touchpad redux

Paul Fox pgf at
Tue Aug 18 17:53:21 EDT 2009

the XO-1 touchpad has recently come under scrutiny once again. 
most long-time readers of this list (and most users of the XO-1)
are aware that the touchpad doesn't always behave as a proper
touchpad should.

as a result of more reports from the field, and with better data
than we've had before [1], we've been taking a fresh look at what
goes wrong when the touchpad acts up, and have taken a fresh stab
at trying to work around some of the problems.

(for various reasons a different touchpad was slipped into
production earlier this year, and initial field reports are much
more favorable.  but that still leaves something on the order of
a million laptops out there, built with the older device.  and to
answer the first question everyone asks:  no, it's not a drop-in
replacement.  most of the lower half of the laptop needs to be
replaced to accomodate the newer touchpad.)

so, i've put together a wiki page detailing what we think goes
wrong, and what we've done so far to try and reduce the impact of
those failures.  most of what i've done is really just
refinements on what was already there (standing on the shoulders
of giants, and all that), but we're hoping that those small
changes may help a lot:

the new driver has been commited to the olpc-2.6.30 branch, so
the next build(s) of F11-on-XO1, or SoaS-on-XO1, should pick up
the new driver.  if you've had trouble with the touchpad in the
past, and can try one of those releases when it's available,
please let me know what you think now.  (if you've resorted to
using a mouse, you might go back to the touchpad for a while
before upgrading to the new driver, just to remind yourself how
poorly, or well, it behaved.)

one unconfirmed theory we've heard from folks in the field is
that the touchpad behavior gets worse over the course of several
hours of continuous use.  if anyone can comment on that, or would
like to design (and execute :-) an appropriate test, please shout!

i'm now in the process of backporting the changes to the
"testing" branch which was the basis for the kernel released for
802.  we don't have plan yet for creating a new release -- i'm
sure the first step will be a kernel rpm, or other installable
package of some sort.


[1] getting better data from the field was made much easier with
the help of a script put together by James Cameron.  See the
final section ("Large Jump Logging Technique") of

 paul fox, pgf at

More information about the Devel mailing list