Kai Lüke

The name_to_handle_at() value (even with FID) changed directly during the systemd-sysext run for the parent directory `/etc/extensions/` probably because the directory got a new timestamp. I didn't look into this...

> hmm, the fid should _not_ change during an mtime fix. Good to check this again and of course I can't reproduce this any more… Maybe due to a reboot....

The JSON library was nice to work with and the result is much better now. That also allowed to leave out the weaker identifiers when we have the verity hash....

The cert comes from --root= for the extensions in --root=. For example, the initrd sets --root=/sysroot/ and then it will use certs from /usr or /etc to load extensions from...

The tumbleweed CI failure looks strange and I think it only came up after I pushed a rebase: ``` [ 152.334524][ T5597] request_module fs-erofs succeeded, but still no fs? [...

> [ 152.334524][ T5597] request_module fs-erofs succeeded, but still no fs? This showed up again and it looks like tumbleweed has broken erofs support? Or the way the erofs image...

Small in terms of RAM usage or disk usage? If we diverge from the standard 1 GB + 1 GB disk partitions this is more involved. One can already use...

The PXE squashfs is a separate initrd but we could maybe support to load it from disk though a cmdline parameter that we evaluate in the initrd. This would save...

With splitting out some base image content into a Flatcar "extras" extension we can reduce the size of the squashfs that the PXE boot uses. For A/B updates I think...

Thank you for the detailed report! Right, as always I think IPv6 wasn't really considered when this code was written. Adding support for dual stack now would be really nice,...