[bug?][10ii][5.0.0.73] Slow wakeup with yellow LED

REPRODUCIBILITY: intermittent
OS VERSION: 5.0.0.73
HARDWARE: Xperia 10II
UI LANGUAGE: -
REGRESSION: Seemingly

DESCRIPTION:

Recently I have noticed my device if often very slow to wake up, the screen remains blank for several long seconds (maybe tens of seconds, I never measured it) and the notification LED turns yellow.

I am not sure if this is related to the recent EA release of 5.0.0.73 or some other issue, I have tried to observe what happens over a USB cable with htop but was unable to catch the behaviour like that.

Any pointers on how to investigate this more are very welcome.

(I have rebooted and hard reset the phone all to no effect)

2 Likes

I have that sometimes on my XA2.
Rarely, like once over 2 month.
I panic, frenetically press everywhere, it does nothing, i wait, it works :melting_face:.
No idea from where it comes, though.

Currently it is happening multiple times every day, my gut says that it happens when the phone is not plugged in an there is no program running preventing deep sleep, but I can’t prove that without knowing how I can document what was running when I was trying to unlock the phone.

Did you try to look for some logs?
Here some doc. about this subject:

1 Like

Would be nice to know what the LED color means.

It sounds a bit like the UI restarting, because that always takes really long on the X10 II.

You mean factory reset, I guess?

Logs would be crucial I guess (after ruling out the usual suspects with htop or Crest).

Also all filesystems’ verbose checks.

I’ll try to get some logs.

This also happened during a phonecall so not sure if my earlier guess about sleep states was correct.

I did not mean factory reset, I meant the forced action where you hold all buttons.

/Edit - my battery is also draining much faster than it used to.

A hard poweroff, I suppose.

Wait, during phonecalls? How does that work? How do you even know there’s a problem?

If you did not reset the device it could well be that some stray, maybe recurring process is taking up all resources. The battery drain would speak for that, too. Use something like Crest or htop+ssh to analyze.

I use a bluetooth headset, wanted to see the time or activate the flashlight during the call for whatever reason.

I tried to turn on htop and nothing happened while it was running, the paranoid in me is starting to get worried about malware, the other side says I didn’t do any meaningful changes to the phone recently to incur this.

But the phone call continues normally? That’s interesting. Not sure what exactly it means though.

You really need to catch the right moment.

Yeah the phonecall was running as expected.

Maybe phonecalls can be handled in deeper sleep than podcast listening.

This continues with 5.0.0.76 (maybe a tiny bit faster than before but that is a subjective feeling not measured).

Based on htop system load while the screen is off is roughly 1.2-1.4 with a spike to 2.4 which I suspect happened during scheduled mail checks.

Based on powertop there are 450-900 wakeups/s while the screen is off.

Screen also comes on very slowly when I get called to show me who is calling.

And it doesn’t tell you what causes these wakeups?
In that case - in any case really - you might want to enable persistent logging (I’d reboot just to be sure) and check what is happening there.

It does, I didn’t see weird processes in the list which is the recipe for not remembering….

I’ll run powertop again later.

There’s a bunch of Timer and kWork lines that don’t have any further info.

[PID ] /usr/bin/lipstick -plugin evdevtouch -plugin evdevmouse -plugin evdevkeyboard:keymap=/usr/share/qt5/keymaps/droid.qm

and

[PID ] /usr/bin/voicecall-ui -prestart

Are consistently in the top 5.

I forgot to mention that journalctl output will need to be parsed as superuser. But it looks like you did that.

Anyhow

journalctl -b | grep -F 'voicecall-ui -prestart'

returns nothing for me.

Neither does the first line, but if I cut it down a little I get quite a few:

journalctl -b | grep -F '/usr/bin/lipstick -plugin evdevtouch -plugin evdevmouse' | wc -l
59

HOWEVER I’m on SFOS 5.0.0.72, not the early .73

That was output from powertop, journalctl shows mostly battery related messages.

Ah.

FWIW, lipstick is usually the #1 consumer on my phone, at least while idle.

Not voicecall-ui -prestart though. It just sits in the background at basically 0% CPU/RAM.

So unless you were using it while measuring - hmm, how does powertop work actually? Does it collect data over a long time?

WRT journalctl, yes, it logs a lot of uselss crap and it’s really hard to see the relevant stuff sometimes. I recommend using | grep -v ... to filter some of it out. Should work for journalctl -f too.