I own a Nvidia TNT2 Ultra from Creative Labs and I'm using the XFree86
4.0.1 drivers and when I move the mouse fast, the cursor changes briefly to
a square (blinks). This doesn't happen if I use the XF86_SVGA server.
If I run the XF86_SVGA server (or reboot to Windows and then back into
Linux) and then run X (using the nv.o driver) this problem does not happen
again until I poweroff the machine and then turn it back on (not a reset,
really poweroff). This also doesn't happen if I use the option SoftCursor
It is annoying to have to log in as root, run XF86_SVGA, kill it and then
use startx normally as a regular user!
Try the 4.0.2-11.4.0 release of XFree86 from:
Does it work for you?
I'm afraid not, it continues to do exactly the same thing.
I've recently tried to use the binary only drivers from nvidia and they don't
exibit this behaviour. In fact, if I run X after running it with nvidia's
drivers, instead of a mouse pointer, what I get is a square made out of random
colored pixels and transparent areas (like a spray), and if I run XF86_SVGA I
get no mouse pointer.
Oh, I forgot to state that it is "run X (with nv_drv.o) after running it with
THe XFree86 4.0.3 release may fix this problem for you. Mark Vojcovich
from mvidia commited many nv driver fixes just prior to the release of
4.0.3, and I believe that it should solve the problem for you. If not,
please update the bug report so that I know, and I will contact Mark.
Oops, forgot to tell you where to get it:
Well, I was hoping that it would be fixed in 4.0.3, but... no. Still the same
problem. What really annoys me is that nvidia's drivers seem to get it right, if
only they were opensource...
The problem still persists with XFree86-4.1.0-3 shipped with RedHat 7.2.
If you use: Option "swcursor" in the config file in the Device section,
does the problem go away?
With SWCursor the problem goes away.
I think I'm not very far from the truth by saying this is a initialization
problem with hardware mouse cursor. As a side note there was recent version of
nvidia's drivers (1.0-12xx I think) that also showed up this behaviour but
they seem to have fixed it now.
I've updated our hardware configuration database to default this
particular card to use "swcursor". This will appear shortly
in rawhide. Also, you might want to test the rawhide version of
XFree86 4.2.0 out to see if perhaps the problem is now gone with
the hardware cursor. If so, I can revert the swcursor option.