bridge_driver.c 174.1 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2016 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
 *
 * 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 <pwd.h>
#include <sys/wait.h>
#include <sys/ioctl.h>
42
#include <net/if.h>
43
#include <dirent.h>
44 45 46
#if HAVE_SYS_SYSCTL_H
# include <sys/sysctl.h>
#endif
47

48
#include "virerror.h"
49
#include "datatypes.h"
50
#include "bridge_driver.h"
51
#include "bridge_driver_platform.h"
52
#include "network_conf.h"
53
#include "device_conf.h"
54
#include "driver.h"
55
#include "virbuffer.h"
56
#include "virpidfile.h"
57
#include "vircommand.h"
58
#include "viralloc.h"
59
#include "viruuid.h"
60
#include "viriptables.h"
61
#include "virlog.h"
62
#include "virdnsmasq.h"
63
#include "configmake.h"
64
#include "virnetdev.h"
65
#include "virpci.h"
66 67
#include "virnetdevbridge.h"
#include "virnetdevtap.h"
68
#include "virnetdevvportprofile.h"
69
#include "virdbus.h"
70
#include "virfile.h"
71
#include "virstring.h"
72
#include "viraccessapicheck.h"
73
#include "network_event.h"
74
#include "virhook.h"
75
#include "virjson.h"
76

77
#define VIR_FROM_THIS VIR_FROM_NETWORK
78
#define MAX_BRIDGE_ID 256
79

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

87 88
VIR_LOG_INIT("network.bridge_driver");

89
static virNetworkDriverStatePtr network_driver;
90

91 92 93 94 95 96 97 98
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;
}
99

100
static void networkDriverLock(virNetworkDriverStatePtr driver)
101
{
102
    virMutexLock(&driver->lock);
103
}
104
static void networkDriverUnlock(virNetworkDriverStatePtr driver)
105
{
106
    virMutexUnlock(&driver->lock);
107 108
}

109 110 111 112
static dnsmasqCapsPtr
networkGetDnsmasqCaps(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr ret;
113
    networkDriverLock(driver);
114
    ret = virObjectRef(driver->dnsmasqCaps);
115
    networkDriverUnlock(driver);
116 117 118 119 120 121 122 123 124 125 126
    return ret;
}

static int
networkDnsmasqCapsRefresh(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr caps;

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

127
    networkDriverLock(driver);
128 129
    virObjectUnref(driver->dnsmasqCaps);
    driver->dnsmasqCaps = caps;
130
    networkDriverUnlock(driver);
131 132 133
    return 0;
}

134
static int networkStateCleanup(void);
135

136 137
static int networkStartNetwork(virNetworkDriverStatePtr driver,
                               virNetworkObjPtr network);
138

139 140
static int networkShutdownNetwork(virNetworkDriverStatePtr driver,
                                  virNetworkObjPtr network);
141

142 143
static int networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
                                      virNetworkObjPtr network);
144

145 146
static int networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
                                         virNetworkObjPtr network);
147

148
static int networkStartNetworkExternal(virNetworkObjPtr network);
149

150
static int networkShutdownNetworkExternal(virNetworkObjPtr network);
151

152 153
static void networkReloadFirewallRules(virNetworkDriverStatePtr driver);
static void networkRefreshDaemons(virNetworkDriverStatePtr driver);
154

155 156 157 158 159
static int networkPlugBandwidth(virNetworkObjPtr net,
                                virDomainNetDefPtr iface);
static int networkUnplugBandwidth(virNetworkObjPtr net,
                                  virDomainNetDefPtr iface);

160
static void networkNetworkObjTaint(virNetworkObjPtr net,
161
                                   virNetworkTaintFlags taint);
162

163 164 165
static virNetworkObjPtr
networkObjFromNetwork(virNetworkPtr net)
{
166
    virNetworkDriverStatePtr driver = networkGetDriver();
167 168 169
    virNetworkObjPtr network;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

170
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
171 172 173 174 175 176 177 178 179 180
    if (!network) {
        virUUIDFormat(net->uuid, uuidstr);
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching uuid '%s' (%s)"),
                       uuidstr, net->name);
    }

    return network;
}

181 182 183
static int
networkRunHook(virNetworkObjPtr network,
               virDomainDefPtr dom,
184
               virDomainNetDefPtr iface,
185 186 187 188 189 190 191 192 193
               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)) {
194 195 196 197 198 199
        if (!network) {
            VIR_DEBUG("Not running hook as @network is NULL");
            ret = 0;
            goto cleanup;
        }

200 201
        virBufferAddLit(&buf, "<hookData>\n");
        virBufferAdjustIndent(&buf, 2);
202
        if (iface && virDomainNetDefFormat(&buf, iface, NULL, 0) < 0)
203
            goto cleanup;
204 205
        if (virNetworkDefFormatBuf(&buf, network->def, 0) < 0)
            goto cleanup;
206
        if (dom && virDomainDefFormatInternal(dom, NULL, 0, &buf) < 0)
207 208 209 210 211
            goto cleanup;

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

212
        if (virBufferCheckError(&buf) < 0)
213 214
            goto cleanup;

215
        xml = virBufferContentAndReset(&buf);
216 217 218 219 220 221 222 223
        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;
224 225

        networkNetworkObjTaint(network, VIR_NETWORK_TAINT_HOOK);
226 227 228
    }

    ret = 0;
229
 cleanup:
230 231 232 233 234 235 236
    virBufferFreeAndReset(&buf);
    VIR_FREE(xml);
    VIR_FREE(net_xml);
    VIR_FREE(dom_xml);
    return ret;
}

237
static char *
238 239
networkDnsmasqLeaseFileNameDefault(virNetworkDriverStatePtr driver,
                                   const char *netname)
240 241 242
{
    char *leasefile;

243
    ignore_value(virAsprintf(&leasefile, "%s/%s.leases",
244
                             driver->dnsmasqStateDir, netname));
245 246 247
    return leasefile;
}

248
static char *
249 250
networkDnsmasqLeaseFileNameCustom(virNetworkDriverStatePtr driver,
                                  const char *bridge)
251 252 253 254
{
    char *leasefile;

    ignore_value(virAsprintf(&leasefile, "%s/%s.status",
255
                             driver->dnsmasqStateDir, bridge));
256 257 258
    return leasefile;
}

259
static char *
260 261
networkDnsmasqConfigFileName(virNetworkDriverStatePtr driver,
                             const char *netname)
262 263 264
{
    char *conffile;

265
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
266
                             driver->dnsmasqStateDir, netname));
267 268 269
    return conffile;
}

270 271 272 273 274 275
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

276
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
277 278 279 280
    return pidfilebase;
}

static char *
281 282
networkRadvdConfigFileName(virNetworkDriverStatePtr driver,
                           const char *netname)
283 284 285
{
    char *configfile;

286
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
287
                             driver->radvdStateDir, netname));
288 289
    return configfile;
}
290

291 292
/* do needed cleanup steps and remove the network from the list */
static int
293 294
networkRemoveInactive(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr net)
295 296
{
    char *leasefile = NULL;
297
    char *customleasefile = NULL;
298
    char *radvdconfigfile = NULL;
299
    char *configfile = NULL;
300
    char *radvdpidbase = NULL;
301
    char *statusfile = NULL;
302 303 304 305 306 307
    dnsmasqContext *dctx = NULL;
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(net);

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
308
    if (!(dctx = dnsmasqContextNew(def->name,
309
                                   driver->dnsmasqStateDir))) {
310
        goto cleanup;
311
    }
312

313
    if (!(leasefile = networkDnsmasqLeaseFileNameDefault(driver, def->name)))
314 315
        goto cleanup;

316
    if (!(customleasefile = networkDnsmasqLeaseFileNameCustom(driver, def->bridge)))
317 318
        goto cleanup;

319
    if (!(radvdconfigfile = networkRadvdConfigFileName(driver, def->name)))
320
        goto cleanup;
321 322

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
323
        goto cleanup;
324

325
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
326
        goto cleanup;
327

328
    if (!(statusfile = virNetworkConfigFile(driver->stateDir, def->name)))
329
        goto cleanup;
330

331 332 333
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
334
    unlink(customleasefile);
335
    unlink(configfile);
336 337 338

    /* radvd */
    unlink(radvdconfigfile);
339
    virPidFileDelete(driver->pidDir, radvdpidbase);
340

341 342 343
    /* remove status file */
    unlink(statusfile);

344
    /* remove the network definition */
345
    virNetworkRemoveInactive(driver->networks, net);
346 347 348

    ret = 0;

349
 cleanup:
350
    VIR_FREE(leasefile);
351
    VIR_FREE(configfile);
352
    VIR_FREE(customleasefile);
353 354
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
355
    VIR_FREE(statusfile);
356 357 358 359
    dnsmasqContextFree(dctx);
    return ret;
}

360 361 362
static char *
networkBridgeDummyNicName(const char *brname)
{
363
    static const char dummyNicSuffix[] = "-nic";
364 365
    char *nicname;

366 367 368 369 370 371 372
    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.
         */
373 374 375 376 377
        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));
378
    } else {
379
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
380
    }
381 382 383
    return nicname;
}

384 385
static int
networkUpdateState(virNetworkObjPtr obj,
386
                   void *opaque)
387
{
388
    virNetworkDriverStatePtr driver = opaque;
389
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
390
    int ret = -1;
391

392
    virObjectLock(obj);
393
    if (!virNetworkObjIsActive(obj)) {
394 395
        ret = 0;
        goto cleanup;
396
    }
397

398 399 400 401 402 403 404 405
    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;
406

407
    case VIR_NETWORK_FORWARD_BRIDGE:
408 409
        if (obj->def->bridge) {
            if (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
#ifdef HAVE_FIREWALLD
    if (!(sysbus = virDBusGetSystemBus())) {
        VIR_WARN("DBus not available, disabling firewalld support "
699
                 "in bridge_network_driver: %s", virGetLastErrorMessage());
700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716
    } 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,
717
                                   network_driver, NULL);
718 719 720
    }
#endif

721
    ret = 0;
722
 cleanup:
723 724 725
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
726

727
 error:
728 729
    if (network_driver)
        networkDriverUnlock(network_driver);
730
    networkStateCleanup();
731
    goto cleanup;
732 733
}

734 735 736 737 738 739 740 741
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
742
    if (!network_driver)
743 744
        return;

745
    virNetworkObjListForEach(network_driver->networks,
746
                             networkAutostartConfig,
747
                             network_driver);
748 749
}

750
/**
751
 * networkStateReload:
752 753 754 755 756
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
757 758
networkStateReload(void)
{
759
    if (!network_driver)
760 761
        return 0;

762 763 764 765 766 767 768 769
    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,
770
                             networkAutostartConfig,
771
                             network_driver);
772 773 774 775 776
    return 0;
}


/**
777
 * networkStateCleanup:
778 779 780 781
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
782 783
networkStateCleanup(void)
{
784
    if (!network_driver)
785 786
        return -1;

787
    virObjectEventStateFree(network_driver->networkEventState);
788

789
    /* free inactive networks */
790
    virObjectUnref(network_driver->networks);
791

792 793 794 795 796 797
    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);
798

799
    virObjectUnref(network_driver->dnsmasqCaps);
800

801
    virMutexDestroy(&network_driver->lock);
802

803
    VIR_FREE(network_driver);
804 805 806 807 808

    return 0;
}


809 810 811 812 813 814 815
/* 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)
{
816 817
    size_t i;
    int ret = -1;
818 819 820 821 822 823 824
    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.
     */
825
    for (i = 0; i < 25; i++) {
826
        int signum = 0;
827
        if (i == 0) {
828
            signum = SIGTERM;
829
        } else if (i == 15) {
830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864
            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);
865
 cleanup:
866 867 868
    return ret;
}

J
Ján Tomko 已提交
869 870 871
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
G
Gene Czarcinski 已提交
872

873
static int
G
Gene Czarcinski 已提交
874 875
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
                                 virNetworkIpDefPtr ipdef)
876
{
877
    size_t i;
G
Gene Czarcinski 已提交
878
    bool ipv6 = false;
879

G
Gene Czarcinski 已提交
880 881
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
882 883
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
884
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
885 886
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
887
                return -1;
888
    }
889

G
Gene Czarcinski 已提交
890 891 892 893 894 895 896
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
897
    size_t i, j;
G
Gene Czarcinski 已提交
898

899 900
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
901
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
902
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
903
                for (j = 0; j < host->nnames; j++)
904 905
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
906 907
            }
        }
908 909
    }

910
    return 0;
911 912 913
}


914 915
int
networkDnsmasqConfContents(virNetworkObjPtr network,
916 917 918 919
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
920
{
921
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
922
    int r, ret = -1;
923
    int nbleases = 0;
924
    size_t i;
925
    virNetworkDNSDefPtr dns = &network->def->dns;
G
Gene Czarcinski 已提交
926 927
    virNetworkIpDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
    bool ipv6SLAAC;
928
    char *saddr = NULL, *eaddr = NULL;
929

930 931
    *configstr = NULL;

932
    /*
933 934 935
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
936
     *
937 938
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
939
     */
940 941 942 943 944 945

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

946 947
    /* create dnsmasq config file appropriate for this network */
    virBufferAsprintf(&configbuf,
948 949 950 951 952 953 954
                      "##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"
955
                      "strict-order\n",
956 957
                      network->def->name);

958 959 960 961
    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
962
                              network->def->dns.forwarders[i]);
963 964 965
        }
    }

966
    if (network->def->domain) {
967 968 969 970 971
        if (network->def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
                              network->def->domain);
        }
972
        virBufferAsprintf(&configbuf,
973 974 975 976
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
977

J
Ján Tomko 已提交
978
    if (network->def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
979 980 981 982
        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)
983
         */
984
        virBufferAddLit(&configbuf, "local=//\n");
985
    }
986

987
    if (pidfile)
988
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
989

990 991 992
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

993 994 995 996 997 998 999 1000
    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.
         */
1001
        virBufferAsprintf(&configbuf,
1002 1003 1004
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
1005
    } else {
1006
        virBufferAddLit(&configbuf, "bind-interfaces\n");
1007 1008 1009 1010 1011 1012 1013 1014
        /*
         * --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
         */
1015 1016 1017
        for (i = 0;
             (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
             i++) {
1018 1019 1020 1021
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
1022

1023
            /* also part of CVE 2012-3411 - if the host's version of
1024
             * dnsmasq doesn't have bind-dynamic, only allow listening on
1025 1026
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
1027 1028
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
1029 1030 1031 1032
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
1033
                                 "The version of dnsmasq on this host (%d.%d) "
1034 1035 1036 1037
                                 "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 "
1038 1039 1040 1041 1042 1043
                                 "(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);
1044
                VIR_FREE(ipaddr);
1045 1046
                goto cleanup;
            }
1047
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1048 1049 1050
            VIR_FREE(ipaddr);
        }
    }
1051

1052 1053
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1054
     * guaranteed to not work, and set no-resolv so that no dns
1055 1056 1057
     * 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).
1058
     */
1059
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1060
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1061
                        "no-resolv\n");
1062
    }
1063

1064
    for (i = 0; i < dns->ntxts; i++) {
1065
        virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
1066 1067
                          dns->txts[i].name,
                          dns->txts[i].value);
1068
    }
1069

1070
    for (i = 0; i < dns->nsrvs; i++) {
1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092
        /* 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);
1093

1094 1095 1096
        /* domain is optional - it defaults to the domain of this network */
        if (dns->srvs[i].domain)
            virBufferAsprintf(&configbuf, ".%s", dns->srvs[i].domain);
1097

1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119
        /* 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);
1120
        }
1121
        virBufferAddLit(&configbuf, "\n");
1122 1123
    }

G
Gene Czarcinski 已提交
1124
    /* Find the first dhcp for both IPv4 and IPv6 */
1125 1126 1127
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
1128 1129 1130 1131
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1132 1133
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144
                    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,
1145 1146 1147 1148 1149 1150 1151 1152 1153
                                   _("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
1154 1155 1156 1157
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1158 1159
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172
                    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 "
1173 1174 1175 1176
                 "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
1177 1178 1179 1180 1181
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1182 1183 1184 1185 1186 1187 1188 1189 1190
        int prefix;

        prefix = virNetworkIpDefPrefix(ipdef);
        if (prefix < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
            goto cleanup;
        }
1191
        for (r = 0; r < ipdef->nranges; r++) {
1192 1193
            int thisRange;

1194 1195
            if (!(saddr = virSocketAddrFormat(&ipdef->ranges[r].start)) ||
                !(eaddr = virSocketAddrFormat(&ipdef->ranges[r].end)))
1196
                goto cleanup;
1197

1198
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s",
1199
                              saddr, eaddr);
1200 1201 1202 1203
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
               virBufferAsprintf(&configbuf, ",%d", prefix);
            virBufferAddLit(&configbuf, "\n");

1204
            VIR_FREE(saddr);
1205
            VIR_FREE(eaddr);
1206
            thisRange = virSocketAddrGetRange(&ipdef->ranges[r].start,
1207 1208 1209
                                              &ipdef->ranges[r].end,
                                              &ipdef->address,
                                              virNetworkIpDefPrefix(ipdef));
1210 1211 1212
            if (thisRange < 0)
                goto cleanup;
            nbleases += thisRange;
1213
        }
1214

1215
        /*
1216 1217 1218 1219
         * 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)
1220 1221
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1222
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1223 1224
            if (!bridgeaddr)
                goto cleanup;
1225 1226 1227 1228 1229
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static",
                              bridgeaddr);
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
               virBufferAsprintf(&configbuf, ",%d", prefix);
            virBufferAddLit(&configbuf, "\n");
1230 1231
            VIR_FREE(bridgeaddr);
        }
1232

G
Gene Czarcinski 已提交
1233 1234
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1235

G
Gene Czarcinski 已提交
1236 1237 1238
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts)
1239
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1240

G
Gene Czarcinski 已提交
1241
            if (ipdef->tftproot) {
1242 1243
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1244
            }
1245

G
Gene Czarcinski 已提交
1246 1247 1248
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1249

1250
                    if (!bootserver)
G
Gene Czarcinski 已提交
1251
                        goto cleanup;
1252
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1253
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1254 1255
                    VIR_FREE(bootserver);
                } else {
1256
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1257 1258 1259 1260 1261
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1262

1263
    if (nbleases > 0)
1264
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1265

G
Gene Czarcinski 已提交
1266 1267
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1268
        goto cleanup;
G
Gene Czarcinski 已提交
1269 1270 1271 1272 1273 1274

    /* 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)
1275 1276
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1277

1278 1279
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1280
     */
1281
    virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
1282
                      dctx->addnhostsfile->path);
G
Gene Czarcinski 已提交
1283 1284 1285

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1286
        if (ipv6def) {
1287
            virBufferAddLit(&configbuf, "enable-ra\n");
1288
        } else {
1289 1290 1291
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
                 i++) {
G
Gene Czarcinski 已提交
1292 1293 1294 1295
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1296 1297
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1298 1299
                    VIR_FREE(bridgeaddr);
                }
1300
            }
1301
        }
1302 1303
    }

1304 1305 1306
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1307
    ret = 0;
G
Gene Czarcinski 已提交
1308

1309
 cleanup:
1310 1311
    VIR_FREE(saddr);
    VIR_FREE(eaddr);
1312
    virBufferFreeAndReset(&configbuf);
1313
    return ret;
1314 1315
}

1316
/* build the dnsmasq command line */
1317 1318 1319
static int ATTRIBUTE_NONNULL(3)
networkBuildDhcpDaemonCommandLine(virNetworkDriverStatePtr driver,
                                  virNetworkObjPtr network,
1320
                                  virCommandPtr *cmdout,
1321 1322
                                  char *pidfile,
                                  dnsmasqContext *dctx)
1323
{
1324
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1325
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1326
    int ret = -1;
1327 1328
    char *configfile = NULL;
    char *configstr = NULL;
1329
    char *leaseshelper_path = NULL;
1330 1331

    network->dnsmasqPid = -1;
1332

1333
    if (networkDnsmasqConfContents(network, pidfile, &configstr,
1334
                                   dctx, dnsmasq_caps) < 0)
1335 1336 1337 1338 1339
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1340
    if (!(configfile = networkDnsmasqConfigFileName(driver, network->def->name)))
1341 1342 1343 1344 1345
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1346 1347
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1348 1349 1350
        goto cleanup;
    }

1351 1352
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1353
                                                  abs_topbuilddir "/src",
1354 1355 1356
                                                  LIBEXECDIR)))
        goto cleanup;

1357
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(dnsmasq_caps));
1358
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1359 1360
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1361
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1362
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", network->def->bridge);
1363

1364
    *cmdout = cmd;
1365
    ret = 0;
1366
 cleanup:
1367
    virObjectUnref(dnsmasq_caps);
1368 1369
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1370
    VIR_FREE(leaseshelper_path);
1371 1372 1373 1374
    return ret;
}

static int
1375 1376
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
1377 1378 1379 1380
{
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1381
    dnsmasqContext *dctx = NULL;
1382

1383
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0)) {
G
Gene Czarcinski 已提交
1384
        /* no IP addresses, so we don't need to run */
1385 1386 1387 1388
        ret = 0;
        goto cleanup;
    }

1389
    if (virFileMakePath(driver->pidDir) < 0) {
1390
        virReportSystemError(errno,
1391
                             _("cannot create directory %s"),
1392
                             driver->pidDir);
1393
        goto cleanup;
1394 1395
    }

1396
    if (!(pidfile = virPidFileBuildPath(driver->pidDir,
1397
                                        network->def->name)))
1398
        goto cleanup;
1399

1400
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1401
        virReportSystemError(errno,
1402
                             _("cannot create directory %s"),
1403
                             driver->dnsmasqStateDir);
1404 1405 1406
        goto cleanup;
    }

1407
    dctx = dnsmasqContextNew(network->def->name, driver->dnsmasqStateDir);
1408 1409 1410
    if (dctx == NULL)
        goto cleanup;

1411
    if (networkDnsmasqCapsRefresh(driver) < 0)
1412
        goto cleanup;
1413

1414
    ret = networkBuildDhcpDaemonCommandLine(driver, network, &cmd, pidfile, dctx);
1415 1416 1417 1418 1419
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1420
        goto cleanup;
1421

G
Guido Günther 已提交
1422
    ret = virCommandRun(cmd, NULL);
1423
    if (ret < 0)
1424 1425 1426
        goto cleanup;

    /*
1427 1428 1429 1430 1431
     * 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
1432 1433
     */

1434
    ret = virPidFileRead(driver->pidDir, network->def->name,
1435 1436
                         &network->dnsmasqPid);
    if (ret < 0)
1437
        goto cleanup;
1438

1439
    ret = 0;
1440
 cleanup:
1441
    VIR_FREE(pidfile);
1442
    virCommandFree(cmd);
1443
    dnsmasqContextFree(dctx);
1444 1445 1446
    return ret;
}

1447 1448
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1449 1450
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1451 1452 1453
 *
 *  Returns 0 on success, -1 on failure.
 */
1454
static int
1455 1456
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1457
{
1458 1459
    int ret = -1;
    size_t i;
G
Gene Czarcinski 已提交
1460
    virNetworkIpDefPtr ipdef, ipv4def, ipv6def;
1461
    dnsmasqContext *dctx = NULL;
1462

G
Gene Czarcinski 已提交
1463
    /* if no IP addresses specified, nothing to do */
1464
    if (!virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1465 1466
        return 0;

1467 1468
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1469
        return networkStartDhcpDaemon(driver, network);
1470

G
Gene Czarcinski 已提交
1471
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1472
    if (!(dctx = dnsmasqContextNew(network->def->name,
1473
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1474
        goto cleanup;
1475
    }
G
Gene Czarcinski 已提交
1476 1477 1478 1479 1480 1481

    /* 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;
1482 1483 1484
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
G
Gene Czarcinski 已提交
1485 1486
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1487 1488
    }

G
Gene Czarcinski 已提交
1489
    ipv6def = NULL;
1490 1491 1492
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1493 1494
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1495 1496
    }

G
Gene Czarcinski 已提交
1497
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1498
        goto cleanup;
G
Gene Czarcinski 已提交
1499 1500

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1501
        goto cleanup;
1502

G
Gene Czarcinski 已提交
1503
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
J
Ján Tomko 已提交
1504
        goto cleanup;
1505 1506

    if ((ret = dnsmasqSave(dctx)) < 0)
1507
        goto cleanup;
1508 1509

    ret = kill(network->dnsmasqPid, SIGHUP);
1510
 cleanup:
1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522
    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
1523 1524
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1525 1526 1527 1528 1529 1530
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1531
    }
1532
    /* now start dnsmasq if it should be started */
1533
    return networkStartDhcpDaemon(driver, network);
1534 1535
}

G
Gene Czarcinski 已提交
1536 1537 1538 1539 1540 1541
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";

1542 1543 1544
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1545
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1546 1547
    int ret = -1;
    size_t i;
1548
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
1549
    bool v6present = false, dhcp6 = false;
1550 1551

    *configstr = NULL;
1552

G
Gene Czarcinski 已提交
1553
    /* Check if DHCPv6 is needed */
1554 1555 1556
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
G
Gene Czarcinski 已提交
1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569
        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;
    }

1570 1571 1572
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1573
    virBufferAsprintf(&configbuf, "interface %s\n"
1574 1575
                      "{\n"
                      "  AdvSendAdvert on;\n"
1576
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1577 1578 1579 1580 1581
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1582 1583

    /* add a section for each IPv6 address in the config */
1584 1585 1586
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET6, i));
         i++) {
1587 1588 1589 1590 1591
        int prefix;
        char *netaddr;

        prefix = virNetworkIpDefPrefix(ipdef);
        if (prefix < 0) {
1592 1593 1594
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1595 1596
            goto cleanup;
        }
1597
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1598
            goto cleanup;
1599
        virBufferAsprintf(&configbuf,
1600
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1601 1602 1603
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1604 1605 1606
        VIR_FREE(netaddr);
    }

1607
    virBufferAddLit(&configbuf, "};\n");
1608

1609
    if (virBufferCheckError(&configbuf) < 0)
1610
        goto cleanup;
1611

1612 1613
    *configstr = virBufferContentAndReset(&configbuf);

1614
    ret = 0;
1615
 cleanup:
1616 1617 1618 1619
    virBufferFreeAndReset(&configbuf);
    return ret;
}

1620
/* write file and return its name (which must be freed by caller) */
1621
static int
1622 1623 1624
networkRadvdConfWrite(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr network,
                      char **configFile)
1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639
{
    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;
1640 1641 1642 1643
        goto cleanup;
    }

    /* construct the filename */
1644
    if (!(*configFile = networkRadvdConfigFileName(driver, network->def->name)))
1645 1646
        goto cleanup;
    /* write the file */
1647
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1648 1649
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1650 1651 1652 1653 1654
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1655
 cleanup:
1656 1657 1658 1659 1660 1661
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1662 1663
networkStartRadvd(virNetworkDriverStatePtr driver,
                  virNetworkObjPtr network)
1664
{
1665
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1666 1667 1668 1669 1670 1671 1672 1673
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1674
    /* Is dnsmasq handling RA? */
1675
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
G
Gene Czarcinski 已提交
1676 1677 1678 1679
        ret = 0;
        goto cleanup;
    }

1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690
    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);
1691 1692 1693
        goto cleanup;
    }

1694
    if (virFileMakePath(driver->pidDir) < 0) {
1695 1696
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1697
                             driver->pidDir);
1698 1699
        goto cleanup;
    }
1700
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1701 1702
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1703
                             driver->radvdStateDir);
1704 1705 1706 1707
        goto cleanup;
    }

    /* construct pidfile name */
1708
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1709
        goto cleanup;
1710
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1711 1712
        goto cleanup;

1713
    if (networkRadvdConfWrite(driver, network, &configfile) < 0)
1714 1715
        goto cleanup;

1716 1717 1718 1719
    /* 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
1720
     * virPidFileRead() below will fail if we use them).
1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735
     * 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;

1736
    if (virPidFileRead(driver->pidDir, radvdpidbase, &network->radvdPid) < 0)
1737 1738 1739
        goto cleanup;

    ret = 0;
1740
 cleanup:
1741
    virObjectUnref(dnsmasq_caps);
1742 1743 1744 1745 1746 1747 1748
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1749
static int
1750 1751
networkRefreshRadvd(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
1752
{
1753
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
G
Gene Czarcinski 已提交
1754 1755 1756
    char *radvdpidbase;

    /* Is dnsmasq handling RA? */
1757 1758
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
        virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1759 1760 1761 1762 1763 1764 1765
        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)) {
1766
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1767 1768 1769 1770 1771
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }
1772
    virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1773

1774 1775
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
1776
        return networkStartRadvd(driver, network);
1777 1778 1779 1780 1781 1782

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

1783
    if (networkRadvdConfWrite(driver, network, NULL) < 0)
1784 1785 1786 1787 1788
        return -1;

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

1789 1790
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1791
static int
1792
networkRestartRadvd(virNetworkObjPtr network)
1793 1794 1795 1796 1797 1798 1799 1800 1801
{
    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
1802
        if ((networkKillDaemon(network->radvdPid, "radvd",
1803 1804 1805
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1806
            virPidFileDelete(driver->pidDir, radvdpidbase);
1807 1808 1809 1810 1811 1812 1813 1814 1815
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1816 1817
static int
networkRefreshDaemonsHelper(virNetworkObjPtr net,
1818
                            void *opaque)
1819
{
1820
    virNetworkDriverStatePtr driver = opaque;
1821

1822
    virObjectLock(net);
1823 1824 1825 1826 1827 1828 1829 1830 1831 1832
    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.
         */
1833 1834
        networkRefreshDhcpDaemon(driver, net);
        networkRefreshRadvd(driver, net);
1835
    }
1836
    virObjectUnlock(net);
1837 1838 1839
    return 0;
}

1840 1841 1842 1843
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
1844
networkRefreshDaemons(virNetworkDriverStatePtr driver)
1845 1846
{
    VIR_INFO("Refreshing network daemons");
1847 1848
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
1849
                             driver);
1850
}
1851

1852 1853 1854 1855 1856
static int
networkReloadFirewallRulesHelper(virNetworkObjPtr net,
                                 void *opaque ATTRIBUTE_UNUSED)
{

1857
    virObjectLock(net);
1858 1859 1860 1861 1862 1863 1864 1865 1866 1867
    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 */
1868 1869
        }
    }
1870
    virObjectUnlock(net);
1871
    return 0;
1872 1873
}

1874
static void
1875
networkReloadFirewallRules(virNetworkDriverStatePtr driver)
1876
{
1877
    VIR_INFO("Reloading iptables rules");
1878 1879 1880
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
1881 1882
}

1883
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
1884
static int
1885
networkEnableIpForwarding(bool enableIPv4, bool enableIPv6)
1886
{
1887
    int ret = 0;
1888 1889 1890 1891
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
1892
                           &enabled, sizeof(enabled));
1893 1894
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
1895
                           &enabled, sizeof(enabled));
1896
#else
1897 1898 1899 1900
    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);
1901
#endif
1902
    return ret;
1903 1904
}

1905 1906
#define SYSCTL_PATH "/proc/sys"

1907 1908
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
1909 1910 1911
{
    char *field = NULL;
    int ret = -1;
1912
    bool enableIPv6 =  !!virNetworkDefGetIpByIndex(network->def, AF_INET6, 0);
1913

1914 1915 1916 1917 1918 1919 1920
    /* 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;
1921

1922 1923
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
1924 1925
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
1926 1927 1928
        ret = 0;
        goto cleanup;
    }
1929

1930 1931 1932 1933 1934
    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;
1935
    }
1936
    VIR_FREE(field);
1937

1938 1939
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
1940 1941 1942 1943 1944 1945
     */

    /* 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",
1946
                    network->def->bridge) < 0)
1947 1948
        goto cleanup;

1949
    if (virFileWriteStr(field, "0", 0) < 0) {
1950
        virReportSystemError(errno,
1951 1952 1953 1954 1955
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

1956 1957 1958 1959
    /* 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",
1960
                    network->def->bridge) < 0)
1961 1962
        goto cleanup;

1963
    if (virFileWriteStr(field, "0", 0) < 0) {
1964
        virReportSystemError(errno,
1965
                             _("cannot disable %s"), field);
1966 1967 1968 1969
        goto cleanup;
    }

    ret = 0;
1970
 cleanup:
1971 1972 1973 1974
    VIR_FREE(field);
    return ret;
}

1975
/* add an IP address to a bridge */
1976
static int
D
Daniel P. Berrange 已提交
1977
networkAddAddrToBridge(virNetworkObjPtr network,
1978
                       virNetworkIpDefPtr ipdef)
1979
{
1980 1981 1982
    int prefix = virNetworkIpDefPrefix(ipdef);

    if (prefix < 0) {
1983 1984 1985
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
1986 1987 1988
        return -1;
    }

1989
    if (virNetDevSetIPAddress(network->def->bridge,
1990
                              &ipdef->address, NULL, prefix) < 0)
1991 1992 1993 1994 1995
        return -1;

    return 0;
}

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

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;
}


2019 2020 2021 2022 2023
/* add an IP (static) route to a bridge */
static int
networkAddRouteToBridge(virNetworkObjPtr network,
                        virNetworkRouteDefPtr routedef)
{
2024 2025 2026 2027
    int prefix = virNetworkRouteDefGetPrefix(routedef);
    unsigned int metric = virNetworkRouteDefGetMetric(routedef);
    virSocketAddrPtr addr = virNetworkRouteDefGetAddress(routedef);
    virSocketAddrPtr gateway = virNetworkRouteDefGetGateway(routedef);
2028 2029 2030 2031 2032 2033 2034 2035 2036

    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;
    }

2037 2038
    if (virNetDevAddRoute(network->def->bridge, addr,
                          prefix, gateway, metric) < 0) {
2039 2040 2041 2042 2043
        return -1;
    }
    return 0;
}

2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069
static int
networkWaitDadFinish(virNetworkObjPtr network)
{
    virNetworkIpDefPtr ipdef;
    virSocketAddrPtr *addrs = NULL, addr = NULL;
    size_t naddrs = 0;
    int ret = -1;

    VIR_DEBUG("Begin waiting for IPv6 DAD on network %s", network->def->name);

    while ((ipdef = virNetworkDefGetIpByIndex(network->def,
                                              AF_INET6, naddrs))) {
        addr = &ipdef->address;
        if (VIR_APPEND_ELEMENT_COPY(addrs, naddrs, addr) < 0)
            goto cleanup;
    }

    ret = (naddrs == 0) ? 0 : virNetDevWaitDadFinish(addrs, naddrs);

 cleanup:
    VIR_FREE(addrs);
    VIR_DEBUG("Finished waiting for IPv6 DAD on network %s with status %d",
              network->def->name, ret);
    return ret;
}

2070
static int
2071 2072
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
                           virNetworkObjPtr network)
2073
{
2074
    size_t i;
2075
    bool v4present = false, v6present = false;
2076 2077
    virErrorPtr save_err = NULL;
    virNetworkIpDefPtr ipdef;
2078
    virNetworkRouteDefPtr routedef;
2079
    char *macTapIfName = NULL;
2080
    int tapfd = -1;
2081

2082
    /* Check to see if any network IP collides with an existing route */
2083
    if (networkCheckRouteCollision(network->def) < 0)
2084 2085
        return -1;

2086
    /* Create and configure the bridge device */
2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100
    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;
    }
2101
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
2102 2103
        return -1;

2104 2105 2106 2107 2108 2109 2110 2111
    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);
2112
        if (!macTapIfName)
2113
            goto err0;
2114
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2115
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
2116
                                           &macTapIfName, &network->def->mac,
2117
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2118 2119 2120
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2121 2122 2123 2124 2125
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2126
    /* Set bridge options */
2127 2128 2129 2130

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2131
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2132
                                   network->def->delay * 1000) < 0)
2133
        goto err1;
2134

2135
    if (virNetDevBridgeSetSTP(network->def->bridge,
2136
                              network->def->stp ? true : false) < 0)
2137
        goto err1;
2138

2139 2140 2141 2142
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2143
        goto err1;
2144

2145
    /* Add "once per network" rules */
2146
    if (networkAddFirewallRules(network->def) < 0)
2147 2148
        goto err1;

2149 2150 2151
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, i));
         i++) {
2152
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2153
            v4present = true;
2154
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2155
            v6present = true;
2156

2157
        /* Add the IP address/netmask to the bridge */
2158
        if (networkAddAddrToBridge(network, ipdef) < 0)
2159
            goto err2;
2160 2161
    }

2162 2163 2164
    if (networkStartHandleMACTableManagerMode(network, macTapIfName) < 0)
        goto err2;

2165
    /* Bring up the bridge interface */
2166
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2167
        goto err2;
2168

2169
    for (i = 0; i < network->def->nroutes; i++) {
2170 2171 2172 2173 2174
        virSocketAddrPtr gateway = NULL;

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

2175 2176 2177
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2178
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2179 2180 2181 2182 2183 2184 2185
            if (networkAddRouteToBridge(network, routedef) < 0) {
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2186 2187
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2188
        networkEnableIpForwarding(v4present, v6present) < 0) {
2189
        virReportSystemError(errno, "%s",
2190
                             _("failed to enable IP forwarding"));
2191
        goto err3;
2192 2193
    }

2194

2195
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2196
    if ((v4present || v6present) &&
2197
        networkStartDhcpDaemon(driver, network) < 0)
2198
        goto err3;
2199

2200
    /* start radvd if there are any ipv6 addresses */
2201
    if (v6present && networkStartRadvd(driver, network) < 0)
2202 2203
        goto err4;

2204 2205 2206 2207 2208 2209 2210 2211
    /* dnsmasq does not wait for DAD to complete before daemonizing,
     * so we need to wait for it ourselves.
     */
    if (v6present && networkWaitDadFinish(network) < 0)
        goto err4;

    /* DAD has finished, dnsmasq is now bound to the
     * bridge's IPv6 address, so we can set the dummy tun down.
2212 2213 2214 2215 2216 2217 2218
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
            goto err4;
        VIR_FORCE_CLOSE(tapfd);
    }

2219
    if (virNetDevBandwidthSet(network->def->bridge,
2220
                              network->def->bandwidth, true) < 0)
2221 2222
        goto err5;

2223
    VIR_FREE(macTapIfName);
2224 2225 2226

    return 0;

2227
 err5:
2228 2229
    if (network->def->bandwidth)
       virNetDevBandwidthClear(network->def->bridge);
2230

2231 2232 2233 2234
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2235 2236 2237 2238 2239
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2240 2241 2242
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2243
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2244

2245 2246 2247
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2248
    networkRemoveFirewallRules(network->def);
2249 2250

 err1:
2251 2252 2253
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2254
    if (macTapIfName) {
2255
        VIR_FORCE_CLOSE(tapfd);
2256
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2257 2258
        VIR_FREE(macTapIfName);
    }
2259 2260

 err0:
2261 2262
    if (!save_err)
        save_err = virSaveLastError();
2263
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2264

2265 2266 2267 2268
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2269
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2270 2271 2272
    return -1;
}

2273 2274 2275
static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
                              virNetworkObjPtr network)
2276
{
2277 2278
    if (network->def->bandwidth)
        virNetDevBandwidthClear(network->def->bridge);
2279

2280 2281 2282 2283 2284
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2285
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2286
            virPidFileDelete(driver->pidDir, radvdpidbase);
2287 2288 2289 2290
            VIR_FREE(radvdpidbase);
        }
    }

2291 2292 2293
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2294
    if (network->def->mac_specified) {
2295
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2296
        if (macTapIfName) {
2297
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2298 2299 2300 2301
            VIR_FREE(macTapIfName);
        }
    }

2302
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2303

2304
    networkRemoveFirewallRules(network->def);
2305

2306
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2307

2308
    /* See if its still alive and really really kill it */
2309
    if (network->dnsmasqPid > 0 &&
2310
        (kill(network->dnsmasqPid, 0) == 0))
2311 2312
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2313 2314 2315 2316 2317 2318

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

2319 2320 2321
    return 0;
}

2322

2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343
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;
}


2344 2345 2346 2347 2348 2349 2350 2351 2352
/* 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;
2353
    unsigned int maxVirtFns = 0;
2354 2355 2356 2357 2358 2359
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2360 2361 2362
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2363 2364
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev, &vfNames,
                                      &virtFns, &numVirtFns, &maxVirtFns)) < 0) {
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 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450
        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;
}


2451
static int
2452
networkStartNetworkExternal(virNetworkObjPtr network)
2453 2454
{
    /* put anything here that needs to be done each time a network of
2455
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2456 2457 2458
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2459
    return networkCreateInterfacePool(network->def);
2460 2461
}

2462
static int networkShutdownNetworkExternal(virNetworkObjPtr network ATTRIBUTE_UNUSED)
2463 2464
{
    /* put anything here that needs to be done each time a network of
2465
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2466 2467 2468 2469 2470 2471 2472
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2473 2474
networkStartNetwork(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
2475
{
2476 2477 2478
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2479 2480

    if (virNetworkObjIsActive(network)) {
2481 2482
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2483
        return ret;
2484 2485
    }

2486 2487 2488
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2489
    if (virNetworkObjSetDefTransient(network, true) < 0)
2490
        goto cleanup;
2491

2492 2493
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2494
    if (networkRunHook(network, NULL, NULL,
2495 2496 2497 2498
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2499
    switch (network->def->forward.type) {
2500 2501 2502 2503

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2504
        if (networkStartNetworkVirtual(driver, network) < 0)
2505
            goto cleanup;
2506 2507 2508
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2509 2510 2511 2512 2513 2514 2515 2516 2517
        if (network->def->bridge) {
            if (networkStartNetworkBridge(network) < 0)
                goto cleanup;
            break;
        }
        /* intentionally fall through to the macvtap/direct case for
         * VIR_NETWORK_FORWARD_BRIDGE with no bridge device defined
         * (since that is macvtap bridge mode).
         */
2518 2519 2520
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2521
    case VIR_NETWORK_FORWARD_HOSTDEV:
2522
        if (networkStartNetworkExternal(network) < 0)
2523
            goto cleanup;
2524 2525 2526
        break;
    }

2527
    /* finally we can call the 'started' hook script if any */
2528
    if (networkRunHook(network, NULL, NULL,
2529 2530 2531 2532
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2533 2534 2535
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2536
    VIR_DEBUG("Writing network status to disk");
2537
    if (virNetworkSaveStatus(driver->stateDir, network) < 0)
2538
        goto cleanup;
2539 2540

    network->active = 1;
2541 2542
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2543

2544
 cleanup:
2545
    if (ret < 0) {
2546
        virNetworkObjUnsetDefTransient(network);
2547 2548
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2549
        networkShutdownNetwork(driver, network);
2550 2551 2552 2553 2554 2555 2556
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2557 2558 2559
static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
2560 2561 2562 2563 2564 2565 2566 2567 2568
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

2569
    stateFile = virNetworkConfigFile(driver->stateDir,
2570
                                     network->def->name);
2571 2572 2573 2574 2575 2576
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2577
    switch (network->def->forward.type) {
2578 2579 2580 2581

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2582
        ret = networkShutdownNetworkVirtual(driver, network);
2583 2584 2585
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2586 2587 2588 2589 2590 2591 2592 2593
        if (network->def->bridge) {
            ret = networkShutdownNetworkBridge(network);
            break;
        }
        /* intentionally fall through to the macvtap/direct case for
         * VIR_NETWORK_FORWARD_BRIDGE with no bridge device defined
         * (since that is macvtap bridge mode).
         */
2594 2595 2596
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2597
    case VIR_NETWORK_FORWARD_HOSTDEV:
2598
        ret = networkShutdownNetworkExternal(network);
2599 2600 2601
        break;
    }

2602
    /* now that we know it's stopped call the hook if present */
2603
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2604 2605
                   VIR_HOOK_SUBOP_END);

2606
    network->active = 0;
2607
    virNetworkObjUnsetDefTransient(network);
2608
    return ret;
2609 2610 2611
}


2612
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2613 2614
                                         const unsigned char *uuid)
{
2615
    virNetworkDriverStatePtr driver = networkGetDriver();
2616 2617
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2618

2619
    network = virNetworkObjFindByUUID(driver->networks, uuid);
2620
    if (!network) {
2621 2622
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2623
        virReportError(VIR_ERR_NO_NETWORK,
2624 2625
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2626
        goto cleanup;
2627 2628
    }

2629 2630 2631
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2632 2633
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2634
 cleanup:
2635
    virNetworkObjEndAPI(&network);
2636
    return ret;
2637 2638
}

2639
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2640 2641
                                         const char *name)
{
2642
    virNetworkDriverStatePtr driver = networkGetDriver();
2643 2644 2645
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2646
    network = virNetworkObjFindByName(driver->networks, name);
2647
    if (!network) {
2648 2649
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2650
        goto cleanup;
2651 2652
    }

2653 2654 2655
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2656 2657
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2658
 cleanup:
2659
    virNetworkObjEndAPI(&network);
2660
    return ret;
2661 2662
}

2663 2664
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2665
    virNetworkDriverStatePtr driver = networkGetDriver();
2666
    int nactive;
2667

2668 2669 2670
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2671 2672 2673 2674
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
2675

2676 2677 2678
    return nactive;
}

2679 2680 2681 2682 2683
static int networkConnectListNetworks(virConnectPtr conn,
                                      char **const names,
                                      int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2684
    int got = 0;
2685

2686 2687 2688
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2689 2690 2691 2692
    got = virNetworkObjListGetNames(driver->networks,
                                    true, names, nnames,
                                    virConnectListNetworksCheckACL,
                                    conn);
2693

2694 2695 2696
    return got;
}

2697 2698
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2699
    virNetworkDriverStatePtr driver = networkGetDriver();
2700
    int ninactive = 0;
2701

2702 2703 2704
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2705 2706 2707 2708
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
2709

2710 2711 2712
    return ninactive;
}

2713 2714 2715
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2716
    int got = 0;
2717

2718 2719 2720
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2721 2722 2723 2724
    got = virNetworkObjListGetNames(driver->networks,
                                    false, names, nnames,
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
2725 2726 2727
    return got;
}

2728
static int
2729 2730 2731
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2732
{
2733
    virNetworkDriverStatePtr driver = networkGetDriver();
2734 2735 2736 2737
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2738 2739 2740
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2741
    ret = virNetworkObjListExport(conn, driver->networks, nets,
2742 2743
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2744

2745
 cleanup:
2746 2747
    return ret;
}
2748

2749 2750 2751 2752 2753 2754 2755 2756
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
2757
    virNetworkDriverStatePtr driver = networkGetDriver();
2758 2759 2760 2761 2762 2763
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
2764
                                       net, eventID, callback,
2765 2766 2767
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2768
 cleanup:
2769 2770 2771 2772 2773 2774 2775
    return ret;
}

static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
2776
    virNetworkDriverStatePtr driver = networkGetDriver();
2777 2778 2779 2780 2781
    int ret = -1;

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

2782 2783 2784 2785 2786 2787
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
2788

2789
 cleanup:
2790 2791 2792
    return ret;
}

2793 2794 2795 2796 2797
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2798 2799
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2800 2801 2802 2803

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

2804 2805
    ret = virNetworkObjIsActive(obj);

2806
 cleanup:
2807
    virNetworkObjEndAPI(&obj);
2808 2809 2810 2811 2812 2813 2814 2815
    return ret;
}

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

2816 2817
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2818 2819 2820 2821

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

2822 2823
    ret = obj->persistent;

2824
 cleanup:
2825
    virNetworkObjEndAPI(&obj);
2826 2827 2828 2829
    return ret;
}


2830 2831
/*
 * networkFindUnusedBridgeName() - try to find a bridge name that is
2832 2833 2834
 * 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.
2835 2836 2837 2838 2839 2840 2841 2842
 */
static int
networkFindUnusedBridgeName(virNetworkObjListPtr nets,
                            virNetworkDefPtr def)
{

    int ret = -1, id = 0;
    char *newname = NULL;
2843 2844 2845 2846 2847
    const char *templ = "virbr%d";
    const char *p;

    if (def->bridge &&
        (p = strchr(def->bridge, '%')) == strrchr(def->bridge, '%') &&
2848
        p && p[1] == 'd')
2849
        templ = def->bridge;
2850 2851 2852 2853

    do {
        if (virAsprintf(&newname, templ, id) < 0)
            goto cleanup;
2854 2855 2856 2857 2858 2859 2860
        /* 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)) {
2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912
            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;
}


2913
static int
2914
networkValidate(virNetworkDriverStatePtr driver,
2915
                virNetworkDefPtr def)
2916
{
2917
    size_t i, j;
2918 2919
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2920
    virNetworkIpDefPtr ipdef;
G
Gene Czarcinski 已提交
2921
    bool ipv4def = false, ipv6def = false;
2922
    bool bandwidthAllowed = true;
2923
    bool usesInterface = false, usesAddress = false;
2924 2925 2926 2927

    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
2928 2929 2930
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2931

2932 2933 2934 2935
        /* 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)
2936 2937 2938
            return -1;

        virNetworkSetBridgeMacAddr(def);
2939 2940
    } else {
        /* They are also the only types that currently support setting
2941 2942
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
2943
         */
2944 2945 2946 2947 2948 2949 2950 2951
        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;
        }
2952 2953 2954 2955 2956
        if (virNetworkDefGetIpByIndex(def, AF_UNSPEC, 0)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2957
                           virNetworkForwardTypeToString(def->forward.type));
2958 2959
            return -1;
        }
2960
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
2961 2962 2963 2964
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2965
                           virNetworkForwardTypeToString(def->forward.type));
2966 2967 2968 2969 2970 2971 2972
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2973
                           virNetworkForwardTypeToString(def->forward.type));
2974 2975
            return -1;
        }
2976 2977 2978 2979 2980 2981 2982 2983
        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;
        }
2984
        bandwidthAllowed = false;
2985 2986
    }

2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020
    /* 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
3021 3022 3023
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
3024 3025 3026
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(def, AF_UNSPEC, i));
         i++) {
G
Gene Czarcinski 已提交
3027 3028 3029 3030 3031
        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 -- "
3032 3033
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050
                    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;
                }
3051 3052 3053
            }
        }
    }
3054 3055 3056 3057 3058 3059

    /* 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.
     */

3060 3061 3062
    vlanAllowed = (def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV ||
                   def->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH ||
                   (def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
J
Ján Tomko 已提交
3063 3064
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
3065
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH));
3066 3067

    vlanUsed = def->vlan.nTags > 0;
3068 3069
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3070 3071 3072 3073 3074
            /* 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.
             */
3075
            if (def->portGroups[i].virtPortProfile) {
3076
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3077
                    def->portGroups[i].virtPortProfile->virtPortType
3078 3079 3080 3081 3082 3083 3084
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3085
        }
3086
        if (def->portGroups[i].isDefault) {
3087 3088 3089 3090 3091
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3092
                               def->name, defaultPortGroup->name,
3093
                               def->portGroups[i].name);
3094
                return -1;
3095
            }
3096
            defaultPortGroup = &def->portGroups[i];
3097
        }
3098 3099 3100 3101 3102 3103 3104 3105 3106
        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;
            }
        }
3107 3108 3109 3110 3111 3112 3113 3114
        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;
        }
3115
    }
3116 3117 3118 3119 3120 3121 3122
    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.
         */
3123 3124 3125 3126 3127 3128 3129 3130 3131
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
    return 0;
}

3132 3133
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
3134
    virNetworkDriverStatePtr driver = networkGetDriver();
3135
    virNetworkDefPtr def;
3136
    virNetworkObjPtr network = NULL;
3137
    virNetworkPtr ret = NULL;
3138
    virObjectEventPtr event = NULL;
3139

3140
    if (!(def = virNetworkDefParseString(xml)))
3141
        goto cleanup;
3142

3143 3144 3145
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3146
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3147
        goto cleanup;
3148

3149 3150 3151
    /* 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.
3152
     */
3153 3154 3155
    if (!(network = virNetworkAssignDef(driver->networks, def,
                                        VIR_NETWORK_OBJ_LIST_ADD_LIVE |
                                        VIR_NETWORK_OBJ_LIST_ADD_CHECK_LIVE)))
3156 3157
        goto cleanup;
    def = NULL;
3158

3159
    if (networkStartNetwork(driver, network) < 0) {
3160
        virNetworkRemoveInactive(driver->networks,
3161
                                 network);
3162
        goto cleanup;
3163 3164
    }

3165 3166
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3167 3168
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3169

3170
    VIR_INFO("Creating network '%s'", network->def->name);
3171 3172
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

3173
 cleanup:
3174
    virNetworkDefFree(def);
3175 3176
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3177
    virNetworkObjEndAPI(&network);
3178
    return ret;
3179 3180
}

3181 3182
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
3183
    virNetworkDriverStatePtr driver = networkGetDriver();
3184
    virNetworkDefPtr def = NULL;
3185
    bool freeDef = true;
3186
    virNetworkObjPtr network = NULL;
3187
    virNetworkPtr ret = NULL;
3188
    virObjectEventPtr event = NULL;
3189

3190
    if (!(def = virNetworkDefParseString(xml)))
3191
        goto cleanup;
3192

3193 3194 3195
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3196
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3197
        goto cleanup;
3198

3199
    if (!(network = virNetworkAssignDef(driver->networks, def, 0)))
J
Ján Tomko 已提交
3200
        goto cleanup;
3201

3202
    /* def was assigned to network object */
3203
    freeDef = false;
3204 3205

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3206
        if (!virNetworkObjIsActive(network)) {
3207
            virNetworkRemoveInactive(driver->networks, network);
3208 3209
            goto cleanup;
        }
3210 3211 3212 3213 3214
        /* 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);
3215 3216 3217
        goto cleanup;
    }

3218
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3219 3220
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3221

3222 3223
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
3224

3225
 cleanup:
3226 3227
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3228
    if (freeDef)
J
Ján Tomko 已提交
3229
        virNetworkDefFree(def);
3230
    virNetworkObjEndAPI(&network);
3231
    return ret;
3232 3233
}

3234
static int
3235 3236
networkUndefine(virNetworkPtr net)
{
3237
    virNetworkDriverStatePtr driver = networkGetDriver();
3238
    virNetworkObjPtr network;
3239
    int ret = -1;
3240
    bool active = false;
3241
    virObjectEventPtr event = NULL;
3242

3243
    if (!(network = networkObjFromNetwork(net)))
3244
        goto cleanup;
3245

3246 3247 3248
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3249 3250
    if (virNetworkObjIsActive(network))
        active = true;
3251

3252 3253 3254 3255 3256 3257
    if (!network->persistent) {
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                       _("can't undefine transient network"));
        goto cleanup;
    }

3258
    /* remove autostart link */
3259
    if (virNetworkDeleteConfig(driver->networkConfigDir,
3260 3261
                               driver->networkAutostartDir,
                               network) < 0)
3262
        goto cleanup;
3263

3264 3265
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3266 3267
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3268

3269
    VIR_INFO("Undefining network '%s'", network->def->name);
3270
    if (!active) {
3271
        if (networkRemoveInactive(driver, network) < 0)
3272
            goto cleanup;
3273 3274 3275 3276 3277 3278
    } 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);
3279 3280
    }

3281
    ret = 0;
3282

3283
 cleanup:
3284 3285
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3286
    virNetworkObjEndAPI(&network);
3287
    return ret;
3288 3289
}

3290 3291 3292 3293 3294 3295 3296 3297
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3298
    virNetworkDriverStatePtr driver = networkGetDriver();
3299
    virNetworkObjPtr network = NULL;
3300 3301
    int isActive, ret = -1;
    size_t i;
3302 3303
    virNetworkIpDefPtr ipdef;
    bool oldDhcpActive = false;
3304
    bool needFirewallRefresh = false;
3305

3306 3307 3308 3309 3310

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3311
    if (!(network = networkObjFromNetwork(net)))
3312 3313
        goto cleanup;

3314 3315 3316
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3317
    /* see if we are listening for dhcp pre-modification */
3318 3319 3320
    for (i = 0;
         (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
         i++) {
3321 3322 3323 3324 3325 3326
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3327 3328
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3329
     */
3330
    isActive = virNetworkObjIsActive(network);
3331 3332
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3333 3334 3335 3336 3337 3338 3339
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356
    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).
                 */
3357
                networkRemoveFirewallRules(network->def);
3358 3359 3360 3361 3362 3363 3364 3365
                needFirewallRefresh = true;
                break;
            default:
                break;
            }
        }
    }

3366
    /* update the network config in memory/on disk */
3367 3368
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3369
            ignore_value(networkAddFirewallRules(network->def));
3370 3371 3372
        goto cleanup;
    }

3373
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396
        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.
             */
3397
            if (networkRestartDhcpDaemon(driver, network) < 0)
3398 3399
                goto cleanup;

3400 3401 3402 3403 3404 3405 3406 3407
        } 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;

3408 3409 3410
            for (i = 0;
                 (ipdef = virNetworkDefGetIpByIndex(network->def, AF_INET, i));
                 i++) {
3411 3412 3413 3414 3415 3416 3417
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3418 3419
                 networkRestartDhcpDaemon(driver, network) < 0) ||
                networkRefreshDhcpDaemon(driver, network) < 0) {
3420 3421 3422 3423
                goto cleanup;
            }

        } else if (section == VIR_NETWORK_SECTION_DNS_HOST ||
3424 3425 3426 3427 3428 3429
                   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.
             */
3430
            if (networkRefreshDhcpDaemon(driver, network) < 0)
3431 3432 3433 3434 3435 3436 3437 3438
                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.
             */
3439
            if (networkRefreshRadvd(driver, network) < 0)
3440 3441 3442 3443
                goto cleanup;
        }

        /* save current network state to disk */
3444
        if ((ret = virNetworkSaveStatus(driver->stateDir,
3445
                                        network)) < 0) {
3446
            goto cleanup;
3447
        }
3448 3449
    }
    ret = 0;
3450
 cleanup:
3451
    virNetworkObjEndAPI(&network);
3452 3453 3454
    return ret;
}

3455 3456
static int networkCreate(virNetworkPtr net)
{
3457
    virNetworkDriverStatePtr driver = networkGetDriver();
3458 3459
    virNetworkObjPtr network;
    int ret = -1;
3460
    virObjectEventPtr event = NULL;
3461

3462
    if (!(network = networkObjFromNetwork(net)))
3463
        goto cleanup;
3464

3465 3466 3467
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3468
    if ((ret = networkStartNetwork(driver, network)) < 0)
3469
        goto cleanup;
3470

3471 3472
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3473 3474
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3475

3476
 cleanup:
3477 3478
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3479
    virNetworkObjEndAPI(&network);
3480
    return ret;
3481 3482
}

3483 3484
static int networkDestroy(virNetworkPtr net)
{
3485
    virNetworkDriverStatePtr driver = networkGetDriver();
3486 3487
    virNetworkObjPtr network;
    int ret = -1;
3488
    virObjectEventPtr event = NULL;
3489

3490
    if (!(network = networkObjFromNetwork(net)))
3491
        goto cleanup;
3492

3493 3494 3495
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3496
    if (!virNetworkObjIsActive(network)) {
3497
        virReportError(VIR_ERR_OPERATION_INVALID,
3498 3499
                       _("network '%s' is not active"),
                       network->def->name);
3500 3501 3502
        goto cleanup;
    }

3503
    if ((ret = networkShutdownNetwork(driver, network)) < 0)
3504 3505
        goto cleanup;

3506 3507
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3508 3509
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3510

3511
    if (!network->persistent &&
3512
        networkRemoveInactive(driver, network) < 0) {
3513 3514
        ret = -1;
        goto cleanup;
3515
    }
3516

3517
 cleanup:
3518 3519
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3520
    virNetworkObjEndAPI(&network);
3521 3522 3523
    return ret;
}

3524
static char *networkGetXMLDesc(virNetworkPtr net,
3525
                               unsigned int flags)
3526
{
3527
    virNetworkObjPtr network;
3528
    virNetworkDefPtr def;
3529
    char *ret = NULL;
3530

3531
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3532

3533 3534
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3535

3536 3537 3538
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3539 3540 3541 3542 3543 3544
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3545

3546
 cleanup:
3547
    virNetworkObjEndAPI(&network);
3548
    return ret;
3549 3550 3551
}

static char *networkGetBridgeName(virNetworkPtr net) {
3552 3553 3554
    virNetworkObjPtr network;
    char *bridge = NULL;

3555 3556
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3557

3558 3559 3560
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3561
    if (!(network->def->bridge)) {
3562 3563 3564
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3565 3566 3567
        goto cleanup;
    }

3568
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3569

3570
 cleanup:
3571
    virNetworkObjEndAPI(&network);
3572 3573 3574 3575
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
J
Ján Tomko 已提交
3576
                               int *autostart)
3577
{
3578 3579
    virNetworkObjPtr network;
    int ret = -1;
3580

3581 3582
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3583

3584 3585 3586
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3587
    *autostart = network->autostart;
3588
    ret = 0;
3589

3590
 cleanup:
3591
    virNetworkObjEndAPI(&network);
3592
    return ret;
3593 3594 3595
}

static int networkSetAutostart(virNetworkPtr net,
3596 3597
                               int autostart)
{
3598
    virNetworkDriverStatePtr driver = networkGetDriver();
3599
    virNetworkObjPtr network;
3600
    char *configFile = NULL, *autostartLink = NULL;
3601
    int ret = -1;
3602

3603

3604
    if (!(network = networkObjFromNetwork(net)))
3605
        goto cleanup;
3606

3607 3608 3609
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3610
    if (!network->persistent) {
3611 3612
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3613 3614 3615
        goto cleanup;
    }

3616 3617
    autostart = (autostart != 0);

3618
    if (network->autostart != autostart) {
3619
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3620
            goto cleanup;
3621
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3622 3623
            goto cleanup;

3624
        if (autostart) {
3625
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3626
                virReportSystemError(errno,
3627 3628
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3629 3630
                goto cleanup;
            }
3631

3632
            if (symlink(configFile, autostartLink) < 0) {
3633
                virReportSystemError(errno,
3634
                                     _("Failed to create symlink '%s' to '%s'"),
3635
                                     autostartLink, configFile);
3636 3637 3638
                goto cleanup;
            }
        } else {
3639
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3640
                virReportSystemError(errno,
3641
                                     _("Failed to delete symlink '%s'"),
3642
                                     autostartLink);
3643 3644
                goto cleanup;
            }
3645 3646
        }

3647
        network->autostart = autostart;
3648
    }
3649
    ret = 0;
3650

3651
 cleanup:
3652 3653
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3654
    virNetworkObjEndAPI(&network);
3655
    return ret;
3656 3657
}

3658
static int
3659 3660 3661 3662
networkGetDHCPLeases(virNetworkPtr network,
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
3663
{
3664
    virNetworkDriverStatePtr driver = networkGetDriver();
3665 3666 3667
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
3668
    ssize_t size = 0;
3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682
    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;
3683
    virNetworkObjPtr obj;
3684
    virMacAddr mac_addr;
3685 3686 3687

    virCheckFlags(0, -1);

3688 3689 3690 3691 3692 3693
    /* only to check if the MAC is valid */
    if (mac && virMacAddrParse(mac, &mac_addr) < 0) {
        virReportError(VIR_ERR_INVALID_MAC, "%s", mac);
        return -1;
    }

3694 3695 3696 3697 3698
    if (!(obj = networkObjFromNetwork(network)))
        return -1;

    if (virNetworkGetDHCPLeasesEnsureACL(network->conn, obj->def) < 0)
        goto cleanup;
3699 3700

    /* Retrieve custom leases file location */
3701
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(driver, obj->def->bridge);
3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744

    /* 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;
        }

3745
        if (mac && virMacAddrCompare(mac, mac_tmp))
3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795
            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) ||
3796
                (VIR_STRDUP(lease->iface, obj->def->bridge) < 0))
3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828
                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);
3829
    VIR_FREE(lease_entries);
3830 3831
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
3832

3833
    virNetworkObjEndAPI(&obj);
3834

3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845
    return rv;

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

3846 3847

static virNetworkDriver networkDriver = {
3848
    .name = "bridge",
3849 3850 3851 3852 3853
    .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 */
3854 3855
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
3856 3857
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3858 3859
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3860
    .networkUndefine = networkUndefine, /* 0.2.0 */
3861
    .networkUpdate = networkUpdate, /* 0.10.2 */
3862
    .networkCreate = networkCreate, /* 0.2.0 */
3863 3864 3865 3866 3867 3868 3869
    .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 */
3870
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
3871 3872 3873
};

static virStateDriver networkStateDriver = {
3874
    .name = "bridge",
3875
    .stateInitialize  = networkStateInitialize,
3876
    .stateAutoStart  = networkStateAutoStart,
3877 3878
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3879 3880
};

3881 3882
int networkRegister(void)
{
3883
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
3884
        return -1;
3885 3886
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
3887 3888
    return 0;
}
3889 3890 3891

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

3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926
/* A unified function to log network connections and disconnections */

static void
networkLogAllocation(virNetworkDefPtr netdef,
                     virDomainNetType actualType,
                     virNetworkForwardIfDefPtr dev,
                     virDomainNetDefPtr iface,
                     bool inUse)
{
    char macStr[VIR_MAC_STRING_BUFLEN];
    const char *verb = inUse ? "using" : "releasing";

    if (!dev) {
        VIR_INFO("MAC %s %s network %s (%d connections)",
                 virMacAddrFormat(&iface->mac, macStr), verb,
                 netdef->name, netdef->connections);
    } else {
        if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) {
            VIR_INFO("MAC %s %s network %s (%d connections) "
                     "physical device %04x:%02x:%02x.%x (%d connections)",
                     virMacAddrFormat(&iface->mac, macStr), verb,
                     netdef->name, netdef->connections,
                     dev->device.pci.domain, dev->device.pci.bus,
                     dev->device.pci.slot, dev->device.pci.function,
                     dev->connections);
        } else {
            VIR_INFO("MAC %s %s network %s (%d connections) "
                     "physical device %s (%d connections)",
                     virMacAddrFormat(&iface->mac, macStr), verb,
                     netdef->name, netdef->connections,
                     dev->device.dev, dev->connections);
        }
    }
}

3927 3928 3929 3930 3931 3932 3933 3934 3935
/* 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:
3936
 * @dom: domain definition that @iface belongs to
3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947
 * @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
3948 3949
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
3950
{
3951
    virNetworkDriverStatePtr driver = networkGetDriver();
3952
    virDomainNetType actualType = iface->type;
3953 3954
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
3955
    virNetDevBandwidthPtr bandwidth = NULL;
3956 3957 3958
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3959
    virNetworkForwardIfDefPtr dev = NULL;
3960
    size_t i;
3961 3962 3963
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3964
        goto validate;
3965 3966 3967 3968

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

3969
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
3970
    if (!network) {
3971 3972 3973
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3974
        goto error;
3975 3976
    }
    netdef = network->def;
3977

3978 3979 3980 3981 3982 3983 3984
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

3985 3986 3987
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

3988 3989 3990
    /* 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
3991
     */
3992 3993 3994 3995 3996 3997
    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.
     */
3998 3999 4000 4001 4002 4003

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

4004 4005
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
4006
        goto error;
4007

4008 4009 4010 4011 4012 4013 4014 4015
    /* 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;

4016 4017
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
4018

4019 4020 4021 4022 4023 4024 4025 4026 4027 4028
    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;

4029 4030 4031
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
4032
        /* for these forward types, the actual net type really *is*
4033
         * NETWORK; we just keep the info from the portgroup in
4034
         * iface->data.network.actual
J
Ján Tomko 已提交
4035
         */
4036
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
4037

4038
        /* we also store the bridge device and macTableManager settings
4039 4040 4041 4042 4043 4044 4045
         * 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;
4046 4047
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4048

4049 4050 4051
        if (networkPlugBandwidth(network, iface) < 0)
            goto error;

4052
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
4053
               netdef->bridge) {
4054 4055 4056 4057 4058

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

4059
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
4060 4061
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
4062
            goto error;
4063 4064
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4065

4066 4067 4068 4069 4070 4071 4072 4073
        /* 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) {
4074
            goto error;
4075 4076 4077 4078 4079 4080 4081 4082 4083 4084
        }
        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);
4085
                goto error;
4086 4087 4088
            }
        }

4089
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4090

4091
        virDomainHostdevSubsysPCIBackendType backend;
4092

4093
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
4094
        if (networkCreateInterfacePool(netdef) < 0)
4095 4096 4097
            goto error;

        /* pick first dev with 0 connections */
4098 4099 4100
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114
                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;
4115
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
4116
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
4117
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
4118

E
Eric Blake 已提交
4119
        switch (netdef->forward.driverName) {
4120
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
4121
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
4122 4123
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
4124
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
4125 4126
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
4127
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138
            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;

4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163
        /* 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;
            }
        }

4164 4165 4166 4167
    } 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)) {
4168 4169 4170 4171 4172 4173

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
4174
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4175
        switch (netdef->forward.type) {
4176
        case VIR_NETWORK_FORWARD_BRIDGE:
4177
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
4178 4179
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
4180
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
4181 4182
            break;
        case VIR_NETWORK_FORWARD_VEPA:
4183
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
4184 4185
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
4186
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
4187 4188 4189
            break;
        }

4190 4191 4192 4193 4194 4195 4196 4197
        /* 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) {
4198
            goto error;
4199
        }
4200
        virtport = iface->data.network.actual->virtPortProfile;
4201
        if (virtport) {
4202 4203 4204 4205 4206 4207 4208 4209
            /* 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);
4210
                goto error;
4211 4212
            }
        }
4213

4214 4215 4216
        /* 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).
         */
4217
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4218 4219 4220 4221
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4222
            goto error;
4223 4224 4225
        } else {
            /* pick an interface from the pool */

4226
            if (networkCreateInterfacePool(netdef) < 0)
4227 4228
                goto error;

4229 4230 4231 4232 4233
            /* 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.
4234
             */
4235 4236
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4237 4238 4239
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4240

4241
                /* pick first dev with 0 connections */
4242 4243 4244
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4245 4246 4247 4248 4249
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4250
                dev = &netdef->forward.ifs[0];
4251 4252 4253
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4254 4255 4256 4257
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4258 4259 4260 4261
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4262
                goto error;
4263
            }
4264 4265
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4266
                goto error;
4267 4268 4269
        }
    }

4270
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4271
        goto error;
4272

4273
 validate:
4274 4275 4276 4277 4278
    /* 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.
     */

4279
    if (virDomainNetGetActualVlan(iface)) {
4280 4281 4282 4283
        /* vlan configuration via libvirt is only supported for PCI
         * Passthrough SR-IOV devices (hostdev or macvtap passthru
         * mode) and openvswitch bridges. Otherwise log an error and
         * fail
4284 4285
         */
        if (!(actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV ||
4286 4287 4288
              (actualType == VIR_DOMAIN_NET_TYPE_DIRECT &&
               virDomainNetGetActualDirectMode(iface)
               == VIR_NETDEV_MACVLAN_MODE_PASSTHRU) ||
4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310
              (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++;
4311
        if (dev)
4312 4313 4314 4315 4316 4317
            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 */
4318
            netdef->connections--;
4319 4320 4321 4322
            if (dev)
                dev->connections--;
            goto error;
        }
4323
        networkLogAllocation(netdef, actualType, dev, iface, true);
4324 4325
    }

4326
    ret = 0;
4327

4328
 cleanup:
4329
    virNetworkObjEndAPI(&network);
4330 4331
    return ret;

4332
 error:
4333
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4334 4335 4336
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4337
    goto cleanup;
4338 4339 4340
}

/* networkNotifyActualDevice:
4341
 * @dom: domain definition that @iface belongs to
4342 4343 4344 4345 4346 4347 4348 4349 4350 4351
 * @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
4352 4353
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4354
{
4355
    virNetworkDriverStatePtr driver = networkGetDriver();
4356
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4357 4358
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4359
    virNetworkForwardIfDefPtr dev = NULL;
4360 4361
    size_t i;
    int ret = -1;
4362 4363

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4364
        return 0;
4365

4366
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4367
    if (!network) {
4368 4369 4370
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4371 4372 4373 4374
        goto error;
    }
    netdef = network->def;

4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385
    /* 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;

4386
    if (!iface->data.network.actual ||
4387 4388
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4389 4390
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4391 4392
    }

4393
    if (networkCreateInterfacePool(netdef) < 0)
4394
        goto error;
4395

4396
    if (netdef->forward.nifs == 0) {
4397
        virReportError(VIR_ERR_INTERNAL_ERROR,
4398 4399
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4400
                       netdef->name);
4401
        goto error;
4402
    }
4403

4404 4405
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4406

4407 4408 4409 4410 4411 4412 4413 4414 4415
        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 */
4416 4417
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4418
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4419 4420
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4421 4422 4423 4424 4425
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4426
            virReportError(VIR_ERR_INTERNAL_ERROR,
4427 4428
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4429
                           netdef->name, actualDev);
4430
            goto error;
4431 4432
        }

4433
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4434 4435
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4436
         */
4437
        if ((dev->connections > 0) &&
4438 4439
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4440 4441
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4442
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4443
            virReportError(VIR_ERR_INTERNAL_ERROR,
4444 4445
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4446
                           netdef->name, actualDev);
4447
            goto error;
4448
        }
4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460
    }  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 */
4461 4462
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4463
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4464
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4465 4466
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4467 4468 4469 4470 4471 4472 4473 4474 4475
                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,
4476 4477 4478 4479
                           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
4480
            goto error;
4481 4482 4483 4484 4485 4486 4487
        }

        /* 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) &&
4488
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4489 4490 4491 4492 4493 4494 4495 4496 4497
            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;
        }
4498 4499
    }

4500
 success:
4501
    netdef->connections++;
4502 4503
    if (dev)
        dev->connections++;
4504 4505 4506 4507 4508 4509 4510 4511 4512
    /* 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;
    }
4513
    networkLogAllocation(netdef, actualType, dev, iface, true);
4514

4515
    ret = 0;
4516
 cleanup:
4517
    virNetworkObjEndAPI(&network);
4518
    return ret;
4519

4520
 error:
4521
    goto cleanup;
4522 4523 4524
}


4525

4526
/* networkReleaseActualDevice:
4527
 * @dom: domain definition that @iface belongs to
4528 4529 4530 4531 4532 4533 4534 4535 4536 4537
 * @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
4538 4539
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4540
{
4541
    virNetworkDriverStatePtr driver = networkGetDriver();
4542
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4543
    virNetworkObjPtr network;
4544
    virNetworkDefPtr netdef;
4545
    virNetworkForwardIfDefPtr dev = NULL;
4546 4547
    size_t i;
    int ret = -1;
4548 4549

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4550
        return 0;
4551

4552
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4553
    if (!network) {
4554 4555 4556
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4557 4558 4559 4560
        goto error;
    }
    netdef = network->def;

4561 4562
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4563 4564 4565 4566 4567
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) &&
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4568 4569 4570
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4571 4572
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4573 4574
    }

4575
    if (netdef->forward.nifs == 0) {
4576
        virReportError(VIR_ERR_INTERNAL_ERROR,
4577
                       _("network '%s' uses a direct/hostdev mode, but "
4578 4579
                         "has no forward dev and no interface pool"),
                       netdef->name);
4580
        goto error;
4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592
    }

    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;
        }
4593

4594 4595
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4596
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4597 4598
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4599 4600 4601
                break;
            }
        }
4602

4603
        if (!dev) {
4604
            virReportError(VIR_ERR_INTERNAL_ERROR,
4605 4606
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4607
                           netdef->name, actualDev);
4608
            goto error;
4609
        }
4610 4611 4612 4613 4614 4615 4616 4617 4618 4619
    } 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;
        }

4620 4621
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4622
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4623
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4624 4625
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4626 4627 4628 4629 4630 4631 4632 4633 4634
                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,
4635 4636 4637 4638
                           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
4639
            goto error;
4640
        }
J
Ján Tomko 已提交
4641
    }
4642

4643
 success:
4644
    if (iface->data.network.actual) {
4645
        netdef->connections--;
4646 4647
        if (dev)
            dev->connections--;
4648 4649 4650
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
4651
        networkLogAllocation(netdef, actualType, dev, iface, false);
4652
    }
4653
    ret = 0;
4654
 cleanup:
4655
    virNetworkObjEndAPI(&network);
4656 4657 4658 4659
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4660
    return ret;
4661

4662
 error:
4663
    goto cleanup;
4664
}
4665 4666 4667 4668 4669 4670

/*
 * networkGetNetworkAddress:
 * @netname: the name of a network
 * @netaddr: string representation of IP address for that network.
 *
4671
 * Attempt to return an IP address associated with the named
4672 4673 4674 4675
 * 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.
 *
4676
 * Note: This function returns the first IP address it finds. It might
4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687
 * 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)
{
4688
    virNetworkDriverStatePtr driver = networkGetDriver();
4689
    int ret = -1;
4690
    virNetworkObjPtr network;
4691 4692 4693 4694
    virNetworkDefPtr netdef;
    virNetworkIpDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4695
    char *dev_name = NULL;
4696 4697

    *netaddr = NULL;
4698
    network = virNetworkObjFindByName(driver->networks, netname);
4699
    if (!network) {
4700 4701 4702
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4703
        goto cleanup;
4704 4705 4706
    }
    netdef = network->def;

4707
    switch (netdef->forward.type) {
4708 4709 4710
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
4711
        ipdef = virNetworkDefGetIpByIndex(netdef, AF_UNSPEC, 0);
4712
        if (!ipdef) {
4713
            virReportError(VIR_ERR_INTERNAL_ERROR,
4714
                           _("network '%s' doesn't have an IP address"),
4715
                           netdef->name);
4716
            goto cleanup;
4717 4718 4719 4720 4721
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4722
        if ((dev_name = netdef->bridge))
4723 4724
            break;
        /*
4725 4726
         * fall through if netdef->bridge wasn't set, since that is
         * macvtap bridge mode network.
4727 4728 4729 4730
         */
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
4731 4732
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4733

4734
        if (!dev_name) {
4735 4736 4737
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4738
            goto cleanup;
4739 4740 4741 4742
        }
        break;
    }

4743
    if (dev_name) {
4744
        if (virNetDevGetIPAddress(dev_name, &addr) < 0)
4745
            goto cleanup;
4746
        addrptr = &addr;
4747 4748
    }

4749 4750
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
4751
        goto cleanup;
4752 4753
    }

4754
    ret = 0;
4755
 cleanup:
4756
    virNetworkObjEndAPI(&network);
4757 4758
    return ret;
}
4759

4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831
/* networkGetActualType:
 * @dom: domain definition that @iface belongs to
 * @iface: the original NetDef from the domain
 *
 * Looks up the network reference by iface, and returns the actual
 * type of the connection without allocating any resources.
 *
 * Returns 0 on success, -1 on failure.
 */
int
networkGetActualType(virDomainNetDefPtr iface)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
    int ret = -1;

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

    if (iface->data.network.actual)
        return iface->data.network.actual->type;

    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 -1;
    }
    netdef = network->def;

    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
        /* for these forward types, the actual net type really *is*
         * NETWORK; we just keep the info from the portgroup in
         * iface->data.network.actual
         */
        ret = VIR_DOMAIN_NET_TYPE_NETWORK;

    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
               netdef->bridge) {

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

        ret = VIR_DOMAIN_NET_TYPE_BRIDGE;

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

        ret = VIR_DOMAIN_NET_TYPE_HOSTDEV;

    } 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)) {

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

        ret = VIR_DOMAIN_NET_TYPE_DIRECT;

    }

    virNetworkObjEndAPI(&network);
    return ret;
}


4832 4833 4834
/**
 * networkCheckBandwidth:
 * @net: network QoS
4835
 * @ifaceBand: interface QoS (may be NULL if no QoS)
4836
 * @oldBandwidth: new interface QoS (may be NULL if no QoS)
4837
 * @ifaceMac: interface MAC (used in error messages for identification)
4838 4839
 * @new_rate: new rate for non guaranteed class
 *
4840 4841 4842 4843 4844 4845 4846 4847
 * 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.
 *
4848 4849 4850 4851 4852 4853
 * 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,
4854
                      virNetDevBandwidthPtr ifaceBand,
4855
                      virNetDevBandwidthPtr oldBandwidth,
4856
                      virMacAddr ifaceMac,
4857 4858 4859 4860 4861 4862 4863 4864
                      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];

4865
    virMacAddrFormat(&ifaceMac, ifmac);
4866 4867 4868 4869 4870 4871 4872 4873 4874 4875

    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;
    }

4876 4877
    if (((!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor) &&
         (!oldBandwidth || !oldBandwidth->in || !oldBandwidth->in->floor)) ||
4878 4879
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4880
        return 1;
4881
    }
4882 4883

    tmp_new_rate = netBand->in->average;
4884 4885 4886 4887
    if (oldBandwidth && oldBandwidth->in)
        tmp_floor_sum -= oldBandwidth->in->floor;
    if (ifaceBand && ifaceBand->in)
        tmp_floor_sum += ifaceBand->in->floor;
4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912

    /* 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;
    }

4913 4914
    if (new_rate)
        *new_rate = tmp_new_rate;
4915 4916
    ret = 0;

4917
 cleanup:
4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933
    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)
{
4934
    ssize_t ret = 0;
4935

4936
    ret = virBitmapNextClearBit(net->class_id, -1);
4937

4938
    if (ret < 0 || virBitmapSetBit(net->class_id, ret) < 0)
4939 4940 4941 4942 4943
        return -1;

    return ret;
}

4944

4945
static int
4946 4947 4948 4949
networkPlugBandwidthImpl(virNetworkObjPtr net,
                         virDomainNetDefPtr iface,
                         virNetDevBandwidthPtr ifaceBand,
                         unsigned long long new_rate)
4950
{
4951
    virNetworkDriverStatePtr driver = networkGetDriver();
4952
    ssize_t class_id = 0;
4953 4954
    int plug_ret;
    int ret = -1;
4955 4956 4957 4958 4959 4960 4961 4962

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

4963 4964
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
4965 4966 4967 4968 4969 4970 4971 4972
    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 */
4973
    net->floor_sum += ifaceBand->in->floor;
4974
    /* update status file */
4975
    if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
4976
        ignore_value(virBitmapClearBit(net->class_id, class_id));
4977
        net->floor_sum -= ifaceBand->in->floor;
4978 4979 4980 4981
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
4982 4983
    /* update rate for non guaranteed NICs */
    new_rate -= net->floor_sum;
4984
    if (virNetDevBandwidthUpdateRate(net->def->bridge, 2,
4985 4986 4987 4988 4989
                                     net->def->bandwidth, new_rate) < 0)
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                 net->def->bridge);

    ret = 0;
4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029
 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;
5030

5031
 cleanup:
5032 5033 5034 5035 5036 5037 5038
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
5039
    virNetworkDriverStatePtr driver = networkGetDriver();
5040 5041
    int ret = 0;
    unsigned long long new_rate;
5042
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
5043 5044 5045

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
5046 5047 5048 5049 5050
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061
        /* 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 */
5062
        net->floor_sum -= ifaceBand->in->floor;
5063 5064 5065 5066
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
5067
        if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
5068
            net->floor_sum += ifaceBand->in->floor;
5069 5070 5071 5072
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
5073 5074
        /* update rate for non guaranteed NICs */
        new_rate -= net->floor_sum;
5075
        if (virNetDevBandwidthUpdateRate(net->def->bridge, 2,
5076 5077 5078 5079 5080 5081 5082
                                         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;
    }

5083
 cleanup:
5084 5085
    return ret;
}
5086 5087 5088

static void
networkNetworkObjTaint(virNetworkObjPtr net,
5089
                       virNetworkTaintFlags taint)
5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100
{
    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));
    }
}
5101 5102 5103 5104 5105 5106


static bool
networkBandwidthGenericChecks(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
5107
    virNetDevBandwidthPtr ifaceBand;
5108 5109 5110 5111 5112 5113 5114 5115
    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;
    }

5116
    ifaceBand = virDomainNetGetActualBandwidth(iface);
5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156
    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;
}
5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194


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: */

5195 5196
    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
        newBandwidth && newBandwidth->in && newBandwidth->in->floor) {
5197 5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238
        /* 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;
}