| 07:43 | suihkulokki> | Artox: can you edit the EFI variables from EFI menu= |
| 12:04 | Artox> | suihkulokki: yes I can. I went into add boot entry, and called it "Debian (eMMC)" |
| 12:04 | Artox> | then back on main page -> reset |
| 12:04 | Artox> | then it was in the boot manager menu |
| 12:04 | Artox> | and is bootable |
| 13:50 | suihkulokki> | Artox: I really wonder what difference our setup then has |
| 13:50 | suihkulokki> | there is the emmc/vs sata install difference at least |
| 13:54 | Artox> | suihkulokki: true, but the next time I try anything, I'll target sata (want to keep the emmc installation in working condition) |
| 13:55 | Artox> | btw the installer has been hanging before making grub.cfg, and before installing files in the EFI partition |
| 13:55 | Artox> | I fixed it by booting the installer in rescue mode, and running grub-install and update-grub by hand |
| 13:55 | Artox> | so now I do have a working debian system on there |
| 13:59 | suihkulokki> | looking back at the patch that "fixed" efi variable writing |
| 14:00 | suihkulokki> | I think it shouldn't depend on kernel knowing that right clocks |
| 14:01 | suihkulokki> | If I understand correctly, kernel jumps into some EFI code that handles writing the variables to SPI |
| 14:02 | suihkulokki> | it would seem more natural for EFI to not trust kernel to have set up clocks correctly |
| 14:02 | suihkulokki> | but perhaps I misunderstand the contract between firmware and kernel |
| 14:02 | suihkulokki> | Artox: which kernel version you have? |
| 14:04 | Artox> | suihkulokki: 4.15.0-2-arm64 |
| 14:04 | Artox> | I can't remember what version the mini.iso came witjh |
| 14:06 | suihkulokki> | if it's from this month, the version should be same |
| 14:07 | Artox> | okay |
| 14:07 | Artox> | yes it is from this month |
| 14:07 | Artox> | downloaded april 6th |
| 14:15 | suihkulokki> | same version |
| 15:10 | jnettlet[m]> | suihkulokki: do you have a link to the patch? |
| 15:28 | suihkulokki> | jnettlet[m]: https://github.com/MarvellEmbeddedProcessors/edk2-open-platform/commit/b316449f24fdac9fc14c87d23f4680dfd789c7b8 |
| 15:33 | jnettlet[m] | 15:33 * jnettlet[m] just shakes head. The entire idea of device-tree is that there is one device-tree that is shared throughout the stack |
| 15:34 | Ke> | I think that is sort of the point |