[SOLVED] System freezing randomly on Kernel 5.10.4

[aRandomDev]

Hi, this is my first post on this forum, I've read the rules and FAQ but forgive me if i make some mistakes.

I've been using arcolinuxd-plasma for about 6 months now as a daily development machine on a newly purchased xps 13 9300, with intel i7 1065G7 processor and Iris Plus Graphics.

It has been working wonders so far!

Recently though I've been getting random system freezes, even the cursor, trying to switch to a different TTY using Ctrl + Alt + Fx does not work either. If I have music playing during the freeze, it continues playing for about 2 minutes, then it also freezes as well. Only option after that is to power off by holding power button.

Looking at the journal logs I can see that the last error that gets logged every time after the freeze is the following (it only logs these after about 2 minutes after the freeze, if i power off before waiting i get no error logs)

Code: Select all

jan 10 11:17:49 xps9300 kernel: INFO: task kworker/4:1H:146 blocked for more than 122 seconds.
jan 10 11:17:49 xps9300 kernel:       Tainted: P           OE     5.10.6-arch1-1 #1
jan 10 11:17:49 xps9300 kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
jan 10 11:17:49 xps9300 kernel: task:kworker/4:1H    state:D stack:    0 pid:  146 ppid:     2 flags:0x00004000
jan 10 11:17:49 xps9300 kernel: Workqueue: events_highpri intel_atomic_cleanup_work [i915]
jan 10 11:17:49 xps9300 kernel: Call Trace:
jan 10 11:17:49 xps9300 kernel:  __schedule+0x295/0x810
jan 10 11:17:49 xps9300 kernel:  schedule+0x5b/0xc0
jan 10 11:17:49 xps9300 kernel:  schedule_preempt_disabled+0x11/0x20
jan 10 11:17:49 xps9300 kernel:  __ww_mutex_lock.constprop.0+0x4bd/0x810
jan 10 11:17:49 xps9300 kernel:  ? dequeue_entity+0xc6/0x460
jan 10 11:17:49 xps9300 kernel:  intel_unpin_fb_vma+0x25/0xa0 [i915]
jan 10 11:17:49 xps9300 kernel:  drm_atomic_helper_cleanup_planes+0x52/0x70 [drm_kms_helper]
jan 10 11:17:49 xps9300 kernel:  intel_atomic_cleanup_work+0x67/0x110 [i915]
jan 10 11:17:49 xps9300 kernel:  process_one_work+0x1d6/0x3a0
jan 10 11:17:49 xps9300 kernel:  worker_thread+0x4d/0x3d0
jan 10 11:17:49 xps9300 kernel:  ? rescuer_thread+0x410/0x410
jan 10 11:17:49 xps9300 kernel:  kthread+0x133/0x150
jan 10 11:17:49 xps9300 kernel:  ? __kthread_bind_mask+0x60/0x60
jan 10 11:17:49 xps9300 kernel:  ret_from_fork+0x1f/0x30
jan 10 11:18:48 xps9300 systemd-logind[1507]: Power key pressed.

This error happens about 2 to 3 times every day, which is quite frustating and unfortunately I didn't found a way to reproduce it, it is very random in nature.

About two days ago, I did a rollback to kernel 5.9.14 and since then the freeze disappeared!

Googling this log message i found this post https://bbs.archlinux.org/viewtopic.php?id=262422 on arch linux forums, which also happened recently and seems to be exactly the same issue I'm having. On it apparently the issue has something to do with xf86-video-intel. It was suggested to remove the package and apparently it did solve the issue.

Unfortunately, I'm not well versed in messing with video drivers, and can't afford to lose access to this work machine, so I tried the kernel rollback and it solved my issue for now.

What do you reccomend me to do in this case? Is there a kernel bug report place where I can report this, as it doesn't look to be anything arco-linux specific?

Thanks!

[aRandomDev]

I've also found similar issue here, and posted a reply there as well

[erikdubois]

We have a similar post on our own forum. Recently discovered.

Starting with the last kernels and latest hardware it is better not to have the xf86-video-intel installed.

So uninstall it and see if the freezes come back if you move back to the last kernel.

[aRandomDev]

Thanks! I've read in the other post that uninstalling the xf86-video-intel might cause an unbootable system and tearing issues, and it was recommended to use another compositor. I don't understand very well how that works, but will look into it and test removing it and moving back to latest kernel after making a full backup when i have some time this weekend. I'll post the results here.

[erikdubois]

Great - hope to hear from you again.

ps. all my computers are NOT using xf86-video-intel - only one old computer of 10 years (tested on 13 pcs) needs the package and then looks great again

[aRandomDev]

Hello, just an update, I've been following this bug and seems the fix is about to be merged into Kernel 5.10.15.

From the discussion there, seems like this can happen even with modesetting driver, though much less frequently.

Once 5.10.15 releases I'll upgrade to it and stay a few days testing it to see if the issue is gone for good.

[erikdubois]

My computer that needed the xf86-video-intel does not need it anymore.

New kernel came in.
No extras to install.

And so it all changes.

[aRandomDev]

Hi, it's been two weeks since updated kernel to 5.10.15 and didn't have a single freeze since then. I guess this issue can be marked as solved. Thanks!