Red Hat Bugzilla – Bug 82982
xrandr resolution changes require VC flip to activate completely
Last modified: 2007-04-18 12:50:26 EDT
Playing with xrandr on my beat-up old laptop (Trident Cyberblade card, lcd
hard-locked at 1024x768 resolution only)
I start out in 1024x768x75 and then see what else is available:
[kaboom@skuld kaboom]$ xrandr -q
SZ: Pixels Physical Refresh
*0 1024 x 768 ( 347mm x 260mm ) *75 70 60
1 800 x 600 ( 347mm x 260mm ) 75 72 60 56
2 640 x 480 ( 347mm x 260mm ) 75 72 60
3 832 x 624 ( 347mm x 260mm ) 74
Current rotation - normal
Current reflection - none
Rotations possible - normal
Reflections possible - none
So then I switch to a lower resolution (xrandr -s 1). It correctly shrinks.
Since my lcd is hardlocked at 1024x768, it just displays as an 800x600 window w/
a 1024x768 border (as expected). It doesn't center the 800x600 w/in the 1024x768
frame though -- instead it anchers the 800x600 display at (0,0). No big deal,
just different than what I might have expected / what usually happens when
running lower res on a fixed res lcd....
At any rate, I then switch back (xrandr -s 0). It appears to go back to
1024x768, but the desktop is really still only 800x600. The last 224 pixels
(from 576 to 800) are simply copied and displayed again as pixels 801 - 1024.
For example, if I move my mouse to about (700,10), it shows up as (700,10) and
as (924,10). Ditto for dialog boxes, stuff in the gnome panel at the bottom of
the screen, etc.
To make things weirder:
* its just the horizontal stretching that's screwed. Vertically, everything's
fine (going from 600 to 768 actually did go to 768, rather than copying the
bottom 168 pixels and displaying them twice)
* I tried using the gimp to do a screen capture of the effect. In the screen
capture, everything's completely normal
If I flip to a text VC and then back to X, everything straightens out, and I get
real 1024 resolution
Created attachment 89674 [details]
Created attachment 89675 [details]
see this on beta5 as well
This problem is probably best to be reported to XFree86.org so the driver
maintainer can investigate it. The URL is http://bugs.xfree86.org if you'd
like to do that. It'll increase the chances of a faster solution.
Hope this helps.
Filed upstream as Bug 197, and removed beta-only bit so upstream could read this
The link above is broken. Upstream bug is:
Resolving as UPSTREAM. Will consider backport of any upstream fixes if and
when they become available.
New upstream URL: http://bugs.xfree86.org/show_bug.cgi?id=197
The upstream report indicates this was fixed in XFree86 CVS in late
2003. X.Org X11 forked the XFree86 tree just prior to the 4.4.0
XFree86 release, so the X.Org sources should contain this fix as
Closing as "RAWHIDE"