PixelPaint: Add a general-purpose parallel image processing pipeline

The ImageProcessor singleton is intended to be used by all sorts of
image processing which might take some time to complete; or other
background actions. We're not using BackgroundTask here because this
system is specifically designed to work with task queues and PixelPaint
interaction; e.g. it provides common image processing tasks such as
filter application.

Userland/Applications/PixelPaint/ImageProcessor.h

@@ -0,0 +1,76 @@
+/*
+ * Copyright (c) 2022, kleines Filmröllchen <filmroellchen@serenityos.org>.
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include "Filters/Filter.h"
+#include <AK/Function.h>
+#include <AK/NonnullRefPtr.h>
+#include <AK/RefCounted.h>
+#include <AK/Singleton.h>
+#include <LibCore/SharedCircularQueue.h>
+#include <LibThreading/ConditionVariable.h>
+#include <LibThreading/Thread.h>
+
+namespace PixelPaint {
+
+class ImageProcessingCommand : public RefCounted<ImageProcessingCommand> {
+public:
+    virtual void execute() = 0;
+    virtual ~ImageProcessingCommand() = default;
+};
+
+// A command applying a filter from a source to a target bitmap.
+class FilterApplicationCommand : public ImageProcessingCommand {
+public:
+    FilterApplicationCommand(NonnullRefPtr<Filter>, NonnullRefPtr<Layer>);
+
+    virtual void execute() override;
+    virtual ~FilterApplicationCommand() = default;
+
+private:
+    NonnullRefPtr<Filter> m_filter;
+    NonnullRefPtr<Layer> m_target_layer;
+};
+
+// A command based on a custom user function.
+class FunctionCommand : public ImageProcessingCommand {
+public:
+    FunctionCommand(Function<void()> function)
+        : m_function(move(function))
+    {
+    }
+
+    virtual void execute() override { m_function(); }
+
+private:
+    Function<void()> m_function;
+};
+
+// A utility class that allows various PixelPaint systems to execute image processing commands asynchronously on another thread.
+class ImageProcessor final {
+    friend struct AK::SingletonInstanceCreator<ImageProcessor>;
+
+public:
+    static ImageProcessor* the();
+
+    ErrorOr<void> enqueue_command(NonnullRefPtr<ImageProcessingCommand> command);
+
+private:
+    ImageProcessor();
+    void processor_main();
+
+    // Only the memory in the queue is in shared memory, i.e. the smart pointers themselves.
+    // The actual data will remain in normal memory, but for this application we're not using multiple processes so it's fine.
+    using Queue = Core::SharedSingleProducerCircularQueue<RefPtr<ImageProcessingCommand>>;
+    Queue m_command_queue;
+
+    NonnullRefPtr<Threading::Thread> m_processor_thread;
+    Threading::Mutex m_wakeup_mutex {};
+    Threading::ConditionVariable m_wakeup_variable;
+};
+
+}