device-types: fix broken xilinx-zcu102 device-type

See merge request lava/lava!1519

Add asus-C523NA-A20057-coral device type template

See merge request lava/lava!1512

device-types: add BeagleBone Enhanced

See merge request lava/lava!1511

Add hp-x360-12b-n4000-octopus device type template

See merge request lava/lava!1513

Add asus-C433TA-AJ0005-rammus device type template

See merge request lava/lava!1514

Add acer-mt8183-cp311-3h-jacuzzi device type template

See merge request lava/lava!1515

debian: lava-lxc-mocker: don't provide lxc

See merge request lava/lava!1517

Current xilinx-zcu102.jinja2 only setenv some variables for tftpboot, but
they are not executed actually. As a result, the board could not boot up.
This commit remove the setenv and make it tftpboot directly.

We also set console_device and boud_rate.

Signed-off-by: Hiraku Toyooka <hiraku.toyooka@cybertrust.co.jp>

lava_rest_app: test_api: don't subclass pathlib.Path

See merge request lava/lava!1518

pathlib.Path can't really be subclassed directly, at least on Python 3.9

More info on
https://stackoverflow.com/questions/29850801/subclass-pathlib-path-fails

Saying that lava-lxc-mocker *provides* lxc is too strong a message, and
has impact on other packages:

- any other package that depends on lxc will have it's dependency
  satisfied by lava-lxc-mocker, but will probably not really work.
- when lava gets uploaded to Debian, it triggers running tests against
  packages that depend on lxc and that's pointless.

lava-lxc-mocker can look like lxc, but only for lava. So allow
lava-lxc-mocker where we depend/recommend lxc.

`lava` already depends on `lava-dispatcher`, and that already recommends
`lxc`. This gives admins a little more flexibility, allowing `lxc` to
not be installed if not needed.

Support security fix in pyyaml 5.4

Closes #488

See merge request lava/lava!1516

A few tests need to explicitly use an unsafe loader to be able to test
YAML roundtrips. As a fix for a security issue, PyYAML 5.4 requires us
to be even more specific to request an unsafe loader, what is a
backwards-incompatible change.

On older versions of PyYAML that don't provide UnsafeLoader, we need
to fallback to the regular Loader class.

Signed-off-by: Marc Deslauriers <marc.deslauriers@ubuntu.com>
Signed-off-by: Antonio Terceiro <antonio.terceiro@linaro.org>

Ricardo Cañuelo Navarro authored Apr 12, 2021 and Mylène Josserand committed May 21, 2021

Add a template for the acer-mt8183-cp311-3h-jacuzzi arm64 Chromebook
device type using Depthcharge.

Signed-off-by: Ricardo Cañuelo <ricardo.canuelo@collabora.com>

Add a template for the asus-C433TA-AJ0005-rammus x86_64 Chromebook
device type using Depthcharge.

Signed-off-by: Mylène Josserand <mylene.josserand@collabora.com>

Add a template for the hp-x360-12b-n4000-octopus x86_64 Chromebook device type
using Depthcharge.

Signed-off-by: Mylène Josserand <mylene.josserand@collabora.com>

api: add devices validate api

See merge request lava/lava!1510

Add a template for the asus-C523NA-A20057-coral x86_64 Chromebook device type
using Depthcharge.

Signed-off-by: Mylène Josserand <mylene.josserand@collabora.com>

This patchs adds device-type for BBE (BeagleBone Enhanced).
Tested in LAVA AGL lab.

Signed-off-by: Chase Qi <chase.qi@linaro.org>

postinst: lava-scheduler is again a lava service

See merge request lava/lava!1509

Signed-off-by: Rémi Duraffort <remi.duraffort@linaro.org>

Set environment for qemu in docker

See merge request lava/lava!1500

Add more uboot_error_message

See merge request lava/lava!1504

device-types: add avenger96

See merge request lava/lava!1506

device-types: Add stm32l562e-dk

See merge request lava/lava!1503

Reintroduce wrongly removed /etc/logrotate.d/lava-scheduler-log

See merge request lava/lava!1508

This was a tricky one. What happened:

- The version in stretch had /etc/logrotate.d/lava-scheduler-log
- The version in buster dropped that configuration file, but didn't
  remove it correctly (i.e. it was removed from the filesystem, but was
  still in the dpkg database.
- The version in bullseye adds that file again.

In a stretch -> buster -> bullseye upgrade scenario, the user gets
prompted about changes to the file even if they didn't touch the file
before upgrading to buster, because it was removed incorrectly in the
buster version. We could fix this in buster first, but then the upgraded
would only work if the user is using the very latest stable update
before upgrading.

Instead, on the buster -> bullseye upgrade, if the file is known by dpkg
-S (meaning it was installed by the stretch version) and missing from
the filesystem (i.e. it was removed by the upgrade to buster), we
reinstate the file with its original contents from stretch, so that the
bullseye upgrade will detect it hasn't been changed and just override it
with the new version.

I tested both stretch -> buster -> bullseye and buster -> bullseye
upgrades and both cases are handled correctly.

Closes Debian bug https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=988362

doc: reminder to add the site to Chosen sites

See merge request lava/lava!1505

Signed-off-by: Chase Qi <chase.qi@linaro.org>

A pitfall that may be encountered and can be easily covered by
an additional bullet point.

Signed-off-by: Gustavo Noronha Silva <gustavo.noronha@collabora.com>

This patch adds more uboot error message encoutered on kernelci labs.

When using qemu in docker, environment are not passed to it.
For the moment, only QEMU_AUDIO_DRV is usefull for qemu, so we pass only
it for limiting the length of the command line.

Add new "n1sdp" device type

See merge request lava/lava!1501

This patch adds stm32l562e-dk device-type

Signed-off-by: Arthur She <arthur.she@linaro.org>

server docker: generate pgsql config when running all services

See merge request lava/lava!1499

When running every services in the same container, generate the configuration.

Signed-off-by: Rémi Duraffort <remi.duraffort@linaro.org>

doc: improve installation guides

See merge request lava/lava!1498