Panasonic Toughbook CF-29 Touch Screen

Quick note.
The sequence from the origional driver again:
http://parzelle.de/Linux/Lifebook/index-1.html#ss1.1
The explanation of those from the origional driver source:
// d0 ??????????????
// e1 ??
// e2 ??
// E6 Reset Scaling
// E7 Set scaling
// E8 Set resolution
// E9 Determine status
// Ea Set stream modus
// eb read data
// ec reset wrap mode
// ee set wrap mode
// f0 set remote mode
// f2 identify unit
// f3 set sample rate
// f4 enable
// f5 disable
// f5 set standard
// fe resend
// ff reset
The psmouse kernel driver using the same codes:
http://www.gelato.unsw.edu.au/lxr/so...e/psmouse.h#L4
And the relevant part from the bugzilla link:
"Clarification: On the CF-18, the change made to lifebook.c that causes the
touchpad to be seen as an absolute device (touchscreen) does get the touchscreem
working. Using evtest against the device assigned to the
touchscreen, events are generated. When configured in xorg, the touchscreen can
be used."

Cheers,
Stan

Nobles, thanks for the link. I remember seeing the page before and not reading through it, wish I had now. It seems to be mostly obsolete as the psmouse driver has gone through a lot of changes and now works with the lifebook and a single report of touchscreen output on a cf-18.
Need some idea of resources, who has a cf 18, 28, 29 or 30 and what kernel are they using?
With a kernel more recent than 2.6.13, lifebook.c modified to check for toughbooks and MOUSE_PS2_LIFEBOOK compiled into the kernel as a module it should be possible to load and unload the module to find out what is and isn't happening.

Cheers,
Stan

CF-29 running 2.6-18 here.

Do you have both the psmouse and lifebook drivers as modules and xorg.conf set up?
I'm still trying to find the links between xorg.conf and the lifebook module. It looks like the values ie. 0x28, Ox3 and 0x7 are now supplied by xorg.conf.
Can you post the settings from xorg.conf and also load and unload both the psmouse and lifebook modules and post the response from dmesg and od /dev/psaux

It should be simple to send the control commands to /dev/psaux directly from the shell, I'll check that and see how it works on the touchpad. An absolute touchpad sounds weird

Cheers,
Stan

Another quote from somebody that managed to change lifebook.c in a similiar way to what Azizcoos did last fall (looks like it is with NetBSD instead of Linux this time but the results sound the same) on a CF-18 and at least get some absolute output out of both the touchpad and touchscreen - seems to confirm what the bugzilla link stated. I haven't actually seen any identical reports for a CF-28/CF-29 or CF-30 - both pieces of information I have seen describe this behaviour on a CF-18 which could be different than the others. If we could get this far and prove the other Toughbooks act the same, the calibration may be a minor issue.

http://archive.netbsd.se/?ml=xorg&a=2006-05&t=2024739
---- Quote from site above -----
I hacked at the kernel source (drivers/input/mouse/lifebook.c) in order
do get the touchscreen seen as a touchscreen by the input system. My
shotgun approach also made the touchpad act like a touchscreen. With
this change, X was able to work with the touchscreen. I tried evtouch,
evdev, fpit and maybe another Driver. They all acted generally the
same. Issues I had maybe were due to lack of calibration.
--- end of quote -----

Nice one.
Not found a simple command to send instructions to /dev/psaux yet, however did just find this:
http://www.cs.washington.edu/homes/oskin/setglide.c
copy and paste to a text file ie. setglide
change string
cd to the directory with setglide
$ gcc setglide.c
will have an executable called a.out
$ sudo ./a.out reset

Not tried it to reset the touchpad yet, seems to work ok as is but its hard to tell.

If we can modify the lifebook.c and get the kernel patched to get to the point where the absolute touchpad works as Azizcoos had, I wonder if tapping the screen while doing the following might let you know if you are getting any thing out of the touchscreen - this might be a way to test for some output without relying on a correct Xorg configuration (which will be the next major step if this works).

cat /dev/input/event1

I got this idea from reading the link below and the one I sent before and seeing that the same event1 shows up as the touchscreen device.

http://coffee.geek.nz/2006/07/25/ux1...een_howto.html

I don't have a CF-28/CF-29 with Linux installed right now but if we get close on this, I might have to take one home to play with over the weekend - let me know if anybody can try this out.

It will verify it if the touchscreen gets initialized, the values should give everything needed for xorg.conf. Just the number of events in /dev/input/ may change when the driver module is loaded too.
udevinfo -a -p /sys/class/input/eventX
Where X is the event number should get some info on what the device reports its self as. Apparently the screen doesn't report a whole lot but the controller should give some kind of clue.

The setglide code isnt making any changes for me, it looks like the psaux wont accept the write command, it isnt giving any errors either though. There will be something to send to the device, god bless open source

Cheers

This is weird. With a default installation (kubuntu here) on a cf-27 the touchscreen shows up as /dev/input/ts0 through this driver:
ts: Compaq touchscreen protocol output
I see that horrible word in there, the kernel did it , but the driver is connected to the touchpad in absolute mode. Weird, the same f***up through 4 generations of toughbooks

One other thing I noticed as I looked into this is that the DMI product name that you need to know to insert into the lifebook.c file is not just CF-29 on the CF-29 Toughbook I have according to Panasonic's DMI Viewer utility. I don't know if you need an exact match (looks like from the links back to the contents of lifebook.c and the various Lifebook models that they get specific in the product name) or not for the routine to initialize properly - on the CF-29 I am trying this on it is CF-29CTKGZKM which includes the more specific part of the model number. The DMI Viewer utility for all models (Windows version) is available on the panasonic.ca support site.

Is there only one entry? I have 2 on the cf-27:

Handle 0x0001, DMI type 1, 25 bytes.
System Information
Manufacturer: Matsushita Electric Industrial Co.,Ltd.
Product Name: CF27
Version: 002
Serial Number:
UUID: Not Settable
Wake-up Type: Unknown

Handle 0x0002, DMI type 2, 8 bytes.
Base Board Information
Manufacturer: Matsushita Electric Industrial Co.,Ltd.
Product Name: CF27-2
Version: 001
Serial Number: None

That was using dmidecode on linux. Does the dmi viewer from Panasonic have the option of dumping the information in hex? It may be interpreting the dmi info into the full code to be more usefull for Panasonic.
43 46 2D 32 39 43 54 4B 47 5A 4B 4D 00
should be the right string for CF-29CTKGZKM
Azizcoos mentioned adding CF-29 to lifebook.c got a responce, but only from the touchpad so it seems to be finding something.

Cheers,
Stan

Here is the info from the Panasonic DMI Viewer (I erased the serial number for privacy reasons) for a CF-29. I think the name the lifebook.c routine wants is the Product Name from the System Information section - dmidecode kind of confuses things by calling the name from both sections "Product Name" but below the second is just called "Product" - in any case the name is not just CF-29 for this model. I will have to try a CF-28 when I get a chance and see what comes up.

[System Information]
Manufacturer = Matsushita Electric Industrial Co.,Ltd.
Product Name = CF-29CTKGZKM
Version = 001
Wake-up Type = Power Switch
Serial Number =
[Base Board Information]
Manufacturer = Matsushita Electric Industrial Co.,Ltd.
Product = CF29-1
Version = 001

Here is the info from dmidecode from Opensuse Linux on the same CF-29:

Handle 0x0001, DMI type 1, 25 bytes
System Information
Manufacturer: Matsushita Electric Industrial Co.,Ltd.
Product Name: CF-29CTKGZKM
Version: 001
Serial Number:
UUID: 00000000-0000-1000-8000-008045132E3F
Wake-up Type: Other

Handle 0x0002, DMI type 2, 8 bytes
Base Board Information
Manufacturer: Matsushita Electric Industrial Co.,Ltd.
Product Name: CF29-1
Version: 001
Serial Number: None

Just got info back from a cf-28, product name CF-28S3QGBDM, product CF28-3.
Not too sure how the parser works for the DMI, it should accept wildcards though.
Azizcoos, you added a line with CF-29 to lifebook.c and it loaded the module automatically?

Cheers,
Stan

Hi Guys,

Just wanted to let you know that after a very long time I was finally able to get the Touchscreen working on my ToughBook CF-28 under Linux tonight..

It would seem that the CF-28's, and CF-29's both used the same touchscreen. I have a 500mhz CF-27, but have not tried this new setup. Previous attempts had failed. I have got the touchscreen to work on a 300mhz CF-27, but that wasn't the big deal that getting the CF-28 to work was. Sorry, but I don't have a CF-29, althought it should be the same touchscreen. The CF-30 is completely different and has good Linux support.

It is 4:00am and I have to leave in about 4 hours, so sorry about the lack of details at the moment, but I will try to get them posted a little later.

Just couldn't wait to let you know it finally works.

James Jenkins

Good to hear james_jenkins - thanks for taking the time to report so that we know it is at least possible. When you get time it would be great if you could let us know which Linux distro you managed to get it working with as well as letting us know the steps required to get it working (as it is likely a complicated process from what we know so far and each step is critical) and if anything that was discussed here helped.