SCRYMAMEReferenceWritereset

MAMEWrite

reset

Soft-reset the machine without restarting MAME, and stop where its boot begins.

reset()

Try it

On Gridlee, a free ROM from mamedev.org, at any stop. The example session resets at frame 630, with a game under way.
Ask

Reset the game without restarting MAME.

Call
reset()
Reply 6 of 16 fields
{
  "summary": "reset: softreset applied at frame 630, pc 0x0000f000 verified True",
  "kind": "softreset",
  "pc": "0x0000f000",
  "boot_pc": "0x0000f000",
  "verified": true,
  "frame_at_reset": 630
}

The machine is back at 0xf000, the address start first stopped it at, so verified is true. The frame counter did not restart: frame_at_reset is 630.

The whole reply
JSON
{
  "summary": "reset: softreset applied at frame 630, pc 0x0000f000 verified True",
  "status": "ok",
  "kind": "softreset",
  "pc": "0x0000f000",
  "boot_pc": "0x0000f000",
  "verified": true,
  "frame_at_reset": 630,
  "taps_across": [
    {"ref": "film", "cpu": ":maincpu"}
  ],
  "cause": "softreset",
  "stop_seq": 61,
  "cpu": ":maincpu",
  "frame": 630,
  "regs": {
    "A": "0x42",
    "B": "0xd8",
    "D": "0x42d8",
    "X": "0x0014",
    "Y": "0xf2e6",
    "U": "0x4071",
    "PC": "0xf000",
    "S": "0x07f4",
    "CC": "0xd0",
    "DP": "0x00"
  },
  "instruction": {"address": "F000", "bytes": "1A50", "text": "ORCC   #$50", "symbol": "take_credit", "delta": 6538},
  "following": ["F002: 4F          CLRA", "F003: 1F 8B       TFR    A,DP", "F005: 10 CE 08 00 LDS    #$0800", "F009: BD D1 21    JSR    $D121", "F00C: B6 95 02    LDA    $9502", "F00F: 97 BC       STA    <$BC", "... 6 more"],
  "errorlog": "<volatile>"
}

Arguments

Takes no argument.

What comes back

kind
Always softreset: the only reset this tool makes.
pc
Where the machine is after the reset.
boot_pc
Where start left the machine at boot. Unavailable when start read no address then.
verified
true when pc and boot_pc agree: the machine is back where it boots.
frame_at_reset
The frame counter at the reset. It does not restart, so subtract this before comparing a frame with a fresh start's.
taps_across
Each count or capture that was running across the reset, by ref, cpu and, for a port, space. Its totals carry on.
All 16 reply fields
  • boot_pc (string or object)
  • cpu (string)
  • following (array of string)
  • instruction (StopInstruction)
  • errorlog (object)
  • frame (integer)
  • frame_at_reset (integer)
  • kind (string)
  • pc (string)
  • regpoint_hit (object)
  • regs (object)
  • stop_seq (integer)
  • taps_across (array)
  • verified (boolean)
  • watch_hit (object)
  • watches (array)

What it refuses

Why it works this way

The reset is a stop of its own: MAME applies it only when the machine next runs, so the call steps one instruction. That stop is recorded with the readings every stop returns, described on the Run page.

There is no hard reset: it re-creates the machine and ends the debugger connection, which is close then start. It would save only MAME's launch, none of the emulated time to the screen that matters.

A reset is a write: the session block of every later reply carries altered, which says this session has written to the machine.

More on why

A count or capture watch counts on across a reset, because its tables live outside the machine. For a fresh count, declaration_clear(kind="watch") and add the watch again.

The contract your client receives
Text
Soft reset, applied, without relaunching. Modifies the machine; needs it stopped. A hard
reset is `close` then `start`.

Reply: `kind`, `pc`, `boot_pc`, `verified` (the two agree), `frame_at_reset`, `taps_across` (a
tap was live), and the stop's survey (`cause`, `stop_seq`, `cpu`, `regs`, `frame`,
`instruction`, `following`, `errorlog`, `watches`, `watch_hit`, `regpoint_hit`). The frame
counter does not restart: subtract `frame_at_reset` before comparing with a cold boot.

Reference: REFERENCE.md, Reading and writing.
Tools