On Wed, 12 Mar 2003 [EMAIL PROTECTED] wrote:

> In theory, is there any limit to the number of mice that can be handled
> simultaneously in XF86? I can't think of a reason why one would want 4 or 5
> people controlling the same pointer, but would be fun to try.

No limit, at least under 2.4.19 and greater kernels.

Here's actually what happens (I examined the kernel code a few months ago
for a strange project --- I even modified the kernel code to try to do
something similar with keyboards, but stopped when I realised I'd have to
replicate all the kernel keyboard buffer code and scancode translation
logic).

When you plug a new USB mouse into your computer the kernel creates a file
/dev/mice/mouse? (where ? is a number). If you ignore this file then all
events from that mouse get redirected to /dev/misc/psaux (the standard
PS/2 mouse device). If you tell X that your mouse device is
/dev/misc/psaux then you can use as many USB mice as you want to.

If you open and read that file (/dev/mice/mouse?) then you get the input
stream from the mouse, and the mouse events are no longer sent to
/dev/misc/psaux

ooh, I'm assuming you're using devfs. Otherwise I have no idea the file
names, and things could easily work differently.

> Also, does a usb mouse provide a better performance than the equivalent
> mouse, in a PS/2 port? I would have thought that any differences would have
> been almost indiscernible.

USB mice update on the screen many times more per second than PS/2 mice.
It's especially noticeable when moving the mouse in circles, and very very
noticeable when playing games like Quake.

Tim Wright

Assistant Lecturer
Department of Computer Science
University of Canterbury

http://www.cosc.canterbury.ac.nz/~tnw13

Reply via email to