redonkable/remarkable-linux
redonkable/remarkable-linux
1
0
Fork 0
Code Releases Activity

[NETFILTER]: x_tables: add port of hashlimit match for IPv4 and IPv6

Browse source 
Signed-off-by: Patrick McHardy <kaber@trash.net>
This commit is contained in:
Patrick McHardy 2006-11-29 02:35:36 +01:00 committed by David S. Miller
parent d7a5c32442
commit 39b46fc6f0
8 changed files with 340 additions and 286 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
include/linux/netfilter/Kbuild
View file

@ -14,6 +14,7 @@ header-y += xt_dscp.h
header-y += xt_DSCP.h header-y += xt_DSCP.h
header-y += xt_esp.h header-y += xt_esp.h
header-y += xt_helper.h header-y += xt_helper.h
header-y += xt_hashlimit.h
header-y += xt_length.h header-y += xt_length.h
header-y += xt_limit.h header-y += xt_limit.h
header-y += xt_mac.h header-y += xt_mac.h

40
include/linux/netfilter/xt_hashlimit.h Normal file
View file

@ -0,0 +1,40 @@
#ifndef _XT_HASHLIMIT_H
#define _XT_HASHLIMIT_H
/* timings are in milliseconds. */
#define XT_HASHLIMIT_SCALE 10000
/* 1/10,000 sec period => max of 10,000/sec. Min rate is then 429490
seconds, or one every 59 hours. */
/* details of this structure hidden by the implementation */
struct xt_hashlimit_htable;
#define XT_HASHLIMIT_HASH_DIP 0x0001
#define XT_HASHLIMIT_HASH_DPT 0x0002
#define XT_HASHLIMIT_HASH_SIP 0x0004
#define XT_HASHLIMIT_HASH_SPT 0x0008
struct hashlimit_cfg {
u_int32_t mode; /* bitmask of IPT_HASHLIMIT_HASH_* */
u_int32_t avg; /* Average secs between packets * scale */
u_int32_t burst; /* Period multiplier for upper limit. */
/* user specified */
u_int32_t size; /* how many buckets */
u_int32_t max; /* max number of entries */
u_int32_t gc_interval; /* gc interval */
u_int32_t expire; /* when do entries expire? */
};
struct xt_hashlimit_info {
char name [IFNAMSIZ]; /* name */
struct hashlimit_cfg cfg;
struct xt_hashlimit_htable *hinfo;
/* Used internally by the kernel */
union {
void *ptr;
struct xt_hashlimit_info *master;
} u;
};
#endif /*_XT_HASHLIMIT_H*/

40
include/linux/netfilter_ipv4/ipt_hashlimit.h
View file

@ -1,40 +1,14 @@
#ifndef _IPT_HASHLIMIT_H #ifndef _IPT_HASHLIMIT_H
#define _IPT_HASHLIMIT_H #define _IPT_HASHLIMIT_H
/* timings are in milliseconds. */ #include <linux/netfilter/xt_hashlimit.h>
#define IPT_HASHLIMIT_SCALE 10000
/* 1/10,000 sec period => max of 10,000/sec. Min rate is then 429490
seconds, or one every 59 hours. */
/* details of this structure hidden by the implementation */ #define IPT_HASHLIMIT_SCALE XT_HASHLIMIT_SCALE
struct ipt_hashlimit_htable; #define IPT_HASHLIMIT_HASH_DIP XT_HASHLIMIT_HASH_DIP
#define IPT_HASHLIMIT_HASH_DPT XT_HASHLIMIT_HASH_DPT
#define IPT_HASHLIMIT_HASH_SIP XT_HASHLIMIT_HASH_SIP
#define IPT_HASHLIMIT_HASH_SPT XT_HASHLIMIT_HASH_SPT
#define IPT_HASHLIMIT_HASH_DIP 0x0001 #define ipt_hashlimit_info xt_hashlimit_info
#define IPT_HASHLIMIT_HASH_DPT 0x0002
#define IPT_HASHLIMIT_HASH_SIP 0x0004
#define IPT_HASHLIMIT_HASH_SPT 0x0008
struct hashlimit_cfg {
u_int32_t mode; /* bitmask of IPT_HASHLIMIT_HASH_* */
u_int32_t avg; /* Average secs between packets * scale */
u_int32_t burst; /* Period multiplier for upper limit. */
/* user specified */
u_int32_t size; /* how many buckets */
u_int32_t max; /* max number of entries */
u_int32_t gc_interval; /* gc interval */
u_int32_t expire; /* when do entries expire? */
};
struct ipt_hashlimit_info {
char name [IFNAMSIZ]; /* name */
struct hashlimit_cfg cfg;
struct ipt_hashlimit_htable *hinfo;
/* Used internally by the kernel */
union {
void *ptr;
struct ipt_hashlimit_info *master;
} u;
};
#endif /* _IPT_HASHLIMIT_H */ #endif /* _IPT_HASHLIMIT_H */

14
net/ipv4/netfilter/Kconfig
View file

@ -326,20 +326,6 @@ config IP_NF_MATCH_ADDRTYPE
If you want to compile it as a module, say M here and read If you want to compile it as a module, say M here and read
<file:Documentation/modules.txt>. If unsure, say `N'. <file:Documentation/modules.txt>. If unsure, say `N'.
config IP_NF_MATCH_HASHLIMIT
tristate 'hashlimit match support'
depends on IP_NF_IPTABLES
help
This option adds a new iptables `hashlimit' match.
As opposed to `limit', this match dynamically creates a hash table
of limit buckets, based on your selection of source/destination
ip addresses and/or ports.
It enables you to express policies like `10kpps for any given
destination IP' or `500pps from any given source IP' with a single
IPtables rule.
# `filter', generic and specific targets # `filter', generic and specific targets
config IP_NF_FILTER config IP_NF_FILTER
tristate "Packet filtering" tristate "Packet filtering"

1
net/ipv4/netfilter/Makefile
View file

@ -53,7 +53,6 @@ obj-$(CONFIG_IP_NF_NAT) += iptable_nat.o
obj-$(CONFIG_IP_NF_RAW) += iptable_raw.o obj-$(CONFIG_IP_NF_RAW) += iptable_raw.o
# matches # matches
obj-$(CONFIG_IP_NF_MATCH_HASHLIMIT) += ipt_hashlimit.o
obj-$(CONFIG_IP_NF_MATCH_IPRANGE) += ipt_iprange.o obj-$(CONFIG_IP_NF_MATCH_IPRANGE) += ipt_iprange.o
obj-$(CONFIG_IP_NF_MATCH_OWNER) += ipt_owner.o obj-$(CONFIG_IP_NF_MATCH_OWNER) += ipt_owner.o
obj-$(CONFIG_IP_NF_MATCH_TOS) += ipt_tos.o obj-$(CONFIG_IP_NF_MATCH_TOS) += ipt_tos.o

14
net/netfilter/Kconfig
View file

@ -464,5 +464,19 @@ config NETFILTER_XT_MATCH_TCPMSS
To compile it as a module, choose M here. If unsure, say N. To compile it as a module, choose M here. If unsure, say N.
config NETFILTER_XT_MATCH_HASHLIMIT
tristate '"hashlimit" match support'
depends on NETFILTER_XTABLES
help
This option adds a `hashlimit' match.
As opposed to `limit', this match dynamically creates a hash table
of limit buckets, based on your selection of source/destination
addresses and/or ports.
It enables you to express policies like `10kpps for any given
destination address' or `500pps from any given source address'
with a single rule.
endmenu endmenu

1
net/netfilter/Makefile
View file

@ -59,3 +59,4 @@ obj-$(CONFIG_NETFILTER_XT_MATCH_STATISTIC) += xt_statistic.o
obj-$(CONFIG_NETFILTER_XT_MATCH_STRING) += xt_string.o obj-$(CONFIG_NETFILTER_XT_MATCH_STRING) += xt_string.o
obj-$(CONFIG_NETFILTER_XT_MATCH_TCPMSS) += xt_tcpmss.o obj-$(CONFIG_NETFILTER_XT_MATCH_TCPMSS) += xt_tcpmss.o
obj-$(CONFIG_NETFILTER_XT_MATCH_PHYSDEV) += xt_physdev.o obj-$(CONFIG_NETFILTER_XT_MATCH_PHYSDEV) += xt_physdev.o
obj-$(CONFIG_NETFILTER_XT_MATCH_HASHLIMIT) += xt_hashlimit.o

471
net/ipv4/netfilter/ipt_hashlimit.c → net/netfilter/xt_hashlimit.c
View file

@ -6,23 +6,8 @@
* $Id: ipt_hashlimit.c 3244 2004-10-20 16:24:29Z laforge@netfilter.org $ * $Id: ipt_hashlimit.c 3244 2004-10-20 16:24:29Z laforge@netfilter.org $
* *
* Development of this code was funded by Astaro AG, http://www.astaro.com/ * Development of this code was funded by Astaro AG, http://www.astaro.com/
*
* based on ipt_limit.c by:
* Jérôme de Vivie <devivie@info.enserb.u-bordeaux.fr>
* Hervé Eychenne <eychenne@info.enserb.u-bordeaux.fr>
* Rusty Russell <rusty@rustcorp.com.au>
*
* The general idea is to create a hash table for every dstip and have a
* seperate limit counter per tuple. This way you can do something like 'limit
* the number of syn packets for each of my internal addresses.
*
* Ideally this would just be implemented as a general 'hash' match, which would
* allow us to attach any iptables target to it's hash buckets. But this is
* not possible in the current iptables architecture. As always, pkttables for
* 2.7.x will help ;)
*/ */
#include <linux/module.h> #include <linux/module.h>
#include <linux/skbuff.h>
#include <linux/spinlock.h> #include <linux/spinlock.h>
#include <linux/random.h> #include <linux/random.h>
#include <linux/jhash.h> #include <linux/jhash.h>
@ -31,28 +16,40 @@
#include <linux/proc_fs.h> #include <linux/proc_fs.h>
#include <linux/seq_file.h> #include <linux/seq_file.h>
#include <linux/list.h> #include <linux/list.h>
#include <linux/skbuff.h>
#include <linux/in.h>
#include <linux/ip.h>
#include <linux/ipv6.h>
#include <linux/netfilter/x_tables.h>
#include <linux/netfilter_ipv4/ip_tables.h> #include <linux/netfilter_ipv4/ip_tables.h>
#include <linux/netfilter_ipv4/ipt_hashlimit.h> #include <linux/netfilter_ipv6/ip6_tables.h>
#include <linux/netfilter/xt_hashlimit.h>
/* FIXME: this is just for IP_NF_ASSERRT */
#include <linux/netfilter_ipv4/ip_conntrack.h>
#include <linux/mutex.h> #include <linux/mutex.h>
MODULE_LICENSE("GPL"); MODULE_LICENSE("GPL");
MODULE_AUTHOR("Harald Welte <laforge@netfilter.org>"); MODULE_AUTHOR("Harald Welte <laforge@netfilter.org>");
MODULE_DESCRIPTION("iptables match for limiting per hash-bucket"); MODULE_DESCRIPTION("iptables match for limiting per hash-bucket");
MODULE_ALIAS("ipt_hashlimit");
MODULE_ALIAS("ip6t_hashlimit");
/* need to declare this at the top */ /* need to declare this at the top */
static struct proc_dir_entry *hashlimit_procdir; static struct proc_dir_entry *hashlimit_procdir4;
static struct proc_dir_entry *hashlimit_procdir6;
static struct file_operations dl_file_ops; static struct file_operations dl_file_ops;
/* hash table crap */ /* hash table crap */
struct dsthash_dst { struct dsthash_dst {
__be32 src_ip; union {
__be32 dst_ip; struct {
/* ports have to be consecutive !!! */ __be32 src;
__be32 dst;
} ip;
struct {
__be32 src[4];
__be32 dst[4];
} ip6;
} addr;
__be16 src_port; __be16 src_port;
__be16 dst_port; __be16 dst_port;
}; };
@ -71,9 +68,10 @@ struct dsthash_ent {
} rateinfo; } rateinfo;
}; };
struct ipt_hashlimit_htable { struct xt_hashlimit_htable {
struct hlist_node node; /* global list of all htables */ struct hlist_node node; /* global list of all htables */
atomic_t use; atomic_t use;
int family;
struct hashlimit_cfg cfg; /* config */ struct hashlimit_cfg cfg; /* config */
@ -81,8 +79,8 @@ struct ipt_hashlimit_htable {
spinlock_t lock; /* lock for list_head */ spinlock_t lock; /* lock for list_head */
u_int32_t rnd; /* random seed for hash */ u_int32_t rnd; /* random seed for hash */
int rnd_initialized; int rnd_initialized;
unsigned int count; /* number entries in table */
struct timer_list timer; /* timer for gc */ struct timer_list timer; /* timer for gc */
atomic_t count; /* number entries in table */
/* seq_file stuff */ /* seq_file stuff */
struct proc_dir_entry *pde; struct proc_dir_entry *pde;
@ -97,41 +95,33 @@ static kmem_cache_t *hashlimit_cachep __read_mostly;
static inline int dst_cmp(const struct dsthash_ent *ent, struct dsthash_dst *b) static inline int dst_cmp(const struct dsthash_ent *ent, struct dsthash_dst *b)
{ {
return (ent->dst.dst_ip == b->dst_ip return !memcmp(&ent->dst, b, sizeof(ent->dst));
&& ent->dst.dst_port == b->dst_port
&& ent->dst.src_port == b->src_port
&& ent->dst.src_ip == b->src_ip);
} }
static inline u_int32_t static u_int32_t
hash_dst(const struct ipt_hashlimit_htable *ht, const struct dsthash_dst *dst) hash_dst(const struct xt_hashlimit_htable *ht, const struct dsthash_dst *dst)
{ {
return (jhash_3words((__force u32)dst->dst_ip, return jhash(dst, sizeof(*dst), ht->rnd) % ht->cfg.size;
((__force u32)dst->dst_port<<16 |
(__force u32)dst->src_port),
(__force u32)dst->src_ip, ht->rnd) % ht->cfg.size);
} }
static inline struct dsthash_ent * static struct dsthash_ent *
__dsthash_find(const struct ipt_hashlimit_htable *ht, struct dsthash_dst *dst) dsthash_find(const struct xt_hashlimit_htable *ht, struct dsthash_dst *dst)
{ {
struct dsthash_ent *ent; struct dsthash_ent *ent;
struct hlist_node *pos; struct hlist_node *pos;
u_int32_t hash = hash_dst(ht, dst); u_int32_t hash = hash_dst(ht, dst);
if (!hlist_empty(&ht->hash[hash])) if (!hlist_empty(&ht->hash[hash])) {
hlist_for_each_entry(ent, pos, &ht->hash[hash], node) { hlist_for_each_entry(ent, pos, &ht->hash[hash], node)
if (dst_cmp(ent, dst)) { if (dst_cmp(ent, dst))
return ent; return ent;
} }
}
return NULL; return NULL;
} }
/* allocate dsthash_ent, initialize dst, put in htable and lock it */ /* allocate dsthash_ent, initialize dst, put in htable and lock it */
static struct dsthash_ent * static struct dsthash_ent *
__dsthash_alloc_init(struct ipt_hashlimit_htable *ht, struct dsthash_dst *dst) dsthash_alloc_init(struct xt_hashlimit_htable *ht, struct dsthash_dst *dst)
{ {
struct dsthash_ent *ent; struct dsthash_ent *ent;
@ -142,12 +132,11 @@ __dsthash_alloc_init(struct ipt_hashlimit_htable *ht, struct dsthash_dst *dst)
ht->rnd_initialized = 1; ht->rnd_initialized = 1;
} }
if (ht->cfg.max && if (ht->cfg.max && ht->count >= ht->cfg.max) {
atomic_read(&ht->count) >= ht->cfg.max) {
/* FIXME: do something. question is what.. */ /* FIXME: do something. question is what.. */
if (net_ratelimit()) if (net_ratelimit())
printk(KERN_WARNING printk(KERN_WARNING
"ipt_hashlimit: max count of %u reached\n", "xt_hashlimit: max count of %u reached\n",
ht->cfg.max); ht->cfg.max);
return NULL; return NULL;
} }
@ -156,52 +145,46 @@ __dsthash_alloc_init(struct ipt_hashlimit_htable *ht, struct dsthash_dst *dst)
if (!ent) { if (!ent) {
if (net_ratelimit()) if (net_ratelimit())
printk(KERN_ERR printk(KERN_ERR
"ipt_hashlimit: can't allocate dsthash_ent\n"); "xt_hashlimit: can't allocate dsthash_ent\n");
return NULL; return NULL;
} }
memcpy(&ent->dst, dst, sizeof(ent->dst));
atomic_inc(&ht->count);
ent->dst.dst_ip = dst->dst_ip;
ent->dst.dst_port = dst->dst_port;
ent->dst.src_ip = dst->src_ip;
ent->dst.src_port = dst->src_port;
hlist_add_head(&ent->node, &ht->hash[hash_dst(ht, dst)]); hlist_add_head(&ent->node, &ht->hash[hash_dst(ht, dst)]);
ht->count++;
return ent; return ent;
} }
static inline void static inline void
__dsthash_free(struct ipt_hashlimit_htable *ht, struct dsthash_ent *ent) dsthash_free(struct xt_hashlimit_htable *ht, struct dsthash_ent *ent)
{ {
hlist_del(&ent->node); hlist_del(&ent->node);
kmem_cache_free(hashlimit_cachep, ent); kmem_cache_free(hashlimit_cachep, ent);
atomic_dec(&ht->count); ht->count--;
} }
static void htable_gc(unsigned long htlong); static void htable_gc(unsigned long htlong);
static int htable_create(struct ipt_hashlimit_info *minfo) static int htable_create(struct xt_hashlimit_info *minfo, int family)
{ {
int i; struct xt_hashlimit_htable *hinfo;
unsigned int size; unsigned int size;
struct ipt_hashlimit_htable *hinfo; unsigned int i;
if (minfo->cfg.size) if (minfo->cfg.size)
size = minfo->cfg.size; size = minfo->cfg.size;
else { else {
size = (((num_physpages << PAGE_SHIFT) / 16384) size = ((num_physpages << PAGE_SHIFT) / 16384) /
/ sizeof(struct list_head)); sizeof(struct list_head);
if (num_physpages > (1024 * 1024 * 1024 / PAGE_SIZE)) if (num_physpages > (1024 * 1024 * 1024 / PAGE_SIZE))
size = 8192; size = 8192;
if (size < 16) if (size < 16)
size = 16; size = 16;
} }
/* FIXME: don't use vmalloc() here or anywhere else -HW */ /* FIXME: don't use vmalloc() here or anywhere else -HW */
hinfo = vmalloc(sizeof(struct ipt_hashlimit_htable) hinfo = vmalloc(sizeof(struct xt_hashlimit_htable) +
+ (sizeof(struct list_head) * size)); sizeof(struct list_head) * size);
if (!hinfo) { if (!hinfo) {
printk(KERN_ERR "ipt_hashlimit: Unable to create hashtable\n"); printk(KERN_ERR "xt_hashlimit: unable to create hashtable\n");
return -1; return -1;
} }
minfo->hinfo = hinfo; minfo->hinfo = hinfo;
@ -217,11 +200,14 @@ static int htable_create(struct ipt_hashlimit_info *minfo)
for (i = 0; i < hinfo->cfg.size; i++) for (i = 0; i < hinfo->cfg.size; i++)
INIT_HLIST_HEAD(&hinfo->hash[i]); INIT_HLIST_HEAD(&hinfo->hash[i]);
atomic_set(&hinfo->count, 0);
atomic_set(&hinfo->use, 1); atomic_set(&hinfo->use, 1);
hinfo->count = 0;
hinfo->family = family;
hinfo->rnd_initialized = 0; hinfo->rnd_initialized = 0;
spin_lock_init(&hinfo->lock); spin_lock_init(&hinfo->lock);
hinfo->pde = create_proc_entry(minfo->name, 0, hashlimit_procdir); hinfo->pde = create_proc_entry(minfo->name, 0,
family == AF_INET ? hashlimit_procdir4 :
hashlimit_procdir6);
if (!hinfo->pde) { if (!hinfo->pde) {
vfree(hinfo); vfree(hinfo);
return -1; return -1;
@ -242,23 +228,21 @@ static int htable_create(struct ipt_hashlimit_info *minfo)
return 0; return 0;
} }
static int select_all(struct ipt_hashlimit_htable *ht, struct dsthash_ent *he) static int select_all(struct xt_hashlimit_htable *ht, struct dsthash_ent *he)
{ {
return 1; return 1;
} }
static int select_gc(struct ipt_hashlimit_htable *ht, struct dsthash_ent *he) static int select_gc(struct xt_hashlimit_htable *ht, struct dsthash_ent *he)
{ {
return (jiffies >= he->expires); return (jiffies >= he->expires);
} }
static void htable_selective_cleanup(struct ipt_hashlimit_htable *ht, static void htable_selective_cleanup(struct xt_hashlimit_htable *ht,
int (*select)(struct ipt_hashlimit_htable *ht, int (*select)(struct xt_hashlimit_htable *ht,
struct dsthash_ent *he)) struct dsthash_ent *he))
{ {
int i; unsigned int i;
IP_NF_ASSERT(ht->cfg.size && ht->cfg.max);
/* lock hash table and iterate over it */ /* lock hash table and iterate over it */
spin_lock_bh(&ht->lock); spin_lock_bh(&ht->lock);
@ -267,7 +251,7 @@ static void htable_selective_cleanup(struct ipt_hashlimit_htable *ht,
struct hlist_node *pos, *n; struct hlist_node *pos, *n;
hlist_for_each_entry_safe(dh, pos, n, &ht->hash[i], node) { hlist_for_each_entry_safe(dh, pos, n, &ht->hash[i], node) {
if ((*select)(ht, dh)) if ((*select)(ht, dh))
__dsthash_free(ht, dh); dsthash_free(ht, dh);
} }
} }
spin_unlock_bh(&ht->lock); spin_unlock_bh(&ht->lock);
@ -276,7 +260,7 @@ static void htable_selective_cleanup(struct ipt_hashlimit_htable *ht,
/* hash table garbage collector, run by timer */ /* hash table garbage collector, run by timer */
static void htable_gc(unsigned long htlong) static void htable_gc(unsigned long htlong)
{ {
struct ipt_hashlimit_htable *ht = (struct ipt_hashlimit_htable *)htlong; struct xt_hashlimit_htable *ht = (struct xt_hashlimit_htable *)htlong;
htable_selective_cleanup(ht, select_gc); htable_selective_cleanup(ht, select_gc);
@ -285,38 +269,39 @@ static void htable_gc(unsigned long htlong)
add_timer(&ht->timer); add_timer(&ht->timer);
} }
static void htable_destroy(struct ipt_hashlimit_htable *hinfo) static void htable_destroy(struct xt_hashlimit_htable *hinfo)
{ {
/* remove timer, if it is pending */ /* remove timer, if it is pending */
if (timer_pending(&hinfo->timer)) if (timer_pending(&hinfo->timer))
del_timer(&hinfo->timer); del_timer(&hinfo->timer);
/* remove proc entry */ /* remove proc entry */
remove_proc_entry(hinfo->pde->name, hashlimit_procdir); remove_proc_entry(hinfo->pde->name,
hinfo->family == AF_INET ? hashlimit_procdir4 :
hashlimit_procdir6);
htable_selective_cleanup(hinfo, select_all); htable_selective_cleanup(hinfo, select_all);
vfree(hinfo); vfree(hinfo);
} }
static struct ipt_hashlimit_htable *htable_find_get(char *name) static struct xt_hashlimit_htable *htable_find_get(char *name, int family)
{ {
struct ipt_hashlimit_htable *hinfo; struct xt_hashlimit_htable *hinfo;
struct hlist_node *pos; struct hlist_node *pos;
spin_lock_bh(&hashlimit_lock); spin_lock_bh(&hashlimit_lock);
hlist_for_each_entry(hinfo, pos, &hashlimit_htables, node) { hlist_for_each_entry(hinfo, pos, &hashlimit_htables, node) {
if (!strcmp(name, hinfo->pde->name)) { if (!strcmp(name, hinfo->pde->name) &&
hinfo->family == family) {
atomic_inc(&hinfo->use); atomic_inc(&hinfo->use);
spin_unlock_bh(&hashlimit_lock); spin_unlock_bh(&hashlimit_lock);
return hinfo; return hinfo;
} }
} }
spin_unlock_bh(&hashlimit_lock); spin_unlock_bh(&hashlimit_lock);
return NULL; return NULL;
} }
static void htable_put(struct ipt_hashlimit_htable *hinfo) static void htable_put(struct xt_hashlimit_htable *hinfo)
{ {
if (atomic_dec_and_test(&hinfo->use)) { if (atomic_dec_and_test(&hinfo->use)) {
spin_lock_bh(&hashlimit_lock); spin_lock_bh(&hashlimit_lock);
@ -326,7 +311,6 @@ static void htable_put(struct ipt_hashlimit_htable *hinfo)
} }
} }
/* The algorithm used is the Simple Token Bucket Filter (TBF) /* The algorithm used is the Simple Token Bucket Filter (TBF)
* see net/sched/sch_tbf.c in the linux source tree * see net/sched/sch_tbf.c in the linux source tree
*/ */
@ -370,17 +354,82 @@ user2credits(u_int32_t user)
/* If multiplying would overflow... */ /* If multiplying would overflow... */
if (user > 0xFFFFFFFF / (HZ*CREDITS_PER_JIFFY)) if (user > 0xFFFFFFFF / (HZ*CREDITS_PER_JIFFY))
/* Divide first. */ /* Divide first. */
return (user / IPT_HASHLIMIT_SCALE) * HZ * CREDITS_PER_JIFFY; return (user / XT_HASHLIMIT_SCALE) * HZ * CREDITS_PER_JIFFY;
return (user * HZ * CREDITS_PER_JIFFY) / IPT_HASHLIMIT_SCALE; return (user * HZ * CREDITS_PER_JIFFY) / XT_HASHLIMIT_SCALE;
} }
static inline void rateinfo_recalc(struct dsthash_ent *dh, unsigned long now) static inline void rateinfo_recalc(struct dsthash_ent *dh, unsigned long now)
{ {
dh->rateinfo.credit += (now - xchg(&dh->rateinfo.prev, now)) dh->rateinfo.credit += (now - dh->rateinfo.prev) * CREDITS_PER_JIFFY;
* CREDITS_PER_JIFFY;
if (dh->rateinfo.credit > dh->rateinfo.credit_cap) if (dh->rateinfo.credit > dh->rateinfo.credit_cap)
dh->rateinfo.credit = dh->rateinfo.credit_cap; dh->rateinfo.credit = dh->rateinfo.credit_cap;
dh->rateinfo.prev = now;
}
static int
hashlimit_init_dst(struct xt_hashlimit_htable *hinfo, struct dsthash_dst *dst,
const struct sk_buff *skb, unsigned int protoff)
{
__be16 _ports[2], *ports;
int nexthdr;
memset(dst, 0, sizeof(*dst));
switch (hinfo->family) {
case AF_INET:
if (hinfo->cfg.mode & XT_HASHLIMIT_HASH_DIP)
dst->addr.ip.dst = skb->nh.iph->daddr;
if (hinfo->cfg.mode & XT_HASHLIMIT_HASH_SIP)
dst->addr.ip.src = skb->nh.iph->saddr;
if (!(hinfo->cfg.mode &
(XT_HASHLIMIT_HASH_DPT | XT_HASHLIMIT_HASH_SPT)))
return 0;
nexthdr = skb->nh.iph->protocol;
break;
#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
case AF_INET6:
if (hinfo->cfg.mode & XT_HASHLIMIT_HASH_DIP)
memcpy(&dst->addr.ip6.dst, &skb->nh.ipv6h->daddr,
sizeof(dst->addr.ip6.dst));
if (hinfo->cfg.mode & XT_HASHLIMIT_HASH_SIP)
memcpy(&dst->addr.ip6.src, &skb->nh.ipv6h->saddr,
sizeof(dst->addr.ip6.src));
if (!(hinfo->cfg.mode &
(XT_HASHLIMIT_HASH_DPT | XT_HASHLIMIT_HASH_SPT)))
return 0;
nexthdr = ipv6_find_hdr(skb, &protoff, -1, NULL);
if (nexthdr < 0)
return -1;
break;
#endif
default:
BUG();
return 0;
}
switch (nexthdr) {
case IPPROTO_TCP:
case IPPROTO_UDP:
case IPPROTO_SCTP:
case IPPROTO_DCCP:
ports = skb_header_pointer(skb, protoff, sizeof(_ports),
&_ports);
break;
default:
_ports[0] = _ports[1] = 0;
ports = _ports;
break;
}
if (!ports)
return -1;
if (hinfo->cfg.mode & XT_HASHLIMIT_HASH_SPT)
dst->src_port = ports[0];
if (hinfo->cfg.mode & XT_HASHLIMIT_HASH_DPT)
dst->dst_port = ports[1];
return 0;
} }
static int static int
@ -393,63 +442,26 @@ hashlimit_match(const struct sk_buff *skb,
unsigned int protoff, unsigned int protoff,
int *hotdrop) int *hotdrop)
{ {
struct ipt_hashlimit_info *r = struct xt_hashlimit_info *r =
((struct ipt_hashlimit_info *)matchinfo)->u.master; ((struct xt_hashlimit_info *)matchinfo)->u.master;
struct ipt_hashlimit_htable *hinfo = r->hinfo; struct xt_hashlimit_htable *hinfo = r->hinfo;
unsigned long now = jiffies; unsigned long now = jiffies;
struct dsthash_ent *dh; struct dsthash_ent *dh;
struct dsthash_dst dst; struct dsthash_dst dst;
/* build 'dst' according to hinfo->cfg and current packet */ if (hashlimit_init_dst(hinfo, &dst, skb, protoff) < 0)
memset(&dst, 0, sizeof(dst)); goto hotdrop;
if (hinfo->cfg.mode & IPT_HASHLIMIT_HASH_DIP)
dst.dst_ip = skb->nh.iph->daddr;
if (hinfo->cfg.mode & IPT_HASHLIMIT_HASH_SIP)
dst.src_ip = skb->nh.iph->saddr;
if (hinfo->cfg.mode & IPT_HASHLIMIT_HASH_DPT
||hinfo->cfg.mode & IPT_HASHLIMIT_HASH_SPT) {
__be16 _ports[2], *ports;
switch (skb->nh.iph->protocol) {
case IPPROTO_TCP:
case IPPROTO_UDP:
case IPPROTO_SCTP:
case IPPROTO_DCCP:
ports = skb_header_pointer(skb, skb->nh.iph->ihl*4,
sizeof(_ports), &_ports);
break;
default:
_ports[0] = _ports[1] = 0;
ports = _ports;
break;
}
if (!ports) {
/* We've been asked to examine this packet, and we
can't. Hence, no choice but to drop. */
*hotdrop = 1;
return 0;
}
if (hinfo->cfg.mode & IPT_HASHLIMIT_HASH_SPT)
dst.src_port = ports[0];
if (hinfo->cfg.mode & IPT_HASHLIMIT_HASH_DPT)
dst.dst_port = ports[1];
}
spin_lock_bh(&hinfo->lock); spin_lock_bh(&hinfo->lock);
dh = __dsthash_find(hinfo, &dst); dh = dsthash_find(hinfo, &dst);
if (!dh) { if (!dh) {
dh = __dsthash_alloc_init(hinfo, &dst); dh = dsthash_alloc_init(hinfo, &dst);
if (!dh) { if (!dh) {
/* enomem... don't match == DROP */
if (net_ratelimit())
printk(KERN_ERR "%s: ENOMEM\n", __FUNCTION__);
spin_unlock_bh(&hinfo->lock); spin_unlock_bh(&hinfo->lock);
return 0; goto hotdrop;
} }
dh->expires = jiffies + msecs_to_jiffies(hinfo->cfg.expire); dh->expires = jiffies + msecs_to_jiffies(hinfo->cfg.expire);
dh->rateinfo.prev = jiffies; dh->rateinfo.prev = jiffies;
dh->rateinfo.credit = user2credits(hinfo->cfg.avg * dh->rateinfo.credit = user2credits(hinfo->cfg.avg *
hinfo->cfg.burst); hinfo->cfg.burst);
@ -473,6 +485,10 @@ hashlimit_match(const struct sk_buff *skb,
/* default case: we're overlimit, thus don't match */ /* default case: we're overlimit, thus don't match */
return 0; return 0;
hotdrop:
*hotdrop = 1;
return 0;
} }
static int static int
@ -482,42 +498,37 @@ hashlimit_checkentry(const char *tablename,
void *matchinfo, void *matchinfo,
unsigned int hook_mask) unsigned int hook_mask)
{ {
struct ipt_hashlimit_info *r = matchinfo; struct xt_hashlimit_info *r = matchinfo;
/* Check for overflow. */ /* Check for overflow. */
if (r->cfg.burst == 0 if (r->cfg.burst == 0 ||
|| user2credits(r->cfg.avg * r->cfg.burst) < user2credits(r->cfg.avg * r->cfg.burst) < user2credits(r->cfg.avg)) {
user2credits(r->cfg.avg)) { printk(KERN_ERR "xt_hashlimit: overflow, try lower: %u/%u\n",
printk(KERN_ERR "ipt_hashlimit: Overflow, try lower: %u/%u\n",
r->cfg.avg, r->cfg.burst); r->cfg.avg, r->cfg.burst);
return 0; return 0;
} }
if (r->cfg.mode == 0 ||
if (r->cfg.mode == 0 r->cfg.mode > (XT_HASHLIMIT_HASH_DPT |
|| r->cfg.mode > (IPT_HASHLIMIT_HASH_DPT XT_HASHLIMIT_HASH_DIP |
|IPT_HASHLIMIT_HASH_DIP XT_HASHLIMIT_HASH_SIP |
|IPT_HASHLIMIT_HASH_SIP XT_HASHLIMIT_HASH_SPT))
|IPT_HASHLIMIT_HASH_SPT))
return 0; return 0;
if (!r->cfg.gc_interval) if (!r->cfg.gc_interval)
return 0; return 0;
if (!r->cfg.expire) if (!r->cfg.expire)
return 0; return 0;
if (r->name[sizeof(r->name) - 1] != '\0') if (r->name[sizeof(r->name) - 1] != '\0')
return 0; return 0;
/* This is the best we've got: We cannot release and re-grab lock, /* This is the best we've got: We cannot release and re-grab lock,
* since checkentry() is called before ip_tables.c grabs ipt_mutex. * since checkentry() is called before x_tables.c grabs xt_mutex.
* We also cannot grab the hashtable spinlock, since htable_create will * We also cannot grab the hashtable spinlock, since htable_create will
* call vmalloc, and that can sleep. And we cannot just re-search * call vmalloc, and that can sleep. And we cannot just re-search
* the list of htable's in htable_create(), since then we would * the list of htable's in htable_create(), since then we would
* create duplicate proc files. -HW */ * create duplicate proc files. -HW */
mutex_lock(&hlimit_mutex); mutex_lock(&hlimit_mutex);
r->hinfo = htable_find_get(r->name); r->hinfo = htable_find_get(r->name, match->family);
if (!r->hinfo && (htable_create(r) != 0)) { if (!r->hinfo && htable_create(r, match->family) != 0) {
mutex_unlock(&hlimit_mutex); mutex_unlock(&hlimit_mutex);
return 0; return 0;
} }
@ -525,20 +536,19 @@ hashlimit_checkentry(const char *tablename,
/* Ugly hack: For SMP, we only want to use one set */ /* Ugly hack: For SMP, we only want to use one set */
r->u.master = r; r->u.master = r;
return 1; return 1;
} }
static void static void
hashlimit_destroy(const struct xt_match *match, void *matchinfo) hashlimit_destroy(const struct xt_match *match, void *matchinfo)
{ {
struct ipt_hashlimit_info *r = matchinfo; struct xt_hashlimit_info *r = matchinfo;
htable_put(r->hinfo); htable_put(r->hinfo);
} }
#ifdef CONFIG_COMPAT #ifdef CONFIG_COMPAT
struct compat_ipt_hashlimit_info { struct compat_xt_hashlimit_info {
char name[IFNAMSIZ]; char name[IFNAMSIZ];
struct hashlimit_cfg cfg; struct hashlimit_cfg cfg;
compat_uptr_t hinfo; compat_uptr_t hinfo;
@ -547,40 +557,56 @@ struct compat_ipt_hashlimit_info {
static void compat_from_user(void *dst, void *src) static void compat_from_user(void *dst, void *src)
{ {
int off = offsetof(struct compat_ipt_hashlimit_info, hinfo); int off = offsetof(struct compat_xt_hashlimit_info, hinfo);
memcpy(dst, src, off); memcpy(dst, src, off);
memset(dst + off, 0, sizeof(struct compat_ipt_hashlimit_info) - off); memset(dst + off, 0, sizeof(struct compat_xt_hashlimit_info) - off);
} }
static int compat_to_user(void __user *dst, void *src) static int compat_to_user(void __user *dst, void *src)
{ {
int off = offsetof(struct compat_ipt_hashlimit_info, hinfo); int off = offsetof(struct compat_xt_hashlimit_info, hinfo);
return copy_to_user(dst, src, off) ? -EFAULT : 0; return copy_to_user(dst, src, off) ? -EFAULT : 0;
} }
#endif #endif
static struct ipt_match ipt_hashlimit = { static struct xt_match xt_hashlimit[] = {
{
.name = "hashlimit", .name = "hashlimit",
.family = AF_INET,
.match = hashlimit_match, .match = hashlimit_match,
.matchsize = sizeof(struct ipt_hashlimit_info), .matchsize = sizeof(struct xt_hashlimit_info),
#ifdef CONFIG_COMPAT #ifdef CONFIG_COMPAT
.compatsize = sizeof(struct compat_ipt_hashlimit_info), .compatsize = sizeof(struct compat_xt_hashlimit_info),
.compat_from_user = compat_from_user, .compat_from_user = compat_from_user,
.compat_to_user = compat_to_user, .compat_to_user = compat_to_user,
#endif #endif
.checkentry = hashlimit_checkentry, .checkentry = hashlimit_checkentry,
.destroy = hashlimit_destroy, .destroy = hashlimit_destroy,
.me = THIS_MODULE .me = THIS_MODULE
},
{
.name = "hashlimit",
.family = AF_INET6,
.match = hashlimit_match,
.matchsize = sizeof(struct xt_hashlimit_info),
#ifdef CONFIG_COMPAT
.compatsize = sizeof(struct compat_xt_hashlimit_info),
.compat_from_user = compat_from_user,
.compat_to_user = compat_to_user,
#endif
.checkentry = hashlimit_checkentry,
.destroy = hashlimit_destroy,
.me = THIS_MODULE
},
}; };
/* PROC stuff */ /* PROC stuff */
static void *dl_seq_start(struct seq_file *s, loff_t *pos) static void *dl_seq_start(struct seq_file *s, loff_t *pos)
{ {
struct proc_dir_entry *pde = s->private; struct proc_dir_entry *pde = s->private;
struct ipt_hashlimit_htable *htable = pde->data; struct xt_hashlimit_htable *htable = pde->data;
unsigned int *bucket; unsigned int *bucket;
spin_lock_bh(&htable->lock); spin_lock_bh(&htable->lock);
@ -598,7 +624,7 @@ static void *dl_seq_start(struct seq_file *s, loff_t *pos)
static void *dl_seq_next(struct seq_file *s, void *v, loff_t *pos) static void *dl_seq_next(struct seq_file *s, void *v, loff_t *pos)
{ {
struct proc_dir_entry *pde = s->private; struct proc_dir_entry *pde = s->private;
struct ipt_hashlimit_htable *htable = pde->data; struct xt_hashlimit_htable *htable = pde->data;
unsigned int *bucket = (unsigned int *)v; unsigned int *bucket = (unsigned int *)v;
*pos = ++(*bucket); *pos = ++(*bucket);
@ -612,43 +638,59 @@ static void *dl_seq_next(struct seq_file *s, void *v, loff_t *pos)
static void dl_seq_stop(struct seq_file *s, void *v) static void dl_seq_stop(struct seq_file *s, void *v)
{ {
struct proc_dir_entry *pde = s->private; struct proc_dir_entry *pde = s->private;
struct ipt_hashlimit_htable *htable = pde->data; struct xt_hashlimit_htable *htable = pde->data;
unsigned int *bucket = (unsigned int *)v; unsigned int *bucket = (unsigned int *)v;
kfree(bucket); kfree(bucket);
spin_unlock_bh(&htable->lock); spin_unlock_bh(&htable->lock);
} }
static inline int dl_seq_real_show(struct dsthash_ent *ent, struct seq_file *s) static int dl_seq_real_show(struct dsthash_ent *ent, int family,
struct seq_file *s)
{ {
/* recalculate to show accurate numbers */ /* recalculate to show accurate numbers */
rateinfo_recalc(ent, jiffies); rateinfo_recalc(ent, jiffies);
return seq_printf(s, "%ld %u.%u.%u.%u:%u->%u.%u.%u.%u:%u %u %u %u\n", switch (family) {
case AF_INET:
return seq_printf(s, "%ld %u.%u.%u.%u:%u->"
"%u.%u.%u.%u:%u %u %u %u\n",
(long)(ent->expires - jiffies)/HZ, (long)(ent->expires - jiffies)/HZ,
NIPQUAD(ent->dst.src_ip), ntohs(ent->dst.src_port), NIPQUAD(ent->dst.addr.ip.src),
NIPQUAD(ent->dst.dst_ip), ntohs(ent->dst.dst_port), ntohs(ent->dst.src_port),
NIPQUAD(ent->dst.addr.ip.dst),
ntohs(ent->dst.dst_port),
ent->rateinfo.credit, ent->rateinfo.credit_cap, ent->rateinfo.credit, ent->rateinfo.credit_cap,
ent->rateinfo.cost); ent->rateinfo.cost);
case AF_INET6:
return seq_printf(s, "%ld " NIP6_FMT ":%u->"
NIP6_FMT ":%u %u %u %u\n",
(long)(ent->expires - jiffies)/HZ,
NIP6(*(struct in6_addr *)&ent->dst.addr.ip6.src),
ntohs(ent->dst.src_port),
NIP6(*(struct in6_addr *)&ent->dst.addr.ip6.dst),
ntohs(ent->dst.dst_port),
ent->rateinfo.credit, ent->rateinfo.credit_cap,
ent->rateinfo.cost);
default:
BUG();
return 0;
}
} }
static int dl_seq_show(struct seq_file *s, void *v) static int dl_seq_show(struct seq_file *s, void *v)
{ {
struct proc_dir_entry *pde = s->private; struct proc_dir_entry *pde = s->private;
struct ipt_hashlimit_htable *htable = pde->data; struct xt_hashlimit_htable *htable = pde->data;
unsigned int *bucket = (unsigned int *)v; unsigned int *bucket = (unsigned int *)v;
struct dsthash_ent *ent; struct dsthash_ent *ent;
struct hlist_node *pos; struct hlist_node *pos;
if (!hlist_empty(&htable->hash[*bucket])) if (!hlist_empty(&htable->hash[*bucket])) {
hlist_for_each_entry(ent, pos, &htable->hash[*bucket], node) { hlist_for_each_entry(ent, pos, &htable->hash[*bucket], node)
if (dl_seq_real_show(ent, s)) { if (dl_seq_real_show(ent, htable->family, s))
/* buffer was filled and unable to print that tuple */
return 1; return 1;
} }
}
return 0; return 0;
} }
@ -678,56 +720,53 @@ static struct file_operations dl_file_ops = {
.release = seq_release .release = seq_release
}; };
static int init_or_fini(int fini) static int __init xt_hashlimit_init(void)
{ {
int ret = 0; int err;
if (fini) err = xt_register_matches(xt_hashlimit, ARRAY_SIZE(xt_hashlimit));
goto cleanup; if (err < 0)
goto err1;
if (ipt_register_match(&ipt_hashlimit)) { err = -ENOMEM;
ret = -EINVAL; hashlimit_cachep = kmem_cache_create("xt_hashlimit",
goto cleanup_nothing; sizeof(struct dsthash_ent), 0, 0,
} NULL, NULL);
hashlimit_cachep = kmem_cache_create("ipt_hashlimit",
sizeof(struct dsthash_ent), 0,
0, NULL, NULL);
if (!hashlimit_cachep) { if (!hashlimit_cachep) {
printk(KERN_ERR "Unable to create ipt_hashlimit slab cache\n"); printk(KERN_ERR "xt_hashlimit: unable to create slab cache\n");
ret = -ENOMEM; goto err2;
goto cleanup_unreg_match;
} }
hashlimit_procdir4 = proc_mkdir("ipt_hashlimit", proc_net);
hashlimit_procdir = proc_mkdir("ipt_hashlimit", proc_net); if (!hashlimit_procdir4) {
if (!hashlimit_procdir) { printk(KERN_ERR "xt_hashlimit: unable to create proc dir "
printk(KERN_ERR "Unable to create proc dir entry\n"); "entry\n");
ret = -ENOMEM; goto err3;
goto cleanup_free_slab;
} }
hashlimit_procdir6 = proc_mkdir("ip6t_hashlimit", proc_net);
return ret; if (!hashlimit_procdir6) {
printk(KERN_ERR "xt_hashlimit: tnable to create proc dir "
cleanup: "entry\n");
goto err4;
}
return 0;
err4:
remove_proc_entry("ipt_hashlimit", proc_net); remove_proc_entry("ipt_hashlimit", proc_net);
cleanup_free_slab: err3:
kmem_cache_destroy(hashlimit_cachep); kmem_cache_destroy(hashlimit_cachep);
cleanup_unreg_match: err2:
ipt_unregister_match(&ipt_hashlimit); xt_unregister_matches(xt_hashlimit, ARRAY_SIZE(xt_hashlimit));
cleanup_nothing: err1:
return ret; return err;
} }
static int __init ipt_hashlimit_init(void) static void __exit xt_hashlimit_fini(void)
{ {
return init_or_fini(0); remove_proc_entry("ipt_hashlimit", proc_net);
remove_proc_entry("ip6t_hashlimit", proc_net);
kmem_cache_destroy(hashlimit_cachep);
xt_unregister_matches(xt_hashlimit, ARRAY_SIZE(xt_hashlimit));
} }
static void __exit ipt_hashlimit_fini(void) module_init(xt_hashlimit_init);
{ module_exit(xt_hashlimit_fini);
init_or_fini(1);
}
module_init(ipt_hashlimit_init);
module_exit(ipt_hashlimit_fini);