Commit Graph
6 Commits
Author SHA1 Message Date
Daniel Samson 767a2a9a7c Notify dispatch and GPE handlers (M21.2)
The AML interpreter now handles the Notify opcode (0x86, previously
unhandled): it resolves the target device, evaluates the code, and
records the pair in a bounded per-evaluate queue the caller drains with
takeNotifications. A host unit test with hand-encoded AML — a method that
issues Notify(DEV_, 0x80) — proves the device and code come back; aml.zig
joins the zig build test loop so the interpreter is covered on the host.

The acpi service's SCI handler now services general-purpose events too:
for each set-and-enabled GPE bit it evaluates the \_GPE._Lxx (level) or
_Exx (edge) handler method, drains the Notify queue that produced, and
publishes a domain event per notified device — PNP0C0A battery, ACPI0003
AC, PNP0C0D lid, else generic notify — then clears the status bit and
acks. The embedded controller's _Qxx queries are out of scope (hardware
track). QEMU raises no GPEs on this config, so the QEMU suite is the
regression net (the power button still works with GPE servicing in the
path); correctness is the unit test. Suite 59/59.
2026-07-13 05:44:58 +01:00
Daniel Samson 5725d35e5b Name the attach reply statuses instead of magic numbers
Add an AttachStatus enum to ps2-library.zig for AttachReply.status,
distinguishing the three failure causes handleAttach previously
collapsed into a bare -1: invalid_request (message too short),
missing_endpoint (no capability passed), and no_such_device (no port
identified the requested device type). The keyboard and mouse drivers
check against AttachStatus.ok rather than a literal 0.
2026-07-11 23:11:45 +01:00
Daniel Samson 8c95525793 Wire real PS/2 mouse packets through to input events
Replace the mouse driver's synthetic stream with the real path, the
way the keyboard was wired:

- The auxiliary port's IRQ12 is enumerated on the mouse's own ACPI
  node (PNP0F13), and the kernel only lets a device's claimer bind or
  ack its IRQs — so ps2-bus now claims that node alongside the
  controller whenever port 2 carries a device, binds IRQ12 to its one
  endpoint, and re-arms whichever line the notification's badge names.
  The forwarding loop already routed auxiliary bytes by status bit 5.
- mouse-packet.zig (new, pure, host-tested): three-byte stream-mode
  packet assembly — bit-3 sync with resynchronization, ACK/BAT bytes
  dropped at packet start, nine-bit two's-complement movement,
  overflow packets discarded, and PS/2 positive-Y-up converted to the
  screen convention (positive down).
- mouse.zig mirrors the keyboard driver: no hardware claim, attaches
  to the bus as its mouse, and publishes button_down/button_up per
  changed button plus motion events with the pressed-button mask.

Verified end to end in QEMU via monitor mouse_move/mouse_button:
motion round-trips in screen coordinates, buttons transition with the
right mask, and keyboard events keep flowing alongside. The follow-up
is the IntelliMouse magic-knock for a scroll wheel (four-byte packets)
and scroll events.
2026-07-11 23:09:04 +01:00
Daniel Samson aa0c97353a fixing arguments 2026-07-11 21:39:27 +01:00
Daniel Samson 1bf91115dd Generalize input module to mouse and joystick/gamepad events
Extend the input service beyond the keyboard so mouse and joystick/gamepad
drivers can broadcast too, with per-device publish and subscribe methods.

- protocol: KeyEvent joins MouseEvent (motion/buttons/scroll) and
  JoystickEvent (axes/buttons), all carried in a common InputEvent envelope
  tagged with a DeviceKind. A subscribe request carries a device_mask, so a
  subscriber names the classes it wants and the service routes each event only
  to interested subscribers (a mouse-only listener never wakes for keystrokes).
- runtime: per-device publish methods (publishKeyboardEvent/publishMouseEvent/
  publishJoystickEvent) and subscribe helpers (subscribeKeyboard/Mouse/Joystick,
  each typed, plus subscribe(mask)/subscribeAll returning the tagged envelope).
- service: subscriber table gains a device_mask; broadcast routes by the
  event's device class.
- mouse driver now publishes (synthetic) mouse events like the keyboard driver;
  input-source cycles all three classes; input-test subscribes to all and only
  emits its "ok" marker once it has received one of each class — so the passing
  test proves per-device routing, not just delivery. Real HID decoding stays a
  follow-up.

No kernel changes: ipc_send is generic and the 36-byte InputEvent fits its
64-byte payload. Full QEMU suite 48/48; serial log confirms keyboard, mouse,
and joystick all reach one subscription.
2026-07-11 15:21:09 +01:00
Daniel Samson 2a583d55a8 Finishing PS/2 bus driver 2026-07-11 14:12:33 +01:00