At the point this script is running from udev (the system log is not yet
available), so to avoid debug messages from qualcomm wireless module
reduce the console log level.
https://jira.digi.com/browse/DEL-4570
Signed-off-by: Arturo Buzarra <arturo.buzarra@digi.com>
At the point this script is run (from udev) the system log is not yet
available, so use the kernel log buffer instead.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Use the same udev script that loads the qualcomm wireless module to
create the 'wlan1' virtual interface.
https://jira.digi.com/browse/DEL-4453
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Udev comes with a bunch of generic rules for the hardware. One of them:
/lib/udev/rules.d/80-drivers.rules
is able to load the qualcomm wireless module.
At the same time our new rule is needed, so it prepares the firmware
before the module is loaded.
Both rules running collide with following error message:
udevd[181]: '/etc/udev/scripts/qualcomm.sh' [184] terminated by signal 1 (Hangup)
So to fix this, we rename the specific rule so it's parsed later, and
then we use the hard assignation '=' instead of '+=' to override the
values that come from the previous parsed rule.
Also clean-up a bit the 'qualcomm.sh' script and the modprobe file,
given that now we know the standard drivers rule won't run for this
interface.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
NetworkManager is not able to detect and manage the wireless interface
until is actually created, so let udev load the kernel module and create
the interface on boot.
Signed-off-by: Mike Engel <Mike.Engel@digi.com>
Since 30c253e108 the module is compiled from a repository which already
includes the patches, so we can remove them from DEY.
Signed-off-by: Diaz de Grenu, Jose <Jose.DiazdeGrenu@digi.com>
Tweaked to maintain some recipes' revisions to AUTOREV instead of the
fixed SHA1s from the tag.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
This commit adds support to read the GPU reserved memory
from the DT.
Signed-off-by: Mike Engel <Mike.Engel@digi.com>
https://jira.digi.com/browse/DEL-3868
The firmware configuration and the driver configuration are part of the
driver repository. Stop duplicating the content in meta-digi so they
are updated with the driver itself.
https://jira.digi.com/browse/DEL-4021
Signed-off-by: Alex Gonzalez <alex.gonzalez@digi.com>
The amount of patches to this driver maintained in meta-digi is
substantial, so we are moving to a git repository release model.
https://jira.digi.com/browse/DEL-3710
Signed-off-by: Alex Gonzalez <alex.gonzalez@digi.com>
Seems that this is needed for the bluetooth interface to work. Without
this options, the interface is not able to initialice (hciattach command
fails).
https://jira.digi.com/browse/DEL-3855
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Loading firmware by userspace helpers has been deprecated from device
managers (like udev, eudev, etc). Firmware loading is supposed to be
performed by the kernel.
This fixes the loading of an external kernel module (qualcomm wlan.ko)
that was failing to load because of a missing firmware file.
In that case the kernel was requesting the firmware load from userspace:
ar6k_wlan mmc0:0001:1: Direct firmware load for qsetup30.bin failed with error -2
ar6k_wlan mmc0:0001:1: Falling back to user helper
The device manager we are using in DEY-2.2 (eudev) does not support
firmware loading and is not giving an error return value to the kernel
either, so at one point the kernel throws an oops:
Hardware name: Freescale i.MX6 Ultralite (Device Tree)
task: 88228000 ti: 88cbe000 task.ti: 88cbe000
PC is at 0x7f183758
LR is at request_firmware+0x38/0x40
pc : [<7f183758>] lr : [<802f80bc>] psr: 600f0013
sp : 88cbfdb8 ip : 00000000 fp : 88644c00
r10: 0000dac0 r9 : 88cbfefc r8 : 0000000c
r7 : 00007d08 r6 : 88cbfde4 r5 : 88644ea2 r4 : 00000005
r3 : 88cbfde4 r2 : 00000080 r1 : 00000000 r0 : fffffff5
and the wireless driver is not loaded.
https://jira.digi.com/browse/DEL-3856
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
The way the kernel artifacts are generated has change as of Yocto 2.2.
Also some of the variables (e.g. KERNEL_IMAGE_SYMLINK_NAME) have changed
their default values.
Thus the trustfence_sign function needed some tweaks to continue working
properly.
https://jira.digi.com/browse/DEL-3834
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
KERNEL_IMAGE_BASE_NAME and KERNEL_IMAGE_SYMLINK_NAME default values have
changed in Yocto 2.2, so now this appended command is failing because
it's translated to:
ln -sf -4.1-r0.2-ccimx6ulstarter-20170216122147.bin ccimx6ulstarter
which fails with:
ln: invalid option -- '4'
Just remove it, because we don't need that symlink anymore. New U-Boot's
'zimage' and 'uimage' environment variables have default values ending in
'.bin' which is what Yocto provides.
https://jira.digi.com/browse/DEL-3451
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Tweaked to maintain the u-boot and linux revisions to AUTOREV instead of
the fixed SHA1s from the tag.
Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Protect the invocation of the _scan_done() callback function with the
global lock to avoid that it is called while the module is being unloaded
and the data structures have been freed.
Additionally, the commit also adds a patch to reduce the log level of the
driver that is logging some annoying messages.
https://jira.digi.com/browse/DEL-3607https://jira.digi.com/browse/DEL-3393
Signed-off-by: Pedro Perez de Heredia <pedro.perez@digi.com>
Arturo Buzarra
Diaz de Grenu, Jose
Javier Viguera
Mike Engel
Hector Palacios
Alex Gonzalez
Tatiana Leon
Sebastian Pastor
David Escalona
Pedro Perez de Heredia