Thanks Simon, Yes, ultimately I want to boot a UKI containing systemd-stub, but I am also interested in running systemd-boot itself as an EFI application.
The chain I have in mind on arm64 is roughly: U-Boot -> U-Boot EFI implementation -> systemd-boot -> UKI / systemd-stub -> Linux My main question is whether U-Boot’s EFI implementation provides enough of the UEFI services/protocols expected by systemd-boot and systemd-stub, without having to use something like an EDK II payload. The EFI call logging you mention sounds particularly useful for this. Do you have a pointer to those patches? I would be interested in using it to boot systemd-boot/UKI and compare the EFI calls actually exercised against what U-Boot currently implements. Regards, Umut On Sat, Oct 3, 2026 at 4:30 PM Simon Glass <[email protected]> wrote: > Hi Umut, > > On Thu, 10 Sept 2026 at 13:14, Umut Tezduyar Lindskog <[email protected]> > wrote: > > > > Hello, > > > > My understanding is that u-boot has its own limited UEFI provider > implementation unlike some other boot loaders bundling open source UEFI > payloads. > > > > Has anyone tried to use u-boot's UEFI provider implementation on arm64 > with EFI applications systemd-boot and systemd-stub? > > > > Using AI, I mapped all the API calls done by these programs to UEFI > service / protocol. Most of the text input output functionality and > graphics functionality is irrelevant for my purposes. > > > > ASCII text mapping table below: > > I think you are referring to UKI (which includes the stub)? I have not > tried it, but there is a good chance that it works. I did send some > patches for an EFI log which keeps track of the calls which are made. > > Regards, > Simon >
