Add Logitech HID++ (HIDPP 2.0) support via hidapi

[PradyumnaKrishna]

Fixes #1277 — the original issue I opened about a year ago.

Description

After a year, I’m back — and I finally got my Logitech Signature M650 to expose button 4 and 5 correctly. The HID++ code here was built with heavy AI help (I’m not a C/C++ expert), but it works and I wanted to upstream it. Feel free to ask for changes you want to make, or if you like you can take it from here.

This change introduces first‑class Logitech HID++ support in the helper so devices that expose buttons only through HID++ can be handled without the vendor driver. This PR doesn't interfere with any code other than HIDPP for logitech devices.

Design Decision

Logitech HID++ exposes multiple button‑capture features. We chose REPROG_CONTROLS_V4 because it is available on our working test device (M650) and Solaar already supports/uses it. That suggests it is a practical, widely‑enough feature to target first, and it likely explains why certain Logitech buttons were previously not visible to MMF. We therefore divert controls via REPROG_CONTROLS_V4 and restore diversion on shutdown.

What’s implemented

• Adds a dedicated HIDPPListener and wires it into DeviceManager for attach/remove.
• Enumerates Logitech HID interfaces through hidapi.
• Discovers HID++ 2.0 features via FEATURE_SET and only enables REPROG_CONTROLS_V4 when advertised.
• When diversion is active, Back/Forward CIDs are translated into CG button down/up events and we synthesize drag events while a diverted button is held.
• Best‑effort: if HID++ probing fails we fall back to normal IOHID input.

Licensing / attribution

hidapi is pulled in as a submodule and remains under its upstream licensing. hidapi is tri‑licensed (BSD‑style/original/GPL) per its LICENSE*.txt files; this PR does not alter those terms and only links against it from our helper.

Thanks to @pwr-Solaar/Solaar — their codebase and docs were a primary reference for HID++ protocol handling and discovery patterns.

Impact

Logitech mice that require HID++ now work without vendor software, while existing non‑Logitech behavior remains unchanged.

PS (testing with logitech devices):

• Signature M650 — supports REPROG_CONTROLS_V4 → works (buttons 4/5 captured via HID++).
• G304 Wireless — does not support REPROG_CONTROLS_V4, but buttons 4/5 already work → app falls back cleanly and behaves as before.

[PradyumnaKrishna]

@noah-nuebling Please take a look when you have a moment.

[pkrh39]

Can confirm this works on M650L sequoia 15.7.3, but the side buttons stop working after waking device up from sleep. I wanted to post logs here though I have no idea how to get logs after a sleep cycle..

[PradyumnaKrishna]

Yes @pkrh39, I’m seeing this too, but I don’t think it’s related to the sleep events. I’m not sure how to reproduce it yet, but I’ll dig into it.

[miguelAngelo1999]

Hey @PradyumnaKrishna — great work tackling this! I hit the same problem and went down a similar path. I ended up with a different implementation that might be worth looking at before this merges: #1904

The main differences in my approach (`LogitechCIDActivator`):

No external dependency — uses `IOHIDLib` directly instead of hidapi, so no submodule or C++ required. Pure Obj-C.

All divertable CIDs, not just Back/Forward — the code enumerates `GetCidInfo` and diverts every button with the divert flag set, so thumb buttons, tilt wheel, precision mode, and side buttons all work without hardcoding CID values.

Unifying/Bolt receiver support — probes slots 1–6 on the receiver so wireless mice via USB dongle work, not just Bluetooth.

Sleep/wake handled — subscribes to `NSWorkspaceDidWakeNotification` + `NSWorkspaceScreensDidWakeNotification` and re-diverts on wake (HID++ diversion is per-connection and resets on sleep). Also has a 25s keep-alive timer for firmware resets during normal use. One gotcha I hit: `IORegisterForSystemPower` with an invalid `io_connect_t` corrupts the run loop on startup — avoid it entirely.

Not saying this to block your PR — just thought the overlap was worth flagging so Noah can pick the best approach or combine the two.

[PradyumnaKrishna]

Saw your work around @miguelAngelo1999. If your work solves the major problem around logitech devices and does work with logitech devices like M650 and other I don't have any problem. I was doing my best to understand the problem and fix it.