kernel: laptop touchpads over I2C-HID

A laptop's touchpad is on the chipset's I2C bus, found only through ACPI
(PNP0C50 / ACPI0C50), and SchOS had no I2C at all, so the touchpad of
every recent laptop did nothing.

kernel/acpi/i2c_hid.c drives the DesignWare I2C controller every x86
chipset uses (AMDI0010 on Ryzen) by polling, speaks HID over I2C to the
device the _DSM points at, and feeds the reports through the USB HID
parser. On AMD the interrupt pin's level is read every tick, so a report
is read exactly when one is waiting; elsewhere the input register is polled.

The touchpad is driven in mouse mode, which it starts in after a reset:
motion, clicks and taps become /dev/input/event1 like any mouse. The HID
parser now prefers a descriptor's relative pointer over an absolute one
for this, and takes descriptors from transports other than USB.

Untested on hardware: QEMU emulates no I2C touchpad.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01S1giNvrHTrPMPw7V6x1kQ4

kernel/core/main.c

@@ -12,6 +12,7 @@
 #include "core/klog.h"
 #include "dev/usb.h"
 #include "dev/acpi.h"
+#include "dev/hid.h"
 #include "arch/vmap.h"
 #include "core/hwinfo.h"
 
@@ -391,6 +392,9 @@ kmain(uint64 fb_addr)
   usb_hotplug_start();
   // Where AML runs for the SCI: the power button, GPEs, notifications.
   acpi_thread_start();
+  // Laptop touchpads, on the chipset's I2C bus; found through ACPI, so after
+  // the namespace is loaded, and polled from a thread of their own.
+  i2c_hid_start();
   // The debugger's own thread, for the same reason netinit's comes after
   // sinit: kthread_create needs a usable process table. From here on Ctrl-G
   // on the console answers, with no filesystem mounted and nothing in