The soft limits cannot go above the hard limits values, so the hard
limits should be configured too, just in case default limits values set
by the kernel or the init process are too small.
E.g. when booting with `init=/bin/sh` we have:
$ cat /proc/1/limits
Limit Soft Limit Hard Limit Units
...
Max open files 1024 4096 files
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35126>
We only have 8GB on the Brya Chromebook used for venus testing, and the
job is unstable because using all the available RAM causes OOM kills.
So reduce launch_cvd memory limit to 4GB by default, still allowing to
change the default by setting the CUTTLEFISH_MEMORY variable in jobs
that might want a different value.
This change has been inspired by CROSVM_MEMORY, but a new variable is
used because technically cuttlefish can also use qemu in some setups,
not only crosvm.
Signed-off-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35126>
Pass XDG_CACHE_HOME=/data/local/tmp to binaries that load mesa
libraries to avoid the following message on the stderr:
Failed to create //.cache for shader cache (Read-only file system)---disabling.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35335>
It is possible that `eglinfo` and `vulkaninfo` provide info on multiple
devices.
Consider only the first device, which is going to be the default one
used by other components, when checking the versions.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35335>
New mechanisms to retrieve the GLES and Vulkan driver versions have been
introduced in
- 3029fdde65 (ci/android: Switch to using eglinfo to check GLES implementation, 2025-05-28)
- 3ba9038648 (ci/android: Check Vulkan driver using vulkaninfo, 2025-05-28)
These mechanisms are more robust than the previous one but they do
change the behavior in that the version is not retrieved by an already
running process (e.g. SurfaceFlinger), but by creating new processes
that load the libraries available on the filesystem.
Because of this change of behavior the original version should be
printed **before** pushing the new libraries to the Android guest, so
that developers are able to compare the old and new versions in the logs.
Moreover, the runtime checks do not answer the original question anymore:
"what GLES/VK libraries is surfaceflinger currently using?"
but rather new question:
"what GLES/VK libraries are services going to use when they load?"
So the shell start/stop can very well performed after the version check,
accompanied by a new check on the PID of SurfaceFlinger to be sure that
it has reloaded consequently picking up the new libraries.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35335>
Not all LAVA jobs appear to require kernel modules, so only apply the
kernel-modules overlay when HWCI_KERNEL_MODULES is explicitly set.
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35129>
Change how we package kernel modules: instead of storing them in
.tar.zst archives with uncompressed .ko files inside, we now compress
each .ko file individually with ZSTD and bundle them into a plain tar
archive.
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35129>
In newer Android versions, SurfaceFlinger uses Vulkan by default,
so `dumpsys SurfaceFlinger` no longer reveals the GLES implementation.
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35232>
This variable appeared twice in debian-testing-msan, removing the first
one because it is probably being overridden by the second one.
Signed-off-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35251>
Filter out the CI_COMMIT_DESCRIPTION, CI_COMMIT_MESSAGE, and
CI_MERGE_REQUEST_DESCRIPTION variables, which were causing issues in LAVA.
Fixes: cc83b3db5f ("ci: Forward all environment variables to DUTs and crosvm")
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35196>
Now that we're no longer using multiple artifacts, we can drop the
LAVA_ prefix from the S3_ARTIFACT_NAME variable name for simplicity.
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34980>
LAVA was the last user of this job. Now that the lava-trigger container
includes all the necessary files, the python-artifacts job is no longer
needed and can be removed.
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34980>
The alpine/x86_64_lava-trigger image was set up in a way that avoids
conflicts with MESA_IMAGE_PATH and FDO_DISTRIBUTION_TAG.
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34980>
Compared to the existing Debian-based x86_64_pyutils container, this
Alpine-based variant reduces the image size by approximately 83%.
Include all the necessary python artifacts, including lava_job_submitter
in the container to avoid having to download them at the start of each
test job.
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34980>
init-stage2.sh and setup-test-env.sh are already downloaded on the DUT as
part of the mesa-build overlay, which downloads the Mesa artifacts from
S3.
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35051>
Instead of the current allowlist in export-gitlab-job-env-for-dut.sh,
filter out unwanted environment variables and forward the rest to
bare-metal and LAVA DUTs, as well as crosvm.
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35051>
With the rest of the Qualcomm devices moving to LAVA, we can remove the
original (!) bare-metal infrastructure, leaving only the Igalia RPi
devices still using bare-metal. When those are converted to b2c, we can
remove the rest of bare-metal.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35148>
Firmware packages continue to grow in size, so stop installing them in
the test-base image.
The necessary firmware is now collected and uploaded per vendor in an
external repository.
LAVA devices can opt into optional firmware by specifying the name of the
archive via LAVA_FIRMWARE.
For bare-metal, Qualcomm firmware required for DUTs in the Google lab is
included in the baremetal image.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/13051
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34861>
We can have jobs with a lower job timeout values, given by
CI_JOB_TIMEOUT environment variable, such as the pytest ones.
The previously hardcoded burst length of 1000 messages at a simulated
rate of 1 msg/sec caused tests to exceed these timeouts and fail
unexpectedly on specific job timeouts.
Reported-by: Eric Engestrom <eric@igalia.com>
Signed-off-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34907>
