Regarding the hid-multitouch on ChromiumOS

[ScottLiu]

Hi all,
There is lack of capability of multitouch touchscreen driver.
So it have to load hid-multitouch manually in console mode. how to load hid-multitouch unconditionally at boot or by udev?

thanks,
Scott

[Dave Parker]

Hi Scott,

Under the 3.4 Linux kernel I've been able to get some hid-multitouch monitors working at boot by adding the matching USB vendor and product IDs (vid and pid) to the hid-mulitouch driver. Here is an example:

https://gerrit.chromium.org/gerrit/#/c/39576/

The vid and pid can be found by running lsusb -v (assuming you have a USB connected touchscreen).

Best,

Dave

--
--
Chromium OS Developers mailing list: chromiu...@chromium.org
View archives, change email options, or unsubscribe:
http://groups.google.com/a/chromium.org/group/chromium-os-dev?hl=en
 
 
 

[ScottLiu]

Hi David,
thanks for reply, so you mean that the hid-multitouch module will autoload without any modification if adding mathing VID PID to it?

Because now, I added 'modprobe hid-multitouch' at buttom of /sbin/chromeos_startup and got upstable issue,
sometime the touchscreen work correctly but fail sometime. (both I've checkd hid-multitouch already be probed)

thanks,
Scott

[Dave Parker]

Yes, if you add the vid:pid hid-mulittouch should be auto load. With later kernels, like 3.8, there is a generic multitouch handler and the vid:pid change isn't needed.

You may be running into a race condition between X starting and multitouch devices getting enumerated. If you look through the X server log you should be able to see what multitouch devices were present when X started.

[ScottLiu]

it's work correctly after add my vid, pid to hid-ids.h and hid-multitouch.c.
thanks, again.

[ScottLiu]

sorry that xorg still got wrong behavior.
sometime ok, sometime bad.

now I didn't add any word in /sbin/chromeos_startup but add the VID,PID in the hid-multitouch.

Do you have any idea?

thanks,
Scott

[Yufeng Shen]

So it seems that the device is managed by hid-multitouch now but the
data format it reports
does not work with X.

Can you run

DISPLAY=:0 xinput list --long

and show the touchscreen part of the result ?

and also if possible, can you run evtest on the touchscreen and post
some sample output ?

BTW, does this touchscreen work with a Linux/Windows machine ? If it
works out-of-box with
a linux machine, ChromeOS should also work with it. And is this a
production touchscreen or
a DUT touchscreen ? I can't seem to find the product info according to

"Vendor 0x4f3 Product 0x1a"

[ScottLiu]

Logs file as attachment.
And touchscreen is Elan production which passed win8 WHCK.
so it should work at ChromiumOS as well.

[Yufeng Shen]

So I am guessing the problem is that the touchscreen FW does not
report complete x/y
position in its first report, e.g. in evtest_fail, the first touch
does not have Y position
at all. That's why Xserver would complain (in your tsbug.rar)
"
[ 100.446] ELAN Touchscreen: Attempted to start touch without x/y (driver bug)
[ 100.453] ELAN Touchscreen: TouchUpdate with no valuators? Driver bug
"

In my opinion, this is the touchscreen FW bug, i.e. you have to
provide both X/Y position
in your first touch report. Though I am not sure how win8 can handle this case.

I would suggest you seek help from either the touchscreen vendor (they
might have a
fix in FW) or from mailing list linux...@vger.kernel.org (there
might be a hid-multitouch
quirk can solve this problem).

[ScottLiu]

thanks, I already using kernel-next v3.8 as chromium os and no problem now. thanks everybody!!