[tshering]

Originally Posted by ipse

Long story short...what is your recommendation for sleep mode? Power off whenever the reader is not used for a longer period? Looks like that was the conclusion of your test...in the second link and pointing to the log.

This is actually what I am doing. But I was doing so already before I had KSM. If I did not follow this style like a religion, I would probably suspend the device if nickel is running, and power off if koreader is running.

Originally Posted by ipse

/EDIT I'll post a reference to the KOreader discussions (and potential solution) in case someone else stumbles across this post.

https://www.mobileread.com/forums/sho...postcount=2449

Even though I recommended myself once (see this post) to try switching off preventfreeze, I doubt that this does really help. As long as the system is in suspended state, preventfreeze is also inactive. I did a small test:

Spoiler:

This is the code of the original preventfreeze.sh:

while :
do
  sleep 1
done

There is an infinite loop. Inside of the loop "sleep 1" tells the kernel that this process does not need and cpu time for the next second.

This is the test script:

while :
do
  sleep 10
  echo "log at: $(date +%Y%m%d_%H%M%S)" >> $debug_logfile
done

The purpose is to log each run through the loop. In order to keep the log output small, I run the loop only once in ten seconds.

As for the test:

• I started the test script.
• After one minute, I put the device to sleep.
• After another minute, I woke the device up.
• After another minute, I terminated the test.

This is the log (plus one comment):

log at: 20151119_202724
log at: 20151119_202734
log at: 20151119_202744
log at: 20151119_202754
log at: 20151119_202804
log at: 20151119_202814
# suspended, no execution of the loop at all
log at: 20151119_202929
log at: 20151119_202939
log at: 20151119_202949
log at: 20151119_202959
log at: 20151119_203009
log at: 20151119_203019
log at: 20151119_203029