PowerStatus: Problem after 100% charged

[dsjonny]

I found a problem: using Samsung RV509 laptop the battery was charging. There was no problem until 99%. When it was fully charged, it has show 899%, and a huge last full charge value. I have changed the language to English, and start the program again, it started to charge again...

Before this, the design capacity was 4400mAh. Corrently I have no image about that. And it was started the charge again:

[dsjonny]

After I restart the app it show real values. But when it full charged, all value goes wrong.

[edglex]

I have this too on my lenovo x61 sometimes, and not only when I have a full charge. I am running a nightly from a few months ago though, I will have to update to a more recent build and check. However I probably wont do this in the very near future as I don't have time for an OS upgrade ATM.

[modeenf]

Replying to dsjonny:

After I restart the app it show real values. But when it full charged, all value goes wrong.

Does it do that allways? Restarting it will show the right amount even if its a full battery?

[ttcoder]

Saw something like that (or like edglex in comment:2) for a long time on my R61, though it occured some days, not always. When it occurs, the percentage can go as high as 85000% or some such (!). More often, the percentage oscillates between 0% and 100%.. Which is possibly a different bug (to file on component "acpi" instead of "powerstatus" I guess).

Since last week (circa rev 48100) I'm not seeing the 85000% issue much or at all but instead have issues that might warrant filing a new ticket: my syslog is getting spammed with this:

KERN: acpi: GPE Event 18 for \_GPE
KERN: Last message repeated 49 times.
KERN: acpi: GPE Event 18 for \_GPE
KERN: Last message repeated 53 times.
KERN: Last message repeated 54 times.
KERN: Last message repeated 45 times.

(about ten per minute, nothing too annoying)

...and mmlr's recent addition of "your battery is critical, please plug your device in immediately" is now popping up every minute or so, whenever my percentage goes down erratically to 0%..

Gotta find out how to disable it. (shame, cause it's interesting to have a notification when the battery level goes down! just wish acpi was reporting the real level, not an erratic one :-). Reportedly it can be configured in notification server. EDIT: Ok it can't (the list of apps is empty in the settings) but there's something else that can be done: quitting PowerStatus solves the notification problem. So back to the main issue at hand, why is ACPI mis-calculating in the first place..

[tqh]

The event spamming is because I've added more logging to find and enable more type of events. It is a temporary measure.

It seems the other issue can be an issue with the ACPI global lock that is shared with the firmware and deciding when it is safe to read values from firmware. However that usually generates a lot of garbage characters.

[stippi]

With the new notification, the problem has become a lot more apparent. For me, there seems to be a correlation between CPU load and how often I see the battery loading warning. It comes up pretty much constantly while I am compiling, almost never when I am not. Hopefully this is helpful.

[pulkomandy]

hrev48633 should fix this. Please test and report :)

[tqh]

Replying to ttcoder:

Since last week (circa rev 48100) I'm not seeing the 85000% issue much or at all but instead have issues that might warrant filing a new ticket: my syslog is getting spammed with this:

KERN: acpi: GPE Event 18 for \_GPE
KERN: Last message repeated 49 times.
KERN: acpi: GPE Event 18 for \_GPE
KERN: Last message repeated 53 times.
KERN: Last message repeated 54 times.
KERN: Last message repeated 45 times.

(about ten per minute, nothing too annoying)

That's because I added logging for ACPI events. It was turned on by default to see if it provides any useful info from users. It doesn't AFAICT so I should add a switch to turn event logging on and have it disabled by default. However, my time has been very limited.

[waddlesplash]

I think we can presume this fixed by now.