Die sehr seltsame Portzuordnung des TP-Link TD-W9980 (Baugleich auch TD-W9980 und TD-W8980)
Außerdem Pfad ROUTERMAC
Signed-off-by: Foliant <foliant@noreply.git.freifunk-franken.de>
Some devices had to be disabled starting with OpenWrt 21.02 due to their
new DSA switch drivers.
As support for DSA configuration has since been added to fff-network,
this makes use of this configuration for devices with DSA drivers.
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Tested-by: Fabian Bläse <fabian@blaese.de>
Starting with OpenWrt 21.02 some devices now use upstream kernel drivers
for the built-in switch instead of relying on OpenWrt's swconfig driver.
The upstream kernel drivers use the Distributed Switch Architecture
(DSA) for configuration of the switch. Instead of explicitly configuring
the hardware switch, all ports appear as distinct interfaces and linux
bridges are offloaded to the hardware switch if possible.
To keep changes small, this patch adds support for DSA devices by
setting up a linux bridge, which is then treated just like the
cpuport-interface of the swconfig driver.
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Tested-by: Fabian Bläse <fabian@blaese.de>
To make it clear why the set sysctls are necessary, add appropriate
comments to them. Also reorder them for improved readability.
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Tested-by: Fabian Bläse <fabian@blaese.de>
Many of the set sysctls are either unnecessary, are already default in
the kernel or in OpenWrts defaults, or the reason for them being
explicitly set is unknown.
Remove all those sysctls from fff-network, as unfounded deviations from
default values will cause hard-to-debug problems in the future.
The original motivation for this patch is the netdev_max_backlog sysctl,
which was set to a very low value without any reason or comment.
This hurt forwarding performance on mt7621 with DSA significantly and
took quite a while to discover.
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Tested-by: Fabian Bläse <fabian@blaese.de>
This migrates the support for the TP-Link CPE210 v1 and CPE510 v1
so they are recognized by the new two-port setup code.
Assignment of ports should be consistent to the ar71xx implementation,
i.e. primary port (PoE in) assigned to WAN and secondary port (PoE out)
assigned to CLIENT by default.
Note that this is the exact opposite of the default configuration of
what OpenWrt does (but both have been consistent in behavior).
Since they work again, also re-enable image transfer.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
Support native two-port devices by adding TWO_PORT variable and
exploiting the LAN0PORT and LAN1PORT variables designed for the
TP-Link CPE devices. Since the latter have been converted to
real two-port devices, we can now repurpose these variable for
the new setup.
This exploits the existing WANDEV and SWITCHDEV variables to
define the primary and secondary interfaces.
Note that by default this takes the initial values from network.mode,
so if initial WAN/CLIENT should swapped, this has to be added to
the relevant network.* file of the devices.
No device-specific changes are done in this patch.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
So far, all one-port devices also triggered the switch-based setup
in configurenetwork, as the one-port condition was not checked
there. While the relevant parts are overwritten by the one-port
config which comes later in the script, it still creates a lot
of useless/broken switch/vlan setup entries in /etc/config/network.
Properly check for one-port vs. non-one-port in the file, without
touching anything else.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Acked-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
Dropbear supports ed25519 keys since OpenWrt 21.02.
Also, ecdsa is supported since v19.07, but disabled in our firmware.
Keep the generated ed25519 and ecdsa host key accross upgrades.
While at it, remove dss host keys, as they are not supported anymore.
5eb7864aadd5 ("dropbear: rewrite init script startup logic to handle both host key files")
8a7a93947004 ("dropbear: remove generation and configuration of DSS keys")
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Because wireguard is part of the Linux kernel starting with version 5.6,
the wireguard packages have been renamed upstream.
Update our dependencies to match this.
This fixes build for the layer3 variant.
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
[add fix comment]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
The Bullet M (XM) was sold with two different SoCs, AR7241 and
AR7240, which cannot be served by one DTS. This implements both
versions as done in OpenWrt.
Note that those variants may not be distinguished from the outside.
The AR7241 version appears to be the more recent and more abundant
version.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
In the latest release, OpenWrt provides the label MAC address for
many devices. All of our devices should be covered.
In can be retrieved by the function
get_mac_label
from /lib/functions/system.sh
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
Some devices were renamed since the last stable release.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
The SPDX license identifier provides a standardized way for specifying
licenses that is both human- and machine-readable. It is used upstream
both in OpenWrt and the Linux kernel.
Replace licenses in our repository by those identifiers.
The full-text licenses corresponding to these identifiers are
provided in the LICENSES folder.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Acked-by: Fabian Bläse <fabian@blaese.de>
This adds an option 'disabled' that will allow to disable
nodewatcher when desired.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
This package is the owner of the config file, so add it to the
Makefile. This will have vanilla OpenWrt copy it during upgrade.
Since we disable this mechanism, it will not change anything for
our standard firmware. But it will improve the situation if this
package is used in vanilla OpenWrt.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
If /etc/config/nodewatcher already exists, a proper uci-defaults
script should not overwrite it. Since this package is the owner
of the config file, this change won't change anything for the
current firmware, but will allow to use this as a package, too.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
The uci-defaults scripts are meant to provide defaults for a
specific package. Distributing them across several packages makes
no sense and just makes maintainance worse.
Thus, move the network part of the initialization back to the
proper package. While at it, suppress output from add commands.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
With commit [1] the ipv4 firewall on wan interface was removed.
This patch adds the ssh connection limit for ipv4.
IPv6 is already limited.
[1] 52e15e072c ("fff-firewall: Remove ssh firewall on WAN interface")
Signed-off-by: Robert Langhammer <rlanghammer@web.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
[improve commit reference]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Removes the firewall ebtables entry.
alfred-monitoring-proxy is only useful in layer3 variant, where no
ebtables rules are set.
With this typo the ebtables command was never active and the
resulting error was never shown.
Fixes: 9b5d3f1aeb ("fff-alfred-monitoring-proxy: add package")
Signed-off-by: Robert Langhammer <rlanghammer@web.de>
[fix typos in message, add Fixes:]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
In '/etc/network.ubnt,unifiac-mesh', we include a file that does
not exist on ath79 anymore. This causes the script execution to
abort and will have configurenetwork not run at all, making the
device quite inaccessible.
Remove the include as it never had any use anyway. Remove another
unused include as well, and add the proper dependency instead.
This fix was first proposed more than 2.5 years ago.
Fixes: #130 (gitea)
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
The "available" amount of memory is helpful for several forensic
and debugging cases. Send it via alfred.
Monitoring support has already been implemented.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
Source address filtering (RFC3704) can be used to mitigate source
address spoofing. However, strict mode only works when routes are
strictly symmetric. If routes are asymmetric, it can happen that
the best route to the source address of a packet is via a different
interface.
Because there is no guarantee that routes have to be symmetric in the
Freifunk Franken backbone network, we cannot use strict mode. Because
default routes are used in the Freifunk Franken backone, loose mode
could be used, but does not make any sense. Instead, revert back to the
kernel default setting, which currently is 0 (disabled).
While this change affects both layer3 and node variant, nothing changes
for the node firmware, because it does not forward packets.
Fixes: #123
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Christian Dresel <freifunk@dresel.systems>
Previously, an additional wan6 interface for SLAAC has been added, which
references the wan interface for its interface.
However, OpenWrt waits until the wan interface is completely up, until
it tries to start up interfaces that depend on it.
This not only can delay the configuration of IPv6 addresses
significantly, but also makes configuration of the wan6 interface
impossible in WAN networks with out a DHCP server.
To solve this issue, a separate interface wan4 for dhcp, which also
reference the wan interface, is created and the proto of wan is set to
none.
Fixes: #114
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
Migration from ar71xx to ath79 included changes to the ethernet/switch
driver for the target. This introduced a peculiar issue where
eth0 and eth1 are swapped for several devices. Most of the relevant
cases were already covered in 1cf4d762ff ("treewide: move devices
from ar71xx to ath79").
This switch also affects the Nanostation M, where the PoE-in port
is now eth1 und the PoE-out port is eth0. However, no action was
taken in the referenced patch, as nobody was aware of it then.
Since the Nanostation M is a two-port device, which we cannot
implement properly so far, it was implemented as a one-port with
the "primary" PoE-in port so far. This was broken by the ath79
introduction and is now fixed in this patch by using the one-port
setup on eth1. That way, the PoE-in port can now be switched by
ETHMODE as usual again.
Note that custom scripts, e.g. to set up the second port, need to
be adjusted manually, as that one is eth0 now.
Fixes: 1cf4d762ff ("treewide: move devices from ar71xx to ath79")
Fixes: #109 (gitea)
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
Tested-by: Fabian Bläse <fabian@blaese.de>
We use the one-port implementation also on two-ports like the
Nanostation M. Therefore, hard-coding eth0 in configurenetwork
will break if the port implemented for the one-port setup
(SWITCHDEV/WANDEV) is not eth0.
Just use SWITCHDEV instead, like done for the rest of the one-port
setup.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
Tested-by: Fabian Bläse <fabian@blaese.de>
When creating a new uci section, the safest way to refer to it is
with index '-1'. While it (probably) does not make a functional
difference for our specific case, since we expect to only have one
section anyway, let's just make sure and use the proper indexing.
While at it, suppress output from the 'add' command.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
So far, we ensure the selection of a specific variant by copying
included Make files. This not only breaks if the packages are used
as a feed, but also is against the concept of how packages are used.
In this patch, the fff package is converted into a set of variants
that allow selection via a FFF_VARIANT variable that is exported by
buildscript. If no export happens, e.g. when using packages in a
feed, no package is selected.
Since the names fff-node and fff-layer3 are not available anymore,
the packages for the variants are called (though irrelevant for
the user):
* fff-variant-layer3
* fff-variant-node
The only drawback is that we now have to specify the list of
available variants in the buildscript. However, these values are
hardcoded in several other places as well, and the former code
based on file names was not really appealing anyway.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Tested-by: Robert Langhammer <rlanghammer@web.de>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
Allow setting WAN vlan/interface from gateway config.
A configuration without explicit definition of a WAN interface is
valid at the moment and results in the default configuration from
fff-network being used.
Originally, it was intended to automatically set WAN to vlan 2, if nothing was
specified. As this would break devices, which don't use swconfig for
WAN, the already configured WAN interface is left untouched.
Fixes: #85 (gitea)
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Christian Dresel <freifunk@dresel.systems>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
Instead of exposing the CPUPORT variable to the calling script
directly, wrap it into a function which can be called there.
Fixes: #52 (gitea)
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Acked-by: Christian Dresel <freifunk@dresel.systems>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
So far, we define PORTORDER individually in each network.* file.
This creates a lot of duplications, and makes the code to parse those
values very ugly (and it's only used outside of configurenetwork
anyway).
Therefore, move the assignment to a library file, and wrap it into
a function for tidyness. This gives us more overview and nicer
implementation of the retrieval.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Acked-by: Christian Dresel <freifunk@dresel.systems>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
With 8d66bdf the port configuration of TP-Link Archer C7 has been
changed to a single-interface configuration.
This had unforseen side effects on upgraded devices. Because WANDEV
is evaluated from the updated network.* file, the port configuration
of the switch is evaluated from the update-safe network.config, which
is now incompatible with the updated interface configuration.
Therefore, a migration script has to be added, which updates the port
configuration in network.config to the new single-interface network
configuration.
Fixes: #60 (gitea)
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
In earlier firmware versions the prefix had to be written
in an incorrect syntax (missing a trailing colon).
To make hoodfiles with this old incorrect syntax work with
newer firmware versions like this one, we have to fix the
incorrect syntax. Both the old, incorrect and the correct
syntax work with this fix, so in the far away future, the
correct syntax can be used in hoodfiles.
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
This reverts commit e5da228cb1.
With the patch there can be situations with just "*" in traceroute,
breaking MTU in wireguard. If R1 with IPv6 address from provider P1
is connected to Freifunk via wireguard, and another R2 with address
from provider P2 is behind it, then R1 won't answer to traceroutes
sent from R2.
Revert the patch for now.
Fixes: #66 (gitea)
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Christian Dresel <freifunk@dresel.systems>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
PKG_BUILD_DIR has the following default values set in include/package.mk,
in case no BUILD_VARIANT is set:
With PKG_VERSION set: $(BUILD_DIR)/$(PKG_NAME)-$(PKG_VERSION)
Without PKG_VERSION: $(BUILD_DIR)/$(PKG_NAME)
Consequently, all PKG_BUILD_DIR definitions in our packages are
redundant. Remove them.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
We do not use the IPv4 functionality of odhcpd, but use dnsmasq
for that. Use odhcpd-ipv6only instead.
This is also the default for OpenWrt.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Johannes Kimmel <fff@bareminimum.eu>
Other than atoi, strtol allows to detect parsing errors.
Therefore atoi is replaced with strtol and appropriate error
checks are added.
Fixes: #33 (gitea)
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Johannes Kimmel <fff@bareminimum.eu>
The owipcalc tool provides an "add" algorithm which can be used
to concateneted IPv6 addresses from prefix and suffix.
Since it's available upstream and our string manipulation is ugly,
let's replace our IP concatenation with that tool. The package
consists of a single .c file with about 1000 lines resulting in
about 4 kB for the ipk package.
This patch does _not_ introduce any conceptual changes yet. Thus,
the "wrong" IPv6 prefix from KeyXchange will be expected in the
same format, it is just healed for the new code for now.
The change allows to get rid of some bloat, i.e. some quite trivial
custom functions on the way. This also drops the ipTidyColon()
function, as owipcalc seems to return the collapsed version by default.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Reviewed-by: Fabian Bläse <fabian@blaese.de>
The variables SERVICE_WRITE_PID and SERVICE_DAEMONIZE are not used by
procd, so they are removed.
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
Reviewed-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
The configuration of our firewall depends on the network configuration.
Most importantly, the firewall has to be restarted, if the WAN-interface
is changed.
Therefore, a procd reload trigger is added to the init-script, so our
firewall is automatically restarted, when the network configuration is
changed.
Fixes: #46 (gitea)
Signed-off-by: Robert Langhammer <rlanghammer@web.de>
[fabian@blaese.de: Remove unrelated changes]
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
Reviewed-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Some sysctls currently are completely useless, as they only exist if
specific kernel configurations are enabled, which we have not.
To hide the error message and prevent them from interfering
unintentionally, if new kernel configurations are activated in the
future, they are removed.
Fixes: #42 (gitea)
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Christian Dresel <freifunk@dresel.systems>
The net.ipv4.netfilter.ip* sysctls have been moved to
net.netfilter.nf* a long time ago, so they have been useless in our
firmware for quite a while.
It probably originally has been added because it was included in the
OpenWrt defaults and in earlier versions of our firmware the OpenWrt
defaults file got overwritten by our own one.
Because there does not seem to be any obvious reason to keep them (they
have been added without a comment in the commit or file) and they have
been inactive ever since they were moved in the kernel, they are removed
completely instead of using the correct path.
Fixes: #42 (gitea)
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Christian Dresel <freifunk@dresel.systems>
The cake scheduler is a popular fair queuing scheduler, which is also
capable of shaping traffic. Due to its sensible defaults it is very
easy to set up.
When tunnel traffic exceeds the capability of the transport connection,
firmware users might want to shape traffic, so meaningful queueing can be done
before tunnel packets are dropped. As this is typically combined with a fair
scheduler, cake provides a simple yet very powerful solution for both problems.
Therefore the cake kernel module is now included in the layer3 variant.
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Christian Dresel <freifunk@dresel.systems>
When neither peer_ip nor ipaddr are set, no ipv4 address for peering
interfaces is available. Therefore, no IPv4 routes can be advertised.
Other than that, a configuration like this is perfectly valid and
configuration is already continued. Therefore, the "FATAL" message might
be misleading so it is replaced with "WARNING" and clarified slightly.
Signed-off-by: Fabian Bläse <fabian@blaese.de>
Reviewed-by: Johannes Kimmel <fff@bareminimum.eu>
[bump PKG_RELEASE]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
On call -t we write the pid on /tmp/configure-layer3-pid.
If the script exits from user we use trap to run the new function keep_changes()
If the connections to the router lost, the user can run configure-layer3 -k after
reconnect to keep changes manually
Signed-off-by: Christian Dresel <freifunk@dresel.systems>
Reviewed-by: Robert Langhammer <rlanghammer@web.de>
[wrap and rephrase exit comment, bump PKG_RELEASE]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Foliant
Fabian Bläse
Adrian Schmutzler
Robert Langhammer
Christian Dresel