acdream/tests/AcDream.App.Tests/Rendering/RetailPViewPassExecutorTests.cs
Erik 5de42a12f3 fix(render): S4 chunk 1 round 1 — the ±12 guard with retail's quantifier, seal count order, the PM/PC depth-event transcript gate
Three-lens review of c7ab5b6d8 found the round-0 port of
DrawPortalPolyInternal's degenerate-input guard inverted retail's
quantifier: it rejected a polygon when ANY vertex sat on ANY of the
four +/-12 planes. The PDB-paired bytes (VA 0x59BCD6-0x59BD66: four
per-plane fld/fcomp 12.0/fnstsw/test ah,0x44/jnp sequences, each
accumulating a "every vertex on this plane" boolean, cleared by the
first vertex NOT on it) and the named decomp's own nested-if reading
of the post-loop branch (0x59bd42-0x59bd6c) agree: retail rejects only
when EVERY local vertex lies on the SAME one of the four planes — a
polygon degenerate onto a plane, not one that merely touches a plane
at one vertex.

F1 — the quantifier. WalkVisibilityMath.IsRejectedByPortalPolygonBoundaryGuard
now tracks four booleans (everyVertexOnPlusX/MinusX/PlusY/MinusY), each
starting true and cleared by the first vertex that fails its own plane
test; the guard rejects iff any of the four survives the whole vertex
loop. T2 (WalkVisibilityMathTests.cs) is rewritten to the five §7
cases: every vertex on x=+12 -> rejected; every vertex on y=-12 ->
rejected; one vertex on x=+12, rest inside -> KEPT; every vertex at
11.999 -> kept; vertices split across x=+12 and y=+12 (no common
plane) -> kept. WalkFrameDriverTests' punch-side pin
(OnPunchGeometry_RejectsOnlyWhenEveryVertexSharesOnePlane_ButPunchesAnyOtherShape)
carries the same correction — its old "one vertex on the plane"
polygon was wrongly rejected under round 0; it punches now, and a new
all-on-plane polygon is added as the actual rejected case. The doc
comment's "essentially never" sentence is deleted and replaced with a
fresh DAT-scan citation (see datScan below).
MUTATION (verified): restore the any-vertex form -> the one-vertex and
split-plane admit cases fail (Assert.False Expected:False Actual:True,
x2; the WalkFrameDriverTests punch pin's Assert.Equal Expected:2
Actual:1). Delete the guard (always return false) -> the all-on-plane
reject cases fail (Assert.True Expected:True Actual:False, x3 —
plusX-plane, minusY-plane, and the vacuous-empty-polygon pin; the
punch pin's Assert.Equal Expected:2 Actual:3).

datScan (DAT counts backing F1's doc comment): a throwaway xunit test
(deleted after this commit — not part of the diff) walked the full
0x0000-0xFFFF landblock-prefix space (LandBlockInfo/EnvCell carry no
usable DatCollection.GetAllIdsOfType range — both are landblock-
relative — matching the method CellStructSurfaceConstructionInstalledDatTests
already established), resolved each EnvCell's Environment/CellStruct,
and for every CellPortal's polygon tested each vertex's local x/y
against +/-12. Against the installed client_cell_1.dat: 3,405
landblocks with cells, 1,854,237 total portal polygons. 2,889 polygons
have at least one vertex on a +/-12 plane; 2,163 have EVERY vertex on
the SAME plane — all 2,163 are EXIT portals (OtherCellId==0xFFFF, out
of 16,939 exit portals total); 0 interior (non-exit) portals ever
qualify (interior: 1,837,298 total, 340 any-vertex, 0 all-on-plane).
Duration ~15s. This exactly reproduces the review's cited numbers.

F2 — seal-path count order. RetailPViewPassExecutor's exit-seal loop
dropped the `localVertices.Length < 3` pre-filter that ran BEFORE both
the boundary guard and the `submitted` increment (round 0's order:
length check -> guard -> transform -> draw -> count). Retail's real
order is guard -> transform -> clip -> count, with the counter
incremented BEFORE polyClipFinish runs (0x59BD70-0x59BD74 precedes
0x59BDB0) — the counter records accepted ATTEMPTS, not successful GPU
fans. `submitted++` now runs immediately after the guard-gated
transform+EmitSeamMask, before DrawDepthFan (whose own `< 3` guard is
this port's stand-in for retail's post-clip `var_4 >= 3` check). A new
pin (RetailPViewPassExecutorTests.DrawExitPortalMask_CountsAnUnclippableTwoVertexPolygon_ButDrawsNothing)
constructs a real PortalDepthMaskRenderer (RecordingGpuDevice-backed)
and a constructor-bypassed executor (RuntimeHelpers.GetUninitializedObject,
the same pattern WalkOutsideViewReassemblyTests already uses) with a
2-vertex ordinary-coordinate portal polygon: DrawExitPortalMask returns
submitted=1 while zero draw calls reach the device.
MUTATION (verified): restore the `< 3` pre-filter ahead of the guard
-> the pin's Assert.Equal fails, Expected:1 Actual:0.

F3 — the depth-event transcript gate, automated. New
WalkAlphaDepthTrace.cs (beside WalkOracleTrace.cs) parses the four
`PM poly=<ptr> mode=<0|1> counterBefore=<hex>` / `PC ov=<n>
counter=<hex> fc=<0|1>` lines per oh-capture/*.alphadepth.log frame,
using the same F-marker frame delimiting as WalkOracleTrace. New
WalkTraceConformanceTests.AlphaDepthTranscript.cs (beside
WalkTraceConformanceTests.cs, which is now `partial`) replays each
pose's frame 2 through the real WalkFrameDriver/RetailFrameWalk over
WalkLandscapeDatBuilder-built installed-DAT world data, with:
  - AlphaDepthCollectSink: an IWalkEventSink decorator forwarding every
    hook to the real driver unchanged, while recording mode=1 punch
    events (OnPunchGeometry, counterBefore = driver.PortalsDrawnCount
    at COLLECT time — Collect finishes entirely before Replay starts,
    so reading the counter at Replay time would already see the SAME
    turn's post-reset value; holtburg-doorway-still's own capture
    proves this: its punches show counterBefore=2 immediately followed
    by its seals' counterBefore=0/1 in the identical turn) and every
    PC tuple (Emit, WalkEventKind.DrawCells — fires for the root's own
    turn AND every building look-in's re-entrant call, matching
    retail's breakpoint not distinguishing them).
  - AlphaDepthLeaf: an IWalkFrameLeafRenderer whose DrawExitSeals
    reproduces DrawPortalDepthWrite's real per-portal loop (F1's guard,
    F2's count-before-clip order) directly over
    driver.InteriorFloodCells / WalkCell.Portals / .PortalPolygons —
    the identical data the render-side production path reads — since
    the interface gives DrawExitSeals no cell/portal parameter at all.
  - Two Collect+Replay passes per pose (a fresh driver reaches the SAME
    steady-state a running session already carries by the time
    retail's capture began — holtburg-doorway-still and foundry-deep
    both show a stable per-frame accepted-seal count that regenerates
    identically every captured frame): the first pass's own output is
    discarded, only the second is compared.
Five Facts (one per pose — the spec's "four" undercounts the five
*.alphadepth.log files actually present in oh-capture/):
  AlphaDepthTranscript_CathedralArrival_MatchesRetailFrame2 — PASS
  AlphaDepthTranscript_CathedralLeak_MatchesRetailFrame2 — PASS
  AlphaDepthTranscript_FoundryDeep_MatchesRetailFrame2 — PASS
  AlphaDepthTranscript_HoltburgDoorwayStill_MatchesRetailFrame2 — PASS
    (proves the incrementing 0/1/2 accepted-seal sequence AND the
    pre-reset counterBefore=2 punches, end to end, against real DAT
    portal data)
  AlphaDepthTranscript_TerraceEdge_MatchesRetailFrame2 — KnownFailure
    (tagged [Trait("Status","KnownFailure")]): terrace-edge never runs
    a qualifying interior turn in its own capture (every PC line reads
    ov=0) — its punches' counterBefore=2 is a session value carried
    over from BEFORE the capture even started (the file's own pre-"F 1"
    content already reads 2; nothing inside the captured frames ever
    touches the counter). A fresh two-pass replay from
    PortalsDrawnCount=0 has no mechanism to derive that leftover value.
    Printed divergence: expected [(mode=1,counterBefore=2) x2], actual
    [(mode=1,counterBefore=0) x2]; the PC sequence (ov=0 both times)
    matches exactly. An initial-condition gap in the fixture, not a
    guard/count defect — the identical harness reproduces holtburg-
    doorway-still's and foundry-deep's own steady states exactly from
    cold starts.
MUTATION (verified): swap AlphaDepthLeaf's recorded seal mode from 0
to 1 -> AlphaDepthTranscript_HoltburgDoorwayStill_MatchesRetailFrame2
fails (Assert.Equal Collections differ at pos 5: expected
(0,0),(0,1)... actual (1,0),(1,1)...).
DispatcherFixture and DrawScope (WalkFrameDriverTests.cs) are widened
from private to internal so the new gate can drive a real
WbDrawDispatcher without duplicating that fixture.

F4 — manifest comment truth. VulkanShaderManifestTests.cs's
portal_depth.vert.spv re-pin comment had the ULP direction backward
(claimed the OLD 0x3F7FFFFE literal was NEARER the camera than
retail's real 0x3F7FFFEF constant; portal_depth.vert's own shader-
source comment already had this correct — FARTHER, larger z/w) and
cited T1 as living in WalkVisibilityMathTests when
PortalDepthVert_FarPunchConstant_MatchesRetailExactBits is in this
same file. Both corrected.

Gates: dotnet build tests/AcDream.App.Tests -c Release -> 0 warnings/0
errors. Hermetic lane (Lane!=InstalledDat&...&Status!=KnownFailure):
6829/6829 passed. InstalledDat lane (real client_cell_1.dat): 253
total, 247 passed, 5 failed — exactly the two #383 layout tests
(LayoutImporterMediaBearingChildSweepTests/LayoutImporterInvisibleSweepTests),
TowerAscentReplayTests' KnownFailure, Oh_doorway_still_first_frame_diff
(#458), and this round's own AlphaDepthTranscript_TerraceEdge
KnownFailure — no other divergence. Shader tests (VulkanShaderDescriptorContractTests
+ VulkanShaderManifestTests + RenderPackSpirvValidatorTests +
SkyVertexLayoutTests): 35/35. Register/divergence tests: 52/52.
dotnet build src/AcDream.App -c Release: 0 warnings/0 errors.

Register: no new AD row. F1/F2 are straight bug fixes toward retail
fidelity (not deviations); AD-119 (the C2 blend-vs-write-mask row)
does not describe the count order, so its neighbour needed no
correction per F2's own instruction.

Deviation from the fix-round spec's literal text: §7 F1 said "swap the
any-vertex hits instead of the four per-plane accumulators" as the
split-plane case's mutation — the code actually ported uses per-vertex
`if (x != 12f) flag = false;` clears, not a per-vertex OR; the mutation
verified above (restore the any-vertex FORM entirely) exercises the
same defect class and is the mutation actually run. §7 also names
"four" alphadepth.log fixtures where five exist on disk
(cathedral-arrival, cathedral-leak, foundry-deep,
holtburg-doorway-still, terrace-edge); all five are gated rather than
four, with cathedral-leak an unremarked addition alongside the four
the spec names.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-03 23:38:52 +02:00

617 lines
31 KiB
C#

using System.Linq;
using System.Numerics;
using System.Reflection;
using System.Reflection.Emit;
using AcDream.App.Composition;
using AcDream.App.Rendering;
using AcDream.App.Rendering.Gpu;
using AcDream.App.Rendering.Gpu.Vk;
using AcDream.App.Rendering.Sky;
using AcDream.App.Rendering.Walk;
using AcDream.App.Tests.Architecture;
using AcDream.App.Tests.Rendering.Gpu;
namespace AcDream.App.Tests.Rendering;
public sealed class RetailPViewPassExecutorTests
{
[Fact]
public void Extracted_contracts_retain_no_window_callbacks_or_visibility_owner()
{
Assert.DoesNotContain(
typeof(RetailPViewFrameInput).GetProperties(),
property => typeof(Delegate).IsAssignableFrom(property.PropertyType));
FieldInfo[] fields = typeof(RetailPViewPassExecutor).GetFields(
BindingFlags.Instance | BindingFlags.NonPublic);
Assert.DoesNotContain(fields, field => field.FieldType == typeof(GameWindow));
Assert.DoesNotContain(fields, field => field.FieldType == typeof(CellVisibility));
Assert.DoesNotContain(fields, field => field.FieldType == typeof(RetailPViewFrameInput));
Assert.DoesNotContain(fields, field => field.FieldType == typeof(RetailPViewFrameResult));
Assert.DoesNotContain(fields, field => field.FieldType == typeof(ClipFrameAssembly));
Assert.DoesNotContain(
fields,
field => typeof(Delegate).IsAssignableFrom(field.FieldType));
}
[Fact]
public void Concrete_executor_accumulates_walk_terrain_batch_timing()
{
// S3 chunk 3 fix round 1 (F3): the walk leaf no longer brackets
// itself with Begin()/Complete() (that stopwatch-restart pair would
// push one timing SAMPLE per batch, not one per frame) — it times
// itself with a raw Stopwatch.GetTimestamp() delta and hands the
// elapsed ticks to AccumulateWalkBatch, which only accumulates.
MethodInfo landscape = typeof(RetailPViewPassExecutor).GetMethod(
"DrawWalkLandCellBatch",
BindingFlags.Instance | BindingFlags.NonPublic)!;
IReadOnlyList<CompiledCall> landscapeCalls = CompiledCallGraph.Read(landscape);
int terrainDraw = RequiredCallIndex(
landscapeCalls,
typeof(TerrainModernRenderer),
nameof(TerrainModernRenderer.DrawLandCells));
int accumulate = RequiredCallIndex(
landscapeCalls,
typeof(TerrainDrawDiagnosticsController),
nameof(TerrainDrawDiagnosticsController.AccumulateWalkBatch));
Assert.True(terrainDraw < accumulate);
Assert.DoesNotContain(
landscapeCalls,
call => call.Target.DeclaringType == typeof(TerrainDrawDiagnosticsController)
&& call.Target.Name == nameof(TerrainDrawDiagnosticsController.Begin));
Assert.DoesNotContain(
landscapeCalls,
call => call.Target.DeclaringType == typeof(TerrainDrawDiagnosticsController)
&& call.Target.Name == nameof(TerrainDrawDiagnosticsController.Complete));
}
[Fact]
public void Concrete_executor_pushes_the_walk_terrain_frame_sample_at_replay_end()
{
// S3 chunk 3 fix round 1 (F3): DrawWalkDrivenStatics is the ONE call
// site of driver.Replay in production — CompleteWalkTerrainFrame
// must run immediately after it, so the frame's sample is pushed
// exactly once, at "the end of the walk replay".
MethodInfo drawWalkDrivenStatics = typeof(RetailPViewRenderer).GetMethod(
"DrawWalkDrivenStatics",
BindingFlags.Instance | BindingFlags.NonPublic)!;
IReadOnlyList<CompiledCall> calls = CompiledCallGraph.Read(drawWalkDrivenStatics);
int replay = RequiredCallIndex(
calls,
typeof(AcDream.App.Rendering.Walk.WalkFrameDriver),
nameof(AcDream.App.Rendering.Walk.WalkFrameDriver.Replay));
int completeWalkFrame = RequiredCallIndex(
calls,
typeof(RetailPViewPassExecutor),
nameof(RetailPViewPassExecutor.CompleteWalkTerrainFrame));
Assert.True(replay < completeWalkFrame);
}
[Fact]
public void Frame_composition_constructs_one_walk_executor()
{
MethodInfo compose = typeof(FrameRootCompositionPhase).GetMethod(
"ComposeCore",
BindingFlags.Instance | BindingFlags.NonPublic)!;
IReadOnlyList<CompiledCall> calls = CompiledCallGraph.Read(compose);
int executor = RequiredCallIndex(
calls,
typeof(RetailPViewPassExecutor),
".ctor");
int renderer = RequiredCallIndex(
calls,
typeof(WorldScenePViewRenderer),
".ctor");
Assert.True(executor < renderer);
Assert.Single(
calls,
call => call.Target.DeclaringType == typeof(RetailPViewPassExecutor)
&& call.Target.Name == ".ctor");
Assert.Single(
calls,
call => call.Target.DeclaringType == typeof(WorldScenePViewRenderer)
&& call.Target.Name == ".ctor");
}
/// <summary>
/// S3 chunk 1 fix round 2 (§11.6 H1): the weather MESH draw + its OC
/// print moved OUT of <see cref="RetailPViewPassExecutor.DrawWeatherOnce"/>
/// entirely — S3 chunk 4 (O3) relocated the print to
/// <c>WalkFrameDriver.OnWeatherTurn</c>, fired at Collect time by
/// <c>RetailFrameWalk.DrawLandscape</c> (see the real transcript pins in
/// <c>WalkFrameDriverTranscriptTests</c>: <c>Collect_OutdoorRoot_...</c>
/// and <c>Collect_InteriorRoot_...</c>). This method now draws the
/// weather MESH and the rain PARTICLES only — the former per-outside-
/// view-slice loop that used to run before this call (the walk's own
/// screen-space terrain-clip writer + its per-frame clip-routing reset
/// call + the old <c>DrawLandscapeSliceLate</c> leaf) is deleted outright (§10.2): retail
/// draws the weather mesh and its
/// rain particles ONCE, unclipped, never once per doorway aperture.
/// MUTATION: re-inlining a
/// <c>WalkTranscriptDump.PrintObjectCellTurn</c> call back into this
/// method makes the <c>Assert.DoesNotContain</c> below fail; deleting
/// either the mesh or the particle call makes the matching
/// <c>Assert.Single</c> fail (zero matches instead of one).
/// </summary>
[Fact]
public void DrawWeatherOnce_DrawsTheWeatherMeshAndParticlesButNeverPrints()
{
MethodInfo method = typeof(RetailPViewPassExecutor).GetMethod(
nameof(RetailPViewPassExecutor.DrawWeatherOnce),
BindingFlags.Instance | BindingFlags.Public)!;
IReadOnlyList<CompiledCall> calls = CompiledCallGraph.Read(method);
Assert.Single(
calls,
call => call.Target.DeclaringType == typeof(SkyRenderer)
&& call.Target.Name == nameof(SkyRenderer.RenderWeather));
Assert.Single(
calls,
call => call.Target.DeclaringType == typeof(ParticleRenderer)
&& call.Target.Name == nameof(ParticleRenderer.Draw));
Assert.DoesNotContain(
calls,
call => call.Target.DeclaringType == typeof(WalkTranscriptDump));
}
/// <summary>
/// S3 chunk 4 (§10.2): the former per-outside-view-slice loop
/// (the walk's own screen-space terrain-clip writer + its per-frame
/// clip-routing reset call + the old <c>DrawLandscapeSliceLate</c> leaf,
/// one call per active landscape view) is deleted — <c>DrawLandscapeDynamicsPhase</c> now calls
/// <see cref="RetailPViewPassExecutor.DrawWeatherOnce"/> exactly once,
/// conditional on <see cref="AcDream.App.Rendering.Walk.WalkFrameDriver.WeatherTurnFired"/>
/// (see <see cref="DrawLandscapeDynamicsPhase_GatesDrawWeatherOnceOnWalkDriverWeatherTurnFired"/>
/// — the L1 pin), with no loop of any kind around it. This
/// <c>Assert.Single</c> alone proved insufficient at fix round 1 (K1):
/// it counts DISTINCT call-site offsets, so it stays green even with a
/// <c>foreach</c> wrapped around the one call site (the exact round-1
/// regression this file's review caught) — see
/// <see cref="DrawLandscapeDynamicsPhase_DrawWeatherOnceCallSiteHasNoEnclosingBackwardBranch"/>
/// for the pin that actually rules that out. Kept as a cheap first-line
/// check: MUTATION: adding a second, textually distinct call site (e.g.
/// a duplicated call, not a loop) makes <c>Assert.Single</c> fail.
/// </summary>
[Fact]
public void DrawLandscapeDynamicsPhase_CallsDrawWeatherOnceExactlyOnce()
{
MethodInfo method = typeof(RetailPViewRenderer).GetMethod(
"DrawLandscapeDynamicsPhase",
BindingFlags.Instance | BindingFlags.NonPublic)!;
IReadOnlyList<CompiledCall> calls = CompiledCallGraph.Read(method);
Assert.Single(
calls,
call => call.Target.DeclaringType == typeof(RetailPViewPassExecutor)
&& call.Target.Name == nameof(RetailPViewPassExecutor.DrawWeatherOnce));
}
/// <summary>
/// S3 chunk 4 fix round 1 (K1, blocking): the real loop-shape pin.
/// <see cref="DrawLandscapeDynamicsPhase_CallsDrawWeatherOnceExactlyOnce"/>'s
/// <c>Assert.Single</c> over call-site offsets still passes when the ONE
/// call site sits inside a <c>foreach</c> — an IL-offset ORDER pin has
/// now failed three times to be discriminating for this exact class of
/// regression, so this asks the LOOP-SHAPE question directly: does any
/// BACKWARD branch (a branch whose target offset is lower than its own
/// offset — the shape every C# loop compiles to, whether
/// <c>for</c>/<c>foreach</c>/<c>while</c>) enclose the
/// <c>DrawWeatherOnce</c> call's own IL offset? A call sitting strictly
/// between a backward branch's target and its own offset is inside that
/// loop's body and can run more than once per method invocation; a call
/// outside every backward branch's span cannot. MUTATION: wrap the call
/// in <c>foreach (var slice in clipAssembly.OutsideViewSlices)</c> —
/// the compiled <c>foreach</c> emits a backward branch (the
/// condition-check jump back to the loop body) whose span now contains
/// the call's offset, so this test fails; restore the single
/// unconditional call to make it pass again. Note for reviewers (added
/// at S3 landing hygiene, H4, matching <see cref="DrawWalkSky_RenderSkyCallSiteHasNoEnclosingBackwardBranch"/>'s
/// own note): <see cref="CompiledCallGraph.ReadBranches"/> reads only
/// <c>br</c>/<c>brtrue</c>/<c>brfalse</c>-family single-target branches —
/// it does not decode a compiled <c>switch</c> jump table, but no C# loop
/// construct (<c>for</c>/<c>foreach</c>/<c>while</c>/<c>do</c>) ever
/// compiles to one, so this pin's blind spot is not a loop shape it
/// could miss.
/// </summary>
[Fact]
public void DrawLandscapeDynamicsPhase_DrawWeatherOnceCallSiteHasNoEnclosingBackwardBranch()
{
MethodInfo method = typeof(RetailPViewRenderer).GetMethod(
"DrawLandscapeDynamicsPhase",
BindingFlags.Instance | BindingFlags.NonPublic)!;
IReadOnlyList<CompiledCall> calls = CompiledCallGraph.Read(method);
int callIndex = RequiredCallIndex(
calls,
typeof(RetailPViewPassExecutor),
nameof(RetailPViewPassExecutor.DrawWeatherOnce));
int callOffset = calls[callIndex].Offset;
IReadOnlyList<CompiledBranch> branches = CompiledCallGraph.ReadBranches(method);
Assert.DoesNotContain(
branches,
branch => branch.TargetOffset < branch.Offset
&& branch.TargetOffset <= callOffset
&& callOffset < branch.Offset);
}
/// <summary>
/// S3 chunk 4 fix round 2 (L1, BLOCKING). Round 1's three-lens review
/// found that neither existing pin above actually looks at the
/// production CONDITION gating <c>DrawWeatherOnce</c>: restoring the
/// pre-fix gate <c>if (clipAssembly.OutsideViewSlices.Length != 0)</c> —
/// the exact regression K2 was supposed to close — leaves both green,
/// because both only ask "is the call site shaped correctly", never
/// "does the call site read <c>WalkFrameDriver.WeatherTurnFired</c>".
/// This pin reads the compiled condition directly: with <c>c</c> the
/// index of the <c>DrawWeatherOnce</c> call, (a) <c>calls[c-1]</c> must
/// be the <c>WeatherTurnFired</c> getter — the LAST call before the draw
/// — and (b) exactly one branch must sit strictly between that getter
/// call and the draw call, be a <c>brfalse</c>/<c>brfalse.s</c>, and
/// jump FORWARD past the draw call — the compiled shape of
/// <c>if (walkDriver.WeatherTurnFired) passes.DrawWeatherOnce(ctx);</c>
/// and nothing else (an inverted test, an unconditional call, or a
/// different condition entirely all fail one of the two checks).
/// MUTATION M1 (restores the pre-fix regression): change the gate back
/// to <c>if (clipAssembly.OutsideViewSlices.Length != 0)</c> — check (a)
/// fails because <c>calls[c-1]</c> is no longer the
/// <c>WeatherTurnFired</c> getter. MUTATION M2 (drops the gate
/// entirely): make the call unconditional — check (b) fails because no
/// branch sits between the getter call and the draw call (in fact the
/// getter call itself disappears with the gate, so check (a) fails
/// first). MUTATION M3 (inverts the condition): change the gate to
/// <c>if (!walkDriver.WeatherTurnFired)</c> — <c>calls[c-1]</c> is still
/// the getter (check (a) passes), but the compiler emits a
/// <c>brtrue</c>/<c>brtrue.s</c> to skip the draw instead of a
/// <c>brfalse</c>/<c>brfalse.s</c>, so check (b)'s opcode filter finds
/// nothing and <c>Assert.Single</c> fails on zero matches.
/// </summary>
[Fact]
public void DrawLandscapeDynamicsPhase_GatesDrawWeatherOnceOnWalkDriverWeatherTurnFired()
{
MethodInfo method = typeof(RetailPViewRenderer).GetMethod(
"DrawLandscapeDynamicsPhase",
BindingFlags.Instance | BindingFlags.NonPublic)!;
IReadOnlyList<CompiledCall> calls = CompiledCallGraph.Read(method);
int callIndex = RequiredCallIndex(
calls,
typeof(RetailPViewPassExecutor),
nameof(RetailPViewPassExecutor.DrawWeatherOnce));
Assert.True(callIndex > 0, "Expected a call before DrawWeatherOnce — the gate condition.");
CompiledCall condition = calls[callIndex - 1];
Assert.Equal(typeof(AcDream.App.Rendering.Walk.WalkFrameDriver), condition.Target.DeclaringType);
Assert.Equal("get_WeatherTurnFired", condition.Target.Name);
int conditionOffset = condition.Offset;
int drawOffset = calls[callIndex].Offset;
IReadOnlyList<CompiledBranch> branches = CompiledCallGraph.ReadBranches(method);
Assert.Single(
branches,
branch => branch.Offset > conditionOffset
&& branch.Offset < drawOffset
&& (branch.OpCode == OpCodes.Brfalse || branch.OpCode == OpCodes.Brfalse_S)
&& branch.TargetOffset > drawOffset);
}
/// <summary>
/// S3 landing hygiene (H5): strengthens the L1 pin above against a
/// conjoined gate the post-hoc three-lens review found it does not
/// reject. L1's check (b) only looks at branches STRICTLY BETWEEN the
/// <c>WeatherTurnFired</c> getter call and the <c>DrawWeatherOnce</c>
/// call — so <c>if (clipAssembly.OutsideViewSlices.Length != 0 &amp;&amp;
/// walkDriver.WeatherTurnFired) passes.DrawWeatherOnce(ctx);</c> still
/// passes it: the compiler evaluates <c>OutsideViewSlices.Length != 0</c>
/// FIRST, so ITS OWN <c>brfalse</c> lands BEFORE the getter call's
/// offset — outside L1's window — while <c>calls[c-1]</c> is still the
/// getter (the added condition reads <c>OutsideViewSlices</c>, a
/// property getter, then <c>.Length</c>, a non-call <c>ldlen</c>, so no
/// OTHER call intervenes before the getter). This pin widens the window
/// to start at the call immediately before the whole gate —
/// <see cref="RetailPViewPassExecutor.DrawUnattachedSceneParticles"/>,
/// the outdoor-emitters call the method's own comment names as the last
/// call before the gate — and counts EVERY branch in that wider window
/// with <c>Assert.Single</c>: production has exactly one, the forward
/// <c>brfalse</c>/<c>brfalse.s</c> right after the getter. A conjoined
/// gate's extra, earlier condition adds a second branch this wider
/// window catches but L1's narrower one cannot.
/// MUTATION: change the gate to <c>if
/// (clipAssembly.OutsideViewSlices.Length != 0 &amp;&amp;
/// walkDriver.WeatherTurnFired)</c> — the branch count in the window
/// goes from 1 to 2 and <c>Assert.Single</c> fails (see the commit body
/// for the exact recorded failure text).
/// </summary>
[Fact]
public void DrawLandscapeDynamicsPhase_ExactlyOneBranchGuardsDrawWeatherOnce()
{
MethodInfo method = typeof(RetailPViewRenderer).GetMethod(
"DrawLandscapeDynamicsPhase",
BindingFlags.Instance | BindingFlags.NonPublic)!;
IReadOnlyList<CompiledCall> calls = CompiledCallGraph.Read(method);
int particlesIndex = RequiredCallIndex(
calls,
typeof(RetailPViewPassExecutor),
nameof(RetailPViewPassExecutor.DrawUnattachedSceneParticles));
int drawIndex = RequiredCallIndex(
calls,
typeof(RetailPViewPassExecutor),
nameof(RetailPViewPassExecutor.DrawWeatherOnce));
int particlesOffset = calls[particlesIndex].Offset;
int drawOffset = calls[drawIndex].Offset;
IReadOnlyList<CompiledBranch> branches = CompiledCallGraph.ReadBranches(method);
CompiledBranch onlyGuard = Assert.Single(
branches,
branch => branch.Offset > particlesOffset && branch.Offset < drawOffset);
Assert.True(
onlyGuard.OpCode == OpCodes.Brfalse || onlyGuard.OpCode == OpCodes.Brfalse_S,
$"Expected the sole branch between DrawUnattachedSceneParticles and "
+ $"DrawWeatherOnce to be a brfalse, was {onlyGuard.OpCode}.");
Assert.True(
onlyGuard.TargetOffset > drawOffset,
"Expected the guard branch to skip forward past DrawWeatherOnce.");
}
/// <summary>
/// S3 chunk 4 fix round 2 (L8): the identical loop-shape question K1
/// asked of <see cref="RetailPViewPassExecutor.DrawWeatherOnce"/>'s call
/// site, applied to <see cref="RetailPViewPassExecutor.DrawWalkSky"/>'s
/// own <see cref="SkyRenderer.RenderSky"/> call — retail draws the sky
/// dome exactly once per frame too (K4's own doc comment on
/// <c>DrawWalkSky</c>), so nothing may wrap this call in a loop either.
/// Note for reviewers: <see cref="CompiledCallGraph.ReadBranches"/> reads
/// only <c>br</c>/<c>brtrue</c>/<c>brfalse</c>-family single-target
/// branches — it does not decode a compiled <c>switch</c> jump table,
/// but no C# loop construct (<c>for</c>/<c>foreach</c>/<c>while</c>/
/// <c>do</c>) ever compiles to one, so this pin's blind spot is not a
/// loop shape it could miss. MUTATION: wrapping the call in
/// <c>for (int i = 0; i &lt; 2; i++) { _sky?.RenderSky(...); }</c> makes
/// this fail; restoring the single unconditional call makes it pass
/// again.
/// </summary>
[Fact]
public void DrawWalkSky_RenderSkyCallSiteHasNoEnclosingBackwardBranch()
{
MethodInfo method = typeof(RetailPViewPassExecutor).GetMethod(
"DrawWalkSky",
BindingFlags.Instance | BindingFlags.NonPublic)!;
IReadOnlyList<CompiledCall> calls = CompiledCallGraph.Read(method);
int callIndex = RequiredCallIndex(
calls,
typeof(SkyRenderer),
nameof(SkyRenderer.RenderSky));
int callOffset = calls[callIndex].Offset;
IReadOnlyList<CompiledBranch> branches = CompiledCallGraph.ReadBranches(method);
Assert.DoesNotContain(
branches,
branch => branch.TargetOffset < branch.Offset
&& branch.TargetOffset <= callOffset
&& callOffset < branch.Offset);
}
/// <summary>
/// S3 chunk 1 fix round 2 (§11.6 H1): <see
/// cref="RetailPViewPassExecutor.ShouldDrawWeatherOnce"/> is
/// <see cref="DrawWeatherOnce_DrawsTheWeatherMeshAndPrintsExactlyOnce"/>'s
/// gate, extracted as a pure predicate so this suite can pin "no OC line
/// while the player stands indoors" without a live GL/DAT
/// <see cref="SkyRenderer"/> — combined with the two structural tests
/// above (moved out of the loop; drawn/printed exactly once per call),
/// this proves both halves of the spec's pin: an outdoor root (or an
/// interior root with several exit-view slices) prints exactly one OC
/// line, and an indoor player prints none. Retail's own check:
/// <c>SmartBox::is_player_outside</c> @0x00451e80,
/// <c>(player objcell_id &amp; 0xFFFF) &lt; 0x100</c>. MUTATION: negating
/// the <c>&lt; 0x100</c> comparison (or dropping either bool AND) makes
/// one of the four rows below fail.
/// </summary>
[Theory]
[InlineData(true, true, 0xF4180003u, true)] // outdoor root, player outside a land cell -> draws
[InlineData(true, true, 0xA9B40100u, false)] // player indoors (local id >= 0x100) -> no draw
[InlineData(false, true, 0xF4180003u, false)] // RenderSky off -> no draw
[InlineData(true, false, 0xF4180003u, false)] // RenderWeather off -> no draw
public void ShouldDrawWeatherOnce_MatchesRetailIsPlayerOutsideGate(
bool renderSky, bool renderWeather, uint playerCellId, bool expected)
{
Assert.Equal(
expected,
RetailPViewPassExecutor.ShouldDrawWeatherOnce(renderSky, renderWeather, playerCellId));
}
/// <summary>
/// S4-c1 C1 (T2, seal half): <c>D3DPolyRender::DrawPortalPolyInternal</c>
/// @0x0059bc90's degenerate-input guard, ported at the exit-seal
/// enumeration (<c>DrawPortalDepthWrite</c> — the SAME loop that reads
/// <c>cell.PortalPolygons[index]</c>, the LOCAL portal-polygon
/// vertices, and is the only production caller of
/// <see cref="RetailPViewPassExecutor.DrawExitPortalMask"/>). A live
/// functional test of this private method needs a real
/// <see cref="PortalDepthMaskRenderer"/> the suite has no fake for (see
/// <c>WalkFrameDriverTests.OnPunchGeometry_RejectsWholePolygonOn...</c>
/// for the punch-fan half's functional proof instead), so this pin asks
/// the compiled-call-graph question this file's other tests already use
/// for exactly this situation: does
/// <see cref="AcDream.App.Rendering.Walk.WalkVisibilityMath.IsRejectedByPortalPolygonBoundaryGuard"/>
/// run BEFORE the vertex loop's <see cref="Vector3.Transform(Vector3,
/// Matrix4x4)"/> calls and BEFORE <see cref="PortalDepthMaskRenderer.DrawDepthFan"/>
/// (retail's reject -&gt; transform -&gt; clip -&gt; count order — a hit
/// must never reach either), with a conditional branch gating that
/// order directly off the guard's own return value.
/// MUTATION: move the guard call to AFTER the transform loop (or
/// delete it) — either check below fails because the guard call index
/// is no longer the smallest, or (deletion) <see cref="RequiredCallIndex"/>
/// throws for finding no call at all.
/// </summary>
[Fact]
public void DrawPortalDepthWrite_RejectsDegenerateLocalPolygons_BeforeTransformOrSubmission()
{
MethodInfo method = typeof(RetailPViewPassExecutor).GetMethod(
"DrawPortalDepthWrite",
BindingFlags.Instance | BindingFlags.NonPublic)!;
IReadOnlyList<CompiledCall> calls = CompiledCallGraph.Read(method);
int guardIndex = RequiredCallIndex(
calls,
typeof(AcDream.App.Rendering.Walk.WalkVisibilityMath),
nameof(AcDream.App.Rendering.Walk.WalkVisibilityMath.IsRejectedByPortalPolygonBoundaryGuard));
int transformIndex = RequiredCallIndex(calls, typeof(Vector3), nameof(Vector3.Transform));
int drawIndex = RequiredCallIndex(
calls, typeof(PortalDepthMaskRenderer), nameof(PortalDepthMaskRenderer.DrawDepthFan));
Assert.True(
guardIndex < transformIndex,
"The boundary guard must run BEFORE the world-transform loop "
+ "(retail's reject -> transform -> clip -> count order).");
Assert.True(
guardIndex < drawIndex,
"The boundary guard must run BEFORE the fan is submitted "
+ "(no draw, no `submitted` increment on a hit).");
int guardOffset = calls[guardIndex].Offset;
int transformOffset = calls[transformIndex].Offset;
IReadOnlyList<CompiledBranch> branches = CompiledCallGraph.ReadBranches(method);
Assert.Contains(
branches,
branch => branch.Offset > guardOffset
&& branch.Offset < transformOffset
&& (branch.OpCode == OpCodes.Brtrue || branch.OpCode == OpCodes.Brtrue_S
|| branch.OpCode == OpCodes.Brfalse || branch.OpCode == OpCodes.Brfalse_S));
}
/// <summary>
/// S4-c1 fix round 1, F2: retail increments <c>portalsDrawnCount</c>
/// (0x59BD70-0x59BD74) BEFORE <c>polyClipFinish</c> runs (0x59BDB0) —
/// the counter records accepted ATTEMPTS, not successful GPU fans
/// (<c>oh1-depth-lifecycle.md</c>'s "Far-Z punches and true-depth exit
/// seals" section). A polygon that survives the boundary guard but has
/// fewer than 3 vertices is still COUNTED by
/// <see cref="RetailPViewPassExecutor.DrawExitPortalMask"/>'s returned
/// <c>submitted</c> total, even though
/// <see cref="PortalDepthMaskRenderer.DrawDepthFan"/> draws nothing (its
/// own <c>&lt; 3</c> guard stands in for retail's post-clip
/// <c>var_4 &gt;= 3</c> check). Round 0 dropped the count too — a
/// <c>&lt; 3</c> continue ahead of both the boundary guard and the
/// count — never observed on authored dat data (every real portal
/// polygon has &gt;= 3 vertices) but the wrong order all the same.
/// MUTATION: move the increment back to AFTER
/// <see cref="PortalDepthMaskRenderer.DrawDepthFan"/> (or restore the
/// old <c>localVertices.Length &lt; 3</c> pre-filter) — this pin's
/// synthetic 2-vertex polygon is no longer counted and the assertion
/// fails (<c>submitted</c> comes back 0, not 1).
/// </summary>
[Fact]
public void DrawExitPortalMask_CountsAnUnclippableTwoVertexPolygon_ButDrawsNothing()
{
var cell = new LoadedCell
{
CellId = 0xA9B40105u,
WorldTransform = Matrix4x4.Identity,
};
cell.Portals.Add(new CellPortalInfo(OtherCellId: 0xFFFF, PolygonId: 0, Flags: 0, OtherPortalId: 0));
// Ordinary (non-degenerate) coordinates — the boundary guard admits
// this polygon — but only TWO vertices: retail's post-clip
// `var_4 >= 3` check (this port's DrawDepthFan `< 3` guard) drops
// the fan submission even though the count already happened.
cell.PortalPolygons.Add(
[
new Vector3(1f, 1f, 0f),
new Vector3(2f, 1f, 0f),
]);
using var device = new RecordingGpuDevice();
var frames = new GpuDeviceFrameLifetime(device);
var scope = new VulkanWorldPassScope(sampleCount: 1);
using var portalDepthMask = new PortalDepthMaskRenderer(device, frames, scope);
var diagnostics = new WorldRenderDiagnostics(new NeverCalledGlStateReader(), new NeverCalledDiagnosticLog());
// DrawPortalDepthWrite only reads _portalDepthMask, frame.Cells,
// frame.RootCell.IsOutdoorNode, frame.ViewProjection, and
// _diagnostics (short-circuited off by RenderingDiagnostics.
// ProbeSeamDrawEnabled's default-off value) — the same
// constructor-bypass pattern WalkOutsideViewReassemblyTests already
// uses for exercising a real leaf without a full GL/DAT renderer
// graph.
var executor = (RetailPViewPassExecutor)System.Runtime.CompilerServices.RuntimeHelpers
.GetUninitializedObject(typeof(RetailPViewPassExecutor));
typeof(RetailPViewPassExecutor)
.GetField("_portalDepthMask", BindingFlags.NonPublic | BindingFlags.Instance)!
.SetValue(executor, portalDepthMask);
typeof(RetailPViewPassExecutor)
.GetField("_diagnostics", BindingFlags.NonPublic | BindingFlags.Instance)!
.SetValue(executor, diagnostics);
var root = new LoadedCell { CellId = 0xF4180003u, IsOutdoorNode = false };
RetailPViewFrameInput frame = new RetailPViewFrameInput().Reset(
rootCell: root,
nearbyBuildingCells: null,
viewerEyePos: Vector3.Zero,
viewProjection: Matrix4x4.Identity,
cells: new SingleCellSource(cell),
camera: null!,
cameraWorldPosition: Vector3.Zero,
frustum: null,
playerLandblockId: null,
animatedEntityIds: null,
renderCenterLbX: 0,
renderCenterLbY: 0,
renderRadius: 0,
landblockEntries: Array.Empty<(uint, Vector3, Vector3,
IReadOnlyList<AcDream.Core.World.WorldEntity>,
IReadOnlyDictionary<uint, AcDream.Core.World.WorldEntity>?)>(),
renderSky: false,
renderWeather: false,
dayFraction: 0f,
activeDayGroup: null,
skyKeyframe: default,
environOverrideActive: false,
viewerCellId: 0,
playerCellId: 0,
playerViewPosition: Vector3.Zero,
cameraView: Matrix4x4.Identity,
cameraCellResolution: default);
int drawsBefore = device.Calls.OfType<GpuRecordedDraw>().Count();
int submitted = executor.DrawExitPortalMask(frame, cell.CellId, ReadOnlySpan<Vector4>.Empty);
int drawsAfter = device.Calls.OfType<GpuRecordedDraw>().Count();
Assert.Equal(1, submitted);
Assert.Equal(drawsBefore, drawsAfter);
}
private sealed class SingleCellSource(LoadedCell cell) : IRetailPViewCellSource
{
public LoadedCell? Find(uint cellId) => cellId == cell.CellId ? cell : null;
}
private sealed class NeverCalledGlStateReader : IRenderGlStateReader
{
public RenderGlStateSnapshot CaptureState() =>
throw new InvalidOperationException("EmitSeamMask must short-circuit before reading GL state.");
public RenderGlScissorSnapshot CaptureScissor() =>
throw new InvalidOperationException("EmitSeamMask must short-circuit before reading GL state.");
}
private sealed class NeverCalledDiagnosticLog : IRenderFrameDiagnosticLog
{
public void WriteLine(string message) =>
throw new InvalidOperationException("EmitSeamMask must short-circuit before logging (ProbeSeamDrawEnabled defaults off).");
}
private static int RequiredCallIndex(
IReadOnlyList<CompiledCall> calls,
Type declaringType,
string methodName)
{
int index = CompiledCallGraph.IndexOf(calls, declaringType, methodName);
Assert.True(
index >= 0,
$"Expected call to {declaringType.Name}.{methodName}.");
return index;
}
}