### System
- Hardware: Lenovo 21DM (ThinkPad/Yoga convertible)
- OS: Omarchy (Arch-based)
- Kernel: 7.1.9-arch1-2

### Symptom
On some cold boots the trackpad is completely unresponsive. `hyprctl devices` shows
an empty `mice:` section — no touchpad object is created at all, even though the
pen digitizer and touchscreen (a separate Wacom chip, `WACF2200`) work fine.

### Root cause
Journal from the affected boot:

```
i2c_hid_acpi i2c-MSFT0001:01: can't add hid device: -110
i2c_hid_acpi i2c-MSFT0001:01: probe with driver i2c_hid_acpi failed with error -110
i2c_designware i2c_designware.1: controller timed out
i2c_hid_acpi i2c-0416C300:00: can't add hid device: -110
i2c_hid_acpi i2c-0416C300:00: probe with driver i2c_hid_acpi failed with error -110
```

The Precision Touchpad (`MSFT0001:01`) and an Elan sensor chip (`0416C300:00`,
binds to hid-sensor-hub) sit behind two separate Intel LPSS I2C controllers
(`i2c_designware.0` at PCI `0000:00:15.0`, and `i2c_designware.1` at PCI
`0000:00:15.2`). During early boot, one or both controllers time out (-110 /
ETIMEDOUT) before the HID devices respond, so `i2c_hid_acpi` never binds and no
input device is ever created for the session. This is a boot-time race, not a
missing driver or missing hardware — the chips work fine once reprobed.

### Manual fix (confirmed working)
Forcing Linux to reinitialize each PCI I2C controller re-triggers a clean probe
of its children and the touchpad comes up correctly:

```
echo 1 | sudo tee /sys/bus/pci/devices/0000:00:15.2/remove
echo 1 | sudo tee /sys/bus/pci/rescan
echo 1 | sudo tee /sys/bus/pci/devices/0000:00:15.0/remove
echo 1 | sudo tee /sys/bus/pci/rescan
```

After this, `hyprctl devices` correctly lists `msft0001:01-06cb:ce2d-mouse` under
`mice:` and the trackpad works normally.

### Suggested persistent workaround
Since this is timing-dependent and not reliably fixed by controller-side driver
rebinding alone (rebinding just the HID driver to the wedged bus still times out —
the bus itself needs a reset, not just the HID device), a systemd oneshot unit
that reruns this reset a few seconds after boot reliably works around it:

```ini
[Unit]
Description=Reset I2C controllers for touchpad/sensor HID devices that fail to probe at boot
After=multi-user.target

[Service]
Type=oneshot
RemainAfterExit=yes
ExecStart=/usr/bin/sh -c 'echo 1 > /sys/bus/pci/devices/0000:00:15.2/remove'
ExecStart=/usr/bin/sh -c 'echo 1 > /sys/bus/pci/rescan'
ExecStart=/usr/bin/sh -c 'sleep 1'
ExecStart=/usr/bin/sh -c 'echo 1 > /sys/bus/pci/devices/0000:00:15.0/remove'
ExecStart=/usr/bin/sh -c 'echo 1 > /sys/bus/pci/rescan'

[Install]
WantedBy=multi-user.target
```

This is obviously hardware/PCI-address-specific (my two controllers happen to be
at 0000:00:15.0 and 0000:00:15.2) and masks rather than fixes the underlying
kernel/ACPI race, but it might be useful as a documented workaround for others
hitting the same class of failure on Omarchy, or as a pointer if Omarchy wants to
ship a generic version of this (e.g. detect any i2c_hid_acpi device that failed
with -110 and auto-reset its parent PCI controller).

Happy to provide more logs/output if useful.

---
Also — unrelated to the bug report, but: thank you for the talks on the Lex
Fridman podcast, they've been genuinely inspirational.