上节主要是分析了通用邻居层邻居项的垃圾回收机制，这一节主要是分析邻居项的创建、查找、删除等相关的函数，这一节只是介绍函数功能，而没有涉及状态机、通用邻居层的架构等。比如邻居项删除函数neigh_destroy，而这个函数主要是通过垃圾回收机制的调用才会执行删除操作；而对于邻居项创建函数neigh_create，在arp协议下，则在路由缓存与邻居项绑定时会触发调用neigh_create创建邻居项，或者接收到arp reply或者arp request数据包时会触发调用neigh_create创建邻居项。本节只是分析这些函数的实现流程，而不分析这些函数在arp协议中或者其他邻居协议中被其他逻辑子层调用的过。
首先介绍neigh_create函数，分析如下
1、调用neigh_alloc，申请一个邻居项缓存
2、设置邻居项的primary_key、dev值，并增加dev的引用计数
3、调用邻居表的constructor，初始化邻居表协议相关的参数
4、如果存在n->parms->neigh_setup，则调用neigh_setup进行初始化(对于arp，该函数
为NULL)
5、设置confirmed时间
6、如果创建的邻居项总数超过了邻居表定义的最大值，则调用neigh_hash_grow扩充
邻居项的hash表容量
7、在将创建的邻居项插入到邻居项hash表之前，需要再次查找hash表里的邻居项:
a)如果该邻居项已经存在hash表里，则增加对该邻居项的引用计数，并释放已
申请的邻居项缓存
b)如果该邻居项还没有在hash表里，则将该邻居项插入的hash表里。
struct neighbour*neigh_create(struct neigh_table *tbl, const void *pkey,
struct net_device *dev)
{
u32 hash_val;
int key_len = tbl->key_len;
int error;
struct neighbour *n1, *rc, *n =neigh_alloc(tbl);
if (!n) {
rc = ERR_PTR(-ENOBUFS);
goto out;
}
memcpy(n->primary_key, pkey, key_len);
n->dev = dev;
dev_hold(dev);
/* Protocol specific setup. */
if (tbl->constructor && (error = tbl->constructor(n)) < 0) {
rc = ERR_PTR(error);
goto out_neigh_release;
}
/* Device specific setup. */
if (n->parms->neigh_setup&&
(error = n->parms->neigh_setup(n)) < 0) {
rc = ERR_PTR(error);
goto out_neigh_release;
}
n->confirmed = jiffies -(n->parms->base_reachable_time << 1);
write_lock_bh(&tbl->lock);
if (atomic_read(&tbl->entries)> (tbl->hash_mask + 1))
neigh_hash_grow(tbl,(tbl->hash_mask + 1) << 1);
hash_val = tbl->hash(pkey, dev) &tbl->hash_mask;
if (n->parms->dead) {
rc = ERR_PTR(-EINVAL);
goto out_tbl_unlock;
}
for (n1 = tbl->hash_buckets[hash_val];n1; n1 = n1->next) {
if (dev == n1->dev &&!memcmp(n1->primary_key, pkey, key_len)) {
neigh_hold(n1);
rc = n1;
goto out_tbl_unlock;
}
}
n->next =tbl->hash_buckets[hash_val];
tbl->hash_buckets[hash_val] = n;
n->dead = 0;
neigh_hold(n);
write_unlock_bh(&tbl->lock);
NEIGH_PRINTK2("neigh %p iscreated.\n", n);
rc = n;
out:
return rc;
out_tbl_unlock:
write_unlock_bh(&tbl->lock);
out_neigh_release:
neigh_release(n);
goto out;
}
由于该函数调用了neigh_alloc，下面我们分析函数neigh_alloc
从邻居表的slab缓存里，申请一个邻居项，并进行通用邻居层的初始化
static structneighbour *neigh_alloc(struct neigh_table *tbl)
{
struct neighbour *n = NULL;
unsigned long now = jiffies;
int entries;
entries =atomic_inc_return(&tbl->entries) – 1;
if (entries >= tbl->gc_thresh3 ||
(entries >= tbl->gc_thresh2 &&
time_after(now, tbl->last_flush + 5 * HZ))) {
if (!neigh_forced_gc(tbl)&&
entries >= tbl->gc_thresh3)
goto out_entries;
}
n =kmem_cache_zalloc(tbl->kmem_cachep, GFP_ATOMIC);
if (!n)
goto out_entries;
skb_queue_head_init(&n->arp_queue);
rwlock_init(&n->lock);
n->updated = n->used = now;
n->nud_state = NUD_NONE;
n->output = neigh_blackhole;
n->parms =neigh_parms_clone(&tbl->parms);//直接从邻居表中克隆
setup_timer(&n->timer,neigh_timer_handler, (unsigned long)n);//创建定时器
NEIGH_CACHE_STAT_INC(tbl, allocs);
n->tbl = tbl;
atomic_set(&n->refcnt, 1);
n->dead = 1;
out:
return n;
out_entries:
atomic_dec(&tbl->entries);
goto out;
}
下面分析通用邻居项的查找函数，主要是有neigh_lookup、neigh_lookup_nodev、__neigh_lookup、__neigh_lookup_errno
函数neigh_lookup的功能是根据关键字net_device与pkey在邻居表里查找一个邻居项
1、根据关键字net_device与pkey，计算hash值
2、通过hash值，在邻居项hash数组中，找到指定的hash链表
3、遍历指定的hash链表，比较net_device与pkey值，查找符合条件的邻居项。
struct neighbour*neigh_lookup(struct neigh_table *tbl, const void *pkey,
struct net_device *dev)
{
struct neighbour *n;
int key_len = tbl->key_len;
u32 hash_val;
NEIGH_CACHE_STAT_INC(tbl, lookups);
read_lock_bh(&tbl->lock);
hash_val = tbl->hash(pkey, dev);
for (n = tbl->hash_buckets[hash_val& tbl->hash_mask]; n; n = n->next) {
if (dev == n->dev &&!memcmp(n->primary_key, pkey, key_len)) {
neigh_hold(n);
NEIGH_CACHE_STAT_INC(tbl,hits);
break;
}
}
read_unlock_bh(&tbl->lock);
return n;
}
函数neigh_lookup_nodev功能是根据关键字net与pkey在邻居表里查找一个邻居项
该函数的功能与neigh_lookup类似
struct neighbour*neigh_lookup_nodev(struct neigh_table *tbl, struct net *net,
const void *pkey)
{
struct neighbour *n;
int key_len = tbl->key_len;
u32 hash_val;
NEIGH_CACHE_STAT_INC(tbl, lookups);
read_lock_bh(&tbl->lock);
hash_val = tbl->hash(pkey, NULL);
for (n = tbl->hash_buckets[hash_val& tbl->hash_mask]; n; n = n->next) {
if (!memcmp(n->primary_key,pkey, key_len) &&
net_eq(dev_net(n->dev), net)) {
neigh_hold(n);
NEIGH_CACHE_STAT_INC(tbl,hits);
break;
}
}
read_unlock_bh(&tbl->lock);
return n;
}
函数__neigh_lookup的功能是同时实现查找邻居项与决定是否创建邻居项，该函数通常被其他协议子层调用。
static inlinestruct neighbour *
__neigh_lookup(structneigh_table *tbl, const void *pkey, struct net_device *dev, int creat)
{
struct neighbour *n = neigh_lookup(tbl,pkey, dev);
if (n || !creat)
return n;
n = neigh_create(tbl, pkey, dev);
return IS_ERR(n) ? NULL : n;
}
该函数与函数__neigh_lookup相比，其在查找不到邻居项的情况下，即会默认创建一个新的邻居项。
static inlinestruct neighbour *
__neigh_lookup_errno(structneigh_table *tbl, const void *pkey,
struct net_device *dev)
{
struct neighbour *n = neigh_lookup(tbl,pkey, dev);
if (n)
return n;
return neigh_create(tbl, pkey, dev);
}
下面分析一下邻居项hash表容量的扩容函数neigh_hash_grow
功能:邻居项散列表的扩容
1、为邻居表申请一个hash bucket数组指针，
2、然后将原hash bucket中所有的邻居项，重新计算hash值后，存放在新的hash链表
中
3、释放原hash bucket数组所占用的缓存。
static voidneigh_hash_grow(struct neigh_table *tbl, unsigned long new_entries)
{
struct neighbour **new_hash, **old_hash;
unsigned int i, new_hash_mask,old_entries;
NEIGH_CACHE_STAT_INC(tbl, hash_grows);
BUG_ON(!is_power_of_2(new_entries));
new_hash = neigh_hash_alloc(new_entries);
if (!new_hash)
return;
old_entries = tbl->hash_mask + 1;
new_hash_mask = new_entries – 1;
old_hash = tbl->hash_buckets;
get_random_bytes(&tbl->hash_rnd,sizeof(tbl->hash_rnd));
for (i = 0; i < old_entries; i++) {
struct neighbour *n, *next;
for (n = old_hash[i]; n; n = next){
unsigned int hash_val =tbl->hash(n->primary_key, n->dev);
hash_val &=new_hash_mask;
next = n->next;
n->next =new_hash[hash_val];
new_hash[hash_val] = n;
}
}
tbl->hash_buckets = new_hash;
tbl->hash_mask = new_hash_mask;
neigh_hash_free(old_hash, old_entries);
}
下面分析邻居项的删除函数neigh_destroy
1、判断该邻居项是否可以删除 dead==1，则可以删除
2、删除该邻居项的定时器
3、对于可以删除的邻居项， 将邻居项所关联的所有二层缓存头的hh_output设置
为neigh_blackhole，即直接丢弃数据包
4、调用skb_queue_purge，丢弃队列中所有待发送的数据包
5、取消对net_dev、neigh_parms的引用
6、最后调用kmem_cache_free，将邻居项的缓存释放给邻居表的slab缓存中。
voidneigh_destroy(struct neighbour *neigh)
{
struct hh_cache *hh;
NEIGH_CACHE_STAT_INC(neigh->tbl,destroys);
if (!neigh->dead) {
printk(KERN_WARNING
"Destroying alive neighbour%p\n", neigh);
dump_stack();
return;
}
if (neigh_del_timer(neigh))
printk(KERN_WARNING"Impossible event.\n");
while ((hh = neigh->hh) != NULL) {
neigh->hh = hh->hh_next;
hh->hh_next = NULL;
write_seqlock_bh(&hh->hh_lock);
hh->hh_output =neigh_blackhole;
write_sequnlock_bh(&hh->hh_lock);
if(atomic_dec_and_test(&hh->hh_refcnt))
kfree(hh);
}
skb_queue_purge(&neigh->arp_queue);
dev_put(neigh->dev);
neigh_parms_put(neigh->parms);
NEIGH_PRINTK2("neigh %p isdestroyed.\n", neigh);
atomic_dec(&neigh->tbl->entries);
kmem_cache_free(neigh->tbl->kmem_cachep,neigh);
}
函数neigh_release是neigh_destroy的包裹函数，增加了对neigh->refcnt的判断。
如果邻居项的引用计数为1时，则调用neigh_destroy释放该邻居项所对应的缓存
static inlinevoid neigh_release(struct neighbour *neigh)
{
if(atomic_dec_and_test(&neigh->refcnt))
neigh_destroy(neigh);
}
下面分析一下一些小函数
neigh_clone，该函数只是增加对邻居项的引用，而并不clone一个邻居项
static inlinestruct neighbour * neigh_clone(struct neighbour *neigh)
{
if (neigh)
atomic_inc(&neigh->refcnt);
return neigh;
}
设置邻居项的confirmed值为当前时间
static inlinevoid neigh_confirm(struct neighbour *neigh)
{
if (neigh)
neigh->confirmed = jiffies;
}
/*
从hh_cache中拷贝二层头部，并调用hh_cache->output输出skb
*/
static inlineint neigh_hh_output(struct hh_cache *hh, struct sk_buff *skb)
{
unsigned seq;
int hh_len;
do {
int hh_alen;
seq =read_seqbegin(&hh->hh_lock);
hh_len = hh->hh_len;
hh_alen = HH_DATA_ALIGN(hh_len);
memcpy(skb->data – hh_alen,hh->hh_data, hh_alen);
} while(read_seqretry(&hh->hh_lock, seq));
skb_push(skb, hh_len);
return hh->hh_output(skb);
}
游手好闲会使人心智生锈
