Add memory mapper support by mullinmax · Pull Request #327 · warped-pinball/vector

mullinmax · 2026-03-11T21:22:19Z

Description

Adds required routes / features to broadcast to a memory mapper tool on the LAN

github-actions · 2026-03-11T21:23:09Z

Developer build links:
Sys11

https://raw.githubusercontent.com/warped-pinball/vector/pr-update-artifacts/pr-artifacts/pr-327/sys11-update.json

Sys11 (Tiny)

https://raw.githubusercontent.com/warped-pinball/vector/pr-update-artifacts/pr-artifacts/pr-327/sys11-tiny-update.json

WPC

https://raw.githubusercontent.com/warped-pinball/vector/pr-update-artifacts/pr-artifacts/pr-327/wpc-update.json

EM

https://raw.githubusercontent.com/warped-pinball/vector/pr-update-artifacts/pr-artifacts/pr-327/em-update.json

WhiteStar

https://raw.githubusercontent.com/warped-pinball/vector/pr-update-artifacts/pr-artifacts/pr-327/whitestar-update.json

DataEast

https://raw.githubusercontent.com/warped-pinball/vector/pr-update-artifacts/pr-artifacts/pr-327/data-east-update.json

…ball/vector into memory-mapper-support

Copilot

Pull request overview

Adds an admin-controlled debug feature to broadcast RAM snapshots over UDP, backed by a new authenticated API route and scheduler support.

Changes:

Add “Broadcast Memory Snapshot” switch to the admin UI and wire it to a new API endpoint.
Implement /api/memory/toggle-broadcast and a scheduled UDP broadcaster for SRAM snapshots.
Add unschedule() to the phew.server scheduler and document the new route in docs/routes.md.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
`src/common/web/js/admin.js`	Adds UI initialization for a new broadcast toggle that calls the backend API.
`src/common/web/html/admin.html`	Adds a debug switch control for enabling/disabling memory snapshot broadcasting.
`src/common/phew/server.py`	Introduces `unschedule()` to remove scheduled tasks.
`src/common/backend.py`	Implements snapshot broadcasting over UDP and an authenticated toggle route using the scheduler.
`docs/routes.md`	Adds the new API route to the route documentation.

You can also share your feedback on Copilot code review. Take the survey.

Copilot · 2026-03-11T21:27:17Z

+def unschedule(func):
+    global _scheduled_tasks
+    _scheduled_tasks = [t for t in _scheduled_tasks if t[0] != func]


unschedule() reassigns _scheduled_tasks to a new list. run_scheduled() iterates _scheduled_tasks and then mutates the global list later in the loop (_scheduled_tasks.remove(t) / _scheduled_tasks[i] = ...). If unschedule() runs concurrently (e.g., via an HTTP route) this can cause ValueError/IndexError and crash the scheduler coroutine. Avoid reassigning the list (keep identity stable) and make add/remove operations safe while run_scheduled() is iterating (e.g., apply pending ops inside run_scheduled, or pause scheduling while mutating).

Copilot · 2026-03-11T21:27:17Z

+def broadcast_memory_snapshot():
+    import discovery
+
+    ram_access = bytes(uctypes.bytearray_at(SRAM_DATA_BASE, SRAM_DATA_LENGTH))
+    chunk_size = 256
+    offset = 0
+
+    while offset < len(ram_access):
+        chunk = ram_access[offset : offset + chunk_size]
+        # Prepend 4-byte offset header to each chunk
+        message = offset.to_bytes(4, "big") + chunk
+        discovery.send_sock.sendto(message, ("255.255.255.255", 2040))
+        offset += chunk_size
+
+    return
+
+
+@add_route("/api/memory/toggle-broadcast", auth=True)
+def app_memory_broadcast(request):
+    data = request.data
+    if data.get("enable", False):
+        # add function call to scheduler
+        freq = data.get("frequency_ms", 100)
+        schedule(broadcast_memory_snapshot, phase_ms=0, frequency_ms=freq)  # broadcast every freq milliseconds
+    else:
+        # remove function call from scheduler
+        unschedule(broadcast_memory_snapshot)


broadcast_memory_snapshot() allocates a full bytes(...) copy of SRAM and then sends the entire snapshot in a tight loop every time it runs. With the default 100ms frequency this is likely to monopolize the scheduler (and can cause allocation pressure / watchdog issues), impacting core tasks like poll_fast. Consider broadcasting incrementally across scheduler ticks (one chunk per run), and/or enforce a much larger minimum interval.

Suggested change

def broadcast_memory_snapshot():

import discovery

ram_access = bytes(uctypes.bytearray_at(SRAM_DATA_BASE, SRAM_DATA_LENGTH))

chunk_size = 256

offset = 0

while offset < len(ram_access):

chunk = ram_access[offset : offset + chunk_size]

# Prepend 4-byte offset header to each chunk

message = offset.to_bytes(4, "big") + chunk

discovery.send_sock.sendto(message, ("255.255.255.255", 2040))

offset += chunk_size

return

@add_route("/api/memory/toggle-broadcast", auth=True)

def app_memory_broadcast(request):

data = request.data

if data.get("enable", False):

# add function call to scheduler

freq = data.get("frequency_ms", 100)

schedule(broadcast_memory_snapshot, phase_ms=0, frequency_ms=freq) # broadcast every freq milliseconds

else:

# remove function call from scheduler

unschedule(broadcast_memory_snapshot)

MEM_SNAPSHOT_CHUNK_SIZE = const(256)

_memory_broadcast_buf = None

_memory_broadcast_offset = 0

_memory_broadcast_active = False

def broadcast_memory_snapshot():

import discovery

global _memory_broadcast_buf, _memory_broadcast_offset, _memory_broadcast_active

# Initialize a new snapshot if one is not already in progress

if not _memory_broadcast_active:

try:

_memory_broadcast_buf = bytes(uctypes.bytearray_at(SRAM_DATA_BASE, SRAM_DATA_LENGTH))

except MemoryError:

# If we cannot allocate the snapshot buffer, skip this cycle

_memory_broadcast_buf = None

_memory_broadcast_offset = 0

_memory_broadcast_active = False

return

_memory_broadcast_offset = 0

_memory_broadcast_active = True

if not _memory_broadcast_buf:

return

# If we've reached the end of the buffer, reset state so the next call starts a new snapshot

if _memory_broadcast_offset >= len(_memory_broadcast_buf):

_memory_broadcast_buf = None

_memory_broadcast_offset = 0

_memory_broadcast_active = False

return

# Send a single chunk per scheduler invocation to avoid monopolizing the scheduler

start = _memory_broadcast_offset

end = start + MEM_SNAPSHOT_CHUNK_SIZE

if end > len(_memory_broadcast_buf):

end = len(_memory_broadcast_buf)

chunk = _memory_broadcast_buf[start:end]

# Prepend 4-byte offset header to each chunk

message = start.to_bytes(4, "big") + chunk

discovery.send_sock.sendto(message, ("255.255.255.255", 2040))

_memory_broadcast_offset = end

# When we've finished sending the entire buffer, clear state so a new snapshot will start next time

if _memory_broadcast_offset >= len(_memory_broadcast_buf):

_memory_broadcast_buf = None

_memory_broadcast_offset = 0

_memory_broadcast_active = False

@add_route("/api/memory/toggle-broadcast", auth=True)

def app_memory_broadcast(request):

global _memory_broadcast_buf, _memory_broadcast_offset, _memory_broadcast_active

data = request.data

if data.get("enable", False):

# Reset broadcast state whenever we (re)enable broadcasting

_memory_broadcast_buf = None

_memory_broadcast_offset = 0

_memory_broadcast_active = False

# Add function call to scheduler with a minimum interval to avoid overload

freq = data.get("frequency_ms", 100)

if freq < 250:

freq = 250

schedule(broadcast_memory_snapshot, phase_ms=0, frequency_ms=freq) # broadcast every freq milliseconds

else:

# Remove function call from scheduler and clear state

unschedule(broadcast_memory_snapshot)

_memory_broadcast_buf = None

_memory_broadcast_offset = 0

_memory_broadcast_active = False

Copilot · 2026-03-11T21:27:17Z

+def broadcast_memory_snapshot():
+    import discovery
+
+    ram_access = bytes(uctypes.bytearray_at(SRAM_DATA_BASE, SRAM_DATA_LENGTH))
+    chunk_size = 256
+    offset = 0
+
+    while offset < len(ram_access):
+        chunk = ram_access[offset : offset + chunk_size]
+        # Prepend 4-byte offset header to each chunk
+        message = offset.to_bytes(4, "big") + chunk
+        discovery.send_sock.sendto(message, ("255.255.255.255", 2040))
+        offset += chunk_size


discovery.send_sock can be None (e.g., after LowMemoryMode closes sockets) and this function bypasses discovery._setup_sockets()/discovery._send(). This will raise and then keep failing on every scheduled run. Re-initialize sockets (or guard/unschedule on failure) before calling sendto.

Copilot · 2026-03-11T21:27:18Z

+    data = request.data
+    if data.get("enable", False):
+        # add function call to scheduler
+        freq = data.get("frequency_ms", 100)


The toggle route schedules a new task every time it receives enable truthy; repeated calls can stack duplicate scheduled entries for the same function. Consider de-duplicating (unschedule first or check existing tasks) and validate/coerce frequency_ms to an int with sane bounds to prevent accidental overload.

Suggested change

freq = data.get("frequency_ms", 100)

raw_freq = data.get("frequency_ms", 100)

try:

freq = int(raw_freq)

except (TypeError, ValueError):

freq = 100

# Clamp frequency to sane bounds to avoid overload or absurd delays

if freq < 10:

freq = 10

elif freq > 60000:

freq = 60000

# Ensure we don't accumulate multiple scheduled entries for the same function

unschedule(broadcast_memory_snapshot)

Copilot · 2026-03-11T21:27:18Z

+    const payload = { enable: memoryBroadcastToggle.checked ? 1 : 0 };
+    try {
+      await window.smartFetch("/api/memory/toggle-broadcast", payload, true);
+    } catch (error) {
+      console.error("Failed to toggle memory broadcast", error);
+      memoryBroadcastToggle.checked = !memoryBroadcastToggle.checked;


This handler assumes smartFetch failures will throw, but fetch() only throws on network errors. Non-2xx responses (including 401/500) won’t hit catch, so the UI can remain checked even though the backend rejected the change. Capture the Response from smartFetch and revert the checkbox if !response.ok.

Suggested change

const payload = { enable: memoryBroadcastToggle.checked ? 1 : 0 };

try {

await window.smartFetch("/api/memory/toggle-broadcast", payload, true);

} catch (error) {

console.error("Failed to toggle memory broadcast", error);

memoryBroadcastToggle.checked = !memoryBroadcastToggle.checked;

const previousChecked = !memoryBroadcastToggle.checked;

const payload = { enable: memoryBroadcastToggle.checked ? 1 : 0 };

try {

const response = await window.smartFetch(

"/api/memory/toggle-broadcast",

payload,

true,

);

if (!response.ok) {

console.error(

"Failed to toggle memory broadcast, server returned status",

response.status,

);

memoryBroadcastToggle.checked = previousChecked;

}

} catch (error) {

console.error("Failed to toggle memory broadcast", error);

memoryBroadcastToggle.checked = previousChecked;

Copilot · 2026-03-11T21:27:18Z

+        <b>Broadcast Memory Snapshots:</b> enables streaming of memory over
+        network for debugging purposes. Use with tools like memory mapper
+      </label>


Broadcasting raw RAM snapshots over UDP is inherently sensitive (it can include credentials/tokens or other secrets in memory) and is visible to anyone on the local network. The UI copy should include an explicit warning about sensitive data and recommending use only on trusted networks / during active debugging.

Suggested change

Broadcast Memory Snapshots: enables streaming of memory over

network for debugging purposes. Use with tools like memory mapper

</label>

Broadcast Memory Snapshots (advanced, sensitive): streams raw memory over the network for debugging purposes (for use with tools like memory mapper).

Warning:

may include passwords, tokens, or other secrets in RAM. Enable only on trusted networks during active debugging and turn off when not in use.

Add memory broadcast functionality with toggle in admin interface

c7c5e6d

mullinmax self-assigned this Mar 11, 2026

Copilot AI review requested due to automatic review settings March 11, 2026 21:22

chore: regenerate API docs [skip ci]

73ca751

Copilot started reviewing on behalf of mullinmax March 11, 2026 21:22 View session

mullinmax and others added 4 commits March 11, 2026 17:24

Update broadcast memory snapshot description for clarity

40aa9a0

Merge branch 'memory-mapper-support' of https://github.com/warped-pin…

edace7b

…ball/vector into memory-mapper-support

ci: auto-bump required patch versions

7a0729b

Merge branch 'main' into memory-mapper-support

4e3eda8

Copilot AI reviewed Mar 11, 2026

View reviewed changes

mullinmax and others added 4 commits April 7, 2026 18:54

Merge branch 'main' into memory-mapper-support

67f57e9

chore: regenerate API docs [skip ci]

9600796

Merge branch 'main' into memory-mapper-support

c2238dc

ci: auto-bump required patch versions

4890d04

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add memory mapper support#327

Add memory mapper support#327
mullinmax wants to merge 10 commits intomainfrom
memory-mapper-support

mullinmax commented Mar 11, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Mar 11, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 11, 2026

Uh oh!

Copilot AI Mar 11, 2026

Uh oh!

Copilot AI Mar 11, 2026

Uh oh!

Copilot AI Mar 11, 2026

Uh oh!

Copilot AI Mar 11, 2026

Uh oh!

Copilot AI Mar 11, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

-        freq = data.get("frequency_ms", 100)
+        raw_freq = data.get("frequency_ms", 100)
+        try:
+            freq = int(raw_freq)
+        except (TypeError, ValueError):
+            freq = 100
+        # Clamp frequency to sane bounds to avoid overload or absurd delays
+        if freq < 10:
+            freq = 10
+        elif freq > 60000:
+            freq = 60000
+        # Ensure we don't accumulate multiple scheduled entries for the same function
+        unschedule(broadcast_memory_snapshot)

Conversation

mullinmax commented Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Uh oh!

github-actions Bot commented Mar 11, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Mar 11, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 11, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 11, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 11, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 11, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 11, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

mullinmax commented Mar 11, 2026 •

edited

Loading