The original ldpd(8) daemon in OpenBSD doesn't allow the user to put
non-existing interfaces in the configuration file. For this reason,
the l2vpn_if_find() and l2vpn_pw_find() functions take an ifindex as
an argument. In FRR's ldpd we can put non-existing interfaces in the
configuration, and they are activated as soon as they are available. For
this reason, we can't lookup interfaces by their ifindex in this port.
The l2vpn_if_find_name() and l2vpn_pw_find_name() functions were
introduced to address this issue. However, since the "find-by-ifindex"
functions are not being used anymore, we can just remove them and rename
the *_find_name() functions removing the "_name" suffix.
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>
The rationale here is to do something very similar to what the original
OpenBSD's ldpd(8) does to reload configuration files:
* create a new empty base configuration (vty_conf);
* parse the configuration file and store the result in vty_conf;
* merge the new configuration into the current configuration.
To do this, introduce a new global variable called "sighup" so that ldpd
doesn't attempt to merge vty_conf into ldpd_conf for each command in
the configuration file, which would be very disruptive. vty_conf should
be merged only after the whole configuration file is parsed in order to
avoid taking actions like shutting down neighbors when not necessary.
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>
The logic to make configuration changes in ldpd is as follows:
1 - duplicate the current configuration;
2 - modify the duplicated configuration (delta);
3 - merge the duplicated configuration into the current configuration.
We can however change things a little bit in order to simplify the code.
If we provide a global vty_conf variable and make sure that it's
always an up-to-date duplicate of ldpd_conf, then we don't need to call
ldp_dup_config() in the callback of each configuration command. Instead
of that, we just need to duplicate ldpd_conf in the startup of the daemon
and in the end of the ldp_reload() function. Doing this is safe because,
due to the single-threaded nature of the parent process, it's impossible
to have multiple requests to change the configuration at the same time.
Also, we can remove the *_ref() functions and do something much simpler
instead. The problem we have to deal with here is that VTY_GET_CONTEXT()
returns a pointer to an element of ldpd_conf, and what we want is a
pointer to an element of vty_conf (e.g. an l2vpn). To solve this,
we just need to look for the equivalent element inside vty_conf using
standard functions from ldpd like l2vpn_find() and others.
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>
Fix two small memleaks in the CLI code and check the return values of
getsockname() and getpeername().
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>
This patch introduces several new configuration commands to ldpd. These
commands should allow the operator to define advanced filtering policies
for things like label advertisement, label allocation, etc.
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>
Using red-black trees instead of linked lists brings the following
benefits:
1 - Elements are naturally ordered (no need to reorder anything before
outputting data to the user);
2 - Faster lookups/deletes: O(log n) time complexity against O(n).
The insert operation with red-black trees is more expensive though,
but that's not a big issue since lookups are much more frequent.
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>
Using red-black trees instead of linked lists brings the following
benefits:
1 - Elements are naturally ordered (no need to reorder anything before
outputting data to the user);
2 - Faster lookups/deletes: O(log n) time complexity against O(n).
The insert operation with red-black trees is more expensive though,
but that's not a big issue since lookups are much more frequent.
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>
Using red-black trees instead of linked lists brings the following
benefits:
1 - Elements are naturally ordered (no need to reorder anything before
outputting data to the user);
2 - Faster lookups/deletes: O(log n) time complexity against O(n).
The insert operation with red-black trees is more expensive though,
but that's not a big issue since lookups are much more frequent.
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>
Using red-black trees instead of linked lists brings the following
benefits:
1 - Elements are naturally ordered (no need to reorder anything before
outputting data to the user);
2 - Faster lookups/deletes: O(log n) time complexity against O(n).
The insert operation with red-black trees is more expensive though,
but that's not a big issue since lookups are much more frequent.
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>
Using red-black trees instead of linked lists brings the following
benefits:
1 - Elements are naturally ordered (no need to reorder anything before
outputting data to the user);
2 - Faster lookups/deletes: O(log n) time complexity against O(n).
The insert operation with red-black trees is more expensive though,
but that's not a big issue since lookups are much more frequent.
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>
Using red-black trees instead of linked lists brings the following
benefits:
1 - Elements are naturally ordered (no need to reorder anything before
outputting data to the user);
2 - Faster lookups/deletes: O(log n) time complexity against O(n).
The insert operation with red-black trees is more expensive though,
but that's not a big issue since lookups are much more frequent.
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>
This moves all install_element calls into the file where the DEFUNs are
located. This fixes several small related bugs:
- ospf6d wasn't installing a "no interface FOO" command
- zebra had a useless copy of "interface FOO"
- pimd's copy of "interface FOO" was not setting qobj_index, which means
"description LINE" commands would fail with an error
The next commit will do the actual act of making "foo_cmd" static.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
ldpd was keeping track of the vty session's position in config editing
with 3 global static variables. This worked because only one vty could
be in configuration-editing mode before.
Replace with vty->qobj_index infrastructure and enable
vty_config_lockless.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
These functions are currently unused but will be used by the Cap'n Proto
interface. They're not a particular burden to maintain in-tree, so here
they go.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
This is necessary to prevent the same old configuration to come back
when the interface is reactivated later for a given address-family.
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>
