automotive-sig February 2025

automotive-sig@lists.centos.org

3 participants
4 discussions

Custom kernel build fails: rsync: Argument list too long
by Sebastian Hetze 28 Feb '25

28 Feb '25

Hi *, when trying to build the centos-stream-9 main-automotive kernel for the x86_64 architecture as described in the automotive sig documentation <https://sigs.centos.org/automotive/building/building_an_os_image_that%20use…>, it fails. The failing command appears to be /usr/bin/make -s 'HOSTCFLAGS=-O2 -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -fstack-protector-strong -specs=/usr/lib/rpm/redhat/redhat-annobin-cc1 -m64 -march=x86-64-v2 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection' 'HOSTLDFLAGS=-Wl,-z,relro -Wl,--as-needed -Wl,-z,now -specs=/usr/lib/rpm/redhat/redhat-hardened-ld -specs=/usr/lib/rpm/redhat/redhat-annobin-cc1 ' -j12 ARCH=x86_64 V=1 'TARGETS=bpf cgroup mm livepatch net net/forwarding net/mptcp netfilter tc-testing memfd drivers/net/bonding iommu cachestat drivers/net' SKIP_TARGETS= FORCE_TARGETS=1 INSTALL_PATH=/home/she/rpmbuild/BUILDROOT/kernel-automotive-5.14.0-570.518_at.el9iv.x86_64/usr/libexec/kselftests VMLINUX_H= DEFAULT_INSTALL_HDR_PATH=0 install with the error make[3]: Entering directory '/home/she/Workspace/centos-stream-9/redhat/rpm/BUILD/kernel-5.14.0-570.518_at.el9iv/linux-5.14.0-570.518_at.el9iv.x86_64/tools/testing/selftests/bpf' Auto-detecting system features: ... llvm: [ OFF ] make[3]: rsync: Argument list too long make[3]: *** [Makefile:768: test_progs-no_alu32-extras] Error 127 make[3]: Leaving directory '/home/she/Workspace/centos-stream-9/redhat/rpm/BUILD/kernel-5.14.0-570.518_at.el9iv/linux-5.14.0-570.518_at.el9iv.x86_64/tools/testing/selftests/bpf' make[2]: *** [Makefile:233: install] Error 2 This is the unmodified main-automotive branch of the current centos-stream-9 kernel on a current centos-stream-9 system. Setting ulimit -S -s unlimited for ARG_MAX 6291456 does not make any difference. It looks like the actual problem is not with the rsync command line. Any hint what is going wrong and how to proceed? Beste Grüße Sebastian Hetze -- Principal Software Engineer Red Hat GmbH. Niederlassung Berlin Leipziger Platz 16, 10117 Berlin Mobil: +49 173 8914205 E-Mail: she(a)redhat.com Registered seat: Werner von Siemens Ring 14, D-85630 Grasbrunn, Germany Commercial register: Amtsgericht Muenchen/Munich, HRB 153243,Managing Directors: Ryan Barnhart, Charles Cachera, Michael O'Neill, Amy Ross

2 1

Re: Request for new git repos
by Pierre-Yves Chibon 27 Feb '25

27 Feb '25

On Wed, Feb 26, 2025 at 03:48:58PM +0100, Alexander Larsson wrote: > I'd lke to request two new gitlab repos: > [1]https://gitlab.com/CentOS/automotive/src/ukiboot > [2]https://gitlab.com/CentOS/automotive/rpms/ukiboot Created and fixed the sig's list address :) Pierre

1 0

Re: autosd install on x86_64 hardware
by Alexander Larsson 19 Feb '25

19 Feb '25

The current automotive images are not compatible with bootc. We hope to eventually change this, but that needs some serious engineering work. The approach for image updates in automotive is based on ostree. automotive-image-builder builds ostree commits, as well as an initial physical image containing that commit. Once the image with the initial commit is deployed, ostree is used to do the update. This happens either by pulling from an ostree repository or by side-loading an offline update file (an ostree static delta) using whatever mechanism you have available. Here are some some docs explaining how this works: https://sigs.centos.org/automotive/building/updating_ostree/#updating-ostre… https://sigs.centos.org/automotive/building/creating_static_deltas/ Also, in general the approach we the automotive images use is to produce physical images that are flashed, we don't use installers like anaconda. Installers are not really a good fit for embedded style hardware. -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Alexander Larsson Red Hat, Inc alexl(a)redhat.com alexander.larsson(a)gmail.com

2 1

autosd install on x86_64 hardware
by shetze＠redhat.com 12 Feb '25

12 Feb '25

Hi *, we are currently trying to install autosd on hardware for our railway project. x86_64 hardware is used in Deutsche Bahn rolling stock. For the Automated Train project, we want to install the CentOS autosd on a duagon G28M processor module. We intend to install a bootc container via Anaconda from an ISO image. On the running system, a rebase to a registry-based image can then be performed for (incremental) updates. Does this make sense, or is there a better approach? As far as I can see, containers or OCI images created by auto-image-builder are not recognized as bootable by bootc-image-builder. With rpm-ostree compose I can create a correct bootc container image from the auto-image-builder's ostree commit. I can perform a rebase on this image, but it doesn't boot because the commitmeta signature for ostree-prepare-root cannot be loaded. I can't make an installable ISO from the bootc container image either, because the osbuild doesn't support the autosd-9 distro. My impression is also that the containers created by the auto-image-builder contain only a single blob, so they are not suitable for incremental updates at all. So we won't reach our goal on the way decribed above. Do you have any hints, feedback or tips?

1 0

2025

2024

2023

2022

2021

automotive-sig February 2025