bug#57490: UPower ignores ‘critical-power-action’

["Ludovic Courtès" <ludo@gnu.org>]

Ludovic Courtès <ludo@gnu.org> skribis:

> Looking at the code, it could be because the ‘CanHybridSleep’ method
> returns false, but why that would happen is unknown to me.

Indeed, if we run ‘dbus-monitor --system’ and
‘herd restart upower-daemon’ (in QEMU), we see this:

--8<---------------cut here---------------start------------->8---
method call time=1661890192.586471 sender=:1.40 -> destination=:1.39 serial=16 path=/org/freedesktop/UPower; interface=org.freedesktop.UPower; member=GetCriticalAction
method call time=1661890192.586862 sender=:1.39 -> destination=:1.1 serial=17 path=/org/freedesktop/login1; interface=org.freedesktop.login1.Manager; member=CanHybridSleep
method return time=1661890192.588676 sender=:1.1 -> destination=:1.39 serial=104 reply_serial=17
   string "na"
method call time=1661890192.589034 sender=:1.39 -> destination=:1.1 serial=18 path=/org/freedesktop/login1; interface=org.freedesktop.login1.Manager; member=CanHibernate
method return time=1661890192.591082 sender=:1.1 -> destination=:1.39 serial=105 reply_serial=18
   string "na"
method return time=1661890192.591405 sender=:1.39 -> destination=:1.40 serial=19 reply_serial=16
   string "PowerOff"
--8<---------------cut here---------------end--------------->8---

That is, elogind returns “na” to the Can* methods.

Same story on my actual laptop:

--8<---------------cut here---------------start------------->8---
$ dbus-send --print-reply --system --dest=org.freedesktop.login1 /org/freedesktop/login1 org.freedesktop.login1.Manager.CanHybridSleep
method return time=1661890748.184775 sender=:1.1 -> destination=:1.130 serial=253 reply_serial=2
   string "na"
$ dbus-send --print-reply --system --dest=org.freedesktop.login1 /org/freedesktop/login1 org.freedesktop.login1.Manager.CanHibernate
method return time=1661890756.999248 sender=:1.1 -> destination=:1.131 serial=254 reply_serial=2
   string "na"
$ dbus-send --print-reply --system --dest=org.freedesktop.login1 /org/freedesktop/login1 org.freedesktop.login1.Manager.CanPowerOff
method return time=1661890761.375007 sender=:1.1 -> destination=:1.132 serial=258 reply_serial=2
   string "yes"
--8<---------------cut here---------------end--------------->8---

This is not surprising since our ‘logind.conf’ reads:

--8<---------------cut here---------------start------------->8---
HybridSleepState=disk
--8<---------------cut here---------------end--------------->8---

… meaning that “hybrid sleep” attempts to suspend-to-disk¹, something
that’s not implemented yet in Guix System².

Setting ‘hybrid-sleep-state’ to '("mem") doesn’t help though:
“CanHybridSleep” still returns “na”.  I’m looking at ‘can_sleep_state’
in elogind without seeing why it doesn’t return true.

To be continued…

Ludo’.

¹ Per <https://man.voidlinux.org/logind.conf.5>.
² But it’s almost there! https://issues.guix.gnu.org/49475