You are not logged in.
I suffered from high-frequent freezes as well (ca every 20min in avg.) but since i disabled the display compositing of my windows manager (xfce) i had no GPU hangs in the last 2 days.
Last edited by dantscher (2020-01-16 15:39:21)
Offline
I'm watching for proper fix for this issue for quite some time for 5.4 kernel, it's not yet ready, however, this helped me to solve hangs on Skylake system for two weeks now (modesetting and intel x11 drivers): https://gitlab.freedesktop.org/drm/inte … ote_376308
I compile my own kernels, but if You want, try this revert until it gets solved properly.
Edit: NVM, it crashed on day 13, so it's not a fix
Last edited by Kirurgs (2020-01-21 13:12:01)
Offline
Same problem. Rolled back to 5.3.0-arch1.
Offline
Just for the record, I have what I believe to be the same problem still with kernel 5.4.13-arch1-1
$ journalctl -xb-1
Jan 20 22:30:51 <hostname> kernel: Linux version 5.4.13-arch1-1 (linux@archlinux) (gcc version 9.2.0 (GCC)) #1 SMP PREEMPT Fri, 17 Jan 2020 23:09:54 +0000
(...)
Jan 21 21:10:01 <hostname> kernel: Rekeying PTK for STA <mac address??> but driver can't safely do that.
Jan 21 21:10:01 <hostname> kernel: i915 0000:00:02.0: GPU HANG: ecode 9:1:0x00000000, hang on rcs0
(...)
We are exactly the people our parents always warned us about.
Offline
Can confirm that this is still an issue in kernel 5.4.14-arch1-1.
Offline
Can confirm that this is still an issue in kernel 5.4.14-arch1-1.
+1
i5-8265u
Offline
Does adding the option i915.enable_dc=0 have any effect on 5.4.14-arch1-1 ?
Offline
Upgraded to linux-mainline (5.5.0-1) with
intel_idle.max_cstate=1 i915.enable_dc=0
kernel options - no more hangs but many still many freeze ups that don't even show up in journalctl anymore. Trying to play with few more kernel parameters but it seem futile at this point - will report if I find something.
In the meantime there are some suggestions in the upstream thread: https://gitlab.freedesktop.org/drm/intel/issues/673 though nothing has worked for me so far. Glad to get rid of full on hangs though!
Offline
What is the latest kernel version unaffected by this bug AFAYK?
Offline
5.5 should (generally, seems there are some outliers here and there) fix what has been discussed here initially.
There seem to be some other factors than just the GPU, FWIW I haven't been able to reproduce any of this on my skylake based Acer V (granted I did not yet extensively use 5.4 I was out of date for a while due to projects)
Last edited by V1del (2020-01-30 08:36:21)
Online
Confirm that kernel 5.4.15 is affected by the issue at i7-8750H. But I have very rare (about 1 per day) freezes maybe because of DRI option in /etc/X11/xorg.conf.d/20-intel.conf
Section "Device"
Identifier "Intel Graphics"
Driver "intel"
Option "DRI" "2"
EndSection
Offline
You should try without xf86-video-intel and that config file. It does have a few known issues that might stem from it.
Last edited by V1del (2020-01-30 08:55:50)
Online
You should try without xf86-video-intel and that config file. It does have a few known issues that might stem from it.
Thanks for your reply. I have very strange image colors without xf86-video-intel and that config file. It's something like brown filter applied or like redshift is on with very red temperature.
Offline
For what it's worth, I haven't had any bugs since I moved to Wayland a month ago, and stopped using chromium on xwayland. Other less resource intensive x11 apps do not trigger this issue.
Offline
I am having the same problem; here is my error log:
an 30 07:40:13 ArchSys kernel: i915 0000:00:02.0: GPU HANG: ecode 9:1:0x00000000, hang on rcs0
Jan 30 07:40:13 ArchSys kernel: GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
Jan 30 07:40:13 ArchSys kernel: Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
Jan 30 07:40:13 ArchSys kernel: drm/i915 developers can then reassign to the right component if it's not a kernel issue.
Jan 30 07:40:13 ArchSys kernel: The GPU crash dump is required to analyze GPU hangs, so please always attach it.
Jan 30 07:40:13 ArchSys kernel: GPU crash dump saved to /sys/class/drm/card0/error
Jan 30 07:40:13 ArchSys kernel: i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
Jan 30 07:40:17 ArchSys kernel: Asynchronous wait on fence i915:cinnamon[1116]:14a48e timed out (hint:intel_atomic_commit_ready+0x0/0x50 [i915])
Jan 30 07:40:21 ArchSys kernel: i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
I managed to capture an actual crash dump yesterday, since for some reason for the first time the system was able to recover without hanging.
A couple of things that I observed:
1) 100% of crashes happen while I am browsing with Qutebrowser; it seems many people are crashing while using Chromium. As far as I understand, these share some aspects of a rendering engine (QtWebEngine?). Maybe there is a connection? Of course, this can be just misleading: I spend 90% of my computer time on Qutebrowser or typing things
2) A moderator asked me why I use the intel driver rather than the modesetting driver; I don't know how to change the driver I am using. I was using xf86-video-intel but I uninstalled it and deleted the 20-intel.conf file that configured it some time ago. Is there anything else I should have done?
Offline
I've started seeing the same issue (hard hang with an oops in the i915 module, which I did manage to capture).
I found this bug report which seems like the exact issue; just thought it would be worth posting here as I didn't see anyone else reference it yet.
Offline
I've started seeing the same issue (hard hang with an oops in the i915 module, which I did manage to capture).
I found this bug report which seems like the exact issue; just thought it would be worth posting here as I didn't see anyone else reference it yet.
I think there might be more than one bug, because I downgraded to 5.3 and I just got an hang a few minutes ago. The bug report you linked talks about 5.4+.
Offline
I think there might be more than one bug, because I downgraded to 5.3 and I just got an hang a few minutes ago. The bug report you linked talks about 5.4+.
Yeah, agreed, in fact I shouldn't have highjacked this thread with a 5.4 issue, but someone already did and I forgot the OP was talking about 5.3. Maybe it will still be useful.
Offline
@merc68k that issue is fixed in 5.5 https://github.com/torvalds/linux/commi … 99fd1721ed
Edit:
It is not fixed in 5.4 as no backport has been submitted https://lore.kernel.org/stable/15783126 … kroah.com/
See also https://bbs.archlinux.org/viewtopic.php … 3#p1884353
Last edited by loqs (2020-01-31 13:44:27)
Offline
For what it's worth, I report this article talking about this very issue: https://linuxreviews.org/Linux_Kernel_5 … nt_Kernels
Offline
Wow ! Sad ... Last time for Intel chip from my side. Time to switch to AMD
Last edited by kikislater (2020-02-02 09:57:59)
Offline
Wow ! Sad ... Last time for Intel chip from my side. Time to switch to AMD
I've been suffering with intel related bugs for 5 years now. My previous notebook had the CPU hang issue, this one the GPU. I'm really looking towards AMD as well.
Offline
Guys, once again, a lot of people are subscribed to this thread to get updates on the actual problem & solution. Please refrain from inane offtopic comments so we don't get notifications about them.
It is better to keep your mouth shut and be thought a fool than to open it and remove all doubt. (Mark Twain)
Offline
A couple of things that I observed:
1) 100% of crashes happen while I am browsing with Qutebrowser; it seems many people are crashing while using Chromium. As far as I understand, these share some aspects of a rendering engine (QtWebEngine?). Maybe there is a connection? Of course, this can be just misleading: I spend 90% of my computer time on Qutebrowser or typing things
This is very similar to my experience. Vast majority of crashes / hangs happen when using Chromium, although I may have had one or two instances when using Slack. Like you, I spend 90% of my time in these two applications so this may be misleading. However hangs very often seemed to be a direct result of an action in Chromium - clicking on a link, opening new tab, switching tabs, etc. Finally I got tired of forced reboots twice a day and I switched to using Firefox some 7-10 days ago. I'm still on kernel 5.4.15 but I haven't had any hangs since the switch. Don't know if this is a coincidence, dumb luck, or whatever else, but I'll take it.
Offline
This is very similar to my experience. Vast majority of crashes / hangs happen when using Chromium, although I may have had one or two instances when using Slack. Like you, I spend 90% of my time in these two applications so this may be misleading. However hangs very often seemed to be a direct result of an action in Chromium - clicking on a link, opening new tab, switching tabs, etc. Finally I got tired of forced reboots twice a day and I switched to using Firefox some 7-10 days ago. I'm still on kernel 5.4.15 but I haven't had any hangs since the switch. Don't know if this is a coincidence, dumb luck, or whatever else, but I'll take it.
I use Firefox and I also experience the crashes. But I think it's also relevant that I have layers.acceleration.force-enabled=true (ie, hardware acceleration enabled).
Offline