[BUG] Intermittent black screen in GLKVM Desktop App v1.4.0 release 2 - kvmd.log shows “resolution: no signal” and encoder reinit on GL-RM10

Hello GL.iNet team,

I am reporting a reproducible/intermittent black screen issue seen from the GLKVM desktop application on Windows (Desktop App v1.4.0 release 2).

Important point: at first I suspected the desktop application itself, but after reviewing the KVM logs, the device-side logs suggest that the KVM is actually losing video signal momentarily and then reinitializing the capture pipeline.

Device / environment

  • Model: GL.iNet Comet Pro GL-RM10.
  • Firmware: V1.8.0 release4.
  • Kernel: Linux 6.1.118 aarch64.
  • Main device log available at: /var/log/kvmd.log.
  • The desktop client identifies itself in kvmd as: gl-kvm/1.4.0 Chrome/132.0.6834.210 Electron/34.5.8 Safari/537.36.

Symptom

  • From the desktop app, the screen occasionally turns black.
  • I am not the only one seeing this; another user I know is experiencing the same issue with the desktop app.
  • The problem appears intermittent rather than a permanent failure.

Why I think this is not only a desktop-app rendering issue
At the exact time of the black-screen event, /var/log/kvmd.log shows the local video pipeline reporting signal loss and reinitializing itself:

2026-05-07 07:22:20,667 - kvmd.apps.kvmd.streamer - INFO - => -- ERROR [0.000 tid=3766] -- get VENC data failed 1 times a004800e
2026-05-07 07:22:20,677 - kvmd.apps.kvmd.streamer - INFO - => -- INFO [0.000 tid=3787] -- resolution: no signal, deinit rv1126 encoder
2026-05-07 07:22:20,678 - kvmd.apps.kvmd.streamer - INFO - => -- INFO [0.000 tid=3787] -- Stopping VO deal thread
2026-05-07 07:22:20,771 - kvmd.apps.kvmd.streamer - INFO - => -- INFO [0.000 tid=3787] -- Stopping MJPEG fetch thread
...
2026-05-07 07:22:21,946 - kvmd.apps.kvmd.streamer - INFO - => -- INFO [0.000 tid=3787] -- resolution: change to 1920x1080@60, reinit vi venc

So from the KVM side, this looks like:

  1. Encoder data retrieval starts failing.
  2. The device reports “no signal”.
  3. It tears down the video pipeline.
  4. It detects 1920x1080@60 again and reinitializes capture.

Additional relevant context

  • When a new session starts, kvmd starts ustreamer normally on /dev/video0 and capture begins correctly.
  • The logs show normal streamer startup, “CAP: Capturing started”, and repeated “Requesting key frame” messages before the signal-loss event.
  • This does not look like a simple frontend-only black overlay, because the KVM itself reports a real signal drop/reset in the capture pipeline.

Why I think this should be investigated separately from previous app/cloud issues
I previously had a different issue where the Windows client could get stuck at 99%, which was traced to blocked STUN/signaling domains and not to the local video capture pipeline.
This new problem is different: the KVM log now explicitly reports “resolution: no signal” and encoder teardown/reinit.
So this seems to be a separate bug or instability path.

Questions for GL.iNet

  • Is this a known issue in the GL-RM10 / current firmware / desktop app combination?
  • Could this be caused by HDMI signal detection / EDID / hotplug handling / resolution renegotiation?
  • Is there any newer firmware, beta build, or debug mode to capture more detailed HDMI/input-state logs?
  • Can GL.iNet confirm what error code a004800e means in this VENC path?
  • Is there any recommended workaround to make the input signal handling more stable?

Useful technical details

  • Device model: GL-RM10.
  • Firmware: V1.8.0 release4.
  • Kernel: 6.1.118 aarch64.
  • Log source: /var/log/kvmd.log.
  • App user-agent seen in kvmd: gl-kvm/1.4.0 / Electron 34.5.8.
  • Signal-loss event observed at: 2026-05-07 07:22:20–07:22:21.
  • Stream recovered by itself after reinitializing to 1920x1080@60.

If needed, I can provide:

  • A larger /var/log/kvmd.log excerpt around the event.
  • Full reproduction steps.
  • Information about the HDMI source, cable, adapter, and display path.

Hi dani10gl, would you mind get me the logs file? click [?Help → Export Log Files], thanks a lot