The _nm_alignas() exists to choose a suitable alignment. Since it's
on the stack, it has (almost) no overhead to just use the maximum
alignment.
That's why gint64 was chosen before. But that isn't the largest
alignment. We rely on C11 already, and we also always include <stddef.h>.
So use max_align_t instead.
NM_CMP_DIRECT_MEMCMP() gets called by NM_CMP_FIELD_MEMCMP_LEN().
For example, if you want to compare a NMIPAddr, it seems sensible
to call
NM_CMP_FIELD_MEMCMP_LEN(obj1, obj2, addr, nm_utils_addr_family_to_size(obj1->addr_family));
Granted, nm_utils_addr_family_to_size() asserts that addr_family is
either AF_INET or AF_INET6. However, if the assertion fails, we don't
want yet another undefined behavior here and do the sensible thing
about n zero.
In general, for a low-level function that uses memcmp(), it's non
obvious to ensure that the caller does not accidentally invoke undefined
behavior. nm_memcmp() avoids that.
- use proper integer types. A netlink message cannot be as large as
size_t, because the length is tracked in an uint32_t. Use the
right types.
- fields like "nlmsg_type" or "nlmsg_flags" are uint16_t. Use the
right types.
- note that nlmsg_size() still returns and accepts "int". Maybe
the should be adjusted too, but we use macros from kernel headers,
which also use int. Even if that is not the type of the length on
the binary protocol. So some of these functions still use int, to
be closer and compatible with <linux/netlink.h>.
For generic netlink, the family-id is important. It changes when
loading/unloading a module, so we should not cache it indefinitely.
To get this right, takes some effort. For "nl80211", "nl802154"
and "wireguard", we only cache the family ID in relation to an
interface. If the module gets unloaded, the family ID also becomes
irrelevant and we need to re-fetch it the next time.
For generic families like "mptcp_pm" or "ethtool", they are commonly not
kernel modules and cannot be unloaded. So caching them would be
(probably) fine.
Still. Some generic netlink families emit notifications, and it will
be interesting to be able to handle them. Since that will be useful later,
start by doing something simple: let the generic netlink family also be
cached this way. Generic netlink will send notifications when a family gets
added/deleted, and we can use that to reliably cache the family ID.
We only care about a well-known set of generic families. Unlike libnl
(which has "struct genl_family" object to handle any family), we can hard
code the few we care about (NMPGenlFamilyType).
This adds the necessary infrastructure of NMLinuxPlatform to listen to
events on the generic netlink socket.
Currently there is no problem. However, DelayedActionType is a packed
enum, and if we add a few more enum values, it might happen that
DELAYED_ACTION_TYPE_MAX is 0x8000 and DelayedActionType effectively
uint16_t.
When that happens, the code would become an infinite loop, because
0x8000 is not larger than DELAYED_ACTION_TYPE_MAX, but `<<= 1`
shifts out the bit, making it zero.
Avoid that.
If nm_platform_get_cache_tc() is disabled, there is no need to refresh
it. Filter those flags out.
Also, don't duplicate the code and add a helper function
delayed_action_schedule_refresh_all().
Reorder fields in DelayedActionWaitForNlResponseData, so that
the struct size is optimal due to the alignment constraints.
Also, when we remember enum values and embed them somewhere, it's nice
if they only take the space actually needed. _nm_packed solves that.
The term "addr_family" is used very frequently, and it usually is an
auto variable or a function parameter.
It is interesting to search where this field is used. So rename to give
it a unique (and better fitting) name.
While at it, use gint8 to encode the addr_family. It's always
sufficient, and this reduces the size of RefreshAllInfo from 8 bytes
to two.
The genl types that we care about are well known. Add an enum
for them, so we can do a lookup by index.
To kernel, the corresponding names (like "wireguard") are also well
known. However, the family-id, that we need when using genl are
allocated dynamically. So we need to lookup the family-id, and by having
an enum for the genl type, we can do so generically.
When NetworkManager runs in initrd, it can be cumbersome to enable debug logging.
Granted, when using dracut, the NetworkManager dracut module will honor "rd.debug".
However, a user may use NetworkManager in initrd without dracut. Then,
the only way to enable debug logging would be by changing
"NetworkManager.conf" and rebuild the initrd (or having some script in
place, that allows to more conveniently enable debug logging for
NetworkManager).
To make it easier for debugging, honor "nm.debug" on the kernel command
line.
Note that if "nm.debug" is set on the kernel command line, it always overrides
both the command line arguments and the configuration from NetworkManager.conf.
That is intentional. The only way to override that is by overriding the
kernel command line with a file "/run/NetworkManager/proc-cmdline".
https://bugzilla.redhat.com/show_bug.cgi?id=2102313
We read /proc/cmdline for "match.kernel-command-line". But next we will
also honor "nm.debug" on the kernel command line, to enable debug
logging. For "nm.debug" it makes sense that it overwrites the debug
options from the command line and from "NetworkManager.conf". That
means, if you set "nm.debug", then verbose logging will be enabled. It
can only be turned off again at runtime (via D-Bus), otherwise, it's
hard to avoid.
It still can make sense to overrule this setting once again. Support
that, by honoring a file "/run/NetworkManager/proc-cmdline" to be used
instead of "/proc/cmdline".
This option is mainly for debugging and testing, but it might be useful
in production too, if you had "nm.debug" enabled during boot, but later
want to disable it until next reboot. Then you could do:
sed 's/ *\<nm\.debug\> */ /g' /proc/cmdline > /run/NetworkManager/proc-cmdline
nmcli general logging level DEFAULT domains DEFAULT
GPtrArray requires an additional heap allocation for the GPtrArray.
Utterly useless in the majority of cases.
Anyway. Allocating (and exponentially grown) a buffer is not too hard,
just slightly more cumbersome. Since nm_utils_strsplit_quoted() is
heavily unit tested and entirely self-contained, let's opt for the
more complicated implementation and avoid the extra allocation.
Before 1.36, manual addresses from the profile were assigned to the
interface; restore that behavior.
The manual IP configuration also contains the DNS priority from the
profile; so this change ensures that the merged l3cd has a DNS
priority and that dynamically discovered DNS servers are not ignored
by the DNS manager.
Fixes: 58287cbcc0 ('core: rework IP configuration in NetworkManager using layer 3 configuration')
Improve documentation by preserving paragraphs in the
nm-settings-nmcli man pages.
To do that structure of src/libnm-client-impl/nm-settings-docs-gir.xml
was changed to have "description" as subnode to property node instead
of attribute of property node. Another subnode "description-docbook"
was added - this node is then used when generating man pages.
tools/generate-docs-nm-settings-docs-gir.py and man/nm-settings-dbus.xsl
were also changed to accomodate for changes mentioned above.
Replace xsltproc tool with python script when generating
./src/libnmc-setting/settings-docs.h.
Deleted settings-docs.xsl since it was replaced by python script.
Change src/libnmc-setting/settings-docs.h.in accodring to newly
generated src/libnmc-setting/settings-docs.h
https://gitlab.freedesktop.org/NetworkManager/NetworkManager/-/issues/661https://gitlab.freedesktop.org/NetworkManager/NetworkManager/-/merge_requests/1260
While iterating the context (once), multiple sources can be dispatched.
So if we get a timeout and shortly after the weak-ref callback, then
we still need to honor the weak-ref.
In particular, because weak_ref.unref() is not safe to do after
the object was already destroyed. So we need to be correct about
tracking destruction.
When creating one profile for each veth during activation the creation
of the veth could fail. When the link for the first profile is created
the link for the peer is generated in kernel. Therefore when trying to
activate the second profile it will fail because the link already
exists. NetworkManager must check if the link already exists and
corresponds to the same veth, if so, it should skip the link creation.
https://bugzilla.redhat.com/show_bug.cgi?id=2036023https://bugzilla.redhat.com/show_bug.cgi?id=2105956
In initrd, a too short carrier timeout means that the machine will
possibly fail to boot. On the other hand, increasing the value doesn't
have side effects, except for a bit longer delay on some machines.
Increase the value to 10 seconds. Note that the default value is not
propagated to the real root.
https://gitlab.freedesktop.org/NetworkManager/NetworkManager/-/merge_requests/1239
Resolve the defaults in build.sh instead of RPM macros. This looks less
terrible maintaining the same defaults as well as options to override it
upstream.
Moving it to the block that downstreams (Fedora, Red Hat) keep
customized makes it possible for them to also maintain customized
defaults here.
In particular, the downstreams should be able to enable bcond_test
at least for their production release (otherwise there's little point in
actually running tests at package build time).
https://gitlab.freedesktop.org/NetworkManager/NetworkManager/-/merge_requests/1286
