Panasonic Toughbook CF-29 Touch Screen

stan.distortion · 03-06-2007, 07:37 AM

You had both pad and screen working ok, or just the pad before the kernel upgrade to 2.6.17-11 ?
Also, with the dmesg output above, you didn't have an external mouse connected?
I will need to have a better look at the od command, it doesn't seem to catch all input events. A tap on the pad for a single left click doesn't get any output from it here.
Thanks for the output. Can you send me the output of lsmod, I'm using 2.6.15 and I think the module name has been changed, it may be lbps2 or something similar.
The xorg module should also be mentioned in /etc/X11/xorg.conf in the Pointer Device section, you should be able to edit that to get things working normaly (without the touchscreen). If you want, post the input device sections and I'll try and post back something. Here is mine as an example:

Quote:

Section "InputDevice"
Identifier "Configured Mouse"
Driver "mouse"
option "CorePointer"
option "Device" "/dev/psaux"
option "Protocol" "ExplorerPS/2"
option "ZAxisMapping" "4 5"
option "Emulate3Buttons" "true"
EndSection

Section "InputDevice"
Identifier "Synaptics Touchpad"
Driver "synaptics"
option "SendCoreEvents" "true"
option "Device" "/dev/psaux"
option "Protocol" "auto-dev"
option "HorizScrollDelta" "0"
EndSection

However, if you are using an external logitec mouse you will need to keep the configuration for it.
Cheers'
Stan

PPaFin · 03-06-2007, 07:56 AM

Quote:

Originally Posted by stan.distortion

You had both pad and screen working ok, or just the pad before the kernel upgrade to 2.6.17-11 ?

Pad was working ok, before upgrade.

Quote:

Originally Posted by stan.distortion

Thanks for the output. Can you send me the output of lsmod, I'm using 2.6.15 and I think the module name has been changed, it may be lbps2 or something similar.

Here it goes (from my second CF-18, which do have same problem):

--

Code:

Module                  Size  Used by
rfcomm                 42260  0
l2cap                  27136  5 rfcomm
bluetooth              53476  4 rfcomm,l2cap
speedstep_centrino      9760  1
cpufreq_userspace       5408  0
cpufreq_stats           7744  0
freq_table              6048  2 speedstep_centrino,cpufreq_stats
cpufreq_powersave       2944  0
cpufreq_ondemand        8876  1
cpufreq_conservative     8712  0
video                  17540  0
tc1100_wmi              8324  0
sbs                    16804  0
sony_acpi               6412  0
pcc_acpi               14080  0
i2c_ec                  6272  1 sbs
i2c_core               23424  1 i2c_ec
hotkey                 11556  0
dev_acpi               12292  0
button                  7952  0
battery                11652  0
container               5632  0
ac                      6788  0
asus_acpi              17688  0
ipv6                  272288  8
parport_pc             37796  0
lp                     12964  0
parport                39496  2 parport_pc,lp
af_packet              24584  2
pcmcia                 40380  0
ipw2100                75828  0
joydev                 11200  0
yenta_socket           28812  2
rsrc_nonstatic         15360  1 yenta_socket
ieee80211              35272  1 ipw2100
ieee80211_crypt         7552  1 ieee80211
tsdev                   9152  0
8139cp                 24832  0
8139too                29056  0
mii                     6912  2 8139cp,8139too
pcmcia_core            43924  3 pcmcia,yenta_socket,rsrc_nonstatic
usbhid                 45152  0
evdev                  11392  1
psmouse                41352  0
serio_raw               8452  0
snd_intel8x0           34844  1
snd_ac97_codec         97696  1 snd_intel8x0
snd_ac97_bus            3456  1 snd_ac97_codec
snd_pcm_oss            47360  0
snd_mixer_oss          19584  1 snd_pcm_oss
snd_pcm                84612  3 snd_intel8x0,snd_ac97_codec,snd_pcm_oss
snd_timer              25348  1 snd_pcm
snd                    58372  8 snd_intel8x0,snd_ac97_codec,snd_pcm_oss,snd_mixer_oss,snd_pcm,snd_timer
shpchp                 42144  0
pcspkr                  4352  0
pci_hotplug            32828  1 shpchp
intel_agp              26012  1
agpgart                34888  2 intel_agp
soundcore              11232  1 snd
snd_page_alloc         11400  2 snd_intel8x0,snd_pcm
ext3                  142856  1
jbd                    62228  1 ext3
ehci_hcd               34696  0
uhci_hcd               24968  0
usbcore               134912  4 usbhid,ehci_hcd,uhci_hcd
ide_generic             2432  0
ide_disk               18560  3
piix                   11780  1
generic                 6276  0
thermal                15624  0
processor              31560  2 speedstep_centrino,thermal
fan                     6020  0
fbcon                  41504  0
tileblit                3840  1 fbcon
font                    9344  1 fbcon
bitblit                 7168  1 fbcon
softcursor              3328  1 bitblit
vesafb                  9244  0
capability              5896  0
commoncap               8704  1 capability

--

xorg.conf:

I have tried different things with this one (and there is now device with path experiment going on). It seems that I can put or take away almost anything in xorg.conf, but touchpad and screen are working just as bad.

--xorg.conf (input part)--

Code:

Section "InputDevice"
  Identifier "Configured Mouse"
  Driver "mouse"
  option "Device" "/dev/input/mice"
  option "Protocol" "ExplorerPS/2"
  option "ZAxisMapping" "4 5"
  option "Emulate3Buttons" "true"
EndSection

Section "InputDevice"
  Identifier "Synaptics Touchpad"
  Driver "synaptics"
  #Option "SendCoreEvents" "true"
  Option "Device" "/dev/input/by-path/pci-0000:00:1d.0-usb-0:1:1.0-mouse"
  Option "Protocol" "auto"
EndSection

--

stan.distortion · 03-06-2007, 09:03 AM

Quick note, try:
sudo rmmod evdev
sudo rmmod psmouse
sudo modprobe psmouse
it probably will refuse with "module evdev in use", thats where the touchscreen input is coming in. You could try and force it to unload with "sudo rmmod -f evdev" but it will probably cause a kernel panic.
cheers,
stan

stan.distortion · 03-06-2007, 09:16 AM

Another. evdev almost certainly wont unload but try the 2 for the psmouse module and post the last few lines of dmesg
thanks

PPaFin · 03-06-2007, 07:38 PM

Quote:

Originally Posted by stan.distortion

Another. evdev almost certainly wont unload but try the 2 for the psmouse module and post the last few lines of dmesg
thanks

I did sudo rmmod psmouse and module unloads, no indication in dmesg. No working touchpad or touchscreen. External mouse is working.

sudo rmmod evdev gives an errror 'module in use'.

I post this and do some experiments now, looking forward to get around that evdev

james_jenkins · 03-07-2007, 04:04 AM

Quote:

Originally Posted by PPaFin

Here it goes (from my second CF-18, which do have same problem):
--

PPaFin,

What are the full model numbers of your CF-18's? There are two distinctly different screens on these:

* Touchscreen PC version: 10.4" 1024 x 768 (XGA) transmissive, outdoor-readable TFT Active Matrix Color LCD

* Tablet PC version: 10.4" 1024 x 768 (XGA) transmissive, outdoor-readable TFT Active Matrix Color LCD with Digitizer

Do you happen to have a spare CF-29 lying around that you could try following my instructions earlier in this thread on? I am very interested if someone can CONFIRM that the same procedure that worked for a CF-28 will work for a CF-29. (I don't have a CF-29 to try it on yet).

Just for grins I think I may image off my CF-28 and try installing kubuntu 6.10 this weekend just to see what happens.

James

PPaFin · 03-07-2007, 04:29 AM

Quote:

Originally Posted by james_jenkins

What are the full model numbers of your CF-18's?

CF-18BHKFXGS = 900 MHz,GPRS (MC35)
CF-18DHAFBGS = 1,1 GHz,GPRS (MC45)

Quote:

Originally Posted by james_jenkins

Do you happen to have a spare CF-29 lying around that you could try following my instructions earlier in this thread on? I am very interested if someone can CONFIRM that the same procedure that worked for a CF-28 will work for a CF-29. (I don't have a CF-29 to try it on yet).

Yes indeed I have one CF-29 with kubuntu 6.10. I could try with that almost anything, could you point out what you wishes are ?

stan.distortion · 03-08-2007, 04:29 AM

Could you send me the output of "sudo od /psaux" again please. The 000010 on line 300 should be switching to absolute, just want to check if it is a glitch or not. Can you tap the screen, then the pad, screen, pad etc around 10 or 15 times each. You can email it to me direct at stan.distortion<at_goes_here>laposte.net to save cluttering up the list. I haven't figured out haw to shrink quotes yet

Thanks a lot

stan.distortion · 03-08-2007, 04:56 PM

PPaFin, what devices do you have in /dev/input on the cf-18? The same as the cf-29? I'm reading through the kernel mailing lists, the intention was to have 2 separate devices for the screen and the pad. If it was implemented and is loading properly there should be at least one extra device node in /dev/input
cheers

stan.distortion · 03-09-2007, 03:49 AM

Thanks for the output. It will be hard to see what is happening with just the od command, the line shows 8 byte and the mouse communicates with 3, so you need 2 to 3 signals before anything shows up.
"od -tx1 -vw3 /dev/psaux" would give a better idea of what is happening but it still needs to be converted from hex (octal with just od) to binary to see what is actually happening. If you set a scientific calc to oct, paste the blocks in and switch to bin you will see the signal. It starts at 100471 above, the next 2 blocks are for the position, then repeats again. The last bit is always 1 which should mean "left mouse button pressed". The only exception above was the 000010, in the output you sent me there are a few more but no positive "switch output here" signals. 3 and 4 (count back from 8 on the right, the first 8 are a repeat of the previous position signal) are showing something but they should also be for the position.
I wrote a few scripts to separate and convert everything, will do a few more to separate each bit and translate the known signals over the weekend. Just hoping all the input gets to /dev/psaux and doesn't get filtered by the driver.
If you do "uname -a" and send me the output, I can recompile the lifebook module to get your touchpad acting normally again. As for a fix, looking at the lifebook module I don't think it will be simple. It looks like once it is loaded it is stuck in absolute, the idea of having 2 separate nodes looks better from here.
Cheers,
Stan

PPaFin · 03-09-2007, 04:48 AM

Quote:

Originally Posted by stan.distortion

"od -tx1 -vw3 /dev/psaux"

Just quick try, pad, touch and pad:

Code:

000000 19 b2 38
0000003 29 00 fa
0000006 29 02 fa
0000011 29 01 f8
0000014 29 01 fc
0000017 29 01 fd
0000022 29 00 fc
0000025 29 00 fb
0000030 39 ff fb
0000033 39 ff fd
0000036 39 ff fe
0000041 39 ff ff
0000044 39 ff fd
0000047 39 ff fe
0000052 29 00 fe
0000055 39 ff fe
0000060 29 00 fd
0000063 29 00 ff
0000066 29 01 ff
0000071 09 01 00
0000074 29 01 ff
0000077 29 00 ff
0000102 29 01 ff
0000105 29 01 ff
0000110 09 01 00
0000113 29 01 ff
0000116 29 01 fe
0000121 09 01 00
0000124 09 01 01
0000127 29 00 ff
0000132 29 01 ff
0000135 29 02 fe
0000140 09 01 00
0000143 09 01 00
0000146 08 00 00
0000151 19 f6 1b
0000154 29 00 ee
0000157 29 01 f6
0000162 29 01 f9
0000165 29 00 fb
0000170 29 01 fd
0000173 29 00 ff
0000176 29 00 ff
0000201 29 00 ff
0000204 09 01 00
0000207 09 01 00
0000212 29 01 ff
0000215 29 00 ff
0000220 29 01 ff
0000223 29 01 ff
0000226 29 01 ff
0000231 09 01 00
0000234 09 01 00
0000237 09 01 00
0000242 29 01 ff
0000245 09 02 00
0000250 09 01 00
0000253 29 01 ff
0000256 29 02 ff
0000261 09 01 00
0000264 29 01 ff
0000267 29 02 ff
0000272 09 02 00
0000275 09 02 01
0000300 09 01 01
0000303 09 01 00
0000306 09 01 00
0000311 09 02 00
0000314 09 01 00
0000317 19 ff 01
0000322 19 fc 02
0000325 19 fa 02
0000330 19 fa 03
0000333 19 fb 03
0000336 19 fc 03
0000341 19 fd 02
0000344 19 ff 02
0000347 19 ff 01
0000352 19 ff 01
0000355 09 00 01
0000360 09 01 00
0000363 09 02 02
0000366 09 03 03
0000371 09 03 05
0000374 08 00 00
0000377 09 7f 08
0000402 09 61 00
0000405 19 ff 00
0000410 29 04 f9
0000413 29 00 fe
0000416 29 00 fe
0000421 29 01 fd
0000424 29 01 fc
0000427 29 00 fd
0000432 29 01 fc
0000435 29 00 fc
0000440 29 00 fc
0000443 29 01 fd
0000446 29 00 fd
0000451 29 00 fd
0000454 29 01 fd
0000457 29 00 ff
0000462 29 00 fe
0000465 29 00 ff
0000470 29 00 ff
0000473 29 00 ff
0000476 09 00 01
0000501 09 00 02
0000504 09 00 03
0000507 09 00 04
0000512 09 01 05
0000515 09 04 12
0000520 08 00 00
0000523 19 81 12
0000526 19 81 00
0000531 19 cc 00
0000534 39 fe fd
0000537 39 fd fe
0000542 39 ff ff
0000545 29 00 ff
0000550 29 01 ff
0000553 29 01 ff
0000556 29 02 fd
0000561 29 01 fe
0000564 29 01 fc
0000567 29 02 fd
0000572 29 02 fd
0000575 29 02 fa
0000600 29 03 fa
0000603 29 03 f9
0000606 29 03 f9
0000611 29 03 fc
0000614 29 02 fe
0000617 29 01 fe
0000622 29 00 fe
0000625 29 01 ff
0000630 29 00 ff
0000633 29 00 ff
0000636 29 01 ff
0000641 29 01 ff
0000644 29 00 ff
0000647 09 00 02
0000652 09 02 02
0000655 09 01 02
0000660 09 01 02
0000663 09 01 03
0000666 09 00 03
0000671 09 00 02
0000674 09 00 01
0000677 09 01 00
0000702 29 01 fe
0000705 09 00 01
0000710 08 00 00

Quote:

Originally Posted by stan.distortion

"uname -a"

Linux cf18 2.6.17-11-generic #2 SMP Thu Feb 1 19:52:28 UTC 2007 i686 GNU/Linux

dmesg:

Code:

[17179573.116000] input: AT Translated Set 2 keyboard as /class/input/input0
[17179591.180000] input: PC Speaker as /class/input/input1
[17179591.560000] input: LBPS/2 Fujitsu Lifebook TouchScreen as /class/input/input2
[17179591.984000] input: LBPS/2 Fujitsu Lifebook TouchScreen as /class/input/input3
[17181703.332000] input: Logitech USB-PS/2 Optical Mouse as /class/input/input4
[17181703.332000] input: USB HID v1.10 Mouse [Logitech USB-PS/2 Optical Mouse] on usb-0000:00:1d.0-1
[17181703.332000] drivers/usb/input/hid-core.c: v2.6:USB HID core driver
[17214211.152000] input: Logitech USB-PS/2 Optical Mouse as /class/input/input5
[17214211.152000] input: USB HID v1.10 Mouse [Logitech USB-PS/2 Optical Mouse] on usb-0000:00:1d.0-1

I think problem might relate fact that it seems that there is two touch screen, which is actually a pad and a touchscreen?

I am really looking forward to have this issue resolved, I take this CF-18 with me on weekend, so if there is anything - I can propably try it out

stan.distortion · 03-09-2007, 12:43 PM

Have the psmouse module compiled for 2.6.17-11-386 without cf-18 support. It should make the touchpad usable again, just need somewhere to upload it to. Run "modinfo psmouse" and make sure it matches "2.6.17.11 preempt mod_unload 486 gcc-4.0", if not I can re-compile for whatever is listed there.
Cheers,
Stan

stan.distortion · 03-09-2007, 06:08 PM

Bit more info that may be of use to anyone else looking at this:
http://www.fujitsu.com/downloads/MIC...-0559-v044.pdf
About half way down are the byte sequences, below that are the commands.
The important part seems to be on page 17, when the panel it touched bit 2 of the first byte is set to 1. When released it is set to 0. Bit 3 is always set to 0.
These 2 bits are opposite to the ps2 mouse specs where bit 2 is set to 0 and bit 3 is set to 1.
From everything I have looked at from the /dev/psaux output, bit 2 is 0 and bit 3 is 1 as per the ps2 specs. Bit 0 is unusual as it is almost always set to 1 ie. left mouse button pressed.
I think either the specs for this controller are different to other Fujitsu controllers or bit 2 and 3 are being set by the psmouse module.
The lifebook part of the psmouse module seems to look for the Fujitsu spec for bit 3 here:
http://www.gelato.unsw.edu.au/lxr/so...lifebook.c#L76
It is switching to absolute mode yet bit 3 is always 1 in the output,
I'm learning a lot trying to figure this out, but it is over my head. If anyone could help out and shed some light on this I would appreciate it.
Thanks,
Stan

PPaFin · 03-10-2007, 12:57 PM

This has been great work from Stan so far. He pointed out the problem in lifebook.c and as I downloaded 2.6.20 kernel version and did modification to lifebook.c file before compiling, I got results!

So far I have refreshed my ability to compile kernel and got my CF-18 working with pad. Looking also forward to see this issue resolved in future, I can offer testing with our various Toughbooks.

This was so nice work, that I might put in some request for our latest CF-Y5 model, which do have integrated 3G/HSDPA module - which you can guess is not working with linux

Module is Novatel Wireless and I have also sent feedback to Panasonic and Novatel. So far there has been no response.

So, now I am back into track with this CF-18. This is nice machine with integrated GPRS and it would be shame to keep it with windows

Azizcoos · 03-10-2007, 01:12 PM

I've been out of the loop here for a while. It seems that LQ didn't send me a notice of new traffic on this thread...

My CF-29, which does have a touchscreen that I have no reason to think was not factory installed, but I can't be sure since I bought it used, reports under dmidecode as:

Code:

System Information
        Product Name: CF-29HWLGZBM
Base Board Information
        Product Name: CF29-3

My try at kernel source hacking last fall was indeed to simply add:

Code:

      {
               .ident = "CF-29",
               .matches = {
                       DMI_MATCH(DMI_PRODUCT_NAME, "CF-29"),
               },
       },

into /usr/src/linux-2.6-2.6.17/drivers/input/mouse/lifebook.c . Of course, I have now upgraded to 2.6.18, so I'll have to update the kernel source before I can try it again... but either "CF-29" or "CF29" would work. While researching last fall, I ran across some discussion on the subject, and the people who put that into the kernel code were also lamenting about the crudeness of the method.

James asked:

Quote:

Do you happen to have a spare CF-29 lying around that you could try following my instructions earlier in this thread on? I am very interested if someone can CONFIRM that the same procedure that worked for a CF-28 will work for a CF-29.

I do, in fact, although I wouldn't call it "a spare"

One question first though; James- what kernel version are you using? I ask because IIRC 2.6.17 was the first one which shipped with the lifebook driver included, so any fix that worked for versions <2.6.17 would likely need to be modified for later ones.