[Haseenesss]

Fix or feature

• Fix
• Feature

What and why

Fixes the Mac part of #971: one Pro Controller was read as two players, and its gyro did not reach the game.

Symptom. The log in #971 (Breath of the Wild, v1.3.0, one Pro Controller over USB) shows the same pad taken twice at the same USB location, then the controller applet over and over (360 times in 30 s):

[hid] 'Pro Controller' (057e:2009) at USB location 0x69b29174: GameController lists 1 of 2 pads of this model: read here, from its report 0x30
controller 'Pro Controller': player 1
[hid] 'Pro Controller' (057e:2009) at USB location 0x69b29174: GameController lists 1 of 2 pads of this model: read here, from its report 0x30
controller 'Pro Controller': player 2
hid controller applet: 0 to 4 players wanted, 2 connected

Cause. The identical-pads reader (pad_hid_mac.h, new in the v1.3.0 pin) matches the Pro Controller's vendor and product besides the gamepad usages. IOHIDManager reports a device once per matching dictionary it meets, each time under a new IOHIDDeviceRef, and hidpad_found compared references only. So one pad counted as two of its model against GameController's one, both references were read as twins (players 1 and 2), and GameController's controller, the one with motion, was left unread. v1.2.3 did not have this reader.

Change. Upstream fixed the same bug on 2026-11-08 in seedds/nxnative a00f03b (from seedds/nxnative#1722, found on Ocarina of Time). This PR first carried it as a backport. Since main moved the pin to c78d7375, which contains a00f03b, the backport is gone and the fix itself comes with the pin. What remains:

• patches/0099-test-one-pro-controller-is-one-pad.patch: the regression test, in t_controllers_mac.m (which comes with patch 0081, Add Steam Controller (2026) support on macOS seedds/nxnative#221, so fork-only until then).

The iOS part of #971 is a separate cause (a controller with no gyroscope that iOS lists with one, which kept the phone's gyro out) and is in #975.

Checked

• Tests: python3 tools/engine.py check (65 patches apply cleanly to c78d7375). t_controllers_mac, t_motion_pro_mac, t_pad_hid, t_hid, t_pad_pro pass on an engine rebuilt from the series.
• Test for a fix: test_hid_same_device_twice in t_controllers_mac.m reports a generic pad and a Pro Controller twice each, under two references to one registry entry: one pad and one player for the first, no twins for the second (GameController keeps reading it). It fails without a00f03b and passes with it.
• Cause confirmed on this Mac (macOS 27.2): a small IOHIDManager probe with two matching dictionaries that both meet one device gets two callbacks with two different IOHIDDeviceRefs for the same registry entry, on a plain and on an independent manager.
• Run on a device or Mac: not run with a Pro Controller, there is none here. Not checked on iOS.

Screenshots

None, no visible change.

Scope

• No engine file (dm/, scripts/, docs/, ref/) is added to the repository: a change there is a patch in patches/ (a commit in engine/, then tools/engine.py export, and tools/engine.py check passes), with its Upstream-Status
• A fix to the engine or a game profile also goes to seedds/nxnative as a PR in parallel: this PR's patch says forwarded seedds/nxnative#..., or pending until that PR is open (upstream already has the fix, a00f03b, in the pin; the test is fork-only with patch 0081)
• A change to a launcher view has before and after screenshots above
• No keys, game files or data extracted from a game in the change or the attachments

🤖 Generated with Claude Code