If the ADV7511 was probe-deferred, then the HDMI codec can have
different IDs. Use wildcard to allow different IDs.
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Correct the DB410c tests to follow upstream changes:
- the SMD device migrated to use new qcom-rpm-proc driver
- smem device name has changed to include address
- APCS clock device name is now autogenerated.
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
For kernel CI purposes it's convenient to always have access to the
latest bootrr.cpio archive. Introduce a workflow that composes this and
generates an artifact.
Signed-off-by: Bjorn Andersson <quic_bjorande@quicinc.com>
Generated using boardrr-generate-template, with test case names adjusted
to be human readable.
Signed-off-by: Bjorn Andersson <quic_bjorande@quicinc.com>
Generated using boardrr-generate-template, with test case names adjusted
to be human readable.
Signed-off-by: Bjorn Andersson <quic_bjorande@quicinc.com>
Generated using boardrr-generate-template, with test case names adjusted
to be human readable.
Signed-off-by: Bjorn Andersson <quic_bjorande@quicinc.com>
`command -v` is invoked to check if lava-test-case exists, to know if we
need to fake it using `echo`. But in the event that lava-test-case
actually exist it also prints the full path of the executable.
Pipe the output to /dev/null, as we only care about the exit code.
Signed-off-by: Bjorn Andersson <quic_bjorande@quicinc.com>
In the typical scenario where the bootrr cpio archive is concatenated
onto some ramdisk, attributes of this cpio archive will overwrite those
previously extracted.
When used in combination with a ramdisk with systemd, this results in
many services failing to start, because the namespaced services does not
have permission to open e.g. /
Fix this my overriding the owner and group.
Signed-off-by: Bjorn Andersson <quic_bjorande@quicinc.com>
Handles drivers with spaces in their names, and simplify slightly by
using globbing to unwrap the bus for loop.
It's necessary to add the final '*' in the "for driver in" loop in
bootrr-generate-template on POSIX sh.
Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
Make use of the newly added assert_usb_probed helper for USB devices on
Spherion, as it relies on documented kernel ABI and will therefore
require less maintenance.
Signed-off-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>
Current bootrr helpers, assert_device_present and assert_driver_present,
use driver and device names, both of which are not part of the kernel's
stable ABI and end up requiring extra maintenance whenever those names
are changed, in order to check for the expected name on each kernel
version.
Introduce a new helper, assert_usb_probed, that verifies a USB device
has been probed, taking as parameters the device's hardware identifying
properties, which are documented kernel ABI: idVendor, idProduct,
bcdDevice, bDeviceClass, bDeviceSubClass, bDeviceProtocol,
bInterfaceClass, bInterfaceSubClass, bInterfaceProtocol,
bInterfaceNumber.
A 'count' parameter is also required, to inform the number of devices
matching these criteria that should have been probed. This allows the
test to verify the probe of multiple identical devices.
A '*' can be used as wildcard for any of the matching fields as
necessary.
Signed-off-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>
This is a modular system, it could be split up further (eg, there will
be a lot shared with the WiFi variant) but this is what I have to hand.
Signed-off-by: Mark Brown <broonie@kernel.org>
