netns: fix GFP flags in rtnl_net_notifyid()

In rtnl_net_notifyid(), we certainly can't pass a null GFP flag to
rtnl_notify(). A GFP_KERNEL flag would be fine in most circumstances,
but there are a few paths calling rtnl_net_notifyid() from atomic
context or from RCU critical sections. The later also precludes the use
of gfp_any() as it wouldn't detect the RCU case. Also, the nlmsg_new()
call is wrong too, as it uses GFP_KERNEL unconditionally.

Therefore, we need to pass the GFP flags as parameter and propagate it
through function calls until the proper flags can be determined.

In most cases, GFP_KERNEL is fine. The exceptions are:
  * openvswitch: ovs_vport_cmd_get() and ovs_vport_cmd_dump()
    indirectly call rtnl_net_notifyid() from RCU critical section,

  * rtnetlink: rtmsg_ifinfo_build_skb() already receives GFP flags as
    parameter.

Also, in ovs_vport_cmd_build_info(), let's change the GFP flags used
by nlmsg_new(). The function is allowed to sleep, so better make the
flags consistent with the ones used in the following
ovs_vport_cmd_fill_info() call.

Found by code inspection.

Fixes: 9a963454 ("netns: notify netns id events")
Signed-off-by: Guillaume Nault <gnault@redhat.com>
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

include/net/net_namespace.h

@@ -342,7 +342,7 @@ static inline struct net *read_pnet(const possible_net_t *pnet)
 #define __net_initconst	__initconst
 #endif
 
-int peernet2id_alloc(struct net *net, struct net *peer);
+int peernet2id_alloc(struct net *net, struct net *peer, gfp_t gfp);
 int peernet2id(struct net *net, struct net *peer);
 bool peernet_has_id(struct net *net, struct net *peer);
 struct net *get_net_ns_by_id(struct net *net, int id);

net/core/dev.c

@@ -9770,7 +9770,7 @@ int dev_change_net_namespace(struct net_device *dev, struct net *net, const char
 	call_netdevice_notifiers(NETDEV_UNREGISTER, dev);
 	rcu_barrier();
 
-	new_nsid = peernet2id_alloc(dev_net(dev), net);
+	new_nsid = peernet2id_alloc(dev_net(dev), net, GFP_KERNEL);
 	/* If there is an ifindex conflict assign a new one */
 	if (__dev_get_by_index(net, dev->ifindex))
 		new_ifindex = dev_new_index(net);

net/core/net_namespace.c

@@ -246,11 +246,11 @@ static int __peernet2id(struct net *net, struct net *peer)
 }
 
 static void rtnl_net_notifyid(struct net *net, int cmd, int id, u32 portid,
-			      struct nlmsghdr *nlh);
+			      struct nlmsghdr *nlh, gfp_t gfp);
 /* This function returns the id of a peer netns. If no id is assigned, one will
  * be allocated and returned.
  */
-int peernet2id_alloc(struct net *net, struct net *peer)
+int peernet2id_alloc(struct net *net, struct net *peer, gfp_t gfp)
 {
 	bool alloc = false, alive = false;
 	int id;
@@ -269,7 +269,7 @@ int peernet2id_alloc(struct net *net, struct net *peer)
 	id = __peernet2id_alloc(net, peer, &alloc);
 	spin_unlock_bh(&net->nsid_lock);
 	if (alloc && id >= 0)
-		rtnl_net_notifyid(net, RTM_NEWNSID, id, 0, NULL);
+		rtnl_net_notifyid(net, RTM_NEWNSID, id, 0, NULL, gfp);
 	if (alive)
 		put_net(peer);
 	return id;
@@ -534,7 +534,8 @@ static void unhash_nsid(struct net *net, struct net *last)
 			idr_remove(&tmp->netns_ids, id);
 		spin_unlock_bh(&tmp->nsid_lock);
 		if (id >= 0)
-			rtnl_net_notifyid(tmp, RTM_DELNSID, id, 0, NULL);
+			rtnl_net_notifyid(tmp, RTM_DELNSID, id, 0, NULL,
+					  GFP_KERNEL);
 		if (tmp == last)
 			break;
 	}
@@ -767,7 +768,7 @@ static int rtnl_net_newid(struct sk_buff *skb, struct nlmsghdr *nlh,
 	spin_unlock_bh(&net->nsid_lock);
 	if (err >= 0) {
 		rtnl_net_notifyid(net, RTM_NEWNSID, err, NETLINK_CB(skb).portid,
-				  nlh);
+				  nlh, GFP_KERNEL);
 		err = 0;
 	} else if (err == -ENOSPC && nsid >= 0) {
 		err = -EEXIST;
@@ -1055,7 +1056,7 @@ static int rtnl_net_dumpid(struct sk_buff *skb, struct netlink_callback *cb)
 }
 
 static void rtnl_net_notifyid(struct net *net, int cmd, int id, u32 portid,
-			      struct nlmsghdr *nlh)
+			      struct nlmsghdr *nlh, gfp_t gfp)
 {
 	struct net_fill_args fillargs = {
 		.portid = portid,
@@ -1066,7 +1067,7 @@ static void rtnl_net_notifyid(struct net *net, int cmd, int id, u32 portid,
 	struct sk_buff *msg;
 	int err = -ENOMEM;
 
-	msg = nlmsg_new(rtnl_net_get_size(), GFP_KERNEL);
+	msg = nlmsg_new(rtnl_net_get_size(), gfp);
 	if (!msg)
 		goto out;
 
@@ -1074,7 +1075,7 @@ static void rtnl_net_notifyid(struct net *net, int cmd, int id, u32 portid,
 	if (err < 0)
 		goto err_out;
 
-	rtnl_notify(msg, net, portid, RTNLGRP_NSID, nlh, 0);
+	rtnl_notify(msg, net, portid, RTNLGRP_NSID, nlh, gfp);
 	return;
 
 err_out:

net/core/rtnetlink.c

@@ -1523,7 +1523,7 @@ static noinline_for_stack int nla_put_ifalias(struct sk_buff *skb,
 
 static int rtnl_fill_link_netnsid(struct sk_buff *skb,
 				  const struct net_device *dev,
-				  struct net *src_net)
+				  struct net *src_net, gfp_t gfp)
 {
 	bool put_iflink = false;
 
@@ -1531,7 +1531,7 @@ static int rtnl_fill_link_netnsid(struct sk_buff *skb,
 		struct net *link_net = dev->rtnl_link_ops->get_link_net(dev);
 
 		if (!net_eq(dev_net(dev), link_net)) {
-			int id = peernet2id_alloc(src_net, link_net);
+			int id = peernet2id_alloc(src_net, link_net, gfp);
 
 			if (nla_put_s32(skb, IFLA_LINK_NETNSID, id))
 				return -EMSGSIZE;
@@ -1589,7 +1589,7 @@ static int rtnl_fill_ifinfo(struct sk_buff *skb,
 			    int type, u32 pid, u32 seq, u32 change,
 			    unsigned int flags, u32 ext_filter_mask,
 			    u32 event, int *new_nsid, int new_ifindex,
-			    int tgt_netnsid)
+			    int tgt_netnsid, gfp_t gfp)
 {
 	struct ifinfomsg *ifm;
 	struct nlmsghdr *nlh;
@@ -1681,7 +1681,7 @@ static int rtnl_fill_ifinfo(struct sk_buff *skb,
 			goto nla_put_failure;
 	}
 
-	if (rtnl_fill_link_netnsid(skb, dev, src_net))
+	if (rtnl_fill_link_netnsid(skb, dev, src_net, gfp))
 		goto nla_put_failure;
 
 	if (new_nsid &&
@@ -2001,7 +2001,7 @@ static int rtnl_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
 					       NETLINK_CB(cb->skb).portid,
 					       nlh->nlmsg_seq, 0, flags,
 					       ext_filter_mask, 0, NULL, 0,
-					       netnsid);
+					       netnsid, GFP_KERNEL);
 
 			if (err < 0) {
 				if (likely(skb->len))
@@ -3360,7 +3360,7 @@ static int rtnl_getlink(struct sk_buff *skb, struct nlmsghdr *nlh,
 	err = rtnl_fill_ifinfo(nskb, dev, net,
 			       RTM_NEWLINK, NETLINK_CB(skb).portid,
 			       nlh->nlmsg_seq, 0, 0, ext_filter_mask,
-			       0, NULL, 0, netnsid);
+			       0, NULL, 0, netnsid, GFP_KERNEL);
 	if (err < 0) {
 		/* -EMSGSIZE implies BUG in if_nlmsg_size */
 		WARN_ON(err == -EMSGSIZE);
@@ -3472,7 +3472,7 @@ struct sk_buff *rtmsg_ifinfo_build_skb(int type, struct net_device *dev,
 
 	err = rtnl_fill_ifinfo(skb, dev, dev_net(dev),
 			       type, 0, 0, change, 0, 0, event,
-			       new_nsid, new_ifindex, -1);
+			       new_nsid, new_ifindex, -1, flags);
 	if (err < 0) {
 		/* -EMSGSIZE implies BUG in if_nlmsg_size() */
 		WARN_ON(err == -EMSGSIZE);

net/openvswitch/datapath.c

@@ -1881,7 +1881,7 @@ static struct genl_family dp_datapath_genl_family __ro_after_init = {
 /* Called with ovs_mutex or RCU read lock. */
 static int ovs_vport_cmd_fill_info(struct vport *vport, struct sk_buff *skb,
 				   struct net *net, u32 portid, u32 seq,
-				   u32 flags, u8 cmd)
+				   u32 flags, u8 cmd, gfp_t gfp)
 {
 	struct ovs_header *ovs_header;
 	struct ovs_vport_stats vport_stats;
@@ -1902,7 +1902,7 @@ static int ovs_vport_cmd_fill_info(struct vport *vport, struct sk_buff *skb,
 		goto nla_put_failure;
 
 	if (!net_eq(net, dev_net(vport->dev))) {
-		int id = peernet2id_alloc(net, dev_net(vport->dev));
+		int id = peernet2id_alloc(net, dev_net(vport->dev), gfp);
 
 		if (nla_put_s32(skb, OVS_VPORT_ATTR_NETNSID, id))
 			goto nla_put_failure;
@@ -1943,11 +1943,12 @@ struct sk_buff *ovs_vport_cmd_build_info(struct vport *vport, struct net *net,
 	struct sk_buff *skb;
 	int retval;
 
-	skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC);
+	skb = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
 	if (!skb)
 		return ERR_PTR(-ENOMEM);
 
-	retval = ovs_vport_cmd_fill_info(vport, skb, net, portid, seq, 0, cmd);
+	retval = ovs_vport_cmd_fill_info(vport, skb, net, portid, seq, 0, cmd,
+					 GFP_KERNEL);
 	BUG_ON(retval < 0);
 
 	return skb;
@@ -2089,7 +2090,7 @@ static int ovs_vport_cmd_new(struct sk_buff *skb, struct genl_info *info)
 
 	err = ovs_vport_cmd_fill_info(vport, reply, genl_info_net(info),
 				      info->snd_portid, info->snd_seq, 0,
-				      OVS_VPORT_CMD_NEW);
+				      OVS_VPORT_CMD_NEW, GFP_KERNEL);
 
 	new_headroom = netdev_get_fwd_headroom(vport->dev);
 
@@ -2150,7 +2151,7 @@ static int ovs_vport_cmd_set(struct sk_buff *skb, struct genl_info *info)
 
 	err = ovs_vport_cmd_fill_info(vport, reply, genl_info_net(info),
 				      info->snd_portid, info->snd_seq, 0,
-				      OVS_VPORT_CMD_SET);
+				      OVS_VPORT_CMD_SET, GFP_KERNEL);
 	BUG_ON(err < 0);
 
 	ovs_unlock();
@@ -2190,7 +2191,7 @@ static int ovs_vport_cmd_del(struct sk_buff *skb, struct genl_info *info)
 
 	err = ovs_vport_cmd_fill_info(vport, reply, genl_info_net(info),
 				      info->snd_portid, info->snd_seq, 0,
-				      OVS_VPORT_CMD_DEL);
+				      OVS_VPORT_CMD_DEL, GFP_KERNEL);
 	BUG_ON(err < 0);
 
 	/* the vport deletion may trigger dp headroom update */
@@ -2237,7 +2238,7 @@ static int ovs_vport_cmd_get(struct sk_buff *skb, struct genl_info *info)
 		goto exit_unlock_free;
 	err = ovs_vport_cmd_fill_info(vport, reply, genl_info_net(info),
 				      info->snd_portid, info->snd_seq, 0,
-				      OVS_VPORT_CMD_GET);
+				      OVS_VPORT_CMD_GET, GFP_ATOMIC);
 	BUG_ON(err < 0);
 	rcu_read_unlock();
 
@@ -2273,7 +2274,8 @@ static int ovs_vport_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb)
 						    NETLINK_CB(cb->skb).portid,
 						    cb->nlh->nlmsg_seq,
 						    NLM_F_MULTI,
-						    OVS_VPORT_CMD_GET) < 0)
+						    OVS_VPORT_CMD_GET,
+						    GFP_ATOMIC) < 0)
 				goto out;
 
 			j++;