Linux queue.h之TAILQ队列分析

这两天想看看memcached的实现,所以先学习了libevent,使用起来还是比较简单的,其实是对select/poll/kqueue等的封装,学习libevent过程中又遇到了linux下队列的使用,简单分析如下,权当做记录:

libevent中的例子中使用的是FreeBSD下的queue.h,在linux的/usr/include/sys/queue.h也有该头文件,但是是一个缩减版本,而且没有看到queue 的access method,不知道是不是跟我们的linux服务器版本有关,没办法google了一下,找到了FreeBSD 下queue.h的定义,我们看一下tail queue的定义

#define TAILQ_HEAD(name, type)				\
struct name {						\
	struct type *tqh_first;	/* first element */	\
	struct type **tqh_last;	/* addr of last next element */\
}

#define TAILQ_ENTRY(type)					\
struct {							\
	struct type *tqe_next;	/* next element */		\
	struct type **tqe_prev;/* addr of previous next element*/ \
}                                                                

#define	TAILQ_INIT(head) do {				\
	(head)->tqh_first = NULL;				\
	(head)->tqh_last = &(head)->tqh_first;		\
} while (0)

#define TAILQ_INSERT_TAIL(head, elm, field) do {		\
	(elm)->field.tqe_next = NULL;			\
	(elm)->field.tqe_prev = (head)->tqh_last;		\
	*(head)->tqh_last = (elm);				\
	(head)->tqh_last = &(elm)->field.tqe_next;		\
} while (0)

#define	TAILQ_INSERT_BEFORE(listelm, elm, field) do {		\
	(elm)->field.tqe_prev = (listelm)->field.tqe_prev;	\
	(elm)->field.tqe_next = (listelm);			\
	*(listelm)->field.tqe_prev = (elm);			\
	(listelm)->field.tqe_prev = &(elm)->field.tqe_next;	\
} while (0)
#define	TAILQ_FIRST(head)		((head)->tqh_first)

#define	TAILQ_NEXT(elm, field)		((elm)->field.tqe_next)
....

我们就先分析上面的这些定义,先看个应用的例子

#include <stdio.h>
#include <stdlib.h>
#include "queue.h"

struct QUEUE_ITEM{
	int value;
	TAILQ_ENTRY(QUEUE_ITEM) entries;
};
TAILQ_HEAD(,QUEUE_ITEM) queue_head;
int main(int argc,char **argv){
	struct QUEUE_ITEM *item;
	struct QUEUE_ITEM *tmp_item;

	TAILQ_INIT(&queue_head);
	int i=0;
	for(i=5;i<10;i+=2){
		item=malloc(sizeof(item));
		item->value=i;
		TAILQ_INSERT_TAIL(&queue_head, item, entries);
	}
	
	struct QUEUE_ITEM *ins_item;
	ins_item=malloc(sizeof(ins_item));

	ins_item->value=100;
	TAILQ_INSERT_BEFORE(item,ins_item,entries);


	tmp_item=TAILQ_FIRST(&queue_head);
	printf("first element is %d\n",tmp_item->value);

	tmp_item=TAILQ_NEXT(tmp_item,entries);
	printf("next element is %d\n",tmp_item->value);

	tmp_item=TAILQ_NEXT(tmp_item,entries);
	printf("next element is %d\n",tmp_item->value);

	tmp_item=TAILQ_NEXT(tmp_item,entries);
	printf("next element is %d\n",tmp_item->value);

}

结果:

first element is 5
next element is 7
next element is 100
next element is 9

分析:
QUEUE_ITEM 是我们定义的存放在队列里的东东,简单起见只包括一个int值
TAILQ_ENTRY(QUEUE_ITEM) entries 主要是存放下一个对象和前一个对象的指针,具体见 header
 
根据头文件进行宏替换后,实际我们声明的是这样的结构:

 struct QUEUE_ITEM{
	int value;
        struct {			
            struct QUEUE_ITEM *tqe_next;	
	    struct QUEUE_ITEM **tqe_prev;
        }entries;
};
 

TAILQ_HEAD(,QUEUE_ITEM) queue_head; 实际是

struct {				
	struct QUEUE_ITEM *tqh_first;	
	struct QUEUE_ITEM **tqh_last;	
}queue_head;

接着我们定义了QUEUE_ITEM的两个指针变量item和tmp_item

TAILQ_INIT(&queue_head); 相当于是

do {
	(&queue_head)->tqh_first = NULL;				
	(&queue_head)->tqh_last = &(&queue_head)->tqh_first;		
} while (0);

head的初始化如 下图1

接着我们通过循环分配了几个元素,并赋值

TAILQ_INSERT_TAIL(&queue_head, item, entries); 相当于执行

do {
	(item)->entries.tqe_next = NULL;	
	(item)->entries.tqe_prev = (&queue_head)->tqh_last;			
	*(&queue_head)->tqh_last = (item);					
	(&queue_head)->tqh_last = &(item)->entries.tqe_next;			
} while (0);

也就是我们的循环执行下面代码段,结果分析见图2,3

for(i=5;i<10;i+=2){
	item=malloc(sizeof(item));
	item->value=i;
	do {
		(item)->entries.tqe_next = NULL;
		//首次执行相当于item->entries.tqe_prev=&(&queue_head)->tqh_first
		//以后执行相当于是(item)->entries.tqe_prev=&(前一个item)->entries.tqe_next;
		(item)->entries.tqe_prev = (&queue_head)->tqh_last;
		//首次执行相当于(&queue_head)->tqh_first=item
		//以后执行相当于是(前一个item)->entries.tqe_next=当前item
		*(&queue_head)->tqh_last = (item);
		(&queue_head)->tqh_last = &(item)->entries.tqe_next;
	} while (0);
}

最终建立的链表结构如图,下面看一下insert操作,经过宏替换后代码如下

struct QUEUE_ITEM *ins_item;
ins_item=malloc(sizeof(ins_item));
ins_item->value=100;

do {
	(ins_item)->entries.tqe_prev = (item)->entries.tqe_prev;
	(ins_item)->entries.tqe_next = (item);
	//这句话体现了TAILQ的特色,tqe_prev是前一个元素的下个元素地址,
	//所以正好应该是当前插入item的地址
	*(item)->entries.tqe_prev = (ins_item);
	(item)->entries.tqe_prev = &(ins_item)->entries.tqe_next;
} while (0);

总结:TAILQ的最大特点就是每个entry的二级指针tqe_prev其存放的是前一个元素的下个元素地址,呵呵,听起来都很拗口
我现在就是不知道为什么linux的queue.h只有建立tailq的宏定义而缺少所有的access method,初涉linux c编程,请大家指教

附经过宏替换后的所有代码

#include "stdio.h"
#include "stdlib.h"
struct QUEUE_ITEM{
	int value;
	struct {
		struct QUEUE_ITEM *tqe_next;
		struct QUEUE_ITEM **tqe_prev;
	}entries;
};
struct {
	struct QUEUE_ITEM *tqh_first;
	struct QUEUE_ITEM **tqh_last;
}queue_head;

int main(int argc,char **argv){
	struct QUEUE_ITEM *item;
	struct QUEUE_ITEM *tmp_item;

	do {
		(&queue_head)->tqh_first = NULL;
		(&queue_head)->tqh_last = &(&queue_head)->tqh_first;
	} while (0);

	int i=0;
	for(i=5;i<10;i+=2){
		item=malloc(sizeof(item));
		item->value=i;
		do {
			(item)->entries.tqe_next = NULL;
			//首次执行相当于item->entries.tqe_prev=&(&queue_head)->tqh_first
			//以后执行相当于是(item)->entries.tqe_prev=&(前一个item)->entries.tqe_next;
			(item)->entries.tqe_prev = (&queue_head)->tqh_last;
			//首次执行相当于(&queue_head)->tqh_first=item
			//以后执行相当于是(前一个item)->entries.tqe_next=当前item
			*(&queue_head)->tqh_last = (item);
			(&queue_head)->tqh_last = &(item)->entries.tqe_next;
		} while (0);
	}

	struct QUEUE_ITEM *ins_item;
	ins_item=malloc(sizeof(ins_item));

	ins_item->value=100;
	do {
		(ins_item)->entries.tqe_prev = (item)->entries.tqe_prev;
		(ins_item)->entries.tqe_next = (item);
		*(item)->entries.tqe_prev = (ins_item);
		(item)->entries.tqe_prev = &(ins_item)->entries.tqe_next;
	} while (0);

	tmp_item=((&queue_head)->tqh_first);
	printf("first element is %d\n",tmp_item->value);

	tmp_item=((tmp_item)->entries.tqe_next);
	printf("next element is %d\n",tmp_item->value);

	tmp_item=((tmp_item)->entries.tqe_next);
	printf("next element is %d\n",tmp_item->value);

	tmp_item=((tmp_item)->entries.tqe_next);
	printf("next element is %d\n",tmp_item->value);

}

>> 不知道为什么linux的queue.h只有建立tailq的宏定义而缺少所有的access method ....

Linux有更好用的list(see: include/linux/list.h), 请看右上角。

看到了,是有一个doubly linked list,多谢指点,呵呵