Re: [PATCH v2 0/3] Fix data-race in deployment of initrd

[Uladzimir Bely <ubely@ilbers.de>]

In the email from Saturday, 18 February 2023 13:30:35 +03 user Felix 
Moessbauer wrote:
> Changes since v1:
> 
> - add API changelog entry
> - fix expression in INITRD_DEPLOY_FILE
> - do not append ${PN} to deploy-dir in image.bb
> 
> This patch fixes a data race in the deployment of the initrd, as discussed
> on the ML in "Issues creating images with custom initramfs". As a measure
> against further races on these files, we also use the sstate cache for
> deployment.
> 
> Please note, that imaging with wic + grub or systemd-boot with a custom
> initramfs is still broken. This is due to the wic logic, which reads the
> initrd in the rootfs, but not the one in the deploy dir.
> 
> Best regards,
> Felix Moessbauer
> Siemens AG
> 
> Felix Moessbauer (3):
>   add initramfs to sstate-cache
>   deploy boot files via sstate-cache
>   fix race-cond between default and custom initrd
> 
>  RECIPE-API-CHANGELOG.md        |  8 ++++++++
>  meta/classes/image.bbclass     | 36 ++++++++++++++++++++++------------
>  meta/classes/initramfs.bbclass | 18 ++++++++++++-----
>  scripts/start_vm               |  4 ++--
>  testsuite/start_vm.py          |  2 +-
>  5 files changed, 48 insertions(+), 20 deletions(-)

While testing this patchset, discovered that initrd files for some targets 
(like qemuarm-buillseye, qemuarm64-bullseye) are not deployed, so run tests 
for these qemu machines fail.

The issue can is also easily reproducible with kas.