Bug 99087

Summary:

Screen flickers - probably double buffered. [regresion]

Product:

DRI

Reporter:

kolorafa <freedesktop_org>

Component:

DRM/Intel

Assignee:

Intel GFX Bugs mailing list <intel-gfx-bugs>

Status:

CLOSED FIXED

QA Contact:

Intel GFX Bugs mailing list <intel-gfx-bugs>

Severity:

normal

Priority:

medium

CC:

intel-gfx-bugs, przanoni

Version:

XOrg git

Hardware:

x86-64 (AMD64)

OS:

Linux (All)

Whiteboard:

i915 platform:

BDW

i915 features:

display/PSR

Attachments:

Description	Flags
dmesg with flickering	none
dmesg without flickering - old lts kernel	none

Description kolorafa 2016-12-14 18:59:21 UTC

Redirected from: https://bugzilla.kernel.org/show_bug.cgi?id=187251

Screen flicker so much that its hard to watch.

Looks like intel buffers do not set correctly and flickers between old and new image.

OS: Arch linux
Problem exists for at least 3 months now.
Hardware: MSI GE72 2QC Apache
00:02.0 VGA compatible controller: Intel Corporation HD Graphics 5600 (rev 0a)
01:00.0 3D controller: NVIDIA Corporation GM107M [GeForce GTX 960M] (rev a2)
00:00.0 Host bridge: Intel Corporation Broadwell-U Host Bridge - DMI (rev 0a)
00:01.0 PCI bridge: Intel Corporation Broadwell-U PCI Express x16 Controller (rev 0a)


Running on intel drivers, on nvidia efect is the same because it's a hybrid so intel still handle final render.

https://dl.dropboxusercontent.com/u/54627341/flicker/MOV_0006.mp4
You see that mostly on non-black parts.

My first guess was that it is backlight pwm problem, i did try that, but then i saw some strange "ghosts" on screen, for example when image change, you still see flickering old image:

https://dl.dropboxusercontent.com/u/54627341/flicker/MOV_20161107115931_TSV.mp4


And it is horrible:
https://dl.dropboxusercontent.com/u/54627341/flicker/MOV_0009.mp4
https://dl.dropboxusercontent.com/u/54627341/flicker/MOV_20161107212024_TSV.mp4


And the flicker speed and intensity change depending on laptop load.
It only flickers on laptop screen, external monitors normal.


Also it shows at boot 2 times: 
[drm:intel_dp_start_link_train [i915]] *ERROR* failed to update link training 
right before it starts to flicker

Comment 1 Paulo Zanoni 2016-12-14 20:51:57 UTC

Please boot with drm.debug=0xe log_buf_len=1M, reproduce the problem, and attach the dmesg output here.

Thanks for the bug report.

Comment 2 kolorafa 2016-12-15 08:06:07 UTC

Created attachment 128478 [details]
dmesg with flickering

Comment 3 kolorafa 2016-12-15 08:06:40 UTC

Created attachment 128479 [details]
dmesg without flickering - old lts kernel

Comment 4 kolorafa 2016-12-15 08:10:03 UTC

I have 2 external screens connected, but it flicker etherway

Comment 5 Paulo Zanoni 2016-12-15 11:33:55 UTC

Does the problem go away if you boot with i915.enable_psr=0 ?

We have a patch to disable it by default on BDW and it's on its way to the stable Kernels. If that's the case, we can just close the bug as fixed since it will eventually hit the distros.

Comment 6 Paulo Zanoni 2016-12-15 11:35:28 UTC

Note: Bug Assignee is the person in charge of solving the bug. I'm setting this back to our team. But if you really want to solve it, feel free to tell us :).

Comment 7 kolorafa 2016-12-15 12:25:12 UTC

i915.enable_psr=0 resolves the problem

screen doesn't flicker, but the error message at boot still present (so it's not related)

(When i add status should i change status? If yes - to what?)

Thanks

Comment 8 Paulo Zanoni 2016-12-15 15:47:14 UTC

(In reply to kolorafa from comment #7)
> i915.enable_psr=0 resolves the problem

Good :).

> 
> screen doesn't flicker, but the error message at boot still present (so it's
> not related)

Yeah, that message is still widely known, but usually harmless. Still needs to be fixed.

> 
> (When i add status should i change status? If yes - to what?)

Usually the "protocol" is that when I ask questions I put it as NEEDINFO, but then when you provide answers you can change back to NEW (or we can do it later too, no problem). But in this case I'll just close the bug since the fix is already on its way to the stable trees.

The following commit should fix it:

commit 2ee7dc497e348eecbb82adbb1ea9e9a7e29fe921
    drm/i915: disable PSR by default on HSW/BDW

Thanks for the bug report!

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.