When comparing settings, nm_setting_compare() performs a complicated
logic, which basically serializes each GObject property to a GVariant
for the D-Bus representation.
That is wrong for example for ipv4.addresses, which don't contain
address labels. That is, the GObject property is called "addresses",
but the D-Bus field "addresses" cannot encode every information
and thus comparison fails. Instead, it would have to look into
"address-data".
Traditionally, we have virtual functions like compare_property() per
NMSetting to do the comparison. That comparison is based on the GObject
properties. I think that is wrong, because we should have a generic
concept of what a property is, independent from GObject properties.
With libnm, we added NMSettingProperty, which indeed is such an
GObject independent representation to define properties.
However, it is not used thoroughly, instead compare_property() is a hack
of special cases, overloads from NMSettingProperty, overloads of
compare_property(), and default behavior based on GParamSpec.
This should be cleaned up.
For now, just hack it by handle the properties with the problems
explicitly.
The Network_ID for generating RFC 7217 stable privacy IPv6 addresses
is by default the UUID of the connection.
Alternatively, prefer "connection.stable-id" as Network_ID to generate
the stable addresses. This allows to configure a set of connections that
all use the same Network_ID for generating stable addresses.
Note that the stable-id and the UUID do no overlap, that is two
connections
[connection]
uuid=uuid1
stable-id=
and
[connection]
uuid=uuid2
stable-id=uuid1
generate distinct addresses.
This new property be used as token to generate stable-ids instead
of the connection's UUID.
Later, this will be used by ipv6.addr-gen-mode=stable-privacy,
ethernet.cloned-mac-address=stable, and wifi.cloned-mac-address=stable
setting. Those generate stable addresses based on the connection's
UUID, but allow to use the stable-id instead.
This allows multiple connections to generate the same addresses
-- on the same machine, because in the above cases a machine
dependant key is also hashed.
The "shared" directory contains files that are possibly used by all components
of NetworkManager repository.
Some of these files are even copied as-is to other projects (VPN plugins, nm-applet)
and used there without modification. Move those files to a separate directory.
By moving them to a common directory, it is clearer that they belong
together. Also, you can easier compare the copied versions to their
original via
$ diff -r ./shared/nm-utils/ /path/to/nm-vpn-plugin/shared/nm-utils/
Let VPN plugins return a virtual function table to extend
the API while bypassing libnm. This allows to add and use
new functionality to VPN plugins without updating libnm.
The actual definitions are in a header-only file
"nm-vpn-editor-plugin-call.h", which can be copied to the
caller/plugin.
The NMVpnPluginInfo is essentially the .name file, that is, a
configuration file about the plugin itself. Via NMVpnPluginInfo
instance, the NMVpnEditorPlugin can be created.
Usually, one would create a NMVpnPluginInfo (that is, reading the
.name file) and then create a NMVpnEditorPlugin instance from there.
In this case, usually the editor-plugin is owned by the plugin-info
instance (although the API allows for creating the editor-plugin
independently).
Now, pass the NMVpnPluginInfo to the editor-plugin too.
This is useful, because then the editor-plugin can look at the .name
file.
The .name file is not user configuration. Instead it is configuration
about the plugin itself. Although the .name file is part of the plugin
build artefacts, it is useful to allow the plugin to access the .name
file. The reason is, that this can allow the user to easily change a
configuration knob of the plugin without requiring to patch or the
plugin.
Refactor code to add function _list_find_by_service(), which will be used in the
next commit.
A notable change is that we now search also through the aliases
together with the service-name.
That makes a difference, if one plugin privdes an "alias" which another
plugin provides as "service". Due to that change, we would also find the
aliased plugin first.
In practice it shouldn't matter, because different plugins are
not supposed to provide identical services.
Commit bdd0e7fec0 which added symbol
nm_setting_ip_config_get_dns_priority to libnm_1_4_0 was backported
to nm-1-2 in commit ad1cdcf657.
Add the backported symbol to master to allow seemless upgrading
from 1.2.4 to 1.4.0.
A failure to g_return*() by default prints a g_critical() with stringifing the
condition. Add a macro NMTST_G_RETURN_MSG() that reproduces that line to more
accurately match the failure message.
Add new Reload D-Bus command to reload NetworkManager configuration.
For now, this is like sending SIGHUP to the process. There are several
advantages here:
- it is guarded via PolicyKit authentication while signals
can only be sent by root.
- the user can wait for the reload to be complete instead of sending
an asynchronous signal. For now, we operation completes after
nm_config_reload() returns, but later we could delay the response
further until specific parts are fully reloaded.
- SIGHUP reloads everything including re-reading configuration from
disk while SIGUSR1 reloads just certain parts such as writing out DNS
configuration anew.
Now, the Reload command has a flags argument which is more granular
in selecting parts which are to be reloaded. For example, via
signals the user can:
1) send SIGUSR1: this writes out the DNS configuration to
resolv.conf and possibly reloads other parts without
re-reading configuration and without restarting the DNS plugin.
2) send SIGHUP: this reloads configuration from disk,
writes out resolv.conf and restarts the DNS plugin.
There is no way, to only restart the DNS plugin without also reloading
everything else.
NMSettingTeam implements a custom compare_property() method in order
to perform a relaxed matching on team configurations when it is
necessary to assume a connection. However, the method is called also
when the core needs to check if a connection has changed before an
update. In that case it is better to use the default string comparison
on the property, otherwise the second of these commands would not have
effect:
$ nmcli connection modify team0 team.config ''
$ nmcli connection modify team0 team.config '{ }'
because compare_property() returns TRUE. Use the @flags argument to
distinguish the two cases.
Fixes: 82f8a54854
This sorts the entries of the [vpn] section alphabetically.
The sorting order doesn't really matter, but having a defined,
stable way to export a connection is preferred. Also, it looks
better when looking at the keyfile with an editor.
No need to require the caller to initialize the optional out-argument.
Otherwise we get:
(nm-connection-editor:2471): libnm-CRITICAL **: nm_utils_file_is_private_key: assertion 'out_encrypted == NULL || *out_encrypted == FALSE' failed
https://bugzilla.gnome.org/show_bug.cgi?id=763578
Fixes: 1c4f41c610
This is required to add objects in the "Types and Values" section and
in the API index. Later, we may want to add useful content in those
empty comments.
<gmodule.h> is implicitly included by <gio/gio.h> which is available
everywhere. For that reason, we would not have to include this header
at all. However, it is recommended to explicitly include <gmodule.h>
where needed.
So, include it where needed -- if <gio/gio.h> wouldn't be there --
and drop it from where it is not needed.
Closing the library will interfere badly as the glib types cannot be
unregistered. We must leak the library handle.
Switch to dlopen() instead of g_module_open(), because the former
supports RTLD_NOLOAD. This is useful for two reasons:
- checking the file prior loading only makes any sense when
dlopen() would actually load a file anew.
- if the library was loaded before, we want to return the handle.
On the other hand, if the library was not loaded, we leak the
handle.
Thereby, refactor the code from if-else blocks to return-early, because
the function nicely does individual steps and if one fails just error
out.
g_object_get_data() and g_object_get_qdata() end up to be identical,
except that g_object_get_data() also requires to intern the string
on every lookup (which involves a hash lookup and locking).
Having a gateway defined when never-default=yes causes troubles in
connection matching and anyway makes no sense.
If the combination is found, remove the gateway during the
normalization phase.
https://bugzilla.redhat.com/show_bug.cgi?id=1313091
libnm-core/nm-vpn-plugin-info.c: In function ‘nm_vpn_plugin_info_get_auth_dialog’:
shared/gsystem-local-alloc.h:53:46: error: ‘prog_basename’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
Fixes: a3f94f451b
Otherwise the types links would be dangling or resolved to slightly
irrelevant documentation in libnm or completely irrelevant documentation
in libnm-util.
nmcli> set 802-1x.ca-cert file:///tmp/certs/eaptest_ca_cert.pem
(process:31015): libnm-CRITICAL **: crypto_is_pkcs12_data: assertion 'data != NULL' failed
Error: failed to set 'ca-cert' property: PEM certificate had no start tag '-----BEGIN CERTIFICATE-----'.
In commit ca000cffbb, we changed to
accept a plugin library name without path. One reason for that
is to keep architecture dependent parts out of the .name file
and possibly support multilib.
However, the shared libraries of VPN plugins are not installed in
a global library search path, but for example into
"/usr/lib64/NetworkManager/libnm-vpn-plugin-openvpn.so".
In that case, specifying "plugin=libnm-vpn-plugin-openvpn.so"
would not be enough to find plugin.
Instead, when configuring a plugin name without path, assume
it is in NMPLUGINDIR directory. Modify nm_vpn_editor_plugin_load_from_file()
to allow path-less plugin-names. Previously such names would be rejected
as not being absolute. This API allows to do file verification
before loading the plugin, but it now supports prepending NMPLUGINDIR
to the plugin name. Basically, this function mangles the plugin_name
argument and checks that such a file exists.
The recently added nm_vpn_editor_plugin_load() continues to behave
as before: it does no checks whatsoever and passes the name directly
to dlopen(). That way, it uses system search paths like LD_LIBRARY_PATH
and performs no checks on the file.
Fixes: ca000cffbb
Contrary to nm_vpn_editor_plugin_load_from_file(), this allows
to specify a library name without path. In this case, g_module_open()
(dlopen()) will search for a library in various system directories.
Since commit 3dfbbb227e, we enforce that
the plugin path in the .name file is absolute and we perform several
checks on the file before loading it (ownership, etc).
Relax that, to also allow libray names without path component.
In that case, g_module_open()/dlopen() will search for a library
in various search paths. This allows, to omit absolute paths
in the .name file. The latter is problematic, because by default
we install the .name file in the architecture independent location
/usr/lib/NetworkManager. As such, it should not contain paths
to architecture dependent libraries. With this change, a .name
file can contain only the library name and it will be loaded
using the usual mechanism.
However, specifying absolute paths is still possible and works
same as before, including checking file permissions.
As such, distributions probably should package the VPN plugins
to have no path in the .name file. On the other hand, a user
compiling from source probably wants to specify an absolute
path. The reason is, that the user probably doesn't build the
plugin for multiple achitectures and that way, he can install
the plugin in a separate (private) prefix.
Instead of looking into the keyfile, lookup the "supports-multiple-connections" setting
in the "keys" hash. This has some behavioral difference:
- g_key_file_get_boolean() first does g_key_file_get_value(), and then
converts the string using the private g_key_file_parse_value_as_boolean()
function -- which is case-sensitive, accepts "true|false|0|1" and
considers only the text until the first whitespace.
- now, we put g_key_file_get_string() into the cache "keys" and
parse it with _nm_utils_ascii_str_to_bool(). The latter is
case insensitive, allows also "yes|no|on|off", strips whitespaces.
However, the difference is subtle and shouldn't matter.
The point of this change is to free "keyfile" after construction.
