on_ac_power possibly returning false when it should return true?

[jc00ke]

regolith-powerd/src/main.rs

Line 163 in ca281fc

if ! on_ac_power; then

I'm still trying to figure out why my desktop is suspending, and I think I've tracked it down to an issue with whatever is setting on_ac_power, though I'm not 100% sure.

Here's the discussion as to why I'm looking into this.

How might I go about debugging this? I'm running Regolith 3.0 that was just released.

What's leading me to this is that upower via DisplayDevice that I use for i3status-rust battery block seems to be returning empty. I'm not sure if this is directly related, but it might be.

[jc00ke]

In the meantime, how can I change my swayidle command?

[SoumyaRanjanPatnaik]

Can you confirm if this is the reason for the behavior you're observing by running this on a bash shell?

if on_ac_power ; then echo YES ; fi

[SoumyaRanjanPatnaik]

regolith-powerd/src/main.rs

Lines 125 to 139 in ca281fc

	let (idle_action_bat_opt, resume_action_bat_opt) = {
	use SleepInactiveBatteryType::*;
	match self.power_settings.sleep_inactive_battery_type() {
	Suspend => (Some("systemctl suspend"), None),
	Hibernate => (Some("systemctl hibernate"), None),
	Blank => (
	Some("swaymsg output * dpms off"),
	Some("swaymsg output * dpms on"),
	),
	Shutdown => (Some("poweroff"), None),
	Logout => (Some("gnome-session-quit --no-prompt"), None),
	Interactive => todo!("Show some gui prompt to the user"),
	Nothing => (None, None),
	}
	};

Regardless I don't think that to be the case. If you've disabled Automatic suspend, idle_action_bat_opt and resume_action_bat_opt should both be None.

[SoumyaRanjanPatnaik]

dconf dump /org/gnome/settings-daemon/plugins/power/

Can you dump the output for this command?

[SoumyaRanjanPatnaik]

In the meantime, how can I change my swayidle command

Uninstalling regolith-powerd should disable swayidle. But it is probably a hard dependency of regolith-sway-session.

[jc00ke]

Thanks for the debugging steps and potential workarounds. I won't be in the office until Tuesday; I'll debug then and report back.

[jc00ke]

Can you confirm if this is the reason for the behavior you're observing by running this on a bash shell?

if on_ac_power ; then echo YES ; fi

jesse@jimmie:~$ if on_ac_power ; then echo YES ; fi
jesse@jimmie:~$ if ! on_ac_power ; then echo NO ; fi
NO

[jc00ke]

dconf dump /org/gnome/settings-daemon/plugins/power/

Can you dump the output for this command?

➜ dconf dump /org/gnome/settings-daemon/plugins/power/
[/]
sleep-inactive-ac-type='nothing'

[jc00ke]

❯ on_ac_power
~
❯ echo $status
255

from man(1):

EXIT STATUS
       0 (true)  System is on mains power
       1 (false) System is not on mains power
       255 (false)    Power status could not be determined

[jc00ke]

Ooh, so a nasty hack would be on this machine to override on_ac_power to always exit with status 0... gross.

[jc00ke]

Seems like these caveats apply for those that use desktops:

CAVEATS
       The vast majority of desktops and servers have no power sensors,  thus  they  return  255  "un‐
       known";  for  most purposes you'd want to consider this answer to imply mains power — note that
       255 evaluates as false for the shell.

       This program currently provides only information known to the kernel; it doesn't  (yet?)  query
       userspace daemons such as nut which may provide further information.

[jc00ke]

Sorry for all the comments, I'm bouncing from this to work and back. I found this bug report, which makes me think that the short term fix is either to rely on something else (override on_ac_power) or change the logic to first use on_ac_power and if it's not conclusive (status 255) then use /lib/systemd/systemd-ac-power or something else.

[jc00ke]

In the meantime, how can I change my swayidle command

Uninstalling regolith-powerd should disable swayidle. But it is probably a hard dependency of regolith-sway-session.

Nope, don't think I want to do that 😆

➜ sudo apt remove regolith-powerd
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
The following packages were automatically installed and are no longer required:
  clipman gtklock i3status-rs kanshi libplayerctl2 light playerctl regolith-avizo regolith-displayd
  regolith-inputd regolith-sway-audio-idle-inhibit regolith-sway-background
  regolith-sway-control-center-regolith regolith-sway-dbus-activation regolith-sway-default-style
  regolith-sway-gaps regolith-sway-gtklock regolith-sway-i3status-rs regolith-sway-ilia
  regolith-sway-media-keys regolith-sway-polkit regolith-sway-screensharing regolith-sway-session
  regolith-sway-unclutter sway-audio-idle-inhibit trawldb xwayland
Use 'sudo apt autoremove' to remove them.
The following packages will be REMOVED:
  regolith-powerd regolith-session-sway regolith-sway-gsd regolith-sway-root-config
0 upgraded, 0 newly installed, 4 to remove and 0 not upgraded.
After this operation, 651 kB disk space will be freed.
Do you want to continue? [Y/n]

[SoumyaRanjanPatnaik]

Hey sorry. i've been quite busy for the last two weeks.

[SoumyaRanjanPatnaik]

Your findings are definitely helpful. I'll implement the fix you suggested. I think I might also look into downgrading regolith-powerd into a soft dep.

[SoumyaRanjanPatnaik]

Also, I see you don't have sleep-inactive-battery-type defined in your dconf dump. This means, the default value of 'suspend' is used whenever regolith-powerd queries gsettings for this property. So a temporary fix should be to set sleep-inactive-battery-type='nothing'.

[jc00ke]

Cool, I can do that. How do I do that? 🤣

[jc00ke]

Figured it out:

> dconf write /org/gnome/settings-daemon/plugins/power/sleep-inactive-battery-type "'nothing'"
# ----------------------------------------------------------------------------------^       ^
# notice the single quotes, which tells dconf this is a string

[jc00ke]

Current settings:

➜ dconf dump /org/gnome/settings-daemon/plugins/power/
[/]
sleep-inactive-ac-type='nothing'
sleep-inactive-battery-type='nothing'

[SoumyaRanjanPatnaik]

I'm hoping this workaround fixes your issue(for now). Regardless, this is a bug that requires fixing in general.

[jc00ke]

So far so good!