bridge_driver.c 184.1 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2016 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

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

48
#include "virerror.h"
49
#include "datatypes.h"
50
#include "bridge_driver.h"
51
#include "bridge_driver_platform.h"
52
#include "device_conf.h"
53
#include "driver.h"
54
#include "virbuffer.h"
55
#include "virpidfile.h"
56
#include "vircommand.h"
57
#include "viralloc.h"
58
#include "viruuid.h"
59
#include "viriptables.h"
60
#include "virlog.h"
61
#include "virdnsmasq.h"
62
#include "configmake.h"
63
#include "virnetlink.h"
64
#include "virnetdev.h"
65
#include "virnetdevip.h"
66 67
#include "virnetdevbridge.h"
#include "virnetdevtap.h"
68
#include "virnetdevvportprofile.h"
69
#include "virpci.h"
70
#include "virdbus.h"
71
#include "virfile.h"
72
#include "virstring.h"
73
#include "viraccessapicheck.h"
74
#include "network_event.h"
75
#include "virhook.h"
76
#include "virjson.h"
77

78
#define VIR_FROM_THIS VIR_FROM_NETWORK
79
#define MAX_BRIDGE_ID 256
80

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

88 89
#define SYSCTL_PATH "/proc/sys"

90 91
VIR_LOG_INIT("network.bridge_driver");

92
static virNetworkDriverStatePtr network_driver;
93

94

95 96 97 98 99 100 101 102
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;
}
103

104 105 106

static void
networkDriverLock(virNetworkDriverStatePtr driver)
107
{
108
    virMutexLock(&driver->lock);
109
}
110 111 112 113


static void
networkDriverUnlock(virNetworkDriverStatePtr driver)
114
{
115
    virMutexUnlock(&driver->lock);
116 117
}

118

119 120 121 122
static dnsmasqCapsPtr
networkGetDnsmasqCaps(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr ret;
123
    networkDriverLock(driver);
124
    ret = virObjectRef(driver->dnsmasqCaps);
125
    networkDriverUnlock(driver);
126 127 128
    return ret;
}

129

130 131 132 133 134 135 136 137
static int
networkDnsmasqCapsRefresh(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr caps;

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

138
    networkDriverLock(driver);
139 140
    virObjectUnref(driver->dnsmasqCaps);
    driver->dnsmasqCaps = caps;
141
    networkDriverUnlock(driver);
142 143 144
    return 0;
}

145

146 147 148 149 150
static int
networkStateCleanup(void);

static int
networkStartNetwork(virNetworkDriverStatePtr driver,
151
                    virNetworkObjPtr obj);
152 153 154

static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
155
                       virNetworkObjPtr obj);
156 157 158

static int
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
159
                           virNetworkObjPtr obj);
160 161 162

static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
163
                              virNetworkObjPtr obj);
164

165
static int
166
networkStartNetworkExternal(virNetworkObjPtr obj);
167

168
static int
169
networkShutdownNetworkExternal(virNetworkObjPtr obj);
170

171 172
static void
networkReloadFirewallRules(virNetworkDriverStatePtr driver);
173

174 175
static void
networkRefreshDaemons(virNetworkDriverStatePtr driver);
176

177
static int
178
networkPlugBandwidth(virNetworkObjPtr obj,
179
                     virDomainNetDefPtr iface);
180

181
static int
182
networkUnplugBandwidth(virNetworkObjPtr obj,
183
                       virDomainNetDefPtr iface);
184

185
static void
186
networkNetworkObjTaint(virNetworkObjPtr obj,
187
                       virNetworkTaintFlags taint);
188

189

190 191 192
static virNetworkObjPtr
networkObjFromNetwork(virNetworkPtr net)
{
193
    virNetworkDriverStatePtr driver = networkGetDriver();
194
    virNetworkObjPtr obj;
195 196
    char uuidstr[VIR_UUID_STRING_BUFLEN];

197 198
    obj = virNetworkObjFindByUUID(driver->networks, net->uuid);
    if (!obj) {
199 200 201 202 203 204
        virUUIDFormat(net->uuid, uuidstr);
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching uuid '%s' (%s)"),
                       uuidstr, net->name);
    }

205
    return obj;
206 207
}

208

209
static int
210
networkRunHook(virNetworkObjPtr obj,
211
               virDomainDefPtr dom,
212
               virDomainNetDefPtr iface,
213 214 215
               int op,
               int sub_op)
{
216
    virNetworkDefPtr def;
217 218 219 220 221 222
    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)) {
223 224
        if (!obj) {
            VIR_DEBUG("Not running hook as @obj is NULL");
225 226 227
            ret = 0;
            goto cleanup;
        }
228
        def = virNetworkObjGetDef(obj);
229

230 231
        virBufferAddLit(&buf, "<hookData>\n");
        virBufferAdjustIndent(&buf, 2);
232
        if (iface && virDomainNetDefFormat(&buf, iface, NULL, 0) < 0)
233
            goto cleanup;
234
        if (virNetworkDefFormatBuf(&buf, def, 0) < 0)
235
            goto cleanup;
236
        if (dom && virDomainDefFormatInternal(dom, NULL, 0, &buf, NULL) < 0)
237 238 239 240 241
            goto cleanup;

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

242
        if (virBufferCheckError(&buf) < 0)
243 244
            goto cleanup;

245
        xml = virBufferContentAndReset(&buf);
246
        hookret = virHookCall(VIR_HOOK_DRIVER_NETWORK, def->name,
247 248 249 250 251 252 253
                              op, sub_op, NULL, xml, NULL);

        /*
         * If the script raised an error, pass it to the callee.
         */
        if (hookret < 0)
            goto cleanup;
254

255
        networkNetworkObjTaint(obj, VIR_NETWORK_TAINT_HOOK);
256 257 258
    }

    ret = 0;
259
 cleanup:
260 261 262 263 264 265 266
    virBufferFreeAndReset(&buf);
    VIR_FREE(xml);
    VIR_FREE(net_xml);
    VIR_FREE(dom_xml);
    return ret;
}

267

268
static char *
269 270
networkDnsmasqLeaseFileNameDefault(virNetworkDriverStatePtr driver,
                                   const char *netname)
271 272 273
{
    char *leasefile;

274
    ignore_value(virAsprintf(&leasefile, "%s/%s.leases",
275
                             driver->dnsmasqStateDir, netname));
276 277 278
    return leasefile;
}

279

280
static char *
281 282
networkDnsmasqLeaseFileNameCustom(virNetworkDriverStatePtr driver,
                                  const char *bridge)
283 284 285 286
{
    char *leasefile;

    ignore_value(virAsprintf(&leasefile, "%s/%s.status",
287
                             driver->dnsmasqStateDir, bridge));
288 289 290
    return leasefile;
}

291

292
static char *
293 294
networkDnsmasqConfigFileName(virNetworkDriverStatePtr driver,
                             const char *netname)
295 296 297
{
    char *conffile;

298
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
299
                             driver->dnsmasqStateDir, netname));
300 301 302
    return conffile;
}

303

304 305 306 307 308 309
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

310
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
311 312 313
    return pidfilebase;
}

314

315
static char *
316 317
networkRadvdConfigFileName(virNetworkDriverStatePtr driver,
                           const char *netname)
318 319 320
{
    char *configfile;

321
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
322
                             driver->radvdStateDir, netname));
323 324
    return configfile;
}
325

326

327 328
/* do needed cleanup steps and remove the network from the list */
static int
329
networkRemoveInactive(virNetworkDriverStatePtr driver,
330
                      virNetworkObjPtr obj)
331 332
{
    char *leasefile = NULL;
333
    char *customleasefile = NULL;
334
    char *radvdconfigfile = NULL;
335
    char *configfile = NULL;
336
    char *radvdpidbase = NULL;
337
    char *statusfile = NULL;
M
Michal Privoznik 已提交
338
    char *macMapFile = NULL;
339
    dnsmasqContext *dctx = NULL;
340
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(obj);
341 342 343 344

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
345
    if (!(dctx = dnsmasqContextNew(def->name,
346
                                   driver->dnsmasqStateDir))) {
347
        goto cleanup;
348
    }
349

350
    if (!(leasefile = networkDnsmasqLeaseFileNameDefault(driver, def->name)))
351 352
        goto cleanup;

353
    if (!(customleasefile = networkDnsmasqLeaseFileNameCustom(driver, def->bridge)))
354 355
        goto cleanup;

356
    if (!(radvdconfigfile = networkRadvdConfigFileName(driver, def->name)))
357
        goto cleanup;
358 359

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
360
        goto cleanup;
361

362
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
363
        goto cleanup;
364

365
    if (!(statusfile = virNetworkConfigFile(driver->stateDir, def->name)))
366
        goto cleanup;
367

368
    if (!(macMapFile = virMacMapFileName(driver->dnsmasqStateDir, def->bridge)))
M
Michal Privoznik 已提交
369 370
        goto cleanup;

371 372 373
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
374
    unlink(customleasefile);
375
    unlink(configfile);
376

M
Michal Privoznik 已提交
377 378 379
    /* MAC map manager */
    unlink(macMapFile);

380 381
    /* radvd */
    unlink(radvdconfigfile);
382
    virPidFileDelete(driver->pidDir, radvdpidbase);
383

384 385 386
    /* remove status file */
    unlink(statusfile);

387
    /* remove the network definition */
388
    virNetworkObjRemoveInactive(driver->networks, obj);
389 390 391

    ret = 0;

392
 cleanup:
393
    VIR_FREE(leasefile);
394
    VIR_FREE(configfile);
395
    VIR_FREE(customleasefile);
396 397
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
398
    VIR_FREE(statusfile);
M
Michal Privoznik 已提交
399
    VIR_FREE(macMapFile);
400 401 402 403
    dnsmasqContextFree(dctx);
    return ret;
}

404

405 406 407
static char *
networkBridgeDummyNicName(const char *brname)
{
408
    static const char dummyNicSuffix[] = "-nic";
409 410
    char *nicname;

411 412 413 414 415 416 417
    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.
         */
418 419 420 421 422
        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));
423
    } else {
424
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
425
    }
426 427 428
    return nicname;
}

429

430 431
static int
networkUpdateState(virNetworkObjPtr obj,
432
                   void *opaque)
433
{
434
    virNetworkDefPtr def;
435
    virNetworkDriverStatePtr driver = opaque;
436
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
437
    virMacMapPtr macmap;
438
    char *macMapFile = NULL;
439
    int ret = -1;
440

441
    virObjectLock(obj);
442
    if (!virNetworkObjIsActive(obj)) {
443 444
        ret = 0;
        goto cleanup;
445
    }
446
    def = virNetworkObjGetDef(obj);
447

448
    switch (def->forward.type) {
449 450 451
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
452
    case VIR_NETWORK_FORWARD_OPEN:
453
        /* If bridge doesn't exist, then mark it inactive */
454
        if (!(def->bridge && virNetDevExists(def->bridge) == 1))
455
            virNetworkObjSetActive(obj, false);
456

457
        if (!(macMapFile = virMacMapFileName(driver->dnsmasqStateDir,
458
                                             def->bridge)))
459 460
            goto cleanup;

461
        if (!(macmap = virMacMapNew(macMapFile)))
462 463
            goto cleanup;

464 465
        virNetworkObjSetMacMap(obj, macmap);

466
        break;
467

468
    case VIR_NETWORK_FORWARD_BRIDGE:
469 470
        if (def->bridge) {
            if (virNetDevExists(def->bridge) != 1)
471
                virNetworkObjSetActive(obj, false);
472
            break;
473
        }
474 475 476 477 478 479 480 481 482
        /* 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;
483

484 485 486
    case VIR_NETWORK_FORWARD_HOSTDEV:
        /* so far no extra checks */
        break;
487
    }
488

489
    /* Try and read dnsmasq/radvd pids of active networks */
490
    if (virNetworkObjIsActive(obj) && def->ips && (def->nips > 0)) {
491 492
        pid_t radvdPid;
        pid_t dnsmasqPid;
493
        char *radvdpidbase;
494

495
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
496
                                           def->name,
497
                                           &dnsmasqPid,
498
                                           dnsmasqCapsGetBinaryPath(dnsmasq_caps)));
499
        virNetworkObjSetDnsmasqPid(obj, dnsmasqPid);
M
Michal Privoznik 已提交
500

501
        radvdpidbase = networkRadvdPidfileBasename(def->name);
502 503
        if (!radvdpidbase)
            goto cleanup;
504

505 506
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           radvdpidbase,
507 508
                                           &radvdPid, RADVD));
        virNetworkObjSetRadvdPid(obj, radvdPid);
509
        VIR_FREE(radvdpidbase);
510
    }
511

512 513
    ret = 0;
 cleanup:
514
    virObjectUnlock(obj);
515
    virObjectUnref(dnsmasq_caps);
516
    VIR_FREE(macMapFile);
517 518
    return ret;
}
519

520

521
static int
522
networkAutostartConfig(virNetworkObjPtr obj,
523
                       void *opaque)
524
{
525
    virNetworkDriverStatePtr driver = opaque;
526
    int ret = -1;
527

528
    virObjectLock(obj);
529
    if (virNetworkObjIsAutostart(obj) &&
530 531
        !virNetworkObjIsActive(obj) &&
        networkStartNetwork(driver, obj) < 0)
532 533 534 535
        goto cleanup;

    ret = 0;
 cleanup:
536
    virObjectUnlock(obj);
537
    return ret;
538 539
}

540

541 542 543
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
544 545
                             DBusMessage *message,
                             void *user_data)
546
{
547 548
    virNetworkDriverStatePtr driver = user_data;

549 550 551 552 553 554
    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.");
555
        networkReloadFirewallRules(driver);
556 557 558 559 560 561
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

562

563
static int
564
networkMigrateStateFiles(virNetworkDriverStatePtr driver)
565 566 567 568 569 570 571 572 573 574 575 576 577
{
    /* 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;
578
    int direrr;
579 580 581
    struct dirent *entry;
    char *oldPath = NULL, *newPath = NULL;
    char *contents = NULL;
J
Ján Tomko 已提交
582
    int rc;
583

J
Ján Tomko 已提交
584 585
    if ((rc = virDirOpenIfExists(&dir, oldStateDir)) <= 0)
        return rc;
586 587 588 589 590 591 592

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

593
    while ((direrr = virDirRead(dir, &entry, oldStateDir)) > 0) {
594 595 596
        if (entry->d_type != DT_UNKNOWN &&
            entry->d_type != DT_REG)
            continue;
597 598 599

        if (virAsprintf(&oldPath, "%s/%s",
                        oldStateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
600
            goto cleanup;
601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617

        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;
            }
        }

618
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
J
Ján Tomko 已提交
619
            goto cleanup;
620 621 622

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

    ret = 0;
 cleanup:
J
Ján Tomko 已提交
641
    VIR_DIR_CLOSE(dir);
642 643 644 645 646 647
    VIR_FREE(oldPath);
    VIR_FREE(newPath);
    VIR_FREE(contents);
    return ret;
}

648

649
/**
650
 * networkStateInitialize:
651
 *
J
Ján Tomko 已提交
652
 * Initialization function for the QEMU daemon
653 654
 */
static int
655 656 657
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
658
{
659 660 661
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
662 663 664
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
665

666
    if (VIR_ALLOC(network_driver) < 0)
667
        goto error;
668

669 670
    if (virMutexInit(&network_driver->lock) < 0) {
        VIR_FREE(network_driver);
671 672
        goto error;
    }
673

674 675
    network_driver->privileged = privileged;

676 677 678 679
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
680
    if (privileged) {
681
        if (VIR_STRDUP(network_driver->networkConfigDir,
682
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
683
            VIR_STRDUP(network_driver->networkAutostartDir,
684
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
685
            VIR_STRDUP(network_driver->stateDir,
686
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
687
            VIR_STRDUP(network_driver->pidDir,
688
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
689
            VIR_STRDUP(network_driver->dnsmasqStateDir,
690
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
691
            VIR_STRDUP(network_driver->radvdStateDir,
692 693
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
694 695 696 697 698

        /* migration from old to new location is only applicable for
         * privileged mode - unprivileged mode directories haven't
         * changed location.
         */
699
        if (networkMigrateStateFiles(network_driver) < 0)
700
            goto error;
701
    } else {
702 703 704
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
705
            goto error;
706

707
        if ((virAsprintf(&network_driver->networkConfigDir,
708
                         "%s/qemu/networks", configdir) < 0) ||
709
            (virAsprintf(&network_driver->networkAutostartDir,
710
                         "%s/qemu/networks/autostart", configdir) < 0) ||
711
            (virAsprintf(&network_driver->stateDir,
712
                         "%s/network/lib", rundir) < 0) ||
713
            (virAsprintf(&network_driver->pidDir,
714
                         "%s/network/run", rundir) < 0) ||
715
            (virAsprintf(&network_driver->dnsmasqStateDir,
716
                         "%s/dnsmasq/lib", rundir) < 0) ||
717
            (virAsprintf(&network_driver->radvdStateDir,
718
                         "%s/radvd/lib", rundir) < 0)) {
719
            goto error;
720
        }
721 722
    }

723
    if (virFileMakePath(network_driver->stateDir) < 0) {
724 725
        virReportSystemError(errno,
                             _("cannot create directory %s"),
726
                             network_driver->stateDir);
727 728 729
        goto error;
    }

730
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
731
    network_driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
732

733
    if (!(network_driver->networks = virNetworkObjListNew()))
734 735
        goto error;

736 737
    if (virNetworkObjLoadAllState(network_driver->networks,
                                  network_driver->stateDir) < 0)
738 739
        goto error;

740 741 742
    if (virNetworkObjLoadAllConfigs(network_driver->networks,
                                    network_driver->networkConfigDir,
                                    network_driver->networkAutostartDir) < 0)
743 744
        goto error;

745 746 747 748
    /* 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). */
749
    virNetworkObjListForEach(network_driver->networks,
750
                             networkUpdateState,
751 752
                             network_driver);
    virNetworkObjListPrune(network_driver->networks,
753 754
                           VIR_CONNECT_LIST_NETWORKS_INACTIVE |
                           VIR_CONNECT_LIST_NETWORKS_TRANSIENT);
755 756
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
757

758
    network_driver->networkEventState = virObjectEventStateNew();
759

760 761 762
#ifdef HAVE_FIREWALLD
    if (!(sysbus = virDBusGetSystemBus())) {
        VIR_WARN("DBus not available, disabling firewalld support "
763
                 "in bridge_network_driver: %s", virGetLastErrorMessage());
764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780
    } 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,
781
                                   network_driver, NULL);
782 783 784
    }
#endif

785
    ret = 0;
786
 cleanup:
787 788 789
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
790

791
 error:
792
    networkStateCleanup();
793
    goto cleanup;
794 795
}

796

797 798 799 800 801 802 803 804
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
805
    if (!network_driver)
806 807
        return;

808
    virNetworkObjListForEach(network_driver->networks,
809
                             networkAutostartConfig,
810
                             network_driver);
811 812
}

813

814
/**
815
 * networkStateReload:
816
 *
J
Ján Tomko 已提交
817
 * Function to restart the QEMU daemon, it will recheck the configuration
818 819 820
 * files and update its state and the networking
 */
static int
821 822
networkStateReload(void)
{
823
    if (!network_driver)
824 825
        return 0;

826 827 828 829 830
    virNetworkObjLoadAllState(network_driver->networks,
                              network_driver->stateDir);
    virNetworkObjLoadAllConfigs(network_driver->networks,
                                network_driver->networkConfigDir,
                                network_driver->networkAutostartDir);
831 832 833
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
    virNetworkObjListForEach(network_driver->networks,
834
                             networkAutostartConfig,
835
                             network_driver);
836 837 838 839 840
    return 0;
}


/**
841
 * networkStateCleanup:
842
 *
J
Ján Tomko 已提交
843
 * Shutdown the QEMU daemon, it will stop all active domains and networks
844 845
 */
static int
846 847
networkStateCleanup(void)
{
848
    if (!network_driver)
849 850
        return -1;

851
    virObjectUnref(network_driver->networkEventState);
852

853
    /* free inactive networks */
854
    virObjectUnref(network_driver->networks);
855

856 857 858 859 860 861
    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);
862

863
    virObjectUnref(network_driver->dnsmasqCaps);
864

865
    virMutexDestroy(&network_driver->lock);
866

867
    VIR_FREE(network_driver);
868 869 870 871 872

    return 0;
}


873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947
static virDrvOpenStatus
networkConnectOpen(virConnectPtr conn,
                   virConnectAuthPtr auth ATTRIBUTE_UNUSED,
                   virConfPtr conf ATTRIBUTE_UNUSED,
                   unsigned int flags)
{
    virCheckFlags(VIR_CONNECT_RO, VIR_DRV_OPEN_ERROR);

    /* Verify uri was specified */
    if (conn->uri == NULL) {
        /* Only hypervisor drivers are permitted to auto-open on NULL uri */
        return VIR_DRV_OPEN_DECLINED;
    } else {
        if (STRNEQ_NULLABLE(conn->uri->scheme, "network"))
            return VIR_DRV_OPEN_DECLINED;

        /* Leave for remote driver */
        if (conn->uri->server != NULL)
            return VIR_DRV_OPEN_DECLINED;

        if (network_driver == NULL) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("network state driver is not active"));
            return VIR_DRV_OPEN_ERROR;
        }

        if (network_driver->privileged) {
            if (STRNEQ(conn->uri->path, "/system")) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("unexpected network URI path '%s', try network:///system"),
                               conn->uri->path);
                return VIR_DRV_OPEN_ERROR;
            }
        } else {
            if (STRNEQ(conn->uri->path, "/session")) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("unexpected network URI path '%s', try network:///session"),
                               conn->uri->path);
                return VIR_DRV_OPEN_ERROR;
            }
        }
    }

    if (virConnectOpenEnsureACL(conn) < 0)
        return VIR_DRV_OPEN_ERROR;

    return VIR_DRV_OPEN_SUCCESS;
}

static int networkConnectClose(virConnectPtr conn ATTRIBUTE_UNUSED)
{
    return 0;
}


static int networkConnectIsSecure(virConnectPtr conn ATTRIBUTE_UNUSED)
{
    /* Trivially secure, since always inside the daemon */
    return 1;
}


static int networkConnectIsEncrypted(virConnectPtr conn ATTRIBUTE_UNUSED)
{
    /* Not encrypted, but remote driver takes care of that */
    return 0;
}


static int networkConnectIsAlive(virConnectPtr conn ATTRIBUTE_UNUSED)
{
    return 1;
}


948 949 950 951 952
/* networkKillDaemon:
 *
 * kill the specified pid/name, and wait a bit to make sure it's dead.
 */
static int
953 954 955
networkKillDaemon(pid_t pid,
                  const char *daemonName,
                  const char *networkName)
956
{
957 958
    size_t i;
    int ret = -1;
959 960 961 962 963 964 965
    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.
     */
966
    for (i = 0; i < 25; i++) {
967
        int signum = 0;
968
        if (i == 0) {
969
            signum = SIGTERM;
970
        } else if (i == 15) {
971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005
            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);
1006
 cleanup:
1007 1008 1009
    return ret;
}

1010

J
Ján Tomko 已提交
1011 1012 1013
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
1014
static int
G
Gene Czarcinski 已提交
1015
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
1016
                                 virNetworkIPDefPtr ipdef)
1017
{
1018
    size_t i;
G
Gene Czarcinski 已提交
1019
    bool ipv6 = false;
1020

G
Gene Czarcinski 已提交
1021 1022
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
1023 1024
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
1025
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
1026 1027
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
1028
                return -1;
1029
    }
1030

G
Gene Czarcinski 已提交
1031 1032 1033
    return 0;
}

1034

G
Gene Czarcinski 已提交
1035 1036 1037 1038
static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
1039
    size_t i, j;
G
Gene Czarcinski 已提交
1040

1041 1042
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
1043
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
1044
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
1045
                for (j = 0; j < host->nnames; j++)
1046 1047
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
1048 1049
            }
        }
1050 1051
    }

1052
    return 0;
1053 1054 1055
}


1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092
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;
}


1093
int
1094
networkDnsmasqConfContents(virNetworkObjPtr obj,
1095 1096 1097 1098
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
1099
{
1100
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1101
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1102
    int r, ret = -1;
1103
    int nbleases = 0;
1104
    size_t i;
1105
    virNetworkDNSDefPtr dns = &def->dns;
1106
    bool wantDNS = dns->enable != VIR_TRISTATE_BOOL_NO;
1107
    virNetworkIPDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
G
Gene Czarcinski 已提交
1108
    bool ipv6SLAAC;
1109
    char *saddr = NULL, *eaddr = NULL;
1110

1111 1112
    *configstr = NULL;

1113
    /*
1114 1115 1116
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
1117
     *
1118 1119
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
1120
     */
1121 1122 1123 1124 1125 1126

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

1127
    /* create dnsmasq config file appropriate for this network */
1128 1129

    /* Don't forget to update networkxml2conftest :-) */
1130
    virBufferAsprintf(&configbuf,
1131 1132 1133 1134 1135 1136 1137
                      "##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"
1138
                      "strict-order\n",
1139
                      def->name);
1140

1141 1142 1143 1144 1145 1146
    /* if dns is disabled, set its listening port to 0, which
     * tells dnsmasq to not listen
     */
    if (!wantDNS)
        virBufferAddLit(&configbuf, "port=0\n");

1147
    if (wantDNS && def->dns.forwarders) {
1148 1149 1150 1151 1152 1153 1154 1155 1156
        /* addNoResolv should be set to true if there are any entries
         * that specify an IP address for requests, but no domain
         * qualifier (implying that all requests otherwise "unclaimed"
         * should be sent to that address). if it is still false when
         * we've looked at all entries, it means we still need the
         * host's resolv.conf for some cases.
         */
        bool addNoResolv = false;

1157 1158
        for (i = 0; i < def->dns.nfwds; i++) {
            virNetworkDNSForwarderPtr fwd = &def->dns.forwarders[i];
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168

            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);
1169
                VIR_FREE(addr);
1170 1171
                if (!fwd->domain)
                    addNoResolv = true;
1172 1173 1174 1175
            } else {
                /* "don't forward requests for this domain" */
                virBufferAddLit(&configbuf, "#\n");
            }
1176
        }
1177 1178
        if (addNoResolv)
            virBufferAddLit(&configbuf, "no-resolv\n");
1179 1180
    }

1181 1182
    if (def->domain) {
        if (def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
1183 1184
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
1185
                              def->domain);
1186
        }
1187
        virBufferAsprintf(&configbuf,
1188 1189
                          "domain=%s\n"
                          "expand-hosts\n",
1190
                          def->domain);
1191
    }
1192

1193
    if (wantDNS &&
1194
        networkDnsmasqConfLocalPTRs(&configbuf, def) < 0)
1195 1196
        goto cleanup;

1197
    if (wantDNS && def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
1198 1199 1200 1201
        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)
1202
         */
1203
        virBufferAddLit(&configbuf, "local=//\n");
1204
    }
1205

1206
    if (pidfile)
1207
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
1208

1209
    /* dnsmasq will *always* listen on localhost unless told otherwise */
P
Pavel Timofeev 已提交
1210
#ifdef __linux__
1211
    virBufferAddLit(&configbuf, "except-interface=lo\n");
P
Pavel Timofeev 已提交
1212 1213 1214 1215
#else
    /* BSD family OSes and Solaris call loopback interface as lo0 */
    virBufferAddLit(&configbuf, "except-interface=lo0\n");
#endif
1216

1217 1218 1219 1220 1221 1222 1223 1224
    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.
         */
1225
        virBufferAsprintf(&configbuf,
1226 1227
                          "bind-dynamic\n"
                          "interface=%s\n",
1228
                          def->bridge);
1229
    } else {
1230
        virBufferAddLit(&configbuf, "bind-interfaces\n");
1231 1232 1233 1234
        /*
         * --interface does not actually work with dnsmasq < 2.47,
         * due to DAD for ipv6 addresses on the interface.
         *
1235
         * virCommandAddArgList(cmd, "--interface", def->bridge, NULL);
1236 1237 1238
         *
         * So listen on all defined IPv[46] addresses
         */
1239
        for (i = 0;
1240
             (tmpipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
1241
             i++) {
1242 1243 1244 1245
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
1246

1247
            /* also part of CVE 2012-3411 - if the host's version of
1248
             * dnsmasq doesn't have bind-dynamic, only allow listening on
1249 1250
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
1251 1252
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
1253 1254 1255 1256
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
1257
                                 "The version of dnsmasq on this host (%d.%d) "
1258 1259 1260 1261
                                 "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 "
1262 1263 1264 1265 1266 1267
                                 "(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);
1268
                VIR_FREE(ipaddr);
1269 1270
                goto cleanup;
            }
1271
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1272 1273 1274
            VIR_FREE(ipaddr);
        }
    }
1275

1276 1277
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1278
     * guaranteed to not work, and set no-resolv so that no dns
1279 1280 1281
     * 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).
1282 1283 1284
     * 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.
1285
     */
1286
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1287
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1288
                        "no-resolv\n");
1289 1290 1291 1292
        if (dnsmasqCapsGet(caps, DNSMASQ_CAPS_RA_PARAM)) {
            /* interface=* (any), interval=0 (default), lifetime=0 (seconds) */
            virBufferAddLit(&configbuf, "ra-param=*,0,0\n");
        }
1293
    }
1294

1295 1296 1297 1298 1299
    if (wantDNS) {
        for (i = 0; i < dns->ntxts; i++) {
            virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
                              dns->txts[i].name,
                              dns->txts[i].value);
1300
        }
1301

1302 1303 1304 1305 1306 1307 1308 1309
        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'"),
1310
                               def->name);
1311 1312 1313 1314 1315 1316
                goto cleanup;
            }
            if (!dns->srvs[i].protocol) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Missing required 'service' "
                                 "attribute in SRV record of network '%s'"),
1317
                               def->name);
1318 1319 1320 1321 1322 1323 1324
                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);
1325

1326 1327 1328 1329 1330 1331 1332
            /* 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.
1333
             */
1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353
            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");
1354
        }
1355 1356
    }

G
Gene Czarcinski 已提交
1357
    /* Find the first dhcp for both IPv4 and IPv6 */
1358
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
1359
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
1360
         i++) {
G
Gene Czarcinski 已提交
1361 1362 1363 1364
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1365 1366
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377
                    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,
1378 1379 1380 1381 1382 1383 1384 1385 1386
                                   _("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
1387 1388 1389 1390
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1391 1392
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405
                    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 "
1406 1407 1408 1409
                 "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
1410 1411 1412 1413 1414
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1415 1416
        int prefix;

1417
        prefix = virNetworkIPDefPrefix(ipdef);
1418 1419 1420
        if (prefix < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
1421
                           def->bridge);
1422 1423
            goto cleanup;
        }
1424
        for (r = 0; r < ipdef->nranges; r++) {
1425 1426
            int thisRange;

1427 1428
            if (!(saddr = virSocketAddrFormat(&ipdef->ranges[r].start)) ||
                !(eaddr = virSocketAddrFormat(&ipdef->ranges[r].end)))
1429
                goto cleanup;
1430

1431
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s",
1432
                              saddr, eaddr);
1433
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1434
                virBufferAsprintf(&configbuf, ",%d", prefix);
1435 1436
            virBufferAddLit(&configbuf, "\n");

1437
            VIR_FREE(saddr);
1438
            VIR_FREE(eaddr);
1439
            thisRange = virSocketAddrGetRange(&ipdef->ranges[r].start,
1440 1441
                                              &ipdef->ranges[r].end,
                                              &ipdef->address,
1442
                                              virNetworkIPDefPrefix(ipdef));
1443 1444 1445
            if (thisRange < 0)
                goto cleanup;
            nbleases += thisRange;
1446
        }
1447

1448
        /*
1449 1450 1451 1452
         * 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)
1453 1454
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1455
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1456 1457
            if (!bridgeaddr)
                goto cleanup;
1458 1459 1460
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static",
                              bridgeaddr);
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
1461
                virBufferAsprintf(&configbuf, ",%d", prefix);
1462
            virBufferAddLit(&configbuf, "\n");
1463 1464
            VIR_FREE(bridgeaddr);
        }
1465

G
Gene Czarcinski 已提交
1466 1467
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1468

G
Gene Czarcinski 已提交
1469 1470
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
1471
            if (ipdef->nranges || ipdef->nhosts) {
1472
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1473 1474
                virBufferAddLit(&configbuf, "dhcp-authoritative\n");
            }
1475

G
Gene Czarcinski 已提交
1476
            if (ipdef->tftproot) {
1477 1478
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1479
            }
1480

G
Gene Czarcinski 已提交
1481 1482 1483
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1484

1485
                    if (!bootserver)
G
Gene Czarcinski 已提交
1486
                        goto cleanup;
1487
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1488
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1489 1490
                    VIR_FREE(bootserver);
                } else {
1491
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1492 1493 1494 1495 1496
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1497

1498
    if (nbleases > 0)
1499
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1500

G
Gene Czarcinski 已提交
1501 1502
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1503
        goto cleanup;
G
Gene Czarcinski 已提交
1504 1505 1506 1507 1508 1509

    /* 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)
1510 1511
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1512

1513 1514
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1515
     */
1516 1517 1518 1519
    if (wantDNS) {
        virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
                          dctx->addnhostsfile->path);
    }
G
Gene Czarcinski 已提交
1520 1521 1522

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1523
        if (ipv6def) {
1524
            virBufferAddLit(&configbuf, "enable-ra\n");
1525
        } else {
1526
            for (i = 0;
1527
                 (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i));
1528
                 i++) {
G
Gene Czarcinski 已提交
1529 1530 1531 1532
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1533 1534
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1535 1536
                    VIR_FREE(bridgeaddr);
                }
1537
            }
1538
        }
1539 1540
    }

1541 1542 1543
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1544
    ret = 0;
G
Gene Czarcinski 已提交
1545

1546
 cleanup:
1547 1548
    VIR_FREE(saddr);
    VIR_FREE(eaddr);
1549
    virBufferFreeAndReset(&configbuf);
1550
    return ret;
1551 1552
}

1553

1554
/* build the dnsmasq command line */
1555 1556
static int ATTRIBUTE_NONNULL(3)
networkBuildDhcpDaemonCommandLine(virNetworkDriverStatePtr driver,
1557
                                  virNetworkObjPtr obj,
1558
                                  virCommandPtr *cmdout,
1559 1560
                                  char *pidfile,
                                  dnsmasqContext *dctx)
1561
{
1562
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1563
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1564
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1565
    int ret = -1;
1566 1567
    char *configfile = NULL;
    char *configstr = NULL;
1568
    char *leaseshelper_path = NULL;
1569

1570
    virNetworkObjSetDnsmasqPid(obj, -1);
1571

1572
    if (networkDnsmasqConfContents(obj, pidfile, &configstr,
1573
                                   dctx, dnsmasq_caps) < 0)
1574 1575 1576 1577 1578
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1579
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
1580 1581 1582 1583 1584
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1585 1586
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1587 1588 1589
        goto cleanup;
    }

1590 1591
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1592
                                                  abs_topbuilddir "/src",
1593 1594 1595
                                                  LIBEXECDIR)))
        goto cleanup;

1596
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(dnsmasq_caps));
1597
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1598 1599
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1600
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1601
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", def->bridge);
1602

1603
    *cmdout = cmd;
1604
    ret = 0;
1605
 cleanup:
1606
    virObjectUnref(dnsmasq_caps);
1607 1608
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1609
    VIR_FREE(leaseshelper_path);
1610 1611 1612
    return ret;
}

1613

1614
static int
1615
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
1616
                       virNetworkObjPtr obj)
1617
{
1618
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1619 1620 1621
    virNetworkIPDefPtr ipdef;
    size_t i;
    bool needDnsmasq = false;
1622 1623
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
1624
    pid_t dnsmasqPid;
1625
    int ret = -1;
1626
    dnsmasqContext *dctx = NULL;
1627

1628
    /* see if there are any IP addresses that need a dhcp server */
1629
    i = 0;
1630
    while ((ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i))) {
1631
        i++;
1632 1633 1634 1635
        if (ipdef->nranges || ipdef->nhosts)
            needDnsmasq = true;
    }

1636 1637 1638 1639 1640 1641
    if (i == 0) {
        /* no IP addresses at all, so we don't need to run */
        ret = 0;
        goto cleanup;
    }

1642
    if (!needDnsmasq && def->dns.enable == VIR_TRISTATE_BOOL_NO) {
1643
        /* no DHCP services needed, and user disabled DNS service */
1644 1645 1646 1647
        ret = 0;
        goto cleanup;
    }

1648
    if (virFileMakePath(driver->pidDir) < 0) {
1649
        virReportSystemError(errno,
1650
                             _("cannot create directory %s"),
1651
                             driver->pidDir);
1652
        goto cleanup;
1653 1654
    }

1655
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, def->name)))
1656
        goto cleanup;
1657

1658
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1659
        virReportSystemError(errno,
1660
                             _("cannot create directory %s"),
1661
                             driver->dnsmasqStateDir);
1662 1663 1664
        goto cleanup;
    }

1665
    dctx = dnsmasqContextNew(def->name, driver->dnsmasqStateDir);
1666 1667 1668
    if (dctx == NULL)
        goto cleanup;

1669
    if (networkDnsmasqCapsRefresh(driver) < 0)
1670
        goto cleanup;
1671

1672
    ret = networkBuildDhcpDaemonCommandLine(driver, obj, &cmd, pidfile, dctx);
1673 1674 1675 1676 1677
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1678
        goto cleanup;
1679

G
Guido Günther 已提交
1680
    ret = virCommandRun(cmd, NULL);
1681
    if (ret < 0)
1682 1683 1684
        goto cleanup;

    /*
1685 1686 1687 1688 1689
     * 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
1690 1691
     */

1692
    ret = virPidFileRead(driver->pidDir, def->name, &dnsmasqPid);
1693
    if (ret < 0)
1694
        goto cleanup;
1695
    virNetworkObjSetDnsmasqPid(obj, dnsmasqPid);
1696

1697
    ret = 0;
1698
 cleanup:
1699
    VIR_FREE(pidfile);
1700
    virCommandFree(cmd);
1701
    dnsmasqContextFree(dctx);
1702 1703 1704
    return ret;
}

1705

1706 1707
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1708 1709
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1710 1711 1712
 *
 *  Returns 0 on success, -1 on failure.
 */
1713
static int
1714
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
1715
                         virNetworkObjPtr obj)
1716
{
1717
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1718 1719
    int ret = -1;
    size_t i;
1720
    pid_t dnsmasqPid;
1721
    virNetworkIPDefPtr ipdef, ipv4def, ipv6def;
1722
    dnsmasqContext *dctx = NULL;
1723

G
Gene Czarcinski 已提交
1724
    /* if no IP addresses specified, nothing to do */
1725
    if (!virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1726 1727
        return 0;

1728
    /* if there's no running dnsmasq, just start it */
1729 1730
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid <= 0 || (kill(dnsmasqPid, 0) < 0))
1731
        return networkStartDhcpDaemon(driver, obj);
1732

1733 1734
    VIR_INFO("Refreshing dnsmasq for network %s", def->bridge);
    if (!(dctx = dnsmasqContextNew(def->name,
1735
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1736
        goto cleanup;
1737
    }
G
Gene Czarcinski 已提交
1738 1739 1740 1741 1742 1743

    /* 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;
1744
    for (i = 0;
1745
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
1746
         i++) {
G
Gene Czarcinski 已提交
1747 1748
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1749 1750
    }

G
Gene Czarcinski 已提交
1751
    ipv6def = NULL;
1752
    for (i = 0;
1753
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i));
1754
         i++) {
G
Gene Czarcinski 已提交
1755 1756
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1757 1758
    }

G
Gene Czarcinski 已提交
1759
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1760
        goto cleanup;
G
Gene Czarcinski 已提交
1761 1762

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1763
        goto cleanup;
1764

1765
    if (networkBuildDnsmasqHostsList(dctx, &def->dns) < 0)
J
Ján Tomko 已提交
1766
        goto cleanup;
1767 1768

    if ((ret = dnsmasqSave(dctx)) < 0)
1769
        goto cleanup;
1770

1771 1772
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    ret = kill(dnsmasqPid, SIGHUP);
1773
 cleanup:
1774 1775 1776 1777
    dnsmasqContextFree(dctx);
    return ret;
}

1778

1779 1780 1781 1782 1783 1784 1785 1786
/* 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
1787
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
1788
                         virNetworkObjPtr obj)
1789
{
1790
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1791 1792
    pid_t dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);

1793
    /* if there is a running dnsmasq, kill it */
1794
    if (dnsmasqPid > 0) {
1795
        networkKillDaemon(dnsmasqPid, "dnsmasq", def->name);
1796
        virNetworkObjSetDnsmasqPid(obj, -1);
1797
    }
1798
    /* now start dnsmasq if it should be started */
1799
    return networkStartDhcpDaemon(driver, obj);
1800 1801
}

1802

G
Gene Czarcinski 已提交
1803 1804 1805 1806 1807 1808
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";

1809
static int
1810
networkRadvdConfContents(virNetworkObjPtr obj,
1811
                         char **configstr)
1812
{
1813
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
E
Eric Blake 已提交
1814
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1815 1816
    int ret = -1;
    size_t i;
1817
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
1818
    bool v6present = false, dhcp6 = false;
1819 1820

    *configstr = NULL;
1821

G
Gene Czarcinski 已提交
1822
    /* Check if DHCPv6 is needed */
1823
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i)); i++) {
G
Gene Czarcinski 已提交
1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836
        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;
    }

1837 1838 1839
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1840
    virBufferAsprintf(&configbuf, "interface %s\n"
1841 1842
                      "{\n"
                      "  AdvSendAdvert on;\n"
1843
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1844 1845
                      "  AdvManagedFlag %s;\n"
                      "%s",
1846
                      def->bridge,
G
Gene Czarcinski 已提交
1847 1848
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1849 1850

    /* add a section for each IPv6 address in the config */
1851
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, i)); i++) {
1852 1853 1854
        int prefix;
        char *netaddr;

1855
        prefix = virNetworkIPDefPrefix(ipdef);
1856
        if (prefix < 0) {
1857 1858
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
1859
                           def->bridge);
1860 1861
            goto cleanup;
        }
1862
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1863
            goto cleanup;
1864
        virBufferAsprintf(&configbuf,
1865
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1866 1867 1868
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1869 1870 1871
        VIR_FREE(netaddr);
    }

1872
    virBufferAddLit(&configbuf, "};\n");
1873

1874
    if (virBufferCheckError(&configbuf) < 0)
1875
        goto cleanup;
1876

1877 1878
    *configstr = virBufferContentAndReset(&configbuf);

1879
    ret = 0;
1880
 cleanup:
1881 1882 1883 1884
    virBufferFreeAndReset(&configbuf);
    return ret;
}

1885

1886
/* write file and return its name (which must be freed by caller) */
1887
static int
1888
networkRadvdConfWrite(virNetworkDriverStatePtr driver,
1889
                      virNetworkObjPtr obj,
1890
                      char **configFile)
1891
{
1892
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1893 1894 1895 1896 1897 1898 1899 1900 1901
    int ret = -1;
    char *configStr = NULL;
    char *myConfigFile = NULL;

    if (!configFile)
        configFile = &myConfigFile;

    *configFile = NULL;

1902
    if (networkRadvdConfContents(obj, &configStr) < 0)
1903 1904 1905 1906
        goto cleanup;

    if (!configStr) {
        ret = 0;
1907 1908 1909 1910
        goto cleanup;
    }

    /* construct the filename */
1911
    if (!(*configFile = networkRadvdConfigFileName(driver, def->name)))
1912 1913
        goto cleanup;
    /* write the file */
1914
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1915 1916
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1917 1918 1919 1920 1921
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1922
 cleanup:
1923 1924 1925 1926 1927
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

1928

1929
static int
1930
networkStartRadvd(virNetworkDriverStatePtr driver,
1931
                  virNetworkObjPtr obj)
1932
{
1933
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
1934
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1935
    pid_t radvdPid;
1936 1937 1938 1939 1940 1941
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

1942
    virNetworkObjSetRadvdPid(obj, -1);
1943

G
Gene Czarcinski 已提交
1944
    /* Is dnsmasq handling RA? */
1945
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
G
Gene Czarcinski 已提交
1946 1947 1948 1949
        ret = 0;
        goto cleanup;
    }

1950
    if (!virNetworkDefGetIPByIndex(def, AF_INET6, 0)) {
1951 1952 1953 1954 1955 1956 1957 1958 1959 1960
        /* 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);
1961 1962 1963
        goto cleanup;
    }

1964
    if (virFileMakePath(driver->pidDir) < 0) {
1965 1966
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1967
                             driver->pidDir);
1968 1969
        goto cleanup;
    }
1970
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1971 1972
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1973
                             driver->radvdStateDir);
1974 1975 1976 1977
        goto cleanup;
    }

    /* construct pidfile name */
1978
    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
1979
        goto cleanup;
1980
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1981 1982
        goto cleanup;

1983
    if (networkRadvdConfWrite(driver, obj, &configfile) < 0)
1984 1985
        goto cleanup;

1986 1987 1988 1989
    /* 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
1990
     * virPidFileRead() below will fail if we use them).
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
     * 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;

2006
    if (virPidFileRead(driver->pidDir, radvdpidbase, &radvdPid) < 0)
2007
        goto cleanup;
2008
    virNetworkObjSetRadvdPid(obj, radvdPid);
2009 2010

    ret = 0;
2011
 cleanup:
2012
    virObjectUnref(dnsmasq_caps);
2013 2014 2015 2016 2017 2018 2019
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

2020

2021
static int
2022
networkRefreshRadvd(virNetworkDriverStatePtr driver,
2023
                    virNetworkObjPtr obj)
2024
{
2025
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2026
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
G
Gene Czarcinski 已提交
2027
    char *radvdpidbase;
2028
    pid_t radvdPid;
G
Gene Czarcinski 已提交
2029 2030

    /* Is dnsmasq handling RA? */
2031 2032
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
        virObjectUnref(dnsmasq_caps);
2033 2034
        radvdPid = virNetworkObjGetRadvdPid(obj);
        if (radvdPid <= 0)
G
Gene Czarcinski 已提交
2035 2036
            return 0;
        /* radvd should not be running but in case it is */
2037 2038
        if ((networkKillDaemon(radvdPid, "radvd", def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(def->name))
G
Gene Czarcinski 已提交
2039
             != NULL)) {
2040
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
2041 2042
            VIR_FREE(radvdpidbase);
        }
2043
        virNetworkObjSetRadvdPid(obj, -1);
G
Gene Czarcinski 已提交
2044 2045
        return 0;
    }
2046
    virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
2047

2048
    /* if there's no running radvd, just start it */
2049 2050
    radvdPid = virNetworkObjGetRadvdPid(obj);
    if (radvdPid <= 0 || (kill(radvdPid, 0) < 0))
2051
        return networkStartRadvd(driver, obj);
2052

2053
    if (!virNetworkDefGetIPByIndex(def, AF_INET6, 0)) {
2054 2055 2056 2057
        /* no IPv6 addresses, so we don't need to run radvd */
        return 0;
    }

2058
    if (networkRadvdConfWrite(driver, obj, NULL) < 0)
2059 2060
        return -1;

2061
    return kill(radvdPid, SIGHUP);
2062 2063
}

2064

2065 2066
#if 0
/* currently unused, so it causes a build error unless we #if it out */
2067
static int
2068
networkRestartRadvd(virNetworkObjPtr obj)
2069
{
2070
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2071
    char *radvdpidbase;
2072
    pid_t radvdPid = virNeworkObjGetRadvdPid(obj);
2073 2074

    /* if there is a running radvd, kill it */
2075
    if (radvdPid > 0) {
2076 2077 2078 2079
        /* 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).
         */
2080 2081
        if ((networkKillDaemon(radvdPid, "radvd", def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(def->name))
2082
             != NULL)) {
2083
            virPidFileDelete(driver->pidDir, radvdpidbase);
2084 2085
            VIR_FREE(radvdpidbase);
        }
2086
        virNetworkObjSetRadvdPid(obj, -1);
2087 2088
    }
    /* now start radvd if it should be started */
2089
    return networkStartRadvd(obj);
2090 2091 2092
}
#endif /* #if 0 */

2093

2094
static int
2095
networkRefreshDaemonsHelper(virNetworkObjPtr obj,
2096
                            void *opaque)
2097
{
2098
    virNetworkDefPtr def;
2099
    virNetworkDriverStatePtr driver = opaque;
2100

2101
    virObjectLock(obj);
2102
    def = virNetworkObjGetDef(obj);
2103
    if (virNetworkObjIsActive(obj) &&
2104 2105 2106 2107
        ((def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
         (def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
         (def->forward.type == VIR_NETWORK_FORWARD_ROUTE) ||
         (def->forward.type == VIR_NETWORK_FORWARD_OPEN))) {
2108 2109 2110 2111 2112 2113
        /* 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.
         */
2114 2115
        networkRefreshDhcpDaemon(driver, obj);
        networkRefreshRadvd(driver, obj);
2116
    }
2117
    virObjectUnlock(obj);
2118 2119 2120
    return 0;
}

2121

2122 2123 2124 2125
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
2126
networkRefreshDaemons(virNetworkDriverStatePtr driver)
2127 2128
{
    VIR_INFO("Refreshing network daemons");
2129 2130
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
2131
                             driver);
2132
}
2133

2134

2135
static int
2136
networkReloadFirewallRulesHelper(virNetworkObjPtr obj,
2137 2138
                                 void *opaque ATTRIBUTE_UNUSED)
{
2139
    virNetworkDefPtr def;
2140

2141
    virObjectLock(obj);
2142
    def = virNetworkObjGetDef(obj);
2143
    if (virNetworkObjIsActive(obj) &&
2144 2145 2146
        ((def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
         (def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
         (def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
2147 2148 2149 2150
        /* 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.
2151
         */
2152 2153
        networkRemoveFirewallRules(def);
        if (networkAddFirewallRules(def) < 0) {
2154
            /* failed to add but already logged */
2155 2156
        }
    }
2157
    virObjectUnlock(obj);
2158
    return 0;
2159 2160
}

2161

2162
static void
2163
networkReloadFirewallRules(virNetworkDriverStatePtr driver)
2164
{
2165
    VIR_INFO("Reloading iptables rules");
2166 2167 2168
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
2169 2170
}

2171

2172
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
2173
static int
2174 2175
networkEnableIPForwarding(bool enableIPv4,
                          bool enableIPv6)
2176
{
2177
    int ret = 0;
2178 2179 2180 2181
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
2182
                           &enabled, sizeof(enabled));
2183 2184
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
2185
                           &enabled, sizeof(enabled));
2186
#else
2187
    if (enableIPv4)
2188
        ret = virFileWriteStr(SYSCTL_PATH "/net/ipv4/ip_forward", "1\n", 0);
2189
    if (enableIPv6 && ret == 0)
2190 2191
        ret = virFileWriteStr(SYSCTL_PATH "/net/ipv6/conf/all/forwarding", "1\n", 0);

2192
#endif
2193
    return ret;
2194 2195
}

2196

2197
static int
2198
networkSetIPv6Sysctls(virNetworkObjPtr obj)
2199
{
2200
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2201 2202
    char *field = NULL;
    int ret = -1;
2203
    bool enableIPv6 = !!virNetworkDefGetIPByIndex(def, AF_INET6, 0);
2204

2205 2206 2207 2208 2209
    /* 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",
2210
                    def->bridge) < 0)
2211
       goto cleanup;
2212

2213 2214
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
2215
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
2216
                      def->bridge);
2217 2218 2219
        ret = 0;
        goto cleanup;
    }
2220

2221 2222 2223
    if (virFileWriteStr(field, enableIPv6 ? "0" : "1", 0) < 0) {
        virReportSystemError(errno,
                             _("cannot write to %s to enable/disable IPv6 "
2224
                               "on bridge %s"), field, def->bridge);
2225
        goto cleanup;
2226
    }
2227
    VIR_FREE(field);
2228

2229 2230
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
2231 2232 2233 2234 2235 2236
     */

    /* 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",
2237
                    def->bridge) < 0)
2238 2239
        goto cleanup;

2240
    if (virFileWriteStr(field, "0", 0) < 0) {
2241
        virReportSystemError(errno,
2242 2243 2244 2245 2246
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

2247 2248 2249 2250
    /* 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",
2251
                    def->bridge) < 0)
2252 2253
        goto cleanup;

2254
    if (virFileWriteStr(field, "0", 0) < 0) {
2255
        virReportSystemError(errno,
2256
                             _("cannot disable %s"), field);
2257 2258 2259 2260
        goto cleanup;
    }

    ret = 0;
2261
 cleanup:
2262 2263 2264 2265
    VIR_FREE(field);
    return ret;
}

2266

2267
/* add an IP address to a bridge */
2268
static int
2269
networkAddAddrToBridge(virNetworkObjPtr obj,
2270
                       virNetworkIPDefPtr ipdef)
2271
{
2272
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2273
    int prefix = virNetworkIPDefPrefix(ipdef);
2274 2275

    if (prefix < 0) {
2276 2277
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
2278
                       def->bridge);
2279 2280 2281
        return -1;
    }

2282
    if (virNetDevIPAddrAdd(def->bridge, &ipdef->address, NULL, prefix) < 0)
2283 2284 2285 2286 2287
        return -1;

    return 0;
}

2288 2289

static int
2290
networkStartHandleMACTableManagerMode(virNetworkObjPtr obj,
2291 2292
                                      const char *macTapIfName)
{
2293 2294
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
    const char *brname = def->bridge;
2295 2296

    if (brname &&
2297
        def->macTableManager == VIR_NETWORK_BRIDGE_MAC_TABLE_MANAGER_LIBVIRT) {
2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310
        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;
}


2311 2312
/* add an IP (static) route to a bridge */
static int
2313
networkAddRouteToBridge(virNetworkObjPtr obj,
2314
                        virNetDevIPRoutePtr routedef)
2315
{
2316
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2317 2318 2319 2320
    int prefix = virNetDevIPRouteGetPrefix(routedef);
    unsigned int metric = virNetDevIPRouteGetMetric(routedef);
    virSocketAddrPtr addr = virNetDevIPRouteGetAddress(routedef);
    virSocketAddrPtr gateway = virNetDevIPRouteGetGateway(routedef);
2321 2322 2323 2324 2325

    if (prefix < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' has an invalid netmask "
                         "or IP address in route definition"),
2326
                       def->name);
2327 2328 2329
        return -1;
    }

2330
    if (virNetDevIPRouteAdd(def->bridge, addr, prefix, gateway, metric) < 0)
2331
        return -1;
2332

2333 2334 2335
    return 0;
}

2336
static int
2337
networkWaitDadFinish(virNetworkObjPtr obj)
2338
{
2339
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2340
    virNetworkIPDefPtr ipdef;
2341 2342 2343 2344
    virSocketAddrPtr *addrs = NULL, addr = NULL;
    size_t naddrs = 0;
    int ret = -1;

2345
    VIR_DEBUG("Begin waiting for IPv6 DAD on network %s", def->name);
2346

2347
    while ((ipdef = virNetworkDefGetIPByIndex(def, AF_INET6, naddrs))) {
2348 2349 2350 2351 2352
        addr = &ipdef->address;
        if (VIR_APPEND_ELEMENT_COPY(addrs, naddrs, addr) < 0)
            goto cleanup;
    }

2353
    ret = (naddrs == 0) ? 0 : virNetDevIPWaitDadFinish(addrs, naddrs);
2354 2355 2356 2357

 cleanup:
    VIR_FREE(addrs);
    VIR_DEBUG("Finished waiting for IPv6 DAD on network %s with status %d",
2358
              def->name, ret);
2359 2360 2361
    return ret;
}

2362

2363
static int
2364
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
2365
                           virNetworkObjPtr obj)
2366
{
2367
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2368
    size_t i;
2369
    bool v4present = false, v6present = false;
2370
    virErrorPtr save_err = NULL;
2371
    virNetworkIPDefPtr ipdef;
2372
    virNetDevIPRoutePtr routedef;
2373
    char *macTapIfName = NULL;
2374
    virMacMapPtr macmap;
M
Michal Privoznik 已提交
2375
    char *macMapFile = NULL;
2376
    int tapfd = -1;
2377
    pid_t dnsmasqPid;
2378

2379
    /* Check to see if any network IP collides with an existing route */
2380
    if (networkCheckRouteCollision(def) < 0)
2381 2382
        return -1;

2383
    /* Create and configure the bridge device */
2384
    if (!def->bridge) {
2385 2386 2387 2388 2389 2390 2391 2392 2393 2394
        /* 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"),
2395
                       def->name);
2396 2397
        return -1;
    }
2398
    if (virNetDevBridgeCreate(def->bridge) < 0)
2399 2400
        return -1;

2401
    if (def->mac_specified) {
2402 2403 2404 2405 2406 2407
        /* 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.
         */
2408
        macTapIfName = networkBridgeDummyNicName(def->bridge);
2409
        if (!macTapIfName)
2410
            goto err0;
2411
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2412 2413
        if (virNetDevTapCreateInBridgePort(def->bridge,
                                           &macTapIfName, &def->mac,
2414
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2415
                                           NULL, def->mtu, NULL,
2416 2417 2418
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2419 2420 2421 2422 2423
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2424
    if (!(macMapFile = virMacMapFileName(driver->dnsmasqStateDir,
2425
                                         def->bridge)) ||
2426
        !(macmap = virMacMapNew(macMapFile)))
M
Michal Privoznik 已提交
2427 2428
        goto err1;

2429 2430
    virNetworkObjSetMacMap(obj, macmap);

2431
    /* Set bridge options */
2432 2433 2434 2435

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2436
    if (virNetDevBridgeSetSTPDelay(def->bridge, def->delay * 1000) < 0)
2437
        goto err1;
2438

2439
    if (virNetDevBridgeSetSTP(def->bridge, def->stp ? true : false) < 0)
2440
        goto err1;
2441

2442 2443 2444
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
2445
    if (networkSetIPv6Sysctls(obj) < 0)
2446
        goto err1;
2447

2448
    /* Add "once per network" rules */
2449 2450
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN &&
        networkAddFirewallRules(def) < 0)
2451 2452
        goto err1;

2453
    for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i)); i++) {
2454
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2455
            v4present = true;
2456
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2457
            v6present = true;
2458

2459
        /* Add the IP address/netmask to the bridge */
2460
        if (networkAddAddrToBridge(obj, ipdef) < 0)
2461
            goto err2;
2462 2463
    }

2464
    if (networkStartHandleMACTableManagerMode(obj, macTapIfName) < 0)
2465 2466
        goto err2;

2467
    /* Bring up the bridge interface */
2468
    if (virNetDevSetOnline(def->bridge, 1) < 0)
2469
        goto err2;
2470

2471
    for (i = 0; i < def->nroutes; i++) {
2472 2473
        virSocketAddrPtr gateway = NULL;

2474
        routedef = def->routes[i];
2475
        gateway = virNetDevIPRouteGetGateway(routedef);
2476

2477 2478 2479
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2480
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2481
            if (networkAddRouteToBridge(obj, routedef) < 0) {
2482 2483 2484 2485 2486 2487
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2488
    /* If forward.type != NONE, turn on global IP forwarding */
2489
    if (def->forward.type != VIR_NETWORK_FORWARD_NONE) {
2490
        if (v6present && !virNetDevIPCheckIPv6Forwarding())
2491 2492 2493 2494 2495 2496 2497
            goto err3; /* Precise error message already provided */

        if (networkEnableIPForwarding(v4present, v6present) < 0) {
            virReportSystemError(errno, "%s",
                                 _("failed to enable IP forwarding"));
            goto err3;
        }
2498 2499
    }

2500

2501
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2502
    if ((v4present || v6present) &&
2503
        networkStartDhcpDaemon(driver, obj) < 0)
2504
        goto err3;
2505

2506
    /* start radvd if there are any ipv6 addresses */
2507
    if (v6present && networkStartRadvd(driver, obj) < 0)
2508 2509
        goto err4;

2510 2511 2512
    /* dnsmasq does not wait for DAD to complete before daemonizing,
     * so we need to wait for it ourselves.
     */
2513
    if (v6present && networkWaitDadFinish(obj) < 0)
2514 2515 2516 2517
        goto err4;

    /* DAD has finished, dnsmasq is now bound to the
     * bridge's IPv6 address, so we can set the dummy tun down.
2518 2519 2520 2521 2522 2523 2524
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
            goto err4;
        VIR_FORCE_CLOSE(tapfd);
    }

2525
    if (virNetDevBandwidthSet(def->bridge, def->bandwidth, true, true) < 0)
2526 2527
        goto err5;

2528
    VIR_FREE(macTapIfName);
M
Michal Privoznik 已提交
2529
    VIR_FREE(macMapFile);
2530 2531 2532

    return 0;

2533
 err5:
2534 2535
    if (def->bandwidth)
       virNetDevBandwidthClear(def->bridge);
2536

2537 2538 2539 2540
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2541 2542 2543 2544
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid > 0) {
        kill(dnsmasqPid, SIGTERM);
        virNetworkObjSetDnsmasqPid(obj, -1);
2545 2546
    }

2547 2548 2549
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2550
    ignore_value(virNetDevSetOnline(def->bridge, 0));
2551

2552 2553 2554
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2555 2556
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(def);
2557 2558

 err1:
2559 2560 2561
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2562
    if (macTapIfName) {
2563
        VIR_FORCE_CLOSE(tapfd);
2564
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2565 2566
        VIR_FREE(macTapIfName);
    }
M
Michal Privoznik 已提交
2567
    VIR_FREE(macMapFile);
2568 2569

 err0:
2570 2571
    if (!save_err)
        save_err = virSaveLastError();
2572
    ignore_value(virNetDevBridgeDelete(def->bridge));
2573

2574 2575 2576 2577
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2578
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2579 2580 2581
    return -1;
}

2582

2583 2584
static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
2585
                              virNetworkObjPtr obj)
2586
{
2587
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2588 2589 2590
    pid_t radvdPid;
    pid_t dnsmasqPid;

2591 2592
    if (def->bandwidth)
        virNetDevBandwidthClear(def->bridge);
2593

2594
    virNetworkObjUnrefMacMap(obj);
M
Michal Privoznik 已提交
2595

2596 2597
    radvdPid = virNetworkObjGetRadvdPid(obj);
    if (radvdPid > 0) {
2598 2599
        char *radvdpidbase;

2600
        kill(radvdPid, SIGTERM);
2601
        /* attempt to delete the pidfile we created */
2602
        if ((radvdpidbase = networkRadvdPidfileBasename(def->name))) {
2603
            virPidFileDelete(driver->pidDir, radvdpidbase);
2604 2605 2606 2607
            VIR_FREE(radvdpidbase);
        }
    }

2608 2609 2610
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid > 0)
        kill(dnsmasqPid, SIGTERM);
2611

2612 2613
    if (def->mac_specified) {
        char *macTapIfName = networkBridgeDummyNicName(def->bridge);
2614
        if (macTapIfName) {
2615
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2616 2617 2618 2619
            VIR_FREE(macTapIfName);
        }
    }

2620
    ignore_value(virNetDevSetOnline(def->bridge, 0));
2621

2622 2623
    if (def->forward.type != VIR_NETWORK_FORWARD_OPEN)
        networkRemoveFirewallRules(def);
2624

2625
    ignore_value(virNetDevBridgeDelete(def->bridge));
2626

2627
    /* See if its still alive and really really kill it */
2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638
    dnsmasqPid = virNetworkObjGetDnsmasqPid(obj);
    if (dnsmasqPid > 0 &&
        (kill(dnsmasqPid, 0) == 0))
        kill(dnsmasqPid, SIGKILL);
    virNetworkObjSetDnsmasqPid(obj, -1);

    radvdPid = virNetworkObjGetRadvdPid(obj);
    if (radvdPid > 0 &&
        (kill(radvdPid, 0) == 0))
        kill(radvdPid, SIGKILL);
    virNetworkObjSetRadvdPid(obj, -1);
2639

2640 2641 2642
    return 0;
}

2643

2644
static int
2645
networkStartNetworkBridge(virNetworkObjPtr obj)
2646 2647 2648 2649 2650
{
    /* 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.
     */
2651
    return networkStartHandleMACTableManagerMode(obj, NULL);
2652 2653
}

2654

2655
static int
2656
networkShutdownNetworkBridge(virNetworkObjPtr obj ATTRIBUTE_UNUSED)
2657 2658 2659 2660 2661 2662 2663 2664 2665
{
    /* 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;
}


2666 2667 2668 2669 2670 2671 2672 2673 2674
/* 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;
2675
    unsigned int maxVirtFns = 0;
2676 2677 2678 2679 2680 2681
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2682 2683 2684
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2685 2686
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev, &vfNames,
                                      &virtFns, &numVirtFns, &maxVirtFns)) < 0) {
2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732
        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:
2733
        case VIR_NETWORK_FORWARD_OPEN:
2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773
        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;
}


2774
static int
2775
networkStartNetworkExternal(virNetworkObjPtr obj)
2776 2777
{
    /* put anything here that needs to be done each time a network of
2778
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2779 2780 2781
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2782
    return networkCreateInterfacePool(virNetworkObjGetDef(obj));
2783 2784
}

2785 2786

static int
2787
networkShutdownNetworkExternal(virNetworkObjPtr obj ATTRIBUTE_UNUSED)
2788 2789
{
    /* put anything here that needs to be done each time a network of
2790
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2791 2792 2793 2794 2795 2796
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

2797

2798
static int
2799
networkStartNetwork(virNetworkDriverStatePtr driver,
2800
                    virNetworkObjPtr obj)
2801
{
2802
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2803 2804
    int ret = -1;

2805
    VIR_DEBUG("driver=%p, network=%p", driver, obj);
2806

2807
    if (virNetworkObjIsActive(obj)) {
2808 2809
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2810
        return ret;
2811 2812
    }

2813 2814 2815
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2816
    if (virNetworkObjSetDefTransient(obj, true) < 0)
2817
        goto cleanup;
2818

2819 2820
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2821
    if (networkRunHook(obj, NULL, NULL,
2822 2823 2824 2825
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2826
    switch (def->forward.type) {
2827 2828 2829 2830

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2831
    case VIR_NETWORK_FORWARD_OPEN:
2832
        if (networkStartNetworkVirtual(driver, obj) < 0)
2833
            goto cleanup;
2834 2835 2836
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2837
        if (def->bridge) {
2838
            if (networkStartNetworkBridge(obj) < 0)
2839 2840 2841 2842 2843 2844 2845
                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).
         */
2846 2847
        ATTRIBUTE_FALLTHROUGH;

2848 2849 2850
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2851
    case VIR_NETWORK_FORWARD_HOSTDEV:
2852
        if (networkStartNetworkExternal(obj) < 0)
2853
            goto cleanup;
2854 2855 2856
        break;
    }

2857
    /* finally we can call the 'started' hook script if any */
2858
    if (networkRunHook(obj, NULL, NULL,
2859 2860 2861 2862
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2863 2864 2865
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2866
    VIR_DEBUG("Writing network status to disk");
2867
    if (virNetworkObjSaveStatus(driver->stateDir, obj) < 0)
2868
        goto cleanup;
2869

2870
    virNetworkObjSetActive(obj, true);
2871
    VIR_INFO("Network '%s' started up", def->name);
2872
    ret = 0;
2873

2874
 cleanup:
2875
    if (ret < 0) {
2876
        virNetworkObjUnsetDefTransient(obj);
2877 2878
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2879
        networkShutdownNetwork(driver, obj);
2880 2881 2882 2883 2884 2885 2886
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2887

2888 2889
static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
2890
                       virNetworkObjPtr obj)
2891
{
2892
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
2893 2894 2895
    int ret = 0;
    char *stateFile;

2896
    VIR_INFO("Shutting down network '%s'", def->name);
2897

2898
    if (!virNetworkObjIsActive(obj))
2899 2900
        return 0;

2901
    stateFile = virNetworkConfigFile(driver->stateDir, def->name);
2902 2903 2904 2905 2906 2907
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2908
    switch (def->forward.type) {
2909 2910 2911 2912

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2913
    case VIR_NETWORK_FORWARD_OPEN:
2914
        ret = networkShutdownNetworkVirtual(driver, obj);
2915 2916 2917
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2918
        if (def->bridge) {
2919
            ret = networkShutdownNetworkBridge(obj);
2920 2921 2922 2923 2924 2925
            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).
         */
2926 2927
        ATTRIBUTE_FALLTHROUGH;

2928 2929 2930
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2931
    case VIR_NETWORK_FORWARD_HOSTDEV:
2932
        ret = networkShutdownNetworkExternal(obj);
2933 2934 2935
        break;
    }

2936
    /* now that we know it's stopped call the hook if present */
2937
    networkRunHook(obj, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2938 2939
                   VIR_HOOK_SUBOP_END);

2940
    virNetworkObjSetActive(obj, false);
2941
    virNetworkObjUnsetDefTransient(obj);
2942
    return ret;
2943 2944 2945
}


2946 2947 2948
static virNetworkPtr
networkLookupByUUID(virConnectPtr conn,
                    const unsigned char *uuid)
2949
{
2950
    virNetworkDriverStatePtr driver = networkGetDriver();
2951
    virNetworkObjPtr obj;
2952
    virNetworkDefPtr def;
2953
    virNetworkPtr net = NULL;
2954

2955 2956
    obj = virNetworkObjFindByUUID(driver->networks, uuid);
    if (!obj) {
2957 2958
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2959
        virReportError(VIR_ERR_NO_NETWORK,
2960 2961
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2962
        goto cleanup;
2963
    }
2964
    def = virNetworkObjGetDef(obj);
2965

2966
    if (virNetworkLookupByUUIDEnsureACL(conn, def) < 0)
2967 2968
        goto cleanup;

2969
    net = virGetNetwork(conn, def->name, def->uuid);
2970

2971
 cleanup:
2972 2973
    virNetworkObjEndAPI(&obj);
    return net;
2974 2975
}

2976 2977 2978 2979

static virNetworkPtr
networkLookupByName(virConnectPtr conn,
                    const char *name)
2980
{
2981
    virNetworkDriverStatePtr driver = networkGetDriver();
2982
    virNetworkObjPtr obj;
2983
    virNetworkDefPtr def;
2984
    virNetworkPtr net = NULL;
2985

2986 2987
    obj = virNetworkObjFindByName(driver->networks, name);
    if (!obj) {
2988 2989
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2990
        goto cleanup;
2991
    }
2992
    def = virNetworkObjGetDef(obj);
2993

2994
    if (virNetworkLookupByNameEnsureACL(conn, def) < 0)
2995 2996
        goto cleanup;

2997
    net = virGetNetwork(conn, def->name, def->uuid);
2998

2999
 cleanup:
3000 3001
    virNetworkObjEndAPI(&obj);
    return net;
3002 3003
}

3004 3005 3006

static int
networkConnectNumOfNetworks(virConnectPtr conn)
3007
{
3008
    virNetworkDriverStatePtr driver = networkGetDriver();
3009
    int nactive;
3010

3011 3012 3013
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

3014 3015 3016 3017
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
3018

3019 3020 3021
    return nactive;
}

3022 3023 3024 3025

static int
networkConnectListNetworks(virConnectPtr conn,
                           char **const names,
3026
                           int maxnames)
3027 3028
{
    virNetworkDriverStatePtr driver = networkGetDriver();
3029
    int got = 0;
3030

3031 3032 3033
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

3034
    got = virNetworkObjListGetNames(driver->networks,
3035
                                    true, names, maxnames,
3036 3037
                                    virConnectListNetworksCheckACL,
                                    conn);
3038

3039 3040 3041
    return got;
}

3042 3043 3044

static int
networkConnectNumOfDefinedNetworks(virConnectPtr conn)
3045
{
3046
    virNetworkDriverStatePtr driver = networkGetDriver();
3047
    int ninactive = 0;
3048

3049 3050 3051
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

3052 3053 3054 3055
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
3056

3057 3058 3059
    return ninactive;
}

3060 3061 3062 3063

static int
networkConnectListDefinedNetworks(virConnectPtr conn,
                                  char **const names,
3064
                                  int maxnames)
3065 3066
{
    virNetworkDriverStatePtr driver = networkGetDriver();
3067
    int got = 0;
3068

3069 3070 3071
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

3072
    got = virNetworkObjListGetNames(driver->networks,
3073
                                    false, names, maxnames,
3074 3075
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
3076 3077 3078
    return got;
}

3079

3080
static int
3081 3082 3083
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
3084
{
3085
    virNetworkDriverStatePtr driver = networkGetDriver();
3086 3087 3088 3089
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

3090 3091 3092
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

3093
    ret = virNetworkObjListExport(conn, driver->networks, nets,
3094 3095
                                  virConnectListAllNetworksCheckACL,
                                  flags);
3096

3097
 cleanup:
3098 3099
    return ret;
}
3100

3101

3102 3103 3104 3105 3106 3107 3108 3109
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
3110
    virNetworkDriverStatePtr driver = networkGetDriver();
3111 3112 3113 3114 3115 3116
    int ret = -1;

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

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
3117
                                       net, eventID, callback,
3118 3119 3120
                                       opaque, freecb, &ret) < 0)
        ret = -1;

3121
 cleanup:
3122 3123 3124
    return ret;
}

3125

3126 3127 3128 3129
static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
3130
    virNetworkDriverStatePtr driver = networkGetDriver();
3131 3132 3133 3134 3135
    int ret = -1;

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

3136 3137
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
3138
                                        callbackID, true) < 0)
3139 3140 3141
        goto cleanup;

    ret = 0;
3142

3143
 cleanup:
3144 3145 3146
    return ret;
}

3147 3148 3149

static int
networkIsActive(virNetworkPtr net)
3150 3151 3152 3153
{
    virNetworkObjPtr obj;
    int ret = -1;

3154 3155
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3156

3157
    if (virNetworkIsActiveEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
3158 3159
        goto cleanup;

3160 3161
    ret = virNetworkObjIsActive(obj);

3162
 cleanup:
3163
    virNetworkObjEndAPI(&obj);
3164 3165 3166
    return ret;
}

3167 3168 3169

static int
networkIsPersistent(virNetworkPtr net)
3170 3171 3172 3173
{
    virNetworkObjPtr obj;
    int ret = -1;

3174 3175
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
3176

3177
    if (virNetworkIsPersistentEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
3178 3179
        goto cleanup;

3180
    ret = virNetworkObjIsPersistent(obj);
3181

3182
 cleanup:
3183
    virNetworkObjEndAPI(&obj);
3184 3185 3186 3187
    return ret;
}


3188 3189
/*
 * networkFindUnusedBridgeName() - try to find a bridge name that is
3190 3191 3192
 * 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.
3193 3194 3195 3196 3197 3198 3199 3200
 */
static int
networkFindUnusedBridgeName(virNetworkObjListPtr nets,
                            virNetworkDefPtr def)
{

    int ret = -1, id = 0;
    char *newname = NULL;
3201 3202 3203 3204 3205
    const char *templ = "virbr%d";
    const char *p;

    if (def->bridge &&
        (p = strchr(def->bridge, '%')) == strrchr(def->bridge, '%') &&
3206
        p && p[1] == 'd')
3207
        templ = def->bridge;
3208 3209 3210 3211

    do {
        if (virAsprintf(&newname, templ, id) < 0)
            goto cleanup;
3212 3213 3214 3215 3216
        /* 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).
         */
3217
        if (!(virNetworkObjBridgeInUse(nets, newname, def->name) ||
3218
              virNetDevExists(newname) == 1)) {
3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251
            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")) {
3252
        if (virNetworkObjBridgeInUse(nets, def->bridge, def->name)) {
3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269
            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;
}


3270
static int
3271
networkValidate(virNetworkDriverStatePtr driver,
3272
                virNetworkDefPtr def)
3273
{
3274
    size_t i, j;
3275 3276
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
3277
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
3278
    bool ipv4def = false, ipv6def = false;
3279
    bool bandwidthAllowed = true;
3280
    bool usesInterface = false, usesAddress = false;
3281

3282 3283 3284
    if (virXMLCheckIllegalChars("name", def->name, "\n") < 0)
        return -1;

3285 3286 3287
    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
3288 3289
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
3290 3291
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE ||
        def->forward.type == VIR_NETWORK_FORWARD_OPEN) {
3292

3293 3294 3295 3296
        /* 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)
3297 3298 3299
            return -1;

        virNetworkSetBridgeMacAddr(def);
3300 3301
    } else {
        /* They are also the only types that currently support setting
3302 3303
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
3304
         */
3305 3306 3307 3308 3309 3310 3311 3312
        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;
        }
3313
        if (virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0)) {
3314 3315 3316 3317
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3318
                           virNetworkForwardTypeToString(def->forward.type));
3319 3320
            return -1;
        }
3321
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
3322 3323 3324 3325
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3326
                           virNetworkForwardTypeToString(def->forward.type));
3327 3328 3329 3330 3331 3332 3333
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
3334
                           virNetworkForwardTypeToString(def->forward.type));
3335 3336
            return -1;
        }
3337 3338 3339 3340 3341 3342 3343 3344
        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;
        }
3345
        bandwidthAllowed = false;
3346 3347
    }

3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360
    /* 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++) {
3361 3362 3363 3364
        virNetworkForwardIfDefPtr iface = &def->forward.ifs[i];
        char *sysfs_path = NULL;

        switch ((virNetworkForwardHostdevDeviceType) iface->type) {
3365 3366
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV:
            usesInterface = true;
3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377

            if (def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("hostdev network '%s' lists '%s' "
                                 "in the device pool, but hostdev "
                                 "networks require all devices to "
                                 "be listed by PCI address, not "
                                 "network device name"),
                               def->name, iface->device.dev);
                return -1;
            }
3378
            break;
3379 3380

        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI: {
3381
            usesAddress = true;
3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406

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

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

            if (!virPCIIsVirtualFunction(sysfs_path)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("device '%s' in network '%s' is not "
                                 "an SR-IOV Virtual Function"),
                               sysfs_path, def->name);
                VIR_FREE(sysfs_path);
                return -1;
            }
            VIR_FREE(sysfs_path);
3407
            break;
3408 3409
        }

3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422
        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
3423 3424 3425
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
3426
    for (i = 0;
3427
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
3428
         i++) {
G
Gene Czarcinski 已提交
3429 3430 3431 3432 3433
        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 -- "
3434 3435
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452
                    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;
                }
3453 3454 3455
            }
        }
    }
3456 3457 3458 3459 3460 3461

    /* 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.
     */

3462 3463 3464
    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
3465 3466
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
3467
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH));
3468 3469

    vlanUsed = def->vlan.nTags > 0;
3470 3471
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3472 3473 3474 3475 3476
            /* 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.
             */
3477
            if (def->portGroups[i].virtPortProfile) {
3478
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3479
                    def->portGroups[i].virtPortProfile->virtPortType
3480 3481 3482 3483 3484 3485 3486
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3487
        }
3488
        if (def->portGroups[i].isDefault) {
3489 3490 3491 3492 3493
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3494
                               def->name, defaultPortGroup->name,
3495
                               def->portGroups[i].name);
3496
                return -1;
3497
            }
3498
            defaultPortGroup = &def->portGroups[i];
3499
        }
3500 3501 3502 3503 3504 3505 3506 3507 3508
        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;
            }
        }
3509 3510 3511 3512 3513 3514 3515 3516
        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;
        }
3517
    }
3518 3519 3520 3521 3522 3523 3524
    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.
         */
3525 3526 3527 3528 3529 3530
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544

    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;
            }
        }
    }
3545 3546 3547
    return 0;
}

3548 3549 3550 3551

static virNetworkPtr
networkCreateXML(virConnectPtr conn,
                 const char *xml)
3552
{
3553
    virNetworkDriverStatePtr driver = networkGetDriver();
3554
    virNetworkDefPtr newDef;
3555
    virNetworkObjPtr obj = NULL;
3556
    virNetworkDefPtr def;
3557
    virNetworkPtr net = NULL;
3558
    virObjectEventPtr event = NULL;
3559

3560
    if (!(newDef = virNetworkDefParseString(xml)))
3561
        goto cleanup;
3562

3563
    if (virNetworkCreateXMLEnsureACL(conn, newDef) < 0)
3564 3565
        goto cleanup;

3566
    if (networkValidate(driver, newDef) < 0)
J
Ján Tomko 已提交
3567
        goto cleanup;
3568

3569 3570 3571
    /* 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.
3572
     */
3573
    if (!(obj = virNetworkObjAssignDef(driver->networks, newDef,
3574 3575
                                       VIR_NETWORK_OBJ_LIST_ADD_LIVE |
                                       VIR_NETWORK_OBJ_LIST_ADD_CHECK_LIVE)))
3576
        goto cleanup;
3577 3578
    newDef = NULL;
    def = virNetworkObjGetDef(obj);
3579

3580 3581
    if (networkStartNetwork(driver, obj) < 0) {
        virNetworkObjRemoveInactive(driver->networks, obj);
3582
        goto cleanup;
3583 3584
    }

3585 3586
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3587 3588
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3589

3590 3591
    VIR_INFO("Creating network '%s'", def->name);
    net = virGetNetwork(conn, def->name, def->uuid);
3592

3593
 cleanup:
3594
    virNetworkDefFree(newDef);
3595 3596
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3597 3598
    virNetworkObjEndAPI(&obj);
    return net;
3599 3600
}

3601 3602 3603 3604

static virNetworkPtr
networkDefineXML(virConnectPtr conn,
                 const char *xml)
3605
{
3606
    virNetworkDriverStatePtr driver = networkGetDriver();
3607
    virNetworkDefPtr def = NULL;
3608
    bool freeDef = true;
3609 3610
    virNetworkObjPtr obj = NULL;
    virNetworkPtr net = NULL;
3611
    virObjectEventPtr event = NULL;
3612

3613
    if (!(def = virNetworkDefParseString(xml)))
3614
        goto cleanup;
3615

3616 3617 3618
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3619
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3620
        goto cleanup;
3621

3622
    if (!(obj = virNetworkObjAssignDef(driver->networks, def, 0)))
J
Ján Tomko 已提交
3623
        goto cleanup;
3624

3625
    /* def was assigned to network object */
3626
    freeDef = false;
3627 3628

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3629 3630
        if (!virNetworkObjIsActive(obj)) {
            virNetworkObjRemoveInactive(driver->networks, obj);
3631 3632
            goto cleanup;
        }
3633 3634 3635 3636
        /* if network was active already, just undo new persistent
         * definition by making it transient.
         * XXX - this isn't necessarily the correct thing to do.
         */
3637
        virNetworkObjUpdateAssignDef(obj, NULL, false);
3638 3639 3640
        goto cleanup;
    }

3641
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3642 3643
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3644

3645
    VIR_INFO("Defining network '%s'", def->name);
3646
    net = virGetNetwork(conn, def->name, def->uuid);
3647

3648
 cleanup:
3649 3650
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3651
    if (freeDef)
J
Ján Tomko 已提交
3652
        virNetworkDefFree(def);
3653 3654
    virNetworkObjEndAPI(&obj);
    return net;
3655 3656
}

3657

3658
static int
3659 3660
networkUndefine(virNetworkPtr net)
{
3661
    virNetworkDriverStatePtr driver = networkGetDriver();
3662
    virNetworkObjPtr obj;
3663
    virNetworkDefPtr def;
3664
    int ret = -1;
3665
    bool active = false;
3666
    virObjectEventPtr event = NULL;
3667

3668
    if (!(obj = networkObjFromNetwork(net)))
3669
        goto cleanup;
3670
    def = virNetworkObjGetDef(obj);
3671

3672
    if (virNetworkUndefineEnsureACL(net->conn, def) < 0)
3673 3674
        goto cleanup;

3675
    if (virNetworkObjIsActive(obj))
3676
        active = true;
3677

3678
    if (!virNetworkObjIsPersistent(obj)) {
3679 3680 3681 3682 3683
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                       _("can't undefine transient network"));
        goto cleanup;
    }

3684
    /* remove autostart link */
3685 3686
    if (virNetworkObjDeleteConfig(driver->networkConfigDir,
                                  driver->networkAutostartDir,
3687
                                  obj) < 0)
3688
        goto cleanup;
3689

3690 3691
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3692 3693
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3694

3695
    VIR_INFO("Undefining network '%s'", def->name);
3696
    if (!active) {
3697
        if (networkRemoveInactive(driver, obj) < 0)
3698
            goto cleanup;
3699 3700 3701 3702 3703
    } else {

        /* if the network still exists, it was active, and we need to make
         * it transient (by deleting the persistent def)
         */
3704
        virNetworkObjUpdateAssignDef(obj, NULL, false);
3705 3706
    }

3707
    ret = 0;
3708

3709
 cleanup:
3710 3711
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3712
    virNetworkObjEndAPI(&obj);
3713
    return ret;
3714 3715
}

3716

3717 3718 3719 3720 3721 3722 3723 3724
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3725
    virNetworkDriverStatePtr driver = networkGetDriver();
3726
    virNetworkObjPtr obj = NULL;
3727
    virNetworkDefPtr def;
3728 3729
    int isActive, ret = -1;
    size_t i;
3730
    virNetworkIPDefPtr ipdef;
3731
    bool oldDhcpActive = false;
3732
    bool needFirewallRefresh = false;
3733

3734 3735 3736 3737 3738

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3739
    if (!(obj = networkObjFromNetwork(net)))
3740
        goto cleanup;
3741
    def = virNetworkObjGetDef(obj);
3742

3743
    if (virNetworkUpdateEnsureACL(net->conn, def, flags) < 0)
3744 3745
        goto cleanup;

3746
    /* see if we are listening for dhcp pre-modification */
3747
    for (i = 0;
3748
         (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
3749
         i++) {
3750 3751 3752 3753 3754 3755
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3756 3757
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3758
     */
3759
    isActive = virNetworkObjIsActive(obj);
3760 3761
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3762 3763 3764 3765 3766 3767 3768
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3769 3770 3771 3772
    if (isActive && (flags & VIR_NETWORK_UPDATE_AFFECT_LIVE)) {
        /* Take care of anything that must be done before updating the
         * live NetworkDef.
         */
3773 3774 3775
        if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
            def->forward.type == VIR_NETWORK_FORWARD_NAT ||
            def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
3776 3777 3778 3779 3780 3781 3782 3783 3784 3785
            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).
                 */
3786 3787
                if (def->forward.type != VIR_NETWORK_FORWARD_OPEN) {
                    networkRemoveFirewallRules(def);
3788 3789
                    needFirewallRefresh = true;
                }
3790 3791 3792 3793 3794 3795 3796
                break;
            default:
                break;
            }
        }
    }

3797
    /* update the network config in memory/on disk */
3798
    if (virNetworkObjUpdate(obj, command, section, parentIndex, xml, flags) < 0) {
3799
        if (needFirewallRefresh)
3800
            ignore_value(networkAddFirewallRules(def));
3801 3802 3803
        goto cleanup;
    }

3804 3805 3806 3807
    /* @def is replaced */
    def = virNetworkObjGetDef(obj);

    if (needFirewallRefresh && networkAddFirewallRules(def) < 0)
3808 3809 3810 3811 3812
        goto cleanup;

    if (flags & VIR_NETWORK_UPDATE_AFFECT_CONFIG) {
        /* save updated persistent config to disk */
        if (virNetworkSaveConfig(driver->networkConfigDir,
3813
                                 virNetworkObjGetPersistentDef(obj)) < 0) {
3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826
            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 ||
3827 3828 3829 3830 3831 3832 3833 3834
            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)
3835
             */
3836
            if (networkRestartDhcpDaemon(driver, obj) < 0)
3837 3838
                goto cleanup;

3839 3840 3841 3842 3843 3844 3845 3846
        } 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;

3847
            for (i = 0; (ipdef = virNetworkDefGetIPByIndex(def, AF_INET, i));
3848
                 i++) {
3849 3850 3851 3852 3853 3854 3855
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3856 3857
                 networkRestartDhcpDaemon(driver, obj) < 0) ||
                networkRefreshDhcpDaemon(driver, obj) < 0) {
3858 3859 3860
                goto cleanup;
            }

3861 3862 3863 3864
        } 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.
3865
             */
3866
            if (networkRefreshDhcpDaemon(driver, obj) < 0)
3867 3868 3869 3870 3871 3872 3873 3874
                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.
             */
3875
            if (networkRefreshRadvd(driver, obj) < 0)
3876 3877 3878 3879
                goto cleanup;
        }

        /* save current network state to disk */
3880
        if ((ret = virNetworkObjSaveStatus(driver->stateDir, obj)) < 0)
3881 3882
            goto cleanup;
    }
3883 3884

    /* call the 'updated' network hook script */
3885
    if (networkRunHook(obj, NULL, NULL, VIR_HOOK_NETWORK_OP_UPDATED,
3886 3887 3888
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

3889
    ret = 0;
3890
 cleanup:
3891
    virNetworkObjEndAPI(&obj);
3892 3893 3894
    return ret;
}

3895 3896 3897

static int
networkCreate(virNetworkPtr net)
3898
{
3899
    virNetworkDriverStatePtr driver = networkGetDriver();
3900
    virNetworkObjPtr obj;
3901
    virNetworkDefPtr def;
3902
    int ret = -1;
3903
    virObjectEventPtr event = NULL;
3904

3905
    if (!(obj = networkObjFromNetwork(net)))
3906
        goto cleanup;
3907
    def = virNetworkObjGetDef(obj);
3908

3909
    if (virNetworkCreateEnsureACL(net->conn, def) < 0)
3910 3911
        goto cleanup;

3912
    if ((ret = networkStartNetwork(driver, obj)) < 0)
3913
        goto cleanup;
3914

3915 3916
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3917 3918
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3919

3920
 cleanup:
3921 3922
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3923
    virNetworkObjEndAPI(&obj);
3924
    return ret;
3925 3926
}

3927 3928 3929

static int
networkDestroy(virNetworkPtr net)
3930
{
3931
    virNetworkDriverStatePtr driver = networkGetDriver();
3932
    virNetworkObjPtr obj;
3933
    virNetworkDefPtr def;
3934
    int ret = -1;
3935
    virObjectEventPtr event = NULL;
3936

3937
    if (!(obj = networkObjFromNetwork(net)))
3938
        goto cleanup;
3939
    def = virNetworkObjGetDef(obj);
3940

3941
    if (virNetworkDestroyEnsureACL(net->conn, def) < 0)
3942 3943
        goto cleanup;

3944
    if (!virNetworkObjIsActive(obj)) {
3945
        virReportError(VIR_ERR_OPERATION_INVALID,
3946
                       _("network '%s' is not active"),
3947
                       def->name);
3948 3949 3950
        goto cleanup;
    }

3951
    if ((ret = networkShutdownNetwork(driver, obj)) < 0)
3952
        goto cleanup;
3953 3954
    /* @def replaced in virNetworkObjUnsetDefTransient*/
    def = virNetworkObjGetDef(obj);
3955

3956 3957
    event = virNetworkEventLifecycleNew(def->name,
                                        def->uuid,
3958 3959
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3960

3961 3962
    if (!virNetworkObjIsPersistent(obj) &&
        networkRemoveInactive(driver, obj) < 0) {
3963 3964
        ret = -1;
        goto cleanup;
3965
    }
3966

3967
 cleanup:
3968 3969
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3970
    virNetworkObjEndAPI(&obj);
3971 3972 3973
    return ret;
}

3974 3975 3976 3977

static char *
networkGetXMLDesc(virNetworkPtr net,
                  unsigned int flags)
3978
{
3979
    virNetworkObjPtr obj;
3980
    virNetworkDefPtr curDef;
3981
    virNetworkDefPtr def;
3982
    virNetworkDefPtr newDef;
3983
    char *ret = NULL;
3984

3985
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3986

3987
    if (!(obj = networkObjFromNetwork(net)))
3988
        return ret;
3989 3990
    def = virNetworkObjGetDef(obj);
    newDef = virNetworkObjGetNewDef(obj);
3991

3992
    if (virNetworkGetXMLDescEnsureACL(net->conn, def) < 0)
3993 3994
        goto cleanup;

3995 3996
    if ((flags & VIR_NETWORK_XML_INACTIVE) && newDef)
        curDef = newDef;
3997
    else
3998
        curDef = def;
3999

4000
    ret = virNetworkDefFormat(curDef, flags);
4001

4002
 cleanup:
4003
    virNetworkObjEndAPI(&obj);
4004
    return ret;
4005 4006
}

4007 4008 4009 4010

static char *
networkGetBridgeName(virNetworkPtr net)
{
4011
    virNetworkObjPtr obj;
4012
    virNetworkDefPtr def;
4013 4014
    char *bridge = NULL;

4015
    if (!(obj = networkObjFromNetwork(net)))
4016
        return bridge;
4017
    def = virNetworkObjGetDef(obj);
4018

4019
    if (virNetworkGetBridgeNameEnsureACL(net->conn, def) < 0)
4020 4021
        goto cleanup;

4022
    if (!(def->bridge)) {
4023 4024
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
4025
                       def->name);
4026 4027 4028
        goto cleanup;
    }

4029
    ignore_value(VIR_STRDUP(bridge, def->bridge));
4030

4031
 cleanup:
4032
    virNetworkObjEndAPI(&obj);
4033 4034 4035
    return bridge;
}

4036 4037 4038 4039

static int
networkGetAutostart(virNetworkPtr net,
                    int *autostart)
4040
{
4041
    virNetworkObjPtr obj;
4042
    int ret = -1;
4043

4044
    if (!(obj = networkObjFromNetwork(net)))
4045
        return ret;
4046

4047
    if (virNetworkGetAutostartEnsureACL(net->conn, virNetworkObjGetDef(obj)) < 0)
4048 4049
        goto cleanup;

4050
    *autostart = virNetworkObjIsAutostart(obj) ? 1 : 0;
4051
    ret = 0;
4052

4053
 cleanup:
4054
    virNetworkObjEndAPI(&obj);
4055
    return ret;
4056 4057
}

4058 4059 4060 4061

static int
networkSetAutostart(virNetworkPtr net,
                    int autostart)
4062
{
4063
    virNetworkDriverStatePtr driver = networkGetDriver();
4064
    virNetworkObjPtr obj;
4065
    virNetworkDefPtr def;
4066
    char *configFile = NULL, *autostartLink = NULL;
4067 4068
    bool new_autostart;
    bool cur_autostart;
4069
    int ret = -1;
4070

4071
    if (!(obj = networkObjFromNetwork(net)))
4072
        goto cleanup;
4073
    def = virNetworkObjGetDef(obj);
4074

4075
    if (virNetworkSetAutostartEnsureACL(net->conn, def) < 0)
4076 4077
        goto cleanup;

4078
    if (!virNetworkObjIsPersistent(obj)) {
4079 4080
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
4081 4082 4083
        goto cleanup;
    }

4084 4085 4086
    new_autostart = (autostart != 0);
    cur_autostart = virNetworkObjIsAutostart(obj);
    if (cur_autostart != new_autostart) {
4087 4088
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir,
                                               def->name)) == NULL)
4089
            goto cleanup;
4090 4091
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir,
                                                  def->name)) == NULL)
4092 4093
            goto cleanup;

4094
        if (new_autostart) {
4095
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
4096
                virReportSystemError(errno,
4097 4098
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
4099 4100
                goto cleanup;
            }
4101

4102
            if (symlink(configFile, autostartLink) < 0) {
4103
                virReportSystemError(errno,
4104
                                     _("Failed to create symlink '%s' to '%s'"),
4105
                                     autostartLink, configFile);
4106 4107 4108
                goto cleanup;
            }
        } else {
4109
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
4110
                virReportSystemError(errno,
4111
                                     _("Failed to delete symlink '%s'"),
4112
                                     autostartLink);
4113 4114
                goto cleanup;
            }
4115 4116
        }

4117
        virNetworkObjSetAutostart(obj, new_autostart);
4118
    }
4119

4120
    ret = 0;
4121

4122
 cleanup:
4123 4124
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
4125
    virNetworkObjEndAPI(&obj);
4126
    return ret;
4127 4128
}

4129

4130
static int
4131
networkGetDHCPLeases(virNetworkPtr net,
4132 4133 4134
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
4135
{
4136
    virNetworkDriverStatePtr driver = networkGetDriver();
4137 4138 4139
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
4140
    ssize_t size = 0;
4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151
    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;
4152
    virNetworkIPDefPtr ipdef_tmp = NULL;
4153 4154
    virNetworkDHCPLeasePtr lease = NULL;
    virNetworkDHCPLeasePtr *leases_ret = NULL;
4155
    virNetworkObjPtr obj;
4156
    virNetworkDefPtr def;
4157
    virMacAddr mac_addr;
4158 4159 4160

    virCheckFlags(0, -1);

4161 4162 4163 4164 4165 4166
    /* only to check if the MAC is valid */
    if (mac && virMacAddrParse(mac, &mac_addr) < 0) {
        virReportError(VIR_ERR_INVALID_MAC, "%s", mac);
        return -1;
    }

4167
    if (!(obj = networkObjFromNetwork(net)))
4168
        return -1;
4169
    def = virNetworkObjGetDef(obj);
4170

4171
    if (virNetworkGetDHCPLeasesEnsureACL(net->conn, def) < 0)
4172
        goto cleanup;
4173 4174

    /* Retrieve custom leases file location */
4175
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(driver, def->bridge);
4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218

    /* 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;
        }

4219
        if (mac && virMacAddrCompare(mac, mac_tmp))
4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250
            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 */
4251 4252
            for (j = 0; j < def->nips; j++) {
                ipdef_tmp = &def->ips[j];
4253 4254 4255 4256 4257 4258 4259 4260

                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)) {
4261
                    lease->prefix = virSocketAddrGetIPPrefix(&ipdef_tmp->address,
4262 4263 4264 4265 4266 4267 4268 4269
                                                             &ipdef_tmp->netmask,
                                                             ipdef_tmp->prefix);
                    break;
                }
            }

            if ((VIR_STRDUP(lease->mac, mac_tmp) < 0) ||
                (VIR_STRDUP(lease->ipaddr, ip_tmp) < 0) ||
4270
                (VIR_STRDUP(lease->iface, def->bridge) < 0))
4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302
                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);
4303
    VIR_FREE(lease_entries);
4304 4305
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
4306

4307
    virNetworkObjEndAPI(&obj);
4308

4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319
    return rv;

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

4320

4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355
/* 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);
        }
    }
}

4356

4357 4358 4359 4360 4361 4362 4363 4364 4365
/* 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:
4366
 * @dom: domain definition that @iface belongs to
4367 4368 4369 4370 4371 4372 4373 4374 4375 4376
 * @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.
 */
4377
static int
4378 4379
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
4380
{
4381
    virNetworkDriverStatePtr driver = networkGetDriver();
4382
    virDomainNetType actualType = iface->type;
4383
    virNetworkObjPtr obj = NULL;
4384
    virNetworkDefPtr netdef = NULL;
4385
    virNetDevBandwidthPtr bandwidth = NULL;
4386 4387 4388
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
4389
    virNetworkForwardIfDefPtr dev = NULL;
4390
    size_t i;
4391 4392 4393
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
4394
        goto validate;
4395 4396 4397 4398

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

4399 4400
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
4401 4402 4403
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4404
        goto error;
4405
    }
4406
    netdef = virNetworkObjGetDef(obj);
4407

4408
    if (!virNetworkObjIsActive(obj)) {
4409 4410 4411 4412 4413 4414
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

4415 4416 4417
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

4418 4419 4420
    /* 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
4421
     */
4422 4423 4424 4425 4426 4427
    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.
     */
4428 4429 4430 4431 4432 4433

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

4434 4435
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
4436
        goto error;
4437

4438 4439 4440 4441 4442 4443 4444 4445
    /* 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;

4446 4447
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
4448

4449 4450 4451 4452 4453 4454 4455 4456 4457 4458
    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;

4459 4460
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
4461 4462
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_OPEN)) {
4463
        /* for these forward types, the actual net type really *is*
4464
         * NETWORK; we just keep the info from the portgroup in
4465
         * iface->data.network.actual
J
Ján Tomko 已提交
4466
         */
4467
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
4468

4469
        /* we also store the bridge device and macTableManager settings
4470 4471 4472 4473 4474 4475 4476
         * 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;
4477 4478
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4479

4480
        if (networkPlugBandwidth(obj, iface) < 0)
4481 4482
            goto error;

4483
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
4484
               netdef->bridge) {
4485 4486 4487 4488 4489

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

4490
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
4491 4492
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
4493
            goto error;
4494 4495
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4496

4497 4498 4499 4500 4501 4502 4503 4504
        /* 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) {
4505
            goto error;
4506 4507 4508 4509 4510 4511 4512 4513 4514 4515
        }
        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);
4516
                goto error;
4517 4518 4519
            }
        }

4520
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4521

4522
        virDomainHostdevSubsysPCIBackendType backend;
4523

4524
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
4525
        if (networkCreateInterfacePool(netdef) < 0)
4526 4527 4528
            goto error;

        /* pick first dev with 0 connections */
4529 4530 4531
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545
                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;
4546
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
4547
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
4548
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
4549

E
Eric Blake 已提交
4550
        switch (netdef->forward.driverName) {
4551
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
4552
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
4553 4554
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
4555
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
4556 4557
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
4558
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569
            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;

4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594
        /* 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;
            }
        }

4595 4596 4597 4598
    } 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)) {
4599 4600 4601 4602 4603 4604

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

        /* Set type=direct and appropriate <source mode='xxx'/> */
4605
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4606
        switch (netdef->forward.type) {
4607
        case VIR_NETWORK_FORWARD_BRIDGE:
4608
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
4609 4610
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
4611
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
4612 4613
            break;
        case VIR_NETWORK_FORWARD_VEPA:
4614
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
4615 4616
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
4617
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
4618 4619 4620
            break;
        }

4621 4622 4623 4624 4625 4626 4627 4628
        /* 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) {
4629
            goto error;
4630
        }
4631
        virtport = iface->data.network.actual->virtPortProfile;
4632
        if (virtport) {
4633 4634 4635 4636 4637 4638 4639 4640
            /* 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);
4641
                goto error;
4642 4643
            }
        }
4644

4645 4646 4647
        /* 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).
         */
4648
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4649 4650 4651 4652
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4653
            goto error;
4654 4655 4656
        } else {
            /* pick an interface from the pool */

4657
            if (networkCreateInterfacePool(netdef) < 0)
4658 4659
                goto error;

4660 4661 4662 4663 4664
            /* 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.
4665
             */
4666 4667
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4668 4669 4670
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4671

4672
                /* pick first dev with 0 connections */
4673 4674 4675
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4676 4677 4678 4679 4680
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4681
                dev = &netdef->forward.ifs[0];
4682 4683 4684
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4685 4686 4687 4688
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4689 4690 4691 4692
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4693
                goto error;
4694
            }
4695 4696
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4697
                goto error;
4698 4699 4700
        }
    }

4701 4702
    if (virNetworkObjMacMgrAdd(obj, driver->dnsmasqStateDir,
                               dom->name, &iface->mac) < 0)
M
Michal Privoznik 已提交
4703 4704
        goto error;

4705
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4706
        goto error;
4707

4708
 validate:
4709 4710 4711 4712 4713
    /* 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.
     */

4714
    if (virDomainNetGetActualVlan(iface)) {
4715 4716 4717 4718
        /* 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
4719 4720
         */
        if (!(actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV ||
4721 4722 4723
              (actualType == VIR_DOMAIN_NET_TYPE_DIRECT &&
               virDomainNetGetActualDirectMode(iface)
               == VIR_NETDEV_MACVLAN_MODE_PASSTHRU) ||
4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742
              (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;
        }
    }
4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753
    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;
        }
    }
4754 4755 4756

    if (netdef) {
        netdef->connections++;
4757
        if (dev)
4758 4759
            dev->connections++;
        /* finally we can call the 'plugged' hook script if any */
4760
        if (networkRunHook(obj, dom, iface,
4761 4762 4763
                           VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
                           VIR_HOOK_SUBOP_BEGIN) < 0) {
            /* adjust for failure */
4764
            netdef->connections--;
4765 4766 4767 4768
            if (dev)
                dev->connections--;
            goto error;
        }
4769
        networkLogAllocation(netdef, actualType, dev, iface, true);
4770 4771
    }

4772
    ret = 0;
4773

4774
 cleanup:
4775
    virNetworkObjEndAPI(&obj);
4776 4777
    return ret;

4778
 error:
4779
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4780 4781 4782
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4783
    goto cleanup;
4784 4785
}

4786

4787
/* networkNotifyActualDevice:
4788
 * @dom: domain definition that @iface belongs to
4789 4790 4791 4792 4793
 * @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
4794 4795
 * order, or re-attach the interface's tap device to the network's
 * bridge.
4796
 *
4797
 * No return value (but does log any failures)
4798
 */
4799
static void
4800 4801
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4802
{
4803
    virNetworkDriverStatePtr driver = networkGetDriver();
4804
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4805
    virNetworkObjPtr obj;
4806
    virNetworkDefPtr netdef;
4807
    virNetworkForwardIfDefPtr dev = NULL;
4808
    size_t i;
4809
    char *master = NULL;
4810 4811

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
4812
        return;
4813

4814 4815
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
4816 4817 4818
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4819 4820
        goto error;
    }
4821
    netdef = virNetworkObjGetDef(obj);
4822

4823
    if (!virNetworkObjIsActive(obj)) {
4824 4825 4826 4827 4828 4829
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840
    /* 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;

4841 4842 4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865
    /* see if we're connected to the correct bridge */
    if (netdef->bridge) {
        if (virNetDevGetMaster(iface->ifname, &master) < 0)
            goto error;

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

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

4866
    if (!iface->data.network.actual ||
4867 4868
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4869 4870
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4871 4872
    }

4873
    if (networkCreateInterfacePool(netdef) < 0)
4874
        goto error;
4875

4876
    if (netdef->forward.nifs == 0) {
4877
        virReportError(VIR_ERR_INTERNAL_ERROR,
4878 4879
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4880
                       netdef->name);
4881
        goto error;
4882
    }
4883

4884 4885
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4886

4887 4888 4889 4890 4891 4892 4893 4894 4895
        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 */
4896 4897
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4898
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4899 4900
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4901 4902 4903 4904 4905
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4906
            virReportError(VIR_ERR_INTERNAL_ERROR,
4907 4908
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4909
                           netdef->name, actualDev);
4910
            goto error;
4911 4912
        }

4913
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4914 4915
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4916
         */
4917
        if ((dev->connections > 0) &&
4918 4919
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4920 4921
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4922
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4923
            virReportError(VIR_ERR_INTERNAL_ERROR,
4924 4925
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4926
                           netdef->name, actualDev);
4927
            goto error;
4928
        }
4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940
    }  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 */
4941 4942
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4943
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4944
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4945 4946
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4947 4948 4949 4950 4951 4952 4953 4954 4955
                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,
4956 4957 4958 4959
                           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
4960
            goto error;
4961 4962 4963 4964 4965 4966 4967
        }

        /* 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) &&
4968
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4969 4970 4971 4972 4973 4974 4975 4976 4977
            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;
        }
4978 4979
    }

4980
 success:
4981
    netdef->connections++;
4982 4983
    if (dev)
        dev->connections++;
4984
    /* finally we can call the 'plugged' hook script if any */
4985
    if (networkRunHook(obj, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
4986 4987 4988 4989 4990 4991 4992
                       VIR_HOOK_SUBOP_BEGIN) < 0) {
        /* adjust for failure */
        if (dev)
            dev->connections--;
        netdef->connections--;
        goto error;
    }
4993
    networkLogAllocation(netdef, actualType, dev, iface, true);
4994

4995
 cleanup:
4996
    virNetworkObjEndAPI(&obj);
4997
    VIR_FREE(master);
4998
    return;
4999

5000
 error:
5001
    goto cleanup;
5002 5003 5004 5005
}


/* networkReleaseActualDevice:
5006
 * @dom: domain definition that @iface belongs to
5007 5008 5009 5010 5011 5012 5013 5014 5015
 * @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.
 */
5016
static int
5017 5018
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
5019
{
5020
    virNetworkDriverStatePtr driver = networkGetDriver();
5021
    virDomainNetType actualType = virDomainNetGetActualType(iface);
5022
    virNetworkObjPtr obj;
5023
    virNetworkDefPtr netdef;
5024
    virNetworkForwardIfDefPtr dev = NULL;
5025 5026
    size_t i;
    int ret = -1;
5027 5028

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
5029
        return 0;
5030

5031 5032
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
5033 5034 5035
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
5036 5037
        goto error;
    }
5038
    netdef = virNetworkObjGetDef(obj);
5039

5040 5041
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
5042
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
5043 5044
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE ||
         netdef->forward.type == VIR_NETWORK_FORWARD_OPEN) &&
5045
        networkUnplugBandwidth(obj, iface) < 0)
5046 5047
        goto error;

5048 5049 5050
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
5051 5052
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
5053 5054
    }

5055
    if (netdef->forward.nifs == 0) {
5056
        virReportError(VIR_ERR_INTERNAL_ERROR,
5057
                       _("network '%s' uses a direct/hostdev mode, but "
5058 5059
                         "has no forward dev and no interface pool"),
                       netdef->name);
5060
        goto error;
5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072
    }

    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;
        }
5073

5074 5075
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
5076
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
5077 5078
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
5079 5080 5081
                break;
            }
        }
5082

5083
        if (!dev) {
5084
            virReportError(VIR_ERR_INTERNAL_ERROR,
5085 5086
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
5087
                           netdef->name, actualDev);
5088
            goto error;
5089
        }
5090 5091 5092 5093 5094 5095 5096 5097 5098 5099
    } 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;
        }

5100 5101
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
5102
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
5103
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
5104 5105
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
5106 5107 5108 5109 5110 5111 5112 5113 5114
                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,
5115 5116 5117 5118
                           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
5119
            goto error;
5120
        }
J
Ján Tomko 已提交
5121
    }
5122

5123
 success:
5124
    virNetworkObjMacMgrDel(obj, driver->dnsmasqStateDir, dom->name, &iface->mac);
M
Michal Privoznik 已提交
5125

5126
    if (iface->data.network.actual) {
5127
        netdef->connections--;
5128 5129
        if (dev)
            dev->connections--;
5130
        /* finally we can call the 'unplugged' hook script if any */
5131
        networkRunHook(obj, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
5132
                       VIR_HOOK_SUBOP_BEGIN);
5133
        networkLogAllocation(netdef, actualType, dev, iface, false);
5134
    }
5135
    ret = 0;
5136
 cleanup:
5137
    virNetworkObjEndAPI(&obj);
5138 5139 5140 5141
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
5142
    return ret;
5143

5144
 error:
5145
    goto cleanup;
5146
}
5147

5148

5149 5150 5151
/**
 * networkCheckBandwidth:
 * @net: network QoS
5152
 * @ifaceBand: interface QoS (may be NULL if no QoS)
5153
 * @oldBandwidth: new interface QoS (may be NULL if no QoS)
5154
 * @ifaceMac: interface MAC (used in error messages for identification)
5155 5156
 * @new_rate: new rate for non guaranteed class
 *
5157 5158 5159 5160 5161 5162 5163 5164
 * 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.
 *
5165 5166 5167 5168 5169
 * 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
5170
networkCheckBandwidth(virNetworkObjPtr obj,
5171
                      virNetDevBandwidthPtr ifaceBand,
5172
                      virNetDevBandwidthPtr oldBandwidth,
5173
                      virMacAddr ifaceMac,
5174 5175 5176
                      unsigned long long *new_rate)
{
    int ret = -1;
5177 5178
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
    virNetDevBandwidthPtr netBand = def->bandwidth;
5179
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5180 5181 5182
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

5183
    virMacAddrFormat(&ifaceMac, ifmac);
5184 5185 5186 5187 5188 5189

    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"),
5190
                       ifmac, def->name);
5191 5192 5193
        return -1;
    }

5194 5195
    if (((!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor) &&
         (!oldBandwidth || !oldBandwidth->in || !oldBandwidth->in->floor)) ||
5196 5197
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
5198
        return 1;
5199
    }
5200 5201

    tmp_new_rate = netBand->in->average;
5202 5203 5204 5205
    if (oldBandwidth && oldBandwidth->in)
        tmp_floor_sum -= oldBandwidth->in->floor;
    if (ifaceBand && ifaceBand->in)
        tmp_floor_sum += ifaceBand->in->floor;
5206 5207 5208 5209 5210 5211 5212 5213 5214

    /* 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,
5215 5216
                           def->bridge,
                           def->name);
5217 5218 5219 5220 5221 5222 5223 5224 5225
            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,
5226 5227
                       def->bridge,
                       def->name);
5228 5229 5230
        goto cleanup;
    }

5231 5232
    if (new_rate)
        *new_rate = tmp_new_rate;
5233 5234
    ret = 0;

5235
 cleanup:
5236 5237 5238
    return ret;
}

5239

5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250
/**
 * 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
5251
networkNextClassID(virNetworkObjPtr obj)
5252
{
5253
    ssize_t ret = 0;
5254
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5255

5256 5257
    if ((ret = virBitmapNextClearBit(classIdMap, -1)) < 0)
        ret = virBitmapSize(classIdMap);
5258

5259
    if (virBitmapSetBitExpand(classIdMap, ret) < 0)
5260 5261 5262 5263 5264
        return -1;

    return ret;
}

5265

5266
static int
5267
networkPlugBandwidthImpl(virNetworkObjPtr obj,
5268 5269 5270
                         virDomainNetDefPtr iface,
                         virNetDevBandwidthPtr ifaceBand,
                         unsigned long long new_rate)
5271
{
5272
    virNetworkDriverStatePtr driver = networkGetDriver();
5273
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
5274
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5275
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5276
    ssize_t class_id = 0;
5277 5278
    int plug_ret;
    int ret = -1;
5279 5280

    /* generate new class_id */
5281
    if ((class_id = networkNextClassID(obj)) < 0) {
5282 5283 5284 5285 5286
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Could not generate next class ID"));
        goto cleanup;
    }

5287
    plug_ret = virNetDevBandwidthPlug(def->bridge, def->bandwidth,
5288
                                      &iface->mac, ifaceBand, class_id);
5289
    if (plug_ret < 0) {
5290
        ignore_value(virNetDevBandwidthUnplug(def->bridge, class_id));
5291 5292 5293 5294 5295 5296
        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 */
5297 5298
    tmp_floor_sum += ifaceBand->in->floor;
    virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5299
    /* update status file */
5300
    if (virNetworkObjSaveStatus(driver->stateDir, obj) < 0) {
5301
        ignore_value(virBitmapClearBit(classIdMap, class_id));
5302 5303
        tmp_floor_sum -= ifaceBand->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5304
        iface->data.network.actual->class_id = 0;
5305
        ignore_value(virNetDevBandwidthUnplug(def->bridge, class_id));
5306 5307
        goto cleanup;
    }
5308
    /* update rate for non guaranteed NICs */
5309
    new_rate -= tmp_floor_sum;
5310 5311
    if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                     def->bandwidth, new_rate) < 0)
5312
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
5313
                 def->bridge);
5314 5315

    ret = 0;
5316 5317 5318 5319 5320 5321
 cleanup:
    return ret;
}


static int
5322
networkPlugBandwidth(virNetworkObjPtr obj,
5323 5324 5325 5326 5327 5328 5329 5330
                     virDomainNetDefPtr iface)
{
    int ret = -1;
    int plug_ret;
    unsigned long long new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);

5331
    if ((plug_ret = networkCheckBandwidth(obj, ifaceBand, NULL,
5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351
                                          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;
    }

5352
    if (networkPlugBandwidthImpl(obj, iface, ifaceBand, new_rate) < 0)
5353 5354 5355
        goto cleanup;

    ret = 0;
5356

5357
 cleanup:
5358 5359 5360
    return ret;
}

5361

5362
static int
5363
networkUnplugBandwidth(virNetworkObjPtr obj,
5364 5365
                       virDomainNetDefPtr iface)
{
5366
    virNetworkDefPtr def = virNetworkObjGetDef(obj);
5367
    virBitmapPtr classIdMap = virNetworkObjGetClassIdMap(obj);
5368
    unsigned long long tmp_floor_sum = virNetworkObjGetFloorSum(obj);
5369
    virNetworkDriverStatePtr driver = networkGetDriver();
5370 5371
    int ret = 0;
    unsigned long long new_rate;
5372
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
5373 5374 5375

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
5376
        if (!def->bandwidth || !def->bandwidth->in) {
5377
            VIR_WARN("Network %s has no bandwidth but unplug requested",
5378
                     def->name);
5379 5380
            goto cleanup;
        }
5381
        /* we must remove class from bridge */
5382
        new_rate = def->bandwidth->in->average;
5383

5384 5385
        if (def->bandwidth->in->peak > 0)
            new_rate = def->bandwidth->in->peak;
5386

5387
        ret = virNetDevBandwidthUnplug(def->bridge,
5388 5389 5390 5391
                                       iface->data.network.actual->class_id);
        if (ret < 0)
            goto cleanup;
        /* update sum of 'floor'-s of attached NICs */
5392 5393 5394
        tmp_floor_sum -= ifaceBand->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);

5395
        /* return class ID */
5396
        ignore_value(virBitmapClearBit(classIdMap,
5397 5398
                                       iface->data.network.actual->class_id));
        /* update status file */
5399
        if (virNetworkObjSaveStatus(driver->stateDir, obj) < 0) {
5400 5401
            tmp_floor_sum += ifaceBand->in->floor;
            virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5402
            ignore_value(virBitmapSetBit(classIdMap,
5403 5404 5405
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
5406
        /* update rate for non guaranteed NICs */
5407
        new_rate -= tmp_floor_sum;
5408 5409
        if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                         def->bandwidth, new_rate) < 0)
5410
            VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
5411
                     def->bridge);
5412 5413 5414 5415
        /* no class is associated any longer */
        iface->data.network.actual->class_id = 0;
    }

5416
 cleanup:
5417 5418
    return ret;
}
5419

5420

5421
static void
5422
networkNetworkObjTaint(virNetworkObjPtr obj,
5423
                       virNetworkTaintFlags taint)
5424
{
5425 5426
    virNetworkDefPtr def = virNetworkObjGetDef(obj);

5427
    if (virNetworkObjTaint(obj, taint)) {
5428
        char uuidstr[VIR_UUID_STRING_BUFLEN];
5429
        virUUIDFormat(def->uuid, uuidstr);
5430 5431

        VIR_WARN("Network name='%s' uuid=%s is tainted: %s",
5432
                 def->name, uuidstr, virNetworkTaintTypeToString(taint));
5433 5434
    }
}
5435 5436 5437 5438 5439 5440


static bool
networkBandwidthGenericChecks(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
5441
    virNetDevBandwidthPtr ifaceBand;
5442 5443 5444 5445 5446 5447 5448 5449
    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;
    }

5450
    ifaceBand = virDomainNetGetActualBandwidth(iface);
5451 5452 5453 5454 5455 5456 5457 5458 5459 5460 5461
    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;
}


5462
static bool
5463 5464 5465 5466
networkBandwidthChangeAllowed(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
5467
    virNetworkObjPtr obj = NULL;
5468 5469 5470 5471 5472 5473
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    bool ret = false;

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

5474 5475
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
5476 5477 5478 5479 5480 5481
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return false;
    }

5482
    if (networkCheckBandwidth(obj, newBandwidth, ifaceBand, iface->mac, NULL) < 0)
5483 5484 5485 5486 5487
        goto cleanup;

    ret = true;

 cleanup:
5488
    virNetworkObjEndAPI(&obj);
5489 5490
    return ret;
}
5491 5492


5493
static int
5494 5495 5496 5497
networkBandwidthUpdate(virDomainNetDefPtr iface,
                       virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
5498
    virNetworkObjPtr obj = NULL;
5499
    virNetworkDefPtr def;
5500
    unsigned long long tmp_floor_sum;
5501 5502 5503 5504 5505 5506 5507 5508
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    unsigned long long new_rate = 0;
    int plug_ret;
    int ret = -1;

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

5509 5510
    obj = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!obj) {
5511 5512 5513 5514 5515
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return ret;
    }
5516
    def = virNetworkObjGetDef(obj);
5517

5518
    if ((plug_ret = networkCheckBandwidth(obj, newBandwidth, ifaceBand,
5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531
                                          iface->mac, &new_rate)) < 0) {
        /* helper reported error */
        goto cleanup;
    }

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

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

5532
    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
5533
        newBandwidth->in && newBandwidth->in->floor) {
5534 5535
        /* Either we just need to update @floor .. */

5536
        if (virNetDevBandwidthUpdateRate(def->bridge,
5537
                                         iface->data.network.actual->class_id,
5538
                                         def->bandwidth,
5539 5540 5541
                                         newBandwidth->in->floor) < 0)
            goto cleanup;

5542 5543 5544 5545 5546
        tmp_floor_sum = virNetworkObjGetFloorSum(obj);
        tmp_floor_sum -= ifaceBand->in->floor;
        tmp_floor_sum += newBandwidth->in->floor;
        virNetworkObjSetFloorSum(obj, tmp_floor_sum);
        new_rate -= tmp_floor_sum;
5547

5548 5549
        if (virNetDevBandwidthUpdateRate(def->bridge, 2,
                                         def->bandwidth, new_rate) < 0 ||
5550
            virNetworkObjSaveStatus(driver->stateDir, obj) < 0) {
5551
            /* Ouch, rollback */
5552 5553 5554
            tmp_floor_sum -= newBandwidth->in->floor;
            tmp_floor_sum += ifaceBand->in->floor;
            virNetworkObjSetFloorSum(obj, tmp_floor_sum);
5555

5556
            ignore_value(virNetDevBandwidthUpdateRate(def->bridge,
5557
                                                      iface->data.network.actual->class_id,
5558
                                                      def->bandwidth,
5559 5560 5561 5562 5563 5564
                                                      ifaceBand->in->floor));
            goto cleanup;
        }
    } else if (newBandwidth->in && newBandwidth->in->floor) {
        /* .. or we need to plug in new .. */

5565
        if (networkPlugBandwidthImpl(obj, iface, newBandwidth, new_rate) < 0)
5566 5567 5568 5569
            goto cleanup;
    } else {
        /* .. or unplug old. */

5570
        if (networkUnplugBandwidth(obj, iface) < 0)
5571 5572 5573 5574 5575
            goto cleanup;
    }

    ret = 0;
 cleanup:
5576
    virNetworkObjEndAPI(&obj);
5577 5578
    return ret;
}
5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606


static virNetworkDriver networkDriver = {
    .name = "bridge",
    .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 */
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
    .networkUndefine = networkUndefine, /* 0.2.0 */
    .networkUpdate = networkUpdate, /* 0.10.2 */
    .networkCreate = networkCreate, /* 0.2.0 */
    .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 */
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
};

5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623

static virHypervisorDriver networkHypervisorDriver = {
    .name = "network",
    .connectOpen = networkConnectOpen, /* 4.1.0 */
    .connectClose = networkConnectClose, /* 4.1.0 */
    .connectIsEncrypted = networkConnectIsEncrypted, /* 4.1.0 */
    .connectIsSecure = networkConnectIsSecure, /* 4.1.0 */
    .connectIsAlive = networkConnectIsAlive, /* 4.1.0 */
};


static virConnectDriver networkConnectDriver = {
    .hypervisorDriver = &networkHypervisorDriver,
    .networkDriver = &networkDriver,
};


5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634
static virStateDriver networkStateDriver = {
    .name = "bridge",
    .stateInitialize  = networkStateInitialize,
    .stateAutoStart  = networkStateAutoStart,
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
};

int
networkRegister(void)
{
5635 5636
    if (virRegisterConnectDriver(&networkConnectDriver, false) < 0)
        return -1;
5637 5638 5639 5640
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
        return -1;
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
5641 5642 5643 5644

    virDomainNetSetDeviceImpl(
        networkAllocateActualDevice,
        networkNotifyActualDevice,
5645 5646
        networkReleaseActualDevice,
        networkBandwidthChangeAllowed,
5647
        networkBandwidthUpdate);
5648

5649 5650
    return 0;
}