/*
* net/dst.c Protocol independent destination cache.
*
* Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
*
*/
#include <asm/system.h>
#include <asm/bitops.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/mm.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/netdevice.h>
#include <linux/skbuff.h>
#include <linux/init.h>
#include <net/dst.h>
/* Locking strategy:
* 1) Garbage collection state of dead destination cache
* entries is protected by dst_lock.
* 2) GC is run only from BH context, and is the only remover
* of entries.
* 3) Entries are added to the garbage list from both BH
* and non-BH context, so local BH disabling is needed.
* 4) All operations modify state, so a spinlock is used.
*/
static struct dst_entry *dst_garbage_list;
static atomic_t dst_total = ATOMIC_INIT(0);
static spinlock_t dst_lock = SPIN_LOCK_UNLOCKED;
static unsigned long dst_gc_timer_expires;
static unsigned long dst_gc_timer_inc = DST_GC_MAX;
static void dst_run_gc(unsigned long);
static struct timer_list dst_gc_timer =
{ data: DST_GC_MIN, function: dst_run_gc };
43 static void dst_run_gc(unsigned long dummy)
{
int delayed = 0;
struct dst_entry * dst, **dstp;
48 if (!spin_trylock(&dst_lock)) {
mod_timer(&dst_gc_timer, jiffies + HZ/10);
50 return;
}
del_timer(&dst_gc_timer);
dstp = &dst_garbage_list;
56 while ((dst = *dstp) != NULL) {
57 if (atomic_read(&dst->__refcnt)) {
dstp = &dst->next;
delayed++;
60 continue;
}
*dstp = dst->next;
dst_destroy(dst);
}
65 if (!dst_garbage_list) {
dst_gc_timer_inc = DST_GC_MAX;
67 goto out;
}
69 if ((dst_gc_timer_expires += dst_gc_timer_inc) > DST_GC_MAX)
dst_gc_timer_expires = DST_GC_MAX;
dst_gc_timer_inc += DST_GC_INC;
dst_gc_timer.expires = jiffies + dst_gc_timer_expires;
#if RT_CACHE_DEBUG >= 2
printk("dst_total: %d/%d %ld\n",
atomic_read(&dst_total), delayed, dst_gc_timer_expires);
#endif
add_timer(&dst_gc_timer);
out:
80 spin_unlock(&dst_lock);
}
83 static int dst_discard(struct sk_buff *skb)
{
kfree_skb(skb);
86 return 0;
}
89 static int dst_blackhole(struct sk_buff *skb)
{
kfree_skb(skb);
92 return 0;
}
95 void * dst_alloc(struct dst_ops * ops)
{
struct dst_entry * dst;
99 if (ops->gc && atomic_read(&ops->entries) > ops->gc_thresh) {
100 if (ops->gc())
101 return NULL;
}
dst = kmem_cache_alloc(ops->kmem_cachep, SLAB_ATOMIC);
104 if (!dst)
105 return NULL;
memset(dst, 0, ops->entry_size);
dst->ops = ops;
dst->lastuse = jiffies;
dst->input = dst_discard;
dst->output = dst_blackhole;
atomic_inc(&dst_total);
atomic_inc(&ops->entries);
113 return dst;
}
116 void __dst_free(struct dst_entry * dst)
{
118 spin_lock_bh(&dst_lock);
/* The first case (dev==NULL) is required, when
protocol module is unloaded.
*/
123 if (dst->dev == NULL || !(dst->dev->flags&IFF_UP)) {
dst->input = dst_discard;
dst->output = dst_blackhole;
}
dst->obsolete = 2;
dst->next = dst_garbage_list;
dst_garbage_list = dst;
130 if (dst_gc_timer_inc > DST_GC_INC) {
del_timer(&dst_gc_timer);
dst_gc_timer_inc = DST_GC_INC;
dst_gc_timer_expires = DST_GC_MIN;
dst_gc_timer.expires = jiffies + dst_gc_timer_expires;
add_timer(&dst_gc_timer);
}
138 spin_unlock_bh(&dst_lock);
}
141 void dst_destroy(struct dst_entry * dst)
{
struct neighbour *neigh = dst->neighbour;
struct hh_cache *hh = dst->hh;
dst->hh = NULL;
147 if (hh && atomic_dec_and_test(&hh->hh_refcnt))
kfree(hh);
150 if (neigh) {
dst->neighbour = NULL;
neigh_release(neigh);
}
atomic_dec(&dst->ops->entries);
157 if (dst->ops->destroy)
dst->ops->destroy(dst);
159 if (dst->dev)
dev_put(dst->dev);
atomic_dec(&dst_total);
kmem_cache_free(dst->ops->kmem_cachep, dst);
}
165 static int dst_dev_event(struct notifier_block *this, unsigned long event, void *ptr)
{
struct net_device *dev = ptr;
struct dst_entry *dst;
170 switch (event) {
171 case NETDEV_UNREGISTER:
172 case NETDEV_DOWN:
173 spin_lock_bh(&dst_lock);
174 for (dst = dst_garbage_list; dst; dst = dst->next) {
175 if (dst->dev == dev) {
/* Dirty hack. We did it in 2.2 (in __dst_free),
we have _very_ good reasons not to repeat
this mistake in 2.3, but we have no choice
now. _It_ _is_ _explicit_ _deliberate_
_race_ _condition_.
*/
if (event!=NETDEV_DOWN &&
!(dev->features & NETIF_F_DYNALLOC) &&
184 dst->output == dst_blackhole) {
dst->dev = &loopback_dev;
dev_put(dev);
dev_hold(&loopback_dev);
dst->output = dst_discard;
189 if (dst->neighbour && dst->neighbour->dev == dev) {
dst->neighbour->dev = &loopback_dev;
dev_put(dev);
dev_hold(&loopback_dev);
}
194 } else {
dst->input = dst_discard;
dst->output = dst_blackhole;
}
}
}
200 spin_unlock_bh(&dst_lock);
201 break;
}
203 return NOTIFY_DONE;
}
struct notifier_block dst_dev_notifier = {
dst_dev_event,
NULL,
0
};
212 void __init dst_init(void)
{
register_netdevice_notifier(&dst_dev_notifier);
}