[Macbook Pro 6,2] Power mode not updating when switching from AC to BAT.

[mightyllama89]

[x] I've read and accepted the Bug Reporting Howto
[x] I've attached all required tlp-stat outputs via Gist (see below)

Describe the bug
When tlp is already running and in AC mode, if you unplug from power the power mode remains unchanged while the power source is updated and accurate.
Mode = AC Power source = battery

I have found the bug. It resides in tlp-func-base, line 789.
is_uint "$wait" 2 || wait=8
The timeout is too short. Changing it to 10 fixed it for me.

Expected behavior
Power_mode should switch to "battery"
A clear and concise description of what you expected to happen.

To Reproduce

Find an ancient 2010 Macbook Pro. Plug it in to AC. Then unplug it.

Additional context
tlp-stat output:
https://gist.github.com/mightyllama89/6b25293c4327fdb4245d36abe67ee0e8

[linrunner]

Hi,

i've noticed your pull request too. Nevertheless i need to check some basic assumptions because there is already a workaround for MacBooks in the code, see tlp-func-base.in#L736.

Quirk: MBP 5.3 reports AC offline when started with charger connected

I would like to understand if this is still the case. Because if the AC status were correct, we would not need the delay loop at all.

So with TLP 1.4.0-beta.1 installed: please power off your MacBook, make sure AC is connected, power on, boot and show:

sudo tlp-stat --psup

[mightyllama89]

The MBP quirk doesn't look necessary on mine. AC status is reported correctly as 'Online' when booting with the power supply plugged in. From what I can tell the quirk is unrelated to the need for the wait time. The wait time is needed anytime the battery status doesn't update immediately regardless of manufacturer. I have attached the information you requested in addition to info on arriving at the fix I suggested.

https://gist.github.com/mightyllama89/31e790ac2374e2125fa59b8814fea508

[linrunner]

From what I can tell the quirk is unrelated to the need for the wait time.

Yes, we do: if the AC status is wrong we can't use it to determine the actual power source. Instead we need a reliable BAT status, therefore the wait time is necessary.

When your MacBook doesn't have the quirk, you may configure

TLP_PS_IGNORE="BAT"

and use the AC status directly, without waiting time.

Unfortunately, one MacBook reporting the status correctly is not enough evidence to remove the quirk completely.

[mightyllama89]

I am sorry if there is confusion.

Unfortunately, one MacBook reporting the status correctly is not enough evidence to remove the quirk completely.

I was not recommending that the quirk be removed.

Because if the AC status were correct, we would not need the delay loop at all.

You asked me to test if the quirk was still necessary. I only mentioned the quirk to clarify that the workaround was not necessary in this case. However the quirk is what makes it check the battery status. If you keep the quirk then the wait time should be changed.

I went through the commits and issues and see that I am not the first one to have this kind of problem.
Issue #443
Commit 338068e

Commit c0c02d8 probably broke the fixes for those two. It looks like this quirk applies to more than just the MBP. The way you currently have it the program checks for AC status immediately upon the udev change. You are assuming that the AC status will update immediately and that any status not online must mean battery. Only AC type ADP1 gets to run any of the battery checks.

In conclusion, it takes longer than 0.8 seconds for the battery status to update. Increasing the wait time to 15 fixes it for me and probably others too. I spent a lot of time figuring out my issue and thought I'd try to contribute what I found out.

[linrunner]

If you keep the quirk then the wait time should be changed.

Right.

You are assuming that the AC status will update immediately and that any status not online must mean battery.

In fact this assumption holds for almost all laptops. For some of the remaining, the quirk solves the problem without users having to configure TLP_PS_IGNORE="AC".

But if the battery status is used to determine the power source, the wait time is needed because the battery status won't change immediately. This is also necessary for TLP_PS_IGNORE="AC".

The changes you list were a consequence of the original quirk, which applied to all ac devices, not just ADP1. Your problem is also a consequence of the quirk. I'm taking this opportunity to rethink the sense and necessity of the quirk as a whole. So i'm grateful for your input.

Because the wait is needed in case of TLP_PS_IGNORE="AC", your pull request makes sense anyway.

I would be happy if you stay tuned here to re-test in case i decide to remove the quirk completely.

[linrunner]

ps. i forgot to mention why i don't like long wait times: udevd kills tasks started with a RUN rule after 4-5 sec, 1.5 sec would already be a third of that.

[mightyllama89]

In fact this assumption holds for almost all laptops.
That is good to know.

Looking at all the commits regarding power supply detection I could tell you hadn't made up your mind on how exactly you wanted it done. I was hoping it would help you if I shared what found out.

consequence of the original quirk, which applied to all ac devices,

The question then becomes how many devices can you not trust the AC offline status. Changing it from all devices (not a quirk) to only ADP1 is a fundamental change and only more testing will tell.

why i don't like long wait times

That makes sense. The status update delay actually varied on mine. Luckily your log reports the time remaining on the wait when it exits the loop. The longest I saw was 1 second so I added .5 more just to be safe. So you might be able to shorten it if you really want to.

Unrelated but probaly not worth opening an issue: in tlp-func-base after line 746 I think you need a ';;' for the case/switch.

Let me know if you need anything else.

[linrunner]

Please test again with 1.4 beta 2 --> https://download.linrunner.de/packages/

Looking at all the commits regarding power supply detection I could tell you hadn't made up your mind on how exactly you wanted it done. I was hoping it would help you if I shared what found out.

Oh, I know where I want to go. Unfortunately, there are conflicting requirements due to a variety of hardware quirks. Also, I don't want too many users to have to adjust their existing configuration. Some - hopefully few - will have to configure

TLP_IGNORE_PS="AC"

retroactively (not you).