Sleepcallback.go: deregister for sleep/wake notifications during shutdown

[Jerry1144]

By adding an additional info string:

diff --git a/sleepcallback.go b/sleepcallback.go
index a3bbcc8..6285739 100644
--- a/sleepcallback.go
+++ b/sleepcallback.go
@@ -161,5 +161,6 @@ func listenNotifications() error {
    if int(C.ListenNotifications()) != 0 {
        return fmt.Errorf("IORegisterForSystemPower failed")
    }
+   logrus.Info("listenNotifications() has returned.")
    return nil
 }

it is clear that listenNotifications() never returns in the happy path, not even during shutdown. This means the sleep listener CFRunLoop() was simply terminated, not given a chance to cleaned up: as per pwr_mgt/IOPMLib.h, the ports listening for sleep notifications "must" be closed by the caller. The code for cleaning up those also appeared in QA1340 listing 4. It's a memory leak.

You may have to investigate passing around the CFRunLoopRef of that thread to the Go side, so as to CFRunLoopStop it during exit, after which CFRunLoop() returns and clean up codes can be run. Or attach another source that sets an exit flag. Doesn't seem simple.

---

On a side note, IOPMLib.h mentions that kIOMessageSystemWillPowerOn is sent before most hardware has powered up, so if you print a log there, it would probably block until hard disks come back online. Maybe you can remove that callback.

[charlie0129]

Thank you for your helpful suggestions. I will try to fix them.

[yexua]

感谢您提供的有用建议。我会尽力修复它们。

看到了这个跟休眠有关系，我在使用batt这段时间，拔掉电源合盖休眠耗电好像有些高，不知道跟你们讨论的是否相关。 我使用该命令看到了一些日志 pmset -g log | grep -e "Sleep.*due to" -e "Wake.*due to"

[charlie0129]

@yexua I see. I think the log you provided is written by macOS and isn't related to batt (this project). The Batt appeared in the log is just a abbreviation of battery.

[yexua]

@yexua I see. I think the log you provided is written by macOS and isn't related to batt (this project). The Batt appeared in the log is just a abbreviation of battery.

我不是太懂这个，我的MBP是刚购买的，之前休眠待机耗电量几乎为1%。使用batt后，一天的休眠耗电量达到了4%左右，我不确认是否与这个有关系。

[charlie0129]

我不是太懂这个，我的MBP是刚购买的，之前休眠待机耗电量几乎为1%。使用batt后，一天的休眠耗电量达到了4%左右，我不确认是否与这个有关系。

Is your 1% (100%->99%) or otherwise (e.g. 80%->79%)? If it 100%->99% then it is expected because that 1% about the same as 5% on other batter levels (Apple lies about battey percentage). batt shouldn't affect sleep/hibernation battery usage if you close the lid. Can you rule out other variables by uninstalling batt?

[yexua]

我不太懂这个，我的MBP是刚购买的，之前的休眠功耗几乎为1%。使用电池后，一天的休眠功耗达到了4%左右，我不确定与这个是否有关系。

你的1%（100%->99%）还是其他（例如80%->79%）？如果它是 100%->99%，那么这是预期的，因为这 1% 与其他电池级别的 5% 大约相同（苹果在电池百分比方面撒了谎）。如果关闭盖子，batt 不应影响睡眠/休眠电池的使用。能否通过卸载batt来排除其他变量？

我可以试试卸载后观察一下