diag(net): #260 outbound/command-gate probe + corrected issue framing

The two-agent investigation refuted #260's as-filed hypotheses: every
UseWithTarget was acked (the J5.2 use gate never latched), and the LOH
leak is bounded sawtooth churn - the real climb is ~2.25 GB of native/
GPU memory (WS 3,261 vs managed 1,015 MiB at wedge). The wedge evidence
also showed why it could hide: the live combat toggle routes through the
generation-gated runtime command seam, and every rejection exit in that
chain (Disposed / StaleGeneration / !IsInWorld at Validate, plus the
operations slot reading IsInWorld=false when unbound) is COMPLETELY
silent - no log, no event.

ACDREAM_PROBE_NET=1 (NetDiagnostics owner, PhysicsDiagnostics pattern)
now arms three probe families, all zero-cost when off:

- [net-out] per reliable send at the SendGameMessage chokepoint: opcode,
  GameAction type+sequence, fragment/packet sequence, managed thread id
  (two tids would prove the cross-thread ISAAC-desync hypothesis alone),
  and state; [net-out-EX] via an exception FILTER that logs without
  catching, so propagation is unchanged.
- [net-tick] 1 Hz cadence from WorldSession.Tick: inbound/s, queue
  depth, budget breaks, worst inter-tick gap (frame-stall witness),
  out/s, acks/s.
- [cmd-gate] every silent runtime-command rejection with expected-vs-
  view generation, lifecycle, and IsInWorld, plus the combat toggle
  result (whose Inactive exit reads a DIFFERENT IsInWorld source).

One walked-portal repro session with this probe distinguishes all
remaining #260 wedge hypotheses. ISSUES.md #260 rewritten to the
corrected two-root framing.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Erik 2026-07-29 08:53:58 +02:00
parent ab3da28c34
commit 534bacbc23
4 changed files with 245 additions and 55 deletions

View file

@ -97,14 +97,15 @@ Copy this block when adding a new issue:
---
## #260 — Portal-network wedge: server-round-trip actions die, LOH leaks — REPRODUCED
## #260 — Portal-network wedge: outbound actions die + native/GPU memory climbs
**Status:** OPEN — reproduced live, forensics captured, investigation not started
**Status:** OPEN — investigated 2026-07-29 (two-agent report-only pass over the
captured evidence); fix authorized by the user. **Verdict: two separate roots.**
**Severity:** HIGH (renders the client unplayable after sustained portal use)
**Filed:** 2026-07-29 (instrumented Coldeve session, Vulkan backend)
**Component:** interaction/use transaction gate + inbound stream + managed memory
**Component:** WorldSession outbound boundary + Vulkan resource lifecycle
**Supersedes framing of:** #256 (invisible portal-network objects) and #257
(~1.5 GB working set) — both are almost certainly facets of this.
(~1.5 GB working set) — both are facets of this.
**Reproduction (the first solid one):** on Coldeve (`play.coldeve.ac`), character
Barris, after several portal-network runs returning to the town portal network,
@ -116,36 +117,41 @@ gcdump taken *in the broken state* (15 MB), full session log with
`[cell-transit]`/`[input]`/`[use-target]`/`equipment` traces, the
`dotnet-counters` CSV, and stderr.
**Two symptoms, established from the traces:**
1. **The outbound action pipeline is wedged.** `CombatToggleCombat Press` × 3
produced **zero outbound send** (movement, being client-predicted, still
produces `[cell-transit]`). Signature of the one-request-at-a-time
interaction/use gate (J5.2) latching closed because a `UseWithTarget` never
received its server ack — a burst of `[use-target] … SEND UseWithTarget`
precedes the hang. Every server-round-trip action (portal use, combat mode)
is then silently dropped; local-only actions still work.
2. **A managed leak on the LOH — this one is ours.** `dotnet-counters` at wedge:
**LOH 601,807,472 B (574 MB)**, process WS 3.2 GB — distinct from the flat
~1.9 GB residency *level* the V11 soak plateaued at (that was benign; this
climbs under live walked play). Fingerprint matches the project's prior LOH
leak class (per-event `float[]`/`byte[]` never released). 185 `equipment:
attached … RightHandCombat` re-attach lines but only 1 CreateObject, so it is
re-attach/replay churn on existing entities, not fresh-object accumulation.
**Root 1 — the wedge (cause still open, field narrowed).** The as-filed
hypothesis (J5.2 use gate latched by an un-acked `UseWithTarget`) is **refuted**:
the log shows every `UseWithTarget` received its matching `[use-done]`, all
reveal generations 2→17 reached `event=complete cancelled=False failures=0`,
and `RuntimeCombatModeState.Toggle()`
(`src/AcDream.Runtime/Gameplay/RuntimeCombatModeState.cs:49-93`) never touches
that gate — it gates only on `IsInWorld`. Remaining ranked hypotheses:
(1) frame-thread saturation — inbound drains from an *unbounded*
`_inboundQueue` on a ~4 ms frame budget (`WorldSession.cs:990-1016`), collapsed
by Root 2's memory pressure; (2) outbound ISAAC/sequence desync — a send off
the frame thread desyncs the cipher (`NextGameActionSequence()` is a non-atomic
`++`, `WorldSession.cs:1618`) and the server silently drops every subsequent
packet, which uniquely explains hard-zero effect while client-predicted
movement survives.
**Hypothesis (unproven, for the investigation):** a jammed inbound stream could
both starve the ack the use-gate waits on AND accumulate LOH buffers — one root,
two symptoms. Or two bugs. The gcdump names the LOH type by retainer; the log
timeline names the unacked request. Not confirmed either way.
**Root 2 — the memory climb is NATIVE/GPU, not managed.** The as-filed "LOH
leak" is wrong: LOH is a bounded sawtooth (318→829→318 MiB; churn, not
retention) and live managed heap at wedge was only 333 MB. The monotonic climb
is **working set 1,295→3,261 MiB (~180 MB/min)**; at peak, WS 3,261 vs managed
committed 1,015 MiB leaves **~2.25 GB unaccounted native/GPU memory**. Prime
suspect: Vulkan device resources minted per equipment re-attach (185187
`equipment: attached … RightHandCombat` lines vs 1 CreateObject) and per vfx
setup, never released.
**Why the scripted gates missed it:** the V11 portal-churn soak used `/teleloc`,
which enters the transit state machine by a different door than a *walked* portal
transit — exactly the gap the walked-play session was designed to probe.
**Next (investigation, user-approved fix only):** read the gcdump retainer graph
per `reference_memory_leak_toolchain.md`; reconstruct the use-gate timeline from
the log to find the unacked request; determine shared-vs-separate root. Do NOT
patch the symptom (a gate timeout would be the classic forbidden workaround) —
find why the ack never lands.
**Next (authorized):** (a) instrument the `WorldSession` outbound boundary —
opcode+sequence per send, swallowed exceptions, `IsInWorld`, per-frame dt,
inbound-queue depth — and reproduce with walked portal play; at a dead combat
toggle, whether the send reaches the wire and whether the sequence still
advances distinguishes the remaining hypotheses. (b) audit the
attach→GPU-resource path for create-without-release. Do NOT patch the symptom
(a gate timeout or retry loop is the classic forbidden workaround).
---

View file

@ -231,6 +231,17 @@ internal sealed class CurrentGameRuntimeCommandAdapter
RuntimeCommandStatus.Inactive,
_ => RuntimeCommandStatus.Rejected,
};
// #260 probe: the toggle's own Inactive exit comes from a
// DIFFERENT IsInWorld source than Validate's (the operations
// slot, which reads false when unbound). Logging the result
// here separates "gate passed, toggle refused" from "gate
// rejected" — the two look identical to the player.
if (AcDream.Core.Net.NetDiagnostics.ProbeNet)
{
Console.WriteLine(
$"[cmd-gate] combat toggle result={result.Status}"
+ $" mode={result.Mode}");
}
}
else
{
@ -753,13 +764,31 @@ internal sealed class CurrentGameRuntimeCommandAdapter
RuntimeGenerationToken expectedGeneration,
bool requireWorld)
{
RuntimeCommandStatus status;
if (_view.Lifecycle.State == RuntimeLifecycleState.Disposed)
return RuntimeCommandStatus.Inactive;
if (expectedGeneration != _view.Generation)
return RuntimeCommandStatus.StaleGeneration;
if (requireWorld && !_session.IsInWorld)
return RuntimeCommandStatus.Inactive;
return RuntimeCommandStatus.Accepted;
status = RuntimeCommandStatus.Inactive;
else if (expectedGeneration != _view.Generation)
status = RuntimeCommandStatus.StaleGeneration;
else if (requireWorld && !_session.IsInWorld)
status = RuntimeCommandStatus.Inactive;
else
status = RuntimeCommandStatus.Accepted;
// #260 probe: a rejected generation-gated command is otherwise
// COMPLETELY silent (no log, no event) — which is exactly how the
// portal-network wedge hid. Log every rejection with the full gate
// state so the failing predicate names itself.
if (status != RuntimeCommandStatus.Accepted
&& AcDream.Core.Net.NetDiagnostics.ProbeNet)
{
Console.WriteLine(
$"[cmd-gate] REJECT status={status}"
+ $" expected={expectedGeneration}"
+ $" view={_view.Generation}"
+ $" lifecycle={_view.Lifecycle.State}"
+ $" inWorld={_session.IsInWorld}");
}
return status;
}
private RuntimeCommandResult Result(

View file

@ -0,0 +1,36 @@
namespace AcDream.Core.Net;
/// <summary>
/// Diagnostic owner for the <c>ACDREAM_PROBE_NET</c> probe family (#260).
/// Read once at startup, following the <c>PhysicsDiagnostics</c> pattern.
///
/// <para>
/// When enabled, three probe line families are emitted:
/// <list type="bullet">
/// <item><c>[net-out]</c> — one line per outbound reliable game message at the
/// <c>WorldSession.SendGameMessage</c> chokepoint: opcode, game-action type +
/// sequence (when the body is a 0xF7B1 GameAction), fragment/packet sequence,
/// managed thread id, and session state. An exception escaping the wire write
/// additionally emits <c>[net-out-EX]</c> via an exception filter (log
/// without catching — behavior is unchanged).</item>
/// <item><c>[net-tick]</c> — a once-per-second cadence summary from
/// <c>WorldSession.Tick</c>: inbound datagrams/s, remaining queue depth,
/// budget-break count, worst inter-tick gap (= worst frame stall as seen by
/// the net pump), outbound sends/s, and acks/s.</item>
/// <item><c>[cmd-gate]</c> — one line per generation-gated runtime command
/// REJECTION in <c>CurrentGameRuntimeCommandAdapter.Validate</c> (status,
/// expected vs view generation, lifecycle, IsInWorld) plus the combat-toggle
/// result. These rejections are otherwise completely silent, which is what
/// let #260's wedge hide.</item>
/// </list>
/// Zero steady-state cost when off: every site is behind this single bool.
/// </para>
/// </summary>
public static class NetDiagnostics
{
/// <summary>
/// <c>ACDREAM_PROBE_NET=1</c> — #260 outbound/command-gate probe family.
/// </summary>
public static bool ProbeNet { get; set; } =
Environment.GetEnvironmentVariable("ACDREAM_PROBE_NET") == "1";
}

View file

@ -990,6 +990,7 @@ public sealed class WorldSession : IDisposable
public int Tick()
{
int processed = 0;
bool budgetBroke = false;
long start = Stopwatch.GetTimestamp();
while (_inboundQueue.Reader.TryRead(
out PooledInboundDatagram datagram))
@ -1010,11 +1011,74 @@ public sealed class WorldSession : IDisposable
// tolerance (holtburger defers acks on a flush cadence). The tail stays queued
// (unbounded channel, FIFO) and drains next frame.
if (InboundBudgetExceeded(CurrentState, start, Stopwatch.GetTimestamp(), InboundBudgetTicks))
{
budgetBroke = true;
break;
}
}
if (NetDiagnostics.ProbeNet)
ProbeNetTickCadence(start, processed, budgetBroke);
return processed;
}
// #260 probe state — only touched when NetDiagnostics.ProbeNet is set.
// The inter-Tick gap doubles as a frame-stall witness: Tick runs once per
// frame on the frame thread, so a GC pause or saturated frame shows up
// directly as maxgap. _probeSendWindow/_probeAckWindow are Interlocked so
// a hypothetical off-thread send can't corrupt the window counters (the
// [net-out] tid field is what would prove such a send exists).
private long _probeLastTickTs;
private long _probeWindowStartTs;
private long _probeMaxGapTicks;
private int _probeProcessedWindow;
private int _probeBudgetBreaks;
private int _probeSendWindow;
private int _probeAckWindow;
/// <summary>
/// #260 probe: accumulate per-Tick cadence facts and emit one
/// <c>[net-tick]</c> summary line per second.
/// </summary>
private void ProbeNetTickCadence(long tickStartTs, int processed, bool budgetBroke)
{
if (_probeLastTickTs != 0)
{
long gap = tickStartTs - _probeLastTickTs;
if (gap > _probeMaxGapTicks)
_probeMaxGapTicks = gap;
}
_probeLastTickTs = tickStartTs;
_probeProcessedWindow += processed;
if (budgetBroke)
_probeBudgetBreaks++;
if (_probeWindowStartTs == 0)
{
_probeWindowStartTs = tickStartTs;
return;
}
long windowTicks = tickStartTs - _probeWindowStartTs;
if (windowTicks < Stopwatch.Frequency)
return;
double windowSeconds = (double)windowTicks / Stopwatch.Frequency;
double maxGapMs = _probeMaxGapTicks * 1000.0 / Stopwatch.Frequency;
int sends = Interlocked.Exchange(ref _probeSendWindow, 0);
int acks = Interlocked.Exchange(ref _probeAckWindow, 0);
Console.WriteLine(
$"[net-tick] in/s={_probeProcessedWindow / windowSeconds:F0}"
+ $" q={_inboundQueue.Reader.Count}"
+ $" budget-breaks={_probeBudgetBreaks}"
+ $" maxgap={maxGapMs:F0}ms"
+ $" out/s={sends / windowSeconds:F0}"
+ $" acks/s={acks / windowSeconds:F0}"
+ $" st={CurrentState}");
_probeWindowStartTs = tickStartTs;
_probeMaxGapTicks = 0;
_probeProcessedWindow = 0;
_probeBudgetBreaks = 0;
}
/// <summary>
/// Pure, testable decision for the per-frame inbound bound: stop draining only when
/// in-world AND the elapsed Stopwatch ticks have reached the budget. Extracted so the
@ -2162,28 +2226,81 @@ public sealed class WorldSession : IDisposable
private void SendGameMessage(byte[] gameMessageBody, GameMessageGroup queue)
{
Span<byte> datagram = stackalloc byte[
PacketHeader.Size
+ MessageFragmentHeader.MaxFragmentSize];
int fragmentLength =
GameMessageFragment.WriteSingleFragment(
datagram.Slice(PacketHeader.Size),
_fragmentSequence++,
queue,
gameMessageBody);
var header = new PacketHeader
// #260 probe: log the send BEFORE the sequence counters are consumed
// so the line carries the values this datagram will actually use. The
// exception filter below logs a wire-write fault WITHOUT catching it
// (the filter returns false), so behavior is byte-identical either way.
if (NetDiagnostics.ProbeNet)
ProbeNetLogOutbound(gameMessageBody, queue);
try
{
Sequence = _clientPacketSequence++,
Flags = PacketHeaderFlags.BlobFragments | PacketHeaderFlags.EncryptedChecksum,
Id = _sessionClientId,
};
int datagramLength = PacketCodec.FinalizeInPlace(
header,
datagram,
fragmentLength,
optionalLength: 0,
_outboundIsaac);
_net.Send(datagram.Slice(0, datagramLength));
Span<byte> datagram = stackalloc byte[
PacketHeader.Size
+ MessageFragmentHeader.MaxFragmentSize];
int fragmentLength =
GameMessageFragment.WriteSingleFragment(
datagram.Slice(PacketHeader.Size),
_fragmentSequence++,
queue,
gameMessageBody);
var header = new PacketHeader
{
Sequence = _clientPacketSequence++,
Flags = PacketHeaderFlags.BlobFragments | PacketHeaderFlags.EncryptedChecksum,
Id = _sessionClientId,
};
int datagramLength = PacketCodec.FinalizeInPlace(
header,
datagram,
fragmentLength,
optionalLength: 0,
_outboundIsaac);
_net.Send(datagram.Slice(0, datagramLength));
}
catch (Exception ex) when (ProbeNetLogOutboundFault(ex))
{
// Unreachable: the filter always returns false so the original
// exception propagates to the caller exactly as before.
throw;
}
if (NetDiagnostics.ProbeNet)
Interlocked.Increment(ref _probeSendWindow);
}
/// <summary>
/// #260 probe: one <c>[net-out]</c> line per outbound reliable game
/// message. The thread id is load-bearing evidence — every send is
/// supposed to happen on the frame thread (the ISAAC keystream is
/// single-threaded); two distinct tids across [net-out] lines would
/// prove the cross-thread-send/cipher-desync hypothesis by itself.
/// </summary>
private void ProbeNetLogOutbound(byte[] body, GameMessageGroup queue)
{
uint op = body.Length >= 4
? BinaryPrimitives.ReadUInt32LittleEndian(body)
: 0u;
string detail = string.Empty;
if (op == 0xF7B1 && body.Length >= 12)
{
uint gseq = BinaryPrimitives.ReadUInt32LittleEndian(body.AsSpan(4));
uint act = BinaryPrimitives.ReadUInt32LittleEndian(body.AsSpan(8));
detail = $" act=0x{act:X4} gseq={gseq}";
}
Console.WriteLine(
$"[net-out] op=0x{op:X4}{detail} q={queue} fseq={_fragmentSequence}"
+ $" pseq={_clientPacketSequence} len={body.Length}"
+ $" tid={Environment.CurrentManagedThreadId} st={CurrentState}");
}
private bool ProbeNetLogOutboundFault(Exception ex)
{
if (NetDiagnostics.ProbeNet)
{
Console.WriteLine(
$"[net-out-EX] {ex.GetType().Name}: {ex.Message}"
+ $" tid={Environment.CurrentManagedThreadId} st={CurrentState}");
}
return false;
}
/// <summary>
@ -2239,6 +2356,8 @@ public sealed class WorldSession : IDisposable
optionalLength: sizeof(uint),
outboundIsaac: null);
_net.Send(datagram.Slice(0, datagramLength));
if (NetDiagnostics.ProbeNet)
Interlocked.Increment(ref _probeAckWindow);
}
private void Transition(State next)