You are not logged in.
Hi all, starting Dec. 31, I frequently experience i915 freezing everyday. I suspect that the guilty packages could be Intel driver or kernel. Here's the timeline:
Dec. 30 / 2020
Noon: Upgraded xf86-video-intel 1:2.99.917+913+g9236c582-1 -> 1:2.99.917+914+ga511f22c-1
Night: Sleep, shutdown.
Dec. 31 / 2020
Afternoon: Upgraded intel-media-driver 20.4.3-1 -> 20.4.5-1, intel-gmmlib 20.4.3-1 -> 20.4.5-1 and linux 5.9.14.arch1-1 -> 5.10.3.arch1-1
I rebooted 38 minutes later (that's not a freeze)
Three hours after boot, it freezed (the first time). I did not find anything in journal, and I forget to check kernel message and /sys. Rebooted, but Xorg and the Intel driver process is in D state so I forced reboot.
Night: Sleep, shutdown.
Jan. 1 / 2021
Night: Freezed. This time, I still did not find anything in journal, but I found some errors in the kernel message (see below). I forget to check /sys. Rebooted, and the process hanged as the previous case
After rebooting I found the WiFi NIC crashes. That's not related to the GPU but I rebooted here.
Five minutes after the reboot, I upgraded mesa 20.3.1-1 -> 20.3.2-2 and vulkan-intel 20.3.1-1 -> 20.3.2-2
After two hours and twenty minutes, it freezed again. Still, I did not check /sys. Rebooted, and processes hanged as the previous case.
Night: Sleep, shutdown.
Jan. 2 / 2021
Afternoon: It freezed again. Same as the previous case.
There's no logs in kernel messages and journal before and after the freeze. The only thing I found in kernel message is:
kernel: Workqueue: events_highpri intel_atomic_cleanup_work [i915]
kernel: Call Trace:
kernel: __schedule+0x295/0x810
kernel: schedule+0x5b/0xc0
kernel: schedule_preempt_disabled+0x11/0x20
kernel: __ww_mutex_lock.constprop.0+0x4bd/0x810
kernel: intel_unpin_fb_vma+0x25/0xa0 [i915]
kernel: drm_atomic_helper_cleanup_planes+0x52/0x70 [drm_kms_helper]
kernel: intel_atomic_cleanup_work+0x67/0x110 [i915]
kernel: process_one_work+0x1d6/0x3a0
kernel: worker_thread+0x4d/0x3d0
kernel: ? rescuer_thread+0x410/0x410
kernel: kthread+0x133/0x150
kernel: ? __kthread_bind_mask+0x60/0x60
kernel: ret_from_fork+0x22/0x30
The freeze happens randomly, and I cannot re-produce it now, but it happens frequently. The suspicious packages include kernel, xf86-video-intel, intel-media-driver and intel-gmmlib. Are there anyone else experienced the same thing? Is it a bug? How can I resolve it? What's the reason? Are there anything I can do to get more information if it happens again? Thanks.
Last edited by YuutaW (2021-01-02 21:29:50)
Offline
You could try removing xf86-video-intel and using the kernel's native modesetting; it works better for most setups.
Offline
You could try removing xf86-video-intel and using the kernel's native modesetting; it works better for most setups.
Thanks. I would try that out. Will there be any performance affects?
Offline
I removed xf86-video-intel, and rebooted. I will report any issues if found.
Offline
Do you by any chance have monitor connected by HDMI? I have (and someone else) a problem with drm/i915/Xorg: after monitor goes to power save mode, Xorg hang indefinitely due to crash in drm/i915 module and keyboard becomes irresponsible. If you have the same problem (I don't use xf86-video-intel almost an year now and didn't see any drawbacks) show your dmesg output. See these posts: https://bbs.archlinux.org/viewtopic.php?pid=1947353 and this https://bbs.archlinux.org/viewtopic.php?pid=1947354
Offline
Do you by any chance have monitor connected by HDMI? I have (and someone else) a problem with drm/i915/Xorg: after monitor goes to power save mode, Xorg hang indefinitely due to crash in drm/i915 module and keyboard becomes irresponsible. If you have the same problem (I don't use xf86-video-intel almost an year now and didn't see any drawbacks) show your dmesg output. See these posts: https://bbs.archlinux.org/viewtopic.php?pid=1947353 and this https://bbs.archlinux.org/viewtopic.php?pid=1947354
Yes, on Jun. 2 I experienced freezing even with an external monitor connected. However, none of the freezes happened when they are in power save mode: in fact, they all happen when I'm working, either chatting online or writing code, etc.
Offline
Update: After removed the driver and switched to mode setting, I have not experienced freezing for one day. However, I encountered problems watching some YouTube video using Chromium. The audio seems working, but the video is always white screen. I had enabled hardware accelerated decoding weeks ago, is the driver required to watch videos?
chrome://gpu report:
Graphics Feature Status
Canvas: Hardware accelerated
Compositing: Hardware accelerated
Multiple Raster Threads: Enabled
Out-of-process Rasterization: Hardware accelerated
OpenGL: Enabled
Hardware Protected Video Decode: Hardware accelerated
Rasterization: Hardware accelerated on all pages
Skia Renderer: Enabled
Video Decode: Hardware accelerated
Vulkan: Enabled
WebGL: Hardware accelerated
WebGL2: Hardware accelerated
Offline
That's a separate issue. Please mark this thread as solved and begin a new thread with details about the Chromium/YouTube problem.
Offline
That's a separate issue. Please mark this thread as solved and begin a new thread with details about the Chromium/YouTube problem.
Yes I know. I'm just mentioning it here.
Offline
jasonwryan wrote:That's a separate issue. Please mark this thread as solved and begin a new thread with details about the Chromium/YouTube problem.
Yes I know. I'm just mentioning it here.
Offline
The freeze happens randomly, and I cannot re-produce it now, but it happens frequently. The suspicious packages include kernel, xf86-video-intel, intel-media-driver and intel-gmmlib. Are there anyone else experienced the same thing? Is it a bug? How can I resolve it? What's the reason? Are there anything I can do to get more information if it happens again? Thanks.
I'm having the same issue. Testing without the xf86-video-intel now. When booting i had to put on the grub line "quiet splash", because for some reason it froze the first time i rebooted. No problems with youtube though.
Offline
I'm also having this issue on one of my four machines using xf86-video-intel. The hardware is the ivybridge integrated HD 4000 GPU. Unfortunately I can't test if this still happens without xf86-video-intel as X won't even start without it. So far this has only happened when playing videos. Since the sound keeps playing I assumed it was just X freezing. The Xorg logfiles don't really reveal anything but looking through journalctl I can see that several graphical applications like my session manager complain about "XIO: fatal IO error".
Offline
YuutaW wrote:jasonwryan wrote:That's a separate issue. Please mark this thread as solved and begin a new thread with details about the Chromium/YouTube problem.
Yes I know. I'm just mentioning it here.
Okey, I will post another post if I have any other issues.
Offline
I found this issue on freedesktop Gitlab: https://gitlab.freedesktop.org/drm/intel/-/issues/2905
Looks similar. Reported my findings there and downgraded to 5.9.14.arch1-1.
Offline
I found this issue on freedesktop Gitlab: https://gitlab.freedesktop.org/drm/intel/-/issues/2905
Looks similar. Reported my findings there and downgraded to 5.9.14.arch1-1.
This looks similar. I will follow this issue as well.
Offline