Completely rework how settings plugin handle connections and how
NMSettings tracks the list of connections.
Previously, settings plugins would return objects of (a subtype of) type
NMSettingsConnection. The NMSettingsConnection was tightly coupled with
the settings plugin. That has a lot of downsides.
Change that. When changing this basic relation how settings connections
are tracked, everything falls appart. That's why this is a huge change.
Also, since I have to largely rewrite the settings plugins, I also
added support for multiple keyfile directories, handle in-memory
connections only by keyfile plugin and (partly) use copy-on-write NMConnection
instances. I don't want to spend effort rewriting large parts while
preserving the old way, that anyway should change. E.g. while rewriting ifcfg-rh,
I don't want to let it handle in-memory connections because that's not right
long-term.
--
If the settings plugins themself create subtypes of NMSettingsConnection
instances, then a lot of knowledge about tracking connections moves
to the plugins.
Just try to follow the code what happend during nm_settings_add_connection().
Note how the logic is spread out:
- nm_settings_add_connection() calls plugin's add_connection()
- add_connection() creates a NMSettingsConnection subtype
- the plugin has to know that it's called during add-connection and
not emit NM_SETTINGS_PLUGIN_CONNECTION_ADDED signal
- NMSettings calls claim_connection() which hocks up the new
NMSettingsConnection instance and configures the instance
(like calling nm_settings_connection_added()).
This summary does not sound like a lot, but try to follow that code. The logic
is all over the place.
Instead, settings plugins should have a very simple API for adding, modifying,
deleting, loading and reloading connections. All the plugin does is to return a
NMSettingsStorage handle. The storage instance is a handle to identify a profile
in storage (e.g. a particular file). The settings plugin is free to subtype
NMSettingsStorage, but it's not necessary.
There are no more events raised, and the settings plugin implements the small
API in a straightforward manner.
NMSettings now drives all of this. Even NMSettingsConnection has now
very little concern about how it's tracked and delegates only to NMSettings.
This should make settings plugins simpler. Currently settings plugins
are so cumbersome to implement, that we avoid having them. It should not be
like that and it should be easy, beneficial and lightweight to create a new
settings plugin.
Note also how the settings plugins no longer care about duplicate UUIDs.
Duplicated UUIDs are a fact of life and NMSettings must handle them. No
need to overly concern settings plugins with that.
--
NMSettingsConnection is exposed directly on D-Bus (being a subtype of
NMDBusObject) but it was also a GObject type provided by the settings
plugin. Hence, it was not possible to migrate a profile from one plugin to
another.
However that would be useful when one profile does not support a
connection type (like ifcfg-rh not supporting VPN). Currently such
migration is not implemented except for migrating them to/from keyfile's
run directory. The problem is that migrating profiles in general is
complicated but in some cases it is important to do.
For example checkpoint rollback should recreate the profile in the right
settings plugin, not just add it to persistent storage. This is not yet
properly implemented.
--
Previously, both keyfile and ifcfg-rh plugin implemented in-memory (unsaved)
profiles, while ifupdown plugin cannot handle them. That meant duplication of code
and a ifupdown profile could not be modified or made unsaved.
This is now unified and only keyfile plugin handles in-memory profiles (bgo #744711).
Also, NMSettings is aware of such profiles and treats them specially.
In particular, NMSettings drives the migration between persistent and non-persistent
storage.
Note that a settings plugins may create truly generated, in-memory profiles.
The settings plugin is free to generate and persist the profiles in any way it
wishes. But the concept of "unsaved" profiles is now something explicitly handled
by keyfile plugin. Also, these "unsaved" keyfile profiles are persisted to file system
too, to the /run directory. This is great for two reasons: first of all, all
profiles from keyfile storage in fact have a backing file -- even the
unsaved ones. It also means you can create "unsaved" profiles in /run
and load them with `nmcli connection load`, meaning there is a file
based API for creating unsaved profiles.
The other advantage is that these profiles now survive restarting
NetworkManager. It's paramount that restarting the daemon is as
non-disruptive as possible. Persisting unsaved files to /run improves
here significantly.
--
In the past, NMSettingsConnection also implemented NMConnection interface.
That was already changed a while ago and instead users call now
nm_settings_connection_get_connection() to delegate to a
NMSimpleConnection. What however still happened was that the NMConnection
instance gets never swapped but instead the instance was modified with
nm_connection_replace_settings_from_connection(), clear-secrets, etc.
Change that and treat the NMConnection instance immutable. Instead of modifying
it, reference/clone a new instance. This changes that previously when somebody
wanted to keep a reference to an NMConnection, then the profile would be cloned.
Now, it is supposed to be safe to reference the instance directly and everybody
must ensure not to modify the instance. nmtst_connection_assert_unchanging()
should help with that.
The point is that the settings plugins may keep references to the
NMConnection instance, and so does the NMSettingsConnection. We want
to avoid cloning the instances as long as they are the same.
Likewise, the device's applied connection can now also be referenced
instead of cloning it. This is not yet done, and possibly there are
further improvements possible.
--
Also implement multiple keyfile directores /usr/lib, /etc, /run (rh #1674545,
bgo #772414).
It was always the case that multiple files could provide the same UUID
(both in case of keyfile and ifcfg-rh). For keyfile plugin, if a profile in
read-only storage in /usr/lib gets modified, then it gets actually stored in
/etc (or /run, if the profile is unsaved).
--
While at it, make /etc/network/interfaces profiles for ifupdown plugin reloadable.
--
https://bugzilla.gnome.org/show_bug.cgi?id=772414https://bugzilla.gnome.org/show_bug.cgi?id=744711https://bugzilla.redhat.com/show_bug.cgi?id=1674545
This is inspired by the existing systemd integration, with a few differences:
* This parses the WPAD option, which systemd requested, but did not use.
* We hook into the DAD handling, only making use of the configured address
once DAD has completed successfully, and declining the lease if it fails.
There are still many areas of possible improvement. In particular, we need
to ensure the parsing of all options are compliant, as n-dhcp4 treats all
options as opaque, unlike sd-dhcp4. We probably also need to look at how
to handle failures and retries (in particular if we decline a lease).
We need to query the current MTU at client startu, as well as the hardware
broadcast address. Both these are provided by the kernel over netlink, so
it should simply be a matter of hooking that up with NM's netlink layer.
Contribution under LGPL2.0+, in addition to stated licenses.
The library is called "libnm_core". So the dependency should be called
"libnm_core_dep", like in all other cases.
(cherry picked from commit c27ad37c27)
Routing rules are unlike addresses or routes not tied to an interface.
NetworkManager thinks in terms of connection profiles. That works well
for addresses and routes, as one profile configures addresses and routes
for one device. For example, when activating a profile on a device, the
configuration does not interfere with the addresses/routes of other
devices. That is not the case for routing rules, which are global, netns-wide
entities.
When one connection profile specifies rules, then this per-device configuration
must be merged with the global configuration. And when a device disconnects later,
the rules must be removed.
Add a new NMPRulesManager API to track/untrack routing rules. Devices can
register/add there the routing rules they require. And the sync method will
apply the configuration. This is be implemented on top of NMPlatform's
caching API.
Like also done for autotools, create and use intermediate libraries
from "shared/nm-utils/".
Also, replace "shared_dep" by "shared_nm_utils_base_dep". We don't
need super fine-grained selection of what we link. We can always
link in "shared/libnm-utils-base.a", and let the linker throw away
unsed parts.
For better or worse, we already pull in large parts of systemd sources.
I need a base64 decode implementation (because glib's g_base64_decode()
cannot reject invalid encodings). Instead of coming up with my own or
copy-paste if from somewhere, reuse systemd's unbase64mem().
But for that, make systemd's basic bits an independent static library
first because I will need it in libnm-core.
This doesn't really change anything except making "libnm-systemd-core.la"
an indpendent static library that could be used from "libnm-core". We
shall still be mindful about which internal code of systemd we use, and only
access functionality that is exposed via "systemd/nm-sd-utils-shared.h".
For P2P connections it makes sense to bind the connection to the status
of the operation that is being done. One example is that a wifi display
(miracast) P2P connection should be shut down when streaming fails for
some reason.
This new helper class allows binding a connection to the presence of a
DBus path meaning that it will be torn down if the process disappears.
This is independent functionality that only depends on linux API
and glib.
Note how "nm-logging" uses this for getting the timestamps. This
makes "nm-logging.c" itself dependen on "src/nm-core-utils.c",
for little reason.
The script didn't include all the symbols needed by plugins because
libNetworkManager.a, as built by meson, doesn't include symbols from
other static libraries that are linked in. Since we used
libNetworkManager.a to know which symbols are potentiall available
from NM, the result was an incomplete list.
Unfortunately, the only way to include the whole static library is to
create a dependency object and use 'link_whole', but this is only
available in meson >= 0.46. Since 'link_whole' is available for
executables in meson >= 0.40, create a fake executable and use that to
enumerate symbols.
Also add tests to check that plugins can be loaded correctly.
Fixes: dfa2a2b40c
- always define the SESSION_TRACKING_* defines to replace
"#ifdef" with "#if".
- drop defining the consolekit database path CKDB_PATH in
config.h. The path was not customizable via configure/meson.
- fix meson build to enable consolekit support for session tracking
without also enabling logind/elogind session tracking.
logind/elogind is mutually exclusive, but consolekit session tracking
goes together just fine.
1) the command line gets shorter. I frequently run `make V=1` to see
the command line arguments for the compiler, and there is a lot
of noise.
2) define each of these variables at one place. This makes it easy
to verify that for all compilation units, a particular
define has the same value. Previously that was not obvious or
even not the case (see commit e5d1a71396
and commit d63cf1ef2f).
The point is to avoid redundancy.
3) not all compilation units need all defines. In fact, most modules
would only need a few of these defines. We aimed to pass the necessary
minium of defines to each compilation unit, but that was non-obvious
to get right and often we set a define that wasn't used. See for example
"src_settings_plugins_ibft_cppflags" which needlessly had "-DSYSCONFDIR".
This question is now entirely avoided by just defining all variables in
a header. We don't care to find the minimum, because every component
gets anyway all defines from the header.
4) this also avoids the situation, where a module that previously did
not use a particular define gets modified to require it. Previously,
that would have required to identify the missing define, and add
it to the CFLAGS of the complation unit. Since every compilation
now includes "config-extra.h", all defines are available everywhere.
5) the fact that each define is now available in all compilation units
could be perceived as a downside. But it isn't, because these defines
should have a unique name and one specific value. Defining the same
name with different values, or refer to the same value by different
names is a bug, not a desirable feature. Since these defines should
be unique accross the entire tree, there is no problem in providing
them to every compilation unit.
6) the reason why we generate "config-extra.h" this way, instead of using
AC_DEFINE() in configure.ac, is due to the particular handling of
autoconf for directory variables. See [1].
With meson, it would be trivial to put them into "config.h.meson".
While that is not easy with autoconf, the "config-extra.h" workaround
seems still preferable to me.
[1] https://www.gnu.org/software/autoconf/manual/autoconf-2.63/html_node/Installation-Directory-Variables.html
We already do conditional build with "#if WITH_CONCHECK".
Get rid of the conditional in the makefile and instead do
conditional compilating inside the source file "nm-connectivity.c".
The advantage is, now if you want to know which parts are build,
you only need to grep for the WITH_CONCHECK preprocessor define
instead of also caring about the conditional in Makefile.am and
meson.build.
It doesn't change the fact of conditional compilation. But it
consistently uses one mechanism to achieve it.
Previously, we used the generated GDBusInterfaceSkeleton types and glued
them via the NMExportedObject base class to our NM types. We also used
GDBusObjectManagerServer.
Don't do that anymore. The resulting code was more complicated despite (or
because?) using generated classes. It was hard to understand, complex, had
ordering-issues, and had a runtime and memory overhead.
This patch refactors this entirely and uses the lower layer API GDBusConnection
directly. It replaces the generated code, GDBusInterfaceSkeleton, and
GDBusObjectManagerServer. All this is now done by NMDbusObject and NMDBusManager
and static descriptor instances of type GDBusInterfaceInfo.
This adds a net plus of more then 1300 lines of hand written code. I claim
that this implementation is easier to understand. Note that previously we
also required extensive and complex glue code to bind our objects to the
generated skeleton objects. Instead, now glue our objects directly to
GDBusConnection. The result is more immediate and gets rid of layers of
code in between.
Now that the D-Bus glue us more under our control, we can address issus and
bottlenecks better, instead of adding code to bend the generated skeletons
to our needs.
Note that the current implementation now only supports one D-Bus connection.
That was effectively the case already, although there were places (and still are)
where the code pretends it could also support connections from a private socket.
We dropped private socket support mainly because it was unused, untested and
buggy, but also because GDBusObjectManagerServer could not export the same
objects on multiple connections. Now, it would be rather straight forward to
fix that and re-introduce ObjectManager on each private connection. But this
commit doesn't do that yet, and the new code intentionally supports only one
D-Bus connection.
Also, the D-Bus startup was simplified. There is no retry, either nm_dbus_manager_start()
succeeds, or it detects the initrd case. In the initrd case, bus manager never tries to
connect to D-Bus. Since the initrd scenario is not yet used/tested, this is good enough
for the moment. It could be easily extended later, for example with polling whether the
system bus appears (like was done previously). Also, restart of D-Bus daemon isn't
supported either -- just like before.
Note how NMDBusManager now implements the ObjectManager D-Bus interface
directly.
Also, this fixes race issues in the server, by no longer delaying
PropertiesChanged signals. NMExportedObject would collect changed
properties and send the signal out in idle_emit_properties_changed()
on idle. This messes up the ordering of change events w.r.t. other
signals and events on the bus. Note that not only NMExportedObject
messed up the ordering. Also the generated code would hook into
notify() and process change events in and idle handle, exhibiting the
same ordering issue too.
No longer do that. PropertiesChanged signals will be sent right away
by hooking into dispatch_properties_changed(). This means, changing
a property in quick succession will no longer be combined and is
guaranteed to emit signals for each individual state. Quite possibly
we emit now more PropertiesChanged signals then before.
However, we are now able to group a set of changes by using standard
g_object_freeze_notify()/g_object_thaw_notify(). We probably should
make more use of that.
Also, now that our signals are all handled in the right order, we
might find places where we still emit them in the wrong order. But that
is then due to the order in which our GObjects emit signals, not due
to an ill behavior of the D-Bus glue. Possibly we need to identify
such ordering issues and fix them.
Numbers (for contrib/rpm --without debug on x86_64):
- the patch changes the code size of NetworkManager by
- 2809360 bytes
+ 2537528 bytes (-9.7%)
- Runtime measurements are harder because there is a large variance
during testing. In other words, the numbers are not reproducible.
Currently, the implementation performs no caching of GVariants at all,
but it would be rather simple to add it, if that turns out to be
useful.
Anyway, without strong claim, it seems that the new form tends to
perform slightly better. That would be no surprise.
$ time (for i in {1..1000}; do nmcli >/dev/null || break; echo -n .; done)
- real 1m39.355s
+ real 1m37.432s
$ time (for i in {1..2000}; do busctl call org.freedesktop.NetworkManager /org/freedesktop org.freedesktop.DBus.ObjectManager GetManagedObjects > /dev/null || break; echo -n .; done)
- real 0m26.843s
+ real 0m25.281s
- Regarding RSS size, just looking at the processes in similar
conditions, doesn't give a large difference. On my system they
consume about 19MB RSS. It seems that the new version has a
slightly smaller RSS size.
- 19356 RSS
+ 18660 RSS
The next commit will completely rework NMBusManager and replace
NMExportedObject by a new type NMDBusObject.
Originally, NMDBusObject was added along NMExportedObject to ease
the rework and have compilable, intermediate stages of refactoring. Now,
I think the new name is better, because NMDBusObject is very strongly related
to the bus manager and the old name NMExportedObject didn't make that
clear.
I also slighly prefer the name NMDBusObject over NMBusObject, hence
for consistancy, also rename NMBusManager to NMDBusManager.
This commit only renames the file for a nicer diff in the next commit.
It does not actually update the type name in sources. That will be done
later.
From libnl3, we only used the helper function to parse/generate netlink
messages and the socket functions to send/receive messages. We don't
need an external dependency to do that, it is simple enough.
Drop the libnl3 dependency, and replace all missing code by directly
copying it from libnl3 sources. At this point, I mostly tried to
import the required bits to make it working with few modifications.
Note that this increases the binary size of NetworkManager by 4736 bytes
for contrib/rpm build on x86_64. In the future, we can simplify the code
further.
A few modifications from libnl3 are:
- netlink errors NLE_* are now in the domain or regular errno.
The distinction of having to bother with two kinds of error
number domains was annoying.
- parts of the callback handling is copied partially and unused parts
are dropped. Especially, the verbose/debug handlers are not used.
In following commits, the callback handling will be significantly
simplified.
- the complex handling of seleting ports was simplified. We now always
let kernel choose the right port automatically.
Especially useful, because we don't link against libnl-genl-3.so
but re-implement generic netlink support. Such code should go there
so it can be used by various components.
We also unconditionally use them with autotools.
Also, the detection for have_version_script does
not seem correct to me. At least, it didn't work
with clang.
The strings holding the names used for libraries have also been
moved to different variables. This way they would be less error
as these variables can be reused easily and any typing error
would be quickly detected.
Some targets are missing dependencies on some generated sources in
the meson port. These makes the build to fail due to missing source
files on a highly parallelized build.
These dependencies have been resolved by taking advantage of meson's
internal dependencies which can be used to pass source files,
include directories, libraries and compiler flags.
One of such internal dependencies called `core_dep` was already in
use. However, in order to avoid any confusion with another new
internal dependency called `nm_core_dep`, which is used to include
directories and source files from the `libnm-core` directory, the
`core_dep` dependency has been renamed to `nm_dep`.
These changes have allowed minimizing the build details which are
inherited by using those dependencies. The parallelized build has
also been improved.
Note that:
- we compile some source files multiple times. Most notably those
under "shared/".
- we include a default header "shared/nm-default.h" in every source
file. This header is supposed to setup a common environment by defining
and including parts that are commonly used. As we always include the
same header, the header must behave differently depending
one whether the compilation is for libnm-core, NetworkManager or
libnm-glib. E.g. it must include <glib/gi18n.h> or <glib/gi18n-lib.h>
depending on whether we compile a library or an application.
For that, the source files need the NETWORKMANAGER_COMPILATION #define
to behave accordingly.
Extend the define to be composed of flags. These flags are all named
NM_NETWORKMANAGER_COMPILATION_WITH_*, they indicate which part of the
build are available. E.g. when building libnm-core.la itself, then
WITH_LIBNM_CORE, WITH_LIBNM_CORE_INTERNAL, and WITH_LIBNM_CORE_PRIVATE
are available. When building NetworkManager, WITH_LIBNM_CORE_PRIVATE
is not available but the internal parts are still accessible. When
building nmcli, only WITH_LIBNM_CORE (the public part) is available.
This granularily controls the build.
Since meson 0.44 there is a new option type called `array`, which
allows to use an array with different values in those options.
These fits the needs of different options that are used to pass
binary paths, which have multiple paths as an alternate locations.
meson's version has been bumped to 0.44 and different options have
been changed to `array` type options.
https://mail.gnome.org/archives/networkmanager-list/2017-December/msg00062.html
Source files for enum types are generated by passing segments of the
source code of the files to the `glib-mkenums` command.
This patch removes those parameters where source code is used from
meson build files by moving those segmeents to template files.
https://mail.gnome.org/archives/networkmanager-list/2017-December/msg00057.html
