pool.h

一个开源SIP协议栈

/* $Id: pool.h 974 2007-02-19 01:13:53Z bennylp $ */
/* 
 * Copyright (C)2003-2007 Benny Prijono <[e-mail]>
 *
 * This program 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 2 of the License, or
 * (at your option) any later version.
 *
 * This program 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 this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA 
 */

#include <pj/list.h>

/* See if we use pool's alternate API.
 * The alternate API is used e.g. to implement pool debugging.
 */
#if PJ_HAS_POOL_ALT_API
#  include <pj/pool_alt.h>
#endif


#ifndef __PJ_POOL_H__
#define __PJ_POOL_H__

/**
 * @file pool.h
 * @brief Memory Pool.
 */

PJ_BEGIN_DECL

/**
 * @defgroup PJ_POOL_GROUP Fast Memory Pool
 * @ingroup PJ
 * @brief
 * Memory pools allow dynamic memory allocation comparable to malloc or the 
 * new in operator C++. Those implementations are not desirable for very
 * high performance applications or real-time systems, because of the 
 * performance bottlenecks and it suffers from fragmentation issue.
 *
 * \section PJ_POOL_INTRO_SEC PJLIB's Memory Pool
 * \subsection PJ_POOL_ADVANTAGE_SUBSEC Advantages
 * 
 * PJLIB's pool has many advantages over traditional malloc/new operator and
 * over other memory pool implementations, because:
 *  - unlike other memory pool implementation, it allows allocation of
 *    memory chunks of different sizes,
 *  - it's very very fast. 
 *    \n
 *    Memory chunk allocation is not only an O(1) 
 *    operation, but it's also very simple (just 
 *    few pointer arithmetic operations) and it doesn't require locking 
 *    any mutex,
 *  - it's memory efficient.
 *    \n
 *    Pool doesn't keep track individual memory chunks allocated by
 *    applications, so there is no additional overhead needed for each
 *    memory allocation (other than possible additional of few bytes, up to
 *    PJ_POOL_ALIGNMENT-1, for aligning the memory). 
 *    But see the @ref PJ_POOL_CAVEATS_SUBSEC below.
 *  - it prevents memory leaks. 
 *    \n
 *    Memory pool inherently has garbage collection functionality. In fact, 
 *    there is no need to free the chunks allocated from the memory pool.
 *    All chunks previously allocated from the pool will be freed once the
 *    pool itself is destroyed. This would prevent memory leaks that haunt
 *    programmers for decades, and it provides additional performance 
 *    advantage over traditional malloc/new operator.
 *
 * Even more, PJLIB's memory pool provides some additional usability and
 * flexibility for applications:
 *  - memory leaks are easily traceable, since memory pool is assigned name,
 *    and application can inspect what pools currently active in the system.
 *  - by design, memory allocation from a pool is not thread safe. We assumed
 *    that a pool will be owned by a higher level object, and thread safety 
 *    should be handled by that object. This enables very fast pool operations
 *    and prevents unnecessary locking operations,
 *  - by default, the memory pool API behaves more like C++ new operator, 
 *    in that it will throw PJ_NO_MEMORY_EXCEPTION exception (see 
 *    @ref PJ_EXCEPT) when memory chunk allocation fails. This enables failure
 *    handling to be done on more high level function (instead of checking
 *    the result of pj_pool_alloc() everytime). If application doesn't like
 *    this, the default behavior can be changed on global basis by supplying 
 *    different policy to the pool factory.
 *  - any memory allocation backend allocator/deallocator may be used. By
 *    default, the policy uses malloc() and free() to manage the pool's block,
 *    but application may use different strategy, for example to allocate
 *    memory blocks from a globally static memory location.
 *
 *
 * \subsection PJ_POOL_PERFORMANCE_SUBSEC Performance
 * 
 * The result of PJLIB's memory design and careful implementation is a
 * memory allocation strategy that can speed-up the memory allocations
 * and deallocations by up to <b>30 times</b> compared to standard
 * malloc()/free() (more than 150 million allocations per second on a
 * P4/3.0GHz Linux machine).
 *
 * (Note: your mileage may vary, of course. You can see how much PJLIB's
 *  pool improves the performance over malloc()/free() in your target
 *  system by running pjlib-test application).
 *
 *
 * \subsection PJ_POOL_CAVEATS_SUBSEC Caveats
 *
 * There are some caveats though!
 *
 * When creating pool, PJLIB requires applications to specify the initial
 * pool size, and as soon as the pool is created, PJLIB allocates memory
 * from the system by that size. Application designers MUST choose the 
 * initial pool size carefully, since choosing too big value will result in
 * wasting system's memory.
 *
 * But the pool can grow. Application designer can specify how the
 * pool will grow in size, by specifying the size increment when creating
 * the pool.
 *
 * The pool, however, <b>cannot</b> shrink! Since there is <b>no</b> 
 * function to deallocate memory chunks, there is no way for the pool to 
 * release back unused memory to the system. 
 * Application designers must be aware that constant memory allocations 
 * from pool that has infinite life-time may cause the memory usage of 
 * the application to grow over time.
 *
 *
 * \section PJ_POOL_USING_SEC Using Memory Pool
 *
 * This section describes how to use PJLIB's memory pool framework.
 * As we hope the readers will witness, PJLIB's memory pool API is quite
 * straightforward. 
 *
 * \subsection PJ_POOL_USING_F Create Pool Factory
 * First, application needs to initialize a pool factory (this normally
 * only needs to be done once in one application). PJLIB provides
 * a pool factory implementation called caching pool (see @ref 
 * PJ_CACHING_POOL), and it is initialized by calling #pj_caching_pool_init().
 *
 * \subsection PJ_POOL_USING_P Create The Pool
 * Then application creates the pool object itself with #pj_pool_create(),
 * specifying among other thing the pool factory where the pool should
 * be created from, the pool name, initial size, and increment/expansion
 * size.
 *
 * \subsection PJ_POOL_USING_M Allocate Memory as Required
 * Then whenever application needs to allocate dynamic memory, it would
 * call #pj_pool_alloc(), #pj_pool_calloc(), or #pj_pool_zalloc() to
 * allocate memory chunks from the pool.
 *
 * \subsection PJ_POOL_USING_DP Destroy the Pool
 * When application has finished with the pool, it should call 
 * #pj_pool_release() to release the pool object back to the factory. 
 * Depending on the types of the factory, this may release the memory back 
 * to the operating system.
 *
 * \subsection PJ_POOL_USING_Dc Destroy the Pool Factory
 * And finally, before application quites, it should deinitialize the
 * pool factory, to make sure that all memory blocks allocated by the
 * factory are released back to the operating system. After this, of 
 * course no more memory pool allocation can be requested.
 *
 * \subsection PJ_POOL_USING_EX Example
 * Below is a sample complete program that utilizes PJLIB's memory pool.
 *
 * \code

   #include <pjlib.h>

   #define THIS_FILE    "pool_sample.c"

   static void my_perror(const char *title, pj_status_t status)
   {
        char errmsg[PJ_ERR_MSG_SIZE];

	pj_strerror(status, errmsg, sizeof(errmsg));
	PJ_LOG(1,(THIS_FILE, "%s: %s [status=%d]", title, errmsg, status));
   }

   static void pool_demo_1(pj_pool_factory *pfactory)
   {
	unsigned i;
	pj_pool_t *pool;

	// Must create pool before we can allocate anything
	pool = pj_pool_create(pfactory,	 // the factory
			      "pool1",	 // pool's name
			      4000,	 // initial size
			      4000,	 // increment size
			      NULL);	 // use default callback.
	if (pool == NULL) {
	    my_perror("Error creating pool", PJ_ENOMEM);
	    return;
	}

	// Demo: allocate some memory chunks
	for (i=0; i<1000; ++i) {
	    void *p;

	    p = pj_pool_alloc(pool, (pj_rand()+1) % 512);

	    // Do something with p
	    ...

	    // Look! No need to free p!!
	}

	// Done with silly demo, must free pool to release all memory.
	pj_pool_release(pool);
   }

   int main()
   {
	pj_caching_pool cp;
	pj_status_t status;

        // Must init PJLIB before anything else
	status = pj_init();
	if (status != PJ_SUCCESS) {
	    my_perror("Error initializing PJLIB", status);
	    return 1;
	}

	// Create the pool factory, in this case, a caching pool.
	pj_caching_pool_init(&cp, &pj_pool_factory_default_policy, 
			     1024*1024 );

	// Do a demo
	pool_demo_1(&cp.factory);

	// Done with demos, destroy caching pool before exiting app.
	pj_caching_pool_destroy(&cp);

	return 0;
   }

   \endcode
 *
 * More information about pool factory, the pool object, and caching pool
 * can be found on the Module Links below.
 */


/**
 * @defgroup PJ_POOL Memory Pool Object
 * @ingroup PJ_POOL_GROUP
 * @brief
 * The memory pool is an opaque object created by pool factory.
 * Application uses this object to request a memory chunk, by calling
 * #pj_pool_alloc(), #pj_pool_calloc(), or #pj_pool_zalloc(). 
 * When the application has finished using
 * the pool, it must call #pj_pool_release() to free all the chunks previously
 * allocated and release the pool back to the factory.
 *
 * A memory pool is initialized with an initial amount of memory, which is
 * called a block. Pool can be configured to dynamically allocate more memory 
 * blocks when it runs out of memory. 
 *
 * The pool doesn't keep track of individual memory allocations
 * by user, and the user doesn't have to free these indidual allocations. This
 * makes memory allocation simple and very fast. All the memory allocated from
 * the pool will be destroyed when the pool itself is destroyed.
 *
 * \section PJ_POOL_THREADING_SEC More on Threading Policies
 * - By design, memory allocation from a pool is not thread safe. We assumed 
 *   that a pool will be owned by an object, and thread safety should be 
 *   handled by that object. Thus these functions are not thread safe: 
 *	- #pj_pool_alloc, 
 *	- #pj_pool_calloc, 
 *	- and other pool statistic functions.
 * - Threading in the pool factory is decided by the policy set for the
 *   factory when it was created.
 *
 * \section PJ_POOL_EXAMPLES_SEC Examples
 *
 * For some sample codes on how to use the pool, please see:
 *  - @ref page_pjlib_pool_test
 *
 * @{
 */

/**
 * The type for function to receive callback from the pool when it is unable
 * to allocate memory. The elegant way to handle this condition is to throw
 * exception, and this is what is expected by most of this library 
 * components.
 */
typedef void pj_pool_callback(pj_pool_t *pool, pj_size_t size);

/**
 * This class, which is used internally by the pool, describes a single 
 * block of memory from which user memory allocations will be allocated from.
 */
typedef struct pj_pool_block
{
    PJ_DECL_LIST_MEMBER(struct pj_pool_block);  /**< List's prev and next.  */
    unsigned char    *buf;                      /**< Start of buffer.       */
    unsigned char    *cur;                      /**< Current alloc ptr.     */
    unsigned char    *end;                      /**< End of buffer.         */
} pj_pool_block;


/**
 * This structure describes the memory pool. Only implementors of pool factory
 * need to care about the contents of this structure.
 */
struct pj_pool_t
{
    PJ_DECL_LIST_MEMBER(struct pj_pool_t);  /**< Standard list elements.    */

    /** Pool name */
    char	    obj_name[PJ_MAX_OBJ_NAME];

    /** Pool factory. */
    pj_pool_factory *factory;

    /** Data put by factory */
    void	    *factory_data;

    /** Current capacity allocated by the pool. */
    pj_size_t	    capacity;

    /** Size of memory block to be allocated when the pool runs out of memory */
    pj_size_t	    increment_size;

    /** List of memory blocks allcoated by the pool. */
    pj_pool_block   block_list;

    /** The callback to be called when the pool is unable to allocate memory. */
    pj_pool_callback *callback;

};


/**
 * Guidance on how much memory required for initial pool administrative data.
 */
#define PJ_POOL_SIZE	        (sizeof(struct pj_pool_t))

/** 
 * Pool memory alignment (must be power of 2). 
 */
#ifndef PJ_POOL_ALIGNMENT
#   define PJ_POOL_ALIGNMENT    4
#endif

/**
 * Create a new pool from the pool factory. This wrapper will call create_pool
 * member of the pool factory.
 *
 * @param factory	    The pool factory.
 * @param name		    The name to be assigned to the pool. The name should 
 *			    not be longer than PJ_MAX_OBJ_NAME (32 chars), or 
 *			    otherwise it will be truncated.
 * @param initial_size	    The size of initial memory blocks taken by the pool.
 *			    Note that the pool will take 68+20 bytes for 
 *			    administrative area from this block.
 * @param increment_size    the size of each additional blocks to be allocated
 *			    when the pool is running out of memory. If user 
 *			    requests memory which is larger than this size, then 
 *			    an error occurs.
 *			    Note that each time a pool allocates additional block, 
 *			    it needs PJ_POOL_SIZE more to store some 
 *			    administrative info.
 * @param callback	    Callback to be called when error occurs in the pool.
 *			    If this value is NULL, then the callback from pool
 *			    factory policy will be used.
 *			    Note that when an error occurs during pool creation, 
 *			    the callback itself is not called. Instead, NULL 
 *			    will be returned.
 *
 * @return                  The memory pool, or NULL.
 */
PJ_IDECL(pj_pool_t*) pj_pool_create(pj_pool_factory *factory, 
				    const char *name,
				    pj_size_t initial_size, 
				    pj_size_t increment_size,
				    pj_pool_callback *callback);

/**
 * Release the pool back to pool factory.
 *
 * @param pool	    Memory pool.
 */
PJ_IDECL(void) pj_pool_release( pj_pool_t *pool );

/**
 * Get pool object name.
 *
 * @param pool the pool.
 *
 * @return pool name as NULL terminated string.
 */
PJ_IDECL(const char *) pj_pool_getobjname( const pj_pool_t *pool );

/**
 * Reset the pool to its state when it was initialized.
 * This means that if additional blocks have been allocated during runtime, 
 * then they will be freed. Only the original block allocated during 
 * initialization is retained. This function will also reset the internal 
 * counters, such as pool capacity and used size.
 *
 * @param pool the pool.
 */
PJ_DECL(void) pj_pool_reset( pj_pool_t *pool );


/**
 * Get the pool capacity, that is, the system storage that have been allocated
 * by the pool, and have been used/will be used to allocate user requests.
 * There's no guarantee that the returned value represent a single
 * contiguous block, because the capacity may be spread in several blocks.
 *
 * @param pool	the pool.
 *
 * @return the capacity.