Repository navigation
Replies: 9 comments
|
Roman, this is an awesome write-up, thank you!! The comparison table and the WOW log lines make the problem really clear. Keeping USB plugged in but dropping only its network session before standby is a clever find. It fits the project's style as an opt-in setting, the same way Wi-Fi as a backup and Block updates harder are off unless you turn them on. I'm not promising it yet though. Before deciding, I want to see how it would sit next to Two things would help in the meantime:
I'll take a proper look at the reference and post here once I know which way it'll go. Thanks again for doing the legwork and keeping it sanitized!! |
|
Thanks! I retested this on v2.10.1 as requested. Same LG OLED65G64LW, webOS 26 / 11.2.0, firmware 43.21.98, RTL8153 on the TV's native r8152 driver. I installed the official v2.10.1 IPK and restarted the TV. I checked the actual script and boot hook, and confirmed that a new --watch process had started automatically. No extra driver module, Realtek mode off, Wi-Fi fallback off. With my USB/Wi-Fi handoff helper stopped, USB was primary and Wi-Fi was connected before standby. The ASUS RT-AX88U router no longer saw the TV associated with Wi-Fi at the first standby check, and it was still absent about seven minutes later. One press of my usual iPhone wake button did not wake the screen within a minute. The remote control brought the TV back. The same v2.10.1 watch process was still running, and network wake was still enabled. I then restored my unchanged, corrected helper and repeated the test. Wi-Fi remained associated for more than 21 minutes of standby. One press of the same iPhone wake button woke the screen. The helper restored USB as primary automatically, with Wi-Fi connected too. There were two transient recovery failures during resume before USB restoration succeeded, without manual intervention. I kept the driver and saved settings unchanged between the two tests. Only my USB/Wi-Fi handoff helper was stopped and restarted. Both used the same phone-triggered wake button on my ASUS RT-AX88U router; Cloud did not send any extra wake packets. The corrected helper also started automatically after the TV restart, with one instance running. Multiple real standby cycles have worked, but I am still testing it during normal use. I will post another update here in a few days, after more standby and wake cycles, including longer overnight standby. I have not tested this on other TV models or webOS versions. So the original failing case is still reproducible on v2.10.1 on my G6. I have left v2.10.1 installed and my helper enabled for normal use. Thanks again, |
|
Perfect, thanks Roman!! That's a really clean A/B: same install, same settings, same wake button, and only your helper turned on and off. So it's confirmed on v2.10.1, and nothing that changed since 2.9.3 helped. Noted on the two transient failures during resume before USB came back. That's the part I'd want to understand best before anything like this goes in, since an opt-in that sometimes leaves someone on Wi-Fi after wake would be a step back for most people. If your helper logs what failed in those two tries, that'd be great to see in your next update. No rush on the multi-day results. I'll wait for those before deciding which way this goes. |
|
Roman, it's in!! v2.11.0 adds Wi-Fi wake in standby ( How it works:
Turn it on in the app (Wi-Fi wake in standby) or with If you can run it through a few standby cycles and a full restart, the log ( |
|
Thanks for adding this so quickly! I've installed v2.11.0 on my LG G6 and enabled Wi-Fi wake in standby. I stopped my helper and removed its boot hook from init.d before testing. After a full restart, the official boot hook and power-state subscription started automatically. The first standby/wake test passed:
This is one successful cycle so far. I'll continue testing over the next few days, including longer standby periods and repeated wakes, and report back with the results and relevant logs. Thanks again! |
|
Awesome, thanks Roman!! That's exactly what I was hoping to see, and thanks for checking the route and the marker yourself too. On try 1: each try first checks the adapter has a cable link, and if it does, connects it and waits up to 10 seconds for an address. So try 1 either ran before the adapter had its link back after standby, or it didn't get an address in those 10 seconds. My guess is the first one, since standby resets the USB ports and the adapter takes a moment to come back, but you're right that the log doesn't say which lol. Looking forward to the longer runs. No rush!! |
|
Quick follow-up Roman: v2.11.2 is out, and now every reconnect try that fails gets its own line in the log with the reason. It'll say no |
|
Follow-up from my LG G6, now on official v2.11.19, using the stock r8152 driver with WATCH and Wi-Fi wake enabled. My custom helper remains stopped and its boot hook remains outside init.d. Today's controlled test (13 November, Kyiv time) passed:
Relevant log lines, 13 November, EET: 19:18:57 awake: try 1 of 10: the adapter has no cable link yet Independent checks at 19:19:17 and 19:20:44 both confirmed the default route on USB eth0, Wi-Fi still connected with a separate working address, no pending handoff marker, unchanged settings and no leftover settings temporary files or lock. Earlier observations, kept separate by version:
These observations span several releases. Today's v2.11.19 result is one successful controlled cycle, not three days of stability testing on that version. The original two generic failures from the old custom helper still have no confirmed cause; the newer cable-link diagnostics do not establish their cause retrospectively. Thanks again! |
|
Roman this is awesome, thank you!! Seriously the most careful testing anyone's done on this lol. Great to see it all come back on its own after a full restart on the official build, with your helper out of the picture. And the try 1 / try 2 pattern makes sense now: on wake the G6 brings the USB port back a few seconds after webOS starts the resume, so try 1 finds no cable link yet and try 2 gets it. That's what the retries are there for, so it's working as intended. Also thanks for confirming the #26 fix held through the update, restarts and the suspend test. I'll close that one. Totally fair point that this is one clean cycle on 2.11.19 and not days of testing. If it ever misses a wake, the log lines will say exactly which step it was stuck on, so just drop them here and I'll dig in. The G6 is already listed as tested in the README and the Wi-Fi wake docs. |
Uh oh!
There was an error while loading. Please reload this page.
What are you trying to do?
Following up on compatibility report #11. USB Ethernet works well on my LG G6, but I also need to wake the TV over Wi-Fi while leaving the adapter plugged in.
The intended setup is simple: USB Ethernet carries the main traffic while the TV is on, Wi-Fi stays connected for wake, and the USB adapter is never physically removed.
The practical goal is reliable remote access to my TV when I am away from home. Before I can connect to the TV and manage it remotely, I need to wake it from standby through my ASUS RT-AX88U router over Tailscale. If Wi-Fi stops listening while the USB Ethernet adapter is plugged in, I cannot reliably wake the TV without someone using the remote control at home. My standby helper addresses that obstacle while keeping USB Ethernet as the main connection when the TV is awake. The profile-selection patch makes this helper choose the connected Wi-Fi network correctly when ConnMan lists several profiles.
Environment
The standby problem
With USB Ethernet as the main connection and Wi-Fi connected, the TV could stop being associated with my ASUS RT-AX88U router after a longer standby. WoL packets sent to the Wi-Fi MAC then failed to turn on the screen. The same sender and target worked in the Wi-Fi-only control test.
I compared these cases:
The driver log also showed these events before deep suspend:
This points to a TV-side network/power interaction, but I have not established the exact internal policy. It is not evidence that the r8152 driver or LG-USB-2-ETH0 itself is at fault.
Why I added my own helper
Sending more wake packets could not preserve the TV's Wi-Fi receive path. The working control suggested preparing the network state on the TV before suspend.
My helper listens to native webOS power-state events. On Active Standby it checks that the expected USB adapter and Wi-Fi are present, Wi-Fi is already connected and network wake is enabled. It records its own pending handoff, disconnects only the USB network session and arms Wi-Fi WOWL. On wake it restores the USB connection and makes it primary again. It does not send power-off or reboot commands, change Wi-Fi passwords, install drivers or modify firmware/firewall settings.
A bug I found in my first helper
After a full restart, ConnMan listed three Wi-Fi service entries for the same radio. My first selector wrongly required exactly one matching Wi-Fi entry. It rejected the list before disconnecting USB, and the same assumption prevented its resume recovery.
This was a bug in my separate helper, not in your script. The corrected selector:
The real service IDs are not included here because they encode the SSID. The attached reference uses dummy adapter identities and made-up profiles.
What was tested
I developed this helper with assistance from OpenAI Codex. I confirmed the physical TV behavior. The public reference has been sanitized and its 20 tests pass, but that dummy-identity copy has not been deployed to the TV.
What would help?
Would an explicitly opt-in standby handoff be useful, or would you prefer to keep this as a separate documented workaround?
I read CONTRIBUTING.md, AGENTS.md and docs/ai/GUIDE.md. I understand that the project uses POSIX sh, leaves Wi-Fi and network settings alone except for its existing opt-in features, and keeps its background checks cheap. My Node.js helper is therefore a reference for discussion, not a drop-in upstream patch or a request to add a new runtime dependency.
If the approach fits, it would need a maintainer-approved design, integration with your existing lifecycle, reversible behavior, busybox tests, documentation and the usual project checks. I have prepared the sanitized source, the selector patch and the tests for review.
Thanks for taking a look,
Roman
LG_G6_WiFi_Wake_GitHub_Review.zip
Area
Script and boot hook
Your TV, if it matters
LG OLED65G64LW, webOS 26 / 11.2.0, firmware 43.21.98, RTL8153 with native r8152.
Before you send it
All reactions