GPU: Pre-emptively flush textures that are flushed often (to imported memory when available) (#4711)

* WIP texture pre-flush

Improve performance of TextureView GetData to buffer

Fix copy/sync ordering

Fix minor bug

Make this actually work

WIP host mapping stuff

* Fix usage flags

* message

* Cleanup 1

* Fix rebase

* Fix

* Improve pre-flush rules

* Fix pre-flush

* A lot of cleanup

* Use the host memory bits

* Select the correct memory type

* Cleanup TextureGroupHandle

* Missing comment

* Remove debugging logs

* Revert BufferHandle _value access modifier

* One interrupt action at a time.

* Support D32S8 to D24S8 conversion, safeguards

* Interrupt cannot happen in sync handle's lock

Waitable needs to be checked twice now, but this should stop it from deadlocking.

* Remove unused using

* Address some feedback

* Address feedback

* Address more feedback

* Address more feedback

* Improve sync rules

Should allow for faster sync in some cases.

10 files changed, 127 insertions, 3 deletions

diff --git a/src/Ryujinx.Graphics.GAL/BufferAccess.cs b/src/Ryujinx.Graphics.GAL/BufferAccess.cs
new file mode 100644
index 00000000..3a2d6c9b
--- /dev/null
+++ b/src/Ryujinx.Graphics.GAL/BufferAccess.cs
@@ -0,0 +1,8 @@
+namespace Ryujinx.Graphics.GAL
+{
+    public enum BufferAccess
+    {
+        Default,
+        FlushPersistent
+    }
+}
diff --git a/src/Ryujinx.Graphics.GAL/IRenderer.cs b/src/Ryujinx.Graphics.GAL/IRenderer.cs
index 2af7b5db..d36dd26b 100644
--- a/src/Ryujinx.Graphics.GAL/IRenderer.cs
+++ b/src/Ryujinx.Graphics.GAL/IRenderer.cs
@@ -21,11 +21,14 @@ namespace Ryujinx.Graphics.GAL
         {
             return CreateBuffer(size, BufferHandle.Null);
         }
+        BufferHandle CreateBuffer(nint pointer, int size);
+        BufferHandle CreateBuffer(int size, BufferAccess access);
 
         IProgram CreateProgram(ShaderSource[] shaders, ShaderInfo info);
 
         ISampler CreateSampler(SamplerCreateInfo info);
         ITexture CreateTexture(TextureCreateInfo info, float scale);
+        bool PrepareHostMapping(nint address, ulong size);
 
         void CreateSync(ulong id, bool strict);
 
diff --git a/src/Ryujinx.Graphics.GAL/ITexture.cs b/src/Ryujinx.Graphics.GAL/ITexture.cs
index 792c863c..2f9f5fbf 100644
--- a/src/Ryujinx.Graphics.GAL/ITexture.cs
+++ b/src/Ryujinx.Graphics.GAL/ITexture.cs
@@ -12,6 +12,7 @@ namespace Ryujinx.Graphics.GAL
         void CopyTo(ITexture destination, int firstLayer, int firstLevel);
         void CopyTo(ITexture destination, int srcLayer, int dstLayer, int srcLevel, int dstLevel);
         void CopyTo(ITexture destination, Extents2D srcRegion, Extents2D dstRegion, bool linearFilter);
+        void CopyTo(BufferRange range, int layer, int level, int stride);
 
         ITexture CreateView(TextureCreateInfo info, int firstLayer, int firstLevel);
 
diff --git a/src/Ryujinx.Graphics.GAL/Multithreading/CommandHelper.cs b/src/Ryujinx.Graphics.GAL/Multithreading/CommandHelper.cs
index 063b7edf..9f6e483c 100644
--- a/src/Ryujinx.Graphics.GAL/Multithreading/CommandHelper.cs
+++ b/src/Ryujinx.Graphics.GAL/Multithreading/CommandHelper.cs
@@ -43,6 +43,8 @@ namespace Ryujinx.Graphics.GAL.Multithreading
 
             Register<ActionCommand>(CommandType.Action);
             Register<CreateBufferCommand>(CommandType.CreateBuffer);
+            Register<CreateBufferAccessCommand>(CommandType.CreateBufferAccess);
+            Register<CreateHostBufferCommand>(CommandType.CreateHostBuffer);
             Register<CreateProgramCommand>(CommandType.CreateProgram);
             Register<CreateSamplerCommand>(CommandType.CreateSampler);
             Register<CreateSyncCommand>(CommandType.CreateSync);
@@ -69,6 +71,7 @@ namespace Ryujinx.Graphics.GAL.Multithreading
             Register<TextureCopyToCommand>(CommandType.TextureCopyTo);
             Register<TextureCopyToScaledCommand>(CommandType.TextureCopyToScaled);
             Register<TextureCopyToSliceCommand>(CommandType.TextureCopyToSlice);
+            Register<TextureCopyToBufferCommand>(CommandType.TextureCopyToBuffer);
             Register<TextureCreateViewCommand>(CommandType.TextureCreateView);
             Register<TextureGetDataCommand>(CommandType.TextureGetData);
             Register<TextureGetDataSliceCommand>(CommandType.TextureGetDataSlice);
diff --git a/src/Ryujinx.Graphics.GAL/Multithreading/CommandType.cs b/src/Ryujinx.Graphics.GAL/Multithreading/CommandType.cs
index 61e729b4..8d9c1ec8 100644
--- a/src/Ryujinx.Graphics.GAL/Multithreading/CommandType.cs
+++ b/src/Ryujinx.Graphics.GAL/Multithreading/CommandType.cs
@@ -4,6 +4,8 @@
     {
         Action,
         CreateBuffer,
+        CreateBufferAccess,
+        CreateHostBuffer,
         CreateProgram,
         CreateSampler,
         CreateSync,
@@ -29,6 +31,7 @@
         SamplerDispose,
 
         TextureCopyTo,
+        TextureCopyToBuffer,
         TextureCopyToScaled,
         TextureCopyToSlice,
         TextureCreateView,
diff --git a/src/Ryujinx.Graphics.GAL/Multithreading/Commands/Renderer/CreateBufferAccessCommand.cs b/src/Ryujinx.Graphics.GAL/Multithreading/Commands/Renderer/CreateBufferAccessCommand.cs
new file mode 100644
index 00000000..ece98b70
--- /dev/null
+++ b/src/Ryujinx.Graphics.GAL/Multithreading/Commands/Renderer/CreateBufferAccessCommand.cs
@@ -0,0 +1,22 @@
+namespace Ryujinx.Graphics.GAL.Multithreading.Commands.Renderer
+{
+    struct CreateBufferAccessCommand : IGALCommand, IGALCommand<CreateBufferAccessCommand>
+    {
+        public CommandType CommandType => CommandType.CreateBufferAccess;
+        private BufferHandle _threadedHandle;
+        private int _size;
+        private BufferAccess _access;
+
+        public void Set(BufferHandle threadedHandle, int size, BufferAccess access)
+        {
+            _threadedHandle = threadedHandle;
+            _size = size;
+            _access = access;
+        }
+
+        public static void Run(ref CreateBufferAccessCommand command, ThreadedRenderer threaded, IRenderer renderer)
+        {
+            threaded.Buffers.AssignBuffer(command._threadedHandle, renderer.CreateBuffer(command._size, command._access));
+        }
+    }
+}
diff --git a/src/Ryujinx.Graphics.GAL/Multithreading/Commands/Renderer/CreateHostBufferCommand.cs b/src/Ryujinx.Graphics.GAL/Multithreading/Commands/Renderer/CreateHostBufferCommand.cs
new file mode 100644
index 00000000..e25f8468
--- /dev/null
+++ b/src/Ryujinx.Graphics.GAL/Multithreading/Commands/Renderer/CreateHostBufferCommand.cs
@@ -0,0 +1,22 @@
+namespace Ryujinx.Graphics.GAL.Multithreading.Commands.Renderer
+{
+    struct CreateHostBufferCommand : IGALCommand, IGALCommand<CreateHostBufferCommand>
+    {
+        public CommandType CommandType => CommandType.CreateHostBuffer;
+        private BufferHandle _threadedHandle;
+        private nint _pointer;
+        private int _size;
+
+        public void Set(BufferHandle threadedHandle, nint pointer, int size)
+        {
+            _threadedHandle = threadedHandle;
+            _pointer = pointer;
+            _size = size;
+        }
+
+        public static void Run(ref CreateHostBufferCommand command, ThreadedRenderer threaded, IRenderer renderer)
+        {
+            threaded.Buffers.AssignBuffer(command._threadedHandle, renderer.CreateBuffer(command._pointer, command._size));
+        }
+    }
+}
diff --git a/src/Ryujinx.Graphics.GAL/Multithreading/Commands/Texture/TextureCopyToBufferCommand.cs b/src/Ryujinx.Graphics.GAL/Multithreading/Commands/Texture/TextureCopyToBufferCommand.cs
new file mode 100644
index 00000000..ac0e07d6
--- /dev/null
+++ b/src/Ryujinx.Graphics.GAL/Multithreading/Commands/Texture/TextureCopyToBufferCommand.cs
@@ -0,0 +1,29 @@
+using Ryujinx.Graphics.GAL.Multithreading.Model;
+using Ryujinx.Graphics.GAL.Multithreading.Resources;
+
+namespace Ryujinx.Graphics.GAL.Multithreading.Commands.Texture
+{
+    struct TextureCopyToBufferCommand : IGALCommand, IGALCommand<TextureCopyToBufferCommand>
+    {
+        public CommandType CommandType => CommandType.TextureCopyToBuffer;
+        private TableRef<ThreadedTexture> _texture;
+        private BufferRange _range;
+        private int _layer;
+        private int _level;
+        private int _stride;
+
+        public void Set(TableRef<ThreadedTexture> texture, BufferRange range, int layer, int level, int stride)
+        {
+            _texture = texture;
+            _range = range;
+            _layer = layer;
+            _level = level;
+            _stride = stride;
+        }
+
+        public static void Run(ref TextureCopyToBufferCommand command, ThreadedRenderer threaded, IRenderer renderer)
+        {
+            command._texture.Get(threaded).Base.CopyTo(threaded.Buffers.MapBufferRange(command._range), command._layer, command._level, command._stride);
+        }
+    }
+}
diff --git a/src/Ryujinx.Graphics.GAL/Multithreading/Resources/ThreadedTexture.cs b/src/Ryujinx.Graphics.GAL/Multithreading/Resources/ThreadedTexture.cs
index ee1cfa29..bb0b05fb 100644
--- a/src/Ryujinx.Graphics.GAL/Multithreading/Resources/ThreadedTexture.cs
+++ b/src/Ryujinx.Graphics.GAL/Multithreading/Resources/ThreadedTexture.cs
@@ -108,6 +108,12 @@ namespace Ryujinx.Graphics.GAL.Multithreading.Resources
             }
         }
 
+        public void CopyTo(BufferRange range, int layer, int level, int stride)
+        {
+            _renderer.New<TextureCopyToBufferCommand>().Set(Ref(this), range, layer, level, stride);
+            _renderer.QueueCommand();
+        }
+
         public void SetData(SpanOrArray<byte> data)
         {
             _renderer.New<TextureSetDataCommand>().Set(Ref(this), Ref(data.ToArray()));
diff --git a/src/Ryujinx.Graphics.GAL/Multithreading/ThreadedRenderer.cs b/src/Ryujinx.Graphics.GAL/Multithreading/ThreadedRenderer.cs
index 2148f43f..3e179621 100644
--- a/src/Ryujinx.Graphics.GAL/Multithreading/ThreadedRenderer.cs
+++ b/src/Ryujinx.Graphics.GAL/Multithreading/ThreadedRenderer.cs
@@ -57,6 +57,7 @@ namespace Ryujinx.Graphics.GAL.Multithreading
         private int _refConsumerPtr;
 
         private Action _interruptAction;
+        private object _interruptLock = new();
 
         public event EventHandler<ScreenCaptureImageInfo> ScreenCaptured;
 
@@ -274,6 +275,24 @@ namespace Ryujinx.Graphics.GAL.Multithreading
             return handle;
         }
 
+        public BufferHandle CreateBuffer(nint pointer, int size)
+        {
+            BufferHandle handle = Buffers.CreateBufferHandle();
+            New<CreateHostBufferCommand>().Set(handle, pointer, size);
+            QueueCommand();
+
+            return handle;
+        }
+
+        public BufferHandle CreateBuffer(int size, BufferAccess access)
+        {
+            BufferHandle handle = Buffers.CreateBufferHandle();
+            New<CreateBufferAccessCommand>().Set(handle, size, access);
+            QueueCommand();
+
+            return handle;
+        }
+
         public IProgram CreateProgram(ShaderSource[] shaders, ShaderInfo info)
         {
             var program = new ThreadedProgram(this);
@@ -448,11 +467,14 @@ namespace Ryujinx.Graphics.GAL.Multithreading
             }
             else
             {
-                while (Interlocked.CompareExchange(ref _interruptAction, action, null) != null) { }
+                lock (_interruptLock)
+                {
+                    while (Interlocked.CompareExchange(ref _interruptAction, action, null) != null) { }
 
-                _galWorkAvailable.Set();
+                    _galWorkAvailable.Set();
 
-                _interruptRun.WaitOne();
+                    _interruptRun.WaitOne();
+                }
             }
         }
 
@@ -461,6 +483,11 @@ namespace Ryujinx.Graphics.GAL.Multithreading
             // Threaded renderer ignores given interrupt action, as it provides its own to the child renderer.
         }
 
+        public bool PrepareHostMapping(nint address, ulong size)
+        {
+            return _baseRenderer.PrepareHostMapping(address, size);
+        }
+
         public void Dispose()
         {
             // Dispose must happen from the render thread, after all commands have completed.