- Reword texts in import dialogs for better clarity, use different
descriptions for full import and peer import
- Allow importing configurations without [Peer] section
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 2ae74b909b4acf35b72142d913f9e61090febf6d)
The /etc/config/ddns in particular might not be present on the system,
don't fail if it is absent.
Fixes: #5838
Fixes: 9ba20645b0 ("luci-proto-wireguard: rewrite protocol handler")
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 340183786e5441197141efaf7d8cf0e682047bfe)
This commit rewrites large chunks of the WireGuard protocol handler in order
to simplify the process of importing and exporting configuration. The major
changes are:
1) The wireguard interface configuration tab (General Settings) gained an
import assistant which allows dragging or pasting a native WireGuard
configuration file in order to import required settrings into uci
2) The peer configuration tab gained a similar import assistant which allows
importing the settings for a WireGuard peer from an existing native
WireGuard configuration file
3) The QR code export feature has been rewritten to make the resulting codes
actually useful for importing into a WireGuard client application.
Additionally the plaintext native WireGuard configuration is displayed
to allow copy-pasting it for use on a Linux or OS X system
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 9ba20645b01aacb391ab8c69f57b3da60db8f630)
Before opening (rendering) a nested modal map, make sure to save the parent
modal map in order to persist any structural uci changes, such as newly added
anonymous sections to prevent the nested map from operating on stale values
or ephemeral config section IDs.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 3be479446bdeaf446468c5e7ac3b9bc49da023da)
The existing logic for cloning section options into the modal form section
container did not properly handle SectionValue objects. Introduce a new
`cloneOptions()` helper to recursively traverse and properly clowning
such nested sections.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit f24606b1ffd59cc82c7251243a18f4cfdc4f78d5)
The nested section modal might have changed values also covered/displayed by
the parent map, so ensure to reload all parent maps in the current modal
view stack to ensure that no stale values are shown on return.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 0c81e25faa9ff9cb6edec477d56145d70be87cf3)
Only stack a new modal CBI map on top of the currently displayed modal one
if the related map or section differs. This prevents misbehavior when
switching the protocol of interfaces.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit f2fa59e89af3d4bdae9b12471184425b307ad5d6)
The variable was party called `res` or `response`, always call it
`response` from now on. This also solves a typo where
`data.request_hash` is never set to a correct value.
Signed-off-by: Paul Spooren <mail@aparcar.org>
(cherry picked from commit af23d0dfc3176511ada5fb3d081f1b0a287a11bc)
It is possible to request a specific filesystem so no other filesystems
are used. This speeds up the build process and may prevent failures in
edge cases.
A recent edge case is installing more packages than ext4 can handle
while squashfs works fine due to compression.
Since `rootfs_type` is now always available via `procd`, remove the
hack of reading `/proc/mounts`.
Signed-off-by: Paul Spooren <mail@aparcar.org>
(cherry picked from commit 8936e3c386e77c2dc090dbbd5a87907196f9149b)
Now it's possible to re-install the currently running version in
advanced mode. This can be useful when installing packages via `opkg`
and then requesting the firmware with the packages stored in squashfs.
FIXES: https://github.com/openwrt/luci/issues/5809
Signed-off-by: Paul Spooren <mail@aparcar.org>
(cherry picked from commit fd72e1c23ef1569fd8f8ab71bdcb4a94f802435a)
This allows the backend to track popular app versions and drop support for
unsed version in case of API changes.
Specifically the app sends a version string in the format "luci/GIT_HASH" to
the backend.
Signed-off-by: Paul Spooren <mail@aparcar.org>
(cherry picked from commit f368faf6789a43c2e1a26b721f8115473672eb29)
The existing implementation was hardcoding the empty TableSection placeholder
instead of invoking `renderSectionPlaceholder()` which might be overridden by
various forms.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 818776451d96e8f1a187eafca4a5ebbf0770fbde)
Theme CSS styles shoudl take care of introducing padding as needed.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit e4e001b59657e4cd0f333a3f43ce66b8f789deb2)
If a whitespace-only description is set on an element, the CSS :empty
selector will not match, causing description icons to be shown when
there's no actual content.
To avoid that, trim the description string when building the element.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit b6ae1d4f4aec94b38e59b11e5fd9c28799dae05b)
- Annotate gridsection textvalue cells as CBI.DummyValue widgets
- Replace `<br>` elements with newlines in `stripTags()`
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 9c136c9beaa7639aecc351c5725de33cfe4d2e5c)
Package luci-proto-ncm depends on comgt-ncm which uses an option
called 'mode' to set the radiomode of the modem. There is no option
'service' in the comgt-ncm scripts.
Suggested-by: breenstorm <49235337+breenstorm@users.noreply.github.com>
[fix commit subject, add commit message, rebase onto master branch]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 4d6642b636a9feaa173c568c99f4129c440ed8b8)
Since the `handleModalSave()` handler of the GridSection class invokes
`handleModalCancel()` after saving the data but before removing the
`addedSection` property, the `handleModalCancel` handler incorrectly
removed the uci section that has just been created.
This bug didn't affect anonymous GridSections because after saving the
id of the created section changes, causing the remove command to fail,
but for named ones with stable section IDs, the bug manifested.
Solve the issue by passing a flag to `handleModalCancel()` indicating
whether the method was called from a safe operation and use it do
decide whether to delete the new staged uci section or not.
Fixes: #5760
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 2771360108cb3e2011c7c9cb4921a9047ea70910)
Under certain circumstances, a staged uci option value might get unset
again before saving the config, leaving an empty section change set
behind, causing the save call to send an empty uci set request via rpc,
triggering an ubus code 4 (Resource not found) error.
In particular this prevented bridge VLANs from getting saved properly.
Fixes: #5876
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 25dcd0b77c4450b83cd8b8307397e58c34ebfcd1)
When kernel bonding module is loaded it will create a special
file /sys/class/net/bonding_masters. This is no network device.
Filter it out for getNetworkDevices() call.
Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
(cherry picked from commit 77d29c9dc0226faa9374ca25fba8dc642c99f303)
For network devices declared in uci but not yet created by netifd, the
runtime status information will be unavailable, causing methods such as
`getType()` to assume plain ethernet interfaces and `getParent()` to fail
resolving parent devices.
Fall back to infer the information from uci configuration settings in such
cases to give accurate type hints to callers.
In particular, this prevents LuCI from turning wireless target networks
containing a to-be-created bridge device into bridges themselves.
Fixes: https://github.com/openwrt/packages/issues/18768
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 8effea58d7281fd4f2b6c31d534195dba27a850b)
Due to improper ordering of regexp alternatives, version constraint operators
might get undermatched.
Fixes: https://github.com/openwrt/openwrt/issues/9982
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit b84791bd029c2aaa67cbc06a15d23828f62f4608)
banIP 0.7.x is not compatible with new nft firewall (default in master and 22.03).
Mark the package as BROKEN for now.
Signed-off-by: Dirk Brenken <dev@brenken.org>
(cherry picked from commit 8837f7608adf770ff33249821d28da097324c54d)
If specific settings such as the protocol, IP address or netmask of an
interface the user is connected to are changed, the apply/rollback
mechanism might interfere. Display an additional warning dialog in this
case, instructing the user to manually reconnect and offering to continue
with a less safe unchecked apply mechanism.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 733ee9a7b6312e9b5fe2a4aa8107baf5379602ab)
Previously, "Apply unchecked" was only possible from the save action of CBI
maps, allow to invoke it from the configuration change dialog as well.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 8583efb5a92785fd56ff4abb0ce56a142eb6602f)
Do not abort the poll loop if some confirm request attempts fail, which
might happen when the device starts rolling back its configuration.
Before that fix, the rollback information dialog offering an unchecked
apply was not reliably presented in such situations.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit b6797821b86678d8dd604ce4805691784e6dc3ce)
On boards with ADSL instead of VDSL support we need to expect an
`/sbin/dsl_cpe_control` instead of an `/sbin/vdsl_cpe_control` executable.
Ref: https://forum.openwrt.org/t/dsl-line-stats/126580
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 111c551cdb8d14e8e5ef7c7a66ffdceb6d3cbb55)
Always offer english, regardless of whether a dummy language pack for it
is installed or not.
Fixes: #5783
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit c2eb30bba7451d16e1de37411624b2bcc449c037)
In the worst case, traceroute might take up to 30s to complete when tracing
a completely unreachable path, due to the default limit of 30 hops multiplied
by the per-hop deadline of 1 second (-w 1).
Limit the maximum amount of hops to the configured rpc execution timeout,
20 second by default, in order to avoid running into timed out rpc requests.
Fixes: #5782
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 1ae142d7d11862b11a6e0cea06562875cc778ae7)
Do not expose the iptables status page as menu item when nftables is present
on the system. Instead add a warning banner to the nftables status page
directing the user to the hidden iptables status page when we encounter
legacy rules on the system.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 2f80fe3767207e2dbb8c5286603f49808b66a60d)
The "absent" dependency type requires the given path to not exist on the
local system for the condition to be satisified. This is useful to disable
menu nodes depending on the presence of specific files.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 2e4b5fb8ff2fb3bfd1400bf7cbf721057f02fab9)
