All Downloads are FREE. Search and download functionalities are using the official Maven repository.

com.sun.electric.tool.util.concurrent.datastructures.FCQueue Maven / Gradle / Ivy

/* -*- tab-width: 4 -*-
 *
 * Electric(tm) VLSI Design System
 *
 * File: FCQueue.java
 * Written by: Itai Incze, Tel-Aviv University
 *
 * Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.
 *
 * Electric(tm) is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 3 of the License, or
 * (at your option) any later version.
 *
 * Electric(tm) is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with Electric(tm); see the file COPYING.  If not, write to
 * the Free Software Foundation, Inc., 59 Temple Place, Suite 330,
 * Boston, Mass 02111-1307, USA.
 */
package com.sun.electric.tool.util.concurrent.datastructures;

import java.util.concurrent.atomic.AtomicInteger;
import java.util.concurrent.atomic.AtomicReferenceFieldUpdater;


/**
 * 
 * @author itai
 */

public class FCQueue extends IStructure {
	// Maximum participating threads
	final int MAX_THREADS = 512;

	static class CombiningNode {
		volatile boolean is_linked;
		int last_request_timestamp;
		// comb_list_head CAS will perform the write for this
		CombiningNode next;
		volatile boolean is_request_valid;
		// membar on item and is_consumer is committed by a write to
		// is_request_valid
		boolean is_consumer;
		T item;

		CombiningNode() {
			is_linked = false;
			next = null;
			is_request_valid = false;
		}
	}

	AtomicInteger fc_lock;
	// used to gather combined enqueued items
	T[] combined_pushed_items;

	volatile int current_timestamp = 0;

	private ThreadLocal> combining_node = new ThreadLocal>() {
		@Override
		protected CombiningNode initialValue() {
			return new CombiningNode();
		}
	};

	volatile CombiningNode comb_list_head;
	// For compareAndSet on the _req_list_head
	@SuppressWarnings("unchecked")
	final private static AtomicReferenceFieldUpdater comb_list_head_updater = AtomicReferenceFieldUpdater
			.newUpdater(FCQueue.class, CombiningNode.class, "comb_list_head");

	static class QueueFatNode {
		T items[];
		int items_left;
		QueueFatNode next;
	}

	volatile QueueFatNode queue_head, queue_tail;

	@SuppressWarnings("unchecked")
	public FCQueue() {
		combined_pushed_items = (T[]) new Object[MAX_THREADS];
		fc_lock = new AtomicInteger(0);
		queue_head = new QueueFatNode();
		queue_tail = queue_head;
		queue_head.next = null;
		queue_head.items_left = 0;
	}

	final int COMBINING_NODE_TIMEOUT = 10000;
	final int COMBINING_NODE_TIMEOUT_CHECK_FREQUENCY = 100;
	final int MAX_COMBINING_ROUNDS = 32;

	@SuppressWarnings("unchecked")
	void doFlatCombining(CombiningNode combiner_thread_node) {
		int combining_rounds = 0;
		int num_pushed_items = 0;
		CombiningNode cur_comb_node = null;
		CombiningNode last_combining_node = null;

		// advance timestamp and sample volatile variables to local variables
		// for reading speed
		int local_current_timestamp = ++current_timestamp;
		QueueFatNode local_queue_head = queue_head;

		boolean check_timestamps = (local_current_timestamp
				% COMBINING_NODE_TIMEOUT_CHECK_FREQUENCY == 0);
		boolean have_work = false;

		while (true) {
			if(this.abort) {
				return;
			}
			// initialize for a new round
			num_pushed_items = 0;
			cur_comb_node = comb_list_head;
			last_combining_node = cur_comb_node;
			have_work = false;

			while (cur_comb_node != null) {
				if (!cur_comb_node.is_request_valid) {
					// After manipulating is_linked the owner thread can change
					// next so we need to save it first
					CombiningNode next_node = cur_comb_node.next;

					// take the node out if its not the first one
					// (we're letting the first one go to avoid CASes)
					if ((check_timestamps)
							&& (cur_comb_node != comb_list_head)
							&& (local_current_timestamp - cur_comb_node.last_request_timestamp > COMBINING_NODE_TIMEOUT)) {
						last_combining_node.next = next_node;
						cur_comb_node.is_linked = false;
					}
					cur_comb_node = next_node;
					continue;
				}

				have_work = true;

				// update combining node last use timestamp
				cur_comb_node.last_request_timestamp = local_current_timestamp;

				if (cur_comb_node.is_consumer) {
					boolean consumer_satisfied = false;
					// check queue first
					while ((local_queue_head.next != null) && !consumer_satisfied) {
						QueueFatNode head_next = local_queue_head.next;
						if (head_next.items_left == 0) {
							local_queue_head = head_next;
						} else {
							head_next.items_left--;
							cur_comb_node.item = head_next.items[head_next.items_left];
							consumer_satisfied = true;
						}
					}

					// if queue is empty, check current pass
					if ((!consumer_satisfied) && (num_pushed_items > 0)) {
						num_pushed_items--;
						cur_comb_node.item = combined_pushed_items[num_pushed_items];
						consumer_satisfied = true;
					}

					if (!consumer_satisfied) {
						// queue empty
						cur_comb_node.item = null;
					}
				} else {
					combined_pushed_items[num_pushed_items] = cur_comb_node.item;
					num_pushed_items++;
				}

				// requesting thread is released
				cur_comb_node.is_request_valid = false;

				// next node
				last_combining_node = cur_comb_node;
				cur_comb_node = cur_comb_node.next;
			}

			// pushed items needs to go into the queue
			if (num_pushed_items > 0) {
				QueueFatNode new_node = new QueueFatNode();
				new_node.items_left = num_pushed_items;
				new_node.items = (T[]) new Object[num_pushed_items];
				System.arraycopy(combined_pushed_items, 0, new_node.items, 0, num_pushed_items);
				new_node.next = null;
				queue_tail.next = new_node;
				queue_tail = new_node;
			}

			combining_rounds++;
			if ((!have_work) || (combining_rounds >= MAX_COMBINING_ROUNDS)) {
				// no more rounds needed

				// Update queue_head.
				// This membar flushes write queue so it also finalize changes
				// made to the queue nodes
				queue_head = local_queue_head;

				return;
			}
		}
	}

	private void link_in_combining(CombiningNode cn) {
		while (true) {
			if (abort) {
				return;
			}
			// snapshot the list head
			CombiningNode cur_head = comb_list_head;
			cn.next = cur_head;

			// try to insert the node
			if (comb_list_head == cur_head) {
				if (comb_list_head_updater.compareAndSet(this, cn.next, cn)) {
					return;
				}
			}
		}
	}

	final int NUM_ROUNDS_IS_LINKED_CHECK_FREQUENCY = 100;

	private void wait_until_fulfilled(CombiningNode comb_node) {
		int rounds = 0;

		while (true) {
			if (abort) {
				return;
			}
			// make sure the combining node is in the list
			if ((rounds % NUM_ROUNDS_IS_LINKED_CHECK_FREQUENCY == 0) && (!comb_node.is_linked)) {
				comb_node.is_linked = true;
				link_in_combining(comb_node);
			}

			if (fc_lock.get() == 0) {
				if (fc_lock.compareAndSet(0, 1)) {
					// combiner
					doFlatCombining(comb_node);
					fc_lock.set(0);
				}
			}

			if (!comb_node.is_request_valid) {
				return;
			}

			rounds++;
		}
	}

	public void add(T value) {
		CombiningNode comb_node = combining_node.get();
		comb_node.is_consumer = false;
		comb_node.item = value;

		comb_node.is_request_valid = true;

		wait_until_fulfilled(comb_node);
	}

	public T remove() {
		CombiningNode comb_node = combining_node.get();
		comb_node.is_consumer = true;

		comb_node.is_request_valid = true;

		wait_until_fulfilled(comb_node);
		return comb_node.item;
	}

	public int size() {
		return 0;
	}

	public String name() {
		return this.getClass().toString();
	}

	/*
	 * (non-Javadoc)
	 * 
	 * @see com.sun.electric.tool.util.IStructure#isEmpty()
	 */
	@Override
	public boolean isEmpty() {
		return false;
	}
}




© 2015 - 2025 Weber Informatics LLC | Privacy Policy