aha/net/core/dst.c
Eric Dumazet ef711cf1d1 net: speedup dst_release()
During tbench/oprofile sessions, I found that dst_release() was in third position.

CPU: Core 2, speed 2999.68 MHz (estimated)
Counted CPU_CLK_UNHALTED events (Clock cycles when not halted) with a unit mask of 0x00 (Unhalted core cycles) count 100000
samples  %        symbol name
483726    9.0185  __copy_user_zeroing_intel
191466    3.5697  __copy_user_intel
185475    3.4580  dst_release
175114    3.2648  ip_queue_xmit
153447    2.8608  tcp_sendmsg
108775    2.0280  tcp_recvmsg
102659    1.9140  sysenter_past_esp
101450    1.8914  tcp_current_mss
95067     1.7724  __copy_from_user_ll
86531     1.6133  tcp_transmit_skb

Of course, all CPUS fight on the dst_entry associated with 127.0.0.1 

Instead of first checking the refcount value, then decrement it,
we use atomic_dec_return() to help CPU to make the right memory transaction
(ie getting the cache line in exclusive mode)

dst_release() is now at the fifth position, and tbench a litle bit faster ;)

CPU: Core 2, speed 3000.1 MHz (estimated)
Counted CPU_CLK_UNHALTED events (Clock cycles when not halted) with a unit mask of 0x00 (Unhalted core cycles) count 100000
samples  %        symbol name
647107    8.8072  __copy_user_zeroing_intel
258840    3.5229  ip_queue_xmit
258302    3.5155  __copy_user_intel
209629    2.8531  tcp_sendmsg
165632    2.2543  dst_release
149232    2.0311  tcp_current_mss
147821    2.0119  tcp_recvmsg
137893    1.8767  sysenter_past_esp
127473    1.7349  __copy_from_user_ll
121308    1.6510  ip_finish_output
118510    1.6129  tcp_transmit_skb
109295    1.4875  tcp_v4_rcv

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2008-11-14 00:53:54 -08:00

349 lines
8.1 KiB
C

/*
* net/core/dst.c Protocol independent destination cache.
*
* Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
*
*/
#include <linux/bitops.h>
#include <linux/errno.h>
#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/workqueue.h>
#include <linux/mm.h>
#include <linux/module.h>
#include <linux/netdevice.h>
#include <linux/skbuff.h>
#include <linux/string.h>
#include <linux/types.h>
#include <net/net_namespace.h>
#include <net/dst.h>
/*
* Theory of operations:
* 1) We use a list, protected by a spinlock, to add
* new entries from both BH and non-BH context.
* 2) In order to keep spinlock held for a small delay,
* we use a second list where are stored long lived
* entries, that are handled by the garbage collect thread
* fired by a workqueue.
* 3) This list is guarded by a mutex,
* so that the gc_task and dst_dev_event() can be synchronized.
*/
#if RT_CACHE_DEBUG >= 2
static atomic_t dst_total = ATOMIC_INIT(0);
#endif
/*
* We want to keep lock & list close together
* to dirty as few cache lines as possible in __dst_free().
* As this is not a very strong hint, we dont force an alignment on SMP.
*/
static struct {
spinlock_t lock;
struct dst_entry *list;
unsigned long timer_inc;
unsigned long timer_expires;
} dst_garbage = {
.lock = __SPIN_LOCK_UNLOCKED(dst_garbage.lock),
.timer_inc = DST_GC_MAX,
};
static void dst_gc_task(struct work_struct *work);
static void ___dst_free(struct dst_entry * dst);
static DECLARE_DELAYED_WORK(dst_gc_work, dst_gc_task);
static DEFINE_MUTEX(dst_gc_mutex);
/*
* long lived entries are maintained in this list, guarded by dst_gc_mutex
*/
static struct dst_entry *dst_busy_list;
static void dst_gc_task(struct work_struct *work)
{
int delayed = 0;
int work_performed = 0;
unsigned long expires = ~0L;
struct dst_entry *dst, *next, head;
struct dst_entry *last = &head;
#if RT_CACHE_DEBUG >= 2
ktime_t time_start = ktime_get();
struct timespec elapsed;
#endif
mutex_lock(&dst_gc_mutex);
next = dst_busy_list;
loop:
while ((dst = next) != NULL) {
next = dst->next;
prefetch(&next->next);
if (likely(atomic_read(&dst->__refcnt))) {
last->next = dst;
last = dst;
delayed++;
continue;
}
work_performed++;
dst = dst_destroy(dst);
if (dst) {
/* NOHASH and still referenced. Unless it is already
* on gc list, invalidate it and add to gc list.
*
* Note: this is temporary. Actually, NOHASH dst's
* must be obsoleted when parent is obsoleted.
* But we do not have state "obsoleted, but
* referenced by parent", so it is right.
*/
if (dst->obsolete > 1)
continue;
___dst_free(dst);
dst->next = next;
next = dst;
}
}
spin_lock_bh(&dst_garbage.lock);
next = dst_garbage.list;
if (next) {
dst_garbage.list = NULL;
spin_unlock_bh(&dst_garbage.lock);
goto loop;
}
last->next = NULL;
dst_busy_list = head.next;
if (!dst_busy_list)
dst_garbage.timer_inc = DST_GC_MAX;
else {
/*
* if we freed less than 1/10 of delayed entries,
* we can sleep longer.
*/
if (work_performed <= delayed/10) {
dst_garbage.timer_expires += dst_garbage.timer_inc;
if (dst_garbage.timer_expires > DST_GC_MAX)
dst_garbage.timer_expires = DST_GC_MAX;
dst_garbage.timer_inc += DST_GC_INC;
} else {
dst_garbage.timer_inc = DST_GC_INC;
dst_garbage.timer_expires = DST_GC_MIN;
}
expires = dst_garbage.timer_expires;
/*
* if the next desired timer is more than 4 seconds in the future
* then round the timer to whole seconds
*/
if (expires > 4*HZ)
expires = round_jiffies_relative(expires);
schedule_delayed_work(&dst_gc_work, expires);
}
spin_unlock_bh(&dst_garbage.lock);
mutex_unlock(&dst_gc_mutex);
#if RT_CACHE_DEBUG >= 2
elapsed = ktime_to_timespec(ktime_sub(ktime_get(), time_start));
printk(KERN_DEBUG "dst_total: %d delayed: %d work_perf: %d"
" expires: %lu elapsed: %lu us\n",
atomic_read(&dst_total), delayed, work_performed,
expires,
elapsed.tv_sec * USEC_PER_SEC + elapsed.tv_nsec / NSEC_PER_USEC);
#endif
}
int dst_discard(struct sk_buff *skb)
{
kfree_skb(skb);
return 0;
}
EXPORT_SYMBOL(dst_discard);
void * dst_alloc(struct dst_ops * ops)
{
struct dst_entry * dst;
if (ops->gc && atomic_read(&ops->entries) > ops->gc_thresh) {
if (ops->gc(ops))
return NULL;
}
dst = kmem_cache_zalloc(ops->kmem_cachep, GFP_ATOMIC);
if (!dst)
return NULL;
atomic_set(&dst->__refcnt, 0);
dst->ops = ops;
dst->lastuse = jiffies;
dst->path = dst;
dst->input = dst->output = dst_discard;
#if RT_CACHE_DEBUG >= 2
atomic_inc(&dst_total);
#endif
atomic_inc(&ops->entries);
return dst;
}
static void ___dst_free(struct dst_entry * dst)
{
/* The first case (dev==NULL) is required, when
protocol module is unloaded.
*/
if (dst->dev == NULL || !(dst->dev->flags&IFF_UP)) {
dst->input = dst->output = dst_discard;
}
dst->obsolete = 2;
}
void __dst_free(struct dst_entry * dst)
{
spin_lock_bh(&dst_garbage.lock);
___dst_free(dst);
dst->next = dst_garbage.list;
dst_garbage.list = dst;
if (dst_garbage.timer_inc > DST_GC_INC) {
dst_garbage.timer_inc = DST_GC_INC;
dst_garbage.timer_expires = DST_GC_MIN;
cancel_delayed_work(&dst_gc_work);
schedule_delayed_work(&dst_gc_work, dst_garbage.timer_expires);
}
spin_unlock_bh(&dst_garbage.lock);
}
struct dst_entry *dst_destroy(struct dst_entry * dst)
{
struct dst_entry *child;
struct neighbour *neigh;
struct hh_cache *hh;
smp_rmb();
again:
neigh = dst->neighbour;
hh = dst->hh;
child = dst->child;
dst->hh = NULL;
if (hh && atomic_dec_and_test(&hh->hh_refcnt))
kfree(hh);
if (neigh) {
dst->neighbour = NULL;
neigh_release(neigh);
}
atomic_dec(&dst->ops->entries);
if (dst->ops->destroy)
dst->ops->destroy(dst);
if (dst->dev)
dev_put(dst->dev);
#if RT_CACHE_DEBUG >= 2
atomic_dec(&dst_total);
#endif
kmem_cache_free(dst->ops->kmem_cachep, dst);
dst = child;
if (dst) {
int nohash = dst->flags & DST_NOHASH;
if (atomic_dec_and_test(&dst->__refcnt)) {
/* We were real parent of this dst, so kill child. */
if (nohash)
goto again;
} else {
/* Child is still referenced, return it for freeing. */
if (nohash)
return dst;
/* Child is still in his hash table */
}
}
return NULL;
}
void dst_release(struct dst_entry *dst)
{
if (dst) {
int newrefcnt;
smp_mb__before_atomic_dec();
newrefcnt = atomic_dec_return(&dst->__refcnt);
WARN_ON(newrefcnt < 0);
}
}
EXPORT_SYMBOL(dst_release);
/* 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_.
*
* Commented and originally written by Alexey.
*/
static inline void dst_ifdown(struct dst_entry *dst, struct net_device *dev,
int unregister)
{
if (dst->ops->ifdown)
dst->ops->ifdown(dst, dev, unregister);
if (dev != dst->dev)
return;
if (!unregister) {
dst->input = dst->output = dst_discard;
} else {
dst->dev = dev_net(dst->dev)->loopback_dev;
dev_hold(dst->dev);
dev_put(dev);
if (dst->neighbour && dst->neighbour->dev == dev) {
dst->neighbour->dev = dst->dev;
dev_hold(dst->dev);
dev_put(dev);
}
}
}
static int dst_dev_event(struct notifier_block *this, unsigned long event, void *ptr)
{
struct net_device *dev = ptr;
struct dst_entry *dst, *last = NULL;
switch (event) {
case NETDEV_UNREGISTER:
case NETDEV_DOWN:
mutex_lock(&dst_gc_mutex);
for (dst = dst_busy_list; dst; dst = dst->next) {
last = dst;
dst_ifdown(dst, dev, event != NETDEV_DOWN);
}
spin_lock_bh(&dst_garbage.lock);
dst = dst_garbage.list;
dst_garbage.list = NULL;
spin_unlock_bh(&dst_garbage.lock);
if (last)
last->next = dst;
else
dst_busy_list = dst;
for (; dst; dst = dst->next) {
dst_ifdown(dst, dev, event != NETDEV_DOWN);
}
mutex_unlock(&dst_gc_mutex);
break;
}
return NOTIFY_DONE;
}
static struct notifier_block dst_dev_notifier = {
.notifier_call = dst_dev_event,
};
void __init dst_init(void)
{
register_netdevice_notifier(&dst_dev_notifier);
}
EXPORT_SYMBOL(__dst_free);
EXPORT_SYMBOL(dst_alloc);
EXPORT_SYMBOL(dst_destroy);