https://bugzilla.redhat.com/show_bug.cgi?id=1325023
--- Comment #20 from Laszlo Ersek lersek@redhat.com --- (In reply to Gerd Hoffmann from comment #19)
(In reply to Laszlo Ersek from comment #18)
The spec file in "edk2-20160412gitba83f7e-0.el7.src.rpm" looks good to me.
However, I'm noticing two things I should have noticed earlier (sorry!):
- We probably should not ship the unified "/usr/share/edk2/ovmf/OVMF.fd"
file, only the split files.
Hmm, why? Can be useful for throwaway virtual machines, such as booting a live iso with efi. Without that you have to copy the varstore to some temp file, boot, then delete the tmpfile even if you don't need the efi variable persistance.
The setup you describe is valid for throwaway VMs; it parallels the case when QEMU is started without a writeable hard disk.
It's just that I've seen reports where a user uses -bios (or an equivalent "hand-crafted" domain XML) for VMs that are meant to be persistent. The resultant confusion is not good.
Personally, I never use the unified file, not even for throwaway VMs. I do the copying (in scripts) the way you describe. It's not a big burden, and the originally unneeded persistence actually turns out welcome in some cases.
I also found it inconsistent that the build without the Secure Boot feature provides a unified binary, while the build with the SB feature doesn't. If the argument is throwaway VMs, then I guess it does make sense (you can't enable SB without enrolling keys and rebooting the VM, hence SB only applies to non-throwaway VMs).
Anyway, up to you.
- The aarch64 build seems to have disappeared, after 2016-04-08, from https://kraxel.fedorapeople.org/review/edk2. Is this intentional?
The old ones are still there. But didn't do a fresh aarch64 build, so there are no updated aarch64 rpms. /me goes kick a build ...
Looks good, thanks.