[EMAIL PROTECTED] (Michal Maruška) writes: > Hello, > I know about 2 bugs in X server which are quite reproducible with my > programs. > .... > 1/ There is a way to see a screen full of pixmaps (instead of what should be > on > the screen). > > I can reproduce this bug with sawfish WM. > ..... > When I do such movement w/ keyboard using mouse emulation, I can trigger > this funny > bug.
I think I found the root of this problem. It's independent of the video driver, just due to a bug in XKB code for "mouse emulation". XkbDDXFakePointerMotion in xc/programs/Xserver/xkb/ddxFakeMtn.c does not check if the (new pointer) position is outside screen limits (at least in my case: xinerama compiled in, but not enabled). ... so it calls (*pScreen->SetCursorPosition)(pScreen, oldX, oldY, TRUE); with wrong oldX,oldY. OTOH xf86PostMotionEvent calls miPointerAbsoluteCursor which does the checks. So, XkbDDXFakePointerMotion should either do the checks or (preferebly) delegate to some mi* function, but it will take me some time before I understand which one. Testing with the "nv" driver I can see some funny little screen shift when I move the pointer (by keyboard) against an edge of the screen. _______________________________________________ Devel mailing list Devel@XFree86.Org http://XFree86.Org/mailman/listinfo/devel
