System freeze during video playback (suspecting intel-media-driver)

dimich · 2024-12-26 04:51:10

Last few weeks I got three times system freeze while watching a video on Youtube in Firefox browser. It looks like a hardware freeze: looping audio fragment, no reaction to Magic SysRq, to Ctrl+Alt+F1, no response from network. Only poweroff by holding power button for a several seconds. On next boot there are no suspicious messages in journal: the very last messages from previous boot are ordinary messages about 15-20 minutes before system got frozen.

The hardware is Dell Latitude 5400 laptop with Intel UHD Graphics 620 integrated GPU:

# lspci -v -d 8086:3ea0
00:02.0 VGA compatible controller: Intel Corporation WhiskeyLake-U GT2 [UHD Graphics 620] (rev 02) (prog-if 00 [VGA controller])
        DeviceName: Onboard - Video
        Subsystem: Dell Device 08b8
        Flags: bus master, fast devsel, latency 0, IRQ 155, IOMMU group 0
        Memory at cb000000 (64-bit, non-prefetchable) [size=16M]
        Memory at 80000000 (64-bit, prefetchable) [size=256M]
        I/O ports at 3000 [size=64]
        Expansion ROM at 000c0000 [virtual] [disabled] [size=128K]
        Capabilities: [40] Vendor Specific Information: Len=0c <?>
        Capabilities: [70] Express Root Complex Integrated Endpoint, IntMsgNum 0
        Capabilities: [ac] MSI: Enable+ Count=1/1 Maskable- 64bit-
        Capabilities: [d0] Power Management version 2
        Capabilities: [100] Process Address Space ID (PASID)
        Capabilities: [200] Address Translation Service (ATS)
        Capabilities: [300] Page Request Interface (PRI)
        Kernel driver in use: i915
        Kernel modules: i915

The only recent change which may be relevant is intel-media-driver installation. It wasn't installed before, and I haven't encountered system freezes in more than two years using this laptop. I uninstalled intel-media-driver a week ago and there are no freezes so far. Obviously, video decoding is not hardware-accelerated without intel-media-driver. I monitor it with intel_gpu_top from intel-gpu-tools: Blitter, Video and VideoEnhance engines are always 0.00% busy.
Direct rendering works well all the time.

I would file a bug to upstream, but it's hard to understand, whether this is a hardware bug, or a software bug (intel-media-driver? libva? firefox?), or even a quirk of particular laptop unit. However, software bug in decoder and above should not lead to a total system freeze, I think. The problem is that the issue is quite long to reproduce, and it is not clear what to do next when it finally managed to reproduce.
Has anyone else encountered the same issue on similar type of GPU?

agatti · 2025-02-01 01:21:32

I'm experiencing the same behaviour, but on a Lenovo T490 and on pretty much any program that uses hardware-accelerated video playback (eg. mpv, firefox, chromium, any electron application, etc.). Incidentally, this laptop has an Intel UHD 620 video controller too:

# lspci -v -d 8086:3ea0
00:02.0 VGA compatible controller: Intel Corporation WhiskeyLake-U GT2 [UHD Graphics 620] (rev 02) (prog-if 00 [VGA controller])
        Subsystem: Lenovo Device 2279
        Flags: bus master, fast devsel, latency 0, IRQ 145, IOMMU group 0
        Memory at d7000000 (64-bit, non-prefetchable) [size=16M]
        Memory at b0000000 (64-bit, prefetchable) [size=256M]
        I/O ports at 3000 [size=64]
        Expansion ROM at 000c0000 [virtual] [disabled] [size=128K]
        Capabilities: [40] Vendor Specific Information: Len=0c <?>
        Capabilities: [70] Express Root Complex Integrated Endpoint, IntMsgNum 0
        Capabilities: [ac] MSI: Enable+ Count=1/1 Maskable- 64bit-
        Capabilities: [d0] Power Management version 2
        Capabilities: [100] Process Address Space ID (PASID)
        Capabilities: [200] Address Translation Service (ATS)
        Capabilities: [300] Page Request Interface (PRI)
        Kernel driver in use: i915
        Kernel modules: i915

I'm tempted to say it's an upstream issue, as I seem to remember this happening some time ago as well like two years ago or something like that, and it disappeared after some time (probably with a newer set of Intel video drivers, didn't keep track of that bit back then). At the time people claimed the issue was fixed with a kernel update but that wasn't the case for me, at least.

I haven't tried to revert the drivers to a version older than 24.4.4-1 yet, but that's something I plan to do soon - maybe that fixes the problem until updated drivers come out.

dimich · 2025-02-01 09:44:19

agatti wrote:

I haven't tried to revert the drivers to a version older than 24.4.4-1 yet, but that's something I plan to do soon - maybe that fixes the problem until updated drivers come out.

I disabled GUC in i915 driver:

$ cat /etc/modprobe.d/intel-gpu.conf
options i915 enable_guc=0

and no freezes for a month. Despite that intel-media-driver is installed and seems to be used sometimes to decode video (intel_gpu_top shows non-zero video engine load during playback).
However, disabling GUC taints the kernel, and upstream won't fix this.

dimich · 2025-02-09 10:32:56

dimich wrote:

and no freezes for a month

Nope, it froze even with enable_guc=0. Interestingly, it happened in a few hours after upgrade to linux-6.13.1.arch2-1. I don't think rebuild linux package with newer binutils affected this, likely it's coincedence. Maybe one of earlier upgrades affected, maybe it was just a luck that there were no freezes for a month.

Jay451 · 2025-02-09 18:31:29

dimich wrote:

dimich wrote:
and no freezes for a month
Nope, it froze even with enable_guc=0. Interestingly, it happened in a few hours after upgrade to linux-6.13.1.arch2-1. I don't think rebuild linux package with newer binutils affected this, likely it's coincedence. Maybe one of earlier upgrades affected, maybe it was just a luck that there were no freezes for a month.

I had just tried this fix myself, after reading your previous post, and the system freeze happened again anyway. I'm not an Arch user (I'm using Debian 12), but I have a similar hardware configuration and I can't find any other thread online mentioning this issue. Unfortunately I can't help out much, since I lack any real technical knowledge.

Interestingly enough, I have another laptop with the same integrated GPU on which does not experience any system freezes. The only difference is that on that laptop (an EliteBook 840 G6) I have a slightly older CPU - an Intel Core i5-8265U, whereas on my current laptop (an EliteBook 830 G7) that's affected by the problem I have an Intel Core i5-10310U.

I might also add that, unlike what another poster has mentioned in this thread, I'm not getting any freezes while watching videos locally, that is on a media player like mpv or VLC.

I really hope to see some more awareness to this issue since it means practically not being able to watch any videos online.

Last edited by Jay451 (2025-02-09 19:44:40)

Jay451 · 2025-02-10 12:51:09

Well, what do you know...it turns out I'm getting the same problem on my other laptop as well. The only difference is that it takes longer for the system to freeze - just a handful of minutes on my newer laptop, whereas it can take well over an hour on the older one. What is also interesting is that I've been using the older laptop for about a year, without ever encountering this issue. Could it have been a kernel update that has introduced this issue? I really can't tell...

Last edited by Jay451 (2025-02-10 12:51:56)

dimich · 2025-02-10 21:23:08

Jay451 wrote:

just a handful of minutes on my newer laptop, whereas it can take well over an hour on the older one.

Wow, that's quick enough to test whether one or another workaround works. In my case sometimes it works for weeks without freezing, althrough video is usually played for several hours per day. Once I left video that had caused freeze to play for weekend in a loop, and it didn't freeze. I was just glad the problem is gone away, but next day it got frozen on another video. And no correlation with temperature, CPU load, memory usage, state of charge etc.

Have you tried any of these three kernel parameters? I set "i915.enable_dc=0" for now, but it my case it may take months to determine if it helped.

Jay451 · 2025-02-13 17:45:52

dimich wrote:

Jay451 wrote:
just a handful of minutes on my newer laptop, whereas it can take well over an hour on the older one.
Wow, that's quick enough to test whether one or another workaround works. In my case sometimes it works for weeks without freezing, althrough video is usually played for several hours per day. Once I left video that had caused freeze to play for weekend in a loop, and it didn't freeze. I was just glad the problem is gone away, but next day it got frozen on another video. And no correlation with temperature, CPU load, memory usage, state of charge etc.
Have you tried any of these three kernel parameters? I set "i915.enable_dc=0" for now, but it my case it may take months to determine if it helped.

Hi there, thanks for your answer. It took me a little while to reply because I was testing out another potential solution I had found on this thread. I've tried out the solution posted in one of the answers, specifically that which suggests disabling a power saving feature by Intel by adding

GRUB_CMDLINE_LINUX_DEFAULT="quiet splash i915.enable_psr=0"

in /etc/default/grub. I haven't experienced any freeze in three days. I'm not too sure if this is related to the modification I've made, or if I've just been lucky - time will tell. However, given that right before that I'd usually experience a freeze in just under ten minutes, it could mean that this change has had an effect.

Thanks for the link, though. I'll be sure to try those workarounds as well in case the issue returns.

EDIT: I've ended up getting another freeze, though it does at least take longer this time around. I'll be checking out the workarounds you've linked...

Last edited by Jay451 (2025-02-15 15:18:17)

dimich · 2025-02-16 18:41:04

Jay451 wrote:

i915.enable_psr=0

Thank you, I will try this as well.

Today it froze again despite "enable_dc=0". A few hours after linux-firmware package updated. Coincidence?

I noticed one interesting thing. My laptop model has keyboard backlight controlled by Fn+F5 hotkey. This hotkey doesn't emit any event visible to the system, as I understand it is handled directly by ACPI. Normally it cycles "off" -> "dim" -> "bright" backlight, one press one toggle. When the system is frozen, it still reacts to F5 press (even without Fn) and makes keyboard backlight to blink randomly.

2mods: "Kernel & Hardware" forum seems more appropriate for this topic. Please move it there if you see fit.

Last edited by dimich (2025-02-16 19:54:55)

dimich · 2025-03-05 16:27:16

The system froze also with i915.enable_psr=0. However, mostly all these three weeks it was working on AC, but today it got frozen in several minutes after AC unplugged during video playback.

adriano.vilela · 2025-12-01 01:25:58

Hello!

Did you guys ever figure this out? I've been having the exact same problem for many months now. I seem to have the same GPU on a Dell Inspiron 15 laptop (I'm running Debian Testing though, not Arch):

00:02.0 VGA compatible controller: Intel Corporation UHD Graphics 620 (rev 07) (prog-if 00 [VGA controller])
	DeviceName:  Onboard IGD
	Subsystem: Dell Device 0829
	Flags: bus master, fast devsel, latency 0, IRQ 136, IOMMU group 0
	Memory at dd000000 (64-bit, non-prefetchable) [size=16M]
	Memory at b0000000 (64-bit, prefetchable) [size=256M]
	I/O ports at f000 [size=64]
	Expansion ROM at 000c0000 [virtual] [disabled] [size=128K]
	Capabilities: [40] Vendor Specific Information: Intel Capabilities v1
		CapA: Peg60Dis- Peg12Dis+ Peg11Dis+ Peg10Dis+ PeLWUDis+ DmiWidth=x4
		      EccDis+ ForceEccEn- VTdDis- DmiG2Dis+ PegG2Dis+ DDRMaxSize=Unlimited
		      1NDis- CDDis- DDPCDis+ X2APICEn+ PDCDis- IGDis- CDID=0 CRID=8
		      DDROCCAP- OCEn- DDRWrtVrefEn- DDR3LEn+
		CapB: ImguDis- OCbySSKUCap- OCbySSKUEn- SMTCap+ CacheSzCap 0x3
		      SoftBinCap- DDR3MaxFreqWithRef100=Disabled PegG3Dis+
		      PkgTyp- AddGfxEn+ AddGfxCap- PegX16Dis+ DmiG3Dis+ GmmDis-
		      DDR3MaxFreq=2666MHz LPDDR3En+
	Capabilities: [70] Express Root Complex Integrated Endpoint, IntMsgNum 0
	Capabilities: [ac] MSI: Enable+ Count=1/1 Maskable- 64bit-
	Capabilities: [d0] Power Management version 2
	Capabilities: [100] Process Address Space ID (PASID)
	Capabilities: [200] Address Translation Service (ATS)
	Capabilities: [300] Page Request Interface (PRI)
	Kernel driver in use: i915
	Kernel modules: i915

I've tried all the settings below (one at a time), to no avail:

i915.enable_guc=0
i915.enable_psr=0
i915.enable_dc=0

seth · 2025-12-01 09:01:19

no reaction to Magic SysRq, to Ctrl+Alt+F1, no response from network

Is this true for youas well?
https://wiki.archlinux.org/title/Keyboa … el_(SysRq) - nb. you'l have to explicitly enable that first!

The only systematic approach to debug an actual kernel panic (where you don't get a fancy QR code) would be https://wiki.archlinux.org/title/Kdump

Lone_Wolf · 2025-12-01 10:19:22

@adriano.vilela
We can't help you with issues on a debian testing installation, please ask for help on debian support channels.

adriano.vilela · 2025-12-01 11:32:30

seth wrote:

no reaction to Magic SysRq, to Ctrl+Alt+F1, no response from network
Is this true for youas well?
https://wiki.archlinux.org/title/Keyboa … el_(SysRq) - nb. you'l have to explicitly enable that first!
The only systematic approach to debug an actual kernel panic (where you don't get a fancy QR code) would be https://wiki.archlinux.org/title/Kdump

Thanks a lot for your feedback. I'll look into your suggestions.

adriano.vilela · 2025-12-01 11:33:45

Lone_Wolf wrote:

@adriano.vilela
We can't help you with issues on a debian testing installation, please ask for help on debian support channels.

I'm sorry about that.

I only posted here because i) this doesn't seem to be distribution-specific, but rather hardware-specific, and ii) the discussion up to this point happened mostly between an Arch (@dimich) and a Debian (@Jay451) user. After suffering with this for months and not really finding many other threads discussing this exact same problem on the Internet, I thought I'd check with them.

I really don't want to bother anyone in here, but if there are people willing to help, I'm afraid I may post again, unless you feel this is really inappropriate.

Last edited by adriano.vilela (2025-12-01 12:15:23)

dimich · 2025-12-01 13:12:26

adriano.vilela wrote:

Did you guys ever figure this out?

I don't know what was the root cause but the issue seems gone. No freezes for the last several months.
Currently I have only "enable_psr=0" option for i915 module (not kernel command line parameter, but it should be the same).

adriano.vilela · 2025-12-01 13:41:22

dimich wrote:

I don't know what was the root cause but the issue seems gone. No freezes for the last several months.
Currently I have only "enable_psr=0" option for i915 module (not kernel command line parameter, but it should be the same).

I see. Do you have intel-media-driver installed?

In my case, uninstalling it seems to solve the problem (though I only tested for a couple of days, and the problem can take various days to show up). But uninstalling intel-media-driver implies no video acceleration :-(

dimich · 2025-12-01 14:07:15

adriano.vilela wrote:

Do you have intel-media-driver installed?

Yes, currently it is 25.3.4-1.

adriano.vilela wrote:

In my case, uninstalling it seems to solve the problem

Yes, there was no single freeze when intel-media-driver wasn't installed.

adriano.vilela wrote:

But uninstalling intel-media-driver implies no video acceleration :-(

Actually I'm not sure whether Firefox uses video acceleration in my system. I gave up figuring out how to check it reliably.
UPD: if someone provides a link to a video which definitely shoud playback via hardware decoder on Intel GPU, I can test it.

Last edited by dimich (2025-12-01 14:11:33)

seth · 2025-12-01 16:29:41

Pretty much everything on youtube, certainly h264.
About making sure to use hw decoding, see about:support - but if it's not tied to the browser I'd just use mpv to control that part.

If you install https://archlinux.org/packages/extra/x8 … media-sdk/ you can also
"mpv --hwdec=qsv" instead of "hwdec=vaapi" and see whether that makes a difference.

dimich · 2025-12-01 18:07:57

seth wrote:

Pretty much everything on youtube, certainly h264.

Thank you, seth.

MPV playing sample from this site with "--hwdec=qsv" shows 0.0% in gputop "video" column and about 60-80% cpu usage in top.
With "--hwdec=vaapi" it shows about 10% in gputop "video" and 10-20% of cpu in top.

Also I can see non-zero "video" value in gputop (jumps up to 5-10%) for Firefox's "RDD Process" after youtube resolution switched manually to 1080p or above.
So I'm concluding that hardware video acceleration still works.

adriano.vilela · 2025-12-01 20:31:51

Yes, as far as I understand, if the column "video" in gputop shows a value different from zero, that means hardware decoding is being used.

I also ran some tests on videos from that site, playing them both from Firefox and from mpv and, in both cases, gputop indicates hardware decoding.

I set up Kdump, as suggested by seth, to see if I can debug this when the next crash happens.

Thanks a lot, folks.

Last edited by adriano.vilela (2025-12-01 20:36:05)

seth · 2025-12-01 20:44:25

MPV playing sample from this site with "--hwdec=qsv" shows 0.0% in gputop "video" column and about 60-80% cpu usage in top.

Oh, qsv doesn't work anymore … I've been decoding in software since a while… fuck.

Arch Linux

#1 2024-12-26 04:51:10

System freeze during video playback (suspecting intel-media-driver)

#2 2025-02-01 01:21:32

Re: System freeze during video playback (suspecting intel-media-driver)

#3 2025-02-01 09:44:19

Re: System freeze during video playback (suspecting intel-media-driver)

#4 2025-02-09 10:32:56

Re: System freeze during video playback (suspecting intel-media-driver)

#5 2025-02-09 18:31:29

Re: System freeze during video playback (suspecting intel-media-driver)

#6 2025-02-10 12:51:09

Re: System freeze during video playback (suspecting intel-media-driver)

#7 2025-02-10 21:23:08

Re: System freeze during video playback (suspecting intel-media-driver)

#8 2025-02-13 17:45:52

Re: System freeze during video playback (suspecting intel-media-driver)

#9 2025-02-16 18:41:04

Re: System freeze during video playback (suspecting intel-media-driver)

#10 2025-03-05 16:27:16

Re: System freeze during video playback (suspecting intel-media-driver)

#11 2025-12-01 01:25:58

Re: System freeze during video playback (suspecting intel-media-driver)

#12 2025-12-01 09:01:19

Re: System freeze during video playback (suspecting intel-media-driver)

#13 2025-12-01 10:19:22

Re: System freeze during video playback (suspecting intel-media-driver)

#14 2025-12-01 11:32:30

Re: System freeze during video playback (suspecting intel-media-driver)

#15 2025-12-01 11:33:45

Re: System freeze during video playback (suspecting intel-media-driver)

#16 2025-12-01 13:12:26

Re: System freeze during video playback (suspecting intel-media-driver)

#17 2025-12-01 13:41:22

Re: System freeze during video playback (suspecting intel-media-driver)

#18 2025-12-01 14:07:15

Re: System freeze during video playback (suspecting intel-media-driver)

#19 2025-12-01 16:29:41

Re: System freeze during video playback (suspecting intel-media-driver)

#20 2025-12-01 18:07:57

Re: System freeze during video playback (suspecting intel-media-driver)

#21 2025-12-01 20:31:51

Re: System freeze during video playback (suspecting intel-media-driver)

#22 2025-12-01 20:44:25

Re: System freeze during video playback (suspecting intel-media-driver)

Board footer