IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
This is necessary for non-ipv4ll hosts to communicate with ipv4ll-only hosts on the same link. Defaults
to being enabled, but can be opted out.
See: <http://avahi.org/wiki/AvahiAutoipd#Routes>
This avoids having to distinguish between IPv4 and IPv6, allowing us
to keep their internal orderings. The consumers now has to turn the
strings into addresses.
All routes added by networkd are currently set RTPROT_BOOT, which according
to the kernel means "Route installed during boot" (rtnetlink.h). But this
is not always the case as networkd changes routing after boot too. Since
the kernel gives more detailed protocols, use them.
With this patch, user-configured static routes now use RTPROT_STATIC (which
they are) and DHCP routes use RTPROT_DHCP. There is no define for IPv4LL
yet, so those are installed as RTPROT_STATIC (though perhaps RTPROT_RA is
better?).
[tomegun: fixup
src/network/networkd-link.c:972:33: error: too few arguments to function 'route_new_dynamic']
Lennart said:
> We have these nice USEC_PER_MSEC-style macro definitions which make it a
> little bit clearer what we are converting here from what into
> what... please use that instead of writing "1000"...
>
> (we stole those from gstreamer btw)
The following bond options are supported by this patch.
MIIMonitorSec:
Specifies the frequency in milli-seconds that MII link
monitoring will occur.
UpDelaySec:
Specifies the delay time in milli-seconds to enable a link
after a link up status has been detected.
DownDelaySec:
Specifies the delay time in milli-seconds to disable a link
after a link failure has been detected.
changes:
1. Added gconf variables.
2. man page
conf:
[NetDev]
Name=bond1
Kind=bond
[Bond]
Mode=802.3ad
TransmitHashPolicy=layer2+3
LacpduTransmitRate=fast
MIIMonitorSec=1s
UpDelaySec=2s
DownDelaySec=8s
cat /proc/net/bonding/bond1
Ethernet Channel Bonding Driver: v3.7.1 (April 27, 2011)
Bonding Mode: IEEE 802.3ad Dynamic link aggregation
Transmit Hash Policy: layer2+3 (2)
MII Status: up
MII Polling Interval (ms): 1000
Up Delay (ms): 2000
Down Delay (ms): 8000
802.3ad info
LACP rate: fast
Min links: 0
Aggregator selection policy (ad_select): stable
bond bond1 has no active aggregator
[tomegun: rephrased manpage, dropped bond_ prefix from variables]
LacpduTransmitRate
option specifies the rate in which link partner to transmit
LACPDU packets in 802.3ad mode. Possible values
slow : Request partner to transmit LACPDUs every 30 seconds
fast : Request partner to transmit LACPDUs every 1 second
The default is slow.
chages:
1. Added enum bond_lacp_rate_table
2. gperf LacpduTransmitRate
Test:
conf file:
[NetDev]
Name=bond1
Kind=bond
[Bond]
Mode=802.3ad
LacpduTransmitRate=fast
test:
cat /proc/net/bonding/bond1
Ethernet Channel Bonding Driver: v3.7.1 (April 27, 2011)
Bonding Mode: IEEE 802.3ad Dynamic link aggregation
Transmit Hash Policy: layer2+3 (2)
MII Status: up
MII Polling Interval (ms): 0
Up Delay (ms): 0
Down Delay (ms): 0
802.3ad info
LACP rate: fast
Min links: 0
Aggregator selection policy (ad_select): stable
bond bond1 has no active aggregator
[tomegun: renamed from LacpduTransmitRate to LACPTransmitRate, manpage fixes and
dropped bond_ prefix from variables]
This patch adds support the transmit hash policy to use
for slave selection in balance-xor, 802.3ad, and tlb modes
layer2, layer3+4, layer2+3, encap3+4, encap3+4
Added:
1. BondXmitHashPolicy
2. conf param TransmitHashPolicy
Test conf:
[NetDev]
Name=bond1
Kind=bond
[Bond]
Mode=802.3ad
TransmitHashPolicy=layer2+3
test output:
cat /proc/net/bonding/bond1
Ethernet Channel Bonding Driver: v3.7.1 (April 27, 2011)
Bonding Mode: IEEE 802.3ad Dynamic link aggregation
Transmit Hash Policy: layer2+3 (2)
MII Status: up
MII Polling Interval (ms): 0
Up Delay (ms): 0
Down Delay (ms): 0
802.3ad info
LACP rate: slow
Min links: 0
Aggregator selection policy (ad_select): stable
bond bond1 has no active aggregator
[tomegun: dropped bond_ prefix from new Bond variable, drop repeated man-page section]
Let's settle on a single type for all address family values, even if
UNIX is very inconsitent on the precise type otherwise. Given that
socket() is the primary entrypoint for the sockets API, and that uses
"int", and "int" is relatively simple and generic, we settle on "int"
for this.
Do not expose link_is_loopback, people should just get this from rtnl directly.
Do not expose NTP servers as IP addresses, these must be strings.
Expose ifindex as int, not unsigned. This is what the kernel (mostly) and glibc uses.
Rather than refetching the link information on ever event, we liston to
rtnl to track them. Much code stolen from resolved.
This will allow us to simplify the sd-network api and don't expose
information available over rtnl.
As long as the number of array entries is relatively small it's nicer to
simply return the number of entries directly, instead of using a size_t*
return parameter for it.
Constructors should return the object they created as first parameter,
except when they are generated as a child/member object of some other
object in which case that should be first.
Special care is needed so that we get an error message if the
file failed to parse, but not when it is missing. To avoid duplicating
the same error check in every caller, add an additional 'warn' boolean
to tell config_parse whether a message should be issued.
This makes things both shorter and more robust wrt. to error reporting.
It appears there is no good way to decide whether or not broadcasts should be enabled,
there is hardware that must have broadcast, and there are networks that only allow
unicast. So we give up and make this configurable.
By default, unicast is used, but if the kernel were to inform us abotu certain
interfaces requiring broadcast, we could change this to opt-in by default in
those cases.
Vendor Class Identifier be used by DHCP clients to identify
their vendor type and configuration. When using this option,
vendors can define their own specific identifier values, such
as to convey a particular hardware or operating system
configuration or other identifying information.
Vendor-specified DHCP options—features that let administrators assign
separate options to clients with similar configuration requirements.
For example, if DHCP-aware clients for example we want to separate
different gateway and option for different set of people
(dev/test/hr/finance) in a org or devices for example web/database
servers or let's say in a embedded device etc and require a different
default gateway or DNS server than the rest of clients.
Now route metric can be configuted via conf file:
example conf:
[Match]
Name=em1
[Route]
Gateway=192.168.1.12
Metric=10
Test:
ip route output
default via 192.168.1.12 dev em1 metric 10
[tomegun: squash TODO update and reword man page a bit]
This patch adds supports networkd to configure bond mode
during creation via persistent conf. Mode can be configured
with conf param 'Mode'. A new section Bond is added to the
conf to support bond mode.
These modes can be configured now.
balance-rr
active-backup
balance-xor
broadcast
802.3ad
balance-tlb
balance-alb
Example conf file: test-bond.conf
[NetDev]
Name=bond1
Kind=bond
[Bond]
Mode=balance-xor
Test case:
1. start networkd service:
12: bond1: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc
noqueue state UNKNOWN mode DEFAULT group default
link/ether 22:89:6c:47:23:d2 brd ff:ff:ff:ff:ff:ff
2. find bond mode:
cat /proc/net/bonding/bond1
Ethernet Channel Bonding Driver: v3.7.1 (April 27, 2011)
Bonding Mode: load balancing (xor)
Transmit Hash Policy: layer2 (0)
MII Status: up
MII Polling Interval (ms): 0
Up Delay (ms): 0
Down Delay (ms): 0
Changes:
1. Added file networkd-bond.c
2. Bond mode enum BondMode
3. conf section [Bond]
[tomegun: whitespace]
This patch adds peer address support for
networkd . In the [Address] a new configurable
param is Peer.
[Match]
Name=ipip-tun
[Address]
Address=10.0.0.1/32
Peer=10.0.0.2/32
When doing a NEWADDR, the reply we get back is the NEWADDR itself, rather
than just an empty ack (unlike how NEWLINK works). For this reason, the
process that did the NEWADDR does not get the broadcast message.
We were only listening for broadcast messages, and hence not tracking the
addresses we added ourselves. This went unnoticed as the kernel will usually
send NEWADDR messages from time to time anyway, so things would mostly work,
but in the worst case we would not notice that a routable address was available
and consider ourselves offline.
This patch introduces TUN/TAP device creation support
to networkd.
Example conf to create a tap device:
file: tap.netdev
------------------
[NetDev]
Name=tap-test
Kind=tap
[Tap]
OneQueue=true
MultiQueue=true
PacketInfo=true
User=sus
Group=sus
------------------
Test:
1. output of ip link
tap-test: tap pi one_queue UNKNOWN_FLAGS:900 user 1000 group 1000
id:
uid=1000(sus) gid=10(wheel) groups=10(wheel),1000(sus)
context=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023
Modifications:
Added:
1. file networkd-tuntap.c
3. netdev kind NETDEV_KIND_TUN and NETDEV_KIND_TAP
2. Tun and Tap Sections and config params to parse
conf and gperf conf parameters
[tomegun: tweak the 'kind' checking for received ifindex]
Send hostname (option 12) in DISCOVER and REQUEST messages so the
DHCP server could use it to register with dynamic DNS and such.
To opt-out of this behaviour set SendHostname to false in [DHCP]
section of .network file
[tomegun: rebased, made sure a failing set_hostname is a noop and moved
config from DHCPv4 to DHCP]
We failed to take a ref when waiting for udev synchronization. Fix that and also
make unreffing in callbacks simpler throughout by using _cleanup_ macros.
Fixes <https://bugs.freedesktop.org/show_bug.cgi?id=80556>.
The logic otherwise is that we leave anything preconfigured alone, but in the case of DHCP
we actually need to update it whenever the lease is renewed.
This adds support for DHCP options 33 and 121: Static Route and
Classless Static Route. To enable this feature, set UseRoutes=true
in .network file. Returned routes are added to the routing table.
If there are v4 or v6 specific options we can keep those in separate sections,
but for the common options, we will use only one.
Moreovere only use DHCP=[yes/both|no/none|v4|v6] to enable or disable the clients.
We used to check if e.g. IFLA_BOND_MAX is defined and provide fallback
values in missing.h is it wasn't. But over time, various kernel
versions added IFLA_* defines, so checking for IFLA_BOND_MAX is not
enough if the kernel is new enough to have some of them but too old to
have all. In case we detect that the latest known enum value is
missing, #define most of them.
https://bugs.freedesktop.org/show_bug.cgi?id=80095
When an address is configured to be all zeroes, networkd will now
automatically find a locally unused network of the right size from a
list of pre-configured pools. Currently those pools are 10.0.0.0/8,
172.16.0.0/12, 192.168.0.0/16 and fc00::/7, i.e. the network ranges for
private networks. They are compiled in, but should be configurable
eventually.
This allows applying the same configuration to a large number of
interfaces with each time a different IP range block, and management of
these IP ranges is fully automatic.
When allocating an address range from the pool it is made sure the range
is not used otherwise.
We could still have an old interface name and/or mac address when libudev
tells us that the device is initialized, as the up-to-date info could still
be on its way from the kernel.
This reverts (and rewrites) commit 7d95c772cb.
The issue blocking this feature has now been fixed in the kernel, and backported
to the various stable kernels.
Our netdevs will now have stable MAC addresses, even if one is not specified.
It may sometimes be necessary to specify the MAC address of a netdev.
Let us set the correct one from the get-go, rather than having the
kernel generate a random one, and then change it after.
It should not be possible to have a DHCP lease on a link without also having
an associated network. Add assert() to avoid compiler warnings.
Reported by Thomas H. P. Andersen
When enabled in [Network] it will set up a dhcp server on the interface, listening
on one of its statically configured IPv4 addresses and with a fixed size pool of
leases determined from it.
Example:
[Match]
Name=ve-arch-tree
[Network]
Address=192.168.12.5/24
DHCPServer=yes
[Route]
Gateway=192.168.12.5
Destination=192.168.12.0/24
In this case we will configure ve-arch-tree with the address 192.168.12.5 and
hand out addresses in the range 192.168.12.6 - 192.168.12.38.
In the future, we should (as suggested by Lennart) introduce a syntax to pick the
server address automatically.
Configuration will be in
root:root /run/systemd/network
and state will be in
systemd-network:systemd-network /run/systemd/netif
This matches what we do for logind's seat/session state.
Rely on modules being built-in or autoloaded on-demand.
As networkd is a network facing service, we want to limits its capabilities,
as much as possible. Also, we may not have CAP_SYS_MODULE in a container,
and we want networkd to work the same there.
Module autoloading does not always work, but should be fixed by the kernel
patch f98f89a0104454f35a: 'net: tunnels - enable module autoloading', which
is currently in net-next and which people may consider backporting if they
want tunneling support without compiling in the modules.
Early adopters may also use a module-load.d snippet and order
systemd-modules-load.service before networkd to force the module
loading of tunneling modules.
This sholud fix the various build issues people have reported.
This patch adds path of mtu discovery for sit tunnel.
To enable/disable DiscoverPathMTU is introduced.
Example configuration
file: sit.netdev
[NetDev]
Name=sit-tun
Kind=sit
MTUBytes=1480
[Tunnel]
DiscoverPathMTU=1
Local=X.X.X.X
Remote=X.X.X.X
By default pmtudisc is turned on , if DiscoverPathMTU
is missing from the config. To turn it off
DiscoverPathMTU=0 needs to be set.
This allows us to run networkd mostly unpriviliged with the exception of
CAP_NET_* and CAP_SYS_MODULE. I'd really like to get rid of the latter
though...
Djalal Harouni <tixxdz@opendz.org>:
There is also this one genrated by LTO, IMO it's a false positive since
we do *check* for "lease" but the code is not consistent since in that
code path, "lease" is initialized to NULL in other places, except for
this one:
src/resolve/resolved-manager.c: In function 'manager_update_resolv_conf':
src/libsystemd-network/sd-dhcp-lease.c:67:18: warning: 'lease' may be used uninitialized in this function [-Wmaybe-uninitialized]
if (lease->dns_size) {
^
src/network/sd-network.c:146:24: note: 'lease' was declared here
sd_dhcp_lease *lease;
^
This makes callback behaviour more like sd-event or sd-resolve, and
creates proper object for unregistering callbacks.
Taking the refernce to the slot is optional. If not taken life time of
the slot will be bound to the underlying bus object (or in the case of
an async call until the reply has been recieved).
The bitmask is deprecated in the kernel, so move to the new interface. At the moment
this does not make a difference for us, but it avoids having to change the API in the future.
Make the logging less verbose by only printing all the changed flags on one line,
at the same time make it more complete by supporting all flags currently supported
by the kernel.
We still fall back to printing the raw flags in case we get something we do not recognize
This may be useful when running on new kernels.
We need to take a refcount on the link whenever we expect a callback. The exceptions
are the ipv4ll/dhcp clients as their lifetimes are guaranteed to be shorter than that
of the link.
Otherwise:
eth0: unknown link flags gained: 0x00040 (ignoring)
[tomegun: hiding these messages is ok, as IFF_RUNNING is redundant
and can be deduced from operstate and IFF_LOWER_UP]
Clarify that we are waiting for any link to reach 'carrier' state, regardless
of who manages it. This will be useful when we add support for waiting for more
operational states.
To make sure we don't delay boot on systems where (some) network links are managed by someone else
we don't block if something else has successfully brought up a link.
We will still block until all links we are aware of that are managed by networkd have been
configured, but if no such links exist, and someone else have configured a link sufficiently
that it has a carrier, it may be that the link is ready so we should no longer block.
Note that in all likelyhood the link is not ready (no addresses/routes configured),
so whatever network managment daemon configured it should provide a similar wait-online
service to block network-online.target until it is ready.
The aim is to block as long as we know networking is not fully configured, but no longer. This
will allow systemd-networkd-wait-online.service to be enabled on any system, even if we don't
know whether networkd is the main/only network manager.
Even in the case networking is fully configured by networkd, the default behavior may not be
sufficient: if two links need to be configured, but the first is fully configured before the
second one appears we will assume the network is up. To work around that, we allow specifying
specific devices to wait for before considering the network up.
This unit is enabled by default, just like systemd-networkd, but will only be pulled in if
anyone pulls in network-online.target.
This properly detects the state of the link based on both the link flags and the
operstate.
Moreover, always log state-changes even if we are not yet managing the link.
Avoid having two code-paths racing with eacother to do the same thing. The change
of flags will be detected in the normal way, so only use the link_up_handler
to detect if the 'up' failed and in that case fail the link.
This reverts commit cdc85c875b.
There appears to be a kernel bug that (among other things) cause bridges
not to get a bridge id set when supplying IFLA_ADDRESS when creating the
netdev.
Simply revert the whole thing until we sort this out in the kernel.
See: http://www.spinics.net/lists/netdev/msg279807.html
Reported-by: C. R. Oldham <cr@saltstack.com>
We may receive RTM_NEWLINK messages with missing LINKINFO after the initial NEWLINK message,
don't bother verifying these, just drop out early after checking that the ifindex is not in conflict.
This essentially swaps the roles of rtnl and udev in networkd. After this
change libudev is only used for waiting for udev to initialize devices and
to get udev-specific information needed for some [Match] attributes.
This in particular simplifies the code in containers where udev is not really
useful, but also simplifies things and reduces round-trips in the non-container
case.
Usually RUNNING implies LOWER_UP, but for drivers that don't support oper state, RUNNING can
also mean that the state is unknown. In that case we should just trust LOWER_UP directly.
The interface is not fully ready until it enterns RUNNING. This was causing
problems with sending out DHCP messages before the interface was ready, so they
would get lost. In particular this affected DHCP INIT-REBOOT, as it relies on
the first package sent being successful (or it will fall back to a full reboot).
Also improve the logging a lot, to make future debugging of link state a lot
easier.
This error should never happen, so replace the check with an assert. The check
was anyway broken due to an uninitialized return value.
Reported by Thomas Hindoe Paaboel Andersen <phomes@gmail.com>.
Similar to DHCP, the IPv4LL library user can decide to free the LL
client any time the callback is called. Guard against freeing the
LL client in the callback by introducing proper reference counting.
Also update code using the IPv4LL library to properly handle a
returned NULL from the notify and stop functions if the IPv4LL
client was freed.
The DHCP library user can decide to free the DHCP client any time
the callback is called. After the callback has been called, other
computations may still be needed - the best example being a full
restart of the DHCP procedure in case of lease expiry.
Fix this by introducing proper reference counting. Properly handle
a returned NULL from the notify and stop functions if the DHCP
client was freed.
Currently when both ipv4ll and dhcp are enabled, ipv4ll
address (if one has been claimed) is removed when dhcp
address is aquired. This is not the best thing to do
since there might be clients unaware of the removal
trying to communicate.
This patch provides a smooth transition between ipv4ll
and dhcp. If ipv4ll address was claimed [1] before dhcp,
address is marked as deprecated. Deprecated address is still
a valid address and packets can be received on it but address
cannot be selected as a source address. If dhcp lease cannot
be extended, then ipv4ll address is marked as valid again.
[1] If there is no collision, claiming IPv4LL takes between 4 to
7 seconds.
Use a static table with all the typing information, rather than repeated
switch statements. This should make it a lot simpler to add new types.
We need to keep all the type info to be able to create containers
without exposing their implementation details to the users of the library.
As a freebee we verify the types of appended/read attributes.
The API is extended to nicely deal with unions of container types.
Firstly, remove stray assert(). Also be a bit stricter when verifying the
received info. If we get an applicable newlink message that we can't make
sense of, we will now enter NETDEV_FAILED, as we cannot reasonably continue
without knowing the ifindex of our device.
This does not belong in shared as it is mostly a detail of our networking subsystem.
Moreover, now we can use libudev here, which will simplify things.
Increase the chance of using the same link local address between reboots. The
pseudo random sequence of addresses we attempt is now seeded with data that is
very likely to stay the same between reboots, but at the same time be unique
to the specific machine/nic.
First we try to use the ID_NET_NAME_* data from the udev db combined with the
machin-id, which is guaranteed to be unique and persistent, if available. If
that is not possible (e.g., in containers where we don't have access to the
udev db) we fallback to using the MAC address of the interface, which is
guaranteed to be unique, and likely to be persistent.
[tomegun: three minor changes:
- don't expose HASH_KEY in the siphash24 header
- get rid of some compile-warnings (and some casts at the same time),
by using uint8_t[8] rather than uint64_t in the api
- added commit message]
Some DHCP servers gives you a netmask of 255.255.255.255 so the gateway is not
routable. Other DHCP client implementations look through the existing routes to
figure out if they should add an explicit host route. See below for a link.
However, it makes sense to just create the route explicitly whether it is
needed or not since it is explicit, makes the dhcp route entries independent of
other entries and saves us from knowing the state of the kernel tables.
After patch route table on a machine with a network (common case):
default via 10.0.2.2 dev ens3
10.0.2.0/24 dev ens3 proto kernel scope link src 10.0.2.15
10.0.2.2 dev ens3 scope link
After patch route table on a machine without a network (this case):
default via 10.240.0.1 dev ens4v1
10.240.0.1 dev ens4v1 scope link
The code from dhcpcd that works around this issue is on line 637.
https://android.googlesource.com/platform/external/dhcpcd/+/master/configure.c
Pass the mac address on to ipv4ll and dhcp clients so they always have
up-to-date information, and may react appropriately to the change.
Also drop setting the mac address from uevent, and only log when the
address actually changes.
safe_close() automatically becomes a NOP when a negative fd is passed,
and returns -1 unconditionally. This makes it easy to write lines like
this:
fd = safe_close(fd);
Which will close an fd if it is open, and reset the fd variable
correctly.
By making use of this new scheme we can drop a > 200 lines of code that
was required to test for non-negative fds or to reset the closed fd
variable afterwards.
Bring some arrays that are used for DEFINE_STRING_TABLE_LOOKUP() in the
same order than the enums they reference.
Also, pass the corresponding _MAX value to the array initalizer where
appropriate.
Bridges will change their MAC address when other devices are enslaved. We need
the correct MAC address to acquire a DHCP lease, so take note of it whenever
it changes.
Implements IPv4LL with respect to RFC 3927
(http://tools.ietf.org/rfc/rfc3927.txt) and integrates it
with networkd. Majority of the IPv4LL state machine is
taken from avahi (http://avahi.org/) project's autoip.
IPv4LL can be enabled by IPv4LL=yes under [Network]
section of .network file.
IPv4LL works independent of DHCP but if DHCP lease is
aquired, then LL address will be dropped.
[tomegun: removed a trailing newline and a compiler warning]
This is mostly a proof of concept to try sd-network, so we don't
hook it up with a .service file quite yet. We probably want it to
be more clever about deciding when we are 'online'.
The binary will wait for at least one network managed by networkd,
and until all networks managed by networkd are configured.
