fix(gta): restore wiped GTA foundation and gameplay path (PPSA04264) (#650)

* fix(kernel): implement APR ResolveFilepathsWithPrefixToIdsAndFileSizes

Resource streamers resolve relative paths against a shared prefix; without
this HLE every call returned NOT_FOUND and assets never got real ids/sizes.

* fix(remoteplay): stub Initialize and GetConnectionStatus as disconnected

Titles probe Remote Play during pad/network bring-up; unresolved imports
returned NOT_FOUND. Report initialized + disconnected so callers take the
normal offline path.

* fix(agc): accept Gen5 hull shaders that omit PGM_LO/HI in CreateShader

Type-5 headers can start with RSRC1/RSRC2; rejecting them left null handles
and Main Thread AVs. Scan the SH table and skip PGM patch when absent.

Co-authored-by: Cursor <cursoragent@cursor.com>

* fix(kernel): reject getdents on file fds and emit . / .. for empty dirs

Returning rax=0 for non-directory or empty listings looked like EOF and
let GTA treat the fd as a pointer (fiWriteAsyncDataWorker AV at 0xB1).

* fix(hle): enable GuestImageWriteTracker CPU sync on Windows

Windows previously hard-disabled the tracker, so CPU-written guest
planes never marked dirty and host textures stayed empty. Arm pages
with VirtualProtect, handle write AVs in VEH, and warm/test on
VirtualAlloc memory so protect cannot poison the CRT heap.

* fix(agc): skip CB metadata draws for EliminateFastClear/Fmask/DCC

CB_COLOR_CONTROL modes 2/5/6 are colour-buffer metadata ops; applying
the bound shader as a normal colour draw corrupts subsequent composites.
Decode MODE from bits [6:4] and return before translate.

* fix(agc): merge Prospero attrib-table formats onto IR vertex inputs

IR-discovered BufferLoadFormat often keeps a stale float sharp format;
patch DataFormat/offset from the AGC attrib table (semantic index),
allow offen fetches, and map quirks 113/121 through NarrowVk for host
vertex input.

* fix(audio): harden AudioOut2 stack out-buffer writes against canary smash

Titles that stack-allocate AudioOut2 outs next to the frame canary were
corrupted by oversized or mistyped HLE writes; keep ContextPush pacing.

* Revert "fix(memory): reserve only large regions (#608)"

This reverts commit 8f9456229a.

* fix(gpu): decode Gen5 R16 and RG32 render-target formats

* fix(audio): AudioOut2 host beds, deeper waveOut queue, AJM MP3

GTA V Enhanced routes intro/menu audio through AudioOut2 and FMOD's AJM
MP3 path. Wire PortCreate/PortSetAttributes/ContextPush to dual host
stereo streams, deepen WinMM queue to 128KiB, and decode AJM codec 0
with a stateful NLayer helper so menu music is not silent.

* fix(agc): map PS interpolants via SPI_PS_INPUT_CNTL semantics

Identity ATTR→param wiring ignored hardware remapping, so UI draws
got wrong (or empty) interpolants. Pack CNTL from matched PS/GS
semantics, thread it into Vulkan/Metal as Location/Flat, and fingerprint
it in the graphics shader cache key.

* fix(agc): rect-list/NGG strips, Index8 expand, and GE_INDX_OFFSET

NGG single-rect UI needs triangle-strip expansion; Prospero Index8 must
expand to host u16; glyphs need base vertex from GE_INDX_OFFSET. Skip
param-less rect-lists instead of inventing colour draws.

* fix(np): report GTA Story Mode addcont entitlements as owned

NpEntitlementAccess was returning an empty add-on list, so GTA V Enhanced offered Buy Story Mode. Publish the installed license labels and stub premium-event registration so offline sessions take the owned path.

* fix(cpu): prefer native workers for all guest entry stubs

Route thread entry, continuation, and main entry through RunGuestEntryStub so guest stubs are not invoked above CLR-managed frames (UnmanagedCallersOnly FailFast). Keep requireNativeWorker for tbb_thead; other paths prefer workers with calli fallback.

* fix(agc): implement Rewind/Jump writers and IT_REWIND waits

GTA Subrender AVs came from AcbJumpGetSize / DcbRewind returning NOT_FOUND as packet sizes. Add IT_REWIND and INDIRECT_BUFFER writers, patch SetRewindState into the GPU wait registry, and nest-parse 4-dword jumps.

* fix(gpu): use AddrLib ExactXor for Gen5 Standard256B (mode 1)

Mode 5 already had Standard4K ExactXor; mode 1 still used the generic StandardSwizzle block table, which mis-detiles Gen5 UI atlases.

* Revert "fix(cpu): prefer native workers for all guest entry stubs"

This reverts commit 31c4db0d38.

* fix(memory): commit-first large maps; reserve only on failure

Replace the #608 always-reserve-only exact-map path with allocate-first and lazy reserve fallback when a huge non-exec commit cannot be satisfied. Prime and widen GetPointer commit so the fallback path is safer for native walkers. Drops the need for a hard #608 revert.

* [Agc] Implement fused shader half exports

* fix(agc): accept optional hull state in CreatePrimState

Port the CreatePrimState hull-optional path from #583 so fused HS pipelines (GTA) are not rejected with INVALID_ARGUMENT. Geometry-derived CX/UC writes are unchanged; hull is traced only.

* fix(videoout): restore thread-safe VulkanHostBufferPool (#564)

The 6db095e wipe dropped CasualcoderDev's lock-ordering-safe pool. Concurrent Return/TryTake without the gate races after the first present and can hang the submit path.

* Revert "fix(agc): implement Rewind/Jump writers and IT_REWIND waits"

This reverts commit bec77bf083.

* test(memory): align lazy-commit expectations with commit-first policy

Fake hosts must reject Allocate so reserve-only paths still run, and GetPointer asserts the 32 MiB prime range including AlignUp spill.

* diag(gpu): log guest-queue backlog breakdown under backpressure

Rate-limit top work types and ordered debugName prefixes when the Vulkan guest work queue stalls, so North Yankton logs show acquire/label vs draw traffic instead of only VulkanOrderedGuestAction.

* perf(agc): coalesce acquire flushes and batch non-DMA label wakes

Flush pending ACQUIRE_MEM invalidation at draw/dispatch/dma/flip boundaries instead of before every packet, and complete release/write-data producers in the same ordered action so load paths enqueue far fewer VulkanOrderedGuestAction items.

* perf(gpu): wait for ordered-action fences and keep draining sync

On Windows/Linux, block briefly for queue-visibility fences instead of deferring the whole logical queue for the tick. Prefer ordered sync/flip heads under backlog pressure, and keep macOS non-blocking defer behavior.

* perf(gpu): raise sync-item ceiling above payload guest-work cap

Apply SHARPEMU_PENDING_GUEST_WORK_ITEMS mainly to compute/draw/image payload work, and allow a higher SHARPEMU_PENDING_GUEST_SYNC_ITEMS ceiling for zero-payload ordered actions and flip markers. Keep the byte budget as the RAM safety valve.

* fix(gta): stub Voice ports and implement sceKernelCheckReachability

Resolve North Yankton-path Voice Create/Delete/Connect/Disconnect/End NIDs and EnumerationThread reachability checks so leftover unresolved imports are not on the critical path.

* diag(gta): arm flip/present/wait probes after North Audio

Rate-limited load_progress TRACE for flip submit, ordered flip enqueue, present taken/not-taken, and GPU wait backlog so North Yankton freezes can be classified without full AGC tracing.

* fix(ampr): restore sequential offset=-1 reads for streamer packs

Re-wire PakDirectoryTracker into sceAmprAprCommandBufferReadFile (dropped in #216) so RAGE sequential pack reads no longer fail while the North Yankton UI keeps flipping. Also rate-limit CheckReachability miss paths for EnumerationThread diagnosis.

* fix(hle/videoout): Windows GuestImage opt-in and keep GTA intro without sync

Default the tracker off on Windows to avoid VirtualProtect thrash, gate AGC
texel-copy skips on Enabled so guest Bink planes keep shipping pixels, and
drain CPU-written images on the present thread when sync is opted in.

* fix(videoout): probe guest content when tracker off so UI can skip copies

Restores upload-known/texture-cache skips for Dead Cells menus, and uses a
sparse guest-memory fingerprint when GuestImageWriteTracker is disabled so
CPU-updated Bink planes still force texel copies for GTA intro.

* fix(audio): keep 128KiB host queue AudioOut2-only

Restore the default 32 KiB (~171 ms) PCM bed for classic AudioOut so
titles like Dreaming Sarah stay in sync; only AudioOut2 opens the deeper
queue needed for bursty FMOD Push on GTA.

---------

Co-authored-by: samto6 <123419830+samto6@users.noreply.github.com>
This commit is contained in:
MarcelMediaDev
2026-07-26 23:58:55 +01:00
committed by GitHub
parent 0535783f46
commit db4339f698
58 changed files with 7301 additions and 642 deletions
+2 -1
View File
@@ -89,7 +89,8 @@ internal sealed record GuestVertexBuffer(
uint OffsetBytes,
byte[] Data,
int Length,
bool Pooled);
bool Pooled,
bool PerInstance = false);
internal sealed record GuestIndexBuffer(
byte[] Data,
+11 -2
View File
@@ -54,6 +54,7 @@ internal interface IGuestGpuBackend
int scalarRegisterBufferIndex = -1,
uint pixelInputEnable = 0,
uint pixelInputAddress = 0,
IReadOnlyList<uint>? pixelInputCntl = null,
ulong storageBufferOffsetAlignment = 1);
bool TryCompileComputeShader(
@@ -108,7 +109,8 @@ internal interface IGuestGpuBackend
GuestIndexBuffer? indexBuffer = null,
IReadOnlyList<GuestVertexBuffer>? vertexBuffers = null,
GuestRenderState? renderState = null,
ulong shaderAddress = 0);
ulong shaderAddress = 0,
int baseVertex = 0);
void SubmitOffscreenTranslatedDraw(
IGuestCompiledShader pixelShader,
@@ -124,7 +126,8 @@ internal interface IGuestGpuBackend
IReadOnlyList<GuestVertexBuffer>? vertexBuffers = null,
GuestRenderState? renderState = null,
GuestDepthTarget? depthTarget = null,
ulong shaderAddress = 0);
ulong shaderAddress = 0,
int baseVertex = 0);
void SubmitStorageTranslatedDraw(
IGuestCompiledShader pixelShader,
@@ -236,6 +239,12 @@ internal interface IGuestGpuBackend
void SubmitGuestImageWrite(ulong address, byte[] pixels);
/// <summary>
/// Asks the presenter to refresh CPU-dirty guest images on its render/present
/// drain. Must not enqueue retained plane copies on the producer path.
/// </summary>
void RequestCpuWrittenGuestImageSync(ulong scopeAddress = 0, ulong scopeByteCount = ulong.MaxValue);
bool TryGetGuestImageExtent(ulong address, out uint width, out uint height, out ulong byteCount);
IReadOnlyList<(ulong Address, uint Width, uint Height, ulong ByteCount)> GetGuestImageExtents();
@@ -218,6 +218,13 @@ internal static class MetalGuestFormats
{
var format = (dataFormat, numberType) switch
{
// Early G-buffer / scene targets (R16 + RG32). Keep in sync with
// VulkanVideoPresenter.TryDecodeRenderTargetFormat.
(2, 0) => MtlPixelFormat.R16Unorm,
(2, 1) => MtlPixelFormat.R16Snorm,
(2, 4) => MtlPixelFormat.R16Uint,
(2, 5) => MtlPixelFormat.R16Sint,
(2, 7) => MtlPixelFormat.R16Float,
(4, 4) => MtlPixelFormat.R32Uint,
(4, 5) => MtlPixelFormat.R32Sint,
(4, 7) => MtlPixelFormat.R32Float,
@@ -230,6 +237,8 @@ internal static class MetalGuestFormats
(10, 5) => MtlPixelFormat.Rgba8Sint,
(10, 9) => MtlPixelFormat.Rgba8UnormSrgb,
(10, _) => MtlPixelFormat.Rgba8Unorm,
(11, 4) => MtlPixelFormat.Rg32Uint,
(11, 5) => MtlPixelFormat.Rg32Sint,
(11, 7) => MtlPixelFormat.Rg32Float,
(12, 4) => MtlPixelFormat.Rgba16Uint,
(12, 5) => MtlPixelFormat.Rgba16Sint,
@@ -258,10 +267,12 @@ internal static class MetalGuestFormats
var outputKind = format switch
{
MtlPixelFormat.R8Uint or MtlPixelFormat.R32Uint or MtlPixelFormat.Rg16Uint or
MtlPixelFormat.Rgba8Uint or MtlPixelFormat.Rgba16Uint => Gen5PixelOutputKind.Uint,
MtlPixelFormat.R32Sint or MtlPixelFormat.Rg16Sint or MtlPixelFormat.Rgba8Sint or
MtlPixelFormat.Rgba16Sint => Gen5PixelOutputKind.Sint,
MtlPixelFormat.R8Uint or MtlPixelFormat.R16Uint or MtlPixelFormat.R32Uint or
MtlPixelFormat.Rg16Uint or MtlPixelFormat.Rg32Uint or MtlPixelFormat.Rgba8Uint or
MtlPixelFormat.Rgba16Uint => Gen5PixelOutputKind.Uint,
MtlPixelFormat.R16Sint or MtlPixelFormat.R32Sint or MtlPixelFormat.Rg16Sint or
MtlPixelFormat.Rg32Sint or MtlPixelFormat.Rgba8Sint or MtlPixelFormat.Rgba16Sint =>
Gen5PixelOutputKind.Sint,
_ => Gen5PixelOutputKind.Float,
};
result = new MetalRenderTargetFormat(format, outputKind);
@@ -70,6 +70,7 @@ internal sealed class MetalGuestGpuBackend : IGuestGpuBackend
int scalarRegisterBufferIndex = -1,
uint pixelInputEnable = 0,
uint pixelInputAddress = 0,
IReadOnlyList<uint>? pixelInputCntl = null,
ulong storageBufferOffsetAlignment = 1)
{
shader = null;
@@ -85,6 +86,7 @@ internal sealed class MetalGuestGpuBackend : IGuestGpuBackend
scalarRegisterBufferIndex,
pixelInputEnable,
pixelInputAddress,
pixelInputCntl,
storageBufferOffsetAlignment))
{
return false;
@@ -251,7 +253,8 @@ internal sealed class MetalGuestGpuBackend : IGuestGpuBackend
GuestIndexBuffer? indexBuffer = null,
IReadOnlyList<GuestVertexBuffer>? vertexBuffers = null,
GuestRenderState? renderState = null,
ulong shaderAddress = 0) =>
ulong shaderAddress = 0,
int baseVertex = 0) =>
MetalVideoPresenter.SubmitDepthOnlyTranslatedDraw(
Msl(pixelShader),
textures,
@@ -265,7 +268,8 @@ internal sealed class MetalGuestGpuBackend : IGuestGpuBackend
indexBuffer,
vertexBuffers,
renderState,
shaderAddress);
shaderAddress,
baseVertex);
public void SubmitOffscreenTranslatedDraw(
IGuestCompiledShader pixelShader,
@@ -281,7 +285,8 @@ internal sealed class MetalGuestGpuBackend : IGuestGpuBackend
IReadOnlyList<GuestVertexBuffer>? vertexBuffers = null,
GuestRenderState? renderState = null,
GuestDepthTarget? depthTarget = null,
ulong shaderAddress = 0) =>
ulong shaderAddress = 0,
int baseVertex = 0) =>
MetalVideoPresenter.SubmitOffscreenTranslatedDraw(
Msl(pixelShader),
textures,
@@ -296,7 +301,8 @@ internal sealed class MetalGuestGpuBackend : IGuestGpuBackend
vertexBuffers,
renderState,
depthTarget,
shaderAddress);
shaderAddress,
baseVertex);
public void SubmitStorageTranslatedDraw(
IGuestCompiledShader pixelShader,
@@ -395,6 +401,9 @@ internal sealed class MetalGuestGpuBackend : IGuestGpuBackend
public void SubmitGuestImageWrite(ulong address, byte[] pixels) =>
MetalVideoPresenter.SubmitGuestImageWrite(address, pixels);
public void RequestCpuWrittenGuestImageSync(ulong scopeAddress = 0, ulong scopeByteCount = ulong.MaxValue) =>
MetalVideoPresenter.RequestCpuWrittenGuestImageSync(scopeAddress, scopeByteCount);
public bool TryGetGuestImageExtent(ulong address, out uint width, out uint height, out ulong byteCount) =>
MetalVideoPresenter.TryGetGuestImageExtent(address, out width, out height, out byteCount);
@@ -1,6 +1,7 @@
// Copyright (C) 2026 SharpEmu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
using System.Buffers.Binary;
using SharpEmu.Libs.Agc;
using SharpEmu.ShaderCompiler;
using SharpEmu.ShaderCompiler.Metal;
@@ -87,7 +88,8 @@ internal static partial class MetalVideoPresenter
uint InstanceCount,
uint PrimitiveType,
GuestIndexBuffer? IndexBuffer,
GuestRenderState RenderState);
GuestRenderState RenderState,
int BaseVertex = 0);
private sealed record OffscreenGuestDraw(
TranslatedGuestDraw Draw,
@@ -245,7 +247,8 @@ internal static partial class MetalVideoPresenter
IReadOnlyList<GuestVertexBuffer>? vertexBuffers,
GuestRenderState? renderState,
GuestDepthTarget? depthTarget,
ulong shaderAddress)
ulong shaderAddress,
int baseVertex = 0)
{
if (targets.Count == 0)
{
@@ -293,7 +296,8 @@ internal static partial class MetalVideoPresenter
instanceCount,
primitiveType,
indexBuffer,
effectiveRenderState),
effectiveRenderState,
baseVertex),
ToArray(targets),
depthTarget,
PublishTarget: true,
@@ -321,7 +325,8 @@ internal static partial class MetalVideoPresenter
GuestIndexBuffer? indexBuffer,
IReadOnlyList<GuestVertexBuffer>? vertexBuffers,
GuestRenderState? renderState,
ulong shaderAddress)
ulong shaderAddress,
int baseVertex = 0)
{
if (depthTarget.Address == 0 || depthTarget.Width == 0 || depthTarget.Height == 0)
{
@@ -348,7 +353,8 @@ internal static partial class MetalVideoPresenter
instanceCount,
primitiveType,
indexBuffer,
renderState ?? GuestRenderState.Default),
renderState ?? GuestRenderState.Default,
baseVertex),
[new GuestRenderTarget(Address: 0, depthTarget.Width, depthTarget.Height, Format: 10, NumberType: 0)],
depthTarget,
PublishTarget: false,
@@ -981,17 +987,38 @@ internal static partial class MetalVideoPresenter
private static void EncodeDrawCall(nint encoder, TranslatedGuestDraw draw)
{
var primitive = GetPrimitiveType(draw.PrimitiveType);
var vertexCount = draw.PrimitiveType == 0x11 && draw.IndexBuffer is null
? 4u
: draw.VertexCount;
var indexed = draw.IndexBuffer is not null;
var hasVertexBuffers = draw.VertexBuffers.Length > 0;
var primitive = GetPrimitiveType(
draw.PrimitiveType,
indexed,
draw.VertexCount,
hasVertexBuffers);
var vertexCount = AgcPrimitiveHelpers.GetRectListDrawVertexCount(
draw.PrimitiveType,
draw.VertexCount,
indexed,
hasVertexBuffers);
var baseVertex = (nuint)Math.Max(draw.BaseVertex, 0);
if (draw.IndexBuffer is { } indexBuffer)
{
var device = MetalNative.Send(encoder, MetalNative.Selector("device"));
var slice = AllocateUpload(
device, Math.Max(indexBuffer.Length, 1), out var buffer, out var offset);
indexBuffer.Data.AsSpan(0, Math.Min(indexBuffer.Length, indexBuffer.Data.Length))
.CopyTo(slice);
var source = indexBuffer.Data.AsSpan(
0,
Math.Min(indexBuffer.Length, indexBuffer.Data.Length));
// Metal drawIndexed without baseVertex: bake GE_INDX_OFFSET into
// the uploaded indices so glyph batches still hit the right verts.
if (draw.BaseVertex != 0)
{
BakeBaseVertexIntoIndices(source, slice, indexBuffer.Is32Bit, draw.BaseVertex);
}
else
{
source.CopyTo(slice);
}
MetalNative.SendDrawIndexedPrimitives(
encoder,
MetalNative.Selector("drawIndexedPrimitives:indexCount:indexType:indexBuffer:indexBufferOffset:instanceCount:"),
@@ -1012,12 +1039,46 @@ internal static partial class MetalVideoPresenter
encoder,
MetalNative.Selector("drawPrimitives:vertexStart:vertexCount:instanceCount:"),
primitive,
0,
baseVertex,
vertexCount,
Math.Max(draw.InstanceCount, 1));
}
}
private static void BakeBaseVertexIntoIndices(
ReadOnlySpan<byte> source,
Span<byte> destination,
bool is32Bit,
int baseVertex)
{
if (is32Bit)
{
var count = source.Length / sizeof(uint);
for (var index = 0; index < count; index++)
{
var value = BinaryPrimitives.ReadUInt32LittleEndian(
source.Slice(index * sizeof(uint), sizeof(uint)));
var adjusted = unchecked((uint)(value + baseVertex));
BinaryPrimitives.WriteUInt32LittleEndian(
destination.Slice(index * sizeof(uint), sizeof(uint)),
adjusted);
}
return;
}
var shortCount = source.Length / sizeof(ushort);
for (var index = 0; index < shortCount; index++)
{
var value = BinaryPrimitives.ReadUInt16LittleEndian(
source.Slice(index * sizeof(ushort), sizeof(ushort)));
var adjusted = unchecked((ushort)(value + baseVertex));
BinaryPrimitives.WriteUInt16LittleEndian(
destination.Slice(index * sizeof(ushort), sizeof(ushort)),
adjusted);
}
}
private static bool TryGetDrawPipeline(
nint device,
TranslatedGuestDraw draw,
@@ -1226,8 +1287,11 @@ internal static partial class MetalVideoPresenter
? vertexBuffer.Stride
: Math.Max(vertexBuffer.ComponentCount, 1) * 4;
MetalNative.Send(layout, MetalNative.Selector("setStride:"), (nint)stride);
// MTLVertexStepFunction.PerVertex = 1.
MetalNative.Send(layout, MetalNative.Selector("setStepFunction:"), 1);
// MTLVertexStepFunction: PerVertex = 1, PerInstance = 2.
MetalNative.Send(
layout,
MetalNative.Selector("setStepFunction:"),
vertexBuffer.PerInstance ? 2 : 1);
}
return descriptor;
@@ -2059,13 +2123,30 @@ internal static partial class MetalVideoPresenter
return 3;
case 6:
case 0x11:
return 4;
default:
return 3;
}
}
private static nuint GetPrimitiveType(
uint guestPrimitiveType,
bool indexed,
uint vertexCount,
bool hasVertexBuffers)
{
if (AgcPrimitiveHelpers.ShouldDrawRectListAsTriangleStrip(
guestPrimitiveType,
indexed,
vertexCount,
hasVertexBuffers))
{
return 4; // MTLPrimitiveTypeTriangleStrip
}
return GetPrimitiveType(guestPrimitiveType);
}
private static bool IsIntegerFormat(Gen5PixelOutputKind kind) =>
kind is Gen5PixelOutputKind.Uint or Gen5PixelOutputKind.Sint;
@@ -128,6 +128,7 @@ internal static partial class MetalVideoPresenter
private static readonly Dictionary<long, GuestImage> _guestImageVersions = new();
private static readonly Dictionary<(int Handle, int BufferIndex), long>
_lastOrderedGuestFlipVersions = new();
private static readonly Dictionary<ulong, ulong> _untrackedGuestImageContentProbes = new();
private static long _orderedGuestFlipVersionSequence;
private static volatile ICpuMemory? _guestMemory;
@@ -162,6 +163,22 @@ internal static partial class MetalVideoPresenter
public static void AttachGuestMemory(ICpuMemory memory) =>
_guestMemory = memory;
private static int _cpuWrittenGuestImageSyncRequested;
private static long _guestImageCpuSyncTraceCount;
public static void RequestCpuWrittenGuestImageSync(
ulong scopeAddress = 0,
ulong scopeByteCount = ulong.MaxValue)
{
_ = scopeAddress;
if (scopeByteCount == 0 || !GuestImageWriteTracker.Enabled)
{
return;
}
Volatile.Write(ref _cpuWrittenGuestImageSyncRequested, 1);
}
public static long SubmitOrderedGuestAction(Action action, string debugName)
{
ArgumentNullException.ThrowIfNull(action);
@@ -253,8 +270,110 @@ internal static partial class MetalVideoPresenter
}
}
public static bool IsGuestImageUploadKnown(ulong address, uint format, uint numberType) =>
IsGuestImageAvailable(address, format, numberType);
public static bool IsGuestImageUploadKnown(ulong address, uint format, uint numberType)
{
var guestFormat = GetGuestTextureFormat(format, numberType);
if (address == 0 || guestFormat == 0)
{
return false;
}
ulong probeByteCount = 0;
lock (_gate)
{
if (!_availableGuestImages.TryGetValue(address, out var availableFormat) ||
availableFormat != guestFormat)
{
return false;
}
if (GuestImageWriteTracker.Enabled)
{
return true;
}
if (_guestImageExtents.TryGetValue(address, out var extent))
{
probeByteCount = extent.ByteCount;
}
}
return IsUntrackedGuestImageContentUnchanged(address, probeByteCount);
}
private static bool IsUntrackedGuestImageContentUnchanged(ulong address, ulong byteCount)
{
var memory = _guestMemory;
if (memory is null || byteCount == 0)
{
return true;
}
var probe = ComputeSparseGuestContentProbe(memory, address, byteCount);
lock (_gate)
{
if (!_untrackedGuestImageContentProbes.TryGetValue(address, out var previous))
{
_untrackedGuestImageContentProbes[address] = probe;
return true;
}
if (previous == probe)
{
return true;
}
_untrackedGuestImageContentProbes[address] = probe;
return false;
}
}
private static ulong ComputeSparseGuestContentProbe(
ICpuMemory memory,
ulong address,
ulong byteCount)
{
Span<byte> sample = stackalloc byte[64];
ulong hash = 14695981039346656037UL;
Span<ulong> offsets = stackalloc ulong[3];
var offsetCount = 0;
offsets[offsetCount++] = 0;
if (byteCount > 128)
{
offsets[offsetCount++] = byteCount / 2;
}
if (byteCount > 64)
{
offsets[offsetCount++] = byteCount - 64;
}
for (var o = 0; o < offsetCount; o++)
{
var offset = offsets[o];
if (offset >= byteCount)
{
continue;
}
var length = (int)Math.Min(64UL, byteCount - offset);
if (!memory.TryRead(address + offset, sample[..length]))
{
hash ^= 0x9E3779B97F4A7C15UL + offset;
continue;
}
for (var i = 0; i < length; i++)
{
hash ^= sample[i];
hash *= 1099511628211UL;
}
hash ^= (ulong)length + offset;
}
return hash ^ byteCount;
}
public static bool GuestImageWantsInitialData(ulong address)
{
@@ -599,7 +718,7 @@ internal static partial class MetalVideoPresenter
var completedWork = 0;
RecycleCompletedUploadPages();
RecycleCompletedSnapshotResources();
EvictDirtyCachedDrawTextures();
DrainGuestImageCpuSync(device);
try
{
while (completedWork < MaxGuestWorkPerRender)
@@ -45,12 +45,11 @@ internal static partial class MetalVideoPresenter
texture.Pitch,
texture.Sampler);
/// <summary>Caching requires the write tracker: without page protection a
/// guest CPU write would never evict the entry and draws would sample
/// stale texels forever. Storage textures are shader-writable on the GPU,
/// so their content identity is not stable either.</summary>
/// <summary>Storage textures are shader-writable on the GPU, so their
/// content identity is not stable. CPU rewrites of protected/CPU-backed
/// images still evict via DrainGuestImageCpuSync when those addresses
/// are dirty.</summary>
private static bool IsCacheableDrawTexture(GuestDrawTexture texture) =>
GuestImageWriteTracker.Enabled &&
texture.Address != 0 &&
!texture.IsStorage &&
!texture.IsFallback;
@@ -69,20 +68,100 @@ internal static partial class MetalVideoPresenter
_ = MetalNative.Send(handle, MetalNative.Selector("retain"));
_drawTextureCache[key] = handle;
_cachedDrawTextureIdentities[key] = 0;
GuestImageWriteTracker.Track(
texture.Address,
(ulong)texture.RgbaPixels.Length,
Volatile.Read(ref _executingGuestWorkSequence),
"metal.texture-cache");
// No GuestImageWriteTracker.Track: watch-only cache registrations
// widened the managed-write hot path. CPU-backed / protected images
// own dirty notifications used for eviction.
}
/// <summary>Runs once per drain, before any queued draw executes: a draw
/// whose texels the submit thread skipped must never resolve to an entry
/// the guest has since rewritten.</summary>
private static void EvictDirtyCachedDrawTextures()
/// <summary>
/// Single dirty consumer per drain: re-upload CPU-written guest images,
/// evict matching draw-texture cache entries, then re-arm once per address.
/// </summary>
private static void DrainGuestImageCpuSync(nint device)
{
if (!GuestImageWriteTracker.Enabled)
{
return;
}
_ = Interlocked.Exchange(ref _cpuWrittenGuestImageSyncRequested, 0);
HashSet<ulong>? dirtyAddresses = null;
List<(ulong Address, uint Width, uint Height, ulong ByteCount)>? extents = null;
lock (_gate)
{
if (_guestImageExtents.Count > 0)
{
extents = new(_guestImageExtents.Count);
foreach (var entry in _guestImageExtents)
{
extents.Add((
entry.Key,
entry.Value.Width,
entry.Value.Height,
entry.Value.ByteCount));
}
}
}
var memory = _guestMemory;
if (extents is not null)
{
foreach (var (address, width, height, byteCount) in extents)
{
if (!GuestImageWriteTracker.ConsumeDirty(address))
{
continue;
}
(dirtyAddresses ??= []).Add(address);
if (memory is null ||
byteCount == 0 ||
byteCount > 128UL * 1024UL * 1024UL)
{
continue;
}
GuestImage? image;
lock (_gate)
{
_guestImages.TryGetValue(address, out image);
}
if (image is null)
{
continue;
}
var pixels = new byte[byteCount];
if (!memory.TryRead(address, pixels) ||
pixels.AsSpan().IndexOfAnyExcept((byte)0) < 0)
{
continue;
}
ExecuteGuestImageWrite(
device,
queue: 0,
new GuestImageWrite(address, pixels, 0));
if (Interlocked.Increment(ref _guestImageCpuSyncTraceCount) <= 64)
{
Console.Error.WriteLine(
$"[SYNC] cpu-write-drain addr=0x{address:X} {width}x{height}");
}
}
}
if (_drawTextureCache.Count == 0)
{
if (dirtyAddresses is not null)
{
foreach (var address in dirtyAddresses)
{
GuestImageWriteTracker.Rearm(address);
}
}
return;
}
@@ -90,17 +169,17 @@ internal static partial class MetalVideoPresenter
// share one source address (same texels, different samplers), and
// ConsumeDirty clears the flag on first read — evicting only the
// first identity would leave the others sampling stale texels.
HashSet<ulong>? dirtyAddresses = null;
foreach (var entry in _drawTextureCache)
{
if (dirtyAddresses is not null && dirtyAddresses.Contains(entry.Key.Address))
var address = entry.Key.Address;
if (dirtyAddresses is not null && dirtyAddresses.Contains(address))
{
continue;
}
if (GuestImageWriteTracker.ConsumeDirty(entry.Key.Address))
if (GuestImageWriteTracker.ConsumeDirty(address))
{
(dirtyAddresses ??= []).Add(entry.Key.Address);
(dirtyAddresses ??= []).Add(address);
}
}
@@ -118,6 +197,14 @@ internal static partial class MetalVideoPresenter
_drawTextureCache.Clear();
_cachedDrawTextureIdentities.Clear();
if (dirtyAddresses is not null)
{
foreach (var address in dirtyAddresses)
{
GuestImageWriteTracker.Rearm(address);
}
}
return;
}
@@ -64,6 +64,7 @@ internal sealed class VulkanGuestGpuBackend : IGuestGpuBackend
int scalarRegisterBufferIndex = -1,
uint pixelInputEnable = 0,
uint pixelInputAddress = 0,
IReadOnlyList<uint>? pixelInputCntl = null,
ulong storageBufferOffsetAlignment = 1)
{
shader = null;
@@ -79,6 +80,7 @@ internal sealed class VulkanGuestGpuBackend : IGuestGpuBackend
scalarRegisterBufferIndex,
pixelInputEnable,
pixelInputAddress,
pixelInputCntl,
storageBufferOffsetAlignment))
{
return false;
@@ -179,7 +181,8 @@ internal sealed class VulkanGuestGpuBackend : IGuestGpuBackend
GuestIndexBuffer? indexBuffer = null,
IReadOnlyList<GuestVertexBuffer>? vertexBuffers = null,
GuestRenderState? renderState = null,
ulong shaderAddress = 0) =>
ulong shaderAddress = 0,
int baseVertex = 0) =>
VulkanVideoPresenter.SubmitDepthOnlyTranslatedDraw(
Spirv(pixelShader),
textures,
@@ -193,7 +196,8 @@ internal sealed class VulkanGuestGpuBackend : IGuestGpuBackend
indexBuffer,
vertexBuffers,
renderState,
shaderAddress);
shaderAddress,
baseVertex);
public void SubmitOffscreenTranslatedDraw(
IGuestCompiledShader pixelShader,
@@ -209,7 +213,8 @@ internal sealed class VulkanGuestGpuBackend : IGuestGpuBackend
IReadOnlyList<GuestVertexBuffer>? vertexBuffers = null,
GuestRenderState? renderState = null,
GuestDepthTarget? depthTarget = null,
ulong shaderAddress = 0) =>
ulong shaderAddress = 0,
int baseVertex = 0) =>
VulkanVideoPresenter.SubmitOffscreenTranslatedDraw(
Spirv(pixelShader),
textures,
@@ -224,7 +229,8 @@ internal sealed class VulkanGuestGpuBackend : IGuestGpuBackend
vertexBuffers,
renderState,
depthTarget,
shaderAddress);
shaderAddress,
baseVertex);
public void SubmitStorageTranslatedDraw(
IGuestCompiledShader pixelShader,
@@ -375,6 +381,9 @@ internal sealed class VulkanGuestGpuBackend : IGuestGpuBackend
public void SubmitGuestImageWrite(ulong address, byte[] pixels) =>
VulkanVideoPresenter.SubmitGuestImageWrite(address, pixels);
public void RequestCpuWrittenGuestImageSync(ulong scopeAddress = 0, ulong scopeByteCount = ulong.MaxValue) =>
VulkanVideoPresenter.RequestCpuWrittenGuestImageSync(scopeAddress, scopeByteCount);
public bool TryGetGuestImageExtent(ulong address, out uint width, out uint height, out ulong byteCount) =>
VulkanVideoPresenter.TryGetGuestImageExtent(address, out width, out height, out byteCount);