The user data values are encoded in shell variables named
prefix "NM_USER_". The variable name is an encoded form of the
data key, consisting only of upper-case letters, digits, and underscore.
The alternative would be something like
NM_USER_1_KEY=my.keys.1
NM_USER_1_VAL='some value'
NM_USER_2_KEY=my.other.KEY.42
NM_USER_2_VAL='other value'
contary to
NM_USER_MY__KEYS__1='some value'
NM_USER_MY__OTHER___K_E_Y__42='other value'
The advantage of the former, numbered scheme is that it may be easier to
find the key of a user-data entry. With the current implementation, the
shell script would have to decode the key, like the ifcfg-rh plugin
does.
However, user data keys are opaque identifers for values. Usually, you
are not concerned with a certain name of the key, you already know it.
Hence, you don't need to write a shell script to decode the key name,
instead, you can use it directly:
if [ -z ${NM_USER_MY__OTHER___K_E_Y__42+x} ]; then
do_something_with_key "$NM_USER_MY__OTHER___K_E_Y__42"
fi
Otherwise, you'd first have to search write a shell script to search
for the interesting key -- in this example "$NM_USER_2_KEY", before being
able to access the value "$NM_USER_2_VAL".
(cherry picked from commit 79be44d990)
Most of the IPv6 methods require a non-tentative link local address
configured on the interface; we look at priv->ip6_config to determine
if such address exist. If the configuration is out-of-sync, we may
proceed with configuration when the link-local address does not exist
or is still tentative, especially because we toggle the "disable_ipv6"
sysctl parameter just before, which clears all IPv6 addresses on the
interface.
Ensure that priv->ip6_config is up-to-date before continuing with the
IPv6 configuration.
Fixes test: @ipv6_set_ra_announced_mtu
(cherry picked from commit 78b43f7ea1)
nm_device_update_firewall_zone() would only reconfigure the firewall
zone when the device is fully activated. That means, while the device
is activating, changing the firewall zone is not working. Activation
might take a long time with DHCP, or with master devices waiting
for their slaves.
For example:
nmcli connection add type team con-name t-team ifname i-team autoconnect no
nmcli connection up t-team
Note how t-team/i-team is waiting for a slave device. During stage3,
we already set firewall.zone to default.
nmcli connection modify t-team connection.zone external
Note how changing the firewall zone does not immidiately take
effect. Only later, during IP_CHECK state the firewall zone
is reset -- but only for devices with differing ip_ifindex.
https://bugzilla.redhat.com/show_bug.cgi?id=1445242
(cherry picked from commit 20ccbb97d5)
For regular devices that don't have a separate ip_iface/ip_ifindex,
the ip_ifindex is left at zero. Hence, the condition is always
true and does not work as intended, resulting in setting the
firewall zone twice.
Fixes: 7cf5c326bc
(cherry picked from commit baa8b4029c)
Commit 850c97795 ("device: track system interface state in NMDevice")
introduced interface states for devices and prevented checking if a
connection should be assumed on already managed devices.
This prevented to properly manage the event of an ip configuration added
externally to NM to a managed but not (yet) activated device.
Fixes: 850c977953
(cherry picked from commit acf1067a45)
When a DHCP connection is active and the DHCP server is temporarily
unreachable, we restart DHCP for some times before failing the
connection. From the user point of view, restarting NM (and thus
assuming the existing connection) should not change this behavior.
However, if NM is restarted while the server is temporarily down, at
the moment we immediately fail because we consider the DHCP
transaction our first try. Fix this by restoring the multiple tries
when we detect that DHCP was active before because the connection is
assumed.
(cherry picked from commit e3113fdc4b)
If we don't have connection checking functionality just avoid adding
a penalty to the defaut route of newly activated connections.
(cherry picked from commit 2524a6f852)
We call nm_device_activate_stage3_ipX_start() in various places,
e.g. after a carrier change or when a master enslaves a new device to
configure IP for the device. If the device is a slave in state
IP_CONFIG, this makes it transition to IP_CHECK, while it should stay
in IP_CONFIG until the master becomes ready. When the master is ready,
it will move slaves directly to SECONDARIES, skipping IP configuration
entirely.
(cherry picked from commit 41f6540afd)
When the operation is cancelled, we must not touch user_data. Note that
NM_POLICY_GET_PRIVATE() theoretically doesn't dereference the pointer
(does it?) but doing pointer arithmetic on a dangling pointer is a very
ugly thing to do.
And of course, the memleak.
Fixes: 5c716c8af8
Fixes: a2cdf63204
(cherry picked from commit 3215508293)
The default timeout in dhclient is 60 seconds; if a lease can't be
obtained during such interval, dhclient sends to NM a FAIL event and
then the IP method fails.
Thus, even if user specified a greater dhcp-timeout, NM terminated
DHCP after 60 seconds. Fix this by passing an explicit timeout to
dhclient.
(cherry picked from commit 82ef497cc9)
py-kickstart writes this out and there apparently are users using this.
Let them have one less problem.
Co-Authored-By: Thomas Haller <thaller@redhat.com>
https://bugzilla.redhat.com/show_bug.cgi?id=1445414
(cherry picked from commit dbe0659ba419a77ad5ff2340bfc93c71a1bec61a)
py-kickstart writes this out. Okay -- we don't care on read and it makes
sense when there actually are addresses.
https://bugzilla.redhat.com/show_bug.cgi?id=1445414
(cherry picked from commit aa50dfc236b3806c6d7161cdea450655a1268f0d)
For example, if you want to test whether a value is present and
reset it to a different value (only if it is present), it would
be reasonable to do
if (svGetValue (s, key, &tmp)) {
svSetValue (s, key, "new-value");
g_free (tmp);
}
Without this patch, you could not be sure that key is not
set to some inparsable value, which svWriteFile() would then
write out as empty string.
Have invalid values returned by svGetValue() as empty string.
That is how svWriteFile() treats them.
(cherry picked from commit 6470bed5f1ad25e20df14b333f1b083c9b390ece)
The dad_counter is hashed into the resulting address. Since we
want the hashing to be independent of the architecture, we always
hash 32 bit of dad_counter. Make the dad_counter argument of
type guint32 for consistency.
In practice this has no effect because:
- for all our (current!) architectues, guint is the same as
guint32.
- all callers of nm_utils_ipv6_addr_set_stable_privacy() keep
their dad-counter argument as guint8, so they never even pass
numbers larger then 255.
- nm_utils_ipv6_addr_set_stable_privacy() limits dad_counter
further against RFC7217_IDGEN_RETRIES.
(cherry picked from commit 951e5f5bf8)
https://tools.ietf.org/html/rfc7217 says:
The resulting Interface Identifier SHOULD be compared against the
reserved IPv6 Interface Identifiers [RFC5453] [IANA-RESERVED-IID]
and against those Interface Identifiers already employed in an
address of the same network interface and the same network
prefix. In the event that an unacceptable identifier has been
generated, this situation SHOULD be handled in the same way as
the case of duplicate addresses (see Section 6).
In case of conflict, this suggests to create a new address incrementing
the DAD counter, etc. Don't do that. If we generate an address of the
reserved region, just rehash it right away. Note that the actual address
anyway appears random, so this re-hashing is just as good as incrementing
the DAD counter and going through the entire process again.
Note that now we no longer generate certain addresses like we did
previously. But realize that we now merely reject (1 + 16777216 + 128)
addresses out of 2^64. So, the likelyhood of of a user accidentally
generating an address that is suddenly rejected is in the order of
10e-13 (1 / 1,099,503,173,697). Which is not astronomically, but still
extreeeemely unlikely.
Also, the whole process is anyway build on the idea that somebody else
might generate conflicting addresses (DAD). It means, there was always
the extremely tiny chance that the address you generated last time is
suddenly taken by somebody else. So, this change appears to a user
like these reserved addresses are now claimed by another (non existing)
host and a different address gets generated -- business as usual, as
far as SLAAC is concerned.
(cherry picked from commit f15c4961ad)
Fixes a crash where the default DNS domain to be announced together with the
prefixes to be delegated is updated at the same time the device is being
unrealized.
https://bugzilla.redhat.com/show_bug.cgi?id=1425818
(cherry picked from commit 3e076cf8b1)
NMDeviceGeneric:check_connection_compatible() doesn't check for a
matching interface name. It relies on the parent implementation to
do that.
The parent implementation calls nm_manager_get_connection_iface().
That fails for NM_SETTING_GENERIC_SETTING_NAME, because that one has
no factory. Maybe this imbalance of having no factory for the Generic device
is wrong, but usually factories only match a distinct set of device
types, while the generic factory would handle them all (as last resort).
Without this, activating a generic connection might activate the
wrong interface.
(cherry picked from commit 3876b10a47)
We have unit tests for writing and re-reading ifcfg file. Those
tests compare whether a file can be successfully read and is
semantically identical.
However, there were no tests that a certain output is written in
a stable format. We aim not to change the output of what we write.
For that, add tests to not only check the semantic of the written
ifcfg file, but their bits and bytes.
Some future changes may well intentionally change the current
output. That will require to update the expected result files
and can be done via
NMTST_IFCFG_RH_UPDATE_EXPECTED=yes src/settings/plugins/ifcfg-rh/tests/test-ifcfg-rh
Note that alias, route, and key files are not checked.
Related: https://bugzilla.redhat.com/show_bug.cgi?id=1445414
(cherry picked from commit f04bf45e84)
As NMDevice now creates the NMPacrunnerManager instance
as needed, it is even more likely that the initial call
to nm_pacrunner_manager_send() will only queue (but not yet
send) the new config.
Later, when the D-Bus proxy is created, we will not get a
name-owner changed signal. We instead have to push the configuration
right away.
(cherry picked from commit 019b9fbfc0)
nm_pacrunner_manager_remove() required a "tag" argument. It was a
bug for callers trying to remove a configuration for a non-existing
tag.
That effectively means, the caller must keep track of whether a certain
"tag" is pending. The caller also must remember the tag -- a tag that he
must choose uniquely in the first place.
Turn that around and have nm_pacrunner_manager_send() return a (non
NULL) call-id. This call-id may later be used to remove the
configuration.
Apparently, previously the tracking of the "tag" was not always correct
and we hit the assertion in nm_pacrunner_manager_remove().
https://bugzilla.redhat.com/show_bug.cgi?id=1444374
(cherry picked from commit b04a9c90eb)
Usually, this "<allow send_destination="..."/>" part is shipped
by firewalld's D-Bus policy. However, if firewalld is initially
not installed with NetworkManager already running, dbus-daemon
seems to cache the missing permission for the D-Bus connection.
As a result, when installing and starting firewalld, NetworkManager
requests fail until restart:
firewall: [0x7f4b83643890,change:"eth1"]: complete: request failed (Rejected send message, 1 matched rules; type="method_call", sender=":1.3" (uid=0 pid=715 comm="/usr/sbin/NetworkManager --no-daemon ") interface="org.fedoraproject.FirewallD1.zone" member="changeZone" error name="(unset)" requested_reply="0" destination=":1.25" (uid=0 pid=1243 comm="/usr/bin/python -Es /usr/sbin/firewalld --nofork -"))
https://bugzilla.redhat.com/show_bug.cgi?id=1436770
(cherry picked from commit cc1d409ba8)
We want to ignore certain errors from firewalld. In the past,
the error message contained only the error code.
Since recently ([1], [2]), the error message contains a longer text:
NetworkManager[647]: <debug> [1492768494.7475] device[0x7f7f21e78f50] (eth0): Activation: setting firewall zone 'default'
NetworkManager[647]: <debug> [1492768494.7475] firewall: [0x7f7f21ed8900,change:"eth0"]: firewall zone change eth0:default
...
firewalld[2342]: ERROR: UNKNOWN_INTERFACE: 'eth0' is not in any zone
NetworkManager[647]: <warn> [1492768494.7832] firewall: [0x7f7f0400c780,remove:"eth0"]: complete: request failed (UNKNOWN_INTERFACE: 'eth0' is not in any zone)
[1] c77156d7f6
[2] 7c6ab456c5
(cherry picked from commit 2ad8bb0ce3)
We now initialize the NMFirewallManager asynchronously. That means, at
first firewalld appears as "not running", for which we usually would
fake-success right away.
It would be complex for callers to wait for firewall-manager to be
ready. So instead, have the asynchronous requests be queued and
complete them once the D-Bus proxy is initialized.
(cherry picked from commit fb7815df6e)
Next we will get another mode, so an is-idle doesn't cut it.
It can be confusing where the mode is set and where it is only
accessed read-only. For that, add mode_mutable.
(cherry picked from commit 04f4e327a9)
Creating it asynchronously changes that on the first call to
nm_firewall_manager_get() the instance is not yet running.
Note that NMPolicy already connects to the "STARTED" signal and
reapplies the zones when firewalld appears. So, this delayed
change of the running state is handled mostly fine already.
One part is still missing, it's to queue add_or_change/remove calls
while the firewall manager is initializing. That follows next.
(cherry picked from commit 753f39fa82)
Since commit 2d1b85f (th/assume-vs-unmanaged-bgo746440), we clearly
distinguish between two modes when encountering devices with external
IP configuration:
a) external devices. For those devices we generate a volatile in-memory
connection and pretend it's active. However, the device must not be
touched by NetworkManager in any way.
b) assume, seamless take over. Mostly for restart of NetworkManager,
we activate a connection gracefully without going through an down-up
cycle. After the device reaches activated state, the device is
considered fully managed. For this only an existing, non volatile
connection can be used.
Before 'th/assume-vs-unmanaged-bgo746440', the behaviors were not
clearly separated.
Since then, we only choose to assume a connection (b) when the state
file indicates a matching connection. Now, extend this to also assume
connections when:
- during first-start (not after a restart) when there is no
state file yet.
- and, if we have an existing, non volatile, connection which
matches the device's configuration.
This patch lets NetworkManager assume connection also on first start.
That is for example useful when handing over network configuration from
initrd.
This only applies to existing, permanent, matching(!) connections, so it is a
good guess that the user wants NM to take over this interface. This brings us
closer to the previous behavior before 'th/assume-vs-unmanaged-bgo746440'.
https://bugzilla.redhat.com/show_bug.cgi?id=1439220
(cherry picked from commit 27b2477cb7)
nm_config_device_state_*() always access the file system directly,
they don't cache data in NMConfig. Hence, they don't use the
@self argument.
Maybe those functions don't belong to nm-config.h, anyway. For lack
of a better place they are there.
(cherry picked from commit 1940be410c)
Before, setting a device to unmanaged causes it to go down and clear
the interface state.
It may be useful to instruct NetworkManager not to touch the device
anymore but leave the current state up. Changing behavior for
nmcli device set "$DEV" managed no
To get the previous behavior, one has to first disconnect the interface
via
nmcli device disconnect "$DEV"
nmcli device set "$DEV" managed no
Note that non-permanent addresses like from DHCP will eventually time
out because NetworkManager stops the DHCP client. When instructing
NetworkManager to let go of the device, you have to take it over in
any way you see fit.
https://bugzilla.redhat.com/show_bug.cgi?id=1371433
(cherry picked from commit 9e8218f99a)
Arguably, we currently only have one instance of NMPlatform,
NMRouteManager, NMDefaultRouteManager -- the one owned by the
NMNetns singleton.
Hence, all these instances we create with "log-with-ptr" set explicitly
to false.
In the future we want to support namespaces, and it will be be common to
have multiple instances. For that we have "log-with-ptr" so we are able
to disambiguiate the logging.
Change the default to TRUE because it makes more sense. It has currently
no effect as the default is never used.
(cherry picked from commit 41148caba8)
Reduce the use of NM_PLATFORM_GET / nm_platform_get() to get
the platform singleton instance.
For one, this is a step towards supporting namespaces, where we need
to use different NMNetns/NMPlatform instances depending on in which
namespace the device lives.
Also, we should reduce our use of singletons. They are difficult to
coordinate on shutdown. Instead there should be a clear order of
dependencies, expressed by owning a reference to those singelton
instances. We already own a reference to the platform singelton,
so use it and avoid NM_PLATFORM_GET.
(cherry picked from commit 94d9ee129d)
This also ensures that we own a reference to the
NMPlatform, NMRouteManager and NMDefaultRouteManager
instances. See bug rh#1440089 where we might access
the singleton getter after destroing the singleton
instance of NMRouteManager. This is prevented by
keeping a reference to those instances -- indirectly
via the netns instance.
Later, we may add support for multiple namespaces. Then it might
make sense to swap the NMNetns instance of a device when moving
the device between namespaces.
Also, drop the use of singelton instances.
https://bugzilla.redhat.com/show_bug.cgi?id=1440089
(cherry picked from commit c48a19b7c6)
NMPlatform, NMRouteManager and NMDefaultRouteManager are singletons
instances. Users of those are for example NMDevice, which registers
to GObject signals of both NMPlatform and NMRouteManager.
Hence, as NMDevice:dispose() disconnects the signal handlers, it must
ensure that those singleton instances live longer then the NMDevice
instance. That is usually accomplished by having users of singleton
instances own a reference to those instances.
For NMDevice that effectively means that it shall own a reference to
several singletons.
NMPlatform, NMRouteManager, and NMDefaultRouteManager are all
per-namespace. In general it doesn't make sense to have more then
one instances of these per name space. Nnote that currently we don't
support multiple namespaces yet. If we will ever support multiple
namespaces, then a NMDevice would have a reference to all of these
manager instances. Hence, introduce a new class NMNetns which bundles
them together.
(cherry picked from commit 0af2f5c28b)
Set the device state as removed when the link disappears, so that in
the call to unrealize() when the device is unmanaged we also perform a
cleanup of it and especially, we terminate any DHCP client instances
running on the device.
If we keep DHCP clients running, we can hit assertions later when we
start another instance on the same interface, because we kill the old
dhclient from the pidfile, and the g_child_watch_add() done by the
first client instance is not able to waitpid() it, complaining with:
GChildWatchSource: Exit status of a child process was requested but
ECHILD was received by waitpid(). Most likely the process is
ignoring SIGCHLD, or some other thread is invoking waitpid() with a
nonpositive first argument; either behavior can break applications
that use g_child_watch_add()/g_spawn_sync() either directly or
indirectly.
https://bugzilla.redhat.com/show_bug.cgi?id=1436602
(cherry picked from commit df537d2eac)
