* [PATCH AUTOSEL 5.13 06/12] netfilter: ipset: Limit the maximal range of consecutive elements to add/delete
[not found] <20210817003536.83063-1-sashal@kernel.org>
@ 2021-08-17 0:35 ` Sasha Levin
2021-08-17 0:35 ` [PATCH AUTOSEL 5.13 07/12] netfilter: conntrack: collect all entries in one cycle Sasha Levin
1 sibling, 0 replies; 2+ messages in thread
From: Sasha Levin @ 2021-08-17 0:35 UTC (permalink / raw)
To: linux-kernel, stable
Cc: Jozsef Kadlecsik, Brad Spengler, Pablo Neira Ayuso, Sasha Levin,
netfilter-devel, coreteam, netdev
From: Jozsef Kadlecsik <kadlec@netfilter.org>
[ Upstream commit 5f7b51bf09baca8e4f80cbe879536842bafb5f31 ]
The range size of consecutive elements were not limited. Thus one could
define a huge range which may result soft lockup errors due to the long
execution time. Now the range size is limited to 2^20 entries.
Reported-by: Brad Spengler <spender@grsecurity.net>
Signed-off-by: Jozsef Kadlecsik <kadlec@netfilter.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
include/linux/netfilter/ipset/ip_set.h | 3 +++
net/netfilter/ipset/ip_set_hash_ip.c | 9 ++++++++-
net/netfilter/ipset/ip_set_hash_ipmark.c | 10 +++++++++-
net/netfilter/ipset/ip_set_hash_ipport.c | 3 +++
net/netfilter/ipset/ip_set_hash_ipportip.c | 3 +++
net/netfilter/ipset/ip_set_hash_ipportnet.c | 3 +++
net/netfilter/ipset/ip_set_hash_net.c | 11 ++++++++++-
net/netfilter/ipset/ip_set_hash_netiface.c | 10 +++++++++-
net/netfilter/ipset/ip_set_hash_netnet.c | 16 +++++++++++++++-
net/netfilter/ipset/ip_set_hash_netport.c | 11 ++++++++++-
net/netfilter/ipset/ip_set_hash_netportnet.c | 16 +++++++++++++++-
11 files changed, 88 insertions(+), 7 deletions(-)
diff --git a/include/linux/netfilter/ipset/ip_set.h b/include/linux/netfilter/ipset/ip_set.h
index 10279c4830ac..ada1296c87d5 100644
--- a/include/linux/netfilter/ipset/ip_set.h
+++ b/include/linux/netfilter/ipset/ip_set.h
@@ -196,6 +196,9 @@ struct ip_set_region {
u32 elements; /* Number of elements vs timeout */
};
+/* Max range where every element is added/deleted in one step */
+#define IPSET_MAX_RANGE (1<<20)
+
/* The max revision number supported by any set type + 1 */
#define IPSET_REVISION_MAX 9
diff --git a/net/netfilter/ipset/ip_set_hash_ip.c b/net/netfilter/ipset/ip_set_hash_ip.c
index d1bef23fd4f5..dd30c03d5a23 100644
--- a/net/netfilter/ipset/ip_set_hash_ip.c
+++ b/net/netfilter/ipset/ip_set_hash_ip.c
@@ -132,8 +132,11 @@ hash_ip4_uadt(struct ip_set *set, struct nlattr *tb[],
ret = ip_set_get_hostipaddr4(tb[IPSET_ATTR_IP_TO], &ip_to);
if (ret)
return ret;
- if (ip > ip_to)
+ if (ip > ip_to) {
+ if (ip_to == 0)
+ return -IPSET_ERR_HASH_ELEM;
swap(ip, ip_to);
+ }
} else if (tb[IPSET_ATTR_CIDR]) {
u8 cidr = nla_get_u8(tb[IPSET_ATTR_CIDR]);
@@ -144,6 +147,10 @@ hash_ip4_uadt(struct ip_set *set, struct nlattr *tb[],
hosts = h->netmask == 32 ? 1 : 2 << (32 - h->netmask - 1);
+ /* 64bit division is not allowed on 32bit */
+ if (((u64)ip_to - ip + 1) >> (32 - h->netmask) > IPSET_MAX_RANGE)
+ return -ERANGE;
+
if (retried) {
ip = ntohl(h->next.ip);
e.ip = htonl(ip);
diff --git a/net/netfilter/ipset/ip_set_hash_ipmark.c b/net/netfilter/ipset/ip_set_hash_ipmark.c
index 18346d18aa16..153de3457423 100644
--- a/net/netfilter/ipset/ip_set_hash_ipmark.c
+++ b/net/netfilter/ipset/ip_set_hash_ipmark.c
@@ -121,6 +121,8 @@ hash_ipmark4_uadt(struct ip_set *set, struct nlattr *tb[],
e.mark = ntohl(nla_get_be32(tb[IPSET_ATTR_MARK]));
e.mark &= h->markmask;
+ if (e.mark == 0 && e.ip == 0)
+ return -IPSET_ERR_HASH_ELEM;
if (adt == IPSET_TEST ||
!(tb[IPSET_ATTR_IP_TO] || tb[IPSET_ATTR_CIDR])) {
@@ -133,8 +135,11 @@ hash_ipmark4_uadt(struct ip_set *set, struct nlattr *tb[],
ret = ip_set_get_hostipaddr4(tb[IPSET_ATTR_IP_TO], &ip_to);
if (ret)
return ret;
- if (ip > ip_to)
+ if (ip > ip_to) {
+ if (e.mark == 0 && ip_to == 0)
+ return -IPSET_ERR_HASH_ELEM;
swap(ip, ip_to);
+ }
} else if (tb[IPSET_ATTR_CIDR]) {
u8 cidr = nla_get_u8(tb[IPSET_ATTR_CIDR]);
@@ -143,6 +148,9 @@ hash_ipmark4_uadt(struct ip_set *set, struct nlattr *tb[],
ip_set_mask_from_to(ip, ip_to, cidr);
}
+ if (((u64)ip_to - ip + 1) > IPSET_MAX_RANGE)
+ return -ERANGE;
+
if (retried)
ip = ntohl(h->next.ip);
for (; ip <= ip_to; ip++) {
diff --git a/net/netfilter/ipset/ip_set_hash_ipport.c b/net/netfilter/ipset/ip_set_hash_ipport.c
index e1ca11196515..7303138e46be 100644
--- a/net/netfilter/ipset/ip_set_hash_ipport.c
+++ b/net/netfilter/ipset/ip_set_hash_ipport.c
@@ -173,6 +173,9 @@ hash_ipport4_uadt(struct ip_set *set, struct nlattr *tb[],
swap(port, port_to);
}
+ if (((u64)ip_to - ip + 1)*(port_to - port + 1) > IPSET_MAX_RANGE)
+ return -ERANGE;
+
if (retried)
ip = ntohl(h->next.ip);
for (; ip <= ip_to; ip++) {
diff --git a/net/netfilter/ipset/ip_set_hash_ipportip.c b/net/netfilter/ipset/ip_set_hash_ipportip.c
index ab179e064597..334fb1ad0e86 100644
--- a/net/netfilter/ipset/ip_set_hash_ipportip.c
+++ b/net/netfilter/ipset/ip_set_hash_ipportip.c
@@ -180,6 +180,9 @@ hash_ipportip4_uadt(struct ip_set *set, struct nlattr *tb[],
swap(port, port_to);
}
+ if (((u64)ip_to - ip + 1)*(port_to - port + 1) > IPSET_MAX_RANGE)
+ return -ERANGE;
+
if (retried)
ip = ntohl(h->next.ip);
for (; ip <= ip_to; ip++) {
diff --git a/net/netfilter/ipset/ip_set_hash_ipportnet.c b/net/netfilter/ipset/ip_set_hash_ipportnet.c
index 8f075b44cf64..7df94f437f60 100644
--- a/net/netfilter/ipset/ip_set_hash_ipportnet.c
+++ b/net/netfilter/ipset/ip_set_hash_ipportnet.c
@@ -253,6 +253,9 @@ hash_ipportnet4_uadt(struct ip_set *set, struct nlattr *tb[],
swap(port, port_to);
}
+ if (((u64)ip_to - ip + 1)*(port_to - port + 1) > IPSET_MAX_RANGE)
+ return -ERANGE;
+
ip2_to = ip2_from;
if (tb[IPSET_ATTR_IP2_TO]) {
ret = ip_set_get_hostipaddr4(tb[IPSET_ATTR_IP2_TO], &ip2_to);
diff --git a/net/netfilter/ipset/ip_set_hash_net.c b/net/netfilter/ipset/ip_set_hash_net.c
index c1a11f041ac6..1422739d9aa2 100644
--- a/net/netfilter/ipset/ip_set_hash_net.c
+++ b/net/netfilter/ipset/ip_set_hash_net.c
@@ -140,7 +140,7 @@ hash_net4_uadt(struct ip_set *set, struct nlattr *tb[],
ipset_adtfn adtfn = set->variant->adt[adt];
struct hash_net4_elem e = { .cidr = HOST_MASK };
struct ip_set_ext ext = IP_SET_INIT_UEXT(set);
- u32 ip = 0, ip_to = 0;
+ u32 ip = 0, ip_to = 0, ipn, n = 0;
int ret;
if (tb[IPSET_ATTR_LINENO])
@@ -188,6 +188,15 @@ hash_net4_uadt(struct ip_set *set, struct nlattr *tb[],
if (ip + UINT_MAX == ip_to)
return -IPSET_ERR_HASH_RANGE;
}
+ ipn = ip;
+ do {
+ ipn = ip_set_range_to_cidr(ipn, ip_to, &e.cidr);
+ n++;
+ } while (ipn++ < ip_to);
+
+ if (n > IPSET_MAX_RANGE)
+ return -ERANGE;
+
if (retried)
ip = ntohl(h->next.ip);
do {
diff --git a/net/netfilter/ipset/ip_set_hash_netiface.c b/net/netfilter/ipset/ip_set_hash_netiface.c
index ddd51c2e1cb3..9810f5bf63f5 100644
--- a/net/netfilter/ipset/ip_set_hash_netiface.c
+++ b/net/netfilter/ipset/ip_set_hash_netiface.c
@@ -202,7 +202,7 @@ hash_netiface4_uadt(struct ip_set *set, struct nlattr *tb[],
ipset_adtfn adtfn = set->variant->adt[adt];
struct hash_netiface4_elem e = { .cidr = HOST_MASK, .elem = 1 };
struct ip_set_ext ext = IP_SET_INIT_UEXT(set);
- u32 ip = 0, ip_to = 0;
+ u32 ip = 0, ip_to = 0, ipn, n = 0;
int ret;
if (tb[IPSET_ATTR_LINENO])
@@ -256,6 +256,14 @@ hash_netiface4_uadt(struct ip_set *set, struct nlattr *tb[],
} else {
ip_set_mask_from_to(ip, ip_to, e.cidr);
}
+ ipn = ip;
+ do {
+ ipn = ip_set_range_to_cidr(ipn, ip_to, &e.cidr);
+ n++;
+ } while (ipn++ < ip_to);
+
+ if (n > IPSET_MAX_RANGE)
+ return -ERANGE;
if (retried)
ip = ntohl(h->next.ip);
diff --git a/net/netfilter/ipset/ip_set_hash_netnet.c b/net/netfilter/ipset/ip_set_hash_netnet.c
index 6532f0505e66..3d09eefe998a 100644
--- a/net/netfilter/ipset/ip_set_hash_netnet.c
+++ b/net/netfilter/ipset/ip_set_hash_netnet.c
@@ -168,7 +168,8 @@ hash_netnet4_uadt(struct ip_set *set, struct nlattr *tb[],
struct hash_netnet4_elem e = { };
struct ip_set_ext ext = IP_SET_INIT_UEXT(set);
u32 ip = 0, ip_to = 0;
- u32 ip2 = 0, ip2_from = 0, ip2_to = 0;
+ u32 ip2 = 0, ip2_from = 0, ip2_to = 0, ipn;
+ u64 n = 0, m = 0;
int ret;
if (tb[IPSET_ATTR_LINENO])
@@ -244,6 +245,19 @@ hash_netnet4_uadt(struct ip_set *set, struct nlattr *tb[],
} else {
ip_set_mask_from_to(ip2_from, ip2_to, e.cidr[1]);
}
+ ipn = ip;
+ do {
+ ipn = ip_set_range_to_cidr(ipn, ip_to, &e.cidr[0]);
+ n++;
+ } while (ipn++ < ip_to);
+ ipn = ip2_from;
+ do {
+ ipn = ip_set_range_to_cidr(ipn, ip2_to, &e.cidr[1]);
+ m++;
+ } while (ipn++ < ip2_to);
+
+ if (n*m > IPSET_MAX_RANGE)
+ return -ERANGE;
if (retried) {
ip = ntohl(h->next.ip[0]);
diff --git a/net/netfilter/ipset/ip_set_hash_netport.c b/net/netfilter/ipset/ip_set_hash_netport.c
index ec1564a1cb5a..09cf72eb37f8 100644
--- a/net/netfilter/ipset/ip_set_hash_netport.c
+++ b/net/netfilter/ipset/ip_set_hash_netport.c
@@ -158,7 +158,8 @@ hash_netport4_uadt(struct ip_set *set, struct nlattr *tb[],
ipset_adtfn adtfn = set->variant->adt[adt];
struct hash_netport4_elem e = { .cidr = HOST_MASK - 1 };
struct ip_set_ext ext = IP_SET_INIT_UEXT(set);
- u32 port, port_to, p = 0, ip = 0, ip_to = 0;
+ u32 port, port_to, p = 0, ip = 0, ip_to = 0, ipn;
+ u64 n = 0;
bool with_ports = false;
u8 cidr;
int ret;
@@ -235,6 +236,14 @@ hash_netport4_uadt(struct ip_set *set, struct nlattr *tb[],
} else {
ip_set_mask_from_to(ip, ip_to, e.cidr + 1);
}
+ ipn = ip;
+ do {
+ ipn = ip_set_range_to_cidr(ipn, ip_to, &cidr);
+ n++;
+ } while (ipn++ < ip_to);
+
+ if (n*(port_to - port + 1) > IPSET_MAX_RANGE)
+ return -ERANGE;
if (retried) {
ip = ntohl(h->next.ip);
diff --git a/net/netfilter/ipset/ip_set_hash_netportnet.c b/net/netfilter/ipset/ip_set_hash_netportnet.c
index 0e91d1e82f1c..19bcdb3141f6 100644
--- a/net/netfilter/ipset/ip_set_hash_netportnet.c
+++ b/net/netfilter/ipset/ip_set_hash_netportnet.c
@@ -182,7 +182,8 @@ hash_netportnet4_uadt(struct ip_set *set, struct nlattr *tb[],
struct hash_netportnet4_elem e = { };
struct ip_set_ext ext = IP_SET_INIT_UEXT(set);
u32 ip = 0, ip_to = 0, p = 0, port, port_to;
- u32 ip2_from = 0, ip2_to = 0, ip2;
+ u32 ip2_from = 0, ip2_to = 0, ip2, ipn;
+ u64 n = 0, m = 0;
bool with_ports = false;
int ret;
@@ -284,6 +285,19 @@ hash_netportnet4_uadt(struct ip_set *set, struct nlattr *tb[],
} else {
ip_set_mask_from_to(ip2_from, ip2_to, e.cidr[1]);
}
+ ipn = ip;
+ do {
+ ipn = ip_set_range_to_cidr(ipn, ip_to, &e.cidr[0]);
+ n++;
+ } while (ipn++ < ip_to);
+ ipn = ip2_from;
+ do {
+ ipn = ip_set_range_to_cidr(ipn, ip2_to, &e.cidr[1]);
+ m++;
+ } while (ipn++ < ip2_to);
+
+ if (n*m*(port_to - port + 1) > IPSET_MAX_RANGE)
+ return -ERANGE;
if (retried) {
ip = ntohl(h->next.ip[0]);
--
2.30.2
^ permalink raw reply related [flat|nested] 2+ messages in thread
* [PATCH AUTOSEL 5.13 07/12] netfilter: conntrack: collect all entries in one cycle
[not found] <20210817003536.83063-1-sashal@kernel.org>
2021-08-17 0:35 ` [PATCH AUTOSEL 5.13 06/12] netfilter: ipset: Limit the maximal range of consecutive elements to add/delete Sasha Levin
@ 2021-08-17 0:35 ` Sasha Levin
1 sibling, 0 replies; 2+ messages in thread
From: Sasha Levin @ 2021-08-17 0:35 UTC (permalink / raw)
To: linux-kernel, stable
Cc: Florian Westphal, Michal Kubecek, Pablo Neira Ayuso, Sasha Levin,
netfilter-devel, coreteam, netdev
From: Florian Westphal <fw@strlen.de>
[ Upstream commit 4608fdfc07e116f9fc0895beb40abad7cdb5ee3d ]
Michal Kubecek reports that conntrack gc is responsible for frequent
wakeups (every 125ms) on idle systems.
On busy systems, timed out entries are evicted during lookup.
The gc worker is only needed to remove entries after system becomes idle
after a busy period.
To resolve this, always scan the entire table.
If the scan is taking too long, reschedule so other work_structs can run
and resume from next bucket.
After a completed scan, wait for 2 minutes before the next cycle.
Heuristics for faster re-schedule are removed.
GC_SCAN_INTERVAL could be exposed as a sysctl in the future to allow
tuning this as-needed or even turn the gc worker off.
Reported-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
net/netfilter/nf_conntrack_core.c | 71 ++++++++++---------------------
1 file changed, 22 insertions(+), 49 deletions(-)
diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c
index 69079a382d3a..6ba9f4b8d145 100644
--- a/net/netfilter/nf_conntrack_core.c
+++ b/net/netfilter/nf_conntrack_core.c
@@ -68,22 +68,17 @@ EXPORT_SYMBOL_GPL(nf_conntrack_hash);
struct conntrack_gc_work {
struct delayed_work dwork;
- u32 last_bucket;
+ u32 next_bucket;
bool exiting;
bool early_drop;
- long next_gc_run;
};
static __read_mostly struct kmem_cache *nf_conntrack_cachep;
static DEFINE_SPINLOCK(nf_conntrack_locks_all_lock);
static __read_mostly bool nf_conntrack_locks_all;
-/* every gc cycle scans at most 1/GC_MAX_BUCKETS_DIV part of table */
-#define GC_MAX_BUCKETS_DIV 128u
-/* upper bound of full table scan */
-#define GC_MAX_SCAN_JIFFIES (16u * HZ)
-/* desired ratio of entries found to be expired */
-#define GC_EVICT_RATIO 50u
+#define GC_SCAN_INTERVAL (120u * HZ)
+#define GC_SCAN_MAX_DURATION msecs_to_jiffies(10)
static struct conntrack_gc_work conntrack_gc_work;
@@ -1359,17 +1354,13 @@ static bool gc_worker_can_early_drop(const struct nf_conn *ct)
static void gc_worker(struct work_struct *work)
{
- unsigned int min_interval = max(HZ / GC_MAX_BUCKETS_DIV, 1u);
- unsigned int i, goal, buckets = 0, expired_count = 0;
- unsigned int nf_conntrack_max95 = 0;
+ unsigned long end_time = jiffies + GC_SCAN_MAX_DURATION;
+ unsigned int i, hashsz, nf_conntrack_max95 = 0;
+ unsigned long next_run = GC_SCAN_INTERVAL;
struct conntrack_gc_work *gc_work;
- unsigned int ratio, scanned = 0;
- unsigned long next_run;
-
gc_work = container_of(work, struct conntrack_gc_work, dwork.work);
- goal = nf_conntrack_htable_size / GC_MAX_BUCKETS_DIV;
- i = gc_work->last_bucket;
+ i = gc_work->next_bucket;
if (gc_work->early_drop)
nf_conntrack_max95 = nf_conntrack_max / 100u * 95u;
@@ -1377,15 +1368,15 @@ static void gc_worker(struct work_struct *work)
struct nf_conntrack_tuple_hash *h;
struct hlist_nulls_head *ct_hash;
struct hlist_nulls_node *n;
- unsigned int hashsz;
struct nf_conn *tmp;
- i++;
rcu_read_lock();
nf_conntrack_get_ht(&ct_hash, &hashsz);
- if (i >= hashsz)
- i = 0;
+ if (i >= hashsz) {
+ rcu_read_unlock();
+ break;
+ }
hlist_nulls_for_each_entry_rcu(h, n, &ct_hash[i], hnnode) {
struct nf_conntrack_net *cnet;
@@ -1393,7 +1384,6 @@ static void gc_worker(struct work_struct *work)
tmp = nf_ct_tuplehash_to_ctrack(h);
- scanned++;
if (test_bit(IPS_OFFLOAD_BIT, &tmp->status)) {
nf_ct_offload_timeout(tmp);
continue;
@@ -1401,7 +1391,6 @@ static void gc_worker(struct work_struct *work)
if (nf_ct_is_expired(tmp)) {
nf_ct_gc_expired(tmp);
- expired_count++;
continue;
}
@@ -1434,7 +1423,14 @@ static void gc_worker(struct work_struct *work)
*/
rcu_read_unlock();
cond_resched();
- } while (++buckets < goal);
+ i++;
+
+ if (time_after(jiffies, end_time) && i < hashsz) {
+ gc_work->next_bucket = i;
+ next_run = 0;
+ break;
+ }
+ } while (i < hashsz);
if (gc_work->exiting)
return;
@@ -1445,40 +1441,17 @@ static void gc_worker(struct work_struct *work)
*
* This worker is only here to reap expired entries when system went
* idle after a busy period.
- *
- * The heuristics below are supposed to balance conflicting goals:
- *
- * 1. Minimize time until we notice a stale entry
- * 2. Maximize scan intervals to not waste cycles
- *
- * Normally, expire ratio will be close to 0.
- *
- * As soon as a sizeable fraction of the entries have expired
- * increase scan frequency.
*/
- ratio = scanned ? expired_count * 100 / scanned : 0;
- if (ratio > GC_EVICT_RATIO) {
- gc_work->next_gc_run = min_interval;
- } else {
- unsigned int max = GC_MAX_SCAN_JIFFIES / GC_MAX_BUCKETS_DIV;
-
- BUILD_BUG_ON((GC_MAX_SCAN_JIFFIES / GC_MAX_BUCKETS_DIV) == 0);
-
- gc_work->next_gc_run += min_interval;
- if (gc_work->next_gc_run > max)
- gc_work->next_gc_run = max;
+ if (next_run) {
+ gc_work->early_drop = false;
+ gc_work->next_bucket = 0;
}
-
- next_run = gc_work->next_gc_run;
- gc_work->last_bucket = i;
- gc_work->early_drop = false;
queue_delayed_work(system_power_efficient_wq, &gc_work->dwork, next_run);
}
static void conntrack_gc_work_init(struct conntrack_gc_work *gc_work)
{
INIT_DEFERRABLE_WORK(&gc_work->dwork, gc_worker);
- gc_work->next_gc_run = HZ;
gc_work->exiting = false;
}
--
2.30.2
^ permalink raw reply related [flat|nested] 2+ messages in thread
end of thread, other threads:[~2021-08-17 0:35 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
[not found] <20210817003536.83063-1-sashal@kernel.org>
2021-08-17 0:35 ` [PATCH AUTOSEL 5.13 06/12] netfilter: ipset: Limit the maximal range of consecutive elements to add/delete Sasha Levin
2021-08-17 0:35 ` [PATCH AUTOSEL 5.13 07/12] netfilter: conntrack: collect all entries in one cycle Sasha Levin
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).