watchthelight labs

mt7925-wifi-native: 2026-11-11 to 2026-11-13

Read from GitHub only. Nothing below is reproduced on our machine.

  • ForbiddenItlwm gives a non-Apple Wi-Fi card a working awdl0 on Tahoe and Sequoia, and AirDrop runs both ways on Intel cards. This is the macOS half of AWDL, which is what WTLWireless #8 (no AWDL interface) and phase C of our Wi-Fi plan are about. Its HOW-IT-WORKS page says: wifip2pd only asks for an AWDL interface if the card reports the AWDL capability; the driver then builds a subclass of Apple's IO80211VirtualInterface and IO80211AWDLProtocol whose virtual table matches Apple's slot for slot (568 slots on 25G241, checked by an audit script); Apple's own peer manager does elections, sync and service discovery; the driver supplies a second firmware context on channel 149 at 80 MHz that time-shares with the station, and a send gate that only transmits while the phone is on our channel. About 20 to 22 MB/s both ways on the author's AX211. It is built on the Realtek AirPort_RTW89 approach and on Intel firmware, so the radio side does not transfer to an MT7925 as is, but the interface side is the macOS piece we were missing. It was created 2026-11-10, has 135 stars, was written mostly with AI help according to its own README, and is GPL-2.0, so ideas only for us. Confidence: High that it works on Intel as claimed (several testers in its issues report two-way AirDrop); Unverified on our hardware. Links: https://github.com/retropineaple/ForbiddenItlwm/blob/main/docs/HOW-IT-WORKS.md , https://github.com/retropineaple/ForbiddenItlwm/releases , retropineaple/ForbiddenItlwm#22

  • A runtime check of Apple's AWDL table is a sturdier guard than a macOS build allow-list. Our allow-list breaks whenever Apple ships a build and we then diff offline (as we did for 25G416). ForbiddenItlwm lists verified kernels but also compares the live IO80211AWDLProtocol table against expected anchor slots, and falls back to plain Wi-Fi if it differs. Worth considering for the native front end's own checks. Confidence: Medium. Link: https://github.com/retropineaple/ForbiddenItlwm/blob/main/docs/HOW-IT-WORKS.md

  • Its Bluetooth helper (BTLEAdvFixup) needed new bluetoothd byte patterns for 26.7.2 and 26.7.3, the same kind of breakage our LegionBtFix guards against. Issue #5 reports all four patch sites missing on 25G342; version 0.8.1 added patterns taken from Apple's installer, and an open PR (#31) adds one for 25G332's extendedConnection site and moves the LE data-length fix into the plugin. We already match by wildcarded bytes and checked 25G241 and 25G332 offline, and 25G416 is installed, so no action beyond keeping the offline check in the update routine. Intel Bluetooth only. Confidence: Medium. Links: retropineaple/ForbiddenItlwm#5 , retropineaple/ForbiddenItlwm#31

  • AWDL on a single-radio card needs the card on the phone's channel at the right moment, and another project says the same. A post on kgp-macPro/AirportItlwm-Tahoe (issue #7) reports AirDrop "hit or miss" on an Intel 9560: with the station and AWDL on different channels the firmware has to hop, and in discovery mode it exchanged sync frames but would not acknowledge unicast data. This matches owly's spec, the ForbiddenItlwm roadmap (Mac to iPhone slow when channels differ, issues #22 and #24) and the hardware-ACK requirement in our owly read. Confidence: Low (a single user's post). Link: kgp-macPro/AirportItlwm-Tahoe#7

  • owly released 0.4.3 on 2026-11-11 (AWDL peers in owly browse, more cards recorded). This is the commit we already read on 2026-11-12, so our notes stand. MIT. Link: https://github.com/t4t5/owly/releases/tag/v0.4.3