Do some preprocessing on the DNS configuration sent to plugins:
- add the '~' default routing (lookup) domain to IP configurations
with the default route or, when there is none, to all non-VPN
IP configurations
- use the dns-priority to decide which connection to use in case
multiple connections have the same domain
- consider a negative dns-priority value as a way to 'shadow' all
subdomains from other connections
- compute reverse DNS domains
and add the resulting domain list to NMDnsIPConfigData so that
split-DNS plugins can use that directly instead of reimplementing the
same logic themselves.
- consistently set options, searches, domains fields to %NULL,
if there are no values.
- in nm_global_dns_config_update_checksum(), ensure that we uniquely
hash values. E.g. a config with "searches[a], options=[b]" should
hash differently from "searches=[ab], options=[]".
- in nm_global_dns_config_to_dbus(), reuse the sorted domain list.
We already have it, and it guarantees a consistent ordering of
fields.
- in global_dns_domain_from_dbus(), fix memleaks if D-Bus strdict
contains duplicate entries.
I dislike the static hash table to cache the integer counter for
numbered paths. Let's instead cache the counter at the class instance
itself -- since the class contains the information how the export
path should be exported.
However, we cannot use a plain integer field inside the class structure,
because the class is copied between derived classes. For example,
NMDeviceEthernet and NMDeviceBridge both get a copy of the NMDeviceClass
instance. Hence, the class doesn't contain the counter directly, but
a pointer to one counter that can be shared between sibling classes.
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.
Previously we always updated resolv.conf on quit. When we are using
systemd-resolved the update is not necessary because the resolver on
127.0.0.53 would still be reachable after NM quits. Also, when NM
manages resolv.conf directly there is no need to update the file
again. Let's rewrite resolv.conf only when using dnsmasq.
https://bugzilla.redhat.com/show_bug.cgi?id=1541031
Fixes the following error when building with gcc 4.8.5 and address
sanitizer:
src/dns/nm-dns-dnsmasq.c: In function 'update':
src/dns/nm-dns-dnsmasq.c:506:44: error: 'first_prio' may be used uninitialized in this function [-Werror=maybe-uninitialized]
} else if (first_prio < 0 && first_prio != prio)
^
Similarly to what systemd-resolved does, introduce the concept of
"routing" domain, which is a domain in the search list that is used
only to decide the interface over which a query must be forwarded, but
is not used to complete unqualified host names. Routing domains are
those starting with a tilde ('~') before the actual domain name.
Domains without the initial tilde are used both for completing
unqualified names and for the routing decision.
The "domain" key of the D-Bus configuration dictionary specifies the
domains a configuration applies to. In DNS code we consider domains
and searches as equivalent, so they should be exported via D-Bus using
the same logic used to populate resolv.conf and for plugins.
The connection.mdns setting is a per-connection setting,
so one might expect that one activated device can only have
one MDNS setting at a time.
However, with certain VPN plugins (those that don't have their
own IP interface, like libreswan), the VPN configuration is merged
into the configuration of the device. So, in this case, there
might be multiple settings for one device that must be merged.
We already have a mechanism for that. It's NMIP4Config. Let NMIP4Config
track this piece of information. Although, stricitly speaking this
is not tied to IPv4, the alternative would be to introduce a new
object to track such data, which would be a tremendous effort
and more complicated then this.
Luckily, NMDnsManager and NMDnsPlugin are already equipped to
handle multiple NMIPConfig instances per device (IPv4 vs. IPv6,
and Device vs. VPN).
Also make "connection.mdns" configurable via global defaults in
NetworkManager.conf.
Don't track the per-device configuration in NMDnsManager by
the ifname, but by the ifindex. We should consistently treat
the ifindex as the ID of a link, like kernel does.
At the few places where we actually need the ifname, resolve
it by looking into the platform cache. That is not necessarily
the same as the ifname that is currently tracked by NMDevice,
because netdev interfaces can be renamed, and NMDevice updates
it's link properties delayed. However, the platform cache has
the most recent notion of the correct interface name for an
ifindex, so if we ever hit a race here, we do it now more
correctly.
This also temporarily drops support for mdns. Will be re-added next,
but differently.
We had two separate queues, one for "SetLinkDNS" and one for
"SetLinkDomains". Merge them into one, and track the operation
as part of the new RequestItem structure.
A visible change to before is that we now would make all requests
per-interface first. Prevously, we would first make all SetLinkDNS
requests (for all interfaces) and then all SetLinkDomains requests.
It feels more correct to order the requests this way, not by
type.
The reason to merge is, that we will next get another operation
and in the current scheme we would need 3 GQueue instances.
While at it, refactor the code to use CList. We now anyway would
need a new struct to track the operation, requiring to allocate
and free it. Previously, we would only track the GVariant argument
as data of the GQueue.
Use a GHashTable instead of a GArray to construct the list of
@interfaces. Also, use NMCListElem instead of GList. With this,
the runtime is O(n*log(n)) instead of O(n^2).
I belive, we should take care that all our code has a reasonable
runtime complexity, even in common use-cases the number of elements
is small. This is not about performace, because likely we expect few
entries anyway, and the direct GArray implementation is likely faster
in those cases. It's about using the data structure that best suits the
access pattern.
The log(n) part comes from sorting the keys. I also believe we should
always aim for a stable behavior. When sending the D-Bus request to
resolved, the order of elements should be in ~some~ defined order.
"UNKNOWN" is not a good name. If you don't set the property
in the connection explicitly, it should be "DEFAULT".
Also, make "DEFAULT" -1. For one, that ensures that the enum's
underlying integer type is signed. Otherwise, it's cumbersome
to test "if (mdns >= DEFAULT)" because in case of unsigned types,
the compiler will warn about the check always being true.
Also, it allows for "NO" to be zero. These are no strong reasons,
but I tend to think this is better.
Also, don't make the property of NMSettingConnection a CONSTRUCT property.
Initialize the default manually in the init function.
Also, order the numeric values so that DEFAULT < NO < RESOLVE < YES with
YES being largest because it enables *the most*.
Update nm-policy.c and nm-dns-manager.c so that the connection-specific
settings get propagated to DNS manger. Currently the only such value is
the mDNS status.
Add update_mdns() function to DNS plugin interface. If a DNS plugin
supports mDNS, it can set an interface with a given index to support
mDNS resolving or also register the current hostname.
The mDNS support is currently added only to systemd-resolved DNS plugin.
The compiler warns when we ignore the return value from write().
And assigning it to an unused variable, causes another warning.
Make some use of it, at least to handle EINTR. All other errors
are still ignored.
While at it, rework the write code to first write to a buffer
in memory.
src/dns/nm-dns-manager.c: In function ‘write_to_netconfig’:
src/dns/nm-dns-manager.c:387:8: error: ignoring return value of ‘write’, declared with attribute warn_unused_result [-Werror=unused-result]
write (fd, str, strlen (str));
^
src/settings/plugins/ifnet/nms-ifnet-connection-parser.c: In function ‘ifnet_update_parsers_by_connection’:
src/settings/plugins/ifnet/nms-ifnet-connection-parser.c:2600:26: error: variable ‘pppoe’ set but not used [-Werror=unused-but-set-variable]
gboolean wired = FALSE, pppoe = TRUE;
^~~~~
While at it, don't log line breaks in ifnet_update_parsers_by_connection().
Fixes: e912b36d95
Fix resolved detection, the symlink target is usually relative to the
root, such that in chroots the file points to a file inside the
chroot. But keep absolute targets too, as these may have been in use
with older version of systemd. Add support for stub-resolv.conf
detection.
https://bugzilla.gnome.org/show_bug.cgi?id=790446
Instead of having 3 properties @gateway, @never_default and @has_gateway
on NMIP4Config/NMIP6Config that determine the default-route, track the
default-route as a regular route.
The gateway setting is the configuration knob for the default-route.
Since an NMIP4Config/NMIP6Config instance only has one gateway property,
it cannot track more then one default-routes (see related bug rh#1445417).
Especially with policy routing, it might be interesting to configure a
default-route in multiple tables.
Also, later it might be interesting to allow adding default-routes as
regular static routes in a connection, so that the user can configure additional
route parameters for the default-route or add default-routes in multiple tables.
With this patch, default-routes now have a rt_source property according to their
origin.
Also, the previous commits of this branch broke handling of the
default-route :) . That should be working now again.
Default-routes are for the most part like regular routes. Add support to
track them like regular routes in NMIP4Config/NMIP6Config.
One thing is, sometimes we need to figure out whether an ip-config
instance has a default-route. For that, keep track of the best
default-route (there might be multiple) and expose it. That is
the most complicated part of this patch, because there are so many
places where the list of routes gets modified (replace, intersect,
subtract, merge, add), and they all need to take care of updating
the best default-route.
In a next patch, NMDefaultRouteManager will be dropped and default-routes
will be tracked by NMIP4Config/NMIP6Config.
Reasons:
- it adds an O(1) lookup index for accessing NMIPxConfig's addresses.
Hence, operations like merge/intersect have now runtime O(n) instead
of O(n^2).
Arguably, we expect low numbers of addresses in general. For low
numbers, the O(n^2) doesn't matter and quite likely in those cases
the previous implementation was just fine -- maybe even faster.
But the simple case works fine either way. It's important to scale
well in the exceptional case.
- the tracked objects can be shared between the various NMPI4Config,
NMIP6Config instances with NMPlatform and everybody else.
- the NMPObject can be treated generically, meaning it enables code to
handle both IPv4 and IPv6, or addresses and routes. See for example
_nm_ip_config_add_obj().
- I want core to evolve to somewhere where we don't keep copies of
NMPlatformIP4Address, et al. instances. Instead they shall all be
shared. I hope this will reduce memory consumption (although tracking a
reference consumes some memory too). Also, it shortcuts nmp_object_equal()
when comparing the same object. Calling nmp_object_equal() on the
identical objects would be a common case after the hash function
pre-evaluates equality.
The DNS manager drops from the search list domains that are public
suffixes to prevent a possible domain hijack when using two-labels
hostnames [1].
This is a problem now that every single-label domain can be a TLD
since this means that such domains can't be used in the search list.
While it's useful to apply such restriction to the domain
automatically derived from the system hostname, it seems wrong to drop
domains specified by users in the configuration or provided by DHCP.
This commit keeps the public-suffix check only for the
hostname-derived domain
[1] https://bugzilla.redhat.com/show_bug.cgi?id=812394https://bugzilla.redhat.com/show_bug.cgi?id=1404350
Eventually, every NMPlatformIP4Route, NMPlatformIP6Route,
NMPlatformIP4Address and NMPlatformIP6Address should be shared
an deduplicated via the global NMDedupMultiIndex instance.
As first proof of concept, refactor NMIP4Config to track
IPv4 routes via the shared multi_idx. There is later potential
for improvement, when we pass (deduplicated) NMPObject instances
around instead of plain NMPlatformIP4Route, which needs still
a lot of comparing and cloning.
No need to clone the list anymore. Unfortunately, GPtrArray is not NULL
terminated (without extra effort), so we have to pass on the GPtrArray
instance for the length.
A negative ipv4.dns-priority and ipv6.dns-priority has the meaning to configure
the DNS information of the connection exclusively. With systemd-resolved, that means
we must explicitly unset the configuration from other interfaces.
https://bugzilla.gnome.org/show_bug.cgi?id=783569
