virnetdevtap.c 27.1 KB
Newer Older
1
/*
E
Eric Blake 已提交
2
 * Copyright (C) 2007-2014 Red Hat, Inc.
3 4 5 6 7 8 9 10 11 12 13 14
 *
 * 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
15
 * License along with this library.  If not, see
O
Osier Yang 已提交
16
 * <http://www.gnu.org/licenses/>.
17 18 19 20
 */

#include <config.h>

21
#include "virmacaddr.h"
22 23 24
#include "virnetdevtap.h"
#include "virnetdev.h"
#include "virnetdevbridge.h"
25
#include "virnetdevmidonet.h"
A
Ansis Atteka 已提交
26
#include "virnetdevopenvswitch.h"
27
#include "virerror.h"
28
#include "virfile.h"
29
#include "viralloc.h"
30
#include "virlog.h"
31
#include "virstring.h"
32
#include "datatypes.h"
33

34
#include <unistd.h>
35
#include <sys/types.h>
36
#include <sys/ioctl.h>
37
#include <net/if.h>
38 39 40
#include <fcntl.h>
#ifdef __linux__
# include <linux/if_tun.h>    /* IFF_TUN, IFF_NO_PI */
41
#elif defined(__FreeBSD__)
42 43
# include <net/if_mib.h>
# include <sys/sysctl.h>
44
#endif
45 46 47
#if defined(HAVE_GETIFADDRS) && defined(AF_LINK)
# include <ifaddrs.h>
#endif
48 49 50

#define VIR_FROM_THIS VIR_FROM_NONE

51 52
VIR_LOG_INIT("util.netdevtap");

53 54 55 56 57 58 59 60 61 62 63
/**
 * virNetDevTapGetName:
 * @tapfd: a tun/tap file descriptor
 * @ifname: a pointer that will receive the interface name
 *
 * Retrieve the interface name given a file descriptor for a tun/tap
 * interface.
 *
 * Returns 0 if the interface name is successfully queried, -1 otherwise
 */
int
J
Ján Tomko 已提交
64
virNetDevTapGetName(int tapfd G_GNUC_UNUSED, char **ifname G_GNUC_UNUSED)
65 66 67 68 69 70 71 72 73 74
{
#ifdef TUNGETIFF
    struct ifreq ifr;

    if (ioctl(tapfd, TUNGETIFF, &ifr) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to query tap interface name"));
        return -1;
    }

75 76
    *ifname = g_strdup(ifr.ifr_name);
    return 0;
77 78 79 80 81
#else
    return -1;
#endif
}

82 83 84 85 86 87 88 89 90 91 92 93 94
/**
 * virNetDevTapGetRealDeviceName:
 * @ifname: the interface name
 *
 * Lookup real interface name (i.e. name of the device entry in /dev),
 * because e.g. on FreeBSD if we rename tap device to vnetN its device
 * entry still remains unchanged (/dev/tapX), but bhyve needs a name
 * that matches /dev entry.
 *
 * Returns the proper interface name or NULL if no corresponding interface
 * found.
 */
char*
J
Ján Tomko 已提交
95
virNetDevTapGetRealDeviceName(char *ifname G_GNUC_UNUSED)
96
{
97 98 99 100
#ifdef IFDATA_DRIVERNAME
    int ifindex = 0;
    int name[6];
    size_t len = 0;
101 102
    char *ret = NULL;

103 104 105 106
    if ((ifindex = if_nametoindex(ifname)) == 0) {
        virReportSystemError(errno,
                             _("Unable to get interface index for '%s'"),
                             ifname);
107
        return NULL;
108 109 110 111 112 113 114 115 116 117 118 119 120 121 122
    }

    name[0] = CTL_NET;
    name[1] = PF_LINK;
    name[2] = NETLINK_GENERIC;
    name[3] = IFMIB_IFDATA;
    name[4] = ifindex;
    name[5] = IFDATA_DRIVERNAME;

    if (sysctl(name, 6, NULL, &len, 0, 0) < 0) {
        virReportSystemError(errno,
                             _("Unable to get driver name for '%s'"),
                             ifname);
        return NULL;
    }
123

124 125 126 127 128 129 130 131 132
    if (VIR_ALLOC_N(ret, len) < 0)
        return NULL;

    if (sysctl(name, 6, ret, &len, 0, 0) < 0) {
        virReportSystemError(errno,
                             _("Unable to get driver name for '%s'"),
                             ifname);
        VIR_FREE(ret);
        return NULL;
133 134 135 136 137 138 139 140
    }

    return ret;
#else
    return NULL;
#endif
}

141

142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203
/**
 * virNetDevProbeVnetHdr:
 * @tapfd: a tun/tap file descriptor
 *
 * Check whether it is safe to enable the IFF_VNET_HDR flag on the
 * tap interface.
 *
 * Setting IFF_VNET_HDR enables QEMU's virtio_net driver to allow
 * guests to pass larger (GSO) packets, with partial checksums, to
 * the host. This greatly increases the achievable throughput.
 *
 * It is only useful to enable this when we're setting up a virtio
 * interface. And it is only *safe* to enable it when we know for
 * sure that a) qemu has support for IFF_VNET_HDR and b) the running
 * kernel implements the TUNGETIFF ioctl(), which qemu needs to query
 * the supplied tapfd.
 *
 * Returns 1 if VnetHdr is supported, 0 if not supported
 */
#ifdef IFF_VNET_HDR
static int
virNetDevProbeVnetHdr(int tapfd)
{
# if defined(IFF_VNET_HDR) && defined(TUNGETFEATURES) && defined(TUNGETIFF)
    unsigned int features;
    struct ifreq dummy;

    if (ioctl(tapfd, TUNGETFEATURES, &features) != 0) {
        VIR_INFO("Not enabling IFF_VNET_HDR; "
                 "TUNGETFEATURES ioctl() not implemented");
        return 0;
    }

    if (!(features & IFF_VNET_HDR)) {
        VIR_INFO("Not enabling IFF_VNET_HDR; "
                 "TUNGETFEATURES ioctl() reports no IFF_VNET_HDR");
        return 0;
    }

    /* The kernel will always return -1 at this point.
     * If TUNGETIFF is not implemented then errno == EBADFD.
     */
    if (ioctl(tapfd, TUNGETIFF, &dummy) != -1 || errno != EBADFD) {
        VIR_INFO("Not enabling IFF_VNET_HDR; "
                 "TUNGETIFF ioctl() not implemented");
        return 0;
    }

    VIR_INFO("Enabling IFF_VNET_HDR");

    return 1;
# else
    (void) tapfd;
    VIR_INFO("Not enabling IFF_VNET_HDR; disabled at build time");
    return 0;
# endif
}
#endif


#ifdef TUNSETIFF
/**
204
 * virNetDevTapCreate:
205
 * @ifname: the interface name
206
 * @tunpath: path to the tun device (if NULL, /dev/net/tun is used)
207 208
 * @tapfds: array of file descriptors return value for the new tap device
 * @tapfdSize: number of file descriptors in @tapfd
209 210 211 212
 * @flags: OR of virNetDevTapCreateFlags. Only one flag is recognized:
 *
 *   VIR_NETDEV_TAP_CREATE_VNET_HDR
 *     - Enable IFF_VNET_HDR on the tap device
213 214
 *   VIR_NETDEV_TAP_CREATE_PERSIST
 *     - The device will persist after the file descriptor is closed
215
 *
216 217 218 219
 * Creates a tap interface. The caller must use virNetDevTapDelete to
 * remove a persistent TAP device when it is no longer needed. In case
 * @tapfdSize is greater than one, multiqueue extension is requested
 * from kernel.
220
 *
221
 * Returns 0 in case of success or -1 on failure.
222 223
 */
int virNetDevTapCreate(char **ifname,
224
                       const char *tunpath,
225
                       int *tapfd,
226
                       size_t tapfdSize,
227
                       unsigned int flags)
228
{
229
    size_t i;
230 231
    struct ifreq ifr;
    int ret = -1;
232
    int fd;
233

234 235 236
    if (!tunpath)
        tunpath = "/dev/net/tun";

237
    memset(&ifr, 0, sizeof(ifr));
238
    for (i = 0; i < tapfdSize; i++) {
239 240 241 242
        if ((fd = open(tunpath, O_RDWR)) < 0) {
            virReportSystemError(errno,
                                 _("Unable to open %s, is tun module loaded?"),
                                 tunpath);
243 244 245 246
            goto cleanup;
        }

        memset(&ifr, 0, sizeof(ifr));
247

248 249
        ifr.ifr_flags = IFF_TAP | IFF_NO_PI;
        /* If tapfdSize is greater than one, request multiqueue */
250 251
        if (tapfdSize > 1) {
# ifdef IFF_MULTI_QUEUE
252
            ifr.ifr_flags |= IFF_MULTI_QUEUE;
253 254 255 256 257 258
# else
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Multiqueue devices are not supported on this system"));
            goto cleanup;
# endif
        }
259 260

# ifdef IFF_VNET_HDR
261 262 263
        if ((flags &  VIR_NETDEV_TAP_CREATE_VNET_HDR) &&
            virNetDevProbeVnetHdr(fd))
            ifr.ifr_flags |= IFF_VNET_HDR;
264 265
# endif

266
        if (virStrcpyStatic(ifr.ifr_name, *ifname) < 0) {
267 268 269 270
            virReportSystemError(ERANGE,
                                 _("Network interface name '%s' is too long"),
                                 *ifname);
            goto cleanup;
271

272
        }
273

274 275 276 277 278 279
        if (ioctl(fd, TUNSETIFF, &ifr) < 0) {
            virReportSystemError(errno,
                                 _("Unable to create tap device %s"),
                                 NULLSTR(*ifname));
            goto cleanup;
        }
280

281 282 283 284
        if (i == 0) {
            /* In case we are looping more than once, set other
             * TAPs to have the same name */
            VIR_FREE(*ifname);
285
            *ifname = g_strdup(ifr.ifr_name);
286
        }
287

288
        if ((flags & VIR_NETDEV_TAP_CREATE_PERSIST) &&
289
            ioctl(fd, TUNSETPERSIST, 1) < 0) {
290 291 292 293 294 295
            virReportSystemError(errno,
                                 _("Unable to set tap device %s to persistent"),
                                 NULLSTR(*ifname));
            goto cleanup;
        }
        tapfd[i] = fd;
296 297 298 299
    }

    ret = 0;

300
 cleanup:
301
    if (ret < 0) {
302
        VIR_FORCE_CLOSE(fd);
303 304 305
        while (i--)
            VIR_FORCE_CLOSE(tapfd[i]);
    }
306 307 308 309 310

    return ret;
}


311 312
int virNetDevTapDelete(const char *ifname,
                       const char *tunpath)
313 314 315 316 317
{
    struct ifreq try;
    int fd;
    int ret = -1;

318 319 320 321 322 323 324
    if (!tunpath)
        tunpath = "/dev/net/tun";

    if ((fd = open(tunpath, O_RDWR)) < 0) {
        virReportSystemError(errno,
                             _("Unable to open %s, is tun module loaded?"),
                             tunpath);
325 326 327 328 329 330
        return -1;
    }

    memset(&try, 0, sizeof(struct ifreq));
    try.ifr_flags = IFF_TAP|IFF_NO_PI;

331
    if (virStrcpyStatic(try.ifr_name, ifname) < 0) {
332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351
        virReportSystemError(ERANGE,
                             _("Network interface name '%s' is too long"),
                             ifname);
        goto cleanup;
    }

    if (ioctl(fd, TUNSETIFF, &try) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to associate TAP device"));
        goto cleanup;
    }

    if (ioctl(fd, TUNSETPERSIST, 0) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to make TAP device non-persistent"));
        goto cleanup;
    }

    ret = 0;

352
 cleanup:
353 354 355
    VIR_FORCE_CLOSE(fd);
    return ret;
}
356
#elif defined(SIOCIFCREATE2) && defined(SIOCIFDESTROY) && defined(IF_MAXUNIT)
357
int virNetDevTapCreate(char **ifname,
J
Ján Tomko 已提交
358
                       const char *tunpath G_GNUC_UNUSED,
359
                       int *tapfd,
360
                       size_t tapfdSize,
J
Ján Tomko 已提交
361
                       unsigned int flags G_GNUC_UNUSED)
362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392
{
    int s;
    struct ifreq ifr;
    int ret = -1;
    char *newifname = NULL;

    if (tapfdSize > 1) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Multiqueue devices are not supported on this system"));
        goto cleanup;
    }

    /* As FreeBSD determines interface type by name,
     * we have to create 'tap' interface first and
     * then rename it to 'vnet'
     */
    if ((s = virNetDevSetupControl("tap", &ifr)) < 0)
        return -1;

    if (ioctl(s, SIOCIFCREATE2, &ifr) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to create tap device"));
        goto cleanup;
    }

    /* In case we were given exact interface name (e.g. 'vnetN'),
     * we just rename to it. If we have format string like
     * 'vnet%d', we need to find the first available name that
     * matches this pattern
     */
    if (strstr(*ifname, "%d") != NULL) {
393
        size_t i;
394
        for (i = 0; i <= IF_MAXUNIT; i++) {
395
            g_autofree char *newname = NULL;
396

397
            newname = g_strdup_printf(*ifname, i);
398 399

            if (virNetDevExists(newname) == 0) {
400
                newifname = g_steal_pointer(&newname);
401 402 403 404 405 406 407 408 409 410 411 412 413 414 415
                break;
            }
        }
        if (newifname) {
            VIR_FREE(*ifname);
            *ifname = newifname;
        } else {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Failed to generate new name for interface %s"),
                           ifr.ifr_name);
            goto cleanup;
        }
    }

    if (tapfd) {
416
        g_autofree char *dev_path = NULL;
417
        dev_path = g_strdup_printf("/dev/%s", ifr.ifr_name);
418 419 420 421 422 423 424 425 426

        if ((*tapfd = open(dev_path, O_RDWR)) < 0) {
            virReportSystemError(errno,
                                 _("Unable to open %s"),
                                 dev_path);
            goto cleanup;
        }
    }

427
    if (virNetDevSetName(ifr.ifr_name, *ifname) == -1)
428 429 430 431
        goto cleanup;


    ret = 0;
432
 cleanup:
433 434 435 436 437
    VIR_FORCE_CLOSE(s);

    return ret;
}

438
int virNetDevTapDelete(const char *ifname,
J
Ján Tomko 已提交
439
                       const char *tunpath G_GNUC_UNUSED)
440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455
{
    int s;
    struct ifreq ifr;
    int ret = -1;

    if ((s = virNetDevSetupControl(ifname, &ifr)) < 0)
        return -1;

    if (ioctl(s, SIOCIFDESTROY, &ifr) < 0) {
        virReportSystemError(errno,
                             _("Unable to remove tap device %s"),
                             ifname);
        goto cleanup;
    }

    ret = 0;
456
 cleanup:
457 458 459 460 461
    VIR_FORCE_CLOSE(s);
    return ret;
}

#else
J
Ján Tomko 已提交
462 463 464 465 466
int virNetDevTapCreate(char **ifname G_GNUC_UNUSED,
                       const char *tunpath G_GNUC_UNUSED,
                       int *tapfd G_GNUC_UNUSED,
                       size_t tapfdSize G_GNUC_UNUSED,
                       unsigned int flags G_GNUC_UNUSED)
467 468 469 470 471
{
    virReportSystemError(ENOSYS, "%s",
                         _("Unable to create TAP devices on this platform"));
    return -1;
}
J
Ján Tomko 已提交
472 473
int virNetDevTapDelete(const char *ifname G_GNUC_UNUSED,
                       const char *tunpath G_GNUC_UNUSED)
474 475 476 477 478
{
    virReportSystemError(ENOSYS, "%s",
                         _("Unable to delete TAP devices on this platform"));
    return -1;
}
479
#endif
480 481


482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501
/**
 * virNetDevTapAttachBridge:
 * @tapname: the tap interface name (or name template)
 * @brname: the bridge name
 * @macaddr: desired MAC address
 * @virtPortProfile: bridge/port specific configuration
 * @virtVlan: vlan tag info
 * @mtu: requested MTU for port (or 0 for "default")
 * @actualMTU: MTU actually set for port (after accounting for bridge's MTU)
 *
 * This attaches an existing tap device (@tapname) to a bridge
 * (@brname).
 *
 * Returns 0 in case of success or -1 on failure
 */
int
virNetDevTapAttachBridge(const char *tapname,
                         const char *brname,
                         const virMacAddr *macaddr,
                         const unsigned char *vmuuid,
502
                         const virNetDevVPortProfile *virtPortProfile,
503
                         const virNetDevVlan *virtVlan,
504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552
                         unsigned int mtu,
                         unsigned int *actualMTU)
{
    /* If an MTU is specified for the new device, set it before
     * attaching the device to the bridge, as it may affect the MTU of
     * the bridge (in particular if it is the first device attached to
     * the bridge, or if it is smaller than the current MTU of the
     * bridge). If MTU isn't specified for the new device (i.e. 0),
     * we need to set the interface MTU to the current MTU of the
     * bridge (to avoid inadvertantly changing the bridge's MTU).
     */
    if (mtu > 0) {
        if (virNetDevSetMTU(tapname, mtu) < 0)
            goto error;
    } else {
        if (virNetDevSetMTUFromDevice(tapname, brname) < 0)
            goto error;
    }
    if (actualMTU) {
        int retMTU = virNetDevGetMTU(tapname);

        if (retMTU < 0)
            goto error;

        *actualMTU = retMTU;
    }


    if (virtPortProfile) {
        if (virtPortProfile->virtPortType == VIR_NETDEV_VPORT_PROFILE_MIDONET) {
            if (virNetDevMidonetBindPort(tapname, virtPortProfile) < 0)
                goto error;
        } else if (virtPortProfile->virtPortType == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
            if (virNetDevOpenvswitchAddPort(brname, tapname, macaddr, vmuuid,
                                            virtPortProfile, virtVlan) < 0)
                goto error;
        }
    } else {
        if (virNetDevBridgeAddPort(brname, tapname) < 0)
            goto error;
    }

    return 0;

 error:
    return -1;
}


553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573
/**
 * virNetDevTapReattachBridge:
 * @tapname: the tap interface name (or name template)
 * @brname: the bridge name
 * @macaddr: desired MAC address
 * @virtPortProfile: bridge/port specific configuration
 * @virtVlan: vlan tag info
 * @mtu: requested MTU for port (or 0 for "default")
 * @actualMTU: MTU actually set for port (after accounting for bridge's MTU)
 *
 * Ensures that the tap device (@tapname) is connected to the bridge
 * (@brname), potentially removing it from any existing bridge that
 * does not match.
 *
 * Returns 0 in case of success or -1 on failure
 */
int
virNetDevTapReattachBridge(const char *tapname,
                           const char *brname,
                           const virMacAddr *macaddr,
                           const unsigned char *vmuuid,
574
                           const virNetDevVPortProfile *virtPortProfile,
575
                           const virNetDevVlan *virtVlan,
576 577 578 579
                           unsigned int mtu,
                           unsigned int *actualMTU)
{
    bool useOVS = false;
580
    g_autofree char *master = NULL;
581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619

    if (virNetDevGetMaster(tapname, &master) < 0)
        return -1;

    /* IFLA_MASTER for a tap on an OVS switch is always "ovs-system" */
    if (STREQ_NULLABLE(master, "ovs-system")) {
        useOVS = true;
        if (virNetDevOpenvswitchInterfaceGetMaster(tapname, &master) < 0)
            return -1;
    }

    /* Nothing more todo if we're on the right bridge already */
    if (STREQ_NULLABLE(brname, master))
        return 0;

    /* disconnect from current (incorrect) bridge, if any  */
    if (master) {
        int ret;
        VIR_INFO("Removing %s from %s", tapname, master);
        if (useOVS)
            ret = virNetDevOpenvswitchRemovePort(master, tapname);
        else
            ret = virNetDevBridgeRemovePort(master, tapname);
        if (ret < 0)
            return -1;
    }

    VIR_INFO("Attaching %s to %s", tapname, brname);
    if (virNetDevTapAttachBridge(tapname, brname,
                                 macaddr, vmuuid,
                                 virtPortProfile,
                                 virtVlan,
                                 mtu, actualMTU) < 0)
        return -1;

    return 0;
}


620 621 622 623
/**
 * virNetDevTapCreateInBridgePort:
 * @brname: the bridge name
 * @ifname: the interface name (or name template)
624
 * @macaddr: desired MAC address
625
 * @tunpath: path to the tun device (if NULL, /dev/net/tun is used)
626 627
 * @tapfd: array of file descriptor return value for the new tap device
 * @tapfdSize: number of file descriptors in @tapfd
A
Ansis Atteka 已提交
628
 * @virtPortProfile: bridge/port specific configuration
629
 * @coalesce: optional coalesce parameters
630 631
 * @mtu: requested MTU for port (or 0 for "default")
 * @actualMTU: MTU actually set for port (after accounting for bridge's MTU)
632 633 634 635 636 637 638 639
 * @flags: OR of virNetDevTapCreateFlags:

 *   VIR_NETDEV_TAP_CREATE_IFUP
 *     - Bring the interface up
 *   VIR_NETDEV_TAP_CREATE_VNET_HDR
 *     - Enable IFF_VNET_HDR on the tap device
 *   VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE
 *     - Set this interface's MAC as the bridge's MAC address
640 641
 *   VIR_NETDEV_TAP_CREATE_PERSIST
 *     - The device will persist after the file descriptor is closed
642 643 644 645
 *
 * This function creates a new tap device on a bridge. @ifname can be either
 * a fixed name or a name template with '%d' for dynamic name allocation.
 * in either case the final name for the bridge will be stored in @ifname.
646 647 648 649
 * If the @tapfd parameter is supplied, the open tap device file descriptor
 * will be returned, otherwise the TAP device will be closed. The caller must
 * use virNetDevTapDelete to remove a persistent TAP device when it is no
 * longer needed.
650 651 652 653 654
 *
 * Returns 0 in case of success or -1 on failure
 */
int virNetDevTapCreateInBridgePort(const char *brname,
                                   char **ifname,
655
                                   const virMacAddr *macaddr,
656
                                   const unsigned char *vmuuid,
657
                                   const char *tunpath,
A
Ansis Atteka 已提交
658
                                   int *tapfd,
659
                                   size_t tapfdSize,
660
                                   const virNetDevVPortProfile *virtPortProfile,
661
                                   const virNetDevVlan *virtVlan,
662
                                   virNetDevCoalescePtr coalesce,
663 664
                                   unsigned int mtu,
                                   unsigned int *actualMTU,
665
                                   unsigned int flags)
666
{
667
    virMacAddr tapmac;
668
    size_t i;
669

670
    if (virNetDevTapCreate(ifname, tunpath, tapfd, tapfdSize, flags) < 0)
671 672 673 674 675 676 677 678
        return -1;

    /* We need to set the interface MAC before adding it
     * to the bridge, because the bridge assumes the lowest
     * MAC of all enslaved interfaces & we don't want it
     * seeing the kernel allocate random MAC for the TAP
     * device before we set our static MAC.
     */
679
    virMacAddrSet(&tapmac, macaddr);
680
    if (!(flags & VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE)) {
681 682 683 684 685 686 687 688 689 690 691 692
        /* The tap device's MAC address cannot match the MAC address
         * used by the guest. This results in "received packet on
         * vnetX with own address as source address" error logs from
         * the kernel. Making the tap address as high as possible
         * discourages the bridge from using this tap's MAC as its own
         * (a Linux host bridge will take on the lowest numbered MAC
         * of all devices attached to it).
         */
        if (tapmac.addr[0] == 0xFE)
            tapmac.addr[0] = 0xFA;
        else
            tapmac.addr[0] = 0xFE;
693
    }
694

695
    if (virNetDevSetMAC(*ifname, &tapmac) < 0)
696 697
        goto error;

698 699 700
    if (virNetDevTapAttachBridge(*ifname, brname, macaddr, vmuuid,
                                 virtPortProfile, virtVlan, mtu, actualMTU) < 0) {
        goto error;
A
Ansis Atteka 已提交
701
    }
702

703
    if (virNetDevSetOnline(*ifname, !!(flags & VIR_NETDEV_TAP_CREATE_IFUP)) < 0)
704 705
        goto error;

706
    if (virNetDevSetCoalesce(*ifname, coalesce, false) < 0)
707 708
        goto error;

709 710
    return 0;

711
 error:
712 713
    for (i = 0; i < tapfdSize && tapfd[i] >= 0; i++)
        VIR_FORCE_CLOSE(tapfd[i]);
714

715
    return -1;
716
}
717 718

/*-------------------- interface stats --------------------*/
719 720 721 722 723 724 725 726 727 728 729 730 731 732 733

/**
 * virNetDevTapInterfaceStats:
 * @ifname: interface
 * @stats: where to store statistics
 * @swapped: whether to swap RX/TX fields
 *
 * Fetch RX/TX statistics for given named interface (@ifname) and
 * store them at @stats. The returned statistics are always from
 * domain POV. Because in some cases this means swapping RX/TX in
 * the stats and in others this means no swapping (consider TAP
 * vs macvtap) caller might choose if the returned stats should
 * be @swapped or not.
 *
 * Returns 0 on success, -1 otherwise (with error reported).
734 735 736 737
 */
#ifdef __linux__
int
virNetDevTapInterfaceStats(const char *ifname,
738 739
                           virDomainInterfaceStatsPtr stats,
                           bool swapped)
740 741 742 743 744
{
    int ifname_len;
    FILE *fp;
    char line[256], *colon;

745 746 747 748 749 750
    if (!ifname) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Interface name not provided"));
        return -1;
    }

751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782
    fp = fopen("/proc/net/dev", "r");
    if (!fp) {
        virReportSystemError(errno, "%s",
                             _("Could not open /proc/net/dev"));
        return -1;
    }

    ifname_len = strlen(ifname);

    while (fgets(line, sizeof(line), fp)) {
        long long dummy;
        long long rx_bytes;
        long long rx_packets;
        long long rx_errs;
        long long rx_drop;
        long long tx_bytes;
        long long tx_packets;
        long long tx_errs;
        long long tx_drop;

        /* The line looks like:
         *   "   eth0:..."
         * Split it at the colon.
         */
        colon = strchr(line, ':');
        if (!colon) continue;
        *colon = '\0';
        if (colon-ifname_len >= line &&
            STREQ(colon-ifname_len, ifname)) {
            if (sscanf(colon+1,
                       "%lld %lld %lld %lld %lld %lld %lld %lld %lld %lld %lld %lld %lld %lld %lld %lld",
                       &rx_bytes, &rx_packets, &rx_errs, &rx_drop,
783 784
                       &dummy, &dummy, &dummy, &dummy,
                       &tx_bytes, &tx_packets, &tx_errs, &tx_drop,
785 786 787
                       &dummy, &dummy, &dummy, &dummy) != 16)
                continue;

788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806
            if (swapped) {
                stats->rx_bytes = tx_bytes;
                stats->rx_packets = tx_packets;
                stats->rx_errs = tx_errs;
                stats->rx_drop = tx_drop;
                stats->tx_bytes = rx_bytes;
                stats->tx_packets = rx_packets;
                stats->tx_errs = rx_errs;
                stats->tx_drop = rx_drop;
            } else {
                stats->rx_bytes = rx_bytes;
                stats->rx_packets = rx_packets;
                stats->rx_errs = rx_errs;
                stats->rx_drop = rx_drop;
                stats->tx_bytes = tx_bytes;
                stats->tx_packets = tx_packets;
                stats->tx_errs = tx_errs;
                stats->tx_drop = tx_drop;
            }
807

808
            VIR_FORCE_FCLOSE(fp);
809 810 811 812 813 814 815 816 817 818 819 820
            return 0;
        }
    }
    VIR_FORCE_FCLOSE(fp);

    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                   _("/proc/net/dev: Interface not found"));
    return -1;
}
#elif defined(HAVE_GETIFADDRS) && defined(AF_LINK)
int
virNetDevTapInterfaceStats(const char *ifname,
821 822
                           virDomainInterfaceStatsPtr stats,
                           bool swapped)
823 824 825 826 827
{
    struct ifaddrs *ifap, *ifa;
    struct if_data *ifd;
    int ret = -1;

828 829 830 831 832 833
    if (!ifname) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Interface name not provided"));
        return -1;
    }

834 835 836 837 838 839 840
    if (getifaddrs(&ifap) < 0) {
        virReportSystemError(errno, "%s",
                             _("Could not get interface list"));
        return -1;
    }

    for (ifa = ifap; ifa; ifa = ifa->ifa_next) {
841 842 843
        if (!ifa->ifa_addr)
            continue;

844 845 846 847 848
        if (ifa->ifa_addr->sa_family != AF_LINK)
            continue;

        if (STREQ(ifa->ifa_name, ifname)) {
            ifd = (struct if_data *)ifa->ifa_data;
849 850 851 852 853 854 855 856
            if (swapped) {
                stats->tx_bytes = ifd->ifi_ibytes;
                stats->tx_packets = ifd->ifi_ipackets;
                stats->tx_errs = ifd->ifi_ierrors;
                stats->tx_drop = ifd->ifi_iqdrops;
                stats->rx_bytes = ifd->ifi_obytes;
                stats->rx_packets = ifd->ifi_opackets;
                stats->rx_errs = ifd->ifi_oerrors;
857
# ifdef HAVE_STRUCT_IF_DATA_IFI_OQDROPS
858
                stats->rx_drop = ifd->ifi_oqdrops;
859
# else
860
                stats->rx_drop = 0;
861
# endif
862 863 864 865 866 867 868 869 870 871 872 873 874 875
            } else {
                stats->tx_bytes = ifd->ifi_obytes;
                stats->tx_packets = ifd->ifi_opackets;
                stats->tx_errs = ifd->ifi_oerrors;
# ifdef HAVE_STRUCT_IF_DATA_IFI_OQDROPS
                stats->tx_drop = ifd->ifi_oqdrops;
# else
                stats->tx_drop = 0;
# endif
                stats->rx_bytes = ifd->ifi_ibytes;
                stats->rx_packets = ifd->ifi_ipackets;
                stats->rx_errs = ifd->ifi_ierrors;
                stats->rx_drop = ifd->ifi_iqdrops;
            }
876 877 878 879 880 881 882 883 884 885 886 887 888 889 890

            ret = 0;
            break;
        }
    }

    if (ret < 0)
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Interface not found"));

    freeifaddrs(ifap);
    return ret;
}
#else
int
J
Ján Tomko 已提交
891 892 893
virNetDevTapInterfaceStats(const char *ifname G_GNUC_UNUSED,
                           virDomainInterfaceStatsPtr stats G_GNUC_UNUSED,
                           bool swapped G_GNUC_UNUSED)
894 895 896 897 898 899 900
{
    virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                   _("interface stats not implemented on this platform"));
    return -1;
}

#endif /* __linux__ */