8.2.x, touchpads and papercuts

Paul Fox pgf at laptop.org
Sun Nov 29 20:26:59 EST 2009

hi martin -- sorry for the delay.  the Weekend of Big Meals got in
the way.  :-)

martin wrote:
 > On Thu, Nov 26, 2009 at 2:38 PM, Paul Fox <pgf at laptop.org> wrote:
 > >  > = CL1 Alps users
 > >  >   - better kmod?
 > >  >   - pgf proposed an xset command to tweak responsiveness?
 > >
 > > both of the above points are covered here:
 > >    http://wiki.laptop.org/go/Touchpad_driver_changes
 > Great. Been reviewing this a bit today. Questions:
 >  - What changes have you applied in it - git tree? The mouse drivers
 > don't seem to be modular...

all of the changes described on that page are on the olpc-2.6 branches
we've been using for 1.5, and are also on the "testing" branch.  the
commit(s) on the testing branch which do this are, i think, the only
commits on that kernel branch since we released 8.2.

 >  - Is the xset command change a good idea even without kernel/driver
 > changes? Or only with them?

i think it's a good idea regardless.  it's possible that the
acceleration value i selected is too "slow", but i think if you
take a child's motor control skills into account, it's probably a
good choice.  (the downside is that it takes more than a single swipe
of the finger to traverse the screen.)

 > >  > = CL1A "New touchpad" users
 > >  >  - split off psmouse kmod?
 > >  >  - kmod or xorg option to disable tapping?
 > >
 > > when we first got the new CL1A touchpads, and enabled the
 > (...)
 > So, not much I can do with a modest effort. Disabling tap-to-click can
 > only happen with the synaptics driver, and that is a longish road.

that's right.  but since i think we want to do this for 1.5 as well,
it might happen "for free" if your schedule stretches enough.  (or,
it might not, depending on what needs to change.)

 > > daniel recently
 > > pointed out to me that we could build in the synaptics protocol,
 > > but explicitly not choose it (i.e., override auto-detection) by
 > > using a module parameter.  this would let deployments and others
 > > experiment with ways to make tap-to-click go away while keeping
 > > everything else working.
 > So you mean CONFIG_MOUSE_PS2_SYNAPTICS=m and that combined with an
 > xorg stanza? Or a udev rule? Or just inelegantly renaming the kmod?

my impression from dsd was that we could turn on the CONFIG_ var, and
then disable the protocol with a module parameter, presumably in

 > > i did propose splitting off the psmouse module, so that it could
 > > be released independently of the kernel.
 > Happy to do that. You mean...?
 > ...

i think it would look like this:

the result would be a psmouse.ko module.

 paul fox, pgf at laptop.org

More information about the Devel mailing list