ModernUO/Projects/UOContent/Engines/Pathing/Cache/StepProbe.cs
Kamron Batman b852bca41e
perf(pathing): pool the StepCache strata buffer, then clean up the pathing engine around it (#2523)
Started as an allocation pass over `StepCache` and grew into a cleanup of the surrounding pathing engine. Four commits, each independently reviewable; net **−560 lines**.

Build clean (0 warnings). All 122 `Server.Tests.Pathfinding` tests pass.

---

## 1. `perf`: pool the strata buffer, cut a hot-path dictionary lookup

**The headline is that `TryGetMask` — the actual hot path — was already allocation-free.** `StepMask` is a readonly struct, `StaticTileEnumerable` is a `ref struct`, `ChunkMissState` is a struct in a `Dictionary`. So most of this is a bake-throughput and GC-churn win, with one exception noted below.

`BuildChunk` accumulated packed multi-Z strata into a `List<byte>` that grew by doubling (256 → 512 → 1024 → …) and then paid a final `ToArray()`. A full map bake runs it ~114k times. It now writes into a `byte[]` rented from `STArrayPool<byte>.Shared` through a span writer, and hands the chunk one exact-size copy.

**This required fixing a latent out-of-bounds guard.** The record-fit check reserved headroom for **8** strata (`StratumByteLength * 8`) while `ComputeStandableSurfaceZs` can return up to **16** — so a cell could write 305 bytes starting from a 65,383-byte offset. Against a `List` that was benign (it just grew past 64 KB, and emitted offsets stayed under the `NoStrata` sentinel). Against a fixed-size rented buffer it is an out-of-bounds write, so tightening it was a *prerequisite* for the pooling, not a drive-by. The guard is now exact, which additionally proves no emitted offset can collide with `NoStrata == ushort.MaxValue`.

**One genuine query-path win:** `ShouldPromoteAfterMiss` did *two* dictionary lookups per miss — a `TryGetValue`, then an indexer assignment that re-hashes and re-probes. It now mutates in place via `CollectionsMarshal.GetValueRefOrNullRef`. This runs on every uncached chunk touch during A* expansion. The window-expiry branch keeps its explicit early return, so `MissPromotionThreshold == 1` still resets rather than promoting.

Also dropped `StepProbe.ComputeStrataAt` / `ComputedStratum` (dead code, zero callers) and collapsed six 18-argument `new StepMask(0, 0, …, kind)` blocks into `Fallthrough(kind)`.

**Considered and rejected:** pooling the `Direction[]` that `Find` returns. It *escapes* the call — `MovementPath` holds it across ticks while `PathFollower` walks `m_Index` through it — so it cannot be rented-and-returned, and it cannot be borrowed from the shared `BitmapAStarAlgorithm.Instance` without one creature clobbering another's in-flight path. `CheckPath` rate-limits repaths to one per 2s per creature, putting this at roughly 60 KB/sec at 1,000 pathing creatures. Not worth a public API break plus a use-after-return footgun.

## 2. `docs`: rewrite the comments for publication

The comments had accumulated as development notes: internal phase jargon (`Tier 4`, `the Phase-2 synthesizer`), change narration aimed at a reviewer (`which the old ComputeStandingZ anchor missed`, `legacy behavior`), benchmark anecdotes (`benchmarked as near-optimal`, `a ~20 ns lookup`), and paragraphs restating the code.

Rewritten to keep the rationale you cannot recover by reading the code — why the source-Z guard cannot be widened, why multis fall through with a halo, why the promotion gate counts Finds rather than calls, why `ComputeFingerprint` must hash the *files* and not the live tile tables — and drop the history that got us there.

Three comments were **factually wrong**, not just wordy:

- `CacheEvictionTimer` and `CacheStats` documented a class called `StaticWalkabilityCache`. No such class exists — it is `StepCache`.
- `StepCacheFile` declared `File layout v8` while `FormatVersion` is 9, and called the current record layout "the v6 layout" in four places. The layout descriptions are now unversioned so they cannot drift again.
- `StepProbe.ComputeStandingZ` claimed `StepCache` uses it to bake `SourceZ`. It has not since the baker moved to the clearance-aware `ComputeStandableSurfaceZs`; only a parity test calls it.

## 3. `refactor`: simplify `StepCacheFile.Write`, consolidate the format tests

`SaveToFile` walked `_keysList` **twice** — once to count the map's chunks, then again through a `ChunkEnumerator` closure to emit them — because `Write` needed the count up front to size its index array. Both loops had the same root cause. Passing a **span** collapses them: the count is just `span.Length`.

That deletes the `ChunkEnumerator` delegate, the closure over the list enumerator, and **both `InvalidOperationException` throws**, which existed only to police the delegate's "yield exactly `chunkCount` chunks" contract — a contract a span makes unrepresentable.

`Write` now patches the header's `IndexOffset` by seeking back to it rather than reaching into the writer's live buffer with `BinaryPrimitives`. That also retires `IndexOffsetFieldPosition`, a hand-maintained byte offset that had to track the header layout, and sidesteps the stale-array hazard that motivated the manual patch (`BufferWriter` reallocates on growth).

**Tests:** `StepCacheFileV6/V7/V8Tests` were named for the format version that introduced each transform — and the format is now **v9**, so all three names described formats the loader rejects outright. Beyond triplicated builders and plumbing, two things were actually broken:

- The three near-identical rejection tests each cited a `MinSupportedVersion` that had since moved (`"version 5 < MinSupportedVersion 6"`, `"6 < 7"`, `"7 < 8"`). They passed for the wrong reason.
- `AssertBaseEqual` (used by V7 and V8) **silently skipped the swim and strata trailers**. A regression dropping either would not have failed those tests.

Now one `StepCacheFileFormatTests`, named for behavior — predictive-Z elision, compression, compact index — with a single `AssertIdentical` that does check both trailers, the three rejection tests folded into one theory that also covers a future version, and a zero-chunk case the delegate-based writer never had coverage for.

## 4. `test`: consolidate the parity and lifecycle tests

Three files tested "parity" and none of the names said *which*. They were three different layers, and the seams are the useful part, so they are now one `StepCacheParityTests` that names them:

| Test | Compares | Answers |
|---|---|---|
| `ProbeMatchesSlowPath` | StepProbe vs MovementImpl | Is the bake right? |
| `CacheMatchesProbe` | StepCache vs StepProbe | Is it stored and returned intact? |
| `CacheServesReachableWalkStates` | StepCache vs MovementImpl | End to end, over the states A* visits |

Merging removed a duplicated stub `Mobile`, duplicated region seeds, and a filename/class mismatch (`StepProbeParityTests.cs` declared `StaticWalkabilityParityTests`). `SwimBake_ProducesWetCells` moved with it — it lived in the cache parity file but never touched the cache.

Tests reached into `StepCache._chunks` via `GetField` in **9 places**, each rebuilding the key encoding and cell-index arithmetic by hand. `StepCache` now exposes `GetResidentChunk` and `ResidentIndexInSync` alongside the internal test hooks it already had (`LazyReaderHasChunk`, `CurrentFindGeneration`), and the shared arithmetic moved to `PathingTestSupport`. All 9 reflection blocks are gone.

`StepCacheLifecycleTests` is regrouped by what it covers — promotion gate, fallthrough routes, strata, swim layer, eviction — with the `Tier4*` names dropped. Removed `Singleton_IsAvailable`, which asserted an inline-initialized static property was not null; that is the entire 123 → 122 test-count delta.

---

## Verification

Tests were mutation-checked rather than just run, since round-trip and parity tests can pass while a transform silently no-ops:

- Injecting an off-by-one into the `IndexOffset` patch fails **15 of 123** — the format tests are load-bearing.
- Offsetting the cache's cell index by one fails **7 of 10** parity cases, and the 3 that stay green are exactly the ones that do not touch the cache. The layering localizes a fault rather than just reporting one.
2026-07-12 20:02:29 -07:00

435 lines
15 KiB
C#

using System;
using CalcMoves = Server.Movement.Movement;
namespace Server.Engines.Pathing.Cache;
/// <summary>
/// Computes the 8-direction "can step" mask and destination Zs for a single cell from land and
/// statics alone. Mirrors <see cref="MovementImpl"/>.Check minus the item and mobile collision
/// phases, which belong to the caller's dynamic-obstacle pass.
///
/// Multis (houses, boats) are excluded from the static bake because they are dynamic content;
/// cells they cover route to the live movement path via <see cref="StepCache"/>'s multi halo.
/// <see cref="ComputeMultiMaskAt"/> is the opt-in exception for those cells.
///
/// Each call bakes both rule sets: walker (canSwim=false, cantWalk=false) and swim-only
/// (canSwim=true, cantWalk=true). Diagonal corner-cut is not applied — callers hold the partner
/// bits in the same mask byte and combine them at query time.
/// </summary>
public static class StepProbe
{
private const int PersonHeight = 16;
private const int StepHeight = 2;
/// <summary>
/// Writes the surface Zs at (x, y) a default walker can actually stand on into
/// <paramref name="zs"/>, ascending, and returns the count. A candidate surface — the
/// walkable land centre, or any walkable static's top — qualifies only if a PersonHeight
/// envelope above it is clear of impassable statics.
///
/// The clearance test is what makes this the exact set of standing Zs the slow path can
/// resolve to: it drops surfaces a creature cannot occupy, like the land beneath a sewer
/// walkway or a low bridge. That in turn means two surviving surfaces are always at least
/// PersonHeight apart (an upper surface any closer would have taken the lower one's
/// clearance away), so one ascending pass with a duplicate skip suffices.
///
/// The baker anchors each cell here so static-over-land geometry — walkways, bridges, raised
/// foundations, upper floors — bakes at the Z a creature stands on rather than the land average.
/// </summary>
public static int ComputeStandableSurfaceZs(Map map, int x, int y, Span<sbyte> zs)
{
if (map == null || map == Map.Internal)
{
return 0;
}
if (x < 0 || y < 0 || x >= map.Width || y >= map.Height)
{
return 0;
}
Span<int> cand = stackalloc int[16];
var count = 0;
var landTile = map.Tiles.GetLandTile(x, y);
var landFlags = TileData.LandTable[landTile.ID & TileData.MaxLandValue].Flags;
if (!landTile.Ignored && (landFlags & TileFlag.Impassable) == 0)
{
map.GetAverageZ(x, y, out _, out var landCenter, out _);
cand[count++] = landCenter;
}
foreach (var tile in map.Tiles.GetStaticTiles(x, y))
{
if (count >= cand.Length)
{
break;
}
var data = TileData.ItemTable[tile.ID & TileData.MaxItemValue];
if (!data.Surface || data.Impassable)
{
continue;
}
cand[count++] = tile.Z + data.CalcHeight;
}
if (count == 0)
{
return 0;
}
cand[..count].Sort();
var n = 0;
for (var i = 0; i < count && n < zs.Length; i++)
{
var cz = (sbyte)Math.Clamp(cand[i], sbyte.MinValue + 1, sbyte.MaxValue);
if (n > 0 && zs[n - 1] == cz)
{
continue;
}
// Standable iff the creature's PersonHeight body envelope above this surface is
// free of impassable statics. The surface itself never blocks (its top == cz,
// which is the envelope floor, not inside it).
if (StaticsBlockAt(map, x, y, cz, cz + PersonHeight))
{
continue;
}
zs[n++] = cz;
}
return n;
}
public static StepMask ComputeMaskAt(Map map, int x, int y, sbyte sourceZ) =>
ComputeMaskCore(map, x, y, sourceZ, includeMultis: false);
/// <summary>
/// Multi-aware counterpart to <see cref="ComputeMaskAt"/>, for cells a multi covers or
/// neighbours: folds house/boat component tiles into the same surface/step logic. Builds the
/// whole 8-direction mask in one pass, where the slow path would run CheckMovement eight times.
/// </summary>
public static StepMask ComputeMultiMaskAt(Map map, int x, int y, sbyte sourceZ) =>
ComputeMaskCore(map, x, y, sourceZ, includeMultis: true);
/// <summary>
/// Shared 8-direction mask builder behind <see cref="ComputeMaskAt"/> and
/// <see cref="ComputeMultiMaskAt"/>. <paramref name="includeMultis"/> is the only difference:
/// it swaps the tile source to GetStaticAndMultiTiles so house and boat components participate.
/// </summary>
private static StepMask ComputeMaskCore(Map map, int x, int y, sbyte sourceZ, bool includeMultis)
{
if (map == null || map == Map.Internal)
{
return default;
}
var srcTiles = includeMultis ? map.Tiles.GetStaticAndMultiTiles(x, y) : map.Tiles.GetStaticTiles(x, y);
GetStaticStartZ(map, x, y, sourceZ, srcTiles, canSwim: false, cantWalk: false,
out var walkStartZ, out var walkStartTop, out _);
GetStaticStartZ(map, x, y, sourceZ, srcTiles, canSwim: true, cantWalk: true,
out var swimStartZ, out var swimStartTop, out _);
byte walkMask = 0;
byte wetMask = 0;
Span<sbyte> walkZs = stackalloc sbyte[8];
Span<sbyte> swimZs = stackalloc sbyte[8];
// stackalloc is not zero-initialized, and the loop below writes a slot only where the
// step succeeds, so blocked directions would otherwise carry stack garbage.
walkZs.Clear();
swimZs.Clear();
for (var d = 0; d < 8; d++)
{
var dx = x;
var dy = y;
CalcMoves.Offset((Direction)d, ref dx, ref dy);
var dTiles = includeMultis ? map.Tiles.GetStaticAndMultiTiles(dx, dy) : map.Tiles.GetStaticTiles(dx, dy);
if (CheckStaticStep(map, dx, dy, dTiles, walkStartZ, walkStartTop,
canSwim: false, cantWalk: false, out var walkZ))
{
walkMask |= (byte)(1 << d);
walkZs[d] = (sbyte)walkZ;
}
if (CheckStaticStep(map, dx, dy, dTiles, swimStartZ, swimStartTop,
canSwim: true, cantWalk: true, out var swimZ))
{
wetMask |= (byte)(1 << d);
swimZs[d] = (sbyte)swimZ;
}
}
return new StepMask(
walkMask, wetMask,
walkZs[0], walkZs[1], walkZs[2], walkZs[3],
walkZs[4], walkZs[5], walkZs[6], walkZs[7],
swimZs[0], swimZs[1], swimZs[2], swimZs[3],
swimZs[4], swimZs[5], swimZs[6], swimZs[7]
);
}
/// <summary>
/// The standing-Z a default walker at (x, y) resolves to under the slow path's
/// surface-selection rules: paver Z+1 over paver-on-ground, land centre on bare land.
/// The baker anchors cells with <see cref="ComputeStandableSurfaceZs"/> instead, which is
/// clearance-aware; this remains the direct MovementImpl equivalent for parity checks.
/// </summary>
public static int ComputeStandingZ(Map map, int x, int y, int locZ)
{
GetStaticStartZ(map, x, y, locZ, map.Tiles.GetStaticTiles(x, y), canSwim: false, cantWalk: false, out _, out _, out var zCenter);
return zCenter;
}
/// <summary>
/// Returns the water-surface standing Z at (x, y) — the Z a swim-only mob would stand
/// at on this cell — or <see cref="int.MinValue"/> if no water surface exists. Used
/// by <see cref="StepCache"/> to detect shore cells (cells with both walk and swim
/// surfaces separated by &gt; StepHeight) and bake their swim layer at swim-perspective Z.
/// </summary>
public static int ComputeSwimStandingZ(Map map, int x, int y)
{
if (map == null || map == Map.Internal || x < 0 || y < 0 || x >= map.Width || y >= map.Height)
{
return int.MinValue;
}
// Land tile flagged Wet — its center Z is the swim surface.
var landTile = map.Tiles.GetLandTile(x, y);
var landFlags = TileData.LandTable[landTile.ID & TileData.MaxLandValue].Flags;
if (!landTile.Ignored && (landFlags & TileFlag.Wet) != 0)
{
map.GetAverageZ(x, y, out _, out var landCenter, out _);
return landCenter;
}
// Otherwise scan statics for a wet surface.
foreach (var tile in map.Tiles.GetStaticTiles(x, y))
{
var data = TileData.ItemTable[tile.ID & TileData.MaxItemValue];
if (data.Wet)
{
return tile.Z + data.CalcHeight;
}
}
return int.MinValue;
}
/// <summary>
/// Mirrors GetStartZ from MovementImpl, parameterized by canSwim / cantWalk.
/// </summary>
private static void GetStaticStartZ(
Map map, int x, int y, int locZ, Map.StaticTileEnumerable tiles,
bool canSwim, bool cantWalk, out int zLow, out int zTop, out int zCenter
)
{
var landTile = map.Tiles.GetLandTile(x, y);
var flags = TileData.LandTable[landTile.ID & TileData.MaxLandValue].Flags;
var impassable = (flags & TileFlag.Impassable) != 0;
// Mirrors MovementImpl: impassable + swim on water is OK; otherwise block on
// cantWalk or impassable.
var landBlocks = (cantWalk || impassable) && !(impassable && canSwim && (flags & TileFlag.Wet) != 0);
map.GetAverageZ(x, y, out var landZ, out var landCenter, out var landTop);
var considerLand = !landTile.Ignored;
zCenter = zLow = zTop = 0;
var isSet = false;
if (considerLand && !landBlocks && locZ >= landCenter)
{
zLow = landZ;
zCenter = landCenter;
zTop = landTop;
isSet = true;
}
foreach (var tile in tiles)
{
var id = TileData.ItemTable[tile.ID & TileData.MaxItemValue];
var calcTop = tile.Z + id.CalcHeight;
if (isSet && calcTop < zCenter || locZ < calcTop || !id.Surface && !(canSwim && id.Wet))
{
continue;
}
zLow = tile.Z;
zCenter = calcTop;
var top = tile.Z + id.Height;
if (!isSet || top > zTop)
{
zTop = top;
}
isSet = true;
}
if (!isSet)
{
zLow = zTop = locZ;
}
else if (locZ > zTop)
{
zTop = locZ;
}
}
/// <summary>
/// Mirrors MovementImpl.Check for static tiles only, parameterized by canSwim / cantWalk.
/// Items and mobile collision phases are omitted.
/// </summary>
private static bool CheckStaticStep(
Map map, int x, int y, Map.StaticTileEnumerable tiles, int startZ, int startTop,
bool canSwim, bool cantWalk, out int newZ
)
{
newZ = 0;
if (x < 0 || y < 0 || x >= map.Width || y >= map.Height)
{
return false;
}
var landTile = map.Tiles.GetLandTile(x, y);
var flags = TileData.LandTable[landTile.ID & TileData.MaxLandValue].Flags;
var impassable = (flags & TileFlag.Impassable) != 0;
var landBlocks = (cantWalk || impassable) && !(impassable && canSwim && (flags & TileFlag.Wet) != 0);
var considerLand = !landTile.Ignored;
map.GetAverageZ(x, y, out var landZ, out var landCenter, out _);
var moveIsOk = false;
var stepTop = startTop + StepHeight;
var checkTop = startZ + PersonHeight;
int testTop;
foreach (var tile in tiles)
{
var itemData = TileData.ItemTable[tile.ID & TileData.MaxItemValue];
var notWater = !itemData.Wet;
// Mirrors MovementImpl: skip if not a passable surface AND not swimmable water,
// OR if the mobile can't walk and this isn't water.
if ((!itemData.Surface || itemData.Impassable) && (!canSwim || notWater)
|| cantWalk && notWater)
{
continue;
}
var itemZ = tile.Z;
var itemTop = itemZ;
var ourZ = itemZ + itemData.CalcHeight;
testTop = checkTop;
if (moveIsOk)
{
var cmp = Math.Abs(ourZ - startZ) - Math.Abs(newZ - startZ);
if (cmp > 0 || cmp == 0 && ourZ > newZ)
{
continue;
}
}
if (ourZ + PersonHeight > testTop)
{
testTop = ourZ + PersonHeight;
}
if (!itemData.Bridge)
{
itemTop += itemData.Height;
}
if (stepTop < itemTop)
{
continue;
}
var landCheck = itemZ + Math.Min(itemData.Height, StepHeight);
if (considerLand && landCheck < landCenter && landCenter > ourZ && testTop > landZ)
{
continue;
}
if (StaticsBlockAt(map, x, y, ourZ, testTop))
{
continue;
}
newZ = ourZ;
moveIsOk = true;
}
if (!considerLand || landBlocks || stepTop < landZ)
{
return moveIsOk;
}
testTop = checkTop;
if (landCenter + PersonHeight > testTop)
{
testTop = landCenter + PersonHeight;
}
var shouldCheck = true;
if (moveIsOk)
{
var cmp = Math.Abs(landCenter - startZ) - Math.Abs(newZ - startZ);
if (cmp > 0 || cmp == 0 && landCenter > newZ)
{
shouldCheck = false;
}
}
if (shouldCheck && !StaticsBlockAt(map, x, y, landCenter, testTop))
{
newZ = landCenter;
moveIsOk = true;
}
return moveIsOk;
}
/// <summary>
/// Mirrors the static-tile portion of IsOk: returns true if any static tile at (x,y)
/// has ImpassableSurface and overlaps the vertical range (ourZ, testTop).
/// </summary>
private static bool StaticsBlockAt(Map map, int x, int y, int ourZ, int testTop)
{
foreach (var check in map.Tiles.GetStaticAndMultiTiles(x, y))
{
var itemData = TileData.ItemTable[check.ID & TileData.MaxItemValue];
if (itemData.ImpassableSurface)
{
var checkZ = check.Z;
var checkTop = checkZ + itemData.CalcHeight;
if (checkTop > ourZ && testTop > checkZ)
{
return true;
}
}
}
return false;
}
}