Bug#1025213: Mutter issue upstream 2602

[Gert van de Kraats]

Sorry, previous message has wrong link.

Mutter issue upstream opened 
athttps://gitlab.gnome.org/GNOME/mutter/-/issues/2602 .

Bug#1025213: Mutter mr 2241

[Gert van de Kraats]

Problem is caused by mutter merge-request 2241:
https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2241 .

Shadow buffer is only used at software rendering.
Damage history is needed at copy from shadow buffer to onscreen,
to synchronize buffers.
I will open an issue upstream.
"

Bug#1025213: mutter

[Gert van de Kraats]

Thank you for adding gallium i915 to mesa 22.3.2-1. Now at least the
default Debian testing release is usable at my laptop.

But the  problem stills persists, if the i915-driver is removed and
kms_swrast is used.

In fact it is a mutter-problem (43.2-4) with shadow buffering.
Problem can be best viewed by opening a Terminal-window at main
Gnome-Shell window.
After some fast flickering the screen is updated every second which is
driven by the blinking of the cursor at the terminal.
Clearly can be seen the current terminal-line is correctly updated with
blinking cursor, but with 3 different old backgrounds.
If the next minute is started the time at the top is switching between 
previous and current time.


With environment variables MUTTER_DEBUG=kms and COGL_DEBUG=clipping this
can be monitored.

The syslog shows messages
gnome-shell[6588]: Device '/dev/dri/card0' prefers shadow buffer
gnome-shell[6588]: Initialized single buffered shadow fb for VGA-1
gnome-shell[6588]: Initialized single buffered shadow fb for LVDS-1

Shadow buffering is only used at software rendering!
The problem disappears if shadow buffering is disabled at
src/backends/native/meta-kms-impl-device.c at
function should_force_shadow_fb:
return FALSE;

Function drmGetCap is called by mutter to check if the device
prefers shadow buffering.