after distribution upgrade from 15.0 to 15.1 - which itself went ok without any error messages - the system fails to boot.
The computer uses EFI, uses LVM and an nvme-SSD for the root partition.
The boot process hangs after displaying the following information:
[OK] Started udev Coldplug all Devices.
Starting dracut initqueue hook...
Starting Show Plymouth Boot Scree...
[OK] Reached target System Initialization.
Sometimes the last line (“System Initialization”) is missing, if that I the case sometimes “dracut initqueue” and “Show Plymouth” change their positions.
There is no timeout, the system does not go into emergency mode.
The LEAP 15.1 rescue system boots successful, and in that case the complete LVM configuration is detected w/o issues; I can mount the different partitions. Doing so, I recreated the initramfs, however w/o success.
Then, I remembered that I had boot issues after upgrading from 42.3 to 15.0 about one year ago, see this thread. The solution was to remove the modules “nvme_core” and “nvme” from the initramfs, without really knowing why this had helped. So I added these modules again, recreated initramfs, however no change - keeps hanging.
So any Ideas what goes wrong here and what I can do ?
There may be more messages to be seen if you disable Plymouth, which could itself be part of the problem. At the Grub menu on the default entry, strike the E key, goto the end of the (usually wrapped) linu line, and append space and plymouth.enable=0 before proceeding as directed at the bottom of that screen.
If the above fails to be helpful, reboot using the failsafe/rescue Grub selection, login, and capture the log from the previous boot:
sudo journalctl -b -1 > sometextfile.txt
Then upload sometextfile.txt to http://susepaste.org/ or http://pastebin.com and paste here the provided URL. Upload also /var/log/Xorg.0.log in same manner. The susepaste -n the_wumpus -e 10080 <filename> command can do this for you, but it usually lies, indicating failure when in fact the upload succeeded. Then it’s necessary to fetch the URL from susepaste.org, which if you do it right away will top the list of recent uploads.
Disabling Plymouth resulting in getting a bit further in booting, but that was it. Looking at the previous boot log was however extremely helpful as it revealed the following lines:
897:Jan 12 16:50:08 hal2000 kernel: mc: Failed to load firmware "amdgpu/polaris11_k_mc.bin"
898:Jan 12 16:50:08 hal2000 kernel: [drm:gmc_v8_0_sw_init [amdgpu]] *ERROR* Failed to load mc firmware!
899:Jan 12 16:50:08 hal2000 kernel: [drm:amdgpu_device_init [amdgpu]] *ERROR* sw_init of IP block <gmc_v8_0> failed -2
900:Jan 12 16:50:08 hal2000 kernel: amdgpu 0000:0a:00.0: amdgpu_device_ip_init failed
901:Jan 12 16:50:08 hal2000 kernel: amdgpu 0000:0a:00.0: Fatal error during GPU init
Booting with parameter “nomodeset” (and teaching myself how to correctly specify runlevels upon boot… ) resulted in a successful boot, but as supposed non-operational graphics card. After some more digging, I found that dracut looks into “/lib/firmware/4.4.74-18.20-default/amdgpu/” for the firmware, but there the above mentioned file “amdgpu/polaris11_k_mc.bin” was missing there and thus not copied into initramfs. After copying that file from “/lib/firmware/amdgpu” and recreating initramfs, everything worked :).
Well I don’t know if this is some bug in the OpenSUSE packages ?
?