By removing the driver on ifdown and loading it on ifup we workaround some
problems found in long run suspend stability testing.
https://jira.digi.com/browse/DEL-1084
Signed-off-by: Alex Gonzalez <alex.gonzalez@digi.com>
This will be the preferred version for ccardimx28. Then use platform
override to set the one for ccimx5 and ccimx6.
Not having a platform override in the machine config file allows to do
a simple version override in the project's local.conf file with:
PREFERRED_VERSION_u-boot-dey = "2009.08"
Otherwise, if we use a platform override in the machine config that
takes precedence over a normal override in the project's local.conf
and we would have needed something like:
PREFERRED_VERSION_u-boot-dey_ccardimx28 ?= "2013.01"
in the project's config.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Adapt the jenkins build script to build for ccardimx28js by default
linux 3.10 and u-boot 2013.01 and as alternative configuration linux
2.6.35.14 and u-boot 2009.08
https://jira.digi.com/browse/DEL-1168
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
This removes the Openembedded logo and leaves the kernel splash until
the desktop is ready.
https://jira.digi.com/browse/DEL-1131
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Override the default from meta-fsl-arm for MX6 which is configured for
eGalax touchscreen.
We create an empty calibration file because the Fusion touch works just
fine without calibration.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
FSL has this enabled but in our case it fails to build for ccardimx28js
using linux 3.10 with error:
fatal error: linux/mxcfb.h: No such file or directory
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Add platform family to SOC_FAMILY, so it's part of the MACHINEOVERRIDES
variable and it can be used to generalize bitbake recipes.
https://jira.digi.com/browse/DEL-1105
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
This updates the driver to upstream AR6003-ISC-v3.4p6-b3.4.0.225.
https://jira.digi.com/browse/DEL-1127
Signed-off-by: Alex Gonzalez <alex.gonzalez@digi.com>
This was needed to overcome some corner cases with storage devices on
suspend/resume using MDEV as device handler.
Now we are using UDEV and after some tests done it seems that the
'resume_storage_devices' hack is not needed anymore.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
udhcpc's '-R' option means 'Release IP on exit'.
This fixes a strange corner case on reboot: if you configure your network
interface to DHCP in '/etc/network/interfaces' AND you have some NFS
share mounted on your system, then running reboot command fails to
complete and it hangs on:
'Unmounting local filesystem'
This happens because on reboot TERM and KILL signals are sent to all
the processes in the system before umounting the local filesystem. When
udhcpc daemon receives the signal it exits and releases the IP (because
it was launched with '-R') leaving an unconfigured network interface
that later is unable to umount a network filesystem (NFS), resulting on
'reboot' command being unable to complete.
Removing the '-R' option from udchpc allows to kill the process without
unconfiguring the network interface.
https://jira.digi.com/browse/DEL-1125
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
The original defconfig came from DEL. Make the defconfig more similar to
the Yocto default one.
This restores the default udhcpc options when running via ifup to '-R
-n'. This is important because without '-n' the target can wait forever
to get an IP on boot while maybe the network is not available.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
This allows to blacklist platforms not officially supported in a DEY
release. It only works at help-level, you can actually create projects
for any platform whose config files exist, but the help of the mkproject
command will only show the officially supported platforms.
At the moment the BLACKLIST variable is not filled. This will be done in
the release branch.
https://jira.digi.com/browse/DEL-1073
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Packagegroups define sets of packages to be installed in the rootfs
*depending* on the hardware variant (~MACHINE_FEATURES). For example
installing or not the bluetooth firmware depending on machine variant
having bluetooth.
It may happen that we build sequentially machines with different
features using the same sstate and thus it may happen we end with a
rootfs lacking bluetooth firmware for a bluetooth variant.
So add all packagegroups to the list of packages to purge state-cache
between platform builds.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
This adds the calibration file for US regulatory (still missing the
world-regulatory calibration file)
https://jira.digi.com/browse/DEL-1102
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
In Yocto Daisy mplayer is not supported. Trying to build it fails with:
ERROR: Nothing PROVIDES 'mplayer'
ERROR: mplayer2 PROVIDES mplayer but was skipped: Recipe is blacklisted: Requires newer libav which has negative D_P
This is because it has been blacklisted in commit:
http://cgit.openembedded.org/meta-openembedded/commit/?id=2e7f3ce93b2c9864037fba8cc68fde46aafac4b0
The reason is that is has a build-time dependence on libav being newer
than the one provided by Yocto.
We don't use mplayer on ccimx6 as it has accelerated video using
gstreamer, so just remove the bbappend for this release.
Probably we need to revisit this when we upgrade our ccardimx28 support
to new versions of Yocto and upstream has removed the recipe from the
blacklist.
https://jira.digi.com/browse/DEL-1075
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
This synchronizes the machine configuration files with the latest
changes in meta-fsl-arm for Daisy maintenance branch.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Those hacks were added to fix a problem when building development images
(with 'dev-pkgs' image feature enabled). In that case the developemnt
packages (libdigi-dev and microwindows-dev) have a dependence on the
normal packages (libdigi and microwindows). But those packages are not
created by default because they are empty, thus the needed ALLOW_EMPTY hack.
Fix the problem by just resetting the runtime dependences of the -dev
packages. This allows to not create empty libdigi and microwindows
packages.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
While on it, rework and simplify atheros firmware recipe:
* remove 'unknown' version (renamed recipe)
* do not create empty debug and development packages
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Javier Viguera
Alex Gonzalez