[SOLVED] Finishing a slackware-current install on mmcblk device with UEFI

rudymartinx · 04-07-2020, 05:40 PM

my mistake was to use the huge.s from install disk on the /boot dir thats why it didn't work the first time I tried.

0XBF · 04-07-2020, 05:51 PM

I was going to suggest an initrd with mmc_core and mmc_block next but it seems you have figured things out. I'm not sure on what to suggest for getting your bootloader to work now since I usually use grub. Perhaps you can run eliloconfig now that the correct modules are loaded?

rudymartinx · 04-07-2020, 08:08 PM

Quote:

Originally Posted by 0XBF

I was going to suggest an initrd with mmc_core and mmc_block next but it seems you have figured things out. I'm not sure on what to suggest for getting your bootloader to work now since I usually use grub. Perhaps you can run eliloconfig now that the correct modules are loaded?

I couldn't. I'm reading/trying to figure out how to install the EFI entry.

So far I have learned efibootmgr is the command used by eliloconfig to create a new EFI but each time I try to run

efibootmgr -C -d /dev/mmcblk0 -p 1 -l "\\EFI\\BOOT\\bootx64.efi" -L Slackware -v

I get

Could not prepare Boot variable: Success
error trace:
linux.c:538 device_get(): parsing emmc failed: Success
creator.c:194 efi_va_generate_file_device_path_from_esp(): could not get ESP disk info: Success
creator.c:349 efi_generate_file_device_path_from_esp(): could not generate File DP from ESP: Success
efi.c:390 make_linux_load_option(): efi_generate_file_device_path_from_esp() = -1 (failed): Success
efibootmgr.c:295 make_var(): make_linux_load_option() failed: Success
efibootmgr.c:344 make_var(): Could not set variable: Success

I was just trying to read the source code. I hve recompiled it but binary seems the same included with slackware (same size at least)

Quote:

root@fvsm:~# ls /boot/efi/EFI/BOOT
bootx64.efi* elilo.conf* elilo.conf~* huge.s* initrd.gz* initrd.gz.0* vmlinuz* vmlinuz-huge*
root@fvsm:~# ls /boot/efi/EFI/BOOT -la
total 46256
drwxr-xr-x 2 root root 4096 Apr 7 23:56 ./
drwxr-xr-x 5 root root 4096 Apr 8 00:32 ../
-rwxr-xr-x 1 root root 238531 Apr 7 23:33 bootx64.efi*
-rwxr-xr-x 1 root root 165 Apr 7 23:33 elilo.conf*
-rwxr-xr-x 1 root root 177 Apr 7 23:33 elilo.conf~*
-rwxr-xr-x 1 root root 10033536 Apr 7 23:33 huge.s*
-rwxr-xr-x 1 root root 10263551 Apr 7 23:33 initrd.gz*
-rwxr-xr-x 1 root root 10240015 Apr 7 23:33 initrd.gz.0*
-rwxr-xr-x 1 root root 6527360 Apr 7 23:33 vmlinuz*
-rwxr-xr-x 1 root root 10033536 Apr 7 23:33 vmlinuz-huge*

I know there are files lying around, I can take care of that later. So far Im trying to create a new EFI entry and mark it as first in the boot order.

seems like both grub and elilo are using this program

of course I have mounted efivars as suggested at https://docs.slackware.com/howtos:sl...ble_filesystem

clueless about how to proceed

edit: this "Success" error reminds me of Windows 98 errors messages

rudymartinx · 04-07-2020, 08:24 PM

nevermind, leaving an \EFI\BOOT\bootx64.efi on efi partition is enough for this to boot when there are no other entries on EFI (which I did delete the others but left the efi shell, just in case something goes wrong)

thanks all for the help.

0XBF · 04-07-2020, 09:07 PM

Good to hear you got things working, I think you did most of the work here.

Keep in mind you will have to remake that initrd if you run any updates on the kernel.

Andypoo · 05-26-2020, 12:07 PM

I just got bitten by this too on a new installation on an Apollo Lake eMMC machine (which I found strange, since I had installed on these machines before).

It looks like efivar is faulty in slackware64-current since:

Code:

Thu Nov 21 22:19:23 UTC 2019
a/efibootmgr-20191011_e8ce9fe-x86_64-1.txz:  Upgraded.
a/efivar-20191015_ca85d51-x86_64-1.txz:  Upgraded.

I believe this version contains a regression in this commit:
https://github.com/rhboot/efivar/com...d000b0296b6544

The fix was committed around a month ago:
https://github.com/rhboot/efivar/com...ec0a82d1ccdd33

I haven't tested yet, but seems to make sense at a glance.

Andypoo · 05-26-2020, 12:17 PM

I built a minimal rebuild of efivar-20191015_ca85d51-x86_64 with just that patch and it indeed fixed the problem.

So I think a new efivar (or if there's something else there preventing it, at least patched efivar) would be great! eMMC users rejoice

Should I report this elsewhere, or is someone monitoring here that can ensure it gets in?

Thanks,
Andypoo.

Alien Bob · 05-26-2020, 02:06 PM

Quote:

Originally Posted by Andypoo

I built a minimal rebuild of efivar-20191015_ca85d51-x86_64 with just that patch and it indeed fixed the problem.

So I think a new efivar (or if there's something else there preventing it, at least patched efivar) would be great! eMMC users rejoice

Should I report this elsewhere, or is someone monitoring here that can ensure it gets in?

Thanks,
Andypoo.

You should report it in the sticky thread in the Slackware forum: https://www.linuxquestions.org/quest...-a-4175620463/