After a fresh boot it fails, but it works after a systemctl --system daemon-reexec $ systemd-analyze Startup finished in 307ms (kernel) + 1.378s (initrd) + 7.028s (userspace) = 8.714s $ sudo systemctl --system daemon-reexec $ systemd-analyze Startup finished in 4.646s (firmware) + 101ms (loader) + 307ms (kernel) + 1.378s (initrd) + 7.028s (userspace) = 13.462s Reverting commit c3a170f3d3358135a39ac6eafe66f18aef0bd67d seems to fix this, so I think the efi timestamp loading should be moved back to where it was, as earlier seems to not work for me. Can anyone else reproduce this bug?
Created attachment 79115 [details] [review] Patch fixing efi timestamp bug
Created attachment 79271 [details] [review] Patch fixing efi timestamp bug This is the same as the previous patch submitted, except it correctly uses spaces instead of tabs
The source of this problem seems to be that c3a170f3 moved the call to efi_get_boot_timestamps to before /sys is mounted in main.c, and so it cannot always read the firmware and loader timestamps
Created attachment 79273 [details] [review] Patch recording efi timestamps after /sys is mounted This patch has a better description than the previous ones
The root cause is calling efi_get_boot_timestamps (which reads from /sys/firmware/efi/efivars) before /sys/firmware/efi/efivars is mounted by systemd, so the read will fail on a fresh boot.
Patch applied to latest git
Thanks!
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.