-
Notifications
You must be signed in to change notification settings - Fork 5.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Kernel 6.13.0 AARCH64 Freezes Up #6624
Comments
A build of d53a520 on a fresh install of RPi OS is running happily for me with either the regular v8. No - spoke too soon. Playing a music video in YouTube while updating to the 16k page build led to:
|
And another, this time trying to exit full screen:
|
Yep. It seems to happen at random times. |
Any ideas, @mairacanal? |
I did revert b9acfd8 yesterday but it did not fix it. |
Trade a crash in v3d_irq for a warning. This is just a temporary sticking plaster over the wound - proper surgery is required. Link: raspberrypi#6624 Signed-off-by: Phil Elwell <[email protected]>
The immediate cause of the crash is a NULL #6625 adds a guard against that, logging a WARN when it happens. It also guards bin_job and csd_job, just in case. After the warning, the GUI freezes, but serial and ssh sessions survive, which might make it easier to check the logs, or at least just sync the disc. |
@pelwell I'll start debugging this issue now, hopefully I'll bring a solution in a couple of hours |
That's great, but this is the bleeding edge kernel, so the stakes are bit lower. By the way, I just saw a NULL bin_job. |
The crash is caused by e4b5ccd ("drm/v3d: Ensure job pointer is set to NULL after job completion"). Therefore, be careful as this patch is also in the stable branches (such as 6.6). I mainly tested unload/load for that patch and basic rendering applications. The fix is already available in this branch. Could you test it and confirm that it fixes the issue for you? I'm playing several YouTube videos for testing and it looks okay here. If it solves the issue for you, let me know and I'll proceed to upstream the patch (your R-b will be welcomed) and send a PR to this repo. Mostly, only the commit message is missing to wrap up the fix. |
Thanks, Maira - it's looking good so far. I've cherry-picked the fix as #6626. |
Do you happen to know why one of the KUnit tests is failing? https://github.com/raspberrypi/linux/actions/runs/12909243870/job/35996480799#step:4:21 |
Link: #6624 Fixes: e4b5ccd ("drm/v3d: Ensure job pointer is set to NULL after job completion") Signed-off-by: Maíra Canal <[email protected]> Co-authored-by: Maíra Canal <[email protected]>
Maira's patch is now merged to rpi-6.13.y. Although the offending commit has been merged to the upstream stable branch, it doesn't yet appear to have been back-ported to any of the LTS branches, so I think we're safe for now. |
Phil, I didn't manage to reproduce this error locally. Could you reproduce it locally? Usually, "Error: 51] [ERROR] Test: : Could not find any KTAP output. Did any KUnit tests run?" indicates some output error, not a test failure specifically. |
Seems to be working good. Thanks everyone!! |
You're right - it works when I run the tests here. Something else is different between the arm and arm64 tests. The arm log begins:
Compare that to the arm64 log:
I don't know why ARCH=arm64 would be different in this regard. |
I sent the patch to the dri-devel mailing list for review [1]. |
Re the arm64 KUnit, re-running a previously successful test using the GitHub actions failed this time. Something in the environment must have changed. Whatever it is, it's not your problem Maíra. |
Link: #6624 Fixes: e4b5ccd ("drm/v3d: Ensure job pointer is set to NULL after job completion") Signed-off-by: Maíra Canal <[email protected]> Co-authored-by: Maíra Canal <[email protected]>
Link: #6624 Fixes: e4b5ccd ("drm/v3d: Ensure job pointer is set to NULL after job completion") Signed-off-by: Maíra Canal <[email protected]> Co-authored-by: Maíra Canal <[email protected]>
Describe the bug
The 6.13.0 aarch64 kernel freezes up at different stages from the login screen onwards. It is most prevelent after opening up firefox or chromium trying to play a video after a few seconds. The screen will lock up and the audio goes into a short loop.
Steps to reproduce the behaviour
Follow the official RPi guide to compile and install the kernel. Kernel compiled at d53a520.
Device (s)
Raspberry Pi 4 Mod. B, Raspberry Pi 5
System
Manjaro ARM
XFCE
Linux jellyfin 6.13.0-v8-16k #1 SMP PREEMPT Wed Jan 22 01:33:02 CST 2025 aarch64 GNU/Linux
[ray@jellyfin ~]$ sudo vcgencmd version
2025/01/14 00:16:48
Copyright (c) 2012 Broadcom
version 0451f142 (release) (embedded)
Logs
Additional context
It is almost impossible to get any errors reported when it freezes. journalctl -f and dmesg -w does not produce anything.
I did get his one time when opening firefox in a terminal but most of the time I see no errors:
Another guy in our forums testing reported getting this:
The text was updated successfully, but these errors were encountered: