[SOLVED] Keyboard issue - fn+key "KeyRelease" event not triggered

Question

Hi Archers :)
First post, sorry if not in the right section.
I have a quite anoying bug. I bought a no name laptop, imported by Textorm, model M116 SU7300, same model as this one (http://www.textorm.com/pc-portable/pc-p … 22%5D.html)
On the keyboard, the '<' and '>' symbols are mapped on "fn+W" and "fn+MAJ+W".
Everything works fine in a TTY console. All others fn-KEY are working well.
The problem is with a Xorg environment : when pressing fn+W or fn+MAJ+W, it seems that the release key is never triggered, acting like if I was pressing the key in continue.
When pressing another key, the crazy '<' input stop, but then, fn+W won't works again... until I trigger a press/release event with 'acpi_fakekey 86'.
Then, same scenario.
I booted a live Archbang,  same result (well, I wasn't very surprised), and a live Sabayon... same result.
Using xev, I can see that when pressing/releasing fn+W, the KeyReleased event is never triggered. I can provide a log if needed.
So it seems related to the Xorg driver....
I tryied to use the kbd driver by adding an xorg.conf
Section "InputDevice"
Identifier "Generic Keyboard"
Driver "kbd"
Option "CoreKeyboard"
Option "XkbRules" "xorg"
Option "XkbModel" "pc105"
Option "XkbLayout" "fr" # "fr-oss" pour Feisty et +
Option "XkbVariant" "latin9"
EndSection
Section "ServerFlags"
Option "AutoAddDevices" "Off"
EndSection
I can confirm that the kbd driver was in use with Xorg log :
[ 900.371] (II) LoadModule: "kbd"
[ 900.371] (II) Loading /usr/lib/xorg/modules/input/kbd_drv.so
[ 900.371] (II) Module kbd: vendor="X.Org Foundation"
[ 900.371] compiled for 1.14.0, module version = 1.7.0
[ 900.371] Module class: X.Org XInput Driver
[ 900.371] ABI class: X.Org XInput driver, version 19.1
[ 900.780] (II) Using input driver 'kbd' for 'Generic Keyboard'
[ 900.780] (**) Option "CoreKeyboard" "on"
[ 900.780] (**) Generic Keyboard: always reports core events
[ 900.780] (**) Generic Keyboard: always reports core events
[ 900.780] (**) Option "Protocol" "standard"
[ 900.780] (**) Option "XkbRules" "xorg"
[ 900.780] (**) Option "XkbModel" "pc105"
[ 900.780] (**) Option "XkbLayout" "fr"
[ 900.780] (**) Option "XkbVariant" "latin9"
[ 900.780] (II) XINPUT: Adding extended input device "Generic Keyboard" (type: KEYBOARD, id 6)
... no luck. Same comportement.
I tryed the xorg-input-evdev git version, no luck. (But if I understand, xorg-input-evdev use kbd as underlying driver?)
I also tryed xorg-server-dev 1.14.99.1-1
So for now, I use a ugly hack : I mapped "W-W" and "S-W-W" in my openbox rc.xml to execute "acpi_fakekey 86". Nearly working all the time.
I know this isn't directly related to Arch, but if someone has any pointers... I take any ideas.
Thanks in advance!
PS :
xorg-server 1.14.1-1
xf86-input-evdev 2.8.0-1
Last edited by jc.saaddupuy (2013-05-28 18:28:05)

Answer 1

Wow thanks!
I confirm that this is working.
Thanks to your infos, I investigate the udev stuff.
You will have to know the vendor and product name of your machine :
cat /sys/class/dmi/id/{sys_vendor,product_name}
Thats the strings your udev rule have to match in ENV{DMI_VENDOR} and ATTR{[dmi/id]product_name}.
For me, both was empty, so I use a wildcard :
ENV{DMI_VENDOR}=="*", ATTR{[dmi/id]product_name}=="*"
Then, you'll have to specify which codes needs to trigger a key release (sames as found with keymap -i)
Add a file in /usr/lib/udev/keymaps/force-release/  whith the hex values (one per line)
I named mine 'textorm'
/echo 0x56 > /usr/lib/udev/keymaps/force-release/textorm
And add it in the RUN of your udev rule (not the full file name):
ENV{DMI_VENDOR}=="*", ATTR{[dmi/id]product_name}=="*", RUN+="keyboard-force-release.sh $devpath textorm"
Knowing your serio ( /sys/bus/serio/devices/serio0), you can test the keyboard-force-release.sh script (the serio parameter HAVE TO NOT begin with /sys/)
/usr/lib/udev/keyboard-force-release.sh bus/serio/devices/serio0 textorm
You can then check the /sys/bus/serio/devices/serio0/force_release content.
Works like a charm :)