The `features` file is converted to a Lua-based DSL.
A helper function `_` is used in the DSL; this will return the original
string for enabled features, and nil for disabled features. This allows
to use boolean operations on features without making the code too
verbose.
Besides having more readable and robust code, this also fixes the bug
that all files `packages/*/features` were evaluated instead of only
using the feature definitions of currently active feeds.
This commit changes the behavior for the wan-dnsmasq to read the
interfaces to update the DNS servers on from flag-files.
This way, external custom packages which add a custom VPN WAN interface
can update the DNS server gluon is using for WAN resolving.
On lantiq, the default WAN ifname is 'dsl0' even on devices that have a
separate WAN port. At least on devices using swconfig we can override
this with the switch port labelled as 'wan'.
When the GTK is offloaded, MT7610 won't transmit any multicast frames.
This is most likely due to a bug in the offloading datapath. MT7612 is
not affected.
Disable GTK offloading for now. It can be re-enabled once the bug in the
offloading path is fixed.
Signed-off-by: David Bauer <mail@david-bauer.net>
This gives us WPA3 support out of the box without having to manually disable
hardware crypto. The driver will fall back to software crypto if the connection
requires management frame protection.
THis allows us to use WPA3 features (Private-WiFi SAE & OWE) on
ramips-mt7620.
When the GTK is offloaded, MT7610 won't transmit any multicast frames.
This is most likely due to a bug in the offloading datapath. MT7612 is
not affected.
Disable GTK offloading for now. It can be re-enabled once the bug in the
offloading path is fixed.
Signed-off-by: David Bauer <mail@david-bauer.net>
With very bad timing, it is possible that the teardown script of a
gluon_mesh interface runs when bat0 was just created, but primary0 is not
yet added to it. Although there is no hardif to remove in this case,
bat0 will still be deleted, because there is no hardif in bat0.
Disable the interface removal logic by passing `-M` to `batctl interface`.
(cherry picked from commit 92647cd47a)
With very bad timing, it is possible that the teardown script of a
gluon_mesh interface runs when bat0 was just created, but primary0 is not
yet added to it. Although there is no hardif to remove in this case,
bat0 will still be deleted, because there is no hardif in bat0.
Disable the interface removal logic by passing `-M` to `batctl interface`.
There is no setting to explicitly enable mesh interfaces in the config
wizard, so we need to enable mesh interfaces by default when the outdoor
mode is disabled (unless site.conf dictates otherwise).
The simplest way to achieve this is to delete the 5GHz mesh interface
sections when outdoor mode is disabled: 200-wireless would delete and
recreate them a moment later anyways, but by deleting them earlier we
ensure that the disabled status is initialized from site.conf again.
(cherry picked from commit 17bc622892)
When disabling outdoor mode, 5GHz mesh interfaces should be enabled by
default again (unless disabled in site.conf).
(cherry picked from commit d2052431dd)
Because is_disabled() was always returning true or false, the
first_non_nil() would never actually check the default setting from
site.conf. This was broken since v2017.1.
Fixes: 6cf03bab37 ("treewide: replace normal uses of luci.model.uci with simple-uci to reduce LuCI dependencies")
(cherry picked from commit 3c2593b684)
Normally, we build all nonshared packages (which includes all kernel
modules) to generate an opkg feed for later package installations by
users. On targets without opkg, this just wastes time - disable it.
(cherry picked from commit b3edfd292a)
There is no setting to explicitly enable mesh interfaces in the config
wizard, so we need to enable mesh interfaces by default when the outdoor
mode is disabled (unless site.conf dictates otherwise).
The simplest way to achieve this is to delete the 5GHz mesh interface
sections when outdoor mode is disabled: 200-wireless would delete and
recreate them a moment later anyways, but by deleting them earlier we
ensure that the disabled status is initialized from site.conf again.
Because is_disabled() was always returning true or false, the
first_non_nil() would never actually check the default setting from
site.conf. This was broken since v2017.1.
Fixes: 6cf03bab37 ("treewide: replace normal uses of luci.model.uci with simple-uci to reduce LuCI dependencies")
Normally, we build all nonshared packages (which includes all kernel
modules) to generate an opkg feed for later package installations by
users. On targets without opkg, this just wastes time - disable it.
This gives us WPA3 support out of the box without having to manually disable
hardware crypto. The driver will fall back to software crypto if the connection
requires management frame protection.
THis allows us to use WPA3 features (Private-WiFi SAE & OWE) on
ramips-mt7620.
The interfaces should always exist and just be disabled when outdoor
mode is enabled. Not creating them at all leads to an issue in the
advanced wifi settings where an additional reload of the page is
necessary after enabling or disabling the ourdoor mode to make the mesh
VIF options appear or disappear.
(cherry picked from commit b15b008e69)
Instead of relying on the existence of interface sections only (which is
still used for the case of interface types that are not configured at
all in the site config), add a proper dependency.
This fixes section visiblity with enabled outdoor mode after the following
fix, and gives the user immediate feedback (allowing to disable outdoor
mode and enable meshing in a single step).
(cherry picked from commit d012d7d32f)
200-wireless will add or remove the mesh network sections of
/etc/config/network. Commit this file, so the modified setting doesn't
get lost on reboot.
Fixes: #2048
(cherry picked from commit 600ab99f80)
Always allow options set to builtin (=y) when modular setting (=m) is
expected. This can happen when a package is added explicitly (in a
target defintion or site.mk) that is also pulled in as a dependency of
another builtin package.
Fixes: 9e23534ec3 ("build: rework config generation")
Fixes: #2046
(cherry picked from commit 8b64517f1b)
With 11a3b56617 ("gluon-web-model: add dependency support for
sections"), a "reset" function was added to all model nodes. This
conflicts with the "reset" field of the Form object, making reset
buttons show labels like "function: 0xbb8d50" and possibly breaking form
processing when packages override the reset field.
The interfaces should always exist and just be disabled when outdoor
mode is enabled. Not creating them at all leads to an issue in the
advanced wifi settings where an additional reload of the page is
necessary after enabling or disabling the ourdoor mode to make the mesh
VIF options appear or disappear.
Instead of relying on the existence of interface sections only (which is
still used for the case of interface types that are not configured at
all in the site config), add a proper dependency.
This fixes section visiblity with enabled outdoor mode after the following
fix, and gives the user immediate feedback (allowing to disable outdoor
mode and enable meshing in a single step).
200-wireless will add or remove the mesh network sections of
/etc/config/network. Commit this file, so the modified setting doesn't
get lost on reboot.
Fixes: #2048
Always allow options set to builtin (=y) when modular setting (=m) is
expected. This can happen when a package is added explicitly (in a
target defintion or site.mk) that is also pulled in as a dependency of
another builtin package.
Fixes: 9e23534ec3 ("build: rework config generation")
Fixes: #2046
Validation errors are mostly handled in the frontend (by displaying
invalid fields with a red background). There was some code left in the
backend for handling different kinds of errors, and returning custom
error messages from validate(), but the resulting value was not used
anywhere.
- Add a dependency array to section template
- Lift dependency handling from AbstractValue to Node
- Split resolve_depends() into a function handling dependencies of a
node itself (resolve_node_depends()), which now is used for all nodes,
and the existing recursive descent through the node tree
- When any node's dependencies are not satisfied, all descendents are
reset recursively, meaning that an option in a section that is hidden
will always be ignored, as if the option's own dependencies were
unsatisfied
Fixes: #1349
