From: Jozsef Kadlecsik <kadlec@xxxxxxxxxxxxxxxxx>
Three types of data need to be protected in the case of the hash types:
a. The hash buckets: standard rcu pointer operations are used.
b. The element blobs in the hash buckets are stored in an array and
a bitmap is used for book-keeping to tell which elements in the array
are used or free.
c. Networks per cidr values and the cidr values themselves are stored
in fix sized arrays and need no protection. The values are modified
in such an order that in the worst case an element testing is repeated
once with the same cidr value.
The ipset hash approach uses arrays instead of lists and therefore is
incompatible with rhashtable.
Performance is tested by Jesper Dangaard Brouer:
Simple drop in FORWARD
~~~~~~~~~~~~~~~~~~~~~~
Dropping via simple iptables net-mask match::
iptables -t raw -N simple || iptables -t raw -F simple
iptables -t raw -I simple -s 198.18.0.0/15 -j DROP
iptables -t raw -D PREROUTING -j simple
iptables -t raw -I PREROUTING -j simple
Drop performance in "raw": 11.3Mpps
Generator: sending 12.2Mpps (tx:12264083 pps)
Drop via original ipset in RAW table
~~~~~~~~~~~~~~~~~~~~~~~~~~~
Create a set with lots of elements::
sudo ./ipset destroy test
echo "create test hash:ip hashsize 65536" > test.set
for x in `seq 0 255`; do
for y in `seq 0 255`; do
echo "add test 198.18.$x.$y" >> test.set
done
done
sudo ./ipset restore < test.set
Dropping via ipset::
iptables -t raw -F
iptables -t raw -N net198 || iptables -t raw -F net198
iptables -t raw -I net198 -m set --match-set test src -j DROP
iptables -t raw -I PREROUTING -j net198
Drop performance in "raw" with ipset: 8Mpps
Perf report numbers ipset drop in "raw"::
+ 24.65% ksoftirqd/1 [ip_set] [k] ip_set_test
- 21.42% ksoftirqd/1 [kernel.kallsyms] [k] _raw_read_lock_bh
- _raw_read_lock_bh
+ 99.88% ip_set_test
- 19.42% ksoftirqd/1 [kernel.kallsyms] [k] _raw_read_unlock_bh
- _raw_read_unlock_bh
+ 99.72% ip_set_test
+ 4.31% ksoftirqd/1 [ip_set_hash_ip] [k] hash_ip4_kadt
+ 2.27% ksoftirqd/1 [ixgbe] [k] ixgbe_fetch_rx_buffer
+ 2.18% ksoftirqd/1 [ip_tables] [k] ipt_do_table
+ 1.81% ksoftirqd/1 [ip_set_hash_ip] [k] hash_ip4_test
+ 1.61% ksoftirqd/1 [kernel.kallsyms] [k] __netif_receive_skb_core
+ 1.44% ksoftirqd/1 [kernel.kallsyms] [k] build_skb
+ 1.42% ksoftirqd/1 [kernel.kallsyms] [k] ip_rcv
+ 1.36% ksoftirqd/1 [kernel.kallsyms] [k] __local_bh_enable_ip
+ 1.16% ksoftirqd/1 [kernel.kallsyms] [k] dev_gro_receive
+ 1.09% ksoftirqd/1 [kernel.kallsyms] [k] __rcu_read_unlock
+ 0.96% ksoftirqd/1 [ixgbe] [k] ixgbe_clean_rx_irq
+ 0.95% ksoftirqd/1 [kernel.kallsyms] [k] __netdev_alloc_frag
+ 0.88% ksoftirqd/1 [kernel.kallsyms] [k] kmem_cache_alloc
+ 0.87% ksoftirqd/1 [xt_set] [k] set_match_v3
+ 0.85% ksoftirqd/1 [kernel.kallsyms] [k] inet_gro_receive
+ 0.83% ksoftirqd/1 [kernel.kallsyms] [k] nf_iterate
+ 0.76% ksoftirqd/1 [kernel.kallsyms] [k] put_compound_page
+ 0.75% ksoftirqd/1 [kernel.kallsyms] [k] __rcu_read_lock
Drop via ipset in RAW table with RCU-locking
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
With RCU locking, the RW-lock is gone.
Drop performance in "raw" with ipset with RCU-locking: 11.3Mpps
Performance-tested-by: Jesper Dangaard Brouer <brouer@xxxxxxxxxx>
Signed-off-by: Jozsef Kadlecsik <kadlec@xxxxxxxxxxxxxxxxx>
---
net/netfilter/ipset/ip_set_hash_gen.h | 589 +++++++++++++++-----------
net/netfilter/ipset/ip_set_hash_ip.c | 1 +
net/netfilter/ipset/ip_set_hash_ipmark.c | 1 +
net/netfilter/ipset/ip_set_hash_ipport.c | 1 +
net/netfilter/ipset/ip_set_hash_ipportip.c | 1 +
net/netfilter/ipset/ip_set_hash_ipportnet.c | 1 +
net/netfilter/ipset/ip_set_hash_mac.c | 1 +
net/netfilter/ipset/ip_set_hash_net.c | 1 +
net/netfilter/ipset/ip_set_hash_netiface.c | 1 +
net/netfilter/ipset/ip_set_hash_netnet.c | 1 +
net/netfilter/ipset/ip_set_hash_netport.c | 1 +
net/netfilter/ipset/ip_set_hash_netportnet.c | 1 +
12 files changed, 364 insertions(+), 236 deletions(-)
diff --git a/net/netfilter/ipset/ip_set_hash_gen.h b/net/netfilter/ipset/ip_set_hash_gen.h
index 5fcf70b..f352cc0 100644
--- a/net/netfilter/ipset/ip_set_hash_gen.h
+++ b/net/netfilter/ipset/ip_set_hash_gen.h
@@ -10,19 +10,19 @@
#include <linux/rcupdate.h>
#include <linux/jhash.h>
+#include <linux/types.h>
#include <linux/netfilter/ipset/ip_set_timeout.h>
-#ifndef rcu_dereference_bh
-#define rcu_dereference_bh(p) rcu_dereference(p)
-#endif
+
+#define __ipset_dereference_protected(p, c) rcu_dereference_protected(p, c)
+#define ipset_dereference_protected(p, set) \
+ __ipset_dereference_protected(p, spin_is_locked(&(set)->lock))
#define rcu_dereference_bh_nfnl(p) rcu_dereference_bh_check(p, 1)
/* Hashing which uses arrays to resolve clashing. The hash table is resized
* (doubled) when searching becomes too long.
* Internally jhash is used with the assumption that the size of the
- * stored data is a multiple of sizeof(u32). If storage supports timeout,
- * the timeout field must be the last one in the data structure - that field
- * is ignored when computing the hash key.
+ * stored data is a multiple of sizeof(u32).
*
* Readers and resizing
*
@@ -36,6 +36,8 @@
#define AHASH_INIT_SIZE 4
/* Max number of elements to store in an array block */
#define AHASH_MAX_SIZE (3*AHASH_INIT_SIZE)
+/* Max muber of elements in the array block when tuned */
+#define AHASH_MAX_TUNED 64
/* Max number of elements can be tuned */
#ifdef IP_SET_HASH_WITH_MULTI
@@ -53,7 +55,7 @@ tune_ahash_max(u8 curr, u32 multi)
/* Currently, at listing one hash bucket must fit into a message.
* Therefore we have a hard limit here.
*/
- return n > curr && n <= 64 ? n : curr;
+ return n > curr && n <= AHASH_MAX_TUNED ? n : curr;
}
#define TUNE_AHASH_MAX(h, multi) \
((h)->ahash_max = tune_ahash_max((h)->ahash_max, multi))
@@ -64,20 +66,23 @@ tune_ahash_max(u8 curr, u32 multi)
/* A hash bucket */
struct hbucket {
- void *value; /* the array of the values */
+ struct rcu_head rcu; /* for call_rcu_bh */
+ /* Which positions are used in the array */
+ DECLARE_BITMAP(used, AHASH_MAX_TUNED);
u8 size; /* size of the array */
u8 pos; /* position of the first free entry */
-};
+ unsigned char value[0]; /* the array of the values */
+} __attribute__ ((aligned));
/* The hash table: the table size stored here in order to make resizing easy */
struct htable {
atomic_t ref; /* References for resizing */
atomic_t uref; /* References for dumping */
u8 htable_bits; /* size of hash table == 2^htable_bits */
- struct hbucket bucket[0]; /* hashtable buckets */
+ struct hbucket __rcu *bucket[0]; /* hashtable buckets */
};
-#define hbucket(h, i) (&((h)->bucket[i]))
+#define hbucket(h, i) ((h)->bucket[i])
#ifndef IPSET_NET_COUNT
#define IPSET_NET_COUNT 1
@@ -85,8 +90,8 @@ struct htable {
/* Book-keeping of the prefixes added to the set */
struct net_prefixes {
- u32 nets[IPSET_NET_COUNT]; /* number of elements per cidr */
- u8 cidr[IPSET_NET_COUNT]; /* the different cidr values in the set */
+ u32 nets[IPSET_NET_COUNT]; /* number of elements for this cidr */
+ u8 cidr[IPSET_NET_COUNT]; /* the cidr value */
};
/* Compute the hash table size */
@@ -99,11 +104,11 @@ htable_size(u8 hbits)
if (hbits > 31)
return 0;
hsize = jhash_size(hbits);
- if ((((size_t)-1) - sizeof(struct htable))/sizeof(struct hbucket)
+ if ((((size_t)-1) - sizeof(struct htable)) / sizeof(struct hbucket *)
< hsize)
return 0;
- return hsize * sizeof(struct hbucket) + sizeof(struct htable);
+ return hsize * sizeof(struct hbucket *) + sizeof(struct htable);
}
/* Compute htable_bits from the user input parameter hashsize */
@@ -112,6 +117,7 @@ htable_bits(u32 hashsize)
{
/* Assume that hashsize == 2^htable_bits */
u8 bits = fls(hashsize - 1);
+
if (jhash_size(bits) != hashsize)
/* Round up to the first 2^n value */
bits = fls(hashsize);
@@ -119,30 +125,6 @@ htable_bits(u32 hashsize)
return bits;
}
-static int
-hbucket_elem_add(struct hbucket *n, u8 ahash_max, size_t dsize)
-{
- if (n->pos >= n->size) {
- void *tmp;
-
- if (n->size >= ahash_max)
- /* Trigger rehashing */
- return -EAGAIN;
-
- tmp = kzalloc((n->size + AHASH_INIT_SIZE) * dsize,
- GFP_ATOMIC);
- if (!tmp)
- return -ENOMEM;
- if (n->size) {
- memcpy(tmp, n->value, n->size * dsize);
- kfree(n->value);
- }
- n->value = tmp;
- n->size += AHASH_INIT_SIZE;
- }
- return 0;
-}
-
#ifdef IP_SET_HASH_WITH_NETS
#if IPSET_NET_COUNT > 1
#define __CIDR(cidr, i) (cidr[i])
@@ -300,9 +282,6 @@ struct htype {
#ifdef IP_SET_HASH_WITH_NETMASK
u8 netmask; /* netmask value for subnets to store */
#endif
-#ifdef IP_SET_HASH_WITH_RBTREE
- struct rb_root rbtree;
-#endif
#ifdef IP_SET_HASH_WITH_NETS
struct net_prefixes nets[0]; /* book-keeping of prefixes */
#endif
@@ -345,8 +324,8 @@ mtype_del_cidr(struct htype *h, u8 cidr, u8 nets_length, u8 n)
for (i = 0; i < nets_length; i++) {
if (h->nets[i].cidr[n] != cidr)
continue;
- h->nets[cidr -1].nets[n]--;
- if (h->nets[cidr -1].nets[n] > 0)
+ h->nets[cidr - 1].nets[n]--;
+ if (h->nets[cidr - 1].nets[n] > 0)
return;
for (j = i; j < net_end && h->nets[j].cidr[n]; j++)
h->nets[j].cidr[n] = h->nets[j + 1].cidr[n];
@@ -362,15 +341,18 @@ mtype_ahash_memsize(const struct htype *h, const struct htable *t,
u8 nets_length, size_t dsize)
{
u32 i;
- size_t memsize = sizeof(*h)
- + sizeof(*t)
+ struct hbucket *n;
+ size_t memsize = sizeof(*h) + sizeof(*t);
+
#ifdef IP_SET_HASH_WITH_NETS
- + sizeof(struct net_prefixes) * nets_length
+ memsize += sizeof(struct net_prefixes) * nets_length;
#endif
- + jhash_size(t->htable_bits) * sizeof(struct hbucket);
-
- for (i = 0; i < jhash_size(t->htable_bits); i++)
- memsize += t->bucket[i].size * dsize;
+ for (i = 0; i < jhash_size(t->htable_bits); i++) {
+ n = rcu_dereference_bh(hbucket(t, i));
+ if (!n)
+ continue;
+ memsize += sizeof(struct hbucket) + n->size * dsize;
+ }
return memsize;
}
@@ -385,7 +367,8 @@ mtype_ext_cleanup(struct ip_set *set, struct hbucket *n)
int i;
for (i = 0; i < n->pos; i++)
- ip_set_ext_destroy(set, ahash_data(n, i, set->dsize));
+ if (test_bit(i, n->used))
+ ip_set_ext_destroy(set, ahash_data(n, i, set->dsize));
}
/* Flush a hash type of set: destroy all elements */
@@ -397,16 +380,16 @@ mtype_flush(struct ip_set *set)
struct hbucket *n;
u32 i;
- t = rcu_dereference_bh_nfnl(h->table);
+ t = ipset_dereference_protected(h->table, set);
for (i = 0; i < jhash_size(t->htable_bits); i++) {
- n = hbucket(t, i);
- if (n->size) {
- if (set->extensions & IPSET_EXT_DESTROY)
- mtype_ext_cleanup(set, n);
- n->size = n->pos = 0;
- /* FIXME: use slab cache */
- kfree(n->value);
- }
+ n = __ipset_dereference_protected(hbucket(t, i), 1);
+ if (!n)
+ continue;
+ if (set->extensions & IPSET_EXT_DESTROY)
+ mtype_ext_cleanup(set, n);
+ /* FIXME: use slab cache */
+ rcu_assign_pointer(hbucket(t, i), NULL);
+ kfree_rcu(n, rcu);
}
#ifdef IP_SET_HASH_WITH_NETS
memset(h->nets, 0, sizeof(struct net_prefixes) * NLEN(set->family));
@@ -422,13 +405,13 @@ mtype_ahash_destroy(struct ip_set *set, struct htable *t, bool ext_destroy)
u32 i;
for (i = 0; i < jhash_size(t->htable_bits); i++) {
- n = hbucket(t, i);
- if (n->size) {
- if (set->extensions & IPSET_EXT_DESTROY && ext_destroy)
- mtype_ext_cleanup(set, n);
- /* FIXME: use slab cache */
- kfree(n->value);
- }
+ n = __ipset_dereference_protected(hbucket(t, i), 1);
+ if (!n)
+ continue;
+ if (set->extensions & IPSET_EXT_DESTROY && ext_destroy)
+ mtype_ext_cleanup(set, n);
+ /* FIXME: use slab cache */
+ kfree(n);
}
ip_set_free(t);
@@ -443,10 +426,8 @@ mtype_destroy(struct ip_set *set)
if (SET_WITH_TIMEOUT(set))
del_timer_sync(&h->gc);
- mtype_ahash_destroy(set, rcu_dereference_bh_nfnl(h->table), true);
-#ifdef IP_SET_HASH_WITH_RBTREE
- rbtree_destroy(&h->rbtree);
-#endif
+ mtype_ahash_destroy(set, __ipset_dereference_protected(h->table, 1),
+ true);
kfree(h);
set->data = NULL;
@@ -491,20 +472,26 @@ mtype_expire(struct ip_set *set, struct htype *h, u8 nets_length, size_t dsize)
struct htable *t;
struct hbucket *n;
struct mtype_elem *data;
- u32 i;
- int j;
+ u32 i, j, d;
#ifdef IP_SET_HASH_WITH_NETS
u8 k;
#endif
- rcu_read_lock_bh();
- t = rcu_dereference_bh(h->table);
+ t = ipset_dereference_protected(h->table, set);
for (i = 0; i < jhash_size(t->htable_bits); i++) {
- n = hbucket(t, i);
- for (j = 0; j < n->pos; j++) {
+ n = __ipset_dereference_protected(hbucket(t, i), 1);
+ if (!n)
+ continue;
+ for (j = 0, d = 0; j < n->pos; j++) {
+ if (!test_bit(j, n->used)) {
+ d++;
+ continue;
+ }
data = ahash_data(n, j, dsize);
if (ip_set_timeout_expired(ext_timeout(data, set))) {
pr_debug("expired %u/%u\n", i, j);
+ clear_bit(j, n->used);
+ smp_mb__after_atomic();
#ifdef IP_SET_HASH_WITH_NETS
for (k = 0; k < IPSET_NET_COUNT; k++)
mtype_del_cidr(h,
@@ -513,29 +500,31 @@ mtype_expire(struct ip_set *set, struct htype *h, u8 nets_length, size_t dsize)
nets_length, k);
#endif
ip_set_ext_destroy(set, data);
- if (j != n->pos - 1)
- /* Not last one */
- memcpy(data,
- ahash_data(n, n->pos - 1, dsize),
- dsize);
- n->pos--;
h->elements--;
+ d++;
}
}
- if (n->pos + AHASH_INIT_SIZE < n->size) {
- void *tmp = kzalloc((n->size - AHASH_INIT_SIZE)
- * dsize,
- GFP_ATOMIC);
+ if (d >= AHASH_INIT_SIZE) {
+ struct hbucket *tmp = kzalloc(sizeof(*tmp) +
+ (n->size - AHASH_INIT_SIZE) * dsize,
+ GFP_ATOMIC);
if (!tmp)
/* Still try to delete expired elements */
continue;
- n->size -= AHASH_INIT_SIZE;
- memcpy(tmp, n->value, n->size * dsize);
- kfree(n->value);
- n->value = tmp;
+ tmp->size = n->size - AHASH_INIT_SIZE;
+ for (j = 0, d = 0; j < n->pos; j++) {
+ if (!test_bit(j, n->used))
+ continue;
+ data = ahash_data(n, j, dsize);
+ memcpy(tmp->value + d * dsize, data, dsize);
+ set_bit(j, tmp->used);
+ d++;
+ }
+ tmp->pos = d;
+ rcu_assign_pointer(hbucket(t, i), tmp);
+ kfree_rcu(n, rcu);
}
}
- rcu_read_unlock_bh();
}
static void
@@ -545,9 +534,9 @@ mtype_gc(unsigned long ul_set)
struct htype *h = set->data;
pr_debug("called\n");
- write_lock_bh(&set->lock);
+ spin_lock_bh(&set->lock);
mtype_expire(set, h, NLEN(set->family), set->dsize);
- write_unlock_bh(&set->lock);
+ spin_unlock_bh(&set->lock);
h->gc.expires = jiffies + IPSET_GC_PERIOD(set->timeout) * HZ;
add_timer(&h->gc);
@@ -560,80 +549,115 @@ static int
mtype_resize(struct ip_set *set, bool retried)
{
struct htype *h = set->data;
- struct htable *t, *orig = rcu_dereference_bh_nfnl(h->table);
- u8 htable_bits = orig->htable_bits;
+ struct htable *t, *orig;
+ u8 htable_bits;
+ size_t dsize = set->dsize;
#ifdef IP_SET_HASH_WITH_NETS
u8 flags;
+ struct mtype_elem *tmp;
#endif
struct mtype_elem *data;
struct mtype_elem *d;
struct hbucket *n, *m;
- u32 i, j;
+ u32 i, j, key;
int ret;
- /* Try to cleanup once */
- if (SET_WITH_TIMEOUT(set) && !retried) {
- i = h->elements;
- write_lock_bh(&set->lock);
- mtype_expire(set, set->data, NLEN(set->family), set->dsize);
- write_unlock_bh(&set->lock);
- if (h->elements < i)
- return 0;
- }
+#ifdef IP_SET_HASH_WITH_NETS
+ tmp = kmalloc(dsize, GFP_KERNEL);
+ if (!tmp)
+ return -ENOMEM;
+#endif
+ rcu_read_lock_bh();
+ orig = rcu_dereference_bh_nfnl(h->table);
+ htable_bits = orig->htable_bits;
+ rcu_read_unlock_bh();
retry:
ret = 0;
htable_bits++;
- pr_debug("attempt to resize set %s from %u to %u, t %p\n",
- set->name, orig->htable_bits, htable_bits, orig);
if (!htable_bits) {
/* In case we have plenty of memory :-) */
pr_warn("Cannot increase the hashsize of set %s further\n",
set->name);
- return -IPSET_ERR_HASH_FULL;
+ ret = -IPSET_ERR_HASH_FULL;
+ goto out;
+ }
+ t = ip_set_alloc(htable_size(htable_bits));
+ if (!t) {
+ ret = -ENOMEM;
+ goto out;
}
- t = ip_set_alloc(sizeof(*t)
- + jhash_size(htable_bits) * sizeof(struct hbucket));
- if (!t)
- return -ENOMEM;
t->htable_bits = htable_bits;
- read_lock_bh(&set->lock);
+ spin_lock_bh(&set->lock);
+ orig = __ipset_dereference_protected(h->table, 1);
/* There can't be another parallel resizing, but dumping is possible */
atomic_set(&orig->ref, 1);
atomic_inc(&orig->uref);
+ pr_debug("attempt to resize set %s from %u to %u, t %p\n",
+ set->name, orig->htable_bits, htable_bits, orig);
for (i = 0; i < jhash_size(orig->htable_bits); i++) {
- n = hbucket(orig, i);
+ n = __ipset_dereference_protected(hbucket(orig, i), 1);
+ if (!n)
+ continue;
for (j = 0; j < n->pos; j++) {
- data = ahash_data(n, j, set->dsize);
+ if (!test_bit(j, n->used))
+ continue;
+ data = ahash_data(n, j, dsize);
#ifdef IP_SET_HASH_WITH_NETS
+ /* We have readers running parallel with us,
+ * so the live data cannot be modified.
+ */
flags = 0;
+ memcpy(tmp, data, dsize);
+ data = tmp;
mtype_data_reset_flags(data, &flags);
#endif
- m = hbucket(t, HKEY(data, h->initval, htable_bits));
- ret = hbucket_elem_add(m, AHASH_MAX(h), set->dsize);
- if (ret < 0) {
-#ifdef IP_SET_HASH_WITH_NETS
- mtype_data_reset_flags(data, &flags);
-#endif
- atomic_set(&orig->ref, 0);
- atomic_dec(&orig->uref);
- read_unlock_bh(&set->lock);
- mtype_ahash_destroy(set, t, false);
- if (ret == -EAGAIN)
- goto retry;
- return ret;
+ key = HKEY(data, h->initval, htable_bits);
+ m = __ipset_dereference_protected(hbucket(t, key), 1);
+ if (!m) {
+ m = kzalloc(sizeof(*m) +
+ AHASH_INIT_SIZE * dsize,
+ GFP_ATOMIC);
+ if (!m) {
+ ret = -ENOMEM;
+ goto cleanup;
+ }
+ m->size = AHASH_INIT_SIZE;
+ RCU_INIT_POINTER(hbucket(t, key), m);
+ } else if (m->pos >= m->size) {
+ struct hbucket *ht;
+
+ if (m->size >= AHASH_MAX(h)) {
+ ret = -EAGAIN;
+ } else {
+ ht = kzalloc(sizeof(*ht) +
+ (m->size + AHASH_INIT_SIZE)
+ * dsize,
+ GFP_ATOMIC);
+ if (!ht)
+ ret = -ENOMEM;
+ }
+ if (ret < 0)
+ goto cleanup;
+ memcpy(ht, m, sizeof(struct hbucket) +
+ m->size * dsize);
+ ht->size = m->size + AHASH_INIT_SIZE;
+ kfree(m);
+ m = ht;
+ RCU_INIT_POINTER(hbucket(t, key), ht);
}
- d = ahash_data(m, m->pos++, set->dsize);
- memcpy(d, data, set->dsize);
+ d = ahash_data(m, m->pos, dsize);
+ memcpy(d, data, dsize);
+ set_bit(m->pos++, m->used);
#ifdef IP_SET_HASH_WITH_NETS
mtype_data_reset_flags(d, &flags);
#endif
}
}
-
rcu_assign_pointer(h->table, t);
- read_unlock_bh(&set->lock);
+
+ spin_unlock_bh(&set->lock);
/* Give time to other readers of the set */
synchronize_rcu_bh();
@@ -646,7 +670,20 @@ retry:
mtype_ahash_destroy(set, orig, false);
}
- return 0;
+out:
+#ifdef IP_SET_HASH_WITH_NETS
+ kfree(tmp);
+#endif
+ return ret;
+
+cleanup:
+ atomic_set(&orig->ref, 0);
+ atomic_dec(&orig->uref);
+ spin_unlock_bh(&set->lock);
+ mtype_ahash_destroy(set, t, false);
+ if (ret == -EAGAIN)
+ goto retry;
+ goto out;
}
/* Add an element to a hash and update the internal counters when succeeded,
@@ -659,17 +696,49 @@ mtype_add(struct ip_set *set, void *value, const struct ip_set_ext *ext,
struct htable *t;
const struct mtype_elem *d = value;
struct mtype_elem *data;
- struct hbucket *n;
- int i, ret = 0;
- int j = AHASH_MAX(h) + 1;
+ struct hbucket *n, *old = ERR_PTR(-ENOENT);
+ int i, j = -1;
bool flag_exist = flags & IPSET_FLAG_EXIST;
+ bool deleted = false, forceadd = false, reuse = false;
u32 key, multi = 0;
- rcu_read_lock_bh();
- t = rcu_dereference_bh(h->table);
+ if (h->elements >= h->maxelem) {
+ if (SET_WITH_TIMEOUT(set))
+ /* FIXME: when set is full, we slow down here */
+ mtype_expire(set, h, NLEN(set->family), set->dsize);
+ if (h->elements >= h->maxelem && SET_WITH_FORCEADD(set))
+ forceadd = true;
+ }
+
+ t = ipset_dereference_protected(h->table, set);
key = HKEY(value, h->initval, t->htable_bits);
- n = hbucket(t, key);
+ n = __ipset_dereference_protected(hbucket(t, key), 1);
+ if (!n) {
+ if (forceadd) {
+ if (net_ratelimit())
+ pr_warn("Set %s is full, maxelem %u reached\n",
+ set->name, h->maxelem);
+ return -IPSET_ERR_HASH_FULL;
+ } else if (h->elements >= h->maxelem) {
+ goto set_full;
+ }
+ old = NULL;
+ n = kzalloc(sizeof(*n) + AHASH_INIT_SIZE * set->dsize,
+ GFP_ATOMIC);
+ if (!n)
+ return -ENOMEM;
+ n->size = AHASH_INIT_SIZE;
+ goto copy_elem;
+ }
for (i = 0; i < n->pos; i++) {
+ if (!test_bit(i, n->used)) {
+ /* Reuse first deleted entry */
+ if (j == -1) {
+ deleted = reuse = true;
+ j = i;
+ }
+ continue;
+ }
data = ahash_data(n, i, set->dsize);
if (mtype_data_equal(data, d, &multi)) {
if (flag_exist ||
@@ -677,85 +746,94 @@ mtype_add(struct ip_set *set, void *value, const struct ip_set_ext *ext,
ip_set_timeout_expired(ext_timeout(data, set)))) {
/* Just the extensions could be overwritten */
j = i;
- goto reuse_slot;
- } else {
- ret = -IPSET_ERR_EXIST;
- goto out;
+ goto overwrite_extensions;
}
+ return -IPSET_ERR_EXIST;
}
/* Reuse first timed out entry */
if (SET_WITH_TIMEOUT(set) &&
ip_set_timeout_expired(ext_timeout(data, set)) &&
- j != AHASH_MAX(h) + 1)
+ j == -1) {
j = i;
+ reuse = true;
+ }
}
- if (h->elements >= h->maxelem && SET_WITH_FORCEADD(set) && n->pos) {
- /* Choosing the first entry in the array to replace */
- j = 0;
- goto reuse_slot;
- }
- if (SET_WITH_TIMEOUT(set) && h->elements >= h->maxelem)
- /* FIXME: when set is full, we slow down here */
- mtype_expire(set, h, NLEN(set->family), set->dsize);
-
- if (h->elements >= h->maxelem) {
- if (net_ratelimit())
- pr_warn("Set %s is full, maxelem %u reached\n",
- set->name, h->maxelem);
- ret = -IPSET_ERR_HASH_FULL;
- goto out;
- }
-
-reuse_slot:
- if (j != AHASH_MAX(h) + 1) {
- /* Fill out reused slot */
+ if (reuse || forceadd) {
data = ahash_data(n, j, set->dsize);
+ if (!deleted) {
#ifdef IP_SET_HASH_WITH_NETS
- for (i = 0; i < IPSET_NET_COUNT; i++) {
- mtype_del_cidr(h, NCIDR_PUT(DCIDR_GET(data->cidr, i)),
- NLEN(set->family), i);
- mtype_add_cidr(h, NCIDR_PUT(DCIDR_GET(d->cidr, i)),
- NLEN(set->family), i);
- }
+ for (i = 0; i < IPSET_NET_COUNT; i++)
+ mtype_del_cidr(h,
+ NCIDR_PUT(DCIDR_GET(data->cidr, i)),
+ NLEN(set->family), i);
#endif
- ip_set_ext_destroy(set, data);
- } else {
- /* Use/create a new slot */
+ ip_set_ext_destroy(set, data);
+ h->elements--;
+ }
+ goto copy_data;
+ }
+ if (h->elements >= h->maxelem)
+ goto set_full;
+ /* Create a new slot */
+ if (n->pos >= n->size) {
TUNE_AHASH_MAX(h, multi);
- ret = hbucket_elem_add(n, AHASH_MAX(h), set->dsize);
- if (ret != 0) {
- if (ret == -EAGAIN)
- mtype_data_next(&h->next, d);
- goto out;
+ if (n->size >= AHASH_MAX(h)) {
+ /* Trigger rehashing */
+ mtype_data_next(&h->next, d);
+ return -EAGAIN;
}
- data = ahash_data(n, n->pos++, set->dsize);
+ old = n;
+ n = kzalloc(sizeof(*n) +
+ (old->size + AHASH_INIT_SIZE) * set->dsize,
+ GFP_ATOMIC);
+ if (!n)
+ return -ENOMEM;
+ memcpy(n, old, sizeof(struct hbucket) +
+ old->size * set->dsize);
+ n->size = old->size + AHASH_INIT_SIZE;
+ }
+
+copy_elem:
+ j = n->pos++;
+ data = ahash_data(n, j, set->dsize);
+copy_data:
+ h->elements++;
#ifdef IP_SET_HASH_WITH_NETS
- for (i = 0; i < IPSET_NET_COUNT; i++)
- mtype_add_cidr(h, NCIDR_PUT(DCIDR_GET(d->cidr, i)),
- NLEN(set->family), i);
+ for (i = 0; i < IPSET_NET_COUNT; i++)
+ mtype_add_cidr(h, NCIDR_PUT(DCIDR_GET(d->cidr, i)),
+ NLEN(set->family), i);
#endif
- h->elements++;
- }
memcpy(data, d, sizeof(struct mtype_elem));
+overwrite_extensions:
#ifdef IP_SET_HASH_WITH_NETS
mtype_data_set_flags(data, flags);
#endif
- if (SET_WITH_TIMEOUT(set))
- ip_set_timeout_set(ext_timeout(data, set), ext->timeout);
if (SET_WITH_COUNTER(set))
ip_set_init_counter(ext_counter(data, set), ext);
if (SET_WITH_COMMENT(set))
ip_set_init_comment(ext_comment(data, set), ext);
if (SET_WITH_SKBINFO(set))
ip_set_init_skbinfo(ext_skbinfo(data, set), ext);
+ /* Must come last for the case when timed out entry is reused */
+ if (SET_WITH_TIMEOUT(set))
+ ip_set_timeout_set(ext_timeout(data, set), ext->timeout);
+ smp_mb__before_atomic();
+ set_bit(j, n->used);
+ if (old != ERR_PTR(-ENOENT)) {
+ rcu_assign_pointer(hbucket(t, key), n);
+ if (old)
+ kfree_rcu(old, rcu);
+ }
-out:
- rcu_read_unlock_bh();
- return ret;
+ return 0;
+set_full:
+ if (net_ratelimit())
+ pr_warn("Set %s is full, maxelem %u reached\n",
+ set->name, h->maxelem);
+ return -IPSET_ERR_HASH_FULL;
}
-/* Delete an element from the hash: swap it with the last element
- * and free up space if possible.
+/* Delete an element from the hash and free up space if possible.
*/
static int
mtype_del(struct ip_set *set, void *value, const struct ip_set_ext *ext,
@@ -766,29 +844,32 @@ mtype_del(struct ip_set *set, void *value, const struct ip_set_ext *ext,
const struct mtype_elem *d = value;
struct mtype_elem *data;
struct hbucket *n;
- int i, ret = -IPSET_ERR_EXIST;
-#ifdef IP_SET_HASH_WITH_NETS
- u8 j;
-#endif
+ int i, j, k, ret = -IPSET_ERR_EXIST;
u32 key, multi = 0;
+ size_t dsize = set->dsize;
- rcu_read_lock_bh();
- t = rcu_dereference_bh(h->table);
+ t = ipset_dereference_protected(h->table, set);
key = HKEY(value, h->initval, t->htable_bits);
- n = hbucket(t, key);
- for (i = 0; i < n->pos; i++) {
- data = ahash_data(n, i, set->dsize);
+ n = __ipset_dereference_protected(hbucket(t, key), 1);
+ if (!n)
+ goto out;
+ for (i = 0, k = 0; i < n->pos; i++) {
+ if (!test_bit(i, n->used)) {
+ k++;
+ continue;
+ }
+ data = ahash_data(n, i, dsize);
if (!mtype_data_equal(data, d, &multi))
continue;
if (SET_WITH_TIMEOUT(set) &&
ip_set_timeout_expired(ext_timeout(data, set)))
goto out;
- if (i != n->pos - 1)
- /* Not last one */
- memcpy(data, ahash_data(n, n->pos - 1, set->dsize),
- set->dsize);
- n->pos--;
+ ret = 0;
+ clear_bit(i, n->used);
+ smp_mb__after_atomic();
+ if (i + 1 == n->pos)
+ n->pos--;
h->elements--;
#ifdef IP_SET_HASH_WITH_NETS
for (j = 0; j < IPSET_NET_COUNT; j++)
@@ -796,25 +877,37 @@ mtype_del(struct ip_set *set, void *value, const struct ip_set_ext *ext,
NLEN(set->family), j);
#endif
ip_set_ext_destroy(set, data);
- if (n->pos + AHASH_INIT_SIZE < n->size) {
- void *tmp = kzalloc((n->size - AHASH_INIT_SIZE)
- * set->dsize,
- GFP_ATOMIC);
- if (!tmp) {
- ret = 0;
+
+ for (; i < n->pos; i++) {
+ if (!test_bit(i, n->used))
+ k++;
+ }
+ if (n->pos == 0 && k == 0) {
+ rcu_assign_pointer(hbucket(t, key), NULL);
+ kfree_rcu(n, rcu);
+ } else if (k >= AHASH_INIT_SIZE) {
+ struct hbucket *tmp = kzalloc(sizeof(*tmp) +
+ (n->size - AHASH_INIT_SIZE) * dsize,
+ GFP_ATOMIC);
+ if (!tmp)
goto out;
+ tmp->size = n->size - AHASH_INIT_SIZE;
+ for (j = 0, k = 0; j < n->pos; j++) {
+ if (!test_bit(j, n->used))
+ continue;
+ data = ahash_data(n, j, dsize);
+ memcpy(tmp->value + k * dsize, data, dsize);
+ set_bit(j, tmp->used);
+ k++;
}
- n->size -= AHASH_INIT_SIZE;
- memcpy(tmp, n->value, n->size * set->dsize);
- kfree(n->value);
- n->value = tmp;
+ tmp->pos = k;
+ rcu_assign_pointer(hbucket(t, key), tmp);
+ kfree_rcu(n, rcu);
}
- ret = 0;
goto out;
}
out:
- rcu_read_unlock_bh();
return ret;
}
@@ -865,8 +958,12 @@ mtype_test_cidrs(struct ip_set *set, struct mtype_elem *d,
mtype_data_netmask(d, NCIDR_GET(h->nets[j].cidr[0]));
#endif
key = HKEY(d, h->initval, t->htable_bits);
- n = hbucket(t, key);
+ n = rcu_dereference_bh(hbucket(t, key));
+ if (!n)
+ continue;
for (i = 0; i < n->pos; i++) {
+ if (!test_bit(i, n->used))
+ continue;
data = ahash_data(n, i, set->dsize);
if (!mtype_data_equal(data, d, &multi))
continue;
@@ -904,7 +1001,6 @@ mtype_test(struct ip_set *set, void *value, const struct ip_set_ext *ext,
int i, ret = 0;
u32 key, multi = 0;
- rcu_read_lock_bh();
t = rcu_dereference_bh(h->table);
#ifdef IP_SET_HASH_WITH_NETS
/* If we test an IP address and not a network address,
@@ -919,8 +1015,14 @@ mtype_test(struct ip_set *set, void *value, const struct ip_set_ext *ext,
#endif
key = HKEY(d, h->initval, t->htable_bits);
- n = hbucket(t, key);
+ n = rcu_dereference_bh(hbucket(t, key));
+ if (!n) {
+ ret = 0;
+ goto out;
+ }
for (i = 0; i < n->pos; i++) {
+ if (!test_bit(i, n->used))
+ continue;
data = ahash_data(n, i, set->dsize);
if (mtype_data_equal(data, d, &multi) &&
!(SET_WITH_TIMEOUT(set) &&
@@ -930,7 +1032,6 @@ mtype_test(struct ip_set *set, void *value, const struct ip_set_ext *ext,
}
}
out:
- rcu_read_unlock_bh();
return ret;
}
@@ -942,15 +1043,19 @@ mtype_head(struct ip_set *set, struct sk_buff *skb)
const struct htable *t;
struct nlattr *nested;
size_t memsize;
+ u8 htable_bits;
+ rcu_read_lock_bh();
t = rcu_dereference_bh_nfnl(h->table);
memsize = mtype_ahash_memsize(h, t, NLEN(set->family), set->dsize);
+ htable_bits = t->htable_bits;
+ rcu_read_unlock_bh();
nested = ipset_nest_start(skb, IPSET_ATTR_DATA);
if (!nested)
goto nla_put_failure;
if (nla_put_net32(skb, IPSET_ATTR_HASHSIZE,
- htonl(jhash_size(t->htable_bits))) ||
+ htonl(jhash_size(htable_bits))) ||
nla_put_net32(skb, IPSET_ATTR_MAXELEM, htonl(h->maxelem)))
goto nla_put_failure;
#ifdef IP_SET_HASH_WITH_NETMASK
@@ -1010,20 +1115,27 @@ mtype_list(const struct ip_set *set,
u32 first = cb->args[IPSET_CB_ARG0];
/* We assume that one hash bucket fills into one page */
void *incomplete;
- int i;
+ int i, ret = 0;
atd = ipset_nest_start(skb, IPSET_ATTR_ADT);
if (!atd)
return -EMSGSIZE;
+
pr_debug("list hash set %s\n", set->name);
t = (const struct htable *)cb->args[IPSET_CB_PRIVATE];
+ /* Expire may replace a hbucket with another one */
+ rcu_read_lock();
for (; cb->args[IPSET_CB_ARG0] < jhash_size(t->htable_bits);
cb->args[IPSET_CB_ARG0]++) {
incomplete = skb_tail_pointer(skb);
- n = hbucket(t, cb->args[IPSET_CB_ARG0]);
+ n = rcu_dereference(hbucket(t, cb->args[IPSET_CB_ARG0]));
pr_debug("cb->arg bucket: %lu, t %p n %p\n",
cb->args[IPSET_CB_ARG0], t, n);
+ if (!n)
+ continue;
for (i = 0; i < n->pos; i++) {
+ if (!test_bit(i, n->used))
+ continue;
e = ahash_data(n, i, set->dsize);
if (SET_WITH_TIMEOUT(set) &&
ip_set_timeout_expired(ext_timeout(e, set)))
@@ -1034,7 +1146,8 @@ mtype_list(const struct ip_set *set,
if (!nested) {
if (cb->args[IPSET_CB_ARG0] == first) {
nla_nest_cancel(skb, atd);
- return -EMSGSIZE;
+ ret = -EMSGSIZE;
+ goto out;
} else
goto nla_put_failure;
}
@@ -1049,7 +1162,7 @@ mtype_list(const struct ip_set *set,
/* Set listing finished */
cb->args[IPSET_CB_ARG0] = 0;
- return 0;
+ goto out;
nla_put_failure:
nlmsg_trim(skb, incomplete);
@@ -1057,10 +1170,12 @@ nla_put_failure:
pr_warn("Can't list set %s: one bucket does not fit into a message. Please report it!\n",
set->name);
cb->args[IPSET_CB_ARG0] = 0;
- return -EMSGSIZE;
- }
- ipset_nest_end(skb, atd);
- return 0;
+ ret = -EMSGSIZE;
+ } else
+ ipset_nest_end(skb, atd);
+out:
+ rcu_read_unlock();
+ return ret;
}
static int
@@ -1122,12 +1237,14 @@ IPSET_TOKEN(HTYPE, _create)(struct net *net, struct ip_set *set,
if (unlikely(!ip_set_optattr_netorder(tb, IPSET_ATTR_HASHSIZE) ||
!ip_set_optattr_netorder(tb, IPSET_ATTR_MAXELEM) ||
-#ifdef IP_SET_HASH_WITH_MARKMASK
- !ip_set_optattr_netorder(tb, IPSET_ATTR_MARKMASK) ||
-#endif
!ip_set_optattr_netorder(tb, IPSET_ATTR_TIMEOUT) ||
!ip_set_optattr_netorder(tb, IPSET_ATTR_CADT_FLAGS)))
return -IPSET_ERR_PROTOCOL;
+#ifdef IP_SET_HASH_WITH_MARKMASK
+ /* Separated condition in order to avoid directive in argument list */
+ if (unlikely(!ip_set_optattr_netorder(tb, IPSET_ATTR_MARKMASK)))
+ return -IPSET_ERR_PROTOCOL;
+#endif
if (tb[IPSET_ATTR_HASHSIZE]) {
hashsize = ip_set_get_h32(tb[IPSET_ATTR_HASHSIZE]);
@@ -1150,7 +1267,7 @@ IPSET_TOKEN(HTYPE, _create)(struct net *net, struct ip_set *set,
#endif
#ifdef IP_SET_HASH_WITH_MARKMASK
if (tb[IPSET_ATTR_MARKMASK]) {
- markmask = ntohl(nla_get_u32(tb[IPSET_ATTR_MARKMASK]));
+ markmask = ntohl(nla_get_be32(tb[IPSET_ATTR_MARKMASK]));
if (markmask == 0)
return -IPSET_ERR_INVALID_MARKMASK;
diff --git a/net/netfilter/ipset/ip_set_hash_ip.c b/net/netfilter/ipset/ip_set_hash_ip.c
index 2bbadcc..f54d706 100644
--- a/net/netfilter/ipset/ip_set_hash_ip.c
+++ b/net/netfilter/ipset/ip_set_hash_ip.c
@@ -315,6 +315,7 @@ hash_ip_init(void)
static void __exit
hash_ip_fini(void)
{
+ rcu_barrier();
ip_set_type_unregister(&hash_ip_type);
}
diff --git a/net/netfilter/ipset/ip_set_hash_ipmark.c b/net/netfilter/ipset/ip_set_hash_ipmark.c
index 3aafb36..f8fbc32 100644
--- a/net/netfilter/ipset/ip_set_hash_ipmark.c
+++ b/net/netfilter/ipset/ip_set_hash_ipmark.c
@@ -319,6 +319,7 @@ hash_ipmark_init(void)
static void __exit
hash_ipmark_fini(void)
{
+ rcu_barrier();
ip_set_type_unregister(&hash_ipmark_type);
}
diff --git a/net/netfilter/ipset/ip_set_hash_ipport.c b/net/netfilter/ipset/ip_set_hash_ipport.c
index 4db1270..9a31db8 100644
--- a/net/netfilter/ipset/ip_set_hash_ipport.c
+++ b/net/netfilter/ipset/ip_set_hash_ipport.c
@@ -382,6 +382,7 @@ hash_ipport_init(void)
static void __exit
hash_ipport_fini(void)
{
+ rcu_barrier();
ip_set_type_unregister(&hash_ipport_type);
}
diff --git a/net/netfilter/ipset/ip_set_hash_ipportip.c b/net/netfilter/ipset/ip_set_hash_ipportip.c
index c01bf68..fc42489 100644
--- a/net/netfilter/ipset/ip_set_hash_ipportip.c
+++ b/net/netfilter/ipset/ip_set_hash_ipportip.c
@@ -397,6 +397,7 @@ hash_ipportip_init(void)
static void __exit
hash_ipportip_fini(void)
{
+ rcu_barrier();
ip_set_type_unregister(&hash_ipportip_type);
}
diff --git a/net/netfilter/ipset/ip_set_hash_ipportnet.c b/net/netfilter/ipset/ip_set_hash_ipportnet.c
index 50248de..2a69b9b 100644
--- a/net/netfilter/ipset/ip_set_hash_ipportnet.c
+++ b/net/netfilter/ipset/ip_set_hash_ipportnet.c
@@ -554,6 +554,7 @@ hash_ipportnet_init(void)
static void __exit
hash_ipportnet_fini(void)
{
+ rcu_barrier();
ip_set_type_unregister(&hash_ipportnet_type);
}
diff --git a/net/netfilter/ipset/ip_set_hash_mac.c b/net/netfilter/ipset/ip_set_hash_mac.c
index 8981c8b..112aff3 100644
--- a/net/netfilter/ipset/ip_set_hash_mac.c
+++ b/net/netfilter/ipset/ip_set_hash_mac.c
@@ -165,6 +165,7 @@ hash_mac_init(void)
static void __exit
hash_mac_fini(void)
{
+ rcu_barrier();
ip_set_type_unregister(&hash_mac_type);
}
diff --git a/net/netfilter/ipset/ip_set_hash_net.c b/net/netfilter/ipset/ip_set_hash_net.c
index 089b23f..e49b1d0 100644
--- a/net/netfilter/ipset/ip_set_hash_net.c
+++ b/net/netfilter/ipset/ip_set_hash_net.c
@@ -392,6 +392,7 @@ hash_net_init(void)
static void __exit
hash_net_fini(void)
{
+ rcu_barrier();
ip_set_type_unregister(&hash_net_type);
}
diff --git a/net/netfilter/ipset/ip_set_hash_netiface.c b/net/netfilter/ipset/ip_set_hash_netiface.c
index 3258189..42c893e 100644
--- a/net/netfilter/ipset/ip_set_hash_netiface.c
+++ b/net/netfilter/ipset/ip_set_hash_netiface.c
@@ -500,6 +500,7 @@ hash_netiface_init(void)
static void __exit
hash_netiface_fini(void)
{
+ rcu_barrier();
ip_set_type_unregister(&hash_netiface_type);
}
diff --git a/net/netfilter/ipset/ip_set_hash_netnet.c b/net/netfilter/ipset/ip_set_hash_netnet.c
index ed9cc45..b5428be 100644
--- a/net/netfilter/ipset/ip_set_hash_netnet.c
+++ b/net/netfilter/ipset/ip_set_hash_netnet.c
@@ -480,6 +480,7 @@ hash_netnet_init(void)
static void __exit
hash_netnet_fini(void)
{
+ rcu_barrier();
ip_set_type_unregister(&hash_netnet_type);
}
diff --git a/net/netfilter/ipset/ip_set_hash_netport.c b/net/netfilter/ipset/ip_set_hash_netport.c
index fbaf813..27307d0 100644
--- a/net/netfilter/ipset/ip_set_hash_netport.c
+++ b/net/netfilter/ipset/ip_set_hash_netport.c
@@ -498,6 +498,7 @@ hash_netport_init(void)
static void __exit
hash_netport_fini(void)
{
+ rcu_barrier();
ip_set_type_unregister(&hash_netport_type);
}
diff --git a/net/netfilter/ipset/ip_set_hash_netportnet.c b/net/netfilter/ipset/ip_set_hash_netportnet.c
index a828cbc..1e0e47a 100644
--- a/net/netfilter/ipset/ip_set_hash_netportnet.c
+++ b/net/netfilter/ipset/ip_set_hash_netportnet.c
@@ -581,6 +581,7 @@ hash_netportnet_init(void)
static void __exit
hash_netportnet_fini(void)
{
+ rcu_barrier();
ip_set_type_unregister(&hash_netportnet_type);
}
--
1.7.10.4
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
