Bug 96380 - Rreverse prime (--setprovideroutput modesetting Intel) produces a corrupted cursor and image
Summary: Rreverse prime (--setprovideroutput modesetting Intel) produces a corrupted c...
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: unspecified
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: Chris Wilson
QA Contact: Intel GFX Bugs mailing list
Depends on:
Reported: 2016-06-04 23:33 UTC by Peter Wu
Modified: 2016-09-17 07:10 UTC (History)
0 users

See Also:
i915 platform:
i915 features:

Picture showing cursor and other corruption (202.28 KB, image/jpeg)
2016-06-04 23:33 UTC, Peter Wu
no flags Details

Description Peter Wu 2016-06-04 23:33:41 UTC
Created attachment 124319 [details]
Picture showing cursor and other corruption

An external screen connected via reverse PRIME works (has a picture), except that the cursor is corrupted in a way such that accurating pointing to text and menus is often impossible. Text in Konsole (KDE) also corrupted. This happens only on the external screen, hovering over the laptop screen is fine.

Attached is a picture showing cursor corruption and issues with a menu in KDE Plasma (I moved from right to left and opened a nested menu by hovering). The cursor trace and corrupted text while typing was also reproduced on a bare X server, so it is not limited to composited desktops.

$ xrandr --setprovideroutputsource modesetting Intel
$ xrandr --output DP-1 --mode 2560x1440 --left-of eDP1
$ xrandr --listproviders
Providers: number : 2
Provider 0: id: 0xcd cap: 0xb, Source Output, Sink Output, Sink Offload crtcs: 3 outputs: 1 associated providers: 1 name:Intel
Provider 1: id: 0x46 cap: 0x3, Source Output, Sink Output crtcs: 4 outputs: 3 associated providers: 1 name:modesetting

# Xorg :0 -nolisten tcp -noreset -verbose 2
# DISPLAY=:0 xterm

- Distribution: Arch Linux
- Linux v4.7-rc1 (also reproduced with v4.6)
- plasma-workspace 5.6.4-1
- xorg-server 1.18.3-1
- xf86-video-intel 2.99.917-662-gb617f80 (tested 2.99.917-654-ga508b11 too)

- i7-6700HQ (eDP1, output source, intel, i915)
- Nvidia GTX 965M (DP-1, output sink, modesetting, nouveau)

Further information:
with modesetting for both the Nvidia and Intel devices, there is no corruption. I do have to set vblank_mode=0, otherwise the external screen refreshes windows with 1fps when glxgears is displayed on just the external screen. If the window is just partially displayed on eDP-1, the frame rate is "normal" again.

Also tried using UXA instead of SNA but that is even worse. When switching tabs / opening a new page in Firefox for example, the left bottom triangle shows the previous content that was supposed to be shown while the right top triangle shows the currently expected picture (if you are lucky). Similar for Konsole terminal output.

Here is another bug about corruption, maybe it is the same issue?
Comment 1 Chris Wilson 2016-09-16 21:59:15 UTC
The kernel bugs have been fixed, quite a few modesetting bugs remain.

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.