elisa-tech / meta-elisa Goto Github PK

With sstate you can save a lot of time during build. This should be referred in the readme.

What?

I propose to add a list of available run and build options at the beginning of the How to build and run the ELISA Cluster demo section of the readme.

The items could be listed in order of increasing complexity, eg.

• fetch & run precompiled image
• build using docker container
• manual build (w/ & w/o cache)

For each item, a link to the document detailing the approach could be provided.
For each item, a very short For whom? and For what? description could be added.

Why?

I personally do not read the full build instructions in their entirety before I start executing them, but instead read and execute them step by step. If one approaches the current run & build section in this manner, one will for example learn about the build via docker only after one has basically finished the more laborious manual build. An overview of the different ways to run and build the demo at the beginning could improve the onboarding experience.

Within meta-elisa/meta-elisa-demo/recipes-elisa/pipe-demo/pipe-demo.bb
we have old comments from Jochen which need to be cleaned up.

DESCRIPTION = "Jochen layers it on ^^"
#that seems to be necessary, the license here makes no sense, I just fumbled it so it would compile ;) LICENSE = "GPLv2"

The latest build is failing while building meta-elisa in Yocto based AGL workspace.

The following patch will fix the issue:
http://cgit.openembedded.org/meta-openembedded/commit/?h=master-next&id=74b66d1911118bac53033f77ba6d3923f4809d5a

xmlsec1: Fix configure QA error caused by host lookup path
The configure script contains hardcoded lookup paths to /usr and other
paths that might interfere with the host. These are overwritten with the
staging dir locations for Poky compatibility.

Signed-off-by: Anatol Belski <[email protected]>
Signed-off-by: Khem Raj <[email protected]>

SBOM generation can be done with latest yocto version used in needlefish. The idea is to switch this on for the instrument cluster demo. This would be input for the SBOM interested within the systems WG.

The list of maintainers in in the meta-elisa readme need to be updated.
See: https://github.com/elisa-tech/meta-elisa#maintainers

We need to change Jochen to e.g. @pahmann
Maybe @AGLExport should be changed to @sudipm-mukherjee
@gabpaoloni to see, if he remains in the list.

We have extensive docker documentation which helps in setting up meta-elisa. The docker information is provided in the wg-automotive under the following link https://github.com/elisa-tech/wg-automotive/tree/master/Docker_container

Information should be added to the enf of the how-to-build-and-run-the-elisa-cluster-demo section.

Text could look like:
Consider to use the prepared docker file and included scripts to simplify the build and handling of buid environment. The necessary steps are documented in wg-automotive Docker_container readme.

Running the demo with the qemu command documented in the readme now results in the following error message:
qemu-system-x86_64: -show-cursor: invalid option

I'm running qemu 6.1, but a little research revealed that the option was deleted in version 6.0
(change log).

The log entry says to use -display sdl,show-cursor=on instead, but do we really need to show the cursor?

The used AGL Kernel is not configured for a constraint environment. The idea of this issue is to reduce kernel and image size by removing not needed modules.

KVM need to be set in the bios setting of the host PC or, if you remove KVM option from qemu call it will work, but performance is poor. The setting will differ based on the motherboard, but could be something like Intel®Virtualization Technology or AMD-V.

Update documentation in readme to remove one remaining old reference to jellyfish

In section https://github.com/elisa-tech/meta-elisa#how-to-build-and-run-the-elisa-cluster-demo
It mentions That is, in ~/whatever/AGL_TOP/AGL_RELEASE. Currently, that would be ~/whatever/AGL_TOP/jellyfish.
Here jellyfish need to be replaced by needlefish.

I sligthly modified the readme:

• I added a section to place the docker file in your desired location where you want to start of with the docker image creation. I could not find this in the readme.
• I had to run the commands for docker build and docker run with sudo rights. This may also require an update.
• Also, I simply installed docker and docker.io from repo without adding the latest version as per docker documentation which is referred to in the readme. Maybe we could say something like use these packages and either follow the docker instructions, if this does not work for your distro or just give it a try with the few packages.
• Next step can be to refer the documentation in the meta-elisa readme readme.md as a good way to start the build. Especially, the script takes away some work for you.
• Change from koi to needlefish once merged

When running the qemu demo, my mouse and keyboard are bound to the qemu window. In order to release the keyboard and mouse from the demo again, since qemu 1.5, you have to type CTRL + ALT + G.

It is worth to mention this in the readme or in the troubleshooting wiki.

qemu v7.2.0 has removed the use of "-watchdog". Instead of -watchdog i6300esb if qemu v7.2.0 is used then the argument needs to be -device i6300esb.

Due to different build environments there was an alignment to have a docker image with the important tools to prepare the build available. This should be placed in a proper repository.

Tasks are:

• Locate the old docker image
• Reproduce the build process using docker
• Document the steps so that other can follow
• Refer to the updated documentation in meta-elisa

according to the docs:
$ repo init -b needlefish -u https://gerrit.automotivelinux.org/gerrit/AGL/AGL-repo
$ repo sync
$ git clone https://github.com/elisa-tech/meta-elisa.git

$ source meta-agl/scripts/aglsetup.sh -f elisa-cluster-demo
$ bitbake elisa-cluster-demo-platform

then reports errors like attached log shows.
bitbake_elisa-cluster-demo-platform.txt

Building the cluster demo involves many packages which become part of the final image. The idea is to reduce the image size step by step towards a minimal image needed to fulfill the use case.

A comparison to the content of the medical devices working group image can make sense as well.

The DANGER warning is no longer visible in the latest execution of the demo in OpenQA testing.

See for example here:
https://openqa.qa.codethink.co.uk/tests/3288#step/safety/8

Compilation in meta-elisa gitlab CI is still successful.

The openQA tests of the demo had been failing for last few days and has
started when AGL updated their needlefish branch to "needlefish_14.0.3".
This is the openQA test from today:
https://openqa.qa.codethink.co.uk/tests/2732.

Also tested the image locally by @sudipm-mukherjee and the "DANGER" sign is not displayed.

The readme should mention that there is a wiki to collect trouble shooting issues.