This partially reverts 95355a281c.
It seems that other parts of link_stop_clients() should be skipped
when restarting, but I don't know enough about those other clients to have
an opinion if it is better to stop&start them on restart or not.
Anyway, that can be done in later patches now that the support for restarts
is there.
Fixes#13625.
When nspawn container with private network starts, networkd creates
the default route for the interface. The route may cause problem on
the host side, and it can be created with DefaultRouteOnDevice= now.
Hence, this makes networkd not create the route implicitly any more.
Closes#13418.
*We* control the sysctl setting. If the user configured IPv6, then we apply the
settings, and just make sure that at some point during the configuration the
sysctl is disabled (i.e. ipv6 enabled) if we have IPv6 configured.
Replaces #13283.
Otherwise, changing the default gateway doesn't purge old gateway routes
left on the system during daemon restart. This also fixes removing other
foreign gateway routes that don't match the expected configuration.
Tested:
Changed gateway addresses prior to the patch and they lingered on
the system during each reconfiguration. Applied this patch and
reconfigured gateways and other routes multiple times and it removed
the foreign routes that had gateways that didn't match.
Signed-off-by: William A. Kennington III <william@wkennington.com>
The flag IFA_F_NOPREFIXROUTE was introduced in kernel-3.14. But even if
the kernel does not support the flag, it should be just ignored. So, it
is not necessary to do the fallback logic. Moreover, the current logic
is not a fallback mechanism but just retrying. So, it should not work.
Let's drop that.
Some path of configuring address, route or etc., go to failed state, but
some do not. E.g., failure in address configuration which is provided by
DHCPv4 goes to failed state, but static address does not.
This is just for consistency. This should not change anything if
everything is fine.
This also voidify manager_rtnl_process_address().
When DHCP or any other dynamic addresses are changed, then
link_request_set_routes() is invoked, and even if no static routes
are configured, the operational state is needlessly changed to
configuring state.
This makes the state is changed only when static routes are configured.
Fixes#12931.
In file included from ../src/basic/macro.h:558,
from ../src/basic/alloc-util.h:9,
from ../src/network/networkd-link.c:7:
../src/network/networkd-link.c: In function ‘link_sysctl_ipv6_enabled’:
../src/basic/log.h:107:9: error: ‘%s’ directive argument is null [-Werror=format-overflow=]
107 | log_internal_realm(LOG_REALM_PLUS_LEVEL(LOG_REALM, (level)), __VA_ARGS__)
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../src/shared/log-link.h:21:25: note: in expansion of macro ‘log_internal’
21 | log_internal(level, error, _FILE_, __LINE__, __func__, ##__VA_ARGS__); \
| ^~~~~~~~~~~~
../src/shared/log-link.h:33:50: note: in expansion of macro ‘log_link_full’
33 | #define log_link_warning_errno(link, error, ...) log_link_full(link, LOG_WARNING, error, ##__VA_ARGS__)
| ^~~~~~~~~~~~~
../src/network/networkd-link.c:83:24: note: in expansion of macro ‘log_link_warning_errno’
83 | return log_link_warning_errno(link, r,
| ^~~~~~~~~~~~~~~~~~~~~~
../src/network/networkd-link.c:84:77: note: format string is defined here
84 | "Failed to read net.ipv6.conf.%s.disable_ipv6 sysctl property: %m",
| ^~
cc1: some warnings being treated as errors
Previously, when a bridge or bonding interface is in degraded-carrier
state, then we cannot judge the interface has addresses or not.
By using the new states, dbus clients can distinguish such situation.
Otherwise, the interface cannot be in "configured" state, as ipv6 link local
addressing is enabled by default. Note that even if ConfigureWithoutCarrier=
is set, all dynamic configurations are checked when the interface has
carrier.
We disabled kernel RA support. Then, we should not send
IFLA_INET6_TOKEN.
Thus, we do not need to send IFLA_INET6_ADDR_GEN_MODE twice.
Follow-up for 0e2fdb83bb and
4eb086a387.
```
May 10 11:08:54 test systemd-networkd[447]: wwan0: Failed to stop LLDP: Success
May 10 11:08:54 test systemd-networkd[447]: wwan0: Gained carrier
May 10 11:08:54 test systemd-networkd[447]: wwan0: Failed to start LLDP: Success
```
Fillup IFLA_INET6_ADDR_GEN_MODE while we do link_up.
Fixes the following error:
```
dummy-test: Could not bring up interface: Invalid argument
```
After reading the kernel code when we do a link up
```
net/core/rtnetlink.c
IFLA_AF_SPEC
af_ops->set_link_af(dev, af);
inet6_set_link_af
if (tb[IFLA_INET6_ADDR_GEN_MODE])
Here it looks for IFLA_INET6_ADDR_GEN_MODE
```
Since link up we didn't filling up that it's failing.
Closes#12504.
The function sd_radv_add_prefix() in dhcp6_pd_prefix_assign() may
return -EEXIST, and in that case the sd_radv_prefix object allocated
in dhcp6_pd_prefix_assign() will be freed when the function returns.
Hence, the key value in Manager::dhcp6_prefixes hashmap is lost.
Fixes#12454.
gcc was complaining that the link->ifname argument is NULL. Adding
assert(link->ifname) right before the call has no effect. It seems that
gcc is confused by the fact that log_link_warning_errno() internally
calls log_object(), with link->ifname passed as the object. log_object()
is also a macro and is does a check whether the passed object is NULL.
So we have a check if something is NULL right next an unconditional use
of it where it cannot be NULL. I think it's a bug in gcc.
Anyway, we don't need to use link->ifname here. log_object() already prepends
the object name to the message.
We not stopping the clients when networkd stops. They
should shut down cleanly and then we need to clean the DS.
One of requirements to implement
https://github.com/systemd/systemd/issues/10820.
```
^CBus bus-api-network: changing state RUNNING → CLOSED
DHCP SERVER: UNREF
DHCP SERVER: STOPPED
DHCP CLIENT (0x60943df0): STOPPED
veth-test: DHCP lease lost
veth-test: Removing address 192.168.5.31
NDISC: Stopping IPv6 Router Solicitation client
DHCP CLIENT (0x0): FREE
==24308==
==24308== HEAP SUMMARY:
==24308== in use at exit: 8,192 bytes in 2 blocks
==24308== total heap usage: 4,230 allocs, 4,228 frees, 1,209,732 bytes allocated
==24308==
==24308== LEAK SUMMARY:
==24308== definitely lost: 0 bytes in 0 blocks
==24308== indirectly lost: 0 bytes in 0 blocks
==24308== possibly lost: 0 bytes in 0 blocks
==24308== still reachable: 8,192 bytes in 2 blocks
==24308== suppressed: 0 bytes in 0 blocks
==24308== Rerun with --leak-check=full to see details of leaked memory
==24308==
==24308== For lists of detected and suppressed errors, rerun with: -s
==24308== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)
==24308== could not unlink /tmp/vgdb-pipe-from-vgdb-to-24308-by-sus-on-Zeus
==24308== could not unlink /tmp/vgdb-pipe-to-vgdb-from-24308-by-sus-on-Zeus
==24308== could not unlink /tmp/vgdb-pipe-shared-mem-vgdb-24308-by-sus-on-Zeus
```
1. When the DHCPv4 lease expires kernel removes the route. So add it back
when we gain lease again.
Closes https://github.com/systemd/systemd/issues/12426
2. When UseRoutes=false do not remove router
The link may not have corresponding .network file.
Note that in that case, link_ipv4ll_enabled() and link_dhcp4_enabled()
returns false. So, it is safe to drop the assertion.
Fixes#12422.
- bridge or bonding master takes a reference of slave links.
- drop link from bridge or bonding master's slave list when slave link
is removed.
- change type of Link::slaves to Set*,
Fixes#12315.
When a uevent is received during the relevant interface is in
LINK_STATE_PENDING, then the interface may be initialized twice.
To prevent that, this introduces LINK_STATE_INITIALIZED.
This is partially a refactoring, but also makes many more places use
unlocked operations implicitly, i.e. all users of fopen_temporary().
AFAICT, the uses are always for short-lived files which are not shared
externally, and are just used within the same context. Locking is not
necessary.
Let's be helpful to static analyzers which care about whether we
knowingly ignore return values. We do in these cases, since they are
usually part of error paths.
Previously, 'degraded' state is ambiguous for bonding or bridge master:
1. one or more slave interfaces does not have carrier,
2. no link local address is assigned to the master,
3. combination of the above two.
This makes the above case 1 and 3 are in the new 'degraded-carrier'
state, and makes 'degraded' state as all slaves are active but no
link local address on master.
When enabled, three samples are used to determine the value of a
received bit by majority rule.
This patch adds support for the TripleSampling= option in the [CAN]
section of .network files.
This effectively reverts 5971cb9de9 and
2b00a4e03d.
Usually, it is not necessary to assign addresses to bridge slaves,
but such functionality is supported by kernel. If users explicitly
request such configuration, networkd should support that.
We were already using OrderedSets in the manager object, but strvs in the
configuration parsing code. Using sets gives us better scaling when many
domains are used.
In oss-fuzz #13059 the attached reproducer takes approximately 30.5 s to be
parsed. Converting to sets makes this go down to 10s. This is not _vastly_
faster, but using sets seems like a nicer approach anyway. In particular, we
avoid the quadratic de-unification operation after each addition.
The DHCP client should not pre-filter addresses beyond what RFC
requires. If a client's user (like networkd) wishes to skip/filter
certain addresses, it's their responsibility.
The point of this is that the DHCP library does not hide/abstract
information that might be relevant for certain users. For example,
NetworkManager exposes DHCP options in its API. When doing that, the
options should be close to the actual lease.
This is related to commit d9ec2e632d
(dhcp4: filter bogus DNS/NTP server addresses silently).
When there is bad link in the network the carrier goes up/down.
This makes networkd stops all the clients and drop config.
But if the remote router/dhcpserver running a prevention
of DHCP Starvation attack or DHCP Flood attack it does not allow
networkd to take a DHCP lease resulting failure in configuration.
This patch allows to keep the client running and keep the conf
also for this scenario.
Closes#9111
If an IPv6 route is added with a source address that is still
tentative, the kernel will refuse to install it.
Previously, once we sent the messages to the kernel to add the
addresses, we would immediately proceed to add the routes. The
addresses would usually still be tentative at this point, so
adding static IPv6 routes was broken - see issue #5882.
Now, only begin to configure routes once the addresses are ready,
by restructuring the state machine, and tracking when addresses are
ready, not just added.
Fixes: #5882
Signed-off-by: Daniel Axtens <dja@axtens.net>
c4397d94c3 introduces
link_detach_from_manager() and netdev_detach_from_manager(), and they
set Link::manager or NetDev::manager NULL.
But, at the time e.g. link is removed, hence link_drop() is called,
there may be still some asynchronous netlink call is waiting, and
their callbacks hit assertion.
This make {link,netdev}_detach_from_manager() just drop all references
from manager, but keep the pointer to manager.
Fixes#11411.
We're about to need it to be later in the file for the next commit.
Moving it now means that when we change it in the next commit, it's
not intermingled with the move.
No functional change intended.
Signed-off-by: Daniel Axtens <dja@axtens.net>
An earlier commit 0e408b82b (dhcp6-client: handle IAID with value zero)
introduced a flag to sd_dhcp6_client to distinguish between an unset
IAID and a value set to zero.
However, that was not sufficient and broke leaving the setting
uninitialized in networkd configuration. The configuration parsing
also must distinguish between the default, unset value and an
explict zero configuration.
Fixes: 0e408b82b8
When using networkd we currently have no way of ensuring that static
neighbor entries are set when our link comes up. This change adds a new
section to the network definition that allows multiple static neighbors
to be set on a link.
This allows us to convey that we are performing multiple link
configuration changes in parallel. This is needed to support configuring
neighbors while simultaneously configuring addresses and routes.
This will be useful to assert that our static route configuration always
happens after address configuration once our individual configure state
goes away.
This splits out a bunch of functions from fileio.c that have to do with
temporary files. Simply to make the header files a bit shorter, and to
group things more nicely.
No code changes, just some rearranging of source files.
Now that we don't (mis-)use the env file parser to parse kernel command
lines there's no need anymore to override the used newline character
set. Let's hence drop the argument and just "\n\r" always. This nicely
simplifies our code.
In case networkd is restarted this prevents a removal of an already existing
route that would be configured using networkd. With the proposed changes the
route will be kept on the interface without removing. This happens only on
physical hosts or VMs since networkd handles interface configuration slightly
different in containers.
In case networkd is restarted this prevents a removal of an already existing IP
address that would be configured using networkd. With the proposed changes the
IP address will be kept on the interface without removing. This happens only on
physical hosts or VMs since networkd handles interface configuration slightly
different in containers.
When a link is configured, wait until there is a Router Advertisement before
attempting to start DHCPv6. The intended DHCPv6 mode will be evaluated in
ndisc_router_handler() in networkd-ndisc.c.
Request prefix delegation for a new downstream link that is enabled
after any number of upstream DHCPv6 links. Submit the request after
the link has been configured with a link-local address.
If the upstream DHCPv6 client has already been configured to request
prefixes, attempt to re-assign any possible prefixes between the
already existing links and the new one. If no prefixes are yet
acquired, nothing will happen right away and any prefixes will be
distributed after a reply from the DHCPv6 server.
If none of the already existing downstream links have requested
DHCPv6 prefixes to be assigned, enable prefix delegation for each
client and restart them one by one if they are already running. This
causes the DHCPv6 clients to re-acquire addresses and prefixes and
to re-distribute them to all links when receiving an updated
response from their respective DHCPv6 servers. If the DHCPv6 client
in question was not already running, it is set to request prefixes
but not restarted.
When an error occurs while setting or restarting the DHCPv6 client,
log the incident and move over to the next link.
Fixes#9758.
Use "falling back" instead of "fallback".
Also, it's not an application-specific machine ID, but rather an
application-and-machine-specific ID. Let's call it "app-machine-speicific" for
short.
Now the setting MTU is embedded into the link_up message which makes it
incapable of setting MTU if link is up. MTU can be set while Link is up.
Closes#9254
This part of the copyright blurb stems from the GPL use recommendations:
https://www.gnu.org/licenses/gpl-howto.en.html
The concept appears to originate in times where version control was per
file, instead of per tree, and was a way to glue the files together.
Ultimately, we nowadays don't live in that world anymore, and this
information is entirely useless anyway, as people are very welcome to
copy these files into any projects they like, and they shouldn't have to
change bits that are part of our copyright header for that.
hence, let's just get rid of this old cruft, and shorten our codebase a
bit.
Most our other parsing functions do this, let's do this here too,
internally we accept that anyway. Also, the closely related
load_env_file() and load_env_file_pairs() also do this, so let's be
systematic.
Rather than choosing to set or unset any of these flag
use kernel defaults. This patch makes following properties to unset.
UseBPDU = unset
HairPin = unset
FastLeave = unset
AllowPortToBeRoot = unset
UnicastFlood = unset
If an interface name is changed, then the link state, especially
managed or not, may need to be updated, as its corresponding
.link or .network files may be different. So, let's once drop
the link and recreate a new link object.
Fixes#8794.
This cleans up handling of MTU values across the codebase. Previously
MTU values where stored sometimes in uint32_t, sometimes in uint16_t,
sometimes unsigned and sometimes in size_t. This now unifies this to
uint32_t across the codebase, as that's what netlink spits out, and what
the majority was already using.
Also, all MTU parameters are now parsed with config_parse_mtu() and
config_parse_ipv6_mtu() is dropped as it is now unneeded.
(Note there is one exception for the MTU typing: in the DCHPv4 code we
continue to process the MTU as uint16_t value, as it is encoded like
that in the protocol, and it's probably better stay close to the
protocol there.)
This drops a good number of type-specific _cleanup_ macros, and patches
all users to just use the generic ones.
In most recent code we abstained from defining type-specific macros, and
this basically removes all those added already, with the exception of
the really low-level ones.
Having explicit macros for this is not too useful, as the expression
without the extra macro is generally just 2ch wider. We should generally
emphesize generic code, unless there are really good reasons for
specific code, hence let's follow this in this case too.
Note that _cleanup_free_ and similar really low-level, libc'ish, Linux
API'ish macros continue to be defined, only the really high-level OO
ones are dropped. From now on this should really be the rule: for really
low-level stuff, such as memory allocation, fd handling and so one, go
ahead and define explicit per-type macros, but for high-level, specific
program code, just use the generic _cleanup_() macro directly, in order
to keep things simple and as readable as possible for the uninitiated.
Note that before this patch some of the APIs (notable libudev ones) were
already used with the high-level macros at some places and with the
generic _cleanup_ macro at others. With this patch we hence unify on the
latter.
Double newlines (i.e. one empty lines) are great to structure code. But
let's avoid triple newlines (i.e. two empty lines), quadruple newlines,
quintuple newlines, …, that's just spurious whitespace.
It's an easy way to drop 121 lines of code, and keeps the coding style
of our sources a bit tigther.
In some situation if networkd fails to get the network file
then networkd crashes becasse the link->network is not initalized;
```
veth99: Failed to get network dhcp-client-ipv4-only: No such file or directory
Segmentation fault
gdb) r
The program being debugged has been started already.
Start it from the beginning? (y or n) y
Starting program: /home/sus/tt/systemd/build/systemd-networkd
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
warning: Loadable section ".note.gnu.property" outside of ELF segments
warning: Loadable section ".note.gnu.property" outside of ELF segments
warning: Loadable section ".note.gnu.property" outside of ELF segments
Program received signal SIGSEGV, Segmentation fault.
link_load (link=0x55555582ccd0) at ../src/network/networkd-link.c:2973
2973 r = sd_dhcp_client_new(&link->dhcp_client, link->network->dhcp_anonymize);
(gdb) bt
(gdb) p link->network
$1 = (Network *) 0x0
(gdb) list
2968 if (r < 0) {
2969 log_link_debug_errno(link, r, "Failed to parse DHCPv4 address %s: %m", dhcp4_address);
2970 goto dhcp4_address_fail;
2971 }
2972
2973 r = sd_dhcp_client_new(&link->dhcp_client, link->network->dhcp_anonymize);
2974 if (r < 0)
2975 return log_link_error_errno(link, r, "Failed to create DHCPv4 client: %m");
2976
2977 r = sd_dhcp_client_set_request_address(link->dhcp_client, &address.in);
(gdb) r
```
Files which are installed as-is (any .service and other unit files, .conf
files, .policy files, etc), are left as is. My assumption is that SPDX
identifiers are not yet that well known, so it's better to retain the
extended header to avoid any doubt.
I also kept any copyright lines. We can probably remove them, but it'd nice to
obtain explicit acks from all involved authors before doing that.
This makes users can configure DHCPv4 client with ClientIdentifier=duid-only.
If set so, then DHCP client sends only DUID as the client identifier.
This may not be RFC compliant, but some setups require this.
Closes#7828.
kernel >= 4.5 (with commit 32bc201e19) supports
RTA_EXPIRES netlink attribute to set router lifetime. This simply detect
the kernel version (>=4.5) and set the lifetime properly, fallback to
expiring route in userspace for kernel that doesnt support it.
Signed-off-by: Daniel Dao <dqminh89@gmail.com>
It turns out that link local doesn't make much sense in its context.
Since link local is disabled by the kernel driver, it's important that
networkd assumes it's off too, so that the link can reach the
"configured" stage, without waiting indefinitely for link local
addresses which will never come.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
When there is a change in mac address we are not currently not changing
the MAC address and resulting v6 connectivity is gone.
When kernel reports a change in mac address change the MAC of ndisc
client too.
Closes # 7806
commit 7715629 (networkd: Fix race condition in [RoutingPolicyRule] handling (#7615)).
Does not fix race. Still there is a race in case of bride because the
bride goes down and up .
calling route_configure then link_set_routing_policy_rule and the
link_check_ready makes a race between routing_policy_rule_messages and route_messages.
While bride comes up and we call the call again route_configure if finds
it self in the callback function LINK_STATE_CONFIGURED networkd dies.
Let's handle first routing policy rules then route_configure. This fixes
the crash.
Closes#7797
The Network section IPv6PrefixDelegation= option takes two new
configuration values, namely "static" and "dhcpv6" in addition
to boolean yes and no values. Static prefixes in IPv6Prefix
sections are used when IPv6PrefixDelegation= option contains
"static", and DHCPv6 is queried for prefixes when the option
contains "dhcpv6". Both DHCPv6 and static prefixes are used when
the option contains a boolean true value. The default value is
false as before, meaning no prefixes are delegated.
Let's replace usage of fputc_unlocked() and friends by __fsetlocking(f,
FSETLOCKING_BYCALLER). This turns off locking for the entire FILE*,
instead of doing individual per-call decision whether to use normal
calls or _unlocked() calls.
This has various benefits:
1. It's easier to read and easier not to forget
2. It's more comprehensive, as fprintf() and friends are covered too
(as these functions have no _unlocked() counterpart)
3. Philosophically, it's a bit more correct, because it's more a
property of the file handle really whether we ever pass it on to another
thread, not of the operations we then apply to it.
This patch reworks all pieces of codes that so far used fxyz_unlocked()
calls to use __fsetlocking() instead. It also reworks all places that
use open_memstream(), i.e. use stdio FILE* for string manipulations.
Note that this in some way a revert of 4b61c87511.
The routing policy rule setup logic is moved to the routes setup phase (rather than the addresses setup phase as it is now). Additionally, a call to `link_check_ready` is added to the routing policy rules setup handler. This prevents a race condition with the routes setup handler.
Also give each async handler its own message counter to prevent race conditions when logging successes.
Fixes: #7614