CPU soft lockup - possibly dualgpu related? #6906
Labels
bug
Something isn't working
Comments
|
Anyway please check out our nightly build, https://github.com/rustdesk/rustdesk/releases/tag/nightly, a lot of related changes there. We will reopen if the issue still in nightly build. |
|
also here something I can see during reboot. if you know how/where to get the rest I can get it |
|
@21pages follow up, rewrite this stupid
|
|
@My1 Thanks |
|
Oh 143 is cute, i think it got over 300 easily the first time if not significantly more I think is more than just 100% load this seems kinda more crazy as the process doesn't get killed and even blocks rebooting |
|
@My1 Does this one work for you? If it works for you, could you try https://github.com/21pages/test/releases/download/test/available_no_loop and see whether it will cause the cpu soft lockup? If "available_no_loop" doesn't cause the cpu soft lockup, could you try https://github.com/21pages/test/releases/download/test/available_loop? If the cpu soft lockup happens, please show the terminal log. |
|
no hwcodec installed and ran fine (except that decoding 4k with cpu is a bit straineous for obvious reasons lol) noloop dies with root, without it doesnt open the gui. logs |
|
It is possible that using VA-API on AMD GPUs may encounter this issue. ValveSoftware/steam-for-linux#8508
|
VAinfolockup on ffmpeg with DRI_PRIME=1 Terminal |
|
Thanks, vaapi cause the issue |
|
https://www.reddit.com/r/linux_gaming/comments/wjkqtw/dri_prime_operates_backwards/ |
|
never used DRI_PRIME=0 explicitly before, but both =0 and not set are intel here, altho vainfo seems to not care about vainfo alone, to get the AMD running on vainfo I need both DRI_PRIME=1 and LIBVA_DRIVER_NAME=radeonsi (output of that in earlier comment). additionally, |
|
Thanks |
|
Edited glxinfo stuff with both DRI_PRIME |
|
the fun question is obviously if there's a way to detect these "broken" implementations of whatever is exactly at fault and try something else (like the intel GPU that can not only encode H264 but also H265) additionally, the 2 Desktops I use with RX560 and RX 580 dont have that issue. |
|
Yes, we'll take the ability table into consideration. |
How about the status of this issue? @21pages |
|
FFmpeg also has the same problem on his computer. I haven't removed the VAAPI decoding because currently it is the only one hardware decoder for Intel and AMD graphics cards on Linux. |
|
@21pages submit an issue to ffmpeg and cite here. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Bug Description
basically after installing rustdesk and maybe running it (it happens delayed), i get notifications on my Laptop about the CPU being stuck having a soft lockup with
rustdesk:rcs0:12345(12345 being the process id) being mentioned.htop also mentions aside from rustdesk either one or multiple
--check-hwcodec-configprocesses from rustdeskInterestingly this only happened on my laptop so far, my theory is it might be because the system has 2 GPUs, the integrated intel HD 620 and the AMD Radeon 520
the Processes don't even respond to SIGKILL and the only possible solution is to apt remove and forcefully shutdown as it hangs even in a reboot.
BIOS Update did not help either
How to Reproduce
install rustdesk (I used QApt with the deb file)
start rustdesk
wait
Expected Behavior
the CPU not locking up
Operating system(s) on local side and remote side
Linux, Kubuntu 22.04 (no remote side)
RustDesk Version(s) on local side and remote side
1.2.3 (no remote)
Screenshots
Additional Context
Laptop Model: HP 17-by0320ng
Intel i3 7020U
Intel HD 620
Radeon 520
16GB RAM
The text was updated successfully, but these errors were encountered: