<div style="line-height:1.7;color:#000000;font-size:14px;font-family:arial"> hi,all I modify the standard driver of fbdev,and add the xv adaptor. So in the fbdev driver ,there are shadow and xv at the same time. Then I found a problem: When I move the video window(which use the xvimagesink),the picture in the window is not the video frame but the data from the shadow buffer.At the same time,When I move the mouse in the video window ,I can see a block data from the shadow buffer instead of the video data. I think it's too bad for the user. How can I improve the experience when I use both shadow and xvideo? If I don't use shadow,I must make the FB cacheable to ensure the result of x11perf is not too bad.But there are lots of stripes ,when I make the FB cacheable. So,I think maybe I could add a cache flush ioctl at the point when the xorg complete a drawing action.I think if i do so,the stripes will be greatly reduced. But where I should add the cache flush ioctl in the source code of xorg? Any advice is welcome! Thanks! Best Regards, David </div>