netfilter: ipset: Rework long task execution when adding/deleting entries

When adding/deleting large number of elements in one step in ipset, it can
take a reasonable amount of time and can result in soft lockup errors. The
patch 5f7b51bf09ba ("netfilter: ipset: Limit the maximal range of
consecutive elements to add/delete") tried to fix it by limiting the max
elements to process at all. However it was not enough, it is still possible
that we get hung tasks. Lowering the limit is not reasonable, so the
approach in this patch is as follows: rely on the method used at resizing
sets and save the state when we reach a smaller internal batch limit,
unlock/lock and proceed from the saved state. Thus we can avoid long
continuous tasks and at the same time removed the limit to add/delete large
number of elements in one step.

The nfnl mutex is held during the whole operation which prevents one to issue
other ipset commands in parallel.

Signed-off-by: Jozsef Kadlecsik <kadlec@netfilter.org>
Reported-by: syzbot+9204e7399656300bf271@syzkaller.appspotmail.com
Fixes: 5f7b51bf09ba ("netfilter: ipset: Limit the maximal range of consecutive elements to add/delete")

11 files changed, 68 insertions, 81 deletions

diff --git a/kernel/include/linux/netfilter/ipset/ip_set.h b/kernel/include/linux/netfilter/ipset/ip_set.h
index efc0e6d..0c6eee2 100644
--- a/kernel/include/linux/netfilter/ipset/ip_set.h
+++ b/kernel/include/linux/netfilter/ipset/ip_set.h
@@ -200,7 +200,7 @@ struct ip_set_region {
 };
 
 /* Max range where every element is added/deleted in one step */
-#define IPSET_MAX_RANGE		(1<<20)
+#define IPSET_MAX_RANGE		(1<<14)
 
 /* The max revision number supported by any set type + 1 */
 #define IPSET_REVISION_MAX	9
diff --git a/kernel/net/netfilter/ipset/ip_set_core.c b/kernel/net/netfilter/ipset/ip_set_core.c
index 77fb55c..a80b999 100644
--- a/kernel/net/netfilter/ipset/ip_set_core.c
+++ b/kernel/net/netfilter/ipset/ip_set_core.c
@@ -1754,9 +1754,10 @@ CALL_AD(struct net *net, struct sock *ctnl, struct sk_buff *skb,
 		ret = set->variant->uadt(set, tb, adt, &lineno, flags, retried);
 		ip_set_unlock(set);
 		retried = true;
-	} while (ret == -EAGAIN &&
-		 set->variant->resize &&
-		 (ret = set->variant->resize(set, retried)) == 0);
+	} while (ret == -ERANGE ||
+		 (ret == -EAGAIN &&
+		  set->variant->resize &&
+		  (ret = set->variant->resize(set, retried)) == 0));
 
 	if (!ret || (ret == -IPSET_ERR_EXIST && eexist))
 		return 0;
diff --git a/kernel/net/netfilter/ipset/ip_set_hash_ip.c b/kernel/net/netfilter/ipset/ip_set_hash_ip.c
index be11354..5a2cb71 100644
--- a/kernel/net/netfilter/ipset/ip_set_hash_ip.c
+++ b/kernel/net/netfilter/ipset/ip_set_hash_ip.c
@@ -101,11 +101,11 @@ static int
 hash_ip4_uadt(struct ip_set *set, struct nlattr *tb[],
 	      enum ipset_adt adt, u32 *lineno, u32 flags, bool retried)
 {
-	const struct hash_ip4 *h = set->data;
+	struct hash_ip4 *h = set->data;
 	ipset_adtfn adtfn = set->variant->adt[adt];
 	struct hash_ip4_elem e = { 0 };
 	struct ip_set_ext ext = IP_SET_INIT_UEXT(set);
-	u32 ip = 0, ip_to = 0, hosts;
+	u32 ip = 0, ip_to = 0, hosts, i = 0;
 	int ret = 0;
 
 	if (tb[IPSET_ATTR_LINENO])
@@ -150,14 +150,14 @@ 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);
-	for (; ip <= ip_to;) {
+	for (; ip <= ip_to; i++) {
 		e.ip = htonl(ip);
+		if (i > IPSET_MAX_RANGE) {
+			hash_ip4_data_next(&h->next, &e);
+			return -ERANGE;
+		}
 		ret = adtfn(set, &e, &ext, &ext, flags);
 		if (ret && !ip_set_eexist(ret, flags))
 			return ret;
diff --git a/kernel/net/netfilter/ipset/ip_set_hash_ipmark.c b/kernel/net/netfilter/ipset/ip_set_hash_ipmark.c
index 7e7eede..ad2a2dc 100644
--- a/kernel/net/netfilter/ipset/ip_set_hash_ipmark.c
+++ b/kernel/net/netfilter/ipset/ip_set_hash_ipmark.c
@@ -99,11 +99,11 @@ static int
 hash_ipmark4_uadt(struct ip_set *set, struct nlattr *tb[],
 		  enum ipset_adt adt, u32 *lineno, u32 flags, bool retried)
 {
-	const struct hash_ipmark4 *h = set->data;
+	struct hash_ipmark4 *h = set->data;
 	ipset_adtfn adtfn = set->variant->adt[adt];
 	struct hash_ipmark4_elem e = { };
 	struct ip_set_ext ext = IP_SET_INIT_UEXT(set);
-	u32 ip, ip_to = 0;
+	u32 ip, ip_to = 0, i = 0;
 	int ret;
 
 	if (tb[IPSET_ATTR_LINENO])
@@ -150,13 +150,14 @@ 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++) {
+	for (; ip <= ip_to; ip++, i++) {
 		e.ip = htonl(ip);
+		if (i > IPSET_MAX_RANGE) {
+			hash_ipmark4_data_next(&h->next, &e);
+			return -ERANGE;
+		}
 		ret = adtfn(set, &e, &ext, &ext, flags);
 
 		if (ret && !ip_set_eexist(ret, flags))
diff --git a/kernel/net/netfilter/ipset/ip_set_hash_ipport.c b/kernel/net/netfilter/ipset/ip_set_hash_ipport.c
index 5ca5df9..46f8bbf 100644
--- a/kernel/net/netfilter/ipset/ip_set_hash_ipport.c
+++ b/kernel/net/netfilter/ipset/ip_set_hash_ipport.c
@@ -113,11 +113,11 @@ static int
 hash_ipport4_uadt(struct ip_set *set, struct nlattr *tb[],
 		  enum ipset_adt adt, u32 *lineno, u32 flags, bool retried)
 {
-	const struct hash_ipport4 *h = set->data;
+	struct hash_ipport4 *h = set->data;
 	ipset_adtfn adtfn = set->variant->adt[adt];
 	struct hash_ipport4_elem e = { .ip = 0 };
 	struct ip_set_ext ext = IP_SET_INIT_UEXT(set);
-	u32 ip, ip_to = 0, p = 0, port, port_to;
+	u32 ip, ip_to = 0, p = 0, port, port_to, i = 0;
 	bool with_ports = false;
 	int ret;
 
@@ -185,17 +185,18 @@ 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++) {
 		p = retried && ip == ntohl(h->next.ip) ? ntohs(h->next.port)
 						       : port;
-		for (; p <= port_to; p++) {
+		for (; p <= port_to; p++, i++) {
 			e.ip = htonl(ip);
 			e.port = htons(p);
+			if (i > IPSET_MAX_RANGE) {
+				hash_ipport4_data_next(&h->next, &e);
+				return -ERANGE;
+			}
 			ret = adtfn(set, &e, &ext, &ext, flags);
 
 			if (ret && !ip_set_eexist(ret, flags))
diff --git a/kernel/net/netfilter/ipset/ip_set_hash_ipportip.c b/kernel/net/netfilter/ipset/ip_set_hash_ipportip.c
index 0be8f53..ca5fe9c 100644
--- a/kernel/net/netfilter/ipset/ip_set_hash_ipportip.c
+++ b/kernel/net/netfilter/ipset/ip_set_hash_ipportip.c
@@ -109,11 +109,11 @@ static int
 hash_ipportip4_uadt(struct ip_set *set, struct nlattr *tb[],
 		    enum ipset_adt adt, u32 *lineno, u32 flags, bool retried)
 {
-	const struct hash_ipportip4 *h = set->data;
+	struct hash_ipportip4 *h = set->data;
 	ipset_adtfn adtfn = set->variant->adt[adt];
 	struct hash_ipportip4_elem e = { .ip = 0 };
 	struct ip_set_ext ext = IP_SET_INIT_UEXT(set);
-	u32 ip, ip_to = 0, p = 0, port, port_to;
+	u32 ip, ip_to = 0, p = 0, port, port_to, i = 0;
 	bool with_ports = false;
 	int ret;
 
@@ -181,17 +181,18 @@ 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++) {
 		p = retried && ip == ntohl(h->next.ip) ? ntohs(h->next.port)
 						       : port;
-		for (; p <= port_to; p++) {
+		for (; p <= port_to; p++, i++) {
 			e.ip = htonl(ip);
 			e.port = htons(p);
+			if (i > IPSET_MAX_RANGE) {
+				hash_ipportip4_data_next(&h->next, &e);
+				return -ERANGE;
+			}
 			ret = adtfn(set, &e, &ext, &ext, flags);
 
 			if (ret && !ip_set_eexist(ret, flags))
diff --git a/kernel/net/netfilter/ipset/ip_set_hash_ipportnet.c b/kernel/net/netfilter/ipset/ip_set_hash_ipportnet.c
index 8f3c069..8f44644 100644
--- a/kernel/net/netfilter/ipset/ip_set_hash_ipportnet.c
+++ b/kernel/net/netfilter/ipset/ip_set_hash_ipportnet.c
@@ -161,12 +161,12 @@ static int
 hash_ipportnet4_uadt(struct ip_set *set, struct nlattr *tb[],
 		     enum ipset_adt adt, u32 *lineno, u32 flags, bool retried)
 {
-	const struct hash_ipportnet4 *h = set->data;
+	struct hash_ipportnet4 *h = set->data;
 	ipset_adtfn adtfn = set->variant->adt[adt];
 	struct hash_ipportnet4_elem e = { .cidr = HOST_MASK - 1 };
 	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, i = 0;
 	bool with_ports = false;
 	u8 cidr;
 	int ret;
@@ -254,9 +254,6 @@ 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);
@@ -283,9 +280,15 @@ hash_ipportnet4_uadt(struct ip_set *set, struct nlattr *tb[],
 		for (; p <= port_to; p++) {
 			e.port = htons(p);
 			do {
+				i++;
 				e.ip2 = htonl(ip2);
 				ip2 = ip_set_range_to_cidr(ip2, ip2_to, &cidr);
 				e.cidr = cidr - 1;
+				if (i > IPSET_MAX_RANGE) {
+					hash_ipportnet4_data_next(&h->next,
+								  &e);
+					return -ERANGE;
+				}
 				ret = adtfn(set, &e, &ext, &ext, flags);
 
 				if (ret && !ip_set_eexist(ret, flags))
diff --git a/kernel/net/netfilter/ipset/ip_set_hash_net.c b/kernel/net/netfilter/ipset/ip_set_hash_net.c
index 5dfd0ed..8baceea 100644
--- a/kernel/net/netfilter/ipset/ip_set_hash_net.c
+++ b/kernel/net/netfilter/ipset/ip_set_hash_net.c
@@ -137,11 +137,11 @@ static int
 hash_net4_uadt(struct ip_set *set, struct nlattr *tb[],
 	       enum ipset_adt adt, u32 *lineno, u32 flags, bool retried)
 {
-	const struct hash_net4 *h = set->data;
+	struct hash_net4 *h = set->data;
 	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, ipn, n = 0;
+	u32 ip = 0, ip_to = 0, i = 0;
 	int ret;
 
 	if (tb[IPSET_ATTR_LINENO])
@@ -189,19 +189,16 @@ 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 {
+		i++;
 		e.ip = htonl(ip);
+		if (i > IPSET_MAX_RANGE) {
+			hash_net4_data_next(&h->next, &e);
+			return -ERANGE;
+		}
 		ip = ip_set_range_to_cidr(ip, ip_to, &e.cidr);
 		ret = adtfn(set, &e, &ext, &ext, flags);
 		if (ret && !ip_set_eexist(ret, flags))
diff --git a/kernel/net/netfilter/ipset/ip_set_hash_netiface.c b/kernel/net/netfilter/ipset/ip_set_hash_netiface.c
index b25400a..af210c7 100644
--- a/kernel/net/netfilter/ipset/ip_set_hash_netiface.c
+++ b/kernel/net/netfilter/ipset/ip_set_hash_netiface.c
@@ -203,7 +203,7 @@ 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, ipn, n = 0;
+	u32 ip = 0, ip_to = 0, i = 0;
 	int ret;
 
 	if (tb[IPSET_ATTR_LINENO])
@@ -257,19 +257,16 @@ 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);
 	do {
+		i++;
 		e.ip = htonl(ip);
+		if (i > IPSET_MAX_RANGE) {
+			hash_netiface4_data_next(&h->next, &e);
+			return -ERANGE;
+		}
 		ip = ip_set_range_to_cidr(ip, ip_to, &e.cidr);
 		ret = adtfn(set, &e, &ext, &ext, flags);
 
diff --git a/kernel/net/netfilter/ipset/ip_set_hash_netnet.c b/kernel/net/netfilter/ipset/ip_set_hash_netnet.c
index cdfb78c..8fbe649 100644
--- a/kernel/net/netfilter/ipset/ip_set_hash_netnet.c
+++ b/kernel/net/netfilter/ipset/ip_set_hash_netnet.c
@@ -166,13 +166,12 @@ static int
 hash_netnet4_uadt(struct ip_set *set, struct nlattr *tb[],
 		  enum ipset_adt adt, u32 *lineno, u32 flags, bool retried)
 {
-	const struct hash_netnet4 *h = set->data;
+	struct hash_netnet4 *h = set->data;
 	ipset_adtfn adtfn = set->variant->adt[adt];
 	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, ipn;
-	u64 n = 0, m = 0;
+	u32 ip2 = 0, ip2_from = 0, ip2_to = 0, i = 0;
 	int ret;
 
 	if (tb[IPSET_ATTR_LINENO])
@@ -248,19 +247,6 @@ 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]);
@@ -273,7 +259,12 @@ hash_netnet4_uadt(struct ip_set *set, struct nlattr *tb[],
 		e.ip[0] = htonl(ip);
 		ip = ip_set_range_to_cidr(ip, ip_to, &e.cidr[0]);
 		do {
+			i++;
 			e.ip[1] = htonl(ip2);
+			if (i > IPSET_MAX_RANGE) {
+				hash_netnet4_data_next(&h->next, &e);
+				return -ERANGE;
+			}
 			ip2 = ip_set_range_to_cidr(ip2, ip2_to, &e.cidr[1]);
 			ret = adtfn(set, &e, &ext, &ext, flags);
 			if (ret && !ip_set_eexist(ret, flags))
diff --git a/kernel/net/netfilter/ipset/ip_set_hash_netport.c b/kernel/net/netfilter/ipset/ip_set_hash_netport.c
index cf70324..8a2a7c5 100644
--- a/kernel/net/netfilter/ipset/ip_set_hash_netport.c
+++ b/kernel/net/netfilter/ipset/ip_set_hash_netport.c
@@ -155,12 +155,11 @@ static int
 hash_netport4_uadt(struct ip_set *set, struct nlattr *tb[],
 		   enum ipset_adt adt, u32 *lineno, u32 flags, bool retried)
 {
-	const struct hash_netport4 *h = set->data;
+	struct hash_netport4 *h = set->data;
 	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, ipn;
-	u64 n = 0;
+	u32 port, port_to, p = 0, ip = 0, ip_to = 0, i = 0;
 	bool with_ports = false;
 	u8 cidr;
 	int ret;
@@ -237,14 +236,6 @@ 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);
@@ -256,8 +247,12 @@ hash_netport4_uadt(struct ip_set *set, struct nlattr *tb[],
 		e.ip = htonl(ip);
 		ip = ip_set_range_to_cidr(ip, ip_to, &cidr);
 		e.cidr = cidr - 1;
-		for (; p <= port_to; p++) {
+		for (; p <= port_to; p++, i++) {
 			e.port = htons(p);
+			if (i > IPSET_MAX_RANGE) {
+				hash_netport4_data_next(&h->next, &e);
+				return -ERANGE;
+			}
 			ret = adtfn(set, &e, &ext, &ext, flags);
 			if (ret && !ip_set_eexist(ret, flags))
 				return ret;