virNetworkObjUnlock(network);
2460
    return ret;
2461 2462
}

2463
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2464 2465
                                         const char *name)
{
2466
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2467 2468 2469
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2470
    networkDriverLock(driver);
2471
    network = virNetworkFindByName(&driver->networks, name);
2472
    networkDriverUnlock(driver);
2473
    if (!network) {
2474 2475
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2476
        goto cleanup;
2477 2478
    }

2479 2480 2481
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2482 2483
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2484
 cleanup:
2485 2486
    if (network)
        virNetworkObjUnlock(network);
2487
    return ret;
2488 2489
}

2490 2491 2492
static virDrvOpenStatus networkOpen(virConnectPtr conn,
                                    virConnectAuthPtr auth ATTRIBUTE_UNUSED,
                                    unsigned int flags)
2493 2494 2495
{
    virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);

2496 2497 2498 2499 2500 2501 2502
    if (!driverState)
        return VIR_DRV_OPEN_DECLINED;

    conn->networkPrivateData = driverState;
    return VIR_DRV_OPEN_SUCCESS;
}

2503 2504
static int networkClose(virConnectPtr conn)
{
2505 2506 2507 2508
    conn->networkPrivateData = NULL;
    return 0;
}

2509 2510
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2511 2512
    int nactive = 0;
    size_t i;
2513
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2514

2515 2516 2517
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2518
    networkDriverLock(driver);
2519
    for (i = 0; i < driver->networks.count; i++) {
2520 2521 2522 2523
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectNumOfNetworksCheckACL(conn, obj->def) &&
            virNetworkObjIsActive(obj))
2524
            nactive++;
2525
        virNetworkObjUnlock(obj);
2526 2527
    }
    networkDriverUnlock(driver);
2528

2529 2530 2531
    return nactive;
}

2532
static int networkConnectListNetworks(virConnectPtr conn, char **const names, int nnames) {
2533
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2534 2535
    int got = 0;
    size_t i;
2536

2537 2538 2539
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2540
    networkDriverLock(driver);
2541
    for (i = 0; i < driver->networks.count && got < nnames; i++) {
2542 2543 2544 2545 2546 2547
        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);
2548 2549 2550 2551
                goto cleanup;
            }
            got++;
        }
2552
        virNetworkObjUnlock(obj);
2553
    }
2554 2555
    networkDriverUnlock(driver);

2556 2557 2558
    return got;

 cleanup:
2559
    networkDriverUnlock(driver);
2560
    for (i = 0; i < got; i++)
2561 2562 2563 2564
        VIR_FREE(names[i]);
    return -1;
}

2565 2566
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2567 2568
    int ninactive = 0;
    size_t i;
2569
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2570

2571 2572 2573
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2574
    networkDriverLock(driver);
2575
    for (i = 0; i < driver->networks.count; i++) {
2576 2577 2578 2579
        virNetworkObjPtr obj = driver->networks.objs[i];
        virNetworkObjLock(obj);
        if (virConnectNumOfDefinedNetworksCheckACL(conn, obj->def) &&
            !virNetworkObjIsActive(obj))
2580
            ninactive++;
2581
        virNetworkObjUnlock(obj);
2582 2583
    }
    networkDriverUnlock(driver);
2584

2585 2586 2587
    return ninactive;
}

2588
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames) {
2589
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2590 2591
    int got = 0;
    size_t i;
2592

2593 2594 2595
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2596
    networkDriverLock(driver);
2597
    for (i = 0; i < driver->networks.count && got < nnames; i++) {
2598 2599 2600 2601 2602 2603
        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);
2604 2605 2606 2607
                goto cleanup;
            }
            got++;
        }
2608
        virNetworkObjUnlock(obj);
2609
    }
2610
    networkDriverUnlock(driver);
2611 2612 2613
    return got;

 cleanup:
2614
    networkDriverUnlock(driver);
2615
    for (i = 0; i < got; i++)
2616 2617 2618 2619
        VIR_FREE(names[i]);
    return -1;
}

2620
static int
2621 2622 2623
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2624
{
2625
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2626 2627 2628 2629
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2630 2631 2632
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2633
    networkDriverLock(driver);
2634 2635 2636
    ret = virNetworkObjListExport(conn, driver->networks, nets,
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2637 2638
    networkDriverUnlock(driver);

2639
 cleanup:
2640 2641
    return ret;
}
2642

2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657
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,
2658
                                       net, eventID, callback,
2659 2660 2661
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2662
 cleanup:
2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675
    return ret;
}

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

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

2676 2677 2678 2679 2680 2681
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
2682

2683
 cleanup:
2684 2685 2686
    return ret;
}

2687 2688 2689 2690 2691
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2692 2693
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2694 2695 2696 2697

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

2698 2699
    ret = virNetworkObjIsActive(obj);

2700
 cleanup:
2701 2702 2703 2704 2705 2706 2707 2708 2709 2710
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}

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

2711 2712
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2713 2714 2715 2716

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

2717 2718
    ret = obj->persistent;

2719
 cleanup:
2720 2721 2722 2723 2724 2725
    if (obj)
        virNetworkObjUnlock(obj);
    return ret;
}


2726
static int
2727
networkValidate(virNetworkDriverStatePtr driver,
2728 2729
                virNetworkDefPtr def,
                bool check_active)
2730
{
2731
    size_t i;
2732 2733
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2734
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
2735
    bool ipv4def = false, ipv6def = false;
2736 2737 2738 2739 2740 2741 2742 2743

    /* 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
     */
2744 2745 2746
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2747 2748 2749 2750 2751

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

        virNetworkSetBridgeMacAddr(def);
2752 2753
    } else {
        /* They are also the only types that currently support setting
2754 2755
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
2756
         */
2757 2758 2759 2760 2761 2762 2763 2764
        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;
        }
2765 2766 2767 2768 2769
        if (virNetworkDefGetIpByIndex(def, AF_UNSPEC, 0)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2770
                           virNetworkForwardTypeToString(def->forward.type));
2771 2772
            return -1;
        }
2773
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
2774 2775 2776 2777
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2778
                           virNetworkForwardTypeToString(def->forward.type));
2779 2780 2781 2782 2783 2784 2785
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2786
                           virNetworkForwardTypeToString(def->forward.type));
2787 2788
            return -1;
        }
2789 2790 2791 2792 2793 2794 2795 2796
        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;
        }
2797 2798
    }

G
Gene Czarcinski 已提交
2799 2800 2801
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
2802 2803 2804
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
2805 2806 2807 2808 2809
        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 -- "
2810 2811
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828
                    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;
                }
2829 2830 2831
            }
        }
    }
2832 2833 2834 2835 2836 2837

    /* 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.
     */

2838
    vlanAllowed = ((def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
J
Ján Tomko 已提交
2839 2840
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
2841 2842
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) ||
                   def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV);
2843 2844

    vlanUsed = def->vlan.nTags > 0;
2845 2846
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
2847 2848 2849 2850 2851
            /* 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.
             */
2852
            if (def->portGroups[i].virtPortProfile) {
2853
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
2854
                    def->portGroups[i].virtPortProfile->virtPortType
2855 2856 2857 2858 2859 2860 2861
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
2862
        }
2863
        if (def->portGroups[i].isDefault) {
2864 2865 2866 2867 2868
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
2869
                               def->name, defaultPortGroup->name,
2870
                               def->portGroups[i].name);
2871
                return -1;
2872
            }
2873
            defaultPortGroup = &def->portGroups[i];
2874
        }
2875
    }
2876 2877 2878 2879 2880 2881 2882
    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.
         */
2883 2884 2885 2886 2887 2888 2889 2890 2891
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
    return 0;
}

2892 2893
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
2894
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2895
    virNetworkDefPtr def;
2896
    virNetworkObjPtr network = NULL;
2897
    virNetworkPtr ret = NULL;
2898
    virObjectEventPtr event = NULL;
2899

2900 2901
    networkDriverLock(driver);

2902
    if (!(def = virNetworkDefParseString(xml)))
2903
        goto cleanup;
2904

2905 2906 2907
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

2908
    if (networkValidate(driver, def, true) < 0)
J
Ján Tomko 已提交
2909
        goto cleanup;
2910

2911 2912 2913
    /* 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.
2914
     */
2915
    if (!(network = virNetworkAssignDef(&driver->networks, def, true)))
2916 2917
        goto cleanup;
    def = NULL;
2918

2919
    if (networkStartNetwork(driver, network) < 0) {
2920 2921
        virNetworkRemoveInactive(&driver->networks,
                                 network);
2922
        network = NULL;
2923
        goto cleanup;
2924 2925
    }

2926 2927
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
2928 2929
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
2930

2931
    VIR_INFO("Creating network '%s'", network->def->name);
2932 2933
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2934
 cleanup:
2935
    virNetworkDefFree(def);
2936 2937
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
2938 2939 2940
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
2941
    return ret;
2942 2943
}

2944 2945
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
2946
    virNetworkDriverStatePtr driver = conn->networkPrivateData;
2947
    virNetworkDefPtr def = NULL;
2948
    bool freeDef = true;
2949
    virNetworkObjPtr network = NULL;
2950
    virNetworkPtr ret = NULL;
2951
    virObjectEventPtr event = NULL;
2952

2953 2954
    networkDriverLock(driver);

2955
    if (!(def = virNetworkDefParseString(xml)))
2956
        goto cleanup;
2957

2958 2959 2960
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

2961
    if (networkValidate(driver, def, false) < 0)
J
Ján Tomko 已提交
2962
        goto cleanup;
2963

2964
    if (!(network = virNetworkAssignDef(&driver->networks, def, false)))
J
Ján Tomko 已提交
2965
        goto cleanup;
2966

2967
    /* def was assigned to network object */
2968
    freeDef = false;
2969 2970

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
2971 2972 2973 2974 2975
        if (!virNetworkObjIsActive(network)) {
            virNetworkRemoveInactive(&driver->networks, network);
            network = NULL;
            goto cleanup;
        }
2976 2977 2978 2979 2980
        /* 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);
2981 2982 2983
        goto cleanup;
    }

2984
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
2985 2986
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
2987

2988 2989
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
2990

2991
 cleanup:
2992 2993
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
2994
    if (freeDef)
J
Ján Tomko 已提交
2995
        virNetworkDefFree(def);
2996 2997 2998
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
2999
    return ret;
3000 3001
}

3002
static int
3003 3004
networkUndefine(virNetworkPtr net)
{
3005
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3006
    virNetworkObjPtr network;
3007
    int ret = -1;
3008
    bool active = false;
3009
    virObjectEventPtr event = NULL;
3010

3011 3012
    networkDriverLock(driver);

3013
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3014
    if (!network) {
3015 3016
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3017
        goto cleanup;
3018 3019
    }

3020 3021 3022
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3023 3024
    if (virNetworkObjIsActive(network))
        active = true;
3025

3026
    /* remove autostart link */
3027
    if (virNetworkDeleteConfig(driver->networkConfigDir,
3028 3029
                               driver->networkAutostartDir,
                               network) < 0)
3030
        goto cleanup;
3031
    network->autostart = 0;
3032

3033 3034
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3035 3036
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3037

3038
    VIR_INFO("Undefining network '%s'", network->def->name);
3039 3040 3041 3042 3043
    if (!active) {
        if (networkRemoveInactive(driver, network) < 0) {
            network = NULL;
            goto cleanup;
        }
3044
        network = NULL;
3045 3046 3047 3048 3049 3050
    } 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);
3051 3052
    }

3053
    ret = 0;
3054

3055
 cleanup:
3056 3057
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3058 3059 3060
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
3061
    return ret;
3062 3063
}

3064 3065 3066 3067 3068 3069 3070 3071
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3072
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3073
    virNetworkObjPtr network = NULL;
3074 3075
    int isActive, ret = -1;
    size_t i;
3076 3077
    virNetworkIpDefPtr ipdef;
    bool oldDhcpActive = false;
3078
    bool needFirewallRefresh = false;
3079

3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093

    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;
    }

3094 3095 3096
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3097
    /* see if we are listening for dhcp pre-modification */
3098 3099 3100
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
3101 3102 3103 3104 3105 3106
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3107 3108
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3109
     */
3110
    isActive = virNetworkObjIsActive(network);
3111 3112
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3113 3114 3115 3116 3117 3118 3119
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136
    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).
                 */
3137
                networkRemoveFirewallRules(network->def);
3138 3139 3140 3141 3142 3143 3144 3145
                needFirewallRefresh = true;
                break;
            default:
                break;
            }
        }
    }

3146
    /* update the network config in memory/on disk */
3147 3148
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3149
            ignore_value(networkAddFirewallRules(network->def));
3150 3151 3152
        goto cleanup;
    }

3153
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176
        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.
             */
3177
            if (networkRestartDhcpDaemon(driver, network) < 0)
3178 3179
                goto cleanup;

3180 3181 3182 3183 3184 3185 3186 3187
        } 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;

3188 3189 3190
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
                 i++) {
3191 3192 3193 3194 3195 3196 3197
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3198 3199
                 networkRestartDhcpDaemon(driver, network) < 0) ||
                networkRefreshDhcpDaemon(driver, network) < 0) {
3200 3201 3202 3203
                goto cleanup;
            }

        } else if (section == VIR_NETWORK_SECTION_DNS_HOST ||
3204 3205 3206 3207 3208 3209
                   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.
             */
3210
            if (networkRefreshDhcpDaemon(driver, network) < 0)
3211 3212 3213 3214 3215 3216 3217 3218
                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.
             */
3219
            if (networkRefreshRadvd(driver, network) < 0)
3220 3221 3222 3223
                goto cleanup;
        }

        /* save current network state to disk */
3224 3225
        if ((ret = virNetworkSaveStatus(driverState->stateDir,
                                        network)) < 0) {
3226
            goto cleanup;
3227
        }
3228 3229
    }
    ret = 0;
3230
 cleanup:
3231 3232 3233 3234 3235 3236
    if (network)
        virNetworkObjUnlock(network);
    networkDriverUnlock(driver);
    return ret;
}

3237 3238
static int networkCreate(virNetworkPtr net)
{
3239
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3240 3241
    virNetworkObjPtr network;
    int ret = -1;
3242
    virObjectEventPtr event = NULL;
3243

3244
    networkDriverLock(driver);
3245
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3246

3247
    if (!network) {
3248 3249
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3250
        goto cleanup;
3251 3252
    }

3253 3254 3255
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3256
    ret = networkStartNetwork(driver, network);
3257

3258 3259
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3260 3261
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3262

3263
 cleanup:
3264 3265
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3266 3267
    if (network)
        virNetworkObjUnlock(network);
3268
    networkDriverUnlock(driver);
3269
    return ret;
3270 3271
}

3272 3273
static int networkDestroy(virNetworkPtr net)
{
3274
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3275 3276
    virNetworkObjPtr network;
    int ret = -1;
3277
    virObjectEventPtr event = NULL;
3278

3279
    networkDriverLock(driver);
3280
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3281

3282
    if (!network) {
3283 3284
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3285
        goto cleanup;
3286 3287
    }

3288 3289 3290
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3291
    if (!virNetworkObjIsActive(network)) {
3292
        virReportError(VIR_ERR_OPERATION_INVALID,
3293 3294
                       _("network '%s' is not active"),
                       network->def->name);
3295 3296 3297
        goto cleanup;
    }

3298 3299 3300
    if ((ret = networkShutdownNetwork(driver, network)) < 0)
        goto cleanup;

3301 3302
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3303 3304
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3305

3306
    if (!network->persistent) {
3307 3308 3309 3310 3311
        if (networkRemoveInactive(driver, network) < 0) {
            network = NULL;
            ret = -1;
            goto cleanup;
        }
3312 3313
        network = NULL;
    }
3314

3315
 cleanup:
3316 3317
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3318 3319
    if (network)
        virNetworkObjUnlock(network);
3320
    networkDriverUnlock(driver);
3321 3322 3323
    return ret;
}

3324
static char *networkGetXMLDesc(virNetworkPtr net,
3325
                               unsigned int flags)
3326
{
3327
    virNetworkObjPtr network;
3328
    virNetworkDefPtr def;
3329
    char *ret = NULL;
3330

3331
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3332

3333 3334
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3335

3336 3337 3338
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3339 3340 3341 3342 3343 3344
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3345

3346
 cleanup:
3347 3348
    if (network)
        virNetworkObjUnlock(network);
3349
    return ret;
3350 3351 3352
}

static char *networkGetBridgeName(virNetworkPtr net) {
3353 3354 3355
    virNetworkObjPtr network;
    char *bridge = NULL;

3356 3357
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3358

3359 3360 3361
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3362
    if (!(network->def->bridge)) {
3363 3364 3365
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3366 3367 3368
        goto cleanup;
    }

3369
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3370

3371
 cleanup:
3372 3373
    if (network)
        virNetworkObjUnlock(network);
3374 3375 3376 3377
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
J
Ján Tomko 已提交
3378
                               int *autostart)
3379
{
3380 3381
    virNetworkObjPtr network;
    int ret = -1;
3382

3383 3384
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3385

3386 3387 3388
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3389
    *autostart = network->autostart;
3390
    ret = 0;
3391

3392
 cleanup:
3393 3394
    if (network)
        virNetworkObjUnlock(network);
3395
    return ret;
3396 3397 3398
}

static int networkSetAutostart(virNetworkPtr net,
3399 3400
                               int autostart)
{
3401
    virNetworkDriverStatePtr driver = net->conn->networkPrivateData;
3402
    virNetworkObjPtr network;
3403
    char *configFile = NULL, *autostartLink = NULL;
3404
    int ret = -1;
3405

3406
    networkDriverLock(driver);
3407
    network = virNetworkFindByUUID(&driver->networks, net->uuid);
3408

3409
    if (!network) {
3410 3411
        virReportError(VIR_ERR_NO_NETWORK,
                       "%s", _("no network with matching uuid"));
3412
        goto cleanup;
3413 3414
    }

3415 3416 3417
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3418
    if (!network->persistent) {
3419 3420
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3421 3422 3423
        goto cleanup;
    }

3424 3425
    autostart = (autostart != 0);

3426
    if (network->autostart != autostart) {
3427
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3428
            goto cleanup;
3429
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3430 3431
            goto cleanup;

3432
        if (autostart) {
3433
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3434
                virReportSystemError(errno,
3435 3436
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3437 3438
                goto cleanup;
            }
3439

3440
            if (symlink(configFile, autostartLink) < 0) {
3441
                virReportSystemError(errno,
3442
                                     _("Failed to create symlink '%s' to '%s'"),
3443
                                     autostartLink, configFile);
3444 3445 3446
                goto cleanup;
            }
        } else {
3447
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3448
                virReportSystemError(errno,
3449
                                     _("Failed to delete symlink '%s'"),
3450
                                     autostartLink);
3451 3452
                goto cleanup;
            }
3453 3454
        }

3455
        network->autostart = autostart;
3456
    }
3457
    ret = 0;
3458

3459
 cleanup:
3460 3461
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3462 3463
    if (network)
        virNetworkObjUnlock(network);
3464
    networkDriverUnlock(driver);
3465
    return ret;
3466 3467
}

3468
static int
3469 3470 3471 3472
networkGetDHCPLeases(virNetworkPtr network,
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491
{
    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;
3492 3493 3494 3495 3496 3497 3498 3499 3500
    virNetworkObjPtr obj;

    virCheckFlags(0, -1);

    if (!(obj = networkObjFromNetwork(network)))
        return -1;

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

    /* 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;
        }

3547
        if (mac && virMacAddrCompare(mac, mac_tmp))
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 3593 3594 3595 3596 3597
            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) ||
3598
                (VIR_STRDUP(lease->iface, obj->def->bridge) < 0))
3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636
                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);
3637
    VIR_FREE(lease_entries);
3638 3639
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
3640 3641 3642 3643

    if (obj)
        virNetworkObjUnlock(obj);

3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654
    return rv;

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

3655 3656 3657

static virNetworkDriver networkDriver = {
    "Network",
3658 3659 3660 3661 3662 3663 3664
    .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 */
3665 3666
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
3667 3668
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3669 3670
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3671
    .networkUndefine = networkUndefine, /* 0.2.0 */
3672
    .networkUpdate = networkUpdate, /* 0.10.2 */
3673
    .networkCreate = networkCreate, /* 0.2.0 */
3674 3675 3676 3677 3678 3679 3680
    .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 */
3681
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
3682 3683 3684
};

static virStateDriver networkStateDriver = {
3685
    .name = "Network",
3686
    .stateInitialize  = networkStateInitialize,
3687
    .stateAutoStart  = networkStateAutoStart,
3688 3689
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3690 3691
};

3692 3693
int networkRegister(void)
{
3694 3695
    if (virRegisterNetworkDriver(&networkDriver) < 0)
        return -1;
3696 3697
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
3698 3699
    return 0;
}
3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711

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

/* Private API to deal with logical switch capabilities.
 * These functions are exported so that other parts of libvirt can
 * call them, but are not part of the public API and not in the
 * driver's function table. If we ever have more than one network
 * driver, we will need to present these functions via a second
 * "backend" function table.
 */

/* networkAllocateActualDevice:
3712
 * @dom: domain definition that @iface belongs to
3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723
 * @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
3724 3725
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
3726
{
3727
    virNetworkDriverStatePtr driver = driverState;
3728
    virDomainNetType actualType = iface->type;
3729 3730
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
3731
    virNetDevBandwidthPtr bandwidth = NULL;
3732 3733 3734
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3735
    virNetworkForwardIfDefPtr dev = NULL;
3736
    size_t i;
3737 3738 3739
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3740
        goto validate;
3741 3742 3743 3744 3745 3746 3747 3748

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

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
3749 3750 3751
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3752
        goto error;
3753 3754
    }
    netdef = network->def;
3755

3756 3757 3758 3759 3760 3761 3762
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

3763 3764 3765
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

3766 3767 3768
    /* portgroup can be present for any type of network, in particular
     * for bandwidth information, so we need to check for that and
     * fill it in appropriately for all forward types.
J
Ján Tomko 已提交
3769
     */
3770 3771 3772 3773 3774 3775
    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.
     */
3776 3777 3778 3779 3780 3781

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

3782 3783
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
3784
        goto error;
3785

3786 3787 3788 3789 3790 3791 3792 3793
    /* 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;

3794 3795
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
3796

3797 3798 3799
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
3800 3801 3802
        /* for these forward types, the actual net type really *is*
         *NETWORK; we just keep the info from the portgroup in
         * iface->data.network.actual
J
Ján Tomko 已提交
3803
         */
3804
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
3805 3806 3807 3808

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

3809
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
3810
               netdef->bridge) {
3811 3812 3813 3814 3815

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

3816
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
3817 3818
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
3819
            goto error;
3820

3821 3822 3823 3824 3825 3826 3827 3828
        /* 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) {
3829
            goto error;
3830 3831 3832 3833 3834 3835 3836 3837 3838 3839
        }
        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);
3840
                goto error;
3841 3842 3843
            }
        }

3844
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3845

3846
        virDomainHostdevSubsysPCIBackendType backend;
3847

3848
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
3849
        if (networkCreateInterfacePool(netdef) < 0)
3850 3851 3852
            goto error;

        /* pick first dev with 0 connections */
3853 3854 3855
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869
                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;
3870
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
3871
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
3872
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
3873

3874 3875 3876
        switch (netdef->forward.driverName)
        {
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
3877
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
3878 3879
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
3880
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
3881 3882
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
3883
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894
            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;

3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919
        /* 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;
            }
        }

3920 3921 3922 3923
    } 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)) {
3924 3925 3926 3927 3928 3929

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
3930
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
3931
        switch (netdef->forward.type) {
3932
        case VIR_NETWORK_FORWARD_BRIDGE:
3933
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
3934 3935
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
3936
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
3937 3938
            break;
        case VIR_NETWORK_FORWARD_VEPA:
3939
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
3940 3941
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
3942
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
3943 3944 3945
            break;
        }

3946 3947 3948 3949 3950 3951 3952 3953
        /* 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) {
3954
            goto error;
3955
        }
3956
        virtport = iface->data.network.actual->virtPortProfile;
3957
        if (virtport) {
3958 3959 3960 3961 3962 3963 3964 3965
            /* 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);
3966
                goto error;
3967 3968
            }
        }
3969

3970 3971 3972
        /* 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).
         */
3973
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
3974 3975 3976 3977
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
3978
            goto error;
3979 3980 3981
        } else {
            /* pick an interface from the pool */

3982
            if (networkCreateInterfacePool(netdef) < 0)
3983 3984
                goto error;

3985 3986 3987 3988 3989
            /* 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.
3990
             */
3991 3992
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
3993 3994 3995
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
3996

3997
                /* pick first dev with 0 connections */
3998 3999 4000
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4001 4002 4003 4004 4005
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4006
                dev = &netdef->forward.ifs[0];
4007 4008 4009
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4010 4011 4012 4013
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4014 4015 4016 4017
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4018
                goto error;
4019
            }
4020 4021
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4022
                goto error;
4023 4024 4025
        }
    }

4026
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4027
        goto error;
4028

4029
 validate:
4030 4031 4032 4033 4034
    /* 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.
     */

4035
    if (virDomainNetGetActualVlan(iface)) {
4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064
        /* 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);
4065

4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084
        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 */
4085
            netdef->connections--;
4086 4087 4088 4089
            if (dev)
                dev->connections--;
            goto error;
        }
4090 4091
    }

4092
    ret = 0;
4093

4094
 cleanup:
4095 4096
    if (network)
        virNetworkObjUnlock(network);
4097 4098
    return ret;

4099
 error:
4100
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4101 4102 4103
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4104
    goto cleanup;
4105 4106 4107
}

/* networkNotifyActualDevice:
4108
 * @dom: domain definition that @iface belongs to
4109 4110 4111 4112 4113 4114 4115 4116 4117 4118
 * @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
4119 4120
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4121
{
4122
    virNetworkDriverStatePtr driver = driverState;
4123
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4124 4125
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4126
    virNetworkForwardIfDefPtr dev = NULL;
4127 4128
    size_t i;
    int ret = -1;
4129 4130

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4131
        return 0;
4132 4133 4134 4135 4136

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
4137 4138 4139
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4140 4141 4142 4143 4144
        goto error;
    }
    netdef = network->def;

    if (!iface->data.network.actual ||
4145 4146
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4147 4148
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4149 4150
    }

4151
    if (networkCreateInterfacePool(netdef) < 0)
4152
        goto error;
4153

4154
    if (netdef->forward.nifs == 0) {
4155
        virReportError(VIR_ERR_INTERNAL_ERROR,
4156 4157
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4158
                       netdef->name);
4159
        goto error;
4160
    }
4161

4162 4163
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4164

4165 4166 4167 4168 4169 4170 4171 4172 4173
        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 */
4174 4175
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4176
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4177 4178
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4179 4180 4181 4182 4183
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4184
            virReportError(VIR_ERR_INTERNAL_ERROR,
4185 4186
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4187
                           netdef->name, actualDev);
4188
            goto error;
4189 4190
        }

4191
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4192 4193
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4194
         */
4195
        if ((dev->connections > 0) &&
4196 4197
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4198 4199
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4200
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4201
            virReportError(VIR_ERR_INTERNAL_ERROR,
4202 4203
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4204
                           netdef->name, actualDev);
4205
            goto error;
4206
        }
4207

4208
        /* we are now assured of success, so mark the allocation */
4209
        dev->connections++;
4210
        VIR_DEBUG("Using physical device %s, connections %d",
4211
                  dev->device.dev, dev->connections);
4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224

    }  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 */
4225 4226
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4227
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4228
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4229 4230
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4231 4232 4233 4234 4235 4236 4237 4238 4239
                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,
4240 4241 4242 4243
                           hostdev->source.subsys.u.pci.addr.domain,
                           hostdev->source.subsys.u.pci.addr.bus,
                           hostdev->source.subsys.u.pci.addr.slot,
                           hostdev->source.subsys.u.pci.addr.function);
J
Ján Tomko 已提交
4244
            goto error;
4245 4246 4247 4248 4249 4250 4251
        }

        /* 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) &&
4252
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268
            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);
4269 4270
    }

4271
 success:
4272 4273 4274
    netdef->connections++;
    VIR_DEBUG("Using network %s, %d connections",
              netdef->name, netdef->connections);
4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285

    /* 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;
    }

4286
    ret = 0;
4287
 cleanup:
4288 4289 4290
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4291

4292
 error:
4293
    goto cleanup;
4294 4295 4296 4297
}


/* networkReleaseActualDevice:
4298
 * @dom: domain definition that @iface belongs to
4299 4300 4301 4302 4303 4304 4305 4306 4307 4308
 * @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
4309 4310
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4311
{
4312
    virNetworkDriverStatePtr driver = driverState;
4313
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4314
    virNetworkObjPtr network;
4315
    virNetworkDefPtr netdef;
4316
    virNetworkForwardIfDefPtr dev = NULL;
4317 4318
    size_t i;
    int ret = -1;
4319 4320

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4321
        return 0;
4322 4323 4324 4325 4326

    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, iface->data.network.name);
    networkDriverUnlock(driver);
    if (!network) {
4327 4328 4329
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4330 4331 4332 4333
        goto error;
    }
    netdef = network->def;

4334 4335
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4336 4337 4338 4339 4340
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) &&
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4341 4342 4343
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4344 4345
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4346 4347
    }

4348
    if (netdef->forward.nifs == 0) {
4349
        virReportError(VIR_ERR_INTERNAL_ERROR,
4350
                       _("network '%s' uses a direct/hostdev mode, but "
4351 4352
                         "has no forward dev and no interface pool"),
                       netdef->name);
4353
        goto error;
4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365
    }

    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;
        }
4366

4367 4368
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4369
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4370 4371
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4372 4373 4374
                break;
            }
        }
4375

4376
        if (!dev) {
4377
            virReportError(VIR_ERR_INTERNAL_ERROR,
4378 4379
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4380
                           netdef->name, actualDev);
4381
            goto error;
4382 4383
        }

4384
        dev->connections--;
4385
        VIR_DEBUG("Releasing physical device %s, connections %d",
4386
                  dev->device.dev, dev->connections);
4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397

    } 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;
        }

4398 4399
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4400
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4401
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4402 4403
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4404 4405 4406 4407 4408 4409 4410 4411 4412
                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,
4413 4414 4415 4416
                           hostdev->source.subsys.u.pci.addr.domain,
                           hostdev->source.subsys.u.pci.addr.bus,
                           hostdev->source.subsys.u.pci.addr.slot,
                           hostdev->source.subsys.u.pci.addr.function);
J
Ján Tomko 已提交
4417
            goto error;
4418 4419 4420 4421 4422 4423 4424
        }

        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);
J
Ján Tomko 已提交
4425
    }
4426

4427
 success:
4428
    if (iface->data.network.actual) {
4429
        netdef->connections--;
4430 4431
        VIR_DEBUG("Releasing network %s, %d connections",
                  netdef->name, netdef->connections);
4432

4433 4434 4435 4436
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
    }
4437
    ret = 0;
4438
 cleanup:
4439 4440
    if (network)
        virNetworkObjUnlock(network);
4441 4442 4443 4444
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4445
    return ret;
4446

4447
 error:
4448
    goto cleanup;
4449
}
4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473

/*
 * 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;
4474
    virNetworkDriverStatePtr driver = driverState;
4475
    virNetworkObjPtr network;
4476 4477 4478 4479
    virNetworkDefPtr netdef;
    virNetworkIpDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4480
    char *dev_name = NULL;
4481 4482 4483 4484 4485 4486

    *netaddr = NULL;
    networkDriverLock(driver);
    network = virNetworkFindByName(&driver->networks, netname);
    networkDriverUnlock(driver);
    if (!network) {
4487 4488 4489
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4490
        goto error;
4491 4492 4493
    }
    netdef = network->def;

4494
    switch (netdef->forward.type) {
4495 4496 4497 4498 4499 4500
    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) {
4501 4502 4503
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have an IPv4 address"),
                           netdef->name);
4504 4505 4506 4507 4508 4509
            break;
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4510
        if ((dev_name = netdef->bridge))
4511 4512 4513 4514 4515 4516 4517 4518
            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:
4519 4520
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4521

4522
        if (!dev_name) {
4523 4524 4525
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4526 4527 4528 4529
        }
        break;
    }

4530
    if (dev_name) {
4531
        if (virNetDevGetIPv4Address(dev_name, &addr) < 0)
4532
            goto error;
4533
        addrptr = &addr;
4534 4535
    }

4536 4537 4538
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
        goto error;
4539 4540
    }

4541
    ret = 0;
4542
 cleanup:
4543 4544 4545
    if (network)
        virNetworkObjUnlock(network);
    return ret;
4546

4547
 error:
4548
    goto cleanup;
4549
}
4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567

/**
 * 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;
4568
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4569 4570 4571 4572
    unsigned long long tmp_floor_sum = net->floor_sum;
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583
    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;
    }

4584
    if (!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor ||
4585 4586
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4587
        return 1;
4588
    }
4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619

    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;

4620
 cleanup:
4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657
    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];
4658
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686

    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;
    }

4687 4688
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
4689 4690 4691 4692 4693 4694 4695 4696
    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 */
4697
    net->floor_sum += ifaceBand->in->floor;
4698
    /* update status file */
4699
    if (virNetworkSaveStatus(driverState->stateDir, net) < 0) {
4700
        ignore_value(virBitmapClearBit(net->class_id, class_id));
4701
        net->floor_sum -= ifaceBand->in->floor;
4702 4703 4704 4705
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
4706 4707 4708 4709 4710 4711 4712 4713 4714
    /* 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;

4715
 cleanup:
4716 4717 4718 4719 4720 4721 4722 4723 4724
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
    int ret = 0;
    unsigned long long new_rate;
4725
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4726 4727 4728

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
4729 4730 4731 4732 4733
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744
        /* we must remove class from bridge */
        new_rate = net->def->bandwidth->in->average;

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

        ret = virNetDevBandwidthUnplug(net->def->bridge,
                                       iface->data.network.actual->class_id);
        if (ret < 0)
            goto cleanup;
        /* update sum of 'floor'-s of attached NICs */
4745
        net->floor_sum -= ifaceBand->in->floor;
4746 4747 4748 4749
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
4750
        if (virNetworkSaveStatus(driverState->stateDir, net) < 0) {
4751
            net->floor_sum += ifaceBand->in->floor;
4752 4753 4754 4755
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
4756 4757 4758 4759 4760 4761 4762 4763 4764 4765
        /* 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;
    }

4766
 cleanup:
4767 4768
    return ret;
}
4769 4770 4771

static void
networkNetworkObjTaint(virNetworkObjPtr net,
4772
                       virNetworkTaintFlags taint)
4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783
{
    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));
    }
}