When the receive buffer is too small, we easily can hit ENOBUFS during recvmsg()
and need to resync the platform cache. But even worse, we possibly also loose
ACKs for pending netlink requests so that requests seem to fail (although they
might have succeeded).
Avoid that harder by increasing the buffer size to 8MB. This is also
done by networkd:
be660c371b
When reading from netlink-socket fails with NLE_NOMEM, messages were
lost. In this case the cache must be resynced and all pending messages
from the socket are rejected via "event_handler_recvmsgs (platform, FALSE)".
In this case, we don't want to handle the received messages as the
cache anyway needs to resync. However, we are still interested in
all queued ACKs that are there.
We are also interested in RTM_NEWADDR messages which we use to detect
kernel support via _support_kernel_extended_ifa_flags_detect().
Coverity: CID 59378 (#1 of 1): Nesting level does not match indentation
(NESTING_INDENT_MISMATCH) This statement is indented to column 41, as if
it were nested within the preceding parent statement, but it is not.
Adding a route via iproute2 with explicit gateway 0.0.0.0
sets the route scope to GLOBAL, contrary to a on-link device-route
which has scope LINK.
Add a test adding two such routes.
We inconsistently use gulong,guint,int types to store signal handler
id, but the type returned by g_signal_connect() is a gulong.
This has no practical consequences because a int/guint is enough to
store the value, however it is better to use a consistent type, also
because nm_clear_g_signal_handler() accepts a pointer to the signal id
and thus it must be always called with the same pointer type.
Up to now, the "include" directory contained (only) header files that were
used project-wide by libs, core, clients, et al.
Since the directory now also contains a non-header file, the "include"
name is misleading. Instead of adding yet another directory that is
project-wide, with non-header-only content, rename the "include"
directory to "shared".
Downsides:
- Add some additional overhead to manage the index
- The NMPCacheId struct grows to 16 bytes, affecting
hashing performance for all object types.
Still do it, based on the assumption that it doesn't matter
for a low number of interfaces. But the O(1) access time matters
when having lots of interfaces.
When deleting an object, we allow failure to delete a non-existing object.
Thus, the only thing we care about is whether the object is no longer
present after deletion. Adjust the return values to reflect that.
The idea was allowing pending delayed-actions and process them in an
idle handler. We dont want to do that, because whenever platform code
returns, we want to have no pending actions -- because otherwise the
platform cache might be in an inconsistent state.
Just drop it.
We would not expect that delayed_action_handle_all() is called recursively.
Assert against that. If we ever happen to call it recursively, we would
need to take care of properly avoiding infinite loops or deep call
stacks.
Track pending netlink requests and properly report the resulting errno
back.
Currently we send only requests in do_request_link() and do_request_all().
These callers don't actually care about the result, all they care that the
request is answered before returning back from platform code to the caller.
Thus, up to now the tracking of the sequence number was pretty simple.
Later we also want to get the errno from a request, thus rework sending
requests to also remember about outstanding sequence numbers and
properly track muliple parallel requests.
Later the synchronous actions (e.g. add-link) will also be handled
via the asynchronous socket.
Calling nl_recvmsgs_default() leads dirctly to recvmsgs() from
"nl.c". This functions reads messages (recvmsg) in a loop and
invokes the callbacks.
Later we want to merge nlh and nlh_event, meaning that we must
anticipate parsing unrelated messages while waiting for an ACK.
While that would be possible by registering different callbacks
and letting them interact, it is actually more complicated. Just
assume full control over the message parsing.
Basically, copy recvmsgs() to event_handler_recvmsg(). For now just
copy the function and do little adjustment (to show the similarity
to the original). Cleanup follows.
