[PATCH 1/5] net: Modify all rtnetlink methods to only work in the initial namespace

Subsystems: ethernet bridge, networking [general], networking [ipv4/ipv6], tc subsystem, the rest

STALE6883d

14 messages, 5 authors, 2007-10-10 · open the first message on its own page

[PATCH 1/5] net: Modify all rtnetlink methods to only work in the initial namespace

From: Eric W. Biederman <hidden>
Date: 2007-09-29 01:01:01 

Before I can enable rtnetlink to work in all network namespaces
I need to be certain that something won't break.  So this
patch deliberately disables all of the rtnletlink methods in everything
except the initial network namespace.  After the methods have been
audited this extra check can be disabled.

Signed-off-by: Eric W. Biederman <redacted>
---
 net/bridge/br_netlink.c |    9 +++++++++
 net/core/fib_rules.c    |   11 +++++++++++
 net/core/neighbour.c    |   18 ++++++++++++++++++
 net/core/rtnetlink.c    |   19 +++++++++++++++++++
 net/decnet/dn_dev.c     |   12 ++++++++++++
 net/decnet/dn_fib.c     |    8 ++++++++
 net/decnet/dn_route.c   |    8 ++++++++
 net/decnet/dn_table.c   |    4 ++++
 net/ipv4/devinet.c      |   12 ++++++++++++
 net/ipv4/fib_frontend.c |   12 ++++++++++++
 net/ipv4/route.c        |    4 ++++
 net/ipv6/addrconf.c     |   31 +++++++++++++++++++++++++++++++
 net/ipv6/ip6_fib.c      |    4 ++++
 net/ipv6/route.c        |   12 ++++++++++++
 net/sched/act_api.c     |   10 ++++++++++
 net/sched/cls_api.c     |   10 ++++++++++
 net/sched/sch_api.c     |   21 +++++++++++++++++++++
 17 files changed, 205 insertions(+), 0 deletions(-)

diff --git a/net/bridge/br_netlink.c b/net/bridge/br_netlink.c
index 53ab8e0..a4ffa2b 100644
--- a/net/bridge/br_netlink.c
+++ b/net/bridge/br_netlink.c
@@ -13,6 +13,7 @@
 #include <linux/kernel.h>
 #include <net/rtnetlink.h>
 #include <net/net_namespace.h>
+#include <net/sock.h>
 #include "br_private.h"
 
 static inline size_t br_nlmsg_size(void)
@@ -107,9 +108,13 @@ errout:
  */
 static int br_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	struct net_device *dev;
 	int idx;
 
+	if (net != &init_net)
+		return 0;
+
 	idx = 0;
 	for_each_netdev(&init_net, dev) {
 		/* not a bridge port */
@@ -135,12 +140,16 @@ skip:
  */
 static int br_rtm_setlink(struct sk_buff *skb,  struct nlmsghdr *nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct ifinfomsg *ifm;
 	struct nlattr *protinfo;
 	struct net_device *dev;
 	struct net_bridge_port *p;
 	u8 new_state;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	if (nlmsg_len(nlh) < sizeof(*ifm))
 		return -EINVAL;
 
diff --git a/net/core/fib_rules.c b/net/core/fib_rules.c
index 13de6f5..357bfa0 100644
--- a/net/core/fib_rules.c
+++ b/net/core/fib_rules.c
@@ -206,6 +206,9 @@ static int fib_nl_newrule(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
 	struct nlattr *tb[FRA_MAX+1];
 	int err = -EINVAL, unresolved = 0;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	if (nlh->nlmsg_len < nlmsg_msg_size(sizeof(*frh)))
 		goto errout;
 
@@ -336,12 +339,16 @@ errout:
 
 static int fib_nl_delrule(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct fib_rule_hdr *frh = nlmsg_data(nlh);
 	struct fib_rules_ops *ops = NULL;
 	struct fib_rule *rule, *tmp;
 	struct nlattr *tb[FRA_MAX+1];
 	int err = -EINVAL;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	if (nlh->nlmsg_len < nlmsg_msg_size(sizeof(*frh)))
 		goto errout;
 
@@ -517,9 +524,13 @@ skip:
 
 static int fib_nl_dumprule(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	struct fib_rules_ops *ops;
 	int idx = 0, family;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	family = rtnl_msg_family(cb->nlh);
 	if (family != AF_UNSPEC) {
 		/* Protocol specific dump request */
diff --git a/net/core/neighbour.c b/net/core/neighbour.c
index c52df85..27001db 100644
--- a/net/core/neighbour.c
+++ b/net/core/neighbour.c
@@ -1448,6 +1448,9 @@ static int neigh_delete(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 	struct net_device *dev = NULL;
 	int err = -EINVAL;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	if (nlmsg_len(nlh) < sizeof(*ndm))
 		goto out;
 
@@ -1514,6 +1517,9 @@ static int neigh_add(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 	struct net_device *dev = NULL;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = nlmsg_parse(nlh, sizeof(*ndm), tb, NDA_MAX, NULL);
 	if (err < 0)
 		goto out;
@@ -1788,11 +1794,15 @@ static const struct nla_policy nl_ntbl_parm_policy[NDTPA_MAX+1] = {
 
 static int neightbl_set(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct neigh_table *tbl;
 	struct ndtmsg *ndtmsg;
 	struct nlattr *tb[NDTA_MAX+1];
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = nlmsg_parse(nlh, sizeof(*ndtmsg), tb, NDTA_MAX,
 			  nl_neightbl_policy);
 	if (err < 0)
@@ -1912,11 +1922,15 @@ errout:
 
 static int neightbl_dump_info(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	int family, tidx, nidx = 0;
 	int tbl_skip = cb->args[0];
 	int neigh_skip = cb->args[1];
 	struct neigh_table *tbl;
 
+	if (net != &init_net)
+		return 0;
+
 	family = ((struct rtgenmsg *) nlmsg_data(cb->nlh))->rtgen_family;
 
 	read_lock(&neigh_tbl_lock);
@@ -2041,9 +2055,13 @@ out:
 
 static int neigh_dump_info(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	struct neigh_table *tbl;
 	int t, family, s_t;
 
+	if (net != &init_net)
+		return 0;
+
 	read_lock(&neigh_tbl_lock);
 	family = ((struct rtgenmsg *) nlmsg_data(cb->nlh))->rtgen_family;
 	s_t = cb->args[0];
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index 8bc68e6..56fc4f9 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -707,6 +707,9 @@ static int rtnl_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
 	int s_idx = cb->args[0];
 	struct net_device *dev;
 
+	if (net != &init_net)
+		return 0;
+
 	idx = 0;
 	for_each_netdev(net, dev) {
 		if (idx < s_idx)
@@ -909,6 +912,9 @@ static int rtnl_setlink(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 	struct nlattr *tb[IFLA_MAX+1];
 	char ifname[IFNAMSIZ];
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFLA_MAX, ifla_policy);
 	if (err < 0)
 		goto errout;
@@ -957,6 +963,9 @@ static int rtnl_dellink(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 	struct nlattr *tb[IFLA_MAX+1];
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFLA_MAX, ifla_policy);
 	if (err < 0)
 		return err;
@@ -1038,6 +1047,9 @@ static int rtnl_newlink(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 	struct nlattr *linkinfo[IFLA_INFO_MAX+1];
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 #ifdef CONFIG_KMOD
 replay:
 #endif
@@ -1164,6 +1176,9 @@ static int rtnl_getlink(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
 	struct sk_buff *nskb;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFLA_MAX, ifla_policy);
 	if (err < 0)
 		return err;
@@ -1199,9 +1214,13 @@ errout:
 
 static int rtnl_dump_all(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	int idx;
 	int s_idx = cb->family;
 
+	if (net != &init_net)
+		return 0;
+
 	if (s_idx == 0)
 		s_idx = 1;
 	for (idx=1; idx<NPROTO; idx++) {
diff --git a/net/decnet/dn_dev.c b/net/decnet/dn_dev.c
index 26130af..6e82f7a 100644
--- a/net/decnet/dn_dev.c
+++ b/net/decnet/dn_dev.c
@@ -647,12 +647,16 @@ static const struct nla_policy dn_ifa_policy[IFA_MAX+1] = {
 
 static int dn_nl_deladdr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct nlattr *tb[IFA_MAX+1];
 	struct dn_dev *dn_db;
 	struct ifaddrmsg *ifm;
 	struct dn_ifaddr *ifa, **ifap;
 	int err = -EADDRNOTAVAIL;
 
+	if (net != &init_net)
+		goto errout;
+
 	err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, dn_ifa_policy);
 	if (err < 0)
 		goto errout;
@@ -679,6 +683,7 @@ errout:
 
 static int dn_nl_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct nlattr *tb[IFA_MAX+1];
 	struct net_device *dev;
 	struct dn_dev *dn_db;
@@ -686,6 +691,9 @@ static int dn_nl_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 	struct dn_ifaddr *ifa;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, dn_ifa_policy);
 	if (err < 0)
 		return err;
@@ -791,11 +799,15 @@ errout:
 
 static int dn_nl_dump_ifaddr(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	int idx, dn_idx = 0, skip_ndevs, skip_naddr;
 	struct net_device *dev;
 	struct dn_dev *dn_db;
 	struct dn_ifaddr *ifa;
 
+	if (net != &init_net)
+		return 0;
+
 	skip_ndevs = cb->args[0];
 	skip_naddr = cb->args[1];
 
diff --git a/net/decnet/dn_fib.c b/net/decnet/dn_fib.c
index 3760a20..5413e1b 100644
--- a/net/decnet/dn_fib.c
+++ b/net/decnet/dn_fib.c
@@ -506,10 +506,14 @@ static int dn_fib_check_attr(struct rtmsg *r, struct rtattr **rta)
 
 static int dn_fib_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct dn_fib_table *tb;
 	struct rtattr **rta = arg;
 	struct rtmsg *r = NLMSG_DATA(nlh);
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	if (dn_fib_check_attr(r, rta))
 		return -EINVAL;
 
@@ -522,10 +526,14 @@ static int dn_fib_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh, void *
 
 static int dn_fib_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct dn_fib_table *tb;
 	struct rtattr **rta = arg;
 	struct rtmsg *r = NLMSG_DATA(nlh);
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	if (dn_fib_check_attr(r, rta))
 		return -EINVAL;
 
diff --git a/net/decnet/dn_route.c b/net/decnet/dn_route.c
index b7ebf99..a16374b 100644
--- a/net/decnet/dn_route.c
+++ b/net/decnet/dn_route.c
@@ -1530,6 +1530,7 @@ rtattr_failure:
  */
 static int dn_cache_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh, void *arg)
 {
+	struct net *net = in_skb->sk->sk_net;
 	struct rtattr **rta = arg;
 	struct rtmsg *rtm = NLMSG_DATA(nlh);
 	struct dn_route *rt = NULL;
@@ -1538,6 +1539,9 @@ static int dn_cache_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh, void
 	struct sk_buff *skb;
 	struct flowi fl;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	memset(&fl, 0, sizeof(fl));
 	fl.proto = DNPROTO_NSP;
 
@@ -1615,10 +1619,14 @@ out_free:
  */
 int dn_cache_dump(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	struct dn_route *rt;
 	int h, s_h;
 	int idx, s_idx;
 
+	if (net != &init_net)
+		return 0;
+
 	if (NLMSG_PAYLOAD(cb->nlh, 0) < sizeof(struct rtmsg))
 		return -EINVAL;
 	if (!(((struct rtmsg *)NLMSG_DATA(cb->nlh))->rtm_flags&RTM_F_CLONED))
diff --git a/net/decnet/dn_table.c b/net/decnet/dn_table.c
index fda0772..a3bdb8d 100644
--- a/net/decnet/dn_table.c
+++ b/net/decnet/dn_table.c
@@ -463,12 +463,16 @@ static int dn_fib_table_dump(struct dn_fib_table *tb, struct sk_buff *skb,
 
 int dn_fib_dump(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	unsigned int h, s_h;
 	unsigned int e = 0, s_e;
 	struct dn_fib_table *tb;
 	struct hlist_node *node;
 	int dumped = 0;
 
+	if (net != &init_net)
+		return 0;
+
 	if (NLMSG_PAYLOAD(cb->nlh, 0) >= sizeof(struct rtmsg) &&
 		((struct rtmsg *)NLMSG_DATA(cb->nlh))->rtm_flags&RTM_F_CLONED)
 			return dn_cache_dump(skb, cb);
diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 55d199e..0ff2ef6 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -441,6 +441,7 @@ struct in_ifaddr *inet_ifa_byprefix(struct in_device *in_dev, __be32 prefix,
 
 static int inet_rtm_deladdr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct nlattr *tb[IFA_MAX+1];
 	struct in_device *in_dev;
 	struct ifaddrmsg *ifm;
@@ -449,6 +450,9 @@ static int inet_rtm_deladdr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg
 
 	ASSERT_RTNL();
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv4_policy);
 	if (err < 0)
 		goto errout;
@@ -561,10 +565,14 @@ errout:
 
 static int inet_rtm_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct in_ifaddr *ifa;
 
 	ASSERT_RTNL();
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	ifa = rtm_to_ifaddr(nlh);
 	if (IS_ERR(ifa))
 		return PTR_ERR(ifa);
@@ -1175,12 +1183,16 @@ nla_put_failure:
 
 static int inet_dump_ifaddr(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	int idx, ip_idx;
 	struct net_device *dev;
 	struct in_device *in_dev;
 	struct in_ifaddr *ifa;
 	int s_ip_idx, s_idx = cb->args[0];
 
+	if (net != &init_net)
+		return 0;
+
 	s_ip_idx = ip_idx = cb->args[1];
 	idx = 0;
 	for_each_netdev(&init_net, dev) {
diff --git a/net/ipv4/fib_frontend.c b/net/ipv4/fib_frontend.c
index f823ca3..84d688a 100644
--- a/net/ipv4/fib_frontend.c
+++ b/net/ipv4/fib_frontend.c
@@ -527,10 +527,14 @@ errout:
 
 static int inet_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct fib_config cfg;
 	struct fib_table *tb;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = rtm_to_fib_config(skb, nlh, &cfg);
 	if (err < 0)
 		goto errout;
@@ -548,10 +552,14 @@ errout:
 
 static int inet_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct fib_config cfg;
 	struct fib_table *tb;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = rtm_to_fib_config(skb, nlh, &cfg);
 	if (err < 0)
 		goto errout;
@@ -569,12 +577,16 @@ errout:
 
 static int inet_dump_fib(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	unsigned int h, s_h;
 	unsigned int e = 0, s_e;
 	struct fib_table *tb;
 	struct hlist_node *node;
 	int dumped = 0;
 
+	if (net != &init_net)
+		return 0;
+
 	if (nlmsg_len(cb->nlh) >= sizeof(struct rtmsg) &&
 	    ((struct rtmsg *) nlmsg_data(cb->nlh))->rtm_flags & RTM_F_CLONED)
 		return ip_rt_dump(skb, cb);
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index c5a06e6..a538a30 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -2552,6 +2552,7 @@ nla_put_failure:
 
 static int inet_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void *arg)
 {
+	struct net *net = in_skb->sk->sk_net;
 	struct rtmsg *rtm;
 	struct nlattr *tb[RTA_MAX+1];
 	struct rtable *rt = NULL;
@@ -2561,6 +2562,9 @@ static int inet_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void
 	int err;
 	struct sk_buff *skb;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv4_policy);
 	if (err < 0)
 		goto errout;
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 6d5c3c2..bdc183e 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -3003,11 +3003,15 @@ static const struct nla_policy ifa_ipv6_policy[IFA_MAX+1] = {
 static int
 inet6_rtm_deladdr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct ifaddrmsg *ifm;
 	struct nlattr *tb[IFA_MAX+1];
 	struct in6_addr *pfx;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
 	if (err < 0)
 		return err;
@@ -3060,6 +3064,7 @@ static int inet6_addr_modify(struct inet6_ifaddr *ifp, u8 ifa_flags,
 static int
 inet6_rtm_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct ifaddrmsg *ifm;
 	struct nlattr *tb[IFA_MAX+1];
 	struct in6_addr *pfx;
@@ -3069,6 +3074,9 @@ inet6_rtm_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
 	u8 ifa_flags;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
 	if (err < 0)
 		return err;
@@ -3342,26 +3350,42 @@ done:
 
 static int inet6_dump_ifaddr(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	enum addr_type_t type = UNICAST_ADDR;
+
+	if (net != &init_net)
+		return 0;
+
 	return inet6_dump_addr(skb, cb, type);
 }
 
 static int inet6_dump_ifmcaddr(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	enum addr_type_t type = MULTICAST_ADDR;
+
+	if (net != &init_net)
+		return 0;
+
 	return inet6_dump_addr(skb, cb, type);
 }
 
 
 static int inet6_dump_ifacaddr(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	enum addr_type_t type = ANYCAST_ADDR;
+
+	if (net != &init_net)
+		return 0;
+
 	return inet6_dump_addr(skb, cb, type);
 }
 
 static int inet6_rtm_getaddr(struct sk_buff *in_skb, struct nlmsghdr* nlh,
 			     void *arg)
 {
+	struct net *net = in_skb->sk->sk_net;
 	struct ifaddrmsg *ifm;
 	struct nlattr *tb[IFA_MAX+1];
 	struct in6_addr *addr = NULL;
@@ -3370,6 +3394,9 @@ static int inet6_rtm_getaddr(struct sk_buff *in_skb, struct nlmsghdr* nlh,
 	struct sk_buff *skb;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv6_policy);
 	if (err < 0)
 		goto errout;
@@ -3587,11 +3614,15 @@ nla_put_failure:
 
 static int inet6_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	int idx, err;
 	int s_idx = cb->args[0];
 	struct net_device *dev;
 	struct inet6_dev *idev;
 
+	if (net != &init_net)
+		return 0;
+
 	read_lock(&dev_base_lock);
 	idx = 0;
 	for_each_netdev(&init_net, dev) {
diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c
index 6a612a7..a731812 100644
--- a/net/ipv6/ip6_fib.c
+++ b/net/ipv6/ip6_fib.c
@@ -361,6 +361,7 @@ end:
 
 static int inet6_dump_fib(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	unsigned int h, s_h;
 	unsigned int e = 0, s_e;
 	struct rt6_rtnl_dump_arg arg;
@@ -369,6 +370,9 @@ static int inet6_dump_fib(struct sk_buff *skb, struct netlink_callback *cb)
 	struct hlist_node *node;
 	int res = 0;
 
+	if (net != &init_net)
+		return 0;
+
 	s_h = cb->args[0];
 	s_e = cb->args[1];
 
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index a7db84c..f0cf11c 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -2080,9 +2080,13 @@ errout:
 
 static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct fib6_config cfg;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = rtm_to_fib6_config(skb, nlh, &cfg);
 	if (err < 0)
 		return err;
@@ -2092,9 +2096,13 @@ static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *a
 
 static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct fib6_config cfg;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = rtm_to_fib6_config(skb, nlh, &cfg);
 	if (err < 0)
 		return err;
@@ -2229,6 +2237,7 @@ int rt6_dump_route(struct rt6_info *rt, void *p_arg)
 
 static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void *arg)
 {
+	struct net *net = in_skb->sk->sk_net;
 	struct nlattr *tb[RTA_MAX+1];
 	struct rt6_info *rt;
 	struct sk_buff *skb;
@@ -2236,6 +2245,9 @@ static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void
 	struct flowi fl;
 	int err, iif = 0;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
 	if (err < 0)
 		goto errout;
diff --git a/net/sched/act_api.c b/net/sched/act_api.c
index 72cdb0f..8528291 100644
--- a/net/sched/act_api.c
+++ b/net/sched/act_api.c
@@ -18,6 +18,8 @@
 #include <linux/skbuff.h>
 #include <linux/init.h>
 #include <linux/kmod.h>
+#include <net/net_namespace.h>
+#include <net/sock.h>
 #include <net/sch_generic.h>
 #include <net/act_api.h>
 #include <net/netlink.h>
@@ -924,10 +926,14 @@ done:
 
 static int tc_ctl_action(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct rtattr **tca = arg;
 	u32 pid = skb ? NETLINK_CB(skb).pid : 0;
 	int ret = 0, ovr = 0;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	if (tca[TCA_ACT_TAB-1] == NULL) {
 		printk("tc_ctl_action: received NO action attribs\n");
 		return -EINVAL;
@@ -997,6 +1003,7 @@ find_dump_kind(struct nlmsghdr *n)
 static int
 tc_dump_action(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	struct nlmsghdr *nlh;
 	unsigned char *b = skb_tail_pointer(skb);
 	struct rtattr *x;
@@ -1006,6 +1013,9 @@ tc_dump_action(struct sk_buff *skb, struct netlink_callback *cb)
 	struct tcamsg *t = (struct tcamsg *) NLMSG_DATA(cb->nlh);
 	struct rtattr *kind = find_dump_kind(cb->nlh);
 
+	if (net != &init_net)
+		return 0;
+
 	if (kind == NULL) {
 		printk("tc_dump_action: action bad kind\n");
 		return 0;
diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
index 0365797..2754e50 100644
--- a/net/sched/cls_api.c
+++ b/net/sched/cls_api.c
@@ -23,6 +23,8 @@
 #include <linux/init.h>
 #include <linux/kmod.h>
 #include <linux/netlink.h>
+#include <net/net_namespace.h>
+#include <net/sock.h>
 #include <net/netlink.h>
 #include <net/pkt_sched.h>
 #include <net/pkt_cls.h>
@@ -119,6 +121,7 @@ static __inline__ u32 tcf_auto_prio(struct tcf_proto *tp)
 
 static int tc_ctl_tfilter(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct rtattr **tca;
 	struct tcmsg *t;
 	u32 protocol;
@@ -135,6 +138,9 @@ static int tc_ctl_tfilter(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
 	unsigned long fh;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 replay:
 	tca = arg;
 	t = NLMSG_DATA(n);
@@ -375,6 +381,7 @@ static int tcf_node_dump(struct tcf_proto *tp, unsigned long n, struct tcf_walke
 
 static int tc_dump_tfilter(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	int t;
 	int s_t;
 	struct net_device *dev;
@@ -385,6 +392,9 @@ static int tc_dump_tfilter(struct sk_buff *skb, struct netlink_callback *cb)
 	struct Qdisc_class_ops *cops;
 	struct tcf_dump_args arg;
 
+	if (net != &init_net)
+		return 0;
+
 	if (cb->nlh->nlmsg_len < NLMSG_LENGTH(sizeof(*tcm)))
 		return skb->len;
 	if ((dev = dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index 39d3278..95eb0a8 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -29,6 +29,7 @@
 #include <linux/hrtimer.h>
 
 #include <net/net_namespace.h>
+#include <net/sock.h>
 #include <net/netlink.h>
 #include <net/pkt_sched.h>
 
@@ -599,6 +600,7 @@ check_loop_fn(struct Qdisc *q, unsigned long cl, struct qdisc_walker *w)
 
 static int tc_get_qdisc(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct tcmsg *tcm = NLMSG_DATA(n);
 	struct rtattr **tca = arg;
 	struct net_device *dev;
@@ -607,6 +609,9 @@ static int tc_get_qdisc(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
 	struct Qdisc *p = NULL;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	if ((dev = __dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
 		return -ENODEV;
 
@@ -660,6 +665,7 @@ static int tc_get_qdisc(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
 
 static int tc_modify_qdisc(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct tcmsg *tcm;
 	struct rtattr **tca;
 	struct net_device *dev;
@@ -667,6 +673,9 @@ static int tc_modify_qdisc(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
 	struct Qdisc *q, *p;
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 replay:
 	/* Reinit, just in case something touches this. */
 	tcm = NLMSG_DATA(n);
@@ -872,11 +881,15 @@ err_out:
 
 static int tc_dump_qdisc(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	int idx, q_idx;
 	int s_idx, s_q_idx;
 	struct net_device *dev;
 	struct Qdisc *q;
 
+	if (net != &init_net)
+		return 0;
+
 	s_idx = cb->args[0];
 	s_q_idx = q_idx = cb->args[1];
 	read_lock(&dev_base_lock);
@@ -920,6 +933,7 @@ done:
 
 static int tc_ctl_tclass(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
 {
+	struct net *net = skb->sk->sk_net;
 	struct tcmsg *tcm = NLMSG_DATA(n);
 	struct rtattr **tca = arg;
 	struct net_device *dev;
@@ -932,6 +946,9 @@ static int tc_ctl_tclass(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
 	u32 qid = TC_H_MAJ(clid);
 	int err;
 
+	if (net != &init_net)
+		return -EINVAL;
+
 	if ((dev = __dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
 		return -ENODEV;
 
@@ -1106,6 +1123,7 @@ static int qdisc_class_dump(struct Qdisc *q, unsigned long cl, struct qdisc_walk
 
 static int tc_dump_tclass(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	int t;
 	int s_t;
 	struct net_device *dev;
@@ -1113,6 +1131,9 @@ static int tc_dump_tclass(struct sk_buff *skb, struct netlink_callback *cb)
 	struct tcmsg *tcm = (struct tcmsg*)NLMSG_DATA(cb->nlh);
 	struct qdisc_dump_args arg;
 
+	if (net != &init_net)
+		return 0;
+
 	if (cb->nlh->nlmsg_len < NLMSG_LENGTH(sizeof(*tcm)))
 		return 0;
 	if ((dev = dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
-- 
1.5.3.rc6.17.g1911

[PATCH 2/5] net: Make rtnetlink infrastructure network namespace aware

From: Eric W. Biederman <hidden>
Date: 2007-09-29 01:02:32 

After this patch none of the netlink callback support anything
except the initial network namespace but the rtnetlink infrastructure
now handles multiple network namespaces.

Signed-off-by: Eric W. Biederman <redacted>
---
 include/linux/rtnetlink.h   |    8 ++--
 include/net/net_namespace.h |    3 +
 net/bridge/br_netlink.c     |    4 +-
 net/core/fib_rules.c        |    4 +-
 net/core/neighbour.c        |    4 +-
 net/core/rtnetlink.c        |   96 +++++++++++++++++++++++++++++++++++++------
 net/decnet/dn_dev.c         |    4 +-
 net/decnet/dn_route.c       |    2 +-
 net/decnet/dn_table.c       |    4 +-
 net/ipv4/devinet.c          |    4 +-
 net/ipv4/fib_semantics.c    |    4 +-
 net/ipv4/ipmr.c             |    4 +-
 net/ipv4/route.c            |    2 +-
 net/ipv6/addrconf.c         |   14 +++---
 net/ipv6/route.c            |    6 +-
 net/sched/act_api.c         |    8 ++--
 net/sched/cls_api.c         |    2 +-
 net/sched/sch_api.c         |    4 +-
 net/wireless/wext.c         |    5 ++-
 19 files changed, 129 insertions(+), 53 deletions(-)

diff --git a/include/linux/rtnetlink.h b/include/linux/rtnetlink.h
index 9c21e45..518247e 100644
--- a/include/linux/rtnetlink.h
+++ b/include/linux/rtnetlink.h
@@ -582,11 +582,11 @@ extern int __rtattr_parse_nested_compat(struct rtattr *tb[], int maxattr,
 ({	data = RTA_PAYLOAD(rta) >= len ? RTA_DATA(rta) : NULL; \
 	__rtattr_parse_nested_compat(tb, max, rta, len); })
 
-extern int rtnetlink_send(struct sk_buff *skb, u32 pid, u32 group, int echo);
-extern int rtnl_unicast(struct sk_buff *skb, u32 pid);
-extern int rtnl_notify(struct sk_buff *skb, u32 pid, u32 group,
+extern int rtnetlink_send(struct sk_buff *skb, struct net *net, u32 pid, u32 group, int echo);
+extern int rtnl_unicast(struct sk_buff *skb, struct net *net, u32 pid);
+extern int rtnl_notify(struct sk_buff *skb, struct net *net, u32 pid, u32 group,
 		       struct nlmsghdr *nlh, gfp_t flags);
-extern void rtnl_set_sk_err(u32 group, int error);
+extern void rtnl_set_sk_err(struct net *net, u32 group, int error);
 extern int rtnetlink_put_metrics(struct sk_buff *skb, u32 *metrics);
 extern int rtnl_put_cacheinfo(struct sk_buff *skb, struct dst_entry *dst,
 			      u32 id, u32 ts, u32 tsage, long expires,
diff --git a/include/net/net_namespace.h b/include/net/net_namespace.h
index 934c840..f75607a 100644
--- a/include/net/net_namespace.h
+++ b/include/net/net_namespace.h
@@ -10,6 +10,7 @@
 
 struct proc_dir_entry;
 struct net_device;
+struct sock;
 struct net {
 	atomic_t		count;		/* To decided when the network
 						 *  namespace should be freed.
@@ -29,6 +30,8 @@ struct net {
 	struct list_head 	dev_base_head;
 	struct hlist_head 	*dev_name_head;
 	struct hlist_head	*dev_index_head;
+
+	struct sock 		*rtnl;			/* rtnetlink socket */
 };
 
 #ifdef CONFIG_NET
diff --git a/net/bridge/br_netlink.c b/net/bridge/br_netlink.c
index a4ffa2b..f5d6933 100644
--- a/net/bridge/br_netlink.c
+++ b/net/bridge/br_netlink.c
@@ -97,10 +97,10 @@ void br_ifinfo_notify(int event, struct net_bridge_port *port)
 		kfree_skb(skb);
 		goto errout;
 	}
-	err = rtnl_notify(skb, 0, RTNLGRP_LINK, NULL, GFP_ATOMIC);
+	err = rtnl_notify(skb, &init_net,0, RTNLGRP_LINK, NULL, GFP_ATOMIC);
 errout:
 	if (err < 0)
-		rtnl_set_sk_err(RTNLGRP_LINK, err);
+		rtnl_set_sk_err(&init_net, RTNLGRP_LINK, err);
 }
 
 /*
diff --git a/net/core/fib_rules.c b/net/core/fib_rules.c
index 357bfa0..03c803c 100644
--- a/net/core/fib_rules.c
+++ b/net/core/fib_rules.c
@@ -577,10 +577,10 @@ static void notify_rule_change(int event, struct fib_rule *rule,
 		kfree_skb(skb);
 		goto errout;
 	}
-	err = rtnl_notify(skb, pid, ops->nlgroup, nlh, GFP_KERNEL);
+	err = rtnl_notify(skb, &init_net, pid, ops->nlgroup, nlh, GFP_KERNEL);
 errout:
 	if (err < 0)
-		rtnl_set_sk_err(ops->nlgroup, err);
+		rtnl_set_sk_err(&init_net, ops->nlgroup, err);
 }
 
 static void attach_rules(struct list_head *rules, struct net_device *dev)
diff --git a/net/core/neighbour.c b/net/core/neighbour.c
index 27001db..c452584 100644
--- a/net/core/neighbour.c
+++ b/net/core/neighbour.c
@@ -2466,10 +2466,10 @@ static void __neigh_notify(struct neighbour *n, int type, int flags)
 		kfree_skb(skb);
 		goto errout;
 	}
-	err = rtnl_notify(skb, 0, RTNLGRP_NEIGH, NULL, GFP_ATOMIC);
+	err = rtnl_notify(skb, &init_net, 0, RTNLGRP_NEIGH, NULL, GFP_ATOMIC);
 errout:
 	if (err < 0)
-		rtnl_set_sk_err(RTNLGRP_NEIGH, err);
+		rtnl_set_sk_err(&init_net, RTNLGRP_NEIGH, err);
 }
 
 #ifdef CONFIG_ARPD
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index 56fc4f9..fc49104 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -60,7 +60,6 @@ struct rtnl_link
 };
 
 static DEFINE_MUTEX(rtnl_mutex);
-static struct sock *rtnl;
 
 void rtnl_lock(void)
 {
@@ -74,9 +73,38 @@ void __rtnl_unlock(void)
 
 void rtnl_unlock(void)
 {
-	mutex_unlock(&rtnl_mutex);
-	if (rtnl && rtnl->sk_receive_queue.qlen)
+	struct net *net;
+
+	/*
+	 * Loop through all of the rtnl sockets until none of them (in
+	 * a live network namespace) have queue packets.
+	 *
+	 * We have to be careful with the locking here as
+	 * sk_data_ready aka rtnetlink_rcv takes the rtnl_mutex.
+	 *
+	 * To ensure the network namespace does not exit while
+	 * we are processing packets on it's rtnl socket we
+	 * grab a reference to the network namespace, ignoring
+	 * it if the network namespace has already exited.
+	 */
+retry:
+	for_each_net(net) {
+		struct sock *rtnl = net->rtnl;
+
+		if (!rtnl || !rtnl->sk_receive_queue.qlen)
+			continue;
+
+		if (!maybe_get_net(net))
+			continue;
+
+		mutex_unlock(&rtnl_mutex);
 		rtnl->sk_data_ready(rtnl, 0);
+		mutex_lock(&rtnl_mutex);
+		put_net(net);
+		goto retry;
+	}
+	mutex_unlock(&rtnl_mutex);
+
 	netdev_run_todo();
 }
 
@@ -462,8 +490,9 @@ size_t rtattr_strlcpy(char *dest, const struct rtattr *rta, size_t size)
 	return ret;
 }
 
-int rtnetlink_send(struct sk_buff *skb, u32 pid, unsigned group, int echo)
+int rtnetlink_send(struct sk_buff *skb, struct net *net, u32 pid, unsigned group, int echo)
 {
+	struct sock *rtnl = net->rtnl;
 	int err = 0;
 
 	NETLINK_CB(skb).dst_group = group;
@@ -475,14 +504,17 @@ int rtnetlink_send(struct sk_buff *skb, u32 pid, unsigned group, int echo)
 	return err;
 }
 
-int rtnl_unicast(struct sk_buff *skb, u32 pid)
+int rtnl_unicast(struct sk_buff *skb, struct net *net, u32 pid)
 {
+	struct sock *rtnl = net->rtnl;
+
 	return nlmsg_unicast(rtnl, skb, pid);
 }
 
-int rtnl_notify(struct sk_buff *skb, u32 pid, u32 group,
+int rtnl_notify(struct sk_buff *skb, struct net *net, u32 pid, u32 group,
 		struct nlmsghdr *nlh, gfp_t flags)
 {
+	struct sock *rtnl = net->rtnl;
 	int report = 0;
 
 	if (nlh)
@@ -491,8 +523,10 @@ int rtnl_notify(struct sk_buff *skb, u32 pid, u32 group,
 	return nlmsg_notify(rtnl, skb, pid, group, report, flags);
 }
 
-void rtnl_set_sk_err(u32 group, int error)
+void rtnl_set_sk_err(struct net *net, u32 group, int error)
 {
+	struct sock *rtnl = net->rtnl;
+
 	netlink_set_err(rtnl, 0, group, error);
 }
 
@@ -1205,7 +1239,7 @@ static int rtnl_getlink(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
 		kfree_skb(nskb);
 		goto errout;
 	}
-	err = rtnl_unicast(nskb, NETLINK_CB(skb).pid);
+	err = rtnl_unicast(nskb, net, NETLINK_CB(skb).pid);
 errout:
 	dev_put(dev);
 
@@ -1256,10 +1290,10 @@ void rtmsg_ifinfo(int type, struct net_device *dev, unsigned change)
 		kfree_skb(skb);
 		goto errout;
 	}
-	err = rtnl_notify(skb, 0, RTNLGRP_LINK, NULL, GFP_KERNEL);
+	err = rtnl_notify(skb, &init_net, 0, RTNLGRP_LINK, NULL, GFP_KERNEL);
 errout:
 	if (err < 0)
-		rtnl_set_sk_err(RTNLGRP_LINK, err);
+		rtnl_set_sk_err(&init_net, RTNLGRP_LINK, err);
 }
 
 /* Protected by RTNL sempahore.  */
@@ -1270,6 +1304,7 @@ static int rtattr_max;
 
 static int rtnetlink_rcv_msg(struct sk_buff *skb, struct nlmsghdr *nlh)
 {
+	struct net *net = skb->sk->sk_net;
 	rtnl_doit_func doit;
 	int sz_idx, kind;
 	int min_len;
@@ -1298,6 +1333,7 @@ static int rtnetlink_rcv_msg(struct sk_buff *skb, struct nlmsghdr *nlh)
 		return -EPERM;
 
 	if (kind == 2 && nlh->nlmsg_flags&NLM_F_DUMP) {
+		struct sock *rtnl;
 		rtnl_dumpit_func dumpit;
 
 		dumpit = rtnl_get_dumpit(family, type);
@@ -1305,6 +1341,7 @@ static int rtnetlink_rcv_msg(struct sk_buff *skb, struct nlmsghdr *nlh)
 			return -EOPNOTSUPP;
 
 		__rtnl_unlock();
+		rtnl = net->rtnl;
 		err = netlink_dump_start(rtnl, skb, nlh, dumpit, NULL);
 		rtnl_lock();
 		return err;
@@ -1383,6 +1420,40 @@ static struct notifier_block rtnetlink_dev_notifier = {
 	.notifier_call	= rtnetlink_event,
 };
 
+
+static int rtnetlink_net_init(struct net *net)
+{
+	struct sock *sk;
+	sk = netlink_kernel_create(net, NETLINK_ROUTE, RTNLGRP_MAX,
+				   rtnetlink_rcv, &rtnl_mutex, THIS_MODULE);
+	if (!sk)
+		return -ENOMEM;
+
+	/* Don't hold an extra reference on the namespace */
+	put_net(sk->sk_net);
+	net->rtnl = sk;
+	return 0;
+}
+
+static void rtnetlink_net_exit(struct net *net)
+{
+	struct sock *sk = net->rtnl;
+	if (sk) {
+		/* At the last minute lie and say this is a socket for the
+		 * initial network namespace.  So the socket will be safe to
+		 * free.
+		 */
+		sk->sk_net = get_net(&init_net);
+		sock_put(sk);
+		net->rtnl = NULL;
+	}
+}
+
+static struct pernet_operations rtnetlink_net_ops = {
+	.init = rtnetlink_net_init,
+	.exit = rtnetlink_net_exit,
+};
+
 void __init rtnetlink_init(void)
 {
 	int i;
@@ -1395,10 +1466,9 @@ void __init rtnetlink_init(void)
 	if (!rta_buf)
 		panic("rtnetlink_init: cannot allocate rta_buf\n");
 
-	rtnl = netlink_kernel_create(&init_net, NETLINK_ROUTE, RTNLGRP_MAX,
-				     rtnetlink_rcv, &rtnl_mutex, THIS_MODULE);
-	if (rtnl == NULL)
+	if (register_pernet_subsys(&rtnetlink_net_ops))
 		panic("rtnetlink_init: cannot initialize rtnetlink\n");
+
 	netlink_set_nonroot(NETLINK_ROUTE, NL_NONROOT_RECV);
 	register_netdevice_notifier(&rtnetlink_dev_notifier);
 
diff --git a/net/decnet/dn_dev.c b/net/decnet/dn_dev.c
index 6e82f7a..8b34b24 100644
--- a/net/decnet/dn_dev.c
+++ b/net/decnet/dn_dev.c
@@ -791,10 +791,10 @@ static void dn_ifaddr_notify(int event, struct dn_ifaddr *ifa)
 		kfree_skb(skb);
 		goto errout;
 	}
-	err = rtnl_notify(skb, 0, RTNLGRP_DECnet_IFADDR, NULL, GFP_KERNEL);
+	err = rtnl_notify(skb, &init_net, 0, RTNLGRP_DECnet_IFADDR, NULL, GFP_KERNEL);
 errout:
 	if (err < 0)
-		rtnl_set_sk_err(RTNLGRP_DECnet_IFADDR, err);
+		rtnl_set_sk_err(&init_net, RTNLGRP_DECnet_IFADDR, err);
 }
 
 static int dn_nl_dump_ifaddr(struct sk_buff *skb, struct netlink_callback *cb)
diff --git a/net/decnet/dn_route.c b/net/decnet/dn_route.c
index a16374b..a5df108 100644
--- a/net/decnet/dn_route.c
+++ b/net/decnet/dn_route.c
@@ -1606,7 +1606,7 @@ static int dn_cache_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh, void
 		goto out_free;
 	}
 
-	return rtnl_unicast(skb, NETLINK_CB(in_skb).pid);
+	return rtnl_unicast(skb, &init_net, NETLINK_CB(in_skb).pid);
 
 out_free:
 	kfree_skb(skb);
diff --git a/net/decnet/dn_table.c b/net/decnet/dn_table.c
index a3bdb8d..e09d915 100644
--- a/net/decnet/dn_table.c
+++ b/net/decnet/dn_table.c
@@ -375,10 +375,10 @@ static void dn_rtmsg_fib(int event, struct dn_fib_node *f, int z, u32 tb_id,
 		kfree_skb(skb);
 		goto errout;
 	}
-	err = rtnl_notify(skb, pid, RTNLGRP_DECnet_ROUTE, nlh, GFP_KERNEL);
+	err = rtnl_notify(skb, &init_net, pid, RTNLGRP_DECnet_ROUTE, nlh, GFP_KERNEL);
 errout:
 	if (err < 0)
-		rtnl_set_sk_err(RTNLGRP_DECnet_ROUTE, err);
+		rtnl_set_sk_err(&init_net, RTNLGRP_DECnet_ROUTE, err);
 }
 
 static __inline__ int dn_hash_dump_bucket(struct sk_buff *skb,
diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 0ff2ef6..1ae7dcf 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -1241,10 +1241,10 @@ static void rtmsg_ifa(int event, struct in_ifaddr* ifa, struct nlmsghdr *nlh,
 		kfree_skb(skb);
 		goto errout;
 	}
-	err = rtnl_notify(skb, pid, RTNLGRP_IPV4_IFADDR, nlh, GFP_KERNEL);
+	err = rtnl_notify(skb, &init_net, pid, RTNLGRP_IPV4_IFADDR, nlh, GFP_KERNEL);
 errout:
 	if (err < 0)
-		rtnl_set_sk_err(RTNLGRP_IPV4_IFADDR, err);
+		rtnl_set_sk_err(&init_net, RTNLGRP_IPV4_IFADDR, err);
 }
 
 #ifdef CONFIG_SYSCTL
diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c
index 1351a26..33ec960 100644
--- a/net/ipv4/fib_semantics.c
+++ b/net/ipv4/fib_semantics.c
@@ -320,11 +320,11 @@ void rtmsg_fib(int event, __be32 key, struct fib_alias *fa,
 		kfree_skb(skb);
 		goto errout;
 	}
-	err = rtnl_notify(skb, info->pid, RTNLGRP_IPV4_ROUTE,
+	err = rtnl_notify(skb, &init_net, info->pid, RTNLGRP_IPV4_ROUTE,
 			  info->nlh, GFP_KERNEL);
 errout:
 	if (err < 0)
-		rtnl_set_sk_err(RTNLGRP_IPV4_ROUTE, err);
+		rtnl_set_sk_err(&init_net, RTNLGRP_IPV4_ROUTE, err);
 }
 
 /* Return the first fib alias matching TOS with
diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
index b8b4b49..c077cc0 100644
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -321,7 +321,7 @@ static void ipmr_destroy_unres(struct mfc_cache *c)
 			e->error = -ETIMEDOUT;
 			memset(&e->msg, 0, sizeof(e->msg));
 
-			rtnl_unicast(skb, NETLINK_CB(skb).pid);
+			rtnl_unicast(skb, &init_net, NETLINK_CB(skb).pid);
 		} else
 			kfree_skb(skb);
 	}
@@ -533,7 +533,7 @@ static void ipmr_cache_resolve(struct mfc_cache *uc, struct mfc_cache *c)
 				memset(&e->msg, 0, sizeof(e->msg));
 			}
 
-			rtnl_unicast(skb, NETLINK_CB(skb).pid);
+			rtnl_unicast(skb, &init_net, NETLINK_CB(skb).pid);
 		} else
 			ip_mr_forward(skb, c, 0);
 	}
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index a538a30..a710b48 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -2635,7 +2635,7 @@ static int inet_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void
 	if (err <= 0)
 		goto errout_free;
 
-	err = rtnl_unicast(skb, NETLINK_CB(in_skb).pid);
+	err = rtnl_unicast(skb, &init_net, NETLINK_CB(in_skb).pid);
 errout:
 	return err;
 
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index bdc183e..7dd3f4f 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -3429,7 +3429,7 @@ static int inet6_rtm_getaddr(struct sk_buff *in_skb, struct nlmsghdr* nlh,
 		kfree_skb(skb);
 		goto errout_ifa;
 	}
-	err = rtnl_unicast(skb, NETLINK_CB(in_skb).pid);
+	err = rtnl_unicast(skb, &init_net, NETLINK_CB(in_skb).pid);
 errout_ifa:
 	in6_ifa_put(ifa);
 errout:
@@ -3452,10 +3452,10 @@ static void inet6_ifa_notify(int event, struct inet6_ifaddr *ifa)
 		kfree_skb(skb);
 		goto errout;
 	}
-	err = rtnl_notify(skb, 0, RTNLGRP_IPV6_IFADDR, NULL, GFP_ATOMIC);
+	err = rtnl_notify(skb, &init_net, 0, RTNLGRP_IPV6_IFADDR, NULL, GFP_ATOMIC);
 errout:
 	if (err < 0)
-		rtnl_set_sk_err(RTNLGRP_IPV6_IFADDR, err);
+		rtnl_set_sk_err(&init_net, RTNLGRP_IPV6_IFADDR, err);
 }
 
 static inline void ipv6_store_devconf(struct ipv6_devconf *cnf,
@@ -3660,10 +3660,10 @@ void inet6_ifinfo_notify(int event, struct inet6_dev *idev)
 		kfree_skb(skb);
 		goto errout;
 	}
-	err = rtnl_notify(skb, 0, RTNLGRP_IPV6_IFADDR, NULL, GFP_ATOMIC);
+	err = rtnl_notify(skb, &init_net, 0, RTNLGRP_IPV6_IFADDR, NULL, GFP_ATOMIC);
 errout:
 	if (err < 0)
-		rtnl_set_sk_err(RTNLGRP_IPV6_IFADDR, err);
+		rtnl_set_sk_err(&init_net, RTNLGRP_IPV6_IFADDR, err);
 }
 
 static inline size_t inet6_prefix_nlmsg_size(void)
@@ -3729,10 +3729,10 @@ static void inet6_prefix_notify(int event, struct inet6_dev *idev,
 		kfree_skb(skb);
 		goto errout;
 	}
-	err = rtnl_notify(skb, 0, RTNLGRP_IPV6_PREFIX, NULL, GFP_ATOMIC);
+	err = rtnl_notify(skb, &init_net, 0, RTNLGRP_IPV6_PREFIX, NULL, GFP_ATOMIC);
 errout:
 	if (err < 0)
-		rtnl_set_sk_err(RTNLGRP_IPV6_PREFIX, err);
+		rtnl_set_sk_err(&init_net, RTNLGRP_IPV6_PREFIX, err);
 }
 
 static void __ipv6_ifa_notify(int event, struct inet6_ifaddr *ifp)
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index f0cf11c..849dfd1 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -2307,7 +2307,7 @@ static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void
 		goto errout;
 	}
 
-	err = rtnl_unicast(skb, NETLINK_CB(in_skb).pid);
+	err = rtnl_unicast(skb, &init_net, NETLINK_CB(in_skb).pid);
 errout:
 	return err;
 }
@@ -2337,10 +2337,10 @@ void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info)
 		kfree_skb(skb);
 		goto errout;
 	}
-	err = rtnl_notify(skb, pid, RTNLGRP_IPV6_ROUTE, nlh, gfp_any());
+	err = rtnl_notify(skb, &init_net, pid, RTNLGRP_IPV6_ROUTE, nlh, gfp_any());
 errout:
 	if (err < 0)
-		rtnl_set_sk_err(RTNLGRP_IPV6_ROUTE, err);
+		rtnl_set_sk_err(&init_net, RTNLGRP_IPV6_ROUTE, err);
 }
 
 /*
diff --git a/net/sched/act_api.c b/net/sched/act_api.c
index 8528291..8150647 100644
--- a/net/sched/act_api.c
+++ b/net/sched/act_api.c
@@ -660,7 +660,7 @@ act_get_notify(u32 pid, struct nlmsghdr *n, struct tc_action *a, int event)
 		return -EINVAL;
 	}
 
-	return rtnl_unicast(skb, pid);
+	return rtnl_unicast(skb, &init_net, pid);
 }
 
 static struct tc_action *
@@ -781,7 +781,7 @@ static int tca_action_flush(struct rtattr *rta, struct nlmsghdr *n, u32 pid)
 	nlh->nlmsg_flags |= NLM_F_ROOT;
 	module_put(a->ops->owner);
 	kfree(a);
-	err = rtnetlink_send(skb, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
+	err = rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
 	if (err > 0)
 		return 0;
 
@@ -844,7 +844,7 @@ tca_action_gd(struct rtattr *rta, struct nlmsghdr *n, u32 pid, int event)
 
 		/* now do the delete */
 		tcf_action_destroy(head, 0);
-		ret = rtnetlink_send(skb, pid, RTNLGRP_TC,
+		ret = rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC,
 				     n->nlmsg_flags&NLM_F_ECHO);
 		if (ret > 0)
 			return 0;
@@ -888,7 +888,7 @@ static int tcf_add_notify(struct tc_action *a, u32 pid, u32 seq, int event,
 	nlh->nlmsg_len = skb_tail_pointer(skb) - b;
 	NETLINK_CB(skb).dst_group = RTNLGRP_TC;
 
-	err = rtnetlink_send(skb, pid, RTNLGRP_TC, flags&NLM_F_ECHO);
+	err = rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, flags&NLM_F_ECHO);
 	if (err > 0)
 		err = 0;
 	return err;
diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
index 2754e50..9dcdc88 100644
--- a/net/sched/cls_api.c
+++ b/net/sched/cls_api.c
@@ -361,7 +361,7 @@ static int tfilter_notify(struct sk_buff *oskb, struct nlmsghdr *n,
 		return -EINVAL;
 	}
 
-	return rtnetlink_send(skb, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
+	return rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
 }
 
 struct tcf_dump_args
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index 95eb0a8..ddbae5a 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -872,7 +872,7 @@ static int qdisc_notify(struct sk_buff *oskb, struct nlmsghdr *n,
 	}
 
 	if (skb->len)
-		return rtnetlink_send(skb, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
+		return rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
 
 err_out:
 	kfree_skb(skb);
@@ -1103,7 +1103,7 @@ static int tclass_notify(struct sk_buff *oskb, struct nlmsghdr *n,
 		return -EINVAL;
 	}
 
-	return rtnetlink_send(skb, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
+	return rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
 }
 
 struct qdisc_dump_args
diff --git a/net/wireless/wext.c b/net/wireless/wext.c
index 85e5f9d..85805f2 100644
--- a/net/wireless/wext.c
+++ b/net/wireless/wext.c
@@ -1137,7 +1137,7 @@ static void wireless_nlevent_process(unsigned long data)
 	struct sk_buff *skb;
 
 	while ((skb = skb_dequeue(&wireless_nlevent_queue)))
-		rtnl_notify(skb, 0, RTNLGRP_LINK, NULL, GFP_ATOMIC);
+		rtnl_notify(skb, &init_net, 0, RTNLGRP_LINK, NULL, GFP_ATOMIC);
 }
 
 static DECLARE_TASKLET(wireless_nlevent_tasklet, wireless_nlevent_process, 0);
@@ -1189,6 +1189,9 @@ static void rtmsg_iwinfo(struct net_device *dev, char *event, int event_len)
 	struct sk_buff *skb;
 	int err;
 
+	if (dev->nd_net != &init_net)
+		return;
+
 	skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC);
 	if (!skb)
 		return;
-- 
1.5.3.rc6.17.g1911

Re: [PATCH 2/5] net: Make rtnetlink infrastructure network namespace aware

From: Patrick McHardy <hidden>
Date: 2007-09-29 15:44:57 

Eric W. Biederman wrote:
 void rtnl_unlock(void)
 {
-	mutex_unlock(&rtnl_mutex);
-	if (rtnl && rtnl->sk_receive_queue.qlen)
+	struct net *net;
+
+	/*
+	 * Loop through all of the rtnl sockets until none of them (in
+	 * a live network namespace) have queue packets.
+	 *
+	 * We have to be careful with the locking here as
+	 * sk_data_ready aka rtnetlink_rcv takes the rtnl_mutex.
+	 *
+	 * To ensure the network namespace does not exit while
+	 * we are processing packets on it's rtnl socket we
+	 * grab a reference to the network namespace, ignoring
+	 * it if the network namespace has already exited.
+	 */
+retry:
+	for_each_net(net) {
+		struct sock *rtnl = net->rtnl;
+
+		if (!rtnl || !rtnl->sk_receive_queue.qlen)
+			continue;
+
+		if (!maybe_get_net(net))
+			continue;
+
+		mutex_unlock(&rtnl_mutex);
 		rtnl->sk_data_ready(rtnl, 0);
+		mutex_lock(&rtnl_mutex);
+		put_net(net);
+		goto retry;
+	}
+	mutex_unlock(&rtnl_mutex);
+
 	netdev_run_todo();
 }

I'm wondering why this receive queue processing on unlock is still
necessary today, we don't do trylock in rtnetlink_rcv anymore, so
all senders will simply wait until the lock is released and then
process the queue.

Re: [PATCH 2/5] net: Make rtnetlink infrastructure network namespace aware

From: Eric W. Biederman <hidden>
Date: 2007-09-29 16:51:47 

Patrick McHardy [off-list ref] writes:

I'm wondering why this receive queue processing on unlock is still
necessary today, we don't do trylock in rtnetlink_rcv anymore, so
all senders will simply wait until the lock is released and then
process the queue.
Good question, I should probably look.  I was lazy and didn't go back
and audit why we were doing this.  I just coded a routine that I was
certain would work.  It does appear that we are processing the queue
with sk_data_read when we add a message, so this may be completely
unnecessary.  I will go back and look.  If we can remove this bit
things should be simpler.

Eric

Re: [PATCH 2/5] net: Make rtnetlink infrastructure network namespace aware

From: Patrick McHardy <hidden>
Date: 2007-09-29 17:48:27 

Eric W. Biederman wrote:
Patrick McHardy [off-list ref] writes:
quoted
I'm wondering why this receive queue processing on unlock is still
necessary today, we don't do trylock in rtnetlink_rcv anymore, so
all senders will simply wait until the lock is released and then
process the queue.

Good question, I should probably look.  I was lazy and didn't go back
and audit why we were doing this.  I just coded a routine that I was
certain would work.  It does appear that we are processing the queue
with sk_data_read when we add a message, so this may be completely
unnecessary.  I will go back and look.  If we can remove this bit
things should be simpler.

Maybe I can save you some time: we used to do down_trylock()
for the rtnl mutex, so senders would simply return if someone
else was already processing the queue *or* the rtnl was locked
for some other reason. In the first case the process already
processing the queue would also process the new messages, but
if it the rtnl was locked for some other reason (for example
during module registration) the message would sit in the
queue until the next rtnetlink sendmsg call, which is why
rtnl_unlock does queue processing. Commit 6756ae4b changed
the down_trylock to mutex_lock, so senders will now simply wait
until the mutex is released and then call netlink_run_queue
themselves. This means its not needed anymore.

Re: [PATCH 2/5] net: Make rtnetlink infrastructure network namespace aware

From: Eric W. Biederman <hidden>
Date: 2007-09-29 21:01:43 

Patrick McHardy [off-list ref] writes:

Maybe I can save you some time: we used to do down_trylock()
for the rtnl mutex, so senders would simply return if someone
else was already processing the queue *or* the rtnl was locked
for some other reason. In the first case the process already
processing the queue would also process the new messages, but
if it the rtnl was locked for some other reason (for example
during module registration) the message would sit in the
queue until the next rtnetlink sendmsg call, which is why
rtnl_unlock does queue processing. Commit 6756ae4b changed
the down_trylock to mutex_lock, so senders will now simply wait
until the mutex is released and then call netlink_run_queue
themselves. This means its not needed anymore.
Sounds reasonable.

I started looking through the code paths and I currently cannot
see anything that would leave a message on a kernel rtnl socket.

However I did a quick test adding a WARN_ON if there were any messages
found in the queue during rtnl_unlock and I found this code path
getting invoked from linkwatch_event.  So there is clearly something I
don't understand, and it sounds at odds just a bit from your
description.

If we can remove the extra queue processing that would be great,
as it looks like a nice way to simplify the locking and the special
cases in the code.

Eric

Re: [Devel] Re: [PATCH 2/5] net: Make rtnetlink infrastructure network namespace aware

From: Denis V. Lunev <hidden>
Date: 2007-09-30 13:12:14 

Hmm, so it looks like we do not need this queue processing at all...

Regards,
	Den

Eric W. Biederman wrote:
Patrick McHardy [off-list ref] writes:
quoted
Maybe I can save you some time: we used to do down_trylock()
for the rtnl mutex, so senders would simply return if someone
else was already processing the queue *or* the rtnl was locked
for some other reason. In the first case the process already
processing the queue would also process the new messages, but
if it the rtnl was locked for some other reason (for example
during module registration) the message would sit in the
queue until the next rtnetlink sendmsg call, which is why
rtnl_unlock does queue processing. Commit 6756ae4b changed
the down_trylock to mutex_lock, so senders will now simply wait
until the mutex is released and then call netlink_run_queue
themselves. This means its not needed anymore.
Sounds reasonable.

I started looking through the code paths and I currently cannot
see anything that would leave a message on a kernel rtnl socket.

However I did a quick test adding a WARN_ON if there were any messages
found in the queue during rtnl_unlock and I found this code path
getting invoked from linkwatch_event.  So there is clearly something I
don't understand, and it sounds at odds just a bit from your
description.

If we can remove the extra queue processing that would be great,
as it looks like a nice way to simplify the locking and the special
cases in the code.

Eric





  

    
Re: [PATCH 2/5] net: Make rtnetlink infrastructure network namespace aware

    

      From: Patrick McHardy <hidden>

      Date: 2007-09-30 15:39:56 
    

  

  
Eric W. Biederman wrote:
Patrick McHardy [off-list ref] writes:


quoted
Maybe I can save you some time: we used to do down_trylock()
for the rtnl mutex, so senders would simply return if someone
else was already processing the queue *or* the rtnl was locked
for some other reason. In the first case the process already
processing the queue would also process the new messages, but
if it the rtnl was locked for some other reason (for example
during module registration) the message would sit in the
queue until the next rtnetlink sendmsg call, which is why
rtnl_unlock does queue processing. Commit 6756ae4b changed
the down_trylock to mutex_lock, so senders will now simply wait
until the mutex is released and then call netlink_run_queue
themselves. This means its not needed anymore.

Sounds reasonable.

I started looking through the code paths and I currently cannot
see anything that would leave a message on a kernel rtnl socket.

However I did a quick test adding a WARN_ON if there were any messages
found in the queue during rtnl_unlock and I found this code path
getting invoked from linkwatch_event.  So there is clearly something I
don't understand, and it sounds at odds just a bit from your
description.

That sounds like a bug. Did you place the WARN_ON before or after
the mutex_unlock()?





  

    
Re: [Devel] Re: [PATCH 2/5] net: Make rtnetlink infrastructure network namespace aware

    

      From: Denis V. Lunev <hidden>

      Date: 2007-10-01 08:25:49 
    

  

  
Patrick McHardy wrote:
Eric W. Biederman wrote:
quoted
Patrick McHardy [off-list ref] writes:


quoted
Maybe I can save you some time: we used to do down_trylock()
for the rtnl mutex, so senders would simply return if someone
else was already processing the queue *or* the rtnl was locked
for some other reason. In the first case the process already
processing the queue would also process the new messages, but
if it the rtnl was locked for some other reason (for example
during module registration) the message would sit in the
queue until the next rtnetlink sendmsg call, which is why
rtnl_unlock does queue processing. Commit 6756ae4b changed
the down_trylock to mutex_lock, so senders will now simply wait
until the mutex is released and then call netlink_run_queue
themselves. This means its not needed anymore.
Sounds reasonable.

I started looking through the code paths and I currently cannot
see anything that would leave a message on a kernel rtnl socket.

However I did a quick test adding a WARN_ON if there were any messages
found in the queue during rtnl_unlock and I found this code path
getting invoked from linkwatch_event.  So there is clearly something I
don't understand, and it sounds at odds just a bit from your
description.

That sounds like a bug. Did you place the WARN_ON before or after
the mutex_unlock()?
The presence of the message in the queue during rtnl_unlock is quite
possible as normal user->kernel message processing path for rtnl is the
following:

netlink_sendmsg
   netlink_unicast
      netlink_sendskb
          skb_queue_tail
          netlink_data_ready
              rtnetlink_rcv
                  mutex_lock(&rtnl_mutex);
                  netlink_run_queue(sk, qlen, &rtnetlink_rcv_msg);
                  mutex_unlock(&rtnl_mutex);

so, the presence of the packet in the rtnl queue on rtnl_unlock is
normal race with a rtnetlink_rcv for me.

Regards,
	Den





  

    
Re: [Devel] Re: [PATCH 2/5] net: Make rtnetlink infrastructure network namespace aware

    

      From: Eric W. Biederman <hidden>

      Date: 2007-10-01 08:46:33 
    

  

  
"Denis V. Lunev" [off-list ref] writes:

The presence of the message in the queue during rtnl_unlock is quite
possible as normal user->kernel message processing path for rtnl is the
following:

netlink_sendmsg
   netlink_unicast
      netlink_sendskb
          skb_queue_tail
          netlink_data_ready
              rtnetlink_rcv
                  mutex_lock(&rtnl_mutex);
                  netlink_run_queue(sk, qlen, &rtnetlink_rcv_msg);
                  mutex_unlock(&rtnl_mutex);

so, the presence of the packet in the rtnl queue on rtnl_unlock is
normal race with a rtnetlink_rcv for me.
Yes.  That is what I saw in practice as well.
Thanks for confirming this.

It happened to reproducible because I had a dhcp client asking
for a list of links in parallel with the actual link coming up
during boot.

Looking at netlink_unicast and netlink_broadcast I am generally
convinced that we can remove the call of sk_data_ready in
rtnl_unlock.   I think those are the only two possible paths
through there and I don't see how we could miss a processing a
packet on the way through there.

What would be nice is if we could figure out how to eliminate
this race.  As that would allow netlink packets to be processed
synchronously and we could actually use current for security
checks, and for getting the context of the calling process.

Right now we are 99% of the way there but because of the above
race the code must all be written as if netlink packets were coming
in completely asynchronously.  Which is unfortunate and a pain.

Eric





  

    
Re: [Devel] [PATCH 1/5] net: Modify all rtnetlink methods to only work in the initial namespace

    

      From: Denis V. Lunev <hidden>

      Date: 2007-10-10 12:33:32 
    

  

  
Eric W. Biederman wrote:
Before I can enable rtnetlink to work in all network namespaces
I need to be certain that something won't break.  So this
patch deliberately disables all of the rtnletlink methods in everything
except the initial network namespace.  After the methods have been
audited this extra check can be disabled.
[...]
 static int br_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	struct net_device *dev;
 	int idx;
 
I've read some code today greping 'init_net.loopback_dev' and found
interesting non-trivial for me issue.

Network namespace is extracted from the packet in two different ways in
TCP. This is a socket for outgoing path and a device for incoming.
Though, there are some places called uniformly both from incoming and
outgoing path.

Typical example is netfilters. They are called uniformly all around the
code. The prototype is the following:

static unsigned int reject6_target(struct sk_buff **pskb,
                           const struct net_device *in,
                           const struct net_device *out,
                           unsigned int hooknum,
                           const struct xt_target *target,
                           const void *targinfo);

So, we are bound to the following options:
- perform additional non-uniform hacks around to place 'struct net' into
  other and other structures like xt_target
- add 7th parameter here and over
- introduce an skb_net field in the 'struct sk_buff' making all code
  uniform, at least when we have an skb

I think that this is not the last place with such a parameter list and
we should make a decision at this point when the code in not mainline yet.

As far as I understand, netfilters are not touched by the Eric and we
can face some non-trivial problems there.

So, if my point about uniformity is valid, this patchset looks wrong and
should be re-worked :(

Regards,
	Den





  

    
Re: [Devel] [PATCH 1/5] net: Modify all rtnetlink methods to only work in the initial namespace

    

      From: Daniel Lezcano <hidden>

      Date: 2007-10-10 14:11:48 
    

  

  
Denis V. Lunev wrote:
Eric W. Biederman wrote:
quoted
Before I can enable rtnetlink to work in all network namespaces
I need to be certain that something won't break.  So this
patch deliberately disables all of the rtnletlink methods in everything
except the initial network namespace.  After the methods have been
audited this extra check can be disabled.
[...]
quoted
 static int br_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	struct net_device *dev;
 	int idx;
 
I've read some code today greping 'init_net.loopback_dev' and found
interesting non-trivial for me issue.

Network namespace is extracted from the packet in two different ways in
TCP. This is a socket for outgoing path and a device for incoming.
Though, there are some places called uniformly both from incoming and
outgoing path.

Typical example is netfilters. They are called uniformly all around the
code. The prototype is the following:

static unsigned int reject6_target(struct sk_buff **pskb,
                           const struct net_device *in,
                           const struct net_device *out,
                           unsigned int hooknum,
                           const struct xt_target *target,
                           const void *targinfo);
Thanks Denis for auditing the code.

As far as I see, struct net_device *in is NULL for outgoing traffic and 
struct net_device *out is NULL for ingress traffic. Except for the 
FORWARD rules where both are filled. If we are following network 
namespace semantic, we should not have two network devices belonging to 
two differents namespaces, right ?
In this case, the following line of code should be sufficient to 
retrieve the network namespace, no ?

struct net *net = in?in->nd_net:out->nd_net;

So, we are bound to the following options:
- perform additional non-uniform hacks around to place 'struct net' into
  other and other structures like xt_target
- add 7th parameter here and over
- introduce an skb_net field in the 'struct sk_buff' making all code
  uniform, at least when we have an skb

I think that this is not the last place with such a parameter list and
we should make a decision at this point when the code in not mainline yet.

As far as I understand, netfilters are not touched by the Eric and we
can face some non-trivial problems there.
In Eric's git tree:
http://git.kernel.org/?p=linux/kernel/git/ebiederm/linux-2.6-netns.git

There are some modifications concerning 
net/ipv4/netfiler/iptable_filter.c and at the ipt_hook function, there is:

struct net *net = (in?in:out)->nd_net;

So, if my point about uniformity is valid, this patchset looks wrong and
should be re-worked :(
As Eric said, we want to build the network namespace step by step, 
taking care of not breaking the init network namespace.

If you want to make iptables per namespace or catch problems before the 
code goes to Dave's tree, IMHO it will be more convenient to post to 
containers@ the patches against netns49, where the modifications will be 
in a network namespace big picture.

Regards.

   -- Daniel






  

    
Re: [Devel] [PATCH 1/5] net: Modify all rtnetlink methods to only work in the initial namespace

    

      From: Denis V. Lunev <hidden>

      Date: 2007-10-10 14:27:35 
    

  

  
Daniel Lezcano wrote:
struct net *net = in?in->nd_net:out->nd_net;

quoted
So, we are bound to the following options:
- perform additional non-uniform hacks around to place 'struct net' into
  other and other structures like xt_target
- add 7th parameter here and over
- introduce an skb_net field in the 'struct sk_buff' making all code
  uniform, at least when we have an skb

I think that this is not the last place with such a parameter list and
we should make a decision at this point when the code in not mainline
yet.

As far as I understand, netfilters are not touched by the Eric and we
can face some non-trivial problems there.
In Eric's git tree:
http://git.kernel.org/?p=linux/kernel/git/ebiederm/linux-2.6-netns.git

There are some modifications concerning
net/ipv4/netfiler/iptable_filter.c and at the ipt_hook function, there is:

struct net *net = (in?in:out)->nd_net;

quoted
So, if my point about uniformity is valid, this patchset looks wrong and
should be re-worked :(
As Eric said, we want to build the network namespace step by step,
taking care of not breaking the init network namespace.

If you want to make iptables per namespace or catch problems before the
code goes to Dave's tree, IMHO it will be more convenient to post to
containers@ the patches against netns49, where the modifications will be
in a network namespace big picture.
my point is somewhat another. Yes, this is enough for that place. If so,
I must scatter these checks all around in the netfilters code. Brr.

In forward chain the situation is different for Layer3 switching. Let's
assume that we have an OpenVZ scheme, where the packet flows from socket
to device and after that from device to device via forwarding path. You
can't call skb_orphan on namespace switching as this breaks UDP flow
regulation. Virtual network device is fast while real Ethernet is slow,
packets will be dropped on queue in real device. So, the situation with
packet on send path with a socket from other namespace is possible :(

I just pray for uniformity to concentrate on the code rather than on
guesses on which path we are :(

Regards,
	Den





  

    
Re: [Devel] [PATCH 1/5] net: Modify all rtnetlink methods to only work in the initial namespace

    

      From: Eric W. Biederman <hidden>

      Date: 2007-10-10 19:38:48 
    

  

  
"Denis V. Lunev" [off-list ref] writes:

Eric W. Biederman wrote:
quoted
Before I can enable rtnetlink to work in all network namespaces
I need to be certain that something won't break.  So this
patch deliberately disables all of the rtnletlink methods in everything
except the initial network namespace.  After the methods have been
audited this extra check can be disabled.
[...]
quoted
 static int br_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
 {
+	struct net *net = skb->sk->sk_net;
 	struct net_device *dev;
 	int idx;
 
I've read some code today greping 'init_net.loopback_dev' and found
interesting non-trivial for me issue.

Network namespace is extracted from the packet in two different ways in
TCP. This is a socket for outgoing path and a device for incoming.
Though, there are some places called uniformly both from incoming and
outgoing path.

Typical example is netfilters. They are called uniformly all around the
code. The prototype is the following:

static unsigned int reject6_target(struct sk_buff **pskb,
                           const struct net_device *in,
                           const struct net_device *out,
                           unsigned int hooknum,
                           const struct xt_target *target,
                           const void *targinfo);

So, we are bound to the following options:
- perform additional non-uniform hacks around to place 'struct net' into
  other and other structures like xt_target
- add 7th parameter here and over
- introduce an skb_net field in the 'struct sk_buff' making all code
  uniform, at least when we have an skb
No.  That bloats a sk_buff, changes the semantics of moving a skb
around, and decreases performance (because we have to maintain the
field on a fast path).  

There will not be a skb_net field.

The entire concept of skb_net is a maintenance disaster.

I think that this is not the last place with such a parameter list and
we should make a decision at this point when the code in not mainline yet.
Certainly that is what I have a proof of concept tree for.  So we can
see how these things look before we merge them. 

As far as I understand, netfilters are not touched by the Eric and we
can face some non-trivial problems there.
No.  In my proof of concept tree I should have working per network
namespace netfilter code.  My intention was to just do enough to see
what the impact would be so most of the netfilter code (in my tree)
insists on running in the initial network namespace.  But there are
a few pieces that are fully converted.  Please take a look.

So, if my point about uniformity is valid, this patchset looks wrong and
should be re-worked :(
This patchset does need to get rebased on top of net-2.6.25 when it
opens and hopefully your patchset to remove the unnecessary work in
rtnl_unlock, and to really process netlink requests in process
context.  I see a need for the more fundamental change you seem to
be advocating.

Differentiating between the incoming and the outgoing code paths is
something we already do permission checking, for locking, for
sleeping, etc.  Modifying the code requires reading and understanding
it in context.  That is the nature of code.

This does make large patches going across the entire networking stack
making something a network namespace parameter difficult, but it
should not cause any problem for maintenance or other work on
the code.  As shown by the fact that even outside the tree rebasing my
network namespace patches has not been all that difficult.

So no I don't think uniformity, or beauty or elegance is what we
are after right now.  Trying to hard in that direction ultimately
obfuscates the code.

What we want is something that is simple, straight forward, and
doesn't require you to be an expert in network namespaces to
understand the code or the patches.

In the particular case of the netfilter hooks we don't have a
network namespace parameter laying around before we call NF_HOOK,
and the idiom "net = (in?in:out)->nd_net" seems perfectly accurate
so it seems reasonable to me to derive the network namespace that
way in generic code.  Although thinking about this.  We know which
hooks we are being called from so we may in fact actually know
if which of in or out must be valid when we get to the netfilter
hook.

Eric




    
  

  
    

      

        Keyboard shortcuts
      

      

        
          
hback out one level

          
jnext message in thread

          
kprevious message in thread

          
ldrill in

          
Escclose help / fold thread tree

          
?toggle this help