This commit removes the default integration of the Flutter framework from the
SDK due to its significant impact on toolchain size, build time, and reliability.
Including Flutter increases the build complexity exponentially, often resulting
in timeouts or failures caused by the large number of recipes involved.
Customers who require Flutter can still enable it manually if needed.
https://onedigi.atlassian.net/browse/DEL-9380
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
Flutter is an open source framework for building multi-platform applications
without a graphical backend running. This commit adds support to create a new
DEY image type based on Flutter ready to use for the users.
https://onedigi.atlassian.net/browse/DEL-9380
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
The IW61x, when enabled as AP or P2P (or bridge with AP), generates udev
triggers that lead the wifi card to fail.
Ignore those events.
The reason of this new behavior is due to poky commit 1e8b49999ad8,
"udev-extraconf: fix network.sh script did not configure hotplugged interfaces"
https://onedigi.atlassian.net/browse/DEL-9465
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
Both python3-cryptography and python3-pyelftools are needed to compile optee-os
with a Yocto-generated SDK, but we were adding these packages to all i.MX SDKs,
even for platforms that don't use optee. This wouldn't be a problem if it
weren't for the fact that python3-cryptography is implemented in rust, and its
inclusion in the SDKs forces bitbake to generate the rust compiler just for
that package.
Only include these python packages for platforms that use optee.
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
Use the generic sys class to check if a mtd device is attached.
The virtual node may not be present in some kernel versions.
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
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>
The nodes "/dev/ramX" and "/dev/loopX" are mounted on boot.
Each node calls the mount.sh script, but they are not mounted
because these nodes are blacklisted in the "blacklist.conf" file.
In the ccmp13 adding this modification in the rule saves
around 4 seconds per boot.
In the ccmp15 and ccimx6ul around 2 seconds are saved.
https://onedigi.atlassian.net/browse/DEL-8725
Signed-off-by: Francisco Gil <francisco.gilmartinez@digi.com>
The appropriate way to add STM signtools to the SDK is via RDEPENDS on
nativesdk-packagegroup-sdk-host, not through the parent recipe of STM
signtools recipe itself.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
https://onedigi.atlassian.net/browse/DEL-8720
The ccmp1 has two MTD partitions (UBI, UBI_2) with different system
volumes.
Previously, the fact of having two ubi devices was taken as proof of
being on a multi-MTD system (one that has one UBI volume per partition).
Instead, this commit reformulates the condition to having a partition of
the same name than the UBI volume.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
When the system runs as read-only with '/etc' overlayfs enabled, '/sbin/init' is not a symlink
to 'systemd', but '/sbin/init.orig' is. This causes the script to wrongly think than 'systemd'
is not being used. The outcome is that partitions are not being automatically mounted in R/O
systems running in mmc with '/etc' overlayfs enabled, as the specific '/usr/bin/systemd-mount'
binary needs to be invoked in these cases.
The proposed fix performs a second check with '/sbin/init.orig' to determine whether systemd is
being used or not. While on it, add an early check to determine if the partition is already
mounted to exit the script.
Signed-off-by: David Escalona <david.escalona@digi.com>
This commit disables the uSD mounting, due to issues
detected during the boot process when UBIFS starts
with the wear-leveling process to erase NAND blocks.
https://onedigi.atlassian.net/browse/DEL-8415
Signed-off-by: Mike Engel <Mike.Engel@digi.com>
The name of the variable was not very intuitive of what
it contains. This variable expands to the SoC vendor
(NXP or STM).
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
For devices running 'systemd', the mount process is done through 'systemd-mount', which
spawns a system unit. Trying to mount the partition again throws an error stating that a
unit with the same name is already running. To avoid it, check if the unit is already
spawned before mounting the partition and, if so, just restart the unit.
Signed-off-by: David Escalona <david.escalona@digi.com>
Our /tmp file is a symbolic link to /var/tmp, therefore a "find /tmp" returns
nothing. Force to search files inside tmp folder by using "find /tmp/".
https://onedigi.atlassian.net/browse/DEL-8410
Signed-off-by: Isaac Hermida <isaac.hermida@digi.com>
In the past, this file was cloned from poky to skip mounting
an already mounted device. This condition was done on poky's
version so the cloned file is no longer needed.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
This commit fixes the issue in Jira
https://onedigi.atlassian.net/browse/DEL-7827 (and in GitHub
91bfa01a52#),
that is being manifest again in a CC6UL with 1GB although it may affect other
hardware.
Partitions are mounted:
* For multi-MTD systems, when an MTD subsystem event is received.
* For single-MTD systems, when a UBI subsystem event is received.
This commit avoids process UBI subsystem events in multi-MTD systems.
The automount rules filter UBI subsystem events to only process 'ubi0*' (the
only UBI device/volumes in a single-MTD) and then the executed script
'mount_digiparts.sh' checks for the existance of '/dev/ubi1' node to consider
the system multi-MTD instead of using 'singlemtdsys' U-Boot variable that only
exists on CC6UL.
Signed-off-by: Tatiana Leon <Tatiana.Leon@digi.com>
* Use 'MOUNT' variable instead of 'mount' command
* On the MTD subsystem event, stop looping once the UBI device number is found
Signed-off-by: Tatiana Leon <Tatiana.Leon@digi.com>
Commit b8d46f01 ("read_only: create folders in build time") creates
unconditionally default folders in spite of we are in a read-only filesystem or
not, however there are different models (dual boot, single boot..) where these
partitions exist or not, so this commit only creates these default partitions
for a read-only filesystem.
https://onedigi.atlassian.net/browse/DEL-8193
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
This commit allows to mount only the linux partition active in read-only mode.
https://onedigi.atlassian.net/browse/DEL-8193
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
According to udev documentation running programs that access to mount/umount
filesystems is not allowed due to the default sandbox that is enforced on
systemd-udevd.service, so on systemd systems we must to use systemd-mount
binary instead of mount. Also systemd-mount does not support mount ubi devices
by volume name, so we need to use the absolute device path.
https://onedigi.atlassian.net/browse/DEL-8193
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
On the CCMP1 SOMs, we only consider a single system partition called UBI,
with many UBI volumes.
The variable 'singlemtdsys' allowed to change between this and
a system with multiple MTD partitions with one UBI volume each
(old approach on CC6UL).
This commit removes the variable and uses the logic as if it was set.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
The modalias changed as part of the kernel update to version 5.15.
See commit dd545833b773b20aa7cec10f89bff96014ccb64f in kernel tree.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Building a toolchain/SDK fails for ccmp15-dvk, because the NXP-based
trustfence tools (for example the cst) are not available. Fix the build
by filtering out those tools when the build platform is not NXP based.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
In order to perform the standalone signature process, it was required
to rebuild the Toolchain with Trustfence support enabled.
CST source code is now available for downloading in the Digi FTP, so add
Trustfence sign scripts and cst/srktool to the default toolchain for it
to be used for standalone signature without rebuilding.
Signed-off-by: Gonzalo Ruiz <Gonzalo.Ruiz@digi.com>
(cherry picked from commit 2c9b721fb9ce38dcd0034e22d95db6e0ee068955)
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
This code path was added in the context of single MTD layout support for
ccimx6ul boards, but it was being executed in the default storage layout use
case as well. This could trigger a race condition in which one code path
deletes the mount point while the other one is still setting it up, causing
partitions to not be mounted automatically (specifically, the "update"
partition).
Since both code paths are mutually exclusive in our implementation, reflect
this by making sure that the single MTD layout path is executed only when said
layout is enabled in the U-Boot environment.
https://onedigi.atlassian.net/browse/DEL-7827
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
(cherry picked from commit 9a7ea10a4dfb74651f865f125372604fe6840a10)
These folders /mnt/linux, /mnt/update and /mnt/data are
created by the automount_block.sh script. In a read only
system, these folders are not created raising some errors.
Creating these folders solves the issue.
Signed-off-by: Francisco Gil <francisco.gilmartinez@digi.com>
Traditionally, platforms based on NAND, used one UBI volume
per MTD partition.
Now it's possible to use only one MTD partition containing many
UBI volumes.
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
https://onedigi.atlassian.net/browse/DEL-7614
Starting with Linux v4.15, regulatory domains are no longer handled via CRDA,
but via the kernel and a flexible database instead. Said database is included
in the wireless-regdb-static package, which conflicts with the old
wireless-regdb package.
To accomodate for this new package, add CRDA as a dependency for ccimx6
platforms only, and incorporate wireless-regdb-static for the rest of the
platforms. Additionally, the ccimx6 kernel should inherit the
kernel_wireless_regdb .bbclass so the plaintext database is copied to the
kernel sources, but since we use the same recipe for all kernel versions and
ccimx6 platforms are not yet supported in zeus, omit this change for now.
Remove the package-base .bbappend, since only ccimx6ulstarter images include
that packagegroup and they require wireless-regdb-static, anyway.
By adding the regulatory database, we avoid delays when going to suspend,
caused by the kernel waiting for the database to appear.
https://jira.digi.com/browse/DEL-7133
Signed-off-by: Gabriel Valcazar <gabriel.valcazar@digi.com>
imx-mkimage is a host recipe to provide the mkimage_imx8 binaries, required
for the trustfence support with platform based on AHAB (ccimx8x). Since
these binaries are required to the sign process we need to export it in the
SDK to allow the standalone sign mode, and with that we can simplify the
mechanism to share these binaries with another recipes (u-boot, linux).
Also the do_deploy() from imx-mkimage recipe was removed to avoid overriding
the implementation from the native class and allow populating the mkimage
binaries.
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
packagegroup-base includes wireless-regdb-static package unconditionally,
however it conflicts with wireless-regdb that it is required by crda,
which is included as a dependency to packagagroup-dey-wireless.
https://jira.digi.com/browse/DEL-7013
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
Default poky's version of the 'mount.sh' script doesn't account
for the possibility of udev calling twice the script for the 'add'
action. This has been observed on USB sticks connected to a USB hub
and, although the script eventually fails, for already mounted
partitions, in that case the script tries to remove the created
mountpoint, when it shouldn't.
This patch checks for already mounted partitions earlier (before
creating the mount point) and returns with no actions.
The patch that it adds (with respect to poky's thud default) is
the following:
diff --git a/meta/recipes-core/udev/udev-extraconf/mount.sh b/meta/recipes-core/udev/udev-extraconf/mount.sh
index 3ee67b13183a..e6882a69c223 100644
--- a/meta/recipes-core/udev/udev-extraconf/mount.sh
+++ b/meta/recipes-core/udev/udev-extraconf/mount.sh
@@ -38,6 +38,12 @@ done
automount_systemd() {
name="`basename "$DEVNAME"`"
+ # Skip already mounted partitions
+ if [ -f /run/systemd/transient/run-media-$name.mount ]; then
+ logger "mount.sh/automount" "/run/media/$name already mounted"
+ return
+ fi
+
# Skip the partition which are already in /etc/fstab
grep "^[[:space:]]*$DEVNAME" /etc/fstab && return
for n in LABEL PARTLABEL UUID PARTUUID; do
Signed-off-by: Hector Palacios <hector.palacios@digi.com>
https://jira.digi.com/browse/DEL-6759
Arturo Buzarra
Isaac Hermida
Gabriel Valcazar
Francisco Gil
Hector Palacios
Mike Engel
David Escalona
Javier Viguera
Tatiana Leon
Gonzalo Ruiz