| Age | Commit message (Collapse) | Author |
|---|
|
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
|
|
Fix macvlan delete via netlink as netlink attribute IFLA_IFNAME data size was incorrect (size reject by the kernel) and NLM_F_REQUEST flag was missing.
In addition some minor improvements (attribute IFLA_INFO_KIND can be left out as RTM_DELLINK does not require the attribute)
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
|
|
Fix vlan delete via netlink as netlink attribute IFLA_IFNAME data size was incorrect.
In addition some minor improvements.
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
|
|
At moment netifd supports just 802.1q vlan, you can configure them using a concise but "hacky"
syntax using an interface config section, with this patch netifd acquire the capability
of configuring 802.1ad and 802.1q vlan using config device sections, so you can define a vlan device
plus interface with something like this:
config device 'test'
option type '8021ad'
option name 'test'
option ifname 'eth0'
option vid '1000'
config interface 'testif'
option ifname 'test'
option proto 'none'
option auto '1'
old syntax for 802.1q keeps working so no retrocompatibility problems,
to keep retrocompatibility means also that user must not use name/ifname like eth0.2
for devices declared with the new style because this would trigger the "old style"
when interface config section is parsed
Signed-off-by: Gioacchino Mazzurco <gmazzurco89@gmail.com>
|
|
Signed-off-by: Steven Barth <steven@midlink.org>
|
|
Signed-off-by: Steven Barth <steven@midlink.org>
|
|
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
|
|
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
|
|
per device
Main use case is being able to disable IPv6 on (a) WAN interface(s) when only IPv4 connectivity is offered or 6rd is used.
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
|
|
Reapply bridge mtu setting as adding a bridge member will override the bridge mtu in the kernel
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
|
|
Device layer is informed by netlink events regarding the link layer status. Link
layer status change results in a DEV_EVENT_LINK_UP/DEV_EVENT_LINK_DOWN broadcast
event for a given device.
Depends on uloop error callback patch.
Solves issue reported in https://dev.openwrt.org/ticket/14590
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Acked-by: Karl Vogel <karl.vogel@gmail.com>
|
|
|
|
Disable netlink auto ack when doing a delete in the get callback
handler to avoid race conditions resulting into stalled message
on the netlink socket.
Solves issue reported in https://dev.openwrt.org/ticket/14590
Signed-off-by: Karl Vogel <karl.vogel@gmail.com>
Acked-by: Hans Dedecker <dedeckeh@gmail.com>
|
|
This reverts commit c439b52400978dd3799c66e1f632ee68d2c7c9eb.
|
|
Signed-off-by: Steven Barth <steven@midlink.org>
|
|
Patch implements link layer state awareness (aka carrier detection) in
netifd on device level.
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
|
|
Make tunnel don't fragment bit configurable via UCI
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
|
|
Tunnel MTU is applied according to the tunnel MTU UCI parameter
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
|
|
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
|
|
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>
|
|
* Set a default policy-failed rule per prefix (based on a patch by Jonas Gorski)
* Use input interface in addition to source for filtering
* Avoid duplicate routing policies
|
|
|
|
|
|
generic multiwan support
|
|
Signed-off-by: Steven Barth <steven@midlink.org>
|
|
Signed-off-by: Steven Barth <steven@midlink.org>
|
|
|
|
|
|
[netifd] add bridge priority option
Using the bridge priority (lower numbers are higher in the
hierarchy), one can ensure that the router is chosen as root
bridge in a setup with spanning tree protocol.
For instance, one can set the priority of network lan to 32767,
causing the router to win over all directly and indirectly
connected nodes that have a default priority of 32768.
The reason for doing that is that otherwise it has a default
priority 32768 as well and any other connected node could win and
get root bridge. In a home setup, those nodes are often desktop
or laptop boxes and get switched off from time to time. As a
consequence, root bridges vanish or new root bridges get chosen
relatively often, resulting in frequent topology changes to the
STP network. While the new topology has not settled, packets can
get lost, causing noticeable interruptions of network traffic.
Setting the router's bridge on a lower numbered priority (and thus
higher in the selection hierarchy) solves the problem in the vast
majority of the cases by ensuring that the device that is most
likely powered on 24/7 gets chosen as root and prevents topology
changes.
Signed-off-by: Niels Boehm <blubberdiblub@gmail.com>
|
|
|
|
|
|
|
|
|
|
Signed-off-by: Steven Barth <steven@midlink.org>
|
|
pass values as u64 blob type
|
|
advertised routes
|
|
RTPROT_BOOT so that routing daemons like quagga will pick it up
|
|
devices
|
|
The system_set_dev_sysctl() function erroneously used the value instead of the
device name to format the sysctl path, this change corrects that.
|
|
|
|
|
|
|
|
do not have 6rd support
|
|
This patch adds support for setting the 6rd tunnel options. These are the same
options normally specified with `ip tunnel 6rd ...`.
Signed-off-by: Stéphan Kochen <stephan@kochen.nl>
|
|
|
|
|
|
|
|
|
|
|
|
|
