S390 options are stored in a separate [ethernet-s390-options] section.
This group must not be interpreted as a NMSetting name, otherwise we
log a bogus warning:
<warn> [1590523563.7757] keyfile: ethernet-s390-options: invalid setting name 'ethernet-s390-options'
Fixes: cf9b8d3bad ('libnm/keyfile: implement ethernet.s390-options in keyfile')
We are going to expose some of this API in libnm.
The name "gendata" (for "generic data") is not very suited. Instead,
call the public API nm_setting_option_*(). This also brings no naming
conflict, because currently no API exists with such naming.
Rename the internal API, so that it matches the API that we are going
to expose next.
We already have NMEthtoolID to handle coalesce options in a way that is
convenient programmatically. That is, we can iterate over all valid
coalesce options (it's just an integer) and use that in a more generic
way.
If NMEthtoolCoalesceState names all fields explicitly, we need explicit
code that names each coalesce option. Especially since NMEthtoolCoalesceState
is an internal and intermediate data structure, this is cumbersome
and unnecessary.
Thereby it also fixes the issue that nm_platform_ethtool_init_coalesce() has a
NMPlatform argument without actually needing it.
nm_platform_ethtool_init_coalesce() does not operate on a NMPlatform
instance, and should not have the appearance of being a method of
NMPlatform.
When configuring with sanitizers enabled, ./configure.ac sets
-DVALGRIND=1 in the CFLAGS.
This causes a compilation error later:
$ /bin/sh ./libtool --tag=CC --mode=compile gcc ... -DVALGRIND=1 ... src/dhcp/nm-dhcp-nettools.c
...
In file included from src/dhcp/nm-dhcp-nettools.c:16:
./shared/systemd/sd-adapt-shared/nm-sd-adapt-shared.h:73: error: "VALGRIND" redefined [-Werror]
#define VALGRIND 0
When parsing user input if is often convenient to allow stripping whitespace.
Especially with escaped strings, the user could still escape the whitespace,
if the space should be taken literally.
Add support for that to nm_utils_buf_utf8safe_unescape().
Note that this is not the same as calling g_strstrip() before/after
unescape. That is, because nm_utils_buf_utf8safe_unescape() correctly
preserves escaped whitespace. If you call g_strstrip() before/after
the unescape, you don't know whether the whitespace is escaped.
We want to use the function to unescape (compress) secrets. As such, we want
to be sure that no secrets are leaked in memory due to growing the buffer with
realloc. In fact, reallocation should never happen. Assert for that.
As reallocation cannot happen, we could directly fill a buffer with
API like nm_utils_strbuf_*(). But NMStrBuf has low overhead even in this
case.
Often it is useful to not accept %NULL as input argument, to catch bugs.
For simple functions like nm_ethtool_id_get_by_name(), such limitations
are more annoying than helpful. Gracefully accept %NULL and treat is
like an invalid ethtool option.
_ASSERT_data() checks static, immutable data. Even with more asserts enabled,
there is no need to do that every time. Use NM_MORE_ASSERT_ONCE().
Note that NM_MORE_ASSERT_ONCE() will return constant FALSE, when build
without a sufficiently high assertion level. That means, the compiler
will just optimize the rest away.
We have nm_str_not_empty() which is the inverse of that. The purpose
of nm_str_not_empty() is to normalize a string to either return
%NULL or a non-empty string, like
const char *
get_name (Object *obj)
{
return nm_str_not_empty (obj->name);
}
Sometimes, we however want to check whether a string is not empty.
So, we previously had two choices:
1) use a temporary variable:
const char *tmp;
tmp = get_string ();
if (tmp && tmp[0])
...
The problem with this variant is that it's more verbose (by requiring a
temporary variable). Another downside is that there are multiple ways
how to check for an empty string (!tmp[0], tmp[0] == '\0', !strlen (tmp),
strlen (tmp) == 0), and sure enough they are all in use.
2) use !nm_str_not_empty(). But this double negation looks really odd
and confusing.
Add nm_str_is_empty() instead.
nm_g_error_matches() can be inlined and first checks whether the error
argument is not NULL. At least from the keyfile accessor functions, use
this macro, as they are called many times.
Macros preferably behave function-like, for example in that they evaluate
arguments exactly ones. Sometimes, we want to evaluate arguments
lazily, like in NM_IN_SET() or nm_g_set_error_take_lazy(). But it
is almost always undesirable to evaluate an argument more than once.
Fix NM_STR_HAS_PREFIX() for that.
Also, rename the local variable to not use the name "_str",
which may be a common name that the caller would like to use.
GPtrArray does not support NULL terminating the pointer array. That
makes it cumbersome to use it for tracking a strv array. Add a few
helper functions nm_strvarray_*() that help using a GArray instead.
We had three callers of nm_keyfile_plugin_kf_get_integer_list(). Two
only wanted to read values in range of guint8. One, wanted to read
unsigned integers (for which nm_keyfile_plugin_kf_get_integer_list()
was not suitable).
Instead, implement a integer list reader ourself.
One change is that g_key_file_get_integer_list() would accept list elements
with a number followed by a white space and garbage ([1]). We don't do that,
so there is a change in behavior here. That seems preferable, we don't
want to accept garbage.
The error reason text from the reader now also changes, and obviously we
no longer fail for integer values larger than G_MAXINT.
[1] c9bf247eb9/glib/gkeyfile.c (L4445)
Keyfile handles GObject properties of type G_TYPE_ARRAY as a GArray
of unsigned ints. That is correct, because all our properties of this
GType happen to be of this kind.
However, then the function was using nm_keyfile_plugin_kf_set_integer_list(),
which only can handle signed integers. There was thus an assertion that all
integers were non-negative. Which, probably was also correct, because NMSettingDcb
would validate that all values of such kind are in fact positive. Anyway, that
is an unexpected limitation (if not a bug).
Fix that by handling the array as unsigned list of integers.
Also, since glib doesn't provide an API for storing lists of unsigend
integers, we have to implement our own. but that is no loss. We probably
do it better anyway.
Previously, we were preallocating a string buffer of fixed size. For guint8
we reserved 3 characters per number, which is sufficient. However, it is
not obviously sufficient. NMStrBuf would grow as needed.
Next, I will add nm_keyfile_plugin_kf_set_integer_list_uint(), where it
is more unclear how large the string can be at most. To avoid that question
from the start, it will use NMStrBuf. To keep the implementations similar,
use NMStrBuf also in this case.
Split the macros to define the setter and getter so that setters
and getters are defined by separate macros. This will be used
to define the boolean getter differently, but still using the
macro to define the setter.
Also, don't construct function names in the macro. Instead, pass
the full names as argument to the macro. This helps with the problem
where ctags/cscope is unable to locate the implementation of the
function. Since we define the function with macro, the tools still
don't recognize this as the location of the definition. But at least
when showing all occurrences of the name, it can be found.
Currently any error encountered in n_dhcp4_c_connection_dispatch_io()
causes a dispatch failure and interrupts the library state
machine. The recvmsg() on the socket can fail for different reasons;
one of these is for example that the UDP request previously sent got a
ICMP port-unreachable response. This can be reproduced in the
following way:
ip netns add ns1
ip link add veth0 type veth peer name veth1
ip link set veth1 netns ns1
ip link set veth0 up
cat > dhcpd.conf <<EOF
server-identifier 172.25.0.1;
max-lease-time 120;
default-lease-time 120;
subnet 172.25.0.0 netmask 255.255.255.0 {
range 172.25.0.100 172.25.0.200;
}
EOF
ip -n ns1 link set veth1 up
ip -n ns1 address add dev veth1 172.25.0.1/24
ip netns exec ns1 iptables -A INPUT -p udp --dport 67 -j REJECT
ip netns exec ns1 dhcpd -4 -cf dhcpd.conf -pf /tmp/dhcp-server.pid
If a client is started on veth0, it is able to obtain a lease despite
the firewall rule blocking DHCP, because dhcpd uses a packet
socket. Then it fails during the renewal because the recvmsg() fails:
dhcp4 (veth0): send REQUEST of 172.25.0.178 to 172.25.0.1
dhcp4 (veth0): error -111 dispatching events
dhcp4 (veth0): state changed bound -> fail
The client should consider such errors non fatal and keep running.
https://bugzilla.redhat.com/show_bug.cgi?id=1829178https://gitlab.freedesktop.org/NetworkManager/NetworkManager/-/merge_requests/486
In general, I like macros. But in this case it seems the make the code harder
to understand than it needs to be. There are repeated patterns in these declarations,
but I feel they are better recognizible by aligning the lines nicely.
When handling a GCancellable, you make decisions based on when the cancelled
property of a GCancellable changes. Correctly handling a cancellable becoming
uncancelled again is really complicated, nor is it clear what it even means:
should the flipping be treated as cancellation or not? Probably if the
cancelled property gets reset, you already start aborting and there is
no way back. So, you would want that a cancellation is always handled.
But it's hard to implement that correctly, and it's odd to claim
something was cancelled, if g_cancellable_is_cancelled() doesn't agree
(anymore).
Avoid such problems by preventing users to call g_cancellable_reset().
Conceptionally, the MUD URL really depends on the device, and not so
much the connection profile. That is, when you have a specific IoT
device, then this device probably should use the same MUD URL for all
profiles (at least by default).
We already have a mechanism for that: global connection defaults. Use
that. This allows a vendor drop pre-install a file
"/usr/lib/NetworkManager/conf.d/10-mud-url.conf" with
[connection-10-mud-url]
connection.mud-url=https://example.com
Note that we introduce the special "connection.mud-url" value "none", to
indicate not to use a MUD URL (but also not to consult the global connection
default).
Add nm_utils_invoke_on_timeout() beside nm_utils_invoke_on_idle().
They are fundamentally similar, except one schedules an idle handler
and the other a timeout.
Also, use the current g_main_context_get_thread_default() as context
instead of the singleton instance. That is a change in behavior, but
the only caller of nm_utils_invoke_on_idle() is the daemon, which
doesn't use different main contexts. Anyway, to avoid anybody being
tripped up by this also change the order of arguments. It anyway
seems nicer to first pass the cancellable, and the callback and user
data as last arguments. It's more in line with glib's asynchronous
methods.
Also, in the unlikely case that the cancellable is already cancelled
from the start, always schedule an idle action to complete fast.
Why "if (length > G_MAXUINT)"? This is never going to hit. Also,
we probably should actual missing keys handle differently from
empty lists. If @error is set, return without setting the property.
g_key_file_get_integer_list() can return %NULL without setting an error.
That is the case if the key is set to an empty value.
For X sake, this API. Read the documentation and figure out whether
the function can return %NULL without reporting an error.
Anyway, avoid the assertion failure.
https://gitlab.freedesktop.org/NetworkManager/NetworkManager/-/issues/412
