bridge_driver.c 182.2 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2016 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

#include <sys/types.h>
#include <sys/poll.h>
#include <limits.h>
#include <string.h>
#include <stdio.h>
#include <stdarg.h>
#include <stdlib.h>
#include <unistd.h>
#include <errno.h>
#include <sys/utsname.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <signal.h>
#include <pwd.h>
#include <sys/wait.h>
#include <sys/ioctl.h>
42
#include <net/if.h>
43
#include <dirent.h>
44 45 46
#if HAVE_SYS_SYSCTL_H
# include <sys/sysctl.h>
#endif
47

48
#include "virerror.h"
49
#include "datatypes.h"
50
#include "bridge_driver.h"
51
#include "bridge_driver_platform.h"
52
#include "network_conf.h"
53
#include "device_conf.h"
54
#include "driver.h"
55
#include "virbuffer.h"
56
#include "virpidfile.h"
57
#include "vircommand.h"
58
#include "viralloc.h"
59
#include "viruuid.h"
60
#include "viriptables.h"
61
#include "virlog.h"
62
#include "virdnsmasq.h"
63
#include "configmake.h"
64
#include "virnetdev.h"
65
#include "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
VIR_LOG_INIT("network.bridge_driver");

90
static virNetworkDriverStatePtr network_driver;
91

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

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

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

static int
networkDnsmasqCapsRefresh(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr caps;

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

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

135
static int networkStateCleanup(void);
136

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

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

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

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

149
static int networkStartNetworkExternal(virNetworkObjPtr network);
150

151
static int networkShutdownNetworkExternal(virNetworkObjPtr network);
152

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

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

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

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

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

    return network;
}

182 183 184
static int
networkRunHook(virNetworkObjPtr network,
               virDomainDefPtr dom,
185
               virDomainNetDefPtr iface,
186 187 188 189 190 191 192 193 194
               int op,
               int sub_op)
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *xml = NULL, *net_xml = NULL, *dom_xml = NULL;
    int hookret;
    int ret = -1;

    if (virHookPresent(VIR_HOOK_DRIVER_NETWORK)) {
195 196 197 198 199 200
        if (!network) {
            VIR_DEBUG("Not running hook as @network is NULL");
            ret = 0;
            goto cleanup;
        }

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

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

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

216
        xml = virBufferContentAndReset(&buf);
217 218 219 220 221 222 223 224
        hookret = virHookCall(VIR_HOOK_DRIVER_NETWORK, network->def->name,
                              op, sub_op, NULL, xml, NULL);

        /*
         * If the script raised an error, pass it to the callee.
         */
        if (hookret < 0)
            goto cleanup;
225 226

        networkNetworkObjTaint(network, VIR_NETWORK_TAINT_HOOK);
227 228 229
    }

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

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

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

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

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

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

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

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

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

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

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

M
Michal Privoznik 已提交
292 293 294 295 296 297 298 299 300 301 302
static char *
networkMacMgrFileName(virNetworkDriverStatePtr driver,
                      const char *bridge)
{
    char *filename;

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

303 304
/* do needed cleanup steps and remove the network from the list */
static int
305 306
networkRemoveInactive(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr net)
307 308
{
    char *leasefile = NULL;
309
    char *customleasefile = NULL;
310
    char *radvdconfigfile = NULL;
311
    char *configfile = NULL;
312
    char *radvdpidbase = NULL;
313
    char *statusfile = NULL;
M
Michal Privoznik 已提交
314
    char *macMapFile = NULL;
315 316 317 318 319 320
    dnsmasqContext *dctx = NULL;
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(net);

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
321
    if (!(dctx = dnsmasqContextNew(def->name,
322
                                   driver->dnsmasqStateDir))) {
323
        goto cleanup;
324
    }
325

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

329
    if (!(customleasefile = networkDnsmasqLeaseFileNameCustom(driver, def->bridge)))
330 331
        goto cleanup;

332
    if (!(radvdconfigfile = networkRadvdConfigFileName(driver, def->name)))
333
        goto cleanup;
334 335

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
336
        goto cleanup;
337

338
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
339
        goto cleanup;
340

341
    if (!(statusfile = virNetworkConfigFile(driver->stateDir, def->name)))
342
        goto cleanup;
343

M
Michal Privoznik 已提交
344 345 346
    if (!(macMapFile = networkMacMgrFileName(driver, def->bridge)))
        goto cleanup;

347 348 349
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
350
    unlink(customleasefile);
351
    unlink(configfile);
352

M
Michal Privoznik 已提交
353 354 355
    /* MAC map manager */
    unlink(macMapFile);

356 357
    /* radvd */
    unlink(radvdconfigfile);
358
    virPidFileDelete(driver->pidDir, radvdpidbase);
359

360 361 362
    /* remove status file */
    unlink(statusfile);

363
    /* remove the network definition */
364
    virNetworkRemoveInactive(driver->networks, net);
365 366 367

    ret = 0;

368
 cleanup:
369
    VIR_FREE(leasefile);
370
    VIR_FREE(configfile);
371
    VIR_FREE(customleasefile);
372 373
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
374
    VIR_FREE(statusfile);
M
Michal Privoznik 已提交
375
    VIR_FREE(macMapFile);
376 377 378 379
    dnsmasqContextFree(dctx);
    return ret;
}

M
Michal Privoznik 已提交
380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439
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;
}

440 441 442
static char *
networkBridgeDummyNicName(const char *brname)
{
443
    static const char dummyNicSuffix[] = "-nic";
444 445
    char *nicname;

446 447 448 449 450 451 452
    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.
         */
453 454 455 456 457
        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));
458
    } else {
459
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
460
    }
461 462 463
    return nicname;
}

464 465
static int
networkUpdateState(virNetworkObjPtr obj,
466
                   void *opaque)
467
{
468
    virNetworkDriverStatePtr driver = opaque;
469
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
470
    int ret = -1;
471

472
    virObjectLock(obj);
473
    if (!virNetworkObjIsActive(obj)) {
474 475
        ret = 0;
        goto cleanup;
476
    }
477

478 479 480 481
    switch (obj->def->forward.type) {
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
482
    case VIR_NETWORK_FORWARD_OPEN:
483 484 485 486
        /* If bridge doesn't exist, then mark it inactive */
        if (!(obj->def->bridge && virNetDevExists(obj->def->bridge) == 1))
            obj->active = 0;
        break;
487

488
    case VIR_NETWORK_FORWARD_BRIDGE:
489 490
        if (obj->def->bridge) {
            if (virNetDevExists(obj->def->bridge) != 1)
491 492
                obj->active = 0;
            break;
493
        }
494 495 496 497 498 499 500 501 502
        /* 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;
503

504 505 506
    case VIR_NETWORK_FORWARD_HOSTDEV:
        /* so far no extra checks */
        break;
507
    }
508

509 510 511
    /* Try and read dnsmasq/radvd pids of active networks */
    if (obj->active && obj->def->ips && (obj->def->nips > 0)) {
        char *radvdpidbase;
M
Michal Privoznik 已提交
512
        char *macMapFile;
513

514 515 516
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           obj->def->name,
                                           &obj->dnsmasqPid,
517
                                           dnsmasqCapsGetBinaryPath(dnsmasq_caps)));
M
Michal Privoznik 已提交
518

519 520 521
        radvdpidbase = networkRadvdPidfileBasename(obj->def->name);
        if (!radvdpidbase)
            goto cleanup;
522

523 524 525 526
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           radvdpidbase,
                                           &obj->radvdPid, RADVD));
        VIR_FREE(radvdpidbase);
M
Michal Privoznik 已提交
527 528 529 530 531 532 533 534 535 536 537

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

        if (virFileExists(macMapFile) &&
            !(obj->macmap = virMacMapNew(macMapFile))) {
            VIR_FREE(macMapFile);
            goto cleanup;
        }

        VIR_FREE(macMapFile);
538
    }
539

540 541
    ret = 0;
 cleanup:
542
    virObjectUnlock(obj);
543
    virObjectUnref(dnsmasq_caps);
544 545
    return ret;
}
546

547

548 549
static int
networkAutostartConfig(virNetworkObjPtr net,
550
                       void *opaque)
551
{
552
    virNetworkDriverStatePtr driver = opaque;
553
    int ret = -1;
554

555
    virObjectLock(net);
556 557
    if (net->autostart &&
        !virNetworkObjIsActive(net) &&
558
        networkStartNetwork(driver, net) < 0)
559 560 561 562
        goto cleanup;

    ret = 0;
 cleanup:
563
    virObjectUnlock(net);
564
    return ret;
565 566
}

567 568 569
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
570
                             DBusMessage *message, void *user_data)
571
{
572 573
    virNetworkDriverStatePtr driver = user_data;

574 575 576 577 578 579
    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.");
580
        networkReloadFirewallRules(driver);
581 582 583 584 585 586
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

587
static int
588
networkMigrateStateFiles(virNetworkDriverStatePtr driver)
589 590 591 592 593 594 595 596 597 598 599 600 601
{
    /* 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;
602
    int direrr;
603 604 605
    struct dirent *entry;
    char *oldPath = NULL, *newPath = NULL;
    char *contents = NULL;
J
Ján Tomko 已提交
606
    int rc;
607

J
Ján Tomko 已提交
608 609
    if ((rc = virDirOpenIfExists(&dir, oldStateDir)) <= 0)
        return rc;
610 611 612 613 614 615 616

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

617
    while ((direrr = virDirRead(dir, &entry, oldStateDir)) > 0) {
618 619 620
        if (entry->d_type != DT_UNKNOWN &&
            entry->d_type != DT_REG)
            continue;
621 622 623

        if (virAsprintf(&oldPath, "%s/%s",
                        oldStateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
624
            goto cleanup;
625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641

        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;
            }
        }

642
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
J
Ján Tomko 已提交
643
            goto cleanup;
644 645 646

        if (virAsprintf(&newPath, "%s/%s",
                        driver->stateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
647
            goto cleanup;
648 649 650 651 652 653 654 655 656 657 658 659
        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);
    }
660
    if (direrr < 0)
J
Ján Tomko 已提交
661
        goto cleanup;
662 663 664

    ret = 0;
 cleanup:
J
Ján Tomko 已提交
665
    VIR_DIR_CLOSE(dir);
666 667 668 669 670 671
    VIR_FREE(oldPath);
    VIR_FREE(newPath);
    VIR_FREE(contents);
    return ret;
}

672
/**
673
 * networkStateInitialize:
674 675 676 677
 *
 * Initialization function for the QEmu daemon
 */
static int
678 679 680
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
681
{
682 683 684
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
685 686 687
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
688

689
    if (VIR_ALLOC(network_driver) < 0)
690
        goto error;
691

692 693
    if (virMutexInit(&network_driver->lock) < 0) {
        VIR_FREE(network_driver);
694 695
        goto error;
    }
696

697 698 699 700
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
701
    if (privileged) {
702
        if (VIR_STRDUP(network_driver->networkConfigDir,
703
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
704
            VIR_STRDUP(network_driver->networkAutostartDir,
705
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
706
            VIR_STRDUP(network_driver->stateDir,
707
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
708
            VIR_STRDUP(network_driver->pidDir,
709
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
710
            VIR_STRDUP(network_driver->dnsmasqStateDir,
711
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
712
            VIR_STRDUP(network_driver->radvdStateDir,
713 714
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
715 716 717 718 719

        /* migration from old to new location is only applicable for
         * privileged mode - unprivileged mode directories haven't
         * changed location.
         */
720
        if (networkMigrateStateFiles(network_driver) < 0)
721
            goto error;
722
    } else {
723 724 725
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
726
            goto error;
727

728
        if ((virAsprintf(&network_driver->networkConfigDir,
729
                         "%s/qemu/networks", configdir) < 0) ||
730
            (virAsprintf(&network_driver->networkAutostartDir,
731
                         "%s/qemu/networks/autostart", configdir) < 0) ||
732
            (virAsprintf(&network_driver->stateDir,
733
                         "%s/network/lib", rundir) < 0) ||
734
            (virAsprintf(&network_driver->pidDir,
735
                         "%s/network/run", rundir) < 0) ||
736
            (virAsprintf(&network_driver->dnsmasqStateDir,
737
                         "%s/dnsmasq/lib", rundir) < 0) ||
738
            (virAsprintf(&network_driver->radvdStateDir,
739
                         "%s/radvd/lib", rundir) < 0)) {
740
            goto error;
741
        }
742 743
    }

744
    if (virFileMakePath(network_driver->stateDir) < 0) {
745 746
        virReportSystemError(errno,
                             _("cannot create directory %s"),
747
                             network_driver->stateDir);
748 749 750
        goto error;
    }

751
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
752
    network_driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
753

754
    if (!(network_driver->networks = virNetworkObjListNew()))
755 756
        goto error;

757 758
    if (virNetworkLoadAllState(network_driver->networks,
                               network_driver->stateDir) < 0)
759 760
        goto error;

761 762 763
    if (virNetworkLoadAllConfigs(network_driver->networks,
                                 network_driver->networkConfigDir,
                                 network_driver->networkAutostartDir) < 0)
764 765
        goto error;

766 767 768 769
    /* 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). */
770
    virNetworkObjListForEach(network_driver->networks,
771
                             networkUpdateState,
772 773
                             network_driver);
    virNetworkObjListPrune(network_driver->networks,
774 775
                           VIR_CONNECT_LIST_NETWORKS_INACTIVE |
                           VIR_CONNECT_LIST_NETWORKS_TRANSIENT);
776 777
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
778

779
    network_driver->networkEventState = virObjectEventStateNew();
780

781 782 783
#ifdef HAVE_FIREWALLD
    if (!(sysbus = virDBusGetSystemBus())) {
        VIR_WARN("DBus not available, disabling firewalld support "
784
                 "in bridge_network_driver: %s", virGetLastErrorMessage());
785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801
    } 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,
802
                                   network_driver, NULL);
803 804 805
    }
#endif

806
    ret = 0;
807
 cleanup:
808 809 810
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
811

812
 error:
813
    networkStateCleanup();
814
    goto cleanup;
815 816
}

817 818 819 820 821 822 823 824
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
825
    if (!network_driver)
826 827
        return;

828
    virNetworkObjListForEach(network_driver->networks,
829
                             networkAutostartConfig,
830
                             network_driver);
831 832
}

833
/**
834
 * networkStateReload:
835 836 837 838 839
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
840 841
networkStateReload(void)
{
842
    if (!network_driver)
843 844
        return 0;

845 846 847 848 849 850 851 852
    virNetworkLoadAllState(network_driver->networks,
                           network_driver->stateDir);
    virNetworkLoadAllConfigs(network_driver->networks,
                             network_driver->networkConfigDir,
                             network_driver->networkAutostartDir);
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
    virNetworkObjListForEach(network_driver->networks,
853
                             networkAutostartConfig,
854
                             network_driver);
855 856 857 858 859
    return 0;
}


/**
860
 * networkStateCleanup:
861 862 863 864
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
865 866
networkStateCleanup(void)
{
867
    if (!network_driver)
868 869
        return -1;

870
    virObjectUnref(network_driver->networkEventState);
871

872
    /* free inactive networks */
873
    virObjectUnref(network_driver->networks);
874

875 876 877 878 879 880
    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);
881

882
    virObjectUnref(network_driver->dnsmasqCaps);
883

884
    virMutexDestroy(&network_driver->lock);
885

886
    VIR_FREE(network_driver);
887 888 889 890 891

    return 0;
}


892 893 894 895 896 897 898
/* 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)
{
899 900
    size_t i;
    int ret = -1;
901 902 903 904 905 906 907
    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.
     */
908
    for (i = 0; i < 25; i++) {
909
        int signum = 0;
910
        if (i == 0) {
911
            signum = SIGTERM;
912
        } else if (i == 15) {
913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947
            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);
948
 cleanup:
949 950 951
    return ret;
}

J
Ján Tomko 已提交
952 953 954
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
G
Gene Czarcinski 已提交
955

956
static int
G
Gene Czarcinski 已提交
957
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
958
                                 virNetworkIPDefPtr ipdef)
959
{
960
    size_t i;
G
Gene Czarcinski 已提交
961
    bool ipv6 = false;
962

G
Gene Czarcinski 已提交
963 964
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
965 966
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
967
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
968 969
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
970
                return -1;
971
    }
972

G
Gene Czarcinski 已提交
973 974 975 976 977 978 979
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
980
    size_t i, j;
G
Gene Czarcinski 已提交
981

982 983
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
984
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
985
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
986
                for (j = 0; j < host->nnames; j++)
987 988
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
989 990
            }
        }
991 992
    }

993
    return 0;
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 1033
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;
}


1034 1035
int
networkDnsmasqConfContents(virNetworkObjPtr network,
1036 1037 1038 1039
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
1040
{
1041
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1042
    int r, ret = -1;
1043
    int nbleases = 0;
1044
    size_t i;
1045
    virNetworkDNSDefPtr dns = &network->def->dns;
1046
    bool wantDNS = dns->enable != VIR_TRISTATE_BOOL_NO;
1047
    virNetworkIPDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
G
Gene Czarcinski 已提交
1048
    bool ipv6SLAAC;
1049
    char *saddr = NULL, *eaddr = NULL;
1050

1051 1052
    *configstr = NULL;

1053
    /*
1054 1055 1056
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
1057
     *
1058 1059
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
1060
     */
1061 1062 1063 1064 1065 1066

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

1067
    /* create dnsmasq config file appropriate for this network */
1068 1069

    /* Don't forget to update networkxml2conftest :-) */
1070
    virBufferAsprintf(&configbuf,
1071 1072 1073 1074 1075 1076 1077
                      "##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"
1078
                      "strict-order\n",
1079 1080
                      network->def->name);

1081 1082 1083 1084 1085 1086 1087
    /* 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) {
1088 1089
        virBufferAddLit(&configbuf, "no-resolv\n");
        for (i = 0; i < network->def->dns.nfwds; i++) {
1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100
            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);
1101
                VIR_FREE(addr);
1102 1103 1104 1105
            } else {
                /* "don't forward requests for this domain" */
                virBufferAddLit(&configbuf, "#\n");
            }
1106 1107 1108
        }
    }

1109
    if (network->def->domain) {
1110 1111 1112 1113 1114
        if (network->def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
                              network->def->domain);
        }
1115
        virBufferAsprintf(&configbuf,
1116 1117 1118 1119
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
1120

1121 1122 1123 1124
    if (wantDNS &&
        networkDnsmasqConfLocalPTRs(&configbuf, network->def) < 0)
        goto cleanup;

1125
    if (wantDNS && network->def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
1126 1127 1128 1129
        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)
1130
         */
1131
        virBufferAddLit(&configbuf, "local=//\n");
1132
    }
1133

1134
    if (pidfile)
1135
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
1136

1137
    /* dnsmasq will *always* listen on localhost unless told otherwise */
P
Pavel Timofeev 已提交
1138
#ifdef __linux__
1139
    virBufferAddLit(&configbuf, "except-interface=lo\n");
P
Pavel Timofeev 已提交
1140 1141 1142 1143
#else
    /* BSD family OSes and Solaris call loopback interface as lo0 */
    virBufferAddLit(&configbuf, "except-interface=lo0\n");
#endif
1144

1145 1146 1147 1148 1149 1150 1151 1152
    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.
         */
1153
        virBufferAsprintf(&configbuf,
1154 1155 1156
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
1157
    } else {
1158
        virBufferAddLit(&configbuf, "bind-interfaces\n");
1159 1160 1161 1162 1163 1164 1165 1166
        /*
         * --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
         */
1167
        for (i = 0;
1168
             (tmpipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i));
1169
             i++) {
1170 1171 1172 1173
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
1174

1175
            /* also part of CVE 2012-3411 - if the host's version of
1176
             * dnsmasq doesn't have bind-dynamic, only allow listening on
1177 1178
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
1179 1180
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
1181 1182 1183 1184
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
1185
                                 "The version of dnsmasq on this host (%d.%d) "
1186 1187 1188 1189
                                 "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 "
1190 1191 1192 1193 1194 1195
                                 "(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);
1196
                VIR_FREE(ipaddr);
1197 1198
                goto cleanup;
            }
1199
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1200 1201 1202
            VIR_FREE(ipaddr);
        }
    }
1203

1204 1205
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1206
     * guaranteed to not work, and set no-resolv so that no dns
1207 1208 1209
     * 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).
1210 1211 1212
     * 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.
1213
     */
1214
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1215
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1216
                        "no-resolv\n");
1217 1218 1219 1220
        if (dnsmasqCapsGet(caps, DNSMASQ_CAPS_RA_PARAM)) {
            /* interface=* (any), interval=0 (default), lifetime=0 (seconds) */
            virBufferAddLit(&configbuf, "ra-param=*,0,0\n");
        }
1221
    }
1222

1223 1224 1225 1226 1227
    if (wantDNS) {
        for (i = 0; i < dns->ntxts; i++) {
            virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
                              dns->txts[i].name,
                              dns->txts[i].value);
1228
        }
1229

1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252
        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);
1253

1254 1255 1256 1257 1258 1259 1260
            /* 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.
1261
             */
1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281
            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");
1282
        }
1283 1284
    }

G
Gene Czarcinski 已提交
1285
    /* Find the first dhcp for both IPv4 and IPv6 */
1286
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
1287
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i));
1288
         i++) {
G
Gene Czarcinski 已提交
1289 1290 1291 1292
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1293 1294
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305
                    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,
1306 1307 1308 1309 1310 1311 1312 1313 1314
                                   _("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
1315 1316 1317 1318
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1319 1320
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333
                    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 "
1334 1335 1336 1337
                 "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
1338 1339 1340 1341 1342
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1343 1344
        int prefix;

1345
        prefix = virNetworkIPDefPrefix(ipdef);
1346 1347 1348 1349 1350 1351
        if (prefix < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
            goto cleanup;
        }
1352
        for (r = 0; r < ipdef->nranges; r++) {
1353 1354
            int thisRange;

1355 1356
            if (!(saddr = virSocketAddrFormat(&ipdef->ranges[r].start)) ||
                !(eaddr = virSocketAddrFormat(&ipdef->ranges[r].end)))
1357
                goto cleanup;
1358

1359
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s",
1360
                              saddr, eaddr);
1361
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1362
                virBufferAsprintf(&configbuf, ",%d", prefix);
1363 1364
            virBufferAddLit(&configbuf, "\n");

1365
            VIR_FREE(saddr);
1366
            VIR_FREE(eaddr);
1367
            thisRange = virSocketAddrGetRange(&ipdef->ranges[r].start,
1368 1369
                                              &ipdef->ranges[r].end,
                                              &ipdef->address,
1370
                                              virNetworkIPDefPrefix(ipdef));
1371 1372 1373
            if (thisRange < 0)
                goto cleanup;
            nbleases += thisRange;
1374
        }
1375

1376
        /*
1377 1378 1379 1380
         * 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)
1381 1382
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1383
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1384 1385
            if (!bridgeaddr)
                goto cleanup;
1386 1387 1388
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static",
                              bridgeaddr);
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1389
                virBufferAsprintf(&configbuf, ",%d", prefix);
1390
            virBufferAddLit(&configbuf, "\n");
1391 1392
            VIR_FREE(bridgeaddr);
        }
1393

G
Gene Czarcinski 已提交
1394 1395
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1396

G
Gene Czarcinski 已提交
1397 1398
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
1399
            if (ipdef->nranges || ipdef->nhosts) {
1400
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1401 1402
                virBufferAddLit(&configbuf, "dhcp-authoritative\n");
            }
1403

G
Gene Czarcinski 已提交
1404
            if (ipdef->tftproot) {
1405 1406
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1407
            }
1408

G
Gene Czarcinski 已提交
1409 1410 1411
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1412

1413
                    if (!bootserver)
G
Gene Czarcinski 已提交
1414
                        goto cleanup;
1415
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1416
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1417 1418
                    VIR_FREE(bootserver);
                } else {
1419
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1420 1421 1422 1423 1424
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1425

1426
    if (nbleases > 0)
1427
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1428

G
Gene Czarcinski 已提交
1429 1430
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1431
        goto cleanup;
G
Gene Czarcinski 已提交
1432 1433 1434 1435 1436 1437

    /* 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)
1438 1439
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1440

1441 1442
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1443
     */
1444 1445 1446 1447
    if (wantDNS) {
        virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
                          dctx->addnhostsfile->path);
    }
G
Gene Czarcinski 已提交
1448 1449 1450

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1451
        if (ipv6def) {
1452
            virBufferAddLit(&configbuf, "enable-ra\n");
1453
        } else {
1454
            for (i = 0;
1455
                 (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1456
                 i++) {
G
Gene Czarcinski 已提交
1457 1458 1459 1460
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1461 1462
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1463 1464
                    VIR_FREE(bridgeaddr);
                }
1465
            }
1466
        }
1467 1468
    }

1469 1470 1471
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1472
    ret = 0;
G
Gene Czarcinski 已提交
1473

1474
 cleanup:
1475 1476
    VIR_FREE(saddr);
    VIR_FREE(eaddr);
1477
    virBufferFreeAndReset(&configbuf);
1478
    return ret;
1479 1480
}

1481
/* build the dnsmasq command line */
1482 1483 1484
static int ATTRIBUTE_NONNULL(3)
networkBuildDhcpDaemonCommandLine(virNetworkDriverStatePtr driver,
                                  virNetworkObjPtr network,
1485
                                  virCommandPtr *cmdout,
1486 1487
                                  char *pidfile,
                                  dnsmasqContext *dctx)
1488
{
1489
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1490
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1491
    int ret = -1;
1492 1493
    char *configfile = NULL;
    char *configstr = NULL;
1494
    char *leaseshelper_path = NULL;
1495 1496

    network->dnsmasqPid = -1;
1497

1498
    if (networkDnsmasqConfContents(network, pidfile, &configstr,
1499
                                   dctx, dnsmasq_caps) < 0)
1500 1501 1502 1503 1504
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1505
    if (!(configfile = networkDnsmasqConfigFileName(driver, network->def->name)))
1506 1507 1508 1509 1510
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1511 1512
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1513 1514 1515
        goto cleanup;
    }

1516 1517
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1518
                                                  abs_topbuilddir "/src",
1519 1520 1521
                                                  LIBEXECDIR)))
        goto cleanup;

1522
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(dnsmasq_caps));
1523
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1524 1525
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1526
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1527
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", network->def->bridge);
1528

1529
    *cmdout = cmd;
1530
    ret = 0;
1531
 cleanup:
1532
    virObjectUnref(dnsmasq_caps);
1533 1534
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1535
    VIR_FREE(leaseshelper_path);
1536 1537 1538 1539
    return ret;
}

static int
1540 1541
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
1542
{
1543 1544 1545
    virNetworkIPDefPtr ipdef;
    size_t i;
    bool needDnsmasq = false;
1546 1547 1548
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1549
    dnsmasqContext *dctx = NULL;
1550

1551
    /* see if there are any IP addresses that need a dhcp server */
1552 1553 1554
    i = 0;
    while ((ipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i))) {
        i++;
1555 1556 1557 1558
        if (ipdef->nranges || ipdef->nhosts)
            needDnsmasq = true;
    }

1559 1560 1561 1562 1563 1564
    if (i == 0) {
        /* no IP addresses at all, so we don't need to run */
        ret = 0;
        goto cleanup;
    }

1565
    if (!needDnsmasq && network->def->dns.enable == VIR_TRISTATE_BOOL_NO) {
1566
        /* no DHCP services needed, and user disabled DNS service */
1567 1568 1569 1570
        ret = 0;
        goto cleanup;
    }

1571
    if (virFileMakePath(driver->pidDir) < 0) {
1572
        virReportSystemError(errno,
1573
                             _("cannot create directory %s"),
1574
                             driver->pidDir);
1575
        goto cleanup;
1576 1577
    }

1578
    if (!(pidfile = virPidFileBuildPath(driver->pidDir,
1579
                                        network->def->name)))
1580
        goto cleanup;
1581

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

1589
    dctx = dnsmasqContextNew(network->def->name, driver->dnsmasqStateDir);
1590 1591 1592
    if (dctx == NULL)
        goto cleanup;

1593
    if (networkDnsmasqCapsRefresh(driver) < 0)
1594
        goto cleanup;
1595

1596
    ret = networkBuildDhcpDaemonCommandLine(driver, network, &cmd, pidfile, dctx);
1597 1598 1599 1600 1601
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1602
        goto cleanup;
1603

G
Guido Günther 已提交
1604
    ret = virCommandRun(cmd, NULL);
1605
    if (ret < 0)
1606 1607 1608
        goto cleanup;

    /*
1609 1610 1611 1612 1613
     * 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
1614 1615
     */

1616
    ret = virPidFileRead(driver->pidDir, network->def->name,
1617 1618
                         &network->dnsmasqPid);
    if (ret < 0)
1619
        goto cleanup;
1620

1621
    ret = 0;
1622
 cleanup:
1623
    VIR_FREE(pidfile);
1624
    virCommandFree(cmd);
1625
    dnsmasqContextFree(dctx);
1626 1627 1628
    return ret;
}

1629 1630
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1631 1632
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1633 1634 1635
 *
 *  Returns 0 on success, -1 on failure.
 */
1636
static int
1637 1638
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1639
{
1640 1641
    int ret = -1;
    size_t i;
1642
    virNetworkIPDefPtr ipdef, ipv4def, ipv6def;
1643
    dnsmasqContext *dctx = NULL;
1644

G
Gene Czarcinski 已提交
1645
    /* if no IP addresses specified, nothing to do */
1646
    if (!virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1647 1648
        return 0;

1649 1650
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1651
        return networkStartDhcpDaemon(driver, network);
1652

G
Gene Czarcinski 已提交
1653
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1654
    if (!(dctx = dnsmasqContextNew(network->def->name,
1655
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1656
        goto cleanup;
1657
    }
G
Gene Czarcinski 已提交
1658 1659 1660 1661 1662 1663

    /* 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;
1664
    for (i = 0;
1665
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET, i));
1666
         i++) {
G
Gene Czarcinski 已提交
1667 1668
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1669 1670
    }

G
Gene Czarcinski 已提交
1671
    ipv6def = NULL;
1672
    for (i = 0;
1673
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1674
         i++) {
G
Gene Czarcinski 已提交
1675 1676
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1677 1678
    }

G
Gene Czarcinski 已提交
1679
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1680
        goto cleanup;
G
Gene Czarcinski 已提交
1681 1682

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1683
        goto cleanup;
1684

G
Gene Czarcinski 已提交
1685
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
J
Ján Tomko 已提交
1686
        goto cleanup;
1687 1688

    if ((ret = dnsmasqSave(dctx)) < 0)
1689
        goto cleanup;
1690 1691

    ret = kill(network->dnsmasqPid, SIGHUP);
1692
 cleanup:
1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704
    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
1705 1706
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1707 1708 1709 1710 1711 1712
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1713
    }
1714
    /* now start dnsmasq if it should be started */
1715
    return networkStartDhcpDaemon(driver, network);
1716 1717
}

G
Gene Czarcinski 已提交
1718 1719 1720 1721 1722 1723
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";

1724 1725 1726
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1727
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1728 1729
    int ret = -1;
    size_t i;
1730
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
1731
    bool v6present = false, dhcp6 = false;
1732 1733

    *configstr = NULL;
1734

G
Gene Czarcinski 已提交
1735
    /* Check if DHCPv6 is needed */
1736
    for (i = 0;
1737
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1738
         i++) {
G
Gene Czarcinski 已提交
1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751
        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;
    }

1752 1753 1754
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1755
    virBufferAsprintf(&configbuf, "interface %s\n"
1756 1757
                      "{\n"
                      "  AdvSendAdvert on;\n"
1758
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1759 1760 1761 1762 1763
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1764 1765

    /* add a section for each IPv6 address in the config */
1766
    for (i = 0;
1767
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1768
         i++) {
1769 1770 1771
        int prefix;
        char *netaddr;

1772
        prefix = virNetworkIPDefPrefix(ipdef);
1773
        if (prefix < 0) {
1774 1775 1776
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1777 1778
            goto cleanup;
        }
1779
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1780
            goto cleanup;
1781
        virBufferAsprintf(&configbuf,
1782
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1783 1784 1785
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1786 1787 1788
        VIR_FREE(netaddr);
    }

1789
    virBufferAddLit(&configbuf, "};\n");
1790

1791
    if (virBufferCheckError(&configbuf) < 0)
1792
        goto cleanup;
1793

1794 1795
    *configstr = virBufferContentAndReset(&configbuf);

1796
    ret = 0;
1797
 cleanup:
1798 1799 1800 1801
    virBufferFreeAndReset(&configbuf);
    return ret;
}

1802
/* write file and return its name (which must be freed by caller) */
1803
static int
1804 1805 1806
networkRadvdConfWrite(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr network,
                      char **configFile)
1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821
{
    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;
1822 1823 1824 1825
        goto cleanup;
    }

    /* construct the filename */
1826
    if (!(*configFile = networkRadvdConfigFileName(driver, network->def->name)))
1827 1828
        goto cleanup;
    /* write the file */
1829
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1830 1831
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1832 1833 1834 1835 1836
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1837
 cleanup:
1838 1839 1840 1841 1842 1843
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1844 1845
networkStartRadvd(virNetworkDriverStatePtr driver,
                  virNetworkObjPtr network)
1846
{
1847
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1848 1849 1850 1851 1852 1853 1854 1855
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1856
    /* Is dnsmasq handling RA? */
1857
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
G
Gene Czarcinski 已提交
1858 1859 1860 1861
        ret = 0;
        goto cleanup;
    }

1862
    if (!virNetworkDefGetIPByIndex(network->def, AF_INET6, 0)) {
1863 1864 1865 1866 1867 1868 1869 1870 1871 1872
        /* 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);
1873 1874 1875
        goto cleanup;
    }

1876
    if (virFileMakePath(driver->pidDir) < 0) {
1877 1878
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1879
                             driver->pidDir);
1880 1881
        goto cleanup;
    }
1882
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1883 1884
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1885
                             driver->radvdStateDir);
1886 1887 1888 1889
        goto cleanup;
    }

    /* construct pidfile name */
1890
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1891
        goto cleanup;
1892
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1893 1894
        goto cleanup;

1895
    if (networkRadvdConfWrite(driver, network, &configfile) < 0)
1896 1897
        goto cleanup;

1898 1899 1900 1901
    /* 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
1902
     * virPidFileRead() below will fail if we use them).
1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917
     * 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;

1918
    if (virPidFileRead(driver->pidDir, radvdpidbase, &network->radvdPid) < 0)
1919 1920 1921
        goto cleanup;

    ret = 0;
1922
 cleanup:
1923
    virObjectUnref(dnsmasq_caps);
1924 1925 1926 1927 1928 1929 1930
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1931
static int
1932 1933
networkRefreshRadvd(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
1934
{
1935
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
G
Gene Czarcinski 已提交
1936 1937 1938
    char *radvdpidbase;

    /* Is dnsmasq handling RA? */
1939 1940
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
        virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1941 1942 1943 1944 1945 1946 1947
        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)) {
1948
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1949 1950 1951 1952 1953
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }
1954
    virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1955

1956 1957
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
1958
        return networkStartRadvd(driver, network);
1959

1960
    if (!virNetworkDefGetIPByIndex(network->def, AF_INET6, 0)) {
1961 1962 1963 1964
        /* no IPv6 addresses, so we don't need to run radvd */
        return 0;
    }

1965
    if (networkRadvdConfWrite(driver, network, NULL) < 0)
1966 1967 1968 1969 1970
        return -1;

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

1971 1972
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1973
static int
1974
networkRestartRadvd(virNetworkObjPtr network)
1975 1976 1977 1978 1979 1980 1981 1982 1983
{
    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
1984
        if ((networkKillDaemon(network->radvdPid, "radvd",
1985 1986 1987
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1988
            virPidFileDelete(driver->pidDir, radvdpidbase);
1989 1990 1991 1992 1993 1994 1995 1996 1997
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1998 1999
static int
networkRefreshDaemonsHelper(virNetworkObjPtr net,
2000
                            void *opaque)
2001
{
2002
    virNetworkDriverStatePtr driver = opaque;
2003

2004
    virObjectLock(net);
2005 2006 2007
    if (virNetworkObjIsActive(net) &&
        ((net->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
2008 2009
         (net->def->forward.type == VIR_NETWORK_FORWARD_ROUTE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_OPEN))) {
2010 2011 2012 2013 2014 2015
        /* 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.
         */
2016 2017
        networkRefreshDhcpDaemon(driver, net);
        networkRefreshRadvd(driver, net);
2018
    }
2019
    virObjectUnlock(net);
2020 2021 2022
    return 0;
}

2023 2024 2025 2026
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
2027
networkRefreshDaemons(virNetworkDriverStatePtr driver)
2028 2029
{
    VIR_INFO("Refreshing network daemons");
2030 2031
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
2032
                             driver);
2033
}
2034

2035 2036 2037 2038 2039
static int
networkReloadFirewallRulesHelper(virNetworkObjPtr net,
                                 void *opaque ATTRIBUTE_UNUSED)
{

2040
    virObjectLock(net);
2041 2042 2043 2044
    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))) {
2045 2046 2047 2048
        /* 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.
2049 2050 2051 2052
         */
        networkRemoveFirewallRules(net->def);
        if (networkAddFirewallRules(net->def) < 0) {
            /* failed to add but already logged */
2053 2054
        }
    }
2055
    virObjectUnlock(net);
2056
    return 0;
2057 2058
}

2059
static void
2060
networkReloadFirewallRules(virNetworkDriverStatePtr driver)
2061
{
2062
    VIR_INFO("Reloading iptables rules");
2063 2064 2065
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
2066 2067
}

2068
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
2069
static int
2070
networkEnableIPForwarding(bool enableIPv4, bool enableIPv6)
2071
{
2072
    int ret = 0;
2073 2074 2075 2076
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
2077
                           &enabled, sizeof(enabled));
2078 2079
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
2080
                           &enabled, sizeof(enabled));
2081
#else
2082 2083 2084 2085
    if (enableIPv4)
        ret = virFileWriteStr("/proc/sys/net/ipv4/ip_forward", "1\n", 0);
    if (enableIPv6 && ret == 0)
        ret = virFileWriteStr("/proc/sys/net/ipv6/conf/all/forwarding", "1\n", 0);
2086
#endif
2087
    return ret;
2088 2089
}

2090 2091
#define SYSCTL_PATH "/proc/sys"

2092 2093
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
2094 2095 2096
{
    char *field = NULL;
    int ret = -1;
2097
    bool enableIPv6 =  !!virNetworkDefGetIPByIndex(network->def, AF_INET6, 0);
2098

2099 2100 2101 2102 2103 2104 2105
    /* 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;
2106

2107 2108
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
2109 2110
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
2111 2112 2113
        ret = 0;
        goto cleanup;
    }
2114

2115 2116 2117 2118 2119
    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;
2120
    }
2121
    VIR_FREE(field);
2122

2123 2124
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
2125 2126 2127 2128 2129 2130
     */

    /* 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",
2131
                    network->def->bridge) < 0)
2132 2133
        goto cleanup;

2134
    if (virFileWriteStr(field, "0", 0) < 0) {
2135
        virReportSystemError(errno,
2136 2137 2138 2139 2140
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

2141 2142 2143 2144
    /* 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",
2145
                    network->def->bridge) < 0)
2146 2147
        goto cleanup;

2148
    if (virFileWriteStr(field, "0", 0) < 0) {
2149
        virReportSystemError(errno,
2150
                             _("cannot disable %s"), field);
2151 2152 2153 2154
        goto cleanup;
    }

    ret = 0;
2155
 cleanup:
2156 2157 2158 2159
    VIR_FREE(field);
    return ret;
}

2160
/* add an IP address to a bridge */
2161
static int
D
Daniel P. Berrange 已提交
2162
networkAddAddrToBridge(virNetworkObjPtr network,
2163
                       virNetworkIPDefPtr ipdef)
2164
{
2165
    int prefix = virNetworkIPDefPrefix(ipdef);
2166 2167

    if (prefix < 0) {
2168 2169 2170
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
2171 2172 2173
        return -1;
    }

2174 2175
    if (virNetDevIPAddrAdd(network->def->bridge,
                           &ipdef->address, NULL, prefix) < 0)
2176 2177 2178 2179 2180
        return -1;

    return 0;
}

2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203

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;
}


2204 2205 2206
/* add an IP (static) route to a bridge */
static int
networkAddRouteToBridge(virNetworkObjPtr network,
2207
                        virNetDevIPRoutePtr routedef)
2208
{
2209 2210 2211 2212
    int prefix = virNetDevIPRouteGetPrefix(routedef);
    unsigned int metric = virNetDevIPRouteGetMetric(routedef);
    virSocketAddrPtr addr = virNetDevIPRouteGetAddress(routedef);
    virSocketAddrPtr gateway = virNetDevIPRouteGetGateway(routedef);
2213 2214 2215 2216 2217 2218 2219 2220 2221

    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;
    }

2222 2223
    if (virNetDevIPRouteAdd(network->def->bridge, addr,
                            prefix, gateway, metric) < 0) {
2224 2225 2226 2227 2228
        return -1;
    }
    return 0;
}

2229 2230 2231
static int
networkWaitDadFinish(virNetworkObjPtr network)
{
2232
    virNetworkIPDefPtr ipdef;
2233 2234 2235 2236 2237 2238
    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);

2239
    while ((ipdef = virNetworkDefGetIPByIndex(network->def,
2240 2241 2242 2243 2244 2245
                                              AF_INET6, naddrs))) {
        addr = &ipdef->address;
        if (VIR_APPEND_ELEMENT_COPY(addrs, naddrs, addr) < 0)
            goto cleanup;
    }

2246
    ret = (naddrs == 0) ? 0 : virNetDevIPWaitDadFinish(addrs, naddrs);
2247 2248 2249 2250 2251 2252 2253 2254

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

2255
static int
2256 2257
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
                           virNetworkObjPtr network)
2258
{
2259
    size_t i;
2260
    bool v4present = false, v6present = false;
2261
    virErrorPtr save_err = NULL;
2262
    virNetworkIPDefPtr ipdef;
2263
    virNetDevIPRoutePtr routedef;
2264
    char *macTapIfName = NULL;
M
Michal Privoznik 已提交
2265
    char *macMapFile = NULL;
2266
    int tapfd = -1;
2267

2268
    /* Check to see if any network IP collides with an existing route */
2269
    if (networkCheckRouteCollision(network->def) < 0)
2270 2271
        return -1;

2272
    /* Create and configure the bridge device */
2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286
    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;
    }
2287
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
2288 2289
        return -1;

2290 2291 2292 2293 2294 2295 2296 2297
    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);
2298
        if (!macTapIfName)
2299
            goto err0;
2300
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2301
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
2302
                                           &macTapIfName, &network->def->mac,
2303
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2304
                                           network->def->mtu, NULL,
2305 2306 2307
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2308 2309 2310 2311 2312
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

M
Michal Privoznik 已提交
2313 2314 2315 2316
    if (!(macMapFile = networkMacMgrFileName(driver, network->def->bridge)) ||
        !(network->macmap = virMacMapNew(macMapFile)))
        goto err1;

2317
    /* Set bridge options */
2318 2319 2320 2321

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2322
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2323
                                   network->def->delay * 1000) < 0)
2324
        goto err1;
2325

2326
    if (virNetDevBridgeSetSTP(network->def->bridge,
2327
                              network->def->stp ? true : false) < 0)
2328
        goto err1;
2329

2330 2331 2332 2333
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2334
        goto err1;
2335

2336
    /* Add "once per network" rules */
2337 2338
    if (network->def->forward.type != VIR_NETWORK_FORWARD_OPEN &&
        networkAddFirewallRules(network->def) < 0)
2339 2340
        goto err1;

2341
    for (i = 0;
2342
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i));
2343
         i++) {
2344
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2345
            v4present = true;
2346
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2347
            v6present = true;
2348

2349
        /* Add the IP address/netmask to the bridge */
2350
        if (networkAddAddrToBridge(network, ipdef) < 0)
2351
            goto err2;
2352 2353
    }

2354 2355 2356
    if (networkStartHandleMACTableManagerMode(network, macTapIfName) < 0)
        goto err2;

2357
    /* Bring up the bridge interface */
2358
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2359
        goto err2;
2360

2361
    for (i = 0; i < network->def->nroutes; i++) {
2362 2363 2364
        virSocketAddrPtr gateway = NULL;

        routedef = network->def->routes[i];
2365
        gateway = virNetDevIPRouteGetGateway(routedef);
2366

2367 2368 2369
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2370
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2371 2372 2373 2374 2375 2376 2377
            if (networkAddRouteToBridge(network, routedef) < 0) {
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2378 2379
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2380
        networkEnableIPForwarding(v4present, v6present) < 0) {
2381
        virReportSystemError(errno, "%s",
2382
                             _("failed to enable IP forwarding"));
2383
        goto err3;
2384 2385
    }

2386

2387
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2388
    if ((v4present || v6present) &&
2389
        networkStartDhcpDaemon(driver, network) < 0)
2390
        goto err3;
2391

2392
    /* start radvd if there are any ipv6 addresses */
2393
    if (v6present && networkStartRadvd(driver, network) < 0)
2394 2395
        goto err4;

2396 2397 2398 2399 2400 2401 2402 2403
    /* 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.
2404 2405 2406 2407 2408 2409 2410
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
            goto err4;
        VIR_FORCE_CLOSE(tapfd);
    }

2411
    if (virNetDevBandwidthSet(network->def->bridge,
2412
                              network->def->bandwidth, true) < 0)
2413 2414
        goto err5;

2415
    VIR_FREE(macTapIfName);
M
Michal Privoznik 已提交
2416
    VIR_FREE(macMapFile);
2417 2418 2419

    return 0;

2420
 err5:
2421 2422
    if (network->def->bandwidth)
       virNetDevBandwidthClear(network->def->bridge);
2423

2424 2425 2426 2427
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2428 2429 2430 2431 2432
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2433 2434 2435
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2436
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2437

2438 2439 2440
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2441 2442
    if (network->def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(network->def);
2443 2444

 err1:
2445 2446 2447
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2448
    if (macTapIfName) {
2449
        VIR_FORCE_CLOSE(tapfd);
2450
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2451 2452
        VIR_FREE(macTapIfName);
    }
M
Michal Privoznik 已提交
2453
    VIR_FREE(macMapFile);
2454 2455

 err0:
2456 2457
    if (!save_err)
        save_err = virSaveLastError();
2458
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2459

2460 2461 2462 2463
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2464
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2465 2466 2467
    return -1;
}

2468 2469 2470
static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
                              virNetworkObjPtr network)
2471
{
2472 2473
    if (network->def->bandwidth)
        virNetDevBandwidthClear(network->def->bridge);
2474

M
Michal Privoznik 已提交
2475 2476
    virObjectUnref(network->macmap);

2477 2478 2479 2480 2481
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2482
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2483
            virPidFileDelete(driver->pidDir, radvdpidbase);
2484 2485 2486 2487
            VIR_FREE(radvdpidbase);
        }
    }

2488 2489 2490
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2491
    if (network->def->mac_specified) {
2492
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2493
        if (macTapIfName) {
2494
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2495 2496 2497 2498
            VIR_FREE(macTapIfName);
        }
    }

2499
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2500

2501 2502
    if (network->def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(network->def);
2503

2504
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2505

2506
    /* See if its still alive and really really kill it */
2507
    if (network->dnsmasqPid > 0 &&
2508
        (kill(network->dnsmasqPid, 0) == 0))
2509 2510
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2511 2512 2513 2514 2515 2516

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

2517 2518 2519
    return 0;
}

2520

2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541
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;
}


2542 2543 2544 2545 2546 2547 2548 2549 2550
/* 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;
2551
    unsigned int maxVirtFns = 0;
2552 2553 2554 2555 2556 2557
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2558 2559 2560
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2561 2562
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev, &vfNames,
                                      &virtFns, &numVirtFns, &maxVirtFns)) < 0) {
2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 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
        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:
2609
        case VIR_NETWORK_FORWARD_OPEN:
2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 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
        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;
}


2650
static int
2651
networkStartNetworkExternal(virNetworkObjPtr network)
2652 2653
{
    /* put anything here that needs to be done each time a network of
2654
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2655 2656 2657
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2658
    return networkCreateInterfacePool(network->def);
2659 2660
}

2661
static int networkShutdownNetworkExternal(virNetworkObjPtr network ATTRIBUTE_UNUSED)
2662 2663
{
    /* put anything here that needs to be done each time a network of
2664
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2665 2666 2667 2668 2669 2670 2671
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2672 2673
networkStartNetwork(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
2674
{
2675 2676 2677
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2678 2679

    if (virNetworkObjIsActive(network)) {
2680 2681
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2682
        return ret;
2683 2684
    }

2685 2686 2687
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2688
    if (virNetworkObjSetDefTransient(network, true) < 0)
2689
        goto cleanup;
2690

2691 2692
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2693
    if (networkRunHook(network, NULL, NULL,
2694 2695 2696 2697
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2698
    switch (network->def->forward.type) {
2699 2700 2701 2702

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2703
    case VIR_NETWORK_FORWARD_OPEN:
2704
        if (networkStartNetworkVirtual(driver, network) < 0)
2705
            goto cleanup;
2706 2707 2708
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2709 2710 2711 2712 2713 2714 2715 2716 2717
        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).
         */
2718 2719
        ATTRIBUTE_FALLTHROUGH;

2720 2721 2722
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2723
    case VIR_NETWORK_FORWARD_HOSTDEV:
2724
        if (networkStartNetworkExternal(network) < 0)
2725
            goto cleanup;
2726 2727 2728
        break;
    }

2729
    /* finally we can call the 'started' hook script if any */
2730
    if (networkRunHook(network, NULL, NULL,
2731 2732 2733 2734
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2735 2736 2737
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2738
    VIR_DEBUG("Writing network status to disk");
2739
    if (virNetworkSaveStatus(driver->stateDir, network) < 0)
2740
        goto cleanup;
2741 2742

    network->active = 1;
2743 2744
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2745

2746
 cleanup:
2747
    if (ret < 0) {
2748
        virNetworkObjUnsetDefTransient(network);
2749 2750
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2751
        networkShutdownNetwork(driver, network);
2752 2753 2754 2755 2756 2757 2758
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2759 2760 2761
static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
2762 2763 2764 2765 2766 2767 2768 2769 2770
{
    int ret = 0;
    char *stateFile;

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

    if (!virNetworkObjIsActive(network))
        return 0;

2771
    stateFile = virNetworkConfigFile(driver->stateDir,
2772
                                     network->def->name);
2773 2774 2775 2776 2777 2778
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2779
    switch (network->def->forward.type) {
2780 2781 2782 2783

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2784
    case VIR_NETWORK_FORWARD_OPEN:
2785
        ret = networkShutdownNetworkVirtual(driver, network);
2786 2787 2788
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2789 2790 2791 2792 2793 2794 2795 2796
        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).
         */
2797 2798
        ATTRIBUTE_FALLTHROUGH;

2799 2800 2801
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2802
    case VIR_NETWORK_FORWARD_HOSTDEV:
2803
        ret = networkShutdownNetworkExternal(network);
2804 2805 2806
        break;
    }

2807
    /* now that we know it's stopped call the hook if present */
2808
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2809 2810
                   VIR_HOOK_SUBOP_END);

2811
    network->active = 0;
2812
    virNetworkObjUnsetDefTransient(network);
2813
    return ret;
2814 2815 2816
}


2817
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2818 2819
                                         const unsigned char *uuid)
{
2820
    virNetworkDriverStatePtr driver = networkGetDriver();
2821 2822
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2823

2824
    network = virNetworkObjFindByUUID(driver->networks, uuid);
2825
    if (!network) {
2826 2827
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2828
        virReportError(VIR_ERR_NO_NETWORK,
2829 2830
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2831
        goto cleanup;
2832 2833
    }

2834 2835 2836
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2837 2838
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2839
 cleanup:
2840
    virNetworkObjEndAPI(&network);
2841
    return ret;
2842 2843
}

2844
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2845 2846
                                         const char *name)
{
2847
    virNetworkDriverStatePtr driver = networkGetDriver();
2848 2849 2850
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2851
    network = virNetworkObjFindByName(driver->networks, name);
2852
    if (!network) {
2853 2854
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2855
        goto cleanup;
2856 2857
    }

2858 2859 2860
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2861 2862
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2863
 cleanup:
2864
    virNetworkObjEndAPI(&network);
2865
    return ret;
2866 2867
}

2868 2869
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2870
    virNetworkDriverStatePtr driver = networkGetDriver();
2871
    int nactive;
2872

2873 2874 2875
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2876 2877 2878 2879
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
2880

2881 2882 2883
    return nactive;
}

2884 2885 2886 2887 2888
static int networkConnectListNetworks(virConnectPtr conn,
                                      char **const names,
                                      int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2889
    int got = 0;
2890

2891 2892 2893
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2894 2895 2896 2897
    got = virNetworkObjListGetNames(driver->networks,
                                    true, names, nnames,
                                    virConnectListNetworksCheckACL,
                                    conn);
2898

2899 2900 2901
    return got;
}

2902 2903
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2904
    virNetworkDriverStatePtr driver = networkGetDriver();
2905
    int ninactive = 0;
2906

2907 2908 2909
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2910 2911 2912 2913
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
2914

2915 2916 2917
    return ninactive;
}

2918 2919 2920
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2921
    int got = 0;
2922

2923 2924 2925
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2926 2927 2928 2929
    got = virNetworkObjListGetNames(driver->networks,
                                    false, names, nnames,
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
2930 2931 2932
    return got;
}

2933
static int
2934 2935 2936
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2937
{
2938
    virNetworkDriverStatePtr driver = networkGetDriver();
2939 2940 2941 2942
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2943 2944 2945
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2946
    ret = virNetworkObjListExport(conn, driver->networks, nets,
2947 2948
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2949

2950
 cleanup:
2951 2952
    return ret;
}
2953

2954 2955 2956 2957 2958 2959 2960 2961
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
2962
    virNetworkDriverStatePtr driver = networkGetDriver();
2963 2964 2965 2966 2967 2968
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
2969
                                       net, eventID, callback,
2970 2971 2972
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2973
 cleanup:
2974 2975 2976 2977 2978 2979 2980
    return ret;
}

static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
2981
    virNetworkDriverStatePtr driver = networkGetDriver();
2982 2983 2984 2985 2986
    int ret = -1;

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

2987 2988 2989 2990 2991 2992
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
2993

2994
 cleanup:
2995 2996 2997
    return ret;
}

2998 2999 3000 3001 3002
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

3003 3004
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3005 3006 3007 3008

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

3009 3010
    ret = virNetworkObjIsActive(obj);

3011
 cleanup:
3012
    virNetworkObjEndAPI(&obj);
3013 3014 3015 3016 3017 3018 3019 3020
    return ret;
}

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

3021 3022
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3023 3024 3025 3026

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

3027 3028
    ret = obj->persistent;

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


3035 3036
/*
 * networkFindUnusedBridgeName() - try to find a bridge name that is
3037 3038 3039
 * 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.
3040 3041 3042 3043 3044 3045 3046 3047
 */
static int
networkFindUnusedBridgeName(virNetworkObjListPtr nets,
                            virNetworkDefPtr def)
{

    int ret = -1, id = 0;
    char *newname = NULL;
3048 3049 3050 3051 3052
    const char *templ = "virbr%d";
    const char *p;

    if (def->bridge &&
        (p = strchr(def->bridge, '%')) == strrchr(def->bridge, '%') &&
3053
        p && p[1] == 'd')
3054
        templ = def->bridge;
3055 3056 3057 3058

    do {
        if (virAsprintf(&newname, templ, id) < 0)
            goto cleanup;
3059 3060 3061 3062 3063 3064 3065
        /* check if this name is used in another libvirt network or
         * there is an existing device with that name. ignore errors
         * from virNetDevExists(), just in case it isn't implemented
         * on this platform (probably impossible).
         */
        if (!(virNetworkBridgeInUse(nets, newname, def->name) ||
              virNetDevExists(newname) == 1)) {
3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 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 3115 3116 3117
            VIR_FREE(def->bridge); /*could contain template */
            def->bridge = newname;
            ret = 0;
            goto cleanup;
        }
        VIR_FREE(newname);
    } while (++id <= MAX_BRIDGE_ID);

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



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

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

    ret = 0;
 cleanup:
    return ret;
}


3118
static int
3119
networkValidate(virNetworkDriverStatePtr driver,
3120
                virNetworkDefPtr def)
3121
{
3122
    size_t i, j;
3123 3124
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
3125
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
3126
    bool ipv4def = false, ipv6def = false;
3127
    bool bandwidthAllowed = true;
3128
    bool usesInterface = false, usesAddress = false;
3129

3130 3131 3132
    if (virXMLCheckIllegalChars("name", def->name, "\n") < 0)
        return -1;

3133 3134 3135
    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
3136 3137
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
3138 3139
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE ||
        def->forward.type == VIR_NETWORK_FORWARD_OPEN) {
3140

3141 3142 3143 3144
        /* 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)
3145 3146 3147
            return -1;

        virNetworkSetBridgeMacAddr(def);
3148 3149
    } else {
        /* They are also the only types that currently support setting
3150 3151
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
3152
         */
3153 3154 3155 3156 3157 3158 3159 3160
        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;
        }
3161
        if (virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0)) {
3162 3163 3164 3165
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3166
                           virNetworkForwardTypeToString(def->forward.type));
3167 3168
            return -1;
        }
3169
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
3170 3171 3172 3173
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3174
                           virNetworkForwardTypeToString(def->forward.type));
3175 3176 3177 3178 3179 3180 3181
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3182
                           virNetworkForwardTypeToString(def->forward.type));
3183 3184
            return -1;
        }
3185 3186 3187 3188 3189 3190 3191 3192
        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;
        }
3193
        bandwidthAllowed = false;
3194 3195
    }

3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229
    /* we support configs with a single PF defined:
     *   <pf dev='eth0'/>
     * or with a list of netdev names:
     *   <interface dev='eth9'/>
     * OR a list of PCI addresses
     *   <address type='pci' domain='0' bus='4' slot='0' function='1'/>
     * but not any combination of those.
     *
     * Since <interface> and <address> are for some strange reason
     * stored in the same array, we need to cycle through it and check
     * the type of each.
     */
    for (i = 0; i < def->forward.nifs; i++) {
        switch ((virNetworkForwardHostdevDeviceType)
                def->forward.ifs[i].type) {
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV:
            usesInterface = true;
            break;
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI:
            usesAddress = true;
            break;
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NONE:
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_LAST:
            break;
        }
    }
    if ((def->forward.npfs > 0) + usesInterface + usesAddress > 1) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<address>, <interface>, and <pf> elements of "
                         "<forward> in network %s are mutually exclusive"),
                       def->name);
        return -1;
    }

G
Gene Czarcinski 已提交
3230 3231 3232
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
3233
    for (i = 0;
3234
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
3235
         i++) {
G
Gene Czarcinski 已提交
3236 3237 3238 3239 3240
        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 -- "
3241 3242
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259
                    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;
                }
3260 3261 3262
            }
        }
    }
3263 3264 3265 3266 3267 3268

    /* 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.
     */

3269 3270 3271
    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
3272 3273
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
3274
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH));
3275 3276

    vlanUsed = def->vlan.nTags > 0;
3277 3278
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3279 3280 3281 3282 3283
            /* 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.
             */
3284
            if (def->portGroups[i].virtPortProfile) {
3285
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3286
                    def->portGroups[i].virtPortProfile->virtPortType
3287 3288 3289 3290 3291 3292 3293
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3294
        }
3295
        if (def->portGroups[i].isDefault) {
3296 3297 3298 3299 3300
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3301
                               def->name, defaultPortGroup->name,
3302
                               def->portGroups[i].name);
3303
                return -1;
3304
            }
3305
            defaultPortGroup = &def->portGroups[i];
3306
        }
3307 3308 3309 3310 3311 3312 3313 3314 3315
        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;
            }
        }
3316 3317 3318 3319 3320 3321 3322 3323
        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;
        }
3324
    }
3325 3326 3327 3328 3329 3330 3331
    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.
         */
3332 3333 3334 3335 3336 3337
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351

    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;
            }
        }
    }
3352 3353 3354
    return 0;
}

3355 3356
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
3357
    virNetworkDriverStatePtr driver = networkGetDriver();
3358
    virNetworkDefPtr def;
3359
    virNetworkObjPtr network = NULL;
3360
    virNetworkPtr ret = NULL;
3361
    virObjectEventPtr event = NULL;
3362

3363
    if (!(def = virNetworkDefParseString(xml)))
3364
        goto cleanup;
3365

3366 3367 3368
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3369
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3370
        goto cleanup;
3371

3372 3373 3374
    /* 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.
3375
     */
3376 3377 3378
    if (!(network = virNetworkAssignDef(driver->networks, def,
                                        VIR_NETWORK_OBJ_LIST_ADD_LIVE |
                                        VIR_NETWORK_OBJ_LIST_ADD_CHECK_LIVE)))
3379 3380
        goto cleanup;
    def = NULL;
3381

3382
    if (networkStartNetwork(driver, network) < 0) {
3383
        virNetworkRemoveInactive(driver->networks,
3384
                                 network);
3385
        goto cleanup;
3386 3387
    }

3388 3389
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3390 3391
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3392

3393
    VIR_INFO("Creating network '%s'", network->def->name);
3394 3395
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

3396
 cleanup:
3397
    virNetworkDefFree(def);
3398 3399
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3400
    virNetworkObjEndAPI(&network);
3401
    return ret;
3402 3403
}

3404 3405
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
3406
    virNetworkDriverStatePtr driver = networkGetDriver();
3407
    virNetworkDefPtr def = NULL;
3408
    bool freeDef = true;
3409
    virNetworkObjPtr network = NULL;
3410
    virNetworkPtr ret = NULL;
3411
    virObjectEventPtr event = NULL;
3412

3413
    if (!(def = virNetworkDefParseString(xml)))
3414
        goto cleanup;
3415

3416 3417 3418
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3419
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3420
        goto cleanup;
3421

3422
    if (!(network = virNetworkAssignDef(driver->networks, def, 0)))
J
Ján Tomko 已提交
3423
        goto cleanup;
3424

3425
    /* def was assigned to network object */
3426
    freeDef = false;
3427 3428

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3429
        if (!virNetworkObjIsActive(network)) {
3430
            virNetworkRemoveInactive(driver->networks, network);
3431 3432
            goto cleanup;
        }
3433 3434 3435 3436 3437
        /* if network was active already, just undo new persistent
         * definition by making it transient.
         * XXX - this isn't necessarily the correct thing to do.
         */
        virNetworkObjAssignDef(network, NULL, false);
3438 3439 3440
        goto cleanup;
    }

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

3445 3446
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
3447

3448
 cleanup:
3449 3450
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3451
    if (freeDef)
J
Ján Tomko 已提交
3452
        virNetworkDefFree(def);
3453
    virNetworkObjEndAPI(&network);
3454
    return ret;
3455 3456
}

3457
static int
3458 3459
networkUndefine(virNetworkPtr net)
{
3460
    virNetworkDriverStatePtr driver = networkGetDriver();
3461
    virNetworkObjPtr network;
3462
    int ret = -1;
3463
    bool active = false;
3464
    virObjectEventPtr event = NULL;
3465

3466
    if (!(network = networkObjFromNetwork(net)))
3467
        goto cleanup;
3468

3469 3470 3471
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3472 3473
    if (virNetworkObjIsActive(network))
        active = true;
3474

3475 3476 3477 3478 3479 3480
    if (!network->persistent) {
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                       _("can't undefine transient network"));
        goto cleanup;
    }

3481
    /* remove autostart link */
3482
    if (virNetworkDeleteConfig(driver->networkConfigDir,
3483 3484
                               driver->networkAutostartDir,
                               network) < 0)
3485
        goto cleanup;
3486

3487 3488
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3489 3490
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3491

3492
    VIR_INFO("Undefining network '%s'", network->def->name);
3493
    if (!active) {
3494
        if (networkRemoveInactive(driver, network) < 0)
3495
            goto cleanup;
3496 3497 3498 3499 3500 3501
    } else {

        /* if the network still exists, it was active, and we need to make
         * it transient (by deleting the persistent def)
         */
        virNetworkObjAssignDef(network, NULL, false);
3502 3503
    }

3504
    ret = 0;
3505

3506
 cleanup:
3507 3508
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3509
    virNetworkObjEndAPI(&network);
3510
    return ret;
3511 3512
}

3513 3514 3515 3516 3517 3518 3519 3520
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3521
    virNetworkDriverStatePtr driver = networkGetDriver();
3522
    virNetworkObjPtr network = NULL;
3523 3524
    int isActive, ret = -1;
    size_t i;
3525
    virNetworkIPDefPtr ipdef;
3526
    bool oldDhcpActive = false;
3527
    bool needFirewallRefresh = false;
3528

3529 3530 3531 3532 3533

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3534
    if (!(network = networkObjFromNetwork(net)))
3535 3536
        goto cleanup;

3537 3538 3539
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3540
    /* see if we are listening for dhcp pre-modification */
3541
    for (i = 0;
3542
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET, i));
3543
         i++) {
3544 3545 3546 3547 3548 3549
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3550 3551
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3552
     */
3553
    isActive = virNetworkObjIsActive(network);
3554 3555
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3556 3557 3558 3559 3560 3561 3562
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579
    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).
                 */
3580 3581 3582 3583
                if (network->def->forward.type != VIR_NETWORK_FORWARD_OPEN) {
                    networkRemoveFirewallRules(network->def);
                    needFirewallRefresh = true;
                }
3584 3585 3586 3587 3588 3589 3590
                break;
            default:
                break;
            }
        }
    }

3591
    /* update the network config in memory/on disk */
3592 3593
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3594
            ignore_value(networkAddFirewallRules(network->def));
3595 3596 3597
        goto cleanup;
    }

3598
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617
        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 ||
3618 3619 3620 3621 3622 3623 3624 3625
            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)
3626
             */
3627
            if (networkRestartDhcpDaemon(driver, network) < 0)
3628 3629
                goto cleanup;

3630 3631 3632 3633 3634 3635 3636 3637
        } 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;

3638
            for (i = 0;
3639
                 (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET, i));
3640
                 i++) {
3641 3642 3643 3644 3645 3646 3647
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3648 3649
                 networkRestartDhcpDaemon(driver, network) < 0) ||
                networkRefreshDhcpDaemon(driver, network) < 0) {
3650 3651 3652
                goto cleanup;
            }

3653 3654 3655 3656
        } 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.
3657
             */
3658
            if (networkRefreshDhcpDaemon(driver, network) < 0)
3659 3660 3661 3662 3663 3664 3665 3666
                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.
             */
3667
            if (networkRefreshRadvd(driver, network) < 0)
3668 3669 3670 3671
                goto cleanup;
        }

        /* save current network state to disk */
3672
        if ((ret = virNetworkSaveStatus(driver->stateDir,
3673
                                        network)) < 0) {
3674
            goto cleanup;
3675
        }
3676
    }
3677 3678 3679 3680 3681 3682

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

3683
    ret = 0;
3684
 cleanup:
3685
    virNetworkObjEndAPI(&network);
3686 3687 3688
    return ret;
}

3689 3690
static int networkCreate(virNetworkPtr net)
{
3691
    virNetworkDriverStatePtr driver = networkGetDriver();
3692 3693
    virNetworkObjPtr network;
    int ret = -1;
3694
    virObjectEventPtr event = NULL;
3695

3696
    if (!(network = networkObjFromNetwork(net)))
3697
        goto cleanup;
3698

3699 3700 3701
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3702
    if ((ret = networkStartNetwork(driver, network)) < 0)
3703
        goto cleanup;
3704

3705 3706
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3707 3708
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3709

3710
 cleanup:
3711 3712
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3713
    virNetworkObjEndAPI(&network);
3714
    return ret;
3715 3716
}

3717 3718
static int networkDestroy(virNetworkPtr net)
{
3719
    virNetworkDriverStatePtr driver = networkGetDriver();
3720 3721
    virNetworkObjPtr network;
    int ret = -1;
3722
    virObjectEventPtr event = NULL;
3723

3724
    if (!(network = networkObjFromNetwork(net)))
3725
        goto cleanup;
3726

3727 3728 3729
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3730
    if (!virNetworkObjIsActive(network)) {
3731
        virReportError(VIR_ERR_OPERATION_INVALID,
3732 3733
                       _("network '%s' is not active"),
                       network->def->name);
3734 3735 3736
        goto cleanup;
    }

3737
    if ((ret = networkShutdownNetwork(driver, network)) < 0)
3738 3739
        goto cleanup;

3740 3741
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3742 3743
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3744

3745
    if (!network->persistent &&
3746
        networkRemoveInactive(driver, network) < 0) {
3747 3748
        ret = -1;
        goto cleanup;
3749
    }
3750

3751
 cleanup:
3752 3753
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3754
    virNetworkObjEndAPI(&network);
3755 3756 3757
    return ret;
}

3758
static char *networkGetXMLDesc(virNetworkPtr net,
3759
                               unsigned int flags)
3760
{
3761
    virNetworkObjPtr network;
3762
    virNetworkDefPtr def;
3763
    char *ret = NULL;
3764

3765
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3766

3767 3768
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3769

3770 3771 3772
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3773 3774 3775 3776 3777 3778
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3779

3780
 cleanup:
3781
    virNetworkObjEndAPI(&network);
3782
    return ret;
3783 3784 3785
}

static char *networkGetBridgeName(virNetworkPtr net) {
3786 3787 3788
    virNetworkObjPtr network;
    char *bridge = NULL;

3789 3790
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3791

3792 3793 3794
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3795
    if (!(network->def->bridge)) {
3796 3797 3798
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3799 3800 3801
        goto cleanup;
    }

3802
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3803

3804
 cleanup:
3805
    virNetworkObjEndAPI(&network);
3806 3807 3808 3809
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
J
Ján Tomko 已提交
3810
                               int *autostart)
3811
{
3812 3813
    virNetworkObjPtr network;
    int ret = -1;
3814

3815 3816
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3817

3818 3819 3820
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3821
    *autostart = network->autostart;
3822
    ret = 0;
3823

3824
 cleanup:
3825
    virNetworkObjEndAPI(&network);
3826
    return ret;
3827 3828 3829
}

static int networkSetAutostart(virNetworkPtr net,
3830 3831
                               int autostart)
{
3832
    virNetworkDriverStatePtr driver = networkGetDriver();
3833
    virNetworkObjPtr network;
3834
    char *configFile = NULL, *autostartLink = NULL;
3835
    int ret = -1;
3836

3837

3838
    if (!(network = networkObjFromNetwork(net)))
3839
        goto cleanup;
3840

3841 3842 3843
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3844
    if (!network->persistent) {
3845 3846
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3847 3848 3849
        goto cleanup;
    }

3850 3851
    autostart = (autostart != 0);

3852
    if (network->autostart != autostart) {
3853
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3854
            goto cleanup;
3855
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3856 3857
            goto cleanup;

3858
        if (autostart) {
3859
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3860
                virReportSystemError(errno,
3861 3862
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3863 3864
                goto cleanup;
            }
3865

3866
            if (symlink(configFile, autostartLink) < 0) {
3867
                virReportSystemError(errno,
3868
                                     _("Failed to create symlink '%s' to '%s'"),
3869
                                     autostartLink, configFile);
3870 3871 3872
                goto cleanup;
            }
        } else {
3873
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3874
                virReportSystemError(errno,
3875
                                     _("Failed to delete symlink '%s'"),
3876
                                     autostartLink);
3877 3878
                goto cleanup;
            }
3879 3880
        }

3881
        network->autostart = autostart;
3882
    }
3883
    ret = 0;
3884

3885
 cleanup:
3886 3887
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3888
    virNetworkObjEndAPI(&network);
3889
    return ret;
3890 3891
}

3892
static int
3893 3894 3895 3896
networkGetDHCPLeases(virNetworkPtr network,
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
3897
{
3898
    virNetworkDriverStatePtr driver = networkGetDriver();
3899 3900 3901
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
3902
    ssize_t size = 0;
3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913
    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;
3914
    virNetworkIPDefPtr ipdef_tmp = NULL;
3915 3916
    virNetworkDHCPLeasePtr lease = NULL;
    virNetworkDHCPLeasePtr *leases_ret = NULL;
3917
    virNetworkObjPtr obj;
3918
    virMacAddr mac_addr;
3919 3920 3921

    virCheckFlags(0, -1);

3922 3923 3924 3925 3926 3927
    /* only to check if the MAC is valid */
    if (mac && virMacAddrParse(mac, &mac_addr) < 0) {
        virReportError(VIR_ERR_INVALID_MAC, "%s", mac);
        return -1;
    }

3928 3929 3930 3931 3932
    if (!(obj = networkObjFromNetwork(network)))
        return -1;

    if (virNetworkGetDHCPLeasesEnsureACL(network->conn, obj->def) < 0)
        goto cleanup;
3933 3934

    /* Retrieve custom leases file location */
3935
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(driver, obj->def->bridge);
3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978

    /* 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;
        }

3979
        if (mac && virMacAddrCompare(mac, mac_tmp))
3980 3981 3982 3983 3984 3985 3986 3987 3988 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
            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)) {
4021
                    lease->prefix = virSocketAddrGetIPPrefix(&ipdef_tmp->address,
4022 4023 4024 4025 4026 4027 4028 4029
                                                             &ipdef_tmp->netmask,
                                                             ipdef_tmp->prefix);
                    break;
                }
            }

            if ((VIR_STRDUP(lease->mac, mac_tmp) < 0) ||
                (VIR_STRDUP(lease->ipaddr, ip_tmp) < 0) ||
4030
                (VIR_STRDUP(lease->iface, obj->def->bridge) < 0))
4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062
                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);
4063
    VIR_FREE(lease_entries);
4064 4065
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
4066

4067
    virNetworkObjEndAPI(&obj);
4068

4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079
    return rv;

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

4080 4081

static virNetworkDriver networkDriver = {
4082
    .name = "bridge",
4083 4084 4085 4086 4087
    .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 */
4088 4089
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
4090 4091
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
4092 4093
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
4094
    .networkUndefine = networkUndefine, /* 0.2.0 */
4095
    .networkUpdate = networkUpdate, /* 0.10.2 */
4096
    .networkCreate = networkCreate, /* 0.2.0 */
4097 4098 4099 4100 4101 4102 4103
    .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 */
4104
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
4105 4106 4107
};

static virStateDriver networkStateDriver = {
4108
    .name = "bridge",
4109
    .stateInitialize  = networkStateInitialize,
4110
    .stateAutoStart  = networkStateAutoStart,
4111 4112
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
4113 4114
};

4115 4116
int networkRegister(void)
{
4117
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
4118
        return -1;
4119 4120
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
4121 4122
    return 0;
}
4123 4124 4125

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

4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160
/* 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);
        }
    }
}

4161 4162 4163 4164 4165 4166 4167 4168 4169
/* 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:
4170
 * @dom: domain definition that @iface belongs to
4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181
 * @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
4182 4183
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
4184
{
4185
    virNetworkDriverStatePtr driver = networkGetDriver();
4186
    virDomainNetType actualType = iface->type;
4187 4188
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
4189
    virNetDevBandwidthPtr bandwidth = NULL;
4190 4191 4192
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
4193
    virNetworkForwardIfDefPtr dev = NULL;
4194
    size_t i;
4195 4196 4197
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
4198
        goto validate;
4199 4200 4201 4202

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

4203
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4204
    if (!network) {
4205 4206 4207
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4208
        goto error;
4209 4210
    }
    netdef = network->def;
4211

4212 4213 4214 4215 4216 4217 4218
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

4219 4220 4221
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

4222 4223 4224
    /* 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
4225
     */
4226 4227 4228 4229 4230 4231
    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.
     */
4232 4233 4234 4235 4236 4237

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

4238 4239
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
4240
        goto error;
4241

4242 4243 4244 4245 4246 4247 4248 4249
    /* 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;

4250 4251
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
4252

4253 4254 4255 4256 4257 4258 4259 4260 4261 4262
    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;

4263 4264
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
4265 4266
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_OPEN)) {
4267
        /* for these forward types, the actual net type really *is*
4268
         * NETWORK; we just keep the info from the portgroup in
4269
         * iface->data.network.actual
J
Ján Tomko 已提交
4270
         */
4271
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
4272

4273
        /* we also store the bridge device and macTableManager settings
4274 4275 4276 4277 4278 4279 4280
         * 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;
4281 4282
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4283

4284 4285 4286
        if (networkPlugBandwidth(network, iface) < 0)
            goto error;

4287
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
4288
               netdef->bridge) {
4289 4290 4291 4292 4293

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

4294
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
4295 4296
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
4297
            goto error;
4298 4299
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4300

4301 4302 4303 4304 4305 4306 4307 4308
        /* 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) {
4309
            goto error;
4310 4311 4312 4313 4314 4315 4316 4317 4318 4319
        }
        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);
4320
                goto error;
4321 4322 4323
            }
        }

4324
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4325

4326
        virDomainHostdevSubsysPCIBackendType backend;
4327

4328
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
4329
        if (networkCreateInterfacePool(netdef) < 0)
4330 4331 4332
            goto error;

        /* pick first dev with 0 connections */
4333 4334 4335
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349
                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;
4350
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
4351
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
4352
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
4353

E
Eric Blake 已提交
4354
        switch (netdef->forward.driverName) {
4355
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
4356
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
4357 4358
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
4359
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
4360 4361
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
4362
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373
            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;

4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398
        /* 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;
            }
        }

4399 4400 4401 4402
    } 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)) {
4403 4404 4405 4406 4407 4408

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
4409
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4410
        switch (netdef->forward.type) {
4411
        case VIR_NETWORK_FORWARD_BRIDGE:
4412
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
4413 4414
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
4415
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
4416 4417
            break;
        case VIR_NETWORK_FORWARD_VEPA:
4418
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
4419 4420
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
4421
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
4422 4423 4424
            break;
        }

4425 4426 4427 4428 4429 4430 4431 4432
        /* 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) {
4433
            goto error;
4434
        }
4435
        virtport = iface->data.network.actual->virtPortProfile;
4436
        if (virtport) {
4437 4438 4439 4440 4441 4442 4443 4444
            /* 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);
4445
                goto error;
4446 4447
            }
        }
4448

4449 4450 4451
        /* 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).
         */
4452
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4453 4454 4455 4456
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4457
            goto error;
4458 4459 4460
        } else {
            /* pick an interface from the pool */

4461
            if (networkCreateInterfacePool(netdef) < 0)
4462 4463
                goto error;

4464 4465 4466 4467 4468
            /* 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.
4469
             */
4470 4471
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4472 4473 4474
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4475

4476
                /* pick first dev with 0 connections */
4477 4478 4479
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4480 4481 4482 4483 4484
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4485
                dev = &netdef->forward.ifs[0];
4486 4487 4488
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4489 4490 4491 4492
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4493 4494 4495 4496
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4497
                goto error;
4498
            }
4499 4500
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4501
                goto error;
4502 4503 4504
        }
    }

M
Michal Privoznik 已提交
4505 4506 4507
    if (networkMacMgrAdd(driver, network, dom->name, &iface->mac) < 0)
        goto error;

4508
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4509
        goto error;
4510

4511
 validate:
4512 4513 4514 4515 4516
    /* 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.
     */

4517
    if (virDomainNetGetActualVlan(iface)) {
4518 4519 4520 4521
        /* 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
4522 4523
         */
        if (!(actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV ||
4524 4525 4526
              (actualType == VIR_DOMAIN_NET_TYPE_DIRECT &&
               virDomainNetGetActualDirectMode(iface)
               == VIR_NETDEV_MACVLAN_MODE_PASSTHRU) ||
4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545
              (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;
        }
    }
4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556
    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;
        }
    }
4557 4558 4559

    if (netdef) {
        netdef->connections++;
4560
        if (dev)
4561 4562 4563 4564 4565 4566
            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 */
4567
            netdef->connections--;
4568 4569 4570 4571
            if (dev)
                dev->connections--;
            goto error;
        }
4572
        networkLogAllocation(netdef, actualType, dev, iface, true);
4573 4574
    }

4575
    ret = 0;
4576

4577
 cleanup:
4578
    virNetworkObjEndAPI(&network);
4579 4580
    return ret;

4581
 error:
4582
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4583 4584 4585
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4586
    goto cleanup;
4587 4588 4589
}

/* networkNotifyActualDevice:
4590
 * @dom: domain definition that @iface belongs to
4591 4592 4593 4594 4595 4596 4597 4598 4599 4600
 * @iface:  the domain's NetDef with an "actual" device already filled in.
 *
 * Called to notify the network driver when libvirtd is restarted and
 * finds an already running domain. If appropriate it will force an
 * allocation of the actual->direct.linkdev to get everything back in
 * order.
 *
 * Returns 0 on success, -1 on failure.
 */
int
4601 4602
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4603
{
4604
    virNetworkDriverStatePtr driver = networkGetDriver();
4605
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4606 4607
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4608
    virNetworkForwardIfDefPtr dev = NULL;
4609 4610
    size_t i;
    int ret = -1;
4611 4612

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4613
        return 0;
4614

4615
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4616
    if (!network) {
4617 4618 4619
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4620 4621 4622 4623
        goto error;
    }
    netdef = network->def;

4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634
    /* 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;

4635
    if (!iface->data.network.actual ||
4636 4637
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4638 4639
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4640 4641
    }

4642
    if (networkCreateInterfacePool(netdef) < 0)
4643
        goto error;
4644

4645
    if (netdef->forward.nifs == 0) {
4646
        virReportError(VIR_ERR_INTERNAL_ERROR,
4647 4648
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4649
                       netdef->name);
4650
        goto error;
4651
    }
4652

4653 4654
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4655

4656 4657 4658 4659 4660 4661 4662 4663 4664
        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 */
4665 4666
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4667
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4668 4669
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4670 4671 4672 4673 4674
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4675
            virReportError(VIR_ERR_INTERNAL_ERROR,
4676 4677
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4678
                           netdef->name, actualDev);
4679
            goto error;
4680 4681
        }

4682
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4683 4684
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4685
         */
4686
        if ((dev->connections > 0) &&
4687 4688
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4689 4690
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4691
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4692
            virReportError(VIR_ERR_INTERNAL_ERROR,
4693 4694
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4695
                           netdef->name, actualDev);
4696
            goto error;
4697
        }
4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709
    }  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 */
4710 4711
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4712
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4713
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4714 4715
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4716 4717 4718 4719 4720 4721 4722 4723 4724
                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,
4725 4726 4727 4728
                           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
4729
            goto error;
4730 4731 4732 4733 4734 4735 4736
        }

        /* 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) &&
4737
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4738 4739 4740 4741 4742 4743 4744 4745 4746
            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;
        }
4747 4748
    }

4749
 success:
4750
    netdef->connections++;
4751 4752
    if (dev)
        dev->connections++;
4753 4754 4755 4756 4757 4758 4759 4760 4761
    /* 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;
    }
4762
    networkLogAllocation(netdef, actualType, dev, iface, true);
4763

4764
    ret = 0;
4765
 cleanup:
4766
    virNetworkObjEndAPI(&network);
4767
    return ret;
4768

4769
 error:
4770
    goto cleanup;
4771 4772 4773
}


4774

4775
/* networkReleaseActualDevice:
4776
 * @dom: domain definition that @iface belongs to
4777 4778 4779 4780 4781 4782 4783 4784 4785 4786
 * @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
4787 4788
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4789
{
4790
    virNetworkDriverStatePtr driver = networkGetDriver();
4791
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4792
    virNetworkObjPtr network;
4793
    virNetworkDefPtr netdef;
4794
    virNetworkForwardIfDefPtr dev = NULL;
4795 4796
    size_t i;
    int ret = -1;
4797 4798

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4799
        return 0;
4800

4801
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4802
    if (!network) {
4803 4804 4805
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4806 4807 4808 4809
        goto error;
    }
    netdef = network->def;

4810 4811
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4812
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
4813 4814
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE ||
         netdef->forward.type == VIR_NETWORK_FORWARD_OPEN) &&
4815 4816 4817
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4818 4819 4820
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4821 4822
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4823 4824
    }

4825
    if (netdef->forward.nifs == 0) {
4826
        virReportError(VIR_ERR_INTERNAL_ERROR,
4827
                       _("network '%s' uses a direct/hostdev mode, but "
4828 4829
                         "has no forward dev and no interface pool"),
                       netdef->name);
4830
        goto error;
4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842
    }

    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;
        }
4843

4844 4845
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4846
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4847 4848
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4849 4850 4851
                break;
            }
        }
4852

4853
        if (!dev) {
4854
            virReportError(VIR_ERR_INTERNAL_ERROR,
4855 4856
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4857
                           netdef->name, actualDev);
4858
            goto error;
4859
        }
4860 4861 4862 4863 4864 4865 4866 4867 4868 4869
    } 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;
        }

4870 4871
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4872
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4873
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4874 4875
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4876 4877 4878 4879 4880 4881 4882 4883 4884
                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,
4885 4886 4887 4888
                           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
4889
            goto error;
4890
        }
J
Ján Tomko 已提交
4891
    }
4892

4893
 success:
M
Michal Privoznik 已提交
4894 4895
    networkMacMgrDel(driver, network, dom->name, &iface->mac);

4896
    if (iface->data.network.actual) {
4897
        netdef->connections--;
4898 4899
        if (dev)
            dev->connections--;
4900 4901 4902
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
4903
        networkLogAllocation(netdef, actualType, dev, iface, false);
4904
    }
4905
    ret = 0;
4906
 cleanup:
4907
    virNetworkObjEndAPI(&network);
4908 4909 4910 4911
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4912
    return ret;
4913

4914
 error:
4915
    goto cleanup;
4916
}
4917 4918 4919 4920 4921 4922

/*
 * networkGetNetworkAddress:
 * @netname: the name of a network
 * @netaddr: string representation of IP address for that network.
 *
4923
 * Attempt to return an IP address associated with the named
4924 4925 4926 4927
 * 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.
 *
4928
 * Note: This function returns the first IP address it finds. It might
4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939
 * 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)
{
4940
    virNetworkDriverStatePtr driver = networkGetDriver();
4941
    int ret = -1;
4942
    virNetworkObjPtr network;
4943
    virNetworkDefPtr netdef;
4944
    virNetworkIPDefPtr ipdef;
4945 4946
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4947
    char *dev_name = NULL;
4948 4949

    *netaddr = NULL;
4950
    network = virNetworkObjFindByName(driver->networks, netname);
4951
    if (!network) {
4952 4953 4954
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4955
        goto cleanup;
4956 4957 4958
    }
    netdef = network->def;

4959
    switch (netdef->forward.type) {
4960 4961 4962
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
4963
    case VIR_NETWORK_FORWARD_OPEN:
4964
        ipdef = virNetworkDefGetIPByIndex(netdef, AF_UNSPEC, 0);
4965
        if (!ipdef) {
4966
            virReportError(VIR_ERR_INTERNAL_ERROR,
4967
                           _("network '%s' doesn't have an IP address"),
4968
                           netdef->name);
4969
            goto cleanup;
4970 4971 4972 4973 4974
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4975
        if ((dev_name = netdef->bridge))
4976 4977
            break;
        /*
4978 4979
         * fall through if netdef->bridge wasn't set, since that is
         * macvtap bridge mode network.
4980
         */
4981 4982
        ATTRIBUTE_FALLTHROUGH;

4983 4984 4985
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
4986 4987
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4988

4989
        if (!dev_name) {
4990 4991 4992
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4993
            goto cleanup;
4994 4995 4996 4997
        }
        break;
    }

4998
    if (dev_name) {
4999
        if (virNetDevIPAddrGet(dev_name, &addr) < 0)
5000
            goto cleanup;
5001
        addrptr = &addr;
5002 5003
    }

5004 5005
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
5006
        goto cleanup;
5007 5008
    }

5009
    ret = 0;
5010
 cleanup:
5011
    virNetworkObjEndAPI(&network);
5012 5013
    return ret;
}
5014

5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048
/* 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) ||
5049 5050
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_OPEN)) {
5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087
        /* 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;
}


5088 5089 5090
/**
 * networkCheckBandwidth:
 * @net: network QoS
5091
 * @ifaceBand: interface QoS (may be NULL if no QoS)
5092
 * @oldBandwidth: new interface QoS (may be NULL if no QoS)
5093
 * @ifaceMac: interface MAC (used in error messages for identification)
5094 5095
 * @new_rate: new rate for non guaranteed class
 *
5096 5097 5098 5099 5100 5101 5102 5103
 * 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.
 *
5104 5105 5106 5107 5108 5109
 * 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,
5110
                      virNetDevBandwidthPtr ifaceBand,
5111
                      virNetDevBandwidthPtr oldBandwidth,
5112
                      virMacAddr ifaceMac,
5113 5114 5115 5116 5117 5118 5119 5120
                      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];

5121
    virMacAddrFormat(&ifaceMac, ifmac);
5122 5123 5124 5125 5126 5127 5128 5129 5130 5131

    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;
    }

5132 5133
    if (((!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor) &&
         (!oldBandwidth || !oldBandwidth->in || !oldBandwidth->in->floor)) ||
5134 5135
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
5136
        return 1;
5137
    }
5138 5139

    tmp_new_rate = netBand->in->average;
5140 5141 5142 5143
    if (oldBandwidth && oldBandwidth->in)
        tmp_floor_sum -= oldBandwidth->in->floor;
    if (ifaceBand && ifaceBand->in)
        tmp_floor_sum += ifaceBand->in->floor;
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

    /* 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;
    }

5169 5170
    if (new_rate)
        *new_rate = tmp_new_rate;
5171 5172
    ret = 0;

5173
 cleanup:
5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189
    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)
{
5190
    ssize_t ret = 0;
5191

5192
    ret = virBitmapNextClearBit(net->class_id, -1);
5193

5194
    if (ret < 0 || virBitmapSetBit(net->class_id, ret) < 0)
5195 5196 5197 5198 5199
        return -1;

    return ret;
}

5200

5201
static int
5202 5203 5204 5205
networkPlugBandwidthImpl(virNetworkObjPtr net,
                         virDomainNetDefPtr iface,
                         virNetDevBandwidthPtr ifaceBand,
                         unsigned long long new_rate)
5206
{
5207
    virNetworkDriverStatePtr driver = networkGetDriver();
5208
    ssize_t class_id = 0;
5209 5210
    int plug_ret;
    int ret = -1;
5211 5212 5213 5214 5215 5216 5217 5218

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

5219 5220
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
5221 5222 5223 5224 5225 5226 5227 5228
    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 */
5229
    net->floor_sum += ifaceBand->in->floor;
5230
    /* update status file */
5231
    if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
5232
        ignore_value(virBitmapClearBit(net->class_id, class_id));
5233
        net->floor_sum -= ifaceBand->in->floor;
5234 5235 5236 5237
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
5238 5239
    /* update rate for non guaranteed NICs */
    new_rate -= net->floor_sum;
5240
    if (virNetDevBandwidthUpdateRate(net->def->bridge, 2,
5241 5242 5243 5244 5245
                                     net->def->bandwidth, new_rate) < 0)
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                 net->def->bridge);

    ret = 0;
5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285
 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;
5286

5287
 cleanup:
5288 5289 5290 5291 5292 5293 5294
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
5295
    virNetworkDriverStatePtr driver = networkGetDriver();
5296 5297
    int ret = 0;
    unsigned long long new_rate;
5298
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
5299 5300 5301

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
5302 5303 5304 5305 5306
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317
        /* 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 */
5318
        net->floor_sum -= ifaceBand->in->floor;
5319 5320 5321 5322
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
5323
        if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
5324
            net->floor_sum += ifaceBand->in->floor;
5325 5326 5327 5328
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
5329 5330
        /* update rate for non guaranteed NICs */
        new_rate -= net->floor_sum;
5331
        if (virNetDevBandwidthUpdateRate(net->def->bridge, 2,
5332 5333 5334 5335 5336 5337 5338
                                         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;
    }

5339
 cleanup:
5340 5341
    return ret;
}
5342 5343 5344

static void
networkNetworkObjTaint(virNetworkObjPtr net,
5345
                       virNetworkTaintFlags taint)
5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356
{
    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));
    }
}
5357 5358 5359 5360 5361 5362


static bool
networkBandwidthGenericChecks(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
5363
    virNetDevBandwidthPtr ifaceBand;
5364 5365 5366 5367 5368 5369 5370 5371
    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;
    }

5372
    ifaceBand = virDomainNetGetActualBandwidth(iface);
5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412
    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;
}
5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450


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: */

5451 5452
    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
        newBandwidth && newBandwidth->in && newBandwidth->in->floor) {
5453 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 5494
        /* Either we just need to update @floor .. */

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

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

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

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

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

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

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