bridge_driver.c 169.2 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2015 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
#define VIR_FROM_THIS VIR_FROM_NETWORK
79
#define MAX_BRIDGE_ID 256
80

81 82 83 84 85 86 87
/**
 * 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)

88 89
VIR_LOG_INIT("network.bridge_driver");

90
static virNetworkDriverStatePtr network_driver;
91

92 93 94 95 96 97 98 99
static virNetworkDriverStatePtr
networkGetDriver(void)
{
    /* Maybe one day we can store @network_driver in the
     * connection object, but until then, it's just a global
     * variable which is returned. */
    return network_driver;
}
100

101
static void networkDriverLock(virNetworkDriverStatePtr driver)
102
{
103
    virMutexLock(&driver->lock);
104
}
105
static void networkDriverUnlock(virNetworkDriverStatePtr driver)
106
{
107
    virMutexUnlock(&driver->lock);
108 109
}

110 111 112 113
static dnsmasqCapsPtr
networkGetDnsmasqCaps(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr ret;
114
    networkDriverLock(driver);
115
    ret = virObjectRef(driver->dnsmasqCaps);
116
    networkDriverUnlock(driver);
117 118 119 120 121 122 123 124 125 126 127
    return ret;
}

static int
networkDnsmasqCapsRefresh(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr caps;

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

128
    networkDriverLock(driver);
129 130
    virObjectUnref(driver->dnsmasqCaps);
    driver->dnsmasqCaps = caps;
131
    networkDriverUnlock(driver);
132 133 134
    return 0;
}

135
static int networkStateCleanup(void);
136

137 138
static int networkStartNetwork(virNetworkDriverStatePtr driver,
                               virNetworkObjPtr network);
139

140 141
static int networkShutdownNetwork(virNetworkDriverStatePtr driver,
                                  virNetworkObjPtr network);
142

143 144
static int networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
                                      virNetworkObjPtr network);
145

146 147
static int networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
                                         virNetworkObjPtr network);
148

149
static int networkStartNetworkExternal(virNetworkObjPtr network);
150

151
static int networkShutdownNetworkExternal(virNetworkObjPtr network);
152

153 154
static void networkReloadFirewallRules(virNetworkDriverStatePtr driver);
static void networkRefreshDaemons(virNetworkDriverStatePtr driver);
155

156 157 158 159 160
static int networkPlugBandwidth(virNetworkObjPtr net,
                                virDomainNetDefPtr iface);
static int networkUnplugBandwidth(virNetworkObjPtr net,
                                  virDomainNetDefPtr iface);

161
static void networkNetworkObjTaint(virNetworkObjPtr net,
162
                                   virNetworkTaintFlags taint);
163

164 165 166
static virNetworkObjPtr
networkObjFromNetwork(virNetworkPtr net)
{
167
    virNetworkDriverStatePtr driver = networkGetDriver();
168 169 170
    virNetworkObjPtr network;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

171
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
172 173 174 175 176 177 178 179 180 181
    if (!network) {
        virUUIDFormat(net->uuid, uuidstr);
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching uuid '%s' (%s)"),
                       uuidstr, net->name);
    }

    return network;
}

182 183 184
static int
networkRunHook(virNetworkObjPtr network,
               virDomainDefPtr dom,
185
               virDomainNetDefPtr iface,
186 187 188 189 190 191 192 193 194
               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)) {
195 196 197 198 199 200
        if (!network) {
            VIR_DEBUG("Not running hook as @network is NULL");
            ret = 0;
            goto cleanup;
        }

201 202
        virBufferAddLit(&buf, "<hookData>\n");
        virBufferAdjustIndent(&buf, 2);
203 204
        if (iface && virDomainNetDefFormat(&buf, iface, 0) < 0)
            goto cleanup;
205 206 207 208 209 210 211 212
        if (virNetworkDefFormatBuf(&buf, network->def, 0) < 0)
            goto cleanup;
        if (dom && virDomainDefFormatInternal(dom, 0, &buf) < 0)
            goto cleanup;

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

213
        if (virBufferCheckError(&buf) < 0)
214 215
            goto cleanup;

216
        xml = virBufferContentAndReset(&buf);
217 218 219 220 221 222 223 224
        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;
225 226

        networkNetworkObjTaint(network, VIR_NETWORK_TAINT_HOOK);
227 228 229
    }

    ret = 0;
230
 cleanup:
231 232 233 234 235 236 237
    virBufferFreeAndReset(&buf);
    VIR_FREE(xml);
    VIR_FREE(net_xml);
    VIR_FREE(dom_xml);
    return ret;
}

238
static char *
239 240
networkDnsmasqLeaseFileNameDefault(virNetworkDriverStatePtr driver,
                                   const char *netname)
241 242 243
{
    char *leasefile;

244
    ignore_value(virAsprintf(&leasefile, "%s/%s.leases",
245
                             driver->dnsmasqStateDir, netname));
246 247 248
    return leasefile;
}

249
static char *
250 251
networkDnsmasqLeaseFileNameCustom(virNetworkDriverStatePtr driver,
                                  const char *bridge)
252 253 254 255
{
    char *leasefile;

    ignore_value(virAsprintf(&leasefile, "%s/%s.status",
256
                             driver->dnsmasqStateDir, bridge));
257 258 259
    return leasefile;
}

260
static char *
261 262
networkDnsmasqConfigFileName(virNetworkDriverStatePtr driver,
                             const char *netname)
263 264 265
{
    char *conffile;

266
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
267
                             driver->dnsmasqStateDir, netname));
268 269 270
    return conffile;
}

271 272 273 274 275 276
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

277
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
278 279 280 281
    return pidfilebase;
}

static char *
282 283
networkRadvdConfigFileName(virNetworkDriverStatePtr driver,
                           const char *netname)
284 285 286
{
    char *configfile;

287
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
288
                             driver->radvdStateDir, netname));
289 290
    return configfile;
}
291

292 293
/* do needed cleanup steps and remove the network from the list */
static int
294 295
networkRemoveInactive(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr net)
296 297
{
    char *leasefile = NULL;
298
    char *customleasefile = NULL;
299
    char *radvdconfigfile = NULL;
300
    char *configfile = NULL;
301
    char *radvdpidbase = NULL;
302
    char *statusfile = NULL;
303 304 305 306 307 308
    dnsmasqContext *dctx = NULL;
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(net);

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
309
    if (!(dctx = dnsmasqContextNew(def->name,
310
                                   driver->dnsmasqStateDir))) {
311
        goto cleanup;
312
    }
313

314
    if (!(leasefile = networkDnsmasqLeaseFileNameDefault(driver, def->name)))
315 316
        goto cleanup;

317
    if (!(customleasefile = networkDnsmasqLeaseFileNameCustom(driver, def->bridge)))
318 319
        goto cleanup;

320
    if (!(radvdconfigfile = networkRadvdConfigFileName(driver, def->name)))
321
        goto cleanup;
322 323

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
324
        goto cleanup;
325

326
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
327
        goto cleanup;
328

329
    if (!(statusfile = virNetworkConfigFile(driver->stateDir, def->name)))
330
        goto cleanup;
331

332 333 334
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
335
    unlink(customleasefile);
336
    unlink(configfile);
337 338 339

    /* radvd */
    unlink(radvdconfigfile);
340
    virPidFileDelete(driver->pidDir, radvdpidbase);
341

342 343 344
    /* remove status file */
    unlink(statusfile);

345
    /* remove the network definition */
346
    virNetworkRemoveInactive(driver->networks, net);
347 348 349

    ret = 0;

350
 cleanup:
351
    VIR_FREE(leasefile);
352
    VIR_FREE(configfile);
353
    VIR_FREE(customleasefile);
354 355
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
356
    VIR_FREE(statusfile);
357 358 359 360
    dnsmasqContextFree(dctx);
    return ret;
}

361 362 363
static char *
networkBridgeDummyNicName(const char *brname)
{
364
    static const char dummyNicSuffix[] = "-nic";
365 366
    char *nicname;

367 368 369 370 371 372 373
    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.
         */
374 375 376 377 378
        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));
379
    } else {
380
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
381
    }
382 383 384
    return nicname;
}

385 386
static int
networkUpdateState(virNetworkObjPtr obj,
387
                   void *opaque)
388
{
389
    virNetworkDriverStatePtr driver = opaque;
390
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
391
    int ret = -1;
392

393
    virObjectLock(obj);
394
    if (!virNetworkObjIsActive(obj)) {
395 396
        ret = 0;
        goto cleanup;
397
    }
398

399 400 401 402 403 404 405 406
    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;
407

408 409
    case VIR_NETWORK_FORWARD_BRIDGE:
        if (!(obj->def->bridge && virNetDevExists(obj->def->bridge) == 1)) {
410 411
                obj->active = 0;
            break;
412
        }
413 414 415 416 417 418 419 420 421
        /* 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;
422

423 424 425
    case VIR_NETWORK_FORWARD_HOSTDEV:
        /* so far no extra checks */
        break;
426
    }
427

428 429 430
    /* Try and read dnsmasq/radvd pids of active networks */
    if (obj->active && obj->def->ips && (obj->def->nips > 0)) {
        char *radvdpidbase;
431

432 433 434
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           obj->def->name,
                                           &obj->dnsmasqPid,
435
                                           dnsmasqCapsGetBinaryPath(dnsmasq_caps)));
436 437 438
        radvdpidbase = networkRadvdPidfileBasename(obj->def->name);
        if (!radvdpidbase)
            goto cleanup;
439

440 441 442 443
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           radvdpidbase,
                                           &obj->radvdPid, RADVD));
        VIR_FREE(radvdpidbase);
444
    }
445

446 447
    ret = 0;
 cleanup:
448
    virObjectUnlock(obj);
449
    virObjectUnref(dnsmasq_caps);
450 451
    return ret;
}
452

453

454 455
static int
networkAutostartConfig(virNetworkObjPtr net,
456
                       void *opaque)
457
{
458
    virNetworkDriverStatePtr driver = opaque;
459
    int ret = -1;
460

461
    virObjectLock(net);
462 463
    if (net->autostart &&
        !virNetworkObjIsActive(net) &&
464
        networkStartNetwork(driver, net) < 0)
465 466 467 468
        goto cleanup;

    ret = 0;
 cleanup:
469
    virObjectUnlock(net);
470
    return ret;
471 472
}

473 474 475
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
476
                             DBusMessage *message, void *user_data)
477
{
478 479
    virNetworkDriverStatePtr driver = user_data;

480 481 482 483 484 485
    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.");
486
        networkReloadFirewallRules(driver);
487 488 489 490 491 492
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

493
static int
494
networkMigrateStateFiles(virNetworkDriverStatePtr driver)
495 496 497 498 499 500 501 502 503 504 505 506 507
{
    /* 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;
508
    int direrr;
509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527
    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;
    }

528
    while ((direrr = virDirRead(dir, &entry, oldStateDir)) > 0) {
529 530 531
        if (entry->d_type != DT_UNKNOWN &&
            entry->d_type != DT_REG)
            continue;
532

533
        if (STREQ(entry->d_name, ".") ||
534 535 536 537 538
            STREQ(entry->d_name, ".."))
            continue;

        if (virAsprintf(&oldPath, "%s/%s",
                        oldStateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
539
            goto cleanup;
540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556

        if (entry->d_type == DT_UNKNOWN) {
            struct stat st;

            if (lstat(oldPath, &st) < 0) {
                virReportSystemError(errno,
                                     _("failed to stat network status file '%s'"),
                                     oldPath);
                goto cleanup;
            }

            if (!S_ISREG(st.st_mode)) {
                VIR_FREE(oldPath);
                continue;
            }
        }

557
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
J
Ján Tomko 已提交
558
            goto cleanup;
559 560 561

        if (virAsprintf(&newPath, "%s/%s",
                        driver->stateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
562
            goto cleanup;
563 564 565 566 567 568 569 570 571 572 573 574
        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);
    }
575
    if (direrr < 0)
J
Ján Tomko 已提交
576
        goto cleanup;
577 578 579 580 581 582 583 584 585 586

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

587
/**
588
 * networkStateInitialize:
589 590 591 592
 *
 * Initialization function for the QEmu daemon
 */
static int
593 594 595
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
596
{
597 598 599
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
600 601 602
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
603

604
    if (VIR_ALLOC(network_driver) < 0)
605
        goto error;
606

607 608
    if (virMutexInit(&network_driver->lock) < 0) {
        VIR_FREE(network_driver);
609 610
        goto error;
    }
611

612 613 614 615
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
616
    if (privileged) {
617
        if (VIR_STRDUP(network_driver->networkConfigDir,
618
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
619
            VIR_STRDUP(network_driver->networkAutostartDir,
620
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
621
            VIR_STRDUP(network_driver->stateDir,
622
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
623
            VIR_STRDUP(network_driver->pidDir,
624
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
625
            VIR_STRDUP(network_driver->dnsmasqStateDir,
626
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
627
            VIR_STRDUP(network_driver->radvdStateDir,
628 629
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
630 631 632 633 634

        /* migration from old to new location is only applicable for
         * privileged mode - unprivileged mode directories haven't
         * changed location.
         */
635
        if (networkMigrateStateFiles(network_driver) < 0)
636
            goto error;
637
    } else {
638 639 640
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
641
            goto error;
642

643
        if ((virAsprintf(&network_driver->networkConfigDir,
644
                         "%s/qemu/networks", configdir) < 0) ||
645
            (virAsprintf(&network_driver->networkAutostartDir,
646
                         "%s/qemu/networks/autostart", configdir) < 0) ||
647
            (virAsprintf(&network_driver->stateDir,
648
                         "%s/network/lib", rundir) < 0) ||
649
            (virAsprintf(&network_driver->pidDir,
650
                         "%s/network/run", rundir) < 0) ||
651
            (virAsprintf(&network_driver->dnsmasqStateDir,
652
                         "%s/dnsmasq/lib", rundir) < 0) ||
653
            (virAsprintf(&network_driver->radvdStateDir,
654
                         "%s/radvd/lib", rundir) < 0)) {
655
            goto error;
656
        }
657 658
    }

659
    if (virFileMakePath(network_driver->stateDir) < 0) {
660 661
        virReportSystemError(errno,
                             _("cannot create directory %s"),
662
                             network_driver->stateDir);
663 664 665
        goto error;
    }

666
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
667
    network_driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
668

669
    if (!(network_driver->networks = virNetworkObjListNew()))
670 671
        goto error;

672 673
    if (virNetworkLoadAllState(network_driver->networks,
                               network_driver->stateDir) < 0)
674 675
        goto error;

676 677 678
    if (virNetworkLoadAllConfigs(network_driver->networks,
                                 network_driver->networkConfigDir,
                                 network_driver->networkAutostartDir) < 0)
679 680
        goto error;

681 682 683 684
    /* 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). */
685
    virNetworkObjListForEach(network_driver->networks,
686
                             networkUpdateState,
687 688
                             network_driver);
    virNetworkObjListPrune(network_driver->networks,
689 690
                           VIR_CONNECT_LIST_NETWORKS_INACTIVE |
                           VIR_CONNECT_LIST_NETWORKS_TRANSIENT);
691 692
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
693

694
    network_driver->networkEventState = virObjectEventStateNew();
695

696 697 698 699
#ifdef HAVE_FIREWALLD
    if (!(sysbus = virDBusGetSystemBus())) {
        virErrorPtr err = virGetLastError();
        VIR_WARN("DBus not available, disabling firewalld support "
700
                 "in bridge_network_driver: %s", err->message);
701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
    } 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,
718
                                   network_driver, NULL);
719 720 721
    }
#endif

722
    ret = 0;
723
 cleanup:
724 725 726
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
727

728
 error:
729 730
    if (network_driver)
        networkDriverUnlock(network_driver);
731
    networkStateCleanup();
732
    goto cleanup;
733 734
}

735 736 737 738 739 740 741 742
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
743
    if (!network_driver)
744 745
        return;

746
    virNetworkObjListForEach(network_driver->networks,
747
                             networkAutostartConfig,
748
                             network_driver);
749 750
}

751
/**
752
 * networkStateReload:
753 754 755 756 757
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
758 759
networkStateReload(void)
{
760
    if (!network_driver)
761 762
        return 0;

763 764 765 766 767 768 769 770
    virNetworkLoadAllState(network_driver->networks,
                           network_driver->stateDir);
    virNetworkLoadAllConfigs(network_driver->networks,
                             network_driver->networkConfigDir,
                             network_driver->networkAutostartDir);
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
    virNetworkObjListForEach(network_driver->networks,
771 772
                             networkAutostartConfig,
                             NULL);
773 774 775 776 777
    return 0;
}


/**
778
 * networkStateCleanup:
779 780 781 782
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
783 784
networkStateCleanup(void)
{
785
    if (!network_driver)
786 787
        return -1;

788
    virObjectEventStateFree(network_driver->networkEventState);
789

790
    /* free inactive networks */
791
    virObjectUnref(network_driver->networks);
792

793 794 795 796 797 798
    VIR_FREE(network_driver->networkConfigDir);
    VIR_FREE(network_driver->networkAutostartDir);
    VIR_FREE(network_driver->stateDir);
    VIR_FREE(network_driver->pidDir);
    VIR_FREE(network_driver->dnsmasqStateDir);
    VIR_FREE(network_driver->radvdStateDir);
799

800
    virObjectUnref(network_driver->dnsmasqCaps);
801

802
    virMutexDestroy(&network_driver->lock);
803

804
    VIR_FREE(network_driver);
805 806 807 808 809

    return 0;
}


810 811 812 813 814 815 816
/* 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)
{
817 818
    size_t i;
    int ret = -1;
819 820 821 822 823 824 825
    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.
     */
826
    for (i = 0; i < 25; i++) {
827
        int signum = 0;
828
        if (i == 0) {
829
            signum = SIGTERM;
830
        } else if (i == 15) {
831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865
            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);
866
 cleanup:
867 868 869
    return ret;
}

J
Ján Tomko 已提交
870 871 872
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
G
Gene Czarcinski 已提交
873

874
static int
G
Gene Czarcinski 已提交
875 876
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
                                 virNetworkIpDefPtr ipdef)
877
{
878
    size_t i;
G
Gene Czarcinski 已提交
879
    bool ipv6 = false;
880

G
Gene Czarcinski 已提交
881 882
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
883 884
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
885
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
886 887
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
888
                return -1;
889
    }
890

G
Gene Czarcinski 已提交
891 892 893 894 895 896 897
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
898
    size_t i, j;
G
Gene Czarcinski 已提交
899

900 901
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
902
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
903
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
904
                for (j = 0; j < host->nnames; j++)
905 906
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
907 908
            }
        }
909 910
    }

911
    return 0;
912 913 914
}


915 916
int
networkDnsmasqConfContents(virNetworkObjPtr network,
917 918 919 920
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
921
{
922
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
923
    int r, ret = -1;
924
    int nbleases = 0;
925
    size_t i;
926
    virNetworkDNSDefPtr dns = &network->def->dns;
G
Gene Czarcinski 已提交
927 928
    virNetworkIpDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
    bool ipv6SLAAC;
929
    char *saddr = NULL, *eaddr = NULL;
930

931 932
    *configstr = NULL;

933
    /*
934 935 936
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
937
     *
938 939
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
940
     */
941 942 943 944 945 946

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

947 948
    /* create dnsmasq config file appropriate for this network */
    virBufferAsprintf(&configbuf,
949 950 951 952 953 954 955
                      "##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"
956
                      "strict-order\n",
957 958
                      network->def->name);

959 960 961 962
    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
963
                              network->def->dns.forwarders[i]);
964 965 966
        }
    }

967
    if (network->def->domain) {
968 969 970 971 972
        if (network->def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
                              network->def->domain);
        }
973
        virBufferAsprintf(&configbuf,
974 975 976 977
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
978

J
Ján Tomko 已提交
979
    if (network->def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
980 981 982 983
        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)
984
         */
985
        virBufferAddLit(&configbuf, "local=//\n");
986
    }
987

988
    if (pidfile)
989
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
990

991 992 993
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

994 995 996 997 998 999 1000 1001
    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.
         */
1002
        virBufferAsprintf(&configbuf,
1003 1004 1005
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
1006
    } else {
1007
        virBufferAddLit(&configbuf, "bind-interfaces\n");
1008 1009 1010 1011 1012 1013 1014 1015
        /*
         * --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
         */
1016 1017 1018
        for (i = 0;
             (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
             i++) {
1019 1020 1021 1022
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
1023

1024
            /* also part of CVE 2012-3411 - if the host's version of
1025
             * dnsmasq doesn't have bind-dynamic, only allow listening on
1026 1027
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
1028 1029
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
1030 1031 1032 1033
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
1034
                                 "The version of dnsmasq on this host (%d.%d) "
1035 1036 1037 1038
                                 "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 "
1039 1040 1041 1042 1043 1044
                                 "(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);
1045
                VIR_FREE(ipaddr);
1046 1047
                goto cleanup;
            }
1048
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1049 1050 1051
            VIR_FREE(ipaddr);
        }
    }
1052

1053 1054
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1055
     * guaranteed to not work, and set no-resolv so that no dns
1056 1057 1058
     * 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).
1059
     */
1060
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1061
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1062
                        "no-resolv\n");
1063
    }
1064

1065
    for (i = 0; i < dns->ntxts; i++) {
1066
        virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
1067 1068
                          dns->txts[i].name,
                          dns->txts[i].value);
1069
    }
1070

1071
    for (i = 0; i < dns->nsrvs; i++) {
1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093
        /* 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);
1094

1095 1096 1097
        /* domain is optional - it defaults to the domain of this network */
        if (dns->srvs[i].domain)
            virBufferAsprintf(&configbuf, ".%s", dns->srvs[i].domain);
1098

1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120
        /* 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);
1121
        }
1122
        virBufferAddLit(&configbuf, "\n");
1123 1124
    }

G
Gene Czarcinski 已提交
1125
    /* Find the first dhcp for both IPv4 and IPv6 */
1126 1127 1128
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
1129 1130 1131 1132
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1133 1134
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145
                    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,
1146 1147 1148 1149 1150 1151 1152 1153 1154
                                   _("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
1155 1156 1157 1158
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1159 1160
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173
                    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 "
1174 1175 1176 1177
                 "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
1178 1179 1180 1181 1182
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1183
        for (r = 0; r < ipdef->nranges; r++) {
1184 1185
            int thisRange;

1186 1187
            if (!(saddr = virSocketAddrFormat(&ipdef->ranges[r].start)) ||
                !(eaddr = virSocketAddrFormat(&ipdef->ranges[r].end)))
1188
                goto cleanup;
1189

1190
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s\n",
1191
                              saddr, eaddr);
1192
            VIR_FREE(saddr);
1193
            VIR_FREE(eaddr);
1194
            thisRange = virSocketAddrGetRange(&ipdef->ranges[r].start,
1195 1196 1197
                                              &ipdef->ranges[r].end,
                                              &ipdef->address,
                                              virNetworkIpDefPrefix(ipdef));
1198 1199 1200
            if (thisRange < 0)
                goto cleanup;
            nbleases += thisRange;
1201
        }
1202

1203
        /*
1204 1205 1206 1207
         * 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)
1208 1209
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1210
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1211 1212
            if (!bridgeaddr)
                goto cleanup;
1213
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static\n", bridgeaddr);
1214 1215
            VIR_FREE(bridgeaddr);
        }
1216

G
Gene Czarcinski 已提交
1217 1218
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1219

G
Gene Czarcinski 已提交
1220 1221 1222
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts)
1223
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1224

G
Gene Czarcinski 已提交
1225
            if (ipdef->tftproot) {
1226 1227
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1228
            }
1229

G
Gene Czarcinski 已提交
1230 1231 1232
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1233

1234
                    if (!bootserver)
G
Gene Czarcinski 已提交
1235
                        goto cleanup;
1236
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1237
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1238 1239
                    VIR_FREE(bootserver);
                } else {
1240
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1241 1242 1243 1244 1245
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1246

1247
    if (nbleases > 0)
1248
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1249

G
Gene Czarcinski 已提交
1250 1251
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1252
        goto cleanup;
G
Gene Czarcinski 已提交
1253 1254 1255 1256 1257 1258

    /* 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)
1259 1260
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1261

1262 1263
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1264
     */
1265
    virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
1266
                      dctx->addnhostsfile->path);
G
Gene Czarcinski 已提交
1267 1268 1269

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1270
        if (ipv6def) {
1271
            virBufferAddLit(&configbuf, "enable-ra\n");
1272
        } else {
1273 1274 1275
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
                 i++) {
G
Gene Czarcinski 已提交
1276 1277 1278 1279
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1280 1281
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1282 1283
                    VIR_FREE(bridgeaddr);
                }
1284
            }
1285
        }
1286 1287
    }

1288 1289 1290
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1291
    ret = 0;
G
Gene Czarcinski 已提交
1292

1293
 cleanup:
1294 1295
    VIR_FREE(saddr);
    VIR_FREE(eaddr);
1296
    virBufferFreeAndReset(&configbuf);
1297
    return ret;
1298 1299
}

1300
/* build the dnsmasq command line */
1301 1302 1303
static int ATTRIBUTE_NONNULL(3)
networkBuildDhcpDaemonCommandLine(virNetworkDriverStatePtr driver,
                                  virNetworkObjPtr network,
1304
                                  virCommandPtr *cmdout,
1305 1306
                                  char *pidfile,
                                  dnsmasqContext *dctx)
1307
{
1308
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1309
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1310
    int ret = -1;
1311 1312
    char *configfile = NULL;
    char *configstr = NULL;
1313
    char *leaseshelper_path = NULL;
1314 1315

    network->dnsmasqPid = -1;
1316

1317
    if (networkDnsmasqConfContents(network, pidfile, &configstr,
1318
                                   dctx, dnsmasq_caps) < 0)
1319 1320 1321 1322 1323
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1324
    if (!(configfile = networkDnsmasqConfigFileName(driver, network->def->name)))
1325 1326 1327 1328 1329
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1330 1331
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1332 1333 1334
        goto cleanup;
    }

1335 1336
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1337
                                                  abs_topbuilddir "/src",
1338 1339 1340
                                                  LIBEXECDIR)))
        goto cleanup;

1341
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(dnsmasq_caps));
1342
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1343 1344
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1345
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1346
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", network->def->bridge);
1347

1348
    *cmdout = cmd;
1349
    ret = 0;
1350
 cleanup:
1351
    virObjectUnref(dnsmasq_caps);
1352 1353
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1354
    VIR_FREE(leaseshelper_path);
1355 1356 1357 1358
    return ret;
}

static int
1359 1360
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
1361 1362 1363 1364
{
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1365
    dnsmasqContext *dctx = NULL;
1366

1367
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0)) {
G
Gene Czarcinski 已提交
1368
        /* no IP addresses, so we don't need to run */
1369 1370 1371 1372
        ret = 0;
        goto cleanup;
    }

1373
    if (virFileMakePath(driver->pidDir) < 0) {
1374
        virReportSystemError(errno,
1375
                             _("cannot create directory %s"),
1376
                             driver->pidDir);
1377
        goto cleanup;
1378 1379
    }

1380
    if (!(pidfile = virPidFileBuildPath(driver->pidDir,
1381
                                        network->def->name)))
1382
        goto cleanup;
1383

1384
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1385
        virReportSystemError(errno,
1386
                             _("cannot create directory %s"),
1387
                             driver->dnsmasqStateDir);
1388 1389 1390
        goto cleanup;
    }

1391
    dctx = dnsmasqContextNew(network->def->name, driver->dnsmasqStateDir);
1392 1393 1394
    if (dctx == NULL)
        goto cleanup;

1395
    if (networkDnsmasqCapsRefresh(driver) < 0)
1396
        goto cleanup;
1397

1398
    ret = networkBuildDhcpDaemonCommandLine(driver, network, &cmd, pidfile, dctx);
1399 1400 1401 1402 1403
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1404
        goto cleanup;
1405

G
Guido Günther 已提交
1406
    ret = virCommandRun(cmd, NULL);
1407
    if (ret < 0)
1408 1409 1410
        goto cleanup;

    /*
1411 1412 1413 1414 1415
     * 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
1416 1417
     */

1418
    ret = virPidFileRead(driver->pidDir, network->def->name,
1419 1420
                         &network->dnsmasqPid);
    if (ret < 0)
1421
        goto cleanup;
1422

1423
    ret = 0;
1424
 cleanup:
1425
    VIR_FREE(pidfile);
1426
    virCommandFree(cmd);
1427
    dnsmasqContextFree(dctx);
1428 1429 1430
    return ret;
}

1431 1432
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1433 1434
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1435 1436 1437
 *
 *  Returns 0 on success, -1 on failure.
 */
1438
static int
1439 1440
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1441
{
1442 1443
    int ret = -1;
    size_t i;
G
Gene Czarcinski 已提交
1444
    virNetworkIpDefPtr ipdef, ipv4def, ipv6def;
1445
    dnsmasqContext *dctx = NULL;
1446

G
Gene Czarcinski 已提交
1447
    /* if no IP addresses specified, nothing to do */
1448
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1449 1450
        return 0;

1451 1452
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1453
        return networkStartDhcpDaemon(driver, network);
1454

G
Gene Czarcinski 已提交
1455
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1456
    if (!(dctx = dnsmasqContextNew(network->def->name,
1457
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1458
        goto cleanup;
1459
    }
G
Gene Czarcinski 已提交
1460 1461 1462 1463 1464 1465

    /* 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;
1466 1467 1468
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
G
Gene Czarcinski 已提交
1469 1470
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1471 1472
    }

G
Gene Czarcinski 已提交
1473
    ipv6def = NULL;
1474 1475 1476
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1477 1478
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1479 1480
    }

G
Gene Czarcinski 已提交
1481
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1482
        goto cleanup;
G
Gene Czarcinski 已提交
1483 1484

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1485
        goto cleanup;
1486

G
Gene Czarcinski 已提交
1487
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
J
Ján Tomko 已提交
1488
        goto cleanup;
1489 1490

    if ((ret = dnsmasqSave(dctx)) < 0)
1491
        goto cleanup;
1492 1493

    ret = kill(network->dnsmasqPid, SIGHUP);
1494
 cleanup:
1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506
    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
1507 1508
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1509 1510 1511 1512 1513 1514
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1515
    }
1516
    /* now start dnsmasq if it should be started */
1517
    return networkStartDhcpDaemon(driver, network);
1518 1519
}

G
Gene Czarcinski 已提交
1520 1521 1522 1523 1524 1525
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";

1526 1527 1528
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1529
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1530 1531
    int ret = -1;
    size_t i;
1532
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
1533
    bool v6present = false, dhcp6 = false;
1534 1535

    *configstr = NULL;
1536

G
Gene Czarcinski 已提交
1537
    /* Check if DHCPv6 is needed */
1538 1539 1540
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553
        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;
    }

1554 1555 1556
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1557
    virBufferAsprintf(&configbuf, "interface %s\n"
1558 1559
                      "{\n"
                      "  AdvSendAdvert on;\n"
1560
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1561 1562 1563 1564 1565
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1566 1567

    /* add a section for each IPv6 address in the config */
1568 1569 1570
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
1571 1572 1573 1574 1575
        int prefix;
        char *netaddr;

        prefix = virNetworkIpDefPrefix(ipdef);
        if (prefix < 0) {
1576 1577 1578
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1579 1580
            goto cleanup;
        }
1581
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1582
            goto cleanup;
1583
        virBufferAsprintf(&configbuf,
1584
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1585 1586 1587
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1588 1589 1590
        VIR_FREE(netaddr);
    }

1591
    virBufferAddLit(&configbuf, "};\n");
1592

1593
    if (virBufferCheckError(&configbuf) < 0)
1594
        goto cleanup;
1595

1596 1597
    *configstr = virBufferContentAndReset(&configbuf);

1598
    ret = 0;
1599
 cleanup:
1600 1601 1602 1603
    virBufferFreeAndReset(&configbuf);
    return ret;
}

1604
/* write file and return its name (which must be freed by caller) */
1605
static int
1606 1607 1608
networkRadvdConfWrite(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr network,
                      char **configFile)
1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623
{
    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;
1624 1625 1626 1627
        goto cleanup;
    }

    /* construct the filename */
1628
    if (!(*configFile = networkRadvdConfigFileName(driver, network->def->name)))
1629 1630
        goto cleanup;
    /* write the file */
1631
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1632 1633
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1634 1635 1636 1637 1638
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1639
 cleanup:
1640 1641 1642 1643 1644 1645
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1646 1647
networkStartRadvd(virNetworkDriverStatePtr driver,
                  virNetworkObjPtr network)
1648
{
1649
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1650 1651 1652 1653 1654 1655 1656 1657
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1658
    /* Is dnsmasq handling RA? */
1659
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
G
Gene Czarcinski 已提交
1660 1661 1662 1663
        ret = 0;
        goto cleanup;
    }

1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674
    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);
1675 1676 1677
        goto cleanup;
    }

1678
    if (virFileMakePath(driver->pidDir) < 0) {
1679 1680
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1681
                             driver->pidDir);
1682 1683
        goto cleanup;
    }
1684
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1685 1686
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1687
                             driver->radvdStateDir);
1688 1689 1690 1691
        goto cleanup;
    }

    /* construct pidfile name */
1692
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1693
        goto cleanup;
1694
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1695 1696
        goto cleanup;

1697
    if (networkRadvdConfWrite(driver, network, &configfile) < 0)
1698 1699
        goto cleanup;

1700 1701 1702 1703
    /* 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
1704
     * virPidFileRead() below will fail if we use them).
1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719
     * 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;

1720
    if (virPidFileRead(driver->pidDir, radvdpidbase, &network->radvdPid) < 0)
1721 1722 1723
        goto cleanup;

    ret = 0;
1724
 cleanup:
1725
    virObjectUnref(dnsmasq_caps);
1726 1727 1728 1729 1730 1731 1732
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1733
static int
1734 1735
networkRefreshRadvd(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
1736
{
1737
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
G
Gene Czarcinski 已提交
1738 1739 1740
    char *radvdpidbase;

    /* Is dnsmasq handling RA? */
1741 1742
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
        virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1743 1744 1745 1746 1747 1748 1749
        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)) {
1750
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1751 1752 1753 1754 1755
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }
1756
    virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1757

1758 1759
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
1760
        return networkStartRadvd(driver, network);
1761 1762 1763 1764 1765 1766

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

1767
    if (networkRadvdConfWrite(driver, network, NULL) < 0)
1768 1769 1770 1771 1772
        return -1;

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

1773 1774
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1775
static int
1776
networkRestartRadvd(virNetworkObjPtr network)
1777 1778 1779 1780 1781 1782 1783 1784 1785
{
    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
1786
        if ((networkKillDaemon(network->radvdPid, "radvd",
1787 1788 1789
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1790
            virPidFileDelete(driver->pidDir, radvdpidbase);
1791 1792 1793 1794 1795 1796 1797 1798 1799
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1800 1801
static int
networkRefreshDaemonsHelper(virNetworkObjPtr net,
1802
                            void *opaque)
1803
{
1804
    virNetworkDriverStatePtr driver = opaque;
1805

1806
    virObjectLock(net);
1807 1808 1809 1810 1811 1812 1813 1814 1815 1816
    if (virNetworkObjIsActive(net) &&
        ((net->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
        /* Only the three L3 network types that are configured by
         * libvirt will have a dnsmasq or radvd daemon associated
         * with them.  Here we send a SIGHUP to an existing
         * dnsmasq and/or radvd, or restart them if they've
         * disappeared.
         */
1817 1818
        networkRefreshDhcpDaemon(driver, net);
        networkRefreshRadvd(driver, net);
1819
    }
1820
    virObjectUnlock(net);
1821 1822 1823
    return 0;
}

1824 1825 1826 1827
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
1828
networkRefreshDaemons(virNetworkDriverStatePtr driver)
1829 1830
{
    VIR_INFO("Refreshing network daemons");
1831 1832
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
1833
                             driver);
1834
}
1835

1836 1837 1838 1839 1840
static int
networkReloadFirewallRulesHelper(virNetworkObjPtr net,
                                 void *opaque ATTRIBUTE_UNUSED)
{

1841
    virObjectLock(net);
1842 1843 1844 1845 1846 1847 1848 1849 1850 1851
    if (virNetworkObjIsActive(net) &&
        ((net->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
        /* Only the three L3 network types that are configured by libvirt
         * need to have iptables rules reloaded.
         */
        networkRemoveFirewallRules(net->def);
        if (networkAddFirewallRules(net->def) < 0) {
            /* failed to add but already logged */
1852 1853
        }
    }
1854
    virObjectUnlock(net);
1855
    return 0;
1856 1857
}

1858
static void
1859
networkReloadFirewallRules(virNetworkDriverStatePtr driver)
1860
{
1861
    VIR_INFO("Reloading iptables rules");
1862 1863 1864
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
1865 1866
}

1867
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
1868
static int
1869
networkEnableIpForwarding(bool enableIPv4, bool enableIPv6)
1870
{
1871
    int ret = 0;
1872 1873 1874 1875
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
1876
                           &enabled, sizeof(enabled));
1877 1878
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
1879
                           &enabled, sizeof(enabled));
1880
#else
1881 1882 1883 1884
    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);
1885
#endif
1886
    return ret;
1887 1888
}

1889 1890
#define SYSCTL_PATH "/proc/sys"

1891 1892
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
1893 1894 1895
{
    char *field = NULL;
    int ret = -1;
1896
    bool enableIPv6 =  !!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0);
1897

1898 1899 1900 1901 1902 1903 1904
    /* 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;
1905

1906 1907
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
1908 1909
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
1910 1911 1912
        ret = 0;
        goto cleanup;
    }
1913

1914 1915 1916 1917 1918
    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;
1919
    }
1920
    VIR_FREE(field);
1921

1922 1923
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
1924 1925 1926 1927 1928 1929
     */

    /* 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",
1930
                    network->def->bridge) < 0)
1931 1932
        goto cleanup;

1933
    if (virFileWriteStr(field, "0", 0) < 0) {
1934
        virReportSystemError(errno,
1935 1936 1937 1938 1939
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

1940 1941 1942 1943
    /* 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",
1944
                    network->def->bridge) < 0)
1945 1946
        goto cleanup;

1947
    if (virFileWriteStr(field, "0", 0) < 0) {
1948
        virReportSystemError(errno,
1949
                             _("cannot disable %s"), field);
1950 1951 1952 1953
        goto cleanup;
    }

    ret = 0;
1954
 cleanup:
1955 1956 1957 1958
    VIR_FREE(field);
    return ret;
}

1959
/* add an IP address to a bridge */
1960
static int
D
Daniel P. Berrange 已提交
1961
networkAddAddrToBridge(virNetworkObjPtr network,
1962
                       virNetworkIpDefPtr ipdef)
1963
{
1964 1965 1966
    int prefix = virNetworkIpDefPrefix(ipdef);

    if (prefix < 0) {
1967 1968 1969
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
1970 1971 1972
        return -1;
    }

1973 1974
    if (virNetDevSetIPAddress(network->def->bridge,
                              &ipdef->address, prefix) < 0)
1975 1976 1977 1978 1979
        return -1;

    return 0;
}

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

static int
networkStartHandleMACTableManagerMode(virNetworkObjPtr network,
                                      const char *macTapIfName)
{
    const char *brname = network->def->bridge;

    if (brname &&
        network->def->macTableManager
        == VIR_NETWORK_BRIDGE_MAC_TABLE_MANAGER_LIBVIRT) {
        if (virNetDevBridgeSetVlanFiltering(brname, true) < 0)
            return -1;
        if (macTapIfName) {
            if (virNetDevBridgePortSetLearning(brname, macTapIfName, false) < 0)
                return -1;
            if (virNetDevBridgePortSetUnicastFlood(brname, macTapIfName, false) < 0)
                return -1;
        }
    }
    return 0;
}


2003 2004 2005 2006 2007
/* add an IP (static) route to a bridge */
static int
networkAddRouteToBridge(virNetworkObjPtr network,
                        virNetworkRouteDefPtr routedef)
{
2008 2009 2010 2011
    int prefix = virNetworkRouteDefGetPrefix(routedef);
    unsigned int metric = virNetworkRouteDefGetMetric(routedef);
    virSocketAddrPtr addr = virNetworkRouteDefGetAddress(routedef);
    virSocketAddrPtr gateway = virNetworkRouteDefGetGateway(routedef);
2012 2013 2014 2015 2016 2017 2018 2019 2020

    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;
    }

2021 2022
    if (virNetDevAddRoute(network->def->bridge, addr,
                          prefix, gateway, metric) < 0) {
2023 2024 2025 2026 2027
        return -1;
    }
    return 0;
}

2028
static int
2029 2030
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
                           virNetworkObjPtr network)
2031
{
2032
    size_t i;
2033
    bool v4present = false, v6present = false;
2034 2035
    virErrorPtr save_err = NULL;
    virNetworkIpDefPtr ipdef;
2036
    virNetworkRouteDefPtr routedef;
2037
    char *macTapIfName = NULL;
2038
    int tapfd = -1;
2039

2040
    /* Check to see if any network IP collides with an existing route */
2041
    if (networkCheckRouteCollision(network->def) < 0)
2042 2043
        return -1;

2044
    /* Create and configure the bridge device */
2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058
    if (!network->def->bridge) {
        /* bridge name can only be empty if the config files were
         * edited directly. Otherwise networkValidate() (called after
         * parsing the XML from networkCreateXML() and
         * networkDefine()) guarantees we will have a valid bridge
         * name before this point. Since hand editing of the config
         * files is explicitly prohibited we can, with clear
         * conscience, log an error and fail at this point.
         */
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' has no bridge name defined"),
                       network->def->name);
        return -1;
    }
2059
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
2060 2061
        return -1;

2062 2063 2064 2065 2066 2067 2068 2069
    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);
2070
        if (!macTapIfName)
2071
            goto err0;
2072
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2073
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
2074
                                           &macTapIfName, &network->def->mac,
2075
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2076 2077 2078
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2079 2080 2081 2082 2083
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2084
    /* Set bridge options */
2085 2086 2087 2088

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2089
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2090
                                   network->def->delay * 1000) < 0)
2091
        goto err1;
2092

2093
    if (virNetDevBridgeSetSTP(network->def->bridge,
2094
                              network->def->stp ? true : false) < 0)
2095
        goto err1;
2096

2097 2098 2099 2100
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2101
        goto err1;
2102

2103
    /* Add "once per network" rules */
2104
    if (networkAddFirewallRules(network->def) < 0)
2105 2106
        goto err1;

2107 2108 2109
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
2110
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2111
            v4present = true;
2112
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2113
            v6present = true;
2114

2115
        /* Add the IP address/netmask to the bridge */
2116
        if (networkAddAddrToBridge(network, ipdef) < 0)
2117
            goto err2;
2118 2119
    }

2120 2121 2122
    if (networkStartHandleMACTableManagerMode(network, macTapIfName) < 0)
        goto err2;

2123
    /* Bring up the bridge interface */
2124
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2125
        goto err2;
2126

2127
    for (i = 0; i < network->def->nroutes; i++) {
2128 2129 2130 2131 2132
        virSocketAddrPtr gateway = NULL;

        routedef = network->def->routes[i];
        gateway = virNetworkRouteDefGetGateway(routedef);

2133 2134 2135
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2136
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2137 2138 2139 2140 2141 2142 2143
            if (networkAddRouteToBridge(network, routedef) < 0) {
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2144 2145
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2146
        networkEnableIpForwarding(v4present, v6present) < 0) {
2147
        virReportSystemError(errno, "%s",
2148
                             _("failed to enable IP forwarding"));
2149
        goto err3;
2150 2151
    }

2152

2153
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2154
    if ((v4present || v6present) &&
2155
        networkStartDhcpDaemon(driver, network) < 0)
2156
        goto err3;
2157

2158
    /* start radvd if there are any ipv6 addresses */
2159
    if (v6present && networkStartRadvd(driver, network) < 0)
2160 2161
        goto err4;

2162 2163 2164 2165 2166 2167 2168 2169 2170
    /* 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);
    }

2171
    if (virNetDevBandwidthSet(network->def->bridge,
2172
                              network->def->bandwidth, true) < 0)
2173 2174
        goto err5;

2175
    VIR_FREE(macTapIfName);
2176 2177 2178

    return 0;

2179
 err5:
2180 2181
    if (network->def->bandwidth)
       virNetDevBandwidthClear(network->def->bridge);
2182

2183 2184 2185 2186
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2187 2188 2189 2190 2191
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2192 2193 2194
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2195
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2196

2197 2198 2199
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2200
    networkRemoveFirewallRules(network->def);
2201 2202

 err1:
2203 2204 2205
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2206
    if (macTapIfName) {
2207
        VIR_FORCE_CLOSE(tapfd);
2208
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2209 2210
        VIR_FREE(macTapIfName);
    }
2211 2212

 err0:
2213 2214
    if (!save_err)
        save_err = virSaveLastError();
2215
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2216

2217 2218 2219 2220
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2221
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2222 2223 2224
    return -1;
}

2225 2226 2227
static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
                              virNetworkObjPtr network)
2228
{
2229 2230
    if (network->def->bandwidth)
        virNetDevBandwidthClear(network->def->bridge);
2231

2232 2233 2234 2235 2236
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2237
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2238
            virPidFileDelete(driver->pidDir, radvdpidbase);
2239 2240 2241 2242
            VIR_FREE(radvdpidbase);
        }
    }

2243 2244 2245
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2246
    if (network->def->mac_specified) {
2247
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2248
        if (macTapIfName) {
2249
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2250 2251 2252 2253
            VIR_FREE(macTapIfName);
        }
    }

2254
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2255

2256
    networkRemoveFirewallRules(network->def);
2257

2258
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2259

2260
    /* See if its still alive and really really kill it */
2261
    if (network->dnsmasqPid > 0 &&
2262
        (kill(network->dnsmasqPid, 0) == 0))
2263 2264
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2265 2266 2267 2268 2269 2270

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

2271 2272 2273
    return 0;
}

2274

2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295
static int
networkStartNetworkBridge(virNetworkObjPtr network)
{
    /* put anything here that needs to be done each time a network of
     * type BRIDGE, is started. On failure, undo anything you've done,
     * and return -1. On success return 0.
     */
    return networkStartHandleMACTableManagerMode(network, NULL);
}

static int
networkShutdownNetworkBridge(virNetworkObjPtr network ATTRIBUTE_UNUSED)
{
    /* put anything here that needs to be done each time a network of
     * type BRIDGE is shutdown. On failure, undo anything you've done,
     * and return -1. On success return 0.
     */
    return 0;
}


2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310
/* 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;

2311 2312 2313
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401
    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;
}


2402
static int
2403
networkStartNetworkExternal(virNetworkObjPtr network)
2404 2405
{
    /* put anything here that needs to be done each time a network of
2406
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2407 2408 2409
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2410
    return networkCreateInterfacePool(network->def);
2411 2412
}

2413
static int networkShutdownNetworkExternal(virNetworkObjPtr network ATTRIBUTE_UNUSED)
2414 2415
{
    /* put anything here that needs to be done each time a network of
2416
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2417 2418 2419 2420 2421 2422 2423
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2424 2425
networkStartNetwork(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
2426
{
2427 2428 2429
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2430 2431

    if (virNetworkObjIsActive(network)) {
2432 2433
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2434
        return ret;
2435 2436
    }

2437 2438 2439
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2440
    if (virNetworkObjSetDefTransient(network, true) < 0)
2441
        goto cleanup;
2442

2443 2444
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2445
    if (networkRunHook(network, NULL, NULL,
2446 2447 2448 2449
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2450
    switch (network->def->forward.type) {
2451 2452 2453 2454

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2455
        if (networkStartNetworkVirtual(driver, network) < 0)
2456
            goto cleanup;
2457 2458 2459
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2460 2461 2462 2463
       if (networkStartNetworkBridge(network) < 0)
          goto cleanup;
       break;

2464 2465 2466
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2467
    case VIR_NETWORK_FORWARD_HOSTDEV:
2468
        if (networkStartNetworkExternal(network) < 0)
2469
            goto cleanup;
2470 2471 2472
        break;
    }

2473
    /* finally we can call the 'started' hook script if any */
2474
    if (networkRunHook(network, NULL, NULL,
2475 2476 2477 2478
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2479 2480 2481
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2482
    VIR_DEBUG("Writing network status to disk");
2483
    if (virNetworkSaveStatus(driver->stateDir, network) < 0)
2484
        goto cleanup;
2485 2486

    network->active = 1;
2487 2488
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2489

2490
 cleanup:
2491
    if (ret < 0) {
2492
        virNetworkObjUnsetDefTransient(network);
2493 2494
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2495
        networkShutdownNetwork(driver, network);
2496 2497 2498 2499 2500 2501 2502
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2503 2504 2505
static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
2506 2507 2508 2509 2510 2511 2512 2513 2514
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

2515
    stateFile = virNetworkConfigFile(driver->stateDir,
2516
                                     network->def->name);
2517 2518 2519 2520 2521 2522
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2523
    switch (network->def->forward.type) {
2524 2525 2526 2527

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2528
        ret = networkShutdownNetworkVirtual(driver, network);
2529 2530 2531
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2532 2533 2534
        ret = networkShutdownNetworkBridge(network);
        break;

2535 2536 2537
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2538
    case VIR_NETWORK_FORWARD_HOSTDEV:
2539
        ret = networkShutdownNetworkExternal(network);
2540 2541 2542
        break;
    }

2543
    /* now that we know it's stopped call the hook if present */
2544
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2545 2546
                   VIR_HOOK_SUBOP_END);

2547
    network->active = 0;
2548
    virNetworkObjUnsetDefTransient(network);
2549
    return ret;
2550 2551 2552
}


2553
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2554 2555
                                         const unsigned char *uuid)
{
2556
    virNetworkDriverStatePtr driver = networkGetDriver();
2557 2558
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2559

2560
    network = virNetworkObjFindByUUID(driver->networks, uuid);
2561
    if (!network) {
2562 2563
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2564
        virReportError(VIR_ERR_NO_NETWORK,
2565 2566
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2567
        goto cleanup;
2568 2569
    }

2570 2571 2572
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2573 2574
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2575
 cleanup:
2576
    virNetworkObjEndAPI(&network);
2577
    return ret;
2578 2579
}

2580
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2581 2582
                                         const char *name)
{
2583
    virNetworkDriverStatePtr driver = networkGetDriver();
2584 2585 2586
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2587
    network = virNetworkObjFindByName(driver->networks, name);
2588
    if (!network) {
2589 2590
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2591
        goto cleanup;
2592 2593
    }

2594 2595 2596
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2597 2598
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2599
 cleanup:
2600
    virNetworkObjEndAPI(&network);
2601
    return ret;
2602 2603
}

2604 2605
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2606
    virNetworkDriverStatePtr driver = networkGetDriver();
2607
    int nactive;
2608

2609 2610 2611
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2612 2613 2614 2615
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
2616

2617 2618 2619
    return nactive;
}

2620 2621 2622 2623 2624
static int networkConnectListNetworks(virConnectPtr conn,
                                      char **const names,
                                      int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2625
    int got = 0;
2626

2627 2628 2629
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2630 2631 2632 2633
    got = virNetworkObjListGetNames(driver->networks,
                                    true, names, nnames,
                                    virConnectListNetworksCheckACL,
                                    conn);
2634

2635 2636 2637
    return got;
}

2638 2639
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2640
    virNetworkDriverStatePtr driver = networkGetDriver();
2641
    int ninactive = 0;
2642

2643 2644 2645
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2646 2647 2648 2649
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
2650

2651 2652 2653
    return ninactive;
}

2654 2655 2656
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2657
    int got = 0;
2658

2659 2660 2661
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2662 2663 2664 2665
    got = virNetworkObjListGetNames(driver->networks,
                                    false, names, nnames,
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
2666 2667 2668
    return got;
}

2669
static int
2670 2671 2672
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2673
{
2674
    virNetworkDriverStatePtr driver = networkGetDriver();
2675 2676 2677 2678
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2679 2680 2681
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2682
    ret = virNetworkObjListExport(conn, driver->networks, nets,
2683 2684
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2685

2686
 cleanup:
2687 2688
    return ret;
}
2689

2690 2691 2692 2693 2694 2695 2696 2697
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
2698
    virNetworkDriverStatePtr driver = networkGetDriver();
2699 2700 2701 2702 2703 2704
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
2705
                                       net, eventID, callback,
2706 2707 2708
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2709
 cleanup:
2710 2711 2712 2713 2714 2715 2716
    return ret;
}

static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
2717
    virNetworkDriverStatePtr driver = networkGetDriver();
2718 2719 2720 2721 2722
    int ret = -1;

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

2723 2724 2725 2726 2727 2728
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
2729

2730
 cleanup:
2731 2732 2733
    return ret;
}

2734 2735 2736 2737 2738
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2739 2740
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2741 2742 2743 2744

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

2745 2746
    ret = virNetworkObjIsActive(obj);

2747
 cleanup:
2748
    virNetworkObjEndAPI(&obj);
2749 2750 2751 2752 2753 2754 2755 2756
    return ret;
}

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

2757 2758
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2759 2760 2761 2762

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

2763 2764
    ret = obj->persistent;

2765
 cleanup:
2766
    virNetworkObjEndAPI(&obj);
2767 2768 2769 2770
    return ret;
}


2771 2772
/*
 * networkFindUnusedBridgeName() - try to find a bridge name that is
2773 2774 2775
 * unused by the currently configured libvirt networks, as well as by
 * the host system itself (possibly created by someone/something other
 * than libvirt). Set this network's name to that new name.
2776 2777 2778 2779 2780 2781 2782 2783
 */
static int
networkFindUnusedBridgeName(virNetworkObjListPtr nets,
                            virNetworkDefPtr def)
{

    int ret = -1, id = 0;
    char *newname = NULL;
2784 2785 2786 2787 2788
    const char *templ = "virbr%d";
    const char *p;

    if (def->bridge &&
        (p = strchr(def->bridge, '%')) == strrchr(def->bridge, '%') &&
2789
        p && p[1] == 'd')
2790
        templ = def->bridge;
2791 2792 2793 2794

    do {
        if (virAsprintf(&newname, templ, id) < 0)
            goto cleanup;
2795 2796 2797 2798 2799 2800 2801
        /* check if this name is used in another libvirt network or
         * there is an existing device with that name. ignore errors
         * from virNetDevExists(), just in case it isn't implemented
         * on this platform (probably impossible).
         */
        if (!(virNetworkBridgeInUse(nets, newname, def->name) ||
              virNetDevExists(newname) == 1)) {
2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853
            VIR_FREE(def->bridge); /*could contain template */
            def->bridge = newname;
            ret = 0;
            goto cleanup;
        }
        VIR_FREE(newname);
    } while (++id <= MAX_BRIDGE_ID);

    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("Bridge generation exceeded max id %d"),
                   MAX_BRIDGE_ID);
    ret = 0;
 cleanup:
    if (ret < 0)
        VIR_FREE(newname);
    return ret;
}



/*
 * networkValidateBridgeName() - if no bridge name is set, or if the
 * bridge name contains a %d (indicating that this is a template for
 * the actual name) try to set an appropriate bridge name.  If a
 * bridge name *is* set, make sure it doesn't conflict with any other
 * network's bridge name.
 */
static int
networkBridgeNameValidate(virNetworkObjListPtr nets,
                          virNetworkDefPtr def)
{
    int ret = -1;

    if (def->bridge && !strstr(def->bridge, "%d")) {
        if (virNetworkBridgeInUse(nets, def->bridge, def->name)) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge name '%s' already in use."),
                           def->bridge);
            goto cleanup;
        }
    } else {
        /* Allocate a bridge name */
        if (networkFindUnusedBridgeName(nets, def) < 0)
            goto cleanup;
    }

    ret = 0;
 cleanup:
    return ret;
}


2854
static int
2855
networkValidate(virNetworkDriverStatePtr driver,
2856
                virNetworkDefPtr def)
2857
{
2858
    size_t i, j;
2859 2860
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2861
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
2862
    bool ipv4def = false, ipv6def = false;
2863
    bool bandwidthAllowed = true;
2864
    bool usesInterface = false, usesAddress = false;
2865 2866 2867 2868

    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
2869 2870 2871
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2872

2873 2874 2875 2876
        /* if no bridge name was given in the config, find a name
         * unused by any other libvirt networks and assign it.
         */
        if (networkBridgeNameValidate(driver->networks, def) < 0)
2877 2878 2879
            return -1;

        virNetworkSetBridgeMacAddr(def);
2880 2881
    } else {
        /* They are also the only types that currently support setting
2882 2883
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
2884
         */
2885 2886 2887 2888 2889 2890 2891 2892
        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;
        }
2893 2894 2895 2896 2897
        if (virNetworkDefGetIpByIndex(def, AF_UNSPEC, 0)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2898
                           virNetworkForwardTypeToString(def->forward.type));
2899 2900
            return -1;
        }
2901
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
2902 2903 2904 2905
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2906
                           virNetworkForwardTypeToString(def->forward.type));
2907 2908 2909 2910 2911 2912 2913
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2914
                           virNetworkForwardTypeToString(def->forward.type));
2915 2916
            return -1;
        }
2917 2918 2919 2920 2921 2922 2923 2924
        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;
        }
2925
        bandwidthAllowed = false;
2926 2927
    }

2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961
    /* we support configs with a single PF defined:
     *   <pf dev='eth0'/>
     * or with a list of netdev names:
     *   <interface dev='eth9'/>
     * OR a list of PCI addresses
     *   <address type='pci' domain='0' bus='4' slot='0' function='1'/>
     * but not any combination of those.
     *
     * Since <interface> and <address> are for some strange reason
     * stored in the same array, we need to cycle through it and check
     * the type of each.
     */
    for (i = 0; i < def->forward.nifs; i++) {
        switch ((virNetworkForwardHostdevDeviceType)
                def->forward.ifs[i].type) {
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV:
            usesInterface = true;
            break;
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI:
            usesAddress = true;
            break;
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NONE:
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_LAST:
            break;
        }
    }
    if ((def->forward.npfs > 0) + usesInterface + usesAddress > 1) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<address>, <interface>, and <pf> elements of "
                         "<forward> in network %s are mutually exclusive"),
                       def->name);
        return -1;
    }

G
Gene Czarcinski 已提交
2962 2963 2964
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
2965 2966 2967
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
2968 2969 2970 2971 2972
        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 -- "
2973 2974
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991
                    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;
                }
2992 2993 2994
            }
        }
    }
2995 2996 2997 2998 2999 3000

    /* 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.
     */

3001
    vlanAllowed = ((def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
J
Ján Tomko 已提交
3002 3003
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
3004 3005
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) ||
                   def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV);
3006 3007

    vlanUsed = def->vlan.nTags > 0;
3008 3009
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3010 3011 3012 3013 3014
            /* 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.
             */
3015
            if (def->portGroups[i].virtPortProfile) {
3016
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3017
                    def->portGroups[i].virtPortProfile->virtPortType
3018 3019 3020 3021 3022 3023 3024
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3025
        }
3026
        if (def->portGroups[i].isDefault) {
3027 3028 3029 3030 3031
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3032
                               def->name, defaultPortGroup->name,
3033
                               def->portGroups[i].name);
3034
                return -1;
3035
            }
3036
            defaultPortGroup = &def->portGroups[i];
3037
        }
3038 3039 3040 3041 3042 3043 3044 3045 3046
        for (j = i + 1; j < def->nPortGroups; j++) {
            if (STREQ(def->portGroups[i].name, def->portGroups[j].name)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("multiple <portgroup> elements with the "
                                 "same name (%s) in network '%s'"),
                               def->portGroups[i].name, def->name);
                return -1;
            }
        }
3047 3048 3049 3050 3051 3052 3053 3054
        if (def->portGroups[i].bandwidth && !bandwidthAllowed) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <bandwidth> element in network '%s' "
                             "in portgroup '%s' with forward mode='%s'"),
                           def->name, def->portGroups[i].name,
                           virNetworkForwardTypeToString(def->forward.type));
            return -1;
        }
3055
    }
3056 3057 3058 3059 3060 3061 3062
    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.
         */
3063 3064 3065 3066 3067 3068 3069 3070 3071
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
    return 0;
}

3072 3073
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
3074
    virNetworkDriverStatePtr driver = networkGetDriver();
3075
    virNetworkDefPtr def;
3076
    virNetworkObjPtr network = NULL;
3077
    virNetworkPtr ret = NULL;
3078
    virObjectEventPtr event = NULL;
3079

3080
    if (!(def = virNetworkDefParseString(xml)))
3081
        goto cleanup;
3082

3083 3084 3085
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3086
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3087
        goto cleanup;
3088

3089 3090 3091
    /* 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.
3092
     */
3093 3094 3095
    if (!(network = virNetworkAssignDef(driver->networks, def,
                                        VIR_NETWORK_OBJ_LIST_ADD_LIVE |
                                        VIR_NETWORK_OBJ_LIST_ADD_CHECK_LIVE)))
3096 3097
        goto cleanup;
    def = NULL;
3098

3099
    if (networkStartNetwork(driver, network) < 0) {
3100
        virNetworkRemoveInactive(driver->networks,
3101
                                 network);
3102
        goto cleanup;
3103 3104
    }

3105 3106
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3107 3108
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3109

3110
    VIR_INFO("Creating network '%s'", network->def->name);
3111 3112
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

3113
 cleanup:
3114
    virNetworkDefFree(def);
3115 3116
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3117
    virNetworkObjEndAPI(&network);
3118
    return ret;
3119 3120
}

3121 3122
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
3123
    virNetworkDriverStatePtr driver = networkGetDriver();
3124
    virNetworkDefPtr def = NULL;
3125
    bool freeDef = true;
3126
    virNetworkObjPtr network = NULL;
3127
    virNetworkPtr ret = NULL;
3128
    virObjectEventPtr event = NULL;
3129

3130
    if (!(def = virNetworkDefParseString(xml)))
3131
        goto cleanup;
3132

3133 3134 3135
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3136
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3137
        goto cleanup;
3138

3139
    if (!(network = virNetworkAssignDef(driver->networks, def, 0)))
J
Ján Tomko 已提交
3140
        goto cleanup;
3141

3142
    /* def was assigned to network object */
3143
    freeDef = false;
3144 3145

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3146
        if (!virNetworkObjIsActive(network)) {
3147
            virNetworkRemoveInactive(driver->networks, network);
3148 3149
            goto cleanup;
        }
3150 3151 3152 3153 3154
        /* 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);
3155 3156 3157
        goto cleanup;
    }

3158
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3159 3160
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3161

3162 3163
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
3164

3165
 cleanup:
3166 3167
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3168
    if (freeDef)
J
Ján Tomko 已提交
3169
        virNetworkDefFree(def);
3170
    virNetworkObjEndAPI(&network);
3171
    return ret;
3172 3173
}

3174
static int
3175 3176
networkUndefine(virNetworkPtr net)
{
3177
    virNetworkDriverStatePtr driver = networkGetDriver();
3178
    virNetworkObjPtr network;
3179
    int ret = -1;
3180
    bool active = false;
3181
    virObjectEventPtr event = NULL;
3182

3183
    if (!(network = networkObjFromNetwork(net)))
3184
        goto cleanup;
3185

3186 3187 3188
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3189 3190
    if (virNetworkObjIsActive(network))
        active = true;
3191

3192
    /* remove autostart link */
3193
    if (virNetworkDeleteConfig(driver->networkConfigDir,
3194 3195
                               driver->networkAutostartDir,
                               network) < 0)
3196
        goto cleanup;
3197

3198 3199
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3200 3201
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3202

3203
    VIR_INFO("Undefining network '%s'", network->def->name);
3204
    if (!active) {
3205
        if (networkRemoveInactive(driver, network) < 0)
3206
            goto cleanup;
3207 3208 3209 3210 3211 3212
    } 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);
3213 3214
    }

3215
    ret = 0;
3216

3217
 cleanup:
3218 3219
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3220
    virNetworkObjEndAPI(&network);
3221
    return ret;
3222 3223
}

3224 3225 3226 3227 3228 3229 3230 3231
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3232
    virNetworkDriverStatePtr driver = networkGetDriver();
3233
    virNetworkObjPtr network = NULL;
3234 3235
    int isActive, ret = -1;
    size_t i;
3236 3237
    virNetworkIpDefPtr ipdef;
    bool oldDhcpActive = false;
3238
    bool needFirewallRefresh = false;
3239

3240 3241 3242 3243 3244

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3245
    if (!(network = networkObjFromNetwork(net)))
3246 3247
        goto cleanup;

3248 3249 3250
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3251
    /* see if we are listening for dhcp pre-modification */
3252 3253 3254
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
3255 3256 3257 3258 3259 3260
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3261 3262
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3263
     */
3264
    isActive = virNetworkObjIsActive(network);
3265 3266
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3267 3268 3269 3270 3271 3272 3273
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290
    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).
                 */
3291
                networkRemoveFirewallRules(network->def);
3292 3293 3294 3295 3296 3297 3298 3299
                needFirewallRefresh = true;
                break;
            default:
                break;
            }
        }
    }

3300
    /* update the network config in memory/on disk */
3301 3302
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3303
            ignore_value(networkAddFirewallRules(network->def));
3304 3305 3306
        goto cleanup;
    }

3307
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330
        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.
             */
3331
            if (networkRestartDhcpDaemon(driver, network) < 0)
3332 3333
                goto cleanup;

3334 3335 3336 3337 3338 3339 3340 3341
        } 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;

3342 3343 3344
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
                 i++) {
3345 3346 3347 3348 3349 3350 3351
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3352 3353
                 networkRestartDhcpDaemon(driver, network) < 0) ||
                networkRefreshDhcpDaemon(driver, network) < 0) {
3354 3355 3356 3357
                goto cleanup;
            }

        } else if (section == VIR_NETWORK_SECTION_DNS_HOST ||
3358 3359 3360 3361 3362 3363
                   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.
             */
3364
            if (networkRefreshDhcpDaemon(driver, network) < 0)
3365 3366 3367 3368 3369 3370 3371 3372
                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.
             */
3373
            if (networkRefreshRadvd(driver, network) < 0)
3374 3375 3376 3377
                goto cleanup;
        }

        /* save current network state to disk */
3378
        if ((ret = virNetworkSaveStatus(driver->stateDir,
3379
                                        network)) < 0) {
3380
            goto cleanup;
3381
        }
3382 3383
    }
    ret = 0;
3384
 cleanup:
3385
    virNetworkObjEndAPI(&network);
3386 3387 3388
    return ret;
}

3389 3390
static int networkCreate(virNetworkPtr net)
{
3391
    virNetworkDriverStatePtr driver = networkGetDriver();
3392 3393
    virNetworkObjPtr network;
    int ret = -1;
3394
    virObjectEventPtr event = NULL;
3395

3396
    if (!(network = networkObjFromNetwork(net)))
3397
        goto cleanup;
3398

3399 3400 3401
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3402
    if ((ret = networkStartNetwork(driver, network)) < 0)
3403
        goto cleanup;
3404

3405 3406
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3407 3408
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3409

3410
 cleanup:
3411 3412
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3413
    virNetworkObjEndAPI(&network);
3414
    return ret;
3415 3416
}

3417 3418
static int networkDestroy(virNetworkPtr net)
{
3419
    virNetworkDriverStatePtr driver = networkGetDriver();
3420 3421
    virNetworkObjPtr network;
    int ret = -1;
3422
    virObjectEventPtr event = NULL;
3423

3424
    if (!(network = networkObjFromNetwork(net)))
3425
        goto cleanup;
3426

3427 3428 3429
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3430
    if (!virNetworkObjIsActive(network)) {
3431
        virReportError(VIR_ERR_OPERATION_INVALID,
3432 3433
                       _("network '%s' is not active"),
                       network->def->name);
3434 3435 3436
        goto cleanup;
    }

3437
    if ((ret = networkShutdownNetwork(driver, network)) < 0)
3438 3439
        goto cleanup;

3440 3441
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3442 3443
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3444

3445
    if (!network->persistent &&
3446
        networkRemoveInactive(driver, network) < 0) {
3447 3448
        ret = -1;
        goto cleanup;
3449
    }
3450

3451
 cleanup:
3452 3453
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3454
    virNetworkObjEndAPI(&network);
3455 3456 3457
    return ret;
}

3458
static char *networkGetXMLDesc(virNetworkPtr net,
3459
                               unsigned int flags)
3460
{
3461
    virNetworkObjPtr network;
3462
    virNetworkDefPtr def;
3463
    char *ret = NULL;
3464

3465
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3466

3467 3468
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3469

3470 3471 3472
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3473 3474 3475 3476 3477 3478
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3479

3480
 cleanup:
3481
    virNetworkObjEndAPI(&network);
3482
    return ret;
3483 3484 3485
}

static char *networkGetBridgeName(virNetworkPtr net) {
3486 3487 3488
    virNetworkObjPtr network;
    char *bridge = NULL;

3489 3490
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3491

3492 3493 3494
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3495
    if (!(network->def->bridge)) {
3496 3497 3498
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3499 3500 3501
        goto cleanup;
    }

3502
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3503

3504
 cleanup:
3505
    virNetworkObjEndAPI(&network);
3506 3507 3508 3509
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
J
Ján Tomko 已提交
3510
                               int *autostart)
3511
{
3512 3513
    virNetworkObjPtr network;
    int ret = -1;
3514

3515 3516
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3517

3518 3519 3520
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3521
    *autostart = network->autostart;
3522
    ret = 0;
3523

3524
 cleanup:
3525
    virNetworkObjEndAPI(&network);
3526
    return ret;
3527 3528 3529
}

static int networkSetAutostart(virNetworkPtr net,
3530 3531
                               int autostart)
{
3532
    virNetworkDriverStatePtr driver = networkGetDriver();
3533
    virNetworkObjPtr network;
3534
    char *configFile = NULL, *autostartLink = NULL;
3535
    int ret = -1;
3536

3537

3538
    if (!(network = networkObjFromNetwork(net)))
3539
        goto cleanup;
3540

3541 3542 3543
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3544
    if (!network->persistent) {
3545 3546
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3547 3548 3549
        goto cleanup;
    }

3550 3551
    autostart = (autostart != 0);

3552
    if (network->autostart != autostart) {
3553
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3554
            goto cleanup;
3555
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3556 3557
            goto cleanup;

3558
        if (autostart) {
3559
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3560
                virReportSystemError(errno,
3561 3562
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3563 3564
                goto cleanup;
            }
3565

3566
            if (symlink(configFile, autostartLink) < 0) {
3567
                virReportSystemError(errno,
3568
                                     _("Failed to create symlink '%s' to '%s'"),
3569
                                     autostartLink, configFile);
3570 3571 3572
                goto cleanup;
            }
        } else {
3573
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3574
                virReportSystemError(errno,
3575
                                     _("Failed to delete symlink '%s'"),
3576
                                     autostartLink);
3577 3578
                goto cleanup;
            }
3579 3580
        }

3581
        network->autostart = autostart;
3582
    }
3583
    ret = 0;
3584

3585
 cleanup:
3586 3587
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3588
    virNetworkObjEndAPI(&network);
3589
    return ret;
3590 3591
}

3592
static int
3593 3594 3595 3596
networkGetDHCPLeases(virNetworkPtr network,
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
3597
{
3598
    virNetworkDriverStatePtr driver = networkGetDriver();
3599 3600 3601
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
3602
    ssize_t size = 0;
3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616
    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;
3617 3618 3619 3620 3621 3622 3623 3624 3625
    virNetworkObjPtr obj;

    virCheckFlags(0, -1);

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

    if (virNetworkGetDHCPLeasesEnsureACL(network->conn, obj->def) < 0)
        goto cleanup;
3626 3627

    /* Retrieve custom leases file location */
3628
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(driver, obj->def->bridge);
3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671

    /* 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;
        }

3672
        if (mac && virMacAddrCompare(mac, mac_tmp))
3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722
            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) ||
3723
                (VIR_STRDUP(lease->iface, obj->def->bridge) < 0))
3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755
                goto error;

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

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

        } else {
            nleases++;
        }

        VIR_FREE(lease);
    }

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

    rv = nleases;

 cleanup:
    VIR_FREE(lease);
3756
    VIR_FREE(lease_entries);
3757 3758
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
3759

3760
    virNetworkObjEndAPI(&obj);
3761

3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772
    return rv;

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

3773 3774

static virNetworkDriver networkDriver = {
3775
    .name = "bridge",
3776 3777 3778 3779 3780
    .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 */
3781 3782
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
3783 3784
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3785 3786
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3787
    .networkUndefine = networkUndefine, /* 0.2.0 */
3788
    .networkUpdate = networkUpdate, /* 0.10.2 */
3789
    .networkCreate = networkCreate, /* 0.2.0 */
3790 3791 3792 3793 3794 3795 3796
    .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 */
3797
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
3798 3799 3800
};

static virStateDriver networkStateDriver = {
3801
    .name = "bridge",
3802
    .stateInitialize  = networkStateInitialize,
3803
    .stateAutoStart  = networkStateAutoStart,
3804 3805
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3806 3807
};

3808 3809
int networkRegister(void)
{
3810
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
3811
        return -1;
3812 3813
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
3814 3815
    return 0;
}
3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827

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

/* 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:
3828
 * @dom: domain definition that @iface belongs to
3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839
 * @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
3840 3841
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
3842
{
3843
    virNetworkDriverStatePtr driver = networkGetDriver();
3844
    virDomainNetType actualType = iface->type;
3845 3846
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
3847
    virNetDevBandwidthPtr bandwidth = NULL;
3848 3849 3850
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3851
    virNetworkForwardIfDefPtr dev = NULL;
3852
    size_t i;
3853 3854 3855
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3856
        goto validate;
3857 3858 3859 3860

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

3861
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
3862
    if (!network) {
3863 3864 3865
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3866
        goto error;
3867 3868
    }
    netdef = network->def;
3869

3870 3871 3872 3873 3874 3875 3876
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

3877 3878 3879
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

3880 3881 3882
    /* 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
3883
     */
3884 3885 3886 3887 3888 3889
    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.
     */
3890 3891 3892 3893 3894 3895

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

3896 3897
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
3898
        goto error;
3899

3900 3901 3902 3903 3904 3905 3906 3907
    /* 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;

3908 3909
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
3910

3911 3912 3913 3914 3915 3916 3917 3918 3919 3920
    if (iface->trustGuestRxFilters)
       iface->data.network.actual->trustGuestRxFilters
          = iface->trustGuestRxFilters;
    else if (portgroup && portgroup->trustGuestRxFilters)
       iface->data.network.actual->trustGuestRxFilters
          = portgroup->trustGuestRxFilters;
    else if (netdef->trustGuestRxFilters)
       iface->data.network.actual->trustGuestRxFilters
          = netdef->trustGuestRxFilters;

3921 3922 3923
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
3924
        /* for these forward types, the actual net type really *is*
3925
         * NETWORK; we just keep the info from the portgroup in
3926
         * iface->data.network.actual
J
Ján Tomko 已提交
3927
         */
3928
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
3929

3930
        /* we also store the bridge device and macTableManager settings
3931 3932 3933 3934 3935 3936 3937
         * in iface->data.network.actual->data.bridge for later use
         * after the domain's tap device is created (to attach to the
         * bridge and set flood/learning mode on the tap device)
         */
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
            goto error;
3938 3939
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
3940

3941 3942 3943
        if (networkPlugBandwidth(network, iface) < 0)
            goto error;

3944
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
3945
               netdef->bridge) {
3946 3947 3948 3949 3950

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

3951
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
3952 3953
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
3954
            goto error;
3955 3956
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
3957

3958 3959 3960 3961 3962 3963 3964 3965
        /* 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) {
3966
            goto error;
3967 3968 3969 3970 3971 3972 3973 3974 3975 3976
        }
        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);
3977
                goto error;
3978 3979 3980
            }
        }

3981
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
3982

3983
        virDomainHostdevSubsysPCIBackendType backend;
3984

3985
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
3986
        if (networkCreateInterfacePool(netdef) < 0)
3987 3988 3989
            goto error;

        /* pick first dev with 0 connections */
3990 3991 3992
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006
                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;
4007
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
4008
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
4009
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
4010

E
Eric Blake 已提交
4011
        switch (netdef->forward.driverName) {
4012
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
4013
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
4014 4015
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
4016
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
4017 4018
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
4019
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030
            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;

4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055
        /* 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;
            }
        }

4056 4057 4058 4059
    } 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)) {
4060 4061 4062 4063 4064 4065

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
4066
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4067
        switch (netdef->forward.type) {
4068
        case VIR_NETWORK_FORWARD_BRIDGE:
4069
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
4070 4071
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
4072
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
4073 4074
            break;
        case VIR_NETWORK_FORWARD_VEPA:
4075
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
4076 4077
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
4078
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
4079 4080 4081
            break;
        }

4082 4083 4084 4085 4086 4087 4088 4089
        /* 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) {
4090
            goto error;
4091
        }
4092
        virtport = iface->data.network.actual->virtPortProfile;
4093
        if (virtport) {
4094 4095 4096 4097 4098 4099 4100 4101
            /* 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);
4102
                goto error;
4103 4104
            }
        }
4105

4106 4107 4108
        /* 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).
         */
4109
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4110 4111 4112 4113
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4114
            goto error;
4115 4116 4117
        } else {
            /* pick an interface from the pool */

4118
            if (networkCreateInterfacePool(netdef) < 0)
4119 4120
                goto error;

4121 4122 4123 4124 4125
            /* 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.
4126
             */
4127 4128
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4129 4130 4131
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4132

4133
                /* pick first dev with 0 connections */
4134 4135 4136
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4137 4138 4139 4140 4141
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4142
                dev = &netdef->forward.ifs[0];
4143 4144 4145
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4146 4147 4148 4149
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4150 4151 4152 4153
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4154
                goto error;
4155
            }
4156 4157
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4158
                goto error;
4159 4160 4161
        }
    }

4162
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4163
        goto error;
4164

4165
 validate:
4166 4167 4168 4169 4170
    /* 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.
     */

4171
    if (virDomainNetGetActualVlan(iface)) {
4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200
        /* 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);
4201

4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220
        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 */
4221
            netdef->connections--;
4222 4223 4224 4225
            if (dev)
                dev->connections--;
            goto error;
        }
4226 4227
    }

4228
    ret = 0;
4229

4230
 cleanup:
4231
    virNetworkObjEndAPI(&network);
4232 4233
    return ret;

4234
 error:
4235
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4236 4237 4238
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4239
    goto cleanup;
4240 4241 4242
}

/* networkNotifyActualDevice:
4243
 * @dom: domain definition that @iface belongs to
4244 4245 4246 4247 4248 4249 4250 4251 4252 4253
 * @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
4254 4255
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4256
{
4257
    virNetworkDriverStatePtr driver = networkGetDriver();
4258
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4259 4260
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4261
    virNetworkForwardIfDefPtr dev = NULL;
4262 4263
    size_t i;
    int ret = -1;
4264 4265

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4266
        return 0;
4267

4268
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4269
    if (!network) {
4270 4271 4272
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4273 4274 4275 4276
        goto error;
    }
    netdef = network->def;

4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287
    /* if we're restarting libvirtd after an upgrade from a version
     * that didn't save bridge name in actualNetDef for
     * actualType==network, we need to copy it in so that it will be
     * available in all cases
     */
    if (actualType == VIR_DOMAIN_NET_TYPE_NETWORK &&
        !iface->data.network.actual->data.bridge.brname &&
        (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                    netdef->bridge) < 0))
            goto error;

4288
    if (!iface->data.network.actual ||
4289 4290
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4291 4292
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4293 4294
    }

4295
    if (networkCreateInterfacePool(netdef) < 0)
4296
        goto error;
4297

4298
    if (netdef->forward.nifs == 0) {
4299
        virReportError(VIR_ERR_INTERNAL_ERROR,
4300 4301
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4302
                       netdef->name);
4303
        goto error;
4304
    }
4305

4306 4307
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4308

4309 4310 4311 4312 4313 4314 4315 4316 4317
        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 */
4318 4319
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4320
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4321 4322
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4323 4324 4325 4326 4327
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4328
            virReportError(VIR_ERR_INTERNAL_ERROR,
4329 4330
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4331
                           netdef->name, actualDev);
4332
            goto error;
4333 4334
        }

4335
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4336 4337
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4338
         */
4339
        if ((dev->connections > 0) &&
4340 4341
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4342 4343
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4344
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4345
            virReportError(VIR_ERR_INTERNAL_ERROR,
4346 4347
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4348
                           netdef->name, actualDev);
4349
            goto error;
4350
        }
4351

4352
        /* we are now assured of success, so mark the allocation */
4353
        dev->connections++;
4354
        VIR_DEBUG("Using physical device %s, connections %d",
4355
                  dev->device.dev, dev->connections);
4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368

    }  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 */
4369 4370
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4371
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4372
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4373 4374
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4375 4376 4377 4378 4379 4380 4381 4382 4383
                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,
4384 4385 4386 4387
                           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
4388
            goto error;
4389 4390 4391 4392 4393 4394 4395
        }

        /* 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) &&
4396
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412
            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);
4413 4414
    }

4415
 success:
4416 4417 4418
    netdef->connections++;
    VIR_DEBUG("Using network %s, %d connections",
              netdef->name, netdef->connections);
4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429

    /* 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;
    }

4430
    ret = 0;
4431
 cleanup:
4432
    virNetworkObjEndAPI(&network);
4433
    return ret;
4434

4435
 error:
4436
    goto cleanup;
4437 4438 4439 4440
}


/* networkReleaseActualDevice:
4441
 * @dom: domain definition that @iface belongs to
4442 4443 4444 4445 4446 4447 4448 4449 4450 4451
 * @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
4452 4453
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4454
{
4455
    virNetworkDriverStatePtr driver = networkGetDriver();
4456
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4457
    virNetworkObjPtr network;
4458
    virNetworkDefPtr netdef;
4459
    virNetworkForwardIfDefPtr dev = NULL;
4460 4461
    size_t i;
    int ret = -1;
4462 4463

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4464
        return 0;
4465

4466
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4467
    if (!network) {
4468 4469 4470
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4471 4472 4473 4474
        goto error;
    }
    netdef = network->def;

4475 4476
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4477 4478 4479 4480 4481
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) &&
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4482 4483 4484
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4485 4486
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4487 4488
    }

4489
    if (netdef->forward.nifs == 0) {
4490
        virReportError(VIR_ERR_INTERNAL_ERROR,
4491
                       _("network '%s' uses a direct/hostdev mode, but "
4492 4493
                         "has no forward dev and no interface pool"),
                       netdef->name);
4494
        goto error;
4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506
    }

    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;
        }
4507

4508 4509
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4510
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4511 4512
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4513 4514 4515
                break;
            }
        }
4516

4517
        if (!dev) {
4518
            virReportError(VIR_ERR_INTERNAL_ERROR,
4519 4520
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4521
                           netdef->name, actualDev);
4522
            goto error;
4523 4524
        }

4525
        dev->connections--;
4526
        VIR_DEBUG("Releasing physical device %s, connections %d",
4527
                  dev->device.dev, dev->connections);
4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538

    } 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;
        }

4539 4540
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4541
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4542
                virDevicePCIAddressEqual(&hostdev->source.subsys.u.pci.addr,
4543 4544
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4545 4546 4547 4548 4549 4550 4551 4552 4553
                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,
4554 4555 4556 4557
                           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
4558
            goto error;
4559 4560 4561 4562 4563 4564 4565
        }

        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
4566
    }
4567

4568
 success:
4569
    if (iface->data.network.actual) {
4570
        netdef->connections--;
4571 4572
        VIR_DEBUG("Releasing network %s, %d connections",
                  netdef->name, netdef->connections);
4573

4574 4575 4576 4577
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
    }
4578
    ret = 0;
4579
 cleanup:
4580
    virNetworkObjEndAPI(&network);
4581 4582 4583 4584
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4585
    return ret;
4586

4587
 error:
4588
    goto cleanup;
4589
}
4590 4591 4592 4593 4594 4595

/*
 * networkGetNetworkAddress:
 * @netname: the name of a network
 * @netaddr: string representation of IP address for that network.
 *
4596
 * Attempt to return an IP address associated with the named
4597 4598 4599 4600
 * 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.
 *
4601
 * Note: This function returns the first IP address it finds. It might
4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612
 * 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)
{
4613
    virNetworkDriverStatePtr driver = networkGetDriver();
4614
    int ret = -1;
4615
    virNetworkObjPtr network;
4616 4617 4618 4619
    virNetworkDefPtr netdef;
    virNetworkIpDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4620
    char *dev_name = NULL;
4621 4622

    *netaddr = NULL;
4623
    network = virNetworkObjFindByName(driver->networks, netname);
4624
    if (!network) {
4625 4626 4627
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4628
        goto cleanup;
4629 4630 4631
    }
    netdef = network->def;

4632
    switch (netdef->forward.type) {
4633 4634 4635
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
4636
        ipdef = virNetworkDefGetIpByIndex(netdef, AF_UNSPEC, 0);
4637
        if (!ipdef) {
4638
            virReportError(VIR_ERR_INTERNAL_ERROR,
4639
                           _("network '%s' doesn't have an IP address"),
4640
                           netdef->name);
4641
            goto cleanup;
4642 4643 4644 4645 4646
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4647
        if ((dev_name = netdef->bridge))
4648 4649 4650 4651 4652 4653 4654 4655
            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:
4656 4657
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4658

4659
        if (!dev_name) {
4660 4661 4662
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4663
            goto cleanup;
4664 4665 4666 4667
        }
        break;
    }

4668
    if (dev_name) {
4669
        if (virNetDevGetIPAddress(dev_name, &addr) < 0)
4670
            goto cleanup;
4671
        addrptr = &addr;
4672 4673
    }

4674 4675
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
4676
        goto cleanup;
4677 4678
    }

4679
    ret = 0;
4680
 cleanup:
4681
    virNetworkObjEndAPI(&network);
4682 4683
    return ret;
}
4684 4685 4686 4687

/**
 * networkCheckBandwidth:
 * @net: network QoS
4688
 * @ifaceBand: interface QoS (may be NULL if no QoS)
4689
 * @oldBandwidth: new interface QoS (may be NULL if no QoS)
4690
 * @ifaceMac: interface MAC (used in error messages for identification)
4691 4692
 * @new_rate: new rate for non guaranteed class
 *
4693 4694 4695 4696 4697 4698 4699 4700
 * Function checks if @ifaceBand can be satisfied on @net. However, sometimes it
 * may happen that the interface that @ifaceBand corresponds to is already
 * plugged into the @net and the bandwidth is to be updated. In that case we
 * need to check if new bandwidth can be satisfied. If that's the case
 * @ifaceBand should point to new bandwidth settings and @oldBandwidth to
 * current ones. If you want to suppress this functionality just pass
 * @oldBandwidth == NULL.
 *
4701 4702 4703 4704 4705 4706
 * 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,
4707
                      virNetDevBandwidthPtr ifaceBand,
4708
                      virNetDevBandwidthPtr oldBandwidth,
4709
                      virMacAddr ifaceMac,
4710 4711 4712 4713 4714 4715 4716 4717
                      unsigned long long *new_rate)
{
    int ret = -1;
    virNetDevBandwidthPtr netBand = net->def->bandwidth;
    unsigned long long tmp_floor_sum = net->floor_sum;
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

4718
    virMacAddrFormat(&ifaceMac, ifmac);
4719 4720 4721 4722 4723 4724 4725 4726 4727 4728

    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;
    }

4729 4730
    if (((!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor) &&
         (!oldBandwidth || !oldBandwidth->in || !oldBandwidth->in->floor)) ||
4731 4732
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4733
        return 1;
4734
    }
4735 4736

    tmp_new_rate = netBand->in->average;
4737 4738 4739 4740
    if (oldBandwidth && oldBandwidth->in)
        tmp_floor_sum -= oldBandwidth->in->floor;
    if (ifaceBand && ifaceBand->in)
        tmp_floor_sum += ifaceBand->in->floor;
4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765

    /* 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;
    }

4766 4767
    if (new_rate)
        *new_rate = tmp_new_rate;
4768 4769
    ret = 0;

4770
 cleanup:
4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786
    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)
{
4787
    ssize_t ret = 0;
4788

4789
    ret = virBitmapNextClearBit(net->class_id, -1);
4790

4791
    if (ret < 0 || virBitmapSetBit(net->class_id, ret) < 0)
4792 4793 4794 4795 4796
        return -1;

    return ret;
}

4797

4798
static int
4799 4800 4801 4802
networkPlugBandwidthImpl(virNetworkObjPtr net,
                         virDomainNetDefPtr iface,
                         virNetDevBandwidthPtr ifaceBand,
                         unsigned long long new_rate)
4803
{
4804
    virNetworkDriverStatePtr driver = networkGetDriver();
4805
    ssize_t class_id = 0;
4806 4807
    int plug_ret;
    int ret = -1;
4808 4809 4810 4811 4812 4813 4814 4815

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

4816 4817
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
4818 4819 4820 4821 4822 4823 4824 4825
    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 */
4826
    net->floor_sum += ifaceBand->in->floor;
4827
    /* update status file */
4828
    if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
4829
        ignore_value(virBitmapClearBit(net->class_id, class_id));
4830
        net->floor_sum -= ifaceBand->in->floor;
4831 4832 4833 4834
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
4835 4836
    /* update rate for non guaranteed NICs */
    new_rate -= net->floor_sum;
4837
    if (virNetDevBandwidthUpdateRate(net->def->bridge, 2,
4838 4839 4840 4841 4842
                                     net->def->bandwidth, new_rate) < 0)
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                 net->def->bridge);

    ret = 0;
4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882
 cleanup:
    return ret;
}


static int
networkPlugBandwidth(virNetworkObjPtr net,
                     virDomainNetDefPtr iface)
{
    int ret = -1;
    int plug_ret;
    unsigned long long new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);

    if ((plug_ret = networkCheckBandwidth(net, ifaceBand, NULL,
                                          iface->mac, &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;
    }

    if (networkPlugBandwidthImpl(net, iface, ifaceBand, new_rate) < 0)
        goto cleanup;

    ret = 0;
4883

4884
 cleanup:
4885 4886 4887 4888 4889 4890 4891
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
4892
    virNetworkDriverStatePtr driver = networkGetDriver();
4893 4894
    int ret = 0;
    unsigned long long new_rate;
4895
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
4896 4897 4898

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
4899 4900 4901 4902 4903
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914
        /* 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 */
4915
        net->floor_sum -= ifaceBand->in->floor;
4916 4917 4918 4919
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
4920
        if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
4921
            net->floor_sum += ifaceBand->in->floor;
4922 4923 4924 4925
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
4926 4927
        /* update rate for non guaranteed NICs */
        new_rate -= net->floor_sum;
4928
        if (virNetDevBandwidthUpdateRate(net->def->bridge, 2,
4929 4930 4931 4932 4933 4934 4935
                                         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;
    }

4936
 cleanup:
4937 4938
    return ret;
}
4939 4940 4941

static void
networkNetworkObjTaint(virNetworkObjPtr net,
4942
                       virNetworkTaintFlags taint)
4943 4944 4945 4946 4947 4948 4949 4950 4951 4952 4953
{
    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));
    }
}
4954 4955 4956 4957 4958 4959


static bool
networkBandwidthGenericChecks(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
4960
    virNetDevBandwidthPtr ifaceBand;
4961 4962 4963 4964 4965 4966 4967 4968
    unsigned long long old_floor, new_floor;

    if (virDomainNetGetActualType(iface) != VIR_DOMAIN_NET_TYPE_NETWORK) {
        /* This is not an interface that's plugged into a network.
         * We don't care. Thus from our POV bandwidth change is allowed. */
        return false;
    }

4969
    ifaceBand = virDomainNetGetActualBandwidth(iface);
4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009
    old_floor = new_floor = 0;

    if (ifaceBand && ifaceBand->in)
        old_floor = ifaceBand->in->floor;
    if (newBandwidth && newBandwidth->in)
        new_floor = newBandwidth->in->floor;

    return new_floor != old_floor;
}


bool
networkBandwidthChangeAllowed(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
    virNetworkObjPtr network = NULL;
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    bool ret = false;

    if (!networkBandwidthGenericChecks(iface, newBandwidth))
        return true;

    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!network) {
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return false;
    }

    if (networkCheckBandwidth(network, newBandwidth, ifaceBand, iface->mac, NULL) < 0)
        goto cleanup;

    ret = true;

 cleanup:
    virNetworkObjEndAPI(&network);
    return ret;
}
5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047


int
networkBandwidthUpdate(virDomainNetDefPtr iface,
                       virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
    virNetworkObjPtr network = NULL;
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    unsigned long long new_rate = 0;
    int plug_ret;
    int ret = -1;

    if (!networkBandwidthGenericChecks(iface, newBandwidth))
        return 0;

    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!network) {
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return ret;
    }

    if ((plug_ret = networkCheckBandwidth(network, newBandwidth, ifaceBand,
                                          iface->mac, &new_rate)) < 0) {
        /* helper reported error */
        goto cleanup;
    }

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

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

5048 5049
    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
        newBandwidth && newBandwidth->in && newBandwidth->in->floor) {
5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091
        /* Either we just need to update @floor .. */

        if (virNetDevBandwidthUpdateRate(network->def->bridge,
                                         iface->data.network.actual->class_id,
                                         network->def->bandwidth,
                                         newBandwidth->in->floor) < 0)
            goto cleanup;

        network->floor_sum -= ifaceBand->in->floor;
        network->floor_sum += newBandwidth->in->floor;
        new_rate -= network->floor_sum;

        if (virNetDevBandwidthUpdateRate(network->def->bridge, 2,
                                         network->def->bandwidth, new_rate) < 0 ||
            virNetworkSaveStatus(driver->stateDir, network) < 0) {
            /* Ouch, rollback */
            network->floor_sum -= newBandwidth->in->floor;
            network->floor_sum += ifaceBand->in->floor;

            ignore_value(virNetDevBandwidthUpdateRate(network->def->bridge,
                                                      iface->data.network.actual->class_id,
                                                      network->def->bandwidth,
                                                      ifaceBand->in->floor));
            goto cleanup;
        }
    } else if (newBandwidth->in && newBandwidth->in->floor) {
        /* .. or we need to plug in new .. */

        if (networkPlugBandwidthImpl(network, iface, newBandwidth, new_rate) < 0)
            goto cleanup;
    } else {
        /* .. or unplug old. */

        if (networkUnplugBandwidth(network, iface) < 0)
            goto cleanup;
    }

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