radut / psmouse-alps Goto Github PK

$Id: README 720 2012-10-02 00:14:28Z dturvene $

History
-------
0.1: hacked in rudimentary support for the Dell IR15 N5110.  Nothing else 
     should work - probably causing a kernel panic.
0.2: add logic to isolate N5110 changes from previously supported devices
     expand diagnostics to capture touchpad protocol
     add kernel parameter to enable/disable runtime packet debug
     enabled edge scrolling for N5110
     add more shell scripts to control/debug device
0.3: clean up a potential null pointer exception
     added signature for Dell I17R 7720 based on user confirmaton to V6 protocol
     disable multi-touch for V6 protocol
     include alps.sh helper scripts
0.4: integrate multi-touch support from florin9doi
     default alps_debug=0, can be enabled through sysfs
     code cleanup
0.5: elevate persistent syslog to psmouse_info for production/release capture
      of alps signature and command response - both of which determine control 
      protocol

1.2: Try to unify the two V5 init sequences.

How to set up
-------------

cd to the directory where this readme sits.
as root, do:
 $ dkms add .
 $ modprobe -r psmouse
 $ dkms build -m psmouse -v custom-1.2 --all
 $ dkms autoinstall --force
 $ modprobe psmouse

There is some reporting data in dmesg | grep psmouse

Background
----------
This README documents my work to reverse engineer the Alps touchpad on a
Dell N5110 for Linux.  When I first purchased the laptop and installed
Ubuntu 12.04, I noticed that the touchpad behaved erratically.  It would
randomly skip and home-to-cursor. A brief search showed that this
behavior has been a long-standing and well publicized issued.  To date
several other Alps touchpads have been reverse engineered but no one had
addressed this particular model.  

Dell has a driver for the touchpad for Vista and Win7.  I repeatedly
contacted Dell support and Alps; the only response I received was "We
don't support Linux"; odd considering Dell ships Ubuntu.

I developed a simple serio_raw program to peek/poke at the touchpad and
found it had characteristics inconsistent with other Alps touchpads.  I
used Seth Forshee's procedure to dump the driver interface using qemu and
vista as the guest OS [1].  This require me to patch the qemu
acpi-dsdt.dsl file to recognize the Alps hardware in order to install
the Dell/Alps Vista driver [2]

Based on the qemu log, I started hacking up a psmouse DLKM from the
linux-3.2.0 source tree.  First I added in Ben Gamari's ALPS_PROTO_V5
fix for the Alps Touchpad on his Dell E6230 [3] and tried it but the
characteristics were too different.  In fact even the enter-command-mode
generic response is different.  I then took the vista log and
created an ALPS_PROTO_V6 hw init by pasting the init sequence from the
qemu log into alps_hw_init_v6.  That got me part of the way, the device
was no longer being handled as a stock ps/2 mouse but now as an ALPS
DualPoint Touchpad.  However, the 6-byte message format was undocumented
[4]. As far as I can tell, it is:

 byte 0:    1    1    0    0    1    0    0    0
 byte 1:    0   x6   x5   x4   x3   x2   x1   x0
 byte 2:    0   y6   y5   y4   y3   y2   y1   y0
 byte 3:    0    ?    ?    ?    1    m    r    l
 byte 4:   y10  y9   y8   y7  x10   x9   x8   x7
 byte 5:    0   z6   z5   z4   z3   z2   z1   z0

So I created alps_process_touchpad_packet_v6 to parse the touchpad
packets.

The ALPS touchpad is working on my N5110.  I can control the touchpad
somewhat via xinput and the System Settings->Touchpad widget.  I think
it can still be tightened up but, at least, it's usable now.

Dave Turvene

[1] http://swapspace.forshee.me/2011/11/touchpad-protocol-reverse-engineering.html
[2] http://www.spinics.net/lists/linux-input/msg21948.html
[3] http://www.spinics.net/lists/linux-input/msg21993.html
[4] file:///usr/src/linux-source-3.2.0/linux-3.2.0/Documentation/input/alps.txt

bash shell scripts
------------------
See alps.sh for support scripts.