2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* Generic address resolution entity
|
|
|
|
*
|
|
|
|
* Authors:
|
|
|
|
* Pedro Roque <roque@di.fc.ul.pt>
|
|
|
|
* Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU General Public License
|
|
|
|
* as published by the Free Software Foundation; either version
|
|
|
|
* 2 of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* Fixes:
|
|
|
|
* Vitaly E. Lavrov releasing NULL neighbor in neigh_add.
|
|
|
|
* Harald Welte Add neighbour cache statistics like rtstat
|
|
|
|
*/
|
|
|
|
|
2012-05-16 21:58:40 +02:00
|
|
|
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
|
|
|
|
|
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 09:04:11 +01:00
|
|
|
#include <linux/slab.h>
|
2005-04-17 00:20:36 +02:00
|
|
|
#include <linux/types.h>
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/socket.h>
|
|
|
|
#include <linux/netdevice.h>
|
|
|
|
#include <linux/proc_fs.h>
|
|
|
|
#ifdef CONFIG_SYSCTL
|
|
|
|
#include <linux/sysctl.h>
|
|
|
|
#endif
|
|
|
|
#include <linux/times.h>
|
2007-09-12 12:01:34 +02:00
|
|
|
#include <net/net_namespace.h>
|
2005-04-17 00:20:36 +02:00
|
|
|
#include <net/neighbour.h>
|
|
|
|
#include <net/dst.h>
|
|
|
|
#include <net/sock.h>
|
2006-07-31 05:43:36 +02:00
|
|
|
#include <net/netevent.h>
|
2006-08-08 02:53:08 +02:00
|
|
|
#include <net/netlink.h>
|
2005-04-17 00:20:36 +02:00
|
|
|
#include <linux/rtnetlink.h>
|
|
|
|
#include <linux/random.h>
|
2005-06-23 09:09:02 +02:00
|
|
|
#include <linux/string.h>
|
2007-08-25 07:27:55 +02:00
|
|
|
#include <linux/log2.h>
|
2013-12-07 19:26:56 +01:00
|
|
|
#include <linux/inetdevice.h>
|
2013-12-07 19:26:57 +01:00
|
|
|
#include <net/addrconf.h>
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2013-04-15 17:17:19 +02:00
|
|
|
#define DEBUG
|
2005-04-17 00:20:36 +02:00
|
|
|
#define NEIGH_DEBUG 1
|
2013-04-15 17:17:19 +02:00
|
|
|
#define neigh_dbg(level, fmt, ...) \
|
|
|
|
do { \
|
|
|
|
if (level <= NEIGH_DEBUG) \
|
|
|
|
pr_debug(fmt, ##__VA_ARGS__); \
|
|
|
|
} while (0)
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
#define PNEIGH_HASHMASK 0xF
|
|
|
|
|
|
|
|
static void neigh_timer_handler(unsigned long arg);
|
2007-08-09 08:12:56 +02:00
|
|
|
static void __neigh_notify(struct neighbour *n, int type, int flags);
|
|
|
|
static void neigh_update_notify(struct neighbour *neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
static int pneigh_ifdown(struct neigh_table *tbl, struct net_device *dev);
|
|
|
|
|
|
|
|
static struct neigh_table *neigh_tables;
|
2005-09-25 01:53:16 +02:00
|
|
|
#ifdef CONFIG_PROC_FS
|
2007-02-12 09:55:35 +01:00
|
|
|
static const struct file_operations neigh_stat_seq_fops;
|
2005-09-25 01:53:16 +02:00
|
|
|
#endif
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
Neighbour hash table buckets are protected with rwlock tbl->lock.
|
|
|
|
|
|
|
|
- All the scans/updates to hash buckets MUST be made under this lock.
|
|
|
|
- NOTHING clever should be made under this lock: no callbacks
|
|
|
|
to protocol backends, no attempts to send something to network.
|
|
|
|
It will result in deadlocks, if backend/driver wants to use neighbour
|
|
|
|
cache.
|
|
|
|
- If the entry requires some non-trivial actions, increase
|
|
|
|
its reference count and release table lock.
|
|
|
|
|
|
|
|
Neighbour entries are protected:
|
|
|
|
- with reference count.
|
|
|
|
- with rwlock neigh->lock
|
|
|
|
|
|
|
|
Reference count prevents destruction.
|
|
|
|
|
|
|
|
neigh->lock mainly serializes ll address data and its validity state.
|
|
|
|
However, the same lock is used to protect another entry fields:
|
|
|
|
- timer
|
|
|
|
- resolution queue
|
|
|
|
|
|
|
|
Again, nothing clever shall be made under neigh->lock,
|
|
|
|
the most complicated procedure, which we allow is dev->hard_header.
|
|
|
|
It is supposed, that dev->hard_header is simplistic and does
|
|
|
|
not make callbacks to neighbour tables.
|
|
|
|
|
|
|
|
The last lock is neigh_tbl_lock. It is pure SMP lock, protecting
|
|
|
|
list of neighbour tables. This list is used only in process context,
|
|
|
|
*/
|
|
|
|
|
|
|
|
static DEFINE_RWLOCK(neigh_tbl_lock);
|
|
|
|
|
2011-07-17 22:34:11 +02:00
|
|
|
static int neigh_blackhole(struct neighbour *neigh, struct sk_buff *skb)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
kfree_skb(skb);
|
|
|
|
return -ENETDOWN;
|
|
|
|
}
|
|
|
|
|
2007-08-09 08:12:36 +02:00
|
|
|
static void neigh_cleanup_and_release(struct neighbour *neigh)
|
|
|
|
{
|
|
|
|
if (neigh->parms->neigh_cleanup)
|
|
|
|
neigh->parms->neigh_cleanup(neigh);
|
|
|
|
|
2007-08-09 08:12:56 +02:00
|
|
|
__neigh_notify(neigh, RTM_DELNEIGH, 0);
|
2007-08-09 08:12:36 +02:00
|
|
|
neigh_release(neigh);
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* It is random distribution in the interval (1/2)*base...(3/2)*base.
|
|
|
|
* It corresponds to default IPv6 settings and is not overridable,
|
|
|
|
* because it is really reasonable choice.
|
|
|
|
*/
|
|
|
|
|
|
|
|
unsigned long neigh_rand_reach_time(unsigned long base)
|
|
|
|
{
|
2014-01-11 13:15:59 +01:00
|
|
|
return base ? (prandom_u32() % base) + (base >> 1) : 0;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_rand_reach_time);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
|
|
|
|
static int neigh_forced_gc(struct neigh_table *tbl)
|
|
|
|
{
|
|
|
|
int shrunk = 0;
|
|
|
|
int i;
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
NEIGH_CACHE_STAT_INC(tbl, forced_gc_runs);
|
|
|
|
|
|
|
|
write_lock_bh(&tbl->lock);
|
2010-10-04 08:15:44 +02:00
|
|
|
nht = rcu_dereference_protected(tbl->nht,
|
|
|
|
lockdep_is_held(&tbl->lock));
|
2011-07-11 10:28:12 +02:00
|
|
|
for (i = 0; i < (1 << nht->hash_shift); i++) {
|
2010-10-07 02:49:21 +02:00
|
|
|
struct neighbour *n;
|
|
|
|
struct neighbour __rcu **np;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
np = &nht->hash_buckets[i];
|
2010-10-07 02:49:21 +02:00
|
|
|
while ((n = rcu_dereference_protected(*np,
|
|
|
|
lockdep_is_held(&tbl->lock))) != NULL) {
|
2005-04-17 00:20:36 +02:00
|
|
|
/* Neighbour record may be discarded if:
|
|
|
|
* - nobody refers to it.
|
|
|
|
* - it is not permanent
|
|
|
|
*/
|
|
|
|
write_lock(&n->lock);
|
|
|
|
if (atomic_read(&n->refcnt) == 1 &&
|
|
|
|
!(n->nud_state & NUD_PERMANENT)) {
|
2010-10-07 02:49:21 +02:00
|
|
|
rcu_assign_pointer(*np,
|
|
|
|
rcu_dereference_protected(n->next,
|
|
|
|
lockdep_is_held(&tbl->lock)));
|
2005-04-17 00:20:36 +02:00
|
|
|
n->dead = 1;
|
|
|
|
shrunk = 1;
|
|
|
|
write_unlock(&n->lock);
|
2007-08-09 08:12:36 +02:00
|
|
|
neigh_cleanup_and_release(n);
|
2005-04-17 00:20:36 +02:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
write_unlock(&n->lock);
|
|
|
|
np = &n->next;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
tbl->last_flush = jiffies;
|
|
|
|
|
|
|
|
write_unlock_bh(&tbl->lock);
|
|
|
|
|
|
|
|
return shrunk;
|
|
|
|
}
|
|
|
|
|
2007-12-21 00:49:05 +01:00
|
|
|
static void neigh_add_timer(struct neighbour *n, unsigned long when)
|
|
|
|
{
|
|
|
|
neigh_hold(n);
|
|
|
|
if (unlikely(mod_timer(&n->timer, when))) {
|
|
|
|
printk("NEIGH: BUG, double timer add, state is %x\n",
|
|
|
|
n->nud_state);
|
|
|
|
dump_stack();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
static int neigh_del_timer(struct neighbour *n)
|
|
|
|
{
|
|
|
|
if ((n->nud_state & NUD_IN_TIMER) &&
|
|
|
|
del_timer(&n->timer)) {
|
|
|
|
neigh_release(n);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pneigh_queue_purge(struct sk_buff_head *list)
|
|
|
|
{
|
|
|
|
struct sk_buff *skb;
|
|
|
|
|
|
|
|
while ((skb = skb_dequeue(list)) != NULL) {
|
|
|
|
dev_put(skb->dev);
|
|
|
|
kfree_skb(skb);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2005-10-23 09:18:00 +02:00
|
|
|
static void neigh_flush_dev(struct neigh_table *tbl, struct net_device *dev)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
int i;
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
nht = rcu_dereference_protected(tbl->nht,
|
|
|
|
lockdep_is_held(&tbl->lock));
|
|
|
|
|
2011-07-11 10:28:12 +02:00
|
|
|
for (i = 0; i < (1 << nht->hash_shift); i++) {
|
2010-10-07 02:49:21 +02:00
|
|
|
struct neighbour *n;
|
|
|
|
struct neighbour __rcu **np = &nht->hash_buckets[i];
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-07 02:49:21 +02:00
|
|
|
while ((n = rcu_dereference_protected(*np,
|
|
|
|
lockdep_is_held(&tbl->lock))) != NULL) {
|
2005-04-17 00:20:36 +02:00
|
|
|
if (dev && n->dev != dev) {
|
|
|
|
np = &n->next;
|
|
|
|
continue;
|
|
|
|
}
|
2010-10-07 02:49:21 +02:00
|
|
|
rcu_assign_pointer(*np,
|
|
|
|
rcu_dereference_protected(n->next,
|
|
|
|
lockdep_is_held(&tbl->lock)));
|
2005-04-17 00:20:36 +02:00
|
|
|
write_lock(&n->lock);
|
|
|
|
neigh_del_timer(n);
|
|
|
|
n->dead = 1;
|
|
|
|
|
|
|
|
if (atomic_read(&n->refcnt) != 1) {
|
|
|
|
/* The most unpleasant situation.
|
|
|
|
We must destroy neighbour entry,
|
|
|
|
but someone still uses it.
|
|
|
|
|
|
|
|
The destroy will be delayed until
|
|
|
|
the last user releases us, but
|
|
|
|
we must kill timers etc. and move
|
|
|
|
it to safe state.
|
|
|
|
*/
|
2013-06-28 11:37:42 +02:00
|
|
|
__skb_queue_purge(&n->arp_queue);
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
n->arp_queue_len_bytes = 0;
|
2005-04-17 00:20:36 +02:00
|
|
|
n->output = neigh_blackhole;
|
|
|
|
if (n->nud_state & NUD_VALID)
|
|
|
|
n->nud_state = NUD_NOARP;
|
|
|
|
else
|
|
|
|
n->nud_state = NUD_NONE;
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(2, "neigh %p is stray\n", n);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
write_unlock(&n->lock);
|
2007-08-09 08:12:36 +02:00
|
|
|
neigh_cleanup_and_release(n);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
}
|
2005-10-23 09:18:00 +02:00
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2005-10-23 09:18:00 +02:00
|
|
|
void neigh_changeaddr(struct neigh_table *tbl, struct net_device *dev)
|
|
|
|
{
|
|
|
|
write_lock_bh(&tbl->lock);
|
|
|
|
neigh_flush_dev(tbl, dev);
|
|
|
|
write_unlock_bh(&tbl->lock);
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_changeaddr);
|
2005-10-23 09:18:00 +02:00
|
|
|
|
|
|
|
int neigh_ifdown(struct neigh_table *tbl, struct net_device *dev)
|
|
|
|
{
|
|
|
|
write_lock_bh(&tbl->lock);
|
|
|
|
neigh_flush_dev(tbl, dev);
|
2005-04-17 00:20:36 +02:00
|
|
|
pneigh_ifdown(tbl, dev);
|
|
|
|
write_unlock_bh(&tbl->lock);
|
|
|
|
|
|
|
|
del_timer_sync(&tbl->proxy_timer);
|
|
|
|
pneigh_queue_purge(&tbl->proxy_queue);
|
|
|
|
return 0;
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_ifdown);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2011-07-25 02:01:25 +02:00
|
|
|
static struct neighbour *neigh_alloc(struct neigh_table *tbl, struct net_device *dev)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
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;
|
|
|
|
}
|
|
|
|
|
2013-01-24 01:44:23 +01:00
|
|
|
n = kzalloc(tbl->entry_size + dev->neigh_priv_len, GFP_ATOMIC);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (!n)
|
|
|
|
goto out_entries;
|
|
|
|
|
2013-06-28 11:37:42 +02:00
|
|
|
__skb_queue_head_init(&n->arp_queue);
|
2005-04-17 00:20:36 +02:00
|
|
|
rwlock_init(&n->lock);
|
2010-10-07 12:44:07 +02:00
|
|
|
seqlock_init(&n->ha_lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
n->updated = n->used = now;
|
|
|
|
n->nud_state = NUD_NONE;
|
|
|
|
n->output = neigh_blackhole;
|
2011-07-14 16:53:20 +02:00
|
|
|
seqlock_init(&n->hh.hh_lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
n->parms = neigh_parms_clone(&tbl->parms);
|
2008-01-24 06:20:07 +01:00
|
|
|
setup_timer(&n->timer, neigh_timer_handler, (unsigned long)n);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
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;
|
|
|
|
}
|
|
|
|
|
2011-12-28 21:06:58 +01:00
|
|
|
static void neigh_get_hash_rnd(u32 *x)
|
|
|
|
{
|
|
|
|
get_random_bytes(x, sizeof(*x));
|
|
|
|
*x |= 1;
|
|
|
|
}
|
|
|
|
|
2011-07-11 10:28:12 +02:00
|
|
|
static struct neigh_hash_table *neigh_hash_alloc(unsigned int shift)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2011-07-11 10:28:12 +02:00
|
|
|
size_t size = (1 << shift) * sizeof(struct neighbour *);
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *ret;
|
2011-01-19 23:02:47 +01:00
|
|
|
struct neighbour __rcu **buckets;
|
2011-12-28 21:06:58 +01:00
|
|
|
int i;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
ret = kmalloc(sizeof(*ret), GFP_ATOMIC);
|
|
|
|
if (!ret)
|
|
|
|
return NULL;
|
|
|
|
if (size <= PAGE_SIZE)
|
|
|
|
buckets = kzalloc(size, GFP_ATOMIC);
|
|
|
|
else
|
2011-01-19 23:02:47 +01:00
|
|
|
buckets = (struct neighbour __rcu **)
|
2010-10-04 08:15:44 +02:00
|
|
|
__get_free_pages(GFP_ATOMIC | __GFP_ZERO,
|
|
|
|
get_order(size));
|
|
|
|
if (!buckets) {
|
|
|
|
kfree(ret);
|
|
|
|
return NULL;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2011-01-19 23:02:47 +01:00
|
|
|
ret->hash_buckets = buckets;
|
2011-07-11 10:28:12 +02:00
|
|
|
ret->hash_shift = shift;
|
2011-12-28 21:06:58 +01:00
|
|
|
for (i = 0; i < NEIGH_NUM_HASH_RND; i++)
|
|
|
|
neigh_get_hash_rnd(&ret->hash_rnd[i]);
|
2005-04-17 00:20:36 +02:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
static void neigh_hash_free_rcu(struct rcu_head *head)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht = container_of(head,
|
|
|
|
struct neigh_hash_table,
|
|
|
|
rcu);
|
2011-07-11 10:28:12 +02:00
|
|
|
size_t size = (1 << nht->hash_shift) * sizeof(struct neighbour *);
|
2011-01-19 23:02:47 +01:00
|
|
|
struct neighbour __rcu **buckets = nht->hash_buckets;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (size <= PAGE_SIZE)
|
2010-10-04 08:15:44 +02:00
|
|
|
kfree(buckets);
|
2005-04-17 00:20:36 +02:00
|
|
|
else
|
2010-10-04 08:15:44 +02:00
|
|
|
free_pages((unsigned long)buckets, get_order(size));
|
|
|
|
kfree(nht);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
static struct neigh_hash_table *neigh_hash_grow(struct neigh_table *tbl,
|
2011-07-11 10:28:12 +02:00
|
|
|
unsigned long new_shift)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2010-10-04 08:15:44 +02:00
|
|
|
unsigned int i, hash;
|
|
|
|
struct neigh_hash_table *new_nht, *old_nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
NEIGH_CACHE_STAT_INC(tbl, hash_grows);
|
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
old_nht = rcu_dereference_protected(tbl->nht,
|
|
|
|
lockdep_is_held(&tbl->lock));
|
2011-07-11 10:28:12 +02:00
|
|
|
new_nht = neigh_hash_alloc(new_shift);
|
2010-10-04 08:15:44 +02:00
|
|
|
if (!new_nht)
|
|
|
|
return old_nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2011-07-11 10:28:12 +02:00
|
|
|
for (i = 0; i < (1 << old_nht->hash_shift); i++) {
|
2005-04-17 00:20:36 +02:00
|
|
|
struct neighbour *n, *next;
|
|
|
|
|
2010-10-07 02:49:21 +02:00
|
|
|
for (n = rcu_dereference_protected(old_nht->hash_buckets[i],
|
|
|
|
lockdep_is_held(&tbl->lock));
|
2010-10-04 08:15:44 +02:00
|
|
|
n != NULL;
|
|
|
|
n = next) {
|
|
|
|
hash = tbl->hash(n->primary_key, n->dev,
|
|
|
|
new_nht->hash_rnd);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2011-07-11 10:28:12 +02:00
|
|
|
hash >>= (32 - new_nht->hash_shift);
|
2010-10-07 02:49:21 +02:00
|
|
|
next = rcu_dereference_protected(n->next,
|
|
|
|
lockdep_is_held(&tbl->lock));
|
|
|
|
|
|
|
|
rcu_assign_pointer(n->next,
|
|
|
|
rcu_dereference_protected(
|
|
|
|
new_nht->hash_buckets[hash],
|
|
|
|
lockdep_is_held(&tbl->lock)));
|
|
|
|
rcu_assign_pointer(new_nht->hash_buckets[hash], n);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
rcu_assign_pointer(tbl->nht, new_nht);
|
|
|
|
call_rcu(&old_nht->rcu, neigh_hash_free_rcu);
|
|
|
|
return new_nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
struct neighbour *neigh_lookup(struct neigh_table *tbl, const void *pkey,
|
|
|
|
struct net_device *dev)
|
|
|
|
{
|
|
|
|
struct neighbour *n;
|
|
|
|
int key_len = tbl->key_len;
|
2008-02-24 04:57:02 +01:00
|
|
|
u32 hash_val;
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht;
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
NEIGH_CACHE_STAT_INC(tbl, lookups);
|
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
rcu_read_lock_bh();
|
|
|
|
nht = rcu_dereference_bh(tbl->nht);
|
2011-07-11 10:28:12 +02:00
|
|
|
hash_val = tbl->hash(pkey, dev, nht->hash_rnd) >> (32 - nht->hash_shift);
|
2010-10-07 02:49:21 +02:00
|
|
|
|
|
|
|
for (n = rcu_dereference_bh(nht->hash_buckets[hash_val]);
|
|
|
|
n != NULL;
|
|
|
|
n = rcu_dereference_bh(n->next)) {
|
2005-04-17 00:20:36 +02:00
|
|
|
if (dev == n->dev && !memcmp(n->primary_key, pkey, key_len)) {
|
2010-10-07 02:49:21 +02:00
|
|
|
if (!atomic_inc_not_zero(&n->refcnt))
|
|
|
|
n = NULL;
|
2005-04-17 00:20:36 +02:00
|
|
|
NEIGH_CACHE_STAT_INC(tbl, hits);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2010-10-07 02:49:21 +02:00
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
rcu_read_unlock_bh();
|
2005-04-17 00:20:36 +02:00
|
|
|
return n;
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_lookup);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2008-01-24 09:13:18 +01:00
|
|
|
struct neighbour *neigh_lookup_nodev(struct neigh_table *tbl, struct net *net,
|
|
|
|
const void *pkey)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct neighbour *n;
|
|
|
|
int key_len = tbl->key_len;
|
2008-02-24 04:57:02 +01:00
|
|
|
u32 hash_val;
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
NEIGH_CACHE_STAT_INC(tbl, lookups);
|
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
rcu_read_lock_bh();
|
|
|
|
nht = rcu_dereference_bh(tbl->nht);
|
2011-07-11 10:28:12 +02:00
|
|
|
hash_val = tbl->hash(pkey, NULL, nht->hash_rnd) >> (32 - nht->hash_shift);
|
2010-10-07 02:49:21 +02:00
|
|
|
|
|
|
|
for (n = rcu_dereference_bh(nht->hash_buckets[hash_val]);
|
|
|
|
n != NULL;
|
|
|
|
n = rcu_dereference_bh(n->next)) {
|
2008-01-24 09:13:18 +01:00
|
|
|
if (!memcmp(n->primary_key, pkey, key_len) &&
|
2008-03-25 19:57:35 +01:00
|
|
|
net_eq(dev_net(n->dev), net)) {
|
2010-10-07 02:49:21 +02:00
|
|
|
if (!atomic_inc_not_zero(&n->refcnt))
|
|
|
|
n = NULL;
|
2005-04-17 00:20:36 +02:00
|
|
|
NEIGH_CACHE_STAT_INC(tbl, hits);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2010-10-07 02:49:21 +02:00
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
rcu_read_unlock_bh();
|
2005-04-17 00:20:36 +02:00
|
|
|
return n;
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_lookup_nodev);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2012-07-02 11:02:15 +02:00
|
|
|
struct neighbour *__neigh_create(struct neigh_table *tbl, const void *pkey,
|
|
|
|
struct net_device *dev, bool want_ref)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
u32 hash_val;
|
|
|
|
int key_len = tbl->key_len;
|
|
|
|
int error;
|
2011-07-25 02:01:25 +02:00
|
|
|
struct neighbour *n1, *rc, *n = neigh_alloc(tbl, dev);
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
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;
|
|
|
|
}
|
|
|
|
|
2011-07-25 02:01:38 +02:00
|
|
|
if (dev->netdev_ops->ndo_neigh_construct) {
|
|
|
|
error = dev->netdev_ops->ndo_neigh_construct(n);
|
|
|
|
if (error < 0) {
|
|
|
|
rc = ERR_PTR(error);
|
|
|
|
goto out_neigh_release;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-12-19 21:04:41 +01:00
|
|
|
/* Device specific setup. */
|
|
|
|
if (n->parms->neigh_setup &&
|
|
|
|
(error = n->parms->neigh_setup(n)) < 0) {
|
|
|
|
rc = ERR_PTR(error);
|
|
|
|
goto out_neigh_release;
|
|
|
|
}
|
|
|
|
|
2013-12-07 19:26:53 +01:00
|
|
|
n->confirmed = jiffies - (NEIGH_VAR(n->parms, BASE_REACHABLE_TIME) << 1);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
write_lock_bh(&tbl->lock);
|
2010-10-04 08:15:44 +02:00
|
|
|
nht = rcu_dereference_protected(tbl->nht,
|
|
|
|
lockdep_is_held(&tbl->lock));
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2011-07-11 10:28:12 +02:00
|
|
|
if (atomic_read(&tbl->entries) > (1 << nht->hash_shift))
|
|
|
|
nht = neigh_hash_grow(tbl, nht->hash_shift + 1);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2011-07-11 10:28:12 +02:00
|
|
|
hash_val = tbl->hash(pkey, dev, nht->hash_rnd) >> (32 - nht->hash_shift);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (n->parms->dead) {
|
|
|
|
rc = ERR_PTR(-EINVAL);
|
|
|
|
goto out_tbl_unlock;
|
|
|
|
}
|
|
|
|
|
2010-10-07 02:49:21 +02:00
|
|
|
for (n1 = rcu_dereference_protected(nht->hash_buckets[hash_val],
|
|
|
|
lockdep_is_held(&tbl->lock));
|
|
|
|
n1 != NULL;
|
|
|
|
n1 = rcu_dereference_protected(n1->next,
|
|
|
|
lockdep_is_held(&tbl->lock))) {
|
2005-04-17 00:20:36 +02:00
|
|
|
if (dev == n1->dev && !memcmp(n1->primary_key, pkey, key_len)) {
|
2012-07-02 11:02:15 +02:00
|
|
|
if (want_ref)
|
|
|
|
neigh_hold(n1);
|
2005-04-17 00:20:36 +02:00
|
|
|
rc = n1;
|
|
|
|
goto out_tbl_unlock;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
n->dead = 0;
|
2012-07-02 11:02:15 +02:00
|
|
|
if (want_ref)
|
|
|
|
neigh_hold(n);
|
2010-10-07 02:49:21 +02:00
|
|
|
rcu_assign_pointer(n->next,
|
|
|
|
rcu_dereference_protected(nht->hash_buckets[hash_val],
|
|
|
|
lockdep_is_held(&tbl->lock)));
|
|
|
|
rcu_assign_pointer(nht->hash_buckets[hash_val], n);
|
2005-04-17 00:20:36 +02:00
|
|
|
write_unlock_bh(&tbl->lock);
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(2, "neigh %p is created\n", n);
|
2005-04-17 00:20:36 +02:00
|
|
|
rc = n;
|
|
|
|
out:
|
|
|
|
return rc;
|
|
|
|
out_tbl_unlock:
|
|
|
|
write_unlock_bh(&tbl->lock);
|
|
|
|
out_neigh_release:
|
|
|
|
neigh_release(n);
|
|
|
|
goto out;
|
|
|
|
}
|
2012-07-02 11:02:15 +02:00
|
|
|
EXPORT_SYMBOL(__neigh_create);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2008-03-28 04:46:53 +01:00
|
|
|
static u32 pneigh_hash(const void *pkey, int key_len)
|
2008-03-24 22:48:59 +01:00
|
|
|
{
|
|
|
|
u32 hash_val = *(u32 *)(pkey + key_len - 4);
|
|
|
|
hash_val ^= (hash_val >> 16);
|
|
|
|
hash_val ^= hash_val >> 8;
|
|
|
|
hash_val ^= hash_val >> 4;
|
|
|
|
hash_val &= PNEIGH_HASHMASK;
|
2008-03-28 04:46:53 +01:00
|
|
|
return hash_val;
|
|
|
|
}
|
2008-03-24 22:48:59 +01:00
|
|
|
|
2008-03-28 04:46:53 +01:00
|
|
|
static struct pneigh_entry *__pneigh_lookup_1(struct pneigh_entry *n,
|
|
|
|
struct net *net,
|
|
|
|
const void *pkey,
|
|
|
|
int key_len,
|
|
|
|
struct net_device *dev)
|
|
|
|
{
|
|
|
|
while (n) {
|
2008-03-24 22:48:59 +01:00
|
|
|
if (!memcmp(n->key, pkey, key_len) &&
|
2008-03-28 04:46:53 +01:00
|
|
|
net_eq(pneigh_net(n), net) &&
|
2008-03-24 22:48:59 +01:00
|
|
|
(n->dev == dev || !n->dev))
|
2008-03-28 04:46:53 +01:00
|
|
|
return n;
|
|
|
|
n = n->next;
|
2008-03-24 22:48:59 +01:00
|
|
|
}
|
2008-03-28 04:46:53 +01:00
|
|
|
return NULL;
|
|
|
|
}
|
2008-03-24 22:48:59 +01:00
|
|
|
|
2008-03-28 04:46:53 +01:00
|
|
|
struct pneigh_entry *__pneigh_lookup(struct neigh_table *tbl,
|
|
|
|
struct net *net, const void *pkey, struct net_device *dev)
|
|
|
|
{
|
|
|
|
int key_len = tbl->key_len;
|
|
|
|
u32 hash_val = pneigh_hash(pkey, key_len);
|
|
|
|
|
|
|
|
return __pneigh_lookup_1(tbl->phash_buckets[hash_val],
|
|
|
|
net, pkey, key_len, dev);
|
2008-03-24 22:48:59 +01:00
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL_GPL(__pneigh_lookup);
|
2008-03-24 22:48:59 +01:00
|
|
|
|
2008-01-24 09:13:18 +01:00
|
|
|
struct pneigh_entry * pneigh_lookup(struct neigh_table *tbl,
|
|
|
|
struct net *net, const void *pkey,
|
2005-04-17 00:20:36 +02:00
|
|
|
struct net_device *dev, int creat)
|
|
|
|
{
|
|
|
|
struct pneigh_entry *n;
|
|
|
|
int key_len = tbl->key_len;
|
2008-03-28 04:46:53 +01:00
|
|
|
u32 hash_val = pneigh_hash(pkey, key_len);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
read_lock_bh(&tbl->lock);
|
2008-03-28 04:46:53 +01:00
|
|
|
n = __pneigh_lookup_1(tbl->phash_buckets[hash_val],
|
|
|
|
net, pkey, key_len, dev);
|
2005-04-17 00:20:36 +02:00
|
|
|
read_unlock_bh(&tbl->lock);
|
2008-03-28 04:46:53 +01:00
|
|
|
|
|
|
|
if (n || !creat)
|
2005-04-17 00:20:36 +02:00
|
|
|
goto out;
|
|
|
|
|
2007-10-15 21:54:15 +02:00
|
|
|
ASSERT_RTNL();
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
n = kmalloc(sizeof(*n) + key_len, GFP_KERNEL);
|
|
|
|
if (!n)
|
|
|
|
goto out;
|
|
|
|
|
2008-11-12 09:54:54 +01:00
|
|
|
write_pnet(&n->net, hold_net(net));
|
2005-04-17 00:20:36 +02:00
|
|
|
memcpy(n->key, pkey, key_len);
|
|
|
|
n->dev = dev;
|
|
|
|
if (dev)
|
|
|
|
dev_hold(dev);
|
|
|
|
|
|
|
|
if (tbl->pconstructor && tbl->pconstructor(n)) {
|
|
|
|
if (dev)
|
|
|
|
dev_put(dev);
|
2008-02-20 09:26:16 +01:00
|
|
|
release_net(net);
|
2005-04-17 00:20:36 +02:00
|
|
|
kfree(n);
|
|
|
|
n = NULL;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
write_lock_bh(&tbl->lock);
|
|
|
|
n->next = tbl->phash_buckets[hash_val];
|
|
|
|
tbl->phash_buckets[hash_val] = n;
|
|
|
|
write_unlock_bh(&tbl->lock);
|
|
|
|
out:
|
|
|
|
return n;
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(pneigh_lookup);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
|
2008-01-24 09:13:18 +01:00
|
|
|
int pneigh_delete(struct neigh_table *tbl, struct net *net, const void *pkey,
|
2005-04-17 00:20:36 +02:00
|
|
|
struct net_device *dev)
|
|
|
|
{
|
|
|
|
struct pneigh_entry *n, **np;
|
|
|
|
int key_len = tbl->key_len;
|
2008-03-28 04:46:53 +01:00
|
|
|
u32 hash_val = pneigh_hash(pkey, key_len);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
write_lock_bh(&tbl->lock);
|
|
|
|
for (np = &tbl->phash_buckets[hash_val]; (n = *np) != NULL;
|
|
|
|
np = &n->next) {
|
2008-01-24 09:13:18 +01:00
|
|
|
if (!memcmp(n->key, pkey, key_len) && n->dev == dev &&
|
2008-03-25 19:57:35 +01:00
|
|
|
net_eq(pneigh_net(n), net)) {
|
2005-04-17 00:20:36 +02:00
|
|
|
*np = n->next;
|
|
|
|
write_unlock_bh(&tbl->lock);
|
|
|
|
if (tbl->pdestructor)
|
|
|
|
tbl->pdestructor(n);
|
|
|
|
if (n->dev)
|
|
|
|
dev_put(n->dev);
|
2008-03-25 19:49:59 +01:00
|
|
|
release_net(pneigh_net(n));
|
2005-04-17 00:20:36 +02:00
|
|
|
kfree(n);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
write_unlock_bh(&tbl->lock);
|
|
|
|
return -ENOENT;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int pneigh_ifdown(struct neigh_table *tbl, struct net_device *dev)
|
|
|
|
{
|
|
|
|
struct pneigh_entry *n, **np;
|
|
|
|
u32 h;
|
|
|
|
|
|
|
|
for (h = 0; h <= PNEIGH_HASHMASK; h++) {
|
|
|
|
np = &tbl->phash_buckets[h];
|
|
|
|
while ((n = *np) != NULL) {
|
|
|
|
if (!dev || n->dev == dev) {
|
|
|
|
*np = n->next;
|
|
|
|
if (tbl->pdestructor)
|
|
|
|
tbl->pdestructor(n);
|
|
|
|
if (n->dev)
|
|
|
|
dev_put(n->dev);
|
2008-03-25 19:49:59 +01:00
|
|
|
release_net(pneigh_net(n));
|
2005-04-17 00:20:36 +02:00
|
|
|
kfree(n);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
np = &n->next;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return -ENOENT;
|
|
|
|
}
|
|
|
|
|
2008-01-24 09:30:58 +01:00
|
|
|
static void neigh_parms_destroy(struct neigh_parms *parms);
|
|
|
|
|
|
|
|
static inline void neigh_parms_put(struct neigh_parms *parms)
|
|
|
|
{
|
|
|
|
if (atomic_dec_and_test(&parms->refcnt))
|
|
|
|
neigh_parms_destroy(parms);
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* neighbour must already be out of the table;
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
void neigh_destroy(struct neighbour *neigh)
|
|
|
|
{
|
2011-07-25 02:01:38 +02:00
|
|
|
struct net_device *dev = neigh->dev;
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
NEIGH_CACHE_STAT_INC(neigh->tbl, destroys);
|
|
|
|
|
|
|
|
if (!neigh->dead) {
|
2012-05-16 21:58:40 +02:00
|
|
|
pr_warn("Destroying alive neighbour %p\n", neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
dump_stack();
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (neigh_del_timer(neigh))
|
2012-05-16 21:58:40 +02:00
|
|
|
pr_warn("Impossible event\n");
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2013-06-28 11:37:42 +02:00
|
|
|
write_lock_bh(&neigh->lock);
|
|
|
|
__skb_queue_purge(&neigh->arp_queue);
|
|
|
|
write_unlock_bh(&neigh->lock);
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
neigh->arp_queue_len_bytes = 0;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2011-12-19 21:04:41 +01:00
|
|
|
if (dev->netdev_ops->ndo_neigh_destroy)
|
|
|
|
dev->netdev_ops->ndo_neigh_destroy(neigh);
|
|
|
|
|
2011-07-25 02:01:38 +02:00
|
|
|
dev_put(dev);
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh_parms_put(neigh->parms);
|
|
|
|
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(2, "neigh %p is destroyed\n", neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
atomic_dec(&neigh->tbl->entries);
|
2011-07-25 02:01:22 +02:00
|
|
|
kfree_rcu(neigh, rcu);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_destroy);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/* Neighbour state is suspicious;
|
|
|
|
disable fast path.
|
|
|
|
|
|
|
|
Called with write_locked neigh.
|
|
|
|
*/
|
|
|
|
static void neigh_suspect(struct neighbour *neigh)
|
|
|
|
{
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(2, "neigh %p is suspected\n", neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
neigh->output = neigh->ops->output;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Neighbour state is OK;
|
|
|
|
enable fast path.
|
|
|
|
|
|
|
|
Called with write_locked neigh.
|
|
|
|
*/
|
|
|
|
static void neigh_connect(struct neighbour *neigh)
|
|
|
|
{
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(2, "neigh %p is connected\n", neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
neigh->output = neigh->ops->connected_output;
|
|
|
|
}
|
|
|
|
|
2009-07-30 05:15:07 +02:00
|
|
|
static void neigh_periodic_work(struct work_struct *work)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2009-07-30 05:15:07 +02:00
|
|
|
struct neigh_table *tbl = container_of(work, struct neigh_table, gc_work.work);
|
2010-10-07 02:49:21 +02:00
|
|
|
struct neighbour *n;
|
|
|
|
struct neighbour __rcu **np;
|
2009-07-30 05:15:07 +02:00
|
|
|
unsigned int i;
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
NEIGH_CACHE_STAT_INC(tbl, periodic_gc_runs);
|
|
|
|
|
2009-07-30 05:15:07 +02:00
|
|
|
write_lock_bh(&tbl->lock);
|
2010-10-04 08:15:44 +02:00
|
|
|
nht = rcu_dereference_protected(tbl->nht,
|
|
|
|
lockdep_is_held(&tbl->lock));
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2013-01-22 06:20:05 +01:00
|
|
|
if (atomic_read(&tbl->entries) < tbl->gc_thresh1)
|
|
|
|
goto out;
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* periodically recompute ReachableTime from random function
|
|
|
|
*/
|
|
|
|
|
2009-07-30 05:15:07 +02:00
|
|
|
if (time_after(jiffies, tbl->last_rand + 300 * HZ)) {
|
2005-04-17 00:20:36 +02:00
|
|
|
struct neigh_parms *p;
|
2009-07-30 05:15:07 +02:00
|
|
|
tbl->last_rand = jiffies;
|
2005-04-17 00:20:36 +02:00
|
|
|
for (p = &tbl->parms; p; p = p->next)
|
|
|
|
p->reachable_time =
|
2013-12-07 19:26:53 +01:00
|
|
|
neigh_rand_reach_time(NEIGH_VAR(p, BASE_REACHABLE_TIME));
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2011-07-11 10:28:12 +02:00
|
|
|
for (i = 0 ; i < (1 << nht->hash_shift); i++) {
|
2010-10-04 08:15:44 +02:00
|
|
|
np = &nht->hash_buckets[i];
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-07 02:49:21 +02:00
|
|
|
while ((n = rcu_dereference_protected(*np,
|
|
|
|
lockdep_is_held(&tbl->lock))) != NULL) {
|
2009-07-30 05:15:07 +02:00
|
|
|
unsigned int state;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2009-07-30 05:15:07 +02:00
|
|
|
write_lock(&n->lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2009-07-30 05:15:07 +02:00
|
|
|
state = n->nud_state;
|
|
|
|
if (state & (NUD_PERMANENT | NUD_IN_TIMER)) {
|
|
|
|
write_unlock(&n->lock);
|
|
|
|
goto next_elt;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2009-07-30 05:15:07 +02:00
|
|
|
if (time_before(n->used, n->confirmed))
|
|
|
|
n->used = n->confirmed;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2009-07-30 05:15:07 +02:00
|
|
|
if (atomic_read(&n->refcnt) == 1 &&
|
|
|
|
(state == NUD_FAILED ||
|
2013-12-07 19:26:53 +01:00
|
|
|
time_after(jiffies, n->used + NEIGH_VAR(n->parms, GC_STALETIME)))) {
|
2009-07-30 05:15:07 +02:00
|
|
|
*np = n->next;
|
|
|
|
n->dead = 1;
|
|
|
|
write_unlock(&n->lock);
|
|
|
|
neigh_cleanup_and_release(n);
|
|
|
|
continue;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
write_unlock(&n->lock);
|
|
|
|
|
|
|
|
next_elt:
|
2009-07-30 05:15:07 +02:00
|
|
|
np = &n->next;
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* It's fine to release lock here, even if hash table
|
|
|
|
* grows while we are preempted.
|
|
|
|
*/
|
|
|
|
write_unlock_bh(&tbl->lock);
|
|
|
|
cond_resched();
|
|
|
|
write_lock_bh(&tbl->lock);
|
2012-02-21 22:04:13 +01:00
|
|
|
nht = rcu_dereference_protected(tbl->nht,
|
|
|
|
lockdep_is_held(&tbl->lock));
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2013-01-22 06:20:05 +01:00
|
|
|
out:
|
2013-12-07 19:26:53 +01:00
|
|
|
/* Cycle through all hash buckets every BASE_REACHABLE_TIME/2 ticks.
|
|
|
|
* ARP entry timeouts range from 1/2 BASE_REACHABLE_TIME to 3/2
|
|
|
|
* BASE_REACHABLE_TIME.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
2014-01-22 07:53:33 +01:00
|
|
|
queue_delayed_work(system_power_efficient_wq, &tbl->gc_work,
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR(&tbl->parms, BASE_REACHABLE_TIME) >> 1);
|
2009-07-30 05:15:07 +02:00
|
|
|
write_unlock_bh(&tbl->lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
static __inline__ int neigh_max_probes(struct neighbour *n)
|
|
|
|
{
|
|
|
|
struct neigh_parms *p = n->parms;
|
2010-09-22 22:43:57 +02:00
|
|
|
return (n->nud_state & NUD_PROBE) ?
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR(p, UCAST_PROBES) :
|
|
|
|
NEIGH_VAR(p, UCAST_PROBES) + NEIGH_VAR(p, APP_PROBES) +
|
|
|
|
NEIGH_VAR(p, MCAST_PROBES);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2009-06-11 13:16:28 +02:00
|
|
|
static void neigh_invalidate(struct neighbour *neigh)
|
2010-03-09 20:40:54 +01:00
|
|
|
__releases(neigh->lock)
|
|
|
|
__acquires(neigh->lock)
|
2009-06-11 13:16:28 +02:00
|
|
|
{
|
|
|
|
struct sk_buff *skb;
|
|
|
|
|
|
|
|
NEIGH_CACHE_STAT_INC(neigh->tbl, res_failed);
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(2, "neigh %p is failed\n", neigh);
|
2009-06-11 13:16:28 +02:00
|
|
|
neigh->updated = jiffies;
|
|
|
|
|
|
|
|
/* It is very thin place. report_unreachable is very complicated
|
|
|
|
routine. Particularly, it can hit the same neighbour entry!
|
|
|
|
|
|
|
|
So that, we try to be accurate and avoid dead loop. --ANK
|
|
|
|
*/
|
|
|
|
while (neigh->nud_state == NUD_FAILED &&
|
|
|
|
(skb = __skb_dequeue(&neigh->arp_queue)) != NULL) {
|
|
|
|
write_unlock(&neigh->lock);
|
|
|
|
neigh->ops->error_report(neigh, skb);
|
|
|
|
write_lock(&neigh->lock);
|
|
|
|
}
|
2013-06-28 11:37:42 +02:00
|
|
|
__skb_queue_purge(&neigh->arp_queue);
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
neigh->arp_queue_len_bytes = 0;
|
2009-06-11 13:16:28 +02:00
|
|
|
}
|
|
|
|
|
2011-08-09 10:15:58 +02:00
|
|
|
static void neigh_probe(struct neighbour *neigh)
|
|
|
|
__releases(neigh->lock)
|
|
|
|
{
|
2013-09-21 06:32:34 +02:00
|
|
|
struct sk_buff *skb = skb_peek_tail(&neigh->arp_queue);
|
2011-08-09 10:15:58 +02:00
|
|
|
/* keep skb alive even if arp_queue overflows */
|
|
|
|
if (skb)
|
|
|
|
skb = skb_copy(skb, GFP_ATOMIC);
|
|
|
|
write_unlock(&neigh->lock);
|
|
|
|
neigh->ops->solicit(neigh, skb);
|
|
|
|
atomic_inc(&neigh->probes);
|
|
|
|
kfree_skb(skb);
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/* Called when a timer expires for a neighbour entry. */
|
|
|
|
|
|
|
|
static void neigh_timer_handler(unsigned long arg)
|
|
|
|
{
|
|
|
|
unsigned long now, next;
|
|
|
|
struct neighbour *neigh = (struct neighbour *)arg;
|
2012-04-15 07:58:06 +02:00
|
|
|
unsigned int state;
|
2005-04-17 00:20:36 +02:00
|
|
|
int notify = 0;
|
|
|
|
|
|
|
|
write_lock(&neigh->lock);
|
|
|
|
|
|
|
|
state = neigh->nud_state;
|
|
|
|
now = jiffies;
|
|
|
|
next = now + HZ;
|
|
|
|
|
2011-11-01 22:45:55 +01:00
|
|
|
if (!(state & NUD_IN_TIMER))
|
2005-04-17 00:20:36 +02:00
|
|
|
goto out;
|
|
|
|
|
|
|
|
if (state & NUD_REACHABLE) {
|
2007-02-09 15:24:36 +01:00
|
|
|
if (time_before_eq(now,
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh->confirmed + neigh->parms->reachable_time)) {
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(2, "neigh %p is still alive\n", neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
next = neigh->confirmed + neigh->parms->reachable_time;
|
|
|
|
} else if (time_before_eq(now,
|
2013-12-07 19:26:53 +01:00
|
|
|
neigh->used +
|
|
|
|
NEIGH_VAR(neigh->parms, DELAY_PROBE_TIME))) {
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(2, "neigh %p is delayed\n", neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh->nud_state = NUD_DELAY;
|
2006-03-21 01:52:52 +01:00
|
|
|
neigh->updated = jiffies;
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh_suspect(neigh);
|
2013-12-07 19:26:53 +01:00
|
|
|
next = now + NEIGH_VAR(neigh->parms, DELAY_PROBE_TIME);
|
2005-04-17 00:20:36 +02:00
|
|
|
} else {
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(2, "neigh %p is suspected\n", neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh->nud_state = NUD_STALE;
|
2006-03-21 01:52:52 +01:00
|
|
|
neigh->updated = jiffies;
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh_suspect(neigh);
|
2006-07-31 05:43:36 +02:00
|
|
|
notify = 1;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
} else if (state & NUD_DELAY) {
|
2007-02-09 15:24:36 +01:00
|
|
|
if (time_before_eq(now,
|
2013-12-07 19:26:53 +01:00
|
|
|
neigh->confirmed +
|
|
|
|
NEIGH_VAR(neigh->parms, DELAY_PROBE_TIME))) {
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(2, "neigh %p is now reachable\n", neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh->nud_state = NUD_REACHABLE;
|
2006-03-21 01:52:52 +01:00
|
|
|
neigh->updated = jiffies;
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh_connect(neigh);
|
2006-07-31 05:43:36 +02:00
|
|
|
notify = 1;
|
2005-04-17 00:20:36 +02:00
|
|
|
next = neigh->confirmed + neigh->parms->reachable_time;
|
|
|
|
} else {
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(2, "neigh %p is probed\n", neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh->nud_state = NUD_PROBE;
|
2006-03-21 01:52:52 +01:00
|
|
|
neigh->updated = jiffies;
|
2005-04-17 00:20:36 +02:00
|
|
|
atomic_set(&neigh->probes, 0);
|
2013-12-07 19:26:53 +01:00
|
|
|
next = now + NEIGH_VAR(neigh->parms, RETRANS_TIME);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
} else {
|
|
|
|
/* NUD_PROBE|NUD_INCOMPLETE */
|
2013-12-07 19:26:53 +01:00
|
|
|
next = now + NEIGH_VAR(neigh->parms, RETRANS_TIME);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
if ((neigh->nud_state & (NUD_INCOMPLETE | NUD_PROBE)) &&
|
|
|
|
atomic_read(&neigh->probes) >= neigh_max_probes(neigh)) {
|
|
|
|
neigh->nud_state = NUD_FAILED;
|
|
|
|
notify = 1;
|
2009-06-11 13:16:28 +02:00
|
|
|
neigh_invalidate(neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
if (neigh->nud_state & NUD_IN_TIMER) {
|
|
|
|
if (time_before(next, jiffies + HZ/2))
|
|
|
|
next = jiffies + HZ/2;
|
2005-10-23 08:37:48 +02:00
|
|
|
if (!mod_timer(&neigh->timer, next))
|
|
|
|
neigh_hold(neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
if (neigh->nud_state & (NUD_INCOMPLETE | NUD_PROBE)) {
|
2011-08-09 10:15:58 +02:00
|
|
|
neigh_probe(neigh);
|
2008-02-18 03:39:54 +01:00
|
|
|
} else {
|
2008-02-12 06:45:44 +01:00
|
|
|
out:
|
2008-02-18 03:39:54 +01:00
|
|
|
write_unlock(&neigh->lock);
|
|
|
|
}
|
2007-08-09 08:12:56 +02:00
|
|
|
|
2006-07-31 05:43:36 +02:00
|
|
|
if (notify)
|
2007-08-09 08:12:56 +02:00
|
|
|
neigh_update_notify(neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
neigh_release(neigh);
|
|
|
|
}
|
|
|
|
|
|
|
|
int __neigh_event_send(struct neighbour *neigh, struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
int rc;
|
2011-08-09 10:15:58 +02:00
|
|
|
bool immediate_probe = false;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
write_lock_bh(&neigh->lock);
|
|
|
|
|
|
|
|
rc = 0;
|
|
|
|
if (neigh->nud_state & (NUD_CONNECTED | NUD_DELAY | NUD_PROBE))
|
|
|
|
goto out_unlock_bh;
|
|
|
|
|
|
|
|
if (!(neigh->nud_state & (NUD_STALE | NUD_INCOMPLETE))) {
|
2013-12-07 19:26:53 +01:00
|
|
|
if (NEIGH_VAR(neigh->parms, MCAST_PROBES) +
|
|
|
|
NEIGH_VAR(neigh->parms, APP_PROBES)) {
|
2011-08-09 10:15:58 +02:00
|
|
|
unsigned long next, now = jiffies;
|
|
|
|
|
2013-12-07 19:26:53 +01:00
|
|
|
atomic_set(&neigh->probes,
|
|
|
|
NEIGH_VAR(neigh->parms, UCAST_PROBES));
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh->nud_state = NUD_INCOMPLETE;
|
2011-08-09 10:15:58 +02:00
|
|
|
neigh->updated = now;
|
2013-12-07 19:26:53 +01:00
|
|
|
next = now + max(NEIGH_VAR(neigh->parms, RETRANS_TIME),
|
|
|
|
HZ/2);
|
2011-08-09 10:15:58 +02:00
|
|
|
neigh_add_timer(neigh, next);
|
|
|
|
immediate_probe = true;
|
2005-04-17 00:20:36 +02:00
|
|
|
} else {
|
|
|
|
neigh->nud_state = NUD_FAILED;
|
2006-03-21 01:52:52 +01:00
|
|
|
neigh->updated = jiffies;
|
2005-04-17 00:20:36 +02:00
|
|
|
write_unlock_bh(&neigh->lock);
|
|
|
|
|
2009-02-25 01:37:32 +01:00
|
|
|
kfree_skb(skb);
|
2005-04-17 00:20:36 +02:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
} else if (neigh->nud_state & NUD_STALE) {
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(2, "neigh %p is delayed\n", neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh->nud_state = NUD_DELAY;
|
2006-03-21 01:52:52 +01:00
|
|
|
neigh->updated = jiffies;
|
2013-12-07 19:26:53 +01:00
|
|
|
neigh_add_timer(neigh, jiffies +
|
|
|
|
NEIGH_VAR(neigh->parms, DELAY_PROBE_TIME));
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
if (neigh->nud_state == NUD_INCOMPLETE) {
|
|
|
|
if (skb) {
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
while (neigh->arp_queue_len_bytes + skb->truesize >
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR(neigh->parms, QUEUE_LEN_BYTES)) {
|
2005-04-17 00:20:36 +02:00
|
|
|
struct sk_buff *buff;
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
|
2008-09-23 10:11:18 +02:00
|
|
|
buff = __skb_dequeue(&neigh->arp_queue);
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
if (!buff)
|
|
|
|
break;
|
|
|
|
neigh->arp_queue_len_bytes -= buff->truesize;
|
2005-04-17 00:20:36 +02:00
|
|
|
kfree_skb(buff);
|
2008-07-17 05:50:49 +02:00
|
|
|
NEIGH_CACHE_STAT_INC(neigh->tbl, unres_discards);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2010-05-28 01:09:39 +02:00
|
|
|
skb_dst_force(skb);
|
2005-04-17 00:20:36 +02:00
|
|
|
__skb_queue_tail(&neigh->arp_queue, skb);
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
neigh->arp_queue_len_bytes += skb->truesize;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
rc = 1;
|
|
|
|
}
|
|
|
|
out_unlock_bh:
|
2011-08-09 10:15:58 +02:00
|
|
|
if (immediate_probe)
|
|
|
|
neigh_probe(neigh);
|
|
|
|
else
|
|
|
|
write_unlock(&neigh->lock);
|
|
|
|
local_bh_enable();
|
2005-04-17 00:20:36 +02:00
|
|
|
return rc;
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(__neigh_event_send);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2011-07-14 16:53:20 +02:00
|
|
|
static void neigh_update_hhs(struct neighbour *neigh)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct hh_cache *hh;
|
2007-10-09 10:40:57 +02:00
|
|
|
void (*update)(struct hh_cache*, const struct net_device*, const unsigned char *)
|
2010-07-15 03:02:16 +02:00
|
|
|
= NULL;
|
|
|
|
|
|
|
|
if (neigh->dev->header_ops)
|
|
|
|
update = neigh->dev->header_ops->cache_update;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (update) {
|
2011-07-14 16:53:20 +02:00
|
|
|
hh = &neigh->hh;
|
|
|
|
if (hh->hh_len) {
|
2006-12-08 00:08:17 +01:00
|
|
|
write_seqlock_bh(&hh->hh_lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
update(hh, neigh->dev, neigh->ha);
|
2006-12-08 00:08:17 +01:00
|
|
|
write_sequnlock_bh(&hh->hh_lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/* Generic update routine.
|
|
|
|
-- lladdr is new lladdr or NULL, if it is not supplied.
|
|
|
|
-- new is new state.
|
|
|
|
-- flags
|
|
|
|
NEIGH_UPDATE_F_OVERRIDE allows to override existing lladdr,
|
|
|
|
if it is different.
|
|
|
|
NEIGH_UPDATE_F_WEAK_OVERRIDE will suspect existing "connected"
|
2007-02-09 15:24:36 +01:00
|
|
|
lladdr instead of overriding it
|
2005-04-17 00:20:36 +02:00
|
|
|
if it is different.
|
|
|
|
It also allows to retain current state
|
|
|
|
if lladdr is unchanged.
|
|
|
|
NEIGH_UPDATE_F_ADMIN means that the change is administrative.
|
|
|
|
|
2007-02-09 15:24:36 +01:00
|
|
|
NEIGH_UPDATE_F_OVERRIDE_ISROUTER allows to override existing
|
2005-04-17 00:20:36 +02:00
|
|
|
NTF_ROUTER flag.
|
|
|
|
NEIGH_UPDATE_F_ISROUTER indicates if the neighbour is known as
|
|
|
|
a router.
|
|
|
|
|
|
|
|
Caller MUST hold reference count on the entry.
|
|
|
|
*/
|
|
|
|
|
|
|
|
int neigh_update(struct neighbour *neigh, const u8 *lladdr, u8 new,
|
|
|
|
u32 flags)
|
|
|
|
{
|
|
|
|
u8 old;
|
|
|
|
int err;
|
|
|
|
int notify = 0;
|
|
|
|
struct net_device *dev;
|
|
|
|
int update_isrouter = 0;
|
|
|
|
|
|
|
|
write_lock_bh(&neigh->lock);
|
|
|
|
|
|
|
|
dev = neigh->dev;
|
|
|
|
old = neigh->nud_state;
|
|
|
|
err = -EPERM;
|
|
|
|
|
2007-02-09 15:24:36 +01:00
|
|
|
if (!(flags & NEIGH_UPDATE_F_ADMIN) &&
|
2005-04-17 00:20:36 +02:00
|
|
|
(old & (NUD_NOARP | NUD_PERMANENT)))
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
if (!(new & NUD_VALID)) {
|
|
|
|
neigh_del_timer(neigh);
|
|
|
|
if (old & NUD_CONNECTED)
|
|
|
|
neigh_suspect(neigh);
|
|
|
|
neigh->nud_state = new;
|
|
|
|
err = 0;
|
|
|
|
notify = old & NUD_VALID;
|
2009-06-11 13:16:28 +02:00
|
|
|
if ((old & (NUD_INCOMPLETE | NUD_PROBE)) &&
|
|
|
|
(new & NUD_FAILED)) {
|
|
|
|
neigh_invalidate(neigh);
|
|
|
|
notify = 1;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Compare new lladdr with cached one */
|
|
|
|
if (!dev->addr_len) {
|
|
|
|
/* First case: device needs no address. */
|
|
|
|
lladdr = neigh->ha;
|
|
|
|
} else if (lladdr) {
|
|
|
|
/* The second case: if something is already cached
|
|
|
|
and a new address is proposed:
|
|
|
|
- compare new & old
|
|
|
|
- if they are different, check override flag
|
|
|
|
*/
|
2007-02-09 15:24:36 +01:00
|
|
|
if ((old & NUD_VALID) &&
|
2005-04-17 00:20:36 +02:00
|
|
|
!memcmp(lladdr, neigh->ha, dev->addr_len))
|
|
|
|
lladdr = neigh->ha;
|
|
|
|
} else {
|
|
|
|
/* No address is supplied; if we know something,
|
|
|
|
use it, otherwise discard the request.
|
|
|
|
*/
|
|
|
|
err = -EINVAL;
|
|
|
|
if (!(old & NUD_VALID))
|
|
|
|
goto out;
|
|
|
|
lladdr = neigh->ha;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (new & NUD_CONNECTED)
|
|
|
|
neigh->confirmed = jiffies;
|
|
|
|
neigh->updated = jiffies;
|
|
|
|
|
|
|
|
/* If entry was valid and address is not changed,
|
|
|
|
do not change entry state, if new one is STALE.
|
|
|
|
*/
|
|
|
|
err = 0;
|
|
|
|
update_isrouter = flags & NEIGH_UPDATE_F_OVERRIDE_ISROUTER;
|
|
|
|
if (old & NUD_VALID) {
|
|
|
|
if (lladdr != neigh->ha && !(flags & NEIGH_UPDATE_F_OVERRIDE)) {
|
|
|
|
update_isrouter = 0;
|
|
|
|
if ((flags & NEIGH_UPDATE_F_WEAK_OVERRIDE) &&
|
|
|
|
(old & NUD_CONNECTED)) {
|
|
|
|
lladdr = neigh->ha;
|
|
|
|
new = NUD_STALE;
|
|
|
|
} else
|
|
|
|
goto out;
|
|
|
|
} else {
|
|
|
|
if (lladdr == neigh->ha && new == NUD_STALE &&
|
|
|
|
((flags & NEIGH_UPDATE_F_WEAK_OVERRIDE) ||
|
|
|
|
(old & NUD_CONNECTED))
|
|
|
|
)
|
|
|
|
new = old;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (new != old) {
|
|
|
|
neigh_del_timer(neigh);
|
2007-12-21 00:49:05 +01:00
|
|
|
if (new & NUD_IN_TIMER)
|
2007-02-09 15:24:36 +01:00
|
|
|
neigh_add_timer(neigh, (jiffies +
|
|
|
|
((new & NUD_REACHABLE) ?
|
2005-09-27 21:07:44 +02:00
|
|
|
neigh->parms->reachable_time :
|
|
|
|
0)));
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh->nud_state = new;
|
2013-12-15 22:39:56 +01:00
|
|
|
notify = 1;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
if (lladdr != neigh->ha) {
|
2010-10-07 12:44:07 +02:00
|
|
|
write_seqlock(&neigh->ha_lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
memcpy(&neigh->ha, lladdr, dev->addr_len);
|
2010-10-07 12:44:07 +02:00
|
|
|
write_sequnlock(&neigh->ha_lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
neigh_update_hhs(neigh);
|
|
|
|
if (!(new & NUD_CONNECTED))
|
|
|
|
neigh->confirmed = jiffies -
|
2013-12-07 19:26:53 +01:00
|
|
|
(NEIGH_VAR(neigh->parms, BASE_REACHABLE_TIME) << 1);
|
2005-04-17 00:20:36 +02:00
|
|
|
notify = 1;
|
|
|
|
}
|
|
|
|
if (new == old)
|
|
|
|
goto out;
|
|
|
|
if (new & NUD_CONNECTED)
|
|
|
|
neigh_connect(neigh);
|
|
|
|
else
|
|
|
|
neigh_suspect(neigh);
|
|
|
|
if (!(old & NUD_VALID)) {
|
|
|
|
struct sk_buff *skb;
|
|
|
|
|
|
|
|
/* Again: avoid dead loop if something went wrong */
|
|
|
|
|
|
|
|
while (neigh->nud_state & NUD_VALID &&
|
|
|
|
(skb = __skb_dequeue(&neigh->arp_queue)) != NULL) {
|
2011-07-18 08:09:49 +02:00
|
|
|
struct dst_entry *dst = skb_dst(skb);
|
|
|
|
struct neighbour *n2, *n1 = neigh;
|
2005-04-17 00:20:36 +02:00
|
|
|
write_unlock_bh(&neigh->lock);
|
2011-10-18 00:32:42 +02:00
|
|
|
|
|
|
|
rcu_read_lock();
|
2012-07-03 07:15:37 +02:00
|
|
|
|
|
|
|
/* Why not just use 'neigh' as-is? The problem is that
|
|
|
|
* things such as shaper, eql, and sch_teql can end up
|
|
|
|
* using alternative, different, neigh objects to output
|
|
|
|
* the packet in the output path. So what we need to do
|
|
|
|
* here is re-lookup the top-level neigh in the path so
|
|
|
|
* we can reinject the packet there.
|
|
|
|
*/
|
|
|
|
n2 = NULL;
|
|
|
|
if (dst) {
|
|
|
|
n2 = dst_neigh_lookup_skb(dst, skb);
|
|
|
|
if (n2)
|
|
|
|
n1 = n2;
|
|
|
|
}
|
2011-07-17 22:34:11 +02:00
|
|
|
n1->output(n1, skb);
|
2012-07-03 07:15:37 +02:00
|
|
|
if (n2)
|
|
|
|
neigh_release(n2);
|
2011-10-18 00:32:42 +02:00
|
|
|
rcu_read_unlock();
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
write_lock_bh(&neigh->lock);
|
|
|
|
}
|
2013-06-28 11:37:42 +02:00
|
|
|
__skb_queue_purge(&neigh->arp_queue);
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
neigh->arp_queue_len_bytes = 0;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
out:
|
|
|
|
if (update_isrouter) {
|
|
|
|
neigh->flags = (flags & NEIGH_UPDATE_F_ISROUTER) ?
|
|
|
|
(neigh->flags | NTF_ROUTER) :
|
|
|
|
(neigh->flags & ~NTF_ROUTER);
|
|
|
|
}
|
|
|
|
write_unlock_bh(&neigh->lock);
|
2006-07-31 05:43:36 +02:00
|
|
|
|
|
|
|
if (notify)
|
2007-08-09 08:12:56 +02:00
|
|
|
neigh_update_notify(neigh);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
return err;
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_update);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2013-12-11 13:48:20 +01:00
|
|
|
/* Update the neigh to listen temporarily for probe responses, even if it is
|
|
|
|
* in a NUD_FAILED state. The caller has to hold neigh->lock for writing.
|
|
|
|
*/
|
|
|
|
void __neigh_set_probe_once(struct neighbour *neigh)
|
|
|
|
{
|
|
|
|
neigh->updated = jiffies;
|
|
|
|
if (!(neigh->nud_state & NUD_FAILED))
|
|
|
|
return;
|
|
|
|
neigh->nud_state = NUD_PROBE;
|
|
|
|
atomic_set(&neigh->probes, NEIGH_VAR(neigh->parms, UCAST_PROBES));
|
|
|
|
neigh_add_timer(neigh,
|
|
|
|
jiffies + NEIGH_VAR(neigh->parms, RETRANS_TIME));
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(__neigh_set_probe_once);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
struct neighbour *neigh_event_ns(struct neigh_table *tbl,
|
|
|
|
u8 *lladdr, void *saddr,
|
|
|
|
struct net_device *dev)
|
|
|
|
{
|
|
|
|
struct neighbour *neigh = __neigh_lookup(tbl, saddr, dev,
|
|
|
|
lladdr || !dev->addr_len);
|
|
|
|
if (neigh)
|
2007-02-09 15:24:36 +01:00
|
|
|
neigh_update(neigh, lladdr, NUD_STALE,
|
2005-04-17 00:20:36 +02:00
|
|
|
NEIGH_UPDATE_F_OVERRIDE);
|
|
|
|
return neigh;
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_event_ns);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-11 18:16:57 +02:00
|
|
|
/* called with read_lock_bh(&n->lock); */
|
2011-07-14 16:53:20 +02:00
|
|
|
static void neigh_hh_init(struct neighbour *n, struct dst_entry *dst)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct net_device *dev = dst->dev;
|
2011-07-14 16:53:20 +02:00
|
|
|
__be16 prot = dst->ops->protocol;
|
|
|
|
struct hh_cache *hh = &n->hh;
|
2010-10-07 12:44:07 +02:00
|
|
|
|
|
|
|
write_lock_bh(&n->lock);
|
2010-10-11 18:16:57 +02:00
|
|
|
|
2011-07-14 16:53:20 +02:00
|
|
|
/* Only one thread can come in here and initialize the
|
|
|
|
* hh_cache entry.
|
|
|
|
*/
|
2011-07-17 02:45:02 +02:00
|
|
|
if (!hh->hh_len)
|
|
|
|
dev->header_ops->cache(n, hh, prot);
|
2010-10-11 18:16:57 +02:00
|
|
|
|
2010-10-07 12:44:07 +02:00
|
|
|
write_unlock_bh(&n->lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/* This function can be used in contexts, where only old dev_queue_xmit
|
2010-10-07 02:49:21 +02:00
|
|
|
* worked, f.e. if you want to override normal output path (eql, shaper),
|
|
|
|
* but resolution is not made yet.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
|
|
|
|
2011-07-17 22:34:11 +02:00
|
|
|
int neigh_compat_output(struct neighbour *neigh, struct sk_buff *skb)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct net_device *dev = skb->dev;
|
|
|
|
|
2007-03-11 02:16:10 +01:00
|
|
|
__skb_pull(skb, skb_network_offset(skb));
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-10-09 10:36:32 +02:00
|
|
|
if (dev_hard_header(skb, dev, ntohs(skb->protocol), NULL, NULL,
|
|
|
|
skb->len) < 0 &&
|
2013-12-31 22:23:35 +01:00
|
|
|
dev_rebuild_header(skb))
|
2005-04-17 00:20:36 +02:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
return dev_queue_xmit(skb);
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_compat_output);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/* Slow and careful. */
|
|
|
|
|
2011-07-17 22:34:11 +02:00
|
|
|
int neigh_resolve_output(struct neighbour *neigh, struct sk_buff *skb)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2009-06-02 07:19:30 +02:00
|
|
|
struct dst_entry *dst = skb_dst(skb);
|
2005-04-17 00:20:36 +02:00
|
|
|
int rc = 0;
|
|
|
|
|
2011-07-17 22:34:11 +02:00
|
|
|
if (!dst)
|
2005-04-17 00:20:36 +02:00
|
|
|
goto discard;
|
|
|
|
|
|
|
|
if (!neigh_event_send(neigh, skb)) {
|
|
|
|
int err;
|
|
|
|
struct net_device *dev = neigh->dev;
|
2010-10-07 12:44:07 +02:00
|
|
|
unsigned int seq;
|
2010-10-11 18:16:57 +02:00
|
|
|
|
2011-07-14 16:53:20 +02:00
|
|
|
if (dev->header_ops->cache && !neigh->hh.hh_len)
|
|
|
|
neigh_hh_init(neigh, dst);
|
2010-10-11 18:16:57 +02:00
|
|
|
|
2010-10-07 12:44:07 +02:00
|
|
|
do {
|
2012-10-05 21:10:15 +02:00
|
|
|
__skb_pull(skb, skb_network_offset(skb));
|
2010-10-07 12:44:07 +02:00
|
|
|
seq = read_seqbegin(&neigh->ha_lock);
|
|
|
|
err = dev_hard_header(skb, dev, ntohs(skb->protocol),
|
|
|
|
neigh->ha, NULL, skb->len);
|
|
|
|
} while (read_seqretry(&neigh->ha_lock, seq));
|
2010-10-11 18:16:57 +02:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
if (err >= 0)
|
2011-07-17 03:06:24 +02:00
|
|
|
rc = dev_queue_xmit(skb);
|
2005-04-17 00:20:36 +02:00
|
|
|
else
|
|
|
|
goto out_kfree_skb;
|
|
|
|
}
|
|
|
|
out:
|
|
|
|
return rc;
|
|
|
|
discard:
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(1, "%s: dst=%p neigh=%p\n", __func__, dst, neigh);
|
2005-04-17 00:20:36 +02:00
|
|
|
out_kfree_skb:
|
|
|
|
rc = -EINVAL;
|
|
|
|
kfree_skb(skb);
|
|
|
|
goto out;
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_resolve_output);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/* As fast as possible without hh cache */
|
|
|
|
|
2011-07-17 22:34:11 +02:00
|
|
|
int neigh_connected_output(struct neighbour *neigh, struct sk_buff *skb)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct net_device *dev = neigh->dev;
|
2010-10-07 12:44:07 +02:00
|
|
|
unsigned int seq;
|
2011-07-17 22:34:11 +02:00
|
|
|
int err;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-07 12:44:07 +02:00
|
|
|
do {
|
2012-10-05 21:10:15 +02:00
|
|
|
__skb_pull(skb, skb_network_offset(skb));
|
2010-10-07 12:44:07 +02:00
|
|
|
seq = read_seqbegin(&neigh->ha_lock);
|
|
|
|
err = dev_hard_header(skb, dev, ntohs(skb->protocol),
|
|
|
|
neigh->ha, NULL, skb->len);
|
|
|
|
} while (read_seqretry(&neigh->ha_lock, seq));
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
if (err >= 0)
|
2011-07-17 03:06:24 +02:00
|
|
|
err = dev_queue_xmit(skb);
|
2005-04-17 00:20:36 +02:00
|
|
|
else {
|
|
|
|
err = -EINVAL;
|
|
|
|
kfree_skb(skb);
|
|
|
|
}
|
|
|
|
return err;
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_connected_output);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2011-07-17 22:34:11 +02:00
|
|
|
int neigh_direct_output(struct neighbour *neigh, struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
return dev_queue_xmit(skb);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(neigh_direct_output);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
static void neigh_proxy_process(unsigned long arg)
|
|
|
|
{
|
|
|
|
struct neigh_table *tbl = (struct neigh_table *)arg;
|
|
|
|
long sched_next = 0;
|
|
|
|
unsigned long now = jiffies;
|
2008-09-23 10:11:18 +02:00
|
|
|
struct sk_buff *skb, *n;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
spin_lock(&tbl->proxy_queue.lock);
|
|
|
|
|
2008-09-23 10:11:18 +02:00
|
|
|
skb_queue_walk_safe(&tbl->proxy_queue, skb, n) {
|
|
|
|
long tdif = NEIGH_CB(skb)->sched_next - now;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (tdif <= 0) {
|
2008-09-23 10:11:18 +02:00
|
|
|
struct net_device *dev = skb->dev;
|
2011-08-22 21:32:42 +02:00
|
|
|
|
2008-09-23 10:11:18 +02:00
|
|
|
__skb_unlink(skb, &tbl->proxy_queue);
|
2011-08-22 21:32:42 +02:00
|
|
|
if (tbl->proxy_redo && netif_running(dev)) {
|
|
|
|
rcu_read_lock();
|
2008-09-23 10:11:18 +02:00
|
|
|
tbl->proxy_redo(skb);
|
2011-08-22 21:32:42 +02:00
|
|
|
rcu_read_unlock();
|
|
|
|
} else {
|
2008-09-23 10:11:18 +02:00
|
|
|
kfree_skb(skb);
|
2011-08-22 21:32:42 +02:00
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
dev_put(dev);
|
|
|
|
} else if (!sched_next || tdif < sched_next)
|
|
|
|
sched_next = tdif;
|
|
|
|
}
|
|
|
|
del_timer(&tbl->proxy_timer);
|
|
|
|
if (sched_next)
|
|
|
|
mod_timer(&tbl->proxy_timer, jiffies + sched_next);
|
|
|
|
spin_unlock(&tbl->proxy_queue.lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
void pneigh_enqueue(struct neigh_table *tbl, struct neigh_parms *p,
|
|
|
|
struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
unsigned long now = jiffies;
|
2014-01-11 13:15:59 +01:00
|
|
|
|
|
|
|
unsigned long sched_next = now + (prandom_u32() %
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR(p, PROXY_DELAY));
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2013-12-07 19:26:53 +01:00
|
|
|
if (tbl->proxy_queue.qlen > NEIGH_VAR(p, PROXY_QLEN)) {
|
2005-04-17 00:20:36 +02:00
|
|
|
kfree_skb(skb);
|
|
|
|
return;
|
|
|
|
}
|
2005-08-15 02:24:31 +02:00
|
|
|
|
|
|
|
NEIGH_CB(skb)->sched_next = sched_next;
|
|
|
|
NEIGH_CB(skb)->flags |= LOCALLY_ENQUEUED;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
spin_lock(&tbl->proxy_queue.lock);
|
|
|
|
if (del_timer(&tbl->proxy_timer)) {
|
|
|
|
if (time_before(tbl->proxy_timer.expires, sched_next))
|
|
|
|
sched_next = tbl->proxy_timer.expires;
|
|
|
|
}
|
2009-06-02 07:19:30 +02:00
|
|
|
skb_dst_drop(skb);
|
2005-04-17 00:20:36 +02:00
|
|
|
dev_hold(skb->dev);
|
|
|
|
__skb_queue_tail(&tbl->proxy_queue, skb);
|
|
|
|
mod_timer(&tbl->proxy_timer, sched_next);
|
|
|
|
spin_unlock(&tbl->proxy_queue.lock);
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(pneigh_enqueue);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2009-07-13 20:17:49 +02:00
|
|
|
static inline struct neigh_parms *lookup_neigh_parms(struct neigh_table *tbl,
|
2008-01-24 09:13:18 +01:00
|
|
|
struct net *net, int ifindex)
|
|
|
|
{
|
|
|
|
struct neigh_parms *p;
|
|
|
|
|
|
|
|
for (p = &tbl->parms; p; p = p->next) {
|
2008-03-25 19:57:35 +01:00
|
|
|
if ((p->dev && p->dev->ifindex == ifindex && net_eq(neigh_parms_net(p), net)) ||
|
2013-06-20 04:01:33 +02:00
|
|
|
(!p->dev && !ifindex && net_eq(net, &init_net)))
|
2008-01-24 09:13:18 +01:00
|
|
|
return p;
|
|
|
|
}
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
struct neigh_parms *neigh_parms_alloc(struct net_device *dev,
|
|
|
|
struct neigh_table *tbl)
|
|
|
|
{
|
2013-06-20 04:01:32 +02:00
|
|
|
struct neigh_parms *p;
|
2008-11-21 05:14:53 +01:00
|
|
|
struct net *net = dev_net(dev);
|
|
|
|
const struct net_device_ops *ops = dev->netdev_ops;
|
2008-01-24 09:13:18 +01:00
|
|
|
|
2013-06-20 04:01:32 +02:00
|
|
|
p = kmemdup(&tbl->parms, sizeof(*p), GFP_KERNEL);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (p) {
|
|
|
|
p->tbl = tbl;
|
|
|
|
atomic_set(&p->refcnt, 1);
|
|
|
|
p->reachable_time =
|
2013-12-07 19:26:53 +01:00
|
|
|
neigh_rand_reach_time(NEIGH_VAR(p, BASE_REACHABLE_TIME));
|
2013-08-02 19:07:38 +02:00
|
|
|
dev_hold(dev);
|
|
|
|
p->dev = dev;
|
|
|
|
write_pnet(&p->net, hold_net(net));
|
|
|
|
p->sysctl_table = NULL;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2008-11-21 05:14:53 +01:00
|
|
|
if (ops->ndo_neigh_setup && ops->ndo_neigh_setup(dev, p)) {
|
2013-08-02 19:07:38 +02:00
|
|
|
release_net(net);
|
|
|
|
dev_put(dev);
|
2008-01-15 07:59:59 +01:00
|
|
|
kfree(p);
|
|
|
|
return NULL;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2008-01-15 07:59:59 +01:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
write_lock_bh(&tbl->lock);
|
|
|
|
p->next = tbl->parms.next;
|
|
|
|
tbl->parms.next = p;
|
|
|
|
write_unlock_bh(&tbl->lock);
|
2013-12-07 19:26:56 +01:00
|
|
|
|
|
|
|
neigh_parms_data_state_cleanall(p);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
return p;
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_parms_alloc);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
static void neigh_rcu_free_parms(struct rcu_head *head)
|
|
|
|
{
|
|
|
|
struct neigh_parms *parms =
|
|
|
|
container_of(head, struct neigh_parms, rcu_head);
|
|
|
|
|
|
|
|
neigh_parms_put(parms);
|
|
|
|
}
|
|
|
|
|
|
|
|
void neigh_parms_release(struct neigh_table *tbl, struct neigh_parms *parms)
|
|
|
|
{
|
|
|
|
struct neigh_parms **p;
|
|
|
|
|
|
|
|
if (!parms || parms == &tbl->parms)
|
|
|
|
return;
|
|
|
|
write_lock_bh(&tbl->lock);
|
|
|
|
for (p = &tbl->parms.next; *p; p = &(*p)->next) {
|
|
|
|
if (*p == parms) {
|
|
|
|
*p = parms->next;
|
|
|
|
parms->dead = 1;
|
|
|
|
write_unlock_bh(&tbl->lock);
|
2008-01-21 01:39:03 +01:00
|
|
|
if (parms->dev)
|
|
|
|
dev_put(parms->dev);
|
2005-04-17 00:20:36 +02:00
|
|
|
call_rcu(&parms->rcu_head, neigh_rcu_free_parms);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
write_unlock_bh(&tbl->lock);
|
2013-04-15 17:17:19 +02:00
|
|
|
neigh_dbg(1, "%s: not found\n", __func__);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_parms_release);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2008-01-24 09:30:58 +01:00
|
|
|
static void neigh_parms_destroy(struct neigh_parms *parms)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2008-03-25 19:49:59 +01:00
|
|
|
release_net(neigh_parms_net(parms));
|
2005-04-17 00:20:36 +02:00
|
|
|
kfree(parms);
|
|
|
|
}
|
|
|
|
|
2007-04-17 21:45:31 +02:00
|
|
|
static struct lock_class_key neigh_table_proxy_queue_class;
|
|
|
|
|
2012-04-13 09:34:44 +02:00
|
|
|
static void neigh_table_init_no_netlink(struct neigh_table *tbl)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
unsigned long now = jiffies;
|
|
|
|
unsigned long phsize;
|
|
|
|
|
2008-11-12 09:54:54 +01:00
|
|
|
write_pnet(&tbl->parms.net, &init_net);
|
2005-04-17 00:20:36 +02:00
|
|
|
atomic_set(&tbl->parms.refcnt, 1);
|
|
|
|
tbl->parms.reachable_time =
|
2013-12-07 19:26:53 +01:00
|
|
|
neigh_rand_reach_time(NEIGH_VAR(&tbl->parms, BASE_REACHABLE_TIME));
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
tbl->stats = alloc_percpu(struct neigh_statistics);
|
|
|
|
if (!tbl->stats)
|
|
|
|
panic("cannot create neighbour cache statistics");
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
#ifdef CONFIG_PROC_FS
|
2008-11-12 01:47:44 +01:00
|
|
|
if (!proc_create_data(tbl->id, 0, init_net.proc_net_stat,
|
|
|
|
&neigh_stat_seq_fops, tbl))
|
2005-04-17 00:20:36 +02:00
|
|
|
panic("cannot create neighbour proc dir entry");
|
|
|
|
#endif
|
|
|
|
|
2011-07-11 10:28:12 +02:00
|
|
|
RCU_INIT_POINTER(tbl->nht, neigh_hash_alloc(3));
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
phsize = (PNEIGH_HASHMASK + 1) * sizeof(struct pneigh_entry *);
|
2006-04-07 23:52:59 +02:00
|
|
|
tbl->phash_buckets = kzalloc(phsize, GFP_KERNEL);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
if (!tbl->nht || !tbl->phash_buckets)
|
2005-04-17 00:20:36 +02:00
|
|
|
panic("cannot allocate neighbour cache hashes");
|
|
|
|
|
2013-01-24 01:44:23 +01:00
|
|
|
if (!tbl->entry_size)
|
|
|
|
tbl->entry_size = ALIGN(offsetof(struct neighbour, primary_key) +
|
|
|
|
tbl->key_len, NEIGH_PRIV_ALIGN);
|
|
|
|
else
|
|
|
|
WARN_ON(tbl->entry_size % NEIGH_PRIV_ALIGN);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
rwlock_init(&tbl->lock);
|
2012-08-21 22:18:23 +02:00
|
|
|
INIT_DEFERRABLE_WORK(&tbl->gc_work, neigh_periodic_work);
|
2014-01-22 07:53:33 +01:00
|
|
|
queue_delayed_work(system_power_efficient_wq, &tbl->gc_work,
|
|
|
|
tbl->parms.reachable_time);
|
2008-01-24 06:20:07 +01:00
|
|
|
setup_timer(&tbl->proxy_timer, neigh_proxy_process, (unsigned long)tbl);
|
2007-04-17 21:45:31 +02:00
|
|
|
skb_queue_head_init_class(&tbl->proxy_queue,
|
|
|
|
&neigh_table_proxy_queue_class);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
tbl->last_flush = now;
|
|
|
|
tbl->last_rand = now + tbl->parms.reachable_time * 20;
|
[NEIGH]: Fix IP-over-ATM and ARP interaction.
The classical IP over ATM code maintains its own IPv4 <-> <ATM stuff>
ARP table, using the standard neighbour-table code. The
neigh_table_init function adds this neighbour table to a linked list
of all neighbor tables which is used by the functions neigh_delete()
neigh_add() and neightbl_set(), all called by the netlink code.
Once the ATM neighbour table is added to the list, there are two
tables with family == AF_INET there, and ARP entries sent via netlink
go into the first table with matching family. This is indeterminate
and often wrong.
To see the bug, on a kernel with CLIP enabled, create a standard IPv4
ARP entry by pinging an unused address on a local subnet. Then attempt
to complete that entry by doing
ip neigh replace <ip address> lladdr <some mac address> nud reachable
Looking at the ARP tables by using
ip neigh show
will reveal two ARP entries for the same address. One of these can be
found in /proc/net/arp, and the other in /proc/net/atm/arp.
This patch adds a new function, neigh_table_init_no_netlink() which
does everything the neigh_table_init() does, except add the table to
the netlink all-arp-tables chain. In addition neigh_table_init() has a
check that all tables on the chain have a distinct address family.
The init call in clip.c is changed to call
neigh_table_init_no_netlink().
Since ATM ARP tables are rather more complicated than can currently be
handled by the available rtattrs in the netlink protocol, no
functionality is lost by this patch, and non-ATM ARP manipulation via
netlink is rescued. A more complete solution would involve a rtattr
for ATM ARP entries and some way for the netlink code to give
neigh_add and friends more information than just address family with
which to find the correct ARP table.
[ I've changed the assertion checking in neigh_table_init() to not
use BUG_ON() while holding neigh_tbl_lock. Instead we remember that
we found an existing tbl with the same family, and after dropping
the lock we'll give a diagnostic kernel log message and a stack dump.
-DaveM ]
Signed-off-by: Simon Kelley <simon@thekelleys.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
2006-05-12 23:56:08 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
void neigh_table_init(struct neigh_table *tbl)
|
|
|
|
{
|
|
|
|
struct neigh_table *tmp;
|
|
|
|
|
|
|
|
neigh_table_init_no_netlink(tbl);
|
2005-04-17 00:20:36 +02:00
|
|
|
write_lock(&neigh_tbl_lock);
|
[NEIGH]: Fix IP-over-ATM and ARP interaction.
The classical IP over ATM code maintains its own IPv4 <-> <ATM stuff>
ARP table, using the standard neighbour-table code. The
neigh_table_init function adds this neighbour table to a linked list
of all neighbor tables which is used by the functions neigh_delete()
neigh_add() and neightbl_set(), all called by the netlink code.
Once the ATM neighbour table is added to the list, there are two
tables with family == AF_INET there, and ARP entries sent via netlink
go into the first table with matching family. This is indeterminate
and often wrong.
To see the bug, on a kernel with CLIP enabled, create a standard IPv4
ARP entry by pinging an unused address on a local subnet. Then attempt
to complete that entry by doing
ip neigh replace <ip address> lladdr <some mac address> nud reachable
Looking at the ARP tables by using
ip neigh show
will reveal two ARP entries for the same address. One of these can be
found in /proc/net/arp, and the other in /proc/net/atm/arp.
This patch adds a new function, neigh_table_init_no_netlink() which
does everything the neigh_table_init() does, except add the table to
the netlink all-arp-tables chain. In addition neigh_table_init() has a
check that all tables on the chain have a distinct address family.
The init call in clip.c is changed to call
neigh_table_init_no_netlink().
Since ATM ARP tables are rather more complicated than can currently be
handled by the available rtattrs in the netlink protocol, no
functionality is lost by this patch, and non-ATM ARP manipulation via
netlink is rescued. A more complete solution would involve a rtattr
for ATM ARP entries and some way for the netlink code to give
neigh_add and friends more information than just address family with
which to find the correct ARP table.
[ I've changed the assertion checking in neigh_table_init() to not
use BUG_ON() while holding neigh_tbl_lock. Instead we remember that
we found an existing tbl with the same family, and after dropping
the lock we'll give a diagnostic kernel log message and a stack dump.
-DaveM ]
Signed-off-by: Simon Kelley <simon@thekelleys.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
2006-05-12 23:56:08 +02:00
|
|
|
for (tmp = neigh_tables; tmp; tmp = tmp->next) {
|
|
|
|
if (tmp->family == tbl->family)
|
|
|
|
break;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
tbl->next = neigh_tables;
|
|
|
|
neigh_tables = tbl;
|
|
|
|
write_unlock(&neigh_tbl_lock);
|
[NEIGH]: Fix IP-over-ATM and ARP interaction.
The classical IP over ATM code maintains its own IPv4 <-> <ATM stuff>
ARP table, using the standard neighbour-table code. The
neigh_table_init function adds this neighbour table to a linked list
of all neighbor tables which is used by the functions neigh_delete()
neigh_add() and neightbl_set(), all called by the netlink code.
Once the ATM neighbour table is added to the list, there are two
tables with family == AF_INET there, and ARP entries sent via netlink
go into the first table with matching family. This is indeterminate
and often wrong.
To see the bug, on a kernel with CLIP enabled, create a standard IPv4
ARP entry by pinging an unused address on a local subnet. Then attempt
to complete that entry by doing
ip neigh replace <ip address> lladdr <some mac address> nud reachable
Looking at the ARP tables by using
ip neigh show
will reveal two ARP entries for the same address. One of these can be
found in /proc/net/arp, and the other in /proc/net/atm/arp.
This patch adds a new function, neigh_table_init_no_netlink() which
does everything the neigh_table_init() does, except add the table to
the netlink all-arp-tables chain. In addition neigh_table_init() has a
check that all tables on the chain have a distinct address family.
The init call in clip.c is changed to call
neigh_table_init_no_netlink().
Since ATM ARP tables are rather more complicated than can currently be
handled by the available rtattrs in the netlink protocol, no
functionality is lost by this patch, and non-ATM ARP manipulation via
netlink is rescued. A more complete solution would involve a rtattr
for ATM ARP entries and some way for the netlink code to give
neigh_add and friends more information than just address family with
which to find the correct ARP table.
[ I've changed the assertion checking in neigh_table_init() to not
use BUG_ON() while holding neigh_tbl_lock. Instead we remember that
we found an existing tbl with the same family, and after dropping
the lock we'll give a diagnostic kernel log message and a stack dump.
-DaveM ]
Signed-off-by: Simon Kelley <simon@thekelleys.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
2006-05-12 23:56:08 +02:00
|
|
|
|
|
|
|
if (unlikely(tmp)) {
|
2012-05-16 21:58:40 +02:00
|
|
|
pr_err("Registering multiple tables for family %d\n",
|
|
|
|
tbl->family);
|
[NEIGH]: Fix IP-over-ATM and ARP interaction.
The classical IP over ATM code maintains its own IPv4 <-> <ATM stuff>
ARP table, using the standard neighbour-table code. The
neigh_table_init function adds this neighbour table to a linked list
of all neighbor tables which is used by the functions neigh_delete()
neigh_add() and neightbl_set(), all called by the netlink code.
Once the ATM neighbour table is added to the list, there are two
tables with family == AF_INET there, and ARP entries sent via netlink
go into the first table with matching family. This is indeterminate
and often wrong.
To see the bug, on a kernel with CLIP enabled, create a standard IPv4
ARP entry by pinging an unused address on a local subnet. Then attempt
to complete that entry by doing
ip neigh replace <ip address> lladdr <some mac address> nud reachable
Looking at the ARP tables by using
ip neigh show
will reveal two ARP entries for the same address. One of these can be
found in /proc/net/arp, and the other in /proc/net/atm/arp.
This patch adds a new function, neigh_table_init_no_netlink() which
does everything the neigh_table_init() does, except add the table to
the netlink all-arp-tables chain. In addition neigh_table_init() has a
check that all tables on the chain have a distinct address family.
The init call in clip.c is changed to call
neigh_table_init_no_netlink().
Since ATM ARP tables are rather more complicated than can currently be
handled by the available rtattrs in the netlink protocol, no
functionality is lost by this patch, and non-ATM ARP manipulation via
netlink is rescued. A more complete solution would involve a rtattr
for ATM ARP entries and some way for the netlink code to give
neigh_add and friends more information than just address family with
which to find the correct ARP table.
[ I've changed the assertion checking in neigh_table_init() to not
use BUG_ON() while holding neigh_tbl_lock. Instead we remember that
we found an existing tbl with the same family, and after dropping
the lock we'll give a diagnostic kernel log message and a stack dump.
-DaveM ]
Signed-off-by: Simon Kelley <simon@thekelleys.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
2006-05-12 23:56:08 +02:00
|
|
|
dump_stack();
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_table_init);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int neigh_table_clear(struct neigh_table *tbl)
|
|
|
|
{
|
|
|
|
struct neigh_table **tp;
|
|
|
|
|
|
|
|
/* It is not clean... Fix it to unload IPv6 module safely */
|
2010-10-19 08:04:42 +02:00
|
|
|
cancel_delayed_work_sync(&tbl->gc_work);
|
2005-04-17 00:20:36 +02:00
|
|
|
del_timer_sync(&tbl->proxy_timer);
|
|
|
|
pneigh_queue_purge(&tbl->proxy_queue);
|
|
|
|
neigh_ifdown(tbl, NULL);
|
|
|
|
if (atomic_read(&tbl->entries))
|
2012-05-16 21:58:40 +02:00
|
|
|
pr_crit("neighbour leakage\n");
|
2005-04-17 00:20:36 +02:00
|
|
|
write_lock(&neigh_tbl_lock);
|
|
|
|
for (tp = &neigh_tables; *tp; tp = &(*tp)->next) {
|
|
|
|
if (*tp == tbl) {
|
|
|
|
*tp = tbl->next;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
write_unlock(&neigh_tbl_lock);
|
|
|
|
|
2011-01-19 23:02:47 +01:00
|
|
|
call_rcu(&rcu_dereference_protected(tbl->nht, 1)->rcu,
|
|
|
|
neigh_hash_free_rcu);
|
2010-10-04 08:15:44 +02:00
|
|
|
tbl->nht = NULL;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
kfree(tbl->phash_buckets);
|
|
|
|
tbl->phash_buckets = NULL;
|
|
|
|
|
2007-11-06 06:28:13 +01:00
|
|
|
remove_proc_entry(tbl->id, init_net.proc_net_stat);
|
|
|
|
|
2006-09-01 10:34:10 +02:00
|
|
|
free_percpu(tbl->stats);
|
|
|
|
tbl->stats = NULL;
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
return 0;
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_table_clear);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2013-03-21 08:45:29 +01:00
|
|
|
static int neigh_delete(struct sk_buff *skb, struct nlmsghdr *nlh)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2008-03-25 18:26:21 +01:00
|
|
|
struct net *net = sock_net(skb->sk);
|
2006-08-08 02:53:08 +02:00
|
|
|
struct ndmsg *ndm;
|
|
|
|
struct nlattr *dst_attr;
|
2005-04-17 00:20:36 +02:00
|
|
|
struct neigh_table *tbl;
|
|
|
|
struct net_device *dev = NULL;
|
2006-08-08 02:53:08 +02:00
|
|
|
int err = -EINVAL;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-04 06:27:36 +02:00
|
|
|
ASSERT_RTNL();
|
2006-08-08 02:53:08 +02:00
|
|
|
if (nlmsg_len(nlh) < sizeof(*ndm))
|
2005-04-17 00:20:36 +02:00
|
|
|
goto out;
|
|
|
|
|
2006-08-08 02:53:08 +02:00
|
|
|
dst_attr = nlmsg_find_attr(nlh, sizeof(*ndm), NDA_DST);
|
|
|
|
if (dst_attr == NULL)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
ndm = nlmsg_data(nlh);
|
|
|
|
if (ndm->ndm_ifindex) {
|
2010-10-04 06:27:36 +02:00
|
|
|
dev = __dev_get_by_index(net, ndm->ndm_ifindex);
|
2006-08-08 02:53:08 +02:00
|
|
|
if (dev == NULL) {
|
|
|
|
err = -ENODEV;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
read_lock(&neigh_tbl_lock);
|
|
|
|
for (tbl = neigh_tables; tbl; tbl = tbl->next) {
|
2006-08-08 02:53:08 +02:00
|
|
|
struct neighbour *neigh;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (tbl->family != ndm->ndm_family)
|
|
|
|
continue;
|
|
|
|
read_unlock(&neigh_tbl_lock);
|
|
|
|
|
2006-08-08 02:53:08 +02:00
|
|
|
if (nla_len(dst_attr) < tbl->key_len)
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (ndm->ndm_flags & NTF_PROXY) {
|
2008-01-24 09:13:18 +01:00
|
|
|
err = pneigh_delete(tbl, net, nla_data(dst_attr), dev);
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2006-08-08 02:53:08 +02:00
|
|
|
if (dev == NULL)
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-08-08 02:53:08 +02:00
|
|
|
neigh = neigh_lookup(tbl, nla_data(dst_attr), dev);
|
|
|
|
if (neigh == NULL) {
|
|
|
|
err = -ENOENT;
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2006-08-08 02:53:08 +02:00
|
|
|
|
|
|
|
err = neigh_update(neigh, NULL, NUD_FAILED,
|
|
|
|
NEIGH_UPDATE_F_OVERRIDE |
|
|
|
|
NEIGH_UPDATE_F_ADMIN);
|
|
|
|
neigh_release(neigh);
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
read_unlock(&neigh_tbl_lock);
|
2006-08-08 02:53:08 +02:00
|
|
|
err = -EAFNOSUPPORT;
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
out:
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2013-03-21 08:45:29 +01:00
|
|
|
static int neigh_add(struct sk_buff *skb, struct nlmsghdr *nlh)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2008-03-25 18:26:21 +01:00
|
|
|
struct net *net = sock_net(skb->sk);
|
2006-08-08 02:55:40 +02:00
|
|
|
struct ndmsg *ndm;
|
|
|
|
struct nlattr *tb[NDA_MAX+1];
|
2005-04-17 00:20:36 +02:00
|
|
|
struct neigh_table *tbl;
|
|
|
|
struct net_device *dev = NULL;
|
2006-08-08 02:55:40 +02:00
|
|
|
int err;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-04 06:27:36 +02:00
|
|
|
ASSERT_RTNL();
|
2006-08-08 02:55:40 +02:00
|
|
|
err = nlmsg_parse(nlh, sizeof(*ndm), tb, NDA_MAX, NULL);
|
|
|
|
if (err < 0)
|
2005-04-17 00:20:36 +02:00
|
|
|
goto out;
|
|
|
|
|
2006-08-08 02:55:40 +02:00
|
|
|
err = -EINVAL;
|
|
|
|
if (tb[NDA_DST] == NULL)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
ndm = nlmsg_data(nlh);
|
|
|
|
if (ndm->ndm_ifindex) {
|
2010-10-04 06:27:36 +02:00
|
|
|
dev = __dev_get_by_index(net, ndm->ndm_ifindex);
|
2006-08-08 02:55:40 +02:00
|
|
|
if (dev == NULL) {
|
|
|
|
err = -ENODEV;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (tb[NDA_LLADDR] && nla_len(tb[NDA_LLADDR]) < dev->addr_len)
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2006-08-08 02:55:40 +02:00
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
read_lock(&neigh_tbl_lock);
|
|
|
|
for (tbl = neigh_tables; tbl; tbl = tbl->next) {
|
2006-08-08 02:55:40 +02:00
|
|
|
int flags = NEIGH_UPDATE_F_ADMIN | NEIGH_UPDATE_F_OVERRIDE;
|
|
|
|
struct neighbour *neigh;
|
|
|
|
void *dst, *lladdr;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (tbl->family != ndm->ndm_family)
|
|
|
|
continue;
|
|
|
|
read_unlock(&neigh_tbl_lock);
|
|
|
|
|
2006-08-08 02:55:40 +02:00
|
|
|
if (nla_len(tb[NDA_DST]) < tbl->key_len)
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2006-08-08 02:55:40 +02:00
|
|
|
dst = nla_data(tb[NDA_DST]);
|
|
|
|
lladdr = tb[NDA_LLADDR] ? nla_data(tb[NDA_LLADDR]) : NULL;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (ndm->ndm_flags & NTF_PROXY) {
|
2006-09-22 23:43:19 +02:00
|
|
|
struct pneigh_entry *pn;
|
|
|
|
|
|
|
|
err = -ENOBUFS;
|
2008-01-24 09:13:18 +01:00
|
|
|
pn = pneigh_lookup(tbl, net, dst, dev, 1);
|
2006-09-22 23:43:19 +02:00
|
|
|
if (pn) {
|
|
|
|
pn->flags = ndm->ndm_flags;
|
|
|
|
err = 0;
|
|
|
|
}
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2006-08-08 02:55:40 +02:00
|
|
|
if (dev == NULL)
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2006-08-08 02:55:40 +02:00
|
|
|
|
|
|
|
neigh = neigh_lookup(tbl, dst, dev);
|
|
|
|
if (neigh == NULL) {
|
|
|
|
if (!(nlh->nlmsg_flags & NLM_F_CREATE)) {
|
|
|
|
err = -ENOENT;
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2006-08-08 02:55:40 +02:00
|
|
|
}
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2006-08-08 02:55:40 +02:00
|
|
|
neigh = __neigh_lookup_errno(tbl, dst, dev);
|
|
|
|
if (IS_ERR(neigh)) {
|
|
|
|
err = PTR_ERR(neigh);
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
} else {
|
2006-08-08 02:55:40 +02:00
|
|
|
if (nlh->nlmsg_flags & NLM_F_EXCL) {
|
|
|
|
err = -EEXIST;
|
|
|
|
neigh_release(neigh);
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2006-08-08 02:55:40 +02:00
|
|
|
if (!(nlh->nlmsg_flags & NLM_F_REPLACE))
|
|
|
|
flags &= ~NEIGH_UPDATE_F_OVERRIDE;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2009-03-04 09:03:08 +01:00
|
|
|
if (ndm->ndm_flags & NTF_USE) {
|
|
|
|
neigh_event_send(neigh, NULL);
|
|
|
|
err = 0;
|
|
|
|
} else
|
|
|
|
err = neigh_update(neigh, lladdr, ndm->ndm_state, flags);
|
2006-08-08 02:55:40 +02:00
|
|
|
neigh_release(neigh);
|
2010-10-04 06:27:36 +02:00
|
|
|
goto out;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
read_unlock(&neigh_tbl_lock);
|
2006-08-08 02:55:40 +02:00
|
|
|
err = -EAFNOSUPPORT;
|
2005-04-17 00:20:36 +02:00
|
|
|
out:
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2005-06-19 07:50:55 +02:00
|
|
|
static int neightbl_fill_parms(struct sk_buff *skb, struct neigh_parms *parms)
|
|
|
|
{
|
2006-08-08 03:00:18 +02:00
|
|
|
struct nlattr *nest;
|
|
|
|
|
|
|
|
nest = nla_nest_start(skb, NDTA_PARMS);
|
|
|
|
if (nest == NULL)
|
|
|
|
return -ENOBUFS;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2012-04-02 02:06:28 +02:00
|
|
|
if ((parms->dev &&
|
|
|
|
nla_put_u32(skb, NDTPA_IFINDEX, parms->dev->ifindex)) ||
|
|
|
|
nla_put_u32(skb, NDTPA_REFCNT, atomic_read(&parms->refcnt)) ||
|
2013-12-07 19:26:53 +01:00
|
|
|
nla_put_u32(skb, NDTPA_QUEUE_LENBYTES,
|
|
|
|
NEIGH_VAR(parms, QUEUE_LEN_BYTES)) ||
|
2012-04-02 02:06:28 +02:00
|
|
|
/* approximative value for deprecated QUEUE_LEN (in packets) */
|
|
|
|
nla_put_u32(skb, NDTPA_QUEUE_LEN,
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR(parms, QUEUE_LEN_BYTES) / SKB_TRUESIZE(ETH_FRAME_LEN)) ||
|
|
|
|
nla_put_u32(skb, NDTPA_PROXY_QLEN, NEIGH_VAR(parms, PROXY_QLEN)) ||
|
|
|
|
nla_put_u32(skb, NDTPA_APP_PROBES, NEIGH_VAR(parms, APP_PROBES)) ||
|
|
|
|
nla_put_u32(skb, NDTPA_UCAST_PROBES,
|
|
|
|
NEIGH_VAR(parms, UCAST_PROBES)) ||
|
|
|
|
nla_put_u32(skb, NDTPA_MCAST_PROBES,
|
|
|
|
NEIGH_VAR(parms, MCAST_PROBES)) ||
|
2012-04-02 02:06:28 +02:00
|
|
|
nla_put_msecs(skb, NDTPA_REACHABLE_TIME, parms->reachable_time) ||
|
|
|
|
nla_put_msecs(skb, NDTPA_BASE_REACHABLE_TIME,
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR(parms, BASE_REACHABLE_TIME)) ||
|
|
|
|
nla_put_msecs(skb, NDTPA_GC_STALETIME,
|
|
|
|
NEIGH_VAR(parms, GC_STALETIME)) ||
|
2012-04-02 02:06:28 +02:00
|
|
|
nla_put_msecs(skb, NDTPA_DELAY_PROBE_TIME,
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR(parms, DELAY_PROBE_TIME)) ||
|
|
|
|
nla_put_msecs(skb, NDTPA_RETRANS_TIME,
|
|
|
|
NEIGH_VAR(parms, RETRANS_TIME)) ||
|
|
|
|
nla_put_msecs(skb, NDTPA_ANYCAST_DELAY,
|
|
|
|
NEIGH_VAR(parms, ANYCAST_DELAY)) ||
|
|
|
|
nla_put_msecs(skb, NDTPA_PROXY_DELAY,
|
|
|
|
NEIGH_VAR(parms, PROXY_DELAY)) ||
|
|
|
|
nla_put_msecs(skb, NDTPA_LOCKTIME,
|
|
|
|
NEIGH_VAR(parms, LOCKTIME)))
|
2012-04-02 02:06:28 +02:00
|
|
|
goto nla_put_failure;
|
2006-08-08 03:00:18 +02:00
|
|
|
return nla_nest_end(skb, nest);
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
nla_put_failure:
|
2008-06-04 01:36:54 +02:00
|
|
|
nla_nest_cancel(skb, nest);
|
|
|
|
return -EMSGSIZE;
|
2005-06-19 07:50:55 +02:00
|
|
|
}
|
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
static int neightbl_fill_info(struct sk_buff *skb, struct neigh_table *tbl,
|
|
|
|
u32 pid, u32 seq, int type, int flags)
|
2005-06-19 07:50:55 +02:00
|
|
|
{
|
|
|
|
struct nlmsghdr *nlh;
|
|
|
|
struct ndtmsg *ndtmsg;
|
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
nlh = nlmsg_put(skb, pid, seq, type, sizeof(*ndtmsg), flags);
|
|
|
|
if (nlh == NULL)
|
2007-02-01 08:16:40 +01:00
|
|
|
return -EMSGSIZE;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
ndtmsg = nlmsg_data(nlh);
|
2005-06-19 07:50:55 +02:00
|
|
|
|
|
|
|
read_lock_bh(&tbl->lock);
|
|
|
|
ndtmsg->ndtm_family = tbl->family;
|
2005-06-28 21:55:30 +02:00
|
|
|
ndtmsg->ndtm_pad1 = 0;
|
|
|
|
ndtmsg->ndtm_pad2 = 0;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2012-04-02 02:06:28 +02:00
|
|
|
if (nla_put_string(skb, NDTA_NAME, tbl->id) ||
|
|
|
|
nla_put_msecs(skb, NDTA_GC_INTERVAL, tbl->gc_interval) ||
|
|
|
|
nla_put_u32(skb, NDTA_THRESH1, tbl->gc_thresh1) ||
|
|
|
|
nla_put_u32(skb, NDTA_THRESH2, tbl->gc_thresh2) ||
|
|
|
|
nla_put_u32(skb, NDTA_THRESH3, tbl->gc_thresh3))
|
|
|
|
goto nla_put_failure;
|
2005-06-19 07:50:55 +02:00
|
|
|
{
|
|
|
|
unsigned long now = jiffies;
|
|
|
|
unsigned int flush_delta = now - tbl->last_flush;
|
|
|
|
unsigned int rand_delta = now - tbl->last_rand;
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht;
|
2005-06-19 07:50:55 +02:00
|
|
|
struct ndt_config ndc = {
|
|
|
|
.ndtc_key_len = tbl->key_len,
|
|
|
|
.ndtc_entry_size = tbl->entry_size,
|
|
|
|
.ndtc_entries = atomic_read(&tbl->entries),
|
|
|
|
.ndtc_last_flush = jiffies_to_msecs(flush_delta),
|
|
|
|
.ndtc_last_rand = jiffies_to_msecs(rand_delta),
|
|
|
|
.ndtc_proxy_qlen = tbl->proxy_queue.qlen,
|
|
|
|
};
|
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
rcu_read_lock_bh();
|
|
|
|
nht = rcu_dereference_bh(tbl->nht);
|
2011-12-28 21:06:58 +01:00
|
|
|
ndc.ndtc_hash_rnd = nht->hash_rnd[0];
|
2011-07-11 10:28:12 +02:00
|
|
|
ndc.ndtc_hash_mask = ((1 << nht->hash_shift) - 1);
|
2010-10-04 08:15:44 +02:00
|
|
|
rcu_read_unlock_bh();
|
|
|
|
|
2012-04-02 02:06:28 +02:00
|
|
|
if (nla_put(skb, NDTA_CONFIG, sizeof(ndc), &ndc))
|
|
|
|
goto nla_put_failure;
|
2005-06-19 07:50:55 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
{
|
|
|
|
int cpu;
|
|
|
|
struct ndt_stats ndst;
|
|
|
|
|
|
|
|
memset(&ndst, 0, sizeof(ndst));
|
|
|
|
|
2006-04-11 07:52:50 +02:00
|
|
|
for_each_possible_cpu(cpu) {
|
2005-06-19 07:50:55 +02:00
|
|
|
struct neigh_statistics *st;
|
|
|
|
|
|
|
|
st = per_cpu_ptr(tbl->stats, cpu);
|
|
|
|
ndst.ndts_allocs += st->allocs;
|
|
|
|
ndst.ndts_destroys += st->destroys;
|
|
|
|
ndst.ndts_hash_grows += st->hash_grows;
|
|
|
|
ndst.ndts_res_failed += st->res_failed;
|
|
|
|
ndst.ndts_lookups += st->lookups;
|
|
|
|
ndst.ndts_hits += st->hits;
|
|
|
|
ndst.ndts_rcv_probes_mcast += st->rcv_probes_mcast;
|
|
|
|
ndst.ndts_rcv_probes_ucast += st->rcv_probes_ucast;
|
|
|
|
ndst.ndts_periodic_gc_runs += st->periodic_gc_runs;
|
|
|
|
ndst.ndts_forced_gc_runs += st->forced_gc_runs;
|
|
|
|
}
|
|
|
|
|
2012-04-02 02:06:28 +02:00
|
|
|
if (nla_put(skb, NDTA_STATS, sizeof(ndst), &ndst))
|
|
|
|
goto nla_put_failure;
|
2005-06-19 07:50:55 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
BUG_ON(tbl->parms.dev);
|
|
|
|
if (neightbl_fill_parms(skb, &tbl->parms) < 0)
|
2006-08-08 03:00:18 +02:00
|
|
|
goto nla_put_failure;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
|
|
|
read_unlock_bh(&tbl->lock);
|
2006-08-08 03:00:18 +02:00
|
|
|
return nlmsg_end(skb, nlh);
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
nla_put_failure:
|
2005-06-19 07:50:55 +02:00
|
|
|
read_unlock_bh(&tbl->lock);
|
2007-02-01 08:16:40 +01:00
|
|
|
nlmsg_cancel(skb, nlh);
|
|
|
|
return -EMSGSIZE;
|
2005-06-19 07:50:55 +02:00
|
|
|
}
|
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
static int neightbl_fill_param_info(struct sk_buff *skb,
|
|
|
|
struct neigh_table *tbl,
|
2005-06-19 07:50:55 +02:00
|
|
|
struct neigh_parms *parms,
|
2006-08-08 03:00:18 +02:00
|
|
|
u32 pid, u32 seq, int type,
|
|
|
|
unsigned int flags)
|
2005-06-19 07:50:55 +02:00
|
|
|
{
|
|
|
|
struct ndtmsg *ndtmsg;
|
|
|
|
struct nlmsghdr *nlh;
|
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
nlh = nlmsg_put(skb, pid, seq, type, sizeof(*ndtmsg), flags);
|
|
|
|
if (nlh == NULL)
|
2007-02-01 08:16:40 +01:00
|
|
|
return -EMSGSIZE;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
ndtmsg = nlmsg_data(nlh);
|
2005-06-19 07:50:55 +02:00
|
|
|
|
|
|
|
read_lock_bh(&tbl->lock);
|
|
|
|
ndtmsg->ndtm_family = tbl->family;
|
2005-06-28 21:55:30 +02:00
|
|
|
ndtmsg->ndtm_pad1 = 0;
|
|
|
|
ndtmsg->ndtm_pad2 = 0;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
if (nla_put_string(skb, NDTA_NAME, tbl->id) < 0 ||
|
|
|
|
neightbl_fill_parms(skb, parms) < 0)
|
|
|
|
goto errout;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
|
|
|
read_unlock_bh(&tbl->lock);
|
2006-08-08 03:00:18 +02:00
|
|
|
return nlmsg_end(skb, nlh);
|
|
|
|
errout:
|
2005-06-19 07:50:55 +02:00
|
|
|
read_unlock_bh(&tbl->lock);
|
2007-02-01 08:16:40 +01:00
|
|
|
nlmsg_cancel(skb, nlh);
|
|
|
|
return -EMSGSIZE;
|
2005-06-19 07:50:55 +02:00
|
|
|
}
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2007-06-05 21:38:30 +02:00
|
|
|
static const struct nla_policy nl_neightbl_policy[NDTA_MAX+1] = {
|
2006-08-08 02:58:53 +02:00
|
|
|
[NDTA_NAME] = { .type = NLA_STRING },
|
|
|
|
[NDTA_THRESH1] = { .type = NLA_U32 },
|
|
|
|
[NDTA_THRESH2] = { .type = NLA_U32 },
|
|
|
|
[NDTA_THRESH3] = { .type = NLA_U32 },
|
|
|
|
[NDTA_GC_INTERVAL] = { .type = NLA_U64 },
|
|
|
|
[NDTA_PARMS] = { .type = NLA_NESTED },
|
|
|
|
};
|
|
|
|
|
2007-06-05 21:38:30 +02:00
|
|
|
static const struct nla_policy nl_ntbl_parm_policy[NDTPA_MAX+1] = {
|
2006-08-08 02:58:53 +02:00
|
|
|
[NDTPA_IFINDEX] = { .type = NLA_U32 },
|
|
|
|
[NDTPA_QUEUE_LEN] = { .type = NLA_U32 },
|
|
|
|
[NDTPA_PROXY_QLEN] = { .type = NLA_U32 },
|
|
|
|
[NDTPA_APP_PROBES] = { .type = NLA_U32 },
|
|
|
|
[NDTPA_UCAST_PROBES] = { .type = NLA_U32 },
|
|
|
|
[NDTPA_MCAST_PROBES] = { .type = NLA_U32 },
|
|
|
|
[NDTPA_BASE_REACHABLE_TIME] = { .type = NLA_U64 },
|
|
|
|
[NDTPA_GC_STALETIME] = { .type = NLA_U64 },
|
|
|
|
[NDTPA_DELAY_PROBE_TIME] = { .type = NLA_U64 },
|
|
|
|
[NDTPA_RETRANS_TIME] = { .type = NLA_U64 },
|
|
|
|
[NDTPA_ANYCAST_DELAY] = { .type = NLA_U64 },
|
|
|
|
[NDTPA_PROXY_DELAY] = { .type = NLA_U64 },
|
|
|
|
[NDTPA_LOCKTIME] = { .type = NLA_U64 },
|
|
|
|
};
|
|
|
|
|
2013-03-21 08:45:29 +01:00
|
|
|
static int neightbl_set(struct sk_buff *skb, struct nlmsghdr *nlh)
|
2005-06-19 07:50:55 +02:00
|
|
|
{
|
2008-03-25 18:26:21 +01:00
|
|
|
struct net *net = sock_net(skb->sk);
|
2005-06-19 07:50:55 +02:00
|
|
|
struct neigh_table *tbl;
|
2006-08-08 02:58:53 +02:00
|
|
|
struct ndtmsg *ndtmsg;
|
|
|
|
struct nlattr *tb[NDTA_MAX+1];
|
|
|
|
int err;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
err = nlmsg_parse(nlh, sizeof(*ndtmsg), tb, NDTA_MAX,
|
|
|
|
nl_neightbl_policy);
|
|
|
|
if (err < 0)
|
|
|
|
goto errout;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
if (tb[NDTA_NAME] == NULL) {
|
|
|
|
err = -EINVAL;
|
|
|
|
goto errout;
|
|
|
|
}
|
|
|
|
|
|
|
|
ndtmsg = nlmsg_data(nlh);
|
2005-06-19 07:50:55 +02:00
|
|
|
read_lock(&neigh_tbl_lock);
|
|
|
|
for (tbl = neigh_tables; tbl; tbl = tbl->next) {
|
|
|
|
if (ndtmsg->ndtm_family && tbl->family != ndtmsg->ndtm_family)
|
|
|
|
continue;
|
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
if (nla_strcmp(tb[NDTA_NAME], tbl->id) == 0)
|
2005-06-19 07:50:55 +02:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (tbl == NULL) {
|
|
|
|
err = -ENOENT;
|
2006-08-08 02:58:53 +02:00
|
|
|
goto errout_locked;
|
2005-06-19 07:50:55 +02:00
|
|
|
}
|
|
|
|
|
2007-02-09 15:24:36 +01:00
|
|
|
/*
|
2005-06-19 07:50:55 +02:00
|
|
|
* We acquire tbl->lock to be nice to the periodic timers and
|
|
|
|
* make sure they always see a consistent set of values.
|
|
|
|
*/
|
|
|
|
write_lock_bh(&tbl->lock);
|
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
if (tb[NDTA_PARMS]) {
|
|
|
|
struct nlattr *tbp[NDTPA_MAX+1];
|
2005-06-19 07:50:55 +02:00
|
|
|
struct neigh_parms *p;
|
2006-08-08 02:58:53 +02:00
|
|
|
int i, ifindex = 0;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
err = nla_parse_nested(tbp, NDTPA_MAX, tb[NDTA_PARMS],
|
|
|
|
nl_ntbl_parm_policy);
|
|
|
|
if (err < 0)
|
|
|
|
goto errout_tbl_lock;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
if (tbp[NDTPA_IFINDEX])
|
|
|
|
ifindex = nla_get_u32(tbp[NDTPA_IFINDEX]);
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2009-07-13 20:17:49 +02:00
|
|
|
p = lookup_neigh_parms(tbl, net, ifindex);
|
2005-06-19 07:50:55 +02:00
|
|
|
if (p == NULL) {
|
|
|
|
err = -ENOENT;
|
2006-08-08 02:58:53 +02:00
|
|
|
goto errout_tbl_lock;
|
2005-06-19 07:50:55 +02:00
|
|
|
}
|
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
for (i = 1; i <= NDTPA_MAX; i++) {
|
|
|
|
if (tbp[i] == NULL)
|
|
|
|
continue;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
switch (i) {
|
|
|
|
case NDTPA_QUEUE_LEN:
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR_SET(p, QUEUE_LEN_BYTES,
|
|
|
|
nla_get_u32(tbp[i]) *
|
|
|
|
SKB_TRUESIZE(ETH_FRAME_LEN));
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
break;
|
|
|
|
case NDTPA_QUEUE_LENBYTES:
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR_SET(p, QUEUE_LEN_BYTES,
|
|
|
|
nla_get_u32(tbp[i]));
|
2006-08-08 02:58:53 +02:00
|
|
|
break;
|
|
|
|
case NDTPA_PROXY_QLEN:
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR_SET(p, PROXY_QLEN,
|
|
|
|
nla_get_u32(tbp[i]));
|
2006-08-08 02:58:53 +02:00
|
|
|
break;
|
|
|
|
case NDTPA_APP_PROBES:
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR_SET(p, APP_PROBES,
|
|
|
|
nla_get_u32(tbp[i]));
|
2006-08-08 02:58:53 +02:00
|
|
|
break;
|
|
|
|
case NDTPA_UCAST_PROBES:
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR_SET(p, UCAST_PROBES,
|
|
|
|
nla_get_u32(tbp[i]));
|
2006-08-08 02:58:53 +02:00
|
|
|
break;
|
|
|
|
case NDTPA_MCAST_PROBES:
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR_SET(p, MCAST_PROBES,
|
|
|
|
nla_get_u32(tbp[i]));
|
2006-08-08 02:58:53 +02:00
|
|
|
break;
|
|
|
|
case NDTPA_BASE_REACHABLE_TIME:
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR_SET(p, BASE_REACHABLE_TIME,
|
|
|
|
nla_get_msecs(tbp[i]));
|
2006-08-08 02:58:53 +02:00
|
|
|
break;
|
|
|
|
case NDTPA_GC_STALETIME:
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR_SET(p, GC_STALETIME,
|
|
|
|
nla_get_msecs(tbp[i]));
|
2006-08-08 02:58:53 +02:00
|
|
|
break;
|
|
|
|
case NDTPA_DELAY_PROBE_TIME:
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR_SET(p, DELAY_PROBE_TIME,
|
|
|
|
nla_get_msecs(tbp[i]));
|
2006-08-08 02:58:53 +02:00
|
|
|
break;
|
|
|
|
case NDTPA_RETRANS_TIME:
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_VAR_SET(p, RETRANS_TIME,
|
|
|
|
nla_get_msecs(tbp[i]));
|
2006-08-08 02:58:53 +02:00
|
|
|
break;
|
|
|
|
case NDTPA_ANYCAST_DELAY:
|
2014-01-14 15:46:07 +01:00
|
|
|
NEIGH_VAR_SET(p, ANYCAST_DELAY,
|
|
|
|
nla_get_msecs(tbp[i]));
|
2006-08-08 02:58:53 +02:00
|
|
|
break;
|
|
|
|
case NDTPA_PROXY_DELAY:
|
2014-01-14 15:46:07 +01:00
|
|
|
NEIGH_VAR_SET(p, PROXY_DELAY,
|
|
|
|
nla_get_msecs(tbp[i]));
|
2006-08-08 02:58:53 +02:00
|
|
|
break;
|
|
|
|
case NDTPA_LOCKTIME:
|
2014-01-14 15:46:07 +01:00
|
|
|
NEIGH_VAR_SET(p, LOCKTIME,
|
|
|
|
nla_get_msecs(tbp[i]));
|
2006-08-08 02:58:53 +02:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2013-06-20 04:01:34 +02:00
|
|
|
err = -ENOENT;
|
|
|
|
if ((tb[NDTA_THRESH1] || tb[NDTA_THRESH2] ||
|
|
|
|
tb[NDTA_THRESH3] || tb[NDTA_GC_INTERVAL]) &&
|
|
|
|
!net_eq(net, &init_net))
|
|
|
|
goto errout_tbl_lock;
|
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
if (tb[NDTA_THRESH1])
|
|
|
|
tbl->gc_thresh1 = nla_get_u32(tb[NDTA_THRESH1]);
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
if (tb[NDTA_THRESH2])
|
|
|
|
tbl->gc_thresh2 = nla_get_u32(tb[NDTA_THRESH2]);
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
if (tb[NDTA_THRESH3])
|
|
|
|
tbl->gc_thresh3 = nla_get_u32(tb[NDTA_THRESH3]);
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
if (tb[NDTA_GC_INTERVAL])
|
|
|
|
tbl->gc_interval = nla_get_msecs(tb[NDTA_GC_INTERVAL]);
|
2005-06-19 07:50:55 +02:00
|
|
|
|
|
|
|
err = 0;
|
|
|
|
|
2006-08-08 02:58:53 +02:00
|
|
|
errout_tbl_lock:
|
2005-06-19 07:50:55 +02:00
|
|
|
write_unlock_bh(&tbl->lock);
|
2006-08-08 02:58:53 +02:00
|
|
|
errout_locked:
|
2005-06-19 07:50:55 +02:00
|
|
|
read_unlock(&neigh_tbl_lock);
|
2006-08-08 02:58:53 +02:00
|
|
|
errout:
|
2005-06-19 07:50:55 +02:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2007-03-22 19:50:06 +01:00
|
|
|
static int neightbl_dump_info(struct sk_buff *skb, struct netlink_callback *cb)
|
2005-06-19 07:50:55 +02:00
|
|
|
{
|
2008-03-25 18:26:21 +01:00
|
|
|
struct net *net = sock_net(skb->sk);
|
2006-08-08 03:00:18 +02:00
|
|
|
int family, tidx, nidx = 0;
|
|
|
|
int tbl_skip = cb->args[0];
|
|
|
|
int neigh_skip = cb->args[1];
|
2005-06-19 07:50:55 +02:00
|
|
|
struct neigh_table *tbl;
|
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
family = ((struct rtgenmsg *) nlmsg_data(cb->nlh))->rtgen_family;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
|
|
|
read_lock(&neigh_tbl_lock);
|
2006-08-08 03:00:18 +02:00
|
|
|
for (tbl = neigh_tables, tidx = 0; tbl; tbl = tbl->next, tidx++) {
|
2005-06-19 07:50:55 +02:00
|
|
|
struct neigh_parms *p;
|
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
if (tidx < tbl_skip || (family && tbl->family != family))
|
2005-06-19 07:50:55 +02:00
|
|
|
continue;
|
|
|
|
|
2012-09-07 22:12:54 +02:00
|
|
|
if (neightbl_fill_info(skb, tbl, NETLINK_CB(cb->skb).portid,
|
2006-08-08 03:00:18 +02:00
|
|
|
cb->nlh->nlmsg_seq, RTM_NEWNEIGHTBL,
|
|
|
|
NLM_F_MULTI) <= 0)
|
2005-06-19 07:50:55 +02:00
|
|
|
break;
|
|
|
|
|
2008-01-24 09:13:18 +01:00
|
|
|
for (nidx = 0, p = tbl->parms.next; p; p = p->next) {
|
2008-03-25 19:57:35 +01:00
|
|
|
if (!net_eq(neigh_parms_net(p), net))
|
2008-01-24 09:13:18 +01:00
|
|
|
continue;
|
|
|
|
|
2009-02-06 09:52:04 +01:00
|
|
|
if (nidx < neigh_skip)
|
|
|
|
goto next;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
if (neightbl_fill_param_info(skb, tbl, p,
|
2012-09-07 22:12:54 +02:00
|
|
|
NETLINK_CB(cb->skb).portid,
|
2006-08-08 03:00:18 +02:00
|
|
|
cb->nlh->nlmsg_seq,
|
|
|
|
RTM_NEWNEIGHTBL,
|
|
|
|
NLM_F_MULTI) <= 0)
|
2005-06-19 07:50:55 +02:00
|
|
|
goto out;
|
2009-02-06 09:52:04 +01:00
|
|
|
next:
|
|
|
|
nidx++;
|
2005-06-19 07:50:55 +02:00
|
|
|
}
|
|
|
|
|
2006-08-08 03:00:18 +02:00
|
|
|
neigh_skip = 0;
|
2005-06-19 07:50:55 +02:00
|
|
|
}
|
|
|
|
out:
|
|
|
|
read_unlock(&neigh_tbl_lock);
|
2006-08-08 03:00:18 +02:00
|
|
|
cb->args[0] = tidx;
|
|
|
|
cb->args[1] = nidx;
|
2005-06-19 07:50:55 +02:00
|
|
|
|
|
|
|
return skb->len;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-08-08 02:56:37 +02:00
|
|
|
static int neigh_fill_info(struct sk_buff *skb, struct neighbour *neigh,
|
|
|
|
u32 pid, u32 seq, int type, unsigned int flags)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
unsigned long now = jiffies;
|
|
|
|
struct nda_cacheinfo ci;
|
2006-08-08 02:56:37 +02:00
|
|
|
struct nlmsghdr *nlh;
|
|
|
|
struct ndmsg *ndm;
|
|
|
|
|
|
|
|
nlh = nlmsg_put(skb, pid, seq, type, sizeof(*ndm), flags);
|
|
|
|
if (nlh == NULL)
|
2007-02-01 08:16:40 +01:00
|
|
|
return -EMSGSIZE;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-08-08 02:56:37 +02:00
|
|
|
ndm = nlmsg_data(nlh);
|
|
|
|
ndm->ndm_family = neigh->ops->family;
|
2005-06-28 21:55:30 +02:00
|
|
|
ndm->ndm_pad1 = 0;
|
|
|
|
ndm->ndm_pad2 = 0;
|
2006-08-08 02:56:37 +02:00
|
|
|
ndm->ndm_flags = neigh->flags;
|
|
|
|
ndm->ndm_type = neigh->type;
|
|
|
|
ndm->ndm_ifindex = neigh->dev->ifindex;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2012-04-02 02:06:28 +02:00
|
|
|
if (nla_put(skb, NDA_DST, neigh->tbl->key_len, neigh->primary_key))
|
|
|
|
goto nla_put_failure;
|
2006-08-08 02:56:37 +02:00
|
|
|
|
|
|
|
read_lock_bh(&neigh->lock);
|
|
|
|
ndm->ndm_state = neigh->nud_state;
|
2010-10-07 12:44:07 +02:00
|
|
|
if (neigh->nud_state & NUD_VALID) {
|
|
|
|
char haddr[MAX_ADDR_LEN];
|
|
|
|
|
|
|
|
neigh_ha_snapshot(haddr, neigh, neigh->dev);
|
|
|
|
if (nla_put(skb, NDA_LLADDR, neigh->dev->addr_len, haddr) < 0) {
|
|
|
|
read_unlock_bh(&neigh->lock);
|
|
|
|
goto nla_put_failure;
|
|
|
|
}
|
2006-08-08 02:56:37 +02:00
|
|
|
}
|
|
|
|
|
2008-06-04 01:03:15 +02:00
|
|
|
ci.ndm_used = jiffies_to_clock_t(now - neigh->used);
|
|
|
|
ci.ndm_confirmed = jiffies_to_clock_t(now - neigh->confirmed);
|
|
|
|
ci.ndm_updated = jiffies_to_clock_t(now - neigh->updated);
|
2006-08-08 02:56:37 +02:00
|
|
|
ci.ndm_refcnt = atomic_read(&neigh->refcnt) - 1;
|
|
|
|
read_unlock_bh(&neigh->lock);
|
|
|
|
|
2012-04-02 02:06:28 +02:00
|
|
|
if (nla_put_u32(skb, NDA_PROBES, atomic_read(&neigh->probes)) ||
|
|
|
|
nla_put(skb, NDA_CACHEINFO, sizeof(ci), &ci))
|
|
|
|
goto nla_put_failure;
|
2006-08-08 02:56:37 +02:00
|
|
|
|
|
|
|
return nlmsg_end(skb, nlh);
|
|
|
|
|
|
|
|
nla_put_failure:
|
2007-02-01 08:16:40 +01:00
|
|
|
nlmsg_cancel(skb, nlh);
|
|
|
|
return -EMSGSIZE;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2012-01-26 23:28:58 +01:00
|
|
|
static int pneigh_fill_info(struct sk_buff *skb, struct pneigh_entry *pn,
|
|
|
|
u32 pid, u32 seq, int type, unsigned int flags,
|
|
|
|
struct neigh_table *tbl)
|
|
|
|
{
|
|
|
|
struct nlmsghdr *nlh;
|
|
|
|
struct ndmsg *ndm;
|
|
|
|
|
|
|
|
nlh = nlmsg_put(skb, pid, seq, type, sizeof(*ndm), flags);
|
|
|
|
if (nlh == NULL)
|
|
|
|
return -EMSGSIZE;
|
|
|
|
|
|
|
|
ndm = nlmsg_data(nlh);
|
|
|
|
ndm->ndm_family = tbl->family;
|
|
|
|
ndm->ndm_pad1 = 0;
|
|
|
|
ndm->ndm_pad2 = 0;
|
|
|
|
ndm->ndm_flags = pn->flags | NTF_PROXY;
|
|
|
|
ndm->ndm_type = NDA_DST;
|
|
|
|
ndm->ndm_ifindex = pn->dev->ifindex;
|
|
|
|
ndm->ndm_state = NUD_NONE;
|
|
|
|
|
2012-04-02 02:06:28 +02:00
|
|
|
if (nla_put(skb, NDA_DST, tbl->key_len, pn->key))
|
|
|
|
goto nla_put_failure;
|
2012-01-26 23:28:58 +01:00
|
|
|
|
|
|
|
return nlmsg_end(skb, nlh);
|
|
|
|
|
|
|
|
nla_put_failure:
|
|
|
|
nlmsg_cancel(skb, nlh);
|
|
|
|
return -EMSGSIZE;
|
|
|
|
}
|
|
|
|
|
2007-08-09 08:12:56 +02:00
|
|
|
static void neigh_update_notify(struct neighbour *neigh)
|
|
|
|
{
|
|
|
|
call_netevent_notifiers(NETEVENT_NEIGH_UPDATE, neigh);
|
|
|
|
__neigh_notify(neigh, RTM_NEWNEIGH, 0);
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
static int neigh_dump_table(struct neigh_table *tbl, struct sk_buff *skb,
|
|
|
|
struct netlink_callback *cb)
|
|
|
|
{
|
2010-10-07 02:49:21 +02:00
|
|
|
struct net *net = sock_net(skb->sk);
|
2005-04-17 00:20:36 +02:00
|
|
|
struct neighbour *n;
|
|
|
|
int rc, h, s_h = cb->args[1];
|
|
|
|
int idx, s_idx = idx = cb->args[2];
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
rcu_read_lock_bh();
|
|
|
|
nht = rcu_dereference_bh(tbl->nht);
|
|
|
|
|
2012-06-07 06:58:35 +02:00
|
|
|
for (h = s_h; h < (1 << nht->hash_shift); h++) {
|
2005-04-17 00:20:36 +02:00
|
|
|
if (h > s_h)
|
|
|
|
s_idx = 0;
|
2010-10-07 02:49:21 +02:00
|
|
|
for (n = rcu_dereference_bh(nht->hash_buckets[h]), idx = 0;
|
|
|
|
n != NULL;
|
|
|
|
n = rcu_dereference_bh(n->next)) {
|
2009-11-26 00:14:13 +01:00
|
|
|
if (!net_eq(dev_net(n->dev), net))
|
2008-01-24 09:13:18 +01:00
|
|
|
continue;
|
2009-02-06 09:52:04 +01:00
|
|
|
if (idx < s_idx)
|
|
|
|
goto next;
|
2012-09-07 22:12:54 +02:00
|
|
|
if (neigh_fill_info(skb, n, NETLINK_CB(cb->skb).portid,
|
2005-04-17 00:20:36 +02:00
|
|
|
cb->nlh->nlmsg_seq,
|
2005-06-19 07:54:12 +02:00
|
|
|
RTM_NEWNEIGH,
|
|
|
|
NLM_F_MULTI) <= 0) {
|
2005-04-17 00:20:36 +02:00
|
|
|
rc = -1;
|
|
|
|
goto out;
|
|
|
|
}
|
2010-10-07 02:49:21 +02:00
|
|
|
next:
|
2009-02-06 09:52:04 +01:00
|
|
|
idx++;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
rc = skb->len;
|
|
|
|
out:
|
2010-10-04 08:15:44 +02:00
|
|
|
rcu_read_unlock_bh();
|
2005-04-17 00:20:36 +02:00
|
|
|
cb->args[1] = h;
|
|
|
|
cb->args[2] = idx;
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
|
2012-01-26 23:28:58 +01:00
|
|
|
static int pneigh_dump_table(struct neigh_table *tbl, struct sk_buff *skb,
|
|
|
|
struct netlink_callback *cb)
|
|
|
|
{
|
|
|
|
struct pneigh_entry *n;
|
|
|
|
struct net *net = sock_net(skb->sk);
|
|
|
|
int rc, h, s_h = cb->args[3];
|
|
|
|
int idx, s_idx = idx = cb->args[4];
|
|
|
|
|
|
|
|
read_lock_bh(&tbl->lock);
|
|
|
|
|
2012-06-07 06:58:35 +02:00
|
|
|
for (h = s_h; h <= PNEIGH_HASHMASK; h++) {
|
2012-01-26 23:28:58 +01:00
|
|
|
if (h > s_h)
|
|
|
|
s_idx = 0;
|
|
|
|
for (n = tbl->phash_buckets[h], idx = 0; n; n = n->next) {
|
|
|
|
if (dev_net(n->dev) != net)
|
|
|
|
continue;
|
|
|
|
if (idx < s_idx)
|
|
|
|
goto next;
|
2012-09-07 22:12:54 +02:00
|
|
|
if (pneigh_fill_info(skb, n, NETLINK_CB(cb->skb).portid,
|
2012-01-26 23:28:58 +01:00
|
|
|
cb->nlh->nlmsg_seq,
|
|
|
|
RTM_NEWNEIGH,
|
|
|
|
NLM_F_MULTI, tbl) <= 0) {
|
|
|
|
read_unlock_bh(&tbl->lock);
|
|
|
|
rc = -1;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
next:
|
|
|
|
idx++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
read_unlock_bh(&tbl->lock);
|
|
|
|
rc = skb->len;
|
|
|
|
out:
|
|
|
|
cb->args[3] = h;
|
|
|
|
cb->args[4] = idx;
|
|
|
|
return rc;
|
|
|
|
|
|
|
|
}
|
|
|
|
|
2007-03-22 19:50:06 +01:00
|
|
|
static int neigh_dump_info(struct sk_buff *skb, struct netlink_callback *cb)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct neigh_table *tbl;
|
|
|
|
int t, family, s_t;
|
2012-01-26 23:28:58 +01:00
|
|
|
int proxy = 0;
|
2012-06-07 06:58:35 +02:00
|
|
|
int err;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
read_lock(&neigh_tbl_lock);
|
2006-08-08 02:56:37 +02:00
|
|
|
family = ((struct rtgenmsg *) nlmsg_data(cb->nlh))->rtgen_family;
|
2012-01-26 23:28:58 +01:00
|
|
|
|
|
|
|
/* check for full ndmsg structure presence, family member is
|
|
|
|
* the same for both structures
|
|
|
|
*/
|
|
|
|
if (nlmsg_len(cb->nlh) >= sizeof(struct ndmsg) &&
|
|
|
|
((struct ndmsg *) nlmsg_data(cb->nlh))->ndm_flags == NTF_PROXY)
|
|
|
|
proxy = 1;
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
s_t = cb->args[0];
|
|
|
|
|
2012-06-07 06:58:35 +02:00
|
|
|
for (tbl = neigh_tables, t = 0; tbl;
|
2012-01-26 23:28:58 +01:00
|
|
|
tbl = tbl->next, t++) {
|
2005-04-17 00:20:36 +02:00
|
|
|
if (t < s_t || (family && tbl->family != family))
|
|
|
|
continue;
|
|
|
|
if (t > s_t)
|
|
|
|
memset(&cb->args[1], 0, sizeof(cb->args) -
|
|
|
|
sizeof(cb->args[0]));
|
2012-01-26 23:28:58 +01:00
|
|
|
if (proxy)
|
|
|
|
err = pneigh_dump_table(tbl, skb, cb);
|
|
|
|
else
|
|
|
|
err = neigh_dump_table(tbl, skb, cb);
|
2012-06-07 06:58:35 +02:00
|
|
|
if (err < 0)
|
|
|
|
break;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
read_unlock(&neigh_tbl_lock);
|
|
|
|
|
|
|
|
cb->args[0] = t;
|
|
|
|
return skb->len;
|
|
|
|
}
|
|
|
|
|
|
|
|
void neigh_for_each(struct neigh_table *tbl, void (*cb)(struct neighbour *, void *), void *cookie)
|
|
|
|
{
|
|
|
|
int chain;
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
rcu_read_lock_bh();
|
|
|
|
nht = rcu_dereference_bh(tbl->nht);
|
|
|
|
|
2010-10-07 02:49:21 +02:00
|
|
|
read_lock(&tbl->lock); /* avoid resizes */
|
2011-07-11 10:28:12 +02:00
|
|
|
for (chain = 0; chain < (1 << nht->hash_shift); chain++) {
|
2005-04-17 00:20:36 +02:00
|
|
|
struct neighbour *n;
|
|
|
|
|
2010-10-07 02:49:21 +02:00
|
|
|
for (n = rcu_dereference_bh(nht->hash_buckets[chain]);
|
|
|
|
n != NULL;
|
|
|
|
n = rcu_dereference_bh(n->next))
|
2005-04-17 00:20:36 +02:00
|
|
|
cb(n, cookie);
|
|
|
|
}
|
2010-10-04 08:15:44 +02:00
|
|
|
read_unlock(&tbl->lock);
|
|
|
|
rcu_read_unlock_bh();
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(neigh_for_each);
|
|
|
|
|
|
|
|
/* The tbl->lock must be held as a writer and BH disabled. */
|
|
|
|
void __neigh_for_each_release(struct neigh_table *tbl,
|
|
|
|
int (*cb)(struct neighbour *))
|
|
|
|
{
|
|
|
|
int chain;
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
nht = rcu_dereference_protected(tbl->nht,
|
|
|
|
lockdep_is_held(&tbl->lock));
|
2011-07-11 10:28:12 +02:00
|
|
|
for (chain = 0; chain < (1 << nht->hash_shift); chain++) {
|
2010-10-07 02:49:21 +02:00
|
|
|
struct neighbour *n;
|
|
|
|
struct neighbour __rcu **np;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
np = &nht->hash_buckets[chain];
|
2010-10-07 02:49:21 +02:00
|
|
|
while ((n = rcu_dereference_protected(*np,
|
|
|
|
lockdep_is_held(&tbl->lock))) != NULL) {
|
2005-04-17 00:20:36 +02:00
|
|
|
int release;
|
|
|
|
|
|
|
|
write_lock(&n->lock);
|
|
|
|
release = cb(n);
|
|
|
|
if (release) {
|
2010-10-07 02:49:21 +02:00
|
|
|
rcu_assign_pointer(*np,
|
|
|
|
rcu_dereference_protected(n->next,
|
|
|
|
lockdep_is_held(&tbl->lock)));
|
2005-04-17 00:20:36 +02:00
|
|
|
n->dead = 1;
|
|
|
|
} else
|
|
|
|
np = &n->next;
|
|
|
|
write_unlock(&n->lock);
|
2007-08-09 08:12:36 +02:00
|
|
|
if (release)
|
|
|
|
neigh_cleanup_and_release(n);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(__neigh_for_each_release);
|
|
|
|
|
|
|
|
#ifdef CONFIG_PROC_FS
|
|
|
|
|
|
|
|
static struct neighbour *neigh_get_first(struct seq_file *seq)
|
|
|
|
{
|
|
|
|
struct neigh_seq_state *state = seq->private;
|
2008-03-25 18:36:06 +01:00
|
|
|
struct net *net = seq_file_net(seq);
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht = state->nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
struct neighbour *n = NULL;
|
|
|
|
int bucket = state->bucket;
|
|
|
|
|
|
|
|
state->flags &= ~NEIGH_SEQ_IS_PNEIGH;
|
2011-07-11 10:28:12 +02:00
|
|
|
for (bucket = 0; bucket < (1 << nht->hash_shift); bucket++) {
|
2010-10-07 02:49:21 +02:00
|
|
|
n = rcu_dereference_bh(nht->hash_buckets[bucket]);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
while (n) {
|
2008-03-25 19:57:35 +01:00
|
|
|
if (!net_eq(dev_net(n->dev), net))
|
2008-01-24 09:13:18 +01:00
|
|
|
goto next;
|
2005-04-17 00:20:36 +02:00
|
|
|
if (state->neigh_sub_iter) {
|
|
|
|
loff_t fakep = 0;
|
|
|
|
void *v;
|
|
|
|
|
|
|
|
v = state->neigh_sub_iter(state, n, &fakep);
|
|
|
|
if (!v)
|
|
|
|
goto next;
|
|
|
|
}
|
|
|
|
if (!(state->flags & NEIGH_SEQ_SKIP_NOARP))
|
|
|
|
break;
|
|
|
|
if (n->nud_state & ~NUD_NOARP)
|
|
|
|
break;
|
2010-10-07 02:49:21 +02:00
|
|
|
next:
|
|
|
|
n = rcu_dereference_bh(n->next);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
if (n)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
state->bucket = bucket;
|
|
|
|
|
|
|
|
return n;
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct neighbour *neigh_get_next(struct seq_file *seq,
|
|
|
|
struct neighbour *n,
|
|
|
|
loff_t *pos)
|
|
|
|
{
|
|
|
|
struct neigh_seq_state *state = seq->private;
|
2008-03-25 18:36:06 +01:00
|
|
|
struct net *net = seq_file_net(seq);
|
2010-10-04 08:15:44 +02:00
|
|
|
struct neigh_hash_table *nht = state->nht;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (state->neigh_sub_iter) {
|
|
|
|
void *v = state->neigh_sub_iter(state, n, pos);
|
|
|
|
if (v)
|
|
|
|
return n;
|
|
|
|
}
|
2010-10-07 02:49:21 +02:00
|
|
|
n = rcu_dereference_bh(n->next);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
while (1) {
|
|
|
|
while (n) {
|
2008-03-25 19:57:35 +01:00
|
|
|
if (!net_eq(dev_net(n->dev), net))
|
2008-01-24 09:13:18 +01:00
|
|
|
goto next;
|
2005-04-17 00:20:36 +02:00
|
|
|
if (state->neigh_sub_iter) {
|
|
|
|
void *v = state->neigh_sub_iter(state, n, pos);
|
|
|
|
if (v)
|
|
|
|
return n;
|
|
|
|
goto next;
|
|
|
|
}
|
|
|
|
if (!(state->flags & NEIGH_SEQ_SKIP_NOARP))
|
|
|
|
break;
|
|
|
|
|
|
|
|
if (n->nud_state & ~NUD_NOARP)
|
|
|
|
break;
|
2010-10-07 02:49:21 +02:00
|
|
|
next:
|
|
|
|
n = rcu_dereference_bh(n->next);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
if (n)
|
|
|
|
break;
|
|
|
|
|
2011-07-11 10:28:12 +02:00
|
|
|
if (++state->bucket >= (1 << nht->hash_shift))
|
2005-04-17 00:20:36 +02:00
|
|
|
break;
|
|
|
|
|
2010-10-07 02:49:21 +02:00
|
|
|
n = rcu_dereference_bh(nht->hash_buckets[state->bucket]);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
if (n && pos)
|
|
|
|
--(*pos);
|
|
|
|
return n;
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct neighbour *neigh_get_idx(struct seq_file *seq, loff_t *pos)
|
|
|
|
{
|
|
|
|
struct neighbour *n = neigh_get_first(seq);
|
|
|
|
|
|
|
|
if (n) {
|
2008-08-03 10:10:55 +02:00
|
|
|
--(*pos);
|
2005-04-17 00:20:36 +02:00
|
|
|
while (*pos) {
|
|
|
|
n = neigh_get_next(seq, n, pos);
|
|
|
|
if (!n)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return *pos ? NULL : n;
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct pneigh_entry *pneigh_get_first(struct seq_file *seq)
|
|
|
|
{
|
|
|
|
struct neigh_seq_state *state = seq->private;
|
2008-03-25 18:36:06 +01:00
|
|
|
struct net *net = seq_file_net(seq);
|
2005-04-17 00:20:36 +02:00
|
|
|
struct neigh_table *tbl = state->tbl;
|
|
|
|
struct pneigh_entry *pn = NULL;
|
|
|
|
int bucket = state->bucket;
|
|
|
|
|
|
|
|
state->flags |= NEIGH_SEQ_IS_PNEIGH;
|
|
|
|
for (bucket = 0; bucket <= PNEIGH_HASHMASK; bucket++) {
|
|
|
|
pn = tbl->phash_buckets[bucket];
|
2008-03-25 19:57:35 +01:00
|
|
|
while (pn && !net_eq(pneigh_net(pn), net))
|
2008-01-24 09:13:18 +01:00
|
|
|
pn = pn->next;
|
2005-04-17 00:20:36 +02:00
|
|
|
if (pn)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
state->bucket = bucket;
|
|
|
|
|
|
|
|
return pn;
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct pneigh_entry *pneigh_get_next(struct seq_file *seq,
|
|
|
|
struct pneigh_entry *pn,
|
|
|
|
loff_t *pos)
|
|
|
|
{
|
|
|
|
struct neigh_seq_state *state = seq->private;
|
2008-03-25 18:36:06 +01:00
|
|
|
struct net *net = seq_file_net(seq);
|
2005-04-17 00:20:36 +02:00
|
|
|
struct neigh_table *tbl = state->tbl;
|
|
|
|
|
2011-11-25 19:24:49 +01:00
|
|
|
do {
|
|
|
|
pn = pn->next;
|
|
|
|
} while (pn && !net_eq(pneigh_net(pn), net));
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
while (!pn) {
|
|
|
|
if (++state->bucket > PNEIGH_HASHMASK)
|
|
|
|
break;
|
|
|
|
pn = tbl->phash_buckets[state->bucket];
|
2008-03-25 19:57:35 +01:00
|
|
|
while (pn && !net_eq(pneigh_net(pn), net))
|
2008-01-24 09:13:18 +01:00
|
|
|
pn = pn->next;
|
2005-04-17 00:20:36 +02:00
|
|
|
if (pn)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (pn && pos)
|
|
|
|
--(*pos);
|
|
|
|
|
|
|
|
return pn;
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct pneigh_entry *pneigh_get_idx(struct seq_file *seq, loff_t *pos)
|
|
|
|
{
|
|
|
|
struct pneigh_entry *pn = pneigh_get_first(seq);
|
|
|
|
|
|
|
|
if (pn) {
|
2008-08-03 10:10:55 +02:00
|
|
|
--(*pos);
|
2005-04-17 00:20:36 +02:00
|
|
|
while (*pos) {
|
|
|
|
pn = pneigh_get_next(seq, pn, pos);
|
|
|
|
if (!pn)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return *pos ? NULL : pn;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void *neigh_get_idx_any(struct seq_file *seq, loff_t *pos)
|
|
|
|
{
|
|
|
|
struct neigh_seq_state *state = seq->private;
|
|
|
|
void *rc;
|
2008-08-03 10:10:55 +02:00
|
|
|
loff_t idxpos = *pos;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2008-08-03 10:10:55 +02:00
|
|
|
rc = neigh_get_idx(seq, &idxpos);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (!rc && !(state->flags & NEIGH_SEQ_NEIGH_ONLY))
|
2008-08-03 10:10:55 +02:00
|
|
|
rc = pneigh_get_idx(seq, &idxpos);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
|
|
|
|
void *neigh_seq_start(struct seq_file *seq, loff_t *pos, struct neigh_table *tbl, unsigned int neigh_seq_flags)
|
2010-10-04 08:15:44 +02:00
|
|
|
__acquires(rcu_bh)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct neigh_seq_state *state = seq->private;
|
|
|
|
|
|
|
|
state->tbl = tbl;
|
|
|
|
state->bucket = 0;
|
|
|
|
state->flags = (neigh_seq_flags & ~NEIGH_SEQ_IS_PNEIGH);
|
|
|
|
|
2010-10-04 08:15:44 +02:00
|
|
|
rcu_read_lock_bh();
|
|
|
|
state->nht = rcu_dereference_bh(tbl->nht);
|
2010-10-07 02:49:21 +02:00
|
|
|
|
2008-08-03 10:10:55 +02:00
|
|
|
return *pos ? neigh_get_idx_any(seq, pos) : SEQ_START_TOKEN;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(neigh_seq_start);
|
|
|
|
|
|
|
|
void *neigh_seq_next(struct seq_file *seq, void *v, loff_t *pos)
|
|
|
|
{
|
|
|
|
struct neigh_seq_state *state;
|
|
|
|
void *rc;
|
|
|
|
|
|
|
|
if (v == SEQ_START_TOKEN) {
|
2008-08-03 10:02:41 +02:00
|
|
|
rc = neigh_get_first(seq);
|
2005-04-17 00:20:36 +02:00
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
state = seq->private;
|
|
|
|
if (!(state->flags & NEIGH_SEQ_IS_PNEIGH)) {
|
|
|
|
rc = neigh_get_next(seq, v, NULL);
|
|
|
|
if (rc)
|
|
|
|
goto out;
|
|
|
|
if (!(state->flags & NEIGH_SEQ_NEIGH_ONLY))
|
|
|
|
rc = pneigh_get_first(seq);
|
|
|
|
} else {
|
|
|
|
BUG_ON(state->flags & NEIGH_SEQ_NEIGH_ONLY);
|
|
|
|
rc = pneigh_get_next(seq, v, NULL);
|
|
|
|
}
|
|
|
|
out:
|
|
|
|
++(*pos);
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(neigh_seq_next);
|
|
|
|
|
|
|
|
void neigh_seq_stop(struct seq_file *seq, void *v)
|
2010-10-04 08:15:44 +02:00
|
|
|
__releases(rcu_bh)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2010-10-04 08:15:44 +02:00
|
|
|
rcu_read_unlock_bh();
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(neigh_seq_stop);
|
|
|
|
|
|
|
|
/* statistics via seq_file */
|
|
|
|
|
|
|
|
static void *neigh_stat_seq_start(struct seq_file *seq, loff_t *pos)
|
|
|
|
{
|
2010-01-22 11:16:05 +01:00
|
|
|
struct neigh_table *tbl = seq->private;
|
2005-04-17 00:20:36 +02:00
|
|
|
int cpu;
|
|
|
|
|
|
|
|
if (*pos == 0)
|
|
|
|
return SEQ_START_TOKEN;
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2008-12-29 13:23:42 +01:00
|
|
|
for (cpu = *pos-1; cpu < nr_cpu_ids; ++cpu) {
|
2005-04-17 00:20:36 +02:00
|
|
|
if (!cpu_possible(cpu))
|
|
|
|
continue;
|
|
|
|
*pos = cpu+1;
|
|
|
|
return per_cpu_ptr(tbl->stats, cpu);
|
|
|
|
}
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void *neigh_stat_seq_next(struct seq_file *seq, void *v, loff_t *pos)
|
|
|
|
{
|
2010-01-22 11:16:05 +01:00
|
|
|
struct neigh_table *tbl = seq->private;
|
2005-04-17 00:20:36 +02:00
|
|
|
int cpu;
|
|
|
|
|
2008-12-29 13:23:42 +01:00
|
|
|
for (cpu = *pos; cpu < nr_cpu_ids; ++cpu) {
|
2005-04-17 00:20:36 +02:00
|
|
|
if (!cpu_possible(cpu))
|
|
|
|
continue;
|
|
|
|
*pos = cpu+1;
|
|
|
|
return per_cpu_ptr(tbl->stats, cpu);
|
|
|
|
}
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void neigh_stat_seq_stop(struct seq_file *seq, void *v)
|
|
|
|
{
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
static int neigh_stat_seq_show(struct seq_file *seq, void *v)
|
|
|
|
{
|
2010-01-22 11:16:05 +01:00
|
|
|
struct neigh_table *tbl = seq->private;
|
2005-04-17 00:20:36 +02:00
|
|
|
struct neigh_statistics *st = v;
|
|
|
|
|
|
|
|
if (v == SEQ_START_TOKEN) {
|
2008-07-17 05:50:49 +02:00
|
|
|
seq_printf(seq, "entries allocs destroys hash_grows lookups hits res_failed rcv_probes_mcast rcv_probes_ucast periodic_gc_runs forced_gc_runs unresolved_discards\n");
|
2005-04-17 00:20:36 +02:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
seq_printf(seq, "%08x %08lx %08lx %08lx %08lx %08lx %08lx "
|
2008-07-17 05:50:49 +02:00
|
|
|
"%08lx %08lx %08lx %08lx %08lx\n",
|
2005-04-17 00:20:36 +02:00
|
|
|
atomic_read(&tbl->entries),
|
|
|
|
|
|
|
|
st->allocs,
|
|
|
|
st->destroys,
|
|
|
|
st->hash_grows,
|
|
|
|
|
|
|
|
st->lookups,
|
|
|
|
st->hits,
|
|
|
|
|
|
|
|
st->res_failed,
|
|
|
|
|
|
|
|
st->rcv_probes_mcast,
|
|
|
|
st->rcv_probes_ucast,
|
|
|
|
|
|
|
|
st->periodic_gc_runs,
|
2008-07-17 05:50:49 +02:00
|
|
|
st->forced_gc_runs,
|
|
|
|
st->unres_discards
|
2005-04-17 00:20:36 +02:00
|
|
|
);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2007-03-12 22:34:29 +01:00
|
|
|
static const struct seq_operations neigh_stat_seq_ops = {
|
2005-04-17 00:20:36 +02:00
|
|
|
.start = neigh_stat_seq_start,
|
|
|
|
.next = neigh_stat_seq_next,
|
|
|
|
.stop = neigh_stat_seq_stop,
|
|
|
|
.show = neigh_stat_seq_show,
|
|
|
|
};
|
|
|
|
|
|
|
|
static int neigh_stat_seq_open(struct inode *inode, struct file *file)
|
|
|
|
{
|
|
|
|
int ret = seq_open(file, &neigh_stat_seq_ops);
|
|
|
|
|
|
|
|
if (!ret) {
|
|
|
|
struct seq_file *sf = file->private_data;
|
2013-04-01 00:16:14 +02:00
|
|
|
sf->private = PDE_DATA(inode);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
return ret;
|
|
|
|
};
|
|
|
|
|
2007-02-12 09:55:35 +01:00
|
|
|
static const struct file_operations neigh_stat_seq_fops = {
|
2005-04-17 00:20:36 +02:00
|
|
|
.owner = THIS_MODULE,
|
|
|
|
.open = neigh_stat_seq_open,
|
|
|
|
.read = seq_read,
|
|
|
|
.llseek = seq_lseek,
|
|
|
|
.release = seq_release,
|
|
|
|
};
|
|
|
|
|
|
|
|
#endif /* CONFIG_PROC_FS */
|
|
|
|
|
2006-11-10 23:10:15 +01:00
|
|
|
static inline size_t neigh_nlmsg_size(void)
|
|
|
|
{
|
|
|
|
return NLMSG_ALIGN(sizeof(struct ndmsg))
|
|
|
|
+ nla_total_size(MAX_ADDR_LEN) /* NDA_DST */
|
|
|
|
+ nla_total_size(MAX_ADDR_LEN) /* NDA_LLADDR */
|
|
|
|
+ nla_total_size(sizeof(struct nda_cacheinfo))
|
|
|
|
+ nla_total_size(4); /* NDA_PROBES */
|
|
|
|
}
|
|
|
|
|
2006-08-15 09:33:14 +02:00
|
|
|
static void __neigh_notify(struct neighbour *n, int type, int flags)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2008-03-25 13:47:49 +01:00
|
|
|
struct net *net = dev_net(n->dev);
|
2006-08-08 02:56:37 +02:00
|
|
|
struct sk_buff *skb;
|
2006-08-15 09:33:14 +02:00
|
|
|
int err = -ENOBUFS;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-11-10 23:10:15 +01:00
|
|
|
skb = nlmsg_new(neigh_nlmsg_size(), GFP_ATOMIC);
|
2006-08-08 02:56:37 +02:00
|
|
|
if (skb == NULL)
|
2006-08-15 09:33:14 +02:00
|
|
|
goto errout;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-08-15 09:33:14 +02:00
|
|
|
err = neigh_fill_info(skb, n, 0, 0, type, flags);
|
2007-02-01 08:16:40 +01:00
|
|
|
if (err < 0) {
|
|
|
|
/* -EMSGSIZE implies BUG in neigh_nlmsg_size() */
|
|
|
|
WARN_ON(err == -EMSGSIZE);
|
|
|
|
kfree_skb(skb);
|
|
|
|
goto errout;
|
|
|
|
}
|
2009-02-25 08:18:28 +01:00
|
|
|
rtnl_notify(skb, net, 0, RTNLGRP_NEIGH, NULL, GFP_ATOMIC);
|
|
|
|
return;
|
2006-08-15 09:33:14 +02:00
|
|
|
errout:
|
|
|
|
if (err < 0)
|
2008-01-24 09:13:18 +01:00
|
|
|
rtnl_set_sk_err(net, RTNLGRP_NEIGH, err);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2006-08-15 09:33:14 +02:00
|
|
|
void neigh_app_ns(struct neighbour *n)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-08-15 09:33:14 +02:00
|
|
|
__neigh_notify(n, RTM_GETNEIGH, NLM_F_REQUEST);
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_app_ns);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
#ifdef CONFIG_SYSCTL
|
2012-12-06 03:04:04 +01:00
|
|
|
static int zero;
|
2013-07-24 10:39:06 +02:00
|
|
|
static int int_max = INT_MAX;
|
2012-12-06 03:04:04 +01:00
|
|
|
static int unres_qlen_max = INT_MAX / SKB_TRUESIZE(ETH_FRAME_LEN);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2013-06-12 08:04:25 +02:00
|
|
|
static int proc_unres_qlen(struct ctl_table *ctl, int write,
|
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
{
|
|
|
|
int size, ret;
|
2013-06-12 08:04:25 +02:00
|
|
|
struct ctl_table tmp = *ctl;
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
|
2012-12-04 19:49:15 +01:00
|
|
|
tmp.extra1 = &zero;
|
|
|
|
tmp.extra2 = &unres_qlen_max;
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
tmp.data = &size;
|
2012-12-04 19:49:15 +01:00
|
|
|
|
|
|
|
size = *(int *)ctl->data / SKB_TRUESIZE(ETH_FRAME_LEN);
|
|
|
|
ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
|
|
|
|
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
if (write && !ret)
|
|
|
|
*(int *)ctl->data = size * SKB_TRUESIZE(ETH_FRAME_LEN);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2013-12-07 19:26:56 +01:00
|
|
|
static struct neigh_parms *neigh_get_dev_parms_rcu(struct net_device *dev,
|
|
|
|
int family)
|
|
|
|
{
|
2013-12-07 19:26:57 +01:00
|
|
|
switch (family) {
|
|
|
|
case AF_INET:
|
2013-12-07 19:26:56 +01:00
|
|
|
return __in_dev_arp_parms_get_rcu(dev);
|
2013-12-07 19:26:57 +01:00
|
|
|
case AF_INET6:
|
|
|
|
return __in6_dev_nd_parms_get_rcu(dev);
|
|
|
|
}
|
2013-12-07 19:26:56 +01:00
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void neigh_copy_dflt_parms(struct net *net, struct neigh_parms *p,
|
|
|
|
int index)
|
|
|
|
{
|
|
|
|
struct net_device *dev;
|
|
|
|
int family = neigh_parms_family(p);
|
|
|
|
|
|
|
|
rcu_read_lock();
|
|
|
|
for_each_netdev_rcu(net, dev) {
|
|
|
|
struct neigh_parms *dst_p =
|
|
|
|
neigh_get_dev_parms_rcu(dev, family);
|
|
|
|
|
|
|
|
if (dst_p && !test_bit(index, dst_p->data_state))
|
|
|
|
dst_p->data[index] = p->data[index];
|
|
|
|
}
|
|
|
|
rcu_read_unlock();
|
|
|
|
}
|
|
|
|
|
|
|
|
static void neigh_proc_update(struct ctl_table *ctl, int write)
|
|
|
|
{
|
|
|
|
struct net_device *dev = ctl->extra1;
|
|
|
|
struct neigh_parms *p = ctl->extra2;
|
2013-12-10 23:55:07 +01:00
|
|
|
struct net *net = neigh_parms_net(p);
|
2013-12-07 19:26:56 +01:00
|
|
|
int index = (int *) ctl->data - p->data;
|
|
|
|
|
|
|
|
if (!write)
|
|
|
|
return;
|
|
|
|
|
|
|
|
set_bit(index, p->data_state);
|
|
|
|
if (!dev) /* NULL dev means this is default value */
|
|
|
|
neigh_copy_dflt_parms(net, p, index);
|
|
|
|
}
|
|
|
|
|
2013-12-07 19:26:53 +01:00
|
|
|
static int neigh_proc_dointvec_zero_intmax(struct ctl_table *ctl, int write,
|
|
|
|
void __user *buffer,
|
|
|
|
size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
|
|
|
struct ctl_table tmp = *ctl;
|
2013-12-07 19:26:56 +01:00
|
|
|
int ret;
|
2013-12-07 19:26:53 +01:00
|
|
|
|
|
|
|
tmp.extra1 = &zero;
|
|
|
|
tmp.extra2 = &int_max;
|
|
|
|
|
2013-12-07 19:26:56 +01:00
|
|
|
ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
|
|
|
|
neigh_proc_update(ctl, write);
|
|
|
|
return ret;
|
2013-12-07 19:26:53 +01:00
|
|
|
}
|
|
|
|
|
2013-12-07 19:26:54 +01:00
|
|
|
int neigh_proc_dointvec(struct ctl_table *ctl, int write,
|
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
2013-12-07 19:26:56 +01:00
|
|
|
int ret = proc_dointvec(ctl, write, buffer, lenp, ppos);
|
|
|
|
|
|
|
|
neigh_proc_update(ctl, write);
|
|
|
|
return ret;
|
2013-12-07 19:26:54 +01:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(neigh_proc_dointvec);
|
|
|
|
|
|
|
|
int neigh_proc_dointvec_jiffies(struct ctl_table *ctl, int write,
|
|
|
|
void __user *buffer,
|
|
|
|
size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
2013-12-07 19:26:56 +01:00
|
|
|
int ret = proc_dointvec_jiffies(ctl, write, buffer, lenp, ppos);
|
|
|
|
|
|
|
|
neigh_proc_update(ctl, write);
|
|
|
|
return ret;
|
2013-12-07 19:26:54 +01:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(neigh_proc_dointvec_jiffies);
|
|
|
|
|
|
|
|
static int neigh_proc_dointvec_userhz_jiffies(struct ctl_table *ctl, int write,
|
|
|
|
void __user *buffer,
|
|
|
|
size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
2013-12-07 19:26:56 +01:00
|
|
|
int ret = proc_dointvec_userhz_jiffies(ctl, write, buffer, lenp, ppos);
|
|
|
|
|
|
|
|
neigh_proc_update(ctl, write);
|
|
|
|
return ret;
|
2013-12-07 19:26:54 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
int neigh_proc_dointvec_ms_jiffies(struct ctl_table *ctl, int write,
|
|
|
|
void __user *buffer,
|
|
|
|
size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
2013-12-07 19:26:56 +01:00
|
|
|
int ret = proc_dointvec_ms_jiffies(ctl, write, buffer, lenp, ppos);
|
|
|
|
|
|
|
|
neigh_proc_update(ctl, write);
|
|
|
|
return ret;
|
2013-12-07 19:26:54 +01:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(neigh_proc_dointvec_ms_jiffies);
|
|
|
|
|
|
|
|
static int neigh_proc_dointvec_unres_qlen(struct ctl_table *ctl, int write,
|
|
|
|
void __user *buffer,
|
|
|
|
size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
2013-12-07 19:26:56 +01:00
|
|
|
int ret = proc_unres_qlen(ctl, write, buffer, lenp, ppos);
|
|
|
|
|
|
|
|
neigh_proc_update(ctl, write);
|
|
|
|
return ret;
|
2013-12-07 19:26:54 +01:00
|
|
|
}
|
|
|
|
|
2013-12-07 19:26:53 +01:00
|
|
|
#define NEIGH_PARMS_DATA_OFFSET(index) \
|
|
|
|
(&((struct neigh_parms *) 0)->data[index])
|
|
|
|
|
|
|
|
#define NEIGH_SYSCTL_ENTRY(attr, data_attr, name, mval, proc) \
|
|
|
|
[NEIGH_VAR_ ## attr] = { \
|
|
|
|
.procname = name, \
|
|
|
|
.data = NEIGH_PARMS_DATA_OFFSET(NEIGH_VAR_ ## data_attr), \
|
|
|
|
.maxlen = sizeof(int), \
|
|
|
|
.mode = mval, \
|
|
|
|
.proc_handler = proc, \
|
|
|
|
}
|
|
|
|
|
|
|
|
#define NEIGH_SYSCTL_ZERO_INTMAX_ENTRY(attr, name) \
|
|
|
|
NEIGH_SYSCTL_ENTRY(attr, attr, name, 0644, neigh_proc_dointvec_zero_intmax)
|
|
|
|
|
|
|
|
#define NEIGH_SYSCTL_JIFFIES_ENTRY(attr, name) \
|
2013-12-07 19:26:54 +01:00
|
|
|
NEIGH_SYSCTL_ENTRY(attr, attr, name, 0644, neigh_proc_dointvec_jiffies)
|
2013-12-07 19:26:53 +01:00
|
|
|
|
|
|
|
#define NEIGH_SYSCTL_USERHZ_JIFFIES_ENTRY(attr, name) \
|
2013-12-07 19:26:54 +01:00
|
|
|
NEIGH_SYSCTL_ENTRY(attr, attr, name, 0644, neigh_proc_dointvec_userhz_jiffies)
|
2013-12-07 19:26:53 +01:00
|
|
|
|
|
|
|
#define NEIGH_SYSCTL_MS_JIFFIES_ENTRY(attr, name) \
|
2013-12-07 19:26:54 +01:00
|
|
|
NEIGH_SYSCTL_ENTRY(attr, attr, name, 0644, neigh_proc_dointvec_ms_jiffies)
|
2013-12-07 19:26:53 +01:00
|
|
|
|
|
|
|
#define NEIGH_SYSCTL_MS_JIFFIES_REUSED_ENTRY(attr, data_attr, name) \
|
2013-12-07 19:26:54 +01:00
|
|
|
NEIGH_SYSCTL_ENTRY(attr, data_attr, name, 0644, neigh_proc_dointvec_ms_jiffies)
|
2013-12-07 19:26:53 +01:00
|
|
|
|
|
|
|
#define NEIGH_SYSCTL_UNRES_QLEN_REUSED_ENTRY(attr, data_attr, name) \
|
2013-12-07 19:26:54 +01:00
|
|
|
NEIGH_SYSCTL_ENTRY(attr, data_attr, name, 0644, neigh_proc_dointvec_unres_qlen)
|
2010-02-14 04:27:03 +01:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
static struct neigh_sysctl_table {
|
|
|
|
struct ctl_table_header *sysctl_header;
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
struct ctl_table neigh_vars[NEIGH_VAR_MAX + 1];
|
2006-09-22 23:15:41 +02:00
|
|
|
} neigh_sysctl_template __read_mostly = {
|
2005-04-17 00:20:36 +02:00
|
|
|
.neigh_vars = {
|
2013-12-07 19:26:53 +01:00
|
|
|
NEIGH_SYSCTL_ZERO_INTMAX_ENTRY(MCAST_PROBES, "mcast_solicit"),
|
|
|
|
NEIGH_SYSCTL_ZERO_INTMAX_ENTRY(UCAST_PROBES, "ucast_solicit"),
|
|
|
|
NEIGH_SYSCTL_ZERO_INTMAX_ENTRY(APP_PROBES, "app_solicit"),
|
|
|
|
NEIGH_SYSCTL_USERHZ_JIFFIES_ENTRY(RETRANS_TIME, "retrans_time"),
|
|
|
|
NEIGH_SYSCTL_JIFFIES_ENTRY(BASE_REACHABLE_TIME, "base_reachable_time"),
|
|
|
|
NEIGH_SYSCTL_JIFFIES_ENTRY(DELAY_PROBE_TIME, "delay_first_probe_time"),
|
|
|
|
NEIGH_SYSCTL_JIFFIES_ENTRY(GC_STALETIME, "gc_stale_time"),
|
|
|
|
NEIGH_SYSCTL_ZERO_INTMAX_ENTRY(QUEUE_LEN_BYTES, "unres_qlen_bytes"),
|
|
|
|
NEIGH_SYSCTL_ZERO_INTMAX_ENTRY(PROXY_QLEN, "proxy_qlen"),
|
|
|
|
NEIGH_SYSCTL_USERHZ_JIFFIES_ENTRY(ANYCAST_DELAY, "anycast_delay"),
|
|
|
|
NEIGH_SYSCTL_USERHZ_JIFFIES_ENTRY(PROXY_DELAY, "proxy_delay"),
|
|
|
|
NEIGH_SYSCTL_USERHZ_JIFFIES_ENTRY(LOCKTIME, "locktime"),
|
|
|
|
NEIGH_SYSCTL_UNRES_QLEN_REUSED_ENTRY(QUEUE_LEN, QUEUE_LEN_BYTES, "unres_qlen"),
|
|
|
|
NEIGH_SYSCTL_MS_JIFFIES_REUSED_ENTRY(RETRANS_TIME_MS, RETRANS_TIME, "retrans_time_ms"),
|
|
|
|
NEIGH_SYSCTL_MS_JIFFIES_REUSED_ENTRY(BASE_REACHABLE_TIME_MS, BASE_REACHABLE_TIME, "base_reachable_time_ms"),
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
[NEIGH_VAR_GC_INTERVAL] = {
|
2005-04-17 00:20:36 +02:00
|
|
|
.procname = "gc_interval",
|
|
|
|
.maxlen = sizeof(int),
|
|
|
|
.mode = 0644,
|
2008-11-04 03:21:05 +01:00
|
|
|
.proc_handler = proc_dointvec_jiffies,
|
2005-04-17 00:20:36 +02:00
|
|
|
},
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
[NEIGH_VAR_GC_THRESH1] = {
|
2005-04-17 00:20:36 +02:00
|
|
|
.procname = "gc_thresh1",
|
|
|
|
.maxlen = sizeof(int),
|
|
|
|
.mode = 0644,
|
2013-07-24 10:39:06 +02:00
|
|
|
.extra1 = &zero,
|
|
|
|
.extra2 = &int_max,
|
|
|
|
.proc_handler = proc_dointvec_minmax,
|
2005-04-17 00:20:36 +02:00
|
|
|
},
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
[NEIGH_VAR_GC_THRESH2] = {
|
2005-04-17 00:20:36 +02:00
|
|
|
.procname = "gc_thresh2",
|
|
|
|
.maxlen = sizeof(int),
|
|
|
|
.mode = 0644,
|
2013-07-24 10:39:06 +02:00
|
|
|
.extra1 = &zero,
|
|
|
|
.extra2 = &int_max,
|
|
|
|
.proc_handler = proc_dointvec_minmax,
|
2005-04-17 00:20:36 +02:00
|
|
|
},
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
[NEIGH_VAR_GC_THRESH3] = {
|
2005-04-17 00:20:36 +02:00
|
|
|
.procname = "gc_thresh3",
|
|
|
|
.maxlen = sizeof(int),
|
|
|
|
.mode = 0644,
|
2013-07-24 10:39:06 +02:00
|
|
|
.extra1 = &zero,
|
|
|
|
.extra2 = &int_max,
|
|
|
|
.proc_handler = proc_dointvec_minmax,
|
2005-04-17 00:20:36 +02:00
|
|
|
},
|
2007-12-01 14:08:16 +01:00
|
|
|
{},
|
2005-04-17 00:20:36 +02:00
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
int neigh_sysctl_register(struct net_device *dev, struct neigh_parms *p,
|
2013-12-07 19:26:55 +01:00
|
|
|
proc_handler *handler)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2013-12-07 19:26:53 +01:00
|
|
|
int i;
|
2007-12-01 14:06:34 +01:00
|
|
|
struct neigh_sysctl_table *t;
|
2013-12-07 19:26:53 +01:00
|
|
|
const char *dev_name_source;
|
2012-04-19 15:38:03 +02:00
|
|
|
char neigh_path[ sizeof("net//neigh/") + IFNAMSIZ + IFNAMSIZ ];
|
2013-12-07 19:26:55 +01:00
|
|
|
char *p_name;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-12-01 14:06:34 +01:00
|
|
|
t = kmemdup(&neigh_sysctl_template, sizeof(*t), GFP_KERNEL);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (!t)
|
2007-12-01 14:06:34 +01:00
|
|
|
goto err;
|
|
|
|
|
2013-12-07 19:26:54 +01:00
|
|
|
for (i = 0; i < ARRAY_SIZE(t->neigh_vars); i++) {
|
2013-12-07 19:26:53 +01:00
|
|
|
t->neigh_vars[i].data += (long) p;
|
2013-12-07 19:26:54 +01:00
|
|
|
t->neigh_vars[i].extra1 = dev;
|
2013-12-07 19:26:56 +01:00
|
|
|
t->neigh_vars[i].extra2 = p;
|
2013-12-07 19:26:54 +01:00
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (dev) {
|
|
|
|
dev_name_source = dev->name;
|
2007-10-18 12:05:25 +02:00
|
|
|
/* Terminate the table early */
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
memset(&t->neigh_vars[NEIGH_VAR_GC_INTERVAL], 0,
|
|
|
|
sizeof(t->neigh_vars[NEIGH_VAR_GC_INTERVAL]));
|
2005-04-17 00:20:36 +02:00
|
|
|
} else {
|
2012-04-19 15:38:03 +02:00
|
|
|
dev_name_source = "default";
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
t->neigh_vars[NEIGH_VAR_GC_INTERVAL].data = (int *)(p + 1);
|
|
|
|
t->neigh_vars[NEIGH_VAR_GC_THRESH1].data = (int *)(p + 1) + 1;
|
|
|
|
t->neigh_vars[NEIGH_VAR_GC_THRESH2].data = (int *)(p + 1) + 2;
|
|
|
|
t->neigh_vars[NEIGH_VAR_GC_THRESH3].data = (int *)(p + 1) + 3;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2009-11-05 22:32:03 +01:00
|
|
|
if (handler) {
|
2005-04-17 00:20:36 +02:00
|
|
|
/* RetransTime */
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
t->neigh_vars[NEIGH_VAR_RETRANS_TIME].proc_handler = handler;
|
2005-04-17 00:20:36 +02:00
|
|
|
/* ReachableTime */
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
t->neigh_vars[NEIGH_VAR_BASE_REACHABLE_TIME].proc_handler = handler;
|
2005-04-17 00:20:36 +02:00
|
|
|
/* RetransTime (in milliseconds)*/
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
t->neigh_vars[NEIGH_VAR_RETRANS_TIME_MS].proc_handler = handler;
|
2005-04-17 00:20:36 +02:00
|
|
|
/* ReachableTime (in milliseconds) */
|
neigh: new unresolved queue limits
Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> > ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit. The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.
Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.
[PATCH V5 net-next] neigh: new unresolved queue limits
unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.
$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-09 13:07:14 +01:00
|
|
|
t->neigh_vars[NEIGH_VAR_BASE_REACHABLE_TIME_MS].proc_handler = handler;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2012-11-16 04:02:59 +01:00
|
|
|
/* Don't export sysctls to unprivileged users */
|
|
|
|
if (neigh_parms_net(p)->user_ns != &init_user_ns)
|
|
|
|
t->neigh_vars[0].procname = NULL;
|
|
|
|
|
2013-12-07 19:26:55 +01:00
|
|
|
switch (neigh_parms_family(p)) {
|
|
|
|
case AF_INET:
|
|
|
|
p_name = "ipv4";
|
|
|
|
break;
|
|
|
|
case AF_INET6:
|
|
|
|
p_name = "ipv6";
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
|
2012-04-19 15:38:03 +02:00
|
|
|
snprintf(neigh_path, sizeof(neigh_path), "net/%s/neigh/%s",
|
|
|
|
p_name, dev_name_source);
|
2008-02-29 05:48:01 +01:00
|
|
|
t->sysctl_header =
|
2012-04-19 15:38:03 +02:00
|
|
|
register_net_sysctl(neigh_parms_net(p), neigh_path, t->neigh_vars);
|
2007-12-01 14:06:34 +01:00
|
|
|
if (!t->sysctl_header)
|
2012-04-19 15:38:03 +02:00
|
|
|
goto free;
|
2007-12-01 14:06:34 +01:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
p->sysctl_table = t;
|
|
|
|
return 0;
|
|
|
|
|
2007-12-01 14:06:34 +01:00
|
|
|
free:
|
2005-04-17 00:20:36 +02:00
|
|
|
kfree(t);
|
2007-12-01 14:06:34 +01:00
|
|
|
err:
|
|
|
|
return -ENOBUFS;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_sysctl_register);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
void neigh_sysctl_unregister(struct neigh_parms *p)
|
|
|
|
{
|
|
|
|
if (p->sysctl_table) {
|
|
|
|
struct neigh_sysctl_table *t = p->sysctl_table;
|
|
|
|
p->sysctl_table = NULL;
|
2012-04-19 15:24:33 +02:00
|
|
|
unregister_net_sysctl_table(t->sysctl_header);
|
2005-04-17 00:20:36 +02:00
|
|
|
kfree(t);
|
|
|
|
}
|
|
|
|
}
|
2008-03-24 10:39:10 +01:00
|
|
|
EXPORT_SYMBOL(neigh_sysctl_unregister);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
#endif /* CONFIG_SYSCTL */
|
|
|
|
|
2007-03-22 19:50:06 +01:00
|
|
|
static int __init neigh_init(void)
|
|
|
|
{
|
2011-06-10 03:27:09 +02:00
|
|
|
rtnl_register(PF_UNSPEC, RTM_NEWNEIGH, neigh_add, NULL, NULL);
|
|
|
|
rtnl_register(PF_UNSPEC, RTM_DELNEIGH, neigh_delete, NULL, NULL);
|
|
|
|
rtnl_register(PF_UNSPEC, RTM_GETNEIGH, NULL, neigh_dump_info, NULL);
|
2007-03-22 19:50:06 +01:00
|
|
|
|
2011-06-10 03:27:09 +02:00
|
|
|
rtnl_register(PF_UNSPEC, RTM_GETNEIGHTBL, NULL, neightbl_dump_info,
|
|
|
|
NULL);
|
|
|
|
rtnl_register(PF_UNSPEC, RTM_SETNEIGHTBL, neightbl_set, NULL, NULL);
|
2007-03-22 19:50:06 +01:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
subsys_initcall(neigh_init);
|
|
|
|
|