Bug 91366 - Vertical Sync broken on Broadwell
Summary: Vertical Sync broken on Broadwell
Status: RESOLVED MOVED
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: unspecified
Hardware: Other All
: medium major
Assignee: Chris Wilson
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2015-07-16 22:27 UTC by Lyude Paul
Modified: 2019-11-27 13:38 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments

Description Lyude Paul 2015-07-16 22:27:31 UTC
Hi! So I've got another bug to report with the Broadwell chipset. Vertical sync seems to be rather broken with Broadwell. I consistently see display tearing around the top of my screen regardless of the conditions. Like the last bug, I've been able to reproduce this with:

- ThinkPad T450s
- ThinkPad X1 carbon (latest revision)
- Chromebook Pixel

All running Fedora 22 and version 2.99.917 of the Intel Xorg driver.
Comment 1 Chris Wilson 2015-07-18 11:13:43 UTC
First question is it even trying to sync the display to the vertical refresh? gnome-shell doesn't by default (only using the vblank as a repaint limiter).
Comment 2 Lyude Paul 2015-07-20 19:47:03 UTC
(In reply to Chris Wilson from comment #1)
> First question is it even trying to sync the display to the vertical
> refresh? gnome-shell doesn't by default (only using the vblank as a repaint
> limiter).

Sorry for the late response. Anyway, how exactly would I check this?
I should also note, the installation on one of the machines I'm using right now (my personal T450s) was literally copied from my T440s where this setup worked perfectly with the Intel drivers, if that helps at all.
Comment 3 James 2015-09-14 17:17:50 UTC
I've got a Broadwell chip with HD5500 and I notice ragged tearing for the top 200 (or so) pixels. This is on a Lenovo C50-30, also F22 with driver 2.99.917. I've a video on RHBZ 1262945 -- is this the same thing?

I find it tears at first login, or after unlocking the screen. Changing to a VT and back helps -- tearing goes away.
Comment 4 James 2015-09-25 21:28:51 UTC
I find booting with i915.enable_ips=0 helps.
Comment 5 Jani Nikula 2015-09-28 08:50:48 UTC
Which kernel version is this? Try latest.
Comment 6 James 2015-09-28 17:40:35 UTC
(In reply to Jani Nikula from comment #5)
> Which kernel version is this? Try latest.

4.1.7-200.fc22.x86_64, the Fedora 22 latest from updates. I'll see if the 4.2.1 RPM from Koji works and how it behaves.

[    1.141888] [drm] Initialized i915 1.6.0 20150327 for 0000:00:02.0 on minor 0
Comment 7 James 2015-09-30 18:06:31 UTC
Still get the tearing with kernel-4.2.2-300.fc23.x86_64 without enable_ips=0.

What's more, with this kernel changing to a VT and back doesn't cure it.

[    1.152465] [drm] Initialized i915 1.6.0 20150522 for 0000:00:02.0 on minor 0
Comment 8 Lyude Paul 2015-10-01 18:24:44 UTC
Alright, tested it with the latest git HEAD. Vertical sync is still just as broken as before. I think I've noticed it actually work every now and then (there doesn't seem to be much of a pattern here)
The kernel option mentioned in the comments here doesn't seem to make any sort of difference in this behavior either. This being said, this is only reproducible under X. The tearing also always happens near the top of the activities bar in a straight line, and doesn't seem to be noticable when 3D acceleration is being used (e.g. if I open up the activities, I see no tearing in the animation, but if I'm typing in a terminal near the top of the screen I'll notice the tearing). If I use xf86-video-modesetting, I can still reproduce tearing but it's quite different. While the intel driver tears at almost a straight line, the modesetting driver tears at a straight line until it almost reaches the right side of the screen, where it suddenly travels 135° downwards for about 200 pixels, then starts going forwards in a straight line up until the end of the screen. The modesetting driver retains this line with the tearing perfectly. In addition, the tearing on the modesetting driver happens regardless of whether or not there's 3D acceleration going on. Running GNOME under Wayland shows no vertical sync issues.
Comment 9 James 2015-10-01 18:39:23 UTC
(In reply to Chandler Paul from comment #8)

> The kernel option mentioned in the comments here doesn't seem to make any
> sort of difference in this behavior either. 

Then we might be seeing different bugs. Sorry for the sidetrack; I'll keep mine over at RHBZ until suggested otherwise.
Comment 10 Martin Peres 2019-11-27 13:38:35 UTC
-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/xorg/driver/xf86-video-intel/issues/61.


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.