search

atar-axis / xpadneo

Advanced Linux Driver for Xbox One Wireless Controller (shipped with Xbox One S)
https://atar-axis.github.io/xpadneo/
GNU General Public License v3.0
1.92k stars 111 forks source link

Xbox Series x Button mapping wrong over bluetooth #326

Closed Kivarnis closed 2 years ago

Kivarnis commented 2 years ago

Version of xpadneo

Controller Model

[*] Xbox One S controller

Connection mode

Installed Software

Severity / Impact

Describe the Bug

The mappings are incorrect, L3/R3 do not function as intended, Share and the new series x middle button are flipped. This works 100% correctly in xow using the dongle, however, as probably aware, this connect/disconnect loops over the dongle and is extremely frustrating, trying bluetooth instead via xpadneo has given these results, which is also unusable.

Steps to Reproduce

Pair to bluetooth, experience issue.

Expected Behavior

L3/R3 and left and middle buttons to be mapped and working correctly.

Screenshots / GIFs / Videos

System Information

# uname -a

# Linux Trinity 5.15.2-zen1-1-zen #1 ZEN SMP PREEMPT Fri, 12 Nov 2021 19:22:07 +0000 x86_64 GNU/Linux

# xxd -c20 -g1 /sys/module/hid_xpadneo/drivers/hid:xpadneo/0005:045E:*/report_descriptor | tee >(cksum)

# fish: No matches for wildcard “/sys/module/hid_xpadneo/drivers/hid:xpadneo/0005:045E:*/report_descriptor”. See `help expand`.
xxd -c20 -g1 /sys/module/hid_xpadneo/drivers/hid:xpadneo/0005:045E:*/report_descriptor | tee >(cksum)
             ^

Controller and Bluetooth Information

Didn't fully understand this, still new to Linux.

Additional Context

Kivarnis commented 2 years ago

Sorry for formatting I also know very little about that too, lol.

kakra commented 2 years ago

The mappings are incorrect, L3/R3 do not function as intended,

By that you mean the thumb stick buttons?

Share and the new series x middle button are flipped

What's the "new series x middle button"? You mean the Xbox button? Those should not be flipped at all but that's probably caused outside of xpadneo.

This works 100% correctly in xow using the dongle, however, as probably aware, this connect/disconnect loops over the dongle and is extremely frustrating, trying bluetooth instead via xpadneo has given these results, which is also unusable.

It's possible that SDL still has read access to the hidraw device, this happens on some distributions and I'm not sure why. Could you try the following to sort things out at the easiest level first:

  1. Stop Steam, connect the controller
  2. Look at dmesg to find the hidraw number for the controller
  3. Run chmod a-rwx /dev/hidrawNUMBER
  4. Start Steam, check your inputs
  5. Still wrong? Disable Steam Input support for Xbox controllers, try again

Thanks.

Kivarnis commented 2 years ago

I do yes.

https://media.gamestop.com/i/gamestop/11108954/Microsoft-Xbox-Series-X-Wireless-Controller-Carbon-Black?$pdp$$&fmt=webp the lower middle button, it's swapped function with the upper left button and that one does absolutely nothing. This all works in xow with the wireless dongle.

I have will try this, I will note however I never had the support enabled to begin with. As a Dual Shock 4 user from Windows, disabling that was habit for use with DS4Windows which was a vastly better service. It was the first thing I did upon booting Steam, before ever using the controller with this install and i do not run my current game through Steam, either.

kakra commented 2 years ago

The button function swap you describe still looks like the current double mapping fix that SDL applies to our hidraw device (although it doesn't need to do that, our hidraw device sends correct mappings already). Thus, disabling access to the hidraw device will force SDL to fall back to normal evdev operation, and it applies no mapping fixups then.

You could run gamepad-tool: If your controller shows up as an ID starting with 03 it would indicate that it runs as an emulated USB controller via hidraw - which causes problems here. If you disable hidraw user permissions, the controller will show up with an ID starting with 05 (which means Bluetooth), so then it would read the controller correctly.

BTW: Don't try to fix your mappings with the gamepad-tool remap function - it won't work because some buttons are simply dead due to double remap. Actually, it's important to not have any manual remapping file at all before trying to fix and verify this.

Kivarnis commented 2 years ago

I'm sorry but this is a huge amount of text and visually parsing I simply found nothing that said anything about hidraw anything. Plenty of entries regarding the controller, but not that, as such no ideas what to put. However I do believe you are correct.

kakra commented 2 years ago

https://github.com/atar-axis/xpadneo/issues/326#issuecomment-972642699 list item 3

Kivarnis commented 2 years ago

I will repeat what I told you, there is thousands of lines from this output. I would need a week of doing nothing else. So I do not have the hidraw'numberhere' if I understood that correctly, I do not have the number to finish that command. Please advise.

Kivarnis commented 2 years ago

I copied literally everything to a text editor, and searched "hidraw" and found the ID, testing now.

Kivarnis commented 2 years ago

This worked, now the question is, this was all done automatically, and without steam ever having support for xbox enabled at any time. Prehaps this interaction should be looked into? As it stands, this seems to be the out of box experience for anyone, and it wasn't like this did nothing. Before installing xpadneo my controller would move the camera indefinitely, installing this, fixed that, but had incorrect mapping of some buttons. So it partially worked, apparently. It is unreasonable to assume most whom need a gamepad will not find themselves in precisely the same situation.

kakra commented 2 years ago

Well, usually, if you just connected the gamepad, the important information will be in the end of the dmesg output (last page usually). So the thousands of lines before do not matter. But sorry, I didn't understand that you were actually referring to the dmesg output when you wrote "huge amount of text", the context wasn't obvious to me.

dmesg output is chronologically, so you'd look from button up to find something about the Xbox controller, then look out for hidraw in a nearby context, no need to search the whole file.

You can use dmesg | less to get the output in a page, then press G (upper case) to get to the end of the file, press ?hidraw to search backward and find something nearby lines about Xbox controller detection.

This worked, now the question is, this was all done automatically

We are already shipping with configuration that would revoke permissions from the hidraw device. But for unknown reasons it doesn't work on Arch Linux (and probably others).

Also, this is not a one-time fix: If you re-connect the controller, the hidraw device will be recreated with wrong permissions for you, and the number may even change.

So this was only meant as steps identifying the problem: Your's is actually the hidraw permissions problems we've been seeing lately. There are some udev files which change uaccess session access to devices on Arch Linux which I even don't have on my system, and which are also not triggered by udev. We currently do not know what's going on, we only know how to identify the problem, and how to solve it temporarily.

We've found a work-around in the Discord channel which can acts as a temporary fix that survives reboots and reconnects.

Kivarnis commented 2 years ago

Well happy to test, as is xow is unsuable for wireless dongle and the series x, so this is my only sustainable option as it sits. But please do understand while I am technically proficient, I am not as savy as some Linux users to this sort of thing.

kakra commented 2 years ago

We have a very nice Discord community which is happy to help with technical and low-level stuff. I'd like to invite you: https://discord.gg/nCqfKa84KA - It has quite a few people across multiple projects (xpadneo, xow, xone) which usually have a good understanding of the current bugs and issues, how to work around them, or even properly identifying what the real underlying problem is.

This issue tracker is more a technical channel, and support usually blows discussions up until I cannot quite well follow them for development.

For now, I'm closing this as a duplicate of #303 so we can keep issue reports tight and clear for development. Your issue is currently already under active investigation, following/subscribing #303 will keep you informed about updates.