The DHCP plugins are linked statically and don't have a plugin mechanism
to be loaded from a shared library. Change the _nm_dhcp_client_register()
mechanism to something more static.
Also, only link the plugins into the binary if they are actually
enabled. Previously, they would always be linked in (and always
register themself). However, nm_dhcp_dh*_get_path() would return NULL
which made the plugin unusable.
The autoconf code to detect the DHCP plugins is still not stellar, but
seems to work well enough for now. At least, we log the result of the
detection at the end of the configure-script, so a user can at least
notice what happend.
- use _NM_GET_PRIVATE() and _NM_GET_PRIVATE_PTR() everywhere.
- reorder statements, to have GObject related functions (init, dispose,
constructed) at the bottom of each file and in a consistent order w.r.t.
each other.
- unify whitespaces in signal and properties declarations.
- use NM_GOBJECT_PROPERTIES_DEFINE() and _notify()
- drop unused signal slots in class structures
- drop unused header files for device factories
- get_property() should be imidiately before set_property().
- type_init() should be before constructing(), constructed()
and type_new().
- dispose() and finalize() should be after object creation.
- at last follows class_init().
Our internal copy of systemd should not be in the search path.
Instead, let users only
#include "systemd/nm-sd.h"
which then includes everything from systemd that we need.
We want to avoid to accidentally include anything from our
systemd-copy. Any user of that should only include "nm-sd.h",
which then includes everything that is needed further.
For example, "src/devices/wwan/nm-modem-manager.c" has
#include <systemd/nm-daemon.h>
which in turn includes
#include "_sd-common.h"
This works all correctly before, because #include "" will first
look in the directory where sd-daemon.h is. However, our mixing of
external systemd library and internal copy is rather dangerous.
Try to avoid it further by keeping the include paths clean.
A D-Bus signal is asynchronous and it can happen that nm-dhcp-helper
emits the "Event" signal before the server is able to register a handler:
NM_DHCP_HELPER=/usr/libexec/nm-dhcp-helper
nmcli general logging level TRACE
for i in `seq 1 500`; do $NM_DHCP_HELPER & done
journalctl -u NetworkManager --since '1 min ago' | grep "didn't have associated interface" | wc -l
499
Avoid that, by calling the synchronous D-Bus method "Notify".
Interestingly, this race seem to exist since 2007.
Actually, we called g_dbus_connection_signal_subscribe() from inside
GDBusServer:new-connection signal. So it is not clear how such a race
could exist. I was not able to reproduce it by putting a sleep
before g_dbus_connection_signal_subscribe(). On the other hand, there
is bug rh#1372854 and above reproducer which strongly indicates that
events can be lost under certain circumstances.
Now we instead g_dbus_connection_register_object() from the
new-connection signal. According to my tests there was no more race
as also backed by glib's documentation. Still, keep a simple retry-loop
in nm-dhcp-helper just to be sure.
https://bugzilla.redhat.com/show_bug.cgi?id=1372854https://bugzilla.redhat.com/show_bug.cgi?id=1373276
- don't include "nm-default.h" in header files. Every source file must
include as first header "nm-default.h", thus our headers get the
default include already implicitly.
- we don't support compiling NetworkManager itself with a C++ compiler. Remove
G_BEGIN_DECLS/G_END_DECLS from internal headers. We do however support
users of libnm to use C++, thus they stay in public headers.
(cherry picked from commit f19aff8909)
Users outside of src/systemd should not make use of internal API.
Currently, "nm-dhcp-systemd.c" still makes use of internal systemd
functions. Instead of letting "nm-dhcp-systemd.c" include internal
headers, handpick the required defines to "nm-sd.h" and hide "nm-sd-adapt.h".
"nm-sd-adapt.h" is now only used to compile internal systemd sources.
- reorder entries in src/Makefile.am so that general names
are all at the beginning (AM_CPPFLAGS, sbin_PROGRAMS)
and the names for a certain library/binary are grouped
together
- have libNetworkManager.la reuse libNetworkManagerBase.la.
- let all components in src/Makefile.am use the same AM_CPPFLAGS,
except libsystem-nm.la.
- move callouts/nm-dispatcher-api.h to shared/ directory. It
is obviously not internal API to callouts, and callouts is
not a library. Thus, the right place is shared/.
The dhclient and dhcpcd clients can be destroyed during disposal of
the DHCP manager singleton and at that point the NMDhcpListener
singleton can be already gone. Reference it in the clients.
When the ipv4.dhcp-fqdn property is set, NM adds the following options
to dhclient.conf:
send fqdn.fqdn "foo.bar";
send fqdn.encoded on;
send fqdn.server-update on;
which enable the S (server-update) and E (encoded) flags in DHCP
option 81, since they are sensible default values and dhclient
requires a "send fqdn.server-update [on|off]" directive in order to
send the option.
Users may want to change these flags according to their server's
configuration, but this is not possible at the moment since NM options
are placed after user's ones, overriding them.
To fix this, collect user's fqdn options and add them after NM
configuration; note that the fqdn.fqdn option still can't be
overridden by users, as NM must control the FQDN sent to server.
Fixes: c3573ebf2b
RFC 4704 ("The DHCPv6 Client FQDN Option") allows both partial and
fully-qualified names in the FQDN option, however dhclient always
appends a terminating zero-length label to the name, so we ignore
unqualified hostnames to prevent a wrong configuration.
Emit a warning when the field is ignored so that users can clearly see
why the hostname is not being sent.
internal systemd code produces logging messages by itself, like
libsystemd: DHCP CLIENT (0x9204b5ce): ACK
Let's log the pointer value initially, to associate the logged "xid"
with the pointer value of the client.
Now we get:
<trace> [1464520695.7655] dhcp4 (enp0s25): dhcp-client4: set 0x556cdd9d6800
<debug> [1464520695.7658] libsystemd: DHCP CLIENT (0x9d87b7c5): STARTED on ifindex 2
A large part of "nm-test-utils.h" is only relevant for tests inside "src/"
directory, as they are helpers related to NetworkManager core part.
Split this part out of "nm-test-utils.h" header.
It is recomended to include <config.h> with angle brackets [1].
Note, that usually we don't include <config.h> directly, except in
two places we have to (because there we include conflicting libraries
that must be included before "nm-default.h").
In that case, define __CONFIG_H__ which is used as include guard around
<config.h> by "nm-default.h".
[1] https://www.gnu.org/software/autoconf/manual/autoconf-2.66/html_node/Configuration-Headers.html
The "source" field of NMPlatformIPRoute (now "rt_source") maps to the
protocol field of the route. The source of NMPlatformIPAddress (now
"addr_source") has no direct equivalent in the kernel.
As their use is different, they should have different names. Also,
the name "source" is used all over the place. Hence give the fields
a more distinct name.
For internal compilation we want to be able to use deprecated
API without warnings.
Define the version min/max macros to effectively disable deprecation
warnings.
However, don't do it via CFLAGS option in the makefiles, instead hack it
to "nm-default.h". After all, *every* source file that is for internal
compilation needs to include this header as first.
If the user specifies an invalid 'dhcp' option in configuration,
currently we disable DHCP. Instead, fall back to other available
clients, as we do for other options.
The macro _LOGx_ENABLED() is defined with a default implementation
that depends on _NMLOG_DOMAIN. Although that default does not
check for LOGD_DHCP4 vs. LOGD_DHCP6, still provide it.
Determining the correct domain might involve a larger performance
impact that what we would safe.
Now we have:
"nm-sd.h" is a header file of NetworkManager with utilities
related to systemd. It can be used anywhere freely.
Also, systemd headers that are considered public API (like
"sd-event.h") can be used without restrictions.
When compiling the systemd sources, we always must include
"nm-sd-adapt.h" as first. Similarly, systemd headers must
not include "nm-sd-adapt.h", because they are either public
(in which case the adapter is not needed) or they are internal
(in which case they are themself included via a systemd source).
Sometimes, we must internal API (like "dhcp-lease-internal.h").
In this case, we also must include "nm-sd-adapt.h".
GError codes are only unique per domain, so logging the code without
also indicating the domain is not helpful. And anyway, if the error
messages are not distinctive enough to tell the whole story then we
should fix the error messages.
Based-on-patch-by: Dan Winship <danw@gnome.org>
Functions that take a GError** MUST fill it in on error. There is no
need to check whether error is NULL if the function it was passed to
had a failing return value.
Likewise, a proper GError must have a non-NULL message, so there's no
need to double-check that either.
Based-on-patch-by: Dan Winship <danw@gnome.org>
Until now the internal DHCP client could start a DHCPv6 transaction
but was not able to parse the lease and pass the information back to
the core. Add the missing glue code to make this work.
https://bugzilla.gnome.org/show_bug.cgi?id=762432
