Bug 104599 - corrupted desktop graphics with latest git intel driver
Summary: corrupted desktop graphics with latest git intel driver
Status: RESOLVED MOVED
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i915 (show other bugs)
Version: git
Hardware: Other All
: medium normal
Assignee: Default DRI bug account
QA Contact: Default DRI bug account
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2018-01-12 12:03 UTC by zebulon
Modified: 2019-09-18 19:40 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments
photo of corrupted graphic on screen (6.58 MB, image/jpeg)
2018-01-12 12:03 UTC, zebulon
Details
uncorrupted desktop (747.41 KB, image/png)
2018-01-12 12:08 UTC, zebulon
Details

Description zebulon 2018-01-12 12:03:20 UTC
Created attachment 136678 [details]
photo of corrupted graphic on screen

I have ubuntu 17.10 with an intel i3-3217U cpu/gpu and using the oibaf PPA repo.

With latest following package:
libgles2-mesa:amd64 (17.4~git1801070730.be144e~oibaf~a
xserver-xorg-video-nouveau:amd64 (1:1.0.15+git1712110734.048baf~oibaf~a
libdrm-nouveau2:amd64 (2.4.89+git1712180630.831036~oibaf~a
libdrm-nouveau2:i386 (2.4.89+git1712180630.831036~oibaf~a
xserver-xorg-video-amdgpu:amd64 (1.4.99+git1712271934.69e208~oibaf~a
libegl1-mesa-dev:amd64 (17.4~git1801070730.be144e~oibaf~a
libglapi-mesa:amd64 (17.4~git1801070730.be144e~oibaf~a
libglapi-mesa:i386 (17.4~git1801070730.be144e~oibaf~a
libxatracker2:amd64 (17.4~git1801070730.be144e~oibaf~a
libegl1-mesa:amd64 (17.4~git1801070730.be144e~oibaf~a
libgbm1:amd64 (17.4~git1801070730.be144e~oibaf~a
libdrm-amdgpu1:amd64 (2.4.89+git1712180630.831036~oibaf~a
libdrm-amdgpu1:i386 (2.4.89+git1712180630.831036~oibaf~a
libllvm5.0:amd64 (1:5.0.1-1~oibaf~a
libllvm5.0:i386 (1:5.0.1-1~oibaf~a
xserver-xorg-video-intel:amd64 (2:2.99.917+git1712201935.708255~oibaf~a
libwayland-egl1-mesa:amd64 (17.4~git1801070730.be144e~oibaf~a
libdrm2:amd64 (2.4.89+git1712180630.831036~oibaf~a
libdrm2:i386 (2.4.89+git1712180630.831036~oibaf~a
xserver-xorg-video-ati:amd64 (1:7.10.99+git1712281934.1fe8ca~oibaf~a
libgl1-mesa-dri:amd64 (17.4~git1801070730.be144e~oibaf~a
libgl1-mesa-dri:i386 (17.4~git1801070730.be144e~oibaf~a
libosmesa6:amd64 (17.4~git1801070730.be144e~oibaf~a
libosmesa6:i386 (17.4~git1801070730.be144e~oibaf~a
libgl1-mesa-glx:amd64 (17.4~git1801070730.be144e~oibaf~a
libgl1-mesa-glx:i386 (17.4~git1801070730.be144e~oibaf~a
xserver-xorg-video-radeon:amd64 (1:7.10.99+git1712281934.1fe8ca~oibaf~a
libdrm-intel1:amd64 (2.4.89+git1712180630.831036~oibaf~a
libdrm-intel1:i386 (2.4.89+git1712180630.831036~oibaf~a
libdrm-radeon1:amd64 (2.4.89+git1712180630.831036~oibaf~a
libdrm-radeon1:i386 (2.4.89+git1712180630.831036~oibaf~a
mesa-vdpau-drivers:amd64 (17.4~git1801070730.be144e~oibaf~a
mesa-vdpau-drivers:i386 (17.4~git1801070730.be144e~oibaf~a
libdrm-dev:amd64 (2.4.89+git1712180630.831036~oibaf~a
mesa-va-drivers:i386 (17.4~git1801070730.be144e~oibaf~a
libdrm-common:amd64 (2.4.89+git1712180630.831036~oibaf~a

I have a desktop screen corrupted like in the photo (screenshot did not wworked)

Windows control did not work (wayland related ?)
Comment 1 zebulon 2018-01-12 12:08:33 UTC
Created attachment 136679 [details]
uncorrupted desktop

To give you a better idea of the corruption, here is a screenshot of the desktop in normal circumstances.
Comment 2 Tapani Pälli 2018-01-12 13:01:55 UTC
possible duplicate for bug #104536 .. the symptoms look a lot worse though
Comment 3 Mario Kleiner 2018-01-13 03:28:03 UTC
This is running the new GNOME based Ubuntu GUI with Gnome-Shell Wayland, right?

With "Windows control did not work" you mean it doesn't respond to mouse-clicks?

In that case it is known bugs/limitations in gnome-shell for depth 30/10bpc mode. GNOME Shells picking code can't handle 10 bit per color channel.

Additionally Gnome-Shells Wayland compositor doesn't handle anything but bgrx8888 or bgra8888 framebuffers in its drm/kms backend. It takes whatever framebuffer format it gets, e.g., bgrx1010102 and hands it to the kernel as bgrx8888, so the gpu misinterprets a 10 bit image as a 8 bit image during scanout, which ends badly.
Comment 4 Tapani Pälli 2018-01-15 06:41:54 UTC
(In reply to Mario Kleiner from comment #3)
> This is running the new GNOME based Ubuntu GUI with Gnome-Shell Wayland,
> right?
> 
> With "Windows control did not work" you mean it doesn't respond to
> mouse-clicks?
> 
> In that case it is known bugs/limitations in gnome-shell for depth 30/10bpc
> mode. GNOME Shells picking code can't handle 10 bit per color channel.
> 
> Additionally Gnome-Shells Wayland compositor doesn't handle anything but
> bgrx8888 or bgra8888 framebuffers in its drm/kms backend. It takes whatever
> framebuffer format it gets, e.g., bgrx1010102 and hands it to the kernel as
> bgrx8888, so the gpu misinterprets a 10 bit image as a 8 bit image during
> scanout, which ends badly.

Right, so this is NOTOURBUG. Zebulon, your issue should disappear when you do apt-get update again since allow_rgb10_configs defaults to 'false'.

Mario, do you know if there us existing Gnome bug about this? (did not find such) Maybe easiest workaround for them would be to ignore 10bpc until picking and kms backend is fixed.
Comment 5 zebulon 2018-01-15 12:03:40 UTC
I can't help you anymore. I removed oibaf repo and switch back to original ubuntu driver.
Comment 6 Mario Kleiner 2018-01-16 03:49:00 UTC
(In reply to Tapani Pälli from comment #4)
> (In reply to Mario Kleiner from comment #3)
> > This is running the new GNOME based Ubuntu GUI with Gnome-Shell Wayland,
> > right?
> > 
> > With "Windows control did not work" you mean it doesn't respond to
> > mouse-clicks?
> > 
> > In that case it is known bugs/limitations in gnome-shell for depth 30/10bpc
> > mode. GNOME Shells picking code can't handle 10 bit per color channel.
> > 
> > Additionally Gnome-Shells Wayland compositor doesn't handle anything but
> > bgrx8888 or bgra8888 framebuffers in its drm/kms backend. It takes whatever
> > framebuffer format it gets, e.g., bgrx1010102 and hands it to the kernel as
> > bgrx8888, so the gpu misinterprets a 10 bit image as a 8 bit image during
> > scanout, which ends badly.
> 
> Right, so this is NOTOURBUG. Zebulon, your issue should disappear when you
> do apt-get update again since allow_rgb10_configs defaults to 'false'.
> 
> Mario, do you know if there us existing Gnome bug about this? (did not find
> such) Maybe easiest workaround for them would be to ignore 10bpc until
> picking and kms backend is fixed.

None that i would know of. I haven't filed one so far.
Comment 7 GitLab Migration User 2019-09-18 19:40:49 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/mesa/mesa/issues/781.


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.