Linux的idr機制(32叉樹)

Linux的idr機制(32叉樹)

一.結構體

1.idr結構體

struct idr {
	struct idr_layer __rcu *top;	//idr_layer頂層,32叉樹的根
	struct idr_layer *id_free;		//指向idr_layer的空閒鏈表
	int	layers;		//idr_layer的層數量
	int	id_free_cnt;	//idr_layer空閒鏈表中剩余的idr_layer個數
	spinlock_t	lock;
};

2.idr_layer結構體

struct idr_layer {
	unsigned long	bitmap;	//標記位圖,標記使用情況
	struct idr_layer __rcu	*ary[1<<IDR_BITS];		//子idr_layer數組ary[32]
	int	count;	//ary數組使用情況
	int	layer;	//層號
	struct rcu_head	rcu_head;
};

在32位系統中IDR_BITS的取值為5

#if BITS_PER_LONG == 32
	# define IDR_BITS 5
	# define IDR_FULL 0xfffffffful
	# define TOP_LEVEL_FULL (IDR_FULL >> 30)
#elif BITS_PER_LONG == 64
	# define IDR_BITS 6
	# define IDR_FULL 0xfffffffffffffffful
	# define TOP_LEVEL_FULL (IDR_FULL >> 62)
#else
	# error "BITS_PER_LONG is not 32 or 64"
#endif

二.idr的初始化

#define IDR_INIT(name)		\
{				\
	.top		= NULL,	\
	.id_free		= NULL,	\
	.layers 		= 0,	\
	.id_free_cnt	= 0,	\
	.lock		= __SPIN_LOCK_UNLOCKED(name.lock),	\
}
#define DEFINE_IDR(name)	struct idr name = IDR_INIT(name)

定義一個idr結構體並賦值

三.分配id

1.idr_pre_get

int idr_pre_get(struct idr *idp, gfp_t gfp_mask)
{
	while (idp->id_free_cnt < IDR_FREE_MAX) {	//IDR_FREE_MAX=14
		struct idr_layer *new;	//定義新的idr_layer結構體指針
		new = kmem_cache_zalloc(idr_layer_cache, gfp_mask);	//分配*new內存空間
		if (new == NULL)
			return (0);
		move_to_free_list(idp, new);	//-->move_to_free_list
	}
	return 1;
}
EXPORT_SYMBOL(idr_pre_get);

move_to_free_list

static void move_to_free_list(struct idr *idp, struct idr_layer *p)
{
	unsigned long flags;
	spin_lock_irqsave(&idp->lock, flags);
	__move_to_free_list(idp, p);	//-->__move_to_free_list
	spin_unlock_irqrestore(&idp->lock, flags);
}

__move_to_free_list

static void __move_to_free_list(struct idr *idp, struct idr_layer *p)
{
	p->ary[0] = idp->id_free;
	idp->id_free = p;
	idp->id_free_cnt++;
}

第一次循環結果

接著循環

再接著

一直這樣下去直到循環結束(14次)

2.idr_get_new和idr_get_new_above

idr_get_new

int idr_get_new(struct idr *idp, void *ptr, int *id)
{
	int rv;
	rv = idr_get_new_above_int(idp, ptr, 0);
	if (rv < 0)
		return _idr_rc_to_errno(rv);
	*id = rv;
	return 0;
}
EXPORT_SYMBOL(idr_get_new);

idr_get_new_above

int idr_get_new_above(struct idr *idp, void *ptr, int starting_id, int *id)
{
	int rv;
	rv = idr_get_new_above_int(idp, ptr, starting_id);
	if (rv < 0)
		return _idr_rc_to_errno(rv);
	*id = rv;
	return 0;
}
EXPORT_SYMBOL(idr_get_new_above);

兩個函數都會調用idr_get_new_above_int函數,差別在於starting_id不同

下面分情況討論,先以id為0走個過場

idr的top簡稱為根top,free簡稱為根free均為idr_layer指針類型,分別指向使用中和空閒idr_layer鏈表頭

static int idr_get_new_above_int(struct idr *idp, void *ptr, int starting_id)
{
	struct idr_layer *pa[MAX_LEVEL];	//MAX_LEVEL=7
	int id;
	id = idr_get_empty_slot(idp, starting_id, pa);	//-->idr_get_empty_slot
	if (id >= 0) {
		rcu_assign_pointer(pa[0]->ary[id & IDR_MASK],(struct idr_layer *)ptr
		//pa[0]->ary[0]=ptr 也就是idr_layer14->ary[0]=ptr
		pa[0]->count++;	//idr_layer14->count++
		idr_mark_full(pa, id);	//設置其位圖-->走完0過場的效果見圖c
	}
	return id;
}

idr_get_empty_slot

static int idr_get_empty_slot(struct idr *idp, int starting_id,struct idr_layer **pa)
{
	struct idr_layer *p, *new;
	int layers, v, id;
	unsigned long flags;

	id = starting_id;	//按常規出牌吧,假設這個為0
build_up:
	p = idp->top;	//根top指向的idr_layer NULL
	layers = idp->layers;	//獲取layers層數量(0)
	if (unlikely(!p)) {	//第一次運行idp->top=NULL,所以if條件為真,執行if分支的結果參考 圖A
		if (!(p = get_from_free_list(idp)))	//>>>1-->get_from_free_list 從根free中獲取一個idr_layer14
			return -1;
		p->layer = 0;	//指定idr_layer14的層號為0
		layers = 1;	//layers層數量設為1
	}
	//layers<6 && id>=2^(layers*5) 看需不需要增加層數 見圖B
	while ((layers < (MAX_LEVEL - 1)) && (id >= (1 << (layers*IDR_BITS)))) {
		layers++;	
		if (!p->count) {
			p->layer++;
			continue;	
		}
		if (!(new = get_from_free_list(idp))) {
			spin_lock_irqsave(&idp->lock, flags);
			for (new = p; p && p != idp->top; new = p) {
				p = p->ary[0];
				new->ary[0] = NULL;
				new->bitmap = new->count = 0;
				__move_to_free_list(idp, new);
			}
			spin_unlock_irqrestore(&idp->lock, flags);
			return -1;
		}
		new->ary[0] = p;
		new->count = 1;
		new->layer = layers-1;
		if (p->bitmap == IDR_FULL)
			__set_bit(0, &new->bitmap);
		p = new;
	}
	rcu_assign_pointer(idp->top, p);	//根top指向idr_layer14
	idp->layers = layers;	//設置更新idr->layers層數量
//----------------------------------------------分割線----------------------------------------------
//以上部分主要處理layer相關,以下部分主要處理id相關
	v = sub_alloc(idp, &id, pa);	//>>>2-->sub_alloc
	if (v == IDR_NEED_TO_GROW)	//IDR_NEED_TO_GROW=-2需要擴大
		goto build_up;
	return(v);
}

圖A：

圖B

>>>get_from_free_list 從idr空閒idr_layer鏈表中獲取第一個idr_layer

static struct idr_layer *get_from_free_list(struct idr *idp)
{
	struct idr_layer *p;	//定義一個idr_layer指針
	unsigned long flags;
	spin_lock_irqsave(&idp->lock, flags);
	if ((p = idp->id_free)) {	//根free獲取一個空閒idr_layer
		idp->id_free = p->ary[0];	//idr空閒鏈表指針指向第二個idr_layer
		idp->id_free_cnt--;	//idr的空閒idr_layer個數減1(14-1)
		p->ary[0] = NULL;	//斷開第一個idr_layer和第二個idr_layer的聯系
	}
	spin_unlock_irqrestore(&idp->lock, flags);
	return(p);
}

這裡先穿插一下32進制的計算,上面圖B中2^0,2^5,2^10,2^15,2^20,2^25可以(32=2^5)理解成32^0,32^1,32^2,32^3,32^3,32^4,32^5

那麼用32進制表達一個十進制數id可以套用一下公式

a的值屬於[0,31]

an的值如何獲得id/(32^n)即可,等同於id/(2^5^n)等同於id/((1<<5)^n)

an-1的值如何獲得id>>(5*(n-1))即可

>>>sub_alloc

static int sub_alloc(struct idr *idp, int *starting_id, struct idr_layer **pa)
{
	int n, m, sh;
	struct idr_layer *p, *new;
	int l, id, oid;
	unsigned long bm;

	id = *starting_id;
 restart:
	p = idp->top;	//根top
	l = idp->layers;	//l=1
	pa[l--] = NULL;	//p[1]=NULL;l=0
	while (1) {
		n = (id >> (IDR_BITS*l)) & IDR_MASK;	//計算對應的n值,屬於[0,31]
		bm = ~p->bitmap;	//取反位圖
		m = find_next_bit(&bm, IDR_SIZE, n);	//>>>1 find_next_bit 位圖中偏移量為n處查找'1'
		if (m == IDR_SIZE) {	//位圖滿了
			l++;
			oid = id;
			id = (id | ((1 << (IDR_BITS * l)) - 1)) + 1;
			if (id >= 1 << (idp->layers * IDR_BITS)) {
				*starting_id = id;
				return IDR_NEED_TO_GROW;
			}
			p = pa[l];
			BUG_ON(!p);
			sh = IDR_BITS * (l + 1);
			if (oid >> sh == id >> sh)
				continue;
			else
				goto restart;
		}
		if (m != n) {	//期望的n值被占用,但可找到可用的m值
			sh = IDR_BITS*l;
			id = ((id >> sh) ^ n ^ m) << sh;	//>>>2 重新計算id值
		}
		if ((id >= MAX_ID_BIT) || (id < 0))
			return IDR_NOMORE_SPACE;
		if (l == 0)	//l==0跳出while循環
			break;
		if (!p->ary[m]) {
			new = get_from_free_list(idp);
			if (!new)
				return -1;
			new->layer = l-1;
			rcu_assign_pointer(p->ary[m], new);
			p->count++;
		}
		pa[l--] = p;
		p = p->ary[m];
	}

	pa[l] = p;	//pa[0]=p 也就是idr_layer14
	return id;
}

>>>find_next_bit

#define find_next_bit(p,sz,off)	_find_next_bit_le(p,sz,off)		//>>_find_next_bit_le

該宏的意思是在p指向的(大小為sz的)位圖表中的第off個位置開始找尋可用(為"1")的格子,找到返回該位

_find_next_bit_le是匯編代碼實現的定義在/arch/arm/lib/findbit.S

ENTRY(_find_next_bit_le)
		teq	r1, #0
		beq	3b
		ands	ip, r2, #7
		beq	1b			@ If new byte, goto old routine
 ARM(		ldrb	r3, [r0, r2, lsr #3]	)
 THUMB(		lsr	r3, r2, #3		)
 THUMB(		ldrb	r3, [r0, r3]		)
		movs	r3, r3, lsr ip		@ shift off unused bits
		bne	.L_found
		orr	r2, r2, #7		@ if zero, then no bits here
		add	r2, r2, #1		@ align bit pointer
		b	2b			@ loop for next bit
ENDPROC(_find_next_bit_le)

.L_found找到合適的跳轉

.L_found:
#if __LINUX_ARM_ARCH__ >= 5
		rsb	r0, r3, #0
		and	r3, r3, r0
		clz	r3, r3
		rsb	r3, r3, #31
		add	r0, r2, r3
#else
		tst	r3, #0x0f
		addeq	r2, r2, #4
		movne	r3, r3, lsl #4
		tst	r3, #0x30
		addeq	r2, r2, #2
		movne	r3, r3, lsl #2
		tst	r3, #0x40
		addeq	r2, r2, #1
		mov	r0, r2
#endif
		cmp	r1, r0			@ Clamp to maxbit
		movlo	r0, r1
		mov	pc, lr

>>>id值的計算的補充說明

首先前面n的取值n = (id >> (IDR_BITS*l)) & IDR_MASK;

IDR_MASK的定義#define IDR_MASK ((1 << IDR_BITS)-1)也就是說IDR_MASK=31等於2進制的1,1111b

所以&IDR_MASK只是框定n值落在0~31之間,掩碼作用

那麼不出意外的話n = (id >> (IDR_BITS*l))

接著

sh = IDR_BITS*l;
id = ((id >> sh) ^ n ^ m) << sh;

帶入表達式中

id=((id >> IDR_BITS*l) ^ (id >> (IDR_BITS*l)) ^ m) << IDR_BITS*l;

異或的操作是相同為1,不同為0,結合起來化簡得

id = ((1 ^ m) << sh=m<<sh

圖C

^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_^_

 已經借用id0走了過場,下面分析下其他情況

static int idr_get_new_above_int(struct idr *idp, void *ptr, int starting_id)
{
	struct idr_layer *pa[MAX_LEVEL];	//定義父idr_layer數組
	int id;
	id = idr_get_empty_slot(idp, starting_id, pa);	//獲取id
	if (id >= 0) {
		rcu_assign_pointer(pa[0]->ary[id & IDR_MASK],(struct idr_layer *)ptr);
		//pa[0]->ary[id]=ptr
		pa[0]->count++;	//idr_layer->count++
		idr_mark_full(pa, id);	//標記id位圖
	}
	return id;
}

static int idr_get_empty_slot(struct idr *idp, int starting_id,struct idr_layer **pa)
{
	struct idr_layer *p, *new;
	int layers, v, id;
	unsigned long flags;

	id = starting_id;
build_up:
	p = idp->top;	//獲取根top
	layers = idp->layers;	//獲取層數量 layers=1
	if (unlikely(!p)) {	//FALSE
		if (!(p = get_from_free_list(idp)))
			return -1;
		p->layer = 0;
		layers = 1;
	}

	while ((layers < 6) && (id >= (1 << (layers*5)))) {	//參考圖B,如果id值超過或等於對應層所能容納的最大數,則進入循環
		layers++;		//增加層數量
		if (!p->count) {	//0~31沒使用,直接使用32就屬於這種情況
			p->layer++;	//由於32需要添加1層的,所以之前的層的層號需要+1
			continue;	 //層數量也需要加1	
		}
		if (!(new = get_from_free_list(idp))) {	//空閒鏈表中獲取新的idr_layer
			spin_lock_irqsave(&idp->lock, flags);	//分配失敗,--空閒idr_layer鏈表缺貨
			for (new = p; p && p != idp->top; new = p) {	//p指針還原
					p = p->ary[0];
				new->ary[0] = NULL;
				new->bitmap = new->count = 0;
				__move_to_free_list(idp, new);	//分配更多空閒鏈表
			}
			spin_unlock_irqrestore(&idp->lock, flags);
			return -1;
		}
		new->ary[0] = p;	//新的idr_layer->ary[0]指向舊的idr_layer
		new->count = 1;	//新的idr_layer計數加1
		new->layer = layers-1;	//設置新的idr_layer的層號
		if (p->bitmap == IDR_FULL)	//若舊的(葉子)idr_layer的id全用過了
			__set_bit(0, &new->bitmap);	//那麼標記下新(父)idr_layer位圖的第0位
		p = new;	//根top指向新的idr_layer
	}
	rcu_assign_pointer(idp->top, p);	//設置根top
	idp->layers = layers;	//更新層數量
	v = sub_alloc(idp, &id, pa);	//獲取id
	if (v == IDR_NEED_TO_GROW)	//該層id號全用完了,必須擴大idr_layer層數量
		goto build_up;
	return(v);
}

static int sub_alloc(struct idr *idp, int *starting_id, struct idr_layer **pa)
{
	int n, m, sh;
	struct idr_layer *p, *new;
	int l, id, oid;
	unsigned long bm;

	id = *starting_id;
 restart:
	p = idp->top;	//獲取根top
	l = idp->layers;	//獲取層數量l=1
	pa[l--] = NULL;	//pa[1]=NULL,l=0
	while (1) {
		n = (id >> (5*l)) & IDR_MASK;	//n做處理 屬於[0,31]
		bm = ~p->bitmap;	//位圖取反
		m = find_next_bit(&bm, IDR_SIZE, n);	//查找n開始能用的位
		if (m == IDR_SIZE) {	//id表滿了
			l++;	層數+1
			oid = id;
			id = (id | ((1 << (5 * l)) - 1)) + 1;	//id或上掩碼再+1	

			if (id >= 1 << (idp->layers * 5)) {	//需要添加層
				*starting_id = id;
				return IDR_NEED_TO_GROW;
			}
			p = pa[l];
			BUG_ON(!p);
			sh = 5 * (l + 1);
			if (oid >> sh == id >> sh)
				continue;
			else
				goto restart;
		}
		if (m != n) {	//期望id給用但有可用id
			sh = 5*l;
			id = ((id >> sh) ^ n ^ m) << sh;	//id設置為可用id
		}
		if ((id >= MAX_ID_BIT) || (id < 0))
			return IDR_NOMORE_SPACE;
		if (l == 0)	//一層層循環計算直到到達葉子處l才為0
			break;
		if (!p->ary[m]) {	//葉子m為空
			new = get_from_free_list(idp);	//從空閒鏈表拿一個idr_layer
			if (!new)
				return -1;
			new->layer = l-1;	//設置新鏈表層數
			rcu_assign_pointer(p->ary[m], new);	//葉子m指向新鏈表
			p->count++;	//使用計數加1
		}
		pa[l--] = p;	//pa[大]=節點
		p = p->ary[m];	//p=節點->葉子m	
	}
	pa[l] = p;		//pa[小]=葉子
	return id;
}

來個效果圖id=4吧

id=32情況(idr_layer13的位圖1標記錯了)

1024情況

四.查找id
1.idr_find

void *idr_find(struct idr *idp, int id)
{
	int n;
	struct idr_layer *p;

	p = rcu_dereference_raw(idp->top);	//獲取根top
	if (!p)
		return NULL;
	n = (p->layer+1) * IDR_BITS;	//計算最外層的n值
	id &= MAX_ID_MASK;
	if (id >= (1 << n))
		return NULL;
	BUG_ON(n == 0);
	while (n > 0 && p) {	//循環一層層查找
		n -= IDR_BITS;
		BUG_ON(n != p->layer*IDR_BITS);
		p = rcu_dereference_raw(p->ary[(id >> n) & IDR_MASK]);	//一次獲取an ... a0 
 	}
	return((void *)p);
}
EXPORT_SYMBOL(idr_find);

前面講過32進制的id值算法

當構建完idr機制之後

id=top->ary[an]->ary[a(n-1)]->....->ary[a0]來獲得

 借助圖片分析下(idr_layer13的位圖標記有錯)

五idr操作

1. idr_remove  idr_remove_all 移除

void idr_remove(struct idr *idp, int id)
{
	struct idr_layer *p;
	struct idr_layer *to_free;
	id &= MAX_ID_MASK;

	sub_remove(idp, (idp->layers - 1) * IDR_BITS, id);
	if (idp->top && idp->top->count == 1 && (idp->layers > 1) &&
	    idp->top->ary[0]) {
		to_free = idp->top;
		p = idp->top->ary[0];
		rcu_assign_pointer(idp->top, p);
		--idp->layers;
		to_free->bitmap = to_free->count = 0;
		free_layer(to_free);
	}
	while (idp->id_free_cnt >= IDR_FREE_MAX) {
		p = get_from_free_list(idp);
		kmem_cache_free(idr_layer_cache, p);
	}
	return;
}
EXPORT_SYMBOL(idr_remove);

移除全部

void idr_remove_all(struct idr *idp)
{
	int n, id, max;
	int bt_mask;
	struct idr_layer *p;
	struct idr_layer *pa[MAX_LEVEL];
	struct idr_layer **paa = &pa[0];

	n = idp->layers * IDR_BITS;
	p = idp->top;
	rcu_assign_pointer(idp->top, NULL);
	max = 1 << n;

	id = 0;
	while (id < max) {
		while (n > IDR_BITS && p) {
			n -= IDR_BITS;
			*paa++ = p;
			p = p->ary[(id >> n) & IDR_MASK];
		}

		bt_mask = id;
		id += 1 << n;
		/* Get the highest bit that the above add changed from 0->1. */
		while (n < fls(id ^ bt_mask)) {
			if (p)
				free_layer(p);
			n += IDR_BITS;
			p = *--paa;
		}
	}
	idp->layers = 0;
}
EXPORT_SYMBOL(idr_remove_all);

2.idr_replace 替換

void *idr_replace(struct idr *idp, void *ptr, int id)
{
	int n;
	struct idr_layer *p, *old_p;

	p = idp->top;
	if (!p)
		return ERR_PTR(-EINVAL);

	n = (p->layer+1) * IDR_BITS;

	id &= MAX_ID_MASK;

	if (id >= (1 << n))
		return ERR_PTR(-EINVAL);

	n -= IDR_BITS;
	while ((n > 0) && p) {
		p = p->ary[(id >> n) & IDR_MASK];
		n -= IDR_BITS;
	}

	n = id & IDR_MASK;
	if (unlikely(p == NULL || !test_bit(n, &p->bitmap)))
		return ERR_PTR(-ENOENT);

	old_p = p->ary[n];
	rcu_assign_pointer(p->ary[n], ptr);

	return old_p;
}
EXPORT_SYMBOL(idr_replace);

六.idr空閒鏈表的銷毀

idr_destroy

void idr_destroy(struct idr *idp)
{
	while (idp->id_free_cnt) {
		struct idr_layer *p = get_from_free_list(idp);
		kmem_cache_free(idr_layer_cache, p);
	}
}
EXPORT_SYMBOL(idr_destroy);

七.用法

1.api函數

void *idr_find(struct idr *idp, int id);	//查找id對應的指針
int idr_pre_get(struct idr *idp, gfp_t gfp_mask);	//分配idr_layer空閒鏈表
int idr_get_new(struct idr *idp, void *ptr, int *id);	//獲取id,捆綁指針ptr
int idr_get_new_above(struct idr *idp, void *ptr, int starting_id, int *id);	//起始數值獲取id,捆綁指針ptr
int idr_for_each(struct idr *idp,int (*fn)(int id, void *p, void *data), void *data);
void *idr_get_next(struct idr *idp, int *nextid);	
void *idr_replace(struct idr *idp, void *ptr, int id);	//替換id捆綁的指針
void idr_remove(struct idr *idp, int id);	//移除id
void idr_remove_all(struct idr *idp);	//移除所有id
void idr_destroy(struct idr *idp);	//銷毀idr_layer空閒鏈表
void idr_init(struct idr *idp);	//初始化idr

2.大致用法

1.idr_init聲明設置idr

2.idr_pre_get分配空閒idr_layer鏈表

3.id_get_new/idr_get_new_above分配id並將id與指針ptr捆綁

4.利用idr_find根據id獲取指針ptr

5.idr_remove/idr_remove_all移除分配的id

6.idr_destroy銷毀空閒idr_layer鏈表

 7.idr_replace替換id