|Summary:||Screen rapidly blinks on and off (UHD Graphics 605)|
|Product:||DRI||Reporter:||Tristan Miller <psychonaut>|
|Component:||DRM/Intel||Assignee:||Intel GFX Bugs mailing list <intel-gfx-bugs>|
|Status:||RESOLVED WORKSFORME||QA Contact:||Intel GFX Bugs mailing list <intel-gfx-bugs>|
|Priority:||not set||CC:||intel-gfx-bugs, psychonaut|
|i915 platform:||GLK||i915 features:||display/eDP|
Description Tristan Miller 2019-11-12 06:57:49 UTC
Created attachment 145936 [details] Video showing bug Summary: On an Acer TravelMate B118 with UHD Graphics 605, the screen rapidly blinks on and off when performing certain activities. System description: I have a new Acer TravelMate B118-M-P98A laptop (Acer article number NX.VHPEG.005) which has an Intel Pentium Silver N5000 and an integrated UHD Graphics 605 video controller. I am running openSUSE Leap 15.1 (x86_64) with its official 4.12.14-lp151.28.25-default kernel. Problem description and how to reproduce: The screen frequently blinks on and off in rapid succession, usually two to eight times in a row. (The screen isn't merely turning black; it looks like it is actually powering off and back on again.) The problem is reproducible in that certain activities consistently trigger the flashing. These include, but are not limited to, the following: * Booting openSUSE 15.1, shortly before the display manager shows the login prompt. * Using the Fn+Up or Fn+Dn keys to adjust the system volume (which also triggers a visual OSD) * Having a system notification appear in KDE Plasma * Booting or shutting down Windows 7 in a VirtualBox * When running Windows 7 in a VirtualBox, switching between folders in the Windows Explorer Once the blinking problem occurs, it can't be triggered again for a few seconds. I am attaching a video showing the blinking problem. In the video, I am in a KDE Plasma session and have just pressed one of the volume control hotkeys (Fn+Up or Fn+Dn), which has triggered the blinking. Diagnostics: Running dmesg reports a bunch of "intel_display_power_put+0xd3/0x100" warnings. These seem to correspond to the instances of blinking: [ 18.703093] WARNING: CPU: 2 PID: 437 at ../drivers/gpu/drm/i915/intel_runtime_pm.c:1692 intel_display_power_put+0xf1/0x100 [i915] [ 18.703308] WARNING: CPU: 2 PID: 437 at ../drivers/gpu/drm/i915/intel_runtime_pm.c:186 intel_display_power_put+0xd3/0x100 [i915] [ 21.724409] WARNING: CPU: 2 PID: 437 at ../drivers/gpu/drm/i915/intel_runtime_pm.c:186 intel_display_power_put+0xd3/0x100 [i915] [ 30.884626] WARNING: CPU: 1 PID: 278 at ../drivers/gpu/drm/i915/intel_runtime_pm.c:186 intel_display_power_put+0xd3/0x100 [i915] [ 33.892611] WARNING: CPU: 3 PID: 276 at ../drivers/gpu/drm/i915/intel_runtime_pm.c:186 intel_display_power_put+0xd3/0x100 [i915] [ 56.932533] WARNING: CPU: 3 PID: 276 at ../drivers/gpu/drm/i915/intel_runtime_pm.c:186 intel_display_power_put+0xd3/0x100 [i915] Attached is a full dmesg with debugging enabled. The dmesg command was run immediately after the system booted up (which, as I mentioned, itself causes the screen to blink). Also attached, in case they are necessary, are outputs of glxinfo and xrandr. After some web searching for similar problems, I have tried booting with the kernel parameters i915.enable_psr=0 or i915.enable_fbc=0. However, neither of these solved the problem.
Comment 1 Tristan Miller 2019-11-12 06:58:29 UTC
Created attachment 145937 [details] Output of dmesg (with debugging) after bootup
Comment 2 Tristan Miller 2019-11-12 06:58:56 UTC
Created attachment 145938 [details] Output of glxinfo
Comment 3 Tristan Miller 2019-11-12 06:59:18 UTC
Created attachment 145939 [details] Output of xrandr
Comment 4 Tristan Miller 2019-11-12 08:29:33 UTC
I forgot to mention that my system uses the following drivers, installed from the official openSUSE 15.1 Leap repositories: libdrm_intel1-2.4.97-lp188.8.131.52.x86_64 libdrm_intel1-32bit-2.4.97-lp184.108.40.206.x86_64
Comment 5 Lakshmi 2019-11-12 13:20:14 UTC
I would recommend to verify the issue with drmtip (https://cgit.freedesktop.org/drm-tip) with kernel parameters drm.debug=0x1e log_buf_len=4M. If the problem persists attach the full dmesg from boot. At the moment the latest drmtip is 5.4.0-rc6+. Attached log is from 4.12.14 which is quiet old.
Comment 6 Tristan Miller 2019-11-15 20:25:29 UTC
(In reply to Lakshmi from comment #5) > I would recommend to verify the issue with drmtip > (https://cgit.freedesktop.org/drm-tip) with kernel parameters drm.debug=0x1e > log_buf_len=4M. If the problem persists attach the full dmesg from boot. > At the moment the latest drmtip is 5.4.0-rc6+. > > Attached log is from 4.12.14 which is quiet old. I'm not really au fait with building my own kernels, so I used the pre-built 5.4.rc7 kernel provided by the openSUSE project at <http://download.opensuse.org/repositories/Kernel:/linux-next/standard/x86_64/>. Supposedly this version includes the latest (or at least a much more recent) version of the i915 driver than the stock 4.12.14 kernel that currently ships with openSUSE 15.1. With this newer kernel package, I'm not able to reproduce the problem. There's no blinking and moreover the whole system runs noticeably faster. So I suppose this means that whatever issue I was experiencing was fixed between 4.12.14 and 5.4.rc7. Is there any easy way of finding out what specific commit fixed the issue? If so I might be able to convince the openSUSE kernel maintainers to apply a patch to the older version of the kernel they're currently distributing.
Comment 7 Tristan Miller 2019-11-15 20:50:15 UTC
Downstream bug report: https://bugzilla.opensuse.org/show_bug.cgi?id=1156928
Comment 8 Lakshmi 2019-11-18 12:26:32 UTC
Thanks for verifying the issue with latest kernel. We always recommned to use the latest kernel to the users. > Is there any easy way of finding out what > specific commit fixed the issue? If so I might be able to convince the > openSUSE kernel maintainers to apply a patch to the older version of the > kernel they're currently distributing. Unless we biset the issue, I am not aware of any other way to find the commit that fixed the issue. Or even multiple commits might helped the issue. Not sure though. Since this issue is not reproducible with latest kernel. I am closing this issue as WORKSFORME. Thanks!