libvirt/src/util/virnetdevtap.c

507 lines
15 KiB
C
Raw Normal View History

/*
* Copyright (C) 2007-2013 Red Hat, Inc.
*
* 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
* License along with this library. If not, see
* <http://www.gnu.org/licenses/>.
*
* Authors:
* Mark McLoughlin <markmc@redhat.com>
* Daniel P. Berrange <berrange@redhat.com>
*/
#include <config.h>
util: centralize tap device MAC address 1st byte "0xFE" modification When a tap device for a domain is created and attached to a bridge, the first byte of the tap device MAC address is set to 0xFE, while the rest is set to match the MAC address that will be presented to the guest as its network device MAC address. Setting this high value in the tap's MAC address discourages the bridge from using the tap device's MAC address as the bridge's own MAC address (Linux bridges always take on the lowest numbered MAC address of all attached devices as their own). In one case within libvirt, a tap device is created and attached to the bridge with the intent that its MAC address be taken on by the bridge as its own (this is used to assure that the bridge has a fixed MAC address to prevent network outages created by the bridge MAC address "flapping" as guests are started and stopped). In this case, the first byte of the mac address is *not* altered to 0xFE. In the current code, callers to virNetDevTapCreateInBridgePort each make the MAC address modification themselves before calling, which leads to code duplication, and also prevents lower level functions from knowing the real MAC address being used by the guest. The problem here is that openvswitch bridges must be informed about this MAC address, or they will be unable to pass traffic to/from the guest. This patch centralizes the location of the MAC address "0xFE fixup" into virNetDevTapCreateInBridgePort(), meaning 1) callers of this function no longer need the extra strange bit of code, and 2) bitNetDevTapCreateBridgeInPort itself now is called with the guest's unaltered MAC address, and can pass it on, unmodified, to virNetDevOpenvswitchAddPort. There is no other behavioral change created by this patch.
2012-02-16 23:49:57 +00:00
#include "virmacaddr.h"
#include "virnetdevtap.h"
#include "virnetdev.h"
#include "virnetdevbridge.h"
#include "virnetdevopenvswitch.h"
#include "virerror.h"
#include "virfile.h"
2012-12-12 18:06:53 +00:00
#include "viralloc.h"
2012-12-12 17:59:27 +00:00
#include "virlog.h"
#include "virstring.h"
#include <sys/ioctl.h>
#include <net/if.h>
#include <fcntl.h>
#ifdef __linux__
# include <linux/if_tun.h> /* IFF_TUN, IFF_NO_PI */
#endif
#define VIR_FROM_THIS VIR_FROM_NONE
/**
* 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
virNetDevTapGetName(int tapfd ATTRIBUTE_UNUSED, char **ifname ATTRIBUTE_UNUSED)
{
#ifdef TUNGETIFF
struct ifreq ifr;
if (ioctl(tapfd, TUNGETIFF, &ifr) < 0) {
virReportSystemError(errno, "%s",
_("Unable to query tap interface name"));
return -1;
}
return VIR_STRDUP(*ifname, ifr.ifr_name) < 0 ? -1 : 0;
#else
return -1;
#endif
}
/**
* 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
/**
* virNetDevTapCreate:
* @ifname: the interface name
* @tapfds: array of file descriptors return value for the new tap device
* @tapfdSize: number of file descriptors in @tapfd
* @flags: OR of virNetDevTapCreateFlags. Only one flag is recognized:
*
* VIR_NETDEV_TAP_CREATE_VNET_HDR
* - Enable IFF_VNET_HDR on the tap device
network: fix dnsmasq/radvd binding to IPv6 on recent kernels I hit this problem recently when trying to create a bridge with an IPv6 address on a 3.2 kernel: dnsmasq (and, further, radvd) would not bind to the given address, waiting 20s and then giving up with -EADDRNOTAVAIL (resp. exiting immediately with "error parsing or activating the config file", without libvirt noticing it, BTW). This can be reproduced with (I think) any kernel >= 2.6.39 and the following XML (to be used with "virsh net-create"): <network> <name>test-bridge</name> <bridge name='testbr0' /> <ip family='ipv6' address='fd00::1' prefix='64'> </ip> </network> (it happens even when you have an IPv4, too) The problem is that since commit [1] (which, ironically, was made to “help IPv6 autoconfiguration”) the linux bridge code makes bridges behave like “real” devices regarding carrier detection. This makes the bridges created by libvirt, which are started without any up devices, stay with the NO-CARRIER flag set, and thus prevents DAD (Duplicate address detection) from happening, thus letting the IPv6 address flagged as “tentative”. Such addresses cannot be bound to (see RFC 2462), so dnsmasq fails binding to it (for radvd, it detects that "interface XXX is not RUNNING", thus that "interface XXX does not exist, ignoring the interface" (sic)). It seems that this behavior was enhanced somehow with commit [2] by avoiding setting NO-CARRIER on empty bridges, but I couldn't reproduce this behavior on my kernel. Anyway, with the “dummy tap to set MAC address” trick, this wouldn't work. To fix this, the idea is to get the bridge's attached device to be up so that DAD can happen (deactivating DAD altogether is not a good idea, I think). Currently, libvirt creates a dummy TAP device to set the MAC address of the bridge, keeping it down. But even if we set this device up, it is not RUNNING as soon as the tap file descriptor attached to it is closed, thus still preventing DAD. So, we must modify the API a bit, so that we can get the fd, keep the tap device persistent, run the daemons, and close it after DAD has taken place. After that, the bridge will be flagged NO-CARRIER again, but the daemons will be running, even if not happy about the device's state (but we don't really care about the bridge's daemons doing anything when no up interface is connected to it). Other solutions that I envisioned were: * Keeping the *-nic interface up: this would waste an fd for each bridge during all its life. May be acceptable, I don't really know. * Stop using the dummy tap trick, and set the MAC address directly on the bridge: it is possible since quite some time it seems, even if then there is the problem of the bridge not being RUNNING when empty, contrary to what [2] says, so this will need fixing (and this fix only happened in 3.1, so it wouldn't work for 2.6.39) * Using the --interface option of dnsmasq, but I saw somewhere that it's not used by libvirt for backward compatibility. I am not sure this would solve this problem, though, as I don't know how dnsmasq binds itself to it with this option. This is why this patch does what's described earlier. This patch also makes radvd start even if the interface is “missing” (i.e. it is not RUNNING), as it daemonizes before binding to it, and thus sometimes does it after the interface has been brought down by us (by closing the tap fd), and then originally stops. This also makes it stop yelling about it in the logs when the interface is down at a later time. [1] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=1faa4356a3bd89ea11fb92752d897cff3a20ec0e [2] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=b64b73d7d0c480f75684519c6134e79d50c1b341
2012-09-26 19:02:20 +00:00
* VIR_NETDEV_TAP_CREATE_PERSIST
* - The device will persist after the file descriptor is closed
*
* 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.
*
* Returns 0 in case of success or -1 on failure.
*/
int virNetDevTapCreate(char **ifname,
int *tapfd,
int tapfdSize,
network: fix dnsmasq/radvd binding to IPv6 on recent kernels I hit this problem recently when trying to create a bridge with an IPv6 address on a 3.2 kernel: dnsmasq (and, further, radvd) would not bind to the given address, waiting 20s and then giving up with -EADDRNOTAVAIL (resp. exiting immediately with "error parsing or activating the config file", without libvirt noticing it, BTW). This can be reproduced with (I think) any kernel >= 2.6.39 and the following XML (to be used with "virsh net-create"): <network> <name>test-bridge</name> <bridge name='testbr0' /> <ip family='ipv6' address='fd00::1' prefix='64'> </ip> </network> (it happens even when you have an IPv4, too) The problem is that since commit [1] (which, ironically, was made to “help IPv6 autoconfiguration”) the linux bridge code makes bridges behave like “real” devices regarding carrier detection. This makes the bridges created by libvirt, which are started without any up devices, stay with the NO-CARRIER flag set, and thus prevents DAD (Duplicate address detection) from happening, thus letting the IPv6 address flagged as “tentative”. Such addresses cannot be bound to (see RFC 2462), so dnsmasq fails binding to it (for radvd, it detects that "interface XXX is not RUNNING", thus that "interface XXX does not exist, ignoring the interface" (sic)). It seems that this behavior was enhanced somehow with commit [2] by avoiding setting NO-CARRIER on empty bridges, but I couldn't reproduce this behavior on my kernel. Anyway, with the “dummy tap to set MAC address” trick, this wouldn't work. To fix this, the idea is to get the bridge's attached device to be up so that DAD can happen (deactivating DAD altogether is not a good idea, I think). Currently, libvirt creates a dummy TAP device to set the MAC address of the bridge, keeping it down. But even if we set this device up, it is not RUNNING as soon as the tap file descriptor attached to it is closed, thus still preventing DAD. So, we must modify the API a bit, so that we can get the fd, keep the tap device persistent, run the daemons, and close it after DAD has taken place. After that, the bridge will be flagged NO-CARRIER again, but the daemons will be running, even if not happy about the device's state (but we don't really care about the bridge's daemons doing anything when no up interface is connected to it). Other solutions that I envisioned were: * Keeping the *-nic interface up: this would waste an fd for each bridge during all its life. May be acceptable, I don't really know. * Stop using the dummy tap trick, and set the MAC address directly on the bridge: it is possible since quite some time it seems, even if then there is the problem of the bridge not being RUNNING when empty, contrary to what [2] says, so this will need fixing (and this fix only happened in 3.1, so it wouldn't work for 2.6.39) * Using the --interface option of dnsmasq, but I saw somewhere that it's not used by libvirt for backward compatibility. I am not sure this would solve this problem, though, as I don't know how dnsmasq binds itself to it with this option. This is why this patch does what's described earlier. This patch also makes radvd start even if the interface is “missing” (i.e. it is not RUNNING), as it daemonizes before binding to it, and thus sometimes does it after the interface has been brought down by us (by closing the tap fd), and then originally stops. This also makes it stop yelling about it in the logs when the interface is down at a later time. [1] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=1faa4356a3bd89ea11fb92752d897cff3a20ec0e [2] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=b64b73d7d0c480f75684519c6134e79d50c1b341
2012-09-26 19:02:20 +00:00
unsigned int flags)
{
size_t i;
struct ifreq ifr;
int ret = -1;
int fd;
memset(&ifr, 0, sizeof(ifr));
for (i = 0; i < tapfdSize; i++) {
if ((fd = open("/dev/net/tun", O_RDWR)) < 0) {
virReportSystemError(errno, "%s",
_("Unable to open /dev/net/tun, is tun module loaded?"));
goto cleanup;
}
memset(&ifr, 0, sizeof(ifr));
ifr.ifr_flags = IFF_TAP | IFF_NO_PI;
/* If tapfdSize is greater than one, request multiqueue */
if (tapfdSize > 1) {
# ifdef IFF_MULTI_QUEUE
ifr.ifr_flags |= IFF_MULTI_QUEUE;
# else
virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
_("Multiqueue devices are not supported on this system"));
goto cleanup;
# endif
}
# ifdef IFF_VNET_HDR
if ((flags & VIR_NETDEV_TAP_CREATE_VNET_HDR) &&
virNetDevProbeVnetHdr(fd))
ifr.ifr_flags |= IFF_VNET_HDR;
# endif
if (virStrcpyStatic(ifr.ifr_name, *ifname) == NULL) {
virReportSystemError(ERANGE,
_("Network interface name '%s' is too long"),
*ifname);
goto cleanup;
}
if (ioctl(fd, TUNSETIFF, &ifr) < 0) {
virReportSystemError(errno,
_("Unable to create tap device %s"),
NULLSTR(*ifname));
goto cleanup;
}
if (i == 0) {
/* In case we are looping more than once, set other
* TAPs to have the same name */
VIR_FREE(*ifname);
if (VIR_STRDUP(*ifname, ifr.ifr_name) < 0)
goto cleanup;
}
if ((flags & VIR_NETDEV_TAP_CREATE_PERSIST) &&
(errno = ioctl(fd, TUNSETPERSIST, 1))) {
virReportSystemError(errno,
_("Unable to set tap device %s to persistent"),
NULLSTR(*ifname));
goto cleanup;
}
tapfd[i] = fd;
}
ret = 0;
cleanup:
if (ret < 0) {
VIR_FORCE_CLOSE(fd);
while (i--)
VIR_FORCE_CLOSE(tapfd[i]);
}
return ret;
}
int virNetDevTapDelete(const char *ifname)
{
struct ifreq try;
int fd;
int ret = -1;
if ((fd = open("/dev/net/tun", O_RDWR)) < 0) {
virReportSystemError(errno, "%s",
_("Unable to open /dev/net/tun, is tun module loaded?"));
return -1;
}
memset(&try, 0, sizeof(struct ifreq));
try.ifr_flags = IFF_TAP|IFF_NO_PI;
if (virStrcpyStatic(try.ifr_name, ifname) == NULL) {
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;
cleanup:
VIR_FORCE_CLOSE(fd);
return ret;
}
#elif defined(SIOCIFCREATE2) && defined(SIOCIFDESTROY) && defined(IF_MAXUNIT)
int virNetDevTapCreate(char **ifname,
int *tapfd,
int tapfdSize,
unsigned int flags ATTRIBUTE_UNUSED)
{
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) {
size_t i;
for (i = 0; i <= IF_MAXUNIT; i++) {
char *newname;
if (virAsprintf(&newname, *ifname, i) < 0)
goto cleanup;
if (virNetDevExists(newname) == 0) {
newifname = newname;
break;
}
VIR_FREE(newname);
}
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) {
char *dev_path = NULL;
if (virAsprintf(&dev_path, "/dev/%s", ifr.ifr_name) < 0)
goto cleanup;
if ((*tapfd = open(dev_path, O_RDWR)) < 0) {
virReportSystemError(errno,
_("Unable to open %s"),
dev_path);
VIR_FREE(dev_path);
goto cleanup;
}
VIR_FREE(dev_path);
}
if (virNetDevSetName(ifr.ifr_name, *ifname) == -1) {
goto cleanup;
}
ret = 0;
cleanup:
VIR_FORCE_CLOSE(s);
return ret;
}
int virNetDevTapDelete(const char *ifname)
{
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;
cleanup:
VIR_FORCE_CLOSE(s);
return ret;
}
#else
int virNetDevTapCreate(char **ifname ATTRIBUTE_UNUSED,
int *tapfd ATTRIBUTE_UNUSED,
int tapfdSize ATTRIBUTE_UNUSED,
unsigned int flags ATTRIBUTE_UNUSED)
{
virReportSystemError(ENOSYS, "%s",
_("Unable to create TAP devices on this platform"));
return -1;
}
int virNetDevTapDelete(const char *ifname ATTRIBUTE_UNUSED)
{
virReportSystemError(ENOSYS, "%s",
_("Unable to delete TAP devices on this platform"));
return -1;
}
#endif
/**
* virNetDevTapCreateInBridgePort:
* @brname: the bridge name
* @ifname: the interface name (or name template)
* @macaddr: desired MAC address
* @tapfd: array of file descriptor return value for the new tap device
* @tapfdSize: number of file descriptors in @tapfd
* @virtPortProfile: bridge/port specific configuration
* @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
network: fix dnsmasq/radvd binding to IPv6 on recent kernels I hit this problem recently when trying to create a bridge with an IPv6 address on a 3.2 kernel: dnsmasq (and, further, radvd) would not bind to the given address, waiting 20s and then giving up with -EADDRNOTAVAIL (resp. exiting immediately with "error parsing or activating the config file", without libvirt noticing it, BTW). This can be reproduced with (I think) any kernel >= 2.6.39 and the following XML (to be used with "virsh net-create"): <network> <name>test-bridge</name> <bridge name='testbr0' /> <ip family='ipv6' address='fd00::1' prefix='64'> </ip> </network> (it happens even when you have an IPv4, too) The problem is that since commit [1] (which, ironically, was made to “help IPv6 autoconfiguration”) the linux bridge code makes bridges behave like “real” devices regarding carrier detection. This makes the bridges created by libvirt, which are started without any up devices, stay with the NO-CARRIER flag set, and thus prevents DAD (Duplicate address detection) from happening, thus letting the IPv6 address flagged as “tentative”. Such addresses cannot be bound to (see RFC 2462), so dnsmasq fails binding to it (for radvd, it detects that "interface XXX is not RUNNING", thus that "interface XXX does not exist, ignoring the interface" (sic)). It seems that this behavior was enhanced somehow with commit [2] by avoiding setting NO-CARRIER on empty bridges, but I couldn't reproduce this behavior on my kernel. Anyway, with the “dummy tap to set MAC address” trick, this wouldn't work. To fix this, the idea is to get the bridge's attached device to be up so that DAD can happen (deactivating DAD altogether is not a good idea, I think). Currently, libvirt creates a dummy TAP device to set the MAC address of the bridge, keeping it down. But even if we set this device up, it is not RUNNING as soon as the tap file descriptor attached to it is closed, thus still preventing DAD. So, we must modify the API a bit, so that we can get the fd, keep the tap device persistent, run the daemons, and close it after DAD has taken place. After that, the bridge will be flagged NO-CARRIER again, but the daemons will be running, even if not happy about the device's state (but we don't really care about the bridge's daemons doing anything when no up interface is connected to it). Other solutions that I envisioned were: * Keeping the *-nic interface up: this would waste an fd for each bridge during all its life. May be acceptable, I don't really know. * Stop using the dummy tap trick, and set the MAC address directly on the bridge: it is possible since quite some time it seems, even if then there is the problem of the bridge not being RUNNING when empty, contrary to what [2] says, so this will need fixing (and this fix only happened in 3.1, so it wouldn't work for 2.6.39) * Using the --interface option of dnsmasq, but I saw somewhere that it's not used by libvirt for backward compatibility. I am not sure this would solve this problem, though, as I don't know how dnsmasq binds itself to it with this option. This is why this patch does what's described earlier. This patch also makes radvd start even if the interface is “missing” (i.e. it is not RUNNING), as it daemonizes before binding to it, and thus sometimes does it after the interface has been brought down by us (by closing the tap fd), and then originally stops. This also makes it stop yelling about it in the logs when the interface is down at a later time. [1] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=1faa4356a3bd89ea11fb92752d897cff3a20ec0e [2] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=b64b73d7d0c480f75684519c6134e79d50c1b341
2012-09-26 19:02:20 +00:00
* VIR_NETDEV_TAP_CREATE_PERSIST
* - The device will persist after the file descriptor is closed
*
* 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.
network: fix dnsmasq/radvd binding to IPv6 on recent kernels I hit this problem recently when trying to create a bridge with an IPv6 address on a 3.2 kernel: dnsmasq (and, further, radvd) would not bind to the given address, waiting 20s and then giving up with -EADDRNOTAVAIL (resp. exiting immediately with "error parsing or activating the config file", without libvirt noticing it, BTW). This can be reproduced with (I think) any kernel >= 2.6.39 and the following XML (to be used with "virsh net-create"): <network> <name>test-bridge</name> <bridge name='testbr0' /> <ip family='ipv6' address='fd00::1' prefix='64'> </ip> </network> (it happens even when you have an IPv4, too) The problem is that since commit [1] (which, ironically, was made to “help IPv6 autoconfiguration”) the linux bridge code makes bridges behave like “real” devices regarding carrier detection. This makes the bridges created by libvirt, which are started without any up devices, stay with the NO-CARRIER flag set, and thus prevents DAD (Duplicate address detection) from happening, thus letting the IPv6 address flagged as “tentative”. Such addresses cannot be bound to (see RFC 2462), so dnsmasq fails binding to it (for radvd, it detects that "interface XXX is not RUNNING", thus that "interface XXX does not exist, ignoring the interface" (sic)). It seems that this behavior was enhanced somehow with commit [2] by avoiding setting NO-CARRIER on empty bridges, but I couldn't reproduce this behavior on my kernel. Anyway, with the “dummy tap to set MAC address” trick, this wouldn't work. To fix this, the idea is to get the bridge's attached device to be up so that DAD can happen (deactivating DAD altogether is not a good idea, I think). Currently, libvirt creates a dummy TAP device to set the MAC address of the bridge, keeping it down. But even if we set this device up, it is not RUNNING as soon as the tap file descriptor attached to it is closed, thus still preventing DAD. So, we must modify the API a bit, so that we can get the fd, keep the tap device persistent, run the daemons, and close it after DAD has taken place. After that, the bridge will be flagged NO-CARRIER again, but the daemons will be running, even if not happy about the device's state (but we don't really care about the bridge's daemons doing anything when no up interface is connected to it). Other solutions that I envisioned were: * Keeping the *-nic interface up: this would waste an fd for each bridge during all its life. May be acceptable, I don't really know. * Stop using the dummy tap trick, and set the MAC address directly on the bridge: it is possible since quite some time it seems, even if then there is the problem of the bridge not being RUNNING when empty, contrary to what [2] says, so this will need fixing (and this fix only happened in 3.1, so it wouldn't work for 2.6.39) * Using the --interface option of dnsmasq, but I saw somewhere that it's not used by libvirt for backward compatibility. I am not sure this would solve this problem, though, as I don't know how dnsmasq binds itself to it with this option. This is why this patch does what's described earlier. This patch also makes radvd start even if the interface is “missing” (i.e. it is not RUNNING), as it daemonizes before binding to it, and thus sometimes does it after the interface has been brought down by us (by closing the tap fd), and then originally stops. This also makes it stop yelling about it in the logs when the interface is down at a later time. [1] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=1faa4356a3bd89ea11fb92752d897cff3a20ec0e [2] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=b64b73d7d0c480f75684519c6134e79d50c1b341
2012-09-26 19:02:20 +00:00
* 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.
*
* Returns 0 in case of success or -1 on failure
*/
int virNetDevTapCreateInBridgePort(const char *brname,
char **ifname,
const virMacAddrPtr macaddr,
const unsigned char *vmuuid,
int *tapfd,
int tapfdSize,
virNetDevVPortProfilePtr virtPortProfile,
virNetDevVlanPtr virtVlan,
unsigned int flags)
{
virMacAddr tapmac;
char macaddrstr[VIR_MAC_STRING_BUFLEN];
size_t i;
util: centralize tap device MAC address 1st byte "0xFE" modification When a tap device for a domain is created and attached to a bridge, the first byte of the tap device MAC address is set to 0xFE, while the rest is set to match the MAC address that will be presented to the guest as its network device MAC address. Setting this high value in the tap's MAC address discourages the bridge from using the tap device's MAC address as the bridge's own MAC address (Linux bridges always take on the lowest numbered MAC address of all attached devices as their own). In one case within libvirt, a tap device is created and attached to the bridge with the intent that its MAC address be taken on by the bridge as its own (this is used to assure that the bridge has a fixed MAC address to prevent network outages created by the bridge MAC address "flapping" as guests are started and stopped). In this case, the first byte of the mac address is *not* altered to 0xFE. In the current code, callers to virNetDevTapCreateInBridgePort each make the MAC address modification themselves before calling, which leads to code duplication, and also prevents lower level functions from knowing the real MAC address being used by the guest. The problem here is that openvswitch bridges must be informed about this MAC address, or they will be unable to pass traffic to/from the guest. This patch centralizes the location of the MAC address "0xFE fixup" into virNetDevTapCreateInBridgePort(), meaning 1) callers of this function no longer need the extra strange bit of code, and 2) bitNetDevTapCreateBridgeInPort itself now is called with the guest's unaltered MAC address, and can pass it on, unmodified, to virNetDevOpenvswitchAddPort. There is no other behavioral change created by this patch.
2012-02-16 23:49:57 +00:00
if (virNetDevTapCreate(ifname, tapfd, tapfdSize, flags) < 0)
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.
*/
virMacAddrSet(&tapmac, macaddr);
if (!(flags & VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE)) {
if (macaddr->addr[0] == 0xFE) {
/* For normal use, 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.
*/
virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
_("Unable to use MAC address starting with "
"reserved value 0xFE - '%s' - "),
virMacAddrFormat(macaddr, macaddrstr));
goto error;
}
tapmac.addr[0] = 0xFE; /* Discourage bridge from using TAP dev MAC */
}
util: centralize tap device MAC address 1st byte "0xFE" modification When a tap device for a domain is created and attached to a bridge, the first byte of the tap device MAC address is set to 0xFE, while the rest is set to match the MAC address that will be presented to the guest as its network device MAC address. Setting this high value in the tap's MAC address discourages the bridge from using the tap device's MAC address as the bridge's own MAC address (Linux bridges always take on the lowest numbered MAC address of all attached devices as their own). In one case within libvirt, a tap device is created and attached to the bridge with the intent that its MAC address be taken on by the bridge as its own (this is used to assure that the bridge has a fixed MAC address to prevent network outages created by the bridge MAC address "flapping" as guests are started and stopped). In this case, the first byte of the mac address is *not* altered to 0xFE. In the current code, callers to virNetDevTapCreateInBridgePort each make the MAC address modification themselves before calling, which leads to code duplication, and also prevents lower level functions from knowing the real MAC address being used by the guest. The problem here is that openvswitch bridges must be informed about this MAC address, or they will be unable to pass traffic to/from the guest. This patch centralizes the location of the MAC address "0xFE fixup" into virNetDevTapCreateInBridgePort(), meaning 1) callers of this function no longer need the extra strange bit of code, and 2) bitNetDevTapCreateBridgeInPort itself now is called with the guest's unaltered MAC address, and can pass it on, unmodified, to virNetDevOpenvswitchAddPort. There is no other behavioral change created by this patch.
2012-02-16 23:49:57 +00:00
if (virNetDevSetMAC(*ifname, &tapmac) < 0)
goto error;
/* We need to set the interface MTU before adding it
* to the bridge, because the bridge will have its
* MTU adjusted automatically when we add the new interface.
*/
if (virNetDevSetMTUFromDevice(*ifname, brname) < 0)
goto error;
if (virtPortProfile) {
if (virNetDevOpenvswitchAddPort(brname, *ifname, macaddr, vmuuid,
virtPortProfile, virtVlan) < 0) {
goto error;
}
} else {
if (virNetDevBridgeAddPort(brname, *ifname) < 0)
goto error;
}
if (virNetDevSetOnline(*ifname, !!(flags & VIR_NETDEV_TAP_CREATE_IFUP)) < 0)
goto error;
return 0;
error:
for (i = 0; i < tapfdSize && tapfd[i] >= 0; i++)
VIR_FORCE_CLOSE(tapfd[i]);
return -1;
}