bridge_driver.c 185.7 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 "device_conf.h"
53
#include "driver.h"
54
#include "virbuffer.h"
55
#include "virpidfile.h"
56
#include "vircommand.h"
57
#include "viralloc.h"
58
#include "viruuid.h"
59
#include "viriptables.h"
60
#include "virlog.h"
61
#include "virdnsmasq.h"
62
#include "configmake.h"
63
#include "virnetlink.h"
64
#include "virnetdev.h"
65
#include "virnetdevip.h"
66 67
#include "virnetdevbridge.h"
#include "virnetdevtap.h"
68
#include "virnetdevvportprofile.h"
69
#include "virpci.h"
70
#include "virdbus.h"
71
#include "virfile.h"
72
#include "virstring.h"
73
#include "viraccessapicheck.h"
74
#include "network_event.h"
75
#include "virhook.h"
76
#include "virjson.h"
77

78
#define VIR_FROM_THIS VIR_FROM_NETWORK
79
#define MAX_BRIDGE_ID 256
80

81 82 83 84 85 86 87
/**
 * VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX:
 *
 * Macro providing the upper limit on the size of leases file
 */
#define VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX (32 * 1024 * 1024)

88 89
#define SYSCTL_PATH "/proc/sys"

90 91
VIR_LOG_INIT("network.bridge_driver");

92
static virNetworkDriverStatePtr network_driver;
93

94 95 96 97 98 99 100 101
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;
}
102

103
static void networkDriverLock(virNetworkDriverStatePtr driver)
104
{
105
    virMutexLock(&driver->lock);
106
}
107
static void networkDriverUnlock(virNetworkDriverStatePtr driver)
108
{
109
    virMutexUnlock(&driver->lock);
110 111
}

112 113 114 115
static dnsmasqCapsPtr
networkGetDnsmasqCaps(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr ret;
116
    networkDriverLock(driver);
117
    ret = virObjectRef(driver->dnsmasqCaps);
118
    networkDriverUnlock(driver);
119 120 121 122 123 124 125 126 127 128 129
    return ret;
}

static int
networkDnsmasqCapsRefresh(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr caps;

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

130
    networkDriverLock(driver);
131 132
    virObjectUnref(driver->dnsmasqCaps);
    driver->dnsmasqCaps = caps;
133
    networkDriverUnlock(driver);
134 135 136
    return 0;
}

137
static int networkStateCleanup(void);
138

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

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

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

148 149
static int networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
                                         virNetworkObjPtr network);
150

151
static int networkStartNetworkExternal(virNetworkObjPtr network);
152

153
static int networkShutdownNetworkExternal(virNetworkObjPtr network);
154

155 156
static void networkReloadFirewallRules(virNetworkDriverStatePtr driver);
static void networkRefreshDaemons(virNetworkDriverStatePtr driver);
157

158 159 160 161 162
static int networkPlugBandwidth(virNetworkObjPtr net,
                                virDomainNetDefPtr iface);
static int networkUnplugBandwidth(virNetworkObjPtr net,
                                  virDomainNetDefPtr iface);

163
static void networkNetworkObjTaint(virNetworkObjPtr net,
164
                                   virNetworkTaintFlags taint);
165

166 167 168
static virNetworkObjPtr
networkObjFromNetwork(virNetworkPtr net)
{
169
    virNetworkDriverStatePtr driver = networkGetDriver();
170 171 172
    virNetworkObjPtr network;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

173
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
174 175 176 177 178 179 180 181 182 183
    if (!network) {
        virUUIDFormat(net->uuid, uuidstr);
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching uuid '%s' (%s)"),
                       uuidstr, net->name);
    }

    return network;
}

184 185 186
static int
networkRunHook(virNetworkObjPtr network,
               virDomainDefPtr dom,
187
               virDomainNetDefPtr iface,
188 189 190 191 192 193 194 195 196
               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)) {
197 198 199 200 201 202
        if (!network) {
            VIR_DEBUG("Not running hook as @network is NULL");
            ret = 0;
            goto cleanup;
        }

203 204
        virBufferAddLit(&buf, "<hookData>\n");
        virBufferAdjustIndent(&buf, 2);
205
        if (iface && virDomainNetDefFormat(&buf, iface, NULL, 0) < 0)
206
            goto cleanup;
207 208
        if (virNetworkDefFormatBuf(&buf, network->def, 0) < 0)
            goto cleanup;
209
        if (dom && virDomainDefFormatInternal(dom, NULL, 0, &buf) < 0)
210 211 212 213 214
            goto cleanup;

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

215
        if (virBufferCheckError(&buf) < 0)
216 217
            goto cleanup;

218
        xml = virBufferContentAndReset(&buf);
219 220 221 222 223 224 225 226
        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;
227 228

        networkNetworkObjTaint(network, VIR_NETWORK_TAINT_HOOK);
229 230 231
    }

    ret = 0;
232
 cleanup:
233 234 235 236 237 238 239
    virBufferFreeAndReset(&buf);
    VIR_FREE(xml);
    VIR_FREE(net_xml);
    VIR_FREE(dom_xml);
    return ret;
}

240
static char *
241 242
networkDnsmasqLeaseFileNameDefault(virNetworkDriverStatePtr driver,
                                   const char *netname)
243 244 245
{
    char *leasefile;

246
    ignore_value(virAsprintf(&leasefile, "%s/%s.leases",
247
                             driver->dnsmasqStateDir, netname));
248 249 250
    return leasefile;
}

251
static char *
252 253
networkDnsmasqLeaseFileNameCustom(virNetworkDriverStatePtr driver,
                                  const char *bridge)
254 255 256 257
{
    char *leasefile;

    ignore_value(virAsprintf(&leasefile, "%s/%s.status",
258
                             driver->dnsmasqStateDir, bridge));
259 260 261
    return leasefile;
}

262
static char *
263 264
networkDnsmasqConfigFileName(virNetworkDriverStatePtr driver,
                             const char *netname)
265 266 267
{
    char *conffile;

268
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
269
                             driver->dnsmasqStateDir, netname));
270 271 272
    return conffile;
}

273 274 275 276 277 278
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

279
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
280 281 282 283
    return pidfilebase;
}

static char *
284 285
networkRadvdConfigFileName(virNetworkDriverStatePtr driver,
                           const char *netname)
286 287 288
{
    char *configfile;

289
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
290
                             driver->radvdStateDir, netname));
291 292
    return configfile;
}
293

M
Michal Privoznik 已提交
294 295 296 297 298 299 300 301 302 303 304
static char *
networkMacMgrFileName(virNetworkDriverStatePtr driver,
                      const char *bridge)
{
    char *filename;

    ignore_value(virAsprintf(&filename, "%s/%s.macs",
                             driver->dnsmasqStateDir, bridge));
    return filename;
}

305 306
/* do needed cleanup steps and remove the network from the list */
static int
307 308
networkRemoveInactive(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr net)
309 310
{
    char *leasefile = NULL;
311
    char *customleasefile = NULL;
312
    char *radvdconfigfile = NULL;
313
    char *configfile = NULL;
314
    char *radvdpidbase = NULL;
315
    char *statusfile = NULL;
M
Michal Privoznik 已提交
316
    char *macMapFile = NULL;
317 318 319 320 321 322
    dnsmasqContext *dctx = NULL;
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(net);

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
323
    if (!(dctx = dnsmasqContextNew(def->name,
324
                                   driver->dnsmasqStateDir))) {
325
        goto cleanup;
326
    }
327

328
    if (!(leasefile = networkDnsmasqLeaseFileNameDefault(driver, def->name)))
329 330
        goto cleanup;

331
    if (!(customleasefile = networkDnsmasqLeaseFileNameCustom(driver, def->bridge)))
332 333
        goto cleanup;

334
    if (!(radvdconfigfile = networkRadvdConfigFileName(driver, def->name)))
335
        goto cleanup;
336 337

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
338
        goto cleanup;
339

340
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
341
        goto cleanup;
342

343
    if (!(statusfile = virNetworkConfigFile(driver->stateDir, def->name)))
344
        goto cleanup;
345

M
Michal Privoznik 已提交
346 347 348
    if (!(macMapFile = networkMacMgrFileName(driver, def->bridge)))
        goto cleanup;

349 350 351
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
352
    unlink(customleasefile);
353
    unlink(configfile);
354

M
Michal Privoznik 已提交
355 356 357
    /* MAC map manager */
    unlink(macMapFile);

358 359
    /* radvd */
    unlink(radvdconfigfile);
360
    virPidFileDelete(driver->pidDir, radvdpidbase);
361

362 363 364
    /* remove status file */
    unlink(statusfile);

365
    /* remove the network definition */
366
    virNetworkObjRemoveInactive(driver->networks, net);
367 368 369

    ret = 0;

370
 cleanup:
371
    VIR_FREE(leasefile);
372
    VIR_FREE(configfile);
373
    VIR_FREE(customleasefile);
374 375
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
376
    VIR_FREE(statusfile);
M
Michal Privoznik 已提交
377
    VIR_FREE(macMapFile);
378 379 380 381
    dnsmasqContextFree(dctx);
    return ret;
}

M
Michal Privoznik 已提交
382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441
static int
networkMacMgrAdd(virNetworkDriverStatePtr driver,
                 virNetworkObjPtr network,
                 const char *domain,
                 const virMacAddr *mac)
{
    char macStr[VIR_MAC_STRING_BUFLEN];
    char *file = NULL;
    int ret = -1;

    if (!network->macmap)
        return 0;

    virMacAddrFormat(mac, macStr);

    if (!(file = networkMacMgrFileName(driver, network->def->bridge)))
        goto cleanup;

    if (virMacMapAdd(network->macmap, domain, macStr) < 0)
        goto cleanup;

    if (virMacMapWriteFile(network->macmap, file) < 0)
        goto cleanup;

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

static int
networkMacMgrDel(virNetworkDriverStatePtr driver,
                 virNetworkObjPtr network,
                 const char *domain,
                 const virMacAddr *mac)
{
    char macStr[VIR_MAC_STRING_BUFLEN];
    char *file = NULL;
    int ret = -1;

    if (!network->macmap)
        return 0;

    virMacAddrFormat(mac, macStr);

    if (!(file = networkMacMgrFileName(driver, network->def->bridge)))
        goto cleanup;

    if (virMacMapRemove(network->macmap, domain, macStr) < 0)
        goto cleanup;

    if (virMacMapWriteFile(network->macmap, file) < 0)
        goto cleanup;

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

442 443 444
static char *
networkBridgeDummyNicName(const char *brname)
{
445
    static const char dummyNicSuffix[] = "-nic";
446 447
    char *nicname;

448 449 450 451 452 453 454
    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.
         */
455 456 457 458 459
        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));
460
    } else {
461
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
462
    }
463 464 465
    return nicname;
}

466 467
static int
networkUpdateState(virNetworkObjPtr obj,
468
                   void *opaque)
469
{
470
    virNetworkDriverStatePtr driver = opaque;
471
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
472
    char *macMapFile = NULL;
473
    int ret = -1;
474

475
    virObjectLock(obj);
476
    if (!virNetworkObjIsActive(obj)) {
477 478
        ret = 0;
        goto cleanup;
479
    }
480

481 482 483 484
    switch (obj->def->forward.type) {
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
485
    case VIR_NETWORK_FORWARD_OPEN:
486 487 488
        /* If bridge doesn't exist, then mark it inactive */
        if (!(obj->def->bridge && virNetDevExists(obj->def->bridge) == 1))
            obj->active = 0;
489 490 491 492 493 494 495

        if (!(macMapFile = networkMacMgrFileName(driver, obj->def->bridge)))
            goto cleanup;

        if (!(obj->macmap = virMacMapNew(macMapFile)))
            goto cleanup;

496
        break;
497

498
    case VIR_NETWORK_FORWARD_BRIDGE:
499 500
        if (obj->def->bridge) {
            if (virNetDevExists(obj->def->bridge) != 1)
501 502
                obj->active = 0;
            break;
503
        }
504 505 506 507 508 509 510 511 512
        /* 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;
513

514 515 516
    case VIR_NETWORK_FORWARD_HOSTDEV:
        /* so far no extra checks */
        break;
517
    }
518

519 520 521
    /* Try and read dnsmasq/radvd pids of active networks */
    if (obj->active && obj->def->ips && (obj->def->nips > 0)) {
        char *radvdpidbase;
522

523 524 525
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           obj->def->name,
                                           &obj->dnsmasqPid,
526
                                           dnsmasqCapsGetBinaryPath(dnsmasq_caps)));
M
Michal Privoznik 已提交
527

528 529 530
        radvdpidbase = networkRadvdPidfileBasename(obj->def->name);
        if (!radvdpidbase)
            goto cleanup;
531

532 533 534 535
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           radvdpidbase,
                                           &obj->radvdPid, RADVD));
        VIR_FREE(radvdpidbase);
536
    }
537

538 539
    ret = 0;
 cleanup:
540
    virObjectUnlock(obj);
541
    virObjectUnref(dnsmasq_caps);
542
    VIR_FREE(macMapFile);
543 544
    return ret;
}
545

546

547 548
static int
networkAutostartConfig(virNetworkObjPtr net,
549
                       void *opaque)
550
{
551
    virNetworkDriverStatePtr driver = opaque;
552
    int ret = -1;
553

554
    virObjectLock(net);
555 556
    if (net->autostart &&
        !virNetworkObjIsActive(net) &&
557
        networkStartNetwork(driver, net) < 0)
558 559 560 561
        goto cleanup;

    ret = 0;
 cleanup:
562
    virObjectUnlock(net);
563
    return ret;
564 565
}

566 567 568
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
569
                             DBusMessage *message, void *user_data)
570
{
571 572
    virNetworkDriverStatePtr driver = user_data;

573 574 575 576 577 578
    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.");
579
        networkReloadFirewallRules(driver);
580 581 582 583 584 585
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

586
static int
587
networkMigrateStateFiles(virNetworkDriverStatePtr driver)
588 589 590 591 592 593 594 595 596 597 598 599 600
{
    /* 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;
601
    int direrr;
602 603 604
    struct dirent *entry;
    char *oldPath = NULL, *newPath = NULL;
    char *contents = NULL;
J
Ján Tomko 已提交
605
    int rc;
606

J
Ján Tomko 已提交
607 608
    if ((rc = virDirOpenIfExists(&dir, oldStateDir)) <= 0)
        return rc;
609 610 611 612 613 614 615

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

616
    while ((direrr = virDirRead(dir, &entry, oldStateDir)) > 0) {
617 618 619
        if (entry->d_type != DT_UNKNOWN &&
            entry->d_type != DT_REG)
            continue;
620 621 622

        if (virAsprintf(&oldPath, "%s/%s",
                        oldStateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
623
            goto cleanup;
624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640

        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;
            }
        }

641
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
J
Ján Tomko 已提交
642
            goto cleanup;
643 644 645

        if (virAsprintf(&newPath, "%s/%s",
                        driver->stateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
646
            goto cleanup;
647 648 649 650 651 652 653 654 655 656 657 658
        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);
    }
659
    if (direrr < 0)
J
Ján Tomko 已提交
660
        goto cleanup;
661 662 663

    ret = 0;
 cleanup:
J
Ján Tomko 已提交
664
    VIR_DIR_CLOSE(dir);
665 666 667 668 669 670
    VIR_FREE(oldPath);
    VIR_FREE(newPath);
    VIR_FREE(contents);
    return ret;
}

671
/**
672
 * networkStateInitialize:
673 674 675 676
 *
 * Initialization function for the QEmu daemon
 */
static int
677 678 679
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
680
{
681 682 683
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
684 685 686
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
687

688
    if (VIR_ALLOC(network_driver) < 0)
689
        goto error;
690

691 692
    if (virMutexInit(&network_driver->lock) < 0) {
        VIR_FREE(network_driver);
693 694
        goto error;
    }
695

696 697 698 699
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
700
    if (privileged) {
701
        if (VIR_STRDUP(network_driver->networkConfigDir,
702
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
703
            VIR_STRDUP(network_driver->networkAutostartDir,
704
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
705
            VIR_STRDUP(network_driver->stateDir,
706
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
707
            VIR_STRDUP(network_driver->pidDir,
708
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
709
            VIR_STRDUP(network_driver->dnsmasqStateDir,
710
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
711
            VIR_STRDUP(network_driver->radvdStateDir,
712 713
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
714 715 716 717 718

        /* migration from old to new location is only applicable for
         * privileged mode - unprivileged mode directories haven't
         * changed location.
         */
719
        if (networkMigrateStateFiles(network_driver) < 0)
720
            goto error;
721
    } else {
722 723 724
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
725
            goto error;
726

727
        if ((virAsprintf(&network_driver->networkConfigDir,
728
                         "%s/qemu/networks", configdir) < 0) ||
729
            (virAsprintf(&network_driver->networkAutostartDir,
730
                         "%s/qemu/networks/autostart", configdir) < 0) ||
731
            (virAsprintf(&network_driver->stateDir,
732
                         "%s/network/lib", rundir) < 0) ||
733
            (virAsprintf(&network_driver->pidDir,
734
                         "%s/network/run", rundir) < 0) ||
735
            (virAsprintf(&network_driver->dnsmasqStateDir,
736
                         "%s/dnsmasq/lib", rundir) < 0) ||
737
            (virAsprintf(&network_driver->radvdStateDir,
738
                         "%s/radvd/lib", rundir) < 0)) {
739
            goto error;
740
        }
741 742
    }

743
    if (virFileMakePath(network_driver->stateDir) < 0) {
744 745
        virReportSystemError(errno,
                             _("cannot create directory %s"),
746
                             network_driver->stateDir);
747 748 749
        goto error;
    }

750
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
751
    network_driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
752

753
    if (!(network_driver->networks = virNetworkObjListNew()))
754 755
        goto error;

756 757
    if (virNetworkObjLoadAllState(network_driver->networks,
                                  network_driver->stateDir) < 0)
758 759
        goto error;

760 761 762
    if (virNetworkObjLoadAllConfigs(network_driver->networks,
                                    network_driver->networkConfigDir,
                                    network_driver->networkAutostartDir) < 0)
763 764
        goto error;

765 766 767 768
    /* 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). */
769
    virNetworkObjListForEach(network_driver->networks,
770
                             networkUpdateState,
771 772
                             network_driver);
    virNetworkObjListPrune(network_driver->networks,
773 774
                           VIR_CONNECT_LIST_NETWORKS_INACTIVE |
                           VIR_CONNECT_LIST_NETWORKS_TRANSIENT);
775 776
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
777

778
    network_driver->networkEventState = virObjectEventStateNew();
779

780 781 782
#ifdef HAVE_FIREWALLD
    if (!(sysbus = virDBusGetSystemBus())) {
        VIR_WARN("DBus not available, disabling firewalld support "
783
                 "in bridge_network_driver: %s", virGetLastErrorMessage());
784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800
    } 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,
801
                                   network_driver, NULL);
802 803 804
    }
#endif

805
    ret = 0;
806
 cleanup:
807 808 809
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
810

811
 error:
812
    networkStateCleanup();
813
    goto cleanup;
814 815
}

816 817 818 819 820 821 822 823
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
824
    if (!network_driver)
825 826
        return;

827
    virNetworkObjListForEach(network_driver->networks,
828
                             networkAutostartConfig,
829
                             network_driver);
830 831
}

832
/**
833
 * networkStateReload:
834 835 836 837 838
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
839 840
networkStateReload(void)
{
841
    if (!network_driver)
842 843
        return 0;

844 845 846 847 848
    virNetworkObjLoadAllState(network_driver->networks,
                              network_driver->stateDir);
    virNetworkObjLoadAllConfigs(network_driver->networks,
                                network_driver->networkConfigDir,
                                network_driver->networkAutostartDir);
849 850 851
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
    virNetworkObjListForEach(network_driver->networks,
852
                             networkAutostartConfig,
853
                             network_driver);
854 855 856 857 858
    return 0;
}


/**
859
 * networkStateCleanup:
860 861 862 863
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
864 865
networkStateCleanup(void)
{
866
    if (!network_driver)
867 868
        return -1;

869
    virObjectUnref(network_driver->networkEventState);
870

871
    /* free inactive networks */
872
    virObjectUnref(network_driver->networks);
873

874 875 876 877 878 879
    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);
880

881
    virObjectUnref(network_driver->dnsmasqCaps);
882

883
    virMutexDestroy(&network_driver->lock);
884

885
    VIR_FREE(network_driver);
886 887 888 889 890

    return 0;
}


891 892 893 894 895 896 897
/* 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)
{
898 899
    size_t i;
    int ret = -1;
900 901 902 903 904 905 906
    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.
     */
907
    for (i = 0; i < 25; i++) {
908
        int signum = 0;
909
        if (i == 0) {
910
            signum = SIGTERM;
911
        } else if (i == 15) {
912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946
            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);
947
 cleanup:
948 949 950
    return ret;
}

J
Ján Tomko 已提交
951 952 953
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
G
Gene Czarcinski 已提交
954

955
static int
G
Gene Czarcinski 已提交
956
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
957
                                 virNetworkIPDefPtr ipdef)
958
{
959
    size_t i;
G
Gene Czarcinski 已提交
960
    bool ipv6 = false;
961

G
Gene Czarcinski 已提交
962 963
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
964 965
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
966
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
967 968
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
969
                return -1;
970
    }
971

G
Gene Czarcinski 已提交
972 973 974 975 976 977 978
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
979
    size_t i, j;
G
Gene Czarcinski 已提交
980

981 982
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
983
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
984
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
985
                for (j = 0; j < host->nnames; j++)
986 987
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
988 989
            }
        }
990 991
    }

992
    return 0;
993 994 995
}


996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032
static int
networkDnsmasqConfLocalPTRs(virBufferPtr buf,
                            virNetworkDefPtr def)
{
    virNetworkIPDefPtr ip;
    size_t i;
    char *ptr = NULL;
    int rc;

    for (i = 0; i < def->nips; i++) {
        ip = def->ips + i;

        if (ip->localPTR != VIR_TRISTATE_BOOL_YES)
            continue;

        if ((rc = virSocketAddrPTRDomain(&ip->address,
                                         virNetworkIPDefPrefix(ip),
                                         &ptr)) < 0) {
            if (rc == -2) {
                int family = VIR_SOCKET_ADDR_FAMILY(&ip->address);
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("PTR domain for %s network with prefix %u "
                                 "cannot be automatically created"),
                               (family == AF_INET) ? "IPv4" : "IPv6",
                               virNetworkIPDefPrefix(ip));
            }
            return -1;
        }

        virBufferAsprintf(buf, "local=/%s/\n", ptr);
        VIR_FREE(ptr);
    }

    return 0;
}


1033 1034
int
networkDnsmasqConfContents(virNetworkObjPtr network,
1035 1036 1037 1038
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
1039
{
1040
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1041
    int r, ret = -1;
1042
    int nbleases = 0;
1043
    size_t i;
1044
    virNetworkDNSDefPtr dns = &network->def->dns;
1045
    bool wantDNS = dns->enable != VIR_TRISTATE_BOOL_NO;
1046
    virNetworkIPDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
G
Gene Czarcinski 已提交
1047
    bool ipv6SLAAC;
1048
    char *saddr = NULL, *eaddr = NULL;
1049

1050 1051
    *configstr = NULL;

1052
    /*
1053 1054 1055
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
1056
     *
1057 1058
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
1059
     */
1060 1061 1062 1063 1064 1065

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

1066
    /* create dnsmasq config file appropriate for this network */
1067 1068

    /* Don't forget to update networkxml2conftest :-) */
1069
    virBufferAsprintf(&configbuf,
1070 1071 1072 1073 1074 1075 1076
                      "##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"
1077
                      "strict-order\n",
1078 1079
                      network->def->name);

1080 1081 1082 1083 1084 1085 1086
    /* if dns is disabled, set its listening port to 0, which
     * tells dnsmasq to not listen
     */
    if (!wantDNS)
        virBufferAddLit(&configbuf, "port=0\n");

    if (wantDNS && network->def->dns.forwarders) {
1087 1088 1089 1090 1091 1092 1093 1094 1095
        /* addNoResolv should be set to true if there are any entries
         * that specify an IP address for requests, but no domain
         * qualifier (implying that all requests otherwise "unclaimed"
         * should be sent to that address). if it is still false when
         * we've looked at all entries, it means we still need the
         * host's resolv.conf for some cases.
         */
        bool addNoResolv = false;

1096
        for (i = 0; i < network->def->dns.nfwds; i++) {
1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107
            virNetworkDNSForwarderPtr fwd = &network->def->dns.forwarders[i];

            virBufferAddLit(&configbuf, "server=");
            if (fwd->domain)
                virBufferAsprintf(&configbuf, "/%s/", fwd->domain);
            if (VIR_SOCKET_ADDR_VALID(&fwd->addr)) {
                char *addr = virSocketAddrFormat(&fwd->addr);

                if (!addr)
                    goto cleanup;
                virBufferAsprintf(&configbuf, "%s\n", addr);
1108
                VIR_FREE(addr);
1109 1110
                if (!fwd->domain)
                    addNoResolv = true;
1111 1112 1113 1114
            } else {
                /* "don't forward requests for this domain" */
                virBufferAddLit(&configbuf, "#\n");
            }
1115
        }
1116 1117
        if (addNoResolv)
            virBufferAddLit(&configbuf, "no-resolv\n");
1118 1119
    }

1120
    if (network->def->domain) {
1121 1122 1123 1124 1125
        if (network->def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
                              network->def->domain);
        }
1126
        virBufferAsprintf(&configbuf,
1127 1128 1129 1130
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
1131

1132 1133 1134 1135
    if (wantDNS &&
        networkDnsmasqConfLocalPTRs(&configbuf, network->def) < 0)
        goto cleanup;

1136
    if (wantDNS && network->def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
1137 1138 1139 1140
        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)
1141
         */
1142
        virBufferAddLit(&configbuf, "local=//\n");
1143
    }
1144

1145
    if (pidfile)
1146
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
1147

1148
    /* dnsmasq will *always* listen on localhost unless told otherwise */
P
Pavel Timofeev 已提交
1149
#ifdef __linux__
1150
    virBufferAddLit(&configbuf, "except-interface=lo\n");
P
Pavel Timofeev 已提交
1151 1152 1153 1154
#else
    /* BSD family OSes and Solaris call loopback interface as lo0 */
    virBufferAddLit(&configbuf, "except-interface=lo0\n");
#endif
1155

1156 1157 1158 1159 1160 1161 1162 1163
    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.
         */
1164
        virBufferAsprintf(&configbuf,
1165 1166 1167
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
1168
    } else {
1169
        virBufferAddLit(&configbuf, "bind-interfaces\n");
1170 1171 1172 1173 1174 1175 1176 1177
        /*
         * --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
         */
1178
        for (i = 0;
1179
             (tmpipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i));
1180
             i++) {
1181 1182 1183 1184
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
1185

1186
            /* also part of CVE 2012-3411 - if the host's version of
1187
             * dnsmasq doesn't have bind-dynamic, only allow listening on
1188 1189
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
1190 1191
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
1192 1193 1194 1195
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
1196
                                 "The version of dnsmasq on this host (%d.%d) "
1197 1198 1199 1200
                                 "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 "
1201 1202 1203 1204 1205 1206
                                 "(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);
1207
                VIR_FREE(ipaddr);
1208 1209
                goto cleanup;
            }
1210
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1211 1212 1213
            VIR_FREE(ipaddr);
        }
    }
1214

1215 1216
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1217
     * guaranteed to not work, and set no-resolv so that no dns
1218 1219 1220
     * 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).
1221 1222 1223
     * IPv6 RA always contains an implicit default route
     * via the sender's link-local address. The only thing we can do
     * is set the lifetime of this route to 0, i.e. disable it.
1224
     */
1225
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1226
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1227
                        "no-resolv\n");
1228 1229 1230 1231
        if (dnsmasqCapsGet(caps, DNSMASQ_CAPS_RA_PARAM)) {
            /* interface=* (any), interval=0 (default), lifetime=0 (seconds) */
            virBufferAddLit(&configbuf, "ra-param=*,0,0\n");
        }
1232
    }
1233

1234 1235 1236 1237 1238
    if (wantDNS) {
        for (i = 0; i < dns->ntxts; i++) {
            virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
                              dns->txts[i].name,
                              dns->txts[i].value);
1239
        }
1240

1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263
        for (i = 0; i < dns->nsrvs; i++) {
            /* 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);
1264

1265 1266 1267 1268 1269 1270 1271
            /* domain is optional - it defaults to the domain of this network */
            if (dns->srvs[i].domain)
                virBufferAsprintf(&configbuf, ".%s", dns->srvs[i].domain);

            /* 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.
1272
             */
1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292
            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);
            }
            virBufferAddLit(&configbuf, "\n");
1293
        }
1294 1295
    }

G
Gene Czarcinski 已提交
1296
    /* Find the first dhcp for both IPv4 and IPv6 */
1297
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
1298
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i));
1299
         i++) {
G
Gene Czarcinski 已提交
1300 1301 1302 1303
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1304 1305
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316
                    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,
1317 1318 1319 1320 1321 1322 1323 1324 1325
                                   _("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
1326 1327 1328 1329
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1330 1331
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344
                    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 "
1345 1346 1347 1348
                 "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
1349 1350 1351 1352 1353
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1354 1355
        int prefix;

1356
        prefix = virNetworkIPDefPrefix(ipdef);
1357 1358 1359 1360 1361 1362
        if (prefix < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
            goto cleanup;
        }
1363
        for (r = 0; r < ipdef->nranges; r++) {
1364 1365
            int thisRange;

1366 1367
            if (!(saddr = virSocketAddrFormat(&ipdef->ranges[r].start)) ||
                !(eaddr = virSocketAddrFormat(&ipdef->ranges[r].end)))
1368
                goto cleanup;
1369

1370
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s",
1371
                              saddr, eaddr);
1372
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1373
                virBufferAsprintf(&configbuf, ",%d", prefix);
1374 1375
            virBufferAddLit(&configbuf, "\n");

1376
            VIR_FREE(saddr);
1377
            VIR_FREE(eaddr);
1378
            thisRange = virSocketAddrGetRange(&ipdef->ranges[r].start,
1379 1380
                                              &ipdef->ranges[r].end,
                                              &ipdef->address,
1381
                                              virNetworkIPDefPrefix(ipdef));
1382 1383 1384
            if (thisRange < 0)
                goto cleanup;
            nbleases += thisRange;
1385
        }
1386

1387
        /*
1388 1389 1390 1391
         * 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)
1392 1393
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1394
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1395 1396
            if (!bridgeaddr)
                goto cleanup;
1397 1398 1399
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static",
                              bridgeaddr);
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1400
                virBufferAsprintf(&configbuf, ",%d", prefix);
1401
            virBufferAddLit(&configbuf, "\n");
1402 1403
            VIR_FREE(bridgeaddr);
        }
1404

G
Gene Czarcinski 已提交
1405 1406
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1407

G
Gene Czarcinski 已提交
1408 1409
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
1410
            if (ipdef->nranges || ipdef->nhosts) {
1411
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1412 1413
                virBufferAddLit(&configbuf, "dhcp-authoritative\n");
            }
1414

G
Gene Czarcinski 已提交
1415
            if (ipdef->tftproot) {
1416 1417
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1418
            }
1419

G
Gene Czarcinski 已提交
1420 1421 1422
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1423

1424
                    if (!bootserver)
G
Gene Czarcinski 已提交
1425
                        goto cleanup;
1426
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1427
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1428 1429
                    VIR_FREE(bootserver);
                } else {
1430
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1431 1432 1433 1434 1435
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1436

1437
    if (nbleases > 0)
1438
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1439

G
Gene Czarcinski 已提交
1440 1441
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1442
        goto cleanup;
G
Gene Czarcinski 已提交
1443 1444 1445 1446 1447 1448

    /* 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)
1449 1450
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1451

1452 1453
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1454
     */
1455 1456 1457 1458
    if (wantDNS) {
        virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
                          dctx->addnhostsfile->path);
    }
G
Gene Czarcinski 已提交
1459 1460 1461

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1462
        if (ipv6def) {
1463
            virBufferAddLit(&configbuf, "enable-ra\n");
1464
        } else {
1465
            for (i = 0;
1466
                 (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1467
                 i++) {
G
Gene Czarcinski 已提交
1468 1469 1470 1471
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1472 1473
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1474 1475
                    VIR_FREE(bridgeaddr);
                }
1476
            }
1477
        }
1478 1479
    }

1480 1481 1482
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1483
    ret = 0;
G
Gene Czarcinski 已提交
1484

1485
 cleanup:
1486 1487
    VIR_FREE(saddr);
    VIR_FREE(eaddr);
1488
    virBufferFreeAndReset(&configbuf);
1489
    return ret;
1490 1491
}

1492
/* build the dnsmasq command line */
1493 1494 1495
static int ATTRIBUTE_NONNULL(3)
networkBuildDhcpDaemonCommandLine(virNetworkDriverStatePtr driver,
                                  virNetworkObjPtr network,
1496
                                  virCommandPtr *cmdout,
1497 1498
                                  char *pidfile,
                                  dnsmasqContext *dctx)
1499
{
1500
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1501
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1502
    int ret = -1;
1503 1504
    char *configfile = NULL;
    char *configstr = NULL;
1505
    char *leaseshelper_path = NULL;
1506 1507

    network->dnsmasqPid = -1;
1508

1509
    if (networkDnsmasqConfContents(network, pidfile, &configstr,
1510
                                   dctx, dnsmasq_caps) < 0)
1511 1512 1513 1514 1515
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1516
    if (!(configfile = networkDnsmasqConfigFileName(driver, network->def->name)))
1517 1518 1519 1520 1521
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1522 1523
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1524 1525 1526
        goto cleanup;
    }

1527 1528
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1529
                                                  abs_topbuilddir "/src",
1530 1531 1532
                                                  LIBEXECDIR)))
        goto cleanup;

1533
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(dnsmasq_caps));
1534
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1535 1536
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1537
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1538
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", network->def->bridge);
1539

1540
    *cmdout = cmd;
1541
    ret = 0;
1542
 cleanup:
1543
    virObjectUnref(dnsmasq_caps);
1544 1545
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1546
    VIR_FREE(leaseshelper_path);
1547 1548 1549 1550
    return ret;
}

static int
1551 1552
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
1553
{
1554 1555 1556
    virNetworkIPDefPtr ipdef;
    size_t i;
    bool needDnsmasq = false;
1557 1558 1559
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1560
    dnsmasqContext *dctx = NULL;
1561

1562
    /* see if there are any IP addresses that need a dhcp server */
1563 1564 1565
    i = 0;
    while ((ipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i))) {
        i++;
1566 1567 1568 1569
        if (ipdef->nranges || ipdef->nhosts)
            needDnsmasq = true;
    }

1570 1571 1572 1573 1574 1575
    if (i == 0) {
        /* no IP addresses at all, so we don't need to run */
        ret = 0;
        goto cleanup;
    }

1576
    if (!needDnsmasq && network->def->dns.enable == VIR_TRISTATE_BOOL_NO) {
1577
        /* no DHCP services needed, and user disabled DNS service */
1578 1579 1580 1581
        ret = 0;
        goto cleanup;
    }

1582
    if (virFileMakePath(driver->pidDir) < 0) {
1583
        virReportSystemError(errno,
1584
                             _("cannot create directory %s"),
1585
                             driver->pidDir);
1586
        goto cleanup;
1587 1588
    }

1589
    if (!(pidfile = virPidFileBuildPath(driver->pidDir,
1590
                                        network->def->name)))
1591
        goto cleanup;
1592

1593
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1594
        virReportSystemError(errno,
1595
                             _("cannot create directory %s"),
1596
                             driver->dnsmasqStateDir);
1597 1598 1599
        goto cleanup;
    }

1600
    dctx = dnsmasqContextNew(network->def->name, driver->dnsmasqStateDir);
1601 1602 1603
    if (dctx == NULL)
        goto cleanup;

1604
    if (networkDnsmasqCapsRefresh(driver) < 0)
1605
        goto cleanup;
1606

1607
    ret = networkBuildDhcpDaemonCommandLine(driver, network, &cmd, pidfile, dctx);
1608 1609 1610 1611 1612
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1613
        goto cleanup;
1614

G
Guido Günther 已提交
1615
    ret = virCommandRun(cmd, NULL);
1616
    if (ret < 0)
1617 1618 1619
        goto cleanup;

    /*
1620 1621 1622 1623 1624
     * 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
1625 1626
     */

1627
    ret = virPidFileRead(driver->pidDir, network->def->name,
1628 1629
                         &network->dnsmasqPid);
    if (ret < 0)
1630
        goto cleanup;
1631

1632
    ret = 0;
1633
 cleanup:
1634
    VIR_FREE(pidfile);
1635
    virCommandFree(cmd);
1636
    dnsmasqContextFree(dctx);
1637 1638 1639
    return ret;
}

1640 1641
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1642 1643
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1644 1645 1646
 *
 *  Returns 0 on success, -1 on failure.
 */
1647
static int
1648 1649
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1650
{
1651 1652
    int ret = -1;
    size_t i;
1653
    virNetworkIPDefPtr ipdef, ipv4def, ipv6def;
1654
    dnsmasqContext *dctx = NULL;
1655

G
Gene Czarcinski 已提交
1656
    /* if no IP addresses specified, nothing to do */
1657
    if (!virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1658 1659
        return 0;

1660 1661
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1662
        return networkStartDhcpDaemon(driver, network);
1663

G
Gene Czarcinski 已提交
1664
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1665
    if (!(dctx = dnsmasqContextNew(network->def->name,
1666
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1667
        goto cleanup;
1668
    }
G
Gene Czarcinski 已提交
1669 1670 1671 1672 1673 1674

    /* 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;
1675
    for (i = 0;
1676
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET, i));
1677
         i++) {
G
Gene Czarcinski 已提交
1678 1679
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1680 1681
    }

G
Gene Czarcinski 已提交
1682
    ipv6def = NULL;
1683
    for (i = 0;
1684
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1685
         i++) {
G
Gene Czarcinski 已提交
1686 1687
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1688 1689
    }

G
Gene Czarcinski 已提交
1690
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1691
        goto cleanup;
G
Gene Czarcinski 已提交
1692 1693

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1694
        goto cleanup;
1695

G
Gene Czarcinski 已提交
1696
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
J
Ján Tomko 已提交
1697
        goto cleanup;
1698 1699

    if ((ret = dnsmasqSave(dctx)) < 0)
1700
        goto cleanup;
1701 1702

    ret = kill(network->dnsmasqPid, SIGHUP);
1703
 cleanup:
1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715
    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
1716 1717
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1718 1719 1720 1721 1722 1723
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1724
    }
1725
    /* now start dnsmasq if it should be started */
1726
    return networkStartDhcpDaemon(driver, network);
1727 1728
}

G
Gene Czarcinski 已提交
1729 1730 1731 1732 1733 1734
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";

1735 1736 1737
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1738
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1739 1740
    int ret = -1;
    size_t i;
1741
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
1742
    bool v6present = false, dhcp6 = false;
1743 1744

    *configstr = NULL;
1745

G
Gene Czarcinski 已提交
1746
    /* Check if DHCPv6 is needed */
1747
    for (i = 0;
1748
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1749
         i++) {
G
Gene Czarcinski 已提交
1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762
        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;
    }

1763 1764 1765
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1766
    virBufferAsprintf(&configbuf, "interface %s\n"
1767 1768
                      "{\n"
                      "  AdvSendAdvert on;\n"
1769
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1770 1771 1772 1773 1774
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1775 1776

    /* add a section for each IPv6 address in the config */
1777
    for (i = 0;
1778
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1779
         i++) {
1780 1781 1782
        int prefix;
        char *netaddr;

1783
        prefix = virNetworkIPDefPrefix(ipdef);
1784
        if (prefix < 0) {
1785 1786 1787
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1788 1789
            goto cleanup;
        }
1790
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1791
            goto cleanup;
1792
        virBufferAsprintf(&configbuf,
1793
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1794 1795 1796
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1797 1798 1799
        VIR_FREE(netaddr);
    }

1800
    virBufferAddLit(&configbuf, "};\n");
1801

1802
    if (virBufferCheckError(&configbuf) < 0)
1803
        goto cleanup;
1804

1805 1806
    *configstr = virBufferContentAndReset(&configbuf);

1807
    ret = 0;
1808
 cleanup:
1809 1810 1811 1812
    virBufferFreeAndReset(&configbuf);
    return ret;
}

1813
/* write file and return its name (which must be freed by caller) */
1814
static int
1815 1816 1817
networkRadvdConfWrite(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr network,
                      char **configFile)
1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832
{
    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;
1833 1834 1835 1836
        goto cleanup;
    }

    /* construct the filename */
1837
    if (!(*configFile = networkRadvdConfigFileName(driver, network->def->name)))
1838 1839
        goto cleanup;
    /* write the file */
1840
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1841 1842
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1843 1844 1845 1846 1847
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1848
 cleanup:
1849 1850 1851 1852 1853 1854
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1855 1856
networkStartRadvd(virNetworkDriverStatePtr driver,
                  virNetworkObjPtr network)
1857
{
1858
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1859 1860 1861 1862 1863 1864 1865 1866
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1867
    /* Is dnsmasq handling RA? */
1868
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
G
Gene Czarcinski 已提交
1869 1870 1871 1872
        ret = 0;
        goto cleanup;
    }

1873
    if (!virNetworkDefGetIPByIndex(network->def, AF_INET6, 0)) {
1874 1875 1876 1877 1878 1879 1880 1881 1882 1883
        /* 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);
1884 1885 1886
        goto cleanup;
    }

1887
    if (virFileMakePath(driver->pidDir) < 0) {
1888 1889
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1890
                             driver->pidDir);
1891 1892
        goto cleanup;
    }
1893
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1894 1895
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1896
                             driver->radvdStateDir);
1897 1898 1899 1900
        goto cleanup;
    }

    /* construct pidfile name */
1901
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1902
        goto cleanup;
1903
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1904 1905
        goto cleanup;

1906
    if (networkRadvdConfWrite(driver, network, &configfile) < 0)
1907 1908
        goto cleanup;

1909 1910 1911 1912
    /* 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
1913
     * virPidFileRead() below will fail if we use them).
1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928
     * 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;

1929
    if (virPidFileRead(driver->pidDir, radvdpidbase, &network->radvdPid) < 0)
1930 1931 1932
        goto cleanup;

    ret = 0;
1933
 cleanup:
1934
    virObjectUnref(dnsmasq_caps);
1935 1936 1937 1938 1939 1940 1941
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1942
static int
1943 1944
networkRefreshRadvd(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
1945
{
1946
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
G
Gene Czarcinski 已提交
1947 1948 1949
    char *radvdpidbase;

    /* Is dnsmasq handling RA? */
1950 1951
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
        virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1952 1953 1954 1955 1956 1957 1958
        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)) {
1959
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1960 1961 1962 1963 1964
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }
1965
    virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1966

1967 1968
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
1969
        return networkStartRadvd(driver, network);
1970

1971
    if (!virNetworkDefGetIPByIndex(network->def, AF_INET6, 0)) {
1972 1973 1974 1975
        /* no IPv6 addresses, so we don't need to run radvd */
        return 0;
    }

1976
    if (networkRadvdConfWrite(driver, network, NULL) < 0)
1977 1978 1979 1980 1981
        return -1;

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

1982 1983
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1984
static int
1985
networkRestartRadvd(virNetworkObjPtr network)
1986 1987 1988 1989 1990 1991 1992 1993 1994
{
    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
1995
        if ((networkKillDaemon(network->radvdPid, "radvd",
1996 1997 1998
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1999
            virPidFileDelete(driver->pidDir, radvdpidbase);
2000 2001 2002 2003 2004 2005 2006 2007 2008
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

2009 2010
static int
networkRefreshDaemonsHelper(virNetworkObjPtr net,
2011
                            void *opaque)
2012
{
2013
    virNetworkDriverStatePtr driver = opaque;
2014

2015
    virObjectLock(net);
2016 2017 2018
    if (virNetworkObjIsActive(net) &&
        ((net->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
2019 2020
         (net->def->forward.type == VIR_NETWORK_FORWARD_ROUTE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_OPEN))) {
2021 2022 2023 2024 2025 2026
        /* 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.
         */
2027 2028
        networkRefreshDhcpDaemon(driver, net);
        networkRefreshRadvd(driver, net);
2029
    }
2030
    virObjectUnlock(net);
2031 2032 2033
    return 0;
}

2034 2035 2036 2037
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
2038
networkRefreshDaemons(virNetworkDriverStatePtr driver)
2039 2040
{
    VIR_INFO("Refreshing network daemons");
2041 2042
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
2043
                             driver);
2044
}
2045

2046 2047 2048 2049 2050
static int
networkReloadFirewallRulesHelper(virNetworkObjPtr net,
                                 void *opaque ATTRIBUTE_UNUSED)
{

2051
    virObjectLock(net);
2052 2053 2054 2055
    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))) {
2056 2057 2058 2059
        /* Only three of the L3 network types that are configured by
         * libvirt need to have iptables rules reloaded. The 4th L3
         * network type, forward='open', doesn't need this because it
         * has no iptables rules.
2060 2061 2062 2063
         */
        networkRemoveFirewallRules(net->def);
        if (networkAddFirewallRules(net->def) < 0) {
            /* failed to add but already logged */
2064 2065
        }
    }
2066
    virObjectUnlock(net);
2067
    return 0;
2068 2069
}

2070
static void
2071
networkReloadFirewallRules(virNetworkDriverStatePtr driver)
2072
{
2073
    VIR_INFO("Reloading iptables rules");
2074 2075 2076
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
2077 2078
}

2079
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
2080
static int
2081
networkEnableIPForwarding(bool enableIPv4, bool enableIPv6)
2082
{
2083
    int ret = 0;
2084 2085 2086 2087
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
2088
                           &enabled, sizeof(enabled));
2089 2090
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
2091
                           &enabled, sizeof(enabled));
2092
#else
2093
    if (enableIPv4)
2094
        ret = virFileWriteStr(SYSCTL_PATH "/net/ipv4/ip_forward", "1\n", 0);
2095
    if (enableIPv6 && ret == 0)
2096 2097
        ret = virFileWriteStr(SYSCTL_PATH "/net/ipv6/conf/all/forwarding", "1\n", 0);

2098
#endif
2099
    return ret;
2100 2101
}

2102 2103
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
2104 2105 2106
{
    char *field = NULL;
    int ret = -1;
2107
    bool enableIPv6 =  !!virNetworkDefGetIPByIndex(network->def, AF_INET6, 0);
2108

2109 2110 2111 2112 2113 2114 2115
    /* 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;
2116

2117 2118
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
2119 2120
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
2121 2122 2123
        ret = 0;
        goto cleanup;
    }
2124

2125 2126 2127 2128 2129
    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;
2130
    }
2131
    VIR_FREE(field);
2132

2133 2134
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
2135 2136 2137 2138 2139 2140
     */

    /* 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",
2141
                    network->def->bridge) < 0)
2142 2143
        goto cleanup;

2144
    if (virFileWriteStr(field, "0", 0) < 0) {
2145
        virReportSystemError(errno,
2146 2147 2148 2149 2150
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

2151 2152 2153 2154
    /* 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",
2155
                    network->def->bridge) < 0)
2156 2157
        goto cleanup;

2158
    if (virFileWriteStr(field, "0", 0) < 0) {
2159
        virReportSystemError(errno,
2160
                             _("cannot disable %s"), field);
2161 2162 2163 2164
        goto cleanup;
    }

    ret = 0;
2165
 cleanup:
2166 2167 2168 2169
    VIR_FREE(field);
    return ret;
}

2170
/* add an IP address to a bridge */
2171
static int
D
Daniel P. Berrange 已提交
2172
networkAddAddrToBridge(virNetworkObjPtr network,
2173
                       virNetworkIPDefPtr ipdef)
2174
{
2175
    int prefix = virNetworkIPDefPrefix(ipdef);
2176 2177

    if (prefix < 0) {
2178 2179 2180
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
2181 2182 2183
        return -1;
    }

2184 2185
    if (virNetDevIPAddrAdd(network->def->bridge,
                           &ipdef->address, NULL, prefix) < 0)
2186 2187 2188 2189 2190
        return -1;

    return 0;
}

2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213

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;
}


2214 2215 2216
/* add an IP (static) route to a bridge */
static int
networkAddRouteToBridge(virNetworkObjPtr network,
2217
                        virNetDevIPRoutePtr routedef)
2218
{
2219 2220 2221 2222
    int prefix = virNetDevIPRouteGetPrefix(routedef);
    unsigned int metric = virNetDevIPRouteGetMetric(routedef);
    virSocketAddrPtr addr = virNetDevIPRouteGetAddress(routedef);
    virSocketAddrPtr gateway = virNetDevIPRouteGetGateway(routedef);
2223 2224 2225 2226 2227 2228 2229 2230 2231

    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;
    }

2232 2233
    if (virNetDevIPRouteAdd(network->def->bridge, addr,
                            prefix, gateway, metric) < 0) {
2234 2235 2236 2237 2238
        return -1;
    }
    return 0;
}

2239 2240 2241
static int
networkWaitDadFinish(virNetworkObjPtr network)
{
2242
    virNetworkIPDefPtr ipdef;
2243 2244 2245 2246 2247 2248
    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);

2249
    while ((ipdef = virNetworkDefGetIPByIndex(network->def,
2250 2251 2252 2253 2254 2255
                                              AF_INET6, naddrs))) {
        addr = &ipdef->address;
        if (VIR_APPEND_ELEMENT_COPY(addrs, naddrs, addr) < 0)
            goto cleanup;
    }

2256
    ret = (naddrs == 0) ? 0 : virNetDevIPWaitDadFinish(addrs, naddrs);
2257 2258 2259 2260 2261 2262 2263 2264

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

2265
static int
2266 2267
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
                           virNetworkObjPtr network)
2268
{
2269
    size_t i;
2270
    bool v4present = false, v6present = false;
2271
    virErrorPtr save_err = NULL;
2272
    virNetworkIPDefPtr ipdef;
2273
    virNetDevIPRoutePtr routedef;
2274
    char *macTapIfName = NULL;
M
Michal Privoznik 已提交
2275
    char *macMapFile = NULL;
2276
    int tapfd = -1;
2277

2278
    /* Check to see if any network IP collides with an existing route */
2279
    if (networkCheckRouteCollision(network->def) < 0)
2280 2281
        return -1;

2282
    /* Create and configure the bridge device */
2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296
    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;
    }
2297
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
2298 2299
        return -1;

2300 2301 2302 2303 2304 2305 2306 2307
    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);
2308
        if (!macTapIfName)
2309
            goto err0;
2310
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2311
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
2312
                                           &macTapIfName, &network->def->mac,
2313
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2314
                                           NULL, network->def->mtu, NULL,
2315 2316 2317
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2318 2319 2320 2321 2322
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

M
Michal Privoznik 已提交
2323 2324 2325 2326
    if (!(macMapFile = networkMacMgrFileName(driver, network->def->bridge)) ||
        !(network->macmap = virMacMapNew(macMapFile)))
        goto err1;

2327
    /* Set bridge options */
2328 2329 2330 2331

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2332
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2333
                                   network->def->delay * 1000) < 0)
2334
        goto err1;
2335

2336
    if (virNetDevBridgeSetSTP(network->def->bridge,
2337
                              network->def->stp ? true : false) < 0)
2338
        goto err1;
2339

2340 2341 2342 2343
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2344
        goto err1;
2345

2346
    /* Add "once per network" rules */
2347 2348
    if (network->def->forward.type != VIR_NETWORK_FORWARD_OPEN &&
        networkAddFirewallRules(network->def) < 0)
2349 2350
        goto err1;

2351
    for (i = 0;
2352
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i));
2353
         i++) {
2354
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2355
            v4present = true;
2356
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2357
            v6present = true;
2358

2359
        /* Add the IP address/netmask to the bridge */
2360
        if (networkAddAddrToBridge(network, ipdef) < 0)
2361
            goto err2;
2362 2363
    }

2364 2365 2366
    if (networkStartHandleMACTableManagerMode(network, macTapIfName) < 0)
        goto err2;

2367
    /* Bring up the bridge interface */
2368
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2369
        goto err2;
2370

2371
    for (i = 0; i < network->def->nroutes; i++) {
2372 2373 2374
        virSocketAddrPtr gateway = NULL;

        routedef = network->def->routes[i];
2375
        gateway = virNetDevIPRouteGetGateway(routedef);
2376

2377 2378 2379
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2380
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2381 2382 2383 2384 2385 2386 2387
            if (networkAddRouteToBridge(network, routedef) < 0) {
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2388
    /* If forward.type != NONE, turn on global IP forwarding */
2389
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE) {
2390
        if (v6present && !virNetDevIPCheckIPv6Forwarding())
2391 2392 2393 2394 2395 2396 2397
            goto err3; /* Precise error message already provided */

        if (networkEnableIPForwarding(v4present, v6present) < 0) {
            virReportSystemError(errno, "%s",
                                 _("failed to enable IP forwarding"));
            goto err3;
        }
2398 2399
    }

2400

2401
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2402
    if ((v4present || v6present) &&
2403
        networkStartDhcpDaemon(driver, network) < 0)
2404
        goto err3;
2405

2406
    /* start radvd if there are any ipv6 addresses */
2407
    if (v6present && networkStartRadvd(driver, network) < 0)
2408 2409
        goto err4;

2410 2411 2412 2413 2414 2415 2416 2417
    /* 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.
2418 2419 2420 2421 2422 2423 2424
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
            goto err4;
        VIR_FORCE_CLOSE(tapfd);
    }

2425
    if (virNetDevBandwidthSet(network->def->bridge,
2426
                              network->def->bandwidth, true) < 0)
2427 2428
        goto err5;

2429
    VIR_FREE(macTapIfName);
M
Michal Privoznik 已提交
2430
    VIR_FREE(macMapFile);
2431 2432 2433

    return 0;

2434
 err5:
2435 2436
    if (network->def->bandwidth)
       virNetDevBandwidthClear(network->def->bridge);
2437

2438 2439 2440 2441
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2442 2443 2444 2445 2446
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2447 2448 2449
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2450
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2451

2452 2453 2454
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2455 2456
    if (network->def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(network->def);
2457 2458

 err1:
2459 2460 2461
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2462
    if (macTapIfName) {
2463
        VIR_FORCE_CLOSE(tapfd);
2464
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2465 2466
        VIR_FREE(macTapIfName);
    }
M
Michal Privoznik 已提交
2467
    VIR_FREE(macMapFile);
2468 2469

 err0:
2470 2471
    if (!save_err)
        save_err = virSaveLastError();
2472
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2473

2474 2475 2476 2477
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2478
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2479 2480 2481
    return -1;
}

2482 2483 2484
static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
                              virNetworkObjPtr network)
2485
{
2486 2487
    if (network->def->bandwidth)
        virNetDevBandwidthClear(network->def->bridge);
2488

2489 2490
    if (!virObjectUnref(network->macmap))
        network->macmap = NULL;
M
Michal Privoznik 已提交
2491

2492 2493 2494 2495 2496
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2497
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2498
            virPidFileDelete(driver->pidDir, radvdpidbase);
2499 2500 2501 2502
            VIR_FREE(radvdpidbase);
        }
    }

2503 2504 2505
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2506
    if (network->def->mac_specified) {
2507
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2508
        if (macTapIfName) {
2509
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2510 2511 2512 2513
            VIR_FREE(macTapIfName);
        }
    }

2514
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2515

2516 2517
    if (network->def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(network->def);
2518

2519
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2520

2521
    /* See if its still alive and really really kill it */
2522
    if (network->dnsmasqPid > 0 &&
2523
        (kill(network->dnsmasqPid, 0) == 0))
2524 2525
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2526 2527 2528 2529 2530 2531

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

2532 2533 2534
    return 0;
}

2535

2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556
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;
}


2557 2558 2559 2560 2561 2562 2563 2564 2565
/* 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;
2566
    unsigned int maxVirtFns = 0;
2567 2568 2569 2570 2571 2572
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2573 2574 2575
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2576 2577
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev, &vfNames,
                                      &virtFns, &numVirtFns, &maxVirtFns)) < 0) {
2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623
        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:
2624
        case VIR_NETWORK_FORWARD_OPEN:
2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664
        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;
}


2665
static int
2666
networkStartNetworkExternal(virNetworkObjPtr network)
2667 2668
{
    /* put anything here that needs to be done each time a network of
2669
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2670 2671 2672
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2673
    return networkCreateInterfacePool(network->def);
2674 2675
}

2676
static int networkShutdownNetworkExternal(virNetworkObjPtr network ATTRIBUTE_UNUSED)
2677 2678
{
    /* put anything here that needs to be done each time a network of
2679
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2680 2681 2682 2683 2684 2685 2686
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2687 2688
networkStartNetwork(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
2689
{
2690 2691 2692
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2693 2694

    if (virNetworkObjIsActive(network)) {
2695 2696
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2697
        return ret;
2698 2699
    }

2700 2701 2702
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2703
    if (virNetworkObjSetDefTransient(network, true) < 0)
2704
        goto cleanup;
2705

2706 2707
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2708
    if (networkRunHook(network, NULL, NULL,
2709 2710 2711 2712
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2713
    switch (network->def->forward.type) {
2714 2715 2716 2717

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2718
    case VIR_NETWORK_FORWARD_OPEN:
2719
        if (networkStartNetworkVirtual(driver, network) < 0)
2720
            goto cleanup;
2721 2722 2723
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2724 2725 2726 2727 2728 2729 2730 2731 2732
        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).
         */
2733 2734
        ATTRIBUTE_FALLTHROUGH;

2735 2736 2737
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2738
    case VIR_NETWORK_FORWARD_HOSTDEV:
2739
        if (networkStartNetworkExternal(network) < 0)
2740
            goto cleanup;
2741 2742 2743
        break;
    }

2744
    /* finally we can call the 'started' hook script if any */
2745
    if (networkRunHook(network, NULL, NULL,
2746 2747 2748 2749
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2750 2751 2752
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2753
    VIR_DEBUG("Writing network status to disk");
2754
    if (virNetworkObjSaveStatus(driver->stateDir, network) < 0)
2755
        goto cleanup;
2756 2757

    network->active = 1;
2758 2759
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2760

2761
 cleanup:
2762
    if (ret < 0) {
2763
        virNetworkObjUnsetDefTransient(network);
2764 2765
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2766
        networkShutdownNetwork(driver, network);
2767 2768 2769 2770 2771 2772 2773
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2774 2775 2776
static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
2777 2778 2779 2780 2781 2782 2783 2784 2785
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

2786
    stateFile = virNetworkConfigFile(driver->stateDir,
2787
                                     network->def->name);
2788 2789 2790 2791 2792 2793
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2794
    switch (network->def->forward.type) {
2795 2796 2797 2798

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2799
    case VIR_NETWORK_FORWARD_OPEN:
2800
        ret = networkShutdownNetworkVirtual(driver, network);
2801 2802 2803
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2804 2805 2806 2807 2808 2809 2810 2811
        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).
         */
2812 2813
        ATTRIBUTE_FALLTHROUGH;

2814 2815 2816
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2817
    case VIR_NETWORK_FORWARD_HOSTDEV:
2818
        ret = networkShutdownNetworkExternal(network);
2819 2820 2821
        break;
    }

2822
    /* now that we know it's stopped call the hook if present */
2823
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2824 2825
                   VIR_HOOK_SUBOP_END);

2826
    network->active = 0;
2827
    virNetworkObjUnsetDefTransient(network);
2828
    return ret;
2829 2830 2831
}


2832
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2833 2834
                                         const unsigned char *uuid)
{
2835
    virNetworkDriverStatePtr driver = networkGetDriver();
2836 2837
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2838

2839
    network = virNetworkObjFindByUUID(driver->networks, uuid);
2840
    if (!network) {
2841 2842
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2843
        virReportError(VIR_ERR_NO_NETWORK,
2844 2845
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2846
        goto cleanup;
2847 2848
    }

2849 2850 2851
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2852 2853
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2854
 cleanup:
2855
    virNetworkObjEndAPI(&network);
2856
    return ret;
2857 2858
}

2859
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2860 2861
                                         const char *name)
{
2862
    virNetworkDriverStatePtr driver = networkGetDriver();
2863 2864 2865
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2866
    network = virNetworkObjFindByName(driver->networks, name);
2867
    if (!network) {
2868 2869
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2870
        goto cleanup;
2871 2872
    }

2873 2874 2875
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2876 2877
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2878
 cleanup:
2879
    virNetworkObjEndAPI(&network);
2880
    return ret;
2881 2882
}

2883 2884
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2885
    virNetworkDriverStatePtr driver = networkGetDriver();
2886
    int nactive;
2887

2888 2889 2890
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2891 2892 2893 2894
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
2895

2896 2897 2898
    return nactive;
}

2899 2900 2901 2902 2903
static int networkConnectListNetworks(virConnectPtr conn,
                                      char **const names,
                                      int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2904
    int got = 0;
2905

2906 2907 2908
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2909 2910 2911 2912
    got = virNetworkObjListGetNames(driver->networks,
                                    true, names, nnames,
                                    virConnectListNetworksCheckACL,
                                    conn);
2913

2914 2915 2916
    return got;
}

2917 2918
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2919
    virNetworkDriverStatePtr driver = networkGetDriver();
2920
    int ninactive = 0;
2921

2922 2923 2924
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2925 2926 2927 2928
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
2929

2930 2931 2932
    return ninactive;
}

2933 2934 2935
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2936
    int got = 0;
2937

2938 2939 2940
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2941 2942 2943 2944
    got = virNetworkObjListGetNames(driver->networks,
                                    false, names, nnames,
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
2945 2946 2947
    return got;
}

2948
static int
2949 2950 2951
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2952
{
2953
    virNetworkDriverStatePtr driver = networkGetDriver();
2954 2955 2956 2957
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2958 2959 2960
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2961
    ret = virNetworkObjListExport(conn, driver->networks, nets,
2962 2963
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2964

2965
 cleanup:
2966 2967
    return ret;
}
2968

2969 2970 2971 2972 2973 2974 2975 2976
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
2977
    virNetworkDriverStatePtr driver = networkGetDriver();
2978 2979 2980 2981 2982 2983
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
2984
                                       net, eventID, callback,
2985 2986 2987
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2988
 cleanup:
2989 2990 2991 2992 2993 2994 2995
    return ret;
}

static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
2996
    virNetworkDriverStatePtr driver = networkGetDriver();
2997 2998 2999 3000 3001
    int ret = -1;

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

3002 3003 3004 3005 3006 3007
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
3008

3009
 cleanup:
3010 3011 3012
    return ret;
}

3013 3014 3015 3016 3017
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

3018 3019
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3020 3021 3022 3023

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

3024 3025
    ret = virNetworkObjIsActive(obj);

3026
 cleanup:
3027
    virNetworkObjEndAPI(&obj);
3028 3029 3030 3031 3032 3033 3034 3035
    return ret;
}

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

3036 3037
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3038 3039 3040 3041

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

3042 3043
    ret = obj->persistent;

3044
 cleanup:
3045
    virNetworkObjEndAPI(&obj);
3046 3047 3048 3049
    return ret;
}


3050 3051
/*
 * networkFindUnusedBridgeName() - try to find a bridge name that is
3052 3053 3054
 * 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.
3055 3056 3057 3058 3059 3060 3061 3062
 */
static int
networkFindUnusedBridgeName(virNetworkObjListPtr nets,
                            virNetworkDefPtr def)
{

    int ret = -1, id = 0;
    char *newname = NULL;
3063 3064 3065 3066 3067
    const char *templ = "virbr%d";
    const char *p;

    if (def->bridge &&
        (p = strchr(def->bridge, '%')) == strrchr(def->bridge, '%') &&
3068
        p && p[1] == 'd')
3069
        templ = def->bridge;
3070 3071 3072 3073

    do {
        if (virAsprintf(&newname, templ, id) < 0)
            goto cleanup;
3074 3075 3076 3077 3078
        /* 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).
         */
3079
        if (!(virNetworkObjBridgeInUse(nets, newname, def->name) ||
3080
              virNetDevExists(newname) == 1)) {
3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114
            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")) {
3115
        if (virNetworkObjBridgeInUse(nets, def->bridge, def->name)) {
3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132
            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;
}


3133
static int
3134
networkValidate(virNetworkDriverStatePtr driver,
3135
                virNetworkDefPtr def)
3136
{
3137
    size_t i, j;
3138 3139
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
3140
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
3141
    bool ipv4def = false, ipv6def = false;
3142
    bool bandwidthAllowed = true;
3143
    bool usesInterface = false, usesAddress = false;
3144

3145 3146 3147
    if (virXMLCheckIllegalChars("name", def->name, "\n") < 0)
        return -1;

3148 3149 3150
    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
3151 3152
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
3153 3154
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE ||
        def->forward.type == VIR_NETWORK_FORWARD_OPEN) {
3155

3156 3157 3158 3159
        /* 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)
3160 3161 3162
            return -1;

        virNetworkSetBridgeMacAddr(def);
3163 3164
    } else {
        /* They are also the only types that currently support setting
3165 3166
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
3167
         */
3168 3169 3170 3171 3172 3173 3174 3175
        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;
        }
3176
        if (virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0)) {
3177 3178 3179 3180
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3181
                           virNetworkForwardTypeToString(def->forward.type));
3182 3183
            return -1;
        }
3184
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
3185 3186 3187 3188
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3189
                           virNetworkForwardTypeToString(def->forward.type));
3190 3191 3192 3193 3194 3195 3196
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3197
                           virNetworkForwardTypeToString(def->forward.type));
3198 3199
            return -1;
        }
3200 3201 3202 3203 3204 3205 3206 3207
        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;
        }
3208
        bandwidthAllowed = false;
3209 3210
    }

3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223
    /* 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++) {
3224 3225 3226 3227
        virNetworkForwardIfDefPtr iface = &def->forward.ifs[i];
        char *sysfs_path = NULL;

        switch ((virNetworkForwardHostdevDeviceType) iface->type) {
3228 3229
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV:
            usesInterface = true;
3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240

            if (def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("hostdev network '%s' lists '%s' "
                                 "in the device pool, but hostdev "
                                 "networks require all devices to "
                                 "be listed by PCI address, not "
                                 "network device name"),
                               def->name, iface->device.dev);
                return -1;
            }
3241
            break;
3242 3243

        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI: {
3244
            usesAddress = true;
3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269

            if (def->forward.type != VIR_NETWORK_FORWARD_HOSTDEV) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has forward mode '%s' "
                                 " but lists a device by PCI address "
                                 "in the device pool. This is only "
                                 "supported for networks with forward "
                                 "mode 'hostdev'"),
                               def->name,
                               virNetworkForwardTypeToString(def->forward.type));
                return -1;
            }

            if (virPCIDeviceAddressGetSysfsFile(&iface->device.pci, &sysfs_path) < 0)
                return -1;

            if (!virPCIIsVirtualFunction(sysfs_path)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("device '%s' in network '%s' is not "
                                 "an SR-IOV Virtual Function"),
                               sysfs_path, def->name);
                VIR_FREE(sysfs_path);
                return -1;
            }
            VIR_FREE(sysfs_path);
3270
            break;
3271 3272
        }

3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285
        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
3286 3287 3288
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
3289
    for (i = 0;
3290
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
3291
         i++) {
G
Gene Czarcinski 已提交
3292 3293 3294 3295 3296
        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 -- "
3297 3298
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315
                    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;
                }
3316 3317 3318
            }
        }
    }
3319 3320 3321 3322 3323 3324

    /* 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.
     */

3325 3326 3327
    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
3328 3329
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
3330
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH));
3331 3332

    vlanUsed = def->vlan.nTags > 0;
3333 3334
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3335 3336 3337 3338 3339
            /* 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.
             */
3340
            if (def->portGroups[i].virtPortProfile) {
3341
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3342
                    def->portGroups[i].virtPortProfile->virtPortType
3343 3344 3345 3346 3347 3348 3349
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3350
        }
3351
        if (def->portGroups[i].isDefault) {
3352 3353 3354 3355 3356
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3357
                               def->name, defaultPortGroup->name,
3358
                               def->portGroups[i].name);
3359
                return -1;
3360
            }
3361
            defaultPortGroup = &def->portGroups[i];
3362
        }
3363 3364 3365 3366 3367 3368 3369 3370 3371
        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;
            }
        }
3372 3373 3374 3375 3376 3377 3378 3379
        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;
        }
3380
    }
3381 3382 3383 3384 3385 3386 3387
    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.
         */
3388 3389 3390 3391 3392 3393
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407

    if (def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
        for (i = 0; i < def->nPortGroups; i++) {
            if (def->portGroups[i].bandwidth) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("unsupported <bandwidth> element "
                                 "in <portgroup name='%s'> of "
                                 "network '%s' with forward mode='%s'"),
                               def->portGroups[i].name, def->name,
                               virNetworkForwardTypeToString(def->forward.type));
                return -1;
            }
        }
    }
3408 3409 3410
    return 0;
}

3411 3412
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
3413
    virNetworkDriverStatePtr driver = networkGetDriver();
3414
    virNetworkDefPtr def;
3415
    virNetworkObjPtr network = NULL;
3416
    virNetworkPtr ret = NULL;
3417
    virObjectEventPtr event = NULL;
3418

3419
    if (!(def = virNetworkDefParseString(xml)))
3420
        goto cleanup;
3421

3422 3423 3424
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3425
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3426
        goto cleanup;
3427

3428 3429 3430
    /* 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.
3431
     */
3432 3433 3434
    if (!(network = virNetworkObjAssignDef(driver->networks, def,
                                           VIR_NETWORK_OBJ_LIST_ADD_LIVE |
                                           VIR_NETWORK_OBJ_LIST_ADD_CHECK_LIVE)))
3435 3436
        goto cleanup;
    def = NULL;
3437

3438
    if (networkStartNetwork(driver, network) < 0) {
3439
        virNetworkObjRemoveInactive(driver->networks, network);
3440
        goto cleanup;
3441 3442
    }

3443 3444
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3445 3446
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3447

3448
    VIR_INFO("Creating network '%s'", network->def->name);
3449 3450
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

3451
 cleanup:
3452
    virNetworkDefFree(def);
3453 3454
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3455
    virNetworkObjEndAPI(&network);
3456
    return ret;
3457 3458
}

3459 3460
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
3461
    virNetworkDriverStatePtr driver = networkGetDriver();
3462
    virNetworkDefPtr def = NULL;
3463
    bool freeDef = true;
3464
    virNetworkObjPtr network = NULL;
3465
    virNetworkPtr ret = NULL;
3466
    virObjectEventPtr event = NULL;
3467

3468
    if (!(def = virNetworkDefParseString(xml)))
3469
        goto cleanup;
3470

3471 3472 3473
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3474
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3475
        goto cleanup;
3476

3477
    if (!(network = virNetworkObjAssignDef(driver->networks, def, 0)))
J
Ján Tomko 已提交
3478
        goto cleanup;
3479

3480
    /* def was assigned to network object */
3481
    freeDef = false;
3482 3483

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3484
        if (!virNetworkObjIsActive(network)) {
3485
            virNetworkObjRemoveInactive(driver->networks, network);
3486 3487
            goto cleanup;
        }
3488 3489 3490 3491
        /* if network was active already, just undo new persistent
         * definition by making it transient.
         * XXX - this isn't necessarily the correct thing to do.
         */
3492
        virNetworkObjUpdateAssignDef(network, NULL, false);
3493 3494 3495
        goto cleanup;
    }

3496
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3497 3498
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3499

3500 3501
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
3502

3503
 cleanup:
3504 3505
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3506
    if (freeDef)
J
Ján Tomko 已提交
3507
        virNetworkDefFree(def);
3508
    virNetworkObjEndAPI(&network);
3509
    return ret;
3510 3511
}

3512
static int
3513 3514
networkUndefine(virNetworkPtr net)
{
3515
    virNetworkDriverStatePtr driver = networkGetDriver();
3516
    virNetworkObjPtr network;
3517
    int ret = -1;
3518
    bool active = false;
3519
    virObjectEventPtr event = NULL;
3520

3521
    if (!(network = networkObjFromNetwork(net)))
3522
        goto cleanup;
3523

3524 3525 3526
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3527 3528
    if (virNetworkObjIsActive(network))
        active = true;
3529

3530 3531 3532 3533 3534 3535
    if (!network->persistent) {
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                       _("can't undefine transient network"));
        goto cleanup;
    }

3536
    /* remove autostart link */
3537 3538 3539
    if (virNetworkObjDeleteConfig(driver->networkConfigDir,
                                  driver->networkAutostartDir,
                                  network) < 0)
3540
        goto cleanup;
3541

3542 3543
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3544 3545
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3546

3547
    VIR_INFO("Undefining network '%s'", network->def->name);
3548
    if (!active) {
3549
        if (networkRemoveInactive(driver, network) < 0)
3550
            goto cleanup;
3551 3552 3553 3554 3555
    } else {

        /* if the network still exists, it was active, and we need to make
         * it transient (by deleting the persistent def)
         */
3556
        virNetworkObjUpdateAssignDef(network, NULL, false);
3557 3558
    }

3559
    ret = 0;
3560

3561
 cleanup:
3562 3563
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3564
    virNetworkObjEndAPI(&network);
3565
    return ret;
3566 3567
}

3568 3569 3570 3571 3572 3573 3574 3575
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3576
    virNetworkDriverStatePtr driver = networkGetDriver();
3577
    virNetworkObjPtr network = NULL;
3578 3579
    int isActive, ret = -1;
    size_t i;
3580
    virNetworkIPDefPtr ipdef;
3581
    bool oldDhcpActive = false;
3582
    bool needFirewallRefresh = false;
3583

3584 3585 3586 3587 3588

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3589
    if (!(network = networkObjFromNetwork(net)))
3590 3591
        goto cleanup;

3592 3593 3594
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3595
    /* see if we are listening for dhcp pre-modification */
3596
    for (i = 0;
3597
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET, i));
3598
         i++) {
3599 3600 3601 3602 3603 3604
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3605 3606
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3607
     */
3608
    isActive = virNetworkObjIsActive(network);
3609 3610
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3611 3612 3613 3614 3615 3616 3617
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634
    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).
                 */
3635 3636 3637 3638
                if (network->def->forward.type != VIR_NETWORK_FORWARD_OPEN) {
                    networkRemoveFirewallRules(network->def);
                    needFirewallRefresh = true;
                }
3639 3640 3641 3642 3643 3644 3645
                break;
            default:
                break;
            }
        }
    }

3646
    /* update the network config in memory/on disk */
3647 3648
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3649
            ignore_value(networkAddFirewallRules(network->def));
3650 3651 3652
        goto cleanup;
    }

3653
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672
        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 ||
3673 3674 3675 3676 3677 3678 3679 3680
            section == VIR_NETWORK_SECTION_IP_DHCP_RANGE ||
            section == VIR_NETWORK_SECTION_DNS_TXT ||
            section == VIR_NETWORK_SECTION_DNS_SRV) {
            /* these sections all change things on the dnsmasq
             * commandline (i.e. in the .conf file), so we need to
             * kill and restart dnsmasq, because dnsmasq sets its uid
             * to "nobody" after it starts, and is unable to re-read
             * the conf file (owned by root, mode 600)
3681
             */
3682
            if (networkRestartDhcpDaemon(driver, network) < 0)
3683 3684
                goto cleanup;

3685 3686 3687 3688 3689 3690 3691 3692
        } 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;

3693
            for (i = 0;
3694
                 (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET, i));
3695
                 i++) {
3696 3697 3698 3699 3700 3701 3702
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3703 3704
                 networkRestartDhcpDaemon(driver, network) < 0) ||
                networkRefreshDhcpDaemon(driver, network) < 0) {
3705 3706 3707
                goto cleanup;
            }

3708 3709 3710 3711
        } else if (section == VIR_NETWORK_SECTION_DNS_HOST) {
            /* this section only changes data in an external file
             * (not the .conf file) so we can just update the config
             * files and send SIGHUP to dnsmasq.
3712
             */
3713
            if (networkRefreshDhcpDaemon(driver, network) < 0)
3714 3715 3716 3717 3718 3719 3720 3721
                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.
             */
3722
            if (networkRefreshRadvd(driver, network) < 0)
3723 3724 3725 3726
                goto cleanup;
        }

        /* save current network state to disk */
3727
        if ((ret = virNetworkObjSaveStatus(driver->stateDir, network)) < 0)
3728 3729
            goto cleanup;
    }
3730 3731 3732 3733 3734 3735

    /* call the 'updated' network hook script */
    if (networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_UPDATED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

3736
    ret = 0;
3737
 cleanup:
3738
    virNetworkObjEndAPI(&network);
3739 3740 3741
    return ret;
}

3742 3743
static int networkCreate(virNetworkPtr net)
{
3744
    virNetworkDriverStatePtr driver = networkGetDriver();
3745 3746
    virNetworkObjPtr network;
    int ret = -1;
3747
    virObjectEventPtr event = NULL;
3748

3749
    if (!(network = networkObjFromNetwork(net)))
3750
        goto cleanup;
3751

3752 3753 3754
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3755
    if ((ret = networkStartNetwork(driver, network)) < 0)
3756
        goto cleanup;
3757

3758 3759
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3760 3761
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3762

3763
 cleanup:
3764 3765
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3766
    virNetworkObjEndAPI(&network);
3767
    return ret;
3768 3769
}

3770 3771
static int networkDestroy(virNetworkPtr net)
{
3772
    virNetworkDriverStatePtr driver = networkGetDriver();
3773 3774
    virNetworkObjPtr network;
    int ret = -1;
3775
    virObjectEventPtr event = NULL;
3776

3777
    if (!(network = networkObjFromNetwork(net)))
3778
        goto cleanup;
3779

3780 3781 3782
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3783
    if (!virNetworkObjIsActive(network)) {
3784
        virReportError(VIR_ERR_OPERATION_INVALID,
3785 3786
                       _("network '%s' is not active"),
                       network->def->name);
3787 3788 3789
        goto cleanup;
    }

3790
    if ((ret = networkShutdownNetwork(driver, network)) < 0)
3791 3792
        goto cleanup;

3793 3794
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3795 3796
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3797

3798
    if (!network->persistent &&
3799
        networkRemoveInactive(driver, network) < 0) {
3800 3801
        ret = -1;
        goto cleanup;
3802
    }
3803

3804
 cleanup:
3805 3806
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3807
    virNetworkObjEndAPI(&network);
3808 3809 3810
    return ret;
}

3811
static char *networkGetXMLDesc(virNetworkPtr net,
3812
                               unsigned int flags)
3813
{
3814
    virNetworkObjPtr network;
3815
    virNetworkDefPtr def;
3816
    char *ret = NULL;
3817

3818
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3819

3820 3821
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3822

3823 3824 3825
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3826 3827 3828 3829 3830 3831
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3832

3833
 cleanup:
3834
    virNetworkObjEndAPI(&network);
3835
    return ret;
3836 3837 3838
}

static char *networkGetBridgeName(virNetworkPtr net) {
3839 3840 3841
    virNetworkObjPtr network;
    char *bridge = NULL;

3842 3843
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3844

3845 3846 3847
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3848
    if (!(network->def->bridge)) {
3849 3850 3851
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3852 3853 3854
        goto cleanup;
    }

3855
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3856

3857
 cleanup:
3858
    virNetworkObjEndAPI(&network);
3859 3860 3861 3862
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
J
Ján Tomko 已提交
3863
                               int *autostart)
3864
{
3865 3866
    virNetworkObjPtr network;
    int ret = -1;
3867

3868 3869
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3870

3871 3872 3873
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3874
    *autostart = network->autostart;
3875
    ret = 0;
3876

3877
 cleanup:
3878
    virNetworkObjEndAPI(&network);
3879
    return ret;
3880 3881 3882
}

static int networkSetAutostart(virNetworkPtr net,
3883 3884
                               int autostart)
{
3885
    virNetworkDriverStatePtr driver = networkGetDriver();
3886
    virNetworkObjPtr network;
3887
    char *configFile = NULL, *autostartLink = NULL;
3888
    int ret = -1;
3889

3890

3891
    if (!(network = networkObjFromNetwork(net)))
3892
        goto cleanup;
3893

3894 3895 3896
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3897
    if (!network->persistent) {
3898 3899
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3900 3901 3902
        goto cleanup;
    }

3903 3904
    autostart = (autostart != 0);

3905
    if (network->autostart != autostart) {
3906
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3907
            goto cleanup;
3908
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3909 3910
            goto cleanup;

3911
        if (autostart) {
3912
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3913
                virReportSystemError(errno,
3914 3915
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3916 3917
                goto cleanup;
            }
3918

3919
            if (symlink(configFile, autostartLink) < 0) {
3920
                virReportSystemError(errno,
3921
                                     _("Failed to create symlink '%s' to '%s'"),
3922
                                     autostartLink, configFile);
3923 3924 3925
                goto cleanup;
            }
        } else {
3926
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3927
                virReportSystemError(errno,
3928
                                     _("Failed to delete symlink '%s'"),
3929
                                     autostartLink);
3930 3931
                goto cleanup;
            }
3932 3933
        }

3934
        network->autostart = autostart;
3935
    }
3936
    ret = 0;
3937

3938
 cleanup:
3939 3940
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3941
    virNetworkObjEndAPI(&network);
3942
    return ret;
3943 3944
}

3945
static int
3946 3947 3948 3949
networkGetDHCPLeases(virNetworkPtr network,
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
3950
{
3951
    virNetworkDriverStatePtr driver = networkGetDriver();
3952 3953 3954
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
3955
    ssize_t size = 0;
3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966
    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;
3967
    virNetworkIPDefPtr ipdef_tmp = NULL;
3968 3969
    virNetworkDHCPLeasePtr lease = NULL;
    virNetworkDHCPLeasePtr *leases_ret = NULL;
3970
    virNetworkObjPtr obj;
3971
    virMacAddr mac_addr;
3972 3973 3974

    virCheckFlags(0, -1);

3975 3976 3977 3978 3979 3980
    /* only to check if the MAC is valid */
    if (mac && virMacAddrParse(mac, &mac_addr) < 0) {
        virReportError(VIR_ERR_INVALID_MAC, "%s", mac);
        return -1;
    }

3981 3982 3983 3984 3985
    if (!(obj = networkObjFromNetwork(network)))
        return -1;

    if (virNetworkGetDHCPLeasesEnsureACL(network->conn, obj->def) < 0)
        goto cleanup;
3986 3987

    /* Retrieve custom leases file location */
3988
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(driver, obj->def->bridge);
3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031

    /* 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;
        }

4032
        if (mac && virMacAddrCompare(mac, mac_tmp))
4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073
            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)) {
4074
                    lease->prefix = virSocketAddrGetIPPrefix(&ipdef_tmp->address,
4075 4076 4077 4078 4079 4080 4081 4082
                                                             &ipdef_tmp->netmask,
                                                             ipdef_tmp->prefix);
                    break;
                }
            }

            if ((VIR_STRDUP(lease->mac, mac_tmp) < 0) ||
                (VIR_STRDUP(lease->ipaddr, ip_tmp) < 0) ||
4083
                (VIR_STRDUP(lease->iface, obj->def->bridge) < 0))
4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115
                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);
4116
    VIR_FREE(lease_entries);
4117 4118
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
4119

4120
    virNetworkObjEndAPI(&obj);
4121

4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132
    return rv;

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

4133 4134

static virNetworkDriver networkDriver = {
4135
    .name = "bridge",
4136 4137 4138 4139 4140
    .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 */
4141 4142
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
4143 4144
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
4145 4146
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
4147
    .networkUndefine = networkUndefine, /* 0.2.0 */
4148
    .networkUpdate = networkUpdate, /* 0.10.2 */
4149
    .networkCreate = networkCreate, /* 0.2.0 */
4150 4151 4152 4153 4154 4155 4156
    .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 */
4157
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
4158 4159 4160
};

static virStateDriver networkStateDriver = {
4161
    .name = "bridge",
4162
    .stateInitialize  = networkStateInitialize,
4163
    .stateAutoStart  = networkStateAutoStart,
4164 4165
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
4166 4167
};

4168 4169
int networkRegister(void)
{
4170
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
4171
        return -1;
4172 4173
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
4174 4175
    return 0;
}
4176 4177 4178

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

4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213
/* 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);
        }
    }
}

4214 4215 4216 4217 4218 4219 4220 4221 4222
/* 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:
4223
 * @dom: domain definition that @iface belongs to
4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234
 * @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
4235 4236
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
4237
{
4238
    virNetworkDriverStatePtr driver = networkGetDriver();
4239
    virDomainNetType actualType = iface->type;
4240 4241
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
4242
    virNetDevBandwidthPtr bandwidth = NULL;
4243 4244 4245
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
4246
    virNetworkForwardIfDefPtr dev = NULL;
4247
    size_t i;
4248 4249 4250
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
4251
        goto validate;
4252 4253 4254 4255

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

4256
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4257
    if (!network) {
4258 4259 4260
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4261
        goto error;
4262 4263
    }
    netdef = network->def;
4264

4265 4266 4267 4268 4269 4270 4271
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

4272 4273 4274
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

4275 4276 4277
    /* 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
4278
     */
4279 4280 4281 4282 4283 4284
    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.
     */
4285 4286 4287 4288 4289 4290

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

4291 4292
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
4293
        goto error;
4294

4295 4296 4297 4298 4299 4300 4301 4302
    /* 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;

4303 4304
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
4305

4306 4307 4308 4309 4310 4311 4312 4313 4314 4315
    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;

4316 4317
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
4318 4319
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_OPEN)) {
4320
        /* for these forward types, the actual net type really *is*
4321
         * NETWORK; we just keep the info from the portgroup in
4322
         * iface->data.network.actual
J
Ján Tomko 已提交
4323
         */
4324
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
4325

4326
        /* we also store the bridge device and macTableManager settings
4327 4328 4329 4330 4331 4332 4333
         * 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;
4334 4335
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4336

4337 4338 4339
        if (networkPlugBandwidth(network, iface) < 0)
            goto error;

4340
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
4341
               netdef->bridge) {
4342 4343 4344 4345 4346

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

4347
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
4348 4349
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
4350
            goto error;
4351 4352
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4353

4354 4355 4356 4357 4358 4359 4360 4361
        /* 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) {
4362
            goto error;
4363 4364 4365 4366 4367 4368 4369 4370 4371 4372
        }
        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);
4373
                goto error;
4374 4375 4376
            }
        }

4377
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4378

4379
        virDomainHostdevSubsysPCIBackendType backend;
4380

4381
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
4382
        if (networkCreateInterfacePool(netdef) < 0)
4383 4384 4385
            goto error;

        /* pick first dev with 0 connections */
4386 4387 4388
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402
                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;
4403
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
4404
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
4405
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
4406

E
Eric Blake 已提交
4407
        switch (netdef->forward.driverName) {
4408
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
4409
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
4410 4411
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
4412
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
4413 4414
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
4415
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426
            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;

4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451
        /* 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;
            }
        }

4452 4453 4454 4455
    } 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)) {
4456 4457 4458 4459 4460 4461

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
4462
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4463
        switch (netdef->forward.type) {
4464
        case VIR_NETWORK_FORWARD_BRIDGE:
4465
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
4466 4467
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
4468
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
4469 4470
            break;
        case VIR_NETWORK_FORWARD_VEPA:
4471
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
4472 4473
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
4474
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
4475 4476 4477
            break;
        }

4478 4479 4480 4481 4482 4483 4484 4485
        /* 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) {
4486
            goto error;
4487
        }
4488
        virtport = iface->data.network.actual->virtPortProfile;
4489
        if (virtport) {
4490 4491 4492 4493 4494 4495 4496 4497
            /* 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);
4498
                goto error;
4499 4500
            }
        }
4501

4502 4503 4504
        /* 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).
         */
4505
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4506 4507 4508 4509
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4510
            goto error;
4511 4512 4513
        } else {
            /* pick an interface from the pool */

4514
            if (networkCreateInterfacePool(netdef) < 0)
4515 4516
                goto error;

4517 4518 4519 4520 4521
            /* 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.
4522
             */
4523 4524
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4525 4526 4527
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4528

4529
                /* pick first dev with 0 connections */
4530 4531 4532
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4533 4534 4535 4536 4537
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4538
                dev = &netdef->forward.ifs[0];
4539 4540 4541
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4542 4543 4544 4545
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4546 4547 4548 4549
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4550
                goto error;
4551
            }
4552 4553
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4554
                goto error;
4555 4556 4557
        }
    }

M
Michal Privoznik 已提交
4558 4559 4560
    if (networkMacMgrAdd(driver, network, dom->name, &iface->mac) < 0)
        goto error;

4561
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4562
        goto error;
4563

4564
 validate:
4565 4566 4567 4568 4569
    /* 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.
     */

4570
    if (virDomainNetGetActualVlan(iface)) {
4571 4572 4573 4574
        /* 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
4575 4576
         */
        if (!(actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV ||
4577 4578 4579
              (actualType == VIR_DOMAIN_NET_TYPE_DIRECT &&
               virDomainNetGetActualDirectMode(iface)
               == VIR_NETDEV_MACVLAN_MODE_PASSTHRU) ||
4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598
              (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;
        }
    }
4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609
    if (virDomainNetGetActualBandwidth(iface)) {
        /* bandwidth configuration via libvirt is not supported for
         * hostdev network devices
         */
        if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("bandwidth settings are not supported "
                             "for hostdev interfaces"));
            goto error;
        }
    }
4610 4611 4612

    if (netdef) {
        netdef->connections++;
4613
        if (dev)
4614 4615 4616 4617 4618 4619
            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 */
4620
            netdef->connections--;
4621 4622 4623 4624
            if (dev)
                dev->connections--;
            goto error;
        }
4625
        networkLogAllocation(netdef, actualType, dev, iface, true);
4626 4627
    }

4628
    ret = 0;
4629

4630
 cleanup:
4631
    virNetworkObjEndAPI(&network);
4632 4633
    return ret;

4634
 error:
4635
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4636 4637 4638
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4639
    goto cleanup;
4640 4641 4642
}

/* networkNotifyActualDevice:
4643
 * @dom: domain definition that @iface belongs to
4644 4645 4646 4647 4648
 * @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
4649 4650
 * order, or re-attach the interface's tap device to the network's
 * bridge.
4651 4652 4653 4654
 *
 * Returns 0 on success, -1 on failure.
 */
int
4655 4656
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4657
{
4658
    virNetworkDriverStatePtr driver = networkGetDriver();
4659
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4660 4661
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4662
    virNetworkForwardIfDefPtr dev = NULL;
4663 4664
    size_t i;
    int ret = -1;
4665
    char *master = NULL;
4666 4667

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4668
        return 0;
4669

4670
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4671
    if (!network) {
4672 4673 4674
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4675 4676 4677 4678
        goto error;
    }
    netdef = network->def;

4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689
    /* 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;

4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714
    /* see if we're connected to the correct bridge */
    if (netdef->bridge) {
        if (virNetDevGetMaster(iface->ifname, &master) < 0)
            goto error;

        if (STRNEQ_NULLABLE(netdef->bridge, master)) {
            /* disconnect from current (incorrect) bridge */
            if (master)
                ignore_value(virNetDevBridgeRemovePort(master, iface->ifname));

            /* attach/reattach to correct bridge.
             * NB: we can't notify the guest of any MTU change anyway,
             * so there is no point in trying to learn the actualMTU
             * (final arg to virNetDevTapAttachBridge())
             */
            if (virNetDevTapAttachBridge(iface->ifname, netdef->bridge,
                                         &iface->mac, dom->uuid,
                                         virDomainNetGetActualVirtPortProfile(iface),
                                         virDomainNetGetActualVlan(iface),
                                         iface->mtu, NULL) < 0) {
                goto error;
            }
        }
    }

4715
    if (!iface->data.network.actual ||
4716 4717
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4718 4719
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4720 4721
    }

4722
    if (networkCreateInterfacePool(netdef) < 0)
4723
        goto error;
4724

4725
    if (netdef->forward.nifs == 0) {
4726
        virReportError(VIR_ERR_INTERNAL_ERROR,
4727 4728
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4729
                       netdef->name);
4730
        goto error;
4731
    }
4732

4733 4734
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4735

4736 4737 4738 4739 4740 4741 4742 4743 4744
        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 */
4745 4746
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4747
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4748 4749
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4750 4751 4752 4753 4754
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4755
            virReportError(VIR_ERR_INTERNAL_ERROR,
4756 4757
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4758
                           netdef->name, actualDev);
4759
            goto error;
4760 4761
        }

4762
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4763 4764
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4765
         */
4766
        if ((dev->connections > 0) &&
4767 4768
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4769 4770
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4771
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4772
            virReportError(VIR_ERR_INTERNAL_ERROR,
4773 4774
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4775
                           netdef->name, actualDev);
4776
            goto error;
4777
        }
4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789
    }  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 */
4790 4791
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4792
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4793
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4794 4795
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4796 4797 4798 4799 4800 4801 4802 4803 4804
                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,
4805 4806 4807 4808
                           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
4809
            goto error;
4810 4811 4812 4813 4814 4815 4816
        }

        /* 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) &&
4817
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4818 4819 4820 4821 4822 4823 4824 4825 4826
            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;
        }
4827 4828
    }

4829
 success:
4830
    netdef->connections++;
4831 4832
    if (dev)
        dev->connections++;
4833 4834 4835 4836 4837 4838 4839 4840 4841
    /* 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;
    }
4842
    networkLogAllocation(netdef, actualType, dev, iface, true);
4843

4844
    ret = 0;
4845
 cleanup:
4846
    virNetworkObjEndAPI(&network);
4847
    VIR_FREE(master);
4848
    return ret;
4849

4850
 error:
4851
    goto cleanup;
4852 4853 4854
}


4855

4856
/* networkReleaseActualDevice:
4857
 * @dom: domain definition that @iface belongs to
4858 4859 4860 4861 4862 4863 4864 4865 4866 4867
 * @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
4868 4869
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4870
{
4871
    virNetworkDriverStatePtr driver = networkGetDriver();
4872
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4873
    virNetworkObjPtr network;
4874
    virNetworkDefPtr netdef;
4875
    virNetworkForwardIfDefPtr dev = NULL;
4876 4877
    size_t i;
    int ret = -1;
4878 4879

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4880
        return 0;
4881

4882
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4883
    if (!network) {
4884 4885 4886
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4887 4888 4889 4890
        goto error;
    }
    netdef = network->def;

4891 4892
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4893
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
4894 4895
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE ||
         netdef->forward.type == VIR_NETWORK_FORWARD_OPEN) &&
4896 4897 4898
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4899 4900 4901
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4902 4903
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4904 4905
    }

4906
    if (netdef->forward.nifs == 0) {
4907
        virReportError(VIR_ERR_INTERNAL_ERROR,
4908
                       _("network '%s' uses a direct/hostdev mode, but "
4909 4910
                         "has no forward dev and no interface pool"),
                       netdef->name);
4911
        goto error;
4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923
    }

    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;
        }
4924

4925 4926
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4927
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4928 4929
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4930 4931 4932
                break;
            }
        }
4933

4934
        if (!dev) {
4935
            virReportError(VIR_ERR_INTERNAL_ERROR,
4936 4937
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4938
                           netdef->name, actualDev);
4939
            goto error;
4940
        }
4941 4942 4943 4944 4945 4946 4947 4948 4949 4950
    } 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;
        }

4951 4952
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4953
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4954
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4955 4956
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4957 4958 4959 4960 4961 4962 4963 4964 4965
                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,
4966 4967 4968 4969
                           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
4970
            goto error;
4971
        }
J
Ján Tomko 已提交
4972
    }
4973

4974
 success:
M
Michal Privoznik 已提交
4975 4976
    networkMacMgrDel(driver, network, dom->name, &iface->mac);

4977
    if (iface->data.network.actual) {
4978
        netdef->connections--;
4979 4980
        if (dev)
            dev->connections--;
4981 4982 4983
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
4984
        networkLogAllocation(netdef, actualType, dev, iface, false);
4985
    }
4986
    ret = 0;
4987
 cleanup:
4988
    virNetworkObjEndAPI(&network);
4989 4990 4991 4992
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4993
    return ret;
4994

4995
 error:
4996
    goto cleanup;
4997
}
4998 4999 5000 5001 5002 5003

/*
 * networkGetNetworkAddress:
 * @netname: the name of a network
 * @netaddr: string representation of IP address for that network.
 *
5004
 * Attempt to return an IP address associated with the named
5005 5006 5007 5008
 * 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.
 *
5009
 * Note: This function returns the first IP address it finds. It might
5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020
 * 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)
{
5021
    virNetworkDriverStatePtr driver = networkGetDriver();
5022
    int ret = -1;
5023
    virNetworkObjPtr network;
5024
    virNetworkDefPtr netdef;
5025
    virNetworkIPDefPtr ipdef;
5026 5027
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
5028
    char *dev_name = NULL;
5029 5030

    *netaddr = NULL;
5031
    network = virNetworkObjFindByName(driver->networks, netname);
5032
    if (!network) {
5033 5034 5035
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
5036
        goto cleanup;
5037 5038 5039
    }
    netdef = network->def;

5040
    switch (netdef->forward.type) {
5041 5042 5043
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
5044
    case VIR_NETWORK_FORWARD_OPEN:
5045
        ipdef = virNetworkDefGetIPByIndex(netdef, AF_UNSPEC, 0);
5046
        if (!ipdef) {
5047
            virReportError(VIR_ERR_INTERNAL_ERROR,
5048
                           _("network '%s' doesn't have an IP address"),
5049
                           netdef->name);
5050
            goto cleanup;
5051 5052 5053 5054 5055
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
5056
        if ((dev_name = netdef->bridge))
5057 5058
            break;
        /*
5059 5060
         * fall through if netdef->bridge wasn't set, since that is
         * macvtap bridge mode network.
5061
         */
5062 5063
        ATTRIBUTE_FALLTHROUGH;

5064 5065 5066
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
5067 5068
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
5069

5070
        if (!dev_name) {
5071 5072 5073
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
5074
            goto cleanup;
5075 5076 5077 5078
        }
        break;
    }

5079
    if (dev_name) {
5080
        if (virNetDevIPAddrGet(dev_name, &addr) < 0)
5081
            goto cleanup;
5082
        addrptr = &addr;
5083 5084
    }

5085 5086
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
5087
        goto cleanup;
5088 5089
    }

5090
    ret = 0;
5091
 cleanup:
5092
    virNetworkObjEndAPI(&network);
5093 5094
    return ret;
}
5095

5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129
/* 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) ||
5130 5131
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_OPEN)) {
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 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168
        /* 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;
}


5169 5170 5171
/**
 * networkCheckBandwidth:
 * @net: network QoS
5172
 * @ifaceBand: interface QoS (may be NULL if no QoS)
5173
 * @oldBandwidth: new interface QoS (may be NULL if no QoS)
5174
 * @ifaceMac: interface MAC (used in error messages for identification)
5175 5176
 * @new_rate: new rate for non guaranteed class
 *
5177 5178 5179 5180 5181 5182 5183 5184
 * 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.
 *
5185 5186 5187 5188 5189 5190
 * 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,
5191
                      virNetDevBandwidthPtr ifaceBand,
5192
                      virNetDevBandwidthPtr oldBandwidth,
5193
                      virMacAddr ifaceMac,
5194 5195 5196 5197 5198 5199 5200 5201
                      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];

5202
    virMacAddrFormat(&ifaceMac, ifmac);
5203 5204 5205 5206 5207 5208 5209 5210 5211 5212

    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;
    }

5213 5214
    if (((!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor) &&
         (!oldBandwidth || !oldBandwidth->in || !oldBandwidth->in->floor)) ||
5215 5216
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
5217
        return 1;
5218
    }
5219 5220

    tmp_new_rate = netBand->in->average;
5221 5222 5223 5224
    if (oldBandwidth && oldBandwidth->in)
        tmp_floor_sum -= oldBandwidth->in->floor;
    if (ifaceBand && ifaceBand->in)
        tmp_floor_sum += ifaceBand->in->floor;
5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249

    /* 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;
    }

5250 5251
    if (new_rate)
        *new_rate = tmp_new_rate;
5252 5253
    ret = 0;

5254
 cleanup:
5255 5256 5257 5258 5259 5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270
    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)
{
5271
    ssize_t ret = 0;
5272

5273
    ret = virBitmapNextClearBit(net->class_id, -1);
5274

5275
    if (ret < 0 || virBitmapSetBit(net->class_id, ret) < 0)
5276 5277 5278 5279 5280
        return -1;

    return ret;
}

5281

5282
static int
5283 5284 5285 5286
networkPlugBandwidthImpl(virNetworkObjPtr net,
                         virDomainNetDefPtr iface,
                         virNetDevBandwidthPtr ifaceBand,
                         unsigned long long new_rate)
5287
{
5288
    virNetworkDriverStatePtr driver = networkGetDriver();
5289
    ssize_t class_id = 0;
5290 5291
    int plug_ret;
    int ret = -1;
5292 5293 5294 5295 5296 5297 5298 5299

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

5300 5301
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
5302 5303 5304 5305 5306 5307 5308 5309
    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 */
5310
    net->floor_sum += ifaceBand->in->floor;
5311
    /* update status file */
5312
    if (virNetworkObjSaveStatus(driver->stateDir, net) < 0) {
5313
        ignore_value(virBitmapClearBit(net->class_id, class_id));
5314
        net->floor_sum -= ifaceBand->in->floor;
5315 5316 5317 5318
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
5319 5320
    /* update rate for non guaranteed NICs */
    new_rate -= net->floor_sum;
5321
    if (virNetDevBandwidthUpdateRate(net->def->bridge, 2,
5322 5323 5324 5325 5326
                                     net->def->bandwidth, new_rate) < 0)
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                 net->def->bridge);

    ret = 0;
5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366
 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;
5367

5368
 cleanup:
5369 5370 5371 5372 5373 5374 5375
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
5376
    virNetworkDriverStatePtr driver = networkGetDriver();
5377 5378
    int ret = 0;
    unsigned long long new_rate;
5379
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
5380 5381 5382

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
5383 5384 5385 5386 5387
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398
        /* 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 */
5399
        net->floor_sum -= ifaceBand->in->floor;
5400 5401 5402 5403
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
5404
        if (virNetworkObjSaveStatus(driver->stateDir, net) < 0) {
5405
            net->floor_sum += ifaceBand->in->floor;
5406 5407 5408 5409
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
5410 5411
        /* update rate for non guaranteed NICs */
        new_rate -= net->floor_sum;
5412
        if (virNetDevBandwidthUpdateRate(net->def->bridge, 2,
5413 5414 5415 5416 5417 5418 5419
                                         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;
    }

5420
 cleanup:
5421 5422
    return ret;
}
5423 5424 5425

static void
networkNetworkObjTaint(virNetworkObjPtr net,
5426
                       virNetworkTaintFlags taint)
5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437
{
    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));
    }
}
5438 5439 5440 5441 5442 5443


static bool
networkBandwidthGenericChecks(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
5444
    virNetDevBandwidthPtr ifaceBand;
5445 5446 5447 5448 5449 5450 5451 5452
    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;
    }

5453
    ifaceBand = virDomainNetGetActualBandwidth(iface);
5454 5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492 5493
    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;
}
5494 5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531


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: */

5532
    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
5533
        newBandwidth->in && newBandwidth->in->floor) {
5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547
        /* 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 ||
5548
            virNetworkObjSaveStatus(driver->stateDir, network) < 0) {
5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575
            /* 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;
}