The result of the modprobe operation doesn't guarantee the driver loaded
correctly. The module may succeed and be loaded, but failures during the
probe may not propagate and result in an error of modprobe.
The valid condition is to check the existence of the interface in
/sys/class/net/wlan0.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
https://onedigi.atlassian.net/browse/DEL-8821
In commit a84d011 this was fixed for mount_digiparts.sh, but
was missing the fix in the mount.sh script.
In that commit there is a full explanation about why this
solves the issue.
https://onedigi.atlassian.net/browse/DEL-8721
Signed-off-by: Francisco Gil <francisco.gilmartinez@digi.com>
With this new rule, only the medias that contain a filesystem
on them are mounted, filtering several calls to mount.sh.
I have checked that this change doesn't increase the boot time
at all.
https://onedigi.atlassian.net/browse/DEL-8826
Signed-off-by: Francisco Gil <francisco.gilmartinez@digi.com>
Backport of graphics package for ccimx93 from NXP's Mickledore-based
lf-6.1.55-2.2.0 release. Mainly copies and appends of graphics recipes
from the new release, and restricted to ccimx93 by changing the
COMPATIBLE_MACHINE, so it does not affect other platforms.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
We removed this package from the recovery initramfs, but it was getting added
as a recommendation to the trustfence initramfs as well.
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
We use cryptsetup in both the recovery and trustfence initramfs to handle block
device encryption on platforms that use eMMC as the internal storage. In
meta-openembedded commit 1ce71d6ec31195280073adec0e400dda7c0dd8a7 (between
DEY 2.6 and 3.0), a lot of PACKAGECONFIG options were added to the cryptsetup
recipe, all of them enabled by default. One of these options is "udev", which
adds a runtime dependency with udev that in turn pulls in a lot of other
dependencies with it. This is unnecessary because we already use mdev in our
recovery image, and the other packages simply take up space as they aren't
needed at all.
Our eMMC partition encryption functionality, which is the reason why we need
cryptsetup to begin with, is unaffected by this change and still works as
expected. Keep in mind that it was working properly without udev in DEY 2.6 and
older DEY releases, so removing udev isn't an issue.
Remove this PACKAGECONFIG option to get rid of udev and its dependencies in
the recovery and trustfence initramfs.
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
In theory, libcap's pam_cap module should use libpam's dlopen runtime linking
mechanism, but starting in libcap 2.53, this changed in order to support
distros that can only link it at build-time. Then, in 2.54, a compile-time
option was added (FORCELINKPAM) to allow users to force the link or not. In the
poky recipe for libcap, this option isn't set, so the library was getting
linked by default, dragging all of its dependencies with it.
For more info: https://bugzilla.kernel.org/show_bug.cgi?id=214023
Set the flag to "no" to avoid linking against libpam, removing its dependency
from the recovery image and saving some space.
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
The TF-A and OP-TEE images have different suffixes depending
on whether TrustFence is enabled or not, but the suffix variables
themselves must exist independently of whether TF is enabled.
Currently, they were defined on the trustfence.bbclass, and the
variables did not exist when TF was disabled, which caused build
problems, for example, building the SWU file.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
This compression method has a better compression ratio than the default (lzo),
but the (de)compression speed is slower, leading to slower read/write speeds.
However, the recovery partition only gets accessed in specific use cases such
as software updates and partition encryption, and it contains an initramfs. The
UBIFS will only be read at boot time to load its elements, but once the
initramfs is loaded in RAM, there are no more read/write operations to the
UBIFS during runtime, so the speed penalty is minimal.
Take advantage of the improved compression ratio to reduce the size of the
recovery image. On the ccimx6ul, the size is reduced by 248 KiB.
https://onedigi.atlassian.net/browse/DEL-8819https://onedigi.atlassian.net/browse/DEL-8825
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
This package is purely vestigial and hasn't been used at all in any of our
initramfs images for over 7 years. It was used in the first implementation
of the trustfence initramfs (commit 4dd7d438af)
to securely erase the key used by cryptsetup, but two weeks later, that logic
was moved to trustfence-tool (commit a8c50c16ea)
and "wipe" was no longer needed. However, the package remained in the
dependencies of the trustfence initramfs, which were then used as reference for
the recovery initramfs, even though this initramfs didn't even need "wipe" to
begin with.
Removing this package saves 30.8 KiB in the ccimx6ul recovery image.
https://onedigi.atlassian.net/browse/DEL-8819
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
This is added as a recommendation for libcrypto and only contains legacy
ciphers which are deemed outdated or unsafe. If any of the packages in the
image actually required this module, they would depend on it explicitly, but
this isn't the case. This saves 66.8 KiB on the ccimx6ul recovery image.
https://onedigi.atlassian.net/browse/DEL-8819
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
We depend on libarchive so that swupdate can perform updates with .tar.gz
files, but by default, libarchive is configured to support a lot of additional
features that pull in even more dependencies. This increases the size of the
recovery image, which barely fits on platforms with limited internal storage
like the ccimx6ul. We already removed zstd support to mitigate this, but the
image size increased again after adding imx-kobs to it in commit
aa2480d1de.
Since libarchive isn't used by any other package in our system, remove the
configurations that bloat the recovery image and are not needed at all for
our file update use case. Note that removing the "lzo" packageconfig doesn't
remove its respective library from the image because mtd-utils also depends on
it, but it still reduces the size of libarchive. On the ccimx6ul, this removes
193 KiB worth of libraries and trims libarchive down by 20.5 KiB in the
recovery image.
https://onedigi.atlassian.net/browse/DEL-8819
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
These fields were added to default files, but not to the
special sw-description files for ccmp1 and cc6ul platforms.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
When TrustFence is enabled, the boot artifacts (TFA and FIP)
have a 'signed' suffix. Handle this case so that the correct
symlinks are created and the correct artifacts are put into the
SWU file.
Signed-off-by: Mike Engel <Mike.Engel@digi.com>
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
For signing SWU files we need to set a couple of variables:
- SWUPDATE_PRIVATE_KEY_TEMPLATE to the private key file
- SWUPDATE_PASSWORD_FILE to the password of the private key
The latter must only contain one password, whereas the current key_pass.txt
file had (for the ccmp13) the eight keys separated by a white space.
This commit:
- If the file key_pass.txt exists, it extracts each key into a separate
file key_pass0X.txt.
- If the keys don't exist, generates separate files per key.
- Changes the permissions of password files to 400.
- Adapts the sign script to use the single password files.
- Fixes a few quotes
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
Starting with NXP release "lf-6.1.55-2.2.0" the IMX optee fork (based on
version 4.0.0) does not support SOC revision A0. This commit recovers
support to build a bootloader for A0, extending the optee patch for
ccimx93 to support A0 with a build time option, and then extending the
optee-os and imx-boot recipes to build two optee binaries and using them
to generate bootloaders for both SOC revisions.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
This includes also an update and rename of the Edgelock Enclave firmware
package (firmware-ele-imx).
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
We were using the .cfg suffix for both the build-time config fragments and the
runtime configuration file. During do_configure(), all files in SRC_URI ending
in .cfg were being merged together to create the final build configuration,
including said runtime file, which has a completely different syntax. In most
cases, the contents of this file were being ignored, but when tweaking
swupdate's configuration and re-building the package, sometimes strange errors
would prevent the build from finishing.
Change the runtime file's suffix entirely to separate it from the config
fragments and prevent it from being treated as such, and reflect the name
change in the defconfig and the recovery script.
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
meta-swupdate recently added support for swupdate 2023.12 and libubootenv
0.3.5. This upgrade conflicts with our custom functionality in both packages,
so keep using 2023.05 and 0.3.2 respectively so DEY 4.0 can be built without
introducing possible functionality changes.
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
Commit 7e81e706a1 modified the logic used to
determine if a NAND is using multi-MTD or single-MTD, but single quotes were
used in the grep pattern, which prevents the variable inside from expanding.
This makes the script always enter the single-MTD code path, even in multi-MTD
systems, which breaks partition formatting in these cases.
Replace the single quotes with double quotes so the variable can expand and the
condition can be properly checked.
https://onedigi.atlassian.net/browse/DEL-8773
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
Fixed via commit 311a492c21f0f3643f4b764f8b743c4c6908d31b in meta-st-stm32mp
This reverts commit 09c4645894.
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
Commit d3f3cfdb62 removed the inclusion of STM's
linux-stm32mp.inc from meta-st-stm32mp in our linux-dey recipe, but this
inadvertently removed the logic in do_configure() necessary to use our custom
ccmp1_defconfig. Since this commit, the kernel was being built with the default
ARM defconfig, which is very different from our custom defconfig and doesn't
even boot on MP1 platforms.
Rework the logic used to copy our platform's defconfigs to prevent this.
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
This commit removes the BOOTLOADER_SEEK_OFFSER, because it has been
moved into the swupdate class and renamed into SWUPDATE_UBOOT_OFFSET.
Signed-off-by: Mike Engel <Mike.Engel@digi.com>
This commit adds u-boot swupdate support for all platforms.
Now u-boot can be updated with all our supported update
options. Currently it will only update first partition
u-boot partition.
https://onedigi.atlassian.net/browse/DEL-8749
Signed-off-by: Mike Engel <Mike.Engel@digi.com>
The 'mtd-blacklist' parameter prevents swupdate from acting upon those
partitions that we consider sensitive.
Make such list platform-dependent.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
The swupdate recipe installs by default a systemd service
and a socket to listen for updates coming from a web server.
DEY only makes use of such service during on-the-fly updates from Cloud
Connector web service.
The default swupdate service fails on images with TrustFence because it's
called with no arguments and there doesn't exist a configuration file.
This commit installs a default configuration file and, if TrustFence is
enabled, sets the parameter 'public-key-file' to point to the public
certificate to use to authenticate SWU packages.
While on it, it removes the same file from the recovery-initramfs recipe
that was the only recipe that was adding such config file for recovery
images only.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
meta-digi layers use many conditionals basing on TRUSTFENCE_SIGN, but this
variable may be disabled when the signing process wants to be isolated
from the image creation.
There are cases when we still need to know if TrustFence is enabled even
if the images are not going to be signed.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
In commit df9b1cf329, the UBOOT_SIGN_ENABLE is set for all
platforms, and should be only added for FIT images.
This is making the process failing in cc8mn/cc8mm platforms
due to the UBOOT_SIGN_ENABLE is also used there to use a dtb
patched with the signature node.
https://onedigi.atlassian.net/browse/DEL-8764
Signed-off-by: Francisco Gil francisco.gilmartinez@digi.com
Otherwise, bitbake's git client won't be able to fetch the sources if there are
no mirrors available.
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
The LEGACY_IMAGE_FORMAT defaults to 'y' if there is no FIT
support, which happens after applying the default configuration.
Then, the FIT support is added in a config fragment, but the
LEGACY_IMAGE_FORMAT is not disabled.
Disabling this is recommended to avoid the possibility to boot
unsigned legacy images.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
As part of the integration of the new ML package, also update the
ethos-u-firmware binary built from Stash:
Repo: emp/ethos_u_firmware.git
Revision: a0352b0cf7267c896b55980dedd08daddc780733
Source: NXP's MCUXpresso SDK_2.14.0_MIMX9352xxxxK
And, drop deepview-rt support for ccimx93, as NXP has done the same in
6.1.36-2.1.0 release.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
The U-Boot bootscript loads the fitImage into RAM memory to run
this bootscript. This bootscript ends up calling 'dboot' command
to run the FIT default configuration.
To avoid 'dboot' re-loading again the fitImage into RAM memory,
set this temporary variable that will be immediately reset
by 'dboot'.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
- Set variables required for FIT signing inside python function, under the
condition of having TRUSTFENCE_SIGN="1".
- Define two sign keys using TRUSTFENCE_ wrapper constants. Default values:
- 'fitcfg' for configuration nodes inside the FIT
- 'fitimg' for image nodes inside the FIT
- Enable FIT_SIGN_INDIVIDUAL to also sign individual images inside the FIT
- Set FIT_GENERATE_KEYS by default (kernel-fitimage.bbclass already checks
if the keys exist before generating new ones)
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
An anonymous function in linux-stm32mp.inc produces a bbfatal error when
KERNEL_DEVICETREE variable contains more than one device tree. This is our
case since we build the main DT plus a number of DT overlays.
This commit removes the dependency to this include file since we have our
own recipe to build the kernel and it is not needed at all.
It also removes the build of a uImage and the need to provide a
LOADADDR.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
This change isn't functional, but necessary if we want to be able to separate
the logic in the include file and the demo recipe. This way the include file is
agnostic and can be included into other recipes if needed.
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
Otherwise, launching a Qt5 app will result in errors and the QPA platform will
need to be set manually to get it to work.
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
Backport of graphics package for ccimx93 from NXP's Mickledore-based
6.1.36-2.1.0 release. Mainly copies and appends of graphics recipes
from the new release, and restricted to ccimx93 by changing the
COMPATIBLE_MACHINE, so it does not affect other platforms.
Notice, that the new version of weston used now by the ccimx93 requires
a different profile file (weston-socket.sh). This profile supercedes the
old 'weston.sh'.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Backport of multimedia package for ccimx93 from NXP's Mickledore-based
6.1.36-2.1.0 release. Mainly copies and appends of multimedia recipes
from the new release, and restricted to ccimx93 by changing the
COMPATIBLE_MACHINE, so it does not affect other platforms.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
This is required to build the new version of gstreamer used by NXP for
the ccimx93 and integrated in following commits.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Hector Palacios
Francisco Gil
Javier Viguera
Hector Bujanda
Gabriel Valcazar
Mike Engel
Isaac Hermida