So, I recently updated my system (Two days ago) after that, I am unable to open any application I installed using third party sources. For Eg.: VSCode, Brave Browser, Firefox Developer Edition. I am just able to use application that came preinstalled Like: Firefox (I am currently visiting this website using the same).Although, Firefox has been crashing too but, not like the other applications. For Eg: I tried starting up Brave Browser and suddenly, my screen froze. I was able to move my mouse but, it was of no avail. Then, I used: Ctrl + Alt + F3 to get into those shells and shutdown(ed) my system from there (Used: shutdown command.).
Again, after restarting my PC, I tried to open Brave and VSCode and still, the problem persisted.
So, I was just trying to fix things up and then, during the next startup, I select the version before 5.7.7 (from GRUB) and everything worked fine.
I did a clean install (Like: Wiped out my Linux Portion of the drive and installed Fedora 32 again) and still the problem persists.
Just wanted to post an Update: So, I selected 5.6.6 version from the GRUB menu and everything seems to work fine. (I just re-installed VSCode and Brave and they are working as intended). The reason, I am fixating on Kernel version is cause, I just read another post where the user is facing issues due to Kernel 5.7.7 and I think, my problem and his are closely related.
@snikhill
Sorry about all the trouble. Usually Fedora keeps three kernels for situations like this. You have succeeded in establishing one thing – which is that Kernel-5.7.7 doesn’t work for you. So file a bug report against it at https://bugzilla.redhat.com. You will provide information to assist developers in resolving the issue. Alternatively, you can wait for another update. Just keep booting the last good Kernel until another working update arrives. That is the design
Interesting I’ve been having similar issues since 5.7.10 and it IS related to AMDGPU. Granted I run risky environments with ROCm and out-of-band OpenCL drivers but suddenly I got no video and even had grub issues since 5.7.10. Turns out there was some kind of regression that broke my AMDGPU dkms in the post script of kernel-core. That includes the latest builds for EL8 and EL7. Trying to reinstall it repeatedly eliminating a few of my AMD packages at a time but I’m pretty close to ripping it out once and for all (and throwing this AMD sidearm into the river because I’m so sick of AMDGPU issues). Sometime along the update the code broke for these ROCm modules which supposedly were finally ready for F32. I’ll disable it and file an issue with AMD.
Building module:
cleaning build area…
make -j40 KERNELRELEASE=5.7.10-201.fc32.x86_64 -j40 TTM_NAME=amdttm SCHED_NAME=amd-sched kdir=/lib/modules/5.7.10-201.fc32.x86_64/build -C /lib/modules/5.7.10-201.fc32.x86_64/build M=/var/lib/dkms/amdgpu/3.5-30.el7/build…(bad exit status: 2)
Error! Bad return status for module build on kernel: 5.7.10-201.fc32.x86_64 (x86_64)
Consult /var/lib/dkms/amdgpu/3.5-30.el7/build/make.log for more information.
Done.
Since last update there is no longer possible to select the 5.6.x kernel. It was removed. I can only select from 5.7.10, 5.7.11, 5.7.12. Unfortunately the issue exist on all of them.
I am typing this message from Fedora 32 with kernel version: 5.6.6.
Well ever since I reported the issue, I haven’t updated my kernel. I exclude the same from the updates.
Also, I have observed that if you install kernel using the official channel then, your current kernel will never be replaced. (For e.g.: If I am on kernel 5.6.6 and run a update using dnf then, kernel 5.6.6 will not be replaced). BUT KEEP IN MIND: THIS TRUE ONLY IF YOU UPDATE USING THE OFFICIAL WAY. (Yes, the current kernel can be overridden if you use any third-party method without proper caution.)
Install those dependencies first: sudo dnf install ./kernel-core-5.6.8-300.fc32.x86_64.rpm sudo dnf install ./kernel-modules-5.6.8-300.fc32.x86_64.rpm
(assuming you are already in the folder holding the packages)
then install the rest likewise.
Yes
Just like @snikhill said, if you booted with it … dnf will not remove it.
Have you looked into log files to see what is actually happening? Switching to an unpatched kernel (5.6.x 5.7.x) cannot be the solution here.
Also, do you get any error messages from the apps that crash? Start them from command line to see their output.
Have you tried to start Gnome on Xorg and Wayland to see whether the problem persists?
Have temporarily created a new user to see whether the crashes are related to your user profile?
I don’t know how to check log files and what to look for.
Yes. I suppose I can’t stay on this long-term. Trying to use lts kernel is probably something I will not be able to do.
For Gnome Video, and for VLC I got no message just the screen freeze.
For Brave I got …ERROR:sandbox_linux.cc(374)] InitializeSandbox() called with multiple threads in process gpu-process.
The problem exists in both Gnome classic session, and in Gnome Xorg.
In both of those even opening a terminal freezes everything.
Freeze is still there.
I’ve tried on 5.8.12-200.fc32.x86_64
I also learnt that opening or trying to play a video in Gnome Videos is also causing the freeze.
The freeze is sometimes causing session to restart. In most cases it just freeze with window of the app that I want to open. Then I can only move mouse cursor but can’t click on anything. Sometimes the screen changes the color when I click. What I can only do is the CTRL+ALT+F3.