[PATCH 1/2] improve-s3c2410_ts-debug-messages.patch
dkogan at cds.caltech.edu
Wed Nov 19 08:44:59 CET 2008
I did some work on this earlier, and submitted two patches, with the
first merged into stable. The second patch is here:
This second patch did a lot more to address the hardware's
deficiencies, and the result was much better for clicking actions, like
typing on an on-screen keyboard, but was not as good for dragging
motions. I unfortunately don't have time to work on this now, but at
least one point from my second patch should improve things:
"3. reject_threshold_vs_avg_x was previously looking for jumps larger
than 5 pixels and throwing those away. Our hardware is much less
consistent than this. I now look for jumps 70 pixels in the long
direction and 20 pixels in the short direction, and throwing those
That patch had a similar debugging mechanism to what you posted, and
plotting the touchscreen output from holding a stylus in one spot
showed that our hardware returns very inconsistent results, requiring
a larger rejection threshold. Also, the jitter is much stronger in the
long direction of the touchscreen. There may be more interesting things
in that patch, but I don't remember, and don't have time to work on it
now. Thanks a lot for looking into it.
On Wed, 19 Nov 2008 02:27:27 -0500
"Nelson Castillo" <nelsoneci at gmail.com> wrote:
> On Wed, Nov 19, 2008 at 2:22 AM, Nelson Castillo
> <nelsoneci at gmail.com> wrote:
> > * Print debugging information when you need to send a missing touch
> > event.
> > * Remove a few warnings in printk (%ld used when %d is enough).
> > With this patch we can see more information about the touchscreen
> > misbehavior and a pattern can be spotted: the data that comes right
> > after the missing touch event that is being sent is always wrong.
> > Also notice that sometimes the messages labeled "First" are also
> > broken. All other readings seem to be OK (confirmed with longer
> > tests). In this example all points pressed were near (X:142, Y:876).
> I am trying to understand the issue.
> Here is a longer test (points near (X:128, Y:872)).
> I wonder if it says something.
More information about the openmoko-kernel