The previous socket wait fixed a race where the login prompt could appear
before Weston had created its Wayland socket. In that case, logging in as
root left WAYLAND_DISPLAY empty and applications could not use waylandsink.
Only wait for the socket when a DRM display is connected, or when the kernel
does not expose DRM connector status files. This keeps the race fix for
display boots while allowing immediate headless logins.
Keep showing a warning when there is no wayland socket.
https://onedigi.atlassian.net/browse/DEL-10141
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
Bootable SD cards may boot with root=/dev/mmcblkXpY instead of
root=PARTUUID=... In that case mount_digiparts.sh treated the system
as NAND/UBI and failed to mount the Digi partitions.
Resolve the active system from the block device mounted as "/" before
falling back to the existing UBI logic.
https://onedigi.atlassian.net/browse/DEL-9676
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
ConnectCore 95 integrates the Murata 2EC Wi-Fi module and requires applying the
vendor custom patch set across the supported releases.
https://onedigi.atlassian.net/browse/DEL-10142
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
Update nl80211 definitions to match Linux v6.6.116, preventing extended feature
mismatches and fixing parsing of driver-reported capabilities.
https://onedigi.atlassian.net/browse/DEL-10142
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
Adjust the ccimx95-specific networking pieces so they no longer inherit
the IW612-specific behavior from ccimx9:
- use wlan1/p2p0 instead of uap0/wfd0 where appropriate
- install hostapd_wlan1.conf instead of hostapd_uap0.conf
- restore virtual WLAN handling for ccimx95
- avoid IW612-specific suspend/resume actions such as moal reload and
load_iw612.sh
Signed-off-by: Francisco Gil <francisco.gilmartinez@digi.com>
Some Bluetooth controllers may expose hci0 even when the firmware
initialization has not completed correctly. In that state, the init
script may report success but bluetoothd cannot use the controller.
Validate the controller through the kernel management interface before
accepting the initialization as successful. This matches the interface
used by bluetoothd and catches controllers that are visible through HCI
but not registered in MGMT yet.
Use a timeout for the MGMT query so a broken controller state cannot
block the init script instead of falling back to the retry loop.
https://onedigi.atlassian.net/browse/DEL-9512
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
Make this service generic so it can be reused on every device containing
a Murata+Infineon chip.
In some platforms Worldwide pseudo-country could be 'XZ', in others it
could be 'WW'. Cover both of them.
In the meantime, improve the service to ensure the wifi device is
available by waiting for udev device discovery to settle, retrying
until the wireless interface responds to wl, and restarting the
oneshot unit on failure.
https://onedigi.atlassian.net/browse/DEL-9798
Signed-off-by: Gonzalo Ruiz <Gonzalo.Ruiz@digi.com>
Add CLM blob files with power limits based on the CCMP25 Certification
process.
There is a different CLM blob file to be used if the device operates
as STATION or Access point (AP) or if it is meant to be used Indoors or
Outdoors. Every file is deployed into the system, but only
STATION+INDOOR files are selected by default.
For each use case there is a US-only and a Worlwide CLM blob file.
Initial list of countries covered by the Worldwide binaries is:
* US
* Canada
* Europe
* Japan
* Australia
* New Zealand
Files:
- cyfmac55500-sdio_US.APIndoor.clm_blob (MD5SUM: 3e6f5fdcf9a88cf425179c69862ac67f)
- cyfmac55500-sdio_US.APOutdoor.clm_blob (MD5SUM: f8655f59b87ac5a71a2a90f08c52755b)
- cyfmac55500-sdio_US.STAIndoor.clm_blob (MD5SUM: e2bf5ed665876203085bd0cc9c90e590)
- cyfmac55500-sdio_US.STAOutdoor.clm_blob (MD5SUM: ddc81881d8ed0b70e2b843bb7899e758)
- cyfmac55500-sdio_World.APIndoor.clm_blob (MD5SUM: e48416c461469652ef5594a402a596cd)
- cyfmac55500-sdio_World.APOutdoor.clm_blob (MD5SUM: 5c57605395d4e186c58b59e4a57d80cd)
- cyfmac55500-sdio_World.STAIndoor.clm_blob (MD5SUM: 3f54dbf06d633d7b1aad816f6d655b9d)
- cyfmac55500-sdio_World.STAOutdoor.clm_blob (MD5SUM: 9f706f9ae5d833e386ebabc21e7c779b)
https://onedigi.atlassian.net/browse/DEL-9798
Signed-off-by: Gonzalo Ruiz <Gonzalo.Ruiz@digi.com>
Add full support to face-recognition demo so it can be integrated in the
webkit demo.
Improvements are:
* support for USB camera
* use 640x480 resolution so it fits in our COG demo
https://onedigi.atlassian.net/browse/DEL-10133
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
dey-image-container is not a full image, so the regular image cleanup
does not remove old DCP artifacts from the deploy directory.
Add a dedicated cleanup task that honors RM_OLD_IMAGE and removes previous
LXC and Podman DCP artifacts before generating the new ones.
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
Generate the LXC payload tarball with config and rootfs/ at the
archive root so it matches the documented DCP format and the strict
installer expectations.
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
Expand the BT_GPIO placeholder after installing the init script, so
optional scripts such as bluetooth-init_atheros are handled the same way.
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
Sync STM32MP-specific RT kernel patches, aligned with the latest
ST release from meta-st-x-linux-rt layer.
Signed-off-by: Mike Engel <Mike.Engel@digi.com>
The udev mount script always queried the active system before mounting
any partition. On eMMC systems this lookup used /dev/disk/by-partlabel,
which is populated by udev itself and may still be incomplete and the
script exited for that partition.
Only resolve the active system for linux_a/linux_b, and read the eMMC
root partition name from sysfs instead of the udev-created by-partlabel
links.
https://onedigi.atlassian.net/browse/DEL-10131
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
create_st_fip_binary.sh converts binary encryption keys to hex before
passing them to encrypt_fw. hexdump may collapse repeated binary data
into '*', corrupting the key and causing encrypt_fw to reject it with an
invalid key size.
Add a bbappend to patch the helper from meta-digi and use hexdump -v so
the full key is emitted.
https://onedigi.atlassian.net/browse/DEL-10115
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
When TrustFence encryption is enabled, the TF-A recipe converts the
binary FIP encryption key to a hexadecimal string and passes it to the
TF-A build as ENC_KEY.
hexdump suppresses repeated output by default and emits '*' when it folds
duplicate data. If that happens while converting the key, the generated
ENC_KEY contains the '*' marker instead of the full hex string. The
malformed value then breaks the make command line, and make can interpret
the remaining key fragment as a target, failing for example with:
make ... ENC_KEY=13eb911dfcc316b1b99bbbbf10a7000c3055e863* 166d3c8c9cd6f888b720a9 all
make: *** No rule to make target '166d3c8c9cd6f888b720a9'. Stop.
Use hexdump -v so every byte of the binary key is emitted and ENC_KEY is
always a complete, contiguous hex string.
https://onedigi.atlassian.net/browse/DEL-10115
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Make the minimal changes needed to adapt the original X-LINUX-AI example and
integrate it into the Digi custom launcher.
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
Restore the V4L2SRC based pipeline for USB webcams that was dropped in newer
X-LINUX-AI releases, fixing USB camera support on the v6.2.0 baseline.
https://onedigi.atlassian.net/browse/DEL-10040
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
Read the QR serial number from /proc/device-tree/serial-number and the
MAC address from the U-Boot ethaddr environment variable.
Include digi,drm-ic in the QR payload only when present, and update the
console output and provisioning message accordingly.
Signed-off-by: Francisco Gil <francisco.gilmartinez@digi.com>
Drop the imx-m7-demos dependency and build only the A55 boot target,
since Cortex-M firmware is board-specific and should not be selected as
a default SoC artifact. This follows the same cleanup done upstream in
meta-freescale commit 598f047ef112a0c88f699714cd10dc87c8460c5a.
https://onedigi.atlassian.net/browse/DEL-9882
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
(cherry picked from commit 81cf596834382f87701d701277e13c97316de978)
After refactoring these scripts in a8c6dcb56e, some platforms are
not setting the enable line correctly.
Regarding the affected platforms, on the 6UL it works because that GPIO
comes enabled by default, but not on the other ones.
For the other platforms refactored, such as the CC91/93 or CCMP25,
it is not needed, as the variable is used directly in the gpioset command.
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
Add the complete Digi recommended set of Wi-Fi tools, if not wpa-cli or
hostapd are not included.
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
The package is no longer fetched from Git, so rename the recipe to a
versioned .bb file and fetch the proprietary artifact from Digi server.
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
The shutdown helper was installed and enabled, but during reboot it was
never started.
Hook the helper directly into shutdown.target.
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
CCCS service starts before the network is fully ready and fail the first
connection attempt. When that happens, the daemon waits for the
configured reconnect interval before trying again, which delays DRM
availability for dependent services.
https://onedigi.atlassian.net/browse/DEL-10083
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
This commit generalizes the BT GPIO value used in the bluetooth-init
script for different platforms.
Signed-off-by: Mike Engel <Mike.Engel@digi.com>
https://onedigi.atlassian.net/browse/DEL-9668
The Cloud Connector depends on syslog for logging, so this package is
required to properly collect its messages.
https://onedigi.atlassian.net/browse/DEL-10080
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
Install the wl tool by default. It is a useful utility for customers to
perform additional wireless configuration and debugging.
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
Create task temporary directories under `${T}` instead of the host `/tmp`
and remove them when the task exits.
This avoids leaving large temporary directories behind during container
artifact generation.
Signed-off-by: Francisco Gil <francisco.gilmartinez@digi.com>
Add an extra step to copy build artifacts after each build. Some build targets
require additional layers and custom configurations, and subsequent build steps
can overwrite or remove previously generated artifacts once the configuration is
no longer compatible.
https://onedigi.atlassian.net/browse/DEL-10026
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
The bbappend hardcoded an old libsocketcan revision. Remove the override
so the recipe uses the revision from meta-openembedded. Also refresh our
patch, so it applies cleanly on top of the new revision.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>