klist

struct klist {
	spinlock_t		k_lock;
	struct list_head	k_list;
	void			(*get)(struct klist_node *);
	void			(*put)(struct klist_node *);
} __attribute__ ((aligned (sizeof(void *))));

struct klist_node {
	void			*n_klist;	/* never access directly */
	struct list_head	n_node;
	struct kref		n_ref;
};

struct klist_iter {
	struct klist		*i_klist;
	struct klist_node	*i_cur;
};

struct klist_waiter {
	struct list_head list;
	struct klist_node *node;
	struct task_struct *process;
	int woken;
};

//=============================================================

// 将 new 插在 head 和 head->next 之间
static inline void list_add(struct list_head *new, struct list_head *head){
	__list_add(new, head, head->next);
}

//判断一个节点是否是第一个的方式如下
static inline void __list_add(struct list_head *new, struct list_head *prev, struct list_head *next)
{
	if (!__list_add_valid(new, prev, next))
		return;

	next->prev = new;
	new->next = next;
	new->prev = prev;
	WRITE_ONCE(prev->next, new);
}

/**
 * list_is_first -- tests whether @list is the first entry in list @head
 * @list: the entry to test
 * @head: the head of the list
 */
static inline int list_is_first(const struct list_head *list,
					const struct list_head *head)
{
	return list->prev == head;
}

//=============================================================

// 将 new 插在 head->prev 和 head 之间
// 那不就是最后一个嘛
/**
 * list_add_tail - add a new entry
 * @new: new entry to be added
 * @head: list head to add it before
 *
 * Insert a new entry before the specified head.
 * This is useful for implementing queues.
 */
static inline void list_add_tail(struct list_head *new, struct list_head *head)
{
	__list_add(new, head->prev, head);
}

//判断一个节点是否是最后一个的方式如下
/**
 * list_is_last - tests whether @list is the last entry in list @head
 * @list: the entry to test
 * @head: the head of the list
 */
static inline int list_is_last(const struct list_head *list,
				const struct list_head *head)
{
	return list->next == head;
}

//=============================================================
【初始化-klist_init】

/**
 * klist_init - Initialize a klist structure.
 * @k: The klist we're initializing.
 * @get: The get function for the embedding object (NULL if none)
 * @put: The put function for the embedding object (NULL if none)
 *
 * Initialises the klist structure.  If the klist_node structures are
 * going to be embedded in refcounted objects (necessary for safe
 * deletion) then the get/put arguments are used to initialise
 * functions that take and release references on the embedding
 * objects.
 */

void klist_init(struct klist *k, void (*get)(struct klist_node *),
		void (*put)(struct klist_node *))
{
	INIT_LIST_HEAD(&k->k_list);
	spin_lock_init(&k->k_lock);
	k->get = get;
	k->put = put;
}

//=============================================================
【增加-klist_add_head】

// 下面统称 节点：klist_node，头部：klist
/**
 * klist_add_head - Initialize a klist_node and add it to front.
 * @n: node we're adding.
 * @k: klist it's going on.
 */
void klist_add_head(struct klist_node *n, struct klist *k)
{
	klist_node_init(k, n); 	// 设置节点的成员
	add_head(k, n);			// 将节点添加到头部的管理列表最前面
}
EXPORT_SYMBOL_GPL(klist_add_head);

static void klist_node_init(struct klist *k, struct klist_node *n)
{
	INIT_LIST_HEAD(&n->n_node); // 初始化节点的 list_head
	kref_init(&n->n_ref);		// 节点的ref值设为1
	knode_set_klist(n, k);		// 节点的n_node插入头部的k_list最前面
	if (k->get)
		k->get(n);
}

static inline void kref_init(struct kref *kref){
	refcount_set(&kref->refcount, 1);
}

static void knode_set_klist(struct klist_node *knode, struct klist *klist)
{
	knode->n_klist = klist;
	/* no knode deserves to start its life dead */
	// 没有一个节点生而为死
	// knode_dead is (unsigned long)knode->n_klist & KNODE_DEAD
	WARN_ON(knode_dead(knode));
}

static void add_head(struct klist *k, struct klist_node *n)
{
	spin_lock(&k->k_lock);
	list_add(&n->n_node, &k->k_list); // 将节点的n_node插入头部的k_list中
	spin_unlock(&k->k_lock);
}

//=============================================================
【增加-klist_add_behind】

/**
 * klist_add_behind - Init a klist_node and add it after an existing node
 * @n: node we're adding.
 * @pos: node to put @n after
 */
void klist_add_behind(struct klist_node *n, struct klist_node *pos)
{
	struct klist *k = knode_klist(pos);

	klist_node_init(k, n);
	spin_lock(&k->k_lock);
	list_add(&n->n_node, &pos->n_node); // 将节点的n_node插入pos的后面
	spin_unlock(&k->k_lock);
}
EXPORT_SYMBOL_GPL(klist_add_behind);

// 取一个knode指向的klist时，要把最低位去掉
// 最低位用于标记dead
static struct klist *knode_klist(struct klist_node *knode){
	return (struct klist *)((unsigned long)knode->n_klist & KNODE_KLIST_MASK);
}

// Use the lowest bit of n_klist to mark deleted nodes and exclude dead ones from iteration.
#define KNODE_DEAD		1LU
#define KNODE_KLIST_MASK	~KNODE_DEAD

//=============================================================
【增加-klist_add_before】

/**
 * klist_add_before - Init a klist_node and add it before an existing node
 * @n: node we're adding.
 * @pos: node to put @n after
 */
void klist_add_before(struct klist_node *n, struct klist_node *pos)
{
	struct klist *k = knode_klist(pos);

	klist_node_init(k, n);
	spin_lock(&k->k_lock);
	list_add_tail(&n->n_node, &pos->n_node);
	spin_unlock(&k->k_lock);
}
EXPORT_SYMBOL_GPL(klist_add_before);

//=============================================================
【增加-klist_add_tail】

/**
 * klist_add_tail - Initialize a klist_node and add it to back.
 * @n: node we're adding.
 * @k: klist it's going on.
 */
void klist_add_tail(struct klist_node *n, struct klist *k)
{
	klist_node_init(k, n);
	add_tail(k, n);
}
EXPORT_SYMBOL_GPL(klist_add_tail);

//=============================================================
【遍历-klist_prev】

/**
 * klist_prev - Ante up prev node in list.
 * @i: Iterator structure.
 *
 * First grab list lock. Decrement the reference count of the previous
 * node, if there was one. Grab the prev node, increment its reference
 * count, drop the lock, and return that prev node.
 */
// 每次迭代的时候要把前一个节点的计数递减，然后把下一个节点的计数递增
// 这些变量的名字起的真好
struct klist_node *klist_prev(struct klist_iter *i)
{
	void (*put)(struct klist_node *) = i->i_klist->put;
	struct klist_node *last = i->i_cur;
	struct klist_node *prev;
	unsigned long flags;

	spin_lock_irqsave(&i->i_klist->k_lock, flags);

	if (last) {
		prev = to_klist_node(last->n_node.prev);
		if (!klist_dec_and_del(last))
			put = NULL;
	} else
		prev = to_klist_node(i->i_klist->k_list.prev);

	i->i_cur = NULL;
	while (prev != to_klist_node(&i->i_klist->k_list)) {
		if (likely(!knode_dead(prev))) { //很有可能没有死
			kref_get(&prev->n_ref);
			i->i_cur = prev;
			break;
		}
		prev = to_klist_node(prev->n_node.prev);
	}

	spin_unlock_irqrestore(&i->i_klist->k_lock, flags);

	if (put && last)
		put(last);
	return i->i_cur;
}
EXPORT_SYMBOL_GPL(klist_prev);

//=============================================================
【遍历-klist_next】

/**
 * klist_next - Ante up next node in list.
 * @i: Iterator structure.
 *
 * First grab list lock. Decrement the reference count of the previous
 * node, if there was one. Grab the next node, increment its reference
 * count, drop the lock, and return that next node.
 */
// 每次迭代的时候要把前一个节点的计数递减，然后把下一个节点的计数递增
struct klist_node *klist_next(struct klist_iter *i)
{
	void (*put)(struct klist_node *) = i->i_klist->put;
	struct klist_node *last = i->i_cur;
	struct klist_node *next;
	unsigned long flags;

	spin_lock_irqsave(&i->i_klist->k_lock, flags);

	if (last) {
		next = to_klist_node(last->n_node.next);
		if (!klist_dec_and_del(last))
			put = NULL;
	} else
		next = to_klist_node(i->i_klist->k_list.next);

	i->i_cur = NULL;
	while (next != to_klist_node(&i->i_klist->k_list)) {
		if (likely(!knode_dead(next))) {
			kref_get(&next->n_ref);
			i->i_cur = next;
			break;
		}
		next = to_klist_node(next->n_node.next);
	}

	spin_unlock_irqrestore(&i->i_klist->k_lock, flags);

	if (put && last)
		put(last);
	return i->i_cur;
}
EXPORT_SYMBOL_GPL(klist_next);

//======================================================================
【删除-klist_del】

// 这个是递减计数，试图remove
/**
 * klist_del - Decrement the reference count of node and try to remove.
 * @n: node we're deleting.
 */
void klist_del(struct klist_node *n)
{
	klist_put(n, true);
}
EXPORT_SYMBOL_GPL(klist_del);

static void klist_put(struct klist_node *n, bool kill)
{
	struct klist *k = knode_klist(n);
	void (*put)(struct klist_node *) = k->put;

	spin_lock(&k->k_lock);
	if (kill)
		knode_kill(n); 			// 将该 klist_node 的 n_klist 指针的值标记为 DEAD
	if (!klist_dec_and_del(n)) 	// 引用计数减1，如果计数为0，则执行 klist_release 操作
		put = NULL;
	spin_unlock(&k->k_lock);
	if (put) 					// 只有在 klist_node 的计数减到0之后，才执行 klist 的 put 操作
		put(n);
}

static void knode_kill(struct klist_node *knode)
{
	/* and no knode should die twice ever either, see we're very humane */
	// 死一次足矣
	WARN_ON(knode_dead(knode));
	// knode->n_klist 该指针指向 klist

	// *((unsigned long *)&(knode->n_klist)) |= KNODE_DEAD;
	// knode->n_klist 是 knode 指针所指的结构体中 n_klist 的值 （该值是管该knode的klist的地址）
	// 此时该值以 void *的数据存放（因为 knode->n_klist 的类型是 void *）
	// 然后对上述值取地址，将地址强转为 unsigned long *
	// 再解引用，那不就是将之前的值从原来类型（void *）转为 unsigned long 类型了
	*(unsigned long *)&knode->n_klist |= KNODE_DEAD;

	// 举例如下：
	// klist 的地址 0x12345678
	// knode 的地址 0x00001212
	// knode 中 n_klist指针的地址为 0x00001212 ，n_klist指针的类型为 (void *) ，n_klist的值为 0x12345678
	// 目前 knode 中 n_klist指针的值 0x12345678 值不可修改？（该值是一个void类型的地址）
	// &knode->n_klist 就是 0x00001212
	// (unsigned long *)&knode->n_klist 就是 (unsigned long *)0x00001212
	// 那么对应 n_klist指针的类型目前变为 (unsigned long *)，n_klist的值不变
	// *(unsigned long *)&knode->n_klist 就是 0x12345678
	// 目前 knode 中 n_klist指针的值 0x12345678 值可修改（该值是一个unsigned long类型的地址）
	// *(unsigned long *)&knode->n_klist |= KNODE_DEAD 就是 (unsigned long)0x12345679
	// 所以 knode 中 n_klist指针的地址为 0x00001212，指针的值为 0x12345679
	// klist 的地址依旧没变，变的是 knode 中 n_klist指针 的值
}

static int klist_dec_and_del(struct klist_node *n){
	return kref_put(&n->n_ref, klist_release); // 自带灭火器
}

/**
 * kref_put - decrement refcount for object.
 * @kref: object.
 * @release: pointer to the function that will clean up the object when the
 *	     last reference to the object is released.
 *	     This pointer is required, and it is not acceptable to pass kfree
 *	     in as this function.
 *
 * Decrement the refcount, and if 0, call release().
 * Return 1 if the object was removed, otherwise return 0.  Beware, if this
 * function returns 0, you still can not count on the kref from remaining in
 * memory.  Only use the return value if you want to see if the kref is now
 * gone, not present.
 */
// 递减引用计数，如果为0，则调用 release函数
// 返回1表示该节点已经被清理
// 返回0表示仅仅递减了引用计数
static inline int kref_put(struct kref *kref, void (*release)(struct kref *kref))
{
	if (refcount_dec_and_test(&kref->refcount)) {
		release(kref);
		return 1;
	}
	return 0;
}

static void klist_release(struct kref *kref)
{
	struct klist_waiter *waiter, *tmp;
	struct klist_node *n = container_of(kref, struct klist_node, n_ref);

	WARN_ON(!knode_dead(n));
	list_del(&n->n_node); 	//从管理链表中移除
	spin_lock(&klist_remove_lock);
	// 遍历等待列表，看哪个 waiter 等待在该 klist_node 上
	// 如果有，则将该 waiter 从等待链表删除，将该 waiter 标记为 1
	// 然后唤醒该 waiter 所在的 process
	list_for_each_entry_safe(waiter, tmp, &klist_remove_waiters, list) {
		if (waiter->node != n)
			continue;

		list_del(&waiter->list);
		waiter->woken = 1;
		mb();
		wake_up_process(waiter->process);
	}
	spin_unlock(&klist_remove_lock);
	knode_set_klist(n, NULL);
}

//======================================================================
【删除-klist_remove】

// 这个是递减计数，一直等到它“灭亡”
**
 * klist_remove - Decrement the refcount of node and wait for it to go away.
 * @n: node we're removing.
 */
void klist_remove(struct klist_node *n)
{
	struct klist_waiter waiter;

	waiter.node = n;
	waiter.process = current;
	waiter.woken = 0;
	spin_lock(&klist_remove_lock);
	list_add(&waiter.list, &klist_remove_waiters); // 将该进程添加到移除队列klist_remove_waiters
	spin_unlock(&klist_remove_lock);

	klist_del(n);

	for (;;) {
		set_current_state(TASK_UNINTERRUPTIBLE);
		if (waiter.woken)
			break;
		schedule();
	}
	__set_current_state(TASK_RUNNING);
}
EXPORT_SYMBOL_GPL(klist_remove);

// 将前面的各个接口都整理进 klist_remove
// 才能发现这个接口设计的美妙
void klist_remove(struct klist_node *n)
{
	struct klist_waiter waiter;

	waiter.node = n;
	waiter.process = current;
	waiter.woken = 0;
	spin_lock(&klist_remove_lock);
	list_add(&waiter.list, &klist_remove_waiters); // 将该进程添加到移除队列klist_remove_waiters
	spin_unlock(&klist_remove_lock);

	struct klist *k = knode_klist(n);
	void (*put)(struct klist_node *) = k->put;

	spin_lock(&k->k_lock);
	knode_kill(n); 									// 将该 klist_node 的 n_klist 指针的值标记为 DEAD
	if (refcount_dec_and_test(&kref->refcount)) {	// 计数减1，计数为0则执行 klist_release 操作
		struct klist_waiter *waiter, *tmp;
		struct klist_node *n = container_of(kref, struct klist_node, n_ref);

		WARN_ON(!knode_dead(n));
		list_del(&n->n_node); 	//从管理链表中移除
		spin_lock(&klist_remove_lock);
		// 遍历等待列表，看哪个 waiter 等待在该 klist_node 上
		// 如果有，则将该 waiter 从等待链表删除，将该 waiter 标记为 1
		// 然后唤醒该 waiter 所在的 process
		list_for_each_entry_safe(waiter, tmp, &klist_remove_waiters, list) {
			if (waiter->node != n)
				continue;

			list_del(&waiter->list);
			waiter->woken = 1;
			mb();
			wake_up_process(waiter->process); // 唤醒等待在该删除node上的waiter process
		}
		spin_unlock(&klist_remove_lock);
		knode_set_klist(n, NULL);
	} else {
		put = NULL;
	}
	spin_unlock(&k->k_lock);

	if (put) 					// 只有在 klist_node 的计数减到0之后，才执行 klist 的 put 操作
		put(n);

	for (;;) {
		set_current_state(TASK_UNINTERRUPTIBLE);
		// woken 主要用在这里，如果前面计数不为0，则调用 klist_remove 的进程都会跑进这里
		// 只有计数减为0的进程才会将“等待删除该节点的进程”的waiter的woken置为1
		// woken 用的真是奇妙
		if (waiter.woken)
			break;
		schedule();
	}
	__set_current_state(TASK_RUNNING);
}

//======================================================================
【判断一个节点是否归属于某个klist】

/**
 * klist_node_attached - Say whether a node is bound to a list or not.
 * @n: Node that we're testing.
 */
int klist_node_attached(struct klist_node *n)
{
	return (n->n_klist != NULL);
}
EXPORT_SYMBOL_GPL(klist_node_attached);

//======================================================================
【klist_iter的初始化-klist_iter_init_node和klist_iter_init】

/**
 * klist_iter_init_node - Initialize a klist_iter structure.
 * @k: klist we're iterating.
 * @i: klist_iter we're filling.
 * @n: node to start with.
 *
 * Similar to klist_iter_init(), but starts the action off with @n,
 * instead of with the list head.
 */
// 初始化一个 klist_iter 指针所指的结构体
// 这个接口：传入的变量有 klist_node ，但是内容中会判断迭代的 klist_node 的计数是否为0
//           为0则代表该 klist_node 并没有初始化完成或者已被删除，因此 i->i_cur = NULL

// kref_get_unless_zero 用于无锁的原子操作
void klist_iter_init_node(struct klist *k, struct klist_iter *i,
			  struct klist_node *n)
{
	i->i_klist = k;
	i->i_cur = NULL;
	if (n && kref_get_unless_zero(&n->n_ref))
		i->i_cur = n;
}
EXPORT_SYMBOL_GPL(klist_iter_init_node);

/**
 * klist_iter_init - Iniitalize a klist_iter structure.
 * @k: klist we're iterating.
 * @i: klist_iter structure we're filling.
 *
 * Similar to klist_iter_init_node(), but start with the list head.
 */
// 从 list head 开始初始化一个 klist_iter
void klist_iter_init(struct klist *k, struct klist_iter *i)
{
	klist_iter_init_node(k, i, NULL);
}
EXPORT_SYMBOL_GPL(klist_iter_init);

/**
 * kref_get_unless_zero - Increment refcount for object unless it is zero.
 * @kref: object.
 *
 * Return non-zero if the increment succeeded. Otherwise return 0.
 *
 * This function is intended to simplify locking around refcounting for
 * objects that can be looked up from a lookup structure, and which are
 * removed from that lookup structure in the object destructor.
 * Operations on such objects require at least a read lock around
 * lookup + kref_get, and a write lock around kref_put + remove from lookup
 * structure. Furthermore, RCU implementations become extremely tricky.
 * With a lookup followed by a kref_get_unless_zero *with return value check*
 * locking in the kref_put path can be deferred to the actual removal from
 * the lookup structure and RCU lookups become trivial.
 */
static inline int __must_check kref_get_unless_zero(struct kref *kref){
	return refcount_inc_not_zero(&kref->refcount);
}

static inline __must_check bool refcount_inc_not_zero(refcount_t *r){
	return __refcount_add_not_zero(1, r, NULL);
}

// 该函数有阉割
static inline __must_check bool __refcount_add_not_zero(int i, refcount_t *r) {
	int old = refcount_read(r);

	do {
		if (!old)
			break;
	} while (!atomic_try_cmpxchg_relaxed(&r->refs, &old, old + i));
	// 原子操作，当第二次读变量的值和第一次old相同，则把新值赋值给该变量

	return old;
}

//======================================================================
【klist_iter的退出-klist_iter_exit】

/**
 * klist_iter_exit - Finish a list iteration.
 * @i: Iterator structure.
 *
 * Must be called when done iterating over list, as it decrements the
 * refcount of the current node. Necessary in case iteration exited before
 * the end of the list was reached, and always good form.
 */
// 迭代结束后，要用该接口
// 如果该 iter 指向某个 klist_node ，则该 klist_node 的引用计数要被减1
// 当迭代还没到达 list 的尾部就退出，调用该接口是一个好习惯
void klist_iter_exit(struct klist_iter *i)
{
	if (i->i_cur) {
		klist_put(i->i_cur, false);
		i->i_cur = NULL;
	}
}
EXPORT_SYMBOL_GPL(klist_iter_exit);

// 该 klist_put 的第二个参数为 false ，代表不用标记 klist_node->n_klist 的值为 DEAD
// klist_node->n_klist 的值标记为 DEAD 后，就不能通过  klist_node 来访问它的头klist了
static void klist_put(struct klist_node *n, bool false)
{
	struct klist *k = knode_klist(n);
	void (*put)(struct klist_node *) = k->put;

	spin_lock(&k->k_lock);
	if (!klist_dec_and_del(n)) 	// 引用计数减1，如果计数为0，则执行 klist_release 操作
		put = NULL;
	spin_unlock(&k->k_lock);
	if (put) 					// 只有在 klist_node 的计数减到0之后，才执行 klist 的 put 操作
		put(n);
}

//======================================================================
【通过 klist_node中的成员list_head来获取结构体地址klist_node】

// list_head* 指向 list_head 的地址，所以实参得是一个 list_head结构体 的地址
// klist_node实例的结构体中，包含一个 list_head结构体
// 所以可以通过 klist_node实例 中的成员 list_head的地址，进行 container_of 计算
// 来获取 klist_node实例 的地址
static struct klist_node *to_klist_node(struct list_head *n){
	return container_of(n, struct klist_node, n_node);
}

//======================================================================
【附件：修改void指针的值】

#include <stdio.h>

void main() {
	int hello = 12;
	void *tmp;
	printf("hello addr %p, hello value %d\n", &hello, hello);
	tmp = &hello;
	*(unsigned long *)&tmp +=1;	
	printf("tmp addr %p, tmp value %p\n", &tmp, tmp);
}

root@ubuntu:/home/kiss/study/test36# ./test 
hello addr 0x7ffee496d8ac, hello value 12
tmp addr 0x7ffee496d8b0, tmp value 0x7ffee496d8ad

//======================================================================
为什么要写它？
本人觉得它是内核开发出来的一个精巧的链表管理，搜索整个kernel 5.15，用的模块很少，但bus总线有用，还好提供了一个示例。