I think we discussed this yesterday that we need to properly test this. A smoke test on amd64 is not adequate, we also need to ensure additionally that
- there are no regressions on arm64
install update in canonistack server and reboot it and make sure it boots.
- there are no regressions in (maas) netbooting.
I don't know how to verify nicely. I have a script you can point at shim and grub efi binaries to
boot them similar to MAAS, but there might be more regressions, I clearly can't cover them all :)
Due to the toolchain being different in the rebuilds happening in bionic/focal instead of hirsute, with much older binutils, and missing patches in gcc-8 likely. Because we know that at least shim did not work with old toolchain at all, so are a bit scared :D
I think we discussed this yesterday that we need to properly test this. A smoke test on amd64 is not adequate, we also need to ensure additionally that
- there are no regressions on arm64
install update in canonistack server and reboot it and make sure it boots.
- there are no regressions in (maas) netbooting.
I don't know how to verify nicely. I have a script you can point at shim and grub efi binaries to
boot them similar to MAAS, but there might be more regressions, I clearly can't cover them all :)
Due to the toolchain being different in the rebuilds happening in bionic/focal instead of hirsute, with much older binutils, and missing patches in gcc-8 likely. Because we know that at least shim did not work with old toolchain at all, so are a bit scared :D