Bug 96533 - Rendering glitches in tk applications with psr enabled on Thinkpad X1 Carbon 2nd gen (Haswell)
Summary: Rendering glitches in tk applications with psr enabled on Thinkpad X1 Carbon ...
Status: CLOSED INVALID
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: unspecified
Hardware: Other All
: medium normal
Assignee: Intel GFX Bugs mailing list
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2016-06-14 23:51 UTC by Benjamin Poirier
Modified: 2016-10-03 11:44 UTC (History)
1 user (show)

See Also:
i915 platform: HSW
i915 features: display/PSR


Attachments
dmesg with drm debug messages (131.58 KB, text/plain)
2016-06-14 23:51 UTC, Benjamin Poirier
no flags Details

Description Benjamin Poirier 2016-06-14 23:51:42 UTC
Created attachment 124539 [details]
dmesg with drm debug messages

Since updating to 4.6 I noticed rendering glitches in tk applications. I
bisected the issue to
9b58e35 drm/i915: Enable PSR by default on Haswell and Broadwell. (v4.6-rc1)

One way to reproduce the issue is to run
	git gui blame --line=4132 v4.6 net/core/dev.c
in a linux kernel git tree,
Click on line 4132
Scroll up and down in the bottom window (the commit log) using the mouse
wheel. Notice that blank lines are not properly refreshed and there is ghost
content from what was there before scrolling. Moving the window fixes the
render.
If you don't have a mouse wheel available ATM, click on lines 4131 and
4132 in alternance and notice that the color highlighting in the
column that has line numbers is not properly updated (if you move the
window, it is updated).

I noticed similar issues in gitk.

I've further tested with 5fb50cc from drm-intel-nightly and the problem
remains.
i915.enable_psr=0 works around the issue
i915.enable_psr=2 and =3 do not work around the issue

root@f1:~# cat /sys/kernel/debug/dri/0/i915_edp_psr_status
Sink_Support: no
Source_OK: no
Enabled: no
Active: no
Busy frontbuffer bits: 0x000
Re-enable work scheduled: no
Main link in standby mode: no
HW Enabled & Active bit: no
Performance_Counter: 0

The platform is
	Lenovo ThinkPad X1 Carbon 2nd, model 20A7CTO1WW
I can see the problem both on the laptop screen or with external monitors.
I am running openbox on Xorg, distro is openSUSE Tumbleweed.
I tried running kms_psr_sink_crc from intel-gpu-tools but it doesn't seem like
it was able to complete:
	IGT-Version: 1.15-NOT-GIT (x86_64) (Linux: 4.6.0-1-default x86_64)
	Test requirement not met in function main, file kms_psr_sink_crc.c:522:
	Test requirement: !(!psr_possible(&data))
	Last errno: 2, No such file or directory
	Subtest psr_basic: SKIP
	Subtest primary_page_flip: SKIP
	Subtest primary_mmap_gtt: SKIP
	Subtest primary_mmap_gtt_waiting: SKIP
	Subtest primary_mmap_cpu: SKIP
	Subtest primary_blt: SKIP
	Subtest primary_render: SKIP
	Subtest sprite_mmap_gtt: SKIP
	Subtest sprite_mmap_gtt_waiting: SKIP
	Subtest sprite_mmap_cpu: SKIP
	Subtest sprite_blt: SKIP
	Subtest sprite_render: SKIP
	Subtest sprite_plane_move: SKIP
	Subtest sprite_plane_onoff: SKIP
	Subtest cursor_mmap_gtt: SKIP
	Subtest cursor_mmap_gtt_waiting: SKIP
	Subtest cursor_mmap_cpu: SKIP
	Subtest cursor_blt: SKIP
	Subtest cursor_render: SKIP
	Subtest cursor_plane_move: SKIP
	Subtest cursor_plane_onoff: SKIP
	Subtest dpms_off_psr_active: SKIP
	Subtest dpms_off_psr_exit: SKIP
	Subtest suspend_psr_active: SKIP
	Subtest suspend_psr_exit: SKIP

Please let me know if you need more info.
Comment 1 Benjamin Poirier 2016-07-13 15:47:57 UTC
FYI, I also notice problems with Firefox, okular and pinentry. If I scroll using the keyboard or type some text, there is a noticeable delay before the window content is refreshed. As expected, moving the mouse continuously works around the issue. Moreover, this only happens on certain display outputs (ie. eDP1, HDMI1, HDMI2) but it is not consistent which one. I can also move the window from one display to another and then the problem is not apparent.
Comment 2 Benjamin Poirier 2016-07-14 18:57:05 UTC
Upon closer inspection, the second set of problems, described in comment 1, was a different problem, bug 96464.
Comment 3 Benjamin Poirier 2016-07-14 18:59:33 UTC
I can no longer reproduce the original issue with git gui/gitk. Beats me as to why since I could reproduce it consistently when bisecting.
Comment 4 Jari Tahvanainen 2016-10-03 11:44:10 UTC
Closing.


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.