hmc
Active Member
deleted
See post #4 - Binky
See post #4 - Binky
Last edited by a moderator:
op_power.sh advice: keyboard light automatic switching
- Thread starter hmc
- Start date
hmc
Active Member
damned, what's up here? After editing the message, it displays totally messed up....
will try again shortly.
will try again shortly.
Last edited by a moderator:
hmc
Active Member
Hi guys,
still working on my configbutton plugin for controlling the keyboard light.
The combination of plugin and scripts now works almost perfectly, except one thing:
If the Pandora is connected to mains power and running, and I "press" the power switch shortly to bring the Pandora into low-power mode, the keyboard light goes off, along with the screen. But when I "press" the power button shortly once more to bring the Pandora to life again, the keyboard light is not enabled again.
For the other situations (lid close/open, suspend/resume) the script works perfectly already.
Maybe someone who has experiences with this script can give me advice where I need to add the call of kbd_light_on() correctly. Of maybe I did not add kbd_light_off call at the proper position so that when switching the Pandora "off" the light state is not properly saved?
Again: this only applies to when the Pandora is connected to mains power. Power button handling if not connected to mains power works correctly (suspend/resume).
Here is my current version of op_power.sh.
I have added the two functions "kbd_light_on()" and "kbd_light_off()", and I have added calls to those functions in functions display_on(), display_off() and suspend(). The obvious addition of kbd_light_on() call in resume() did not have the desired effect, by the way!)
still working on my configbutton plugin for controlling the keyboard light.
The combination of plugin and scripts now works almost perfectly, except one thing:
If the Pandora is connected to mains power and running, and I "press" the power switch shortly to bring the Pandora into low-power mode, the keyboard light goes off, along with the screen. But when I "press" the power button shortly once more to bring the Pandora to life again, the keyboard light is not enabled again.
For the other situations (lid close/open, suspend/resume) the script works perfectly already.
Maybe someone who has experiences with this script can give me advice where I need to add the call of kbd_light_on() correctly. Of maybe I did not add kbd_light_off call at the proper position so that when switching the Pandora "off" the light state is not properly saved?
Again: this only applies to when the Pandora is connected to mains power. Power button handling if not connected to mains power works correctly (suspend/resume).
Here is my current version of op_power.sh.
I have added the two functions "kbd_light_on()" and "kbd_light_off()", and I have added calls to those functions in functions display_on(), display_off() and suspend(). The obvious addition of kbd_light_on() call in resume() did not have the desired effect, by the way!)
Code:
#!/bin/bash
. /usr/pandora/scripts/op_paths.sh
# XXX: better use lockfile (or something), but it's not in current firmware
test -e /tmp/op_power.lock && exit 2
touch /tmp/op_power.lock
highpow="$(cat /etc/pandora/conf/led.conf | grep HighPowerLED: | awk -F\: '{print $2}')"
lowpow="$(cat /etc/pandora/conf/led.conf | grep LowPowerLED: | awk -F\: '{print $2}')"
debug(){
return 1 # 0 when debugging, 1 when not
}
test -e $(grep /etc/passwd -e $(ps u -C xfce4-session | tail -n1 | awk '{print $1}')| cut -f 6 -d ":")/.lidconfig && lidconfig=$(cat $(grep /etc/passwd -e $(ps u -C xfce4-session | tail -n1 | awk '{print $1}')| cut -f 6 -d ":")/.lidconfig) # read lid conf. file if it exists
#powerbuttonconfig=$(cat $(grep /etc/passwd -e $(ps u -C xfce4-session | tail -n1 | awk '{print $1}')| cut -f 6 -d ":")/.powerbuttonconfig)
if [ -e /tmp/powerstate ]; then
powerstate="$(cat /tmp/powerstate)"
else
powerstate="on"
fi
debug && echo "powerstate=$powerstate"
suspend_net() {
hcistate=$(hciconfig hci0 | grep DOWN)
if [ $hcistate ]; then
echo "down" > /tmp/hcistate
else
echo "up" > /tmp/hcistate
hciconfig hci0 down
fi
wlstate=$(lsmod | grep -m1 wl1251)
if [ -z "$wlstate" ]; then
echo "down" > /tmp/wlstate
else
echo "up" > /tmp/wlstate
ifconfig wlan0 down
rmmod board_omap3pandora_wifi 2> /dev/null
rmmod wl1251_sdio wl1251
fi
}
resume_net() {
hcistate=$(cat /tmp/hcistate)
if [ "$hcistate" = "up" ]; then
hciconfig hci0 up pscan
fi
wlstate=$(cat /tmp/wlstate)
if [ "$wlstate" = "up" ]; then
/etc/init.d/wl1251-init start
fi
rm -f /tmp/hcistate /tmp/wlstate
}
kbd_light_on() {
# Switch on keyboard light, if it was enabled before
oldbrightkbd=0
if [ -f /tmp/oldbrightkbd ]; then
oldbrightkbd=$(cat /tmp/oldbrightkbd)
fi
if [ $oldbrightkbd -gt 0 ]; then
/usr/pandora/scripts/op_kbd_light.sh on
fi
}
display_on() {
kbd_light_on
echo 0 > /sys/class/graphics/fb0/blank
# only bother restoring brightness if it's 0
# (old kernel or user messed it up somehow)
brightness=$(cat $SYSFS_BACKLIGHT_BRIGHTNESS)
if [ $brightness -gt 0 ]; then
return 0
fi
maxbright=$(cat $SYSFS_BACKLIGHT/max_brightness)
oldbright=0
if [ -f /tmp/oldbright ]; then
oldbright=$(cat /tmp/oldbright)
fi
if [ $oldbright -eq 0 ]; then
oldbright=$(cat /etc/pandora/conf/brightness.state)
fi
if [ $oldbright -ge 3 ] && [ $oldbright -le $maxbright ]; then
/usr/pandora/scripts/op_bright.sh $oldbright
else
/usr/pandora/scripts/op_bright.sh $maxbright
fi
}
kbd_light_off() {
# Save state of keyboard light, then switch it off:
/usr/pandora/scripts/op_kbd_light.sh
echo $? > /tmp/oldbrightkbd
/usr/pandora/scripts/op_kbd_light.sh off
}
display_off() {
kbd_light_off
brightness=$(cat $SYSFS_BACKLIGHT_BRIGHTNESS)
if [ $brightness -gt 0 ]; then
echo $brightness > /tmp/oldbright
fi
kernel_major=`uname -r | cut -c 1`
if [ "$kernel_major" = "2" ]; then
echo 0 > $SYSFS_BACKLIGHT_BRIGHTNESS
fi
echo 1 > /sys/class/graphics/fb0/blank
}
lowPowerOn(){ #switch from normal to lowpower mode
display_off
pidlist=$(pstree -lpA | grep pnd_run.sh | sed -ne 's/.*(\([0-9]\+\))/\1/p')
for PID in $pidlist
do
kill -STOP $PID
done
suspend_net
cat /proc/pandora/cpu_mhz_max > /tmp/oldspeed
/usr/pandora/scripts/op_cpuspeed.sh -n 125
}
lowPowerOff(){ # switch from lowpower to normal mode
oldspeed=$(cat /tmp/oldspeed)
/usr/pandora/scripts/op_cpuspeed.sh -n $oldspeed
rm -f /tmp/oldspeed
display_on
resume_net
pidlist=$(pstree -lpA | grep pnd_run.sh | sed -ne 's/.*(\([0-9]\+\))/\1/p')
for PID in $pidlist
do
kill -CONT $PID
done
echo $highpow > /sys/class/leds/pandora\:\:power/brightness #power LED bright
}
display_on_with_checks() {
# after turning on the display, we don't want lowpower state
# (which could be active because of some races)
if [ "$powerstate" = "buttonlowpower" -o "$powerstate" = "lidlowpower" -o \
-e /tmp/oldspeed ]
then
lowPowerOff
else
display_on
fi
}
show_message() {
# TODO: check if desktop is visible; maybe use layer3?
xfceuser=$(ps u -C xfce4-session | tail -n1 | awk '{print $1}')
cmd="DISPLAY=:0.0 zenity --info --text \"$1\" --timeout 10"
su -c "$cmd" $xfceuser
}
suspend_real() {
delay=0
if ! [ -e /sys/power/state ]; then
# no kernel suspend support
return 1
fi
current_now="$(cat /sys/class/power_supply/bq27500-0/current_now)"
if [ $current_now -gt 0 ]; then
return 1
#don't suspend while unit is charging
fi
# can't suspend while SGX is in use due to bugs
# (prevents low power states and potential lockup)
if lsof -t /dev/pvrsrvkm > /dev/null; then
return 1
fi
if ! grep -q 'mmc_core.removable=0' /proc/cmdline; then
# must unmount cards because they will be "ejected" on suspend
# (some filesystems may even deadlock if we don't do this due to bugs)
mounts="$(grep "/dev/mmcblk" /proc/mounts | awk '{print $1}' | xargs echo)"
for mnt in $mounts; do
if ! umount $mnt; then
show_message "Could not unmount $mnt, using partial suspend only"
return 1
fi
done
swaps="$(grep "/dev/mmcblk" /proc/swaps | awk '{print $1}' | xargs echo)"
for swp in $swaps; do
if ! swapoff $swp; then
show_message "Could not unmount $swp, using partial suspend only"
return 1
fi
done
else
if [ ! -e /etc/pandora/suspend-warned ]; then
show_message "Pandora will now suspend.\n\n\
Please do not remove SD cards while pandora is suspended, doing so will corrupt them."
touch /etc/pandora/suspend-warned
fi
fi
# FIXME: fix the kernel and get rid of this
suspend_net
# get rid of modules that prevent suspend due to bugs
modules="$(lsmod | awk '{print $1}' | xargs echo)"
blacklist="g_zero g_audio g_ether g_serial g_midi gadgetfs g_file_storage
g_mass_storage g_printer g_cdc g_multi g_hid g_dbgp g_nokia g_webcam g_ncm g_acm_ms
ehci_hcd bridgedriver"
restore_list=""
for mod in $modules; do
if echo $blacklist | grep -q "\<$mod\>"; then
restore_list="$restore_list $mod"
rmmod $mod
delay=1 # enough?
fi
done
sleep $delay
sync
echo mem > /sys/power/state
# if we are here, either we already resumed or the suspend failed
if [ -n "$restore_list" ]; then
for module in $restore_list; do
modprobe $module
done
fi
display_on
resume_net
echo $highpow > /sys/class/leds/pandora\:\:power/brightness
# wait here a bit to prevent this script from running again (keep op_power.lock)
# in case user did resume using the power switch.
sleep 2
return 0
}
suspend_() {
# dim power LED
echo $lowpow > /sys/class/leds/pandora\:\:power/brightness
kbd_light_off
if suspend_real; then
# resumed already
powerstate="on"
else
lowPowerOn
fi
}
resume() {
if [ "$powerstate" = "on" ]; then
# nothing to do
echo "resume called unexpectedly" >&2
else
lowPowerOff
fi
}
shutdown(){ # warns the user and shuts the pandora down
xfceuser=$(ps u -C xfce4-session | tail -n1 | awk '{print $1}')
time=5
countdown () {
for i in $(seq $time); do
precentage=$(echo $i $time | awk '{ printf("%f\n", $1/$2*100) }')
echo $precentage
echo "# Shutdown in $(($time-$i))"
sleep 1
done
}
countdown | su -c 'DISPLAY=:0.0 zenity --progress --auto-close --text "Shutdown in X" --title "Shutdown"' $xfceuser
if [ $? -eq 0 ]; then
/sbin/shutdown -h now
else
su -c 'DISPLAY=:0.0 zenity --error --text "Shutdown aborted!"' $xfceuser
fi
}
if [[ "$2" == "" ]]; then
if [[ "$1" -le 2 ]]; then # power button was pressed 1-2sec, "suspend"
if [[ "$powerstate" == "buttonlowpower" ]]; then
(debug && echo "resume") || resume
powerstate="on"
elif [[ "$powerstate" == "on" ]]; then
powerstate="buttonlowpower"
(debug && echo "suspend") || suspend_
elif [[ "$powerstate" == "liddisplayoff" ]]; then
powerstate="buttonlowpower"
(debug && echo "suspend") || suspend_
fi
elif [[ "$1" -ge 3 ]]; then # power button was pressed 3 sec or longer, shutdown
if [[ "$powerstate" == "on" ]]; then
(debug && echo "shutdown") || shutdown
fi
fi
elif [[ "$2" == "lid" ]]; then
if [[ "$1" == 0 ]]; then # lid was opened
if [[ "$powerstate" == lid* ]]; then
case "$lidconfig" in
"lowpower")
(debug && echo "resume") || resume
powerstate="on"
;;
*)
(debug && echo "display_on") || display_on_with_checks
powerstate="on"
;;
esac
fi
elif [[ "$1" == 1 ]]; then # lid was closed
if [[ "$powerstate" == "on" ]]; then
case "$lidconfig" in
"shutdown")
(debug && echo "shutdown") || shutdown
;;
"lowpower")
powerstate="lidlowpower"
(debug && echo "suspend") || suspend_
;;
*)
(debug && echo "display_off") || display_off
powerstate="liddisplayoff"
;;
esac
fi
fi
elif [[ "$2" == "screensaver" ]]; then
# warning: don't try to interact with X or do real suspend here -
# will cause various deadlocks
unset DISPLAY
if [[ "$1" == 0 ]]; then # deactivate screensaver
display_on_with_checks
powerstate="on"
elif [[ "$1" == 1 ]]; then # activate screensaver
display_off
fi
fi
debug && echo "powerstate=$powerstate"
echo "$powerstate" > /tmp/powerstate
rm -f /tmp/op_power.lock
Last edited by a moderator:
Neelix
Insecticidal Maniac
You need to call your kbd_light_on function from the LowPowerOff function.
Edit: (or in the same section of the code that calls it)
- Neelix
Edit: (or in the same section of the code that calls it)
- Neelix
Last edited by a moderator:
hmc
Active Member
But I do (indirectly):
I call it from display_on(), which is called from lowPowerOff()...
Just to make it sure, I added the kbd_light_on call to lowPowerOff directly, but it indeed makes no difference.
Something else must be wrong.
I call it from display_on(), which is called from lowPowerOff()...
Just to make it sure, I added the kbd_light_on call to lowPowerOff directly, but it indeed makes no difference.
Something else must be wrong.
Neelix
Insecticidal Maniac
have you confirmed that /tmp/oldbrightkbd exists and has a non-zero value after you have triggered this situation?
Edit:
I think I have it:
kbd_light_off() gets called from suspend()
suspend() then calls LowPowerOn, which Calls display_off() which again calls kbd_light_off()
because the light is already off when kbd_light_off() is called the second time it saves the off state instead of the previous on state.
- Neelix
Edit:
I think I have it:
kbd_light_off() gets called from suspend()
suspend() then calls LowPowerOn, which Calls display_off() which again calls kbd_light_off()
because the light is already off when kbd_light_off() is called the second time it saves the off state instead of the previous on state.
- Neelix
Last edited by a moderator:
hmc
Active Member
Yes, you're right.<br /><br />However, when removing it, light stays on when suspending without mains connected.<br /><br />If I put the call into the "if suspend_real; then" branch, then it's funny: When on mains and switching off, light stays on, and then switching the unit on again, light goes off. Probably because kbd_light_off is only executed <i>after</i> suspend_real.<br /><br />So I now put the call into suspend_real (at the beginning, behind all those "return" conditions), and now it works 
<br /><br />Thanks a lot, Neelix.<br /><br /><br />By the way: The behavior before the fix was even different when on mains and the unit was charging vs. when on mains but NOT charging (battery full).<br />I have tested both situations now, and they seem to work well.<br /><br />Everything's fixed now.<br />Now on to creating an installer PND.<br /><br /><br />P.S.:<br />Just encountered another situation that may need fixing, but please comment on this before I try to fix something, because I'm not sure how stuff works.<br />If I close the Pandora, connected to mains power, display goes off and keyboard light goes off (OK).<br />If I open it, display and light are switched on again. (OK)<br />But if I let it alone for some time and then open it again, screen is switched on but keyboard light is not. I may be wrong with this, as it was only an observation and I wasn't sure anymore if I really switched on the light before closing the lid. Needs reproducing to be sure.<br />Is there a timer, that executes some "suspend" function or so, while the unit sits there with a closed lid? That would again lead to double execution of kbd_light_off with storage of lastmode "0"...<br />
<br /><br />Thanks a lot, Neelix.<br /><br /><br />By the way: The behavior before the fix was even different when on mains and the unit was charging vs. when on mains but NOT charging (battery full).<br />I have tested both situations now, and they seem to work well.<br /><br />Everything's fixed now.<br />Now on to creating an installer PND.<br /><br /><br />P.S.:<br />Just encountered another situation that may need fixing, but please comment on this before I try to fix something, because I'm not sure how stuff works.<br />If I close the Pandora, connected to mains power, display goes off and keyboard light goes off (OK).<br />If I open it, display and light are switched on again. (OK)<br />But if I let it alone for some time and then open it again, screen is switched on but keyboard light is not. I may be wrong with this, as it was only an observation and I wasn't sure anymore if I really switched on the light before closing the lid. Needs reproducing to be sure.<br />Is there a timer, that executes some "suspend" function or so, while the unit sits there with a closed lid? That would again lead to double execution of kbd_light_off with storage of lastmode "0"...<br />
Neelix
Insecticidal Maniac
You're welcome.Yes, you're right.
However, when removing it, light stays on when suspending without mains connected.
If I put the call into the "if suspend_real; then" branch, then it's funny: When on mains and switching off, light stays on, and then switching the unit on again, light goes off. Probably because kbd_light_off is only executed after suspend_real.
So I now put the call into suspend_real (at the beginning, behind all those "return" conditions), and now it works
Thanks a lot, Neelix.
I'm not sure... What is your lid switch mode set to?
I suspect these two are related. I have a funny feeling that when you close the lid with the power plugged in but during a discharge cycle, it tries to suspend because it's not charging but fails to actually do so because it's connected to power.
I'm not exactly sure what actually happens in that situation. Perhaps someone more familiar with the systems involved would be able to shed some light on the subject.
- Neelix
Similar threads
