Broken suspend on several devices due to controller emulation

[0xDEADCADE]

A friend of mine recently asked me about an issue with their GPD Win Max 2 (Ryzen 6800U) running Bazzite. When pressing the power button, the device would attempt to suspend, then about half the time wake back up immediately.

After much testing of enabling/disabling different settings, it became clear that the gamepad emulation option in HHD was causing the issue. The issue appears only when enabling controller emulation, on any emulated controller (xbox, steam controller).

Looking through the logs of hhd at the time of suspend (sudo journalctl -f -u hhd), the following can be seen:

Journalctl output snippet

...
Nov 04 19:10:50 GrandpaDeck hhd[1274]:                 OSError: [Errno 19] No such device
Nov 04 19:10:50 GrandpaDeck hhd[1274]: GPDW   INFO     Launching emulated controller.
Nov 04 19:10:50 GrandpaDeck hhd[1274]: GPDW   INFO     Gamepadui closed. Disabling touchpad emulation.
Nov 04 19:10:50 GrandpaDeck hhd[1274]: GPDW   INFO     Found device 'i2c-10EC5280:00' at
Nov 04 19:10:50 GrandpaDeck hhd[1274]:                 /sys/bus/iio/devices/iio:device0
Nov 04 19:10:50 GrandpaDeck hhd[1274]: GPDW   INFO     Found device 'i2c-10EC5280:00' at
Nov 04 19:10:50 GrandpaDeck hhd[1274]:                 /sys/bus/iio/devices/iio:device0
Nov 04 19:10:50 GrandpaDeck hhd[1274]: GPDW   WARNING  Using cached controller node for Steamdeck Controller.
Nov 04 19:10:50 GrandpaDeck hhd[1274]: GPDW   INFO     Starting 'Steam Controller (HHD)'.
Nov 04 19:10:50 GrandpaDeck hhd[1274]: GPDW   INFO     Emulated controller launched, have fun!
Nov 04 19:11:08 GrandpaDeck hhd[1274]: GPDW   WARNING  Caching Steam Controller to avoid reconnection.
Nov 04 19:11:08 GrandpaDeck hhd[1274]: GPDW   ERROR    Assuming controllers disconnected, restarting after 0.3s.
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 Traceback (most recent call last):
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                   File
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 "/usr/lib/python3.13/site-packages/hhd/device/gpd/win/base.py",
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 line 183, in plugin_run
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                     controller_loop(conf.copy(), should_exit, updated, dconf,
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 emit)
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                     ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 ^^^
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                   File
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 "/usr/lib/python3.13/site-packages/hhd/device/gpd/win/base.py",
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 line 383, in controller_loop
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                     evs.extend(d.produce(r))
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                                ~~~~~~~~~^^^
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                   File
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 "/usr/lib/python3.13/site-packages/hhd/controller/physical/evdev
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 .py", line 433, in produce
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                     for e in self.dev.read():
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                              ~~~~~~~~~~~~~^^
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                   File "/usr/lib64/python3.13/site-packages/evdev/eventio.py",
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 line 71, in read
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                     events = _input.device_read_many(self.fd)
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 OSError: [Errno 19] No such device
Nov 04 19:11:08 GrandpaDeck hhd[1274]: GPDW   INFO     Launching emulated controller.
Nov 04 19:11:08 GrandpaDeck hhd[1274]: GPDW   INFO     Gamepadui closed. Disabling touchpad emulation.
Nov 04 19:11:08 GrandpaDeck hhd[1274]: GPDW   INFO     Found device 'i2c-10EC5280:00' at
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 /sys/bus/iio/devices/iio:device0
Nov 04 19:11:08 GrandpaDeck hhd[1274]: GPDW   INFO     Found device 'i2c-10EC5280:00' at
Nov 04 19:11:08 GrandpaDeck hhd[1274]:                 /sys/bus/iio/devices/iio:device0
Nov 04 19:11:08 GrandpaDeck hhd[1274]: GPDW   WARNING  Using cached controller node for Steamdeck Controller.
Nov 04 19:11:08 GrandpaDeck hhd[1274]: GPDW   INFO     Starting 'Steam Controller (HHD)'.
Nov 04 19:11:08 GrandpaDeck hhd[1274]: GPDW   INFO     Emulated controller launched, have fun!
...

The suspend in the above logs was attempted at 19:11:07, after which hhd immediately sees the physical controller as disconnected (which makes sense, as the handheld is suspending USB devices). HHD tries to re-initiate the virtual controller, which somehow cancels the suspend, causing the device to wake up again.

Another friend with an ROG Ally Z1 Extreme running Arch Linux also tested this, with the exact same issue (though I do not have logs from them). Turning off controller emulation resolves it, turning it back on causes suspend to fail randomly.

I'm unsure of the potential fix to this, it would likely involve detecting when the system wants to suspend, then waiting to run controller_loop in device/{model}/base.py until the system has fully resumed.

[antheas]

Please post the full log. This is not the problem. The controller disconnects before sleep

There is a bug currently with the Add TDP controls to /sys for Steam setting that can cause this. But it only causes this if you restart hhd too many times

[0xDEADCADE]

I do not currently have the full log, but will request it from my friend.

The physical controller does indeed disconnect before suspend, that is done intentionally by the kernel before S3 sleep.

As seen in the existing logs: ERROR Assuming controllers disconnected, restarting after 0.3s., which indicates HHD restarts controller_loop after 0.3s. This is the only event that directly relates to the device canceling the suspend. When disabling HHD controller emulation, this issue does not occur. The Add TDP controls to /sys for Steam option was not touched while testing. Toggling the controller emulation option consistently (with multiple tests of suspend) shows this behavior appearing and disappearing.

The friend with the ROG Ally Z1 Extreme rebooted his device directly before testing, and observed the same issue caused by the same setting (controller emulation). Thus, it cannot be caused by restarting hhd too many times.

[0xDEADCADE]

@antheas as requested, the full log: hhd.log

At 13:04:00 in this log, a successful suspend was performed.
At 13:04:30 in the log, an unsuccessful suspend happened.

[antheas]

journalctl -b > log.txt

[0xDEADCADE]

Output of sudo journalctl -b > log.txt: log.txt

[antheas]

Check your fingerprint sensor or another device that might cause a wakeup.

Nov 05 13:04:16 GrandpaDeck kernel: ------------[ cut here ]------------
Nov 05 13:04:16 GrandpaDeck kernel: i2c i2c-2: Transfer while suspended

This here has never caused a spurious wakeup before. We know how to fix it just have not gotten to it yet. If this is the case, it is your gyroscope.

[0xDEADCADE]

13:04:16 was the wake from the successful suspend. 13:04:30 is the unsuccessful suspend. That would be somewhere between here:

...
Nov 05 13:04:31 GrandpaDeck kernel: smpboot: CPU 3 is now offline
Nov 05 13:04:31 GrandpaDeck kernel: smpboot: CPU 2 is now offline
Nov 05 13:04:31 GrandpaDeck kernel: smpboot: CPU 1 is now offline
Nov 05 13:04:31 GrandpaDeck kernel: Spectre V2 : Update user space SMT mitigation: STIBP off
Nov 05 13:04:31 GrandpaDeck kernel: ACPI: PM: Low-level resume complete
Nov 05 13:04:31 GrandpaDeck kernel: ACPI: EC: EC started
Nov 05 13:04:31 GrandpaDeck kernel: ACPI: PM: Restoring platform NVS memory
Nov 05 13:04:31 GrandpaDeck kernel: AMD-Vi: Virtual APIC enabled
Nov 05 13:04:31 GrandpaDeck kernel: AMD-Vi: Virtual APIC enabled
Nov 05 13:04:31 GrandpaDeck kernel: LVT offset 0 assigned for vector 0x400
Nov 05 13:04:31 GrandpaDeck kernel: Enabling non-boot CPUs ...
Nov 05 13:04:31 GrandpaDeck kernel: smpboot: Booting Node 0 Processor 1 APIC 0x1
Nov 05 13:04:31 GrandpaDeck kernel: Spectre V2 : Update user space SMT mitigation: STIBP always-on
Nov 05 13:04:31 GrandpaDeck kernel: CPU1 is up
Nov 05 13:04:31 GrandpaDeck kernel: smpboot: Booting Node 0 Processor 2 APIC 0x2
Nov 05 13:04:31 GrandpaDeck kernel: CPU2 is up
...

It is not specifically a spurious wakeup, it's more a failed suspend. The issue disappears when disabling controller emulation in hhd.

[antheas]

It is a spurious wakeup. Sleep entry is successful. Check if it's the gyroscope by disabling it.

[0xDEADCADE]

The "Motion Support" and "Gyro Output Sync" options were toggled off to test, with controller emulation turned on as "Steam Controller" and "Xbox". This has the same issue.