[Bug] Wireless central/master half fails to wake from deep sleep & drains battery on main (not present in v0.3)

[braun-steven]

Describe the bug
On a split keyboard (Lily58) with nice!nano v2 and nice!view displays, the Central (Master/Left) half fails to wake up from deep sleep and exhibits high battery drain. The Peripheral (Slave/Right) half wakes up correctly.

Reverting the ZMK revision from main to v0.3 in west.yml (specifically for the GitHub Actions manifest) completely resolves the issue, suggesting a regression in the recent main branch or its interaction with Zephyr 4.1.0/fixes.

To Reproduce

1. Setup a split keyboard (Lily58) with nice!nano v2 and nice!view displays.
2. Build firmware using the main branch of ZMK (Jan 15, 2026).
3. Enable Deep Sleep in config (CONFIG_ZMK_SLEEP=y).
4. Allow the keyboard to enter deep sleep (timeout).
5. Attempt to wake the keyboard by pressing a key on the Central half.
6. Result: Central half does not wake, display remains off, and battery drains rapidly (empties overnight), suggesting a crash or a hang in a high-power state instead of proper sleep.
7. Expected: Central half wakes up, display turns on, and it reconnects to the host.

Regression Context

• Working Version: v0.3
• Broken Version: main (current)

Hardware:

• Board: nice!nano v2
• Shield: Lily58 (Left/Central)
• Display: nice!view
• Connectivity: BLE

Additional Context

• CONFIG_ZMK_STUDIO was not enabled in the builds and this issue is thus not related to Central side does not wake from sleep when ZMK Studio is enabled (Zephyr 4.1) #3195 (I've also explicitly disabled it just in case to test if it mistakenly defaults to y).
• The issue persists regardless of the peripheral half's state.
• Flashing the right-half firmware to the Left board shows that wake-up detection works (the hardware registers the keypress and wakes), isolating the failure specifically to the Central role logic/firmware on main.
• The battery drain is significant enough that a full battery is depleted within ~24 hours of "sleeping".
• When the left half is connected via USB, deep sleep works and the half successfully wakes up from sleep.

Configuration (lily58.conf)

CONFIG_ZMK_SLEEP=y
CONFIG_ZMK_DISPLAY=y

West Manifest (west.yml)

manifest:
  projects:
    - name: zmk
      remote: zmkfirmware
      revision: v0.3 # Changing this to main causes the issue
      import: app/west.yml

ZMK Lily 58 Pro Config: https://github.com/braun-steven/lily58-wireless-view-zmk-config

[Alawari]

Seems like I have the same issue on a Chocofi since updating my firmware yesterday. Would be happy to contribute via testing if someone has ideas what to look for!

[BrunoReis-GTT]

I noticed similar issues with my Ferris Sweep. I just discovered the battery to be dead sometimes for no apparent reason. Or, a bit weirder, while using the keyboard with plenty of battery, it just goes dead (no keys response whatsoever) unless I push the reset buttons (and sometimes needing also to turn each half off and on again). I updated to main just before vacation and came back recently, so I think this might be related to @braun-steven's experience. I'l check how it goes withv0.3 and let you guys know.

[repparw]

Do y'all have a date of when you built from main? could help pinpoint a commit. afaict this didn't happen to me on an oct31 build but seems to happen now (both main)

[Alawari]

Do y'all have a date of when you built from main? could help pinpoint a commit. afaict this didn't happen to me on an oct31 build but seems to happen now (both main)

I have a build without the bug from Dec 8th (old Zephyr version) and another build from Dec 31th. The latter is using 4.1 and showed a similar bug - the central half occasionally lost connection and did't reconnect afterwards, though waking from sleep and then reconnect worked. This could also be the same bug in disguise, I guess.
A build from January 21st then showed the bug from this issue.

[BrunoReis-GTT]

Do y'all have a date of when you built from main? could help pinpoint a commit. afaict this didn't happen to me on an oct31 build but seems to happen now (both main)

Yeah, the issue definitively started for me after my build from December 10th. With my previous build (November 19th) I didn't notice anything similar. And just to confirm, after switching back to v0.3 4 days ago, my keyboard seems to be working normally again.

[repparw]

The bug most certainly came with adaption of the new Zephyr version (4.1, I think) or shortly after. I have a build without the bug from Dec 8th (old Zephyr version) and another build from Dec 31th.

any related issue upstream on https://github.com/zephyrproject-rtos/zephyr ?