rtcwake on Kobo?

[z000ao8q]

yes, I found it. thanks.
I will test with the following lines being commented out...

if (strcmp(suspend, "on") && !may_wakeup(rtcname))
bb_error_msg_and_die("%s not enabled for wakeup events", rtcname);

[z000ao8q]

what happens after the standby?
when I have WiFi and Telnet working and I am triggering the RTCWAKE command (which seems to work), do I need to reinitialize everything again after the wakeup - or should it work without doing anything?

If the answer is "everything should still work" then it might not work in my test yet because I cant reconnect telnet

[sven]

Originally Posted by z000ao8q

what happens after the standby?
when I have WiFi and Telnet working and I am triggering the RTCWAKE command (which seems to work), do I need to reinitialize everything again after the wakeup - or should it work without doing anything?

If the answer is "everything should still work" then it might not work in my test yet because I cant reconnect telnet

Normally everything should work without any re-initialization. Currently I do not use the rtc alarm to resume my Kobo from suspend, but I use the power button (quick slide, no boot) to get the Kobo running again. The WiFi and Telnet are instantly working after resume.

[Grunge]

ha, just figured out:

the kobo will not sleep when you have it plugged to usb (usb network, ssh session).
So rtcwake IS working, it was my mistake.

[z000ao8q]

on my Mini the RTCWAKE did work to set it to sleep - only the wakeup did not.
As Sven posted, everything should be there again.
on my side, I could not connect to telnet unless I rebooted the Mini.

I will try a little more today...

[z000ao8q]

it seems to work :-)
but only with re-init of Wifi.

I am running my test.sh in background with "nohup ./test.sh"

echo 1 > /sys/power/state-extended
sh /mnt/onboard/wifidn.sh
/mnt/onboard/busybox_kobo rtcwake -a -s 30
sleep 2
sh /mnt/onboard/wifiup.sh
zcat /mnt/onboard/kobo.raw.gz | /usr/local/Kobo/pickel showpic

and TA-DAAA - a Pic-Refresh and ability to reconnect to Telnet after 30 seconds.
need to do some more tweaking but looks promising :-)

[z000ao8q]

Hello,
just a happy update. It seems to work for me - even so I still need to run it for several days to test stability and runtime.

Basic idea:

• wakeup every hour
• execute Cron (at 1 min after the hour) to update display
• send to sleep until next full hour

As the cron is only executed between 7am and 8pm on my side, I had to go a little extra way. basically I am dividing the timestamp by 3600 seconds to geht the last full hour and then up to the next execution...
this is how my update.sh looks like (which is executed by cron)

sh /mnt/onboard/wifiup.sh

....update display...

sh /mnt/onboard/wifidn.sh

h=`date +%H`
timestp=$((`date +%s`/3600))
if [ "$h" -ge 20 ]; then
multi=24-$h
let timestp+=7
let timestp+=$multi
elif [ "$h" -lt 7 ]; then
multi=7-$h
let timestp+=$multi
else
let timestp+=1
fi
timestp=$((timestp * 3600))
/mnt/onboard/busybox_kobo rtcwake -t $timestp

[z000ao8q]

wakeup with seconds works perfectly - but using a time it is behaving strange.

hwclock is UCT: Thu Jun 27 19:20:31 2013 0.000000 seconds
date (system time) is local: Thu Jun 27 21:20:52 CEST 2013

I try:

timestp=`date +%s`
let timestp+=30
echo `date`
/mnt/onboard/busybox_kobo rtcwake -u -t $timestp -m mem

output:

Thu Jun 27 21:13:45 CEST 2013
wakeup from "mem" at Thu Jun 27 21:14:15 2013

no wakeup

rtcwake -l would give me

wakeup from "mem" at Thu Jun 27 21:14:15 2013

but also no wakeup

my assumption: rtcwake is using the hwclock
but when I adjust the system time backwards with hwclock -s -l (instead of -s -u like in rcS):

hwclock is UTC: Thu Jun 27 19:29:48 2013 0.000000 seconds
date is: Thu Jun 27 19:30:16 CEST 2013

Then I execute

timestp=`date +%s`
let timestp+=30
echo `date`
/mnt/onboard/busybox_kobo rtcwake -a -t $timestp -m mem

output:

Thu Jun 27 19:29:03 CEST 2013
wakeup from "mem" at Thu Jun 27 19:29:33 2013

and yes - a wakeup

seems to work - but now the issue is, that cron does not work anymore because it is using the system time -which is now 2hrs off


any idea?

[Lucas Malor]

let timestp -= 120 ?

[Lucas Malor]

Or
date -u +%s

Or
man date
man hwclock

[z000ao8q]

I guess you did not try your answer before :-)
+%s obviously always outputs the utc format, no matter what additional parameter

[root@(none) /]# date -u +%s
1372441495
[root@(none) /]# date +%s
1372441500
[root@(none) /]# date -u
Fri Jun 28 17:45:04 UTC 2013
[root@(none) /]# date
Fri Jun 28 19:45:06 CEST 2013

and regarding "man": not installed on kobo

If i am right, the implementation of rtcwake has an issue. it compares against the local system time

sys_time = time(NULL);
if (alarm_time < sys_time)
bb_error_msg_and_die("time doesn't go backward to %s", ctime(&alarm_time));

and here is the issue when you want to set it to UTC and this one is 2hrs backwards. I think I read something about this somewhere already but cant find the article

I changed the timezone now to UTC so I only have one time for all clocks now.
now it works - finally

[auspex]

er, ALL the unix/linux man pages are in numerous places on the Internet

[z000ao8q]

a little update:
its running now stable sincs several days with 13 2min wakeups per day.
I am relying on the battery-percentage values of the kobo and the consumption of the mode "mem" (suspend-to-ram) seems to be ~3,5% per day.
this would give a runtime of around 1 month on a single charge.

I want to test the disk-mode next to see if it brings more value...

[sunda]

Hi all,

maybe I missed somthing, but I get the same error:
rtcwake: /dev/rtc0 not enabled for wakeup events

can I "enable" it!?

[z000ao8q]

Hello,

you need a tweaked version of the rtcwake command.
I still need to post it here. hope that I can do it today or tomorrow...