Restore touchpad two-finger scroll and gestures that die after resume
Issue #12409 (open) supports the symptom, the libinput lines, the DPMS-only trigger, the same-second rebind failing, the rebind about 4 s later working, and the reporter's note that omarchy restart trackpad also reseats the I2C keyboard i2c-ITE8350:00. /usr/share/omarchy/bin/omarchy-restart-trackpad on 4.0.4-1 is byte-identical to quattro: requires-sudo=true, loops over every /sys/bus/i2c/drivers/i2c_hid_acpi/i2c-*, reloads intel_quicki2c if loaded. /usr/lib/systemd/system-sleep/ holds only keyboard-backlight, nvidia and unmount-fuse, so nothing runs it on resume, as the cause says. The defect is in the fix order: it tells the reader to run omarchy restart trackpad first and only afterwards says to check for an I2C keyboard, so a reader pasting top to bottom reseats the keyboard before reading the warning. A third commenter on the issue (Framework 13) shows the shipped script also walks other I2C devices such as FRMW0004/FRMW0005. The fix is reordered so the listing comes first. Not exercised: no driver was unbound here.
omarchy restart trackpad unbinds every i2c_hid_acpi device, not only the touchpad. On laptops with an I2C keyboard (the reporter's has i2c-ITE8350:00) the internal keyboard drops out for about a second, and if a rebind fails it stays gone until reboot. Keep a USB keyboard to hand, or use the single-device form.
Symptom
After resuming from suspend, or sometimes just after the screen blanks and comes back, my touchpad still moves the pointer and tap-to-click works, but two-finger scrolling and three-finger gestures stop working until I reboot. There are no i2c_hid errors in dmesg. Hyprland's log shows Touchpad: kernel bug: Touch jump detected and discarded and a stream of palm: keyboard timeout. Seen on a Lenovo Yoga Slim 7 Pro (Synaptics 06CB:CE44 on AMD AMDI0010).
Cause
Some I2C-HID touchpads come back from resume, or from a display power cycle, stuck in a state where they send only single-touch or relative reports and no multitouch reports, while the kernel driver stays bound and reports no error. Re-probing the device through an i2c_hid_acpi unbind and bind restores multitouch, but only if it runs once resume has settled. Running it in the same second as PM: suspend exit did not help in the report. Omarchy 4 ships omarchy-restart-trackpad, which does the reseat, but nothing runs it automatically.
Fix
**1. List the I2C-HID devices first.** No root needed:
ls /sys/bus/i2c/drivers/i2c_hid_acpi/
# e.g. i2c-MSFT0001:00 i2c-ITE8350:00
If only the touchpad is listed, step 2 is safe. If anything else is listed (an internal keyboard such as i2c-ITE8350:00, or other vendor devices), use step 3 instead, because the shipped script reseats every device in that directory.
**2. Omarchy 4, only the touchpad on the bus.** Run the shipped reset from a terminal after resume. It unbinds and rebinds every i2c_hid_acpi device and reloads intel_quicki2c if that module is loaded:
omarchy restart trackpad
Do not prefix it with sudo. The script calls sudo itself.
**3. Plain Arch, or Omarchy 4 with other devices on the bus.** Reseat only the touchpad by hand, using the name from step 1:
echo -n 'i2c-MSFT0001:00' | sudo tee /sys/bus/i2c/drivers/i2c_hid_acpi/unbind
sleep 1
echo -n 'i2c-MSFT0001:00' | sudo tee /sys/bus/i2c/drivers/i2c_hid_acpi/bind
If you automate it with a resume hook, wait until the system has settled (for example until systemctl is-system-running reports running, which took about 4 seconds in the report) rather than running it at the instant of resume. A reseat run in the same second as PM: suspend exit re-probed the device but left multitouch dead.
Verify
Two-finger scroll works again straight after the reset. hyprctl devices still lists the touchpad, and Hyprland's log shows Touchpad: device removed followed by New device ... Touchpad.