| Summary: | Blur & contrast shaders cause flicker on intel i915 | ||
|---|---|---|---|
| Product: | [Plasma] kwin | Reporter: | Pierre Ducroquet <pinaraf> |
| Component: | effects-various | Assignee: | KWin default assignee <kwin-bugs-null> |
| Status: | RESOLVED WORKSFORME | ||
| Severity: | normal | CC: | arojas |
| Priority: | NOR | Flags: | thomas.luebking:
Intel+
|
| Version First Reported In: | 5.3.0 | ||
| Target Milestone: | --- | ||
| Platform: | Fedora RPMs | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed/Implemented In: | ||
| Sentry Crash Report: | |||
|
Description
Pierre Ducroquet
2015-05-30 14:26:15 UTC
Smells like the blur shader being the ultimate trigger. Could you add/override [Effect-Logout] UseBlur=false in ~/.config/kwinrc and disable the blur (and likely the "contrast") effect in the dialog spawned by "kcmshell5 kwineffects" and check the result? Indeed, this "fixes" the issue. And the contrast effect creates flickering when switching between windows using alt-tab. By the way, if you need the OpenGL renderer and version strings to identify the GPU : renderer : Mesa DRI Intel(R) 945GME version : 2.1 Mesa 10.5.4 For a wild shot, re-enable the effect(s) and run export KWIN_USE_INTEL_SWAP_EVENT=0 # should be disabled anyway export KWIN_USE_BUFFER_AGE=0 kwin_x11 --replace & Does this also happen with the EGL backend? This has gotten much worse with linux 4.2. With linux <4.2 this happened for me after almost every restart, but restarting kwin was enough to fix it for the entire session. Now restarting kwin only fixes it for a shot time (less than 1 minute). Still happens with export KWIN_USE_INTEL_SWAP_EVENT=0 export KWIN_USE_BUFFER_AGE=0 Try https://wiki.archlinux.org/index.php/Intel_graphics#SNA_issues Also please attach the output of "qdbus org.kde.KWin /KWin supportInformation" This very bug was caused by the blur shader (in the logout effect) which doesn't really fit your description - however, I assume this is due to the blur & contrast shaders? Works fine for me. If you're still able to reproduce the bug, please reopen this bug report. |