Switch 2 Controller Support #13327

flibitijibibo · 2025-07-05T18:55:43Z

Device init and polling works, but I haven't translated the packets yet since ~~the public code doesn't line up with the automatic mappings~~ and I'm intentionally avoiding doing any further reverse engineering myself. Happy to add patches to get this PR working though!

udev rule for Linux:

SUBSYSTEM=="usb", ENV{DEVTYPE}=="usb_device", ATTRS{idVendor}=="057e", ATTRS{idProduct}=="2073", MODE="0666"

CC @Nohzockt: If you have a preferred name/email for Co-authored-by I'll apply it to the patch that implements the init/polling work.

Nohzockt · 2025-07-05T21:05:41Z

Fixes #13178.

Device init and polling works, but I haven't translated the packets yet since the public code doesn't line up with the automatic mappings and I'm intentionally avoiding doing any further reverse engineering myself. Happy to add patches to get this PR working though!

udev rule for Linux:
SUBSYSTEM=="usb", ENV{DEVTYPE}=="usb_device", ATTRS{idVendor}=="057e", ATTRS{idProduct}=="2073", MODE="0666"
CC @Nohzockt: If you have a preferred name/email for Co-authored-by I'll apply it to the patch that implements the init/polling work.

Well, my preferred name would be just Nohzockt or Noah, no email or something special. I feel honoured that you ask me. Thanks a lot.

flibitijibibo · 2025-07-05T21:33:31Z

Done!

I also cleaned up the button reading and made a working gamepad mapping, once the left/right stick are working this is ready to review.

nilsojc · 2025-07-06T11:26:22Z

Quick question. I was able to generate an SDL3.dll file with your WIP stub, would an SDL2.dll be able to be generated as well or this still a work in progress? I am wondering how these can be built for executables like emulators or steam API.

src/hidapi/libusb/hid.c

endrift · 2025-07-08T01:31:29Z

I have some devices I can test in addition to the GC controller so I might want to add those to this PR if possible. I can create a branch off of your branch, or add them after this is merged.

flibitijibibo · 2025-07-08T01:46:11Z

Whatever works! I can add commits to this PR if Sam doesn't get to it first.

src/hidapi/libusb/hid.c

src/joystick/SDL_joystick.c

src/joystick/hidapi/SDL_hidapi_switch2.c

slouken · 2025-07-11T19:06:17Z

src/joystick/hidapi/SDL_hidapi_switch2.c

+    int size;
+    while ((size = SDL_hid_read_timeout(device->dev, packet, sizeof(packet), 0)) > 0) {
+        if (size > 14) {
+            Uint64 timestamp = SDL_GetTicksNS();


This block should be refactored out into a separate function to handle devices with varying capabilities and report formats.

I believe all NS2 devices report the same layout (passing 0 for unsupported axes like the GCN triggers) but @endrift can confirm this.

slouken · 2025-07-12T00:11:31Z

Does this support Bluetooth? What changes would need to be made for rumble and gyro support?

endrift · 2025-07-12T00:14:06Z

I haven't looked over the code yet, but I tested the controller and discovered a few things that might want improvements:

The ranging of everything analog isn't saturating. Wholly unpressed triggers have non-zero values, and wholly pressed triggers as well as wholly tilted sticks don't reach anywhere close to the full values.
The share and C buttons register, but aren't mapped to any gamepad buttons. The Xbox One controller maps the share button to Misc1, so we might want to do the same thing for the share button here. Though we might want to have a named share button in SDL too.

flibitijibibo · 2025-07-12T00:14:50Z

Does this support Bluetooth? What changes would need to be made for rumble and gyro support?

Sadly no, it seems to communicate entirely via a proprietary BLE service so we would need to communicate directly via dbus+bluez or something. Rumble/Gyro come down to figuring out any more possible bulk output init commands or hid output commands to get more out of the endpoint than what it shares with us today.

endrift · 2025-07-12T00:16:08Z

Rumble is partially documented here: https://gist.github.com/shinyquagsire23/66f006b46c56216acbaac6c1e2279b64#file-00_ns2_sidechannel_notes-txt-L285

slouken · 2025-07-12T00:16:52Z

The ranging of everything analog isn't saturating. Wholly unpressed triggers have non-zero values, and wholly pressed triggers as well as wholly tilted sticks don't reach anywhere close to the full values.

We probably need to get some sort of factory calibration from the controller, like is done with the Switch 1 controllers.

The share and C buttons register, but aren't mapped to any gamepad buttons. The Xbox One controller maps the share button to Misc1, so we might want to do the same thing for the share button here. Though we might want to have a named share button in SDL too.

Yes, share should be mapped to MISC1 and the C button should be mapped to MISC2.

flibitijibibo · 2025-07-12T00:17:07Z

I haven't looked over the code yet, but I tested the controller and discovered a few things that might want improvements:

The ranging of everything analog isn't saturating. Wholly unpressed triggers have non-zero values, and wholly pressed triggers as well as wholly tilted sticks don't reach anywhere close to the full values.

The share and C buttons register, but aren't mapped to any gamepad buttons. The Xbox One controller maps the share button to Misc1, so we might want to do the same thing for the share button here. Though we might want to have a named share button in SDL too.

Huh, I thought I put capture at misc1 and C at misc2, but testcontroller doesn't show these so I couldn't confirm visually.

I can confirm the stick range isn't saturated but I wasn't sure if there was a way to get a calibration range just yet. The triggers are showing full range though, but maybe I'm not looking closely enough.

endrift · 2025-07-12T00:20:48Z

Testing adding the USB ID for the Pro controller (0x2069) to this, it doesn't actually show up as functional. I haven't dug further yet.

Also, Misc1 definitely shows up in testcontroller for the Xbox One controller.

slouken · 2025-07-12T00:24:33Z

Misc2 doesn't show up in the gamepad image, but it will show up in the left-hand input list if it's mapped.

flibitijibibo · 2025-07-12T00:26:37Z

Odd... does anything seem off with the mapping added to gamepad.c?

src/joystick/SDL_gamepad.c

flibitijibibo · 2025-07-12T01:26:20Z

Latest push is confirmed working with misc1/misc2.

endrift

Some minor things I noticed, but until calibration is figured out I don't think this will be ready.

include/SDL3/SDL_hints.h

src/hidapi/SDL_hidapi.c

endrift · 2025-07-16T22:23:26Z

src/hidapi/libusb/hid.c

+	const unsigned char DEFAULT_REPORT_DATA[] = {
+		0x03, 0x91, 0x00, 0x0d, 0x00, 0x08,
+		0x00, 0x00, 0x01, 0x00, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF
+	};
+	const unsigned char SET_LED_DATA[] = {
+		0x09, 0x91, 0x00, 0x07, 0x00, 0x08,
+		0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
+	};


Might want to use some enums for things like the first byte (the command) and the second byte (the direction), but that's not needed for the initial merge.

src/joystick/hidapi/SDL_hidapi_nintendo.h

src/joystick/hidapi/SDL_hidapi_switch2.c

Thanks to kiddkaffeine for the Xcode updates!

Thanks to Nohzockt for the initial libusb init and hidapi polling work!

flibitijibibo · 2025-07-16T22:44:00Z

Latest push addresses most of the review items, left the question for Sam and didn't touch the init commands just yet; presumably we need at least one more to get it to show us calibration data so we can probably make the change at that point.

endrift · 2025-07-16T22:45:26Z

Are you on ndeadly's discord where a bunch of discussion about reversing has been going on? If not, it's probably a good idea to join.

flibitijibibo · 2025-07-16T22:47:47Z

Are you on ndeadly's discord where a bunch of discussion about reversing has been going on? If not, it's probably a good idea to join.

Went ahead and joined, probably won't be able to add very much under the circumstances but I'll keep up as best as I can!

endrift · 2025-07-21T04:10:57Z

I've made a PR on your branch to get the procon working too. However, confining the bulk endpoint comms to hid.c won't be viable for this driver. Things like rumble go over that endpoint, so we need a way to plumb it through to the actual driver.

eloj · 2025-07-21T07:24:41Z

src/joystick/hidapi/SDL_hidapi_switch2.c

+                timestamp,
+                joystick,
+                (Uint8) i,
+                (packet[buttons[i].byte] & buttons[i].mask) != 0


After the latest change to the buttons array, it seems fairly redundant from a purely functional standpoint.

Suggested change

(packet[buttons[i].byte] & buttons[i].mask) != 0

(packet[3 + (i >> 3)] & (1U << (i & 7))) != 0

flibitijibibo · 2025-07-21T12:37:32Z

Merged the ProCon2 support and renamed this accordingly.

flibitijibibo mentioned this pull request Jul 5, 2025

Add support for NS2 GameCube controller #13178

Open

flibitijibibo marked this pull request as draft July 5, 2025 18:58

flibitijibibo force-pushed the s2stub branch from 620f3ba to 20f3114 Compare July 5, 2025 19:10

flibitijibibo force-pushed the s2stub branch from 20f3114 to 864b463 Compare July 5, 2025 21:32

flibitijibibo force-pushed the s2stub branch from 864b463 to 27346e0 Compare July 5, 2025 23:37

flibitijibibo force-pushed the s2stub branch from 27346e0 to ac016b6 Compare July 6, 2025 15:59

flibitijibibo changed the title ~~[DONOTMERGE] Switch 2 GameCube Controller Support~~ Switch 2 GameCube Controller Support Jul 6, 2025

flibitijibibo marked this pull request as ready for review July 6, 2025 16:00

flibitijibibo requested a review from slouken July 6, 2025 16:00

flibitijibibo force-pushed the s2stub branch from ac016b6 to 008be2a Compare July 6, 2025 16:06

endrift reviewed Jul 6, 2025

View reviewed changes

src/hidapi/libusb/hid.c Outdated Show resolved Hide resolved

flibitijibibo force-pushed the s2stub branch from 008be2a to e7dc078 Compare July 6, 2025 23:57

flibitijibibo mentioned this pull request Jul 9, 2025

Implement SInput Device Support #13343

Merged