- nm_streq() is easier to read.
- NM_STR_HAS_PREFIX() works only with string literals, and gets
inlined up to one strncmp() call. There is no need to call glib,
to determine strlen(prefix) (that we already know), and then to
call strncmp().
On Fedora/RHEL, the default for main.plugins is "ifcfg-rh". You would
expect that a single configuration file
[main]
plugins-=ifcfg-rh
would result in an empty list of plugins (which subsequently gets
completed with "keyfile").
That didn't happen due to a bug. Fix it.
lgtm.com warns:
int nm_owned:3;
>> Bit field nm_owned of type int should have explicitly unsigned integral, explicitly signed integral, or enumeration type.
So make it a NMTernary instead. It's nicer anyway.
lgtm.com flags this. The check was there to be better safe than sorry.
Also, it seems better to have code that shows what happens instead
of a verbose code comment (or no comment at all). Anyway, avoid the
false positive.
Reported by coverity:
>>> CID 210222: Null pointer dereferences (NULL_RETURNS)
>>> Dereferencing a pointer that might be "NULL" "f" when calling
"fseek".
Fixes: ac5206aa9c ('2007-11-21')
If the driver is unknown, that doesn't necessarily mean that the match
passes. Instead, the match passes if there is no positive match that
asks for the driver name.
%NULL means that the string is unknown. The pattern should still match
if there are no positive matches that want to match against the string.
For example, the nm_device_get_driver() might return NULL. If we have
a match.driver setting, we still need to handle that somehow that it
makes sense.
- write_match_setting() never fails. Don't let it return a boolean
error result.
- drop "if (!name || !name[0])" checks. It's not possibly to configure
a name %NULL in NMSettingMatch (without triggering assertions). Also,
an empty name "" is not valid, so we wouldn't expect it. There is one
problem with the way how we concatenate the string list: it uses
spaces as separator, while stripping spaces. That means, in the
currenty format, an empty token "" cannot be expressed. On the other
hand, serializing it would lead to duplicate spaces, that get dropped
during re-read. So the empty name wasn't valid from the start, but it
also cannot be encoded.
- use nm_gstring_add_space_delimiter() and nm_gstring_prepare().
Add a new "driver" match option to nm-settings. It allows to disable a
network connection configuration if a pattern is found or is not found
in the device driver name.
Add a new "kernel-command-line" match option to nm-settings. It allows
to disable a network connection configuration if a pattern is found or
is not found in /proc/cmdline.
This warning is from coverity against 1.18.6. But it applies
in a similar manner here.
1. NetworkManager-1.18.6/src/devices/nm-device-macsec.c:811:25: warning: Value stored to 'priv' during its initialization is never read
# NMDeviceMacsecPrivate *priv = NM_DEVICE_MACSEC_GET_PRIVATE (self);
# ^~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4. NetworkManager-1.18.6/src/devices/nm-device-macsec.c:811:25: note: Value stored to 'priv' during its initialization is never read
# NMDeviceMacsecPrivate *priv = NM_DEVICE_MACSEC_GET_PRIVATE (self);
# ^~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# 809| {
# 810| NMDeviceMacsec *self = NM_DEVICE_MACSEC (object);
# 811|-> NMDeviceMacsecPrivate *priv = NM_DEVICE_MACSEC_GET_PRIVATE (self);
# 812|
# 813| macsec_secrets_cancel (self);
Also, silently ignore all environment variables with a name that
is not valid UTF-8. We would hit an assertion trying to put that
in a GVariant (or sending it via D-Bus).
