tech log on gentoo, linux, and random stuff

Posts Tagged ‘actkbd

actkbd and screen rotation in archlinux

leave a comment »

actkbd is in AUR. I need it so the tablet keys on X220 tablet can serve more use (e.g., differentiate between long and short press, etc.)

In order for the rotation script triggered by actkbd to be able to communicate with X, we need to start actkbd with the DISPLAY=:0 environment variable. The AUR package created the /etc/rc.d/actkbd entry. Just change

[ ! -f $PIDFILE ] && /usr/sbin/actkbd -D -q -p $PIDFILE


[ ! -f $PIDFILE ] && DISPLAY=:0 /usr/sbin/actkbd -D -q -p $PIDFILE

Written by zsh

February 20, 2012 at 2:40 pm

Posted in no cat is good cat

Tagged with , , ,

quick note on x61t tuxonice issues

leave a comment »

  1. dhcpcd dies after resume
    fix: add an OnResume entry in /etc/hibernateOnResume scripts

  2. stylus geometry messed up. also reproducible if
    • manually rotate to another direction by
      xrandr -o right && xsetwacom set stylus rotate CW

    • switch to some VT and back
    • now one or more of {Top/Bottom}{X/Y} is wrong

    there’s no problem if both xrandr and xsetwacom are in normal orientation, so probably a problem of wacom driver. Never realized it before as I’ve never switched to VT from tablet mode, but IIRC hibernate-script has this trick of switching to some ficticious VT and then back to ensure X is displayed on resume, and so the weird problem.

    fix: manually by DISPLAY=:0 xsetwacom set stylus xydefault for device named stylus, etc. . Here’s a script to reset all available devices:

    # rst: reset tablet geometry 
    export DISPLAY=:0.0
    devices=`xsetwacom list dev | cut -f1 -d' '`
    for d in $devices; do
        xsetwacom set $d xydefault

    I just bound it to the tiny little reset button on the tablet panel (keycode 199 as revealed by showkey) in actkbd.

    ## tiny little button, code 199
    199	:rep   :grab,noexec	:
    199	:rel   :grabbed,ungrab  :/home/bin/rst
    199	:rel   :ungrabbed	:/home/bin/rst

Written by zsh

April 1, 2009 at 11:05 pm

Posted in /etc

Tagged with , , , ,


leave a comment »

After my X61 tablet arrived, I needed a keyboard daemon that can take care of hotkey events both under console and in X (I’d like to have control of brightness and volume level, etc., in both). After some search, I settled with actkbd, which to me is a good balance of speed (pure C), functionality and complexity (take a look at, say, gizmo daemon). It monitors evdev keyboard (e.g., /dev/input/event1. so need kernel evdev support), and reponds according to a config file.

  1. Simple config example
    A typical snippet in actkbd’s configuration file looks like this:
    ## mute key
    # hard mute if pressed long, otherwise toggle
    #code   event   attrib          ext command
    113	:rep	:grab,noexec	:
    113	:rel	:grabbed,ungrab	:/home/bin/mute
    113	:rel	:ungrabbed	:/home/bin/toggle
    # hard unmute (all muters)if shift+
    42+113:key::/home/bin/unmute Master Headphone Speaker

    With the helper scripts mute and toggle, this snippet either mutes or toggles the mute status of my soundcard depending on whether the Mute key (113) is just pressed or hold for a while.

    Each config line has four fields,

    • code: keycode as shown in showkey. multiple keys concatenated by ‘+’
    • event: one of key, rep (repeat), and rel (release)
    • attrib: (un)grab to block/release device from/to other programs, (un)grabbed to ensure action taking place only when (un)grabbed. This is the major featureset of actkbd. For a complete list, see the official README
    • ext command: external command to call. It could be better if actkbd can export relevant info as some sort of variable that can be passed to external commands (so that e.g. a general purpose hook program can handle all events).
  2. Tablet screen rotation
    The X61 tablet has some buttons alongside the tablet panel. When in tablet mode, these are the only keys accessible as the whole keyboard is covered under the tablet, so we want to use these keys wisely. With an external helper script rotate, the following snippet will
    • reset screen to normal orientation if “rotate” button hold for long
    • rotate screen 90 degrees clock-wise if “rotate” button is merely pressed
    • rotate screen to a relative orientation (left, right, or invert) if the “rotate” button is hold while the orientation arrow keys are pressed
    ## grab as `clean' bit, detectable
    # 191: "rotate" button
    # rotate clockwise if only hold short
    191	:key	:grab,noexec	:
    191	:rel	:grabbed,ungrab	:/home/bin/rotate
    ## 192 as `hold' bit
    # reset orientation if hold long
    191+192	:rel	:grabbed,ungrab,unset(192)	:/home/bin/rotate monitor
    191+192	:rel	:ungrabbed,unset(192),noexec	:
    ## u/d/l/r/enter: 103/108/105/106/28
    # u/d: invert; l/r: rotate; enter: pin down
    # relative rotation
    191+103		:rel	:ungrab	:/home/bin/rotate i
    191+108		:rel	:ungrab	:/home/bin/rotate i
    191+105		:rel	:ungrab	:/home/bin/rotate l
    191+106		:rel	:ungrab	:/home/bin/rotate r
    191+28		:rel	:ungrab	:/home/bin/rotate c
    191+192+103	:rel	:ungrab	:/home/bin/rotate i
    191+192+108	:rel	:ungrab	:/home/bin/rotate i
    191+192+105	:rel	:ungrab	:/home/bin/rotate l
    191+192+106	:rel	:ungrab	:/home/bin/rotate r
    191+192+28	:rel	:ungrab	:/home/bin/rotate c

    Here, the trick is to use ‘grab’ to indicate whether the current 191 key event is being processed or not, and use the ficticious 192 key to indicate whether it’s a long hold or a short one.

  3. LCD brightness control
    Similar trick can be used to control screen brightness. The master button to use now is “application” button right next to the “rotate” button.
    • short and long hold: reserverd for future use (:D)
    • app + up/down: turn brightness up/down by 1.
      app + left/right: turn brightness to min/max
      app + enter: turn brightness midway (7)
    ## similar `grab' and `hold' trick, but for 152
    152	:key	:grab,noexec			:
    152	:rel	:grabbed,ungrab			:echo "short press"
    152	:rep	:grabbed,set(153),noexec	:
    152+153	:rel	:grabbed,ungrab,unset(153)	:echo "long press"
    152+153	:rel	:ungrabbed,unset(153),noexec	:
    ## adjust brightness
    152+103		:rel	:ungrab	:echo up > /proc/acpi/ibm/brightness
    152+108		:rel	:ungrab	:echo down > /proc/acpi/ibm/brightness
    152+105		:rel	:ungrab	:echo level 0 > /proc/acpi/ibm/brightness
    152+106		:rel	:ungrab	:echo level 15 > /proc/acpi/ibm/brightness
    152+28		:rel	:ungrab	:echo level 7 > /proc/acpi/ibm/brightness
    152+153+103	:rel	:ungrab	:echo up > /proc/acpi/ibm/brightness
    152+153+108	:rel	:ungrab	:echo down > /proc/acpi/ibm/brightness
    152+153+105	:rel	:ungrab	:echo level 0 > /proc/acpi/ibm/brightness
    152+153+106	:rel	:ungrab	:echo level 15 > /proc/acpi/ibm/brightness
    152+153+28	:rel	:ungrab	:echo level 7 > /proc/acpi/ibm/brightness
  4. Microsoft Natural Ergonomic Keyboard 4000 (NE4k)
    The problem with NE4k, which I use on my desktop, is that kernel exports it as two devices, one for the normal keys (a,o,e,u,… yes I use dvorak), say /dev/input/event3, and one for all the multimedia/net/hot keys, say /dev/input/event4. Since each process of actkbd can only monitor a single device, there will be no way to capture something like shift+mute, which involves both devices. Even if you run two actkbd daemons monitoring both, there’s no easy way for them to crosstalk. A possible workaround is to somehow combine the two devices into one, or even one further step back, to “redirect” all events from one device to the other. Maybe I was using the wrong keywords, but a google search doesn’t reveal any such things, so here comes the stitch works: a simple “event replicator” which basically is a stripped down version of actkbd. Just gcc evpipe.c -o evpipe, and use ./evpipe /dev/input/event{4,3} to redirect event4 to event3. NB: although I “grabbed” the incoming device, it’d be a good idea to redirect the hotkey device to the standard keyboard instead of the other way around, so that even if something goes wrong you’ll still have a usable device. (I made the mistake of not “grabbing” the incoming device AND redirected the normal keyboard to the hotkeys, making each keypress of “a” generating “aa” and ended up having to kill the process from my laptop).

    * evpipe.c: *
    * basically c-k c-y of linux.c in actkbd. *
    * NO ERROR HANDLING!! (almost) *

    #include #include

    static FILE *from;
    static FILE *to;
    struct input_event ev;
    int grabbed=0;

    int main(int argc, char* argv[]){
    /* Allow SIGTERM to cause graceful termination */
    void close_and_quit(int signum) {
    if (to)
    if (from)
    if (grabbed)

    int pipe(char* ffrom, char* fto){
    signal(SIGTERM, close_and_quit);
    signal(SIGINT, close_and_quit);
    from = fopen(ffrom, “a+”);
    if (from == NULL)
    return 1;
    to = fopen(fto, “a+”);
    if (to == NULL)
    return 2;

    while (get_key() == 0) {
    // printf(“.”);
    // fflush(stdout);

    int get_key(){
    int ret;

    do {
    ret = fread(&ev, sizeof(ev), 1, from);
    if (ret < 1) return -1; /* read error */ } while (ev.type != EV_KEY); return 0; } int snd_key() { int ret; ret = fwrite(&ev, sizeof(ev), 1, to); fflush(to); if (ret < 1) return -2; /* write error */ return 0; } int grab_dev() { int ret; if (grabbed) return 0; ret = ioctl(fileno(from), EVIOCGRAB, (void *)1); if (ret == 0) grabbed = 1; // else // lprintf("Error: could not grab %s: %s\n", device, strerror(errno)); return ret; } int ungrab_dev() { int ret; if (!grabbed) return 0; ret = ioctl(fileno(from), EVIOCGRAB, (void *)0); if (ret == 0) grabbed = 0; // else // lprintf("Error: could not ungrab %s: %s\n", device, strerror(errno)); return ret; } [/sourcecode] Before I forget, there's also a ruby evdev binding. I myself am not using the code above, nor the ruby evdev binding, since I don’t urgently need shift + VolUp, etc.

Written by zsh

March 26, 2009 at 11:28 pm