OPINIONATED OMARCHY
TROUBLESHOOTING CORPUS
790
RECORDS
790
AUDITED
0
UNCHECKED
2388
SOURCES
LIVE
audio-input medium common DISPUTE

Game controller pairs or plugs in but no game sees any input

archendeavouroscachyosmanjaroomarchydesktoplaptopbluetooth
CORRECTED An auditor found the fix wrong or incomplete and rewrote it.
AUDIT NOTE

Most of this is faithfully and precisely sourced, but the last section does not work on the platform this record targets. Verified good: joyutils (extra 1.8.1, ships jstest, and the wiki names exactly this package), linuxconsole (extra, fftest/evdev-joystick), evtest (extra) and usbutils. xpadneo-dkms 0.10.4, xone-dkms 0.5.8, xone-dongle-firmware 2.0.0 and game-devices-udev all live in the AUR. The xone-dkms PKGBUILD really does install -Dm644 install/modprobe.conf "$pkgdir/usr/lib/modprobe.d/xone-blacklist.conf", so the blacklist path and the mandatory reboot are exactly right. The permissions section quotes the wiki's grep, the systemd 664/input-group default and the Steam Controller workaround verbatim (/usr/lib/udev/rules.d/70-steam-controller.rules → /etc/udev/rules.d/99-steam-controller-perms.rules, MODE 0660→0666). Three defects: (1) The final /etc/X11/xorg.conf.d/51-joystick.conf block is presented as the fix "for XWayland apps". That is wrong. XWayland takes its input from the Wayland compositor and does not read xorg.conf.d InputClass sections at all, so on Hyprland/Omarchy (every tag on this record) the file is inert. It also needs Driver "joystick" from xf86-input-joystick, which is NOT in the repos (AUR only, 1.6.4-5) and is unmentioned. (2) [GATT] ReconnectIntervals=1,1,2,3,5,8,13,21,34,55: the Arch Bluetooth page really does print it under [GATT], but bluez's own src/main.conf defines ReconnectIntervals (and ReconnectAttempts, AutoEnable, ResumeDelay) under [Policy]. Under [GATT] it is parsed as an unknown key and ignored. (3) sudo pacman -S steam-devices fails on a box without multilib: steam-devices is multilib/1.0.0.87, not extra.

The cause above was not rewritten and may still contain the error described. The fix below is the corrected version.
RISK

xpadneo-dkms and xone-dkms are DKMS modules: they must rebuild against every new kernel. If linux-headers (or linux-lts-headers, matching your installed kernel) is missing at upgrade time, the build fails silently during the pacman hook and the controller stops working after the next reboot. Keep the matching headers package installed, and check dkms status after a kernel upgrade.

Symptom

An Xbox or generic USB controller shows up in lsusb (or pairs successfully in bluetoothctl) but nothing responds to it. Common variants: ls /dev/input/js* returns No such file or directory, the pad works in Steam Big Picture but not in the game, the buttons are all mapped wrong and the triggers act like buttons, an Xbox Wireless controller over Bluetooth enters an endless connect/disconnect loop, and the Xbox Wireless Adapter dongle appears in lsusb but no controller ever binds to it.

Cause

Four separate causes wear the same symptom. (1) In-kernel xpad only supports Xbox controllers over USB. Over Bluetooth an Xbox One S/Series pad is handled by hid-generic, which gives wrong mappings, no rumble and no battery, and on stock firmware often loops connect/disconnect. (2) The Microsoft Xbox Wireless Adapter dongle has no in-kernel driver at all. It needs the out-of-tree xone driver plus its firmware blob. (3) Device permissions: udev must grant your user access to the /dev/input/event* and hidraw nodes, and if no installed package ships a rule for your pad, it is simply not readable. (4) The pad is fine but the game is using a different API than the one you tested (legacy joydev /dev/input/js* vs evdev vs SDL).

Fix

Three changes.

**1. Replace the "cursor moves when you push the stick" section.** The xorg.conf.d snippet only works under a real X server. XWayland ignores xorg.conf.d entirely, so it does nothing on Hyprland. On Wayland the fix is:


# Steam: Settings > Controller > turn OFF "Enable Steam Input for desktop configuration"
# or make the pad's evdev node invisible as a pointer, per-device:
sudo tee /etc/udev/rules.d/60-joystick-not-a-pointer.rules >/dev/null <<'EOF'
SUBSYSTEM=="input", ATTRS{name}=="*Controller*", ENV{ID_INPUT_MOUSE}="", ENV{ID_INPUT_POINTINGSTICK}=""
EOF
sudo udevadm control --reload && sudo udevadm trigger

Reconnect the pad. Only keep the xorg.conf.d block if you are actually running Xorg, and note it needs the AUR package xf86-input-joystick (it is not in the repos) for Driver "joystick" to resolve.

**2. Fix the main.conf section headers.** ReconnectIntervals is a [Policy] key in bluez, not [GATT] (the Arch wiki misfiles it):


# /etc/bluetooth/main.conf
[General]
JustWorksRepairing = always
FastConnectable = true
Class = 0x000100

[Policy]
AutoEnable = true
ReconnectAttempts = 7
ReconnectIntervals = 1,1,2,3,5,8,13,21,34,55

sudo systemctl restart bluetooth.service

**3. steam-devices is in multilib.** Enable the [multilib] repo in /etc/pacman.conf first, or use the broad AUR alternative:


sudo pacman -Syu steam-devices      # requires [multilib]
yay -S game-devices-udev            # no multilib needed

Also worth adding for the older Xbox 360 pad over Bluetooth, which this record omits: add the kernel parameter bluetooth.disable_ertm=1. And per the wiki, the most reliable cure for an Xbox One S pairing loop is to pair it once in Windows using the *same* Bluetooth adapter before pairing under Linux.

Verify

As your normal user (not root), evtest shows button and axis events. jstest /dev/input/js0 prints changing axis values. Open https://gamepad-tester.com/ in Chromium and confirm the pad appears with working buttons. For xpadneo, cat /sys/class/power_supply/*/capacity reports the controller's battery level, which proves the right driver bound rather than hid-generic.

Sources

gamepad-not-detected-or-no-input