Created attachment 106568 [details] Restoration of one hunk from fb51ec4 I have Lenovo ThinkPad X201 tablet with Intel Core i7 L-620 (Intel HD Graphics gen 5). Somewhere around version 2.99.912 of xf86-video-intel I got a strange bug: when I scroll page in browser the refreshing of screen is delayed or when I type text the symbols appear not after each one but after every second symbol. I found that problem starts after commit: commit fb51ec44f4b6069ab854233d5d0da34eb79f754c Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Mon Jun 23 09:07:13 2014 +0100 sna: Rename IGNORE_CPU to IGNORE_DAMAGE to better reflect its purpose Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> I found that if I restore only one hunk (4th hunk in file «src/sna/sna_accel.c») from this commit the bug disappears. By this hunk the one line in «src/sna/sna_accel.c» was changed but not just renamed: from: (flags & REPLACES || !priv->gpu_damage..... to: (flags & (REPLACES | IGNORE_DAMAGE) || !priv->gpu_damage..... But according to bug #79993 this maybe not a root cause. But I don't know why my patch works... Please see video here: http://youtu.be/JjAScg4MZNk On «test 1» and «test 2» I use clean driver from git and symbols printed with delay. On «test 3» and «test 4» I use attached patch and symbols printed normally.
Can you please also attach your Xorg.0.log? What's the PageFlip setting you are changing there?
Created attachment 106582 [details] Patch not applied and no «"PageFlip" "false"» in xorg.conf. Patch not applied and no «"PageFlip" "false"» in xorg.conf. $ cat /etc/X11/xorg.conf.d/10-intel.conf | grepv Section "Device" Identifier "Intel HD" Driver "intel" Option "FallbackDebug" "true" Option "AccelMethod" "sna" EndSection This is for «test 1».
Created attachment 106583 [details] Patch not applied and «"PageFlip" "false"» in xorg.conf. Patch not applied and «"PageFlip" "false"» in xorg.conf. $ cat /etc/X11/xorg.conf.d/10-intel.conf | grepv Section "Device" Identifier "Intel HD" Driver "intel" Option "FallbackDebug" "true" Option "PageFlip" "false" Option "AccelMethod" "sna" EndSection This is for «test 2».
Created attachment 106584 [details] Patch applied and no «"PageFlip" "false"» in xorg.conf. Patch applied and no «"PageFlip" "false"» in xorg.conf. $ cat /etc/X11/xorg.conf.d/10-intel.conf | grepv Section "Device" Identifier "Intel HD" Driver "intel" Option "FallbackDebug" "true" Option "AccelMethod" "sna" EndSection This is for «test 3».
Created attachment 106585 [details] Patch applied and «"PageFlip" "false"» in xorg.conf. Patch applied and «"PageFlip" "false"» in xorg.conf. $ cat /etc/X11/xorg.conf.d/10-intel.conf | grepv Section "Device" Identifier "Intel HD" Driver "intel" Option "FallbackDebug" "true" Option "PageFlip" "false" Option "AccelMethod" "sna" EndSection This is for «test 4».
Is KDE using DRI3? Could you try Option "DRI" "2" to check?
Created attachment 106598 [details] KDE4 kwin settings KDE uses «OpenGL 2.0» instead of «XRender» for compositing. Qt uses «raster» engine instead of «native». But I don't know how and where to see which DRI type KDE uses...
Created attachment 106599 [details] DRI2. Patch not applied and no «"PageFlip" "false"» in xorg.conf. With «Option DRI 2» the both problems has gone. $ cat /etc/X11/xorg.conf.d/10-intel.conf | grepv Section "Device" Identifier "Intel HD" Driver "intel" Option "DRI" "2" Option "FallbackDebug" "true" Option "AccelMethod" "sna" EndSection Does it mean that DRI3 support is not so good for daily use?
The delayed rendering then makes sense. Currently DRI3 implements fences in only one direction, leaving the question of how to sync rendering to a texture (i.e. the compositor reading pixmaps rendered by X) completely unanswered.
*** This bug has been marked as a duplicate of bug 81551 ***
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.