Annotation of embedaddon/strongswan/src/libstrongswan/processing/scheduler.h, revision 1.1

1.1     ! misho       1: /*
        !             2:  * Copyright (C) 2009-2015 Tobias Brunner
        !             3:  * Copyright (C) 2005-2007 Martin Willi
        !             4:  * Copyright (C) 2005 Jan Hutter
        !             5:  * HSR Hochschule fuer Technik Rapperswil
        !             6:  *
        !             7:  * This program is free software; you can redistribute it and/or modify it
        !             8:  * under the terms of the GNU General Public License as published by the
        !             9:  * Free Software Foundation; either version 2 of the License, or (at your
        !            10:  * option) any later version.  See <http://www.fsf.org/copyleft/gpl.txt>.
        !            11:  *
        !            12:  * This program is distributed in the hope that it will be useful, but
        !            13:  * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
        !            14:  * or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
        !            15:  * for more details.
        !            16:  */
        !            17: 
        !            18: /**
        !            19:  * @defgroup scheduler scheduler
        !            20:  * @{ @ingroup processing
        !            21:  */
        !            22: 
        !            23: #ifndef SCHEDULER_H_
        !            24: #define SCHEDULER_H_
        !            25: 
        !            26: typedef struct scheduler_t scheduler_t;
        !            27: 
        !            28: #include <library.h>
        !            29: #include <processing/jobs/job.h>
        !            30: 
        !            31: /**
        !            32:  * The scheduler queues timed events which are then passed to the processor.
        !            33:  *
        !            34:  * The scheduler is implemented as a heap. A heap is a special kind of tree-
        !            35:  * based data structure that satisfies the following property: if B is a child
        !            36:  * node of A, then key(A) >= (or <=) key(B). So either the element with the
        !            37:  * greatest (max-heap) or the smallest (min-heap) key is the root of the heap.
        !            38:  * We use a min-heap with the key being the absolute unix time at which an
        !            39:  * event is scheduled. So the root is always the event that will fire next.
        !            40:  *
        !            41:  * An earlier implementation of the scheduler used a sorted linked list to store
        !            42:  * the events. That had the advantage that removing the next event was extremely
        !            43:  * fast, also, adding an event scheduled before or after all other events was
        !            44:  * equally fast (all in O(1)). The problem was, though, that adding an event
        !            45:  * in-between got slower, as the number of events grew larger (O(n)).
        !            46:  * For each connection there could be several events: IKE-rekey, NAT-keepalive,
        !            47:  * retransmissions, expire (half-open), and others. So a gateway that probably
        !            48:  * has to handle thousands of concurrent connections has to be able to queue a
        !            49:  * large number of events as fast as possible. Locking makes this even worse, to
        !            50:  * provide thread-safety, no events can be processed, while an event is queued,
        !            51:  * so making the insertion fast is even more important.
        !            52:  *
        !            53:  * That's the advantage of the heap. Adding an element to the heap can be
        !            54:  * achieved in O(log n) - on the other hand, removing the root node also
        !            55:  * requires O(log n) operations. Consider 10000 queued events. Inserting a new
        !            56:  * event in the list implementation required up to 10000 comparisons. In the
        !            57:  * heap implementation, the worst case is about 13.3 comparisons. That's a
        !            58:  * drastic improvement.
        !            59:  *
        !            60:  * The implementation itself uses a binary tree mapped to a one-based array to
        !            61:  * store the elements. This reduces storage overhead and simplifies navigation:
        !            62:  * the children of the node at position n are at position 2n and 2n+1 (likewise
        !            63:  * the parent node of the node at position n is at position [n/2]). Thus,
        !            64:  * navigating up and down the tree is reduced to simple index computations.
        !            65:  *
        !            66:  * Adding an element to the heap works as follows: The heap is always filled
        !            67:  * from left to right, until a row is full, then the next row is filled. Mapped
        !            68:  * to an array this gets as simple as putting the new element to the first free
        !            69:  * position. In a one-based array that position equals the number of elements
        !            70:  * currently stored in the heap. Then the heap property has to be restored, i.e.
        !            71:  * the new element has to be "bubbled up" the tree until the parent node's key
        !            72:  * is smaller or the element got the new root of the tree.
        !            73:  *
        !            74:  * Removing the next event from the heap works similarly. The event itself is
        !            75:  * the root node and stored at position 1 of the array. After removing it, the
        !            76:  * root has to be replaced and the heap property has to be restored. This is
        !            77:  * done by moving the bottom element (last row, rightmost element) to the root
        !            78:  * and then "seep it down" by swapping it with child nodes until none of the
        !            79:  * children has a smaller key or it is again a leaf node.
        !            80:  */
        !            81: struct scheduler_t {
        !            82: 
        !            83:        /**
        !            84:         * Adds a event to the queue, using a relative time offset in s.
        !            85:         *
        !            86:         * @param job                   job to schedule
        !            87:         * @param time                  relative time to schedule job, in s
        !            88:         */
        !            89:        void (*schedule_job) (scheduler_t *this, job_t *job, uint32_t s);
        !            90: 
        !            91:        /**
        !            92:         * Adds a event to the queue, using a relative time offset in ms.
        !            93:         *
        !            94:         * @param job                   job to schedule
        !            95:         * @param time                  relative time to schedule job, in ms
        !            96:         */
        !            97:        void (*schedule_job_ms) (scheduler_t *this, job_t *job, uint32_t ms);
        !            98: 
        !            99:        /**
        !           100:         * Adds a event to the queue, using an absolute time.
        !           101:         *
        !           102:         * The passed timeval should be calculated based on the time_monotonic()
        !           103:         * function.
        !           104:         *
        !           105:         * @param job                   job to schedule
        !           106:         * @param time                  absolute time to schedule job
        !           107:         */
        !           108:        void (*schedule_job_tv) (scheduler_t *this, job_t *job, timeval_t tv);
        !           109: 
        !           110:        /**
        !           111:         * Returns number of jobs scheduled.
        !           112:         *
        !           113:         * @return                              number of scheduled jobs
        !           114:         */
        !           115:        u_int (*get_job_load) (scheduler_t *this);
        !           116: 
        !           117:        /**
        !           118:         * Remove all scheduled jobs.
        !           119:         */
        !           120:        void (*flush)(scheduler_t *this);
        !           121: 
        !           122:        /**
        !           123:         * Destroys a scheduler object.
        !           124:         */
        !           125:        void (*destroy) (scheduler_t *this);
        !           126: };
        !           127: 
        !           128: /**
        !           129:  * Create a scheduler.
        !           130:  *
        !           131:  * @return             scheduler_t object
        !           132:  */
        !           133: scheduler_t *scheduler_create(void);
        !           134: 
        !           135: #endif /** SCHEDULER_H_ @}*/

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>