Description: Lack of support of HTTP/2 by default starts to hurt,
for example with https-dns-proxy package, some DoH resolvers (like mullvad)
no longer support HTTP/1 and are not usable.
This enables HTTP/2 support by default (which would bring ~68Kb libnghttp).
Signed-off-by: Stan Grishin <stangri@melmac.net>
Open vSwitch supports SSL to connect to an OpenFlow controller. This is
recommended for security. Expand the UCI ovs config section to allow
configuring SSL CA, certificate and private key.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
The Open vSwitch init script does not set USE_PROCD=1. Instead, it
defines most of the functions and variables that would be set when
USE_PROCD is set to 1, but with some minor changes.
The basescript variable however, which is used when calling
procd_open_service and procd_kill, is not set. As a result, basename of
the contents of the initscript variable is used as the service name. As
the service is automatically started via its symlink in /etc/rc.d,
S15openvswitch, the service name is S15openvswitch.
Set the basescript variable so that the service name is openvswitch.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
By default, Open vSwitch will generate the OpenFlow datapath ID of a
bridge based on the MAC address of one of its ports. Due to this, it's
possible that the datapath ID changes when new ports are added. When the
datapath ID changes, Open vSwitch disconnects from the controller, as
there is no way to notify the controller that the datapath ID has
changed.
Add an option to set the datapath ID so that the above situation can be
avoided. The option takes either exactly 16 hex characters, or when
prefixed with 0x, between 1 and 16 hex characters.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
The config symbol is named CONFIG_OPENVSWITCH_WITH_LIBUNBOUND, so check
for that instead of the non-existent CONFIG_OPENVSWITCH_WITH_UNBOUND.
Fixes: 45c8cc9d8a ("openvswitch: make libunbound optional")
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
support for 21.02.0-rc2 and up
support for reloading a single interface on ifup/ifupdate
rename config file
updated shellcheck compatibility
remove obsolete create/remove_lock
interface processing optimizations to speed up reloads
drop dependency on curl in user scripts
uniform styling of functions
Signed-off-by: Stan Grishin <stangri@melmac.net>
Installing openvswitch on an x86/64 snapshot image pulls in a bunch of
dependencies, good for a total size of 3648406 byte. Disabling
libunbound reduces that with 559941 byte, for a total of 3088465 byte.
This is quite a big reduction for a small tradeoff: without libunbound,
hostnames can not be used to specify OpenFlow managers or controllers.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Also bump Cython version to 0.29.23.
And add support for OpenBLAS.
Currently optional, but will be enabled by default on some architectures
later.
Depends on PR https://github.com/openwrt/packages/pull/15685
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Initial draft PR is:
https://github.com/openwrt/packages/pull/11894
This one is a bit more complete, and follows packaging practices.
For now, disabling builds on ARC and PowerPC. Will require more work to get
them going.
Explicitly disabling OpenMP support, so that it doesn't get picked by
accident.
Later we may use the `CPU_TYPE` parameter to tweak things a little further.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
* code cleanup
* add auto login script for Julianahoeve beach resort (NL)
* add auto login script for Vodafone hotspots (DE)
* add auto login script for telekom hotspots (DE)
* enhance captive portal detection to support html redirects as well
* change default captive portal detection url to
'detectportal.firefox.com'
Signed-off-by: Dirk Brenken <dev@brenken.org>
* genrate UCI fstab configs for each volume before first 'up'
* remove UCI section on volume remove
* use autofs automounter for read-only volumes
* try umount on 'down'
* emulate hotplug events for UBI volume up/down
* more robust error paths
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
We previously did not identify the correct BSS from the output of
`hostapd_cli -i <phy> status`, because when asked for a vif it will
always respond with information relevant to the whole phy.
The per vif settings will use an iterator and we now try to detect
the correct BSS from that output.
Signed-off-by: Martin Weinelt <hexa@darmstadt.ccc.de>