Add zephyr build and test jobs for the below qemu machines:
* qemu-x86
* qemu-cortex-m3

Closes meta-zephyr#35



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

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

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

Signed-off-by: Pavel Zhukov <pavel.zhukov@huawei.com>

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

Reorganize SQUAD groups and projects to make it easier to find out what
is tested by looking at the group/project overview.

New group/project org.

```
squad/
├── merge-request
│   ├── meta-zephyr
│   └── oniro
├── release
├── scheduled-daily
│   ├── kselftest
│   ├── ltp
│   ├── perf
│   └── zephyr
└── scheduled-weekly
    └── gcc
```

This closes: #731

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

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

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

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

Signed-off-by: Andrei Gherzan <andrei.gherzan@huawei.com>
Signed-off-by: Luca Seritan <luca.seritan@huawei.com>

Chase Qi authored 2 years ago and Stevan Radaković committed 2 years ago

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

Using build job name to download bundles doesn't work. We need to introduce
new var and make it mandatory for lava jobs containing the `sysota`
in job name.

Signed-off-by: Stevan Radaković <stevan.radakovic@linaro.org>

There was some cases that the same variable appears more then once in
the line. Only first appearence in the line was replaced.

Signed-off-by: artem.kondratiuk <artem.kondratiuk@huawei.com>

Artem Kondratiuk authored 2 years ago and Artem Kondratiuk committed 2 years ago

Signed-off-by: artem.kondratiuk <artem.kondratiuk@huawei.com>

    manifests/default.xml:      Add new project meta-seco-rockchip
    local.conf.sample:          Add seco-px30-d23 to MACHINE
    bblayers.conf.sample:       Add meta-seco-rockchip layer to BBLAYERS
    conf-notes.txt              Add seco-px30-d23 to supported machines
    machines-and-flavours.yaml  Add linux-seco-px30-d23 and linux-seco-px30-d23-extra jobs
    linux-flavour.rst:          Add seco-px30-d23 to supported machines
    machines-and-flavours.rst   Add linux-seco-px30-d23 job
    boards/index.rst            Add seco-px30-d23 to supported boards

    Add the seco-px30-d23.rst file representing the new Seco's
    JUNO board (SBC-D23) in the oniro project.

Signed-off-by: Natalia Kovalenko <waykovalenko@gmail.com>

Use squad build ID to query test pass percentage.

The lava-badge job is only needed by scheduled pipeline as calculating
pass rate of full testing makes more sense. Enabling the job here for
MR in draft status is mainly for debugging purpose.

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

This reverts commit a180a4b8.

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

Artem Kondratiuk authored 2 years ago and Pawel Stankiewicz committed 2 years ago

Signed-off-by: artem.kondratiuk <artem.kondratiuk@huawei.com>

There is an incompatibility issue between the sphinx version in Ubuntu
focal and the docutils version pulled in by sphinx-tabs. We need to make
sure that pip doesn't pull in a version of docutils that would break
sphinx:

https://github.com/sphinx-doc/sphinx/issues/9841



Signed-off-by: Andrei Gherzan <andrei.gherzan@huawei.com>

The Ubuntu version we downgraded to doesn't have support for
sphinx-tabs. Switch that to pip until we can upgrade to 22.04.

Signed-off-by: Andrei Gherzan <andrei.gherzan@huawei.com>

The currently used version, impish, is not updatable anymore since July
the 14th and we don't have a newer version in our repository. Downgrade
this to focal to fix the container image builds.

Fixes: #724


Signed-off-by: Andrei Gherzan <andrei.gherzan@huawei.com>

To help Oniro with
eclipsefdn/helpdesk#1565, EF created
a bot for us that is passing the ECA checks. Use it to unblock the docs
pipelines.

Fixes: #723


Signed-off-by: Andrei Gherzan <andrei.gherzan@huawei.com>

Chase Qi authored 2 years ago and Stefan Schmidt committed 2 years ago

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

Chase Qi authored 2 years ago and Stefan Schmidt committed 2 years ago

Check the current version with `bitbake -s` and the last tested version
via squad's api. When version is changed, continue the following tests,
otherwise, skip the tests.

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

Chase Qi authored 2 years ago and Stefan Schmidt committed 2 years ago

gcc test suite has been integrated in oe-selftest framework. It allows
to run gcc test against qemu vm and within the build env using qemu
linux-user mode.

These tests can be run in gitlab CI with qemu friendly runners. This
commit add three oe-selftest jobs to run the tests using qemu
linux-user, qemu x86-64 vm and qemu arm64 vm.

Test log and sum files are collected as job artifacts.

When all test jobs finish, report job will be triggered to parse the
sum files to generate results in the format that supported by squad. The
report job will upload results to squad for reporting.

Closes issues #8 and #9.

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

Pavel Zhukov authored 2 years ago and Andrei Gherzan committed 2 years ago

We need to analyse ABI and keep ABI data for different images/arches
thus artifact name should be job specific

Signed-off-by: Pavel Zhukov <pavel.zhukov@huawei.com>

Pavel Zhukov authored 2 years ago and Andrei Gherzan committed 2 years ago

It was broken because of BITBAKE_TMPDIR variable definition removal as
part of the cleanup. Define BITBAKE_TOPDIR variable instead which serves
needs of ABI checker better.

Signed-off-by: Pavel Zhukov <pavel.zhukov@huawei.com>

Zygmunt Krynicki authored 2 years ago and Stefan Schmidt committed 2 years ago

Some jobs care about images and licenses, those look at the
$TMPDIR/deploy, taking some effort to find the correct tmpdir while
hard-coding the deploy dir. Take a shortcut and query for the right
variable (DEPLOY_DIR).

This allows non-trivial layouts with DEPLOY_DIR and TMPDIR placed on
distinct filesystems, for efficiency.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

Pavel Zhukov authored 2 years ago and Stefan Schmidt committed 2 years ago

ABI checker implementation is meta-binaryaudit. Adding it for CI only
to not extend support surface of the project.
Added to three flavours for now. It can be easily extended to all jobs
if needed in the future

Signed-off-by: Pavel Zhukov <pavel.zhukov@huawei.com>

Part of gcc oeself-test is client/server based. It uses qemu vm for
native tests via ssh. The vm is launched by runqemu command which
requires the following changes in the docker image.

* passwordless sudo permission
* package iptables pre-installed

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

The reference is provided only when repo is used with runner persistent
storage. This is not a major change, just a clean-up of the existing
logic.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

Add missing INHERITs to Linux jobs. Now all
Linux images in CI run "create-spdx.bbclass" correctly.

Signed-off-by: Davide Gardenal <davide.gardenal@huawei.com>

Signed-off-by: Davide Gardenal <davide.gardenal@huawei.com>

The parent of all zephyr jobs is now unified to enable easier
extensibility later on.
Update documentation.

Signed-off-by: Davide Gardenal <davide.gardenal@huawei.com>

Enabling debug mode is not the ideal solution for running
tests because it's responsible for enlarging the rootfs size and causing
tests to fail. More detail reported in #673.

This reverts commit 3ddebb37.

Signed-off-by: Davide Gardenal <davide.gardenal@huawei.com>

Use the Yocto class to generate SBOM. Activate on Linux only,
there is a pending issue on Zephyr [1]

[1] https://lists.openembedded.org/g/openembedded-core/message/167450



Signed-off-by: Marta Rybczynska <marta.rybczynska@huawei.com>

LAVA needs a writable directory to deploy test cases, and many test
cases only work with root user.

The debug mode can be disabled by overwriting the
CI_ONIRO_BB_LOCAL_CONF_plus_equals_INHERIT variable at build job level.

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

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

When the callback not dispatchered correctly, the record can be helpful
for debugging in squad.

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

Given test scope for new MRs is not defined yet and we only have one
rpi4, b68 and c61 in the warsaw lab which are occupied mostly by daily
scheduled pipeline, starting with boot and smoke tests on qemu devices
should be a good starting point for per MR testing.

Jobs that extend `.lava-test` will be added to MR's pipeline and
schedueld pipeline.

`.lava-test-scheduled` extends `.lava-test` with new scheduled pipeline
only rule. Long run jobs like ltp should extends the job so that they
wouldn't prevent MRs from merging.

Solves #486.

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