A post was merged into an existing topic: After updating kernel to 5.18.3, system fails to start
[Stable Update] 2022-06-12 - Kernels, Systemd 251, GNOME 42.2, KDE Gear 20.04.2, NVIDIA, Mesa, Pulseaudio, Pipewire, LibreOffice
I added the following (temporary) solution based on @student975’s link:
Akonadi (KMail / Kontact / KOrganizer…) fails to start
- The issue is being investigated upstream: 454948 – Akonadi server will not start on fresh install; yields "unknown error"
- If you read this before updating, it seems recommended to backup
ib_logfile0before updating, and restore it after rebooting:
cp ~/.local/share/akonadi/db_data/ib_logfile0 ~/.local/share/akonadi/db_data/ib_logfile0.bak pamac update reboot 0 mv ~/.local/share/akonadi/db_data/ib_logfile0.bak ~/.local/share/akonadi/db_data/ib_logfile0 akonadi start
- If you read this after updating and rebooting, you need to create a new
Though i personally haven’t updated yet. So if anyone sees anything wrong with it, please don’t hesitate to fix it above.
I’m having a bit of difficulty getting amd_pstate to work (again). On the previous stable update with kernel 5.17.x I got it working, but noticed on a previous testing update that a new
/etc/mkinitcpio.conf change was needed, otherwise it switches back to acpi-cpufreq.
I’ve updated to today’s stable, running kernel 5.18.3, checked cpupower frequency-info, and it says acpi-cpufreq. I tried following the instructions in the testing thread for adding amd_pstate to the MODULES section of mkinitcpio.conf, but when running the “
sudo mkinitcpio -P linux” command I get this:
==> ERROR: module not found: amd_pstate for each kernel. I undid the suggestions in the testing thread as a precaution.
I tried the instructions for modifying mkinitcpio.conf before updating to this stable release as a test, and had the same error results.
My processor is a Zen 2, Ryzen 7 3800X, /etc/default/grub has the amd_pstate.enable=1 and amd_pstate.shared_mem=1.
Hmmm, should I hold off updating if I know I’m using/need the r8169 driver?
Network: Device-1: Realtek RTL8125 2.5GbE vendor: Micro-Star MSI driver: r8169 v: kernel pcie: gen: 2 speed: 5 GT/s lanes: 1 port: f000 bus-ID: 27:00.0 chip-ID: 10ec:8125 class-ID: 0200 IF: enp39s0 state: up speed: 1000 Mbps duplex: full mac: <filter>
Or perhaps this was an issue with a kernel above 5.15 LTS?
I think I found the answer to my own question… the only
r8168 package beyond 5.15 is 5.17… so I suspect the issue was on kernel 5.16 or 5.18.
I guess there were two (or more) push to mirrors. After the first one multiple problems have appeared. After the second - hour or two late - all problems have gone (except for akonadi start, and it is discussed above).
Again - it’s as I see this update nuances, no more.
Everything is working great, as usual, except Blender won’t start anymore. Have performed a downgrade but the issue stayed.
ERROR (gpu.shader): gpu_shader_2D_widget_base FragShader: | 81 | layout(depth_any) out float gl_FragDepth; | ^ | Error: unrecognized layout identifier `depth_any' Writing: /tmp/blender.crash.txt zsh: segmentation fault (core dumped) blender
# backtrace /usr/bin/blender(BLI_system_backtrace+0x38) [0x55e106095eb8] /usr/bin/blender(+0xff0ef6) [0x55e1053c3ef6] /usr/lib/libc.so.6(+0x3e8e0) [0x7fb34043e8e0] /usr/bin/blender(GPU_shader_get_builtin_uniform+0) [0x55e108052750] /usr/bin/blender(GPU_matrix_bind+0x2b) [0x55e10805180b] /usr/bin/blender(UI_widgetbase_draw_cache_flush+0xe6) [0x55e105d32e36] /usr/bin/blender(+0x1962d67) [0x55e105d35d67] /usr/bin/blender(ui_draw_but+0xdbb) [0x55e105d3ab6b] /usr/bin/blender(UI_block_draw+0x387) [0x55e106b311f7] /usr/bin/blender(UI_blocklist_draw+0x34) [0x55e106b31294] /usr/bin/blender(ED_region_header_draw+0xe9) [0x55e1059e26b9] /usr/bin/blender(ED_region_do_draw+0xb09) [0x55e1059d5829] /usr/bin/blender(wm_draw_update+0xb7c) [0x55e1055540dc] /usr/bin/blender(WM_main+0x34) [0x55e10555cd74] /usr/bin/blender(main+0x361) [0x55e105389d91] /usr/lib/libc.so.6(+0x29290) [0x7fb340429290] /usr/lib/libc.so.6(__libc_start_main+0x8a) [0x7fb34042934a] /usr/bin/blender(_start+0x25) [0x55e1053b6665] # Python backtrace
Great update, minor issue with open razer dkms after updating to kernel 5.18.3. Easily resolved by booting into the new kernel, removing the dkms driver, reinstalling and rebooting.
I’ve 50% of success: On my desktop computer the solution proposed has failed, however it has fixed on my laptop.
I’m sure that I could had messed something, so I think that I’m going to reinstall & config Kmail again…
Thank you very much.
A custom Pacman hook like this:
[Trigger] Operation=Install Operation=Upgrade Operation=Remove Type=Package Target=linux Target=nvidia-dkms Target=nvidia-470xx-dkms Target=openrazer-driver-dkms Target=system76-dkms Target=v4l2loopback-dkms Target=virtualbox-host-dkms [Action] Description=Update dkms modules in Linux initcpio Depends=mkinitcpio When=PostTransaction NeedsTargets Exec=/usr/bin/mkinitcpio -P
Add / remove whatever you need for the Targets.
Maybe we should make this default?
I installed the new experimental 5.19 kernel from manjaro’s GUI. It just installed
linux519 but it didn’t install
linux519-headers. To solve the problem i installed them manually with pacman.
Dolphin now closes shortly after opening it most of the time. So I have to start it a bunch of times until it finally stays open. In the console I sometimes get “tcache_thread_shutdown(): unaligned tcache chunk detected”, but that seems to be unrelated to when it closes, sometimes I only get the line, sometimes it closes without that line, sometimes both, sometimes neither.
You should have installed the kernel from Manjaro Settings Manager.
A problem came back with the latest linux 518 5.18.3, same I experience before with 5.15.X LTS series… When my screen shutdown and I’m moving the mouse, it still in sleep mode. I need to enter the password and press enter to get back to the desktop. If not, the screen will keep sleeping.
Got that trouble with KDE 5.24 install to the latest update. Didn’t have the issue on Kernel 5.16 and 5.17 series.
I don’t think it’s related to Manjaro since I got that trouble on Arch too.
Same thing with a Ryzen 5700G. I’m a bit wary of trying those steps you linked because I never did any of that on 5.17, it just worked (and still does if I boot to 5.17.14 from grub menu). I have no amd_pstate settings in either /etc/default/grub or /etc/mkinitcpio.conf
It’s not causing any problems that I’ve noticed so far but given current energy prices I’d prefer the 400MHz lower limit rather than 1400MHz
cpupower frequency-info on 5.17.14
analyzing CPU 0: driver: amd-pstate CPUs which run at the same hardware frequency: 0 CPUs which need to have their frequency coordinated by software: 0 maximum transition latency: 131 us hardware limits: 400 MHz - 4.67 GHz available cpufreq governors: conservative ondemand userspace powersave performance schedutil current policy: frequency should be within 400 MHz and 4.67 GHz. The governor "schedutil" may decide which speed to use within this range. current CPU frequency: Unable to call hardware current CPU frequency: 3.77 GHz (asserted by call to kernel) boost state support: Supported: yes Active: yes AMD PSTATE Highest Performance: 166. Maximum Frequency: 4.67 GHz. AMD PSTATE Nominal Performance: 135. Nominal Frequency: 3.80 GHz. AMD PSTATE Lowest Non-linear Performance: 85. Lowest Non-linear Frequency: 2.39 GHz. AMD PSTATE Lowest Performance: 15. Lowest Frequency: 400 MHz.
cpupower frequency-info on 5.18.3
analyzing CPU 0: driver: acpi-cpufreq CPUs which run at the same hardware frequency: 0 CPUs which need to have their frequency coordinated by software: 0 maximum transition latency: Cannot determine or is not supported. hardware limits: 1.40 GHz - 4.67 GHz available frequency steps: 3.80 GHz, 1.70 GHz, 1.40 GHz available cpufreq governors: conservative ondemand userspace powersave performance schedutil current policy: frequency should be within 1.40 GHz and 3.80 GHz. The governor "schedutil" may decide which speed to use within this range. current CPU frequency: Unable to call hardware current CPU frequency: 2.54 GHz (asserted by call to kernel) boost state support: Supported: yes Active: no
I looked at the kernel sources on Manjaro gitlab and amd_pstate is now built as a module in 5.18. That’s why it loads automatically for me in 5.17 but not 5.18.
What is the reason for this change?
$ grep PSTATE linux517/config CONFIG_X86_INTEL_PSTATE=y CONFIG_X86_AMD_PSTATE=y $ grep PSTATE linux518/config CONFIG_X86_INTEL_PSTATE=y CONFIG_X86_AMD_PSTATE=m
So the steps given in this post are correct and worked for me. Ignore the module not found errors for all other kernels, it’s just correctly telling you that the module is only in 5.18.
I dare not reboot because of this grub message:
Grub does no longer support side-loading modules when secure boot is enabled. Thus booting will fail, unless you have an efi executable 'grubx64.efi' with bundled modules.
I’m not sure if I have secure boot enabled or not. Where should the ‘grub64.efi’ file be and what should be in it?
check if you have it disabled:
Wonderful. Thank you. It’s disabled.
My history of a failed system update:
pamac gui showed system update notification.
I ran timeshift to take a snapshot of the system.
While timeshift was running, a terminal and gedit were launched. I edited bmenu and tested it in terminal.
By mistake from the terminal command history, I ran sudo pacman -Syu. pacman found system updates and offered to install them (choose: y/n). I chose cancel.
At this time, the system freezes and I could not do anything until the timeshift job was over. At the end of timeshift, the system returned to its normal state.
I opened the pamac gui window and proceeded with the suggested update. The update was not going as usual, the system was constantly freezing, the update took a very long time. At the end of the update, the system returned to working condition, but something was not as usual.
I rebooted my computer and got a bootloader error: Kernel missing, please load the kernel first. Windows boot items disappeared from the grub menu.
I booted the live CD and found that /boot was missing the kernel and the windows and linux images. Root folder was empty…
I restored the system from the last timeshift snapshot: the system started up, but wifi did not work and there was no uefi settings item in the grub menu.
I restored the system from another timeshift snapshot - everything worked as it should. On a clean booted system, I ran the update with the sudo pacman -Syu command. The system has been updated and everything works great.
I concluded for myself: update the system only after creating a clean backup, do not load the system while timeshift is running, run the update on a clean boot system, using the sudo pacman -Syu command, pamac and pacman may conflict with each other.
Maybe someone will benefit from this experience.
Ooof, been a while since I’ve had one of these:
error: failed to prepare transaction (could not satisfy dependencies) :: installing nvidia-utils (515.48.07-2) breaks dependency 'nvidia-utils=510.60.02' required by linux516-nvidia
I can see that kernel 5.16 is no longer supported (whoops, not sure why I am running this one).
Anyway, attempting to install one of the LTS ones, like 5.15 now fails with the exact same error message.
Same for 5.17